{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999542857142857,
  "eval_steps": 500,
  "global_step": 10937,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.530778170634292,
      "learning_rate": 3.0395136778115507e-08,
      "loss": 0.3758,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.27550326563542,
      "learning_rate": 6.079027355623101e-08,
      "loss": 0.2946,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.401044234045929,
      "learning_rate": 9.118541033434651e-08,
      "loss": 0.2811,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5897927931655014,
      "learning_rate": 1.2158054711246203e-07,
      "loss": 0.3238,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.768141493923043,
      "learning_rate": 1.5197568389057753e-07,
      "loss": 0.3192,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.67173285056069,
      "learning_rate": 1.8237082066869301e-07,
      "loss": 0.7868,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6584616658265654,
      "learning_rate": 2.1276595744680852e-07,
      "loss": 0.3423,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.025487768703574,
      "learning_rate": 2.4316109422492405e-07,
      "loss": 0.3172,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.551415769484005,
      "learning_rate": 2.7355623100303953e-07,
      "loss": 0.2692,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.022686163445313,
      "learning_rate": 3.0395136778115507e-07,
      "loss": 0.2903,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6161743245464146,
      "learning_rate": 3.3434650455927055e-07,
      "loss": 0.2711,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.644899724932609,
      "learning_rate": 3.6474164133738603e-07,
      "loss": 0.8621,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.0462537756445895,
      "learning_rate": 3.9513677811550156e-07,
      "loss": 0.917,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.158162567929741,
      "learning_rate": 4.2553191489361704e-07,
      "loss": 0.2994,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.959685015230075,
      "learning_rate": 4.5592705167173257e-07,
      "loss": 0.3842,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8550038686181285,
      "learning_rate": 4.863221884498481e-07,
      "loss": 0.2677,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3331241995214813,
      "learning_rate": 5.167173252279636e-07,
      "loss": 0.3132,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.9437248197789927,
      "learning_rate": 5.471124620060791e-07,
      "loss": 0.2952,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.722249798422083,
      "learning_rate": 5.775075987841945e-07,
      "loss": 0.3137,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.247630106445413,
      "learning_rate": 6.079027355623101e-07,
      "loss": 0.6561,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.8104001483816297,
      "learning_rate": 6.382978723404255e-07,
      "loss": 0.3403,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.611429777759041,
      "learning_rate": 6.686930091185411e-07,
      "loss": 0.3158,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.395083534998535,
      "learning_rate": 6.990881458966566e-07,
      "loss": 0.2227,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.483082556511799,
      "learning_rate": 7.294832826747721e-07,
      "loss": 0.298,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.605848587066687,
      "learning_rate": 7.598784194528875e-07,
      "loss": 0.7199,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.480550142719525,
      "learning_rate": 7.902735562310031e-07,
      "loss": 0.322,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0746913976099015,
      "learning_rate": 8.206686930091186e-07,
      "loss": 0.2723,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.511006007502499,
      "learning_rate": 8.510638297872341e-07,
      "loss": 0.3345,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1471050881915614,
      "learning_rate": 8.814589665653496e-07,
      "loss": 0.3035,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.4762943745354993,
      "learning_rate": 9.118541033434651e-07,
      "loss": 0.3388,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2652228332511974,
      "learning_rate": 9.422492401215806e-07,
      "loss": 0.2765,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.9197233013238195,
      "learning_rate": 9.726443768996962e-07,
      "loss": 0.9197,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2645585568683377,
      "learning_rate": 1.0030395136778117e-06,
      "loss": 0.2917,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.188259800163142,
      "learning_rate": 1.0334346504559272e-06,
      "loss": 0.2782,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.032893164103704,
      "learning_rate": 1.0638297872340427e-06,
      "loss": 0.3128,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.362872180144831,
      "learning_rate": 1.0942249240121581e-06,
      "loss": 0.2712,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1045520715329937,
      "learning_rate": 1.1246200607902736e-06,
      "loss": 0.2258,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0945676307883216,
      "learning_rate": 1.155015197568389e-06,
      "loss": 0.2736,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0305947291775395,
      "learning_rate": 1.1854103343465048e-06,
      "loss": 0.2747,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.013666894055341,
      "learning_rate": 1.2158054711246203e-06,
      "loss": 0.2248,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1196409810889754,
      "learning_rate": 1.2462006079027357e-06,
      "loss": 0.2962,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0508897877757892,
      "learning_rate": 1.276595744680851e-06,
      "loss": 0.2456,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.006759413636361,
      "learning_rate": 1.3069908814589665e-06,
      "loss": 0.2107,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2390595951270296,
      "learning_rate": 1.3373860182370822e-06,
      "loss": 0.2794,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0039127474049305,
      "learning_rate": 1.3677811550151977e-06,
      "loss": 0.244,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9230852937564158,
      "learning_rate": 1.3981762917933131e-06,
      "loss": 0.2639,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2360475892741465,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.2668,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0076593415650468,
      "learning_rate": 1.4589665653495441e-06,
      "loss": 0.2715,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1710325540577644,
      "learning_rate": 1.4893617021276596e-06,
      "loss": 0.2179,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1739412756502157,
      "learning_rate": 1.519756838905775e-06,
      "loss": 0.243,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.108250287148761,
      "learning_rate": 1.5501519756838905e-06,
      "loss": 0.7702,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0299501229586085,
      "learning_rate": 1.5805471124620062e-06,
      "loss": 0.2237,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1034290550660026,
      "learning_rate": 1.6109422492401217e-06,
      "loss": 0.2485,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2335100170796203,
      "learning_rate": 1.6413373860182372e-06,
      "loss": 0.187,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1487499332897047,
      "learning_rate": 1.6717325227963527e-06,
      "loss": 0.232,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.085201578860626,
      "learning_rate": 1.7021276595744682e-06,
      "loss": 0.2322,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1228871931792557,
      "learning_rate": 1.7325227963525836e-06,
      "loss": 0.2419,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2463602314130595,
      "learning_rate": 1.7629179331306991e-06,
      "loss": 0.246,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8226079517838247,
      "learning_rate": 1.7933130699088146e-06,
      "loss": 0.1953,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4849901653402546,
      "learning_rate": 1.8237082066869303e-06,
      "loss": 0.2256,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4271964315726975,
      "learning_rate": 1.8541033434650458e-06,
      "loss": 0.2369,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.286326109039245,
      "learning_rate": 1.8844984802431613e-06,
      "loss": 0.2567,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.148462047185282,
      "learning_rate": 1.9148936170212767e-06,
      "loss": 0.2157,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.073124635093232,
      "learning_rate": 1.9452887537993924e-06,
      "loss": 0.2194,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.384044024436231,
      "learning_rate": 1.9756838905775077e-06,
      "loss": 0.8111,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1107897265607223,
      "learning_rate": 2.0060790273556234e-06,
      "loss": 0.2242,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0807855823138253,
      "learning_rate": 2.0364741641337387e-06,
      "loss": 0.2571,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.17585628498883,
      "learning_rate": 2.0668693009118543e-06,
      "loss": 0.2511,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8900826790155214,
      "learning_rate": 2.0972644376899696e-06,
      "loss": 0.1948,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.96399678896824,
      "learning_rate": 2.1276595744680853e-06,
      "loss": 0.8687,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.254394873415711,
      "learning_rate": 2.158054711246201e-06,
      "loss": 0.2892,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.784424014830949,
      "learning_rate": 2.1884498480243163e-06,
      "loss": 0.9147,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.117077381200809,
      "learning_rate": 2.218844984802432e-06,
      "loss": 0.2595,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0895485094403514,
      "learning_rate": 2.2492401215805472e-06,
      "loss": 0.1782,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2511634795401227,
      "learning_rate": 2.279635258358663e-06,
      "loss": 0.2608,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.69463026618189,
      "learning_rate": 2.310030395136778e-06,
      "loss": 0.7637,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0051532847722515,
      "learning_rate": 2.340425531914894e-06,
      "loss": 0.2418,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.953116463614525,
      "learning_rate": 2.3708206686930096e-06,
      "loss": 0.2369,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0209755119704025,
      "learning_rate": 2.401215805471125e-06,
      "loss": 0.2713,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8413556796703818,
      "learning_rate": 2.4316109422492405e-06,
      "loss": 0.2085,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0386228390152366,
      "learning_rate": 2.462006079027356e-06,
      "loss": 0.268,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.178919902781247,
      "learning_rate": 2.4924012158054715e-06,
      "loss": 0.2193,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.070516823160257,
      "learning_rate": 2.5227963525835868e-06,
      "loss": 0.2612,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.746885051899326,
      "learning_rate": 2.553191489361702e-06,
      "loss": 0.1927,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0756598405270243,
      "learning_rate": 2.5835866261398177e-06,
      "loss": 0.2544,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.105156457496671,
      "learning_rate": 2.613981762917933e-06,
      "loss": 0.2268,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.120081041712307,
      "learning_rate": 2.644376899696049e-06,
      "loss": 0.2718,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.45527999219953,
      "learning_rate": 2.6747720364741644e-06,
      "loss": 0.6715,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.751234783925829,
      "learning_rate": 2.70516717325228e-06,
      "loss": 0.8094,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.059261833476642,
      "learning_rate": 2.7355623100303953e-06,
      "loss": 0.2249,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9409579744625887,
      "learning_rate": 2.765957446808511e-06,
      "loss": 0.2301,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.961399429539251,
      "learning_rate": 2.7963525835866263e-06,
      "loss": 0.2667,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9341199054501244,
      "learning_rate": 2.826747720364742e-06,
      "loss": 0.2238,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.055485718215866,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.2434,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1107389672639076,
      "learning_rate": 2.887537993920973e-06,
      "loss": 0.2211,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9608482977210646,
      "learning_rate": 2.9179331306990882e-06,
      "loss": 0.2329,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.744729785875091,
      "learning_rate": 2.948328267477204e-06,
      "loss": 0.6923,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9430150396206145,
      "learning_rate": 2.978723404255319e-06,
      "loss": 0.2297,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.711591425477986,
      "learning_rate": 3.009118541033435e-06,
      "loss": 0.5834,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7355338126831437,
      "learning_rate": 3.03951367781155e-06,
      "loss": 0.2704,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0033511773276205,
      "learning_rate": 3.0699088145896663e-06,
      "loss": 0.2255,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.035158531700837,
      "learning_rate": 3.100303951367781e-06,
      "loss": 0.2597,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6623453254673155,
      "learning_rate": 3.1306990881458972e-06,
      "loss": 0.1863,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.233156261651148,
      "learning_rate": 3.1610942249240125e-06,
      "loss": 0.7773,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.844943667105061,
      "learning_rate": 3.191489361702128e-06,
      "loss": 0.195,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0134239960107734,
      "learning_rate": 3.2218844984802434e-06,
      "loss": 0.1941,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.194746409734577,
      "learning_rate": 3.252279635258359e-06,
      "loss": 0.8839,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.088455544202752,
      "learning_rate": 3.2826747720364744e-06,
      "loss": 0.2381,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.215159070399499,
      "learning_rate": 3.31306990881459e-06,
      "loss": 0.6033,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.593123187021663,
      "learning_rate": 3.3434650455927054e-06,
      "loss": 0.5951,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.913475323808685,
      "learning_rate": 3.373860182370821e-06,
      "loss": 0.5727,
      "step": 111
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0000080003421963,
      "learning_rate": 3.4042553191489363e-06,
      "loss": 0.2239,
      "step": 112
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.047331578037651,
      "learning_rate": 3.434650455927052e-06,
      "loss": 0.227,
      "step": 113
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1047023234279014,
      "learning_rate": 3.4650455927051673e-06,
      "loss": 0.2305,
      "step": 114
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1018967442757894,
      "learning_rate": 3.495440729483283e-06,
      "loss": 0.1929,
      "step": 115
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.824327210347898,
      "learning_rate": 3.5258358662613982e-06,
      "loss": 0.6185,
      "step": 116
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9501754073935431,
      "learning_rate": 3.5562310030395144e-06,
      "loss": 0.2354,
      "step": 117
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.140546907271745,
      "learning_rate": 3.586626139817629e-06,
      "loss": 0.4522,
      "step": 118
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1227017701507473,
      "learning_rate": 3.6170212765957453e-06,
      "loss": 0.1865,
      "step": 119
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1283152966308765,
      "learning_rate": 3.6474164133738606e-06,
      "loss": 0.2298,
      "step": 120
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.080981395669856,
      "learning_rate": 3.6778115501519763e-06,
      "loss": 0.2335,
      "step": 121
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.135327726792392,
      "learning_rate": 3.7082066869300915e-06,
      "loss": 0.2277,
      "step": 122
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0007963846956436,
      "learning_rate": 3.7386018237082072e-06,
      "loss": 0.2283,
      "step": 123
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.325154961725575,
      "learning_rate": 3.7689969604863225e-06,
      "loss": 0.7307,
      "step": 124
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.281634592266141,
      "learning_rate": 3.799392097264438e-06,
      "loss": 0.2269,
      "step": 125
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9347987017996375,
      "learning_rate": 3.8297872340425535e-06,
      "loss": 0.2367,
      "step": 126
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0195269435225187,
      "learning_rate": 3.860182370820669e-06,
      "loss": 0.2269,
      "step": 127
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.060279641876815,
      "learning_rate": 3.890577507598785e-06,
      "loss": 0.2476,
      "step": 128
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.14581650995802,
      "learning_rate": 3.9209726443769005e-06,
      "loss": 0.6263,
      "step": 129
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.989899416635907,
      "learning_rate": 3.951367781155015e-06,
      "loss": 0.2114,
      "step": 130
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.03565705153693,
      "learning_rate": 3.981762917933131e-06,
      "loss": 0.6145,
      "step": 131
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.080101445893299,
      "learning_rate": 4.012158054711247e-06,
      "loss": 0.7646,
      "step": 132
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0172393355338305,
      "learning_rate": 4.042553191489362e-06,
      "loss": 0.2246,
      "step": 133
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.828073220520803,
      "learning_rate": 4.072948328267477e-06,
      "loss": 0.6863,
      "step": 134
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1939042199818304,
      "learning_rate": 4.103343465045593e-06,
      "loss": 0.2271,
      "step": 135
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1047195391053966,
      "learning_rate": 4.133738601823709e-06,
      "loss": 0.2331,
      "step": 136
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4539997849560464,
      "learning_rate": 4.1641337386018235e-06,
      "loss": 0.2101,
      "step": 137
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9076586802883304,
      "learning_rate": 4.194528875379939e-06,
      "loss": 0.1959,
      "step": 138
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8472988328159374,
      "learning_rate": 4.224924012158055e-06,
      "loss": 0.1669,
      "step": 139
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.39952758899086,
      "learning_rate": 4.255319148936171e-06,
      "loss": 0.2559,
      "step": 140
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2248831063730683,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.1911,
      "step": 141
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9344729753226317,
      "learning_rate": 4.316109422492402e-06,
      "loss": 0.2296,
      "step": 142
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0000623146513323,
      "learning_rate": 4.346504559270517e-06,
      "loss": 0.2197,
      "step": 143
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8495682197915468,
      "learning_rate": 4.3768996960486325e-06,
      "loss": 0.2157,
      "step": 144
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.363095597960798,
      "learning_rate": 4.407294832826748e-06,
      "loss": 0.7121,
      "step": 145
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8295358283733718,
      "learning_rate": 4.437689969604864e-06,
      "loss": 0.1779,
      "step": 146
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9781313686694,
      "learning_rate": 4.468085106382979e-06,
      "loss": 0.1901,
      "step": 147
    },
    {
      "epoch": 0.01,
      "grad_norm": 8.002321715411789,
      "learning_rate": 4.4984802431610945e-06,
      "loss": 0.7203,
      "step": 148
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9137361940873845,
      "learning_rate": 4.52887537993921e-06,
      "loss": 0.1935,
      "step": 149
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7442959092802977,
      "learning_rate": 4.559270516717326e-06,
      "loss": 0.2058,
      "step": 150
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.014109162640188,
      "learning_rate": 4.589665653495441e-06,
      "loss": 0.1888,
      "step": 151
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9472908041223023,
      "learning_rate": 4.620060790273556e-06,
      "loss": 0.2111,
      "step": 152
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7953174672607084,
      "learning_rate": 4.650455927051672e-06,
      "loss": 0.1998,
      "step": 153
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9396049736669463,
      "learning_rate": 4.680851063829788e-06,
      "loss": 0.2465,
      "step": 154
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8502360145362193,
      "learning_rate": 4.711246200607903e-06,
      "loss": 0.1963,
      "step": 155
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8934270689562276,
      "learning_rate": 4.741641337386019e-06,
      "loss": 0.2336,
      "step": 156
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8712606096736728,
      "learning_rate": 4.772036474164134e-06,
      "loss": 0.2135,
      "step": 157
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8835293366083228,
      "learning_rate": 4.80243161094225e-06,
      "loss": 0.2246,
      "step": 158
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.342166783703997,
      "learning_rate": 4.832826747720365e-06,
      "loss": 0.6183,
      "step": 159
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.674812384316777,
      "learning_rate": 4.863221884498481e-06,
      "loss": 0.7285,
      "step": 160
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2263743545791628,
      "learning_rate": 4.893617021276596e-06,
      "loss": 0.2034,
      "step": 161
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8475031564031499,
      "learning_rate": 4.924012158054712e-06,
      "loss": 0.2241,
      "step": 162
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1318957293640857,
      "learning_rate": 4.954407294832827e-06,
      "loss": 0.2615,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8449564826545497,
      "learning_rate": 4.984802431610943e-06,
      "loss": 0.2543,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.891969764761733,
      "learning_rate": 5.015197568389059e-06,
      "loss": 0.216,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1542873457947116,
      "learning_rate": 5.0455927051671735e-06,
      "loss": 0.2456,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1401083757023516,
      "learning_rate": 5.075987841945289e-06,
      "loss": 0.199,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1198149318065913,
      "learning_rate": 5.106382978723404e-06,
      "loss": 0.1988,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.976518951576068,
      "learning_rate": 5.136778115501521e-06,
      "loss": 0.2682,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.375479242053848,
      "learning_rate": 5.1671732522796354e-06,
      "loss": 0.7262,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7069038898479223,
      "learning_rate": 5.197568389057751e-06,
      "loss": 0.1994,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7541724852104172,
      "learning_rate": 5.227963525835866e-06,
      "loss": 0.1894,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.920290708868249,
      "learning_rate": 5.2583586626139825e-06,
      "loss": 0.202,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9793260038472975,
      "learning_rate": 5.288753799392098e-06,
      "loss": 0.1909,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9967558005718824,
      "learning_rate": 5.319148936170213e-06,
      "loss": 0.1887,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 7.347996954916956,
      "learning_rate": 5.349544072948329e-06,
      "loss": 0.6964,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0247210470196864,
      "learning_rate": 5.3799392097264444e-06,
      "loss": 0.2383,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0687929949574206,
      "learning_rate": 5.41033434650456e-06,
      "loss": 0.2422,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.026241317146177,
      "learning_rate": 5.440729483282675e-06,
      "loss": 0.2065,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0372880485341405,
      "learning_rate": 5.471124620060791e-06,
      "loss": 0.2532,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0077297637953038,
      "learning_rate": 5.501519756838906e-06,
      "loss": 0.1773,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5798615723070182,
      "learning_rate": 5.531914893617022e-06,
      "loss": 0.2018,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.08597158252966,
      "learning_rate": 5.562310030395137e-06,
      "loss": 0.2133,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1294619228262834,
      "learning_rate": 5.592705167173253e-06,
      "loss": 0.2391,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9525676973124861,
      "learning_rate": 5.623100303951369e-06,
      "loss": 0.2266,
      "step": 185
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0036446383217488,
      "learning_rate": 5.653495440729484e-06,
      "loss": 0.2214,
      "step": 186
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8938990758525718,
      "learning_rate": 5.683890577507599e-06,
      "loss": 0.2133,
      "step": 187
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9999300812872878,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.2272,
      "step": 188
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0588192283501425,
      "learning_rate": 5.744680851063831e-06,
      "loss": 0.2389,
      "step": 189
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0009248320621613,
      "learning_rate": 5.775075987841946e-06,
      "loss": 0.2379,
      "step": 190
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9381715684758138,
      "learning_rate": 5.805471124620062e-06,
      "loss": 0.1911,
      "step": 191
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1067964743820062,
      "learning_rate": 5.8358662613981764e-06,
      "loss": 0.2155,
      "step": 192
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0384012183516673,
      "learning_rate": 5.866261398176293e-06,
      "loss": 0.1479,
      "step": 193
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0290675496027966,
      "learning_rate": 5.896656534954408e-06,
      "loss": 0.2126,
      "step": 194
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9598438966789722,
      "learning_rate": 5.9270516717325235e-06,
      "loss": 0.1977,
      "step": 195
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.450083151733316,
      "learning_rate": 5.957446808510638e-06,
      "loss": 0.2543,
      "step": 196
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3884283666232307,
      "learning_rate": 5.987841945288755e-06,
      "loss": 0.2886,
      "step": 197
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8437262021004734,
      "learning_rate": 6.01823708206687e-06,
      "loss": 0.1928,
      "step": 198
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9288406340202773,
      "learning_rate": 6.0486322188449854e-06,
      "loss": 0.2489,
      "step": 199
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9408398526138568,
      "learning_rate": 6.0790273556231e-06,
      "loss": 0.2211,
      "step": 200
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.863738611440386,
      "learning_rate": 6.109422492401216e-06,
      "loss": 0.2485,
      "step": 201
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8913254800940573,
      "learning_rate": 6.1398176291793325e-06,
      "loss": 0.2365,
      "step": 202
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2505319260324286,
      "learning_rate": 6.170212765957447e-06,
      "loss": 0.2715,
      "step": 203
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8130898456245346,
      "learning_rate": 6.200607902735562e-06,
      "loss": 0.2336,
      "step": 204
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8580721660187416,
      "learning_rate": 6.231003039513678e-06,
      "loss": 0.2553,
      "step": 205
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8103814798761941,
      "learning_rate": 6.2613981762917944e-06,
      "loss": 0.2477,
      "step": 206
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9135869908200516,
      "learning_rate": 6.291793313069909e-06,
      "loss": 0.2639,
      "step": 207
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.891105405720599,
      "learning_rate": 6.322188449848025e-06,
      "loss": 0.7268,
      "step": 208
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7568177315974207,
      "learning_rate": 6.35258358662614e-06,
      "loss": 0.2168,
      "step": 209
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.995651623642621,
      "learning_rate": 6.382978723404256e-06,
      "loss": 0.2,
      "step": 210
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9366752787255455,
      "learning_rate": 6.413373860182371e-06,
      "loss": 0.2607,
      "step": 211
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9339086351107886,
      "learning_rate": 6.443768996960487e-06,
      "loss": 0.214,
      "step": 212
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.523728593189411,
      "learning_rate": 6.474164133738602e-06,
      "loss": 0.759,
      "step": 213
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9598298192689887,
      "learning_rate": 6.504559270516718e-06,
      "loss": 0.2164,
      "step": 214
    },
    {
      "epoch": 0.02,
      "grad_norm": 13.011750571456918,
      "learning_rate": 6.534954407294833e-06,
      "loss": 0.6765,
      "step": 215
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.138562116586146,
      "learning_rate": 6.565349544072949e-06,
      "loss": 0.2403,
      "step": 216
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.739207768951209,
      "learning_rate": 6.595744680851064e-06,
      "loss": 0.2257,
      "step": 217
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0134825509696785,
      "learning_rate": 6.62613981762918e-06,
      "loss": 0.2101,
      "step": 218
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.934780699103113,
      "learning_rate": 6.656534954407296e-06,
      "loss": 0.2243,
      "step": 219
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.833984064252977,
      "learning_rate": 6.686930091185411e-06,
      "loss": 0.8127,
      "step": 220
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8432531196303703,
      "learning_rate": 6.7173252279635256e-06,
      "loss": 0.2471,
      "step": 221
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.761812307642613,
      "learning_rate": 6.747720364741642e-06,
      "loss": 0.1713,
      "step": 222
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9400918654627868,
      "learning_rate": 6.778115501519758e-06,
      "loss": 0.2594,
      "step": 223
    },
    {
      "epoch": 0.02,
      "grad_norm": 4.4801071944372115,
      "learning_rate": 6.808510638297873e-06,
      "loss": 0.6364,
      "step": 224
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8663933636111942,
      "learning_rate": 6.838905775075988e-06,
      "loss": 0.2524,
      "step": 225
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8553152263481993,
      "learning_rate": 6.869300911854104e-06,
      "loss": 0.2122,
      "step": 226
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9964456700331248,
      "learning_rate": 6.89969604863222e-06,
      "loss": 0.2191,
      "step": 227
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.755585713616869,
      "learning_rate": 6.9300911854103346e-06,
      "loss": 0.2062,
      "step": 228
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.999095442820337,
      "learning_rate": 6.96048632218845e-06,
      "loss": 0.238,
      "step": 229
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8727167074761366,
      "learning_rate": 6.990881458966566e-06,
      "loss": 0.239,
      "step": 230
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2338768583662256,
      "learning_rate": 7.021276595744682e-06,
      "loss": 0.2131,
      "step": 231
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.923543095331135,
      "learning_rate": 7.0516717325227965e-06,
      "loss": 0.197,
      "step": 232
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9265167645773862,
      "learning_rate": 7.082066869300912e-06,
      "loss": 0.2035,
      "step": 233
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.737316570646388,
      "learning_rate": 7.112462006079029e-06,
      "loss": 0.6545,
      "step": 234
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.109929062074609,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.1979,
      "step": 235
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.833039101021244,
      "learning_rate": 7.173252279635258e-06,
      "loss": 0.7363,
      "step": 236
    },
    {
      "epoch": 0.02,
      "grad_norm": 5.654951249276348,
      "learning_rate": 7.203647416413374e-06,
      "loss": 0.5292,
      "step": 237
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8464204791633403,
      "learning_rate": 7.234042553191491e-06,
      "loss": 0.2192,
      "step": 238
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7969949519319874,
      "learning_rate": 7.2644376899696055e-06,
      "loss": 0.1755,
      "step": 239
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0236503565203585,
      "learning_rate": 7.294832826747721e-06,
      "loss": 0.2994,
      "step": 240
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9810968382468181,
      "learning_rate": 7.325227963525836e-06,
      "loss": 0.2782,
      "step": 241
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7828379506800158,
      "learning_rate": 7.3556231003039526e-06,
      "loss": 0.1933,
      "step": 242
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.152466522550283,
      "learning_rate": 7.386018237082067e-06,
      "loss": 0.2689,
      "step": 243
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3680377651260507,
      "learning_rate": 7.416413373860183e-06,
      "loss": 0.2558,
      "step": 244
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.148735054787804,
      "learning_rate": 7.446808510638298e-06,
      "loss": 0.3396,
      "step": 245
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0619240385332134,
      "learning_rate": 7.4772036474164145e-06,
      "loss": 0.2303,
      "step": 246
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8437973884123742,
      "learning_rate": 7.507598784194529e-06,
      "loss": 0.2139,
      "step": 247
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9131847966252207,
      "learning_rate": 7.537993920972645e-06,
      "loss": 0.2291,
      "step": 248
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.831667132098316,
      "learning_rate": 7.56838905775076e-06,
      "loss": 0.2455,
      "step": 249
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.062726658762247,
      "learning_rate": 7.598784194528876e-06,
      "loss": 0.2352,
      "step": 250
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.899045616532663,
      "learning_rate": 7.629179331306992e-06,
      "loss": 0.2193,
      "step": 251
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.067146398083773,
      "learning_rate": 7.659574468085107e-06,
      "loss": 0.2562,
      "step": 252
    },
    {
      "epoch": 0.02,
      "grad_norm": 8.659099397329017,
      "learning_rate": 7.689969604863222e-06,
      "loss": 0.643,
      "step": 253
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8558734138265014,
      "learning_rate": 7.720364741641338e-06,
      "loss": 0.2069,
      "step": 254
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9678162328290127,
      "learning_rate": 7.750759878419453e-06,
      "loss": 0.2023,
      "step": 255
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.809530161386004,
      "learning_rate": 7.78115501519757e-06,
      "loss": 0.1798,
      "step": 256
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.739313476364235,
      "learning_rate": 7.811550151975685e-06,
      "loss": 0.2159,
      "step": 257
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.089078944716058,
      "learning_rate": 7.841945288753801e-06,
      "loss": 0.2624,
      "step": 258
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8312308069390018,
      "learning_rate": 7.872340425531916e-06,
      "loss": 0.2029,
      "step": 259
    },
    {
      "epoch": 0.02,
      "grad_norm": 9.847873516103846,
      "learning_rate": 7.90273556231003e-06,
      "loss": 0.8856,
      "step": 260
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2035788994094783,
      "learning_rate": 7.933130699088146e-06,
      "loss": 0.2945,
      "step": 261
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8122894352402987,
      "learning_rate": 7.963525835866262e-06,
      "loss": 0.2291,
      "step": 262
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9465575954534933,
      "learning_rate": 7.993920972644377e-06,
      "loss": 0.2408,
      "step": 263
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8179180168032132,
      "learning_rate": 8.024316109422494e-06,
      "loss": 0.2091,
      "step": 264
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.978975192342584,
      "learning_rate": 8.054711246200608e-06,
      "loss": 0.2346,
      "step": 265
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7888432891490837,
      "learning_rate": 8.085106382978723e-06,
      "loss": 0.2443,
      "step": 266
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7592188750107096,
      "learning_rate": 8.11550151975684e-06,
      "loss": 0.2509,
      "step": 267
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7798788078399526,
      "learning_rate": 8.145896656534955e-06,
      "loss": 0.2221,
      "step": 268
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7459886871002661,
      "learning_rate": 8.176291793313071e-06,
      "loss": 0.2084,
      "step": 269
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8880007358024042,
      "learning_rate": 8.206686930091186e-06,
      "loss": 0.2083,
      "step": 270
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6369805613504722,
      "learning_rate": 8.237082066869303e-06,
      "loss": 0.2022,
      "step": 271
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8179424143515948,
      "learning_rate": 8.267477203647417e-06,
      "loss": 0.277,
      "step": 272
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.915091216326802,
      "learning_rate": 8.297872340425532e-06,
      "loss": 0.2473,
      "step": 273
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.030634065747183,
      "learning_rate": 8.328267477203647e-06,
      "loss": 0.2513,
      "step": 274
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7824205397697697,
      "learning_rate": 8.358662613981764e-06,
      "loss": 0.1789,
      "step": 275
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.15562882688461,
      "learning_rate": 8.389057750759878e-06,
      "loss": 0.2905,
      "step": 276
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8197662912905246,
      "learning_rate": 8.419452887537995e-06,
      "loss": 0.2342,
      "step": 277
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6113596109563757,
      "learning_rate": 8.44984802431611e-06,
      "loss": 0.1944,
      "step": 278
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.977398207103766,
      "learning_rate": 8.480243161094226e-06,
      "loss": 0.2683,
      "step": 279
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5434688239777148,
      "learning_rate": 8.510638297872341e-06,
      "loss": 0.1872,
      "step": 280
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5378771524019677,
      "learning_rate": 8.541033434650456e-06,
      "loss": 0.2018,
      "step": 281
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.206723124607889,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.2581,
      "step": 282
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.669188579407555,
      "learning_rate": 8.601823708206687e-06,
      "loss": 0.1693,
      "step": 283
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.128991047154538,
      "learning_rate": 8.632218844984804e-06,
      "loss": 0.2179,
      "step": 284
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.772876244613542,
      "learning_rate": 8.662613981762919e-06,
      "loss": 0.7442,
      "step": 285
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9156326481357455,
      "learning_rate": 8.693009118541034e-06,
      "loss": 0.2414,
      "step": 286
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.715415364534055,
      "learning_rate": 8.72340425531915e-06,
      "loss": 0.5681,
      "step": 287
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.320377359879025,
      "learning_rate": 8.753799392097265e-06,
      "loss": 0.5705,
      "step": 288
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.966390057178097,
      "learning_rate": 8.78419452887538e-06,
      "loss": 0.2232,
      "step": 289
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8369638011720675,
      "learning_rate": 8.814589665653496e-06,
      "loss": 0.2721,
      "step": 290
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.921058962697042,
      "learning_rate": 8.844984802431611e-06,
      "loss": 0.2097,
      "step": 291
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6977725932659093,
      "learning_rate": 8.875379939209728e-06,
      "loss": 0.1699,
      "step": 292
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0281338711240564,
      "learning_rate": 8.905775075987843e-06,
      "loss": 0.2422,
      "step": 293
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0546026032363733,
      "learning_rate": 8.936170212765958e-06,
      "loss": 0.2899,
      "step": 294
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9252495523268123,
      "learning_rate": 8.966565349544074e-06,
      "loss": 0.2424,
      "step": 295
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9483703658966516,
      "learning_rate": 8.996960486322189e-06,
      "loss": 0.1871,
      "step": 296
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8462358241100691,
      "learning_rate": 9.027355623100304e-06,
      "loss": 0.2863,
      "step": 297
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7326571581018002,
      "learning_rate": 9.05775075987842e-06,
      "loss": 0.2066,
      "step": 298
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.218415999072072,
      "learning_rate": 9.088145896656537e-06,
      "loss": 0.2934,
      "step": 299
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.987465186608381,
      "learning_rate": 9.118541033434652e-06,
      "loss": 0.2799,
      "step": 300
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.134702081490576,
      "learning_rate": 9.148936170212767e-06,
      "loss": 0.2548,
      "step": 301
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.10018895311198,
      "learning_rate": 9.179331306990881e-06,
      "loss": 0.5444,
      "step": 302
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9095481556862732,
      "learning_rate": 9.209726443768998e-06,
      "loss": 0.25,
      "step": 303
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.007195366243451,
      "learning_rate": 9.240121580547113e-06,
      "loss": 0.6484,
      "step": 304
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.747076708056105,
      "learning_rate": 9.27051671732523e-06,
      "loss": 0.2076,
      "step": 305
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9817838159869463,
      "learning_rate": 9.300911854103344e-06,
      "loss": 0.2628,
      "step": 306
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.884595617738149,
      "learning_rate": 9.33130699088146e-06,
      "loss": 0.2673,
      "step": 307
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9398425703852518,
      "learning_rate": 9.361702127659576e-06,
      "loss": 0.249,
      "step": 308
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.875581352058878,
      "learning_rate": 9.39209726443769e-06,
      "loss": 0.2357,
      "step": 309
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8720786570726082,
      "learning_rate": 9.422492401215805e-06,
      "loss": 0.2416,
      "step": 310
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8888056311575987,
      "learning_rate": 9.452887537993922e-06,
      "loss": 0.2412,
      "step": 311
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.772503476609252,
      "learning_rate": 9.483282674772038e-06,
      "loss": 0.2568,
      "step": 312
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.398392783976805,
      "learning_rate": 9.513677811550153e-06,
      "loss": 0.18,
      "step": 313
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.102675408957947,
      "learning_rate": 9.544072948328268e-06,
      "loss": 0.5649,
      "step": 314
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.059398744022831,
      "learning_rate": 9.574468085106385e-06,
      "loss": 0.6208,
      "step": 315
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8818424954511377,
      "learning_rate": 9.6048632218845e-06,
      "loss": 0.2259,
      "step": 316
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.408866791160655,
      "learning_rate": 9.635258358662614e-06,
      "loss": 0.6226,
      "step": 317
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9719096206849047,
      "learning_rate": 9.66565349544073e-06,
      "loss": 0.1982,
      "step": 318
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8843356075203583,
      "learning_rate": 9.696048632218846e-06,
      "loss": 0.2379,
      "step": 319
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9934524564281286,
      "learning_rate": 9.726443768996962e-06,
      "loss": 0.2549,
      "step": 320
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.787293155224456,
      "learning_rate": 9.756838905775077e-06,
      "loss": 0.1939,
      "step": 321
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7372853879332772,
      "learning_rate": 9.787234042553192e-06,
      "loss": 0.2384,
      "step": 322
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.135251179888554,
      "learning_rate": 9.817629179331308e-06,
      "loss": 0.2968,
      "step": 323
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7036991419591918,
      "learning_rate": 9.848024316109423e-06,
      "loss": 0.1986,
      "step": 324
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.726913588980697,
      "learning_rate": 9.878419452887538e-06,
      "loss": 0.2103,
      "step": 325
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5798501314012396,
      "learning_rate": 9.908814589665655e-06,
      "loss": 0.1959,
      "step": 326
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9068110047651665,
      "learning_rate": 9.939209726443771e-06,
      "loss": 0.2249,
      "step": 327
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6265266101778424,
      "learning_rate": 9.969604863221886e-06,
      "loss": 0.2074,
      "step": 328
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.955002118329717,
      "learning_rate": 1e-05,
      "loss": 0.3017,
      "step": 329
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8412147816883415,
      "learning_rate": 9.999999780733276e-06,
      "loss": 0.2151,
      "step": 330
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9251497630918495,
      "learning_rate": 9.999999122933122e-06,
      "loss": 0.2623,
      "step": 331
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.73647558904776,
      "learning_rate": 9.999998026599596e-06,
      "loss": 0.1987,
      "step": 332
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7374063368705253,
      "learning_rate": 9.999996491732795e-06,
      "loss": 0.204,
      "step": 333
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7803093948085562,
      "learning_rate": 9.999994518332852e-06,
      "loss": 0.2234,
      "step": 334
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7210563051905903,
      "learning_rate": 9.999992106399941e-06,
      "loss": 0.2127,
      "step": 335
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7604555803853112,
      "learning_rate": 9.999989255934275e-06,
      "loss": 0.2512,
      "step": 336
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.073619298318284,
      "learning_rate": 9.9999859669361e-06,
      "loss": 0.2646,
      "step": 337
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7537907248381372,
      "learning_rate": 9.99998223940571e-06,
      "loss": 0.2281,
      "step": 338
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8303087314817224,
      "learning_rate": 9.999978073343426e-06,
      "loss": 0.2989,
      "step": 339
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7332450859304807,
      "learning_rate": 9.999973468749619e-06,
      "loss": 0.2278,
      "step": 340
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7401479069865657,
      "learning_rate": 9.99996842562469e-06,
      "loss": 0.1762,
      "step": 341
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8681081031827902,
      "learning_rate": 9.99996294396908e-06,
      "loss": 0.178,
      "step": 342
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.575366180189789,
      "learning_rate": 9.99995702378327e-06,
      "loss": 0.2052,
      "step": 343
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8114543050366805,
      "learning_rate": 9.999950665067784e-06,
      "loss": 0.2739,
      "step": 344
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7781193427413386,
      "learning_rate": 9.999943867823174e-06,
      "loss": 0.257,
      "step": 345
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6819908688148602,
      "learning_rate": 9.999936632050038e-06,
      "loss": 0.1906,
      "step": 346
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.451183856932117,
      "learning_rate": 9.999928957749012e-06,
      "loss": 0.5413,
      "step": 347
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5883681231706397,
      "learning_rate": 9.999920844920768e-06,
      "loss": 0.1939,
      "step": 348
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8005071355117377,
      "learning_rate": 9.999912293566017e-06,
      "loss": 0.2273,
      "step": 349
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.684155861847664,
      "learning_rate": 9.99990330368551e-06,
      "loss": 0.2279,
      "step": 350
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.658795573279144,
      "learning_rate": 9.999893875280036e-06,
      "loss": 0.2318,
      "step": 351
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6647246075721427,
      "learning_rate": 9.99988400835042e-06,
      "loss": 0.2085,
      "step": 352
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7059711230540482,
      "learning_rate": 9.999873702897528e-06,
      "loss": 0.2501,
      "step": 353
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6481915796940003,
      "learning_rate": 9.999862958922266e-06,
      "loss": 0.1973,
      "step": 354
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7970471404994925,
      "learning_rate": 9.999851776425575e-06,
      "loss": 0.2648,
      "step": 355
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.84671522398079,
      "learning_rate": 9.999840155408433e-06,
      "loss": 0.2394,
      "step": 356
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6448462620068238,
      "learning_rate": 9.999828095871863e-06,
      "loss": 0.1909,
      "step": 357
    },
    {
      "epoch": 0.03,
      "grad_norm": 8.907522156236992,
      "learning_rate": 9.999815597816923e-06,
      "loss": 0.6586,
      "step": 358
    },
    {
      "epoch": 0.03,
      "grad_norm": 6.698518308007166,
      "learning_rate": 9.999802661244704e-06,
      "loss": 0.6199,
      "step": 359
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.780827634347472,
      "learning_rate": 9.999789286156346e-06,
      "loss": 0.2466,
      "step": 360
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7442797419960363,
      "learning_rate": 9.999775472553019e-06,
      "loss": 0.2728,
      "step": 361
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7915611219049001,
      "learning_rate": 9.999761220435937e-06,
      "loss": 0.2586,
      "step": 362
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6217905123941667,
      "learning_rate": 9.99974652980635e-06,
      "loss": 0.2704,
      "step": 363
    },
    {
      "epoch": 0.03,
      "grad_norm": 5.222207812425093,
      "learning_rate": 9.999731400665544e-06,
      "loss": 0.4624,
      "step": 364
    },
    {
      "epoch": 0.03,
      "grad_norm": 30.02081012594024,
      "learning_rate": 9.999715833014846e-06,
      "loss": 0.6001,
      "step": 365
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7415905865610677,
      "learning_rate": 9.999699826855624e-06,
      "loss": 0.2532,
      "step": 366
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8339766041450503,
      "learning_rate": 9.99968338218928e-06,
      "loss": 0.2404,
      "step": 367
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8494954181080976,
      "learning_rate": 9.999666499017257e-06,
      "loss": 0.2478,
      "step": 368
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5811408114838028,
      "learning_rate": 9.999649177341036e-06,
      "loss": 0.1933,
      "step": 369
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7182831050546579,
      "learning_rate": 9.999631417162134e-06,
      "loss": 0.2554,
      "step": 370
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8461185772366515,
      "learning_rate": 9.999613218482112e-06,
      "loss": 0.2238,
      "step": 371
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.742875121892276,
      "learning_rate": 9.999594581302562e-06,
      "loss": 0.6717,
      "step": 372
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6693454975480913,
      "learning_rate": 9.999575505625125e-06,
      "loss": 0.2349,
      "step": 373
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6176944829004256,
      "learning_rate": 9.999555991451467e-06,
      "loss": 0.1975,
      "step": 374
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.670713957409525,
      "learning_rate": 9.999536038783304e-06,
      "loss": 0.1697,
      "step": 375
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5945730772336086,
      "learning_rate": 9.999515647622385e-06,
      "loss": 0.2305,
      "step": 376
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7644206496128931,
      "learning_rate": 9.999494817970498e-06,
      "loss": 0.2891,
      "step": 377
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7664113968758461,
      "learning_rate": 9.99947354982947e-06,
      "loss": 0.2301,
      "step": 378
    },
    {
      "epoch": 0.03,
      "grad_norm": 7.071211208519892,
      "learning_rate": 9.999451843201167e-06,
      "loss": 0.5615,
      "step": 379
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6295272627520265,
      "learning_rate": 9.999429698087491e-06,
      "loss": 0.1785,
      "step": 380
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0094822972085256,
      "learning_rate": 9.999407114490384e-06,
      "loss": 0.25,
      "step": 381
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8309162833306567,
      "learning_rate": 9.999384092411832e-06,
      "loss": 0.2473,
      "step": 382
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.498621246477221,
      "learning_rate": 9.999360631853848e-06,
      "loss": 0.645,
      "step": 383
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6615827715417415,
      "learning_rate": 9.999336732818494e-06,
      "loss": 0.1817,
      "step": 384
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.752141942931631,
      "learning_rate": 9.999312395307861e-06,
      "loss": 0.2946,
      "step": 385
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.737721047563245,
      "learning_rate": 9.99928761932409e-06,
      "loss": 0.1899,
      "step": 386
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8384169766446767,
      "learning_rate": 9.99926240486935e-06,
      "loss": 0.2689,
      "step": 387
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7668962430531936,
      "learning_rate": 9.999236751945851e-06,
      "loss": 0.2353,
      "step": 388
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9017776803913906,
      "learning_rate": 9.999210660555847e-06,
      "loss": 0.2476,
      "step": 389
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.076890120781865,
      "learning_rate": 9.999184130701624e-06,
      "loss": 0.2467,
      "step": 390
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5688514526567943,
      "learning_rate": 9.999157162385511e-06,
      "loss": 0.2217,
      "step": 391
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5784313897275015,
      "learning_rate": 9.99912975560987e-06,
      "loss": 0.1952,
      "step": 392
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5019055587871064,
      "learning_rate": 9.999101910377107e-06,
      "loss": 0.1836,
      "step": 393
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6721227300734345,
      "learning_rate": 9.999073626689664e-06,
      "loss": 0.2302,
      "step": 394
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.93738768810481,
      "learning_rate": 9.999044904550022e-06,
      "loss": 0.2098,
      "step": 395
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5923722271089455,
      "learning_rate": 9.999015743960698e-06,
      "loss": 0.2191,
      "step": 396
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8317152166835546,
      "learning_rate": 9.998986144924253e-06,
      "loss": 0.2514,
      "step": 397
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.683127231588279,
      "learning_rate": 9.998956107443278e-06,
      "loss": 0.2528,
      "step": 398
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.588470914824123,
      "learning_rate": 9.998925631520413e-06,
      "loss": 0.2049,
      "step": 399
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8434271582960615,
      "learning_rate": 9.998894717158328e-06,
      "loss": 0.236,
      "step": 400
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.583573167756623,
      "learning_rate": 9.998863364359734e-06,
      "loss": 0.2653,
      "step": 401
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.571257512701912,
      "learning_rate": 9.998831573127383e-06,
      "loss": 0.2429,
      "step": 402
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.665010721931615,
      "learning_rate": 9.998799343464061e-06,
      "loss": 0.1931,
      "step": 403
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.073462753160062,
      "learning_rate": 9.998766675372595e-06,
      "loss": 0.5725,
      "step": 404
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8528956592164112,
      "learning_rate": 9.998733568855853e-06,
      "loss": 0.2567,
      "step": 405
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7496487832121392,
      "learning_rate": 9.998700023916737e-06,
      "loss": 0.2128,
      "step": 406
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6835938893769928,
      "learning_rate": 9.998666040558187e-06,
      "loss": 0.2471,
      "step": 407
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7453850267141413,
      "learning_rate": 9.998631618783185e-06,
      "loss": 0.1725,
      "step": 408
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.933301105416742,
      "learning_rate": 9.998596758594752e-06,
      "loss": 0.5326,
      "step": 409
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7230787014022457,
      "learning_rate": 9.998561459995944e-06,
      "loss": 0.2732,
      "step": 410
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.882429323506284,
      "learning_rate": 9.998525722989857e-06,
      "loss": 0.7092,
      "step": 411
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6358367950966257,
      "learning_rate": 9.998489547579623e-06,
      "loss": 0.2095,
      "step": 412
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8889663314657197,
      "learning_rate": 9.99845293376842e-06,
      "loss": 0.256,
      "step": 413
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.675473135231128,
      "learning_rate": 9.998415881559455e-06,
      "loss": 0.2276,
      "step": 414
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7249561779805345,
      "learning_rate": 9.998378390955979e-06,
      "loss": 0.2294,
      "step": 415
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.754022131939183,
      "learning_rate": 9.99834046196128e-06,
      "loss": 0.2592,
      "step": 416
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6701695853901117,
      "learning_rate": 9.998302094578685e-06,
      "loss": 0.2779,
      "step": 417
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5726225598346228,
      "learning_rate": 9.998263288811558e-06,
      "loss": 0.2735,
      "step": 418
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7571285324616754,
      "learning_rate": 9.998224044663303e-06,
      "loss": 0.2754,
      "step": 419
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.594772596146678,
      "learning_rate": 9.998184362137364e-06,
      "loss": 0.1803,
      "step": 420
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7119637669399907,
      "learning_rate": 9.998144241237218e-06,
      "loss": 0.2389,
      "step": 421
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7156378386268956,
      "learning_rate": 9.998103681966389e-06,
      "loss": 0.231,
      "step": 422
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5385300508721778,
      "learning_rate": 9.998062684328427e-06,
      "loss": 0.2172,
      "step": 423
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8160884355975235,
      "learning_rate": 9.998021248326935e-06,
      "loss": 0.2315,
      "step": 424
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.000300791628822,
      "learning_rate": 9.997979373965542e-06,
      "loss": 0.6024,
      "step": 425
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7718036775237342,
      "learning_rate": 9.997937061247922e-06,
      "loss": 0.2745,
      "step": 426
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6491159079705342,
      "learning_rate": 9.99789431017779e-06,
      "loss": 0.2569,
      "step": 427
    },
    {
      "epoch": 0.04,
      "grad_norm": 5.934831637486679,
      "learning_rate": 9.99785112075889e-06,
      "loss": 0.559,
      "step": 428
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.753118025809732,
      "learning_rate": 9.99780749299501e-06,
      "loss": 0.2358,
      "step": 429
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.038453258798056,
      "learning_rate": 9.997763426889982e-06,
      "loss": 0.2689,
      "step": 430
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6782258460705965,
      "learning_rate": 9.997718922447669e-06,
      "loss": 0.2635,
      "step": 431
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7464554580626437,
      "learning_rate": 9.99767397967197e-06,
      "loss": 0.2322,
      "step": 432
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.913249147756931,
      "learning_rate": 9.99762859856683e-06,
      "loss": 0.2503,
      "step": 433
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.748474397156549,
      "learning_rate": 9.99758277913623e-06,
      "loss": 0.2418,
      "step": 434
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7158750566445138,
      "learning_rate": 9.997536521384187e-06,
      "loss": 0.2616,
      "step": 435
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6903735667844044,
      "learning_rate": 9.997489825314759e-06,
      "loss": 0.2678,
      "step": 436
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5586267989509845,
      "learning_rate": 9.997442690932043e-06,
      "loss": 0.1948,
      "step": 437
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5608968271550325,
      "learning_rate": 9.997395118240168e-06,
      "loss": 0.2181,
      "step": 438
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.676769409206277,
      "learning_rate": 9.997347107243311e-06,
      "loss": 0.2487,
      "step": 439
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5526051930786378,
      "learning_rate": 9.997298657945682e-06,
      "loss": 0.2399,
      "step": 440
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8275123392481565,
      "learning_rate": 9.997249770351531e-06,
      "loss": 0.3011,
      "step": 441
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6000122571826394,
      "learning_rate": 9.997200444465143e-06,
      "loss": 0.2513,
      "step": 442
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6223313214354727,
      "learning_rate": 9.997150680290849e-06,
      "loss": 0.1976,
      "step": 443
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8004977716895076,
      "learning_rate": 9.997100477833006e-06,
      "loss": 0.246,
      "step": 444
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5944003513024463,
      "learning_rate": 9.997049837096024e-06,
      "loss": 0.2217,
      "step": 445
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.238157263567679,
      "learning_rate": 9.996998758084344e-06,
      "loss": 0.6014,
      "step": 446
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6908923449997297,
      "learning_rate": 9.996947240802442e-06,
      "loss": 0.2345,
      "step": 447
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5981253170378944,
      "learning_rate": 9.996895285254839e-06,
      "loss": 0.2121,
      "step": 448
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.329714618297412,
      "learning_rate": 9.996842891446092e-06,
      "loss": 0.7682,
      "step": 449
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7937534046214967,
      "learning_rate": 9.996790059380796e-06,
      "loss": 0.3007,
      "step": 450
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7853104070244654,
      "learning_rate": 9.996736789063583e-06,
      "loss": 0.2845,
      "step": 451
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8655092996663347,
      "learning_rate": 9.996683080499129e-06,
      "loss": 0.2014,
      "step": 452
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7589101142805283,
      "learning_rate": 9.99662893369214e-06,
      "loss": 0.219,
      "step": 453
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.742014427154616,
      "learning_rate": 9.996574348647368e-06,
      "loss": 0.2762,
      "step": 454
    },
    {
      "epoch": 0.04,
      "grad_norm": 10.7847263982688,
      "learning_rate": 9.996519325369599e-06,
      "loss": 0.6268,
      "step": 455
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.626448161282985,
      "learning_rate": 9.996463863863662e-06,
      "loss": 0.2406,
      "step": 456
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7637549042027811,
      "learning_rate": 9.996407964134416e-06,
      "loss": 0.259,
      "step": 457
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7808844530141805,
      "learning_rate": 9.99635162618677e-06,
      "loss": 0.2233,
      "step": 458
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.751715490565843,
      "learning_rate": 9.996294850025658e-06,
      "loss": 0.2394,
      "step": 459
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7646481700762622,
      "learning_rate": 9.996237635656067e-06,
      "loss": 0.2144,
      "step": 460
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9124184289179456,
      "learning_rate": 9.99617998308301e-06,
      "loss": 0.2771,
      "step": 461
    },
    {
      "epoch": 0.04,
      "grad_norm": 8.774346390722034,
      "learning_rate": 9.996121892311546e-06,
      "loss": 0.6522,
      "step": 462
    },
    {
      "epoch": 0.04,
      "grad_norm": 6.50007910588441,
      "learning_rate": 9.996063363346768e-06,
      "loss": 0.7128,
      "step": 463
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6026748368972725,
      "learning_rate": 9.996004396193813e-06,
      "loss": 0.2459,
      "step": 464
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.867069075732658,
      "learning_rate": 9.995944990857848e-06,
      "loss": 0.2903,
      "step": 465
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.071595609076211,
      "learning_rate": 9.995885147344088e-06,
      "loss": 0.2357,
      "step": 466
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8653438751783447,
      "learning_rate": 9.995824865657778e-06,
      "loss": 0.2028,
      "step": 467
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7601660363410474,
      "learning_rate": 9.995764145804206e-06,
      "loss": 0.2535,
      "step": 468
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.796832955977051,
      "learning_rate": 9.995702987788698e-06,
      "loss": 0.2726,
      "step": 469
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6813226942198753,
      "learning_rate": 9.99564139161662e-06,
      "loss": 0.1869,
      "step": 470
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5234230601135563,
      "learning_rate": 9.99557935729337e-06,
      "loss": 0.2146,
      "step": 471
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.843751573833635,
      "learning_rate": 9.995516884824393e-06,
      "loss": 0.2199,
      "step": 472
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.644355527226282,
      "learning_rate": 9.995453974215164e-06,
      "loss": 0.2673,
      "step": 473
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9865762538848437,
      "learning_rate": 9.995390625471204e-06,
      "loss": 0.2495,
      "step": 474
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6631324249119672,
      "learning_rate": 9.995326838598068e-06,
      "loss": 0.2127,
      "step": 475
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.838053822160713,
      "learning_rate": 9.995262613601352e-06,
      "loss": 0.3404,
      "step": 476
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9925826189533975,
      "learning_rate": 9.995197950486687e-06,
      "loss": 0.2537,
      "step": 477
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.75135713959376,
      "learning_rate": 9.995132849259745e-06,
      "loss": 0.1942,
      "step": 478
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6232246277647173,
      "learning_rate": 9.995067309926237e-06,
      "loss": 0.2298,
      "step": 479
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5336939417146547,
      "learning_rate": 9.995001332491907e-06,
      "loss": 0.2495,
      "step": 480
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8566346381531469,
      "learning_rate": 9.994934916962547e-06,
      "loss": 0.236,
      "step": 481
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8842577258433577,
      "learning_rate": 9.99486806334398e-06,
      "loss": 0.2842,
      "step": 482
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.461718151406493,
      "learning_rate": 9.99480077164207e-06,
      "loss": 0.7225,
      "step": 483
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6611262738259445,
      "learning_rate": 9.994733041862716e-06,
      "loss": 0.2441,
      "step": 484
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6711039542663353,
      "learning_rate": 9.994664874011864e-06,
      "loss": 0.2525,
      "step": 485
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7801937175500537,
      "learning_rate": 9.994596268095487e-06,
      "loss": 0.2106,
      "step": 486
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9868449490031554,
      "learning_rate": 9.994527224119606e-06,
      "loss": 0.2531,
      "step": 487
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7776494640259504,
      "learning_rate": 9.994457742090273e-06,
      "loss": 0.2092,
      "step": 488
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7025607470763506,
      "learning_rate": 9.994387822013586e-06,
      "loss": 0.2392,
      "step": 489
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3944948933011174,
      "learning_rate": 9.994317463895676e-06,
      "loss": 0.1886,
      "step": 490
    },
    {
      "epoch": 0.04,
      "grad_norm": 7.653285085639979,
      "learning_rate": 9.994246667742712e-06,
      "loss": 0.6943,
      "step": 491
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.543790539001803,
      "learning_rate": 9.994175433560907e-06,
      "loss": 0.199,
      "step": 492
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.615381952638979,
      "learning_rate": 9.994103761356506e-06,
      "loss": 0.2131,
      "step": 493
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.75273135192563,
      "learning_rate": 9.994031651135796e-06,
      "loss": 0.7442,
      "step": 494
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6928897229669573,
      "learning_rate": 9.9939591029051e-06,
      "loss": 0.2217,
      "step": 495
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5819325046462756,
      "learning_rate": 9.993886116670783e-06,
      "loss": 0.2424,
      "step": 496
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4554272232173031,
      "learning_rate": 9.993812692439247e-06,
      "loss": 0.187,
      "step": 497
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8533182069235625,
      "learning_rate": 9.993738830216929e-06,
      "loss": 0.2455,
      "step": 498
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6802807951138712,
      "learning_rate": 9.993664530010308e-06,
      "loss": 0.2566,
      "step": 499
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7874248526543088,
      "learning_rate": 9.993589791825903e-06,
      "loss": 0.3071,
      "step": 500
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.778286266781573,
      "learning_rate": 9.993514615670267e-06,
      "loss": 0.251,
      "step": 501
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.466128905419672,
      "learning_rate": 9.993439001549994e-06,
      "loss": 0.2105,
      "step": 502
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.601255996984939,
      "learning_rate": 9.993362949471714e-06,
      "loss": 0.6585,
      "step": 503
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.549372705532987,
      "learning_rate": 9.9932864594421e-06,
      "loss": 0.2478,
      "step": 504
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.698140665598578,
      "learning_rate": 9.99320953146786e-06,
      "loss": 0.2481,
      "step": 505
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.935260810659762,
      "learning_rate": 9.99313216555574e-06,
      "loss": 0.26,
      "step": 506
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6541118036443432,
      "learning_rate": 9.993054361712528e-06,
      "loss": 0.2137,
      "step": 507
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8311885867236932,
      "learning_rate": 9.992976119945044e-06,
      "loss": 0.2515,
      "step": 508
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6860071673586863,
      "learning_rate": 9.992897440260154e-06,
      "loss": 0.2163,
      "step": 509
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.537936403398204,
      "learning_rate": 9.992818322664756e-06,
      "loss": 0.6032,
      "step": 510
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8057918088283813,
      "learning_rate": 9.992738767165791e-06,
      "loss": 0.2545,
      "step": 511
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8150611228494031,
      "learning_rate": 9.992658773770237e-06,
      "loss": 0.2531,
      "step": 512
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5347292560079435,
      "learning_rate": 9.992578342485107e-06,
      "loss": 0.2287,
      "step": 513
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.904291856236873,
      "learning_rate": 9.992497473317459e-06,
      "loss": 0.6473,
      "step": 514
    },
    {
      "epoch": 0.05,
      "grad_norm": 8.371086628068749,
      "learning_rate": 9.992416166274382e-06,
      "loss": 0.587,
      "step": 515
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9562005892138374,
      "learning_rate": 9.99233442136301e-06,
      "loss": 0.2776,
      "step": 516
    },
    {
      "epoch": 0.05,
      "grad_norm": 10.919731627334757,
      "learning_rate": 9.992252238590512e-06,
      "loss": 0.7334,
      "step": 517
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7313681845815745,
      "learning_rate": 9.992169617964095e-06,
      "loss": 0.2562,
      "step": 518
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8196677671275618,
      "learning_rate": 9.992086559491007e-06,
      "loss": 0.2214,
      "step": 519
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.780912936733916,
      "learning_rate": 9.992003063178533e-06,
      "loss": 0.229,
      "step": 520
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.306712237559199,
      "learning_rate": 9.991919129033994e-06,
      "loss": 0.5103,
      "step": 521
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4065358579072549,
      "learning_rate": 9.991834757064751e-06,
      "loss": 0.2045,
      "step": 522
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4576457726808076,
      "learning_rate": 9.991749947278206e-06,
      "loss": 0.2766,
      "step": 523
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7170978291570596,
      "learning_rate": 9.9916646996818e-06,
      "loss": 0.2527,
      "step": 524
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7367361647140727,
      "learning_rate": 9.991579014283004e-06,
      "loss": 0.2435,
      "step": 525
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5432732492431542,
      "learning_rate": 9.991492891089335e-06,
      "loss": 0.207,
      "step": 526
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.885397189047643,
      "learning_rate": 9.99140633010835e-06,
      "loss": 0.2579,
      "step": 527
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6126490890408571,
      "learning_rate": 9.991319331347638e-06,
      "loss": 0.2567,
      "step": 528
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6103460135797516,
      "learning_rate": 9.99123189481483e-06,
      "loss": 0.2259,
      "step": 529
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.678562582903193,
      "learning_rate": 9.991144020517595e-06,
      "loss": 0.2402,
      "step": 530
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8662332734602338,
      "learning_rate": 9.99105570846364e-06,
      "loss": 0.2579,
      "step": 531
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6138194923992664,
      "learning_rate": 9.990966958660709e-06,
      "loss": 0.2759,
      "step": 532
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.187269559443866,
      "learning_rate": 9.990877771116588e-06,
      "loss": 0.6065,
      "step": 533
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5043604958448693,
      "learning_rate": 9.990788145839099e-06,
      "loss": 0.2259,
      "step": 534
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.61711180418912,
      "learning_rate": 9.990698082836102e-06,
      "loss": 0.2264,
      "step": 535
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7802177781730852,
      "learning_rate": 9.990607582115497e-06,
      "loss": 0.2494,
      "step": 536
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6567016331649467,
      "learning_rate": 9.990516643685222e-06,
      "loss": 0.2143,
      "step": 537
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7762338128037396,
      "learning_rate": 9.99042526755325e-06,
      "loss": 0.2169,
      "step": 538
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.943492709151052,
      "learning_rate": 9.990333453727598e-06,
      "loss": 0.2825,
      "step": 539
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.79008811834441,
      "learning_rate": 9.99024120221632e-06,
      "loss": 0.2501,
      "step": 540
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5814779734091862,
      "learning_rate": 9.990148513027503e-06,
      "loss": 0.2033,
      "step": 541
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.854421270051113,
      "learning_rate": 9.990055386169278e-06,
      "loss": 0.2399,
      "step": 542
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.77274619984077,
      "learning_rate": 9.989961821649814e-06,
      "loss": 0.2162,
      "step": 543
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.91801935286915,
      "learning_rate": 9.989867819477318e-06,
      "loss": 0.2484,
      "step": 544
    },
    {
      "epoch": 0.05,
      "grad_norm": 7.621868563059297,
      "learning_rate": 9.98977337966003e-06,
      "loss": 0.6408,
      "step": 545
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6936276311287228,
      "learning_rate": 9.98967850220624e-06,
      "loss": 0.2296,
      "step": 546
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7651192856027438,
      "learning_rate": 9.989583187124262e-06,
      "loss": 0.2715,
      "step": 547
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.587160916273804,
      "learning_rate": 9.989487434422463e-06,
      "loss": 0.2584,
      "step": 548
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.54788420845505,
      "learning_rate": 9.989391244109236e-06,
      "loss": 0.1922,
      "step": 549
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6501448230015834,
      "learning_rate": 9.989294616193018e-06,
      "loss": 0.264,
      "step": 550
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6876387070033372,
      "learning_rate": 9.989197550682287e-06,
      "loss": 0.2675,
      "step": 551
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.592073617006624,
      "learning_rate": 9.989100047585553e-06,
      "loss": 0.2361,
      "step": 552
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6234720114197472,
      "learning_rate": 9.989002106911368e-06,
      "loss": 0.1917,
      "step": 553
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5786103404134577,
      "learning_rate": 9.988903728668324e-06,
      "loss": 0.2237,
      "step": 554
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6034632950399903,
      "learning_rate": 9.988804912865048e-06,
      "loss": 0.2468,
      "step": 555
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.703501899106674,
      "learning_rate": 9.988705659510209e-06,
      "loss": 0.749,
      "step": 556
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6116831209195057,
      "learning_rate": 9.98860596861251e-06,
      "loss": 0.2105,
      "step": 557
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8025003761646656,
      "learning_rate": 9.988505840180693e-06,
      "loss": 0.2115,
      "step": 558
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.594132138483214,
      "learning_rate": 9.988405274223543e-06,
      "loss": 0.189,
      "step": 559
    },
    {
      "epoch": 0.05,
      "grad_norm": 4.934828236316857,
      "learning_rate": 9.98830427074988e-06,
      "loss": 0.6544,
      "step": 560
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9320055750973406,
      "learning_rate": 9.988202829768562e-06,
      "loss": 0.255,
      "step": 561
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5376237718389991,
      "learning_rate": 9.988100951288485e-06,
      "loss": 0.2412,
      "step": 562
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6173092564864975,
      "learning_rate": 9.987998635318586e-06,
      "loss": 0.2213,
      "step": 563
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8487720799662384,
      "learning_rate": 9.987895881867837e-06,
      "loss": 0.225,
      "step": 564
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3863881810151115,
      "learning_rate": 9.987792690945252e-06,
      "loss": 0.1853,
      "step": 565
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5696021089016001,
      "learning_rate": 9.987689062559882e-06,
      "loss": 0.2266,
      "step": 566
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7376036183808248,
      "learning_rate": 9.987584996720813e-06,
      "loss": 0.2336,
      "step": 567
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8220868627762037,
      "learning_rate": 9.987480493437176e-06,
      "loss": 0.2604,
      "step": 568
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8497781718174446,
      "learning_rate": 9.987375552718133e-06,
      "loss": 0.2474,
      "step": 569
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6560169095611161,
      "learning_rate": 9.98727017457289e-06,
      "loss": 0.2234,
      "step": 570
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.722524402264342,
      "learning_rate": 9.987164359010691e-06,
      "loss": 0.2553,
      "step": 571
    },
    {
      "epoch": 0.05,
      "grad_norm": 6.232979051963925,
      "learning_rate": 9.987058106040812e-06,
      "loss": 0.6352,
      "step": 572
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8986204034650016,
      "learning_rate": 9.986951415672576e-06,
      "loss": 0.2608,
      "step": 573
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6034788124629207,
      "learning_rate": 9.98684428791534e-06,
      "loss": 0.2538,
      "step": 574
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7013937554366243,
      "learning_rate": 9.986736722778498e-06,
      "loss": 0.2439,
      "step": 575
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4866462935273255,
      "learning_rate": 9.986628720271487e-06,
      "loss": 0.1992,
      "step": 576
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5105100258400834,
      "learning_rate": 9.986520280403775e-06,
      "loss": 0.2335,
      "step": 577
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6211768520636665,
      "learning_rate": 9.986411403184879e-06,
      "loss": 0.2483,
      "step": 578
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8908421388366425,
      "learning_rate": 9.986302088624344e-06,
      "loss": 0.3214,
      "step": 579
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5293528172883106,
      "learning_rate": 9.986192336731757e-06,
      "loss": 0.2497,
      "step": 580
    },
    {
      "epoch": 0.05,
      "grad_norm": 9.400299241724934,
      "learning_rate": 9.986082147516747e-06,
      "loss": 0.7083,
      "step": 581
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0201295140463786,
      "learning_rate": 9.985971520988978e-06,
      "loss": 0.2952,
      "step": 582
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6396656081420475,
      "learning_rate": 9.98586045715815e-06,
      "loss": 0.1918,
      "step": 583
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5332853692806567,
      "learning_rate": 9.985748956034007e-06,
      "loss": 0.2334,
      "step": 584
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5331788429201652,
      "learning_rate": 9.985637017626326e-06,
      "loss": 0.2592,
      "step": 585
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6750575805775831,
      "learning_rate": 9.985524641944926e-06,
      "loss": 0.2609,
      "step": 586
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.861121200311394,
      "learning_rate": 9.985411828999664e-06,
      "loss": 0.2435,
      "step": 587
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4509910377421487,
      "learning_rate": 9.985298578800434e-06,
      "loss": 0.2209,
      "step": 588
    },
    {
      "epoch": 0.05,
      "grad_norm": 5.543910147125815,
      "learning_rate": 9.985184891357165e-06,
      "loss": 0.555,
      "step": 589
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6382829232970355,
      "learning_rate": 9.985070766679834e-06,
      "loss": 0.2242,
      "step": 590
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5665458176733342,
      "learning_rate": 9.984956204778446e-06,
      "loss": 0.2527,
      "step": 591
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4880875736648367,
      "learning_rate": 9.984841205663052e-06,
      "loss": 0.2441,
      "step": 592
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7095555440054468,
      "learning_rate": 9.984725769343737e-06,
      "loss": 0.2774,
      "step": 593
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5385536935887594,
      "learning_rate": 9.984609895830624e-06,
      "loss": 0.2355,
      "step": 594
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5622265264778155,
      "learning_rate": 9.984493585133878e-06,
      "loss": 0.1773,
      "step": 595
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.555157318420605,
      "learning_rate": 9.984376837263701e-06,
      "loss": 0.2269,
      "step": 596
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6434586298212317,
      "learning_rate": 9.984259652230331e-06,
      "loss": 0.2495,
      "step": 597
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6696173722606948,
      "learning_rate": 9.984142030044043e-06,
      "loss": 0.2408,
      "step": 598
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7750098225157604,
      "learning_rate": 9.98402397071516e-06,
      "loss": 0.2529,
      "step": 599
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5613136660846705,
      "learning_rate": 9.983905474254032e-06,
      "loss": 0.2376,
      "step": 600
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7567625503795314,
      "learning_rate": 9.983786540671052e-06,
      "loss": 0.2679,
      "step": 601
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.58851836928129,
      "learning_rate": 9.983667169976651e-06,
      "loss": 0.1983,
      "step": 602
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.7164822974160785,
      "learning_rate": 9.9835473621813e-06,
      "loss": 0.4583,
      "step": 603
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7433480781003048,
      "learning_rate": 9.983427117295507e-06,
      "loss": 0.2215,
      "step": 604
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9081764988812575,
      "learning_rate": 9.983306435329818e-06,
      "loss": 0.2906,
      "step": 605
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6704092784430975,
      "learning_rate": 9.983185316294818e-06,
      "loss": 0.2337,
      "step": 606
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5747298245773038,
      "learning_rate": 9.98306376020113e-06,
      "loss": 0.2598,
      "step": 607
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7799130519785407,
      "learning_rate": 9.982941767059413e-06,
      "loss": 0.2818,
      "step": 608
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3918181707373283,
      "learning_rate": 9.982819336880369e-06,
      "loss": 0.1629,
      "step": 609
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4043613755736868,
      "learning_rate": 9.982696469674737e-06,
      "loss": 0.1995,
      "step": 610
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5215803461875426,
      "learning_rate": 9.982573165453288e-06,
      "loss": 0.2297,
      "step": 611
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.313554097724333,
      "learning_rate": 9.982449424226843e-06,
      "loss": 0.5566,
      "step": 612
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6688461399082763,
      "learning_rate": 9.982325246006253e-06,
      "loss": 0.2588,
      "step": 613
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5963011841483472,
      "learning_rate": 9.982200630802406e-06,
      "loss": 0.2291,
      "step": 614
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.601066746573854,
      "learning_rate": 9.982075578626235e-06,
      "loss": 0.4787,
      "step": 615
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6203431520150609,
      "learning_rate": 9.981950089488708e-06,
      "loss": 0.1923,
      "step": 616
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6317821719429604,
      "learning_rate": 9.981824163400827e-06,
      "loss": 0.2875,
      "step": 617
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7248653464604753,
      "learning_rate": 9.981697800373642e-06,
      "loss": 0.2798,
      "step": 618
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7227338389147528,
      "learning_rate": 9.981571000418233e-06,
      "loss": 0.2974,
      "step": 619
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4887129460735953,
      "learning_rate": 9.981443763545722e-06,
      "loss": 0.2101,
      "step": 620
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.658764651429913,
      "learning_rate": 9.981316089767269e-06,
      "loss": 0.2153,
      "step": 621
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.886490128434196,
      "learning_rate": 9.981187979094069e-06,
      "loss": 0.6781,
      "step": 622
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5573769400304518,
      "learning_rate": 9.98105943153736e-06,
      "loss": 0.2244,
      "step": 623
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5592570662610032,
      "learning_rate": 9.980930447108418e-06,
      "loss": 0.1982,
      "step": 624
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4323594261619055,
      "learning_rate": 9.980801025818556e-06,
      "loss": 0.1629,
      "step": 625
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5650026353994284,
      "learning_rate": 9.980671167679123e-06,
      "loss": 0.2694,
      "step": 626
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3446651387178776,
      "learning_rate": 9.980540872701508e-06,
      "loss": 0.1499,
      "step": 627
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.507544295607225,
      "learning_rate": 9.98041014089714e-06,
      "loss": 0.6716,
      "step": 628
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6586871379088748,
      "learning_rate": 9.980278972277487e-06,
      "loss": 0.2599,
      "step": 629
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.262624813939047,
      "learning_rate": 9.98014736685405e-06,
      "loss": 0.8009,
      "step": 630
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6895852304030567,
      "learning_rate": 9.980015324638371e-06,
      "loss": 0.2016,
      "step": 631
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.521158925988013,
      "learning_rate": 9.979882845642034e-06,
      "loss": 0.2071,
      "step": 632
    },
    {
      "epoch": 0.06,
      "grad_norm": 12.9457319521458,
      "learning_rate": 9.979749929876658e-06,
      "loss": 0.9066,
      "step": 633
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.714310715188928,
      "learning_rate": 9.979616577353902e-06,
      "loss": 0.2337,
      "step": 634
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5916029784688286,
      "learning_rate": 9.979482788085455e-06,
      "loss": 0.2361,
      "step": 635
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.69135248456899,
      "learning_rate": 9.979348562083058e-06,
      "loss": 0.263,
      "step": 636
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7195145652246517,
      "learning_rate": 9.979213899358482e-06,
      "loss": 0.2697,
      "step": 637
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8091452891490918,
      "learning_rate": 9.979078799923538e-06,
      "loss": 0.2397,
      "step": 638
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6390963134692758,
      "learning_rate": 9.978943263790073e-06,
      "loss": 0.1888,
      "step": 639
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6209396237284228,
      "learning_rate": 9.978807290969976e-06,
      "loss": 0.2491,
      "step": 640
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6013780651389133,
      "learning_rate": 9.978670881475173e-06,
      "loss": 0.2647,
      "step": 641
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7195564102714191,
      "learning_rate": 9.978534035317629e-06,
      "loss": 0.2672,
      "step": 642
    },
    {
      "epoch": 0.06,
      "grad_norm": 7.076630306296654,
      "learning_rate": 9.978396752509342e-06,
      "loss": 0.7242,
      "step": 643
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.49527535037465,
      "learning_rate": 9.978259033062357e-06,
      "loss": 0.2197,
      "step": 644
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.747455321731336,
      "learning_rate": 9.978120876988752e-06,
      "loss": 0.494,
      "step": 645
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7168956329285083,
      "learning_rate": 9.977982284300643e-06,
      "loss": 0.2622,
      "step": 646
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5387696292742992,
      "learning_rate": 9.977843255010186e-06,
      "loss": 0.2225,
      "step": 647
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5630024924339858,
      "learning_rate": 9.977703789129576e-06,
      "loss": 0.2236,
      "step": 648
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.859711464667497,
      "learning_rate": 9.977563886671043e-06,
      "loss": 0.2489,
      "step": 649
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4903912788361589,
      "learning_rate": 9.977423547646858e-06,
      "loss": 0.2164,
      "step": 650
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.850660731634943,
      "learning_rate": 9.977282772069331e-06,
      "loss": 0.6373,
      "step": 651
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.564042731113915,
      "learning_rate": 9.977141559950807e-06,
      "loss": 0.6115,
      "step": 652
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.615134664805221,
      "learning_rate": 9.976999911303675e-06,
      "loss": 0.2569,
      "step": 653
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5024473879924438,
      "learning_rate": 9.976857826140354e-06,
      "loss": 0.1906,
      "step": 654
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.826845486530151,
      "learning_rate": 9.976715304473306e-06,
      "loss": 0.2386,
      "step": 655
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7181979741610762,
      "learning_rate": 9.976572346315035e-06,
      "loss": 0.2551,
      "step": 656
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7165257157199212,
      "learning_rate": 9.976428951678077e-06,
      "loss": 0.2364,
      "step": 657
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.623036495345222,
      "learning_rate": 9.976285120575008e-06,
      "loss": 0.2367,
      "step": 658
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4300902783395675,
      "learning_rate": 9.976140853018445e-06,
      "loss": 0.1858,
      "step": 659
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8177804854395565,
      "learning_rate": 9.975996149021039e-06,
      "loss": 0.2211,
      "step": 660
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.516799335096312,
      "learning_rate": 9.975851008595483e-06,
      "loss": 0.6849,
      "step": 661
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7067499964063282,
      "learning_rate": 9.975705431754505e-06,
      "loss": 0.2689,
      "step": 662
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9673499313525566,
      "learning_rate": 9.975559418510876e-06,
      "loss": 0.2378,
      "step": 663
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7231606547043394,
      "learning_rate": 9.9754129688774e-06,
      "loss": 0.2712,
      "step": 664
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4756881766615373,
      "learning_rate": 9.975266082866923e-06,
      "loss": 0.2126,
      "step": 665
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8180194957287772,
      "learning_rate": 9.975118760492328e-06,
      "loss": 0.26,
      "step": 666
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.815700118339477,
      "learning_rate": 9.974971001766534e-06,
      "loss": 0.2753,
      "step": 667
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.755089355313793,
      "learning_rate": 9.974822806702502e-06,
      "loss": 0.2447,
      "step": 668
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.27759724398308,
      "learning_rate": 9.974674175313228e-06,
      "loss": 0.7704,
      "step": 669
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6327968850823973,
      "learning_rate": 9.974525107611752e-06,
      "loss": 0.2367,
      "step": 670
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.594031560200186,
      "learning_rate": 9.974375603611144e-06,
      "loss": 0.2104,
      "step": 671
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5794555196428008,
      "learning_rate": 9.97422566332452e-06,
      "loss": 0.2309,
      "step": 672
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.549057868691088,
      "learning_rate": 9.974075286765027e-06,
      "loss": 0.2687,
      "step": 673
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6323859389645992,
      "learning_rate": 9.973924473945858e-06,
      "loss": 0.2207,
      "step": 674
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.683805834771808,
      "learning_rate": 9.973773224880236e-06,
      "loss": 0.2471,
      "step": 675
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5675812129378532,
      "learning_rate": 9.973621539581429e-06,
      "loss": 0.2349,
      "step": 676
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.65657245000919,
      "learning_rate": 9.973469418062742e-06,
      "loss": 0.2235,
      "step": 677
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5593060826530303,
      "learning_rate": 9.973316860337516e-06,
      "loss": 0.1797,
      "step": 678
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4492103058971144,
      "learning_rate": 9.97316386641913e-06,
      "loss": 0.2096,
      "step": 679
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7724051602319457,
      "learning_rate": 9.973010436321005e-06,
      "loss": 0.2855,
      "step": 680
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7084795032136018,
      "learning_rate": 9.972856570056594e-06,
      "loss": 0.1853,
      "step": 681
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6621259977902831,
      "learning_rate": 9.972702267639396e-06,
      "loss": 0.2335,
      "step": 682
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5316334806320409,
      "learning_rate": 9.972547529082944e-06,
      "loss": 0.2149,
      "step": 683
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6824158002206047,
      "learning_rate": 9.972392354400807e-06,
      "loss": 0.2535,
      "step": 684
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4587555341887288,
      "learning_rate": 9.972236743606597e-06,
      "loss": 0.1713,
      "step": 685
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5936951701388655,
      "learning_rate": 9.972080696713962e-06,
      "loss": 0.2235,
      "step": 686
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.550652136665215,
      "learning_rate": 9.971924213736588e-06,
      "loss": 0.243,
      "step": 687
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6519153945280718,
      "learning_rate": 9.971767294688198e-06,
      "loss": 0.2446,
      "step": 688
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7063231990315464,
      "learning_rate": 9.971609939582556e-06,
      "loss": 0.2273,
      "step": 689
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.527491011521758,
      "learning_rate": 9.971452148433466e-06,
      "loss": 0.2992,
      "step": 690
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5571669251868785,
      "learning_rate": 9.971293921254763e-06,
      "loss": 0.2063,
      "step": 691
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6217816799030609,
      "learning_rate": 9.971135258060326e-06,
      "loss": 0.2534,
      "step": 692
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7318478226096528,
      "learning_rate": 9.970976158864074e-06,
      "loss": 0.2167,
      "step": 693
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6551834880162033,
      "learning_rate": 9.970816623679956e-06,
      "loss": 0.2208,
      "step": 694
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.553780636758177,
      "learning_rate": 9.970656652521965e-06,
      "loss": 0.27,
      "step": 695
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.6379851632111295,
      "learning_rate": 9.970496245404137e-06,
      "loss": 0.6335,
      "step": 696
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.64445597978665,
      "learning_rate": 9.970335402340534e-06,
      "loss": 0.2437,
      "step": 697
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7149742399985657,
      "learning_rate": 9.970174123345267e-06,
      "loss": 0.2361,
      "step": 698
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4933375306773304,
      "learning_rate": 9.97001240843248e-06,
      "loss": 0.2189,
      "step": 699
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7585774564265815,
      "learning_rate": 9.969850257616356e-06,
      "loss": 0.2554,
      "step": 700
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6458600834953958,
      "learning_rate": 9.969687670911116e-06,
      "loss": 0.2548,
      "step": 701
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.489451262985719,
      "learning_rate": 9.969524648331024e-06,
      "loss": 0.596,
      "step": 702
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6156716876896735,
      "learning_rate": 9.969361189890373e-06,
      "loss": 0.1856,
      "step": 703
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7574321170889942,
      "learning_rate": 9.969197295603505e-06,
      "loss": 0.2321,
      "step": 704
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6337224571001592,
      "learning_rate": 9.969032965484789e-06,
      "loss": 0.2637,
      "step": 705
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.651106419428514,
      "learning_rate": 9.968868199548641e-06,
      "loss": 0.2614,
      "step": 706
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5737196379453449,
      "learning_rate": 9.968702997809512e-06,
      "loss": 0.2422,
      "step": 707
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.97518558083543,
      "learning_rate": 9.968537360281889e-06,
      "loss": 0.6764,
      "step": 708
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.480586808937375,
      "learning_rate": 9.968371286980302e-06,
      "loss": 0.743,
      "step": 709
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5913452617053576,
      "learning_rate": 9.968204777919317e-06,
      "loss": 0.2134,
      "step": 710
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6158067033914119,
      "learning_rate": 9.968037833113536e-06,
      "loss": 0.2115,
      "step": 711
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.355290695412414,
      "learning_rate": 9.967870452577601e-06,
      "loss": 0.6912,
      "step": 712
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7692932628721796,
      "learning_rate": 9.967702636326195e-06,
      "loss": 0.2632,
      "step": 713
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8666287319434063,
      "learning_rate": 9.967534384374036e-06,
      "loss": 0.249,
      "step": 714
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.640624640594718,
      "learning_rate": 9.967365696735878e-06,
      "loss": 0.2713,
      "step": 715
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4246695028616634,
      "learning_rate": 9.967196573426519e-06,
      "loss": 0.2102,
      "step": 716
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.700392671167931,
      "learning_rate": 9.96702701446079e-06,
      "loss": 0.267,
      "step": 717
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.541093310737119,
      "learning_rate": 9.966857019853565e-06,
      "loss": 0.2919,
      "step": 718
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6786955025590695,
      "learning_rate": 9.96668658961975e-06,
      "loss": 0.2876,
      "step": 719
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6535870073321912,
      "learning_rate": 9.966515723774296e-06,
      "loss": 0.2917,
      "step": 720
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6294893728897377,
      "learning_rate": 9.96634442233219e-06,
      "loss": 0.3022,
      "step": 721
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4460352391645632,
      "learning_rate": 9.966172685308453e-06,
      "loss": 0.2408,
      "step": 722
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5657898850406984,
      "learning_rate": 9.966000512718151e-06,
      "loss": 0.2548,
      "step": 723
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.299682417959603,
      "learning_rate": 9.965827904576379e-06,
      "loss": 0.6467,
      "step": 724
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6012097662070324,
      "learning_rate": 9.965654860898283e-06,
      "loss": 0.2509,
      "step": 725
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6451623288799975,
      "learning_rate": 9.965481381699036e-06,
      "loss": 0.2182,
      "step": 726
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9751806539689098,
      "learning_rate": 9.965307466993852e-06,
      "loss": 0.2509,
      "step": 727
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5582006982441055,
      "learning_rate": 9.965133116797989e-06,
      "loss": 0.2652,
      "step": 728
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4041719597684783,
      "learning_rate": 9.964958331126735e-06,
      "loss": 0.1715,
      "step": 729
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6926380358796558,
      "learning_rate": 9.964783109995419e-06,
      "loss": 0.2125,
      "step": 730
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5825630388376732,
      "learning_rate": 9.964607453419415e-06,
      "loss": 0.2575,
      "step": 731
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.470100161105013,
      "learning_rate": 9.964431361414122e-06,
      "loss": 0.2088,
      "step": 732
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4166287160223923,
      "learning_rate": 9.964254833994989e-06,
      "loss": 0.229,
      "step": 733
    },
    {
      "epoch": 0.07,
      "grad_norm": 11.654589488631698,
      "learning_rate": 9.964077871177498e-06,
      "loss": 0.7759,
      "step": 734
    },
    {
      "epoch": 0.07,
      "grad_norm": 15.443074125101528,
      "learning_rate": 9.963900472977168e-06,
      "loss": 0.5618,
      "step": 735
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6335368875532728,
      "learning_rate": 9.96372263940956e-06,
      "loss": 0.2849,
      "step": 736
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5182449459528864,
      "learning_rate": 9.96354437049027e-06,
      "loss": 0.2058,
      "step": 737
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5118374330054583,
      "learning_rate": 9.963365666234935e-06,
      "loss": 0.2168,
      "step": 738
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5755130309327054,
      "learning_rate": 9.963186526659226e-06,
      "loss": 0.2138,
      "step": 739
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5424410886006152,
      "learning_rate": 9.963006951778858e-06,
      "loss": 0.2142,
      "step": 740
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7695939041568223,
      "learning_rate": 9.962826941609577e-06,
      "loss": 0.2738,
      "step": 741
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5221201638058013,
      "learning_rate": 9.962646496167173e-06,
      "loss": 0.2428,
      "step": 742
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5709926667641063,
      "learning_rate": 9.962465615467474e-06,
      "loss": 0.2384,
      "step": 743
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5128592384857527,
      "learning_rate": 9.962284299526341e-06,
      "loss": 0.226,
      "step": 744
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.468985652567334,
      "learning_rate": 9.96210254835968e-06,
      "loss": 0.2081,
      "step": 745
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.827487415495247,
      "learning_rate": 9.96192036198343e-06,
      "loss": 0.5985,
      "step": 746
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6003304268327123,
      "learning_rate": 9.961737740413569e-06,
      "loss": 0.2592,
      "step": 747
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.4784702181649,
      "learning_rate": 9.961554683666117e-06,
      "loss": 0.8217,
      "step": 748
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6209766780369546,
      "learning_rate": 9.961371191757126e-06,
      "loss": 0.2469,
      "step": 749
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7008347156420147,
      "learning_rate": 9.961187264702692e-06,
      "loss": 0.2681,
      "step": 750
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3941236557293788,
      "learning_rate": 9.961002902518945e-06,
      "loss": 0.2294,
      "step": 751
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6601682444322061,
      "learning_rate": 9.960818105222057e-06,
      "loss": 0.2374,
      "step": 752
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.576236134953168,
      "learning_rate": 9.960632872828233e-06,
      "loss": 0.2207,
      "step": 753
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.515565782265065,
      "learning_rate": 9.960447205353723e-06,
      "loss": 0.2048,
      "step": 754
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5549660594571182,
      "learning_rate": 9.960261102814806e-06,
      "loss": 0.2111,
      "step": 755
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.477644505699103,
      "learning_rate": 9.960074565227808e-06,
      "loss": 0.2397,
      "step": 756
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8702149068408855,
      "learning_rate": 9.95988759260909e-06,
      "loss": 0.2283,
      "step": 757
    },
    {
      "epoch": 0.07,
      "grad_norm": 7.916628807327948,
      "learning_rate": 9.95970018497505e-06,
      "loss": 0.6216,
      "step": 758
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5660266066634116,
      "learning_rate": 9.959512342342124e-06,
      "loss": 0.2167,
      "step": 759
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6079496811925416,
      "learning_rate": 9.959324064726787e-06,
      "loss": 0.2318,
      "step": 760
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.640562324379634,
      "learning_rate": 9.959135352145552e-06,
      "loss": 0.2257,
      "step": 761
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4276776316307138,
      "learning_rate": 9.958946204614974e-06,
      "loss": 0.1993,
      "step": 762
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.49398095251267,
      "learning_rate": 9.958756622151637e-06,
      "loss": 0.1856,
      "step": 763
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3776042127313801,
      "learning_rate": 9.958566604772173e-06,
      "loss": 0.2021,
      "step": 764
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5201314756315403,
      "learning_rate": 9.958376152493246e-06,
      "loss": 0.1966,
      "step": 765
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.935779705925596,
      "learning_rate": 9.95818526533156e-06,
      "loss": 0.7417,
      "step": 766
    },
    {
      "epoch": 0.07,
      "grad_norm": 4.359775942518877,
      "learning_rate": 9.957993943303857e-06,
      "loss": 0.6812,
      "step": 767
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8128175967675253,
      "learning_rate": 9.957802186426917e-06,
      "loss": 0.2209,
      "step": 768
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5854602805367286,
      "learning_rate": 9.957609994717559e-06,
      "loss": 0.2336,
      "step": 769
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.524150883089539,
      "learning_rate": 9.957417368192639e-06,
      "loss": 0.2105,
      "step": 770
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.515220202649567,
      "learning_rate": 9.957224306869053e-06,
      "loss": 0.1764,
      "step": 771
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5371517653096105,
      "learning_rate": 9.957030810763732e-06,
      "loss": 0.221,
      "step": 772
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5531891982221884,
      "learning_rate": 9.956836879893647e-06,
      "loss": 0.223,
      "step": 773
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.534849138460873,
      "learning_rate": 9.956642514275809e-06,
      "loss": 0.2263,
      "step": 774
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5960290062622526,
      "learning_rate": 9.956447713927261e-06,
      "loss": 0.2489,
      "step": 775
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5149554674296444,
      "learning_rate": 9.956252478865093e-06,
      "loss": 0.2088,
      "step": 776
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.0828388243121605,
      "learning_rate": 9.956056809106426e-06,
      "loss": 0.596,
      "step": 777
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5255385862781479,
      "learning_rate": 9.955860704668425e-06,
      "loss": 0.2057,
      "step": 778
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.74459259825781,
      "learning_rate": 9.955664165568283e-06,
      "loss": 0.2605,
      "step": 779
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6712233941140935,
      "learning_rate": 9.955467191823243e-06,
      "loss": 0.2358,
      "step": 780
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.376064974962242,
      "learning_rate": 9.955269783450578e-06,
      "loss": 0.2018,
      "step": 781
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4735486608861448,
      "learning_rate": 9.955071940467605e-06,
      "loss": 0.1971,
      "step": 782
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8624627697530411,
      "learning_rate": 9.954873662891674e-06,
      "loss": 0.2544,
      "step": 783
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5428312028034077,
      "learning_rate": 9.954674950740175e-06,
      "loss": 0.2834,
      "step": 784
    },
    {
      "epoch": 0.07,
      "grad_norm": 14.015776455703804,
      "learning_rate": 9.954475804030539e-06,
      "loss": 0.5756,
      "step": 785
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5957034306677988,
      "learning_rate": 9.954276222780231e-06,
      "loss": 0.2476,
      "step": 786
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3031214138541232,
      "learning_rate": 9.954076207006752e-06,
      "loss": 0.1762,
      "step": 787
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5795397854648603,
      "learning_rate": 9.95387575672765e-06,
      "loss": 0.2228,
      "step": 788
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4967353029382104,
      "learning_rate": 9.953674871960507e-06,
      "loss": 0.2356,
      "step": 789
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3817879189958813,
      "learning_rate": 9.953473552722934e-06,
      "loss": 0.1957,
      "step": 790
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5190301459605642,
      "learning_rate": 9.953271799032595e-06,
      "loss": 0.2267,
      "step": 791
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5388193577560438,
      "learning_rate": 9.953069610907185e-06,
      "loss": 0.2457,
      "step": 792
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4978440819700856,
      "learning_rate": 9.952866988364431e-06,
      "loss": 0.1956,
      "step": 793
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5960524177738418,
      "learning_rate": 9.952663931422112e-06,
      "loss": 0.2129,
      "step": 794
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3080783067364243,
      "learning_rate": 9.952460440098034e-06,
      "loss": 0.2063,
      "step": 795
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4463929523042047,
      "learning_rate": 9.952256514410042e-06,
      "loss": 0.2284,
      "step": 796
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4498847372311854,
      "learning_rate": 9.952052154376027e-06,
      "loss": 0.2018,
      "step": 797
    },
    {
      "epoch": 0.07,
      "grad_norm": 8.077613901886973,
      "learning_rate": 9.951847360013909e-06,
      "loss": 0.7717,
      "step": 798
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3746423685878217,
      "learning_rate": 9.95164213134165e-06,
      "loss": 0.2088,
      "step": 799
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7934208976774486,
      "learning_rate": 9.95143646837725e-06,
      "loss": 0.2561,
      "step": 800
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.004497970802525,
      "learning_rate": 9.95123037113875e-06,
      "loss": 0.2349,
      "step": 801
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8293142684483812,
      "learning_rate": 9.951023839644224e-06,
      "loss": 0.2686,
      "step": 802
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6243945150316492,
      "learning_rate": 9.950816873911786e-06,
      "loss": 0.2588,
      "step": 803
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6388937471952925,
      "learning_rate": 9.950609473959588e-06,
      "loss": 0.2271,
      "step": 804
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8235331883610157,
      "learning_rate": 9.950401639805822e-06,
      "loss": 0.2745,
      "step": 805
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6168573467459935,
      "learning_rate": 9.950193371468712e-06,
      "loss": 0.2333,
      "step": 806
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.2878554813687595,
      "learning_rate": 9.949984668966531e-06,
      "loss": 0.6802,
      "step": 807
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4839911639318448,
      "learning_rate": 9.949775532317579e-06,
      "loss": 0.227,
      "step": 808
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5467815415297717,
      "learning_rate": 9.9495659615402e-06,
      "loss": 0.2348,
      "step": 809
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7088968636604445,
      "learning_rate": 9.949355956652773e-06,
      "loss": 0.2669,
      "step": 810
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.687743372102919,
      "learning_rate": 9.949145517673722e-06,
      "loss": 0.2271,
      "step": 811
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.661881493212693,
      "learning_rate": 9.948934644621499e-06,
      "loss": 0.2529,
      "step": 812
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4572749776086746,
      "learning_rate": 9.9487233375146e-06,
      "loss": 0.255,
      "step": 813
    },
    {
      "epoch": 0.07,
      "grad_norm": 6.540397214067111,
      "learning_rate": 9.948511596371558e-06,
      "loss": 0.5158,
      "step": 814
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5560347873166567,
      "learning_rate": 9.948299421210946e-06,
      "loss": 0.2324,
      "step": 815
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.737140062754329,
      "learning_rate": 9.948086812051371e-06,
      "loss": 0.2998,
      "step": 816
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4812179273613817,
      "learning_rate": 9.947873768911483e-06,
      "loss": 0.2294,
      "step": 817
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3748402143818328,
      "learning_rate": 9.947660291809962e-06,
      "loss": 0.2051,
      "step": 818
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4633305664868757,
      "learning_rate": 9.947446380765538e-06,
      "loss": 0.1916,
      "step": 819
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5038607576990854,
      "learning_rate": 9.947232035796967e-06,
      "loss": 0.229,
      "step": 820
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5821455912294642,
      "learning_rate": 9.947017256923051e-06,
      "loss": 0.2315,
      "step": 821
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.73789320993524,
      "learning_rate": 9.946802044162629e-06,
      "loss": 0.2178,
      "step": 822
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4408164519777924,
      "learning_rate": 9.946586397534572e-06,
      "loss": 0.2372,
      "step": 823
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5537102167479162,
      "learning_rate": 9.946370317057798e-06,
      "loss": 0.1955,
      "step": 824
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6893750650383037,
      "learning_rate": 9.946153802751257e-06,
      "loss": 0.2203,
      "step": 825
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.256390216575877,
      "learning_rate": 9.945936854633939e-06,
      "loss": 0.1728,
      "step": 826
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4122013439044006,
      "learning_rate": 9.945719472724874e-06,
      "loss": 0.2025,
      "step": 827
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5284087299680198,
      "learning_rate": 9.945501657043121e-06,
      "loss": 0.2288,
      "step": 828
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5155936024694456,
      "learning_rate": 9.945283407607792e-06,
      "loss": 0.2387,
      "step": 829
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.591839242791317,
      "learning_rate": 9.945064724438026e-06,
      "loss": 0.2217,
      "step": 830
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.484763233638824,
      "learning_rate": 9.944845607552999e-06,
      "loss": 0.2461,
      "step": 831
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6638420747805178,
      "learning_rate": 9.944626056971935e-06,
      "loss": 0.2689,
      "step": 832
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3902465033103337,
      "learning_rate": 9.944406072714086e-06,
      "loss": 0.1621,
      "step": 833
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.487856225278641,
      "learning_rate": 9.944185654798748e-06,
      "loss": 0.2194,
      "step": 834
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.350864240985189,
      "learning_rate": 9.943964803245252e-06,
      "loss": 0.5451,
      "step": 835
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7485892135667378,
      "learning_rate": 9.943743518072971e-06,
      "loss": 0.1954,
      "step": 836
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6389631756261165,
      "learning_rate": 9.943521799301309e-06,
      "loss": 0.2434,
      "step": 837
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6780231912018626,
      "learning_rate": 9.943299646949715e-06,
      "loss": 0.2575,
      "step": 838
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6221215009081817,
      "learning_rate": 9.943077061037672e-06,
      "loss": 0.2416,
      "step": 839
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.136015438214619,
      "learning_rate": 9.942854041584703e-06,
      "loss": 0.5345,
      "step": 840
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6236045601807951,
      "learning_rate": 9.942630588610368e-06,
      "loss": 0.2274,
      "step": 841
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.50843021135481,
      "learning_rate": 9.942406702134266e-06,
      "loss": 0.1927,
      "step": 842
    },
    {
      "epoch": 0.08,
      "grad_norm": 6.864613162186772,
      "learning_rate": 9.942182382176031e-06,
      "loss": 0.7915,
      "step": 843
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6020330318181286,
      "learning_rate": 9.941957628755342e-06,
      "loss": 0.2257,
      "step": 844
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4975017114507696,
      "learning_rate": 9.941732441891906e-06,
      "loss": 0.2337,
      "step": 845
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4681216286598928,
      "learning_rate": 9.941506821605477e-06,
      "loss": 0.2438,
      "step": 846
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7781146496371236,
      "learning_rate": 9.941280767915841e-06,
      "loss": 0.2962,
      "step": 847
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4216952674892422,
      "learning_rate": 9.941054280842828e-06,
      "loss": 0.2155,
      "step": 848
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4561829784300024,
      "learning_rate": 9.940827360406297e-06,
      "loss": 0.1922,
      "step": 849
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.499137951439764,
      "learning_rate": 9.940600006626155e-06,
      "loss": 0.2419,
      "step": 850
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6517983996894345,
      "learning_rate": 9.940372219522343e-06,
      "loss": 0.2764,
      "step": 851
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5864012104913954,
      "learning_rate": 9.940143999114836e-06,
      "loss": 0.2273,
      "step": 852
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6856875452317632,
      "learning_rate": 9.939915345423652e-06,
      "loss": 0.2396,
      "step": 853
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.617142623710166,
      "learning_rate": 9.939686258468844e-06,
      "loss": 0.2404,
      "step": 854
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4291842391832585,
      "learning_rate": 9.939456738270507e-06,
      "loss": 0.216,
      "step": 855
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.61234443965128,
      "learning_rate": 9.93922678484877e-06,
      "loss": 0.1987,
      "step": 856
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6007031922990926,
      "learning_rate": 9.938996398223802e-06,
      "loss": 0.2491,
      "step": 857
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.726509901766618,
      "learning_rate": 9.93876557841581e-06,
      "loss": 0.2251,
      "step": 858
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7271895504756645,
      "learning_rate": 9.938534325445035e-06,
      "loss": 0.2272,
      "step": 859
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5199206310019855,
      "learning_rate": 9.938302639331764e-06,
      "loss": 0.216,
      "step": 860
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6541795041631664,
      "learning_rate": 9.938070520096314e-06,
      "loss": 0.2399,
      "step": 861
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.551303061104106,
      "learning_rate": 9.937837967759047e-06,
      "loss": 0.245,
      "step": 862
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5656421514092895,
      "learning_rate": 9.937604982340356e-06,
      "loss": 0.2633,
      "step": 863
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.718633275267488,
      "learning_rate": 9.937371563860678e-06,
      "loss": 0.5578,
      "step": 864
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4441783272474271,
      "learning_rate": 9.937137712340483e-06,
      "loss": 0.2424,
      "step": 865
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3999249236985731,
      "learning_rate": 9.936903427800281e-06,
      "loss": 0.2344,
      "step": 866
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4687253820382173,
      "learning_rate": 9.936668710260625e-06,
      "loss": 0.1988,
      "step": 867
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6134891877215394,
      "learning_rate": 9.936433559742096e-06,
      "loss": 0.2051,
      "step": 868
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.602946562289177,
      "learning_rate": 9.93619797626532e-06,
      "loss": 0.1895,
      "step": 869
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6492169741002785,
      "learning_rate": 9.935961959850959e-06,
      "loss": 0.219,
      "step": 870
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5484785464652173,
      "learning_rate": 9.935725510519716e-06,
      "loss": 0.2234,
      "step": 871
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8303338418591573,
      "learning_rate": 9.935488628292325e-06,
      "loss": 0.2936,
      "step": 872
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9129011771341111,
      "learning_rate": 9.935251313189564e-06,
      "loss": 0.2819,
      "step": 873
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.463776165756342,
      "learning_rate": 9.93501356523225e-06,
      "loss": 0.2016,
      "step": 874
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4302515173880388,
      "learning_rate": 9.93477538444123e-06,
      "loss": 0.2052,
      "step": 875
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.745310110530251,
      "learning_rate": 9.934536770837395e-06,
      "loss": 0.2258,
      "step": 876
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5910304649836102,
      "learning_rate": 9.934297724441677e-06,
      "loss": 0.2467,
      "step": 877
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5265105942172525,
      "learning_rate": 9.934058245275038e-06,
      "loss": 0.2187,
      "step": 878
    },
    {
      "epoch": 0.08,
      "grad_norm": 11.428881718572242,
      "learning_rate": 9.933818333358482e-06,
      "loss": 0.4718,
      "step": 879
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4391312431197443,
      "learning_rate": 9.933577988713055e-06,
      "loss": 0.2625,
      "step": 880
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5938730647799755,
      "learning_rate": 9.933337211359833e-06,
      "loss": 0.2054,
      "step": 881
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6148298737378517,
      "learning_rate": 9.933096001319934e-06,
      "loss": 0.2462,
      "step": 882
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5647856446130184,
      "learning_rate": 9.932854358614514e-06,
      "loss": 0.2079,
      "step": 883
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.66335087604354,
      "learning_rate": 9.932612283264768e-06,
      "loss": 0.2481,
      "step": 884
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.418048176655788,
      "learning_rate": 9.932369775291925e-06,
      "loss": 0.4107,
      "step": 885
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3759397313467059,
      "learning_rate": 9.93212683471726e-06,
      "loss": 0.2053,
      "step": 886
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.494634553090629,
      "learning_rate": 9.931883461562074e-06,
      "loss": 0.2512,
      "step": 887
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6245530921905404,
      "learning_rate": 9.931639655847716e-06,
      "loss": 0.2509,
      "step": 888
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.654768433337211,
      "learning_rate": 9.931395417595568e-06,
      "loss": 0.2335,
      "step": 889
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7391533903363683,
      "learning_rate": 9.931150746827055e-06,
      "loss": 0.2562,
      "step": 890
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5017274722099074,
      "learning_rate": 9.930905643563631e-06,
      "loss": 0.2104,
      "step": 891
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5131005359091114,
      "learning_rate": 9.930660107826795e-06,
      "loss": 0.2546,
      "step": 892
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5238750825850143,
      "learning_rate": 9.930414139638083e-06,
      "loss": 0.2684,
      "step": 893
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7215697242523975,
      "learning_rate": 9.930167739019067e-06,
      "loss": 0.2325,
      "step": 894
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5046482114072648,
      "learning_rate": 9.929920905991361e-06,
      "loss": 0.258,
      "step": 895
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5483638162778057,
      "learning_rate": 9.929673640576608e-06,
      "loss": 0.2768,
      "step": 896
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4549794392083506,
      "learning_rate": 9.929425942796502e-06,
      "loss": 0.2634,
      "step": 897
    },
    {
      "epoch": 0.08,
      "grad_norm": 7.485815777079187,
      "learning_rate": 9.929177812672763e-06,
      "loss": 0.5625,
      "step": 898
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.510540534710968,
      "learning_rate": 9.928929250227155e-06,
      "loss": 0.2527,
      "step": 899
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.677695821464582,
      "learning_rate": 9.928680255481477e-06,
      "loss": 0.2514,
      "step": 900
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5609358832156408,
      "learning_rate": 9.92843082845757e-06,
      "loss": 0.1783,
      "step": 901
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.424345510351746,
      "learning_rate": 9.92818096917731e-06,
      "loss": 0.2498,
      "step": 902
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5587698215654586,
      "learning_rate": 9.92793067766261e-06,
      "loss": 0.2659,
      "step": 903
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4829816958018291,
      "learning_rate": 9.927679953935423e-06,
      "loss": 0.2446,
      "step": 904
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6146633005137265,
      "learning_rate": 9.927428798017738e-06,
      "loss": 0.2241,
      "step": 905
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4632012820165878,
      "learning_rate": 9.927177209931586e-06,
      "loss": 0.2015,
      "step": 906
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.499117677385342,
      "learning_rate": 9.92692518969903e-06,
      "loss": 0.2552,
      "step": 907
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7252028191737003,
      "learning_rate": 9.926672737342175e-06,
      "loss": 0.2256,
      "step": 908
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3678379794740976,
      "learning_rate": 9.926419852883162e-06,
      "loss": 0.1638,
      "step": 909
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.000356332702719,
      "learning_rate": 9.926166536344173e-06,
      "loss": 0.7986,
      "step": 910
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.486470485429745,
      "learning_rate": 9.925912787747423e-06,
      "loss": 0.2652,
      "step": 911
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4883122927419574,
      "learning_rate": 9.925658607115167e-06,
      "loss": 0.2208,
      "step": 912
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5852773031093639,
      "learning_rate": 9.925403994469702e-06,
      "loss": 0.2938,
      "step": 913
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5351554640887692,
      "learning_rate": 9.925148949833356e-06,
      "loss": 0.2052,
      "step": 914
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5510130016370671,
      "learning_rate": 9.924893473228498e-06,
      "loss": 0.1935,
      "step": 915
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7110725612963282,
      "learning_rate": 9.924637564677538e-06,
      "loss": 0.2431,
      "step": 916
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3902716598876024,
      "learning_rate": 9.924381224202917e-06,
      "loss": 0.2521,
      "step": 917
    },
    {
      "epoch": 0.08,
      "grad_norm": 9.083170635908544,
      "learning_rate": 9.924124451827121e-06,
      "loss": 0.6018,
      "step": 918
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.9129228677787158,
      "learning_rate": 9.92386724757267e-06,
      "loss": 0.2684,
      "step": 919
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5817488418210177,
      "learning_rate": 9.92360961146212e-06,
      "loss": 0.236,
      "step": 920
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5925316565059302,
      "learning_rate": 9.92335154351807e-06,
      "loss": 0.2221,
      "step": 921
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5214645776548792,
      "learning_rate": 9.923093043763154e-06,
      "loss": 0.2666,
      "step": 922
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7136978735027983,
      "learning_rate": 9.922834112220043e-06,
      "loss": 0.2512,
      "step": 923
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7513549076130193,
      "learning_rate": 9.922574748911449e-06,
      "loss": 0.2585,
      "step": 924
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.625300530798959,
      "learning_rate": 9.922314953860119e-06,
      "loss": 0.2498,
      "step": 925
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.522027357582532,
      "learning_rate": 9.922054727088837e-06,
      "loss": 0.1951,
      "step": 926
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6929194930164335,
      "learning_rate": 9.92179406862043e-06,
      "loss": 0.1822,
      "step": 927
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4922615886458794,
      "learning_rate": 9.921532978477755e-06,
      "loss": 0.2199,
      "step": 928
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6076367507674796,
      "learning_rate": 9.921271456683716e-06,
      "loss": 0.2286,
      "step": 929
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.703683866895293,
      "learning_rate": 9.921009503261246e-06,
      "loss": 0.5515,
      "step": 930
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5990397557625846,
      "learning_rate": 9.920747118233324e-06,
      "loss": 0.2557,
      "step": 931
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5117803739586604,
      "learning_rate": 9.920484301622961e-06,
      "loss": 0.2282,
      "step": 932
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5539517539994352,
      "learning_rate": 9.920221053453208e-06,
      "loss": 0.2277,
      "step": 933
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.453600785237198,
      "learning_rate": 9.919957373747153e-06,
      "loss": 0.2042,
      "step": 934
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4853405214236632,
      "learning_rate": 9.919693262527923e-06,
      "loss": 0.2171,
      "step": 935
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7375150873439835,
      "learning_rate": 9.919428719818682e-06,
      "loss": 0.2527,
      "step": 936
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6898696190454165,
      "learning_rate": 9.919163745642633e-06,
      "loss": 0.2446,
      "step": 937
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.875308169150922,
      "learning_rate": 9.918898340023015e-06,
      "loss": 0.5974,
      "step": 938
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.707889143681717,
      "learning_rate": 9.918632502983107e-06,
      "loss": 0.2332,
      "step": 939
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.517278781673,
      "learning_rate": 9.918366234546223e-06,
      "loss": 0.2147,
      "step": 940
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5336235641461633,
      "learning_rate": 9.91809953473572e-06,
      "loss": 0.1728,
      "step": 941
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.94552379376375,
      "learning_rate": 9.917832403574985e-06,
      "loss": 0.8124,
      "step": 942
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5270184042006925,
      "learning_rate": 9.917564841087448e-06,
      "loss": 0.2218,
      "step": 943
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6047134086970922,
      "learning_rate": 9.917296847296578e-06,
      "loss": 0.2303,
      "step": 944
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7081274870521845,
      "learning_rate": 9.91702842222588e-06,
      "loss": 0.1929,
      "step": 945
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5222148608784944,
      "learning_rate": 9.916759565898894e-06,
      "loss": 0.2151,
      "step": 946
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5475265255422586,
      "learning_rate": 9.916490278339204e-06,
      "loss": 0.2787,
      "step": 947
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.365731511384655,
      "learning_rate": 9.916220559570426e-06,
      "loss": 0.2424,
      "step": 948
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.647425276424929,
      "learning_rate": 9.915950409616217e-06,
      "loss": 0.2632,
      "step": 949
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.316362853451814,
      "learning_rate": 9.91567982850027e-06,
      "loss": 0.6241,
      "step": 950
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.770223129213491,
      "learning_rate": 9.915408816246317e-06,
      "loss": 0.4973,
      "step": 951
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4373604316633122,
      "learning_rate": 9.915137372878127e-06,
      "loss": 0.203,
      "step": 952
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5230055374168185,
      "learning_rate": 9.91486549841951e-06,
      "loss": 0.2154,
      "step": 953
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6280610963592297,
      "learning_rate": 9.914593192894309e-06,
      "loss": 0.2374,
      "step": 954
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6762585718556333,
      "learning_rate": 9.914320456326407e-06,
      "loss": 0.2504,
      "step": 955
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5069986256290884,
      "learning_rate": 9.914047288739726e-06,
      "loss": 0.2559,
      "step": 956
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.604555795822262,
      "learning_rate": 9.913773690158223e-06,
      "loss": 0.2533,
      "step": 957
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5734229950845229,
      "learning_rate": 9.913499660605896e-06,
      "loss": 0.2476,
      "step": 958
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5893204768956668,
      "learning_rate": 9.91322520010678e-06,
      "loss": 0.2681,
      "step": 959
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5309410952969464,
      "learning_rate": 9.912950308684945e-06,
      "loss": 0.2376,
      "step": 960
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5128712689687744,
      "learning_rate": 9.912674986364502e-06,
      "loss": 0.2903,
      "step": 961
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.522925567988772,
      "learning_rate": 9.912399233169596e-06,
      "loss": 0.2176,
      "step": 962
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5741747766096448,
      "learning_rate": 9.912123049124415e-06,
      "loss": 0.2643,
      "step": 963
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6255569846589473,
      "learning_rate": 9.911846434253183e-06,
      "loss": 0.2588,
      "step": 964
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4661524019270207,
      "learning_rate": 9.911569388580158e-06,
      "loss": 0.2313,
      "step": 965
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6345898800402376,
      "learning_rate": 9.911291912129642e-06,
      "loss": 0.1926,
      "step": 966
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.511811322186476,
      "learning_rate": 9.91101400492597e-06,
      "loss": 0.234,
      "step": 967
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.613812119910068,
      "learning_rate": 9.910735666993515e-06,
      "loss": 0.2639,
      "step": 968
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4803109653796314,
      "learning_rate": 9.91045689835669e-06,
      "loss": 0.2024,
      "step": 969
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6979791421225428,
      "learning_rate": 9.910177699039946e-06,
      "loss": 0.2828,
      "step": 970
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4540091824039094,
      "learning_rate": 9.909898069067772e-06,
      "loss": 0.1968,
      "step": 971
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6212040386909947,
      "learning_rate": 9.909618008464688e-06,
      "loss": 0.2121,
      "step": 972
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5500125651015009,
      "learning_rate": 9.909337517255262e-06,
      "loss": 0.2332,
      "step": 973
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.438891944764058,
      "learning_rate": 9.909056595464093e-06,
      "loss": 0.1889,
      "step": 974
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4930622681696173,
      "learning_rate": 9.908775243115822e-06,
      "loss": 0.2401,
      "step": 975
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3982596693575406,
      "learning_rate": 9.908493460235122e-06,
      "loss": 0.2532,
      "step": 976
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8134208752657173,
      "learning_rate": 9.908211246846708e-06,
      "loss": 0.187,
      "step": 977
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5272325623370606,
      "learning_rate": 9.907928602975335e-06,
      "loss": 0.2008,
      "step": 978
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5516883479771946,
      "learning_rate": 9.907645528645791e-06,
      "loss": 0.2484,
      "step": 979
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.486051174195556,
      "learning_rate": 9.9073620238829e-06,
      "loss": 0.227,
      "step": 980
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4749929231064387,
      "learning_rate": 9.907078088711533e-06,
      "loss": 0.2155,
      "step": 981
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5481202256640652,
      "learning_rate": 9.906793723156588e-06,
      "loss": 0.2124,
      "step": 982
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6767749766696016,
      "learning_rate": 9.906508927243013e-06,
      "loss": 0.2472,
      "step": 983
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6136235817734446,
      "learning_rate": 9.906223700995778e-06,
      "loss": 0.2284,
      "step": 984
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.731633873108982,
      "learning_rate": 9.905938044439904e-06,
      "loss": 0.2411,
      "step": 985
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6155142314892073,
      "learning_rate": 9.905651957600444e-06,
      "loss": 0.2381,
      "step": 986
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5966832194238254,
      "learning_rate": 9.905365440502487e-06,
      "loss": 0.2445,
      "step": 987
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.421775484308334,
      "learning_rate": 9.90507849317117e-06,
      "loss": 0.1843,
      "step": 988
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7687844030399014,
      "learning_rate": 9.904791115631652e-06,
      "loss": 0.2759,
      "step": 989
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6766201081669527,
      "learning_rate": 9.904503307909143e-06,
      "loss": 0.2587,
      "step": 990
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.838926753438958,
      "learning_rate": 9.904215070028883e-06,
      "loss": 0.5475,
      "step": 991
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.512458166181056,
      "learning_rate": 9.903926402016153e-06,
      "loss": 0.1918,
      "step": 992
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5129581783034411,
      "learning_rate": 9.903637303896272e-06,
      "loss": 0.2611,
      "step": 993
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.649456501720178,
      "learning_rate": 9.903347775694597e-06,
      "loss": 0.259,
      "step": 994
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.583054574939023,
      "learning_rate": 9.903057817436517e-06,
      "loss": 0.3134,
      "step": 995
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5218417183422626,
      "learning_rate": 9.902767429147467e-06,
      "loss": 0.1818,
      "step": 996
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.747574042579083,
      "learning_rate": 9.902476610852916e-06,
      "loss": 0.4423,
      "step": 997
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6929576654426977,
      "learning_rate": 9.90218536257837e-06,
      "loss": 0.2328,
      "step": 998
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5227659129417885,
      "learning_rate": 9.90189368434937e-06,
      "loss": 0.2199,
      "step": 999
    },
    {
      "epoch": 0.09,
      "grad_norm": 5.375394425241627,
      "learning_rate": 9.901601576191506e-06,
      "loss": 0.6013,
      "step": 1000
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5138189926462013,
      "learning_rate": 9.901309038130392e-06,
      "loss": 0.2616,
      "step": 1001
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.501606804982349,
      "learning_rate": 9.901016070191685e-06,
      "loss": 0.226,
      "step": 1002
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5062442535392424,
      "learning_rate": 9.900722672401081e-06,
      "loss": 0.2149,
      "step": 1003
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6723565689038722,
      "learning_rate": 9.900428844784318e-06,
      "loss": 0.2653,
      "step": 1004
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.444729909271482,
      "learning_rate": 9.90013458736716e-06,
      "loss": 0.2504,
      "step": 1005
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6890830045789178,
      "learning_rate": 9.899839900175418e-06,
      "loss": 0.2592,
      "step": 1006
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6531792353199823,
      "learning_rate": 9.89954478323494e-06,
      "loss": 0.2357,
      "step": 1007
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6571741454669064,
      "learning_rate": 9.899249236571607e-06,
      "loss": 0.215,
      "step": 1008
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7659345092908567,
      "learning_rate": 9.89895326021134e-06,
      "loss": 0.2477,
      "step": 1009
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.020098084619039,
      "learning_rate": 9.8986568541801e-06,
      "loss": 0.6104,
      "step": 1010
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4571160018033054,
      "learning_rate": 9.898360018503883e-06,
      "loss": 0.1951,
      "step": 1011
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4828686975097922,
      "learning_rate": 9.898062753208723e-06,
      "loss": 0.2301,
      "step": 1012
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6773393266672996,
      "learning_rate": 9.897765058320691e-06,
      "loss": 0.209,
      "step": 1013
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.780953869150563,
      "learning_rate": 9.897466933865903e-06,
      "loss": 0.6353,
      "step": 1014
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7522264517833679,
      "learning_rate": 9.897168379870499e-06,
      "loss": 0.2272,
      "step": 1015
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7296766245217114,
      "learning_rate": 9.896869396360667e-06,
      "loss": 0.2377,
      "step": 1016
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6127916486988447,
      "learning_rate": 9.896569983362632e-06,
      "loss": 0.2071,
      "step": 1017
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5433110531124339,
      "learning_rate": 9.89627014090265e-06,
      "loss": 0.2368,
      "step": 1018
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.75038375050885,
      "learning_rate": 9.895969869007025e-06,
      "loss": 0.2632,
      "step": 1019
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4277617599565162,
      "learning_rate": 9.895669167702087e-06,
      "loss": 0.2363,
      "step": 1020
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5950654658825931,
      "learning_rate": 9.895368037014214e-06,
      "loss": 0.2397,
      "step": 1021
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.567944989009336,
      "learning_rate": 9.895066476969814e-06,
      "loss": 0.2529,
      "step": 1022
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4953008699689085,
      "learning_rate": 9.894764487595338e-06,
      "loss": 0.2142,
      "step": 1023
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5731249833207568,
      "learning_rate": 9.894462068917271e-06,
      "loss": 0.2726,
      "step": 1024
    },
    {
      "epoch": 0.09,
      "grad_norm": 7.252000404229288,
      "learning_rate": 9.894159220962138e-06,
      "loss": 0.5962,
      "step": 1025
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5249961915040506,
      "learning_rate": 9.8938559437565e-06,
      "loss": 0.2398,
      "step": 1026
    },
    {
      "epoch": 0.09,
      "grad_norm": 9.238937346945171,
      "learning_rate": 9.893552237326958e-06,
      "loss": 0.5336,
      "step": 1027
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6821753356565168,
      "learning_rate": 9.893248101700148e-06,
      "loss": 0.2198,
      "step": 1028
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5479195487631967,
      "learning_rate": 9.892943536902744e-06,
      "loss": 0.1945,
      "step": 1029
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6220872854646693,
      "learning_rate": 9.892638542961461e-06,
      "loss": 0.251,
      "step": 1030
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4870312405702693,
      "learning_rate": 9.892333119903045e-06,
      "loss": 0.2534,
      "step": 1031
    },
    {
      "epoch": 0.09,
      "grad_norm": 6.183994400144805,
      "learning_rate": 9.892027267754287e-06,
      "loss": 0.5931,
      "step": 1032
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3897977010157192,
      "learning_rate": 9.891720986542011e-06,
      "loss": 0.2199,
      "step": 1033
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3893974019270645,
      "learning_rate": 9.89141427629308e-06,
      "loss": 0.2412,
      "step": 1034
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5178204187815243,
      "learning_rate": 9.891107137034396e-06,
      "loss": 0.256,
      "step": 1035
    },
    {
      "epoch": 0.09,
      "grad_norm": 16.878214894417823,
      "learning_rate": 9.890799568792894e-06,
      "loss": 0.5454,
      "step": 1036
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.473925853706892,
      "learning_rate": 9.890491571595551e-06,
      "loss": 0.218,
      "step": 1037
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6558832937735748,
      "learning_rate": 9.890183145469383e-06,
      "loss": 0.2312,
      "step": 1038
    },
    {
      "epoch": 0.09,
      "grad_norm": 4.960289356457262,
      "learning_rate": 9.889874290441439e-06,
      "loss": 0.5592,
      "step": 1039
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6168568284655396,
      "learning_rate": 9.889565006538807e-06,
      "loss": 0.2379,
      "step": 1040
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.466738871293739,
      "learning_rate": 9.889255293788613e-06,
      "loss": 0.2103,
      "step": 1041
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6060349052251193,
      "learning_rate": 9.888945152218022e-06,
      "loss": 0.2488,
      "step": 1042
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4481475422846783,
      "learning_rate": 9.888634581854235e-06,
      "loss": 0.196,
      "step": 1043
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7698012580398008,
      "learning_rate": 9.888323582724493e-06,
      "loss": 0.2652,
      "step": 1044
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4812434411237883,
      "learning_rate": 9.88801215485607e-06,
      "loss": 0.2022,
      "step": 1045
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.581060922457455,
      "learning_rate": 9.887700298276281e-06,
      "loss": 0.253,
      "step": 1046
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3804997872976355,
      "learning_rate": 9.88738801301248e-06,
      "loss": 0.23,
      "step": 1047
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.37478449622713,
      "learning_rate": 9.887075299092052e-06,
      "loss": 0.7422,
      "step": 1048
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4731735977430573,
      "learning_rate": 9.886762156542428e-06,
      "loss": 0.2692,
      "step": 1049
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5248284762968984,
      "learning_rate": 9.886448585391074e-06,
      "loss": 0.2338,
      "step": 1050
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.350484985349908,
      "learning_rate": 9.886134585665488e-06,
      "loss": 0.202,
      "step": 1051
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4490916499790742,
      "learning_rate": 9.885820157393212e-06,
      "loss": 0.2188,
      "step": 1052
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4927379858579093,
      "learning_rate": 9.885505300601823e-06,
      "loss": 0.2348,
      "step": 1053
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5490794241616743,
      "learning_rate": 9.885190015318936e-06,
      "loss": 0.2166,
      "step": 1054
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3604141055692147,
      "learning_rate": 9.884874301572204e-06,
      "loss": 0.1798,
      "step": 1055
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7646745673334199,
      "learning_rate": 9.884558159389317e-06,
      "loss": 0.2582,
      "step": 1056
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4449087666477818,
      "learning_rate": 9.884241588798004e-06,
      "loss": 0.2654,
      "step": 1057
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3487499480823315,
      "learning_rate": 9.883924589826029e-06,
      "loss": 0.2097,
      "step": 1058
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3463916388920483,
      "learning_rate": 9.883607162501195e-06,
      "loss": 0.2164,
      "step": 1059
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9549008032225574,
      "learning_rate": 9.883289306851342e-06,
      "loss": 0.2294,
      "step": 1060
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6489985156651354,
      "learning_rate": 9.88297102290435e-06,
      "loss": 0.2311,
      "step": 1061
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6638158396460678,
      "learning_rate": 9.882652310688132e-06,
      "loss": 0.246,
      "step": 1062
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4368664112881635,
      "learning_rate": 9.882333170230645e-06,
      "loss": 0.233,
      "step": 1063
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5166354512593676,
      "learning_rate": 9.882013601559874e-06,
      "loss": 0.2432,
      "step": 1064
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.74691903998192,
      "learning_rate": 9.881693604703853e-06,
      "loss": 0.5666,
      "step": 1065
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6591974625088015,
      "learning_rate": 9.881373179690645e-06,
      "loss": 0.2516,
      "step": 1066
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3592011417203567,
      "learning_rate": 9.881052326548353e-06,
      "loss": 0.172,
      "step": 1067
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.178317254310779,
      "learning_rate": 9.88073104530512e-06,
      "loss": 0.5532,
      "step": 1068
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5771240106086812,
      "learning_rate": 9.880409335989123e-06,
      "loss": 0.2737,
      "step": 1069
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4908595487948417,
      "learning_rate": 9.880087198628579e-06,
      "loss": 0.1925,
      "step": 1070
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6957779072144639,
      "learning_rate": 9.879764633251741e-06,
      "loss": 0.2257,
      "step": 1071
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5566643408525422,
      "learning_rate": 9.8794416398869e-06,
      "loss": 0.2432,
      "step": 1072
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7750042052148647,
      "learning_rate": 9.879118218562384e-06,
      "loss": 0.1996,
      "step": 1073
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.321655505743404,
      "learning_rate": 9.878794369306561e-06,
      "loss": 0.2042,
      "step": 1074
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.340320658671947,
      "learning_rate": 9.878470092147833e-06,
      "loss": 0.7587,
      "step": 1075
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4953608195460182,
      "learning_rate": 9.878145387114644e-06,
      "loss": 0.2181,
      "step": 1076
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5319743017469822,
      "learning_rate": 9.87782025423547e-06,
      "loss": 0.2386,
      "step": 1077
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7140019846729075,
      "learning_rate": 9.877494693538829e-06,
      "loss": 0.2286,
      "step": 1078
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.464506730187972,
      "learning_rate": 9.877168705053273e-06,
      "loss": 0.2314,
      "step": 1079
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6579461319288833,
      "learning_rate": 9.876842288807394e-06,
      "loss": 0.2276,
      "step": 1080
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4204374413373217,
      "learning_rate": 9.876515444829822e-06,
      "loss": 0.2093,
      "step": 1081
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6481967072907655,
      "learning_rate": 9.876188173149224e-06,
      "loss": 0.2341,
      "step": 1082
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.377892118925886,
      "learning_rate": 9.875860473794302e-06,
      "loss": 0.2219,
      "step": 1083
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5670081145594648,
      "learning_rate": 9.875532346793798e-06,
      "loss": 0.2611,
      "step": 1084
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5089747724092055,
      "learning_rate": 9.875203792176488e-06,
      "loss": 0.2008,
      "step": 1085
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5912716525415802,
      "learning_rate": 9.874874809971195e-06,
      "loss": 0.2356,
      "step": 1086
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5782808891671,
      "learning_rate": 9.87454540020677e-06,
      "loss": 0.2556,
      "step": 1087
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4529916066079755,
      "learning_rate": 9.874215562912102e-06,
      "loss": 0.2065,
      "step": 1088
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.68630217091828,
      "learning_rate": 9.873885298116123e-06,
      "loss": 0.2545,
      "step": 1089
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5229380895442877,
      "learning_rate": 9.873554605847797e-06,
      "loss": 0.2318,
      "step": 1090
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7926880265863794,
      "learning_rate": 9.87322348613613e-06,
      "loss": 0.2755,
      "step": 1091
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5665804579605005,
      "learning_rate": 9.872891939010163e-06,
      "loss": 0.2412,
      "step": 1092
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.705503578766063,
      "learning_rate": 9.872559964498974e-06,
      "loss": 0.2717,
      "step": 1093
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4845005130500226,
      "learning_rate": 9.87222756263168e-06,
      "loss": 0.2994,
      "step": 1094
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5099533247743542,
      "learning_rate": 9.871894733437433e-06,
      "loss": 0.2723,
      "step": 1095
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4362323288370136,
      "learning_rate": 9.871561476945429e-06,
      "loss": 0.2187,
      "step": 1096
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5360084048661486,
      "learning_rate": 9.871227793184893e-06,
      "loss": 0.2588,
      "step": 1097
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4531180853263277,
      "learning_rate": 9.87089368218509e-06,
      "loss": 0.1588,
      "step": 1098
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5825642510486797,
      "learning_rate": 9.870559143975328e-06,
      "loss": 0.1807,
      "step": 1099
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3757138050189461,
      "learning_rate": 9.870224178584947e-06,
      "loss": 0.2172,
      "step": 1100
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.432213979906721,
      "learning_rate": 9.869888786043323e-06,
      "loss": 0.2125,
      "step": 1101
    },
    {
      "epoch": 0.1,
      "grad_norm": 5.8063900700443325,
      "learning_rate": 9.869552966379874e-06,
      "loss": 0.6337,
      "step": 1102
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5450561415757171,
      "learning_rate": 9.869216719624055e-06,
      "loss": 0.2502,
      "step": 1103
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7532250784969317,
      "learning_rate": 9.868880045805356e-06,
      "loss": 0.2409,
      "step": 1104
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3920830610429151,
      "learning_rate": 9.868542944953304e-06,
      "loss": 0.1944,
      "step": 1105
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3432112619246859,
      "learning_rate": 9.868205417097467e-06,
      "loss": 0.2199,
      "step": 1106
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4848686400052566,
      "learning_rate": 9.867867462267449e-06,
      "loss": 0.2917,
      "step": 1107
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.732605122010063,
      "learning_rate": 9.867529080492888e-06,
      "loss": 0.2343,
      "step": 1108
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.786720216546288,
      "learning_rate": 9.867190271803466e-06,
      "loss": 0.7718,
      "step": 1109
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.605402121732165,
      "learning_rate": 9.866851036228895e-06,
      "loss": 0.2455,
      "step": 1110
    },
    {
      "epoch": 0.1,
      "grad_norm": 9.405078914462543,
      "learning_rate": 9.86651137379893e-06,
      "loss": 0.5801,
      "step": 1111
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4745339253314913,
      "learning_rate": 9.866171284543362e-06,
      "loss": 0.2441,
      "step": 1112
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5605030676437175,
      "learning_rate": 9.865830768492019e-06,
      "loss": 0.2271,
      "step": 1113
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6128098832023408,
      "learning_rate": 9.865489825674768e-06,
      "loss": 0.3075,
      "step": 1114
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7195187783921626,
      "learning_rate": 9.865148456121509e-06,
      "loss": 0.2729,
      "step": 1115
    },
    {
      "epoch": 0.1,
      "grad_norm": 8.075823386765283,
      "learning_rate": 9.864806659862184e-06,
      "loss": 0.5723,
      "step": 1116
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5675610930343515,
      "learning_rate": 9.864464436926769e-06,
      "loss": 0.2337,
      "step": 1117
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5083950207305599,
      "learning_rate": 9.864121787345282e-06,
      "loss": 0.2378,
      "step": 1118
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4172948407825676,
      "learning_rate": 9.863778711147775e-06,
      "loss": 0.1685,
      "step": 1119
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4921653792060032,
      "learning_rate": 9.863435208364335e-06,
      "loss": 0.2186,
      "step": 1120
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.884665036357697,
      "learning_rate": 9.863091279025095e-06,
      "loss": 0.2662,
      "step": 1121
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.329104483129984,
      "learning_rate": 9.862746923160215e-06,
      "loss": 0.1667,
      "step": 1122
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.435381420292863,
      "learning_rate": 9.862402140799901e-06,
      "loss": 0.5163,
      "step": 1123
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.584884103183068,
      "learning_rate": 9.862056931974392e-06,
      "loss": 0.2099,
      "step": 1124
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.529665271809933,
      "learning_rate": 9.86171129671396e-06,
      "loss": 0.2389,
      "step": 1125
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4516693813982895,
      "learning_rate": 9.861365235048927e-06,
      "loss": 0.2064,
      "step": 1126
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8979255060523421,
      "learning_rate": 9.86101874700964e-06,
      "loss": 0.2984,
      "step": 1127
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5143343977939472,
      "learning_rate": 9.86067183262649e-06,
      "loss": 0.1921,
      "step": 1128
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4224684399164784,
      "learning_rate": 9.860324491929905e-06,
      "loss": 0.1867,
      "step": 1129
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3695560869364598,
      "learning_rate": 9.859976724950346e-06,
      "loss": 0.2262,
      "step": 1130
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4488681360278406,
      "learning_rate": 9.859628531718317e-06,
      "loss": 0.2063,
      "step": 1131
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4373431199180713,
      "learning_rate": 9.859279912264355e-06,
      "loss": 0.2064,
      "step": 1132
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.499948821474026,
      "learning_rate": 9.858930866619038e-06,
      "loss": 0.2835,
      "step": 1133
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3624984108089158,
      "learning_rate": 9.858581394812979e-06,
      "loss": 0.1988,
      "step": 1134
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4805566735292013,
      "learning_rate": 9.85823149687683e-06,
      "loss": 0.1885,
      "step": 1135
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9184499635391843,
      "learning_rate": 9.857881172841275e-06,
      "loss": 0.2691,
      "step": 1136
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4799048403777804,
      "learning_rate": 9.857530422737045e-06,
      "loss": 0.2228,
      "step": 1137
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5358839818228192,
      "learning_rate": 9.857179246594903e-06,
      "loss": 0.2261,
      "step": 1138
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.465903841293275,
      "learning_rate": 9.856827644445645e-06,
      "loss": 0.2454,
      "step": 1139
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4637033112269189,
      "learning_rate": 9.856475616320113e-06,
      "loss": 0.2312,
      "step": 1140
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5182644149672908,
      "learning_rate": 9.85612316224918e-06,
      "loss": 0.2502,
      "step": 1141
    },
    {
      "epoch": 0.1,
      "grad_norm": 6.28815070669023,
      "learning_rate": 9.855770282263761e-06,
      "loss": 0.4131,
      "step": 1142
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.518874419420727,
      "learning_rate": 9.855416976394803e-06,
      "loss": 0.2167,
      "step": 1143
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5713109234330431,
      "learning_rate": 9.855063244673295e-06,
      "loss": 0.1779,
      "step": 1144
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.496336251448894,
      "learning_rate": 9.854709087130261e-06,
      "loss": 0.2257,
      "step": 1145
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.649011023852194,
      "learning_rate": 9.854354503796763e-06,
      "loss": 0.5476,
      "step": 1146
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7397685044555473,
      "learning_rate": 9.8539994947039e-06,
      "loss": 0.3077,
      "step": 1147
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.598230749996207,
      "learning_rate": 9.853644059882812e-06,
      "loss": 0.6452,
      "step": 1148
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.390644504804812,
      "learning_rate": 9.853288199364667e-06,
      "loss": 0.2587,
      "step": 1149
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6081732233797352,
      "learning_rate": 9.85293191318068e-06,
      "loss": 0.236,
      "step": 1150
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6459544984878465,
      "learning_rate": 9.852575201362103e-06,
      "loss": 0.2552,
      "step": 1151
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3364246173283447,
      "learning_rate": 9.852218063940216e-06,
      "loss": 0.2215,
      "step": 1152
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.646369651972716,
      "learning_rate": 9.851860500946342e-06,
      "loss": 0.4738,
      "step": 1153
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5757771393860365,
      "learning_rate": 9.851502512411847e-06,
      "loss": 0.2214,
      "step": 1154
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4965408024953535,
      "learning_rate": 9.851144098368127e-06,
      "loss": 0.2631,
      "step": 1155
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.741094515326322,
      "learning_rate": 9.850785258846614e-06,
      "loss": 0.2982,
      "step": 1156
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5821434410604045,
      "learning_rate": 9.850425993878785e-06,
      "loss": 0.239,
      "step": 1157
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.566404646546354,
      "learning_rate": 9.850066303496148e-06,
      "loss": 0.2207,
      "step": 1158
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.207794337811969,
      "learning_rate": 9.84970618773025e-06,
      "loss": 0.1813,
      "step": 1159
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6336473550337058,
      "learning_rate": 9.849345646612676e-06,
      "loss": 0.243,
      "step": 1160
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4606957863047219,
      "learning_rate": 9.848984680175049e-06,
      "loss": 0.246,
      "step": 1161
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.405354776691957,
      "learning_rate": 9.848623288449026e-06,
      "loss": 0.2202,
      "step": 1162
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5175453074219265,
      "learning_rate": 9.848261471466304e-06,
      "loss": 0.2595,
      "step": 1163
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4965227982373852,
      "learning_rate": 9.847899229258617e-06,
      "loss": 0.2298,
      "step": 1164
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6345124718096897,
      "learning_rate": 9.847536561857739e-06,
      "loss": 0.2389,
      "step": 1165
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4446150533328017,
      "learning_rate": 9.847173469295473e-06,
      "loss": 0.242,
      "step": 1166
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6356076579243866,
      "learning_rate": 9.846809951603667e-06,
      "loss": 0.2066,
      "step": 1167
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3676527054340093,
      "learning_rate": 9.846446008814205e-06,
      "loss": 0.1915,
      "step": 1168
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5227043047928026,
      "learning_rate": 9.846081640959008e-06,
      "loss": 0.2324,
      "step": 1169
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5964407919089114,
      "learning_rate": 9.84571684807003e-06,
      "loss": 0.3078,
      "step": 1170
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5889166325206243,
      "learning_rate": 9.84535163017927e-06,
      "loss": 0.2768,
      "step": 1171
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.919978562797967,
      "learning_rate": 9.844985987318756e-06,
      "loss": 0.852,
      "step": 1172
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5446373801708948,
      "learning_rate": 9.844619919520559e-06,
      "loss": 0.1906,
      "step": 1173
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.618805082797594,
      "learning_rate": 9.844253426816785e-06,
      "loss": 0.1973,
      "step": 1174
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4003568716270092,
      "learning_rate": 9.843886509239582e-06,
      "loss": 0.197,
      "step": 1175
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4417740622598065,
      "learning_rate": 9.843519166821125e-06,
      "loss": 0.2514,
      "step": 1176
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4504486849330316,
      "learning_rate": 9.843151399593636e-06,
      "loss": 0.299,
      "step": 1177
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.553168220250697,
      "learning_rate": 9.842783207589368e-06,
      "loss": 0.6849,
      "step": 1178
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.830308394008952,
      "learning_rate": 9.842414590840618e-06,
      "loss": 0.2617,
      "step": 1179
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.753190577791111,
      "learning_rate": 9.842045549379713e-06,
      "loss": 0.773,
      "step": 1180
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5257706498401498,
      "learning_rate": 9.84167608323902e-06,
      "loss": 0.2806,
      "step": 1181
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4173103349942027,
      "learning_rate": 9.841306192450946e-06,
      "loss": 0.2066,
      "step": 1182
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4692126279514244,
      "learning_rate": 9.840935877047932e-06,
      "loss": 0.2414,
      "step": 1183
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4258674164613558,
      "learning_rate": 9.840565137062456e-06,
      "loss": 0.1943,
      "step": 1184
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7362718372010586,
      "learning_rate": 9.840193972527037e-06,
      "loss": 0.2441,
      "step": 1185
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4575013542563775,
      "learning_rate": 9.839822383474225e-06,
      "loss": 0.2338,
      "step": 1186
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.43145693964649,
      "learning_rate": 9.839450369936615e-06,
      "loss": 0.2452,
      "step": 1187
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5803765962205156,
      "learning_rate": 9.83907793194683e-06,
      "loss": 0.2362,
      "step": 1188
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3548753258776773,
      "learning_rate": 9.838705069537539e-06,
      "loss": 0.203,
      "step": 1189
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3551174823894951,
      "learning_rate": 9.838331782741444e-06,
      "loss": 0.2239,
      "step": 1190
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5755589545558244,
      "learning_rate": 9.837958071591285e-06,
      "loss": 0.221,
      "step": 1191
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4528128211709341,
      "learning_rate": 9.837583936119838e-06,
      "loss": 0.2282,
      "step": 1192
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7000249234276406,
      "learning_rate": 9.837209376359918e-06,
      "loss": 0.2517,
      "step": 1193
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4969654804703267,
      "learning_rate": 9.836834392344374e-06,
      "loss": 0.2107,
      "step": 1194
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.579632088208198,
      "learning_rate": 9.836458984106099e-06,
      "loss": 0.2925,
      "step": 1195
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6987476820203402,
      "learning_rate": 9.836083151678014e-06,
      "loss": 0.2759,
      "step": 1196
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4731161919146947,
      "learning_rate": 9.835706895093086e-06,
      "loss": 0.2249,
      "step": 1197
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.031416674339686,
      "learning_rate": 9.835330214384313e-06,
      "loss": 0.6617,
      "step": 1198
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3884831001111357,
      "learning_rate": 9.834953109584733e-06,
      "loss": 0.216,
      "step": 1199
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5438359212080641,
      "learning_rate": 9.83457558072742e-06,
      "loss": 0.2465,
      "step": 1200
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4302809119877573,
      "learning_rate": 9.834197627845488e-06,
      "loss": 0.2124,
      "step": 1201
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3782529005510942,
      "learning_rate": 9.833819250972084e-06,
      "loss": 0.2303,
      "step": 1202
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.46496822719778,
      "learning_rate": 9.833440450140394e-06,
      "loss": 0.214,
      "step": 1203
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6702639790776213,
      "learning_rate": 9.833061225383642e-06,
      "loss": 0.2647,
      "step": 1204
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4644127808756968,
      "learning_rate": 9.832681576735087e-06,
      "loss": 0.2376,
      "step": 1205
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.59752056902952,
      "learning_rate": 9.83230150422803e-06,
      "loss": 0.2685,
      "step": 1206
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5782842803881239,
      "learning_rate": 9.831921007895806e-06,
      "loss": 0.2493,
      "step": 1207
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6288208478843755,
      "learning_rate": 9.831540087771782e-06,
      "loss": 0.2669,
      "step": 1208
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.40977208848425,
      "learning_rate": 9.831158743889373e-06,
      "loss": 0.2157,
      "step": 1209
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4918511659586309,
      "learning_rate": 9.830776976282022e-06,
      "loss": 0.2252,
      "step": 1210
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5922784247862725,
      "learning_rate": 9.830394784983213e-06,
      "loss": 0.2292,
      "step": 1211
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4866033829829008,
      "learning_rate": 9.830012170026465e-06,
      "loss": 0.247,
      "step": 1212
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.696015960214825,
      "learning_rate": 9.829629131445342e-06,
      "loss": 0.2524,
      "step": 1213
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5188438700036275,
      "learning_rate": 9.829245669273434e-06,
      "loss": 0.1921,
      "step": 1214
    },
    {
      "epoch": 0.11,
      "grad_norm": 4.981499552215183,
      "learning_rate": 9.828861783544374e-06,
      "loss": 0.4717,
      "step": 1215
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5562365742227142,
      "learning_rate": 9.828477474291831e-06,
      "loss": 0.2235,
      "step": 1216
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6701644181852595,
      "learning_rate": 9.828092741549513e-06,
      "loss": 0.2203,
      "step": 1217
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5408085249861543,
      "learning_rate": 9.827707585351163e-06,
      "loss": 0.2383,
      "step": 1218
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6410499907534268,
      "learning_rate": 9.82732200573056e-06,
      "loss": 0.1957,
      "step": 1219
    },
    {
      "epoch": 0.11,
      "grad_norm": 7.884218138442543,
      "learning_rate": 9.826936002721526e-06,
      "loss": 0.5195,
      "step": 1220
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6670229708663609,
      "learning_rate": 9.826549576357911e-06,
      "loss": 0.244,
      "step": 1221
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6798993315124289,
      "learning_rate": 9.826162726673611e-06,
      "loss": 0.3034,
      "step": 1222
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4784801054942627,
      "learning_rate": 9.825775453702553e-06,
      "loss": 0.2083,
      "step": 1223
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6450189095497987,
      "learning_rate": 9.825387757478705e-06,
      "loss": 0.281,
      "step": 1224
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5926953703040256,
      "learning_rate": 9.82499963803607e-06,
      "loss": 0.2252,
      "step": 1225
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4778440016747625,
      "learning_rate": 9.824611095408691e-06,
      "loss": 0.196,
      "step": 1226
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4862223397823315,
      "learning_rate": 9.82422212963064e-06,
      "loss": 0.2228,
      "step": 1227
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3700091505588172,
      "learning_rate": 9.823832740736037e-06,
      "loss": 0.2482,
      "step": 1228
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2799246948993428,
      "learning_rate": 9.823442928759033e-06,
      "loss": 0.2007,
      "step": 1229
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5723042157765523,
      "learning_rate": 9.823052693733814e-06,
      "loss": 0.1851,
      "step": 1230
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4934483100672433,
      "learning_rate": 9.822662035694612e-06,
      "loss": 0.247,
      "step": 1231
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5181373328763055,
      "learning_rate": 9.822270954675685e-06,
      "loss": 0.2136,
      "step": 1232
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6774851031032423,
      "learning_rate": 9.821879450711336e-06,
      "loss": 0.2339,
      "step": 1233
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5981714320450104,
      "learning_rate": 9.8214875238359e-06,
      "loss": 0.2519,
      "step": 1234
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5560495870710036,
      "learning_rate": 9.821095174083758e-06,
      "loss": 0.2451,
      "step": 1235
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4675982047736842,
      "learning_rate": 9.820702401489312e-06,
      "loss": 0.1986,
      "step": 1236
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.539563213079687,
      "learning_rate": 9.82030920608702e-06,
      "loss": 0.2314,
      "step": 1237
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3810878344015851,
      "learning_rate": 9.819915587911362e-06,
      "loss": 0.2266,
      "step": 1238
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8804378947534781,
      "learning_rate": 9.819521546996864e-06,
      "loss": 0.2413,
      "step": 1239
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.569915854359324,
      "learning_rate": 9.819127083378084e-06,
      "loss": 0.2126,
      "step": 1240
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5327523482160017,
      "learning_rate": 9.81873219708962e-06,
      "loss": 0.2484,
      "step": 1241
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6346444250270518,
      "learning_rate": 9.818336888166106e-06,
      "loss": 0.2382,
      "step": 1242
    },
    {
      "epoch": 0.11,
      "grad_norm": 6.462502267132339,
      "learning_rate": 9.817941156642215e-06,
      "loss": 0.7101,
      "step": 1243
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5114944376087847,
      "learning_rate": 9.817545002552653e-06,
      "loss": 0.2178,
      "step": 1244
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5332112089607013,
      "learning_rate": 9.817148425932165e-06,
      "loss": 0.232,
      "step": 1245
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8050430842000356,
      "learning_rate": 9.816751426815535e-06,
      "loss": 0.2501,
      "step": 1246
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4962027291454203,
      "learning_rate": 9.816354005237583e-06,
      "loss": 0.2628,
      "step": 1247
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3866523537152724,
      "learning_rate": 9.815956161233162e-06,
      "loss": 0.1901,
      "step": 1248
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.640626789415663,
      "learning_rate": 9.815557894837171e-06,
      "loss": 0.2449,
      "step": 1249
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6408390155772716,
      "learning_rate": 9.815159206084537e-06,
      "loss": 0.3035,
      "step": 1250
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4671452014398,
      "learning_rate": 9.814760095010227e-06,
      "loss": 0.2318,
      "step": 1251
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6232166268892725,
      "learning_rate": 9.814360561649247e-06,
      "loss": 0.2722,
      "step": 1252
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5020260040516298,
      "learning_rate": 9.813960606036641e-06,
      "loss": 0.1931,
      "step": 1253
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5668721297943802,
      "learning_rate": 9.813560228207484e-06,
      "loss": 0.2396,
      "step": 1254
    },
    {
      "epoch": 0.11,
      "grad_norm": 5.489481836529407,
      "learning_rate": 9.813159428196894e-06,
      "loss": 0.5785,
      "step": 1255
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5326573822623402,
      "learning_rate": 9.812758206040023e-06,
      "loss": 0.2144,
      "step": 1256
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6058434557588617,
      "learning_rate": 9.81235656177206e-06,
      "loss": 0.2432,
      "step": 1257
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4709612118015507,
      "learning_rate": 9.811954495428235e-06,
      "loss": 0.2306,
      "step": 1258
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5831037198002198,
      "learning_rate": 9.81155200704381e-06,
      "loss": 0.255,
      "step": 1259
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6908838534354629,
      "learning_rate": 9.811149096654085e-06,
      "loss": 0.2889,
      "step": 1260
    },
    {
      "epoch": 0.12,
      "grad_norm": 12.28010292629828,
      "learning_rate": 9.810745764294399e-06,
      "loss": 0.5485,
      "step": 1261
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4677947413689365,
      "learning_rate": 9.810342010000127e-06,
      "loss": 0.2431,
      "step": 1262
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2993951996357807,
      "learning_rate": 9.80993783380668e-06,
      "loss": 0.1807,
      "step": 1263
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.390660736009998,
      "learning_rate": 9.809533235749508e-06,
      "loss": 0.2246,
      "step": 1264
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5809087012148157,
      "learning_rate": 9.809128215864096e-06,
      "loss": 0.2307,
      "step": 1265
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4089767419045036,
      "learning_rate": 9.808722774185967e-06,
      "loss": 0.178,
      "step": 1266
    },
    {
      "epoch": 0.12,
      "grad_norm": 17.092756241068958,
      "learning_rate": 9.808316910750683e-06,
      "loss": 0.5925,
      "step": 1267
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5843425054794924,
      "learning_rate": 9.80791062559384e-06,
      "loss": 0.2386,
      "step": 1268
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4481548351008486,
      "learning_rate": 9.807503918751071e-06,
      "loss": 0.1806,
      "step": 1269
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4720989210709503,
      "learning_rate": 9.807096790258045e-06,
      "loss": 0.223,
      "step": 1270
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2688304380638662,
      "learning_rate": 9.806689240150475e-06,
      "loss": 0.1889,
      "step": 1271
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6018531866786285,
      "learning_rate": 9.806281268464102e-06,
      "loss": 0.2812,
      "step": 1272
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3119975953845067,
      "learning_rate": 9.80587287523471e-06,
      "loss": 0.2024,
      "step": 1273
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.6875107574391714,
      "learning_rate": 9.805464060498115e-06,
      "loss": 0.7555,
      "step": 1274
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5023865130015932,
      "learning_rate": 9.805054824290175e-06,
      "loss": 0.1892,
      "step": 1275
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.387978724262368,
      "learning_rate": 9.804645166646782e-06,
      "loss": 0.2003,
      "step": 1276
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.576539887206123,
      "learning_rate": 9.804235087603867e-06,
      "loss": 0.2261,
      "step": 1277
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4953246684838217,
      "learning_rate": 9.803824587197395e-06,
      "loss": 0.1912,
      "step": 1278
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.107466851988401,
      "learning_rate": 9.80341366546337e-06,
      "loss": 0.7182,
      "step": 1279
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.7511842613015665,
      "learning_rate": 9.803002322437833e-06,
      "loss": 0.6791,
      "step": 1280
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5317373193392139,
      "learning_rate": 9.802590558156863e-06,
      "loss": 0.2367,
      "step": 1281
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5604658525349635,
      "learning_rate": 9.80217837265657e-06,
      "loss": 0.2604,
      "step": 1282
    },
    {
      "epoch": 0.12,
      "grad_norm": 10.291514778662403,
      "learning_rate": 9.80176576597311e-06,
      "loss": 0.5992,
      "step": 1283
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4833616735251935,
      "learning_rate": 9.80135273814267e-06,
      "loss": 0.2202,
      "step": 1284
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.691854716839821,
      "learning_rate": 9.800939289201475e-06,
      "loss": 0.5764,
      "step": 1285
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6096064249079658,
      "learning_rate": 9.800525419185786e-06,
      "loss": 0.2671,
      "step": 1286
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.511207124689196,
      "learning_rate": 9.800111128131903e-06,
      "loss": 0.2182,
      "step": 1287
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5062372241176452,
      "learning_rate": 9.799696416076166e-06,
      "loss": 0.2349,
      "step": 1288
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4368474929314936,
      "learning_rate": 9.79928128305494e-06,
      "loss": 0.2002,
      "step": 1289
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3875094814256834,
      "learning_rate": 9.798865729104642e-06,
      "loss": 0.2188,
      "step": 1290
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5668982794250443,
      "learning_rate": 9.798449754261716e-06,
      "loss": 0.2463,
      "step": 1291
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.542486006248439,
      "learning_rate": 9.798033358562647e-06,
      "loss": 0.2018,
      "step": 1292
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.238366137644412,
      "learning_rate": 9.797616542043953e-06,
      "loss": 0.655,
      "step": 1293
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.5554602100043886,
      "learning_rate": 9.797199304742194e-06,
      "loss": 0.6147,
      "step": 1294
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4613547152994575,
      "learning_rate": 9.796781646693964e-06,
      "loss": 0.2205,
      "step": 1295
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.710434289735432,
      "learning_rate": 9.796363567935893e-06,
      "loss": 0.2105,
      "step": 1296
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5362417298041149,
      "learning_rate": 9.795945068504654e-06,
      "loss": 0.2438,
      "step": 1297
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.515987686311798,
      "learning_rate": 9.795526148436945e-06,
      "loss": 0.2398,
      "step": 1298
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.639302275640009,
      "learning_rate": 9.795106807769514e-06,
      "loss": 0.2539,
      "step": 1299
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4559939795234584,
      "learning_rate": 9.794687046539137e-06,
      "loss": 0.2137,
      "step": 1300
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.853504417286309,
      "learning_rate": 9.794266864782632e-06,
      "loss": 0.2282,
      "step": 1301
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3191649911469356,
      "learning_rate": 9.793846262536847e-06,
      "loss": 0.1953,
      "step": 1302
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5865533395406177,
      "learning_rate": 9.793425239838679e-06,
      "loss": 0.2203,
      "step": 1303
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4278308676354468,
      "learning_rate": 9.793003796725049e-06,
      "loss": 0.2025,
      "step": 1304
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.512336296548195,
      "learning_rate": 9.792581933232924e-06,
      "loss": 0.7128,
      "step": 1305
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.58195830952609,
      "learning_rate": 9.792159649399299e-06,
      "loss": 0.2295,
      "step": 1306
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5301850695258075,
      "learning_rate": 9.791736945261215e-06,
      "loss": 0.2201,
      "step": 1307
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5035709780501565,
      "learning_rate": 9.791313820855748e-06,
      "loss": 0.2112,
      "step": 1308
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.556000081946997,
      "learning_rate": 9.790890276220003e-06,
      "loss": 0.1973,
      "step": 1309
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.586515979185901,
      "learning_rate": 9.790466311391132e-06,
      "loss": 0.2385,
      "step": 1310
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4316714822694754,
      "learning_rate": 9.79004192640632e-06,
      "loss": 0.1943,
      "step": 1311
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5540140666665927,
      "learning_rate": 9.789617121302784e-06,
      "loss": 0.2559,
      "step": 1312
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.474987790745863,
      "learning_rate": 9.789191896117786e-06,
      "loss": 0.1948,
      "step": 1313
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.467724090041633,
      "learning_rate": 9.788766250888622e-06,
      "loss": 0.2244,
      "step": 1314
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.498370498658575,
      "learning_rate": 9.78834018565262e-06,
      "loss": 0.1955,
      "step": 1315
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6674408401598377,
      "learning_rate": 9.787913700447152e-06,
      "loss": 0.2266,
      "step": 1316
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6667390072739972,
      "learning_rate": 9.787486795309621e-06,
      "loss": 0.2284,
      "step": 1317
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4570678296691502,
      "learning_rate": 9.787059470277474e-06,
      "loss": 0.194,
      "step": 1318
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7657039850893035,
      "learning_rate": 9.786631725388184e-06,
      "loss": 0.2769,
      "step": 1319
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4778325426910708,
      "learning_rate": 9.786203560679273e-06,
      "loss": 0.2138,
      "step": 1320
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5659656117897238,
      "learning_rate": 9.78577497618829e-06,
      "loss": 0.2265,
      "step": 1321
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4874023141286152,
      "learning_rate": 9.785345971952825e-06,
      "loss": 0.2674,
      "step": 1322
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5345452477372705,
      "learning_rate": 9.784916548010507e-06,
      "loss": 0.2056,
      "step": 1323
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3923046069797396,
      "learning_rate": 9.784486704398997e-06,
      "loss": 0.2279,
      "step": 1324
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4435927408098292,
      "learning_rate": 9.784056441155995e-06,
      "loss": 0.2202,
      "step": 1325
    },
    {
      "epoch": 0.12,
      "grad_norm": 14.260563739174533,
      "learning_rate": 9.78362575831924e-06,
      "loss": 0.7983,
      "step": 1326
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3945308914309604,
      "learning_rate": 9.783194655926505e-06,
      "loss": 0.2184,
      "step": 1327
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.681732427144389,
      "learning_rate": 9.7827631340156e-06,
      "loss": 0.6833,
      "step": 1328
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4954933286115126,
      "learning_rate": 9.782331192624372e-06,
      "loss": 0.2286,
      "step": 1329
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5030373215545665,
      "learning_rate": 9.781898831790706e-06,
      "loss": 0.2072,
      "step": 1330
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7058599181989664,
      "learning_rate": 9.781466051552524e-06,
      "loss": 0.296,
      "step": 1331
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.534558549626928,
      "learning_rate": 9.78103285194778e-06,
      "loss": 0.265,
      "step": 1332
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2974407199301679,
      "learning_rate": 9.780599233014474e-06,
      "loss": 0.1756,
      "step": 1333
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8190696297314055,
      "learning_rate": 9.780165194790631e-06,
      "loss": 0.2478,
      "step": 1334
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.787632605832483,
      "learning_rate": 9.779730737314324e-06,
      "loss": 0.4865,
      "step": 1335
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4706994356750667,
      "learning_rate": 9.779295860623656e-06,
      "loss": 0.2096,
      "step": 1336
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6108759335483047,
      "learning_rate": 9.778860564756769e-06,
      "loss": 0.2359,
      "step": 1337
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7933895244000553,
      "learning_rate": 9.77842484975184e-06,
      "loss": 0.2467,
      "step": 1338
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.62890708181144,
      "learning_rate": 9.777988715647086e-06,
      "loss": 0.2744,
      "step": 1339
    },
    {
      "epoch": 0.12,
      "grad_norm": 16.46262544170884,
      "learning_rate": 9.777552162480758e-06,
      "loss": 0.5294,
      "step": 1340
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4209293268954342,
      "learning_rate": 9.777115190291145e-06,
      "loss": 0.2139,
      "step": 1341
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.72764171118372,
      "learning_rate": 9.776677799116572e-06,
      "loss": 0.5607,
      "step": 1342
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7403734309714833,
      "learning_rate": 9.776239988995401e-06,
      "loss": 0.2592,
      "step": 1343
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7737311454161038,
      "learning_rate": 9.77580175996603e-06,
      "loss": 0.2216,
      "step": 1344
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5267192561354603,
      "learning_rate": 9.775363112066897e-06,
      "loss": 0.2125,
      "step": 1345
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4963680371679078,
      "learning_rate": 9.774924045336474e-06,
      "loss": 0.1983,
      "step": 1346
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8079639922623398,
      "learning_rate": 9.774484559813266e-06,
      "loss": 0.2203,
      "step": 1347
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7514724674918378,
      "learning_rate": 9.774044655535825e-06,
      "loss": 0.2325,
      "step": 1348
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6516521196514247,
      "learning_rate": 9.77360433254273e-06,
      "loss": 0.2334,
      "step": 1349
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.619978162877564,
      "learning_rate": 9.773163590872598e-06,
      "loss": 0.167,
      "step": 1350
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3781408435257887,
      "learning_rate": 9.772722430564092e-06,
      "loss": 0.2031,
      "step": 1351
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4534652274103534,
      "learning_rate": 9.772280851655897e-06,
      "loss": 0.2379,
      "step": 1352
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6194328708807977,
      "learning_rate": 9.771838854186748e-06,
      "loss": 0.2404,
      "step": 1353
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6053140746467178,
      "learning_rate": 9.771396438195407e-06,
      "loss": 0.2291,
      "step": 1354
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7736142578224894,
      "learning_rate": 9.770953603720682e-06,
      "loss": 0.2549,
      "step": 1355
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.561439816245013,
      "learning_rate": 9.770510350801407e-06,
      "loss": 0.2298,
      "step": 1356
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7336290524638738,
      "learning_rate": 9.770066679476461e-06,
      "loss": 0.2887,
      "step": 1357
    },
    {
      "epoch": 0.12,
      "grad_norm": 4.887494246619368,
      "learning_rate": 9.769622589784756e-06,
      "loss": 0.6989,
      "step": 1358
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4755083723765021,
      "learning_rate": 9.769178081765243e-06,
      "loss": 0.1889,
      "step": 1359
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.197029803121975,
      "learning_rate": 9.76873315545691e-06,
      "loss": 0.7061,
      "step": 1360
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5647618203087805,
      "learning_rate": 9.768287810898773e-06,
      "loss": 0.2707,
      "step": 1361
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6157034641005685,
      "learning_rate": 9.7678420481299e-06,
      "loss": 0.2276,
      "step": 1362
    },
    {
      "epoch": 0.12,
      "grad_norm": 6.4654884907908885,
      "learning_rate": 9.767395867189382e-06,
      "loss": 0.7364,
      "step": 1363
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.433019053114797,
      "learning_rate": 9.766949268116354e-06,
      "loss": 0.2343,
      "step": 1364
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5538597346702323,
      "learning_rate": 9.766502250949985e-06,
      "loss": 0.2503,
      "step": 1365
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3646695833843991,
      "learning_rate": 9.766054815729481e-06,
      "loss": 0.2232,
      "step": 1366
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5970171125572576,
      "learning_rate": 9.765606962494088e-06,
      "loss": 0.2345,
      "step": 1367
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6556289825769401,
      "learning_rate": 9.765158691283083e-06,
      "loss": 0.1729,
      "step": 1368
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5322586406656418,
      "learning_rate": 9.764710002135784e-06,
      "loss": 0.2062,
      "step": 1369
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8255673691069925,
      "learning_rate": 9.764260895091541e-06,
      "loss": 0.2862,
      "step": 1370
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3597204919507264,
      "learning_rate": 9.763811370189747e-06,
      "loss": 0.2212,
      "step": 1371
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.2911290957161565,
      "learning_rate": 9.763361427469826e-06,
      "loss": 0.5243,
      "step": 1372
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6172603682641917,
      "learning_rate": 9.762911066971244e-06,
      "loss": 0.2044,
      "step": 1373
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.8489493778794825,
      "learning_rate": 9.762460288733497e-06,
      "loss": 0.6775,
      "step": 1374
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5719614905211408,
      "learning_rate": 9.762009092796124e-06,
      "loss": 0.2542,
      "step": 1375
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4730680902766344,
      "learning_rate": 9.761557479198697e-06,
      "loss": 0.2207,
      "step": 1376
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8140321492705818,
      "learning_rate": 9.761105447980824e-06,
      "loss": 0.2665,
      "step": 1377
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3652875725780154,
      "learning_rate": 9.760652999182153e-06,
      "loss": 0.2117,
      "step": 1378
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3426529641731115,
      "learning_rate": 9.760200132842368e-06,
      "loss": 0.1858,
      "step": 1379
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6067898350234744,
      "learning_rate": 9.759746849001184e-06,
      "loss": 0.2643,
      "step": 1380
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.384507786544646,
      "learning_rate": 9.759293147698362e-06,
      "loss": 0.5563,
      "step": 1381
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.397886108033073,
      "learning_rate": 9.758839028973693e-06,
      "loss": 0.2325,
      "step": 1382
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4171389846139166,
      "learning_rate": 9.758384492867004e-06,
      "loss": 0.1737,
      "step": 1383
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3069559889914888,
      "learning_rate": 9.757929539418163e-06,
      "loss": 0.1571,
      "step": 1384
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.455238289206748,
      "learning_rate": 9.757474168667072e-06,
      "loss": 0.2157,
      "step": 1385
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6148798576656356,
      "learning_rate": 9.757018380653668e-06,
      "loss": 0.2677,
      "step": 1386
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.840002053902008,
      "learning_rate": 9.756562175417932e-06,
      "loss": 0.2567,
      "step": 1387
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6660978642026054,
      "learning_rate": 9.756105552999873e-06,
      "loss": 0.2459,
      "step": 1388
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6217362629568028,
      "learning_rate": 9.755648513439538e-06,
      "loss": 0.2575,
      "step": 1389
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9194841216901495,
