{
  "best_metric": 1.4320533275604248,
  "best_model_checkpoint": "model/chessformer-3/checkpoint-22000",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 22491,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.892445867235783e-05,
      "grad_norm": 15.056964874267578,
      "learning_rate": 2e-06,
      "loss": 8.6442,
      "step": 2
    },
    {
      "epoch": 0.00017784891734471566,
      "grad_norm": 12.359731674194336,
      "learning_rate": 4e-06,
      "loss": 8.4787,
      "step": 4
    },
    {
      "epoch": 0.0002667733760170735,
      "grad_norm": 6.386526584625244,
      "learning_rate": 6e-06,
      "loss": 8.1571,
      "step": 6
    },
    {
      "epoch": 0.0003556978346894313,
      "grad_norm": 3.8906912803649902,
      "learning_rate": 8e-06,
      "loss": 7.9194,
      "step": 8
    },
    {
      "epoch": 0.00044462229336178917,
      "grad_norm": 2.871436357498169,
      "learning_rate": 1e-05,
      "loss": 7.7565,
      "step": 10
    },
    {
      "epoch": 0.000533546752034147,
      "grad_norm": 2.3168184757232666,
      "learning_rate": 1.2e-05,
      "loss": 7.6304,
      "step": 12
    },
    {
      "epoch": 0.0006224712107065049,
      "grad_norm": 1.9175453186035156,
      "learning_rate": 1.4e-05,
      "loss": 7.5403,
      "step": 14
    },
    {
      "epoch": 0.0007113956693788626,
      "grad_norm": 3.3437652587890625,
      "learning_rate": 1.6e-05,
      "loss": 7.464,
      "step": 16
    },
    {
      "epoch": 0.0008003201280512205,
      "grad_norm": 1.4156451225280762,
      "learning_rate": 1.8e-05,
      "loss": 7.4051,
      "step": 18
    },
    {
      "epoch": 0.0008892445867235783,
      "grad_norm": 2.325925588607788,
      "learning_rate": 2e-05,
      "loss": 7.3434,
      "step": 20
    },
    {
      "epoch": 0.0009781690453959362,
      "grad_norm": 1.561901569366455,
      "learning_rate": 2.2e-05,
      "loss": 7.2769,
      "step": 22
    },
    {
      "epoch": 0.001067093504068294,
      "grad_norm": 1.9596675634384155,
      "learning_rate": 2.4e-05,
      "loss": 7.1864,
      "step": 24
    },
    {
      "epoch": 0.0011560179627406518,
      "grad_norm": 3.5041377544403076,
      "learning_rate": 2.6e-05,
      "loss": 7.0938,
      "step": 26
    },
    {
      "epoch": 0.0012449424214130097,
      "grad_norm": 9.32341480255127,
      "learning_rate": 2.8e-05,
      "loss": 7.0071,
      "step": 28
    },
    {
      "epoch": 0.0013338668800853674,
      "grad_norm": 7.433465480804443,
      "learning_rate": 3e-05,
      "loss": 6.9707,
      "step": 30
    },
    {
      "epoch": 0.0014227913387577253,
      "grad_norm": 32.543617248535156,
      "learning_rate": 3.2e-05,
      "loss": 6.9459,
      "step": 32
    },
    {
      "epoch": 0.0015117157974300832,
      "grad_norm": 19.157636642456055,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 6.9043,
      "step": 34
    },
    {
      "epoch": 0.001600640256102441,
      "grad_norm": 4.467665672302246,
      "learning_rate": 3.6e-05,
      "loss": 6.8362,
      "step": 36
    },
    {
      "epoch": 0.0016895647147747988,
      "grad_norm": 9.15857982635498,
      "learning_rate": 3.8e-05,
      "loss": 6.7782,
      "step": 38
    },
    {
      "epoch": 0.0017784891734471567,
      "grad_norm": 11.840293884277344,
      "learning_rate": 4e-05,
      "loss": 6.7454,
      "step": 40
    },
    {
      "epoch": 0.0018674136321195146,
      "grad_norm": 5.0099263191223145,
      "learning_rate": 4.2000000000000004e-05,
      "loss": 6.699,
      "step": 42
    },
    {
      "epoch": 0.0019563380907918725,
      "grad_norm": 5.58508825302124,
      "learning_rate": 4.4e-05,
      "loss": 6.6336,
      "step": 44
    },
    {
      "epoch": 0.00204526254946423,
      "grad_norm": 4.943291187286377,
      "learning_rate": 4.6e-05,
      "loss": 6.6033,
      "step": 46
    },
    {
      "epoch": 0.002134187008136588,
      "grad_norm": 3.4954986572265625,
      "learning_rate": 4.8e-05,
      "loss": 6.5511,
      "step": 48
    },
    {
      "epoch": 0.002223111466808946,
      "grad_norm": 1.8841090202331543,
      "learning_rate": 5e-05,
      "loss": 6.4977,
      "step": 50
    },
    {
      "epoch": 0.0023120359254813036,
      "grad_norm": 2.3768599033355713,
      "learning_rate": 5.2e-05,
      "loss": 6.4315,
      "step": 52
    },
    {
      "epoch": 0.0024009603841536613,
      "grad_norm": 2.597606658935547,
      "learning_rate": 5.4e-05,
      "loss": 6.3706,
      "step": 54
    },
    {
      "epoch": 0.0024898848428260194,
      "grad_norm": 5.908978462219238,
      "learning_rate": 5.6e-05,
      "loss": 6.3226,
      "step": 56
    },
    {
      "epoch": 0.002578809301498377,
      "grad_norm": 3.5320820808410645,
      "learning_rate": 5.800000000000001e-05,
      "loss": 6.2571,
      "step": 58
    },
    {
      "epoch": 0.0026677337601707348,
      "grad_norm": 3.2798655033111572,
      "learning_rate": 6e-05,
      "loss": 6.2272,
      "step": 60
    },
    {
      "epoch": 0.002756658218843093,
      "grad_norm": 4.321415901184082,
      "learning_rate": 6.2e-05,
      "loss": 6.1702,
      "step": 62
    },
    {
      "epoch": 0.0028455826775154506,
      "grad_norm": 3.398996353149414,
      "learning_rate": 6.4e-05,
      "loss": 6.1203,
      "step": 64
    },
    {
      "epoch": 0.0029345071361878083,
      "grad_norm": 5.903692245483398,
      "learning_rate": 6.6e-05,
      "loss": 6.0742,
      "step": 66
    },
    {
      "epoch": 0.0030234315948601664,
      "grad_norm": 2.3560030460357666,
      "learning_rate": 6.800000000000001e-05,
      "loss": 6.0274,
      "step": 68
    },
    {
      "epoch": 0.003112356053532524,
      "grad_norm": 3.1397175788879395,
      "learning_rate": 7.000000000000001e-05,
      "loss": 5.9813,
      "step": 70
    },
    {
      "epoch": 0.003201280512204882,
      "grad_norm": 4.45662784576416,
      "learning_rate": 7.2e-05,
      "loss": 5.9562,
      "step": 72
    },
    {
      "epoch": 0.00329020497087724,
      "grad_norm": 5.479487895965576,
      "learning_rate": 7.4e-05,
      "loss": 5.9073,
      "step": 74
    },
    {
      "epoch": 0.0033791294295495975,
      "grad_norm": 4.0991363525390625,
      "learning_rate": 7.6e-05,
      "loss": 5.8533,
      "step": 76
    },
    {
      "epoch": 0.0034680538882219556,
      "grad_norm": 2.800793170928955,
      "learning_rate": 7.8e-05,
      "loss": 5.8174,
      "step": 78
    },
    {
      "epoch": 0.0035569783468943133,
      "grad_norm": 3.3895621299743652,
      "learning_rate": 8e-05,
      "loss": 5.7922,
      "step": 80
    },
    {
      "epoch": 0.003645902805566671,
      "grad_norm": 4.780695915222168,
      "learning_rate": 8.2e-05,
      "loss": 5.7626,
      "step": 82
    },
    {
      "epoch": 0.003734827264239029,
      "grad_norm": 5.153146743774414,
      "learning_rate": 8.400000000000001e-05,
      "loss": 5.726,
      "step": 84
    },
    {
      "epoch": 0.003823751722911387,
      "grad_norm": 6.493163108825684,
      "learning_rate": 8.599999999999999e-05,
      "loss": 5.7056,
      "step": 86
    },
    {
      "epoch": 0.003912676181583745,
      "grad_norm": 7.0106916427612305,
      "learning_rate": 8.8e-05,
      "loss": 5.6722,
      "step": 88
    },
    {
      "epoch": 0.004001600640256103,
      "grad_norm": 4.670555591583252,
      "learning_rate": 8.999999999999999e-05,
      "loss": 5.6304,
      "step": 90
    },
    {
      "epoch": 0.00409052509892846,
      "grad_norm": 5.772502899169922,
      "learning_rate": 9.2e-05,
      "loss": 5.634,
      "step": 92
    },
    {
      "epoch": 0.004179449557600818,
      "grad_norm": 4.904117107391357,
      "learning_rate": 9.400000000000001e-05,
      "loss": 5.5903,
      "step": 94
    },
    {
      "epoch": 0.004268374016273176,
      "grad_norm": 4.786822319030762,
      "learning_rate": 9.6e-05,
      "loss": 5.5855,
      "step": 96
    },
    {
      "epoch": 0.004357298474945534,
      "grad_norm": 6.82847261428833,
      "learning_rate": 9.800000000000001e-05,
      "loss": 5.5506,
      "step": 98
    },
    {
      "epoch": 0.004446222933617892,
      "grad_norm": 5.267941474914551,
      "learning_rate": 0.0001,
      "loss": 5.5599,
      "step": 100
    },
    {
      "epoch": 0.0045351473922902496,
      "grad_norm": 4.727493762969971,
      "learning_rate": 0.000102,
      "loss": 5.536,
      "step": 102
    },
    {
      "epoch": 0.004624071850962607,
      "grad_norm": 5.104313373565674,
      "learning_rate": 0.000104,
      "loss": 5.4914,
      "step": 104
    },
    {
      "epoch": 0.004712996309634965,
      "grad_norm": 5.250439167022705,
      "learning_rate": 0.000106,
      "loss": 5.4865,
      "step": 106
    },
    {
      "epoch": 0.004801920768307323,
      "grad_norm": 6.170891761779785,
      "learning_rate": 0.000108,
      "loss": 5.4896,
      "step": 108
    },
    {
      "epoch": 0.004890845226979681,
      "grad_norm": 6.01248025894165,
      "learning_rate": 0.00011,
      "loss": 5.4465,
      "step": 110
    },
    {
      "epoch": 0.004979769685652039,
      "grad_norm": 7.449190139770508,
      "learning_rate": 0.000112,
      "loss": 5.4305,
      "step": 112
    },
    {
      "epoch": 0.0050686941443243965,
      "grad_norm": 4.532025337219238,
      "learning_rate": 0.000114,
      "loss": 5.4133,
      "step": 114
    },
    {
      "epoch": 0.005157618602996754,
      "grad_norm": 6.034667491912842,
      "learning_rate": 0.00011600000000000001,
      "loss": 5.403,
      "step": 116
    },
    {
      "epoch": 0.005246543061669112,
      "grad_norm": 8.826780319213867,
      "learning_rate": 0.000118,
      "loss": 5.4014,
      "step": 118
    },
    {
      "epoch": 0.0053354675203414696,
      "grad_norm": 6.5053606033325195,
      "learning_rate": 0.00012,
      "loss": 5.3666,
      "step": 120
    },
    {
      "epoch": 0.005424391979013828,
      "grad_norm": 5.656647205352783,
      "learning_rate": 0.000122,
      "loss": 5.3378,
      "step": 122
    },
    {
      "epoch": 0.005513316437686186,
      "grad_norm": 4.237445831298828,
      "learning_rate": 0.000124,
      "loss": 5.331,
      "step": 124
    },
    {
      "epoch": 0.0056022408963585435,
      "grad_norm": 5.622873306274414,
      "learning_rate": 0.000126,
      "loss": 5.3327,
      "step": 126
    },
    {
      "epoch": 0.005691165355030901,
      "grad_norm": 5.530211925506592,
      "learning_rate": 0.000128,
      "loss": 5.3176,
      "step": 128
    },
    {
      "epoch": 0.005780089813703259,
      "grad_norm": 6.660238742828369,
      "learning_rate": 0.00013000000000000002,
      "loss": 5.3224,
      "step": 130
    },
    {
      "epoch": 0.0058690142723756165,
      "grad_norm": 5.829392910003662,
      "learning_rate": 0.000132,
      "loss": 5.2855,
      "step": 132
    },
    {
      "epoch": 0.005957938731047975,
      "grad_norm": 4.142682075500488,
      "learning_rate": 0.000134,
      "loss": 5.2567,
      "step": 134
    },
    {
      "epoch": 0.006046863189720333,
      "grad_norm": 4.414917469024658,
      "learning_rate": 0.00013600000000000003,
      "loss": 5.2591,
      "step": 136
    },
    {
      "epoch": 0.00613578764839269,
      "grad_norm": 5.137867450714111,
      "learning_rate": 0.00013800000000000002,
      "loss": 5.2214,
      "step": 138
    },
    {
      "epoch": 0.006224712107065048,
      "grad_norm": 4.432452201843262,
      "learning_rate": 0.00014000000000000001,
      "loss": 5.2201,
      "step": 140
    },
    {
      "epoch": 0.006313636565737406,
      "grad_norm": 3.24981951713562,
      "learning_rate": 0.00014199999999999998,
      "loss": 5.222,
      "step": 142
    },
    {
      "epoch": 0.006402561024409764,
      "grad_norm": 7.362234592437744,
      "learning_rate": 0.000144,
      "loss": 5.2342,
      "step": 144
    },
    {
      "epoch": 0.006491485483082122,
      "grad_norm": 6.990148067474365,
      "learning_rate": 0.000146,
      "loss": 5.1899,
      "step": 146
    },
    {
      "epoch": 0.00658040994175448,
      "grad_norm": 9.778069496154785,
      "learning_rate": 0.000148,
      "loss": 5.2028,
      "step": 148
    },
    {
      "epoch": 0.006669334400426837,
      "grad_norm": 7.326447486877441,
      "learning_rate": 0.00015,
      "loss": 5.1736,
      "step": 150
    },
    {
      "epoch": 0.006758258859099195,
      "grad_norm": 6.0494065284729,
      "learning_rate": 0.000152,
      "loss": 5.1704,
      "step": 152
    },
    {
      "epoch": 0.006847183317771553,
      "grad_norm": 7.345653057098389,
      "learning_rate": 0.000154,
      "loss": 5.1535,
      "step": 154
    },
    {
      "epoch": 0.006936107776443911,
      "grad_norm": 5.5805768966674805,
      "learning_rate": 0.000156,
      "loss": 5.1525,
      "step": 156
    },
    {
      "epoch": 0.007025032235116269,
      "grad_norm": 5.274758815765381,
      "learning_rate": 0.000158,
      "loss": 5.1331,
      "step": 158
    },
    {
      "epoch": 0.007113956693788627,
      "grad_norm": 4.2305169105529785,
      "learning_rate": 0.00016,
      "loss": 5.1053,
      "step": 160
    },
    {
      "epoch": 0.007202881152460984,
      "grad_norm": 6.961759567260742,
      "learning_rate": 0.000162,
      "loss": 5.1053,
      "step": 162
    },
    {
      "epoch": 0.007291805611133342,
      "grad_norm": 6.679699420928955,
      "learning_rate": 0.000164,
      "loss": 5.0955,
      "step": 164
    },
    {
      "epoch": 0.0073807300698057,
      "grad_norm": 6.162731647491455,
      "learning_rate": 0.00016600000000000002,
      "loss": 5.0938,
      "step": 166
    },
    {
      "epoch": 0.007469654528478058,
      "grad_norm": 7.818715572357178,
      "learning_rate": 0.00016800000000000002,
      "loss": 5.0897,
      "step": 168
    },
    {
      "epoch": 0.007558578987150416,
      "grad_norm": 5.024120807647705,
      "learning_rate": 0.00017,
      "loss": 5.0611,
      "step": 170
    },
    {
      "epoch": 0.007647503445822774,
      "grad_norm": 6.050165176391602,
      "learning_rate": 0.00017199999999999998,
      "loss": 5.053,
      "step": 172
    },
    {
      "epoch": 0.007736427904495131,
      "grad_norm": 4.531955242156982,
      "learning_rate": 0.000174,
      "loss": 5.021,
      "step": 174
    },
    {
      "epoch": 0.00782535236316749,
      "grad_norm": 4.957339286804199,
      "learning_rate": 0.000176,
      "loss": 5.003,
      "step": 176
    },
    {
      "epoch": 0.007914276821839848,
      "grad_norm": 5.476766586303711,
      "learning_rate": 0.000178,
      "loss": 5.0004,
      "step": 178
    },
    {
      "epoch": 0.008003201280512205,
      "grad_norm": 3.472067356109619,
      "learning_rate": 0.00017999999999999998,
      "loss": 4.9962,
      "step": 180
    },
    {
      "epoch": 0.008092125739184563,
      "grad_norm": 4.526614665985107,
      "learning_rate": 0.000182,
      "loss": 4.9533,
      "step": 182
    },
    {
      "epoch": 0.00818105019785692,
      "grad_norm": 4.315027236938477,
      "learning_rate": 0.000184,
      "loss": 4.9476,
      "step": 184
    },
    {
      "epoch": 0.008269974656529278,
      "grad_norm": 5.466373920440674,
      "learning_rate": 0.000186,
      "loss": 4.9424,
      "step": 186
    },
    {
      "epoch": 0.008358899115201636,
      "grad_norm": 4.121331691741943,
      "learning_rate": 0.00018800000000000002,
      "loss": 4.8852,
      "step": 188
    },
    {
      "epoch": 0.008447823573873994,
      "grad_norm": 5.156192302703857,
      "learning_rate": 0.00019,
      "loss": 4.8728,
      "step": 190
    },
    {
      "epoch": 0.008536748032546351,
      "grad_norm": 2.8727962970733643,
      "learning_rate": 0.000192,
      "loss": 4.8646,
      "step": 192
    },
    {
      "epoch": 0.008625672491218709,
      "grad_norm": 3.9635331630706787,
      "learning_rate": 0.000194,
      "loss": 4.8479,
      "step": 194
    },
    {
      "epoch": 0.008714596949891068,
      "grad_norm": 2.293034315109253,
      "learning_rate": 0.00019600000000000002,
      "loss": 4.8295,
      "step": 196
    },
    {
      "epoch": 0.008803521408563426,
      "grad_norm": 4.057344436645508,
      "learning_rate": 0.00019800000000000002,
      "loss": 4.8595,
      "step": 198
    },
    {
      "epoch": 0.008892445867235784,
      "grad_norm": 6.884203910827637,
      "learning_rate": 0.0002,
      "loss": 4.8364,
      "step": 200
    },
    {
      "epoch": 0.008981370325908141,
      "grad_norm": 3.6881725788116455,
      "learning_rate": 0.000202,
      "loss": 4.8192,
      "step": 202
    },
    {
      "epoch": 0.009070294784580499,
      "grad_norm": 3.199767827987671,
      "learning_rate": 0.000204,
      "loss": 4.7844,
      "step": 204
    },
    {
      "epoch": 0.009159219243252857,
      "grad_norm": 3.4823689460754395,
      "learning_rate": 0.000206,
      "loss": 4.7691,
      "step": 206
    },
    {
      "epoch": 0.009248143701925214,
      "grad_norm": 3.9715564250946045,
      "learning_rate": 0.000208,
      "loss": 4.7412,
      "step": 208
    },
    {
      "epoch": 0.009337068160597572,
      "grad_norm": 4.0734710693359375,
      "learning_rate": 0.00021,
      "loss": 4.7082,
      "step": 210
    },
    {
      "epoch": 0.00942599261926993,
      "grad_norm": 3.408423662185669,
      "learning_rate": 0.000212,
      "loss": 4.6992,
      "step": 212
    },
    {
      "epoch": 0.009514917077942288,
      "grad_norm": 3.8470771312713623,
      "learning_rate": 0.000214,
      "loss": 4.6654,
      "step": 214
    },
    {
      "epoch": 0.009603841536614645,
      "grad_norm": 3.259704113006592,
      "learning_rate": 0.000216,
      "loss": 4.6517,
      "step": 216
    },
    {
      "epoch": 0.009692765995287003,
      "grad_norm": 3.3092963695526123,
      "learning_rate": 0.000218,
      "loss": 4.6479,
      "step": 218
    },
    {
      "epoch": 0.009781690453959362,
      "grad_norm": 3.3664305210113525,
      "learning_rate": 0.00022,
      "loss": 4.6234,
      "step": 220
    },
    {
      "epoch": 0.00987061491263172,
      "grad_norm": 2.985428810119629,
      "learning_rate": 0.000222,
      "loss": 4.5937,
      "step": 222
    },
    {
      "epoch": 0.009959539371304078,
      "grad_norm": 4.405549049377441,
      "learning_rate": 0.000224,
      "loss": 4.5766,
      "step": 224
    },
    {
      "epoch": 0.010048463829976435,
      "grad_norm": 3.270683765411377,
      "learning_rate": 0.00022600000000000002,
      "loss": 4.5669,
      "step": 226
    },
    {
      "epoch": 0.010137388288648793,
      "grad_norm": 2.5708577632904053,
      "learning_rate": 0.000228,
      "loss": 4.52,
      "step": 228
    },
    {
      "epoch": 0.01022631274732115,
      "grad_norm": 4.556532859802246,
      "learning_rate": 0.00023,
      "loss": 4.5271,
      "step": 230
    },
    {
      "epoch": 0.010315237205993508,
      "grad_norm": 5.423923492431641,
      "learning_rate": 0.00023200000000000003,
      "loss": 4.5133,
      "step": 232
    },
    {
      "epoch": 0.010404161664665866,
      "grad_norm": 3.195823907852173,
      "learning_rate": 0.00023400000000000002,
      "loss": 4.4832,
      "step": 234
    },
    {
      "epoch": 0.010493086123338224,
      "grad_norm": 2.918025016784668,
      "learning_rate": 0.000236,
      "loss": 4.4876,
      "step": 236
    },
    {
      "epoch": 0.010582010582010581,
      "grad_norm": 3.2062084674835205,
      "learning_rate": 0.00023799999999999998,
      "loss": 4.4573,
      "step": 238
    },
    {
      "epoch": 0.010670935040682939,
      "grad_norm": 3.347294330596924,
      "learning_rate": 0.00024,
      "loss": 4.4288,
      "step": 240
    },
    {
      "epoch": 0.010759859499355299,
      "grad_norm": 2.824416160583496,
      "learning_rate": 0.000242,
      "loss": 4.3892,
      "step": 242
    },
    {
      "epoch": 0.010848783958027656,
      "grad_norm": 2.5121309757232666,
      "learning_rate": 0.000244,
      "loss": 4.3841,
      "step": 244
    },
    {
      "epoch": 0.010937708416700014,
      "grad_norm": 3.1845433712005615,
      "learning_rate": 0.000246,
      "loss": 4.3617,
      "step": 246
    },
    {
      "epoch": 0.011026632875372372,
      "grad_norm": 3.0984976291656494,
      "learning_rate": 0.000248,
      "loss": 4.3292,
      "step": 248
    },
    {
      "epoch": 0.01111555733404473,
      "grad_norm": 2.4962711334228516,
      "learning_rate": 0.00025,
      "loss": 4.292,
      "step": 250
    },
    {
      "epoch": 0.011204481792717087,
      "grad_norm": 3.3320326805114746,
      "learning_rate": 0.000252,
      "loss": 4.2964,
      "step": 252
    },
    {
      "epoch": 0.011293406251389445,
      "grad_norm": 3.5653789043426514,
      "learning_rate": 0.000254,
      "loss": 4.2572,
      "step": 254
    },
    {
      "epoch": 0.011382330710061802,
      "grad_norm": 2.946533679962158,
      "learning_rate": 0.000256,
      "loss": 4.2446,
      "step": 256
    },
    {
      "epoch": 0.01147125516873416,
      "grad_norm": 3.6157164573669434,
      "learning_rate": 0.00025800000000000004,
      "loss": 4.2731,
      "step": 258
    },
    {
      "epoch": 0.011560179627406518,
      "grad_norm": 3.1283135414123535,
      "learning_rate": 0.00026000000000000003,
      "loss": 4.2333,
      "step": 260
    },
    {
      "epoch": 0.011649104086078875,
      "grad_norm": 3.6416194438934326,
      "learning_rate": 0.000262,
      "loss": 4.1951,
      "step": 262
    },
    {
      "epoch": 0.011738028544751233,
      "grad_norm": 2.4387340545654297,
      "learning_rate": 0.000264,
      "loss": 4.1772,
      "step": 264
    },
    {
      "epoch": 0.011826953003423592,
      "grad_norm": 2.776474714279175,
      "learning_rate": 0.000266,
      "loss": 4.1483,
      "step": 266
    },
    {
      "epoch": 0.01191587746209595,
      "grad_norm": 3.2721757888793945,
      "learning_rate": 0.000268,
      "loss": 4.1457,
      "step": 268
    },
    {
      "epoch": 0.012004801920768308,
      "grad_norm": 3.0057179927825928,
      "learning_rate": 0.00027,
      "loss": 4.0978,
      "step": 270
    },
    {
      "epoch": 0.012093726379440665,
      "grad_norm": 3.400449514389038,
      "learning_rate": 0.00027200000000000005,
      "loss": 4.0854,
      "step": 272
    },
    {
      "epoch": 0.012182650838113023,
      "grad_norm": 2.8374624252319336,
      "learning_rate": 0.00027400000000000005,
      "loss": 4.0779,
      "step": 274
    },
    {
      "epoch": 0.01227157529678538,
      "grad_norm": 3.7050421237945557,
      "learning_rate": 0.00027600000000000004,
      "loss": 4.0588,
      "step": 276
    },
    {
      "epoch": 0.012360499755457739,
      "grad_norm": 3.274221181869507,
      "learning_rate": 0.00027800000000000004,
      "loss": 4.0334,
      "step": 278
    },
    {
      "epoch": 0.012449424214130096,
      "grad_norm": 2.5664162635803223,
      "learning_rate": 0.00028000000000000003,
      "loss": 4.0303,
      "step": 280
    },
    {
      "epoch": 0.012538348672802454,
      "grad_norm": 2.751819133758545,
      "learning_rate": 0.00028199999999999997,
      "loss": 3.9866,
      "step": 282
    },
    {
      "epoch": 0.012627273131474812,
      "grad_norm": 2.635796546936035,
      "learning_rate": 0.00028399999999999996,
      "loss": 3.967,
      "step": 284
    },
    {
      "epoch": 0.01271619759014717,
      "grad_norm": 2.1804494857788086,
      "learning_rate": 0.00028599999999999996,
      "loss": 3.9361,
      "step": 286
    },
    {
      "epoch": 0.012805122048819529,
      "grad_norm": 2.3362250328063965,
      "learning_rate": 0.000288,
      "loss": 3.9309,
      "step": 288
    },
    {
      "epoch": 0.012894046507491886,
      "grad_norm": 2.179434061050415,
      "learning_rate": 0.00029,
      "loss": 3.9056,
      "step": 290
    },
    {
      "epoch": 0.012982970966164244,
      "grad_norm": 2.6744179725646973,
      "learning_rate": 0.000292,
      "loss": 3.9158,
      "step": 292
    },
    {
      "epoch": 0.013071895424836602,
      "grad_norm": 2.8267312049865723,
      "learning_rate": 0.000294,
      "loss": 3.9134,
      "step": 294
    },
    {
      "epoch": 0.01316081988350896,
      "grad_norm": 2.420217990875244,
      "learning_rate": 0.000296,
      "loss": 3.8633,
      "step": 296
    },
    {
      "epoch": 0.013249744342181317,
      "grad_norm": 3.261594533920288,
      "learning_rate": 0.000298,
      "loss": 3.8427,
      "step": 298
    },
    {
      "epoch": 0.013338668800853675,
      "grad_norm": 2.8842456340789795,
      "learning_rate": 0.0003,
      "loss": 3.8155,
      "step": 300
    },
    {
      "epoch": 0.013427593259526032,
      "grad_norm": 2.741398334503174,
      "learning_rate": 0.000302,
      "loss": 3.7851,
      "step": 302
    },
    {
      "epoch": 0.01351651771819839,
      "grad_norm": 2.377107620239258,
      "learning_rate": 0.000304,
      "loss": 3.7656,
      "step": 304
    },
    {
      "epoch": 0.013605442176870748,
      "grad_norm": 3.3047285079956055,
      "learning_rate": 0.000306,
      "loss": 3.7708,
      "step": 306
    },
    {
      "epoch": 0.013694366635543105,
      "grad_norm": 3.042447090148926,
      "learning_rate": 0.000308,
      "loss": 3.7339,
      "step": 308
    },
    {
      "epoch": 0.013783291094215463,
      "grad_norm": 2.671283721923828,
      "learning_rate": 0.00031,
      "loss": 3.7213,
      "step": 310
    },
    {
      "epoch": 0.013872215552887823,
      "grad_norm": 3.2665302753448486,
      "learning_rate": 0.000312,
      "loss": 3.7418,
      "step": 312
    },
    {
      "epoch": 0.01396114001156018,
      "grad_norm": 2.6072757244110107,
      "learning_rate": 0.000314,
      "loss": 3.7062,
      "step": 314
    },
    {
      "epoch": 0.014050064470232538,
      "grad_norm": 2.4585163593292236,
      "learning_rate": 0.000316,
      "loss": 3.675,
      "step": 316
    },
    {
      "epoch": 0.014138988928904896,
      "grad_norm": 2.604253053665161,
      "learning_rate": 0.00031800000000000003,
      "loss": 3.6659,
      "step": 318
    },
    {
      "epoch": 0.014227913387577253,
      "grad_norm": 2.0046472549438477,
      "learning_rate": 0.00032,
      "loss": 3.6397,
      "step": 320
    },
    {
      "epoch": 0.014316837846249611,
      "grad_norm": 2.0930159091949463,
      "learning_rate": 0.000322,
      "loss": 3.6476,
      "step": 322
    },
    {
      "epoch": 0.014405762304921969,
      "grad_norm": 2.7829599380493164,
      "learning_rate": 0.000324,
      "loss": 3.6397,
      "step": 324
    },
    {
      "epoch": 0.014494686763594326,
      "grad_norm": 2.372708559036255,
      "learning_rate": 0.000326,
      "loss": 3.6136,
      "step": 326
    },
    {
      "epoch": 0.014583611222266684,
      "grad_norm": 2.2821056842803955,
      "learning_rate": 0.000328,
      "loss": 3.5951,
      "step": 328
    },
    {
      "epoch": 0.014672535680939042,
      "grad_norm": 2.620718002319336,
      "learning_rate": 0.00033,
      "loss": 3.5921,
      "step": 330
    },
    {
      "epoch": 0.0147614601396114,
      "grad_norm": 2.5476973056793213,
      "learning_rate": 0.00033200000000000005,
      "loss": 3.5755,
      "step": 332
    },
    {
      "epoch": 0.014850384598283759,
      "grad_norm": 2.17576003074646,
      "learning_rate": 0.00033400000000000004,
      "loss": 3.5587,
      "step": 334
    },
    {
      "epoch": 0.014939309056956116,
      "grad_norm": 2.368635416030884,
      "learning_rate": 0.00033600000000000004,
      "loss": 3.5409,
      "step": 336
    },
    {
      "epoch": 0.015028233515628474,
      "grad_norm": 3.101987361907959,
      "learning_rate": 0.00033800000000000003,
      "loss": 3.5305,
      "step": 338
    },
    {
      "epoch": 0.015117157974300832,
      "grad_norm": 2.805263042449951,
      "learning_rate": 0.00034,
      "loss": 3.5345,
      "step": 340
    },
    {
      "epoch": 0.01520608243297319,
      "grad_norm": 2.2070181369781494,
      "learning_rate": 0.000342,
      "loss": 3.5125,
      "step": 342
    },
    {
      "epoch": 0.015295006891645547,
      "grad_norm": 2.235079765319824,
      "learning_rate": 0.00034399999999999996,
      "loss": 3.4963,
      "step": 344
    },
    {
      "epoch": 0.015383931350317905,
      "grad_norm": 2.6045584678649902,
      "learning_rate": 0.000346,
      "loss": 3.5081,
      "step": 346
    },
    {
      "epoch": 0.015472855808990263,
      "grad_norm": 2.463226318359375,
      "learning_rate": 0.000348,
      "loss": 3.4777,
      "step": 348
    },
    {
      "epoch": 0.01556178026766262,
      "grad_norm": 2.3299500942230225,
      "learning_rate": 0.00035,
      "loss": 3.4477,
      "step": 350
    },
    {
      "epoch": 0.01565070472633498,
      "grad_norm": 1.9035766124725342,
      "learning_rate": 0.000352,
      "loss": 3.4405,
      "step": 352
    },
    {
      "epoch": 0.015739629185007337,
      "grad_norm": 2.3295583724975586,
      "learning_rate": 0.000354,
      "loss": 3.4328,
      "step": 354
    },
    {
      "epoch": 0.015828553643679695,
      "grad_norm": 2.2347898483276367,
      "learning_rate": 0.000356,
      "loss": 3.4162,
      "step": 356
    },
    {
      "epoch": 0.015917478102352053,
      "grad_norm": 2.777460813522339,
      "learning_rate": 0.000358,
      "loss": 3.4241,
      "step": 358
    },
    {
      "epoch": 0.01600640256102441,
      "grad_norm": 1.8630869388580322,
      "learning_rate": 0.00035999999999999997,
      "loss": 3.408,
      "step": 360
    },
    {
      "epoch": 0.016095327019696768,
      "grad_norm": 2.6221816539764404,
      "learning_rate": 0.000362,
      "loss": 3.4151,
      "step": 362
    },
    {
      "epoch": 0.016184251478369126,
      "grad_norm": 2.0977978706359863,
      "learning_rate": 0.000364,
      "loss": 3.3877,
      "step": 364
    },
    {
      "epoch": 0.016273175937041483,
      "grad_norm": 1.9285433292388916,
      "learning_rate": 0.000366,
      "loss": 3.3893,
      "step": 366
    },
    {
      "epoch": 0.01636210039571384,
      "grad_norm": 1.8546204566955566,
      "learning_rate": 0.000368,
      "loss": 3.3685,
      "step": 368
    },
    {
      "epoch": 0.0164510248543862,
      "grad_norm": 1.7032506465911865,
      "learning_rate": 0.00037,
      "loss": 3.3536,
      "step": 370
    },
    {
      "epoch": 0.016539949313058556,
      "grad_norm": 1.8106862306594849,
      "learning_rate": 0.000372,
      "loss": 3.3425,
      "step": 372
    },
    {
      "epoch": 0.016628873771730914,
      "grad_norm": 1.9233118295669556,
      "learning_rate": 0.000374,
      "loss": 3.3492,
      "step": 374
    },
    {
      "epoch": 0.016717798230403272,
      "grad_norm": 1.8728731870651245,
      "learning_rate": 0.00037600000000000003,
      "loss": 3.3253,
      "step": 376
    },
    {
      "epoch": 0.01680672268907563,
      "grad_norm": 2.1671135425567627,
      "learning_rate": 0.000378,
      "loss": 3.3353,
      "step": 378
    },
    {
      "epoch": 0.016895647147747987,
      "grad_norm": 2.369257926940918,
      "learning_rate": 0.00038,
      "loss": 3.303,
      "step": 380
    },
    {
      "epoch": 0.016984571606420345,
      "grad_norm": 2.2066473960876465,
      "learning_rate": 0.000382,
      "loss": 3.2988,
      "step": 382
    },
    {
      "epoch": 0.017073496065092703,
      "grad_norm": 2.1249470710754395,
      "learning_rate": 0.000384,
      "loss": 3.2837,
      "step": 384
    },
    {
      "epoch": 0.01716242052376506,
      "grad_norm": 2.062621831893921,
      "learning_rate": 0.000386,
      "loss": 3.269,
      "step": 386
    },
    {
      "epoch": 0.017251344982437418,
      "grad_norm": 1.8696043491363525,
      "learning_rate": 0.000388,
      "loss": 3.2692,
      "step": 388
    },
    {
      "epoch": 0.017340269441109776,
      "grad_norm": 1.7806510925292969,
      "learning_rate": 0.00039000000000000005,
      "loss": 3.2694,
      "step": 390
    },
    {
      "epoch": 0.017429193899782137,
      "grad_norm": 1.5019872188568115,
      "learning_rate": 0.00039200000000000004,
      "loss": 3.2581,
      "step": 392
    },
    {
      "epoch": 0.017518118358454494,
      "grad_norm": 1.6843113899230957,
      "learning_rate": 0.00039400000000000004,
      "loss": 3.248,
      "step": 394
    },
    {
      "epoch": 0.017607042817126852,
      "grad_norm": 1.682606816291809,
      "learning_rate": 0.00039600000000000003,
      "loss": 3.2564,
      "step": 396
    },
    {
      "epoch": 0.01769596727579921,
      "grad_norm": 2.0928115844726562,
      "learning_rate": 0.000398,
      "loss": 3.2256,
      "step": 398
    },
    {
      "epoch": 0.017784891734471567,
      "grad_norm": 1.9646433591842651,
      "learning_rate": 0.0004,
      "loss": 3.2244,
      "step": 400
    },
    {
      "epoch": 0.017873816193143925,
      "grad_norm": 1.8194555044174194,
      "learning_rate": 0.000402,
      "loss": 3.1879,
      "step": 402
    },
    {
      "epoch": 0.017962740651816283,
      "grad_norm": 1.7736834287643433,
      "learning_rate": 0.000404,
      "loss": 3.1818,
      "step": 404
    },
    {
      "epoch": 0.01805166511048864,
      "grad_norm": 1.6785839796066284,
      "learning_rate": 0.00040600000000000006,
      "loss": 3.1718,
      "step": 406
    },
    {
      "epoch": 0.018140589569160998,
      "grad_norm": 1.5538625717163086,
      "learning_rate": 0.000408,
      "loss": 3.1648,
      "step": 408
    },
    {
      "epoch": 0.018229514027833356,
      "grad_norm": 1.4424693584442139,
      "learning_rate": 0.00041,
      "loss": 3.1574,
      "step": 410
    },
    {
      "epoch": 0.018318438486505714,
      "grad_norm": 1.8182153701782227,
      "learning_rate": 0.000412,
      "loss": 3.1613,
      "step": 412
    },
    {
      "epoch": 0.01840736294517807,
      "grad_norm": 1.6871416568756104,
      "learning_rate": 0.000414,
      "loss": 3.1314,
      "step": 414
    },
    {
      "epoch": 0.01849628740385043,
      "grad_norm": 1.3707678318023682,
      "learning_rate": 0.000416,
      "loss": 3.1252,
      "step": 416
    },
    {
      "epoch": 0.018585211862522787,
      "grad_norm": 1.730461597442627,
      "learning_rate": 0.00041799999999999997,
      "loss": 3.1065,
      "step": 418
    },
    {
      "epoch": 0.018674136321195144,
      "grad_norm": 1.8215389251708984,
      "learning_rate": 0.00042,
      "loss": 3.1239,
      "step": 420
    },
    {
      "epoch": 0.018763060779867502,
      "grad_norm": 1.7602427005767822,
      "learning_rate": 0.000422,
      "loss": 3.108,
      "step": 422
    },
    {
      "epoch": 0.01885198523853986,
      "grad_norm": 1.5596345663070679,
      "learning_rate": 0.000424,
      "loss": 3.0993,
      "step": 424
    },
    {
      "epoch": 0.018940909697212217,
      "grad_norm": 1.3841055631637573,
      "learning_rate": 0.000426,
      "loss": 3.0894,
      "step": 426
    },
    {
      "epoch": 0.019029834155884575,
      "grad_norm": 1.6600890159606934,
      "learning_rate": 0.000428,
      "loss": 3.1196,
      "step": 428
    },
    {
      "epoch": 0.019118758614556933,
      "grad_norm": 1.7408195734024048,
      "learning_rate": 0.00043,
      "loss": 3.08,
      "step": 430
    },
    {
      "epoch": 0.01920768307322929,
      "grad_norm": 1.5704591274261475,
      "learning_rate": 0.000432,
      "loss": 3.0946,
      "step": 432
    },
    {
      "epoch": 0.019296607531901648,
      "grad_norm": 1.6484616994857788,
      "learning_rate": 0.00043400000000000003,
      "loss": 3.0796,
      "step": 434
    },
    {
      "epoch": 0.019385531990574006,
      "grad_norm": 1.5855478048324585,
      "learning_rate": 0.000436,
      "loss": 3.0624,
      "step": 436
    },
    {
      "epoch": 0.019474456449246367,
      "grad_norm": 1.427921175956726,
      "learning_rate": 0.000438,
      "loss": 3.0426,
      "step": 438
    },
    {
      "epoch": 0.019563380907918725,
      "grad_norm": 1.3351646661758423,
      "learning_rate": 0.00044,
      "loss": 3.0436,
      "step": 440
    },
    {
      "epoch": 0.019652305366591082,
      "grad_norm": 1.4644699096679688,
      "learning_rate": 0.000442,
      "loss": 3.0203,
      "step": 442
    },
    {
      "epoch": 0.01974122982526344,
      "grad_norm": 1.3958910703659058,
      "learning_rate": 0.000444,
      "loss": 3.0173,
      "step": 444
    },
    {
      "epoch": 0.019830154283935798,
      "grad_norm": 1.179125428199768,
      "learning_rate": 0.000446,
      "loss": 3.0025,
      "step": 446
    },
    {
      "epoch": 0.019919078742608155,
      "grad_norm": 1.343889594078064,
      "learning_rate": 0.000448,
      "loss": 2.9919,
      "step": 448
    },
    {
      "epoch": 0.020008003201280513,
      "grad_norm": 1.136616587638855,
      "learning_rate": 0.00045000000000000004,
      "loss": 2.9778,
      "step": 450
    },
    {
      "epoch": 0.02009692765995287,
      "grad_norm": 1.4144681692123413,
      "learning_rate": 0.00045200000000000004,
      "loss": 2.9919,
      "step": 452
    },
    {
      "epoch": 0.02018585211862523,
      "grad_norm": 1.300412654876709,
      "learning_rate": 0.00045400000000000003,
      "loss": 2.9753,
      "step": 454
    },
    {
      "epoch": 0.020274776577297586,
      "grad_norm": 1.2180825471878052,
      "learning_rate": 0.000456,
      "loss": 2.9706,
      "step": 456
    },
    {
      "epoch": 0.020363701035969944,
      "grad_norm": 1.2152589559555054,
      "learning_rate": 0.000458,
      "loss": 2.9791,
      "step": 458
    },
    {
      "epoch": 0.0204526254946423,
      "grad_norm": 1.3307461738586426,
      "learning_rate": 0.00046,
      "loss": 2.9722,
      "step": 460
    },
    {
      "epoch": 0.02054154995331466,
      "grad_norm": 1.3811430931091309,
      "learning_rate": 0.000462,
      "loss": 2.9623,
      "step": 462
    },
    {
      "epoch": 0.020630474411987017,
      "grad_norm": 1.6451685428619385,
      "learning_rate": 0.00046400000000000006,
      "loss": 2.9489,
      "step": 464
    },
    {
      "epoch": 0.020719398870659374,
      "grad_norm": 1.451253890991211,
      "learning_rate": 0.00046600000000000005,
      "loss": 2.9488,
      "step": 466
    },
    {
      "epoch": 0.020808323329331732,
      "grad_norm": 1.1156989336013794,
      "learning_rate": 0.00046800000000000005,
      "loss": 2.9374,
      "step": 468
    },
    {
      "epoch": 0.02089724778800409,
      "grad_norm": 1.3331708908081055,
      "learning_rate": 0.00047,
      "loss": 2.9271,
      "step": 470
    },
    {
      "epoch": 0.020986172246676447,
      "grad_norm": 1.2767914533615112,
      "learning_rate": 0.000472,
      "loss": 2.9176,
      "step": 472
    },
    {
      "epoch": 0.021075096705348805,
      "grad_norm": 1.5849168300628662,
      "learning_rate": 0.000474,
      "loss": 2.916,
      "step": 474
    },
    {
      "epoch": 0.021164021164021163,
      "grad_norm": 1.1405562162399292,
      "learning_rate": 0.00047599999999999997,
      "loss": 2.9101,
      "step": 476
    },
    {
      "epoch": 0.02125294562269352,
      "grad_norm": 1.2228617668151855,
      "learning_rate": 0.00047799999999999996,
      "loss": 2.9069,
      "step": 478
    },
    {
      "epoch": 0.021341870081365878,
      "grad_norm": 1.2419967651367188,
      "learning_rate": 0.00048,
      "loss": 2.8927,
      "step": 480
    },
    {
      "epoch": 0.021430794540038236,
      "grad_norm": 1.4789477586746216,
      "learning_rate": 0.000482,
      "loss": 2.8991,
      "step": 482
    },
    {
      "epoch": 0.021519718998710597,
      "grad_norm": 1.4335638284683228,
      "learning_rate": 0.000484,
      "loss": 2.8878,
      "step": 484
    },
    {
      "epoch": 0.021608643457382955,
      "grad_norm": 1.608608603477478,
      "learning_rate": 0.000486,
      "loss": 2.8878,
      "step": 486
    },
    {
      "epoch": 0.021697567916055312,
      "grad_norm": 1.1100876331329346,
      "learning_rate": 0.000488,
      "loss": 2.8723,
      "step": 488
    },
    {
      "epoch": 0.02178649237472767,
      "grad_norm": 1.4269640445709229,
      "learning_rate": 0.00049,
      "loss": 2.8704,
      "step": 490
    },
    {
      "epoch": 0.021875416833400028,
      "grad_norm": 1.4346448183059692,
      "learning_rate": 0.000492,
      "loss": 2.8758,
      "step": 492
    },
    {
      "epoch": 0.021964341292072385,
      "grad_norm": 1.3498926162719727,
      "learning_rate": 0.000494,
      "loss": 2.8468,
      "step": 494
    },
    {
      "epoch": 0.022053265750744743,
      "grad_norm": 1.1565560102462769,
      "learning_rate": 0.000496,
      "loss": 2.8485,
      "step": 496
    },
    {
      "epoch": 0.0221421902094171,
      "grad_norm": 1.066014289855957,
      "learning_rate": 0.000498,
      "loss": 2.8366,
      "step": 498
    },
    {
      "epoch": 0.02223111466808946,
      "grad_norm": 0.9729352593421936,
      "learning_rate": 0.0005,
      "loss": 2.8219,
      "step": 500
    },
    {
      "epoch": 0.02223111466808946,
      "eval_loss": 2.707019329071045,
      "eval_runtime": 12.411,
      "eval_samples_per_second": 556.764,
      "eval_steps_per_second": 69.616,
      "step": 500
    },
    {
      "epoch": 0.022320039126761816,
      "grad_norm": 1.277807354927063,
      "learning_rate": 0.0005020000000000001,
      "loss": 2.8322,
      "step": 502
    },
    {
      "epoch": 0.022408963585434174,
      "grad_norm": 1.2700484991073608,
      "learning_rate": 0.000504,
      "loss": 2.8281,
      "step": 504
    },
    {
      "epoch": 0.02249788804410653,
      "grad_norm": 1.3185160160064697,
      "learning_rate": 0.000506,
      "loss": 2.8253,
      "step": 506
    },
    {
      "epoch": 0.02258681250277889,
      "grad_norm": 1.1390444040298462,
      "learning_rate": 0.000508,
      "loss": 2.8258,
      "step": 508
    },
    {
      "epoch": 0.022675736961451247,
      "grad_norm": 1.3068996667861938,
      "learning_rate": 0.00051,
      "loss": 2.7982,
      "step": 510
    },
    {
      "epoch": 0.022764661420123605,
      "grad_norm": 1.2303944826126099,
      "learning_rate": 0.000512,
      "loss": 2.7974,
      "step": 512
    },
    {
      "epoch": 0.022853585878795962,
      "grad_norm": 1.3849083185195923,
      "learning_rate": 0.000514,
      "loss": 2.7972,
      "step": 514
    },
    {
      "epoch": 0.02294251033746832,
      "grad_norm": 0.9999197125434875,
      "learning_rate": 0.0005160000000000001,
      "loss": 2.7885,
      "step": 516
    },
    {
      "epoch": 0.023031434796140678,
      "grad_norm": 1.1116942167282104,
      "learning_rate": 0.000518,
      "loss": 2.7936,
      "step": 518
    },
    {
      "epoch": 0.023120359254813035,
      "grad_norm": 1.0687438249588013,
      "learning_rate": 0.0005200000000000001,
      "loss": 2.7732,
      "step": 520
    },
    {
      "epoch": 0.023209283713485393,
      "grad_norm": 1.048213243484497,
      "learning_rate": 0.000522,
      "loss": 2.7626,
      "step": 522
    },
    {
      "epoch": 0.02329820817215775,
      "grad_norm": 1.0101920366287231,
      "learning_rate": 0.000524,
      "loss": 2.7557,
      "step": 524
    },
    {
      "epoch": 0.02338713263083011,
      "grad_norm": 1.3061760663986206,
      "learning_rate": 0.000526,
      "loss": 2.7727,
      "step": 526
    },
    {
      "epoch": 0.023476057089502466,
      "grad_norm": 1.0070077180862427,
      "learning_rate": 0.000528,
      "loss": 2.7614,
      "step": 528
    },
    {
      "epoch": 0.023564981548174827,
      "grad_norm": 0.9444275498390198,
      "learning_rate": 0.0005300000000000001,
      "loss": 2.7597,
      "step": 530
    },
    {
      "epoch": 0.023653906006847185,
      "grad_norm": 1.201491355895996,
      "learning_rate": 0.000532,
      "loss": 2.751,
      "step": 532
    },
    {
      "epoch": 0.023742830465519543,
      "grad_norm": 1.0685005187988281,
      "learning_rate": 0.0005340000000000001,
      "loss": 2.7524,
      "step": 534
    },
    {
      "epoch": 0.0238317549241919,
      "grad_norm": 1.089680790901184,
      "learning_rate": 0.000536,
      "loss": 2.7506,
      "step": 536
    },
    {
      "epoch": 0.023920679382864258,
      "grad_norm": 1.0986379384994507,
      "learning_rate": 0.0005380000000000001,
      "loss": 2.7398,
      "step": 538
    },
    {
      "epoch": 0.024009603841536616,
      "grad_norm": 1.0519707202911377,
      "learning_rate": 0.00054,
      "loss": 2.7296,
      "step": 540
    },
    {
      "epoch": 0.024098528300208973,
      "grad_norm": 1.0870462656021118,
      "learning_rate": 0.0005420000000000001,
      "loss": 2.7153,
      "step": 542
    },
    {
      "epoch": 0.02418745275888133,
      "grad_norm": 0.8601863980293274,
      "learning_rate": 0.0005440000000000001,
      "loss": 2.7028,
      "step": 544
    },
    {
      "epoch": 0.02427637721755369,
      "grad_norm": 0.96519935131073,
      "learning_rate": 0.000546,
      "loss": 2.6951,
      "step": 546
    },
    {
      "epoch": 0.024365301676226046,
      "grad_norm": 1.0897146463394165,
      "learning_rate": 0.0005480000000000001,
      "loss": 2.6911,
      "step": 548
    },
    {
      "epoch": 0.024454226134898404,
      "grad_norm": 0.9683467149734497,
      "learning_rate": 0.00055,
      "loss": 2.6904,
      "step": 550
    },
    {
      "epoch": 0.02454315059357076,
      "grad_norm": 1.0787107944488525,
      "learning_rate": 0.0005520000000000001,
      "loss": 2.69,
      "step": 552
    },
    {
      "epoch": 0.02463207505224312,
      "grad_norm": 1.0309861898422241,
      "learning_rate": 0.000554,
      "loss": 2.691,
      "step": 554
    },
    {
      "epoch": 0.024720999510915477,
      "grad_norm": 0.9567499756813049,
      "learning_rate": 0.0005560000000000001,
      "loss": 2.6843,
      "step": 556
    },
    {
      "epoch": 0.024809923969587835,
      "grad_norm": 0.8283863663673401,
      "learning_rate": 0.000558,
      "loss": 2.6758,
      "step": 558
    },
    {
      "epoch": 0.024898848428260192,
      "grad_norm": 0.8621578216552734,
      "learning_rate": 0.0005600000000000001,
      "loss": 2.6639,
      "step": 560
    },
    {
      "epoch": 0.02498777288693255,
      "grad_norm": 0.9517662525177002,
      "learning_rate": 0.0005620000000000001,
      "loss": 2.6644,
      "step": 562
    },
    {
      "epoch": 0.025076697345604908,
      "grad_norm": 0.9091687202453613,
      "learning_rate": 0.0005639999999999999,
      "loss": 2.6538,
      "step": 564
    },
    {
      "epoch": 0.025165621804277265,
      "grad_norm": 0.8210693597793579,
      "learning_rate": 0.000566,
      "loss": 2.6505,
      "step": 566
    },
    {
      "epoch": 0.025254546262949623,
      "grad_norm": 0.906134843826294,
      "learning_rate": 0.0005679999999999999,
      "loss": 2.6422,
      "step": 568
    },
    {
      "epoch": 0.02534347072162198,
      "grad_norm": 0.7618714570999146,
      "learning_rate": 0.00057,
      "loss": 2.6204,
      "step": 570
    },
    {
      "epoch": 0.02543239518029434,
      "grad_norm": 0.7566989064216614,
      "learning_rate": 0.0005719999999999999,
      "loss": 2.6204,
      "step": 572
    },
    {
      "epoch": 0.025521319638966696,
      "grad_norm": 0.8227992057800293,
      "learning_rate": 0.000574,
      "loss": 2.6237,
      "step": 574
    },
    {
      "epoch": 0.025610244097639057,
      "grad_norm": 0.7188429236412048,
      "learning_rate": 0.000576,
      "loss": 2.6089,
      "step": 576
    },
    {
      "epoch": 0.025699168556311415,
      "grad_norm": 0.7563546895980835,
      "learning_rate": 0.000578,
      "loss": 2.594,
      "step": 578
    },
    {
      "epoch": 0.025788093014983773,
      "grad_norm": 0.7829350233078003,
      "learning_rate": 0.00058,
      "loss": 2.6021,
      "step": 580
    },
    {
      "epoch": 0.02587701747365613,
      "grad_norm": 0.7614354491233826,
      "learning_rate": 0.0005819999999999999,
      "loss": 2.6012,
      "step": 582
    },
    {
      "epoch": 0.025965941932328488,
      "grad_norm": 0.9515312314033508,
      "learning_rate": 0.000584,
      "loss": 2.5882,
      "step": 584
    },
    {
      "epoch": 0.026054866391000846,
      "grad_norm": 0.8379482626914978,
      "learning_rate": 0.0005859999999999999,
      "loss": 2.6015,
      "step": 586
    },
    {
      "epoch": 0.026143790849673203,
      "grad_norm": 0.8818963170051575,
      "learning_rate": 0.000588,
      "loss": 2.5888,
      "step": 588
    },
    {
      "epoch": 0.02623271530834556,
      "grad_norm": 0.7838352918624878,
      "learning_rate": 0.00059,
      "loss": 2.5806,
      "step": 590
    },
    {
      "epoch": 0.02632163976701792,
      "grad_norm": 0.7371238470077515,
      "learning_rate": 0.000592,
      "loss": 2.5727,
      "step": 592
    },
    {
      "epoch": 0.026410564225690276,
      "grad_norm": 0.776374340057373,
      "learning_rate": 0.000594,
      "loss": 2.5669,
      "step": 594
    },
    {
      "epoch": 0.026499488684362634,
      "grad_norm": 0.8051691651344299,
      "learning_rate": 0.000596,
      "loss": 2.5712,
      "step": 596
    },
    {
      "epoch": 0.026588413143034992,
      "grad_norm": 0.8638246655464172,
      "learning_rate": 0.000598,
      "loss": 2.5693,
      "step": 598
    },
    {
      "epoch": 0.02667733760170735,
      "grad_norm": 0.7371112704277039,
      "learning_rate": 0.0006,
      "loss": 2.5555,
      "step": 600
    },
    {
      "epoch": 0.026766262060379707,
      "grad_norm": 0.7257919311523438,
      "learning_rate": 0.000602,
      "loss": 2.5526,
      "step": 602
    },
    {
      "epoch": 0.026855186519052065,
      "grad_norm": 0.6093848943710327,
      "learning_rate": 0.000604,
      "loss": 2.5433,
      "step": 604
    },
    {
      "epoch": 0.026944110977724423,
      "grad_norm": 0.8363651633262634,
      "learning_rate": 0.000606,
      "loss": 2.5358,
      "step": 606
    },
    {
      "epoch": 0.02703303543639678,
      "grad_norm": 0.9323018193244934,
      "learning_rate": 0.000608,
      "loss": 2.554,
      "step": 608
    },
    {
      "epoch": 0.027121959895069138,
      "grad_norm": 0.743377149105072,
      "learning_rate": 0.00061,
      "loss": 2.5392,
      "step": 610
    },
    {
      "epoch": 0.027210884353741496,
      "grad_norm": 0.6902635097503662,
      "learning_rate": 0.000612,
      "loss": 2.5257,
      "step": 612
    },
    {
      "epoch": 0.027299808812413853,
      "grad_norm": 0.6234912276268005,
      "learning_rate": 0.000614,
      "loss": 2.521,
      "step": 614
    },
    {
      "epoch": 0.02738873327108621,
      "grad_norm": 0.6421347260475159,
      "learning_rate": 0.000616,
      "loss": 2.5161,
      "step": 616
    },
    {
      "epoch": 0.02747765772975857,
      "grad_norm": 0.6125656962394714,
      "learning_rate": 0.0006180000000000001,
      "loss": 2.5081,
      "step": 618
    },
    {
      "epoch": 0.027566582188430926,
      "grad_norm": 0.6754232048988342,
      "learning_rate": 0.00062,
      "loss": 2.5047,
      "step": 620
    },
    {
      "epoch": 0.027655506647103287,
      "grad_norm": 0.6989823579788208,
      "learning_rate": 0.000622,
      "loss": 2.5019,
      "step": 622
    },
    {
      "epoch": 0.027744431105775645,
      "grad_norm": 0.572574257850647,
      "learning_rate": 0.000624,
      "loss": 2.4895,
      "step": 624
    },
    {
      "epoch": 0.027833355564448003,
      "grad_norm": 0.6032222509384155,
      "learning_rate": 0.000626,
      "loss": 2.4796,
      "step": 626
    },
    {
      "epoch": 0.02792228002312036,
      "grad_norm": 0.64015793800354,
      "learning_rate": 0.000628,
      "loss": 2.4823,
      "step": 628
    },
    {
      "epoch": 0.028011204481792718,
      "grad_norm": 1.0496829748153687,
      "learning_rate": 0.00063,
      "loss": 2.4876,
      "step": 630
    },
    {
      "epoch": 0.028100128940465076,
      "grad_norm": 0.7946370244026184,
      "learning_rate": 0.000632,
      "loss": 2.4895,
      "step": 632
    },
    {
      "epoch": 0.028189053399137434,
      "grad_norm": 0.7404975295066833,
      "learning_rate": 0.000634,
      "loss": 2.4849,
      "step": 634
    },
    {
      "epoch": 0.02827797785780979,
      "grad_norm": 0.7491384148597717,
      "learning_rate": 0.0006360000000000001,
      "loss": 2.4866,
      "step": 636
    },
    {
      "epoch": 0.02836690231648215,
      "grad_norm": 0.7001233696937561,
      "learning_rate": 0.000638,
      "loss": 2.4697,
      "step": 638
    },
    {
      "epoch": 0.028455826775154507,
      "grad_norm": 0.6039974689483643,
      "learning_rate": 0.00064,
      "loss": 2.4638,
      "step": 640
    },
    {
      "epoch": 0.028544751233826864,
      "grad_norm": 0.5607019662857056,
      "learning_rate": 0.000642,
      "loss": 2.459,
      "step": 642
    },
    {
      "epoch": 0.028633675692499222,
      "grad_norm": 0.7814773321151733,
      "learning_rate": 0.000644,
      "loss": 2.4615,
      "step": 644
    },
    {
      "epoch": 0.02872260015117158,
      "grad_norm": 0.8695835471153259,
      "learning_rate": 0.000646,
      "loss": 2.472,
      "step": 646
    },
    {
      "epoch": 0.028811524609843937,
      "grad_norm": 0.6280689239501953,
      "learning_rate": 0.000648,
      "loss": 2.4554,
      "step": 648
    },
    {
      "epoch": 0.028900449068516295,
      "grad_norm": 0.6447476148605347,
      "learning_rate": 0.0006500000000000001,
      "loss": 2.4426,
      "step": 650
    },
    {
      "epoch": 0.028989373527188653,
      "grad_norm": 0.7183859944343567,
      "learning_rate": 0.000652,
      "loss": 2.445,
      "step": 652
    },
    {
      "epoch": 0.02907829798586101,
      "grad_norm": 0.657177746295929,
      "learning_rate": 0.0006540000000000001,
      "loss": 2.4433,
      "step": 654
    },
    {
      "epoch": 0.029167222444533368,
      "grad_norm": 0.635399341583252,
      "learning_rate": 0.000656,
      "loss": 2.4327,
      "step": 656
    },
    {
      "epoch": 0.029256146903205726,
      "grad_norm": 0.519986629486084,
      "learning_rate": 0.0006580000000000001,
      "loss": 2.4328,
      "step": 658
    },
    {
      "epoch": 0.029345071361878083,
      "grad_norm": 0.6304401755332947,
      "learning_rate": 0.00066,
      "loss": 2.4328,
      "step": 660
    },
    {
      "epoch": 0.02943399582055044,
      "grad_norm": 0.6917545199394226,
      "learning_rate": 0.000662,
      "loss": 2.4259,
      "step": 662
    },
    {
      "epoch": 0.0295229202792228,