      "learning_rate": 9.755191056777013e-06,
      "loss": 0.2449,
      "step": 1390
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6537192705174653,
      "learning_rate": 9.754733183052424e-06,
      "loss": 0.2619,
      "step": 1391
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6398327250532594,
      "learning_rate": 9.754274892305926e-06,
      "loss": 0.3254,
      "step": 1392
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.242598002863911,
      "learning_rate": 9.753816184577715e-06,
      "loss": 0.7074,
      "step": 1393
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5360053315111086,
      "learning_rate": 9.753357059908023e-06,
      "loss": 0.1879,
      "step": 1394
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.560605521751944,
      "learning_rate": 9.752897518337117e-06,
      "loss": 0.6717,
      "step": 1395
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6970952588509614,
      "learning_rate": 9.752437559905303e-06,
      "loss": 0.2863,
      "step": 1396
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6245591777615067,
      "learning_rate": 9.751977184652923e-06,
      "loss": 0.2506,
      "step": 1397
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4553317243615769,
      "learning_rate": 9.751516392620354e-06,
      "loss": 0.2059,
      "step": 1398
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3703405397187842,
      "learning_rate": 9.751055183848012e-06,
      "loss": 0.201,
      "step": 1399
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3462476232989684,
      "learning_rate": 9.750593558376347e-06,
      "loss": 0.2109,
      "step": 1400
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3568632156414813,
      "learning_rate": 9.750131516245844e-06,
      "loss": 0.1865,
      "step": 1401
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.067128437647094,
      "learning_rate": 9.749669057497033e-06,
      "loss": 0.625,
      "step": 1402
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7877764548781894,
      "learning_rate": 9.749206182170469e-06,
      "loss": 0.2509,
      "step": 1403
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5760283801193635,
      "learning_rate": 9.748742890306753e-06,
      "loss": 0.2204,
      "step": 1404
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4036997063304733,
      "learning_rate": 9.748279181946518e-06,
      "loss": 0.2653,
      "step": 1405
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.63476479840845,
      "learning_rate": 9.747815057130432e-06,
      "loss": 0.6292,
      "step": 1406
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.567922977874336,
      "learning_rate": 9.747350515899205e-06,
      "loss": 0.2223,
      "step": 1407
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.401483697990862,
      "learning_rate": 9.74688555829358e-06,
      "loss": 0.2188,
      "step": 1408
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5119329655951996,
      "learning_rate": 9.746420184354334e-06,
      "loss": 0.2581,
      "step": 1409
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4536326467241005,
      "learning_rate": 9.745954394122288e-06,
      "loss": 0.218,
      "step": 1410
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.695547820614073,
      "learning_rate": 9.74548818763829e-06,
      "loss": 0.2622,
      "step": 1411
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4647864988621508,
      "learning_rate": 9.745021564943234e-06,
      "loss": 0.1925,
      "step": 1412
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5198145439694182,
      "learning_rate": 9.74455452607804e-06,
      "loss": 0.2206,
      "step": 1413
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4442459231786284,
      "learning_rate": 9.744087071083679e-06,
      "loss": 0.2641,
      "step": 1414
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.568248759165224,
      "learning_rate": 9.743619200001141e-06,
      "loss": 0.5312,
      "step": 1415
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.361606845474261,
      "learning_rate": 9.743150912871468e-06,
      "loss": 0.1975,
      "step": 1416
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5863645861662339,
      "learning_rate": 9.742682209735727e-06,
      "loss": 0.2312,
      "step": 1417
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4316601263213196,
      "learning_rate": 9.742213090635033e-06,
      "loss": 0.233,
      "step": 1418
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5107755960978746,
      "learning_rate": 9.741743555610523e-06,
      "loss": 0.2639,
      "step": 1419
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6825936711935017,
      "learning_rate": 9.741273604703383e-06,
      "loss": 0.2527,
      "step": 1420
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4347770983502635,
      "learning_rate": 9.74080323795483e-06,
      "loss": 0.1879,
      "step": 1421
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4492366696477448,
      "learning_rate": 9.740332455406119e-06,
      "loss": 0.1968,
      "step": 1422
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4753023509916041,
      "learning_rate": 9.739861257098538e-06,
      "loss": 0.2367,
      "step": 1423
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.443532131261639,
      "learning_rate": 9.739389643073418e-06,
      "loss": 0.2417,
      "step": 1424
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.389970753446936,
      "learning_rate": 9.738917613372121e-06,
      "loss": 0.1911,
      "step": 1425
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6686591218076927,
      "learning_rate": 9.738445168036046e-06,
      "loss": 0.2497,
      "step": 1426
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6836727205379722,
      "learning_rate": 9.73797230710663e-06,
      "loss": 0.2221,
      "step": 1427
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5163169899030604,
      "learning_rate": 9.737499030625349e-06,
      "loss": 0.2526,
      "step": 1428
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4864815423421556,
      "learning_rate": 9.73702533863371e-06,
      "loss": 0.2212,
      "step": 1429
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.814910811489299,
      "learning_rate": 9.736551231173259e-06,
      "loss": 0.2226,
      "step": 1430
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.458506153625006,
      "learning_rate": 9.736076708285578e-06,
      "loss": 0.2472,
      "step": 1431
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.76384294903257,
      "learning_rate": 9.735601770012287e-06,
      "loss": 0.5109,
      "step": 1432
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3619174120404127,
      "learning_rate": 9.73512641639504e-06,
      "loss": 0.214,
      "step": 1433
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7753481843874195,
      "learning_rate": 9.73465064747553e-06,
      "loss": 0.2289,
      "step": 1434
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3555156516873919,
      "learning_rate": 9.734174463295483e-06,
      "loss": 0.2215,
      "step": 1435
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6644703620964105,
      "learning_rate": 9.733697863896667e-06,
      "loss": 0.2563,
      "step": 1436
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5389741547371774,
      "learning_rate": 9.733220849320881e-06,
      "loss": 0.2138,
      "step": 1437
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5342279404153054,
      "learning_rate": 9.732743419609961e-06,
      "loss": 0.2376,
      "step": 1438
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.417505801024901,
      "learning_rate": 9.732265574805783e-06,
      "loss": 0.1965,
      "step": 1439
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4602823394061561,
      "learning_rate": 9.731787314950257e-06,
      "loss": 0.2189,
      "step": 1440
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4821086694772796,
      "learning_rate": 9.731308640085329e-06,
      "loss": 0.216,
      "step": 1441
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4023358219372093,
      "learning_rate": 9.73082955025298e-06,
      "loss": 0.205,
      "step": 1442
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4310263685842781,
      "learning_rate": 9.730350045495233e-06,
      "loss": 0.2326,
      "step": 1443
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.602868529721414,
      "learning_rate": 9.729870125854144e-06,
      "loss": 0.2553,
      "step": 1444
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.397618593498871,
      "learning_rate": 9.729389791371799e-06,
      "loss": 0.2154,
      "step": 1445
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6499921532142794,
      "learning_rate": 9.728909042090335e-06,
      "loss": 0.2401,
      "step": 1446
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3688389737613749,
      "learning_rate": 9.72842787805191e-06,
      "loss": 0.2105,
      "step": 1447
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.19552193675555,
      "learning_rate": 9.727946299298729e-06,
      "loss": 0.7123,
      "step": 1448
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.653805216742925,
      "learning_rate": 9.72746430587303e-06,
      "loss": 0.2885,
      "step": 1449
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2988681907259605,
      "learning_rate": 9.726981897817082e-06,
      "loss": 0.2125,
      "step": 1450
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.322684548927425,
      "learning_rate": 9.726499075173201e-06,
      "loss": 0.2647,
      "step": 1451
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6355570361047997,
      "learning_rate": 9.726015837983733e-06,
      "loss": 0.2769,
      "step": 1452
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6672418455102909,
      "learning_rate": 9.72553218629106e-06,
      "loss": 0.2583,
      "step": 1453
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6797575559914304,
      "learning_rate": 9.7250481201376e-06,
      "loss": 0.2675,
      "step": 1454
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5953859721300965,
      "learning_rate": 9.724563639565814e-06,
      "loss": 0.2656,
      "step": 1455
    },
    {
      "epoch": 0.13,
      "grad_norm": 6.545419524254801,
      "learning_rate": 9.724078744618187e-06,
      "loss": 0.5678,
      "step": 1456
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4076353661038261,
      "learning_rate": 9.723593435337252e-06,
      "loss": 0.194,
      "step": 1457
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3695557615416394,
      "learning_rate": 9.723107711765574e-06,
      "loss": 0.1817,
      "step": 1458
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5056216077026754,
      "learning_rate": 9.722621573945752e-06,
      "loss": 0.2541,
      "step": 1459
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4086986331812184,
      "learning_rate": 9.722135021920427e-06,
      "loss": 0.2253,
      "step": 1460
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6736213070170278,
      "learning_rate": 9.721648055732269e-06,
      "loss": 0.2741,
      "step": 1461
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.289780414111577,
      "learning_rate": 9.721160675423989e-06,
      "loss": 0.6023,
      "step": 1462
    },
    {
      "epoch": 0.13,
      "grad_norm": 9.977225439482737,
      "learning_rate": 9.720672881038335e-06,
      "loss": 0.6945,
      "step": 1463
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6093627662291667,
      "learning_rate": 9.720184672618091e-06,
      "loss": 0.2055,
      "step": 1464
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4712892881817583,
      "learning_rate": 9.719696050206072e-06,
      "loss": 0.1919,
      "step": 1465
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.659513707756264,
      "learning_rate": 9.719207013845137e-06,
      "loss": 0.2505,
      "step": 1466
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5243958373305075,
      "learning_rate": 9.718717563578176e-06,
      "loss": 0.2311,
      "step": 1467
    },
    {
      "epoch": 0.13,
      "grad_norm": 35.588888489020526,
      "learning_rate": 9.718227699448117e-06,
      "loss": 0.6085,
      "step": 1468
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.644649015970009,
      "learning_rate": 9.717737421497926e-06,
      "loss": 0.2753,
      "step": 1469
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5463009947388136,
      "learning_rate": 9.717246729770602e-06,
      "loss": 0.2283,
      "step": 1470
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5054326660250126,
      "learning_rate": 9.716755624309185e-06,
      "loss": 0.2071,
      "step": 1471
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6588733308638681,
      "learning_rate": 9.716264105156742e-06,
      "loss": 0.2256,
      "step": 1472
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.544093895929859,
      "learning_rate": 9.715772172356388e-06,
      "loss": 0.2439,
      "step": 1473
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.612701265618232,
      "learning_rate": 9.715279825951267e-06,
      "loss": 0.2168,
      "step": 1474
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5483721292409873,
      "learning_rate": 9.714787065984561e-06,
      "loss": 0.2727,
      "step": 1475
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7387351473185197,
      "learning_rate": 9.71429389249949e-06,
      "loss": 0.2994,
      "step": 1476
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5916130630184355,
      "learning_rate": 9.713800305539306e-06,
      "loss": 0.2437,
      "step": 1477
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5751941961444946,
      "learning_rate": 9.713306305147301e-06,
      "loss": 0.2664,
      "step": 1478
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4668817768435833,
      "learning_rate": 9.7128118913668e-06,
      "loss": 0.2043,
      "step": 1479
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5778937902063592,
      "learning_rate": 9.712317064241173e-06,
      "loss": 0.1992,
      "step": 1480
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4071469369426912,
      "learning_rate": 9.711821823813812e-06,
      "loss": 0.2073,
      "step": 1481
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4621448359809046,
      "learning_rate": 9.711326170128157e-06,
      "loss": 0.2123,
      "step": 1482
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5115480096553133,
      "learning_rate": 9.710830103227679e-06,
      "loss": 0.2196,
      "step": 1483
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.3262509310442026,
      "learning_rate": 9.710333623155888e-06,
      "loss": 0.6464,
      "step": 1484
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6554070073672746,
      "learning_rate": 9.709836729956326e-06,
      "loss": 0.2978,
      "step": 1485
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1643847357864274,
      "learning_rate": 9.709339423672575e-06,
      "loss": 0.3452,
      "step": 1486
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7874338185863419,
      "learning_rate": 9.708841704348251e-06,
      "loss": 0.2858,
      "step": 1487
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.414075135867976,
      "learning_rate": 9.70834357202701e-06,
      "loss": 0.2625,
      "step": 1488
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2314861320118315,
      "learning_rate": 9.70784502675254e-06,
      "loss": 0.2782,
      "step": 1489
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.569800864707917,
      "learning_rate": 9.707346068568568e-06,
      "loss": 0.6556,
      "step": 1490
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.4010433667869258,
      "learning_rate": 9.706846697518853e-06,
      "loss": 0.2048,
      "step": 1491
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.558032476565784,
      "learning_rate": 9.706346913647196e-06,
      "loss": 0.2452,
      "step": 1492
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6051856338178325,
      "learning_rate": 9.70584671699743e-06,
      "loss": 0.2372,
      "step": 1493
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6038488145711653,
      "learning_rate": 9.705346107613426e-06,
      "loss": 0.2279,
      "step": 1494
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.542574697116791,
      "learning_rate": 9.704845085539091e-06,
      "loss": 0.2569,
      "step": 1495
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5089928697572916,
      "learning_rate": 9.704343650818367e-06,
      "loss": 0.1943,
      "step": 1496
    },
    {
      "epoch": 0.14,
      "grad_norm": 19.17228023004775,
      "learning_rate": 9.703841803495234e-06,
      "loss": 0.5982,
      "step": 1497
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.802055499782346,
      "learning_rate": 9.703339543613708e-06,
      "loss": 0.247,
      "step": 1498
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.060447217686787,
      "learning_rate": 9.702836871217838e-06,
      "loss": 0.7621,
      "step": 1499
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7153440068849464,
      "learning_rate": 9.702333786351716e-06,
      "loss": 0.2828,
      "step": 1500
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6696937495643533,
      "learning_rate": 9.701830289059461e-06,
      "loss": 0.2511,
      "step": 1501
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7526031022780046,
      "learning_rate": 9.701326379385238e-06,
      "loss": 0.2412,
      "step": 1502
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6644634636620863,
      "learning_rate": 9.70082205737324e-06,
      "loss": 0.2325,
      "step": 1503
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5190482241292544,
      "learning_rate": 9.700317323067701e-06,
      "loss": 0.2371,
      "step": 1504
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6597065058947078,
      "learning_rate": 9.699812176512887e-06,
      "loss": 0.2756,
      "step": 1505
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6788979778258644,
      "learning_rate": 9.699306617753104e-06,
      "loss": 0.2335,
      "step": 1506
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.34889510056142,
      "learning_rate": 9.698800646832696e-06,
      "loss": 0.7511,
      "step": 1507
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5206134733439736,
      "learning_rate": 9.698294263796035e-06,
      "loss": 0.2305,
      "step": 1508
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4592531397949637,
      "learning_rate": 9.697787468687538e-06,
      "loss": 0.183,
      "step": 1509
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.238361294596597,
      "learning_rate": 9.697280261551655e-06,
      "loss": 0.5984,
      "step": 1510
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5202307367597663,
      "learning_rate": 9.696772642432866e-06,
      "loss": 0.2188,
      "step": 1511
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.785182256185916,
      "learning_rate": 9.696264611375697e-06,
      "loss": 0.3013,
      "step": 1512
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.613498447267519,
      "learning_rate": 9.695756168424703e-06,
      "loss": 0.2159,
      "step": 1513
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.360304850727104,
      "learning_rate": 9.695247313624483e-06,
      "loss": 0.2247,
      "step": 1514
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.829902991065343,
      "learning_rate": 9.69473804701966e-06,
      "loss": 0.6448,
      "step": 1515
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.388858500709557,
      "learning_rate": 9.694228368654906e-06,
      "loss": 0.2198,
      "step": 1516
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4334035889711854,
      "learning_rate": 9.69371827857492e-06,
      "loss": 0.2022,
      "step": 1517
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.903483441701676,
      "learning_rate": 9.693207776824443e-06,
      "loss": 0.6136,
      "step": 1518
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4800221428797617,
      "learning_rate": 9.692696863448246e-06,
      "loss": 0.2429,
      "step": 1519
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.509708855861796,
      "learning_rate": 9.69218553849114e-06,
      "loss": 0.2079,
      "step": 1520
    },
    {
      "epoch": 0.14,
      "grad_norm": 4.466364830728979,
      "learning_rate": 9.691673801997974e-06,
      "loss": 0.4337,
      "step": 1521
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7355075618870217,
      "learning_rate": 9.691161654013628e-06,
      "loss": 0.3147,
      "step": 1522
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6767332182713546,
      "learning_rate": 9.690649094583025e-06,
      "loss": 0.2757,
      "step": 1523
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.60261344021058,
      "learning_rate": 9.690136123751115e-06,
      "loss": 0.2268,
      "step": 1524
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5975917986066306,
      "learning_rate": 9.689622741562891e-06,
      "loss": 0.2275,
      "step": 1525
    },
    {
      "epoch": 0.14,
      "grad_norm": 9.214200655248009,
      "learning_rate": 9.68910894806338e-06,
      "loss": 0.7624,
      "step": 1526
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4338629365556717,
      "learning_rate": 9.688594743297646e-06,
      "loss": 0.1939,
      "step": 1527
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4487997605862575,
      "learning_rate": 9.688080127310788e-06,
      "loss": 0.1901,
      "step": 1528
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6793086975285052,
      "learning_rate": 9.68756510014794e-06,
      "loss": 0.2375,
      "step": 1529
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7786184820924373,
      "learning_rate": 9.687049661854274e-06,
      "loss": 0.2559,
      "step": 1530
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4837469586784466,
      "learning_rate": 9.686533812474997e-06,
      "loss": 0.2234,
      "step": 1531
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5184614756769166,
      "learning_rate": 9.686017552055353e-06,
      "loss": 0.2497,
      "step": 1532
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.373839168555958,
      "learning_rate": 9.685500880640623e-06,
      "loss": 0.196,
      "step": 1533
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4967024572015408,
      "learning_rate": 9.68498379827612e-06,
      "loss": 0.2746,
      "step": 1534
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4557726982782924,
      "learning_rate": 9.684466305007197e-06,
      "loss": 0.2107,
      "step": 1535
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3613747660911153,
      "learning_rate": 9.683948400879239e-06,
      "loss": 0.1702,
      "step": 1536
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.581924473185747,
      "learning_rate": 9.683430085937672e-06,
      "loss": 0.6703,
      "step": 1537
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.654619872673201,
      "learning_rate": 9.682911360227958e-06,
      "loss": 0.2463,
      "step": 1538
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.717339824127487,
      "learning_rate": 9.682392223795588e-06,
      "loss": 0.2612,
      "step": 1539
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.870179856742636,
      "learning_rate": 9.681872676686097e-06,
      "loss": 0.5091,
      "step": 1540
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2687919437204354,
      "learning_rate": 9.68135271894505e-06,
      "loss": 0.1916,
      "step": 1541
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4977883347388068,
      "learning_rate": 9.680832350618055e-06,
      "loss": 0.2236,
      "step": 1542
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3333121128460268,
      "learning_rate": 9.680311571750747e-06,
      "loss": 0.198,
      "step": 1543
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.362686623879738,
      "learning_rate": 9.679790382388805e-06,
      "loss": 0.2306,
      "step": 1544
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.374259724616099,
      "learning_rate": 9.67926878257794e-06,
      "loss": 0.2333,
      "step": 1545
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.914492492270594,
      "learning_rate": 9.678746772363898e-06,
      "loss": 0.651,
      "step": 1546
    },
    {
      "epoch": 0.14,
      "grad_norm": 5.1646795095843085,
      "learning_rate": 9.678224351792467e-06,
      "loss": 0.5575,
      "step": 1547
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4586040250670604,
      "learning_rate": 9.677701520909462e-06,
      "loss": 0.2163,
      "step": 1548
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7128062201475298,
      "learning_rate": 9.677178279760741e-06,
      "loss": 0.2747,
      "step": 1549
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8728336850803935,
      "learning_rate": 9.676654628392198e-06,
      "loss": 0.1964,
      "step": 1550
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4372128907560218,
      "learning_rate": 9.676130566849757e-06,
      "loss": 0.2361,
      "step": 1551
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4521916504912729,
      "learning_rate": 9.675606095179382e-06,
      "loss": 0.2063,
      "step": 1552
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4639792740940447,
      "learning_rate": 9.675081213427076e-06,
      "loss": 0.2159,
      "step": 1553
    },
    {
      "epoch": 0.14,
      "grad_norm": 8.083926799490218,
      "learning_rate": 9.674555921638871e-06,
      "loss": 0.5338,
      "step": 1554
    },
    {
      "epoch": 0.14,
      "grad_norm": 7.474451903833828,
      "learning_rate": 9.67403021986084e-06,
      "loss": 0.5834,
      "step": 1555
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6098550031319276,
      "learning_rate": 9.67350410813909e-06,
      "loss": 0.2447,
      "step": 1556
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6169250175705265,
      "learning_rate": 9.672977586519766e-06,
      "loss": 0.1985,
      "step": 1557
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5307988607133403,
      "learning_rate": 9.672450655049046e-06,
      "loss": 0.199,
      "step": 1558
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.559119635214225,
      "learning_rate": 9.671923313773146e-06,
      "loss": 0.2466,
      "step": 1559
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.627966269738106,
      "learning_rate": 9.671395562738319e-06,
      "loss": 0.2337,
      "step": 1560
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6491996743080286,
      "learning_rate": 9.67086740199085e-06,
      "loss": 0.1773,
      "step": 1561
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.356600835413021,
      "learning_rate": 9.670338831577063e-06,
      "loss": 0.2125,
      "step": 1562
    },
    {
      "epoch": 0.14,
      "grad_norm": 63.35696412463156,
      "learning_rate": 9.669809851543314e-06,
      "loss": 0.4855,
      "step": 1563
    },
    {
      "epoch": 0.14,
      "grad_norm": 3.049787239827408,
      "learning_rate": 9.669280461936004e-06,
      "loss": 0.2279,
      "step": 1564
    },
    {
      "epoch": 0.14,
      "grad_norm": 12.24799275614197,
      "learning_rate": 9.66875066280156e-06,
      "loss": 0.759,
      "step": 1565
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.660602281351028,
      "learning_rate": 9.66822045418645e-06,
      "loss": 0.232,
      "step": 1566
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5281161930145386,
      "learning_rate": 9.667689836137178e-06,
      "loss": 0.231,
      "step": 1567
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5534391558957814,
      "learning_rate": 9.667158808700283e-06,
      "loss": 0.2736,
      "step": 1568
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4269460442533408,
      "learning_rate": 9.666627371922335e-06,
      "loss": 0.2483,
      "step": 1569
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7446302257096002,
      "learning_rate": 9.66609552584995e-06,
      "loss": 0.2479,
      "step": 1570
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2624806892002467,
      "learning_rate": 9.665563270529772e-06,
      "loss": 0.1843,
      "step": 1571
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3535202675470834,
      "learning_rate": 9.665030606008482e-06,
      "loss": 0.1808,
      "step": 1572
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.401102646989681,
      "learning_rate": 9.664497532332801e-06,
      "loss": 0.2389,
      "step": 1573
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6559525302787184,
      "learning_rate": 9.663964049549483e-06,
      "loss": 0.2973,
      "step": 1574
    },
    {
      "epoch": 0.14,
      "grad_norm": 6.413560985689603,
      "learning_rate": 9.663430157705317e-06,
      "loss": 0.6922,
      "step": 1575
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4264527787838344,
      "learning_rate": 9.662895856847128e-06,
      "loss": 0.2494,
      "step": 1576
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6127958918792484,
      "learning_rate": 9.66236114702178e-06,
      "loss": 0.1753,
      "step": 1577
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3585435056716502,
      "learning_rate": 9.66182602827617e-06,
      "loss": 0.213,
      "step": 1578
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5438781532433266,
      "learning_rate": 9.661290500657227e-06,
      "loss": 0.2069,
      "step": 1579
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.543665874819746,
      "learning_rate": 9.660754564211927e-06,
      "loss": 0.2294,
      "step": 1580
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7067035412791207,
      "learning_rate": 9.660218218987272e-06,
      "loss": 0.2265,
      "step": 1581
    },
    {
      "epoch": 0.14,
      "grad_norm": 10.467776533858736,
      "learning_rate": 9.659681465030303e-06,
      "loss": 0.5051,
      "step": 1582
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.574338018320039,
      "learning_rate": 9.659144302388098e-06,
      "loss": 0.2205,
      "step": 1583
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4633130468716418,
      "learning_rate": 9.65860673110777e-06,
      "loss": 0.2177,
      "step": 1584
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.406872180474505,
      "learning_rate": 9.658068751236464e-06,
      "loss": 0.1986,
      "step": 1585
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.0551623438948905,
      "learning_rate": 9.65753036282137e-06,
      "loss": 0.5363,
      "step": 1586
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.019332690263125,
      "learning_rate": 9.656991565909703e-06,
      "loss": 0.5602,
      "step": 1587
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.883665270309864,
      "learning_rate": 9.656452360548724e-06,
      "loss": 0.7415,
      "step": 1588
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4271980463641296,
      "learning_rate": 9.65591274678572e-06,
      "loss": 0.2084,
      "step": 1589
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5027874601647546,
      "learning_rate": 9.655372724668022e-06,
      "loss": 0.2463,
      "step": 1590
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.508088812349196,
      "learning_rate": 9.654832294242995e-06,
      "loss": 0.2127,
      "step": 1591
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6664828194904484,
      "learning_rate": 9.654291455558033e-06,
      "loss": 0.2251,
      "step": 1592
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4732462714063825,
      "learning_rate": 9.653750208660577e-06,
      "loss": 0.2645,
      "step": 1593
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5207671790633346,
      "learning_rate": 9.653208553598093e-06,
      "loss": 0.3045,
      "step": 1594
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4360280628511952,
      "learning_rate": 9.65266649041809e-06,
      "loss": 0.2348,
      "step": 1595
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3142393687880591,
      "learning_rate": 9.652124019168113e-06,
      "loss": 0.2136,
      "step": 1596
    },
    {
      "epoch": 0.15,
      "grad_norm": 9.855407305034326,
      "learning_rate": 9.651581139895735e-06,
      "loss": 0.6933,
      "step": 1597
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.811905923793135,
      "learning_rate": 9.651037852648575e-06,
      "loss": 0.6245,
      "step": 1598
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4301352512236345,
      "learning_rate": 9.65049415747428e-06,
      "loss": 0.1991,
      "step": 1599
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3256508818251829,
      "learning_rate": 9.649950054420536e-06,
      "loss": 0.2044,
      "step": 1600
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5528515781826262,
      "learning_rate": 9.649405543535067e-06,
      "loss": 0.2346,
      "step": 1601
    },
    {
      "epoch": 0.15,
      "grad_norm": 7.861474722874935,
      "learning_rate": 9.648860624865628e-06,
      "loss": 0.4583,
      "step": 1602
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.481493087281962,
      "learning_rate": 9.64831529846001e-06,
      "loss": 0.2416,
      "step": 1603
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4419688655925735,
      "learning_rate": 9.647769564366048e-06,
      "loss": 0.1856,
      "step": 1604
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.487419250696316,
      "learning_rate": 9.647223422631601e-06,
      "loss": 0.6506,
      "step": 1605
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5658454985672334,
      "learning_rate": 9.646676873304572e-06,
      "loss": 0.239,
      "step": 1606
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.346123568300846,
      "learning_rate": 9.646129916432894e-06,
      "loss": 0.1883,
      "step": 1607
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7159474090110591,
      "learning_rate": 9.645582552064544e-06,
      "loss": 0.2526,
      "step": 1608
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4798291218874422,
      "learning_rate": 9.645034780247521e-06,
      "loss": 0.1969,
      "step": 1609
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7089209988549716,
      "learning_rate": 9.644486601029879e-06,
      "loss": 0.266,
      "step": 1610
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7175003373586193,
      "learning_rate": 9.64393801445969e-06,
      "loss": 0.2294,
      "step": 1611
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.983137675669334,
      "learning_rate": 9.64338902058507e-06,
      "loss": 0.6136,
      "step": 1612
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4217464171186251,
      "learning_rate": 9.64283961945417e-06,
      "loss": 0.2548,
      "step": 1613
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.393172465022725,
      "learning_rate": 9.642289811115175e-06,
      "loss": 0.2556,
      "step": 1614
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3534823337985566,
      "learning_rate": 9.641739595616309e-06,
      "loss": 0.1808,
      "step": 1615
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5483100360182829,
      "learning_rate": 9.641188973005826e-06,
      "loss": 0.2355,
      "step": 1616
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8076734057261539,
      "learning_rate": 9.640637943332025e-06,
      "loss": 0.2969,
      "step": 1617
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6138397566969358,
      "learning_rate": 9.64008650664323e-06,
      "loss": 0.2606,
      "step": 1618
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3390441563242057,
      "learning_rate": 9.639534662987807e-06,
      "loss": 0.1927,
      "step": 1619
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.34650652845546,
      "learning_rate": 9.638982412414156e-06,
      "loss": 0.5727,
      "step": 1620
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.315290420885151,
      "learning_rate": 9.638429754970715e-06,
      "loss": 0.2022,
      "step": 1621
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.400888220823928,
      "learning_rate": 9.637876690705953e-06,
      "loss": 0.2193,
      "step": 1622
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6686685770956966,
      "learning_rate": 9.637323219668382e-06,
      "loss": 0.2297,
      "step": 1623
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5626242312424636,
      "learning_rate": 9.63676934190654e-06,
      "loss": 0.2169,
      "step": 1624
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4632741434886631,
      "learning_rate": 9.636215057469009e-06,
      "loss": 0.2104,
      "step": 1625
    },
    {
      "epoch": 0.15,
      "grad_norm": 15.828851560895881,
      "learning_rate": 9.6356603664044e-06,
      "loss": 0.749,
      "step": 1626
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5388895413504848,
      "learning_rate": 9.635105268761368e-06,
      "loss": 0.2521,
      "step": 1627
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6910449756552697,
      "learning_rate": 9.634549764588594e-06,
      "loss": 0.2518,
      "step": 1628
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5677465887335296,
      "learning_rate": 9.633993853934803e-06,
      "loss": 0.2482,
      "step": 1629
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4282670169582334,
      "learning_rate": 9.633437536848751e-06,
      "loss": 0.1935,
      "step": 1630
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6325628667352976,
      "learning_rate": 9.63288081337923e-06,
      "loss": 0.2273,
      "step": 1631
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.439972637445628,
      "learning_rate": 9.632323683575069e-06,
      "loss": 0.2547,
      "step": 1632
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.184542901472758,
      "learning_rate": 9.631766147485131e-06,
      "loss": 0.6734,
      "step": 1633
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.0807375552727425,
      "learning_rate": 9.631208205158317e-06,
      "loss": 0.6731,
      "step": 1634
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.486988051414305,
      "learning_rate": 9.63064985664356e-06,
      "loss": 0.591,
      "step": 1635
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6305986831513088,
      "learning_rate": 9.630091101989836e-06,
      "loss": 0.2631,
      "step": 1636
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.417227489942563,
      "learning_rate": 9.629531941246147e-06,
      "loss": 0.2159,
      "step": 1637
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.48756808889817,
      "learning_rate": 9.628972374461537e-06,
      "loss": 0.1947,
      "step": 1638
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.477435180547878,
      "learning_rate": 9.62841240168508e-06,
      "loss": 0.2224,
      "step": 1639
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4291047854269876,
      "learning_rate": 9.627852022965894e-06,
      "loss": 0.2212,
      "step": 1640
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5068867930497738,
      "learning_rate": 9.627291238353127e-06,
      "loss": 0.2322,
      "step": 1641
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.496245347011364,
      "learning_rate": 9.626730047895962e-06,
      "loss": 0.6696,
      "step": 1642
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.383793980971439,
      "learning_rate": 9.62616845164362e-06,
      "loss": 0.2039,
      "step": 1643
    },
    {
      "epoch": 0.15,
      "grad_norm": 24.317838006562706,
      "learning_rate": 9.62560644964536e-06,
      "loss": 0.6135,
      "step": 1644
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.593472147877941,
      "learning_rate": 9.625044041950466e-06,
      "loss": 0.5055,
      "step": 1645
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3597339223638432,
      "learning_rate": 9.624481228608272e-06,
      "loss": 0.2097,
      "step": 1646
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.588489772852717,
      "learning_rate": 9.623918009668136e-06,
      "loss": 0.5146,
      "step": 1647
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.55861020597593,
      "learning_rate": 9.623354385179458e-06,
      "loss": 0.2482,
      "step": 1648
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.334423566276226,
      "learning_rate": 9.622790355191672e-06,
      "loss": 0.7785,
      "step": 1649
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3211799280103262,
      "learning_rate": 9.622225919754247e-06,
      "loss": 0.1852,
      "step": 1650
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5842804817000864,
      "learning_rate": 9.621661078916686e-06,
      "loss": 0.1914,
      "step": 1651
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4028547963089166,
      "learning_rate": 9.62109583272853e-06,
      "loss": 0.1923,
      "step": 1652
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5124860142504737,
      "learning_rate": 9.620530181239356e-06,
      "loss": 0.229,
      "step": 1653
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5078185629198702,
      "learning_rate": 9.619964124498775e-06,
      "loss": 0.1911,
      "step": 1654
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3660575150719596,
      "learning_rate": 9.619397662556434e-06,
      "loss": 0.2387,
      "step": 1655
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5012646747642822,
      "learning_rate": 9.618830795462016e-06,
      "loss": 0.2529,
      "step": 1656
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.428578496129174,
      "learning_rate": 9.618263523265238e-06,
      "loss": 0.2144,
      "step": 1657
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5630320271399487,
      "learning_rate": 9.617695846015853e-06,
      "loss": 0.2544,
      "step": 1658
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.458792827552947,
      "learning_rate": 9.617127763763653e-06,
      "loss": 0.256,
      "step": 1659
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6192060758620648,
      "learning_rate": 9.616559276558458e-06,
      "loss": 0.2734,
      "step": 1660
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5803800233680296,
      "learning_rate": 9.615990384450132e-06,
      "loss": 0.2147,
      "step": 1661
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6670656791358216,
      "learning_rate": 9.61542108748857e-06,
      "loss": 0.2326,
      "step": 1662
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.335241064425907,
      "learning_rate": 9.614851385723703e-06,
      "loss": 0.2446,
      "step": 1663
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5079296746057829,
      "learning_rate": 9.614281279205495e-06,
      "loss": 0.2236,
      "step": 1664
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.356947379679163,
      "learning_rate": 9.613710767983953e-06,
      "loss": 0.2456,
      "step": 1665
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.588576495563664,
      "learning_rate": 9.61313985210911e-06,
      "loss": 0.2839,
      "step": 1666
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5593693495391978,
      "learning_rate": 9.612568531631042e-06,
      "loss": 0.2289,
      "step": 1667
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5060195879803264,
      "learning_rate": 9.611996806599858e-06,
      "loss": 0.2281,
      "step": 1668
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.74534950700552,
      "learning_rate": 9.6114246770657e-06,
      "loss": 0.2505,
      "step": 1669
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.566826788255634,
      "learning_rate": 9.610852143078747e-06,
      "loss": 0.2195,
      "step": 1670
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.536878627587062,
      "learning_rate": 9.610279204689218e-06,
      "loss": 0.2166,
      "step": 1671
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.576518200985818,
      "learning_rate": 9.609705861947359e-06,
      "loss": 0.2245,
      "step": 1672
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2777300950684896,
      "learning_rate": 9.609132114903458e-06,
      "loss": 0.1989,
      "step": 1673
    },
    {
      "epoch": 0.15,
      "grad_norm": 8.754397428465243,
      "learning_rate": 9.60855796360784e-06,
      "loss": 0.5689,
      "step": 1674
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4718898756729804,
      "learning_rate": 9.607983408110856e-06,
      "loss": 0.1836,
      "step": 1675
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4737509856042128,
      "learning_rate": 9.607408448462897e-06,
      "loss": 0.218,
      "step": 1676
    },
    {
      "epoch": 0.15,
      "grad_norm": 6.055928925745671,
      "learning_rate": 9.6068330847144e-06,
      "loss": 0.7879,
      "step": 1677
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3942236580894056,
      "learning_rate": 9.606257316915821e-06,
      "loss": 0.2101,
      "step": 1678
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6929523957076655,
      "learning_rate": 9.605681145117662e-06,
      "loss": 0.2428,
      "step": 1679
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3096505016181976,
      "learning_rate": 9.605104569370453e-06,
      "loss": 0.2302,
      "step": 1680
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.4064354537343675,
      "learning_rate": 9.60452758972477e-06,
      "loss": 0.5666,
      "step": 1681
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.625007617143567,
      "learning_rate": 9.603950206231211e-06,
      "loss": 0.7341,
      "step": 1682
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3741984745814113,
      "learning_rate": 9.603372418940419e-06,
      "loss": 0.2548,
      "step": 1683
    },
    {
      "epoch": 0.15,
      "grad_norm": 12.682412101451417,
      "learning_rate": 9.602794227903074e-06,
      "loss": 0.6832,
      "step": 1684
    },
    {
      "epoch": 0.15,
      "grad_norm": 5.688424272472998,
      "learning_rate": 9.602215633169881e-06,
      "loss": 0.619,
      "step": 1685
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3848998484697377,
      "learning_rate": 9.60163663479159e-06,
      "loss": 0.2088,
      "step": 1686
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5506030738911725,
      "learning_rate": 9.601057232818982e-06,
      "loss": 0.2583,
      "step": 1687
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5789600033940787,
      "learning_rate": 9.600477427302876e-06,
      "loss": 0.2253,
      "step": 1688
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5119246610465558,
      "learning_rate": 9.599897218294122e-06,
      "loss": 0.2924,
      "step": 1689
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3524728571556988,
      "learning_rate": 9.599316605843609e-06,
      "loss": 0.2006,
      "step": 1690
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3010897168574613,
      "learning_rate": 9.598735590002264e-06,
      "loss": 0.2586,
      "step": 1691
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.772268556407157,
      "learning_rate": 9.598154170821041e-06,
      "loss": 0.6554,
      "step": 1692
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5898056200942203,
      "learning_rate": 9.597572348350939e-06,
      "loss": 0.2244,
      "step": 1693
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4382789476410769,
      "learning_rate": 9.596990122642984e-06,
      "loss": 0.2059,
      "step": 1694
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4033019129801536,
      "learning_rate": 9.596407493748242e-06,
      "loss": 0.2228,
      "step": 1695
    },
    {
      "epoch": 0.16,
      "grad_norm": 10.086205021247554,
      "learning_rate": 9.595824461717815e-06,
      "loss": 0.7729,
      "step": 1696
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2915928132431373,
      "learning_rate": 9.595241026602836e-06,
      "loss": 0.2164,
      "step": 1697
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6914517602621428,
      "learning_rate": 9.594657188454479e-06,
      "loss": 0.2337,
      "step": 1698
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3550194966557993,
      "learning_rate": 9.59407294732395e-06,
      "loss": 0.2088,
      "step": 1699
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.41278297645253,
      "learning_rate": 9.59348830326249e-06,
      "loss": 0.2229,
      "step": 1700
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4768212918657342,
      "learning_rate": 9.592903256321376e-06,
      "loss": 0.2524,
      "step": 1701
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3663173050923996,
      "learning_rate": 9.592317806551922e-06,
      "loss": 0.2122,
      "step": 1702
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3437209812820747,
      "learning_rate": 9.591731954005475e-06,
      "loss": 0.2077,
      "step": 1703
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3191030051945833,
      "learning_rate": 9.591145698733416e-06,
      "loss": 0.1649,
      "step": 1704
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4463982210927189,
      "learning_rate": 9.590559040787168e-06,
      "loss": 0.2051,
      "step": 1705
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.416082684164785,
      "learning_rate": 9.58997198021818e-06,
      "loss": 0.254,
      "step": 1706
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3316888534540015,
      "learning_rate": 9.589384517077945e-06,
      "loss": 0.1935,
      "step": 1707
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4959853094583024,
      "learning_rate": 9.588796651417987e-06,
      "loss": 0.2937,
      "step": 1708
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4277333634646208,
      "learning_rate": 9.588208383289864e-06,
      "loss": 0.1836,
      "step": 1709
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5709279809719878,
      "learning_rate": 9.58761971274517e-06,
      "loss": 0.2354,
      "step": 1710
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4840259621254646,
      "learning_rate": 9.587030639835538e-06,
      "loss": 0.2767,
      "step": 1711
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5176066850115608,
      "learning_rate": 9.586441164612633e-06,
      "loss": 0.2553,
      "step": 1712
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.359011886393938,
      "learning_rate": 9.585851287128157e-06,
      "loss": 0.1995,
      "step": 1713
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4621332969048462,
      "learning_rate": 9.585261007433844e-06,
      "loss": 0.2138,
      "step": 1714
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3503185595360898,
      "learning_rate": 9.584670325581465e-06,
      "loss": 0.209,
      "step": 1715
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4210948142327642,
      "learning_rate": 9.58407924162283e-06,
      "loss": 0.2439,
      "step": 1716
    },
    {
      "epoch": 0.16,
      "grad_norm": 12.471453375676537,
      "learning_rate": 9.583487755609778e-06,
      "loss": 0.7198,
      "step": 1717
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5210165307926893,
      "learning_rate": 9.582895867594188e-06,
      "loss": 0.1791,
      "step": 1718
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3607368465889638,
      "learning_rate": 9.582303577627972e-06,
      "loss": 0.1911,
      "step": 1719
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4744510667317785,
      "learning_rate": 9.581710885763077e-06,
      "loss": 0.1977,
      "step": 1720
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.314667559285805,
      "learning_rate": 9.581117792051487e-06,
      "loss": 0.6765,
      "step": 1721
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2698605528838194,
      "learning_rate": 9.58052429654522e-06,
      "loss": 0.1637,
      "step": 1722
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5624767265522292,
      "learning_rate": 9.57993039929633e-06,
      "loss": 0.2302,
      "step": 1723
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.388541887864509,
      "learning_rate": 9.579336100356906e-06,
      "loss": 0.2229,
      "step": 1724
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6216429758987436,
      "learning_rate": 9.57874139977907e-06,
      "loss": 0.2735,
      "step": 1725
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5150333076008375,
      "learning_rate": 9.578146297614984e-06,
      "loss": 0.2197,
      "step": 1726
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7598886477461966,
      "learning_rate": 9.577550793916839e-06,
      "loss": 0.2523,
      "step": 1727
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.509813378780454,
      "learning_rate": 9.57695488873687e-06,
      "loss": 0.2421,
      "step": 1728
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6347144654457106,
      "learning_rate": 9.576358582127334e-06,
      "loss": 0.229,
      "step": 1729
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5157082924462812,
      "learning_rate": 9.575761874140538e-06,
      "loss": 0.2073,
      "step": 1730
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.463207371942395,
      "learning_rate": 9.575164764828816e-06,
      "loss": 0.2364,
      "step": 1731
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6354369608807213,
      "learning_rate": 9.574567254244535e-06,
      "loss": 0.19,
      "step": 1732
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.438630188012058,
      "learning_rate": 9.573969342440107e-06,
      "loss": 0.2175,
      "step": 1733
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6774789206914862,
      "learning_rate": 9.573371029467966e-06,
      "loss": 0.2648,
      "step": 1734
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.294950946986401,
      "learning_rate": 9.57277231538059e-06,
      "loss": 0.8724,
      "step": 1735
    },
    {
      "epoch": 0.16,
      "grad_norm": 8.832091778292007,
      "learning_rate": 9.572173200230492e-06,
      "loss": 0.7605,
      "step": 1736
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.2558991458679545,
      "learning_rate": 9.57157368407022e-06,
      "loss": 0.642,
      "step": 1737
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3862982391662204,
      "learning_rate": 9.57097376695235e-06,
      "loss": 0.2379,
      "step": 1738
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.434744957306862,
      "learning_rate": 9.570373448929504e-06,
      "loss": 0.6206,
      "step": 1739
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3353043913594196,
      "learning_rate": 9.569772730054329e-06,
      "loss": 0.2071,
      "step": 1740
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.51483072401552,
      "learning_rate": 9.569171610379518e-06,
      "loss": 0.2347,
      "step": 1741
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3791513892830054,
      "learning_rate": 9.568570089957789e-06,
      "loss": 0.206,
      "step": 1742
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.339595300993974,
      "learning_rate": 9.567968168841901e-06,
      "loss": 0.2143,
      "step": 1743
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6227402728900355,
      "learning_rate": 9.567365847084645e-06,
      "loss": 0.2409,
      "step": 1744
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4793391083207807,
      "learning_rate": 9.56676312473885e-06,
      "loss": 0.2416,
      "step": 1745
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5480949838232494,
      "learning_rate": 9.566160001857378e-06,
      "loss": 0.248,
      "step": 1746
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4101405391641066,
      "learning_rate": 9.56555647849313e-06,
      "loss": 0.1961,
      "step": 1747
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4263902426508175,
      "learning_rate": 9.564952554699034e-06,
      "loss": 0.2568,
      "step": 1748
    },
    {
      "epoch": 0.16,
      "grad_norm": 7.391147644850832,
      "learning_rate": 9.56434823052806e-06,
      "loss": 0.6842,
      "step": 1749
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3838850012611161,
      "learning_rate": 9.563743506033214e-06,
      "loss": 0.1976,
      "step": 1750
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.472461416836681,
      "learning_rate": 9.56313838126753e-06,
      "loss": 0.1871,
      "step": 1751
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5356436938092761,
      "learning_rate": 9.562532856284085e-06,
      "loss": 0.2763,
      "step": 1752
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4618838917284156,
      "learning_rate": 9.561926931135985e-06,
      "loss": 0.1828,
      "step": 1753
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5861074318573951,
      "learning_rate": 9.561320605876376e-06,
      "loss": 0.2737,
      "step": 1754
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2975723842804394,
      "learning_rate": 9.560713880558435e-06,
      "loss": 0.1826,
      "step": 1755
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5748606966399608,
      "learning_rate": 9.560106755235377e-06,
      "loss": 0.2706,
      "step": 1756
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5289721748594354,
      "learning_rate": 9.55949922996045e-06,
      "loss": 0.2354,
      "step": 1757
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3019416777828121,
      "learning_rate": 9.55889130478694e-06,
      "loss": 0.1789,
      "step": 1758
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4622487834256355,
      "learning_rate": 9.558282979768164e-06,
      "loss": 0.246,
      "step": 1759
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4549088036232936,
      "learning_rate": 9.557674254957477e-06,
      "loss": 0.2481,
      "step": 1760
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3626545178530778,
      "learning_rate": 9.557065130408267e-06,
      "loss": 0.232,
      "step": 1761
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.494346035025263,
      "learning_rate": 9.556455606173961e-06,
      "loss": 0.2176,
      "step": 1762
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4840470477226886,
      "learning_rate": 9.555845682308016e-06,
      "loss": 0.2397,
      "step": 1763
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.461546660921417,
      "learning_rate": 9.555235358863928e-06,
      "loss": 0.1827,
      "step": 1764
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4659496126940277,
      "learning_rate": 9.554624635895224e-06,
      "loss": 0.2079,
      "step": 1765
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5189430512567699,
      "learning_rate": 9.554013513455471e-06,
      "loss": 0.1969,
      "step": 1766
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4995221279972353,
      "learning_rate": 9.553401991598268e-06,
      "loss": 0.2375,
      "step": 1767
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.671966738391788,
      "learning_rate": 9.552790070377249e-06,
      "loss": 0.2443,
      "step": 1768
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4861994933040914,
      "learning_rate": 9.552177749846083e-06,
      "loss": 0.2684,
      "step": 1769
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8666518060449773,
      "learning_rate": 9.551565030058476e-06,
      "loss": 0.1824,
      "step": 1770
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3675529225821865,
      "learning_rate": 9.550951911068167e-06,
      "loss": 0.204,
      "step": 1771
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4844193271412005,
      "learning_rate": 9.550338392928931e-06,
      "loss": 0.2132,
      "step": 1772
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4844131919496366,
      "learning_rate": 9.549724475694576e-06,
      "loss": 0.22,
      "step": 1773
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6166074881302341,
      "learning_rate": 9.549110159418948e-06,
      "loss": 0.264,
      "step": 1774
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5391379610799525,
      "learning_rate": 9.548495444155927e-06,
      "loss": 0.2152,
      "step": 1775
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5832027693911936,
      "learning_rate": 9.547880329959429e-06,
      "loss": 0.2362,
      "step": 1776
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2960314694207513,
      "learning_rate": 9.5472648168834e-06,
      "loss": 0.1882,
      "step": 1777
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5709253649526218,
      "learning_rate": 9.546648904981826e-06,
      "loss": 0.2637,
      "step": 1778
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5007495960853918,
      "learning_rate": 9.546032594308728e-06,
      "loss": 0.2691,
      "step": 1779
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.328105546854352,
      "learning_rate": 9.54541588491816e-06,
      "loss": 0.5874,
      "step": 1780
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5554411934777017,
      "learning_rate": 9.544798776864211e-06,
      "loss": 0.2376,
      "step": 1781
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.746508209082215,
      "learning_rate": 9.544181270201005e-06,
      "loss": 0.2848,
      "step": 1782
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5398808995799425,
      "learning_rate": 9.543563364982702e-06,
      "loss": 0.2357,
      "step": 1783
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4208012663930674,
      "learning_rate": 9.542945061263497e-06,
      "loss": 0.2081,
      "step": 1784
    },
    {
      "epoch": 0.16,
      "grad_norm": 5.547447368081283,
      "learning_rate": 9.542326359097619e-06,
      "loss": 0.6716,
      "step": 1785
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.360690365077677,
      "learning_rate": 9.541707258539332e-06,
      "loss": 0.2118,
      "step": 1786
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6540982796589296,
      "learning_rate": 9.541087759642935e-06,
      "loss": 0.2531,
      "step": 1787
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.368876221143342,
      "learning_rate": 9.540467862462765e-06,
      "loss": 0.205,
      "step": 1788
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.547149325555399,
      "learning_rate": 9.539847567053186e-06,
      "loss": 0.2277,
      "step": 1789
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5557557805783482,
      "learning_rate": 9.539226873468606e-06,
      "loss": 0.2581,
      "step": 1790
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4876159466192924,
      "learning_rate": 9.538605781763464e-06,
      "loss": 0.2128,
      "step": 1791
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.51307371659347,
      "learning_rate": 9.537984291992232e-06,
      "loss": 0.2429,
      "step": 1792
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4271891451546066,
      "learning_rate": 9.537362404209419e-06,
      "loss": 0.2285,
      "step": 1793
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.48325296155953,
      "learning_rate": 9.53674011846957e-06,
      "loss": 0.1979,
      "step": 1794
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5210481648310104,
      "learning_rate": 9.536117434827262e-06,
      "loss": 0.2541,
      "step": 1795
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.453540684890752,
      "learning_rate": 9.53549435333711e-06,
      "loss": 0.2364,
      "step": 1796
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.587034390577919,
      "learning_rate": 9.534870874053761e-06,
      "loss": 0.2285,
      "step": 1797
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.471158277917216,
      "learning_rate": 9.534246997031901e-06,
      "loss": 0.2339,
      "step": 1798
    },
    {
      "epoch": 0.16,
      "grad_norm": 6.775926640795709,
      "learning_rate": 9.533622722326246e-06,
      "loss": 0.5968,
      "step": 1799
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4051242209000123,
      "learning_rate": 9.532998049991547e-06,
      "loss": 0.2146,
      "step": 1800
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.484183481639132,
      "learning_rate": 9.532372980082598e-06,
      "loss": 0.2137,
      "step": 1801
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4883950751466977,
      "learning_rate": 9.531747512654215e-06,
      "loss": 0.2022,
      "step": 1802
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4703915802428769,
      "learning_rate": 9.531121647761262e-06,
      "loss": 0.2573,
      "step": 1803
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5200537112041415,
      "learning_rate": 9.530495385458625e-06,
      "loss": 0.2527,
      "step": 1804
    },
    {
      "epoch": 0.17,
      "grad_norm": 4.730634619102503,
      "learning_rate": 9.529868725801237e-06,
      "loss": 0.4678,
      "step": 1805
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4840842263122411,
      "learning_rate": 9.529241668844057e-06,
      "loss": 0.1981,
      "step": 1806
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6214873458175263,
      "learning_rate": 9.528614214642083e-06,
      "loss": 0.2458,
      "step": 1807
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2913123888453046,
      "learning_rate": 9.527986363250348e-06,
      "loss": 0.2041,
      "step": 1808
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4041659311238452,
      "learning_rate": 9.527358114723917e-06,
      "loss": 0.2384,
      "step": 1809
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.844511808629284,
      "learning_rate": 9.526729469117892e-06,
      "loss": 0.5478,
      "step": 1810
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4590683200016028,
      "learning_rate": 9.52610042648741e-06,
      "loss": 0.239,
      "step": 1811
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5737531960563351,
      "learning_rate": 9.525470986887643e-06,
      "loss": 0.2527,
      "step": 1812
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4006986643072987,
      "learning_rate": 9.524841150373796e-06,
      "loss": 0.2226,
      "step": 1813
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2316538649254696,
      "learning_rate": 9.524210917001111e-06,
      "loss": 0.2135,
      "step": 1814
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6361088647288142,
      "learning_rate": 9.52358028682486e-06,
      "loss": 0.2414,
      "step": 1815
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.634050878478791,
      "learning_rate": 9.522949259900358e-06,
      "loss": 0.2064,
      "step": 1816
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5691208591481725,
      "learning_rate": 9.522317836282949e-06,
      "loss": 0.2075,
      "step": 1817
    },
    {
      "epoch": 0.17,
      "grad_norm": 8.999421623889686,
      "learning_rate": 9.52168601602801e-06,
      "loss": 0.6285,
      "step": 1818
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3624775356948762,
      "learning_rate": 9.521053799190962e-06,
      "loss": 0.2449,
      "step": 1819
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5946254673480211,
      "learning_rate": 9.520421185827248e-06,
      "loss": 0.2507,
      "step": 1820
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.022191216394653,
      "learning_rate": 9.519788175992357e-06,
      "loss": 0.5637,
      "step": 1821
    },
    {
      "epoch": 0.17,
      "grad_norm": 11.269404865042128,
      "learning_rate": 9.519154769741807e-06,
      "loss": 0.7342,
      "step": 1822
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6470203723767796,
      "learning_rate": 9.51852096713115e-06,
      "loss": 0.2882,
      "step": 1823
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3685709304108729,
      "learning_rate": 9.517886768215978e-06,
      "loss": 0.2078,
      "step": 1824
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.434592914817289,
      "learning_rate": 9.517252173051912e-06,
      "loss": 0.4653,
      "step": 1825
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.415243637892122,
      "learning_rate": 9.516617181694611e-06,
      "loss": 0.232,
      "step": 1826
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.991819776004684,
      "learning_rate": 9.515981794199768e-06,
      "loss": 0.629,
      "step": 1827
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3073317900647117,
      "learning_rate": 9.515346010623112e-06,
      "loss": 0.239,
      "step": 1828
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6074882378941866,
      "learning_rate": 9.514709831020403e-06,
      "loss": 0.2225,
      "step": 1829
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.443392370986208,
      "learning_rate": 9.51407325544744e-06,
      "loss": 0.2151,
      "step": 1830
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.365295308923036,
      "learning_rate": 9.513436283960057e-06,
      "loss": 0.2491,
      "step": 1831
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.647701757458352,
      "learning_rate": 9.512798916614115e-06,
      "loss": 0.648,
      "step": 1832
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3766804136926998,
      "learning_rate": 9.512161153465518e-06,
      "loss": 0.2091,
      "step": 1833
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.313473982658426,
      "learning_rate": 9.511522994570204e-06,
      "loss": 0.221,
      "step": 1834
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5445963719898352,
      "learning_rate": 9.510884439984141e-06,
      "loss": 0.2503,
      "step": 1835
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.37804349923222,
      "learning_rate": 9.510245489763338e-06,
      "loss": 0.2103,
      "step": 1836
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.080335531245107,
      "learning_rate": 9.509606143963832e-06,
      "loss": 0.6794,
      "step": 1837
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.307500889704778,
      "learning_rate": 9.5089664026417e-06,
      "loss": 0.6574,
      "step": 1838
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4395843737500356,
      "learning_rate": 9.508326265853049e-06,
      "loss": 0.2385,
      "step": 1839
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.508282214547347,
      "learning_rate": 9.507685733654024e-06,
      "loss": 0.2435,
      "step": 1840
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8950554500504335,
      "learning_rate": 9.507044806100806e-06,
      "loss": 0.2281,
      "step": 1841
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.0919515008549405,
      "learning_rate": 9.506403483249609e-06,
      "loss": 0.4225,
      "step": 1842
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5533730007866795,
      "learning_rate": 9.505761765156679e-06,
      "loss": 0.217,
      "step": 1843
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4366866865675842,
      "learning_rate": 9.505119651878298e-06,
      "loss": 0.2303,
      "step": 1844
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.972079007011514,
      "learning_rate": 9.504477143470787e-06,
      "loss": 0.5988,
      "step": 1845
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5241396480223073,
      "learning_rate": 9.503834239990497e-06,
      "loss": 0.243,
      "step": 1846
    },
    {
      "epoch": 0.17,
      "grad_norm": 9.208617432525031,
      "learning_rate": 9.503190941493814e-06,
      "loss": 0.7426,
      "step": 1847
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6038113278934383,
      "learning_rate": 9.502547248037159e-06,
      "loss": 0.2432,
      "step": 1848
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6472547306346392,
      "learning_rate": 9.501903159676993e-06,
      "loss": 0.2733,
      "step": 1849
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.369274872338326,
      "learning_rate": 9.5012586764698e-06,
      "loss": 0.2007,
      "step": 1850
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5631170021116667,
      "learning_rate": 9.500613798472109e-06,
      "loss": 0.2045,
      "step": 1851
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4024422850950533,
      "learning_rate": 9.49996852574048e-06,
      "loss": 0.2137,
      "step": 1852
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3917299295877825,
      "learning_rate": 9.499322858331507e-06,
      "loss": 0.2385,
      "step": 1853
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5127525768722168,
      "learning_rate": 9.498676796301822e-06,
      "loss": 0.2581,
      "step": 1854
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.722034109037352,
      "learning_rate": 9.498030339708086e-06,
      "loss": 0.7753,
      "step": 1855
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4800099777629894,
      "learning_rate": 9.497383488606998e-06,
      "loss": 0.2426,
      "step": 1856
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5778327778437071,
      "learning_rate": 9.496736243055293e-06,
      "loss": 0.282,
      "step": 1857
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5287417648809225,
      "learning_rate": 9.496088603109735e-06,
      "loss": 0.2594,
      "step": 1858
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3899120746593208,
      "learning_rate": 9.49544056882713e-06,
      "loss": 0.2277,
      "step": 1859
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5797588554316968,
      "learning_rate": 9.494792140264315e-06,
      "loss": 0.2497,
      "step": 1860
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5984399083216567,
      "learning_rate": 9.49414331747816e-06,
      "loss": 0.22,
      "step": 1861
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6666836564446217,
      "learning_rate": 9.493494100525569e-06,
      "loss": 0.2731,
      "step": 1862
    },
    {
      "epoch": 0.17,
      "grad_norm": 16.484243353703352,
      "learning_rate": 9.492844489463486e-06,
      "loss": 0.5353,
      "step": 1863
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.498231197056075,
      "learning_rate": 9.492194484348886e-06,
      "loss": 0.1982,
      "step": 1864
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.213797970984464,
      "learning_rate": 9.491544085238778e-06,
      "loss": 0.7117,
      "step": 1865
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4152940431081804,
      "learning_rate": 9.490893292190205e-06,
      "loss": 0.2204,
      "step": 1866
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.411153672991245,
      "learning_rate": 9.490242105260248e-06,
      "loss": 0.2,
      "step": 1867
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.078133601817216,
      "learning_rate": 9.48959052450602e-06,
      "loss": 0.5724,
      "step": 1868
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6696497783321413,
      "learning_rate": 9.48893854998467e-06,
      "loss": 0.2281,
      "step": 1869
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7527930516547903,
      "learning_rate": 9.488286181753377e-06,
      "loss": 0.2304,
      "step": 1870
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5212090348552283,
      "learning_rate": 9.487633419869362e-06,
      "loss": 0.2137,
      "step": 1871
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.244431219361301,
      "learning_rate": 9.486980264389875e-06,
      "loss": 0.2173,
      "step": 1872
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.567586313125447,
      "learning_rate": 9.486326715372201e-06,
      "loss": 0.6466,
      "step": 1873
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.404162025320571,
      "learning_rate": 9.485672772873662e-06,
      "loss": 0.2348,
      "step": 1874
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3668027839670567,
      "learning_rate": 9.485018436951611e-06,
      "loss": 0.2378,
      "step": 1875
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.453240387380786,
      "learning_rate": 9.484363707663443e-06,
      "loss": 0.2275,
      "step": 1876
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3643294160018173,
      "learning_rate": 9.483708585066576e-06,
      "loss": 0.2361,
      "step": 1877
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2223364588081413,
      "learning_rate": 9.48305306921847e-06,
      "loss": 0.2124,
      "step": 1878
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5520986812132804,
      "learning_rate": 9.482397160176623e-06,
      "loss": 0.1872,
      "step": 1879
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5103562765290093,
      "learning_rate": 9.481740857998557e-06,
      "loss": 0.2361,
      "step": 1880
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3506359892798687,
      "learning_rate": 9.481084162741835e-06,
      "loss": 0.2413,
      "step": 1881
    },
    {
      "epoch": 0.17,
      "grad_norm": 7.4498586493992205,
      "learning_rate": 9.480427074464056e-06,
      "loss": 0.7785,
      "step": 1882
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.479875281604057,
      "learning_rate": 9.479769593222849e-06,
      "loss": 0.5474,
      "step": 1883
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.490833208806634,
      "learning_rate": 9.47911171907588e-06,
      "loss": 0.7915,
      "step": 1884
    },
    {
      "epoch": 0.17,
      "grad_norm": 6.928013765075317,
      "learning_rate": 9.47845345208085e-06,
      "loss": 0.6336,
      "step": 1885
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5147213367710604,
      "learning_rate": 9.477794792295492e-06,
      "loss": 0.2463,
      "step": 1886
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.439520693699015,
      "learning_rate": 9.477135739777575e-06,
      "loss": 0.1946,
      "step": 1887
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5917705502932207,
      "learning_rate": 9.476476294584902e-06,
      "loss": 0.2379,
      "step": 1888
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.68903647359145,
      "learning_rate": 9.475816456775313e-06,
      "loss": 0.2486,
      "step": 1889
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.525000325667257,
      "learning_rate": 9.475156226406677e-06,
      "loss": 0.2463,
      "step": 1890
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.409866139505887,
      "learning_rate": 9.474495603536904e-06,
      "loss": 0.2259,
      "step": 1891
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4176890149236505,
      "learning_rate": 9.473834588223931e-06,
      "loss": 0.2295,
      "step": 1892
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5469317255304873,
      "learning_rate": 9.473173180525737e-06,
      "loss": 0.2627,
      "step": 1893
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2704902245136258,
      "learning_rate": 9.47251138050033e-06,
      "loss": 0.1901,
      "step": 1894
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3833558549778973,
      "learning_rate": 9.471849188205758e-06,
      "loss": 0.2087,
      "step": 1895
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4714612112998962,
      "learning_rate": 9.471186603700094e-06,
      "loss": 0.242,
      "step": 1896
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.515364420274641,
      "learning_rate": 9.470523627041452e-06,
      "loss": 0.2257,
      "step": 1897
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.460065129107397,
      "learning_rate": 9.469860258287984e-06,
      "loss": 0.2091,
      "step": 1898
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5333190721684633,
      "learning_rate": 9.469196497497868e-06,
      "loss": 0.1643,
      "step": 1899
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5843949373599224,
      "learning_rate": 9.468532344729321e-06,
      "loss": 0.2448,
      "step": 1900
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3598069559719503,
      "learning_rate": 9.467867800040595e-06,
      "loss": 0.2288,
      "step": 1901
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5136680033853138,
      "learning_rate": 9.467202863489974e-06,
      "loss": 0.2114,
      "step": 1902
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.491913047893721,
      "learning_rate": 9.466537535135776e-06,
      "loss": 0.2224,
      "step": 1903
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.531567588646997,
      "learning_rate": 9.465871815036358e-06,
      "loss": 0.2308,
      "step": 1904
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.164180950841117,
      "learning_rate": 9.465205703250105e-06,
      "loss": 0.4424,
      "step": 1905
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3607881786295053,
      "learning_rate": 9.46453919983544e-06,
      "loss": 0.1936,
      "step": 1906
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2671667727905802,
      "learning_rate": 9.463872304850821e-06,
      "loss": 0.1778,
      "step": 1907
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3779728328686611,
      "learning_rate": 9.46320501835474e-06,
      "loss": 0.1913,
      "step": 1908
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3886716228128781,
      "learning_rate": 9.46253734040572e-06,
      "loss": 0.2432,
      "step": 1909
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2570302821158104,
      "learning_rate": 9.461869271062322e-06,
      "loss": 0.2234,
      "step": 1910
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3641715021794858,
      "learning_rate": 9.46120081038314e-06,
      "loss": 0.2317,
      "step": 1911
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3373579806919775,
      "learning_rate": 9.460531958426803e-06,
      "loss": 0.1871,
      "step": 1912
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3964672173739268,
      "learning_rate": 9.459862715251973e-06,
      "loss": 0.1963,
      "step": 1913
    },
    {
      "epoch": 0.17,
      "grad_norm": 5.251351105850906,
      "learning_rate": 9.459193080917349e-06,
      "loss": 0.6709,
      "step": 1914
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5354434946875375,
      "learning_rate": 9.458523055481658e-06,
      "loss": 0.2265,
      "step": 1915
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5302418071238362,
      "learning_rate": 9.457852639003671e-06,
      "loss": 0.2195,
      "step": 1916
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.834734891366557,
      "learning_rate": 9.457181831542183e-06,
      "loss": 0.2373,
      "step": 1917
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.745580492854764,
      "learning_rate": 9.456510633156033e-06,
      "loss": 0.6598,
      "step": 1918
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5773146849864954,
      "learning_rate": 9.455839043904088e-06,
      "loss": 0.2179,
      "step": 1919
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3677250821886275,
      "learning_rate": 9.45516706384525e-06,
      "loss": 0.2222,
      "step": 1920
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3294839698719094,
      "learning_rate": 9.454494693038455e-06,
      "loss": 0.1975,
      "step": 1921
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4660782966991526,
      "learning_rate": 9.453821931542678e-06,
      "loss": 0.2247,
      "step": 1922
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.680239121384478,
      "learning_rate": 9.45314877941692e-06,
      "loss": 0.2444,
      "step": 1923
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5487657667792198,
      "learning_rate": 9.452475236720226e-06,
      "loss": 0.2162,
      "step": 1924
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.60223328803334,
      "learning_rate": 9.451801303511666e-06,
      "loss": 0.2999,
      "step": 1925
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4951801305811736,
      "learning_rate": 9.45112697985035e-06,
      "loss": 0.2138,
      "step": 1926
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3305163864516092,
      "learning_rate": 9.450452265795423e-06,
      "loss": 0.1874,
      "step": 1927
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.974421849259398,
      "learning_rate": 9.44977716140606e-06,
      "loss": 0.7397,
      "step": 1928
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6899935649146556,
      "learning_rate": 9.44910166674147e-06,
      "loss": 0.2882,
      "step": 1929
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3541840413481732,
      "learning_rate": 9.448425781860902e-06,
      "loss": 0.1735,
      "step": 1930
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.426569694215913,
      "learning_rate": 9.447749506823633e-06,
      "loss": 0.2089,
      "step": 1931
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6395133121925516,
      "learning_rate": 9.447072841688977e-06,
      "loss": 0.2712,
      "step": 1932
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.438828353082915,
      "learning_rate": 9.446395786516285e-06,
      "loss": 0.2474,
      "step": 1933
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3819560671837992,
      "learning_rate": 9.445718341364936e-06,
      "loss": 0.2064,
      "step": 1934
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.552254642870175,
      "learning_rate": 9.445040506294347e-06,
      "loss": 0.5362,
      "step": 1935
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5771020538607436,
      "learning_rate": 9.44436228136397e-06,
      "loss": 0.2641,
      "step": 1936
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.416982007576552,
      "learning_rate": 9.44368366663329e-06,
      "loss": 0.1943,
      "step": 1937
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5944203159559822,
      "learning_rate": 9.443004662161825e-06,
      "loss": 0.2392,
      "step": 1938
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.664335093877917,
      "learning_rate": 9.44232526800913e-06,
      "loss": 0.255,
      "step": 1939
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.652304857772123,
      "learning_rate": 9.441645484234788e-06,
      "loss": 0.5377,
      "step": 1940
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.513408104362655,
      "learning_rate": 9.440965310898425e-06,
      "loss": 0.6179,
      "step": 1941
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3863392618171924,
      "learning_rate": 9.440284748059697e-06,
      "loss": 0.2038,
      "step": 1942
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3955437447239682,
      "learning_rate": 9.439603795778287e-06,
      "loss": 0.2365,
      "step": 1943
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5501715500628275,
      "learning_rate": 9.438922454113928e-06,
      "loss": 0.2062,
      "step": 1944
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5124966443134196,
      "learning_rate": 9.438240723126376e-06,
      "loss": 0.2112,
      "step": 1945
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5097260238493997,
      "learning_rate": 9.437558602875418e-06,
      "loss": 0.2387,
      "step": 1946
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5881182837503829,
      "learning_rate": 9.436876093420885e-06,
      "loss": 0.2498,
      "step": 1947
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4999531188406645,
      "learning_rate": 9.436193194822638e-06,
      "loss": 0.1856,
      "step": 1948
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.3155774124073885,
      "learning_rate": 9.435509907140571e-06,
      "loss": 0.6138,
      "step": 1949
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5246795127768844,
      "learning_rate": 9.434826230434615e-06,
      "loss": 0.2439,
      "step": 1950
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4768804574114238,
      "learning_rate": 9.434142164764727e-06,
      "loss": 0.2391,
      "step": 1951
    },
    {
      "epoch": 0.18,
      "grad_norm": 20.31062854463794,
      "learning_rate": 9.433457710190908e-06,
      "loss": 0.615,
      "step": 1952
    },
    {
      "epoch": 0.18,
      "grad_norm": 8.174200299270654,
      "learning_rate": 9.43277286677319e-06,
      "loss": 0.7555,
      "step": 1953
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4779372823511412,
      "learning_rate": 9.432087634571638e-06,
      "loss": 0.2373,
      "step": 1954
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6731689422760563,
      "learning_rate": 9.431402013646351e-06,
      "loss": 0.2503,
      "step": 1955
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4391727455250904,
      "learning_rate": 9.43071600405746e-06,
      "loss": 0.2576,
      "step": 1956
    },
    {
      "epoch": 0.18,
      "grad_norm": 4.421627270122378,
      "learning_rate": 9.430029605865137e-06,
      "loss": 0.4287,
      "step": 1957
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3904195134902184,
      "learning_rate": 9.429342819129583e-06,
      "loss": 0.1891,
      "step": 1958
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.388934086371612,
      "learning_rate": 9.42865564391103e-06,
      "loss": 0.7388,
      "step": 1959
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4687593143883821,
      "learning_rate": 9.42796808026975e-06,
      "loss": 0.2126,
      "step": 1960
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.514220048360977,
      "learning_rate": 9.427280128266049e-06,
      "loss": 0.1958,
      "step": 1961
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.454862528250326,
      "learning_rate": 9.426591787960264e-06,
      "loss": 0.2641,
      "step": 1962
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3754723693335553,
      "learning_rate": 9.425903059412765e-06,
      "loss": 0.2117,
      "step": 1963
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.7038201142875185,
      "learning_rate": 9.425213942683961e-06,
      "loss": 0.7892,
      "step": 1964
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4626777405670561,
      "learning_rate": 9.424524437834288e-06,
      "loss": 0.2217,
      "step": 1965
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4913945108342086,
      "learning_rate": 9.423834544924226e-06,
      "loss": 0.259,
      "step": 1966
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5307444240231316,
      "learning_rate": 9.423144264014278e-06,
      "loss": 0.232,
      "step": 1967
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5092603835761338,
      "learning_rate": 9.422453595164988e-06,
      "loss": 0.2514,
      "step": 1968
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5777745093400999,
      "learning_rate": 9.421762538436933e-06,
      "loss": 0.2211,
      "step": 1969
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.368395241907501,
      "learning_rate": 9.421071093890723e-06,
      "loss": 0.2271,
      "step": 1970
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4137222956184876,
      "learning_rate": 9.420379261587003e-06,
      "loss": 0.2293,
      "step": 1971
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4967153729883045,
      "learning_rate": 9.41968704158645e-06,
      "loss": 0.2162,
      "step": 1972
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.41922508028487,
      "learning_rate": 9.418994433949776e-06,
      "loss": 0.2476,
      "step": 1973
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4508375995368958,
      "learning_rate": 9.418301438737729e-06,
      "loss": 0.2235,
      "step": 1974
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4630261232441553,
      "learning_rate": 9.417608056011087e-06,
      "loss": 0.2641,
      "step": 1975
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5719272794302575,
      "learning_rate": 9.416914285830668e-06,
      "loss": 0.2197,
      "step": 1976
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4880551227241139,
      "learning_rate": 9.416220128257317e-06,
      "loss": 0.1994,
      "step": 1977
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4023030481455685,
      "learning_rate": 9.415525583351916e-06,
      "loss": 0.2335,
      "step": 1978
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4206553097700263,
      "learning_rate": 9.414830651175384e-06,
      "loss": 0.2078,
      "step": 1979
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4404520549421995,
      "learning_rate": 9.41413533178867e-06,
      "loss": 0.2684,
      "step": 1980
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5596505526091518,
      "learning_rate": 9.413439625252758e-06,
      "loss": 0.2152,
      "step": 1981
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3727663697938663,
      "learning_rate": 9.412743531628665e-06,
      "loss": 0.2039,
      "step": 1982
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.975106095353146,
      "learning_rate": 9.412047050977444e-06,
      "loss": 0.7616,
      "step": 1983
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.644231552087236,
      "learning_rate": 9.411350183360181e-06,
      "loss": 0.2223,
      "step": 1984
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.511658004174918,
      "learning_rate": 9.410652928837998e-06,
      "loss": 0.2156,
      "step": 1985
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.2305082031487204,
      "learning_rate": 9.409955287472045e-06,
      "loss": 0.8001,
      "step": 1986
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3897475487151265,
      "learning_rate": 9.409257259323511e-06,
      "loss": 0.2152,
      "step": 1987
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5053586516392101,
      "learning_rate": 9.408558844453622e-06,
      "loss": 0.2517,
      "step": 1988
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2287618012162544,
      "learning_rate": 9.407860042923627e-06,
      "loss": 0.219,
      "step": 1989
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5512250570356292,
      "learning_rate": 9.407160854794821e-06,
      "loss": 0.2318,
      "step": 1990
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.739564112823615,
      "learning_rate": 9.406461280128523e-06,
      "loss": 0.517,
      "step": 1991
    },
    {
      "epoch": 0.18,
      "grad_norm": 9.698531503191084,
      "learning_rate": 9.405761318986095e-06,
      "loss": 0.7778,
      "step": 1992
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5035824668293731,
      "learning_rate": 9.405060971428924e-06,
      "loss": 0.2171,
      "step": 1993
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4601013017996334,
      "learning_rate": 9.404360237518438e-06,
      "loss": 0.2287,
      "step": 1994
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4657930832148522,
      "learning_rate": 9.403659117316093e-06,
      "loss": 0.231,
      "step": 1995
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3865881512715101,
      "learning_rate": 9.402957610883386e-06,
      "loss": 0.2122,
      "step": 1996
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5275358673570993,
      "learning_rate": 9.40225571828184e-06,
      "loss": 0.3006,
      "step": 1997
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4245932859509307,
      "learning_rate": 9.40155343957302e-06,
      "loss": 0.1955,
      "step": 1998
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4949230019569737,
      "learning_rate": 9.400850774818518e-06,
      "loss": 0.2249,
      "step": 1999
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.442621878222878,
      "learning_rate": 9.40014772407996e-06,
      "loss": 0.176,
      "step": 2000
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.555262586809395,
      "learning_rate": 9.399444287419012e-06,
      "loss": 0.2464,
      "step": 2001
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3725084671226389,
      "learning_rate": 9.398740464897368e-06,
      "loss": 0.2081,
      "step": 2002
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4370842734551497,
      "learning_rate": 9.39803625657676e-06,
      "loss": 0.2318,
      "step": 2003
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.532789341432881,
      "learning_rate": 9.39733166251895e-06,
      "loss": 0.1885,
      "step": 2004
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.579481484012104,
      "learning_rate": 9.396626682785736e-06,
      "loss": 0.2504,
      "step": 2005
    },
    {
      "epoch": 0.18,
      "grad_norm": 5.551532997278856,
      "learning_rate": 9.39592131743895e-06,
      "loss": 0.6397,
      "step": 2006
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4506041004580283,
      "learning_rate": 9.395215566540457e-06,
      "loss": 0.2356,
      "step": 2007
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4463965830596361,
      "learning_rate": 9.394509430152154e-06,
      "loss": 0.2456,
      "step": 2008
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7038259022300903,
      "learning_rate": 9.393802908335978e-06,
      "loss": 0.2449,
      "step": 2009
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7291264567696967,
      "learning_rate": 9.393096001153892e-06,
      "loss": 0.2645,
      "step": 2010
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4865938752462426,
      "learning_rate": 9.392388708667898e-06,
      "loss": 0.2026,
      "step": 2011
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4736707040976182,
      "learning_rate": 9.391681030940031e-06,
      "loss": 0.2268,
      "step": 2012
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4482982232110473,
      "learning_rate": 9.390972968032356e-06,
      "loss": 0.2523,
      "step": 2013
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4256158003927537,
      "learning_rate": 9.390264520006981e-06,
      "loss": 0.1976,
      "step": 2014
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5425954092795529,
      "learning_rate": 9.389555686926035e-06,
      "loss": 0.2215,
      "step": 2015
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4560435311607598,
      "learning_rate": 9.388846468851693e-06,
      "loss": 0.236,
      "step": 2016
    },
    {
      "epoch": 0.18,
      "grad_norm": 7.933119060196217,
      "learning_rate": 9.388136865846153e-06,
      "loss": 0.676,
      "step": 2017
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8864013025494077,
      "learning_rate": 9.387426877971654e-06,
      "loss": 0.1989,
      "step": 2018
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4608016883127088,
      "learning_rate": 9.386716505290467e-06,
      "loss": 0.2527,
      "step": 2019
    },
    {
      "epoch": 0.18,
      "grad_norm": 3.2628395573770868,
      "learning_rate": 9.3860057478649e-06,
      "loss": 0.6768,
      "step": 2020
    },
    {
      "epoch": 0.18,
      "grad_norm": 6.5457133902484195,
      "learning_rate": 9.385294605757283e-06,
      "loss": 0.5292,
      "step": 2021
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2335684032936989,
      "learning_rate": 9.384583079029993e-06,
      "loss": 0.1778,
      "step": 2022
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4523720115495629,
      "learning_rate": 9.383871167745438e-06,
      "loss": 0.2451,
      "step": 2023
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6061182835442407,
      "learning_rate": 9.383158871966054e-06,
      "loss": 0.2962,
      "step": 2024
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.588610979894669,
      "learning_rate": 9.382446191754313e-06,
      "loss": 0.2193,
      "step": 2025
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5984068063299823,
      "learning_rate": 9.381733127172723e-06,
      "loss": 0.2957,
      "step": 2026
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.531104794444474,
      "learning_rate": 9.381019678283827e-06,
      "loss": 0.2022,
      "step": 2027
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.510465735920445,
      "learning_rate": 9.380305845150197e-06,
      "loss": 0.2445,
      "step": 2028
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.278943012150109,
      "learning_rate": 9.37959162783444e-06,
      "loss": 0.2007,
      "step": 2029
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.597482956276203,
      "learning_rate": 9.3788770263992e-06,
      "loss": 0.2015,
      "step": 2030
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.482303515919779,
      "learning_rate": 9.37816204090715e-06,
      "loss": 0.2382,
      "step": 2031
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.381804273396025,
      "learning_rate": 9.377446671421001e-06,
      "loss": 0.1944,
      "step": 2032
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.152276790952037,
      "learning_rate": 9.376730918003495e-06,
      "loss": 0.7465,
      "step": 2033
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.324152795315225,
      "learning_rate": 9.376014780717408e-06,
      "loss": 0.7593,
      "step": 2034
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.33192056761173,
      "learning_rate": 9.375298259625549e-06,
      "loss": 0.2013,
      "step": 2035
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.650347014868354,
      "learning_rate": 9.374581354790764e-06,
      "loss": 0.2087,
      "step": 2036
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3127900602915021,
      "learning_rate": 9.37386406627593e-06,
      "loss": 0.2088,
      "step": 2037
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4672954458761187,
      "learning_rate": 9.373146394143957e-06,
      "loss": 0.2398,
      "step": 2038
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2195171665345368,
      "learning_rate": 9.37242833845779e-06,
      "loss": 0.1613,
      "step": 2039
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.458583269188955,
      "learning_rate": 9.371709899280406e-06,
      "loss": 0.1705,
      "step": 2040
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.183577233518575,
      "learning_rate": 9.370991076674821e-06,
      "loss": 0.538,
      "step": 2041
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.558190165615435,
      "learning_rate": 9.370271870704074e-06,
      "loss": 0.2171,
      "step": 2042
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4937014153954165,
      "learning_rate": 9.369552281431248e-06,
      "loss": 0.2608,
      "step": 2043
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.874274380898526,
      "learning_rate": 9.36883230891946e-06,
      "loss": 0.5508,
      "step": 2044
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.55227002561057,
      "learning_rate": 9.368111953231849e-06,
      "loss": 0.2257,
      "step": 2045
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4063959390245528,
      "learning_rate": 9.367391214431598e-06,
      "loss": 0.21,
      "step": 2046
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4374571430724283,
      "learning_rate": 9.366670092581922e-06,
      "loss": 0.1983,
      "step": 2047
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.320917382688473,
      "learning_rate": 9.365948587746065e-06,
      "loss": 0.6571,
      "step": 2048
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.597758522916908,
      "learning_rate": 9.36522669998731e-06,
      "loss": 0.2153,
      "step": 2049
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4513899872055627,
      "learning_rate": 9.364504429368972e-06,
      "loss": 0.2433,
      "step": 2050
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5324123708114017,
      "learning_rate": 9.363781775954399e-06,
      "loss": 0.259,
      "step": 2051
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6584282466947038,
      "learning_rate": 9.36305873980697e-06,
      "loss": 0.2479,
      "step": 2052
    },
    {
      "epoch": 0.19,
      "grad_norm": 8.661849462778147,
      "learning_rate": 9.362335320990104e-06,
      "loss": 0.7946,
      "step": 2053
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4391505585260618,
      "learning_rate": 9.361611519567244e-06,
      "loss": 0.1916,
      "step": 2054
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.506808776648301,
      "learning_rate": 9.360887335601879e-06,
      "loss": 0.2621,
      "step": 2055
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.35233633743219,
      "learning_rate": 9.360162769157517e-06,
      "loss": 0.2091,
      "step": 2056
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4248250177102058,
      "learning_rate": 9.359437820297716e-06,
      "loss": 0.2514,
      "step": 2057
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.29058065265366,
      "learning_rate": 9.358712489086053e-06,
      "loss": 0.2308,
      "step": 2058
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3925023115154256,
      "learning_rate": 9.357986775586146e-06,
      "loss": 0.1959,
      "step": 2059
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.458746832658592,
      "learning_rate": 9.357260679861644e-06,
      "loss": 0.6976,
      "step": 2060
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5030903032628133,
      "learning_rate": 9.356534201976233e-06,
      "loss": 0.1996,
      "step": 2061
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.459957601108532,
      "learning_rate": 9.355807341993629e-06,
      "loss": 0.2591,
      "step": 2062
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4470237634470513,
      "learning_rate": 9.355080099977579e-06,
      "loss": 0.1994,
      "step": 2063
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5557719948004098,
      "learning_rate": 9.35435247599187e-06,
      "loss": 0.2098,
      "step": 2064
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.284447892125381,
      "learning_rate": 9.353624470100321e-06,
      "loss": 0.2203,
      "step": 2065
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3645599529982246,
      "learning_rate": 9.352896082366782e-06,
      "loss": 0.2336,
      "step": 2066
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5990422505366588,
      "learning_rate": 9.352167312855134e-06,
      "loss": 0.2337,
      "step": 2067
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4707012875971532,
      "learning_rate": 9.351438161629301e-06,
      "loss": 0.2162,
      "step": 2068
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4985281297257664,
      "learning_rate": 9.350708628753228e-06,
      "loss": 0.2539,
      "step": 2069
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.6133046650380445,
      "learning_rate": 9.349978714290904e-06,
      "loss": 0.627,
      "step": 2070
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3970575549182545,
      "learning_rate": 9.349248418306347e-06,
      "loss": 0.2283,
      "step": 2071
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4394280940605495,
      "learning_rate": 9.348517740863608e-06,
      "loss": 0.2207,
      "step": 2072
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4404022424195528,
      "learning_rate": 9.347786682026774e-06,
      "loss": 0.2175,
      "step": 2073
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4948032448535749,
      "learning_rate": 9.34705524185996e-06,
      "loss": 0.2289,
      "step": 2074
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.1547990351942,
      "learning_rate": 9.346323420427323e-06,
      "loss": 0.6068,
      "step": 2075
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5740385697457966,
      "learning_rate": 9.345591217793044e-06,
      "loss": 0.2702,
      "step": 2076
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.437017903524951,
      "learning_rate": 9.344858634021345e-06,
      "loss": 0.2306,
      "step": 2077
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.116624479714142,
      "learning_rate": 9.344125669176478e-06,
      "loss": 0.6348,
      "step": 2078
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.915657855574931,
      "learning_rate": 9.34339232332273e-06,
      "loss": 0.7532,
      "step": 2079
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3911402854764396,
      "learning_rate": 9.342658596524416e-06,
      "loss": 0.2393,
      "step": 2080
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.498774423492753,
      "learning_rate": 9.341924488845892e-06,
      "loss": 0.2671,
      "step": 2081
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5163968632870466,
      "learning_rate": 9.341190000351546e-06,
      "loss": 0.2653,
      "step": 2082
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3239893945381143,
      "learning_rate": 9.340455131105796e-06,
      "loss": 0.2121,
      "step": 2083
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6366537883877903,
      "learning_rate": 9.339719881173093e-06,
      "loss": 0.2169,
      "step": 2084
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.559788067882474,
      "learning_rate": 9.338984250617924e-06,
      "loss": 0.2621,
      "step": 2085
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.05352679592871,
      "learning_rate": 9.33824823950481e-06,
      "loss": 0.5141,
      "step": 2086
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.459008404146402,
      "learning_rate": 9.337511847898303e-06,
      "loss": 0.2407,
      "step": 2087
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.813767338252843,
      "learning_rate": 9.336775075862991e-06,
      "loss": 0.6289,
      "step": 2088
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.062381348316698,
      "learning_rate": 9.336037923463494e-06,
      "loss": 0.4547,
      "step": 2089
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5268372203394658,
      "learning_rate": 9.335300390764464e-06,
      "loss": 0.2636,
      "step": 2090
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4277271674706975,
      "learning_rate": 9.334562477830586e-06,
      "loss": 0.2167,
      "step": 2091
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4088499317862733,
      "learning_rate": 9.33382418472658e-06,
      "loss": 0.1997,
      "step": 2092
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4762655758431042,
      "learning_rate": 9.333085511517204e-06,
      "loss": 0.2646,
      "step": 2093
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.699342133358629,
      "learning_rate": 9.33234645826724e-06,
      "loss": 0.2802,
      "step": 2094
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5158917195881398,
      "learning_rate": 9.331607025041509e-06,
      "loss": 0.171,
      "step": 2095
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4587928506122825,
      "learning_rate": 9.330867211904865e-06,
      "loss": 0.2461,
      "step": 2096
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4549880521491958,
      "learning_rate": 9.330127018922195e-06,
      "loss": 0.2257,
      "step": 2097
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4024049019491291,
      "learning_rate": 9.329386446158416e-06,
      "loss": 0.2367,
      "step": 2098
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5664306616311205,
      "learning_rate": 9.328645493678483e-06,
      "loss": 0.2132,
      "step": 2099
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4985635389119705,
      "learning_rate": 9.327904161547386e-06,
      "loss": 0.2409,
      "step": 2100
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4169771327310279,
      "learning_rate": 9.327162449830139e-06,
      "loss": 0.1702,
      "step": 2101
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.924870384502962,
      "learning_rate": 9.326420358591797e-06,
      "loss": 0.5816,
      "step": 2102
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5632349988137866,
      "learning_rate": 9.325677887897447e-06,
      "loss": 0.2563,
      "step": 2103
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6671523500569791,
      "learning_rate": 9.324935037812208e-06,
      "loss": 0.2764,
      "step": 2104
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.802139437359112,
      "learning_rate": 9.324191808401235e-06,
      "loss": 0.4382,
      "step": 2105
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4477260053285244,
      "learning_rate": 9.323448199729713e-06,
      "loss": 0.2423,
      "step": 2106
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5145599008090043,
      "learning_rate": 9.322704211862858e-06,
      "loss": 0.228,
      "step": 2107
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6052387216034072,
      "learning_rate": 9.321959844865928e-06,
      "loss": 0.2078,
      "step": 2108
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6196397439268364,
      "learning_rate": 9.321215098804205e-06,
      "loss": 0.2245,
      "step": 2109
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4292649467988283,
      "learning_rate": 9.320469973743013e-06,
      "loss": 0.244,
      "step": 2110
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6282318574953984,
      "learning_rate": 9.3197244697477e-06,
      "loss": 0.2663,
      "step": 2111
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6402755615677946,
      "learning_rate": 9.318978586883652e-06,
      "loss": 0.2386,
      "step": 2112
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3324286033474997,
      "learning_rate": 9.31823232521629e-06,
      "loss": 0.186,
      "step": 2113
    },
    {
      "epoch": 0.19,
      "grad_norm": 9.064114509501852,
      "learning_rate": 9.317485684811065e-06,
      "loss": 0.7643,
      "step": 2114
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6089711395664068,
      "learning_rate": 9.316738665733462e-06,
      "loss": 0.2763,
      "step": 2115
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.364624383575771,
      "learning_rate": 9.315991268049001e-06,
      "loss": 0.2094,
      "step": 2116
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3283255591968788,
      "learning_rate": 9.315243491823234e-06,
      "loss": 0.2002,
      "step": 2117
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5842708238421166,
      "learning_rate": 9.314495337121742e-06,
      "loss": 0.226,
      "step": 2118
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3420665446697149,
      "learning_rate": 9.313746804010148e-06,
      "loss": 0.2192,
      "step": 2119
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4610954638252078,
      "learning_rate": 9.312997892554101e-06,
      "loss": 0.2431,
      "step": 2120
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5134890498953364,
      "learning_rate": 9.312248602819284e-06,
      "loss": 0.2573,
      "step": 2121
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.585823484775488,
      "learning_rate": 9.31149893487142e-06,
      "loss": 0.2237,
      "step": 2122
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.464708787514176,
      "learning_rate": 9.310748888776254e-06,
      "loss": 0.2399,
      "step": 2123
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5195500569102176,
      "learning_rate": 9.309998464599574e-06,
      "loss": 0.2366,
      "step": 2124
    },
    {
      "epoch": 0.19,
      "grad_norm": 10.546932152072934,
      "learning_rate": 9.309247662407194e-06,
      "loss": 0.6806,
      "step": 2125
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3352583589629325,
      "learning_rate": 9.308496482264968e-06,
      "loss": 0.2539,
      "step": 2126
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5074205658846265,
      "learning_rate": 9.307744924238777e-06,
      "loss": 0.2606,
      "step": 2127
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5268725573512913,
      "learning_rate": 9.306992988394539e-06,
      "loss": 0.2487,
      "step": 2128
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3951400271249286,
      "learning_rate": 9.306240674798203e-06,
      "loss": 0.2011,
      "step": 2129
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4879004001576255,
      "learning_rate": 9.305487983515751e-06,
      "loss": 0.2181,
      "step": 2130
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.149381681187313,
      "learning_rate": 9.3047349146132e-06,
      "loss": 0.5885,
      "step": 2131
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3497524493135666,
      "learning_rate": 9.303981468156601e-06,
      "loss": 0.2775,
      "step": 2132
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.169187331303034,
      "learning_rate": 9.303227644212034e-06,
      "loss": 0.5057,
      "step": 2133
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3370017993791494,
      "learning_rate": 9.302473442845615e-06,
      "loss": 0.2118,
      "step": 2134
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.312528451001121,
      "learning_rate": 9.301718864123492e-06,
      "loss": 0.1932,
      "step": 2135
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.748610552417936,
      "learning_rate": 9.30096390811185e-06,
      "loss": 0.6496,
      "step": 2136
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.478490513521484,
      "learning_rate": 9.300208574876897e-06,
      "loss": 0.532,
      "step": 2137
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.376478536664148,
      "learning_rate": 9.299452864484885e-06,
      "loss": 0.192,
      "step": 2138
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4542369271242097,
      "learning_rate": 9.298696777002096e-06,
      "loss": 0.1849,
      "step": 2139
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2291984178834732,
      "learning_rate": 9.297940312494843e-06,
      "loss": 0.1603,
      "step": 2140
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4088862837946718,
      "learning_rate": 9.297183471029472e-06,
      "loss": 0.1904,
      "step": 2141
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4040123705627983,
      "learning_rate": 9.296426252672363e-06,
      "loss": 0.2127,
      "step": 2142
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4790357622912371,
      "learning_rate": 9.295668657489929e-06,
      "loss": 0.1876,
      "step": 2143
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5119996491662946,
      "learning_rate": 9.294910685548617e-06,
      "loss": 0.2198,
      "step": 2144
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4097793256852647,
      "learning_rate": 9.294152336914907e-06,
      "loss": 0.2481,
      "step": 2145
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3802642747228295,
      "learning_rate": 9.293393611655311e-06,
      "loss": 0.2268,
      "step": 2146
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4494636619158765,
      "learning_rate": 9.29263450983637e-06,
      "loss": 0.2039,
      "step": 2147
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4621558832571107,
      "learning_rate": 9.291875031524669e-06,
      "loss": 0.1925,
      "step": 2148
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7008709685838508,
      "learning_rate": 9.291115176786814e-06,
      "loss": 0.2313,
      "step": 2149
    },
    {
      "epoch": 0.2,
      "grad_norm": 13.661440072596172,
      "learning_rate": 9.290354945689452e-06,
      "loss": 0.6164,
      "step": 2150
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6688688444565691,
      "learning_rate": 9.289594338299262e-06,
      "loss": 0.2534,
      "step": 2151
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4484505309325129,
      "learning_rate": 9.28883335468295e-06,
      "loss": 0.3063,
      "step": 2152
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2599598963213758,
      "learning_rate": 9.288071994907262e-06,
      "loss": 0.1977,
      "step": 2153
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.512904772630652,
      "learning_rate": 9.287310259038973e-06,
      "loss": 0.2667,
      "step": 2154
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.382380866941268,
      "learning_rate": 9.286548147144895e-06,
      "loss": 0.2247,
      "step": 2155
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4536370168969464,
      "learning_rate": 9.285785659291868e-06,
      "loss": 0.2249,
      "step": 2156
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2677030040792712,
      "learning_rate": 9.285022795546766e-06,
      "loss": 0.2033,
      "step": 2157
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4585048964155702,
      "learning_rate": 9.2842595559765e-06,
      "loss": 0.1889,
      "step": 2158
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4910783827656626,
      "learning_rate": 9.283495940648013e-06,
      "loss": 0.2567,
      "step": 2159
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6116759933888267,
      "learning_rate": 9.282731949628273e-06,
      "loss": 0.3076,
      "step": 2160
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.444633432168669,
      "learning_rate": 9.281967582984292e-06,
      "loss": 0.6085,
      "step": 2161
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5719585047327733,
      "learning_rate": 9.281202840783109e-06,
      "loss": 0.1994,
      "step": 2162
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4507519624389702,
      "learning_rate": 9.280437723091794e-06,
      "loss": 0.1944,
      "step": 2163
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3211862738573497,
      "learning_rate": 9.279672229977458e-06,
      "loss": 0.2008,
      "step": 2164
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4325523593786955,
      "learning_rate": 9.278906361507238e-06,
      "loss": 0.1926,
      "step": 2165
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.307431872303585,
      "learning_rate": 9.278140117748304e-06,
      "loss": 0.7175,
      "step": 2166
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5089229716293506,
      "learning_rate": 9.277373498767863e-06,
      "loss": 0.1945,
      "step": 2167
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4556546450898507,
      "learning_rate": 9.276606504633151e-06,
      "loss": 0.2042,
      "step": 2168
    },
    {
      "epoch": 0.2,
      "grad_norm": 11.984433231663472,
      "learning_rate": 9.275839135411439e-06,
      "loss": 0.851,
      "step": 2169
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3560767521240293,
      "learning_rate": 9.275071391170031e-06,
      "loss": 0.178,
      "step": 2170
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3342886458685028,
      "learning_rate": 9.274303271976262e-06,
      "loss": 0.225,
      "step": 2171
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5682824884156739,
      "learning_rate": 9.273534777897503e-06,
      "loss": 0.252,
      "step": 2172
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3894845500612965,
      "learning_rate": 9.272765909001154e-06,
      "loss": 0.2406,
      "step": 2173
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4295992383020806,
      "learning_rate": 9.27199666535465e-06,
      "loss": 0.2133,
      "step": 2174
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.618141078411502,
      "learning_rate": 9.271227047025462e-06,
      "loss": 0.2444,
      "step": 2175
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2622820472285086,
      "learning_rate": 9.270457054081088e-06,
      "loss": 0.1772,
      "step": 2176
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4524875656486351,
      "learning_rate": 9.269686686589063e-06,
      "loss": 0.1911,
      "step": 2177
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2685040030961516,
      "learning_rate": 9.26891594461695e-06,
      "loss": 0.1884,
      "step": 2178
    },
    {
      "epoch": 0.2,
      "grad_norm": 6.031535547575513,
      "learning_rate": 9.268144828232352e-06,
      "loss": 0.5658,
      "step": 2179
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6461512767386437,
      "learning_rate": 9.2673733375029e-06,
      "loss": 0.2651,
      "step": 2180
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.794943740838582,
      "learning_rate": 9.266601472496257e-06,
      "loss": 0.5676,
      "step": 2181
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4801964964218497,
      "learning_rate": 9.265829233280124e-06,
      "loss": 0.2552,
      "step": 2182
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.370056115917438,
      "learning_rate": 9.26505661992223e-06,
      "loss": 0.2422,
      "step": 2183
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3613674717664175,
      "learning_rate": 9.264283632490338e-06,
      "loss": 0.2352,
      "step": 2184
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.383662063539557,
      "learning_rate": 9.263510271052243e-06,
      "loss": 0.2294,
      "step": 2185
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.358824921183217,
      "learning_rate": 9.262736535675778e-06,
      "loss": 0.2737,
      "step": 2186
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4053962491430956,
      "learning_rate": 9.261962426428798e-06,
      "loss": 0.1919,
      "step": 2187
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3589067459446642,
      "learning_rate": 9.261187943379206e-06,
      "loss": 0.2699,
      "step": 2188
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4634698210565034,
      "learning_rate": 9.260413086594923e-06,
      "loss": 0.2024,
      "step": 2189
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4942039733047605,
      "learning_rate": 9.259637856143911e-06,
      "loss": 0.2322,
      "step": 2190
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3539034602864164,
      "learning_rate": 9.258862252094164e-06,
      "loss": 0.1677,
      "step": 2191
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5613160934647514,
      "learning_rate": 9.258086274513707e-06,
      "loss": 0.2212,
      "step": 2192
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3196586975379934,
      "learning_rate": 9.257309923470596e-06,
      "loss": 0.1954,
      "step": 2193
    },
    {
      "epoch": 0.2,
      "grad_norm": 9.202645161595825,
      "learning_rate": 9.256533199032926e-06,
      "loss": 0.6373,
      "step": 2194
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3524475745799533,
      "learning_rate": 9.25575610126882e-06,
      "loss": 0.2217,
      "step": 2195
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.44308072028083,
      "learning_rate": 9.254978630246432e-06,
      "loss": 0.6279,
      "step": 2196
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.415909780669994,
      "learning_rate": 9.254200786033956e-06,
      "loss": 0.1911,
      "step": 2197
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.411890033847372,
      "learning_rate": 9.253422568699608e-06,
      "loss": 0.2242,
      "step": 2198
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.440655963054898,
      "learning_rate": 9.252643978311649e-06,
      "loss": 0.228,
      "step": 2199
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3892936708821009,
      "learning_rate": 9.251865014938364e-06,
      "loss": 0.2303,
      "step": 2200
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5490298207341369,
      "learning_rate": 9.251085678648072e-06,
      "loss": 0.2475,
      "step": 2201
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3564836657911377,
      "learning_rate": 9.250305969509128e-06,
      "loss": 0.1962,
      "step": 2202
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3527310602396359,
      "learning_rate": 9.249525887589917e-06,
      "loss": 0.2461,
      "step": 2203
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4081438844944132,
      "learning_rate": 9.248745432958856e-06,
      "loss": 0.26,
      "step": 2204
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2196721428668622,
      "learning_rate": 9.247964605684399e-06,
      "loss": 0.2193,
      "step": 2205
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5338154927454075,
      "learning_rate": 9.247183405835027e-06,
      "loss": 0.2462,
      "step": 2206
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.32848635086449,
      "learning_rate": 9.246401833479257e-06,
      "loss": 0.5271,
      "step": 2207
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5327752611088792,
      "learning_rate": 9.245619888685642e-06,
      "loss": 0.2027,
      "step": 2208
    },
    {
      "epoch": 0.2,
      "grad_norm": 7.792617431769593,
      "learning_rate": 9.244837571522758e-06,
      "loss": 0.4643,
      "step": 2209
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6288110815769823,
      "learning_rate": 9.244054882059224e-06,
      "loss": 0.2164,
      "step": 2210
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4728399114449306,
      "learning_rate": 9.243271820363684e-06,
      "loss": 0.2455,
      "step": 2211
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6512415357817782,
      "learning_rate": 9.242488386504817e-06,
      "loss": 0.2799,
      "step": 2212
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3770309359747055,
      "learning_rate": 9.24170458055134e-06,
      "loss": 0.2102,
      "step": 2213
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3802764068093347,
      "learning_rate": 9.240920402571995e-06,
      "loss": 0.254,
      "step": 2214
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3701318911012283,
      "learning_rate": 9.24013585263556e-06,
      "loss": 0.24,
      "step": 2215
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.480927217834954,
      "learning_rate": 9.239350930810843e-06,
      "loss": 0.2381,
      "step": 2216
    },
    {
      "epoch": 0.2,
      "grad_norm": 8.070915100017734,
      "learning_rate": 9.238565637166692e-06,
      "loss": 0.7198,
      "step": 2217
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3406105990788262,
      "learning_rate": 9.237779971771977e-06,
      "loss": 0.2023,
      "step": 2218
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3472878890759143,
      "learning_rate": 9.23699393469561e-06,
      "loss": 0.2173,
      "step": 2219
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4432801746884347,
      "learning_rate": 9.23620752600653e-06,
      "loss": 0.2489,
      "step": 2220
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6865061937512404,
      "learning_rate": 9.235420745773714e-06,
      "loss": 0.2876,
      "step": 2221
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3992250710309593,
      "learning_rate": 9.234633594066161e-06,
      "loss": 0.2079,
      "step": 2222
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.824664823828008,
      "learning_rate": 9.233846070952915e-06,
      "loss": 0.2374,
      "step": 2223
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4033250779652655,
      "learning_rate": 9.233058176503043e-06,
      "loss": 0.2716,
      "step": 2224
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3517432653019186,
      "learning_rate": 9.232269910785651e-06,
      "loss": 0.2202,
      "step": 2225
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4102686025610234,
      "learning_rate": 9.231481273869878e-06,
      "loss": 0.2303,
      "step": 2226
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.510664477579311,
      "learning_rate": 9.230692265824888e-06,
      "loss": 0.2435,
      "step": 2227
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5721066760996056,
      "learning_rate": 9.229902886719883e-06,
      "loss": 0.1986,
      "step": 2228
    },
    {
      "epoch": 0.2,
      "grad_norm": 5.736137913231066,
      "learning_rate": 9.229113136624099e-06,
      "loss": 0.4572,
      "step": 2229
    },
    {
      "epoch": 0.2,
      "grad_norm": 19.944553180183892,
      "learning_rate": 9.228323015606801e-06,
      "loss": 0.704,
      "step": 2230
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3364147559034536,
      "learning_rate": 9.227532523737288e-06,
      "loss": 0.2505,
      "step": 2231
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4802782714886418,
      "learning_rate": 9.22674166108489e-06,
      "loss": 0.1828,
      "step": 2232
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5430617467051875,
      "learning_rate": 9.225950427718974e-06,
      "loss": 0.2327,
      "step": 2233
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.376315095561674,
      "learning_rate": 9.225158823708936e-06,
      "loss": 0.1874,
      "step": 2234
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8899084413418166,
      "learning_rate": 9.224366849124202e-06,
      "loss": 0.1949,
      "step": 2235
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6014138549058374,
      "learning_rate": 9.223574504034236e-06,
      "loss": 0.2407,
      "step": 2236
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1815879050360765,
      "learning_rate": 9.222781788508531e-06,
      "loss": 0.1864,
      "step": 2237
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.65759986508392,
      "learning_rate": 9.221988702616613e-06,
      "loss": 0.512,
      "step": 2238
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5463303614591746,
      "learning_rate": 9.221195246428043e-06,
      "loss": 0.2623,
      "step": 2239
    },
    {
      "epoch": 0.2,
      "grad_norm": 4.217057096787312,
      "learning_rate": 9.220401420012412e-06,
      "loss": 0.5557,
      "step": 2240
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3764067779603872,
      "learning_rate": 9.219607223439343e-06,
      "loss": 0.2328,
      "step": 2241
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4734663903509824,
      "learning_rate": 9.21881265677849e-06,
      "loss": 0.2045,
      "step": 2242
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5249144840514095,
      "learning_rate": 9.218017720099545e-06,
      "loss": 0.2687,
      "step": 2243
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4286474313374444,
      "learning_rate": 9.217222413472229e-06,
      "loss": 0.2393,
      "step": 2244
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4641412319546394,
      "learning_rate": 9.216426736966296e-06,
      "loss": 0.2279,
      "step": 2245
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3021014693061321,
      "learning_rate": 9.21563069065153e-06,
      "loss": 0.2141,
      "step": 2246
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5743939733583434,
      "learning_rate": 9.214834274597751e-06,
      "loss": 0.1869,
      "step": 2247
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.025511084113215,
      "learning_rate": 9.21403748887481e-06,
      "loss": 0.5012,
      "step": 2248
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.523165721500062,
      "learning_rate": 9.213240333552589e-06,
      "loss": 0.1867,
      "step": 2249
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.544952055897225,
      "learning_rate": 9.212442808701008e-06,
      "loss": 0.6458,
      "step": 2250
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.342522503517935,
      "learning_rate": 9.21164491439001e-06,
      "loss": 0.2169,
      "step": 2251
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5751347996238527,
      "learning_rate": 9.21084665068958e-06,
      "loss": 0.2327,
      "step": 2252
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5618651138835054,
      "learning_rate": 9.210048017669727e-06,
      "loss": 0.2424,
      "step": 2253
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2215465404232604,
      "learning_rate": 9.209249015400501e-06,
      "loss": 0.2129,
      "step": 2254
    },
    {
      "epoch": 0.21,
      "grad_norm": 11.365593649138534,
      "learning_rate": 9.208449643951976e-06,
      "loss": 0.6572,
      "step": 2255
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.249314702581518,
      "learning_rate": 9.207649903394264e-06,
      "loss": 0.1685,
      "step": 2256
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4616216876114345,
      "learning_rate": 9.206849793797508e-06,
      "loss": 0.191,
      "step": 2257
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.8169547388108285,
      "learning_rate": 9.206049315231882e-06,
      "loss": 0.6294,
      "step": 2258
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3233400692660413,
      "learning_rate": 9.205248467767592e-06,
      "loss": 0.2356,
      "step": 2259
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3903198005480726,
      "learning_rate": 9.204447251474881e-06,
      "loss": 0.2632,
      "step": 2260
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4379654871780436,
      "learning_rate": 9.20364566642402e-06,
      "loss": 0.2373,
      "step": 2261
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4399308945460019,
      "learning_rate": 9.202843712685311e-06,
      "loss": 0.1999,
      "step": 2262
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5533839974949795,
      "learning_rate": 9.202041390329092e-06,
      "loss": 0.2311,
      "step": 2263
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.156589570804053,
      "learning_rate": 9.201238699425735e-06,
      "loss": 0.5975,
      "step": 2264
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.327794893497749,
      "learning_rate": 9.200435640045637e-06,
      "loss": 0.1677,
      "step": 2265
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.098752761073723,
      "learning_rate": 9.199632212259233e-06,
      "loss": 0.5657,
      "step": 2266
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.650541340354742,
      "learning_rate": 9.198828416136991e-06,
      "loss": 0.2311,
      "step": 2267
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6427291337904535,
      "learning_rate": 9.198024251749406e-06,
      "loss": 0.2592,
      "step": 2268
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6352576263032732,
      "learning_rate": 9.197219719167013e-06,
      "loss": 0.2601,
      "step": 2269
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5417585439597399,
      "learning_rate": 9.19641481846037e-06,
      "loss": 0.2256,
      "step": 2270
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4796870687971453,
      "learning_rate": 9.195609549700075e-06,
      "loss": 0.2129,
      "step": 2271
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3121335194887016,
      "learning_rate": 9.194803912956753e-06,
      "loss": 0.1988,
      "step": 2272
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2550755595180483,
      "learning_rate": 9.193997908301069e-06,
      "loss": 0.1482,
      "step": 2273
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4903324676341865,
      "learning_rate": 9.193191535803709e-06,
      "loss": 0.2313,
      "step": 2274
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4122202509955426,
      "learning_rate": 9.192384795535402e-06,
      "loss": 0.1962,
      "step": 2275
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1947612509639614,
      "learning_rate": 9.191577687566902e-06,
      "loss": 0.1953,
      "step": 2276
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.501612060246724,
      "learning_rate": 9.190770211968995e-06,
      "loss": 0.2156,
      "step": 2277
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5828974277523866,
      "learning_rate": 9.189962368812509e-06,
      "loss": 0.2145,
      "step": 2278
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.445738435344419,
      "learning_rate": 9.189154158168293e-06,
      "loss": 0.2061,
      "step": 2279
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5819140468919832,
      "learning_rate": 9.188345580107232e-06,
      "loss": 0.2512,
      "step": 2280
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2386989073446104,
      "learning_rate": 9.187536634700244e-06,
      "loss": 0.1744,
      "step": 2281
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3870981336368617,
      "learning_rate": 9.186727322018281e-06,
      "loss": 0.2032,
      "step": 2282
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4890945457350249,
      "learning_rate": 9.185917642132322e-06,
      "loss": 0.2498,
      "step": 2283
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.400946429888759,
      "learning_rate": 9.185107595113384e-06,
      "loss": 0.1991,
      "step": 2284
    },
    {
      "epoch": 0.21,
      "grad_norm": 9.233617017940409,
      "learning_rate": 9.184297181032513e-06,
      "loss": 0.6954,
      "step": 2285
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4744628328768299,
      "learning_rate": 9.183486399960787e-06,
      "loss": 0.2495,
      "step": 2286
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.358424485530751,
      "learning_rate": 9.182675251969316e-06,
      "loss": 0.1954,
      "step": 2287
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.485573981224487,
      "learning_rate": 9.181863737129245e-06,
      "loss": 0.2317,
      "step": 2288
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.687035239794732,
      "learning_rate": 9.181051855511749e-06,
      "loss": 0.2323,
      "step": 2289
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4816460707841124,
      "learning_rate": 9.180239607188035e-06,
      "loss": 0.2152,
      "step": 2290
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4744175360691794,
      "learning_rate": 9.179426992229343e-06,
      "loss": 0.1817,
      "step": 2291
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3605501224304921,
      "learning_rate": 9.178614010706942e-06,
      "loss": 0.2085,
      "step": 2292
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.3334780373980575,
      "learning_rate": 9.177800662692142e-06,
      "loss": 0.7303,
      "step": 2293
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4605279030240135,
      "learning_rate": 9.176986948256272e-06,
      "loss": 0.1951,
      "step": 2294
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.611517795913014,
      "learning_rate": 9.176172867470707e-06,
      "loss": 0.2889,
      "step": 2295
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3970105020419314,
      "learning_rate": 9.175358420406842e-06,
      "loss": 0.2206,
      "step": 2296
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4718875390678154,
      "learning_rate": 9.174543607136111e-06,
      "loss": 0.2181,
      "step": 2297
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5786741382527039,
      "learning_rate": 9.17372842772998e-06,
      "loss": 0.2279,
      "step": 2298
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2239850191234911,
      "learning_rate": 9.172912882259944e-06,
      "loss": 0.213,
      "step": 2299
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5670650002997062,
      "learning_rate": 9.172096970797533e-06,
      "loss": 0.2403,
      "step": 2300
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4872733770045377,
      "learning_rate": 9.171280693414307e-06,
      "loss": 0.254,
      "step": 2301
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2711134605431957,
      "learning_rate": 9.17046405018186e-06,
      "loss": 0.1897,
      "step": 2302
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4596115548520978,
      "learning_rate": 9.169647041171817e-06,
      "loss": 0.1904,
      "step": 2303
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4193596141427818,
      "learning_rate": 9.168829666455832e-06,
      "loss": 0.2279,
      "step": 2304
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.333121905617738,
      "learning_rate": 9.168011926105598e-06,
      "loss": 0.5427,
      "step": 2305
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2339098250934724,
      "learning_rate": 9.167193820192836e-06,
      "loss": 0.1956,
      "step": 2306
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.413809615544483,
      "learning_rate": 9.166375348789297e-06,
      "loss": 0.2408,
      "step": 2307
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6209455246399467,
      "learning_rate": 9.165556511966768e-06,
      "loss": 0.255,
      "step": 2308
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2583062594646415,
      "learning_rate": 9.164737309797068e-06,
      "loss": 0.2087,
      "step": 2309
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.366250684563212,
      "learning_rate": 9.163917742352044e-06,
      "loss": 0.2156,
      "step": 2310
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4590321138126863,
      "learning_rate": 9.163097809703579e-06,
      "loss": 0.2562,
      "step": 2311
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6016005979364543,
      "learning_rate": 9.162277511923583e-06,
      "loss": 0.2584,
      "step": 2312
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.761693891652089,
      "learning_rate": 9.161456849084007e-06,
      "loss": 0.7394,
      "step": 2313
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5544670378646772,
      "learning_rate": 9.160635821256827e-06,
      "loss": 0.2202,
      "step": 2314
    },
    {
      "epoch": 0.21,
      "grad_norm": 8.671840132563764,
      "learning_rate": 9.15981442851405e-06,
      "loss": 0.6819,
      "step": 2315
    },
    {
      "epoch": 0.21,
      "grad_norm": 7.333971246323631,
      "learning_rate": 9.15899267092772e-06,
      "loss": 0.5414,
      "step": 2316
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4083523346702318,
      "learning_rate": 9.158170548569912e-06,
      "loss": 0.2188,
      "step": 2317
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7757399800601708,
      "learning_rate": 9.157348061512728e-06,
      "loss": 0.2806,
      "step": 2318
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4552770875549774,
      "learning_rate": 9.156525209828306e-06,
      "loss": 0.1894,
      "step": 2319
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.444775054819836,
      "learning_rate": 9.15570199358882e-06,
      "loss": 0.2189,
      "step": 2320
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6519985377849602,
      "learning_rate": 9.154878412866465e-06,
      "loss": 0.2586,
      "step": 2321
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3324352777022952,
      "learning_rate": 9.15405446773348e-06,
      "loss": 0.1696,
      "step": 2322
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5088770366052087,
      "learning_rate": 9.153230158262129e-06,
      "loss": 0.2573,
      "step": 2323
    },
    {
      "epoch": 0.21,
      "grad_norm": 6.2339530536987215,
      "learning_rate": 9.152405484524709e-06,
      "loss": 0.6046,
      "step": 2324
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5285163667081982,
      "learning_rate": 9.151580446593548e-06,
      "loss": 0.1963,
      "step": 2325
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5111682331590581,
      "learning_rate": 9.150755044541011e-06,
      "loss": 0.205,
      "step": 2326
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3032338818235667,
      "learning_rate": 9.149929278439487e-06,
      "loss": 0.1988,
      "step": 2327
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7792765904751935,
      "learning_rate": 9.149103148361404e-06,
      "loss": 0.1992,
      "step": 2328
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.372483998955009,
      "learning_rate": 9.14827665437922e-06,
      "loss": 0.2398,
      "step": 2329
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4624772476882903,
      "learning_rate": 9.14744979656542e-06,
      "loss": 0.2389,
      "step": 2330
    },
    {
      "epoch": 0.21,
      "grad_norm": 5.984294726645087,
      "learning_rate": 9.146622574992528e-06,
      "loss": 0.6011,
      "step": 2331
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4023494464853763,
      "learning_rate": 9.145794989733096e-06,
      "loss": 0.2292,
      "step": 2332
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.684667320009622,
      "learning_rate": 9.14496704085971e-06,
      "loss": 0.5536,
      "step": 2333
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.497395238797467,
      "learning_rate": 9.144138728444986e-06,
      "loss": 0.2233,
      "step": 2334
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5921287887399467,
      "learning_rate": 9.14331005256157e-06,
      "loss": 0.2018,
      "step": 2335
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4136604339580152,
      "learning_rate": 9.142481013282148e-06,
      "loss": 0.2116,
      "step": 2336
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5128609079706759,
      "learning_rate": 9.141651610679427e-06,
      "loss": 0.2671,
      "step": 2337
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5117355917619386,
      "learning_rate": 9.140821844826153e-06,
      "loss": 0.2711,
      "step": 2338
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5798663657987777,
      "learning_rate": 9.139991715795102e-06,
      "loss": 0.281,
      "step": 2339
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.7124516165301398,
      "learning_rate": 9.139161223659082e-06,
      "loss": 0.2903,
      "step": 2340
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4811842935258268,
      "learning_rate": 9.138330368490934e-06,
      "loss": 0.1704,
      "step": 2341
    },
    {
      "epoch": 0.21,
      "grad_norm": 13.585148147580867,
      "learning_rate": 9.137499150363526e-06,
      "loss": 0.5445,
      "step": 2342
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5951762868379524,
      "learning_rate": 9.136667569349767e-06,
      "loss": 0.2997,
      "step": 2343
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3743586345153644,
      "learning_rate": 9.135835625522585e-06,
      "loss": 0.1901,
      "step": 2344
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1279758788683805,
      "learning_rate": 9.135003318954954e-06,
      "loss": 0.1526,
      "step": 2345
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2289853597994116,
      "learning_rate": 9.134170649719867e-06,
      "loss": 0.1877,
      "step": 2346
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.340780362131192,
      "learning_rate": 9.13333761789036e-06,
      "loss": 0.2281,
      "step": 2347
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2384151939019294,
      "learning_rate": 9.132504223539491e-06,
      "loss": 0.1757,
      "step": 2348
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6098515514637717,
      "learning_rate": 9.131670466740357e-06,
      "loss": 0.2238,
      "step": 2349
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.433923227210694,
      "learning_rate": 9.130836347566083e-06,
      "loss": 0.2069,
      "step": 2350
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.384991800908738,
      "learning_rate": 9.130001866089828e-06,
      "loss": 0.1821,
      "step": 2351
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5571412905619246,
      "learning_rate": 9.129167022384779e-06,
      "loss": 0.2503,
      "step": 2352
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5136224044645292,
      "learning_rate": 9.12833181652416e-06,
      "loss": 0.1894,
      "step": 2353
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3821296647388055,
      "learning_rate": 9.127496248581223e-06,
      "loss": 0.186,
      "step": 2354
    },
    {
      "epoch": 0.22,
      "grad_norm": 10.874122434269399,
      "learning_rate": 9.126660318629252e-06,
      "loss": 0.6865,
      "step": 2355
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2841603015763647,
      "learning_rate": 9.125824026741566e-06,
      "loss": 0.1711,
      "step": 2356
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4090295287158399,
      "learning_rate": 9.124987372991512e-06,
      "loss": 0.2376,
      "step": 2357
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.393383643444646,
      "learning_rate": 9.12415035745247e-06,
      "loss": 0.2312,
      "step": 2358
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4267426673784496,
      "learning_rate": 9.123312980197852e-06,
      "loss": 0.2321,
      "step": 2359
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.350951560995063,
      "learning_rate": 9.122475241301102e-06,
      "loss": 0.2036,
      "step": 2360
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4111971227309232,
      "learning_rate": 9.121637140835696e-06,
      "loss": 0.2251,
      "step": 2361
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7453606777916042,
      "learning_rate": 9.120798678875137e-06,
      "loss": 0.2214,
      "step": 2362
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6148753849661437,
      "learning_rate": 9.119959855492969e-06,
      "loss": 0.2149,
      "step": 2363
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4968631022752203,
      "learning_rate": 9.11912067076276e-06,
      "loss": 0.23,
      "step": 2364
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5149927012417095,
      "learning_rate": 9.11828112475811e-06,
      "loss": 0.1995,
      "step": 2365
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2734098727364118,
      "learning_rate": 9.117441217552658e-06,
      "loss": 0.2122,
      "step": 2366
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3144453522645023,
      "learning_rate": 9.116600949220066e-06,
      "loss": 0.2012,
      "step": 2367
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.163685675225926,
      "learning_rate": 9.11576031983403e-06,
      "loss": 0.5917,
      "step": 2368
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3436050738605891,
      "learning_rate": 9.114919329468283e-06,
      "loss": 0.2305,
      "step": 2369
    },
    {
      "epoch": 0.22,
      "grad_norm": 4.083597273107623,
      "learning_rate": 9.11407797819658e-06,
      "loss": 0.4774,
      "step": 2370
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.781961014503872,
      "learning_rate": 9.11323626609272e-06,
      "loss": 0.2084,
      "step": 2371
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.875512834331647,
      "learning_rate": 9.11239419323052e-06,
      "loss": 0.6749,
      "step": 2372
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.218106208839189,
      "learning_rate": 9.111551759683839e-06,
      "loss": 0.2123,
      "step": 2373
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4299848629820733,
      "learning_rate": 9.110708965526562e-06,
      "loss": 0.231,
      "step": 2374
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3994827996025863,
      "learning_rate": 9.109865810832612e-06,
      "loss": 0.2238,
      "step": 2375
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.167716141456229,
      "learning_rate": 9.109022295675934e-06,
      "loss": 0.466,
      "step": 2376
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3192008218190907,
      "learning_rate": 9.108178420130514e-06,
      "loss": 0.1963,
      "step": 2377
    },
    {
      "epoch": 0.22,
      "grad_norm": 11.75709855089902,
      "learning_rate": 9.107334184270362e-06,
      "loss": 0.4941,
      "step": 2378
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4569084431992292,
      "learning_rate": 9.106489588169528e-06,
      "loss": 0.2128,
      "step": 2379
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.54877777621254,
      "learning_rate": 9.105644631902081e-06,
      "loss": 0.2518,
      "step": 2380
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4413866102732604,
      "learning_rate": 9.104799315542138e-06,
      "loss": 0.2067,
      "step": 2381
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.229250601559573,
      "learning_rate": 9.103953639163834e-06,
      "loss": 0.7144,
      "step": 2382
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5440441566368075,
      "learning_rate": 9.103107602841341e-06,
      "loss": 0.2019,
      "step": 2383
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3533301595083875,
      "learning_rate": 9.102261206648862e-06,
      "loss": 0.1856,
      "step": 2384
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4298779891155367,
      "learning_rate": 9.101414450660633e-06,
      "loss": 0.1942,
      "step": 2385
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.254739037138083,
      "learning_rate": 9.100567334950919e-06,
      "loss": 0.5008,
      "step": 2386
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3423491999577704,
      "learning_rate": 9.099719859594018e-06,
      "loss": 0.264,
      "step": 2387
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5695438845243035,
      "learning_rate": 9.098872024664258e-06,
      "loss": 0.2473,
      "step": 2388
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2752756070585023,
      "learning_rate": 9.098023830236002e-06,
      "loss": 0.205,
      "step": 2389
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3938439747375764,
      "learning_rate": 9.09717527638364e-06,
      "loss": 0.1778,
      "step": 2390
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.33487949880131,
      "learning_rate": 9.0963263631816e-06,
      "loss": 0.1615,
      "step": 2391
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5539823948112352,
      "learning_rate": 9.095477090704332e-06,
      "loss": 0.19,
      "step": 2392
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.643595854738931,
      "learning_rate": 9.094627459026326e-06,
      "loss": 0.2764,
      "step": 2393
    },
    {
      "epoch": 0.22,
      "grad_norm": 7.029966528288545,
      "learning_rate": 9.0937774682221e-06,
      "loss": 0.4967,
      "step": 2394
    },
    {
      "epoch": 0.22,
      "grad_norm": 6.266432186620175,
      "learning_rate": 9.092927118366203e-06,
      "loss": 0.5802,
      "step": 2395
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2947634836327029,
      "learning_rate": 9.092076409533218e-06,
      "loss": 0.1857,
      "step": 2396
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5313422792129037,
      "learning_rate": 9.091225341797757e-06,
      "loss": 0.2242,
      "step": 2397
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.425056775235877,
      "learning_rate": 9.090373915234463e-06,
      "loss": 0.2334,
      "step": 2398
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4159747671947036,
      "learning_rate": 9.089522129918014e-06,
      "loss": 0.2651,
      "step": 2399
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.608995227409002,
      "learning_rate": 9.088669985923115e-06,
      "loss": 0.6294,
      "step": 2400
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5150314415522301,
      "learning_rate": 9.087817483324507e-06,
      "loss": 0.2404,
      "step": 2401
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.591294674221441,
      "learning_rate": 9.086964622196962e-06,
      "loss": 0.2655,
      "step": 2402
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6997028302948638,
      "learning_rate": 9.086111402615274e-06,
      "loss": 0.2964,
      "step": 2403
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5980687983420672,
      "learning_rate": 9.085257824654283e-06,
      "loss": 0.2746,
      "step": 2404
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4142313480433069,
      "learning_rate": 9.084403888388852e-06,
      "loss": 0.193,
      "step": 2405
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5251031673874063,
      "learning_rate": 9.083549593893876e-06,
      "loss": 0.2414,
      "step": 2406
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2904820733165676,
      "learning_rate": 9.082694941244282e-06,
      "loss": 0.2084,
      "step": 2407
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7080778933502183,
      "learning_rate": 9.08183993051503e-06,
      "loss": 0.2735,
      "step": 2408
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.63947137748463,
      "learning_rate": 9.08098456178111e-06,
      "loss": 0.5407,
      "step": 2409
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4109611404766345,
      "learning_rate": 9.080128835117541e-06,
      "loss": 0.2279,
      "step": 2410
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3749050285511626,
      "learning_rate": 9.079272750599381e-06,
      "loss": 0.1987,
      "step": 2411
    },
    {
      "epoch": 0.22,
      "grad_norm": 8.225327925288067,
      "learning_rate": 9.07841630830171e-06,
      "loss": 0.4676,
      "step": 2412
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.382516038326227,
      "learning_rate": 9.077559508299644e-06,
      "loss": 0.2393,
      "step": 2413
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.717810494352102,
      "learning_rate": 9.076702350668335e-06,
      "loss": 0.3135,
      "step": 2414
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3506768820660116,
      "learning_rate": 9.075844835482952e-06,
      "loss": 0.204,
      "step": 2415
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4747335799474417,
      "learning_rate": 9.074986962818714e-06,
      "loss": 0.2614,
      "step": 2416
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.381854877626078,
      "learning_rate": 9.074128732750859e-06,
      "loss": 0.2351,
      "step": 2417
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3645225224166553,
      "learning_rate": 9.073270145354658e-06,
      "loss": 0.2128,
      "step": 2418
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6427287745443184,
      "learning_rate": 9.072411200705417e-06,
      "loss": 0.2563,
      "step": 2419
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3720856370903431,
      "learning_rate": 9.071551898878471e-06,
      "loss": 0.2177,
      "step": 2420
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.27828774825116,
      "learning_rate": 9.070692239949183e-06,
      "loss": 0.5844,
      "step": 2421
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4772150192626732,
      "learning_rate": 9.069832223992955e-06,
      "loss": 0.2315,
      "step": 2422
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.53132623091745,
      "learning_rate": 9.068971851085215e-06,
      "loss": 0.2155,
      "step": 2423
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4713519074278427,
      "learning_rate": 9.068111121301424e-06,
      "loss": 0.2545,
      "step": 2424
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.507672635593707,
      "learning_rate": 9.067250034717072e-06,
      "loss": 0.2484,
      "step": 2425
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4920626050923238,
      "learning_rate": 9.066388591407683e-06,
      "loss": 0.2576,
      "step": 2426
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4703686845160404,
      "learning_rate": 9.065526791448812e-06,
      "loss": 0.2019,
      "step": 2427
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4602130094331591,
      "learning_rate": 9.064664634916042e-06,
      "loss": 0.2553,
      "step": 2428
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5427287199066582,
      "learning_rate": 9.063802121884995e-06,
      "loss": 0.2276,
      "step": 2429
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.698370274263152,
      "learning_rate": 9.062939252431315e-06,
      "loss": 0.5867,
      "step": 2430
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7204400341752888,
      "learning_rate": 9.062076026630682e-06,
      "loss": 0.2499,
      "step": 2431
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3991764828051338,
      "learning_rate": 9.061212444558807e-06,
      "loss": 0.2678,
      "step": 2432
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3759832836677897,
      "learning_rate": 9.060348506291432e-06,
      "loss": 0.2243,
      "step": 2433
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4777654657804038,
      "learning_rate": 9.05948421190433e-06,
      "loss": 0.2181,
      "step": 2434
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6083061426264071,
      "learning_rate": 9.058619561473308e-06,
      "loss": 0.2528,
      "step": 2435
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6282617037765443,
      "learning_rate": 9.057754555074196e-06,
      "loss": 0.2178,
      "step": 2436
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3143309378893813,
      "learning_rate": 9.056889192782865e-06,
      "loss": 0.1679,
      "step": 2437
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3098306954644983,
      "learning_rate": 9.056023474675213e-06,
      "loss": 0.1782,
      "step": 2438
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4728864654938099,
      "learning_rate": 9.055157400827168e-06,
      "loss": 0.2641,
      "step": 2439
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.69039928526605,
      "learning_rate": 9.054290971314692e-06,
      "loss": 0.2292,
      "step": 2440
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3410119446162514,
      "learning_rate": 9.053424186213776e-06,
      "loss": 0.191,
      "step": 2441
    },
    {
      "epoch": 0.22,
      "grad_norm": 3.9487301594333775,
      "learning_rate": 9.052557045600442e-06,
      "loss": 0.6536,
      "step": 2442
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.353652767460656,
      "learning_rate": 9.051689549550741e-06,
      "loss": 0.6777,
      "step": 2443
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.411590170784529,
      "learning_rate": 9.050821698140765e-06,
      "loss": 0.2285,
      "step": 2444
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5510198892414024,
      "learning_rate": 9.049953491446627e-06,
      "loss": 0.2444,
      "step": 2445
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5269264363439847,
      "learning_rate": 9.049084929544474e-06,
      "loss": 0.2375,
      "step": 2446
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2653238309245274,
      "learning_rate": 9.048216012510485e-06,
      "loss": 0.1857,
      "step": 2447
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3281026161933835,
      "learning_rate": 9.04734674042087e-06,
      "loss": 0.1586,
      "step": 2448
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3739598273212639,
      "learning_rate": 9.046477113351871e-06,
      "loss": 0.2235,
      "step": 2449
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3215202626027684,
      "learning_rate": 9.04560713137976e-06,
      "loss": 0.2083,
      "step": 2450
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5883860233709652,
      "learning_rate": 9.044736794580837e-06,
      "loss": 0.2541,
      "step": 2451
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3768230328397313,
      "learning_rate": 9.04386610303144e-06,
      "loss": 0.2112,
      "step": 2452
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4491749205389082,
      "learning_rate": 9.042995056807933e-06,
      "loss": 0.2347,
      "step": 2453
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4174533843731005,
      "learning_rate": 9.042123655986713e-06,
      "loss": 0.2323,
      "step": 2454
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1226044924576053,
      "learning_rate": 9.041251900644207e-06,
      "loss": 0.1558,
      "step": 2455
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4802507364759765,
      "learning_rate": 9.040379790856875e-06,
      "loss": 0.23,
      "step": 2456
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3332339557403636,
      "learning_rate": 9.039507326701207e-06,
      "loss": 0.179,
      "step": 2457
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5423121555440193,
      "learning_rate": 9.038634508253722e-06,
      "loss": 0.2313,
      "step": 2458
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.496093195151772,
      "learning_rate": 9.037761335590972e-06,
      "loss": 0.2499,
      "step": 2459
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3699199225917162,
      "learning_rate": 9.036887808789544e-06,
      "loss": 0.2428,
      "step": 2460
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7158532223584517,
      "learning_rate": 9.036013927926049e-06,
      "loss": 0.2588,
      "step": 2461
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3749622810267255,
      "learning_rate": 9.03513969307713e-06,
      "loss": 0.1807,
      "step": 2462
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2357246421664596,
      "learning_rate": 9.034265104319468e-06,
      "loss": 0.1963,
      "step": 2463
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4267566251533155,
      "learning_rate": 9.033390161729767e-06,
      "loss": 0.1834,
      "step": 2464
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4643555584513361,
      "learning_rate": 9.032514865384767e-06,
      "loss": 0.2158,
      "step": 2465
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3445333571376408,
      "learning_rate": 9.031639215361237e-06,
      "loss": 0.2181,
      "step": 2466
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5186474042335205,
      "learning_rate": 9.030763211735977e-06,
      "loss": 0.2523,
      "step": 2467
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6866024616029587,
      "learning_rate": 9.02988685458582e-06,
      "loss": 0.2331,
      "step": 2468
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3994278394859412,
      "learning_rate": 9.029010143987624e-06,
      "loss": 0.2276,
      "step": 2469
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1130427144682027,
      "learning_rate": 9.028133080018287e-06,
      "loss": 0.1481,
      "step": 2470
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4377911754521617,
      "learning_rate": 9.02725566275473e-06,
      "loss": 0.2408,
      "step": 2471
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5874534130102171,
      "learning_rate": 9.026377892273911e-06,
      "loss": 0.2379,
      "step": 2472
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4349382288192365,
      "learning_rate": 9.025499768652817e-06,
      "loss": 0.2315,
      "step": 2473
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.346383008241994,
      "learning_rate": 9.024621291968461e-06,
      "loss": 0.1839,
      "step": 2474
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.208280432285236,
      "learning_rate": 9.023742462297895e-06,
      "loss": 0.5619,
      "step": 2475
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5193228322108059,
      "learning_rate": 9.022863279718197e-06,
      "loss": 0.2241,
      "step": 2476
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4263866767059217,
      "learning_rate": 9.021983744306478e-06,
      "loss": 0.2132,
      "step": 2477
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3864601771803573,
      "learning_rate": 9.021103856139879e-06,
      "loss": 0.2217,
      "step": 2478
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3663729619769034,
      "learning_rate": 9.020223615295572e-06,
      "loss": 0.1937,
      "step": 2479
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4219284831592054,
      "learning_rate": 9.019343021850758e-06,
      "loss": 0.2045,
      "step": 2480
    },
    {
      "epoch": 0.23,
      "grad_norm": 8.384804276670044,
      "learning_rate": 9.018462075882673e-06,
      "loss": 0.598,
      "step": 2481
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.483539906172262,
      "learning_rate": 9.01758077746858e-06,
      "loss": 0.258,
      "step": 2482
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3941841803425277,
      "learning_rate": 9.016699126685777e-06,
      "loss": 0.236,
      "step": 2483
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3308799120454509,
      "learning_rate": 9.015817123611591e-06,
      "loss": 0.2091,
      "step": 2484
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.937257767279001,
      "learning_rate": 9.014934768323378e-06,
      "loss": 0.5499,
      "step": 2485
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2565841541883123,
      "learning_rate": 9.014052060898527e-06,
      "loss": 0.1663,
      "step": 2486
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5424631717074226,
      "learning_rate": 9.013169001414458e-06,
      "loss": 0.2623,
      "step": 2487
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.437863222355023,
      "learning_rate": 9.012285589948618e-06,
      "loss": 0.2714,
      "step": 2488
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4447806765295557,
      "learning_rate": 9.011401826578492e-06,
      "loss": 0.1911,
      "step": 2489
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.4875419318882885,
      "learning_rate": 9.01051771138159e-06,
      "loss": 0.5482,
      "step": 2490
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.538460493732001,
      "learning_rate": 9.009633244435456e-06,
      "loss": 0.7021,
      "step": 2491
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5548163438465974,
      "learning_rate": 9.008748425817664e-06,
      "loss": 0.2269,
      "step": 2492
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.264924572233476,
      "learning_rate": 9.007863255605815e-06,
      "loss": 0.1355,
      "step": 2493
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.322701599990516,
      "learning_rate": 9.006977733877549e-06,
      "loss": 0.205,
      "step": 2494
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.587653083648968,
      "learning_rate": 9.006091860710529e-06,
      "loss": 0.2134,
      "step": 2495
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5676591788048748,
      "learning_rate": 9.005205636182452e-06,
      "loss": 0.2628,
      "step": 2496
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4011512268116708,
      "learning_rate": 9.00431906037105e-06,
      "loss": 0.1896,
      "step": 2497
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2137366157373952,
      "learning_rate": 9.003432133354075e-06,
      "loss": 0.2459,
      "step": 2498
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3371145430854994,
      "learning_rate": 9.002544855209322e-06,
      "loss": 0.1921,
      "step": 2499
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5272688453155576,
      "learning_rate": 9.001657226014609e-06,
      "loss": 0.2266,
      "step": 2500
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5562182010594083,
      "learning_rate": 9.000769245847786e-06,
      "loss": 0.273,
      "step": 2501
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3311394950534514,
      "learning_rate": 8.999880914786737e-06,
      "loss": 0.2331,
      "step": 2502
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.417123855926398,
      "learning_rate": 8.998992232909374e-06,
      "loss": 0.5815,
      "step": 2503
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.586377619874347,
      "learning_rate": 8.998103200293638e-06,
      "loss": 0.2536,
      "step": 2504
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.690536192911846,
      "learning_rate": 8.997213817017508e-06,
      "loss": 0.6745,
      "step": 2505
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.727277458335071,
      "learning_rate": 8.996324083158982e-06,
      "loss": 0.5102,
      "step": 2506
    },
    {
      "epoch": 0.23,
      "grad_norm": 5.998705744685242,
      "learning_rate": 8.995433998796102e-06,
      "loss": 0.7419,
      "step": 2507
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.21588696416881,
      "learning_rate": 8.994543564006931e-06,
      "loss": 0.1912,
      "step": 2508
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4602403403961255,
      "learning_rate": 8.993652778869569e-06,
      "loss": 0.2273,
      "step": 2509
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5207909612809123,
      "learning_rate": 8.992761643462137e-06,
      "loss": 0.2168,
      "step": 2510
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.490702457100676,
      "learning_rate": 8.991870157862803e-06,
      "loss": 0.2203,
      "step": 2511
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2534777802164054,
      "learning_rate": 8.99097832214975e-06,
      "loss": 0.1923,
      "step": 2512
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.308855569994701,
      "learning_rate": 8.990086136401199e-06,
      "loss": 0.2392,
      "step": 2513
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.511436116101338,
      "learning_rate": 8.989193600695401e-06,
      "loss": 0.212,
      "step": 2514
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.319814267552795,
      "learning_rate": 8.988300715110637e-06,
      "loss": 0.1917,
      "step": 2515
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.426086126711358,
      "learning_rate": 8.987407479725221e-06,
      "loss": 0.2339,
      "step": 2516
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3284088765334046,
      "learning_rate": 8.986513894617493e-06,
      "loss": 0.1973,
      "step": 2517
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.51641291593791,
      "learning_rate": 8.985619959865828e-06,
      "loss": 0.2563,
      "step": 2518
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4553330902198947,
      "learning_rate": 8.98472567554863e-06,
      "loss": 0.2426,
      "step": 2519
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3746933879453835,
      "learning_rate": 8.983831041744333e-06,
      "loss": 0.2112,
      "step": 2520
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.429775698338732,
      "learning_rate": 8.982936058531403e-06,
      "loss": 0.2248,
      "step": 2521
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6721586799612957,
      "learning_rate": 8.982040725988337e-06,
      "loss": 0.2925,
      "step": 2522
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4408970759109248,
      "learning_rate": 8.981145044193659e-06,
      "loss": 0.2408,
      "step": 2523
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4038321969054184,
      "learning_rate": 8.980249013225928e-06,
      "loss": 0.2135,
      "step": 2524
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3925701311200445,
      "learning_rate": 8.979352633163731e-06,
      "loss": 0.2178,
      "step": 2525
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.310671234578445,
      "learning_rate": 8.978455904085687e-06,
      "loss": 0.2401,
      "step": 2526
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3114976764120063,
      "learning_rate": 8.977558826070447e-06,
      "loss": 0.218,
      "step": 2527
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2471191352428226,
      "learning_rate": 8.976661399196689e-06,
      "loss": 0.2354,
      "step": 2528
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5038962087328853,
      "learning_rate": 8.975763623543121e-06,
      "loss": 0.2367,
      "step": 2529
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.41629294862454,
      "learning_rate": 8.974865499188488e-06,
      "loss": 0.2518,
      "step": 2530
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.680754670208158,
      "learning_rate": 8.97396702621156e-06,
      "loss": 0.2688,
      "step": 2531
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3977376756136946,
      "learning_rate": 8.973068204691139e-06,
      "loss": 0.1927,
      "step": 2532
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4877382080580406,
      "learning_rate": 8.972169034706057e-06,
      "loss": 0.2633,
      "step": 2533
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4231064553945656,
      "learning_rate": 8.971269516335178e-06,
      "loss": 0.1901,
      "step": 2534
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.323963089281683,
      "learning_rate": 8.970369649657396e-06,
      "loss": 0.1833,
      "step": 2535
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3349424326402914,
      "learning_rate": 8.969469434751633e-06,
      "loss": 0.234,
      "step": 2536
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4161073322698323,
      "learning_rate": 8.968568871696847e-06,
      "loss": 0.2261,
      "step": 2537
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3341632910416517,
      "learning_rate": 8.967667960572024e-06,
      "loss": 0.2006,
      "step": 2538
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.272103660971173,
      "learning_rate": 8.966766701456177e-06,
      "loss": 0.2071,
      "step": 2539
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2824487644035496,
      "learning_rate": 8.965865094428352e-06,
      "loss": 0.1729,
      "step": 2540
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.450339812574304,
      "learning_rate": 8.96496313956763e-06,
      "loss": 0.2178,
      "step": 2541
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.528822162949716,
      "learning_rate": 8.964060836953115e-06,
      "loss": 0.2658,
      "step": 2542
    },
    {
      "epoch": 0.23,
      "grad_norm": 3.633164502520148,
      "learning_rate": 8.963158186663945e-06,
      "loss": 0.6095,
      "step": 2543
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.388888997051944,
      "learning_rate": 8.96225518877929e-06,
      "loss": 0.5248,
      "step": 2544
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4930851687600113,
      "learning_rate": 8.961351843378349e-06,
      "loss": 0.2252,
      "step": 2545
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2135229008764363,
      "learning_rate": 8.960448150540351e-06,
      "loss": 0.161,
      "step": 2546
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4361667840702064,
      "learning_rate": 8.959544110344554e-06,
      "loss": 0.2032,
      "step": 2547
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5628292865453395,
      "learning_rate": 8.958639722870253e-06,
      "loss": 0.2401,
      "step": 2548
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2642095192372014,
      "learning_rate": 8.957734988196762e-06,
      "loss": 0.1827,
      "step": 2549
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2940874200140966,
      "learning_rate": 8.95682990640344e-06,
      "loss": 0.2436,
      "step": 2550
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.493491781341229,
      "learning_rate": 8.955924477569663e-06,
      "loss": 0.2424,
      "step": 2551
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5107666663828494,
      "learning_rate": 8.955018701774848e-06,
      "loss": 0.2002,
      "step": 2552
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3462376214163252,
      "learning_rate": 8.95411257909843e-06,
      "loss": 0.1922,
      "step": 2553
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2694988178638034,
      "learning_rate": 8.95320610961989e-06,
      "loss": 0.2226,
      "step": 2554
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2461455285208332,
      "learning_rate": 8.952299293418726e-06,
      "loss": 0.2037,
      "step": 2555
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6771359976684608,
      "learning_rate": 8.951392130574476e-06,
      "loss": 0.2793,
      "step": 2556
    },
    {
      "epoch": 0.23,
      "grad_norm": 4.925307218824036,
      "learning_rate": 8.950484621166702e-06,
      "loss": 0.6704,
      "step": 2557
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2715113920665337,
      "learning_rate": 8.949576765274999e-06,
      "loss": 0.2044,
      "step": 2558
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.346442730921242,
      "learning_rate": 8.948668562978992e-06,
      "loss": 0.2121,
      "step": 2559
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.409302058127317,
      "learning_rate": 8.947760014358335e-06,
      "loss": 0.1994,
      "step": 2560
    },
    {
      "epoch": 0.23,
      "grad_norm": 6.200921978600393,
      "learning_rate": 8.946851119492717e-06,
      "loss": 0.5483,
      "step": 2561
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5878062973495015,
      "learning_rate": 8.945941878461851e-06,
      "loss": 0.2187,
      "step": 2562
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.399355554248126,
      "learning_rate": 8.945032291345486e-06,
      "loss": 0.1891,
      "step": 2563
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3067553299005166,
      "learning_rate": 8.944122358223398e-06,
      "loss": 0.1927,
      "step": 2564
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.505321516573947,
      "learning_rate": 8.943212079175392e-06,
      "loss": 0.2145,
      "step": 2565
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3514756748901853,
      "learning_rate": 8.942301454281308e-06,
      "loss": 0.1909,
      "step": 2566
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3797599907295233,
      "learning_rate": 8.941390483621015e-06,
      "loss": 0.2533,
      "step": 2567
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.470512177818092,
      "learning_rate": 8.940479167274409e-06,
      "loss": 0.2394,
      "step": 2568
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4611728809890305,
      "learning_rate": 8.939567505321418e-06,
      "loss": 0.2086,
      "step": 2569
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4930124455617746,
      "learning_rate": 8.938655497842004e-06,
      "loss": 0.2177,
      "step": 2570
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.052805964450612,
      "learning_rate": 8.937743144916153e-06,
      "loss": 0.5323,
      "step": 2571
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4965312600331215,
      "learning_rate": 8.936830446623886e-06,
      "loss": 0.2409,
      "step": 2572
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3809651768316271,
      "learning_rate": 8.935917403045251e-06,
      "loss": 0.2538,
      "step": 2573
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4461061546590126,
      "learning_rate": 8.935004014260331e-06,
      "loss": 0.2322,
      "step": 2574
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5515834392407202,
      "learning_rate": 8.934090280349234e-06,
      "loss": 0.2104,
      "step": 2575
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6614233223747519,
      "learning_rate": 8.933176201392102e-06,
      "loss": 0.2948,
      "step": 2576
    },
    {
      "epoch": 0.24,
      "grad_norm": 6.023262902675161,
      "learning_rate": 8.932261777469105e-06,
      "loss": 0.6442,
      "step": 2577
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5010311935966822,
      "learning_rate": 8.931347008660443e-06,
      "loss": 0.2086,
      "step": 2578
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.460558266944312,
      "learning_rate": 8.930431895046349e-06,
      "loss": 0.5409,
      "step": 2579
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3446641610495387,
      "learning_rate": 8.929516436707086e-06,
      "loss": 0.2357,
      "step": 2580
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.9645764809683377,
      "learning_rate": 8.928600633722944e-06,
      "loss": 0.5205,
      "step": 2581
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4877684355319465,
      "learning_rate": 8.927684486174242e-06,
      "loss": 0.2492,
      "step": 2582
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4315197500981882,
      "learning_rate": 8.926767994141337e-06,
      "loss": 0.2395,
      "step": 2583
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3417980031635441,
      "learning_rate": 8.92585115770461e-06,
      "loss": 0.2572,
      "step": 2584
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3547637663766292,
      "learning_rate": 8.924933976944474e-06,
      "loss": 0.227,
      "step": 2585
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4027747816887266,
      "learning_rate": 8.924016451941371e-06,
      "loss": 0.2223,
      "step": 2586
    },
    {
      "epoch": 0.24,
      "grad_norm": 8.92354310485635,
      "learning_rate": 8.923098582775773e-06,
      "loss": 0.6524,
      "step": 2587
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.353179916248741,
      "learning_rate": 8.922180369528185e-06,
      "loss": 0.193,
      "step": 2588
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5681628527089477,
      "learning_rate": 8.921261812279144e-06,
      "loss": 0.2483,
      "step": 2589
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4139623186305983,
      "learning_rate": 8.920342911109206e-06,
      "loss": 0.2036,
      "step": 2590
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.163014234802205,
      "learning_rate": 8.91942366609897e-06,
      "loss": 0.5861,
      "step": 2591
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4520588310408702,
      "learning_rate": 8.918504077329057e-06,
      "loss": 0.222,
      "step": 2592
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4701383089683207,
      "learning_rate": 8.917584144880124e-06,
      "loss": 0.244,
      "step": 2593
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4493429974967542,
      "learning_rate": 8.916663868832854e-06,
      "loss": 0.2372,
      "step": 2594
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.322608858569593,
      "learning_rate": 8.915743249267959e-06,
      "loss": 0.1956,
      "step": 2595
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3568480954837687,
      "learning_rate": 8.914822286266187e-06,
      "loss": 0.2287,
      "step": 2596
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.643685070236296,
      "learning_rate": 8.913900979908311e-06,
      "loss": 0.6945,
      "step": 2597
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.408885220805745,
      "learning_rate": 8.912979330275136e-06,
      "loss": 0.2278,
      "step": 2598
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3963861476907977,
      "learning_rate": 8.912057337447497e-06,
      "loss": 0.233,
      "step": 2599
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4202802798541063,
      "learning_rate": 8.91113500150626e-06,
      "loss": 0.2696,
      "step": 2600
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3878661204754137,
      "learning_rate": 8.910212322532317e-06,
      "loss": 0.2575,
      "step": 2601
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2944770770197287,
      "learning_rate": 8.909289300606597e-06,
      "loss": 0.1772,
      "step": 2602
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3039252251083997,
      "learning_rate": 8.908365935810052e-06,
      "loss": 0.2106,
      "step": 2603
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3561506610993648,
      "learning_rate": 8.907442228223668e-06,
      "loss": 0.1665,
      "step": 2604
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3636331432129045,
      "learning_rate": 8.90651817792846e-06,
      "loss": 0.1997,
      "step": 2605
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3864313496253062,
      "learning_rate": 8.905593785005476e-06,
      "loss": 0.2124,
      "step": 2606
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4545889614091785,
      "learning_rate": 8.90466904953579e-06,
      "loss": 0.2027,
      "step": 2607
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5911132967331678,
      "learning_rate": 8.903743971600504e-06,
      "loss": 0.2109,
      "step": 2608
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.8550489484479336,
      "learning_rate": 8.902818551280758e-06,
      "loss": 0.6142,
      "step": 2609
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5926670491707802,
      "learning_rate": 8.901892788657718e-06,
      "loss": 0.2508,
      "step": 2610
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2454608644696885,
      "learning_rate": 8.900966683812575e-06,
      "loss": 0.1915,
      "step": 2611
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3786893693519582,
      "learning_rate": 8.900040236826559e-06,
      "loss": 0.1835,
      "step": 2612
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5507356160858219,
      "learning_rate": 8.899113447780923e-06,
      "loss": 0.2451,
      "step": 2613
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6982410303461453,
      "learning_rate": 8.898186316756955e-06,
      "loss": 0.2456,
      "step": 2614
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.282708220421205,
      "learning_rate": 8.897258843835968e-06,
      "loss": 0.1958,
      "step": 2615
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5660946070879418,
      "learning_rate": 8.896331029099308e-06,
      "loss": 0.2495,
      "step": 2616
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.450172872942331,
      "learning_rate": 8.895402872628352e-06,
      "loss": 0.216,
      "step": 2617
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2972466552825948,
      "learning_rate": 8.894474374504506e-06,
      "loss": 0.1935,
      "step": 2618
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3048852678478027,
      "learning_rate": 8.893545534809202e-06,
      "loss": 0.1896,
      "step": 2619
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4419369912511197,
      "learning_rate": 8.89261635362391e-06,
      "loss": 0.2205,
      "step": 2620
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3809129159411464,
      "learning_rate": 8.891686831030121e-06,
      "loss": 0.2193,
      "step": 2621
    },
    {
      "epoch": 0.24,
      "grad_norm": 5.745947306907133,
      "learning_rate": 8.890756967109364e-06,
      "loss": 0.6133,
      "step": 2622
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3250898928672727,
      "learning_rate": 8.889826761943192e-06,
      "loss": 0.1868,
      "step": 2623
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.530758374227946,
      "learning_rate": 8.888896215613192e-06,
      "loss": 0.175,
      "step": 2624
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.403943782630684,
      "learning_rate": 8.887965328200975e-06,
      "loss": 0.4559,
      "step": 2625
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2438972328397786,
      "learning_rate": 8.887034099788194e-06,
      "loss": 0.2005,
      "step": 2626
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6894603416547025,
      "learning_rate": 8.886102530456514e-06,
      "loss": 0.226,
      "step": 2627
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4858070603114066,
      "learning_rate": 8.885170620287649e-06,
      "loss": 0.2194,
      "step": 2628
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3487539449300299,
      "learning_rate": 8.884238369363329e-06,
      "loss": 0.2194,
      "step": 2629
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.707240166366971,
      "learning_rate": 8.883305777765319e-06,
      "loss": 0.1985,
      "step": 2630
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3998711968184818,
      "learning_rate": 8.882372845575414e-06,
      "loss": 0.2015,
      "step": 2631
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6308182507556723,
      "learning_rate": 8.881439572875436e-06,
      "loss": 0.2541,
      "step": 2632
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4501278859253715,
      "learning_rate": 8.880505959747245e-06,
      "loss": 0.2425,
      "step": 2633
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.457476565490881,
      "learning_rate": 8.879572006272718e-06,
      "loss": 0.2201,
      "step": 2634
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5124161112526948,
      "learning_rate": 8.878637712533776e-06,
      "loss": 0.1892,
      "step": 2635
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5438943124107825,
      "learning_rate": 8.877703078612358e-06,
      "loss": 0.1972,
      "step": 2636
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4196616879066415,
      "learning_rate": 8.876768104590441e-06,
      "loss": 0.198,
      "step": 2637
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4186373195296849,
      "learning_rate": 8.875832790550024e-06,
      "loss": 0.1877,
      "step": 2638
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5340305414103181,
      "learning_rate": 8.874897136573143e-06,
      "loss": 0.2115,
      "step": 2639
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4881616258922918,
      "learning_rate": 8.873961142741862e-06,
      "loss": 0.2137,
      "step": 2640
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4227854174556658,
      "learning_rate": 8.873024809138272e-06,
      "loss": 0.2418,
      "step": 2641
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.333227769061717,
      "learning_rate": 8.8720881358445e-06,
      "loss": 0.1601,
      "step": 2642
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3240228299423566,
      "learning_rate": 8.871151122942692e-06,
      "loss": 0.1919,
      "step": 2643
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.386376039477144,
      "learning_rate": 8.870213770515034e-06,
      "loss": 0.1944,
      "step": 2644
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3056169832412083,
      "learning_rate": 8.869276078643739e-06,
      "loss": 0.2211,
      "step": 2645
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4703056571302606,
      "learning_rate": 8.868338047411045e-06,
      "loss": 0.2245,
      "step": 2646
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.388987156426266,
      "learning_rate": 8.867399676899227e-06,
      "loss": 0.1988,
      "step": 2647
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2342980182308303,
      "learning_rate": 8.866460967190584e-06,
      "loss": 0.1913,
      "step": 2648
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5702963312472642,
      "learning_rate": 8.86552191836745e-06,
      "loss": 0.2513,
      "step": 2649
    },
    {
      "epoch": 0.24,
      "grad_norm": 12.819905614450835,
      "learning_rate": 8.864582530512185e-06,
      "loss": 0.5581,
      "step": 2650
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3199110408392098,
      "learning_rate": 8.863642803707177e-06,
      "loss": 0.2119,
      "step": 2651
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.406917541563749,
      "learning_rate": 8.862702738034851e-06,
      "loss": 0.2123,
      "step": 2652
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.411445116685656,
      "learning_rate": 8.861762333577653e-06,
      "loss": 0.1867,
      "step": 2653
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.364076684376785,
      "learning_rate": 8.860821590418062e-06,
      "loss": 0.2431,
      "step": 2654
    },
    {
      "epoch": 0.24,
      "grad_norm": 7.4248967424925745,
      "learning_rate": 8.859880508638592e-06,
      "loss": 0.6468,
      "step": 2655
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3683364783743535,
      "learning_rate": 8.858939088321781e-06,
      "loss": 0.2083,
      "step": 2656
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5112487706131637,
      "learning_rate": 8.857997329550195e-06,
      "loss": 0.2269,
      "step": 2657
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4141794836826682,
      "learning_rate": 8.857055232406432e-06,
      "loss": 0.2191,
      "step": 2658
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4532104894883155,
      "learning_rate": 8.856112796973125e-06,
      "loss": 0.2391,
      "step": 2659
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.271410558122883,
      "learning_rate": 8.855170023332928e-06,
      "loss": 0.2037,
      "step": 2660
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4724919579553468,
      "learning_rate": 8.854226911568532e-06,
      "loss": 0.239,
      "step": 2661
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.868518481417162,
      "learning_rate": 8.85328346176265e-06,
      "loss": 0.6202,
      "step": 2662
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.236455031226945,
      "learning_rate": 8.85233967399803e-06,
      "loss": 0.1739,
      "step": 2663
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3505030346821674,
      "learning_rate": 8.85139554835745e-06,
      "loss": 0.2124,
      "step": 2664
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3925128675558625,
      "learning_rate": 8.850451084923717e-06,
      "loss": 0.2122,
      "step": 2665
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.280585752347895,
      "learning_rate": 8.849506283779665e-06,
      "loss": 0.1933,
      "step": 2666
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4682331045184873,
      "learning_rate": 8.848561145008158e-06,
      "loss": 0.192,
      "step": 2667
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3720553737036847,
      "learning_rate": 8.847615668692094e-06,
      "loss": 0.24,
      "step": 2668
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.684412511193261,
      "learning_rate": 8.846669854914395e-06,
      "loss": 0.6648,
      "step": 2669
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4445516762135102,
      "learning_rate": 8.845723703758017e-06,
      "loss": 0.2055,
      "step": 2670
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3349785626190371,
      "learning_rate": 8.844777215305944e-06,
      "loss": 0.1795,
      "step": 2671
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2671912863468733,
      "learning_rate": 8.843830389641189e-06,
      "loss": 0.2033,
      "step": 2672
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.35836160884706,
      "learning_rate": 8.842883226846792e-06,
      "loss": 0.1871,
      "step": 2673
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4360291277778017,
      "learning_rate": 8.84193572700583e-06,
      "loss": 0.2348,
      "step": 2674
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3913305326089387,
      "learning_rate": 8.840987890201404e-06,
      "loss": 0.2307,
      "step": 2675
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2240753399762305,
      "learning_rate": 8.840039716516642e-06,
      "loss": 0.1592,
      "step": 2676
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6207326923905498,
      "learning_rate": 8.839091206034712e-06,
      "loss": 0.1948,
      "step": 2677
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4348246092988404,
      "learning_rate": 8.838142358838797e-06,
      "loss": 0.2505,
      "step": 2678
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.073780668133791,
      "learning_rate": 8.837193175012121e-06,
      "loss": 0.4231,
      "step": 2679
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.397780559539884,
      "learning_rate": 8.836243654637936e-06,
      "loss": 0.5545,
      "step": 2680
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.523885080179518,
      "learning_rate": 8.835293797799517e-06,
      "loss": 0.1843,
      "step": 2681
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2467049669942634,
      "learning_rate": 8.834343604580175e-06,
      "loss": 0.2074,
      "step": 2682
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.684166882131869,
      "learning_rate": 8.833393075063248e-06,
      "loss": 0.2551,
      "step": 2683
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3944272809224858,
      "learning_rate": 8.832442209332104e-06,
      "loss": 0.2096,
      "step": 2684
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.109012438068451,
      "learning_rate": 8.831491007470142e-06,
      "loss": 0.5713,
      "step": 2685
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4522548734707974,
      "learning_rate": 8.830539469560784e-06,
      "loss": 0.2216,
      "step": 2686
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.416977885544616,
      "learning_rate": 8.829587595687493e-06,
      "loss": 0.1934,
      "step": 2687
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.36674208475292,
      "learning_rate": 8.828635385933747e-06,
      "loss": 0.6204,
      "step": 2688
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5161504784553719,
      "learning_rate": 8.827682840383065e-06,
      "loss": 0.2499,
      "step": 2689
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3172746538371762,
      "learning_rate": 8.826729959118993e-06,
      "loss": 0.1695,
      "step": 2690
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5226642947698525,
      "learning_rate": 8.825776742225104e-06,
      "loss": 0.2331,
      "step": 2691
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4227270091459985,
      "learning_rate": 8.824823189785001e-06,
      "loss": 0.254,
      "step": 2692
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.29644855254401,
      "learning_rate": 8.823869301882318e-06,
      "loss": 0.5948,
      "step": 2693
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4727084551945542,
      "learning_rate": 8.822915078600715e-06,
      "loss": 0.2739,
      "step": 2694
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.657368682014743,
      "learning_rate": 8.821960520023884e-06,
      "loss": 0.1991,
      "step": 2695
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.19259717881787,
      "learning_rate": 8.82100562623555e-06,
      "loss": 0.162,
      "step": 2696
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.452019495214713,
      "learning_rate": 8.82005039731946e-06,
      "loss": 0.233,
      "step": 2697
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3059258253664485,
      "learning_rate": 8.819094833359395e-06,
      "loss": 0.2064,
      "step": 2698
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5354910604892271,
      "learning_rate": 8.818138934439165e-06,
      "loss": 0.2678,
      "step": 2699
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5960173719456165,
      "learning_rate": 8.817182700642606e-06,
      "loss": 0.2999,
      "step": 2700
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4296261158178816,
      "learning_rate": 8.816226132053588e-06,
      "loss": 0.206,
      "step": 2701
    },
    {
      "epoch": 0.25,
      "grad_norm": 13.302388968034903,
      "learning_rate": 8.815269228756012e-06,
      "loss": 0.6181,
      "step": 2702
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.43210593046343,
      "learning_rate": 8.8143119908338e-06,
      "loss": 0.2514,
      "step": 2703
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4651523318074482,
      "learning_rate": 8.813354418370908e-06,
      "loss": 0.2386,
      "step": 2704
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.510769828598478,
      "learning_rate": 8.812396511451324e-06,
      "loss": 0.2143,
      "step": 2705
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5037535871823873,
      "learning_rate": 8.811438270159064e-06,
      "loss": 0.1928,
      "step": 2706
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.452124601202633,
      "learning_rate": 8.810479694578166e-06,
      "loss": 0.2427,
      "step": 2707
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.481177869652,
      "learning_rate": 8.80952078479271e-06,
      "loss": 0.2035,
      "step": 2708
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.528365627938233,
      "learning_rate": 8.808561540886796e-06,
      "loss": 0.2523,
      "step": 2709
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2407619476385168,
      "learning_rate": 8.807601962944556e-06,
      "loss": 0.1643,
      "step": 2710
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4309919405991598,
      "learning_rate": 8.806642051050154e-06,
      "loss": 0.2447,
      "step": 2711
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.446179654260049,
      "learning_rate": 8.805681805287776e-06,
      "loss": 0.2147,
      "step": 2712
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6344846182331243,
      "learning_rate": 8.804721225741646e-06,
      "loss": 0.2295,
      "step": 2713
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4480957939943162,
      "learning_rate": 8.803760312496012e-06,
      "loss": 0.1934,
      "step": 2714
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7764548088510819,
      "learning_rate": 8.802799065635152e-06,
      "loss": 0.2728,
      "step": 2715
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3911427081582952,
      "learning_rate": 8.801837485243373e-06,
      "loss": 0.2405,
      "step": 2716
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5308378515032564,
      "learning_rate": 8.800875571405015e-06,
      "loss": 0.2827,
      "step": 2717
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.500783160663961,
      "learning_rate": 8.799913324204441e-06,
      "loss": 0.2738,
      "step": 2718
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3751540126540214,
      "learning_rate": 8.79895074372605e-06,
      "loss": 0.1807,
      "step": 2719
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.452181812093203,
      "learning_rate": 8.797987830054264e-06,
      "loss": 0.6029,
      "step": 2720
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2532993820945355,
      "learning_rate": 8.797024583273536e-06,
      "loss": 0.195,
      "step": 2721
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4344683509171596,
      "learning_rate": 8.796061003468353e-06,
      "loss": 0.2436,
      "step": 2722
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.967819872173226,
      "learning_rate": 8.795097090723224e-06,
      "loss": 0.609,
      "step": 2723
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.027407254999556,
      "learning_rate": 8.794132845122692e-06,
      "loss": 0.5701,
      "step": 2724
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.362925959966901,
      "learning_rate": 8.793168266751327e-06,
      "loss": 0.2282,
      "step": 2725
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3202421670629054,
      "learning_rate": 8.792203355693731e-06,
      "loss": 0.1861,
      "step": 2726
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.638930983612814,
      "learning_rate": 8.791238112034531e-06,
      "loss": 0.2798,
      "step": 2727
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.520527987712115,
      "learning_rate": 8.790272535858387e-06,
      "loss": 0.2544,
      "step": 2728
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.355730492973801,
      "learning_rate": 8.789306627249985e-06,
      "loss": 0.2601,
      "step": 2729
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2706234103337484,
      "learning_rate": 8.78834038629404e-06,
      "loss": 0.1501,
      "step": 2730
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5548619729148645,
      "learning_rate": 8.787373813075304e-06,
      "loss": 0.2786,
      "step": 2731
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.480156602607674,
      "learning_rate": 8.786406907678547e-06,
      "loss": 0.1971,
      "step": 2732
    },
    {
      "epoch": 0.25,
      "grad_norm": 6.344777076506671,
      "learning_rate": 8.78543967018857e-06,
      "loss": 0.6209,
      "step": 2733
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3191865135897693,
      "learning_rate": 8.784472100690215e-06,
      "loss": 0.236,
      "step": 2734
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.137803084449063,
      "learning_rate": 8.783504199268339e-06,
      "loss": 0.5338,
      "step": 2735
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4959337976084868,
      "learning_rate": 8.782535966007832e-06,
      "loss": 0.2647,
      "step": 2736
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.541060287598577,
      "learning_rate": 8.781567400993617e-06,
      "loss": 0.2302,
      "step": 2737
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.514560986116313,
      "learning_rate": 8.780598504310645e-06,
      "loss": 0.2152,
      "step": 2738
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5111282681400804,
      "learning_rate": 8.779629276043891e-06,
      "loss": 0.2173,
      "step": 2739
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5344120334178297,
      "learning_rate": 8.778659716278366e-06,
      "loss": 0.2114,
      "step": 2740
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3768201396102293,
      "learning_rate": 8.777689825099104e-06,
      "loss": 0.2075,
      "step": 2741
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3786914576084828,
      "learning_rate": 8.776719602591176e-06,
      "loss": 0.2413,
      "step": 2742
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3649061700793936,
      "learning_rate": 8.77574904883967e-06,
      "loss": 0.1871,
      "step": 2743
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5613242955184297,
      "learning_rate": 8.774778163929714e-06,
      "loss": 0.1691,
      "step": 2744
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.267176581924431,
      "learning_rate": 8.77380694794646e-06,
      "loss": 0.16,
      "step": 2745
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5390211486178609,
      "learning_rate": 8.772835400975093e-06,
      "loss": 0.252,
      "step": 2746
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4634852453730784,
      "learning_rate": 8.771863523100821e-06,
      "loss": 0.2507,
      "step": 2747
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3571363804454724,
      "learning_rate": 8.770891314408885e-06,
      "loss": 0.1744,
      "step": 2748
    },
    {
      "epoch": 0.25,
      "grad_norm": 8.595144007390239,
      "learning_rate": 8.769918774984553e-06,
      "loss": 0.6036,
      "step": 2749
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4530430251876856,
      "learning_rate": 8.768945904913127e-06,
      "loss": 0.2438,
      "step": 2750
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.685706931835743,
      "learning_rate": 8.767972704279932e-06,
      "loss": 0.6698,
      "step": 2751
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5297656184685322,
      "learning_rate": 8.766999173170323e-06,
      "loss": 0.1943,
      "step": 2752
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3523528753648464,
      "learning_rate": 8.766025311669685e-06,
      "loss": 0.2063,
      "step": 2753
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4185878566677856,
      "learning_rate": 8.765051119863435e-06,
      "loss": 0.2264,
      "step": 2754
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.333203564546402,
      "learning_rate": 8.764076597837014e-06,
      "loss": 0.1917,
      "step": 2755
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4168313630037517,
      "learning_rate": 8.763101745675894e-06,
      "loss": 0.201,
      "step": 2756
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4304595958077646,
      "learning_rate": 8.762126563465578e-06,
      "loss": 0.2184,
      "step": 2757
    },
    {
      "epoch": 0.25,
      "grad_norm": 10.113653547431802,