      "grad_norm": 0.844325840473175,
      "learning_rate": 0.0006640000000000001,
      "loss": 2.4414,
      "step": 664
    },
    {
      "epoch": 0.029611844737895156,
      "grad_norm": 0.8940752744674683,
      "learning_rate": 0.000666,
      "loss": 2.4353,
      "step": 666
    },
    {
      "epoch": 0.029700769196567518,
      "grad_norm": 0.8456497192382812,
      "learning_rate": 0.0006680000000000001,
      "loss": 2.4598,
      "step": 668
    },
    {
      "epoch": 0.029789693655239875,
      "grad_norm": 0.7692534327507019,
      "learning_rate": 0.00067,
      "loss": 2.4437,
      "step": 670
    },
    {
      "epoch": 0.029878618113912233,
      "grad_norm": 0.5847371816635132,
      "learning_rate": 0.0006720000000000001,
      "loss": 2.4399,
      "step": 672
    },
    {
      "epoch": 0.02996754257258459,
      "grad_norm": 0.505262017250061,
      "learning_rate": 0.000674,
      "loss": 2.4225,
      "step": 674
    },
    {
      "epoch": 0.03005646703125695,
      "grad_norm": 0.4932805001735687,
      "learning_rate": 0.0006760000000000001,
      "loss": 2.4107,
      "step": 676
    },
    {
      "epoch": 0.030145391489929306,
      "grad_norm": 0.4631360173225403,
      "learning_rate": 0.0006780000000000001,
      "loss": 2.4122,
      "step": 678
    },
    {
      "epoch": 0.030234315948601664,
      "grad_norm": 0.49897080659866333,
      "learning_rate": 0.00068,
      "loss": 2.4029,
      "step": 680
    },
    {
      "epoch": 0.03032324040727402,
      "grad_norm": 0.5904172658920288,
      "learning_rate": 0.0006820000000000001,
      "loss": 2.396,
      "step": 682
    },
    {
      "epoch": 0.03041216486594638,
      "grad_norm": 0.6271270513534546,
      "learning_rate": 0.000684,
      "loss": 2.3943,
      "step": 684
    },
    {
      "epoch": 0.030501089324618737,
      "grad_norm": 0.7415940761566162,
      "learning_rate": 0.0006860000000000001,
      "loss": 2.3821,
      "step": 686
    },
    {
      "epoch": 0.030590013783291094,
      "grad_norm": 0.9829635620117188,
      "learning_rate": 0.0006879999999999999,
      "loss": 2.4093,
      "step": 688
    },
    {
      "epoch": 0.030678938241963452,
      "grad_norm": 0.7612989544868469,
      "learning_rate": 0.00069,
      "loss": 2.4255,
      "step": 690
    },
    {
      "epoch": 0.03076786270063581,
      "grad_norm": 0.6028714776039124,
      "learning_rate": 0.000692,
      "loss": 2.4156,
      "step": 692
    },
    {
      "epoch": 0.030856787159308167,
      "grad_norm": 0.4892686903476715,
      "learning_rate": 0.000694,
      "loss": 2.3899,
      "step": 694
    },
    {
      "epoch": 0.030945711617980525,
      "grad_norm": 0.49779361486434937,
      "learning_rate": 0.000696,
      "loss": 2.388,
      "step": 696
    },
    {
      "epoch": 0.031034636076652883,
      "grad_norm": 0.43636444211006165,
      "learning_rate": 0.0006979999999999999,
      "loss": 2.3775,
      "step": 698
    },
    {
      "epoch": 0.03112356053532524,
      "grad_norm": 0.5327876806259155,
      "learning_rate": 0.0007,
      "loss": 2.3719,
      "step": 700
    },
    {
      "epoch": 0.031212484993997598,
      "grad_norm": 0.5242130160331726,
      "learning_rate": 0.0007019999999999999,
      "loss": 2.3703,
      "step": 702
    },
    {
      "epoch": 0.03130140945266996,
      "grad_norm": 0.523585319519043,
      "learning_rate": 0.000704,
      "loss": 2.3688,
      "step": 704
    },
    {
      "epoch": 0.031390333911342314,
      "grad_norm": 0.5895810127258301,
      "learning_rate": 0.0007059999999999999,
      "loss": 2.3672,
      "step": 706
    },
    {
      "epoch": 0.031479258370014675,
      "grad_norm": 0.7120672464370728,
      "learning_rate": 0.000708,
      "loss": 2.3507,
      "step": 708
    },
    {
      "epoch": 0.03156818282868703,
      "grad_norm": 0.8193224668502808,
      "learning_rate": 0.00071,
      "loss": 2.3585,
      "step": 710
    },
    {
      "epoch": 0.03165710728735939,
      "grad_norm": 0.7592872977256775,
      "learning_rate": 0.000712,
      "loss": 2.3881,
      "step": 712
    },
    {
      "epoch": 0.031746031746031744,
      "grad_norm": 0.6383097171783447,
      "learning_rate": 0.000714,
      "loss": 2.3597,
      "step": 714
    },
    {
      "epoch": 0.031834956204704105,
      "grad_norm": 0.5765336751937866,
      "learning_rate": 0.000716,
      "loss": 2.3569,
      "step": 716
    },
    {
      "epoch": 0.03192388066337646,
      "grad_norm": 0.44413337111473083,
      "learning_rate": 0.000718,
      "loss": 2.3484,
      "step": 718
    },
    {
      "epoch": 0.03201280512204882,
      "grad_norm": 0.42275282740592957,
      "learning_rate": 0.0007199999999999999,
      "loss": 2.3459,
      "step": 720
    },
    {
      "epoch": 0.032101729580721175,
      "grad_norm": 0.4223826229572296,
      "learning_rate": 0.000722,
      "loss": 2.3355,
      "step": 722
    },
    {
      "epoch": 0.032190654039393536,
      "grad_norm": 0.45038753747940063,
      "learning_rate": 0.000724,
      "loss": 2.3321,
      "step": 724
    },
    {
      "epoch": 0.03227957849806589,
      "grad_norm": 0.5487307906150818,
      "learning_rate": 0.000726,
      "loss": 2.3356,
      "step": 726
    },
    {
      "epoch": 0.03236850295673825,
      "grad_norm": 0.5726903676986694,
      "learning_rate": 0.000728,
      "loss": 2.3356,
      "step": 728
    },
    {
      "epoch": 0.032457427415410606,
      "grad_norm": 0.6729259490966797,
      "learning_rate": 0.00073,
      "loss": 2.3351,
      "step": 730
    },
    {
      "epoch": 0.03254635187408297,
      "grad_norm": 0.6266768574714661,
      "learning_rate": 0.000732,
      "loss": 2.323,
      "step": 732
    },
    {
      "epoch": 0.03263527633275532,
      "grad_norm": 0.5825262069702148,
      "learning_rate": 0.000734,
      "loss": 2.3358,
      "step": 734
    },
    {
      "epoch": 0.03272420079142768,
      "grad_norm": 0.5942145586013794,
      "learning_rate": 0.000736,
      "loss": 2.3235,
      "step": 736
    },
    {
      "epoch": 0.03281312525010004,
      "grad_norm": 0.5730122923851013,
      "learning_rate": 0.000738,
      "loss": 2.3229,
      "step": 738
    },
    {
      "epoch": 0.0329020497087724,
      "grad_norm": 0.6102321743965149,
      "learning_rate": 0.00074,
      "loss": 2.3301,
      "step": 740
    },
    {
      "epoch": 0.03299097416744476,
      "grad_norm": 0.4807248115539551,
      "learning_rate": 0.000742,
      "loss": 2.3169,
      "step": 742
    },
    {
      "epoch": 0.03307989862611711,
      "grad_norm": 0.4981859028339386,
      "learning_rate": 0.000744,
      "loss": 2.3238,
      "step": 744
    },
    {
      "epoch": 0.033168823084789474,
      "grad_norm": 0.5408335328102112,
      "learning_rate": 0.000746,
      "loss": 2.305,
      "step": 746
    },
    {
      "epoch": 0.03325774754346183,
      "grad_norm": 0.5518751740455627,
      "learning_rate": 0.000748,
      "loss": 2.313,
      "step": 748
    },
    {
      "epoch": 0.03334667200213419,
      "grad_norm": 0.5658276677131653,
      "learning_rate": 0.00075,
      "loss": 2.3078,
      "step": 750
    },
    {
      "epoch": 0.033435596460806544,
      "grad_norm": 0.481430321931839,
      "learning_rate": 0.0007520000000000001,
      "loss": 2.3019,
      "step": 752
    },
    {
      "epoch": 0.033524520919478905,
      "grad_norm": 0.5349968075752258,
      "learning_rate": 0.000754,
      "loss": 2.3053,
      "step": 754
    },
    {
      "epoch": 0.03361344537815126,
      "grad_norm": 0.5499293208122253,
      "learning_rate": 0.000756,
      "loss": 2.3026,
      "step": 756
    },
    {
      "epoch": 0.03370236983682362,
      "grad_norm": 0.6066279411315918,
      "learning_rate": 0.000758,
      "loss": 2.3015,
      "step": 758
    },
    {
      "epoch": 0.033791294295495974,
      "grad_norm": 0.5410886406898499,
      "learning_rate": 0.00076,
      "loss": 2.2962,
      "step": 760
    },
    {
      "epoch": 0.033880218754168336,
      "grad_norm": 0.5119853615760803,
      "learning_rate": 0.000762,
      "loss": 2.3011,
      "step": 762
    },
    {
      "epoch": 0.03396914321284069,
      "grad_norm": 0.5836531519889832,
      "learning_rate": 0.000764,
      "loss": 2.2847,
      "step": 764
    },
    {
      "epoch": 0.03405806767151305,
      "grad_norm": 0.5594723224639893,
      "learning_rate": 0.0007660000000000001,
      "loss": 2.2891,
      "step": 766
    },
    {
      "epoch": 0.034146992130185405,
      "grad_norm": 0.7394406199455261,
      "learning_rate": 0.000768,
      "loss": 2.301,
      "step": 768
    },
    {
      "epoch": 0.034235916588857766,
      "grad_norm": 0.45566725730895996,
      "learning_rate": 0.0007700000000000001,
      "loss": 2.2951,
      "step": 770
    },
    {
      "epoch": 0.03432484104753012,
      "grad_norm": 0.4442650079727173,
      "learning_rate": 0.000772,
      "loss": 2.2941,
      "step": 772
    },
    {
      "epoch": 0.03441376550620248,
      "grad_norm": 0.4281231164932251,
      "learning_rate": 0.0007740000000000001,
      "loss": 2.2621,
      "step": 774
    },
    {
      "epoch": 0.034502689964874836,
      "grad_norm": 0.47158682346343994,
      "learning_rate": 0.000776,
      "loss": 2.2682,
      "step": 776
    },
    {
      "epoch": 0.0345916144235472,
      "grad_norm": 0.46278801560401917,
      "learning_rate": 0.000778,
      "loss": 2.2633,
      "step": 778
    },
    {
      "epoch": 0.03468053888221955,
      "grad_norm": 0.5910720229148865,
      "learning_rate": 0.0007800000000000001,
      "loss": 2.2861,
      "step": 780
    },
    {
      "epoch": 0.03476946334089191,
      "grad_norm": 0.6585924625396729,
      "learning_rate": 0.000782,
      "loss": 2.281,
      "step": 782
    },
    {
      "epoch": 0.034858387799564274,
      "grad_norm": 0.5449870824813843,
      "learning_rate": 0.0007840000000000001,
      "loss": 2.2765,
      "step": 784
    },
    {
      "epoch": 0.03494731225823663,
      "grad_norm": 0.4451574981212616,
      "learning_rate": 0.000786,
      "loss": 2.2599,
      "step": 786
    },
    {
      "epoch": 0.03503623671690899,
      "grad_norm": 0.4292662441730499,
      "learning_rate": 0.0007880000000000001,
      "loss": 2.2572,
      "step": 788
    },
    {
      "epoch": 0.03512516117558134,
      "grad_norm": 0.446139931678772,
      "learning_rate": 0.00079,
      "loss": 2.2538,
      "step": 790
    },
    {
      "epoch": 0.035214085634253704,
      "grad_norm": 0.5321487784385681,
      "learning_rate": 0.0007920000000000001,
      "loss": 2.2516,
      "step": 792
    },
    {
      "epoch": 0.03530301009292606,
      "grad_norm": 0.5208760499954224,
      "learning_rate": 0.0007940000000000001,
      "loss": 2.2596,
      "step": 794
    },
    {
      "epoch": 0.03539193455159842,
      "grad_norm": 0.5374959707260132,
      "learning_rate": 0.000796,
      "loss": 2.2597,
      "step": 796
    },
    {
      "epoch": 0.035480859010270774,
      "grad_norm": 0.6247416734695435,
      "learning_rate": 0.0007980000000000001,
      "loss": 2.2682,
      "step": 798
    },
    {
      "epoch": 0.035569783468943135,
      "grad_norm": 0.720618724822998,
      "learning_rate": 0.0008,
      "loss": 2.2675,
      "step": 800
    },
    {
      "epoch": 0.03565870792761549,
      "grad_norm": 0.5454574227333069,
      "learning_rate": 0.0008020000000000001,
      "loss": 2.2625,
      "step": 802
    },
    {
      "epoch": 0.03574763238628785,
      "grad_norm": 0.47981786727905273,
      "learning_rate": 0.000804,
      "loss": 2.2511,
      "step": 804
    },
    {
      "epoch": 0.035836556844960205,
      "grad_norm": 0.4263577461242676,
      "learning_rate": 0.0008060000000000001,
      "loss": 2.2463,
      "step": 806
    },
    {
      "epoch": 0.035925481303632566,
      "grad_norm": 0.41618812084198,
      "learning_rate": 0.000808,
      "loss": 2.2463,
      "step": 808
    },
    {
      "epoch": 0.03601440576230492,
      "grad_norm": 0.3966350257396698,
      "learning_rate": 0.0008100000000000001,
      "loss": 2.2392,
      "step": 810
    },
    {
      "epoch": 0.03610333022097728,
      "grad_norm": 0.48056676983833313,
      "learning_rate": 0.0008120000000000001,
      "loss": 2.2379,
      "step": 812
    },
    {
      "epoch": 0.036192254679649635,
      "grad_norm": 0.4377332031726837,
      "learning_rate": 0.0008139999999999999,
      "loss": 2.237,
      "step": 814
    },
    {
      "epoch": 0.036281179138321996,
      "grad_norm": 0.3729042410850525,
      "learning_rate": 0.000816,
      "loss": 2.2328,
      "step": 816
    },
    {
      "epoch": 0.03637010359699435,
      "grad_norm": 0.3967163562774658,
      "learning_rate": 0.0008179999999999999,
      "loss": 2.2196,
      "step": 818
    },
    {
      "epoch": 0.03645902805566671,
      "grad_norm": 0.44842958450317383,
      "learning_rate": 0.00082,
      "loss": 2.2195,
      "step": 820
    },
    {
      "epoch": 0.036547952514339066,
      "grad_norm": 0.46168598532676697,
      "learning_rate": 0.0008219999999999999,
      "loss": 2.2219,
      "step": 822
    },
    {
      "epoch": 0.03663687697301143,
      "grad_norm": 0.5996253490447998,
      "learning_rate": 0.000824,
      "loss": 2.2359,
      "step": 824
    },
    {
      "epoch": 0.03672580143168378,
      "grad_norm": 0.575162410736084,
      "learning_rate": 0.000826,
      "loss": 2.2405,
      "step": 826
    },
    {
      "epoch": 0.03681472589035614,
      "grad_norm": 0.7242479920387268,
      "learning_rate": 0.000828,
      "loss": 2.2319,
      "step": 828
    },
    {
      "epoch": 0.036903650349028504,
      "grad_norm": 0.5244578719139099,
      "learning_rate": 0.00083,
      "loss": 2.2303,
      "step": 830
    },
    {
      "epoch": 0.03699257480770086,
      "grad_norm": 0.5107928514480591,
      "learning_rate": 0.000832,
      "loss": 2.2203,
      "step": 832
    },
    {
      "epoch": 0.03708149926637322,
      "grad_norm": 0.43508070707321167,
      "learning_rate": 0.000834,
      "loss": 2.2151,
      "step": 834
    },
    {
      "epoch": 0.03717042372504557,
      "grad_norm": 0.4192095398902893,
      "learning_rate": 0.0008359999999999999,
      "loss": 2.2141,
      "step": 836
    },
    {
      "epoch": 0.037259348183717934,
      "grad_norm": 0.49640336632728577,
      "learning_rate": 0.000838,
      "loss": 2.2206,
      "step": 838
    },
    {
      "epoch": 0.03734827264239029,
      "grad_norm": 0.4244445264339447,
      "learning_rate": 0.00084,
      "loss": 2.213,
      "step": 840
    },
    {
      "epoch": 0.03743719710106265,
      "grad_norm": 0.5675143003463745,
      "learning_rate": 0.000842,
      "loss": 2.2096,
      "step": 842
    },
    {
      "epoch": 0.037526121559735004,
      "grad_norm": 0.5136843323707581,
      "learning_rate": 0.000844,
      "loss": 2.2086,
      "step": 844
    },
    {
      "epoch": 0.037615046018407365,
      "grad_norm": 0.660275399684906,
      "learning_rate": 0.000846,
      "loss": 2.2125,
      "step": 846
    },
    {
      "epoch": 0.03770397047707972,
      "grad_norm": 0.4898814558982849,
      "learning_rate": 0.000848,
      "loss": 2.2159,
      "step": 848
    },
    {
      "epoch": 0.03779289493575208,
      "grad_norm": 0.47421759366989136,
      "learning_rate": 0.00085,
      "loss": 2.2037,
      "step": 850
    },
    {
      "epoch": 0.037881819394424435,
      "grad_norm": 0.46137571334838867,
      "learning_rate": 0.000852,
      "loss": 2.1975,
      "step": 852
    },
    {
      "epoch": 0.037970743853096796,
      "grad_norm": 0.4468640089035034,
      "learning_rate": 0.000854,
      "loss": 2.1951,
      "step": 854
    },
    {
      "epoch": 0.03805966831176915,
      "grad_norm": 0.43135735392570496,
      "learning_rate": 0.000856,
      "loss": 2.1939,
      "step": 856
    },
    {
      "epoch": 0.03814859277044151,
      "grad_norm": 0.5193342566490173,
      "learning_rate": 0.000858,
      "loss": 2.199,
      "step": 858
    },
    {
      "epoch": 0.038237517229113865,
      "grad_norm": 0.4511534571647644,
      "learning_rate": 0.00086,
      "loss": 2.1918,
      "step": 860
    },
    {
      "epoch": 0.03832644168778623,
      "grad_norm": 0.40734702348709106,
      "learning_rate": 0.000862,
      "loss": 2.1835,
      "step": 862
    },
    {
      "epoch": 0.03841536614645858,
      "grad_norm": 0.38885200023651123,
      "learning_rate": 0.000864,
      "loss": 2.1822,
      "step": 864
    },
    {
      "epoch": 0.03850429060513094,
      "grad_norm": 0.4552465081214905,
      "learning_rate": 0.000866,
      "loss": 2.1789,
      "step": 866
    },
    {
      "epoch": 0.038593215063803296,
      "grad_norm": 0.45240339636802673,
      "learning_rate": 0.0008680000000000001,
      "loss": 2.1767,
      "step": 868
    },
    {
      "epoch": 0.03868213952247566,
      "grad_norm": 0.3916630744934082,
      "learning_rate": 0.00087,
      "loss": 2.1763,
      "step": 870
    },
    {
      "epoch": 0.03877106398114801,
      "grad_norm": 0.34862634539604187,
      "learning_rate": 0.000872,
      "loss": 2.1735,
      "step": 872
    },
    {
      "epoch": 0.03885998843982037,
      "grad_norm": 0.5051468014717102,
      "learning_rate": 0.000874,
      "loss": 2.1749,
      "step": 874
    },
    {
      "epoch": 0.038948912898492734,
      "grad_norm": 0.43139076232910156,
      "learning_rate": 0.000876,
      "loss": 2.1651,
      "step": 876
    },
    {
      "epoch": 0.03903783735716509,
      "grad_norm": 0.5403070449829102,
      "learning_rate": 0.000878,
      "loss": 2.1745,
      "step": 878
    },
    {
      "epoch": 0.03912676181583745,
      "grad_norm": 0.4276416599750519,
      "learning_rate": 0.00088,
      "loss": 2.1771,
      "step": 880
    },
    {
      "epoch": 0.0392156862745098,
      "grad_norm": 0.4407658576965332,
      "learning_rate": 0.000882,
      "loss": 2.1773,
      "step": 882
    },
    {
      "epoch": 0.039304610733182165,
      "grad_norm": 0.42780712246894836,
      "learning_rate": 0.000884,
      "loss": 2.175,
      "step": 884
    },
    {
      "epoch": 0.03939353519185452,
      "grad_norm": 0.4067372977733612,
      "learning_rate": 0.0008860000000000001,
      "loss": 2.1651,
      "step": 886
    },
    {
      "epoch": 0.03948245965052688,
      "grad_norm": 0.40394145250320435,
      "learning_rate": 0.000888,
      "loss": 2.1654,
      "step": 888
    },
    {
      "epoch": 0.039571384109199234,
      "grad_norm": 0.4600394666194916,
      "learning_rate": 0.0008900000000000001,
      "loss": 2.158,
      "step": 890
    },
    {
      "epoch": 0.039660308567871595,
      "grad_norm": 0.5146481990814209,
      "learning_rate": 0.000892,
      "loss": 2.175,
      "step": 892
    },
    {
      "epoch": 0.03974923302654395,
      "grad_norm": 0.5569943189620972,
      "learning_rate": 0.000894,
      "loss": 2.1744,
      "step": 894
    },
    {
      "epoch": 0.03983815748521631,
      "grad_norm": 0.5006479024887085,
      "learning_rate": 0.000896,
      "loss": 2.1685,
      "step": 896
    },
    {
      "epoch": 0.039927081943888665,
      "grad_norm": 0.45075637102127075,
      "learning_rate": 0.000898,
      "loss": 2.1591,
      "step": 898
    },
    {
      "epoch": 0.040016006402561026,
      "grad_norm": 0.450245201587677,
      "learning_rate": 0.0009000000000000001,
      "loss": 2.1606,
      "step": 900
    },
    {
      "epoch": 0.04010493086123338,
      "grad_norm": 0.41936612129211426,
      "learning_rate": 0.000902,
      "loss": 2.16,
      "step": 902
    },
    {
      "epoch": 0.04019385531990574,
      "grad_norm": 0.44605305790901184,
      "learning_rate": 0.0009040000000000001,
      "loss": 2.1485,
      "step": 904
    },
    {
      "epoch": 0.040282779778578096,
      "grad_norm": 0.4443049430847168,
      "learning_rate": 0.000906,
      "loss": 2.1607,
      "step": 906
    },
    {
      "epoch": 0.04037170423725046,
      "grad_norm": 0.40073588490486145,
      "learning_rate": 0.0009080000000000001,
      "loss": 2.1574,
      "step": 908
    },
    {
      "epoch": 0.04046062869592281,
      "grad_norm": 0.35070598125457764,
      "learning_rate": 0.00091,
      "loss": 2.1515,
      "step": 910
    },
    {
      "epoch": 0.04054955315459517,
      "grad_norm": 0.37149950861930847,
      "learning_rate": 0.000912,
      "loss": 2.1477,
      "step": 912
    },
    {
      "epoch": 0.040638477613267526,
      "grad_norm": 0.3967283368110657,
      "learning_rate": 0.0009140000000000001,
      "loss": 2.1482,
      "step": 914
    },
    {
      "epoch": 0.04072740207193989,
      "grad_norm": 0.3983295261859894,
      "learning_rate": 0.000916,
      "loss": 2.1413,
      "step": 916
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 0.7367068529129028,
      "learning_rate": 0.0009180000000000001,
      "loss": 2.164,
      "step": 918
    },
    {
      "epoch": 0.0409052509892846,
      "grad_norm": 0.5214952230453491,
      "learning_rate": 0.00092,
      "loss": 2.1595,
      "step": 920
    },
    {
      "epoch": 0.040994175447956964,
      "grad_norm": 0.4257142245769501,
      "learning_rate": 0.0009220000000000001,
      "loss": 2.149,
      "step": 922
    },
    {
      "epoch": 0.04108309990662932,
      "grad_norm": 0.3842937648296356,
      "learning_rate": 0.000924,
      "loss": 2.1419,
      "step": 924
    },
    {
      "epoch": 0.04117202436530168,
      "grad_norm": 0.37283626198768616,
      "learning_rate": 0.0009260000000000001,
      "loss": 2.1455,
      "step": 926
    },
    {
      "epoch": 0.041260948823974034,
      "grad_norm": 0.38478460907936096,
      "learning_rate": 0.0009280000000000001,
      "loss": 2.1416,
      "step": 928
    },
    {
      "epoch": 0.041349873282646395,
      "grad_norm": 0.37205952405929565,
      "learning_rate": 0.00093,
      "loss": 2.147,
      "step": 930
    },
    {
      "epoch": 0.04143879774131875,
      "grad_norm": 0.38429123163223267,
      "learning_rate": 0.0009320000000000001,
      "loss": 2.1287,
      "step": 932
    },
    {
      "epoch": 0.04152772219999111,
      "grad_norm": 0.38537994027137756,
      "learning_rate": 0.000934,
      "loss": 2.129,
      "step": 934
    },
    {
      "epoch": 0.041616646658663464,
      "grad_norm": 0.35850268602371216,
      "learning_rate": 0.0009360000000000001,
      "loss": 2.1279,
      "step": 936
    },
    {
      "epoch": 0.041705571117335825,
      "grad_norm": 0.40889298915863037,
      "learning_rate": 0.0009379999999999999,
      "loss": 2.1328,
      "step": 938
    },
    {
      "epoch": 0.04179449557600818,
      "grad_norm": 0.44349735975265503,
      "learning_rate": 0.00094,
      "loss": 2.1313,
      "step": 940
    },
    {
      "epoch": 0.04188342003468054,
      "grad_norm": 0.38156867027282715,
      "learning_rate": 0.000942,
      "loss": 2.1207,
      "step": 942
    },
    {
      "epoch": 0.041972344493352895,
      "grad_norm": 0.34405317902565,
      "learning_rate": 0.000944,
      "loss": 2.115,
      "step": 944
    },
    {
      "epoch": 0.042061268952025256,
      "grad_norm": 0.3376498818397522,
      "learning_rate": 0.000946,
      "loss": 2.1125,
      "step": 946
    },
    {
      "epoch": 0.04215019341069761,
      "grad_norm": 0.4137539565563202,
      "learning_rate": 0.000948,
      "loss": 2.1226,
      "step": 948
    },
    {
      "epoch": 0.04223911786936997,
      "grad_norm": 0.4108043611049652,
      "learning_rate": 0.00095,
      "loss": 2.1165,
      "step": 950
    },
    {
      "epoch": 0.042328042328042326,
      "grad_norm": 0.46412286162376404,
      "learning_rate": 0.0009519999999999999,
      "loss": 2.1204,
      "step": 952
    },
    {
      "epoch": 0.04241696678671469,
      "grad_norm": 0.5203374624252319,
      "learning_rate": 0.000954,
      "loss": 2.1255,
      "step": 954
    },
    {
      "epoch": 0.04250589124538704,
      "grad_norm": 0.5785240530967712,
      "learning_rate": 0.0009559999999999999,
      "loss": 2.1322,
      "step": 956
    },
    {
      "epoch": 0.0425948157040594,
      "grad_norm": 0.43567749857902527,
      "learning_rate": 0.000958,
      "loss": 2.1276,
      "step": 958
    },
    {
      "epoch": 0.042683740162731756,
      "grad_norm": 0.40512555837631226,
      "learning_rate": 0.00096,
      "loss": 2.1237,
      "step": 960
    },
    {
      "epoch": 0.04277266462140412,
      "grad_norm": 0.3421748876571655,
      "learning_rate": 0.000962,
      "loss": 2.1184,
      "step": 962
    },
    {
      "epoch": 0.04286158908007647,
      "grad_norm": 0.3585096001625061,
      "learning_rate": 0.000964,
      "loss": 2.1104,
      "step": 964
    },
    {
      "epoch": 0.04295051353874883,
      "grad_norm": 0.37870660424232483,
      "learning_rate": 0.000966,
      "loss": 2.112,
      "step": 966
    },
    {
      "epoch": 0.043039437997421194,
      "grad_norm": 0.37261155247688293,
      "learning_rate": 0.000968,
      "loss": 2.1118,
      "step": 968
    },
    {
      "epoch": 0.04312836245609355,
      "grad_norm": 0.3406919538974762,
      "learning_rate": 0.0009699999999999999,
      "loss": 2.1024,
      "step": 970
    },
    {
      "epoch": 0.04321728691476591,
      "grad_norm": 0.3668791651725769,
      "learning_rate": 0.000972,
      "loss": 2.1062,
      "step": 972
    },
    {
      "epoch": 0.043306211373438264,
      "grad_norm": 0.40684422850608826,
      "learning_rate": 0.000974,
      "loss": 2.0927,
      "step": 974
    },
    {
      "epoch": 0.043395135832110625,
      "grad_norm": 0.4066929221153259,
      "learning_rate": 0.000976,
      "loss": 2.1012,
      "step": 976
    },
    {
      "epoch": 0.04348406029078298,
      "grad_norm": 0.34670764207839966,
      "learning_rate": 0.000978,
      "loss": 2.0961,
      "step": 978
    },
    {
      "epoch": 0.04357298474945534,
      "grad_norm": 0.3352697789669037,
      "learning_rate": 0.00098,
      "loss": 2.0942,
      "step": 980
    },
    {
      "epoch": 0.043661909208127694,
      "grad_norm": 0.3769042491912842,
      "learning_rate": 0.000982,
      "loss": 2.0997,
      "step": 982
    },
    {
      "epoch": 0.043750833666800056,
      "grad_norm": 0.3640815317630768,
      "learning_rate": 0.000984,
      "loss": 2.107,
      "step": 984
    },
    {
      "epoch": 0.04383975812547241,
      "grad_norm": 0.40777748823165894,
      "learning_rate": 0.0009860000000000001,
      "loss": 2.0952,
      "step": 986
    },
    {
      "epoch": 0.04392868258414477,
      "grad_norm": 0.32192882895469666,
      "learning_rate": 0.000988,
      "loss": 2.0917,
      "step": 988
    },
    {
      "epoch": 0.044017607042817125,
      "grad_norm": 0.3336566090583801,
      "learning_rate": 0.00099,
      "loss": 2.0988,
      "step": 990
    },
    {
      "epoch": 0.044106531501489486,
      "grad_norm": 0.31402355432510376,
      "learning_rate": 0.000992,
      "loss": 2.0945,
      "step": 992
    },
    {
      "epoch": 0.04419545596016184,
      "grad_norm": 0.37855321168899536,
      "learning_rate": 0.000994,
      "loss": 2.0981,
      "step": 994
    },
    {
      "epoch": 0.0442843804188342,
      "grad_norm": 0.4493234157562256,
      "learning_rate": 0.000996,
      "loss": 2.0848,
      "step": 996
    },
    {
      "epoch": 0.044373304877506556,
      "grad_norm": 0.5948104858398438,
      "learning_rate": 0.000998,
      "loss": 2.1067,
      "step": 998
    },
    {
      "epoch": 0.04446222933617892,
      "grad_norm": 0.5480212569236755,
      "learning_rate": 0.001,
      "loss": 2.1027,
      "step": 1000
    },
    {
      "epoch": 0.04446222933617892,
      "eval_loss": 2.0268025398254395,
      "eval_runtime": 12.884,
      "eval_samples_per_second": 536.325,
      "eval_steps_per_second": 67.06,
      "step": 1000
    },
    {
      "epoch": 0.04455115379485127,
      "grad_norm": 0.41122549772262573,
      "learning_rate": 0.0009999999987529585,
      "loss": 2.0929,
      "step": 1002
    },
    {
      "epoch": 0.04464007825352363,
      "grad_norm": 0.33955812454223633,
      "learning_rate": 0.000999999995011834,
      "loss": 2.0879,
      "step": 1004
    },
    {
      "epoch": 0.04472900271219599,
      "grad_norm": 0.3389456570148468,
      "learning_rate": 0.0009999999887766267,
      "loss": 2.0822,
      "step": 1006
    },
    {
      "epoch": 0.04481792717086835,
      "grad_norm": 0.356757253408432,
      "learning_rate": 0.0009999999800473362,
      "loss": 2.0849,
      "step": 1008
    },
    {
      "epoch": 0.0449068516295407,
      "grad_norm": 0.3514195680618286,
      "learning_rate": 0.000999999968823963,
      "loss": 2.0822,
      "step": 1010
    },
    {
      "epoch": 0.04499577608821306,
      "grad_norm": 0.3277719020843506,
      "learning_rate": 0.0009999999551065068,
      "loss": 2.0798,
      "step": 1012
    },
    {
      "epoch": 0.045084700546885424,
      "grad_norm": 0.33637166023254395,
      "learning_rate": 0.000999999938894968,
      "loss": 2.0844,
      "step": 1014
    },
    {
      "epoch": 0.04517362500555778,
      "grad_norm": 0.3239494860172272,
      "learning_rate": 0.0009999999201893463,
      "loss": 2.0747,
      "step": 1016
    },
    {
      "epoch": 0.04526254946423014,
      "grad_norm": 0.3227521777153015,
      "learning_rate": 0.000999999898989642,
      "loss": 2.0656,
      "step": 1018
    },
    {
      "epoch": 0.045351473922902494,
      "grad_norm": 0.3502488136291504,
      "learning_rate": 0.0009999998752958554,
      "loss": 2.0771,
      "step": 1020
    },
    {
      "epoch": 0.045440398381574855,
      "grad_norm": 0.33191025257110596,
      "learning_rate": 0.0009999998491079863,
      "loss": 2.0785,
      "step": 1022
    },
    {
      "epoch": 0.04552932284024721,
      "grad_norm": 0.31606367230415344,
      "learning_rate": 0.000999999820426035,
      "loss": 2.0669,
      "step": 1024
    },
    {
      "epoch": 0.04561824729891957,
      "grad_norm": 0.3388826847076416,
      "learning_rate": 0.0009999997892500016,
      "loss": 2.0602,
      "step": 1026
    },
    {
      "epoch": 0.045707171757591925,
      "grad_norm": 0.32445207238197327,
      "learning_rate": 0.0009999997555798863,
      "loss": 2.06,
      "step": 1028
    },
    {
      "epoch": 0.045796096216264286,
      "grad_norm": 0.2825400233268738,
      "learning_rate": 0.000999999719415689,
      "loss": 2.0588,
      "step": 1030
    },
    {
      "epoch": 0.04588502067493664,
      "grad_norm": 0.3642183542251587,
      "learning_rate": 0.0009999996807574104,
      "loss": 2.0636,
      "step": 1032
    },
    {
      "epoch": 0.045973945133609,
      "grad_norm": 0.39157211780548096,
      "learning_rate": 0.0009999996396050502,
      "loss": 2.0637,
      "step": 1034
    },
    {
      "epoch": 0.046062869592281355,
      "grad_norm": 0.36561650037765503,
      "learning_rate": 0.000999999595958609,
      "loss": 2.0696,
      "step": 1036
    },
    {
      "epoch": 0.046151794050953716,
      "grad_norm": 0.3913203477859497,
      "learning_rate": 0.0009999995498180868,
      "loss": 2.0639,
      "step": 1038
    },
    {
      "epoch": 0.04624071850962607,
      "grad_norm": 0.31236109137535095,
      "learning_rate": 0.0009999995011834835,
      "loss": 2.058,
      "step": 1040
    },
    {
      "epoch": 0.04632964296829843,
      "grad_norm": 0.3150652050971985,
      "learning_rate": 0.0009999994500548,
      "loss": 2.061,
      "step": 1042
    },
    {
      "epoch": 0.046418567426970786,
      "grad_norm": 0.31297731399536133,
      "learning_rate": 0.0009999993964320362,
      "loss": 2.059,
      "step": 1044
    },
    {
      "epoch": 0.04650749188564315,
      "grad_norm": 0.35480573773384094,
      "learning_rate": 0.0009999993403151924,
      "loss": 2.0582,
      "step": 1046
    },
    {
      "epoch": 0.0465964163443155,
      "grad_norm": 0.34173378348350525,
      "learning_rate": 0.000999999281704269,
      "loss": 2.0601,
      "step": 1048
    },
    {
      "epoch": 0.04668534080298786,
      "grad_norm": 0.3285476267337799,
      "learning_rate": 0.000999999220599266,
      "loss": 2.0555,
      "step": 1050
    },
    {
      "epoch": 0.04677426526166022,
      "grad_norm": 0.3582735061645508,
      "learning_rate": 0.000999999157000184,
      "loss": 2.0468,
      "step": 1052
    },
    {
      "epoch": 0.04686318972033258,
      "grad_norm": 0.3618029057979584,
      "learning_rate": 0.0009999990909070231,
      "loss": 2.0443,
      "step": 1054
    },
    {
      "epoch": 0.04695211417900493,
      "grad_norm": 0.3185672461986542,
      "learning_rate": 0.000999999022319784,
      "loss": 2.0531,
      "step": 1056
    },
    {
      "epoch": 0.04704103863767729,
      "grad_norm": 0.38706305623054504,
      "learning_rate": 0.0009999989512384665,
      "loss": 2.0531,
      "step": 1058
    },
    {
      "epoch": 0.047129963096349654,
      "grad_norm": 0.28708359599113464,
      "learning_rate": 0.0009999988776630713,
      "loss": 2.0482,
      "step": 1060
    },
    {
      "epoch": 0.04721888755502201,
      "grad_norm": 0.37909284234046936,
      "learning_rate": 0.0009999988015935988,
      "loss": 2.0437,
      "step": 1062
    },
    {
      "epoch": 0.04730781201369437,
      "grad_norm": 0.35751160979270935,
      "learning_rate": 0.0009999987230300492,
      "loss": 2.0528,
      "step": 1064
    },
    {
      "epoch": 0.047396736472366724,
      "grad_norm": 0.31660065054893494,
      "learning_rate": 0.000999998641972423,
      "loss": 2.0399,
      "step": 1066
    },
    {
      "epoch": 0.047485660931039085,
      "grad_norm": 0.3356022238731384,
      "learning_rate": 0.0009999985584207205,
      "loss": 2.039,
      "step": 1068
    },
    {
      "epoch": 0.04757458538971144,
      "grad_norm": 0.30728912353515625,
      "learning_rate": 0.0009999984723749423,
      "loss": 2.0428,
      "step": 1070
    },
    {
      "epoch": 0.0476635098483838,
      "grad_norm": 0.3009743094444275,
      "learning_rate": 0.0009999983838350887,
      "loss": 2.0365,
      "step": 1072
    },
    {
      "epoch": 0.047752434307056155,
      "grad_norm": 0.308032751083374,
      "learning_rate": 0.0009999982928011603,
      "loss": 2.0353,
      "step": 1074
    },
    {
      "epoch": 0.047841358765728516,
      "grad_norm": 0.30482032895088196,
      "learning_rate": 0.0009999981992731572,
      "loss": 2.0267,
      "step": 1076
    },
    {
      "epoch": 0.04793028322440087,
      "grad_norm": 0.2963328957557678,
      "learning_rate": 0.0009999981032510802,
      "loss": 2.0352,
      "step": 1078
    },
    {
      "epoch": 0.04801920768307323,
      "grad_norm": 0.32439178228378296,
      "learning_rate": 0.0009999980047349295,
      "loss": 2.0265,
      "step": 1080
    },
    {
      "epoch": 0.048108132141745585,
      "grad_norm": 0.3173850178718567,
      "learning_rate": 0.000999997903724706,
      "loss": 2.0351,
      "step": 1082
    },
    {
      "epoch": 0.04819705660041795,
      "grad_norm": 0.33313167095184326,
      "learning_rate": 0.0009999978002204098,
      "loss": 2.0257,
      "step": 1084
    },
    {
      "epoch": 0.0482859810590903,
      "grad_norm": 0.36748653650283813,
      "learning_rate": 0.0009999976942220417,
      "loss": 2.0357,
      "step": 1086
    },
    {
      "epoch": 0.04837490551776266,
      "grad_norm": 0.33924633264541626,
      "learning_rate": 0.000999997585729602,
      "loss": 2.0275,
      "step": 1088
    },
    {
      "epoch": 0.048463829976435016,
      "grad_norm": 0.3379874527454376,
      "learning_rate": 0.0009999974747430913,
      "loss": 2.0361,
      "step": 1090
    },
    {
      "epoch": 0.04855275443510738,
      "grad_norm": 0.2935180068016052,
      "learning_rate": 0.0009999973612625102,
      "loss": 2.0229,
      "step": 1092
    },
    {
      "epoch": 0.04864167889377973,
      "grad_norm": 0.3268764317035675,
      "learning_rate": 0.0009999972452878593,
      "loss": 2.0253,
      "step": 1094
    },
    {
      "epoch": 0.04873060335245209,
      "grad_norm": 0.27827081084251404,
      "learning_rate": 0.0009999971268191393,
      "loss": 2.0234,
      "step": 1096
    },
    {
      "epoch": 0.04881952781112445,
      "grad_norm": 0.3332640528678894,
      "learning_rate": 0.0009999970058563506,
      "loss": 2.023,
      "step": 1098
    },
    {
      "epoch": 0.04890845226979681,
      "grad_norm": 0.2747075855731964,
      "learning_rate": 0.0009999968823994937,
      "loss": 2.0189,
      "step": 1100
    },
    {
      "epoch": 0.04899737672846916,
      "grad_norm": 0.32247695326805115,
      "learning_rate": 0.0009999967564485695,
      "loss": 2.0231,
      "step": 1102
    },
    {
      "epoch": 0.04908630118714152,
      "grad_norm": 0.2975604832172394,
      "learning_rate": 0.0009999966280035784,
      "loss": 2.0139,
      "step": 1104
    },
    {
      "epoch": 0.049175225645813885,
      "grad_norm": 0.3203318417072296,
      "learning_rate": 0.000999996497064521,
      "loss": 2.0207,
      "step": 1106
    },
    {
      "epoch": 0.04926415010448624,
      "grad_norm": 0.3753434717655182,
      "learning_rate": 0.0009999963636313982,
      "loss": 2.0153,
      "step": 1108
    },
    {
      "epoch": 0.0493530745631586,
      "grad_norm": 0.35570961236953735,
      "learning_rate": 0.0009999962277042107,
      "loss": 2.0186,
      "step": 1110
    },
    {
      "epoch": 0.049441999021830954,
      "grad_norm": 0.3200981020927429,
      "learning_rate": 0.0009999960892829588,
      "loss": 2.0169,
      "step": 1112
    },
    {
      "epoch": 0.049530923480503315,
      "grad_norm": 0.31688785552978516,
      "learning_rate": 0.0009999959483676436,
      "loss": 2.0098,
      "step": 1114
    },
    {
      "epoch": 0.04961984793917567,
      "grad_norm": 0.2900930941104889,
      "learning_rate": 0.0009999958049582655,
      "loss": 2.0136,
      "step": 1116
    },
    {
      "epoch": 0.04970877239784803,
      "grad_norm": 0.337090402841568,
      "learning_rate": 0.000999995659054825,
      "loss": 2.0138,
      "step": 1118
    },
    {
      "epoch": 0.049797696856520385,
      "grad_norm": 0.41873857378959656,
      "learning_rate": 0.0009999955106573238,
      "loss": 2.0086,
      "step": 1120
    },
    {
      "epoch": 0.049886621315192746,
      "grad_norm": 0.3219401240348816,
      "learning_rate": 0.0009999953597657617,
      "loss": 2.018,
      "step": 1122
    },
    {
      "epoch": 0.0499755457738651,
      "grad_norm": 0.337938517332077,
      "learning_rate": 0.0009999952063801396,
      "loss": 2.0093,
      "step": 1124
    },
    {
      "epoch": 0.05006447023253746,
      "grad_norm": 0.3835712969303131,
      "learning_rate": 0.0009999950505004587,
      "loss": 2.0144,
      "step": 1126
    },
    {
      "epoch": 0.050153394691209816,
      "grad_norm": 0.3164381682872772,
      "learning_rate": 0.0009999948921267192,
      "loss": 2.0111,
      "step": 1128
    },
    {
      "epoch": 0.05024231914988218,
      "grad_norm": 0.2837233245372772,
      "learning_rate": 0.0009999947312589225,
      "loss": 2.0099,
      "step": 1130
    },
    {
      "epoch": 0.05033124360855453,
      "grad_norm": 0.30504587292671204,
      "learning_rate": 0.0009999945678970687,
      "loss": 2.0049,
      "step": 1132
    },
    {
      "epoch": 0.05042016806722689,
      "grad_norm": 0.30361393094062805,
      "learning_rate": 0.0009999944020411592,
      "loss": 2.0056,
      "step": 1134
    },
    {
      "epoch": 0.050509092525899246,
      "grad_norm": 0.2646671533584595,
      "learning_rate": 0.0009999942336911948,
      "loss": 1.9977,
      "step": 1136
    },
    {
      "epoch": 0.05059801698457161,
      "grad_norm": 0.3026483952999115,
      "learning_rate": 0.000999994062847176,
      "loss": 2.0039,
      "step": 1138
    },
    {
      "epoch": 0.05068694144324396,
      "grad_norm": 0.2975611388683319,
      "learning_rate": 0.0009999938895091035,
      "loss": 2.0043,
      "step": 1140
    },
    {
      "epoch": 0.05077586590191632,
      "grad_norm": 0.3021259605884552,
      "learning_rate": 0.000999993713676979,
      "loss": 2.0006,
      "step": 1142
    },
    {
      "epoch": 0.05086479036058868,
      "grad_norm": 0.2845633924007416,
      "learning_rate": 0.0009999935353508028,
      "loss": 1.9903,
      "step": 1144
    },
    {
      "epoch": 0.05095371481926104,
      "grad_norm": 0.2758135497570038,
      "learning_rate": 0.0009999933545305757,
      "loss": 2.0041,
      "step": 1146
    },
    {
      "epoch": 0.05104263927793339,
      "grad_norm": 0.2746117413043976,
      "learning_rate": 0.0009999931712162985,
      "loss": 1.9894,
      "step": 1148
    },
    {
      "epoch": 0.051131563736605753,
      "grad_norm": 0.2912278473377228,
      "learning_rate": 0.0009999929854079728,
      "loss": 1.9964,
      "step": 1150
    },
    {
      "epoch": 0.051220488195278115,
      "grad_norm": 0.28964367508888245,
      "learning_rate": 0.0009999927971055989,
      "loss": 1.9883,
      "step": 1152
    },
    {
      "epoch": 0.05130941265395047,
      "grad_norm": 0.2836471498012543,
      "learning_rate": 0.0009999926063091779,
      "loss": 1.9925,
      "step": 1154
    },
    {
      "epoch": 0.05139833711262283,
      "grad_norm": 0.2557382583618164,
      "learning_rate": 0.000999992413018711,
      "loss": 1.996,
      "step": 1156
    },
    {
      "epoch": 0.051487261571295184,
      "grad_norm": 0.3542958199977875,
      "learning_rate": 0.0009999922172341988,
      "loss": 1.9893,
      "step": 1158
    },
    {
      "epoch": 0.051576186029967545,
      "grad_norm": 0.4211042821407318,
      "learning_rate": 0.0009999920189556425,
      "loss": 1.9965,
      "step": 1160
    },
    {
      "epoch": 0.0516651104886399,
      "grad_norm": 0.3262304365634918,
      "learning_rate": 0.0009999918181830428,
      "loss": 1.9993,
      "step": 1162
    },
    {
      "epoch": 0.05175403494731226,
      "grad_norm": 0.33062461018562317,
      "learning_rate": 0.0009999916149164012,
      "loss": 1.9873,
      "step": 1164
    },
    {
      "epoch": 0.051842959405984615,
      "grad_norm": 0.2799008786678314,
      "learning_rate": 0.0009999914091557182,
      "loss": 1.9846,
      "step": 1166
    },
    {
      "epoch": 0.051931883864656976,
      "grad_norm": 0.2756935656070709,
      "learning_rate": 0.0009999912009009953,
      "loss": 1.9872,
      "step": 1168
    },
    {
      "epoch": 0.05202080832332933,
      "grad_norm": 0.2897654175758362,
      "learning_rate": 0.000999990990152233,
      "loss": 1.9908,
      "step": 1170
    },
    {
      "epoch": 0.05210973278200169,
      "grad_norm": 0.27705973386764526,
      "learning_rate": 0.0009999907769094327,
      "loss": 1.9873,
      "step": 1172
    },
    {
      "epoch": 0.052198657240674046,
      "grad_norm": 0.3140580654144287,
      "learning_rate": 0.0009999905611725957,
      "loss": 1.9881,
      "step": 1174
    },
    {
      "epoch": 0.05228758169934641,
      "grad_norm": 0.27007198333740234,
      "learning_rate": 0.0009999903429417226,
      "loss": 1.9862,
      "step": 1176
    },
    {
      "epoch": 0.05237650615801876,
      "grad_norm": 0.2722287178039551,
      "learning_rate": 0.0009999901222168144,
      "loss": 1.9773,
      "step": 1178
    },
    {
      "epoch": 0.05246543061669112,
      "grad_norm": 0.2842351496219635,
      "learning_rate": 0.0009999898989978728,
      "loss": 1.9826,
      "step": 1180
    },
    {
      "epoch": 0.052554355075363476,
      "grad_norm": 0.29141294956207275,
      "learning_rate": 0.0009999896732848985,
      "loss": 1.9808,
      "step": 1182
    },
    {
      "epoch": 0.05264327953403584,
      "grad_norm": 0.2940419316291809,
      "learning_rate": 0.000999989445077893,
      "loss": 1.9789,
      "step": 1184
    },
    {
      "epoch": 0.05273220399270819,
      "grad_norm": 0.30048856139183044,
      "learning_rate": 0.0009999892143768567,
      "loss": 1.979,
      "step": 1186
    },
    {
      "epoch": 0.05282112845138055,
      "grad_norm": 0.33986565470695496,
      "learning_rate": 0.0009999889811817915,
      "loss": 1.9688,
      "step": 1188
    },
    {
      "epoch": 0.05291005291005291,
      "grad_norm": 0.27422598004341125,
      "learning_rate": 0.0009999887454926982,
      "loss": 1.9818,
      "step": 1190
    },
    {
      "epoch": 0.05299897736872527,
      "grad_norm": 0.3004554808139801,
      "learning_rate": 0.0009999885073095779,
      "loss": 1.9704,
      "step": 1192
    },
    {
      "epoch": 0.05308790182739762,
      "grad_norm": 0.2710491716861725,
      "learning_rate": 0.0009999882666324321,
      "loss": 1.9735,
      "step": 1194
    },
    {
      "epoch": 0.053176826286069984,
      "grad_norm": 0.28977084159851074,
      "learning_rate": 0.0009999880234612616,
      "loss": 1.9682,
      "step": 1196
    },
    {
      "epoch": 0.053265750744742345,
      "grad_norm": 0.28438761830329895,
      "learning_rate": 0.0009999877777960681,
      "loss": 1.9728,
      "step": 1198
    },
    {
      "epoch": 0.0533546752034147,
      "grad_norm": 0.25546666979789734,
      "learning_rate": 0.0009999875296368525,
      "loss": 1.9651,
      "step": 1200
    },
    {
      "epoch": 0.05344359966208706,
      "grad_norm": 0.2793266773223877,
      "learning_rate": 0.000999987278983616,
      "loss": 1.9721,
      "step": 1202
    },
    {
      "epoch": 0.053532524120759414,
      "grad_norm": 0.3147725462913513,
      "learning_rate": 0.00099998702583636,
      "loss": 1.9682,
      "step": 1204
    },
    {
      "epoch": 0.053621448579431776,
      "grad_norm": 0.29513031244277954,
      "learning_rate": 0.000999986770195086,
      "loss": 1.9685,
      "step": 1206
    },
    {
      "epoch": 0.05371037303810413,
      "grad_norm": 0.25496989488601685,
      "learning_rate": 0.0009999865120597948,
      "loss": 1.9681,
      "step": 1208
    },
    {
      "epoch": 0.05379929749677649,
      "grad_norm": 0.2535437345504761,
      "learning_rate": 0.0009999862514304878,
      "loss": 1.9657,
      "step": 1210
    },
    {
      "epoch": 0.053888221955448845,
      "grad_norm": 0.9937949180603027,
      "learning_rate": 0.0009999859883071665,
      "loss": 1.9724,
      "step": 1212
    },
    {
      "epoch": 0.053977146414121206,
      "grad_norm": 0.7421650290489197,
      "learning_rate": 0.000999985722689832,
      "loss": 2.2602,
      "step": 1214
    },
    {
      "epoch": 0.05406607087279356,
      "grad_norm": 1.2442095279693604,
      "learning_rate": 0.0009999854545784859,
      "loss": 2.1722,
      "step": 1216
    },
    {
      "epoch": 0.05415499533146592,
      "grad_norm": 1.6552417278289795,
      "learning_rate": 0.0009999851839731292,
      "loss": 2.1954,
      "step": 1218
    },
    {
      "epoch": 0.054243919790138276,
      "grad_norm": 0.8579690456390381,
      "learning_rate": 0.0009999849108737633,
      "loss": 2.263,
      "step": 1220
    },
    {
      "epoch": 0.05433284424881064,
      "grad_norm": 1.149450421333313,
      "learning_rate": 0.00099998463528039,
      "loss": 2.1687,
      "step": 1222
    },
    {
      "epoch": 0.05442176870748299,
      "grad_norm": 3.364642858505249,
      "learning_rate": 0.00099998435719301,
      "loss": 2.3092,
      "step": 1224
    },
    {
      "epoch": 0.05451069316615535,
      "grad_norm": 2.966097593307495,
      "learning_rate": 0.0009999840766116252,
      "loss": 2.2565,
      "step": 1226
    },
    {
      "epoch": 0.054599617624827707,
      "grad_norm": 1.0643404722213745,
      "learning_rate": 0.0009999837935362368,
      "loss": 2.2603,
      "step": 1228
    },
    {
      "epoch": 0.05468854208350007,
      "grad_norm": 1.2306264638900757,
      "learning_rate": 0.0009999835079668462,
      "loss": 2.2179,
      "step": 1230
    },
    {
      "epoch": 0.05477746654217242,
      "grad_norm": 3.267204999923706,
      "learning_rate": 0.000999983219903455,
      "loss": 2.2186,
      "step": 1232
    },
    {
      "epoch": 0.05486639100084478,
      "grad_norm": 1.2005592584609985,
      "learning_rate": 0.0009999829293460645,
      "loss": 2.2835,
      "step": 1234
    },
    {
      "epoch": 0.05495531545951714,
      "grad_norm": 7.615020275115967,
      "learning_rate": 0.000999982636294676,
      "loss": 2.2995,
      "step": 1236
    },
    {
      "epoch": 0.0550442399181895,
      "grad_norm": 1.7953623533248901,
      "learning_rate": 0.0009999823407492912,
      "loss": 2.2996,
      "step": 1238
    },
    {
      "epoch": 0.05513316437686185,
      "grad_norm": 1.1320980787277222,
      "learning_rate": 0.0009999820427099115,
      "loss": 2.1913,
      "step": 1240
    },
    {
      "epoch": 0.055222088835534214,
      "grad_norm": 1.0222786664962769,
      "learning_rate": 0.0009999817421765384,
      "loss": 2.1571,
      "step": 1242
    },
    {
      "epoch": 0.055311013294206575,
      "grad_norm": 0.7209634780883789,
      "learning_rate": 0.0009999814391491732,
      "loss": 2.1174,
      "step": 1244
    },
    {
      "epoch": 0.05539993775287893,
      "grad_norm": 0.489495187997818,
      "learning_rate": 0.000999981133627818,
      "loss": 2.1004,
      "step": 1246
    },
    {
      "epoch": 0.05548886221155129,
      "grad_norm": 0.357696533203125,
      "learning_rate": 0.0009999808256124737,
      "loss": 2.0732,
      "step": 1248
    },
    {
      "epoch": 0.055577786670223644,
      "grad_norm": 0.41004520654678345,
      "learning_rate": 0.0009999805151031417,
      "loss": 2.0589,
      "step": 1250
    },
    {
      "epoch": 0.055666711128896006,
      "grad_norm": 0.29356011748313904,
      "learning_rate": 0.0009999802020998244,
      "loss": 2.0436,
      "step": 1252
    },
    {
      "epoch": 0.05575563558756836,
      "grad_norm": 0.2679998278617859,
      "learning_rate": 0.0009999798866025225,
      "loss": 2.0365,
      "step": 1254
    },
    {
      "epoch": 0.05584456004624072,
      "grad_norm": 0.28258460760116577,
      "learning_rate": 0.0009999795686112382,
      "loss": 2.0301,
      "step": 1256
    },
    {
      "epoch": 0.055933484504913075,
      "grad_norm": 0.24409818649291992,
      "learning_rate": 0.0009999792481259727,
      "loss": 2.0139,
      "step": 1258
    },
    {
      "epoch": 0.056022408963585436,
      "grad_norm": 0.2604723870754242,
      "learning_rate": 0.0009999789251467278,
      "loss": 2.0132,
      "step": 1260
    },
    {
      "epoch": 0.05611133342225779,
      "grad_norm": 0.23944151401519775,
      "learning_rate": 0.000999978599673505,
      "loss": 2.0037,
      "step": 1262
    },
    {
      "epoch": 0.05620025788093015,
      "grad_norm": 0.24858349561691284,
      "learning_rate": 0.0009999782717063058,
      "loss": 2.0112,
      "step": 1264
    },
    {
      "epoch": 0.056289182339602506,
      "grad_norm": 0.22487503290176392,
      "learning_rate": 0.0009999779412451322,
      "loss": 1.9946,
      "step": 1266
    },
    {
      "epoch": 0.05637810679827487,
      "grad_norm": 0.23384034633636475,
      "learning_rate": 0.0009999776082899854,
      "loss": 1.9879,
      "step": 1268
    },
    {
      "epoch": 0.05646703125694722,
      "grad_norm": 0.23720683157444,
      "learning_rate": 0.0009999772728408675,
      "loss": 1.9923,
      "step": 1270
    },
    {
      "epoch": 0.05655595571561958,
      "grad_norm": 0.23450630903244019,
      "learning_rate": 0.00099997693489778,
      "loss": 1.9795,
      "step": 1272
    },
    {
      "epoch": 0.05664488017429194,
      "grad_norm": 0.2357773780822754,
      "learning_rate": 0.0009999765944607244,
      "loss": 1.9775,
      "step": 1274
    },
    {
      "epoch": 0.0567338046329643,
      "grad_norm": 0.24982024729251862,
      "learning_rate": 0.0009999762515297025,
      "loss": 1.9817,
      "step": 1276
    },
    {
      "epoch": 0.05682272909163665,
      "grad_norm": 0.24399292469024658,
      "learning_rate": 0.0009999759061047162,
      "loss": 1.9772,
      "step": 1278
    },
    {
      "epoch": 0.05691165355030901,
      "grad_norm": 0.23936818540096283,
      "learning_rate": 0.000999975558185767,
      "loss": 1.9714,
      "step": 1280
    },
    {
      "epoch": 0.05700057800898137,
      "grad_norm": 0.24449017643928528,
      "learning_rate": 0.000999975207772857,
      "loss": 1.9671,
      "step": 1282
    },
    {
      "epoch": 0.05708950246765373,
      "grad_norm": 0.24265700578689575,
      "learning_rate": 0.0009999748548659875,
      "loss": 1.9614,
      "step": 1284
    },
    {
      "epoch": 0.05717842692632608,
      "grad_norm": 0.22446395456790924,
      "learning_rate": 0.0009999744994651602,
      "loss": 1.9688,
      "step": 1286
    },
    {
      "epoch": 0.057267351384998444,
      "grad_norm": 0.21999609470367432,
      "learning_rate": 0.0009999741415703774,
      "loss": 1.9591,
      "step": 1288
    },
    {
      "epoch": 0.057356275843670805,
      "grad_norm": 0.21769745647907257,
      "learning_rate": 0.0009999737811816406,
      "loss": 1.9581,
      "step": 1290
    },
    {
      "epoch": 0.05744520030234316,
      "grad_norm": 0.2255491316318512,
      "learning_rate": 0.0009999734182989514,
      "loss": 1.9604,
      "step": 1292
    },
    {
      "epoch": 0.05753412476101552,
      "grad_norm": 0.22669969499111176,
      "learning_rate": 0.0009999730529223119,
      "loss": 1.9538,
      "step": 1294
    },
    {
      "epoch": 0.057623049219687875,
      "grad_norm": 0.22954627871513367,
      "learning_rate": 0.0009999726850517237,
      "loss": 1.9534,
      "step": 1296
    },
    {
      "epoch": 0.057711973678360236,
      "grad_norm": 0.22537246346473694,
      "learning_rate": 0.0009999723146871889,
      "loss": 1.951,
      "step": 1298
    },
    {
      "epoch": 0.05780089813703259,
      "grad_norm": 0.2216878980398178,
      "learning_rate": 0.0009999719418287093,
      "loss": 1.9498,
      "step": 1300
    },
    {
      "epoch": 0.05788982259570495,
      "grad_norm": 0.23779501020908356,
      "learning_rate": 0.0009999715664762866,
      "loss": 1.951,
      "step": 1302
    },
    {
      "epoch": 0.057978747054377305,
      "grad_norm": 0.23081955313682556,
      "learning_rate": 0.0009999711886299226,
      "loss": 1.9514,
      "step": 1304
    },
    {
      "epoch": 0.058067671513049667,
      "grad_norm": 0.22819961607456207,
      "learning_rate": 0.0009999708082896192,