      "learning_rate": 8.761151051291593e-06,
      "loss": 0.5299,
      "step": 2758
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.532783482239178,
      "learning_rate": 8.760175209239501e-06,
      "loss": 0.207,
      "step": 2759
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5596264911651263,
      "learning_rate": 8.759199037394888e-06,
      "loss": 0.3011,
      "step": 2760
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.330097600683855,
      "learning_rate": 8.75822253584337e-06,
      "loss": 0.226,
      "step": 2761
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.2387324479183075,
      "learning_rate": 8.757245704670596e-06,
      "loss": 0.4224,
      "step": 2762
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5759969835822911,
      "learning_rate": 8.756268543962238e-06,
      "loss": 0.2583,
      "step": 2763
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3646219167455416,
      "learning_rate": 8.755291053804e-06,
      "loss": 0.2075,
      "step": 2764
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4023658568338615,
      "learning_rate": 8.754313234281614e-06,
      "loss": 0.1853,
      "step": 2765
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.101742846123356,
      "learning_rate": 8.753335085480844e-06,
      "loss": 0.6147,
      "step": 2766
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4959685576121116,
      "learning_rate": 8.752356607487479e-06,
      "loss": 0.2325,
      "step": 2767
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.68607552650466,
      "learning_rate": 8.751377800387333e-06,
      "loss": 0.5041,
      "step": 2768
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4173995225004448,
      "learning_rate": 8.75039866426626e-06,
      "loss": 0.2385,
      "step": 2769
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2593608122397117,
      "learning_rate": 8.749419199210135e-06,
      "loss": 0.2153,
      "step": 2770
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5691009845547563,
      "learning_rate": 8.748439405304863e-06,
      "loss": 0.2443,
      "step": 2771
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4211562040688785,
      "learning_rate": 8.747459282636378e-06,
      "loss": 0.1774,
      "step": 2772
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2346441712272551,
      "learning_rate": 8.746478831290648e-06,
      "loss": 0.1909,
      "step": 2773
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4100785069252146,
      "learning_rate": 8.745498051353658e-06,
      "loss": 0.2261,
      "step": 2774
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4943052020190613,
      "learning_rate": 8.744516942911433e-06,
      "loss": 0.2693,
      "step": 2775
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.443620905579626,
      "learning_rate": 8.743535506050023e-06,
      "loss": 0.6,
      "step": 2776
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2478153676600177,
      "learning_rate": 8.742553740855507e-06,
      "loss": 0.2303,
      "step": 2777
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4064386472029762,
      "learning_rate": 8.741571647413989e-06,
      "loss": 0.2194,
      "step": 2778
    },
    {
      "epoch": 0.25,
      "grad_norm": 5.214070455054304,
      "learning_rate": 8.740589225811608e-06,
      "loss": 0.5159,
      "step": 2779
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.511830235483976,
      "learning_rate": 8.739606476134528e-06,
      "loss": 0.202,
      "step": 2780
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4385608881884053,
      "learning_rate": 8.738623398468945e-06,
      "loss": 0.2473,
      "step": 2781
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2053591206038663,
      "learning_rate": 8.737639992901076e-06,
      "loss": 0.1801,
      "step": 2782
    },
    {
      "epoch": 0.25,
      "grad_norm": 14.818510583104214,
      "learning_rate": 8.736656259517178e-06,
      "loss": 0.6006,
      "step": 2783
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.384744541828624,
      "learning_rate": 8.73567219840353e-06,
      "loss": 0.2235,
      "step": 2784
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5688169706338626,
      "learning_rate": 8.734687809646437e-06,
      "loss": 0.247,
      "step": 2785
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4013559410330212,
      "learning_rate": 8.733703093332237e-06,
      "loss": 0.193,
      "step": 2786
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1730937172313867,
      "learning_rate": 8.732718049547302e-06,
      "loss": 0.1628,
      "step": 2787
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5456196050546154,
      "learning_rate": 8.731732678378022e-06,
      "loss": 0.2432,
      "step": 2788
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.413401126258453,
      "learning_rate": 8.730746979910823e-06,
      "loss": 0.6863,
      "step": 2789
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3664194049941756,
      "learning_rate": 8.729760954232154e-06,
      "loss": 0.2451,
      "step": 2790
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3861534335171242,
      "learning_rate": 8.728774601428498e-06,
      "loss": 0.1777,
      "step": 2791
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3018429663259843,
      "learning_rate": 8.727787921586366e-06,
      "loss": 0.1938,
      "step": 2792
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4999838173904525,
      "learning_rate": 8.726800914792296e-06,
      "loss": 0.2145,
      "step": 2793
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2281126322478102,
      "learning_rate": 8.725813581132852e-06,
      "loss": 0.1917,
      "step": 2794
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4036862457173787,
      "learning_rate": 8.724825920694632e-06,
      "loss": 0.2041,
      "step": 2795
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5909706020252834,
      "learning_rate": 8.723837933564263e-06,
      "loss": 0.2518,
      "step": 2796
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.517621220228533,
      "learning_rate": 8.722849619828395e-06,
      "loss": 0.2033,
      "step": 2797
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2814172033046325,
      "learning_rate": 8.72186097957371e-06,
      "loss": 0.1834,
      "step": 2798
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5329356756110564,
      "learning_rate": 8.720872012886918e-06,
      "loss": 0.2388,
      "step": 2799
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.08922651367435,
      "learning_rate": 8.71988271985476e-06,
      "loss": 0.556,
      "step": 2800
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.327901323499165,
      "learning_rate": 8.718893100564002e-06,
      "loss": 0.2195,
      "step": 2801
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.683514002723553,
      "learning_rate": 8.717903155101441e-06,
      "loss": 0.6331,
      "step": 2802
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5259548301580645,
      "learning_rate": 8.7169128835539e-06,
      "loss": 0.2266,
      "step": 2803
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.633843113527126,
      "learning_rate": 8.715922286008234e-06,
      "loss": 0.7369,
      "step": 2804
    },
    {
      "epoch": 0.26,
      "grad_norm": 22.130424143693208,
      "learning_rate": 8.714931362551326e-06,
      "loss": 0.6185,
      "step": 2805
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3746923448768322,
      "learning_rate": 8.713940113270085e-06,
      "loss": 0.2117,
      "step": 2806
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5347932736964316,
      "learning_rate": 8.71294853825145e-06,
      "loss": 0.2489,
      "step": 2807
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.622240699289381,
      "learning_rate": 8.711956637582391e-06,
      "loss": 0.2507,
      "step": 2808
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.53468165135485,
      "learning_rate": 8.710964411349902e-06,
      "loss": 0.2408,
      "step": 2809
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3678325502697504,
      "learning_rate": 8.70997185964101e-06,
      "loss": 0.2431,
      "step": 2810
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6774769438030313,
      "learning_rate": 8.708978982542766e-06,
      "loss": 0.2273,
      "step": 2811
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.369966543615706,
      "learning_rate": 8.707985780142252e-06,
      "loss": 0.2059,
      "step": 2812
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3201022812152818,
      "learning_rate": 8.706992252526581e-06,
      "loss": 0.1851,
      "step": 2813
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3170577374171581,
      "learning_rate": 8.70599839978289e-06,
      "loss": 0.206,
      "step": 2814
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2750840767622171,
      "learning_rate": 8.705004221998348e-06,
      "loss": 0.2002,
      "step": 2815
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5352527555807254,
      "learning_rate": 8.70400971926015e-06,
      "loss": 0.238,
      "step": 2816
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.2308701024456505,
      "learning_rate": 8.703014891655518e-06,
      "loss": 0.5392,
      "step": 2817
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4596448621313691,
      "learning_rate": 8.70201973927171e-06,
      "loss": 0.2002,
      "step": 2818
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.595251934953582,
      "learning_rate": 8.701024262196005e-06,
      "loss": 0.207,
      "step": 2819
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.565453492353938,
      "learning_rate": 8.700028460515714e-06,
      "loss": 0.4497,
      "step": 2820
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2866090699074593,
      "learning_rate": 8.699032334318173e-06,
      "loss": 0.1793,
      "step": 2821
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3666587293182937,
      "learning_rate": 8.698035883690752e-06,
      "loss": 0.2395,
      "step": 2822
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.624972269877882,
      "learning_rate": 8.697039108720844e-06,
      "loss": 0.6661,
      "step": 2823
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.370477768409505,
      "learning_rate": 8.696042009495876e-06,
      "loss": 0.2185,
      "step": 2824
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.454207139099368,
      "learning_rate": 8.695044586103297e-06,
      "loss": 0.5622,
      "step": 2825
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4940853320522993,
      "learning_rate": 8.694046838630588e-06,
      "loss": 0.2259,
      "step": 2826
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.359624529226148,
      "learning_rate": 8.69304876716526e-06,
      "loss": 0.2217,
      "step": 2827
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3401376433958347,
      "learning_rate": 8.692050371794849e-06,
      "loss": 0.1855,
      "step": 2828
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5071133274971111,
      "learning_rate": 8.69105165260692e-06,
      "loss": 0.2186,
      "step": 2829
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3348957354273863,
      "learning_rate": 8.690052609689071e-06,
      "loss": 0.2126,
      "step": 2830
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.433672816205031,
      "learning_rate": 8.689053243128921e-06,
      "loss": 0.423,
      "step": 2831
    },
    {
      "epoch": 0.26,
      "grad_norm": 6.39034779019586,
      "learning_rate": 8.688053553014124e-06,
      "loss": 0.643,
      "step": 2832
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4551062556812213,
      "learning_rate": 8.687053539432358e-06,
      "loss": 0.1911,
      "step": 2833
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4494836437200287,
      "learning_rate": 8.686053202471331e-06,
      "loss": 0.2201,
      "step": 2834
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3857721597209247,
      "learning_rate": 8.68505254221878e-06,
      "loss": 0.2224,
      "step": 2835
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4558465929769808,
      "learning_rate": 8.684051558762468e-06,
      "loss": 0.2377,
      "step": 2836
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.369862197206661,
      "learning_rate": 8.683050252190189e-06,
      "loss": 0.2204,
      "step": 2837
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2424351704929033,
      "learning_rate": 8.682048622589764e-06,
      "loss": 0.1916,
      "step": 2838
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5742490373642295,
      "learning_rate": 8.681046670049043e-06,
      "loss": 0.2569,
      "step": 2839
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.579582826853813,
      "learning_rate": 8.680044394655905e-06,
      "loss": 0.2621,
      "step": 2840
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.601027752662453,
      "learning_rate": 8.679041796498253e-06,
      "loss": 0.3248,
      "step": 2841
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4364991466756853,
      "learning_rate": 8.678038875664024e-06,
      "loss": 0.2455,
      "step": 2842
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2715448728219532,
      "learning_rate": 8.677035632241181e-06,
      "loss": 0.1859,
      "step": 2843
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.526460426672062,
      "learning_rate": 8.676032066317715e-06,
      "loss": 0.2471,
      "step": 2844
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.429854389835442,
      "learning_rate": 8.675028177981643e-06,
      "loss": 0.2336,
      "step": 2845
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.535945476252859,
      "learning_rate": 8.674023967321018e-06,
      "loss": 0.2047,
      "step": 2846
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4055949785684188,
      "learning_rate": 8.67301943442391e-06,
      "loss": 0.2154,
      "step": 2847
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4649400027668291,
      "learning_rate": 8.672014579378428e-06,
      "loss": 0.2216,
      "step": 2848
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5847666107816125,
      "learning_rate": 8.6710094022727e-06,
      "loss": 0.2209,
      "step": 2849
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3248446577161528,
      "learning_rate": 8.67000390319489e-06,
      "loss": 0.1948,
      "step": 2850
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3344703126462387,
      "learning_rate": 8.668998082233186e-06,
      "loss": 0.1629,
      "step": 2851
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4364880922372045,
      "learning_rate": 8.667991939475808e-06,
      "loss": 0.19,
      "step": 2852
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.219931472703374,
      "learning_rate": 8.666985475010995e-06,
      "loss": 0.2124,
      "step": 2853
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4237794789235996,
      "learning_rate": 8.665978688927029e-06,
      "loss": 0.2414,
      "step": 2854
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.328100817587429,
      "learning_rate": 8.664971581312205e-06,
      "loss": 0.2122,
      "step": 2855
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3390131255293887,
      "learning_rate": 8.663964152254855e-06,
      "loss": 0.2114,
      "step": 2856
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4451226853758583,
      "learning_rate": 8.66295640184334e-06,
      "loss": 0.2229,
      "step": 2857
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4255479533445061,
      "learning_rate": 8.661948330166042e-06,
      "loss": 0.2355,
      "step": 2858
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.357321883440389,
      "learning_rate": 8.66093993731138e-06,
      "loss": 0.4447,
      "step": 2859
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2579147426611657,
      "learning_rate": 8.659931223367792e-06,
      "loss": 0.2018,
      "step": 2860
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3914243770219143,
      "learning_rate": 8.658922188423753e-06,
      "loss": 0.2118,
      "step": 2861
    },
    {
      "epoch": 0.26,
      "grad_norm": 5.940170952638645,
      "learning_rate": 8.65791283256776e-06,
      "loss": 0.6528,
      "step": 2862
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2818134279250104,
      "learning_rate": 8.656903155888343e-06,
      "loss": 0.1973,
      "step": 2863
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3337457071476044,
      "learning_rate": 8.655893158474056e-06,
      "loss": 0.2523,
      "step": 2864
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.468452075269291,
      "learning_rate": 8.65488284041348e-06,
      "loss": 0.1874,
      "step": 2865
    },
    {
      "epoch": 0.26,
      "grad_norm": 4.673275978093116,
      "learning_rate": 8.653872201795229e-06,
      "loss": 0.6377,
      "step": 2866
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4626333327889345,
      "learning_rate": 8.652861242707941e-06,
      "loss": 0.2126,
      "step": 2867
    },
    {
      "epoch": 0.26,
      "grad_norm": 3.9707541133120574,
      "learning_rate": 8.651849963240287e-06,
      "loss": 0.6012,
      "step": 2868
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3705851126652517,
      "learning_rate": 8.650838363480961e-06,
      "loss": 0.2337,
      "step": 2869
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4674890777887228,
      "learning_rate": 8.649826443518687e-06,
      "loss": 0.2272,
      "step": 2870
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4824773461085612,
      "learning_rate": 8.648814203442217e-06,
      "loss": 0.2421,
      "step": 2871
    },
    {
      "epoch": 0.26,
      "grad_norm": 7.082845758093708,
      "learning_rate": 8.647801643340332e-06,
      "loss": 0.6489,
      "step": 2872
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4504532060811035,
      "learning_rate": 8.646788763301842e-06,
      "loss": 0.2194,
      "step": 2873
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.685524959611446,
      "learning_rate": 8.645775563415579e-06,
      "loss": 0.2297,
      "step": 2874
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4561450534990392,
      "learning_rate": 8.64476204377041e-06,
      "loss": 0.1892,
      "step": 2875
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2439055085193225,
      "learning_rate": 8.643748204455228e-06,
      "loss": 0.1879,
      "step": 2876
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4064612186336503,
      "learning_rate": 8.642734045558952e-06,
      "loss": 0.2363,
      "step": 2877
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2181209066511636,
      "learning_rate": 8.641719567170532e-06,
      "loss": 0.1645,
      "step": 2878
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2427301480218735,
      "learning_rate": 8.640704769378943e-06,
      "loss": 0.2057,
      "step": 2879
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4327281381064976,
      "learning_rate": 8.639689652273189e-06,
      "loss": 0.1807,
      "step": 2880
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3146661723276016,
      "learning_rate": 8.638674215942307e-06,
      "loss": 0.1862,
      "step": 2881
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.385556041678885,
      "learning_rate": 8.637658460475352e-06,
      "loss": 0.2115,
      "step": 2882
    },
    {
      "epoch": 0.26,
      "grad_norm": 10.922651270346815,
      "learning_rate": 8.636642385961416e-06,
      "loss": 0.4967,
      "step": 2883
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.591098024083333,
      "learning_rate": 8.635625992489613e-06,
      "loss": 0.264,
      "step": 2884
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3418714599421615,
      "learning_rate": 8.634609280149091e-06,
      "loss": 0.2138,
      "step": 2885
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3567547922409562,
      "learning_rate": 8.633592249029021e-06,
      "loss": 0.2351,
      "step": 2886
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4741969954130034,
      "learning_rate": 8.6325748992186e-06,
      "loss": 0.2043,
      "step": 2887
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4983552750843872,
      "learning_rate": 8.631557230807062e-06,
      "loss": 0.1555,
      "step": 2888
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5269267837145426,
      "learning_rate": 8.630539243883659e-06,
      "loss": 0.2159,
      "step": 2889
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5447903052965142,
      "learning_rate": 8.629520938537677e-06,
      "loss": 0.2573,
      "step": 2890
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.443768983099507,
      "learning_rate": 8.628502314858429e-06,
      "loss": 0.2252,
      "step": 2891
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.394074721954043,
      "learning_rate": 8.627483372935253e-06,
      "loss": 0.2303,
      "step": 2892
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4955298121280056,
      "learning_rate": 8.626464112857519e-06,
      "loss": 0.2317,
      "step": 2893
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.425513302302495,
      "learning_rate": 8.625444534714623e-06,
      "loss": 0.232,
      "step": 2894
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2528633994390528,
      "learning_rate": 8.624424638595985e-06,
      "loss": 0.2021,
      "step": 2895
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2884855991965127,
      "learning_rate": 8.623404424591062e-06,
      "loss": 0.275,
      "step": 2896
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5588455948193598,
      "learning_rate": 8.62238389278933e-06,
      "loss": 0.229,
      "step": 2897
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3371167749161788,
      "learning_rate": 8.6213630432803e-06,
      "loss": 0.1854,
      "step": 2898
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.247787195196842,
      "learning_rate": 8.620341876153505e-06,
      "loss": 0.1756,
      "step": 2899
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4871710148993158,
      "learning_rate": 8.619320391498507e-06,
      "loss": 0.2244,
      "step": 2900
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5712997423012736,
      "learning_rate": 8.618298589404896e-06,
      "loss": 0.2388,
      "step": 2901
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.611888609220342,
      "learning_rate": 8.617276469962296e-06,
      "loss": 0.5517,
      "step": 2902
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.66082275002613,
      "learning_rate": 8.616254033260351e-06,
      "loss": 0.2859,
      "step": 2903
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.434678184466059,
      "learning_rate": 8.615231279388734e-06,
      "loss": 0.2388,
      "step": 2904
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.056239814202053,
      "learning_rate": 8.61420820843715e-06,
      "loss": 0.6534,
      "step": 2905
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4217992566993243,
      "learning_rate": 8.613184820495326e-06,
      "loss": 0.2096,
      "step": 2906
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6606335621820318,
      "learning_rate": 8.612161115653023e-06,
      "loss": 0.2301,
      "step": 2907
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.527041033360132,
      "learning_rate": 8.611137094000028e-06,
      "loss": 0.1904,
      "step": 2908
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5726136246444682,
      "learning_rate": 8.61011275562615e-06,
      "loss": 0.2732,
      "step": 2909
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5120665371189788,
      "learning_rate": 8.609088100621232e-06,
      "loss": 0.2864,
      "step": 2910
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.423583272299464,
      "learning_rate": 8.608063129075143e-06,
      "loss": 0.2292,
      "step": 2911
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.930820835171568,
      "learning_rate": 8.607037841077782e-06,
      "loss": 0.6303,
      "step": 2912
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.198980919676222,
      "learning_rate": 8.606012236719073e-06,
      "loss": 0.6511,
      "step": 2913
    },
    {
      "epoch": 0.27,
      "grad_norm": 11.654773999478653,
      "learning_rate": 8.604986316088967e-06,
      "loss": 0.6719,
      "step": 2914
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4088028138363284,
      "learning_rate": 8.603960079277445e-06,
      "loss": 0.2171,
      "step": 2915
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4381984998770976,
      "learning_rate": 8.602933526374512e-06,
      "loss": 0.2615,
      "step": 2916
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.821719461030636,
      "learning_rate": 8.601906657470209e-06,
      "loss": 0.5482,
      "step": 2917
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.33488206854032,
      "learning_rate": 8.600879472654595e-06,
      "loss": 0.206,
      "step": 2918
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.401810654701214,
      "learning_rate": 8.599851972017761e-06,
      "loss": 0.1862,
      "step": 2919
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3640767770846927,
      "learning_rate": 8.598824155649828e-06,
      "loss": 0.2184,
      "step": 2920
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.559532031353858,
      "learning_rate": 8.59779602364094e-06,
      "loss": 0.2637,
      "step": 2921
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3860838107267808,
      "learning_rate": 8.596767576081275e-06,
      "loss": 0.2132,
      "step": 2922
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4740937277064725,
      "learning_rate": 8.595738813061031e-06,
      "loss": 0.2223,
      "step": 2923
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3625075769379356,
      "learning_rate": 8.59470973467044e-06,
      "loss": 0.2426,
      "step": 2924
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5176856275919002,
      "learning_rate": 8.593680340999755e-06,
      "loss": 0.2288,
      "step": 2925
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4945829462422509,
      "learning_rate": 8.592650632139264e-06,
      "loss": 0.2469,
      "step": 2926
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.461511293714239,
      "learning_rate": 8.591620608179276e-06,
      "loss": 0.2863,
      "step": 2927
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.603212127377697,
      "learning_rate": 8.590590269210137e-06,
      "loss": 0.1987,
      "step": 2928
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.461691249185531,
      "learning_rate": 8.58955961532221e-06,
      "loss": 0.2522,
      "step": 2929
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.373096609494322,
      "learning_rate": 8.588528646605893e-06,
      "loss": 0.2748,
      "step": 2930
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4039496441211932,
      "learning_rate": 8.587497363151604e-06,
      "loss": 0.2007,
      "step": 2931
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.473983460243633,
      "learning_rate": 8.586465765049801e-06,
      "loss": 0.6727,
      "step": 2932
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2824547365901973,
      "learning_rate": 8.585433852390955e-06,
      "loss": 0.1801,
      "step": 2933
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4917134340226035,
      "learning_rate": 8.584401625265574e-06,
      "loss": 0.241,
      "step": 2934
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.399521418081149,
      "learning_rate": 8.583369083764194e-06,
      "loss": 0.2082,
      "step": 2935
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.315320263358807,
      "learning_rate": 8.582336227977372e-06,
      "loss": 0.618,
      "step": 2936
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3086166035700784,
      "learning_rate": 8.581303057995697e-06,
      "loss": 0.1623,
      "step": 2937
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.10631753675346,
      "learning_rate": 8.580269573909789e-06,
      "loss": 0.6223,
      "step": 2938
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4624148298825073,
      "learning_rate": 8.579235775810287e-06,
      "loss": 0.2232,
      "step": 2939
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3164409808716224,
      "learning_rate": 8.57820166378786e-06,
      "loss": 0.1947,
      "step": 2940
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4519940720593623,
      "learning_rate": 8.577167237933213e-06,
      "loss": 0.2625,
      "step": 2941
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3326773695285954,
      "learning_rate": 8.576132498337069e-06,
      "loss": 0.2268,
      "step": 2942
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5603497747907284,
      "learning_rate": 8.57509744509018e-06,
      "loss": 0.2225,
      "step": 2943
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4855256407929758,
      "learning_rate": 8.574062078283328e-06,
      "loss": 0.2267,
      "step": 2944
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.372077404835741,
      "learning_rate": 8.573026398007323e-06,
      "loss": 0.2331,
      "step": 2945
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.99752112198902,
      "learning_rate": 8.571990404353002e-06,
      "loss": 0.63,
      "step": 2946
    },
    {
      "epoch": 0.27,
      "grad_norm": 6.061502945599847,
      "learning_rate": 8.570954097411224e-06,
      "loss": 0.5903,
      "step": 2947
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.258664375907557,
      "learning_rate": 8.569917477272885e-06,
      "loss": 0.645,
      "step": 2948
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4644236001438897,
      "learning_rate": 8.5688805440289e-06,
      "loss": 0.2374,
      "step": 2949
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.404367456948624,
      "learning_rate": 8.567843297770215e-06,
      "loss": 0.2389,
      "step": 2950
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.440659744330427,
      "learning_rate": 8.566805738587806e-06,
      "loss": 0.1863,
      "step": 2951
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3620334423441391,
      "learning_rate": 8.565767866572674e-06,
      "loss": 0.2199,
      "step": 2952
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.406635630101618,
      "learning_rate": 8.564729681815846e-06,
      "loss": 0.2257,
      "step": 2953
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2381890124232307,
      "learning_rate": 8.563691184408376e-06,
      "loss": 0.19,
      "step": 2954
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.410449437638908,
      "learning_rate": 8.56265237444135e-06,
      "loss": 0.2263,
      "step": 2955
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.188336669346204,
      "learning_rate": 8.561613252005878e-06,
      "loss": 0.1553,
      "step": 2956
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2789371218075705,
      "learning_rate": 8.560573817193096e-06,
      "loss": 0.2378,
      "step": 2957
    },
    {
      "epoch": 0.27,
      "grad_norm": 8.333154423187542,
      "learning_rate": 8.559534070094172e-06,
      "loss": 0.6942,
      "step": 2958
    },
    {
      "epoch": 0.27,
      "grad_norm": 9.955741183614002,
      "learning_rate": 8.558494010800297e-06,
      "loss": 0.6673,
      "step": 2959
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2849002041349207,
      "learning_rate": 8.557453639402692e-06,
      "loss": 0.2199,
      "step": 2960
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4149118514957524,
      "learning_rate": 8.556412955992604e-06,
      "loss": 0.2033,
      "step": 2961
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3182316455139278,
      "learning_rate": 8.55537196066131e-06,
      "loss": 0.2211,
      "step": 2962
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.469965624550941,
      "learning_rate": 8.554330653500107e-06,
      "loss": 0.2437,
      "step": 2963
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3193908549908249,
      "learning_rate": 8.55328903460033e-06,
      "loss": 0.2314,
      "step": 2964
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3689074609004974,
      "learning_rate": 8.552247104053335e-06,
      "loss": 0.1944,
      "step": 2965
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4951693080277773,
      "learning_rate": 8.551204861950503e-06,
      "loss": 0.2288,
      "step": 2966
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.478773789156093,
      "learning_rate": 8.55016230838325e-06,
      "loss": 0.2322,
      "step": 2967
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2774302213389745,
      "learning_rate": 8.549119443443013e-06,
      "loss": 0.2115,
      "step": 2968
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.340236391054055,
      "learning_rate": 8.548076267221258e-06,
      "loss": 0.1718,
      "step": 2969
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.362043957471619,
      "learning_rate": 8.547032779809477e-06,
      "loss": 0.2356,
      "step": 2970
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.378099738454324,
      "learning_rate": 8.545988981299192e-06,
      "loss": 0.7548,
      "step": 2971
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5245827589532897,
      "learning_rate": 8.544944871781953e-06,
      "loss": 0.2226,
      "step": 2972
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3788337493079295,
      "learning_rate": 8.543900451349334e-06,
      "loss": 0.2203,
      "step": 2973
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4695330911726474,
      "learning_rate": 8.542855720092937e-06,
      "loss": 0.198,
      "step": 2974
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.425543651085425,
      "learning_rate": 8.541810678104394e-06,
      "loss": 0.5549,
      "step": 2975
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.389355795081214,
      "learning_rate": 8.540765325475358e-06,
      "loss": 0.2319,
      "step": 2976
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5144248622679308,
      "learning_rate": 8.539719662297519e-06,
      "loss": 0.1971,
      "step": 2977
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3092391418474099,
      "learning_rate": 8.538673688662583e-06,
      "loss": 0.1908,
      "step": 2978
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3328842526910494,
      "learning_rate": 8.537627404662293e-06,
      "loss": 0.1783,
      "step": 2979
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.550470017985465,
      "learning_rate": 8.536580810388414e-06,
      "loss": 0.2411,
      "step": 2980
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4188356252737506,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.2365,
      "step": 2981
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.558566013477577,
      "learning_rate": 8.534486691387087e-06,
      "loss": 0.2175,
      "step": 2982
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4048992301708763,
      "learning_rate": 8.533439166843309e-06,
      "loss": 0.2159,
      "step": 2983
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4178719179838797,
      "learning_rate": 8.532391332393278e-06,
      "loss": 0.2151,
      "step": 2984
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3443460264104958,
      "learning_rate": 8.531343188128896e-06,
      "loss": 0.2449,
      "step": 2985
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7358477450390113,
      "learning_rate": 8.530294734142092e-06,
      "loss": 0.2188,
      "step": 2986
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2990095306534934,
      "learning_rate": 8.529245970524824e-06,
      "loss": 0.2107,
      "step": 2987
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1584775732714525,
      "learning_rate": 8.528196897369074e-06,
      "loss": 0.189,
      "step": 2988
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.519414666782397,
      "learning_rate": 8.527147514766854e-06,
      "loss": 0.2406,
      "step": 2989
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3526558372607929,
      "learning_rate": 8.526097822810202e-06,
      "loss": 0.2085,
      "step": 2990
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2087482327820085,
      "learning_rate": 8.52504782159118e-06,
      "loss": 0.1692,
      "step": 2991
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3935331518520169,
      "learning_rate": 8.523997511201884e-06,
      "loss": 0.2464,
      "step": 2992
    },
    {
      "epoch": 0.27,
      "grad_norm": 7.842793228834663,
      "learning_rate": 8.52294689173443e-06,
      "loss": 0.8182,
      "step": 2993
    },
    {
      "epoch": 0.27,
      "grad_norm": 5.947887845278049,
      "learning_rate": 8.521895963280967e-06,
      "loss": 0.6606,
      "step": 2994
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4031325604999436,
      "learning_rate": 8.520844725933668e-06,
      "loss": 0.2392,
      "step": 2995
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3017837933722771,
      "learning_rate": 8.519793179784732e-06,
      "loss": 0.1876,
      "step": 2996
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.422138253317163,
      "learning_rate": 8.518741324926388e-06,
      "loss": 0.2464,
      "step": 2997
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4524898767982426,
      "learning_rate": 8.51768916145089e-06,
      "loss": 0.1857,
      "step": 2998
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5702805909459245,
      "learning_rate": 8.516636689450519e-06,
      "loss": 0.3148,
      "step": 2999
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3979626508287815,
      "learning_rate": 8.515583909017585e-06,
      "loss": 0.2001,
      "step": 3000
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3710111689207112,
      "learning_rate": 8.514530820244427e-06,
      "loss": 0.2164,
      "step": 3001
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2405218438458936,
      "learning_rate": 8.513477423223401e-06,
      "loss": 0.2225,
      "step": 3002
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2877924476146168,
      "learning_rate": 8.512423718046903e-06,
      "loss": 0.205,
      "step": 3003
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3734945847240205,
      "learning_rate": 8.511369704807347e-06,
      "loss": 0.161,
      "step": 3004
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.526743341091488,
      "learning_rate": 8.510315383597178e-06,
      "loss": 0.231,
      "step": 3005
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.215710148171461,
      "learning_rate": 8.509260754508866e-06,
      "loss": 0.1408,
      "step": 3006
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1887698298973903,
      "learning_rate": 8.508205817634908e-06,
      "loss": 0.1773,
      "step": 3007
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1745755855815907,
      "learning_rate": 8.507150573067834e-06,
      "loss": 0.1977,
      "step": 3008
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.628676375530482,
      "learning_rate": 8.506095020900192e-06,
      "loss": 0.2349,
      "step": 3009
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.502199990614777,
      "learning_rate": 8.50503916122456e-06,
      "loss": 0.2284,
      "step": 3010
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4019533937401383,
      "learning_rate": 8.503982994133547e-06,
      "loss": 0.2341,
      "step": 3011
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4375649158214765,
      "learning_rate": 8.502926519719785e-06,
      "loss": 0.2355,
      "step": 3012
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.133139039267726,
      "learning_rate": 8.501869738075932e-06,
      "loss": 0.5452,
      "step": 3013
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4377597525539938,
      "learning_rate": 8.500812649294679e-06,
      "loss": 0.2438,
      "step": 3014
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.147749520002581,
      "learning_rate": 8.499755253468732e-06,
      "loss": 0.6623,
      "step": 3015
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3693121331413884,
      "learning_rate": 8.498697550690841e-06,
      "loss": 0.1991,
      "step": 3016
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2684794199535256,
      "learning_rate": 8.497639541053769e-06,
      "loss": 0.1784,
      "step": 3017
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4008883715329576,
      "learning_rate": 8.496581224650311e-06,
      "loss": 0.2182,
      "step": 3018
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4162578048991328,
      "learning_rate": 8.495522601573288e-06,
      "loss": 0.2095,
      "step": 3019
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7471140816597674,
      "learning_rate": 8.494463671915547e-06,
      "loss": 0.2848,
      "step": 3020
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.016760813469047,
      "learning_rate": 8.493404435769967e-06,
      "loss": 0.7982,
      "step": 3021
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6599659315432291,
      "learning_rate": 8.492344893229447e-06,
      "loss": 0.2728,
      "step": 3022
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.615604482059825,
      "learning_rate": 8.491285044386917e-06,
      "loss": 0.5486,
      "step": 3023
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3989170479998738,
      "learning_rate": 8.490224889335332e-06,
      "loss": 0.2029,
      "step": 3024
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.409021321818854,
      "learning_rate": 8.489164428167677e-06,
      "loss": 0.2155,
      "step": 3025
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4304840426638454,
      "learning_rate": 8.488103660976959e-06,
      "loss": 0.1997,
      "step": 3026
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4668609160504915,
      "learning_rate": 8.487042587856215e-06,
      "loss": 0.1914,
      "step": 3027
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3084805731274092,
      "learning_rate": 8.48598120889851e-06,
      "loss": 0.2342,
      "step": 3028
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3001114365589774,
      "learning_rate": 8.484919524196932e-06,
      "loss": 0.2064,
      "step": 3029
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3664265853717075,
      "learning_rate": 8.483857533844598e-06,
      "loss": 0.2035,
      "step": 3030
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.666058197352472,
      "learning_rate": 8.482795237934652e-06,
      "loss": 0.2858,
      "step": 3031
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3545253621538518,
      "learning_rate": 8.481732636560266e-06,
      "loss": 0.1838,
      "step": 3032
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3893393794541704,
      "learning_rate": 8.480669729814635e-06,
      "loss": 0.2272,
      "step": 3033
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5415812878644304,
      "learning_rate": 8.479606517790982e-06,
      "loss": 0.2615,
      "step": 3034
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3380580312904937,
      "learning_rate": 8.478543000582562e-06,
      "loss": 0.217,
      "step": 3035
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.903206599425237,
      "learning_rate": 8.47747917828265e-06,
      "loss": 0.7132,
      "step": 3036
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6756846749252865,
      "learning_rate": 8.47641505098455e-06,
      "loss": 0.1949,
      "step": 3037
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.504881579953699,
      "learning_rate": 8.475350618781593e-06,
      "loss": 0.7143,
      "step": 3038
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2681008220875574,
      "learning_rate": 8.474285881767138e-06,
      "loss": 0.2014,
      "step": 3039
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4089171647761898,
      "learning_rate": 8.47322084003457e-06,
      "loss": 0.242,
      "step": 3040
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3462532921147063,
      "learning_rate": 8.472155493677299e-06,
      "loss": 0.2188,
      "step": 3041
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.257034482834706,
      "learning_rate": 8.471089842788762e-06,
      "loss": 0.7795,
      "step": 3042
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4138281634650984,
      "learning_rate": 8.470023887462426e-06,
      "loss": 0.1953,
      "step": 3043
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3753552736477763,
      "learning_rate": 8.468957627791781e-06,
      "loss": 0.216,
      "step": 3044
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5023213000004534,
      "learning_rate": 8.467891063870346e-06,
      "loss": 0.2462,
      "step": 3045
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.183059936377528,
      "learning_rate": 8.466824195791665e-06,
      "loss": 0.5644,
      "step": 3046
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.34317544069724,
      "learning_rate": 8.465757023649311e-06,
      "loss": 0.2252,
      "step": 3047
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2659715866873145,
      "learning_rate": 8.46468954753688e-06,
      "loss": 0.1733,
      "step": 3048
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6155316686340022,
      "learning_rate": 8.463621767547998e-06,
      "loss": 0.2222,
      "step": 3049
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.770613827381001,
      "learning_rate": 8.462553683776316e-06,
      "loss": 0.2507,
      "step": 3050
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3068721744405078,
      "learning_rate": 8.461485296315512e-06,
      "loss": 0.2135,
      "step": 3051
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3010641059583974,
      "learning_rate": 8.460416605259292e-06,
      "loss": 0.1662,
      "step": 3052
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4786039040465506,
      "learning_rate": 8.459347610701386e-06,
      "loss": 0.22,
      "step": 3053
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3433666034110257,
      "learning_rate": 8.458278312735552e-06,
      "loss": 0.1971,
      "step": 3054
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2889287935247695,
      "learning_rate": 8.457208711455574e-06,
      "loss": 0.1932,
      "step": 3055
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3445976437319083,
      "learning_rate": 8.456138806955266e-06,
      "loss": 0.2395,
      "step": 3056
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4303379791103776,
      "learning_rate": 8.455068599328462e-06,
      "loss": 0.2465,
      "step": 3057
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4420692774540091,
      "learning_rate": 8.45399808866903e-06,
      "loss": 0.2153,
      "step": 3058
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4623864527580401,
      "learning_rate": 8.452927275070858e-06,
      "loss": 0.1891,
      "step": 3059
    },
    {
      "epoch": 0.28,
      "grad_norm": 29.820475918419877,
      "learning_rate": 8.451856158627864e-06,
      "loss": 0.6767,
      "step": 3060
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.031042473667646,
      "learning_rate": 8.450784739433996e-06,
      "loss": 0.5995,
      "step": 3061
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.452554401085638,
      "learning_rate": 8.449713017583218e-06,
      "loss": 0.7901,
      "step": 3062
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3263306721358163,
      "learning_rate": 8.448640993169533e-06,
      "loss": 0.1969,
      "step": 3063
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.338211046788763,
      "learning_rate": 8.447568666286961e-06,
      "loss": 0.1996,
      "step": 3064
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3145759182648227,
      "learning_rate": 8.446496037029555e-06,
      "loss": 0.2113,
      "step": 3065
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.421209626013811,
      "learning_rate": 8.445423105491387e-06,
      "loss": 0.2078,
      "step": 3066
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.431615710055734,
      "learning_rate": 8.444349871766566e-06,
      "loss": 0.2135,
      "step": 3067
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3903674757658668,
      "learning_rate": 8.44327633594922e-06,
      "loss": 0.2124,
      "step": 3068
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5647797951726121,
      "learning_rate": 8.442202498133506e-06,
      "loss": 0.2422,
      "step": 3069
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2158503381792956,
      "learning_rate": 8.441128358413602e-06,
      "loss": 0.1468,
      "step": 3070
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4671989554096443,
      "learning_rate": 8.440053916883721e-06,
      "loss": 0.2602,
      "step": 3071
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4264201080183205,
      "learning_rate": 8.4389791736381e-06,
      "loss": 0.2182,
      "step": 3072
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2240994677062427,
      "learning_rate": 8.437904128770999e-06,
      "loss": 0.1854,
      "step": 3073
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3422793355130131,
      "learning_rate": 8.436828782376707e-06,
      "loss": 0.2101,
      "step": 3074
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3159513990334315,
      "learning_rate": 8.43575313454954e-06,
      "loss": 0.1819,
      "step": 3075
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5294704467237932,
      "learning_rate": 8.434677185383838e-06,
      "loss": 0.2345,
      "step": 3076
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5081078123517397,
      "learning_rate": 8.43360093497397e-06,
      "loss": 0.2329,
      "step": 3077
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4363913780283313,
      "learning_rate": 8.432524383414329e-06,
      "loss": 0.2167,
      "step": 3078
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.162480891643165,
      "learning_rate": 8.431447530799338e-06,
      "loss": 0.1828,
      "step": 3079
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3523944817136377,
      "learning_rate": 8.430370377223442e-06,
      "loss": 0.1816,
      "step": 3080
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5157707610259938,
      "learning_rate": 8.429292922781115e-06,
      "loss": 0.2306,
      "step": 3081
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3167758435006256,
      "learning_rate": 8.42821516756686e-06,
      "loss": 0.2163,
      "step": 3082
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3922512638980014,
      "learning_rate": 8.4271371116752e-06,
      "loss": 0.2365,
      "step": 3083
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4553798728983744,
      "learning_rate": 8.426058755200688e-06,
      "loss": 0.2152,
      "step": 3084
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6034055383195183,
      "learning_rate": 8.424980098237904e-06,
      "loss": 0.2197,
      "step": 3085
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3163205534446545,
      "learning_rate": 8.423901140881452e-06,
      "loss": 0.242,
      "step": 3086
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.424930650509104,
      "learning_rate": 8.422821883225966e-06,
      "loss": 0.2167,
      "step": 3087
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5064160379944709,
      "learning_rate": 8.421742325366103e-06,
      "loss": 0.1961,
      "step": 3088
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3933854617898855,
      "learning_rate": 8.420662467396548e-06,
      "loss": 0.2442,
      "step": 3089
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.940014363866124,
      "learning_rate": 8.41958230941201e-06,
      "loss": 0.5862,
      "step": 3090
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.210066517604482,
      "learning_rate": 8.418501851507228e-06,
      "loss": 0.1841,
      "step": 3091
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4437973121608354,
      "learning_rate": 8.417421093776963e-06,
      "loss": 0.1759,
      "step": 3092
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3018461469396716,
      "learning_rate": 8.416340036316008e-06,
      "loss": 0.2082,
      "step": 3093
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3894556257162263,
      "learning_rate": 8.415258679219179e-06,
      "loss": 0.2414,
      "step": 3094
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3277673412784081,
      "learning_rate": 8.414177022581314e-06,
      "loss": 0.1941,
      "step": 3095
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2282232236368016,
      "learning_rate": 8.413095066497283e-06,
      "loss": 0.194,
      "step": 3096
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.573001939098727,
      "learning_rate": 8.412012811061985e-06,
      "loss": 0.2608,
      "step": 3097
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4409698992859783,
      "learning_rate": 8.410930256370337e-06,
      "loss": 0.2262,
      "step": 3098
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3146806596087388,
      "learning_rate": 8.409847402517286e-06,
      "loss": 0.1754,
      "step": 3099
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4731188859414748,
      "learning_rate": 8.408764249597807e-06,
      "loss": 0.2379,
      "step": 3100
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3219050259152456,
      "learning_rate": 8.407680797706901e-06,
      "loss": 0.2341,
      "step": 3101
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2878470737226626,
      "learning_rate": 8.406597046939591e-06,
      "loss": 0.1742,
      "step": 3102
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.037637291595524,
      "learning_rate": 8.405512997390932e-06,
      "loss": 0.6454,
      "step": 3103
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4418898322289104,
      "learning_rate": 8.404428649156e-06,
      "loss": 0.2569,
      "step": 3104
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5634207607079926,
      "learning_rate": 8.403344002329901e-06,
      "loss": 0.2349,
      "step": 3105
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.369045480267953,
      "learning_rate": 8.402259057007766e-06,
      "loss": 0.2148,
      "step": 3106
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4049411495821904,
      "learning_rate": 8.40117381328475e-06,
      "loss": 0.2594,
      "step": 3107
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5694257920392054,
      "learning_rate": 8.400088271256038e-06,
      "loss": 0.2275,
      "step": 3108
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2969841450786963,
      "learning_rate": 8.39900243101684e-06,
      "loss": 0.2335,
      "step": 3109
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.427733855450143,
      "learning_rate": 8.397916292662389e-06,
      "loss": 0.2423,
      "step": 3110
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3301443455480895,
      "learning_rate": 8.39682985628795e-06,
      "loss": 0.2059,
      "step": 3111
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4051118250288774,
      "learning_rate": 8.395743121988807e-06,
      "loss": 0.2539,
      "step": 3112
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3488280179008287,
      "learning_rate": 8.394656089860274e-06,
      "loss": 0.2423,
      "step": 3113
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.475129047250131,
      "learning_rate": 8.393568759997694e-06,
      "loss": 0.2374,
      "step": 3114
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.176090084222605,
      "learning_rate": 8.392481132496434e-06,
      "loss": 0.6436,
      "step": 3115
    },
    {
      "epoch": 0.28,
      "grad_norm": 7.0493267614758155,
      "learning_rate": 8.391393207451879e-06,
      "loss": 0.6142,
      "step": 3116
    },
    {
      "epoch": 0.28,
      "grad_norm": 6.776388773601277,
      "learning_rate": 8.390304984959455e-06,
      "loss": 0.4807,
      "step": 3117
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4725567133055826,
      "learning_rate": 8.389216465114602e-06,
      "loss": 0.2252,
      "step": 3118
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3047435421929947,
      "learning_rate": 8.388127648012794e-06,
      "loss": 0.1787,
      "step": 3119
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3183652481366077,
      "learning_rate": 8.387038533749524e-06,
      "loss": 0.2168,
      "step": 3120
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4134779613646948,
      "learning_rate": 8.385949122420318e-06,
      "loss": 0.1692,
      "step": 3121
    },
    {
      "epoch": 0.29,
      "grad_norm": 20.253463271298948,
      "learning_rate": 8.384859414120721e-06,
      "loss": 0.6323,
      "step": 3122
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7043606200107981,
      "learning_rate": 8.383769408946309e-06,
      "loss": 0.2285,
      "step": 3123
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3296328883378632,
      "learning_rate": 8.382679106992687e-06,
      "loss": 0.1929,
      "step": 3124
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.417149789411225,
      "learning_rate": 8.381588508355473e-06,
      "loss": 0.2053,
      "step": 3125
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4645788490885607,
      "learning_rate": 8.380497613130327e-06,
      "loss": 0.2421,
      "step": 3126
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4783886538087398,
      "learning_rate": 8.379406421412926e-06,
      "loss": 0.2555,
      "step": 3127
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3080558875625243,
      "learning_rate": 8.378314933298974e-06,
      "loss": 0.2026,
      "step": 3128
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6072997648721172,
      "learning_rate": 8.377223148884202e-06,
      "loss": 0.268,
      "step": 3129
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3941035114885638,
      "learning_rate": 8.376131068264368e-06,
      "loss": 0.208,
      "step": 3130
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4244440277229393,
      "learning_rate": 8.375038691535253e-06,
      "loss": 0.2165,
      "step": 3131
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4404526495179306,
      "learning_rate": 8.373946018792666e-06,
      "loss": 0.2002,
      "step": 3132
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3040067115812959,
      "learning_rate": 8.372853050132443e-06,
      "loss": 0.2026,
      "step": 3133
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.506529147130273,
      "learning_rate": 8.371759785650444e-06,
      "loss": 0.2254,
      "step": 3134
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5269746935423574,
      "learning_rate": 8.370666225442557e-06,
      "loss": 0.2474,
      "step": 3135
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4471604223739667,
      "learning_rate": 8.36957236960469e-06,
      "loss": 0.2247,
      "step": 3136
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4681278014193313,
      "learning_rate": 8.368478218232787e-06,
      "loss": 0.2121,
      "step": 3137
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.288330316362755,
      "learning_rate": 8.36738377142281e-06,
      "loss": 0.2011,
      "step": 3138
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4208946139374619,
      "learning_rate": 8.366289029270746e-06,
      "loss": 0.2245,
      "step": 3139
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.568869503850206,
      "learning_rate": 8.365193991872618e-06,
      "loss": 0.2374,
      "step": 3140
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.59031530180309,
      "learning_rate": 8.364098659324464e-06,
      "loss": 0.6473,
      "step": 3141
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.655617970816917,
      "learning_rate": 8.363003031722352e-06,
      "loss": 0.2162,
      "step": 3142
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3365875577069768,
      "learning_rate": 8.361907109162377e-06,
      "loss": 0.2477,
      "step": 3143
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2470074666164985,
      "learning_rate": 8.360810891740656e-06,
      "loss": 0.1915,
      "step": 3144
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4463608605497407,
      "learning_rate": 8.359714379553338e-06,
      "loss": 0.2509,
      "step": 3145
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3174868344057027,
      "learning_rate": 8.358617572696595e-06,
      "loss": 0.2206,
      "step": 3146
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2102205768797185,
      "learning_rate": 8.35752047126662e-06,
      "loss": 0.1879,
      "step": 3147
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3533050550281687,
      "learning_rate": 8.35642307535964e-06,
      "loss": 0.1877,
      "step": 3148
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.509613575283223,
      "learning_rate": 8.355325385071904e-06,
      "loss": 0.1817,
      "step": 3149
    },
    {
      "epoch": 0.29,
      "grad_norm": 7.374600154006957,
      "learning_rate": 8.354227400499684e-06,
      "loss": 0.7547,
      "step": 3150
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4523527730668069,
      "learning_rate": 8.35312912173928e-06,
      "loss": 0.2296,
      "step": 3151
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5037886841453196,
      "learning_rate": 8.352030548887024e-06,
      "loss": 0.2026,
      "step": 3152
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.432785879449174,
      "learning_rate": 8.350931682039262e-06,
      "loss": 0.1942,
      "step": 3153
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.485191954747989,
      "learning_rate": 8.349832521292375e-06,
      "loss": 0.1762,
      "step": 3154
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4530478486863228,
      "learning_rate": 8.348733066742766e-06,
      "loss": 0.2059,
      "step": 3155
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4124798337905005,
      "learning_rate": 8.347633318486868e-06,
      "loss": 0.2863,
      "step": 3156
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.249297789388605,
      "learning_rate": 8.346533276621129e-06,
      "loss": 0.1809,
      "step": 3157
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.595629070610278,
      "learning_rate": 8.345432941242038e-06,
      "loss": 0.2449,
      "step": 3158
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3925151361351311,
      "learning_rate": 8.344332312446095e-06,
      "loss": 0.2016,
      "step": 3159
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.84915477052672,
      "learning_rate": 8.343231390329836e-06,
      "loss": 0.7656,
      "step": 3160
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.311236264774522,
      "learning_rate": 8.342130174989819e-06,
      "loss": 0.6432,
      "step": 3161
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3539719447891634,
      "learning_rate": 8.341028666522629e-06,
      "loss": 0.2039,
      "step": 3162
    },
    {
      "epoch": 0.29,
      "grad_norm": 27.16647759988819,
      "learning_rate": 8.339926865024871e-06,
      "loss": 0.4502,
      "step": 3163
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4371677038557955,
      "learning_rate": 8.338824770593185e-06,
      "loss": 0.2018,
      "step": 3164
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4491997195368167,
      "learning_rate": 8.337722383324232e-06,
      "loss": 0.1597,
      "step": 3165
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.425224643830377,
      "learning_rate": 8.336619703314696e-06,
      "loss": 0.2164,
      "step": 3166
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5428888423661975,
      "learning_rate": 8.335516730661292e-06,
      "loss": 0.2613,
      "step": 3167
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4217151787333597,
      "learning_rate": 8.334413465460756e-06,
      "loss": 0.2229,
      "step": 3168
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4000058307441627,
      "learning_rate": 8.333309907809852e-06,
      "loss": 0.2002,
      "step": 3169
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2008720281965026,
      "learning_rate": 8.332206057805372e-06,
      "loss": 0.1665,
      "step": 3170
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3444621803715904,
      "learning_rate": 8.331101915544128e-06,
      "loss": 0.2071,
      "step": 3171
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2552992685631428,
      "learning_rate": 8.329997481122962e-06,
      "loss": 0.1805,
      "step": 3172
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3485685123118645,
      "learning_rate": 8.32889275463874e-06,
      "loss": 0.2755,
      "step": 3173
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2147507508599233,
      "learning_rate": 8.327787736188354e-06,
      "loss": 0.1713,
      "step": 3174
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.402259218198423,
      "learning_rate": 8.32668242586872e-06,
      "loss": 0.4802,
      "step": 3175
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.720021391735566,
      "learning_rate": 8.325576823776785e-06,
      "loss": 0.297,
      "step": 3176
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5478305507846644,
      "learning_rate": 8.324470930009514e-06,
      "loss": 0.2134,
      "step": 3177
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4582639042812482,
      "learning_rate": 8.323364744663902e-06,
      "loss": 0.2291,
      "step": 3178
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.593088885664943,
      "learning_rate": 8.32225826783697e-06,
      "loss": 0.2232,
      "step": 3179
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.444514772502851,
      "learning_rate": 8.321151499625764e-06,
      "loss": 0.2035,
      "step": 3180
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2965419891322887,
      "learning_rate": 8.32004444012735e-06,
      "loss": 0.2011,
      "step": 3181
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6340778420786952,
      "learning_rate": 8.318937089438833e-06,
      "loss": 0.2199,
      "step": 3182
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5004390160444552,
      "learning_rate": 8.317829447657327e-06,
      "loss": 0.2378,
      "step": 3183
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4687443673718281,
      "learning_rate": 8.316721514879984e-06,
      "loss": 0.2176,
      "step": 3184
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4030841417785993,
      "learning_rate": 8.315613291203977e-06,
      "loss": 0.2339,
      "step": 3185
    },
    {
      "epoch": 0.29,
      "grad_norm": 16.09365243601699,
      "learning_rate": 8.314504776726503e-06,
      "loss": 0.6632,
      "step": 3186
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3320871529739713,
      "learning_rate": 8.313395971544787e-06,
      "loss": 0.2254,
      "step": 3187
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2706625715295878,
      "learning_rate": 8.312286875756078e-06,
      "loss": 0.1929,
      "step": 3188
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4778816155532801,
      "learning_rate": 8.311177489457653e-06,
      "loss": 0.2407,
      "step": 3189
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5709818206429025,
      "learning_rate": 8.310067812746809e-06,
      "loss": 0.2879,
      "step": 3190
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3940277721089116,
      "learning_rate": 8.308957845720875e-06,
      "loss": 0.2369,
      "step": 3191
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3321679021747144,
      "learning_rate": 8.307847588477203e-06,
      "loss": 0.2219,
      "step": 3192
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4421550587664635,
      "learning_rate": 8.306737041113169e-06,
      "loss": 0.2305,
      "step": 3193
    },
    {
      "epoch": 0.29,
      "grad_norm": 37.39103045346438,
      "learning_rate": 8.305626203726174e-06,
      "loss": 0.5085,
      "step": 3194
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3054714812535413,
      "learning_rate": 8.304515076413647e-06,
      "loss": 0.1845,
      "step": 3195
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4979409578455283,
      "learning_rate": 8.303403659273042e-06,
      "loss": 0.2281,
      "step": 3196
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4599372089660598,
      "learning_rate": 8.302291952401837e-06,
      "loss": 0.2435,
      "step": 3197
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.366694083033715,
      "learning_rate": 8.301179955897535e-06,
      "loss": 0.5685,
      "step": 3198
    },
    {
      "epoch": 0.29,
      "grad_norm": 9.68674381438151,
      "learning_rate": 8.300067669857668e-06,
      "loss": 0.6016,
      "step": 3199
    },
    {
      "epoch": 0.29,
      "grad_norm": 18.916279030450983,
      "learning_rate": 8.298955094379789e-06,
      "loss": 0.5989,
      "step": 3200
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.375534129809793,
      "learning_rate": 8.29784222956148e-06,
      "loss": 0.2264,
      "step": 3201
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4725728663289301,
      "learning_rate": 8.296729075500345e-06,
      "loss": 0.2389,
      "step": 3202
    },
    {
      "epoch": 0.29,
      "grad_norm": 4.49224237625698,
      "learning_rate": 8.295615632294017e-06,
      "loss": 0.5929,
      "step": 3203
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.370168913775399,
      "learning_rate": 8.29450190004015e-06,
      "loss": 0.5278,
      "step": 3204
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4036638564074693,
      "learning_rate": 8.293387878836428e-06,
      "loss": 0.2381,
      "step": 3205
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.344032585896445,
      "learning_rate": 8.292273568780555e-06,
      "loss": 0.2261,
      "step": 3206
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4673372143318222,
      "learning_rate": 8.291158969970267e-06,
      "loss": 0.2001,
      "step": 3207
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6148501366527013,
      "learning_rate": 8.290044082503319e-06,
      "loss": 0.2364,
      "step": 3208
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1941787473683336,
      "learning_rate": 8.288928906477497e-06,
      "loss": 0.1881,
      "step": 3209
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4376366335239519,
      "learning_rate": 8.287813441990607e-06,
      "loss": 0.2345,
      "step": 3210
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2085655871051504,
      "learning_rate": 8.286697689140483e-06,
      "loss": 0.1792,
      "step": 3211
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.32555571210724,
      "learning_rate": 8.285581648024984e-06,
      "loss": 0.2256,
      "step": 3212
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1613771469875096,
      "learning_rate": 8.284465318741997e-06,
      "loss": 0.184,
      "step": 3213
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2829411557033,
      "learning_rate": 8.283348701389429e-06,
      "loss": 0.2362,
      "step": 3214
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4100432263096276,
      "learning_rate": 8.282231796065215e-06,
      "loss": 0.2188,
      "step": 3215
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3677155995178247,
      "learning_rate": 8.281114602867312e-06,
      "loss": 0.2335,
      "step": 3216
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3733713499108737,
      "learning_rate": 8.279997121893713e-06,
      "loss": 0.2262,
      "step": 3217
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.456112510955722,
      "learning_rate": 8.278879353242421e-06,
      "loss": 0.2306,
      "step": 3218
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4041778643154916,
      "learning_rate": 8.277761297011475e-06,
      "loss": 0.1929,
      "step": 3219
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4823136714901013,
      "learning_rate": 8.276642953298937e-06,
      "loss": 0.2075,
      "step": 3220
    },
    {
      "epoch": 0.29,
      "grad_norm": 6.552608833752154,
      "learning_rate": 8.27552432220289e-06,
      "loss": 0.6423,
      "step": 3221
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4151251046575393,
      "learning_rate": 8.27440540382145e-06,
      "loss": 0.194,
      "step": 3222
    },
    {
      "epoch": 0.29,
      "grad_norm": 11.377274280426924,
      "learning_rate": 8.27328619825275e-06,
      "loss": 0.7022,
      "step": 3223
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.477951971344679,
      "learning_rate": 8.272166705594953e-06,
      "loss": 0.2132,
      "step": 3224
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3106287408365431,
      "learning_rate": 8.271046925946247e-06,
      "loss": 0.1753,
      "step": 3225
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4140865878332567,
      "learning_rate": 8.26992685940484e-06,
      "loss": 0.2558,
      "step": 3226
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.451179425755891,
      "learning_rate": 8.268806506068976e-06,
      "loss": 0.2503,
      "step": 3227
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.9286856928784895,
      "learning_rate": 8.267685866036912e-06,
      "loss": 0.5827,
      "step": 3228
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5619197696781615,
      "learning_rate": 8.266564939406938e-06,
      "loss": 0.2609,
      "step": 3229
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5262075552987338,
      "learning_rate": 8.265443726277365e-06,
      "loss": 0.2551,
      "step": 3230
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5922475034302694,
      "learning_rate": 8.264322226746532e-06,
      "loss": 0.2387,
      "step": 3231
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7736577263498592,
      "learning_rate": 8.263200440912805e-06,
      "loss": 0.2535,
      "step": 3232
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.323985659083663,
      "learning_rate": 8.262078368874566e-06,
      "loss": 0.1797,
      "step": 3233
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4095353327213314,
      "learning_rate": 8.260956010730234e-06,
      "loss": 0.2566,
      "step": 3234
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6564836806452454,
      "learning_rate": 8.259833366578245e-06,
      "loss": 0.269,
      "step": 3235
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3225239826665893,
      "learning_rate": 8.25871043651706e-06,
      "loss": 0.2026,
      "step": 3236
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.334559592039823,
      "learning_rate": 8.257587220645172e-06,
      "loss": 0.2181,
      "step": 3237
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4046235597531693,
      "learning_rate": 8.256463719061092e-06,
      "loss": 0.2378,
      "step": 3238
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3400060894549493,
      "learning_rate": 8.255339931863358e-06,
      "loss": 0.2157,
      "step": 3239
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4718944707714765,
      "learning_rate": 8.254215859150535e-06,
      "loss": 0.2078,
      "step": 3240
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4105170667041922,
      "learning_rate": 8.25309150102121e-06,
      "loss": 0.2399,
      "step": 3241
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.324008194942578,
      "learning_rate": 8.251966857574002e-06,
      "loss": 0.7369,
      "step": 3242
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4251088527872975,
      "learning_rate": 8.250841928907541e-06,
      "loss": 0.2149,
      "step": 3243
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3170679917807075,
      "learning_rate": 8.249716715120497e-06,
      "loss": 0.2612,
      "step": 3244
    },
    {
      "epoch": 0.3,
      "grad_norm": 2.0946624944404793,
      "learning_rate": 8.248591216311559e-06,
      "loss": 0.2474,
      "step": 3245
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5658247441519946,
      "learning_rate": 8.247465432579436e-06,
      "loss": 0.2353,
      "step": 3246
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3837919893503043,
      "learning_rate": 8.246339364022871e-06,
      "loss": 0.2133,
      "step": 3247
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.662514086248661,
      "learning_rate": 8.245213010740628e-06,
      "loss": 0.776,
      "step": 3248
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3687176435199846,
      "learning_rate": 8.244086372831492e-06,
      "loss": 0.1614,
      "step": 3249
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3085094806760176,
      "learning_rate": 8.242959450394281e-06,
      "loss": 0.2119,
      "step": 3250
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.321439287326036,
      "learning_rate": 8.241832243527831e-06,
      "loss": 0.2163,
      "step": 3251
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4231581033188816,
      "learning_rate": 8.240704752331006e-06,
      "loss": 0.2479,
      "step": 3252
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.632746456533457,
      "learning_rate": 8.239576976902694e-06,
      "loss": 0.6087,
      "step": 3253
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3647060886109141,
      "learning_rate": 8.23844891734181e-06,
      "loss": 0.2351,
      "step": 3254
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4250919392768433,
      "learning_rate": 8.237320573747292e-06,
      "loss": 0.2295,
      "step": 3255
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3953049594141913,
      "learning_rate": 8.236191946218101e-06,
      "loss": 0.2002,
      "step": 3256
    },
    {
      "epoch": 0.3,
      "grad_norm": 5.5015628123802935,
      "learning_rate": 8.235063034853228e-06,
      "loss": 0.7956,
      "step": 3257
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.598395149200212,
      "learning_rate": 8.233933839751686e-06,
      "loss": 0.2252,
      "step": 3258
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2920445528275752,
      "learning_rate": 8.232804361012511e-06,
      "loss": 0.1934,
      "step": 3259
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.298026522982601,
      "learning_rate": 8.231674598734766e-06,
      "loss": 0.2076,
      "step": 3260
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.409018085343699,
      "learning_rate": 8.23054455301754e-06,
      "loss": 0.2337,
      "step": 3261
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3055444557826785,
      "learning_rate": 8.229414223959946e-06,
      "loss": 0.1892,
      "step": 3262
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4187494344821736,
      "learning_rate": 8.228283611661122e-06,
      "loss": 0.2172,
      "step": 3263
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5707596444929988,
      "learning_rate": 8.227152716220227e-06,
      "loss": 0.2319,
      "step": 3264
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.439175341089941,
      "learning_rate": 8.226021537736449e-06,
      "loss": 0.1952,
      "step": 3265
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5243414761383516,
      "learning_rate": 8.224890076309003e-06,
      "loss": 0.1736,
      "step": 3266
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2851434780016477,
      "learning_rate": 8.223758332037121e-06,
      "loss": 0.2046,
      "step": 3267
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.12398142687226,
      "learning_rate": 8.222626305020069e-06,
      "loss": 0.658,
      "step": 3268
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.189710531587487,
      "learning_rate": 8.22149399535713e-06,
      "loss": 0.179,
      "step": 3269
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4289923245151728,
      "learning_rate": 8.220361403147618e-06,
      "loss": 0.212,
      "step": 3270
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2275388834426104,
      "learning_rate": 8.219228528490868e-06,
      "loss": 0.2155,
      "step": 3271
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.8561396263415855,
      "learning_rate": 8.21809537148624e-06,
      "loss": 0.6083,
      "step": 3272
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3269721132521135,
      "learning_rate": 8.216961932233118e-06,
      "loss": 0.2423,
      "step": 3273
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4470699653340613,
      "learning_rate": 8.215828210830915e-06,
      "loss": 0.2474,
      "step": 3274
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.400663088207073,
      "learning_rate": 8.214694207379064e-06,
      "loss": 0.1918,
      "step": 3275
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3186937845308726,
      "learning_rate": 8.213559921977026e-06,
      "loss": 0.2015,
      "step": 3276
    },
    {
      "epoch": 0.3,
      "grad_norm": 4.663654796374677,
      "learning_rate": 8.212425354724284e-06,
      "loss": 0.5548,
      "step": 3277
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3527650940585338,
      "learning_rate": 8.21129050572035e-06,
      "loss": 0.2402,
      "step": 3278
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.338622982683647,
      "learning_rate": 8.210155375064756e-06,
      "loss": 0.1848,
      "step": 3279
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2817878373419056,
      "learning_rate": 8.209019962857058e-06,
      "loss": 0.2041,
      "step": 3280
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3261851843335741,
      "learning_rate": 8.207884269196845e-06,
      "loss": 0.2174,
      "step": 3281
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3153309687870016,
      "learning_rate": 8.206748294183718e-06,
      "loss": 0.2256,
      "step": 3282
    },
    {
      "epoch": 0.3,
      "grad_norm": 9.755470437294749,
      "learning_rate": 8.205612037917316e-06,
      "loss": 0.6347,
      "step": 3283
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3665694145929386,
      "learning_rate": 8.20447550049729e-06,
      "loss": 0.2519,
      "step": 3284
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4595546731070028,
      "learning_rate": 8.20333868202333e-06,
      "loss": 0.2316,
      "step": 3285
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.3638048514944545,
      "learning_rate": 8.202201582595135e-06,
      "loss": 0.6588,
      "step": 3286
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.243070700791882,
      "learning_rate": 8.20106420231244e-06,
      "loss": 0.1807,
      "step": 3287
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3839248866625247,
      "learning_rate": 8.199926541275e-06,
      "loss": 0.1888,
      "step": 3288
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4706918131000428,
      "learning_rate": 8.198788599582596e-06,
      "loss": 0.2512,
      "step": 3289
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3185944310356266,
      "learning_rate": 8.197650377335034e-06,
      "loss": 0.218,
      "step": 3290
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2743931993441686,
      "learning_rate": 8.19651187463214e-06,
      "loss": 0.2411,
      "step": 3291
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.468093741965218,
      "learning_rate": 8.195373091573771e-06,
      "loss": 0.2124,
      "step": 3292
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3481219070055457,
      "learning_rate": 8.194234028259806e-06,
      "loss": 0.1773,
      "step": 3293
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3439985740425269,
      "learning_rate": 8.19309468479015e-06,
      "loss": 0.2126,
      "step": 3294
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4374514072230893,
      "learning_rate": 8.191955061264727e-06,
      "loss": 0.2166,
      "step": 3295
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3297829486217125,
      "learning_rate": 8.190815157783492e-06,
      "loss": 0.2098,
      "step": 3296
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3141211928986756,
      "learning_rate": 8.189674974446423e-06,
      "loss": 0.207,
      "step": 3297
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3036630570589736,
      "learning_rate": 8.188534511353518e-06,
      "loss": 0.2124,
      "step": 3298
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3189151517821571,
      "learning_rate": 8.187393768604808e-06,
      "loss": 0.2054,
      "step": 3299
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6054091282567724,
      "learning_rate": 8.186252746300342e-06,
      "loss": 0.2177,
      "step": 3300
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3917294413595107,
      "learning_rate": 8.185111444540193e-06,
      "loss": 0.2082,
      "step": 3301
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4243683008206998,
      "learning_rate": 8.183969863424463e-06,
      "loss": 0.2162,
      "step": 3302
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3548608839298624,
      "learning_rate": 8.182828003053278e-06,
      "loss": 0.2355,
      "step": 3303
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.240103014727166,
      "learning_rate": 8.181685863526783e-06,
      "loss": 0.2117,
      "step": 3304
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5112892533640887,
      "learning_rate": 8.180543444945154e-06,
      "loss": 0.2604,
      "step": 3305
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6074344043373143,
      "learning_rate": 8.179400747408586e-06,
      "loss": 0.2407,
      "step": 3306
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3496761789815221,
      "learning_rate": 8.178257771017307e-06,
      "loss": 0.2109,
      "step": 3307
    },
    {
      "epoch": 0.3,
      "grad_norm": 10.504585883125163,
      "learning_rate": 8.177114515871556e-06,
      "loss": 0.645,
      "step": 3308
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3969335551667326,
      "learning_rate": 8.175970982071607e-06,
      "loss": 0.2615,
      "step": 3309
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4305620101573213,
      "learning_rate": 8.17482716971776e-06,
      "loss": 0.223,
      "step": 3310
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3664978387656612,
      "learning_rate": 8.17368307891033e-06,
      "loss": 0.2058,
      "step": 3311
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3333620595488012,
      "learning_rate": 8.172538709749663e-06,
      "loss": 0.2103,
      "step": 3312
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1746530130532418,
      "learning_rate": 8.171394062336127e-06,
      "loss": 0.1577,
      "step": 3313
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3280486356863432,
      "learning_rate": 8.170249136770114e-06,
      "loss": 0.2381,
      "step": 3314
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3691010196804576,
      "learning_rate": 8.169103933152047e-06,
      "loss": 0.2427,
      "step": 3315
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3628715864654921,
      "learning_rate": 8.167958451582364e-06,
      "loss": 0.2384,
      "step": 3316
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5645841965849707,
      "learning_rate": 8.16681269216153e-06,
      "loss": 0.2431,
      "step": 3317
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.745963049132989,
      "learning_rate": 8.165666654990038e-06,
      "loss": 0.6722,
      "step": 3318
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3660500397278563,
      "learning_rate": 8.164520340168404e-06,
      "loss": 0.1963,
      "step": 3319
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5318066695937471,
      "learning_rate": 8.163373747797166e-06,
      "loss": 0.2405,
      "step": 3320
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.169232405689942,
      "learning_rate": 8.162226877976886e-06,
      "loss": 0.1938,
      "step": 3321
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3085888594101607,
      "learning_rate": 8.161079730808156e-06,
      "loss": 0.1771,
      "step": 3322
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3177625620057256,
      "learning_rate": 8.159932306391587e-06,
      "loss": 0.1759,
      "step": 3323
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3813773910060458,
      "learning_rate": 8.158784604827816e-06,
      "loss": 0.2695,
      "step": 3324
    },
    {
      "epoch": 0.3,
      "grad_norm": 8.056596449729756,
      "learning_rate": 8.157636626217504e-06,
      "loss": 0.5987,
      "step": 3325
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2508994169852572,
      "learning_rate": 8.156488370661336e-06,
      "loss": 0.1958,
      "step": 3326
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4303416410939307,
      "learning_rate": 8.155339838260018e-06,
      "loss": 0.2344,
      "step": 3327
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.474110975919125,
      "learning_rate": 8.154191029114292e-06,
      "loss": 0.2422,
      "step": 3328
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.480411080526752,
      "learning_rate": 8.153041943324912e-06,
      "loss": 0.251,
      "step": 3329
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3815206054725437,
      "learning_rate": 8.151892580992657e-06,
      "loss": 0.248,
      "step": 3330
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.237494407904517,
      "learning_rate": 8.15074294221834e-06,
      "loss": 0.1798,
      "step": 3331
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4726440983349627,
      "learning_rate": 8.14959302710279e-06,
      "loss": 0.1977,
      "step": 3332
    },
    {
      "epoch": 0.3,
      "grad_norm": 6.162921075350874,
      "learning_rate": 8.14844283574686e-06,
      "loss": 0.6754,
      "step": 3333
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3400215082319078,
      "learning_rate": 8.14729236825143e-06,
      "loss": 0.2451,
      "step": 3334
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4288268355320985,
      "learning_rate": 8.146141624717406e-06,
      "loss": 0.2929,
      "step": 3335
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4753678284102245,
      "learning_rate": 8.144990605245714e-06,
      "loss": 0.2703,
      "step": 3336
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6642558075755276,
      "learning_rate": 8.143839309937307e-06,
      "loss": 0.214,
      "step": 3337
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.522712845848004,
      "learning_rate": 8.142687738893161e-06,
      "loss": 0.2742,
      "step": 3338
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4257288852608512,
      "learning_rate": 8.141535892214277e-06,
      "loss": 0.2366,
      "step": 3339
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4732438598022648,
      "learning_rate": 8.140383770001678e-06,
      "loss": 0.2028,
      "step": 3340
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.466562976866913,
      "learning_rate": 8.139231372356415e-06,
      "loss": 0.1888,
      "step": 3341
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2630910876629475,
      "learning_rate": 8.138078699379559e-06,
      "loss": 0.2158,
      "step": 3342
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.44154763065511,
      "learning_rate": 8.13692575117221e-06,
      "loss": 0.5838,
      "step": 3343
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.215439678326269,
      "learning_rate": 8.135772527835485e-06,
      "loss": 0.2079,
      "step": 3344
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.027684026090245,
      "learning_rate": 8.134619029470535e-06,
      "loss": 0.4153,
      "step": 3345
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2726837363781678,
      "learning_rate": 8.133465256178522e-06,
      "loss": 0.179,
      "step": 3346
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.00909985951613,
      "learning_rate": 8.132311208060645e-06,
      "loss": 0.504,
      "step": 3347
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4109002189115039,
      "learning_rate": 8.131156885218123e-06,
      "loss": 0.1772,
      "step": 3348
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.694161638397515,
      "learning_rate": 8.130002287752194e-06,
      "loss": 0.7549,
      "step": 3349
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4999058287123974,
      "learning_rate": 8.128847415764124e-06,
      "loss": 0.2454,
      "step": 3350
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3910089314928347,
      "learning_rate": 8.127692269355207e-06,
      "loss": 0.2313,
      "step": 3351
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2804401109660912,
      "learning_rate": 8.126536848626754e-06,
      "loss": 0.2047,
      "step": 3352
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3652395848772374,
      "learning_rate": 8.125381153680103e-06,
      "loss": 0.2109,
      "step": 3353
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.326437937966779,
      "learning_rate": 8.124225184616617e-06,
      "loss": 0.2002,
      "step": 3354
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.484250019492903,
      "learning_rate": 8.123068941537681e-06,
      "loss": 0.2293,
      "step": 3355
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3323255928101982,
      "learning_rate": 8.121912424544706e-06,
      "loss": 0.2107,
      "step": 3356
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.533198435717517,
      "learning_rate": 8.120755633739128e-06,
      "loss": 0.2337,
      "step": 3357
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3442604961675764,
      "learning_rate": 8.119598569222404e-06,
      "loss": 0.2066,
      "step": 3358
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4309843869770806,
      "learning_rate": 8.118441231096016e-06,
      "loss": 0.1968,
      "step": 3359
    },
    {
      "epoch": 0.31,
      "grad_norm": 7.115306231288505,
      "learning_rate": 8.117283619461469e-06,
      "loss": 0.6336,
      "step": 3360
    },
    {
      "epoch": 0.31,
      "grad_norm": 12.36131139811745,
      "learning_rate": 8.116125734420297e-06,
      "loss": 0.674,
      "step": 3361
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.836204895594616,
      "learning_rate": 8.11496757607405e-06,
      "loss": 0.4783,
      "step": 3362
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.356619320215389,
      "learning_rate": 8.11380914452431e-06,
      "loss": 0.213,
      "step": 3363
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.448328811327283,
      "learning_rate": 8.112650439872674e-06,
      "loss": 0.2495,
      "step": 3364
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4713758132947912,
      "learning_rate": 8.111491462220776e-06,
      "loss": 0.2259,
      "step": 3365
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2916957165753382,
      "learning_rate": 8.110332211670257e-06,
      "loss": 0.2293,
      "step": 3366
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4778631246733884,
      "learning_rate": 8.1091726883228e-06,
      "loss": 0.1945,
      "step": 3367
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4166347138179343,
      "learning_rate": 8.108012892280094e-06,
      "loss": 0.2085,
      "step": 3368
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4686291676878531,
      "learning_rate": 8.10685282364387e-06,
      "loss": 0.2223,
      "step": 3369
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4180568481399458,
      "learning_rate": 8.105692482515868e-06,
      "loss": 0.2409,
      "step": 3370
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3602454190417306,
      "learning_rate": 8.104531868997858e-06,
      "loss": 0.216,
      "step": 3371
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4123647774095724,
      "learning_rate": 8.103370983191634e-06,
      "loss": 0.2432,
      "step": 3372
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.444879301146441,
      "learning_rate": 8.102209825199016e-06,
      "loss": 0.2468,
      "step": 3373
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3676163787223046,
      "learning_rate": 8.101048395121843e-06,
      "loss": 0.2079,
      "step": 3374
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.332280338682339,
      "learning_rate": 8.099886693061978e-06,
      "loss": 0.2296,
      "step": 3375
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.871208677121005,
      "learning_rate": 8.098724719121314e-06,
      "loss": 0.6097,
      "step": 3376
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4122288810140533,
      "learning_rate": 8.097562473401764e-06,
      "loss": 0.2368,
      "step": 3377
    },
    {
      "epoch": 0.31,
      "grad_norm": 10.595787650240085,
      "learning_rate": 8.09639995600526e-06,
      "loss": 0.6467,
      "step": 3378
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3757335598651117,
      "learning_rate": 8.095237167033769e-06,
      "loss": 0.2402,
      "step": 3379
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.618548940220795,
      "learning_rate": 8.094074106589269e-06,
      "loss": 0.5548,
      "step": 3380
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3644097584574773,
      "learning_rate": 8.092910774773773e-06,
      "loss": 0.1765,
      "step": 3381
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.115677711303015,
      "learning_rate": 8.091747171689312e-06,
      "loss": 0.5864,
      "step": 3382
    },
    {
      "epoch": 0.31,
      "grad_norm": 9.301298117633134,
      "learning_rate": 8.090583297437939e-06,
      "loss": 0.5693,
      "step": 3383
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4187978596995958,
      "learning_rate": 8.089419152121736e-06,
      "loss": 0.2443,
      "step": 3384
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4114506564064275,
      "learning_rate": 8.088254735842808e-06,
      "loss": 0.2213,
      "step": 3385
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3845116674360987,
      "learning_rate": 8.087090048703278e-06,
      "loss": 0.2226,
      "step": 3386
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2462491409205745,
      "learning_rate": 8.085925090805298e-06,
      "loss": 0.1994,
      "step": 3387
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.69120537653236,
      "learning_rate": 8.084759862251047e-06,
      "loss": 0.6229,
      "step": 3388
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.2632978198784715,
      "learning_rate": 8.083594363142717e-06,
      "loss": 0.633,
      "step": 3389
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3753395104219905,
      "learning_rate": 8.082428593582531e-06,
      "loss": 0.2061,
      "step": 3390
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3798300606268321,
      "learning_rate": 8.081262553672742e-06,
      "loss": 0.2225,
      "step": 3391
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4785806181490455,
      "learning_rate": 8.08009624351561e-06,
      "loss": 0.2577,
      "step": 3392
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.37814383888239,
      "learning_rate": 8.078929663213432e-06,
      "loss": 0.6263,
      "step": 3393
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2646471642085269,
      "learning_rate": 8.077762812868527e-06,
      "loss": 0.2075,
      "step": 3394
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2790476674421933,
      "learning_rate": 8.076595692583235e-06,
      "loss": 0.2251,
      "step": 3395
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.1653570291934665,
      "learning_rate": 8.075428302459916e-06,
      "loss": 0.5764,
      "step": 3396
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.242856502163035,
      "learning_rate": 8.074260642600963e-06,
      "loss": 0.6207,
      "step": 3397
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.44894121688754,
      "learning_rate": 8.073092713108785e-06,
      "loss": 0.1726,
      "step": 3398
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.32034401779048,
      "learning_rate": 8.071924514085818e-06,
      "loss": 0.1697,
      "step": 3399
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.7150660525513945,
      "learning_rate": 8.070756045634522e-06,
      "loss": 0.6259,
      "step": 3400
    },
    {
      "epoch": 0.31,
      "grad_norm": 5.377873961808898,
      "learning_rate": 8.069587307857377e-06,
      "loss": 0.559,
      "step": 3401
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.451554279641545,
      "learning_rate": 8.06841830085689e-06,
      "loss": 0.1915,
      "step": 3402
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4506466638953268,
      "learning_rate": 8.06724902473559e-06,
      "loss": 0.2789,
      "step": 3403
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3349420510717882,
      "learning_rate": 8.066079479596035e-06,
      "loss": 0.1756,
      "step": 3404
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2098982678772896,
      "learning_rate": 8.064909665540794e-06,
      "loss": 0.1835,
      "step": 3405
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5563984663309096,
      "learning_rate": 8.063739582672474e-06,
      "loss": 0.2535,
      "step": 3406
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2799857827424113,
      "learning_rate": 8.062569231093698e-06,
      "loss": 0.2089,
      "step": 3407
    },
    {
      "epoch": 0.31,
      "grad_norm": 4.187158379260508,
      "learning_rate": 8.06139861090711e-06,
      "loss": 0.5654,
      "step": 3408
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5603797855008321,
      "learning_rate": 8.060227722215385e-06,
      "loss": 0.2395,
      "step": 3409
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2111654142537087,
      "learning_rate": 8.059056565121218e-06,
      "loss": 0.2005,
      "step": 3410
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3335267866116907,
      "learning_rate": 8.057885139727324e-06,
      "loss": 0.1594,
      "step": 3411
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4238964650392474,
      "learning_rate": 8.056713446136447e-06,
      "loss": 0.1903,
      "step": 3412
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3388219707423532,
      "learning_rate": 8.055541484451353e-06,
      "loss": 0.22,
      "step": 3413
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4854656607253045,
      "learning_rate": 8.05436925477483e-06,
      "loss": 0.2024,
      "step": 3414
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.83887899909533,
      "learning_rate": 8.05319675720969e-06,
      "loss": 0.6465,
      "step": 3415
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.259961960425171,
      "learning_rate": 8.052023991858769e-06,
      "loss": 0.4978,
      "step": 3416
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6304935985752187,
      "learning_rate": 8.050850958824926e-06,
      "loss": 0.2456,
      "step": 3417
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3690692015520334,
      "learning_rate": 8.049677658211045e-06,
      "loss": 0.2084,
      "step": 3418
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4240761160439723,
      "learning_rate": 8.04850409012003e-06,
      "loss": 0.2912,
      "step": 3419
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5819424151618058,
      "learning_rate": 8.047330254654815e-06,
      "loss": 0.2546,
      "step": 3420
    },
    {
      "epoch": 0.31,
      "grad_norm": 8.960776734239195,
      "learning_rate": 8.046156151918351e-06,
      "loss": 0.6467,
      "step": 3421
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4706945811877332,
      "learning_rate": 8.044981782013614e-06,
      "loss": 0.2115,
      "step": 3422
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.379773672110705,
      "learning_rate": 8.043807145043604e-06,
      "loss": 0.2127,
      "step": 3423
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.9071245796949925,
      "learning_rate": 8.042632241111346e-06,
      "loss": 0.4458,
      "step": 3424
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4378696142999348,
      "learning_rate": 8.041457070319884e-06,
      "loss": 0.2217,
      "step": 3425
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2075302091281537,
      "learning_rate": 8.040281632772292e-06,
      "loss": 0.1741,
      "step": 3426
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4250233307188251,
      "learning_rate": 8.039105928571662e-06,
      "loss": 0.2019,
      "step": 3427
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4452620383059285,
      "learning_rate": 8.037929957821109e-06,
      "loss": 0.2368,
      "step": 3428
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7336633351292492,
      "learning_rate": 8.036753720623778e-06,
      "loss": 0.2366,
      "step": 3429
    },
    {
      "epoch": 0.31,
      "grad_norm": 6.139265301295478,
      "learning_rate": 8.03557721708283e-06,
      "loss": 0.8848,
      "step": 3430
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3298164306905158,
      "learning_rate": 8.034400447301454e-06,
      "loss": 0.2131,
      "step": 3431
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3763708139479534,
      "learning_rate": 8.033223411382858e-06,
      "loss": 0.217,
      "step": 3432
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2152977335346469,
      "learning_rate": 8.032046109430276e-06,
      "loss": 0.1565,
      "step": 3433
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.66289938026063,
      "learning_rate": 8.03086854154697e-06,
      "loss": 0.2355,
      "step": 3434
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2053764218059393,
      "learning_rate": 8.029690707836213e-06,
      "loss": 0.1397,
      "step": 3435
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3139152915192764,
      "learning_rate": 8.028512608401315e-06,
      "loss": 0.241,
      "step": 3436
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3475565966353336,
      "learning_rate": 8.027334243345601e-06,
      "loss": 0.2503,
      "step": 3437
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.418463619105596,
      "learning_rate": 8.026155612772424e-06,
      "loss": 0.2146,
      "step": 3438
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3979657095877467,
      "learning_rate": 8.024976716785153e-06,
      "loss": 0.2141,
      "step": 3439
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6221476059884998,
      "learning_rate": 8.023797555487188e-06,
      "loss": 0.1898,
      "step": 3440
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4273636498796192,
      "learning_rate": 8.02261812898195e-06,
      "loss": 0.1978,
      "step": 3441
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4257229038426542,
      "learning_rate": 8.021438437372879e-06,
      "loss": 0.1754,
      "step": 3442
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4053719329886611,
      "learning_rate": 8.020258480763448e-06,
      "loss": 0.1993,
      "step": 3443
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4941364241556614,
      "learning_rate": 8.01907825925714e-06,
      "loss": 0.2713,
      "step": 3444
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.30703905893331,
      "learning_rate": 8.017897772957473e-06,
      "loss": 0.1881,
      "step": 3445
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.712499369596143,
      "learning_rate": 8.016717021967983e-06,
      "loss": 0.6525,
      "step": 3446
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3685874177785746,
      "learning_rate": 8.015536006392229e-06,
      "loss": 0.2183,
      "step": 3447
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4329246963173419,
      "learning_rate": 8.014354726333793e-06,
      "loss": 0.2205,
      "step": 3448
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.167507505465687,
      "learning_rate": 8.013173181896283e-06,
      "loss": 0.1694,
      "step": 3449
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4979242653009848,
      "learning_rate": 8.011991373183327e-06,
      "loss": 0.2571,
      "step": 3450
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.209290731489003,
      "learning_rate": 8.010809300298578e-06,
      "loss": 0.6635,
      "step": 3451
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4205762981472483,
      "learning_rate": 8.009626963345712e-06,
      "loss": 0.1936,
      "step": 3452
    },
    {
      "epoch": 0.32,
      "grad_norm": 7.484172959091661,
      "learning_rate": 8.008444362428428e-06,
      "loss": 0.591,
      "step": 3453
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.398160685506038,
      "learning_rate": 8.007261497650448e-06,
      "loss": 0.1978,
      "step": 3454
    },
    {
      "epoch": 0.32,
      "grad_norm": 11.067881640953463,
      "learning_rate": 8.006078369115516e-06,
      "loss": 0.5092,
      "step": 3455
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4388722136438292,
      "learning_rate": 8.004894976927402e-06,
      "loss": 0.2252,
      "step": 3456
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.099090514125947,
      "learning_rate": 8.003711321189895e-06,
      "loss": 0.5031,
      "step": 3457
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.279979315089309,
      "learning_rate": 8.002527402006813e-06,
      "loss": 0.1878,
      "step": 3458
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4194675604311298,
      "learning_rate": 8.00134321948199e-06,
      "loss": 0.2522,
      "step": 3459
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3341658621605317,
      "learning_rate": 8.000158773719288e-06,
      "loss": 0.2336,
      "step": 3460
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3677612182386971,
      "learning_rate": 7.998974064822593e-06,
      "loss": 0.2228,
      "step": 3461
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4622346237283694,
      "learning_rate": 7.997789092895809e-06,
      "loss": 0.2219,
      "step": 3462
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.388402620219671,
      "learning_rate": 7.996603858042867e-06,
      "loss": 0.2108,
      "step": 3463
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.041903455651804,
      "learning_rate": 7.995418360367722e-06,
      "loss": 0.2351,
      "step": 3464
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5242476690462365,
      "learning_rate": 7.994232599974346e-06,
      "loss": 0.2699,
      "step": 3465
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4042622701851721,
      "learning_rate": 7.993046576966742e-06,
      "loss": 0.1979,
      "step": 3466
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.064265640610629,
      "learning_rate": 7.991860291448927e-06,
      "loss": 0.4503,
      "step": 3467
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4974577894807723,
      "learning_rate": 7.990673743524953e-06,
      "loss": 0.2466,
      "step": 3468
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.370011910298337,
      "learning_rate": 7.989486933298884e-06,
      "loss": 0.266,
      "step": 3469
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3767572565247628,
      "learning_rate": 7.98829986087481e-06,
      "loss": 0.1772,
      "step": 3470
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3512987019197202,
      "learning_rate": 7.98711252635685e-06,
      "loss": 0.243,
      "step": 3471
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3008023951022325,
      "learning_rate": 7.985924929849139e-06,
      "loss": 0.1755,
      "step": 3472
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5379079155954398,
      "learning_rate": 7.984737071455834e-06,
      "loss": 0.2364,
      "step": 3473
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2474977298779097,
      "learning_rate": 7.983548951281121e-06,
      "loss": 0.2063,
      "step": 3474
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.276340324762916,
      "learning_rate": 7.982360569429206e-06,
      "loss": 0.1951,
      "step": 3475
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.313641643281499,
      "learning_rate": 7.981171926004319e-06,
      "loss": 0.1905,
      "step": 3476
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5813797202337827,
      "learning_rate": 7.97998302111071e-06,
      "loss": 0.2287,
      "step": 3477
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4301369738739909,
      "learning_rate": 7.978793854852655e-06,
      "loss": 0.1913,
      "step": 3478
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4745327180736343,
      "learning_rate": 7.977604427334452e-06,
      "loss": 0.2062,
      "step": 3479
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4359898724939106,
      "learning_rate": 7.976414738660418e-06,
      "loss": 0.2168,
      "step": 3480
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5647673651574114,
      "learning_rate": 7.975224788934903e-06,
      "loss": 0.2507,
      "step": 3481
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2717973956948834,
      "learning_rate": 7.97403457826227e-06,
      "loss": 0.1821,
      "step": 3482
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5539972551246253,
      "learning_rate": 7.972844106746908e-06,
      "loss": 0.2492,
      "step": 3483
    },
    {
      "epoch": 0.32,
      "grad_norm": 20.420218797544965,
      "learning_rate": 7.97165337449323e-06,
      "loss": 0.4692,
      "step": 3484
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3101742534012757,
      "learning_rate": 7.970462381605675e-06,
      "loss": 0.231,
      "step": 3485
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.260193387994176,
      "learning_rate": 7.969271128188693e-06,
      "loss": 0.1827,
      "step": 3486
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3919539879376595,
      "learning_rate": 7.968079614346771e-06,
      "loss": 0.2349,
      "step": 3487
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5975530339439301,
      "learning_rate": 7.966887840184412e-06,
      "loss": 0.2608,
      "step": 3488
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5743532368560562,
      "learning_rate": 7.965695805806141e-06,
      "loss": 0.2564,
      "step": 3489
    },
    {
      "epoch": 0.32,
      "grad_norm": 12.308908128887996,
      "learning_rate": 7.964503511316509e-06,
      "loss": 0.4749,
      "step": 3490
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3049772995215791,
      "learning_rate": 7.963310956820085e-06,
      "loss": 0.2171,
      "step": 3491
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3099192570116072,
      "learning_rate": 7.962118142421467e-06,
      "loss": 0.1955,
      "step": 3492
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.333147567055041,
      "learning_rate": 7.960925068225272e-06,
      "loss": 0.2066,
      "step": 3493
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.8530477107305983,
      "learning_rate": 7.959731734336143e-06,
      "loss": 0.2455,
      "step": 3494
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.343052992694613,
      "learning_rate": 7.958538140858739e-06,
      "loss": 0.2273,
      "step": 3495
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3577130182743935,
      "learning_rate": 7.957344287897748e-06,
      "loss": 0.2121,
      "step": 3496
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3293458174105797,
      "learning_rate": 7.95615017555788e-06,
      "loss": 0.2117,
      "step": 3497
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5523590045425149,
      "learning_rate": 7.954955803943864e-06,
      "loss": 0.2762,
      "step": 3498
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2834229756959878,
      "learning_rate": 7.953761173160456e-06,
      "loss": 0.1735,
      "step": 3499
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.54735086839714,
      "learning_rate": 7.952566283312435e-06,
      "loss": 0.2167,
      "step": 3500
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.452406666655797,
      "learning_rate": 7.951371134504599e-06,
      "loss": 0.1801,
      "step": 3501
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4176996995057773,
      "learning_rate": 7.95017572684177e-06,
      "loss": 0.2363,
      "step": 3502
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2864636852567044,
      "learning_rate": 7.948980060428792e-06,
      "loss": 0.1778,
      "step": 3503
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3064430457109513,
      "learning_rate": 7.947784135370537e-06,
      "loss": 0.2035,
      "step": 3504
    },
    {
      "epoch": 0.32,
      "grad_norm": 8.027781889374655,
      "learning_rate": 7.946587951771894e-06,
      "loss": 0.5275,
      "step": 3505
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5591253113431647,
      "learning_rate": 7.945391509737774e-06,
      "loss": 0.2973,
      "step": 3506
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3915706491289297,
      "learning_rate": 7.944194809373115e-06,
      "loss": 0.227,
      "step": 3507
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.428087543863104,
      "learning_rate": 7.942997850782874e-06,
      "loss": 0.233,
      "step": 3508
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4854119029443889,
      "learning_rate": 7.941800634072036e-06,
      "loss": 0.2086,
      "step": 3509
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2178819896869586,
      "learning_rate": 7.940603159345601e-06,
      "loss": 0.1707,
      "step": 3510
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4116699271897806,
      "learning_rate": 7.939405426708598e-06,
      "loss": 0.2021,
      "step": 3511
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.499072277318319,
      "learning_rate": 7.938207436266075e-06,
      "loss": 0.2147,
      "step": 3512
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3004070658421671,
      "learning_rate": 7.937009188123102e-06,
      "loss": 0.194,
      "step": 3513
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3305444404557503,
      "learning_rate": 7.935810682384777e-06,
      "loss": 0.1999,
      "step": 3514
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3293622786657686,
      "learning_rate": 7.934611919156214e-06,
      "loss": 0.1964,
      "step": 3515
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4059370193934595,
      "learning_rate": 7.933412898542556e-06,
      "loss": 0.2079,
      "step": 3516
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3209234092870288,
      "learning_rate": 7.932213620648961e-06,
      "loss": 0.1913,
      "step": 3517
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0893863254908271,
      "learning_rate": 7.931014085580617e-06,
      "loss": 0.1522,
      "step": 3518
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.9195403472040455,
      "learning_rate": 7.92981429344273e-06,
      "loss": 0.5447,
      "step": 3519
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6291911737133111,
      "learning_rate": 7.928614244340529e-06,
      "loss": 0.2121,
      "step": 3520
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3758926750030414,
      "learning_rate": 7.927413938379268e-06,
      "loss": 0.215,
      "step": 3521
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.373960565536078,
      "learning_rate": 7.926213375664219e-06,
      "loss": 0.585,
      "step": 3522
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2585170789312004,
      "learning_rate": 7.925012556300681e-06,
      "loss": 0.212,
      "step": 3523
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1681035307340875,
      "learning_rate": 7.923811480393977e-06,
      "loss": 0.1677,
      "step": 3524
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5641109357682406,
      "learning_rate": 7.922610148049445e-06,
      "loss": 0.199,
      "step": 3525
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2862137007929377,
      "learning_rate": 7.921408559372453e-06,
      "loss": 0.1996,
      "step": 3526
    },
    {
      "epoch": 0.32,
      "grad_norm": 4.8158607979473365,
      "learning_rate": 7.920206714468383e-06,
      "loss": 0.5519,
      "step": 3527
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4250153053380923,
      "learning_rate": 7.919004613442652e-06,
      "loss": 0.2139,
      "step": 3528
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2763614612925374,
      "learning_rate": 7.917802256400688e-06,
      "loss": 0.2019,
      "step": 3529
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4170360460859523,
      "learning_rate": 7.916599643447947e-06,
      "loss": 0.2338,
      "step": 3530
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3407834488859711,
      "learning_rate": 7.915396774689906e-06,
      "loss": 0.196,
      "step": 3531
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.305770636607573,
      "learning_rate": 7.914193650232064e-06,
      "loss": 0.2187,
      "step": 3532
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.360434643902874,
      "learning_rate": 7.912990270179944e-06,
      "loss": 0.5255,
      "step": 3533
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3000186775801894,
      "learning_rate": 7.91178663463909e-06,
      "loss": 0.209,
      "step": 3534
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3295047786751613,
      "learning_rate": 7.91058274371507e-06,
      "loss": 0.2345,
      "step": 3535
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.279802757792276,
      "learning_rate": 7.90937859751347e-06,
      "loss": 0.5751,
      "step": 3536
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3380010427987454,
      "learning_rate": 7.908174196139907e-06,
      "loss": 0.2243,
      "step": 3537
    },
    {
      "epoch": 0.32,
      "grad_norm": 5.43083332425324,
      "learning_rate": 7.90696953970001e-06,
      "loss": 0.634,
      "step": 3538
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.256596179730038,
      "learning_rate": 7.905764628299438e-06,
      "loss": 0.2012,
      "step": 3539
    },
    {
      "epoch": 0.32,
      "grad_norm": 6.805603393593466,
      "learning_rate": 7.904559462043869e-06,
      "loss": 0.822,
      "step": 3540
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.210792648092062,
      "learning_rate": 7.903354041039004e-06,
      "loss": 0.2114,
      "step": 3541
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3941216553472335,
      "learning_rate": 7.902148365390567e-06,
      "loss": 0.1929,
      "step": 3542
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3758667763712926,
      "learning_rate": 7.900942435204304e-06,
      "loss": 0.2391,
      "step": 3543
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3154758595061835,
      "learning_rate": 7.899736250585983e-06,
      "loss": 0.193,
      "step": 3544
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4282603317032914,
      "learning_rate": 7.898529811641393e-06,
      "loss": 0.274,
      "step": 3545
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.292724105574128,
      "learning_rate": 7.89732311847635e-06,
      "loss": 0.2439,
      "step": 3546
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4363138300368354,
      "learning_rate": 7.896116171196686e-06,
      "loss": 0.2411,
      "step": 3547
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2856324158525265,
      "learning_rate": 7.89490896990826e-06,
      "loss": 0.2035,
      "step": 3548
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6093307806719814,
      "learning_rate": 7.893701514716951e-06,
      "loss": 0.2559,
      "step": 3549
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.354532628907328,
      "learning_rate": 7.89249380572866e-06,
      "loss": 0.2212,
      "step": 3550
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4386428295653428,
      "learning_rate": 7.891285843049313e-06,
      "loss": 0.2639,
      "step": 3551
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4385377063276723,
      "learning_rate": 7.890077626784855e-06,
      "loss": 0.2321,
      "step": 3552
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6201863960734624,
      "learning_rate": 7.888869157041257e-06,
      "loss": 0.2306,
      "step": 3553
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3561192321335542,
      "learning_rate": 7.887660433924507e-06,
      "loss": 0.2565,
      "step": 3554
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.323803701010625,
      "learning_rate": 7.886451457540619e-06,
      "loss": 0.1863,
      "step": 3555
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.695966746437119,
      "learning_rate": 7.885242227995628e-06,
      "loss": 0.2422,
      "step": 3556
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.344611788552816,
      "learning_rate": 7.884032745395592e-06,
      "loss": 0.208,
      "step": 3557
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.246452212753757,
      "learning_rate": 7.882823009846592e-06,
      "loss": 0.1693,
      "step": 3558
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3156921253485863,
      "learning_rate": 7.881613021454728e-06,
      "loss": 0.2061,
      "step": 3559
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5011884208006971,
      "learning_rate": 7.880402780326123e-06,
      "loss": 0.2421,
      "step": 3560
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2742348384687208,
      "learning_rate": 7.879192286566929e-06,
      "loss": 0.1685,
      "step": 3561
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.343991435661978,
      "learning_rate": 7.877981540283306e-06,
      "loss": 0.2379,
      "step": 3562
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3232619179498541,
      "learning_rate": 7.87677054158145e-06,
      "loss": 0.197,
      "step": 3563
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5191133426451475,
      "learning_rate": 7.875559290567576e-06,
      "loss": 0.2304,
      "step": 3564
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.310367805437937,
      "learning_rate": 7.874347787347913e-06,
      "loss": 0.8066,
      "step": 3565
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4749532993149659,
      "learning_rate": 7.873136032028719e-06,
      "loss": 0.2023,
      "step": 3566
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5719843182995763,
      "learning_rate": 7.871924024716276e-06,
      "loss": 0.2873,
      "step": 3567
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2908956391059947,
      "learning_rate": 7.870711765516882e-06,
      "loss": 0.2326,
      "step": 3568
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.412227860293391,
      "learning_rate": 7.869499254536865e-06,
      "loss": 0.2464,
      "step": 3569
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2902646707537204,
      "learning_rate": 7.868286491882565e-06,
      "loss": 0.1939,
      "step": 3570
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3383137633847393,
      "learning_rate": 7.867073477660352e-06,
      "loss": 0.2201,
      "step": 3571
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4380804331666193,
      "learning_rate": 7.865860211976615e-06,
      "loss": 0.2282,
      "step": 3572
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5710253548992414,
      "learning_rate": 7.864646694937765e-06,
      "loss": 0.2252,
      "step": 3573
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.294391301174928,
      "learning_rate": 7.863432926650237e-06,
      "loss": 0.2269,
      "step": 3574
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5685004994978058,
      "learning_rate": 7.862218907220484e-06,
      "loss": 0.2138,
      "step": 3575
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3415706041980437,
      "learning_rate": 7.861004636754986e-06,
      "loss": 0.2279,
      "step": 3576
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3133758883169464,
      "learning_rate": 7.859790115360243e-06,
      "loss": 0.1942,
      "step": 3577
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4158923523094527,
      "learning_rate": 7.858575343142774e-06,
      "loss": 0.2534,
      "step": 3578
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2561740748904309,
      "learning_rate": 7.857360320209126e-06,
      "loss": 0.2176,
      "step": 3579
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3174586225081248,
      "learning_rate": 7.856145046665861e-06,
      "loss": 0.191,
      "step": 3580
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3677321305680115,
      "learning_rate": 7.85492952261957e-06,
      "loss": 0.1908,
      "step": 3581
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4288926044608417,
      "learning_rate": 7.85371374817686e-06,
      "loss": 0.2394,
      "step": 3582
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.472774924760673,
      "learning_rate": 7.852497723444364e-06,
      "loss": 0.6164,
      "step": 3583
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4529283484942448,
      "learning_rate": 7.851281448528735e-06,
      "loss": 0.2079,
      "step": 3584
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3510219988264034,
      "learning_rate": 7.850064923536649e-06,
      "loss": 0.2041,
      "step": 3585
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.686447515314759,
      "learning_rate": 7.848848148574803e-06,
      "loss": 0.4665,
      "step": 3586
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3194653830133825,
      "learning_rate": 7.847631123749915e-06,
      "loss": 0.2068,
      "step": 3587
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2739195376968488,
      "learning_rate": 7.84641384916873e-06,
      "loss": 0.185,
      "step": 3588
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3678112654493695,
      "learning_rate": 7.845196324938006e-06,
      "loss": 0.2184,
      "step": 3589
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3198122641770276,
      "learning_rate": 7.843978551164531e-06,
      "loss": 0.1746,
      "step": 3590
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.538317390518146,
      "learning_rate": 7.842760527955114e-06,
      "loss": 0.5936,
      "step": 3591
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5637423713522607,
      "learning_rate": 7.841542255416579e-06,
      "loss": 0.2234,
      "step": 3592
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6164045066373722,
      "learning_rate": 7.84032373365578e-06,
      "loss": 0.1909,
      "step": 3593
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3448087868760168,
      "learning_rate": 7.839104962779586e-06,
      "loss": 0.1789,
      "step": 3594
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2938267049789187,
      "learning_rate": 7.837885942894899e-06,
      "loss": 0.2015,
      "step": 3595
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2486396478836779,
      "learning_rate": 7.836666674108626e-06,
      "loss": 0.162,
      "step": 3596
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.403095064733091,
      "learning_rate": 7.835447156527709e-06,
      "loss": 0.2548,
      "step": 3597
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2617704194588473,
      "learning_rate": 7.83422739025911e-06,
      "loss": 0.1973,
      "step": 3598
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.381324763924456,
      "learning_rate": 7.833007375409807e-06,
      "loss": 0.2431,
      "step": 3599
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4540578461422033,
      "learning_rate": 7.831787112086806e-06,
      "loss": 0.2156,
      "step": 3600
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.887924956920562,
      "learning_rate": 7.83056660039713e-06,
      "loss": 0.5853,
      "step": 3601
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.445991795779826,
      "learning_rate": 7.829345840447828e-06,
      "loss": 0.2495,
      "step": 3602
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5435930358992753,
      "learning_rate": 7.82812483234597e-06,
      "loss": 0.271,
      "step": 3603
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5733169541809413,
      "learning_rate": 7.826903576198643e-06,
      "loss": 0.2407,
      "step": 3604
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.406845015046844,
      "learning_rate": 7.82568207211296e-06,
      "loss": 0.2109,
      "step": 3605
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4413953011736114,
      "learning_rate": 7.824460320196057e-06,
      "loss": 0.1993,
      "step": 3606
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5664173536546813,
      "learning_rate": 7.82323832055509e-06,
      "loss": 0.2629,
      "step": 3607
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.328652520060878,
      "learning_rate": 7.822016073297233e-06,
      "loss": 0.1964,
      "step": 3608
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4162120774582059,
      "learning_rate": 7.82079357852969e-06,
      "loss": 0.2239,
      "step": 3609
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.0456810965041825,
      "learning_rate": 7.81957083635968e-06,
      "loss": 0.7011,
      "step": 3610
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.330807364730372,
      "learning_rate": 7.818347846894445e-06,
      "loss": 0.2096,
      "step": 3611
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.358352548560822,
      "learning_rate": 7.817124610241249e-06,
      "loss": 0.2175,
      "step": 3612
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4808999980847082,
      "learning_rate": 7.81590112650738e-06,
      "loss": 0.2159,
      "step": 3613
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.16120549782127,
      "learning_rate": 7.814677395800143e-06,
      "loss": 0.5499,
      "step": 3614
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.285741839728571,
      "learning_rate": 7.81345341822687e-06,
      "loss": 0.1819,
      "step": 3615
    },
    {
      "epoch": 0.33,
      "grad_norm": 5.305638320021748,
      "learning_rate": 7.81222919389491e-06,
      "loss": 0.6595,
      "step": 3616
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4835931193502905,
      "learning_rate": 7.811004722911637e-06,
      "loss": 0.2614,
      "step": 3617
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4021884066080792,
      "learning_rate": 7.809780005384446e-06,
      "loss": 0.2469,
      "step": 3618
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.174556796639947,
      "learning_rate": 7.80855504142075e-06,
      "loss": 0.1638,
      "step": 3619
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3667807139521295,
      "learning_rate": 7.80732983112799e-06,
      "loss": 0.2146,
      "step": 3620
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.362237921676946,
      "learning_rate": 7.806104374613622e-06,
      "loss": 0.2532,
      "step": 3621
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4527387600990576,
      "learning_rate": 7.804878671985129e-06,
      "loss": 0.2253,
      "step": 3622
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3068311276436229,
      "learning_rate": 7.803652723350013e-06,
      "loss": 0.213,
      "step": 3623
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3402516151566057,
      "learning_rate": 7.802426528815796e-06,
      "loss": 0.2048,
      "step": 3624
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2616795123100333,
      "learning_rate": 7.801200088490026e-06,
      "loss": 0.1781,
      "step": 3625
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.293564312515553,
      "learning_rate": 7.799973402480267e-06,
      "loss": 0.2159,
      "step": 3626
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.460410797577893,
      "learning_rate": 7.798746470894113e-06,
      "loss": 0.1791,
      "step": 3627
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5958127849485944,
      "learning_rate": 7.797519293839166e-06,
      "loss": 0.216,
      "step": 3628
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4759031711597799,
      "learning_rate": 7.796291871423065e-06,
      "loss": 0.2397,
      "step": 3629
    },
    {
      "epoch": 0.33,
      "grad_norm": 23.368073315198583,
      "learning_rate": 7.79506420375346e-06,
      "loss": 0.4688,
      "step": 3630
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4454274314787257,
      "learning_rate": 7.793836290938026e-06,
      "loss": 0.2241,
      "step": 3631
    },
    {
      "epoch": 0.33,
      "grad_norm": 9.221823433090915,
      "learning_rate": 7.792608133084459e-06,
      "loss": 0.5853,
      "step": 3632
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2588643360895033,
      "learning_rate": 7.791379730300476e-06,
      "loss": 0.1899,
      "step": 3633
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3495236169273273,
      "learning_rate": 7.79015108269382e-06,
      "loss": 0.1612,
      "step": 3634
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5261932914440768,
      "learning_rate": 7.788922190372245e-06,
      "loss": 0.2518,
      "step": 3635
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.885480163391736,
      "learning_rate": 7.78769305344354e-06,
      "loss": 0.559,
      "step": 3636
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3976485518245214,
      "learning_rate": 7.786463672015503e-06,
      "loss": 0.2081,
      "step": 3637
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6731536926883104,
      "learning_rate": 7.785234046195962e-06,
      "loss": 0.2071,
      "step": 3638
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3118632252239577,
      "learning_rate": 7.784004176092762e-06,
      "loss": 0.204,
      "step": 3639
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2961981270423857,
      "learning_rate": 7.782774061813771e-06,
      "loss": 0.1668,
      "step": 3640
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4251058484521462,
      "learning_rate": 7.781543703466881e-06,
      "loss": 0.2217,
      "step": 3641
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.30576599339657,
      "learning_rate": 7.78031310116e-06,
      "loss": 0.6551,
      "step": 3642
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3334814671521353,
      "learning_rate": 7.779082255001058e-06,
      "loss": 0.1787,
      "step": 3643
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3317038242004735,
      "learning_rate": 7.777851165098012e-06,
      "loss": 0.1612,
      "step": 3644
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.480071323926619,
      "learning_rate": 7.776619831558836e-06,
      "loss": 0.23,
      "step": 3645
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2775874262217366,
      "learning_rate": 7.775388254491526e-06,
      "loss": 0.1695,
      "step": 3646
    },
    {
      "epoch": 0.33,
      "grad_norm": 7.661638656755845,
      "learning_rate": 7.774156434004099e-06,
      "loss": 0.5157,
      "step": 3647
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5642464210380098,
      "learning_rate": 7.772924370204593e-06,
      "loss": 0.2722,
      "step": 3648
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1096456811168656,
      "learning_rate": 7.771692063201072e-06,
      "loss": 0.1748,
      "step": 3649
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.396082943441589,
      "learning_rate": 7.770459513101613e-06,
      "loss": 0.1878,
      "step": 3650
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5770948748930849,
      "learning_rate": 7.769226720014322e-06,
      "loss": 0.2363,
      "step": 3651
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5023547011655947,
      "learning_rate": 7.767993684047323e-06,
      "loss": 0.2086,
      "step": 3652
    },
    {
      "epoch": 0.33,
      "grad_norm": 6.517732990387394,
      "learning_rate": 7.766760405308759e-06,
      "loss": 0.687,
      "step": 3653
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2970926947554118,
      "learning_rate": 7.7655268839068e-06,
      "loss": 0.198,
      "step": 3654
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3807055965397368,
      "learning_rate": 7.764293119949634e-06,
      "loss": 0.1882,
      "step": 3655
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2438573058032376,
      "learning_rate": 7.763059113545466e-06,
      "loss": 0.1491,
      "step": 3656
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4589157717887153,
      "learning_rate": 7.76182486480253e-06,
      "loss": 0.2419,
      "step": 3657
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.378835214777887,
      "learning_rate": 7.760590373829078e-06,
      "loss": 0.2315,
      "step": 3658
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3928648151182743,
      "learning_rate": 7.759355640733381e-06,
      "loss": 0.217,
      "step": 3659
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3630047909105845,
      "learning_rate": 7.75812066562374e-06,
      "loss": 0.2103,
      "step": 3660
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.320233345037559,
      "learning_rate": 7.75688544860846e-06,
      "loss": 0.2094,
      "step": 3661
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1686845565435957,
      "learning_rate": 7.755649989795883e-06,
      "loss": 0.1316,
      "step": 3662
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4783303049300605,
      "learning_rate": 7.75441428929437e-06,
      "loss": 0.2715,
      "step": 3663
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.464231177177223,
      "learning_rate": 7.753178347212299e-06,
      "loss": 0.2401,
      "step": 3664
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.545274150899674,
      "learning_rate": 7.751942163658066e-06,
      "loss": 0.1926,
      "step": 3665
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3639465537916318,
      "learning_rate": 7.750705738740095e-06,
      "loss": 0.2236,
      "step": 3666
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2835866879425495,
      "learning_rate": 7.749469072566832e-06,
      "loss": 0.2138,
      "step": 3667
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3483886908239848,
      "learning_rate": 7.748232165246735e-06,
      "loss": 0.2054,
      "step": 3668
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2318050685117683,
      "learning_rate": 7.746995016888294e-06,
      "loss": 0.1938,
      "step": 3669
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3562716874373635,
      "learning_rate": 7.745757627600013e-06,
      "loss": 0.1699,
      "step": 3670
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.448707908268264,
      "learning_rate": 7.744519997490419e-06,
      "loss": 0.5504,
      "step": 3671
    },
    {
      "epoch": 0.34,
      "grad_norm": 13.069765057589299,
      "learning_rate": 7.743282126668061e-06,
      "loss": 0.7101,
      "step": 3672
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4480840072718761,
      "learning_rate": 7.742044015241508e-06,
      "loss": 0.2448,
      "step": 3673
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.398078829150046,
      "learning_rate": 7.740805663319352e-06,
      "loss": 0.1764,
      "step": 3674
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4361278794157197,
      "learning_rate": 7.739567071010205e-06,
      "loss": 0.2273,
      "step": 3675
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3357358883112873,
      "learning_rate": 7.738328238422698e-06,
      "loss": 0.1914,
      "step": 3676
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.335829393195378,
      "learning_rate": 7.737089165665486e-06,
      "loss": 0.1831,
      "step": 3677
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.3469185499783904,
      "learning_rate": 7.735849852847244e-06,
      "loss": 0.4779,
      "step": 3678
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.368180914668152,
      "learning_rate": 7.734610300076666e-06,
      "loss": 0.2481,
      "step": 3679
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.412127480431939,
      "learning_rate": 7.733370507462473e-06,
      "loss": 0.1767,
      "step": 3680
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4744973268164028,
      "learning_rate": 7.7321304751134e-06,
      "loss": 0.1835,
      "step": 3681
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2006188032362923,
      "learning_rate": 7.730890203138206e-06,
      "loss": 0.1934,
      "step": 3682
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.412750512558883,
      "learning_rate": 7.729649691645673e-06,
      "loss": 0.1979,
      "step": 3683
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2758884549388048,
      "learning_rate": 7.728408940744602e-06,
      "loss": 0.2147,
      "step": 3684
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2747172588582387,
      "learning_rate": 7.727167950543813e-06,
      "loss": 0.1882,
      "step": 3685
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3709515847822513,
      "learning_rate": 7.725926721152153e-06,
      "loss": 0.21,
      "step": 3686
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4225855190141306,
      "learning_rate": 7.724685252678481e-06,
      "loss": 0.2414,
      "step": 3687
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6046247527997128,
      "learning_rate": 7.723443545231687e-06,
      "loss": 0.2049,
      "step": 3688
    },
    {
      "epoch": 0.34,
      "grad_norm": 8.207001915079571,
      "learning_rate": 7.722201598920673e-06,
      "loss": 0.5176,
      "step": 3689
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2126715425895227,
      "learning_rate": 7.72095941385437e-06,
      "loss": 0.1715,
      "step": 3690
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3142797674042668,
      "learning_rate": 7.719716990141724e-06,
      "loss": 0.1824,
      "step": 3691
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.354886202520737,
      "learning_rate": 7.718474327891702e-06,
      "loss": 0.2393,
      "step": 3692
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2939768279430441,
      "learning_rate": 7.717231427213296e-06,
      "loss": 0.2474,
      "step": 3693
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.476725754786754,
      "learning_rate": 7.715988288215518e-06,
      "loss": 0.2287,
      "step": 3694
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2062784073349766,
      "learning_rate": 7.714744911007395e-06,
      "loss": 0.1767,
      "step": 3695
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.288648961986427,
      "learning_rate": 7.713501295697983e-06,
      "loss": 0.1794,
      "step": 3696
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4048028955667702,
      "learning_rate": 7.712257442396355e-06,
      "loss": 0.1934,
      "step": 3697
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3262887833288306,
      "learning_rate": 7.711013351211605e-06,
      "loss": 0.1831,
      "step": 3698
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4820703269364177,
      "learning_rate": 7.709769022252849e-06,
      "loss": 0.2154,
      "step": 3699
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5628372569363953,
      "learning_rate": 7.70852445562922e-06,
      "loss": 0.2436,
      "step": 3700
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4682596827370253,
      "learning_rate": 7.70727965144988e-06,
      "loss": 0.2139,
      "step": 3701
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4636856534098,
      "learning_rate": 7.706034609824e-06,
      "loss": 0.213,
      "step": 3702
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3091231022030647,
      "learning_rate": 7.704789330860784e-06,
      "loss": 0.202,
      "step": 3703
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4556265763895504,
      "learning_rate": 7.703543814669449e-06,
      "loss": 0.1834,
      "step": 3704
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5418958214334886,
      "learning_rate": 7.702298061359236e-06,
      "loss": 0.2462,
      "step": 3705
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.343515371240495,
      "learning_rate": 7.701052071039405e-06,
      "loss": 0.1913,
      "step": 3706
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.288878649240474,
      "learning_rate": 7.699805843819237e-06,
      "loss": 0.1774,
      "step": 3707
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.388937496771091,
      "learning_rate": 7.698559379808034e-06,
      "loss": 0.2011,
      "step": 3708
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.29195264963434,
      "learning_rate": 7.697312679115126e-06,
      "loss": 0.1962,
      "step": 3709
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2802282469362924,
      "learning_rate": 7.696065741849848e-06,
      "loss": 0.2379,
      "step": 3710
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3431278488015084,
      "learning_rate": 7.69481856812157e-06,
      "loss": 0.2269,
      "step": 3711
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4424391126145293,
      "learning_rate": 7.693571158039673e-06,
      "loss": 0.2154,
      "step": 3712
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2794348262380515,
      "learning_rate": 7.692323511713568e-06,
      "loss": 0.1832,
      "step": 3713
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.948362500402499,
      "learning_rate": 7.69107562925268e-06,
      "loss": 0.569,
      "step": 3714
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5094205200396142,
      "learning_rate": 7.689827510766457e-06,
      "loss": 0.2512,
      "step": 3715
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.392760619487376,
      "learning_rate": 7.688579156364366e-06,
      "loss": 0.1985,
      "step": 3716
    },
    {
      "epoch": 0.34,
      "grad_norm": 10.671653737699588,
      "learning_rate": 7.687330566155897e-06,
      "loss": 0.736,
      "step": 3717
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3565987426414658,
      "learning_rate": 7.686081740250562e-06,
      "loss": 0.1776,
      "step": 3718
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4657443566870627,
      "learning_rate": 7.684832678757888e-06,
      "loss": 0.2181,
      "step": 3719
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2798817279381318,
      "learning_rate": 7.683583381787425e-06,
      "loss": 0.2066,
      "step": 3720
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2242593235376003,
      "learning_rate": 7.682333849448749e-06,
      "loss": 0.1762,
      "step": 3721
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3211807108026061,
      "learning_rate": 7.68108408185145e-06,
      "loss": 0.2127,
      "step": 3722
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4812170343087647,
      "learning_rate": 7.679834079105142e-06,
      "loss": 0.2197,
      "step": 3723
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.007292412656627,
      "learning_rate": 7.678583841319456e-06,
      "loss": 0.6047,
      "step": 3724
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3949805897312046,
      "learning_rate": 7.677333368604048e-06,
      "loss": 0.1801,
      "step": 3725
    },
    {
      "epoch": 0.34,
      "grad_norm": 6.811397647045549,
      "learning_rate": 7.676082661068595e-06,
      "loss": 0.7079,
      "step": 3726
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5046221212054685,
      "learning_rate": 7.674831718822789e-06,
      "loss": 0.2548,
      "step": 3727
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.252235880228154,
      "learning_rate": 7.673580541976346e-06,
      "loss": 0.1583,
      "step": 3728
    },
    {
      "epoch": 0.34,
      "grad_norm": 7.506913111175142,
      "learning_rate": 7.672329130639007e-06,
      "loss": 0.6448,
      "step": 3729
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4052618416304992,
      "learning_rate": 7.671077484920523e-06,
      "loss": 0.1986,
      "step": 3730
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.30377631072308,
      "learning_rate": 7.669825604930676e-06,
      "loss": 0.2261,
      "step": 3731
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5341540421023243,
      "learning_rate": 7.668573490779263e-06,
      "loss": 0.2254,
      "step": 3732
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3120633786368094,
      "learning_rate": 7.667321142576104e-06,
      "loss": 0.1667,
      "step": 3733
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2917580907710855,
      "learning_rate": 7.666068560431035e-06,
      "loss": 0.1926,
      "step": 3734
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3659369845836435,
      "learning_rate": 7.664815744453918e-06,
      "loss": 0.193,
      "step": 3735
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.937057418636659,
      "learning_rate": 7.663562694754635e-06,
      "loss": 0.6241,
      "step": 3736
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2268199348084476,
      "learning_rate": 7.662309411443084e-06,
      "loss": 0.1933,
      "step": 3737
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3633063686165852,
      "learning_rate": 7.661055894629186e-06,
      "loss": 0.1916,
      "step": 3738
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4017956131203984,
      "learning_rate": 7.659802144422885e-06,
      "loss": 0.2148,
      "step": 3739
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.583434748708022,
      "learning_rate": 7.658548160934143e-06,
      "loss": 0.2005,
      "step": 3740
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.465062281164179,
      "learning_rate": 7.657293944272944e-06,
      "loss": 0.3759,
      "step": 3741
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4578231298392494,
      "learning_rate": 7.656039494549288e-06,
      "loss": 0.2064,
      "step": 3742
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5409244332037737,
      "learning_rate": 7.654784811873199e-06,
      "loss": 0.2316,
      "step": 3743
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4336637892880153,
      "learning_rate": 7.653529896354724e-06,
      "loss": 0.194,
      "step": 3744
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3372129938236008,
      "learning_rate": 7.652274748103924e-06,
      "loss": 0.1979,
      "step": 3745
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3560199387756355,
      "learning_rate": 7.651019367230886e-06,
      "loss": 0.2198,
      "step": 3746
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5845967036628386,
      "learning_rate": 7.649763753845714e-06,
      "loss": 0.2483,
      "step": 3747
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6295250655555003,
      "learning_rate": 7.648507908058536e-06,
      "loss": 0.2584,
      "step": 3748
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.26000610598188,
      "learning_rate": 7.647251829979495e-06,
      "loss": 0.2024,
      "step": 3749
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.746358504944656,
      "learning_rate": 7.645995519718761e-06,
      "loss": 0.4879,
      "step": 3750
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2520525271690746,
      "learning_rate": 7.644738977386519e-06,
      "loss": 0.1893,
      "step": 3751
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3163151481215039,
      "learning_rate": 7.643482203092974e-06,
      "loss": 0.2092,
      "step": 3752
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4232334003434064,
      "learning_rate": 7.642225196948357e-06,
      "loss": 0.2258,
      "step": 3753
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.8147952371741045,
      "learning_rate": 7.640967959062914e-06,
      "loss": 0.5285,
      "step": 3754
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.348203584939301,
      "learning_rate": 7.639710489546913e-06,
      "loss": 0.2167,
      "step": 3755
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3039174156702489,
      "learning_rate": 7.638452788510645e-06,
      "loss": 0.205,
      "step": 3756
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3403906574177633,
      "learning_rate": 7.637194856064415e-06,
      "loss": 0.2462,
      "step": 3757
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.482024589210738,
      "learning_rate": 7.635936692318554e-06,
      "loss": 0.2108,
      "step": 3758
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5116013758637907,
      "learning_rate": 7.634678297383411e-06,
      "loss": 0.2421,
      "step": 3759
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.425863080457926,
      "learning_rate": 7.633419671369357e-06,
      "loss": 0.2089,
      "step": 3760
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3012298479795832,
      "learning_rate": 7.63216081438678e-06,
      "loss": 0.1964,
      "step": 3761
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.617317082320442,
      "learning_rate": 7.630901726546091e-06,
      "loss": 0.2689,
      "step": 3762
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6259288492156443,
      "learning_rate": 7.62964240795772e-06,
      "loss": 0.2059,
      "step": 3763
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2857481661320003,
      "learning_rate": 7.628382858732119e-06,
      "loss": 0.1679,
      "step": 3764
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3317458462435048,
      "learning_rate": 7.627123078979759e-06,
      "loss": 0.2034,
      "step": 3765
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3037455225205563,
      "learning_rate": 7.625863068811129e-06,
      "loss": 0.1932,
      "step": 3766
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2040953060542023,
      "learning_rate": 7.6246028283367425e-06,
      "loss": 0.176,
      "step": 3767
    },
    {
      "epoch": 0.34,
      "grad_norm": 3.696932865049907,
      "learning_rate": 7.62334235766713e-06,
      "loss": 0.5781,
      "step": 3768
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5321614221088022,
      "learning_rate": 7.622081656912842e-06,
      "loss": 0.2219,
      "step": 3769
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3210141473209749,
      "learning_rate": 7.620820726184452e-06,
      "loss": 0.199,
      "step": 3770
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1976743505415144,
      "learning_rate": 7.619559565592552e-06,
      "loss": 0.1894,
      "step": 3771
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1466516406136706,
      "learning_rate": 7.618298175247754e-06,
      "loss": 0.176,
      "step": 3772
    },
    {
      "epoch": 0.34,
      "grad_norm": 11.536497582979802,
      "learning_rate": 7.617036555260689e-06,
      "loss": 0.5682,
      "step": 3773
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.521162715937966,
      "learning_rate": 7.615774705742012e-06,
      "loss": 0.217,
      "step": 3774
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5646565297713926,
      "learning_rate": 7.614512626802393e-06,
      "loss": 0.2233,
      "step": 3775
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4349151425398596,
      "learning_rate": 7.613250318552527e-06,
      "loss": 0.2003,
      "step": 3776
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4022904505193665,
      "learning_rate": 7.611987781103128e-06,
      "loss": 0.1983,
      "step": 3777
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.47732143345657,
      "learning_rate": 7.610725014564924e-06,
      "loss": 0.2445,
      "step": 3778
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4960907050535,
      "learning_rate": 7.609462019048672e-06,
      "loss": 0.2133,
      "step": 3779
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2038043066112398,
      "learning_rate": 7.608198794665144e-06,
      "loss": 0.1757,
      "step": 3780
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5575350642446295,
      "learning_rate": 7.606935341525134e-06,
      "loss": 0.2288,
      "step": 3781
    },
    {
      "epoch": 0.35,
      "grad_norm": 7.790129088204815,
      "learning_rate": 7.605671659739455e-06,
      "loss": 0.621,
      "step": 3782
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3340792235889332,
      "learning_rate": 7.60440774941894e-06,
      "loss": 0.1476,
      "step": 3783
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2584616370664745,
      "learning_rate": 7.603143610674444e-06,
      "loss": 0.1825,
      "step": 3784
    },
    {
      "epoch": 0.35,
      "grad_norm": 72.90338862271946,
      "learning_rate": 7.601879243616838e-06,
      "loss": 0.7327,
      "step": 3785
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4074642425268087,
      "learning_rate": 7.600614648357016e-06,
      "loss": 0.2228,
      "step": 3786
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4634367796630694,
      "learning_rate": 7.599349825005892e-06,
      "loss": 0.1918,
      "step": 3787
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5348656978839021,
      "learning_rate": 7.598084773674399e-06,
      "loss": 0.2651,
      "step": 3788
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3807852598798551,
      "learning_rate": 7.596819494473491e-06,
      "loss": 0.1812,
      "step": 3789
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4226078554123478,
      "learning_rate": 7.59555398751414e-06,
      "loss": 0.1863,
      "step": 3790
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.366517906864069,
      "learning_rate": 7.594288252907342e-06,
      "loss": 0.2326,
      "step": 3791
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3171924841806222,
      "learning_rate": 7.593022290764109e-06,
      "loss": 0.1761,
      "step": 3792
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1760016041413983,
      "learning_rate": 7.5917561011954755e-06,
      "loss": 0.1336,
      "step": 3793
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4417539733999982,
      "learning_rate": 7.590489684312492e-06,
      "loss": 0.2271,
      "step": 3794
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4373533586549663,
      "learning_rate": 7.589223040226234e-06,
      "loss": 0.2351,
      "step": 3795
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.523819200879201,
      "learning_rate": 7.587956169047794e-06,
      "loss": 0.4766,
      "step": 3796
    },
    {
      "epoch": 0.35,
      "grad_norm": 4.8863480846842595,
      "learning_rate": 7.586689070888284e-06,
      "loss": 0.422,
      "step": 3797
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2428938426576266,
      "learning_rate": 7.5854217458588396e-06,
      "loss": 0.1573,
      "step": 3798
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.495859469544707,
      "learning_rate": 7.584154194070612e-06,
      "loss": 0.1909,
      "step": 3799
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3120750074563237,
      "learning_rate": 7.5828864156347735e-06,
      "loss": 0.2073,
      "step": 3800
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3424735242647539,
      "learning_rate": 7.581618410662519e-06,
      "loss": 0.2261,
      "step": 3801
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3346128550858873,
      "learning_rate": 7.580350179265059e-06,
      "loss": 0.2239,
      "step": 3802
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5000946484255357,
      "learning_rate": 7.579081721553626e-06,
      "loss": 0.2047,
      "step": 3803
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4182744114396264,
      "learning_rate": 7.577813037639472e-06,
      "loss": 0.21,
      "step": 3804
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4293335659660218,
      "learning_rate": 7.576544127633871e-06,
      "loss": 0.195,
      "step": 3805
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.423707409502982,
      "learning_rate": 7.575274991648113e-06,
      "loss": 0.2637,
      "step": 3806
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.600571587464459,
      "learning_rate": 7.57400562979351e-06,
      "loss": 0.4757,
      "step": 3807
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1676939859261575,
      "learning_rate": 7.572736042181393e-06,
      "loss": 0.1764,
      "step": 3808
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2836745996822243,
      "learning_rate": 7.571466228923115e-06,
      "loss": 0.1972,
      "step": 3809
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2516296677617498,
      "learning_rate": 7.570196190130048e-06,
      "loss": 0.2063,
      "step": 3810
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2186185746405718,
      "learning_rate": 7.56892592591358e-06,
      "loss": 0.1836,
      "step": 3811
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4581923545714879,
      "learning_rate": 7.567655436385121e-06,
      "loss": 0.2381,
      "step": 3812
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3393256863877743,
      "learning_rate": 7.566384721656103e-06,
      "loss": 0.2188,
      "step": 3813
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3652830644462175,
      "learning_rate": 7.565113781837979e-06,
      "loss": 0.2251,
      "step": 3814
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.262432760893215,
      "learning_rate": 7.563842617042214e-06,
      "loss": 0.1667,
      "step": 3815
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4110634574289693,
      "learning_rate": 7.562571227380302e-06,
      "loss": 0.1944,
      "step": 3816
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2569239063921551,
      "learning_rate": 7.56129961296375e-06,
      "loss": 0.1715,
      "step": 3817
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4269515836052535,
      "learning_rate": 7.560027773904086e-06,
      "loss": 0.1972,
      "step": 3818
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4630052755928138,
      "learning_rate": 7.558755710312862e-06,
      "loss": 0.1916,
      "step": 3819
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3328601542127583,
      "learning_rate": 7.557483422301645e-06,
      "loss": 0.2109,
      "step": 3820
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2217469501438452,
      "learning_rate": 7.5562109099820215e-06,
      "loss": 0.1684,
      "step": 3821
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6514811170261448,
      "learning_rate": 7.5549381734656e-06,
      "loss": 0.177,
      "step": 3822
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2139286581029423,
      "learning_rate": 7.553665212864011e-06,
      "loss": 0.1867,
      "step": 3823
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.145738698227842,
      "learning_rate": 7.552392028288898e-06,
      "loss": 0.5735,
      "step": 3824
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.393044652720146,
      "learning_rate": 7.551118619851929e-06,
      "loss": 0.6481,
      "step": 3825
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.339395935071511,
      "learning_rate": 7.549844987664792e-06,
      "loss": 0.2225,
      "step": 3826
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5569840813214169,
      "learning_rate": 7.5485711318391916e-06,
      "loss": 0.2043,
      "step": 3827
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.46375730256853,
      "learning_rate": 7.547297052486853e-06,
      "loss": 0.2618,
      "step": 3828
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.309375358705764,
      "learning_rate": 7.546022749719523e-06,
      "loss": 0.2396,
      "step": 3829
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0712538072858449,
      "learning_rate": 7.544748223648964e-06,
      "loss": 0.1253,
      "step": 3830
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4027695553313355,
      "learning_rate": 7.543473474386962e-06,
      "loss": 0.2037,
      "step": 3831
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1639251982454264,
      "learning_rate": 7.5421985020453224e-06,
      "loss": 0.1857,
      "step": 3832
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3290275187909173,
      "learning_rate": 7.540923306735868e-06,
      "loss": 0.2146,
      "step": 3833
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3424211613360224,
      "learning_rate": 7.539647888570439e-06,
      "loss": 0.2164,
      "step": 3834
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.315903228736228,
      "learning_rate": 7.538372247660902e-06,
      "loss": 0.2043,
      "step": 3835
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.365060789439123,
      "learning_rate": 7.537096384119138e-06,
      "loss": 0.1937,
      "step": 3836
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.289260667591457,
      "learning_rate": 7.535820298057048e-06,
      "loss": 0.2283,
      "step": 3837
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4883472718175346,
      "learning_rate": 7.534543989586556e-06,
      "loss": 0.2236,
      "step": 3838
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2918707581766424,
      "learning_rate": 7.533267458819597e-06,
      "loss": 0.2083,
      "step": 3839
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3754687545052848,
      "learning_rate": 7.531990705868137e-06,
      "loss": 0.1821,
      "step": 3840
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7428611774858271,
      "learning_rate": 7.530713730844153e-06,
      "loss": 0.2292,
      "step": 3841
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3649121507658017,
      "learning_rate": 7.529436533859645e-06,
      "loss": 0.2129,
      "step": 3842
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3219782601069336,
      "learning_rate": 7.528159115026632e-06,
      "loss": 0.2465,
      "step": 3843
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4937112677203939,
      "learning_rate": 7.526881474457152e-06,
      "loss": 0.1975,
      "step": 3844
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4506565588992268,
      "learning_rate": 7.525603612263263e-06,
      "loss": 0.2318,
      "step": 3845
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.388618235152554,
      "learning_rate": 7.5243255285570414e-06,
      "loss": 0.2041,
      "step": 3846
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3832866471569694,
      "learning_rate": 7.523047223450582e-06,
      "loss": 0.1714,
      "step": 3847
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4820216929575953,
      "learning_rate": 7.521768697056004e-06,
      "loss": 0.2256,
      "step": 3848
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.217897520959832,
      "learning_rate": 7.5204899494854415e-06,
      "loss": 0.2002,
      "step": 3849
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3514685488664535,
      "learning_rate": 7.519210980851048e-06,
      "loss": 0.2143,
      "step": 3850
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.383277025962602,
      "learning_rate": 7.517931791264999e-06,
      "loss": 0.2275,
      "step": 3851
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4489318325627245,
      "learning_rate": 7.516652380839488e-06,
      "loss": 0.2158,
      "step": 3852
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3998169670678346,
      "learning_rate": 7.515372749686727e-06,
      "loss": 0.2317,
      "step": 3853
    },
    {
      "epoch": 0.35,
      "grad_norm": 11.430718730533343,
      "learning_rate": 7.514092897918948e-06,
      "loss": 0.4665,
      "step": 3854
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3490593826415322,
      "learning_rate": 7.512812825648404e-06,
      "loss": 0.1962,
      "step": 3855
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.606669793894019,
      "learning_rate": 7.511532532987366e-06,
      "loss": 0.2363,
      "step": 3856
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5931247439654266,
      "learning_rate": 7.510252020048121e-06,
      "loss": 0.209,
      "step": 3857
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4742800342213576,
      "learning_rate": 7.508971286942983e-06,
      "loss": 0.2416,
      "step": 3858
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4282540280667873,
      "learning_rate": 7.507690333784277e-06,
      "loss": 0.2055,
      "step": 3859
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5723105512422828,
      "learning_rate": 7.5064091606843535e-06,
      "loss": 0.2193,
      "step": 3860
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5082096827184457,
      "learning_rate": 7.5051277677555805e-06,
      "loss": 0.2565,
      "step": 3861
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.373668885875216,
      "learning_rate": 7.503846155110343e-06,
      "loss": 0.2436,
      "step": 3862
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3664958553403987,
      "learning_rate": 7.502564322861047e-06,
      "loss": 0.2447,
      "step": 3863
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3810953452891919,
      "learning_rate": 7.501282271120118e-06,
      "loss": 0.2004,
      "step": 3864
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.372898072781204,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.2132,
      "step": 3865
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.224791139880947,
      "learning_rate": 7.49871750961316e-06,
      "loss": 0.1846,
      "step": 3866
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.236053942524455,
      "learning_rate": 7.497434800072078e-06,
      "loss": 0.1728,
      "step": 3867
    },
    {
      "epoch": 0.35,
      "grad_norm": 5.303962789491088,
      "learning_rate": 7.4961518714892555e-06,
      "loss": 0.6001,
      "step": 3868
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1831325158732777,
      "learning_rate": 7.494868723977215e-06,
      "loss": 0.1546,
      "step": 3869
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3817173263131322,
      "learning_rate": 7.493585357648501e-06,
      "loss": 0.1617,
      "step": 3870
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2497145078099354,
      "learning_rate": 7.492301772615666e-06,
      "loss": 0.192,
      "step": 3871
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3467813330512928,
      "learning_rate": 7.491017968991294e-06,
      "loss": 0.2127,
      "step": 3872
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.317372038780547,
      "learning_rate": 7.489733946887982e-06,
      "loss": 0.5969,
      "step": 3873
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3874790885625765,
      "learning_rate": 7.488449706418346e-06,
      "loss": 0.2311,
      "step": 3874
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.376577474216688,
      "learning_rate": 7.487165247695024e-06,
      "loss": 0.2333,
      "step": 3875
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5466821407734939,
      "learning_rate": 7.485880570830671e-06,
      "loss": 0.2234,
      "step": 3876
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2725232624059855,
      "learning_rate": 7.484595675937962e-06,
      "loss": 0.1944,
      "step": 3877
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5568367897519346,
      "learning_rate": 7.483310563129592e-06,
      "loss": 0.2186,
      "step": 3878
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.1111766915136405,
      "learning_rate": 7.482025232518273e-06,
      "loss": 0.594,
      "step": 3879
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3401031117808009,
      "learning_rate": 7.480739684216736e-06,
      "loss": 0.2142,
      "step": 3880
    },
    {
      "epoch": 0.35,
      "grad_norm": 6.958311832811153,
      "learning_rate": 7.479453918337733e-06,
      "loss": 0.6963,
      "step": 3881
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2897915232877781,
      "learning_rate": 7.478167934994035e-06,
      "loss": 0.2248,
      "step": 3882
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.613088294581768,
      "learning_rate": 7.476881734298431e-06,
      "loss": 0.682,
      "step": 3883
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2375631883203124,
      "learning_rate": 7.475595316363728e-06,
      "loss": 0.2287,
      "step": 3884
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.897986352778526,
      "learning_rate": 7.474308681302755e-06,
      "loss": 0.5808,
      "step": 3885
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1382992823197855,
      "learning_rate": 7.47302182922836e-06,
      "loss": 0.1847,
      "step": 3886
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3081138302225783,
      "learning_rate": 7.471734760253404e-06,
      "loss": 0.2178,
      "step": 3887
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5626435550056208,
      "learning_rate": 7.4704474744907765e-06,
      "loss": 0.2764,
      "step": 3888
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3848167838800034,
      "learning_rate": 7.469159972053377e-06,
      "loss": 0.2251,
      "step": 3889
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2586594621289011,
      "learning_rate": 7.467872253054131e-06,
      "loss": 0.1346,
      "step": 3890
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.360518684816263,
      "learning_rate": 7.466584317605978e-06,
      "loss": 0.5582,
      "step": 3891
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.560539389434415,
      "learning_rate": 7.46529616582188e-06,
      "loss": 0.2598,
      "step": 3892
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4143306062580205,
      "learning_rate": 7.4640077978148155e-06,
      "loss": 0.1873,
      "step": 3893
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2007560379376194,
      "learning_rate": 7.462719213697785e-06,
      "loss": 0.1881,
      "step": 3894
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.469203697304166,
      "learning_rate": 7.4614304135838026e-06,
      "loss": 0.2284,
      "step": 3895
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5537064906994176,
      "learning_rate": 7.460141397585907e-06,
      "loss": 0.2446,
      "step": 3896
    },
    {
      "epoch": 0.36,
      "grad_norm": 6.8809207650764375,
      "learning_rate": 7.458852165817153e-06,
      "loss": 0.5813,
      "step": 3897
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2259883641510885,
      "learning_rate": 7.457562718390616e-06,
      "loss": 0.1811,
      "step": 3898
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3803611865436787,
      "learning_rate": 7.4562730554193875e-06,
      "loss": 0.1911,
      "step": 3899
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1631960468220448,
      "learning_rate": 7.454983177016579e-06,
      "loss": 0.1704,
      "step": 3900
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.562321453238662,
      "learning_rate": 7.453693083295325e-06,
      "loss": 0.2313,
      "step": 3901
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5809721184359355,
      "learning_rate": 7.452402774368771e-06,
      "loss": 0.2787,
      "step": 3902
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4799879507624263,
      "learning_rate": 7.45111225035009e-06,
      "loss": 0.2396,
      "step": 3903
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.470837418789056,
      "learning_rate": 7.449821511352465e-06,
      "loss": 0.2247,
      "step": 3904
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1890427412459574,
      "learning_rate": 7.448530557489105e-06,
      "loss": 0.1822,
      "step": 3905
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4662028262645475,
      "learning_rate": 7.4472393888732365e-06,
      "loss": 0.1634,
      "step": 3906
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4769863951460538,
      "learning_rate": 7.445948005618101e-06,
      "loss": 0.2378,
      "step": 3907
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6905069128266283,
      "learning_rate": 7.444656407836963e-06,
      "loss": 0.2245,
      "step": 3908
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.370997048819361,
      "learning_rate": 7.443364595643103e-06,
      "loss": 0.2288,
      "step": 3909
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5395136796015219,
      "learning_rate": 7.442072569149823e-06,
      "loss": 0.2083,
      "step": 3910
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3679577886502556,
      "learning_rate": 7.440780328470443e-06,
      "loss": 0.2458,
      "step": 3911
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3360136413540709,
      "learning_rate": 7.439487873718298e-06,
      "loss": 0.1889,
      "step": 3912
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6023988286639719,
      "learning_rate": 7.438195205006749e-06,
      "loss": 0.222,
      "step": 3913
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.319606858759861,
      "learning_rate": 7.436902322449168e-06,
      "loss": 0.2176,
      "step": 3914
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4672693156448098,
      "learning_rate": 7.435609226158951e-06,
      "loss": 0.2322,
      "step": 3915
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4154805299205881,
      "learning_rate": 7.434315916249513e-06,
      "loss": 0.2485,
      "step": 3916
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2716977071317346,
      "learning_rate": 7.4330223928342814e-06,
      "loss": 0.1754,
      "step": 3917
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4361601759177713,
      "learning_rate": 7.431728656026711e-06,
      "loss": 0.2212,
      "step": 3918
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5396322167498135,
      "learning_rate": 7.43043470594027e-06,
      "loss": 0.2332,
      "step": 3919
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.317691552312896,
      "learning_rate": 7.429140542688447e-06,
      "loss": 0.2186,
      "step": 3920
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3075405795911568,
      "learning_rate": 7.427846166384747e-06,
      "loss": 0.1928,
      "step": 3921
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.135284468559158,
      "learning_rate": 7.4265515771426964e-06,
      "loss": 0.6047,
      "step": 3922
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.370210744677206,
      "learning_rate": 7.425256775075839e-06,
      "loss": 0.2108,
      "step": 3923
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3492476118216583,
      "learning_rate": 7.4239617602977395e-06,
      "loss": 0.1647,
      "step": 3924
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3727965540717417,
      "learning_rate": 7.422666532921977e-06,
      "loss": 0.2052,
      "step": 3925
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.084834194030965,
      "learning_rate": 7.421371093062153e-06,
      "loss": 0.5644,
      "step": 3926
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.474295909732121,
      "learning_rate": 7.420075440831886e-06,
      "loss": 0.256,
      "step": 3927
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4786744472630715,
      "learning_rate": 7.418779576344812e-06,
      "loss": 0.2322,
      "step": 3928
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.341864639460881,
      "learning_rate": 7.417483499714589e-06,
      "loss": 0.214,
      "step": 3929
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4293930969886168,
      "learning_rate": 7.416187211054891e-06,
      "loss": 0.1856,
      "step": 3930
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2278289915784528,
      "learning_rate": 7.41489071047941e-06,
      "loss": 0.1855,
      "step": 3931
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3589637286296061,
      "learning_rate": 7.4135939981018575e-06,
      "loss": 0.223,
      "step": 3932
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2981689887935817,
      "learning_rate": 7.412297074035968e-06,
      "loss": 0.2199,
      "step": 3933
    },
    {
      "epoch": 0.36,
      "grad_norm": 8.012587681575935,
      "learning_rate": 7.410999938395484e-06,
      "loss": 0.7536,
      "step": 3934
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4677571045135498,
      "learning_rate": 7.409702591294179e-06,
      "loss": 0.1998,
      "step": 3935
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5048839181255713,
      "learning_rate": 7.408405032845836e-06,
      "loss": 0.2193,
      "step": 3936
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3895638948036928,
      "learning_rate": 7.40710726316426e-06,
      "loss": 0.2208,
      "step": 3937
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3257422367298635,
      "learning_rate": 7.4058092823632745e-06,
      "loss": 0.1969,
      "step": 3938
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2872478507040788,
      "learning_rate": 7.4045110905567185e-06,
      "loss": 0.1553,
      "step": 3939
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3882843557088014,
      "learning_rate": 7.4032126878584565e-06,
      "loss": 0.2139,
      "step": 3940
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4299668337525824,
      "learning_rate": 7.401914074382364e-06,
      "loss": 0.1978,
      "step": 3941
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2341805355518833,
      "learning_rate": 7.400615250242338e-06,
      "loss": 0.1796,
      "step": 3942
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4007157554092784,
      "learning_rate": 7.399316215552296e-06,
      "loss": 0.1914,
      "step": 3943
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5675958889930681,
      "learning_rate": 7.398016970426171e-06,
      "loss": 0.2298,
      "step": 3944
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.410030991607614,
      "learning_rate": 7.396717514977916e-06,
      "loss": 0.5656,
      "step": 3945
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.545469586719706,
      "learning_rate": 7.395417849321501e-06,
      "loss": 0.2649,
      "step": 3946
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3034816106045555,
      "learning_rate": 7.394117973570915e-06,
      "loss": 0.1912,
      "step": 3947
    },
    {
      "epoch": 0.36,
      "grad_norm": 11.225506284400124,
      "learning_rate": 7.392817887840169e-06,
      "loss": 0.6852,
      "step": 3948
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4104937043205343,
      "learning_rate": 7.3915175922432845e-06,
      "loss": 0.2215,
      "step": 3949
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4351702343305601,
      "learning_rate": 7.390217086894309e-06,
      "loss": 0.1857,
      "step": 3950
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3410115639942424,
      "learning_rate": 7.388916371907307e-06,
      "loss": 0.1908,
      "step": 3951
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5707662389477688,
      "learning_rate": 7.387615447396355e-06,
      "loss": 0.2093,
      "step": 3952
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4242895963836255,
      "learning_rate": 7.386314313475557e-06,
      "loss": 0.2285,
      "step": 3953
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3350948401909297,
      "learning_rate": 7.38501297025903e-06,
      "loss": 0.2341,
      "step": 3954
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3874307133000257,
      "learning_rate": 7.38371141786091e-06,
      "loss": 0.2104,
      "step": 3955
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2993790810712658,
      "learning_rate": 7.382409656395353e-06,
      "loss": 0.1579,
      "step": 3956
    },
    {
      "epoch": 0.36,
      "grad_norm": 7.149219959022006,
      "learning_rate": 7.381107685976531e-06,
      "loss": 0.5358,
      "step": 3957
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4471935948297128,
      "learning_rate": 7.379805506718636e-06,
      "loss": 0.227,
      "step": 3958
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4534483797155129,
      "learning_rate": 7.378503118735877e-06,
      "loss": 0.2161,
      "step": 3959
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1663771537548095,
      "learning_rate": 7.377200522142482e-06,
      "loss": 0.1681,
      "step": 3960
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4027403360910646,
      "learning_rate": 7.3758977170527e-06,
      "loss": 0.2336,
      "step": 3961
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5064973041855574,
      "learning_rate": 7.374594703580794e-06,
      "loss": 0.2309,
      "step": 3962
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.405802965500346,
      "learning_rate": 7.373291481841047e-06,
      "loss": 0.2413,
      "step": 3963
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4160029227804405,
      "learning_rate": 7.37198805194776e-06,
      "loss": 0.2047,
      "step": 3964
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4151711799387021,
      "learning_rate": 7.370684414015253e-06,
      "loss": 0.2246,
      "step": 3965
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.415350076088018,
      "learning_rate": 7.369380568157863e-06,
      "loss": 0.218,
      "step": 3966
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.237707925958351,
      "learning_rate": 7.368076514489947e-06,
      "loss": 0.1605,
      "step": 3967
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2560540068528703,
      "learning_rate": 7.366772253125879e-06,
      "loss": 0.2045,
      "step": 3968
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3779649790408548,
      "learning_rate": 7.365467784180051e-06,
      "loss": 0.1937,
      "step": 3969
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3798255448795678,
      "learning_rate": 7.364163107766874e-06,
      "loss": 0.201,
      "step": 3970
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4550995366971107,
      "learning_rate": 7.362858224000776e-06,
      "loss": 0.1942,
      "step": 3971
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3389973841715184,
      "learning_rate": 7.361553132996206e-06,
      "loss": 0.1832,
      "step": 3972
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3392024180882824,
      "learning_rate": 7.3602478348676285e-06,
      "loss": 0.1923,
      "step": 3973
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5590819650414833,
      "learning_rate": 7.358942329729526e-06,
      "loss": 0.2579,
      "step": 3974
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4702688515682227,
      "learning_rate": 7.357636617696399e-06,
      "loss": 0.2367,
      "step": 3975
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.403911310475652,
      "learning_rate": 7.35633069888277e-06,
      "loss": 0.1914,
      "step": 3976
    },
    {
      "epoch": 0.36,
      "grad_norm": 12.958640557536496,
      "learning_rate": 7.355024573403174e-06,
      "loss": 0.668,
      "step": 3977
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2574633064493506,
      "learning_rate": 7.353718241372169e-06,
      "loss": 0.2116,
      "step": 3978
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3861752569072632,
      "learning_rate": 7.352411702904329e-06,
      "loss": 0.2257,
      "step": 3979
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2832027068291416,
      "learning_rate": 7.351104958114247e-06,
      "loss": 0.2053,
      "step": 3980
    },
    {
      "epoch": 0.36,
      "grad_norm": 5.784365949955208,
      "learning_rate": 7.34979800711653e-06,
      "loss": 0.5746,
      "step": 3981
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4243776683463227,
      "learning_rate": 7.348490850025808e-06,
      "loss": 0.2362,
      "step": 3982
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3884237181933827,
      "learning_rate": 7.347183486956728e-06,
      "loss": 0.233,
      "step": 3983
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3806027735012745,
      "learning_rate": 7.345875918023954e-06,
      "loss": 0.1979,
      "step": 3984
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4243663013009087,
      "learning_rate": 7.3445681433421675e-06,
      "loss": 0.2243,
      "step": 3985
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2743727967366913,
      "learning_rate": 7.343260163026073e-06,
      "loss": 0.1764,
      "step": 3986
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.390587254631357,
      "learning_rate": 7.341951977190385e-06,
      "loss": 0.1712,
      "step": 3987
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5796020351021853,
      "learning_rate": 7.34064358594984e-06,
      "loss": 0.2408,
      "step": 3988
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.420720808547284,
      "learning_rate": 7.3393349894191966e-06,
      "loss": 0.2399,
      "step": 3989
    },
    {
      "epoch": 0.36,
      "grad_norm": 4.385332387383504,
      "learning_rate": 7.338026187713224e-06,
      "loss": 0.5974,
      "step": 3990
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.325758701340424,
      "learning_rate": 7.336717180946714e-06,
      "loss": 0.188,
      "step": 3991
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4427696270472892,
      "learning_rate": 7.3354079692344745e-06,
      "loss": 0.262,
      "step": 3992
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2167484277272427,
      "learning_rate": 7.3340985526913335e-06,
      "loss": 0.2123,
      "step": 3993
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4841022977069274,
      "learning_rate": 7.332788931432134e-06,
      "loss": 0.2419,
      "step": 3994
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.529158047532292,
      "learning_rate": 7.33147910557174e-06,
      "loss": 0.2403,
      "step": 3995
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4910260940793572,
      "learning_rate": 7.330169075225032e-06,
      "loss": 0.1702,
      "step": 3996
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.291444421777299,
      "learning_rate": 7.328858840506909e-06,
      "loss": 0.2251,
      "step": 3997
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5586065013043506,
      "learning_rate": 7.327548401532284e-06,
      "loss": 0.2542,
      "step": 3998
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3578227408176917,
      "learning_rate": 7.326237758416094e-06,
      "loss": 0.199,
      "step": 3999
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3554966021610986,
      "learning_rate": 7.32492691127329e-06,
      "loss": 0.1958,
      "step": 4000
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4856709718007486,
      "learning_rate": 7.323615860218844e-06,
      "loss": 0.2138,
      "step": 4001
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5167558738659173,
      "learning_rate": 7.322304605367742e-06,
      "loss": 0.1731,
      "step": 4002
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3593008181236723,
      "learning_rate": 7.320993146834989e-06,
      "loss": 0.2042,
      "step": 4003
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.224951469394474,
      "learning_rate": 7.319681484735612e-06,
      "loss": 0.1895,
      "step": 4004
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3453379003172417,
      "learning_rate": 7.318369619184651e-06,
      "loss": 0.1934,
      "step": 4005
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3864836436246868,
      "learning_rate": 7.317057550297164e-06,
      "loss": 0.2041,
      "step": 4006
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3580189784779029,
      "learning_rate": 7.31574527818823e-06,
      "loss": 0.1787,
      "step": 4007
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3836027909000579,
      "learning_rate": 7.314432802972943e-06,
      "loss": 0.2561,
      "step": 4008
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7082964959742006,
      "learning_rate": 7.313120124766417e-06,
      "loss": 0.3164,
      "step": 4009
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3964793190166547,
      "learning_rate": 7.311807243683779e-06,
      "loss": 0.2779,
      "step": 4010
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3542244360562465,
      "learning_rate": 7.310494159840182e-06,
      "loss": 0.2011,
      "step": 4011
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6254709333583643,
      "learning_rate": 7.309180873350792e-06,
      "loss": 0.2429,
      "step": 4012
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5679562728615166,
      "learning_rate": 7.30786738433079e-06,
      "loss": 0.2271,
      "step": 4013
    },
    {
      "epoch": 0.37,
      "grad_norm": 10.015354791218066,
      "learning_rate": 7.30655369289538e-06,
      "loss": 0.6731,
      "step": 4014
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5144220213082813,
      "learning_rate": 7.305239799159781e-06,
      "loss": 0.2196,
      "step": 4015
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.390020504909237,
      "learning_rate": 7.3039257032392294e-06,
      "loss": 0.1947,
      "step": 4016
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3686914168005808,
      "learning_rate": 7.30261140524898e-06,
      "loss": 0.1683,
      "step": 4017
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5082297017432622,
      "learning_rate": 7.301296905304309e-06,
      "loss": 0.1837,
      "step": 4018
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3512960803798861,
      "learning_rate": 7.299982203520503e-06,
      "loss": 0.2241,
      "step": 4019
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3462291288287387,
      "learning_rate": 7.29866730001287e-06,
      "loss": 0.1988,
      "step": 4020
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.172556659882297,
      "learning_rate": 7.297352194896738e-06,
      "loss": 0.6818,
      "step": 4021
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.485778047855276,
      "learning_rate": 7.296036888287451e-06,
      "loss": 0.2176,
      "step": 4022
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2823931991330135,
      "learning_rate": 7.294721380300369e-06,
      "loss": 0.1933,
      "step": 4023
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.427689567234733,
      "learning_rate": 7.2934056710508705e-06,
      "loss": 0.2462,
      "step": 4024
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.458272866316415,
      "learning_rate": 7.292089760654352e-06,
      "loss": 0.2228,
      "step": 4025
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4125163562798055,
      "learning_rate": 7.290773649226228e-06,
      "loss": 0.1925,
      "step": 4026
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4371457206784226,
      "learning_rate": 7.2894573368819295e-06,
      "loss": 0.1981,
      "step": 4027
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3695680414169005,
      "learning_rate": 7.288140823736908e-06,
      "loss": 0.2285,
      "step": 4028
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2454715594636623,
      "learning_rate": 7.286824109906627e-06,
      "loss": 0.1878,
      "step": 4029
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3970729091436673,
      "learning_rate": 7.285507195506576e-06,
      "loss": 0.1946,
      "step": 4030
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.458015298700494,
      "learning_rate": 7.284190080652253e-06,
      "loss": 0.1908,
      "step": 4031
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3015735089664002,
      "learning_rate": 7.28287276545918e-06,
      "loss": 0.185,
      "step": 4032
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.208125594076188,
      "learning_rate": 7.281555250042893e-06,
      "loss": 0.7188,
      "step": 4033
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6080721626664072,
      "learning_rate": 7.280237534518948e-06,
      "loss": 0.229,
      "step": 4034
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.500233689016217,
      "learning_rate": 7.2789196190029155e-06,
      "loss": 0.2039,
      "step": 4035
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.308870776078561,
      "learning_rate": 7.2776015036103885e-06,
      "loss": 0.1605,
      "step": 4036
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4338382796784657,
      "learning_rate": 7.2762831884569714e-06,
      "loss": 0.2401,
      "step": 4037
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3704826961179675,
      "learning_rate": 7.274964673658293e-06,
      "loss": 0.2052,
      "step": 4038
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3824561836207705,
      "learning_rate": 7.273645959329993e-06,
      "loss": 0.2437,
      "step": 4039
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2862530347422456,
      "learning_rate": 7.272327045587733e-06,
      "loss": 0.2015,
      "step": 4040
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4251352962317034,
      "learning_rate": 7.271007932547188e-06,
      "loss": 0.228,
      "step": 4041
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3196892656095993,
      "learning_rate": 7.269688620324056e-06,
      "loss": 0.235,
      "step": 4042
    },
    {
      "epoch": 0.37,
      "grad_norm": 8.36769296439412,
      "learning_rate": 7.268369109034048e-06,
      "loss": 0.5401,
      "step": 4043
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6874778307929545,
      "learning_rate": 7.267049398792894e-06,
      "loss": 0.2585,
      "step": 4044
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2515466749502222,
      "learning_rate": 7.265729489716342e-06,
      "loss": 0.2018,
      "step": 4045
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3986341983939972,
      "learning_rate": 7.264409381920157e-06,
      "loss": 0.2204,
      "step": 4046
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2506416402448561,
      "learning_rate": 7.26308907552012e-06,
      "loss": 0.2161,
      "step": 4047
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3679578543997999,
      "learning_rate": 7.261768570632032e-06,
      "loss": 0.2228,
      "step": 4048
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3485275378022252,
      "learning_rate": 7.2604478673717095e-06,
      "loss": 0.1748,
      "step": 4049
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5128395025581935,
      "learning_rate": 7.259126965854988e-06,
      "loss": 0.2138,
      "step": 4050
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2895451043565185,
      "learning_rate": 7.257805866197718e-06,
      "loss": 0.1327,
      "step": 4051
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.086722339431394,
      "learning_rate": 7.256484568515769e-06,
      "loss": 0.1543,
      "step": 4052
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4778935425698092,
      "learning_rate": 7.255163072925028e-06,
      "loss": 0.2277,
      "step": 4053
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3032719179491619,
      "learning_rate": 7.2538413795414e-06,
      "loss": 0.2173,
      "step": 4054
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4008106242819365,
      "learning_rate": 7.252519488480804e-06,
      "loss": 0.1831,
      "step": 4055
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.651741563044316,
      "learning_rate": 7.25119739985918e-06,
      "loss": 0.2108,
      "step": 4056
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2325770016860782,
      "learning_rate": 7.249875113792485e-06,
      "loss": 0.1569,
      "step": 4057
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5909005336920488,
      "learning_rate": 7.2485526303966925e-06,
      "loss": 0.2295,
      "step": 4058
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.413859171415866,
      "learning_rate": 7.24722994978779e-06,
      "loss": 0.7841,
      "step": 4059
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.321617256254489,
      "learning_rate": 7.245907072081787e-06,
      "loss": 0.2281,
      "step": 4060
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5457837832066672,
      "learning_rate": 7.244583997394711e-06,
      "loss": 0.2443,
      "step": 4061
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3198063446023602,
      "learning_rate": 7.2432607258426024e-06,
      "loss": 0.1855,
      "step": 4062
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2446929861903715,
      "learning_rate": 7.24193725754152e-06,
      "loss": 0.1768,
      "step": 4063
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3664731980648974,
      "learning_rate": 7.240613592607544e-06,
      "loss": 0.225,
      "step": 4064
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4864972978644284,
      "learning_rate": 7.239289731156767e-06,
      "loss": 0.2255,
      "step": 4065
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4672401072625307,
      "learning_rate": 7.237965673305298e-06,
      "loss": 0.1987,
      "step": 4066
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3978708476503432,
      "learning_rate": 7.236641419169269e-06,
      "loss": 0.206,
      "step": 4067
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.254557184645988,
      "learning_rate": 7.2353169688648254e-06,
      "loss": 0.1657,
      "step": 4068
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.2603261409663595,
      "learning_rate": 7.2339923225081296e-06,
      "loss": 0.5843,
      "step": 4069
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3317835005017116,
      "learning_rate": 7.2326674802153615e-06,
      "loss": 0.2011,
      "step": 4070
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.4566838409134935,
      "learning_rate": 7.23134244210272e-06,
      "loss": 0.4756,
      "step": 4071
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4004787435264652,
      "learning_rate": 7.23001720828642e-06,
      "loss": 0.1883,
      "step": 4072
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1940854966774368,
      "learning_rate": 7.2286917788826926e-06,
      "loss": 0.1855,
      "step": 4073
    },
    {
      "epoch": 0.37,
      "grad_norm": 7.4567643974606925,
      "learning_rate": 7.227366154007785e-06,
      "loss": 0.4299,
      "step": 4074
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2818990247142739,
      "learning_rate": 7.226040333777966e-06,
      "loss": 0.2081,
      "step": 4075
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3571519017635139,
      "learning_rate": 7.2247143183095195e-06,
      "loss": 0.1863,
      "step": 4076
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3606110029766074,
      "learning_rate": 7.2233881077187425e-06,
      "loss": 0.2311,
      "step": 4077
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3404689227851623,
      "learning_rate": 7.222061702121955e-06,
      "loss": 0.2707,
      "step": 4078
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.88687200008903,
      "learning_rate": 7.220735101635492e-06,
      "loss": 0.6102,
      "step": 4079
    },
    {
      "epoch": 0.37,
      "grad_norm": 4.9768823981060075,
      "learning_rate": 7.219408306375703e-06,
      "loss": 0.5664,
      "step": 4080
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.441754114617436,
      "learning_rate": 7.218081316458959e-06,
      "loss": 0.2457,
      "step": 4081
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4467645411232135,
      "learning_rate": 7.216754132001647e-06,
      "loss": 0.2295,
      "step": 4082
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3108479510551914,
      "learning_rate": 7.215426753120165e-06,
      "loss": 0.2256,
      "step": 4083
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.402262961401674,
      "learning_rate": 7.214099179930938e-06,
      "loss": 0.2205,
      "step": 4084
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7432168793130185,
      "learning_rate": 7.2127714125504e-06,
      "loss": 0.2286,
      "step": 4085
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4183454582277135,
      "learning_rate": 7.211443451095007e-06,
      "loss": 0.1909,
      "step": 4086
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3798517278133837,
      "learning_rate": 7.210115295681228e-06,
      "loss": 0.2381,
      "step": 4087
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3644554592980038,
      "learning_rate": 7.208786946425555e-06,
      "loss": 0.1899,
      "step": 4088
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2621998876325582,
      "learning_rate": 7.207458403444488e-06,
      "loss": 0.1722,
      "step": 4089
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.49137505154637,
      "learning_rate": 7.206129666854553e-06,
      "loss": 0.2797,
      "step": 4090
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.477151550635239,
      "learning_rate": 7.204800736772288e-06,
      "loss": 0.2637,
      "step": 4091
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5250388520278386,
      "learning_rate": 7.2034716133142475e-06,
      "loss": 0.2721,
      "step": 4092
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4169939355701628,
      "learning_rate": 7.202142296597007e-06,
      "loss": 0.1889,
      "step": 4093
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.868701683378816,
      "learning_rate": 7.200812786737154e-06,
      "loss": 0.6651,
      "step": 4094
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2270017668901874,
      "learning_rate": 7.199483083851299e-06,
      "loss": 0.1852,
      "step": 4095
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2927530457312624,
      "learning_rate": 7.198153188056061e-06,
      "loss": 0.1841,
      "step": 4096
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.301987471903623,
      "learning_rate": 7.196823099468084e-06,
      "loss": 0.1846,
      "step": 4097
    },
    {
      "epoch": 0.37,
      "grad_norm": 6.374839745201719,
      "learning_rate": 7.1954928182040265e-06,
      "loss": 0.6303,
      "step": 4098
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.339798777360862,
      "learning_rate": 7.194162344380561e-06,
      "loss": 0.2061,
      "step": 4099
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5030762038899426,
      "learning_rate": 7.192831678114378e-06,
      "loss": 0.2486,
      "step": 4100
    },
    {
      "epoch": 0.37,
      "grad_norm": 5.039177809252851,
      "learning_rate": 7.191500819522188e-06,
      "loss": 0.5059,
      "step": 4101
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4257008924371477,
      "learning_rate": 7.1901697687207164e-06,
      "loss": 0.2132,
      "step": 4102
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1901671317356628,
      "learning_rate": 7.188838525826702e-06,
      "loss": 0.1668,
      "step": 4103
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2815541332693423,
      "learning_rate": 7.187507090956909e-06,
      "loss": 0.2052,
      "step": 4104
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.408121795187174,
      "learning_rate": 7.186175464228109e-06,
      "loss": 0.5525,
      "step": 4105
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5166819425135796,
      "learning_rate": 7.184843645757096e-06,
      "loss": 0.2366,
      "step": 4106
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2415572885213346,
      "learning_rate": 7.18351163566068e-06,
      "loss": 0.2014,
      "step": 4107
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.266551134776023,
      "learning_rate": 7.182179434055685e-06,
      "loss": 0.6,
      "step": 4108
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.461273671366144,
      "learning_rate": 7.180847041058955e-06,
      "loss": 0.2325,
      "step": 4109
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5761423475002732,
      "learning_rate": 7.1795144567873505e-06,
      "loss": 0.256,
      "step": 4110
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.197298729309014,
      "learning_rate": 7.178181681357748e-06,
      "loss": 0.1807,
      "step": 4111
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4523077302339915,
      "learning_rate": 7.176848714887042e-06,
      "loss": 0.207,
      "step": 4112
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3483563397056795,
      "learning_rate": 7.175515557492139e-06,
      "loss": 0.18,
      "step": 4113
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.899757527537496,
      "learning_rate": 7.174182209289968e-06,
      "loss": 0.6086,
      "step": 4114
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4618162018962815,
      "learning_rate": 7.1728486703974735e-06,
      "loss": 0.2573,
      "step": 4115
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4184853279133667,
      "learning_rate": 7.171514940931615e-06,
      "loss": 0.2231,
      "step": 4116
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4017334116351485,
      "learning_rate": 7.170181021009368e-06,
      "loss": 0.2224,
      "step": 4117
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.320755382399439,
      "learning_rate": 7.1688469107477285e-06,
      "loss": 0.1772,
      "step": 4118
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5627360164660502,
      "learning_rate": 7.167512610263706e-06,
      "loss": 0.1789,
      "step": 4119
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3937837035284897,
      "learning_rate": 7.166178119674327e-06,
      "loss": 0.1955,
      "step": 4120
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1222962172810673,
      "learning_rate": 7.1648434390966356e-06,
      "loss": 0.1496,
      "step": 4121
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.066264785951316,
      "learning_rate": 7.163508568647692e-06,
      "loss": 0.6748,
      "step": 4122
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4558187071005226,
      "learning_rate": 7.162173508444574e-06,
      "loss": 0.2184,
      "step": 4123
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1961099959189903,
      "learning_rate": 7.160838258604375e-06,
      "loss": 0.1729,
      "step": 4124
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4881283013277018,
      "learning_rate": 7.159502819244206e-06,
      "loss": 0.1701,
      "step": 4125