      "loss": 1.9488,
      "step": 1306
    },
    {
      "epoch": 0.05815659597172202,
      "grad_norm": 0.23504947125911713,
      "learning_rate": 0.0009999704254553788,
      "loss": 1.9488,
      "step": 1308
    },
    {
      "epoch": 0.05824552043039438,
      "grad_norm": 0.24413910508155823,
      "learning_rate": 0.0009999700401272025,
      "loss": 1.9443,
      "step": 1310
    },
    {
      "epoch": 0.058334444889066736,
      "grad_norm": 0.22513625025749207,
      "learning_rate": 0.000999969652305093,
      "loss": 1.9458,
      "step": 1312
    },
    {
      "epoch": 0.0584233693477391,
      "grad_norm": 0.22811587154865265,
      "learning_rate": 0.0009999692619890516,
      "loss": 1.9439,
      "step": 1314
    },
    {
      "epoch": 0.05851229380641145,
      "grad_norm": 0.21283623576164246,
      "learning_rate": 0.0009999688691790809,
      "loss": 1.9435,
      "step": 1316
    },
    {
      "epoch": 0.05860121826508381,
      "grad_norm": 0.22420313954353333,
      "learning_rate": 0.000999968473875182,
      "loss": 1.9389,
      "step": 1318
    },
    {
      "epoch": 0.05869014272375617,
      "grad_norm": 0.23344479501247406,
      "learning_rate": 0.000999968076077358,
      "loss": 1.9442,
      "step": 1320
    },
    {
      "epoch": 0.05877906718242853,
      "grad_norm": 0.21948975324630737,
      "learning_rate": 0.0009999676757856098,
      "loss": 1.9358,
      "step": 1322
    },
    {
      "epoch": 0.05886799164110088,
      "grad_norm": 0.24658890068531036,
      "learning_rate": 0.0009999672729999398,
      "loss": 1.9407,
      "step": 1324
    },
    {
      "epoch": 0.05895691609977324,
      "grad_norm": 0.23471440374851227,
      "learning_rate": 0.0009999668677203501,
      "loss": 1.942,
      "step": 1326
    },
    {
      "epoch": 0.0590458405584456,
      "grad_norm": 0.23664714395999908,
      "learning_rate": 0.000999966459946843,
      "loss": 1.9393,
      "step": 1328
    },
    {
      "epoch": 0.05913476501711796,
      "grad_norm": 0.21111926436424255,
      "learning_rate": 0.0009999660496794198,
      "loss": 1.9344,
      "step": 1330
    },
    {
      "epoch": 0.05922368947579031,
      "grad_norm": 0.20922254025936127,
      "learning_rate": 0.0009999656369180832,
      "loss": 1.9326,
      "step": 1332
    },
    {
      "epoch": 0.059312613934462674,
      "grad_norm": 0.2358391135931015,
      "learning_rate": 0.0009999652216628348,
      "loss": 1.9361,
      "step": 1334
    },
    {
      "epoch": 0.059401538393135035,
      "grad_norm": 0.2448922097682953,
      "learning_rate": 0.000999964803913677,
      "loss": 1.931,
      "step": 1336
    },
    {
      "epoch": 0.05949046285180739,
      "grad_norm": 0.2407871037721634,
      "learning_rate": 0.0009999643836706119,
      "loss": 1.9348,
      "step": 1338
    },
    {
      "epoch": 0.05957938731047975,
      "grad_norm": 0.22217388451099396,
      "learning_rate": 0.0009999639609336412,
      "loss": 1.9406,
      "step": 1340
    },
    {
      "epoch": 0.059668311769152105,
      "grad_norm": 0.2148897796869278,
      "learning_rate": 0.0009999635357027675,
      "loss": 1.927,
      "step": 1342
    },
    {
      "epoch": 0.059757236227824466,
      "grad_norm": 0.22098839282989502,
      "learning_rate": 0.0009999631079779926,
      "loss": 1.9321,
      "step": 1344
    },
    {
      "epoch": 0.05984616068649682,
      "grad_norm": 0.22021619975566864,
      "learning_rate": 0.0009999626777593188,
      "loss": 1.9246,
      "step": 1346
    },
    {
      "epoch": 0.05993508514516918,
      "grad_norm": 0.23338301479816437,
      "learning_rate": 0.000999962245046748,
      "loss": 1.9298,
      "step": 1348
    },
    {
      "epoch": 0.060024009603841535,
      "grad_norm": 0.2286379635334015,
      "learning_rate": 0.0009999618098402827,
      "loss": 1.9241,
      "step": 1350
    },
    {
      "epoch": 0.0601129340625139,
      "grad_norm": 0.23167788982391357,
      "learning_rate": 0.000999961372139925,
      "loss": 1.9308,
      "step": 1352
    },
    {
      "epoch": 0.06020185852118625,
      "grad_norm": 0.23941820859909058,
      "learning_rate": 0.0009999609319456767,
      "loss": 1.922,
      "step": 1354
    },
    {
      "epoch": 0.06029078297985861,
      "grad_norm": 0.23236848413944244,
      "learning_rate": 0.0009999604892575406,
      "loss": 1.9277,
      "step": 1356
    },
    {
      "epoch": 0.060379707438530966,
      "grad_norm": 0.2312156707048416,
      "learning_rate": 0.0009999600440755182,
      "loss": 1.9266,
      "step": 1358
    },
    {
      "epoch": 0.06046863189720333,
      "grad_norm": 0.2197667956352234,
      "learning_rate": 0.0009999595963996123,
      "loss": 1.9266,
      "step": 1360
    },
    {
      "epoch": 0.06055755635587568,
      "grad_norm": 0.22230570018291473,
      "learning_rate": 0.000999959146229825,
      "loss": 1.9171,
      "step": 1362
    },
    {
      "epoch": 0.06064648081454804,
      "grad_norm": 0.22791069746017456,
      "learning_rate": 0.0009999586935661585,
      "loss": 1.9194,
      "step": 1364
    },
    {
      "epoch": 0.0607354052732204,
      "grad_norm": 0.21917258203029633,
      "learning_rate": 0.000999958238408615,
      "loss": 1.9182,
      "step": 1366
    },
    {
      "epoch": 0.06082432973189276,
      "grad_norm": 0.2185303419828415,
      "learning_rate": 0.0009999577807571968,
      "loss": 1.9117,
      "step": 1368
    },
    {
      "epoch": 0.06091325419056511,
      "grad_norm": 0.21194784343242645,
      "learning_rate": 0.000999957320611906,
      "loss": 1.9163,
      "step": 1370
    },
    {
      "epoch": 0.06100217864923747,
      "grad_norm": 0.24148310720920563,
      "learning_rate": 0.0009999568579727456,
      "loss": 1.9209,
      "step": 1372
    },
    {
      "epoch": 0.06109110310790983,
      "grad_norm": 0.2278190702199936,
      "learning_rate": 0.0009999563928397171,
      "loss": 1.9118,
      "step": 1374
    },
    {
      "epoch": 0.06118002756658219,
      "grad_norm": 0.23531115055084229,
      "learning_rate": 0.000999955925212823,
      "loss": 1.9211,
      "step": 1376
    },
    {
      "epoch": 0.06126895202525454,
      "grad_norm": 0.23974572122097015,
      "learning_rate": 0.0009999554550920659,
      "loss": 1.9173,
      "step": 1378
    },
    {
      "epoch": 0.061357876483926904,
      "grad_norm": 0.2158331573009491,
      "learning_rate": 0.000999954982477448,
      "loss": 1.9132,
      "step": 1380
    },
    {
      "epoch": 0.061446800942599265,
      "grad_norm": 0.21482673287391663,
      "learning_rate": 0.0009999545073689717,
      "loss": 1.9197,
      "step": 1382
    },
    {
      "epoch": 0.06153572540127162,
      "grad_norm": 0.23469914495944977,
      "learning_rate": 0.0009999540297666394,
      "loss": 1.9115,
      "step": 1384
    },
    {
      "epoch": 0.06162464985994398,
      "grad_norm": 0.22895728051662445,
      "learning_rate": 0.0009999535496704533,
      "loss": 1.9155,
      "step": 1386
    },
    {
      "epoch": 0.061713574318616335,
      "grad_norm": 0.22930032014846802,
      "learning_rate": 0.0009999530670804157,
      "loss": 1.909,
      "step": 1388
    },
    {
      "epoch": 0.061802498777288696,
      "grad_norm": 0.21696867048740387,
      "learning_rate": 0.0009999525819965297,
      "loss": 1.9058,
      "step": 1390
    },
    {
      "epoch": 0.06189142323596105,
      "grad_norm": 0.22047823667526245,
      "learning_rate": 0.0009999520944187967,
      "loss": 1.907,
      "step": 1392
    },
    {
      "epoch": 0.06198034769463341,
      "grad_norm": 0.2223616987466812,
      "learning_rate": 0.00099995160434722,
      "loss": 1.9068,
      "step": 1394
    },
    {
      "epoch": 0.062069272153305766,
      "grad_norm": 0.22555837035179138,
      "learning_rate": 0.0009999511117818017,
      "loss": 1.9109,
      "step": 1396
    },
    {
      "epoch": 0.06215819661197813,
      "grad_norm": 0.20461831986904144,
      "learning_rate": 0.000999950616722544,
      "loss": 1.9028,
      "step": 1398
    },
    {
      "epoch": 0.06224712107065048,
      "grad_norm": 0.22076964378356934,
      "learning_rate": 0.00099995011916945,
      "loss": 1.8984,
      "step": 1400
    },
    {
      "epoch": 0.06233604552932284,
      "grad_norm": 0.20543377101421356,
      "learning_rate": 0.0009999496191225217,
      "loss": 1.9059,
      "step": 1402
    },
    {
      "epoch": 0.062424969987995196,
      "grad_norm": 0.20205794274806976,
      "learning_rate": 0.0009999491165817616,
      "loss": 1.91,
      "step": 1404
    },
    {
      "epoch": 0.06251389444666755,
      "grad_norm": 0.22159075736999512,
      "learning_rate": 0.0009999486115471723,
      "loss": 1.9041,
      "step": 1406
    },
    {
      "epoch": 0.06260281890533992,
      "grad_norm": 0.20887672901153564,
      "learning_rate": 0.0009999481040187566,
      "loss": 1.9044,
      "step": 1408
    },
    {
      "epoch": 0.06269174336401227,
      "grad_norm": 0.22122983634471893,
      "learning_rate": 0.0009999475939965164,
      "loss": 1.906,
      "step": 1410
    },
    {
      "epoch": 0.06278066782268463,
      "grad_norm": 0.2313462495803833,
      "learning_rate": 0.0009999470814804547,
      "loss": 1.9103,
      "step": 1412
    },
    {
      "epoch": 0.06286959228135698,
      "grad_norm": 0.22419576346874237,
      "learning_rate": 0.0009999465664705743,
      "loss": 1.9024,
      "step": 1414
    },
    {
      "epoch": 0.06295851674002935,
      "grad_norm": 0.21045628190040588,
      "learning_rate": 0.000999946048966877,
      "loss": 1.9055,
      "step": 1416
    },
    {
      "epoch": 0.0630474411987017,
      "grad_norm": 0.2150772362947464,
      "learning_rate": 0.0009999455289693663,
      "loss": 1.9068,
      "step": 1418
    },
    {
      "epoch": 0.06313636565737406,
      "grad_norm": 0.22035369277000427,
      "learning_rate": 0.000999945006478044,
      "loss": 1.9028,
      "step": 1420
    },
    {
      "epoch": 0.06322529011604641,
      "grad_norm": 0.22336137294769287,
      "learning_rate": 0.000999944481492913,
      "loss": 1.906,
      "step": 1422
    },
    {
      "epoch": 0.06331421457471878,
      "grad_norm": 0.22286519408226013,
      "learning_rate": 0.000999943954013976,
      "loss": 1.9031,
      "step": 1424
    },
    {
      "epoch": 0.06340313903339113,
      "grad_norm": 0.2087024748325348,
      "learning_rate": 0.0009999434240412356,
      "loss": 1.9006,
      "step": 1426
    },
    {
      "epoch": 0.06349206349206349,
      "grad_norm": 0.22912132740020752,
      "learning_rate": 0.0009999428915746943,
      "loss": 1.9031,
      "step": 1428
    },
    {
      "epoch": 0.06358098795073586,
      "grad_norm": 0.21829701960086823,
      "learning_rate": 0.000999942356614355,
      "loss": 1.8998,
      "step": 1430
    },
    {
      "epoch": 0.06366991240940821,
      "grad_norm": 0.22255627810955048,
      "learning_rate": 0.0009999418191602202,
      "loss": 1.8958,
      "step": 1432
    },
    {
      "epoch": 0.06375883686808057,
      "grad_norm": 0.2457275241613388,
      "learning_rate": 0.0009999412792122927,
      "loss": 1.8966,
      "step": 1434
    },
    {
      "epoch": 0.06384776132675292,
      "grad_norm": 0.22053119540214539,
      "learning_rate": 0.000999940736770575,
      "loss": 1.8982,
      "step": 1436
    },
    {
      "epoch": 0.06393668578542529,
      "grad_norm": 0.21990668773651123,
      "learning_rate": 0.00099994019183507,
      "loss": 1.8965,
      "step": 1438
    },
    {
      "epoch": 0.06402561024409764,
      "grad_norm": 0.21976561844348907,
      "learning_rate": 0.0009999396444057803,
      "loss": 1.8872,
      "step": 1440
    },
    {
      "epoch": 0.06411453470277,
      "grad_norm": 0.2285648137331009,
      "learning_rate": 0.0009999390944827086,
      "loss": 1.9001,
      "step": 1442
    },
    {
      "epoch": 0.06420345916144235,
      "grad_norm": 0.20996513962745667,
      "learning_rate": 0.0009999385420658578,
      "loss": 1.8845,
      "step": 1444
    },
    {
      "epoch": 0.06429238362011472,
      "grad_norm": 0.2080724537372589,
      "learning_rate": 0.0009999379871552304,
      "loss": 1.8932,
      "step": 1446
    },
    {
      "epoch": 0.06438130807878707,
      "grad_norm": 0.22280140221118927,
      "learning_rate": 0.0009999374297508295,
      "loss": 1.8875,
      "step": 1448
    },
    {
      "epoch": 0.06447023253745943,
      "grad_norm": 0.2291831374168396,
      "learning_rate": 0.0009999368698526577,
      "loss": 1.8935,
      "step": 1450
    },
    {
      "epoch": 0.06455915699613178,
      "grad_norm": 0.21631364524364471,
      "learning_rate": 0.0009999363074607178,
      "loss": 1.8842,
      "step": 1452
    },
    {
      "epoch": 0.06464808145480415,
      "grad_norm": 0.22055967152118683,
      "learning_rate": 0.0009999357425750127,
      "loss": 1.8922,
      "step": 1454
    },
    {
      "epoch": 0.0647370059134765,
      "grad_norm": 0.2344065010547638,
      "learning_rate": 0.000999935175195545,
      "loss": 1.8895,
      "step": 1456
    },
    {
      "epoch": 0.06482593037214886,
      "grad_norm": 0.2481888383626938,
      "learning_rate": 0.0009999346053223175,
      "loss": 1.8887,
      "step": 1458
    },
    {
      "epoch": 0.06491485483082121,
      "grad_norm": 0.21356667578220367,
      "learning_rate": 0.0009999340329553334,
      "loss": 1.8872,
      "step": 1460
    },
    {
      "epoch": 0.06500377928949358,
      "grad_norm": 0.22886356711387634,
      "learning_rate": 0.0009999334580945956,
      "loss": 1.8847,
      "step": 1462
    },
    {
      "epoch": 0.06509270374816593,
      "grad_norm": 0.2166658341884613,
      "learning_rate": 0.0009999328807401065,
      "loss": 1.8917,
      "step": 1464
    },
    {
      "epoch": 0.06518162820683829,
      "grad_norm": 0.217306986451149,
      "learning_rate": 0.000999932300891869,
      "loss": 1.8965,
      "step": 1466
    },
    {
      "epoch": 0.06527055266551064,
      "grad_norm": 0.20982764661312103,
      "learning_rate": 0.0009999317185498864,
      "loss": 1.8833,
      "step": 1468
    },
    {
      "epoch": 0.06535947712418301,
      "grad_norm": 0.2036888599395752,
      "learning_rate": 0.0009999311337141614,
      "loss": 1.8866,
      "step": 1470
    },
    {
      "epoch": 0.06544840158285536,
      "grad_norm": 0.20320679247379303,
      "learning_rate": 0.000999930546384697,
      "loss": 1.8832,
      "step": 1472
    },
    {
      "epoch": 0.06553732604152772,
      "grad_norm": 0.20226582884788513,
      "learning_rate": 0.000999929956561496,
      "loss": 1.8863,
      "step": 1474
    },
    {
      "epoch": 0.06562625050020009,
      "grad_norm": 0.19525253772735596,
      "learning_rate": 0.0009999293642445613,
      "loss": 1.8798,
      "step": 1476
    },
    {
      "epoch": 0.06571517495887244,
      "grad_norm": 0.20026175677776337,
      "learning_rate": 0.000999928769433896,
      "loss": 1.8813,
      "step": 1478
    },
    {
      "epoch": 0.0658040994175448,
      "grad_norm": 0.21397806704044342,
      "learning_rate": 0.0009999281721295031,
      "loss": 1.8751,
      "step": 1480
    },
    {
      "epoch": 0.06589302387621715,
      "grad_norm": 0.22252975404262543,
      "learning_rate": 0.0009999275723313853,
      "loss": 1.8884,
      "step": 1482
    },
    {
      "epoch": 0.06598194833488952,
      "grad_norm": 0.2118113934993744,
      "learning_rate": 0.0009999269700395458,
      "loss": 1.8773,
      "step": 1484
    },
    {
      "epoch": 0.06607087279356187,
      "grad_norm": 0.21514491736888885,
      "learning_rate": 0.0009999263652539878,
      "loss": 1.8834,
      "step": 1486
    },
    {
      "epoch": 0.06615979725223423,
      "grad_norm": 0.20763011276721954,
      "learning_rate": 0.000999925757974714,
      "loss": 1.876,
      "step": 1488
    },
    {
      "epoch": 0.06624872171090658,
      "grad_norm": 0.21687519550323486,
      "learning_rate": 0.0009999251482017276,
      "loss": 1.8788,
      "step": 1490
    },
    {
      "epoch": 0.06633764616957895,
      "grad_norm": 0.22184109687805176,
      "learning_rate": 0.0009999245359350315,
      "loss": 1.8724,
      "step": 1492
    },
    {
      "epoch": 0.0664265706282513,
      "grad_norm": 0.22225528955459595,
      "learning_rate": 0.0009999239211746288,
      "loss": 1.8765,
      "step": 1494
    },
    {
      "epoch": 0.06651549508692366,
      "grad_norm": 0.20817163586616516,
      "learning_rate": 0.0009999233039205226,
      "loss": 1.8764,
      "step": 1496
    },
    {
      "epoch": 0.06660441954559601,
      "grad_norm": 0.2178126573562622,
      "learning_rate": 0.000999922684172716,
      "loss": 1.886,
      "step": 1498
    },
    {
      "epoch": 0.06669334400426838,
      "grad_norm": 0.23448695242404938,
      "learning_rate": 0.0009999220619312122,
      "loss": 1.8872,
      "step": 1500
    },
    {
      "epoch": 0.06669334400426838,
      "eval_loss": 1.8269543647766113,
      "eval_runtime": 12.3418,
      "eval_samples_per_second": 559.886,
      "eval_steps_per_second": 70.006,
      "step": 1500
    },
    {
      "epoch": 0.06678226846294073,
      "grad_norm": 0.21950295567512512,
      "learning_rate": 0.000999921437196014,
      "loss": 1.8797,
      "step": 1502
    },
    {
      "epoch": 0.06687119292161309,
      "grad_norm": 0.21364814043045044,
      "learning_rate": 0.0009999208099671247,
      "loss": 1.8754,
      "step": 1504
    },
    {
      "epoch": 0.06696011738028544,
      "grad_norm": 0.20813779532909393,
      "learning_rate": 0.0009999201802445474,
      "loss": 1.8765,
      "step": 1506
    },
    {
      "epoch": 0.06704904183895781,
      "grad_norm": 0.2039484828710556,
      "learning_rate": 0.0009999195480282855,
      "loss": 1.8757,
      "step": 1508
    },
    {
      "epoch": 0.06713796629763016,
      "grad_norm": 0.22736653685569763,
      "learning_rate": 0.0009999189133183416,
      "loss": 1.8799,
      "step": 1510
    },
    {
      "epoch": 0.06722689075630252,
      "grad_norm": 0.20011046528816223,
      "learning_rate": 0.0009999182761147192,
      "loss": 1.8747,
      "step": 1512
    },
    {
      "epoch": 0.06731581521497487,
      "grad_norm": 0.1882735788822174,
      "learning_rate": 0.0009999176364174217,
      "loss": 1.866,
      "step": 1514
    },
    {
      "epoch": 0.06740473967364724,
      "grad_norm": 0.2117367535829544,
      "learning_rate": 0.0009999169942264518,
      "loss": 1.8735,
      "step": 1516
    },
    {
      "epoch": 0.0674936641323196,
      "grad_norm": 0.20652443170547485,
      "learning_rate": 0.000999916349541813,
      "loss": 1.8761,
      "step": 1518
    },
    {
      "epoch": 0.06758258859099195,
      "grad_norm": 0.20789290964603424,
      "learning_rate": 0.0009999157023635088,
      "loss": 1.8637,
      "step": 1520
    },
    {
      "epoch": 0.06767151304966432,
      "grad_norm": 0.19228237867355347,
      "learning_rate": 0.0009999150526915418,
      "loss": 1.8699,
      "step": 1522
    },
    {
      "epoch": 0.06776043750833667,
      "grad_norm": 0.20845188200473785,
      "learning_rate": 0.0009999144005259156,
      "loss": 1.8741,
      "step": 1524
    },
    {
      "epoch": 0.06784936196700903,
      "grad_norm": 0.20529267191886902,
      "learning_rate": 0.0009999137458666333,
      "loss": 1.868,
      "step": 1526
    },
    {
      "epoch": 0.06793828642568138,
      "grad_norm": 0.20668277144432068,
      "learning_rate": 0.0009999130887136983,
      "loss": 1.8751,
      "step": 1528
    },
    {
      "epoch": 0.06802721088435375,
      "grad_norm": 0.21079662442207336,
      "learning_rate": 0.0009999124290671138,
      "loss": 1.8683,
      "step": 1530
    },
    {
      "epoch": 0.0681161353430261,
      "grad_norm": 0.21383321285247803,
      "learning_rate": 0.0009999117669268831,
      "loss": 1.8794,
      "step": 1532
    },
    {
      "epoch": 0.06820505980169846,
      "grad_norm": 0.20688050985336304,
      "learning_rate": 0.0009999111022930096,
      "loss": 1.8711,
      "step": 1534
    },
    {
      "epoch": 0.06829398426037081,
      "grad_norm": 0.2304670512676239,
      "learning_rate": 0.0009999104351654964,
      "loss": 1.8715,
      "step": 1536
    },
    {
      "epoch": 0.06838290871904318,
      "grad_norm": 0.22473397850990295,
      "learning_rate": 0.0009999097655443473,
      "loss": 1.8662,
      "step": 1538
    },
    {
      "epoch": 0.06847183317771553,
      "grad_norm": 0.2325805276632309,
      "learning_rate": 0.0009999090934295649,
      "loss": 1.8631,
      "step": 1540
    },
    {
      "epoch": 0.06856075763638789,
      "grad_norm": 0.21634715795516968,
      "learning_rate": 0.0009999084188211532,
      "loss": 1.8677,
      "step": 1542
    },
    {
      "epoch": 0.06864968209506024,
      "grad_norm": 0.21803300082683563,
      "learning_rate": 0.0009999077417191153,
      "loss": 1.8657,
      "step": 1544
    },
    {
      "epoch": 0.06873860655373261,
      "grad_norm": 0.20068170130252838,
      "learning_rate": 0.0009999070621234546,
      "loss": 1.862,
      "step": 1546
    },
    {
      "epoch": 0.06882753101240496,
      "grad_norm": 0.19963085651397705,
      "learning_rate": 0.0009999063800341745,
      "loss": 1.868,
      "step": 1548
    },
    {
      "epoch": 0.06891645547107732,
      "grad_norm": 0.19617609679698944,
      "learning_rate": 0.0009999056954512783,
      "loss": 1.8689,
      "step": 1550
    },
    {
      "epoch": 0.06900537992974967,
      "grad_norm": 0.20461757481098175,
      "learning_rate": 0.0009999050083747696,
      "loss": 1.8631,
      "step": 1552
    },
    {
      "epoch": 0.06909430438842204,
      "grad_norm": 0.21538877487182617,
      "learning_rate": 0.0009999043188046518,
      "loss": 1.8703,
      "step": 1554
    },
    {
      "epoch": 0.0691832288470944,
      "grad_norm": 0.20673204958438873,
      "learning_rate": 0.0009999036267409284,
      "loss": 1.8687,
      "step": 1556
    },
    {
      "epoch": 0.06927215330576675,
      "grad_norm": 0.24245448410511017,
      "learning_rate": 0.0009999029321836024,
      "loss": 1.8601,
      "step": 1558
    },
    {
      "epoch": 0.0693610777644391,
      "grad_norm": 0.2253614217042923,
      "learning_rate": 0.0009999022351326776,
      "loss": 1.8716,
      "step": 1560
    },
    {
      "epoch": 0.06945000222311147,
      "grad_norm": 0.21160924434661865,
      "learning_rate": 0.0009999015355881577,
      "loss": 1.8608,
      "step": 1562
    },
    {
      "epoch": 0.06953892668178382,
      "grad_norm": 0.20701120793819427,
      "learning_rate": 0.000999900833550046,
      "loss": 1.8621,
      "step": 1564
    },
    {
      "epoch": 0.06962785114045618,
      "grad_norm": 0.210323303937912,
      "learning_rate": 0.0009999001290183457,
      "loss": 1.8667,
      "step": 1566
    },
    {
      "epoch": 0.06971677559912855,
      "grad_norm": 0.21616849303245544,
      "learning_rate": 0.0009998994219930606,
      "loss": 1.8677,
      "step": 1568
    },
    {
      "epoch": 0.0698057000578009,
      "grad_norm": 0.21297746896743774,
      "learning_rate": 0.0009998987124741944,
      "loss": 1.869,
      "step": 1570
    },
    {
      "epoch": 0.06989462451647326,
      "grad_norm": 0.2072431594133377,
      "learning_rate": 0.0009998980004617504,
      "loss": 1.8584,
      "step": 1572
    },
    {
      "epoch": 0.06998354897514561,
      "grad_norm": 0.2180069535970688,
      "learning_rate": 0.000999897285955732,
      "loss": 1.8578,
      "step": 1574
    },
    {
      "epoch": 0.07007247343381798,
      "grad_norm": 0.23733854293823242,
      "learning_rate": 0.0009998965689561432,
      "loss": 1.8605,
      "step": 1576
    },
    {
      "epoch": 0.07016139789249033,
      "grad_norm": 0.19760969281196594,
      "learning_rate": 0.0009998958494629874,
      "loss": 1.8575,
      "step": 1578
    },
    {
      "epoch": 0.07025032235116269,
      "grad_norm": 0.21075817942619324,
      "learning_rate": 0.0009998951274762678,
      "loss": 1.87,
      "step": 1580
    },
    {
      "epoch": 0.07033924680983504,
      "grad_norm": 0.2083798050880432,
      "learning_rate": 0.0009998944029959884,
      "loss": 1.8528,
      "step": 1582
    },
    {
      "epoch": 0.07042817126850741,
      "grad_norm": 0.2075483202934265,
      "learning_rate": 0.000999893676022153,
      "loss": 1.8584,
      "step": 1584
    },
    {
      "epoch": 0.07051709572717976,
      "grad_norm": 0.2059869021177292,
      "learning_rate": 0.0009998929465547647,
      "loss": 1.8558,
      "step": 1586
    },
    {
      "epoch": 0.07060602018585212,
      "grad_norm": 0.2247011363506317,
      "learning_rate": 0.0009998922145938276,
      "loss": 1.8625,
      "step": 1588
    },
    {
      "epoch": 0.07069494464452447,
      "grad_norm": 0.21843503415584564,
      "learning_rate": 0.0009998914801393449,
      "loss": 1.8617,
      "step": 1590
    },
    {
      "epoch": 0.07078386910319684,
      "grad_norm": 0.1983787566423416,
      "learning_rate": 0.0009998907431913206,
      "loss": 1.8538,
      "step": 1592
    },
    {
      "epoch": 0.0708727935618692,
      "grad_norm": 0.19233089685440063,
      "learning_rate": 0.0009998900037497586,
      "loss": 1.8622,
      "step": 1594
    },
    {
      "epoch": 0.07096171802054155,
      "grad_norm": 0.22240038216114044,
      "learning_rate": 0.000999889261814662,
      "loss": 1.8567,
      "step": 1596
    },
    {
      "epoch": 0.0710506424792139,
      "grad_norm": 0.20009489357471466,
      "learning_rate": 0.0009998885173860348,
      "loss": 1.8603,
      "step": 1598
    },
    {
      "epoch": 0.07113956693788627,
      "grad_norm": 0.21011285483837128,
      "learning_rate": 0.0009998877704638807,
      "loss": 1.8492,
      "step": 1600
    },
    {
      "epoch": 0.07122849139655862,
      "grad_norm": 0.2004971206188202,
      "learning_rate": 0.0009998870210482034,
      "loss": 1.851,
      "step": 1602
    },
    {
      "epoch": 0.07131741585523098,
      "grad_norm": 0.20903609693050385,
      "learning_rate": 0.0009998862691390066,
      "loss": 1.8618,
      "step": 1604
    },
    {
      "epoch": 0.07140634031390333,
      "grad_norm": 0.21211233735084534,
      "learning_rate": 0.0009998855147362944,
      "loss": 1.8508,
      "step": 1606
    },
    {
      "epoch": 0.0714952647725757,
      "grad_norm": 0.19902560114860535,
      "learning_rate": 0.0009998847578400701,
      "loss": 1.852,
      "step": 1608
    },
    {
      "epoch": 0.07158418923124805,
      "grad_norm": 0.19855988025665283,
      "learning_rate": 0.0009998839984503376,
      "loss": 1.8467,
      "step": 1610
    },
    {
      "epoch": 0.07167311368992041,
      "grad_norm": 0.1922021359205246,
      "learning_rate": 0.000999883236567101,
      "loss": 1.8498,
      "step": 1612
    },
    {
      "epoch": 0.07176203814859278,
      "grad_norm": 0.18983250856399536,
      "learning_rate": 0.0009998824721903635,
      "loss": 1.8572,
      "step": 1614
    },
    {
      "epoch": 0.07185096260726513,
      "grad_norm": 0.20658980309963226,
      "learning_rate": 0.0009998817053201295,
      "loss": 1.8525,
      "step": 1616
    },
    {
      "epoch": 0.07193988706593749,
      "grad_norm": 0.21265624463558197,
      "learning_rate": 0.0009998809359564023,
      "loss": 1.8482,
      "step": 1618
    },
    {
      "epoch": 0.07202881152460984,
      "grad_norm": 0.21423251926898956,
      "learning_rate": 0.0009998801640991861,
      "loss": 1.8608,
      "step": 1620
    },
    {
      "epoch": 0.07211773598328221,
      "grad_norm": 0.20722512900829315,
      "learning_rate": 0.000999879389748485,
      "loss": 1.8522,
      "step": 1622
    },
    {
      "epoch": 0.07220666044195456,
      "grad_norm": 0.20465902984142303,
      "learning_rate": 0.0009998786129043022,
      "loss": 1.8502,
      "step": 1624
    },
    {
      "epoch": 0.07229558490062692,
      "grad_norm": 0.20114745199680328,
      "learning_rate": 0.000999877833566642,
      "loss": 1.8477,
      "step": 1626
    },
    {
      "epoch": 0.07238450935929927,
      "grad_norm": 0.2192513346672058,
      "learning_rate": 0.0009998770517355082,
      "loss": 1.8445,
      "step": 1628
    },
    {
      "epoch": 0.07247343381797164,
      "grad_norm": 0.2060314267873764,
      "learning_rate": 0.0009998762674109046,
      "loss": 1.8545,
      "step": 1630
    },
    {
      "epoch": 0.07256235827664399,
      "grad_norm": 0.20097291469573975,
      "learning_rate": 0.0009998754805928354,
      "loss": 1.8489,
      "step": 1632
    },
    {
      "epoch": 0.07265128273531635,
      "grad_norm": 0.21980713307857513,
      "learning_rate": 0.000999874691281304,
      "loss": 1.8508,
      "step": 1634
    },
    {
      "epoch": 0.0727402071939887,
      "grad_norm": 0.20329026877880096,
      "learning_rate": 0.000999873899476315,
      "loss": 1.8407,
      "step": 1636
    },
    {
      "epoch": 0.07282913165266107,
      "grad_norm": 0.20128802955150604,
      "learning_rate": 0.0009998731051778717,
      "loss": 1.8521,
      "step": 1638
    },
    {
      "epoch": 0.07291805611133342,
      "grad_norm": 0.20749594271183014,
      "learning_rate": 0.0009998723083859786,
      "loss": 1.8492,
      "step": 1640
    },
    {
      "epoch": 0.07300698057000578,
      "grad_norm": 0.20556269586086273,
      "learning_rate": 0.0009998715091006393,
      "loss": 1.839,
      "step": 1642
    },
    {
      "epoch": 0.07309590502867813,
      "grad_norm": 0.20244896411895752,
      "learning_rate": 0.0009998707073218583,
      "loss": 1.8373,
      "step": 1644
    },
    {
      "epoch": 0.0731848294873505,
      "grad_norm": 0.2151784747838974,
      "learning_rate": 0.0009998699030496388,
      "loss": 1.8453,
      "step": 1646
    },
    {
      "epoch": 0.07327375394602285,
      "grad_norm": 0.21280355751514435,
      "learning_rate": 0.0009998690962839856,
      "loss": 1.8447,
      "step": 1648
    },
    {
      "epoch": 0.07336267840469521,
      "grad_norm": 0.20697881281375885,
      "learning_rate": 0.000999868287024902,
      "loss": 1.8475,
      "step": 1650
    },
    {
      "epoch": 0.07345160286336756,
      "grad_norm": 0.19814634323120117,
      "learning_rate": 0.0009998674752723926,
      "loss": 1.8451,
      "step": 1652
    },
    {
      "epoch": 0.07354052732203993,
      "grad_norm": 0.192799910902977,
      "learning_rate": 0.0009998666610264613,
      "loss": 1.837,
      "step": 1654
    },
    {
      "epoch": 0.07362945178071229,
      "grad_norm": 0.2121358960866928,
      "learning_rate": 0.0009998658442871122,
      "loss": 1.8423,
      "step": 1656
    },
    {
      "epoch": 0.07371837623938464,
      "grad_norm": 0.21163220703601837,
      "learning_rate": 0.000999865025054349,
      "loss": 1.839,
      "step": 1658
    },
    {
      "epoch": 0.07380730069805701,
      "grad_norm": 0.2047502100467682,
      "learning_rate": 0.0009998642033281764,
      "loss": 1.8361,
      "step": 1660
    },
    {
      "epoch": 0.07389622515672936,
      "grad_norm": 0.2158685177564621,
      "learning_rate": 0.000999863379108598,
      "loss": 1.8407,
      "step": 1662
    },
    {
      "epoch": 0.07398514961540172,
      "grad_norm": 0.21105967462062836,
      "learning_rate": 0.0009998625523956182,
      "loss": 1.8494,
      "step": 1664
    },
    {
      "epoch": 0.07407407407407407,
      "grad_norm": 0.2116025984287262,
      "learning_rate": 0.000999861723189241,
      "loss": 1.8446,
      "step": 1666
    },
    {
      "epoch": 0.07416299853274644,
      "grad_norm": 0.19938991963863373,
      "learning_rate": 0.0009998608914894706,
      "loss": 1.8381,
      "step": 1668
    },
    {
      "epoch": 0.07425192299141879,
      "grad_norm": 0.20230786502361298,
      "learning_rate": 0.0009998600572963107,
      "loss": 1.8361,
      "step": 1670
    },
    {
      "epoch": 0.07434084745009115,
      "grad_norm": 0.20542068779468536,
      "learning_rate": 0.0009998592206097662,
      "loss": 1.8327,
      "step": 1672
    },
    {
      "epoch": 0.0744297719087635,
      "grad_norm": 0.20988449454307556,
      "learning_rate": 0.000999858381429841,
      "loss": 1.8391,
      "step": 1674
    },
    {
      "epoch": 0.07451869636743587,
      "grad_norm": 0.20840682089328766,
      "learning_rate": 0.0009998575397565392,
      "loss": 1.8346,
      "step": 1676
    },
    {
      "epoch": 0.07460762082610822,
      "grad_norm": 0.1958722621202469,
      "learning_rate": 0.000999856695589865,
      "loss": 1.8342,
      "step": 1678
    },
    {
      "epoch": 0.07469654528478058,
      "grad_norm": 0.20700527727603912,
      "learning_rate": 0.0009998558489298225,
      "loss": 1.8323,
      "step": 1680
    },
    {
      "epoch": 0.07478546974345293,
      "grad_norm": 0.20546066761016846,
      "learning_rate": 0.000999854999776416,
      "loss": 1.8312,
      "step": 1682
    },
    {
      "epoch": 0.0748743942021253,
      "grad_norm": 0.2187362015247345,
      "learning_rate": 0.00099985414812965,
      "loss": 1.8375,
      "step": 1684
    },
    {
      "epoch": 0.07496331866079765,
      "grad_norm": 0.1964145451784134,
      "learning_rate": 0.0009998532939895285,
      "loss": 1.8339,
      "step": 1686
    },
    {
      "epoch": 0.07505224311947001,
      "grad_norm": 0.20831523835659027,
      "learning_rate": 0.0009998524373560557,
      "loss": 1.8381,
      "step": 1688
    },
    {
      "epoch": 0.07514116757814236,
      "grad_norm": 0.21572189033031464,
      "learning_rate": 0.0009998515782292361,
      "loss": 1.8409,
      "step": 1690
    },
    {
      "epoch": 0.07523009203681473,
      "grad_norm": 0.21819041669368744,
      "learning_rate": 0.0009998507166090738,
      "loss": 1.8326,
      "step": 1692
    },
    {
      "epoch": 0.07531901649548708,
      "grad_norm": 0.20171982049942017,
      "learning_rate": 0.000999849852495573,
      "loss": 1.8375,
      "step": 1694
    },
    {
      "epoch": 0.07540794095415944,
      "grad_norm": 0.20552539825439453,
      "learning_rate": 0.0009998489858887383,
      "loss": 1.8351,
      "step": 1696
    },
    {
      "epoch": 0.07549686541283179,
      "grad_norm": 0.19860213994979858,
      "learning_rate": 0.000999848116788574,
      "loss": 1.8321,
      "step": 1698
    },
    {
      "epoch": 0.07558578987150416,
      "grad_norm": 0.20631815493106842,
      "learning_rate": 0.0009998472451950842,
      "loss": 1.8372,
      "step": 1700
    },
    {
      "epoch": 0.07567471433017652,
      "grad_norm": 0.2032255083322525,
      "learning_rate": 0.0009998463711082732,
      "loss": 1.8302,
      "step": 1702
    },
    {
      "epoch": 0.07576363878884887,
      "grad_norm": 0.22408516705036163,
      "learning_rate": 0.000999845494528146,
      "loss": 1.8288,
      "step": 1704
    },
    {
      "epoch": 0.07585256324752124,
      "grad_norm": 0.2116139978170395,
      "learning_rate": 0.000999844615454706,
      "loss": 1.8269,
      "step": 1706
    },
    {
      "epoch": 0.07594148770619359,
      "grad_norm": 0.21274049580097198,
      "learning_rate": 0.0009998437338879583,
      "loss": 1.8308,
      "step": 1708
    },
    {
      "epoch": 0.07603041216486595,
      "grad_norm": 0.2098899781703949,
      "learning_rate": 0.0009998428498279072,
      "loss": 1.8334,
      "step": 1710
    },
    {
      "epoch": 0.0761193366235383,
      "grad_norm": 0.20081061124801636,
      "learning_rate": 0.0009998419632745567,
      "loss": 1.8261,
      "step": 1712
    },
    {
      "epoch": 0.07620826108221067,
      "grad_norm": 0.19824229180812836,
      "learning_rate": 0.0009998410742279118,
      "loss": 1.8391,
      "step": 1714
    },
    {
      "epoch": 0.07629718554088302,
      "grad_norm": 0.19863907992839813,
      "learning_rate": 0.0009998401826879766,
      "loss": 1.8336,
      "step": 1716
    },
    {
      "epoch": 0.07638610999955538,
      "grad_norm": 0.19055074453353882,
      "learning_rate": 0.0009998392886547557,
      "loss": 1.8333,
      "step": 1718
    },
    {
      "epoch": 0.07647503445822773,
      "grad_norm": 0.20255160331726074,
      "learning_rate": 0.0009998383921282531,
      "loss": 1.8275,
      "step": 1720
    },
    {
      "epoch": 0.0765639589169001,
      "grad_norm": 0.19958724081516266,
      "learning_rate": 0.000999837493108474,
      "loss": 1.8305,
      "step": 1722
    },
    {
      "epoch": 0.07665288337557245,
      "grad_norm": 0.2084696739912033,
      "learning_rate": 0.0009998365915954222,
      "loss": 1.8254,
      "step": 1724
    },
    {
      "epoch": 0.07674180783424481,
      "grad_norm": 0.19476568698883057,
      "learning_rate": 0.0009998356875891028,
      "loss": 1.8294,
      "step": 1726
    },
    {
      "epoch": 0.07683073229291716,
      "grad_norm": 0.20423997938632965,
      "learning_rate": 0.00099983478108952,
      "loss": 1.8298,
      "step": 1728
    },
    {
      "epoch": 0.07691965675158953,
      "grad_norm": 0.20384973287582397,
      "learning_rate": 0.000999833872096678,
      "loss": 1.8219,
      "step": 1730
    },
    {
      "epoch": 0.07700858121026188,
      "grad_norm": 0.22047263383865356,
      "learning_rate": 0.000999832960610582,
      "loss": 1.8308,
      "step": 1732
    },
    {
      "epoch": 0.07709750566893424,
      "grad_norm": 0.2119770050048828,
      "learning_rate": 0.000999832046631236,
      "loss": 1.827,
      "step": 1734
    },
    {
      "epoch": 0.07718643012760659,
      "grad_norm": 0.20165476202964783,
      "learning_rate": 0.000999831130158645,
      "loss": 1.8247,
      "step": 1736
    },
    {
      "epoch": 0.07727535458627896,
      "grad_norm": 0.20627140998840332,
      "learning_rate": 0.0009998302111928132,
      "loss": 1.8295,
      "step": 1738
    },
    {
      "epoch": 0.07736427904495131,
      "grad_norm": 0.19049489498138428,
      "learning_rate": 0.0009998292897337455,
      "loss": 1.8265,
      "step": 1740
    },
    {
      "epoch": 0.07745320350362367,
      "grad_norm": 0.2037346363067627,
      "learning_rate": 0.0009998283657814463,
      "loss": 1.8253,
      "step": 1742
    },
    {
      "epoch": 0.07754212796229602,
      "grad_norm": 0.19989527761936188,
      "learning_rate": 0.00099982743933592,
      "loss": 1.8227,
      "step": 1744
    },
    {
      "epoch": 0.07763105242096839,
      "grad_norm": 0.19065894186496735,
      "learning_rate": 0.0009998265103971717,
      "loss": 1.8203,
      "step": 1746
    },
    {
      "epoch": 0.07771997687964075,
      "grad_norm": 0.20793575048446655,
      "learning_rate": 0.0009998255789652058,
      "loss": 1.8259,
      "step": 1748
    },
    {
      "epoch": 0.0778089013383131,
      "grad_norm": 0.19519171118736267,
      "learning_rate": 0.000999824645040027,
      "loss": 1.8263,
      "step": 1750
    },
    {
      "epoch": 0.07789782579698547,
      "grad_norm": 0.18882858753204346,
      "learning_rate": 0.0009998237086216398,
      "loss": 1.8275,
      "step": 1752
    },
    {
      "epoch": 0.07798675025565782,
      "grad_norm": 0.19698387384414673,
      "learning_rate": 0.000999822769710049,
      "loss": 1.8215,
      "step": 1754
    },
    {
      "epoch": 0.07807567471433018,
      "grad_norm": 0.1997542679309845,
      "learning_rate": 0.0009998218283052591,
      "loss": 1.8214,
      "step": 1756
    },
    {
      "epoch": 0.07816459917300253,
      "grad_norm": 0.20124317705631256,
      "learning_rate": 0.000999820884407275,
      "loss": 1.8274,
      "step": 1758
    },
    {
      "epoch": 0.0782535236316749,
      "grad_norm": 0.2041187584400177,
      "learning_rate": 0.0009998199380161015,
      "loss": 1.8218,
      "step": 1760
    },
    {
      "epoch": 0.07834244809034725,
      "grad_norm": 0.18576522171497345,
      "learning_rate": 0.0009998189891317433,
      "loss": 1.8191,
      "step": 1762
    },
    {
      "epoch": 0.0784313725490196,
      "grad_norm": 0.17914514243602753,
      "learning_rate": 0.0009998180377542047,
      "loss": 1.8233,
      "step": 1764
    },
    {
      "epoch": 0.07852029700769196,
      "grad_norm": 0.20891062915325165,
      "learning_rate": 0.000999817083883491,
      "loss": 1.8232,
      "step": 1766
    },
    {
      "epoch": 0.07860922146636433,
      "grad_norm": 0.19134004414081573,
      "learning_rate": 0.0009998161275196068,
      "loss": 1.8189,
      "step": 1768
    },
    {
      "epoch": 0.07869814592503668,
      "grad_norm": 0.18456409871578217,
      "learning_rate": 0.0009998151686625566,
      "loss": 1.8209,
      "step": 1770
    },
    {
      "epoch": 0.07878707038370904,
      "grad_norm": 0.1992667019367218,
      "learning_rate": 0.0009998142073123454,
      "loss": 1.8167,
      "step": 1772
    },
    {
      "epoch": 0.07887599484238139,
      "grad_norm": 0.21500921249389648,
      "learning_rate": 0.000999813243468978,
      "loss": 1.8153,
      "step": 1774
    },
    {
      "epoch": 0.07896491930105376,
      "grad_norm": 0.2065228521823883,
      "learning_rate": 0.000999812277132459,
      "loss": 1.8155,
      "step": 1776
    },
    {
      "epoch": 0.07905384375972611,
      "grad_norm": 0.20505499839782715,
      "learning_rate": 0.0009998113083027936,
      "loss": 1.8183,
      "step": 1778
    },
    {
      "epoch": 0.07914276821839847,
      "grad_norm": 0.1991395652294159,
      "learning_rate": 0.0009998103369799863,
      "loss": 1.8083,
      "step": 1780
    },
    {
      "epoch": 0.07923169267707082,
      "grad_norm": 0.19899892807006836,
      "learning_rate": 0.0009998093631640422,
      "loss": 1.8152,
      "step": 1782
    },
    {
      "epoch": 0.07932061713574319,
      "grad_norm": 0.21442580223083496,
      "learning_rate": 0.0009998083868549658,
      "loss": 1.8218,
      "step": 1784
    },
    {
      "epoch": 0.07940954159441554,
      "grad_norm": 0.20083315670490265,
      "learning_rate": 0.0009998074080527625,
      "loss": 1.8157,
      "step": 1786
    },
    {
      "epoch": 0.0794984660530879,
      "grad_norm": 0.19688989222049713,
      "learning_rate": 0.000999806426757437,
      "loss": 1.8132,
      "step": 1788
    },
    {
      "epoch": 0.07958739051176025,
      "grad_norm": 0.1846112161874771,
      "learning_rate": 0.0009998054429689936,
      "loss": 1.8219,
      "step": 1790
    },
    {
      "epoch": 0.07967631497043262,
      "grad_norm": 0.1728316843509674,
      "learning_rate": 0.000999804456687438,
      "loss": 1.8128,
      "step": 1792
    },
    {
      "epoch": 0.07976523942910498,
      "grad_norm": 0.18066954612731934,
      "learning_rate": 0.0009998034679127748,
      "loss": 1.8129,
      "step": 1794
    },
    {
      "epoch": 0.07985416388777733,
      "grad_norm": 0.1870201677083969,
      "learning_rate": 0.000999802476645009,
      "loss": 1.8113,
      "step": 1796
    },
    {
      "epoch": 0.0799430883464497,
      "grad_norm": 0.2108452171087265,
      "learning_rate": 0.0009998014828841452,
      "loss": 1.8179,
      "step": 1798
    },
    {
      "epoch": 0.08003201280512205,
      "grad_norm": 0.19203181564807892,
      "learning_rate": 0.000999800486630189,
      "loss": 1.8089,
      "step": 1800
    },
    {
      "epoch": 0.0801209372637944,
      "grad_norm": 0.1975417286157608,
      "learning_rate": 0.0009997994878831447,
      "loss": 1.8162,
      "step": 1802
    },
    {
      "epoch": 0.08020986172246676,
      "grad_norm": 0.19258712232112885,
      "learning_rate": 0.0009997984866430176,
      "loss": 1.8176,
      "step": 1804
    },
    {
      "epoch": 0.08029878618113913,
      "grad_norm": 0.1956678330898285,
      "learning_rate": 0.0009997974829098128,
      "loss": 1.8093,
      "step": 1806
    },
    {
      "epoch": 0.08038771063981148,
      "grad_norm": 0.19371239840984344,
      "learning_rate": 0.0009997964766835352,
      "loss": 1.8101,
      "step": 1808
    },
    {
      "epoch": 0.08047663509848384,
      "grad_norm": 0.18697892129421234,
      "learning_rate": 0.00099979546796419,
      "loss": 1.8084,
      "step": 1810
    },
    {
      "epoch": 0.08056555955715619,
      "grad_norm": 0.1840774416923523,
      "learning_rate": 0.0009997944567517816,
      "loss": 1.8017,
      "step": 1812
    },
    {
      "epoch": 0.08065448401582856,
      "grad_norm": 0.1876949518918991,
      "learning_rate": 0.000999793443046316,
      "loss": 1.8098,
      "step": 1814
    },
    {
      "epoch": 0.08074340847450091,
      "grad_norm": 0.19937770068645477,
      "learning_rate": 0.0009997924268477973,
      "loss": 1.8146,
      "step": 1816
    },
    {
      "epoch": 0.08083233293317327,
      "grad_norm": 0.18788865208625793,
      "learning_rate": 0.0009997914081562311,
      "loss": 1.8212,
      "step": 1818
    },
    {
      "epoch": 0.08092125739184562,
      "grad_norm": 0.17869558930397034,
      "learning_rate": 0.0009997903869716225,
      "loss": 1.8126,
      "step": 1820
    },
    {
      "epoch": 0.08101018185051799,
      "grad_norm": 0.19068653881549835,
      "learning_rate": 0.0009997893632939766,
      "loss": 1.8133,
      "step": 1822
    },
    {
      "epoch": 0.08109910630919034,
      "grad_norm": 0.18737611174583435,
      "learning_rate": 0.0009997883371232982,
      "loss": 1.8099,
      "step": 1824
    },
    {
      "epoch": 0.0811880307678627,
      "grad_norm": 0.19693921506404877,
      "learning_rate": 0.0009997873084595927,
      "loss": 1.8155,
      "step": 1826
    },
    {
      "epoch": 0.08127695522653505,
      "grad_norm": 0.19971245527267456,
      "learning_rate": 0.000999786277302865,
      "loss": 1.8165,
      "step": 1828
    },
    {
      "epoch": 0.08136587968520742,
      "grad_norm": 0.19381284713745117,
      "learning_rate": 0.0009997852436531205,
      "loss": 1.8095,
      "step": 1830
    },
    {
      "epoch": 0.08145480414387977,
      "grad_norm": 0.18925556540489197,
      "learning_rate": 0.0009997842075103642,
      "loss": 1.8148,
      "step": 1832
    },
    {
      "epoch": 0.08154372860255213,
      "grad_norm": 0.19684088230133057,
      "learning_rate": 0.0009997831688746016,
      "loss": 1.8153,
      "step": 1834
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 0.1987828016281128,
      "learning_rate": 0.0009997821277458372,
      "loss": 1.8126,
      "step": 1836
    },
    {
      "epoch": 0.08172157751989685,
      "grad_norm": 0.19290605187416077,
      "learning_rate": 0.0009997810841240768,
      "loss": 1.8085,
      "step": 1838
    },
    {
      "epoch": 0.0818105019785692,
      "grad_norm": 0.19497103989124298,
      "learning_rate": 0.0009997800380093253,
      "loss": 1.81,
      "step": 1840
    },
    {
      "epoch": 0.08189942643724156,
      "grad_norm": 0.20089541375637054,
      "learning_rate": 0.000999778989401588,
      "loss": 1.809,
      "step": 1842
    },
    {
      "epoch": 0.08198835089591393,
      "grad_norm": 0.19982276856899261,
      "learning_rate": 0.0009997779383008702,
      "loss": 1.8178,
      "step": 1844
    },
    {
      "epoch": 0.08207727535458628,
      "grad_norm": 0.20949578285217285,
      "learning_rate": 0.0009997768847071773,
      "loss": 1.8105,
      "step": 1846
    },
    {
      "epoch": 0.08216619981325864,
      "grad_norm": 0.20833182334899902,
      "learning_rate": 0.000999775828620514,
      "loss": 1.8152,
      "step": 1848
    },
    {
      "epoch": 0.08225512427193099,
      "grad_norm": 0.21134817600250244,
      "learning_rate": 0.000999774770040886,
      "loss": 1.805,
      "step": 1850
    },
    {
      "epoch": 0.08234404873060336,
      "grad_norm": 0.21612076461315155,
      "learning_rate": 0.0009997737089682986,
      "loss": 1.8034,
      "step": 1852
    },
    {
      "epoch": 0.08243297318927571,
      "grad_norm": 0.18973609805107117,
      "learning_rate": 0.000999772645402757,
      "loss": 1.8057,
      "step": 1854
    },
    {
      "epoch": 0.08252189764794807,
      "grad_norm": 0.19349431991577148,
      "learning_rate": 0.0009997715793442663,
      "loss": 1.807,
      "step": 1856
    },
    {
      "epoch": 0.08261082210662042,
      "grad_norm": 0.19564490020275116,
      "learning_rate": 0.0009997705107928322,
      "loss": 1.8048,
      "step": 1858
    },
    {
      "epoch": 0.08269974656529279,
      "grad_norm": 0.18188372254371643,
      "learning_rate": 0.0009997694397484596,
      "loss": 1.8138,
      "step": 1860
    },
    {
      "epoch": 0.08278867102396514,
      "grad_norm": 0.18335072696208954,
      "learning_rate": 0.0009997683662111541,
      "loss": 1.8099,
      "step": 1862
    },
    {
      "epoch": 0.0828775954826375,
      "grad_norm": 0.18123309314250946,
      "learning_rate": 0.000999767290180921,
      "loss": 1.7991,
      "step": 1864
    },
    {
      "epoch": 0.08296651994130985,
      "grad_norm": 0.1729104071855545,
      "learning_rate": 0.000999766211657766,
      "loss": 1.8033,
      "step": 1866
    },
    {
      "epoch": 0.08305544439998222,
      "grad_norm": 0.1979542374610901,
      "learning_rate": 0.000999765130641694,
      "loss": 1.8014,
      "step": 1868
    },
    {
      "epoch": 0.08314436885865457,
      "grad_norm": 0.19370914995670319,
      "learning_rate": 0.0009997640471327107,
      "loss": 1.8032,
      "step": 1870
    },
    {
      "epoch": 0.08323329331732693,
      "grad_norm": 0.19891789555549622,
      "learning_rate": 0.0009997629611308212,
      "loss": 1.8034,
      "step": 1872
    },
    {
      "epoch": 0.08332221777599928,
      "grad_norm": 0.1980421394109726,
      "learning_rate": 0.0009997618726360312,
      "loss": 1.8026,
      "step": 1874
    },
    {
      "epoch": 0.08341114223467165,
      "grad_norm": 0.18609797954559326,
      "learning_rate": 0.000999760781648346,
      "loss": 1.8015,
      "step": 1876
    },
    {
      "epoch": 0.083500066693344,
      "grad_norm": 0.18574324250221252,
      "learning_rate": 0.000999759688167771,
      "loss": 1.8089,
      "step": 1878
    },
    {
      "epoch": 0.08358899115201636,
      "grad_norm": 0.1960146427154541,
      "learning_rate": 0.0009997585921943117,
      "loss": 1.7994,
      "step": 1880
    },
    {
      "epoch": 0.08367791561068871,
      "grad_norm": 0.19020849466323853,
      "learning_rate": 0.0009997574937279736,
      "loss": 1.8026,
      "step": 1882
    },
    {
      "epoch": 0.08376684006936108,
      "grad_norm": 0.18406033515930176,
      "learning_rate": 0.0009997563927687623,
      "loss": 1.804,
      "step": 1884
    },
    {
      "epoch": 0.08385576452803344,
      "grad_norm": 0.1974494904279709,
      "learning_rate": 0.000999755289316683,
      "loss": 1.8038,
      "step": 1886
    },
    {
      "epoch": 0.08394468898670579,
      "grad_norm": 0.1874603033065796,
      "learning_rate": 0.0009997541833717415,
      "loss": 1.7976,
      "step": 1888
    },
    {
      "epoch": 0.08403361344537816,
      "grad_norm": 0.1843891739845276,
      "learning_rate": 0.000999753074933943,
      "loss": 1.8011,
      "step": 1890
    },
    {
      "epoch": 0.08412253790405051,
      "grad_norm": 0.17537960410118103,
      "learning_rate": 0.0009997519640032935,
      "loss": 1.8076,
      "step": 1892
    },
    {
      "epoch": 0.08421146236272287,
      "grad_norm": 0.17957009375095367,
      "learning_rate": 0.000999750850579798,
      "loss": 1.7974,
      "step": 1894
    },
    {
      "epoch": 0.08430038682139522,
      "grad_norm": 0.18118196725845337,
      "learning_rate": 0.0009997497346634623,
      "loss": 1.7928,
      "step": 1896
    },
    {
      "epoch": 0.08438931128006759,
      "grad_norm": 0.19623412191867828,
      "learning_rate": 0.0009997486162542921,
      "loss": 1.7973,
      "step": 1898
    },
    {
      "epoch": 0.08447823573873994,
      "grad_norm": 0.20058177411556244,
      "learning_rate": 0.0009997474953522929,
      "loss": 1.799,
      "step": 1900
    },
    {
      "epoch": 0.0845671601974123,
      "grad_norm": 0.19219110906124115,
      "learning_rate": 0.00099974637195747,
      "loss": 1.8008,
      "step": 1902
    },
    {
      "epoch": 0.08465608465608465,
      "grad_norm": 0.18925371766090393,
      "learning_rate": 0.0009997452460698292,
      "loss": 1.7954,
      "step": 1904
    },
    {
      "epoch": 0.08474500911475702,
      "grad_norm": 0.183928444981575,
      "learning_rate": 0.0009997441176893764,
      "loss": 1.8056,
      "step": 1906
    },
    {
      "epoch": 0.08483393357342937,
      "grad_norm": 0.18831098079681396,
      "learning_rate": 0.000999742986816117,
      "loss": 1.799,
      "step": 1908
    },
    {
      "epoch": 0.08492285803210173,
      "grad_norm": 0.19508256018161774,
      "learning_rate": 0.0009997418534500567,
      "loss": 1.8007,
      "step": 1910
    },
    {
      "epoch": 0.08501178249077408,
      "grad_norm": 0.188640758395195,
      "learning_rate": 0.0009997407175912007,
      "loss": 1.7926,
      "step": 1912
    },
    {
      "epoch": 0.08510070694944645,
      "grad_norm": 0.18694116175174713,
      "learning_rate": 0.0009997395792395553,
      "loss": 1.7915,
      "step": 1914
    },
    {
      "epoch": 0.0851896314081188,
      "grad_norm": 0.19411993026733398,
      "learning_rate": 0.000999738438395126,
      "loss": 1.7981,
      "step": 1916
    },
    {
      "epoch": 0.08527855586679116,
      "grad_norm": 0.2087039351463318,
      "learning_rate": 0.0009997372950579183,
      "loss": 1.7954,
      "step": 1918
    },
    {
      "epoch": 0.08536748032546351,
      "grad_norm": 0.20052039623260498,
      "learning_rate": 0.000999736149227938,
      "loss": 1.7936,
      "step": 1920
    },
    {
      "epoch": 0.08545640478413588,
      "grad_norm": 0.21173499524593353,
      "learning_rate": 0.000999735000905191,
      "loss": 1.8015,
      "step": 1922
    },
    {
      "epoch": 0.08554532924280824,
      "grad_norm": 0.17747758328914642,
      "learning_rate": 0.0009997338500896827,
      "loss": 1.7894,
      "step": 1924
    },
    {
      "epoch": 0.08563425370148059,
      "grad_norm": 0.1921745240688324,
      "learning_rate": 0.000999732696781419,
      "loss": 1.7967,
      "step": 1926
    },
    {
      "epoch": 0.08572317816015294,
      "grad_norm": 0.18577106297016144,
      "learning_rate": 0.0009997315409804057,
      "loss": 1.7903,
      "step": 1928
    },
    {
      "epoch": 0.08581210261882531,
      "grad_norm": 0.1862160861492157,