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1368530119846596,
      "learning_rate": 7.158167190481192e-06,
      "loss": 0.1734,
      "step": 4126
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.415301399318853,
      "learning_rate": 7.156831372432478e-06,
      "loss": 0.2803,
      "step": 4127
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4747690815511916,
      "learning_rate": 7.155495365215224e-06,
      "loss": 0.2019,
      "step": 4128
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5133227180073208,
      "learning_rate": 7.154159168946607e-06,
      "loss": 0.2204,
      "step": 4129
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5548587815113777,
      "learning_rate": 7.15282278374382e-06,
      "loss": 0.2172,
      "step": 4130
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.421686856608464,
      "learning_rate": 7.151486209724072e-06,
      "loss": 0.225,
      "step": 4131
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2767192707833137,
      "learning_rate": 7.150149447004592e-06,
      "loss": 0.1946,
      "step": 4132
    },
    {
      "epoch": 0.38,
      "grad_norm": 6.357501482091333,
      "learning_rate": 7.148812495702622e-06,
      "loss": 0.6444,
      "step": 4133
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3645585200189294,
      "learning_rate": 7.14747535593542e-06,
      "loss": 0.2689,
      "step": 4134
    },
    {
      "epoch": 0.38,
      "grad_norm": 26.210142454342126,
      "learning_rate": 7.146138027820263e-06,
      "loss": 0.6568,
      "step": 4135
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2971056752108208,
      "learning_rate": 7.144800511474443e-06,
      "loss": 0.2052,
      "step": 4136
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3135946488967356,
      "learning_rate": 7.143462807015271e-06,
      "loss": 0.1777,
      "step": 4137
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2878519598243232,
      "learning_rate": 7.142124914560073e-06,
      "loss": 0.202,
      "step": 4138
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4868477710863683,
      "learning_rate": 7.140786834226188e-06,
      "loss": 0.2423,
      "step": 4139
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3340701061706852,
      "learning_rate": 7.139448566130975e-06,
      "loss": 0.19,
      "step": 4140
    },
    {
      "epoch": 0.38,
      "grad_norm": 8.479392936590765,
      "learning_rate": 7.138110110391813e-06,
      "loss": 0.5529,
      "step": 4141
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.808308356066614,
      "learning_rate": 7.13677146712609e-06,
      "loss": 0.5355,
      "step": 4142
    },
    {
      "epoch": 0.38,
      "grad_norm": 10.088043814903523,
      "learning_rate": 7.135432636451215e-06,
      "loss": 0.7623,
      "step": 4143
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3453745223066047,
      "learning_rate": 7.1340936184846125e-06,
      "loss": 0.1982,
      "step": 4144
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.177567934347355,
      "learning_rate": 7.132754413343721e-06,
      "loss": 0.1701,
      "step": 4145
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2401333475683771,
      "learning_rate": 7.131415021146002e-06,
      "loss": 0.2023,
      "step": 4146
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4239182989759374,
      "learning_rate": 7.130075442008925e-06,
      "loss": 0.277,
      "step": 4147
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4008663697242334,
      "learning_rate": 7.1287356760499845e-06,
      "loss": 0.215,
      "step": 4148
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5269668963672989,
      "learning_rate": 7.127395723386683e-06,
      "loss": 0.2462,
      "step": 4149
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.476397201858881,
      "learning_rate": 7.126055584136546e-06,
      "loss": 0.5499,
      "step": 4150
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3930748057218456,
      "learning_rate": 7.124715258417111e-06,
      "loss": 0.2295,
      "step": 4151
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.874295645592017,
      "learning_rate": 7.123374746345932e-06,
      "loss": 0.7522,
      "step": 4152
    },
    {
      "epoch": 0.38,
      "grad_norm": 7.100724784945761,
      "learning_rate": 7.122034048040586e-06,
      "loss": 0.6813,
      "step": 4153
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.246521771319379,
      "learning_rate": 7.120693163618656e-06,
      "loss": 0.152,
      "step": 4154
    },
    {
      "epoch": 0.38,
      "grad_norm": 4.712287951062589,
      "learning_rate": 7.119352093197749e-06,
      "loss": 0.6051,
      "step": 4155
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.352834283561449,
      "learning_rate": 7.118010836895486e-06,
      "loss": 0.2331,
      "step": 4156
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2917822950509381,
      "learning_rate": 7.116669394829503e-06,
      "loss": 0.1722,
      "step": 4157
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5781403500666158,
      "learning_rate": 7.115327767117455e-06,
      "loss": 0.2702,
      "step": 4158
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.303134180118966,
      "learning_rate": 7.1139859538770105e-06,
      "loss": 0.1891,
      "step": 4159
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3650972789938631,
      "learning_rate": 7.112643955225855e-06,
      "loss": 0.2063,
      "step": 4160
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0683145787613029,
      "learning_rate": 7.111301771281692e-06,
      "loss": 0.1436,
      "step": 4161
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3828979765748903,
      "learning_rate": 7.1099594021622394e-06,
      "loss": 0.1938,
      "step": 4162
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2720169339652199,
      "learning_rate": 7.108616847985232e-06,
      "loss": 0.1914,
      "step": 4163
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2826179434853142,
      "learning_rate": 7.107274108868422e-06,
      "loss": 0.1927,
      "step": 4164
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4806405311037296,
      "learning_rate": 7.105931184929574e-06,
      "loss": 0.2273,
      "step": 4165
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.034827314616841,
      "learning_rate": 7.104588076286474e-06,
      "loss": 0.4849,
      "step": 4166
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.35681519388895,
      "learning_rate": 7.103244783056919e-06,
      "loss": 0.6432,
      "step": 4167
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3177674715137677,
      "learning_rate": 7.101901305358729e-06,
      "loss": 0.222,
      "step": 4168
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.429394063709667,
      "learning_rate": 7.100557643309732e-06,
      "loss": 0.2185,
      "step": 4169
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2658139903358354,
      "learning_rate": 7.099213797027777e-06,
      "loss": 0.1862,
      "step": 4170
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3133830453635695,
      "learning_rate": 7.09786976663073e-06,
      "loss": 0.1925,
      "step": 4171
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.261288859737051,
      "learning_rate": 7.096525552236469e-06,
      "loss": 0.2202,
      "step": 4172
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.362481783817376,
      "learning_rate": 7.095181153962891e-06,
      "loss": 0.1856,
      "step": 4173
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3277951457379007,
      "learning_rate": 7.09383657192791e-06,
      "loss": 0.1798,
      "step": 4174
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4389456547136517,
      "learning_rate": 7.092491806249455e-06,
      "loss": 0.2118,
      "step": 4175
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4006411950106952,
      "learning_rate": 7.091146857045472e-06,
      "loss": 0.2344,
      "step": 4176
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.402683394330109,
      "learning_rate": 7.089801724433918e-06,
      "loss": 0.2422,
      "step": 4177
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4652233859158117,
      "learning_rate": 7.088456408532774e-06,
      "loss": 0.2086,
      "step": 4178
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4478191989449718,
      "learning_rate": 7.087110909460031e-06,
      "loss": 0.2432,
      "step": 4179
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3462274649699324,
      "learning_rate": 7.085765227333699e-06,
      "loss": 0.1845,
      "step": 4180
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.293174523073039,
      "learning_rate": 7.084419362271802e-06,
      "loss": 0.1741,
      "step": 4181
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4132246092383687,
      "learning_rate": 7.083073314392383e-06,
      "loss": 0.2367,
      "step": 4182
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3798264895238819,
      "learning_rate": 7.081727083813501e-06,
      "loss": 0.2451,
      "step": 4183
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3720841371810326,
      "learning_rate": 7.080380670653226e-06,
      "loss": 0.2029,
      "step": 4184
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3277040222290486,
      "learning_rate": 7.079034075029651e-06,
      "loss": 0.1737,
      "step": 4185
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.425524282784045,
      "learning_rate": 7.077687297060877e-06,
      "loss": 0.4489,
      "step": 4186
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3244266983533324,
      "learning_rate": 7.076340336865028e-06,
      "loss": 0.199,
      "step": 4187
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5137088515432726,
      "learning_rate": 7.0749931945602425e-06,
      "loss": 0.2292,
      "step": 4188
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2404697121020978,
      "learning_rate": 7.073645870264672e-06,
      "loss": 0.1773,
      "step": 4189
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3779161922838898,
      "learning_rate": 7.072298364096486e-06,
      "loss": 0.1821,
      "step": 4190
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.325315085583261,
      "learning_rate": 7.070950676173871e-06,
      "loss": 0.199,
      "step": 4191
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2605650542883062,
      "learning_rate": 7.069602806615027e-06,
      "loss": 0.2021,
      "step": 4192
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3837359058681342,
      "learning_rate": 7.0682547555381734e-06,
      "loss": 0.2432,
      "step": 4193
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.31897708869819,
      "learning_rate": 7.06690652306154e-06,
      "loss": 0.199,
      "step": 4194
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.5956083857395609,
      "learning_rate": 7.065558109303377e-06,
      "loss": 0.2236,
      "step": 4195
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.355740442340127,
      "learning_rate": 7.06420951438195e-06,
      "loss": 0.1982,
      "step": 4196
    },
    {
      "epoch": 0.38,
      "grad_norm": 13.416421457678467,
      "learning_rate": 7.06286073841554e-06,
      "loss": 0.712,
      "step": 4197
    },
    {
      "epoch": 0.38,
      "grad_norm": 18.640383694100127,
      "learning_rate": 7.0615117815224435e-06,
      "loss": 0.4441,
      "step": 4198
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.427438727363187,
      "learning_rate": 7.060162643820971e-06,
      "loss": 0.2015,
      "step": 4199
    },
    {
      "epoch": 0.38,
      "grad_norm": 5.764207506751616,
      "learning_rate": 7.058813325429454e-06,
      "loss": 0.5648,
      "step": 4200
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3298190252595272,
      "learning_rate": 7.057463826466235e-06,
      "loss": 0.1919,
      "step": 4201
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2318880364643536,
      "learning_rate": 7.056114147049677e-06,
      "loss": 0.1909,
      "step": 4202
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4256746259362831,
      "learning_rate": 7.05476428729815e-06,
      "loss": 0.2204,
      "step": 4203
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3411635070122123,
      "learning_rate": 7.053414247330051e-06,
      "loss": 0.2011,
      "step": 4204
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.623130142370163,
      "learning_rate": 7.052064027263785e-06,
      "loss": 0.2071,
      "step": 4205
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4870505631149133,
      "learning_rate": 7.050713627217775e-06,
      "loss": 0.2177,
      "step": 4206
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3156472752145807,
      "learning_rate": 7.049363047310463e-06,
      "loss": 0.1832,
      "step": 4207
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.309614214838596,
      "learning_rate": 7.0480122876603015e-06,
      "loss": 0.1962,
      "step": 4208
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4462874164667299,
      "learning_rate": 7.0466613483857615e-06,
      "loss": 0.2187,
      "step": 4209
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4742016070299224,
      "learning_rate": 7.0453102296053286e-06,
      "loss": 0.2187,
      "step": 4210
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.243406329168864,
      "learning_rate": 7.043958931437508e-06,
      "loss": 0.1776,
      "step": 4211
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5448614507658547,
      "learning_rate": 7.0426074540008134e-06,
      "loss": 0.2205,
      "step": 4212
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2368815129336583,
      "learning_rate": 7.041255797413782e-06,
      "loss": 0.1737,
      "step": 4213
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3567609035930297,
      "learning_rate": 7.03990396179496e-06,
      "loss": 0.2326,
      "step": 4214
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2597561230323178,
      "learning_rate": 7.038551947262916e-06,
      "loss": 0.1711,
      "step": 4215
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4688588057639098,
      "learning_rate": 7.037199753936228e-06,
      "loss": 0.2134,
      "step": 4216
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2848690939865444,
      "learning_rate": 7.035847381933494e-06,
      "loss": 0.1666,
      "step": 4217
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2462141033716807,
      "learning_rate": 7.034494831373324e-06,
      "loss": 0.1858,
      "step": 4218
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5968008866551169,
      "learning_rate": 7.033142102374349e-06,
      "loss": 0.2335,
      "step": 4219
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5221642991519186,
      "learning_rate": 7.031789195055207e-06,
      "loss": 0.2137,
      "step": 4220
    },
    {
      "epoch": 0.39,
      "grad_norm": 14.57761063006889,
      "learning_rate": 7.030436109534563e-06,
      "loss": 0.4605,
      "step": 4221
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.9935735913490475,
      "learning_rate": 7.029082845931088e-06,
      "loss": 0.772,
      "step": 4222
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3719120825319946,
      "learning_rate": 7.027729404363474e-06,
      "loss": 0.1826,
      "step": 4223
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.9438717444605835,
      "learning_rate": 7.026375784950426e-06,
      "loss": 0.5165,
      "step": 4224
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.316945704349391,
      "learning_rate": 7.025021987810664e-06,
      "loss": 0.1919,
      "step": 4225
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2353134481588055,
      "learning_rate": 7.0236680130629296e-06,
      "loss": 0.1695,
      "step": 4226
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3452598524588877,
      "learning_rate": 7.022313860825971e-06,
      "loss": 0.2033,
      "step": 4227
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4325144667592473,
      "learning_rate": 7.020959531218558e-06,
      "loss": 0.1843,
      "step": 4228
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.606891296423726,
      "learning_rate": 7.019605024359475e-06,
      "loss": 0.5062,
      "step": 4229
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.266339688474492,
      "learning_rate": 7.018250340367521e-06,
      "loss": 0.5923,
      "step": 4230
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.393320873718726,
      "learning_rate": 7.016895479361509e-06,
      "loss": 0.185,
      "step": 4231
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5770026045483505,
      "learning_rate": 7.015540441460272e-06,
      "loss": 0.245,
      "step": 4232
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4296868264562688,
      "learning_rate": 7.014185226782655e-06,
      "loss": 0.2419,
      "step": 4233
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6184187239813246,
      "learning_rate": 7.01282983544752e-06,
      "loss": 0.2305,
      "step": 4234
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4786806406095787,
      "learning_rate": 7.0114742675737425e-06,
      "loss": 0.19,
      "step": 4235
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.3873788084156065,
      "learning_rate": 7.010118523280214e-06,
      "loss": 0.5191,
      "step": 4236
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.409644928273423,
      "learning_rate": 7.008762602685845e-06,
      "loss": 0.203,
      "step": 4237
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4545197877530627,
      "learning_rate": 7.007406505909558e-06,
      "loss": 0.1847,
      "step": 4238
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3793094887954138,
      "learning_rate": 7.006050233070289e-06,
      "loss": 0.1755,
      "step": 4239
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4493380532274343,
      "learning_rate": 7.004693784286998e-06,
      "loss": 0.2782,
      "step": 4240
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3555702296377248,
      "learning_rate": 7.003337159678649e-06,
      "loss": 0.1671,
      "step": 4241
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4850657531915445,
      "learning_rate": 7.00198035936423e-06,
      "loss": 0.1837,
      "step": 4242
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4253756415222012,
      "learning_rate": 7.000623383462741e-06,
      "loss": 0.2192,
      "step": 4243
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5063570552771357,
      "learning_rate": 6.999266232093197e-06,
      "loss": 0.2226,
      "step": 4244
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1708517433045185,
      "learning_rate": 6.99790890537463e-06,
      "loss": 0.1802,
      "step": 4245
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3864614974377292,
      "learning_rate": 6.9965514034260864e-06,
      "loss": 0.2367,
      "step": 4246
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.449575164845682,
      "learning_rate": 6.995193726366628e-06,
      "loss": 0.2034,
      "step": 4247
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3628091629720822,
      "learning_rate": 6.993835874315334e-06,
      "loss": 0.1881,
      "step": 4248
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.6085332211602505,
      "learning_rate": 6.992477847391292e-06,
      "loss": 0.6099,
      "step": 4249
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.390506937068009,
      "learning_rate": 6.991119645713617e-06,
      "loss": 0.2736,
      "step": 4250
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1713229074887683,
      "learning_rate": 6.989761269401427e-06,
      "loss": 0.1689,
      "step": 4251
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3947134214715036,
      "learning_rate": 6.988402718573865e-06,
      "loss": 0.1937,
      "step": 4252
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5601655376998007,
      "learning_rate": 6.987043993350081e-06,
      "loss": 0.2004,
      "step": 4253
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3102158091385498,
      "learning_rate": 6.985685093849246e-06,
      "loss": 0.1757,
      "step": 4254
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7094720660655107,
      "learning_rate": 6.984326020190544e-06,
      "loss": 0.2564,
      "step": 4255
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3717043570001881,
      "learning_rate": 6.982966772493176e-06,
      "loss": 0.1827,
      "step": 4256
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4839567093304162,
      "learning_rate": 6.981607350876357e-06,
      "loss": 0.2088,
      "step": 4257
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4179731503599362,
      "learning_rate": 6.9802477554593165e-06,
      "loss": 0.2044,
      "step": 4258
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3097861899350853,
      "learning_rate": 6.9788879863613e-06,
      "loss": 0.2106,
      "step": 4259
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.278638861622817,
      "learning_rate": 6.977528043701569e-06,
      "loss": 0.1986,
      "step": 4260
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5462567475672333,
      "learning_rate": 6.9761679275994e-06,
      "loss": 0.2217,
      "step": 4261
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.580148122904253,
      "learning_rate": 6.9748076381740836e-06,
      "loss": 0.773,
      "step": 4262
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3973008674398755,
      "learning_rate": 6.973447175544925e-06,
      "loss": 0.1787,
      "step": 4263
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.382888473203345,
      "learning_rate": 6.972086539831249e-06,
      "loss": 0.182,
      "step": 4264
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.414898204728901,
      "learning_rate": 6.970725731152389e-06,
      "loss": 0.1948,
      "step": 4265
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3183144535718974,
      "learning_rate": 6.969364749627699e-06,
      "loss": 0.243,
      "step": 4266
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.910926199976956,
      "learning_rate": 6.9680035953765464e-06,
      "loss": 0.6803,
      "step": 4267
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.462374549102954,
      "learning_rate": 6.966642268518313e-06,
      "loss": 0.2017,
      "step": 4268
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3278777438560583,
      "learning_rate": 6.9652807691723975e-06,
      "loss": 0.2445,
      "step": 4269
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3623878994886547,
      "learning_rate": 6.96391909745821e-06,
      "loss": 0.2246,
      "step": 4270
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4262517498495988,
      "learning_rate": 6.962557253495179e-06,
      "loss": 0.1987,
      "step": 4271
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.380383164176899,
      "learning_rate": 6.961195237402749e-06,
      "loss": 0.2522,
      "step": 4272
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3378716151092125,
      "learning_rate": 6.959833049300376e-06,
      "loss": 0.2111,
      "step": 4273
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.7816349962489735,
      "learning_rate": 6.958470689307534e-06,
      "loss": 0.6236,
      "step": 4274
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4690577723718383,
      "learning_rate": 6.95710815754371e-06,
      "loss": 0.2304,
      "step": 4275
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3386869603332034,
      "learning_rate": 6.955745454128409e-06,
      "loss": 0.2065,
      "step": 4276
    },
    {
      "epoch": 0.39,
      "grad_norm": 6.885067668513142,
      "learning_rate": 6.954382579181149e-06,
      "loss": 0.6582,
      "step": 4277
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.270212055047014,
      "learning_rate": 6.953019532821461e-06,
      "loss": 0.1852,
      "step": 4278
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3902399305444462,
      "learning_rate": 6.951656315168896e-06,
      "loss": 0.1981,
      "step": 4279
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3770504027389776,
      "learning_rate": 6.9502929263430155e-06,
      "loss": 0.2248,
      "step": 4280
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4349319460907737,
      "learning_rate": 6.948929366463397e-06,
      "loss": 0.2153,
      "step": 4281
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.31383557877762,
      "learning_rate": 6.947565635649637e-06,
      "loss": 0.1987,
      "step": 4282
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4164709717130501,
      "learning_rate": 6.94620173402134e-06,
      "loss": 0.1932,
      "step": 4283
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2962680209479098,
      "learning_rate": 6.944837661698134e-06,
      "loss": 0.1758,
      "step": 4284
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.12436698701193,
      "learning_rate": 6.943473418799653e-06,
      "loss": 0.4899,
      "step": 4285
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3967449035975168,
      "learning_rate": 6.942109005445553e-06,
      "loss": 0.2599,
      "step": 4286
    },
    {
      "epoch": 0.39,
      "grad_norm": 8.257331378043657,
      "learning_rate": 6.9407444217555e-06,
      "loss": 0.5397,
      "step": 4287
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4561104593759535,
      "learning_rate": 6.939379667849178e-06,
      "loss": 0.238,
      "step": 4288
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.398788302597858,
      "learning_rate": 6.938014743846285e-06,
      "loss": 0.1849,
      "step": 4289
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.430959088514304,
      "learning_rate": 6.936649649866534e-06,
      "loss": 0.2267,
      "step": 4290
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3251837348025803,
      "learning_rate": 6.935284386029654e-06,
      "loss": 0.1866,
      "step": 4291
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3130809566650887,
      "learning_rate": 6.933918952455385e-06,
      "loss": 0.2487,
      "step": 4292
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3092927510365497,
      "learning_rate": 6.932553349263489e-06,
      "loss": 0.2066,
      "step": 4293
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7742528142496177,
      "learning_rate": 6.931187576573733e-06,
      "loss": 0.2057,
      "step": 4294
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3011831784739907,
      "learning_rate": 6.929821634505911e-06,
      "loss": 0.2174,
      "step": 4295
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1893204044031065,
      "learning_rate": 6.928455523179819e-06,
      "loss": 0.1906,
      "step": 4296
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.247365407648213,
      "learning_rate": 6.927089242715277e-06,
      "loss": 0.1843,
      "step": 4297
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3194033065404402,
      "learning_rate": 6.925722793232117e-06,
      "loss": 0.2587,
      "step": 4298
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3380480215423927,
      "learning_rate": 6.924356174850185e-06,
      "loss": 0.2146,
      "step": 4299
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4039150589280038,
      "learning_rate": 6.922989387689343e-06,
      "loss": 0.1957,
      "step": 4300
    },
    {
      "epoch": 0.39,
      "grad_norm": 7.05829222104441,
      "learning_rate": 6.921622431869468e-06,
      "loss": 0.6599,
      "step": 4301
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3800675474900372,
      "learning_rate": 6.92025530751045e-06,
      "loss": 0.1965,
      "step": 4302
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4010911921809144,
      "learning_rate": 6.918888014732197e-06,
      "loss": 0.2629,
      "step": 4303
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.315908367418908,
      "learning_rate": 6.917520553654627e-06,
      "loss": 0.195,
      "step": 4304
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4515297904954594,
      "learning_rate": 6.916152924397676e-06,
      "loss": 0.2918,
      "step": 4305
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3990660558709542,
      "learning_rate": 6.914785127081295e-06,
      "loss": 0.2369,
      "step": 4306
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2401448582739387,
      "learning_rate": 6.913417161825449e-06,
      "loss": 0.2108,
      "step": 4307
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2425558764410054,
      "learning_rate": 6.9120490287501184e-06,
      "loss": 0.1678,
      "step": 4308
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3243233259102223,
      "learning_rate": 6.9106807279752964e-06,
      "loss": 0.1837,
      "step": 4309
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4442734968399613,
      "learning_rate": 6.909312259620993e-06,
      "loss": 0.2167,
      "step": 4310
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3173070387511103,
      "learning_rate": 6.907943623807231e-06,
      "loss": 0.2376,
      "step": 4311
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4105419848687755,
      "learning_rate": 6.906574820654051e-06,
      "loss": 0.232,
      "step": 4312
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3004668209857413,
      "learning_rate": 6.905205850281502e-06,
      "loss": 0.2011,
      "step": 4313
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2402872812030112,
      "learning_rate": 6.903836712809658e-06,
      "loss": 0.1283,
      "step": 4314
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3699128122027864,
      "learning_rate": 6.9024674083585955e-06,
      "loss": 0.2523,
      "step": 4315
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4446007410905257,
      "learning_rate": 6.901097937048414e-06,
      "loss": 0.2713,
      "step": 4316
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.279099543735031,
      "learning_rate": 6.899728298999226e-06,
      "loss": 0.1765,
      "step": 4317
    },
    {
      "epoch": 0.39,
      "grad_norm": 4.752297623847678,
      "learning_rate": 6.898358494331159e-06,
      "loss": 0.61,
      "step": 4318
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2700114622766723,
      "learning_rate": 6.896988523164351e-06,
      "loss": 0.1906,
      "step": 4319
    },
    {
      "epoch": 0.39,
      "grad_norm": 5.522196154329298,
      "learning_rate": 6.89561838561896e-06,
      "loss": 0.74,
      "step": 4320
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3858765326795204,
      "learning_rate": 6.894248081815155e-06,
      "loss": 0.2284,
      "step": 4321
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3527094870741503,
      "learning_rate": 6.89287761187312e-06,
      "loss": 0.1965,
      "step": 4322
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3421013947767453,
      "learning_rate": 6.891506975913057e-06,
      "loss": 0.2107,
      "step": 4323
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.8900826910109805,
      "learning_rate": 6.890136174055178e-06,
      "loss": 0.6607,
      "step": 4324
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.488473188062189,
      "learning_rate": 6.888765206419712e-06,
      "loss": 0.276,
      "step": 4325
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2368512563272451,
      "learning_rate": 6.887394073126902e-06,
      "loss": 0.1887,
      "step": 4326
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.308310242779351,
      "learning_rate": 6.886022774297006e-06,
      "loss": 0.1999,
      "step": 4327
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4402118395379335,
      "learning_rate": 6.884651310050294e-06,
      "loss": 0.2267,
      "step": 4328
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3293864761709395,
      "learning_rate": 6.883279680507057e-06,
      "loss": 0.1966,
      "step": 4329
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4029207182949857,
      "learning_rate": 6.8819078857875905e-06,
      "loss": 0.1896,
      "step": 4330
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.449528681583464,
      "learning_rate": 6.880535926012215e-06,
      "loss": 0.1895,
      "step": 4331
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5032010120178165,
      "learning_rate": 6.879163801301257e-06,
      "loss": 0.1882,
      "step": 4332
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3905172464067035,
      "learning_rate": 6.877791511775064e-06,
      "loss": 0.1821,
      "step": 4333
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3999984240891523,
      "learning_rate": 6.8764190575539925e-06,
      "loss": 0.1914,
      "step": 4334
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1968335604087363,
      "learning_rate": 6.875046438758417e-06,
      "loss": 0.1717,
      "step": 4335
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1657530113168513,
      "learning_rate": 6.873673655508724e-06,
      "loss": 0.1661,
      "step": 4336
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3923199517951619,
      "learning_rate": 6.872300707925319e-06,
      "loss": 0.1691,
      "step": 4337
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.412670742587497,
      "learning_rate": 6.870927596128617e-06,
      "loss": 0.2174,
      "step": 4338
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.570718411720269,
      "learning_rate": 6.869554320239047e-06,
      "loss": 0.2139,
      "step": 4339
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2834980678966381,
      "learning_rate": 6.8681808803770565e-06,
      "loss": 0.2254,
      "step": 4340
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.451905601779468,
      "learning_rate": 6.8668072766631054e-06,
      "loss": 0.5382,
      "step": 4341
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4333025793318384,
      "learning_rate": 6.865433509217669e-06,
      "loss": 0.2265,
      "step": 4342
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.330385073098159,
      "learning_rate": 6.864059578161233e-06,
      "loss": 0.2288,
      "step": 4343
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1688596503670623,
      "learning_rate": 6.862685483614302e-06,
      "loss": 0.184,
      "step": 4344
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2181036154980611,
      "learning_rate": 6.861311225697392e-06,
      "loss": 0.2084,
      "step": 4345
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5138551310103536,
      "learning_rate": 6.859936804531039e-06,
      "loss": 0.2572,
      "step": 4346
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4082534197295387,
      "learning_rate": 6.858562220235784e-06,
      "loss": 0.2063,
      "step": 4347
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2268559447232836,
      "learning_rate": 6.857187472932188e-06,
      "loss": 0.1873,
      "step": 4348
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4035406546630944,
      "learning_rate": 6.855812562740828e-06,
      "loss": 0.2389,
      "step": 4349
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3427371484410526,
      "learning_rate": 6.85443748978229e-06,
      "loss": 0.203,
      "step": 4350
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.783103573636663,
      "learning_rate": 6.853062254177178e-06,
      "loss": 0.6026,
      "step": 4351
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4138020356367518,
      "learning_rate": 6.851686856046111e-06,
      "loss": 0.2095,
      "step": 4352
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3147098936130317,
      "learning_rate": 6.850311295509719e-06,
      "loss": 0.1897,
      "step": 4353
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3012097881830218,
      "learning_rate": 6.8489355726886484e-06,
      "loss": 0.1587,
      "step": 4354
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4489266859056873,
      "learning_rate": 6.847559687703561e-06,
      "loss": 0.2264,
      "step": 4355
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4109044156689259,
      "learning_rate": 6.846183640675127e-06,
      "loss": 0.2072,
      "step": 4356
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2438407774415223,
      "learning_rate": 6.8448074317240375e-06,
      "loss": 0.1933,
      "step": 4357
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.413620743753725,
      "learning_rate": 6.843431060970995e-06,
      "loss": 0.1852,
      "step": 4358
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4130400570162764,
      "learning_rate": 6.842054528536717e-06,
      "loss": 0.1934,
      "step": 4359
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3050971806808287,
      "learning_rate": 6.840677834541934e-06,
      "loss": 0.1923,
      "step": 4360
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3765130486525317,
      "learning_rate": 6.8393009791073895e-06,
      "loss": 0.1857,
      "step": 4361
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5355414809765116,
      "learning_rate": 6.837923962353845e-06,
      "loss": 0.1881,
      "step": 4362
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4738743221745634,
      "learning_rate": 6.836546784402076e-06,
      "loss": 0.2006,
      "step": 4363
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3769908662335841,
      "learning_rate": 6.835169445372867e-06,
      "loss": 0.2201,
      "step": 4364
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2577177369589074,
      "learning_rate": 6.833791945387019e-06,
      "loss": 0.1538,
      "step": 4365
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4116672911546335,
      "learning_rate": 6.832414284565351e-06,
      "loss": 0.2258,
      "step": 4366
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5702986241143662,
      "learning_rate": 6.831036463028692e-06,
      "loss": 0.2677,
      "step": 4367
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4074879800418967,
      "learning_rate": 6.829658480897885e-06,
      "loss": 0.2655,
      "step": 4368
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3079851886505671,
      "learning_rate": 6.828280338293792e-06,
      "loss": 0.1651,
      "step": 4369
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4787909511217616,
      "learning_rate": 6.82690203533728e-06,
      "loss": 0.2528,
      "step": 4370
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3477477866737553,
      "learning_rate": 6.8255235721492375e-06,
      "loss": 0.2121,
      "step": 4371
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.123804593439188,
      "learning_rate": 6.824144948850567e-06,
      "loss": 0.6163,
      "step": 4372
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.477485578586739,
      "learning_rate": 6.822766165562181e-06,
      "loss": 0.2529,
      "step": 4373
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4619286661358466,
      "learning_rate": 6.821387222405009e-06,
      "loss": 0.2117,
      "step": 4374
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2210121190898782,
      "learning_rate": 6.820008119499992e-06,
      "loss": 0.2069,
      "step": 4375
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5116388428799825,
      "learning_rate": 6.818628856968088e-06,
      "loss": 0.2111,
      "step": 4376
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3865206568319488,
      "learning_rate": 6.817249434930267e-06,
      "loss": 0.2389,
      "step": 4377
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3546986746831864,
      "learning_rate": 6.8158698535075126e-06,
      "loss": 0.1665,
      "step": 4378
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.328391091229481,
      "learning_rate": 6.814490112820825e-06,
      "loss": 0.1738,
      "step": 4379
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.239027788405847,
      "learning_rate": 6.813110212991216e-06,
      "loss": 0.1834,
      "step": 4380
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.409767592601601,
      "learning_rate": 6.811730154139714e-06,
      "loss": 0.2095,
      "step": 4381
    },
    {
      "epoch": 0.4,
      "grad_norm": 10.424884294241506,
      "learning_rate": 6.810349936387355e-06,
      "loss": 0.6365,
      "step": 4382
    },
    {
      "epoch": 0.4,
      "grad_norm": 6.369430036509902,
      "learning_rate": 6.808969559855196e-06,
      "loss": 0.6521,
      "step": 4383
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3054770516464476,
      "learning_rate": 6.807589024664306e-06,
      "loss": 0.1972,
      "step": 4384
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.297815648887617,
      "learning_rate": 6.806208330935766e-06,
      "loss": 0.6064,
      "step": 4385
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.398371496585226,
      "learning_rate": 6.804827478790672e-06,
      "loss": 0.2118,
      "step": 4386
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.103671793938504,
      "learning_rate": 6.803446468350134e-06,
      "loss": 0.4827,
      "step": 4387
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6343015933357803,
      "learning_rate": 6.802065299735276e-06,
      "loss": 0.1872,
      "step": 4388
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2951992060371136,
      "learning_rate": 6.800683973067237e-06,
      "loss": 0.2259,
      "step": 4389
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.334914609578658,
      "learning_rate": 6.799302488467166e-06,
      "loss": 0.1684,
      "step": 4390
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.535632409494203,
      "learning_rate": 6.79792084605623e-06,
      "loss": 0.2109,
      "step": 4391
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3949510375779852,
      "learning_rate": 6.796539045955609e-06,
      "loss": 0.2155,
      "step": 4392
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4952509779839747,
      "learning_rate": 6.7951570882864944e-06,
      "loss": 0.2518,
      "step": 4393
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2536188630111755,
      "learning_rate": 6.793774973170094e-06,
      "loss": 0.1995,
      "step": 4394
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1998810066248318,
      "learning_rate": 6.792392700727629e-06,
      "loss": 0.2116,
      "step": 4395
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5586616131385984,
      "learning_rate": 6.791010271080335e-06,
      "loss": 0.208,
      "step": 4396
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.387700452041078,
      "learning_rate": 6.789627684349457e-06,
      "loss": 0.2081,
      "step": 4397
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4451650713657436,
      "learning_rate": 6.788244940656259e-06,
      "loss": 0.2267,
      "step": 4398
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6496777730508818,
      "learning_rate": 6.786862040122017e-06,
      "loss": 0.2562,
      "step": 4399
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4877551510684461,
      "learning_rate": 6.78547898286802e-06,
      "loss": 0.1798,
      "step": 4400
    },
    {
      "epoch": 0.4,
      "grad_norm": 5.64323053912259,
      "learning_rate": 6.784095769015573e-06,
      "loss": 0.524,
      "step": 4401
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2814480166811857,
      "learning_rate": 6.78271239868599e-06,
      "loss": 0.2138,
      "step": 4402
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5888439013711488,
      "learning_rate": 6.781328872000605e-06,
      "loss": 0.252,
      "step": 4403
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3040456123678676,
      "learning_rate": 6.7799451890807625e-06,
      "loss": 0.1856,
      "step": 4404
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.537153519407418,
      "learning_rate": 6.778561350047818e-06,
      "loss": 0.5212,
      "step": 4405
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2471111237199755,
      "learning_rate": 6.7771773550231445e-06,
      "loss": 0.1982,
      "step": 4406
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.051883139881495,
      "learning_rate": 6.775793204128128e-06,
      "loss": 0.5414,
      "step": 4407
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5614967306910101,
      "learning_rate": 6.7744088974841686e-06,
      "loss": 0.2597,
      "step": 4408
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.849091108622289,
      "learning_rate": 6.773024435212678e-06,
      "loss": 0.5405,
      "step": 4409
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3026682501176963,
      "learning_rate": 6.771639817435084e-06,
      "loss": 0.173,
      "step": 4410
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4293204038850384,
      "learning_rate": 6.770255044272826e-06,
      "loss": 0.2067,
      "step": 4411
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3534276320286938,
      "learning_rate": 6.768870115847358e-06,
      "loss": 0.2047,
      "step": 4412
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3674498680720344,
      "learning_rate": 6.767485032280148e-06,
      "loss": 0.2019,
      "step": 4413
    },
    {
      "epoch": 0.4,
      "grad_norm": 12.817463982652837,
      "learning_rate": 6.766099793692676e-06,
      "loss": 0.6498,
      "step": 4414
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3747515717329553,
      "learning_rate": 6.764714400206438e-06,
      "loss": 0.234,
      "step": 4415
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5691817255674594,
      "learning_rate": 6.763328851942941e-06,
      "loss": 0.2404,
      "step": 4416
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3872021855113315,
      "learning_rate": 6.761943149023706e-06,
      "loss": 0.2028,
      "step": 4417
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3489443192710457,
      "learning_rate": 6.760557291570272e-06,
      "loss": 0.2234,
      "step": 4418
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4645829111672934,
      "learning_rate": 6.759171279704185e-06,
      "loss": 0.2289,
      "step": 4419
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4129930866247784,
      "learning_rate": 6.757785113547008e-06,
      "loss": 0.2077,
      "step": 4420
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3205717538421125,
      "learning_rate": 6.756398793220318e-06,
      "loss": 0.2144,
      "step": 4421
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.444427938147897,
      "learning_rate": 6.755012318845705e-06,
      "loss": 0.2169,
      "step": 4422
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3301660045440014,
      "learning_rate": 6.753625690544769e-06,
      "loss": 0.1923,
      "step": 4423
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4242562276286452,
      "learning_rate": 6.75223890843913e-06,
      "loss": 0.2281,
      "step": 4424
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2713191640995867,
      "learning_rate": 6.750851972650416e-06,
      "loss": 0.1886,
      "step": 4425
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3243952300959347,
      "learning_rate": 6.749464883300272e-06,
      "loss": 0.2093,
      "step": 4426
    },
    {
      "epoch": 0.4,
      "grad_norm": 8.59223688822236,
      "learning_rate": 6.7480776405103535e-06,
      "loss": 0.6634,
      "step": 4427
    },
    {
      "epoch": 0.4,
      "grad_norm": 7.6319371080586125,
      "learning_rate": 6.74669024440233e-06,
      "loss": 0.7322,
      "step": 4428
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2704753725401403,
      "learning_rate": 6.74530269509789e-06,
      "loss": 0.2132,
      "step": 4429
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3850756074702877,
      "learning_rate": 6.743914992718728e-06,
      "loss": 0.2144,
      "step": 4430
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3417284556355615,
      "learning_rate": 6.742527137386553e-06,
      "loss": 0.18,
      "step": 4431
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4190116959478873,
      "learning_rate": 6.741139129223091e-06,
      "loss": 0.2175,
      "step": 4432
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3093160981251464,
      "learning_rate": 6.739750968350081e-06,
      "loss": 0.1645,
      "step": 4433
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.457420717330241,
      "learning_rate": 6.738362654889271e-06,
      "loss": 0.2305,
      "step": 4434
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.699495134176554,
      "learning_rate": 6.7369741889624285e-06,
      "loss": 0.6382,
      "step": 4435
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2681983051944123,
      "learning_rate": 6.735585570691328e-06,
      "loss": 0.1792,
      "step": 4436
    },
    {
      "epoch": 0.41,
      "grad_norm": 8.473557795072301,
      "learning_rate": 6.734196800197763e-06,
      "loss": 0.48,
      "step": 4437
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.327731552483371,
      "learning_rate": 6.732807877603538e-06,
      "loss": 0.2265,
      "step": 4438
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4651051647969453,
      "learning_rate": 6.7314188030304705e-06,
      "loss": 0.2143,
      "step": 4439
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.530949060170279,
      "learning_rate": 6.73002957660039e-06,
      "loss": 0.2081,
      "step": 4440
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4427482427421583,
      "learning_rate": 6.728640198435143e-06,
      "loss": 0.2493,
      "step": 4441
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2440339183089026,
      "learning_rate": 6.727250668656585e-06,
      "loss": 0.1907,
      "step": 4442
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.348088222153972,
      "learning_rate": 6.7258609873865895e-06,
      "loss": 0.6156,
      "step": 4443
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3947822381315473,
      "learning_rate": 6.724471154747039e-06,
      "loss": 0.2246,
      "step": 4444
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.918952704980463,
      "learning_rate": 6.7230811708598335e-06,
      "loss": 0.5773,
      "step": 4445
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5216649699647857,
      "learning_rate": 6.72169103584688e-06,
      "loss": 0.2298,
      "step": 4446
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.346046608829269,
      "learning_rate": 6.7203007498301075e-06,
      "loss": 0.2126,
      "step": 4447
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.305931248402418,
      "learning_rate": 6.71891031293145e-06,
      "loss": 0.2047,
      "step": 4448
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4861507517839843,
      "learning_rate": 6.717519725272859e-06,
      "loss": 0.2221,
      "step": 4449
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4403882455560981,
      "learning_rate": 6.716128986976297e-06,
      "loss": 0.2273,
      "step": 4450
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4140353277664697,
      "learning_rate": 6.7147380981637435e-06,
      "loss": 0.2249,
      "step": 4451
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4415600192689175,
      "learning_rate": 6.713347058957187e-06,
      "loss": 0.1894,
      "step": 4452
    },
    {
      "epoch": 0.41,
      "grad_norm": 9.256320478396093,
      "learning_rate": 6.711955869478632e-06,
      "loss": 0.4121,
      "step": 4453
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2898673374771206,
      "learning_rate": 6.710564529850094e-06,
      "loss": 0.2184,
      "step": 4454
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2622211753697832,
      "learning_rate": 6.709173040193605e-06,
      "loss": 0.1863,
      "step": 4455
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.322788465087741,
      "learning_rate": 6.707781400631206e-06,
      "loss": 0.1795,
      "step": 4456
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3631221807218459,
      "learning_rate": 6.706389611284953e-06,
      "loss": 0.2327,
      "step": 4457
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6171569618267625,
      "learning_rate": 6.7049976722769165e-06,
      "loss": 0.2615,
      "step": 4458
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2905913112460086,
      "learning_rate": 6.703605583729177e-06,
      "loss": 0.1702,
      "step": 4459
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.412674529342629,
      "learning_rate": 6.70221334576383e-06,
      "loss": 0.2166,
      "step": 4460
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.318301332659725,
      "learning_rate": 6.700820958502987e-06,
      "loss": 0.7022,
      "step": 4461
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2806479904690868,
      "learning_rate": 6.699428422068768e-06,
      "loss": 0.2297,
      "step": 4462
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5859021768709196,
      "learning_rate": 6.698035736583307e-06,
      "loss": 0.2201,
      "step": 4463
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5760802332114046,
      "learning_rate": 6.696642902168753e-06,
      "loss": 0.2064,
      "step": 4464
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.385183406278292,
      "learning_rate": 6.6952499189472665e-06,
      "loss": 0.2434,
      "step": 4465
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3527308033859333,
      "learning_rate": 6.6938567870410206e-06,
      "loss": 0.2122,
      "step": 4466
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4401297025875308,
      "learning_rate": 6.692463506572203e-06,
      "loss": 0.2164,
      "step": 4467
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.437768920840015,
      "learning_rate": 6.691070077663014e-06,
      "loss": 0.1964,
      "step": 4468
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.395193826997989,
      "learning_rate": 6.689676500435667e-06,
      "loss": 0.2237,
      "step": 4469
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2524046683692227,
      "learning_rate": 6.688282775012386e-06,
      "loss": 0.2018,
      "step": 4470
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.543190132177635,
      "learning_rate": 6.686888901515412e-06,
      "loss": 0.2251,
      "step": 4471
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.430227913184342,
      "learning_rate": 6.685494880066996e-06,
      "loss": 0.2284,
      "step": 4472
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3660629670276023,
      "learning_rate": 6.684100710789405e-06,
      "loss": 0.2413,
      "step": 4473
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.362348497802394,
      "learning_rate": 6.6827063938049144e-06,
      "loss": 0.1896,
      "step": 4474
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2342659258732265,
      "learning_rate": 6.681311929235817e-06,
      "loss": 0.1975,
      "step": 4475
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.342293252893099,
      "learning_rate": 6.679917317204414e-06,
      "loss": 0.1999,
      "step": 4476
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.7786224572235145,
      "learning_rate": 6.678522557833025e-06,
      "loss": 0.7824,
      "step": 4477
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2744867867447627,
      "learning_rate": 6.677127651243979e-06,
      "loss": 0.1796,
      "step": 4478
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1835326027442987,
      "learning_rate": 6.675732597559618e-06,
      "loss": 0.1599,
      "step": 4479
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3540799854803913,
      "learning_rate": 6.674337396902297e-06,
      "loss": 0.1969,
      "step": 4480
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.402189529962834,
      "learning_rate": 6.6729420493943875e-06,
      "loss": 0.2274,
      "step": 4481
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.026322783641129,
      "learning_rate": 6.671546555158268e-06,
      "loss": 0.6401,
      "step": 4482
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4708894658633196,
      "learning_rate": 6.6701509143163335e-06,
      "loss": 0.2321,
      "step": 4483
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3673248495389116,
      "learning_rate": 6.66875512699099e-06,
      "loss": 0.1555,
      "step": 4484
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2200823631706903,
      "learning_rate": 6.667359193304659e-06,
      "loss": 0.1738,
      "step": 4485
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3575201207874208,
      "learning_rate": 6.665963113379772e-06,
      "loss": 0.227,
      "step": 4486
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.204100084525027,
      "learning_rate": 6.664566887338776e-06,
      "loss": 0.6481,
      "step": 4487
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6052930555353626,
      "learning_rate": 6.663170515304127e-06,
      "loss": 0.2733,
      "step": 4488
    },
    {
      "epoch": 0.41,
      "grad_norm": 5.429658549222327,
      "learning_rate": 6.6617739973982985e-06,
      "loss": 0.5009,
      "step": 4489
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2220921381168237,
      "learning_rate": 6.660377333743775e-06,
      "loss": 0.1867,
      "step": 4490
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4631797688869748,
      "learning_rate": 6.65898052446305e-06,
      "loss": 0.2401,
      "step": 4491
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1814013445078573,
      "learning_rate": 6.657583569678635e-06,
      "loss": 0.2008,
      "step": 4492
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4457368907803676,
      "learning_rate": 6.656186469513052e-06,
      "loss": 0.1963,
      "step": 4493
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.204792542377426,
      "learning_rate": 6.654789224088837e-06,
      "loss": 0.1896,
      "step": 4494
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2198793296084085,
      "learning_rate": 6.653391833528535e-06,
      "loss": 0.1784,
      "step": 4495
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5293496583123682,
      "learning_rate": 6.65199429795471e-06,
      "loss": 0.1952,
      "step": 4496
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.161190193407612,
      "learning_rate": 6.6505966174899326e-06,
      "loss": 0.1975,
      "step": 4497
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3918678490495788,
      "learning_rate": 6.649198792256791e-06,
      "loss": 0.1701,
      "step": 4498
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.310211017710711,
      "learning_rate": 6.647800822377883e-06,
      "loss": 0.1941,
      "step": 4499
    },
    {
      "epoch": 0.41,
      "grad_norm": 7.333968654204564,
      "learning_rate": 6.6464027079758186e-06,
      "loss": 0.6346,
      "step": 4500
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5875927578839646,
      "learning_rate": 6.645004449173223e-06,
      "loss": 0.2417,
      "step": 4501
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3359158391105872,
      "learning_rate": 6.643606046092732e-06,
      "loss": 0.154,
      "step": 4502
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3646101726218403,
      "learning_rate": 6.642207498856995e-06,
      "loss": 0.1801,
      "step": 4503
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4932781220490494,
      "learning_rate": 6.640808807588675e-06,
      "loss": 0.2456,
      "step": 4504
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3623367778221522,
      "learning_rate": 6.639409972410446e-06,
      "loss": 0.2155,
      "step": 4505
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.108698375847646,
      "learning_rate": 6.638010993444994e-06,
      "loss": 0.197,
      "step": 4506
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2416137622048553,
      "learning_rate": 6.636611870815022e-06,
      "loss": 0.2231,
      "step": 4507
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4010421001384081,
      "learning_rate": 6.635212604643239e-06,
      "loss": 0.1981,
      "step": 4508
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2945375148231926,
      "learning_rate": 6.633813195052373e-06,
      "loss": 0.1997,
      "step": 4509
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3370781698461789,
      "learning_rate": 6.632413642165158e-06,
      "loss": 0.1774,
      "step": 4510
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.415706528347574,
      "learning_rate": 6.631013946104348e-06,
      "loss": 0.2296,
      "step": 4511
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2617586042389592,
      "learning_rate": 6.6296141069927024e-06,
      "loss": 0.2074,
      "step": 4512
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2811960961601896,
      "learning_rate": 6.628214124952999e-06,
      "loss": 0.1975,
      "step": 4513
    },
    {
      "epoch": 0.41,
      "grad_norm": 6.808834931582189,
      "learning_rate": 6.626814000108023e-06,
      "loss": 0.5963,
      "step": 4514
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2185788869580207,
      "learning_rate": 6.625413732580577e-06,
      "loss": 0.1905,
      "step": 4515
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.58994380908534,
      "learning_rate": 6.624013322493474e-06,
      "loss": 0.2457,
      "step": 4516
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3889247400374451,
      "learning_rate": 6.622612769969537e-06,
      "loss": 0.2171,
      "step": 4517
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3998204504723537,
      "learning_rate": 6.621212075131605e-06,
      "loss": 0.2022,
      "step": 4518
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3293746101094066,
      "learning_rate": 6.619811238102529e-06,
      "loss": 0.2003,
      "step": 4519
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.389656810092131,
      "learning_rate": 6.618410259005172e-06,
      "loss": 0.2165,
      "step": 4520
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4579652488120503,
      "learning_rate": 6.617009137962407e-06,
      "loss": 0.1931,
      "step": 4521
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3832400482275915,
      "learning_rate": 6.615607875097124e-06,
      "loss": 0.1908,
      "step": 4522
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3472966088275873,
      "learning_rate": 6.614206470532223e-06,
      "loss": 0.1974,
      "step": 4523
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3279200329158147,
      "learning_rate": 6.6128049243906144e-06,
      "loss": 0.1908,
      "step": 4524
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3294129182294947,
      "learning_rate": 6.611403236795228e-06,
      "loss": 0.1866,
      "step": 4525
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4116212676815385,
      "learning_rate": 6.610001407868995e-06,
      "loss": 0.211,
      "step": 4526
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.484598042084023,
      "learning_rate": 6.608599437734867e-06,
      "loss": 0.2153,
      "step": 4527
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4591296706616883,
      "learning_rate": 6.607197326515808e-06,
      "loss": 0.1883,
      "step": 4528
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2845219726561807,
      "learning_rate": 6.605795074334793e-06,
      "loss": 0.1915,
      "step": 4529
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4952208415050146,
      "learning_rate": 6.604392681314806e-06,
      "loss": 0.2713,
      "step": 4530
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3939936776954045,
      "learning_rate": 6.60299014757885e-06,
      "loss": 0.1938,
      "step": 4531
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2749667124996538,
      "learning_rate": 6.601587473249933e-06,
      "loss": 0.1869,
      "step": 4532
    },
    {
      "epoch": 0.41,
      "grad_norm": 4.397852195679747,
      "learning_rate": 6.600184658451081e-06,
      "loss": 0.3887,
      "step": 4533
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7598184982165879,
      "learning_rate": 6.59878170330533e-06,
      "loss": 0.2105,
      "step": 4534
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3772677315172912,
      "learning_rate": 6.597378607935726e-06,
      "loss": 0.2145,
      "step": 4535
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3580138588842985,
      "learning_rate": 6.595975372465333e-06,
      "loss": 0.1986,
      "step": 4536
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3477127952495502,
      "learning_rate": 6.594571997017224e-06,
      "loss": 0.2132,
      "step": 4537
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2148989490926343,
      "learning_rate": 6.593168481714483e-06,
      "loss": 0.2029,
      "step": 4538
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4406337618612186,
      "learning_rate": 6.591764826680208e-06,
      "loss": 0.188,
      "step": 4539
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3779509487200594,
      "learning_rate": 6.590361032037509e-06,
      "loss": 0.2208,
      "step": 4540
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.306635571750856,
      "learning_rate": 6.588957097909509e-06,
      "loss": 0.1918,
      "step": 4541
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.179551345041812,
      "learning_rate": 6.5875530244193415e-06,
      "loss": 0.4875,
      "step": 4542
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3020299889356515,
      "learning_rate": 6.586148811690154e-06,
      "loss": 0.1745,
      "step": 4543
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4988330100561056,
      "learning_rate": 6.5847444598451045e-06,
      "loss": 0.2261,
      "step": 4544
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3909928684397737,
      "learning_rate": 6.583339969007364e-06,
      "loss": 0.2034,
      "step": 4545
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.338687223646486,
      "learning_rate": 6.5819353393001165e-06,
      "loss": 0.2209,
      "step": 4546
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4678242252912383,
      "learning_rate": 6.580530570846556e-06,
      "loss": 0.2102,
      "step": 4547
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3510118668041158,
      "learning_rate": 6.579125663769892e-06,
      "loss": 0.2103,
      "step": 4548
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.354770985878043,
      "learning_rate": 6.577720618193342e-06,
      "loss": 0.1837,
      "step": 4549
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.35804199853474,
      "learning_rate": 6.5763154342401405e-06,
      "loss": 0.2328,
      "step": 4550
    },
    {
      "epoch": 0.42,
      "grad_norm": 49.05541872748496,
      "learning_rate": 6.57491011203353e-06,
      "loss": 0.5201,
      "step": 4551
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.21041043717854,
      "learning_rate": 6.5735046516967675e-06,
      "loss": 0.178,
      "step": 4552
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3800287403006462,
      "learning_rate": 6.57209905335312e-06,
      "loss": 0.2174,
      "step": 4553
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4095333523264422,
      "learning_rate": 6.570693317125868e-06,
      "loss": 0.1547,
      "step": 4554
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3954510958484527,
      "learning_rate": 6.5692874431383055e-06,
      "loss": 0.2161,
      "step": 4555
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.322569667787132,
      "learning_rate": 6.567881431513734e-06,
      "loss": 0.6504,
      "step": 4556
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.310634833138595,
      "learning_rate": 6.566475282375475e-06,
      "loss": 0.2053,
      "step": 4557
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5280867428850284,
      "learning_rate": 6.565068995846853e-06,
      "loss": 0.2309,
      "step": 4558
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2750534376879714,
      "learning_rate": 6.5636625720512105e-06,
      "loss": 0.2229,
      "step": 4559
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.322110310331337,
      "learning_rate": 6.5622560111119e-06,
      "loss": 0.2195,
      "step": 4560
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3862047426197475,
      "learning_rate": 6.560849313152287e-06,
      "loss": 0.1969,
      "step": 4561
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2275250073005448,
      "learning_rate": 6.559442478295745e-06,
      "loss": 0.1912,
      "step": 4562
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3205416783807797,
      "learning_rate": 6.5580355066656675e-06,
      "loss": 0.208,
      "step": 4563
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3823962119860584,
      "learning_rate": 6.556628398385452e-06,
      "loss": 0.2035,
      "step": 4564
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.47986965427082,
      "learning_rate": 6.555221153578514e-06,
      "loss": 0.1873,
      "step": 4565
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.457008244786316,
      "learning_rate": 6.553813772368276e-06,
      "loss": 0.6431,
      "step": 4566
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3825592014752424,
      "learning_rate": 6.552406254878175e-06,
      "loss": 0.2016,
      "step": 4567
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4287784283087626,
      "learning_rate": 6.550998601231662e-06,
      "loss": 0.244,
      "step": 4568
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3625318910573199,
      "learning_rate": 6.549590811552193e-06,
      "loss": 0.183,
      "step": 4569
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5088224376096449,
      "learning_rate": 6.548182885963246e-06,
      "loss": 0.2311,
      "step": 4570
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4237967848630142,
      "learning_rate": 6.5467748245883015e-06,
      "loss": 0.1769,
      "step": 4571
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3606490987937911,
      "learning_rate": 6.545366627550858e-06,
      "loss": 0.1849,
      "step": 4572
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3564802744514868,
      "learning_rate": 6.543958294974422e-06,
      "loss": 0.1762,
      "step": 4573
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2458753146241008,
      "learning_rate": 6.5425498269825164e-06,
      "loss": 0.2011,
      "step": 4574
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3798413939916538,
      "learning_rate": 6.541141223698672e-06,
      "loss": 0.2076,
      "step": 4575
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4514155072355348,
      "learning_rate": 6.539732485246431e-06,
      "loss": 0.1979,
      "step": 4576
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6172467644682063,
      "learning_rate": 6.538323611749351e-06,
      "loss": 0.2416,
      "step": 4577
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6445930813077587,
      "learning_rate": 6.536914603330999e-06,
      "loss": 0.2553,
      "step": 4578
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.362868625452791,
      "learning_rate": 6.5355054601149545e-06,
      "loss": 0.1905,
      "step": 4579
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2429239091417703,
      "learning_rate": 6.534096182224809e-06,
      "loss": 0.1918,
      "step": 4580
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4032893900554309,
      "learning_rate": 6.532686769784167e-06,
      "loss": 0.2068,
      "step": 4581
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.184442299187689,
      "learning_rate": 6.531277222916641e-06,
      "loss": 0.1582,
      "step": 4582
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4365675147952108,
      "learning_rate": 6.529867541745859e-06,
      "loss": 0.1947,
      "step": 4583
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3045849723857839,
      "learning_rate": 6.528457726395458e-06,
      "loss": 0.2329,
      "step": 4584
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.326569031584843,
      "learning_rate": 6.5270477769890906e-06,
      "loss": 0.1748,
      "step": 4585
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2940361234715656,
      "learning_rate": 6.525637693650417e-06,
      "loss": 0.1705,
      "step": 4586
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.446961539658579,
      "learning_rate": 6.5242274765031124e-06,
      "loss": 0.2171,
      "step": 4587
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4555463804994277,
      "learning_rate": 6.522817125670861e-06,
      "loss": 0.2161,
      "step": 4588
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5789838040448354,
      "learning_rate": 6.521406641277359e-06,
      "loss": 0.2074,
      "step": 4589
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3672773025039895,
      "learning_rate": 6.519996023446318e-06,
      "loss": 0.2573,
      "step": 4590
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3268083407829407,
      "learning_rate": 6.518585272301458e-06,
      "loss": 0.2042,
      "step": 4591
    },
    {
      "epoch": 0.42,
      "grad_norm": 9.82247499961122,
      "learning_rate": 6.517174387966509e-06,
      "loss": 0.7378,
      "step": 4592
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5740422456799001,
      "learning_rate": 6.515763370565218e-06,
      "loss": 0.2659,
      "step": 4593
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3919956427055427,
      "learning_rate": 6.51435222022134e-06,
      "loss": 0.2025,
      "step": 4594
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.478911091819167,
      "learning_rate": 6.51294093705864e-06,
      "loss": 0.5212,
      "step": 4595
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.759211628861096,
      "learning_rate": 6.5115295212008985e-06,
      "loss": 0.7072,
      "step": 4596
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4045527394431552,
      "learning_rate": 6.510117972771906e-06,
      "loss": 0.2226,
      "step": 4597
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3244031882331573,
      "learning_rate": 6.5087062918954655e-06,
      "loss": 0.2116,
      "step": 4598
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2232541553290448,
      "learning_rate": 6.507294478695392e-06,
      "loss": 0.2153,
      "step": 4599
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.366730530673693,
      "learning_rate": 6.505882533295508e-06,
      "loss": 0.2182,
      "step": 4600
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.355420785807701,
      "learning_rate": 6.504470455819651e-06,
      "loss": 0.2077,
      "step": 4601
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2772509378144847,
      "learning_rate": 6.503058246391671e-06,
      "loss": 0.1888,
      "step": 4602
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.853779537359721,
      "learning_rate": 6.501645905135428e-06,
      "loss": 0.565,
      "step": 4603
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.52887893937297,
      "learning_rate": 6.500233432174794e-06,
      "loss": 0.2251,
      "step": 4604
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3706191063887951,
      "learning_rate": 6.498820827633651e-06,
      "loss": 0.1976,
      "step": 4605
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.893646848704328,
      "learning_rate": 6.4974080916358935e-06,
      "loss": 0.5523,
      "step": 4606
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.270889804905594,
      "learning_rate": 6.495995224305431e-06,
      "loss": 0.1487,
      "step": 4607
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4394925265269203,
      "learning_rate": 6.494582225766179e-06,
      "loss": 0.2173,
      "step": 4608
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3324691013383931,
      "learning_rate": 6.493169096142068e-06,
      "loss": 0.1782,
      "step": 4609
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3170291998082928,
      "learning_rate": 6.491755835557038e-06,
      "loss": 0.1986,
      "step": 4610
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3991184081653028,
      "learning_rate": 6.4903424441350405e-06,
      "loss": 0.2263,
      "step": 4611
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2773583815139724,
      "learning_rate": 6.488928922000042e-06,
      "loss": 0.1679,
      "step": 4612
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3561642809532908,
      "learning_rate": 6.487515269276015e-06,
      "loss": 0.1957,
      "step": 4613
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.373668841518135,
      "learning_rate": 6.486101486086949e-06,
      "loss": 0.4842,
      "step": 4614
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.287166092398606,
      "learning_rate": 6.4846875725568405e-06,
      "loss": 0.1829,
      "step": 4615
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.334977404765034,
      "learning_rate": 6.483273528809699e-06,
      "loss": 0.1912,
      "step": 4616
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6366590190713002,
      "learning_rate": 6.481859354969549e-06,
      "loss": 0.2107,
      "step": 4617
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4974992776350158,
      "learning_rate": 6.480445051160418e-06,
      "loss": 0.2182,
      "step": 4618
    },
    {
      "epoch": 0.42,
      "grad_norm": 11.617853997630425,
      "learning_rate": 6.4790306175063535e-06,
      "loss": 0.5754,
      "step": 4619
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3424948636431786,
      "learning_rate": 6.477616054131408e-06,
      "loss": 0.198,
      "step": 4620
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.32965865569824,
      "learning_rate": 6.4762013611596506e-06,
      "loss": 0.193,
      "step": 4621
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0979580446401223,
      "learning_rate": 6.474786538715158e-06,
      "loss": 0.1679,
      "step": 4622
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.479683172632388,
      "learning_rate": 6.47337158692202e-06,
      "loss": 0.5511,
      "step": 4623
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2960053277780532,
      "learning_rate": 6.4719565059043365e-06,
      "loss": 0.241,
      "step": 4624
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4683785152652042,
      "learning_rate": 6.470541295786222e-06,
      "loss": 0.226,
      "step": 4625
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2613683969142966,
      "learning_rate": 6.469125956691798e-06,
      "loss": 0.2175,
      "step": 4626
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3974288831066939,
      "learning_rate": 6.467710488745199e-06,
      "loss": 0.205,
      "step": 4627
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3556216293264696,
      "learning_rate": 6.466294892070571e-06,
      "loss": 0.1973,
      "step": 4628
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4383859457691341,
      "learning_rate": 6.464879166792073e-06,
      "loss": 0.203,
      "step": 4629
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3470162179581797,
      "learning_rate": 6.463463313033872e-06,
      "loss": 0.1855,
      "step": 4630
    },
    {
      "epoch": 0.42,
      "grad_norm": 5.19626145913329,
      "learning_rate": 6.4620473309201485e-06,
      "loss": 0.611,
      "step": 4631
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5112332404899222,
      "learning_rate": 6.460631220575092e-06,
      "loss": 0.2215,
      "step": 4632
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2924631544859357,
      "learning_rate": 6.4592149821229064e-06,
      "loss": 0.1723,
      "step": 4633
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.304530023201297,
      "learning_rate": 6.457798615687807e-06,
      "loss": 0.2025,
      "step": 4634
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3507584820642902,
      "learning_rate": 6.4563821213940156e-06,
      "loss": 0.176,
      "step": 4635
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2937740839847107,
      "learning_rate": 6.4549654993657695e-06,
      "loss": 0.1799,
      "step": 4636
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3868099091279027,
      "learning_rate": 6.453548749727316e-06,
      "loss": 0.2325,
      "step": 4637
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.327052652286221,
      "learning_rate": 6.452131872602911e-06,
      "loss": 0.1936,
      "step": 4638
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6529484478390128,
      "learning_rate": 6.450714868116829e-06,
      "loss": 0.2593,
      "step": 4639
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3884987416479941,
      "learning_rate": 6.4492977363933465e-06,
      "loss": 0.231,
      "step": 4640
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3101845139558115,
      "learning_rate": 6.447880477556757e-06,
      "loss": 0.1832,
      "step": 4641
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3602180680616818,
      "learning_rate": 6.446463091731366e-06,
      "loss": 0.1993,
      "step": 4642
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.298487781335301,
      "learning_rate": 6.445045579041484e-06,
      "loss": 0.2013,
      "step": 4643
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5093918952032577,
      "learning_rate": 6.443627939611437e-06,
      "loss": 0.2251,
      "step": 4644
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.332129134277426,
      "learning_rate": 6.442210173565562e-06,
      "loss": 0.1856,
      "step": 4645
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.450861513550017,
      "learning_rate": 6.4407922810282075e-06,
      "loss": 0.2269,
      "step": 4646
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5536975730360434,
      "learning_rate": 6.439374262123731e-06,
      "loss": 0.2342,
      "step": 4647
    },
    {
      "epoch": 0.42,
      "grad_norm": 6.82050974718885,
      "learning_rate": 6.4379561169765024e-06,
      "loss": 0.7429,
      "step": 4648
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.047311166764569,
      "learning_rate": 6.436537845710904e-06,
      "loss": 0.5912,
      "step": 4649
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3729477437517335,
      "learning_rate": 6.435119448451326e-06,
      "loss": 0.2145,
      "step": 4650
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.115084598471266,
      "learning_rate": 6.4337009253221704e-06,
      "loss": 0.5776,
      "step": 4651
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2565548793419676,
      "learning_rate": 6.432282276447855e-06,
      "loss": 0.176,
      "step": 4652
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.524764622757212,
      "learning_rate": 6.430863501952801e-06,
      "loss": 0.2118,
      "step": 4653
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.349428455248947,
      "learning_rate": 6.429444601961445e-06,
      "loss": 0.2381,
      "step": 4654
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3290302050701377,
      "learning_rate": 6.428025576598235e-06,
      "loss": 0.1806,
      "step": 4655
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2127596888277024,
      "learning_rate": 6.426606425987631e-06,
      "loss": 0.1972,
      "step": 4656
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.273101842485808,
      "learning_rate": 6.425187150254097e-06,
      "loss": 0.16,
      "step": 4657
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2467616184677255,
      "learning_rate": 6.423767749522116e-06,
      "loss": 0.1751,
      "step": 4658
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.501009773512423,
      "learning_rate": 6.422348223916181e-06,
      "loss": 0.6418,
      "step": 4659
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3108305531387459,
      "learning_rate": 6.420928573560789e-06,
      "loss": 0.2169,
      "step": 4660
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4441745539810873,
      "learning_rate": 6.419508798580458e-06,
      "loss": 0.2214,
      "step": 4661
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2825998331549515,
      "learning_rate": 6.418088899099707e-06,
      "loss": 0.159,
      "step": 4662
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3682496322789262,
      "learning_rate": 6.416668875243074e-06,
      "loss": 0.1977,
      "step": 4663
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.2315456974830115,
      "learning_rate": 6.415248727135103e-06,
      "loss": 0.6051,
      "step": 4664
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3093508214234963,
      "learning_rate": 6.413828454900351e-06,
      "loss": 0.1751,
      "step": 4665
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1804847097034121,
      "learning_rate": 6.412408058663385e-06,
      "loss": 0.1558,
      "step": 4666
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2614480755184807,
      "learning_rate": 6.410987538548784e-06,
      "loss": 0.181,
      "step": 4667
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3829502870233044,
      "learning_rate": 6.4095668946811364e-06,
      "loss": 0.2076,
      "step": 4668
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1983148645770028,
      "learning_rate": 6.408146127185044e-06,
      "loss": 0.1948,
      "step": 4669
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2948988058781983,
      "learning_rate": 6.406725236185114e-06,
      "loss": 0.1821,
      "step": 4670
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.473093097214795,
      "learning_rate": 6.405304221805972e-06,
      "loss": 0.2157,
      "step": 4671
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4253261560176453,
      "learning_rate": 6.4038830841722465e-06,
      "loss": 0.1746,
      "step": 4672
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3130351488645966,
      "learning_rate": 6.402461823408584e-06,
      "loss": 0.2292,
      "step": 4673
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2678616445472504,
      "learning_rate": 6.401040439639638e-06,
      "loss": 0.2122,
      "step": 4674
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.376952233011698,
      "learning_rate": 6.3996189329900716e-06,
      "loss": 0.2047,
      "step": 4675
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4024737452169915,
      "learning_rate": 6.398197303584562e-06,
      "loss": 0.2141,
      "step": 4676
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1765147261827653,
      "learning_rate": 6.396775551547796e-06,
      "loss": 0.1548,
      "step": 4677
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6098496512081455,
      "learning_rate": 6.39535367700447e-06,
      "loss": 0.2931,
      "step": 4678
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2445292290146257,
      "learning_rate": 6.393931680079293e-06,
      "loss": 0.1949,
      "step": 4679
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5634254776282088,
      "learning_rate": 6.39250956089698e-06,
      "loss": 0.1681,
      "step": 4680
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4277691930308085,
      "learning_rate": 6.391087319582264e-06,
      "loss": 0.2474,
      "step": 4681
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3895605026663223,
      "learning_rate": 6.3896649562598845e-06,
      "loss": 0.2127,
      "step": 4682
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5458908024364495,
      "learning_rate": 6.388242471054591e-06,
      "loss": 0.2594,
      "step": 4683
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.362268321782668,
      "learning_rate": 6.386819864091146e-06,
      "loss": 0.1879,
      "step": 4684
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.406859881678999,
      "learning_rate": 6.385397135494323e-06,
      "loss": 0.2068,
      "step": 4685
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5236978281247182,
      "learning_rate": 6.383974285388902e-06,
      "loss": 0.2597,
      "step": 4686
    },
    {
      "epoch": 0.43,
      "grad_norm": 8.791641117515594,
      "learning_rate": 6.382551313899677e-06,
      "loss": 0.6302,
      "step": 4687
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.508228422970902,
      "learning_rate": 6.381128221151453e-06,
      "loss": 0.5981,
      "step": 4688
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.33583446897593,
      "learning_rate": 6.379705007269046e-06,
      "loss": 0.1936,
      "step": 4689
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.44130662253187,
      "learning_rate": 6.378281672377279e-06,
      "loss": 0.2238,
      "step": 4690
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.326664496668688,
      "learning_rate": 6.3768582166009884e-06,
      "loss": 0.1755,
      "step": 4691
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2426342424176673,
      "learning_rate": 6.375434640065023e-06,
      "loss": 0.2003,
      "step": 4692
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.42636654851433,
      "learning_rate": 6.374010942894237e-06,
      "loss": 0.2162,
      "step": 4693
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4381094873943352,
      "learning_rate": 6.3725871252134984e-06,
      "loss": 0.2306,
      "step": 4694
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4280763829484047,
      "learning_rate": 6.37116318714769e-06,
      "loss": 0.2036,
      "step": 4695
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.234857486440399,
      "learning_rate": 6.369739128821695e-06,
      "loss": 0.639,
      "step": 4696
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.414968848527583,
      "learning_rate": 6.368314950360416e-06,
      "loss": 0.2435,
      "step": 4697
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4764132027305,
      "learning_rate": 6.366890651888762e-06,
      "loss": 0.232,
      "step": 4698
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4188290541681947,
      "learning_rate": 6.365466233531652e-06,
      "loss": 0.2063,
      "step": 4699
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3979288043447768,
      "learning_rate": 6.36404169541402e-06,
      "loss": 0.1807,
      "step": 4700
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4309792524910012,
      "learning_rate": 6.3626170376608066e-06,
      "loss": 0.2272,
      "step": 4701
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.368267121201294,
      "learning_rate": 6.361192260396963e-06,
      "loss": 0.2442,
      "step": 4702
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2876803911607964,
      "learning_rate": 6.359767363747453e-06,
      "loss": 0.181,
      "step": 4703
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2879412083520965,
      "learning_rate": 6.358342347837248e-06,
      "loss": 0.1826,
      "step": 4704
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.402598739193164,
      "learning_rate": 6.356917212791332e-06,
      "loss": 0.1974,
      "step": 4705
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0173074497228667,
      "learning_rate": 6.355491958734699e-06,
      "loss": 0.1291,
      "step": 4706
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2959155337126809,
      "learning_rate": 6.354066585792354e-06,
      "loss": 0.244,
      "step": 4707
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5286927441281273,
      "learning_rate": 6.352641094089309e-06,
      "loss": 0.1879,
      "step": 4708
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3334382895422878,
      "learning_rate": 6.351215483750594e-06,
      "loss": 0.2106,
      "step": 4709
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.818372568750834,
      "learning_rate": 6.3497897549012385e-06,
      "loss": 0.6055,
      "step": 4710
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4499177789388396,
      "learning_rate": 6.348363907666294e-06,
      "loss": 0.234,
      "step": 4711
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.697786699149427,
      "learning_rate": 6.346937942170815e-06,
      "loss": 0.606,
      "step": 4712
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.584548915113418,
      "learning_rate": 6.3455118585398676e-06,
      "loss": 0.2111,
      "step": 4713
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.336864318394207,
      "learning_rate": 6.344085656898528e-06,
      "loss": 0.3728,
      "step": 4714
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.67204677531407,
      "learning_rate": 6.342659337371884e-06,
      "loss": 0.2429,
      "step": 4715
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.644610392684093,
      "learning_rate": 6.341232900085036e-06,
      "loss": 0.5275,
      "step": 4716
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2893518703052875,
      "learning_rate": 6.339806345163089e-06,
      "loss": 0.1975,
      "step": 4717
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.382671804236823,
      "learning_rate": 6.338379672731163e-06,
      "loss": 0.2121,
      "step": 4718
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.335398797418444,
      "learning_rate": 6.336952882914385e-06,
      "loss": 0.1723,
      "step": 4719
    },
    {
      "epoch": 0.43,
      "grad_norm": 5.532115036755214,
      "learning_rate": 6.335525975837897e-06,
      "loss": 0.4579,
      "step": 4720
    },
    {
      "epoch": 0.43,
      "grad_norm": 9.827359783725393,
      "learning_rate": 6.334098951626847e-06,
      "loss": 0.6614,
      "step": 4721
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4161073634546815,
      "learning_rate": 6.332671810406392e-06,
      "loss": 0.2049,
      "step": 4722
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3238743143920457,
      "learning_rate": 6.331244552301705e-06,
      "loss": 0.1925,
      "step": 4723
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3060069666722087,
      "learning_rate": 6.329817177437965e-06,
      "loss": 0.1705,
      "step": 4724
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4436877582530425,
      "learning_rate": 6.3283896859403624e-06,
      "loss": 0.1816,
      "step": 4725
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.340500651344923,
      "learning_rate": 6.326962077934098e-06,
      "loss": 0.2423,
      "step": 4726
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1743892432472225,
      "learning_rate": 6.325534353544382e-06,
      "loss": 0.1676,
      "step": 4727
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4010133618498413,
      "learning_rate": 6.324106512896436e-06,
      "loss": 0.1963,
      "step": 4728
    },
    {
      "epoch": 0.43,
      "grad_norm": 6.263080409188105,
      "learning_rate": 6.3226785561154914e-06,
      "loss": 0.4825,
      "step": 4729
    },
    {
      "epoch": 0.43,
      "grad_norm": 4.706784633949319,
      "learning_rate": 6.321250483326789e-06,
      "loss": 0.589,
      "step": 4730
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2975912656892161,
      "learning_rate": 6.319822294655579e-06,
      "loss": 0.1789,
      "step": 4731
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2399037888401705,
      "learning_rate": 6.318393990227125e-06,
      "loss": 0.1508,
      "step": 4732
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.719602067180062,
      "learning_rate": 6.316965570166698e-06,
      "loss": 0.2549,
      "step": 4733
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5206703715762786,
      "learning_rate": 6.315537034599582e-06,
      "loss": 0.2309,
      "step": 4734
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.466037694046505,
      "learning_rate": 6.314108383651065e-06,
      "loss": 0.2014,
      "step": 4735
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5785461875835125,
      "learning_rate": 6.312679617446453e-06,
      "loss": 0.2155,
      "step": 4736
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4926926245706627,
      "learning_rate": 6.311250736111058e-06,
      "loss": 0.2282,
      "step": 4737
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.449187209098158,
      "learning_rate": 6.3098217397702e-06,
      "loss": 0.2155,
      "step": 4738
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6491083529874766,
      "learning_rate": 6.308392628549214e-06,
      "loss": 0.2221,
      "step": 4739
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3107553010111768,
      "learning_rate": 6.30696340257344e-06,
      "loss": 0.2018,
      "step": 4740
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2124455418537146,
      "learning_rate": 6.305534061968235e-06,
      "loss": 0.1977,
      "step": 4741
    },
    {
      "epoch": 0.43,
      "grad_norm": 17.3142238896579,
      "learning_rate": 6.304104606858956e-06,
      "loss": 0.5753,
      "step": 4742
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2605635667040522,
      "learning_rate": 6.302675037370981e-06,
      "loss": 0.1838,
      "step": 4743
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4113713870475122,
      "learning_rate": 6.301245353629691e-06,
      "loss": 0.1984,
      "step": 4744
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2854259334572542,
      "learning_rate": 6.299815555760478e-06,
      "loss": 0.1723,
      "step": 4745
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5325071003937991,
      "learning_rate": 6.298385643888744e-06,
      "loss": 0.2078,
      "step": 4746
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5006640397166748,
      "learning_rate": 6.296955618139904e-06,
      "loss": 0.2787,
      "step": 4747
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.250707632593525,
      "learning_rate": 6.295525478639379e-06,
      "loss": 0.1612,
      "step": 4748
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4234992297850955,
      "learning_rate": 6.294095225512604e-06,
      "loss": 0.2032,
      "step": 4749
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2603360717372,
      "learning_rate": 6.29266485888502e-06,
      "loss": 0.1874,
      "step": 4750
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3063752078700195,
      "learning_rate": 6.291234378882081e-06,
      "loss": 0.2028,
      "step": 4751
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2338969761050427,
      "learning_rate": 6.289803785629248e-06,
      "loss": 0.1683,
      "step": 4752
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4396438649919974,
      "learning_rate": 6.288373079251996e-06,
      "loss": 0.1684,
      "step": 4753
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2856990975727536,
      "learning_rate": 6.286942259875805e-06,
      "loss": 0.1707,
      "step": 4754
    },
    {
      "epoch": 0.43,
      "grad_norm": 7.142911345545256,
      "learning_rate": 6.285511327626168e-06,
      "loss": 0.7632,
      "step": 4755
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3312541192550633,
      "learning_rate": 6.284080282628588e-06,
      "loss": 0.2127,
      "step": 4756
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4069529003858336,
      "learning_rate": 6.282649125008578e-06,
      "loss": 0.1995,
      "step": 4757
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4202788718899495,
      "learning_rate": 6.281217854891658e-06,
      "loss": 0.2105,
      "step": 4758
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3106016398621878,
      "learning_rate": 6.279786472403362e-06,
      "loss": 0.1815,
      "step": 4759
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2726335100779729,
      "learning_rate": 6.27835497766923e-06,
      "loss": 0.2073,
      "step": 4760
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2419441160620859,
      "learning_rate": 6.276923370814815e-06,
      "loss": 0.2229,
      "step": 4761
    },
    {
      "epoch": 0.44,
      "grad_norm": 17.48995382992211,
      "learning_rate": 6.275491651965678e-06,
      "loss": 0.6398,
      "step": 4762
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2775392114494137,
      "learning_rate": 6.274059821247392e-06,
      "loss": 0.1723,
      "step": 4763
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1668199484876842,
      "learning_rate": 6.272627878785533e-06,
      "loss": 0.1865,
      "step": 4764
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.381284115674678,
      "learning_rate": 6.271195824705698e-06,
      "loss": 0.1799,
      "step": 4765
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.948840787969439,
      "learning_rate": 6.269763659133486e-06,
      "loss": 0.6997,
      "step": 4766
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4056416509309684,
      "learning_rate": 6.268331382194504e-06,
      "loss": 0.1841,
      "step": 4767
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3192518325252351,
      "learning_rate": 6.266898994014376e-06,
      "loss": 0.1941,
      "step": 4768
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.335547076487659,
      "learning_rate": 6.265466494718731e-06,
      "loss": 0.227,
      "step": 4769
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3193567680578695,
      "learning_rate": 6.26403388443321e-06,
      "loss": 0.1993,
      "step": 4770
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.529354506105235,
      "learning_rate": 6.2626011632834605e-06,
      "loss": 0.2446,
      "step": 4771
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3178841081128962,
      "learning_rate": 6.2611683313951424e-06,
      "loss": 0.2178,
      "step": 4772
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.246463058556305,
      "learning_rate": 6.2597353888939234e-06,
      "loss": 0.1872,
      "step": 4773
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.498794469328186,
      "learning_rate": 6.258302335905485e-06,
      "loss": 0.202,
      "step": 4774
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.615891528492879,
      "learning_rate": 6.2568691725555144e-06,
      "loss": 0.217,
      "step": 4775
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3127091790676912,
      "learning_rate": 6.2554358989697086e-06,
      "loss": 0.1691,
      "step": 4776
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3487358577369517,
      "learning_rate": 6.254002515273775e-06,
      "loss": 0.21,
      "step": 4777
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.413444276855874,
      "learning_rate": 6.252569021593432e-06,
      "loss": 0.2164,
      "step": 4778
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2180911786983273,
      "learning_rate": 6.251135418054408e-06,
      "loss": 0.1615,
      "step": 4779
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3890308235251658,
      "learning_rate": 6.249701704782437e-06,
      "loss": 0.1843,
      "step": 4780
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3340913471278564,
      "learning_rate": 6.248267881903267e-06,
      "loss": 0.2095,
      "step": 4781
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.115449075302745,
      "learning_rate": 6.246833949542652e-06,
      "loss": 0.6402,
      "step": 4782
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2314362687786153,
      "learning_rate": 6.2453999078263596e-06,
      "loss": 0.2091,
      "step": 4783
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4154607584353665,
      "learning_rate": 6.243965756880163e-06,
      "loss": 0.203,
      "step": 4784
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4324168168888967,
      "learning_rate": 6.242531496829848e-06,
      "loss": 0.1976,
      "step": 4785
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6392801885027,
      "learning_rate": 6.241097127801208e-06,
      "loss": 0.217,
      "step": 4786
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.534279611579517,
      "learning_rate": 6.239662649920047e-06,
      "loss": 0.5035,
      "step": 4787
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.417305833594074,
      "learning_rate": 6.238228063312181e-06,
      "loss": 0.6296,
      "step": 4788
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.063878928187685,
      "learning_rate": 6.2367933681034285e-06,
      "loss": 0.7229,
      "step": 4789
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.364875266964722,
      "learning_rate": 6.235358564419625e-06,
      "loss": 0.2171,
      "step": 4790
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2206856627312457,
      "learning_rate": 6.233923652386609e-06,
      "loss": 0.1423,
      "step": 4791
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1590073195556936,
      "learning_rate": 6.232488632130236e-06,
      "loss": 0.1566,
      "step": 4792
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.313073548655557,
      "learning_rate": 6.231053503776363e-06,
      "loss": 0.1923,
      "step": 4793
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5759377560408387,
      "learning_rate": 6.229618267450863e-06,
      "loss": 0.2525,
      "step": 4794
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2895760703488597,
      "learning_rate": 6.2281829232796165e-06,
      "loss": 0.1978,
      "step": 4795
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2334951080005137,
      "learning_rate": 6.226747471388511e-06,
      "loss": 0.1912,
      "step": 4796
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.404845923326964,
      "learning_rate": 6.2253119119034445e-06,
      "loss": 0.2399,
      "step": 4797
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4133637911953745,
      "learning_rate": 6.2238762449503265e-06,
      "loss": 0.1874,
      "step": 4798
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4163136964773402,
      "learning_rate": 6.222440470655075e-06,
      "loss": 0.214,
      "step": 4799
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.353889834983274,
      "learning_rate": 6.221004589143615e-06,
      "loss": 0.2146,
      "step": 4800
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4498168497832153,
      "learning_rate": 6.219568600541886e-06,
      "loss": 0.2268,
      "step": 4801
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.966454711864838,
      "learning_rate": 6.218132504975831e-06,
      "loss": 0.6105,
      "step": 4802
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3947849157908387,
      "learning_rate": 6.216696302571408e-06,
      "loss": 0.2264,
      "step": 4803
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4212975703829778,
      "learning_rate": 6.215259993454578e-06,
      "loss": 0.2115,
      "step": 4804
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2534734984418936,
      "learning_rate": 6.213823577751319e-06,
      "loss": 0.1849,
      "step": 4805
    },
    {
      "epoch": 0.44,
      "grad_norm": 14.60945081469179,
      "learning_rate": 6.21238705558761e-06,
      "loss": 0.5141,
      "step": 4806
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3741179582438667,
      "learning_rate": 6.210950427089446e-06,
      "loss": 0.2296,
      "step": 4807
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2891652155979105,
      "learning_rate": 6.209513692382828e-06,
      "loss": 0.1744,
      "step": 4808
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.25476548082558,
      "learning_rate": 6.208076851593768e-06,
      "loss": 0.6478,
      "step": 4809
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.702561736041564,
      "learning_rate": 6.206639904848287e-06,
      "loss": 0.6669,
      "step": 4810
    },
    {
      "epoch": 0.44,
      "grad_norm": 11.517669336788632,
      "learning_rate": 6.205202852272415e-06,
      "loss": 0.6256,
      "step": 4811
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3154243783758115,
      "learning_rate": 6.203765693992189e-06,
      "loss": 0.2288,
      "step": 4812
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4138702113035753,
      "learning_rate": 6.202328430133661e-06,
      "loss": 0.2175,
      "step": 4813
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.68669330702574,
      "learning_rate": 6.200891060822884e-06,
      "loss": 0.61,
      "step": 4814
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.440172963371198,
      "learning_rate": 6.199453586185928e-06,
      "loss": 0.4767,
      "step": 4815
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.390725327627807,
      "learning_rate": 6.198016006348868e-06,
      "loss": 0.1785,
      "step": 4816
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3192490499138432,
      "learning_rate": 6.1965783214377895e-06,
      "loss": 0.1964,
      "step": 4817
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.393276696280079,
      "learning_rate": 6.1951405315787886e-06,
      "loss": 0.2312,
      "step": 4818
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2999058646367325,
      "learning_rate": 6.193702636897967e-06,
      "loss": 0.2003,
      "step": 4819
    },
    {
      "epoch": 0.44,
      "grad_norm": 8.88944367931529,
      "learning_rate": 6.192264637521439e-06,
      "loss": 0.6991,
      "step": 4820
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.232290516600067,
      "learning_rate": 6.190826533575326e-06,
      "loss": 0.1769,
      "step": 4821
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2855471637097606,
      "learning_rate": 6.18938832518576e-06,
      "loss": 0.1803,
      "step": 4822
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3900110908371688,
      "learning_rate": 6.187950012478881e-06,
      "loss": 0.2119,
      "step": 4823
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5210909907288606,
      "learning_rate": 6.186511595580838e-06,
      "loss": 0.2168,
      "step": 4824
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.574957448237485,
      "learning_rate": 6.185073074617793e-06,
      "loss": 0.5535,
      "step": 4825
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4370383016414656,
      "learning_rate": 6.183634449715909e-06,
      "loss": 0.184,
      "step": 4826
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5315483411039308,
      "learning_rate": 6.182195721001366e-06,
      "loss": 0.2357,
      "step": 4827
    },
    {
      "epoch": 0.44,
      "grad_norm": 4.854346056409095,
      "learning_rate": 6.18075688860035e-06,
      "loss": 0.4608,
      "step": 4828
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3764957302774197,
      "learning_rate": 6.179317952639057e-06,
      "loss": 0.1944,
      "step": 4829
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3275131268819236,
      "learning_rate": 6.177878913243689e-06,
      "loss": 0.2042,
      "step": 4830
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2240854004857484,
      "learning_rate": 6.176439770540461e-06,
      "loss": 0.1679,
      "step": 4831
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3528687581956425,
      "learning_rate": 6.175000524655596e-06,
      "loss": 0.2242,
      "step": 4832
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4101640961310458,
      "learning_rate": 6.173561175715323e-06,
      "loss": 0.2001,
      "step": 4833
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1648500479787813,
      "learning_rate": 6.172121723845883e-06,
      "loss": 0.1536,
      "step": 4834
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.684053801073858,
      "learning_rate": 6.170682169173528e-06,
      "loss": 0.5907,
      "step": 4835
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3293739673189275,
      "learning_rate": 6.169242511824516e-06,
      "loss": 0.169,
      "step": 4836
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.449532228188714,
      "learning_rate": 6.167802751925112e-06,
      "loss": 0.6167,
      "step": 4837
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4386696411093616,
      "learning_rate": 6.166362889601596e-06,
      "loss": 0.1908,
      "step": 4838
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.320895054313711,
      "learning_rate": 6.164922924980252e-06,
      "loss": 0.202,
      "step": 4839
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3348494217614502,
      "learning_rate": 6.163482858187373e-06,
      "loss": 0.1951,
      "step": 4840
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3698129944809776,
      "learning_rate": 6.1620426893492645e-06,
      "loss": 0.2311,
      "step": 4841
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6030151145756937,
      "learning_rate": 6.160602418592239e-06,
      "loss": 0.2712,
      "step": 4842
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3344381774955687,
      "learning_rate": 6.159162046042616e-06,
      "loss": 0.1785,
      "step": 4843
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4482536023758077,
      "learning_rate": 6.157721571826728e-06,
      "loss": 0.2501,
      "step": 4844
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.481797324782395,
      "learning_rate": 6.156280996070913e-06,
      "loss": 0.1864,
      "step": 4845
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4873990009597444,
      "learning_rate": 6.154840318901518e-06,
      "loss": 0.209,
      "step": 4846
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4059145737126082,
      "learning_rate": 6.153399540444903e-06,
      "loss": 0.1716,
      "step": 4847
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3464414068934225,
      "learning_rate": 6.151958660827434e-06,
      "loss": 0.1762,
      "step": 4848
    },
    {
      "epoch": 0.44,
      "grad_norm": 5.268687925215583,
      "learning_rate": 6.150517680175482e-06,
      "loss": 0.4762,
      "step": 4849
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2844635579248214,
      "learning_rate": 6.149076598615433e-06,
      "loss": 0.1922,
      "step": 4850
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2517840773159798,
      "learning_rate": 6.147635416273679e-06,
      "loss": 0.1717,
      "step": 4851
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3803880670069881,
      "learning_rate": 6.146194133276622e-06,
      "loss": 0.1944,
      "step": 4852
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3894630346436425,
      "learning_rate": 6.144752749750671e-06,
      "loss": 0.2393,
      "step": 4853
    },
    {
      "epoch": 0.44,
      "grad_norm": 11.520152892099603,
      "learning_rate": 6.143311265822246e-06,
      "loss": 0.8701,
      "step": 4854
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3192542967934107,
      "learning_rate": 6.141869681617775e-06,
      "loss": 0.1808,
      "step": 4855
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.343273783013129,
      "learning_rate": 6.140427997263695e-06,
      "loss": 0.1976,
      "step": 4856
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5820842810550493,
      "learning_rate": 6.13898621288645e-06,
      "loss": 0.2273,
      "step": 4857
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4765241308212975,
      "learning_rate": 6.137544328612494e-06,
      "loss": 0.2095,
      "step": 4858
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6146179740882973,
      "learning_rate": 6.13610234456829e-06,
      "loss": 0.2045,
      "step": 4859
    },
    {
      "epoch": 0.44,
      "grad_norm": 7.468821377512437,
      "learning_rate": 6.13466026088031e-06,
      "loss": 0.5534,
      "step": 4860
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3092966018646013,
      "learning_rate": 6.133218077675035e-06,
      "loss": 0.1874,
      "step": 4861
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3804761112295225,
      "learning_rate": 6.1317757950789535e-06,
      "loss": 0.2341,
      "step": 4862
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.821538593015031,
      "learning_rate": 6.1303334132185645e-06,
      "loss": 0.5444,
      "step": 4863
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3933415073193691,
      "learning_rate": 6.128890932220373e-06,
      "loss": 0.1776,
      "step": 4864
    },
    {
      "epoch": 0.44,
      "grad_norm": 6.394139581141159,
      "learning_rate": 6.127448352210894e-06,
      "loss": 0.5707,
      "step": 4865
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4446042082509545,
      "learning_rate": 6.126005673316653e-06,
      "loss": 0.2257,
      "step": 4866
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2729179460393114,
      "learning_rate": 6.124562895664181e-06,
      "loss": 0.2117,
      "step": 4867
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.6071821124015635,
      "learning_rate": 6.123120019380021e-06,
      "loss": 0.5398,
      "step": 4868
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4090926452407408,
      "learning_rate": 6.121677044590721e-06,
      "loss": 0.2412,
      "step": 4869
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4336586999260292,
      "learning_rate": 6.120233971422841e-06,
      "loss": 0.2525,
      "step": 4870
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5683597783449243,
      "learning_rate": 6.118790800002947e-06,
      "loss": 0.2078,
      "step": 4871
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2514545211414312,
      "learning_rate": 6.117347530457616e-06,
      "loss": 0.1688,
      "step": 4872
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4203928454253072,
      "learning_rate": 6.115904162913431e-06,
      "loss": 0.2262,
      "step": 4873
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3886943993442842,
      "learning_rate": 6.114460697496987e-06,
      "loss": 0.1884,
      "step": 4874
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4804503910883677,
      "learning_rate": 6.113017134334883e-06,
      "loss": 0.2045,
      "step": 4875
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5317072308164614,
      "learning_rate": 6.1115734735537316e-06,
      "loss": 0.1988,
      "step": 4876
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4478202260978994,
      "learning_rate": 6.11012971528015e-06,
      "loss": 0.2029,
      "step": 4877
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4203645083395608,
      "learning_rate": 6.108685859640766e-06,
      "loss": 0.1849,
      "step": 4878
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.390580931223967,
      "learning_rate": 6.107241906762214e-06,
      "loss": 0.1977,
      "step": 4879
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2815498742854778,
      "learning_rate": 6.105797856771142e-06,
      "loss": 0.1798,
      "step": 4880
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.498794261913591,
      "learning_rate": 6.1043537097941985e-06,
      "loss": 0.4694,
      "step": 4881
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1823384240665977,
      "learning_rate": 6.102909465958049e-06,
      "loss": 0.1411,
      "step": 4882
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2942190814478014,
      "learning_rate": 6.1014651253893586e-06,
      "loss": 0.1837,
      "step": 4883
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4119334068970224,
      "learning_rate": 6.10002068821481e-06,
      "loss": 0.2318,
      "step": 4884
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3442244731072546,
      "learning_rate": 6.0985761545610865e-06,
      "loss": 0.194,
      "step": 4885
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5929058852218976,
      "learning_rate": 6.097131524554885e-06,
      "loss": 0.212,
      "step": 4886
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2776857274437636,
      "learning_rate": 6.095686798322911e-06,
      "loss": 0.1704,
      "step": 4887
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2697920091082806,
      "learning_rate": 6.0942419759918725e-06,
      "loss": 0.1716,
      "step": 4888
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3891011642385884,
      "learning_rate": 6.092797057688496e-06,
      "loss": 0.1856,
      "step": 4889
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4030203586216679,
      "learning_rate": 6.0913520435395045e-06,
      "loss": 0.2146,
      "step": 4890
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.445862331202187,
      "learning_rate": 6.089906933671637e-06,
      "loss": 0.1828,
      "step": 4891
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3167342466582286,
      "learning_rate": 6.088461728211642e-06,
      "loss": 0.1854,
      "step": 4892
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3915754444536448,
      "learning_rate": 6.087016427286272e-06,
      "loss": 0.1889,
      "step": 4893
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3958154683755806,
      "learning_rate": 6.085571031022288e-06,
      "loss": 0.2401,
      "step": 4894
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4295284310181473,
      "learning_rate": 6.084125539546464e-06,
      "loss": 0.1782,
      "step": 4895
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4664580190975267,
      "learning_rate": 6.082679952985577e-06,
      "loss": 0.2031,
      "step": 4896
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.342509676885103,
      "learning_rate": 6.081234271466416e-06,
      "loss": 0.1845,
      "step": 4897
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.269581513461251,
      "learning_rate": 6.079788495115775e-06,
      "loss": 0.182,
      "step": 4898
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.095094370167326,
      "learning_rate": 6.078342624060461e-06,
      "loss": 0.1424,
      "step": 4899
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2755993322556605,
      "learning_rate": 6.076896658427283e-06,
      "loss": 0.1975,
      "step": 4900
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5415542176448662,
      "learning_rate": 6.075450598343065e-06,
      "loss": 0.218,
      "step": 4901
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3192701209609938,
      "learning_rate": 6.074004443934635e-06,
      "loss": 0.2221,
      "step": 4902
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.41773036273898,
      "learning_rate": 6.072558195328829e-06,
      "loss": 0.1988,
      "step": 4903
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.376699927549922,
      "learning_rate": 6.071111852652496e-06,
      "loss": 0.207,
      "step": 4904
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3366520713839247,
      "learning_rate": 6.0696654160324875e-06,
      "loss": 0.1898,
      "step": 4905
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.226190451475774,
      "learning_rate": 6.068218885595667e-06,
      "loss": 0.1617,
      "step": 4906
    },
    {
      "epoch": 0.45,
      "grad_norm": 22.1882055569329,
      "learning_rate": 6.066772261468903e-06,
      "loss": 0.7153,
      "step": 4907
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.7656801542217515,
      "learning_rate": 6.065325543779075e-06,
      "loss": 0.4956,
      "step": 4908
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4874332479687618,
      "learning_rate": 6.06387873265307e-06,
      "loss": 0.194,
      "step": 4909
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3683704238369165,
      "learning_rate": 6.0624318282177845e-06,
      "loss": 0.1996,
      "step": 4910
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.749403022013599,
      "learning_rate": 6.060984830600118e-06,
      "loss": 0.7908,
      "step": 4911
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3766929941029216,
      "learning_rate": 6.059537739926986e-06,
      "loss": 0.2019,
      "step": 4912
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.741468437864505,
      "learning_rate": 6.058090556325305e-06,
      "loss": 0.1651,
      "step": 4913
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3229335990826872,
      "learning_rate": 6.056643279922005e-06,
      "loss": 0.1774,
      "step": 4914
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.189909861676738,
      "learning_rate": 6.055195910844021e-06,
      "loss": 0.2414,
      "step": 4915
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.08068719453425,
      "learning_rate": 6.053748449218295e-06,
      "loss": 0.6759,
      "step": 4916
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1646469340032493,
      "learning_rate": 6.052300895171782e-06,
      "loss": 0.1586,
      "step": 4917
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2586616801574533,
      "learning_rate": 6.05085324883144e-06,
      "loss": 0.1896,
      "step": 4918
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.309954835712529,
      "learning_rate": 6.049405510324237e-06,
      "loss": 0.6681,
      "step": 4919
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4013160004389484,
      "learning_rate": 6.047957679777153e-06,
      "loss": 0.2362,
      "step": 4920
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4324578119335782,
      "learning_rate": 6.046509757317168e-06,
      "loss": 0.2177,
      "step": 4921
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4534349235947948,
      "learning_rate": 6.045061743071278e-06,
      "loss": 0.1897,
      "step": 4922
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2031955755529327,
      "learning_rate": 6.043613637166481e-06,
      "loss": 0.174,
      "step": 4923
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5979426985637042,
      "learning_rate": 6.042165439729787e-06,
      "loss": 0.2307,
      "step": 4924
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2970875787706504,
      "learning_rate": 6.040717150888211e-06,
      "loss": 0.1883,
      "step": 4925
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4333559215151155,
      "learning_rate": 6.039268770768781e-06,
      "loss": 0.2059,
      "step": 4926
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2774207950222372,
      "learning_rate": 6.037820299498525e-06,
      "loss": 0.1751,
      "step": 4927
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.217079457199542,
      "learning_rate": 6.036371737204486e-06,
      "loss": 0.1948,
      "step": 4928
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4082648784390883,
      "learning_rate": 6.034923084013713e-06,
      "loss": 0.2384,
      "step": 4929
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.452405141359888,
      "learning_rate": 6.033474340053263e-06,
      "loss": 0.2341,
      "step": 4930
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4774562460646288,
      "learning_rate": 6.0320255054501985e-06,
      "loss": 0.2163,
      "step": 4931
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.206737359847539,
      "learning_rate": 6.030576580331595e-06,
      "loss": 0.1691,
      "step": 4932
    },
    {
      "epoch": 0.45,
      "grad_norm": 13.42211216615921,
      "learning_rate": 6.02912756482453e-06,
      "loss": 0.5738,
      "step": 4933
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4013037335284244,
      "learning_rate": 6.027678459056092e-06,
      "loss": 0.2052,
      "step": 4934
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4908281402450165,
      "learning_rate": 6.026229263153379e-06,
      "loss": 0.256,
      "step": 4935
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5678749586851284,
      "learning_rate": 6.024779977243495e-06,
      "loss": 0.2125,
      "step": 4936
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.481241482268394,
      "learning_rate": 6.0233306014535505e-06,
      "loss": 0.2487,
      "step": 4937
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5349506106268511,
      "learning_rate": 6.021881135910667e-06,
      "loss": 0.2362,
      "step": 4938
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4565575780989086,
      "learning_rate": 6.020431580741971e-06,
      "loss": 0.2301,
      "step": 4939
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2708537775321664,
      "learning_rate": 6.018981936074601e-06,
      "loss": 0.222,
      "step": 4940
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3337219646099086,
      "learning_rate": 6.017532202035697e-06,
      "loss": 0.1678,
      "step": 4941
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4818364776610677,
      "learning_rate": 6.016082378752411e-06,
      "loss": 0.2017,
      "step": 4942
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4585108131919187,
      "learning_rate": 6.014632466351904e-06,
      "loss": 0.1902,
      "step": 4943
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3567018987328476,
      "learning_rate": 6.013182464961341e-06,
      "loss": 0.2038,
      "step": 4944
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.860720819091109,
      "learning_rate": 6.0117323747079e-06,
      "loss": 0.679,
      "step": 4945
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.411988649690339,
      "learning_rate": 6.01028219571876e-06,
      "loss": 0.2058,
      "step": 4946
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6571488855083254,
      "learning_rate": 6.008831928121113e-06,
      "loss": 0.1973,
      "step": 4947
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4652585038121833,
      "learning_rate": 6.007381572042157e-06,
      "loss": 0.2354,
      "step": 4948
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.19543288374418,
      "learning_rate": 6.005931127609099e-06,
      "loss": 0.1645,
      "step": 4949
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3566613305115274,
      "learning_rate": 6.00448059494915e-06,
      "loss": 0.1872,
      "step": 4950
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3379821614158354,
      "learning_rate": 6.003029974189535e-06,
      "loss": 0.1707,
      "step": 4951
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.481446862930584,
      "learning_rate": 6.00157926545748e-06,
      "loss": 0.1977,
      "step": 4952
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.489566697963104,
      "learning_rate": 6.000128468880223e-06,
      "loss": 0.2224,
      "step": 4953
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3091915982164242,
      "learning_rate": 5.998677584585009e-06,
      "loss": 0.2021,
      "step": 4954
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3032975016934327,
      "learning_rate": 5.99722661269909e-06,
      "loss": 0.1624,
      "step": 4955
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3247711816977459,
      "learning_rate": 5.9957755533497255e-06,
      "loss": 0.1763,
      "step": 4956
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3541063192692093,
      "learning_rate": 5.994324406664184e-06,
      "loss": 0.2035,
      "step": 4957
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2077862515214086,
      "learning_rate": 5.99287317276974e-06,
      "loss": 0.1822,
      "step": 4958
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.302026846970816,
      "learning_rate": 5.991421851793677e-06,
      "loss": 0.1953,
      "step": 4959
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2763521427509459,
      "learning_rate": 5.989970443863285e-06,
      "loss": 0.2144,
      "step": 4960
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.267319694338375,
      "learning_rate": 5.988518949105862e-06,
      "loss": 0.1729,
      "step": 4961
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3118349746454672,
      "learning_rate": 5.9870673676487134e-06,
      "loss": 0.1929,
      "step": 4962
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.309217579667421,
      "learning_rate": 5.985615699619155e-06,
      "loss": 0.1562,
      "step": 4963
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3173575209149657,
      "learning_rate": 5.984163945144506e-06,
      "loss": 0.1927,
      "step": 4964
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.908258452106244,
      "learning_rate": 5.982712104352094e-06,
      "loss": 0.6657,
      "step": 4965
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.377256259954862,
      "learning_rate": 5.9812601773692594e-06,
      "loss": 0.1983,
      "step": 4966
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1239595339470358,
      "learning_rate": 5.97980816432334e-06,
      "loss": 0.1437,
      "step": 4967
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.013023805083497,
      "learning_rate": 5.978356065341691e-06,
      "loss": 0.4971,
      "step": 4968
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.239680685072475,
      "learning_rate": 5.976903880551669e-06,
      "loss": 0.7277,
      "step": 4969
    },
    {
      "epoch": 0.45,
      "grad_norm": 9.391433828978892,
      "learning_rate": 5.975451610080643e-06,
      "loss": 0.5508,
      "step": 4970
    },
    {
      "epoch": 0.45,
      "grad_norm": 8.080939300256977,
      "learning_rate": 5.973999254055983e-06,
      "loss": 0.6939,
      "step": 4971
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.262515824083976,
      "learning_rate": 5.9725468126050746e-06,
      "loss": 0.1832,
      "step": 4972
    },
    {
      "epoch": 0.45,
      "grad_norm": 6.19490856006001,
      "learning_rate": 5.971094285855303e-06,
      "loss": 0.663,
      "step": 4973
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2085037232669391,
      "learning_rate": 5.9696416739340676e-06,
      "loss": 0.1543,
      "step": 4974
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.446779353871554,
      "learning_rate": 5.968188976968772e-06,
      "loss": 0.2584,
      "step": 4975
    },
    {
      "epoch": 0.45,
      "grad_norm": 5.874987251103141,
      "learning_rate": 5.966736195086824e-06,
      "loss": 0.6331,
      "step": 4976
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4530462281703413,
      "learning_rate": 5.965283328415644e-06,
      "loss": 0.2162,
      "step": 4977
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5276433947295383,
      "learning_rate": 5.96383037708266e-06,
      "loss": 0.2497,
      "step": 4978
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.305268268290392,
      "learning_rate": 5.9623773412153045e-06,
      "loss": 0.5431,
      "step": 4979
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2305973436398827,
      "learning_rate": 5.9609242209410165e-06,
      "loss": 0.1765,
      "step": 4980
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.149717068448712,
      "learning_rate": 5.959471016387246e-06,
      "loss": 0.6375,
      "step": 4981
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.944607165332889,
      "learning_rate": 5.95801772768145e-06,
      "loss": 0.4381,
      "step": 4982
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4462535230676186,
      "learning_rate": 5.956564354951091e-06,
      "loss": 0.2124,
      "step": 4983
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.403250136376451,
      "learning_rate": 5.955110898323637e-06,
      "loss": 0.1817,
      "step": 4984
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4448091960328593,
      "learning_rate": 5.953657357926569e-06,
      "loss": 0.2039,
      "step": 4985
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4980154934777157,
      "learning_rate": 5.952203733887371e-06,
      "loss": 0.2124,
      "step": 4986
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.140261433765539,
      "learning_rate": 5.950750026333534e-06,
      "loss": 0.6269,
      "step": 4987
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3007541175897281,
      "learning_rate": 5.94929623539256e-06,
      "loss": 0.2085,
      "step": 4988
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2672227037222255,
      "learning_rate": 5.947842361191957e-06,
      "loss": 0.1828,
      "step": 4989
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4963763818437434,
      "learning_rate": 5.946388403859236e-06,
      "loss": 0.2441,
      "step": 4990
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.538529108500912,
      "learning_rate": 5.944934363521923e-06,
      "loss": 0.5688,
      "step": 4991
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2514428395694461,
      "learning_rate": 5.943480240307544e-06,
      "loss": 0.1964,
      "step": 4992
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2173806685309296,
      "learning_rate": 5.942026034343636e-06,
      "loss": 0.1472,
      "step": 4993
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5523121894940983,
      "learning_rate": 5.940571745757743e-06,
      "loss": 0.2502,
      "step": 4994
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2908381665561715,
      "learning_rate": 5.939117374677416e-06,
      "loss": 0.225,
      "step": 4995
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2276465025846657,
      "learning_rate": 5.937662921230212e-06,
      "loss": 0.1841,
      "step": 4996
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3122790586817752,
      "learning_rate": 5.936208385543698e-06,
      "loss": 0.1776,
      "step": 4997
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3607888017879548,
      "learning_rate": 5.934753767745445e-06,
      "loss": 0.203,
      "step": 4998
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4647256816898064,
      "learning_rate": 5.933299067963033e-06,
      "loss": 0.1892,
      "step": 4999
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4897976503741854,
      "learning_rate": 5.931844286324049e-06,
      "loss": 0.1933,
      "step": 5000
    },
    {
      "epoch": 0.46,
      "grad_norm": 7.064310351122346,
      "learning_rate": 5.930389422956088e-06,
      "loss": 0.5449,
      "step": 5001
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5113843189215046,
      "learning_rate": 5.928934477986752e-06,
      "loss": 0.207,
      "step": 5002
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5156735207635055,
      "learning_rate": 5.927479451543647e-06,
      "loss": 0.1946,
      "step": 5003
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3578512422140578,
      "learning_rate": 5.9260243437543874e-06,
      "loss": 0.2118,
      "step": 5004
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.36357408922126,
      "learning_rate": 5.9245691547466e-06,
      "loss": 0.6247,
      "step": 5005
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4291280577659995,
      "learning_rate": 5.923113884647912e-06,
      "loss": 0.2318,
      "step": 5006
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.445031996816869,
      "learning_rate": 5.92165853358596e-06,
      "loss": 0.211,
      "step": 5007
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.453321145433299,
      "learning_rate": 5.92020310168839e-06,
      "loss": 0.1926,
      "step": 5008
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5002901706954348,
      "learning_rate": 5.918747589082853e-06,
      "loss": 0.1869,
      "step": 5009
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4636464194809131,
      "learning_rate": 5.917291995897005e-06,
      "loss": 0.2216,
      "step": 5010
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.356649909729755,
      "learning_rate": 5.915836322258512e-06,
      "loss": 0.1911,
      "step": 5011
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.369957886511173,
      "learning_rate": 5.914380568295048e-06,
      "loss": 0.2385,
      "step": 5012
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3534648270909655,
      "learning_rate": 5.912924734134291e-06,
      "loss": 0.1964,
      "step": 5013
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3624989660850384,
      "learning_rate": 5.911468819903927e-06,
      "loss": 0.1549,
      "step": 5014
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3837318244239503,
      "learning_rate": 5.9100128257316505e-06,
      "loss": 0.2323,
      "step": 5015
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4863118369396555,
      "learning_rate": 5.90855675174516e-06,
      "loss": 0.2406,
      "step": 5016
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.378469145988232,
      "learning_rate": 5.907100598072166e-06,
      "loss": 0.2364,
      "step": 5017
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3498486523568258,
      "learning_rate": 5.9056443648403815e-06,
      "loss": 0.1841,
      "step": 5018
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2941327366586726,
      "learning_rate": 5.904188052177527e-06,
      "loss": 0.201,
      "step": 5019
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0370663374573716,
      "learning_rate": 5.902731660211332e-06,
      "loss": 0.119,
      "step": 5020
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2637445969951113,
      "learning_rate": 5.90127518906953e-06,
      "loss": 0.1894,
      "step": 5021
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.450918663648108,
      "learning_rate": 5.899818638879865e-06,
      "loss": 0.5448,
      "step": 5022
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.740166251412088,
      "learning_rate": 5.898362009770087e-06,
      "loss": 0.7449,
      "step": 5023
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3751456390345513,
      "learning_rate": 5.896905301867949e-06,
      "loss": 0.1863,
      "step": 5024
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2990998959475473,
      "learning_rate": 5.895448515301218e-06,
      "loss": 0.2012,
      "step": 5025
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4814245674771047,
      "learning_rate": 5.893991650197661e-06,
      "loss": 0.1855,
      "step": 5026
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4236251960810924,
      "learning_rate": 5.892534706685056e-06,
      "loss": 0.2283,
      "step": 5027
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2124472405345608,
      "learning_rate": 5.8910776848911856e-06,
      "loss": 0.2083,
      "step": 5028
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.547654964877837,
      "learning_rate": 5.889620584943842e-06,
      "loss": 0.5165,
      "step": 5029
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3034231699883523,
      "learning_rate": 5.88816340697082e-06,
      "loss": 0.1774,
      "step": 5030
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3547415042204254,
      "learning_rate": 5.886706151099927e-06,
      "loss": 0.1961,
      "step": 5031
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2182711304015943,
      "learning_rate": 5.885248817458973e-06,
      "loss": 0.1612,
      "step": 5032
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3266392221497252,
      "learning_rate": 5.883791406175775e-06,
      "loss": 0.1919,
      "step": 5033
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4653387484613551,
      "learning_rate": 5.8823339173781575e-06,
      "loss": 0.2524,
      "step": 5034
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.264149527954371,
      "learning_rate": 5.880876351193956e-06,
      "loss": 0.181,
      "step": 5035
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.955447714417884,
      "learning_rate": 5.879418707751003e-06,
      "loss": 0.6733,
      "step": 5036
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4177776586608173,
      "learning_rate": 5.877960987177147e-06,
      "loss": 0.2228,
      "step": 5037
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.260180743675449,
      "learning_rate": 5.876503189600239e-06,
      "loss": 0.1982,
      "step": 5038
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.380352168494488,
      "learning_rate": 5.8750453151481366e-06,
      "loss": 0.2314,
      "step": 5039
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1988624323981953,
      "learning_rate": 5.873587363948707e-06,
      "loss": 0.1363,
      "step": 5040
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.410261378547138,
      "learning_rate": 5.872129336129821e-06,
      "loss": 0.2083,
      "step": 5041
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.357866924210142,
      "learning_rate": 5.8706712318193594e-06,
      "loss": 0.5671,
      "step": 5042
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5006280322048078,
      "learning_rate": 5.869213051145205e-06,
      "loss": 0.2086,
      "step": 5043
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1780870680580733,
      "learning_rate": 5.867754794235253e-06,
      "loss": 0.1775,
      "step": 5044
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.073000421732859,
      "learning_rate": 5.866296461217399e-06,
      "loss": 0.6173,
      "step": 5045
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4510413023077084,
      "learning_rate": 5.864838052219549e-06,
      "loss": 0.2157,
      "step": 5046
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.471858981488025,
      "learning_rate": 5.863379567369618e-06,
      "loss": 0.2266,
      "step": 5047
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.502983097335393,
      "learning_rate": 5.861921006795522e-06,
      "loss": 0.2388,
      "step": 5048
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3644220833316816,
      "learning_rate": 5.860462370625189e-06,
      "loss": 0.2188,
      "step": 5049
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.505968751841777,
      "learning_rate": 5.859003658986547e-06,
      "loss": 0.2311,
      "step": 5050
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3348964778272017,
      "learning_rate": 5.85754487200754e-06,
      "loss": 0.2219,
      "step": 5051
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4734616285149689,
      "learning_rate": 5.856086009816111e-06,
      "loss": 0.2427,
      "step": 5052
    },
    {
      "epoch": 0.46,
      "grad_norm": 5.441524291907658,
      "learning_rate": 5.854627072540211e-06,
      "loss": 0.6914,
      "step": 5053
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.324674438085143,
      "learning_rate": 5.853168060307799e-06,
      "loss": 0.596,
      "step": 5054
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3917747338496733,
      "learning_rate": 5.851708973246841e-06,
      "loss": 0.2244,
      "step": 5055
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3070402143504103,
      "learning_rate": 5.850249811485309e-06,
      "loss": 0.1747,
      "step": 5056
    },
    {
      "epoch": 0.46,
      "grad_norm": 6.2633710726373435,
      "learning_rate": 5.848790575151181e-06,
      "loss": 0.5298,
      "step": 5057
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3080795294552252,
      "learning_rate": 5.84733126437244e-06,
      "loss": 0.1879,
      "step": 5058
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.231927424350263,
      "learning_rate": 5.845871879277081e-06,
      "loss": 0.188,
      "step": 5059
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3195464216725301,
      "learning_rate": 5.844412419993098e-06,
      "loss": 0.1905,
      "step": 5060
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.457254088690117,
      "learning_rate": 5.842952886648496e-06,
      "loss": 0.1739,
      "step": 5061
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.309932932704829,
      "learning_rate": 5.841493279371287e-06,
      "loss": 0.1882,
      "step": 5062
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.09353703172549,
      "learning_rate": 5.84003359828949e-06,
      "loss": 0.4963,
      "step": 5063
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3232132701833723,
      "learning_rate": 5.838573843531125e-06,
      "loss": 0.191,
      "step": 5064
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2893667175302908,
      "learning_rate": 5.837114015224223e-06,
      "loss": 0.1875,
      "step": 5065
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.200821552638944,
      "learning_rate": 5.8356541134968235e-06,
      "loss": 0.1719,
      "step": 5066
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4891856168958328,
      "learning_rate": 5.834194138476968e-06,
      "loss": 0.1928,
      "step": 5067
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.435710006636725,
      "learning_rate": 5.8327340902927065e-06,
      "loss": 0.2156,
      "step": 5068
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4114795815021255,
      "learning_rate": 5.831273969072093e-06,
      "loss": 0.2114,
      "step": 5069
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1892658174201731,
      "learning_rate": 5.829813774943191e-06,
      "loss": 0.1604,
      "step": 5070
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3433272814087438,
      "learning_rate": 5.828353508034071e-06,
      "loss": 0.2093,
      "step": 5071
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.369302431025333,
      "learning_rate": 5.826893168472807e-06,
      "loss": 0.2261,
      "step": 5072
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3016601244657389,
      "learning_rate": 5.8254327563874794e-06,
      "loss": 0.2032,
      "step": 5073
    },
    {
      "epoch": 0.46,
      "grad_norm": 8.257303989911657,
      "learning_rate": 5.823972271906177e-06,
      "loss": 0.6091,
      "step": 5074
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5323894229672177,
      "learning_rate": 5.822511715156996e-06,
      "loss": 0.1964,
      "step": 5075
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3743283036120884,
      "learning_rate": 5.821051086268033e-06,
      "loss": 0.1766,
      "step": 5076
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3593008659151056,
      "learning_rate": 5.8195903853674e-06,
      "loss": 0.1636,
      "step": 5077
    },
    {
      "epoch": 0.46,
      "grad_norm": 4.8092505568744155,
      "learning_rate": 5.818129612583204e-06,
      "loss": 0.4367,
      "step": 5078
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3578847122462847,
      "learning_rate": 5.81666876804357e-06,
      "loss": 0.2135,
      "step": 5079
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3162950852610211,
      "learning_rate": 5.815207851876621e-06,
      "loss": 0.1869,
      "step": 5080
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3650154700958708,
      "learning_rate": 5.813746864210489e-06,
      "loss": 0.2063,
      "step": 5081
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4878657180151544,
      "learning_rate": 5.812285805173314e-06,
      "loss": 0.202,
      "step": 5082
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3010270868202818,
      "learning_rate": 5.810824674893241e-06,
      "loss": 0.1964,
      "step": 5083
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3164682202049904,
      "learning_rate": 5.809363473498417e-06,
      "loss": 0.1907,
      "step": 5084
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3113902630844614,
      "learning_rate": 5.807902201117006e-06,
      "loss": 0.1752,
      "step": 5085
    },
    {
      "epoch": 0.47,
      "grad_norm": 7.476771645375195,
      "learning_rate": 5.806440857877165e-06,
      "loss": 0.729,
      "step": 5086
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4641070052139518,
      "learning_rate": 5.804979443907065e-06,
      "loss": 0.1772,
      "step": 5087
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3631598772482698,
      "learning_rate": 5.803517959334883e-06,
      "loss": 0.1961,
      "step": 5088
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.157267421728568,
      "learning_rate": 5.8020564042888015e-06,
      "loss": 0.1384,
      "step": 5089
    },
    {
      "epoch": 0.47,
      "grad_norm": 10.134735107429817,
      "learning_rate": 5.800594778897007e-06,
      "loss": 0.6809,
      "step": 5090
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.305357620138046,
      "learning_rate": 5.7991330832876955e-06,
      "loss": 0.1821,
      "step": 5091
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3764847414423993,
      "learning_rate": 5.797671317589066e-06,
      "loss": 0.2419,
      "step": 5092
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.519316833396908,
      "learning_rate": 5.796209481929327e-06,
      "loss": 0.171,
      "step": 5093
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.414398291930829,
      "learning_rate": 5.794747576436689e-06,
      "loss": 0.2207,
      "step": 5094
    },
    {
      "epoch": 0.47,
      "grad_norm": 8.45783858819554,
      "learning_rate": 5.793285601239371e-06,
      "loss": 0.5699,
      "step": 5095
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6716497555511178,
      "learning_rate": 5.7918235564656e-06,
      "loss": 0.2277,
      "step": 5096
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3998130647853138,
      "learning_rate": 5.790361442243605e-06,
      "loss": 0.178,
      "step": 5097
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2504166985015905,
      "learning_rate": 5.788899258701625e-06,
      "loss": 0.1946,
      "step": 5098
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4342013687158761,
      "learning_rate": 5.787437005967902e-06,
      "loss": 0.2047,
      "step": 5099
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.314293031246258,
      "learning_rate": 5.785974684170685e-06,
      "loss": 0.2005,
      "step": 5100
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5703561868547866,
      "learning_rate": 5.784512293438233e-06,
      "loss": 0.193,
      "step": 5101
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5367326210449235,
      "learning_rate": 5.783049833898803e-06,
      "loss": 0.2224,
      "step": 5102
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.609111950688373,
      "learning_rate": 5.781587305680664e-06,
      "loss": 0.2583,
      "step": 5103
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.0675337750712774,
      "learning_rate": 5.7801247089120895e-06,
      "loss": 0.6229,
      "step": 5104
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.383984134097282,
      "learning_rate": 5.778662043721359e-06,
      "loss": 0.1935,
      "step": 5105
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.425508764738369,
      "learning_rate": 5.777199310236758e-06,
      "loss": 0.562,
      "step": 5106
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3569357763554222,
      "learning_rate": 5.77573650858658e-06,
      "loss": 0.1728,
      "step": 5107
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2254683071753514,
      "learning_rate": 5.77427363889912e-06,
      "loss": 0.1941,
      "step": 5108
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3296392845467049,
      "learning_rate": 5.772810701302681e-06,
      "loss": 0.1998,
      "step": 5109
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.328467314705122,
      "learning_rate": 5.771347695925576e-06,
      "loss": 0.2117,
      "step": 5110
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.276972709300457,
      "learning_rate": 5.769884622896116e-06,
      "loss": 0.1782,
      "step": 5111
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5882081658141007,
      "learning_rate": 5.7684214823426235e-06,
      "loss": 0.2133,
      "step": 5112
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2995340886170004,
      "learning_rate": 5.766958274393428e-06,
      "loss": 0.1736,
      "step": 5113
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.32498275454034,
      "learning_rate": 5.765494999176861e-06,
      "loss": 0.1781,
      "step": 5114
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4172252609517646,
      "learning_rate": 5.764031656821261e-06,
      "loss": 0.2298,
      "step": 5115
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3986775161531955,
      "learning_rate": 5.762568247454973e-06,
      "loss": 0.2004,
      "step": 5116
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4267929303121278,
      "learning_rate": 5.7611047712063485e-06,
      "loss": 0.209,
      "step": 5117
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.6332425859210833,
      "learning_rate": 5.7596412282037435e-06,
      "loss": 0.3972,
      "step": 5118
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.355513844025024,
      "learning_rate": 5.758177618575523e-06,
      "loss": 0.1813,
      "step": 5119
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4878895207961338,
      "learning_rate": 5.756713942450053e-06,
      "loss": 0.2139,
      "step": 5120
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1920910890910186,
      "learning_rate": 5.7552501999557065e-06,
      "loss": 0.1804,
      "step": 5121
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4392421539998717,
      "learning_rate": 5.753786391220866e-06,
      "loss": 0.2328,
      "step": 5122
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3762089626114251,
      "learning_rate": 5.752322516373916e-06,
      "loss": 0.1962,
      "step": 5123
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.398326261453707,
      "learning_rate": 5.750858575543248e-06,
      "loss": 0.1807,
      "step": 5124
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3608138046306668,
      "learning_rate": 5.74939456885726e-06,
      "loss": 0.2115,
      "step": 5125
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4944981869986267,
      "learning_rate": 5.747930496444356e-06,
      "loss": 0.2023,
      "step": 5126
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3231843367549687,
      "learning_rate": 5.746466358432942e-06,
      "loss": 0.2132,
      "step": 5127
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.320195451845588,
      "learning_rate": 5.745002154951438e-06,
      "loss": 0.2099,
      "step": 5128
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.524526139192832,
      "learning_rate": 5.743537886128258e-06,
      "loss": 0.1626,
      "step": 5129
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3093934270575645,
      "learning_rate": 5.742073552091834e-06,
      "loss": 0.1983,
      "step": 5130
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4340598253503887,
      "learning_rate": 5.740609152970593e-06,
      "loss": 0.2216,
      "step": 5131
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.32732747911828,
      "learning_rate": 5.739144688892976e-06,
      "loss": 0.1597,
      "step": 5132
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3274486962637397,
      "learning_rate": 5.737680159987424e-06,
      "loss": 0.2116,
      "step": 5133
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4637326956882355,
      "learning_rate": 5.736215566382388e-06,
      "loss": 0.233,
      "step": 5134
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3723609932732739,
      "learning_rate": 5.734750908206321e-06,
      "loss": 0.188,
      "step": 5135
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4292621080735568,
      "learning_rate": 5.733286185587685e-06,
      "loss": 0.1958,
      "step": 5136
    },
    {
      "epoch": 0.47,
      "grad_norm": 11.523566147152643,
      "learning_rate": 5.731821398654944e-06,
      "loss": 0.5913,
      "step": 5137
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.595136204555328,
      "learning_rate": 5.730356547536572e-06,
      "loss": 0.7499,
      "step": 5138
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.606587866914565,
      "learning_rate": 5.728891632361043e-06,
      "loss": 0.2316,
      "step": 5139
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4260229425927222,
      "learning_rate": 5.727426653256844e-06,
      "loss": 0.2117,
      "step": 5140
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.345898132223914,
      "learning_rate": 5.725961610352458e-06,
      "loss": 0.1674,
      "step": 5141
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1494635754776228,
      "learning_rate": 5.724496503776384e-06,
      "loss": 0.1852,
      "step": 5142
    },
    {
      "epoch": 0.47,
      "grad_norm": 9.495640508773585,
      "learning_rate": 5.723031333657121e-06,
      "loss": 0.6471,
      "step": 5143
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5470619048551522,
      "learning_rate": 5.721566100123171e-06,
      "loss": 0.2358,
      "step": 5144
    },
    {
      "epoch": 0.47,
      "grad_norm": 6.121256313922405,
      "learning_rate": 5.72010080330305e-06,
      "loss": 0.5378,
      "step": 5145
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4791785107843398,
      "learning_rate": 5.718635443325269e-06,
      "loss": 0.1852,
      "step": 5146
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4800585471750372,
      "learning_rate": 5.7171700203183524e-06,
      "loss": 0.2204,
      "step": 5147
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4236440228600078,
      "learning_rate": 5.715704534410828e-06,
      "loss": 0.2208,
      "step": 5148
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2672330359173989,
      "learning_rate": 5.714238985731226e-06,
      "loss": 0.199,
      "step": 5149
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3074565100346314,
      "learning_rate": 5.712773374408089e-06,
      "loss": 0.1717,
      "step": 5150
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.257191638253504,
      "learning_rate": 5.711307700569958e-06,
      "loss": 0.225,
      "step": 5151
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3993866043877994,
      "learning_rate": 5.709841964345383e-06,
      "loss": 0.2379,
      "step": 5152
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4080747850298074,
      "learning_rate": 5.708376165862921e-06,
      "loss": 0.2246,
      "step": 5153
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.00132243702012,
      "learning_rate": 5.7069103052511274e-06,
      "loss": 0.5647,
      "step": 5154
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2917750457309196,
      "learning_rate": 5.705444382638573e-06,
      "loss": 0.1775,
      "step": 5155
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.222021752496229,
      "learning_rate": 5.703978398153827e-06,
      "loss": 0.1905,
      "step": 5156
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.312098924818084,
      "learning_rate": 5.7025123519254644e-06,
      "loss": 0.2244,
      "step": 5157
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4841447148179785,
      "learning_rate": 5.701046244082069e-06,
      "loss": 0.2286,
      "step": 5158
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.356710803593515,
      "learning_rate": 5.699580074752229e-06,
      "loss": 0.1855,
      "step": 5159
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3089624190970413,
      "learning_rate": 5.698113844064537e-06,
      "loss": 0.2269,
      "step": 5160
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.500437016270642,
      "learning_rate": 5.696647552147589e-06,
      "loss": 0.2039,
      "step": 5161
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5205727662066897,
      "learning_rate": 5.6951811991299935e-06,
      "loss": 0.1917,
      "step": 5162
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3354744784186883,
      "learning_rate": 5.693714785140354e-06,
      "loss": 0.1782,
      "step": 5163
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3804411508280034,
      "learning_rate": 5.692248310307287e-06,
      "loss": 0.1939,
      "step": 5164
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.312961217400101,
      "learning_rate": 5.690781774759412e-06,
      "loss": 0.1852,
      "step": 5165
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.292568640616332,
      "learning_rate": 5.689315178625354e-06,
      "loss": 0.2035,
      "step": 5166
    },
    {
      "epoch": 0.47,
      "grad_norm": 4.5572983061563175,
      "learning_rate": 5.6878485220337446e-06,
      "loss": 0.5701,
      "step": 5167
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4203342812927293,
      "learning_rate": 5.686381805113217e-06,
      "loss": 0.2012,
      "step": 5168
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.842821158609528,
      "learning_rate": 5.684915027992415e-06,
      "loss": 0.5293,
      "step": 5169
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3702467848403708,
      "learning_rate": 5.683448190799981e-06,
      "loss": 0.2334,
      "step": 5170
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3734330954748837,
      "learning_rate": 5.681981293664572e-06,
      "loss": 0.2134,
      "step": 5171
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3906392190126442,
      "learning_rate": 5.680514336714839e-06,
      "loss": 0.2274,
      "step": 5172
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3841605282094853,
      "learning_rate": 5.679047320079445e-06,
      "loss": 0.1959,
      "step": 5173
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3534034895529612,
      "learning_rate": 5.6775802438870596e-06,
      "loss": 0.1667,
      "step": 5174
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.5289566722675465,
      "learning_rate": 5.676113108266355e-06,
      "loss": 0.6666,
      "step": 5175
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3869680162974145,
      "learning_rate": 5.674645913346005e-06,
      "loss": 0.1964,
      "step": 5176
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4934099734413522,
      "learning_rate": 5.673178659254698e-06,
      "loss": 0.2301,
      "step": 5177
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4429039142515978,
      "learning_rate": 5.671711346121117e-06,
      "loss": 0.1796,
      "step": 5178
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.385313286084446,
      "learning_rate": 5.67024397407396e-06,
      "loss": 0.1925,
      "step": 5179
    },
    {
      "epoch": 0.47,
      "grad_norm": 12.241932195422258,
      "learning_rate": 5.66877654324192e-06,
      "loss": 0.523,
      "step": 5180
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5830785813502202,
      "learning_rate": 5.667309053753706e-06,
      "loss": 0.2517,
      "step": 5181
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2940072975587642,
      "learning_rate": 5.665841505738021e-06,
      "loss": 0.1883,
      "step": 5182
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4647097132680469,
      "learning_rate": 5.664373899323583e-06,
      "loss": 0.2457,
      "step": 5183
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3210368814796831,
      "learning_rate": 5.662906234639109e-06,
      "loss": 0.1605,
      "step": 5184
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2390081608099193,
      "learning_rate": 5.661438511813324e-06,
      "loss": 0.178,
      "step": 5185
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4271674481699972,
      "learning_rate": 5.6599707309749565e-06,
      "loss": 0.2049,
      "step": 5186
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4314170565057291,
      "learning_rate": 5.658502892252741e-06,
      "loss": 0.2214,
      "step": 5187
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4336721407116793,
      "learning_rate": 5.657034995775418e-06,
      "loss": 0.2207,
      "step": 5188
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.334687116668403,
      "learning_rate": 5.6555670416717275e-06,
      "loss": 0.2155,
      "step": 5189
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2355429963727769,
      "learning_rate": 5.654099030070424e-06,
      "loss": 0.1769,
      "step": 5190
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2458153398463976,
      "learning_rate": 5.65263096110026e-06,
      "loss": 0.1845,
      "step": 5191
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4322137518330236,
      "learning_rate": 5.651162834889993e-06,
      "loss": 0.2069,
      "step": 5192
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2708641243584775,
      "learning_rate": 5.64969465156839e-06,
      "loss": 0.1875,
      "step": 5193
    },
    {
      "epoch": 0.47,
      "grad_norm": 5.053880737556917,
      "learning_rate": 5.648226411264219e-06,
      "loss": 0.6705,
      "step": 5194
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4190750817021602,
      "learning_rate": 5.646758114106256e-06,
      "loss": 0.1657,
      "step": 5195
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4549979602868857,
      "learning_rate": 5.645289760223281e-06,
      "loss": 0.243,
      "step": 5196
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.092288455788529,
      "learning_rate": 5.643821349744076e-06,
      "loss": 0.7192,
      "step": 5197
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2318360808870463,
      "learning_rate": 5.642352882797433e-06,
      "loss": 0.1943,
      "step": 5198
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3118116665864288,
      "learning_rate": 5.640884359512143e-06,
      "loss": 0.2016,
      "step": 5199
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3461121896648585,
      "learning_rate": 5.639415780017008e-06,
      "loss": 0.1576,
      "step": 5200
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3127446090420214,
      "learning_rate": 5.637947144440832e-06,
      "loss": 0.2077,
      "step": 5201
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4822485315392635,
      "learning_rate": 5.6364784529124225e-06,
      "loss": 0.227,
      "step": 5202
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3087661624279712,
      "learning_rate": 5.635009705560595e-06,
      "loss": 0.2004,
      "step": 5203
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.327157669758748,
      "learning_rate": 5.63354090251417e-06,
      "loss": 0.2206,
      "step": 5204
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2412950601585215,
      "learning_rate": 5.632072043901969e-06,
      "loss": 0.1947,
      "step": 5205
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3482974082878367,
      "learning_rate": 5.6306031298528205e-06,
      "loss": 0.2306,
      "step": 5206
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3158092334042923,
      "learning_rate": 5.6291341604955594e-06,
      "loss": 0.2024,
      "step": 5207
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.457958017321023,
      "learning_rate": 5.627665135959024e-06,
      "loss": 0.2338,
      "step": 5208
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4346972275560732,
      "learning_rate": 5.626196056372056e-06,
      "loss": 0.2589,
      "step": 5209
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4066220876432791,
      "learning_rate": 5.624726921863506e-06,
      "loss": 0.2203,
      "step": 5210
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5103651460275962,
      "learning_rate": 5.623257732562225e-06,
      "loss": 0.2287,
      "step": 5211
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.211764558490554,
      "learning_rate": 5.621788488597071e-06,
      "loss": 0.6545,
      "step": 5212
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5395898555094487,
      "learning_rate": 5.620319190096908e-06,
      "loss": 0.2008,
      "step": 5213
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4735857305504383,
      "learning_rate": 5.618849837190603e-06,
      "loss": 0.1907,
      "step": 5214
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3721287530320445,
      "learning_rate": 5.6173804300070254e-06,
      "loss": 0.1868,
      "step": 5215
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.375639008103119,
      "learning_rate": 5.615910968675055e-06,
      "loss": 0.2166,
      "step": 5216
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.874584336046937,
      "learning_rate": 5.614441453323571e-06,
      "loss": 0.5784,
      "step": 5217
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.737394246971639,
      "learning_rate": 5.612971884081463e-06,
      "loss": 0.1898,
      "step": 5218
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3380964186446918,
      "learning_rate": 5.611502261077618e-06,
      "loss": 0.2118,
      "step": 5219
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4452683379065026,
      "learning_rate": 5.610032584440937e-06,
      "loss": 0.251,
      "step": 5220
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2363593822510635,
      "learning_rate": 5.608562854300315e-06,
      "loss": 0.1747,
      "step": 5221
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5433957092879145,
      "learning_rate": 5.60709307078466e-06,
      "loss": 0.2177,
      "step": 5222
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.3432647594570994,
      "learning_rate": 5.605623234022881e-06,
      "loss": 0.5385,
      "step": 5223
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.353060900763661,
      "learning_rate": 5.604153344143894e-06,
      "loss": 0.2109,
      "step": 5224
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4449632800953276,
      "learning_rate": 5.6026834012766155e-06,
      "loss": 0.1846,
      "step": 5225
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3126476277332169,
      "learning_rate": 5.601213405549973e-06,
      "loss": 0.1708,
      "step": 5226
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2300879637155435,
      "learning_rate": 5.59974335709289e-06,
      "loss": 0.1858,
      "step": 5227
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4049237870270064,
      "learning_rate": 5.598273256034305e-06,
      "loss": 0.1716,
      "step": 5228
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1999609285144612,
      "learning_rate": 5.596803102503152e-06,
      "loss": 0.1651,
      "step": 5229
    },
    {
      "epoch": 0.48,
      "grad_norm": 10.443418670105356,
      "learning_rate": 5.595332896628375e-06,
      "loss": 0.7556,
      "step": 5230
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6180800068413228,
      "learning_rate": 5.593862638538921e-06,
      "loss": 0.231,
      "step": 5231
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3463383527667205,
      "learning_rate": 5.592392328363739e-06,
      "loss": 0.1628,
      "step": 5232
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.769933261394983,
      "learning_rate": 5.590921966231788e-06,
      "loss": 0.636,
      "step": 5233
    },
    {
      "epoch": 0.48,
      "grad_norm": 22.38077746996907,
      "learning_rate": 5.589451552272027e-06,
      "loss": 0.4938,
      "step": 5234
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.542129956381683,
      "learning_rate": 5.587981086613421e-06,
      "loss": 0.2533,
      "step": 5235
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5488551188306257,
      "learning_rate": 5.586510569384941e-06,
      "loss": 0.2314,
      "step": 5236
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4796623146764427,
      "learning_rate": 5.58504000071556e-06,
      "loss": 0.2386,
      "step": 5237
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2657639809805357,
      "learning_rate": 5.5835693807342595e-06,
      "loss": 0.1786,
      "step": 5238
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.896467797744524,
      "learning_rate": 5.582098709570018e-06,
      "loss": 0.4482,
      "step": 5239
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.490742024525751,
      "learning_rate": 5.5806279873518265e-06,
      "loss": 0.1972,
      "step": 5240
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.718351942919172,
      "learning_rate": 5.579157214208675e-06,
      "loss": 0.2981,
      "step": 5241
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3888157231325275,
      "learning_rate": 5.577686390269562e-06,
      "loss": 0.2065,
      "step": 5242
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4752085099746797,
      "learning_rate": 5.576215515663489e-06,
      "loss": 0.2093,
      "step": 5243
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.372074216177419,
      "learning_rate": 5.5747445905194586e-06,
      "loss": 0.2437,
      "step": 5244
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2758593689017381,
      "learning_rate": 5.573273614966485e-06,
      "loss": 0.16,
      "step": 5245
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5338156546791566,
      "learning_rate": 5.571802589133579e-06,
      "loss": 0.2346,
      "step": 5246
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.601357496696569,
      "learning_rate": 5.5703315131497615e-06,
      "loss": 0.6553,
      "step": 5247
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.437401961857391,
      "learning_rate": 5.568860387144054e-06,
      "loss": 0.2222,
      "step": 5248
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2511718737599742,
      "learning_rate": 5.567389211245486e-06,
      "loss": 0.2034,
      "step": 5249
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2427883302228844,
      "learning_rate": 5.565917985583087e-06,
      "loss": 0.1577,
      "step": 5250
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4583905864032007,
      "learning_rate": 5.564446710285895e-06,
      "loss": 0.1836,
      "step": 5251
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4947281791145115,
      "learning_rate": 5.56297538548295e-06,
      "loss": 0.1725,
      "step": 5252
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3845680852306081,
      "learning_rate": 5.561504011303299e-06,
      "loss": 0.2158,
      "step": 5253
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.331311885528706,
      "learning_rate": 5.5600325878759895e-06,
      "loss": 0.1798,
      "step": 5254
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3006952051623262,
      "learning_rate": 5.558561115330077e-06,
      "loss": 0.1687,
      "step": 5255
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3078816998336933,
      "learning_rate": 5.557089593794616e-06,
      "loss": 0.2174,
      "step": 5256
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3407967035938673,
      "learning_rate": 5.555618023398671e-06,
      "loss": 0.2232,
      "step": 5257
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.307303883318132,
      "learning_rate": 5.554146404271309e-06,
      "loss": 0.2032,
      "step": 5258
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6423956929888064,
      "learning_rate": 5.5526747365416e-06,
      "loss": 0.2395,
      "step": 5259
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.40016281144859,
      "learning_rate": 5.551203020338619e-06,
      "loss": 0.7265,
      "step": 5260
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5779615632615382,
      "learning_rate": 5.549731255791446e-06,
      "loss": 0.2544,
      "step": 5261
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.469935413104567,
      "learning_rate": 5.5482594430291644e-06,
      "loss": 0.1932,
      "step": 5262
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5359861270899666,
      "learning_rate": 5.546787582180864e-06,
      "loss": 0.2234,
      "step": 5263
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3022396499660933,
      "learning_rate": 5.545315673375632e-06,
      "loss": 0.2026,
      "step": 5264
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4162302681190526,
      "learning_rate": 5.5438437167425675e-06,
      "loss": 0.2048,
      "step": 5265
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3718030213371324,
      "learning_rate": 5.5423717124107715e-06,
      "loss": 0.1826,
      "step": 5266
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5158211028958661,
      "learning_rate": 5.540899660509348e-06,
      "loss": 0.2096,
      "step": 5267
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4212778747713184,
      "learning_rate": 5.539427561167405e-06,
      "loss": 0.1901,
      "step": 5268
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4051570121669932,
      "learning_rate": 5.537955414514058e-06,
      "loss": 0.1838,
      "step": 5269
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4345971956463905,
      "learning_rate": 5.53648322067842e-06,
      "loss": 0.2051,
      "step": 5270
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3131277635404546,
      "learning_rate": 5.5350109797896165e-06,
      "loss": 0.2182,
      "step": 5271
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4733955329590132,
      "learning_rate": 5.533538691976772e-06,
      "loss": 0.1839,
      "step": 5272
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2932686410838594,
      "learning_rate": 5.532066357369012e-06,
      "loss": 0.1875,
      "step": 5273
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.555681994028562,
      "learning_rate": 5.530593976095475e-06,
      "loss": 0.5675,
      "step": 5274
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.933226738725018,
      "learning_rate": 5.529121548285297e-06,
      "loss": 0.5175,
      "step": 5275
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4201120449316187,
      "learning_rate": 5.527649074067618e-06,
      "loss": 0.2121,
      "step": 5276
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3302651824500002,
      "learning_rate": 5.526176553571586e-06,
      "loss": 0.1786,
      "step": 5277
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3397116084176344,
      "learning_rate": 5.52470398692635e-06,
      "loss": 0.1989,
      "step": 5278
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.452916374534749,
      "learning_rate": 5.5232313742610644e-06,
      "loss": 0.2609,
      "step": 5279
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.265591204479746,
      "learning_rate": 5.521758715704887e-06,
      "loss": 0.1953,
      "step": 5280
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3052875942817415,
      "learning_rate": 5.52028601138698e-06,
      "loss": 0.1813,
      "step": 5281
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2812737379032901,
      "learning_rate": 5.51881326143651e-06,
      "loss": 0.1774,
      "step": 5282
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3838397782021294,
      "learning_rate": 5.517340465982646e-06,
      "loss": 0.2045,
      "step": 5283
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.848931524298109,
      "learning_rate": 5.51586762515456e-06,
      "loss": 0.7676,
      "step": 5284
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3398313542844345,
      "learning_rate": 5.514394739081434e-06,
      "loss": 0.1985,
      "step": 5285
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4686600221095873,
      "learning_rate": 5.5129218078924485e-06,
      "loss": 0.1481,
      "step": 5286
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.077817152627956,
      "learning_rate": 5.511448831716788e-06,
      "loss": 0.6036,
      "step": 5287
    },
    {
      "epoch": 0.48,
      "grad_norm": 6.866481372961101,
      "learning_rate": 5.509975810683644e-06,
      "loss": 0.6051,
      "step": 5288
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3303846302500493,
      "learning_rate": 5.508502744922212e-06,
      "loss": 0.2063,
      "step": 5289
    },
    {
      "epoch": 0.48,
      "grad_norm": 7.830821314011497,
      "learning_rate": 5.507029634561685e-06,
      "loss": 0.6819,
      "step": 5290
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.898358661487691,
      "learning_rate": 5.505556479731268e-06,
      "loss": 0.4822,
      "step": 5291
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5429984649288848,
      "learning_rate": 5.504083280560165e-06,
      "loss": 0.1751,
      "step": 5292
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4065268187457112,
      "learning_rate": 5.502610037177586e-06,
      "loss": 0.2138,
      "step": 5293
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.329230665651124,
      "learning_rate": 5.501136749712744e-06,
      "loss": 0.1828,
      "step": 5294
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2936271797936714,
      "learning_rate": 5.499663418294858e-06,
      "loss": 0.1948,
      "step": 5295
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.152311420151807,
      "learning_rate": 5.498190043053147e-06,
      "loss": 0.5265,
      "step": 5296
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.593897619557058,
      "learning_rate": 5.496716624116836e-06,
      "loss": 0.2393,
      "step": 5297
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4807647455125637,
      "learning_rate": 5.495243161615156e-06,
      "loss": 0.211,
      "step": 5298
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3877887401229465,
      "learning_rate": 5.493769655677335e-06,
      "loss": 0.2077,
      "step": 5299
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2991508837538903,
      "learning_rate": 5.4922961064326144e-06,
      "loss": 0.2038,
      "step": 5300
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2706299459304768,
      "learning_rate": 5.49082251401023e-06,
      "loss": 0.2072,
      "step": 5301
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1910361729621184,
      "learning_rate": 5.489348878539429e-06,
      "loss": 0.1561,
      "step": 5302
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4655723084795969,
      "learning_rate": 5.487875200149456e-06,
      "loss": 0.159,
      "step": 5303
    },
    {
      "epoch": 0.48,
      "grad_norm": 5.127411940798793,
      "learning_rate": 5.486401478969565e-06,
      "loss": 0.5648,
      "step": 5304
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3405623243222278,
      "learning_rate": 5.484927715129011e-06,
      "loss": 0.213,
      "step": 5305
    },
    {
      "epoch": 0.49,
      "grad_norm": 13.660227380469793,
      "learning_rate": 5.483453908757054e-06,
      "loss": 0.6827,
      "step": 5306
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4376791430236266,
      "learning_rate": 5.481980059982952e-06,
      "loss": 0.2058,
      "step": 5307
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.212317714381318,
      "learning_rate": 5.4805061689359754e-06,
      "loss": 0.1888,
      "step": 5308
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3678737902666696,
      "learning_rate": 5.479032235745394e-06,
      "loss": 0.1629,
      "step": 5309
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.239659389777924,
      "learning_rate": 5.4775582605404795e-06,
      "loss": 0.225,
      "step": 5310
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3872035412020396,
      "learning_rate": 5.476084243450511e-06,
      "loss": 0.1981,
      "step": 5311
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2393216687713704,
      "learning_rate": 5.47461018460477e-06,
      "loss": 0.1756,
      "step": 5312
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4098697687124575,
      "learning_rate": 5.4731360841325405e-06,
      "loss": 0.2324,
      "step": 5313
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2943459926164158,
      "learning_rate": 5.471661942163112e-06,
      "loss": 0.1784,
      "step": 5314
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5410528096904355,
      "learning_rate": 5.470187758825777e-06,
      "loss": 0.221,
      "step": 5315
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4347283339268095,
      "learning_rate": 5.468713534249827e-06,
      "loss": 0.2212,
      "step": 5316
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2970356401226049,
      "learning_rate": 5.467239268564567e-06,
      "loss": 0.1876,
      "step": 5317
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2747738114519616,
      "learning_rate": 5.465764961899296e-06,
      "loss": 0.1975,
      "step": 5318
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4339073431965381,
      "learning_rate": 5.464290614383323e-06,
      "loss": 0.1872,
      "step": 5319
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6484369311634066,
      "learning_rate": 5.462816226145956e-06,
      "loss": 0.2091,
      "step": 5320
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1997331437336456,
      "learning_rate": 5.46134179731651e-06,
      "loss": 0.1545,
      "step": 5321
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3635054242329527,
      "learning_rate": 5.459867328024303e-06,
      "loss": 0.209,
      "step": 5322
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1821899562260594,
      "learning_rate": 5.458392818398655e-06,
      "loss": 0.1748,
      "step": 5323
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4098830327419192,
      "learning_rate": 5.45691826856889e-06,
      "loss": 0.233,
      "step": 5324
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2571789976170358,
      "learning_rate": 5.455443678664336e-06,
      "loss": 0.1814,
      "step": 5325
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.461689067781723,
      "learning_rate": 5.453969048814324e-06,
      "loss": 0.4865,
      "step": 5326
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2936534238067146,
      "learning_rate": 5.45249437914819e-06,
      "loss": 0.1878,
      "step": 5327
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3625964849856171,
      "learning_rate": 5.4510196697952725e-06,
      "loss": 0.2014,
      "step": 5328
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2706113564173684,
      "learning_rate": 5.449544920884912e-06,
      "loss": 0.1694,
      "step": 5329
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.357838815499863,
      "learning_rate": 5.448070132546453e-06,
      "loss": 0.2105,
      "step": 5330
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5740837863963468,
      "learning_rate": 5.446595304909248e-06,
      "loss": 0.198,
      "step": 5331
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6529949101627557,
      "learning_rate": 5.445120438102648e-06,
      "loss": 0.2146,
      "step": 5332
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.509886400350842,
      "learning_rate": 5.443645532256007e-06,
      "loss": 0.2459,
      "step": 5333
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.354716759569837,
      "learning_rate": 5.442170587498684e-06,
      "loss": 0.1867,
      "step": 5334
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4121926966127571,
      "learning_rate": 5.440695603960042e-06,
      "loss": 0.234,
      "step": 5335
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.473196937069528,
      "learning_rate": 5.4392205817694485e-06,
      "loss": 0.1924,
      "step": 5336
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4202091729623538,
      "learning_rate": 5.437745521056272e-06,
      "loss": 0.2065,
      "step": 5337
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4921279586997795,
      "learning_rate": 5.436270421949884e-06,
      "loss": 0.2165,
      "step": 5338
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4334125782709293,
      "learning_rate": 5.434795284579661e-06,
      "loss": 0.2009,
      "step": 5339
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3595671483544334,
      "learning_rate": 5.4333201090749846e-06,
      "loss": 0.2023,
      "step": 5340
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3808425965569966,
      "learning_rate": 5.431844895565236e-06,
      "loss": 0.2277,
      "step": 5341
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4914314350526747,
      "learning_rate": 5.430369644179801e-06,
      "loss": 0.2223,
      "step": 5342
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.693182217970552,
      "learning_rate": 5.428894355048066e-06,
      "loss": 0.4503,
      "step": 5343
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.248611182209059,
      "learning_rate": 5.42741902829943e-06,
      "loss": 0.1752,
      "step": 5344
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.162794264600739,
      "learning_rate": 5.425943664063284e-06,
      "loss": 0.569,
      "step": 5345
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2633093586929578,
      "learning_rate": 5.4244682624690315e-06,
      "loss": 0.1741,
      "step": 5346
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.201677784461721,
      "learning_rate": 5.4229928236460705e-06,
      "loss": 0.1954,
      "step": 5347
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4985856804831026,
      "learning_rate": 5.42151734772381e-06,
      "loss": 0.1999,
      "step": 5348
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.381489742019592,
      "learning_rate": 5.42004183483166e-06,
      "loss": 0.2247,
      "step": 5349
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5419319096679245,
      "learning_rate": 5.41856628509903e-06,
      "loss": 0.2268,
      "step": 5350
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.770613544598515,
      "learning_rate": 5.417090698655335e-06,
      "loss": 0.6463,
      "step": 5351
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.002849593166509,
      "learning_rate": 5.415615075629996e-06,
      "loss": 0.774,
      "step": 5352
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2327348112229062,
      "learning_rate": 5.414139416152435e-06,
      "loss": 0.17,
      "step": 5353
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2295421400767241,
      "learning_rate": 5.412663720352077e-06,
      "loss": 0.1756,
      "step": 5354
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3262229558998424,
      "learning_rate": 5.411187988358351e-06,
      "loss": 0.2322,
      "step": 5355
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2903357607042933,
      "learning_rate": 5.409712220300686e-06,
      "loss": 0.2255,
      "step": 5356
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2987393449118856,
      "learning_rate": 5.408236416308519e-06,
      "loss": 0.2022,
      "step": 5357
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3122355788455284,
      "learning_rate": 5.406760576511288e-06,
      "loss": 0.1604,
      "step": 5358
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3617574577143916,
      "learning_rate": 5.4052847010384325e-06,
      "loss": 0.2136,
      "step": 5359
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.204331192590547,
      "learning_rate": 5.4038087900193985e-06,
      "loss": 0.1955,
      "step": 5360
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2521741149496552,
      "learning_rate": 5.402332843583631e-06,
      "loss": 0.1739,
      "step": 5361
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3970436462421856,
      "learning_rate": 5.400856861860582e-06,
      "loss": 0.1602,
      "step": 5362
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2684908112511595,
      "learning_rate": 5.399380844979704e-06,
      "loss": 0.192,
      "step": 5363
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.405489594107299,
      "learning_rate": 5.397904793070454e-06,
      "loss": 0.1846,
      "step": 5364
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2269994749031288,
      "learning_rate": 5.3964287062622925e-06,
      "loss": 0.1767,
      "step": 5365
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4816572613956143,
      "learning_rate": 5.394952584684683e-06,
      "loss": 0.2244,
      "step": 5366
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3749674419197837,
      "learning_rate": 5.393476428467087e-06,
      "loss": 0.2261,
      "step": 5367
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.406716944156697,
      "learning_rate": 5.392000237738978e-06,
      "loss": 0.2168,
      "step": 5368
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.449785863784339,
      "learning_rate": 5.390524012629824e-06,
      "loss": 0.1958,
      "step": 5369
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5572873124006004,
      "learning_rate": 5.389047753269102e-06,
      "loss": 0.2169,
      "step": 5370
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2725545783100165,
      "learning_rate": 5.38757145978629e-06,
      "loss": 0.1908,
      "step": 5371
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3368518381486514,
      "learning_rate": 5.3860951323108676e-06,
      "loss": 0.1805,
      "step": 5372
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3091148969620974,
      "learning_rate": 5.3846187709723195e-06,
      "loss": 0.1403,
      "step": 5373
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1504907552422556,
      "learning_rate": 5.383142375900132e-06,
      "loss": 0.1771,
      "step": 5374
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3762526650196716,
      "learning_rate": 5.381665947223796e-06,
      "loss": 0.212,
      "step": 5375
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.5422691044635215,
      "learning_rate": 5.380189485072803e-06,
      "loss": 0.593,
      "step": 5376
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.48809485790328,
      "learning_rate": 5.3787129895766484e-06,
      "loss": 0.171,
      "step": 5377
    },
    {
      "epoch": 0.49,
      "grad_norm": 7.161000059386557,
      "learning_rate": 5.3772364608648304e-06,
      "loss": 0.6056,
      "step": 5378
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1340704008195468,
      "learning_rate": 5.375759899066853e-06,
      "loss": 0.156,
      "step": 5379
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2803242532192185,
      "learning_rate": 5.374283304312216e-06,
      "loss": 0.2054,
      "step": 5380
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4093411867814634,
      "learning_rate": 5.372806676730431e-06,
      "loss": 0.208,
      "step": 5381
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.492212066074004,
      "learning_rate": 5.371330016451006e-06,
      "loss": 0.2074,
      "step": 5382
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3281445201646849,
      "learning_rate": 5.3698533236034545e-06,
      "loss": 0.1876,
      "step": 5383
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3585715150721236,
      "learning_rate": 5.3683765983172945e-06,
      "loss": 0.1782,
      "step": 5384
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5493239789081161,
      "learning_rate": 5.3668998407220385e-06,
      "loss": 0.2259,
      "step": 5385
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4366052877755742,
      "learning_rate": 5.365423050947214e-06,
      "loss": 0.1951,
      "step": 5386
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4188518862111692,
      "learning_rate": 5.363946229122343e-06,
      "loss": 0.2216,
      "step": 5387
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4575683711697949,
      "learning_rate": 5.362469375376953e-06,
      "loss": 0.1942,
      "step": 5388
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4590714152069377,
      "learning_rate": 5.3609924898405745e-06,
      "loss": 0.1994,
      "step": 5389
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4597982370147757,
      "learning_rate": 5.359515572642741e-06,
      "loss": 0.229,
      "step": 5390
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.336078834592265,
      "learning_rate": 5.358038623912985e-06,
      "loss": 0.2291,
      "step": 5391
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.208120252520734,
      "learning_rate": 5.356561643780846e-06,
      "loss": 0.1856,
      "step": 5392
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.715347919723074,
      "learning_rate": 5.3550846323758666e-06,
      "loss": 0.4931,
      "step": 5393
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.34234430161024,
      "learning_rate": 5.353607589827589e-06,
      "loss": 0.2419,
      "step": 5394
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2376794231849206,
      "learning_rate": 5.35213051626556e-06,
      "loss": 0.1911,
      "step": 5395
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5250259091121583,
      "learning_rate": 5.350653411819329e-06,
      "loss": 0.2244,
      "step": 5396
    },
    {
      "epoch": 0.49,
      "grad_norm": 8.41279542444585,
      "learning_rate": 5.349176276618449e-06,
      "loss": 0.6253,
      "step": 5397
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.294440994626182,
      "learning_rate": 5.347699110792472e-06,
      "loss": 0.191,
      "step": 5398
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2652257874316848,
      "learning_rate": 5.346221914470959e-06,
      "loss": 0.1982,
      "step": 5399
    },
    {
      "epoch": 0.49,
      "grad_norm": 4.61867347284111,
      "learning_rate": 5.344744687783467e-06,
      "loss": 0.5791,
      "step": 5400
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.116422040249046,
      "learning_rate": 5.343267430859559e-06,
      "loss": 0.5767,
      "step": 5401
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4805851708285291,
      "learning_rate": 5.341790143828802e-06,
      "loss": 0.1641,
      "step": 5402
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4310338691750006,
      "learning_rate": 5.3403128268207615e-06,
      "loss": 0.1802,
      "step": 5403
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3687710916998028,
      "learning_rate": 5.338835479965011e-06,
      "loss": 0.1629,
      "step": 5404
    },
    {
      "epoch": 0.49,
      "grad_norm": 6.1242408837463795,
      "learning_rate": 5.337358103391122e-06,
      "loss": 0.5641,
      "step": 5405
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4114747051919032,
      "learning_rate": 5.335880697228669e-06,
      "loss": 0.1977,
      "step": 5406
    },
    {
      "epoch": 0.49,
      "grad_norm": 5.186885758044176,
      "learning_rate": 5.334403261607233e-06,
      "loss": 0.4762,
      "step": 5407
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3478466919950736,
      "learning_rate": 5.332925796656394e-06,
      "loss": 0.1739,
      "step": 5408
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6986226225099006,
      "learning_rate": 5.331448302505736e-06,
      "loss": 0.2275,
      "step": 5409
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3839930543905143,
      "learning_rate": 5.329970779284843e-06,
      "loss": 0.1613,
      "step": 5410
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3735787873647536,
      "learning_rate": 5.328493227123306e-06,
      "loss": 0.201,
      "step": 5411
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3006722470702057,
      "learning_rate": 5.327015646150716e-06,
      "loss": 0.2042,
      "step": 5412
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3224450649324198,
      "learning_rate": 5.325538036496665e-06,
      "loss": 0.176,
      "step": 5413
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3419786541938472,
      "learning_rate": 5.324060398290751e-06,
      "loss": 0.1832,
      "step": 5414
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.238089189873905,
      "learning_rate": 5.322582731662572e-06,
      "loss": 0.6406,
      "step": 5415
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2649409556566282,
      "learning_rate": 5.3211050367417305e-06,
      "loss": 0.1919,
      "step": 5416
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1649202269356185,
      "learning_rate": 5.319627313657829e-06,
      "loss": 0.1622,
      "step": 5417
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.370447787205625,
      "learning_rate": 5.318149562540473e-06,
      "loss": 0.2117,
      "step": 5418
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1327226567872761,
      "learning_rate": 5.316671783519272e-06,
      "loss": 0.134,
      "step": 5419
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4866038535727988,
      "learning_rate": 5.315193976723837e-06,
      "loss": 0.1906,
      "step": 5420
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2006911954775945,
      "learning_rate": 5.313716142283782e-06,
      "loss": 0.1879,
      "step": 5421
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.241882225255,
      "learning_rate": 5.312238280328723e-06,
      "loss": 0.6343,
      "step": 5422
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.504432396717288,
      "learning_rate": 5.310760390988278e-06,
      "loss": 0.608,
      "step": 5423
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3205924606864374,
      "learning_rate": 5.309282474392067e-06,
      "loss": 0.2013,
      "step": 5424
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2513919243054759,
      "learning_rate": 5.3078045306697154e-06,
      "loss": 0.1844,
      "step": 5425
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6160582471478149,
      "learning_rate": 5.306326559950846e-06,
      "loss": 0.2274,
      "step": 5426
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4776820126789942,
      "learning_rate": 5.304848562365088e-06,
      "loss": 0.2186,
      "step": 5427
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.488320991231062,
      "learning_rate": 5.303370538042072e-06,
      "loss": 0.2047,
      "step": 5428
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.567595075016131,
      "learning_rate": 5.301892487111431e-06,
      "loss": 0.2384,
      "step": 5429
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.332676261059836,
      "learning_rate": 5.300414409702799e-06,
      "loss": 0.1939,
      "step": 5430
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2908834313018196,
      "learning_rate": 5.298936305945812e-06,
      "loss": 0.1981,
      "step": 5431
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4280943663664771,
      "learning_rate": 5.297458175970113e-06,
      "loss": 0.2031,
      "step": 5432
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3556248035499507,
      "learning_rate": 5.295980019905342e-06,
      "loss": 0.207,
      "step": 5433
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5241610661509384,
      "learning_rate": 5.294501837881143e-06,
      "loss": 0.1945,
      "step": 5434
    },
    {
      "epoch": 0.5,
      "grad_norm": 10.462500169608136,
      "learning_rate": 5.293023630027162e-06,
      "loss": 0.5001,
      "step": 5435
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5112358172495695,
      "learning_rate": 5.291545396473049e-06,
      "loss": 0.206,
      "step": 5436
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5115846821900694,
      "learning_rate": 5.290067137348455e-06,
      "loss": 0.2041,
      "step": 5437
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3571335582938797,
      "learning_rate": 5.288588852783031e-06,
      "loss": 0.1679,
      "step": 5438
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9857237830696282,
      "learning_rate": 5.2871105429064354e-06,
      "loss": 0.1783,
      "step": 5439
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5685296955244918,
      "learning_rate": 5.285632207848323e-06,
      "loss": 0.2582,
      "step": 5440
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2760704162991496,
      "learning_rate": 5.284153847738356e-06,
      "loss": 0.2179,
      "step": 5441
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5216389628902798,
      "learning_rate": 5.282675462706197e-06,
      "loss": 0.1882,
      "step": 5442
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5658895197751455,
      "learning_rate": 5.2811970528815074e-06,
      "loss": 0.2004,
      "step": 5443
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2914709435833693,
      "learning_rate": 5.279718618393955e-06,
      "loss": 0.2052,
      "step": 5444
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.482996988005178,
      "learning_rate": 5.278240159373208e-06,
      "loss": 0.2196,
      "step": 5445
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.50615764329,
      "learning_rate": 5.2767616759489395e-06,
      "loss": 0.264,
      "step": 5446
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2699785562716637,
      "learning_rate": 5.275283168250818e-06,
      "loss": 0.209,
      "step": 5447
    },
    {
      "epoch": 0.5,
      "grad_norm": 11.848729171539578,
      "learning_rate": 5.273804636408523e-06,
      "loss": 0.6325,
      "step": 5448
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1867785872978276,
      "learning_rate": 5.272326080551729e-06,
      "loss": 0.1676,
      "step": 5449
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3139006384362286,
      "learning_rate": 5.270847500810117e-06,
      "loss": 0.2135,
      "step": 5450
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3424960319584833,
      "learning_rate": 5.2693688973133675e-06,
      "loss": 0.2041,
      "step": 5451
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2681927907379025,
      "learning_rate": 5.2678902701911614e-06,
      "loss": 0.1861,
      "step": 5452
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4605139467962969,
      "learning_rate": 5.266411619573189e-06,
      "loss": 0.233,
      "step": 5453
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3476540555023724,
      "learning_rate": 5.264932945589133e-06,
      "loss": 0.2076,
      "step": 5454
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3026090952971967,
      "learning_rate": 5.263454248368687e-06,
      "loss": 0.1666,
      "step": 5455
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3594160493525507,
      "learning_rate": 5.261975528041539e-06,
      "loss": 0.161,
      "step": 5456
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6554440060560138,
      "learning_rate": 5.260496784737386e-06,
      "loss": 0.2032,
      "step": 5457
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2296684889754588,
      "learning_rate": 5.259018018585923e-06,
      "loss": 0.1777,
      "step": 5458
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3663575083279267,
      "learning_rate": 5.257539229716847e-06,
      "loss": 0.1879,
      "step": 5459
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5059969552290986,
      "learning_rate": 5.256060418259858e-06,
      "loss": 0.2055,
      "step": 5460
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2747600367668424,
      "learning_rate": 5.254581584344656e-06,
      "loss": 0.1831,
      "step": 5461
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.3456538832563405,
      "learning_rate": 5.253102728100946e-06,
      "loss": 0.5124,
      "step": 5462
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3309919926893985,
      "learning_rate": 5.251623849658434e-06,
      "loss": 0.1992,
      "step": 5463
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1754613656885013,
      "learning_rate": 5.250144949146827e-06,
      "loss": 0.1767,
      "step": 5464
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5514344588215756,
      "learning_rate": 5.248666026695835e-06,
      "loss": 0.2469,
      "step": 5465
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4897028594072408,
      "learning_rate": 5.247187082435168e-06,
      "loss": 0.1973,
      "step": 5466
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.548126290911371,
      "learning_rate": 5.24570811649454e-06,
      "loss": 0.4514,
      "step": 5467
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2146468718551093,
      "learning_rate": 5.244229129003668e-06,
      "loss": 0.1876,
      "step": 5468
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2543518410161856,
      "learning_rate": 5.242750120092267e-06,
      "loss": 0.167,
      "step": 5469
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.521916780679478,
      "learning_rate": 5.241271089890057e-06,
      "loss": 0.5568,
      "step": 5470
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3497403160658203,
      "learning_rate": 5.239792038526757e-06,
      "loss": 0.1787,
      "step": 5471
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2617207254885927,
      "learning_rate": 5.238312966132091e-06,
      "loss": 0.1623,
      "step": 5472
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3952947168005223,
      "learning_rate": 5.236833872835785e-06,
      "loss": 0.1897,
      "step": 5473
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5727203640975653,
      "learning_rate": 5.235354758767562e-06,
      "loss": 0.2443,
      "step": 5474
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4300051768208202,
      "learning_rate": 5.233875624057154e-06,
      "loss": 0.1677,
      "step": 5475
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3659733482447134,
      "learning_rate": 5.23239646883429e-06,
      "loss": 0.1857,
      "step": 5476
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3004154976667117,
      "learning_rate": 5.230917293228699e-06,
      "loss": 0.1803,
      "step": 5477
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3664427565952364,
      "learning_rate": 5.229438097370118e-06,
      "loss": 0.2186,
      "step": 5478
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.298257051096371,
      "learning_rate": 5.22795888138828e-06,
      "loss": 0.2032,
      "step": 5479
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.264264613656242,
      "learning_rate": 5.226479645412923e-06,
      "loss": 0.1755,
      "step": 5480
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3334202623672196,
      "learning_rate": 5.2250003895737865e-06,
      "loss": 0.1878,
      "step": 5481
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.317512835948999,
      "learning_rate": 5.22352111400061e-06,
      "loss": 0.2044,
      "step": 5482
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.261258158929048,
      "learning_rate": 5.222041818823137e-06,
      "loss": 0.1778,
      "step": 5483
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.410157806166686,
      "learning_rate": 5.220562504171111e-06,
      "loss": 0.1926,
      "step": 5484
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7313606587217922,
      "learning_rate": 5.21908317017428e-06,
      "loss": 0.2155,
      "step": 5485
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4272904298646292,
      "learning_rate": 5.217603816962386e-06,
      "loss": 0.2307,
      "step": 5486
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2583129626512812,
      "learning_rate": 5.216124444665183e-06,
      "loss": 0.1978,
      "step": 5487
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2275391805974691,
      "learning_rate": 5.21464505341242e-06,
      "loss": 0.2042,
      "step": 5488
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4052571590147325,
      "learning_rate": 5.213165643333851e-06,
      "loss": 0.2362,
      "step": 5489
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2321971821672482,
      "learning_rate": 5.211686214559227e-06,
      "loss": 0.1754,
      "step": 5490
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4587721260199518,
      "learning_rate": 5.210206767218308e-06,
      "loss": 0.2173,
      "step": 5491
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3629402254276353,