      "learning_rate": 0.0009997303826866485,
      "loss": 1.7978,
      "step": 1930
    },
    {
      "epoch": 0.08590102707749767,
      "grad_norm": 0.18871815502643585,
      "learning_rate": 0.0009997292219001532,
      "loss": 1.7985,
      "step": 1932
    },
    {
      "epoch": 0.08598995153617002,
      "grad_norm": 0.18492014706134796,
      "learning_rate": 0.0009997280586209257,
      "loss": 1.795,
      "step": 1934
    },
    {
      "epoch": 0.08607887599484239,
      "grad_norm": 0.17806781828403473,
      "learning_rate": 0.0009997268928489717,
      "loss": 1.7889,
      "step": 1936
    },
    {
      "epoch": 0.08616780045351474,
      "grad_norm": 0.18848101794719696,
      "learning_rate": 0.0009997257245842968,
      "loss": 1.7931,
      "step": 1938
    },
    {
      "epoch": 0.0862567249121871,
      "grad_norm": 0.18113906681537628,
      "learning_rate": 0.000999724553826907,
      "loss": 1.7927,
      "step": 1940
    },
    {
      "epoch": 0.08634564937085945,
      "grad_norm": 0.19498959183692932,
      "learning_rate": 0.0009997233805768085,
      "loss": 1.7927,
      "step": 1942
    },
    {
      "epoch": 0.08643457382953182,
      "grad_norm": 0.19100996851921082,
      "learning_rate": 0.0009997222048340065,
      "loss": 1.7918,
      "step": 1944
    },
    {
      "epoch": 0.08652349828820417,
      "grad_norm": 0.19174836575984955,
      "learning_rate": 0.0009997210265985073,
      "loss": 1.7889,
      "step": 1946
    },
    {
      "epoch": 0.08661242274687653,
      "grad_norm": 0.20086248219013214,
      "learning_rate": 0.0009997198458703168,
      "loss": 1.7958,
      "step": 1948
    },
    {
      "epoch": 0.08670134720554888,
      "grad_norm": 0.18942195177078247,
      "learning_rate": 0.0009997186626494407,
      "loss": 1.7884,
      "step": 1950
    },
    {
      "epoch": 0.08679027166422125,
      "grad_norm": 0.17851027846336365,
      "learning_rate": 0.000999717476935885,
      "loss": 1.7867,
      "step": 1952
    },
    {
      "epoch": 0.0868791961228936,
      "grad_norm": 0.1856180876493454,
      "learning_rate": 0.0009997162887296553,
      "loss": 1.7886,
      "step": 1954
    },
    {
      "epoch": 0.08696812058156596,
      "grad_norm": 0.18615032732486725,
      "learning_rate": 0.0009997150980307582,
      "loss": 1.7869,
      "step": 1956
    },
    {
      "epoch": 0.08705704504023831,
      "grad_norm": 0.18606998026371002,
      "learning_rate": 0.000999713904839199,
      "loss": 1.7897,
      "step": 1958
    },
    {
      "epoch": 0.08714596949891068,
      "grad_norm": 0.19354918599128723,
      "learning_rate": 0.000999712709154984,
      "loss": 1.7938,
      "step": 1960
    },
    {
      "epoch": 0.08723489395758303,
      "grad_norm": 0.19163134694099426,
      "learning_rate": 0.0009997115109781189,
      "loss": 1.7934,
      "step": 1962
    },
    {
      "epoch": 0.08732381841625539,
      "grad_norm": 0.17643679678440094,
      "learning_rate": 0.00099971031030861,
      "loss": 1.7914,
      "step": 1964
    },
    {
      "epoch": 0.08741274287492774,
      "grad_norm": 0.19244788587093353,
      "learning_rate": 0.000999709107146463,
      "loss": 1.7927,
      "step": 1966
    },
    {
      "epoch": 0.08750166733360011,
      "grad_norm": 0.1964365541934967,
      "learning_rate": 0.000999707901491684,
      "loss": 1.7919,
      "step": 1968
    },
    {
      "epoch": 0.08759059179227247,
      "grad_norm": 0.19603557884693146,
      "learning_rate": 0.0009997066933442793,
      "loss": 1.7924,
      "step": 1970
    },
    {
      "epoch": 0.08767951625094482,
      "grad_norm": 0.1899658441543579,
      "learning_rate": 0.0009997054827042544,
      "loss": 1.7871,
      "step": 1972
    },
    {
      "epoch": 0.08776844070961717,
      "grad_norm": 0.1820671409368515,
      "learning_rate": 0.0009997042695716158,
      "loss": 1.7918,
      "step": 1974
    },
    {
      "epoch": 0.08785736516828954,
      "grad_norm": 0.1810377836227417,
      "learning_rate": 0.000999703053946369,
      "loss": 1.7916,
      "step": 1976
    },
    {
      "epoch": 0.0879462896269619,
      "grad_norm": 0.182911679148674,
      "learning_rate": 0.000999701835828521,
      "loss": 1.787,
      "step": 1978
    },
    {
      "epoch": 0.08803521408563425,
      "grad_norm": 0.17979872226715088,
      "learning_rate": 0.000999700615218077,
      "loss": 1.7915,
      "step": 1980
    },
    {
      "epoch": 0.08812413854430662,
      "grad_norm": 0.18918099999427795,
      "learning_rate": 0.0009996993921150433,
      "loss": 1.7764,
      "step": 1982
    },
    {
      "epoch": 0.08821306300297897,
      "grad_norm": 0.18799486756324768,
      "learning_rate": 0.000999698166519426,
      "loss": 1.7828,
      "step": 1984
    },
    {
      "epoch": 0.08830198746165133,
      "grad_norm": 0.1944817304611206,
      "learning_rate": 0.0009996969384312316,
      "loss": 1.7836,
      "step": 1986
    },
    {
      "epoch": 0.08839091192032368,
      "grad_norm": 0.17584578692913055,
      "learning_rate": 0.0009996957078504658,
      "loss": 1.7905,
      "step": 1988
    },
    {
      "epoch": 0.08847983637899605,
      "grad_norm": 0.1802207976579666,
      "learning_rate": 0.000999694474777135,
      "loss": 1.7863,
      "step": 1990
    },
    {
      "epoch": 0.0885687608376684,
      "grad_norm": 0.18599218130111694,
      "learning_rate": 0.0009996932392112448,
      "loss": 1.7833,
      "step": 1992
    },
    {
      "epoch": 0.08865768529634076,
      "grad_norm": 0.1814405471086502,
      "learning_rate": 0.0009996920011528022,
      "loss": 1.7845,
      "step": 1994
    },
    {
      "epoch": 0.08874660975501311,
      "grad_norm": 0.1817612200975418,
      "learning_rate": 0.0009996907606018126,
      "loss": 1.7884,
      "step": 1996
    },
    {
      "epoch": 0.08883553421368548,
      "grad_norm": 0.19590908288955688,
      "learning_rate": 0.0009996895175582827,
      "loss": 1.7754,
      "step": 1998
    },
    {
      "epoch": 0.08892445867235783,
      "grad_norm": 0.17747312784194946,
      "learning_rate": 0.0009996882720222186,
      "loss": 1.7815,
      "step": 2000
    },
    {
      "epoch": 0.08892445867235783,
      "eval_loss": 1.7405998706817627,
      "eval_runtime": 12.3499,
      "eval_samples_per_second": 559.518,
      "eval_steps_per_second": 69.96,
      "step": 2000
    },
    {
      "epoch": 0.08901338313103019,
      "grad_norm": 0.18658480048179626,
      "learning_rate": 0.0009996870239936265,
      "loss": 1.7879,
      "step": 2002
    },
    {
      "epoch": 0.08910230758970254,
      "grad_norm": 0.18885967135429382,
      "learning_rate": 0.0009996857734725125,
      "loss": 1.7915,
      "step": 2004
    },
    {
      "epoch": 0.08919123204837491,
      "grad_norm": 0.19159406423568726,
      "learning_rate": 0.0009996845204588828,
      "loss": 1.7847,
      "step": 2006
    },
    {
      "epoch": 0.08928015650704726,
      "grad_norm": 0.18446260690689087,
      "learning_rate": 0.000999683264952744,
      "loss": 1.7862,
      "step": 2008
    },
    {
      "epoch": 0.08936908096571962,
      "grad_norm": 0.18793538212776184,
      "learning_rate": 0.000999682006954102,
      "loss": 1.7804,
      "step": 2010
    },
    {
      "epoch": 0.08945800542439197,
      "grad_norm": 0.18664175271987915,
      "learning_rate": 0.0009996807464629632,
      "loss": 1.7826,
      "step": 2012
    },
    {
      "epoch": 0.08954692988306434,
      "grad_norm": 0.19432403147220612,
      "learning_rate": 0.0009996794834793339,
      "loss": 1.7839,
      "step": 2014
    },
    {
      "epoch": 0.0896358543417367,
      "grad_norm": 0.18868719041347504,
      "learning_rate": 0.0009996782180032204,
      "loss": 1.7919,
      "step": 2016
    },
    {
      "epoch": 0.08972477880040905,
      "grad_norm": 0.19868771731853485,
      "learning_rate": 0.0009996769500346292,
      "loss": 1.7812,
      "step": 2018
    },
    {
      "epoch": 0.0898137032590814,
      "grad_norm": 0.2011130154132843,
      "learning_rate": 0.0009996756795735663,
      "loss": 1.7797,
      "step": 2020
    },
    {
      "epoch": 0.08990262771775377,
      "grad_norm": 0.19635726511478424,
      "learning_rate": 0.0009996744066200379,
      "loss": 1.7816,
      "step": 2022
    },
    {
      "epoch": 0.08999155217642613,
      "grad_norm": 0.17903649806976318,
      "learning_rate": 0.000999673131174051,
      "loss": 1.783,
      "step": 2024
    },
    {
      "epoch": 0.09008047663509848,
      "grad_norm": 0.18663077056407928,
      "learning_rate": 0.0009996718532356112,
      "loss": 1.7857,
      "step": 2026
    },
    {
      "epoch": 0.09016940109377085,
      "grad_norm": 0.1833386868238449,
      "learning_rate": 0.0009996705728047256,
      "loss": 1.7836,
      "step": 2028
    },
    {
      "epoch": 0.0902583255524432,
      "grad_norm": 0.19123496115207672,
      "learning_rate": 0.0009996692898814003,
      "loss": 1.7755,
      "step": 2030
    },
    {
      "epoch": 0.09034725001111556,
      "grad_norm": 0.1912190318107605,
      "learning_rate": 0.0009996680044656412,
      "loss": 1.7766,
      "step": 2032
    },
    {
      "epoch": 0.09043617446978791,
      "grad_norm": 0.18371352553367615,
      "learning_rate": 0.0009996667165574555,
      "loss": 1.7774,
      "step": 2034
    },
    {
      "epoch": 0.09052509892846028,
      "grad_norm": 0.18822483718395233,
      "learning_rate": 0.0009996654261568492,
      "loss": 1.7775,
      "step": 2036
    },
    {
      "epoch": 0.09061402338713263,
      "grad_norm": 0.17826074361801147,
      "learning_rate": 0.0009996641332638287,
      "loss": 1.7792,
      "step": 2038
    },
    {
      "epoch": 0.09070294784580499,
      "grad_norm": 0.1842014193534851,
      "learning_rate": 0.0009996628378784007,
      "loss": 1.7845,
      "step": 2040
    },
    {
      "epoch": 0.09079187230447734,
      "grad_norm": 0.1995820701122284,
      "learning_rate": 0.0009996615400005716,
      "loss": 1.7765,
      "step": 2042
    },
    {
      "epoch": 0.09088079676314971,
      "grad_norm": 0.19667015969753265,
      "learning_rate": 0.0009996602396303476,
      "loss": 1.771,
      "step": 2044
    },
    {
      "epoch": 0.09096972122182206,
      "grad_norm": 0.2025892585515976,
      "learning_rate": 0.0009996589367677355,
      "loss": 1.7775,
      "step": 2046
    },
    {
      "epoch": 0.09105864568049442,
      "grad_norm": 0.1986626386642456,
      "learning_rate": 0.0009996576314127417,
      "loss": 1.78,
      "step": 2048
    },
    {
      "epoch": 0.09114757013916677,
      "grad_norm": 0.20763103663921356,
      "learning_rate": 0.0009996563235653727,
      "loss": 1.7747,
      "step": 2050
    },
    {
      "epoch": 0.09123649459783914,
      "grad_norm": 0.1954943984746933,
      "learning_rate": 0.000999655013225635,
      "loss": 1.7739,
      "step": 2052
    },
    {
      "epoch": 0.0913254190565115,
      "grad_norm": 0.18959374725818634,
      "learning_rate": 0.000999653700393535,
      "loss": 1.7786,
      "step": 2054
    },
    {
      "epoch": 0.09141434351518385,
      "grad_norm": 0.17867428064346313,
      "learning_rate": 0.0009996523850690794,
      "loss": 1.7786,
      "step": 2056
    },
    {
      "epoch": 0.0915032679738562,
      "grad_norm": 0.19500364363193512,
      "learning_rate": 0.000999651067252275,
      "loss": 1.7711,
      "step": 2058
    },
    {
      "epoch": 0.09159219243252857,
      "grad_norm": 0.18299424648284912,
      "learning_rate": 0.000999649746943128,
      "loss": 1.7742,
      "step": 2060
    },
    {
      "epoch": 0.09168111689120093,
      "grad_norm": 0.21155034005641937,
      "learning_rate": 0.0009996484241416453,
      "loss": 1.771,
      "step": 2062
    },
    {
      "epoch": 0.09177004134987328,
      "grad_norm": 0.19016344845294952,
      "learning_rate": 0.000999647098847833,
      "loss": 1.7853,
      "step": 2064
    },
    {
      "epoch": 0.09185896580854563,
      "grad_norm": 0.19537319242954254,
      "learning_rate": 0.0009996457710616984,
      "loss": 1.7822,
      "step": 2066
    },
    {
      "epoch": 0.091947890267218,
      "grad_norm": 0.1982775330543518,
      "learning_rate": 0.0009996444407832474,
      "loss": 1.7752,
      "step": 2068
    },
    {
      "epoch": 0.09203681472589036,
      "grad_norm": 0.1813453882932663,
      "learning_rate": 0.0009996431080124873,
      "loss": 1.7731,
      "step": 2070
    },
    {
      "epoch": 0.09212573918456271,
      "grad_norm": 0.18596340715885162,
      "learning_rate": 0.0009996417727494242,
      "loss": 1.7777,
      "step": 2072
    },
    {
      "epoch": 0.09221466364323508,
      "grad_norm": 0.17686715722084045,
      "learning_rate": 0.000999640434994065,
      "loss": 1.7787,
      "step": 2074
    },
    {
      "epoch": 0.09230358810190743,
      "grad_norm": 0.1908753216266632,
      "learning_rate": 0.0009996390947464167,
      "loss": 1.7713,
      "step": 2076
    },
    {
      "epoch": 0.09239251256057979,
      "grad_norm": 0.1761447936296463,
      "learning_rate": 0.0009996377520064853,
      "loss": 1.7727,
      "step": 2078
    },
    {
      "epoch": 0.09248143701925214,
      "grad_norm": 0.18040457367897034,
      "learning_rate": 0.000999636406774278,
      "loss": 1.7783,
      "step": 2080
    },
    {
      "epoch": 0.09257036147792451,
      "grad_norm": 0.17208784818649292,
      "learning_rate": 0.0009996350590498015,
      "loss": 1.7716,
      "step": 2082
    },
    {
      "epoch": 0.09265928593659686,
      "grad_norm": 0.19492624700069427,
      "learning_rate": 0.0009996337088330623,
      "loss": 1.7755,
      "step": 2084
    },
    {
      "epoch": 0.09274821039526922,
      "grad_norm": 0.1833573430776596,
      "learning_rate": 0.000999632356124067,
      "loss": 1.7769,
      "step": 2086
    },
    {
      "epoch": 0.09283713485394157,
      "grad_norm": 0.1697169542312622,
      "learning_rate": 0.000999631000922823,
      "loss": 1.7732,
      "step": 2088
    },
    {
      "epoch": 0.09292605931261394,
      "grad_norm": 0.17216144502162933,
      "learning_rate": 0.000999629643229336,
      "loss": 1.7658,
      "step": 2090
    },
    {
      "epoch": 0.0930149837712863,
      "grad_norm": 0.17798788845539093,
      "learning_rate": 0.0009996282830436142,
      "loss": 1.7753,
      "step": 2092
    },
    {
      "epoch": 0.09310390822995865,
      "grad_norm": 0.17972996830940247,
      "learning_rate": 0.000999626920365663,
      "loss": 1.7783,
      "step": 2094
    },
    {
      "epoch": 0.093192832688631,
      "grad_norm": 0.1812281608581543,
      "learning_rate": 0.0009996255551954901,
      "loss": 1.7725,
      "step": 2096
    },
    {
      "epoch": 0.09328175714730337,
      "grad_norm": 0.18036170303821564,
      "learning_rate": 0.0009996241875331016,
      "loss": 1.77,
      "step": 2098
    },
    {
      "epoch": 0.09337068160597572,
      "grad_norm": 0.1780519038438797,
      "learning_rate": 0.000999622817378505,
      "loss": 1.773,
      "step": 2100
    },
    {
      "epoch": 0.09345960606464808,
      "grad_norm": 0.17772600054740906,
      "learning_rate": 0.000999621444731707,
      "loss": 1.7757,
      "step": 2102
    },
    {
      "epoch": 0.09354853052332043,
      "grad_norm": 0.1823798418045044,
      "learning_rate": 0.000999620069592714,
      "loss": 1.7643,
      "step": 2104
    },
    {
      "epoch": 0.0936374549819928,
      "grad_norm": 0.18164494633674622,
      "learning_rate": 0.0009996186919615333,
      "loss": 1.7707,
      "step": 2106
    },
    {
      "epoch": 0.09372637944066516,
      "grad_norm": 0.17441585659980774,
      "learning_rate": 0.0009996173118381718,
      "loss": 1.7731,
      "step": 2108
    },
    {
      "epoch": 0.09381530389933751,
      "grad_norm": 0.18183459341526031,
      "learning_rate": 0.0009996159292226358,
      "loss": 1.7637,
      "step": 2110
    },
    {
      "epoch": 0.09390422835800986,
      "grad_norm": 0.17130981385707855,
      "learning_rate": 0.000999614544114933,
      "loss": 1.7723,
      "step": 2112
    },
    {
      "epoch": 0.09399315281668223,
      "grad_norm": 0.18831345438957214,
      "learning_rate": 0.00099961315651507,
      "loss": 1.7687,
      "step": 2114
    },
    {
      "epoch": 0.09408207727535459,
      "grad_norm": 0.17629346251487732,
      "learning_rate": 0.0009996117664230534,
      "loss": 1.7694,
      "step": 2116
    },
    {
      "epoch": 0.09417100173402694,
      "grad_norm": 0.17761678993701935,
      "learning_rate": 0.0009996103738388901,
      "loss": 1.7664,
      "step": 2118
    },
    {
      "epoch": 0.09425992619269931,
      "grad_norm": 0.17654430866241455,
      "learning_rate": 0.0009996089787625878,
      "loss": 1.7683,
      "step": 2120
    },
    {
      "epoch": 0.09434885065137166,
      "grad_norm": 0.16883589327335358,
      "learning_rate": 0.0009996075811941527,
      "loss": 1.7703,
      "step": 2122
    },
    {
      "epoch": 0.09443777511004402,
      "grad_norm": 0.1744341105222702,
      "learning_rate": 0.0009996061811335922,
      "loss": 1.7722,
      "step": 2124
    },
    {
      "epoch": 0.09452669956871637,
      "grad_norm": 0.1795886754989624,
      "learning_rate": 0.0009996047785809131,
      "loss": 1.7625,
      "step": 2126
    },
    {
      "epoch": 0.09461562402738874,
      "grad_norm": 0.18121252954006195,
      "learning_rate": 0.0009996033735361226,
      "loss": 1.7678,
      "step": 2128
    },
    {
      "epoch": 0.0947045484860611,
      "grad_norm": 0.17234009504318237,
      "learning_rate": 0.0009996019659992274,
      "loss": 1.7731,
      "step": 2130
    },
    {
      "epoch": 0.09479347294473345,
      "grad_norm": 0.18222364783287048,
      "learning_rate": 0.0009996005559702345,
      "loss": 1.7666,
      "step": 2132
    },
    {
      "epoch": 0.0948823974034058,
      "grad_norm": 0.18646016716957092,
      "learning_rate": 0.0009995991434491513,
      "loss": 1.7708,
      "step": 2134
    },
    {
      "epoch": 0.09497132186207817,
      "grad_norm": 0.16806988418102264,
      "learning_rate": 0.0009995977284359846,
      "loss": 1.7684,
      "step": 2136
    },
    {
      "epoch": 0.09506024632075052,
      "grad_norm": 0.17961154878139496,
      "learning_rate": 0.0009995963109307414,
      "loss": 1.7631,
      "step": 2138
    },
    {
      "epoch": 0.09514917077942288,
      "grad_norm": 0.18496328592300415,
      "learning_rate": 0.000999594890933429,
      "loss": 1.7682,
      "step": 2140
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 0.17461057007312775,
      "learning_rate": 0.0009995934684440544,
      "loss": 1.7697,
      "step": 2142
    },
    {
      "epoch": 0.0953270196967676,
      "grad_norm": 0.17441897094249725,
      "learning_rate": 0.0009995920434626244,
      "loss": 1.7668,
      "step": 2144
    },
    {
      "epoch": 0.09541594415543996,
      "grad_norm": 0.183127298951149,
      "learning_rate": 0.0009995906159891465,
      "loss": 1.7578,
      "step": 2146
    },
    {
      "epoch": 0.09550486861411231,
      "grad_norm": 0.1828969568014145,
      "learning_rate": 0.0009995891860236277,
      "loss": 1.7704,
      "step": 2148
    },
    {
      "epoch": 0.09559379307278466,
      "grad_norm": 0.17396622896194458,
      "learning_rate": 0.0009995877535660751,
      "loss": 1.7636,
      "step": 2150
    },
    {
      "epoch": 0.09568271753145703,
      "grad_norm": 0.17213039100170135,
      "learning_rate": 0.000999586318616496,
      "loss": 1.7667,
      "step": 2152
    },
    {
      "epoch": 0.09577164199012939,
      "grad_norm": 0.16386523842811584,
      "learning_rate": 0.0009995848811748973,
      "loss": 1.7696,
      "step": 2154
    },
    {
      "epoch": 0.09586056644880174,
      "grad_norm": 0.16317681968212128,
      "learning_rate": 0.0009995834412412862,
      "loss": 1.7689,
      "step": 2156
    },
    {
      "epoch": 0.0959494909074741,
      "grad_norm": 0.16907323896884918,
      "learning_rate": 0.00099958199881567,
      "loss": 1.7692,
      "step": 2158
    },
    {
      "epoch": 0.09603841536614646,
      "grad_norm": 0.1724671572446823,
      "learning_rate": 0.000999580553898056,
      "loss": 1.7614,
      "step": 2160
    },
    {
      "epoch": 0.09612733982481882,
      "grad_norm": 0.17434446513652802,
      "learning_rate": 0.000999579106488451,
      "loss": 1.762,
      "step": 2162
    },
    {
      "epoch": 0.09621626428349117,
      "grad_norm": 0.17296640574932098,
      "learning_rate": 0.0009995776565868628,
      "loss": 1.7674,
      "step": 2164
    },
    {
      "epoch": 0.09630518874216354,
      "grad_norm": 0.1922890841960907,
      "learning_rate": 0.0009995762041932982,
      "loss": 1.7709,
      "step": 2166
    },
    {
      "epoch": 0.0963941132008359,
      "grad_norm": 0.15948012471199036,
      "learning_rate": 0.0009995747493077645,
      "loss": 1.7622,
      "step": 2168
    },
    {
      "epoch": 0.09648303765950825,
      "grad_norm": 0.1770942509174347,
      "learning_rate": 0.0009995732919302691,
      "loss": 1.7594,
      "step": 2170
    },
    {
      "epoch": 0.0965719621181806,
      "grad_norm": 0.17478297650814056,
      "learning_rate": 0.0009995718320608192,
      "loss": 1.76,
      "step": 2172
    },
    {
      "epoch": 0.09666088657685297,
      "grad_norm": 0.1589890867471695,
      "learning_rate": 0.000999570369699422,
      "loss": 1.7549,
      "step": 2174
    },
    {
      "epoch": 0.09674981103552532,
      "grad_norm": 0.17255498468875885,
      "learning_rate": 0.000999568904846085,
      "loss": 1.7638,
      "step": 2176
    },
    {
      "epoch": 0.09683873549419768,
      "grad_norm": 0.17751558125019073,
      "learning_rate": 0.0009995674375008153,
      "loss": 1.7597,
      "step": 2178
    },
    {
      "epoch": 0.09692765995287003,
      "grad_norm": 0.16570937633514404,
      "learning_rate": 0.0009995659676636205,
      "loss": 1.7608,
      "step": 2180
    },
    {
      "epoch": 0.0970165844115424,
      "grad_norm": 0.16247688233852386,
      "learning_rate": 0.0009995644953345072,
      "loss": 1.7596,
      "step": 2182
    },
    {
      "epoch": 0.09710550887021475,
      "grad_norm": 0.16889050602912903,
      "learning_rate": 0.000999563020513484,
      "loss": 1.7627,
      "step": 2184
    },
    {
      "epoch": 0.09719443332888711,
      "grad_norm": 0.16616292297840118,
      "learning_rate": 0.000999561543200557,
      "loss": 1.7563,
      "step": 2186
    },
    {
      "epoch": 0.09728335778755946,
      "grad_norm": 0.18441399931907654,
      "learning_rate": 0.0009995600633957342,
      "loss": 1.7615,
      "step": 2188
    },
    {
      "epoch": 0.09737228224623183,
      "grad_norm": 0.17722539603710175,
      "learning_rate": 0.0009995585810990229,
      "loss": 1.7568,
      "step": 2190
    },
    {
      "epoch": 0.09746120670490419,
      "grad_norm": 0.17260760068893433,
      "learning_rate": 0.0009995570963104304,
      "loss": 1.758,
      "step": 2192
    },
    {
      "epoch": 0.09755013116357654,
      "grad_norm": 0.17014533281326294,
      "learning_rate": 0.0009995556090299643,
      "loss": 1.7556,
      "step": 2194
    },
    {
      "epoch": 0.0976390556222489,
      "grad_norm": 0.1753702163696289,
      "learning_rate": 0.000999554119257632,
      "loss": 1.7636,
      "step": 2196
    },
    {
      "epoch": 0.09772798008092126,
      "grad_norm": 0.1720827966928482,
      "learning_rate": 0.0009995526269934406,
      "loss": 1.7559,
      "step": 2198
    },
    {
      "epoch": 0.09781690453959362,
      "grad_norm": 0.16954128444194794,
      "learning_rate": 0.000999551132237398,
      "loss": 1.7599,
      "step": 2200
    },
    {
      "epoch": 0.09790582899826597,
      "grad_norm": 0.1615387201309204,
      "learning_rate": 0.0009995496349895112,
      "loss": 1.7527,
      "step": 2202
    },
    {
      "epoch": 0.09799475345693832,
      "grad_norm": 0.1675902158021927,
      "learning_rate": 0.000999548135249788,
      "loss": 1.7502,
      "step": 2204
    },
    {
      "epoch": 0.09808367791561069,
      "grad_norm": 0.15957540273666382,
      "learning_rate": 0.0009995466330182357,
      "loss": 1.7642,
      "step": 2206
    },
    {
      "epoch": 0.09817260237428305,
      "grad_norm": 0.1576092690229416,
      "learning_rate": 0.000999545128294862,
      "loss": 1.7599,
      "step": 2208
    },
    {
      "epoch": 0.0982615268329554,
      "grad_norm": 0.17128455638885498,
      "learning_rate": 0.0009995436210796743,
      "loss": 1.754,
      "step": 2210
    },
    {
      "epoch": 0.09835045129162777,
      "grad_norm": 0.16489969193935394,
      "learning_rate": 0.00099954211137268,
      "loss": 1.754,
      "step": 2212
    },
    {
      "epoch": 0.09843937575030012,
      "grad_norm": 0.16618114709854126,
      "learning_rate": 0.0009995405991738869,
      "loss": 1.7533,
      "step": 2214
    },
    {
      "epoch": 0.09852830020897248,
      "grad_norm": 0.17597708106040955,
      "learning_rate": 0.000999539084483302,
      "loss": 1.7581,
      "step": 2216
    },
    {
      "epoch": 0.09861722466764483,
      "grad_norm": 0.1678888350725174,
      "learning_rate": 0.0009995375673009337,
      "loss": 1.7543,
      "step": 2218
    },
    {
      "epoch": 0.0987061491263172,
      "grad_norm": 0.1661226749420166,
      "learning_rate": 0.0009995360476267887,
      "loss": 1.753,
      "step": 2220
    },
    {
      "epoch": 0.09879507358498955,
      "grad_norm": 0.17764368653297424,
      "learning_rate": 0.0009995345254608752,
      "loss": 1.7608,
      "step": 2222
    },
    {
      "epoch": 0.09888399804366191,
      "grad_norm": 0.16960063576698303,
      "learning_rate": 0.0009995330008032005,
      "loss": 1.7567,
      "step": 2224
    },
    {
      "epoch": 0.09897292250233426,
      "grad_norm": 0.16657236218452454,
      "learning_rate": 0.0009995314736537724,
      "loss": 1.7529,
      "step": 2226
    },
    {
      "epoch": 0.09906184696100663,
      "grad_norm": 0.17307406663894653,
      "learning_rate": 0.0009995299440125982,
      "loss": 1.7584,
      "step": 2228
    },
    {
      "epoch": 0.09915077141967898,
      "grad_norm": 0.18175233900547028,
      "learning_rate": 0.0009995284118796857,
      "loss": 1.7607,
      "step": 2230
    },
    {
      "epoch": 0.09923969587835134,
      "grad_norm": 0.18064169585704803,
      "learning_rate": 0.0009995268772550428,
      "loss": 1.7559,
      "step": 2232
    },
    {
      "epoch": 0.0993286203370237,
      "grad_norm": 0.18270429968833923,
      "learning_rate": 0.0009995253401386768,
      "loss": 1.7546,
      "step": 2234
    },
    {
      "epoch": 0.09941754479569606,
      "grad_norm": 0.17944471538066864,
      "learning_rate": 0.0009995238005305954,
      "loss": 1.7585,
      "step": 2236
    },
    {
      "epoch": 0.09950646925436842,
      "grad_norm": 0.1786056011915207,
      "learning_rate": 0.0009995222584308066,
      "loss": 1.7606,
      "step": 2238
    },
    {
      "epoch": 0.09959539371304077,
      "grad_norm": 0.18306146562099457,
      "learning_rate": 0.0009995207138393176,
      "loss": 1.745,
      "step": 2240
    },
    {
      "epoch": 0.09968431817171312,
      "grad_norm": 0.17886687815189362,
      "learning_rate": 0.0009995191667561364,
      "loss": 1.7594,
      "step": 2242
    },
    {
      "epoch": 0.09977324263038549,
      "grad_norm": 0.17537564039230347,
      "learning_rate": 0.0009995176171812708,
      "loss": 1.7536,
      "step": 2244
    },
    {
      "epoch": 0.09986216708905785,
      "grad_norm": 0.18771038949489594,
      "learning_rate": 0.0009995160651147283,
      "loss": 1.7574,
      "step": 2246
    },
    {
      "epoch": 0.0999510915477302,
      "grad_norm": 0.1864713877439499,
      "learning_rate": 0.0009995145105565167,
      "loss": 1.7607,
      "step": 2248
    },
    {
      "epoch": 0.10004001600640255,
      "grad_norm": 0.1830584704875946,
      "learning_rate": 0.000999512953506644,
      "loss": 1.7579,
      "step": 2250
    },
    {
      "epoch": 0.10012894046507492,
      "grad_norm": 0.16705258190631866,
      "learning_rate": 0.0009995113939651177,
      "loss": 1.7557,
      "step": 2252
    },
    {
      "epoch": 0.10021786492374728,
      "grad_norm": 0.17379559576511383,
      "learning_rate": 0.0009995098319319456,
      "loss": 1.755,
      "step": 2254
    },
    {
      "epoch": 0.10030678938241963,
      "grad_norm": 0.17130069434642792,
      "learning_rate": 0.0009995082674071356,
      "loss": 1.7519,
      "step": 2256
    },
    {
      "epoch": 0.100395713841092,
      "grad_norm": 0.16507191956043243,
      "learning_rate": 0.0009995067003906954,
      "loss": 1.7497,
      "step": 2258
    },
    {
      "epoch": 0.10048463829976435,
      "grad_norm": 0.17065347731113434,
      "learning_rate": 0.0009995051308826328,
      "loss": 1.7598,
      "step": 2260
    },
    {
      "epoch": 0.10057356275843671,
      "grad_norm": 0.15946049988269806,
      "learning_rate": 0.0009995035588829556,
      "loss": 1.7528,
      "step": 2262
    },
    {
      "epoch": 0.10066248721710906,
      "grad_norm": 0.16204677522182465,
      "learning_rate": 0.0009995019843916722,
      "loss": 1.75,
      "step": 2264
    },
    {
      "epoch": 0.10075141167578143,
      "grad_norm": 0.17083314061164856,
      "learning_rate": 0.0009995004074087896,
      "loss": 1.7561,
      "step": 2266
    },
    {
      "epoch": 0.10084033613445378,
      "grad_norm": 0.16661955416202545,
      "learning_rate": 0.0009994988279343163,
      "loss": 1.7493,
      "step": 2268
    },
    {
      "epoch": 0.10092926059312614,
      "grad_norm": 0.17094558477401733,
      "learning_rate": 0.0009994972459682597,
      "loss": 1.7547,
      "step": 2270
    },
    {
      "epoch": 0.10101818505179849,
      "grad_norm": 0.1688528060913086,
      "learning_rate": 0.000999495661510628,
      "loss": 1.753,
      "step": 2272
    },
    {
      "epoch": 0.10110710951047086,
      "grad_norm": 0.16342481970787048,
      "learning_rate": 0.000999494074561429,
      "loss": 1.7535,
      "step": 2274
    },
    {
      "epoch": 0.10119603396914321,
      "grad_norm": 0.16310431063175201,
      "learning_rate": 0.0009994924851206707,
      "loss": 1.7499,
      "step": 2276
    },
    {
      "epoch": 0.10128495842781557,
      "grad_norm": 0.17421621084213257,
      "learning_rate": 0.000999490893188361,
      "loss": 1.752,
      "step": 2278
    },
    {
      "epoch": 0.10137388288648792,
      "grad_norm": 0.16932517290115356,
      "learning_rate": 0.0009994892987645076,
      "loss": 1.7496,
      "step": 2280
    },
    {
      "epoch": 0.10146280734516029,
      "grad_norm": 0.17567944526672363,
      "learning_rate": 0.000999487701849119,
      "loss": 1.7521,
      "step": 2282
    },
    {
      "epoch": 0.10155173180383265,
      "grad_norm": 0.1631414145231247,
      "learning_rate": 0.0009994861024422027,
      "loss": 1.7529,
      "step": 2284
    },
    {
      "epoch": 0.101640656262505,
      "grad_norm": 0.16340436041355133,
      "learning_rate": 0.0009994845005437667,
      "loss": 1.7517,
      "step": 2286
    },
    {
      "epoch": 0.10172958072117735,
      "grad_norm": 0.1647539883852005,
      "learning_rate": 0.0009994828961538192,
      "loss": 1.7488,
      "step": 2288
    },
    {
      "epoch": 0.10181850517984972,
      "grad_norm": 0.16869130730628967,
      "learning_rate": 0.0009994812892723682,
      "loss": 1.7569,
      "step": 2290
    },
    {
      "epoch": 0.10190742963852208,
      "grad_norm": 0.1707681268453598,
      "learning_rate": 0.0009994796798994214,
      "loss": 1.7453,
      "step": 2292
    },
    {
      "epoch": 0.10199635409719443,
      "grad_norm": 0.17477324604988098,
      "learning_rate": 0.0009994780680349872,
      "loss": 1.753,
      "step": 2294
    },
    {
      "epoch": 0.10208527855586678,
      "grad_norm": 0.16798841953277588,
      "learning_rate": 0.0009994764536790733,
      "loss": 1.7492,
      "step": 2296
    },
    {
      "epoch": 0.10217420301453915,
      "grad_norm": 0.17107439041137695,
      "learning_rate": 0.0009994748368316881,
      "loss": 1.7394,
      "step": 2298
    },
    {
      "epoch": 0.10226312747321151,
      "grad_norm": 0.17812353372573853,
      "learning_rate": 0.0009994732174928396,
      "loss": 1.755,
      "step": 2300
    },
    {
      "epoch": 0.10235205193188386,
      "grad_norm": 0.16799861192703247,
      "learning_rate": 0.0009994715956625356,
      "loss": 1.7451,
      "step": 2302
    },
    {
      "epoch": 0.10244097639055623,
      "grad_norm": 0.17055779695510864,
      "learning_rate": 0.0009994699713407845,
      "loss": 1.748,
      "step": 2304
    },
    {
      "epoch": 0.10252990084922858,
      "grad_norm": 0.17974479496479034,
      "learning_rate": 0.0009994683445275943,
      "loss": 1.7459,
      "step": 2306
    },
    {
      "epoch": 0.10261882530790094,
      "grad_norm": 0.1801697462797165,
      "learning_rate": 0.000999466715222973,
      "loss": 1.7471,
      "step": 2308
    },
    {
      "epoch": 0.10270774976657329,
      "grad_norm": 0.17756927013397217,
      "learning_rate": 0.0009994650834269287,
      "loss": 1.7523,
      "step": 2310
    },
    {
      "epoch": 0.10279667422524566,
      "grad_norm": 0.173659086227417,
      "learning_rate": 0.0009994634491394697,
      "loss": 1.7455,
      "step": 2312
    },
    {
      "epoch": 0.10288559868391801,
      "grad_norm": 0.16928839683532715,
      "learning_rate": 0.0009994618123606042,
      "loss": 1.744,
      "step": 2314
    },
    {
      "epoch": 0.10297452314259037,
      "grad_norm": 0.1663655787706375,
      "learning_rate": 0.0009994601730903402,
      "loss": 1.7498,
      "step": 2316
    },
    {
      "epoch": 0.10306344760126272,
      "grad_norm": 0.16375510394573212,
      "learning_rate": 0.000999458531328686,
      "loss": 1.7437,
      "step": 2318
    },
    {
      "epoch": 0.10315237205993509,
      "grad_norm": 0.16213606297969818,
      "learning_rate": 0.0009994568870756498,
      "loss": 1.745,
      "step": 2320
    },
    {
      "epoch": 0.10324129651860744,
      "grad_norm": 0.1694078892469406,
      "learning_rate": 0.0009994552403312397,
      "loss": 1.7449,
      "step": 2322
    },
    {
      "epoch": 0.1033302209772798,
      "grad_norm": 0.1763123720884323,
      "learning_rate": 0.000999453591095464,
      "loss": 1.7382,
      "step": 2324
    },
    {
      "epoch": 0.10341914543595215,
      "grad_norm": 0.16301725804805756,
      "learning_rate": 0.000999451939368331,
      "loss": 1.7417,
      "step": 2326
    },
    {
      "epoch": 0.10350806989462452,
      "grad_norm": 0.17242752015590668,
      "learning_rate": 0.0009994502851498484,
      "loss": 1.74,
      "step": 2328
    },
    {
      "epoch": 0.10359699435329688,
      "grad_norm": 0.16804571449756622,
      "learning_rate": 0.0009994486284400253,
      "loss": 1.7447,
      "step": 2330
    },
    {
      "epoch": 0.10368591881196923,
      "grad_norm": 0.1732250601053238,
      "learning_rate": 0.0009994469692388693,
      "loss": 1.7471,
      "step": 2332
    },
    {
      "epoch": 0.10377484327064158,
      "grad_norm": 0.1738041341304779,
      "learning_rate": 0.000999445307546389,
      "loss": 1.7442,
      "step": 2334
    },
    {
      "epoch": 0.10386376772931395,
      "grad_norm": 0.18072937428951263,
      "learning_rate": 0.0009994436433625926,
      "loss": 1.7448,
      "step": 2336
    },
    {
      "epoch": 0.1039526921879863,
      "grad_norm": 0.16701507568359375,
      "learning_rate": 0.0009994419766874883,
      "loss": 1.7394,
      "step": 2338
    },
    {
      "epoch": 0.10404161664665866,
      "grad_norm": 0.17945359647274017,
      "learning_rate": 0.0009994403075210846,
      "loss": 1.7454,
      "step": 2340
    },
    {
      "epoch": 0.10413054110533101,
      "grad_norm": 0.17203912138938904,
      "learning_rate": 0.0009994386358633898,
      "loss": 1.7446,
      "step": 2342
    },
    {
      "epoch": 0.10421946556400338,
      "grad_norm": 0.17877012491226196,
      "learning_rate": 0.0009994369617144121,
      "loss": 1.7451,
      "step": 2344
    },
    {
      "epoch": 0.10430839002267574,
      "grad_norm": 0.18563194572925568,
      "learning_rate": 0.00099943528507416,
      "loss": 1.7403,
      "step": 2346
    },
    {
      "epoch": 0.10439731448134809,
      "grad_norm": 0.1772448718547821,
      "learning_rate": 0.0009994336059426416,
      "loss": 1.7503,
      "step": 2348
    },
    {
      "epoch": 0.10448623894002046,
      "grad_norm": 0.17919184267520905,
      "learning_rate": 0.0009994319243198657,
      "loss": 1.7465,
      "step": 2350
    },
    {
      "epoch": 0.10457516339869281,
      "grad_norm": 0.18327289819717407,
      "learning_rate": 0.0009994302402058404,
      "loss": 1.7366,
      "step": 2352
    },
    {
      "epoch": 0.10466408785736517,
      "grad_norm": 0.17099310457706451,
      "learning_rate": 0.0009994285536005741,
      "loss": 1.7488,
      "step": 2354
    },
    {
      "epoch": 0.10475301231603752,
      "grad_norm": 0.1615380197763443,
      "learning_rate": 0.0009994268645040754,
      "loss": 1.7382,
      "step": 2356
    },
    {
      "epoch": 0.10484193677470989,
      "grad_norm": 0.18831703066825867,
      "learning_rate": 0.0009994251729163524,
      "loss": 1.7417,
      "step": 2358
    },
    {
      "epoch": 0.10493086123338224,
      "grad_norm": 0.16562096774578094,
      "learning_rate": 0.0009994234788374139,
      "loss": 1.7382,
      "step": 2360
    },
    {
      "epoch": 0.1050197856920546,
      "grad_norm": 0.1602797955274582,
      "learning_rate": 0.0009994217822672682,
      "loss": 1.7417,
      "step": 2362
    },
    {
      "epoch": 0.10510871015072695,
      "grad_norm": 0.15765704214572906,
      "learning_rate": 0.0009994200832059237,
      "loss": 1.7419,
      "step": 2364
    },
    {
      "epoch": 0.10519763460939932,
      "grad_norm": 0.165849968791008,
      "learning_rate": 0.0009994183816533888,
      "loss": 1.7475,
      "step": 2366
    },
    {
      "epoch": 0.10528655906807168,
      "grad_norm": 0.17611372470855713,
      "learning_rate": 0.0009994166776096723,
      "loss": 1.7364,
      "step": 2368
    },
    {
      "epoch": 0.10537548352674403,
      "grad_norm": 0.18146196007728577,
      "learning_rate": 0.0009994149710747823,
      "loss": 1.738,
      "step": 2370
    },
    {
      "epoch": 0.10546440798541638,
      "grad_norm": 0.17269714176654816,
      "learning_rate": 0.0009994132620487278,
      "loss": 1.7429,
      "step": 2372
    },
    {
      "epoch": 0.10555333244408875,
      "grad_norm": 0.1751071959733963,
      "learning_rate": 0.0009994115505315168,
      "loss": 1.7418,
      "step": 2374
    },
    {
      "epoch": 0.1056422569027611,
      "grad_norm": 0.1733344942331314,
      "learning_rate": 0.0009994098365231584,
      "loss": 1.7327,
      "step": 2376
    },
    {
      "epoch": 0.10573118136143346,
      "grad_norm": 0.17954912781715393,
      "learning_rate": 0.0009994081200236605,
      "loss": 1.741,
      "step": 2378
    },
    {
      "epoch": 0.10582010582010581,
      "grad_norm": 0.16836385428905487,
      "learning_rate": 0.000999406401033032,
      "loss": 1.736,
      "step": 2380
    },
    {
      "epoch": 0.10590903027877818,
      "grad_norm": 0.181427001953125,
      "learning_rate": 0.000999404679551282,
      "loss": 1.7416,
      "step": 2382
    },
    {
      "epoch": 0.10599795473745054,
      "grad_norm": 0.17160336673259735,
      "learning_rate": 0.0009994029555784182,
      "loss": 1.7419,
      "step": 2384
    },
    {
      "epoch": 0.10608687919612289,
      "grad_norm": 0.1804126501083374,
      "learning_rate": 0.0009994012291144494,
      "loss": 1.74,
      "step": 2386
    },
    {
      "epoch": 0.10617580365479524,
      "grad_norm": 0.1687408834695816,
      "learning_rate": 0.0009993995001593846,
      "loss": 1.7425,
      "step": 2388
    },
    {
      "epoch": 0.10626472811346761,
      "grad_norm": 0.16643251478672028,
      "learning_rate": 0.000999397768713232,
      "loss": 1.7386,
      "step": 2390
    },
    {
      "epoch": 0.10635365257213997,
      "grad_norm": 0.16881844401359558,
      "learning_rate": 0.0009993960347760005,
      "loss": 1.7383,
      "step": 2392
    },
    {
      "epoch": 0.10644257703081232,
      "grad_norm": 0.1696399301290512,
      "learning_rate": 0.0009993942983476988,
      "loss": 1.7429,
      "step": 2394
    },
    {
      "epoch": 0.10653150148948469,
      "grad_norm": 0.1724826991558075,
      "learning_rate": 0.0009993925594283352,
      "loss": 1.7405,
      "step": 2396
    },
    {
      "epoch": 0.10662042594815704,
      "grad_norm": 0.17419011890888214,
      "learning_rate": 0.000999390818017919,
      "loss": 1.7365,
      "step": 2398
    },
    {
      "epoch": 0.1067093504068294,
      "grad_norm": 0.16340835392475128,
      "learning_rate": 0.0009993890741164582,
      "loss": 1.734,
      "step": 2400
    },
    {
      "epoch": 0.10679827486550175,
      "grad_norm": 0.17877715826034546,
      "learning_rate": 0.000999387327723962,
      "loss": 1.74,
      "step": 2402
    },
    {
      "epoch": 0.10688719932417412,
      "grad_norm": 0.17038090527057648,
      "learning_rate": 0.0009993855788404387,
      "loss": 1.7377,
      "step": 2404
    },
    {
      "epoch": 0.10697612378284647,
      "grad_norm": 0.16269196569919586,
      "learning_rate": 0.0009993838274658972,
      "loss": 1.737,
      "step": 2406
    },
    {
      "epoch": 0.10706504824151883,
      "grad_norm": 0.1645708680152893,
      "learning_rate": 0.0009993820736003466,
      "loss": 1.7404,
      "step": 2408
    },
    {
      "epoch": 0.10715397270019118,
      "grad_norm": 0.16457295417785645,
      "learning_rate": 0.0009993803172437953,
      "loss": 1.7386,
      "step": 2410
    },
    {
      "epoch": 0.10724289715886355,
      "grad_norm": 0.1686396598815918,
      "learning_rate": 0.0009993785583962517,
      "loss": 1.7322,
      "step": 2412
    },
    {
      "epoch": 0.1073318216175359,
      "grad_norm": 0.1839965432882309,
      "learning_rate": 0.0009993767970577251,
      "loss": 1.7396,
      "step": 2414
    },
    {
      "epoch": 0.10742074607620826,
      "grad_norm": 0.178563192486763,
      "learning_rate": 0.0009993750332282245,
      "loss": 1.7399,
      "step": 2416
    },
    {
      "epoch": 0.10750967053488061,
      "grad_norm": 0.18056493997573853,
      "learning_rate": 0.000999373266907758,
      "loss": 1.7386,
      "step": 2418
    },
    {
      "epoch": 0.10759859499355298,
      "grad_norm": 0.17721746861934662,
      "learning_rate": 0.0009993714980963348,
      "loss": 1.73,
      "step": 2420
    },
    {
      "epoch": 0.10768751945222534,
      "grad_norm": 0.1596176028251648,
      "learning_rate": 0.0009993697267939637,
      "loss": 1.7382,
      "step": 2422
    },
    {
      "epoch": 0.10777644391089769,
      "grad_norm": 0.17123466730117798,
      "learning_rate": 0.0009993679530006536,
      "loss": 1.7313,
      "step": 2424
    },
    {
      "epoch": 0.10786536836957004,
      "grad_norm": 0.163445383310318,
      "learning_rate": 0.000999366176716413,
      "loss": 1.739,
      "step": 2426
    },
    {
      "epoch": 0.10795429282824241,
      "grad_norm": 0.16375136375427246,
      "learning_rate": 0.0009993643979412513,
      "loss": 1.7306,
      "step": 2428
    },
    {
      "epoch": 0.10804321728691477,
      "grad_norm": 0.16650699079036713,
      "learning_rate": 0.000999362616675177,
      "loss": 1.74,
      "step": 2430
    },
    {
      "epoch": 0.10813214174558712,
      "grad_norm": 0.16717134416103363,
      "learning_rate": 0.0009993608329181992,
      "loss": 1.7321,
      "step": 2432
    },
    {
      "epoch": 0.10822106620425948,
      "grad_norm": 0.1608593463897705,
      "learning_rate": 0.0009993590466703265,
      "loss": 1.7389,
      "step": 2434
    },
    {
      "epoch": 0.10830999066293184,
      "grad_norm": 0.16169093549251556,
      "learning_rate": 0.000999357257931568,
      "loss": 1.7329,
      "step": 2436
    },
    {
      "epoch": 0.1083989151216042,
      "grad_norm": 0.15770640969276428,
      "learning_rate": 0.0009993554667019327,
      "loss": 1.7344,
      "step": 2438
    },
    {
      "epoch": 0.10848783958027655,
      "grad_norm": 0.16067162156105042,
      "learning_rate": 0.0009993536729814294,
      "loss": 1.7307,
      "step": 2440
    },
    {
      "epoch": 0.10857676403894892,
      "grad_norm": 0.16905395686626434,
      "learning_rate": 0.0009993518767700668,
      "loss": 1.7335,
      "step": 2442
    },
    {
      "epoch": 0.10866568849762127,
      "grad_norm": 0.16266003251075745,
      "learning_rate": 0.0009993500780678547,
      "loss": 1.7362,
      "step": 2444
    },
    {
      "epoch": 0.10875461295629363,
      "grad_norm": 0.16899318993091583,
      "learning_rate": 0.000999348276874801,
      "loss": 1.7361,
      "step": 2446
    },
    {
      "epoch": 0.10884353741496598,
      "grad_norm": 0.16284890472888947,
      "learning_rate": 0.0009993464731909156,
      "loss": 1.7325,
      "step": 2448
    },
    {
      "epoch": 0.10893246187363835,
      "grad_norm": 0.16005484759807587,
      "learning_rate": 0.000999344667016207,
      "loss": 1.7263,
      "step": 2450
    },
    {
      "epoch": 0.1090213863323107,
      "grad_norm": 0.16001056134700775,
      "learning_rate": 0.0009993428583506842,
      "loss": 1.7337,
      "step": 2452
    },
    {
      "epoch": 0.10911031079098306,
      "grad_norm": 0.1561092734336853,
      "learning_rate": 0.0009993410471943564,
      "loss": 1.7349,
      "step": 2454
    },
    {
      "epoch": 0.10919923524965541,
      "grad_norm": 0.16243596374988556,
      "learning_rate": 0.0009993392335472327,
      "loss": 1.7319,
      "step": 2456
    },
    {
      "epoch": 0.10928815970832778,
      "grad_norm": 0.15440620481967926,
      "learning_rate": 0.0009993374174093218,
      "loss": 1.7362,
      "step": 2458
    },
    {
      "epoch": 0.10937708416700014,
      "grad_norm": 0.16307926177978516,
      "learning_rate": 0.000999335598780633,
      "loss": 1.7422,
      "step": 2460
    },
    {
      "epoch": 0.10946600862567249,
      "grad_norm": 0.16545718908309937,
      "learning_rate": 0.0009993337776611755,
      "loss": 1.7348,
      "step": 2462
    },
    {
      "epoch": 0.10955493308434484,
      "grad_norm": 0.16467604041099548,
      "learning_rate": 0.0009993319540509582,
      "loss": 1.7311,
      "step": 2464
    },
    {
      "epoch": 0.10964385754301721,
      "grad_norm": 0.15704773366451263,
      "learning_rate": 0.0009993301279499901,
      "loss": 1.7317,
      "step": 2466
    },
    {
      "epoch": 0.10973278200168957,
      "grad_norm": 0.1750647872686386,
      "learning_rate": 0.0009993282993582804,
      "loss": 1.7354,
      "step": 2468
    },
    {
      "epoch": 0.10982170646036192,
      "grad_norm": 0.17140831053256989,
      "learning_rate": 0.0009993264682758385,
      "loss": 1.7328,
      "step": 2470
    },
    {
      "epoch": 0.10991063091903427,
      "grad_norm": 0.18565557897090912,
      "learning_rate": 0.0009993246347026732,
      "loss": 1.732,
      "step": 2472
    },
    {
      "epoch": 0.10999955537770664,
      "grad_norm": 0.16787555813789368,
      "learning_rate": 0.0009993227986387935,
      "loss": 1.7332,
      "step": 2474
    },
    {
      "epoch": 0.110088479836379,
      "grad_norm": 0.16842927038669586,
      "learning_rate": 0.0009993209600842093,
      "loss": 1.7336,
      "step": 2476
    },
    {
      "epoch": 0.11017740429505135,
      "grad_norm": 0.16819126904010773,
      "learning_rate": 0.0009993191190389287,
      "loss": 1.7303,
      "step": 2478
    },
    {
      "epoch": 0.1102663287537237,
      "grad_norm": 0.1662745177745819,
      "learning_rate": 0.0009993172755029618,
      "loss": 1.7277,
      "step": 2480
    },
    {
      "epoch": 0.11035525321239607,
      "grad_norm": 0.15442778170108795,
      "learning_rate": 0.0009993154294763173,
      "loss": 1.7296,
      "step": 2482
    },
    {
      "epoch": 0.11044417767106843,
      "grad_norm": 0.1560961753129959,
      "learning_rate": 0.0009993135809590048,
      "loss": 1.7373,
      "step": 2484
    },
    {
      "epoch": 0.11053310212974078,
      "grad_norm": 0.15413883328437805,
      "learning_rate": 0.0009993117299510331,
      "loss": 1.7303,
      "step": 2486
    },
    {
      "epoch": 0.11062202658841315,
      "grad_norm": 0.148091122508049,
      "learning_rate": 0.0009993098764524116,
      "loss": 1.7288,
      "step": 2488
    },
    {
      "epoch": 0.1107109510470855,
      "grad_norm": 0.1534961462020874,
      "learning_rate": 0.0009993080204631496,
      "loss": 1.7328,
      "step": 2490
    },
    {
      "epoch": 0.11079987550575786,
      "grad_norm": 0.15761974453926086,
      "learning_rate": 0.0009993061619832562,
      "loss": 1.7275,
      "step": 2492
    },
    {
      "epoch": 0.11088879996443021,
      "grad_norm": 0.15099415183067322,
      "learning_rate": 0.000999304301012741,
      "loss": 1.7273,
      "step": 2494
    },
    {
      "epoch": 0.11097772442310258,
      "grad_norm": 0.16000492870807648,
      "learning_rate": 0.0009993024375516129,
      "loss": 1.7294,
      "step": 2496
    },
    {
      "epoch": 0.11106664888177493,
      "grad_norm": 0.1632835417985916,
      "learning_rate": 0.0009993005715998812,
      "loss": 1.7367,
      "step": 2498
    },
    {
      "epoch": 0.11115557334044729,
      "grad_norm": 0.16005057096481323,
      "learning_rate": 0.0009992987031575556,
      "loss": 1.7306,
      "step": 2500
    },
    {
      "epoch": 0.11115557334044729,
      "eval_loss": 1.6908477544784546,
      "eval_runtime": 12.3532,
      "eval_samples_per_second": 559.37,
      "eval_steps_per_second": 69.942,
      "step": 2500
    },
    {
      "epoch": 0.11124449779911964,
      "grad_norm": 0.16231466829776764,
      "learning_rate": 0.000999296832224645,
      "loss": 1.7317,
      "step": 2502
    },
    {
      "epoch": 0.11133342225779201,
      "grad_norm": 0.1661044955253601,
      "learning_rate": 0.0009992949588011593,
      "loss": 1.7268,
      "step": 2504
    },
    {
      "epoch": 0.11142234671646437,
      "grad_norm": 0.16056203842163086,
      "learning_rate": 0.0009992930828871072,
      "loss": 1.7362,
      "step": 2506
    },
    {
      "epoch": 0.11151127117513672,
      "grad_norm": 0.16574431955814362,
      "learning_rate": 0.0009992912044824984,
      "loss": 1.7286,
      "step": 2508
    },
    {
      "epoch": 0.11160019563380907,
      "grad_norm": 0.15495635569095612,
      "learning_rate": 0.0009992893235873422,
      "loss": 1.7297,
      "step": 2510
    },
    {
      "epoch": 0.11168912009248144,
      "grad_norm": 0.15354973077774048,
      "learning_rate": 0.0009992874402016479,
      "loss": 1.7335,
      "step": 2512
    },
    {
      "epoch": 0.1117780445511538,
      "grad_norm": 0.15808668732643127,
      "learning_rate": 0.000999285554325425,
      "loss": 1.7216,
      "step": 2514
    },
    {
      "epoch": 0.11186696900982615,
      "grad_norm": 0.17658215761184692,
      "learning_rate": 0.0009992836659586828,
      "loss": 1.7262,
      "step": 2516
    },
    {
      "epoch": 0.1119558934684985,
      "grad_norm": 0.15850138664245605,
      "learning_rate": 0.0009992817751014308,
      "loss": 1.7308,
      "step": 2518
    },
    {
      "epoch": 0.11204481792717087,
      "grad_norm": 0.14654642343521118,
      "learning_rate": 0.0009992798817536786,
      "loss": 1.7206,
      "step": 2520
    },
    {
      "epoch": 0.11213374238584323,
      "grad_norm": 0.1507309526205063,
      "learning_rate": 0.0009992779859154355,
      "loss": 1.7274,
      "step": 2522
    },
    {
      "epoch": 0.11222266684451558,
      "grad_norm": 0.15434762835502625,
      "learning_rate": 0.0009992760875867106,
      "loss": 1.7224,
      "step": 2524
    },
    {
      "epoch": 0.11231159130318794,
      "grad_norm": 0.15948152542114258,
      "learning_rate": 0.0009992741867675139,
      "loss": 1.7265,
      "step": 2526
    },
    {
      "epoch": 0.1124005157618603,
      "grad_norm": 0.18099045753479004,
      "learning_rate": 0.0009992722834578547,
      "loss": 1.7229,
      "step": 2528
    },
    {
      "epoch": 0.11248944022053266,
      "grad_norm": 0.17173022031784058,
      "learning_rate": 0.0009992703776577426,
      "loss": 1.7246,
      "step": 2530
    },
    {
      "epoch": 0.11257836467920501,
      "grad_norm": 0.16352210938930511,
      "learning_rate": 0.0009992684693671868,
      "loss": 1.7226,
      "step": 2532
    },
    {
      "epoch": 0.11266728913787738,
      "grad_norm": 0.1548294872045517,
      "learning_rate": 0.000999266558586197,
      "loss": 1.728,
      "step": 2534
    },
    {
      "epoch": 0.11275621359654973,
      "grad_norm": 0.16450268030166626,
      "learning_rate": 0.0009992646453147829,
      "loss": 1.7232,
      "step": 2536
    },
    {
      "epoch": 0.11284513805522209,
      "grad_norm": 0.1629658341407776,
      "learning_rate": 0.0009992627295529537,
      "loss": 1.7282,
      "step": 2538
    },
    {
      "epoch": 0.11293406251389444,
      "grad_norm": 0.145051047205925,
      "learning_rate": 0.0009992608113007192,
      "loss": 1.7257,
      "step": 2540
    },
    {
      "epoch": 0.11302298697256681,
      "grad_norm": 0.1702987104654312,
      "learning_rate": 0.000999258890558089,
      "loss": 1.7199,
      "step": 2542
    },
    {
      "epoch": 0.11311191143123916,
      "grad_norm": 0.1623816341161728,
      "learning_rate": 0.0009992569673250723,
      "loss": 1.7191,
      "step": 2544
    },
    {
      "epoch": 0.11320083588991152,
      "grad_norm": 0.15296286344528198,
      "learning_rate": 0.000999255041601679,
      "loss": 1.7202,
      "step": 2546
    },
    {
      "epoch": 0.11328976034858387,
      "grad_norm": 0.1567697525024414,
      "learning_rate": 0.000999253113387919,
      "loss": 1.7256,
      "step": 2548
    },
    {
      "epoch": 0.11337868480725624,
      "grad_norm": 0.153935045003891,
      "learning_rate": 0.0009992511826838013,
      "loss": 1.7252,
      "step": 2550