      "learning_rate": 5.208727301440848e-06,
      "loss": 0.2107,
      "step": 5492
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2719891379067083,
      "learning_rate": 5.207247817356607e-06,
      "loss": 0.2075,
      "step": 5493
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2637762937847774,
      "learning_rate": 5.205768315095348e-06,
      "loss": 0.1722,
      "step": 5494
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.427021652843613,
      "learning_rate": 5.204288794786829e-06,
      "loss": 0.1936,
      "step": 5495
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.675189757195057,
      "learning_rate": 5.202809256560817e-06,
      "loss": 0.3808,
      "step": 5496
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3740327309561595,
      "learning_rate": 5.201329700547077e-06,
      "loss": 0.2121,
      "step": 5497
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4421979350555383,
      "learning_rate": 5.199850126875374e-06,
      "loss": 0.1767,
      "step": 5498
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.453295782769005,
      "learning_rate": 5.198370535675479e-06,
      "loss": 0.2053,
      "step": 5499
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.334810476846506,
      "learning_rate": 5.1968909270771604e-06,
      "loss": 0.2014,
      "step": 5500
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4843425427287225,
      "learning_rate": 5.19541130121019e-06,
      "loss": 0.2172,
      "step": 5501
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3952561455715093,
      "learning_rate": 5.193931658204343e-06,
      "loss": 0.1473,
      "step": 5502
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3389731017793287,
      "learning_rate": 5.192451998189392e-06,
      "loss": 0.2076,
      "step": 5503
    },
    {
      "epoch": 0.5,
      "grad_norm": 5.438100592835502,
      "learning_rate": 5.190972321295112e-06,
      "loss": 0.5112,
      "step": 5504
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2783515167062898,
      "learning_rate": 5.1894926276512824e-06,
      "loss": 0.188,
      "step": 5505
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3274245442398442,
      "learning_rate": 5.188012917387681e-06,
      "loss": 0.1809,
      "step": 5506
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.121203958178049,
      "learning_rate": 5.18653319063409e-06,
      "loss": 0.6078,
      "step": 5507
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4061486772155702,
      "learning_rate": 5.18505344752029e-06,
      "loss": 0.1959,
      "step": 5508
    },
    {
      "epoch": 0.5,
      "grad_norm": 6.984176726951175,
      "learning_rate": 5.1835736881760635e-06,
      "loss": 0.6246,
      "step": 5509
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.384588707584414,
      "learning_rate": 5.182093912731198e-06,
      "loss": 0.1912,
      "step": 5510
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3063028002996622,
      "learning_rate": 5.180614121315478e-06,
      "loss": 0.1536,
      "step": 5511
    },
    {
      "epoch": 0.5,
      "grad_norm": 7.60617011173385,
      "learning_rate": 5.179134314058692e-06,
      "loss": 0.6351,
      "step": 5512
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2344133326964613,
      "learning_rate": 5.177654491090627e-06,
      "loss": 0.1656,
      "step": 5513
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3291825136217055,
      "learning_rate": 5.176174652541074e-06,
      "loss": 0.1852,
      "step": 5514
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.408134478441794,
      "learning_rate": 5.1746947985398265e-06,
      "loss": 0.1653,
      "step": 5515
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4127203563740995,
      "learning_rate": 5.173214929216677e-06,
      "loss": 0.221,
      "step": 5516
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.991401687613672,
      "learning_rate": 5.171735044701418e-06,
      "loss": 0.4645,
      "step": 5517
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.231805951246055,
      "learning_rate": 5.170255145123848e-06,
      "loss": 0.1859,
      "step": 5518
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3281821835208207,
      "learning_rate": 5.168775230613763e-06,
      "loss": 0.1997,
      "step": 5519
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2992834299068472,
      "learning_rate": 5.16729530130096e-06,
      "loss": 0.1574,
      "step": 5520
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3234297404898683,
      "learning_rate": 5.1658153573152405e-06,
      "loss": 0.1717,
      "step": 5521
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3725540148327586,
      "learning_rate": 5.164335398786405e-06,
      "loss": 0.2148,
      "step": 5522
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5024587280070012,
      "learning_rate": 5.162855425844255e-06,
      "loss": 0.1957,
      "step": 5523
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.55875493077369,
      "learning_rate": 5.161375438618595e-06,
      "loss": 0.7139,
      "step": 5524
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4406482435986319,
      "learning_rate": 5.159895437239231e-06,
      "loss": 0.176,
      "step": 5525
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.463679784165347,
      "learning_rate": 5.158415421835965e-06,
      "loss": 0.2124,
      "step": 5526
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2762643022128495,
      "learning_rate": 5.156935392538608e-06,
      "loss": 0.171,
      "step": 5527
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4331631941987935,
      "learning_rate": 5.155455349476967e-06,
      "loss": 0.2164,
      "step": 5528
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3161661381691763,
      "learning_rate": 5.153975292780852e-06,
      "loss": 0.2022,
      "step": 5529
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2787807828644229,
      "learning_rate": 5.152495222580075e-06,
      "loss": 0.1577,
      "step": 5530
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2675523848535413,
      "learning_rate": 5.151015139004445e-06,
      "loss": 0.1682,
      "step": 5531
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5148436878569778,
      "learning_rate": 5.149535042183778e-06,
      "loss": 0.2152,
      "step": 5532
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2206544843315512,
      "learning_rate": 5.148054932247887e-06,
      "loss": 0.1809,
      "step": 5533
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4867144063454776,
      "learning_rate": 5.1465748093265875e-06,
      "loss": 0.1419,
      "step": 5534
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.407532343682772,
      "learning_rate": 5.145094673549698e-06,
      "loss": 0.7097,
      "step": 5535
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.240595203416136,
      "learning_rate": 5.1436145250470335e-06,
      "loss": 0.1978,
      "step": 5536
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5540300691163949,
      "learning_rate": 5.1421343639484165e-06,
      "loss": 0.2039,
      "step": 5537
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3123964915442745,
      "learning_rate": 5.140654190383663e-06,
      "loss": 0.1908,
      "step": 5538
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.269582516757633,
      "learning_rate": 5.139174004482597e-06,
      "loss": 0.1816,
      "step": 5539
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.386101264135285,
      "learning_rate": 5.137693806375039e-06,
      "loss": 0.5649,
      "step": 5540
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3764873434228162,
      "learning_rate": 5.1362135961908135e-06,
      "loss": 0.1728,
      "step": 5541
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4507985955695821,
      "learning_rate": 5.134733374059745e-06,
      "loss": 0.1813,
      "step": 5542
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5865411831045793,
      "learning_rate": 5.133253140111657e-06,
      "loss": 0.2053,
      "step": 5543
    },
    {
      "epoch": 0.51,
      "grad_norm": 13.64041482314365,
      "learning_rate": 5.131772894476378e-06,
      "loss": 0.673,
      "step": 5544
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.525302907896863,
      "learning_rate": 5.130292637283735e-06,
      "loss": 0.1926,
      "step": 5545
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3617927502123635,
      "learning_rate": 5.128812368663556e-06,
      "loss": 0.1978,
      "step": 5546
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5406039575360524,
      "learning_rate": 5.12733208874567e-06,
      "loss": 0.2028,
      "step": 5547
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3191189994649748,
      "learning_rate": 5.125851797659908e-06,
      "loss": 0.1937,
      "step": 5548
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2693633930152683,
      "learning_rate": 5.124371495536102e-06,
      "loss": 0.1571,
      "step": 5549
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.928587051910914,
      "learning_rate": 5.122891182504081e-06,
      "loss": 0.5918,
      "step": 5550
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.456550662280522,
      "learning_rate": 5.1214108586936836e-06,
      "loss": 0.2091,
      "step": 5551
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3928665768515407,
      "learning_rate": 5.119930524234742e-06,
      "loss": 0.1921,
      "step": 5552
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.202792283630807,
      "learning_rate": 5.118450179257091e-06,
      "loss": 0.1393,
      "step": 5553
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3795803585839754,
      "learning_rate": 5.116969823890566e-06,
      "loss": 0.1696,
      "step": 5554
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5959131956053196,
      "learning_rate": 5.115489458265006e-06,
      "loss": 0.2745,
      "step": 5555
    },
    {
      "epoch": 0.51,
      "grad_norm": 4.947533087109854,
      "learning_rate": 5.114009082510247e-06,
      "loss": 0.6024,
      "step": 5556
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3999417542623551,
      "learning_rate": 5.112528696756129e-06,
      "loss": 0.1847,
      "step": 5557
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.601714011609243,
      "learning_rate": 5.111048301132492e-06,
      "loss": 0.5396,
      "step": 5558
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3186797842319367,
      "learning_rate": 5.109567895769175e-06,
      "loss": 0.191,
      "step": 5559
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.376069102826562,
      "learning_rate": 5.1080874807960235e-06,
      "loss": 0.1859,
      "step": 5560
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.460427484065786,
      "learning_rate": 5.1066070563428736e-06,
      "loss": 0.212,
      "step": 5561
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4063095402409724,
      "learning_rate": 5.105126622539573e-06,
      "loss": 0.1924,
      "step": 5562
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3314347815522112,
      "learning_rate": 5.103646179515964e-06,
      "loss": 0.201,
      "step": 5563
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3121734881059461,
      "learning_rate": 5.1021657274018925e-06,
      "loss": 0.1749,
      "step": 5564
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2915445191834274,
      "learning_rate": 5.100685266327202e-06,
      "loss": 0.188,
      "step": 5565
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3011223202108357,
      "learning_rate": 5.0992047964217405e-06,
      "loss": 0.19,
      "step": 5566
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.48005774111876,
      "learning_rate": 5.097724317815354e-06,
      "loss": 0.6905,
      "step": 5567
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3636445374456088,
      "learning_rate": 5.0962438306378924e-06,
      "loss": 0.2057,
      "step": 5568
    },
    {
      "epoch": 0.51,
      "grad_norm": 8.719786916269111,
      "learning_rate": 5.0947633350192035e-06,
      "loss": 0.5857,
      "step": 5569
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2684452371295876,
      "learning_rate": 5.093282831089136e-06,
      "loss": 0.1589,
      "step": 5570
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3398712806066682,
      "learning_rate": 5.091802318977539e-06,
      "loss": 0.1575,
      "step": 5571
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.377438210572566,
      "learning_rate": 5.090321798814264e-06,
      "loss": 0.1876,
      "step": 5572
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4003255289555052,
      "learning_rate": 5.088841270729164e-06,
      "loss": 0.2021,
      "step": 5573
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3756084162259814,
      "learning_rate": 5.0873607348520904e-06,
      "loss": 0.2136,
      "step": 5574
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1442205823894154,
      "learning_rate": 5.085880191312895e-06,
      "loss": 0.16,
      "step": 5575
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2068282278513671,
      "learning_rate": 5.084399640241433e-06,
      "loss": 0.1575,
      "step": 5576
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3947225057573225,
      "learning_rate": 5.082919081767558e-06,
      "loss": 0.1722,
      "step": 5577
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6465875746150995,
      "learning_rate": 5.0814385160211256e-06,
      "loss": 0.182,
      "step": 5578
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5907780653242327,
      "learning_rate": 5.07995794313199e-06,
      "loss": 0.2567,
      "step": 5579
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4447280142010615,
      "learning_rate": 5.078477363230007e-06,
      "loss": 0.2128,
      "step": 5580
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.437605902415969,
      "learning_rate": 5.0769967764450345e-06,
      "loss": 0.21,
      "step": 5581
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4574717198141205,
      "learning_rate": 5.07551618290693e-06,
      "loss": 0.1729,
      "step": 5582
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3696650121665672,
      "learning_rate": 5.074035582745551e-06,
      "loss": 0.1834,
      "step": 5583
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3079865922334455,
      "learning_rate": 5.072554976090756e-06,
      "loss": 0.2322,
      "step": 5584
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3476927431254513,
      "learning_rate": 5.071074363072403e-06,
      "loss": 0.1717,
      "step": 5585
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1375920424015118,
      "learning_rate": 5.069593743820356e-06,
      "loss": 0.1557,
      "step": 5586
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3054484862393152,
      "learning_rate": 5.068113118464471e-06,
      "loss": 0.1759,
      "step": 5587
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3168553174658841,
      "learning_rate": 5.0666324871346085e-06,
      "loss": 0.218,
      "step": 5588
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3052968595147059,
      "learning_rate": 5.0651518499606314e-06,
      "loss": 0.2277,
      "step": 5589
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4440083317162398,
      "learning_rate": 5.063671207072402e-06,
      "loss": 0.1995,
      "step": 5590
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.433989395865072,
      "learning_rate": 5.062190558599781e-06,
      "loss": 0.2184,
      "step": 5591
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3828295680908897,
      "learning_rate": 5.060709904672632e-06,
      "loss": 0.2147,
      "step": 5592
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3043681854627578,
      "learning_rate": 5.059229245420819e-06,
      "loss": 0.1785,
      "step": 5593
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2738273954490484,
      "learning_rate": 5.057748580974204e-06,
      "loss": 0.168,
      "step": 5594
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2418816429479451,
      "learning_rate": 5.056267911462653e-06,
      "loss": 0.1598,
      "step": 5595
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.223670579336651,
      "learning_rate": 5.05478723701603e-06,
      "loss": 0.553,
      "step": 5596
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2619968236566241,
      "learning_rate": 5.053306557764199e-06,
      "loss": 0.1539,
      "step": 5597
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3139259360439122,
      "learning_rate": 5.051825873837026e-06,
      "loss": 0.2054,
      "step": 5598
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3695190871850014,
      "learning_rate": 5.050345185364378e-06,
      "loss": 0.1512,
      "step": 5599
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3975113814268199,
      "learning_rate": 5.04886449247612e-06,
      "loss": 0.1933,
      "step": 5600
    },
    {
      "epoch": 0.51,
      "grad_norm": 23.52830036853133,
      "learning_rate": 5.047383795302119e-06,
      "loss": 0.6323,
      "step": 5601
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2946737343321024,
      "learning_rate": 5.045903093972242e-06,
      "loss": 0.1804,
      "step": 5602
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3102259641433023,
      "learning_rate": 5.044422388616358e-06,
      "loss": 0.1768,
      "step": 5603
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4515949982521847,
      "learning_rate": 5.042941679364332e-06,
      "loss": 0.1924,
      "step": 5604
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.046741242052823,
      "learning_rate": 5.041460966346034e-06,
      "loss": 0.1441,
      "step": 5605
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.441478502450984,
      "learning_rate": 5.039980249691332e-06,
      "loss": 0.2212,
      "step": 5606
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3769052438476486,
      "learning_rate": 5.038499529530094e-06,
      "loss": 0.2092,
      "step": 5607
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3086862209248702,
      "learning_rate": 5.0370188059921896e-06,
      "loss": 0.2139,
      "step": 5608
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3977565528235436,
      "learning_rate": 5.035538079207488e-06,
      "loss": 0.1859,
      "step": 5609
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.336241183796019,
      "learning_rate": 5.034057349305859e-06,
      "loss": 0.6747,
      "step": 5610
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.312770029139897,
      "learning_rate": 5.0325766164171745e-06,
      "loss": 0.1699,
      "step": 5611
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4282308964219876,
      "learning_rate": 5.031095880671303e-06,
      "loss": 0.2086,
      "step": 5612
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3528692003763598,
      "learning_rate": 5.0296151421981136e-06,
      "loss": 0.2107,
      "step": 5613
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.197022405329983,
      "learning_rate": 5.028134401127477e-06,
      "loss": 0.5117,
      "step": 5614
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.387191791000173,
      "learning_rate": 5.026653657589267e-06,
      "loss": 0.2429,
      "step": 5615
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3813811689438038,
      "learning_rate": 5.025172911713352e-06,
      "loss": 0.2332,
      "step": 5616
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.479931770536127,
      "learning_rate": 5.023692163629603e-06,
      "loss": 0.2041,
      "step": 5617
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.402726764878429,
      "learning_rate": 5.022211413467895e-06,
      "loss": 0.4393,
      "step": 5618
    },
    {
      "epoch": 0.51,
      "grad_norm": 7.5315653818788935,
      "learning_rate": 5.020730661358096e-06,
      "loss": 0.5129,
      "step": 5619
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2933486402756549,
      "learning_rate": 5.019249907430079e-06,
      "loss": 0.1968,
      "step": 5620
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2784364589155015,
      "learning_rate": 5.017769151813718e-06,
      "loss": 0.1595,
      "step": 5621
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.172245467870352,
      "learning_rate": 5.016288394638883e-06,
      "loss": 0.1713,
      "step": 5622
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2451051347420432,
      "learning_rate": 5.014807636035447e-06,
      "loss": 0.1707,
      "step": 5623
    },
    {
      "epoch": 0.51,
      "grad_norm": 5.489082219319016,
      "learning_rate": 5.013326876133282e-06,
      "loss": 0.4475,
      "step": 5624
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.39901276185357,
      "learning_rate": 5.01184611506226e-06,
      "loss": 0.688,
      "step": 5625
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4068962236075977,
      "learning_rate": 5.010365352952256e-06,
      "loss": 0.1984,
      "step": 5626
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.325813067694516,
      "learning_rate": 5.008884589933139e-06,
      "loss": 0.1881,
      "step": 5627
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3410664301563144,
      "learning_rate": 5.007403826134786e-06,
      "loss": 0.2277,
      "step": 5628
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3686238367084644,
      "learning_rate": 5.005923061687068e-06,
      "loss": 0.1842,
      "step": 5629
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.454541549283674,
      "learning_rate": 5.004442296719857e-06,
      "loss": 0.2043,
      "step": 5630
    },
    {
      "epoch": 0.51,
      "grad_norm": 9.635603193467439,
      "learning_rate": 5.002961531363026e-06,
      "loss": 0.6756,
      "step": 5631
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.152102831599322,
      "learning_rate": 5.00148076574645e-06,
      "loss": 0.1448,
      "step": 5632
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.17455503952277,
      "learning_rate": 5e-06,
      "loss": 0.55,
      "step": 5633
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2605342769946926,
      "learning_rate": 4.9985192342535515e-06,
      "loss": 0.1766,
      "step": 5634
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4872047005349787,
      "learning_rate": 4.997038468636975e-06,
      "loss": 0.2354,
      "step": 5635
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4277969324950346,
      "learning_rate": 4.995557703280145e-06,
      "loss": 0.2203,
      "step": 5636
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4465008545062281,
      "learning_rate": 4.994076938312934e-06,
      "loss": 0.1905,
      "step": 5637
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1983295133899505,
      "learning_rate": 4.992596173865216e-06,
      "loss": 0.1635,
      "step": 5638
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.274179151594958,
      "learning_rate": 4.991115410066862e-06,
      "loss": 0.1639,
      "step": 5639
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.488339015746579,
      "learning_rate": 4.989634647047746e-06,
      "loss": 0.2787,
      "step": 5640
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1911613754872883,
      "learning_rate": 4.988153884937742e-06,
      "loss": 0.1993,
      "step": 5641
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3568853510510503,
      "learning_rate": 4.98667312386672e-06,
      "loss": 0.1927,
      "step": 5642
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4673569132943494,
      "learning_rate": 4.985192363964554e-06,
      "loss": 0.2327,
      "step": 5643
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.596462252311872,
      "learning_rate": 4.983711605361118e-06,
      "loss": 0.451,
      "step": 5644
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3524002757073466,
      "learning_rate": 4.9822308481862825e-06,
      "loss": 0.1572,
      "step": 5645
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.362575018174975,
      "learning_rate": 4.9807500925699215e-06,
      "loss": 0.21,
      "step": 5646
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2840578865947434,
      "learning_rate": 4.9792693386419056e-06,
      "loss": 0.1697,
      "step": 5647
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4827696520956548,
      "learning_rate": 4.977788586532107e-06,
      "loss": 0.1863,
      "step": 5648
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3538460548804505,
      "learning_rate": 4.9763078363703975e-06,
      "loss": 0.1768,
      "step": 5649
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.7821094411735485,
      "learning_rate": 4.97482708828665e-06,
      "loss": 0.6042,
      "step": 5650
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.220597310235528,
      "learning_rate": 4.973346342410735e-06,
      "loss": 0.1693,
      "step": 5651
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4487334854912906,
      "learning_rate": 4.971865598872524e-06,
      "loss": 0.2048,
      "step": 5652
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3748424155319023,
      "learning_rate": 4.970384857801888e-06,
      "loss": 0.2274,
      "step": 5653
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3672349130516976,
      "learning_rate": 4.968904119328699e-06,
      "loss": 0.1947,
      "step": 5654
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.354102968799807,
      "learning_rate": 4.967423383582826e-06,
      "loss": 0.1893,
      "step": 5655
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.425333801836749,
      "learning_rate": 4.9659426506941415e-06,
      "loss": 0.2468,
      "step": 5656
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2524556298988194,
      "learning_rate": 4.964461920792512e-06,
      "loss": 0.1473,
      "step": 5657
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4831714519312533,
      "learning_rate": 4.962981194007811e-06,
      "loss": 0.1986,
      "step": 5658
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4180223383084487,
      "learning_rate": 4.961500470469908e-06,
      "loss": 0.2036,
      "step": 5659
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3266463812460574,
      "learning_rate": 4.96001975030867e-06,
      "loss": 0.2574,
      "step": 5660
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3166633914618247,
      "learning_rate": 4.9585390336539675e-06,
      "loss": 0.1674,
      "step": 5661
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2607287526848272,
      "learning_rate": 4.9570583206356685e-06,
      "loss": 0.1994,
      "step": 5662
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3004105224613245,
      "learning_rate": 4.955577611383643e-06,
      "loss": 0.1896,
      "step": 5663
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7657974474772422,
      "learning_rate": 4.954096906027759e-06,
      "loss": 0.234,
      "step": 5664
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3492488416079313,
      "learning_rate": 4.952616204697882e-06,
      "loss": 0.1906,
      "step": 5665
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2544874627673157,
      "learning_rate": 4.9511355075238815e-06,
      "loss": 0.2023,
      "step": 5666
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3002934092309233,
      "learning_rate": 4.949654814635623e-06,
      "loss": 0.1833,
      "step": 5667
    },
    {
      "epoch": 0.52,
      "grad_norm": 10.127040018716366,
      "learning_rate": 4.948174126162975e-06,
      "loss": 0.7794,
      "step": 5668
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4265757221991129,
      "learning_rate": 4.946693442235802e-06,
      "loss": 0.1863,
      "step": 5669
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4039923703972264,
      "learning_rate": 4.945212762983971e-06,
      "loss": 0.2215,
      "step": 5670
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5410856471830616,
      "learning_rate": 4.943732088537348e-06,
      "loss": 0.2206,
      "step": 5671
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.416165074873131,
      "learning_rate": 4.942251419025797e-06,
      "loss": 0.2205,
      "step": 5672
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2867040477423175,
      "learning_rate": 4.940770754579183e-06,
      "loss": 0.177,
      "step": 5673
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4892645813481178,
      "learning_rate": 4.939290095327369e-06,
      "loss": 0.2286,
      "step": 5674
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3161304854965372,
      "learning_rate": 4.937809441400221e-06,
      "loss": 0.1807,
      "step": 5675
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.061813976139321,
      "learning_rate": 4.9363287929276e-06,
      "loss": 0.6073,
      "step": 5676
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2944891524075313,
      "learning_rate": 4.934848150039369e-06,
      "loss": 0.1802,
      "step": 5677
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1622328909449582,
      "learning_rate": 4.933367512865393e-06,
      "loss": 0.2026,
      "step": 5678
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2605426814421117,
      "learning_rate": 4.931886881535531e-06,
      "loss": 0.1717,
      "step": 5679
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1907401589063926,
      "learning_rate": 4.930406256179645e-06,
      "loss": 0.2064,
      "step": 5680
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3095531341117488,
      "learning_rate": 4.928925636927597e-06,
      "loss": 0.1868,
      "step": 5681
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.438649237752857,
      "learning_rate": 4.927445023909246e-06,
      "loss": 0.2166,
      "step": 5682
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.650181134532012,
      "learning_rate": 4.925964417254451e-06,
      "loss": 0.486,
      "step": 5683
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.497629527965488,
      "learning_rate": 4.924483817093071e-06,
      "loss": 0.2264,
      "step": 5684
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.4643460153827546,
      "learning_rate": 4.923003223554967e-06,
      "loss": 0.5501,
      "step": 5685
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.483576987054975,
      "learning_rate": 4.921522636769994e-06,
      "loss": 0.2143,
      "step": 5686
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.482191542664832,
      "learning_rate": 4.920042056868012e-06,
      "loss": 0.1975,
      "step": 5687
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4015332309680901,
      "learning_rate": 4.918561483978876e-06,
      "loss": 0.1879,
      "step": 5688
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3154463866249448,
      "learning_rate": 4.917080918232444e-06,
      "loss": 0.1801,
      "step": 5689
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.284546646717108,
      "learning_rate": 4.915600359758568e-06,
      "loss": 0.7465,
      "step": 5690
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.465080345212381,
      "learning_rate": 4.914119808687106e-06,
      "loss": 0.5438,
      "step": 5691
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4266577166883698,
      "learning_rate": 4.912639265147911e-06,
      "loss": 0.2134,
      "step": 5692
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4073800089159911,
      "learning_rate": 4.911158729270837e-06,
      "loss": 0.2064,
      "step": 5693
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2970938636927731,
      "learning_rate": 4.909678201185737e-06,
      "loss": 0.1713,
      "step": 5694
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.296036292392722,
      "learning_rate": 4.908197681022463e-06,
      "loss": 0.1659,
      "step": 5695
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3375169665410576,
      "learning_rate": 4.906717168910866e-06,
      "loss": 0.2258,
      "step": 5696
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.256998352914722,
      "learning_rate": 4.905236664980797e-06,
      "loss": 0.5313,
      "step": 5697
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3267538183081458,
      "learning_rate": 4.903756169362109e-06,
      "loss": 0.2277,
      "step": 5698
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5694642203104374,
      "learning_rate": 4.9022756821846466e-06,
      "loss": 0.2138,
      "step": 5699
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2948977811482807,
      "learning_rate": 4.900795203578261e-06,
      "loss": 0.1846,
      "step": 5700
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5674209248481958,
      "learning_rate": 4.8993147336728e-06,
      "loss": 0.2158,
      "step": 5701
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.267778563531199,
      "learning_rate": 4.897834272598109e-06,
      "loss": 0.1635,
      "step": 5702
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2460985293351212,
      "learning_rate": 4.896353820484037e-06,
      "loss": 0.2206,
      "step": 5703
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2974717038705532,
      "learning_rate": 4.894873377460428e-06,
      "loss": 0.17,
      "step": 5704
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1915190241909015,
      "learning_rate": 4.893392943657127e-06,
      "loss": 0.1499,
      "step": 5705
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4355620390753103,
      "learning_rate": 4.891912519203978e-06,
      "loss": 0.2324,
      "step": 5706
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.356593304737148,
      "learning_rate": 4.890432104230826e-06,
      "loss": 0.1665,
      "step": 5707
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5654317534717872,
      "learning_rate": 4.888951698867509e-06,
      "loss": 0.2143,
      "step": 5708
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.138075458416247,
      "learning_rate": 4.887471303243872e-06,
      "loss": 0.155,
      "step": 5709
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4354629683295073,
      "learning_rate": 4.885990917489754e-06,
      "loss": 0.2447,
      "step": 5710
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.450687985249076,
      "learning_rate": 4.8845105417349955e-06,
      "loss": 0.1948,
      "step": 5711
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3483359242078259,
      "learning_rate": 4.883030176109435e-06,
      "loss": 0.1816,
      "step": 5712
    },
    {
      "epoch": 0.52,
      "grad_norm": 5.15436013989326,
      "learning_rate": 4.88154982074291e-06,
      "loss": 0.6252,
      "step": 5713
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.193939445176183,
      "learning_rate": 4.880069475765259e-06,
      "loss": 0.1549,
      "step": 5714
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1701376358631244,
      "learning_rate": 4.878589141306318e-06,
      "loss": 0.1546,
      "step": 5715
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.326953286718918,
      "learning_rate": 4.87710881749592e-06,
      "loss": 0.2004,
      "step": 5716
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3358176158724497,
      "learning_rate": 4.875628504463902e-06,
      "loss": 0.2162,
      "step": 5717
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5775407222237507,
      "learning_rate": 4.8741482023400945e-06,
      "loss": 0.2211,
      "step": 5718
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2408116282124027,
      "learning_rate": 4.872667911254333e-06,
      "loss": 0.1665,
      "step": 5719
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.461352458778088,
      "learning_rate": 4.871187631336447e-06,
      "loss": 0.202,
      "step": 5720
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2596465244973125,
      "learning_rate": 4.8697073627162675e-06,
      "loss": 0.1537,
      "step": 5721
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.238098717531903,
      "learning_rate": 4.868227105523624e-06,
      "loss": 0.1801,
      "step": 5722
    },
    {
      "epoch": 0.52,
      "grad_norm": 7.383749320613764,
      "learning_rate": 4.866746859888345e-06,
      "loss": 0.5936,
      "step": 5723
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1904467939171572,
      "learning_rate": 4.8652666259402584e-06,
      "loss": 0.1807,
      "step": 5724
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.68496595688303,
      "learning_rate": 4.863786403809189e-06,
      "loss": 0.624,
      "step": 5725
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.298674880656128,
      "learning_rate": 4.862306193624963e-06,
      "loss": 0.199,
      "step": 5726
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2387731719364667,
      "learning_rate": 4.8608259955174055e-06,
      "loss": 0.1388,
      "step": 5727
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.155853015933902,
      "learning_rate": 4.859345809616339e-06,
      "loss": 0.1364,
      "step": 5728
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3677767308449347,
      "learning_rate": 4.857865636051586e-06,
      "loss": 0.2148,
      "step": 5729
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2956131483787279,
      "learning_rate": 4.856385474952967e-06,
      "loss": 0.1835,
      "step": 5730
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3923864794059464,
      "learning_rate": 4.854905326450304e-06,
      "loss": 0.2405,
      "step": 5731
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3940215437555579,
      "learning_rate": 4.853425190673415e-06,
      "loss": 0.216,
      "step": 5732
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.282786991359001,
      "learning_rate": 4.8519450677521155e-06,
      "loss": 0.1351,
      "step": 5733
    },
    {
      "epoch": 0.52,
      "grad_norm": 4.9246615550737225,
      "learning_rate": 4.850464957816224e-06,
      "loss": 0.468,
      "step": 5734
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3050521079918391,
      "learning_rate": 4.848984860995557e-06,
      "loss": 0.1783,
      "step": 5735
    },
    {
      "epoch": 0.52,
      "grad_norm": 6.3400332388747795,
      "learning_rate": 4.847504777419927e-06,
      "loss": 0.5577,
      "step": 5736
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3286198167637877,
      "learning_rate": 4.846024707219149e-06,
      "loss": 0.1943,
      "step": 5737
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3904569368615447,
      "learning_rate": 4.844544650523034e-06,
      "loss": 0.2104,
      "step": 5738
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4389485917718452,
      "learning_rate": 4.8430646074613936e-06,
      "loss": 0.2066,
      "step": 5739
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.349007812757157,
      "learning_rate": 4.8415845781640365e-06,
      "loss": 0.1831,
      "step": 5740
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.436047970584274,
      "learning_rate": 4.840104562760773e-06,
      "loss": 0.2474,
      "step": 5741
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.560938603517133,
      "learning_rate": 4.838624561381407e-06,
      "loss": 0.2098,
      "step": 5742
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2758468567398864,
      "learning_rate": 4.837144574155747e-06,
      "loss": 0.2124,
      "step": 5743
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4031600874373789,
      "learning_rate": 4.835664601213598e-06,
      "loss": 0.2052,
      "step": 5744
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3597999752082712,
      "learning_rate": 4.834184642684762e-06,
      "loss": 0.2233,
      "step": 5745
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5422674727408185,
      "learning_rate": 4.832704698699042e-06,
      "loss": 0.1945,
      "step": 5746
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.381881066045822,
      "learning_rate": 4.831224769386239e-06,
      "loss": 0.2208,
      "step": 5747
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.733761636716384,
      "learning_rate": 4.829744854876154e-06,
      "loss": 0.5253,
      "step": 5748
    },
    {
      "epoch": 0.53,
      "grad_norm": 7.2188159485754255,
      "learning_rate": 4.828264955298583e-06,
      "loss": 0.6523,
      "step": 5749
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3209810071269592,
      "learning_rate": 4.826785070783326e-06,
      "loss": 0.1824,
      "step": 5750
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2415639755935566,
      "learning_rate": 4.825305201460176e-06,
      "loss": 0.1612,
      "step": 5751
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5559005399431152,
      "learning_rate": 4.823825347458928e-06,
      "loss": 0.1957,
      "step": 5752
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2623494558318638,
      "learning_rate": 4.822345508909376e-06,
      "loss": 0.1654,
      "step": 5753
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3346319269082196,
      "learning_rate": 4.820865685941311e-06,
      "loss": 0.1919,
      "step": 5754
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3147433542602927,
      "learning_rate": 4.819385878684524e-06,
      "loss": 0.1645,
      "step": 5755
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.462374632203303,
      "learning_rate": 4.817906087268804e-06,
      "loss": 0.2021,
      "step": 5756
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.0019893973505365,
      "learning_rate": 4.816426311823938e-06,
      "loss": 0.5718,
      "step": 5757
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2768678827058084,
      "learning_rate": 4.8149465524797134e-06,
      "loss": 0.2101,
      "step": 5758
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4266806712211084,
      "learning_rate": 4.813466809365913e-06,
      "loss": 0.2057,
      "step": 5759
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2440380051510296,
      "learning_rate": 4.811987082612322e-06,
      "loss": 0.1568,
      "step": 5760
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2314621846938967,
      "learning_rate": 4.810507372348721e-06,
      "loss": 0.1644,
      "step": 5761
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3159437937127665,
      "learning_rate": 4.8090276787048905e-06,
      "loss": 0.1875,
      "step": 5762
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.652825783948506,
      "learning_rate": 4.807548001810611e-06,
      "loss": 0.6951,
      "step": 5763
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2891519242075675,
      "learning_rate": 4.8060683417956586e-06,
      "loss": 0.1946,
      "step": 5764
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4082650921664022,
      "learning_rate": 4.804588698789811e-06,
      "loss": 0.2133,
      "step": 5765
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.222528120812223,
      "learning_rate": 4.80310907292284e-06,
      "loss": 0.1696,
      "step": 5766
    },
    {
      "epoch": 0.53,
      "grad_norm": 5.5501100490770545,
      "learning_rate": 4.801629464324524e-06,
      "loss": 0.5393,
      "step": 5767
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.276315554496272,
      "learning_rate": 4.800149873124628e-06,
      "loss": 0.1602,
      "step": 5768
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.44189505007537,
      "learning_rate": 4.798670299452926e-06,
      "loss": 0.2252,
      "step": 5769
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.381668454748118,
      "learning_rate": 4.797190743439185e-06,
      "loss": 0.2033,
      "step": 5770
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4196601123403538,
      "learning_rate": 4.795711205213173e-06,
      "loss": 0.1957,
      "step": 5771
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4424083362409175,
      "learning_rate": 4.794231684904655e-06,
      "loss": 0.1727,
      "step": 5772
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3976665584773906,
      "learning_rate": 4.792752182643394e-06,
      "loss": 0.2383,
      "step": 5773
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4759039063603738,
      "learning_rate": 4.791272698559154e-06,
      "loss": 0.2245,
      "step": 5774
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3219269890529155,
      "learning_rate": 4.789793232781696e-06,
      "loss": 0.1911,
      "step": 5775
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2947118517629959,
      "learning_rate": 4.7883137854407755e-06,
      "loss": 0.196,
      "step": 5776
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3026734026356444,
      "learning_rate": 4.786834356666153e-06,
      "loss": 0.2016,
      "step": 5777
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.157174392023031,
      "learning_rate": 4.785354946587582e-06,
      "loss": 0.1771,
      "step": 5778
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1597694479233511,
      "learning_rate": 4.783875555334819e-06,
      "loss": 0.1689,
      "step": 5779
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4077434764006662,
      "learning_rate": 4.782396183037616e-06,
      "loss": 0.2133,
      "step": 5780
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3856046604148557,
      "learning_rate": 4.7809168298257235e-06,
      "loss": 0.216,
      "step": 5781
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.469470263126206,
      "learning_rate": 4.77943749582889e-06,
      "loss": 0.2452,
      "step": 5782
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3062542932917904,
      "learning_rate": 4.777958181176864e-06,
      "loss": 0.1611,
      "step": 5783
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3842208532696172,
      "learning_rate": 4.7764788859993925e-06,
      "loss": 0.1989,
      "step": 5784
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4076687955084737,
      "learning_rate": 4.774999610426216e-06,
      "loss": 0.1459,
      "step": 5785
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.388285444930387,
      "learning_rate": 4.7735203545870794e-06,
      "loss": 0.2043,
      "step": 5786
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.333783307574661,
      "learning_rate": 4.7720411186117235e-06,
      "loss": 0.1793,
      "step": 5787
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4011513185886275,
      "learning_rate": 4.770561902629885e-06,
      "loss": 0.2306,
      "step": 5788
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.29320827491304,
      "learning_rate": 4.7690827067713035e-06,
      "loss": 0.1939,
      "step": 5789
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4033827350330206,
      "learning_rate": 4.767603531165713e-06,
      "loss": 0.2336,
      "step": 5790
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.509947065397123,
      "learning_rate": 4.766124375942848e-06,
      "loss": 0.2116,
      "step": 5791
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.416796684776607,
      "learning_rate": 4.7646452412324404e-06,
      "loss": 0.1937,
      "step": 5792
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2995580871701498,
      "learning_rate": 4.7631661271642185e-06,
      "loss": 0.2159,
      "step": 5793
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2263676755311597,
      "learning_rate": 4.761687033867912e-06,
      "loss": 0.1983,
      "step": 5794
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3528493467847453,
      "learning_rate": 4.760207961473246e-06,
      "loss": 0.1852,
      "step": 5795
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.508469564173116,
      "learning_rate": 4.7587289101099465e-06,
      "loss": 0.2316,
      "step": 5796
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4172499872162825,
      "learning_rate": 4.757249879907736e-06,
      "loss": 0.2007,
      "step": 5797
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.5689825918552,
      "learning_rate": 4.755770870996334e-06,
      "loss": 0.4459,
      "step": 5798
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.883676058929396,
      "learning_rate": 4.75429188350546e-06,
      "loss": 0.5145,
      "step": 5799
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2828982865897425,
      "learning_rate": 4.752812917564833e-06,
      "loss": 0.161,
      "step": 5800
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3165873249104618,
      "learning_rate": 4.751333973304166e-06,
      "loss": 0.1959,
      "step": 5801
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2827915796290572,
      "learning_rate": 4.749855050853174e-06,
      "loss": 0.1662,
      "step": 5802
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2810918249016847,
      "learning_rate": 4.748376150341566e-06,
      "loss": 0.1823,
      "step": 5803
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3594366300518663,
      "learning_rate": 4.7468972718990545e-06,
      "loss": 0.1702,
      "step": 5804
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1721417009310489,
      "learning_rate": 4.745418415655344e-06,
      "loss": 0.1614,
      "step": 5805
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2657022127014552,
      "learning_rate": 4.743939581740144e-06,
      "loss": 0.1732,
      "step": 5806
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.677575185621529,
      "learning_rate": 4.7424607702831535e-06,
      "loss": 0.238,
      "step": 5807
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4848603917216947,
      "learning_rate": 4.740981981414077e-06,
      "loss": 0.2671,
      "step": 5808
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3801893643088168,
      "learning_rate": 4.739503215262614e-06,
      "loss": 0.1519,
      "step": 5809
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2827375860214023,
      "learning_rate": 4.738024471958461e-06,
      "loss": 0.1703,
      "step": 5810
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2309735901625467,
      "learning_rate": 4.736545751631313e-06,
      "loss": 0.165,
      "step": 5811
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4850655929835457,
      "learning_rate": 4.735067054410867e-06,
      "loss": 0.2471,
      "step": 5812
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4676968056617894,
      "learning_rate": 4.733588380426812e-06,
      "loss": 0.1808,
      "step": 5813
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.310516185368121,
      "learning_rate": 4.732109729808838e-06,
      "loss": 0.1902,
      "step": 5814
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3759491630824008,
      "learning_rate": 4.730631102686635e-06,
      "loss": 0.2009,
      "step": 5815
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2891851306847208,
      "learning_rate": 4.729152499189884e-06,
      "loss": 0.1694,
      "step": 5816
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4061986033890592,
      "learning_rate": 4.727673919448271e-06,
      "loss": 0.2476,
      "step": 5817
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3609077870974942,
      "learning_rate": 4.726195363591477e-06,
      "loss": 0.2388,
      "step": 5818
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.435781136067207,
      "learning_rate": 4.724716831749182e-06,
      "loss": 0.2143,
      "step": 5819
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2493568997671172,
      "learning_rate": 4.723238324051061e-06,
      "loss": 0.183,
      "step": 5820
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.275601427862246,
      "learning_rate": 4.721759840626792e-06,
      "loss": 0.1644,
      "step": 5821
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2189129520973512,
      "learning_rate": 4.720281381606046e-06,
      "loss": 0.1925,
      "step": 5822
    },
    {
      "epoch": 0.53,
      "grad_norm": 6.520373177154753,
      "learning_rate": 4.718802947118494e-06,
      "loss": 0.6287,
      "step": 5823
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2251527906191855,
      "learning_rate": 4.7173245372938045e-06,
      "loss": 0.1866,
      "step": 5824
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2890630160869572,
      "learning_rate": 4.715846152261645e-06,
      "loss": 0.1623,
      "step": 5825
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.354265983084915,
      "learning_rate": 4.7143677921516775e-06,
      "loss": 0.1996,
      "step": 5826
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4098577480220629,
      "learning_rate": 4.712889457093565e-06,
      "loss": 0.1906,
      "step": 5827
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.440933070402594,
      "learning_rate": 4.711411147216969e-06,
      "loss": 0.1778,
      "step": 5828
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4865997650491831,
      "learning_rate": 4.709932862651546e-06,
      "loss": 0.1858,
      "step": 5829
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3379582128048977,
      "learning_rate": 4.7084546035269516e-06,
      "loss": 0.1815,
      "step": 5830
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1854684616206375,
      "learning_rate": 4.706976369972838e-06,
      "loss": 0.143,
      "step": 5831
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.517677587076647,
      "learning_rate": 4.705498162118858e-06,
      "loss": 0.1868,
      "step": 5832
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2874422500746157,
      "learning_rate": 4.704019980094659e-06,
      "loss": 0.1627,
      "step": 5833
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3437720211389275,
      "learning_rate": 4.702541824029887e-06,
      "loss": 0.196,
      "step": 5834
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2851061134610344,
      "learning_rate": 4.701063694054188e-06,
      "loss": 0.189,
      "step": 5835
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4414121297470661,
      "learning_rate": 4.699585590297202e-06,
      "loss": 0.1982,
      "step": 5836
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.568683955132922,
      "learning_rate": 4.69810751288857e-06,
      "loss": 0.6746,
      "step": 5837
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2543241751351735,
      "learning_rate": 4.6966294619579274e-06,
      "loss": 0.1774,
      "step": 5838
    },
    {
      "epoch": 0.53,
      "grad_norm": 4.619105328176056,
      "learning_rate": 4.6951514376349125e-06,
      "loss": 0.3489,
      "step": 5839
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2731234518218788,
      "learning_rate": 4.693673440049155e-06,
      "loss": 0.1602,
      "step": 5840
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3011615347805765,
      "learning_rate": 4.692195469330286e-06,
      "loss": 0.2029,
      "step": 5841
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3497163270183328,
      "learning_rate": 4.690717525607933e-06,
      "loss": 0.1655,
      "step": 5842
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1967734186204695,
      "learning_rate": 4.689239609011723e-06,
      "loss": 0.157,
      "step": 5843
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3964818944095654,
      "learning_rate": 4.687761719671278e-06,
      "loss": 0.1933,
      "step": 5844
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3033797138218262,
      "learning_rate": 4.6862838577162186e-06,
      "loss": 0.172,
      "step": 5845
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3667680908315494,
      "learning_rate": 4.684806023276163e-06,
      "loss": 0.1621,
      "step": 5846
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4390266101479714,
      "learning_rate": 4.683328216480728e-06,
      "loss": 0.2008,
      "step": 5847
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.406755708135987,
      "learning_rate": 4.681850437459528e-06,
      "loss": 0.1903,
      "step": 5848
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.394291157899469,
      "learning_rate": 4.680372686342173e-06,
      "loss": 0.2029,
      "step": 5849
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4887264911988545,
      "learning_rate": 4.67889496325827e-06,
      "loss": 0.196,
      "step": 5850
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3288802556286252,
      "learning_rate": 4.677417268337428e-06,
      "loss": 0.2103,
      "step": 5851
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2803284226112845,
      "learning_rate": 4.675939601709249e-06,
      "loss": 0.1704,
      "step": 5852
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3860944540015465,
      "learning_rate": 4.674461963503335e-06,
      "loss": 0.1782,
      "step": 5853
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4155567039606245,
      "learning_rate": 4.672984353849285e-06,
      "loss": 0.1901,
      "step": 5854
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3258419920771254,
      "learning_rate": 4.671506772876694e-06,
      "loss": 0.2113,
      "step": 5855
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4101992707520004,
      "learning_rate": 4.670029220715157e-06,
      "loss": 0.189,
      "step": 5856
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3567078040935783,
      "learning_rate": 4.668551697494265e-06,
      "loss": 0.1605,
      "step": 5857
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.917472140795904,
      "learning_rate": 4.667074203343607e-06,
      "loss": 0.4593,
      "step": 5858
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2064724157765678,
      "learning_rate": 4.665596738392767e-06,
      "loss": 0.166,
      "step": 5859
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4445601932381118,
      "learning_rate": 4.664119302771331e-06,
      "loss": 0.2156,
      "step": 5860
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3105627057908578,
      "learning_rate": 4.662641896608879e-06,
      "loss": 0.1936,
      "step": 5861
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5392378353955225,
      "learning_rate": 4.66116452003499e-06,
      "loss": 0.2203,
      "step": 5862
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.507826949941993,
      "learning_rate": 4.6596871731792385e-06,
      "loss": 0.6308,
      "step": 5863
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2621760891175766,
      "learning_rate": 4.658209856171199e-06,
      "loss": 0.1835,
      "step": 5864
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2306395128278775,
      "learning_rate": 4.656732569140441e-06,
      "loss": 0.1825,
      "step": 5865
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3815585290404273,
      "learning_rate": 4.6552553122165345e-06,
      "loss": 0.1898,
      "step": 5866
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2971214583282031,
      "learning_rate": 4.653778085529043e-06,
      "loss": 0.1339,
      "step": 5867
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2151363255477197,
      "learning_rate": 4.652300889207529e-06,
      "loss": 0.1941,
      "step": 5868
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3983638557463152,
      "learning_rate": 4.650823723381552e-06,
      "loss": 0.2351,
      "step": 5869
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5517118008611168,
      "learning_rate": 4.649346588180672e-06,
      "loss": 0.2259,
      "step": 5870
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4687172612915573,
      "learning_rate": 4.647869483734441e-06,
      "loss": 0.2122,
      "step": 5871
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.513657694219756,
      "learning_rate": 4.646392410172411e-06,
      "loss": 0.5183,
      "step": 5872
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8522514265462529,
      "learning_rate": 4.644915367624134e-06,
      "loss": 0.1891,
      "step": 5873
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4356476019267232,
      "learning_rate": 4.643438356219155e-06,
      "loss": 0.2085,
      "step": 5874
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5053691324293363,
      "learning_rate": 4.641961376087017e-06,
      "loss": 0.2219,
      "step": 5875
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1766246520741337,
      "learning_rate": 4.640484427357261e-06,
      "loss": 0.1425,
      "step": 5876
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.248792184694948,
      "learning_rate": 4.6390075101594255e-06,
      "loss": 0.1706,
      "step": 5877
    },
    {
      "epoch": 0.54,
      "grad_norm": 7.968592476884842,
      "learning_rate": 4.637530624623046e-06,
      "loss": 0.5316,
      "step": 5878
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4498762959116345,
      "learning_rate": 4.636053770877657e-06,
      "loss": 0.2351,
      "step": 5879
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4303089831646378,
      "learning_rate": 4.634576949052786e-06,
      "loss": 0.1944,
      "step": 5880
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3954347364833881,
      "learning_rate": 4.6331001592779615e-06,
      "loss": 0.2118,
      "step": 5881
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.5063276946925654,
      "learning_rate": 4.631623401682708e-06,
      "loss": 0.5183,
      "step": 5882
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3256804562651112,
      "learning_rate": 4.630146676396546e-06,
      "loss": 0.1901,
      "step": 5883
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.205817377450685,
      "learning_rate": 4.628669983548995e-06,
      "loss": 0.1651,
      "step": 5884
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6125031446618883,
      "learning_rate": 4.627193323269571e-06,
      "loss": 0.2378,
      "step": 5885
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.304750241393002,
      "learning_rate": 4.625716695687785e-06,
      "loss": 0.2148,
      "step": 5886
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2762861064073676,
      "learning_rate": 4.62424010093315e-06,
      "loss": 0.1926,
      "step": 5887
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5344073529014128,
      "learning_rate": 4.62276353913517e-06,
      "loss": 0.1941,
      "step": 5888
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3704991338631616,
      "learning_rate": 4.621287010423353e-06,
      "loss": 0.1744,
      "step": 5889
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3546147882054609,
      "learning_rate": 4.619810514927198e-06,
      "loss": 0.1953,
      "step": 5890
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.424898300411858,
      "learning_rate": 4.618334052776205e-06,
      "loss": 0.2001,
      "step": 5891
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3991600444910766,
      "learning_rate": 4.616857624099869e-06,
      "loss": 0.1887,
      "step": 5892
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2007645805606735,
      "learning_rate": 4.615381229027681e-06,
      "loss": 0.1634,
      "step": 5893
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2520743064911952,
      "learning_rate": 4.613904867689133e-06,
      "loss": 0.1933,
      "step": 5894
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.256674027023574,
      "learning_rate": 4.612428540213711e-06,
      "loss": 0.598,
      "step": 5895
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3913542298792103,
      "learning_rate": 4.610952246730899e-06,
      "loss": 0.1994,
      "step": 5896
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3742161636805441,
      "learning_rate": 4.609475987370177e-06,
      "loss": 0.1948,
      "step": 5897
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.703710599667075,
      "learning_rate": 4.607999762261024e-06,
      "loss": 0.6883,
      "step": 5898
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3838766677393584,
      "learning_rate": 4.606523571532914e-06,
      "loss": 0.2015,
      "step": 5899
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4115046538299,
      "learning_rate": 4.605047415315319e-06,
      "loss": 0.2021,
      "step": 5900
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2752573440214747,
      "learning_rate": 4.603571293737708e-06,
      "loss": 0.1983,
      "step": 5901
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.325021106971625,
      "learning_rate": 4.602095206929547e-06,
      "loss": 0.2094,
      "step": 5902
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2107776295687678,
      "learning_rate": 4.600619155020298e-06,
      "loss": 0.1689,
      "step": 5903
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.402971870030341,
      "learning_rate": 4.59914313813942e-06,
      "loss": 0.5574,
      "step": 5904
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2821544418141175,
      "learning_rate": 4.597667156416371e-06,
      "loss": 0.1635,
      "step": 5905
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.200552570744236,
      "learning_rate": 4.596191209980604e-06,
      "loss": 0.1334,
      "step": 5906
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3342145791171685,
      "learning_rate": 4.594715298961568e-06,
      "loss": 0.172,
      "step": 5907
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2668601934268762,
      "learning_rate": 4.593239423488713e-06,
      "loss": 0.204,
      "step": 5908
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1882361769412095,
      "learning_rate": 4.5917635836914825e-06,
      "loss": 0.215,
      "step": 5909
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4206592823395463,
      "learning_rate": 4.590287779699315e-06,
      "loss": 0.1302,
      "step": 5910
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4850021168519227,
      "learning_rate": 4.588812011641652e-06,
      "loss": 0.2454,
      "step": 5911
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3290675125186662,
      "learning_rate": 4.5873362796479236e-06,
      "loss": 0.1927,
      "step": 5912
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.133995727754334,
      "learning_rate": 4.585860583847566e-06,
      "loss": 0.1653,
      "step": 5913
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3442870900566548,
      "learning_rate": 4.584384924370005e-06,
      "loss": 0.1732,
      "step": 5914
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3908328993202068,
      "learning_rate": 4.5829093013446666e-06,
      "loss": 0.1823,
      "step": 5915
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3837684614554464,
      "learning_rate": 4.581433714900973e-06,
      "loss": 0.1934,
      "step": 5916
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2741000930436215,
      "learning_rate": 4.579958165168342e-06,
      "loss": 0.1989,
      "step": 5917
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.226575936648929,
      "learning_rate": 4.578482652276191e-06,
      "loss": 0.5436,
      "step": 5918
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4149006652192653,
      "learning_rate": 4.577007176353931e-06,
      "loss": 0.1962,
      "step": 5919
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1469415347336178,
      "learning_rate": 4.57553173753097e-06,
      "loss": 0.1539,
      "step": 5920
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.397210709444152,
      "learning_rate": 4.5740563359367164e-06,
      "loss": 0.207,
      "step": 5921
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2656886564328984,
      "learning_rate": 4.572580971700571e-06,
      "loss": 0.1802,
      "step": 5922
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3066309758122054,
      "learning_rate": 4.5711056449519345e-06,
      "loss": 0.1674,
      "step": 5923
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4801289420333397,
      "learning_rate": 4.569630355820202e-06,
      "loss": 0.1978,
      "step": 5924
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.18330068991044,
      "learning_rate": 4.568155104434765e-06,
      "loss": 0.1789,
      "step": 5925
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.201043681133201,
      "learning_rate": 4.566679890925017e-06,
      "loss": 0.599,
      "step": 5926
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.37519637912811,
      "learning_rate": 4.5652047154203395e-06,
      "loss": 0.1788,
      "step": 5927
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.430130471226371,
      "learning_rate": 4.563729578050118e-06,
      "loss": 0.1962,
      "step": 5928
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3639565248296832,
      "learning_rate": 4.562254478943729e-06,
      "loss": 0.1991,
      "step": 5929
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.829792005292734,
      "learning_rate": 4.560779418230552e-06,
      "loss": 0.2392,
      "step": 5930
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6183286944098287,
      "learning_rate": 4.559304396039959e-06,
      "loss": 0.184,
      "step": 5931
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5132126687620493,
      "learning_rate": 4.5578294125013175e-06,
      "loss": 0.2461,
      "step": 5932
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5809038929063528,
      "learning_rate": 4.556354467743995e-06,
      "loss": 0.2602,
      "step": 5933
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.3921933552170955,
      "learning_rate": 4.5548795618973536e-06,
      "loss": 0.5885,
      "step": 5934
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3382736631039023,
      "learning_rate": 4.553404695090753e-06,
      "loss": 0.2192,
      "step": 5935
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2607280894899193,
      "learning_rate": 4.551929867453548e-06,
      "loss": 0.1902,
      "step": 5936
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4312825962884546,
      "learning_rate": 4.550455079115091e-06,
      "loss": 0.1819,
      "step": 5937
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.294755829146495,
      "learning_rate": 4.54898033020473e-06,
      "loss": 0.1937,
      "step": 5938
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2780953124328838,
      "learning_rate": 4.547505620851812e-06,
      "loss": 0.1792,
      "step": 5939
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.122377162092497,
      "learning_rate": 4.546030951185677e-06,
      "loss": 0.4793,
      "step": 5940
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4317890380668987,
      "learning_rate": 4.5445563213356655e-06,
      "loss": 0.1846,
      "step": 5941
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.294966622042207,
      "learning_rate": 4.5430817314311115e-06,
      "loss": 0.1773,
      "step": 5942
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.32762574333295,
      "learning_rate": 4.541607181601346e-06,
      "loss": 0.1711,
      "step": 5943
    },
    {
      "epoch": 0.54,
      "grad_norm": 10.487026309201621,
      "learning_rate": 4.5401326719756985e-06,
      "loss": 0.6009,
      "step": 5944
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4151279945749264,
      "learning_rate": 4.53865820268349e-06,
      "loss": 0.2149,
      "step": 5945
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.345685950655604,
      "learning_rate": 4.537183773854045e-06,
      "loss": 0.1853,
      "step": 5946
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4554081151455798,
      "learning_rate": 4.535709385616679e-06,
      "loss": 0.1801,
      "step": 5947
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4230978787298074,
      "learning_rate": 4.5342350381007046e-06,
      "loss": 0.2082,
      "step": 5948
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4529932008041808,
      "learning_rate": 4.532760731435434e-06,
      "loss": 0.177,
      "step": 5949
    },
    {
      "epoch": 0.54,
      "grad_norm": 4.926441104987107,
      "learning_rate": 4.5312864657501735e-06,
      "loss": 0.3945,
      "step": 5950
    },
    {
      "epoch": 0.54,
      "grad_norm": 6.205049743818295,
      "learning_rate": 4.529812241174225e-06,
      "loss": 0.5583,
      "step": 5951
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3508453558079845,
      "learning_rate": 4.528338057836889e-06,
      "loss": 0.1799,
      "step": 5952
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5773604988806167,
      "learning_rate": 4.52686391586746e-06,
      "loss": 0.2281,
      "step": 5953
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2182301068703603,
      "learning_rate": 4.5253898153952306e-06,
      "loss": 0.1615,
      "step": 5954
    },
    {
      "epoch": 0.54,
      "grad_norm": 5.25131863167822,
      "learning_rate": 4.52391575654949e-06,
      "loss": 0.5912,
      "step": 5955
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3603072628480215,
      "learning_rate": 4.522441739459521e-06,
      "loss": 0.1846,
      "step": 5956
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4040030048267738,
      "learning_rate": 4.520967764254608e-06,
      "loss": 0.2478,
      "step": 5957
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2962325119445282,
      "learning_rate": 4.519493831064025e-06,
      "loss": 0.188,
      "step": 5958
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.195016744016805,
      "learning_rate": 4.51801994001705e-06,
      "loss": 0.1697,
      "step": 5959
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3367640758653574,
      "learning_rate": 4.516546091242948e-06,
      "loss": 0.2061,
      "step": 5960
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.29124262839155,
      "learning_rate": 4.51507228487099e-06,
      "loss": 0.5301,
      "step": 5961
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2894292938573026,
      "learning_rate": 4.513598521030436e-06,
      "loss": 0.1682,
      "step": 5962
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3409866279138072,
      "learning_rate": 4.512124799850545e-06,
      "loss": 0.1785,
      "step": 5963
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.441303729800367,
      "learning_rate": 4.510651121460573e-06,
      "loss": 0.2026,
      "step": 5964
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2800099927918556,
      "learning_rate": 4.509177485989771e-06,
      "loss": 0.1755,
      "step": 5965
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4884575662139197,
      "learning_rate": 4.507703893567387e-06,
      "loss": 0.2005,
      "step": 5966
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.375160036305886,
      "learning_rate": 4.5062303443226655e-06,
      "loss": 0.4725,
      "step": 5967
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2723310381399728,
      "learning_rate": 4.504756838384845e-06,
      "loss": 0.1482,
      "step": 5968
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3757815560846098,
      "learning_rate": 4.503283375883165e-06,
      "loss": 0.1767,
      "step": 5969
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.510600872801466,
      "learning_rate": 4.501809956946854e-06,
      "loss": 0.1925,
      "step": 5970
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1102127126294414,
      "learning_rate": 4.5003365817051434e-06,
      "loss": 0.1563,
      "step": 5971
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4645751432628036,
      "learning_rate": 4.4988632502872565e-06,
      "loss": 0.1862,
      "step": 5972
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5664146052481462,
      "learning_rate": 4.497389962822416e-06,
      "loss": 0.2084,
      "step": 5973
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2951243893381912,
      "learning_rate": 4.495916719439837e-06,
      "loss": 0.1492,
      "step": 5974
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.161297114605533,
      "learning_rate": 4.494443520268734e-06,
      "loss": 0.1365,
      "step": 5975
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.407261434765898,
      "learning_rate": 4.492970365438317e-06,
      "loss": 0.6574,
      "step": 5976
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5072255091664224,
      "learning_rate": 4.49149725507779e-06,
      "loss": 0.2221,
      "step": 5977
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.555536535027938,
      "learning_rate": 4.490024189316356e-06,
      "loss": 0.2129,
      "step": 5978
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5331807655346665,
      "learning_rate": 4.488551168283213e-06,
      "loss": 0.2272,
      "step": 5979
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4898593458847311,
      "learning_rate": 4.487078192107553e-06,
      "loss": 0.1733,
      "step": 5980
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4876897964047961,
      "learning_rate": 4.4856052609185665e-06,
      "loss": 0.1733,
      "step": 5981
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4399293052843287,
      "learning_rate": 4.48413237484544e-06,
      "loss": 0.1749,
      "step": 5982
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2283091300344715,
      "learning_rate": 4.482659534017356e-06,
      "loss": 0.1707,
      "step": 5983
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.327610670282453,
      "learning_rate": 4.4811867385634916e-06,
      "loss": 0.1586,
      "step": 5984
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4624237689275439,
      "learning_rate": 4.479713988613021e-06,
      "loss": 0.202,
      "step": 5985
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.858266812459728,
      "learning_rate": 4.478241284295115e-06,
      "loss": 0.3764,
      "step": 5986
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4535798088349479,
      "learning_rate": 4.476768625738937e-06,
      "loss": 0.173,
      "step": 5987
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3387019242749063,
      "learning_rate": 4.475296013073651e-06,
      "loss": 0.233,
      "step": 5988
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3884584939592117,
      "learning_rate": 4.473823446428416e-06,
      "loss": 0.1817,
      "step": 5989
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.248796382330011,
      "learning_rate": 4.472350925932384e-06,
      "loss": 0.1358,
      "step": 5990
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.416807961829454,
      "learning_rate": 4.470878451714705e-06,
      "loss": 0.1946,
      "step": 5991
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3230644704171564,
      "learning_rate": 4.4694060239045264e-06,
      "loss": 0.1741,
      "step": 5992
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3352652732494028,
      "learning_rate": 4.467933642630989e-06,
      "loss": 0.2212,
      "step": 5993
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4348353375484182,
      "learning_rate": 4.46646130802323e-06,
      "loss": 0.1932,
      "step": 5994
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1590181175847796,
      "learning_rate": 4.464989020210385e-06,
      "loss": 0.1588,
      "step": 5995
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5020633257933418,
      "learning_rate": 4.463516779321581e-06,
      "loss": 0.212,
      "step": 5996
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.544082750330183,
      "learning_rate": 4.462044585485944e-06,
      "loss": 0.6081,
      "step": 5997
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.887424631036927,
      "learning_rate": 4.4605724388325956e-06,
      "loss": 0.6829,
      "step": 5998
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4105741883737486,
      "learning_rate": 4.459100339490654e-06,
      "loss": 0.1779,
      "step": 5999
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.275863848941055,
      "learning_rate": 4.457628287589229e-06,
      "loss": 0.2041,
      "step": 6000
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2912768933998262,
      "learning_rate": 4.456156283257433e-06,
      "loss": 0.2175,
      "step": 6001
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3949443465739348,
      "learning_rate": 4.454684326624369e-06,
      "loss": 0.1823,
      "step": 6002
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.499756335923067,
      "learning_rate": 4.453212417819138e-06,
      "loss": 0.2268,
      "step": 6003
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3241358423281626,
      "learning_rate": 4.451740556970836e-06,
      "loss": 0.1714,
      "step": 6004
    },
    {
      "epoch": 0.55,
      "grad_norm": 10.615502806614323,
      "learning_rate": 4.450268744208555e-06,
      "loss": 0.6419,
      "step": 6005
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4931659028569617,
      "learning_rate": 4.448796979661382e-06,
      "loss": 0.2376,
      "step": 6006
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2652607870046604,
      "learning_rate": 4.447325263458401e-06,
      "loss": 0.2018,
      "step": 6007
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3124930552981988,
      "learning_rate": 4.445853595728692e-06,
      "loss": 0.2079,
      "step": 6008
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.261205179049543,
      "learning_rate": 4.44438197660133e-06,
      "loss": 0.17,
      "step": 6009
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3206201830432673,
      "learning_rate": 4.442910406205385e-06,
      "loss": 0.187,
      "step": 6010
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.728146199221785,
      "learning_rate": 4.441438884669925e-06,
      "loss": 0.7349,
      "step": 6011
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3456771212056087,
      "learning_rate": 4.439967412124011e-06,
      "loss": 0.1953,
      "step": 6012
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.123419109307755,
      "learning_rate": 4.438495988696703e-06,
      "loss": 0.5883,
      "step": 6013
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2487869620610375,
      "learning_rate": 4.437024614517051e-06,
      "loss": 0.1814,
      "step": 6014
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6883864867808294,
      "learning_rate": 4.435553289714106e-06,
      "loss": 0.2115,
      "step": 6015
    },
    {
      "epoch": 0.55,
      "grad_norm": 27.87513761500664,
      "learning_rate": 4.434082014416915e-06,
      "loss": 0.6044,
      "step": 6016
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.449096521429678,
      "learning_rate": 4.432610788754517e-06,
      "loss": 0.2225,
      "step": 6017
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4340074538393173,
      "learning_rate": 4.431139612855948e-06,
      "loss": 0.2155,
      "step": 6018
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3118638654659835,
      "learning_rate": 4.42966848685024e-06,
      "loss": 0.2089,
      "step": 6019
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2625402535769559,
      "learning_rate": 4.428197410866422e-06,
      "loss": 0.1938,
      "step": 6020
    },
    {
      "epoch": 0.55,
      "grad_norm": 8.348943014824544,
      "learning_rate": 4.426726385033518e-06,
      "loss": 0.4478,
      "step": 6021
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3964128639620315,
      "learning_rate": 4.425255409480542e-06,
      "loss": 0.195,
      "step": 6022
    },
    {
      "epoch": 0.55,
      "grad_norm": 5.836850533775301,
      "learning_rate": 4.4237844843365126e-06,
      "loss": 0.5045,
      "step": 6023
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3359933928687677,
      "learning_rate": 4.422313609730439e-06,
      "loss": 0.1776,
      "step": 6024
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4015744579515539,
      "learning_rate": 4.420842785791326e-06,
      "loss": 0.233,
      "step": 6025
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4156127532428693,
      "learning_rate": 4.419372012648175e-06,
      "loss": 0.2381,
      "step": 6026
    },
    {
      "epoch": 0.55,
      "grad_norm": 11.650149819527824,
      "learning_rate": 4.4179012904299835e-06,
      "loss": 0.7105,
      "step": 6027
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3648990749923369,
      "learning_rate": 4.416430619265742e-06,
      "loss": 0.1913,
      "step": 6028
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1340359220138558,
      "learning_rate": 4.414959999284441e-06,
      "loss": 0.1645,
      "step": 6029
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3321065902445766,
      "learning_rate": 4.41348943061506e-06,
      "loss": 0.1924,
      "step": 6030
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3913640393347897,
      "learning_rate": 4.41201891338658e-06,
      "loss": 0.1682,
      "step": 6031
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3419364697956213,
      "learning_rate": 4.4105484477279745e-06,
      "loss": 0.2013,
      "step": 6032
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.298914927442202,
      "learning_rate": 4.409078033768214e-06,
      "loss": 0.2066,
      "step": 6033
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3274376253061992,
      "learning_rate": 4.407607671636263e-06,
      "loss": 0.2055,
      "step": 6034
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2985122097181665,
      "learning_rate": 4.406137361461081e-06,
      "loss": 0.2146,
      "step": 6035
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4573767538223708,
      "learning_rate": 4.404667103371626e-06,
      "loss": 0.1724,
      "step": 6036
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3138109922903163,
      "learning_rate": 4.403196897496849e-06,
      "loss": 0.1765,
      "step": 6037
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3909088153876865,
      "learning_rate": 4.401726743965697e-06,
      "loss": 0.2067,
      "step": 6038
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5505279645778127,
      "learning_rate": 4.4002566429071105e-06,
      "loss": 0.2071,
      "step": 6039
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2960909285571849,
      "learning_rate": 4.39878659445003e-06,
      "loss": 0.1687,
      "step": 6040
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3859350600551712,
      "learning_rate": 4.397316598723385e-06,
      "loss": 0.1824,
      "step": 6041
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.313993008810389,
      "learning_rate": 4.395846655856109e-06,
      "loss": 0.1915,
      "step": 6042
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2656735585673327,
      "learning_rate": 4.394376765977121e-06,
      "loss": 0.1641,
      "step": 6043
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2783512742044227,
      "learning_rate": 4.392906929215342e-06,
      "loss": 0.1935,
      "step": 6044
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4245732419082129,
      "learning_rate": 4.391437145699687e-06,
      "loss": 0.1795,
      "step": 6045
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3976403147355427,
      "learning_rate": 4.389967415559066e-06,
      "loss": 0.1715,
      "step": 6046
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3429075899608438,
      "learning_rate": 4.388497738922384e-06,
      "loss": 0.1975,
      "step": 6047
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.164496315679673,
      "learning_rate": 4.38702811591854e-06,
      "loss": 0.6029,
      "step": 6048
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3409365145015322,
      "learning_rate": 4.3855585466764305e-06,
      "loss": 0.1995,
      "step": 6049
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.19796329004212,
      "learning_rate": 4.384089031324947e-06,
      "loss": 0.1414,
      "step": 6050
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4645501230066642,
      "learning_rate": 4.382619569992977e-06,
      "loss": 0.2501,
      "step": 6051
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4302963084666376,
      "learning_rate": 4.3811501628094e-06,
      "loss": 0.1864,
      "step": 6052
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.920811206468133,
      "learning_rate": 4.379680809903094e-06,
      "loss": 0.5383,
      "step": 6053
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5488101261817615,
      "learning_rate": 4.37821151140293e-06,
      "loss": 0.1983,
      "step": 6054
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3538014728878143,
      "learning_rate": 4.376742267437778e-06,
      "loss": 0.1784,
      "step": 6055
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5314616442985,
      "learning_rate": 4.375273078136496e-06,
      "loss": 0.2188,
      "step": 6056
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3451321168884192,
      "learning_rate": 4.373803943627946e-06,
      "loss": 0.1894,
      "step": 6057
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2215659056750832,
      "learning_rate": 4.372334864040979e-06,
      "loss": 0.1737,
      "step": 6058
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3733725155100869,
      "learning_rate": 4.370865839504443e-06,
      "loss": 0.2192,
      "step": 6059
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5784003135962883,
      "learning_rate": 4.369396870147182e-06,
      "loss": 0.1694,
      "step": 6060
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3491738971063474,
      "learning_rate": 4.367927956098033e-06,
      "loss": 0.1702,
      "step": 6061
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.38180873239824,
      "learning_rate": 4.366459097485832e-06,
      "loss": 0.2351,
      "step": 6062
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.38105330597719,
      "learning_rate": 4.364990294439405e-06,
      "loss": 0.2215,
      "step": 6063
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3799804262314523,
      "learning_rate": 4.36352154708758e-06,
      "loss": 0.2278,
      "step": 6064
    },
    {
      "epoch": 0.55,
      "grad_norm": 4.9853959038424005,
      "learning_rate": 4.362052855559171e-06,
      "loss": 0.584,
      "step": 6065
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4786072615419923,
      "learning_rate": 4.360584219982994e-06,
      "loss": 0.1959,
      "step": 6066
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.405633268999085,
      "learning_rate": 4.359115640487859e-06,
      "loss": 0.2078,
      "step": 6067
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3969193702439997,
      "learning_rate": 4.357647117202571e-06,
      "loss": 0.207,
      "step": 6068
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2156322902948429,
      "learning_rate": 4.356178650255925e-06,
      "loss": 0.1936,
      "step": 6069
    },
    {
      "epoch": 0.55,
      "grad_norm": 6.1518512836564225,
      "learning_rate": 4.35471023977672e-06,
      "loss": 0.5603,
      "step": 6070
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4268212502458144,
      "learning_rate": 4.353241885893745e-06,
      "loss": 0.1901,
      "step": 6071
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4145840876840874,
      "learning_rate": 4.351773588735783e-06,
      "loss": 0.1636,
      "step": 6072
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2838641250078087,
      "learning_rate": 4.350305348431612e-06,
      "loss": 0.1748,
      "step": 6073
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5034497884617657,
      "learning_rate": 4.34883716511001e-06,
      "loss": 0.2017,
      "step": 6074
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4683029783694872,
      "learning_rate": 4.347369038899744e-06,
      "loss": 0.1917,
      "step": 6075
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.331362354684668,
      "learning_rate": 4.345900969929579e-06,
      "loss": 0.1785,
      "step": 6076
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5262162245947164,
      "learning_rate": 4.344432958328274e-06,
      "loss": 0.2033,
      "step": 6077
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3986753508167244,
      "learning_rate": 4.342965004224585e-06,
      "loss": 0.1751,
      "step": 6078
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3003253630659748,
      "learning_rate": 4.3414971077472606e-06,
      "loss": 0.1966,
      "step": 6079
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3829600653397278,
      "learning_rate": 4.340029269025045e-06,
      "loss": 0.1876,
      "step": 6080
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.295838891936822,
      "learning_rate": 4.338561488186678e-06,
      "loss": 0.1933,
      "step": 6081
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5102869566907962,
      "learning_rate": 4.337093765360893e-06,
      "loss": 0.2224,
      "step": 6082
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2869481253457058,
      "learning_rate": 4.335626100676419e-06,
      "loss": 0.2066,
      "step": 6083
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3577354736138985,
      "learning_rate": 4.334158494261982e-06,
      "loss": 0.175,
      "step": 6084
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2558675480194594,
      "learning_rate": 4.332690946246298e-06,
      "loss": 0.1854,
      "step": 6085
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5522518052733996,
      "learning_rate": 4.3312234567580815e-06,
      "loss": 0.2161,
      "step": 6086
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2227402168886718,
      "learning_rate": 4.329756025926043e-06,
      "loss": 0.1722,
      "step": 6087
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.30345024940112,
      "learning_rate": 4.328288653878884e-06,
      "loss": 0.5059,
      "step": 6088
    },
    {
      "epoch": 0.56,
      "grad_norm": 9.984579327619958,
      "learning_rate": 4.326821340745304e-06,
      "loss": 0.4771,
      "step": 6089
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1324618720581145,
      "learning_rate": 4.3253540866539965e-06,
      "loss": 0.1462,
      "step": 6090
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.490185955042744,
      "learning_rate": 4.3238868917336485e-06,
      "loss": 0.2196,
      "step": 6091
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2486197485491957,
      "learning_rate": 4.322419756112943e-06,
      "loss": 0.1818,
      "step": 6092
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3693379728454158,
      "learning_rate": 4.320952679920557e-06,
      "loss": 0.2233,
      "step": 6093
    },
    {
      "epoch": 0.56,
      "grad_norm": 43.99541762316239,
      "learning_rate": 4.319485663285165e-06,
      "loss": 0.54,
      "step": 6094
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.547087043762263,
      "learning_rate": 4.318018706335431e-06,
      "loss": 0.2255,
      "step": 6095
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.8948222017618415,
      "learning_rate": 4.31655180920002e-06,
      "loss": 0.6673,
      "step": 6096
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.990686610347999,
      "learning_rate": 4.315084972007587e-06,
      "loss": 0.6049,
      "step": 6097
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.249350142046865,
      "learning_rate": 4.313618194886785e-06,
      "loss": 0.1868,
      "step": 6098
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.355571475067444,
      "learning_rate": 4.312151477966258e-06,
      "loss": 0.1752,
      "step": 6099
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1801203362048518,
      "learning_rate": 4.3106848213746486e-06,
      "loss": 0.168,
      "step": 6100
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.321184250989106,
      "learning_rate": 4.309218225240591e-06,
      "loss": 0.5766,
      "step": 6101
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4534094429178355,
      "learning_rate": 4.3077516896927155e-06,
      "loss": 0.2296,
      "step": 6102
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3483786487493332,
      "learning_rate": 4.306285214859649e-06,
      "loss": 0.2091,
      "step": 6103
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.173030053872959,
      "learning_rate": 4.304818800870009e-06,
      "loss": 0.5601,
      "step": 6104
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3020230681830436,
      "learning_rate": 4.303352447852412e-06,
      "loss": 0.2086,
      "step": 6105
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1964111933471704,
      "learning_rate": 4.301886155935465e-06,
      "loss": 0.1599,
      "step": 6106
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.6401212283557225,
      "learning_rate": 4.300419925247773e-06,
      "loss": 0.5109,
      "step": 6107
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3938043257169972,
      "learning_rate": 4.298953755917933e-06,
      "loss": 0.1724,
      "step": 6108
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.727868524846495,
      "learning_rate": 4.297487648074538e-06,
      "loss": 0.5038,
      "step": 6109
    },
    {
      "epoch": 0.56,
      "grad_norm": 7.61969601996812,
      "learning_rate": 4.2960216018461765e-06,
      "loss": 0.5207,
      "step": 6110
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4504904896522677,
      "learning_rate": 4.2945556173614296e-06,
      "loss": 0.1864,
      "step": 6111
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5086263825755135,
      "learning_rate": 4.293089694748874e-06,
      "loss": 0.2312,
      "step": 6112
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3441207652478229,
      "learning_rate": 4.291623834137082e-06,
      "loss": 0.1832,
      "step": 6113
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5063433746436332,
      "learning_rate": 4.290158035654618e-06,
      "loss": 0.2798,
      "step": 6114
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5428650220247668,
      "learning_rate": 4.288692299430045e-06,
      "loss": 0.2335,
      "step": 6115
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.228950868547029,
      "learning_rate": 4.287226625591914e-06,
      "loss": 0.1655,
      "step": 6116
    },
    {
      "epoch": 0.56,
      "grad_norm": 4.7648037303119315,
      "learning_rate": 4.285761014268775e-06,
      "loss": 0.5448,
      "step": 6117
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.43748567997142,
      "learning_rate": 4.284295465589175e-06,
      "loss": 0.2207,
      "step": 6118
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.329118425931067,
      "learning_rate": 4.28282997968165e-06,
      "loss": 0.1728,
      "step": 6119
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.289365135910745,
      "learning_rate": 4.281364556674733e-06,
      "loss": 0.2223,
      "step": 6120
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3833914058138272,
      "learning_rate": 4.279899196696953e-06,
      "loss": 0.1963,
      "step": 6121
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3847155820797163,
      "learning_rate": 4.27843389987683e-06,
      "loss": 0.1849,
      "step": 6122
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.273335860911896,
      "learning_rate": 4.27696866634288e-06,
      "loss": 0.1813,
      "step": 6123
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3417104427980806,
      "learning_rate": 4.2755034962236175e-06,
      "loss": 0.1718,
      "step": 6124
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4220710526343836,
      "learning_rate": 4.2740383896475434e-06,
      "loss": 0.2159,
      "step": 6125
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.058699426147524,
      "learning_rate": 4.27257334674316e-06,
      "loss": 0.6166,
      "step": 6126
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3816744175326645,
      "learning_rate": 4.271108367638959e-06,
      "loss": 0.1563,
      "step": 6127
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1602597348941524,
      "learning_rate": 4.2696434524634314e-06,
      "loss": 0.1284,
      "step": 6128
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6089074890284745,
      "learning_rate": 4.268178601345057e-06,
      "loss": 0.2617,
      "step": 6129
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4719560999398245,
      "learning_rate": 4.2667138144123175e-06,
      "loss": 0.2473,
      "step": 6130
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3618367981622823,
      "learning_rate": 4.26524909179368e-06,
      "loss": 0.2183,
      "step": 6131
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4487439601033862,
      "learning_rate": 4.263784433617613e-06,
      "loss": 0.1985,
      "step": 6132
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3710825606783064,
      "learning_rate": 4.2623198400125765e-06,
      "loss": 0.1675,
      "step": 6133
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4454534657762712,
      "learning_rate": 4.260855311107025e-06,
      "loss": 0.1693,
      "step": 6134
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4138093623094088,
      "learning_rate": 4.259390847029407e-06,
      "loss": 0.2103,
      "step": 6135
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.520631414844303,
      "learning_rate": 4.257926447908167e-06,
      "loss": 0.1739,
      "step": 6136
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3156129502244371,
      "learning_rate": 4.256462113871741e-06,
      "loss": 0.1668,
      "step": 6137
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.410139543176199,
      "learning_rate": 4.254997845048563e-06,
      "loss": 0.2,
      "step": 6138
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.359778998139213,
      "learning_rate": 4.253533641567058e-06,
      "loss": 0.1959,
      "step": 6139
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2058060801684523,
      "learning_rate": 4.252069503555645e-06,
      "loss": 0.1608,
      "step": 6140
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4732384215031435,
      "learning_rate": 4.250605431142741e-06,
      "loss": 0.1723,
      "step": 6141
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4187090666860158,
      "learning_rate": 4.2491414244567515e-06,
      "loss": 0.1709,
      "step": 6142
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3919027931069532,
      "learning_rate": 4.247677483626085e-06,
      "loss": 0.1831,
      "step": 6143
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.270749565578066,
      "learning_rate": 4.246213608779134e-06,
      "loss": 0.2071,
      "step": 6144
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.6112254452443384,
      "learning_rate": 4.2447498000442935e-06,
      "loss": 0.2008,
      "step": 6145
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4128990605527751,
      "learning_rate": 4.243286057549949e-06,
      "loss": 0.1599,
      "step": 6146
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4914835545330254,
      "learning_rate": 4.241822381424478e-06,
      "loss": 0.2546,
      "step": 6147
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.739043861630807,
      "learning_rate": 4.2403587717962564e-06,
      "loss": 0.6225,
      "step": 6148
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4725959580857018,
      "learning_rate": 4.2388952287936515e-06,
      "loss": 0.2012,
      "step": 6149
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3489045581733106,
      "learning_rate": 4.237431752545028e-06,
      "loss": 0.1863,
      "step": 6150
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4717095945727856,
      "learning_rate": 4.23596834317874e-06,
      "loss": 0.2387,
      "step": 6151
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4568249624366338,
      "learning_rate": 4.23450500082314e-06,
      "loss": 0.1985,
      "step": 6152
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2879214765098421,
      "learning_rate": 4.233041725606573e-06,
      "loss": 0.1835,
      "step": 6153
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2250708238511616,
      "learning_rate": 4.2315785176573765e-06,
      "loss": 0.1703,
      "step": 6154
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.324602875190679,
      "learning_rate": 4.230115377103886e-06,
      "loss": 0.2021,
      "step": 6155
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3391510950010954,
      "learning_rate": 4.228652304074426e-06,
      "loss": 0.2266,
      "step": 6156
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.415130381408991,
      "learning_rate": 4.22718929869732e-06,
      "loss": 0.2007,
      "step": 6157
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4939452023124817,
      "learning_rate": 4.225726361100881e-06,
      "loss": 0.1982,
      "step": 6158
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2286136295556565,
      "learning_rate": 4.224263491413421e-06,
      "loss": 0.1768,
      "step": 6159
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.368726309172582,
      "learning_rate": 4.222800689763241e-06,
      "loss": 0.1897,
      "step": 6160
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2460950283354364,
      "learning_rate": 4.2213379562786406e-06,
      "loss": 0.1629,
      "step": 6161
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5117976235619488,
      "learning_rate": 4.2198752910879105e-06,
      "loss": 0.1861,
      "step": 6162
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.247959430864002,
      "learning_rate": 4.218412694319337e-06,
      "loss": 0.1963,
      "step": 6163
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3751751877124352,
      "learning_rate": 4.216950166101199e-06,
      "loss": 0.1732,
      "step": 6164
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.944417028751711,
      "learning_rate": 4.215487706561768e-06,
      "loss": 0.7145,
      "step": 6165
    },
    {
      "epoch": 0.56,
      "grad_norm": 8.604108034359768,
      "learning_rate": 4.214025315829314e-06,
      "loss": 0.4172,
      "step": 6166
    },
    {
      "epoch": 0.56,
      "grad_norm": 6.173138548269165,
      "learning_rate": 4.212562994032099e-06,
      "loss": 0.6774,
      "step": 6167
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1981054311777768,
      "learning_rate": 4.211100741298375e-06,
      "loss": 0.1629,
      "step": 6168
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5869575914336023,
      "learning_rate": 4.209638557756396e-06,
      "loss": 0.2292,
      "step": 6169
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2089214504179986,
      "learning_rate": 4.208176443534401e-06,
      "loss": 0.1821,
      "step": 6170
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2054697096473814,
      "learning_rate": 4.206714398760629e-06,
      "loss": 0.1796,
      "step": 6171
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3204709751894235,
      "learning_rate": 4.205252423563313e-06,
      "loss": 0.2087,
      "step": 6172
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2607308135589195,
      "learning_rate": 4.203790518070675e-06,
      "loss": 0.1694,
      "step": 6173
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.236143008549389,
      "learning_rate": 4.202328682410935e-06,
      "loss": 0.1611,
      "step": 6174
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3763547023165963,
      "learning_rate": 4.200866916712305e-06,
      "loss": 0.1822,
      "step": 6175
    },
    {
      "epoch": 0.56,
      "grad_norm": 14.290832036215983,
      "learning_rate": 4.199405221102993e-06,
      "loss": 0.5703,
      "step": 6176
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3531715603684242,
      "learning_rate": 4.1979435957111984e-06,
      "loss": 0.2046,
      "step": 6177
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1674976146002964,
      "learning_rate": 4.196482040665117e-06,
      "loss": 0.1879,
      "step": 6178
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4769513775730159,
      "learning_rate": 4.195020556092935e-06,
      "loss": 0.2068,
      "step": 6179
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2363618543569546,
      "learning_rate": 4.193559142122837e-06,
      "loss": 0.1925,
      "step": 6180
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3858868193513674,
      "learning_rate": 4.192097798882996e-06,
      "loss": 0.2316,
      "step": 6181
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4048486805977252,
      "learning_rate": 4.190636526501582e-06,
      "loss": 0.1646,
      "step": 6182
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.47783365821116,
      "learning_rate": 4.18917532510676e-06,
      "loss": 0.6222,
      "step": 6183
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2865523951910771,
      "learning_rate": 4.1877141948266866e-06,
      "loss": 0.1457,
      "step": 6184
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3528497159712554,
      "learning_rate": 4.186253135789511e-06,
      "loss": 0.1933,
      "step": 6185
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4744262550851506,
      "learning_rate": 4.18479214812338e-06,
      "loss": 0.1784,
      "step": 6186
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3213368268288281,
      "learning_rate": 4.18333123195643e-06,
      "loss": 0.2172,
      "step": 6187
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.20639209545753,
      "learning_rate": 4.181870387416796e-06,
      "loss": 0.1611,
      "step": 6188
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.270679695897556,
      "learning_rate": 4.1804096146326025e-06,
      "loss": 0.1823,
      "step": 6189
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4653273839501428,
      "learning_rate": 4.1789489137319675e-06,
      "loss": 0.1677,
      "step": 6190
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3019748991836113,
      "learning_rate": 4.177488284843005e-06,
      "loss": 0.191,
      "step": 6191
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4186636778407664,
      "learning_rate": 4.1760277280938226e-06,
      "loss": 0.1715,
      "step": 6192
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5347110277803502,
      "learning_rate": 4.1745672436125205e-06,
      "loss": 0.2252,
      "step": 6193
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2705540862185527,
      "learning_rate": 4.173106831527194e-06,
      "loss": 0.1795,
      "step": 6194
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.22063181365621,
      "learning_rate": 4.1716464919659295e-06,
      "loss": 0.1674,
      "step": 6195
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3765713986439896,
      "learning_rate": 4.170186225056809e-06,
      "loss": 0.1885,
      "step": 6196
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4902537057281608,
      "learning_rate": 4.168726030927908e-06,
      "loss": 0.2205,
      "step": 6197
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2442796886098244,
      "learning_rate": 4.167265909707295e-06,
      "loss": 0.1961,
      "step": 6198
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.14210430896355,
      "learning_rate": 4.165805861523033e-06,
      "loss": 0.1575,
      "step": 6199
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4342130731217697,
      "learning_rate": 4.164345886503177e-06,
      "loss": 0.22,
      "step": 6200
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5174040844226742,
      "learning_rate": 4.162885984775777e-06,
      "loss": 0.2197,
      "step": 6201
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4124863666936287,
      "learning_rate": 4.161426156468876e-06,
      "loss": 0.2013,
      "step": 6202
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2724965268994206,
      "learning_rate": 4.159966401710511e-06,
      "loss": 0.1772,
      "step": 6203
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4540867121794585,
      "learning_rate": 4.158506720628713e-06,
      "loss": 0.1751,
      "step": 6204
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4588048117460706,
      "learning_rate": 4.157047113351504e-06,
      "loss": 0.1917,
      "step": 6205
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.259764433807396,
      "learning_rate": 4.155587580006903e-06,
      "loss": 0.1683,
      "step": 6206
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2858322293362618,
      "learning_rate": 4.154128120722921e-06,
      "loss": 0.1747,
      "step": 6207
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.776308694540035,
      "learning_rate": 4.15266873562756e-06,
      "loss": 0.541,
      "step": 6208
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4018356427100582,
      "learning_rate": 4.15120942484882e-06,
      "loss": 0.2099,
      "step": 6209
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2398111071519573,
      "learning_rate": 4.149750188514691e-06,
      "loss": 0.1558,
      "step": 6210
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3780745506573158,
      "learning_rate": 4.148291026753159e-06,
      "loss": 0.1903,
      "step": 6211
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4909719892627953,
      "learning_rate": 4.146831939692201e-06,
      "loss": 0.2217,
      "step": 6212
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.201275879014935,
      "learning_rate": 4.14537292745979e-06,
      "loss": 0.4852,
      "step": 6213
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3531995644917578,
      "learning_rate": 4.1439139901838896e-06,
      "loss": 0.1679,
      "step": 6214
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2798534691355785,
      "learning_rate": 4.142455127992461e-06,
      "loss": 0.2197,
      "step": 6215
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4036280846155063,
      "learning_rate": 4.140996341013453e-06,
      "loss": 0.1662,
      "step": 6216
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.119364922604595,
      "learning_rate": 4.139537629374814e-06,
      "loss": 0.6448,
      "step": 6217
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3923742216149446,
      "learning_rate": 4.1380789932044794e-06,
      "loss": 0.2179,
      "step": 6218
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2191804656375873,
      "learning_rate": 4.136620432630384e-06,
      "loss": 0.1755,
      "step": 6219
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2899579076871437,
      "learning_rate": 4.135161947780452e-06,
      "loss": 0.1669,
      "step": 6220
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2051320157521552,
      "learning_rate": 4.133703538782603e-06,
      "loss": 0.1797,
      "step": 6221
    },
    {
      "epoch": 0.57,
      "grad_norm": 8.376782290556054,
      "learning_rate": 4.132245205764749e-06,
      "loss": 0.5776,
      "step": 6222
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.41743675862204,
      "learning_rate": 4.130786948854796e-06,
      "loss": 0.219,
      "step": 6223
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2744948243350815,
      "learning_rate": 4.129328768180642e-06,
      "loss": 0.1814,
      "step": 6224
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2886877798935605,
      "learning_rate": 4.12787066387018e-06,
      "loss": 0.2012,
      "step": 6225
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2668795482550401,
      "learning_rate": 4.126412636051294e-06,
      "loss": 0.2397,
      "step": 6226
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2756738080513557,
      "learning_rate": 4.124954684851864e-06,
      "loss": 0.1515,
      "step": 6227
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.258549556675159,
      "learning_rate": 4.123496810399763e-06,
      "loss": 0.1818,
      "step": 6228
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.365140975463284,
      "learning_rate": 4.1220390128228545e-06,
      "loss": 0.1622,
      "step": 6229
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5407840587650938,
      "learning_rate": 4.120581292248998e-06,
      "loss": 0.2637,
      "step": 6230
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.452078092754755,
      "learning_rate": 4.119123648806046e-06,
      "loss": 0.19,
      "step": 6231
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.859838677883448,
      "learning_rate": 4.117666082621843e-06,
      "loss": 0.7618,
      "step": 6232
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4996162472892687,
      "learning_rate": 4.116208593824227e-06,
      "loss": 0.1555,
      "step": 6233
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.9455425827054365,
      "learning_rate": 4.114751182541029e-06,
      "loss": 0.6432,
      "step": 6234
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2984345505042523,
      "learning_rate": 4.113293848900074e-06,
      "loss": 0.177,
      "step": 6235
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3595411096217425,
      "learning_rate": 4.1118365930291805e-06,
      "loss": 0.2153,
      "step": 6236
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2752463358057322,
      "learning_rate": 4.110379415056161e-06,
      "loss": 0.1672,
      "step": 6237
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1939863662866168,
      "learning_rate": 4.108922315108816e-06,
      "loss": 0.1709,
      "step": 6238
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2344788975234204,
      "learning_rate": 4.107465293314946e-06,
      "loss": 0.1609,
      "step": 6239
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4582699614954082,
      "learning_rate": 4.106008349802341e-06,
      "loss": 0.1961,
      "step": 6240
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.115082495187881,
      "learning_rate": 4.104551484698785e-06,
      "loss": 0.4828,
      "step": 6241
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.34734668672082,
      "learning_rate": 4.103094698132052e-06,
      "loss": 0.1518,
      "step": 6242
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3543475503123632,
      "learning_rate": 4.101637990229915e-06,
      "loss": 0.1969,
      "step": 6243
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3760661138120662,
      "learning_rate": 4.100181361120136e-06,
      "loss": 0.2018,
      "step": 6244
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2519011523370318,
      "learning_rate": 4.098724810930472e-06,
      "loss": 0.1471,
      "step": 6245
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5013816793007595,
      "learning_rate": 4.09726833978867e-06,
      "loss": 0.2159,
      "step": 6246
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1392665610151056,
      "learning_rate": 4.095811947822474e-06,
      "loss": 0.1528,
      "step": 6247
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2913242348520275,
      "learning_rate": 4.094355635159619e-06,
      "loss": 0.1803,
      "step": 6248
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3539475944344779,
      "learning_rate": 4.092899401927836e-06,
      "loss": 0.149,
      "step": 6249
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.496282744759611,
      "learning_rate": 4.091443248254841e-06,
      "loss": 0.2176,
      "step": 6250
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.247680123068209,
      "learning_rate": 4.089987174268352e-06,
      "loss": 0.1758,
      "step": 6251
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4467384132457881,
      "learning_rate": 4.088531180096075e-06,
      "loss": 0.193,
      "step": 6252
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.340202700767549,
      "learning_rate": 4.087075265865711e-06,
      "loss": 0.6944,
      "step": 6253
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3962804919122376,
      "learning_rate": 4.0856194317049535e-06,
      "loss": 0.1969,
      "step": 6254
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2433936264472925,
      "learning_rate": 4.0841636777414885e-06,
      "loss": 0.2192,
      "step": 6255
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6416314608121516,
      "learning_rate": 4.082708004102996e-06,
      "loss": 0.2296,
      "step": 6256
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3364002096298702,
      "learning_rate": 4.081252410917148e-06,
      "loss": 0.2007,
      "step": 6257
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2335329226063274,
      "learning_rate": 4.0797968983116106e-06,
      "loss": 0.1823,
      "step": 6258
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3290078068730091,
      "learning_rate": 4.0783414664140406e-06,
      "loss": 0.1803,
      "step": 6259
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3211358026137447,
      "learning_rate": 4.07688611535209e-06,
      "loss": 0.1833,
      "step": 6260
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.439087028791798,
      "learning_rate": 4.075430845253402e-06,
      "loss": 0.1941,
      "step": 6261
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.457363556203414,
      "learning_rate": 4.073975656245613e-06,
      "loss": 0.2334,
      "step": 6262
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5314532169948225,
      "learning_rate": 4.072520548456356e-06,
      "loss": 0.2235,
      "step": 6263
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4263220946460051,
      "learning_rate": 4.07106552201325e-06,
      "loss": 0.1861,
      "step": 6264
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.296717158149837,
      "learning_rate": 4.069610577043912e-06,
      "loss": 0.2005,
      "step": 6265
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.33828720923131,
      "learning_rate": 4.068155713675952e-06,
      "loss": 0.18,
      "step": 6266
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4386478002905907,
      "learning_rate": 4.066700932036969e-06,
      "loss": 0.2031,
      "step": 6267
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3588765716592406,
      "learning_rate": 4.065246232254557e-06,
      "loss": 0.1844,
      "step": 6268
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4125828242342757,
      "learning_rate": 4.063791614456304e-06,
      "loss": 0.1702,
      "step": 6269
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3960546588084235,
      "learning_rate": 4.062337078769789e-06,
      "loss": 0.2112,
      "step": 6270
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3592043101792377,
      "learning_rate": 4.060882625322586e-06,
      "loss": 0.1866,
      "step": 6271
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2537227081931304,
      "learning_rate": 4.059428254242259e-06,
      "loss": 0.1831,
      "step": 6272
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4903594573582983,
      "learning_rate": 4.057973965656365e-06,
      "loss": 0.1936,
      "step": 6273
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4913939328936565,
      "learning_rate": 4.056519759692457e-06,
      "loss": 0.1908,
      "step": 6274
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2512951692525096,
      "learning_rate": 4.05506563647808e-06,
      "loss": 0.2089,
      "step": 6275
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.196756356725271,
      "learning_rate": 4.053611596140765e-06,
      "loss": 0.7029,
      "step": 6276
    },
    {
      "epoch": 0.57,
      "grad_norm": 7.410320631939319,
      "learning_rate": 4.0521576388080455e-06,
      "loss": 0.5767,
      "step": 6277
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4537608725412756,
      "learning_rate": 4.050703764607441e-06,
      "loss": 0.2379,
      "step": 6278
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3414341548497373,
      "learning_rate": 4.049249973666468e-06,
      "loss": 0.1506,
      "step": 6279
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.431099683954377,
      "learning_rate": 4.047796266112631e-06,
      "loss": 0.1772,
      "step": 6280
    },
    {
      "epoch": 0.57,
      "grad_norm": 9.261441224042292,
      "learning_rate": 4.046342642073433e-06,
      "loss": 0.7867,
      "step": 6281
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.274512511211927,
      "learning_rate": 4.0448891016763645e-06,
      "loss": 0.1819,
      "step": 6282
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4385875862316115,
      "learning_rate": 4.043435645048911e-06,
      "loss": 0.1929,
      "step": 6283
    },
    {
      "epoch": 0.57,
      "grad_norm": 6.572430420556811,
      "learning_rate": 4.041982272318552e-06,
      "loss": 0.5676,
      "step": 6284
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3478780401127919,
      "learning_rate": 4.040528983612755e-06,
      "loss": 0.1463,
      "step": 6285
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5001931377538005,
      "learning_rate": 4.039075779058985e-06,
      "loss": 0.1724,
      "step": 6286
    },
    {
      "epoch": 0.57,
      "grad_norm": 5.020525459139824,
      "learning_rate": 4.037622658784698e-06,
      "loss": 0.5648,
      "step": 6287
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2466642480914405,
      "learning_rate": 4.0361696229173415e-06,
      "loss": 0.147,
      "step": 6288
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3337259178946326,
      "learning_rate": 4.034716671584357e-06,
      "loss": 0.1643,
      "step": 6289
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.289813192853057,
      "learning_rate": 4.033263804913178e-06,
      "loss": 0.2061,
      "step": 6290
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.024157407891659,
      "learning_rate": 4.03181102303123e-06,
      "loss": 0.656,
      "step": 6291
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2768471283610832,
      "learning_rate": 4.030358326065934e-06,
      "loss": 0.1906,
      "step": 6292
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.364461025715933,
      "learning_rate": 4.028905714144698e-06,
      "loss": 0.1586,
      "step": 6293
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.25669704081196,
      "learning_rate": 4.027453187394927e-06,
      "loss": 0.1913,
      "step": 6294
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3949337518932212,
      "learning_rate": 4.026000745944018e-06,
      "loss": 0.1967,
      "step": 6295
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3331258102561008,
      "learning_rate": 4.02454838991936e-06,
      "loss": 0.1645,
      "step": 6296
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.17077041024621,
      "learning_rate": 4.0230961194483325e-06,
      "loss": 0.1474,
      "step": 6297
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3856316101364774,
      "learning_rate": 4.021643934658311e-06,
      "loss": 0.197,
      "step": 6298
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2863047307102449,
      "learning_rate": 4.020191835676662e-06,
      "loss": 0.1582,
      "step": 6299
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4464272867051335,
      "learning_rate": 4.018739822630743e-06,
      "loss": 0.199,
      "step": 6300
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2815176025737072,
      "learning_rate": 4.017287895647907e-06,
      "loss": 0.2054,
      "step": 6301
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.496602026067309,
      "learning_rate": 4.015836054855496e-06,
      "loss": 0.2275,
      "step": 6302
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4597439190098032,
      "learning_rate": 4.014384300380847e-06,
      "loss": 0.1829,
      "step": 6303
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3980790035811912,
      "learning_rate": 4.012932632351287e-06,
      "loss": 0.2073,
      "step": 6304
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2986519648575257,
      "learning_rate": 4.01148105089414e-06,
      "loss": 0.202,
      "step": 6305
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3963473322673936,
      "learning_rate": 4.010029556136717e-06,
      "loss": 0.2036,
      "step": 6306
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6132675598148674,
      "learning_rate": 4.0085781482063245e-06,
      "loss": 0.2214,
      "step": 6307
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.507446985578268,
      "learning_rate": 4.007126827230261e-06,
      "loss": 0.7811,
      "step": 6308
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5161461143450565,
      "learning_rate": 4.005675593335818e-06,
      "loss": 0.234,
      "step": 6309
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.078576442159267,
      "learning_rate": 4.004224446650277e-06,
      "loss": 0.4504,
      "step": 6310
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4898401406376218,
      "learning_rate": 4.002773387300912e-06,
      "loss": 0.1931,
      "step": 6311
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4619431945218861,
      "learning_rate": 4.001322415414993e-06,
      "loss": 0.2395,
      "step": 6312
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4584947888114415,
      "learning_rate": 3.999871531119779e-06,
      "loss": 0.198,
      "step": 6313
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3510781570968742,
      "learning_rate": 3.998420734542522e-06,
      "loss": 0.1826,
      "step": 6314
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3129154314729938,
      "learning_rate": 3.996970025810467e-06,
      "loss": 0.2408,
      "step": 6315
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3658884079459344,
      "learning_rate": 3.995519405050851e-06,
      "loss": 0.1695,
      "step": 6316
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.471421057050766,
      "learning_rate": 3.9940688723909025e-06,
      "loss": 0.6024,
      "step": 6317
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1747765520439832,
      "learning_rate": 3.992618427957845e-06,
      "loss": 0.1508,
      "step": 6318
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1982756685709892,
      "learning_rate": 3.991168071878889e-06,
      "loss": 0.157,
      "step": 6319
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.237307858752615,
      "learning_rate": 3.989717804281242e-06,
      "loss": 0.2005,
      "step": 6320
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2466881876373073,
      "learning_rate": 3.988267625292102e-06,
      "loss": 0.1575,
      "step": 6321
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.172942414459212,
      "learning_rate": 3.98681753503866e-06,
      "loss": 0.1931,
      "step": 6322
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3800968192740315,
      "learning_rate": 3.985367533648097e-06,
      "loss": 0.1989,
      "step": 6323
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3361740302245297,
      "learning_rate": 3.983917621247591e-06,
      "loss": 0.2228,
      "step": 6324
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4023256501650194,
      "learning_rate": 3.982467797964305e-06,
      "loss": 0.2287,
      "step": 6325
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2070142721430295,
      "learning_rate": 3.981018063925401e-06,
      "loss": 0.1538,
      "step": 6326
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.251336101330605,
      "learning_rate": 3.979568419258031e-06,
      "loss": 0.5796,
      "step": 6327
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3152497190629404,
      "learning_rate": 3.978118864089335e-06,
      "loss": 0.1725,
      "step": 6328
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3109179018220605,
      "learning_rate": 3.976669398546451e-06,
      "loss": 0.1761,
      "step": 6329
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4483531288055203,
      "learning_rate": 3.975220022756507e-06,
      "loss": 0.1518,
      "step": 6330
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.692415863810205,
      "learning_rate": 3.973770736846622e-06,
      "loss": 0.4163,
      "step": 6331
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3764726884215963,
      "learning_rate": 3.972321540943909e-06,
      "loss": 0.1885,
      "step": 6332
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5203919720910768,
      "learning_rate": 3.970872435175472e-06,
      "loss": 0.2296,
      "step": 6333
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4402128823706228,
      "learning_rate": 3.969423419668406e-06,
      "loss": 0.1909,
      "step": 6334
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.321772259841988,
      "learning_rate": 3.967974494549803e-06,
      "loss": 0.1506,
      "step": 6335
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3728033468842005,
      "learning_rate": 3.966525659946739e-06,
      "loss": 0.1931,
      "step": 6336
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.134661751968677,
      "learning_rate": 3.9650769159862875e-06,
      "loss": 0.6947,
      "step": 6337
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.22529365810849,
      "learning_rate": 3.963628262795515e-06,
      "loss": 0.1429,
      "step": 6338
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3130460026896176,
      "learning_rate": 3.962179700501477e-06,
      "loss": 0.1817,
      "step": 6339
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1923937888315683,
      "learning_rate": 3.960731229231222e-06,
      "loss": 0.1623,
      "step": 6340
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3753848716035888,
      "learning_rate": 3.95928284911179e-06,
      "loss": 0.2029,
      "step": 6341
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.498410396325084,
      "learning_rate": 3.957834560270214e-06,
      "loss": 0.1805,
      "step": 6342
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4991124429109701,
      "learning_rate": 3.95638636283352e-06,
      "loss": 0.207,
      "step": 6343
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3030339182504433,
      "learning_rate": 3.954938256928724e-06,
      "loss": 0.1759,
      "step": 6344
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5505533021222233,
      "learning_rate": 3.9534902426828325e-06,
      "loss": 0.1809,
      "step": 6345
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.713930464231492,
      "learning_rate": 3.952042320222848e-06,
      "loss": 0.5262,
      "step": 6346
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.412215439384078,
      "learning_rate": 3.9505944896757635e-06,
      "loss": 0.2017,
      "step": 6347
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.262420031461272,
      "learning_rate": 3.949146751168562e-06,
      "loss": 0.1848,
      "step": 6348
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2787908670162709,
      "learning_rate": 3.94769910482822e-06,
      "loss": 0.138,
      "step": 6349
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.039837552611814,
      "learning_rate": 3.946251550781706e-06,
      "loss": 0.5142,
      "step": 6350
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.502146700317018,
      "learning_rate": 3.94480408915598e-06,
      "loss": 0.2234,
      "step": 6351
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.405601893831673,
      "learning_rate": 3.943356720077996e-06,
      "loss": 0.1896,
      "step": 6352
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4196398873896907,
      "learning_rate": 3.941909443674696e-06,
      "loss": 0.2205,
      "step": 6353
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4256655011489394,
      "learning_rate": 3.940462260073016e-06,
      "loss": 0.1989,
      "step": 6354
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5602783962647855,
      "learning_rate": 3.939015169399883e-06,
      "loss": 0.2141,
      "step": 6355
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.410875697807696,
      "learning_rate": 3.937568171782217e-06,
      "loss": 0.2434,
      "step": 6356
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3475131185002176,
      "learning_rate": 3.936121267346931e-06,
      "loss": 0.2089,
      "step": 6357
    },
    {
      "epoch": 0.58,
      "grad_norm": 7.039301292348148,
      "learning_rate": 3.934674456220926e-06,
      "loss": 0.5153,
      "step": 6358
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4327048993852434,
      "learning_rate": 3.9332277385310985e-06,
      "loss": 0.2105,
      "step": 6359
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4314440406580535,
      "learning_rate": 3.931781114404335e-06,
      "loss": 0.1876,
      "step": 6360
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.462880720316695,
      "learning_rate": 3.930334583967514e-06,
      "loss": 0.2247,
      "step": 6361
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4188185858381173,
      "learning_rate": 3.9288881473475055e-06,
      "loss": 0.2045,
      "step": 6362
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3852959295278482,
      "learning_rate": 3.9274418046711714e-06,
      "loss": 0.1769,
      "step": 6363
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3028714045923366,
      "learning_rate": 3.925995556065367e-06,
      "loss": 0.1648,
      "step": 6364
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3831390073711927,
      "learning_rate": 3.924549401656937e-06,
      "loss": 0.228,
      "step": 6365
    },
    {
      "epoch": 0.58,
      "grad_norm": 9.021587562281818,
      "learning_rate": 3.923103341572718e-06,
      "loss": 0.6818,
      "step": 6366
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2669351020818214,
      "learning_rate": 3.921657375939541e-06,
      "loss": 0.1861,
      "step": 6367
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4731891607877567,
      "learning_rate": 3.920211504884226e-06,
      "loss": 0.1814,
      "step": 6368
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5379542089557259,
      "learning_rate": 3.918765728533586e-06,
      "loss": 0.2025,
      "step": 6369
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2618316270728962,
      "learning_rate": 3.917320047014424e-06,
      "loss": 0.203,
      "step": 6370
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4306274703175175,
      "learning_rate": 3.915874460453537e-06,
      "loss": 0.1976,
      "step": 6371
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4173846432394164,
      "learning_rate": 3.9144289689777126e-06,
      "loss": 0.2021,
      "step": 6372
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.321650365439275,
      "learning_rate": 3.912983572713729e-06,
      "loss": 0.1556,
      "step": 6373
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.347979840725068,
      "learning_rate": 3.911538271788359e-06,
      "loss": 0.1805,
      "step": 6374
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.276578805682579,
      "learning_rate": 3.9100930663283634e-06,
      "loss": 0.4331,
      "step": 6375
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2331829004598513,
      "learning_rate": 3.908647956460497e-06,
      "loss": 0.1614,
      "step": 6376
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.461332141964392,
      "learning_rate": 3.907202942311506e-06,
      "loss": 0.1966,
      "step": 6377
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.288639595989816,
      "learning_rate": 3.905758024008129e-06,
      "loss": 0.1678,
      "step": 6378
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3039002737320005,
      "learning_rate": 3.904313201677092e-06,
      "loss": 0.1754,
      "step": 6379
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.187446714729525,
      "learning_rate": 3.902868475445117e-06,
      "loss": 0.1291,
      "step": 6380
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.258906072662518,
      "learning_rate": 3.901423845438916e-06,
      "loss": 0.144,
      "step": 6381
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4563568574424886,
      "learning_rate": 3.899979311785193e-06,
      "loss": 0.1594,
      "step": 6382
    },
    {
      "epoch": 0.58,
      "grad_norm": 5.168170502984,
      "learning_rate": 3.898534874610643e-06,
      "loss": 0.6085,
      "step": 6383
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3972766540413797,
      "learning_rate": 3.897090534041954e-06,
      "loss": 0.2228,
      "step": 6384
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.058419953865987,
      "learning_rate": 3.895646290205803e-06,
      "loss": 0.6039,
      "step": 6385
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5543577833703992,
      "learning_rate": 3.8942021432288604e-06,
      "loss": 0.2463,
      "step": 6386
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3840734569825184,
      "learning_rate": 3.892758093237788e-06,
      "loss": 0.2106,
      "step": 6387
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3986765357188824,
      "learning_rate": 3.891314140359237e-06,
      "loss": 0.198,
      "step": 6388
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.287749011200535,
      "learning_rate": 3.889870284719852e-06,
      "loss": 0.1571,
      "step": 6389
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4291053898451582,
      "learning_rate": 3.888426526446271e-06,
      "loss": 0.2059,
      "step": 6390
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4776741777325104,
      "learning_rate": 3.8869828656651184e-06,
      "loss": 0.1978,
      "step": 6391
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6381813419886713,
      "learning_rate": 3.8855393025030156e-06,
      "loss": 0.2683,
      "step": 6392
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4891356731426464,
      "learning_rate": 3.884095837086571e-06,
      "loss": 0.2155,
      "step": 6393
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6132325024930159,
      "learning_rate": 3.882652469542386e-06,
      "loss": 0.1966,
      "step": 6394
    },
    {
      "epoch": 0.58,
      "grad_norm": 6.549878279321853,
      "learning_rate": 3.881209199997055e-06,
      "loss": 0.5608,
      "step": 6395
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2522171649187717,
      "learning_rate": 3.879766028577162e-06,
      "loss": 0.1804,
      "step": 6396
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5059143338111922,
      "learning_rate": 3.878322955409281e-06,
      "loss": 0.1815,
      "step": 6397
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5364453358561987,
      "learning_rate": 3.876879980619982e-06,
      "loss": 0.203,
      "step": 6398
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3094599042513801,
      "learning_rate": 3.875437104335821e-06,
      "loss": 0.2062,
      "step": 6399
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2734263289052492,
      "learning_rate": 3.873994326683349e-06,
      "loss": 0.1776,
      "step": 6400
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3096951691828707,
      "learning_rate": 3.872551647789108e-06,
      "loss": 0.1947,
      "step": 6401
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.98366397704182,
      "learning_rate": 3.8711090677796295e-06,
      "loss": 0.5782,
      "step": 6402
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.360143081153869,
      "learning_rate": 3.869666586781438e-06,
      "loss": 0.4274,
      "step": 6403
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3630455139373652,
      "learning_rate": 3.868224204921048e-06,
      "loss": 0.2001,
      "step": 6404
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2690182264421273,
      "learning_rate": 3.866781922324967e-06,
      "loss": 0.1802,
      "step": 6405
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4490598857835266,
      "learning_rate": 3.865339739119692e-06,
      "loss": 0.167,
      "step": 6406
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.246486839921685,
      "learning_rate": 3.863897655431712e-06,
      "loss": 0.7516,
      "step": 6407
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.410385145995033,
      "learning_rate": 3.86245567138751e-06,
      "loss": 0.7105,
      "step": 6408
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.062132137119639,
      "learning_rate": 3.861013787113553e-06,
      "loss": 0.5929,
      "step": 6409
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.223698190241511,
      "learning_rate": 3.859572002736307e-06,
      "loss": 0.1839,
      "step": 6410
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.424015528394149,
      "learning_rate": 3.858130318382226e-06,
      "loss": 0.197,
      "step": 6411
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3120284709096548,
      "learning_rate": 3.856688734177755e-06,
      "loss": 0.1706,
      "step": 6412
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2333107931907035,
      "learning_rate": 3.855247250249331e-06,
      "loss": 0.1464,
      "step": 6413
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4350602831523511,
      "learning_rate": 3.853805866723381e-06,
      "loss": 0.1883,
      "step": 6414
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5083740878272018,
      "learning_rate": 3.852364583726324e-06,
      "loss": 0.2237,
      "step": 6415
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3620503386872682,
      "learning_rate": 3.850923401384569e-06,
      "loss": 0.2061,
      "step": 6416
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.316668116244767,
      "learning_rate": 3.849482319824521e-06,
      "loss": 0.1924,
      "step": 6417
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4231875364913062,
      "learning_rate": 3.848041339172569e-06,
      "loss": 0.1857,
      "step": 6418
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.418193209902338,
      "learning_rate": 3.846600459555098e-06,
      "loss": 0.2074,
      "step": 6419
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.480126606898877,
      "learning_rate": 3.845159681098483e-06,
      "loss": 0.2112,
      "step": 6420
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3504083280118413,
      "learning_rate": 3.84371900392909e-06,
      "loss": 0.1833,
      "step": 6421
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4039198385630935,
      "learning_rate": 3.842278428173275e-06,
      "loss": 0.204,
      "step": 6422
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3441646721569884,
      "learning_rate": 3.840837953957387e-06,
      "loss": 0.1759,
      "step": 6423
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2501131389974558,
      "learning_rate": 3.839397581407764e-06,
      "loss": 0.183,
      "step": 6424
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.869008920348603,
      "learning_rate": 3.837957310650738e-06,
      "loss": 0.6864,
      "step": 6425
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.432654518392217,
      "learning_rate": 3.836517141812629e-06,
      "loss": 0.1927,
      "step": 6426
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5489954188390551,
      "learning_rate": 3.835077075019751e-06,
      "loss": 0.1911,
      "step": 6427
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3029747881761728,
      "learning_rate": 3.833637110398406e-06,
      "loss": 0.1712,
      "step": 6428
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.300160092349486,
      "learning_rate": 3.832197248074889e-06,
      "loss": 0.4125,
      "step": 6429
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3160056826278153,
      "learning_rate": 3.830757488175487e-06,
      "loss": 0.1776,
      "step": 6430
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.34102486438877,
      "learning_rate": 3.829317830826474e-06,
      "loss": 0.1601,
      "step": 6431
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3350646888428057,
      "learning_rate": 3.827878276154119e-06,
      "loss": 0.1797,
      "step": 6432
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.320235302846297,
      "learning_rate": 3.82643882428468e-06,
      "loss": 0.1527,
      "step": 6433
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3349143742997096,
      "learning_rate": 3.8249994753444075e-06,
      "loss": 0.2003,
      "step": 6434
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6029478940211201,
      "learning_rate": 3.8235602294595405e-06,
      "loss": 0.1902,
      "step": 6435
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3685739979650038,
      "learning_rate": 3.822121086756313e-06,
      "loss": 0.2153,
      "step": 6436
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.16747924402424,
      "learning_rate": 3.820682047360945e-06,
      "loss": 0.6475,
      "step": 6437
    },
    {
      "epoch": 0.59,
      "grad_norm": 9.89127551732955,
      "learning_rate": 3.819243111399652e-06,
      "loss": 0.6281,
      "step": 6438
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3113576883022218,
      "learning_rate": 3.8178042789986355e-06,
      "loss": 0.1791,
      "step": 6439
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2324905084595466,
      "learning_rate": 3.816365550284093e-06,
      "loss": 0.2043,
      "step": 6440
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2326698583175597,
      "learning_rate": 3.81492692538221e-06,
      "loss": 0.1667,
      "step": 6441
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.33079858433721,
      "learning_rate": 3.8134884044191634e-06,
      "loss": 0.4488,
      "step": 6442
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.445871083397518,
      "learning_rate": 3.8120499875211216e-06,
      "loss": 0.679,
      "step": 6443
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.626306500751672,
      "learning_rate": 3.810611674814242e-06,
      "loss": 0.7246,
      "step": 6444
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4834212115301244,
      "learning_rate": 3.809173466424676e-06,
      "loss": 0.1865,
      "step": 6445
    },
    {
      "epoch": 0.59,
      "grad_norm": 17.26028565598055,
      "learning_rate": 3.8077353624785627e-06,
      "loss": 0.5868,
      "step": 6446
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4904324425385647,
      "learning_rate": 3.8062973631020363e-06,
      "loss": 0.2056,
      "step": 6447
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3950848917628615,
      "learning_rate": 3.8048594684212144e-06,
      "loss": 0.1995,
      "step": 6448
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3496557458575273,
      "learning_rate": 3.803421678562213e-06,
      "loss": 0.2214,
      "step": 6449
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4637344211467955,
      "learning_rate": 3.801983993651135e-06,
      "loss": 0.1878,
      "step": 6450
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.258610555782424,
      "learning_rate": 3.8005464138140753e-06,
      "loss": 0.1727,
      "step": 6451
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4175014615944659,
      "learning_rate": 3.7991089391771185e-06,
      "loss": 0.2054,
      "step": 6452
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2791899605143189,
      "learning_rate": 3.7976715698663425e-06,
      "loss": 0.1517,
      "step": 6453
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.215709236455457,
      "learning_rate": 3.796234306007812e-06,
      "loss": 0.1455,
      "step": 6454
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3099003211524247,
      "learning_rate": 3.794797147727588e-06,
      "loss": 0.1838,
      "step": 6455
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4470261090438625,
      "learning_rate": 3.7933600951517148e-06,
      "loss": 0.2073,
      "step": 6456
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4712636349948787,
      "learning_rate": 3.7919231484062334e-06,
      "loss": 0.2163,
      "step": 6457
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.401662372826977,
      "learning_rate": 3.790486307617174e-06,
      "loss": 0.1853,
      "step": 6458
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3582763793886625,
      "learning_rate": 3.789049572910557e-06,
      "loss": 0.1973,
      "step": 6459
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2147969681046542,
      "learning_rate": 3.7876129444123927e-06,
      "loss": 0.1869,
      "step": 6460
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.282244211756957,
      "learning_rate": 3.7861764222486843e-06,
      "loss": 0.6119,
      "step": 6461
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3248489349152932,
      "learning_rate": 3.784740006545422e-06,
      "loss": 0.1464,
      "step": 6462
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4522975157298952,
      "learning_rate": 3.783303697428593e-06,
      "loss": 0.1807,
      "step": 6463
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.45542544219993,
      "learning_rate": 3.7818674950241685e-06,
      "loss": 0.8159,
      "step": 6464
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.599544606623719,
      "learning_rate": 3.7804313994581143e-06,
      "loss": 0.2264,
      "step": 6465
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4726388455270956,
      "learning_rate": 3.7789954108563844e-06,
      "loss": 0.1913,
      "step": 6466
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3580453659896847,
      "learning_rate": 3.777559529344925e-06,
      "loss": 0.1762,
      "step": 6467
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.589967283194611,
      "learning_rate": 3.776123755049673e-06,
      "loss": 0.2474,
      "step": 6468
    },
    {
      "epoch": 0.59,
      "grad_norm": 4.15428521461457,
      "learning_rate": 3.7746880880965563e-06,
      "loss": 0.4838,
      "step": 6469
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.380209388410635,
      "learning_rate": 3.7732525286114906e-06,
      "loss": 0.1995,
      "step": 6470
    },
    {
      "epoch": 0.59,
      "grad_norm": 8.327705622378476,
      "learning_rate": 3.7718170767203843e-06,
      "loss": 0.5951,
      "step": 6471
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.0722217087841015,
      "learning_rate": 3.7703817325491367e-06,
      "loss": 0.4089,
      "step": 6472
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2448603919831587,
      "learning_rate": 3.7689464962236367e-06,
      "loss": 0.1578,
      "step": 6473
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6161872134182345,
      "learning_rate": 3.767511367869765e-06,
      "loss": 0.2029,
      "step": 6474
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3043877809323032,
      "learning_rate": 3.766076347613391e-06,
      "loss": 0.178,
      "step": 6475
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2049221749940346,
      "learning_rate": 3.7646414355803764e-06,
      "loss": 0.1683,
      "step": 6476
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4741029371631351,
      "learning_rate": 3.7632066318965715e-06,
      "loss": 0.1725,
      "step": 6477
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2691628851249115,
      "learning_rate": 3.7617719366878204e-06,
      "loss": 0.1671,
      "step": 6478
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3237361232780422,
      "learning_rate": 3.7603373500799527e-06,
      "loss": 0.1965,
      "step": 6479
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2271358002444874,
      "learning_rate": 3.7589028721987927e-06,
      "loss": 0.1551,
      "step": 6480
    },
    {
      "epoch": 0.59,
      "grad_norm": 27.782412040346923,
      "learning_rate": 3.757468503170153e-06,
      "loss": 0.8149,
      "step": 6481
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3139075930200979,
      "learning_rate": 3.7560342431198377e-06,
      "loss": 0.1876,
      "step": 6482
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4879521983748467,
      "learning_rate": 3.7546000921736413e-06,
      "loss": 0.2391,
      "step": 6483
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2852235940595735,
      "learning_rate": 3.7531660504573485e-06,
      "loss": 0.2017,
      "step": 6484
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4311637439886997,
      "learning_rate": 3.7517321180967335e-06,
      "loss": 0.1895,
      "step": 6485
    },
    {
      "epoch": 0.59,
      "grad_norm": 6.622226785010074,
      "learning_rate": 3.7502982952175644e-06,
      "loss": 0.7702,
      "step": 6486
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.151698196558661,
      "learning_rate": 3.748864581945593e-06,
      "loss": 0.612,
      "step": 6487
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.444885471707848,
      "learning_rate": 3.7474309784065676e-06,
      "loss": 0.1973,
      "step": 6488
    },
    {
      "epoch": 0.59,
      "grad_norm": 7.0200022662500885,
      "learning_rate": 3.7459974847262253e-06,
      "loss": 0.6186,
      "step": 6489
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0611982692010942,
      "learning_rate": 3.7445641010302922e-06,
      "loss": 0.1317,
      "step": 6490
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3883543417021578,
      "learning_rate": 3.743130827444487e-06,
      "loss": 0.1893,
      "step": 6491
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5654394659628272,
      "learning_rate": 3.7416976640945146e-06,
      "loss": 0.1897,
      "step": 6492
    },
    {
      "epoch": 0.59,
      "grad_norm": 3.4586371074936166,
      "learning_rate": 3.7402646111060757e-06,
      "loss": 0.3544,
      "step": 6493
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3729676699004936,
      "learning_rate": 3.738831668604858e-06,
      "loss": 0.1816,
      "step": 6494
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3010941539481695,
      "learning_rate": 3.7373988367165408e-06,
      "loss": 0.2089,
      "step": 6495
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3055595818501713,
      "learning_rate": 3.7359661155667916e-06,
      "loss": 0.1849,
      "step": 6496
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3307396280711767,
      "learning_rate": 3.734533505281269e-06,
      "loss": 0.1832,
      "step": 6497
    },
    {
      "epoch": 0.59,
      "grad_norm": 16.399507150045327,
      "learning_rate": 3.7331010059856245e-06,
      "loss": 0.5852,
      "step": 6498
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.373814408156403,
      "learning_rate": 3.7316686178054963e-06,
      "loss": 0.168,
      "step": 6499
    },
    {
      "epoch": 0.59,
      "grad_norm": 5.049246229545484,
      "learning_rate": 3.7302363408665155e-06,
      "loss": 0.5749,
      "step": 6500
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3684011051894427,
      "learning_rate": 3.7288041752943022e-06,
      "loss": 0.213,
      "step": 6501
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.476316987459021,
      "learning_rate": 3.7273721212144663e-06,
      "loss": 0.2139,
      "step": 6502
    },
    {
      "epoch": 0.59,
      "grad_norm": 13.878691028265319,
      "learning_rate": 3.7259401787526104e-06,
      "loss": 0.5244,
      "step": 6503
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3444088761623332,
      "learning_rate": 3.7245083480343225e-06,
      "loss": 0.209,
      "step": 6504
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4685633275433796,
      "learning_rate": 3.723076629185186e-06,
      "loss": 0.2103,
      "step": 6505
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4143022399487661,
      "learning_rate": 3.721645022330771e-06,
      "loss": 0.2249,
      "step": 6506
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1965631692803516,
      "learning_rate": 3.720213527596639e-06,
      "loss": 0.1648,
      "step": 6507
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2889032580412378,
      "learning_rate": 3.718782145108342e-06,
      "loss": 0.2158,
      "step": 6508
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.216813286452686,
      "learning_rate": 3.717350874991423e-06,
      "loss": 0.1884,
      "step": 6509
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3424907212898058,
      "learning_rate": 3.715919717371412e-06,
      "loss": 0.1739,
      "step": 6510
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1715025431616595,
      "learning_rate": 3.7144886723738322e-06,
      "loss": 0.1779,
      "step": 6511
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1372495601231203,
      "learning_rate": 3.7130577401241963e-06,
      "loss": 0.1527,
      "step": 6512
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4074580046472283,
      "learning_rate": 3.7116269207480055e-06,
      "loss": 0.2033,
      "step": 6513
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.336000474683285,
      "learning_rate": 3.7101962143707528e-06,
      "loss": 0.155,
      "step": 6514
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4979604569583829,
      "learning_rate": 3.7087656211179198e-06,
      "loss": 0.1918,
      "step": 6515
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2024470379224332,
      "learning_rate": 3.70733514111498e-06,
      "loss": 0.1715,
      "step": 6516
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2131869197169711,
      "learning_rate": 3.705904774487396e-06,
      "loss": 0.1784,
      "step": 6517
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3680870597052814,
      "learning_rate": 3.7044745213606205e-06,
      "loss": 0.1827,
      "step": 6518
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3107721289988368,
      "learning_rate": 3.7030443818600962e-06,
      "loss": 0.1641,
      "step": 6519
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4403523670992093,
      "learning_rate": 3.701614356111256e-06,
      "loss": 0.1956,
      "step": 6520
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.2484605525427535,
      "learning_rate": 3.700184444239524e-06,
      "loss": 0.538,
      "step": 6521
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.482885600536359,
      "learning_rate": 3.6987546463703106e-06,
      "loss": 0.2573,
      "step": 6522
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.480117662968099,
      "learning_rate": 3.6973249626290197e-06,
      "loss": 0.1829,
      "step": 6523
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3382783305793509,
      "learning_rate": 3.6958953931410434e-06,
      "loss": 0.2039,
      "step": 6524
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2893361993874803,
      "learning_rate": 3.694465938031766e-06,
      "loss": 0.1957,
      "step": 6525
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3163950522315775,
      "learning_rate": 3.6930365974265596e-06,
      "loss": 0.2035,
      "step": 6526
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.137924294910774,
      "learning_rate": 3.691607371450787e-06,
      "loss": 0.1663,
      "step": 6527
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3308642101272252,
      "learning_rate": 3.6901782602298e-06,
      "loss": 0.1316,
      "step": 6528
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4308964238387454,
      "learning_rate": 3.6887492638889433e-06,
      "loss": 0.1937,
      "step": 6529
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.284403892079936,
      "learning_rate": 3.6873203825535473e-06,
      "loss": 0.6731,
      "step": 6530
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.600154286262828,
      "learning_rate": 3.6858916163489355e-06,
      "loss": 0.2287,
      "step": 6531
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4773443777809938,
      "learning_rate": 3.6844629654004194e-06,
      "loss": 0.228,
      "step": 6532
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3989489021225419,
      "learning_rate": 3.683034429833302e-06,
      "loss": 0.2041,
      "step": 6533
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3538334925673712,
      "learning_rate": 3.6816060097728758e-06,
      "loss": 0.201,
      "step": 6534
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.575394978085883,
      "learning_rate": 3.6801777053444217e-06,
      "loss": 0.1789,
      "step": 6535
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2723980268535422,
      "learning_rate": 3.6787495166732125e-06,
      "loss": 0.1816,
      "step": 6536
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2729921655924725,
      "learning_rate": 3.677321443884509e-06,
      "loss": 0.2059,
      "step": 6537
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3697593622283923,
      "learning_rate": 3.6758934871035646e-06,
      "loss": 0.1707,
      "step": 6538
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1666525835101818,
      "learning_rate": 3.6744656464556184e-06,
      "loss": 0.1519,
      "step": 6539
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4820521682128922,
      "learning_rate": 3.673037922065903e-06,
      "loss": 0.1946,
      "step": 6540
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.464578489276624,
      "learning_rate": 3.6716103140596384e-06,
      "loss": 0.2182,
      "step": 6541
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2444375098887879,
      "learning_rate": 3.670182822562035e-06,
      "loss": 0.1599,
      "step": 6542
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.682777520676065,
      "learning_rate": 3.6687554476982954e-06,
      "loss": 0.2149,
      "step": 6543
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2746931958356966,
      "learning_rate": 3.6673281895936084e-06,
      "loss": 0.1743,
      "step": 6544
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4387393901043721,
      "learning_rate": 3.6659010483731543e-06,
      "loss": 0.1627,
      "step": 6545
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3691923761068567,
      "learning_rate": 3.664474024162104e-06,
      "loss": 0.2064,
      "step": 6546
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.222219254624736,
      "learning_rate": 3.663047117085615e-06,
      "loss": 0.1635,
      "step": 6547
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3363203457598065,
      "learning_rate": 3.661620327268839e-06,
      "loss": 0.1943,
      "step": 6548
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.466839353039284,
      "learning_rate": 3.660193654836912e-06,
      "loss": 0.2,
      "step": 6549
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4701194958963593,
      "learning_rate": 3.6587670999149656e-06,
      "loss": 0.1677,
      "step": 6550
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3883679160160158,
      "learning_rate": 3.657340662628116e-06,
      "loss": 0.1803,
      "step": 6551
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2665701986718487,
      "learning_rate": 3.6559143431014733e-06,
      "loss": 0.1971,
      "step": 6552
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.433153557672973,
      "learning_rate": 3.654488141460134e-06,
      "loss": 0.1662,
      "step": 6553
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3784874782663918,
      "learning_rate": 3.653062057829186e-06,
      "loss": 0.1904,
      "step": 6554
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4701158612050496,
      "learning_rate": 3.651636092333707e-06,
      "loss": 0.2081,
      "step": 6555
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.532080480337296,
      "learning_rate": 3.650210245098762e-06,
      "loss": 0.2195,
      "step": 6556
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4602513129083219,
      "learning_rate": 3.6487845162494083e-06,
      "loss": 0.184,
      "step": 6557
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.312110087229507,
      "learning_rate": 3.6473589059106916e-06,
      "loss": 0.2054,
      "step": 6558
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3564057386951753,
      "learning_rate": 3.645933414207648e-06,
      "loss": 0.1874,
      "step": 6559
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4408931180157836,
      "learning_rate": 3.6445080412653024e-06,
      "loss": 0.1976,
      "step": 6560
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3715560524258903,
      "learning_rate": 3.6430827872086694e-06,
      "loss": 0.1653,
      "step": 6561
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.398396833420926,
      "learning_rate": 3.6416576521627535e-06,
      "loss": 0.2357,
      "step": 6562
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.181949425146332,
      "learning_rate": 3.640232636252548e-06,
      "loss": 0.1712,
      "step": 6563
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.342377346019373,
      "learning_rate": 3.6388077396030388e-06,
      "loss": 0.1886,
      "step": 6564
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4154954733720437,
      "learning_rate": 3.6373829623391943e-06,
      "loss": 0.1558,
      "step": 6565
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3079780870073434,
      "learning_rate": 3.6359583045859807e-06,
      "loss": 0.193,
      "step": 6566
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.363330336547966,
      "learning_rate": 3.634533766468349e-06,
      "loss": 0.2049,
      "step": 6567
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2729722501365017,
      "learning_rate": 3.63310934811124e-06,
      "loss": 0.1852,
      "step": 6568
    },
    {
      "epoch": 0.6,
      "grad_norm": 7.366135775494421,
      "learning_rate": 3.6316850496395863e-06,
      "loss": 0.6457,
      "step": 6569
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2596192755031212,
      "learning_rate": 3.630260871178306e-06,
      "loss": 0.1482,
      "step": 6570
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3494216271484372,
      "learning_rate": 3.6288368128523117e-06,
      "loss": 0.1779,
      "step": 6571
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1801777703186733,
      "learning_rate": 3.6274128747865024e-06,
      "loss": 0.1656,
      "step": 6572
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.292459478346994,
      "learning_rate": 3.625989057105765e-06,
      "loss": 0.1666,
      "step": 6573
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3832314146898843,
      "learning_rate": 3.6245653599349793e-06,
      "loss": 0.2499,
      "step": 6574
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3104787212169022,
      "learning_rate": 3.6231417833990124e-06,
      "loss": 0.1822,
      "step": 6575
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.465633559732013,
      "learning_rate": 3.621718327622723e-06,
      "loss": 0.2254,
      "step": 6576
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3773933486217829,
      "learning_rate": 3.6202949927309555e-06,
      "loss": 0.1639,
      "step": 6577
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.355945566280058,
      "learning_rate": 3.618871778848547e-06,
      "loss": 0.209,
      "step": 6578
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2017344660320033,
      "learning_rate": 3.6174486861003244e-06,
      "loss": 0.1752,
      "step": 6579
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.353922267298631,
      "learning_rate": 3.6160257146111e-06,
      "loss": 0.1958,
      "step": 6580
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3213516598442636,
      "learning_rate": 3.61460286450568e-06,
      "loss": 0.1753,
      "step": 6581
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.696847070271441,
      "learning_rate": 3.613180135908855e-06,
      "loss": 0.4687,
      "step": 6582
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2118477354517896,
      "learning_rate": 3.61175752894541e-06,
      "loss": 0.1595,
      "step": 6583
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3180945518637497,
      "learning_rate": 3.6103350437401176e-06,
      "loss": 0.1966,
      "step": 6584
    },
    {
      "epoch": 0.6,
      "grad_norm": 9.060268645569742,
      "learning_rate": 3.6089126804177373e-06,
      "loss": 0.5097,
      "step": 6585
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3537806843183287,
      "learning_rate": 3.607490439103021e-06,
      "loss": 0.1906,
      "step": 6586
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.291588217914926,
      "learning_rate": 3.606068319920709e-06,
      "loss": 0.454,
      "step": 6587
    },
    {
      "epoch": 0.6,
      "grad_norm": 8.881483585626288,
      "learning_rate": 3.604646322995531e-06,
      "loss": 0.7016,
      "step": 6588
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1365920036090407,
      "learning_rate": 3.603224448452205e-06,
      "loss": 0.1734,
      "step": 6589
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3512786499248204,
      "learning_rate": 3.601802696415439e-06,
      "loss": 0.1699,
      "step": 6590
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4756402400995998,
      "learning_rate": 3.6003810670099292e-06,
      "loss": 0.269,
      "step": 6591
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.597203077974484,
      "learning_rate": 3.5989595603603634e-06,
      "loss": 0.2131,
      "step": 6592
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5286345828102401,
      "learning_rate": 3.597538176591417e-06,
      "loss": 0.2408,
      "step": 6593
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3143343532523466,
      "learning_rate": 3.5961169158277543e-06,
      "loss": 0.1867,
      "step": 6594
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3362599472955305,
      "learning_rate": 3.5946957781940296e-06,
      "loss": 0.1771,
      "step": 6595
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3598430167002755,
      "learning_rate": 3.5932747638148868e-06,
      "loss": 0.1968,
      "step": 6596
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.459727845412049,
      "learning_rate": 3.5918538728149578e-06,
      "loss": 0.2107,
      "step": 6597
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3373378126056097,
      "learning_rate": 3.5904331053188644e-06,
      "loss": 0.2034,
      "step": 6598
    },
    {
      "epoch": 0.6,
      "grad_norm": 5.465539429761765,
      "learning_rate": 3.5890124614512175e-06,
      "loss": 0.426,
      "step": 6599
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2558036229967025,
      "learning_rate": 3.587591941336616e-06,
      "loss": 0.1578,
      "step": 6600
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2389310188509437,
      "learning_rate": 3.5861715450996505e-06,
      "loss": 0.1549,
      "step": 6601
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3418226827707433,
      "learning_rate": 3.584751272864899e-06,
      "loss": 0.1437,
      "step": 6602
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2802729114239204,
      "learning_rate": 3.5833311247569276e-06,
      "loss": 0.197,
      "step": 6603
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3347207549840883,
      "learning_rate": 3.581911100900294e-06,
      "loss": 0.2011,
      "step": 6604
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4083947287170608,
      "learning_rate": 3.5804912014195435e-06,
      "loss": 0.2062,
      "step": 6605
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1357527771488365,
      "learning_rate": 3.5790714264392112e-06,
      "loss": 0.142,
      "step": 6606
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.569929607359071,
      "learning_rate": 3.5776517760838216e-06,
      "loss": 0.2787,
      "step": 6607
    },
    {
      "epoch": 0.6,
      "grad_norm": 6.702281653537641,
      "learning_rate": 3.5762322504778846e-06,
      "loss": 0.5258,
      "step": 6608
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.7115298503301515,
      "learning_rate": 3.5748128497459044e-06,
      "loss": 0.2313,
      "step": 6609
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2549838510191178,
      "learning_rate": 3.5733935740123715e-06,
      "loss": 0.1801,
      "step": 6610
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4484537624729221,
      "learning_rate": 3.5719744234017654e-06,
      "loss": 0.2061,
      "step": 6611
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3280427078139017,
      "learning_rate": 3.570555398038556e-06,
      "loss": 0.1668,
      "step": 6612
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4882080368333825,
      "learning_rate": 3.5691364980472004e-06,
      "loss": 0.195,
      "step": 6613
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4258242057376642,
      "learning_rate": 3.5677177235521467e-06,
      "loss": 0.1702,
      "step": 6614
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3711623295602566,
      "learning_rate": 3.5662990746778304e-06,
      "loss": 0.1788,
      "step": 6615
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3520597831647285,
      "learning_rate": 3.564880551548676e-06,
      "loss": 0.2009,
      "step": 6616
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2256201385803558,
      "learning_rate": 3.563462154289098e-06,
      "loss": 0.1745,
      "step": 6617
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2678909869245194,
      "learning_rate": 3.5620438830234984e-06,
      "loss": 0.1538,
      "step": 6618
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4669072036552604,
      "learning_rate": 3.56062573787627e-06,
      "loss": 0.1605,
      "step": 6619
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4585039049410164,
      "learning_rate": 3.5592077189717937e-06,
      "loss": 0.2274,
      "step": 6620
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3789373367563569,
      "learning_rate": 3.557789826434439e-06,
      "loss": 0.2004,
      "step": 6621
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.310578808219339,
      "learning_rate": 3.5563720603885643e-06,
      "loss": 0.1717,
      "step": 6622
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.906923523912089,
      "learning_rate": 3.554954420958518e-06,
      "loss": 0.627,
      "step": 6623
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1675906151412265,
      "learning_rate": 3.5535369082686365e-06,
      "loss": 0.1338,
      "step": 6624
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.436701064629977,
      "learning_rate": 3.5521195224432436e-06,
      "loss": 0.1772,
      "step": 6625
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3663096178383358,
      "learning_rate": 3.5507022636066547e-06,
      "loss": 0.1867,
      "step": 6626
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.456241410427993,
      "learning_rate": 3.5492851318831733e-06,
      "loss": 0.2244,
      "step": 6627
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.407974205123852,
      "learning_rate": 3.54786812739709e-06,
      "loss": 0.2119,
      "step": 6628
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.723420796140971,
      "learning_rate": 3.546451250272687e-06,
      "loss": 0.6727,
      "step": 6629
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2862494033284961,
      "learning_rate": 3.5450345006342326e-06,
      "loss": 0.1776,
      "step": 6630
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.662718225182757,
      "learning_rate": 3.543617878605986e-06,
      "loss": 0.5666,
      "step": 6631
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3164605766220114,
      "learning_rate": 3.542201384312195e-06,
      "loss": 0.1957,
      "step": 6632
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3704745857917378,
      "learning_rate": 3.5407850178770944e-06,
      "loss": 0.2025,
      "step": 6633
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.349703168851246,
      "learning_rate": 3.5393687794249093e-06,
      "loss": 0.1949,
      "step": 6634
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.517458744401848,
      "learning_rate": 3.5379526690798536e-06,
      "loss": 0.6286,
      "step": 6635
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4086223298961864,
      "learning_rate": 3.53653668696613e-06,
      "loss": 0.1797,
      "step": 6636
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3037569100889967,
      "learning_rate": 3.535120833207928e-06,
      "loss": 0.1769,
      "step": 6637
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5223109651000184,
      "learning_rate": 3.5337051079294295e-06,
      "loss": 0.2094,
      "step": 6638
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2795967280783853,
      "learning_rate": 3.532289511254802e-06,
      "loss": 0.1673,
      "step": 6639
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5147185629199225,
      "learning_rate": 3.5308740433082034e-06,
      "loss": 0.2182,
      "step": 6640
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3840041779842553,
      "learning_rate": 3.5294587042137796e-06,
      "loss": 0.173,
      "step": 6641
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4390943471052078,
      "learning_rate": 3.5280434940956643e-06,
      "loss": 0.1789,
      "step": 6642
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4276269516091933,
      "learning_rate": 3.5266284130779815e-06,
      "loss": 0.1957,
      "step": 6643
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3470430552124493,
      "learning_rate": 3.525213461284844e-06,
      "loss": 0.1648,
      "step": 6644
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.351593665270956,
      "learning_rate": 3.523798638840351e-06,
      "loss": 0.1966,
      "step": 6645
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2702571316046798,
      "learning_rate": 3.5223839458685934e-06,
      "loss": 0.1544,
      "step": 6646
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2485188685726034,
      "learning_rate": 3.5209693824936486e-06,
      "loss": 0.151,
      "step": 6647
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.331774181097016,
      "learning_rate": 3.519554948839583e-06,
      "loss": 0.2039,
      "step": 6648
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5315331332384776,
      "learning_rate": 3.5181406450304536e-06,
      "loss": 0.2017,
      "step": 6649
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.7063016194218232,
      "learning_rate": 3.5167264711903013e-06,
      "loss": 0.2225,
      "step": 6650
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3728707794475148,
      "learning_rate": 3.5153124274431603e-06,
      "loss": 0.1962,
      "step": 6651
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6146001698194519,
      "learning_rate": 3.5138985139130523e-06,
      "loss": 0.1972,
      "step": 6652
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3859541716107928,
      "learning_rate": 3.5124847307239863e-06,
      "loss": 0.1804,
      "step": 6653
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4636765604054458,
      "learning_rate": 3.51107107799996e-06,
      "loss": 0.1949,
      "step": 6654
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3869600995113216,
      "learning_rate": 3.5096575558649603e-06,
      "loss": 0.2035,
      "step": 6655
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2290626304694543,
      "learning_rate": 3.5082441644429638e-06,
      "loss": 0.1948,
      "step": 6656
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4815911806775348,
      "learning_rate": 3.506830903857933e-06,
      "loss": 0.2243,
      "step": 6657
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.307952693204746,
      "learning_rate": 3.5054177742338224e-06,
      "loss": 0.1522,
      "step": 6658
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3781188408171716,
      "learning_rate": 3.5040047756945707e-06,
      "loss": 0.1907,
      "step": 6659
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3176062991308408,
      "learning_rate": 3.502591908364107e-06,
      "loss": 0.1764,
      "step": 6660
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5258242169082745,
      "learning_rate": 3.501179172366351e-06,
      "loss": 0.2434,
      "step": 6661
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4058065880278847,
      "learning_rate": 3.4997665678252078e-06,
      "loss": 0.2215,
      "step": 6662
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3585141258548974,
      "learning_rate": 3.4983540948645733e-06,
      "loss": 0.1823,
      "step": 6663
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.2097632216689425,
      "learning_rate": 3.4969417536083294e-06,
      "loss": 0.53,
      "step": 6664
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.546730747252059,
      "learning_rate": 3.49552954418035e-06,
      "loss": 0.5715,
      "step": 6665
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.536179525366414,
      "learning_rate": 3.4941174667044936e-06,
      "loss": 0.2395,
      "step": 6666
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2955602635697596,
      "learning_rate": 3.49270552130461e-06,
      "loss": 0.1847,
      "step": 6667
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6208550062329743,
      "learning_rate": 3.491293708104535e-06,
      "loss": 0.2154,
      "step": 6668
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2365652079253975,
      "learning_rate": 3.489882027228095e-06,
      "loss": 0.162,
      "step": 6669
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.682516371801576,
      "learning_rate": 3.488470478799103e-06,
      "loss": 0.5375,
      "step": 6670
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3841551252842927,
      "learning_rate": 3.487059062941362e-06,
      "loss": 0.1969,
      "step": 6671
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2853917572559734,
      "learning_rate": 3.4856477797786625e-06,
      "loss": 0.1888,
      "step": 6672
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3159942618353924,
      "learning_rate": 3.484236629434783e-06,
      "loss": 0.1985,
      "step": 6673
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3296091907946088,
      "learning_rate": 3.4828256120334913e-06,
      "loss": 0.1811,
      "step": 6674
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3458662343383003,
      "learning_rate": 3.4814147276985442e-06,
      "loss": 0.1501,
      "step": 6675
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2056456723224906,
      "learning_rate": 3.480003976553683e-06,
      "loss": 0.1468,
      "step": 6676
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1958863746083768,
      "learning_rate": 3.478593358722642e-06,
      "loss": 0.173,
      "step": 6677
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4278739771510671,
      "learning_rate": 3.477182874329141e-06,
      "loss": 0.2664,
      "step": 6678
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2098159287242711,
      "learning_rate": 3.4757725234968892e-06,
      "loss": 0.175,
      "step": 6679
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3409847908478603,
      "learning_rate": 3.474362306349584e-06,
      "loss": 0.1944,
      "step": 6680
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2386846827674407,
      "learning_rate": 3.4729522230109103e-06,
      "loss": 0.1638,
      "step": 6681
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3636715950260645,
      "learning_rate": 3.4715422736045423e-06,
      "loss": 0.1682,
      "step": 6682
    },
    {
      "epoch": 0.61,
      "grad_norm": 8.886609695628675,
      "learning_rate": 3.4701324582541423e-06,
      "loss": 0.7007,
      "step": 6683
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3648321223569648,
      "learning_rate": 3.468722777083361e-06,
      "loss": 0.1829,
      "step": 6684
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.561669157854957,
      "learning_rate": 3.467313230215835e-06,
      "loss": 0.1774,
      "step": 6685
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4119758605005388,
      "learning_rate": 3.4659038177751918e-06,
      "loss": 0.1924,
      "step": 6686
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5315383361210568,
      "learning_rate": 3.464494539885047e-06,
      "loss": 0.2301,
      "step": 6687
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.211578040908468,
      "learning_rate": 3.4630853966690026e-06,
      "loss": 0.1782,
      "step": 6688
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1764945138124683,
      "learning_rate": 3.461676388250651e-06,
      "loss": 0.1439,
      "step": 6689
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5278170466223833,
      "learning_rate": 3.4602675147535707e-06,
      "loss": 0.2057,
      "step": 6690
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.34753887659162,
      "learning_rate": 3.4588587763013303e-06,
      "loss": 0.4774,
      "step": 6691
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4562076211888326,
      "learning_rate": 3.457450173017486e-06,
      "loss": 0.1751,
      "step": 6692
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4239436969433332,
      "learning_rate": 3.456041705025579e-06,
      "loss": 0.2177,
      "step": 6693
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3071804856666043,
      "learning_rate": 3.454633372449144e-06,
      "loss": 0.1725,
      "step": 6694
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3250115666676534,
      "learning_rate": 3.4532251754117e-06,
      "loss": 0.1846,
      "step": 6695
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4419409601920579,
      "learning_rate": 3.4518171140367563e-06,
      "loss": 0.1842,
      "step": 6696
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.153362937681891,
      "learning_rate": 3.4504091884478076e-06,
      "loss": 0.1573,
      "step": 6697
    },
    {
      "epoch": 0.61,
      "grad_norm": 5.261118529083179,
      "learning_rate": 3.4490013987683403e-06,
      "loss": 0.4914,
      "step": 6698
    },
    {
      "epoch": 0.61,
      "grad_norm": 6.743502410453799,
      "learning_rate": 3.4475937451218257e-06,
      "loss": 0.576,
      "step": 6699
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4613535213837299,
      "learning_rate": 3.4461862276317246e-06,
      "loss": 0.1921,
      "step": 6700
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.582377117016841,
      "learning_rate": 3.444778846421488e-06,
      "loss": 0.2045,
      "step": 6701
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4418836063958964,
      "learning_rate": 3.4433716016145492e-06,
      "loss": 0.2071,
      "step": 6702
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3817869623461372,
      "learning_rate": 3.441964493334334e-06,
      "loss": 0.2026,
      "step": 6703
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2964343347249816,
      "learning_rate": 3.440557521704256e-06,
      "loss": 0.1536,
      "step": 6704
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3307466402488275,
      "learning_rate": 3.4391506868477153e-06,
      "loss": 0.2271,
      "step": 6705
    },
    {
      "epoch": 0.61,
      "grad_norm": 7.549841357364181,
      "learning_rate": 3.4377439888881014e-06,
      "loss": 0.5157,
      "step": 6706
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2630397688024195,
      "learning_rate": 3.4363374279487904e-06,
      "loss": 0.1829,
      "step": 6707
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.428506948731569,
      "learning_rate": 3.434931004153148e-06,
      "loss": 0.195,
      "step": 6708
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2982566402001932,
      "learning_rate": 3.4335247176245263e-06,
      "loss": 0.1498,
      "step": 6709
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3076417585160642,
      "learning_rate": 3.432118568486267e-06,
      "loss": 0.1906,
      "step": 6710
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1951368061194625,
      "learning_rate": 3.430712556861697e-06,
      "loss": 0.1586,
      "step": 6711
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4341019287551138,
      "learning_rate": 3.429306682874134e-06,
      "loss": 0.2201,
      "step": 6712
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5029302594373488,
      "learning_rate": 3.4279009466468825e-06,
      "loss": 0.2138,
      "step": 6713
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4746023820936855,
      "learning_rate": 3.4264953483032346e-06,
      "loss": 0.1614,
      "step": 6714
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1195278610365103,
      "learning_rate": 3.4250898879664717e-06,
      "loss": 0.1701,
      "step": 6715
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2517518054608605,
      "learning_rate": 3.423684565759861e-06,
      "loss": 0.1245,
      "step": 6716
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2591198462131592,
      "learning_rate": 3.422279381806659e-06,
      "loss": 0.1779,
      "step": 6717
    },
    {
      "epoch": 0.61,
      "grad_norm": 11.77664454245237,
      "learning_rate": 3.420874336230111e-06,
      "loss": 0.6129,
      "step": 6718
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3348211063901716,
      "learning_rate": 3.419469429153447e-06,
      "loss": 0.2009,
      "step": 6719
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4296116075939502,
      "learning_rate": 3.418064660699887e-06,
      "loss": 0.1867,
      "step": 6720
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1906175327981598,
      "learning_rate": 3.416660030992639e-06,
      "loss": 0.2015,
      "step": 6721
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.382318061452498,
      "learning_rate": 3.415255540154898e-06,
      "loss": 0.1684,
      "step": 6722
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1396266850936358,
      "learning_rate": 3.413851188309848e-06,
      "loss": 0.1642,
      "step": 6723
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4011331455048084,
      "learning_rate": 3.41244697558066e-06,
      "loss": 0.1844,
      "step": 6724
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.358440404404315,
      "learning_rate": 3.4110429020904924e-06,
      "loss": 0.1959,
      "step": 6725
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5278373699610333,
      "learning_rate": 3.409638967962492e-06,
      "loss": 0.2209,
      "step": 6726
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.257756642520655,
      "learning_rate": 3.408235173319795e-06,
      "loss": 0.1891,
      "step": 6727
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.628011871448975,
      "learning_rate": 3.4068315182855195e-06,
      "loss": 0.6418,
      "step": 6728
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4485826512591715,
      "learning_rate": 3.405428002982779e-06,
      "loss": 0.1954,
      "step": 6729
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4691200830464917,
      "learning_rate": 3.404024627534669e-06,
      "loss": 0.2051,
      "step": 6730
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2415729015008397,
      "learning_rate": 3.4026213920642758e-06,
      "loss": 0.181,
      "step": 6731
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1953557801166976,
      "learning_rate": 3.4012182966946733e-06,
      "loss": 0.1628,
      "step": 6732
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.333480445232498,
      "learning_rate": 3.399815341548921e-06,
      "loss": 0.1713,
      "step": 6733
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4754470360874925,
      "learning_rate": 3.398412526750069e-06,
      "loss": 0.2132,
      "step": 6734
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3940806006345088,
      "learning_rate": 3.3970098524211524e-06,
      "loss": 0.1861,
      "step": 6735
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.298937633816926,
      "learning_rate": 3.3956073186851957e-06,
      "loss": 0.4714,
      "step": 6736
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.89653054388889,
      "learning_rate": 3.3942049256652093e-06,
      "loss": 0.5329,
      "step": 6737
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3955576803097332,
      "learning_rate": 3.3928026734841935e-06,
      "loss": 0.1889,
      "step": 6738
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.319887929765894,
      "learning_rate": 3.391400562265135e-06,
      "loss": 0.1814,
      "step": 6739
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.280458633713256,
      "learning_rate": 3.389998592131008e-06,
      "loss": 0.1923,
      "step": 6740
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3321191690424075,
      "learning_rate": 3.3885967632047756e-06,
      "loss": 0.1812,
      "step": 6741
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4912098895053318,
      "learning_rate": 3.3871950756093864e-06,
      "loss": 0.2132,
      "step": 6742
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3905189255042625,
      "learning_rate": 3.3857935294677786e-06,
      "loss": 0.1824,
      "step": 6743
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2735265143246042,
      "learning_rate": 3.384392124902878e-06,
      "loss": 0.1666,
      "step": 6744
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.306691203568938,
      "learning_rate": 3.3829908620375953e-06,
      "loss": 0.164,
      "step": 6745
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5190741440933533,
      "learning_rate": 3.3815897409948305e-06,
      "loss": 0.2562,
      "step": 6746
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2141563247210019,
      "learning_rate": 3.3801887618974725e-06,
      "loss": 0.164,
      "step": 6747
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.398527907670781,
      "learning_rate": 3.3787879248683968e-06,
      "loss": 0.1862,
      "step": 6748
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4586596951629778,
      "learning_rate": 3.3773872300304653e-06,
      "loss": 0.2328,
      "step": 6749
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3554745635589327,
      "learning_rate": 3.3759866775065285e-06,
      "loss": 0.21,
      "step": 6750
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3662062739859455,
      "learning_rate": 3.3745862674194246e-06,
      "loss": 0.2348,
      "step": 6751
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2601755795478897,
      "learning_rate": 3.3731859998919785e-06,
      "loss": 0.1561,
      "step": 6752
    },
    {
      "epoch": 0.62,
      "grad_norm": 5.903891104381597,
      "learning_rate": 3.3717858750470046e-06,
      "loss": 0.6535,
      "step": 6753
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.675799829167926,
      "learning_rate": 3.3703858930072996e-06,
      "loss": 0.5325,
      "step": 6754
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.429565736119932,
      "learning_rate": 3.3689860538956547e-06,
      "loss": 0.1766,
      "step": 6755
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.248850312818966,
      "learning_rate": 3.3675863578348434e-06,
      "loss": 0.1675,
      "step": 6756
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3223463674193237,
      "learning_rate": 3.366186804947629e-06,
      "loss": 0.165,
      "step": 6757
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3125128383463234,
      "learning_rate": 3.3647873953567623e-06,
      "loss": 0.1368,
      "step": 6758
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3487761403181615,
      "learning_rate": 3.3633881291849796e-06,
      "loss": 0.1761,
      "step": 6759
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4163038763480256,
      "learning_rate": 3.3619890065550064e-06,
      "loss": 0.1843,
      "step": 6760
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4624706989061045,
      "learning_rate": 3.3605900275895565e-06,
      "loss": 0.2018,
      "step": 6761
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4311523457392366,
      "learning_rate": 3.3591911924113274e-06,
      "loss": 0.1884,
      "step": 6762
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3860695983816982,
      "learning_rate": 3.357792501143007e-06,
      "loss": 0.2058,
      "step": 6763
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2474536147067044,
      "learning_rate": 3.356393953907271e-06,
      "loss": 0.1734,
      "step": 6764
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2994760999037138,
      "learning_rate": 3.3549955508267793e-06,
      "loss": 0.1858,
      "step": 6765
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2963514176852415,
      "learning_rate": 3.353597292024184e-06,
      "loss": 0.1985,
      "step": 6766
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4362601373652626,
      "learning_rate": 3.352199177622119e-06,
      "loss": 0.2098,
      "step": 6767
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3398112664221888,
      "learning_rate": 3.35080120774321e-06,
      "loss": 0.2058,
      "step": 6768
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5581821092644368,
      "learning_rate": 3.349403382510068e-06,
      "loss": 0.1757,
      "step": 6769
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.217904430551844,
      "learning_rate": 3.3480057020452922e-06,
      "loss": 0.6537,
      "step": 6770
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5293531342398197,
      "learning_rate": 3.3466081664714666e-06,
      "loss": 0.1962,
      "step": 6771
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.03295551223274,
      "learning_rate": 3.3452107759111665e-06,
      "loss": 0.3695,
      "step": 6772
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4862876497056707,
      "learning_rate": 3.34381353048695e-06,
      "loss": 0.179,
      "step": 6773
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3391183811824592,
      "learning_rate": 3.3424164303213673e-06,
      "loss": 0.1803,
      "step": 6774
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2217645724643114,
      "learning_rate": 3.3410194755369523e-06,
      "loss": 0.1557,
      "step": 6775
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.333529759644691,
      "learning_rate": 3.339622666256228e-06,
      "loss": 0.1656,
      "step": 6776
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3351654736033793,
      "learning_rate": 3.3382260026017027e-06,
      "loss": 0.2146,
      "step": 6777
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.508511742188706,
      "learning_rate": 3.336829484695875e-06,
      "loss": 0.2236,
      "step": 6778
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3634159585870655,
      "learning_rate": 3.335433112661227e-06,
      "loss": 0.2121,
      "step": 6779
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3985939834001846,
      "learning_rate": 3.33403688662023e-06,
      "loss": 0.214,
      "step": 6780
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4428049171636608,
      "learning_rate": 3.332640806695343e-06,
      "loss": 0.1988,
      "step": 6781
    },
    {
      "epoch": 0.62,
      "grad_norm": 6.405778256376369,
      "learning_rate": 3.331244873009012e-06,
      "loss": 0.6423,
      "step": 6782
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4760688756358977,
      "learning_rate": 3.329849085683669e-06,
      "loss": 0.1889,
      "step": 6783
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.402512489853884,
      "learning_rate": 3.328453444841734e-06,
      "loss": 0.2256,
      "step": 6784
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.227999360086007,
      "learning_rate": 3.3270579506056146e-06,
      "loss": 0.1679,
      "step": 6785
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2505762355903904,
      "learning_rate": 3.3256626030977033e-06,
      "loss": 0.1871,
      "step": 6786
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3423090450850683,
      "learning_rate": 3.324267402440385e-06,
      "loss": 0.1733,
      "step": 6787
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2945627330150267,
      "learning_rate": 3.322872348756024e-06,
      "loss": 0.1475,
      "step": 6788
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.437017811514275,
      "learning_rate": 3.3214774421669777e-06,
      "loss": 0.1996,
      "step": 6789
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5048138296269453,
      "learning_rate": 3.3200826827955882e-06,
      "loss": 0.1758,
      "step": 6790
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4831917432009865,
      "learning_rate": 3.3186880707641866e-06,
      "loss": 0.229,
      "step": 6791
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3060634338420485,
      "learning_rate": 3.3172936061950877e-06,
      "loss": 0.1539,
      "step": 6792
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3502418647611099,
      "learning_rate": 3.3158992892105975e-06,
      "loss": 0.154,
      "step": 6793
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4298702593657855,
      "learning_rate": 3.314505119933004e-06,
      "loss": 0.2129,
      "step": 6794
    },
    {
      "epoch": 0.62,
      "grad_norm": 10.854596252000245,
      "learning_rate": 3.3131110984845883e-06,
      "loss": 0.5271,
      "step": 6795
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.406372061526399,
      "learning_rate": 3.3117172249876144e-06,
      "loss": 0.2014,
      "step": 6796
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.285858345037813,
      "learning_rate": 3.310323499564334e-06,
      "loss": 0.2006,
      "step": 6797
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4147497137473108,
      "learning_rate": 3.3089299223369862e-06,
      "loss": 0.1889,
      "step": 6798
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.529446958930779,
      "learning_rate": 3.307536493427797e-06,
      "loss": 0.1911,
      "step": 6799
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.668140571237147,
      "learning_rate": 3.3061432129589794e-06,
      "loss": 0.2415,
      "step": 6800
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.271323821870088,
      "learning_rate": 3.3047500810527343e-06,
      "loss": 0.188,
      "step": 6801
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3496284274694883,
      "learning_rate": 3.3033570978312478e-06,
      "loss": 0.2019,
      "step": 6802
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5418959885998051,
      "learning_rate": 3.301964263416693e-06,
      "loss": 0.2057,
      "step": 6803
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.228171868176907,
      "learning_rate": 3.3005715779312318e-06,
      "loss": 0.1598,
      "step": 6804
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2792751784325374,
      "learning_rate": 3.2991790414970125e-06,
      "loss": 0.1849,
      "step": 6805
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2844181069218321,
      "learning_rate": 3.297786654236169e-06,
      "loss": 0.1914,
      "step": 6806
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1506469275676998,
      "learning_rate": 3.296394416270824e-06,
      "loss": 0.1598,
      "step": 6807
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3335217771281407,
      "learning_rate": 3.295002327723085e-06,
      "loss": 0.1614,
      "step": 6808
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5681231124401402,
      "learning_rate": 3.2936103887150484e-06,
      "loss": 0.2039,
      "step": 6809
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4009768320743365,
      "learning_rate": 3.292218599368795e-06,
      "loss": 0.1693,
      "step": 6810
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.30786099387134,
      "learning_rate": 3.2908269598063957e-06,
      "loss": 0.1757,
      "step": 6811
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1856848345137918,
      "learning_rate": 3.2894354701499053e-06,
      "loss": 0.1391,
      "step": 6812
    },
    {
      "epoch": 0.62,
      "grad_norm": 8.76067247163164,
      "learning_rate": 3.2880441305213686e-06,
      "loss": 0.6011,
      "step": 6813
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.28796010330946,
      "learning_rate": 3.286652941042813e-06,
      "loss": 0.2,
      "step": 6814
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.306802912649143,
      "learning_rate": 3.285261901836257e-06,
      "loss": 0.1735,
      "step": 6815
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5144547741487702,
      "learning_rate": 3.2838710130237027e-06,
      "loss": 0.206,
      "step": 6816
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2478759342929586,
      "learning_rate": 3.2824802747271424e-06,
      "loss": 0.171,
      "step": 6817
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4913707156775062,
      "learning_rate": 3.2810896870685514e-06,
      "loss": 0.1904,
      "step": 6818
    },
    {
      "epoch": 0.62,
      "grad_norm": 7.206410903202754,
      "learning_rate": 3.2796992501698934e-06,
      "loss": 0.4583,
      "step": 6819
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2896957990255031,
      "learning_rate": 3.2783089641531198e-06,
      "loss": 0.1543,
      "step": 6820
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3021086001836693,
      "learning_rate": 3.2769188291401678e-06,
      "loss": 0.1954,
      "step": 6821
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5703396687867701,
      "learning_rate": 3.275528845252961e-06,
      "loss": 0.1814,
      "step": 6822
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5249503748892501,
      "learning_rate": 3.274139012613411e-06,
      "loss": 0.1737,
      "step": 6823
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4835583693480305,
      "learning_rate": 3.2727493313434155e-06,
      "loss": 0.2095,
      "step": 6824
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1598786681808846,
      "learning_rate": 3.271359801564858e-06,
      "loss": 0.1632,
      "step": 6825
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4993991537086255,
      "learning_rate": 3.269970423399611e-06,
      "loss": 0.2059,
      "step": 6826
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2458860155064457,
      "learning_rate": 3.268581196969531e-06,
      "loss": 0.1619,
      "step": 6827
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2383781957980742,
      "learning_rate": 3.2671921223964625e-06,
      "loss": 0.1405,
      "step": 6828
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.153075554269837,
      "learning_rate": 3.2658031998022368e-06,
      "loss": 0.1492,
      "step": 6829
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5417989769825844,
      "learning_rate": 3.264414429308672e-06,
      "loss": 0.1564,
      "step": 6830
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3304812357317362,
      "learning_rate": 3.2630258110375723e-06,
      "loss": 0.1714,
      "step": 6831
    },
    {
      "epoch": 0.62,
      "grad_norm": 4.643142141730674,
      "learning_rate": 3.2616373451107285e-06,
      "loss": 0.5006,
      "step": 6832
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.516753730716037,
      "learning_rate": 3.2602490316499197e-06,
      "loss": 0.2369,
      "step": 6833
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3080441453037253,
      "learning_rate": 3.2588608707769088e-06,
      "loss": 0.1771,
      "step": 6834
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3980042075267263,
      "learning_rate": 3.257472862613448e-06,
      "loss": 0.2051,
      "step": 6835
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.381504538687691,
      "learning_rate": 3.2560850072812738e-06,
      "loss": 0.1494,
      "step": 6836
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3107746151188238,
      "learning_rate": 3.2546973049021107e-06,
      "loss": 0.1855,
      "step": 6837
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.549928080579972,
      "learning_rate": 3.2533097555976696e-06,
      "loss": 0.204,
      "step": 6838
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3645890089293955,
      "learning_rate": 3.2519223594896478e-06,
      "loss": 0.1748,
      "step": 6839
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.427259277002975,
      "learning_rate": 3.250535116699729e-06,
      "loss": 0.2065,
      "step": 6840
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.221021087105162,
      "learning_rate": 3.2491480273495847e-06,
      "loss": 0.1547,
      "step": 6841
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.151482381296683,
      "learning_rate": 3.2477610915608705e-06,
      "loss": 0.1271,
      "step": 6842
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2394089399254984,
      "learning_rate": 3.2463743094552318e-06,
      "loss": 0.178,
      "step": 6843
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5388294121728185,
      "learning_rate": 3.2449876811542967e-06,
      "loss": 0.2018,
      "step": 6844
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4911975221871618,
      "learning_rate": 3.2436012067796825e-06,
      "loss": 0.1861,
      "step": 6845
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3257799230707903,
      "learning_rate": 3.2422148864529924e-06,
      "loss": 0.1893,
      "step": 6846
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2512811704863143,
      "learning_rate": 3.2408287202958156e-06,
      "loss": 0.1731,
      "step": 6847
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4489010331367465,
      "learning_rate": 3.239442708429729e-06,
      "loss": 0.2056,
      "step": 6848
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.279627548966902,
      "learning_rate": 3.2380568509762935e-06,
      "loss": 0.1691,
      "step": 6849
    },
    {
      "epoch": 0.63,
      "grad_norm": 9.93088227043295,
      "learning_rate": 3.23667114805706e-06,
      "loss": 0.6746,
      "step": 6850
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2781498641098636,
      "learning_rate": 3.235285599793563e-06,
      "loss": 0.1781,
      "step": 6851
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1997586807795526,
      "learning_rate": 3.2339002063073253e-06,
      "loss": 0.1926,
      "step": 6852
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.487058076201707,
      "learning_rate": 3.2325149677198532e-06,
      "loss": 0.1968,
      "step": 6853
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4835188772601409,
      "learning_rate": 3.231129884152643e-06,
      "loss": 0.1968,
      "step": 6854
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2866953666447771,
      "learning_rate": 3.2297449557271743e-06,
      "loss": 0.1746,
      "step": 6855
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4503856057521838,
      "learning_rate": 3.2283601825649164e-06,
      "loss": 0.1678,
      "step": 6856
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5204727611612296,