    },
    {
      "epoch": 0.1134676092659286,
      "grad_norm": 0.15539416670799255,
      "learning_rate": 0.000999249249489336,
      "loss": 1.7257,
      "step": 2552
    },
    {
      "epoch": 0.11355653372460095,
      "grad_norm": 0.15414440631866455,
      "learning_rate": 0.0009992473138045327,
      "loss": 1.7196,
      "step": 2554
    },
    {
      "epoch": 0.1136454581832733,
      "grad_norm": 0.15253779292106628,
      "learning_rate": 0.0009992453756294006,
      "loss": 1.7202,
      "step": 2556
    },
    {
      "epoch": 0.11373438264194567,
      "grad_norm": 0.1521858274936676,
      "learning_rate": 0.00099924343496395,
      "loss": 1.7215,
      "step": 2558
    },
    {
      "epoch": 0.11382330710061803,
      "grad_norm": 0.15332087874412537,
      "learning_rate": 0.00099924149180819,
      "loss": 1.7324,
      "step": 2560
    },
    {
      "epoch": 0.11391223155929038,
      "grad_norm": 0.16587688028812408,
      "learning_rate": 0.0009992395461621306,
      "loss": 1.7251,
      "step": 2562
    },
    {
      "epoch": 0.11400115601796273,
      "grad_norm": 0.16591493785381317,
      "learning_rate": 0.0009992375980257817,
      "loss": 1.7292,
      "step": 2564
    },
    {
      "epoch": 0.1140900804766351,
      "grad_norm": 0.15625201165676117,
      "learning_rate": 0.0009992356473991525,
      "loss": 1.7215,
      "step": 2566
    },
    {
      "epoch": 0.11417900493530746,
      "grad_norm": 0.15243349969387054,
      "learning_rate": 0.0009992336942822532,
      "loss": 1.7175,
      "step": 2568
    },
    {
      "epoch": 0.11426792939397981,
      "grad_norm": 0.15183353424072266,
      "learning_rate": 0.0009992317386750933,
      "loss": 1.7262,
      "step": 2570
    },
    {
      "epoch": 0.11435685385265217,
      "grad_norm": 0.15364834666252136,
      "learning_rate": 0.0009992297805776827,
      "loss": 1.726,
      "step": 2572
    },
    {
      "epoch": 0.11444577831132453,
      "grad_norm": 0.1533900946378708,
      "learning_rate": 0.000999227819990031,
      "loss": 1.7252,
      "step": 2574
    },
    {
      "epoch": 0.11453470276999689,
      "grad_norm": 0.15749764442443848,
      "learning_rate": 0.000999225856912148,
      "loss": 1.7221,
      "step": 2576
    },
    {
      "epoch": 0.11462362722866924,
      "grad_norm": 0.15611757338047028,
      "learning_rate": 0.000999223891344044,
      "loss": 1.7161,
      "step": 2578
    },
    {
      "epoch": 0.11471255168734161,
      "grad_norm": 0.15750350058078766,
      "learning_rate": 0.0009992219232857278,
      "loss": 1.7249,
      "step": 2580
    },
    {
      "epoch": 0.11480147614601396,
      "grad_norm": 0.16888903081417084,
      "learning_rate": 0.00099921995273721,
      "loss": 1.7217,
      "step": 2582
    },
    {
      "epoch": 0.11489040060468632,
      "grad_norm": 0.1606723666191101,
      "learning_rate": 0.0009992179796985004,
      "loss": 1.7213,
      "step": 2584
    },
    {
      "epoch": 0.11497932506335867,
      "grad_norm": 0.15816037356853485,
      "learning_rate": 0.0009992160041696083,
      "loss": 1.7144,
      "step": 2586
    },
    {
      "epoch": 0.11506824952203104,
      "grad_norm": 0.16229891777038574,
      "learning_rate": 0.0009992140261505442,
      "loss": 1.7215,
      "step": 2588
    },
    {
      "epoch": 0.1151571739807034,
      "grad_norm": 0.16987872123718262,
      "learning_rate": 0.0009992120456413175,
      "loss": 1.7208,
      "step": 2590
    },
    {
      "epoch": 0.11524609843937575,
      "grad_norm": 0.17010580003261566,
      "learning_rate": 0.0009992100626419382,
      "loss": 1.7199,
      "step": 2592
    },
    {
      "epoch": 0.1153350228980481,
      "grad_norm": 0.1615486890077591,
      "learning_rate": 0.0009992080771524165,
      "loss": 1.7174,
      "step": 2594
    },
    {
      "epoch": 0.11542394735672047,
      "grad_norm": 0.15959839522838593,
      "learning_rate": 0.000999206089172762,
      "loss": 1.7177,
      "step": 2596
    },
    {
      "epoch": 0.11551287181539283,
      "grad_norm": 0.1626252979040146,
      "learning_rate": 0.0009992040987029844,
      "loss": 1.7155,
      "step": 2598
    },
    {
      "epoch": 0.11560179627406518,
      "grad_norm": 0.15380766987800598,
      "learning_rate": 0.000999202105743094,
      "loss": 1.7182,
      "step": 2600
    },
    {
      "epoch": 0.11569072073273753,
      "grad_norm": 0.15339282155036926,
      "learning_rate": 0.0009992001102931008,
      "loss": 1.7212,
      "step": 2602
    },
    {
      "epoch": 0.1157796451914099,
      "grad_norm": 0.15971560776233673,
      "learning_rate": 0.0009991981123530144,
      "loss": 1.7232,
      "step": 2604
    },
    {
      "epoch": 0.11586856965008226,
      "grad_norm": 0.1581919938325882,
      "learning_rate": 0.000999196111922845,
      "loss": 1.7221,
      "step": 2606
    },
    {
      "epoch": 0.11595749410875461,
      "grad_norm": 0.1494348645210266,
      "learning_rate": 0.0009991941090026026,
      "loss": 1.7173,
      "step": 2608
    },
    {
      "epoch": 0.11604641856742696,
      "grad_norm": 0.14686119556427002,
      "learning_rate": 0.0009991921035922972,
      "loss": 1.72,
      "step": 2610
    },
    {
      "epoch": 0.11613534302609933,
      "grad_norm": 0.15541252493858337,
      "learning_rate": 0.0009991900956919384,
      "loss": 1.7245,
      "step": 2612
    },
    {
      "epoch": 0.11622426748477169,
      "grad_norm": 0.15264230966567993,
      "learning_rate": 0.000999188085301537,
      "loss": 1.7158,
      "step": 2614
    },
    {
      "epoch": 0.11631319194344404,
      "grad_norm": 0.1619824767112732,
      "learning_rate": 0.000999186072421102,
      "loss": 1.7179,
      "step": 2616
    },
    {
      "epoch": 0.1164021164021164,
      "grad_norm": 0.16652943193912506,
      "learning_rate": 0.0009991840570506443,
      "loss": 1.7214,
      "step": 2618
    },
    {
      "epoch": 0.11649104086078876,
      "grad_norm": 0.1658804565668106,
      "learning_rate": 0.0009991820391901738,
      "loss": 1.721,
      "step": 2620
    },
    {
      "epoch": 0.11657996531946112,
      "grad_norm": 0.1601102203130722,
      "learning_rate": 0.0009991800188397003,
      "loss": 1.7252,
      "step": 2622
    },
    {
      "epoch": 0.11666888977813347,
      "grad_norm": 0.147894486784935,
      "learning_rate": 0.000999177995999234,
      "loss": 1.7161,
      "step": 2624
    },
    {
      "epoch": 0.11675781423680584,
      "grad_norm": 0.1540108621120453,
      "learning_rate": 0.000999175970668785,
      "loss": 1.7226,
      "step": 2626
    },
    {
      "epoch": 0.1168467386954782,
      "grad_norm": 0.15112482011318207,
      "learning_rate": 0.0009991739428483635,
      "loss": 1.7159,
      "step": 2628
    },
    {
      "epoch": 0.11693566315415055,
      "grad_norm": 0.15371938049793243,
      "learning_rate": 0.0009991719125379795,
      "loss": 1.719,
      "step": 2630
    },
    {
      "epoch": 0.1170245876128229,
      "grad_norm": 0.15657579898834229,
      "learning_rate": 0.000999169879737643,
      "loss": 1.719,
      "step": 2632
    },
    {
      "epoch": 0.11711351207149527,
      "grad_norm": 0.15088775753974915,
      "learning_rate": 0.0009991678444473643,
      "loss": 1.71,
      "step": 2634
    },
    {
      "epoch": 0.11720243653016763,
      "grad_norm": 0.15197664499282837,
      "learning_rate": 0.0009991658066671536,
      "loss": 1.7177,
      "step": 2636
    },
    {
      "epoch": 0.11729136098883998,
      "grad_norm": 0.15489469468593597,
      "learning_rate": 0.000999163766397021,
      "loss": 1.7077,
      "step": 2638
    },
    {
      "epoch": 0.11738028544751233,
      "grad_norm": 0.14296165108680725,
      "learning_rate": 0.0009991617236369767,
      "loss": 1.7175,
      "step": 2640
    },
    {
      "epoch": 0.1174692099061847,
      "grad_norm": 0.14150473475456238,
      "learning_rate": 0.0009991596783870307,
      "loss": 1.7163,
      "step": 2642
    },
    {
      "epoch": 0.11755813436485706,
      "grad_norm": 0.1558849960565567,
      "learning_rate": 0.0009991576306471936,
      "loss": 1.7169,
      "step": 2644
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.16008678078651428,
      "learning_rate": 0.0009991555804174752,
      "loss": 1.7156,
      "step": 2646
    },
    {
      "epoch": 0.11773598328220176,
      "grad_norm": 0.1638006567955017,
      "learning_rate": 0.000999153527697886,
      "loss": 1.7208,
      "step": 2648
    },
    {
      "epoch": 0.11782490774087413,
      "grad_norm": 0.15358804166316986,
      "learning_rate": 0.0009991514724884362,
      "loss": 1.7161,
      "step": 2650
    },
    {
      "epoch": 0.11791383219954649,
      "grad_norm": 0.14830666780471802,
      "learning_rate": 0.000999149414789136,
      "loss": 1.7219,
      "step": 2652
    },
    {
      "epoch": 0.11800275665821884,
      "grad_norm": 0.15406952798366547,
      "learning_rate": 0.0009991473545999956,
      "loss": 1.71,
      "step": 2654
    },
    {
      "epoch": 0.1180916811168912,
      "grad_norm": 0.15282905101776123,
      "learning_rate": 0.0009991452919210252,
      "loss": 1.7205,
      "step": 2656
    },
    {
      "epoch": 0.11818060557556356,
      "grad_norm": 0.15068869292736053,
      "learning_rate": 0.0009991432267522353,
      "loss": 1.7166,
      "step": 2658
    },
    {
      "epoch": 0.11826953003423592,
      "grad_norm": 0.14806392788887024,
      "learning_rate": 0.0009991411590936363,
      "loss": 1.7165,
      "step": 2660
    },
    {
      "epoch": 0.11835845449290827,
      "grad_norm": 0.15650790929794312,
      "learning_rate": 0.0009991390889452383,
      "loss": 1.7121,
      "step": 2662
    },
    {
      "epoch": 0.11844737895158063,
      "grad_norm": 0.16825319826602936,
      "learning_rate": 0.0009991370163070516,
      "loss": 1.7079,
      "step": 2664
    },
    {
      "epoch": 0.118536303410253,
      "grad_norm": 0.15932542085647583,
      "learning_rate": 0.0009991349411790868,
      "loss": 1.7133,
      "step": 2666
    },
    {
      "epoch": 0.11862522786892535,
      "grad_norm": 0.15582598745822906,
      "learning_rate": 0.0009991328635613537,
      "loss": 1.7171,
      "step": 2668
    },
    {
      "epoch": 0.1187141523275977,
      "grad_norm": 0.1434444785118103,
      "learning_rate": 0.0009991307834538632,
      "loss": 1.7143,
      "step": 2670
    },
    {
      "epoch": 0.11880307678627007,
      "grad_norm": 0.1549002230167389,
      "learning_rate": 0.0009991287008566255,
      "loss": 1.7218,
      "step": 2672
    },
    {
      "epoch": 0.11889200124494242,
      "grad_norm": 0.15441077947616577,
      "learning_rate": 0.0009991266157696511,
      "loss": 1.7141,
      "step": 2674
    },
    {
      "epoch": 0.11898092570361478,
      "grad_norm": 0.1501181274652481,
      "learning_rate": 0.0009991245281929502,
      "loss": 1.7079,
      "step": 2676
    },
    {
      "epoch": 0.11906985016228713,
      "grad_norm": 0.1563824862241745,
      "learning_rate": 0.0009991224381265332,
      "loss": 1.7179,
      "step": 2678
    },
    {
      "epoch": 0.1191587746209595,
      "grad_norm": 0.15084655582904816,
      "learning_rate": 0.0009991203455704107,
      "loss": 1.7089,
      "step": 2680
    },
    {
      "epoch": 0.11924769907963186,
      "grad_norm": 0.15098781883716583,
      "learning_rate": 0.000999118250524593,
      "loss": 1.7131,
      "step": 2682
    },
    {
      "epoch": 0.11933662353830421,
      "grad_norm": 0.1477523148059845,
      "learning_rate": 0.0009991161529890908,
      "loss": 1.712,
      "step": 2684
    },
    {
      "epoch": 0.11942554799697656,
      "grad_norm": 0.15165407955646515,
      "learning_rate": 0.0009991140529639142,
      "loss": 1.7142,
      "step": 2686
    },
    {
      "epoch": 0.11951447245564893,
      "grad_norm": 0.14917202293872833,
      "learning_rate": 0.000999111950449074,
      "loss": 1.7139,
      "step": 2688
    },
    {
      "epoch": 0.11960339691432129,
      "grad_norm": 0.15467822551727295,
      "learning_rate": 0.0009991098454445806,
      "loss": 1.7058,
      "step": 2690
    },
    {
      "epoch": 0.11969232137299364,
      "grad_norm": 0.1571405827999115,
      "learning_rate": 0.0009991077379504442,
      "loss": 1.72,
      "step": 2692
    },
    {
      "epoch": 0.119781245831666,
      "grad_norm": 0.15948142111301422,
      "learning_rate": 0.0009991056279666758,
      "loss": 1.7136,
      "step": 2694
    },
    {
      "epoch": 0.11987017029033836,
      "grad_norm": 0.15302774310112,
      "learning_rate": 0.0009991035154932855,
      "loss": 1.7084,
      "step": 2696
    },
    {
      "epoch": 0.11995909474901072,
      "grad_norm": 0.14382019639015198,
      "learning_rate": 0.0009991014005302841,
      "loss": 1.7078,
      "step": 2698
    },
    {
      "epoch": 0.12004801920768307,
      "grad_norm": 0.14728422462940216,
      "learning_rate": 0.000999099283077682,
      "loss": 1.7078,
      "step": 2700
    },
    {
      "epoch": 0.12013694366635543,
      "grad_norm": 0.1458166539669037,
      "learning_rate": 0.00099909716313549,
      "loss": 1.714,
      "step": 2702
    },
    {
      "epoch": 0.1202258681250278,
      "grad_norm": 0.15718547999858856,
      "learning_rate": 0.0009990950407037184,
      "loss": 1.7146,
      "step": 2704
    },
    {
      "epoch": 0.12031479258370015,
      "grad_norm": 0.1658591479063034,
      "learning_rate": 0.000999092915782378,
      "loss": 1.7134,
      "step": 2706
    },
    {
      "epoch": 0.1204037170423725,
      "grad_norm": 0.1692623794078827,
      "learning_rate": 0.000999090788371479,
      "loss": 1.7091,
      "step": 2708
    },
    {
      "epoch": 0.12049264150104486,
      "grad_norm": 0.16184183955192566,
      "learning_rate": 0.0009990886584710324,
      "loss": 1.7098,
      "step": 2710
    },
    {
      "epoch": 0.12058156595971722,
      "grad_norm": 0.14483095705509186,
      "learning_rate": 0.0009990865260810487,
      "loss": 1.7069,
      "step": 2712
    },
    {
      "epoch": 0.12067049041838958,
      "grad_norm": 0.15486344695091248,
      "learning_rate": 0.0009990843912015387,
      "loss": 1.7116,
      "step": 2714
    },
    {
      "epoch": 0.12075941487706193,
      "grad_norm": 0.14510615170001984,
      "learning_rate": 0.000999082253832513,
      "loss": 1.7196,
      "step": 2716
    },
    {
      "epoch": 0.1208483393357343,
      "grad_norm": 0.15198862552642822,
      "learning_rate": 0.000999080113973982,
      "loss": 1.7065,
      "step": 2718
    },
    {
      "epoch": 0.12093726379440665,
      "grad_norm": 0.14839208126068115,
      "learning_rate": 0.0009990779716259567,
      "loss": 1.7075,
      "step": 2720
    },
    {
      "epoch": 0.12102618825307901,
      "grad_norm": 0.14327092468738556,
      "learning_rate": 0.0009990758267884475,
      "loss": 1.7133,
      "step": 2722
    },
    {
      "epoch": 0.12111511271175136,
      "grad_norm": 0.1524917483329773,
      "learning_rate": 0.0009990736794614652,
      "loss": 1.7002,
      "step": 2724
    },
    {
      "epoch": 0.12120403717042373,
      "grad_norm": 0.14945504069328308,
      "learning_rate": 0.0009990715296450206,
      "loss": 1.714,
      "step": 2726
    },
    {
      "epoch": 0.12129296162909609,
      "grad_norm": 0.15353934466838837,
      "learning_rate": 0.0009990693773391244,
      "loss": 1.7118,
      "step": 2728
    },
    {
      "epoch": 0.12138188608776844,
      "grad_norm": 0.14560256898403168,
      "learning_rate": 0.0009990672225437872,
      "loss": 1.7053,
      "step": 2730
    },
    {
      "epoch": 0.1214708105464408,
      "grad_norm": 0.15236537158489227,
      "learning_rate": 0.00099906506525902,
      "loss": 1.71,
      "step": 2732
    },
    {
      "epoch": 0.12155973500511316,
      "grad_norm": 0.15007594227790833,
      "learning_rate": 0.0009990629054848332,
      "loss": 1.7085,
      "step": 2734
    },
    {
      "epoch": 0.12164865946378552,
      "grad_norm": 0.1456535905599594,
      "learning_rate": 0.000999060743221238,
      "loss": 1.7157,
      "step": 2736
    },
    {
      "epoch": 0.12173758392245787,
      "grad_norm": 0.15266625583171844,
      "learning_rate": 0.0009990585784682448,
      "loss": 1.711,
      "step": 2738
    },
    {
      "epoch": 0.12182650838113022,
      "grad_norm": 0.16604028642177582,
      "learning_rate": 0.000999056411225865,
      "loss": 1.7128,
      "step": 2740
    },
    {
      "epoch": 0.12191543283980259,
      "grad_norm": 0.15921242535114288,
      "learning_rate": 0.0009990542414941084,
      "loss": 1.7046,
      "step": 2742
    },
    {
      "epoch": 0.12200435729847495,
      "grad_norm": 0.1624920517206192,
      "learning_rate": 0.0009990520692729867,
      "loss": 1.7087,
      "step": 2744
    },
    {
      "epoch": 0.1220932817571473,
      "grad_norm": 0.15475033223628998,
      "learning_rate": 0.0009990498945625103,
      "loss": 1.7067,
      "step": 2746
    },
    {
      "epoch": 0.12218220621581966,
      "grad_norm": 0.1475726068019867,
      "learning_rate": 0.0009990477173626904,
      "loss": 1.7074,
      "step": 2748
    },
    {
      "epoch": 0.12227113067449202,
      "grad_norm": 0.15255337953567505,
      "learning_rate": 0.0009990455376735375,
      "loss": 1.7045,
      "step": 2750
    },
    {
      "epoch": 0.12236005513316438,
      "grad_norm": 0.14628992974758148,
      "learning_rate": 0.0009990433554950627,
      "loss": 1.7138,
      "step": 2752
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 0.15724748373031616,
      "learning_rate": 0.0009990411708272768,
      "loss": 1.714,
      "step": 2754
    },
    {
      "epoch": 0.12253790405050909,
      "grad_norm": 0.15618425607681274,
      "learning_rate": 0.0009990389836701907,
      "loss": 1.6981,
      "step": 2756
    },
    {
      "epoch": 0.12262682850918145,
      "grad_norm": 0.15823392570018768,
      "learning_rate": 0.0009990367940238153,
      "loss": 1.7111,
      "step": 2758
    },
    {
      "epoch": 0.12271575296785381,
      "grad_norm": 0.1651688665151596,
      "learning_rate": 0.0009990346018881615,
      "loss": 1.7038,
      "step": 2760
    },
    {
      "epoch": 0.12280467742652616,
      "grad_norm": 0.1569395363330841,
      "learning_rate": 0.0009990324072632402,
      "loss": 1.7033,
      "step": 2762
    },
    {
      "epoch": 0.12289360188519853,
      "grad_norm": 0.15873000025749207,
      "learning_rate": 0.0009990302101490626,
      "loss": 1.7047,
      "step": 2764
    },
    {
      "epoch": 0.12298252634387088,
      "grad_norm": 0.1520274430513382,
      "learning_rate": 0.0009990280105456393,
      "loss": 1.6992,
      "step": 2766
    },
    {
      "epoch": 0.12307145080254324,
      "grad_norm": 0.14999651908874512,
      "learning_rate": 0.0009990258084529816,
      "loss": 1.7109,
      "step": 2768
    },
    {
      "epoch": 0.1231603752612156,
      "grad_norm": 0.16025303304195404,
      "learning_rate": 0.0009990236038711002,
      "loss": 1.7068,
      "step": 2770
    },
    {
      "epoch": 0.12324929971988796,
      "grad_norm": 0.15831394493579865,
      "learning_rate": 0.0009990213968000064,
      "loss": 1.703,
      "step": 2772
    },
    {
      "epoch": 0.12333822417856032,
      "grad_norm": 0.16204720735549927,
      "learning_rate": 0.0009990191872397108,
      "loss": 1.7041,
      "step": 2774
    },
    {
      "epoch": 0.12342714863723267,
      "grad_norm": 0.15055668354034424,
      "learning_rate": 0.000999016975190225,
      "loss": 1.6996,
      "step": 2776
    },
    {
      "epoch": 0.12351607309590502,
      "grad_norm": 0.15014596283435822,
      "learning_rate": 0.0009990147606515594,
      "loss": 1.7061,
      "step": 2778
    },
    {
      "epoch": 0.12360499755457739,
      "grad_norm": 0.15276427567005157,
      "learning_rate": 0.0009990125436237253,
      "loss": 1.7085,
      "step": 2780
    },
    {
      "epoch": 0.12369392201324975,
      "grad_norm": 0.14822447299957275,
      "learning_rate": 0.000999010324106734,
      "loss": 1.7116,
      "step": 2782
    },
    {
      "epoch": 0.1237828464719221,
      "grad_norm": 0.15309688448905945,
      "learning_rate": 0.0009990081021005964,
      "loss": 1.7078,
      "step": 2784
    },
    {
      "epoch": 0.12387177093059445,
      "grad_norm": 0.15499570965766907,
      "learning_rate": 0.0009990058776053233,
      "loss": 1.7011,
      "step": 2786
    },
    {
      "epoch": 0.12396069538926682,
      "grad_norm": 0.15672209858894348,
      "learning_rate": 0.0009990036506209263,
      "loss": 1.706,
      "step": 2788
    },
    {
      "epoch": 0.12404961984793918,
      "grad_norm": 0.15120433270931244,
      "learning_rate": 0.000999001421147416,
      "loss": 1.7017,
      "step": 2790
    },
    {
      "epoch": 0.12413854430661153,
      "grad_norm": 0.14683088660240173,
      "learning_rate": 0.000998999189184804,
      "loss": 1.7024,
      "step": 2792
    },
    {
      "epoch": 0.12422746876528389,
      "grad_norm": 0.15255948901176453,
      "learning_rate": 0.0009989969547331012,
      "loss": 1.7104,
      "step": 2794
    },
    {
      "epoch": 0.12431639322395625,
      "grad_norm": 0.15793496370315552,
      "learning_rate": 0.0009989947177923189,
      "loss": 1.707,
      "step": 2796
    },
    {
      "epoch": 0.12440531768262861,
      "grad_norm": 0.15464596450328827,
      "learning_rate": 0.0009989924783624677,
      "loss": 1.7053,
      "step": 2798
    },
    {
      "epoch": 0.12449424214130096,
      "grad_norm": 0.15060003101825714,
      "learning_rate": 0.0009989902364435596,
      "loss": 1.7017,
      "step": 2800
    },
    {
      "epoch": 0.12458316659997332,
      "grad_norm": 0.1549662947654724,
      "learning_rate": 0.0009989879920356051,
      "loss": 1.6983,
      "step": 2802
    },
    {
      "epoch": 0.12467209105864568,
      "grad_norm": 0.1606224775314331,
      "learning_rate": 0.000998985745138616,
      "loss": 1.7036,
      "step": 2804
    },
    {
      "epoch": 0.12476101551731804,
      "grad_norm": 0.14715637266635895,
      "learning_rate": 0.0009989834957526028,
      "loss": 1.7038,
      "step": 2806
    },
    {
      "epoch": 0.12484993997599039,
      "grad_norm": 0.16040703654289246,
      "learning_rate": 0.0009989812438775771,
      "loss": 1.7035,
      "step": 2808
    },
    {
      "epoch": 0.12493886443466276,
      "grad_norm": 0.14962682127952576,
      "learning_rate": 0.0009989789895135503,
      "loss": 1.6948,
      "step": 2810
    },
    {
      "epoch": 0.1250277888933351,
      "grad_norm": 0.14733915030956268,
      "learning_rate": 0.0009989767326605335,
      "loss": 1.6969,
      "step": 2812
    },
    {
      "epoch": 0.12511671335200747,
      "grad_norm": 0.15031510591506958,
      "learning_rate": 0.000998974473318538,
      "loss": 1.7009,
      "step": 2814
    },
    {
      "epoch": 0.12520563781067984,
      "grad_norm": 0.14500756561756134,
      "learning_rate": 0.0009989722114875747,
      "loss": 1.6987,
      "step": 2816
    },
    {
      "epoch": 0.12529456226935218,
      "grad_norm": 0.14894454181194305,
      "learning_rate": 0.0009989699471676554,
      "loss": 1.6968,
      "step": 2818
    },
    {
      "epoch": 0.12538348672802455,
      "grad_norm": 0.14639148116111755,
      "learning_rate": 0.0009989676803587914,
      "loss": 1.699,
      "step": 2820
    },
    {
      "epoch": 0.1254724111866969,
      "grad_norm": 0.1445195972919464,
      "learning_rate": 0.0009989654110609933,
      "loss": 1.7026,
      "step": 2822
    },
    {
      "epoch": 0.12556133564536925,
      "grad_norm": 0.1507134586572647,
      "learning_rate": 0.0009989631392742732,
      "loss": 1.7009,
      "step": 2824
    },
    {
      "epoch": 0.12565026010404162,
      "grad_norm": 0.16465866565704346,
      "learning_rate": 0.000998960864998642,
      "loss": 1.7034,
      "step": 2826
    },
    {
      "epoch": 0.12573918456271396,
      "grad_norm": 0.15758581459522247,
      "learning_rate": 0.0009989585882341113,
      "loss": 1.7026,
      "step": 2828
    },
    {
      "epoch": 0.12582810902138633,
      "grad_norm": 0.15651194751262665,
      "learning_rate": 0.0009989563089806924,
      "loss": 1.7085,
      "step": 2830
    },
    {
      "epoch": 0.1259170334800587,
      "grad_norm": 0.16078348457813263,
      "learning_rate": 0.0009989540272383962,
      "loss": 1.6964,
      "step": 2832
    },
    {
      "epoch": 0.12600595793873104,
      "grad_norm": 0.16136224567890167,
      "learning_rate": 0.000998951743007235,
      "loss": 1.6983,
      "step": 2834
    },
    {
      "epoch": 0.1260948823974034,
      "grad_norm": 0.1584000438451767,
      "learning_rate": 0.0009989494562872194,
      "loss": 1.6981,
      "step": 2836
    },
    {
      "epoch": 0.12618380685607578,
      "grad_norm": 0.15041996538639069,
      "learning_rate": 0.0009989471670783612,
      "loss": 1.7036,
      "step": 2838
    },
    {
      "epoch": 0.12627273131474812,
      "grad_norm": 0.14316673576831818,
      "learning_rate": 0.0009989448753806717,
      "loss": 1.7013,
      "step": 2840
    },
    {
      "epoch": 0.12636165577342048,
      "grad_norm": 0.1562669724225998,
      "learning_rate": 0.0009989425811941623,
      "loss": 1.7016,
      "step": 2842
    },
    {
      "epoch": 0.12645058023209282,
      "grad_norm": 0.15850582718849182,
      "learning_rate": 0.0009989402845188445,
      "loss": 1.6942,
      "step": 2844
    },
    {
      "epoch": 0.1265395046907652,
      "grad_norm": 0.14375914633274078,
      "learning_rate": 0.0009989379853547298,
      "loss": 1.699,
      "step": 2846
    },
    {
      "epoch": 0.12662842914943756,
      "grad_norm": 0.1460895985364914,
      "learning_rate": 0.0009989356837018296,
      "loss": 1.7018,
      "step": 2848
    },
    {
      "epoch": 0.1267173536081099,
      "grad_norm": 0.1511843204498291,
      "learning_rate": 0.0009989333795601553,
      "loss": 1.7008,
      "step": 2850
    },
    {
      "epoch": 0.12680627806678227,
      "grad_norm": 0.15189401805400848,
      "learning_rate": 0.0009989310729297185,
      "loss": 1.6961,
      "step": 2852
    },
    {
      "epoch": 0.12689520252545464,
      "grad_norm": 0.15254703164100647,
      "learning_rate": 0.0009989287638105307,
      "loss": 1.6967,
      "step": 2854
    },
    {
      "epoch": 0.12698412698412698,
      "grad_norm": 0.1455395370721817,
      "learning_rate": 0.0009989264522026033,
      "loss": 1.7,
      "step": 2856
    },
    {
      "epoch": 0.12707305144279935,
      "grad_norm": 0.14095422625541687,
      "learning_rate": 0.0009989241381059481,
      "loss": 1.6935,
      "step": 2858
    },
    {
      "epoch": 0.1271619759014717,
      "grad_norm": 0.14039580523967743,
      "learning_rate": 0.0009989218215205764,
      "loss": 1.7008,
      "step": 2860
    },
    {
      "epoch": 0.12725090036014405,
      "grad_norm": 0.14331957697868347,
      "learning_rate": 0.0009989195024465,
      "loss": 1.7059,
      "step": 2862
    },
    {
      "epoch": 0.12733982481881642,
      "grad_norm": 0.14794786274433136,
      "learning_rate": 0.0009989171808837303,
      "loss": 1.7019,
      "step": 2864
    },
    {
      "epoch": 0.12742874927748876,
      "grad_norm": 0.14051014184951782,
      "learning_rate": 0.0009989148568322786,
      "loss": 1.6935,
      "step": 2866
    },
    {
      "epoch": 0.12751767373616113,
      "grad_norm": 0.14383691549301147,
      "learning_rate": 0.0009989125302921572,
      "loss": 1.6944,
      "step": 2868
    },
    {
      "epoch": 0.1276065981948335,
      "grad_norm": 0.149238720536232,
      "learning_rate": 0.0009989102012633768,
      "loss": 1.7026,
      "step": 2870
    },
    {
      "epoch": 0.12769552265350584,
      "grad_norm": 0.1515042781829834,
      "learning_rate": 0.0009989078697459498,
      "loss": 1.6958,
      "step": 2872
    },
    {
      "epoch": 0.1277844471121782,
      "grad_norm": 0.14632724225521088,
      "learning_rate": 0.0009989055357398876,
      "loss": 1.7018,
      "step": 2874
    },
    {
      "epoch": 0.12787337157085057,
      "grad_norm": 0.14849677681922913,
      "learning_rate": 0.0009989031992452015,
      "loss": 1.6933,
      "step": 2876
    },
    {
      "epoch": 0.12796229602952292,
      "grad_norm": 0.15248118340969086,
      "learning_rate": 0.0009989008602619036,
      "loss": 1.6951,
      "step": 2878
    },
    {
      "epoch": 0.12805122048819528,
      "grad_norm": 0.14534400403499603,
      "learning_rate": 0.0009988985187900053,
      "loss": 1.6988,
      "step": 2880
    },
    {
      "epoch": 0.12814014494686762,
      "grad_norm": 0.14211168885231018,
      "learning_rate": 0.0009988961748295185,
      "loss": 1.695,
      "step": 2882
    },
    {
      "epoch": 0.12822906940554,
      "grad_norm": 0.14200276136398315,
      "learning_rate": 0.0009988938283804548,
      "loss": 1.6924,
      "step": 2884
    },
    {
      "epoch": 0.12831799386421236,
      "grad_norm": 0.13837294280529022,
      "learning_rate": 0.0009988914794428257,
      "loss": 1.7047,
      "step": 2886
    },
    {
      "epoch": 0.1284069183228847,
      "grad_norm": 0.14080971479415894,
      "learning_rate": 0.0009988891280166433,
      "loss": 1.6943,
      "step": 2888
    },
    {
      "epoch": 0.12849584278155707,
      "grad_norm": 0.14714248478412628,
      "learning_rate": 0.000998886774101919,
      "loss": 1.6969,
      "step": 2890
    },
    {
      "epoch": 0.12858476724022944,
      "grad_norm": 0.13803038001060486,
      "learning_rate": 0.0009988844176986645,
      "loss": 1.7008,
      "step": 2892
    },
    {
      "epoch": 0.12867369169890178,
      "grad_norm": 0.1391960233449936,
      "learning_rate": 0.0009988820588068918,
      "loss": 1.6993,
      "step": 2894
    },
    {
      "epoch": 0.12876261615757414,
      "grad_norm": 0.1502789705991745,
      "learning_rate": 0.0009988796974266126,
      "loss": 1.6965,
      "step": 2896
    },
    {
      "epoch": 0.12885154061624648,
      "grad_norm": 0.16213124990463257,
      "learning_rate": 0.0009988773335578387,
      "loss": 1.7027,
      "step": 2898
    },
    {
      "epoch": 0.12894046507491885,
      "grad_norm": 0.1447189748287201,
      "learning_rate": 0.0009988749672005819,
      "loss": 1.6945,
      "step": 2900
    },
    {
      "epoch": 0.12902938953359122,
      "grad_norm": 0.14806969463825226,
      "learning_rate": 0.000998872598354854,
      "loss": 1.6946,
      "step": 2902
    },
    {
      "epoch": 0.12911831399226356,
      "grad_norm": 0.14672061800956726,
      "learning_rate": 0.0009988702270206666,
      "loss": 1.6867,
      "step": 2904
    },
    {
      "epoch": 0.12920723845093593,
      "grad_norm": 0.1454840451478958,
      "learning_rate": 0.0009988678531980318,
      "loss": 1.6891,
      "step": 2906
    },
    {
      "epoch": 0.1292961629096083,
      "grad_norm": 0.13582098484039307,
      "learning_rate": 0.0009988654768869614,
      "loss": 1.6986,
      "step": 2908
    },
    {
      "epoch": 0.12938508736828064,
      "grad_norm": 0.13995283842086792,
      "learning_rate": 0.000998863098087467,
      "loss": 1.6971,
      "step": 2910
    },
    {
      "epoch": 0.129474011826953,
      "grad_norm": 0.1353904753923416,
      "learning_rate": 0.0009988607167995607,
      "loss": 1.6926,
      "step": 2912
    },
    {
      "epoch": 0.12956293628562537,
      "grad_norm": 0.13783226907253265,
      "learning_rate": 0.0009988583330232546,
      "loss": 1.6971,
      "step": 2914
    },
    {
      "epoch": 0.12965186074429771,
      "grad_norm": 0.14349259436130524,
      "learning_rate": 0.00099885594675856,
      "loss": 1.6915,
      "step": 2916
    },
    {
      "epoch": 0.12974078520297008,
      "grad_norm": 0.1502244919538498,
      "learning_rate": 0.0009988535580054893,
      "loss": 1.6919,
      "step": 2918
    },
    {
      "epoch": 0.12982970966164242,
      "grad_norm": 0.15048769116401672,
      "learning_rate": 0.0009988511667640542,
      "loss": 1.6992,
      "step": 2920
    },
    {
      "epoch": 0.1299186341203148,
      "grad_norm": 0.1476777046918869,
      "learning_rate": 0.0009988487730342667,
      "loss": 1.6874,
      "step": 2922
    },
    {
      "epoch": 0.13000755857898716,
      "grad_norm": 0.1491258591413498,
      "learning_rate": 0.0009988463768161387,
      "loss": 1.6969,
      "step": 2924
    },
    {
      "epoch": 0.1300964830376595,
      "grad_norm": 0.14041897654533386,
      "learning_rate": 0.0009988439781096821,
      "loss": 1.6917,
      "step": 2926
    },
    {
      "epoch": 0.13018540749633187,
      "grad_norm": 0.14257672429084778,
      "learning_rate": 0.0009988415769149092,
      "loss": 1.6875,
      "step": 2928
    },
    {
      "epoch": 0.13027433195500424,
      "grad_norm": 0.1400248110294342,
      "learning_rate": 0.0009988391732318313,
      "loss": 1.6904,
      "step": 2930
    },
    {
      "epoch": 0.13036325641367658,
      "grad_norm": 0.13869576156139374,
      "learning_rate": 0.000998836767060461,
      "loss": 1.6931,
      "step": 2932
    },
    {
      "epoch": 0.13045218087234894,
      "grad_norm": 0.15129977464675903,
      "learning_rate": 0.0009988343584008102,
      "loss": 1.6964,
      "step": 2934
    },
    {
      "epoch": 0.13054110533102128,
      "grad_norm": 0.14765389263629913,
      "learning_rate": 0.0009988319472528909,
      "loss": 1.691,
      "step": 2936
    },
    {
      "epoch": 0.13063002978969365,
      "grad_norm": 0.14335504174232483,
      "learning_rate": 0.0009988295336167147,
      "loss": 1.6875,
      "step": 2938
    },
    {
      "epoch": 0.13071895424836602,
      "grad_norm": 0.1386997550725937,
      "learning_rate": 0.0009988271174922942,
      "loss": 1.6867,
      "step": 2940
    },
    {
      "epoch": 0.13080787870703836,
      "grad_norm": 0.15370407700538635,
      "learning_rate": 0.0009988246988796412,
      "loss": 1.6949,
      "step": 2942
    },
    {
      "epoch": 0.13089680316571073,
      "grad_norm": 0.15418584644794464,
      "learning_rate": 0.000998822277778768,
      "loss": 1.6916,
      "step": 2944
    },
    {
      "epoch": 0.1309857276243831,
      "grad_norm": 0.15106600522994995,
      "learning_rate": 0.0009988198541896862,
      "loss": 1.695,
      "step": 2946
    },
    {
      "epoch": 0.13107465208305544,
      "grad_norm": 0.15028196573257446,
      "learning_rate": 0.000998817428112408,
      "loss": 1.6886,
      "step": 2948
    },
    {
      "epoch": 0.1311635765417278,
      "grad_norm": 0.14836204051971436,
      "learning_rate": 0.000998814999546946,
      "loss": 1.6949,
      "step": 2950
    },
    {
      "epoch": 0.13125250100040017,
      "grad_norm": 0.14906133711338043,
      "learning_rate": 0.000998812568493312,
      "loss": 1.6894,
      "step": 2952
    },
    {
      "epoch": 0.1313414254590725,
      "grad_norm": 0.14908622205257416,
      "learning_rate": 0.000998810134951518,
      "loss": 1.6903,
      "step": 2954
    },
    {
      "epoch": 0.13143034991774488,
      "grad_norm": 0.14440849423408508,
      "learning_rate": 0.0009988076989215762,
      "loss": 1.6966,
      "step": 2956
    },
    {
      "epoch": 0.13151927437641722,
      "grad_norm": 0.14841178059577942,
      "learning_rate": 0.0009988052604034989,
      "loss": 1.6882,
      "step": 2958
    },
    {
      "epoch": 0.1316081988350896,
      "grad_norm": 0.14035280048847198,
      "learning_rate": 0.0009988028193972983,
      "loss": 1.682,
      "step": 2960
    },
    {
      "epoch": 0.13169712329376196,
      "grad_norm": 0.1416475474834442,
      "learning_rate": 0.000998800375902986,
      "loss": 1.6944,
      "step": 2962
    },
    {
      "epoch": 0.1317860477524343,
      "grad_norm": 0.14471939206123352,
      "learning_rate": 0.0009987979299205748,
      "loss": 1.6878,
      "step": 2964
    },
    {
      "epoch": 0.13187497221110667,
      "grad_norm": 0.14979827404022217,
      "learning_rate": 0.000998795481450077,
      "loss": 1.6979,
      "step": 2966
    },
    {
      "epoch": 0.13196389666977903,
      "grad_norm": 0.14293459057807922,
      "learning_rate": 0.0009987930304915043,
      "loss": 1.697,
      "step": 2968
    },
    {
      "epoch": 0.13205282112845138,
      "grad_norm": 0.13399027287960052,
      "learning_rate": 0.0009987905770448694,
      "loss": 1.6866,
      "step": 2970
    },
    {
      "epoch": 0.13214174558712374,
      "grad_norm": 0.14252635836601257,
      "learning_rate": 0.0009987881211101842,
      "loss": 1.6921,
      "step": 2972
    },
    {
      "epoch": 0.13223067004579608,
      "grad_norm": 0.14642393589019775,
      "learning_rate": 0.0009987856626874608,
      "loss": 1.6899,
      "step": 2974
    },
    {
      "epoch": 0.13231959450446845,
      "grad_norm": 0.15148261189460754,
      "learning_rate": 0.000998783201776712,
      "loss": 1.6916,
      "step": 2976
    },
    {
      "epoch": 0.13240851896314082,
      "grad_norm": 0.14195556938648224,
      "learning_rate": 0.0009987807383779496,
      "loss": 1.691,
      "step": 2978
    },
    {
      "epoch": 0.13249744342181316,
      "grad_norm": 0.14468583464622498,
      "learning_rate": 0.0009987782724911862,
      "loss": 1.6911,
      "step": 2980
    },
    {
      "epoch": 0.13258636788048553,
      "grad_norm": 0.15198451280593872,
      "learning_rate": 0.0009987758041164342,
      "loss": 1.6966,
      "step": 2982
    },
    {
      "epoch": 0.1326752923391579,
      "grad_norm": 0.15600472688674927,
      "learning_rate": 0.0009987733332537053,
      "loss": 1.6958,
      "step": 2984
    },
    {
      "epoch": 0.13276421679783024,
      "grad_norm": 0.14877556264400482,
      "learning_rate": 0.0009987708599030125,
      "loss": 1.6921,
      "step": 2986
    },
    {
      "epoch": 0.1328531412565026,
      "grad_norm": 0.14063440263271332,
      "learning_rate": 0.0009987683840643677,
      "loss": 1.6859,
      "step": 2988
    },
    {
      "epoch": 0.13294206571517495,
      "grad_norm": 0.14707665145397186,
      "learning_rate": 0.0009987659057377834,
      "loss": 1.6789,
      "step": 2990
    },
    {
      "epoch": 0.1330309901738473,
      "grad_norm": 0.15149006247520447,
      "learning_rate": 0.000998763424923272,
      "loss": 1.6913,
      "step": 2992
    },
    {
      "epoch": 0.13311991463251968,
      "grad_norm": 0.14212511479854584,
      "learning_rate": 0.000998760941620846,
      "loss": 1.6941,
      "step": 2994
    },
    {
      "epoch": 0.13320883909119202,
      "grad_norm": 0.15261532366275787,
      "learning_rate": 0.0009987584558305174,
      "loss": 1.6848,
      "step": 2996
    },
    {
      "epoch": 0.1332977635498644,
      "grad_norm": 0.13972164690494537,
      "learning_rate": 0.000998755967552299,
      "loss": 1.6903,
      "step": 2998
    },
    {
      "epoch": 0.13338668800853676,
      "grad_norm": 0.16085447371006012,
      "learning_rate": 0.000998753476786203,
      "loss": 1.6976,
      "step": 3000
    },
    {
      "epoch": 0.13338668800853676,
      "eval_loss": 1.6553056240081787,
      "eval_runtime": 12.3492,
      "eval_samples_per_second": 559.55,
      "eval_steps_per_second": 69.964,
      "step": 3000
    },
    {
      "epoch": 0.1334756124672091,
      "grad_norm": 0.161129429936409,
      "learning_rate": 0.0009987509835322418,
      "loss": 1.685,
      "step": 3002
    },
    {
      "epoch": 0.13356453692588147,
      "grad_norm": 0.1477159708738327,
      "learning_rate": 0.000998748487790428,
      "loss": 1.687,
      "step": 3004
    },
    {
      "epoch": 0.13365346138455383,
      "grad_norm": 0.1458180546760559,
      "learning_rate": 0.000998745989560774,
      "loss": 1.6878,
      "step": 3006
    },
    {
      "epoch": 0.13374238584322617,
      "grad_norm": 0.15459588170051575,
      "learning_rate": 0.000998743488843292,
      "loss": 1.6907,
      "step": 3008
    },
    {
      "epoch": 0.13383131030189854,
      "grad_norm": 0.15089759230613708,
      "learning_rate": 0.000998740985637995,
      "loss": 1.6887,
      "step": 3010
    },
    {
      "epoch": 0.13392023476057088,
      "grad_norm": 0.14244233071804047,
      "learning_rate": 0.000998738479944895,
      "loss": 1.6896,
      "step": 3012
    },
    {
      "epoch": 0.13400915921924325,
      "grad_norm": 0.14067134261131287,
      "learning_rate": 0.0009987359717640047,
      "loss": 1.686,
      "step": 3014
    },
    {
      "epoch": 0.13409808367791562,
      "grad_norm": 0.1399071216583252,
      "learning_rate": 0.0009987334610953365,
      "loss": 1.6862,
      "step": 3016
    },
    {
      "epoch": 0.13418700813658796,
      "grad_norm": 0.14086857438087463,
      "learning_rate": 0.0009987309479389031,
      "loss": 1.6891,
      "step": 3018
    },
    {
      "epoch": 0.13427593259526033,
      "grad_norm": 0.14159750938415527,
      "learning_rate": 0.000998728432294717,
      "loss": 1.6865,
      "step": 3020
    },
    {
      "epoch": 0.1343648570539327,
      "grad_norm": 0.1503407210111618,
      "learning_rate": 0.0009987259141627904,
      "loss": 1.6943,
      "step": 3022
    },
    {
      "epoch": 0.13445378151260504,
      "grad_norm": 0.14096394181251526,
      "learning_rate": 0.0009987233935431366,
      "loss": 1.6877,
      "step": 3024
    },
    {
      "epoch": 0.1345427059712774,
      "grad_norm": 0.13867942988872528,
      "learning_rate": 0.0009987208704357673,
      "loss": 1.6827,
      "step": 3026
    },
    {
      "epoch": 0.13463163042994974,
      "grad_norm": 0.1480480581521988,
      "learning_rate": 0.0009987183448406957,
      "loss": 1.6863,
      "step": 3028
    },
    {
      "epoch": 0.1347205548886221,
      "grad_norm": 0.14531119167804718,
      "learning_rate": 0.0009987158167579344,
      "loss": 1.6917,
      "step": 3030
    },
    {
      "epoch": 0.13480947934729448,
      "grad_norm": 0.13942700624465942,
      "learning_rate": 0.0009987132861874956,
      "loss": 1.6876,
      "step": 3032
    },
    {
      "epoch": 0.13489840380596682,
      "grad_norm": 0.14429835975170135,
      "learning_rate": 0.0009987107531293923,
      "loss": 1.6851,
      "step": 3034
    },
    {
      "epoch": 0.1349873282646392,
      "grad_norm": 0.1691877692937851,
      "learning_rate": 0.000998708217583637,
      "loss": 1.6865,
      "step": 3036
    },
    {
      "epoch": 0.13507625272331156,
      "grad_norm": 0.15932905673980713,
      "learning_rate": 0.0009987056795502422,
      "loss": 1.6861,
      "step": 3038
    },
    {
      "epoch": 0.1351651771819839,
      "grad_norm": 0.15188652276992798,
      "learning_rate": 0.0009987031390292205,
      "loss": 1.6876,
      "step": 3040
    },
    {
      "epoch": 0.13525410164065627,
      "grad_norm": 0.1434570848941803,
      "learning_rate": 0.000998700596020585,
      "loss": 1.6842,
      "step": 3042
    },
    {
      "epoch": 0.13534302609932863,
      "grad_norm": 0.1466592699289322,
      "learning_rate": 0.000998698050524348,
      "loss": 1.6867,
      "step": 3044
    },
    {
      "epoch": 0.13543195055800097,
      "grad_norm": 0.1536661833524704,
      "learning_rate": 0.0009986955025405226,
      "loss": 1.6888,
      "step": 3046
    },
    {
      "epoch": 0.13552087501667334,
      "grad_norm": 0.14562800526618958,
      "learning_rate": 0.000998692952069121,
      "loss": 1.6877,
      "step": 3048
    },
    {
      "epoch": 0.13560979947534568,
      "grad_norm": 0.14590971171855927,
      "learning_rate": 0.0009986903991101564,
      "loss": 1.6895,
      "step": 3050
    },
    {
      "epoch": 0.13569872393401805,
      "grad_norm": 0.1405206024646759,
      "learning_rate": 0.0009986878436636412,
      "loss": 1.6892,
      "step": 3052
    },
    {
      "epoch": 0.13578764839269042,
      "grad_norm": 0.13950257003307343,
      "learning_rate": 0.0009986852857295882,
      "loss": 1.6866,
      "step": 3054
    },
    {
      "epoch": 0.13587657285136276,
      "grad_norm": 0.14230984449386597,
      "learning_rate": 0.0009986827253080101,
      "loss": 1.6809,
      "step": 3056
    },
    {
      "epoch": 0.13596549731003513,
      "grad_norm": 0.13012036681175232,
      "learning_rate": 0.00099868016239892,
      "loss": 1.6902,
      "step": 3058
    },
    {
      "epoch": 0.1360544217687075,
      "grad_norm": 0.13848593831062317,
      "learning_rate": 0.0009986775970023305,
      "loss": 1.6759,
      "step": 3060
    },
    {
      "epoch": 0.13614334622737984,
      "grad_norm": 0.14082428812980652,
      "learning_rate": 0.0009986750291182543,
      "loss": 1.6812,
      "step": 3062
    },
    {
      "epoch": 0.1362322706860522,
      "grad_norm": 0.1358221471309662,
      "learning_rate": 0.0009986724587467044,
      "loss": 1.6845,
      "step": 3064
    },
    {
      "epoch": 0.13632119514472454,
      "grad_norm": 0.13674212992191315,
      "learning_rate": 0.0009986698858876933,
      "loss": 1.6857,
      "step": 3066
    },
    {
      "epoch": 0.1364101196033969,
      "grad_norm": 0.13396227359771729,
      "learning_rate": 0.000998667310541234,
      "loss": 1.6859,
      "step": 3068
    },
    {
      "epoch": 0.13649904406206928,
      "grad_norm": 0.1398070603609085,
      "learning_rate": 0.0009986647327073397,
      "loss": 1.6786,
      "step": 3070
    },
    {
      "epoch": 0.13658796852074162,
      "grad_norm": 0.136416956782341,
      "learning_rate": 0.0009986621523860226,
      "loss": 1.6759,
      "step": 3072
    },
    {
      "epoch": 0.136676892979414,
      "grad_norm": 0.13669733703136444,
      "learning_rate": 0.0009986595695772962,
      "loss": 1.6856,
      "step": 3074
    },
    {
      "epoch": 0.13676581743808636,
      "grad_norm": 0.14353300631046295,
      "learning_rate": 0.0009986569842811727,
      "loss": 1.688,
      "step": 3076
    },
    {
      "epoch": 0.1368547418967587,
      "grad_norm": 0.1414812058210373,
      "learning_rate": 0.0009986543964976657,
      "loss": 1.6793,
      "step": 3078
    },
    {
      "epoch": 0.13694366635543107,
      "grad_norm": 0.13965506851673126,
      "learning_rate": 0.0009986518062267877,
      "loss": 1.6824,
      "step": 3080
    },
    {
      "epoch": 0.1370325908141034,
      "grad_norm": 0.14354394376277924,
      "learning_rate": 0.000998649213468552,
      "loss": 1.6847,
      "step": 3082
    },
    {
      "epoch": 0.13712151527277577,
      "grad_norm": 0.13872870802879333,
      "learning_rate": 0.0009986466182229708,
      "loss": 1.684,
      "step": 3084
    },
    {
      "epoch": 0.13721043973144814,
      "grad_norm": 0.1326591521501541,
      "learning_rate": 0.0009986440204900577,
      "loss": 1.6856,
      "step": 3086
    },
    {
      "epoch": 0.13729936419012048,
      "grad_norm": 0.14356881380081177,
      "learning_rate": 0.0009986414202698254,
      "loss": 1.6777,
      "step": 3088
    },
    {
      "epoch": 0.13738828864879285,
      "grad_norm": 0.14380203187465668,
      "learning_rate": 0.0009986388175622872,
      "loss": 1.6864,
      "step": 3090
    },
    {
      "epoch": 0.13747721310746522,
      "grad_norm": 0.3013915419578552,
      "learning_rate": 0.0009986362123674554,
      "loss": 1.6923,
      "step": 3092
    },
    {
      "epoch": 0.13756613756613756,
      "grad_norm": 0.26866304874420166,
      "learning_rate": 0.0009986336046853438,
      "loss": 1.6887,
      "step": 3094
    },
    {
      "epoch": 0.13765506202480993,
      "grad_norm": 0.16091522574424744,
      "learning_rate": 0.0009986309945159647,
      "loss": 1.6846,
      "step": 3096
    },
    {
      "epoch": 0.1377439864834823,
      "grad_norm": 0.22362981736660004,
      "learning_rate": 0.0009986283818593314,
      "loss": 1.6896,
      "step": 3098
    },
    {
      "epoch": 0.13783291094215464,
      "grad_norm": 0.2674184739589691,
      "learning_rate": 0.0009986257667154571,
      "loss": 1.6874,
      "step": 3100
    },
    {
      "epoch": 0.137921835400827,
      "grad_norm": 0.17050045728683472,
      "learning_rate": 0.0009986231490843548,
      "loss": 1.6888,
      "step": 3102
    },
    {
      "epoch": 0.13801075985949934,
      "grad_norm": 0.1663951575756073,
      "learning_rate": 0.0009986205289660372,
      "loss": 1.6818,
      "step": 3104
    },
    {
      "epoch": 0.1380996843181717,
      "grad_norm": 0.14594919979572296,
      "learning_rate": 0.000998617906360518,
      "loss": 1.6865,
      "step": 3106
    },
    {
      "epoch": 0.13818860877684408,
      "grad_norm": 0.17260988056659698,
      "learning_rate": 0.0009986152812678096,
      "loss": 1.6838,
      "step": 3108
    },
    {
      "epoch": 0.13827753323551642,
      "grad_norm": 0.15733318030834198,
      "learning_rate": 0.0009986126536879255,
      "loss": 1.6901,
      "step": 3110
    },
    {
      "epoch": 0.1383664576941888,
      "grad_norm": 0.1493389755487442,
      "learning_rate": 0.0009986100236208787,
      "loss": 1.6869,
      "step": 3112
    },
    {
      "epoch": 0.13845538215286116,
      "grad_norm": 0.16421248018741608,
      "learning_rate": 0.0009986073910666824,
      "loss": 1.6875,
      "step": 3114
    },
    {
      "epoch": 0.1385443066115335,
      "grad_norm": 0.15142826735973358,
      "learning_rate": 0.0009986047560253497,
      "loss": 1.6803,
      "step": 3116
    },
    {
      "epoch": 0.13863323107020586,
      "grad_norm": 0.148903951048851,
      "learning_rate": 0.0009986021184968936,
      "loss": 1.6907,
      "step": 3118
    },
    {
      "epoch": 0.1387221555288782,
      "grad_norm": 0.1498846411705017,
      "learning_rate": 0.0009985994784813274,
      "loss": 1.6831,
      "step": 3120
    },
    {
      "epoch": 0.13881107998755057,
      "grad_norm": 0.14996740221977234,
      "learning_rate": 0.0009985968359786644,
      "loss": 1.6851,
      "step": 3122
    },
    {
      "epoch": 0.13890000444622294,
      "grad_norm": 0.15415999293327332,
      "learning_rate": 0.0009985941909889174,
      "loss": 1.6834,
      "step": 3124
    },
    {
      "epoch": 0.13898892890489528,
      "grad_norm": 0.14532987773418427,
      "learning_rate": 0.0009985915435121,
      "loss": 1.6852,
      "step": 3126
    },
    {
      "epoch": 0.13907785336356765,
      "grad_norm": 0.1408744752407074,
      "learning_rate": 0.0009985888935482251,
      "loss": 1.6789,
      "step": 3128
    },
    {
      "epoch": 0.13916677782224002,
      "grad_norm": 0.14001396298408508,
      "learning_rate": 0.000998586241097306,
      "loss": 1.6841,
      "step": 3130
    },
    {
      "epoch": 0.13925570228091236,
      "grad_norm": 0.1425395905971527,
      "learning_rate": 0.000998583586159356,
      "loss": 1.6739,
      "step": 3132
    },
    {
      "epoch": 0.13934462673958473,
      "grad_norm": 0.13320642709732056,
      "learning_rate": 0.0009985809287343882,
      "loss": 1.6881,
      "step": 3134
    },
    {
      "epoch": 0.1394335511982571,
      "grad_norm": 0.13793528079986572,
      "learning_rate": 0.0009985782688224163,
      "loss": 1.6859,
      "step": 3136
    },
    {
      "epoch": 0.13952247565692943,
      "grad_norm": 0.14627783000469208,
      "learning_rate": 0.000998575606423453,
      "loss": 1.6844,
      "step": 3138
    },
    {
      "epoch": 0.1396114001156018,
      "grad_norm": 0.1399577409029007,
      "learning_rate": 0.0009985729415375119,
      "loss": 1.6793,
      "step": 3140
    },
    {
      "epoch": 0.13970032457427414,
      "grad_norm": 0.14193718135356903,
      "learning_rate": 0.000998570274164606,
      "loss": 1.6811,
      "step": 3142
    },
    {
      "epoch": 0.1397892490329465,
      "grad_norm": 0.13408489525318146,
      "learning_rate": 0.0009985676043047492,
      "loss": 1.6811,
      "step": 3144
    },
    {
      "epoch": 0.13987817349161888,
      "grad_norm": 0.13769420981407166,
      "learning_rate": 0.0009985649319579541,
      "loss": 1.6786,
      "step": 3146
    },
    {
      "epoch": 0.13996709795029122,
      "grad_norm": 0.1458839327096939,
      "learning_rate": 0.0009985622571242346,
      "loss": 1.6786,
      "step": 3148
    },
    {
      "epoch": 0.1400560224089636,
      "grad_norm": 0.15078197419643402,
      "learning_rate": 0.0009985595798036036,
      "loss": 1.6774,
      "step": 3150
    },
    {
      "epoch": 0.14014494686763596,
      "grad_norm": 0.13975287973880768,
      "learning_rate": 0.000998556899996075,
      "loss": 1.677,
      "step": 3152
    },
    {
      "epoch": 0.1402338713263083,
      "grad_norm": 0.13561749458312988,
      "learning_rate": 0.0009985542177016614,
      "loss": 1.6851,
      "step": 3154
    },
    {
      "epoch": 0.14032279578498066,
      "grad_norm": 0.1398334950208664,
      "learning_rate": 0.0009985515329203767,
      "loss": 1.6826,
      "step": 3156
    },
    {
      "epoch": 0.140411720243653,
      "grad_norm": 0.137189120054245,
      "learning_rate": 0.0009985488456522341,
      "loss": 1.6829,
      "step": 3158
    },
    {
      "epoch": 0.14050064470232537,
      "grad_norm": 0.14057347178459167,
      "learning_rate": 0.0009985461558972475,
      "loss": 1.6825,
      "step": 3160
    },
    {
      "epoch": 0.14058956916099774,
      "grad_norm": 0.14627771079540253,
      "learning_rate": 0.0009985434636554295,
      "loss": 1.6822,
      "step": 3162
    },
    {
      "epoch": 0.14067849361967008,
      "grad_norm": 0.14206919074058533,
      "learning_rate": 0.0009985407689267942,
      "loss": 1.6782,
      "step": 3164
    },
    {
      "epoch": 0.14076741807834245,
      "grad_norm": 0.13843832910060883,
      "learning_rate": 0.0009985380717113547,
      "loss": 1.6851,
      "step": 3166
    },
    {
      "epoch": 0.14085634253701482,
      "grad_norm": 0.13097672164440155,
      "learning_rate": 0.0009985353720091245,
      "loss": 1.6769,
      "step": 3168
    },
    {
      "epoch": 0.14094526699568716,
      "grad_norm": 0.1339416652917862,
      "learning_rate": 0.0009985326698201174,
      "loss": 1.6851,
      "step": 3170
    },
    {
      "epoch": 0.14103419145435953,
      "grad_norm": 0.1318987011909485,
      "learning_rate": 0.0009985299651443463,
      "loss": 1.6804,
      "step": 3172
    },
    {
      "epoch": 0.14112311591303187,
      "grad_norm": 0.12795791029930115,
      "learning_rate": 0.000998527257981825,
      "loss": 1.6775,
      "step": 3174
    },
    {