      "learning_rate": 3.226975564787322e-06,
      "loss": 0.2032,
      "step": 6857
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2205278789847576,
      "learning_rate": 3.2255911025158314e-06,
      "loss": 0.1618,
      "step": 6858
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5246012786097731,
      "learning_rate": 3.224206795871872e-06,
      "loss": 0.1989,
      "step": 6859
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3852606415387716,
      "learning_rate": 3.2228226449768563e-06,
      "loss": 0.1981,
      "step": 6860
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7105609758700961,
      "learning_rate": 3.2214386499521837e-06,
      "loss": 0.1799,
      "step": 6861
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.491645660641873,
      "learning_rate": 3.2200548109192396e-06,
      "loss": 0.4504,
      "step": 6862
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.430598121162589,
      "learning_rate": 3.2186711279993953e-06,
      "loss": 0.1761,
      "step": 6863
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3706888160195145,
      "learning_rate": 3.21728760131401e-06,
      "loss": 0.177,
      "step": 6864
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2630992877269789,
      "learning_rate": 3.215904230984428e-06,
      "loss": 0.2005,
      "step": 6865
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.018499827495881,
      "learning_rate": 3.21452101713198e-06,
      "loss": 0.5493,
      "step": 6866
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4659989694357367,
      "learning_rate": 3.2131379598779835e-06,
      "loss": 0.1929,
      "step": 6867
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1079797386435737,
      "learning_rate": 3.2117550593437412e-06,
      "loss": 0.1509,
      "step": 6868
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4920429176270948,
      "learning_rate": 3.210372315650544e-06,
      "loss": 0.2185,
      "step": 6869
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.404120773744644,
      "learning_rate": 3.2089897289196666e-06,
      "loss": 0.1926,
      "step": 6870
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.0495270456882757,
      "learning_rate": 3.2076072992723705e-06,
      "loss": 0.1737,
      "step": 6871
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3748745745742008,
      "learning_rate": 3.2062250268299054e-06,
      "loss": 0.1475,
      "step": 6872
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.003110269273089,
      "learning_rate": 3.204842911713506e-06,
      "loss": 0.5658,
      "step": 6873
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.349827768740363,
      "learning_rate": 3.2034609540443917e-06,
      "loss": 0.1411,
      "step": 6874
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1972839273797211,
      "learning_rate": 3.20207915394377e-06,
      "loss": 0.1339,
      "step": 6875
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.25588540148455,
      "learning_rate": 3.2006975115328342e-06,
      "loss": 0.208,
      "step": 6876
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3216585328005725,
      "learning_rate": 3.199316026932764e-06,
      "loss": 0.1993,
      "step": 6877
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4972456425894365,
      "learning_rate": 3.1979347002647253e-06,
      "loss": 0.219,
      "step": 6878
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6246891783433204,
      "learning_rate": 3.1965535316498675e-06,
      "loss": 0.1992,
      "step": 6879
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.312434777980299,
      "learning_rate": 3.19517252120933e-06,
      "loss": 0.1463,
      "step": 6880
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.375199193105949,
      "learning_rate": 3.1937916690642356e-06,
      "loss": 0.2087,
      "step": 6881
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4782629748377751,
      "learning_rate": 3.192410975335695e-06,
      "loss": 0.2415,
      "step": 6882
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4989102215358936,
      "learning_rate": 3.1910304401448044e-06,
      "loss": 0.2122,
      "step": 6883
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1617935764940668,
      "learning_rate": 3.1896500636126457e-06,
      "loss": 0.1442,
      "step": 6884
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2050841980427012,
      "learning_rate": 3.188269845860288e-06,
      "loss": 0.1624,
      "step": 6885
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3845444236548865,
      "learning_rate": 3.1868897870087844e-06,
      "loss": 0.1747,
      "step": 6886
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3474373714152805,
      "learning_rate": 3.1855098871791756e-06,
      "loss": 0.2018,
      "step": 6887
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.187385065272721,
      "learning_rate": 3.1841301464924883e-06,
      "loss": 0.5597,
      "step": 6888
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.438917312693474,
      "learning_rate": 3.182750565069735e-06,
      "loss": 0.2068,
      "step": 6889
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.496479490787991,
      "learning_rate": 3.181371143031914e-06,
      "loss": 0.208,
      "step": 6890
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.529890655156429,
      "learning_rate": 3.1799918805000097e-06,
      "loss": 0.1984,
      "step": 6891
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2322162076841163,
      "learning_rate": 3.178612777594993e-06,
      "loss": 0.1262,
      "step": 6892
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.340251071874398,
      "learning_rate": 3.1772338344378194e-06,
      "loss": 0.2246,
      "step": 6893
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3423954482296923,
      "learning_rate": 3.1758550511494336e-06,
      "loss": 0.2054,
      "step": 6894
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.366526852070082,
      "learning_rate": 3.1744764278507633e-06,
      "loss": 0.1736,
      "step": 6895
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4568930849801278,
      "learning_rate": 3.1730979646627224e-06,
      "loss": 0.2118,
      "step": 6896
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2165291773235611,
      "learning_rate": 3.171719661706211e-06,
      "loss": 0.1595,
      "step": 6897
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5657543809932244,
      "learning_rate": 3.1703415191021156e-06,
      "loss": 0.2126,
      "step": 6898
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4556800538761534,
      "learning_rate": 3.1689635369713096e-06,
      "loss": 0.2479,
      "step": 6899
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3996302932315574,
      "learning_rate": 3.1675857154346502e-06,
      "loss": 0.2011,
      "step": 6900
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3156631241111636,
      "learning_rate": 3.1662080546129826e-06,
      "loss": 0.1462,
      "step": 6901
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1309902181707683,
      "learning_rate": 3.164830554627135e-06,
      "loss": 0.1296,
      "step": 6902
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4689123022349222,
      "learning_rate": 3.1634532155979257e-06,
      "loss": 0.235,
      "step": 6903
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.368892319712636,
      "learning_rate": 3.1620760376461557e-06,
      "loss": 0.1685,
      "step": 6904
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.197612215971727,
      "learning_rate": 3.1606990208926125e-06,
      "loss": 0.1471,
      "step": 6905
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3309340608171891,
      "learning_rate": 3.1593221654580684e-06,
      "loss": 0.2263,
      "step": 6906
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3198688588003156,
      "learning_rate": 3.1579454714632853e-06,
      "loss": 0.1944,
      "step": 6907
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6216153275462868,
      "learning_rate": 3.1565689390290067e-06,
      "loss": 0.2194,
      "step": 6908
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4004854099264743,
      "learning_rate": 3.1551925682759638e-06,
      "loss": 0.2195,
      "step": 6909
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4111140338599635,
      "learning_rate": 3.153816359324874e-06,
      "loss": 0.1828,
      "step": 6910
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.276430572706485,
      "learning_rate": 3.152440312296441e-06,
      "loss": 0.1715,
      "step": 6911
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2116229407496624,
      "learning_rate": 3.1510644273113524e-06,
      "loss": 0.1718,
      "step": 6912
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3309619968830775,
      "learning_rate": 3.1496887044902815e-06,
      "loss": 0.1991,
      "step": 6913
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4329574760846584,
      "learning_rate": 3.1483131439538895e-06,
      "loss": 0.1916,
      "step": 6914
    },
    {
      "epoch": 0.63,
      "grad_norm": 8.091571247179099,
      "learning_rate": 3.146937745822822e-06,
      "loss": 0.3442,
      "step": 6915
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.469053506248382,
      "learning_rate": 3.1455625102177114e-06,
      "loss": 0.2205,
      "step": 6916
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1409477820421312,
      "learning_rate": 3.1441874372591743e-06,
      "loss": 0.1822,
      "step": 6917
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6018948697429884,
      "learning_rate": 3.1428125270678135e-06,
      "loss": 0.2189,
      "step": 6918
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5387960344552054,
      "learning_rate": 3.141437779764218e-06,
      "loss": 0.233,
      "step": 6919
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3561370804022048,
      "learning_rate": 3.1400631954689626e-06,
      "loss": 0.1839,
      "step": 6920
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2876467845679698,
      "learning_rate": 3.1386887743026083e-06,
      "loss": 0.1891,
      "step": 6921
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4729209477711251,
      "learning_rate": 3.1373145163857e-06,
      "loss": 0.2072,
      "step": 6922
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2782295829442774,
      "learning_rate": 3.135940421838769e-06,
      "loss": 0.1915,
      "step": 6923
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3938793345087677,
      "learning_rate": 3.134566490782333e-06,
      "loss": 0.1717,
      "step": 6924
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.344826108117254,
      "learning_rate": 3.1331927233368954e-06,
      "loss": 0.1658,
      "step": 6925
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2404592518700375,
      "learning_rate": 3.1318191196229443e-06,
      "loss": 0.1656,
      "step": 6926
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.271234853471498,
      "learning_rate": 3.130445679760954e-06,
      "loss": 0.2132,
      "step": 6927
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.975181935539228,
      "learning_rate": 3.129072403871385e-06,
      "loss": 0.6451,
      "step": 6928
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2752685277701523,
      "learning_rate": 3.127699292074683e-06,
      "loss": 0.1277,
      "step": 6929
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5232140121385342,
      "learning_rate": 3.126326344491277e-06,
      "loss": 0.1959,
      "step": 6930
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3652315052516506,
      "learning_rate": 3.124953561241585e-06,
      "loss": 0.1845,
      "step": 6931
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4979779431991498,
      "learning_rate": 3.1235809424460096e-06,
      "loss": 0.1957,
      "step": 6932
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2096921966738206,
      "learning_rate": 3.1222084882249375e-06,
      "loss": 0.1792,
      "step": 6933
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3549532176220032,
      "learning_rate": 3.1208361986987434e-06,
      "loss": 0.1933,
      "step": 6934
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1402194381630995,
      "learning_rate": 3.1194640739877865e-06,
      "loss": 0.1258,
      "step": 6935
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4333508908917783,
      "learning_rate": 3.11809211421241e-06,
      "loss": 0.2046,
      "step": 6936
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.676102627517125,
      "learning_rate": 3.1167203194929447e-06,
      "loss": 0.6428,
      "step": 6937
    },
    {
      "epoch": 0.63,
      "grad_norm": 5.940161665188165,
      "learning_rate": 3.1153486899497073e-06,
      "loss": 0.6815,
      "step": 6938
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.282444115199352,
      "learning_rate": 3.1139772257029967e-06,
      "loss": 0.1873,
      "step": 6939
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4073792974957888,
      "learning_rate": 3.1126059268731e-06,
      "loss": 0.2272,
      "step": 6940
    },
    {
      "epoch": 0.63,
      "grad_norm": 7.136570448177949,
      "learning_rate": 3.11123479358029e-06,
      "loss": 0.5803,
      "step": 6941
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4088204232939743,
      "learning_rate": 3.1098638259448234e-06,
      "loss": 0.2094,
      "step": 6942
    },
    {
      "epoch": 0.63,
      "grad_norm": 6.314420606994232,
      "learning_rate": 3.108493024086945e-06,
      "loss": 0.5398,
      "step": 6943
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2867387687480054,
      "learning_rate": 3.1071223881268806e-06,
      "loss": 0.1895,
      "step": 6944
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.537115312910949,
      "learning_rate": 3.1057519181848474e-06,
      "loss": 0.2229,
      "step": 6945
    },
    {
      "epoch": 0.64,
      "grad_norm": 12.749837823066892,
      "learning_rate": 3.104381614381041e-06,
      "loss": 0.3803,
      "step": 6946
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.199829578646033,
      "learning_rate": 3.1030114768356513e-06,
      "loss": 0.1814,
      "step": 6947
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2490496307412593,
      "learning_rate": 3.1016415056688432e-06,
      "loss": 0.1631,
      "step": 6948
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3054076087072988,
      "learning_rate": 3.100271701000775e-06,
      "loss": 0.1904,
      "step": 6949
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.298512712496573,
      "learning_rate": 3.0989020629515875e-06,
      "loss": 0.1555,
      "step": 6950
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.755081168333445,
      "learning_rate": 3.0975325916414065e-06,
      "loss": 0.2215,
      "step": 6951
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.247007804696543,
      "learning_rate": 3.096163287190345e-06,
      "loss": 0.1956,
      "step": 6952
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4142953464271697,
      "learning_rate": 3.0947941497184985e-06,
      "loss": 0.1775,
      "step": 6953
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3622349172418773,
      "learning_rate": 3.0934251793459506e-06,
      "loss": 0.1331,
      "step": 6954
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2386089011616102,
      "learning_rate": 3.09205637619277e-06,
      "loss": 0.1669,
      "step": 6955
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2855132836632197,
      "learning_rate": 3.0906877403790086e-06,
      "loss": 0.2009,
      "step": 6956
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5011776905976595,
      "learning_rate": 3.0893192720247044e-06,
      "loss": 0.212,
      "step": 6957
    },
    {
      "epoch": 0.64,
      "grad_norm": 8.76899184315638,
      "learning_rate": 3.087950971249883e-06,
      "loss": 0.6238,
      "step": 6958
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4663852822625918,
      "learning_rate": 3.0865828381745515e-06,
      "loss": 0.1936,
      "step": 6959
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7663324263194062,
      "learning_rate": 3.0852148729187058e-06,
      "loss": 0.1828,
      "step": 6960
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.336785953348962,
      "learning_rate": 3.0838470756023253e-06,
      "loss": 0.1828,
      "step": 6961
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2638156133235237,
      "learning_rate": 3.082479446345375e-06,
      "loss": 0.153,
      "step": 6962
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8197119750205608,
      "learning_rate": 3.081111985267805e-06,
      "loss": 0.1501,
      "step": 6963
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2260657331142186,
      "learning_rate": 3.0797446924895514e-06,
      "loss": 0.1862,
      "step": 6964
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.293984954470355,
      "learning_rate": 3.078377568130534e-06,
      "loss": 0.1918,
      "step": 6965
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.277659826849924,
      "learning_rate": 3.077010612310658e-06,
      "loss": 0.1946,
      "step": 6966
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.63304322142784,
      "learning_rate": 3.0756438251498167e-06,
      "loss": 0.6161,
      "step": 6967
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3285647737776904,
      "learning_rate": 3.0742772067678845e-06,
      "loss": 0.1829,
      "step": 6968
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.750779755170254,
      "learning_rate": 3.0729107572847244e-06,
      "loss": 0.7242,
      "step": 6969
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.426227210355278,
      "learning_rate": 3.0715444768201826e-06,
      "loss": 0.1954,
      "step": 6970
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3542208590547948,
      "learning_rate": 3.070178365494091e-06,
      "loss": 0.1907,
      "step": 6971
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.824401662056625,
      "learning_rate": 3.0688124234262677e-06,
      "loss": 0.5492,
      "step": 6972
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2301120902316305,
      "learning_rate": 3.067446650736513e-06,
      "loss": 0.177,
      "step": 6973
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.409388524148663,
      "learning_rate": 3.0660810475446156e-06,
      "loss": 0.1841,
      "step": 6974
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4275640492555115,
      "learning_rate": 3.0647156139703477e-06,
      "loss": 0.1747,
      "step": 6975
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3953442215720049,
      "learning_rate": 3.0633503501334673e-06,
      "loss": 0.1847,
      "step": 6976
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2897650700001841,
      "learning_rate": 3.0619852561537165e-06,
      "loss": 0.1866,
      "step": 6977
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.507107874456671,
      "learning_rate": 3.060620332150823e-06,
      "loss": 0.1814,
      "step": 6978
    },
    {
      "epoch": 0.64,
      "grad_norm": 6.032668433232076,
      "learning_rate": 3.0592555782445017e-06,
      "loss": 0.4693,
      "step": 6979
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2188937783956775,
      "learning_rate": 3.057890994554449e-06,
      "loss": 0.1528,
      "step": 6980
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4214869874802272,
      "learning_rate": 3.0565265812003488e-06,
      "loss": 0.1756,
      "step": 6981
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1848531859624445,
      "learning_rate": 3.055162338301868e-06,
      "loss": 0.1379,
      "step": 6982
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.356555168188353,
      "learning_rate": 3.0537982659786604e-06,
      "loss": 0.1721,
      "step": 6983
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3299314538809595,
      "learning_rate": 3.0524343643503645e-06,
      "loss": 0.1939,
      "step": 6984
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4337206900376298,
      "learning_rate": 3.0510706335366034e-06,
      "loss": 0.2242,
      "step": 6985
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3733046594378029,
      "learning_rate": 3.0497070736569866e-06,
      "loss": 0.1657,
      "step": 6986
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3831334032459157,
      "learning_rate": 3.0483436848311055e-06,
      "loss": 0.1683,
      "step": 6987
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3572169356807506,
      "learning_rate": 3.0469804671785395e-06,
      "loss": 0.1477,
      "step": 6988
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2900387264831978,
      "learning_rate": 3.045617420818852e-06,
      "loss": 0.1652,
      "step": 6989
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4128823326545676,
      "learning_rate": 3.0442545458715922e-06,
      "loss": 0.2201,
      "step": 6990
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1788644499114924,
      "learning_rate": 3.0428918424562907e-06,
      "loss": 0.1174,
      "step": 6991
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5088349210862508,
      "learning_rate": 3.0415293106924675e-06,
      "loss": 0.1721,
      "step": 6992
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3191443733268509,
      "learning_rate": 3.040166950699626e-06,
      "loss": 0.1735,
      "step": 6993
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4866467051055483,
      "learning_rate": 3.038804762597253e-06,
      "loss": 0.205,
      "step": 6994
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.246773555196905,
      "learning_rate": 3.0374427465048217e-06,
      "loss": 0.1972,
      "step": 6995
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.505973863413649,
      "learning_rate": 3.0360809025417915e-06,
      "loss": 0.1963,
      "step": 6996
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2710007899550797,
      "learning_rate": 3.034719230827604e-06,
      "loss": 0.1988,
      "step": 6997
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2323426649459386,
      "learning_rate": 3.0333577314816883e-06,
      "loss": 0.1427,
      "step": 6998
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.209514419676392,
      "learning_rate": 3.0319964046234544e-06,
      "loss": 0.5162,
      "step": 6999
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2678674661954161,
      "learning_rate": 3.030635250372302e-06,
      "loss": 0.1715,
      "step": 7000
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3303395901939286,
      "learning_rate": 3.0292742688476125e-06,
      "loss": 0.1684,
      "step": 7001
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.432022965548628,
      "learning_rate": 3.027913460168753e-06,
      "loss": 0.1526,
      "step": 7002
    },
    {
      "epoch": 0.64,
      "grad_norm": 9.91132262105535,
      "learning_rate": 3.0265528244550756e-06,
      "loss": 0.4117,
      "step": 7003
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3379218229544918,
      "learning_rate": 3.025192361825918e-06,
      "loss": 0.2032,
      "step": 7004
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.418803539458398,
      "learning_rate": 3.023832072400601e-06,
      "loss": 0.2099,
      "step": 7005
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3290617028311664,
      "learning_rate": 3.0224719562984315e-06,
      "loss": 0.148,
      "step": 7006
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3470284500801917,
      "learning_rate": 3.021112013638702e-06,
      "loss": 0.1561,
      "step": 7007
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4680510593954346,
      "learning_rate": 3.0197522445406855e-06,
      "loss": 0.1886,
      "step": 7008
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3137632451619818,
      "learning_rate": 3.018392649123645e-06,
      "loss": 0.1601,
      "step": 7009
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2465662831633706,
      "learning_rate": 3.0170332275068247e-06,
      "loss": 0.1602,
      "step": 7010
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3384812567712998,
      "learning_rate": 3.015673979809457e-06,
      "loss": 0.1548,
      "step": 7011
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.379636965990238,
      "learning_rate": 3.0143149061507555e-06,
      "loss": 0.2057,
      "step": 7012
    },
    {
      "epoch": 0.64,
      "grad_norm": 13.938893664709212,
      "learning_rate": 3.0129560066499204e-06,
      "loss": 0.6084,
      "step": 7013
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4631420297695814,
      "learning_rate": 3.0115972814261362e-06,
      "loss": 0.1983,
      "step": 7014
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3949457597954464,
      "learning_rate": 3.0102387305985737e-06,
      "loss": 0.2012,
      "step": 7015
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.390650944041313,
      "learning_rate": 3.008880354286384e-06,
      "loss": 0.1606,
      "step": 7016
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4191223678172498,
      "learning_rate": 3.0075221526087083e-06,
      "loss": 0.1496,
      "step": 7017
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2538289825820528,
      "learning_rate": 3.0061641256846685e-06,
      "loss": 0.1502,
      "step": 7018
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3101840950205386,
      "learning_rate": 3.0048062736333733e-06,
      "loss": 0.162,
      "step": 7019
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3173482783431143,
      "learning_rate": 3.0034485965739152e-06,
      "loss": 0.1905,
      "step": 7020
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4923820372057865,
      "learning_rate": 3.0020910946253713e-06,
      "loss": 0.1753,
      "step": 7021
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2107650548949136,
      "learning_rate": 3.0007337679068044e-06,
      "loss": 0.1846,
      "step": 7022
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4567510245558537,
      "learning_rate": 2.9993766165372606e-06,
      "loss": 0.1604,
      "step": 7023
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5123942852930623,
      "learning_rate": 2.998019640635772e-06,
      "loss": 0.2014,
      "step": 7024
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4032857444138884,
      "learning_rate": 2.9966628403213528e-06,
      "loss": 0.1865,
      "step": 7025
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2282525286960295,
      "learning_rate": 2.995306215713004e-06,
      "loss": 0.1749,
      "step": 7026
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1773896903793266,
      "learning_rate": 2.993949766929711e-06,
      "loss": 0.1414,
      "step": 7027
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3919328065725496,
      "learning_rate": 2.992593494090444e-06,
      "loss": 0.1816,
      "step": 7028
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4643699251313376,
      "learning_rate": 2.9912373973141563e-06,
      "loss": 0.2168,
      "step": 7029
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.351163179200742,
      "learning_rate": 2.989881476719787e-06,
      "loss": 0.1903,
      "step": 7030
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1512844550558865,
      "learning_rate": 2.9885257324262596e-06,
      "loss": 0.1485,
      "step": 7031
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4609175582549412,
      "learning_rate": 2.9871701645524816e-06,
      "loss": 0.1776,
      "step": 7032
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7180767359602183,
      "learning_rate": 2.985814773217346e-06,
      "loss": 0.2016,
      "step": 7033
    },
    {
      "epoch": 0.64,
      "grad_norm": 5.958759016755205,
      "learning_rate": 2.984459558539729e-06,
      "loss": 0.4985,
      "step": 7034
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.400335155233277,
      "learning_rate": 2.983104520638492e-06,
      "loss": 0.2054,
      "step": 7035
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3524055210652317,
      "learning_rate": 2.9817496596324808e-06,
      "loss": 0.2352,
      "step": 7036
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3777954665168146,
      "learning_rate": 2.980394975640526e-06,
      "loss": 0.142,
      "step": 7037
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.493584552904603,
      "learning_rate": 2.979040468781443e-06,
      "loss": 0.2151,
      "step": 7038
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.47089025738635,
      "learning_rate": 2.97768613917403e-06,
      "loss": 0.2073,
      "step": 7039
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4184651039843335,
      "learning_rate": 2.976331986937072e-06,
      "loss": 0.1692,
      "step": 7040
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4065322943583094,
      "learning_rate": 2.9749780121893366e-06,
      "loss": 0.2079,
      "step": 7041
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4397560524440045,
      "learning_rate": 2.9736242150495765e-06,
      "loss": 0.2028,
      "step": 7042
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2845649257513438,
      "learning_rate": 2.9722705956365284e-06,
      "loss": 0.1528,
      "step": 7043
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6042372513259286,
      "learning_rate": 2.970917154068914e-06,
      "loss": 0.2244,
      "step": 7044
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3275570837713118,
      "learning_rate": 2.9695638904654395e-06,
      "loss": 0.1651,
      "step": 7045
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3638895378180873,
      "learning_rate": 2.9682108049447946e-06,
      "loss": 0.2007,
      "step": 7046
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4609296071640325,
      "learning_rate": 2.9668578976256545e-06,
      "loss": 0.1877,
      "step": 7047
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3711916023198718,
      "learning_rate": 2.9655051686266777e-06,
      "loss": 0.2079,
      "step": 7048
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3497504391538737,
      "learning_rate": 2.964152618066508e-06,
      "loss": 0.1806,
      "step": 7049
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1092679662776816,
      "learning_rate": 2.962800246063774e-06,
      "loss": 0.1464,
      "step": 7050
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2999886370454328,
      "learning_rate": 2.961448052737086e-06,
      "loss": 0.1994,
      "step": 7051
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.28510831265969,
      "learning_rate": 2.9600960382050413e-06,
      "loss": 0.2018,
      "step": 7052
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2769511998806076,
      "learning_rate": 2.9587442025862203e-06,
      "loss": 0.161,
      "step": 7053
    },
    {
      "epoch": 0.64,
      "grad_norm": 7.439822955872501,
      "learning_rate": 2.9573925459991882e-06,
      "loss": 0.6346,
      "step": 7054
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2756429429834546,
      "learning_rate": 2.956041068562494e-06,
      "loss": 0.1597,
      "step": 7055
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.381956749300612,
      "learning_rate": 2.9546897703946727e-06,
      "loss": 0.1683,
      "step": 7056
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.440303354355733,
      "learning_rate": 2.9533386516142402e-06,
      "loss": 0.1747,
      "step": 7057
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4565931978594344,
      "learning_rate": 2.951987712339701e-06,
      "loss": 0.1956,
      "step": 7058
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1366991523333299,
      "learning_rate": 2.9506369526895394e-06,
      "loss": 0.1522,
      "step": 7059
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.105365684777746,
      "learning_rate": 2.9492863727822264e-06,
      "loss": 0.5254,
      "step": 7060
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3099858540401774,
      "learning_rate": 2.947935972736217e-06,
      "loss": 0.1709,
      "step": 7061
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4434614749142485,
      "learning_rate": 2.9465857526699517e-06,
      "loss": 0.2054,
      "step": 7062
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5206117037205311,
      "learning_rate": 2.9452357127018516e-06,
      "loss": 0.1967,
      "step": 7063
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2459401475669452,
      "learning_rate": 2.9438858529503255e-06,
      "loss": 0.1614,
      "step": 7064
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.579364076362977,
      "learning_rate": 2.9425361735337655e-06,
      "loss": 0.5628,
      "step": 7065
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3545978340110465,
      "learning_rate": 2.9411866745705466e-06,
      "loss": 0.1547,
      "step": 7066
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4183303516179753,
      "learning_rate": 2.9398373561790303e-06,
      "loss": 0.2139,
      "step": 7067
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4374083562402782,
      "learning_rate": 2.93848821847756e-06,
      "loss": 0.2188,
      "step": 7068
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6235507275576813,
      "learning_rate": 2.9371392615844623e-06,
      "loss": 0.209,
      "step": 7069
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.269051287948674,
      "learning_rate": 2.935790485618052e-06,
      "loss": 0.6048,
      "step": 7070
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2567272258248876,
      "learning_rate": 2.9344418906966252e-06,
      "loss": 0.1653,
      "step": 7071
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.953161630207478,
      "learning_rate": 2.9330934769384633e-06,
      "loss": 0.5684,
      "step": 7072
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3079207436304385,
      "learning_rate": 2.93174524446183e-06,
      "loss": 0.173,
      "step": 7073
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.317534014153782,
      "learning_rate": 2.930397193384975e-06,
      "loss": 0.6548,
      "step": 7074
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3378214852482797,
      "learning_rate": 2.929049323826131e-06,
      "loss": 0.1977,
      "step": 7075
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3444848053791734,
      "learning_rate": 2.9277016359035165e-06,
      "loss": 0.206,
      "step": 7076
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3472282240841165,
      "learning_rate": 2.926354129735331e-06,
      "loss": 0.1869,
      "step": 7077
    },
    {
      "epoch": 0.65,
      "grad_norm": 8.226752898815388,
      "learning_rate": 2.92500680543976e-06,
      "loss": 0.4891,
      "step": 7078
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.438088135919928,
      "learning_rate": 2.923659663134974e-06,
      "loss": 0.6641,
      "step": 7079
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3065248975249655,
      "learning_rate": 2.922312702939125e-06,
      "loss": 0.1611,
      "step": 7080
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.424551768824686,
      "learning_rate": 2.920965924970352e-06,
      "loss": 0.1808,
      "step": 7081
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4508204081042517,
      "learning_rate": 2.919619329346776e-06,
      "loss": 0.218,
      "step": 7082
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2763350662469828,
      "learning_rate": 2.918272916186501e-06,
      "loss": 0.173,
      "step": 7083
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2620257782712145,
      "learning_rate": 2.9169266856076183e-06,
      "loss": 0.1984,
      "step": 7084
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4349644817369998,
      "learning_rate": 2.9155806377281994e-06,
      "loss": 0.1669,
      "step": 7085
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.382708727354416,
      "learning_rate": 2.9142347726663044e-06,
      "loss": 0.1665,
      "step": 7086
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.0519879102180845,
      "learning_rate": 2.912889090539971e-06,
      "loss": 0.4921,
      "step": 7087
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.598348924934024,
      "learning_rate": 2.9115435914672287e-06,
      "loss": 0.6836,
      "step": 7088
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3390734960400896,
      "learning_rate": 2.910198275566085e-06,
      "loss": 0.2033,
      "step": 7089
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3830527593047697,
      "learning_rate": 2.9088531429545306e-06,
      "loss": 0.1768,
      "step": 7090
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4166044320096944,
      "learning_rate": 2.9075081937505467e-06,
      "loss": 0.1867,
      "step": 7091
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3016371279957053,
      "learning_rate": 2.9061634280720907e-06,
      "loss": 0.164,
      "step": 7092
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.324011043770951,
      "learning_rate": 2.904818846037111e-06,
      "loss": 0.1686,
      "step": 7093
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3750275200382611,
      "learning_rate": 2.9034744477635337e-06,
      "loss": 0.2101,
      "step": 7094
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1795746368541826,
      "learning_rate": 2.9021302333692738e-06,
      "loss": 0.1616,
      "step": 7095
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4650866982214696,
      "learning_rate": 2.9007862029722244e-06,
      "loss": 0.1987,
      "step": 7096
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4768978806265725,
      "learning_rate": 2.899442356690271e-06,
      "loss": 0.2001,
      "step": 7097
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4260896956475555,
      "learning_rate": 2.898098694641274e-06,
      "loss": 0.1905,
      "step": 7098
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1064354721547758,
      "learning_rate": 2.896755216943081e-06,
      "loss": 0.1193,
      "step": 7099
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3869451037489988,
      "learning_rate": 2.8954119237135287e-06,
      "loss": 0.1958,
      "step": 7100
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2833885433618868,
      "learning_rate": 2.894068815070427e-06,
      "loss": 0.1923,
      "step": 7101
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3814787637505122,
      "learning_rate": 2.892725891131581e-06,
      "loss": 0.1861,
      "step": 7102
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2517113867870358,
      "learning_rate": 2.891383152014769e-06,
      "loss": 0.1646,
      "step": 7103
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5124028309062278,
      "learning_rate": 2.8900405978377627e-06,
      "loss": 0.1781,
      "step": 7104
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2914080141738409,
      "learning_rate": 2.8886982287183092e-06,
      "loss": 0.1546,
      "step": 7105
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3400150817949037,
      "learning_rate": 2.8873560447741473e-06,
      "loss": 0.1793,
      "step": 7106
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4708444836476176,
      "learning_rate": 2.886014046122993e-06,
      "loss": 0.2116,
      "step": 7107
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3056077592871733,
      "learning_rate": 2.884672232882547e-06,
      "loss": 0.1633,
      "step": 7108
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1746530116239047,
      "learning_rate": 2.8833306051704996e-06,
      "loss": 0.1326,
      "step": 7109
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5244248711912713,
      "learning_rate": 2.881989163104516e-06,
      "loss": 0.2123,
      "step": 7110
    },
    {
      "epoch": 0.65,
      "grad_norm": 15.44986120846426,
      "learning_rate": 2.880647906802253e-06,
      "loss": 0.5176,
      "step": 7111
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4426411125935625,
      "learning_rate": 2.879306836381345e-06,
      "loss": 0.2146,
      "step": 7112
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.080864696802248,
      "learning_rate": 2.8779659519594173e-06,
      "loss": 0.7303,
      "step": 7113
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.484904368604764,
      "learning_rate": 2.8766252536540705e-06,
      "loss": 0.1524,
      "step": 7114
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4691829359465407,
      "learning_rate": 2.8752847415828923e-06,
      "loss": 0.1821,
      "step": 7115
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2169017244634823,
      "learning_rate": 2.8739444158634578e-06,
      "loss": 0.1741,
      "step": 7116
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4076752444258238,
      "learning_rate": 2.872604276613319e-06,
      "loss": 0.1755,
      "step": 7117
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5188829071150158,
      "learning_rate": 2.8712643239500184e-06,
      "loss": 0.1992,
      "step": 7118
    },
    {
      "epoch": 0.65,
      "grad_norm": 12.85125629107116,
      "learning_rate": 2.8699245579910754e-06,
      "loss": 0.4809,
      "step": 7119
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4287934742528847,
      "learning_rate": 2.868584978854001e-06,
      "loss": 0.1967,
      "step": 7120
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.400209787498388,
      "learning_rate": 2.8672455866562797e-06,
      "loss": 0.4259,
      "step": 7121
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2609470019793536,
      "learning_rate": 2.865906381515391e-06,
      "loss": 0.1717,
      "step": 7122
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4524415611063526,
      "learning_rate": 2.8645673635487888e-06,
      "loss": 0.2315,
      "step": 7123
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4693638510011962,
      "learning_rate": 2.863228532873912e-06,
      "loss": 0.1967,
      "step": 7124
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2827799311516497,
      "learning_rate": 2.861889889608188e-06,
      "loss": 0.1603,
      "step": 7125
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2698806025006315,
      "learning_rate": 2.8605514338690233e-06,
      "loss": 0.1604,
      "step": 7126
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4314049025004627,
      "learning_rate": 2.859213165773813e-06,
      "loss": 0.1847,
      "step": 7127
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1322074151489843,
      "learning_rate": 2.8578750854399294e-06,
      "loss": 0.1265,
      "step": 7128
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.296604034668371,
      "learning_rate": 2.8565371929847286e-06,
      "loss": 0.1877,
      "step": 7129
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.386686567439765,
      "learning_rate": 2.8551994885255573e-06,
      "loss": 0.1911,
      "step": 7130
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5829619118204707,
      "learning_rate": 2.853861972179738e-06,
      "loss": 0.2133,
      "step": 7131
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3389212836273787,
      "learning_rate": 2.8525246440645814e-06,
      "loss": 0.1808,
      "step": 7132
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4168174950888262,
      "learning_rate": 2.8511875042973788e-06,
      "loss": 0.1708,
      "step": 7133
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3808412306120983,
      "learning_rate": 2.8498505529954084e-06,
      "loss": 0.203,
      "step": 7134
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4609483064748963,
      "learning_rate": 2.8485137902759262e-06,
      "loss": 0.2024,
      "step": 7135
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4791461663994965,
      "learning_rate": 2.8471772162561804e-06,
      "loss": 0.2213,
      "step": 7136
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1354650375897608,
      "learning_rate": 2.8458408310533948e-06,
      "loss": 0.1621,
      "step": 7137
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2105606037288072,
      "learning_rate": 2.844504634784776e-06,
      "loss": 0.146,
      "step": 7138
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2775613038483138,
      "learning_rate": 2.843168627567523e-06,
      "loss": 0.156,
      "step": 7139
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.667537588411422,
      "learning_rate": 2.841832809518808e-06,
      "loss": 0.4869,
      "step": 7140
    },
    {
      "epoch": 0.65,
      "grad_norm": 12.093453556945605,
      "learning_rate": 2.8404971807557957e-06,
      "loss": 0.3814,
      "step": 7141
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.375577831159084,
      "learning_rate": 2.839161741395624e-06,
      "loss": 0.1729,
      "step": 7142
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3016080748611867,
      "learning_rate": 2.8378264915554264e-06,
      "loss": 0.1667,
      "step": 7143
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5121634449876653,
      "learning_rate": 2.8364914313523075e-06,
      "loss": 0.2132,
      "step": 7144
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3879664647785535,
      "learning_rate": 2.835156560903365e-06,
      "loss": 0.2116,
      "step": 7145
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.421173255540491,
      "learning_rate": 2.8338218803256745e-06,
      "loss": 0.1832,
      "step": 7146
    },
    {
      "epoch": 0.65,
      "grad_norm": 7.673154047363919,
      "learning_rate": 2.8324873897362947e-06,
      "loss": 0.5196,
      "step": 7147
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3540354418576959,
      "learning_rate": 2.8311530892522727e-06,
      "loss": 0.2083,
      "step": 7148
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4093948437222517,
      "learning_rate": 2.8298189789906317e-06,
      "loss": 0.1947,
      "step": 7149
    },
    {
      "epoch": 0.65,
      "grad_norm": 4.799236996429594,
      "learning_rate": 2.8284850590683867e-06,
      "loss": 0.5754,
      "step": 7150
    },
    {
      "epoch": 0.65,
      "grad_norm": 9.682884639750416,
      "learning_rate": 2.827151329602526e-06,
      "loss": 0.4985,
      "step": 7151
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2336531949179161,
      "learning_rate": 2.8258177907100326e-06,
      "loss": 0.17,
      "step": 7152
    },
    {
      "epoch": 0.65,
      "grad_norm": 6.2781827479945616,
      "learning_rate": 2.824484442507863e-06,
      "loss": 0.5692,
      "step": 7153
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2924277545275478,
      "learning_rate": 2.8231512851129596e-06,
      "loss": 0.1599,
      "step": 7154
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3658857605609254,
      "learning_rate": 2.8218183186422534e-06,
      "loss": 0.1794,
      "step": 7155
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5813094787451756,
      "learning_rate": 2.82048554321265e-06,
      "loss": 0.2239,
      "step": 7156
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4777174570312848,
      "learning_rate": 2.8191529589410467e-06,
      "loss": 0.2068,
      "step": 7157
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3344145358796307,
      "learning_rate": 2.817820565944316e-06,
      "loss": 0.1783,
      "step": 7158
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.340969135077072,
      "learning_rate": 2.8164883643393224e-06,
      "loss": 0.2033,
      "step": 7159
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2861286706993849,
      "learning_rate": 2.815156354242904e-06,
      "loss": 0.18,
      "step": 7160
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3865514238024237,
      "learning_rate": 2.813824535771892e-06,
      "loss": 0.2075,
      "step": 7161
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.395927095691754,
      "learning_rate": 2.812492909043093e-06,
      "loss": 0.6807,
      "step": 7162
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5635844243500243,
      "learning_rate": 2.8111614741732975e-06,
      "loss": 0.2164,
      "step": 7163
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.74778823046956,
      "learning_rate": 2.8098302312792857e-06,
      "loss": 0.6239,
      "step": 7164
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4168893295504916,
      "learning_rate": 2.8084991804778124e-06,
      "loss": 0.1652,
      "step": 7165
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3688507016198603,
      "learning_rate": 2.807168321885623e-06,
      "loss": 0.1827,
      "step": 7166
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.090879335355454,
      "learning_rate": 2.80583765561944e-06,
      "loss": 0.4889,
      "step": 7167
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2403085149107362,
      "learning_rate": 2.8045071817959748e-06,
      "loss": 0.1594,
      "step": 7168
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3051080942775812,
      "learning_rate": 2.803176900531915e-06,
      "loss": 0.1796,
      "step": 7169
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.317861911877252,
      "learning_rate": 2.801846811943939e-06,
      "loss": 0.1833,
      "step": 7170
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1106656646544766,
      "learning_rate": 2.8005169161487035e-06,
      "loss": 0.1928,
      "step": 7171
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2875783820019617,
      "learning_rate": 2.7991872132628457e-06,
      "loss": 0.171,
      "step": 7172
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2803580154590306,
      "learning_rate": 2.797857703402994e-06,
      "loss": 0.168,
      "step": 7173
    },
    {
      "epoch": 0.66,
      "grad_norm": 13.884986368474145,
      "learning_rate": 2.7965283866857525e-06,
      "loss": 0.6986,
      "step": 7174
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3899736661188802,
      "learning_rate": 2.7951992632277137e-06,
      "loss": 0.2354,
      "step": 7175
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3958934788551873,
      "learning_rate": 2.7938703331454465e-06,
      "loss": 0.1626,
      "step": 7176
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4366594143486093,
      "learning_rate": 2.7925415965555126e-06,
      "loss": 0.2006,
      "step": 7177
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2475667533301638,
      "learning_rate": 2.7912130535744453e-06,
      "loss": 0.1516,
      "step": 7178
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.373282468897723,
      "learning_rate": 2.7898847043187717e-06,
      "loss": 0.1833,
      "step": 7179
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.857347184940021,
      "learning_rate": 2.7885565489049948e-06,
      "loss": 0.4815,
      "step": 7180
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3310891338130815,
      "learning_rate": 2.7872285874496002e-06,
      "loss": 0.2083,
      "step": 7181
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3946431439037743,
      "learning_rate": 2.7859008200690636e-06,
      "loss": 0.2217,
      "step": 7182
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3278776568481763,
      "learning_rate": 2.784573246879835e-06,
      "loss": 0.1812,
      "step": 7183
    },
    {
      "epoch": 0.66,
      "grad_norm": 6.130505219790346,
      "learning_rate": 2.783245867998355e-06,
      "loss": 0.5405,
      "step": 7184
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3748171379266316,
      "learning_rate": 2.78191868354104e-06,
      "loss": 0.1864,
      "step": 7185
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5704451872377214,
      "learning_rate": 2.780591693624297e-06,
      "loss": 0.1415,
      "step": 7186
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4525024221515912,
      "learning_rate": 2.7792648983645077e-06,
      "loss": 0.1929,
      "step": 7187
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4799890832467506,
      "learning_rate": 2.777938297878045e-06,
      "loss": 0.1712,
      "step": 7188
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.344151192374556,
      "learning_rate": 2.776611892281259e-06,
      "loss": 0.1497,
      "step": 7189
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.3161718407007,
      "learning_rate": 2.7752856816904817e-06,
      "loss": 0.4156,
      "step": 7190
    },
    {
      "epoch": 0.66,
      "grad_norm": 9.961374845894102,
      "learning_rate": 2.773959666222035e-06,
      "loss": 0.5835,
      "step": 7191
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3291916731628186,
      "learning_rate": 2.7726338459922146e-06,
      "loss": 0.1541,
      "step": 7192
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.619133303421193,
      "learning_rate": 2.771308221117309e-06,
      "loss": 0.2147,
      "step": 7193
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.32073525825343,
      "learning_rate": 2.76998279171358e-06,
      "loss": 0.1829,
      "step": 7194
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.209039193502856,
      "learning_rate": 2.76865755789728e-06,
      "loss": 0.165,
      "step": 7195
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4073306485623411,
      "learning_rate": 2.7673325197846397e-06,
      "loss": 0.2168,
      "step": 7196
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.430328161929756,
      "learning_rate": 2.766007677491871e-06,
      "loss": 0.4787,
      "step": 7197
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3779671578326385,
      "learning_rate": 2.7646830311351762e-06,
      "loss": 0.1589,
      "step": 7198
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4153498796197785,
      "learning_rate": 2.763358580830731e-06,
      "loss": 0.162,
      "step": 7199
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2876775330614847,
      "learning_rate": 2.7620343266947036e-06,
      "loss": 0.1636,
      "step": 7200
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4753197116746488,
      "learning_rate": 2.760710268843234e-06,
      "loss": 0.1849,
      "step": 7201
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3998410872378173,
      "learning_rate": 2.759386407392457e-06,
      "loss": 0.1671,
      "step": 7202
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4929914389124448,
      "learning_rate": 2.7580627424584795e-06,
      "loss": 0.206,
      "step": 7203
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3261111283397453,
      "learning_rate": 2.7567392741573984e-06,
      "loss": 0.1911,
      "step": 7204
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2137049002331493,
      "learning_rate": 2.7554160026052907e-06,
      "loss": 0.168,
      "step": 7205
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.179889100435968,
      "learning_rate": 2.754092927918213e-06,
      "loss": 0.1685,
      "step": 7206
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2907900597611068,
      "learning_rate": 2.7527700502122124e-06,
      "loss": 0.1583,
      "step": 7207
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4407003656613089,
      "learning_rate": 2.7514473696033096e-06,
      "loss": 0.1961,
      "step": 7208
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.348787391902651,
      "learning_rate": 2.7501248862075163e-06,
      "loss": 0.2058,
      "step": 7209
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6819867183200095,
      "learning_rate": 2.7488026001408196e-06,
      "loss": 0.2155,
      "step": 7210
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.519751192473585,
      "learning_rate": 2.747480511519198e-06,
      "loss": 0.218,
      "step": 7211
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4345291129020068,
      "learning_rate": 2.746158620458601e-06,
      "loss": 0.1795,
      "step": 7212
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4153030116862775,
      "learning_rate": 2.744836927074973e-06,
      "loss": 0.1835,
      "step": 7213
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3514030845105447,
      "learning_rate": 2.7435154314842337e-06,
      "loss": 0.2134,
      "step": 7214
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5668417701404107,
      "learning_rate": 2.7421941338022838e-06,
      "loss": 0.1732,
      "step": 7215
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3316619869692283,
      "learning_rate": 2.7408730341450147e-06,
      "loss": 0.1373,
      "step": 7216
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7497405128849817,
      "learning_rate": 2.7395521326282913e-06,
      "loss": 0.2148,
      "step": 7217
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5008513221707775,
      "learning_rate": 2.73823142936797e-06,
      "loss": 0.202,
      "step": 7218
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.337413183081187,
      "learning_rate": 2.736910924479881e-06,
      "loss": 0.214,
      "step": 7219
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.406304743739041,
      "learning_rate": 2.7355906180798453e-06,
      "loss": 0.1524,
      "step": 7220
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.565559342287884,
      "learning_rate": 2.734270510283659e-06,
      "loss": 0.5774,
      "step": 7221
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3526729749665702,
      "learning_rate": 2.7329506012071074e-06,
      "loss": 0.1889,
      "step": 7222
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1597757248425846,
      "learning_rate": 2.731630890965955e-06,
      "loss": 0.1648,
      "step": 7223
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.317062566181406,
      "learning_rate": 2.7303113796759457e-06,
      "loss": 0.1413,
      "step": 7224
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3848440809544234,
      "learning_rate": 2.7289920674528142e-06,
      "loss": 0.1964,
      "step": 7225
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.38581376809736,
      "learning_rate": 2.7276729544122692e-06,
      "loss": 0.2109,
      "step": 7226
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.31403653560159,
      "learning_rate": 2.726354040670009e-06,
      "loss": 0.1891,
      "step": 7227
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.218671079967854,
      "learning_rate": 2.7250353263417073e-06,
      "loss": 0.2169,
      "step": 7228
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2623756546136273,
      "learning_rate": 2.723716811543029e-06,
      "loss": 0.171,
      "step": 7229
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4265886047833225,
      "learning_rate": 2.7223984963896123e-06,
      "loss": 0.2115,
      "step": 7230
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.363051838742029,
      "learning_rate": 2.7210803809970853e-06,
      "loss": 0.2008,
      "step": 7231
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1965849133035291,
      "learning_rate": 2.719762465481055e-06,
      "loss": 0.1457,
      "step": 7232
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3376433286938672,
      "learning_rate": 2.718444749957109e-06,
      "loss": 0.2059,
      "step": 7233
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.271509103580493,
      "learning_rate": 2.717127234540823e-06,
      "loss": 0.1681,
      "step": 7234
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3818073190842715,
      "learning_rate": 2.7158099193477477e-06,
      "loss": 0.1773,
      "step": 7235
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3665631821986202,
      "learning_rate": 2.714492804493426e-06,
      "loss": 0.1679,
      "step": 7236
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1018398219602317,
      "learning_rate": 2.7131758900933724e-06,
      "loss": 0.1775,
      "step": 7237
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5642258450657984,
      "learning_rate": 2.7118591762630943e-06,
      "loss": 0.1998,
      "step": 7238
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.316772574442578,
      "learning_rate": 2.7105426631180726e-06,
      "loss": 0.2221,
      "step": 7239
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.182798967628208,
      "learning_rate": 2.709226350773774e-06,
      "loss": 0.1653,
      "step": 7240
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3395639714710454,
      "learning_rate": 2.7079102393456503e-06,
      "loss": 0.1875,
      "step": 7241
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3489262759528924,
      "learning_rate": 2.7065943289491307e-06,
      "loss": 0.1923,
      "step": 7242
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3786089018682015,
      "learning_rate": 2.705278619699633e-06,
      "loss": 0.1621,
      "step": 7243
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.416832208845116,
      "learning_rate": 2.7039631117125487e-06,
      "loss": 0.1856,
      "step": 7244
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4794479342372466,
      "learning_rate": 2.7026478051032625e-06,
      "loss": 0.1802,
      "step": 7245
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.41492493167022,
      "learning_rate": 2.7013326999871297e-06,
      "loss": 0.1726,
      "step": 7246
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3409442294274898,
      "learning_rate": 2.700017796479499e-06,
      "loss": 0.1716,
      "step": 7247
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4200164204182653,
      "learning_rate": 2.6987030946956937e-06,
      "loss": 0.146,
      "step": 7248
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3916727082161489,
      "learning_rate": 2.69738859475102e-06,
      "loss": 0.1751,
      "step": 7249
    },
    {
      "epoch": 0.66,
      "grad_norm": 4.95611617747841,
      "learning_rate": 2.6960742967607726e-06,
      "loss": 0.3433,
      "step": 7250
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3620527806182223,
      "learning_rate": 2.6947602008402207e-06,
      "loss": 0.1617,
      "step": 7251
    },
    {
      "epoch": 0.66,
      "grad_norm": 11.514941335749178,
      "learning_rate": 2.693446307104622e-06,
      "loss": 0.5719,
      "step": 7252
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3494436098479632,
      "learning_rate": 2.6921326156692105e-06,
      "loss": 0.2239,
      "step": 7253
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3706154264672386,
      "learning_rate": 2.6908191266492095e-06,
      "loss": 0.2351,
      "step": 7254
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.288866053400438,
      "learning_rate": 2.689505840159817e-06,
      "loss": 0.1583,
      "step": 7255
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.309051907396566,
      "learning_rate": 2.6881927563162215e-06,
      "loss": 0.1821,
      "step": 7256
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4060460503472174,
      "learning_rate": 2.6868798752335867e-06,
      "loss": 0.2068,
      "step": 7257
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3290995268074637,
      "learning_rate": 2.685567197027058e-06,
      "loss": 0.2004,
      "step": 7258
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.375682922815664,
      "learning_rate": 2.684254721811772e-06,
      "loss": 0.1651,
      "step": 7259
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.373208787968508,
      "learning_rate": 2.6829424497028365e-06,
      "loss": 0.1782,
      "step": 7260
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4613876219477437,
      "learning_rate": 2.681630380815351e-06,
      "loss": 0.1864,
      "step": 7261
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.500524242808137,
      "learning_rate": 2.6803185152643885e-06,
      "loss": 0.1851,
      "step": 7262
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5529991754870882,
      "learning_rate": 2.6790068531650116e-06,
      "loss": 0.2042,
      "step": 7263
    },
    {
      "epoch": 0.66,
      "grad_norm": 8.29327010447963,
      "learning_rate": 2.677695394632259e-06,
      "loss": 0.7513,
      "step": 7264
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.576688998914018,
      "learning_rate": 2.6763841397811576e-06,
      "loss": 0.1869,
      "step": 7265
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.407072543185537,
      "learning_rate": 2.6750730887267117e-06,
      "loss": 0.1778,
      "step": 7266
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3750547132972892,
      "learning_rate": 2.673762241583907e-06,
      "loss": 0.2314,
      "step": 7267
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.414537618106395,
      "learning_rate": 2.6724515984677187e-06,
      "loss": 0.1842,
      "step": 7268
    },
    {
      "epoch": 0.66,
      "grad_norm": 5.684046033065117,
      "learning_rate": 2.6711411594930927e-06,
      "loss": 0.375,
      "step": 7269
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3901653177537525,
      "learning_rate": 2.669830924774969e-06,
      "loss": 0.1638,
      "step": 7270
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1060040441895551,
      "learning_rate": 2.668520894428259e-06,
      "loss": 0.1431,
      "step": 7271
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.371896794082595,
      "learning_rate": 2.6672110685678665e-06,
      "loss": 0.2116,
      "step": 7272
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7218095666031852,
      "learning_rate": 2.6659014473086665e-06,
      "loss": 0.21,
      "step": 7273
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4091647424171188,
      "learning_rate": 2.6645920307655263e-06,
      "loss": 0.181,
      "step": 7274
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2712847019332425,
      "learning_rate": 2.663282819053288e-06,
      "loss": 0.1485,
      "step": 7275
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5627086533397,
      "learning_rate": 2.6619738122867774e-06,
      "loss": 0.224,
      "step": 7276
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3205091271700553,
      "learning_rate": 2.660665010580805e-06,
      "loss": 0.1863,
      "step": 7277
    },
    {
      "epoch": 0.67,
      "grad_norm": 9.377584180652404,
      "learning_rate": 2.65935641405016e-06,
      "loss": 0.4665,
      "step": 7278
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.50492408199879,
      "learning_rate": 2.658048022809617e-06,
      "loss": 0.5799,
      "step": 7279
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.339725660446328,
      "learning_rate": 2.6567398369739282e-06,
      "loss": 0.1892,
      "step": 7280
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3159695716916657,
      "learning_rate": 2.655431856657833e-06,
      "loss": 0.1842,
      "step": 7281
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6649111780978714,
      "learning_rate": 2.6541240819760485e-06,
      "loss": 0.1922,
      "step": 7282
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2956544690265936,
      "learning_rate": 2.6528165130432735e-06,
      "loss": 0.1545,
      "step": 7283
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3854740647680348,
      "learning_rate": 2.6515091499741946e-06,
      "loss": 0.1543,
      "step": 7284
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4118026229447902,
      "learning_rate": 2.650201992883472e-06,
      "loss": 0.1701,
      "step": 7285
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4191403390738644,
      "learning_rate": 2.648895041885756e-06,
      "loss": 0.2228,
      "step": 7286
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.133388656783793,
      "learning_rate": 2.647588297095671e-06,
      "loss": 0.6245,
      "step": 7287
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4858657285500032,
      "learning_rate": 2.6462817586278315e-06,
      "loss": 0.1938,
      "step": 7288
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5363052738518819,
      "learning_rate": 2.6449754265968263e-06,
      "loss": 0.168,
      "step": 7289
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.330533010711495,
      "learning_rate": 2.643669301117232e-06,
      "loss": 0.2112,
      "step": 7290
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.328234094860274,
      "learning_rate": 2.642363382303603e-06,
      "loss": 0.1721,
      "step": 7291
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3441824671451506,
      "learning_rate": 2.641057670270476e-06,
      "loss": 0.133,
      "step": 7292
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.310531109354768,
      "learning_rate": 2.6397521651323744e-06,
      "loss": 0.1902,
      "step": 7293
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4747737979661726,
      "learning_rate": 2.6384468670037946e-06,
      "loss": 0.2412,
      "step": 7294
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4592551275040346,
      "learning_rate": 2.6371417759992246e-06,
      "loss": 0.1544,
      "step": 7295
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3850106189471019,
      "learning_rate": 2.6358368922331266e-06,
      "loss": 0.1628,
      "step": 7296
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.453289110750966,
      "learning_rate": 2.6345322158199503e-06,
      "loss": 0.2072,
      "step": 7297
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.364705331784067,
      "learning_rate": 2.633227746874122e-06,
      "loss": 0.1767,
      "step": 7298
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3054850647255927,
      "learning_rate": 2.631923485510054e-06,
      "loss": 0.1563,
      "step": 7299
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2812663139511975,
      "learning_rate": 2.630619431842139e-06,
      "loss": 0.1423,
      "step": 7300
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2080087989628328,
      "learning_rate": 2.6293155859847487e-06,
      "loss": 0.1625,
      "step": 7301
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3543885234442328,
      "learning_rate": 2.628011948052242e-06,
      "loss": 0.1697,
      "step": 7302
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2225741113186361,
      "learning_rate": 2.6267085181589537e-06,
      "loss": 0.1635,
      "step": 7303
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5475286501265864,
      "learning_rate": 2.6254052964192077e-06,
      "loss": 0.1961,
      "step": 7304
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.305736125306503,
      "learning_rate": 2.6241022829473e-06,
      "loss": 0.1813,
      "step": 7305
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2769463645993724,
      "learning_rate": 2.622799477857518e-06,
      "loss": 0.2066,
      "step": 7306
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4134049174829186,
      "learning_rate": 2.6214968812641237e-06,
      "loss": 0.1819,
      "step": 7307
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2578944012093571,
      "learning_rate": 2.6201944932813663e-06,
      "loss": 0.194,
      "step": 7308
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4809813540504828,
      "learning_rate": 2.618892314023471e-06,
      "loss": 0.1979,
      "step": 7309
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.300796741104481,
      "learning_rate": 2.617590343604648e-06,
      "loss": 0.1807,
      "step": 7310
    },
    {
      "epoch": 0.67,
      "grad_norm": 13.667474575780952,
      "learning_rate": 2.6162885821390914e-06,
      "loss": 0.5674,
      "step": 7311
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6086258159156996,
      "learning_rate": 2.6149870297409707e-06,
      "loss": 0.2133,
      "step": 7312
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.370479046878188,
      "learning_rate": 2.6136856865244443e-06,
      "loss": 0.1718,
      "step": 7313
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.477137127593329,
      "learning_rate": 2.6123845526036453e-06,
      "loss": 0.1907,
      "step": 7314
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.367227935786881,
      "learning_rate": 2.6110836280926952e-06,
      "loss": 0.1597,
      "step": 7315
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.720315694626513,
      "learning_rate": 2.609782913105691e-06,
      "loss": 0.4981,
      "step": 7316
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.621908168361162,
      "learning_rate": 2.6084824077567163e-06,
      "loss": 0.5879,
      "step": 7317
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4487303636061242,
      "learning_rate": 2.607182112159834e-06,
      "loss": 0.1757,
      "step": 7318
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.8352144537263495,
      "learning_rate": 2.605882026429085e-06,
      "loss": 0.7307,
      "step": 7319
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3446353531774067,
      "learning_rate": 2.6045821506785008e-06,
      "loss": 0.2132,
      "step": 7320
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3070625488407195,
      "learning_rate": 2.603282485022085e-06,
      "loss": 0.1667,
      "step": 7321
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2663786362417775,
      "learning_rate": 2.60198302957383e-06,
      "loss": 0.153,
      "step": 7322
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4006047393623828,
      "learning_rate": 2.600683784447704e-06,
      "loss": 0.1587,
      "step": 7323
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2801747008909807,
      "learning_rate": 2.599384749757663e-06,
      "loss": 0.1912,
      "step": 7324
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.417832838089726,
      "learning_rate": 2.598085925617639e-06,
      "loss": 0.7232,
      "step": 7325
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3783520943346101,
      "learning_rate": 2.5967873121415455e-06,
      "loss": 0.1725,
      "step": 7326
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5729989903150075,
      "learning_rate": 2.5954889094432835e-06,
      "loss": 0.1925,
      "step": 7327
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2258475196848455,
      "learning_rate": 2.5941907176367276e-06,
      "loss": 0.1609,
      "step": 7328
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5139328038711355,
      "learning_rate": 2.592892736835742e-06,
      "loss": 0.1985,
      "step": 7329
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3863478932795912,
      "learning_rate": 2.5915949671541646e-06,
      "loss": 0.1946,
      "step": 7330
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2659234879438963,
      "learning_rate": 2.590297408705822e-06,
      "loss": 0.2068,
      "step": 7331
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1999648925458142,
      "learning_rate": 2.5890000616045154e-06,
      "loss": 0.1564,
      "step": 7332
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3523758868390385,
      "learning_rate": 2.587702925964034e-06,
      "loss": 0.1876,
      "step": 7333
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.176071222385831,
      "learning_rate": 2.586406001898144e-06,
      "loss": 0.6009,
      "step": 7334
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3929415819164883,
      "learning_rate": 2.585109289520592e-06,
      "loss": 0.1972,
      "step": 7335
    },
    {
      "epoch": 0.67,
      "grad_norm": 4.714717800941463,
      "learning_rate": 2.5838127889451113e-06,
      "loss": 0.4237,
      "step": 7336
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4620407834176847,
      "learning_rate": 2.5825165002854124e-06,
      "loss": 0.2068,
      "step": 7337
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.672599315462404,
      "learning_rate": 2.5812204236551898e-06,
      "loss": 0.5511,
      "step": 7338
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.162909374776181,
      "learning_rate": 2.5799245591681156e-06,
      "loss": 0.6468,
      "step": 7339
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1660599519699781,
      "learning_rate": 2.5786289069378482e-06,
      "loss": 0.1326,
      "step": 7340
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5768108330151296,
      "learning_rate": 2.577333467078023e-06,
      "loss": 0.1971,
      "step": 7341
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3533441404588304,
      "learning_rate": 2.576038239702262e-06,
      "loss": 0.2015,
      "step": 7342
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.828619076160219,
      "learning_rate": 2.5747432249241623e-06,
      "loss": 0.4917,
      "step": 7343
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.533493019359749,
      "learning_rate": 2.5734484228573044e-06,
      "loss": 0.5306,
      "step": 7344
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4390470544921863,
      "learning_rate": 2.5721538336152553e-06,
      "loss": 0.2267,
      "step": 7345
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4472538460595308,
      "learning_rate": 2.570859457311554e-06,
      "loss": 0.1586,
      "step": 7346
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2133874725765381,
      "learning_rate": 2.5695652940597315e-06,
      "loss": 0.1707,
      "step": 7347
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3022627260493693,
      "learning_rate": 2.5682713439732887e-06,
      "loss": 0.1485,
      "step": 7348
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2427475909158012,
      "learning_rate": 2.5669776071657194e-06,
      "loss": 0.1615,
      "step": 7349
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0787564857286787,
      "learning_rate": 2.5656840837504884e-06,
      "loss": 0.1207,
      "step": 7350
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4691733983810618,
      "learning_rate": 2.56439077384105e-06,
      "loss": 0.1862,
      "step": 7351
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3414009651922447,
      "learning_rate": 2.5630976775508344e-06,
      "loss": 0.2141,
      "step": 7352
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.419708580256591,
      "learning_rate": 2.5618047949932524e-06,
      "loss": 0.1853,
      "step": 7353
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3398463604950523,
      "learning_rate": 2.5605121262817033e-06,
      "loss": 0.1728,
      "step": 7354
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4738838026273924,
      "learning_rate": 2.559219671529558e-06,
      "loss": 0.1825,
      "step": 7355
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3445977921650358,
      "learning_rate": 2.557927430850178e-06,
      "loss": 0.1776,
      "step": 7356
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3558399511122743,
      "learning_rate": 2.5566354043568977e-06,
      "loss": 0.1599,
      "step": 7357
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1661815732467802,
      "learning_rate": 2.5553435921630387e-06,
      "loss": 0.1503,
      "step": 7358
    },
    {
      "epoch": 0.67,
      "grad_norm": 7.471907284225069,
      "learning_rate": 2.5540519943818996e-06,
      "loss": 0.6201,
      "step": 7359
    },
    {
      "epoch": 0.67,
      "grad_norm": 11.38741744806102,
      "learning_rate": 2.552760611126765e-06,
      "loss": 0.5345,
      "step": 7360
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.330028371615192,
      "learning_rate": 2.5514694425108968e-06,
      "loss": 0.174,
      "step": 7361
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.301666060401134,
      "learning_rate": 2.550178488647536e-06,
      "loss": 0.1804,
      "step": 7362
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2819574183853937,
      "learning_rate": 2.5488877496499133e-06,
      "loss": 0.2084,
      "step": 7363
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6319232600385174,
      "learning_rate": 2.5475972256312292e-06,
      "loss": 0.1888,
      "step": 7364
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3763529589667396,
      "learning_rate": 2.5463069167046774e-06,
      "loss": 0.1609,
      "step": 7365
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6482555536505727,
      "learning_rate": 2.5450168229834205e-06,
      "loss": 0.2491,
      "step": 7366
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2899973368030773,
      "learning_rate": 2.5437269445806146e-06,
      "loss": 0.1877,
      "step": 7367
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.443929942888394,
      "learning_rate": 2.542437281609387e-06,
      "loss": 0.1812,
      "step": 7368
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3559706855899878,
      "learning_rate": 2.5411478341828475e-06,
      "loss": 0.1831,
      "step": 7369
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4341026971477493,
      "learning_rate": 2.5398586024140953e-06,
      "loss": 0.2156,
      "step": 7370
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.18226408637218,
      "learning_rate": 2.5385695864161987e-06,
      "loss": 0.1295,
      "step": 7371
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.510127680883411,
      "learning_rate": 2.5372807863022186e-06,
      "loss": 0.1762,
      "step": 7372
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3184748301233462,
      "learning_rate": 2.5359922021851858e-06,
      "loss": 0.1312,
      "step": 7373
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4047992078364675,
      "learning_rate": 2.5347038341781228e-06,
      "loss": 0.1798,
      "step": 7374
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4877330974189844,
      "learning_rate": 2.5334156823940237e-06,
      "loss": 0.1646,
      "step": 7375
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4499768967898738,
      "learning_rate": 2.5321277469458715e-06,
      "loss": 0.1984,
      "step": 7376
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.368741031577755,
      "learning_rate": 2.5308400279466262e-06,
      "loss": 0.2605,
      "step": 7377
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4574416039303846,
      "learning_rate": 2.5295525255092256e-06,
      "loss": 0.1681,
      "step": 7378
    },
    {
      "epoch": 0.67,
      "grad_norm": 5.455549766906659,
      "learning_rate": 2.5282652397465984e-06,
      "loss": 0.672,
      "step": 7379
    },
    {
      "epoch": 0.67,
      "grad_norm": 6.321504867759982,
      "learning_rate": 2.5269781707716423e-06,
      "loss": 0.5851,
      "step": 7380
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4008177364358607,
      "learning_rate": 2.5256913186972465e-06,
      "loss": 0.1866,
      "step": 7381
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2731188563391374,
      "learning_rate": 2.524404683636273e-06,
      "loss": 0.1738,
      "step": 7382
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1701920224757931,
      "learning_rate": 2.5231182657015717e-06,
      "loss": 0.1374,
      "step": 7383
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.652679870760581,
      "learning_rate": 2.521832065005966e-06,
      "loss": 0.2095,
      "step": 7384
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.395036725348142,
      "learning_rate": 2.5205460816622684e-06,
      "loss": 0.1945,
      "step": 7385
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4301740251012383,
      "learning_rate": 2.519260315783267e-06,
      "loss": 0.1871,
      "step": 7386
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2120206820628805,
      "learning_rate": 2.517974767481729e-06,
      "loss": 0.1606,
      "step": 7387
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4103502145297773,
      "learning_rate": 2.51668943687041e-06,
      "loss": 0.1586,
      "step": 7388
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5186444662889789,
      "learning_rate": 2.5154043240620385e-06,
      "loss": 0.2155,
      "step": 7389
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5841244023293983,
      "learning_rate": 2.5141194291693306e-06,
      "loss": 0.1807,
      "step": 7390
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.411780916296455,
      "learning_rate": 2.5128347523049774e-06,
      "loss": 0.1901,
      "step": 7391
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4351678202915157,
      "learning_rate": 2.5115502935816565e-06,
      "loss": 0.1895,
      "step": 7392
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1560992186122039,
      "learning_rate": 2.5102660531120204e-06,
      "loss": 0.15,
      "step": 7393
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4149970338973754,
      "learning_rate": 2.508982031008709e-06,
      "loss": 0.1947,
      "step": 7394
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3524100260797804,
      "learning_rate": 2.507698227384337e-06,
      "loss": 0.1936,
      "step": 7395
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3387492141278807,
      "learning_rate": 2.5064146423515024e-06,
      "loss": 0.1454,
      "step": 7396
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2152532673262657,
      "learning_rate": 2.5051312760227864e-06,
      "loss": 0.1695,
      "step": 7397
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2870005552171895,
      "learning_rate": 2.5038481285107458e-06,
      "loss": 0.1968,
      "step": 7398
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4210147817555059,
      "learning_rate": 2.502565199927925e-06,
      "loss": 0.2252,
      "step": 7399
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5439721965334272,
      "learning_rate": 2.5012824903868415e-06,
      "loss": 0.1818,
      "step": 7400
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4056916158364368,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.2019,
      "step": 7401
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5152870953670439,
      "learning_rate": 2.4987177288798837e-06,
      "loss": 0.1987,
      "step": 7402
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2906004119185572,
      "learning_rate": 2.497435677138956e-06,
      "loss": 0.1703,
      "step": 7403
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3322635967680747,
      "learning_rate": 2.496153844889661e-06,
      "loss": 0.1616,
      "step": 7404
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.993569738311419,
      "learning_rate": 2.494872232244422e-06,
      "loss": 0.4819,
      "step": 7405
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.254548755841848,
      "learning_rate": 2.493590839315649e-06,
      "loss": 0.1735,
      "step": 7406
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.714109525390379,
      "learning_rate": 2.4923096662157244e-06,
      "loss": 0.6618,
      "step": 7407
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3183663494403832,
      "learning_rate": 2.49102871305702e-06,
      "loss": 0.1932,
      "step": 7408
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3962602453218549,
      "learning_rate": 2.4897479799518797e-06,
      "loss": 0.1695,
      "step": 7409
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5111443437667762,
      "learning_rate": 2.4884674670126367e-06,
      "loss": 0.1951,
      "step": 7410
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3777596184723757,
      "learning_rate": 2.4871871743515987e-06,
      "loss": 0.1949,
      "step": 7411
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.30067546592178,
      "learning_rate": 2.4859071020810533e-06,
      "loss": 0.1711,
      "step": 7412
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4203748396091138,
      "learning_rate": 2.484627250313276e-06,
      "loss": 0.1806,
      "step": 7413
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4347504951297583,
      "learning_rate": 2.4833476191605136e-06,
      "loss": 0.183,
      "step": 7414
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5013475843670216,
      "learning_rate": 2.4820682087350033e-06,
      "loss": 0.1814,
      "step": 7415
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2781321288899905,
      "learning_rate": 2.480789019148953e-06,
      "loss": 0.175,
      "step": 7416
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4348344785626534,
      "learning_rate": 2.479510050514561e-06,
      "loss": 0.1957,
      "step": 7417
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3846206302484414,
      "learning_rate": 2.478231302943997e-06,
      "loss": 0.1776,
      "step": 7418
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.058576344824263,
      "learning_rate": 2.4769527765494194e-06,
      "loss": 0.4969,
      "step": 7419
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3674657118501552,
      "learning_rate": 2.475674471442962e-06,
      "loss": 0.1858,
      "step": 7420
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4810982111264883,
      "learning_rate": 2.4743963877367388e-06,
      "loss": 0.1975,
      "step": 7421
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3447696223825087,
      "learning_rate": 2.47311852554285e-06,
      "loss": 0.1931,
      "step": 7422
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4646476972185332,
      "learning_rate": 2.471840884973369e-06,
      "loss": 0.2051,
      "step": 7423
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5820499847272342,
      "learning_rate": 2.4705634661403567e-06,
      "loss": 0.2116,
      "step": 7424
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6440658494794318,
      "learning_rate": 2.469286269155848e-06,
      "loss": 0.2372,
      "step": 7425
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.70252193820834,
      "learning_rate": 2.468009294131865e-06,
      "loss": 0.5197,
      "step": 7426
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3122592522458747,
      "learning_rate": 2.466732541180404e-06,
      "loss": 0.189,
      "step": 7427
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3510492545418005,
      "learning_rate": 2.465456010413447e-06,
      "loss": 0.1759,
      "step": 7428
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.304267182643206,
      "learning_rate": 2.4641797019429544e-06,
      "loss": 0.4256,
      "step": 7429
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2082975333505097,
      "learning_rate": 2.4629036158808633e-06,
      "loss": 0.1621,
      "step": 7430
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2693905275666002,
      "learning_rate": 2.4616277523391e-06,
      "loss": 0.1516,
      "step": 7431
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4078598372798334,
      "learning_rate": 2.460352111429562e-06,
      "loss": 0.1764,
      "step": 7432
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4122764892169655,
      "learning_rate": 2.4590766932641353e-06,
      "loss": 0.1891,
      "step": 7433
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.0885737810189955,
      "learning_rate": 2.457801497954679e-06,
      "loss": 0.5422,
      "step": 7434
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.509401515074584,
      "learning_rate": 2.4565265256130393e-06,
      "loss": 0.575,
      "step": 7435
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.251239144740075,
      "learning_rate": 2.4552517763510376e-06,
      "loss": 0.1721,
      "step": 7436
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1776838662550144,
      "learning_rate": 2.4539772502804797e-06,
      "loss": 0.1468,
      "step": 7437
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3820040633109136,
      "learning_rate": 2.4527029475131495e-06,
      "loss": 0.2089,
      "step": 7438
    },
    {
      "epoch": 0.68,
      "grad_norm": 18.816929253485746,
      "learning_rate": 2.4514288681608105e-06,
      "loss": 0.707,
      "step": 7439
    },
    {
      "epoch": 0.68,
      "grad_norm": 9.082288657982202,
      "learning_rate": 2.4501550123352105e-06,
      "loss": 0.4844,
      "step": 7440
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.457658132046944,
      "learning_rate": 2.4488813801480717e-06,
      "loss": 0.2364,
      "step": 7441
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.56133627974026,
      "learning_rate": 2.447607971711104e-06,
      "loss": 0.5688,
      "step": 7442
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4231194373584783,
      "learning_rate": 2.4463347871359906e-06,
      "loss": 0.2164,
      "step": 7443
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4912032645347664,
      "learning_rate": 2.445061826534401e-06,
      "loss": 0.1957,
      "step": 7444
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4269102936631832,
      "learning_rate": 2.44378909001798e-06,
      "loss": 0.1687,
      "step": 7445
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4342877962554759,
      "learning_rate": 2.4425165776983577e-06,
      "loss": 0.2,
      "step": 7446
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.353608047591945,
      "learning_rate": 2.4412442896871407e-06,
      "loss": 0.1772,
      "step": 7447
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3730953609826178,
      "learning_rate": 2.4399722260959152e-06,
      "loss": 0.1321,
      "step": 7448
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.485276882649135,
      "learning_rate": 2.438700387036253e-06,
      "loss": 0.3701,
      "step": 7449
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3787414551569817,
      "learning_rate": 2.4374287726196994e-06,
      "loss": 0.169,
      "step": 7450
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4063914684865235,
      "learning_rate": 2.4361573829577874e-06,
      "loss": 0.1629,
      "step": 7451
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5707199104405742,
      "learning_rate": 2.4348862181620225e-06,
      "loss": 0.1873,
      "step": 7452
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3811576747254026,
      "learning_rate": 2.4336152783438984e-06,
      "loss": 0.2066,
      "step": 7453
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3663259586140624,
      "learning_rate": 2.4323445636148824e-06,
      "loss": 0.1913,
      "step": 7454
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4258243530390677,
      "learning_rate": 2.4310740740864237e-06,
      "loss": 0.1887,
      "step": 7455
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.39195797795973,
      "learning_rate": 2.429803809869956e-06,
      "loss": 0.1837,
      "step": 7456
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2869243004581041,
      "learning_rate": 2.4285337710768843e-06,
      "loss": 0.1535,
      "step": 7457
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.290201178562365,
      "learning_rate": 2.427263957818607e-06,
      "loss": 0.1692,
      "step": 7458
    },
    {
      "epoch": 0.68,
      "grad_norm": 7.374051590127174,
      "learning_rate": 2.425994370206492e-06,
      "loss": 0.583,
      "step": 7459
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5141039436370922,
      "learning_rate": 2.4247250083518875e-06,
      "loss": 0.1724,
      "step": 7460
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4249938097628942,
      "learning_rate": 2.4234558723661306e-06,
      "loss": 0.1631,
      "step": 7461
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3661086743230875,
      "learning_rate": 2.4221869623605275e-06,
      "loss": 0.1647,
      "step": 7462
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.476936232080037,
      "learning_rate": 2.4209182784463754e-06,
      "loss": 0.2024,
      "step": 7463
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.344276301466382,
      "learning_rate": 2.4196498207349412e-06,
      "loss": 0.1994,
      "step": 7464
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5813608721271293,
      "learning_rate": 2.4183815893374817e-06,
      "loss": 0.1925,
      "step": 7465
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4178456364993894,
      "learning_rate": 2.4171135843652256e-06,
      "loss": 0.1837,
      "step": 7466
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4274231805098245,
      "learning_rate": 2.415845805929388e-06,
      "loss": 0.1762,
      "step": 7467
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4828082437300067,
      "learning_rate": 2.4145782541411617e-06,
      "loss": 0.1912,
      "step": 7468
    },
    {
      "epoch": 0.68,
      "grad_norm": 5.005571756251913,
      "learning_rate": 2.4133109291117156e-06,
      "loss": 0.6039,
      "step": 7469
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4711820066316712,
      "learning_rate": 2.4120438309522077e-06,
      "loss": 0.2167,
      "step": 7470
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.20075836911554,
      "learning_rate": 2.410776959773767e-06,
      "loss": 0.1713,
      "step": 7471
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1696799256362544,
      "learning_rate": 2.4095103156875093e-06,
      "loss": 0.1624,
      "step": 7472
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.262605669992556,
      "learning_rate": 2.4082438988045253e-06,
      "loss": 0.1725,
      "step": 7473
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3796441175810643,
      "learning_rate": 2.4069777092358914e-06,
      "loss": 0.1786,
      "step": 7474
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4648120689962938,
      "learning_rate": 2.4057117470926574e-06,
      "loss": 0.1828,
      "step": 7475
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3134310849685205,
      "learning_rate": 2.40444601248586e-06,
      "loss": 0.1587,
      "step": 7476
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3564777270345845,
      "learning_rate": 2.403180505526511e-06,
      "loss": 0.1947,
      "step": 7477
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4212212413099854,
      "learning_rate": 2.401915226325602e-06,
      "loss": 0.1934,
      "step": 7478
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2808420095077586,
      "learning_rate": 2.4006501749941097e-06,
      "loss": 0.1122,
      "step": 7479
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2759295378825097,
      "learning_rate": 2.399385351642984e-06,
      "loss": 0.16,
      "step": 7480
    },
    {
      "epoch": 0.68,
      "grad_norm": 6.15050415033382,
      "learning_rate": 2.3981207563831633e-06,
      "loss": 0.5275,
      "step": 7481
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4097692438945748,
      "learning_rate": 2.396856389325556e-06,
      "loss": 0.1729,
      "step": 7482
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5096262900145674,
      "learning_rate": 2.395592250581059e-06,
      "loss": 0.1974,
      "step": 7483
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4226827444789518,
      "learning_rate": 2.394328340260543e-06,
      "loss": 0.1855,
      "step": 7484
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.391367706431759,
      "learning_rate": 2.3930646584748652e-06,
      "loss": 0.1724,
      "step": 7485
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3303078152061176,
      "learning_rate": 2.391801205334856e-06,
      "loss": 0.1436,
      "step": 7486
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7117039833326468,
      "learning_rate": 2.390537980951328e-06,
      "loss": 0.1694,
      "step": 7487
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.186665270692279,
      "learning_rate": 2.389274985435077e-06,
      "loss": 0.1677,
      "step": 7488
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1221029063255321,
      "learning_rate": 2.388012218896873e-06,
      "loss": 0.1341,
      "step": 7489
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.472820497228309,
      "learning_rate": 2.386749681447473e-06,
      "loss": 0.1921,
      "step": 7490
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2958690440944727,
      "learning_rate": 2.3854873731976065e-06,
      "loss": 0.1365,
      "step": 7491
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3414014486919816,
      "learning_rate": 2.384225294257989e-06,
      "loss": 0.1855,
      "step": 7492
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3097898850946532,
      "learning_rate": 2.382963444739311e-06,
      "loss": 0.1442,
      "step": 7493
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4273788866566894,
      "learning_rate": 2.3817018247522477e-06,
      "loss": 0.2104,
      "step": 7494
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5156098848433588,
      "learning_rate": 2.38044043440745e-06,
      "loss": 0.1781,
      "step": 7495
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3652976241974382,
      "learning_rate": 2.379179273815549e-06,
      "loss": 0.1856,
      "step": 7496
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4418064423827066,
      "learning_rate": 2.3779183430871596e-06,
      "loss": 0.2108,
      "step": 7497
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.458146120781622,
      "learning_rate": 2.3766576423328707e-06,
      "loss": 0.1854,
      "step": 7498
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3294572050456697,
      "learning_rate": 2.3753971716632583e-06,
      "loss": 0.1641,
      "step": 7499
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3455437989746248,
      "learning_rate": 2.3741369311888702e-06,
      "loss": 0.1681,
      "step": 7500
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.265986822989652,
      "learning_rate": 2.3728769210202413e-06,
      "loss": 0.1754,
      "step": 7501
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4250887124301614,
      "learning_rate": 2.371617141267881e-06,
      "loss": 0.1904,
      "step": 7502
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3623976130760864,
      "learning_rate": 2.3703575920422793e-06,
      "loss": 0.2197,
      "step": 7503
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3506241734549247,
      "learning_rate": 2.3690982734539104e-06,
      "loss": 0.1714,
      "step": 7504
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5865846136892754,
      "learning_rate": 2.3678391856132203e-06,
      "loss": 0.1766,
      "step": 7505
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.25856587949767,
      "learning_rate": 2.366580328630645e-06,
      "loss": 0.1803,
      "step": 7506
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4444895308497967,
      "learning_rate": 2.3653217026165894e-06,
      "loss": 0.1863,
      "step": 7507
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1975170110411828,
      "learning_rate": 2.3640633076814477e-06,
      "loss": 0.1867,
      "step": 7508
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.3019346686716915,
      "learning_rate": 2.3628051439355858e-06,
      "loss": 0.5196,
      "step": 7509
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.294763870223664,
      "learning_rate": 2.361547211489357e-06,
      "loss": 0.1725,
      "step": 7510
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.226709435689056,
      "learning_rate": 2.360289510453088e-06,
      "loss": 0.1627,
      "step": 7511
    },
    {
      "epoch": 0.69,
      "grad_norm": 4.393051888757235,
      "learning_rate": 2.3590320409370865e-06,
      "loss": 0.3701,
      "step": 7512
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4408026057461163,
      "learning_rate": 2.3577748030516443e-06,
      "loss": 0.2295,
      "step": 7513
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3329241059906176,
      "learning_rate": 2.3565177969070257e-06,
      "loss": 0.1618,
      "step": 7514
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4414152389382315,
      "learning_rate": 2.3552610226134827e-06,
      "loss": 0.2038,
      "step": 7515
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5132887099436518,
      "learning_rate": 2.3540044802812382e-06,
      "loss": 0.2025,
      "step": 7516
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3521610824238754,
      "learning_rate": 2.3527481700205045e-06,
      "loss": 0.1827,
      "step": 7517
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.416982957211356,
      "learning_rate": 2.3514920919414636e-06,
      "loss": 0.1791,
      "step": 7518
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4853618292966517,
      "learning_rate": 2.3502362461542853e-06,
      "loss": 0.1809,
      "step": 7519
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.388022188836477,
      "learning_rate": 2.3489806327691156e-06,
      "loss": 0.1858,
      "step": 7520
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2527985812023323,
      "learning_rate": 2.3477252518960764e-06,
      "loss": 0.205,
      "step": 7521
    },
    {
      "epoch": 0.69,
      "grad_norm": 5.993720605733375,
      "learning_rate": 2.346470103645278e-06,
      "loss": 0.5316,
      "step": 7522
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.365872625286382,
      "learning_rate": 2.345215188126801e-06,
      "loss": 0.1629,
      "step": 7523
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3597832255632,
      "learning_rate": 2.3439605054507135e-06,
      "loss": 0.1758,
      "step": 7524
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.344655326143506,
      "learning_rate": 2.3427060557270563e-06,
      "loss": 0.2052,
      "step": 7525
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2525417862234587,
      "learning_rate": 2.3414518390658562e-06,
      "loss": 0.1372,
      "step": 7526
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3940172636660058,
      "learning_rate": 2.3401978555771136e-06,
      "loss": 0.1725,
      "step": 7527
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3773643643949538,
      "learning_rate": 2.338944105370814e-06,
      "loss": 0.1323,
      "step": 7528
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4287504577550378,
      "learning_rate": 2.3376905885569185e-06,
      "loss": 0.1908,
      "step": 7529
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4228198716661997,
      "learning_rate": 2.3364373052453667e-06,
      "loss": 0.2209,
      "step": 7530
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4870118476828698,
      "learning_rate": 2.335184255546083e-06,
      "loss": 0.1943,
      "step": 7531
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3802011904717417,
      "learning_rate": 2.3339314395689656e-06,
      "loss": 0.2133,
      "step": 7532
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3454685433788338,
      "learning_rate": 2.3326788574238985e-06,
      "loss": 0.1672,
      "step": 7533
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3438840402753789,
      "learning_rate": 2.3314265092207365e-06,
      "loss": 0.2195,
      "step": 7534
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.275123750563231,
      "learning_rate": 2.3301743950693247e-06,
      "loss": 0.1777,
      "step": 7535
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1937274949348398,
      "learning_rate": 2.3289225150794766e-06,
      "loss": 0.1679,
      "step": 7536
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4120613409689278,
      "learning_rate": 2.3276708693609947e-06,
      "loss": 0.1924,
      "step": 7537
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3700907276676257,
      "learning_rate": 2.326419458023655e-06,
      "loss": 0.198,
      "step": 7538
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3554060826808039,
      "learning_rate": 2.325168281177212e-06,
      "loss": 0.1557,
      "step": 7539
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3140611879832067,
      "learning_rate": 2.3239173389314066e-06,
      "loss": 0.1645,
      "step": 7540
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.220167910020197,
      "learning_rate": 2.322666631395952e-06,
      "loss": 0.1415,
      "step": 7541
    },
    {
      "epoch": 0.69,
      "grad_norm": 8.350631655655079,
      "learning_rate": 2.3214161586805455e-06,
      "loss": 0.6236,
      "step": 7542
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5068173805206115,
      "learning_rate": 2.3201659208948597e-06,
      "loss": 0.2082,
      "step": 7543
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3906737102449889,
      "learning_rate": 2.3189159181485517e-06,
      "loss": 0.1847,
      "step": 7544
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4343502795229572,
      "learning_rate": 2.3176661505512534e-06,
      "loss": 0.2116,
      "step": 7545
    },
    {
      "epoch": 0.69,
      "grad_norm": 7.0822872973796915,
      "learning_rate": 2.316416618212576e-06,
      "loss": 0.4572,
      "step": 7546
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1434876562185075,
      "learning_rate": 2.3151673212421157e-06,
      "loss": 0.1314,
      "step": 7547
    },
    {
      "epoch": 0.69,
      "grad_norm": 25.553306549110477,
      "learning_rate": 2.3139182597494398e-06,
      "loss": 0.5302,
      "step": 7548
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3417941958368873,
      "learning_rate": 2.312669433844104e-06,
      "loss": 0.1709,
      "step": 7549
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3809864928131634,
      "learning_rate": 2.3114208436356346e-06,
      "loss": 0.1793,
      "step": 7550
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2100238891055228,
      "learning_rate": 2.310172489233545e-06,
      "loss": 0.1444,
      "step": 7551
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.479686834821447,
      "learning_rate": 2.3089243707473208e-06,
      "loss": 0.1876,
      "step": 7552
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2198205608106796,
      "learning_rate": 2.3076764882864333e-06,
      "loss": 0.1672,
      "step": 7553
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3606599198516205,
      "learning_rate": 2.306428841960329e-06,
      "loss": 0.169,
      "step": 7554
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3458231743451083,
      "learning_rate": 2.3051814318784325e-06,
      "loss": 0.1255,
      "step": 7555
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.382192143591041,
      "learning_rate": 2.3039342581501544e-06,
      "loss": 0.1877,
      "step": 7556
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5163694638643972,
      "learning_rate": 2.302687320884876e-06,
      "loss": 0.2118,
      "step": 7557
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.332473900618379,
      "learning_rate": 2.3014406201919653e-06,
      "loss": 0.1687,
      "step": 7558
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.308906930135461,
      "learning_rate": 2.300194156180764e-06,
      "loss": 0.1829,
      "step": 7559
    },
    {
      "epoch": 0.69,
      "grad_norm": 6.108795542651529,
      "learning_rate": 2.298947928960597e-06,
      "loss": 0.5282,
      "step": 7560
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2516518441864397,
      "learning_rate": 2.2977019386407653e-06,
      "loss": 0.1499,
      "step": 7561
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3829795532411147,
      "learning_rate": 2.2964561853305522e-06,
      "loss": 0.2072,
      "step": 7562
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4785371304707309,
      "learning_rate": 2.295210669139218e-06,
      "loss": 0.205,
      "step": 7563
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3852717107116472,
      "learning_rate": 2.2939653901760007e-06,
      "loss": 0.1587,
      "step": 7564
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5591403598561147,
      "learning_rate": 2.292720348550123e-06,
      "loss": 0.2024,
      "step": 7565
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2939002028109803,
      "learning_rate": 2.2914755443707804e-06,
      "loss": 0.1644,
      "step": 7566
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5986280581865664,
      "learning_rate": 2.290230977747153e-06,
      "loss": 0.2174,
      "step": 7567
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4653600644156022,
      "learning_rate": 2.2889866487883957e-06,
      "loss": 0.2293,
      "step": 7568
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4029845151372242,
      "learning_rate": 2.2877425576036467e-06,
      "loss": 0.1907,
      "step": 7569
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2241650884919149,
      "learning_rate": 2.2864987043020176e-06,
      "loss": 0.1888,
      "step": 7570
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3389812950153221,
      "learning_rate": 2.2852550889926067e-06,
      "loss": 0.1938,
      "step": 7571
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1813601240357121,
      "learning_rate": 2.284011711784486e-06,
      "loss": 0.1475,
      "step": 7572
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3758482509621264,
      "learning_rate": 2.2827685727867055e-06,
      "loss": 0.1465,
      "step": 7573
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3673971549299075,
      "learning_rate": 2.2815256721083e-06,
      "loss": 0.1698,
      "step": 7574
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2949955612024904,
      "learning_rate": 2.280283009858278e-06,
      "loss": 0.1792,
      "step": 7575
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1916307226044471,
      "learning_rate": 2.279040586145632e-06,
      "loss": 0.1451,
      "step": 7576
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7685616903443802,
      "learning_rate": 2.2777984010793264e-06,
      "loss": 0.2027,
      "step": 7577
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4081017490390038,
      "learning_rate": 2.2765564547683145e-06,
      "loss": 0.1923,
      "step": 7578
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.425225285330366,
      "learning_rate": 2.2753147473215187e-06,
      "loss": 0.1752,
      "step": 7579
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4364059074851423,
      "learning_rate": 2.2740732788478493e-06,
      "loss": 0.181,
      "step": 7580
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4779330368118984,
      "learning_rate": 2.2728320494561884e-06,
      "loss": 0.2021,
      "step": 7581
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.347869137559017,
      "learning_rate": 2.2715910592553998e-06,
      "loss": 0.1717,
      "step": 7582
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2779823772265535,
      "learning_rate": 2.2703503083543288e-06,
      "loss": 0.1527,
      "step": 7583
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3588928106490477,
      "learning_rate": 2.2691097968617946e-06,
      "loss": 0.1729,
      "step": 7584
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3099152129571936,
      "learning_rate": 2.267869524886603e-06,
      "loss": 0.1591,
      "step": 7585
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3937300607835126,
      "learning_rate": 2.2666294925375287e-06,
      "loss": 0.1855,
      "step": 7586
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1528500264918482,
      "learning_rate": 2.2653896999233353e-06,
      "loss": 0.1275,
      "step": 7587
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3269196557296385,
      "learning_rate": 2.2641501471527594e-06,
      "loss": 0.1588,
      "step": 7588
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.298029115302002,
      "learning_rate": 2.262910834334516e-06,
      "loss": 0.1659,
      "step": 7589
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.450087983728294,
      "learning_rate": 2.2616717615773044e-06,
      "loss": 0.2253,
      "step": 7590
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0637837765678992,
      "learning_rate": 2.2604329289897965e-06,
      "loss": 0.1226,
      "step": 7591
    },
    {
      "epoch": 0.69,
      "grad_norm": 9.516285174297181,
      "learning_rate": 2.259194336680649e-06,
      "loss": 0.4924,
      "step": 7592
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.110645716935891,
      "learning_rate": 2.2579559847584924e-06,
      "loss": 0.1471,
      "step": 7593
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4329271318626355,
      "learning_rate": 2.256717873331941e-06,
      "loss": 0.1572,
      "step": 7594
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.161828272635256,
      "learning_rate": 2.2554800025095826e-06,
      "loss": 0.1461,
      "step": 7595
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5013746751992998,
      "learning_rate": 2.2542423723999896e-06,
      "loss": 0.1819,
      "step": 7596
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.342561918751822,
      "learning_rate": 2.253004983111709e-06,
      "loss": 0.2044,
      "step": 7597
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3655635235719628,
      "learning_rate": 2.251767834753266e-06,
      "loss": 0.1574,
      "step": 7598
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6053685411632361,
      "learning_rate": 2.2505309274331717e-06,
      "loss": 0.1784,
      "step": 7599
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3261121530971527,
      "learning_rate": 2.2492942612599058e-06,
      "loss": 0.1442,
      "step": 7600
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2382073488246248,
      "learning_rate": 2.2480578363419363e-06,
      "loss": 0.1574,
      "step": 7601
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3997465078390088,
      "learning_rate": 2.246821652787703e-06,
      "loss": 0.167,
      "step": 7602
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.338598789706711,
      "learning_rate": 2.2455857107056307e-06,
      "loss": 0.1943,
      "step": 7603
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.448151548460313,
      "learning_rate": 2.244350010204116e-06,
      "loss": 0.5688,
      "step": 7604
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2552471801449914,
      "learning_rate": 2.243114551391542e-06,
      "loss": 0.1594,
      "step": 7605
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5484111016970958,
      "learning_rate": 2.2418793343762646e-06,
      "loss": 0.2229,
      "step": 7606
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4335033308974026,
      "learning_rate": 2.2406443592666187e-06,
      "loss": 0.158,
      "step": 7607
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4765844607732677,
      "learning_rate": 2.239409626170924e-06,
      "loss": 0.2021,
      "step": 7608
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2774042794204896,
      "learning_rate": 2.238175135197471e-06,
      "loss": 0.1397,
      "step": 7609
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.177132011540004,
      "learning_rate": 2.236940886454536e-06,
      "loss": 0.7608,
      "step": 7610
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4188266661305802,
      "learning_rate": 2.2357068800503684e-06,
      "loss": 0.2083,
      "step": 7611
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1630900213569952,
      "learning_rate": 2.234473116093201e-06,
      "loss": 0.1627,
      "step": 7612
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4273030024911755,
      "learning_rate": 2.2332395946912406e-06,
      "loss": 0.2106,
      "step": 7613
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1954113784100018,
      "learning_rate": 2.2320063159526784e-06,
      "loss": 0.125,
      "step": 7614
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5416626033625351,
      "learning_rate": 2.230773279985679e-06,
      "loss": 0.2097,
      "step": 7615
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2481572670921328,
      "learning_rate": 2.2295404868983874e-06,
      "loss": 0.1463,
      "step": 7616
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.442161661913989,
      "learning_rate": 2.2283079367989303e-06,
      "loss": 0.2182,
      "step": 7617
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.308101571833179,
      "learning_rate": 2.227075629795407e-06,
      "loss": 0.1891,
      "step": 7618
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5800247541179515,
      "learning_rate": 2.225843565995903e-06,
      "loss": 0.1671,
      "step": 7619
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3641904465320598,
      "learning_rate": 2.2246117455084756e-06,
      "loss": 0.1652,
      "step": 7620
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1842052340853282,
      "learning_rate": 2.2233801684411655e-06,
      "loss": 0.1674,
      "step": 7621
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4931441977246371,
      "learning_rate": 2.2221488349019903e-06,
      "loss": 0.2015,
      "step": 7622
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3417479801592807,
      "learning_rate": 2.2209177449989434e-06,
      "loss": 0.1209,
      "step": 7623
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2293764899838142,
      "learning_rate": 2.2196868988400032e-06,
      "loss": 0.1531,
      "step": 7624
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2303252909745914,
      "learning_rate": 2.2184562965331203e-06,
      "loss": 0.1514,
      "step": 7625
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2954177751441196,
      "learning_rate": 2.21722593818623e-06,
      "loss": 0.1687,
      "step": 7626
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4298841866137957,
      "learning_rate": 2.2159958239072386e-06,
      "loss": 0.1804,
      "step": 7627
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2950312912286737,
      "learning_rate": 2.21476595380404e-06,
      "loss": 0.1826,
      "step": 7628
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.409784014407426,
      "learning_rate": 2.2135363279844973e-06,
      "loss": 0.2093,
      "step": 7629
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2466135843864246,
      "learning_rate": 2.2123069465564617e-06,
      "loss": 0.1645,
      "step": 7630
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3795231365244331,
      "learning_rate": 2.211077809627756e-06,
      "loss": 0.1902,
      "step": 7631
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3141634500951256,
      "learning_rate": 2.2098489173061815e-06,
      "loss": 0.1717,
      "step": 7632
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7739959382265713,
      "learning_rate": 2.2086202696995248e-06,
      "loss": 0.2237,
      "step": 7633
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3189162597410593,
      "learning_rate": 2.2073918669155416e-06,
      "loss": 0.1856,
      "step": 7634
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3494792265503377,
      "learning_rate": 2.206163709061976e-06,
      "loss": 0.1517,
      "step": 7635
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3445933515601245,
      "learning_rate": 2.2049357962465405e-06,
      "loss": 0.1657,
      "step": 7636
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1790906567734207,
      "learning_rate": 2.203708128576936e-06,
      "loss": 0.1307,
      "step": 7637
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.456826386850268,
      "learning_rate": 2.202480706160834e-06,
      "loss": 0.2159,
      "step": 7638
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5126123526235047,
      "learning_rate": 2.20125352910589e-06,
      "loss": 0.1849,
      "step": 7639
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.8205363293275285,
      "learning_rate": 2.200026597519735e-06,
      "loss": 0.5442,
      "step": 7640
    },
    {
      "epoch": 0.7,
      "grad_norm": 6.460914879941328,
      "learning_rate": 2.1987999115099763e-06,
      "loss": 0.4462,
      "step": 7641
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.449458141915972,
      "learning_rate": 2.1975734711842067e-06,
      "loss": 0.1868,
      "step": 7642
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2648012014057954,
      "learning_rate": 2.1963472766499894e-06,
      "loss": 0.1731,
      "step": 7643
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3547951965554268,
      "learning_rate": 2.1951213280148733e-06,
      "loss": 0.159,
      "step": 7644
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.807011123673549,
      "learning_rate": 2.1938956253863787e-06,
      "loss": 0.7202,
      "step": 7645
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5150930003475287,
      "learning_rate": 2.192670168872012e-06,
      "loss": 0.1821,
      "step": 7646
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4853066581246563,
      "learning_rate": 2.1914449585792504e-06,
      "loss": 0.173,
      "step": 7647
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4927292968657377,
      "learning_rate": 2.1902199946155555e-06,
      "loss": 0.2257,
      "step": 7648
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5286439202318671,
      "learning_rate": 2.1889952770883644e-06,
      "loss": 0.1614,
      "step": 7649
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6062950372338485,
      "learning_rate": 2.1877708061050908e-06,
      "loss": 0.2229,
      "step": 7650
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2737266476557827,
      "learning_rate": 2.186546581773132e-06,
      "loss": 0.1799,
      "step": 7651
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.501540701211405,
      "learning_rate": 2.1853226041998577e-06,
      "loss": 0.1784,
      "step": 7652
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2179144215315951,
      "learning_rate": 2.184098873492622e-06,
      "loss": 0.1678,
      "step": 7653
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.5403884094157458,
      "learning_rate": 2.1828753897587512e-06,
      "loss": 0.1971,
      "step": 7654
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4319870799801213,
      "learning_rate": 2.1816521531055566e-06,
      "loss": 0.1885,
      "step": 7655
    },
    {
      "epoch": 0.7,
      "grad_norm": 8.321208458428757,
      "learning_rate": 2.1804291636403204e-06,
      "loss": 0.6692,
      "step": 7656
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.976304029126993,
      "learning_rate": 2.17920642147031e-06,
      "loss": 0.6904,
      "step": 7657
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4760793435524517,
      "learning_rate": 2.1779839267027676e-06,
      "loss": 0.1788,
      "step": 7658
    },
    {
      "epoch": 0.7,
      "grad_norm": 9.121395314372515,
      "learning_rate": 2.1767616794449113e-06,
      "loss": 0.606,
      "step": 7659
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.333704762869364,
      "learning_rate": 2.1755396798039446e-06,
      "loss": 0.1784,
      "step": 7660
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2702187726580514,
      "learning_rate": 2.174317927887041e-06,
      "loss": 0.1845,
      "step": 7661
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.241173570997097,
      "learning_rate": 2.1730964238013595e-06,
      "loss": 0.1781,
      "step": 7662
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.491214200360045,
      "learning_rate": 2.1718751676540315e-06,
      "loss": 0.1901,
      "step": 7663
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1985410247250259,
      "learning_rate": 2.170654159552173e-06,
      "loss": 0.1824,
      "step": 7664
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2153882814629107,
      "learning_rate": 2.169433399602872e-06,
      "loss": 0.1656,
      "step": 7665
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3791192288685943,
      "learning_rate": 2.1682128879131956e-06,
      "loss": 0.2117,
      "step": 7666
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3359009540242435,
      "learning_rate": 2.1669926245901953e-06,
      "loss": 0.1741,
      "step": 7667
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2004400727446267,
      "learning_rate": 2.1657726097408914e-06,
      "loss": 0.1774,
      "step": 7668
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.543577464154628,
      "learning_rate": 2.164552843472292e-06,
      "loss": 0.552,
      "step": 7669
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1508667943974427,
      "learning_rate": 2.1633333258913753e-06,
      "loss": 0.1588,
      "step": 7670
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2844890561404696,
      "learning_rate": 2.162114057105104e-06,
      "loss": 0.1765,
      "step": 7671
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.870382795521232,
      "learning_rate": 2.160895037220413e-06,
      "loss": 0.6602,
      "step": 7672
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.59745579059482,
      "learning_rate": 2.159676266344222e-06,
      "loss": 0.179,
      "step": 7673
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2533817309631143,
      "learning_rate": 2.1584577445834234e-06,
      "loss": 0.1485,
      "step": 7674
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.105855312085888,
      "learning_rate": 2.1572394720448876e-06,
      "loss": 0.6101,
      "step": 7675
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4011164533718978,
      "learning_rate": 2.1560214488354697e-06,
      "loss": 0.1749,
      "step": 7676
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.961108194318026,
      "learning_rate": 2.1548036750619943e-06,
      "loss": 0.5779,
      "step": 7677
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.362702559052844,
      "learning_rate": 2.1535861508312726e-06,
      "loss": 0.1999,
      "step": 7678
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4788263809864688,
      "learning_rate": 2.152368876250085e-06,
      "loss": 0.2136,
      "step": 7679
    },
    {
      "epoch": 0.7,
      "grad_norm": 5.279899467831149,
      "learning_rate": 2.151151851425199e-06,
      "loss": 0.5225,
      "step": 7680
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.269626606885254,
      "learning_rate": 2.1499350764633513e-06,
      "loss": 0.1565,
      "step": 7681
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3778655027075843,
      "learning_rate": 2.148718551471266e-06,
      "loss": 0.1838,
      "step": 7682
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2448148264656504,
      "learning_rate": 2.147502276555638e-06,
      "loss": 0.1427,
      "step": 7683
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2244459081631924,
      "learning_rate": 2.1462862518231414e-06,
      "loss": 0.1738,
      "step": 7684
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3395049140417858,
      "learning_rate": 2.1450704773804325e-06,
      "loss": 0.1693,
      "step": 7685
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1889716087945494,
      "learning_rate": 2.1438549533341396e-06,
      "loss": 0.1603,
      "step": 7686
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3747637565653121,
      "learning_rate": 2.1426396797908764e-06,
      "loss": 0.1499,
      "step": 7687
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.194077419426404,
      "learning_rate": 2.1414246568572266e-06,
      "loss": 0.5965,
      "step": 7688
    },
    {
      "epoch": 0.7,
      "grad_norm": 4.662409539556531,
      "learning_rate": 2.140209884639759e-06,
      "loss": 0.441,
      "step": 7689
    },
    {
      "epoch": 0.7,
      "grad_norm": 10.005134406046874,
      "learning_rate": 2.1389953632450144e-06,
      "loss": 0.6383,
      "step": 7690
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.252721956684366,
      "learning_rate": 2.1377810927795173e-06,
      "loss": 0.1441,
      "step": 7691
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4983504211858674,
      "learning_rate": 2.136567073349766e-06,
      "loss": 0.2128,
      "step": 7692
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.385311254682837,
      "learning_rate": 2.1353533050622364e-06,
      "loss": 0.1657,
      "step": 7693
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6276424965353116,
      "learning_rate": 2.1341397880233874e-06,
      "loss": 0.1915,
      "step": 7694
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3996160922588143,
      "learning_rate": 2.132926522339649e-06,
      "loss": 0.1725,
      "step": 7695
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4511135418401413,
      "learning_rate": 2.1317135081174366e-06,
      "loss": 0.1656,
      "step": 7696
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4966242192674482,
      "learning_rate": 2.130500745463136e-06,
      "loss": 0.1954,
      "step": 7697
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2931541922763172,
      "learning_rate": 2.129288234483118e-06,
      "loss": 0.1534,
      "step": 7698
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.7544438730875596,
      "learning_rate": 2.1280759752837245e-06,
      "loss": 0.2023,
      "step": 7699
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4076944410978292,
      "learning_rate": 2.1268639679712814e-06,
      "loss": 0.2146,
      "step": 7700
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3166382194454784,
      "learning_rate": 2.12565221265209e-06,
      "loss": 0.1832,
      "step": 7701
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4852000985019649,
      "learning_rate": 2.124440709432426e-06,
      "loss": 0.2032,
      "step": 7702
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4425741775421885,
      "learning_rate": 2.1232294584185505e-06,
      "loss": 0.2288,
      "step": 7703
    },
    {
      "epoch": 0.7,
      "grad_norm": 7.010903856742783,
      "learning_rate": 2.1220184597166944e-06,
      "loss": 0.5349,
      "step": 7704
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6292979521825166,
      "learning_rate": 2.120807713433074e-06,
      "loss": 0.2231,
      "step": 7705
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.418537103796292,
      "learning_rate": 2.1195972196738767e-06,
      "loss": 0.1509,
      "step": 7706
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.384412394222819,
      "learning_rate": 2.1183869785452744e-06,
      "loss": 0.1829,
      "step": 7707
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4157515314627018,
      "learning_rate": 2.1171769901534112e-06,
      "loss": 0.1971,
      "step": 7708
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2827057763883034,
      "learning_rate": 2.115967254604409e-06,
      "loss": 0.1566,
      "step": 7709
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4905055831559193,
      "learning_rate": 2.114757772004375e-06,
      "loss": 0.1874,
      "step": 7710
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.149417329270534,
      "learning_rate": 2.113548542459383e-06,
      "loss": 0.4936,
      "step": 7711
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4156734686402999,
      "learning_rate": 2.112339566075496e-06,
      "loss": 0.1861,
      "step": 7712
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.375646832305071,
      "learning_rate": 2.1111308429587446e-06,
      "loss": 0.2046,
      "step": 7713
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5151772863352289,
      "learning_rate": 2.1099223732151464e-06,
      "loss": 0.1694,
      "step": 7714
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4310781687665237,
      "learning_rate": 2.108714156950688e-06,
      "loss": 0.1918,
      "step": 7715
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4730548292129675,
      "learning_rate": 2.1075061942713415e-06,
      "loss": 0.1781,
      "step": 7716
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3602852974469173,
      "learning_rate": 2.1062984852830525e-06,
      "loss": 0.203,
      "step": 7717
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4026920879272735,
      "learning_rate": 2.105091030091742e-06,
      "loss": 0.1806,
      "step": 7718
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6570364938399564,
      "learning_rate": 2.1038838288033164e-06,
      "loss": 0.2466,
      "step": 7719
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.374122670895493,
      "learning_rate": 2.102676881523652e-06,
      "loss": 0.1678,
      "step": 7720
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2247531362615136,
      "learning_rate": 2.1014701883586087e-06,
      "loss": 0.1578,
      "step": 7721
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2943333266836599,
      "learning_rate": 2.100263749414019e-06,
      "loss": 0.1239,
      "step": 7722
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2885154356021227,
      "learning_rate": 2.0990575647956983e-06,
      "loss": 0.16,
      "step": 7723
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4495723470599091,
      "learning_rate": 2.0978516346094342e-06,
      "loss": 0.1828,
      "step": 7724
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3265394034366964,
      "learning_rate": 2.0966459589609983e-06,
      "loss": 0.1779,
      "step": 7725
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5591521953020087,
      "learning_rate": 2.0954405379561345e-06,
      "loss": 0.2045,
      "step": 7726
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4241044669027136,
      "learning_rate": 2.0942353717005647e-06,
      "loss": 0.1356,
      "step": 7727
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3762269649764223,
      "learning_rate": 2.093030460299993e-06,
      "loss": 0.1699,
      "step": 7728
    },
    {
      "epoch": 0.71,
      "grad_norm": 11.891131346839355,
      "learning_rate": 2.091825803860095e-06,
      "loss": 0.4887,
      "step": 7729
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3405857288720173,
      "learning_rate": 2.0906214024865318e-06,
      "loss": 0.1693,
      "step": 7730
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4058671525796644,
      "learning_rate": 2.089417256284932e-06,
      "loss": 0.1745,
      "step": 7731
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3351999538610926,
      "learning_rate": 2.0882133653609117e-06,
      "loss": 0.1578,
      "step": 7732
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.375247455408831,
      "learning_rate": 2.087009729820057e-06,
      "loss": 0.1785,
      "step": 7733
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.9792887019464604,
      "learning_rate": 2.0858063497679375e-06,
      "loss": 0.6376,
      "step": 7734
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.320243148346544,
      "learning_rate": 2.0846032253100967e-06,
      "loss": 0.1596,
      "step": 7735
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2824987292657395,
      "learning_rate": 2.0834003565520544e-06,
      "loss": 0.1851,
      "step": 7736
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3972217934860125,
      "learning_rate": 2.082197743599314e-06,
      "loss": 0.19,
      "step": 7737
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4022053737386948,
      "learning_rate": 2.080995386557349e-06,
      "loss": 0.1828,
      "step": 7738
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.479767398698103,
      "learning_rate": 2.0797932855316183e-06,
      "loss": 0.5284,
      "step": 7739
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3094920858201151,
      "learning_rate": 2.07859144062755e-06,
      "loss": 0.1832,
      "step": 7740
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4810753465033573,
      "learning_rate": 2.077389851950557e-06,
      "loss": 0.195,
      "step": 7741
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4097575809801361,
      "learning_rate": 2.0761885196060242e-06,
      "loss": 0.1837,
      "step": 7742
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2279367469510962,
      "learning_rate": 2.0749874436993196e-06,
      "loss": 0.2125,
      "step": 7743
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.8753601871329533,
      "learning_rate": 2.073786624335784e-06,
      "loss": 0.1918,
      "step": 7744
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.4601725958872604,
      "learning_rate": 2.072586061620735e-06,
      "loss": 0.4902,
      "step": 7745
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2712226330204115,
      "learning_rate": 2.071385755659474e-06,
      "loss": 0.1626,
      "step": 7746
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3008296838781999,
      "learning_rate": 2.0701857065572724e-06,
      "loss": 0.2015,
      "step": 7747
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3340408786198183,
      "learning_rate": 2.0689859144193854e-06,
      "loss": 0.1926,
      "step": 7748
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.486726869625602,
      "learning_rate": 2.06778637935104e-06,
      "loss": 0.1565,
      "step": 7749
    },
    {
      "epoch": 0.71,
      "grad_norm": 9.676771822648707,
      "learning_rate": 2.0665871014574464e-06,
      "loss": 0.6454,
      "step": 7750
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3599899180679353,
      "learning_rate": 2.065388080843788e-06,
      "loss": 0.17,
      "step": 7751
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4674253938374808,
      "learning_rate": 2.064189317615225e-06,
      "loss": 0.1609,
      "step": 7752
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5141802106414508,
      "learning_rate": 2.0629908118769004e-06,
      "loss": 0.2059,
      "step": 7753
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2890411641630009,
      "learning_rate": 2.061792563733928e-06,
      "loss": 0.1393,
      "step": 7754
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3864342999387966,
      "learning_rate": 2.060594573291405e-06,
      "loss": 0.1795,
      "step": 7755
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5752263910987725,
      "learning_rate": 2.059396840654401e-06,
      "loss": 0.1697,
      "step": 7756
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2919161863519588,
      "learning_rate": 2.058199365927967e-06,
      "loss": 0.1855,
      "step": 7757
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4921146452483123,
      "learning_rate": 2.0570021492171268e-06,
      "loss": 0.1929,
      "step": 7758
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5135458528322527,
      "learning_rate": 2.0558051906268877e-06,
      "loss": 0.1806,
      "step": 7759
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.408220406027463,
      "learning_rate": 2.0546084902622294e-06,
      "loss": 0.1664,
      "step": 7760
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4679340874719347,
      "learning_rate": 2.0534120482281087e-06,
      "loss": 0.1964,
      "step": 7761
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.353073675024125,
      "learning_rate": 2.0522158646294653e-06,
      "loss": 0.1903,
      "step": 7762
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2995287039903722,
      "learning_rate": 2.051019939571209e-06,
      "loss": 0.1813,
      "step": 7763
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.068909964294098,
      "learning_rate": 2.049824273158233e-06,
      "loss": 0.7226,
      "step": 7764
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.515479828083029,
      "learning_rate": 2.048628865495403e-06,
      "loss": 0.2204,
      "step": 7765
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.641056834284202,
      "learning_rate": 2.0474337166875667e-06,
      "loss": 0.185,
      "step": 7766
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.301711182745952,
      "learning_rate": 2.046238826839544e-06,
      "loss": 0.1399,
      "step": 7767
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.204170403688399,
      "learning_rate": 2.045044196056138e-06,
      "loss": 0.192,
      "step": 7768
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.967484620665515,
      "learning_rate": 2.043849824442124e-06,
      "loss": 0.6414,
      "step": 7769
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2992311458487293,
      "learning_rate": 2.0426557121022545e-06,
      "loss": 0.1592,
      "step": 7770
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4917170135923294,
      "learning_rate": 2.0414618591412644e-06,
      "loss": 0.1801,
      "step": 7771
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3673651523942734,
      "learning_rate": 2.0402682656638597e-06,
      "loss": 0.2023,
      "step": 7772
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4536795413548311,
      "learning_rate": 2.0390749317747293e-06,
      "loss": 0.2085,
      "step": 7773
    },
    {
      "epoch": 0.71,
      "grad_norm": 7.91979019656981,
      "learning_rate": 2.0378818575785342e-06,
      "loss": 0.6426,
      "step": 7774
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.28639074427011,
      "learning_rate": 2.036689043179917e-06,
      "loss": 0.1834,
      "step": 7775
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4322681056812303,
      "learning_rate": 2.035496488683493e-06,
      "loss": 0.2178,
      "step": 7776
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.532923707937776,
      "learning_rate": 2.034304194193861e-06,
      "loss": 0.4434,
      "step": 7777
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3548045110873457,
      "learning_rate": 2.0331121598155905e-06,
      "loss": 0.1845,
      "step": 7778
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5672242877910998,
      "learning_rate": 2.03192038565323e-06,
      "loss": 0.1873,
      "step": 7779
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4318321149018793,
      "learning_rate": 2.030728871811309e-06,
      "loss": 0.1864,
      "step": 7780
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3655637038419537,
      "learning_rate": 2.029537618394328e-06,
      "loss": 0.1937,
      "step": 7781
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5570158256968096,
      "learning_rate": 2.0283466255067707e-06,
      "loss": 0.177,
      "step": 7782
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2266797297832706,
      "learning_rate": 2.027155893253093e-06,
      "loss": 0.1788,
      "step": 7783
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2476620862632457,
      "learning_rate": 2.0259654217377322e-06,
      "loss": 0.1799,
      "step": 7784
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3143902446919047,
      "learning_rate": 2.024775211065098e-06,
      "loss": 0.175,
      "step": 7785
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4737677777944806,
      "learning_rate": 2.0235852613395833e-06,
      "loss": 0.1988,
      "step": 7786
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2592433662505251,
      "learning_rate": 2.0223955726655527e-06,
      "loss": 0.1492,
      "step": 7787
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3586087949724217,
      "learning_rate": 2.021206145147346e-06,
      "loss": 0.2078,
      "step": 7788
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.150546798895481,
      "learning_rate": 2.020016978889291e-06,
      "loss": 0.1669,
      "step": 7789
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3291323365267693,
      "learning_rate": 2.018828073995681e-06,
      "loss": 0.1645,
      "step": 7790
    },
    {
      "epoch": 0.71,
      "grad_norm": 5.467035389685612,
      "learning_rate": 2.017639430570794e-06,
      "loss": 0.5012,
      "step": 7791
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6520900600725332,
      "learning_rate": 2.01645104871888e-06,
      "loss": 0.2242,
      "step": 7792
    },
    {
      "epoch": 0.71,
      "grad_norm": 8.103149975073643,
      "learning_rate": 2.0152629285441668e-06,
      "loss": 0.4698,
      "step": 7793
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.268901855520419,
      "learning_rate": 2.0140750701508632e-06,
      "loss": 0.1705,
      "step": 7794
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.339755182562323,
      "learning_rate": 2.0128874736431496e-06,
      "loss": 0.1345,
      "step": 7795
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4806889474288318,
      "learning_rate": 2.0117001391251893e-06,
      "loss": 0.2129,
      "step": 7796
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3898589828134291,
      "learning_rate": 2.0105130667011165e-06,
      "loss": 0.1657,
      "step": 7797
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2139776521343757,
      "learning_rate": 2.0093262564750478e-06,
      "loss": 0.1403,
      "step": 7798
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5755058407147458,
      "learning_rate": 2.0081397085510736e-06,
      "loss": 0.2336,
      "step": 7799
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5012282843122984,
      "learning_rate": 2.0069534230332597e-06,
      "loss": 0.1761,
      "step": 7800
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.480247190518003,
      "learning_rate": 2.0057674000256556e-06,
      "loss": 0.1855,
      "step": 7801
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3095823571077398,
      "learning_rate": 2.0045816396322794e-06,
      "loss": 0.1631,
      "step": 7802
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2072459570722747,
      "learning_rate": 2.0033961419571333e-06,
      "loss": 0.1625,
      "step": 7803
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.299245467632577,
      "learning_rate": 2.0022109071041905e-06,
      "loss": 0.1753,
      "step": 7804
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3033740645178482,
      "learning_rate": 2.0010259351774082e-06,
      "loss": 0.189,
      "step": 7805
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4099698334139397,
      "learning_rate": 1.999841226280711e-06,
      "loss": 0.1932,
      "step": 7806
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2427073887186932,
      "learning_rate": 1.9986567805180106e-06,
      "loss": 0.1606,
      "step": 7807
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3471853267137317,
      "learning_rate": 1.9974725979931893e-06,
      "loss": 0.1956,
      "step": 7808
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3686429480193976,
      "learning_rate": 1.996288678810105e-06,
      "loss": 0.1939,
      "step": 7809
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.369629151416732,
      "learning_rate": 1.9951050230726e-06,
      "loss": 0.1621,
      "step": 7810
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2295527639618309,
      "learning_rate": 1.9939216308844845e-06,
      "loss": 0.1979,
      "step": 7811
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.299622189706434,
      "learning_rate": 1.9927385023495532e-06,
      "loss": 0.1806,
      "step": 7812
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.2594756851339355,
      "learning_rate": 1.9915556375715715e-06,
      "loss": 0.5563,
      "step": 7813
    },
    {
      "epoch": 0.71,
      "grad_norm": 6.009062415791806,
      "learning_rate": 1.9903730366542885e-06,
      "loss": 0.5398,
      "step": 7814
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3362521178611584,
      "learning_rate": 1.9891906997014214e-06,
      "loss": 0.1427,
      "step": 7815
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.264565161925314,
      "learning_rate": 1.9880086268166736e-06,
      "loss": 0.1764,
      "step": 7816
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.141339347889898,
      "learning_rate": 1.9868268181037186e-06,
      "loss": 0.1661,
      "step": 7817
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3174434195358944,
      "learning_rate": 1.9856452736662074e-06,
      "loss": 0.1744,
      "step": 7818
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4359792129965891,
      "learning_rate": 1.984463993607773e-06,
      "loss": 0.1478,
      "step": 7819
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2741811990805716,
      "learning_rate": 1.9832829780320174e-06,
      "loss": 0.1875,
      "step": 7820
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3725769960869882,
      "learning_rate": 1.9821022270425276e-06,
      "loss": 0.1596,
      "step": 7821
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3129083278950415,
      "learning_rate": 1.98092174074286e-06,
      "loss": 0.17,
      "step": 7822
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5335248265561516,
      "learning_rate": 1.9797415192365537e-06,
      "loss": 0.1966,
      "step": 7823
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3431918939490886,
      "learning_rate": 1.97856156262712e-06,
      "loss": 0.1739,
      "step": 7824
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.718606924711379,
      "learning_rate": 1.9773818710180514e-06,
      "loss": 0.6569,
      "step": 7825
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2754234644895435,
      "learning_rate": 1.976202444512813e-06,
      "loss": 0.1715,
      "step": 7826
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.220707311273098,
      "learning_rate": 1.9750232832148474e-06,
      "loss": 0.503,
      "step": 7827
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1716385813191554,
      "learning_rate": 1.9738443872275777e-06,
      "loss": 0.1235,
      "step": 7828
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.633529864118448,
      "learning_rate": 1.972665756654398e-06,
      "loss": 0.5454,
      "step": 7829
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4398310520248059,
      "learning_rate": 1.9714873915986848e-06,
      "loss": 0.2075,
      "step": 7830
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.417600998754826,
      "learning_rate": 1.9703092921637864e-06,
      "loss": 0.1764,
      "step": 7831
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1519553612694853,
      "learning_rate": 1.969131458453032e-06,
      "loss": 0.1567,
      "step": 7832
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1531790263865747,
      "learning_rate": 1.967953890569723e-06,
      "loss": 0.1672,
      "step": 7833
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3495434763432406,
      "learning_rate": 1.966776588617143e-06,
      "loss": 0.1618,
      "step": 7834
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.378929209404365,
      "learning_rate": 1.9655995526985484e-06,
      "loss": 0.2058,
      "step": 7835
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.065340532178163,
      "learning_rate": 1.9644227829171703e-06,
      "loss": 0.5749,
      "step": 7836
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2671264803357856,
      "learning_rate": 1.963246279376223e-06,
      "loss": 0.171,
      "step": 7837
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4437434599370964,
      "learning_rate": 1.96207004217889e-06,
      "loss": 0.2059,
      "step": 7838
    },
    {
      "epoch": 0.72,
      "grad_norm": 12.828644893835214,
      "learning_rate": 1.96089407142834e-06,
      "loss": 0.4112,
      "step": 7839
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3995823562435314,
      "learning_rate": 1.959718367227708e-06,
      "loss": 0.1894,
      "step": 7840
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.419945817366746,
      "learning_rate": 1.958542929680117e-06,
      "loss": 0.2106,
      "step": 7841
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4017715684651653,
      "learning_rate": 1.9573677588886564e-06,
      "loss": 0.1583,
      "step": 7842
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.349561111308033,
      "learning_rate": 1.956192854956397e-06,
      "loss": 0.1591,
      "step": 7843
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4912402987356839,
      "learning_rate": 1.9550182179863875e-06,
      "loss": 0.1969,
      "step": 7844
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6027359304104365,
      "learning_rate": 1.953843848081649e-06,
      "loss": 0.2084,
      "step": 7845
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3807626125731796,
      "learning_rate": 1.952669745345185e-06,
      "loss": 0.1771,
      "step": 7846
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4121516513814327,
      "learning_rate": 1.9514959098799687e-06,
      "loss": 0.189,
      "step": 7847
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2638172315041878,
      "learning_rate": 1.950322341788956e-06,
      "loss": 0.1561,
      "step": 7848
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5328926321611278,
      "learning_rate": 1.9491490411750745e-06,
      "loss": 0.1652,
      "step": 7849
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3001781284872664,
      "learning_rate": 1.947976008141233e-06,
      "loss": 0.1578,
      "step": 7850
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.708986237944675,
      "learning_rate": 1.946803242790313e-06,
      "loss": 0.5576,
      "step": 7851
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3625610788628788,
      "learning_rate": 1.9456307452251715e-06,
      "loss": 0.1647,
      "step": 7852
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3785335407784514,
      "learning_rate": 1.9444585155486484e-06,
      "loss": 0.1792,
      "step": 7853
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2540089626486042,
      "learning_rate": 1.943286553863553e-06,
      "loss": 0.14,
      "step": 7854
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3054968247721628,
      "learning_rate": 1.9421148602726774e-06,
      "loss": 0.2212,
      "step": 7855
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3174976739000166,
      "learning_rate": 1.9409434348787824e-06,
      "loss": 0.166,
      "step": 7856
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2672199023568815,
      "learning_rate": 1.9397722777846153e-06,
      "loss": 0.1578,
      "step": 7857
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.718015267704107,
      "learning_rate": 1.9386013890928897e-06,
      "loss": 0.5284,
      "step": 7858
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2536324240100853,
      "learning_rate": 1.9374307689063037e-06,
      "loss": 0.1216,
      "step": 7859
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3434112585451383,
      "learning_rate": 1.9362604173275268e-06,
      "loss": 0.1611,
      "step": 7860
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2062048479263887,
      "learning_rate": 1.935090334459206e-06,
      "loss": 0.1612,
      "step": 7861
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3823247739833675,
      "learning_rate": 1.9339205204039675e-06,
      "loss": 0.1554,
      "step": 7862
    },
    {
      "epoch": 0.72,
      "grad_norm": 25.31075682887252,
      "learning_rate": 1.932750975264409e-06,
      "loss": 0.6323,
      "step": 7863
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5139516172822218,
      "learning_rate": 1.9315816991431114e-06,
      "loss": 0.1951,
      "step": 7864
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2068720743054084,
      "learning_rate": 1.9304126921426235e-06,
      "loss": 0.142,
      "step": 7865
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2819624378668175,
      "learning_rate": 1.9292439543654796e-06,
      "loss": 0.1949,
      "step": 7866
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.433231109447931,
      "learning_rate": 1.9280754859141808e-06,
      "loss": 0.1593,
      "step": 7867
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.2635826820160085,
      "learning_rate": 1.9269072868912147e-06,
      "loss": 0.5254,
      "step": 7868
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3050327948130203,
      "learning_rate": 1.925739357399038e-06,
      "loss": 0.1634,
      "step": 7869
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4305769240798873,
      "learning_rate": 1.9245716975400833e-06,
      "loss": 0.2078,
      "step": 7870
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2543158938012569,
      "learning_rate": 1.923404307416767e-06,
      "loss": 0.1524,
      "step": 7871
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1802304229765694,
      "learning_rate": 1.9222371871314725e-06,
      "loss": 0.1476,
      "step": 7872
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2298621767019617,
      "learning_rate": 1.921070336786568e-06,
      "loss": 0.1748,
      "step": 7873
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4679653647373785,
      "learning_rate": 1.9199037564843907e-06,
      "loss": 0.2011,
      "step": 7874
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.939181514472262,
      "learning_rate": 1.9187374463272605e-06,
      "loss": 0.5977,
      "step": 7875
    },
    {
      "epoch": 0.72,
      "grad_norm": 7.597216158596209,
      "learning_rate": 1.9175714064174677e-06,
      "loss": 0.6917,
      "step": 7876
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7932306463007055,
      "learning_rate": 1.9164056368572847e-06,
      "loss": 0.2099,
      "step": 7877
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.862822310374318,
      "learning_rate": 1.915240137748956e-06,
      "loss": 0.4911,
      "step": 7878
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2742735936380756,
      "learning_rate": 1.914074909194702e-06,
      "loss": 0.1727,
      "step": 7879
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4657333613302967,
      "learning_rate": 1.912909951296724e-06,
      "loss": 0.1726,
      "step": 7880
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5362835809502944,
      "learning_rate": 1.9117452641571934e-06,
      "loss": 0.2283,
      "step": 7881
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3403699330259073,
      "learning_rate": 1.9105808478782644e-06,
      "loss": 0.1918,
      "step": 7882
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.464273729694331,
      "learning_rate": 1.9094167025620615e-06,
      "loss": 0.1596,
      "step": 7883
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2455249292488504,
      "learning_rate": 1.90825282831069e-06,
      "loss": 0.1581,
      "step": 7884
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0433771485650285,
      "learning_rate": 1.9070892252262284e-06,
      "loss": 0.1064,
      "step": 7885
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.4787514548025325,
      "learning_rate": 1.9059258934107316e-06,
      "loss": 0.4581,
      "step": 7886
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3766949896372125,
      "learning_rate": 1.9047628329662338e-06,
      "loss": 0.1931,
      "step": 7887
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2872019524610987,
      "learning_rate": 1.90360004399474e-06,
      "loss": 0.1575,
      "step": 7888
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2035598287008482,
      "learning_rate": 1.9024375265982386e-06,
      "loss": 0.1282,
      "step": 7889
    },
    {
      "epoch": 0.72,
      "grad_norm": 9.249139816432221,
      "learning_rate": 1.9012752808786866e-06,
      "loss": 0.5638,
      "step": 7890
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2615252169526074,
      "learning_rate": 1.9001133069380234e-06,
      "loss": 0.1864,
      "step": 7891
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4605240640284292,
      "learning_rate": 1.8989516048781592e-06,
      "loss": 0.1767,
      "step": 7892
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4139117590140267,
      "learning_rate": 1.897790174800986e-06,
      "loss": 0.1982,
      "step": 7893
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.564794908420042,
      "learning_rate": 1.8966290168083678e-06,
      "loss": 0.1908,
      "step": 7894
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5464630294419472,
      "learning_rate": 1.8954681310021434e-06,
      "loss": 0.205,
      "step": 7895
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5903022792585755,
      "learning_rate": 1.8943075174841347e-06,
      "loss": 0.2123,
      "step": 7896
    },
    {
      "epoch": 0.72,
      "grad_norm": 8.508465282680932,
      "learning_rate": 1.893147176356131e-06,
      "loss": 0.4611,
      "step": 7897
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4432229510185575,
      "learning_rate": 1.891987107719906e-06,
      "loss": 0.1883,
      "step": 7898
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2897111649119772,
      "learning_rate": 1.8908273116772018e-06,
      "loss": 0.1512,
      "step": 7899
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4973134710062856,
      "learning_rate": 1.8896677883297437e-06,
      "loss": 0.2155,
      "step": 7900
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.259303222494471,
      "learning_rate": 1.8885085377792262e-06,
      "loss": 0.14,
      "step": 7901
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.530549128991134,
      "learning_rate": 1.8873495601273267e-06,
      "loss": 0.4895,
      "step": 7902
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.494279320262269,
      "learning_rate": 1.8861908554756935e-06,
      "loss": 0.3798,
      "step": 7903
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4394662105997886,
      "learning_rate": 1.885032423925951e-06,
      "loss": 0.1794,
      "step": 7904
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3923974438203006,
      "learning_rate": 1.8838742655797053e-06,
      "loss": 0.2416,
      "step": 7905
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4372693660350342,
      "learning_rate": 1.8827163805385312e-06,
      "loss": 0.1874,
      "step": 7906
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.284843796696772,
      "learning_rate": 1.881558768903986e-06,
      "loss": 0.1731,
      "step": 7907
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5691033249332624,
      "learning_rate": 1.8804014307775965e-06,
      "loss": 0.2032,
      "step": 7908
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3048161756534207,
      "learning_rate": 1.8792443662608727e-06,
      "loss": 0.1466,
      "step": 7909
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4759042626472425,
      "learning_rate": 1.878087575455293e-06,
      "loss": 0.1759,
      "step": 7910
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6094514902681,
      "learning_rate": 1.87693105846232e-06,
      "loss": 0.186,
      "step": 7911
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2360252593116254,
      "learning_rate": 1.8757748153833854e-06,
      "loss": 0.1631,
      "step": 7912
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5300869025684236,
      "learning_rate": 1.8746188463198983e-06,
      "loss": 0.1744,
      "step": 7913
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3223062058497466,
      "learning_rate": 1.8734631513732482e-06,
      "loss": 0.1458,
      "step": 7914
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.574028463337141,
      "learning_rate": 1.8723077306447934e-06,
      "loss": 0.183,
      "step": 7915
    },
    {
      "epoch": 0.72,
      "grad_norm": 5.445561703546853,
      "learning_rate": 1.8711525842358764e-06,
      "loss": 0.5079,
      "step": 7916
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.993236079043797,
      "learning_rate": 1.869997712247807e-06,
      "loss": 0.5913,
      "step": 7917
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2362003429012716,
      "learning_rate": 1.8688431147818787e-06,
      "loss": 0.2086,
      "step": 7918
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.320777044191807,
      "learning_rate": 1.8676887919393543e-06,
      "loss": 0.5255,
      "step": 7919
    },
    {
      "epoch": 0.72,
      "grad_norm": 4.341520737898426,
      "learning_rate": 1.8665347438214788e-06,
      "loss": 0.6038,
      "step": 7920
    },
    {
      "epoch": 0.72,
      "grad_norm": 12.155899312744504,
      "learning_rate": 1.865380970529469e-06,
      "loss": 0.486,
      "step": 7921
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1919076368833046,
      "learning_rate": 1.8642274721645154e-06,
      "loss": 0.1529,
      "step": 7922
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3733751056242045,
      "learning_rate": 1.8630742488277926e-06,
      "loss": 0.1602,
      "step": 7923
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3395392958859207,
      "learning_rate": 1.8619213006204411e-06,
      "loss": 0.1885,
      "step": 7924
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.364040764700063,
      "learning_rate": 1.8607686276435865e-06,
      "loss": 0.1874,
      "step": 7925
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.467614463891958,
      "learning_rate": 1.8596162299983222e-06,
      "loss": 0.2165,
      "step": 7926
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3570728528717335,
      "learning_rate": 1.8584641077857247e-06,
      "loss": 0.1661,
      "step": 7927
    },
    {
      "epoch": 0.72,
      "grad_norm": 6.166872937272724,
      "learning_rate": 1.8573122611068406e-06,
      "loss": 0.5142,
      "step": 7928
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3759409348931144,
      "learning_rate": 1.8561606900626938e-06,
      "loss": 0.1545,
      "step": 7929
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2859246938016549,
      "learning_rate": 1.8550093947542875e-06,
      "loss": 0.1331,
      "step": 7930
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2918423272008772,
      "learning_rate": 1.853858375282595e-06,
      "loss": 0.1752,
      "step": 7931
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.160790136980249,
      "learning_rate": 1.8527076317485714e-06,
      "loss": 0.1579,
      "step": 7932
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.357798724021326,
      "learning_rate": 1.8515571642531415e-06,
      "loss": 0.1829,
      "step": 7933
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.463335337293943,
      "learning_rate": 1.8504069728972124e-06,
      "loss": 0.2069,
      "step": 7934
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2170932620960715,
      "learning_rate": 1.8492570577816604e-06,
      "loss": 0.1713,
      "step": 7935
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3795659891315744,
      "learning_rate": 1.8481074190073434e-06,
      "loss": 0.1791,
      "step": 7936
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.724281492428427,
      "learning_rate": 1.8469580566750911e-06,
      "loss": 0.6128,
      "step": 7937
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1956525628988675,
      "learning_rate": 1.845808970885709e-06,
      "loss": 0.1451,
      "step": 7938
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4348277638454268,
      "learning_rate": 1.8446601617399823e-06,
      "loss": 0.1746,
      "step": 7939
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7334648680216305,
      "learning_rate": 1.8435116293386662e-06,
      "loss": 0.217,
      "step": 7940
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.492325712032056,
      "learning_rate": 1.842363373782498e-06,
      "loss": 0.5714,
      "step": 7941
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2055729860675564,
      "learning_rate": 1.8412153951721845e-06,
      "loss": 0.1751,
      "step": 7942
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.32690086163513,
      "learning_rate": 1.8400676936084134e-06,
      "loss": 0.2164,
      "step": 7943
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.626428214092839,
      "learning_rate": 1.8389202691918435e-06,
      "loss": 0.2275,
      "step": 7944
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.540748054998021,
      "learning_rate": 1.8377731220231144e-06,
      "loss": 0.4285,
      "step": 7945
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4076065956577233,
      "learning_rate": 1.8366262522028367e-06,
      "loss": 0.152,
      "step": 7946
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3597382170440664,
      "learning_rate": 1.8354796598315977e-06,
      "loss": 0.1681,
      "step": 7947
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4195781454652772,
      "learning_rate": 1.8343333450099638e-06,
      "loss": 0.2298,
      "step": 7948
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.303018723985122,
      "learning_rate": 1.8331873078384714e-06,
      "loss": 0.1622,
      "step": 7949
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.41364560856361,
      "learning_rate": 1.832041548417639e-06,
      "loss": 0.1611,
      "step": 7950
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.543959932104808,
      "learning_rate": 1.8308960668479541e-06,
      "loss": 0.6341,
      "step": 7951
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4169253995477524,
      "learning_rate": 1.8297508632298866e-06,
      "loss": 0.1789,
      "step": 7952
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2623457046798137,
      "learning_rate": 1.8286059376638748e-06,
      "loss": 0.1737,
      "step": 7953
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.410715030517687,
      "learning_rate": 1.8274612902503398e-06,
      "loss": 0.1643,
      "step": 7954
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3325062590901315,
      "learning_rate": 1.8263169210896725e-06,
      "loss": 0.1542,
      "step": 7955
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.815720617066585,
      "learning_rate": 1.8251728302822414e-06,
      "loss": 0.359,
      "step": 7956
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0543261035214724,
      "learning_rate": 1.8240290179283938e-06,
      "loss": 0.1209,
      "step": 7957
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2821215981175935,
      "learning_rate": 1.8228854841284454e-06,
      "loss": 0.1866,
      "step": 7958
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4855740479982662,
      "learning_rate": 1.8217422289826964e-06,
      "loss": 0.1829,
      "step": 7959
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3765313255193803,
      "learning_rate": 1.8205992525914135e-06,
      "loss": 0.2047,
      "step": 7960
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4151567453349139,
      "learning_rate": 1.8194565550548477e-06,
      "loss": 0.1743,
      "step": 7961
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4573915310390368,
      "learning_rate": 1.818314136473217e-06,
      "loss": 0.1631,
      "step": 7962
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2379416959351384,
      "learning_rate": 1.8171719969467232e-06,
      "loss": 0.1626,
      "step": 7963
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.266693454390465,
      "learning_rate": 1.8160301365755378e-06,
      "loss": 0.1426,
      "step": 7964
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3938800743810391,
      "learning_rate": 1.8148885554598079e-06,
      "loss": 0.176,
      "step": 7965
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3706052153893828,
      "learning_rate": 1.8137472536996603e-06,
      "loss": 0.1954,
      "step": 7966
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4309318190158296,
      "learning_rate": 1.8126062313951926e-06,
      "loss": 0.1547,
      "step": 7967
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.351481607599842,
      "learning_rate": 1.8114654886464828e-06,
      "loss": 0.1695,
      "step": 7968
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4229775230793993,
      "learning_rate": 1.810325025553578e-06,
      "loss": 0.1588,
      "step": 7969
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3127613620568372,
      "learning_rate": 1.8091848422165093e-06,
      "loss": 0.1718,
      "step": 7970
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3626478303894463,
      "learning_rate": 1.808044938735275e-06,
      "loss": 0.1889,
      "step": 7971
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2213366156048158,
      "learning_rate": 1.8069053152098515e-06,
      "loss": 0.152,
      "step": 7972
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.268046633675456,
      "learning_rate": 1.8057659717401948e-06,
      "loss": 0.1651,
      "step": 7973
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4341944735902505,
      "learning_rate": 1.8046269084262291e-06,
      "loss": 0.2109,
      "step": 7974
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2466020908653501,
      "learning_rate": 1.8034881253678615e-06,
      "loss": 0.1499,
      "step": 7975
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.378018984351007,
      "learning_rate": 1.8023496226649673e-06,
      "loss": 0.216,
      "step": 7976
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3100218669461483,
      "learning_rate": 1.8012114004174048e-06,
      "loss": 0.1619,
      "step": 7977
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.026603794421502,
      "learning_rate": 1.8000734587249996e-06,
      "loss": 0.6437,
      "step": 7978
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.258639661381233,
      "learning_rate": 1.7989357976875603e-06,
      "loss": 0.1521,
      "step": 7979
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.403746146533392,
      "learning_rate": 1.7977984174048663e-06,
      "loss": 0.1963,
      "step": 7980
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4230446549806406,
      "learning_rate": 1.7966613179766712e-06,
      "loss": 0.2084,
      "step": 7981
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.525350657482692,
      "learning_rate": 1.7955244995027099e-06,
      "loss": 0.5404,
      "step": 7982
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.275841588571398,
      "learning_rate": 1.7943879620826853e-06,
      "loss": 0.1621,
      "step": 7983
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4229654519244497,
      "learning_rate": 1.793251705816283e-06,
      "loss": 0.1846,
      "step": 7984
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4254783620193707,
      "learning_rate": 1.7921157308031567e-06,
      "loss": 0.1611,
      "step": 7985
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2970375538309509,
      "learning_rate": 1.7909800371429425e-06,
      "loss": 0.1749,
      "step": 7986
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4316806490910359,
      "learning_rate": 1.7898446249352452e-06,
      "loss": 0.2098,
      "step": 7987
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3907677106310665,
      "learning_rate": 1.7887094942796507e-06,
      "loss": 0.1816,
      "step": 7988
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4156749346938864,
      "learning_rate": 1.7875746452757165e-06,
      "loss": 0.1656,
      "step": 7989
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2847730615425257,
      "learning_rate": 1.7864400780229746e-06,
      "loss": 0.16,
      "step": 7990
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4956845099054594,
      "learning_rate": 1.7853057926209371e-06,
      "loss": 0.1814,
      "step": 7991
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1614502694422673,
      "learning_rate": 1.7841717891690858e-06,
      "loss": 0.1619,
      "step": 7992
    },
    {
      "epoch": 0.73,
      "grad_norm": 8.374063181438306,
      "learning_rate": 1.7830380677668836e-06,
      "loss": 0.662,
      "step": 7993
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.381648686713934,
      "learning_rate": 1.7819046285137614e-06,
      "loss": 0.1916,
      "step": 7994
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.246176568690652,
      "learning_rate": 1.7807714715091334e-06,
      "loss": 0.1527,
      "step": 7995
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4446292002905063,
      "learning_rate": 1.7796385968523817e-06,
      "loss": 0.1743,
      "step": 7996
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.482901743248128,
      "learning_rate": 1.7785060046428698e-06,
      "loss": 0.1796,
      "step": 7997
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6426134964325825,
      "learning_rate": 1.777373694979933e-06,
      "loss": 0.14,
      "step": 7998
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.655407280442881,
      "learning_rate": 1.7762416679628792e-06,
      "loss": 0.4784,
      "step": 7999
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5839574436573074,
      "learning_rate": 1.7751099236909996e-06,
      "loss": 0.1844,
      "step": 8000
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2706122069694017,
      "learning_rate": 1.7739784622635514e-06,
      "loss": 0.1533,
      "step": 8001
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3854102434771134,
      "learning_rate": 1.772847283779775e-06,
      "loss": 0.1613,
      "step": 8002
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4035295571796162,
      "learning_rate": 1.7717163883388793e-06,
      "loss": 0.2001,
      "step": 8003
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.28457059727192,
      "learning_rate": 1.7705857760400542e-06,
      "loss": 0.1559,
      "step": 8004
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2397646041011015,
      "learning_rate": 1.7694554469824592e-06,
      "loss": 0.164,
      "step": 8005
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3059166440599403,
      "learning_rate": 1.7683254012652346e-06,
      "loss": 0.1501,
      "step": 8006
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3017565938234754,
      "learning_rate": 1.7671956389874912e-06,
      "loss": 0.1986,
      "step": 8007
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5133006608127102,
      "learning_rate": 1.7660661602483153e-06,
      "loss": 0.2087,
      "step": 8008
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.104933951684027,
      "learning_rate": 1.764936965146773e-06,
      "loss": 0.1256,
      "step": 8009
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.308898469351845,
      "learning_rate": 1.7638080537818991e-06,
      "loss": 0.1605,
      "step": 8010
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.39491165438738,
      "learning_rate": 1.7626794262527104e-06,
      "loss": 0.156,
      "step": 8011
    },
    {
      "epoch": 0.73,
      "grad_norm": 6.491728553476785,
      "learning_rate": 1.7615510826581906e-06,
      "loss": 0.6047,
      "step": 8012
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.388525269055001,
      "learning_rate": 1.7604230230973068e-06,
      "loss": 0.2236,
      "step": 8013
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3992762022795024,
      "learning_rate": 1.7592952476689962e-06,
      "loss": 0.1879,
      "step": 8014
    },
    {
      "epoch": 0.73,
      "grad_norm": 4.763909424106235,
      "learning_rate": 1.7581677564721706e-06,
      "loss": 0.5465,
      "step": 8015
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.239906432783578,
      "learning_rate": 1.7570405496057208e-06,
      "loss": 0.1796,
      "step": 8016
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.221530274889384,
      "learning_rate": 1.7559136271685079e-06,
      "loss": 0.1298,
      "step": 8017
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.205902265307385,
      "learning_rate": 1.7547869892593738e-06,
      "loss": 0.1432,
      "step": 8018
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4116284513801454,
      "learning_rate": 1.753660635977129e-06,
      "loss": 0.2094,
      "step": 8019
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.251152510498843,
      "learning_rate": 1.752534567420565e-06,
      "loss": 0.1547,
      "step": 8020
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3355535765030377,
      "learning_rate": 1.751408783688443e-06,
      "loss": 0.1477,
      "step": 8021
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3852745282214802,
      "learning_rate": 1.7502832848795043e-06,
      "loss": 0.1757,
      "step": 8022
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.383386844443127,
      "learning_rate": 1.749158071092461e-06,
      "loss": 0.1562,
      "step": 8023
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3647913072753797,
      "learning_rate": 1.748033142426001e-06,
      "loss": 0.1671,
      "step": 8024
    },
    {
      "epoch": 0.73,
      "grad_norm": 5.5853980504646215,
      "learning_rate": 1.746908498978791e-06,
      "loss": 0.66,
      "step": 8025
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4786489795447118,
      "learning_rate": 1.7457841408494658e-06,
      "loss": 0.1896,
      "step": 8026
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2707000664315389,
      "learning_rate": 1.7446600681366438e-06,
      "loss": 0.1522,
      "step": 8027
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3800534150386186,
      "learning_rate": 1.743536280938909e-06,
      "loss": 0.1464,
      "step": 8028
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6677384359887246,
      "learning_rate": 1.742412779354829e-06,
      "loss": 0.2024,
      "step": 8029
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4118515208287996,
      "learning_rate": 1.7412895634829391e-06,
      "loss": 0.1763,
      "step": 8030
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5319806326968362,
      "learning_rate": 1.7401666334217566e-06,
      "loss": 0.1893,
      "step": 8031
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5497020722384147,
      "learning_rate": 1.7390439892697675e-06,
      "loss": 0.1846,
      "step": 8032
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4943001205150277,
      "learning_rate": 1.7379216311254339e-06,
      "loss": 0.2349,
      "step": 8033
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3386795211037488,
      "learning_rate": 1.7367995590871972e-06,
      "loss": 0.1584,
      "step": 8034
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5509129242502255,
      "learning_rate": 1.735677773253468e-06,
      "loss": 0.221,
      "step": 8035
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.473341445033521,
      "learning_rate": 1.7345562737226368e-06,
      "loss": 0.1778,
      "step": 8036
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.34519737510539,
      "learning_rate": 1.7334350605930638e-06,
      "loss": 0.1694,
      "step": 8037
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2899414843393324,
      "learning_rate": 1.7323141339630905e-06,
      "loss": 0.164,
      "step": 8038
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2543740229287137,
      "learning_rate": 1.731193493931026e-06,
      "loss": 0.1744,
      "step": 8039
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2469260903254182,
      "learning_rate": 1.7300731405951609e-06,
      "loss": 0.147,
      "step": 8040
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4112442829796727,
      "learning_rate": 1.7289530740537569e-06,
      "loss": 0.2054,
      "step": 8041
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3811502163280778,
      "learning_rate": 1.7278332944050485e-06,
      "loss": 0.2093,
      "step": 8042
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.530052990715965,
      "learning_rate": 1.7267138017472524e-06,
      "loss": 0.1777,
      "step": 8043
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3059975575199092,
      "learning_rate": 1.7255945961785515e-06,
      "loss": 0.1462,
      "step": 8044
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2301164999485887,
      "learning_rate": 1.724475677797111e-06,
      "loss": 0.182,
      "step": 8045
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4193889410957352,
      "learning_rate": 1.7233570467010645e-06,
      "loss": 0.1985,
      "step": 8046
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.367680087209939,
      "learning_rate": 1.7222387029885268e-06,
      "loss": 0.1757,
      "step": 8047
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2494085742887937,
      "learning_rate": 1.7211206467575802e-06,
      "loss": 0.1961,
      "step": 8048
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2791163311755096,
      "learning_rate": 1.72000287810629e-06,
      "loss": 0.2152,
      "step": 8049
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3325052871274885,
      "learning_rate": 1.7188853971326897e-06,
      "loss": 0.1912,
      "step": 8050
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6099279976328844,
      "learning_rate": 1.7177682039347875e-06,
      "loss": 0.2283,
      "step": 8051
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.68391846083831,
      "learning_rate": 1.7166512986105737e-06,
      "loss": 0.3841,
      "step": 8052
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3302167223435932,
      "learning_rate": 1.7155346812580037e-06,
      "loss": 0.1897,
      "step": 8053
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.735287860299543,
      "learning_rate": 1.7144183519750163e-06,
      "loss": 0.617,
      "step": 8054
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3111348045974638,
      "learning_rate": 1.7133023108595177e-06,
      "loss": 0.1623,
      "step": 8055
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4194475833429285,
      "learning_rate": 1.712186558009395e-06,
      "loss": 0.2142,
      "step": 8056
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.472819764369202,
      "learning_rate": 1.7110710935225055e-06,
      "loss": 0.1948,
      "step": 8057
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3818161483692772,
      "learning_rate": 1.7099559174966823e-06,
      "loss": 0.1593,
      "step": 8058
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.746431106182746,
      "learning_rate": 1.708841030029736e-06,
      "loss": 0.5329,
      "step": 8059
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3778795809049542,
      "learning_rate": 1.7077264312194464e-06,
      "loss": 0.1693,
      "step": 8060
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4796892295392565,
      "learning_rate": 1.7066121211635755e-06,
      "loss": 0.1805,
      "step": 8061
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5606362729452756,
      "learning_rate": 1.705498099959852e-06,
      "loss": 0.2109,
      "step": 8062
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.430560881752865,
      "learning_rate": 1.7043843677059857e-06,
      "loss": 0.4995,
      "step": 8063
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1931103546547301,
      "learning_rate": 1.7032709244996559e-06,
      "loss": 0.1734,
      "step": 8064
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.039555276892425,
      "learning_rate": 1.7021577704385218e-06,
      "loss": 0.4385,
      "step": 8065
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.528046160828198,
      "learning_rate": 1.7010449056202132e-06,
      "loss": 0.2115,
      "step": 8066
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3289418972890035,
      "learning_rate": 1.6999323301423337e-06,
      "loss": 0.1832,
      "step": 8067
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5731202299933356,
      "learning_rate": 1.698820044102467e-06,
      "loss": 0.209,
      "step": 8068
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2763514888034106,
      "learning_rate": 1.6977080475981655e-06,
      "loss": 0.1466,
      "step": 8069
    },
    {
      "epoch": 0.74,
      "grad_norm": 21.909272597019616,
      "learning_rate": 1.696596340726961e-06,
      "loss": 0.6292,
      "step": 8070
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5140351207635026,
      "learning_rate": 1.6954849235863547e-06,
      "loss": 0.1861,
      "step": 8071
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.447214096158724,
      "learning_rate": 1.6943737962738289e-06,
      "loss": 0.2117,
      "step": 8072
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.426755605175759,
      "learning_rate": 1.6932629588868332e-06,
      "loss": 0.1835,
      "step": 8073
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.560797652514139,
      "learning_rate": 1.6921524115227988e-06,
      "loss": 0.2104,
      "step": 8074
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3537732751266018,
      "learning_rate": 1.6910421542791272e-06,
      "loss": 0.2025,
      "step": 8075
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3589911352571238,
      "learning_rate": 1.6899321872531926e-06,
      "loss": 0.1881,
      "step": 8076
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7538695986939445,
      "learning_rate": 1.6888225105423505e-06,
      "loss": 0.1927,
      "step": 8077
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.530804108042174,
      "learning_rate": 1.6877131242439232e-06,
      "loss": 0.5078,
      "step": 8078
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4695339341500684,
      "learning_rate": 1.6866040284552155e-06,
      "loss": 0.173,
      "step": 8079
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2610602803205502,
      "learning_rate": 1.6854952232734984e-06,
      "loss": 0.1709,
      "step": 8080
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4715945847165408,
      "learning_rate": 1.6843867087960252e-06,
      "loss": 0.1476,
      "step": 8081
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3516365006562483,
      "learning_rate": 1.6832784851200163e-06,
      "loss": 0.1501,
      "step": 8082
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.431631549002694,
      "learning_rate": 1.6821705523426747e-06,
      "loss": 0.2123,
      "step": 8083
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3567510794403754,
      "learning_rate": 1.6810629105611704e-06,
      "loss": 0.1928,
      "step": 8084
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4287061149586093,
      "learning_rate": 1.6799555598726502e-06,
      "loss": 0.2165,
      "step": 8085
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.810490762006272,
      "learning_rate": 1.6788485003742394e-06,
      "loss": 0.6073,
      "step": 8086
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.564381319669485,
      "learning_rate": 1.6777417321630312e-06,
      "loss": 0.1935,
      "step": 8087
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2878327188342382,
      "learning_rate": 1.6766352553361004e-06,
      "loss": 0.1948,
      "step": 8088
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4765297936028257,
      "learning_rate": 1.6755290699904881e-06,
      "loss": 0.2018,
      "step": 8089
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.502406861637745,
      "learning_rate": 1.6744231762232178e-06,
      "loss": 0.2071,
      "step": 8090
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4088667890155602,
      "learning_rate": 1.6733175741312807e-06,
      "loss": 0.1725,
      "step": 8091
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3920428540736332,
      "learning_rate": 1.6722122638116483e-06,
      "loss": 0.1527,
      "step": 8092
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0405484014711632,
      "learning_rate": 1.671107245361263e-06,
      "loss": 0.1562,
      "step": 8093
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3675785162048404,
      "learning_rate": 1.6700025188770403e-06,
      "loss": 0.1594,
      "step": 8094
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6082606297123356,
      "learning_rate": 1.6688980844558745e-06,
      "loss": 0.1948,
      "step": 8095
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2302014932969643,
      "learning_rate": 1.66779394219463e-06,
      "loss": 0.1286,
      "step": 8096
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5574700569641227,
      "learning_rate": 1.6666900921901497e-06,
      "loss": 0.2415,
      "step": 8097
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.834567460165467,
      "learning_rate": 1.6655865345392458e-06,
      "loss": 0.5361,
      "step": 8098
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2798224161430574,
      "learning_rate": 1.6644832693387104e-06,
      "loss": 0.1909,
      "step": 8099
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4381367789847717,
      "learning_rate": 1.663380296685307e-06,
      "loss": 0.1613,
      "step": 8100
    },
    {
      "epoch": 0.74,
      "grad_norm": 8.904660260106567,
      "learning_rate": 1.6622776166757704e-06,
      "loss": 0.6514,
      "step": 8101
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2708692002735393,
      "learning_rate": 1.6611752294068173e-06,
      "loss": 0.1868,
      "step": 8102
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.365815920996107,
      "learning_rate": 1.6600731349751303e-06,
      "loss": 0.1493,
      "step": 8103
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.566334487082573,
      "learning_rate": 1.6589713334773749e-06,
      "loss": 0.6084,
      "step": 8104
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4591761950888107,
      "learning_rate": 1.6578698250101828e-06,
      "loss": 0.1722,
      "step": 8105
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4172250568544786,
      "learning_rate": 1.6567686096701662e-06,
      "loss": 0.181,
      "step": 8106
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3361220920568573,
      "learning_rate": 1.655667687553907e-06,
      "loss": 0.1756,
      "step": 8107
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5547391779277289,
      "learning_rate": 1.6545670587579655e-06,
      "loss": 0.2577,
      "step": 8108
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4567507010113263,
      "learning_rate": 1.6534667233788726e-06,
      "loss": 0.2189,
      "step": 8109
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.344922083361939,
      "learning_rate": 1.6523666815131345e-06,
      "loss": 0.1978,
      "step": 8110
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3822303100697573,
      "learning_rate": 1.651266933257235e-06,
      "loss": 0.1485,
      "step": 8111
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5093057603177058,
      "learning_rate": 1.650167478707626e-06,
      "loss": 0.1909,
      "step": 8112
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4379886903697654,
      "learning_rate": 1.6490683179607403e-06,
      "loss": 0.1973,
      "step": 8113
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.215901936302743,
      "learning_rate": 1.6479694511129779e-06,
      "loss": 0.4191,
      "step": 8114
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.543456528275588,
      "learning_rate": 1.6468708782607213e-06,
      "loss": 0.1779,
      "step": 8115
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2768877370397003,
      "learning_rate": 1.6457725995003182e-06,
      "loss": 0.137,
      "step": 8116
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.264487661257691,
      "learning_rate": 1.6446746149280984e-06,
      "loss": 0.5113,
      "step": 8117
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3120639315824953,
      "learning_rate": 1.6435769246403622e-06,
      "loss": 0.1808,
      "step": 8118
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.350924744823735,
      "learning_rate": 1.642479528733381e-06,
      "loss": 0.1724,
      "step": 8119
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4574780618027026,
      "learning_rate": 1.6413824273034063e-06,
      "loss": 0.2062,
      "step": 8120
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.469711578961423,
      "learning_rate": 1.6402856204466611e-06,
      "loss": 0.1836,
      "step": 8121
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3731165973950867,
      "learning_rate": 1.6391891082593441e-06,
      "loss": 0.1692,
      "step": 8122
    },
    {
      "epoch": 0.74,
      "grad_norm": 5.198759534124845,
      "learning_rate": 1.6380928908376252e-06,
      "loss": 0.4433,
      "step": 8123
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3854043874682083,
      "learning_rate": 1.6369969682776487e-06,
      "loss": 0.2066,
      "step": 8124
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3863495215476096,
      "learning_rate": 1.6359013406755374e-06,
      "loss": 0.1758,
      "step": 8125
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3893692324703037,
      "learning_rate": 1.634806008127382e-06,
      "loss": 0.1503,
      "step": 8126
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3761196675886975,
      "learning_rate": 1.6337109707292538e-06,
      "loss": 0.1737,
      "step": 8127
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3020708761364908,
      "learning_rate": 1.6326162285771912e-06,
      "loss": 0.1722,
      "step": 8128
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4438208348391521,
      "learning_rate": 1.6315217817672142e-06,
      "loss": 0.1504,
      "step": 8129
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.220194544158918,
      "learning_rate": 1.6304276303953092e-06,
      "loss": 0.1461,
      "step": 8130
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.332887523711864,
      "learning_rate": 1.6293337745574444e-06,
      "loss": 0.199,
      "step": 8131
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.67845585062058,
      "learning_rate": 1.6282402143495568e-06,
      "loss": 0.6937,
      "step": 8132
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.238165353441387,
      "learning_rate": 1.6271469498675568e-06,
      "loss": 0.1545,
      "step": 8133
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.654592611826069,
      "learning_rate": 1.6260539812073344e-06,
      "loss": 0.4176,
      "step": 8134
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2977811961285526,
      "learning_rate": 1.624961308464747e-06,
      "loss": 0.179,
      "step": 8135
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.263806128896416,
      "learning_rate": 1.6238689317356332e-06,
      "loss": 0.155,
      "step": 8136
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.684758149552574,
      "learning_rate": 1.6227768511157976e-06,
      "loss": 0.1811,
      "step": 8137
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6156881939319476,
      "learning_rate": 1.6216850667010264e-06,
      "loss": 0.1823,
      "step": 8138
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4434778112470008,
      "learning_rate": 1.6205935785870751e-06,
      "loss": 0.2161,
      "step": 8139
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.52319314108432,
      "learning_rate": 1.619502386869673e-06,
      "loss": 0.1798,
      "step": 8140
    },
    {
      "epoch": 0.74,
      "grad_norm": 6.50298171108289,
      "learning_rate": 1.6184114916445282e-06,
      "loss": 0.6791,
      "step": 8141
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2912720926626013,
      "learning_rate": 1.6173208930073152e-06,
      "loss": 0.1939,
      "step": 8142
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3143611691238064,
      "learning_rate": 1.6162305910536913e-06,
      "loss": 0.1976,
      "step": 8143
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3105873846977933,
      "learning_rate": 1.615140585879279e-06,
      "loss": 0.1773,
      "step": 8144
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2890906597675367,
      "learning_rate": 1.6140508775796832e-06,
      "loss": 0.154,
      "step": 8145
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2679416088299997,
      "learning_rate": 1.612961466250475e-06,
      "loss": 0.1558,
      "step": 8146
    },
    {
      "epoch": 0.74,
      "grad_norm": 7.343276067108657,
      "learning_rate": 1.6118723519872065e-06,
      "loss": 0.5564,
      "step": 8147
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2414469973128643,
      "learning_rate": 1.6107835348853983e-06,
      "loss": 0.1673,
      "step": 8148
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4789966117450966,
      "learning_rate": 1.6096950150405454e-06,
      "loss": 0.206,
      "step": 8149
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3137083911557035,
      "learning_rate": 1.6086067925481218e-06,
      "loss": 0.1855,
      "step": 8150
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.1341241626610845,
      "learning_rate": 1.607518867503568e-06,
      "loss": 0.5183,
      "step": 8151
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1665943960293694,
      "learning_rate": 1.6064312400023063e-06,
      "loss": 0.1529,
      "step": 8152
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.348141313356375,
      "learning_rate": 1.6053439101397257e-06,
      "loss": 0.1554,
      "step": 8153
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2800624091911132,
      "learning_rate": 1.604256878011195e-06,
      "loss": 0.1732,
      "step": 8154
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1145852022865577,
      "learning_rate": 1.6031701437120512e-06,
      "loss": 0.1445,
      "step": 8155
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.666752418289231,
      "learning_rate": 1.6020837073376112e-06,
      "loss": 0.2024,
      "step": 8156
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.376886397111822,
      "learning_rate": 1.6009975689831614e-06,
      "loss": 0.177,
      "step": 8157
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1116487102959467,
      "learning_rate": 1.5999117287439619e-06,
      "loss": 0.1417,
      "step": 8158
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4570371752764135,
      "learning_rate": 1.598826186715251e-06,
      "loss": 0.2132,
      "step": 8159
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3595991210392784,
      "learning_rate": 1.5977409429922348e-06,
      "loss": 0.1646,
      "step": 8160
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3791259137467744,
      "learning_rate": 1.5966559976701e-06,
      "loss": 0.1744,
      "step": 8161
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4526751011622816,
      "learning_rate": 1.5955713508440002e-06,
      "loss": 0.2139,
      "step": 8162
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4121570835700878,
      "learning_rate": 1.5944870026090692e-06,
      "loss": 0.1597,
      "step": 8163
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3045414253000007,
      "learning_rate": 1.5934029530604084e-06,
      "loss": 0.1712,
      "step": 8164
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2995851543989527,
      "learning_rate": 1.5923192022931e-06,
      "loss": 0.1932,
      "step": 8165
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.381685247683135,
      "learning_rate": 1.5912357504021937e-06,
      "loss": 0.193,
      "step": 8166
    },
    {
      "epoch": 0.75,
      "grad_norm": 8.218548295658577,
      "learning_rate": 1.5901525974827143e-06,
      "loss": 0.5331,
      "step": 8167
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4352665645222542,
      "learning_rate": 1.5890697436296648e-06,
      "loss": 0.1988,
      "step": 8168
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2407277678126918,
      "learning_rate": 1.5879871889380155e-06,
      "loss": 0.1637,
      "step": 8169
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2672482601094512,
      "learning_rate": 1.586904933502717e-06,
      "loss": 0.1652,
      "step": 8170
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3775546560834482,
      "learning_rate": 1.5858229774186866e-06,
      "loss": 0.1726,
      "step": 8171
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1693222035167585,
      "learning_rate": 1.5847413207808231e-06,
      "loss": 0.1406,
      "step": 8172
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.315314245194929,
      "learning_rate": 1.583659963683991e-06,
      "loss": 0.1686,
      "step": 8173
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3253303801647907,
      "learning_rate": 1.5825789062230367e-06,
      "loss": 0.1695,
      "step": 8174
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.201595250532351,
      "learning_rate": 1.5814981484927738e-06,
      "loss": 0.1657,
      "step": 8175
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.28924296593638,
      "learning_rate": 1.580417690587991e-06,
      "loss": 0.1773,
      "step": 8176
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5768588883909784,
      "learning_rate": 1.5793375326034539e-06,
      "loss": 0.2238,
      "step": 8177
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2726823364368163,
      "learning_rate": 1.5782576746338974e-06,
      "loss": 0.1566,
      "step": 8178
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3082039864824753,
      "learning_rate": 1.5771781167740347e-06,
      "loss": 0.1677,
      "step": 8179
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.72544248203779,
      "learning_rate": 1.5760988591185478e-06,
      "loss": 0.5518,
      "step": 8180
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3570779042012517,
      "learning_rate": 1.575019901762097e-06,
      "loss": 0.2191,
      "step": 8181
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1769881437518943,
      "learning_rate": 1.573941244799314e-06,
      "loss": 0.1298,
      "step": 8182
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3980717409726362,
      "learning_rate": 1.572862888324801e-06,
      "loss": 0.1806,
      "step": 8183
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3439994539048916,
      "learning_rate": 1.5717848324331415e-06,
      "loss": 0.1606,
      "step": 8184
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5433889406273988,
      "learning_rate": 1.5707070772188843e-06,
      "loss": 0.1595,
      "step": 8185
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.188224008333354,
      "learning_rate": 1.569629622776559e-06,
      "loss": 0.1416,
      "step": 8186
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.353584109165849,
      "learning_rate": 1.5685524692006627e-06,
      "loss": 0.1972,
      "step": 8187
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5912369481165458,
      "learning_rate": 1.567475616585672e-06,
      "loss": 0.1963,
      "step": 8188
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2457985091642187,
      "learning_rate": 1.5663990650260309e-06,
      "loss": 0.1585,
      "step": 8189
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5235152495592743,
      "learning_rate": 1.5653228146161636e-06,
      "loss": 0.1798,
      "step": 8190
    },
    {
      "epoch": 0.75,
      "grad_norm": 5.9067022767572706,
      "learning_rate": 1.5642468654504621e-06,
      "loss": 0.5259,
      "step": 8191
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5539305805047228,
      "learning_rate": 1.5631712176232939e-06,
      "loss": 0.2567,
      "step": 8192
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4987384156871133,
      "learning_rate": 1.5620958712290023e-06,
      "loss": 0.227,
      "step": 8193
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3140784030872275,
      "learning_rate": 1.5610208263619002e-06,
      "loss": 0.2192,
      "step": 8194
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.398741399380852,
      "learning_rate": 1.5599460831162793e-06,
      "loss": 0.1992,
      "step": 8195
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.41397157287056,
      "learning_rate": 1.5588716415863985e-06,
      "loss": 0.1815,
      "step": 8196
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4795612643219993,
      "learning_rate": 1.5577975018664966e-06,
      "loss": 0.2301,
      "step": 8197
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4062094368828968,
      "learning_rate": 1.55672366405078e-06,
      "loss": 0.2271,
      "step": 8198
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.899902123455237,
      "learning_rate": 1.5556501282334335e-06,
      "loss": 0.5303,
      "step": 8199
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3297827640237065,
      "learning_rate": 1.5545768945086132e-06,
      "loss": 0.1843,
      "step": 8200
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.373972553856761,
      "learning_rate": 1.5535039629704467e-06,
      "loss": 0.1913,
      "step": 8201
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3610966658036854,
      "learning_rate": 1.5524313337130404e-06,
      "loss": 0.1998,
      "step": 8202
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3637448670285075,
      "learning_rate": 1.551359006830468e-06,
      "loss": 0.1327,
      "step": 8203
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.451423868244094,
      "learning_rate": 1.5502869824167826e-06,
      "loss": 0.1581,
      "step": 8204
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6570176698288501,
      "learning_rate": 1.549215260566005e-06,
      "loss": 0.2143,
      "step": 8205
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.663653037318881,
      "learning_rate": 1.548143841372136e-06,
      "loss": 0.1764,
      "step": 8206
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4468863418130746,
      "learning_rate": 1.5470727249291423e-06,
      "loss": 0.1742,
      "step": 8207
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2903273710841876,
      "learning_rate": 1.5460019113309715e-06,
      "loss": 0.1716,
      "step": 8208
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2373722008034553,
      "learning_rate": 1.5449314006715394e-06,
      "loss": 0.1611,
      "step": 8209
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3546831883096078,
      "learning_rate": 1.5438611930447355e-06,
      "loss": 0.1798,
      "step": 8210
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3348269386343645,
      "learning_rate": 1.5427912885444275e-06,
      "loss": 0.1542,
      "step": 8211
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3363186448067785,
      "learning_rate": 1.5417216872644492e-06,
      "loss": 0.1842,
      "step": 8212
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4027176916405084,
      "learning_rate": 1.5406523892986163e-06,
      "loss": 0.1514,
      "step": 8213
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3757050304751268,
      "learning_rate": 1.539583394740709e-06,
      "loss": 0.1674,
      "step": 8214
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3606233985307232,
      "learning_rate": 1.5385147036844889e-06,
      "loss": 0.1669,
      "step": 8215
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0935807124758563,
      "learning_rate": 1.5374463162236846e-06,
      "loss": 0.1653,
      "step": 8216
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.444733769523496,
      "learning_rate": 1.5363782324520033e-06,
      "loss": 0.186,
      "step": 8217
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3197126865134938,
      "learning_rate": 1.5353104524631218e-06,
      "loss": 0.1957,
      "step": 8218
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3253386266518412,
      "learning_rate": 1.53424297635069e-06,
      "loss": 0.2039,
      "step": 8219
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4996928857412204,
      "learning_rate": 1.5331758042083355e-06,
      "loss": 0.2081,
      "step": 8220
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2773786932405495,
      "learning_rate": 1.5321089361296542e-06,
      "loss": 0.1813,
      "step": 8221
    },
    {
      "epoch": 0.75,
      "grad_norm": 7.357524045355727,
      "learning_rate": 1.5310423722082196e-06,
      "loss": 0.5661,
      "step": 8222
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2321929693959666,
      "learning_rate": 1.5299761125375745e-06,
      "loss": 0.1701,
      "step": 8223
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.258767933189854,
      "learning_rate": 1.5289101572112392e-06,
      "loss": 0.1962,
      "step": 8224
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4190037057935283,
      "learning_rate": 1.5278445063227038e-06,
      "loss": 0.1841,
      "step": 8225
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.54933319545894,
      "learning_rate": 1.5267791599654315e-06,
      "loss": 0.1801,
      "step": 8226
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1787294400524515,
      "learning_rate": 1.5257141182328634e-06,
      "loss": 0.1315,
      "step": 8227
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3802161775309734,
      "learning_rate": 1.5246493812184077e-06,
      "loss": 0.1707,
      "step": 8228
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3205943275617378,
      "learning_rate": 1.523584949015452e-06,
      "loss": 0.1729,
      "step": 8229
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4070824483018982,
      "learning_rate": 1.5225208217173514e-06,
      "loss": 0.1647,
      "step": 8230
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.548330608035589,
      "learning_rate": 1.5214569994174394e-06,
      "loss": 0.1845,
      "step": 8231
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3356056697259289,
      "learning_rate": 1.5203934822090178e-06,
      "loss": 0.1572,
      "step": 8232
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.963566842622165,
      "learning_rate": 1.5193302701853674e-06,
      "loss": 0.5252,
      "step": 8233
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3716704819336893,
      "learning_rate": 1.5182673634397365e-06,
      "loss": 0.1436,
      "step": 8234
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2015920602327241,
      "learning_rate": 1.5172047620653486e-06,
      "loss": 0.1684,
      "step": 8235
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2934202221034539,
      "learning_rate": 1.5161424661554035e-06,
      "loss": 0.17,
      "step": 8236
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4543940635280337,
      "learning_rate": 1.515080475803069e-06,
      "loss": 0.2171,
      "step": 8237
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2845029030870814,
      "learning_rate": 1.5140187911014914e-06,
      "loss": 0.1527,
      "step": 8238
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.314338222533717,
      "learning_rate": 1.5129574121437852e-06,
      "loss": 0.1436,
      "step": 8239
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3642860650496658,
      "learning_rate": 1.5118963390230423e-06,
      "loss": 0.1734,
      "step": 8240
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3523881081366487,
      "learning_rate": 1.5108355718323236e-06,
      "loss": 0.1843,
      "step": 8241
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2451955303626367,
      "learning_rate": 1.5097751106646685e-06,
      "loss": 0.1766,
      "step": 8242
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.545297787748457,
      "learning_rate": 1.5087149556130848e-06,
      "loss": 0.1941,
      "step": 8243
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4372992472181678,
      "learning_rate": 1.507655106770554e-06,
      "loss": 0.16,
      "step": 8244
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4789776532955876,
      "learning_rate": 1.506595564230035e-06,
      "loss": 0.202,
      "step": 8245
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2175195124538924,
      "learning_rate": 1.505536328084453e-06,
      "loss": 0.1806,
      "step": 8246
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3663159828998137,
      "learning_rate": 1.5044773984267141e-06,
      "loss": 0.1537,
      "step": 8247
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1739837779759308,
      "learning_rate": 1.5034187753496898e-06,
      "loss": 0.1686,
      "step": 8248
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4748374002200983,
      "learning_rate": 1.502360458946232e-06,
      "loss": 0.2013,
      "step": 8249
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3131197379478923,
      "learning_rate": 1.5013024493091582e-06,
      "loss": 0.1604,
      "step": 8250
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5056762866899314,
      "learning_rate": 1.5002447465312675e-06,
      "loss": 0.1792,
      "step": 8251
    },
    {
      "epoch": 0.75,
      "grad_norm": 6.360483297201026,
      "learning_rate": 1.4991873507053246e-06,
      "loss": 0.7802,
      "step": 8252
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5159469584720247,
      "learning_rate": 1.4981302619240683e-06,
      "loss": 0.1898,
      "step": 8253
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.415685208209731,
      "learning_rate": 1.4970734802802173e-06,
      "loss": 0.1644,
      "step": 8254
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.40978079746201,
      "learning_rate": 1.4960170058664536e-06,
      "loss": 0.1658,
      "step": 8255
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2210776911546255,
      "learning_rate": 1.494960838775441e-06,
      "loss": 0.1518,
      "step": 8256
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2991527478866385,
      "learning_rate": 1.4939049790998095e-06,
      "loss": 0.1814,
      "step": 8257
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3153457463551526,
      "learning_rate": 1.4928494269321674e-06,
      "loss": 0.1757,
      "step": 8258
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.467813101121754,
      "learning_rate": 1.4917941823650917e-06,
      "loss": 0.1598,
      "step": 8259
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3707384597430632,
      "learning_rate": 1.4907392454911363e-06,
      "loss": 0.1523,
      "step": 8260
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4013039039029114,
      "learning_rate": 1.489684616402825e-06,
      "loss": 0.1836,
      "step": 8261
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3552937830574279,
      "learning_rate": 1.4886302951926546e-06,
      "loss": 0.1869,
      "step": 8262
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2032507752693558,
      "learning_rate": 1.4875762819530986e-06,
      "loss": 0.1718,
      "step": 8263
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4667802235456333,
      "learning_rate": 1.486522576776599e-06,
      "loss": 0.157,
      "step": 8264
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.283168944217346,
      "learning_rate": 1.4854691797555753e-06,
      "loss": 0.4563,
      "step": 8265
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.704111014959695,
      "learning_rate": 1.484416090982414e-06,
      "loss": 0.5986,
      "step": 8266
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.028687360320003,
      "learning_rate": 1.4833633105494817e-06,
      "loss": 0.6568,
      "step": 8267
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.276863577162427,
      "learning_rate": 1.4823108385491125e-06,
      "loss": 0.6054,
      "step": 8268
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.497474486430763,
      "learning_rate": 1.4812586750736135e-06,
      "loss": 0.5811,
      "step": 8269
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.498074671451125,
      "learning_rate": 1.4802068202152702e-06,
      "loss": 0.1662,
      "step": 8270
    },
    {
      "epoch": 0.76,
      "grad_norm": 17.227149025844113,
      "learning_rate": 1.4791552740663334e-06,
      "loss": 0.6665,
      "step": 8271
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2563516021313184,
      "learning_rate": 1.478104036719034e-06,
      "loss": 0.1639,
      "step": 8272
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3347464007412582,
      "learning_rate": 1.4770531082655704e-06,
      "loss": 0.177,
      "step": 8273
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1866361663552432,
      "learning_rate": 1.4760024887981177e-06,
      "loss": 0.152,
      "step": 8274
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.611653288712884,
      "learning_rate": 1.4749521784088205e-06,
      "loss": 0.2246,
      "step": 8275
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4169237792576155,
      "learning_rate": 1.4739021771898e-06,
      "loss": 0.1781,
      "step": 8276
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.352632062190317,
      "learning_rate": 1.4728524852331477e-06,
      "loss": 0.1655,
      "step": 8277
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5884908891480187,
      "learning_rate": 1.471803102630926e-06,
      "loss": 0.2295,
      "step": 8278
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.4945132927379134,
      "learning_rate": 1.4707540294751772e-06,
      "loss": 0.5931,
      "step": 8279
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.413621321337149,
      "learning_rate": 1.469705265857908e-06,
      "loss": 0.1828,
      "step": 8280
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2841125252019259,
      "learning_rate": 1.4686568118711054e-06,
      "loss": 0.1505,
      "step": 8281
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3619229405734614,
      "learning_rate": 1.4676086676067224e-06,
      "loss": 0.1719,
      "step": 8282
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.82286878623829,
      "learning_rate": 1.4665608331566922e-06,
      "loss": 0.5243,
      "step": 8283
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2853414403346473,
      "learning_rate": 1.465513308612913e-06,
      "loss": 0.1795,
      "step": 8284
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4429073382490494,
      "learning_rate": 1.4644660940672628e-06,
      "loss": 0.1858,
      "step": 8285
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2327213764770704,
      "learning_rate": 1.4634191896115884e-06,
      "loss": 0.1747,
      "step": 8286
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1333869718911924,
      "learning_rate": 1.4623725953377082e-06,
      "loss": 0.1286,
      "step": 8287
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.358652254913664,
      "learning_rate": 1.4613263113374187e-06,
      "loss": 0.1866,
      "step": 8288
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.32814488969084,
      "learning_rate": 1.4602803377024833e-06,
      "loss": 0.1385,
      "step": 8289
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.8798321864206855,
      "learning_rate": 1.4592346745246432e-06,
      "loss": 0.6606,
      "step": 8290
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3571091420555792,
      "learning_rate": 1.4581893218956077e-06,
      "loss": 0.1473,
      "step": 8291
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.454108275210295,
      "learning_rate": 1.457144279907064e-06,
      "loss": 0.6487,
      "step": 8292
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.492404158413746,
      "learning_rate": 1.4560995486506663e-06,
      "loss": 0.1613,
      "step": 8293
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4651337452134119,
      "learning_rate": 1.455055128218048e-06,
      "loss": 0.1578,
      "step": 8294
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2884355921598503,
      "learning_rate": 1.4540110187008094e-06,
      "loss": 0.1585,
      "step": 8295
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4634924134740932,
      "learning_rate": 1.4529672201905248e-06,
      "loss": 0.1813,
      "step": 8296
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4515133094509154,
      "learning_rate": 1.451923732778745e-06,
      "loss": 0.1994,
      "step": 8297
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.392229406719439,
      "learning_rate": 1.4508805565569883e-06,
      "loss": 0.1935,
      "step": 8298
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3595780466524066,
      "learning_rate": 1.449837691616751e-06,
      "loss": 0.1715,
      "step": 8299
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3847522345730479,
      "learning_rate": 1.448795138049497e-06,
      "loss": 0.196,
      "step": 8300
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5053698332681336,
      "learning_rate": 1.447752895946667e-06,
      "loss": 0.1731,
      "step": 8301
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5160288443960892,
      "learning_rate": 1.4467109653996697e-06,
      "loss": 0.1941,
      "step": 8302
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2132167526835784,
      "learning_rate": 1.4456693464998938e-06,
      "loss": 0.1494,
      "step": 8303
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3909595366474738,
      "learning_rate": 1.4446280393386936e-06,
      "loss": 0.1902,
      "step": 8304
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.294350575246388,
      "learning_rate": 1.4435870440073968e-06,
      "loss": 0.1601,
      "step": 8305
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3073601598935463,
      "learning_rate": 1.44254636059731e-06,
      "loss": 0.1465,
      "step": 8306
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3172835567029153,
      "learning_rate": 1.441505989199704e-06,
      "loss": 0.1773,
      "step": 8307
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.458395978161336,
      "learning_rate": 1.44046592990583e-06,
      "loss": 0.5306,
      "step": 8308
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4600198963346172,
      "learning_rate": 1.4394261828069046e-06,
      "loss": 0.1713,
      "step": 8309
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1823112406021308,
      "learning_rate": 1.4383867479941238e-06,
      "loss": 0.1556,
      "step": 8310
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2306551069064604,
      "learning_rate": 1.4373476255586515e-06,
      "loss": 0.1621,
      "step": 8311
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3887731868634752,
      "learning_rate": 1.436308815591625e-06,
      "loss": 0.1784,
      "step": 8312
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2458065777740033,
      "learning_rate": 1.435270318184156e-06,
      "loss": 0.1719,
      "step": 8313
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.582749194700988,
      "learning_rate": 1.4342321334273262e-06,
      "loss": 0.6404,
      "step": 8314
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.273592696034716,
      "learning_rate": 1.4331942614121941e-06,
      "loss": 0.1496,
      "step": 8315
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.419752964036411,
      "learning_rate": 1.4321567022297849e-06,
      "loss": 0.152,
      "step": 8316
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.375614878947233,
      "learning_rate": 1.4311194559711022e-06,
      "loss": 0.178,
      "step": 8317
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3952033311031757,
      "learning_rate": 1.4300825227271164e-06,
      "loss": 0.1795,
      "step": 8318
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2746359356223707,
      "learning_rate": 1.4290459025887771e-06,
      "loss": 0.1573,
      "step": 8319
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3822669519969757,
      "learning_rate": 1.4280095956470014e-06,
      "loss": 0.193,
      "step": 8320
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1744359588685493,
      "learning_rate": 1.4269736019926778e-06,
      "loss": 0.1419,
      "step": 8321
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.524305038725684,
      "learning_rate": 1.4259379217166735e-06,
      "loss": 0.521,
      "step": 8322
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4183126109610409,
      "learning_rate": 1.4249025549098217e-06,
      "loss": 0.1952,
      "step": 8323
    },
    {
      "epoch": 0.76,
      "grad_norm": 4.26898197315685,
      "learning_rate": 1.423867501662934e-06,
      "loss": 0.5842,
      "step": 8324
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2291232674052586,
      "learning_rate": 1.4228327620667881e-06,
      "loss": 0.1992,
      "step": 8325
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2806258655820835,
      "learning_rate": 1.4217983362121407e-06,
      "loss": 0.1803,
      "step": 8326
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.302045869778176,
      "learning_rate": 1.420764224189715e-06,
      "loss": 0.1593,
      "step": 8327
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3372134770343056,
      "learning_rate": 1.4197304260902134e-06,
      "loss": 0.1511,
      "step": 8328
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.334585872380472,
      "learning_rate": 1.418696942004304e-06,
      "loss": 0.2034,
      "step": 8329
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.5344687460833,
      "learning_rate": 1.4176637720226288e-06,
      "loss": 0.4764,
      "step": 8330
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3141751796887429,
      "learning_rate": 1.4166309162358083e-06,
      "loss": 0.1592,
      "step": 8331
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.457253230508812,
      "learning_rate": 1.415598374734426e-06,
      "loss": 0.1788,
      "step": 8332
    },
    {
      "epoch": 0.76,
      "grad_norm": 9.259660106974644,
      "learning_rate": 1.4145661476090466e-06,
      "loss": 0.4736,
      "step": 8333
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6343489650819893,
      "learning_rate": 1.4135342349502007e-06,
      "loss": 0.1968,
      "step": 8334
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.273392815111158,
      "learning_rate": 1.4125026368483963e-06,
      "loss": 0.1412,
      "step": 8335
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3746546382904348,
      "learning_rate": 1.4114713533941082e-06,
      "loss": 0.1497,
      "step": 8336
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.483499931737661,
      "learning_rate": 1.410440384677791e-06,
      "loss": 0.1665,
      "step": 8337
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3192061921668436,
      "learning_rate": 1.4094097307898652e-06,
      "loss": 0.168,
      "step": 8338
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3307751671801746,
      "learning_rate": 1.4083793918207246e-06,
      "loss": 0.1413,
      "step": 8339
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.741908759555876,
      "learning_rate": 1.4073493678607397e-06,
      "loss": 0.5312,
      "step": 8340
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3809604327630456,
      "learning_rate": 1.4063196590002476e-06,
      "loss": 0.1763,
      "step": 8341
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3968960395877454,
      "learning_rate": 1.405290265329564e-06,
      "loss": 0.2025,
      "step": 8342
    },
    {
      "epoch": 0.76,
      "grad_norm": 14.243304885507227,
      "learning_rate": 1.40426118693897e-06,
      "loss": 0.3329,
      "step": 8343
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3280979126850956,
      "learning_rate": 1.4032324239187266e-06,
      "loss": 0.1839,
      "step": 8344
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2975489244033565,
      "learning_rate": 1.4022039763590595e-06,
      "loss": 0.1513,
      "step": 8345
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3984068891169057,
      "learning_rate": 1.4011758443501732e-06,
      "loss": 0.1766,
      "step": 8346
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2881126416798163,
      "learning_rate": 1.400148027982241e-06,
      "loss": 0.1724,
      "step": 8347
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.310147672983513,
      "learning_rate": 1.3991205273454072e-06,
      "loss": 0.1639,
      "step": 8348
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3917538962061993,
      "learning_rate": 1.3980933425297937e-06,
      "loss": 0.1972,
      "step": 8349
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.383925267597688,
      "learning_rate": 1.3970664736254885e-06,
      "loss": 0.1797,
      "step": 8350
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.621196811281956,
      "learning_rate": 1.3960399207225579e-06,
      "loss": 0.3823,
      "step": 8351
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3510314884962489,
      "learning_rate": 1.395013683911034e-06,
      "loss": 0.1791,
      "step": 8352
    },
    {
      "epoch": 0.76,
      "grad_norm": 8.3545908956294,
      "learning_rate": 1.3939877632809279e-06,
      "loss": 0.5747,
      "step": 8353
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3418676926501616,
      "learning_rate": 1.3929621589222186e-06,
      "loss": 0.1923,
      "step": 8354
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.417349228935806,
      "learning_rate": 1.3919368709248572e-06,
      "loss": 0.1605,
      "step": 8355
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.481104178501008,
      "learning_rate": 1.3909118993787702e-06,
      "loss": 0.2134,
      "step": 8356
    },
    {
      "epoch": 0.76,
      "grad_norm": 5.739242940967031,
      "learning_rate": 1.389887244373852e-06,
      "loss": 0.6214,
      "step": 8357
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3294870391517029,
      "learning_rate": 1.3888629059999747e-06,
      "loss": 0.1733,
      "step": 8358
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4726378988633166,
      "learning_rate": 1.387838884346977e-06,
      "loss": 0.216,
      "step": 8359
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.334189910537414,
      "learning_rate": 1.3868151795046746e-06,
      "loss": 0.1947,
      "step": 8360
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4849906978213234,
      "learning_rate": 1.3857917915628516e-06,
      "loss": 0.1833,
      "step": 8361
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3960241063403418,
      "learning_rate": 1.3847687206112676e-06,
      "loss": 0.2287,
      "step": 8362
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2585923598608504,
      "learning_rate": 1.383745966739652e-06,
      "loss": 0.1672,
      "step": 8363
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2416461423231278,
      "learning_rate": 1.3827235300377052e-06,
      "loss": 0.1555,
      "step": 8364
    },
    {
      "epoch": 0.76,
      "grad_norm": 6.60530580427243,
      "learning_rate": 1.3817014105951055e-06,
      "loss": 0.464,
      "step": 8365
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3389849296116836,
      "learning_rate": 1.3806796085014957e-06,
      "loss": 0.1586,
      "step": 8366
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2406497349801204,
      "learning_rate": 1.3796581238464984e-06,
      "loss": 0.1579,
      "step": 8367
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4511190551628612,
      "learning_rate": 1.378636956719701e-06,
      "loss": 0.1919,
      "step": 8368
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.352685525840772,
      "learning_rate": 1.3776161072106703e-06,
      "loss": 0.1663,
      "step": 8369
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3638936877515357,
      "learning_rate": 1.3765955754089382e-06,
      "loss": 0.1471,
      "step": 8370
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.064646197698674,
      "learning_rate": 1.3755753614040158e-06,
      "loss": 0.4804,
      "step": 8371
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4103533076943002,
      "learning_rate": 1.3745554652853805e-06,
      "loss": 0.1881,
      "step": 8372
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4824028998777516,
      "learning_rate": 1.3735358871424825e-06,
      "loss": 0.2082,
      "step": 8373
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.568610837127698,
      "learning_rate": 1.372516627064749e-06,
      "loss": 0.1892,
      "step": 8374
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3548603333996616,
      "learning_rate": 1.371497685141573e-06,
      "loss": 0.1586,
      "step": 8375
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3660627153346367,
      "learning_rate": 1.3704790614623248e-06,
      "loss": 0.2095,
      "step": 8376
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6905332654023264,
      "learning_rate": 1.369460756116342e-06,
      "loss": 0.2219,
      "step": 8377
    },
    {
      "epoch": 0.77,
      "grad_norm": 10.522198924122282,
      "learning_rate": 1.3684427691929403e-06,
      "loss": 0.5467,
      "step": 8378
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.400102105854044,
      "learning_rate": 1.3674251007814004e-06,
      "loss": 0.5967,
      "step": 8379
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3711256203659625,
      "learning_rate": 1.3664077509709817e-06,
      "loss": 0.1852,
      "step": 8380
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3425777482157746,
      "learning_rate": 1.3653907198509109e-06,
      "loss": 0.167,
      "step": 8381
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2386728622968137,
      "learning_rate": 1.3643740075103874e-06,
      "loss": 0.1436,
      "step": 8382
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.3337151942492,
      "learning_rate": 1.3633576140385863e-06,
      "loss": 0.4785,
      "step": 8383
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3024910546196642,
      "learning_rate": 1.3623415395246487e-06,
      "loss": 0.1635,
      "step": 8384
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3725001402386738,
      "learning_rate": 1.3613257840576954e-06,
      "loss": 0.1601,
      "step": 8385
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4738163661857189,
      "learning_rate": 1.3603103477268109e-06,
      "loss": 0.2036,
      "step": 8386
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.457015582746539,
      "learning_rate": 1.3592952306210589e-06,
      "loss": 0.1615,
      "step": 8387
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.360876402443144,
      "learning_rate": 1.358280432829469e-06,
      "loss": 0.5716,
      "step": 8388
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4052505976474077,
      "learning_rate": 1.3572659544410493e-06,
      "loss": 0.1833,
      "step": 8389
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.157220467795166,
      "learning_rate": 1.356251795544774e-06,
      "loss": 0.5819,
      "step": 8390
    },
    {
      "epoch": 0.77,
      "grad_norm": 4.571433919806657,
      "learning_rate": 1.3552379562295908e-06,
      "loss": 0.4974,
      "step": 8391
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6743795017822687,
      "learning_rate": 1.3542244365844227e-06,
      "loss": 0.1974,
      "step": 8392
    },
    {
      "epoch": 0.77,
      "grad_norm": 9.638975052571453,
      "learning_rate": 1.3532112366981598e-06,
      "loss": 0.352,
      "step": 8393
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2889343461446827,
      "learning_rate": 1.3521983566596685e-06,
      "loss": 0.1636,
      "step": 8394
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.206661070864568,
      "learning_rate": 1.3511857965577829e-06,
      "loss": 0.1631,
      "step": 8395
    },
    {
      "epoch": 0.77,
      "grad_norm": 13.466475572977997,
      "learning_rate": 1.350173556481314e-06,
      "loss": 0.5573,
      "step": 8396
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2940287447437546,
      "learning_rate": 1.3491616365190414e-06,
      "loss": 0.1658,
      "step": 8397
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.425201674949447,
      "learning_rate": 1.3481500367597139e-06,
      "loss": 0.1819,
      "step": 8398
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3756885859951669,
      "learning_rate": 1.3471387572920602e-06,
      "loss": 0.1405,
      "step": 8399
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.236257636989753,
      "learning_rate": 1.3461277982047733e-06,
      "loss": 0.1394,
      "step": 8400
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3206467588892226,
      "learning_rate": 1.3451171595865226e-06,
      "loss": 0.1749,
      "step": 8401
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.347742676741923,
      "learning_rate": 1.3441068415259462e-06,
      "loss": 0.162,
      "step": 8402
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3691756391336423,
      "learning_rate": 1.343096844111658e-06,
      "loss": 0.1657,
      "step": 8403
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.508527581075846,
      "learning_rate": 1.3420871674322394e-06,
      "loss": 0.7573,
      "step": 8404
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4156884171676016,
      "learning_rate": 1.341077811576248e-06,
      "loss": 0.1635,
      "step": 8405
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.226597583793121,
      "learning_rate": 1.34006877663221e-06,
      "loss": 0.1652,
      "step": 8406
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5108192606033293,
      "learning_rate": 1.3390600626886225e-06,
      "loss": 0.1717,
      "step": 8407
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2951884549523112,
      "learning_rate": 1.3380516698339601e-06,
      "loss": 0.1785,
      "step": 8408
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4480272566266024,
      "learning_rate": 1.3370435981566622e-06,
      "loss": 0.1676,
      "step": 8409
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3049307025146342,
      "learning_rate": 1.3360358477451469e-06,
      "loss": 0.1848,
      "step": 8410
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4570489035028298,
      "learning_rate": 1.3350284186877966e-06,
      "loss": 0.1754,
      "step": 8411
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3206721096313898,
      "learning_rate": 1.3340213110729733e-06,
      "loss": 0.1385,
      "step": 8412
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4255847755318791,
      "learning_rate": 1.3330145249890047e-06,
      "loss": 0.1615,
      "step": 8413
    },
    {
      "epoch": 0.77,
      "grad_norm": 8.196171477026247,
      "learning_rate": 1.332008060524194e-06,
      "loss": 0.4388,
      "step": 8414
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.317011519773128,
      "learning_rate": 1.3310019177668154e-06,
      "loss": 0.5618,
      "step": 8415
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6158362009239666,
      "learning_rate": 1.3299960968051112e-06,
      "loss": 0.1659,
      "step": 8416
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2804066519550563,
      "learning_rate": 1.3289905977273027e-06,
      "loss": 0.1776,
      "step": 8417
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.146484016481389,
      "learning_rate": 1.3279854206215747e-06,
      "loss": 0.5579,
      "step": 8418
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3264604306562962,
      "learning_rate": 1.3269805655760926e-06,
      "loss": 0.143,
      "step": 8419
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7542065864031524,
      "learning_rate": 1.3259760326789845e-06,
      "loss": 0.172,
      "step": 8420
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.025517553457639,
      "learning_rate": 1.3249718220183583e-06,
      "loss": 0.1775,
      "step": 8421
    },
    {
      "epoch": 0.77,
      "grad_norm": 7.745354244124625,
      "learning_rate": 1.323967933682287e-06,
      "loss": 0.5476,
      "step": 8422
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2934225098423124,
      "learning_rate": 1.3229643677588205e-06,
      "loss": 0.1895,
      "step": 8423
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4009212304312177,
      "learning_rate": 1.3219611243359782e-06,
      "loss": 0.1827,
      "step": 8424
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4098249152230191,
      "learning_rate": 1.3209582035017487e-06,
      "loss": 0.1524,
      "step": 8425
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.269086532621238,
      "learning_rate": 1.319955605344098e-06,
      "loss": 0.1609,
      "step": 8426
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.557667523236311,
      "learning_rate": 1.3189533299509577e-06,
      "loss": 0.5065,
      "step": 8427
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1673610174232865,
      "learning_rate": 1.3179513774102376e-06,
      "loss": 0.1289,
      "step": 8428
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.778902074686792,
      "learning_rate": 1.316949747809812e-06,
      "loss": 0.5578,
      "step": 8429
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4087743816202065,
      "learning_rate": 1.315948441237534e-06,
      "loss": 0.1954,
      "step": 8430
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1524518278356757,
      "learning_rate": 1.314947457781221e-06,
      "loss": 0.1309,
      "step": 8431
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2955315800103004,
      "learning_rate": 1.3139467975286702e-06,
      "loss": 0.149,
      "step": 8432
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2349398501979956,
      "learning_rate": 1.312946460567644e-06,
      "loss": 0.1834,
      "step": 8433
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3017080046018303,
      "learning_rate": 1.3119464469858767e-06,
      "loss": 0.1849,
      "step": 8434
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3689479510831057,
      "learning_rate": 1.3109467568710798e-06,
      "loss": 0.1448,
      "step": 8435
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4639552910744624,
      "learning_rate": 1.30994739031093e-06,
      "loss": 0.1752,
      "step": 8436
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3641830773949104,
      "learning_rate": 1.308948347393081e-06,
      "loss": 0.2015,
      "step": 8437
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3743020796170968,
      "learning_rate": 1.307949628205153e-06,
      "loss": 0.2007,
      "step": 8438
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.241697281287567,
      "learning_rate": 1.3069512328347421e-06,
      "loss": 0.1737,
      "step": 8439
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2843892078950918,
      "learning_rate": 1.3059531613694143e-06,
      "loss": 0.1403,
      "step": 8440
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3439884566977114,
      "learning_rate": 1.3049554138967052e-06,
      "loss": 0.1467,
      "step": 8441
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3402897257367477,
      "learning_rate": 1.3039579905041267e-06,
      "loss": 0.1603,
      "step": 8442
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.320194521977236,
      "learning_rate": 1.3029608912791564e-06,
      "loss": 0.1627,
      "step": 8443
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2148116700051974,
      "learning_rate": 1.3019641163092494e-06,
      "loss": 0.1537,
      "step": 8444
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3857145001491653,
      "learning_rate": 1.3009676656818277e-06,
      "loss": 0.1913,
      "step": 8445
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2694406830535243,
      "learning_rate": 1.2999715394842882e-06,
      "loss": 0.1774,
      "step": 8446
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2948019209643826,
      "learning_rate": 1.2989757378039958e-06,
      "loss": 0.1731,
      "step": 8447
    },
    {
      "epoch": 0.77,
      "grad_norm": 6.060133183694328,
      "learning_rate": 1.2979802607282915e-06,
      "loss": 0.6272,
      "step": 8448
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3983185506987723,
      "learning_rate": 1.2969851083444834e-06,
      "loss": 0.2161,
      "step": 8449
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2678144311792636,
      "learning_rate": 1.2959902807398528e-06,
      "loss": 0.1444,
      "step": 8450
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.275804570460314,
      "learning_rate": 1.2949957780016536e-06,
      "loss": 0.1092,
      "step": 8451
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4695208204818917,
      "learning_rate": 1.2940016002171097e-06,
      "loss": 0.1956,
      "step": 8452
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.305333109725317,
      "learning_rate": 1.2930077474734192e-06,
      "loss": 0.1606,
      "step": 8453
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.399502752022421,
      "learning_rate": 1.2920142198577484e-06,
      "loss": 0.1954,
      "step": 8454
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3946995891083174,
      "learning_rate": 1.2910210174572346e-06,
      "loss": 0.1937,
      "step": 8455
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.280049019079319,
      "learning_rate": 1.2900281403589916e-06,
      "loss": 0.1931,
      "step": 8456
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4132987665897778,
      "learning_rate": 1.2890355886500971e-06,
      "loss": 0.1795,
      "step": 8457
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4858471815791352,
      "learning_rate": 1.2880433624176092e-06,
      "loss": 0.1974,
      "step": 8458
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2777187302021296,
      "learning_rate": 1.2870514617485486e-06,
      "loss": 0.1546,
      "step": 8459
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.528495396056384,
      "learning_rate": 1.2860598867299152e-06,
      "loss": 0.1886,
      "step": 8460
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.457123674278731,
      "learning_rate": 1.2850686374486732e-06,
      "loss": 0.1908,
      "step": 8461
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3751217933049984,
      "learning_rate": 1.284077713991766e-06,
      "loss": 0.1943,
      "step": 8462
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7892933602684382,
      "learning_rate": 1.2830871164461012e-06,
      "loss": 0.1928,
      "step": 8463
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4685166880411713,
      "learning_rate": 1.2820968448985598e-06,
      "loss": 0.1981,
      "step": 8464
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.8941062025474917,
      "learning_rate": 1.2811068994359992e-06,
      "loss": 0.1897,
      "step": 8465
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2977222291050883,
      "learning_rate": 1.2801172801452406e-06,
      "loss": 0.1579,
      "step": 8466
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.514481248007388,
      "learning_rate": 1.2791279871130824e-06,
      "loss": 0.1796,
      "step": 8467
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.372478317125059,
      "learning_rate": 1.2781390204262905e-06,
      "loss": 0.214,
      "step": 8468
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2820290880324448,
      "learning_rate": 1.2771503801716061e-06,
      "loss": 0.1762,
      "step": 8469
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.353980452536283,
      "learning_rate": 1.2761620664357366e-06,
      "loss": 0.168,
      "step": 8470
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2098572148782691,
      "learning_rate": 1.2751740793053674e-06,
      "loss": 0.1503,
      "step": 8471
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2867944999506549,
      "learning_rate": 1.2741864188671494e-06,
      "loss": 0.1246,
      "step": 8472
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2758355439452542,
      "learning_rate": 1.273199085207706e-06,
      "loss": 0.1608,
      "step": 8473
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2057398788416531,
      "learning_rate": 1.2722120784136355e-06,
      "loss": 0.1375,
      "step": 8474
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2898651510420247,
      "learning_rate": 1.2712253985715023e-06,
      "loss": 0.1903,
      "step": 8475
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3981924449731995,
      "learning_rate": 1.2702390457678476e-06,
      "loss": 0.1461,
      "step": 8476
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3250365715482613,
      "learning_rate": 1.2692530200891783e-06,
      "loss": 0.1453,
      "step": 8477
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3128261489540969,
      "learning_rate": 1.2682673216219788e-06,
      "loss": 0.146,
      "step": 8478
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3578698720328441,
      "learning_rate": 1.267281950452699e-06,
      "loss": 0.1456,
      "step": 8479
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4451438049316545,
      "learning_rate": 1.266296906667762e-06,
      "loss": 0.191,
      "step": 8480
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4439946942433577,
      "learning_rate": 1.2653121903535653e-06,
      "loss": 0.1675,
      "step": 8481
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2974106229793703,
      "learning_rate": 1.2643278015964722e-06,
      "loss": 0.1828,
      "step": 8482
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4249825109011338,
      "learning_rate": 1.263343740482823e-06,
      "loss": 0.1851,
      "step": 8483
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.532571535066409,
      "learning_rate": 1.262360007098924e-06,
      "loss": 0.2151,
      "step": 8484
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2649323308971396,
      "learning_rate": 1.2613766015310575e-06,
      "loss": 0.141,
      "step": 8485
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2169781259400383,
      "learning_rate": 1.260393523865472e-06,
      "loss": 0.1426,
      "step": 8486
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3393510364667087,
      "learning_rate": 1.2594107741883927e-06,
      "loss": 0.1808,
      "step": 8487
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.419176207430988,
      "learning_rate": 1.2584283525860126e-06,
      "loss": 0.1885,
      "step": 8488
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4736179646631895,
      "learning_rate": 1.257446259144494e-06,
      "loss": 0.2481,
      "step": 8489
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2829550693516507,
      "learning_rate": 1.2564644939499775e-06,
      "loss": 0.1552,
      "step": 8490
    },
    {
      "epoch": 0.78,
      "grad_norm": 11.274059780110953,
      "learning_rate": 1.2554830570885668e-06,
      "loss": 0.5537,
      "step": 8491
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3498201950958442,
      "learning_rate": 1.2545019486463428e-06,
      "loss": 0.1817,
      "step": 8492
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3533149900585115,
      "learning_rate": 1.2535211687093535e-06,
      "loss": 0.1837,
      "step": 8493
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.655389960135336,
      "learning_rate": 1.2525407173636218e-06,
      "loss": 0.2167,
      "step": 8494
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5278669016517281,
      "learning_rate": 1.251560594695137e-06,
      "loss": 0.1854,
      "step": 8495
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.04342505746839,
      "learning_rate": 1.2505808007898662e-06,
      "loss": 0.4473,
      "step": 8496
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3795225204285446,
      "learning_rate": 1.2496013357337416e-06,
      "loss": 0.1933,
      "step": 8497
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3519180442663192,
      "learning_rate": 1.248622199612668e-06,
      "loss": 0.1893,
      "step": 8498
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3202694486012863,
      "learning_rate": 1.2476433925125248e-06,
      "loss": 0.1863,
      "step": 8499
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3630005651931492,
      "learning_rate": 1.2466649145191567e-06,
      "loss": 0.1416,
      "step": 8500
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4723096487052811,
      "learning_rate": 1.2456867657183857e-06,
      "loss": 0.1888,
      "step": 8501
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.190399327644042,
      "learning_rate": 1.244708946196e-06,
      "loss": 0.5888,
      "step": 8502
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.352505566195008,
      "learning_rate": 1.2437314560377629e-06,
      "loss": 0.1454,
      "step": 8503
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4400442194920096,
      "learning_rate": 1.2427542953294036e-06,
      "loss": 0.1644,
      "step": 8504
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.931360124634717,
      "learning_rate": 1.2417774641566298e-06,
      "loss": 0.5093,
      "step": 8505
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2905311570537472,
      "learning_rate": 1.2408009626051137e-06,
      "loss": 0.1599,
      "step": 8506
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4591362237213343,
      "learning_rate": 1.2398247907605e-06,
      "loss": 0.2171,
      "step": 8507
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3371993136304332,
      "learning_rate": 1.2388489487084082e-06,
      "loss": 0.1652,
      "step": 8508
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3874093348293812,
      "learning_rate": 1.237873436534423e-06,
      "loss": 0.1516,
      "step": 8509
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.8640631655137225,
      "learning_rate": 1.236898254324107e-06,
      "loss": 0.7071,
      "step": 8510
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2742754804643308,
      "learning_rate": 1.2359234021629868e-06,
      "loss": 0.1782,
      "step": 8511
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2143487538983597,
      "learning_rate": 1.2349488801365666e-06,
      "loss": 0.145,
      "step": 8512
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.510600434362621,
      "learning_rate": 1.233974688330315e-06,
      "loss": 0.2169,
      "step": 8513
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5678359637556112,
      "learning_rate": 1.2330008268296789e-06,
      "loss": 0.1899,
      "step": 8514
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4549176566222883,
      "learning_rate": 1.2320272957200702e-06,
      "loss": 0.1668,
      "step": 8515
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3902926127432753,
      "learning_rate": 1.2310540950868732e-06,
      "loss": 0.1683,
      "step": 8516
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.956203358185076,
      "learning_rate": 1.230081225015447e-06,
      "loss": 0.415,
      "step": 8517
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.0771215382147785,
      "learning_rate": 1.229108685591116e-06,
      "loss": 0.3426,
      "step": 8518
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.331724602124372,
      "learning_rate": 1.2281364768991804e-06,
      "loss": 0.1695,
      "step": 8519
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.359573835114451,
      "learning_rate": 1.2271645990249075e-06,
      "loss": 0.2056,
      "step": 8520
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4423999391889095,
      "learning_rate": 1.2261930520535403e-06,
      "loss": 0.1718,
      "step": 8521
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3457174993496634,
      "learning_rate": 1.2252218360702884e-06,
      "loss": 0.1369,
      "step": 8522
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4159484057742588,
      "learning_rate": 1.2242509511603318e-06,
      "loss": 0.1473,
      "step": 8523
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4719351544615777,
      "learning_rate": 1.2232803974088276e-06,
      "loss": 0.174,
      "step": 8524
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3569531566569033,
      "learning_rate": 1.222310174900896e-06,
      "loss": 0.1956,
      "step": 8525
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4399553858895762,
      "learning_rate": 1.2213402837216354e-06,
      "loss": 0.1858,
      "step": 8526
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.391573370708581,
      "learning_rate": 1.2203707239561092e-06,
      "loss": 0.1638,
      "step": 8527
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5059393019546046,
      "learning_rate": 1.2194014956893562e-06,
      "loss": 0.1832,
      "step": 8528
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.213753325621608,
      "learning_rate": 1.2184325990063822e-06,
      "loss": 0.1571,
      "step": 8529
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1067083424324635,
      "learning_rate": 1.2174640339921683e-06,
      "loss": 0.1463,
      "step": 8530
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.367618473462389,
      "learning_rate": 1.216495800731663e-06,
      "loss": 0.1857,
      "step": 8531
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5236981852306666,
      "learning_rate": 1.2155278993097853e-06,
      "loss": 0.1773,
      "step": 8532
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4904517481935236,
      "learning_rate": 1.2145603298114294e-06,
      "loss": 0.1775,
      "step": 8533
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5772330546462126,
      "learning_rate": 1.213593092321455e-06,
      "loss": 0.2201,
      "step": 8534
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.972703717182438,
      "learning_rate": 1.2126261869246976e-06,
      "loss": 0.6178,
      "step": 8535
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4689550787979613,
      "learning_rate": 1.2116596137059589e-06,
      "loss": 0.2101,
      "step": 8536
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1831705572180138,
      "learning_rate": 1.210693372750017e-06,
      "loss": 0.1436,
      "step": 8537
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3372464584227783,
      "learning_rate": 1.209727464141614e-06,
      "loss": 0.1933,
      "step": 8538
    },
    {
      "epoch": 0.78,
      "grad_norm": 7.435559762001543,
      "learning_rate": 1.20876188796547e-06,
      "loss": 0.5768,
      "step": 8539
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.53681196636163,
      "learning_rate": 1.2077966443062706e-06,
      "loss": 0.4436,
      "step": 8540
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.335514644969168,
      "learning_rate": 1.2068317332486735e-06,
      "loss": 0.5852,
      "step": 8541
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4002810450259249,
      "learning_rate": 1.20586715487731e-06,
      "loss": 0.1612,
      "step": 8542
    },
    {
      "epoch": 0.78,
      "grad_norm": 5.395939557492612,
      "learning_rate": 1.2049029092767773e-06,
      "loss": 0.5852,
      "step": 8543
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3655341168888837,
      "learning_rate": 1.203938996531649e-06,
      "loss": 0.1322,
      "step": 8544
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2982355672426278,
      "learning_rate": 1.202975416726464e-06,
      "loss": 0.1752,
      "step": 8545
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3592342380707598,
      "learning_rate": 1.2020121699457376e-06,
      "loss": 0.1875,
      "step": 8546
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3094102539329575,
      "learning_rate": 1.2010492562739505e-06,
      "loss": 0.1932,
      "step": 8547
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3314517166181885,
      "learning_rate": 1.2000866757955588e-06,
      "loss": 0.2031,
      "step": 8548
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3940992287845029,
      "learning_rate": 1.199124428594986e-06,
      "loss": 0.172,
      "step": 8549
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3022468319064082,
      "learning_rate": 1.1981625147566268e-06,
      "loss": 0.1935,
      "step": 8550
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4743402230757816,
      "learning_rate": 1.1972009343648494e-06,
      "loss": 0.1725,
      "step": 8551
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3980797019066842,
      "learning_rate": 1.1962396875039884e-06,
      "loss": 0.1711,
      "step": 8552
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6473682796690963,
      "learning_rate": 1.1952787742583549e-06,
      "loss": 0.2028,
      "step": 8553
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4810872969183007,
      "learning_rate": 1.1943181947122235e-06,
      "loss": 0.185,
      "step": 8554
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4599229428863065,
      "learning_rate": 1.1933579489498475e-06,
      "loss": 0.1894,
      "step": 8555
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3827925179554224,
      "learning_rate": 1.1923980370554434e-06,
      "loss": 0.2047,
      "step": 8556
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2546758622651968,
      "learning_rate": 1.1914384591132045e-06,
      "loss": 0.1402,
      "step": 8557
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.374089404233112,
      "learning_rate": 1.1904792152072914e-06,
      "loss": 0.181,
      "step": 8558
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4697975814653894,
      "learning_rate": 1.1895203054218341e-06,
      "loss": 0.1805,
      "step": 8559
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5187800848025363,
      "learning_rate": 1.188561729840939e-06,
      "loss": 0.1945,
      "step": 8560
    },
    {
      "epoch": 0.78,
      "grad_norm": 8.486772933488345,
      "learning_rate": 1.1876034885486764e-06,
      "loss": 0.4497,
      "step": 8561
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.416541982536195,
      "learning_rate": 1.186645581629093e-06,
      "loss": 0.1433,
      "step": 8562
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4007750518574402,
      "learning_rate": 1.185688009166201e-06,
      "loss": 0.1369,
      "step": 8563
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.430712483815631,
      "learning_rate": 1.1847307712439893e-06,
      "loss": 0.1836,
      "step": 8564
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3549625247049082,
      "learning_rate": 1.1837738679464117e-06,
      "loss": 0.1632,
      "step": 8565
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.536633009809763,
      "learning_rate": 1.1828172993573944e-06,
      "loss": 0.1944,
      "step": 8566
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2138395124390262,
      "learning_rate": 1.1818610655608376e-06,
      "loss": 0.1417,
      "step": 8567
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3195867257139824,
      "learning_rate": 1.1809051666406058e-06,
      "loss": 0.1556,
      "step": 8568
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4602835507646414,
      "learning_rate": 1.1799496026805413e-06,
      "loss": 0.1612,
      "step": 8569
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3722922141129987,
      "learning_rate": 1.1789943737644505e-06,
      "loss": 0.1798,
      "step": 8570
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.369904141210929,
      "learning_rate": 1.1780394799761163e-06,
      "loss": 0.1623,
      "step": 8571
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.125919991898343,
      "learning_rate": 1.1770849213992864e-06,
      "loss": 0.2244,
      "step": 8572
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.391883683098805,
      "learning_rate": 1.176130698117684e-06,
      "loss": 0.1694,
      "step": 8573
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3431908835570552,
      "learning_rate": 1.1751768102150008e-06,
      "loss": 0.1963,
      "step": 8574
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4444668753406815,
      "learning_rate": 1.174223257774897e-06,
      "loss": 0.2083,
      "step": 8575
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6629030329126213,
      "learning_rate": 1.1732700408810077e-06,
      "loss": 0.1861,
      "step": 8576
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4084329528614274,
      "learning_rate": 1.1723171596169353e-06,
      "loss": 0.1824,
      "step": 8577
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2095350164382508,
      "learning_rate": 1.171364614066255e-06,
      "loss": 0.142,
      "step": 8578
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3142494010027037,
      "learning_rate": 1.1704124043125098e-06,
      "loss": 0.1821,
      "step": 8579
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4672375880894188,
      "learning_rate": 1.1694605304392164e-06,
      "loss": 0.1783,
      "step": 8580
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2045029399474223,
      "learning_rate": 1.1685089925298593e-06,
      "loss": 0.142,
      "step": 8581
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.29694953392117,
      "learning_rate": 1.1675577906678959e-06,
      "loss": 0.1455,
      "step": 8582
    },
    {
      "epoch": 0.78,
      "grad_norm": 6.053238972130766,
      "learning_rate": 1.166606924936753e-06,
      "loss": 0.5739,
      "step": 8583
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.311757902166503,
      "learning_rate": 1.1656563954198258e-06,
      "loss": 0.1687,
      "step": 8584
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.37957727887493,
      "learning_rate": 1.1647062022004845e-06,
      "loss": 0.1968,
      "step": 8585
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5440977422095707,
      "learning_rate": 1.1637563453620654e-06,
      "loss": 0.2135,
      "step": 8586
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2092233556320569,
      "learning_rate": 1.1628068249878798e-06,
      "loss": 0.1484,
      "step": 8587
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.417324585910642,
      "learning_rate": 1.1618576411612042e-06,
      "loss": 0.2139,
      "step": 8588
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4672124707731276,
      "learning_rate": 1.1609087939652908e-06,
      "loss": 0.1855,
      "step": 8589
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4179164168600005,
      "learning_rate": 1.1599602834833574e-06,
      "loss": 0.1707,
      "step": 8590
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4307989962485268,
      "learning_rate": 1.159012109798598e-06,
      "loss": 0.1673,
      "step": 8591
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.181854710532463,
      "learning_rate": 1.158064272994171e-06,
      "loss": 0.5429,
      "step": 8592
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.55089308079013,
      "learning_rate": 1.157116773153208e-06,
      "loss": 0.1861,
      "step": 8593
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2104148128430352,
      "learning_rate": 1.1561696103588132e-06,
      "loss": 0.141,
      "step": 8594
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4228337716674933,
      "learning_rate": 1.1552227846940567e-06,
      "loss": 0.1855,
      "step": 8595
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3000841220453911,
      "learning_rate": 1.154276296241984e-06,
      "loss": 0.1616,
      "step": 8596
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3729459144148926,
      "learning_rate": 1.1533301450856054e-06,
      "loss": 0.1757,
      "step": 8597
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.345014242732362,
      "learning_rate": 1.1523843313079075e-06,
      "loss": 0.2285,
      "step": 8598
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2148985319225,
      "learning_rate": 1.1514388549918426e-06,
      "loss": 0.1527,
      "step": 8599
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.209784601845919,
      "learning_rate": 1.1504937162203373e-06,
      "loss": 0.5985,
      "step": 8600
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4852637105002675,
      "learning_rate": 1.1495489150762851e-06,
      "loss": 0.1677,
      "step": 8601
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3462072528799536,
      "learning_rate": 1.1486044516425504e-06,
      "loss": 0.1506,
      "step": 8602
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2142199011883195,
      "learning_rate": 1.1476603260019714e-06,
      "loss": 0.1491,
      "step": 8603
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4186248275522797,
      "learning_rate": 1.1467165382373518e-06,
      "loss": 0.1914,
      "step": 8604
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4039994426833178,
      "learning_rate": 1.145773088431471e-06,
      "loss": 0.1973,
      "step": 8605
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4948956373505269,
      "learning_rate": 1.1448299766670723e-06,
      "loss": 0.1601,
      "step": 8606
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2603389325057743,
      "learning_rate": 1.1438872030268766e-06,
      "loss": 0.1481,
      "step": 8607
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4690490922386856,
      "learning_rate": 1.1429447675935695e-06,
      "loss": 0.1913,
      "step": 8608
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.422370949385181,
      "learning_rate": 1.1420026704498077e-06,
      "loss": 0.5603,
      "step": 8609
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3671827378587154,
      "learning_rate": 1.1410609116782218e-06,
      "loss": 0.1716,
      "step": 8610
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3543027973425157,
      "learning_rate": 1.1401194913614084e-06,
      "loss": 0.1692,
      "step": 8611
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.642362779215581,
      "learning_rate": 1.139178409581938e-06,
      "loss": 0.5273,
      "step": 8612
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4246493044999526,
      "learning_rate": 1.1382376664223482e-06,
      "loss": 0.2003,
      "step": 8613
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.270551036445925,
      "learning_rate": 1.1372972619651506e-06,
      "loss": 0.1623,
      "step": 8614
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.317318765277382,
      "learning_rate": 1.1363571962928227e-06,
      "loss": 0.1671,
      "step": 8615
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3576233644294946,
      "learning_rate": 1.1354174694878161e-06,
      "loss": 0.1804,
      "step": 8616
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.406243888870676,
      "learning_rate": 1.1344780816325512e-06,
      "loss": 0.1666,
      "step": 8617
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.400405108541393,
      "learning_rate": 1.1335390328094164e-06,
      "loss": 0.1562,
      "step": 8618
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8864251908875675,
      "learning_rate": 1.1326003231007753e-06,
      "loss": 0.2146,
      "step": 8619
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.280271395798489,
      "learning_rate": 1.1316619525889566e-06,
      "loss": 0.1809,
      "step": 8620
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4248569213805509,
      "learning_rate": 1.1307239213562643e-06,
      "loss": 0.1921,
      "step": 8621
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2949120588639014,
      "learning_rate": 1.1297862294849671e-06,
      "loss": 0.1771,
      "step": 8622
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4931205332579698,
      "learning_rate": 1.1288488770573097e-06,
      "loss": 0.1631,
      "step": 8623
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0909692826888364,
      "learning_rate": 1.1279118641555016e-06,
      "loss": 0.1188,
      "step": 8624
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.302967420748658,
      "learning_rate": 1.1269751908617277e-06,
      "loss": 0.1445,
      "step": 8625
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4657578474857773,
      "learning_rate": 1.1260388572581387e-06,
      "loss": 0.1895,
      "step": 8626
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2667808888281968,
      "learning_rate": 1.125102863426857e-06,
      "loss": 0.1511,
      "step": 8627
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.28935207411956,
      "learning_rate": 1.124167209449978e-06,
      "loss": 0.1969,
      "step": 8628
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3621390268098643,
      "learning_rate": 1.123231895409561e-06,
      "loss": 0.1513,
      "step": 8629
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6178971285758121,
      "learning_rate": 1.1222969213876427e-06,
      "loss": 0.1904,
      "step": 8630
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.008937580261453,
      "learning_rate": 1.1213622874662244e-06,
      "loss": 0.3687,
      "step": 8631
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.347745993639479,
      "learning_rate": 1.120427993727282e-06,
      "loss": 0.1856,
      "step": 8632
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4367049127264229,
      "learning_rate": 1.1194940402527566e-06,
      "loss": 0.1948,
      "step": 8633
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4476953693072996,
      "learning_rate": 1.1185604271245643e-06,
      "loss": 0.1837,
      "step": 8634
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1650479297159935,
      "learning_rate": 1.1176271544245892e-06,
      "loss": 0.1375,
      "step": 8635
    },
    {
      "epoch": 0.79,
      "grad_norm": 6.439084782382419,
      "learning_rate": 1.1166942222346828e-06,
      "loss": 0.5717,
      "step": 8636
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2757478032001404,
      "learning_rate": 1.115761630636673e-06,
      "loss": 0.155,
      "step": 8637
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4013265143265168,
      "learning_rate": 1.1148293797123516e-06,
      "loss": 0.166,
      "step": 8638
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4188482132901301,
      "learning_rate": 1.1138974695434863e-06,
      "loss": 0.2187,
      "step": 8639
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2311112703996014,
      "learning_rate": 1.1129659002118075e-06,
      "loss": 0.203,
      "step": 8640
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2893992586646124,
      "learning_rate": 1.112034671799025e-06,
      "loss": 0.1398,
      "step": 8641
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4881083372892812,
      "learning_rate": 1.1111037843868095e-06,
      "loss": 0.2082,
      "step": 8642
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2634936985418053,
      "learning_rate": 1.1101732380568091e-06,
      "loss": 0.1626,
      "step": 8643
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3567544443133608,
      "learning_rate": 1.109243032890638e-06,
      "loss": 0.1663,
      "step": 8644
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.00012319498089,
      "learning_rate": 1.1083131689698796e-06,
      "loss": 0.6473,
      "step": 8645
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5111040255431347,
      "learning_rate": 1.1073836463760923e-06,
      "loss": 0.2145,
      "step": 8646
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3385802985878528,
      "learning_rate": 1.1064544651907982e-06,
      "loss": 0.1749,
      "step": 8647
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3351188909846956,
      "learning_rate": 1.105525625495496e-06,
      "loss": 0.1865,
      "step": 8648
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.305506666026339,
      "learning_rate": 1.1045971273716476e-06,
      "loss": 0.1857,
      "step": 8649
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3670048859516146,
      "learning_rate": 1.1036689709006925e-06,
      "loss": 0.1779,
      "step": 8650
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.448939934710532,
      "learning_rate": 1.102741156164034e-06,
      "loss": 0.1947,
      "step": 8651
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.516807301059232,
      "learning_rate": 1.101813683243046e-06,
      "loss": 0.1681,
      "step": 8652
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2868959392747397,
      "learning_rate": 1.1008865522190775e-06,
      "loss": 0.1608,
      "step": 8653
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3602557382744607,
      "learning_rate": 1.0999597631734415e-06,
      "loss": 0.1551,
      "step": 8654
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3761904455438225,
      "learning_rate": 1.0990333161874257e-06,
      "loss": 0.1698,
      "step": 8655
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.419465768617789,
      "learning_rate": 1.0981072113422836e-06,
      "loss": 0.1765,
      "step": 8656
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3192076338225704,
      "learning_rate": 1.0971814487192429e-06,
      "loss": 0.1915,
      "step": 8657
    },
    {
      "epoch": 0.79,
      "grad_norm": 10.69869494940607,
      "learning_rate": 1.096256028399496e-06,
      "loss": 0.6768,
      "step": 8658
    },
    {
      "epoch": 0.79,
      "grad_norm": 7.995547281983222,
      "learning_rate": 1.095330950464213e-06,
      "loss": 0.6147,
      "step": 8659
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3288965868629645,
      "learning_rate": 1.094406214994526e-06,
      "loss": 0.1931,
      "step": 8660
    },
    {
      "epoch": 0.79,
      "grad_norm": 8.92150062291338,
      "learning_rate": 1.0934818220715403e-06,
      "loss": 0.5605,
      "step": 8661
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.128606116261427,
      "learning_rate": 1.0925577717763342e-06,
      "loss": 0.1232,
      "step": 8662
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2098658276081635,
      "learning_rate": 1.0916340641899499e-06,
      "loss": 0.186,
      "step": 8663
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4844155943284008,
      "learning_rate": 1.0907106993934052e-06,
      "loss": 0.1619,
      "step": 8664
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5467913173959722,
      "learning_rate": 1.089787677467683e-06,
      "loss": 0.2128,
      "step": 8665
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4683682606825665,
      "learning_rate": 1.0888649984937416e-06,
      "loss": 0.1924,
      "step": 8666
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.769581913359929,
      "learning_rate": 1.087942662552503e-06,
      "loss": 0.5414,
      "step": 8667
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3391444697668071,
      "learning_rate": 1.0870206697248648e-06,
      "loss": 0.1823,
      "step": 8668
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.440431313537403,
      "learning_rate": 1.086099020091691e-06,
      "loss": 0.2167,
      "step": 8669
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3106148479989204,
      "learning_rate": 1.0851777137338142e-06,
      "loss": 0.162,
      "step": 8670
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7612676536243923,
      "learning_rate": 1.0842567507320429e-06,
      "loss": 0.1419,
      "step": 8671
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5008373168462288,
      "learning_rate": 1.0833361311671486e-06,
      "loss": 0.1914,
      "step": 8672
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3391419773958266,
      "learning_rate": 1.0824158551198783e-06,
      "loss": 0.1795,
      "step": 8673
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2908366981332209,
      "learning_rate": 1.0814959226709443e-06,
      "loss": 0.1624,
      "step": 8674
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4961563975888477,
      "learning_rate": 1.0805763339010329e-06,
      "loss": 0.2181,
      "step": 8675
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.464494380594849,
      "learning_rate": 1.0796570888907953e-06,
      "loss": 0.1781,
      "step": 8676
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3221977542922965,
      "learning_rate": 1.0787381877208586e-06,
      "loss": 0.1492,
      "step": 8677
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3635585161681474,
      "learning_rate": 1.0778196304718153e-06,
      "loss": 0.1761,
      "step": 8678
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4051282420028737,
      "learning_rate": 1.076901417224227e-06,
      "loss": 0.1765,
      "step": 8679
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3051318935201037,
      "learning_rate": 1.0759835480586312e-06,
      "loss": 0.1899,
      "step": 8680
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3672006881798064,
      "learning_rate": 1.075066023055527e-06,
      "loss": 0.1847,
      "step": 8681
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3929830505248804,
      "learning_rate": 1.074148842295391e-06,
      "loss": 0.1589,
      "step": 8682
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2286533733768332,
      "learning_rate": 1.0732320058586632e-06,
      "loss": 0.1672,
      "step": 8683
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.364513295290568,
      "learning_rate": 1.0723155138257586e-06,
      "loss": 0.1529,
      "step": 8684
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3552993608608124,
      "learning_rate": 1.0713993662770578e-06,
      "loss": 0.1766,
      "step": 8685
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3892798461108342,
      "learning_rate": 1.070483563292915e-06,
      "loss": 0.1882,
      "step": 8686
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3644335359683923,
      "learning_rate": 1.0695681049536515e-06,
      "loss": 0.1737,
      "step": 8687
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3393991375444734,
      "learning_rate": 1.0686529913395572e-06,
      "loss": 0.1499,
      "step": 8688
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.289150604637478,
      "learning_rate": 1.0677382225308969e-06,
      "loss": 0.2061,
      "step": 8689
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3324859429651938,
      "learning_rate": 1.0668237986078988e-06,
      "loss": 0.1495,
      "step": 8690
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.615292136968704,
      "learning_rate": 1.0659097196507672e-06,
      "loss": 0.5905,
      "step": 8691
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5035076231382805,
      "learning_rate": 1.0649959857396697e-06,
      "loss": 0.2428,
      "step": 8692
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4103063084925744,
      "learning_rate": 1.0640825969547498e-06,
      "loss": 0.1981,
      "step": 8693
    },
    {
      "epoch": 0.79,
      "grad_norm": 5.768583222044134,
      "learning_rate": 1.0631695533761165e-06,
      "loss": 0.5826,
      "step": 8694
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3617094232837887,
      "learning_rate": 1.0622568550838485e-06,
      "loss": 0.1855,
      "step": 8695
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3383675797859247,
      "learning_rate": 1.0613445021579983e-06,
      "loss": 0.1493,
      "step": 8696
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3725387648793175,
      "learning_rate": 1.0604324946785826e-06,
      "loss": 0.161,
      "step": 8697
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1326662109933792,
      "learning_rate": 1.059520832725593e-06,
      "loss": 0.112,
      "step": 8698
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3135306276479088,
      "learning_rate": 1.0586095163789862e-06,
      "loss": 0.1602,
      "step": 8699
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3962791019354972,
      "learning_rate": 1.0576985457186928e-06,
      "loss": 0.1673,
      "step": 8700
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3616302354731782,
      "learning_rate": 1.0567879208246084e-06,
      "loss": 0.1775,
      "step": 8701
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3628567122083537,
      "learning_rate": 1.0558776417766043e-06,
      "loss": 0.1532,
      "step": 8702
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.634687423281601,
      "learning_rate": 1.0549677086545161e-06,
      "loss": 0.531,
      "step": 8703
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.688054344722962,
      "learning_rate": 1.0540581215381495e-06,
      "loss": 0.2915,
      "step": 8704
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4892167291940037,
      "learning_rate": 1.0531488805072848e-06,
      "loss": 0.2127,
      "step": 8705
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4694804037168399,
      "learning_rate": 1.0522399856416653e-06,
      "loss": 0.1866,
      "step": 8706
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4720016844164234,
      "learning_rate": 1.0513314370210098e-06,
      "loss": 0.1782,
      "step": 8707
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2792507620149467,
      "learning_rate": 1.0504232347250021e-06,
      "loss": 0.1438,
      "step": 8708
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3264707671876534,
      "learning_rate": 1.0495153788332991e-06,
      "loss": 0.1963,
      "step": 8709
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4658455894479967,
      "learning_rate": 1.0486078694255242e-06,
      "loss": 0.2074,
      "step": 8710
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3861774583750406,
      "learning_rate": 1.0477007065812744e-06,
      "loss": 0.1563,
      "step": 8711
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2205169451629658,
      "learning_rate": 1.0467938903801123e-06,
      "loss": 0.155,
      "step": 8712
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.319568500550792,
      "learning_rate": 1.0458874209015708e-06,
      "loss": 0.1444,
      "step": 8713
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.277474642040767,
      "learning_rate": 1.0449812982251556e-06,
      "loss": 0.123,
      "step": 8714
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.460570518027027,
      "learning_rate": 1.0440755224303379e-06,
      "loss": 0.1822,
      "step": 8715
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7289338243636425,
      "learning_rate": 1.0431700935965616e-06,
      "loss": 0.2284,
      "step": 8716
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2483560620965841,
      "learning_rate": 1.042265011803238e-06,
      "loss": 0.1532,
      "step": 8717
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.522344628939267,
      "learning_rate": 1.0413602771297499e-06,
      "loss": 0.1572,
      "step": 8718
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.324921968507895,
      "learning_rate": 1.0404558896554462e-06,
      "loss": 0.1931,
      "step": 8719
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3846580244028175,
      "learning_rate": 1.0395518494596512e-06,
      "loss": 0.1846,
      "step": 8720
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.445584180776854,
      "learning_rate": 1.0386481566216532e-06,
      "loss": 0.1662,
      "step": 8721
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5226337055821098,
      "learning_rate": 1.0377448112207112e-06,
      "loss": 0.1817,
      "step": 8722
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3216731980665604,
      "learning_rate": 1.0368418133360569e-06,
      "loss": 0.1496,
      "step": 8723
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3962170016603017,
      "learning_rate": 1.0359391630468869e-06,
      "loss": 0.1402,
      "step": 8724
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3040760092974688,
      "learning_rate": 1.0350368604323723e-06,
      "loss": 0.1647,
      "step": 8725
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.265383981173362,
      "learning_rate": 1.0341349055716482e-06,
      "loss": 0.1568,
      "step": 8726
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.548215548928925,
      "learning_rate": 1.0332332985438248e-06,
      "loss": 0.1689,
      "step": 8727
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.305401217519291,
      "learning_rate": 1.0323320394279769e-06,
      "loss": 0.1632,
      "step": 8728
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4486667589257392,
      "learning_rate": 1.0314311283031531e-06,
      "loss": 0.184,
      "step": 8729
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4352283100817633,
      "learning_rate": 1.0305305652483678e-06,
      "loss": 0.1564,
      "step": 8730
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4527190289278338,
      "learning_rate": 1.0296303503426058e-06,
      "loss": 0.1479,
      "step": 8731
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4042822363395229,
      "learning_rate": 1.0287304836648238e-06,
      "loss": 0.1871,
      "step": 8732
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3864359555998078,
      "learning_rate": 1.0278309652939439e-06,
      "loss": 0.1558,
      "step": 8733
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.53681360012889,
      "learning_rate": 1.026931795308863e-06,
      "loss": 0.2091,
      "step": 8734
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4355119610080884,
      "learning_rate": 1.0260329737884406e-06,
      "loss": 0.1921,
      "step": 8735
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.365149487933766,
      "learning_rate": 1.0251345008115132e-06,
      "loss": 0.1756,
      "step": 8736
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3309164842467998,
      "learning_rate": 1.0242363764568808e-06,
      "loss": 0.1186,
      "step": 8737
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2793245098832136,
      "learning_rate": 1.0233386008033136e-06,
      "loss": 0.1547,
      "step": 8738
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4526990019072061,
      "learning_rate": 1.0224411739295552e-06,
      "loss": 0.2026,
      "step": 8739
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4880084773173832,
      "learning_rate": 1.0215440959143137e-06,
      "loss": 0.162,
      "step": 8740
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.435519827686088,
      "learning_rate": 1.020647366836271e-06,
      "loss": 0.1569,
      "step": 8741
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2955166648521264,
      "learning_rate": 1.0197509867740734e-06,
      "loss": 0.1711,
      "step": 8742
    },
    {
      "epoch": 0.8,
      "grad_norm": 11.248741782815587,
      "learning_rate": 1.0188549558063433e-06,
      "loss": 0.6343,
      "step": 8743
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4031938231635122,
      "learning_rate": 1.017959274011665e-06,
      "loss": 0.1753,
      "step": 8744
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3609352641470946,
      "learning_rate": 1.0170639414685985e-06,
      "loss": 0.1644,
      "step": 8745
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.742408143641095,
      "learning_rate": 1.0161689582556688e-06,
      "loss": 0.5333,
      "step": 8746
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4123929572124478,
      "learning_rate": 1.0152743244513712e-06,
      "loss": 0.1917,
      "step": 8747
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.3652577154056775,
      "learning_rate": 1.0143800401341736e-06,
      "loss": 0.4565,
      "step": 8748
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4804113703265662,
      "learning_rate": 1.0134861053825078e-06,
      "loss": 0.1774,
      "step": 8749
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4556326273893325,
      "learning_rate": 1.0125925202747811e-06,
      "loss": 0.1962,
      "step": 8750
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5020203162034451,
      "learning_rate": 1.0116992848893636e-06,
      "loss": 0.1795,
      "step": 8751
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4926649402861516,
      "learning_rate": 1.0108063993046013e-06,
      "loss": 0.1525,
      "step": 8752
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.336732960032424,
      "learning_rate": 1.0099138635988026e-06,
      "loss": 0.1645,
      "step": 8753
    },
    {
      "epoch": 0.8,
      "grad_norm": 6.635853156970372,
      "learning_rate": 1.0090216778502527e-06,
      "loss": 0.5204,
      "step": 8754
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.325819409043826,
      "learning_rate": 1.0081298421371998e-06,
      "loss": 0.1654,
      "step": 8755
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3545674148598426,
      "learning_rate": 1.007238356537863e-06,
      "loss": 0.1711,
      "step": 8756
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3251402929047102,
      "learning_rate": 1.006347221130435e-06,
      "loss": 0.184,
      "step": 8757
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.298454685591963,
      "learning_rate": 1.00545643599307e-06,
      "loss": 0.1503,
      "step": 8758
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.435956480870236,
      "learning_rate": 1.0045660012039e-06,
      "loss": 0.1311,
      "step": 8759
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4288453568494368,
      "learning_rate": 1.0036759168410187e-06,
      "loss": 0.1703,
      "step": 8760
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2858077436681399,
      "learning_rate": 1.0027861829824953e-06,
      "loss": 0.1606,
      "step": 8761
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.38714925693023,
      "learning_rate": 1.0018967997063628e-06,
      "loss": 0.1479,
      "step": 8762
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.304974523847873,
      "learning_rate": 1.0010077670906282e-06,
      "loss": 0.564,
      "step": 8763
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3361387619568754,
      "learning_rate": 1.0001190852132654e-06,
      "loss": 0.1406,
      "step": 8764
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3622272513307634,
      "learning_rate": 9.99230754152215e-07,
      "loss": 0.1628,
      "step": 8765
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.228112243841836,
      "learning_rate": 9.983427739853935e-07,
      "loss": 0.1403,
      "step": 8766
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3640702565880152,
      "learning_rate": 9.974551447906793e-07,
      "loss": 0.1741,
      "step": 8767
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3308383340034904,
      "learning_rate": 9.965678666459262e-07,
      "loss": 0.1734,
      "step": 8768
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1824019603412665,
      "learning_rate": 9.956809396289519e-07,
      "loss": 0.1503,
      "step": 8769
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.361050012043237,
      "learning_rate": 9.94794363817549e-07,
      "loss": 0.1555,
      "step": 8770
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2108367418351393,
      "learning_rate": 9.939081392894724e-07,
      "loss": 0.141,
      "step": 8771
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.387283855696306,
      "learning_rate": 9.930222661224532e-07,
      "loss": 0.1816,
      "step": 8772
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4303884757426255,
      "learning_rate": 9.921367443941866e-07,
      "loss": 0.1752,
      "step": 8773
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.943165459759204,
      "learning_rate": 9.91251574182338e-07,
      "loss": 0.3638,
      "step": 8774
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4922115738898925,
      "learning_rate": 9.903667555645458e-07,
      "loss": 0.2049,
      "step": 8775
    },
    {
      "epoch": 0.8,
      "grad_norm": 10.87982283119492,
      "learning_rate": 9.894822886184108e-07,
      "loss": 0.61,
      "step": 8776
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2334526742945684,
      "learning_rate": 9.885981734215094e-07,
      "loss": 0.1529,
      "step": 8777
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3185404811394392,
      "learning_rate": 9.87714410051383e-07,
      "loss": 0.137,
      "step": 8778
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5849306609921914,
      "learning_rate": 9.868309985855446e-07,
      "loss": 0.2121,
      "step": 8779
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.443535373046627,
      "learning_rate": 9.859479391014753e-07,
      "loss": 0.1623,
      "step": 8780
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2544293310235832,
      "learning_rate": 9.850652316766236e-07,
      "loss": 0.1715,
      "step": 8781
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6204881465694654,
      "learning_rate": 9.84182876388411e-07,
      "loss": 0.1749,
      "step": 8782
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5963719915857717,
      "learning_rate": 9.833008733142224e-07,
      "loss": 0.2233,
      "step": 8783
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2879203274638817,
      "learning_rate": 9.824192225314206e-07,
      "loss": 0.1844,
      "step": 8784
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.34083568856688,
      "learning_rate": 9.815379241173295e-07,
      "loss": 0.1738,
      "step": 8785
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4451031912648393,
      "learning_rate": 9.806569781492436e-07,
      "loss": 0.2202,
      "step": 8786
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2704782370308796,
      "learning_rate": 9.797763847044305e-07,
      "loss": 0.1524,
      "step": 8787
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3242463871888739,
      "learning_rate": 9.788961438601219e-07,
      "loss": 0.2059,
      "step": 8788
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4802946215523896,
      "learning_rate": 9.780162556935224e-07,
      "loss": 0.1723,
      "step": 8789
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3979119284768227,
      "learning_rate": 9.771367202818027e-07,
      "loss": 0.2232,
      "step": 8790
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3897397508388865,
      "learning_rate": 9.762575377021055e-07,
      "loss": 0.1875,
      "step": 8791
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4035555578381944,
      "learning_rate": 9.753787080315385e-07,
      "loss": 0.1615,
      "step": 8792
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4534951968870158,
      "learning_rate": 9.745002313471847e-07,
      "loss": 0.1963,
      "step": 8793
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.268282066890395,
      "learning_rate": 9.7362210772609e-07,
      "loss": 0.1487,
      "step": 8794
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4557825075019897,
      "learning_rate": 9.7274433724527e-07,
      "loss": 0.1698,
      "step": 8795
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3094944045994006,
      "learning_rate": 9.71866919981715e-07,
      "loss": 0.1505,
      "step": 8796
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3334239495280549,
      "learning_rate": 9.709898560123766e-07,
      "loss": 0.1647,
      "step": 8797
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.279830615456434,
      "learning_rate": 9.701131454141821e-07,
      "loss": 0.1706,
      "step": 8798
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5482150184257564,
      "learning_rate": 9.69236788264023e-07,
      "loss": 0.1928,
      "step": 8799
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.223904706732008,
      "learning_rate": 9.683607846387632e-07,
      "loss": 0.1565,
      "step": 8800
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4801262286363592,
      "learning_rate": 9.67485134615232e-07,
      "loss": 0.1533,
      "step": 8801
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5552696065744764,
      "learning_rate": 9.666098382702326e-07,
      "loss": 0.211,
      "step": 8802
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.481328897929841,
      "learning_rate": 9.65734895680533e-07,
      "loss": 0.5428,
      "step": 8803
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.352706977059975,
      "learning_rate": 9.648603069228696e-07,
      "loss": 0.1636,
      "step": 8804
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3574125870833407,
      "learning_rate": 9.639860720739524e-07,
      "loss": 0.2009,
      "step": 8805
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1972904475639952,
      "learning_rate": 9.631121912104557e-07,
      "loss": 0.1606,
      "step": 8806
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.239139943662002,
      "learning_rate": 9.62238664409027e-07,
      "loss": 0.1501,
      "step": 8807
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6444863266502876,
      "learning_rate": 9.613654917462778e-07,
      "loss": 0.2129,
      "step": 8808
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2919020480385608,
      "learning_rate": 9.60492673298794e-07,
      "loss": 0.1486,
      "step": 8809
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5973137213038984,
      "learning_rate": 9.596202091431245e-07,
      "loss": 0.181,
      "step": 8810
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.3476408673337295,
      "learning_rate": 9.58748099355793e-07,
      "loss": 0.606,
      "step": 8811
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3770909787712642,
      "learning_rate": 9.578763440132882e-07,
      "loss": 0.1664,
      "step": 8812
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.341837853482818,
      "learning_rate": 9.570049431920674e-07,
      "loss": 0.2105,
      "step": 8813
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4788931237014495,
      "learning_rate": 9.561338969685614e-07,
      "loss": 0.153,
      "step": 8814
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3985317984325056,
      "learning_rate": 9.552632054191636e-07,
      "loss": 0.1876,
      "step": 8815
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3849779023427782,
      "learning_rate": 9.543928686202425e-07,
      "loss": 0.1658,
      "step": 8816
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3499324516288176,
      "learning_rate": 9.535228866481295e-07,
      "loss": 0.1796,
      "step": 8817
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.42907503813758,
      "learning_rate": 9.526532595791305e-07,
      "loss": 0.1806,
      "step": 8818
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5538558296192964,
      "learning_rate": 9.517839874895151e-07,
      "loss": 0.2178,
      "step": 8819
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3283141796021614,
      "learning_rate": 9.50915070455527e-07,
      "loss": 0.1521,
      "step": 8820
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4660950947059783,
      "learning_rate": 9.50046508553375e-07,
      "loss": 0.2059,
      "step": 8821
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4017227012834321,
      "learning_rate": 9.491783018592354e-07,
      "loss": 0.1904,
      "step": 8822
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.347209471207204,
      "learning_rate": 9.483104504492596e-07,
      "loss": 0.176,
      "step": 8823
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2822716891865422,
      "learning_rate": 9.474429543995606e-07,
      "loss": 0.1551,
      "step": 8824
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2266239828836127,
      "learning_rate": 9.465758137862264e-07,
      "loss": 0.1647,
      "step": 8825
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5788982814752694,
      "learning_rate": 9.457090286853082e-07,
      "loss": 0.167,
      "step": 8826
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.449491723573011,
      "learning_rate": 9.448425991728322e-07,
      "loss": 0.161,
      "step": 8827
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5591758719065867,
      "learning_rate": 9.439765253247879e-07,
      "loss": 0.1852,
      "step": 8828
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3999977845582043,
      "learning_rate": 9.431108072171346e-07,
      "loss": 0.1972,
      "step": 8829
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4212557879746353,
      "learning_rate": 9.42245444925805e-07,
      "loss": 0.1514,
      "step": 8830
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3671163081291071,
      "learning_rate": 9.41380438526694e-07,
      "loss": 0.1818,
      "step": 8831
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4031562863077602,
      "learning_rate": 9.405157880956706e-07,
      "loss": 0.1946,
      "step": 8832
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.322450633114087,
      "learning_rate": 9.396514937085682e-07,
      "loss": 0.1623,
      "step": 8833
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3067316345984181,
      "learning_rate": 9.38787555441194e-07,
      "loss": 0.164,
      "step": 8834
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3954231025533708,
      "learning_rate": 9.379239733693185e-07,
      "loss": 0.193,
      "step": 8835
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2860365666829376,
      "learning_rate": 9.370607475686866e-07,
      "loss": 0.1533,
      "step": 8836
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3210049788115037,
      "learning_rate": 9.361978781150066e-07,
      "loss": 0.1516,
      "step": 8837
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.350179997411609,
      "learning_rate": 9.353353650839575e-07,
      "loss": 0.1894,
      "step": 8838
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4150411581056865,
      "learning_rate": 9.344732085511892e-07,
      "loss": 0.1856,
      "step": 8839
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4511694098203032,
      "learning_rate": 9.336114085923176e-07,
      "loss": 0.1832,
      "step": 8840
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4125470409609995,
      "learning_rate": 9.327499652829292e-07,
      "loss": 0.1583,
      "step": 8841
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3938267814262786,
      "learning_rate": 9.318888786985764e-07,
      "loss": 0.2177,
      "step": 8842
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3525694812892692,
      "learning_rate": 9.310281489147854e-07,
      "loss": 0.1902,
      "step": 8843
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4776920616749876,
      "learning_rate": 9.301677760070449e-07,
      "loss": 0.1739,
      "step": 8844
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.400814252807214,
      "learning_rate": 9.293077600508177e-07,
      "loss": 0.1837,
      "step": 8845
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1525573751405913,
      "learning_rate": 9.284481011215318e-07,
      "loss": 0.1565,
      "step": 8846
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2657063541916456,
      "learning_rate": 9.275887992945837e-07,
      "loss": 0.1359,
      "step": 8847
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5182669496065857,
      "learning_rate": 9.267298546453429e-07,
      "loss": 0.2172,
      "step": 8848
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.920470664742591,
      "learning_rate": 9.258712672491416e-07,
      "loss": 0.6797,
      "step": 8849
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.476294422145195,
      "learning_rate": 9.250130371812865e-07,
      "loss": 0.1927,
      "step": 8850
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3182699803065963,
      "learning_rate": 9.241551645170471e-07,
      "loss": 0.1578,
      "step": 8851
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.365492542115594,
      "learning_rate": 9.232976493316681e-07,
      "loss": 0.5342,
      "step": 8852
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4465878819208617,
      "learning_rate": 9.22440491700356e-07,
      "loss": 0.1881,
      "step": 8853
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2336452220968805,
      "learning_rate": 9.215836916982912e-07,
      "loss": 0.163,
      "step": 8854
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2575655573730804,
      "learning_rate": 9.207272494006208e-07,
      "loss": 0.1413,
      "step": 8855
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3465205632500477,
      "learning_rate": 9.198711648824588e-07,
      "loss": 0.1739,
      "step": 8856
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6614980365149858,
      "learning_rate": 9.190154382188921e-07,
      "loss": 0.1791,
      "step": 8857
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.338585303784745,
      "learning_rate": 9.181600694849707e-07,
      "loss": 0.1792,
      "step": 8858
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3000179088930464,
      "learning_rate": 9.173050587557192e-07,
      "loss": 0.1657,
      "step": 8859
    },
    {
      "epoch": 0.81,
      "grad_norm": 10.902443668850019,
      "learning_rate": 9.164504061061247e-07,
      "loss": 0.5633,
      "step": 8860
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5704247991225166,
      "learning_rate": 9.155961116111489e-07,
      "loss": 0.1806,
      "step": 8861
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.553105294859498,
      "learning_rate": 9.147421753457164e-07,
      "loss": 0.7088,
      "step": 8862
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2698791898313453,
      "learning_rate": 9.138885973847262e-07,
      "loss": 0.1802,
      "step": 8863
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.479308239700672,
      "learning_rate": 9.13035377803041e-07,
      "loss": 0.1558,
      "step": 8864
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3475095257816827,
      "learning_rate": 9.121825166754927e-07,
      "loss": 0.1668,
      "step": 8865
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.441924648831618,
      "learning_rate": 9.11330014076885e-07,
      "loss": 0.1752,
      "step": 8866
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.406771305904077,
      "learning_rate": 9.104778700819866e-07,
      "loss": 0.1968,
      "step": 8867
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.554440529817756,
      "learning_rate": 9.096260847655375e-07,
      "loss": 0.45,
      "step": 8868
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.423338136383676,
      "learning_rate": 9.087746582022439e-07,
      "loss": 0.1352,
      "step": 8869
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3230465779732505,
      "learning_rate": 9.079235904667826e-07,
      "loss": 0.166,
      "step": 8870
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4775020948122735,
      "learning_rate": 9.070728816337982e-07,
      "loss": 0.1934,
      "step": 8871
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5055056499154316,
      "learning_rate": 9.062225317779011e-07,
      "loss": 0.1855,
      "step": 8872
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.387757324032783,
      "learning_rate": 9.053725409736752e-07,
      "loss": 0.518,
      "step": 8873
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1537197612046668,
      "learning_rate": 9.045229092956687e-07,
      "loss": 0.1255,
      "step": 8874
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.414717630981344,
      "learning_rate": 9.036736368184024e-07,
      "loss": 0.565,
      "step": 8875
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5322580458785373,
      "learning_rate": 9.028247236163596e-07,
      "loss": 0.1874,
      "step": 8876
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2891605436858316,
      "learning_rate": 9.019761697639995e-07,
      "loss": 0.1607,
      "step": 8877
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.209985748708505,
      "learning_rate": 9.011279753357427e-07,
      "loss": 0.1642,
      "step": 8878
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4158382552378599,
      "learning_rate": 9.002801404059841e-07,
      "loss": 0.1729,
      "step": 8879
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.48763874193533,
      "learning_rate": 8.994326650490832e-07,
      "loss": 0.2117,
      "step": 8880
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4765342842761295,
      "learning_rate": 8.98585549339368e-07,
      "loss": 0.1977,
      "step": 8881
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1965395495123106,
      "learning_rate": 8.977387933511389e-07,
      "loss": 0.151,
      "step": 8882
    },
    {
      "epoch": 0.81,
      "grad_norm": 8.584335393207382,
      "learning_rate": 8.968923971586596e-07,
      "loss": 0.5923,
      "step": 8883
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5506029787183235,
      "learning_rate": 8.960463608361674e-07,
      "loss": 0.1846,
      "step": 8884
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5137087400442757,
      "learning_rate": 8.95200684457862e-07,
      "loss": 0.1739,
      "step": 8885
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.408131065227428,
      "learning_rate": 8.943553680979183e-07,
      "loss": 0.1568,
      "step": 8886
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3770703732720906,
      "learning_rate": 8.935104118304733e-07,
      "loss": 0.1608,
      "step": 8887
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.390827795372674,
      "learning_rate": 8.92665815729638e-07,
      "loss": 0.213,
      "step": 8888
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.998910497964643,
      "learning_rate": 8.918215798694879e-07,
      "loss": 0.4911,
      "step": 8889
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.538232223609781,
      "learning_rate": 8.909777043240664e-07,
      "loss": 0.1927,
      "step": 8890
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2995585827615164,
      "learning_rate": 8.901341891673898e-07,
      "loss": 0.1998,
      "step": 8891
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.456021257718599,
      "learning_rate": 8.892910344734373e-07,
      "loss": 0.2096,
      "step": 8892
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3535367473318907,
      "learning_rate": 8.884482403161626e-07,
      "loss": 0.1703,
      "step": 8893
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3683924999970645,
      "learning_rate": 8.876058067694804e-07,
      "loss": 0.1703,
      "step": 8894
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.187668680937316,
      "learning_rate": 8.867637339072821e-07,
      "loss": 0.58,
      "step": 8895
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5271747119540238,
      "learning_rate": 8.859220218034192e-07,
      "loss": 0.1786,
      "step": 8896
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3938811282651236,
      "learning_rate": 8.850806705317183e-07,
      "loss": 0.2088,
      "step": 8897
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2827335774351847,
      "learning_rate": 8.842396801659703e-07,
      "loss": 0.1759,
      "step": 8898
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5330880799573527,
      "learning_rate": 8.83399050779935e-07,
      "loss": 0.1993,
      "step": 8899
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.711177837932409,
      "learning_rate": 8.825587824473431e-07,
      "loss": 0.1399,
      "step": 8900
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4048160336193,
      "learning_rate": 8.817188752418892e-07,
      "loss": 0.1867,
      "step": 8901
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3281966015342235,
      "learning_rate": 8.808793292372414e-07,
      "loss": 0.1498,
      "step": 8902
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4794237663945569,
      "learning_rate": 8.800401445070311e-07,
      "loss": 0.1788,
      "step": 8903
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2933490585986833,
      "learning_rate": 8.792013211248635e-07,
      "loss": 0.1293,
      "step": 8904
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.098482749125733,
      "learning_rate": 8.783628591643056e-07,
      "loss": 0.1055,
      "step": 8905
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4252161064401012,
      "learning_rate": 8.775247586988988e-07,
      "loss": 0.1759,
      "step": 8906
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2027305191387423,
      "learning_rate": 8.766870198021493e-07,
      "loss": 0.1373,
      "step": 8907
    },
    {
      "epoch": 0.81,
      "grad_norm": 6.371362860283578,
      "learning_rate": 8.758496425475305e-07,
      "loss": 0.6986,
      "step": 8908
    },
    {
      "epoch": 0.81,
      "grad_norm": 5.618988014383454,
      "learning_rate": 8.750126270084891e-07,
      "loss": 0.5059,
      "step": 8909
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.802252641763886,
      "learning_rate": 8.741759732584343e-07,
      "loss": 0.4404,
      "step": 8910
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3479386570142942,
      "learning_rate": 8.733396813707485e-07,
      "loss": 0.1898,
      "step": 8911
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3627225491102202,
      "learning_rate": 8.725037514187778e-07,
      "loss": 0.1741,
      "step": 8912
    },
    {
      "epoch": 0.81,
      "grad_norm": 7.333236828822463,
      "learning_rate": 8.716681834758411e-07,
      "loss": 0.4695,
      "step": 8913
    },
    {
      "epoch": 0.81,
      "grad_norm": 9.186774454601299,
      "learning_rate": 8.708329776152224e-07,
      "loss": 0.5462,
      "step": 8914
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.323773141637584,
      "learning_rate": 8.699981339101738e-07,
      "loss": 0.5537,
      "step": 8915
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3420853368582275,
      "learning_rate": 8.691636524339181e-07,
      "loss": 0.1729,
      "step": 8916
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4537385921567587,
      "learning_rate": 8.683295332596437e-07,
      "loss": 0.2252,
      "step": 8917
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.660135757635199,
      "learning_rate": 8.674957764605102e-07,
      "loss": 0.5354,
      "step": 8918
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3718391043290026,
      "learning_rate": 8.666623821096409e-07,
      "loss": 0.1676,
      "step": 8919
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3159421465558725,
      "learning_rate": 8.658293502801335e-07,
      "loss": 0.1859,
      "step": 8920
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.117455345018023,
      "learning_rate": 8.649966810450472e-07,
      "loss": 0.6871,
      "step": 8921
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1705865290344286,
      "learning_rate": 8.641643744774159e-07,
      "loss": 0.1513,
      "step": 8922
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1766000268114865,
      "learning_rate": 8.633324306502361e-07,
      "loss": 0.1487,
      "step": 8923
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1209107196134067,
      "learning_rate": 8.625008496364745e-07,
      "loss": 0.1305,
      "step": 8924
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1989889811060932,
      "learning_rate": 8.616696315090684e-07,
      "loss": 0.151,
      "step": 8925
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2799985480684621,
      "learning_rate": 8.608387763409187e-07,
      "loss": 0.1784,
      "step": 8926
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.260481000391451,
      "learning_rate": 8.600082842048996e-07,
      "loss": 0.1477,
      "step": 8927
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2784800828471055,
      "learning_rate": 8.591781551738481e-07,
      "loss": 0.2078,
      "step": 8928
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.506918049902612,
      "learning_rate": 8.583483893205746e-07,
      "loss": 0.4796,
      "step": 8929
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.961591252061192,
      "learning_rate": 8.575189867178535e-07,
      "loss": 0.6353,
      "step": 8930
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3426992269681044,
      "learning_rate": 8.566899474384299e-07,
      "loss": 0.1966,
      "step": 8931
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.377468993641758,
      "learning_rate": 8.558612715550163e-07,
      "loss": 0.1488,
      "step": 8932
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.33390062888924,
      "learning_rate": 8.550329591402906e-07,
      "loss": 0.1731,
      "step": 8933
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3552549039991593,
      "learning_rate": 8.542050102669047e-07,
      "loss": 0.1569,
      "step": 8934
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.561154982194475,
      "learning_rate": 8.533774250074727e-07,
      "loss": 0.4817,
      "step": 8935
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4802038771764467,
      "learning_rate": 8.525502034345817e-07,
      "loss": 0.203,
      "step": 8936
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4740472576525054,
      "learning_rate": 8.517233456207819e-07,
      "loss": 0.2362,
      "step": 8937
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.5908163500356265,
      "learning_rate": 8.508968516385968e-07,
      "loss": 0.4925,
      "step": 8938
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3873546644281058,
      "learning_rate": 8.50070721560513e-07,
      "loss": 0.1804,
      "step": 8939
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4188611385047187,
      "learning_rate": 8.492449554589904e-07,
      "loss": 0.1495,
      "step": 8940
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.401919492725426,
      "learning_rate": 8.484195534064527e-07,
      "loss": 0.1654,
      "step": 8941
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3995077449752644,
      "learning_rate": 8.475945154752924e-07,
      "loss": 0.1551,
      "step": 8942
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2036711171688914,
      "learning_rate": 8.467698417378723e-07,
      "loss": 0.163,
      "step": 8943
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3263715116982693,
      "learning_rate": 8.4594553226652e-07,
      "loss": 0.1844,
      "step": 8944
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2064639053240362,
      "learning_rate": 8.451215871335355e-07,
      "loss": 0.1667,
      "step": 8945
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.351517445514625,
      "learning_rate": 8.442980064111816e-07,
      "loss": 0.5493,
      "step": 8946
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4323915640643263,
      "learning_rate": 8.434747901716949e-07,
      "loss": 0.1938,
      "step": 8947
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3874257562942638,
      "learning_rate": 8.426519384872733e-07,
      "loss": 0.1633,
      "step": 8948
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.337852319079808,
      "learning_rate": 8.418294514300901e-07,
      "loss": 0.1776,
      "step": 8949
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2657426861196865,
      "learning_rate": 8.410073290722809e-07,
      "loss": 0.1597,
      "step": 8950
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2733095147189595,
      "learning_rate": 8.401855714859508e-07,
      "loss": 0.1257,
      "step": 8951
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6291071349387765,
      "learning_rate": 8.39364178743175e-07,
      "loss": 0.1816,
      "step": 8952
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2430387602857509,
      "learning_rate": 8.38543150915993e-07,
      "loss": 0.1465,
      "step": 8953
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4404675525683865,
      "learning_rate": 8.377224880764173e-07,
      "loss": 0.1949,
      "step": 8954
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2203424018317235,
      "learning_rate": 8.369021902964231e-07,
      "loss": 0.1564,
      "step": 8955
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2760255012093298,
      "learning_rate": 8.360822576479577e-07,
      "loss": 0.1618,
      "step": 8956
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.4690902208137695,
      "learning_rate": 8.352626902029337e-07,
      "loss": 0.433,
      "step": 8957
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4488706022563322,
      "learning_rate": 8.344434880332325e-07,
      "loss": 0.1912,
      "step": 8958
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4279770397541849,
      "learning_rate": 8.336246512107043e-07,
      "loss": 0.1989,
      "step": 8959
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.276301920152674,
      "learning_rate": 8.328061798071652e-07,
      "loss": 0.1695,
      "step": 8960
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4411414283000847,
      "learning_rate": 8.31988073894403e-07,
      "loss": 0.1699,
      "step": 8961
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5023348385832687,
      "learning_rate": 8.311703335441689e-07,
      "loss": 0.1926,
      "step": 8962
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3369801157331098,
      "learning_rate": 8.303529588281856e-07,
      "loss": 0.1578,
      "step": 8963
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4556678837402603,
      "learning_rate": 8.29535949818141e-07,
      "loss": 0.1682,
      "step": 8964
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.8607862205486425,
      "learning_rate": 8.287193065856936e-07,
      "loss": 0.6691,
      "step": 8965
    },
    {
      "epoch": 0.82,
      "grad_norm": 9.516600162825524,
      "learning_rate": 8.279030292024686e-07,
      "loss": 0.4773,
      "step": 8966
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3200203313624332,
      "learning_rate": 8.270871177400563e-07,
      "loss": 0.1606,
      "step": 8967
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6419087139542128,
      "learning_rate": 8.262715722700215e-07,
      "loss": 0.208,
      "step": 8968
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.355596532076566,
      "learning_rate": 8.254563928638892e-07,
      "loss": 0.1717,
      "step": 8969
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4164416290064634,
      "learning_rate": 8.246415795931595e-07,
      "loss": 0.1715,
      "step": 8970
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.785005287481494,
      "learning_rate": 8.238271325292941e-07,
      "loss": 0.5976,
      "step": 8971
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.172249004193302,
      "learning_rate": 8.230130517437279e-07,
      "loss": 0.164,
      "step": 8972
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2064373851128412,
      "learning_rate": 8.221993373078591e-07,
      "loss": 0.169,
      "step": 8973
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7307096161338198,
      "learning_rate": 8.213859892930581e-07,
      "loss": 0.2599,
      "step": 8974
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2459737958124357,
      "learning_rate": 8.205730077706592e-07,
      "loss": 0.1639,
      "step": 8975
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6072682843358421,
      "learning_rate": 8.197603928119663e-07,
      "loss": 0.1529,
      "step": 8976
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.455953737376644,
      "learning_rate": 8.189481444882524e-07,
      "loss": 0.4967,
      "step": 8977
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2491306708528245,
      "learning_rate": 8.181362628707557e-07,
      "loss": 0.1679,
      "step": 8978
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.748962499217604,
      "learning_rate": 8.173247480306857e-07,
      "loss": 0.5341,
      "step": 8979
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4588208693692606,
      "learning_rate": 8.165136000392149e-07,
      "loss": 0.1798,
      "step": 8980
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3965624139804895,
      "learning_rate": 8.157028189674892e-07,
      "loss": 0.1843,
      "step": 8981
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3784422211811695,
      "learning_rate": 8.148924048866169e-07,
      "loss": 0.1754,
      "step": 8982
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3243598639535463,
      "learning_rate": 8.140823578676793e-07,
      "loss": 0.1281,
      "step": 8983
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6206975112051663,
      "learning_rate": 8.132726779817218e-07,
      "loss": 0.1504,
      "step": 8984
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.631077618003463,
      "learning_rate": 8.124633652997571e-07,
      "loss": 0.5537,
      "step": 8985
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5096670134064696,
      "learning_rate": 8.1165441989277e-07,
      "loss": 0.1945,
      "step": 8986
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.301151071977981,
      "learning_rate": 8.108458418317089e-07,
      "loss": 0.1679,
      "step": 8987
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3990634091743952,
      "learning_rate": 8.100376311874924e-07,
      "loss": 0.1912,
      "step": 8988
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3962474884806668,
      "learning_rate": 8.09229788031004e-07,
      "loss": 0.1875,
      "step": 8989
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5972440352331017,
      "learning_rate": 8.084223124331003e-07,
      "loss": 0.2256,
      "step": 8990
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3868509478475053,
      "learning_rate": 8.076152044646002e-07,
      "loss": 0.1914,
      "step": 8991
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2192866117342493,
      "learning_rate": 8.068084641962915e-07,
      "loss": 0.1727,
      "step": 8992
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3373378495663089,
      "learning_rate": 8.060020916989331e-07,
      "loss": 0.1518,
      "step": 8993
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.704209029032557,
      "learning_rate": 8.051960870432468e-07,
      "loss": 0.5749,
      "step": 8994
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.376852513720436,
      "learning_rate": 8.043904502999272e-07,
      "loss": 0.1924,
      "step": 8995
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3722319664349358,
      "learning_rate": 8.035851815396317e-07,
      "loss": 0.1594,
      "step": 8996
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4080348617769778,
      "learning_rate": 8.027802808329899e-07,
      "loss": 0.218,
      "step": 8997
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3129408415608632,
      "learning_rate": 8.019757482505946e-07,
      "loss": 0.1704,
      "step": 8998
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.416397787664698,
      "learning_rate": 8.011715838630107e-07,
      "loss": 0.1997,
      "step": 8999
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.42407598369629,
      "learning_rate": 8.00367787740769e-07,
      "loss": 0.1421,
      "step": 9000
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.388895111672584,
      "learning_rate": 7.995643599543645e-07,
      "loss": 0.1954,
      "step": 9001
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3545882461134464,
      "learning_rate": 7.987613005742678e-07,
      "loss": 0.1671,
      "step": 9002
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4695134706316721,
      "learning_rate": 7.979586096709086e-07,
      "loss": 0.1839,
      "step": 9003
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4549391415967292,
      "learning_rate": 7.97156287314691e-07,
      "loss": 0.2003,
      "step": 9004
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.409111359610834,
      "learning_rate": 7.963543335759821e-07,
      "loss": 0.1997,
      "step": 9005
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3450742863222644,
      "learning_rate": 7.955527485251203e-07,
      "loss": 0.1864,
      "step": 9006
    },
    {
      "epoch": 0.82,
      "grad_norm": 11.497764987982679,
      "learning_rate": 7.947515322324084e-07,
      "loss": 0.5402,
      "step": 9007
    },
    {
      "epoch": 0.82,
      "grad_norm": 7.796153499388182,
      "learning_rate": 7.939506847681206e-07,
      "loss": 0.4836,
      "step": 9008
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5498460709450583,
      "learning_rate": 7.931502062024949e-07,
      "loss": 0.1763,
      "step": 9009
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4505084332263785,
      "learning_rate": 7.923500966057374e-07,
      "loss": 0.1509,
      "step": 9010
    },
    {
      "epoch": 0.82,
      "grad_norm": 16.08981486445869,
      "learning_rate": 7.915503560480259e-07,
      "loss": 0.6193,
      "step": 9011
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3628229118906925,
      "learning_rate": 7.90750984599501e-07,
      "loss": 0.1689,
      "step": 9012
    },
    {
      "epoch": 0.82,
      "grad_norm": 6.070311346496922,
      "learning_rate": 7.899519823302743e-07,
      "loss": 0.4964,
      "step": 9013
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3890919254682939,
      "learning_rate": 7.89153349310422e-07,
      "loss": 0.1866,
      "step": 9014
    },
    {
      "epoch": 0.82,
      "grad_norm": 13.715165449106163,
      "learning_rate": 7.883550856099914e-07,
      "loss": 0.4671,
      "step": 9015
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5113117325367393,
      "learning_rate": 7.875571912989938e-07,
      "loss": 0.187,
      "step": 9016
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3468120655818572,
      "learning_rate": 7.86759666447412e-07,
      "loss": 0.2099,
      "step": 9017
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.6202019971921064,
      "learning_rate": 7.859625111251929e-07,
      "loss": 0.6586,
      "step": 9018
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.276147822545171,
      "learning_rate": 7.85165725402251e-07,
      "loss": 0.1684,
      "step": 9019
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4236677600213818,
      "learning_rate": 7.843693093484728e-07,
      "loss": 0.1634,
      "step": 9020
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.351178763115741,
      "learning_rate": 7.835732630337062e-07,
      "loss": 0.1728,
      "step": 9021
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3359163425692377,
      "learning_rate": 7.827775865277732e-07,
      "loss": 0.1726,
      "step": 9022
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2999820167747158,
      "learning_rate": 7.819822799004561e-07,
      "loss": 0.1806,
      "step": 9023
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.407824465497688,
      "learning_rate": 7.811873432215117e-07,
      "loss": 0.2087,
      "step": 9024
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.43595347052484,
      "learning_rate": 7.803927765606595e-07,
      "loss": 0.1593,
      "step": 9025
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4771116090114758,
      "learning_rate": 7.7959857998759e-07,
      "loss": 0.1947,
      "step": 9026
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3914771723290211,
      "learning_rate": 7.788047535719579e-07,
      "loss": 0.2154,
      "step": 9027
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2653454883031114,
      "learning_rate": 7.78011297383387e-07,
      "loss": 0.1935,
      "step": 9028
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4011844352135845,
      "learning_rate": 7.772182114914706e-07,
      "loss": 0.172,
      "step": 9029
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4870742368171757,
      "learning_rate": 7.764254959657652e-07,
      "loss": 0.1788,
      "step": 9030
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3393592249409119,
      "learning_rate": 7.756331508757997e-07,
      "loss": 0.1444,
      "step": 9031
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3982851498870879,
      "learning_rate": 7.748411762910657e-07,
      "loss": 0.1857,
      "step": 9032
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3370925183891291,
      "learning_rate": 7.740495722810271e-07,
      "loss": 0.1867,
      "step": 9033
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5191832694668403,
      "learning_rate": 7.73258338915111e-07,
      "loss": 0.1813,
      "step": 9034
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2753472998674316,
      "learning_rate": 7.724674762627138e-07,
      "loss": 0.1621,
      "step": 9035
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.457256357284098,
      "learning_rate": 7.716769843932015e-07,
      "loss": 0.234,
      "step": 9036
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.248095557495265,
      "learning_rate": 7.708868633759025e-07,
      "loss": 0.1469,
      "step": 9037
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1766476279643348,
      "learning_rate": 7.700971132801188e-07,
      "loss": 0.1663,
      "step": 9038
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3255685178212842,
      "learning_rate": 7.693077341751138e-07,
      "loss": 0.1735,
      "step": 9039
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5194546077095896,
      "learning_rate": 7.685187261301242e-07,
      "loss": 0.1942,
      "step": 9040
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2643015182581228,
      "learning_rate": 7.677300892143485e-07,
      "loss": 0.1767,
      "step": 9041
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.765110942448002,
      "learning_rate": 7.669418234969584e-07,
      "loss": 0.5585,
      "step": 9042
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3183551211509175,
      "learning_rate": 7.661539290470882e-07,
      "loss": 0.156,
      "step": 9043
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4240606045599873,
      "learning_rate": 7.653664059338406e-07,
      "loss": 0.1881,
      "step": 9044
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.380847209060617,
      "learning_rate": 7.645792542262892e-07,
      "loss": 0.1529,
      "step": 9045
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5716658289807652,
      "learning_rate": 7.637924739934699e-07,
      "loss": 0.1916,
      "step": 9046
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3818517948012239,
      "learning_rate": 7.630060653043914e-07,
      "loss": 0.1737,
      "step": 9047
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.104425672687015,
      "learning_rate": 7.622200282280234e-07,
      "loss": 0.3798,
      "step": 9048
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3238515398499728,
      "learning_rate": 7.614343628333104e-07,
      "loss": 0.1716,
      "step": 9049
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6517436760837707,
      "learning_rate": 7.606490691891577e-07,
      "loss": 0.2073,
      "step": 9050
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6431464508295226,
      "learning_rate": 7.59864147364443e-07,
      "loss": 0.2103,
      "step": 9051
    },
    {
      "epoch": 0.83,
      "grad_norm": 9.091194435929,
      "learning_rate": 7.590795974280079e-07,
      "loss": 0.5222,
      "step": 9052
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4270353764732013,
      "learning_rate": 7.582954194486614e-07,
      "loss": 0.1818,
      "step": 9053
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.184925815885643,
      "learning_rate": 7.575116134951843e-07,
      "loss": 0.1665,
      "step": 9054
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.760686605043155,
      "learning_rate": 7.567281796363185e-07,
      "loss": 0.654,
      "step": 9055
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.9179356104584935,
      "learning_rate": 7.55945117940779e-07,
      "loss": 0.5969,
      "step": 9056
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4714930375627202,
      "learning_rate": 7.55162428477243e-07,
      "loss": 0.2178,
      "step": 9057
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.963897921401916,
      "learning_rate": 7.543801113143601e-07,
      "loss": 0.4121,
      "step": 9058
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.844370074692142,
      "learning_rate": 7.535981665207426e-07,
      "loss": 0.3346,
      "step": 9059
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5086587614412246,
      "learning_rate": 7.528165941649746e-07,
      "loss": 0.1561,
      "step": 9060
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3470705176837312,
      "learning_rate": 7.520353943156039e-07,
      "loss": 0.1958,
      "step": 9061
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.391507255504773,
      "learning_rate": 7.512545670411453e-07,
      "loss": 0.1565,
      "step": 9062
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7154184106969366,
      "learning_rate": 7.504741124100862e-07,
      "loss": 0.1748,
      "step": 9063
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4899262901567858,
      "learning_rate": 7.49694030490874e-07,
      "loss": 0.1827,
      "step": 9064
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.00742000805336,
      "learning_rate": 7.489143213519301e-07,
      "loss": 0.5946,
      "step": 9065
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3650392785622747,
      "learning_rate": 7.481349850616376e-07,
      "loss": 0.1385,
      "step": 9066
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3297513427021486,
      "learning_rate": 7.473560216883524e-07,
      "loss": 0.1743,
      "step": 9067
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3869371680984706,
      "learning_rate": 7.46577431300392e-07,
      "loss": 0.1928,
      "step": 9068
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3251831904281441,
      "learning_rate": 7.457992139660464e-07,
      "loss": 0.1803,
      "step": 9069
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4541532972162396,
      "learning_rate": 7.450213697535696e-07,
      "loss": 0.1872,
      "step": 9070
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4065102624766797,
      "learning_rate": 7.442438987311817e-07,
      "loss": 0.1734,
      "step": 9071
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3978091189469157,
      "learning_rate": 7.434668009670753e-07,
      "loss": 0.1914,
      "step": 9072
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4635926619802107,
      "learning_rate": 7.426900765294043e-07,
      "loss": 0.1993,
      "step": 9073
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.26564737010364,
      "learning_rate": 7.419137254862957e-07,
      "loss": 0.1633,
      "step": 9074
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4088935472725597,
      "learning_rate": 7.411377479058374e-07,
      "loss": 0.1963,
      "step": 9075
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3735024766344834,
      "learning_rate": 7.403621438560904e-07,
      "loss": 0.1476,
      "step": 9076
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6030249840279338,
      "learning_rate": 7.395869134050793e-07,
      "loss": 0.1944,
      "step": 9077
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.372910293941756,
      "learning_rate": 7.388120566207957e-07,
      "loss": 0.2023,
      "step": 9078
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5762653964801754,
      "learning_rate": 7.380375735712025e-07,
      "loss": 0.152,
      "step": 9079
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4476742511895064,
      "learning_rate": 7.372634643242244e-07,
      "loss": 0.1731,
      "step": 9080
    },
    {
      "epoch": 0.83,
      "grad_norm": 13.203294607480737,
      "learning_rate": 7.364897289477585e-07,
      "loss": 0.5036,
      "step": 9081
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3624393731956383,
      "learning_rate": 7.35716367509664e-07,
      "loss": 0.1872,
      "step": 9082
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3609250264197355,
      "learning_rate": 7.349433800777722e-07,
      "loss": 0.1624,
      "step": 9083
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3529643222182435,
      "learning_rate": 7.341707667198766e-07,
      "loss": 0.1766,
      "step": 9084
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.278169671034989,
      "learning_rate": 7.333985275037436e-07,
      "loss": 0.1388,
      "step": 9085
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5242508812432016,
      "learning_rate": 7.326266624971024e-07,
      "loss": 0.2114,
      "step": 9086
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2050582187966643,
      "learning_rate": 7.318551717676492e-07,
      "loss": 0.1587,
      "step": 9087
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.354019817508874,
      "learning_rate": 7.310840553830517e-07,
      "loss": 0.1658,
      "step": 9088
    },
    {
      "epoch": 0.83,
      "grad_norm": 4.7949148350099025,
      "learning_rate": 7.303133134109391e-07,
      "loss": 0.4871,
      "step": 9089
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3796786201629143,
      "learning_rate": 7.295429459189135e-07,
      "loss": 0.1819,
      "step": 9090
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5062396611219568,
      "learning_rate": 7.287729529745386e-07,
      "loss": 0.1706,
      "step": 9091
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2362121613746118,
      "learning_rate": 7.280033346453503e-07,
      "loss": 0.1454,
      "step": 9092
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4198766354713048,
      "learning_rate": 7.272340909988473e-07,
      "loss": 0.127,
      "step": 9093
    },
    {
      "epoch": 0.83,
      "grad_norm": 8.147875400693064,
      "learning_rate": 7.264652221024992e-07,
      "loss": 0.6662,
      "step": 9094
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.556634999495305,
      "learning_rate": 7.256967280237398e-07,
      "loss": 0.5665,
      "step": 9095
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2786260060347476,
      "learning_rate": 7.249286088299706e-07,
      "loss": 0.1567,
      "step": 9096
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.213003219186564,
      "learning_rate": 7.241608645885629e-07,
      "loss": 0.4357,
      "step": 9097
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.645258631403885,
      "learning_rate": 7.233934953668503e-07,
      "loss": 0.5962,
      "step": 9098
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.871904150219692,
      "learning_rate": 7.226265012321388e-07,
      "loss": 0.6579,
      "step": 9099
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.045778166594505,
      "learning_rate": 7.218598822516965e-07,
      "loss": 0.7008,
      "step": 9100
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5338762170525666,
      "learning_rate": 7.210936384927631e-07,
      "loss": 0.182,
      "step": 9101
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2582670385949455,
      "learning_rate": 7.203277700225419e-07,
      "loss": 0.1845,
      "step": 9102
    },
    {
      "epoch": 0.83,
      "grad_norm": 7.452330968931774,
      "learning_rate": 7.195622769082061e-07,
      "loss": 0.5043,
      "step": 9103
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.563936373736288,
      "learning_rate": 7.187971592168936e-07,
      "loss": 0.2049,
      "step": 9104
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.46739099968578,
      "learning_rate": 7.180324170157094e-07,
      "loss": 0.1744,
      "step": 9105
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.427444783575,
      "learning_rate": 7.172680503717289e-07,
      "loss": 0.2124,
      "step": 9106
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4818648594259616,
      "learning_rate": 7.165040593519896e-07,
      "loss": 0.1841,
      "step": 9107
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3087915253527795,
      "learning_rate": 7.157404440235006e-07,
      "loss": 0.2024,
      "step": 9108
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3155584556200395,
      "learning_rate": 7.149772044532343e-07,
      "loss": 0.1656,
      "step": 9109
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.085677691076392,
      "learning_rate": 7.142143407081348e-07,
      "loss": 0.5025,
      "step": 9110
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1188388400751568,
      "learning_rate": 7.134518528551065e-07,
      "loss": 0.124,
      "step": 9111
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.238714143831025,
      "learning_rate": 7.126897409610284e-07,
      "loss": 0.1691,
      "step": 9112
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3855352450716862,
      "learning_rate": 7.119280050927407e-07,
      "loss": 0.1811,
      "step": 9113
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2828673668325892,
      "learning_rate": 7.111666453170507e-07,
      "loss": 0.1582,
      "step": 9114
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4078531286343963,
      "learning_rate": 7.104056617007399e-07,
      "loss": 0.1733,
      "step": 9115
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3631778630842837,
      "learning_rate": 7.096450543105471e-07,
      "loss": 0.1707,
      "step": 9116
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2863427409403019,
      "learning_rate": 7.088848232131862e-07,
      "loss": 0.1423,
      "step": 9117
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.693652004533877,
      "learning_rate": 7.081249684753327e-07,
      "loss": 0.6083,
      "step": 9118
    },
    {
      "epoch": 0.83,
      "grad_norm": 5.1632526263594825,
      "learning_rate": 7.073654901636301e-07,
      "loss": 0.5372,
      "step": 9119
    },
    {
      "epoch": 0.83,
      "grad_norm": 6.594825385997862,
      "learning_rate": 7.066063883446916e-07,
      "loss": 0.6254,
      "step": 9120
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2290255889561927,
      "learning_rate": 7.058476630850935e-07,
      "loss": 0.1633,
      "step": 9121
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6193900422365088,
      "learning_rate": 7.050893144513831e-07,
      "loss": 0.2064,
      "step": 9122
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1909858350438982,
      "learning_rate": 7.043313425100712e-07,
      "loss": 0.145,
      "step": 9123
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7412726082865562,
      "learning_rate": 7.035737473276383e-07,
      "loss": 0.1654,
      "step": 9124
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2699847819112215,
      "learning_rate": 7.028165289705296e-07,
      "loss": 0.1706,
      "step": 9125
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4203403218749324,
      "learning_rate": 7.020596875051572e-07,
      "loss": 0.1929,
      "step": 9126
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5187246428047843,
      "learning_rate": 7.013032229979039e-07,
      "loss": 0.2107,
      "step": 9127
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5551976382584138,
      "learning_rate": 7.005471355151139e-07,
      "loss": 0.155,
      "step": 9128
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.440245181465122,
      "learning_rate": 6.997914251231036e-07,
      "loss": 0.1625,
      "step": 9129
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.329178731970748,
      "learning_rate": 6.990360918881517e-07,
      "loss": 0.1969,
      "step": 9130
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2749291528460085,
      "learning_rate": 6.982811358765079e-07,
      "loss": 0.1886,
      "step": 9131
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.258199151495306,
      "learning_rate": 6.975265571543849e-07,
      "loss": 0.1658,
      "step": 9132
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4654785251460551,
      "learning_rate": 6.967723557879663e-07,
      "loss": 0.2205,
      "step": 9133
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4956408029914676,
      "learning_rate": 6.960185318433999e-07,
      "loss": 0.212,
      "step": 9134
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4120829103109527,
      "learning_rate": 6.952650853867993e-07,
      "loss": 0.1675,
      "step": 9135
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3376954794606253,
      "learning_rate": 6.945120164842501e-07,
      "loss": 0.1778,
      "step": 9136
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.402868232693703,
      "learning_rate": 6.937593252017983e-07,
      "loss": 0.155,
      "step": 9137
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2628338281984282,
      "learning_rate": 6.930070116054621e-07,
      "loss": 0.1741,
      "step": 9138
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.300992698300528,
      "learning_rate": 6.922550757612229e-07,
      "loss": 0.1608,
      "step": 9139
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.508737564532336,
      "learning_rate": 6.915035177350327e-07,
      "loss": 0.6548,
      "step": 9140
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4507269087404475,
      "learning_rate": 6.907523375928055e-07,
      "loss": 0.1725,
      "step": 9141
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3701527906760187,
      "learning_rate": 6.900015354004269e-07,
      "loss": 0.1572,
      "step": 9142
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4784180897397645,
      "learning_rate": 6.892511112237472e-07,
      "loss": 0.1987,
      "step": 9143
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.359954509298618,
      "learning_rate": 6.885010651285812e-07,
      "loss": 0.1917,
      "step": 9144
    },
    {
      "epoch": 0.84,
      "grad_norm": 10.507292492643412,
      "learning_rate": 6.87751397180716e-07,
      "loss": 0.6823,
      "step": 9145
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.489912918493555,
      "learning_rate": 6.870021074458999e-07,
      "loss": 0.1612,
      "step": 9146
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.364771808738156,
      "learning_rate": 6.862531959898533e-07,
      "loss": 0.1518,
      "step": 9147
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1222055304496519,
      "learning_rate": 6.85504662878258e-07,
      "loss": 0.1018,
      "step": 9148
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2499745800814273,
      "learning_rate": 6.847565081767676e-07,
      "loss": 0.1463,
      "step": 9149
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.425248802490335,
      "learning_rate": 6.840087319509986e-07,
      "loss": 0.1979,
      "step": 9150
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3951081562048493,
      "learning_rate": 6.832613342665379e-07,
      "loss": 0.1884,
      "step": 9151
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.314367523885589,
      "learning_rate": 6.825143151889358e-07,
      "loss": 0.1778,
      "step": 9152
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3761858325571654,
      "learning_rate": 6.817676747837104e-07,
      "loss": 0.1963,
      "step": 9153
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.234873567304519,
      "learning_rate": 6.81021413116349e-07,
      "loss": 0.1504,
      "step": 9154
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7014198673092973,
      "learning_rate": 6.802755302523012e-07,
      "loss": 0.1861,
      "step": 9155
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.376667668103344,
      "learning_rate": 6.795300262569882e-07,
      "loss": 0.14,
      "step": 9156
    },
    {
      "epoch": 0.84,
      "grad_norm": 8.321204240272984,
      "learning_rate": 6.787849011957942e-07,
      "loss": 0.428,
      "step": 9157
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4297412074161704,
      "learning_rate": 6.780401551340732e-07,
      "loss": 0.1328,
      "step": 9158
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2571598993236104,
      "learning_rate": 6.772957881371423e-07,
      "loss": 0.166,
      "step": 9159
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.532365608101695,
      "learning_rate": 6.765518002702898e-07,
      "loss": 0.5598,
      "step": 9160
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4746046942195332,
      "learning_rate": 6.758081915987669e-07,
      "loss": 0.1782,
      "step": 9161
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.342758245211818,
      "learning_rate": 6.750649621877925e-07,
      "loss": 0.1311,
      "step": 9162
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4510357323029426,
      "learning_rate": 6.74322112102555e-07,
      "loss": 0.2109,
      "step": 9163
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4372699553876238,
      "learning_rate": 6.735796414082046e-07,
      "loss": 0.1412,
      "step": 9164
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3403214633402993,
      "learning_rate": 6.728375501698637e-07,
      "loss": 0.1692,
      "step": 9165
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7201289390396168,
      "learning_rate": 6.72095838452616e-07,
      "loss": 0.1779,
      "step": 9166
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3956458512026275,
      "learning_rate": 6.713545063215171e-07,
      "loss": 0.2229,
      "step": 9167
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3377987023500715,
      "learning_rate": 6.706135538415854e-07,
      "loss": 0.1717,
      "step": 9168
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4852425244637995,
      "learning_rate": 6.698729810778065e-07,
      "loss": 0.1977,
      "step": 9169
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4391133746090619,
      "learning_rate": 6.691327880951359e-07,
      "loss": 0.1751,
      "step": 9170
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4569587860627586,
      "learning_rate": 6.683929749584911e-07,
      "loss": 0.1719,
      "step": 9171
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.363698979144191,
      "learning_rate": 6.676535417327612e-07,
      "loss": 0.1565,
      "step": 9172
    },
    {
      "epoch": 0.84,
      "grad_norm": 17.626757957819006,
      "learning_rate": 6.669144884827966e-07,
      "loss": 0.4439,
      "step": 9173
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3688034318852424,
      "learning_rate": 6.661758152734199e-07,
      "loss": 0.1455,
      "step": 9174
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1967473004035385,
      "learning_rate": 6.654375221694153e-07,
      "loss": 0.1295,
      "step": 9175
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.221818419780837,
      "learning_rate": 6.646996092355384e-07,
      "loss": 0.17,
      "step": 9176
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.345422719586483,
      "learning_rate": 6.639620765365074e-07,
      "loss": 0.1901,
      "step": 9177
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5209940411522882,
      "learning_rate": 6.632249241370087e-07,
      "loss": 0.1591,
      "step": 9178
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3414676178554505,
      "learning_rate": 6.624881521016968e-07,
      "loss": 0.1637,
      "step": 9179
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3376871769672358,
      "learning_rate": 6.617517604951901e-07,
      "loss": 0.1549,
      "step": 9180
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3016733850568523,
      "learning_rate": 6.610157493820768e-07,
      "loss": 0.1877,
      "step": 9181
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.283460749947041,
      "learning_rate": 6.602801188269081e-07,
      "loss": 0.1484,
      "step": 9182
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.155531148877504,
      "learning_rate": 6.595448688942058e-07,
      "loss": 0.457,
      "step": 9183
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5602729083785574,
      "learning_rate": 6.588099996484537e-07,
      "loss": 0.2393,
      "step": 9184
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3658208107276752,
      "learning_rate": 6.580755111541076e-07,
      "loss": 0.1659,
      "step": 9185
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.40729996598768,
      "learning_rate": 6.573414034755854e-07,
      "loss": 0.1817,
      "step": 9186
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2085894316688217,
      "learning_rate": 6.566076766772722e-07,
      "loss": 0.1525,
      "step": 9187
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1833860492682877,
      "learning_rate": 6.558743308235232e-07,
      "loss": 0.1445,
      "step": 9188
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3073009074042383,
      "learning_rate": 6.551413659786554e-07,
      "loss": 0.1303,
      "step": 9189
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4056192756133437,
      "learning_rate": 6.544087822069573e-07,
      "loss": 0.18,
      "step": 9190
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3867512940720346,
      "learning_rate": 6.536765795726785e-07,
      "loss": 0.1898,
      "step": 9191
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4580599712157134,
      "learning_rate": 6.529447581400406e-07,
      "loss": 0.1855,
      "step": 9192
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4294957312547352,
      "learning_rate": 6.522133179732271e-07,
      "loss": 0.1549,
      "step": 9193
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.887587935053699,
      "learning_rate": 6.514822591363923e-07,
      "loss": 0.6194,
      "step": 9194
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.225654432605148,
      "learning_rate": 6.507515816936538e-07,
      "loss": 0.156,
      "step": 9195
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3962471764961637,
      "learning_rate": 6.500212857090959e-07,
      "loss": 0.1681,
      "step": 9196
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.280079427730603,
      "learning_rate": 6.49291371246773e-07,
      "loss": 0.1519,
      "step": 9197
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2786189957742036,
      "learning_rate": 6.48561838370701e-07,
      "loss": 0.1965,
      "step": 9198
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.54726978133092,
      "learning_rate": 6.478326871448665e-07,
      "loss": 0.2101,
      "step": 9199
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.114857006585926,
      "learning_rate": 6.471039176332194e-07,
      "loss": 0.5011,
      "step": 9200
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.346637563035334,
      "learning_rate": 6.463755298996799e-07,
      "loss": 0.1473,
      "step": 9201
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3878773933793482,
      "learning_rate": 6.456475240081294e-07,
      "loss": 0.2232,
      "step": 9202
    },
    {
      "epoch": 0.84,
      "grad_norm": 4.980853205979796,
      "learning_rate": 6.449199000224221e-07,
      "loss": 0.646,
      "step": 9203
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.346325952690006,
      "learning_rate": 6.44192658006374e-07,
      "loss": 0.1791,
      "step": 9204
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4507408000112814,
      "learning_rate": 6.434657980237679e-07,
      "loss": 0.1815,
      "step": 9205
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3191184870793788,
      "learning_rate": 6.427393201383569e-07,
      "loss": 0.1313,
      "step": 9206
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.533270522539676,
      "learning_rate": 6.42013224413855e-07,
      "loss": 0.1945,
      "step": 9207
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2630665698442993,
      "learning_rate": 6.412875109139482e-07,
      "loss": 0.1507,
      "step": 9208
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.5660092785740805,
      "learning_rate": 6.405621797022848e-07,
      "loss": 0.706,
      "step": 9209
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1841907042011344,
      "learning_rate": 6.398372308424827e-07,
      "loss": 0.1445,
      "step": 9210
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4148927491885872,
      "learning_rate": 6.391126643981238e-07,
      "loss": 0.1665,
      "step": 9211
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6185146404995536,
      "learning_rate": 6.383884804327567e-07,
      "loss": 0.1769,
      "step": 9212
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4060019524918905,
      "learning_rate": 6.376646790098984e-07,
      "loss": 0.1815,
      "step": 9213
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4875612682793307,
      "learning_rate": 6.369412601930303e-07,
      "loss": 0.186,
      "step": 9214
    },
    {
      "epoch": 0.84,
      "grad_norm": 9.724996544105647,
      "learning_rate": 6.362182240456022e-07,
      "loss": 0.7087,
      "step": 9215
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3057496224556049,
      "learning_rate": 6.354955706310278e-07,
      "loss": 0.1669,
      "step": 9216
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1956755795410545,
      "learning_rate": 6.347733000126899e-07,
      "loss": 0.1472,
      "step": 9217
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4501128568483204,
      "learning_rate": 6.340514122539354e-07,
      "loss": 0.1791,
      "step": 9218
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3118211498738142,
      "learning_rate": 6.333299074180799e-07,
      "loss": 0.1526,
      "step": 9219
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4086312209280598,
      "learning_rate": 6.326087855684038e-07,
      "loss": 0.1491,
      "step": 9220
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.49718103346276,
      "learning_rate": 6.318880467681527e-07,
      "loss": 0.2219,
      "step": 9221
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5612655993204703,
      "learning_rate": 6.31167691080542e-07,
      "loss": 0.1907,
      "step": 9222
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2988988046264134,
      "learning_rate": 6.304477185687508e-07,
      "loss": 0.166,
      "step": 9223
    },
    {
      "epoch": 0.84,
      "grad_norm": 6.918398735707323,
      "learning_rate": 6.297281292959268e-07,
      "loss": 0.4076,
      "step": 9224
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6477285009750515,
      "learning_rate": 6.290089233251811e-07,
      "loss": 0.1829,
      "step": 9225
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.276226889559467,
      "learning_rate": 6.282901007195946e-07,
      "loss": 0.1648,
      "step": 9226
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3075210977293117,
      "learning_rate": 6.275716615422111e-07,
      "loss": 0.136,
      "step": 9227
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4799359764686,
      "learning_rate": 6.268536058560442e-07,
      "loss": 0.1897,
      "step": 9228
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.864039362963984,
      "learning_rate": 6.261359337240713e-07,
      "loss": 0.5681,
      "step": 9229
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.147240908443607,
      "learning_rate": 6.254186452092359e-07,
      "loss": 0.1262,
      "step": 9230
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5267325186822862,
      "learning_rate": 6.247017403744515e-07,
      "loss": 0.2024,
      "step": 9231
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.337476057352763,
      "learning_rate": 6.239852192825929e-07,
      "loss": 0.1454,
      "step": 9232
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5999174713133806,
      "learning_rate": 6.232690819965065e-07,
      "loss": 0.1996,
      "step": 9233
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3081440833100324,
      "learning_rate": 6.225533285789997e-07,
      "loss": 0.1384,
      "step": 9234
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2364709228285133,
      "learning_rate": 6.218379590928509e-07,
      "loss": 0.163,
      "step": 9235
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4840861823281652,
      "learning_rate": 6.211229736008007e-07,
      "loss": 0.1653,
      "step": 9236
    },
    {
      "epoch": 0.84,
      "grad_norm": 7.014824646834957,
      "learning_rate": 6.204083721655607e-07,
      "loss": 0.6607,
      "step": 9237
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3171222792584005,
      "learning_rate": 6.196941548498048e-07,
      "loss": 0.1561,
      "step": 9238
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2944491572355137,
      "learning_rate": 6.189803217161739e-07,
      "loss": 0.1926,
      "step": 9239
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.285593751891947,
      "learning_rate": 6.182668728272778e-07,
      "loss": 0.6129,
      "step": 9240
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4551534241277941,
      "learning_rate": 6.175538082456883e-07,
      "loss": 0.2128,
      "step": 9241
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4269756809267748,
      "learning_rate": 6.168411280339481e-07,
      "loss": 0.1728,
      "step": 9242
    },
    {
      "epoch": 0.85,
      "grad_norm": 24.81109948768854,
      "learning_rate": 6.161288322545627e-07,
      "loss": 0.422,
      "step": 9243
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4272679792947296,
      "learning_rate": 6.154169209700067e-07,
      "loss": 0.1595,
      "step": 9244
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5361255363931392,
      "learning_rate": 6.147053942427173e-07,
      "loss": 0.184,
      "step": 9245
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.454708549083225,
      "learning_rate": 6.139942521351028e-07,
      "loss": 0.1562,
      "step": 9246
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3339716510458928,
      "learning_rate": 6.132834947095334e-07,
      "loss": 0.1701,
      "step": 9247
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.509633318503748,
      "learning_rate": 6.125731220283465e-07,
      "loss": 0.6972,
      "step": 9248
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.556095485812444,
      "learning_rate": 6.118631341538489e-07,
      "loss": 0.1656,
      "step": 9249
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4280061155778023,
      "learning_rate": 6.111535311483085e-07,
      "loss": 0.1516,
      "step": 9250
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2947743053402547,
      "learning_rate": 6.104443130739657e-07,
      "loss": 0.1288,
      "step": 9251
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2970523838364383,
      "learning_rate": 6.097354799930199e-07,
      "loss": 0.1476,
      "step": 9252
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.435260819031232,
      "learning_rate": 6.090270319676434e-07,
      "loss": 0.166,
      "step": 9253
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1720019221341376,
      "learning_rate": 6.083189690599712e-07,
      "loss": 0.1322,
      "step": 9254
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.401482129644872,
      "learning_rate": 6.076112913321031e-07,
      "loss": 0.2045,
      "step": 9255
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4758916820439854,
      "learning_rate": 6.069039988461101e-07,
      "loss": 0.2074,
      "step": 9256
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5192514751612467,
      "learning_rate": 6.061970916640236e-07,
      "loss": 0.1903,
      "step": 9257
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3031972484316006,
      "learning_rate": 6.054905698478475e-07,
      "loss": 0.137,
      "step": 9258
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.961563222953485,
      "learning_rate": 6.047844334595449e-07,
      "loss": 0.5238,
      "step": 9259
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.531191910868044,
      "learning_rate": 6.040786825610518e-07,
      "loss": 0.4551,
      "step": 9260
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4321810222479,
      "learning_rate": 6.033733172142647e-07,
      "loss": 0.179,
      "step": 9261
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.446893834933065,
      "learning_rate": 6.026683374810516e-07,
      "loss": 0.2087,
      "step": 9262
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2359116375013368,
      "learning_rate": 6.019637434232417e-07,
      "loss": 0.1522,
      "step": 9263
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4849507879322936,
      "learning_rate": 6.012595351026324e-07,
      "loss": 0.1728,
      "step": 9264
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.546932305143697,
      "learning_rate": 6.005557125809896e-07,
      "loss": 0.584,
      "step": 9265
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1359023609415573,
      "learning_rate": 5.998522759200409e-07,
      "loss": 0.1377,
      "step": 9266
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.385350024890974,
      "learning_rate": 5.991492251814846e-07,
      "loss": 0.6567,
      "step": 9267
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4072128772564234,
      "learning_rate": 5.984465604269807e-07,
      "loss": 0.1512,
      "step": 9268
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3844964757473053,
      "learning_rate": 5.9774428171816e-07,
      "loss": 0.1705,
      "step": 9269
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.434084466055941,
      "learning_rate": 5.970423891166149e-07,
      "loss": 0.191,
      "step": 9270
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2736620651230959,
      "learning_rate": 5.963408826839079e-07,
      "loss": 0.1816,
      "step": 9271
    },
    {
      "epoch": 0.85,
      "grad_norm": 9.249843354440364,
      "learning_rate": 5.956397624815646e-07,
      "loss": 0.512,
      "step": 9272
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3762157262739825,
      "learning_rate": 5.949390285710777e-07,
      "loss": 0.1449,
      "step": 9273
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.369200504532378,
      "learning_rate": 5.942386810139072e-07,
      "loss": 0.6703,
      "step": 9274
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3786349106930949,
      "learning_rate": 5.935387198714776e-07,
      "loss": 0.1413,
      "step": 9275
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.420729145833441,
      "learning_rate": 5.928391452051807e-07,
      "loss": 0.2195,
      "step": 9276
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3901981139142134,
      "learning_rate": 5.921399570763731e-07,
      "loss": 0.1446,
      "step": 9277
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.445157383679451,
      "learning_rate": 5.914411555463795e-07,
      "loss": 0.2028,
      "step": 9278
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3242896238578297,
      "learning_rate": 5.907427406764882e-07,
      "loss": 0.1615,
      "step": 9279
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.319238472787639,
      "learning_rate": 5.900447125279557e-07,
      "loss": 0.1326,
      "step": 9280
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.324235302466202,
      "learning_rate": 5.893470711620036e-07,
      "loss": 0.1515,
      "step": 9281
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3470278160405196,
      "learning_rate": 5.886498166398191e-07,
      "loss": 0.219,
      "step": 9282
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3029979823567708,
      "learning_rate": 5.879529490225571e-07,
      "loss": 0.1583,
      "step": 9283
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3424216943375646,
      "learning_rate": 5.872564683713361e-07,
      "loss": 0.1779,
      "step": 9284
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5539166009333607,
      "learning_rate": 5.865603747472443e-07,
      "loss": 0.213,
      "step": 9285
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.562034502878179,
      "learning_rate": 5.858646682113306e-07,
      "loss": 0.4733,
      "step": 9286
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.389364089449417,
      "learning_rate": 5.851693488246169e-07,
      "loss": 0.1599,
      "step": 9287
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.415457716510168,
      "learning_rate": 5.844744166480842e-07,
      "loss": 0.1733,
      "step": 9288
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4208486671116083,
      "learning_rate": 5.837798717426846e-07,
      "loss": 0.1714,
      "step": 9289
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4752251745171918,
      "learning_rate": 5.830857141693341e-07,
      "loss": 0.1819,
      "step": 9290
    },
    {
      "epoch": 0.85,
      "grad_norm": 10.820764128192211,
      "learning_rate": 5.823919439889136e-07,
      "loss": 0.5349,
      "step": 9291
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4092800870803501,
      "learning_rate": 5.816985612622728e-07,
      "loss": 0.1797,
      "step": 9292
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.307683602919559,
      "learning_rate": 5.81005566050225e-07,
      "loss": 0.161,
      "step": 9293
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3918318605931066,
      "learning_rate": 5.80312958413552e-07,
      "loss": 0.1929,
      "step": 9294
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5267565898022932,
      "learning_rate": 5.796207384129981e-07,
      "loss": 0.2101,
      "step": 9295
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3977722432252955,
      "learning_rate": 5.789289061092778e-07,
      "loss": 0.1598,
      "step": 9296
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3909863189817824,
      "learning_rate": 5.782374615630682e-07,
      "loss": 0.1739,
      "step": 9297
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2769640452679987,
      "learning_rate": 5.775464048350127e-07,
      "loss": 0.1703,
      "step": 9298
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3600162571595702,
      "learning_rate": 5.768557359857241e-07,
      "loss": 0.196,
      "step": 9299
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6307563012185136,
      "learning_rate": 5.761654550757756e-07,
      "loss": 0.1756,
      "step": 9300
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5489242988980987,
      "learning_rate": 5.754755621657121e-07,
      "loss": 0.139,
      "step": 9301
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5587384538299338,
      "learning_rate": 5.747860573160402e-07,
      "loss": 0.227,
      "step": 9302
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1025739443949396,
      "learning_rate": 5.740969405872354e-07,
      "loss": 0.1223,
      "step": 9303
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2894203278855172,
      "learning_rate": 5.734082120397367e-07,
      "loss": 0.167,
      "step": 9304
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2473154221517642,
      "learning_rate": 5.727198717339511e-07,
      "loss": 0.1651,
      "step": 9305
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4286166486023009,
      "learning_rate": 5.720319197302498e-07,
      "loss": 0.1969,
      "step": 9306
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4495180872554652,
      "learning_rate": 5.713443560889709e-07,
      "loss": 0.1992,
      "step": 9307
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5875996534772159,
      "learning_rate": 5.706571808704192e-07,
      "loss": 0.144,
      "step": 9308
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3570222028613557,
      "learning_rate": 5.699703941348633e-07,
      "loss": 0.1582,
      "step": 9309
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5883625461845416,
      "learning_rate": 5.6928399594254e-07,
      "loss": 0.2161,
      "step": 9310
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6651818740343254,
      "learning_rate": 5.685979863536506e-07,
      "loss": 0.19,
      "step": 9311
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.452036699675577,
      "learning_rate": 5.67912365428363e-07,
      "loss": 0.2016,
      "step": 9312
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.97462724701292,
      "learning_rate": 5.672271332268098e-07,
      "loss": 0.5119,
      "step": 9313
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1702546991209848,
      "learning_rate": 5.665422898090922e-07,
      "loss": 0.1428,
      "step": 9314
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3748137820851458,
      "learning_rate": 5.65857835235275e-07,
      "loss": 0.152,
      "step": 9315
    },
    {
      "epoch": 0.85,
      "grad_norm": 4.543423422330169,
      "learning_rate": 5.651737695653875e-07,
      "loss": 0.6346,
      "step": 9316
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3051406178097689,
      "learning_rate": 5.644900928594294e-07,
      "loss": 0.162,
      "step": 9317
    },
    {
      "epoch": 0.85,
      "grad_norm": 11.37861685561779,
      "learning_rate": 5.638068051773615e-07,
      "loss": 0.5838,
      "step": 9318
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.30156592936268,
      "learning_rate": 5.63123906579115e-07,
      "loss": 0.1816,
      "step": 9319
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.387764075730026,
      "learning_rate": 5.624413971245824e-07,
      "loss": 0.1588,
      "step": 9320
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3084140626158638,
      "learning_rate": 5.617592768736269e-07,
      "loss": 0.1668,
      "step": 9321
    },
    {
      "epoch": 0.85,
      "grad_norm": 18.284945218469566,
      "learning_rate": 5.610775458860717e-07,
      "loss": 0.5271,
      "step": 9322
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2839782875238712,
      "learning_rate": 5.603962042217126e-07,
      "loss": 0.1644,
      "step": 9323
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3702167574488953,
      "learning_rate": 5.597152519403065e-07,
      "loss": 0.1802,
      "step": 9324
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4800733870304261,
      "learning_rate": 5.590346891015758e-07,
      "loss": 0.2135,
      "step": 9325
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3315390687924604,
      "learning_rate": 5.583545157652132e-07,
      "loss": 0.1856,
      "step": 9326
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.44358116073562,
      "learning_rate": 5.576747319908721e-07,
      "loss": 0.1662,
      "step": 9327
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5364307209109982,
      "learning_rate": 5.569953378381754e-07,
      "loss": 0.1925,
      "step": 9328
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.286173466040506,
      "learning_rate": 5.563163333667098e-07,
      "loss": 0.179,
      "step": 9329
    },
    {
      "epoch": 0.85,
      "grad_norm": 5.713034744737994,
      "learning_rate": 5.556377186360301e-07,
      "loss": 0.3878,
      "step": 9330
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5450578940886204,
      "learning_rate": 5.54959493705653e-07,
      "loss": 0.1634,
      "step": 9331
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.1503582116179585,
      "learning_rate": 5.542816586350653e-07,
      "loss": 0.6551,
      "step": 9332
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5818146940868765,
      "learning_rate": 5.536042134837171e-07,
      "loss": 0.1873,
      "step": 9333
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4287954431895156,
      "learning_rate": 5.529271583110235e-07,
      "loss": 0.1856,
      "step": 9334
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5119388476435398,
      "learning_rate": 5.52250493176369e-07,
      "loss": 0.1475,
      "step": 9335
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4608388863974715,
      "learning_rate": 5.515742181390999e-07,
      "loss": 0.1763,
      "step": 9336
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4255645424955088,
      "learning_rate": 5.508983332585316e-07,
      "loss": 0.1742,
      "step": 9337
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3677344441674877,
      "learning_rate": 5.502228385939418e-07,
      "loss": 0.1636,
      "step": 9338
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3166304726042024,
      "learning_rate": 5.49547734204578e-07,
      "loss": 0.1576,
      "step": 9339
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3891729539562105,
      "learning_rate": 5.488730201496501e-07,
      "loss": 0.1727,
      "step": 9340
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3425053042833284,
      "learning_rate": 5.481986964883345e-07,
      "loss": 0.1448,
      "step": 9341
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2647013404807295,
      "learning_rate": 5.475247632797759e-07,
      "loss": 0.1704,
      "step": 9342
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2869265893469568,
      "learning_rate": 5.468512205830806e-07,
      "loss": 0.1482,
      "step": 9343
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4650591720061856,
      "learning_rate": 5.461780684573248e-07,
      "loss": 0.1866,
      "step": 9344
    },
    {
      "epoch": 0.85,
      "grad_norm": 8.747459836546453,
      "learning_rate": 5.455053069615456e-07,
      "loss": 0.5654,
      "step": 9345
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2655436839232537,
      "learning_rate": 5.448329361547522e-07,
      "loss": 0.162,
      "step": 9346
    },
    {
      "epoch": 0.85,
      "grad_norm": 7.021664215239333,
      "learning_rate": 5.441609560959132e-07,
      "loss": 0.595,
      "step": 9347
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2664475657397802,
      "learning_rate": 5.434893668439678e-07,
      "loss": 0.1569,
      "step": 9348
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4773967349689834,
      "learning_rate": 5.428181684578177e-07,
      "loss": 0.1903,
      "step": 9349
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3026201725259337,
      "learning_rate": 5.421473609963307e-07,
      "loss": 0.1687,
      "step": 9350
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2966210466893149,
      "learning_rate": 5.414769445183432e-07,
      "loss": 0.1509,
      "step": 9351
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.432472687166909,
      "learning_rate": 5.408069190826531e-07,
      "loss": 0.1828,
      "step": 9352
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.41653331828677,
      "learning_rate": 5.401372847480285e-07,
      "loss": 0.5373,
      "step": 9353
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2873892582240054,
      "learning_rate": 5.394680415731979e-07,
      "loss": 0.1559,
      "step": 9354
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4400446575538344,
      "learning_rate": 5.387991896168609e-07,
      "loss": 0.2569,
      "step": 9355
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.34582685055546,
      "learning_rate": 5.381307289376786e-07,
      "loss": 0.14,
      "step": 9356
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3924183195641333,
      "learning_rate": 5.374626595942811e-07,
      "loss": 0.1555,
      "step": 9357
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2745271877868543,
      "learning_rate": 5.367949816452616e-07,
      "loss": 0.1596,
      "step": 9358
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.202896979985382,
      "learning_rate": 5.361276951491789e-07,
      "loss": 0.6459,
      "step": 9359
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.309081911009978,
      "learning_rate": 5.354608001645611e-07,
      "loss": 0.1901,
      "step": 9360
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.989493701565725,
      "learning_rate": 5.347942967498965e-07,
      "loss": 0.5692,
      "step": 9361
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.436369690761993,
      "learning_rate": 5.341281849636437e-07,
      "loss": 0.2074,
      "step": 9362
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4192325857981551,
      "learning_rate": 5.334624648642245e-07,
      "loss": 0.2216,
      "step": 9363
    },
    {
      "epoch": 0.86,
      "grad_norm": 15.37469481125021,
      "learning_rate": 5.327971365100276e-07,
      "loss": 0.5525,
      "step": 9364
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4993130141871025,
      "learning_rate": 5.321321999594054e-07,
      "loss": 0.1441,
      "step": 9365
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3277638331703576,
      "learning_rate": 5.314676552706793e-07,
      "loss": 0.1559,
      "step": 9366
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4264297970791961,
      "learning_rate": 5.308035025021335e-07,
      "loss": 0.1852,
      "step": 9367
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5934008864234488,
      "learning_rate": 5.301397417120169e-07,
      "loss": 0.2226,
      "step": 9368
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4590118038608932,
      "learning_rate": 5.294763729585484e-07,
      "loss": 0.1701,
      "step": 9369
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4870361507448309,
      "learning_rate": 5.288133962999076e-07,
      "loss": 0.1874,
      "step": 9370
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3302540817753647,
      "learning_rate": 5.281508117942446e-07,
      "loss": 0.1483,
      "step": 9371
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5085443062845538,
      "learning_rate": 5.274886194996698e-07,
      "loss": 0.1431,
      "step": 9372
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2687858736404387,
      "learning_rate": 5.268268194742638e-07,
      "loss": 0.173,
      "step": 9373
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.568599770725523,
      "learning_rate": 5.26165411776069e-07,
      "loss": 0.1978,
      "step": 9374
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.307948533877031,
      "learning_rate": 5.25504396463098e-07,
      "loss": 0.1715,
      "step": 9375
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.516446603665749,
      "learning_rate": 5.248437735933242e-07,
      "loss": 0.212,
      "step": 9376
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0948188115823974,
      "learning_rate": 5.241835432246888e-07,
      "loss": 0.1033,
      "step": 9377
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5702512266577469,
      "learning_rate": 5.235237054150994e-07,
      "loss": 0.1999,
      "step": 9378
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5395941071874026,
      "learning_rate": 5.228642602224266e-07,
      "loss": 0.1796,
      "step": 9379
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2008206085173925,
      "learning_rate": 5.222052077045103e-07,
      "loss": 0.1328,
      "step": 9380
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3996926945441703,
      "learning_rate": 5.215465479191511e-07,
      "loss": 0.1713,
      "step": 9381
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.509504517006286,
      "learning_rate": 5.208882809241211e-07,
      "loss": 0.2288,
      "step": 9382
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3090815217950174,
      "learning_rate": 5.202304067771524e-07,
      "loss": 0.1921,
      "step": 9383
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.636217681562517,
      "learning_rate": 5.19572925535945e-07,
      "loss": 0.5954,
      "step": 9384
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.279015728506597,
      "learning_rate": 5.18915837258166e-07,
      "loss": 0.1527,
      "step": 9385
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4214638795109689,
      "learning_rate": 5.182591420014443e-07,
      "loss": 0.2091,
      "step": 9386
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3095237505074733,
      "learning_rate": 5.17602839823379e-07,
      "loss": 0.1757,
      "step": 9387
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3650204704227096,
      "learning_rate": 5.169469307815294e-07,
      "loss": 0.19,
      "step": 9388
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4730614141601504,
      "learning_rate": 5.162914149334258e-07,
      "loss": 0.2091,
      "step": 9389
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.40277717973386,
      "learning_rate": 5.156362923365587e-07,
      "loss": 0.1842,
      "step": 9390
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3730947121479855,
      "learning_rate": 5.14981563048389e-07,
      "loss": 0.1733,
      "step": 9391
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.642782449324843,
      "learning_rate": 5.1432722712634e-07,
      "loss": 0.62,
      "step": 9392
    },
    {
      "epoch": 0.86,
      "grad_norm": 10.202393569299591,
      "learning_rate": 5.136732846278003e-07,
      "loss": 0.4753,
      "step": 9393
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4628460752689034,
      "learning_rate": 5.130197356101274e-07,
      "loss": 0.1532,
      "step": 9394
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.531941723597389,
      "learning_rate": 5.123665801306388e-07,
      "loss": 0.226,
      "step": 9395
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2115158449872525,
      "learning_rate": 5.117138182466235e-07,
      "loss": 0.118,
      "step": 9396
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3510277986639367,
      "learning_rate": 5.110614500153311e-07,
      "loss": 0.1936,
      "step": 9397
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3077104287904953,
      "learning_rate": 5.104094754939804e-07,
      "loss": 0.1641,
      "step": 9398
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.409415358017663,
      "learning_rate": 5.097578947397519e-07,
      "loss": 0.1867,
      "step": 9399
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2401610782788717,
      "learning_rate": 5.091067078097956e-07,
      "loss": 0.1615,
      "step": 9400
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2948868510263296,
      "learning_rate": 5.084559147612244e-07,
      "loss": 0.1633,
      "step": 9401
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4312468265006018,
      "learning_rate": 5.078055156511152e-07,
      "loss": 0.1586,
      "step": 9402
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.261256459692085,
      "learning_rate": 5.071555105365156e-07,
      "loss": 0.1433,
      "step": 9403
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2446781121325625,
      "learning_rate": 5.065058994744321e-07,
      "loss": 0.1551,
      "step": 9404
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3898649423125842,
      "learning_rate": 5.05856682521843e-07,
      "loss": 0.1942,
      "step": 9405
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5065759152995424,
      "learning_rate": 5.052078597356868e-07,
      "loss": 0.1578,
      "step": 9406
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4335558277196776,
      "learning_rate": 5.045594311728708e-07,
      "loss": 0.1739,
      "step": 9407
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3082875397089744,
      "learning_rate": 5.039113968902654e-07,
      "loss": 0.1437,
      "step": 9408
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.368092555470712,
      "learning_rate": 5.032637569447091e-07,
      "loss": 0.1689,
      "step": 9409
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6273068774190507,
      "learning_rate": 5.026165113930037e-07,
      "loss": 0.2324,
      "step": 9410
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3976012322856342,
      "learning_rate": 5.019696602919155e-07,
      "loss": 0.1549,
      "step": 9411
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3627908991099245,
      "learning_rate": 5.013232036981797e-07,
      "loss": 0.1693,
      "step": 9412
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4074569869942384,
      "learning_rate": 5.00677141668493e-07,
      "loss": 0.1442,
      "step": 9413
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2558670553912668,
      "learning_rate": 5.000314742595214e-07,
      "loss": 0.1541,
      "step": 9414
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3122343094707605,
      "learning_rate": 4.993862015278922e-07,
      "loss": 0.1663,
      "step": 9415
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5314330624924097,
      "learning_rate": 4.987413235302025e-07,
      "loss": 0.2183,
      "step": 9416
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3614450028326528,
      "learning_rate": 4.980968403230097e-07,
      "loss": 0.1979,
      "step": 9417
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4201893182877166,
      "learning_rate": 4.974527519628413e-07,
      "loss": 0.1937,
      "step": 9418
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.719505004791233,
      "learning_rate": 4.96809058506188e-07,
      "loss": 0.4494,
      "step": 9419
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.371651444221982,
      "learning_rate": 4.961657600095043e-07,
      "loss": 0.1585,
      "step": 9420
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3334570927468827,
      "learning_rate": 4.955228565292142e-07,
      "loss": 0.175,
      "step": 9421
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.240834192363153,
      "learning_rate": 4.948803481217024e-07,
      "loss": 0.1532,
      "step": 9422
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3339602641309933,
      "learning_rate": 4.942382348433233e-07,
      "loss": 0.1445,
      "step": 9423
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5322906798606113,
      "learning_rate": 4.93596516750392e-07,
      "loss": 0.1979,
      "step": 9424
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2610433494750304,
      "learning_rate": 4.929551938991945e-07,
      "loss": 0.1276,
      "step": 9425
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.133698573089895,
      "learning_rate": 4.923142663459774e-07,
      "loss": 0.479,
      "step": 9426
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.452079595142217,
      "learning_rate": 4.916737341469529e-07,
      "loss": 0.1895,
      "step": 9427
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3752718347997352,
      "learning_rate": 4.91033597358303e-07,
      "loss": 0.1801,
      "step": 9428
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4941497604911993,
      "learning_rate": 4.903938560361698e-07,
      "loss": 0.1905,
      "step": 9429
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4469779237390583,
      "learning_rate": 4.897545102366641e-07,
      "loss": 0.2006,
      "step": 9430
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.496846262607309,
      "learning_rate": 4.891155600158592e-07,
      "loss": 0.1832,
      "step": 9431
    },
    {
      "epoch": 0.86,
      "grad_norm": 5.861994233396674,
      "learning_rate": 4.884770054297977e-07,
      "loss": 0.4506,
      "step": 9432
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4146660541075275,
      "learning_rate": 4.87838846534483e-07,
      "loss": 0.1871,
      "step": 9433
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2956208649961736,
      "learning_rate": 4.872010833858875e-07,
      "loss": 0.1261,
      "step": 9434
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5715978983402636,
      "learning_rate": 4.865637160399467e-07,
      "loss": 0.1841,
      "step": 9435
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4448579049154788,
      "learning_rate": 4.859267445525606e-07,
      "loss": 0.192,
      "step": 9436
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2278460109388016,
      "learning_rate": 4.85290168979598e-07,
      "loss": 0.1868,
      "step": 9437
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4058882656441507,
      "learning_rate": 4.846539893768893e-07,
      "loss": 0.1487,
      "step": 9438
    },
    {
      "epoch": 0.86,
      "grad_norm": 6.2324198056996725,
      "learning_rate": 4.840182058002329e-07,
      "loss": 0.7239,
      "step": 9439
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.523153511512153,
      "learning_rate": 4.833828183053896e-07,
      "loss": 0.19,
      "step": 9440
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3157625664698123,
      "learning_rate": 4.827478269480895e-07,
      "loss": 0.158,
      "step": 9441
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3237990446534706,
      "learning_rate": 4.821132317840233e-07,
      "loss": 0.169,
      "step": 9442
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2788999054840153,
      "learning_rate": 4.814790328688506e-07,
      "loss": 0.1856,
      "step": 9443
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2488922779039244,
      "learning_rate": 4.808452302581951e-07,
      "loss": 0.1248,
      "step": 9444
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3880318707564314,
      "learning_rate": 4.802118240076442e-07,
      "loss": 0.1611,
      "step": 9445
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5087991266652123,
      "learning_rate": 4.79578814172752e-07,
      "loss": 0.1923,
      "step": 9446
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.468709744784216,
      "learning_rate": 4.789462008090389e-07,
      "loss": 0.1863,
      "step": 9447
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.565552660785216,
      "learning_rate": 4.783139839719891e-07,
      "loss": 0.1898,
      "step": 9448
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0933051390326918,
      "learning_rate": 4.776821637170525e-07,
      "loss": 0.1263,
      "step": 9449
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4675000459975054,
      "learning_rate": 4.770507400996422e-07,
      "loss": 0.2014,
      "step": 9450
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.339800967611032,
      "learning_rate": 4.7641971317514025e-07,
      "loss": 0.1812,
      "step": 9451
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.32220718278629,
      "learning_rate": 4.757890829988904e-07,
      "loss": 0.1775,
      "step": 9452
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2244141057239468,
      "learning_rate": 4.751588496262044e-07,
      "loss": 0.1413,
      "step": 9453
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.366885722478068,
      "learning_rate": 4.74529013112357e-07,
      "loss": 0.1426,
      "step": 9454
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.54591555748931,
      "learning_rate": 4.738995735125895e-07,
      "loss": 0.1896,
      "step": 9455
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.471576198731406,
      "learning_rate": 4.7327053088210786e-07,
      "loss": 0.1876,
      "step": 9456
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.484984975773857,
      "learning_rate": 4.726418852760839e-07,
      "loss": 0.1635,
      "step": 9457
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4290385181493899,
      "learning_rate": 4.720136367496536e-07,
      "loss": 0.1819,
      "step": 9458
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.425948634128643,
      "learning_rate": 4.7138578535791737e-07,
      "loss": 0.2098,
      "step": 9459
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2430089571590832,
      "learning_rate": 4.7075833115594437e-07,
      "loss": 0.1174,
      "step": 9460
    },
    {
      "epoch": 0.87,
      "grad_norm": 23.479050450668087,
      "learning_rate": 4.70131274198764e-07,
      "loss": 0.5255,
      "step": 9461
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4094667504025051,
      "learning_rate": 4.695046145413762e-07,
      "loss": 0.1313,
      "step": 9462
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3067726423192565,
      "learning_rate": 4.688783522387402e-07,
      "loss": 0.1857,
      "step": 9463
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5658184737725582,
      "learning_rate": 4.6825248734578553e-07,
      "loss": 0.2245,
      "step": 9464
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4713755099454304,
      "learning_rate": 4.6762701991740434e-07,
      "loss": 0.2035,
      "step": 9465
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5843412441635218,
      "learning_rate": 4.670019500084527e-07,
      "loss": 0.1687,
      "step": 9466
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.378740974678732,
      "learning_rate": 4.6637727767375573e-07,
      "loss": 0.1579,
      "step": 9467
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.050573299930204,
      "learning_rate": 4.6575300296809956e-07,
      "loss": 0.5888,
      "step": 9468
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2872547268199652,
      "learning_rate": 4.651291259462393e-07,
      "loss": 0.1448,
      "step": 9469
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6760396268436895,
      "learning_rate": 4.6450564666289053e-07,
      "loss": 0.1908,
      "step": 9470
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.62109390581649,
      "learning_rate": 4.638825651727391e-07,
      "loss": 0.5457,
      "step": 9471
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3361904298515708,
      "learning_rate": 4.632598815304312e-07,
      "loss": 0.1552,
      "step": 9472
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3425078487993822,
      "learning_rate": 4.626375957905821e-07,
      "loss": 0.1979,
      "step": 9473
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4618183439753052,
      "learning_rate": 4.620157080077697e-07,
      "loss": 0.1945,
      "step": 9474
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3903694423596407,
      "learning_rate": 4.6139421823653716e-07,
      "loss": 0.1353,
      "step": 9475
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6627419883342185,
      "learning_rate": 4.6077312653139414e-07,
      "loss": 0.1839,
      "step": 9476
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2958789247265616,
      "learning_rate": 4.601524329468138e-07,
      "loss": 0.1492,
      "step": 9477
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4652818780640653,
      "learning_rate": 4.5953213753723637e-07,
      "loss": 0.167,
      "step": 9478
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2146039505279498,
      "learning_rate": 4.589122403570645e-07,
      "loss": 0.1644,
      "step": 9479
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.527278849763563,
      "learning_rate": 4.582927414606686e-07,
      "loss": 0.1599,
      "step": 9480
    },
    {
      "epoch": 0.87,
      "grad_norm": 13.435148237704007,
      "learning_rate": 4.576736409023813e-07,
      "loss": 0.4905,
      "step": 9481
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1460406815614133,
      "learning_rate": 4.570549387365036e-07,
      "loss": 0.1413,
      "step": 9482
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4644569073795983,
      "learning_rate": 4.5643663501729885e-07,
      "loss": 0.1814,
      "step": 9483
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1986614814432879,
      "learning_rate": 4.5581872979899576e-07,
      "loss": 0.1438,
      "step": 9484
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4208200511249403,
      "learning_rate": 4.5520122313579054e-07,
      "loss": 0.1815,
      "step": 9485
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.391537937500784,
      "learning_rate": 4.5458411508184084e-07,
      "loss": 0.1913,
      "step": 9486
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2789623217819392,
      "learning_rate": 4.539674056912724e-07,
      "loss": 0.1613,
      "step": 9487
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3905032783461309,
      "learning_rate": 4.5335109501817407e-07,
      "loss": 0.1641,
      "step": 9488
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4771272974797687,
      "learning_rate": 4.5273518311660103e-07,
      "loss": 0.1697,
      "step": 9489
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3743972752874065,
      "learning_rate": 4.521196700405717e-07,
      "loss": 0.1518,
      "step": 9490
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5136490039028196,
      "learning_rate": 4.515045558440728e-07,
      "loss": 0.1883,
      "step": 9491
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.342131882770714,
      "learning_rate": 4.5088984058105246e-07,
      "loss": 0.3879,
      "step": 9492
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3249605159695903,
      "learning_rate": 4.5027552430542467e-07,
      "loss": 0.1593,
      "step": 9493
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2669944383623455,
      "learning_rate": 4.4966160707107075e-07,
      "loss": 0.144,
      "step": 9494
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.631280141319869,
      "learning_rate": 4.490480889318338e-07,
      "loss": 0.5212,
      "step": 9495
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2151154375309725,
      "learning_rate": 4.4843496994152527e-07,
      "loss": 0.1156,
      "step": 9496
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2926742905200148,
      "learning_rate": 4.4782225015391754e-07,
      "loss": 0.1655,
      "step": 9497
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.632330376974551,
      "learning_rate": 4.4720992962275223e-07,
      "loss": 0.4951,
      "step": 9498
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3208822156450286,
      "learning_rate": 4.465980084017324e-07,
      "loss": 0.1662,
      "step": 9499
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2792775997529788,
      "learning_rate": 4.459864865445296e-07,
      "loss": 0.1836,
      "step": 9500
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2975846558854154,
      "learning_rate": 4.45375364104777e-07,
      "loss": 0.1705,
      "step": 9501
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3509294805145873,
      "learning_rate": 4.4476464113607285e-07,
      "loss": 0.1613,
      "step": 9502
    },
    {
      "epoch": 0.87,
      "grad_norm": 15.191306014649834,
      "learning_rate": 4.441543176919849e-07,
      "loss": 0.5074,
      "step": 9503
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2729644144890775,
      "learning_rate": 4.435443938260392e-07,
      "loss": 0.1728,
      "step": 9504
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5176660228219114,
      "learning_rate": 4.429348695917329e-07,
      "loss": 0.178,
      "step": 9505
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3607789555968988,
      "learning_rate": 4.4232574504252337e-07,
      "loss": 0.1639,
      "step": 9506
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3790066251711275,
      "learning_rate": 4.4171702023183663e-07,
      "loss": 0.1534,
      "step": 9507
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.546342237743706,
      "learning_rate": 4.411086952130611e-07,
      "loss": 0.5098,
      "step": 9508
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5015819286796344,
      "learning_rate": 4.405007700395497e-07,
      "loss": 0.1742,
      "step": 9509
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.46309309623645,
      "learning_rate": 4.398932447646237e-07,
      "loss": 0.1744,
      "step": 9510
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5294115532606118,
      "learning_rate": 4.392861194415654e-07,
      "loss": 0.1535,
      "step": 9511
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.523433908977804,
      "learning_rate": 4.38679394123625e-07,
      "loss": 0.1774,
      "step": 9512
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4125886643023766,
      "learning_rate": 4.3807306886401555e-07,
      "loss": 0.1687,
      "step": 9513
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4204561741574495,
      "learning_rate": 4.374671437159167e-07,
      "loss": 0.1689,
      "step": 9514
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4708859605418854,
      "learning_rate": 4.3686161873247037e-07,
      "loss": 0.2113,
      "step": 9515
    },
    {
      "epoch": 0.87,
      "grad_norm": 5.667763364792595,
      "learning_rate": 4.3625649396678795e-07,
      "loss": 0.5984,
      "step": 9516
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3370346104296529,
      "learning_rate": 4.356517694719409e-07,
      "loss": 0.1641,
      "step": 9517
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7388546631847794,
      "learning_rate": 4.3504744530096734e-07,
      "loss": 0.1827,
      "step": 9518
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.202404928642059,
      "learning_rate": 4.3444352150687206e-07,
      "loss": 0.165,
      "step": 9519
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3366787156391664,
      "learning_rate": 4.338399981426211e-07,
      "loss": 0.1616,
      "step": 9520
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.6171972614391805,
      "learning_rate": 4.3323687526115045e-07,
      "loss": 0.5214,
      "step": 9521
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1405465881414683,
      "learning_rate": 4.3263415291535495e-07,
      "loss": 0.1352,
      "step": 9522
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4162754579305836,
      "learning_rate": 4.320318311581001e-07,
      "loss": 0.1849,
      "step": 9523
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.585856990863089,
      "learning_rate": 4.314299100422109e-07,
      "loss": 0.6105,
      "step": 9524
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.75613475422491,
      "learning_rate": 4.3082838962048237e-07,
      "loss": 0.4393,
      "step": 9525
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3242760165155696,
      "learning_rate": 4.302272699456711e-07,
      "loss": 0.1718,
      "step": 9526
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.150235987150477,
      "learning_rate": 4.2962655107049723e-07,
      "loss": 0.5381,
      "step": 9527
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3827111317575733,
      "learning_rate": 4.2902623304765134e-07,
      "loss": 0.1534,
      "step": 9528
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5694731924682268,
      "learning_rate": 4.284263159297819e-07,
      "loss": 0.1445,
      "step": 9529
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2943261745137933,
      "learning_rate": 4.278267997695085e-07,
      "loss": 0.1689,
      "step": 9530
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3498838465051062,
      "learning_rate": 4.272276846194101e-07,
      "loss": 0.1601,
      "step": 9531
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3807716349456984,
      "learning_rate": 4.2662897053203587e-07,
      "loss": 0.1549,
      "step": 9532
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5024761708044174,
      "learning_rate": 4.2603065755989493e-07,
      "loss": 0.1515,
      "step": 9533
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3191528619835715,
      "learning_rate": 4.254327457554647e-07,
      "loss": 0.1624,
      "step": 9534
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5047512800854004,
      "learning_rate": 4.248352351711854e-07,
      "loss": 0.1737,
      "step": 9535
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.336603564174764,
      "learning_rate": 4.242381258594619e-07,
      "loss": 0.1669,
      "step": 9536
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4893081988727397,
      "learning_rate": 4.2364141787266613e-07,
      "loss": 0.2,
      "step": 9537
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.423640905940152,
      "learning_rate": 4.230451112631323e-07,
      "loss": 0.1993,
      "step": 9538
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3097483855361245,
      "learning_rate": 4.2244920608316144e-07,
      "loss": 0.1605,
      "step": 9539
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2979406307129016,
      "learning_rate": 4.218537023850172e-07,
      "loss": 0.1835,
      "step": 9540
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3960566268785255,
      "learning_rate": 4.2125860022093066e-07,
      "loss": 0.1364,
      "step": 9541
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4943858374836136,
      "learning_rate": 4.2066389964309507e-07,
      "loss": 0.1878,
      "step": 9542
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4290243854883422,
      "learning_rate": 4.200696007036703e-07,
      "loss": 0.1456,
      "step": 9543
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2915323619599801,
      "learning_rate": 4.1947570345478085e-07,
      "loss": 0.1508,
      "step": 9544
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2804738956807475,
      "learning_rate": 4.1888220794851386e-07,
      "loss": 0.1395,
      "step": 9545
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.591377229287582,
      "learning_rate": 4.182891142369244e-07,
      "loss": 0.2099,
      "step": 9546
    },
    {
      "epoch": 0.87,
      "grad_norm": 11.848176739841614,
      "learning_rate": 4.1769642237202924e-07,
      "loss": 0.5046,
      "step": 9547
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3688267631679698,
      "learning_rate": 4.171041324058134e-07,
      "loss": 0.1583,
      "step": 9548
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.483397061751091,
      "learning_rate": 4.165122443902225e-07,
      "loss": 0.1931,
      "step": 9549
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2512464592040635,
      "learning_rate": 4.1592075837717126e-07,
      "loss": 0.1694,
      "step": 9550
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4953045973419188,
      "learning_rate": 4.1532967441853643e-07,
      "loss": 0.1669,
      "step": 9551
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3388302001429506,
      "learning_rate": 4.147389925661577e-07,
      "loss": 0.1778,
      "step": 9552
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5175635348725953,
      "learning_rate": 4.141487128718452e-07,
      "loss": 0.2074,
      "step": 9553
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4508028498552938,
      "learning_rate": 4.135588353873671e-07,
      "loss": 0.201,
      "step": 9554
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.817632484631189,
      "learning_rate": 4.1296936016446297e-07,
      "loss": 0.5064,
      "step": 9555
    },
    {
      "epoch": 0.87,
      "grad_norm": 7.133711127632838,
      "learning_rate": 4.1238028725483105e-07,
      "loss": 0.4922,
      "step": 9556
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3305790991580164,
      "learning_rate": 4.1179161671013876e-07,
      "loss": 0.1461,
      "step": 9557
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3457500030412886,
      "learning_rate": 4.112033485820144e-07,
      "loss": 0.1661,
      "step": 9558
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.293723395743828,
      "learning_rate": 4.10615482922056e-07,
      "loss": 0.2087,
      "step": 9559
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3066738930999973,
      "learning_rate": 4.100280197818207e-07,
      "loss": 0.1646,
      "step": 9560
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4474688878231126,
      "learning_rate": 4.0944095921283347e-07,
      "loss": 0.1773,
      "step": 9561
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4895770965244304,
      "learning_rate": 4.088543012665852e-07,
      "loss": 0.1593,
      "step": 9562
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2444932158125532,
      "learning_rate": 4.082680459945265e-07,
      "loss": 0.1477,
      "step": 9563
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4875554915443814,
      "learning_rate": 4.0768219344807955e-07,
      "loss": 0.1777,
      "step": 9564
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.344461814171348,
      "learning_rate": 4.070967436786244e-07,
      "loss": 0.1506,
      "step": 9565
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3273125191103015,
      "learning_rate": 4.06511696737511e-07,
      "loss": 0.1771,
      "step": 9566
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4806577258610403,
      "learning_rate": 4.059270526760506e-07,
      "loss": 0.1959,
      "step": 9567
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4747065229545628,
      "learning_rate": 4.053428115455216e-07,
      "loss": 0.1629,
      "step": 9568
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.226794987717282,
      "learning_rate": 4.0475897339716466e-07,
      "loss": 0.1615,
      "step": 9569
    },
    {
      "epoch": 0.87,
      "grad_norm": 6.381049298458121,
      "learning_rate": 4.0417553828218656e-07,
      "loss": 0.5217,
      "step": 9570
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3705044351699265,
      "learning_rate": 4.035925062517593e-07,
      "loss": 0.2062,
      "step": 9571
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.398124324868973,
      "learning_rate": 4.0300987735701733e-07,
      "loss": 0.1569,
      "step": 9572
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3308088371472044,
      "learning_rate": 4.024276516490627e-07,
      "loss": 0.171,
      "step": 9573
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.412564457157437,
      "learning_rate": 4.01845829178959e-07,
      "loss": 0.1664,
      "step": 9574
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.362626163577762,
      "learning_rate": 4.012644099977375e-07,
      "loss": 0.1489,
      "step": 9575
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3019455635297081,
      "learning_rate": 4.0068339415639033e-07,
      "loss": 0.1203,
      "step": 9576
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.229291411907437,
      "learning_rate": 4.001027817058789e-07,
      "loss": 0.1603,
      "step": 9577
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.417664842018877,
      "learning_rate": 3.995225726971258e-07,
      "loss": 0.1363,
      "step": 9578
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3856391422253618,
      "learning_rate": 3.9894276718101865e-07,
      "loss": 0.1574,
      "step": 9579
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4153695039393124,
      "learning_rate": 3.9836336520841124e-07,
      "loss": 0.187,
      "step": 9580
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4071592857828994,
      "learning_rate": 3.9778436683011957e-07,
      "loss": 0.1848,
      "step": 9581
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5135538738890062,
      "learning_rate": 3.9720577209692747e-07,
      "loss": 0.1611,
      "step": 9582
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4021311228803781,