      "epoch": 0.14121204037170423,
      "grad_norm": 0.1405027210712433,
      "learning_rate": 0.000998524548332567,
      "loss": 1.6753,
      "step": 3176
    },
    {
      "epoch": 0.1413009648303766,
      "grad_norm": 0.13704566657543182,
      "learning_rate": 0.000998521836196586,
      "loss": 1.6747,
      "step": 3178
    },
    {
      "epoch": 0.14138988928904894,
      "grad_norm": 0.13784196972846985,
      "learning_rate": 0.000998519121573895,
      "loss": 1.6833,
      "step": 3180
    },
    {
      "epoch": 0.1414788137477213,
      "grad_norm": 0.14620931446552277,
      "learning_rate": 0.0009985164044645082,
      "loss": 1.6831,
      "step": 3182
    },
    {
      "epoch": 0.14156773820639368,
      "grad_norm": 0.13323108851909637,
      "learning_rate": 0.0009985136848684386,
      "loss": 1.6781,
      "step": 3184
    },
    {
      "epoch": 0.14165666266506602,
      "grad_norm": 0.13440033793449402,
      "learning_rate": 0.0009985109627857,
      "loss": 1.6794,
      "step": 3186
    },
    {
      "epoch": 0.1417455871237384,
      "grad_norm": 0.13603028655052185,
      "learning_rate": 0.000998508238216306,
      "loss": 1.6712,
      "step": 3188
    },
    {
      "epoch": 0.14183451158241075,
      "grad_norm": 0.14003223180770874,
      "learning_rate": 0.0009985055111602704,
      "loss": 1.6754,
      "step": 3190
    },
    {
      "epoch": 0.1419234360410831,
      "grad_norm": 0.13278284668922424,
      "learning_rate": 0.0009985027816176062,
      "loss": 1.6805,
      "step": 3192
    },
    {
      "epoch": 0.14201236049975546,
      "grad_norm": 0.1399509310722351,
      "learning_rate": 0.0009985000495883276,
      "loss": 1.6724,
      "step": 3194
    },
    {
      "epoch": 0.1421012849584278,
      "grad_norm": 0.14798682928085327,
      "learning_rate": 0.0009984973150724478,
      "loss": 1.6767,
      "step": 3196
    },
    {
      "epoch": 0.14219020941710017,
      "grad_norm": 0.14598892629146576,
      "learning_rate": 0.000998494578069981,
      "loss": 1.6795,
      "step": 3198
    },
    {
      "epoch": 0.14227913387577254,
      "grad_norm": 0.14363674819469452,
      "learning_rate": 0.0009984918385809402,
      "loss": 1.6739,
      "step": 3200
    },
    {
      "epoch": 0.14236805833444488,
      "grad_norm": 0.14036902785301208,
      "learning_rate": 0.0009984890966053396,
      "loss": 1.6797,
      "step": 3202
    },
    {
      "epoch": 0.14245698279311725,
      "grad_norm": 0.13619792461395264,
      "learning_rate": 0.0009984863521431925,
      "loss": 1.6821,
      "step": 3204
    },
    {
      "epoch": 0.14254590725178962,
      "grad_norm": 0.14569631218910217,
      "learning_rate": 0.0009984836051945124,
      "loss": 1.6698,
      "step": 3206
    },
    {
      "epoch": 0.14263483171046196,
      "grad_norm": 0.14593179523944855,
      "learning_rate": 0.0009984808557593138,
      "loss": 1.6768,
      "step": 3208
    },
    {
      "epoch": 0.14272375616913432,
      "grad_norm": 0.1467268168926239,
      "learning_rate": 0.0009984781038376099,
      "loss": 1.6805,
      "step": 3210
    },
    {
      "epoch": 0.14281268062780667,
      "grad_norm": 0.13134878873825073,
      "learning_rate": 0.0009984753494294144,
      "loss": 1.6728,
      "step": 3212
    },
    {
      "epoch": 0.14290160508647903,
      "grad_norm": 0.13109003007411957,
      "learning_rate": 0.000998472592534741,
      "loss": 1.6745,
      "step": 3214
    },
    {
      "epoch": 0.1429905295451514,
      "grad_norm": 0.14403113722801208,
      "learning_rate": 0.0009984698331536035,
      "loss": 1.6733,
      "step": 3216
    },
    {
      "epoch": 0.14307945400382374,
      "grad_norm": 0.14508023858070374,
      "learning_rate": 0.000998467071286016,
      "loss": 1.6752,
      "step": 3218
    },
    {
      "epoch": 0.1431683784624961,
      "grad_norm": 0.13933469355106354,
      "learning_rate": 0.0009984643069319919,
      "loss": 1.676,
      "step": 3220
    },
    {
      "epoch": 0.14325730292116848,
      "grad_norm": 0.13462568819522858,
      "learning_rate": 0.000998461540091545,
      "loss": 1.6755,
      "step": 3222
    },
    {
      "epoch": 0.14334622737984082,
      "grad_norm": 0.13455134630203247,
      "learning_rate": 0.0009984587707646893,
      "loss": 1.678,
      "step": 3224
    },
    {
      "epoch": 0.1434351518385132,
      "grad_norm": 0.13943925499916077,
      "learning_rate": 0.0009984559989514386,
      "loss": 1.6716,
      "step": 3226
    },
    {
      "epoch": 0.14352407629718555,
      "grad_norm": 0.14231517910957336,
      "learning_rate": 0.0009984532246518065,
      "loss": 1.6776,
      "step": 3228
    },
    {
      "epoch": 0.1436130007558579,
      "grad_norm": 0.14278076589107513,
      "learning_rate": 0.000998450447865807,
      "loss": 1.6738,
      "step": 3230
    },
    {
      "epoch": 0.14370192521453026,
      "grad_norm": 0.1441042423248291,
      "learning_rate": 0.000998447668593454,
      "loss": 1.6722,
      "step": 3232
    },
    {
      "epoch": 0.1437908496732026,
      "grad_norm": 0.13601696491241455,
      "learning_rate": 0.000998444886834761,
      "loss": 1.6692,
      "step": 3234
    },
    {
      "epoch": 0.14387977413187497,
      "grad_norm": 0.14175257086753845,
      "learning_rate": 0.0009984421025897425,
      "loss": 1.6691,
      "step": 3236
    },
    {
      "epoch": 0.14396869859054734,
      "grad_norm": 0.1315532624721527,
      "learning_rate": 0.000998439315858412,
      "loss": 1.6729,
      "step": 3238
    },
    {
      "epoch": 0.14405762304921968,
      "grad_norm": 0.13552317023277283,
      "learning_rate": 0.0009984365266407834,
      "loss": 1.6766,
      "step": 3240
    },
    {
      "epoch": 0.14414654750789205,
      "grad_norm": 0.13469535112380981,
      "learning_rate": 0.0009984337349368707,
      "loss": 1.6747,
      "step": 3242
    },
    {
      "epoch": 0.14423547196656442,
      "grad_norm": 0.1256866753101349,
      "learning_rate": 0.0009984309407466878,
      "loss": 1.6759,
      "step": 3244
    },
    {
      "epoch": 0.14432439642523676,
      "grad_norm": 0.13308246433734894,
      "learning_rate": 0.0009984281440702486,
      "loss": 1.6753,
      "step": 3246
    },
    {
      "epoch": 0.14441332088390912,
      "grad_norm": 0.134820818901062,
      "learning_rate": 0.0009984253449075672,
      "loss": 1.683,
      "step": 3248
    },
    {
      "epoch": 0.14450224534258146,
      "grad_norm": 0.13477694988250732,
      "learning_rate": 0.0009984225432586571,
      "loss": 1.6735,
      "step": 3250
    },
    {
      "epoch": 0.14459116980125383,
      "grad_norm": 0.13332480192184448,
      "learning_rate": 0.000998419739123533,
      "loss": 1.6705,
      "step": 3252
    },
    {
      "epoch": 0.1446800942599262,
      "grad_norm": 0.13237187266349792,
      "learning_rate": 0.0009984169325022083,
      "loss": 1.6745,
      "step": 3254
    },
    {
      "epoch": 0.14476901871859854,
      "grad_norm": 0.1395091861486435,
      "learning_rate": 0.0009984141233946972,
      "loss": 1.6736,
      "step": 3256
    },
    {
      "epoch": 0.1448579431772709,
      "grad_norm": 0.13899393379688263,
      "learning_rate": 0.0009984113118010139,
      "loss": 1.6721,
      "step": 3258
    },
    {
      "epoch": 0.14494686763594328,
      "grad_norm": 0.1368132382631302,
      "learning_rate": 0.0009984084977211721,
      "loss": 1.6735,
      "step": 3260
    },
    {
      "epoch": 0.14503579209461562,
      "grad_norm": 0.13790659606456757,
      "learning_rate": 0.000998405681155186,
      "loss": 1.6792,
      "step": 3262
    },
    {
      "epoch": 0.14512471655328799,
      "grad_norm": 0.13561876118183136,
      "learning_rate": 0.0009984028621030697,
      "loss": 1.6753,
      "step": 3264
    },
    {
      "epoch": 0.14521364101196033,
      "grad_norm": 0.14658573269844055,
      "learning_rate": 0.000998400040564837,
      "loss": 1.6835,
      "step": 3266
    },
    {
      "epoch": 0.1453025654706327,
      "grad_norm": 0.14082130789756775,
      "learning_rate": 0.0009983972165405024,
      "loss": 1.6713,
      "step": 3268
    },
    {
      "epoch": 0.14539148992930506,
      "grad_norm": 0.13444170355796814,
      "learning_rate": 0.0009983943900300795,
      "loss": 1.6771,
      "step": 3270
    },
    {
      "epoch": 0.1454804143879774,
      "grad_norm": 0.1398838758468628,
      "learning_rate": 0.000998391561033583,
      "loss": 1.677,
      "step": 3272
    },
    {
      "epoch": 0.14556933884664977,
      "grad_norm": 0.1403886079788208,
      "learning_rate": 0.0009983887295510264,
      "loss": 1.6713,
      "step": 3274
    },
    {
      "epoch": 0.14565826330532214,
      "grad_norm": 0.12883295118808746,
      "learning_rate": 0.000998385895582424,
      "loss": 1.6749,
      "step": 3276
    },
    {
      "epoch": 0.14574718776399448,
      "grad_norm": 0.1351172924041748,
      "learning_rate": 0.0009983830591277902,
      "loss": 1.669,
      "step": 3278
    },
    {
      "epoch": 0.14583611222266685,
      "grad_norm": 0.12804821133613586,
      "learning_rate": 0.000998380220187139,
      "loss": 1.6777,
      "step": 3280
    },
    {
      "epoch": 0.14592503668133922,
      "grad_norm": 0.13905996084213257,
      "learning_rate": 0.0009983773787604847,
      "loss": 1.6725,
      "step": 3282
    },
    {
      "epoch": 0.14601396114001156,
      "grad_norm": 0.14166028797626495,
      "learning_rate": 0.000998374534847841,
      "loss": 1.6792,
      "step": 3284
    },
    {
      "epoch": 0.14610288559868392,
      "grad_norm": 0.13185730576515198,
      "learning_rate": 0.0009983716884492224,
      "loss": 1.6752,
      "step": 3286
    },
    {
      "epoch": 0.14619181005735626,
      "grad_norm": 0.12768588960170746,
      "learning_rate": 0.0009983688395646434,
      "loss": 1.6759,
      "step": 3288
    },
    {
      "epoch": 0.14628073451602863,
      "grad_norm": 0.13263854384422302,
      "learning_rate": 0.0009983659881941177,
      "loss": 1.6696,
      "step": 3290
    },
    {
      "epoch": 0.146369658974701,
      "grad_norm": 0.12498720735311508,
      "learning_rate": 0.0009983631343376596,
      "loss": 1.674,
      "step": 3292
    },
    {
      "epoch": 0.14645858343337334,
      "grad_norm": 0.13268351554870605,
      "learning_rate": 0.0009983602779952837,
      "loss": 1.6687,
      "step": 3294
    },
    {
      "epoch": 0.1465475078920457,
      "grad_norm": 0.13163034617900848,
      "learning_rate": 0.0009983574191670037,
      "loss": 1.667,
      "step": 3296
    },
    {
      "epoch": 0.14663643235071808,
      "grad_norm": 0.13976821303367615,
      "learning_rate": 0.0009983545578528344,
      "loss": 1.6731,
      "step": 3298
    },
    {
      "epoch": 0.14672535680939042,
      "grad_norm": 0.1371973156929016,
      "learning_rate": 0.0009983516940527898,
      "loss": 1.6723,
      "step": 3300
    },
    {
      "epoch": 0.14681428126806279,
      "grad_norm": 0.13065384328365326,
      "learning_rate": 0.0009983488277668841,
      "loss": 1.6682,
      "step": 3302
    },
    {
      "epoch": 0.14690320572673513,
      "grad_norm": 0.1396087110042572,
      "learning_rate": 0.000998345958995132,
      "loss": 1.6707,
      "step": 3304
    },
    {
      "epoch": 0.1469921301854075,
      "grad_norm": 0.13367894291877747,
      "learning_rate": 0.0009983430877375475,
      "loss": 1.6715,
      "step": 3306
    },
    {
      "epoch": 0.14708105464407986,
      "grad_norm": 0.14132438600063324,
      "learning_rate": 0.0009983402139941448,
      "loss": 1.6709,
      "step": 3308
    },
    {
      "epoch": 0.1471699791027522,
      "grad_norm": 0.13156946003437042,
      "learning_rate": 0.0009983373377649384,
      "loss": 1.6656,
      "step": 3310
    },
    {
      "epoch": 0.14725890356142457,
      "grad_norm": 0.12967397272586823,
      "learning_rate": 0.0009983344590499426,
      "loss": 1.6671,
      "step": 3312
    },
    {
      "epoch": 0.14734782802009694,
      "grad_norm": 0.12589845061302185,
      "learning_rate": 0.000998331577849172,
      "loss": 1.6755,
      "step": 3314
    },
    {
      "epoch": 0.14743675247876928,
      "grad_norm": 0.12698271870613098,
      "learning_rate": 0.0009983286941626407,
      "loss": 1.6676,
      "step": 3316
    },
    {
      "epoch": 0.14752567693744165,
      "grad_norm": 0.1336132436990738,
      "learning_rate": 0.000998325807990363,
      "loss": 1.6753,
      "step": 3318
    },
    {
      "epoch": 0.14761460139611401,
      "grad_norm": 0.1344204992055893,
      "learning_rate": 0.0009983229193323536,
      "loss": 1.6665,
      "step": 3320
    },
    {
      "epoch": 0.14770352585478635,
      "grad_norm": 0.13292238116264343,
      "learning_rate": 0.0009983200281886266,
      "loss": 1.6717,
      "step": 3322
    },
    {
      "epoch": 0.14779245031345872,
      "grad_norm": 0.13287590444087982,
      "learning_rate": 0.0009983171345591968,
      "loss": 1.6657,
      "step": 3324
    },
    {
      "epoch": 0.14788137477213106,
      "grad_norm": 0.13067129254341125,
      "learning_rate": 0.0009983142384440782,
      "loss": 1.6659,
      "step": 3326
    },
    {
      "epoch": 0.14797029923080343,
      "grad_norm": 0.1255059391260147,
      "learning_rate": 0.0009983113398432857,
      "loss": 1.6669,
      "step": 3328
    },
    {
      "epoch": 0.1480592236894758,
      "grad_norm": 0.13463814556598663,
      "learning_rate": 0.0009983084387568334,
      "loss": 1.6683,
      "step": 3330
    },
    {
      "epoch": 0.14814814814814814,
      "grad_norm": 0.13709482550621033,
      "learning_rate": 0.0009983055351847358,
      "loss": 1.6648,
      "step": 3332
    },
    {
      "epoch": 0.1482370726068205,
      "grad_norm": 0.13506925106048584,
      "learning_rate": 0.0009983026291270076,
      "loss": 1.6638,
      "step": 3334
    },
    {
      "epoch": 0.14832599706549288,
      "grad_norm": 0.13321250677108765,
      "learning_rate": 0.0009982997205836633,
      "loss": 1.6754,
      "step": 3336
    },
    {
      "epoch": 0.14841492152416522,
      "grad_norm": 0.12868818640708923,
      "learning_rate": 0.000998296809554717,
      "loss": 1.6656,
      "step": 3338
    },
    {
      "epoch": 0.14850384598283758,
      "grad_norm": 0.12971720099449158,
      "learning_rate": 0.0009982938960401838,
      "loss": 1.6697,
      "step": 3340
    },
    {
      "epoch": 0.14859277044150992,
      "grad_norm": 0.13777963817119598,
      "learning_rate": 0.0009982909800400777,
      "loss": 1.6691,
      "step": 3342
    },
    {
      "epoch": 0.1486816949001823,
      "grad_norm": 0.134913831949234,
      "learning_rate": 0.0009982880615544134,
      "loss": 1.6654,
      "step": 3344
    },
    {
      "epoch": 0.14877061935885466,
      "grad_norm": 0.14073821902275085,
      "learning_rate": 0.0009982851405832057,
      "loss": 1.6667,
      "step": 3346
    },
    {
      "epoch": 0.148859543817527,
      "grad_norm": 0.12406160682439804,
      "learning_rate": 0.000998282217126469,
      "loss": 1.6641,
      "step": 3348
    },
    {
      "epoch": 0.14894846827619937,
      "grad_norm": 0.13188298046588898,
      "learning_rate": 0.000998279291184218,
      "loss": 1.6682,
      "step": 3350
    },
    {
      "epoch": 0.14903739273487174,
      "grad_norm": 0.13892415165901184,
      "learning_rate": 0.000998276362756467,
      "loss": 1.6669,
      "step": 3352
    },
    {
      "epoch": 0.14912631719354408,
      "grad_norm": 0.13346289098262787,
      "learning_rate": 0.0009982734318432307,
      "loss": 1.6626,
      "step": 3354
    },
    {
      "epoch": 0.14921524165221645,
      "grad_norm": 0.1317860335111618,
      "learning_rate": 0.000998270498444524,
      "loss": 1.664,
      "step": 3356
    },
    {
      "epoch": 0.1493041661108888,
      "grad_norm": 0.13750551640987396,
      "learning_rate": 0.0009982675625603613,
      "loss": 1.669,
      "step": 3358
    },
    {
      "epoch": 0.14939309056956115,
      "grad_norm": 0.13547618687152863,
      "learning_rate": 0.0009982646241907575,
      "loss": 1.664,
      "step": 3360
    },
    {
      "epoch": 0.14948201502823352,
      "grad_norm": 0.13562791049480438,
      "learning_rate": 0.0009982616833357268,
      "loss": 1.6637,
      "step": 3362
    },
    {
      "epoch": 0.14957093948690586,
      "grad_norm": 0.13718074560165405,
      "learning_rate": 0.0009982587399952841,
      "loss": 1.6725,
      "step": 3364
    },
    {
      "epoch": 0.14965986394557823,
      "grad_norm": 0.14120802283287048,
      "learning_rate": 0.0009982557941694442,
      "loss": 1.6668,
      "step": 3366
    },
    {
      "epoch": 0.1497487884042506,
      "grad_norm": 0.13966819643974304,
      "learning_rate": 0.0009982528458582216,
      "loss": 1.6678,
      "step": 3368
    },
    {
      "epoch": 0.14983771286292294,
      "grad_norm": 0.13596081733703613,
      "learning_rate": 0.000998249895061631,
      "loss": 1.6628,
      "step": 3370
    },
    {
      "epoch": 0.1499266373215953,
      "grad_norm": 0.1467939019203186,
      "learning_rate": 0.0009982469417796874,
      "loss": 1.6667,
      "step": 3372
    },
    {
      "epoch": 0.15001556178026768,
      "grad_norm": 0.13276167213916779,
      "learning_rate": 0.0009982439860124054,
      "loss": 1.6716,
      "step": 3374
    },
    {
      "epoch": 0.15010448623894002,
      "grad_norm": 0.1353362649679184,
      "learning_rate": 0.0009982410277597994,
      "loss": 1.6746,
      "step": 3376
    },
    {
      "epoch": 0.15019341069761238,
      "grad_norm": 0.13252989947795868,
      "learning_rate": 0.0009982380670218848,
      "loss": 1.6693,
      "step": 3378
    },
    {
      "epoch": 0.15028233515628472,
      "grad_norm": 0.1437995731830597,
      "learning_rate": 0.000998235103798676,
      "loss": 1.6689,
      "step": 3380
    },
    {
      "epoch": 0.1503712596149571,
      "grad_norm": 0.13310275971889496,
      "learning_rate": 0.0009982321380901874,
      "loss": 1.6703,
      "step": 3382
    },
    {
      "epoch": 0.15046018407362946,
      "grad_norm": 0.13596755266189575,
      "learning_rate": 0.0009982291698964348,
      "loss": 1.673,
      "step": 3384
    },
    {
      "epoch": 0.1505491085323018,
      "grad_norm": 0.13616810739040375,
      "learning_rate": 0.000998226199217432,
      "loss": 1.67,
      "step": 3386
    },
    {
      "epoch": 0.15063803299097417,
      "grad_norm": 0.13550317287445068,
      "learning_rate": 0.0009982232260531943,
      "loss": 1.663,
      "step": 3388
    },
    {
      "epoch": 0.15072695744964654,
      "grad_norm": 0.14142318069934845,
      "learning_rate": 0.0009982202504037363,
      "loss": 1.6656,
      "step": 3390
    },
    {
      "epoch": 0.15081588190831888,
      "grad_norm": 0.13731466233730316,
      "learning_rate": 0.0009982172722690732,
      "loss": 1.6683,
      "step": 3392
    },
    {
      "epoch": 0.15090480636699125,
      "grad_norm": 0.14071650803089142,
      "learning_rate": 0.0009982142916492196,
      "loss": 1.6575,
      "step": 3394
    },
    {
      "epoch": 0.15099373082566359,
      "grad_norm": 0.13527441024780273,
      "learning_rate": 0.0009982113085441903,
      "loss": 1.6707,
      "step": 3396
    },
    {
      "epoch": 0.15108265528433595,
      "grad_norm": 0.1322237253189087,
      "learning_rate": 0.0009982083229540003,
      "loss": 1.6704,
      "step": 3398
    },
    {
      "epoch": 0.15117157974300832,
      "grad_norm": 0.13059936463832855,
      "learning_rate": 0.0009982053348786648,
      "loss": 1.6671,
      "step": 3400
    },
    {
      "epoch": 0.15126050420168066,
      "grad_norm": 0.1345200389623642,
      "learning_rate": 0.000998202344318198,
      "loss": 1.6644,
      "step": 3402
    },
    {
      "epoch": 0.15134942866035303,
      "grad_norm": 0.13381880521774292,
      "learning_rate": 0.0009981993512726153,
      "loss": 1.6649,
      "step": 3404
    },
    {
      "epoch": 0.1514383531190254,
      "grad_norm": 0.12679660320281982,
      "learning_rate": 0.0009981963557419318,
      "loss": 1.6672,
      "step": 3406
    },
    {
      "epoch": 0.15152727757769774,
      "grad_norm": 0.12676650285720825,
      "learning_rate": 0.000998193357726162,
      "loss": 1.6654,
      "step": 3408
    },
    {
      "epoch": 0.1516162020363701,
      "grad_norm": 0.13394631445407867,
      "learning_rate": 0.0009981903572253209,
      "loss": 1.6672,
      "step": 3410
    },
    {
      "epoch": 0.15170512649504247,
      "grad_norm": 0.1335240751504898,
      "learning_rate": 0.0009981873542394238,
      "loss": 1.6594,
      "step": 3412
    },
    {
      "epoch": 0.15179405095371482,
      "grad_norm": 0.13069608807563782,
      "learning_rate": 0.0009981843487684855,
      "loss": 1.6684,
      "step": 3414
    },
    {
      "epoch": 0.15188297541238718,
      "grad_norm": 0.1335192620754242,
      "learning_rate": 0.0009981813408125207,
      "loss": 1.6606,
      "step": 3416
    },
    {
      "epoch": 0.15197189987105952,
      "grad_norm": 0.13265785574913025,
      "learning_rate": 0.0009981783303715447,
      "loss": 1.6701,
      "step": 3418
    },
    {
      "epoch": 0.1520608243297319,
      "grad_norm": 0.13182604312896729,
      "learning_rate": 0.0009981753174455727,
      "loss": 1.6665,
      "step": 3420
    },
    {
      "epoch": 0.15214974878840426,
      "grad_norm": 0.13200457394123077,
      "learning_rate": 0.0009981723020346196,
      "loss": 1.6655,
      "step": 3422
    },
    {
      "epoch": 0.1522386732470766,
      "grad_norm": 0.13492505252361298,
      "learning_rate": 0.0009981692841387001,
      "loss": 1.669,
      "step": 3424
    },
    {
      "epoch": 0.15232759770574897,
      "grad_norm": 0.13645698130130768,
      "learning_rate": 0.0009981662637578298,
      "loss": 1.6681,
      "step": 3426
    },
    {
      "epoch": 0.15241652216442134,
      "grad_norm": 0.1314222663640976,
      "learning_rate": 0.0009981632408920234,
      "loss": 1.6653,
      "step": 3428
    },
    {
      "epoch": 0.15250544662309368,
      "grad_norm": 0.1272493451833725,
      "learning_rate": 0.000998160215541296,
      "loss": 1.6654,
      "step": 3430
    },
    {
      "epoch": 0.15259437108176604,
      "grad_norm": 0.13651682436466217,
      "learning_rate": 0.0009981571877056628,
      "loss": 1.6594,
      "step": 3432
    },
    {
      "epoch": 0.15268329554043839,
      "grad_norm": 0.137037456035614,
      "learning_rate": 0.0009981541573851388,
      "loss": 1.6631,
      "step": 3434
    },
    {
      "epoch": 0.15277221999911075,
      "grad_norm": 0.13613489270210266,
      "learning_rate": 0.0009981511245797392,
      "loss": 1.6649,
      "step": 3436
    },
    {
      "epoch": 0.15286114445778312,
      "grad_norm": 0.14442335069179535,
      "learning_rate": 0.000998148089289479,
      "loss": 1.6635,
      "step": 3438
    },
    {
      "epoch": 0.15295006891645546,
      "grad_norm": 0.13192929327487946,
      "learning_rate": 0.0009981450515143735,
      "loss": 1.6667,
      "step": 3440
    },
    {
      "epoch": 0.15303899337512783,
      "grad_norm": 0.13971135020256042,
      "learning_rate": 0.000998142011254438,
      "loss": 1.6672,
      "step": 3442
    },
    {
      "epoch": 0.1531279178338002,
      "grad_norm": 0.1356842964887619,
      "learning_rate": 0.0009981389685096871,
      "loss": 1.6578,
      "step": 3444
    },
    {
      "epoch": 0.15321684229247254,
      "grad_norm": 0.12647144496440887,
      "learning_rate": 0.0009981359232801366,
      "loss": 1.6622,
      "step": 3446
    },
    {
      "epoch": 0.1533057667511449,
      "grad_norm": 0.1329360455274582,
      "learning_rate": 0.0009981328755658014,
      "loss": 1.6615,
      "step": 3448
    },
    {
      "epoch": 0.15339469120981725,
      "grad_norm": 0.13213002681732178,
      "learning_rate": 0.0009981298253666966,
      "loss": 1.6571,
      "step": 3450
    },
    {
      "epoch": 0.15348361566848961,
      "grad_norm": 0.13628779351711273,
      "learning_rate": 0.0009981267726828379,
      "loss": 1.6632,
      "step": 3452
    },
    {
      "epoch": 0.15357254012716198,
      "grad_norm": 0.1299857646226883,
      "learning_rate": 0.0009981237175142397,
      "loss": 1.6656,
      "step": 3454
    },
    {
      "epoch": 0.15366146458583432,
      "grad_norm": 0.14375044405460358,
      "learning_rate": 0.000998120659860918,
      "loss": 1.6644,
      "step": 3456
    },
    {
      "epoch": 0.1537503890445067,
      "grad_norm": 0.1285606473684311,
      "learning_rate": 0.0009981175997228876,
      "loss": 1.6627,
      "step": 3458
    },
    {
      "epoch": 0.15383931350317906,
      "grad_norm": 0.1306368112564087,
      "learning_rate": 0.000998114537100164,
      "loss": 1.6652,
      "step": 3460
    },
    {
      "epoch": 0.1539282379618514,
      "grad_norm": 0.1342526525259018,
      "learning_rate": 0.0009981114719927625,
      "loss": 1.6633,
      "step": 3462
    },
    {
      "epoch": 0.15401716242052377,
      "grad_norm": 0.13883987069129944,
      "learning_rate": 0.0009981084044006982,
      "loss": 1.6666,
      "step": 3464
    },
    {
      "epoch": 0.15410608687919614,
      "grad_norm": 0.12659454345703125,
      "learning_rate": 0.0009981053343239865,
      "loss": 1.6635,
      "step": 3466
    },
    {
      "epoch": 0.15419501133786848,
      "grad_norm": 0.13172872364521027,
      "learning_rate": 0.0009981022617626427,
      "loss": 1.6643,
      "step": 3468
    },
    {
      "epoch": 0.15428393579654084,
      "grad_norm": 0.1316794753074646,
      "learning_rate": 0.0009980991867166822,
      "loss": 1.6593,
      "step": 3470
    },
    {
      "epoch": 0.15437286025521318,
      "grad_norm": 0.14180782437324524,
      "learning_rate": 0.0009980961091861202,
      "loss": 1.6608,
      "step": 3472
    },
    {
      "epoch": 0.15446178471388555,
      "grad_norm": 0.13621553778648376,
      "learning_rate": 0.0009980930291709723,
      "loss": 1.6631,
      "step": 3474
    },
    {
      "epoch": 0.15455070917255792,
      "grad_norm": 0.12693116068840027,
      "learning_rate": 0.0009980899466712535,
      "loss": 1.6623,
      "step": 3476
    },
    {
      "epoch": 0.15463963363123026,
      "grad_norm": 0.12886109948158264,
      "learning_rate": 0.0009980868616869794,
      "loss": 1.6589,
      "step": 3478
    },
    {
      "epoch": 0.15472855808990263,
      "grad_norm": 0.12660932540893555,
      "learning_rate": 0.0009980837742181654,
      "loss": 1.6668,
      "step": 3480
    },
    {
      "epoch": 0.154817482548575,
      "grad_norm": 0.12759321928024292,
      "learning_rate": 0.0009980806842648268,
      "loss": 1.66,
      "step": 3482
    },
    {
      "epoch": 0.15490640700724734,
      "grad_norm": 0.12718096375465393,
      "learning_rate": 0.0009980775918269791,
      "loss": 1.6659,
      "step": 3484
    },
    {
      "epoch": 0.1549953314659197,
      "grad_norm": 0.12915131449699402,
      "learning_rate": 0.0009980744969046377,
      "loss": 1.6595,
      "step": 3486
    },
    {
      "epoch": 0.15508425592459205,
      "grad_norm": 0.12834011018276215,
      "learning_rate": 0.0009980713994978182,
      "loss": 1.6652,
      "step": 3488
    },
    {
      "epoch": 0.15517318038326441,
      "grad_norm": 0.12587936222553253,
      "learning_rate": 0.0009980682996065355,
      "loss": 1.6654,
      "step": 3490
    },
    {
      "epoch": 0.15526210484193678,
      "grad_norm": 0.13400059938430786,
      "learning_rate": 0.0009980651972308057,
      "loss": 1.6603,
      "step": 3492
    },
    {
      "epoch": 0.15535102930060912,
      "grad_norm": 0.13040411472320557,
      "learning_rate": 0.000998062092370644,
      "loss": 1.6602,
      "step": 3494
    },
    {
      "epoch": 0.1554399537592815,
      "grad_norm": 0.12811923027038574,
      "learning_rate": 0.000998058985026066,
      "loss": 1.6603,
      "step": 3496
    },
    {
      "epoch": 0.15552887821795386,
      "grad_norm": 0.12436926364898682,
      "learning_rate": 0.000998055875197087,
      "loss": 1.6562,
      "step": 3498
    },
    {
      "epoch": 0.1556178026766262,
      "grad_norm": 0.13370974361896515,
      "learning_rate": 0.0009980527628837227,
      "loss": 1.666,
      "step": 3500
    },
    {
      "epoch": 0.1556178026766262,
      "eval_loss": 1.6284940242767334,
      "eval_runtime": 12.3496,
      "eval_samples_per_second": 559.532,
      "eval_steps_per_second": 69.962,
      "step": 3500
    },
    {
      "epoch": 0.15570672713529857,
      "grad_norm": 0.13278929889202118,
      "learning_rate": 0.0009980496480859883,
      "loss": 1.6623,
      "step": 3502
    },
    {
      "epoch": 0.15579565159397094,
      "grad_norm": 0.13166457414627075,
      "learning_rate": 0.0009980465308038998,
      "loss": 1.6608,
      "step": 3504
    },
    {
      "epoch": 0.15588457605264328,
      "grad_norm": 0.12355690449476242,
      "learning_rate": 0.0009980434110374724,
      "loss": 1.6639,
      "step": 3506
    },
    {
      "epoch": 0.15597350051131564,
      "grad_norm": 0.12370448559522629,
      "learning_rate": 0.000998040288786722,
      "loss": 1.6587,
      "step": 3508
    },
    {
      "epoch": 0.15606242496998798,
      "grad_norm": 0.12431325763463974,
      "learning_rate": 0.0009980371640516638,
      "loss": 1.6608,
      "step": 3510
    },
    {
      "epoch": 0.15615134942866035,
      "grad_norm": 0.1281343400478363,
      "learning_rate": 0.0009980340368323135,
      "loss": 1.6655,
      "step": 3512
    },
    {
      "epoch": 0.15624027388733272,
      "grad_norm": 0.12553107738494873,
      "learning_rate": 0.0009980309071286868,
      "loss": 1.6553,
      "step": 3514
    },
    {
      "epoch": 0.15632919834600506,
      "grad_norm": 0.1262499839067459,
      "learning_rate": 0.0009980277749407995,
      "loss": 1.6578,
      "step": 3516
    },
    {
      "epoch": 0.15641812280467743,
      "grad_norm": 0.12978388369083405,
      "learning_rate": 0.0009980246402686668,
      "loss": 1.6575,
      "step": 3518
    },
    {
      "epoch": 0.1565070472633498,
      "grad_norm": 0.1310720145702362,
      "learning_rate": 0.0009980215031123046,
      "loss": 1.6541,
      "step": 3520
    },
    {
      "epoch": 0.15659597172202214,
      "grad_norm": 0.12396594882011414,
      "learning_rate": 0.0009980183634717284,
      "loss": 1.655,
      "step": 3522
    },
    {
      "epoch": 0.1566848961806945,
      "grad_norm": 0.1306004524230957,
      "learning_rate": 0.0009980152213469539,
      "loss": 1.6586,
      "step": 3524
    },
    {
      "epoch": 0.15677382063936685,
      "grad_norm": 0.13290682435035706,
      "learning_rate": 0.000998012076737997,
      "loss": 1.6586,
      "step": 3526
    },
    {
      "epoch": 0.1568627450980392,
      "grad_norm": 0.13578933477401733,
      "learning_rate": 0.000998008929644873,
      "loss": 1.657,
      "step": 3528
    },
    {
      "epoch": 0.15695166955671158,
      "grad_norm": 0.1279151439666748,
      "learning_rate": 0.0009980057800675977,
      "loss": 1.6591,
      "step": 3530
    },
    {
      "epoch": 0.15704059401538392,
      "grad_norm": 0.12457801401615143,
      "learning_rate": 0.0009980026280061872,
      "loss": 1.6639,
      "step": 3532
    },
    {
      "epoch": 0.1571295184740563,
      "grad_norm": 0.12850531935691833,
      "learning_rate": 0.0009979994734606568,
      "loss": 1.6618,
      "step": 3534
    },
    {
      "epoch": 0.15721844293272866,
      "grad_norm": 0.1258937120437622,
      "learning_rate": 0.0009979963164310224,
      "loss": 1.6541,
      "step": 3536
    },
    {
      "epoch": 0.157307367391401,
      "grad_norm": 0.12144545465707779,
      "learning_rate": 0.0009979931569172994,
      "loss": 1.6592,
      "step": 3538
    },
    {
      "epoch": 0.15739629185007337,
      "grad_norm": 0.1225188747048378,
      "learning_rate": 0.0009979899949195043,
      "loss": 1.6578,
      "step": 3540
    },
    {
      "epoch": 0.1574852163087457,
      "grad_norm": 0.13467775285243988,
      "learning_rate": 0.0009979868304376522,
      "loss": 1.6604,
      "step": 3542
    },
    {
      "epoch": 0.15757414076741807,
      "grad_norm": 0.13501925766468048,
      "learning_rate": 0.0009979836634717591,
      "loss": 1.6567,
      "step": 3544
    },
    {
      "epoch": 0.15766306522609044,
      "grad_norm": 0.1286332756280899,
      "learning_rate": 0.000997980494021841,
      "loss": 1.6608,
      "step": 3546
    },
    {
      "epoch": 0.15775198968476278,
      "grad_norm": 0.12571607530117035,
      "learning_rate": 0.0009979773220879136,
      "loss": 1.6518,
      "step": 3548
    },
    {
      "epoch": 0.15784091414343515,
      "grad_norm": 0.12651760876178741,
      "learning_rate": 0.0009979741476699923,
      "loss": 1.6562,
      "step": 3550
    },
    {
      "epoch": 0.15792983860210752,
      "grad_norm": 0.12374807894229889,
      "learning_rate": 0.0009979709707680937,
      "loss": 1.6635,
      "step": 3552
    },
    {
      "epoch": 0.15801876306077986,
      "grad_norm": 0.12591153383255005,
      "learning_rate": 0.000997967791382233,
      "loss": 1.6606,
      "step": 3554
    },
    {
      "epoch": 0.15810768751945223,
      "grad_norm": 0.13102535903453827,
      "learning_rate": 0.0009979646095124265,
      "loss": 1.6571,
      "step": 3556
    },
    {
      "epoch": 0.1581966119781246,
      "grad_norm": 0.1338418573141098,
      "learning_rate": 0.0009979614251586894,
      "loss": 1.6582,
      "step": 3558
    },
    {
      "epoch": 0.15828553643679694,
      "grad_norm": 0.13042640686035156,
      "learning_rate": 0.0009979582383210386,
      "loss": 1.6553,
      "step": 3560
    },
    {
      "epoch": 0.1583744608954693,
      "grad_norm": 0.13031484186649323,
      "learning_rate": 0.0009979550489994892,
      "loss": 1.6604,
      "step": 3562
    },
    {
      "epoch": 0.15846338535414164,
      "grad_norm": 0.12561386823654175,
      "learning_rate": 0.0009979518571940574,
      "loss": 1.6619,
      "step": 3564
    },
    {
      "epoch": 0.158552309812814,
      "grad_norm": 0.13122697174549103,
      "learning_rate": 0.0009979486629047589,
      "loss": 1.658,
      "step": 3566
    },
    {
      "epoch": 0.15864123427148638,
      "grad_norm": 0.13682737946510315,
      "learning_rate": 0.0009979454661316102,
      "loss": 1.6619,
      "step": 3568
    },
    {
      "epoch": 0.15873015873015872,
      "grad_norm": 0.12418843060731888,
      "learning_rate": 0.0009979422668746265,
      "loss": 1.6576,
      "step": 3570
    },
    {
      "epoch": 0.1588190831888311,
      "grad_norm": 0.13619369268417358,
      "learning_rate": 0.0009979390651338243,
      "loss": 1.6581,
      "step": 3572
    },
    {
      "epoch": 0.15890800764750346,
      "grad_norm": 0.12203360348939896,
      "learning_rate": 0.0009979358609092193,
      "loss": 1.657,
      "step": 3574
    },
    {
      "epoch": 0.1589969321061758,
      "grad_norm": 0.12217724323272705,
      "learning_rate": 0.0009979326542008276,
      "loss": 1.6533,
      "step": 3576
    },
    {
      "epoch": 0.15908585656484817,
      "grad_norm": 0.13109327852725983,
      "learning_rate": 0.0009979294450086652,
      "loss": 1.6624,
      "step": 3578
    },
    {
      "epoch": 0.1591747810235205,
      "grad_norm": 0.123811274766922,
      "learning_rate": 0.0009979262333327482,
      "loss": 1.6538,
      "step": 3580
    },
    {
      "epoch": 0.15926370548219287,
      "grad_norm": 0.12288839370012283,
      "learning_rate": 0.000997923019173092,
      "loss": 1.6553,
      "step": 3582
    },
    {
      "epoch": 0.15935262994086524,
      "grad_norm": 0.12873615324497223,
      "learning_rate": 0.0009979198025297138,
      "loss": 1.6612,
      "step": 3584
    },
    {
      "epoch": 0.15944155439953758,
      "grad_norm": 0.12492024898529053,
      "learning_rate": 0.0009979165834026285,
      "loss": 1.6557,
      "step": 3586
    },
    {
      "epoch": 0.15953047885820995,
      "grad_norm": 0.12257129698991776,
      "learning_rate": 0.0009979133617918528,
      "loss": 1.6571,
      "step": 3588
    },
    {
      "epoch": 0.15961940331688232,
      "grad_norm": 0.13002300262451172,
      "learning_rate": 0.0009979101376974026,
      "loss": 1.6584,
      "step": 3590
    },
    {
      "epoch": 0.15970832777555466,
      "grad_norm": 0.12845563888549805,
      "learning_rate": 0.000997906911119294,
      "loss": 1.6524,
      "step": 3592
    },
    {
      "epoch": 0.15979725223422703,
      "grad_norm": 0.1296927034854889,
      "learning_rate": 0.0009979036820575431,
      "loss": 1.6599,
      "step": 3594
    },
    {
      "epoch": 0.1598861766928994,
      "grad_norm": 0.11898598074913025,
      "learning_rate": 0.000997900450512166,
      "loss": 1.6547,
      "step": 3596
    },
    {
      "epoch": 0.15997510115157174,
      "grad_norm": 0.12477612495422363,
      "learning_rate": 0.0009978972164831786,
      "loss": 1.6576,
      "step": 3598
    },
    {
      "epoch": 0.1600640256102441,
      "grad_norm": 0.12979185581207275,
      "learning_rate": 0.0009978939799705974,
      "loss": 1.6585,
      "step": 3600
    },
    {
      "epoch": 0.16015295006891644,
      "grad_norm": 0.1325928121805191,
      "learning_rate": 0.0009978907409744385,
      "loss": 1.6561,
      "step": 3602
    },
    {
      "epoch": 0.1602418745275888,
      "grad_norm": 0.12693507969379425,
      "learning_rate": 0.0009978874994947178,
      "loss": 1.6591,
      "step": 3604
    },
    {
      "epoch": 0.16033079898626118,
      "grad_norm": 0.12471980601549149,
      "learning_rate": 0.0009978842555314516,
      "loss": 1.6559,
      "step": 3606
    },
    {
      "epoch": 0.16041972344493352,
      "grad_norm": 0.13122089207172394,
      "learning_rate": 0.000997881009084656,
      "loss": 1.6615,
      "step": 3608
    },
    {
      "epoch": 0.1605086479036059,
      "grad_norm": 0.12727150321006775,
      "learning_rate": 0.0009978777601543473,
      "loss": 1.6517,
      "step": 3610
    },
    {
      "epoch": 0.16059757236227826,
      "grad_norm": 0.13121071457862854,
      "learning_rate": 0.0009978745087405418,
      "loss": 1.6547,
      "step": 3612
    },
    {
      "epoch": 0.1606864968209506,
      "grad_norm": 0.13580727577209473,
      "learning_rate": 0.0009978712548432557,
      "loss": 1.6545,
      "step": 3614
    },
    {
      "epoch": 0.16077542127962297,
      "grad_norm": 0.13076774775981903,
      "learning_rate": 0.0009978679984625049,
      "loss": 1.658,
      "step": 3616
    },
    {
      "epoch": 0.1608643457382953,
      "grad_norm": 0.1376674324274063,
      "learning_rate": 0.000997864739598306,
      "loss": 1.6551,
      "step": 3618
    },
    {
      "epoch": 0.16095327019696767,
      "grad_norm": 0.13014668226242065,
      "learning_rate": 0.0009978614782506752,
      "loss": 1.6476,
      "step": 3620
    },
    {
      "epoch": 0.16104219465564004,
      "grad_norm": 0.12911993265151978,
      "learning_rate": 0.0009978582144196287,
      "loss": 1.6458,
      "step": 3622
    },
    {
      "epoch": 0.16113111911431238,
      "grad_norm": 0.12219536304473877,
      "learning_rate": 0.0009978549481051825,
      "loss": 1.6512,
      "step": 3624
    },
    {
      "epoch": 0.16122004357298475,
      "grad_norm": 0.12736880779266357,
      "learning_rate": 0.0009978516793073535,
      "loss": 1.6565,
      "step": 3626
    },
    {
      "epoch": 0.16130896803165712,
      "grad_norm": 0.12791965901851654,
      "learning_rate": 0.0009978484080261575,
      "loss": 1.6543,
      "step": 3628
    },
    {
      "epoch": 0.16139789249032946,
      "grad_norm": 0.12389549612998962,
      "learning_rate": 0.0009978451342616112,
      "loss": 1.653,
      "step": 3630
    },
    {
      "epoch": 0.16148681694900183,
      "grad_norm": 0.12796027958393097,
      "learning_rate": 0.0009978418580137308,
      "loss": 1.6529,
      "step": 3632
    },
    {
      "epoch": 0.16157574140767417,
      "grad_norm": 0.13139387965202332,
      "learning_rate": 0.0009978385792825323,
      "loss": 1.6517,
      "step": 3634
    },
    {
      "epoch": 0.16166466586634654,
      "grad_norm": 0.12868015468120575,
      "learning_rate": 0.0009978352980680324,
      "loss": 1.6522,
      "step": 3636
    },
    {
      "epoch": 0.1617535903250189,
      "grad_norm": 0.12967824935913086,
      "learning_rate": 0.0009978320143702475,
      "loss": 1.6532,
      "step": 3638
    },
    {
      "epoch": 0.16184251478369124,
      "grad_norm": 0.13034017384052277,
      "learning_rate": 0.0009978287281891937,
      "loss": 1.6552,
      "step": 3640
    },
    {
      "epoch": 0.1619314392423636,
      "grad_norm": 0.12667222321033478,
      "learning_rate": 0.0009978254395248878,
      "loss": 1.6524,
      "step": 3642
    },
    {
      "epoch": 0.16202036370103598,
      "grad_norm": 0.1248355358839035,
      "learning_rate": 0.0009978221483773458,
      "loss": 1.651,
      "step": 3644
    },
    {
      "epoch": 0.16210928815970832,
      "grad_norm": 0.12515072524547577,
      "learning_rate": 0.0009978188547465842,
      "loss": 1.6517,
      "step": 3646
    },
    {
      "epoch": 0.1621982126183807,
      "grad_norm": 0.12195508927106857,
      "learning_rate": 0.0009978155586326198,
      "loss": 1.6487,
      "step": 3648
    },
    {
      "epoch": 0.16228713707705306,
      "grad_norm": 0.12744548916816711,
      "learning_rate": 0.0009978122600354687,
      "loss": 1.6517,
      "step": 3650
    },
    {
      "epoch": 0.1623760615357254,
      "grad_norm": 0.1203659400343895,
      "learning_rate": 0.0009978089589551473,
      "loss": 1.6527,
      "step": 3652
    },
    {
      "epoch": 0.16246498599439776,
      "grad_norm": 0.12796948850154877,
      "learning_rate": 0.0009978056553916722,
      "loss": 1.6491,
      "step": 3654
    },
    {
      "epoch": 0.1625539104530701,
      "grad_norm": 0.12525366246700287,
      "learning_rate": 0.0009978023493450597,
      "loss": 1.6569,
      "step": 3656
    },
    {
      "epoch": 0.16264283491174247,
      "grad_norm": 0.12322328239679337,
      "learning_rate": 0.0009977990408153268,
      "loss": 1.6653,
      "step": 3658
    },
    {
      "epoch": 0.16273175937041484,
      "grad_norm": 0.13084837794303894,
      "learning_rate": 0.0009977957298024894,
      "loss": 1.6531,
      "step": 3660
    },
    {
      "epoch": 0.16282068382908718,
      "grad_norm": 0.12508228421211243,
      "learning_rate": 0.0009977924163065643,
      "loss": 1.6563,
      "step": 3662
    },
    {
      "epoch": 0.16290960828775955,
      "grad_norm": 0.12103430181741714,
      "learning_rate": 0.000997789100327568,
      "loss": 1.6506,
      "step": 3664
    },
    {
      "epoch": 0.16299853274643192,
      "grad_norm": 0.12172827869653702,
      "learning_rate": 0.0009977857818655172,
      "loss": 1.6474,
      "step": 3666
    },
    {
      "epoch": 0.16308745720510426,
      "grad_norm": 0.12971435487270355,
      "learning_rate": 0.000997782460920428,
      "loss": 1.6508,
      "step": 3668
    },
    {
      "epoch": 0.16317638166377663,
      "grad_norm": 0.1291271448135376,
      "learning_rate": 0.0009977791374923175,
      "loss": 1.6535,
      "step": 3670
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 0.13160645961761475,
      "learning_rate": 0.000997775811581202,
      "loss": 1.6513,
      "step": 3672
    },
    {
      "epoch": 0.16335423058112133,
      "grad_norm": 0.1266569048166275,
      "learning_rate": 0.000997772483187098,
      "loss": 1.6537,
      "step": 3674
    },
    {
      "epoch": 0.1634431550397937,
      "grad_norm": 0.12977223098278046,
      "learning_rate": 0.0009977691523100224,
      "loss": 1.6543,
      "step": 3676
    },
    {
      "epoch": 0.16353207949846604,
      "grad_norm": 0.1290341168642044,
      "learning_rate": 0.0009977658189499914,
      "loss": 1.6526,
      "step": 3678
    },
    {
      "epoch": 0.1636210039571384,
      "grad_norm": 0.12717925012111664,
      "learning_rate": 0.0009977624831070222,
      "loss": 1.6534,
      "step": 3680
    },
    {
      "epoch": 0.16370992841581078,
      "grad_norm": 0.12076234817504883,
      "learning_rate": 0.0009977591447811308,
      "loss": 1.6484,
      "step": 3682
    },
    {
      "epoch": 0.16379885287448312,
      "grad_norm": 0.12450318038463593,
      "learning_rate": 0.0009977558039723344,
      "loss": 1.6525,
      "step": 3684
    },
    {
      "epoch": 0.1638877773331555,
      "grad_norm": 0.12309740483760834,
      "learning_rate": 0.0009977524606806494,
      "loss": 1.6511,
      "step": 3686
    },
    {
      "epoch": 0.16397670179182786,
      "grad_norm": 0.12498360872268677,
      "learning_rate": 0.0009977491149060925,
      "loss": 1.6513,
      "step": 3688
    },
    {
      "epoch": 0.1640656262505002,
      "grad_norm": 0.13169927895069122,
      "learning_rate": 0.0009977457666486803,
      "loss": 1.6528,
      "step": 3690
    },
    {
      "epoch": 0.16415455070917256,
      "grad_norm": 0.12063473463058472,
      "learning_rate": 0.0009977424159084295,
      "loss": 1.6449,
      "step": 3692
    },
    {
      "epoch": 0.1642434751678449,
      "grad_norm": 0.12622106075286865,
      "learning_rate": 0.000997739062685357,
      "loss": 1.6565,
      "step": 3694
    },
    {
      "epoch": 0.16433239962651727,
      "grad_norm": 0.12294983863830566,
      "learning_rate": 0.0009977357069794795,
      "loss": 1.6491,
      "step": 3696
    },
    {
      "epoch": 0.16442132408518964,
      "grad_norm": 0.12392249703407288,
      "learning_rate": 0.0009977323487908136,
      "loss": 1.6513,
      "step": 3698
    },
    {
      "epoch": 0.16451024854386198,
      "grad_norm": 0.1264803558588028,
      "learning_rate": 0.0009977289881193763,
      "loss": 1.6466,
      "step": 3700
    },
    {
      "epoch": 0.16459917300253435,
      "grad_norm": 0.12120287120342255,
      "learning_rate": 0.0009977256249651839,
      "loss": 1.6532,
      "step": 3702
    },
    {
      "epoch": 0.16468809746120672,
      "grad_norm": 0.12672477960586548,
      "learning_rate": 0.0009977222593282535,
      "loss": 1.6501,
      "step": 3704
    },
    {
      "epoch": 0.16477702191987906,
      "grad_norm": 0.12327267974615097,
      "learning_rate": 0.0009977188912086022,
      "loss": 1.649,
      "step": 3706
    },
    {
      "epoch": 0.16486594637855143,
      "grad_norm": 0.12144723534584045,
      "learning_rate": 0.000997715520606246,
      "loss": 1.6497,
      "step": 3708
    },
    {
      "epoch": 0.16495487083722377,
      "grad_norm": 0.12345416843891144,
      "learning_rate": 0.0009977121475212025,
      "loss": 1.6519,
      "step": 3710
    },
    {
      "epoch": 0.16504379529589613,
      "grad_norm": 0.12130998075008392,
      "learning_rate": 0.000997708771953488,
      "loss": 1.6541,
      "step": 3712
    },
    {
      "epoch": 0.1651327197545685,
      "grad_norm": 0.1252111792564392,
      "learning_rate": 0.0009977053939031197,
      "loss": 1.6551,
      "step": 3714
    },
    {
      "epoch": 0.16522164421324084,
      "grad_norm": 0.1185317412018776,
      "learning_rate": 0.0009977020133701143,
      "loss": 1.6497,
      "step": 3716
    },
    {
      "epoch": 0.1653105686719132,
      "grad_norm": 0.13791343569755554,
      "learning_rate": 0.0009976986303544884,
      "loss": 1.6527,
      "step": 3718
    },
    {
      "epoch": 0.16539949313058558,
      "grad_norm": 0.1374451071023941,
      "learning_rate": 0.0009976952448562592,
      "loss": 1.6527,
      "step": 3720
    },
    {
      "epoch": 0.16548841758925792,
      "grad_norm": 0.12747934460639954,
      "learning_rate": 0.0009976918568754436,
      "loss": 1.6481,
      "step": 3722
    },
    {
      "epoch": 0.1655773420479303,
      "grad_norm": 0.13003109395503998,
      "learning_rate": 0.0009976884664120583,
      "loss": 1.654,
      "step": 3724
    },
    {
      "epoch": 0.16566626650660263,
      "grad_norm": 0.11762271076440811,
      "learning_rate": 0.0009976850734661204,
      "loss": 1.645,
      "step": 3726
    },
    {
      "epoch": 0.165755190965275,
      "grad_norm": 0.13263928890228271,
      "learning_rate": 0.0009976816780376467,
      "loss": 1.6455,
      "step": 3728
    },
    {
      "epoch": 0.16584411542394736,
      "grad_norm": 0.1377555876970291,
      "learning_rate": 0.0009976782801266542,
      "loss": 1.6414,
      "step": 3730
    },
    {
      "epoch": 0.1659330398826197,
      "grad_norm": 0.12296448647975922,
      "learning_rate": 0.0009976748797331598,
      "loss": 1.6466,
      "step": 3732
    },
    {
      "epoch": 0.16602196434129207,
      "grad_norm": 0.12507364153862,
      "learning_rate": 0.0009976714768571806,
      "loss": 1.6475,
      "step": 3734
    },
    {
      "epoch": 0.16611088879996444,
      "grad_norm": 0.12280484288930893,
      "learning_rate": 0.0009976680714987333,
      "loss": 1.6461,
      "step": 3736
    },
    {
      "epoch": 0.16619981325863678,
      "grad_norm": 0.12638729810714722,
      "learning_rate": 0.0009976646636578352,
      "loss": 1.6534,
      "step": 3738
    },
    {
      "epoch": 0.16628873771730915,
      "grad_norm": 0.12049468606710434,
      "learning_rate": 0.000997661253334503,
      "loss": 1.6484,
      "step": 3740
    },
    {
      "epoch": 0.16637766217598152,
      "grad_norm": 0.12127727270126343,
      "learning_rate": 0.000997657840528754,
      "loss": 1.6468,
      "step": 3742
    },
    {
      "epoch": 0.16646658663465386,
      "grad_norm": 0.1225084736943245,
      "learning_rate": 0.000997654425240605,
      "loss": 1.6486,
      "step": 3744
    },
    {
      "epoch": 0.16655551109332623,
      "grad_norm": 0.12508682906627655,
      "learning_rate": 0.0009976510074700732,
      "loss": 1.6429,
      "step": 3746
    },
    {
      "epoch": 0.16664443555199857,
      "grad_norm": 0.12496629357337952,
      "learning_rate": 0.0009976475872171754,
      "loss": 1.6486,
      "step": 3748
    },
    {
      "epoch": 0.16673336001067093,
      "grad_norm": 0.1250324249267578,
      "learning_rate": 0.000997644164481929,
      "loss": 1.6467,
      "step": 3750
    },
    {
      "epoch": 0.1668222844693433,
      "grad_norm": 0.1318887323141098,
      "learning_rate": 0.0009976407392643508,
      "loss": 1.6502,
      "step": 3752
    },
    {
      "epoch": 0.16691120892801564,
      "grad_norm": 0.12485989183187485,
      "learning_rate": 0.000997637311564458,
      "loss": 1.6462,
      "step": 3754
    },
    {
      "epoch": 0.167000133386688,
      "grad_norm": 0.12936322391033173,
      "learning_rate": 0.0009976338813822679,
      "loss": 1.6469,
      "step": 3756
    },
    {
      "epoch": 0.16708905784536038,
      "grad_norm": 0.12338431924581528,
      "learning_rate": 0.000997630448717797,
      "loss": 1.6471,
      "step": 3758
    },
    {
      "epoch": 0.16717798230403272,
      "grad_norm": 0.12906041741371155,
      "learning_rate": 0.0009976270135710632,
      "loss": 1.6512,
      "step": 3760
    },
    {
      "epoch": 0.1672669067627051,
      "grad_norm": 0.12762369215488434,
      "learning_rate": 0.000997623575942083,
      "loss": 1.6502,
      "step": 3762
    },
    {
      "epoch": 0.16735583122137743,
      "grad_norm": 0.12695105373859406,
      "learning_rate": 0.000997620135830874,
      "loss": 1.6524,
      "step": 3764
    },
    {
      "epoch": 0.1674447556800498,
      "grad_norm": 0.12652401626110077,
      "learning_rate": 0.000997616693237453,
      "loss": 1.6479,
      "step": 3766
    },
    {
      "epoch": 0.16753368013872216,
      "grad_norm": 0.12700212001800537,
      "learning_rate": 0.0009976132481618375,
      "loss": 1.6463,
      "step": 3768
    },
    {
      "epoch": 0.1676226045973945,
      "grad_norm": 0.1289287805557251,
      "learning_rate": 0.0009976098006040442,
      "loss": 1.6458,
      "step": 3770
    },
    {
      "epoch": 0.16771152905606687,
      "grad_norm": 0.12287376075983047,
      "learning_rate": 0.0009976063505640908,
      "loss": 1.6516,
      "step": 3772
    },
    {
      "epoch": 0.16780045351473924,
      "grad_norm": 0.1236732006072998,
      "learning_rate": 0.0009976028980419945,
      "loss": 1.649,
      "step": 3774
    },
    {
      "epoch": 0.16788937797341158,
      "grad_norm": 0.12876683473587036,
      "learning_rate": 0.000997599443037772,
      "loss": 1.6499,
      "step": 3776
    },
    {
      "epoch": 0.16797830243208395,
      "grad_norm": 0.12273410707712173,
      "learning_rate": 0.0009975959855514412,
      "loss": 1.6491,
      "step": 3778
    },
    {
      "epoch": 0.16806722689075632,
      "grad_norm": 0.12249413132667542,
      "learning_rate": 0.000997592525583019,
      "loss": 1.6492,
      "step": 3780
    },
    {
      "epoch": 0.16815615134942866,
      "grad_norm": 0.1305217146873474,
      "learning_rate": 0.0009975890631325223,
      "loss": 1.6416,
      "step": 3782
    },
    {
      "epoch": 0.16824507580810102,
      "grad_norm": 0.13063625991344452,
      "learning_rate": 0.0009975855981999692,
      "loss": 1.6455,
      "step": 3784
    },
    {
      "epoch": 0.16833400026677336,
      "grad_norm": 0.12041649222373962,
      "learning_rate": 0.0009975821307853763,
      "loss": 1.6451,
      "step": 3786
    },
    {
      "epoch": 0.16842292472544573,
      "grad_norm": 0.12157703936100006,
      "learning_rate": 0.0009975786608887613,
      "loss": 1.6447,
      "step": 3788
    },
    {
      "epoch": 0.1685118491841181,
      "grad_norm": 0.12154638022184372,
      "learning_rate": 0.0009975751885101412,
      "loss": 1.6527,
      "step": 3790
    },
    {
      "epoch": 0.16860077364279044,
      "grad_norm": 0.11739989370107651,
      "learning_rate": 0.0009975717136495336,
      "loss": 1.6468,
      "step": 3792
    },
    {
      "epoch": 0.1686896981014628,
      "grad_norm": 0.11798025667667389,
      "learning_rate": 0.0009975682363069557,
      "loss": 1.6459,
      "step": 3794
    },
    {
      "epoch": 0.16877862256013518,
      "grad_norm": 0.11994331330060959,
      "learning_rate": 0.0009975647564824247,
      "loss": 1.6495,
      "step": 3796
    },
    {
      "epoch": 0.16886754701880752,
      "grad_norm": 0.1198289692401886,
      "learning_rate": 0.0009975612741759583,