      "learning_rate": 3.9662758105957986e-07,
      "loss": 0.1981,
      "step": 9583
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4572324472494755,
      "learning_rate": 3.960497937687902e-07,
      "loss": 0.1941,
      "step": 9584
    },
    {
      "epoch": 0.88,
      "grad_norm": 13.426960024702707,
      "learning_rate": 3.9547241027523164e-07,
      "loss": 0.5768,
      "step": 9585
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3632643480785749,
      "learning_rate": 3.948954306295466e-07,
      "loss": 0.165,
      "step": 9586
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2818972581633428,
      "learning_rate": 3.943188548823396e-07,
      "loss": 0.1615,
      "step": 9587
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3149776654017507,
      "learning_rate": 3.9374268308417895e-07,
      "loss": 0.1535,
      "step": 9588
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5348462347762013,
      "learning_rate": 3.9316691528560103e-07,
      "loss": 0.1545,
      "step": 9589
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5760687365981603,
      "learning_rate": 3.9259155153710207e-07,
      "loss": 0.1577,
      "step": 9590
    },
    {
      "epoch": 0.88,
      "grad_norm": 9.02809957529744,
      "learning_rate": 3.920165918891472e-07,
      "loss": 0.7024,
      "step": 9591
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.446054541114556,
      "learning_rate": 3.9144203639216285e-07,
      "loss": 0.1849,
      "step": 9592
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.317230986116325,
      "learning_rate": 3.908678850965425e-07,
      "loss": 0.1962,
      "step": 9593
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4517579007549792,
      "learning_rate": 3.902941380526426e-07,
      "loss": 0.2107,
      "step": 9594
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.301460653732728,
      "learning_rate": 3.8972079531078454e-07,
      "loss": 0.1773,
      "step": 9595
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3598843096191746,
      "learning_rate": 3.8914785692125423e-07,
      "loss": 0.1767,
      "step": 9596
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3796569725792862,
      "learning_rate": 3.88575322934302e-07,
      "loss": 0.1718,
      "step": 9597
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3128592410986333,
      "learning_rate": 3.8800319340014437e-07,
      "loss": 0.1633,
      "step": 9598
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.37271007357051,
      "learning_rate": 3.874314683689584e-07,
      "loss": 0.4632,
      "step": 9599
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.654943107615491,
      "learning_rate": 3.868601478908912e-07,
      "loss": 0.488,
      "step": 9600
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.327020000643534,
      "learning_rate": 3.862892320160483e-07,
      "loss": 0.1744,
      "step": 9601
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4498788095000799,
      "learning_rate": 3.857187207945057e-07,
      "loss": 0.2272,
      "step": 9602
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.524662277580298,
      "learning_rate": 3.851486142762995e-07,
      "loss": 0.1816,
      "step": 9603
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3623029245962137,
      "learning_rate": 3.8457891251143076e-07,
      "loss": 0.1426,
      "step": 9604
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1793176663114062,
      "learning_rate": 3.8400961554986906e-07,
      "loss": 0.1309,
      "step": 9605
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4878384516078698,
      "learning_rate": 3.834407234415427e-07,
      "loss": 0.1654,
      "step": 9606
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.380097254069067,
      "learning_rate": 3.828722362363491e-07,
      "loss": 0.1655,
      "step": 9607
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4526810888738217,
      "learning_rate": 3.823041539841477e-07,
      "loss": 0.2067,
      "step": 9608
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3865189742247495,
      "learning_rate": 3.8173647673476366e-07,
      "loss": 0.2083,
      "step": 9609
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4833862915604843,
      "learning_rate": 3.81169204537985e-07,
      "loss": 0.1883,
      "step": 9610
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5763569353615432,
      "learning_rate": 3.8060233744356634e-07,
      "loss": 0.2139,
      "step": 9611
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4981354728499796,
      "learning_rate": 3.800358755012257e-07,
      "loss": 0.2217,
      "step": 9612
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.384069009380949,
      "learning_rate": 3.7946981876064435e-07,
      "loss": 0.2051,
      "step": 9613
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4984494469854128,
      "learning_rate": 3.7890416727147105e-07,
      "loss": 0.196,
      "step": 9614
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2262181537485601,
      "learning_rate": 3.7833892108331547e-07,
      "loss": 0.1321,
      "step": 9615
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4372901862496152,
      "learning_rate": 3.7777408024575523e-07,
      "loss": 0.1503,
      "step": 9616
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.565536622141539,
      "learning_rate": 3.7720964480832847e-07,
      "loss": 0.1893,
      "step": 9617
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5701190830938097,
      "learning_rate": 3.766456148205422e-07,
      "loss": 0.1763,
      "step": 9618
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.454172498725269,
      "learning_rate": 3.760819903318641e-07,
      "loss": 0.1683,
      "step": 9619
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.71309905860544,
      "learning_rate": 3.755187713917291e-07,
      "loss": 0.5481,
      "step": 9620
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4695110820906914,
      "learning_rate": 3.7495595804953435e-07,
      "loss": 0.1693,
      "step": 9621
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2807987737995765,
      "learning_rate": 3.7439355035464144e-07,
      "loss": 0.1566,
      "step": 9622
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.491860460677599,
      "learning_rate": 3.738315483563792e-07,
      "loss": 0.5379,
      "step": 9623
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3085578677115697,
      "learning_rate": 3.732699521040378e-07,
      "loss": 0.1712,
      "step": 9624
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.518756000565078,
      "learning_rate": 3.727087616468739e-07,
      "loss": 0.1769,
      "step": 9625
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4757084582132387,
      "learning_rate": 3.7214797703410576e-07,
      "loss": 0.1829,
      "step": 9626
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2688712947248246,
      "learning_rate": 3.7158759831492084e-07,
      "loss": 0.1658,
      "step": 9627
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.704790544533859,
      "learning_rate": 3.710276255384654e-07,
      "loss": 0.5975,
      "step": 9628
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2973992790082298,
      "learning_rate": 3.704680587538545e-07,
      "loss": 0.1688,
      "step": 9629
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3718914672254898,
      "learning_rate": 3.699088980101656e-07,
      "loss": 0.165,
      "step": 9630
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5988147148463723,
      "learning_rate": 3.693501433564395e-07,
      "loss": 0.1911,
      "step": 9631
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.511981594752711,
      "learning_rate": 3.687917948416847e-07,
      "loss": 0.1739,
      "step": 9632
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4240646256220606,
      "learning_rate": 3.682338525148699e-07,
      "loss": 0.1585,
      "step": 9633
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3607993631233555,
      "learning_rate": 3.676763164249331e-07,
      "loss": 0.164,
      "step": 9634
    },
    {
      "epoch": 0.88,
      "grad_norm": 4.870482826757756,
      "learning_rate": 3.6711918662077126e-07,
      "loss": 0.6026,
      "step": 9635
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2979140346760467,
      "learning_rate": 3.6656246315125045e-07,
      "loss": 0.1556,
      "step": 9636
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3173846310117159,
      "learning_rate": 3.660061460651981e-07,
      "loss": 0.1698,
      "step": 9637
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2905349083907904,
      "learning_rate": 3.6545023541140643e-07,
      "loss": 0.1433,
      "step": 9638
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.257002586345758,
      "learning_rate": 3.6489473123863363e-07,
      "loss": 0.1508,
      "step": 9639
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3622382102115809,
      "learning_rate": 3.643396335956001e-07,
      "loss": 0.1534,
      "step": 9640
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2868379580413807,
      "learning_rate": 3.6378494253099307e-07,
      "loss": 0.1581,
      "step": 9641
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3840071727590904,
      "learning_rate": 3.6323065809346083e-07,
      "loss": 0.1936,
      "step": 9642
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4243063518097876,
      "learning_rate": 3.6267678033161947e-07,
      "loss": 0.1703,
      "step": 9643
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3115830744811752,
      "learning_rate": 3.6212330929404627e-07,
      "loss": 0.1692,
      "step": 9644
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2012493270607076,
      "learning_rate": 3.615702450292857e-07,
      "loss": 0.1512,
      "step": 9645
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2955677625260806,
      "learning_rate": 3.6101758758584517e-07,
      "loss": 0.1223,
      "step": 9646
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.786526688348888,
      "learning_rate": 3.6046533701219465e-07,
      "loss": 0.601,
      "step": 9647
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3276677464991997,
      "learning_rate": 3.5991349335677215e-07,
      "loss": 0.15,
      "step": 9648
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3912120272454058,
      "learning_rate": 3.5936205666797675e-07,
      "loss": 0.1799,
      "step": 9649
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3690968697051118,
      "learning_rate": 3.588110269941747e-07,
      "loss": 0.1628,
      "step": 9650
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2948446937904636,
      "learning_rate": 3.582604043836929e-07,
      "loss": 0.1783,
      "step": 9651
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.251641708893239,
      "learning_rate": 3.577101888848267e-07,
      "loss": 0.1707,
      "step": 9652
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3246159954283423,
      "learning_rate": 3.571603805458318e-07,
      "loss": 0.1721,
      "step": 9653
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3758336181443314,
      "learning_rate": 3.566109794149319e-07,
      "loss": 0.1625,
      "step": 9654
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.773366222189255,
      "learning_rate": 3.560619855403119e-07,
      "loss": 0.2173,
      "step": 9655
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3725621772188312,
      "learning_rate": 3.5551339897012206e-07,
      "loss": 0.1932,
      "step": 9656
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.303759243830807,
      "learning_rate": 3.549652197524783e-07,
      "loss": 0.1701,
      "step": 9657
    },
    {
      "epoch": 0.88,
      "grad_norm": 5.35183704481413,
      "learning_rate": 3.544174479354584e-07,
      "loss": 0.5101,
      "step": 9658
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.373015290812523,
      "learning_rate": 3.5387008356710715e-07,
      "loss": 0.1638,
      "step": 9659
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3785410928503616,
      "learning_rate": 3.5332312669542956e-07,
      "loss": 0.1772,
      "step": 9660
    },
    {
      "epoch": 0.88,
      "grad_norm": 11.621520512695794,
      "learning_rate": 3.5277657736839997e-07,
      "loss": 0.5013,
      "step": 9661
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4133992619676872,
      "learning_rate": 3.522304356339529e-07,
      "loss": 0.181,
      "step": 9662
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3909107503684883,
      "learning_rate": 3.5168470153998937e-07,
      "loss": 0.1596,
      "step": 9663
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3745689607330027,
      "learning_rate": 3.511393751343739e-07,
      "loss": 0.1604,
      "step": 9664
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4414240030481171,
      "learning_rate": 3.505944564649344e-07,
      "loss": 0.1777,
      "step": 9665
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4827758640453574,
      "learning_rate": 3.500499455794648e-07,
      "loss": 0.1614,
      "step": 9666
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3297532338343216,
      "learning_rate": 3.495058425257214e-07,
      "loss": 0.1528,
      "step": 9667
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.432852043462935,
      "learning_rate": 3.4896214735142706e-07,
      "loss": 0.186,
      "step": 9668
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.518376011877333,
      "learning_rate": 3.4841886010426593e-07,
      "loss": 0.1892,
      "step": 9669
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4522916291823476,
      "learning_rate": 3.478759808318893e-07,
      "loss": 0.1851,
      "step": 9670
    },
    {
      "epoch": 0.88,
      "grad_norm": 6.250219286088677,
      "learning_rate": 3.4733350958191026e-07,
      "loss": 0.5758,
      "step": 9671
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2997281688385325,
      "learning_rate": 3.4679144640190845e-07,
      "loss": 0.1495,
      "step": 9672
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.167012556809159,
      "learning_rate": 3.462497913394258e-07,
      "loss": 0.1351,
      "step": 9673
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4263484460603268,
      "learning_rate": 3.4570854444196777e-07,
      "loss": 0.1832,
      "step": 9674
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2652310780162688,
      "learning_rate": 3.451677057570074e-07,
      "loss": 0.1386,
      "step": 9675
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3778020185847484,
      "learning_rate": 3.4462727533197794e-07,
      "loss": 0.1719,
      "step": 9676
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3819169203829862,
      "learning_rate": 3.440872532142808e-07,
      "loss": 0.2113,
      "step": 9677
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3572733372791412,
      "learning_rate": 3.435476394512771e-07,
      "loss": 0.1507,
      "step": 9678
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4499413204787615,
      "learning_rate": 3.430084340902973e-07,
      "loss": 0.2005,
      "step": 9679
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.672070606123463,
      "learning_rate": 3.424696371786318e-07,
      "loss": 0.6588,
      "step": 9680
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.38830147872937,
      "learning_rate": 3.419312487635362e-07,
      "loss": 0.1433,
      "step": 9681
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.375873608107084,
      "learning_rate": 3.413932688922322e-07,
      "loss": 0.1558,
      "step": 9682
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.292657972534402,
      "learning_rate": 3.4085569761190264e-07,
      "loss": 0.1606,
      "step": 9683
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4276450490593742,
      "learning_rate": 3.4031853496969803e-07,
      "loss": 0.1706,
      "step": 9684
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2601082523388265,
      "learning_rate": 3.397817810127291e-07,
      "loss": 0.1453,
      "step": 9685
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.331077411632531,
      "learning_rate": 3.3924543578807425e-07,
      "loss": 0.1568,
      "step": 9686
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.47366244859435,
      "learning_rate": 3.387094993427731e-07,
      "loss": 0.1994,
      "step": 9687
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3312911871784974,
      "learning_rate": 3.3817397172383303e-07,
      "loss": 0.2027,
      "step": 9688
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3140223507939175,
      "learning_rate": 3.3763885297822153e-07,
      "loss": 0.1478,
      "step": 9689
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.46450636207469,
      "learning_rate": 3.371041431528727e-07,
      "loss": 0.1556,
      "step": 9690
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4800683459008726,
      "learning_rate": 3.3656984229468447e-07,
      "loss": 0.1772,
      "step": 9691
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3742732212490993,
      "learning_rate": 3.360359504505173e-07,
      "loss": 0.1837,
      "step": 9692
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5294734258423166,
      "learning_rate": 3.355024676671992e-07,
      "loss": 0.1863,
      "step": 9693
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5473477428878644,
      "learning_rate": 3.349693939915177e-07,
      "loss": 0.2028,
      "step": 9694
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.424585218008168,
      "learning_rate": 3.3443672947023e-07,
      "loss": 0.1759,
      "step": 9695
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.57976360968374,
      "learning_rate": 3.339044741500508e-07,
      "loss": 0.1482,
      "step": 9696
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2836829730795496,
      "learning_rate": 3.333726280776656e-07,
      "loss": 0.1528,
      "step": 9697
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2976475790295268,
      "learning_rate": 3.328411912997198e-07,
      "loss": 0.1578,
      "step": 9698
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3042524855636068,
      "learning_rate": 3.3231016386282233e-07,
      "loss": 0.1602,
      "step": 9699
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4747128443059476,
      "learning_rate": 3.3177954581355033e-07,
      "loss": 0.1498,
      "step": 9700
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4270966591108676,
      "learning_rate": 3.312493371984404e-07,
      "loss": 0.2025,
      "step": 9701
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3845354243974934,
      "learning_rate": 3.307195380639977e-07,
      "loss": 0.1703,
      "step": 9702
    },
    {
      "epoch": 0.89,
      "grad_norm": 3.4276745463721143,
      "learning_rate": 3.3019014845668663e-07,
      "loss": 0.3179,
      "step": 9703
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.707263147681044,
      "learning_rate": 3.2966116842294007e-07,
      "loss": 0.5467,
      "step": 9704
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4049287040395642,
      "learning_rate": 3.2913259800915196e-07,
      "loss": 0.1468,
      "step": 9705
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3771124915628314,
      "learning_rate": 3.2860443726168254e-07,
      "loss": 0.1724,
      "step": 9706
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5138891146822542,
      "learning_rate": 3.2807668622685475e-07,
      "loss": 0.2336,
      "step": 9707
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.251380096462661,
      "learning_rate": 3.2754934495095483e-07,
      "loss": 0.1415,
      "step": 9708
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.74223058060713,
      "learning_rate": 3.270224134802358e-07,
      "loss": 0.5517,
      "step": 9709
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.440042921148299,
      "learning_rate": 3.2649589186091125e-07,
      "loss": 0.1586,
      "step": 9710
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4276362815471122,
      "learning_rate": 3.25969780139162e-07,
      "loss": 0.1568,
      "step": 9711
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.551241437807585,
      "learning_rate": 3.254440783611307e-07,
      "loss": 0.6544,
      "step": 9712
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5167066891247036,
      "learning_rate": 3.2491878657292643e-07,
      "loss": 0.1866,
      "step": 9713
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4159037262496903,
      "learning_rate": 3.243939048206185e-07,
      "loss": 0.1894,
      "step": 9714
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.371269716042797,
      "learning_rate": 3.238694331502451e-07,
      "loss": 0.1412,
      "step": 9715
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3388015431224778,
      "learning_rate": 3.233453716078039e-07,
      "loss": 0.1678,
      "step": 9716
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3121912232013127,
      "learning_rate": 3.228217202392592e-07,
      "loss": 0.1392,
      "step": 9717
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5065667437594552,
      "learning_rate": 3.2229847909053925e-07,
      "loss": 0.1489,
      "step": 9718
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.340288428797547,
      "learning_rate": 3.217756482075346e-07,
      "loss": 0.1901,
      "step": 9719
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4611555320860297,
      "learning_rate": 3.212532276361025e-07,
      "loss": 0.1719,
      "step": 9720
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2432301136382045,
      "learning_rate": 3.2073121742206117e-07,
      "loss": 0.1659,
      "step": 9721
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7264877749592,
      "learning_rate": 3.202096176111963e-07,
      "loss": 0.1877,
      "step": 9722
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3539952456168762,
      "learning_rate": 3.196884282492546e-07,
      "loss": 0.1402,
      "step": 9723
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.319160734807771,
      "learning_rate": 3.1916764938194676e-07,
      "loss": 0.1663,
      "step": 9724
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5746017836453612,
      "learning_rate": 3.186472810549507e-07,
      "loss": 0.1766,
      "step": 9725
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2675683239313813,
      "learning_rate": 3.1812732331390483e-07,
      "loss": 0.1769,
      "step": 9726
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.130432298772296,
      "learning_rate": 3.1760777620441327e-07,
      "loss": 0.5109,
      "step": 9727
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1828548486815575,
      "learning_rate": 3.170886397720435e-07,
      "loss": 0.1494,
      "step": 9728
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4946186125734031,
      "learning_rate": 3.165699140623285e-07,
      "loss": 0.1396,
      "step": 9729
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5562790486908609,
      "learning_rate": 3.1605159912076186e-07,
      "loss": 0.2114,
      "step": 9730
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3659457456023196,
      "learning_rate": 3.155336949928056e-07,
      "loss": 0.1576,
      "step": 9731
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2424065712001418,
      "learning_rate": 3.1501620172388225e-07,
      "loss": 0.1639,
      "step": 9732
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4287695664014155,
      "learning_rate": 3.144991193593783e-07,
      "loss": 0.1867,
      "step": 9733
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2471536878500162,
      "learning_rate": 3.139824479446474e-07,
      "loss": 0.1581,
      "step": 9734
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5060821248741638,
      "learning_rate": 3.1346618752500335e-07,
      "loss": 0.1639,
      "step": 9735
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.340997482570412,
      "learning_rate": 3.1295033814572716e-07,
      "loss": 0.1514,
      "step": 9736
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2073354248763104,
      "learning_rate": 3.1243489985206097e-07,
      "loss": 0.1428,
      "step": 9737
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3301638275612515,
      "learning_rate": 3.119198726892136e-07,
      "loss": 0.1824,
      "step": 9738
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1171315532733597,
      "learning_rate": 3.1140525670235454e-07,
      "loss": 0.1441,
      "step": 9739
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3717515041998465,
      "learning_rate": 3.108910519366204e-07,
      "loss": 0.1647,
      "step": 9740
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2657107938001941,
      "learning_rate": 3.103772584371106e-07,
      "loss": 0.1522,
      "step": 9741
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2274168877278449,
      "learning_rate": 3.0986387624888647e-07,
      "loss": 0.1439,
      "step": 9742
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3965930987159918,
      "learning_rate": 3.0935090541697743e-07,
      "loss": 0.1736,
      "step": 9743
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3306616778727898,
      "learning_rate": 3.0883834598637197e-07,
      "loss": 0.1575,
      "step": 9744
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4839971314901108,
      "learning_rate": 3.0832619800202746e-07,
      "loss": 0.1927,
      "step": 9745
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.306872882970711,
      "learning_rate": 3.0781446150886074e-07,
      "loss": 0.1432,
      "step": 9746
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6084913978680968,
      "learning_rate": 3.0730313655175647e-07,
      "loss": 0.1951,
      "step": 9747
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.65970499821819,
      "learning_rate": 3.067922231755588e-07,
      "loss": 0.6265,
      "step": 9748
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2867219632319509,
      "learning_rate": 3.0628172142508074e-07,
      "loss": 0.1756,
      "step": 9749
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4548999239704101,
      "learning_rate": 3.0577163134509536e-07,
      "loss": 0.1875,
      "step": 9750
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5366160705478182,
      "learning_rate": 3.052619529803402e-07,
      "loss": 0.1834,
      "step": 9751
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4644111879564128,
      "learning_rate": 3.047526863755196e-07,
      "loss": 0.1568,
      "step": 9752
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3505809334482393,
      "learning_rate": 3.0424383157529716e-07,
      "loss": 0.1764,
      "step": 9753
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5520382586951036,
      "learning_rate": 3.037353886243055e-07,
      "loss": 0.1827,
      "step": 9754
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4458199683381878,
      "learning_rate": 3.032273575671357e-07,
      "loss": 0.2087,
      "step": 9755
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.276069834756909,
      "learning_rate": 3.0271973844834825e-07,
      "loss": 0.5174,
      "step": 9756
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3298962414522706,
      "learning_rate": 3.0221253131246243e-07,
      "loss": 0.1773,
      "step": 9757
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3500788738310787,
      "learning_rate": 3.017057362039655e-07,
      "loss": 0.165,
      "step": 9758
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.488791076286128,
      "learning_rate": 3.0119935316730574e-07,
      "loss": 0.1675,
      "step": 9759
    },
    {
      "epoch": 0.89,
      "grad_norm": 18.535186774218566,
      "learning_rate": 3.0069338224689604e-07,
      "loss": 0.6339,
      "step": 9760
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2439904665187758,
      "learning_rate": 3.001878234871147e-07,
      "loss": 0.1472,
      "step": 9761
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4915432688291816,
      "learning_rate": 2.996826769323008e-07,
      "loss": 0.1868,
      "step": 9762
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.446067385590648,
      "learning_rate": 2.9917794262676156e-07,
      "loss": 0.1709,
      "step": 9763
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3603034107563585,
      "learning_rate": 2.986736206147628e-07,
      "loss": 0.1924,
      "step": 9764
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4834963004210022,
      "learning_rate": 2.98169710940539e-07,
      "loss": 0.1845,
      "step": 9765
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3457442204402812,
      "learning_rate": 2.976662136482855e-07,
      "loss": 0.1619,
      "step": 9766
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.499377902351153,
      "learning_rate": 2.9716312878216194e-07,
      "loss": 0.1558,
      "step": 9767
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3004638282595145,
      "learning_rate": 2.9666045638629406e-07,
      "loss": 0.1717,
      "step": 9768
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3790885666887782,
      "learning_rate": 2.961581965047672e-07,
      "loss": 0.1846,
      "step": 9769
    },
    {
      "epoch": 0.89,
      "grad_norm": 5.688496866168599,
      "learning_rate": 2.9565634918163444e-07,
      "loss": 0.5533,
      "step": 9770
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3783797518415577,
      "learning_rate": 2.951549144609106e-07,
      "loss": 0.1603,
      "step": 9771
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.784832210397092,
      "learning_rate": 2.946538923865755e-07,
      "loss": 0.6331,
      "step": 9772
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2876606001313056,
      "learning_rate": 2.941532830025712e-07,
      "loss": 0.1581,
      "step": 9773
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3242648039547933,
      "learning_rate": 2.936530863528053e-07,
      "loss": 0.1866,
      "step": 9774
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6079685884253865,
      "learning_rate": 2.9315330248114835e-07,
      "loss": 0.1484,
      "step": 9775
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.439037990259018,
      "learning_rate": 2.926539314314336e-07,
      "loss": 0.2082,
      "step": 9776
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5097707325968894,
      "learning_rate": 2.921549732474599e-07,
      "loss": 0.1635,
      "step": 9777
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3781267586514845,
      "learning_rate": 2.916564279729894e-07,
      "loss": 0.1573,
      "step": 9778
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3169312831575923,
      "learning_rate": 2.911582956517489e-07,
      "loss": 0.1785,
      "step": 9779
    },
    {
      "epoch": 0.89,
      "grad_norm": 7.189293319064046,
      "learning_rate": 2.9066057632742674e-07,
      "loss": 0.4994,
      "step": 9780
    },
    {
      "epoch": 0.89,
      "grad_norm": 6.381119309889038,
      "learning_rate": 2.901632700436757e-07,
      "loss": 0.4098,
      "step": 9781
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3019180403147652,
      "learning_rate": 2.896663768441138e-07,
      "loss": 0.1984,
      "step": 9782
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.335435240353899,
      "learning_rate": 2.8916989677232166e-07,
      "loss": 0.1982,
      "step": 9783
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5221487272251202,
      "learning_rate": 2.886738298718439e-07,
      "loss": 0.2137,
      "step": 9784
    },
    {
      "epoch": 0.89,
      "grad_norm": 9.06864977718582,
      "learning_rate": 2.8817817618618846e-07,
      "loss": 0.5935,
      "step": 9785
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3388350852209867,
      "learning_rate": 2.87682935758829e-07,
      "loss": 0.1659,
      "step": 9786
    },
    {
      "epoch": 0.89,
      "grad_norm": 8.508503471753045,
      "learning_rate": 2.87188108633199e-07,
      "loss": 0.6376,
      "step": 9787
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1946589299157373,
      "learning_rate": 2.866936948527005e-07,
      "loss": 0.1449,
      "step": 9788
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.429317122634134,
      "learning_rate": 2.8619969446069594e-07,
      "loss": 0.1622,
      "step": 9789
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3420321529822594,
      "learning_rate": 2.857061075005113e-07,
      "loss": 0.151,
      "step": 9790
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5419253447185788,
      "learning_rate": 2.852129340154397e-07,
      "loss": 0.1693,
      "step": 9791
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2604108606591393,
      "learning_rate": 2.847201740487332e-07,
      "loss": 0.1684,
      "step": 9792
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4887457734365268,
      "learning_rate": 2.842278276436128e-07,
      "loss": 0.1789,
      "step": 9793
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4143562169663804,
      "learning_rate": 2.8373589484325836e-07,
      "loss": 0.1606,
      "step": 9794
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3107896544843973,
      "learning_rate": 2.8324437569081707e-07,
      "loss": 0.1783,
      "step": 9795
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4638683973676099,
      "learning_rate": 2.827532702293978e-07,
      "loss": 0.1782,
      "step": 9796
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.455709384252527,
      "learning_rate": 2.822625785020744e-07,
      "loss": 0.173,
      "step": 9797
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4549711278125408,
      "learning_rate": 2.817723005518835e-07,
      "loss": 0.1821,
      "step": 9798
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4084759277015924,
      "learning_rate": 2.812824364218247e-07,
      "loss": 0.1894,
      "step": 9799
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.310085708827099,
      "learning_rate": 2.807929861548642e-07,
      "loss": 0.1613,
      "step": 9800
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.372866705721366,
      "learning_rate": 2.803039497939281e-07,
      "loss": 0.1558,
      "step": 9801
    },
    {
      "epoch": 0.9,
      "grad_norm": 7.836718917212141,
      "learning_rate": 2.7981532738191044e-07,
      "loss": 0.6276,
      "step": 9802
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.378225961130227,
      "learning_rate": 2.793271189616642e-07,
      "loss": 0.1688,
      "step": 9803
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4982920923519647,
      "learning_rate": 2.7883932457601127e-07,
      "loss": 0.1946,
      "step": 9804
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3756990242705827,
      "learning_rate": 2.7835194426773236e-07,
      "loss": 0.1915,
      "step": 9805
    },
    {
      "epoch": 0.9,
      "grad_norm": 15.91227290236527,
      "learning_rate": 2.778649780795739e-07,
      "loss": 0.571,
      "step": 9806
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3772097682169253,
      "learning_rate": 2.773784260542478e-07,
      "loss": 0.1637,
      "step": 9807
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3707905178028317,
      "learning_rate": 2.7689228823442603e-07,
      "loss": 0.1589,
      "step": 9808
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3137842967927598,
      "learning_rate": 2.7640656466274785e-07,
      "loss": 0.162,
      "step": 9809
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4226450424560293,
      "learning_rate": 2.75921255381813e-07,
      "loss": 0.191,
      "step": 9810
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5965588685776397,
      "learning_rate": 2.754363604341875e-07,
      "loss": 0.2101,
      "step": 9811
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.233548646007321,
      "learning_rate": 2.74951879862399e-07,
      "loss": 0.1344,
      "step": 9812
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.365365077843215,
      "learning_rate": 2.744678137089407e-07,
      "loss": 0.1789,
      "step": 9813
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3708734045082969,
      "learning_rate": 2.7398416201626756e-07,
      "loss": 0.1251,
      "step": 9814
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.293462008496437,
      "learning_rate": 2.7350092482679836e-07,
      "loss": 0.1527,
      "step": 9815
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5015192023202792,
      "learning_rate": 2.730181021829187e-07,
      "loss": 0.2103,
      "step": 9816
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3995644675735837,
      "learning_rate": 2.7253569412697244e-07,
      "loss": 0.1905,
      "step": 9817
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5022954549112704,
      "learning_rate": 2.720537007012725e-07,
      "loss": 0.1811,
      "step": 9818
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.549203933026406,
      "learning_rate": 2.71572121948091e-07,
      "loss": 0.1977,
      "step": 9819
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3127923242280288,
      "learning_rate": 2.710909579096671e-07,
      "loss": 0.1563,
      "step": 9820
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5048243421258856,
      "learning_rate": 2.706102086282003e-07,
      "loss": 0.168,
      "step": 9821
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2638496011400697,
      "learning_rate": 2.7012987414585803e-07,
      "loss": 0.1699,
      "step": 9822
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3910832369082238,
      "learning_rate": 2.696499545047671e-07,
      "loss": 0.1824,
      "step": 9823
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5309998799007398,
      "learning_rate": 2.6917044974701933e-07,
      "loss": 0.1733,
      "step": 9824
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3802455579879231,
      "learning_rate": 2.686913599146723e-07,
      "loss": 0.2134,
      "step": 9825
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5583095410781278,
      "learning_rate": 2.6821268504974353e-07,
      "loss": 0.1762,
      "step": 9826
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3541078288007429,
      "learning_rate": 2.677344251942171e-07,
      "loss": 0.18,
      "step": 9827
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.292123458636392,
      "learning_rate": 2.67256580390039e-07,
      "loss": 0.2002,
      "step": 9828
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3373556291669724,
      "learning_rate": 2.667791506791201e-07,
      "loss": 0.1615,
      "step": 9829
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.825261588358136,
      "learning_rate": 2.6630213610333353e-07,
      "loss": 0.5727,
      "step": 9830
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.247617307120198,
      "learning_rate": 2.6582553670451706e-07,
      "loss": 0.1868,
      "step": 9831
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3235180251689764,
      "learning_rate": 2.653493525244721e-07,
      "loss": 0.1568,
      "step": 9832
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.062617072254113,
      "learning_rate": 2.648735836049615e-07,
      "loss": 0.192,
      "step": 9833
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.747108921664362,
      "learning_rate": 2.6439822998771516e-07,
      "loss": 0.5096,
      "step": 9834
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2547792569964717,
      "learning_rate": 2.6392329171442356e-07,
      "loss": 0.1314,
      "step": 9835
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2645075872248313,
      "learning_rate": 2.634487688267429e-07,
      "loss": 0.1941,
      "step": 9836
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.412010736863739,
      "learning_rate": 2.629746613662909e-07,
      "loss": 0.1923,
      "step": 9837
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4421989166432927,
      "learning_rate": 2.625009693746516e-07,
      "loss": 0.1478,
      "step": 9838
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.450293742970325,
      "learning_rate": 2.6202769289336905e-07,
      "loss": 0.1438,
      "step": 9839
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3558047396094974,
      "learning_rate": 2.6155483196395495e-07,
      "loss": 0.161,
      "step": 9840
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3833727743302904,
      "learning_rate": 2.6108238662788057e-07,
      "loss": 0.1325,
      "step": 9841
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3595638889177302,
      "learning_rate": 2.6061035692658276e-07,
      "loss": 0.1477,
      "step": 9842
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3556164545027638,
      "learning_rate": 2.6013874290146237e-07,
      "loss": 0.1673,
      "step": 9843
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5057716913508652,
      "learning_rate": 2.596675445938823e-07,
      "loss": 0.1657,
      "step": 9844
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2848650027312465,
      "learning_rate": 2.5919676204517073e-07,
      "loss": 0.1649,
      "step": 9845
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.395450463638806,
      "learning_rate": 2.5872639529661737e-07,
      "loss": 0.1773,
      "step": 9846
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.150028693452673,
      "learning_rate": 2.5825644438947803e-07,
      "loss": 0.6961,
      "step": 9847
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5193700105326395,
      "learning_rate": 2.577869093649693e-07,
      "loss": 0.228,
      "step": 9848
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4373096633763112,
      "learning_rate": 2.573177902642726e-07,
      "loss": 0.1447,
      "step": 9849
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.350870319120556,
      "learning_rate": 2.568490871285334e-07,
      "loss": 0.1535,
      "step": 9850
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.459787350207169,
      "learning_rate": 2.563807999988588e-07,
      "loss": 0.1371,
      "step": 9851
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4631754437714126,
      "learning_rate": 2.5591292891632315e-07,
      "loss": 0.1701,
      "step": 9852
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.458375867957808,
      "learning_rate": 2.5544547392195917e-07,
      "loss": 0.1846,
      "step": 9853
    },
    {
      "epoch": 0.9,
      "grad_norm": 12.209169368003906,
      "learning_rate": 2.54978435056768e-07,
      "loss": 0.6632,
      "step": 9854
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.11872403251276,
      "learning_rate": 2.545118123617102e-07,
      "loss": 0.5591,
      "step": 9855
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3000593064297763,
      "learning_rate": 2.5404560587771356e-07,
      "loss": 0.1666,
      "step": 9856
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3455844238045969,
      "learning_rate": 2.5357981564566647e-07,
      "loss": 0.18,
      "step": 9857
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.368674744457856,
      "learning_rate": 2.531144417064213e-07,
      "loss": 0.179,
      "step": 9858
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.452134380846958,
      "learning_rate": 2.526494841007959e-07,
      "loss": 0.1797,
      "step": 9859
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2701038902918873,
      "learning_rate": 2.521849428695683e-07,
      "loss": 0.1492,
      "step": 9860
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4704150837217296,
      "learning_rate": 2.5172081805348413e-07,
      "loss": 0.1924,
      "step": 9861
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3654086625763904,
      "learning_rate": 2.512571096932481e-07,
      "loss": 0.1914,
      "step": 9862
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.659228813424832,
      "learning_rate": 2.5079381782953215e-07,
      "loss": 0.2029,
      "step": 9863
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3343708901752762,
      "learning_rate": 2.5033094250296877e-07,
      "loss": 0.173,
      "step": 9864
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.420185696002089,
      "learning_rate": 2.4986848375415653e-07,
      "loss": 0.152,
      "step": 9865
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3596247830202932,
      "learning_rate": 2.4940644162365523e-07,
      "loss": 0.1431,
      "step": 9866
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5249164495485195,
      "learning_rate": 2.4894481615198916e-07,
      "loss": 0.154,
      "step": 9867
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4025847600071588,
      "learning_rate": 2.484836073796465e-07,
      "loss": 0.186,
      "step": 9868
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3490736430077341,
      "learning_rate": 2.4802281534707706e-07,
      "loss": 0.1512,
      "step": 9869
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3706108521317935,
      "learning_rate": 2.475624400946974e-07,
      "loss": 0.1651,
      "step": 9870
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2794666894972806,
      "learning_rate": 2.471024816628836e-07,
      "loss": 0.1678,
      "step": 9871
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5370599191062893,
      "learning_rate": 2.466429400919784e-07,
      "loss": 0.1835,
      "step": 9872
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5690400570800858,
      "learning_rate": 2.4618381542228565e-07,
      "loss": 0.2108,
      "step": 9873
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6184651720767838,
      "learning_rate": 2.4572510769407473e-07,
      "loss": 0.1968,
      "step": 9874
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.540257612881793,
      "learning_rate": 2.4526681694757694e-07,
      "loss": 0.1735,
      "step": 9875
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.208537667284189,
      "learning_rate": 2.448089432229867e-07,
      "loss": 0.6525,
      "step": 9876
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.312859689827257,
      "learning_rate": 2.4435148656046347e-07,
      "loss": 0.154,
      "step": 9877
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.508065095733836,
      "learning_rate": 2.438944470001287e-07,
      "loss": 0.1535,
      "step": 9878
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3268957897706823,
      "learning_rate": 2.43437824582069e-07,
      "loss": 0.1403,
      "step": 9879
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.277661955034315,
      "learning_rate": 2.4298161934633136e-07,
      "loss": 0.1286,
      "step": 9880
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2344605093694814,
      "learning_rate": 2.4252583133292927e-07,
      "loss": 0.1226,
      "step": 9881
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5402505913287765,
      "learning_rate": 2.42070460581838e-07,
      "loss": 0.1658,
      "step": 9882
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3200288902721293,
      "learning_rate": 2.416155071329973e-07,
      "loss": 0.1684,
      "step": 9883
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4210692587929397,
      "learning_rate": 2.411609710263091e-07,
      "loss": 0.1898,
      "step": 9884
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3005726193695895,
      "learning_rate": 2.407068523016387e-07,
      "loss": 0.1805,
      "step": 9885
    },
    {
      "epoch": 0.9,
      "grad_norm": 9.504641414208463,
      "learning_rate": 2.4025315099881595e-07,
      "loss": 0.5961,
      "step": 9886
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4635696687806905,
      "learning_rate": 2.397998671576335e-07,
      "loss": 0.1717,
      "step": 9887
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3026755614235201,
      "learning_rate": 2.393470008178472e-07,
      "loss": 0.1521,
      "step": 9888
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.435230536001582,
      "learning_rate": 2.3889455201917655e-07,
      "loss": 0.195,
      "step": 9889
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4576547385009957,
      "learning_rate": 2.384425208013047e-07,
      "loss": 0.2057,
      "step": 9890
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.382240507134395,
      "learning_rate": 2.379909072038772e-07,
      "loss": 0.2109,
      "step": 9891
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.513362623827593,
      "learning_rate": 2.3753971126650344e-07,
      "loss": 0.2265,
      "step": 9892
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5236790162433875,
      "learning_rate": 2.3708893302875736e-07,
      "loss": 0.194,
      "step": 9893
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2927971059801182,
      "learning_rate": 2.3663857253017398e-07,
      "loss": 0.1659,
      "step": 9894
    },
    {
      "epoch": 0.9,
      "grad_norm": 5.553156516298993,
      "learning_rate": 2.3618862981025393e-07,
      "loss": 0.5722,
      "step": 9895
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4069041780627978,
      "learning_rate": 2.3573910490845898e-07,
      "loss": 0.184,
      "step": 9896
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.249759452107933,
      "learning_rate": 2.3528999786421758e-07,
      "loss": 0.1538,
      "step": 9897
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2816545775265553,
      "learning_rate": 2.3484130871691713e-07,
      "loss": 0.1542,
      "step": 9898
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3238224683990525,
      "learning_rate": 2.3439303750591224e-07,
      "loss": 0.1504,
      "step": 9899
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4039354886373532,
      "learning_rate": 2.339451842705187e-07,
      "loss": 0.1597,
      "step": 9900
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4044145946264512,
      "learning_rate": 2.3349774905001566e-07,
      "loss": 0.1895,
      "step": 9901
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.518676949846834,
      "learning_rate": 2.3305073188364723e-07,
      "loss": 0.1747,
      "step": 9902
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.206879575069779,
      "learning_rate": 2.3260413281061877e-07,
      "loss": 0.1504,
      "step": 9903
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2015194037212165,
      "learning_rate": 2.3215795187010172e-07,
      "loss": 0.1165,
      "step": 9904
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4025233930219982,
      "learning_rate": 2.3171218910122695e-07,
      "loss": 0.1834,
      "step": 9905
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.271116754604078,
      "learning_rate": 2.3126684454309266e-07,
      "loss": 0.1716,
      "step": 9906
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3691997295420564,
      "learning_rate": 2.3082191823475708e-07,
      "loss": 0.1549,
      "step": 9907
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3143055860084325,
      "learning_rate": 2.3037741021524452e-07,
      "loss": 0.1362,
      "step": 9908
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6135590868870069,
      "learning_rate": 2.2993332052354044e-07,
      "loss": 0.2027,
      "step": 9909
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.259148619340648,
      "learning_rate": 2.2948964919859428e-07,
      "loss": 0.1631,
      "step": 9910
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3542832209451876,
      "learning_rate": 2.2904639627932045e-07,
      "loss": 0.1339,
      "step": 9911
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3235409755199468,
      "learning_rate": 2.2860356180459286e-07,
      "loss": 0.1237,
      "step": 9912
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2228219801056726,
      "learning_rate": 2.2816114581325377e-07,
      "loss": 0.1316,
      "step": 9913
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4426706746612588,
      "learning_rate": 2.277191483441038e-07,
      "loss": 0.1766,
      "step": 9914
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.593279498892089,
      "learning_rate": 2.272775694359103e-07,
      "loss": 0.1974,
      "step": 9915
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3855347599314165,
      "learning_rate": 2.2683640912740168e-07,
      "loss": 0.1595,
      "step": 9916
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6908757922552389,
      "learning_rate": 2.2639566745727203e-07,
      "loss": 0.1947,
      "step": 9917
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.582545344901536,
      "learning_rate": 2.2595534446417656e-07,
      "loss": 0.1808,
      "step": 9918
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3761902127779568,
      "learning_rate": 2.2551544018673376e-07,
      "loss": 0.1952,
      "step": 9919
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.26717929262116,
      "learning_rate": 2.2507595466352783e-07,
      "loss": 0.5013,
      "step": 9920
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2494434934276457,
      "learning_rate": 2.2463688793310345e-07,
      "loss": 0.1297,
      "step": 9921
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.350102241364642,
      "learning_rate": 2.241982400339704e-07,
      "loss": 0.1712,
      "step": 9922
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6102191295907065,
      "learning_rate": 2.237600110046001e-07,
      "loss": 0.1598,
      "step": 9923
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3318133420516833,
      "learning_rate": 2.2332220088342903e-07,
      "loss": 0.1804,
      "step": 9924
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1637384653553926,
      "learning_rate": 2.228848097088554e-07,
      "loss": 0.1238,
      "step": 9925
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.056205232048585,
      "learning_rate": 2.2244783751924237e-07,
      "loss": 0.4968,
      "step": 9926
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5779664412406547,
      "learning_rate": 2.220112843529143e-07,
      "loss": 0.185,
      "step": 9927
    },
    {
      "epoch": 0.91,
      "grad_norm": 7.455433952833147,
      "learning_rate": 2.2157515024816002e-07,
      "loss": 0.6175,
      "step": 9928
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2546091991738952,
      "learning_rate": 2.2113943524323167e-07,
      "loss": 0.1204,
      "step": 9929
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2625346270370108,
      "learning_rate": 2.2070413937634427e-07,
      "loss": 0.1655,
      "step": 9930
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.529455740071935,
      "learning_rate": 2.2026926268567616e-07,
      "loss": 0.1962,
      "step": 9931
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.426446390098119,
      "learning_rate": 2.198348052093685e-07,
      "loss": 0.1821,
      "step": 9932
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2632871477847645,
      "learning_rate": 2.1940076698552748e-07,
      "loss": 0.1913,
      "step": 9933
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3389924036910843,
      "learning_rate": 2.1896714805222041e-07,
      "loss": 0.1965,
      "step": 9934
    },
    {
      "epoch": 0.91,
      "grad_norm": 11.134912414698745,
      "learning_rate": 2.1853394844747743e-07,
      "loss": 0.5688,
      "step": 9935
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5119012397328715,
      "learning_rate": 2.1810116820929427e-07,
      "loss": 0.1768,
      "step": 9936
    },
    {
      "epoch": 0.91,
      "grad_norm": 9.345699416184857,
      "learning_rate": 2.1766880737562833e-07,
      "loss": 0.52,
      "step": 9937
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.515594481830176,
      "learning_rate": 2.1723686598440098e-07,
      "loss": 0.1883,
      "step": 9938
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3254401875535187,
      "learning_rate": 2.1680534407349574e-07,
      "loss": 0.1516,
      "step": 9939
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6392960066622846,
      "learning_rate": 2.163742416807607e-07,
      "loss": 0.2409,
      "step": 9940
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.253996042243659,
      "learning_rate": 2.159435588440051e-07,
      "loss": 0.1642,
      "step": 9941
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4452975147564044,
      "learning_rate": 2.155132956010042e-07,
      "loss": 0.1817,
      "step": 9942
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.466402459683431,
      "learning_rate": 2.1508345198949453e-07,
      "loss": 0.192,
      "step": 9943
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3891498293432294,
      "learning_rate": 2.146540280471754e-07,
      "loss": 0.1715,
      "step": 9944
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2491814176436182,
      "learning_rate": 2.1422502381171163e-07,
      "loss": 0.1259,
      "step": 9945
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4187197427506937,
      "learning_rate": 2.1379643932072812e-07,
      "loss": 0.1592,
      "step": 9946
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4243633133328781,
      "learning_rate": 2.1336827461181597e-07,
      "loss": 0.1474,
      "step": 9947
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.666602991078262,
      "learning_rate": 2.1294052972252677e-07,
      "loss": 0.2136,
      "step": 9948
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5181432714600667,
      "learning_rate": 2.1251320469037827e-07,
      "loss": 0.1725,
      "step": 9949
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.392169113548567,
      "learning_rate": 2.1208629955284833e-07,
      "loss": 0.1445,
      "step": 9950
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.622863372679212,
      "learning_rate": 2.1165981434738025e-07,
      "loss": 0.6615,
      "step": 9951
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4653959208929743,
      "learning_rate": 2.1123374911137918e-07,
      "loss": 0.1654,
      "step": 9952
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3561612148351803,
      "learning_rate": 2.1080810388221406e-07,
      "loss": 0.1591,
      "step": 9953
    },
    {
      "epoch": 0.91,
      "grad_norm": 10.553549404684611,
      "learning_rate": 2.1038287869721674e-07,
      "loss": 0.5906,
      "step": 9954
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3881864253440641,
      "learning_rate": 2.099580735936818e-07,
      "loss": 0.2239,
      "step": 9955
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3351621932811515,
      "learning_rate": 2.0953368860886892e-07,
      "loss": 0.1671,
      "step": 9956
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.481557572658908,
      "learning_rate": 2.0910972377999773e-07,
      "loss": 0.1598,
      "step": 9957
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.396048696335575,
      "learning_rate": 2.0868617914425403e-07,
      "loss": 0.1379,
      "step": 9958
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4553344260615888,
      "learning_rate": 2.0826305473878484e-07,
      "loss": 0.2197,
      "step": 9959
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.477575293462935,
      "learning_rate": 2.0784035060070207e-07,
      "loss": 0.1877,
      "step": 9960
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.208342873635631,
      "learning_rate": 2.0741806676707887e-07,
      "loss": 0.1345,
      "step": 9961
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3863386522652672,
      "learning_rate": 2.0699620327495174e-07,
      "loss": 0.1791,
      "step": 9962
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4109195557403378,
      "learning_rate": 2.0657476016132217e-07,
      "loss": 0.1901,
      "step": 9963
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.480144672550358,
      "learning_rate": 2.0615373746315225e-07,
      "loss": 0.1849,
      "step": 9964
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6122380055555416,
      "learning_rate": 2.0573313521737027e-07,
      "loss": 0.1782,
      "step": 9965
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.59311474747289,
      "learning_rate": 2.0531295346086387e-07,
      "loss": 0.5564,
      "step": 9966
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2547775345722678,
      "learning_rate": 2.0489319223048753e-07,
      "loss": 0.1719,
      "step": 9967
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2755401368317756,
      "learning_rate": 2.0447385156305565e-07,
      "loss": 0.1786,
      "step": 9968
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4500738063252574,
      "learning_rate": 2.0405493149534828e-07,
      "loss": 0.1858,
      "step": 9969
    },
    {
      "epoch": 0.91,
      "grad_norm": 8.70945513559887,
      "learning_rate": 2.0363643206410765e-07,
      "loss": 0.5877,
      "step": 9970
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3075612258669627,
      "learning_rate": 2.0321835330603724e-07,
      "loss": 0.1473,
      "step": 9971
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.38097722606896,
      "learning_rate": 2.0280069525780766e-07,
      "loss": 0.1644,
      "step": 9972
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2731567531482892,
      "learning_rate": 2.0238345795604798e-07,
      "loss": 0.1887,
      "step": 9973
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.947787132369203,
      "learning_rate": 2.01966641437355e-07,
      "loss": 0.5815,
      "step": 9974
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.593388824359585,
      "learning_rate": 2.0155024573828452e-07,
      "loss": 0.1798,
      "step": 9975
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3695397779998768,
      "learning_rate": 2.0113427089535896e-07,
      "loss": 0.1723,
      "step": 9976
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.320748235135971,
      "learning_rate": 2.007187169450603e-07,
      "loss": 0.1596,
      "step": 9977
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3965183756674133,
      "learning_rate": 2.0030358392383598e-07,
      "loss": 0.1869,
      "step": 9978
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2436690993182502,
      "learning_rate": 1.9988887186809692e-07,
      "loss": 0.1223,
      "step": 9979
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3810081228912807,
      "learning_rate": 1.9947458081421457e-07,
      "loss": 0.1745,
      "step": 9980
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3314570170426785,
      "learning_rate": 1.990607107985265e-07,
      "loss": 0.1628,
      "step": 9981
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4379281579842267,
      "learning_rate": 1.9864726185733097e-07,
      "loss": 0.1802,
      "step": 9982
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3299953783533378,
      "learning_rate": 1.9823423402689057e-07,
      "loss": 0.1729,
      "step": 9983
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3206111079012206,
      "learning_rate": 1.9782162734343026e-07,
      "loss": 0.1493,
      "step": 9984
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4944598270177967,
      "learning_rate": 1.9740944184313882e-07,
      "loss": 0.1831,
      "step": 9985
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3661176380378621,
      "learning_rate": 1.9699767756216737e-07,
      "loss": 0.1722,
      "step": 9986
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3061765545210533,
      "learning_rate": 1.9658633453663034e-07,
      "loss": 0.1405,
      "step": 9987
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2571268710474608,
      "learning_rate": 1.961754128026061e-07,
      "loss": 0.152,
      "step": 9988
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3218283687802026,
      "learning_rate": 1.957649123961336e-07,
      "loss": 0.1653,
      "step": 9989
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2777030633996662,
      "learning_rate": 1.9535483335321848e-07,
      "loss": 0.1348,
      "step": 9990
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4159716372955018,
      "learning_rate": 1.9494517570982529e-07,
      "loss": 0.1829,
      "step": 9991
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4057326871767724,
      "learning_rate": 1.9453593950188587e-07,
      "loss": 0.1833,
      "step": 9992
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.886883949749345,
      "learning_rate": 1.941271247652915e-07,
      "loss": 0.4811,
      "step": 9993
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2657350313068199,
      "learning_rate": 1.937187315358985e-07,
      "loss": 0.1515,
      "step": 9994
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4274314147201255,
      "learning_rate": 1.9331075984952595e-07,
      "loss": 0.1865,
      "step": 9995
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3684947235722285,
      "learning_rate": 1.9290320974195476e-07,
      "loss": 0.1791,
      "step": 9996
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3208405568754493,
      "learning_rate": 1.924960812489307e-07,
      "loss": 0.1672,
      "step": 9997
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3678681413409515,
      "learning_rate": 1.9208937440616082e-07,
      "loss": 0.1744,
      "step": 9998
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5563687169733553,
      "learning_rate": 1.9168308924931712e-07,
      "loss": 0.189,
      "step": 9999
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3440066892580138,
      "learning_rate": 1.9127722581403275e-07,
      "loss": 0.1734,
      "step": 10000
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.492042288393485,
      "learning_rate": 1.908717841359048e-07,
      "loss": 0.1932,
      "step": 10001
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3736208674192667,
      "learning_rate": 1.9046676425049316e-07,
      "loss": 0.1747,
      "step": 10002
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7475111572494846,
      "learning_rate": 1.9006216619332163e-07,
      "loss": 0.1719,
      "step": 10003
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.233068302447286,
      "learning_rate": 1.8965798999987518e-07,
      "loss": 0.1587,
      "step": 10004
    },
    {
      "epoch": 0.91,
      "grad_norm": 6.522397121841394,
      "learning_rate": 1.8925423570560265e-07,
      "loss": 0.4635,
      "step": 10005
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.472765305457941,
      "learning_rate": 1.8885090334591626e-07,
      "loss": 0.1844,
      "step": 10006
    },
    {
      "epoch": 0.91,
      "grad_norm": 5.398859885665314,
      "learning_rate": 1.884479929561911e-07,
      "loss": 0.5326,
      "step": 10007
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2710272486595,
      "learning_rate": 1.8804550457176552e-07,
      "loss": 0.1589,
      "step": 10008
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4558739300324872,
      "learning_rate": 1.8764343822793962e-07,
      "loss": 0.1688,
      "step": 10009
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.188975388345186,
      "learning_rate": 1.8724179395997798e-07,
      "loss": 0.1419,
      "step": 10010
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.10934392771702,
      "learning_rate": 1.8684057180310688e-07,
      "loss": 0.1464,
      "step": 10011
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5073129534828167,
      "learning_rate": 1.8643977179251704e-07,
      "loss": 0.1755,
      "step": 10012
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.469039147862632,
      "learning_rate": 1.8603939396336036e-07,
      "loss": 0.146,
      "step": 10013
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4490934301591163,
      "learning_rate": 1.8563943835075315e-07,
      "loss": 0.1744,
      "step": 10014
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4061552003267381,
      "learning_rate": 1.8523990498977407e-07,
      "loss": 0.1658,
      "step": 10015
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.44982816776585,
      "learning_rate": 1.848407939154645e-07,
      "loss": 0.1567,
      "step": 10016
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3046850404748214,
      "learning_rate": 1.8444210516283035e-07,
      "loss": 0.1551,
      "step": 10017
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4613651316498033,
      "learning_rate": 1.8404383876683806e-07,
      "loss": 0.18,
      "step": 10018
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3708168493444217,
      "learning_rate": 1.8364599476241862e-07,
      "loss": 0.1554,
      "step": 10019
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4200409587668885,
      "learning_rate": 1.8324857318446632e-07,
      "loss": 0.1935,
      "step": 10020
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2590610724409363,
      "learning_rate": 1.8285157406783605e-07,
      "loss": 0.1412,
      "step": 10021
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.189377516135267,
      "learning_rate": 1.8245499744734885e-07,
      "loss": 0.1583,
      "step": 10022
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.361145190199787,
      "learning_rate": 1.8205884335778634e-07,
      "loss": 0.1577,
      "step": 10023
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6083147171728776,
      "learning_rate": 1.8166311183389463e-07,
      "loss": 0.2274,
      "step": 10024
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4435858560924621,
      "learning_rate": 1.8126780291038037e-07,
      "loss": 0.1872,
      "step": 10025
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4055921386669101,
      "learning_rate": 1.8087291662191697e-07,
      "loss": 0.1689,
      "step": 10026
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3103584419359864,
      "learning_rate": 1.8047845300313726e-07,
      "loss": 0.1574,
      "step": 10027
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.547753085117449,
      "learning_rate": 1.8008441208863914e-07,
      "loss": 0.1548,
      "step": 10028
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4029804879594956,
      "learning_rate": 1.7969079391298162e-07,
      "loss": 0.1667,
      "step": 10029
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.240705336174564,
      "learning_rate": 1.7929759851068817e-07,
      "loss": 0.6145,
      "step": 10030
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3892925799600404,
      "learning_rate": 1.7890482591624515e-07,
      "loss": 0.1771,
      "step": 10031
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5044364649480133,
      "learning_rate": 1.7851247616409994e-07,
      "loss": 0.157,
      "step": 10032
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.374722909942731,
      "learning_rate": 1.7812054928866617e-07,
      "loss": 0.17,
      "step": 10033
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5320705780382569,
      "learning_rate": 1.7772904532431634e-07,
      "loss": 0.185,
      "step": 10034
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.3914499003257195,
      "learning_rate": 1.7733796430539018e-07,
      "loss": 0.434,
      "step": 10035
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3296200644152374,
      "learning_rate": 1.7694730626618585e-07,
      "loss": 0.1519,
      "step": 10036
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4727419054069644,
      "learning_rate": 1.7655707124096865e-07,
      "loss": 0.1762,
      "step": 10037
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3918675255094128,
      "learning_rate": 1.7616725926396406e-07,
      "loss": 0.1675,
      "step": 10038
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2707001046541264,
      "learning_rate": 1.7577787036936024e-07,
      "loss": 0.1368,
      "step": 10039
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4045569670897071,
      "learning_rate": 1.7538890459131098e-07,
      "loss": 0.2383,
      "step": 10040
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.237156497588941,
      "learning_rate": 1.7500036196392956e-07,
      "loss": 0.1696,
      "step": 10041
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3701647985217411,
      "learning_rate": 1.746122425212954e-07,
      "loss": 0.1652,
      "step": 10042
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4070425469868568,
      "learning_rate": 1.7422454629744733e-07,
      "loss": 0.1858,
      "step": 10043
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.318643627823436,
      "learning_rate": 1.7383727332639044e-07,
      "loss": 0.1788,
      "step": 10044
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5567890977886047,
      "learning_rate": 1.7345042364208976e-07,
      "loss": 0.2076,
      "step": 10045
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2515521946105206,
      "learning_rate": 1.7306399727847588e-07,
      "loss": 0.1607,
      "step": 10046
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.429849824482191,
      "learning_rate": 1.7267799426944066e-07,
      "loss": 0.1634,
      "step": 10047
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4042555522163767,
      "learning_rate": 1.722924146488386e-07,
      "loss": 0.2064,
      "step": 10048
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5429042573878518,
      "learning_rate": 1.7190725845048827e-07,
      "loss": 0.2012,
      "step": 10049
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3850786695202986,
      "learning_rate": 1.7152252570816985e-07,
      "loss": 0.1675,
      "step": 10050
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3879423362596783,
      "learning_rate": 1.7113821645562744e-07,
      "loss": 0.152,
      "step": 10051
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4447134886964592,
      "learning_rate": 1.707543307265669e-07,
      "loss": 0.1981,
      "step": 10052
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3685473407794408,
      "learning_rate": 1.7037086855465902e-07,
      "loss": 0.1848,
      "step": 10053
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3116609997584794,
      "learning_rate": 1.6998782997353413e-07,
      "loss": 0.1698,
      "step": 10054
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2300639793172077,
      "learning_rate": 1.6960521501678872e-07,
      "loss": 0.1657,
      "step": 10055
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3403723227096467,
      "learning_rate": 1.6922302371797984e-07,
      "loss": 0.1869,
      "step": 10056
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3025941506906773,
      "learning_rate": 1.688412561106284e-07,
      "loss": 0.1824,
      "step": 10057
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5155998616840356,
      "learning_rate": 1.6845991222821877e-07,
      "loss": 0.1988,
      "step": 10058
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3298905392260187,
      "learning_rate": 1.6807899210419533e-07,
      "loss": 0.1744,
      "step": 10059
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3472800816190182,
      "learning_rate": 1.6769849577197018e-07,
      "loss": 0.1302,
      "step": 10060
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4566897372259853,
      "learning_rate": 1.6731842326491277e-07,
      "loss": 0.1838,
      "step": 10061
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5980756862571903,
      "learning_rate": 1.6693877461635976e-07,
      "loss": 0.2009,
      "step": 10062
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.499195970212406,
      "learning_rate": 1.6655954985960788e-07,
      "loss": 0.1663,
      "step": 10063
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2829404222602623,
      "learning_rate": 1.661807490279177e-07,
      "loss": 0.1592,
      "step": 10064
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.192814556805397,
      "learning_rate": 1.6580237215451378e-07,
      "loss": 0.6084,
      "step": 10065
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5716380410821977,
      "learning_rate": 1.6542441927258068e-07,
      "loss": 0.1936,
      "step": 10066
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3079170434927123,
      "learning_rate": 1.6504689041526855e-07,
      "loss": 0.1796,
      "step": 10067
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4463960760521963,
      "learning_rate": 1.6466978561568812e-07,
      "loss": 0.1695,
      "step": 10068
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2875442031582296,
      "learning_rate": 1.6429310490691575e-07,
      "loss": 0.1389,
      "step": 10069
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.49592737154851,
      "learning_rate": 1.639168483219872e-07,
      "loss": 0.1815,
      "step": 10070
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2455474398927395,
      "learning_rate": 1.6354101589390325e-07,
      "loss": 0.1358,
      "step": 10071
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4152689402850174,
      "learning_rate": 1.631656076556276e-07,
      "loss": 0.1717,
      "step": 10072
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4150159133331517,
      "learning_rate": 1.6279062364008446e-07,
      "loss": 0.1864,
      "step": 10073
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2040575621470417,
      "learning_rate": 1.6241606388016362e-07,
      "loss": 0.1656,
      "step": 10074
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.272830454700548,
      "learning_rate": 1.6204192840871602e-07,
      "loss": 0.1478,
      "step": 10075
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3319400951258227,
      "learning_rate": 1.6166821725855708e-07,
      "loss": 0.1724,
      "step": 10076
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3050396900242582,
      "learning_rate": 1.612949304624617e-07,
      "loss": 0.1637,
      "step": 10077
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3646365140774206,
      "learning_rate": 1.6092206805317146e-07,
      "loss": 0.1736,
      "step": 10078
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3161676620423444,
      "learning_rate": 1.6054963006338742e-07,
      "loss": 0.1684,
      "step": 10079
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4421783773786478,
      "learning_rate": 1.601776165257757e-07,
      "loss": 0.2083,
      "step": 10080
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3860915751922314,
      "learning_rate": 1.5980602747296513e-07,
      "loss": 0.1598,
      "step": 10081
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4376509329736844,
      "learning_rate": 1.594348629375442e-07,
      "loss": 0.1687,
      "step": 10082
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.327795678628248,
      "learning_rate": 1.5906412295206897e-07,
      "loss": 0.161,
      "step": 10083
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4462766646592276,
      "learning_rate": 1.586938075490546e-07,
      "loss": 0.1948,
      "step": 10084
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3995847314015206,
      "learning_rate": 1.583239167609807e-07,
      "loss": 0.156,
      "step": 10085
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.287501100938203,
      "learning_rate": 1.5795445062028848e-07,
      "loss": 0.1288,
      "step": 10086
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4611409129062507,
      "learning_rate": 1.575854091593837e-07,
      "loss": 0.1794,
      "step": 10087
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3568059799023242,
      "learning_rate": 1.5721679241063213e-07,
      "loss": 0.1633,
      "step": 10088
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3110625287937396,
      "learning_rate": 1.5684860040636573e-07,
      "loss": 0.138,
      "step": 10089
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.39271805960131,
      "learning_rate": 1.5648083317887696e-07,
      "loss": 0.1929,
      "step": 10090
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1920068808880968,
      "learning_rate": 1.5611349076042004e-07,
      "loss": 0.1469,
      "step": 10091
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2671403327999033,
      "learning_rate": 1.5574657318321528e-07,
      "loss": 0.1797,
      "step": 10092
    },
    {
      "epoch": 0.92,
      "grad_norm": 12.952425781296878,
      "learning_rate": 1.55380080479442e-07,
      "loss": 0.5946,
      "step": 10093
    },
    {
      "epoch": 0.92,
      "grad_norm": 7.430517637526269,
      "learning_rate": 1.5501401268124605e-07,
      "loss": 0.621,
      "step": 10094
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.368712206428652,
      "learning_rate": 1.5464836982073184e-07,
      "loss": 0.1758,
      "step": 10095
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4326583388095562,
      "learning_rate": 1.542831519299709e-07,
      "loss": 0.1921,
      "step": 10096
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.592606516350593,
      "learning_rate": 1.5391835904099316e-07,
      "loss": 0.1778,
      "step": 10097
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.464079139653311,
      "learning_rate": 1.535539911857953e-07,
      "loss": 0.1479,
      "step": 10098
    },
    {
      "epoch": 0.92,
      "grad_norm": 8.31077933953723,
      "learning_rate": 1.531900483963339e-07,
      "loss": 0.5332,
      "step": 10099
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4103140487385872,
      "learning_rate": 1.5282653070452845e-07,
      "loss": 0.1438,
      "step": 10100
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2753779862751535,
      "learning_rate": 1.524634381422635e-07,
      "loss": 0.1374,
      "step": 10101
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3381244889270694,
      "learning_rate": 1.521007707413835e-07,
      "loss": 0.1546,
      "step": 10102
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.40231942022799,
      "learning_rate": 1.5173852853369753e-07,
      "loss": 0.18,
      "step": 10103
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.318293556744689,
      "learning_rate": 1.5137671155097567e-07,
      "loss": 0.1601,
      "step": 10104
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.323546107961835,
      "learning_rate": 1.510153198249531e-07,
      "loss": 0.1695,
      "step": 10105
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3480092454618862,
      "learning_rate": 1.5065435338732504e-07,
      "loss": 0.1915,
      "step": 10106
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2674801806723333,
      "learning_rate": 1.5029381226975116e-07,
      "loss": 0.145,
      "step": 10107
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3726144137896115,
      "learning_rate": 1.4993369650385337e-07,
      "loss": 0.1813,
      "step": 10108
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1790588627356637,
      "learning_rate": 1.4957400612121532e-07,
      "loss": 0.1252,
      "step": 10109
    },
    {
      "epoch": 0.92,
      "grad_norm": 5.600392258850896,
      "learning_rate": 1.492147411533862e-07,
      "loss": 0.5434,
      "step": 10110
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3362215047260155,
      "learning_rate": 1.488559016318747e-07,
      "loss": 0.154,
      "step": 10111
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2024454170249381,
      "learning_rate": 1.4849748758815342e-07,
      "loss": 0.1521,
      "step": 10112
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4324400482699056,
      "learning_rate": 1.4813949905365833e-07,
      "loss": 0.2178,
      "step": 10113
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4039972765910784,
      "learning_rate": 1.4778193605978597e-07,
      "loss": 0.1558,
      "step": 10114
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2453081814616036,
      "learning_rate": 1.4742479863789905e-07,
      "loss": 0.1632,
      "step": 10115
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3405808168977182,
      "learning_rate": 1.470680868193186e-07,
      "loss": 0.1478,
      "step": 10116
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3637102200246027,
      "learning_rate": 1.4671180063533296e-07,
      "loss": 0.1406,
      "step": 10117
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.337510033464553,
      "learning_rate": 1.4635594011718935e-07,
      "loss": 0.468,
      "step": 10118
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3808032411274243,
      "learning_rate": 1.4600050529609944e-07,
      "loss": 0.1929,
      "step": 10119
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.619427348255946,
      "learning_rate": 1.4564549620323775e-07,
      "loss": 0.1572,
      "step": 10120
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5449343067343164,
      "learning_rate": 1.4529091286973994e-07,
      "loss": 0.2024,
      "step": 10121
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3472279189342318,
      "learning_rate": 1.4493675532670615e-07,
      "loss": 0.1662,
      "step": 10122
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.663171079998095,
      "learning_rate": 1.445830236051976e-07,
      "loss": 0.2278,
      "step": 10123
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.300431165637502,
      "learning_rate": 1.4422971773624005e-07,
      "loss": 0.1629,
      "step": 10124
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.435171198461285,
      "learning_rate": 1.4387683775081985e-07,
      "loss": 0.1598,
      "step": 10125
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.239670678364585,
      "learning_rate": 1.4352438367988775e-07,
      "loss": 0.1469,
      "step": 10126
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4552650996077934,
      "learning_rate": 1.4317235555435516e-07,
      "loss": 0.1955,
      "step": 10127
    },
    {
      "epoch": 0.93,
      "grad_norm": 4.70713609197745,
      "learning_rate": 1.428207534050985e-07,
      "loss": 0.4828,
      "step": 10128
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3486791409863614,
      "learning_rate": 1.424695772629553e-07,
      "loss": 0.1301,
      "step": 10129
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4078094914297026,
      "learning_rate": 1.4211882715872538e-07,
      "loss": 0.1547,
      "step": 10130
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1727964272268294,
      "learning_rate": 1.4176850312317246e-07,
      "loss": 0.1356,
      "step": 10131
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3674764622325053,
      "learning_rate": 1.41418605187022e-07,
      "loss": 0.1655,
      "step": 10132
    },
    {
      "epoch": 0.93,
      "grad_norm": 10.770028244352071,
      "learning_rate": 1.410691333809633e-07,
      "loss": 0.4328,
      "step": 10133
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2189672871456054,
      "learning_rate": 1.407200877356457e-07,
      "loss": 0.1277,
      "step": 10134
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4329783153683584,
      "learning_rate": 1.403714682816848e-07,
      "loss": 0.1892,
      "step": 10135
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.022225889440993,
      "learning_rate": 1.4002327504965495e-07,
      "loss": 0.5176,
      "step": 10136
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1184871802397203,
      "learning_rate": 1.3967550807009677e-07,
      "loss": 0.1297,
      "step": 10137
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3352529328641995,
      "learning_rate": 1.3932816737351084e-07,
      "loss": 0.1625,
      "step": 10138
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4166096929336247,
      "learning_rate": 1.389812529903606e-07,
      "loss": 0.1595,
      "step": 10139
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2825265877152376,
      "learning_rate": 1.3863476495107442e-07,
      "loss": 0.1356,
      "step": 10140
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.372471293473126,
      "learning_rate": 1.3828870328603973e-07,
      "loss": 0.1615,
      "step": 10141
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.338811578614515,
      "learning_rate": 1.3794306802560998e-07,
      "loss": 0.1767,
      "step": 10142
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.558309948457504,
      "learning_rate": 1.3759785920009928e-07,
      "loss": 0.1545,
      "step": 10143
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.325798280377775,
      "learning_rate": 1.372530768397845e-07,
      "loss": 0.1561,
      "step": 10144
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3840875880415455,
      "learning_rate": 1.3690872097490481e-07,
      "loss": 0.1822,
      "step": 10145
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2143115655186805,
      "learning_rate": 1.365647916356644e-07,
      "loss": 0.1444,
      "step": 10146
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4470218241251134,
      "learning_rate": 1.362212888522263e-07,
      "loss": 0.1893,
      "step": 10147
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4405935074361478,
      "learning_rate": 1.3587821265471866e-07,
      "loss": 0.1675,
      "step": 10148
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3897953216812666,
      "learning_rate": 1.355355630732319e-07,
      "loss": 0.1391,
      "step": 10149
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3752417739223284,
      "learning_rate": 1.3519334013781803e-07,
      "loss": 0.1359,
      "step": 10150
    },
    {
      "epoch": 0.93,
      "grad_norm": 7.102206187855264,
      "learning_rate": 1.3485154387849252e-07,
      "loss": 0.6868,
      "step": 10151
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.438542183982347,
      "learning_rate": 1.345101743252336e-07,
      "loss": 0.2107,
      "step": 10152
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4731794522774555,
      "learning_rate": 1.3416923150798123e-07,
      "loss": 0.1716,
      "step": 10153
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2358812696900978,
      "learning_rate": 1.3382871545663866e-07,
      "loss": 0.1455,
      "step": 10154
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4025791865534623,
      "learning_rate": 1.3348862620107038e-07,
      "loss": 0.1931,
      "step": 10155
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3545990709453852,
      "learning_rate": 1.3314896377110642e-07,
      "loss": 0.1669,
      "step": 10156
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.6119136581172935,
      "learning_rate": 1.328097281965357e-07,
      "loss": 0.4845,
      "step": 10157
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2993484355140734,
      "learning_rate": 1.3247091950711277e-07,
      "loss": 0.1896,
      "step": 10158
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4206398512985192,
      "learning_rate": 1.3213253773255218e-07,
      "loss": 0.194,
      "step": 10159
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.508715675052906,
      "learning_rate": 1.31794582902533e-07,
      "loss": 0.1972,
      "step": 10160
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2944812633388298,
      "learning_rate": 1.3145705504669592e-07,
      "loss": 0.1898,
      "step": 10161
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.78138775677847,
      "learning_rate": 1.3111995419464507e-07,
      "loss": 0.6071,
      "step": 10162
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.398659091668771,
      "learning_rate": 1.3078328037594567e-07,
      "loss": 0.1922,
      "step": 10163
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5605484671837084,
      "learning_rate": 1.3044703362012634e-07,
      "loss": 0.1536,
      "step": 10164
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2908957750720287,
      "learning_rate": 1.3011121395667848e-07,
      "loss": 0.1673,
      "step": 10165
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.942181587092205,
      "learning_rate": 1.2977582141505464e-07,
      "loss": 0.5723,
      "step": 10166
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2539870015982895,
      "learning_rate": 1.2944085602467294e-07,
      "loss": 0.1436,
      "step": 10167
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.349952874020824,
      "learning_rate": 1.2910631781491044e-07,
      "loss": 0.163,
      "step": 10168
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5511680494258049,
      "learning_rate": 1.2877220681510927e-07,
      "loss": 0.1721,
      "step": 10169
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4273275266375895,
      "learning_rate": 1.284385230545726e-07,
      "loss": 0.1794,
      "step": 10170
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.35933351096935,
      "learning_rate": 1.2810526656256705e-07,
      "loss": 0.191,
      "step": 10171
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3831956637118952,
      "learning_rate": 1.2777243736832202e-07,
      "loss": 0.171,
      "step": 10172
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4366587298309859,
      "learning_rate": 1.2744003550102746e-07,
      "loss": 0.1791,
      "step": 10173
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4572471502058941,
      "learning_rate": 1.2710806098983842e-07,
      "loss": 0.221,
      "step": 10174
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4212231496579257,
      "learning_rate": 1.2677651386387046e-07,
      "loss": 0.2084,
      "step": 10175
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3078909824823286,
      "learning_rate": 1.2644539415220369e-07,
      "loss": 0.168,
      "step": 10176
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2378946546899445,
      "learning_rate": 1.261147018838782e-07,
      "loss": 0.1415,
      "step": 10177
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2208410452948293,
      "learning_rate": 1.2578443708789855e-07,
      "loss": 0.1226,
      "step": 10178
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.440450431710611,
      "learning_rate": 1.2545459979323104e-07,
      "loss": 0.1476,
      "step": 10179
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3071871962946038,
      "learning_rate": 1.251251900288053e-07,
      "loss": 0.1511,
      "step": 10180
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3246063375471249,
      "learning_rate": 1.2479620782351154e-07,
      "loss": 0.1785,
      "step": 10181
    },
    {
      "epoch": 0.93,
      "grad_norm": 12.012254849812132,
      "learning_rate": 1.244676532062039e-07,
      "loss": 0.6405,
      "step": 10182
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.363471320170603,
      "learning_rate": 1.241395262056999e-07,
      "loss": 0.167,
      "step": 10183
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.444743047236859,
      "learning_rate": 1.2381182685077763e-07,
      "loss": 0.1726,
      "step": 10184
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5772854780545078,
      "learning_rate": 1.2348455517017855e-07,
      "loss": 0.1979,
      "step": 10185
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4872692466588666,
      "learning_rate": 1.2315771119260633e-07,
      "loss": 0.1678,
      "step": 10186
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.92969664268411,
      "learning_rate": 1.2283129494672862e-07,
      "loss": 0.5273,
      "step": 10187
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.304380783808884,
      "learning_rate": 1.2250530646117254e-07,
      "loss": 0.6261,
      "step": 10188
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.380181071515254,
      "learning_rate": 1.2217974576453072e-07,
      "loss": 0.1791,
      "step": 10189
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.09429546440145,
      "learning_rate": 1.2185461288535706e-07,
      "loss": 0.6341,
      "step": 10190
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4432533900896491,
      "learning_rate": 1.2152990785216645e-07,
      "loss": 0.156,
      "step": 10191
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3150365079366397,
      "learning_rate": 1.212056306934395e-07,
      "loss": 0.1634,
      "step": 10192
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.409854814847087,
      "learning_rate": 1.208817814376162e-07,
      "loss": 0.2146,
      "step": 10193
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.50728683030081,
      "learning_rate": 1.2055836011310108e-07,
      "loss": 0.1938,
      "step": 10194
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.287367759876239,
      "learning_rate": 1.2023536674825974e-07,
      "loss": 0.1829,
      "step": 10195
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2854727027064232,
      "learning_rate": 1.199128013714218e-07,
      "loss": 0.1511,
      "step": 10196
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2306882387477633,
      "learning_rate": 1.1959066401087739e-07,
      "loss": 0.1528,
      "step": 10197
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3646374480660197,
      "learning_rate": 1.1926895469488052e-07,
      "loss": 0.1724,
      "step": 10198
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5066805397262477,
      "learning_rate": 1.189476734516476e-07,
      "loss": 0.1641,
      "step": 10199
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5665219989616144,
      "learning_rate": 1.1862682030935602e-07,
      "loss": 0.1704,
      "step": 10200
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3455139718695472,
      "learning_rate": 1.1830639529614774e-07,
      "loss": 0.1482,
      "step": 10201
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.396107716030546,
      "learning_rate": 1.1798639844012583e-07,
      "loss": 0.1698,
      "step": 10202
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.360574179452762,
      "learning_rate": 1.1766682976935672e-07,
      "loss": 0.2013,
      "step": 10203
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3355877218299221,
      "learning_rate": 1.1734768931186801e-07,
      "loss": 0.1475,
      "step": 10204
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4194987590397592,
      "learning_rate": 1.170289770956512e-07,
      "loss": 0.2009,
      "step": 10205
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2313633236276433,
      "learning_rate": 1.1671069314865835e-07,
      "loss": 0.1405,
      "step": 10206
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3591256612593603,
      "learning_rate": 1.1639283749880604e-07,
      "loss": 0.2163,
      "step": 10207
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2942910723022838,
      "learning_rate": 1.1607541017397194e-07,
      "loss": 0.1208,
      "step": 10208
    },
    {
      "epoch": 0.93,
      "grad_norm": 8.43058163638681,
      "learning_rate": 1.157584112019966e-07,
      "loss": 0.45,
      "step": 10209
    },
    {
      "epoch": 0.93,
      "grad_norm": 6.039330109923727,
      "learning_rate": 1.1544184061068275e-07,
      "loss": 0.5847,
      "step": 10210
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3484721686444816,
      "learning_rate": 1.1512569842779653e-07,
      "loss": 0.1882,
      "step": 10211
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.381184689692604,
      "learning_rate": 1.1480998468106464e-07,
      "loss": 0.1611,
      "step": 10212
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4282282067882788,
      "learning_rate": 1.1449469939817825e-07,
      "loss": 0.2068,
      "step": 10213
    },
    {
      "epoch": 0.93,
      "grad_norm": 11.77223075498985,
      "learning_rate": 1.1417984260678915e-07,
      "loss": 0.7132,
      "step": 10214
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.290680026196007,
      "learning_rate": 1.1386541433451359e-07,
      "loss": 0.1508,
      "step": 10215
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3503419573622857,
      "learning_rate": 1.1355141460892726e-07,
      "loss": 0.1677,
      "step": 10216
    },
    {
      "epoch": 0.93,
      "grad_norm": 5.0070807534613,
      "learning_rate": 1.1323784345757205e-07,
      "loss": 0.5248,
      "step": 10217
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3385269467702883,
      "learning_rate": 1.1292470090794815e-07,
      "loss": 0.201,
      "step": 10218
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.53849139318754,
      "learning_rate": 1.1261198698752196e-07,
      "loss": 0.1963,
      "step": 10219
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3604056899524672,
      "learning_rate": 1.1229970172371985e-07,
      "loss": 0.169,
      "step": 10220
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4238762395504219,
      "learning_rate": 1.1198784514393157e-07,
      "loss": 0.176,
      "step": 10221
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2892085215856335,
      "learning_rate": 1.1167641727550804e-07,
      "loss": 0.194,
      "step": 10222
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2946307539686606,
      "learning_rate": 1.1136541814576574e-07,
      "loss": 0.1684,
      "step": 10223
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3370076473644976,
      "learning_rate": 1.110548477819795e-07,
      "loss": 0.1817,
      "step": 10224
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4902652408431054,
      "learning_rate": 1.1074470621138866e-07,
      "loss": 0.1971,
      "step": 10225
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3401988897335189,
      "learning_rate": 1.1043499346119535e-07,
      "loss": 0.1639,
      "step": 10226
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.741666041012493,
      "learning_rate": 1.1012570955856283e-07,
      "loss": 0.4867,
      "step": 10227
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.481705753171839,
      "learning_rate": 1.0981685453061775e-07,
      "loss": 0.2026,
      "step": 10228
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3857030138585547,
      "learning_rate": 1.0950842840444842e-07,
      "loss": 0.1826,
      "step": 10229
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.152612786388112,
      "learning_rate": 1.092004312071071e-07,
      "loss": 0.5655,
      "step": 10230
    },
    {
      "epoch": 0.94,
      "grad_norm": 4.775843152247178,
      "learning_rate": 1.0889286296560497e-07,
      "loss": 0.4209,
      "step": 10231
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.392124050208001,
      "learning_rate": 1.0858572370691989e-07,
      "loss": 0.5446,
      "step": 10232
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.125301314687862,
      "learning_rate": 1.0827901345798919e-07,
      "loss": 0.1309,
      "step": 10233
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6208550537500084,
      "learning_rate": 1.0797273224571303e-07,
      "loss": 0.1967,
      "step": 10234
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3881514811281315,
      "learning_rate": 1.0766688009695548e-07,
      "loss": 0.1795,
      "step": 10235
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2484073510513913,
      "learning_rate": 1.0736145703854006e-07,
      "loss": 0.1624,
      "step": 10236
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.285603715612004,
      "learning_rate": 1.0705646309725593e-07,
      "loss": 0.1423,
      "step": 10237
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5171180825263066,
      "learning_rate": 1.0675189829985222e-07,
      "loss": 0.174,
      "step": 10238
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.326520255796291,
      "learning_rate": 1.0644776267304257e-07,
      "loss": 0.7134,
      "step": 10239
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3222135687732215,
      "learning_rate": 1.061440562435001e-07,
      "loss": 0.1655,
      "step": 10240
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.0694128040576425,
      "learning_rate": 1.0584077903786238e-07,
      "loss": 0.4466,
      "step": 10241
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3936689554292772,
      "learning_rate": 1.055379310827298e-07,
      "loss": 0.1413,
      "step": 10242
    },
    {
      "epoch": 0.94,
      "grad_norm": 22.440206374363907,
      "learning_rate": 1.0523551240466278e-07,
      "loss": 0.5845,
      "step": 10243
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2764353355030373,
      "learning_rate": 1.0493352303018678e-07,
      "loss": 0.1663,
      "step": 10244
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.579024192703873,
      "learning_rate": 1.0463196298578726e-07,
      "loss": 0.1813,
      "step": 10245
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1563724423438753,
      "learning_rate": 1.0433083229791308e-07,
      "loss": 0.1451,
      "step": 10246
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3691520883415351,
      "learning_rate": 1.0403013099297587e-07,
      "loss": 0.166,
      "step": 10247
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.115886252917227,
      "learning_rate": 1.0372985909734956e-07,
      "loss": 0.5821,
      "step": 10248
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3242805455261664,
      "learning_rate": 1.0343001663736918e-07,
      "loss": 0.1785,
      "step": 10249
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.428834641776908,
      "learning_rate": 1.0313060363933314e-07,
      "loss": 0.1471,
      "step": 10250
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.274814469103373,
      "learning_rate": 1.028316201295021e-07,
      "loss": 0.1155,
      "step": 10251
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.502437099012124,
      "learning_rate": 1.0253306613409842e-07,
      "loss": 0.1764,
      "step": 10252
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2760547257218173,
      "learning_rate": 1.0223494167930837e-07,
      "loss": 0.1373,
      "step": 10253
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3542548005736403,
      "learning_rate": 1.0193724679127826e-07,
      "loss": 0.2009,
      "step": 10254
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2871254581609333,
      "learning_rate": 1.0163998149611887e-07,
      "loss": 0.1311,
      "step": 10255
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5288078327078332,
      "learning_rate": 1.0134314581990156e-07,
      "loss": 0.1702,
      "step": 10256
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.218760892247743,
      "learning_rate": 1.0104673978866164e-07,
      "loss": 0.1489,
      "step": 10257
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5005979539558274,
      "learning_rate": 1.0075076342839552e-07,
      "loss": 0.2408,
      "step": 10258
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3708870067477452,
      "learning_rate": 1.0045521676506132e-07,
      "loss": 0.17,
      "step": 10259
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5623912567215263,
      "learning_rate": 1.0016009982458219e-07,
      "loss": 0.1703,
      "step": 10260
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4254587925627873,
      "learning_rate": 9.986541263284077e-08,
      "loss": 0.1862,
      "step": 10261
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3234346013870004,
      "learning_rate": 9.957115521568362e-08,
      "loss": 0.1889,
      "step": 10262
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.177516911504544,
      "learning_rate": 9.927732759891839e-08,
      "loss": 0.5123,
      "step": 10263
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5233559404259085,
      "learning_rate": 9.898392980831673e-08,
      "loss": 0.1975,
      "step": 10264
    },
    {
      "epoch": 0.94,
      "grad_norm": 5.346982306907959,
      "learning_rate": 9.869096186961025e-08,
      "loss": 0.5531,
      "step": 10265
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.274987238289018,
      "learning_rate": 9.839842380849563e-08,
      "loss": 0.1531,
      "step": 10266
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3675691811147974,
      "learning_rate": 9.810631565063012e-08,
      "loss": 0.1772,
      "step": 10267
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5486276563619021,
      "learning_rate": 9.78146374216321e-08,
      "loss": 0.1684,
      "step": 10268
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3592437220235791,
      "learning_rate": 9.752338914708559e-08,
      "loss": 0.1787,
      "step": 10269
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2626261773466863,
      "learning_rate": 9.723257085253401e-08,
      "loss": 0.13,
      "step": 10270
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4378369403443891,
      "learning_rate": 9.69421825634842e-08,
      "loss": 0.1697,
      "step": 10271
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.246877502433305,
      "learning_rate": 9.665222430540522e-08,
      "loss": 0.5591,
      "step": 10272
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3502451385817804,
      "learning_rate": 9.636269610372895e-08,
      "loss": 0.1603,
      "step": 10273
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.471493602046358,
      "learning_rate": 9.607359798384785e-08,
      "loss": 0.1893,
      "step": 10274
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5978703482221182,
      "learning_rate": 9.578492997111833e-08,
      "loss": 0.1917,
      "step": 10275
    },
    {
      "epoch": 0.94,
      "grad_norm": 8.722585336995168,
      "learning_rate": 9.549669209085899e-08,
      "loss": 0.5119,
      "step": 10276
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.238259611383801,
      "learning_rate": 9.520888436834852e-08,
      "loss": 0.1643,
      "step": 10277
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.478359592902307,
      "learning_rate": 9.492150682883172e-08,
      "loss": 0.1652,
      "step": 10278
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.324415605279585,
      "learning_rate": 9.463455949751232e-08,
      "loss": 0.1524,
      "step": 10279
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5245713948478714,
      "learning_rate": 9.434804239955797e-08,
      "loss": 0.2152,
      "step": 10280
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3215338821966396,
      "learning_rate": 9.406195556009745e-08,
      "loss": 0.2049,
      "step": 10281
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4112887283040831,
      "learning_rate": 9.377629900422291e-08,
      "loss": 0.1484,
      "step": 10282
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4659996274726483,
      "learning_rate": 9.349107275698932e-08,
      "loss": 0.2207,
      "step": 10283
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3895062611031657,
      "learning_rate": 9.32062768434111e-08,
      "loss": 0.1808,
      "step": 10284
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2177622382373774,
      "learning_rate": 9.292191128846772e-08,
      "loss": 0.145,
      "step": 10285
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4104645701060596,
      "learning_rate": 9.263797611710035e-08,
      "loss": 0.1564,
      "step": 10286
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3869331874769044,
      "learning_rate": 9.235447135421127e-08,
      "loss": 0.1531,
      "step": 10287
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2596877687465549,
      "learning_rate": 9.207139702466561e-08,
      "loss": 0.1736,
      "step": 10288
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.3511027386648164,
      "learning_rate": 9.178875315329183e-08,
      "loss": 0.6305,
      "step": 10289
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3734353345026584,
      "learning_rate": 9.1506539764879e-08,
      "loss": 0.1736,
      "step": 10290
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2371871904764755,
      "learning_rate": 9.122475688417954e-08,
      "loss": 0.1404,
      "step": 10291
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.462322299580734,
      "learning_rate": 9.0943404535907e-08,
      "loss": 0.1612,
      "step": 10292
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.490458571971432,
      "learning_rate": 9.066248274473832e-08,
      "loss": 0.1515,
      "step": 10293
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3026853182748475,
      "learning_rate": 9.038199153531213e-08,
      "loss": 0.1741,
      "step": 10294
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5689378744930629,
      "learning_rate": 9.010193093222986e-08,
      "loss": 0.1924,
      "step": 10295
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3308027831230842,
      "learning_rate": 8.982230096005406e-08,
      "loss": 0.1598,
      "step": 10296
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.268693403385036,
      "learning_rate": 8.954310164331015e-08,
      "loss": 0.1451,
      "step": 10297
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.245451925695151,
      "learning_rate": 8.926433300648629e-08,
      "loss": 0.496,
      "step": 10298
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.309661584113704,
      "learning_rate": 8.898599507403127e-08,
      "loss": 0.1495,
      "step": 10299
    },
    {
      "epoch": 0.94,
      "grad_norm": 7.359974405936344,
      "learning_rate": 8.87080878703589e-08,
      "loss": 0.6918,
      "step": 10300
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3780993867593538,
      "learning_rate": 8.843061141984244e-08,
      "loss": 0.145,
      "step": 10301
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2207895009008267,
      "learning_rate": 8.815356574681855e-08,
      "loss": 0.1371,
      "step": 10302
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3829063751382873,
      "learning_rate": 8.787695087558556e-08,
      "loss": 0.1691,
      "step": 10303
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5211479691965957,
      "learning_rate": 8.760076683040519e-08,
      "loss": 0.2089,
      "step": 10304
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4096976337358735,
      "learning_rate": 8.732501363550028e-08,
      "loss": 0.2211,
      "step": 10305
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.340537583162746,
      "learning_rate": 8.704969131505648e-08,
      "loss": 0.1321,
      "step": 10306
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2312881150314314,
      "learning_rate": 8.677479989322113e-08,
      "loss": 0.1494,
      "step": 10307
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.34304651921825,
      "learning_rate": 8.650033939410385e-08,
      "loss": 0.1632,
      "step": 10308
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.321663341657466,
      "learning_rate": 8.622630984177704e-08,
      "loss": 0.1478,
      "step": 10309
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.537288002232365,
      "learning_rate": 8.595271126027538e-08,
      "loss": 0.1986,
      "step": 10310
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4337788534246974,
      "learning_rate": 8.567954367359355e-08,
      "loss": 0.1679,
      "step": 10311
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4391655641905778,
      "learning_rate": 8.540680710569238e-08,
      "loss": 0.2013,
      "step": 10312
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.507556400682295,
      "learning_rate": 8.513450158049109e-08,
      "loss": 0.1874,
      "step": 10313
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4597175865517948,
      "learning_rate": 8.486262712187387e-08,
      "loss": 0.1899,
      "step": 10314
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2028747164403777,
      "learning_rate": 8.459118375368447e-08,
      "loss": 0.1079,
      "step": 10315
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4308051567037683,
      "learning_rate": 8.432017149973216e-08,
      "loss": 0.1786,
      "step": 10316
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2678941574683438,
      "learning_rate": 8.404959038378513e-08,
      "loss": 0.1519,
      "step": 10317
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3348062007481039,
      "learning_rate": 8.377944042957498e-08,
      "loss": 0.1653,
      "step": 10318
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4949918429817248,
      "learning_rate": 8.350972166079663e-08,
      "loss": 0.1823,
      "step": 10319
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4390100428597752,
      "learning_rate": 8.324043410110561e-08,
      "loss": 0.1624,
      "step": 10320
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4813041352485108,
      "learning_rate": 8.29715777741208e-08,
      "loss": 0.1983,
      "step": 10321
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4299034143199052,
      "learning_rate": 8.270315270342222e-08,
      "loss": 0.1636,
      "step": 10322
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2829532630243046,
      "learning_rate": 8.243515891255272e-08,
      "loss": 0.1702,
      "step": 10323
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3550934929614171,
      "learning_rate": 8.216759642501681e-08,
      "loss": 0.1667,
      "step": 10324
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4478548064551977,
      "learning_rate": 8.190046526428241e-08,
      "loss": 0.2202,
      "step": 10325
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6537294966101082,
      "learning_rate": 8.163376545377744e-08,
      "loss": 0.1519,
      "step": 10326
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3719627402647514,
      "learning_rate": 8.136749701689428e-08,
      "loss": 0.2019,
      "step": 10327
    },
    {
      "epoch": 0.94,
      "grad_norm": 6.3387749830318905,
      "learning_rate": 8.110165997698593e-08,
      "loss": 0.5888,
      "step": 10328
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5356591718655657,
      "learning_rate": 8.08362543573682e-08,
      "loss": 0.1743,
      "step": 10329
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3134056615600995,
      "learning_rate": 8.05712801813191e-08,
      "loss": 0.1607,
      "step": 10330
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5061883329382117,
      "learning_rate": 8.03067374720784e-08,
      "loss": 0.2152,
      "step": 10331
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6346740280960201,
      "learning_rate": 8.004262625284864e-08,
      "loss": 0.1708,
      "step": 10332
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5535533206300989,
      "learning_rate": 7.977894654679352e-08,
      "loss": 0.2122,
      "step": 10333
    },
    {
      "epoch": 0.94,
      "grad_norm": 11.297203278637827,
      "learning_rate": 7.951569837704009e-08,
      "loss": 0.5804,
      "step": 10334
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2226269279182187,
      "learning_rate": 7.925288176667711e-08,
      "loss": 0.1291,
      "step": 10335
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5465582599109036,
      "learning_rate": 7.899049673875447e-08,
      "loss": 0.1811,
      "step": 10336
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3894168753015934,
      "learning_rate": 7.872854331628599e-08,
      "loss": 0.1579,
      "step": 10337
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.18838822069206,
      "learning_rate": 7.846702152224606e-08,
      "loss": 0.1447,
      "step": 10338
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.432079855254805,
      "learning_rate": 7.820593137957244e-08,
      "loss": 0.1466,
      "step": 10339
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.383686110422532,
      "learning_rate": 7.794527291116404e-08,
      "loss": 0.1475,
      "step": 10340
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3781481261405073,
      "learning_rate": 7.768504613988259e-08,
      "loss": 0.1794,
      "step": 10341
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4622406834063135,
      "learning_rate": 7.742525108855148e-08,
      "loss": 0.1479,
      "step": 10342
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.439670984895812,
      "learning_rate": 7.716588777995693e-08,
      "loss": 0.1449,
      "step": 10343
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.812525602574268,
      "learning_rate": 7.690695623684685e-08,
      "loss": 0.468,
      "step": 10344
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.359246506770542,
      "learning_rate": 7.664845648193087e-08,
      "loss": 0.1647,
      "step": 10345
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.575799903973583,
      "learning_rate": 7.639038853788083e-08,
      "loss": 0.2096,
      "step": 10346
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.39548961065227,
      "learning_rate": 7.613275242733198e-08,
      "loss": 0.511,
      "step": 10347
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4127448678750525,
      "learning_rate": 7.587554817288012e-08,
      "loss": 0.1777,
      "step": 10348
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.403971264413945,
      "learning_rate": 7.561877579708332e-08,
      "loss": 0.1974,
      "step": 10349
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1341289524034563,
      "learning_rate": 7.536243532246356e-08,
      "loss": 0.1361,
      "step": 10350
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3317324254704912,
      "learning_rate": 7.510652677150232e-08,
      "loss": 0.1675,
      "step": 10351
    },
    {
      "epoch": 0.95,
      "grad_norm": 12.373516806295772,
      "learning_rate": 7.485105016664551e-08,
      "loss": 0.4391,
      "step": 10352
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2907389023301175,
      "learning_rate": 7.459600553029966e-08,
      "loss": 0.1599,
      "step": 10353
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3862100343964219,
      "learning_rate": 7.434139288483355e-08,
      "loss": 0.1977,
      "step": 10354
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.28141907154086,
      "learning_rate": 7.408721225257876e-08,
      "loss": 0.6144,
      "step": 10355
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4033340716802027,
      "learning_rate": 7.383346365582856e-08,
      "loss": 0.1359,
      "step": 10356
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5291720958607324,
      "learning_rate": 7.358014711683903e-08,
      "loss": 0.211,
      "step": 10357
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3175440889033319,
      "learning_rate": 7.33272626578263e-08,
      "loss": 0.1848,
      "step": 10358
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.531804112231102,
      "learning_rate": 7.307481030097152e-08,
      "loss": 0.513,
      "step": 10359
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.570544211325991,
      "learning_rate": 7.282279006841587e-08,
      "loss": 0.1738,
      "step": 10360
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.393178721173774,
      "learning_rate": 7.257120198226219e-08,
      "loss": 0.1563,
      "step": 10361
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5856700879537882,
      "learning_rate": 7.232004606457844e-08,
      "loss": 0.1938,
      "step": 10362
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3595050597917675,
      "learning_rate": 7.206932233739084e-08,
      "loss": 0.1494,
      "step": 10363
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4781553609990736,
      "learning_rate": 7.181903082269126e-08,
      "loss": 0.1781,
      "step": 10364
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4400625014846749,
      "learning_rate": 7.156917154243048e-08,
      "loss": 0.1757,
      "step": 10365
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3675514662800854,
      "learning_rate": 7.131974451852319e-08,
      "loss": 0.1336,
      "step": 10366
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2666649323605772,
      "learning_rate": 7.107074977284634e-08,
      "loss": 0.1808,
      "step": 10367
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.347710072976657,
      "learning_rate": 7.0822187327238e-08,
      "loss": 0.1518,
      "step": 10368
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3285353766560648,
      "learning_rate": 7.057405720349853e-08,
      "loss": 0.1672,
      "step": 10369
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.393636339043218,
      "learning_rate": 7.032635942339106e-08,
      "loss": 0.1664,
      "step": 10370
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.299639616860739,
      "learning_rate": 7.007909400864044e-08,
      "loss": 0.1713,
      "step": 10371
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3686839601395673,
      "learning_rate": 6.98322609809332e-08,
      "loss": 0.1545,
      "step": 10372
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.392901255875332,
      "learning_rate": 6.958586036191817e-08,
      "loss": 0.1762,
      "step": 10373
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3026528521415341,
      "learning_rate": 6.933989217320636e-08,
      "loss": 0.1522,
      "step": 10374
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.0348414986064745,
      "learning_rate": 6.909435643637108e-08,
      "loss": 0.5295,
      "step": 10375
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3033799553137757,
      "learning_rate": 6.884925317294678e-08,
      "loss": 0.1677,
      "step": 10376
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4482345262577858,
      "learning_rate": 6.860458240443179e-08,
      "loss": 0.1523,
      "step": 10377
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4560151245209163,
      "learning_rate": 6.836034415228509e-08,
      "loss": 0.1965,
      "step": 10378
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.27571666048097,
      "learning_rate": 6.811653843792676e-08,
      "loss": 0.138,
      "step": 10379
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.934345200281895,
      "learning_rate": 6.78731652827419e-08,
      "loss": 0.7354,
      "step": 10380
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1235007972598972,
      "learning_rate": 6.763022470807457e-08,
      "loss": 0.1588,
      "step": 10381
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.588455415969361,
      "learning_rate": 6.738771673523325e-08,
      "loss": 0.7204,
      "step": 10382
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.265857353360008,
      "learning_rate": 6.714564138548707e-08,
      "loss": 0.1521,
      "step": 10383
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3367691983550538,
      "learning_rate": 6.690399868006791e-08,
      "loss": 0.1911,
      "step": 10384
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4151772369027718,
      "learning_rate": 6.666278864016884e-08,
      "loss": 0.1871,
      "step": 10385
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5300473670921015,
      "learning_rate": 6.642201128694625e-08,
      "loss": 0.1933,
      "step": 10386
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4656334847024997,
      "learning_rate": 6.618166664151827e-08,
      "loss": 0.1726,
      "step": 10387
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4003437573859687,
      "learning_rate": 6.594175472496301e-08,
      "loss": 0.1512,
      "step": 10388
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4785670991145496,
      "learning_rate": 6.570227555832475e-08,
      "loss": 0.1459,
      "step": 10389
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7355185367186272,
      "learning_rate": 6.546322916260505e-08,
      "loss": 0.1739,
      "step": 10390
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.365240229261152,
      "learning_rate": 6.522461555877213e-08,
      "loss": 0.1818,
      "step": 10391
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1801595594478607,
      "learning_rate": 6.498643476775201e-08,
      "loss": 0.1197,
      "step": 10392
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.346771354096725,
      "learning_rate": 6.474868681043578e-08,
      "loss": 0.1626,
      "step": 10393
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3414350219422002,
      "learning_rate": 6.451137170767563e-08,
      "loss": 0.2031,
      "step": 10394
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5085881053942602,
      "learning_rate": 6.427448948028492e-08,
      "loss": 0.1997,
      "step": 10395
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2271330103495133,
      "learning_rate": 6.403804014904092e-08,
      "loss": 0.1951,
      "step": 10396
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3614936866369145,
      "learning_rate": 6.380202373468092e-08,
      "loss": 0.1483,
      "step": 10397
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1993965284144272,
      "learning_rate": 6.35664402579056e-08,
      "loss": 0.1674,
      "step": 10398
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5348810234128218,
      "learning_rate": 6.333128973937619e-08,
      "loss": 0.1813,
      "step": 10399
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5190652145936803,
      "learning_rate": 6.309657219971843e-08,
      "loss": 0.1522,
      "step": 10400
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3471930086121031,
      "learning_rate": 6.286228765951807e-08,
      "loss": 0.183,
      "step": 10401
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3431507410774213,
      "learning_rate": 6.262843613932312e-08,
      "loss": 0.1569,
      "step": 10402
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3094825980200382,
      "learning_rate": 6.23950176596444e-08,
      "loss": 0.1356,
      "step": 10403
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3266360799088581,
      "learning_rate": 6.216203224095386e-08,
      "loss": 0.149,
      "step": 10404
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4396616430199107,
      "learning_rate": 6.19294799036857e-08,
      "loss": 0.1552,
      "step": 10405
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.680411398768758,
      "learning_rate": 6.169736066823695e-08,
      "loss": 0.5678,
      "step": 10406
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3638710946048056,
      "learning_rate": 6.146567455496577e-08,
      "loss": 0.1831,
      "step": 10407
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2550753386676288,
      "learning_rate": 6.123442158419257e-08,
      "loss": 0.1804,
      "step": 10408
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3762967583608041,
      "learning_rate": 6.100360177619946e-08,
      "loss": 0.1773,
      "step": 10409
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2393796879948504,
      "learning_rate": 6.077321515123136e-08,
      "loss": 0.1402,
      "step": 10410
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4287703626333814,
      "learning_rate": 6.054326172949432e-08,
      "loss": 0.14,
      "step": 10411
    },
    {
      "epoch": 0.95,
      "grad_norm": 9.939042319551573,
      "learning_rate": 6.031374153115722e-08,
      "loss": 0.5569,
      "step": 10412
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4701976740319802,
      "learning_rate": 6.008465457635004e-08,
      "loss": 0.1615,
      "step": 10413
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5405332770333449,
      "learning_rate": 5.985600088516619e-08,
      "loss": 0.212,
      "step": 10414
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2932088125223504,
      "learning_rate": 5.96277804776585e-08,
      "loss": 0.1749,
      "step": 10415
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5087837211013224,
      "learning_rate": 5.939999337384483e-08,
      "loss": 0.162,
      "step": 10416
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4356077837515246,
      "learning_rate": 5.917263959370312e-08,
      "loss": 0.149,
      "step": 10417
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.306297951840575,
      "learning_rate": 5.894571915717406e-08,
      "loss": 0.18,
      "step": 10418
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5354706658698825,
      "learning_rate": 5.8719232084159525e-08,
      "loss": 0.1948,
      "step": 10419
    },
    {
      "epoch": 0.95,
      "grad_norm": 7.601701998632939,
      "learning_rate": 5.8493178394524174e-08,
      "loss": 0.5278,
      "step": 10420
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.315143751158972,
      "learning_rate": 5.8267558108094924e-08,
      "loss": 0.153,
      "step": 10421
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6328800926974865,
      "learning_rate": 5.8042371244659836e-08,
      "loss": 0.178,
      "step": 10422
    },
    {
      "epoch": 0.95,
      "grad_norm": 8.419508999937376,
      "learning_rate": 5.781761782396922e-08,
      "loss": 0.7672,
      "step": 10423
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4778508843816414,
      "learning_rate": 5.759329786573509e-08,
      "loss": 0.1547,
      "step": 10424
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5632514182591297,
      "learning_rate": 5.736941138963281e-08,
      "loss": 0.1918,
      "step": 10425
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3368544565004294,
      "learning_rate": 5.71459584152978e-08,
      "loss": 0.1764,
      "step": 10426
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6233960407136614,
      "learning_rate": 5.6922938962329364e-08,
      "loss": 0.1997,
      "step": 10427
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3603221718083929,
      "learning_rate": 5.6700353050286316e-08,
      "loss": 0.1626,
      "step": 10428
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2998972167159297,
      "learning_rate": 5.647820069869247e-08,
      "loss": 0.1517,
      "step": 10429
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.300985077780548,
      "learning_rate": 5.625648192703115e-08,
      "loss": 0.1579,
      "step": 10430
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.413143488793899,
      "learning_rate": 5.6035196754748444e-08,
      "loss": 0.1489,
      "step": 10431
    },
    {
      "epoch": 0.95,
      "grad_norm": 6.837307737541399,
      "learning_rate": 5.581434520125328e-08,
      "loss": 0.5412,
      "step": 10432
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4469477792702374,
      "learning_rate": 5.559392728591517e-08,
      "loss": 0.1911,
      "step": 10433
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3873025751284203,
      "learning_rate": 5.537394302806698e-08,
      "loss": 0.1481,
      "step": 10434
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2513270116396005,
      "learning_rate": 5.5154392447001605e-08,
      "loss": 0.16,
      "step": 10435
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2642380616335906,
      "learning_rate": 5.493527556197642e-08,
      "loss": 0.1659,
      "step": 10436
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2487915271566332,
      "learning_rate": 5.4716592392208834e-08,
      "loss": 0.1711,
      "step": 10437
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5287207060896988,
      "learning_rate": 5.44983429568785e-08,
      "loss": 0.2067,
      "step": 10438
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.434298295442813,
      "learning_rate": 5.428052727512845e-08,
      "loss": 0.2005,
      "step": 10439
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4494952642485386,
      "learning_rate": 5.406314536606061e-08,
      "loss": 0.1633,
      "step": 10440
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2378451548542142,
      "learning_rate": 5.384619724874307e-08,
      "loss": 0.1363,
      "step": 10441
    },
    {
      "epoch": 0.95,
      "grad_norm": 5.346877365224179,
      "learning_rate": 5.362968294220172e-08,
      "loss": 0.557,
      "step": 10442
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.242597847930935,
      "learning_rate": 5.341360246542804e-08,
      "loss": 0.124,
      "step": 10443
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3604516007284548,
      "learning_rate": 5.319795583737242e-08,
      "loss": 0.1888,
      "step": 10444
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3106479357426064,
      "learning_rate": 5.298274307694917e-08,
      "loss": 0.1413,
      "step": 10445
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4391633207203314,
      "learning_rate": 5.276796420303376e-08,
      "loss": 0.1521,
      "step": 10446
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4210991960986248,
      "learning_rate": 5.2553619234463335e-08,
      "loss": 0.1622,
      "step": 10447
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3528172134312957,
      "learning_rate": 5.2339708190038417e-08,
      "loss": 0.1885,
      "step": 10448
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.511892626753977,
      "learning_rate": 5.2126231088519e-08,
      "loss": 0.1656,
      "step": 10449
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2345362558678024,
      "learning_rate": 5.191318794862954e-08,
      "loss": 0.1566,
      "step": 10450
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3487077758916826,
      "learning_rate": 5.17005787890551e-08,
      "loss": 0.1684,
      "step": 10451
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5838185695337472,
      "learning_rate": 5.148840362844243e-08,
      "loss": 0.2045,
      "step": 10452
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5170226916771345,
      "learning_rate": 5.1276662485401086e-08,
      "loss": 0.1734,
      "step": 10453
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3780811352541937,
      "learning_rate": 5.106535537850232e-08,
      "loss": 0.154,
      "step": 10454
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.51689893075814,
      "learning_rate": 5.0854482326279654e-08,
      "loss": 0.1492,
      "step": 10455
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.383449703224305,
      "learning_rate": 5.0644043347226615e-08,
      "loss": 0.1785,
      "step": 10456
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5124107051353883,
      "learning_rate": 5.0434038459801213e-08,
      "loss": 0.1791,
      "step": 10457
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2161248922963481,
      "learning_rate": 5.022446768242206e-08,
      "loss": 0.1151,
      "step": 10458
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3249462811926371,
      "learning_rate": 5.001533103347056e-08,
      "loss": 0.1515,
      "step": 10459
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3946360899709596,
      "learning_rate": 4.9806628531288146e-08,
      "loss": 0.1384,
      "step": 10460
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3137295816128698,
      "learning_rate": 4.959836019417963e-08,
      "loss": 0.1583,
      "step": 10461
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2144577086721944,
      "learning_rate": 4.9390526040412613e-08,
      "loss": 0.1596,
      "step": 10462
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4744953924733089,
      "learning_rate": 4.9183126088214185e-08,
      "loss": 0.1909,
      "step": 10463
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.554499456595901,
      "learning_rate": 4.8976160355775906e-08,
      "loss": 0.1741,
      "step": 10464
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4466992997429533,
      "learning_rate": 4.876962886124936e-08,
      "loss": 0.1623,
      "step": 10465
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3185214866414783,
      "learning_rate": 4.856353162274896e-08,
      "loss": 0.1663,
      "step": 10466
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.207420903904331,
      "learning_rate": 4.835786865835024e-08,
      "loss": 0.4367,
      "step": 10467
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.381484481514656,
      "learning_rate": 4.8152639986092095e-08,
      "loss": 0.6181,
      "step": 10468
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3981316011079783,
      "learning_rate": 4.794784562397459e-08,
      "loss": 0.1987,
      "step": 10469
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2148327198528268,
      "learning_rate": 4.774348558995834e-08,
      "loss": 0.16,
      "step": 10470
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3999007137120494,
      "learning_rate": 4.75395599019679e-08,
      "loss": 0.1928,
      "step": 10471
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3175811111118283,
      "learning_rate": 4.733606857788897e-08,
      "loss": 0.1781,
      "step": 10472
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.153525431266877,
      "learning_rate": 4.713301163556894e-08,
      "loss": 0.5905,
      "step": 10473
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3683050329729685,
      "learning_rate": 4.6930389092816883e-08,
      "loss": 0.1957,
      "step": 10474
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2595833537404904,
      "learning_rate": 4.672820096740527e-08,
      "loss": 0.1287,
      "step": 10475
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.542669781350257,
      "learning_rate": 4.652644727706601e-08,
      "loss": 0.3818,
      "step": 10476
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5454733883526053,
      "learning_rate": 4.6325128039494957e-08,
      "loss": 0.2027,
      "step": 10477
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.35707305571806,
      "learning_rate": 4.612424327234966e-08,
      "loss": 0.1488,
      "step": 10478
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.488987164925197,
      "learning_rate": 4.592379299324767e-08,
      "loss": 0.1837,
      "step": 10479
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.744501887928056,
      "learning_rate": 4.572377721977106e-08,
      "loss": 0.5928,
      "step": 10480
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3150040105536653,
      "learning_rate": 4.5524195969461895e-08,
      "loss": 0.1586,
      "step": 10481
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.691422881510411,
      "learning_rate": 4.532504925982506e-08,
      "loss": 0.4907,
      "step": 10482
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5109166955761404,
      "learning_rate": 4.512633710832659e-08,
      "loss": 0.1854,
      "step": 10483
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1722295646606469,
      "learning_rate": 4.4928059532395875e-08,
      "loss": 0.1386,
      "step": 10484
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.348534893079874,
      "learning_rate": 4.473021654942178e-08,
      "loss": 0.1813,
      "step": 10485
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3371156543577805,
      "learning_rate": 4.4532808176758204e-08,
      "loss": 0.1498,
      "step": 10486
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3672771409355031,
      "learning_rate": 4.4335834431717964e-08,
      "loss": 0.1671,
      "step": 10487
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5185019940363544,
      "learning_rate": 4.413929533157668e-08,
      "loss": 0.1524,
      "step": 10488
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2847715221744143,
      "learning_rate": 4.394319089357335e-08,
      "loss": 0.1434,
      "step": 10489
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3979159012123499,
      "learning_rate": 4.374752113490643e-08,
      "loss": 0.1668,
      "step": 10490
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.049184548014845,
      "learning_rate": 4.355228607273887e-08,
      "loss": 0.2039,
      "step": 10491
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3523737769545034,
      "learning_rate": 4.335748572419252e-08,
      "loss": 0.1825,
      "step": 10492
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.318833698249875,
      "learning_rate": 4.316312010635371e-08,
      "loss": 0.14,
      "step": 10493
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.354813644556685,
      "learning_rate": 4.296918923626936e-08,
      "loss": 0.1584,
      "step": 10494
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4029139486036748,
      "learning_rate": 4.2775693130948094e-08,
      "loss": 0.1926,
      "step": 10495
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5218314125527979,
      "learning_rate": 4.258263180736133e-08,
      "loss": 0.1736,
      "step": 10496
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.233495866532101,
      "learning_rate": 4.239000528244164e-08,
      "loss": 0.5792,
      "step": 10497
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.887483194873758,
      "learning_rate": 4.219781357308383e-08,
      "loss": 0.5242,
      "step": 10498
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.565459379483048,
      "learning_rate": 4.2006056696143885e-08,
      "loss": 0.1862,
      "step": 10499
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1989589014377402,
      "learning_rate": 4.18147346684411e-08,
      "loss": 0.1739,
      "step": 10500
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.923557854920391,
      "learning_rate": 4.1623847506754855e-08,
      "loss": 0.7179,
      "step": 10501
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4232030441067947,
      "learning_rate": 4.1433395227827856e-08,
      "loss": 0.1671,
      "step": 10502
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.423691467243962,
      "learning_rate": 4.124337784836341e-08,
      "loss": 0.5153,
      "step": 10503
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.252678862519982,
      "learning_rate": 4.1053795385027074e-08,
      "loss": 0.1904,
      "step": 10504
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3518460222579016,
      "learning_rate": 4.086464785444777e-08,
      "loss": 0.168,
      "step": 10505
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2008187837172213,
      "learning_rate": 4.0675935273213894e-08,
      "loss": 0.1648,
      "step": 10506
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.248353364983745,
      "learning_rate": 4.048765765787776e-08,
      "loss": 0.1284,
      "step": 10507
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2323965561154653,
      "learning_rate": 4.029981502495117e-08,
      "loss": 0.1348,
      "step": 10508
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4820435999089023,
      "learning_rate": 4.011240739091038e-08,
      "loss": 0.1879,
      "step": 10509
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4225799444963498,
      "learning_rate": 3.9925434772191685e-08,
      "loss": 0.1572,
      "step": 10510
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2628655576375367,
      "learning_rate": 3.973889718519475e-08,
      "loss": 0.171,
      "step": 10511
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4311910085703021,
      "learning_rate": 3.955279464627926e-08,
      "loss": 0.1779,
      "step": 10512
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.664052955247472,
      "learning_rate": 3.936712717176716e-08,
      "loss": 0.5195,
      "step": 10513
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3135523684504973,
      "learning_rate": 3.918189477794432e-08,
      "loss": 0.1575,
      "step": 10514
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4839564250071884,
      "learning_rate": 3.8997097481054955e-08,
      "loss": 0.1922,
      "step": 10515
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6090272067995763,
      "learning_rate": 3.881273529730889e-08,
      "loss": 0.208,
      "step": 10516
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2765551026597441,
      "learning_rate": 3.862880824287485e-08,
      "loss": 0.1277,
      "step": 10517
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3406013271682322,
      "learning_rate": 3.844531633388437e-08,
      "loss": 0.1347,
      "step": 10518
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.1071877884440555,
      "learning_rate": 3.8262259586431815e-08,
      "loss": 0.478,
      "step": 10519
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.563964024959524,
      "learning_rate": 3.807963801657155e-08,
      "loss": 0.1693,
      "step": 10520
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5653129129270573,
      "learning_rate": 3.7897451640321326e-08,
      "loss": 0.1932,
      "step": 10521
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2904338277689937,
      "learning_rate": 3.771570047365947e-08,
      "loss": 0.1864,
      "step": 10522
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.35306793620285,
      "learning_rate": 3.753438453252767e-08,
      "loss": 0.1662,
      "step": 10523
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2561145909313907,
      "learning_rate": 3.735350383282765e-08,
      "loss": 0.1587,
      "step": 10524
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3125209537732445,
      "learning_rate": 3.717305839042451e-08,
      "loss": 0.1408,
      "step": 10525
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4114779306177836,
      "learning_rate": 3.699304822114391e-08,
      "loss": 0.1756,
      "step": 10526
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.359928221098799,
      "learning_rate": 3.681347334077434e-08,
      "loss": 0.5409,
      "step": 10527
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1999661076482386,
      "learning_rate": 3.663433376506598e-08,
      "loss": 0.1482,
      "step": 10528
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3741410356302448,
      "learning_rate": 3.645562950973014e-08,
      "loss": 0.1506,
      "step": 10529
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3792504467494602,
      "learning_rate": 3.62773605904404e-08,
      "loss": 0.1723,
      "step": 10530
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4169116170559335,
      "learning_rate": 3.6099527022832035e-08,
      "loss": 0.1698,
      "step": 10531
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4815260189161772,
      "learning_rate": 3.592212882250312e-08,
      "loss": 0.2051,
      "step": 10532
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3335917714625218,
      "learning_rate": 3.574516600501121e-08,
      "loss": 0.1739,
      "step": 10533
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3334845623850426,
      "learning_rate": 3.556863858587833e-08,
      "loss": 0.1768,
      "step": 10534
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3476566432343549,
      "learning_rate": 3.539254658058655e-08,
      "loss": 0.1602,
      "step": 10535
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.574383156984476,
      "learning_rate": 3.521689000458073e-08,
      "loss": 0.1877,
      "step": 10536
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3449665518742593,
      "learning_rate": 3.504166887326688e-08,
      "loss": 0.1866,
      "step": 10537
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.196003411634916,
      "learning_rate": 3.48668832020127e-08,
      "loss": 0.6922,
      "step": 10538
    },
    {
      "epoch": 0.96,
      "grad_norm": 9.167212531959674,
      "learning_rate": 3.469253300614817e-08,
      "loss": 0.7223,
      "step": 10539
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4033067041301481,
      "learning_rate": 3.451861830096548e-08,
      "loss": 0.1699,
      "step": 10540
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5830844949870655,
      "learning_rate": 3.434513910171799e-08,
      "loss": 0.1802,
      "step": 10541
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3248969605008885,
      "learning_rate": 3.417209542362077e-08,
      "loss": 0.1531,
      "step": 10542
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.368725970754114,
      "learning_rate": 3.39994872818511e-08,
      "loss": 0.1521,
      "step": 10543
    },
    {
      "epoch": 0.96,
      "grad_norm": 8.637095749428092,
      "learning_rate": 3.3827314691547434e-08,
      "loss": 0.6699,
      "step": 10544
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.367879129727182,
      "learning_rate": 3.365557766781047e-08,
      "loss": 0.4126,
      "step": 10545
    },
    {
      "epoch": 0.96,
      "grad_norm": 4.87162444776474,
      "learning_rate": 3.3484276225703714e-08,
      "loss": 0.507,
      "step": 10546
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3603516768940151,
      "learning_rate": 3.3313410380250157e-08,
      "loss": 0.1726,
      "step": 10547
    },
    {
      "epoch": 0.96,
      "grad_norm": 5.877360881426624,
      "learning_rate": 3.3142980146436685e-08,
      "loss": 0.498,
      "step": 10548
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3928321160122938,
      "learning_rate": 3.29729855392108e-08,
      "loss": 0.1677,
      "step": 10549
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4067324060902004,
      "learning_rate": 3.2803426573482786e-08,
      "loss": 0.1841,
      "step": 10550
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4712498838531283,
      "learning_rate": 3.263430326412298e-08,
      "loss": 0.2183,
      "step": 10551
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3263698627904288,
      "learning_rate": 3.246561562596562e-08,
      "loss": 0.1637,
      "step": 10552
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.701241889655023,
      "learning_rate": 3.229736367380498e-08,
      "loss": 0.2266,
      "step": 10553
    },
    {
      "epoch": 0.96,
      "grad_norm": 7.25062500088458,
      "learning_rate": 3.212954742239871e-08,
      "loss": 0.3701,
      "step": 10554
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3673545979722033,
      "learning_rate": 3.196216688646503e-08,
      "loss": 0.138,
      "step": 10555
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5669505227165448,
      "learning_rate": 3.1795222080683864e-08,
      "loss": 0.1778,
      "step": 10556
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2213941970431266,
      "learning_rate": 3.1628713019697946e-08,
      "loss": 0.1809,
      "step": 10557
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2546484882278501,
      "learning_rate": 3.146263971811114e-08,
      "loss": 0.1803,
      "step": 10558
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.313836590753206,
      "learning_rate": 3.1297002190489034e-08,
      "loss": 0.1451,
      "step": 10559
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3286831408066229,
      "learning_rate": 3.113180045135944e-08,
      "loss": 0.2069,
      "step": 10560
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5454137789868474,
      "learning_rate": 3.0967034515211323e-08,
      "loss": 0.1892,
      "step": 10561
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4574445153269233,
      "learning_rate": 3.080270439649591e-08,
      "loss": 0.1969,
      "step": 10562
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.475092327832501,
      "learning_rate": 3.063881010962611e-08,
      "loss": 0.4076,
      "step": 10563
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4542810952704188,
      "learning_rate": 3.0475351668976526e-08,
      "loss": 0.1423,
      "step": 10564
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3801482055302523,
      "learning_rate": 3.0312329088883484e-08,
      "loss": 0.1424,
      "step": 10565
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3759476603798837,
      "learning_rate": 3.0149742383644984e-08,
      "loss": 0.1702,
      "step": 10566
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4148779552875006,
      "learning_rate": 2.998759156752129e-08,
      "loss": 0.1414,
      "step": 10567
    },
    {
      "epoch": 0.97,
      "grad_norm": 10.616140851627828,
      "learning_rate": 2.982587665473435e-08,
      "loss": 0.4053,
      "step": 10568
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6389917369203237,
      "learning_rate": 2.966459765946672e-08,
      "loss": 0.1402,
      "step": 10569
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5786436469157428,
      "learning_rate": 2.9503754595864297e-08,
      "loss": 0.1615,
      "step": 10570
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.317924190270061,
      "learning_rate": 2.9343347478034133e-08,
      "loss": 0.1506,
      "step": 10571
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2574969502985898,
      "learning_rate": 2.9183376320044975e-08,
      "loss": 0.1644,
      "step": 10572
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.164392797752642,
      "learning_rate": 2.9023841135927822e-08,
      "loss": 0.5446,
      "step": 10573
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4290757109943788,
      "learning_rate": 2.8864741939673703e-08,
      "loss": 0.18,
      "step": 10574
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2532677545612623,
      "learning_rate": 2.8706078745237564e-08,
      "loss": 0.1357,
      "step": 10575
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.289731578230168,
      "learning_rate": 2.8547851566534947e-08,
      "loss": 0.1566,
      "step": 10576
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4682665267758879,
      "learning_rate": 2.8390060417443632e-08,
      "loss": 0.1925,
      "step": 10577
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.262232130634507,
      "learning_rate": 2.8232705311802556e-08,
      "loss": 0.1574,
      "step": 10578
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.472666719713499,
      "learning_rate": 2.8075786263414006e-08,
      "loss": 0.2117,
      "step": 10579
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2749138914803937,
      "learning_rate": 2.7919303286039202e-08,
      "loss": 0.136,
      "step": 10580
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3459658648752872,
      "learning_rate": 2.7763256393403824e-08,
      "loss": 0.1711,
      "step": 10581
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5201380475859874,
      "learning_rate": 2.7607645599194155e-08,
      "loss": 0.1623,
      "step": 10582
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3868558989106814,
      "learning_rate": 2.7452470917057605e-08,
      "loss": 0.1234,
      "step": 10583
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4220601121971126,
      "learning_rate": 2.72977323606044e-08,
      "loss": 0.1778,
      "step": 10584
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5275636057306206,
      "learning_rate": 2.714342994340646e-08,
      "loss": 0.2008,
      "step": 10585
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.099127354704506,
      "learning_rate": 2.6989563678996856e-08,
      "loss": 0.6806,
      "step": 10586
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.973984025309771,
      "learning_rate": 2.6836133580870895e-08,
      "loss": 0.6378,
      "step": 10587
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3182719252176773,
      "learning_rate": 2.668313966248559e-08,
      "loss": 0.1771,
      "step": 10588
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4298001579253823,
      "learning_rate": 2.653058193725855e-08,
      "loss": 0.192,
      "step": 10589
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4559983470896067,
      "learning_rate": 2.6378460418571284e-08,
      "loss": 0.1711,
      "step": 10590
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3463931089738406,
      "learning_rate": 2.6226775119765345e-08,
      "loss": 0.1784,
      "step": 10591
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3564811756967148,
      "learning_rate": 2.6075526054143986e-08,
      "loss": 0.1674,
      "step": 10592
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3816083752575583,
      "learning_rate": 2.592471323497381e-08,
      "loss": 0.1833,
      "step": 10593
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2054955030235899,
      "learning_rate": 2.577433667548146e-08,
      "loss": 0.1484,
      "step": 10594
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.369673514456428,
      "learning_rate": 2.5624396388856387e-08,
      "loss": 0.1703,
      "step": 10595
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3004530629984385,
      "learning_rate": 2.5474892388249185e-08,
      "loss": 0.163,
      "step": 10596
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6023857366718015,
      "learning_rate": 2.5325824686772138e-08,
      "loss": 0.1766,
      "step": 10597
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3627754356625659,
      "learning_rate": 2.5177193297499236e-08,
      "loss": 0.1738,
      "step": 10598
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2867563083553841,
      "learning_rate": 2.5028998233467272e-08,
      "loss": 0.155,
      "step": 10599
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5857984690149758,
      "learning_rate": 2.4881239507673627e-08,
      "loss": 0.2157,
      "step": 10600
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3771936794192514,
      "learning_rate": 2.4733917133077378e-08,
      "loss": 0.1728,
      "step": 10601
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4808471538301649,
      "learning_rate": 2.4587031122599857e-08,
      "loss": 0.1872,
      "step": 10602
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4085061273923283,
      "learning_rate": 2.4440581489124095e-08,
      "loss": 0.1861,
      "step": 10603
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3142209233305848,
      "learning_rate": 2.429456824549481e-08,
      "loss": 0.1537,
      "step": 10604
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2951727626947886,
      "learning_rate": 2.4148991404517873e-08,
      "loss": 0.1416,
      "step": 10605
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4124226806508273,
      "learning_rate": 2.400385097896141e-08,
      "loss": 0.1989,
      "step": 10606
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.256590734438919,
      "learning_rate": 2.3859146981556337e-08,
      "loss": 0.1722,
      "step": 10607
    },
    {
      "epoch": 0.97,
      "grad_norm": 7.369870075033695,
      "learning_rate": 2.371487942499251e-08,
      "loss": 0.5916,
      "step": 10608
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2086187064581158,
      "learning_rate": 2.3571048321923694e-08,
      "loss": 0.1869,
      "step": 10609
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.32058482139907,
      "learning_rate": 2.3427653684965357e-08,
      "loss": 0.1774,
      "step": 10610
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.874943258854614,
      "learning_rate": 2.32846955266941e-08,
      "loss": 0.423,
      "step": 10611
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2827531593973172,
      "learning_rate": 2.3142173859647675e-08,
      "loss": 0.1527,
      "step": 10612
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5267191159691202,
      "learning_rate": 2.3000088696326637e-08,
      "loss": 0.2061,
      "step": 10613
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3346629283716955,
      "learning_rate": 2.285844004919324e-08,
      "loss": 0.1659,
      "step": 10614
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.380355703956074,
      "learning_rate": 2.2717227930669772e-08,
      "loss": 0.1825,
      "step": 10615
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4980658423535351,
      "learning_rate": 2.2576452353142986e-08,
      "loss": 0.2182,
      "step": 10616
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5622061090031962,
      "learning_rate": 2.2436113328958565e-08,
      "loss": 0.1812,
      "step": 10617
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2853722000276386,
      "learning_rate": 2.2296210870426105e-08,
      "loss": 0.1557,
      "step": 10618
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4476464245226426,
      "learning_rate": 2.2156744989814683e-08,
      "loss": 0.1669,
      "step": 10619
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3745163977343084,
      "learning_rate": 2.20177156993584e-08,
      "loss": 0.1627,
      "step": 10620
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4956959085844121,
      "learning_rate": 2.187912301124917e-08,
      "loss": 0.1611,
      "step": 10621
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.839957534913192,
      "learning_rate": 2.1740966937643938e-08,
      "loss": 0.4634,
      "step": 10622
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3391416160839098,
      "learning_rate": 2.1603247490658563e-08,
      "loss": 0.1747,
      "step": 10623
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3759164078319837,
      "learning_rate": 2.146596468237283e-08,
      "loss": 0.1951,
      "step": 10624
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4788591688225872,
      "learning_rate": 2.1329118524827662e-08,
      "loss": 0.2009,
      "step": 10625
    },
    {
      "epoch": 0.97,
      "grad_norm": 6.693739773919755,
      "learning_rate": 2.119270903002457e-08,
      "loss": 0.5503,
      "step": 10626
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2269322394079045,
      "learning_rate": 2.1056736209927874e-08,
      "loss": 0.1431,
      "step": 10627
    },
    {
      "epoch": 0.97,
      "grad_norm": 4.002369620582992,
      "learning_rate": 2.0921200076463034e-08,
      "loss": 0.4674,
      "step": 10628
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5102309040224897,
      "learning_rate": 2.0786100641518313e-08,
      "loss": 0.2008,
      "step": 10629
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4722086004147827,
      "learning_rate": 2.065143791694202e-08,
      "loss": 0.1469,
      "step": 10630
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2462380264415087,
      "learning_rate": 2.0517211914545254e-08,
      "loss": 0.1396,
      "step": 10631
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1993092773908045,
      "learning_rate": 2.0383422646100824e-08,
      "loss": 0.143,
      "step": 10632
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2450355786838783,
      "learning_rate": 2.0250070123342124e-08,
      "loss": 0.1578,
      "step": 10633
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1419036968673895,
      "learning_rate": 2.0117154357965908e-08,
      "loss": 0.1201,
      "step": 10634
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4888073676414875,
      "learning_rate": 1.9984675361629512e-08,
      "loss": 0.2001,
      "step": 10635
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2717511730483257,
      "learning_rate": 1.9852633145951984e-08,
      "loss": 0.1549,
      "step": 10636
    },
    {
      "epoch": 0.97,
      "grad_norm": 8.554145966041867,
      "learning_rate": 1.9721027722514608e-08,
      "loss": 0.5497,
      "step": 10637
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.553153622901369,
      "learning_rate": 1.958985910285982e-08,
      "loss": 0.1643,
      "step": 10638
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4353174087186187,
      "learning_rate": 1.945912729849231e-08,
      "loss": 0.1524,
      "step": 10639
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.436919327906195,
      "learning_rate": 1.9328832320877898e-08,
      "loss": 0.1833,
      "step": 10640
    },
    {
      "epoch": 0.97,
      "grad_norm": 11.27292100365966,
      "learning_rate": 1.9198974181444675e-08,
      "loss": 0.5429,
      "step": 10641
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3012476817104364,
      "learning_rate": 1.90695528915813e-08,
      "loss": 0.1573,
      "step": 10642
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.316401643355429,
      "learning_rate": 1.894056846263981e-08,
      "loss": 0.1581,
      "step": 10643
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2491071820168724,
      "learning_rate": 1.881202090593226e-08,
      "loss": 0.1328,
      "step": 10644
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2053466580071468,
      "learning_rate": 1.8683910232733527e-08,
      "loss": 0.146,
      "step": 10645
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5030493174767114,
      "learning_rate": 1.855623645427962e-08,
      "loss": 0.1747,
      "step": 10646
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4441164553633103,
      "learning_rate": 1.8428999581768247e-08,
      "loss": 0.2254,
      "step": 10647
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3146960832191659,
      "learning_rate": 1.830219962635882e-08,
      "loss": 0.1763,
      "step": 10648
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3843651519313689,
      "learning_rate": 1.8175836599173545e-08,
      "loss": 0.1596,
      "step": 10649
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3928031111132635,
      "learning_rate": 1.8049910511294677e-08,
      "loss": 0.1943,
      "step": 10650
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1763808305267554,
      "learning_rate": 1.7924421373766153e-08,
      "loss": 0.1231,
      "step": 10651
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.320882170014934,
      "learning_rate": 1.779936919759473e-08,
      "loss": 0.1722,
      "step": 10652
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6350099118880874,
      "learning_rate": 1.767475399374885e-08,
      "loss": 0.1771,
      "step": 10653
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4458188407530876,
      "learning_rate": 1.7550575773156996e-08,
      "loss": 0.1536,
      "step": 10654
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3624372800760822,
      "learning_rate": 1.7426834546711567e-08,
      "loss": 0.2135,
      "step": 10655
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4646751416158597,
      "learning_rate": 1.730353032526444e-08,
      "loss": 0.1794,
      "step": 10656
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4424043513335052,
      "learning_rate": 1.7180663119630846e-08,
      "loss": 0.1656,
      "step": 10657
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.345151859957602,
      "learning_rate": 1.705823294058717e-08,
      "loss": 0.1622,
      "step": 10658
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.341055124401631,
      "learning_rate": 1.693623979887149e-08,
      "loss": 0.1687,
      "step": 10659
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5400247695422495,
      "learning_rate": 1.6814683705182467e-08,
      "loss": 0.1982,
      "step": 10660
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.311035780197929,
      "learning_rate": 1.6693564670182126e-08,
      "loss": 0.1818,
      "step": 10661
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2985743125689648,
      "learning_rate": 1.6572882704493086e-08,
      "loss": 0.1507,
      "step": 10662
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6383407279554514,
      "learning_rate": 1.6452637818700766e-08,
      "loss": 0.2159,
      "step": 10663
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.228781425503421,
      "learning_rate": 1.6332830023350065e-08,
      "loss": 0.1426,
      "step": 10664
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3947939003296252,
      "learning_rate": 1.6213459328950355e-08,
      "loss": 0.1598,
      "step": 10665
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.204988395077158,
      "learning_rate": 1.6094525745969923e-08,
      "loss": 0.5661,
      "step": 10666
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3424590307636521,
      "learning_rate": 1.5976029284840987e-08,
      "loss": 0.1666,
      "step": 10667
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.286541623555598,
      "learning_rate": 1.5857969955956344e-08,
      "loss": 0.1601,
      "step": 10668
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5839712413681442,
      "learning_rate": 1.5740347769670482e-08,
      "loss": 0.1725,
      "step": 10669
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1719644359973969,
      "learning_rate": 1.56231627362996e-08,
      "loss": 0.1667,
      "step": 10670
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.623014462420815,
      "learning_rate": 1.550641486612159e-08,
      "loss": 0.18,
      "step": 10671
    },
    {
      "epoch": 0.98,
      "grad_norm": 10.203662498322524,
      "learning_rate": 1.539010416937603e-08,
      "loss": 0.642,
      "step": 10672
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6453805393355099,
      "learning_rate": 1.5274230656263656e-08,
      "loss": 0.2028,
      "step": 10673
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5036168586805647,
      "learning_rate": 1.5158794336948556e-08,
      "loss": 0.1743,
      "step": 10674
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4076810782113367,
      "learning_rate": 1.50437952215543e-08,
      "loss": 0.1519,
      "step": 10675
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2716942775947728,
      "learning_rate": 1.492923332016727e-08,
      "loss": 0.1367,
      "step": 10676
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.151087496541748,
      "learning_rate": 1.481510864283553e-08,
      "loss": 0.5396,
      "step": 10677
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2145294199644692,
      "learning_rate": 1.47014211995683e-08,
      "loss": 0.1368,
      "step": 10678
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.063922414810894,
      "learning_rate": 1.4588171000337048e-08,
      "loss": 0.5218,
      "step": 10679
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3840042067071925,
      "learning_rate": 1.4475358055074384e-08,
      "loss": 0.1948,
      "step": 10680
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.349048830721258,
      "learning_rate": 1.4362982373675171e-08,
      "loss": 0.171,
      "step": 10681
    },
    {
      "epoch": 0.98,
      "grad_norm": 11.01938800853334,
      "learning_rate": 1.4251043965994304e-08,
      "loss": 0.5794,
      "step": 10682
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.527362874705129,
      "learning_rate": 1.4139542841850595e-08,
      "loss": 0.1522,
      "step": 10683
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2719656231477245,
      "learning_rate": 1.4028479011023444e-08,
      "loss": 0.175,
      "step": 10684
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2741613106510588,
      "learning_rate": 1.3917852483253391e-08,
      "loss": 0.1538,
      "step": 10685
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.278898476546717,
      "learning_rate": 1.3807663268243231e-08,
      "loss": 0.1881,
      "step": 10686
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3511741283744345,
      "learning_rate": 1.3697911375657457e-08,
      "loss": 0.1516,
      "step": 10687
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3415478355654267,
      "learning_rate": 1.3588596815122257e-08,
      "loss": 0.1861,
      "step": 10688
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3184451147599088,
      "learning_rate": 1.347971959622496e-08,
      "loss": 0.1734,
      "step": 10689
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4404824633752373,
      "learning_rate": 1.3371279728515152e-08,
      "loss": 0.1577,
      "step": 10690
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.197604437739786,
      "learning_rate": 1.3263277221503002e-08,
      "loss": 0.1597,
      "step": 10691
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2699385283561315,
      "learning_rate": 1.3155712084661487e-08,
      "loss": 0.1242,
      "step": 10692
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3550425626355374,
      "learning_rate": 1.304858432742473e-08,
      "loss": 0.1705,
      "step": 10693
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3179168051797334,
      "learning_rate": 1.2941893959189101e-08,
      "loss": 0.1771,
      "step": 10694
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5180650924808428,
      "learning_rate": 1.2835640989311005e-08,
      "loss": 0.2215,
      "step": 10695
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.244783021719751,
      "learning_rate": 1.2729825427110209e-08,
      "loss": 0.1511,
      "step": 10696
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2186773227803926,
      "learning_rate": 1.2624447281867625e-08,
      "loss": 0.1483,
      "step": 10697
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3822808954984556,
      "learning_rate": 1.2519506562825301e-08,
      "loss": 0.1721,
      "step": 10698
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3861673836304012,
      "learning_rate": 1.2415003279186988e-08,
      "loss": 0.1649,
      "step": 10699
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.287349431183091,
      "learning_rate": 1.2310937440119241e-08,
      "loss": 0.681,
      "step": 10700
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.9321672746432395,
      "learning_rate": 1.2207309054748096e-08,
      "loss": 0.4286,
      "step": 10701
    },
    {
      "epoch": 0.98,
      "grad_norm": 4.536827390537436,
      "learning_rate": 1.21041181321635e-08,
      "loss": 0.5235,
      "step": 10702
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3021843404136577,
      "learning_rate": 1.200136468141544e-08,
      "loss": 0.1848,
      "step": 10703
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3905856524364562,
      "learning_rate": 1.1899048711516148e-08,
      "loss": 0.18,
      "step": 10704
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.265334725406974,
      "learning_rate": 1.1797170231439004e-08,
      "loss": 0.1491,
      "step": 10705
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2553703439220532,
      "learning_rate": 1.1695729250120747e-08,
      "loss": 0.1384,
      "step": 10706
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.079326131895326,
      "learning_rate": 1.159472577645704e-08,
      "loss": 0.7143,
      "step": 10707
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.370896089268078,
      "learning_rate": 1.1494159819306905e-08,
      "loss": 0.1846,
      "step": 10708
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3515328431627425,
      "learning_rate": 1.1394031387491622e-08,
      "loss": 0.1571,
      "step": 10709
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.634097202421853,
      "learning_rate": 1.1294340489791944e-08,
      "loss": 0.6347,
      "step": 10710
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.511651988856716,
      "learning_rate": 1.1195087134951431e-08,
      "loss": 0.1626,
      "step": 10711
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.181446371749484,
      "learning_rate": 1.1096271331675902e-08,
      "loss": 0.5143,
      "step": 10712
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.358131920172868,
      "learning_rate": 1.0997893088632306e-08,
      "loss": 0.1499,
      "step": 10713
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2859355012277713,
      "learning_rate": 1.089995241444819e-08,
      "loss": 0.192,
      "step": 10714
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.993645658578931,
      "learning_rate": 1.0802449317714459e-08,
      "loss": 0.5673,
      "step": 10715
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2677553338064114,
      "learning_rate": 1.0705383806982606e-08,
      "loss": 0.1711,
      "step": 10716
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4282521488547624,
      "learning_rate": 1.0608755890765265e-08,
      "loss": 0.152,
      "step": 10717
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1765422251286144,
      "learning_rate": 1.0512565577538437e-08,
      "loss": 0.1555,
      "step": 10718
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4790297905912764,
      "learning_rate": 1.0416812875737591e-08,
      "loss": 0.1992,
      "step": 10719
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.516249074196245,
      "learning_rate": 1.032149779376157e-08,
      "loss": 0.2072,
      "step": 10720
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.367160957112398,
      "learning_rate": 1.0226620339969795e-08,
      "loss": 0.1983,
      "step": 10721
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3331284222113875,
      "learning_rate": 1.0132180522683943e-08,
      "loss": 0.1865,
      "step": 10722
    },
    {
      "epoch": 0.98,
      "grad_norm": 8.027439989142263,
      "learning_rate": 1.0038178350186834e-08,
      "loss": 0.8282,
      "step": 10723
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2914601239147587,
      "learning_rate": 9.944613830722982e-09,
      "loss": 0.1667,
      "step": 10724
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.301984385152835,
      "learning_rate": 9.851486972498603e-09,
      "loss": 0.5507,
      "step": 10725
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3746932071547624,
      "learning_rate": 9.758797783682162e-09,
      "loss": 0.1706,
      "step": 10726
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4213179393757713,
      "learning_rate": 9.666546272402711e-09,
      "loss": 0.2284,
      "step": 10727
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3143662923868946,
      "learning_rate": 9.574732446750445e-09,
      "loss": 0.1428,
      "step": 10728
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.397860034376929,
      "learning_rate": 9.48335631477948e-09,
      "loss": 0.5575,
      "step": 10729
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4541437585616148,
      "learning_rate": 9.392417884503401e-09,
      "loss": 0.1798,
      "step": 10730
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3901926910949958,
      "learning_rate": 9.30191716389861e-09,
      "loss": 0.1632,
      "step": 10731
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.309305480982929,
      "learning_rate": 9.211854160902089e-09,
      "loss": 0.152,
      "step": 10732
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3565772225463115,
      "learning_rate": 9.12222888341252e-09,
      "loss": 0.1619,
      "step": 10733
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.249242705491855,
      "learning_rate": 9.033041339291948e-09,
      "loss": 0.1631,
      "step": 10734
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.357087448968722,
      "learning_rate": 8.944291536361893e-09,
      "loss": 0.1606,
      "step": 10735
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4082939474377547,
      "learning_rate": 8.85597948240613e-09,
      "loss": 0.133,
      "step": 10736
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.295724949492852,
      "learning_rate": 8.768105185170683e-09,
      "loss": 0.2208,
      "step": 10737
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5366944246698349,
      "learning_rate": 8.680668652362168e-09,
      "loss": 0.1683,
      "step": 10738
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2768957378090409,
      "learning_rate": 8.593669891650004e-09,
      "loss": 0.154,
      "step": 10739
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4784390171584387,
      "learning_rate": 8.507108910664752e-09,
      "loss": 0.1685,
      "step": 10740
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4300369390852194,
      "learning_rate": 8.42098571699701e-09,
      "loss": 0.1652,
      "step": 10741
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.329437182077071,
      "learning_rate": 8.335300318201844e-09,
      "loss": 0.166,
      "step": 10742
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3992133950971601,
      "learning_rate": 8.2500527217938e-09,
      "loss": 0.1618,
      "step": 10743
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5424427391374058,
      "learning_rate": 8.165242935250228e-09,
      "loss": 0.1708,
      "step": 10744
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4281799009152836,
      "learning_rate": 8.080870966008513e-09,
      "loss": 0.19,
      "step": 10745
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2351837875719538,
      "learning_rate": 7.996936821468848e-09,
      "loss": 0.1426,
      "step": 10746
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.596963518090819,
      "learning_rate": 7.913440508993674e-09,
      "loss": 0.2044,
      "step": 10747
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.422661531704769,
      "learning_rate": 7.830382035904915e-09,
      "loss": 0.593,
      "step": 10748
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2863412730459391,
      "learning_rate": 7.747761409488964e-09,
      "loss": 0.1488,
      "step": 10749
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4820933122291415,
      "learning_rate": 7.665578636990578e-09,
      "loss": 0.1839,
      "step": 10750
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4245591195491332,
      "learning_rate": 7.583833725618439e-09,
      "loss": 0.1972,
      "step": 10751
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.218683768922726,
      "learning_rate": 7.502526682542921e-09,
      "loss": 0.4247,
      "step": 10752
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.8009550105169865,
      "learning_rate": 7.421657514893321e-09,
      "loss": 0.481,
      "step": 10753
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3370449369878157,
      "learning_rate": 7.34122622976452e-09,
      "loss": 0.1589,
      "step": 10754
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4276038889565035,
      "learning_rate": 7.261232834209208e-09,
      "loss": 0.1752,
      "step": 10755
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.322819913798217,
      "learning_rate": 7.181677335243997e-09,
      "loss": 0.1802,
      "step": 10756
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4185881347934448,
      "learning_rate": 7.102559739846637e-09,
      "loss": 0.1765,
      "step": 10757
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3855945019503493,
      "learning_rate": 7.023880054956022e-09,
      "loss": 0.161,
      "step": 10758
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2624224672421185,
      "learning_rate": 6.9456382874727445e-09,
      "loss": 0.1616,
      "step": 10759
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3804105603743075,
      "learning_rate": 6.867834444259647e-09,
      "loss": 0.1296,
      "step": 10760
    },
    {
      "epoch": 0.98,
      "grad_norm": 7.698602562418363,
      "learning_rate": 6.79046853214016e-09,
      "loss": 0.6059,
      "step": 10761
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3087145589555667,
      "learning_rate": 6.713540557899967e-09,
      "loss": 0.1619,
      "step": 10762
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.436325133358082,
      "learning_rate": 6.637050528286448e-09,
      "loss": 0.5285,
      "step": 10763
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2573766046280068,
      "learning_rate": 6.5609984500075715e-09,
      "loss": 0.1636,
      "step": 10764
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6185261457867675,
      "learning_rate": 6.485384329734112e-09,
      "loss": 0.1823,
      "step": 10765
    },
    {
      "epoch": 0.98,
      "grad_norm": 20.77386192723569,
      "learning_rate": 6.410208174097432e-09,
      "loss": 0.5732,
      "step": 10766
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2457502918521743,
      "learning_rate": 6.335469989692255e-09,
      "loss": 0.1758,
      "step": 10767
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4961238179554597,
      "learning_rate": 6.2611697830722295e-09,
      "loss": 0.1691,
      "step": 10768
    },
    {
      "epoch": 0.98,
      "grad_norm": 5.729207286470654,
      "learning_rate": 6.187307560754363e-09,
      "loss": 0.455,
      "step": 10769
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3469458732590256,
      "learning_rate": 6.113883329217363e-09,
      "loss": 0.1763,
      "step": 10770
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3684837201503879,
      "learning_rate": 6.040897094900522e-09,
      "loss": 0.1741,
      "step": 10771
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5020843068601626,
      "learning_rate": 5.9683488642053865e-09,
      "loss": 0.1666,
      "step": 10772
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.992667116326784,
      "learning_rate": 5.8962386434952e-09,
      "loss": 0.5526,
      "step": 10773
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.999133848583205,
      "learning_rate": 5.824566439093793e-09,
      "loss": 0.7616,
      "step": 10774
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.423461812120471,
      "learning_rate": 5.753332257287802e-09,
      "loss": 0.1856,
      "step": 10775
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3802987186968005,
      "learning_rate": 5.6825361043250095e-09,
      "loss": 0.164,
      "step": 10776
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2127824039190893,
      "learning_rate": 5.612177986414891e-09,
      "loss": 0.1245,
      "step": 10777
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.500317533741518,
      "learning_rate": 5.542257909727511e-09,
      "loss": 0.1714,
      "step": 10778
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4137895156438252,
      "learning_rate": 5.472775880395742e-09,
      "loss": 0.1804,
      "step": 10779
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5781766290448827,
      "learning_rate": 5.4037319045141534e-09,
      "loss": 0.2181,
      "step": 10780
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.294661351287554,
      "learning_rate": 5.3351259881379016e-09,
      "loss": 0.1294,
      "step": 10781
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2899883767941178,
      "learning_rate": 5.266958137283839e-09,
      "loss": 0.1303,
      "step": 10782
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4309085053827288,
      "learning_rate": 5.199228357931075e-09,
      "loss": 0.1932,
      "step": 10783
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3329594095159902,
      "learning_rate": 5.131936656020409e-09,
      "loss": 0.1591,
      "step": 10784
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4360739838552636,
      "learning_rate": 5.065083037453234e-09,
      "loss": 0.1687,
      "step": 10785
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3162605134811416,
      "learning_rate": 4.998667508093191e-09,
      "loss": 0.1418,
      "step": 10786
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2659303893932354,
      "learning_rate": 4.932690073765068e-09,
      "loss": 0.1438,
      "step": 10787
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3485440547691856,
      "learning_rate": 4.867150740255899e-09,
      "loss": 0.1609,
      "step": 10788
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2889630354369486,
      "learning_rate": 4.802049513313866e-09,
      "loss": 0.151,
      "step": 10789
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4540430441597783,
      "learning_rate": 4.737386398648846e-09,
      "loss": 0.1834,
      "step": 10790
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2236682711850353,
      "learning_rate": 4.673161401932413e-09,
      "loss": 0.1388,
      "step": 10791
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1711020867190198,
      "learning_rate": 4.609374528796728e-09,
      "loss": 0.1269,
      "step": 10792
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.332787972058241,
      "learning_rate": 4.546025784837316e-09,
      "loss": 0.1485,
      "step": 10793
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5138478299868632,
      "learning_rate": 4.4831151756091766e-09,
      "loss": 0.216,
      "step": 10794
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.7346371711748105,
      "learning_rate": 4.420642706631229e-09,
      "loss": 0.4771,
      "step": 10795
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5650233798984943,
      "learning_rate": 4.358608383381868e-09,
      "loss": 0.182,
      "step": 10796
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4113111060532724,
      "learning_rate": 4.297012211302298e-09,
      "loss": 0.1858,
      "step": 10797
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3260518484012584,
      "learning_rate": 4.235854195794309e-09,
      "loss": 0.1389,
      "step": 10798
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4014592006207787,
      "learning_rate": 4.175134342223053e-09,
      "loss": 0.1804,
      "step": 10799
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3455737609003986,
      "learning_rate": 4.1148526559131595e-09,
      "loss": 0.1634,
      "step": 10800
    },
    {
      "epoch": 0.99,
      "grad_norm": 9.211168821863629,
      "learning_rate": 4.055009142152066e-09,
      "loss": 0.665,
      "step": 10801
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.610875804765072,
      "learning_rate": 3.9956038061877975e-09,
      "loss": 0.2105,
      "step": 10802
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4340021508454324,
      "learning_rate": 3.93663665323174e-09,
      "loss": 0.1917,
      "step": 10803
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.192634404965198,
      "learning_rate": 3.878107688454758e-09,
      "loss": 0.1387,
      "step": 10804
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7570381341626335,
      "learning_rate": 3.820016916990521e-09,
      "loss": 0.202,
      "step": 10805
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4799211295877015,
      "learning_rate": 3.762364343934399e-09,
      "loss": 0.179,
      "step": 10806
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.417241766713626,
      "learning_rate": 3.705149974342348e-09,
      "loss": 0.1569,
      "step": 10807
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1390157335558395,
      "learning_rate": 3.648373813232575e-09,
      "loss": 0.1449,
      "step": 10808
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.732514453946708,
      "learning_rate": 3.5920358655844312e-09,
      "loss": 0.4902,
      "step": 10809
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3215677648120923,
      "learning_rate": 3.5361361363400735e-09,
      "loss": 0.1686,
      "step": 10810
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.440492620915381,
      "learning_rate": 3.4806746304016924e-09,
      "loss": 0.5473,
      "step": 10811
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7083795820415555,
      "learning_rate": 3.425651352633175e-09,
      "loss": 0.2182,
      "step": 10812
    },
    {
      "epoch": 0.99,
      "grad_norm": 13.965290511520413,
      "learning_rate": 3.371066307861215e-09,
      "loss": 0.421,
      "step": 10813
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5058954791218142,
      "learning_rate": 3.316919500873095e-09,
      "loss": 0.1726,
      "step": 10814
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3518288398319842,
      "learning_rate": 3.2632109364177934e-09,
      "loss": 0.1448,
      "step": 10815
    },
    {
      "epoch": 0.99,
      "grad_norm": 13.554330821889875,
      "learning_rate": 3.2099406192054315e-09,
      "loss": 0.4392,
      "step": 10816
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.240961439389439,
      "learning_rate": 3.1571085539089384e-09,
      "loss": 0.5064,
      "step": 10817
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3792237768480233,
      "learning_rate": 3.1047147451618297e-09,
      "loss": 0.1618,
      "step": 10818
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3977050973738756,
      "learning_rate": 3.0527591975587632e-09,
      "loss": 0.1906,
      "step": 10819
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5650435740907966,
      "learning_rate": 3.0012419156572047e-09,
      "loss": 0.1833,
      "step": 10820
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.281998986029381,
      "learning_rate": 2.950162903975762e-09,
      "loss": 0.1738,
      "step": 10821
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3302260494155027,
      "learning_rate": 2.8995221669941843e-09,
      "loss": 0.1821,
      "step": 10822
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.236583697306866,
      "learning_rate": 2.849319709153364e-09,
      "loss": 0.6602,
      "step": 10823
    },
    {
      "epoch": 0.99,
      "grad_norm": 6.132742025221653,
      "learning_rate": 2.7995555348570014e-09,
      "loss": 0.4725,
      "step": 10824
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3116976801503697,
      "learning_rate": 2.7502296484699374e-09,
      "loss": 0.1726,
      "step": 10825
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2798902063751336,
      "learning_rate": 2.701342054318157e-09,
      "loss": 0.1468,
      "step": 10826
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0714895354395688,
      "learning_rate": 2.6528927566893405e-09,
      "loss": 0.1448,
      "step": 10827
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.422957131719553,
      "learning_rate": 2.6048817598328667e-09,
      "loss": 0.1969,
      "step": 10828
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.883844807805086,
      "learning_rate": 2.5573090679592573e-09,
      "loss": 0.6939,
      "step": 10829
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.238436339827715,
      "learning_rate": 2.51017468524184e-09,
      "loss": 0.6307,
      "step": 10830
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3058286818691989,
      "learning_rate": 2.4634786158139747e-09,
      "loss": 0.1737,
      "step": 10831
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5121270804023796,
      "learning_rate": 2.4172208637707195e-09,
      "loss": 0.1723,
      "step": 10832
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5082435725811962,
      "learning_rate": 2.371401433170495e-09,
      "loss": 0.172,
      "step": 10833
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1524326582042699,
      "learning_rate": 2.326020328031198e-09,
      "loss": 0.1522,
      "step": 10834
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4919295909977708,
      "learning_rate": 2.2810775523329775e-09,
      "loss": 0.214,
      "step": 10835
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.455663967305135,
      "learning_rate": 2.2365731100182363e-09,
      "loss": 0.1862,
      "step": 10836
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2026856761124103,
      "learning_rate": 2.192507004989408e-09,
      "loss": 0.1479,
      "step": 10837
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4281258256391027,
      "learning_rate": 2.148879241111734e-09,
      "loss": 0.1498,
      "step": 10838
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3975993933041797,
      "learning_rate": 2.1056898222121535e-09,
      "loss": 0.1681,
      "step": 10839
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4360142995146348,
      "learning_rate": 2.062938752078192e-09,
      "loss": 0.1706,
      "step": 10840
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5336242588424145,
      "learning_rate": 2.0206260344590724e-09,
      "loss": 0.1681,
      "step": 10841
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.607906917889805,
      "learning_rate": 1.9787516730668253e-09,
      "loss": 0.1824,
      "step": 10842
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6028032151042757,
      "learning_rate": 1.9373156715735143e-09,
      "loss": 0.1682,
      "step": 10843
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5111012361716476,
      "learning_rate": 1.8963180336128983e-09,
      "loss": 0.1686,
      "step": 10844
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2826272265247096,
      "learning_rate": 1.8557587627821005e-09,
      "loss": 0.153,
      "step": 10845
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.440574213664256,
      "learning_rate": 1.815637862637165e-09,
      "loss": 0.5472,
      "step": 10846
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4251142088393831,
      "learning_rate": 1.775955336697499e-09,
      "loss": 0.1877,
      "step": 10847
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.103181245093823,
      "learning_rate": 1.7367111884436516e-09,
      "loss": 0.468,
      "step": 10848
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.866046199260645,
      "learning_rate": 1.6979054213173141e-09,
      "loss": 0.5313,
      "step": 10849
    },
    {
      "epoch": 0.99,
      "grad_norm": 13.427526332527274,
      "learning_rate": 1.659538038721875e-09,
      "loss": 0.6029,
      "step": 10850
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5112976049001297,
      "learning_rate": 1.6216090440224207e-09,
      "loss": 0.1801,
      "step": 10851
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1440834003077895,
      "learning_rate": 1.5841184405462895e-09,
      "loss": 0.1296,
      "step": 10852
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5258831757444071,
      "learning_rate": 1.5470662315808516e-09,
      "loss": 0.1815,
      "step": 10853
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4108851622152334,
      "learning_rate": 1.51045242037684e-09,
      "loss": 0.1917,
      "step": 10854
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4203441167537756,
      "learning_rate": 1.4742770101439097e-09,
      "loss": 0.1476,
      "step": 10855
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.357441954509869,
      "learning_rate": 1.4385400040561882e-09,
      "loss": 0.1945,
      "step": 10856
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.453064233809701,
      "learning_rate": 1.4032414052478348e-09,
      "loss": 0.1693,
      "step": 10857
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5335998262654271,
      "learning_rate": 1.368381216814707e-09,
      "loss": 0.2076,
      "step": 10858
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3123918519300515,
      "learning_rate": 1.3339594418138036e-09,
      "loss": 0.1379,
      "step": 10859
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3607344210904144,
      "learning_rate": 1.299976083264376e-09,
      "loss": 0.1709,
      "step": 10860
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4800459666421006,
      "learning_rate": 1.2664311441473732e-09,
      "loss": 0.1959,
      "step": 10861
    },
    {
      "epoch": 0.99,
      "grad_norm": 7.092953244513387,
      "learning_rate": 1.233324627404886e-09,
      "loss": 0.5642,
      "step": 10862
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2395852200995525,
      "learning_rate": 1.2006565359401479e-09,
      "loss": 0.1724,
      "step": 10863
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.558927664804295,
      "learning_rate": 1.1684268726186443e-09,
      "loss": 0.155,
      "step": 10864
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2971870405401011,
      "learning_rate": 1.136635640267003e-09,
      "loss": 0.174,
      "step": 10865
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2675199869253102,
      "learning_rate": 1.1052828416735495e-09,
      "loss": 0.1612,
      "step": 10866
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.279083397636326,
      "learning_rate": 1.0743684795883057e-09,
      "loss": 0.1803,
      "step": 10867
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3468822603544406,
      "learning_rate": 1.0438925567229919e-09,
      "loss": 0.1677,
      "step": 10868
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.386493901444284,
      "learning_rate": 1.0138550757493592e-09,
      "loss": 0.1524,
      "step": 10869
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7631198833909336,
      "learning_rate": 9.842560393030776e-10,
      "loss": 0.1742,
      "step": 10870
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.292388875308847,
      "learning_rate": 9.550954499792931e-10,
      "loss": 0.1751,
      "step": 10871
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3018389219252164,
      "learning_rate": 9.263733103365147e-10,
      "loss": 0.1532,
      "step": 10872
    },
    {
      "epoch": 0.99,
      "grad_norm": 8.680436397672153,
      "learning_rate": 8.980896228932834e-10,
      "loss": 0.6088,
      "step": 10873
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2957926202844954,
      "learning_rate": 8.702443901303925e-10,
      "loss": 0.179,
      "step": 10874
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2907013355918473,
      "learning_rate": 8.428376144897776e-10,
      "loss": 0.1648,
      "step": 10875
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.190469628268926,
      "learning_rate": 8.158692983756266e-10,
      "loss": 0.1437,
      "step": 10876
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.371547629142922,
      "learning_rate": 7.893394441532698e-10,
      "loss": 0.1946,
      "step": 10877
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4394088693768905,
      "learning_rate": 7.632480541491793e-10,
      "loss": 0.189,
      "step": 10878
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4772864241968229,
      "learning_rate": 7.37595130651525e-10,
      "loss": 0.1813,
      "step": 10879
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3571333868582856,
      "learning_rate": 7.123806759112839e-10,
      "loss": 0.1759,
      "step": 10880
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3380517979274926,
      "learning_rate": 6.876046921389102e-10,
      "loss": 0.1528,
      "step": 10881
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4510333771545527,
      "learning_rate": 6.632671815076653e-10,
      "loss": 0.162,
      "step": 10882
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3723886530877727,
      "learning_rate": 6.393681461525081e-10,
      "loss": 0.202,
      "step": 10883
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.469903290420173,
      "learning_rate": 6.159075881689846e-10,
      "loss": 0.1516,
      "step": 10884
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3223752217853655,
      "learning_rate": 5.928855096154485e-10,
      "loss": 0.1779,
      "step": 10885
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4718036622145363,
      "learning_rate": 5.703019125102849e-10,
      "loss": 0.1708,
      "step": 10886
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1651367869804743,
      "learning_rate": 5.48156798834687e-10,
      "loss": 0.1436,
      "step": 10887
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2786565230872906,
      "learning_rate": 5.264501705309899e-10,
      "loss": 0.1807,
      "step": 10888
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4834662221862838,
      "learning_rate": 5.051820295032262e-10,
      "loss": 0.1776,
      "step": 10889
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5964331888607435,
      "learning_rate": 4.843523776160153e-10,
      "loss": 0.1898,
      "step": 10890
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.383384133586915,
      "learning_rate": 4.639612166967844e-10,
      "loss": 0.1657,
      "step": 10891
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3162131054128157,
      "learning_rate": 4.440085485335477e-10,
      "loss": 0.1671,
      "step": 10892
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2609610428720004,
      "learning_rate": 4.2449437487657173e-10,
      "loss": 0.1394,
      "step": 10893
    },
    {
      "epoch": 1.0,
      "grad_norm": 10.181383943767681,
      "learning_rate": 4.054186974378205e-10,
      "loss": 0.668,
      "step": 10894
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3621085107581037,
      "learning_rate": 3.867815178892897e-10,
      "loss": 0.1587,
      "step": 10895
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4651400097115077,
      "learning_rate": 3.685828378668932e-10,
      "loss": 0.1777,
      "step": 10896
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3810517363046522,
      "learning_rate": 3.508226589660213e-10,
      "loss": 0.1863,
      "step": 10897
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8620127637482398,
      "learning_rate": 3.335009827437619e-10,
      "loss": 0.2028,
      "step": 10898
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4674430686880717,
      "learning_rate": 3.166178107205653e-10,
      "loss": 0.1913,
      "step": 10899
    },
    {
      "epoch": 1.0,
      "grad_norm": 7.914440747762613,
      "learning_rate": 3.00173144376914e-10,
      "loss": 0.5008,
      "step": 10900
    },
    {
      "epoch": 1.0,
      "grad_norm": 8.54444969338757,
      "learning_rate": 2.8416698515443263e-10,
      "loss": 0.6739,
      "step": 10901
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5083049138514835,
      "learning_rate": 2.685993344575533e-10,
      "loss": 0.1991,
      "step": 10902
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1755157056806778,
      "learning_rate": 2.534701936512951e-10,
      "loss": 0.1191,
      "step": 10903
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6248316422796991,
      "learning_rate": 2.387795640629298e-10,
      "loss": 0.2142,
      "step": 10904
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2386817250496553,
      "learning_rate": 2.2452744698087114e-10,
      "loss": 0.167,
      "step": 10905
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.264900392824305,
      "learning_rate": 2.107138436546752e-10,
      "loss": 0.1526,
      "step": 10906
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2290559500603924,
      "learning_rate": 1.9733875529670542e-10,
      "loss": 0.121,
      "step": 10907
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5012042312670526,
      "learning_rate": 1.844021830793574e-10,
      "loss": 0.1639,
      "step": 10908
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3621259858143306,
      "learning_rate": 1.719041281378342e-10,
      "loss": 0.1569,
      "step": 10909
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.314745781322409,
      "learning_rate": 1.5984459156737076e-10,
      "loss": 0.1668,
      "step": 10910
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.423329998515394,
      "learning_rate": 1.4822357442656475e-10,
      "loss": 0.1835,
      "step": 10911
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5607944746052465,
      "learning_rate": 1.3704107773460095e-10,
      "loss": 0.1624,
      "step": 10912
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.368356133257762,
      "learning_rate": 1.2629710247180626e-10,
      "loss": 0.1493,
      "step": 10913
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3961883856950204,
      "learning_rate": 1.1599164958020492e-10,
      "loss": 0.1701,
      "step": 10914
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3429211902478448,
      "learning_rate": 1.061247199646287e-10,
      "loss": 0.1538,
      "step": 10915
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3037904591994756,
      "learning_rate": 9.669631448994132e-11,
      "loss": 0.1745,
      "step": 10916
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2967998560015501,
      "learning_rate": 8.770643398325896e-11,
      "loss": 0.1718,
      "step": 10917
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3845629496084144,
      "learning_rate": 7.915507923283994e-11,
      "loss": 0.1525,
      "step": 10918
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3584639643172074,
      "learning_rate": 7.104225098863992e-11,
      "loss": 0.1997,
      "step": 10919
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3064593702451792,
      "learning_rate": 6.336794996231188e-11,
      "loss": 0.1773,
      "step": 10920
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5219304266889706,
      "learning_rate": 5.613217682720606e-11,
      "loss": 0.1781,
      "step": 10921
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3667444479207913,
      "learning_rate": 4.9334932217814934e-11,
      "loss": 0.1766,
      "step": 10922
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.31493255561236,
      "learning_rate": 4.2976216729773145e-11,
      "loss": 0.1563,
      "step": 10923
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1559444729552368,
      "learning_rate": 3.7056030921522877e-11,
      "loss": 0.1394,
      "step": 10924
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2469713194538998,
      "learning_rate": 3.1574375312093396e-11,
      "loss": 0.1574,
      "step": 10925
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.513962356158132,
      "learning_rate": 2.653125038165616e-11,
      "loss": 0.1653,
      "step": 10926
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3292050812469425,
      "learning_rate": 2.1926656573745263e-11,
      "loss": 0.1506,
      "step": 10927
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.26651176140228,
      "learning_rate": 1.7760594290816556e-11,
      "loss": 0.1385,
      "step": 10928
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2326677826152548,
      "learning_rate": 1.4033063899243637e-11,
      "loss": 0.1653,
      "step": 10929
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3994313881148177,
      "learning_rate": 1.0744065725987185e-11,
      "loss": 0.174,
      "step": 10930
    },
    {
      "epoch": 1.0,
      "grad_norm": 5.310423397705133,
      "learning_rate": 7.893600059150075e-12,
      "loss": 0.5443,
      "step": 10931
    },
    {
      "epoch": 1.0,
      "grad_norm": 11.044111509116139,
      "learning_rate": 5.481667148532488e-12,
      "loss": 0.5778,
      "step": 10932
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.707690077325594,
      "learning_rate": 3.50826720563191e-12,
      "loss": 0.514,
      "step": 10933
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3673223408635384,
      "learning_rate": 1.973400404198245e-12,
      "loss": 0.188,
      "step": 10934
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2770298796837296,
      "learning_rate": 8.770668785684777e-13,
      "loss": 0.1592,
      "step": 10935
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4645349824955611,
      "learning_rate": 2.192667242217894e-13,
      "loss": 0.1892,
      "step": 10936
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.900706792298434,
      "learning_rate": 0.0,
      "loss": 0.3235,
      "step": 10937
    },
    {
      "epoch": 1.0,
      "step": 10937,
      "total_flos": 5082876500559872.0,
      "train_loss": 0.2547208672946121,
      "train_runtime": 32462.8361,
      "train_samples_per_second": 5.391,
      "train_steps_per_second": 0.337
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 10937,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 5082876500559872.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}