      "loss": 1.6497,
      "step": 3798
    },
    {
      "epoch": 0.16895647147747989,
      "grad_norm": 0.12364807724952698,
      "learning_rate": 0.0009975577893875734,
      "loss": 1.6444,
      "step": 3800
    },
    {
      "epoch": 0.16904539593615223,
      "grad_norm": 0.1229090690612793,
      "learning_rate": 0.0009975543021172879,
      "loss": 1.6499,
      "step": 3802
    },
    {
      "epoch": 0.1691343203948246,
      "grad_norm": 0.1250782459974289,
      "learning_rate": 0.000997550812365119,
      "loss": 1.6417,
      "step": 3804
    },
    {
      "epoch": 0.16922324485349696,
      "grad_norm": 0.12151206284761429,
      "learning_rate": 0.0009975473201310839,
      "loss": 1.6451,
      "step": 3806
    },
    {
      "epoch": 0.1693121693121693,
      "grad_norm": 0.12076232582330704,
      "learning_rate": 0.0009975438254152,
      "loss": 1.6476,
      "step": 3808
    },
    {
      "epoch": 0.16940109377084167,
      "grad_norm": 0.1340232938528061,
      "learning_rate": 0.0009975403282174852,
      "loss": 1.6504,
      "step": 3810
    },
    {
      "epoch": 0.16949001822951404,
      "grad_norm": 0.12269595265388489,
      "learning_rate": 0.0009975368285379564,
      "loss": 1.6462,
      "step": 3812
    },
    {
      "epoch": 0.16957894268818638,
      "grad_norm": 0.1238735169172287,
      "learning_rate": 0.0009975333263766316,
      "loss": 1.6437,
      "step": 3814
    },
    {
      "epoch": 0.16966786714685875,
      "grad_norm": 0.13427115976810455,
      "learning_rate": 0.0009975298217335278,
      "loss": 1.6458,
      "step": 3816
    },
    {
      "epoch": 0.1697567916055311,
      "grad_norm": 0.12311917543411255,
      "learning_rate": 0.0009975263146086628,
      "loss": 1.6445,
      "step": 3818
    },
    {
      "epoch": 0.16984571606420346,
      "grad_norm": 0.13276326656341553,
      "learning_rate": 0.000997522805002054,
      "loss": 1.6489,
      "step": 3820
    },
    {
      "epoch": 0.16993464052287582,
      "grad_norm": 0.12898485362529755,
      "learning_rate": 0.0009975192929137188,
      "loss": 1.6513,
      "step": 3822
    },
    {
      "epoch": 0.17002356498154816,
      "grad_norm": 0.12796077132225037,
      "learning_rate": 0.0009975157783436745,
      "loss": 1.6499,
      "step": 3824
    },
    {
      "epoch": 0.17011248944022053,
      "grad_norm": 0.11903078854084015,
      "learning_rate": 0.0009975122612919391,
      "loss": 1.6457,
      "step": 3826
    },
    {
      "epoch": 0.1702014138988929,
      "grad_norm": 0.11384111642837524,
      "learning_rate": 0.00099750874175853,
      "loss": 1.6383,
      "step": 3828
    },
    {
      "epoch": 0.17029033835756524,
      "grad_norm": 0.11926476657390594,
      "learning_rate": 0.0009975052197434648,
      "loss": 1.6462,
      "step": 3830
    },
    {
      "epoch": 0.1703792628162376,
      "grad_norm": 0.11415476351976395,
      "learning_rate": 0.0009975016952467608,
      "loss": 1.657,
      "step": 3832
    },
    {
      "epoch": 0.17046818727490998,
      "grad_norm": 0.1166810616850853,
      "learning_rate": 0.000997498168268436,
      "loss": 1.646,
      "step": 3834
    },
    {
      "epoch": 0.17055711173358232,
      "grad_norm": 0.13669024407863617,
      "learning_rate": 0.0009974946388085076,
      "loss": 1.6467,
      "step": 3836
    },
    {
      "epoch": 0.17064603619225469,
      "grad_norm": 0.12202465534210205,
      "learning_rate": 0.0009974911068669934,
      "loss": 1.643,
      "step": 3838
    },
    {
      "epoch": 0.17073496065092703,
      "grad_norm": 0.12026860564947128,
      "learning_rate": 0.0009974875724439109,
      "loss": 1.6458,
      "step": 3840
    },
    {
      "epoch": 0.1708238851095994,
      "grad_norm": 0.12143933773040771,
      "learning_rate": 0.0009974840355392778,
      "loss": 1.6471,
      "step": 3842
    },
    {
      "epoch": 0.17091280956827176,
      "grad_norm": 0.12261870503425598,
      "learning_rate": 0.000997480496153112,
      "loss": 1.6384,
      "step": 3844
    },
    {
      "epoch": 0.1710017340269441,
      "grad_norm": 0.12399966269731522,
      "learning_rate": 0.0009974769542854306,
      "loss": 1.6459,
      "step": 3846
    },
    {
      "epoch": 0.17109065848561647,
      "grad_norm": 0.12557664513587952,
      "learning_rate": 0.0009974734099362516,
      "loss": 1.6463,
      "step": 3848
    },
    {
      "epoch": 0.17117958294428884,
      "grad_norm": 0.1280864030122757,
      "learning_rate": 0.0009974698631055928,
      "loss": 1.6447,
      "step": 3850
    },
    {
      "epoch": 0.17126850740296118,
      "grad_norm": 0.12775865197181702,
      "learning_rate": 0.0009974663137934717,
      "loss": 1.6477,
      "step": 3852
    },
    {
      "epoch": 0.17135743186163355,
      "grad_norm": 0.12285788357257843,
      "learning_rate": 0.000997462761999906,
      "loss": 1.6416,
      "step": 3854
    },
    {
      "epoch": 0.1714463563203059,
      "grad_norm": 0.12305966019630432,
      "learning_rate": 0.0009974592077249137,
      "loss": 1.6478,
      "step": 3856
    },
    {
      "epoch": 0.17153528077897826,
      "grad_norm": 0.12091002613306046,
      "learning_rate": 0.000997455650968512,
      "loss": 1.6458,
      "step": 3858
    },
    {
      "epoch": 0.17162420523765062,
      "grad_norm": 0.12569975852966309,
      "learning_rate": 0.000997452091730719,
      "loss": 1.6378,
      "step": 3860
    },
    {
      "epoch": 0.17171312969632296,
      "grad_norm": 0.13074176013469696,
      "learning_rate": 0.0009974485300115526,
      "loss": 1.647,
      "step": 3862
    },
    {
      "epoch": 0.17180205415499533,
      "grad_norm": 0.11964616179466248,
      "learning_rate": 0.0009974449658110302,
      "loss": 1.6445,
      "step": 3864
    },
    {
      "epoch": 0.1718909786136677,
      "grad_norm": 0.11686211824417114,
      "learning_rate": 0.0009974413991291698,
      "loss": 1.6438,
      "step": 3866
    },
    {
      "epoch": 0.17197990307234004,
      "grad_norm": 0.11895184218883514,
      "learning_rate": 0.000997437829965989,
      "loss": 1.6416,
      "step": 3868
    },
    {
      "epoch": 0.1720688275310124,
      "grad_norm": 0.12349550426006317,
      "learning_rate": 0.0009974342583215058,
      "loss": 1.6466,
      "step": 3870
    },
    {
      "epoch": 0.17215775198968478,
      "grad_norm": 0.12045907974243164,
      "learning_rate": 0.000997430684195738,
      "loss": 1.6422,
      "step": 3872
    },
    {
      "epoch": 0.17224667644835712,
      "grad_norm": 0.12487021833658218,
      "learning_rate": 0.0009974271075887033,
      "loss": 1.645,
      "step": 3874
    },
    {
      "epoch": 0.17233560090702948,
      "grad_norm": 0.12949977815151215,
      "learning_rate": 0.0009974235285004195,
      "loss": 1.6428,
      "step": 3876
    },
    {
      "epoch": 0.17242452536570183,
      "grad_norm": 0.129159078001976,
      "learning_rate": 0.0009974199469309048,
      "loss": 1.6378,
      "step": 3878
    },
    {
      "epoch": 0.1725134498243742,
      "grad_norm": 0.12624463438987732,
      "learning_rate": 0.000997416362880177,
      "loss": 1.6427,
      "step": 3880
    },
    {
      "epoch": 0.17260237428304656,
      "grad_norm": 0.11681614816188812,
      "learning_rate": 0.0009974127763482535,
      "loss": 1.6428,
      "step": 3882
    },
    {
      "epoch": 0.1726912987417189,
      "grad_norm": 0.11823716759681702,
      "learning_rate": 0.0009974091873351525,
      "loss": 1.6421,
      "step": 3884
    },
    {
      "epoch": 0.17278022320039127,
      "grad_norm": 0.12332173436880112,
      "learning_rate": 0.000997405595840892,
      "loss": 1.6423,
      "step": 3886
    },
    {
      "epoch": 0.17286914765906364,
      "grad_norm": 0.1153373122215271,
      "learning_rate": 0.0009974020018654898,
      "loss": 1.649,
      "step": 3888
    },
    {
      "epoch": 0.17295807211773598,
      "grad_norm": 0.11574465781450272,
      "learning_rate": 0.000997398405408964,
      "loss": 1.6347,
      "step": 3890
    },
    {
      "epoch": 0.17304699657640835,
      "grad_norm": 0.11528904736042023,
      "learning_rate": 0.0009973948064713322,
      "loss": 1.6458,
      "step": 3892
    },
    {
      "epoch": 0.1731359210350807,
      "grad_norm": 0.11927869915962219,
      "learning_rate": 0.0009973912050526125,
      "loss": 1.6434,
      "step": 3894
    },
    {
      "epoch": 0.17322484549375305,
      "grad_norm": 0.12420177459716797,
      "learning_rate": 0.000997387601152823,
      "loss": 1.6429,
      "step": 3896
    },
    {
      "epoch": 0.17331376995242542,
      "grad_norm": 0.11699125170707703,
      "learning_rate": 0.0009973839947719817,
      "loss": 1.6461,
      "step": 3898
    },
    {
      "epoch": 0.17340269441109776,
      "grad_norm": 0.1246216669678688,
      "learning_rate": 0.0009973803859101064,
      "loss": 1.644,
      "step": 3900
    },
    {
      "epoch": 0.17349161886977013,
      "grad_norm": 0.13008350133895874,
      "learning_rate": 0.0009973767745672152,
      "loss": 1.6469,
      "step": 3902
    },
    {
      "epoch": 0.1735805433284425,
      "grad_norm": 0.12936833500862122,
      "learning_rate": 0.000997373160743326,
      "loss": 1.6459,
      "step": 3904
    },
    {
      "epoch": 0.17366946778711484,
      "grad_norm": 0.11948209255933762,
      "learning_rate": 0.000997369544438457,
      "loss": 1.6425,
      "step": 3906
    },
    {
      "epoch": 0.1737583922457872,
      "grad_norm": 0.11415665596723557,
      "learning_rate": 0.000997365925652626,
      "loss": 1.6449,
      "step": 3908
    },
    {
      "epoch": 0.17384731670445955,
      "grad_norm": 0.12373475730419159,
      "learning_rate": 0.0009973623043858513,
      "loss": 1.6428,
      "step": 3910
    },
    {
      "epoch": 0.17393624116313192,
      "grad_norm": 0.11973095685243607,
      "learning_rate": 0.0009973586806381507,
      "loss": 1.6411,
      "step": 3912
    },
    {
      "epoch": 0.17402516562180428,
      "grad_norm": 0.12207634001970291,
      "learning_rate": 0.0009973550544095427,
      "loss": 1.6408,
      "step": 3914
    },
    {
      "epoch": 0.17411409008047662,
      "grad_norm": 0.12171240895986557,
      "learning_rate": 0.0009973514257000451,
      "loss": 1.6446,
      "step": 3916
    },
    {
      "epoch": 0.174203014539149,
      "grad_norm": 0.12418859452009201,
      "learning_rate": 0.000997347794509676,
      "loss": 1.6475,
      "step": 3918
    },
    {
      "epoch": 0.17429193899782136,
      "grad_norm": 0.12470457702875137,
      "learning_rate": 0.0009973441608384535,
      "loss": 1.6453,
      "step": 3920
    },
    {
      "epoch": 0.1743808634564937,
      "grad_norm": 0.12560194730758667,
      "learning_rate": 0.0009973405246863957,
      "loss": 1.6393,
      "step": 3922
    },
    {
      "epoch": 0.17446978791516607,
      "grad_norm": 0.1209321841597557,
      "learning_rate": 0.0009973368860535208,
      "loss": 1.647,
      "step": 3924
    },
    {
      "epoch": 0.17455871237383844,
      "grad_norm": 0.11720415949821472,
      "learning_rate": 0.000997333244939847,
      "loss": 1.6403,
      "step": 3926
    },
    {
      "epoch": 0.17464763683251078,
      "grad_norm": 0.1247502788901329,
      "learning_rate": 0.0009973296013453927,
      "loss": 1.6419,
      "step": 3928
    },
    {
      "epoch": 0.17473656129118315,
      "grad_norm": 0.12291789054870605,
      "learning_rate": 0.0009973259552701755,
      "loss": 1.6477,
      "step": 3930
    },
    {
      "epoch": 0.17482548574985549,
      "grad_norm": 0.12356559187173843,
      "learning_rate": 0.000997322306714214,
      "loss": 1.6493,
      "step": 3932
    },
    {
      "epoch": 0.17491441020852785,
      "grad_norm": 0.12039901316165924,
      "learning_rate": 0.0009973186556775262,
      "loss": 1.6419,
      "step": 3934
    },
    {
      "epoch": 0.17500333466720022,
      "grad_norm": 0.11828312277793884,
      "learning_rate": 0.0009973150021601305,
      "loss": 1.6374,
      "step": 3936
    },
    {
      "epoch": 0.17509225912587256,
      "grad_norm": 0.12081318348646164,
      "learning_rate": 0.0009973113461620448,
      "loss": 1.6384,
      "step": 3938
    },
    {
      "epoch": 0.17518118358454493,
      "grad_norm": 0.11971206218004227,
      "learning_rate": 0.000997307687683288,
      "loss": 1.6403,
      "step": 3940
    },
    {
      "epoch": 0.1752701080432173,
      "grad_norm": 0.1235707625746727,
      "learning_rate": 0.0009973040267238775,
      "loss": 1.6432,
      "step": 3942
    },
    {
      "epoch": 0.17535903250188964,
      "grad_norm": 0.12012319266796112,
      "learning_rate": 0.000997300363283832,
      "loss": 1.6455,
      "step": 3944
    },
    {
      "epoch": 0.175447956960562,
      "grad_norm": 0.11857002973556519,
      "learning_rate": 0.0009972966973631698,
      "loss": 1.6372,
      "step": 3946
    },
    {
      "epoch": 0.17553688141923435,
      "grad_norm": 0.12104901671409607,
      "learning_rate": 0.000997293028961909,
      "loss": 1.6464,
      "step": 3948
    },
    {
      "epoch": 0.17562580587790672,
      "grad_norm": 0.12087158113718033,
      "learning_rate": 0.0009972893580800684,
      "loss": 1.6348,
      "step": 3950
    },
    {
      "epoch": 0.17571473033657908,
      "grad_norm": 0.11955670267343521,
      "learning_rate": 0.0009972856847176657,
      "loss": 1.6369,
      "step": 3952
    },
    {
      "epoch": 0.17580365479525142,
      "grad_norm": 0.12120344489812851,
      "learning_rate": 0.0009972820088747195,
      "loss": 1.6386,
      "step": 3954
    },
    {
      "epoch": 0.1758925792539238,
      "grad_norm": 0.12605014443397522,
      "learning_rate": 0.000997278330551248,
      "loss": 1.6385,
      "step": 3956
    },
    {
      "epoch": 0.17598150371259616,
      "grad_norm": 0.12289277464151382,
      "learning_rate": 0.0009972746497472697,
      "loss": 1.6435,
      "step": 3958
    },
    {
      "epoch": 0.1760704281712685,
      "grad_norm": 0.11947757750749588,
      "learning_rate": 0.0009972709664628028,
      "loss": 1.6271,
      "step": 3960
    },
    {
      "epoch": 0.17615935262994087,
      "grad_norm": 0.11600028723478317,
      "learning_rate": 0.000997267280697866,
      "loss": 1.6419,
      "step": 3962
    },
    {
      "epoch": 0.17624827708861324,
      "grad_norm": 0.11856868118047714,
      "learning_rate": 0.0009972635924524772,
      "loss": 1.6373,
      "step": 3964
    },
    {
      "epoch": 0.17633720154728558,
      "grad_norm": 0.11774558573961258,
      "learning_rate": 0.0009972599017266553,
      "loss": 1.6403,
      "step": 3966
    },
    {
      "epoch": 0.17642612600595795,
      "grad_norm": 0.1212194487452507,
      "learning_rate": 0.0009972562085204183,
      "loss": 1.6383,
      "step": 3968
    },
    {
      "epoch": 0.17651505046463029,
      "grad_norm": 0.11406757682561874,
      "learning_rate": 0.0009972525128337849,
      "loss": 1.6429,
      "step": 3970
    },
    {
      "epoch": 0.17660397492330265,
      "grad_norm": 0.11931362748146057,
      "learning_rate": 0.0009972488146667733,
      "loss": 1.6388,
      "step": 3972
    },
    {
      "epoch": 0.17669289938197502,
      "grad_norm": 0.12099288403987885,
      "learning_rate": 0.0009972451140194023,
      "loss": 1.6395,
      "step": 3974
    },
    {
      "epoch": 0.17678182384064736,
      "grad_norm": 0.1235419511795044,
      "learning_rate": 0.00099724141089169,
      "loss": 1.6444,
      "step": 3976
    },
    {
      "epoch": 0.17687074829931973,
      "grad_norm": 0.12513847649097443,
      "learning_rate": 0.000997237705283655,
      "loss": 1.6411,
      "step": 3978
    },
    {
      "epoch": 0.1769596727579921,
      "grad_norm": 0.11601508408784866,
      "learning_rate": 0.0009972339971953156,
      "loss": 1.6372,
      "step": 3980
    },
    {
      "epoch": 0.17704859721666444,
      "grad_norm": 0.11849774420261383,
      "learning_rate": 0.000997230286626691,
      "loss": 1.6378,
      "step": 3982
    },
    {
      "epoch": 0.1771375216753368,
      "grad_norm": 0.1219199150800705,
      "learning_rate": 0.0009972265735777987,
      "loss": 1.6404,
      "step": 3984
    },
    {
      "epoch": 0.17722644613400915,
      "grad_norm": 0.11740437895059586,
      "learning_rate": 0.000997222858048658,
      "loss": 1.6357,
      "step": 3986
    },
    {
      "epoch": 0.17731537059268151,
      "grad_norm": 0.11547745764255524,
      "learning_rate": 0.000997219140039287,
      "loss": 1.6347,
      "step": 3988
    },
    {
      "epoch": 0.17740429505135388,
      "grad_norm": 0.12119320780038834,
      "learning_rate": 0.0009972154195497043,
      "loss": 1.6375,
      "step": 3990
    },
    {
      "epoch": 0.17749321951002622,
      "grad_norm": 0.11464416235685349,
      "learning_rate": 0.0009972116965799287,
      "loss": 1.6361,
      "step": 3992
    },
    {
      "epoch": 0.1775821439686986,
      "grad_norm": 0.11988995969295502,
      "learning_rate": 0.0009972079711299787,
      "loss": 1.6456,
      "step": 3994
    },
    {
      "epoch": 0.17767106842737096,
      "grad_norm": 0.12310869991779327,
      "learning_rate": 0.0009972042431998726,
      "loss": 1.6377,
      "step": 3996
    },
    {
      "epoch": 0.1777599928860433,
      "grad_norm": 0.11940263211727142,
      "learning_rate": 0.0009972005127896295,
      "loss": 1.6408,
      "step": 3998
    },
    {
      "epoch": 0.17784891734471567,
      "grad_norm": 0.11837908625602722,
      "learning_rate": 0.0009971967798992675,
      "loss": 1.6408,
      "step": 4000
    },
    {
      "epoch": 0.17784891734471567,
      "eval_loss": 1.6105084419250488,
      "eval_runtime": 12.5593,
      "eval_samples_per_second": 550.192,
      "eval_steps_per_second": 68.794,
      "step": 4000
    },
    {
      "epoch": 0.177937841803388,
      "grad_norm": 0.12262357026338577,
      "learning_rate": 0.0009971930445288055,
      "loss": 1.6371,
      "step": 4002
    },
    {
      "epoch": 0.17802676626206038,
      "grad_norm": 0.1207030713558197,
      "learning_rate": 0.0009971893066782619,
      "loss": 1.6371,
      "step": 4004
    },
    {
      "epoch": 0.17811569072073274,
      "grad_norm": 0.12452663481235504,
      "learning_rate": 0.0009971855663476557,
      "loss": 1.6377,
      "step": 4006
    },
    {
      "epoch": 0.17820461517940508,
      "grad_norm": 0.12130036950111389,
      "learning_rate": 0.0009971818235370053,
      "loss": 1.6384,
      "step": 4008
    },
    {
      "epoch": 0.17829353963807745,
      "grad_norm": 0.125193789601326,
      "learning_rate": 0.0009971780782463294,
      "loss": 1.6409,
      "step": 4010
    },
    {
      "epoch": 0.17838246409674982,
      "grad_norm": 0.11911486834287643,
      "learning_rate": 0.0009971743304756468,
      "loss": 1.6442,
      "step": 4012
    },
    {
      "epoch": 0.17847138855542216,
      "grad_norm": 0.11762598156929016,
      "learning_rate": 0.000997170580224976,
      "loss": 1.6446,
      "step": 4014
    },
    {
      "epoch": 0.17856031301409453,
      "grad_norm": 0.12029220908880234,
      "learning_rate": 0.000997166827494336,
      "loss": 1.6428,
      "step": 4016
    },
    {
      "epoch": 0.1786492374727669,
      "grad_norm": 0.11997038871049881,
      "learning_rate": 0.000997163072283745,
      "loss": 1.6333,
      "step": 4018
    },
    {
      "epoch": 0.17873816193143924,
      "grad_norm": 0.11605632305145264,
      "learning_rate": 0.0009971593145932224,
      "loss": 1.6405,
      "step": 4020
    },
    {
      "epoch": 0.1788270863901116,
      "grad_norm": 0.11872642487287521,
      "learning_rate": 0.0009971555544227865,
      "loss": 1.6388,
      "step": 4022
    },
    {
      "epoch": 0.17891601084878395,
      "grad_norm": 0.11706759035587311,
      "learning_rate": 0.0009971517917724561,
      "loss": 1.6308,
      "step": 4024
    },
    {
      "epoch": 0.17900493530745631,
      "grad_norm": 0.11391259729862213,
      "learning_rate": 0.0009971480266422503,
      "loss": 1.636,
      "step": 4026
    },
    {
      "epoch": 0.17909385976612868,
      "grad_norm": 0.11665000766515732,
      "learning_rate": 0.0009971442590321873,
      "loss": 1.6341,
      "step": 4028
    },
    {
      "epoch": 0.17918278422480102,
      "grad_norm": 0.11588221043348312,
      "learning_rate": 0.0009971404889422865,
      "loss": 1.6384,
      "step": 4030
    },
    {
      "epoch": 0.1792717086834734,
      "grad_norm": 0.12535299360752106,
      "learning_rate": 0.0009971367163725661,
      "loss": 1.6425,
      "step": 4032
    },
    {
      "epoch": 0.17936063314214576,
      "grad_norm": 0.11556640267372131,
      "learning_rate": 0.0009971329413230455,
      "loss": 1.6384,
      "step": 4034
    },
    {
      "epoch": 0.1794495576008181,
      "grad_norm": 0.11973097175359726,
      "learning_rate": 0.0009971291637937431,
      "loss": 1.6318,
      "step": 4036
    },
    {
      "epoch": 0.17953848205949047,
      "grad_norm": 0.11719883978366852,
      "learning_rate": 0.000997125383784678,
      "loss": 1.6386,
      "step": 4038
    },
    {
      "epoch": 0.1796274065181628,
      "grad_norm": 0.11612795293331146,
      "learning_rate": 0.0009971216012958691,
      "loss": 1.6363,
      "step": 4040
    },
    {
      "epoch": 0.17971633097683518,
      "grad_norm": 0.11973612755537033,
      "learning_rate": 0.000997117816327335,
      "loss": 1.6354,
      "step": 4042
    },
    {
      "epoch": 0.17980525543550754,
      "grad_norm": 0.11819841712713242,
      "learning_rate": 0.0009971140288790948,
      "loss": 1.6378,
      "step": 4044
    },
    {
      "epoch": 0.17989417989417988,
      "grad_norm": 0.11543485522270203,
      "learning_rate": 0.000997110238951167,
      "loss": 1.6354,
      "step": 4046
    },
    {
      "epoch": 0.17998310435285225,
      "grad_norm": 0.11066845804452896,
      "learning_rate": 0.000997106446543571,
      "loss": 1.6385,
      "step": 4048
    },
    {
      "epoch": 0.18007202881152462,
      "grad_norm": 0.11605249345302582,
      "learning_rate": 0.0009971026516563258,
      "loss": 1.644,
      "step": 4050
    },
    {
      "epoch": 0.18016095327019696,
      "grad_norm": 0.11649958044290543,
      "learning_rate": 0.0009970988542894496,
      "loss": 1.6386,
      "step": 4052
    },
    {
      "epoch": 0.18024987772886933,
      "grad_norm": 0.11853307485580444,
      "learning_rate": 0.0009970950544429622,
      "loss": 1.6339,
      "step": 4054
    },
    {
      "epoch": 0.1803388021875417,
      "grad_norm": 0.12470807880163193,
      "learning_rate": 0.000997091252116882,
      "loss": 1.6345,
      "step": 4056
    },
    {
      "epoch": 0.18042772664621404,
      "grad_norm": 0.12232770025730133,
      "learning_rate": 0.000997087447311228,
      "loss": 1.6334,
      "step": 4058
    },
    {
      "epoch": 0.1805166511048864,
      "grad_norm": 0.11800771951675415,
      "learning_rate": 0.0009970836400260195,
      "loss": 1.6344,
      "step": 4060
    },
    {
      "epoch": 0.18060557556355875,
      "grad_norm": 0.12016160786151886,
      "learning_rate": 0.0009970798302612752,
      "loss": 1.6384,
      "step": 4062
    },
    {
      "epoch": 0.1806945000222311,
      "grad_norm": 0.11976221203804016,
      "learning_rate": 0.000997076018017014,
      "loss": 1.6344,
      "step": 4064
    },
    {
      "epoch": 0.18078342448090348,
      "grad_norm": 0.11883070319890976,
      "learning_rate": 0.0009970722032932554,
      "loss": 1.6326,
      "step": 4066
    },
    {
      "epoch": 0.18087234893957582,
      "grad_norm": 0.11358685046434402,
      "learning_rate": 0.000997068386090018,
      "loss": 1.6369,
      "step": 4068
    },
    {
      "epoch": 0.1809612733982482,
      "grad_norm": 0.1167619526386261,
      "learning_rate": 0.000997064566407321,
      "loss": 1.6324,
      "step": 4070
    },
    {
      "epoch": 0.18105019785692056,
      "grad_norm": 0.11498478800058365,
      "learning_rate": 0.0009970607442451834,
      "loss": 1.6365,
      "step": 4072
    },
    {
      "epoch": 0.1811391223155929,
      "grad_norm": 0.11521459370851517,
      "learning_rate": 0.0009970569196036244,
      "loss": 1.6374,
      "step": 4074
    },
    {
      "epoch": 0.18122804677426527,
      "grad_norm": 0.12538723647594452,
      "learning_rate": 0.000997053092482663,
      "loss": 1.6398,
      "step": 4076
    },
    {
      "epoch": 0.1813169712329376,
      "grad_norm": 0.1221718043088913,
      "learning_rate": 0.0009970492628823182,
      "loss": 1.6334,
      "step": 4078
    },
    {
      "epoch": 0.18140589569160998,
      "grad_norm": 0.11676152795553207,
      "learning_rate": 0.000997045430802609,
      "loss": 1.6289,
      "step": 4080
    },
    {
      "epoch": 0.18149482015028234,
      "grad_norm": 0.11201895028352737,
      "learning_rate": 0.000997041596243555,
      "loss": 1.637,
      "step": 4082
    },
    {
      "epoch": 0.18158374460895468,
      "grad_norm": 0.12027441710233688,
      "learning_rate": 0.0009970377592051747,
      "loss": 1.6364,
      "step": 4084
    },
    {
      "epoch": 0.18167266906762705,
      "grad_norm": 0.11957783252000809,
      "learning_rate": 0.0009970339196874879,
      "loss": 1.633,
      "step": 4086
    },
    {
      "epoch": 0.18176159352629942,
      "grad_norm": 0.1207532063126564,
      "learning_rate": 0.0009970300776905132,
      "loss": 1.6328,
      "step": 4088
    },
    {
      "epoch": 0.18185051798497176,
      "grad_norm": 0.11650995910167694,
      "learning_rate": 0.00099702623321427,
      "loss": 1.6376,
      "step": 4090
    },
    {
      "epoch": 0.18193944244364413,
      "grad_norm": 0.127812460064888,
      "learning_rate": 0.0009970223862587773,
      "loss": 1.6435,
      "step": 4092
    },
    {
      "epoch": 0.18202836690231647,
      "grad_norm": 0.12352623790502548,
      "learning_rate": 0.0009970185368240547,
      "loss": 1.6333,
      "step": 4094
    },
    {
      "epoch": 0.18211729136098884,
      "grad_norm": 0.12078031152486801,
      "learning_rate": 0.0009970146849101209,
      "loss": 1.6332,
      "step": 4096
    },
    {
      "epoch": 0.1822062158196612,
      "grad_norm": 0.1223541796207428,
      "learning_rate": 0.0009970108305169957,
      "loss": 1.636,
      "step": 4098
    },
    {
      "epoch": 0.18229514027833355,
      "grad_norm": 0.12119990587234497,
      "learning_rate": 0.0009970069736446976,
      "loss": 1.6355,
      "step": 4100
    },
    {
      "epoch": 0.1823840647370059,
      "grad_norm": 0.11915529519319534,
      "learning_rate": 0.0009970031142932463,
      "loss": 1.6365,
      "step": 4102
    },
    {
      "epoch": 0.18247298919567828,
      "grad_norm": 0.1180507242679596,
      "learning_rate": 0.000996999252462661,
      "loss": 1.6277,
      "step": 4104
    },
    {
      "epoch": 0.18256191365435062,
      "grad_norm": 0.11856456100940704,
      "learning_rate": 0.000996995388152961,
      "loss": 1.6403,
      "step": 4106
    },
    {
      "epoch": 0.182650838113023,
      "grad_norm": 0.11693500727415085,
      "learning_rate": 0.0009969915213641654,
      "loss": 1.6333,
      "step": 4108
    },
    {
      "epoch": 0.18273976257169536,
      "grad_norm": 0.11582940071821213,
      "learning_rate": 0.0009969876520962938,
      "loss": 1.6371,
      "step": 4110
    },
    {
      "epoch": 0.1828286870303677,
      "grad_norm": 0.11472149193286896,
      "learning_rate": 0.000996983780349365,
      "loss": 1.636,
      "step": 4112
    },
    {
      "epoch": 0.18291761148904007,
      "grad_norm": 0.11126035451889038,
      "learning_rate": 0.0009969799061233988,
      "loss": 1.6344,
      "step": 4114
    },
    {
      "epoch": 0.1830065359477124,
      "grad_norm": 0.11547966301441193,
      "learning_rate": 0.0009969760294184144,
      "loss": 1.6331,
      "step": 4116
    },
    {
      "epoch": 0.18309546040638477,
      "grad_norm": 0.11593402922153473,
      "learning_rate": 0.0009969721502344308,
      "loss": 1.6322,
      "step": 4118
    },
    {
      "epoch": 0.18318438486505714,
      "grad_norm": 0.11121030151844025,
      "learning_rate": 0.0009969682685714678,
      "loss": 1.6341,
      "step": 4120
    },
    {
      "epoch": 0.18327330932372948,
      "grad_norm": 0.11206244677305222,
      "learning_rate": 0.0009969643844295445,
      "loss": 1.6368,
      "step": 4122
    },
    {
      "epoch": 0.18336223378240185,
      "grad_norm": 0.12471252679824829,
      "learning_rate": 0.0009969604978086806,
      "loss": 1.6369,
      "step": 4124
    },
    {
      "epoch": 0.18345115824107422,
      "grad_norm": 0.12352842092514038,
      "learning_rate": 0.000996956608708895,
      "loss": 1.6323,
      "step": 4126
    },
    {
      "epoch": 0.18354008269974656,
      "grad_norm": 0.11872182041406631,
      "learning_rate": 0.0009969527171302076,
      "loss": 1.6305,
      "step": 4128
    },
    {
      "epoch": 0.18362900715841893,
      "grad_norm": 0.11692368239164352,
      "learning_rate": 0.0009969488230726373,
      "loss": 1.6387,
      "step": 4130
    },
    {
      "epoch": 0.18371793161709127,
      "grad_norm": 0.12516173720359802,
      "learning_rate": 0.000996944926536204,
      "loss": 1.6325,
      "step": 4132
    },
    {
      "epoch": 0.18380685607576364,
      "grad_norm": 0.1195317879319191,
      "learning_rate": 0.0009969410275209266,
      "loss": 1.6341,
      "step": 4134
    },
    {
      "epoch": 0.183895780534436,
      "grad_norm": 0.117120161652565,
      "learning_rate": 0.000996937126026825,
      "loss": 1.6323,
      "step": 4136
    },
    {
      "epoch": 0.18398470499310834,
      "grad_norm": 0.11949758231639862,
      "learning_rate": 0.0009969332220539186,
      "loss": 1.6385,
      "step": 4138
    },
    {
      "epoch": 0.1840736294517807,
      "grad_norm": 0.12189013510942459,
      "learning_rate": 0.0009969293156022268,
      "loss": 1.6276,
      "step": 4140
    },
    {
      "epoch": 0.18416255391045308,
      "grad_norm": 0.12576699256896973,
      "learning_rate": 0.000996925406671769,
      "loss": 1.6291,
      "step": 4142
    },
    {
      "epoch": 0.18425147836912542,
      "grad_norm": 0.12639765441417694,
      "learning_rate": 0.000996921495262565,
      "loss": 1.6325,
      "step": 4144
    },
    {
      "epoch": 0.1843404028277978,
      "grad_norm": 0.11879011988639832,
      "learning_rate": 0.000996917581374634,
      "loss": 1.6319,
      "step": 4146
    },
    {
      "epoch": 0.18442932728647016,
      "grad_norm": 0.11668401211500168,
      "learning_rate": 0.0009969136650079954,
      "loss": 1.6372,
      "step": 4148
    },
    {
      "epoch": 0.1845182517451425,
      "grad_norm": 0.11744611710309982,
      "learning_rate": 0.000996909746162669,
      "loss": 1.6369,
      "step": 4150
    },
    {
      "epoch": 0.18460717620381487,
      "grad_norm": 0.12382924556732178,
      "learning_rate": 0.0009969058248386745,
      "loss": 1.6325,
      "step": 4152
    },
    {
      "epoch": 0.1846961006624872,
      "grad_norm": 0.1276998668909073,
      "learning_rate": 0.0009969019010360313,
      "loss": 1.6285,
      "step": 4154
    },
    {
      "epoch": 0.18478502512115957,
      "grad_norm": 0.1217699944972992,
      "learning_rate": 0.0009968979747547587,
      "loss": 1.6361,
      "step": 4156
    },
    {
      "epoch": 0.18487394957983194,
      "grad_norm": 0.12193545699119568,
      "learning_rate": 0.0009968940459948769,
      "loss": 1.637,
      "step": 4158
    },
    {
      "epoch": 0.18496287403850428,
      "grad_norm": 0.11924280226230621,
      "learning_rate": 0.0009968901147564047,
      "loss": 1.641,
      "step": 4160
    },
    {
      "epoch": 0.18505179849717665,
      "grad_norm": 0.11626352369785309,
      "learning_rate": 0.0009968861810393623,
      "loss": 1.6306,
      "step": 4162
    },
    {
      "epoch": 0.18514072295584902,
      "grad_norm": 0.1163124293088913,
      "learning_rate": 0.000996882244843769,
      "loss": 1.6318,
      "step": 4164
    },
    {
      "epoch": 0.18522964741452136,
      "grad_norm": 0.11747079342603683,
      "learning_rate": 0.0009968783061696448,
      "loss": 1.6351,
      "step": 4166
    },
    {
      "epoch": 0.18531857187319373,
      "grad_norm": 0.11650542914867401,
      "learning_rate": 0.000996874365017009,
      "loss": 1.6323,
      "step": 4168
    },
    {
      "epoch": 0.18540749633186607,
      "grad_norm": 0.12692221999168396,
      "learning_rate": 0.0009968704213858814,
      "loss": 1.6342,
      "step": 4170
    },
    {
      "epoch": 0.18549642079053844,
      "grad_norm": 0.11849970370531082,
      "learning_rate": 0.0009968664752762818,
      "loss": 1.6372,
      "step": 4172
    },
    {
      "epoch": 0.1855853452492108,
      "grad_norm": 0.11646901071071625,
      "learning_rate": 0.0009968625266882295,
      "loss": 1.6309,
      "step": 4174
    },
    {
      "epoch": 0.18567426970788314,
      "grad_norm": 0.1176559180021286,
      "learning_rate": 0.0009968585756217447,
      "loss": 1.6308,
      "step": 4176
    },
    {
      "epoch": 0.1857631941665555,
      "grad_norm": 0.11736711859703064,
      "learning_rate": 0.0009968546220768468,
      "loss": 1.638,
      "step": 4178
    },
    {
      "epoch": 0.18585211862522788,
      "grad_norm": 0.11711812019348145,
      "learning_rate": 0.0009968506660535555,
      "loss": 1.6333,
      "step": 4180
    },
    {
      "epoch": 0.18594104308390022,
      "grad_norm": 0.12779754400253296,
      "learning_rate": 0.0009968467075518906,
      "loss": 1.6394,
      "step": 4182
    },
    {
      "epoch": 0.1860299675425726,
      "grad_norm": 0.11566952615976334,
      "learning_rate": 0.000996842746571872,
      "loss": 1.6316,
      "step": 4184
    },
    {
      "epoch": 0.18611889200124493,
      "grad_norm": 0.11619218438863754,
      "learning_rate": 0.000996838783113519,
      "loss": 1.6308,
      "step": 4186
    },
    {
      "epoch": 0.1862078164599173,
      "grad_norm": 0.11629801988601685,
      "learning_rate": 0.000996834817176852,
      "loss": 1.6298,
      "step": 4188
    },
    {
      "epoch": 0.18629674091858967,
      "grad_norm": 0.11523006111383438,
      "learning_rate": 0.0009968308487618902,
      "loss": 1.6338,
      "step": 4190
    },
    {
      "epoch": 0.186385665377262,
      "grad_norm": 0.11307571083307266,
      "learning_rate": 0.0009968268778686538,
      "loss": 1.6333,
      "step": 4192
    },
    {
      "epoch": 0.18647458983593437,
      "grad_norm": 0.11505776643753052,
      "learning_rate": 0.0009968229044971624,
      "loss": 1.6365,
      "step": 4194
    },
    {
      "epoch": 0.18656351429460674,
      "grad_norm": 0.11206153780221939,
      "learning_rate": 0.000996818928647436,
      "loss": 1.6315,
      "step": 4196
    },
    {
      "epoch": 0.18665243875327908,
      "grad_norm": 0.11358271539211273,
      "learning_rate": 0.0009968149503194943,
      "loss": 1.625,
      "step": 4198
    },
    {
      "epoch": 0.18674136321195145,
      "grad_norm": 0.11253488063812256,
      "learning_rate": 0.000996810969513357,
      "loss": 1.6295,
      "step": 4200
    },
    {
      "epoch": 0.18683028767062382,
      "grad_norm": 0.10948602855205536,
      "learning_rate": 0.0009968069862290442,
      "loss": 1.6361,
      "step": 4202
    },
    {
      "epoch": 0.18691921212929616,
      "grad_norm": 0.11469581723213196,
      "learning_rate": 0.0009968030004665759,
      "loss": 1.6329,
      "step": 4204
    },
    {
      "epoch": 0.18700813658796853,
      "grad_norm": 0.12032172828912735,
      "learning_rate": 0.0009967990122259714,
      "loss": 1.6314,
      "step": 4206
    },
    {
      "epoch": 0.18709706104664087,
      "grad_norm": 0.12292454391717911,
      "learning_rate": 0.0009967950215072512,
      "loss": 1.6348,
      "step": 4208
    },
    {
      "epoch": 0.18718598550531323,
      "grad_norm": 0.1108386218547821,
      "learning_rate": 0.0009967910283104349,
      "loss": 1.6317,
      "step": 4210
    },
    {
      "epoch": 0.1872749099639856,
      "grad_norm": 0.11762715131044388,
      "learning_rate": 0.0009967870326355423,
      "loss": 1.6301,
      "step": 4212
    },
    {
      "epoch": 0.18736383442265794,
      "grad_norm": 0.1212281808257103,
      "learning_rate": 0.0009967830344825939,
      "loss": 1.6304,
      "step": 4214
    },
    {
      "epoch": 0.1874527588813303,
      "grad_norm": 0.11403360962867737,
      "learning_rate": 0.0009967790338516092,
      "loss": 1.6324,
      "step": 4216
    },
    {
      "epoch": 0.18754168334000268,
      "grad_norm": 0.11312601715326309,
      "learning_rate": 0.000996775030742608,
      "loss": 1.634,
      "step": 4218
    },
    {
      "epoch": 0.18763060779867502,
      "grad_norm": 0.11520449817180634,
      "learning_rate": 0.0009967710251556105,
      "loss": 1.6309,
      "step": 4220
    },
    {
      "epoch": 0.1877195322573474,
      "grad_norm": 0.11794621497392654,
      "learning_rate": 0.0009967670170906368,
      "loss": 1.633,
      "step": 4222
    },
    {
      "epoch": 0.18780845671601973,
      "grad_norm": 0.11893275380134583,
      "learning_rate": 0.0009967630065477067,
      "loss": 1.6297,
      "step": 4224
    },
    {
      "epoch": 0.1878973811746921,
      "grad_norm": 0.11467410624027252,
      "learning_rate": 0.0009967589935268402,
      "loss": 1.63,
      "step": 4226
    },
    {
      "epoch": 0.18798630563336446,
      "grad_norm": 0.11384375393390656,
      "learning_rate": 0.0009967549780280575,
      "loss": 1.624,
      "step": 4228
    },
    {
      "epoch": 0.1880752300920368,
      "grad_norm": 0.11653546243906021,
      "learning_rate": 0.0009967509600513785,
      "loss": 1.6286,
      "step": 4230
    },
    {
      "epoch": 0.18816415455070917,
      "grad_norm": 0.11761261522769928,
      "learning_rate": 0.0009967469395968233,
      "loss": 1.629,
      "step": 4232
    },
    {
      "epoch": 0.18825307900938154,
      "grad_norm": 0.11235671490430832,
      "learning_rate": 0.0009967429166644118,
      "loss": 1.6301,
      "step": 4234
    },
    {
      "epoch": 0.18834200346805388,
      "grad_norm": 0.11122483015060425,
      "learning_rate": 0.000996738891254164,
      "loss": 1.6322,
      "step": 4236
    },
    {
      "epoch": 0.18843092792672625,
      "grad_norm": 0.11003600060939789,
      "learning_rate": 0.0009967348633661004,
      "loss": 1.6245,
      "step": 4238
    },
    {
      "epoch": 0.18851985238539862,
      "grad_norm": 0.12158048897981644,
      "learning_rate": 0.000996730833000241,
      "loss": 1.6395,
      "step": 4240
    },
    {
      "epoch": 0.18860877684407096,
      "grad_norm": 0.12556138634681702,
      "learning_rate": 0.0009967268001566053,
      "loss": 1.6307,
      "step": 4242
    },
    {
      "epoch": 0.18869770130274333,
      "grad_norm": 0.1180553287267685,
      "learning_rate": 0.0009967227648352143,
      "loss": 1.625,
      "step": 4244
    },
    {
      "epoch": 0.18878662576141567,
      "grad_norm": 0.11684153228998184,
      "learning_rate": 0.0009967187270360875,
      "loss": 1.6378,
      "step": 4246
    },
    {
      "epoch": 0.18887555022008803,
      "grad_norm": 0.11572734266519547,
      "learning_rate": 0.000996714686759245,
      "loss": 1.6324,
      "step": 4248
    },
    {
      "epoch": 0.1889644746787604,
      "grad_norm": 0.1246868148446083,
      "learning_rate": 0.0009967106440047077,
      "loss": 1.6308,
      "step": 4250
    },
    {
      "epoch": 0.18905339913743274,
      "grad_norm": 0.13137935101985931,
      "learning_rate": 0.0009967065987724951,
      "loss": 1.627,
      "step": 4252
    },
    {
      "epoch": 0.1891423235961051,
      "grad_norm": 0.12096533179283142,
      "learning_rate": 0.0009967025510626273,
      "loss": 1.6227,
      "step": 4254
    },
    {
      "epoch": 0.18923124805477748,
      "grad_norm": 0.11495944112539291,
      "learning_rate": 0.0009966985008751249,
      "loss": 1.6305,
      "step": 4256
    },
    {
      "epoch": 0.18932017251344982,
      "grad_norm": 0.10990019887685776,
      "learning_rate": 0.000996694448210008,
      "loss": 1.6281,
      "step": 4258
    },
    {
      "epoch": 0.1894090969721222,
      "grad_norm": 0.12176516652107239,
      "learning_rate": 0.0009966903930672965,
      "loss": 1.6315,
      "step": 4260
    },
    {
      "epoch": 0.18949802143079453,
      "grad_norm": 0.11939161270856857,
      "learning_rate": 0.000996686335447011,
      "loss": 1.6318,
      "step": 4262
    },
    {
      "epoch": 0.1895869458894669,
      "grad_norm": 0.11153752356767654,
      "learning_rate": 0.0009966822753491717,
      "loss": 1.6257,
      "step": 4264
    },
    {
      "epoch": 0.18967587034813926,
      "grad_norm": 0.10733985155820847,
      "learning_rate": 0.0009966782127737988,
      "loss": 1.6348,
      "step": 4266
    },
    {
      "epoch": 0.1897647948068116,
      "grad_norm": 0.1079716607928276,
      "learning_rate": 0.0009966741477209125,
      "loss": 1.6324,
      "step": 4268
    },
    {
      "epoch": 0.18985371926548397,
      "grad_norm": 0.11477085947990417,
      "learning_rate": 0.0009966700801905329,
      "loss": 1.6324,
      "step": 4270
    },
    {
      "epoch": 0.18994264372415634,
      "grad_norm": 0.12020217627286911,
      "learning_rate": 0.0009966660101826807,
      "loss": 1.6279,
      "step": 4272
    },
    {
      "epoch": 0.19003156818282868,
      "grad_norm": 0.1192086786031723,
      "learning_rate": 0.0009966619376973757,
      "loss": 1.6281,
      "step": 4274
    },
    {
      "epoch": 0.19012049264150105,
      "grad_norm": 0.11376933008432388,
      "learning_rate": 0.000996657862734639,
      "loss": 1.6346,
      "step": 4276
    },
    {
      "epoch": 0.1902094171001734,
      "grad_norm": 0.12346582859754562,
      "learning_rate": 0.0009966537852944901,
      "loss": 1.634,
      "step": 4278
    },
    {
      "epoch": 0.19029834155884576,
      "grad_norm": 0.11636168509721756,
      "learning_rate": 0.0009966497053769498,
      "loss": 1.6221,
      "step": 4280
    },
    {
      "epoch": 0.19038726601751813,
      "grad_norm": 0.11525320261716843,
      "learning_rate": 0.0009966456229820381,
      "loss": 1.6296,
      "step": 4282
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 0.11658184975385666,
      "learning_rate": 0.000996641538109776,
      "loss": 1.6273,
      "step": 4284
    },
    {
      "epoch": 0.19056511493486283,
      "grad_norm": 0.11971661448478699,
      "learning_rate": 0.0009966374507601832,
      "loss": 1.6275,
      "step": 4286
    },
    {
      "epoch": 0.1906540393935352,
      "grad_norm": 0.11325690895318985,
      "learning_rate": 0.0009966333609332805,
      "loss": 1.6265,
      "step": 4288
    },
    {
      "epoch": 0.19074296385220754,
      "grad_norm": 0.1084250956773758,
      "learning_rate": 0.000996629268629088,
      "loss": 1.6298,
      "step": 4290
    },
    {
      "epoch": 0.1908318883108799,
      "grad_norm": 0.11336717754602432,
      "learning_rate": 0.0009966251738476265,
      "loss": 1.6285,
      "step": 4292
    },
    {
      "epoch": 0.19092081276955228,
      "grad_norm": 0.11266341805458069,
      "learning_rate": 0.0009966210765889158,
      "loss": 1.6251,
      "step": 4294
    },
    {
      "epoch": 0.19100973722822462,
      "grad_norm": 0.1167646199464798,
      "learning_rate": 0.000996616976852977,
      "loss": 1.6338,
      "step": 4296
    },
    {
      "epoch": 0.191098661686897,
      "grad_norm": 0.1174037903547287,
      "learning_rate": 0.0009966128746398302,
      "loss": 1.6337,
      "step": 4298
    },
    {
      "epoch": 0.19118758614556933,
      "grad_norm": 0.12295513600111008,
      "learning_rate": 0.000996608769949496,
      "loss": 1.6288,
      "step": 4300
    },
    {
      "epoch": 0.1912765106042417,
      "grad_norm": 0.11581303924322128,
      "learning_rate": 0.000996604662781995,
      "loss": 1.6275,
      "step": 4302
    },
    {
      "epoch": 0.19136543506291406,
      "grad_norm": 0.11440426111221313,
      "learning_rate": 0.0009966005531373474,
      "loss": 1.6265,
      "step": 4304
    },
    {
      "epoch": 0.1914543595215864,
      "grad_norm": 0.11269820481538773,
      "learning_rate": 0.0009965964410155737,
      "loss": 1.6274,
      "step": 4306
    },
    {
      "epoch": 0.19154328398025877,
      "grad_norm": 0.11619145423173904,
      "learning_rate": 0.0009965923264166946,
      "loss": 1.6227,
      "step": 4308
    },
    {
      "epoch": 0.19163220843893114,
      "grad_norm": 0.10982932150363922,
      "learning_rate": 0.0009965882093407304,
      "loss": 1.6331,
      "step": 4310
    },
    {
      "epoch": 0.19172113289760348,
      "grad_norm": 0.11294446885585785,
      "learning_rate": 0.0009965840897877018,
      "loss": 1.6257,
      "step": 4312
    },
    {
      "epoch": 0.19181005735627585,
      "grad_norm": 0.11385387182235718,
      "learning_rate": 0.0009965799677576295,
      "loss": 1.631,
      "step": 4314
    },
    {
      "epoch": 0.1918989818149482,
      "grad_norm": 0.11767963320016861,
      "learning_rate": 0.0009965758432505338,
      "loss": 1.6264,
      "step": 4316
    },
    {
      "epoch": 0.19198790627362056,
      "grad_norm": 0.11952953040599823,
      "learning_rate": 0.0009965717162664354,
      "loss": 1.6238,
      "step": 4318
    },
    {
      "epoch": 0.19207683073229292,
      "grad_norm": 0.11729772388935089,
      "learning_rate": 0.000996567586805355,
      "loss": 1.6218,
      "step": 4320
    },
    {
      "epoch": 0.19216575519096527,
      "grad_norm": 0.12014216929674149,
      "learning_rate": 0.0009965634548673129,
      "loss": 1.6236,
      "step": 4322
    },
    {
      "epoch": 0.19225467964963763,
      "grad_norm": 0.11585142463445663,
      "learning_rate": 0.0009965593204523298,
      "loss": 1.6249,
      "step": 4324
    },
    {
      "epoch": 0.19234360410831,
      "grad_norm": 0.11284307390451431,
      "learning_rate": 0.0009965551835604265,
      "loss": 1.6247,
      "step": 4326
    },
    {
      "epoch": 0.19243252856698234,
      "grad_norm": 0.11561114341020584,
      "learning_rate": 0.0009965510441916235,
      "loss": 1.6257,
      "step": 4328
    },
    {
      "epoch": 0.1925214530256547,
      "grad_norm": 0.11217344552278519,
      "learning_rate": 0.0009965469023459413,
      "loss": 1.6261,
      "step": 4330
    },
    {
      "epoch": 0.19261037748432708,
      "grad_norm": 0.11755629628896713,
      "learning_rate": 0.000996542758023401,
      "loss": 1.633,
      "step": 4332
    },
    {
      "epoch": 0.19269930194299942,
      "grad_norm": 0.11235187947750092,
      "learning_rate": 0.0009965386112240227,
      "loss": 1.627,
      "step": 4334
    },
    {
      "epoch": 0.1927882264016718,
      "grad_norm": 0.11830323189496994,
      "learning_rate": 0.0009965344619478275,
      "loss": 1.6281,
      "step": 4336
    },
    {
      "epoch": 0.19287715086034413,
      "grad_norm": 0.11256743967533112,
      "learning_rate": 0.0009965303101948362,
      "loss": 1.6258,
      "step": 4338
    },
    {
      "epoch": 0.1929660753190165,
      "grad_norm": 0.11304402351379395,
      "learning_rate": 0.0009965261559650692,
      "loss": 1.6223,
      "step": 4340
    },
    {
      "epoch": 0.19305499977768886,
      "grad_norm": 0.11517365276813507,
      "learning_rate": 0.000996521999258547,
      "loss": 1.6252,
      "step": 4342
    },
    {
      "epoch": 0.1931439242363612,
      "grad_norm": 0.11518683284521103,
      "learning_rate": 0.000996517840075291,
      "loss": 1.6261,
      "step": 4344
    },
    {
      "epoch": 0.19323284869503357,
      "grad_norm": 0.10763294994831085,
      "learning_rate": 0.0009965136784153214,
      "loss": 1.6236,
      "step": 4346
    },
    {
      "epoch": 0.19332177315370594,
      "grad_norm": 0.11887557059526443,
      "learning_rate": 0.0009965095142786594,
      "loss": 1.6274,
      "step": 4348
    },
    {
      "epoch": 0.19341069761237828,
      "grad_norm": 0.11843670904636383,
      "learning_rate": 0.000996505347665325,
      "loss": 1.6258,
      "step": 4350
    },
    {
      "epoch": 0.19349962207105065,
      "grad_norm": 0.12076882272958755,
      "learning_rate": 0.0009965011785753402,
      "loss": 1.6246,
      "step": 4352
    },
    {
      "epoch": 0.193588546529723,
      "grad_norm": 0.11038120090961456,
      "learning_rate": 0.0009964970070087246,
      "loss": 1.6222,
      "step": 4354
    },
    {
      "epoch": 0.19367747098839536,
      "grad_norm": 0.11580009013414383,
      "learning_rate": 0.0009964928329654998,
      "loss": 1.6271,
      "step": 4356
    },
    {
      "epoch": 0.19376639544706772,
      "grad_norm": 0.11985108256340027,
      "learning_rate": 0.0009964886564456862,
      "loss": 1.6326,
      "step": 4358
    },
    {
      "epoch": 0.19385531990574006,
      "grad_norm": 0.11022962629795074,
      "learning_rate": 0.0009964844774493049,
      "loss": 1.6277,
      "step": 4360
    },
    {
      "epoch": 0.19394424436441243,
      "grad_norm": 0.11388500034809113,
      "learning_rate": 0.0009964802959763765,
      "loss": 1.6208,
      "step": 4362
    },
    {
      "epoch": 0.1940331688230848,
      "grad_norm": 0.11820525676012039,
      "learning_rate": 0.0009964761120269219,
      "loss": 1.6211,
      "step": 4364
    },
    {
      "epoch": 0.19412209328175714,
      "grad_norm": 0.1088062971830368,
      "learning_rate": 0.0009964719256009623,
      "loss": 1.6206,
      "step": 4366
    },
    {
      "epoch": 0.1942110177404295,
      "grad_norm": 0.11145839095115662,
      "learning_rate": 0.000996467736698518,
      "loss": 1.6248,
      "step": 4368
    },
    {
      "epoch": 0.19429994219910185,
      "grad_norm": 0.11437372118234634,
      "learning_rate": 0.0009964635453196104,
      "loss": 1.629,
      "step": 4370
    },
    {
      "epoch": 0.19438886665777422,
      "grad_norm": 0.11181259900331497,
      "learning_rate": 0.0009964593514642601,
      "loss": 1.6234,
      "step": 4372
    },
    {
      "epoch": 0.19447779111644659,
      "grad_norm": 0.10853833705186844,
      "learning_rate": 0.0009964551551324883,
      "loss": 1.6206,
      "step": 4374
    },
    {
      "epoch": 0.19456671557511893,
      "grad_norm": 0.11572292447090149,
      "learning_rate": 0.0009964509563243158,
      "loss": 1.6234,
      "step": 4376
    },
    {
      "epoch": 0.1946556400337913,
      "grad_norm": 0.1133660301566124,
      "learning_rate": 0.0009964467550397634,
      "loss": 1.6261,
      "step": 4378
    },
    {
      "epoch": 0.19474456449246366,
      "grad_norm": 0.1107514351606369,
      "learning_rate": 0.0009964425512788522,
      "loss": 1.6302,
      "step": 4380
    },
    {
      "epoch": 0.194833488951136,
      "grad_norm": 0.11202812194824219,
      "learning_rate": 0.0009964383450416032,
      "loss": 1.6259,
      "step": 4382
    },
    {
      "epoch": 0.19492241340980837,
      "grad_norm": 0.11862754821777344,
      "learning_rate": 0.0009964341363280374,
      "loss": 1.6295,
      "step": 4384
    },
    {
      "epoch": 0.19501133786848074,
      "grad_norm": 0.11439205706119537,
      "learning_rate": 0.0009964299251381756,
      "loss": 1.6283,
      "step": 4386
    },
    {
      "epoch": 0.19510026232715308,
      "grad_norm": 0.11098136007785797,
      "learning_rate": 0.000996425711472039,
      "loss": 1.6241,
      "step": 4388
    },
    {
      "epoch": 0.19518918678582545,
      "grad_norm": 0.12104503810405731,
      "learning_rate": 0.0009964214953296483,
      "loss": 1.6298,
      "step": 4390
    },
    {
      "epoch": 0.1952781112444978,
      "grad_norm": 0.1151038259267807,
      "learning_rate": 0.000996417276711025,
      "loss": 1.6237,
      "step": 4392
    },
    {
      "epoch": 0.19536703570317016,
      "grad_norm": 0.12040484696626663,
      "learning_rate": 0.00099641305561619,
      "loss": 1.6284,
      "step": 4394
    },
    {
      "epoch": 0.19545596016184252,
      "grad_norm": 0.11409257352352142,
      "learning_rate": 0.0009964088320451642,
      "loss": 1.6252,
      "step": 4396
    },
    {
      "epoch": 0.19554488462051486,
      "grad_norm": 0.1078067272901535,
      "learning_rate": 0.0009964046059979688,
      "loss": 1.6247,
      "step": 4398
    },
    {
      "epoch": 0.19563380907918723,
      "grad_norm": 0.11023259162902832,
      "learning_rate": 0.0009964003774746246,
      "loss": 1.6276,
      "step": 4400
    },
    {
      "epoch": 0.1957227335378596,
      "grad_norm": 0.11113858222961426,
      "learning_rate": 0.0009963961464751533,
      "loss": 1.6256,
      "step": 4402
    },
    {
      "epoch": 0.19581165799653194,
      "grad_norm": 0.11487720906734467,
      "learning_rate": 0.0009963919129995754,
      "loss": 1.6264,
      "step": 4404
    },
    {
      "epoch": 0.1959005824552043,
      "grad_norm": 0.10912340879440308,
      "learning_rate": 0.0009963876770479121,
      "loss": 1.632,
      "step": 4406
    },
    {
      "epoch": 0.19598950691387665,
      "grad_norm": 0.11209183931350708,
      "learning_rate": 0.0009963834386201849,
      "loss": 1.6318,
      "step": 4408
    },
    {
      "epoch": 0.19607843137254902,
      "grad_norm": 0.10975425690412521,
      "learning_rate": 0.0009963791977164146,
      "loss": 1.6252,
      "step": 4410
    },
    {
      "epoch": 0.19616735583122138,
      "grad_norm": 0.11096843332052231,
      "learning_rate": 0.0009963749543366225,
      "loss": 1.623,
      "step": 4412
    },
    {
      "epoch": 0.19625628028989373,
      "grad_norm": 0.10731083899736404,
      "learning_rate": 0.0009963707084808299,
      "loss": 1.6265,
      "step": 4414
    },
    {
      "epoch": 0.1963452047485661,
      "grad_norm": 0.10330415517091751,
      "learning_rate": 0.0009963664601490574,
      "loss": 1.6283,
      "step": 4416
    },
    {
      "epoch": 0.19643412920723846,
      "grad_norm": 0.10680415481328964,
      "learning_rate": 0.000996362209341327,
      "loss": 1.6248,
      "step": 4418
    },
    {
      "epoch": 0.1965230536659108,
      "grad_norm": 0.11378171294927597,
      "learning_rate": 0.0009963579560576593,
      "loss": 1.6208,
      "step": 4420
    },
    {
      "epoch": 0.19661197812458317,
      "grad_norm": 0.11050064116716385,
      "learning_rate": 0.0009963537002980758,
      "loss": 1.619,
      "step": 4422
    },
    {
      "epoch": 0.19670090258325554,
      "grad_norm": 0.11730819940567017,
      "learning_rate": 0.0009963494420625976,
      "loss": 1.6275,
      "step": 4424
    },
    {
      "epoch": 0.19678982704192788,
      "grad_norm": 0.11182375252246857,
      "learning_rate": 0.000996345181351246,
      "loss": 1.6219,
      "step": 4426
    },
    {
      "epoch": 0.19687875150060025,
      "grad_norm": 0.10767224431037903,
      "learning_rate": 0.0009963409181640423,
      "loss": 1.6238,
      "step": 4428
    },
    {
      "epoch": 0.1969676759592726,
      "grad_norm": 0.10671267658472061,
      "learning_rate": 0.0009963366525010077,
      "loss": 1.6212,
      "step": 4430
    },
    {
      "epoch": 0.19705660041794495,
      "grad_norm": 0.10829255729913712,
      "learning_rate": 0.0009963323843621634,
      "loss": 1.6243,
      "step": 4432
    },
    {
      "epoch": 0.19714552487661732,
      "grad_norm": 0.11310666054487228,
      "learning_rate": 0.0009963281137475308,
      "loss": 1.6223,
      "step": 4434
    },
    {
      "epoch": 0.19723444933528966,
      "grad_norm": 0.110753633081913,
      "learning_rate": 0.0009963238406571313,
      "loss": 1.6255,
      "step": 4436
    },
    {
      "epoch": 0.19732337379396203,
      "grad_norm": 0.11032912880182266,
      "learning_rate": 0.000996319565090986,
      "loss": 1.6242,
      "step": 4438
    },
    {
      "epoch": 0.1974122982526344,
      "grad_norm": 0.11288750916719437,
      "learning_rate": 0.0009963152870491163,
      "loss": 1.6233,
      "step": 4440
    },
    {
      "epoch": 0.19750122271130674,
      "grad_norm": 0.11633003503084183,
      "learning_rate": 0.0009963110065315437,
      "loss": 1.6237,
      "step": 4442
    },
    {
      "epoch": 0.1975901471699791,
      "grad_norm": 0.11502166092395782,
      "learning_rate": 0.0009963067235382894,
      "loss": 1.6245,
      "step": 4444
    },
    {
      "epoch": 0.19767907162865145,
      "grad_norm": 0.11149843782186508,
      "learning_rate": 0.0009963024380693745,
      "loss": 1.6264,
      "step": 4446
    },
    {
      "epoch": 0.19776799608732382,
      "grad_norm": 0.10773973912000656,
      "learning_rate": 0.0009962981501248208,
      "loss": 1.6213,
      "step": 4448
    },
    {
      "epoch": 0.19785692054599618,
      "grad_norm": 0.1118425726890564,
      "learning_rate": 0.0009962938597046498,
      "loss": 1.6326,
      "step": 4450
    },
    {
      "epoch": 0.19794584500466852,
      "grad_norm": 0.1150190681219101,
      "learning_rate": 0.0009962895668088824,
      "loss": 1.6223,
      "step": 4452
    },
    {
      "epoch": 0.1980347694633409,
      "grad_norm": 0.11391531676054001,
      "learning_rate": 0.0009962852714375404,
      "loss": 1.6274,
      "step": 4454
    },
    {
      "epoch": 0.19812369392201326,
      "grad_norm": 0.11059886962175369,
      "learning_rate": 0.0009962809735906452,
      "loss": 1.6264,
      "step": 4456
    },
    {
      "epoch": 0.1982126183806856,
      "grad_norm": 0.11909958720207214,
      "learning_rate": 0.0009962766732682178,
      "loss": 1.6188,
      "step": 4458
    },
    {
      "epoch": 0.19830154283935797,
      "grad_norm": 0.11559052765369415,
      "learning_rate": 0.00099627237047028,
      "loss": 1.6273,
      "step": 4460
    },
    {
      "epoch": 0.1983904672980303,
      "grad_norm": 0.11191637068986893,
      "learning_rate": 0.0009962680651968534,
      "loss": 1.6214,
      "step": 4462
    },
    {
      "epoch": 0.19847939175670268,
      "grad_norm": 0.113275907933712,
      "learning_rate": 0.0009962637574479592,
      "loss": 1.6233,
      "step": 4464
    },
    {
      "epoch": 0.19856831621537505,
      "grad_norm": 0.10795215517282486,
      "learning_rate": 0.0009962594472236192,
      "loss": 1.6249,
      "step": 4466
    },
    {
      "epoch": 0.1986572406740474,
      "grad_norm": 0.11079131066799164,
      "learning_rate": 0.0009962551345238544,
      "loss": 1.6265,
      "step": 4468
    },
    {
      "epoch": 0.19874616513271975,
      "grad_norm": 0.10867098718881607,
      "learning_rate": 0.000996250819348687,
      "loss": 1.6255,
      "step": 4470
    },
    {
      "epoch": 0.19883508959139212,
      "grad_norm": 0.10837139934301376,
      "learning_rate": 0.000996246501698138,
      "loss": 1.6283,
      "step": 4472
    },
    {
      "epoch": 0.19892401405006446,
      "grad_norm": 0.10983740538358688,
      "learning_rate": 0.0009962421815722289,
      "loss": 1.6196,
      "step": 4474
    },
    {
      "epoch": 0.19901293850873683,
      "grad_norm": 0.107894666492939,
      "learning_rate": 0.0009962378589709814,
      "loss": 1.6216,
      "step": 4476
    },
    {
      "epoch": 0.1991018629674092,
      "grad_norm": 0.10687065124511719,
      "learning_rate": 0.0009962335338944172,
      "loss": 1.6226,
      "step": 4478
    },
    {
      "epoch": 0.19919078742608154,
      "grad_norm": 0.10734596103429794,
      "learning_rate": 0.0009962292063425578,
      "loss": 1.6241,
      "step": 4480
    },
    {
      "epoch": 0.1992797118847539,
      "grad_norm": 0.1052248552441597,
      "learning_rate": 0.0009962248763154247,
      "loss": 1.6134,
      "step": 4482
    },
    {
      "epoch": 0.19936863634342625,
      "grad_norm": 0.10848673433065414,
      "learning_rate": 0.0009962205438130397,
      "loss": 1.6259,
      "step": 4484
    },
    {
      "epoch": 0.19945756080209862,
      "grad_norm": 0.10908637940883636,
      "learning_rate": 0.000996216208835424,
      "loss": 1.6229,
      "step": 4486
    },
    {
      "epoch": 0.19954648526077098,
      "grad_norm": 0.10700564086437225,
      "learning_rate": 0.0009962118713825996,
      "loss": 1.6207,
      "step": 4488
    },
    {
      "epoch": 0.19963540971944332,
      "grad_norm": 0.11299486458301544,
      "learning_rate": 0.0009962075314545883,
      "loss": 1.6296,
      "step": 4490
    },
    {
      "epoch": 0.1997243341781157,
      "grad_norm": 0.11249647289514542,
      "learning_rate": 0.000996203189051411,
      "loss": 1.6251,
      "step": 4492
    },
    {
      "epoch": 0.19981325863678806,
      "grad_norm": 0.10399112105369568,
      "learning_rate": 0.00099619884417309,
      "loss": 1.6197,
      "step": 4494
    },
    {
      "epoch": 0.1999021830954604,
      "grad_norm": 0.10901904106140137,
      "learning_rate": 0.000996194496819647,
      "loss": 1.6202,
      "step": 4496
    },
    {
      "epoch": 0.19999110755413277,
      "grad_norm": 0.11132825911045074,
      "learning_rate": 0.0009961901469911032,
      "loss": 1.6222,
      "step": 4498
    },
    {
      "epoch": 0.2000800320128051,
      "grad_norm": 0.10988521575927734,
      "learning_rate": 0.0009961857946874807,
      "loss": 1.6208,
      "step": 4500
    },
    {
      "epoch": 0.2000800320128051,
      "eval_loss": 1.5955942869186401,
      "eval_runtime": 12.3611,
      "eval_samples_per_second": 559.011,
      "eval_steps_per_second": 69.897,
      "step": 4500
    },
    {
      "epoch": 0.20016895647147748,
      "grad_norm": 0.11773038655519485,
      "learning_rate": 0.000996181439908801,
      "loss": 1.6182,
      "step": 4502
    },
    {
      "epoch": 0.20025788093014985,
      "grad_norm": 0.11245664954185486,
      "learning_rate": 0.0009961770826550861,
      "loss": 1.6263,
      "step": 4504
    },
    {
      "epoch": 0.20034680538882219,
      "grad_norm": 0.11493460088968277,
      "learning_rate": 0.0009961727229263575,
      "loss": 1.6264,
      "step": 4506
    },
    {
      "epoch": 0.20043572984749455,
      "grad_norm": 0.10975135862827301,
      "learning_rate": 0.0009961683607226368,
      "loss": 1.6218,
      "step": 4508
    },
    {
      "epoch": 0.20052465430616692,
      "grad_norm": 0.1100163534283638,
      "learning_rate": 0.0009961639960439463,
      "loss": 1.6255,
      "step": 4510
    },
    {
      "epoch": 0.20061357876483926,
      "grad_norm": 0.10604830086231232,
      "learning_rate": 0.0009961596288903072,
      "loss": 1.6221,
      "step": 4512
    },
    {
      "epoch": 0.20070250322351163,
      "grad_norm": 0.10983182489871979,
      "learning_rate": 0.0009961552592617415,
      "loss": 1.6193,
      "step": 4514
    },
    {
      "epoch": 0.200791427682184,
      "grad_norm": 0.1092047244310379,
      "learning_rate": 0.000996150887158271,
      "loss": 1.6215,
      "step": 4516
    },
    {
      "epoch": 0.20088035214085634,
      "grad_norm": 0.1078716516494751,
      "learning_rate": 0.0009961465125799176,
      "loss": 1.6246,
      "step": 4518
    },
    {
      "epoch": 0.2009692765995287,
      "grad_norm": 0.11041826754808426,
      "learning_rate": 0.0009961421355267031,
      "loss": 1.6225,
      "step": 4520
    },
    {
      "epoch": 0.20105820105820105,
      "grad_norm": 0.11641491949558258,
      "learning_rate": 0.0009961377559986491,
      "loss": 1.6173,
      "step": 4522
    },
    {
      "epoch": 0.20114712551687342,
      "grad_norm": 0.11122878640890121,
      "learning_rate": 0.0009961333739957779,
      "loss": 1.6179,
      "step": 4524
    },
    {
      "epoch": 0.20123604997554578,
      "grad_norm": 0.11828659474849701,
      "learning_rate": 0.0009961289895181107,
      "loss": 1.6227,
      "step": 4526
    },
    {
      "epoch": 0.20132497443421812,
      "grad_norm": 0.10966084152460098,
      "learning_rate": 0.00099612460256567,
      "loss": 1.6308,
      "step": 4528
    },
    {
      "epoch": 0.2014138988928905,
      "grad_norm": 0.11296512931585312,
      "learning_rate": 0.0009961202131384774,
      "loss": 1.6207,
      "step": 4530
    },
    {
      "epoch": 0.20150282335156286,
      "grad_norm": 0.10687365382909775,
      "learning_rate": 0.0009961158212365548,
      "loss": 1.6192,
      "step": 4532
    },
    {
      "epoch": 0.2015917478102352,
      "grad_norm": 0.10820367932319641,
      "learning_rate": 0.000996111426859924,
      "loss": 1.6254,
      "step": 4534
    },
    {
      "epoch": 0.20168067226890757,
      "grad_norm": 0.1103745698928833,
      "learning_rate": 0.0009961070300086073,
      "loss": 1.6171,
      "step": 4536
    },
    {
      "epoch": 0.2017695967275799,
      "grad_norm": 0.11085197329521179,
      "learning_rate": 0.0009961026306826261,
      "loss": 1.6233,
      "step": 4538
    },
    {
      "epoch": 0.20185852118625228,
      "grad_norm": 0.10982104390859604,
      "learning_rate": 0.000996098228882003,
      "loss": 1.6198,
      "step": 4540
    },
    {
      "epoch": 0.20194744564492464,
      "grad_norm": 0.11612139642238617,
      "learning_rate": 0.0009960938246067592,
      "loss": 1.6163,
      "step": 4542
    },
    {
      "epoch": 0.20203637010359699,
      "grad_norm": 0.11329114437103271,
      "learning_rate": 0.0009960894178569174,
      "loss": 1.6203,
      "step": 4544
    },
    {
      "epoch": 0.20212529456226935,
      "grad_norm": 0.11832458525896072,
      "learning_rate": 0.0009960850086324989,
      "loss": 1.6259,
      "step": 4546
    },
    {
      "epoch": 0.20221421902094172,
      "grad_norm": 0.1062120795249939,
      "learning_rate": 0.0009960805969335264,
      "loss": 1.6238,
      "step": 4548
    },
    {
      "epoch": 0.20230314347961406,
      "grad_norm": 0.10603608936071396,
      "learning_rate": 0.0009960761827600211,
      "loss": 1.6186,
      "step": 4550
    },
    {
      "epoch": 0.20239206793828643,
      "grad_norm": 0.10631296783685684,
      "learning_rate": 0.0009960717661120057,
      "loss": 1.6155,
      "step": 4552
    },
    {
      "epoch": 0.20248099239695877,
      "grad_norm": 0.11160627007484436,
      "learning_rate": 0.000996067346989502,
      "loss": 1.6248,
      "step": 4554
    },
    {
      "epoch": 0.20256991685563114,
      "grad_norm": 0.12129071354866028,
      "learning_rate": 0.0009960629253925322,
      "loss": 1.6211,
      "step": 4556
    },
    {
      "epoch": 0.2026588413143035,
      "grad_norm": 0.11960272490978241,
      "learning_rate": 0.0009960585013211178,
      "loss": 1.6191,
      "step": 4558
    },
    {
      "epoch": 0.20274776577297585,
      "grad_norm": 0.11304063349962234,
      "learning_rate": 0.0009960540747752816,
      "loss": 1.6207,
      "step": 4560
    },
    {
      "epoch": 0.20283669023164821,
      "grad_norm": 0.11380619555711746,
      "learning_rate": 0.000996049645755045,
      "loss": 1.615,
      "step": 4562
    },
    {
      "epoch": 0.20292561469032058,
      "grad_norm": 0.11661424487829208,
      "learning_rate": 0.0009960452142604307,
      "loss": 1.6163,
      "step": 4564
    },
    {
      "epoch": 0.20301453914899292,
      "grad_norm": 0.1175897866487503,
      "learning_rate": 0.0009960407802914605,
      "loss": 1.6242,
      "step": 4566
    },
    {
      "epoch": 0.2031034636076653,
      "grad_norm": 0.11345373094081879,
      "learning_rate": 0.0009960363438481564,
      "loss": 1.6157,
      "step": 4568
    },
    {
      "epoch": 0.20319238806633766,
      "grad_norm": 0.10861898958683014,
      "learning_rate": 0.000996031904930541,
      "loss": 1.6193,
      "step": 4570
    },
    {
      "epoch": 0.20328131252501,
      "grad_norm": 0.10877344012260437,
      "learning_rate": 0.0009960274635386357,
      "loss": 1.6176,
      "step": 4572
    },
    {
      "epoch": 0.20337023698368237,
      "grad_norm": 0.10538928955793381,
      "learning_rate": 0.0009960230196724633,
      "loss": 1.6164,
      "step": 4574
    },
    {
      "epoch": 0.2034591614423547,
      "grad_norm": 0.10377932339906693,
      "learning_rate": 0.0009960185733320457,
      "loss": 1.6172,
      "step": 4576
    },
    {
      "epoch": 0.20354808590102708,
      "grad_norm": 0.11017095297574997,
      "learning_rate": 0.000996014124517405,
      "loss": 1.61,
      "step": 4578
    },
    {
      "epoch": 0.20363701035969944,
      "grad_norm": 0.10805568099021912,
      "learning_rate": 0.0009960096732285638,
      "loss": 1.6151,
      "step": 4580
    },
    {
      "epoch": 0.20372593481837178,
      "grad_norm": 0.11047879606485367,
      "learning_rate": 0.0009960052194655437,
      "loss": 1.6191,
      "step": 4582
    },
    {
      "epoch": 0.20381485927704415,
      "grad_norm": 0.11243947595357895,
      "learning_rate": 0.0009960007632283674,
      "loss": 1.6166,
      "step": 4584
    },
    {
      "epoch": 0.20390378373571652,
      "grad_norm": 0.11238662898540497,
      "learning_rate": 0.000995996304517057,
      "loss": 1.6176,
      "step": 4586
    },
    {
      "epoch": 0.20399270819438886,
      "grad_norm": 0.11109507828950882,
      "learning_rate": 0.0009959918433316347,
      "loss": 1.6119,
      "step": 4588
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 0.11443541198968887,
      "learning_rate": 0.0009959873796721226,
      "loss": 1.6252,
      "step": 4590
    },
    {
      "epoch": 0.20417055711173357,
      "grad_norm": 0.11701706796884537,
      "learning_rate": 0.0009959829135385433,
      "loss": 1.6235,
      "step": 4592
    },
    {
      "epoch": 0.20425948157040594,
      "grad_norm": 0.11349605768918991,
      "learning_rate": 0.0009959784449309187,
      "loss": 1.617,
      "step": 4594
    },
    {
      "epoch": 0.2043484060290783,
      "grad_norm": 0.11320377141237259,
      "learning_rate": 0.0009959739738492717,
      "loss": 1.6217,
      "step": 4596
    },
    {
      "epoch": 0.20443733048775065,
      "grad_norm": 0.10560119152069092,
      "learning_rate": 0.0009959695002936238,
      "loss": 1.6176,
      "step": 4598
    },
    {
      "epoch": 0.20452625494642301,
      "grad_norm": 0.11200706660747528,
      "learning_rate": 0.0009959650242639976,
      "loss": 1.6199,
      "step": 4600
    },
    {
      "epoch": 0.20461517940509538,
      "grad_norm": 0.11499829590320587,
      "learning_rate": 0.000995960545760416,
      "loss": 1.6197,
      "step": 4602
    },
    {
      "epoch": 0.20470410386376772,
      "grad_norm": 0.10916129499673843,
      "learning_rate": 0.0009959560647829005,
      "loss": 1.6166,
      "step": 4604
    },
    {
      "epoch": 0.2047930283224401,
      "grad_norm": 0.11446643620729446,
      "learning_rate": 0.000995951581331474,
      "loss": 1.6173,
      "step": 4606
    },
    {
      "epoch": 0.20488195278111246,
      "grad_norm": 0.11069472879171371,
      "learning_rate": 0.0009959470954061585,
      "loss": 1.6155,
      "step": 4608
    },
    {
      "epoch": 0.2049708772397848,
      "grad_norm": 0.11012015491724014,
      "learning_rate": 0.000995942607006977,
      "loss": 1.6196,
      "step": 4610
    },
    {
      "epoch": 0.20505980169845717,
      "grad_norm": 0.10821570456027985,
      "learning_rate": 0.000995938116133951,
      "loss": 1.6182,
      "step": 4612
    },
    {
      "epoch": 0.2051487261571295,
      "grad_norm": 0.11206888407468796,
      "learning_rate": 0.0009959336227871034,
      "loss": 1.618,
      "step": 4614
    },
    {
      "epoch": 0.20523765061580188,
      "grad_norm": 0.11033768951892853,
      "learning_rate": 0.0009959291269664567,
      "loss": 1.6268,
      "step": 4616
    },
    {
      "epoch": 0.20532657507447424,
      "grad_norm": 0.10916636139154434,
      "learning_rate": 0.000995924628672033,
      "loss": 1.6132,
      "step": 4618
    },
    {
      "epoch": 0.20541549953314658,
      "grad_norm": 0.10948532074689865,
      "learning_rate": 0.0009959201279038552,
      "loss": 1.6132,
      "step": 4620
    },
    {
      "epoch": 0.20550442399181895,
      "grad_norm": 0.1057814285159111,
      "learning_rate": 0.0009959156246619454,
      "loss": 1.6169,
      "step": 4622
    },
    {
      "epoch": 0.20559334845049132,
      "grad_norm": 0.10795042663812637,
      "learning_rate": 0.000995911118946326,
      "loss": 1.6187,
      "step": 4624
    },
    {
      "epoch": 0.20568227290916366,
      "grad_norm": 0.10685425996780396,
      "learning_rate": 0.00099590661075702,
      "loss": 1.6212,
      "step": 4626
    },
    {
      "epoch": 0.20577119736783603,
      "grad_norm": 0.11638550460338593,
      "learning_rate": 0.000995902100094049,
      "loss": 1.6247,
      "step": 4628
    },
    {
      "epoch": 0.20586012182650837,
      "grad_norm": 0.1139526218175888,
      "learning_rate": 0.0009958975869574366,
      "loss": 1.6143,
      "step": 4630
    },
    {
      "epoch": 0.20594904628518074,
      "grad_norm": 0.1081729456782341,
      "learning_rate": 0.0009958930713472043,
      "loss": 1.617,
      "step": 4632
    },
    {
      "epoch": 0.2060379707438531,
      "grad_norm": 0.11640775948762894,
      "learning_rate": 0.0009958885532633751,
      "loss": 1.6124,
      "step": 4634
    },
    {
      "epoch": 0.20612689520252545,
      "grad_norm": 0.11649386584758759,
      "learning_rate": 0.0009958840327059714,
      "loss": 1.6203,
      "step": 4636
    },
    {
      "epoch": 0.2062158196611978,
      "grad_norm": 0.10759291797876358,
      "learning_rate": 0.000995879509675016,
      "loss": 1.617,
      "step": 4638
    },
    {
      "epoch": 0.20630474411987018,
      "grad_norm": 0.10583692044019699,
      "learning_rate": 0.0009958749841705312,
      "loss": 1.6198,
      "step": 4640
    },
    {
      "epoch": 0.20639366857854252,
      "grad_norm": 0.10626080632209778,
      "learning_rate": 0.0009958704561925397,
      "loss": 1.6194,
      "step": 4642
    },
    {
      "epoch": 0.2064825930372149,
      "grad_norm": 0.10711178928613663,
      "learning_rate": 0.0009958659257410642,
      "loss": 1.6159,
      "step": 4644
    },
    {
      "epoch": 0.20657151749588723,
      "grad_norm": 0.11228474974632263,
      "learning_rate": 0.000995861392816127,
      "loss": 1.622,
      "step": 4646
    },
    {
      "epoch": 0.2066604419545596,
      "grad_norm": 0.10673855245113373,
      "learning_rate": 0.000995856857417751,
      "loss": 1.6169,
      "step": 4648
    },
    {
      "epoch": 0.20674936641323197,
      "grad_norm": 0.10420949757099152,
      "learning_rate": 0.0009958523195459585,
      "loss": 1.6168,
      "step": 4650
    },
    {
      "epoch": 0.2068382908719043,
      "grad_norm": 0.10757365077733994,
      "learning_rate": 0.0009958477792007723,
      "loss": 1.6188,
      "step": 4652
    },
    {
      "epoch": 0.20692721533057667,
      "grad_norm": 0.10664011538028717,
      "learning_rate": 0.0009958432363822152,
      "loss": 1.6173,
      "step": 4654
    },
    {
      "epoch": 0.20701613978924904,
      "grad_norm": 0.11007332801818848,
      "learning_rate": 0.0009958386910903097,
      "loss": 1.6143,
      "step": 4656
    },
    {
      "epoch": 0.20710506424792138,
      "grad_norm": 0.11352553218603134,
      "learning_rate": 0.0009958341433250785,
      "loss": 1.6141,
      "step": 4658
    },
    {
      "epoch": 0.20719398870659375,
      "grad_norm": 0.10892893373966217,
      "learning_rate": 0.0009958295930865442,
      "loss": 1.6205,
      "step": 4660
    },
    {
      "epoch": 0.20728291316526612,
      "grad_norm": 0.10752708464860916,
      "learning_rate": 0.0009958250403747297,
      "loss": 1.6129,
      "step": 4662
    },
    {
      "epoch": 0.20737183762393846,
      "grad_norm": 0.10832353681325912,
      "learning_rate": 0.0009958204851896577,
      "loss": 1.6208,
      "step": 4664
    },
    {
      "epoch": 0.20746076208261083,
      "grad_norm": 0.11221420764923096,
      "learning_rate": 0.0009958159275313505,
      "loss": 1.6194,
      "step": 4666
    },
    {
      "epoch": 0.20754968654128317,
      "grad_norm": 0.10880377888679504,
      "learning_rate": 0.0009958113673998316,
      "loss": 1.6143,
      "step": 4668
    },
    {
      "epoch": 0.20763861099995554,
      "grad_norm": 0.10734687000513077,
      "learning_rate": 0.0009958068047951228,
      "loss": 1.6223,
      "step": 4670
    },
    {
      "epoch": 0.2077275354586279,
      "grad_norm": 0.11376973241567612,
      "learning_rate": 0.0009958022397172476,
      "loss": 1.6145,
      "step": 4672
    },
    {
      "epoch": 0.20781645991730024,
      "grad_norm": 0.11082089692354202,
      "learning_rate": 0.0009957976721662286,
      "loss": 1.6207,
      "step": 4674
    },
    {
      "epoch": 0.2079053843759726,
      "grad_norm": 0.1084550991654396,
      "learning_rate": 0.0009957931021420884,
      "loss": 1.6231,
      "step": 4676
    },
    {
      "epoch": 0.20799430883464498,
      "grad_norm": 0.11070792376995087,
      "learning_rate": 0.00099578852964485,
      "loss": 1.6124,
      "step": 4678
    },
    {
      "epoch": 0.20808323329331732,
      "grad_norm": 0.1046636626124382,
      "learning_rate": 0.000995783954674536,
      "loss": 1.619,
      "step": 4680
    },
    {
      "epoch": 0.2081721577519897,
      "grad_norm": 0.11071905493736267,
      "learning_rate": 0.0009957793772311695,
      "loss": 1.6193,
      "step": 4682
    },
    {
      "epoch": 0.20826108221066203,
      "grad_norm": 0.1142248883843422,
      "learning_rate": 0.000995774797314773,
      "loss": 1.6173,
      "step": 4684
    },
    {
      "epoch": 0.2083500066693344,
      "grad_norm": 0.10734932124614716,
      "learning_rate": 0.0009957702149253696,
      "loss": 1.618,
      "step": 4686
    },
    {
      "epoch": 0.20843893112800677,
      "grad_norm": 0.1084061935544014,
      "learning_rate": 0.000995765630062982,
      "loss": 1.6178,
      "step": 4688
    },
    {
      "epoch": 0.2085278555866791,
      "grad_norm": 0.11053569614887238,
      "learning_rate": 0.0009957610427276335,
      "loss": 1.6205,
      "step": 4690
    },
    {
      "epoch": 0.20861678004535147,
      "grad_norm": 0.10548976063728333,
      "learning_rate": 0.0009957564529193461,
      "loss": 1.6156,
      "step": 4692
    },
    {
      "epoch": 0.20870570450402384,
      "grad_norm": 0.10936005413532257,
      "learning_rate": 0.0009957518606381434,
      "loss": 1.6242,
      "step": 4694
    },
    {
      "epoch": 0.20879462896269618,
      "grad_norm": 0.11412529647350311,
      "learning_rate": 0.0009957472658840482,
      "loss": 1.6182,
      "step": 4696
    },
    {
      "epoch": 0.20888355342136855,
      "grad_norm": 0.1128493994474411,
      "learning_rate": 0.0009957426686570835,
      "loss": 1.6187,
      "step": 4698
    },
    {
      "epoch": 0.20897247788004092,
      "grad_norm": 0.11843925714492798,
      "learning_rate": 0.0009957380689572719,
      "loss": 1.619,
      "step": 4700
    },
    {
      "epoch": 0.20906140233871326,
      "grad_norm": 0.12056998908519745,
      "learning_rate": 0.0009957334667846365,
      "loss": 1.6178,
      "step": 4702
    },
    {
      "epoch": 0.20915032679738563,
      "grad_norm": 0.1123088002204895,
      "learning_rate": 0.0009957288621392003,
      "loss": 1.6228,
      "step": 4704
    },
    {
      "epoch": 0.20923925125605797,
      "grad_norm": 0.10781093686819077,
      "learning_rate": 0.0009957242550209862,
      "loss": 1.6149,
      "step": 4706
    },
    {
      "epoch": 0.20932817571473034,
      "grad_norm": 0.10738489776849747,
      "learning_rate": 0.0009957196454300173,
      "loss": 1.6187,
      "step": 4708
    },
    {
      "epoch": 0.2094171001734027,
      "grad_norm": 0.11667206138372421,
      "learning_rate": 0.0009957150333663165,
      "loss": 1.6207,
      "step": 4710
    },
    {
      "epoch": 0.20950602463207504,
      "grad_norm": 0.11121194809675217,
      "learning_rate": 0.0009957104188299069,
      "loss": 1.6148,
      "step": 4712
    },
    {
      "epoch": 0.2095949490907474,
      "grad_norm": 0.1158638447523117,
      "learning_rate": 0.0009957058018208113,
      "loss": 1.613,
      "step": 4714
    },
    {
      "epoch": 0.20968387354941978,
      "grad_norm": 0.11311011016368866,
      "learning_rate": 0.000995701182339053,
      "loss": 1.6145,
      "step": 4716
    },
    {
      "epoch": 0.20977279800809212,
      "grad_norm": 0.11132704466581345,
      "learning_rate": 0.000995696560384655,
      "loss": 1.6165,
      "step": 4718
    },
    {
      "epoch": 0.2098617224667645,
      "grad_norm": 0.11866897344589233,
      "learning_rate": 0.00099569193595764,
      "loss": 1.6116,
      "step": 4720
    },
    {
      "epoch": 0.20995064692543683,
      "grad_norm": 0.1159665435552597,
      "learning_rate": 0.0009956873090580316,
      "loss": 1.617,
      "step": 4722
    },
    {
      "epoch": 0.2100395713841092,
      "grad_norm": 0.10780125856399536,
      "learning_rate": 0.0009956826796858524,
      "loss": 1.6134,
      "step": 4724
    },
    {
      "epoch": 0.21012849584278157,
      "grad_norm": 0.10876625776290894,
      "learning_rate": 0.000995678047841126,
      "loss": 1.6187,
      "step": 4726
    },
    {
      "epoch": 0.2102174203014539,
      "grad_norm": 0.1066119521856308,
      "learning_rate": 0.0009956734135238752,
      "loss": 1.6158,
      "step": 4728
    },
    {
      "epoch": 0.21030634476012627,
      "grad_norm": 0.10660472512245178,
      "learning_rate": 0.0009956687767341229,
      "loss": 1.61,
      "step": 4730
    },
    {
      "epoch": 0.21039526921879864,
      "grad_norm": 0.10808862745761871,
      "learning_rate": 0.0009956641374718924,
      "loss": 1.6131,
      "step": 4732
    },
    {
      "epoch": 0.21048419367747098,
      "grad_norm": 0.10631510615348816,
      "learning_rate": 0.0009956594957372072,
      "loss": 1.6142,
      "step": 4734
    },
    {
      "epoch": 0.21057311813614335,
      "grad_norm": 0.10543810576200485,
      "learning_rate": 0.00099565485153009,
      "loss": 1.6175,
      "step": 4736
    },
    {
      "epoch": 0.2106620425948157,
      "grad_norm": 0.11249963939189911,
      "learning_rate": 0.0009956502048505643,
      "loss": 1.6134,
      "step": 4738
    },
    {
      "epoch": 0.21075096705348806,
      "grad_norm": 0.10649897158145905,
      "learning_rate": 0.0009956455556986528,
      "loss": 1.6176,
      "step": 4740
    },
    {
      "epoch": 0.21083989151216043,
      "grad_norm": 0.11082588136196136,
      "learning_rate": 0.0009956409040743793,
      "loss": 1.6141,
      "step": 4742
    },
    {
      "epoch": 0.21092881597083277,
      "grad_norm": 0.11297667771577835,
      "learning_rate": 0.0009956362499777664,
      "loss": 1.6122,
      "step": 4744
    },
    {
      "epoch": 0.21101774042950514,
      "grad_norm": 0.10962715744972229,
      "learning_rate": 0.0009956315934088378,
      "loss": 1.6147,
      "step": 4746
    },
    {
      "epoch": 0.2111066648881775,
      "grad_norm": 0.10940365493297577,
      "learning_rate": 0.0009956269343676166,
      "loss": 1.617,
      "step": 4748
    },
    {
      "epoch": 0.21119558934684984,
      "grad_norm": 0.10843819379806519,
      "learning_rate": 0.0009956222728541258,
      "loss": 1.6163,
      "step": 4750
    },
    {
      "epoch": 0.2112845138055222,
      "grad_norm": 0.10900399088859558,
      "learning_rate": 0.000995617608868389,
      "loss": 1.6211,
      "step": 4752
    },
    {
      "epoch": 0.21137343826419458,
      "grad_norm": 0.11029382050037384,
      "learning_rate": 0.0009956129424104293,
      "loss": 1.6172,
      "step": 4754
    },
    {
      "epoch": 0.21146236272286692,
      "grad_norm": 0.11329919844865799,
      "learning_rate": 0.00099560827348027,
      "loss": 1.6157,
      "step": 4756
    },
    {
      "epoch": 0.2115512871815393,
      "grad_norm": 0.11223863065242767,
      "learning_rate": 0.000995603602077934,
      "loss": 1.6195,
      "step": 4758
    },
    {
      "epoch": 0.21164021164021163,
      "grad_norm": 0.11206690222024918,
      "learning_rate": 0.0009955989282034452,
      "loss": 1.6098,
      "step": 4760
    },
    {
      "epoch": 0.211729136098884,
      "grad_norm": 0.10785891860723495,
      "learning_rate": 0.0009955942518568269,
      "loss": 1.6205,
      "step": 4762
    },
    {
      "epoch": 0.21181806055755636,
      "grad_norm": 0.110198475420475,
      "learning_rate": 0.0009955895730381018,
      "loss": 1.6116,
      "step": 4764
    },
    {
      "epoch": 0.2119069850162287,
      "grad_norm": 0.10387492179870605,
      "learning_rate": 0.000995584891747294,
      "loss": 1.6151,
      "step": 4766
    },
    {
      "epoch": 0.21199590947490107,
      "grad_norm": 0.10851506143808365,
      "learning_rate": 0.0009955802079844264,
      "loss": 1.6107,
      "step": 4768
    },
    {
      "epoch": 0.21208483393357344,
      "grad_norm": 0.11151215434074402,
      "learning_rate": 0.0009955755217495223,
      "loss": 1.6143,
      "step": 4770
    },
    {
      "epoch": 0.21217375839224578,
      "grad_norm": 0.10874391347169876,
      "learning_rate": 0.0009955708330426055,
      "loss": 1.6167,
      "step": 4772
    },
    {
      "epoch": 0.21226268285091815,
      "grad_norm": 0.10614147037267685,
      "learning_rate": 0.0009955661418636988,
      "loss": 1.6114,
      "step": 4774
    },
    {
      "epoch": 0.2123516073095905,
      "grad_norm": 0.110053151845932,
      "learning_rate": 0.000995561448212826,
      "loss": 1.6092,
      "step": 4776
    },
    {
      "epoch": 0.21244053176826286,
      "grad_norm": 0.11465108394622803,
      "learning_rate": 0.0009955567520900108,
      "loss": 1.6082,
      "step": 4778
    },
    {
      "epoch": 0.21252945622693523,
      "grad_norm": 0.11055173724889755,
      "learning_rate": 0.000995552053495276,
      "loss": 1.6118,
      "step": 4780
    },
    {
      "epoch": 0.21261838068560757,
      "grad_norm": 0.10470802336931229,
      "learning_rate": 0.0009955473524286454,
      "loss": 1.617,
      "step": 4782
    },
    {
      "epoch": 0.21270730514427993,
      "grad_norm": 0.10800164937973022,
      "learning_rate": 0.0009955426488901422,
      "loss": 1.6144,
      "step": 4784
    },
    {
      "epoch": 0.2127962296029523,
      "grad_norm": 0.10720106959342957,
      "learning_rate": 0.00099553794287979,
      "loss": 1.6125,
      "step": 4786
    },
    {
      "epoch": 0.21288515406162464,
      "grad_norm": 0.1036008968949318,
      "learning_rate": 0.0009955332343976123,
      "loss": 1.6188,
      "step": 4788
    },
    {
      "epoch": 0.212974078520297,
      "grad_norm": 0.1018817201256752,
      "learning_rate": 0.0009955285234436327,
      "loss": 1.6117,
      "step": 4790
    },
    {
      "epoch": 0.21306300297896938,
      "grad_norm": 0.10686477273702621,
      "learning_rate": 0.0009955238100178745,
      "loss": 1.6121,
      "step": 4792
    },
    {
      "epoch": 0.21315192743764172,
      "grad_norm": 0.10223913937807083,
      "learning_rate": 0.0009955190941203613,
      "loss": 1.6085,
      "step": 4794
    },
    {
      "epoch": 0.2132408518963141,
      "grad_norm": 0.10361933708190918,
      "learning_rate": 0.0009955143757511165,
      "loss": 1.6149,
      "step": 4796
    },
    {
      "epoch": 0.21332977635498643,
      "grad_norm": 0.10988039523363113,
      "learning_rate": 0.000995509654910164,
      "loss": 1.6095,
      "step": 4798
    },
    {
      "epoch": 0.2134187008136588,
      "grad_norm": 0.10833392292261124,
      "learning_rate": 0.0009955049315975267,
      "loss": 1.6164,
      "step": 4800
    },
    {
      "epoch": 0.21350762527233116,
      "grad_norm": 0.10468830913305283,
      "learning_rate": 0.0009955002058132287,
      "loss": 1.607,
      "step": 4802
    },
    {
      "epoch": 0.2135965497310035,
      "grad_norm": 0.1071217730641365,
      "learning_rate": 0.0009954954775572937,
      "loss": 1.6122,
      "step": 4804
    },
    {
      "epoch": 0.21368547418967587,
      "grad_norm": 0.10310684144496918,
      "learning_rate": 0.0009954907468297447,
      "loss": 1.6144,
      "step": 4806
    },
    {
      "epoch": 0.21377439864834824,
      "grad_norm": 0.10384196788072586,
      "learning_rate": 0.0009954860136306058,
      "loss": 1.6118,
      "step": 4808
    },
    {
      "epoch": 0.21386332310702058,
      "grad_norm": 0.10493996739387512,
      "learning_rate": 0.0009954812779599002,
      "loss": 1.6141,
      "step": 4810
    },
    {
      "epoch": 0.21395224756569295,
      "grad_norm": 0.10700862854719162,
      "learning_rate": 0.0009954765398176519,
      "loss": 1.6131,
      "step": 4812
    },
    {
      "epoch": 0.2140411720243653,
      "grad_norm": 0.10613391548395157,
      "learning_rate": 0.0009954717992038844,
      "loss": 1.6165,
      "step": 4814
    },
    {
      "epoch": 0.21413009648303766,
      "grad_norm": 0.10231418162584305,
      "learning_rate": 0.0009954670561186211,
      "loss": 1.6095,
      "step": 4816
    },
    {
      "epoch": 0.21421902094171003,
      "grad_norm": 0.10666482895612717,
      "learning_rate": 0.0009954623105618862,
      "loss": 1.6107,
      "step": 4818
    },
    {
      "epoch": 0.21430794540038237,
      "grad_norm": 0.10890256613492966,
      "learning_rate": 0.000995457562533703,
      "loss": 1.6116,
      "step": 4820
    },
    {
      "epoch": 0.21439686985905473,
      "grad_norm": 0.10631954669952393,
      "learning_rate": 0.000995452812034095,
      "loss": 1.6096,
      "step": 4822
    },
    {
      "epoch": 0.2144857943177271,
      "grad_norm": 0.10400256514549255,
      "learning_rate": 0.0009954480590630864,
      "loss": 1.612,
      "step": 4824
    },
    {
      "epoch": 0.21457471877639944,
      "grad_norm": 0.1029767394065857,
      "learning_rate": 0.0009954433036207006,
      "loss": 1.6084,
      "step": 4826
    },
    {
      "epoch": 0.2146636432350718,
      "grad_norm": 0.10465129464864731,
      "learning_rate": 0.000995438545706961,
      "loss": 1.6101,
      "step": 4828
    },
    {
      "epoch": 0.21475256769374415,
      "grad_norm": 0.10955265164375305,
      "learning_rate": 0.000995433785321892,
      "loss": 1.6092,
      "step": 4830
    },
    {
      "epoch": 0.21484149215241652,
      "grad_norm": 0.10269038379192352,
      "learning_rate": 0.000995429022465517,
      "loss": 1.6195,
      "step": 4832
    },
    {
      "epoch": 0.2149304166110889,
      "grad_norm": 0.1033727377653122,
      "learning_rate": 0.0009954242571378598,
      "loss": 1.6076,
      "step": 4834
    },
    {
      "epoch": 0.21501934106976123,
      "grad_norm": 0.10561516880989075,
      "learning_rate": 0.0009954194893389441,
      "loss": 1.6069,
      "step": 4836
    },
    {
      "epoch": 0.2151082655284336,
      "grad_norm": 0.10597854852676392,
      "learning_rate": 0.000995414719068794,
      "loss": 1.6085,
      "step": 4838
    },
    {
      "epoch": 0.21519718998710596,
      "grad_norm": 0.10583419352769852,
      "learning_rate": 0.000995409946327433,
      "loss": 1.6096,
      "step": 4840
    },
    {
      "epoch": 0.2152861144457783,
      "grad_norm": 0.11089032143354416,
      "learning_rate": 0.0009954051711148846,
      "loss": 1.6115,
      "step": 4842
    },
    {
      "epoch": 0.21537503890445067,
      "grad_norm": 0.11158238351345062,
      "learning_rate": 0.0009954003934311733,
      "loss": 1.6138,
      "step": 4844
    },
    {
      "epoch": 0.21546396336312304,
      "grad_norm": 0.10959797352552414,
      "learning_rate": 0.0009953956132763224,
      "loss": 1.6092,
      "step": 4846
    },
    {
      "epoch": 0.21555288782179538,
      "grad_norm": 0.11067400872707367,
      "learning_rate": 0.000995390830650356,
      "loss": 1.6079,
      "step": 4848
    },
    {
      "epoch": 0.21564181228046775,
      "grad_norm": 0.10605499148368835,
      "learning_rate": 0.0009953860455532981,
      "loss": 1.6174,
      "step": 4850
    },
    {
      "epoch": 0.2157307367391401,
      "grad_norm": 0.10270148515701294,
      "learning_rate": 0.0009953812579851721,
      "loss": 1.6129,
      "step": 4852
    },
    {
      "epoch": 0.21581966119781246,
      "grad_norm": 0.10574102401733398,
      "learning_rate": 0.0009953764679460025,
      "loss": 1.6113,
      "step": 4854
    },
    {
      "epoch": 0.21590858565648482,
      "grad_norm": 0.10553975403308868,
      "learning_rate": 0.0009953716754358126,
      "loss": 1.6095,
      "step": 4856
    },
    {
      "epoch": 0.21599751011515717,
      "grad_norm": 0.10573318600654602,
      "learning_rate": 0.0009953668804546264,
      "loss": 1.6124,
      "step": 4858
    },
    {
      "epoch": 0.21608643457382953,
      "grad_norm": 0.10750672966241837,
      "learning_rate": 0.0009953620830024682,
      "loss": 1.6107,
      "step": 4860
    },
    {
      "epoch": 0.2161753590325019,
      "grad_norm": 0.10438815504312515,
      "learning_rate": 0.0009953572830793617,
      "loss": 1.6129,
      "step": 4862
    },
    {
      "epoch": 0.21626428349117424,
      "grad_norm": 0.10748635977506638,
      "learning_rate": 0.0009953524806853309,
      "loss": 1.6101,
      "step": 4864
    },
    {
      "epoch": 0.2163532079498466,
      "grad_norm": 0.1096944510936737,
      "learning_rate": 0.0009953476758203996,
      "loss": 1.6175,
      "step": 4866
    },
    {
      "epoch": 0.21644213240851895,
      "grad_norm": 0.10906648635864258,
      "learning_rate": 0.000995342868484592,
      "loss": 1.6118,
      "step": 4868
    },
    {
      "epoch": 0.21653105686719132,
      "grad_norm": 0.10521861910820007,
      "learning_rate": 0.0009953380586779318,
      "loss": 1.6142,
      "step": 4870
    },
    {
      "epoch": 0.2166199813258637,
      "grad_norm": 0.10765370726585388,
      "learning_rate": 0.0009953332464004432,
      "loss": 1.6113,
      "step": 4872
    },
    {
      "epoch": 0.21670890578453603,
      "grad_norm": 0.10835171490907669,
      "learning_rate": 0.00099532843165215,
      "loss": 1.6175,
      "step": 4874
    },
    {
      "epoch": 0.2167978302432084,
      "grad_norm": 0.10787054151296616,
      "learning_rate": 0.0009953236144330767,
      "loss": 1.6158,
      "step": 4876
    },
    {
      "epoch": 0.21688675470188076,
      "grad_norm": 0.1088169738650322,
      "learning_rate": 0.0009953187947432465,
      "loss": 1.606,
      "step": 4878
    },
    {
      "epoch": 0.2169756791605531,
      "grad_norm": 0.10647636651992798,
      "learning_rate": 0.0009953139725826844,
      "loss": 1.6161,
      "step": 4880
    },
    {
      "epoch": 0.21706460361922547,
      "grad_norm": 0.10560780018568039,
      "learning_rate": 0.0009953091479514137,
      "loss": 1.6118,
      "step": 4882
    },
    {
      "epoch": 0.21715352807789784,
      "grad_norm": 0.10568542778491974,
      "learning_rate": 0.0009953043208494588,
      "loss": 1.6149,
      "step": 4884
    },
    {
      "epoch": 0.21724245253657018,
      "grad_norm": 0.10434123873710632,
      "learning_rate": 0.000995299491276844,
      "loss": 1.615,
      "step": 4886
    },
    {
      "epoch": 0.21733137699524255,
      "grad_norm": 0.10796734690666199,
      "learning_rate": 0.0009952946592335927,
      "loss": 1.6103,
      "step": 4888
    },
    {
      "epoch": 0.2174203014539149,
      "grad_norm": 0.11157654225826263,
      "learning_rate": 0.0009952898247197297,
      "loss": 1.6161,
      "step": 4890
    },
    {
      "epoch": 0.21750922591258726,
      "grad_norm": 0.10369045287370682,
      "learning_rate": 0.0009952849877352787,
      "loss": 1.6128,
      "step": 4892
    },
    {
      "epoch": 0.21759815037125962,
      "grad_norm": 0.10328072309494019,
      "learning_rate": 0.0009952801482802642,
      "loss": 1.6128,
      "step": 4894
    },
    {
      "epoch": 0.21768707482993196,
      "grad_norm": 0.10743004828691483,
      "learning_rate": 0.00099527530635471,
      "loss": 1.608,
      "step": 4896
    },
    {
      "epoch": 0.21777599928860433,
      "grad_norm": 0.11129618436098099,
      "learning_rate": 0.00099527046195864,
      "loss": 1.6098,
      "step": 4898
    },
    {
      "epoch": 0.2178649237472767,
      "grad_norm": 0.10475630313158035,
      "learning_rate": 0.0009952656150920792,
      "loss": 1.6102,
      "step": 4900
    },
    {
      "epoch": 0.21795384820594904,
      "grad_norm": 0.10529398918151855,
      "learning_rate": 0.0009952607657550511,
      "loss": 1.6052,
      "step": 4902
    },
    {
      "epoch": 0.2180427726646214,
      "grad_norm": 0.10349459946155548,
      "learning_rate": 0.0009952559139475802,
      "loss": 1.6098,
      "step": 4904
    },
    {
      "epoch": 0.21813169712329375,
      "grad_norm": 0.10338271409273148,
      "learning_rate": 0.0009952510596696904,
      "loss": 1.6127,
      "step": 4906
    },
    {
      "epoch": 0.21822062158196612,
      "grad_norm": 0.11201298981904984,
      "learning_rate": 0.0009952462029214064,
      "loss": 1.6127,
      "step": 4908
    },
    {
      "epoch": 0.21830954604063849,
      "grad_norm": 0.10434871912002563,
      "learning_rate": 0.0009952413437027518,
      "loss": 1.6086,
      "step": 4910
    },
    {
      "epoch": 0.21839847049931083,
      "grad_norm": 0.10629836469888687,
      "learning_rate": 0.0009952364820137514,
      "loss": 1.6101,
      "step": 4912
    },
    {
      "epoch": 0.2184873949579832,
      "grad_norm": 0.09886250644922256,
      "learning_rate": 0.0009952316178544291,
      "loss": 1.6038,
      "step": 4914
    },
    {
      "epoch": 0.21857631941665556,
      "grad_norm": 0.09968285262584686,
      "learning_rate": 0.0009952267512248095,
      "loss": 1.6106,
      "step": 4916
    },
    {
      "epoch": 0.2186652438753279,
      "grad_norm": 0.1019214391708374,
      "learning_rate": 0.0009952218821249164,
      "loss": 1.6043,
      "step": 4918
    },
    {
      "epoch": 0.21875416833400027,
      "grad_norm": 0.10381198674440384,
      "learning_rate": 0.0009952170105547743,
      "loss": 1.6032,
      "step": 4920
    },
    {
      "epoch": 0.2188430927926726,
      "grad_norm": 0.10088684409856796,
      "learning_rate": 0.0009952121365144077,
      "loss": 1.6129,
      "step": 4922
    },
    {
      "epoch": 0.21893201725134498,
      "grad_norm": 0.09936795383691788,
      "learning_rate": 0.0009952072600038408,
      "loss": 1.6079,
      "step": 4924
    },
    {
      "epoch": 0.21902094171001735,
      "grad_norm": 0.10552465170621872,
      "learning_rate": 0.000995202381023098,
      "loss": 1.6062,
      "step": 4926
    },
    {
      "epoch": 0.2191098661686897,
      "grad_norm": 0.09979760646820068,
      "learning_rate": 0.0009951974995722032,
      "loss": 1.6061,
      "step": 4928
    },
    {
      "epoch": 0.21919879062736206,
      "grad_norm": 0.10540541261434555,
      "learning_rate": 0.0009951926156511815,
      "loss": 1.6019,
      "step": 4930
    },
    {
      "epoch": 0.21928771508603442,
      "grad_norm": 0.10734565556049347,
      "learning_rate": 0.0009951877292600564,
      "loss": 1.6059,
      "step": 4932
    },
    {
      "epoch": 0.21937663954470676,
      "grad_norm": 0.10099034011363983,
      "learning_rate": 0.000995182840398853,
      "loss": 1.6102,
      "step": 4934
    },
    {
      "epoch": 0.21946556400337913,
      "grad_norm": 0.10821837186813354,
      "learning_rate": 0.0009951779490675952,
      "loss": 1.6089,
      "step": 4936
    },
    {
      "epoch": 0.2195544884620515,
      "grad_norm": 0.10176271945238113,
      "learning_rate": 0.0009951730552663078,
      "loss": 1.61,
      "step": 4938
    },
    {
      "epoch": 0.21964341292072384,
      "grad_norm": 0.10236615687608719,
      "learning_rate": 0.0009951681589950147,
      "loss": 1.6074,
      "step": 4940
    },
    {
      "epoch": 0.2197323373793962,
      "grad_norm": 0.10596957057714462,
      "learning_rate": 0.000995163260253741,
      "loss": 1.6151,
      "step": 4942
    },
    {
      "epoch": 0.21982126183806855,
      "grad_norm": 0.10543814301490784,
      "learning_rate": 0.0009951583590425105,
      "loss": 1.6109,
      "step": 4944
    },
    {
      "epoch": 0.21991018629674092,
      "grad_norm": 0.10567460209131241,
      "learning_rate": 0.000995153455361348,
      "loss": 1.6031,
      "step": 4946
    },
    {
      "epoch": 0.21999911075541329,
      "grad_norm": 0.1093137338757515,
      "learning_rate": 0.000995148549210278,
      "loss": 1.6047,
      "step": 4948
    },
    {
      "epoch": 0.22008803521408563,
      "grad_norm": 0.11018449813127518,
      "learning_rate": 0.0009951436405893246,
      "loss": 1.6101,
      "step": 4950
    },
    {
      "epoch": 0.220176959672758,
      "grad_norm": 0.10580731928348541,
      "learning_rate": 0.0009951387294985126,
      "loss": 1.6132,
      "step": 4952
    },
    {
      "epoch": 0.22026588413143036,
      "grad_norm": 0.10643559694290161,
      "learning_rate": 0.0009951338159378664,
      "loss": 1.6078,
      "step": 4954
    },
    {
      "epoch": 0.2203548085901027,
      "grad_norm": 0.10614238679409027,
      "learning_rate": 0.0009951288999074106,
      "loss": 1.6054,
      "step": 4956
    },
    {
      "epoch": 0.22044373304877507,
      "grad_norm": 0.10441026836633682,
      "learning_rate": 0.0009951239814071698,
      "loss": 1.6083,
      "step": 4958
    },
    {
      "epoch": 0.2205326575074474,
      "grad_norm": 0.10442102700471878,
      "learning_rate": 0.0009951190604371683,
      "loss": 1.6069,
      "step": 4960
    },
    {
      "epoch": 0.22062158196611978,
      "grad_norm": 0.1024702936410904,
      "learning_rate": 0.0009951141369974306,
      "loss": 1.6092,
      "step": 4962
    },
    {
      "epoch": 0.22071050642479215,
      "grad_norm": 0.09814217686653137,
      "learning_rate": 0.0009951092110879816,
      "loss": 1.6095,
      "step": 4964
    },
    {
      "epoch": 0.2207994308834645,
      "grad_norm": 0.09846185147762299,
      "learning_rate": 0.0009951042827088454,
      "loss": 1.6036,
      "step": 4966
    },
    {
      "epoch": 0.22088835534213686,
      "grad_norm": 0.10347746312618256,
      "learning_rate": 0.0009950993518600471,
      "loss": 1.6062,
      "step": 4968
    },
    {
      "epoch": 0.22097727980080922,
      "grad_norm": 0.101874940097332,
      "learning_rate": 0.000995094418541611,
      "loss": 1.6067,
      "step": 4970
    },
    {
      "epoch": 0.22106620425948156,
      "grad_norm": 0.1038205623626709,
      "learning_rate": 0.0009950894827535617,
      "loss": 1.6083,
      "step": 4972
    },
    {
      "epoch": 0.22115512871815393,
      "grad_norm": 0.10341985523700714,
      "learning_rate": 0.000995084544495924,
      "loss": 1.6091,
      "step": 4974
    },
    {
      "epoch": 0.2212440531768263,
      "grad_norm": 0.10060903429985046,
      "learning_rate": 0.0009950796037687224,
      "loss": 1.6075,
      "step": 4976
    },
    {
      "epoch": 0.22133297763549864,
      "grad_norm": 0.10430523008108139,
      "learning_rate": 0.0009950746605719813,
      "loss": 1.6024,
      "step": 4978
    },
    {
      "epoch": 0.221421902094171,
      "grad_norm": 0.11263576149940491,
      "learning_rate": 0.0009950697149057259,
      "loss": 1.6101,
      "step": 4980
    },
    {
      "epoch": 0.22151082655284335,
      "grad_norm": 0.10367169231176376,
      "learning_rate": 0.0009950647667699806,
      "loss": 1.6015,
      "step": 4982
    },
    {
      "epoch": 0.22159975101151572,
      "grad_norm": 0.10355625301599503,
      "learning_rate": 0.0009950598161647698,
      "loss": 1.6094,
      "step": 4984
    },
    {
      "epoch": 0.22168867547018808,
      "grad_norm": 0.09836798161268234,
      "learning_rate": 0.0009950548630901187,
      "loss": 1.606,
      "step": 4986
    },
    {
      "epoch": 0.22177759992886042,
      "grad_norm": 0.10221516340970993,
      "learning_rate": 0.0009950499075460518,
      "loss": 1.6084,
      "step": 4988
    },
    {
      "epoch": 0.2218665243875328,
      "grad_norm": 0.10238867253065109,
      "learning_rate": 0.0009950449495325934,
      "loss": 1.6016,
      "step": 4990
    },
    {
      "epoch": 0.22195544884620516,
      "grad_norm": 0.1028597429394722,
      "learning_rate": 0.000995039989049769,
      "loss": 1.6079,
      "step": 4992
    },
    {
      "epoch": 0.2220443733048775,
      "grad_norm": 0.10329331457614899,
      "learning_rate": 0.0009950350260976028,
      "loss": 1.6096,
      "step": 4994
    },
    {
      "epoch": 0.22213329776354987,
      "grad_norm": 0.10824605077505112,
      "learning_rate": 0.0009950300606761198,
      "loss": 1.6103,
      "step": 4996
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.10527484118938446,
      "learning_rate": 0.0009950250927853448,
      "loss": 1.6112,
      "step": 4998
    },
    {
      "epoch": 0.22231114668089458,
      "grad_norm": 0.10636089742183685,
      "learning_rate": 0.0009950201224253022,
      "loss": 1.6092,
      "step": 5000
    },
    {
      "epoch": 0.22231114668089458,
      "eval_loss": 1.5817906856536865,
      "eval_runtime": 12.3795,
      "eval_samples_per_second": 558.179,
      "eval_steps_per_second": 69.793,
      "step": 5000
    },
    {
      "epoch": 0.22240007113956695,
      "grad_norm": 0.09906064718961716,
      "learning_rate": 0.0009950151495960171,
      "loss": 1.6069,
      "step": 5002
    },
    {
      "epoch": 0.2224889955982393,
      "grad_norm": 0.10404371470212936,
      "learning_rate": 0.0009950101742975143,
      "loss": 1.6045,
      "step": 5004
    },
    {
      "epoch": 0.22257792005691165,
      "grad_norm": 0.1025133803486824,
      "learning_rate": 0.0009950051965298189,
      "loss": 1.607,
      "step": 5006
    },
    {
      "epoch": 0.22266684451558402,
      "grad_norm": 0.10632102936506271,
      "learning_rate": 0.000995000216292955,
      "loss": 1.603,
      "step": 5008
    },
    {
      "epoch": 0.22275576897425636,
      "grad_norm": 0.10821643471717834,
      "learning_rate": 0.000994995233586948,
      "loss": 1.6062,
      "step": 5010
    },
    {
      "epoch": 0.22284469343292873,
      "grad_norm": 0.10510935634374619,
      "learning_rate": 0.0009949902484118226,
      "loss": 1.6077,
      "step": 5012
    },
    {
      "epoch": 0.22293361789160107,
      "grad_norm": 0.1051439642906189,
      "learning_rate": 0.0009949852607676038,
      "loss": 1.6069,
      "step": 5014
    },
    {
      "epoch": 0.22302254235027344,
      "grad_norm": 0.10179907083511353,
      "learning_rate": 0.0009949802706543161,
      "loss": 1.6097,
      "step": 5016
    },
    {
      "epoch": 0.2231114668089458,
      "grad_norm": 0.10336954891681671,
      "learning_rate": 0.0009949752780719849,
      "loss": 1.6074,
      "step": 5018
    },
    {
      "epoch": 0.22320039126761815,
      "grad_norm": 0.10205300152301788,
      "learning_rate": 0.0009949702830206347,
      "loss": 1.6064,
      "step": 5020
    },
    {
      "epoch": 0.22328931572629052,
      "grad_norm": 0.10609652101993561,
      "learning_rate": 0.0009949652855002907,
      "loss": 1.6048,
      "step": 5022
    },
    {
      "epoch": 0.22337824018496288,
      "grad_norm": 0.10084786266088486,
      "learning_rate": 0.0009949602855109775,
      "loss": 1.6025,
      "step": 5024
    },
    {
      "epoch": 0.22346716464363522,
      "grad_norm": 0.10056611895561218,
      "learning_rate": 0.0009949552830527202,
      "loss": 1.6073,
      "step": 5026
    },
    {
      "epoch": 0.2235560891023076,
      "grad_norm": 0.09772695600986481,
      "learning_rate": 0.0009949502781255442,
      "loss": 1.6077,
      "step": 5028
    },
    {
      "epoch": 0.22364501356097996,
      "grad_norm": 0.10336261987686157,
      "learning_rate": 0.0009949452707294737,
      "loss": 1.6084,
      "step": 5030
    },
    {
      "epoch": 0.2237339380196523,
      "grad_norm": 0.1037675067782402,
      "learning_rate": 0.0009949402608645341,
      "loss": 1.6126,
      "step": 5032
    },
    {
      "epoch": 0.22382286247832467,
      "grad_norm": 0.10318697243928909,
      "learning_rate": 0.0009949352485307504,
      "loss": 1.6088,
      "step": 5034
    },
    {
      "epoch": 0.223911786936997,
      "grad_norm": 0.10445655882358551,
      "learning_rate": 0.0009949302337281473,
      "loss": 1.6139,
      "step": 5036
    },
    {
      "epoch": 0.22400071139566938,
      "grad_norm": 0.10120239853858948,
      "learning_rate": 0.0009949252164567503,
      "loss": 1.6073,
      "step": 5038
    },
    {
      "epoch": 0.22408963585434175,
      "grad_norm": 0.1085609570145607,
      "learning_rate": 0.000994920196716584,
      "loss": 1.6062,
      "step": 5040
    },
    {
      "epoch": 0.22417856031301409,
      "grad_norm": 0.10521785914897919,
      "learning_rate": 0.0009949151745076736,
      "loss": 1.6031,
      "step": 5042
    },
    {
      "epoch": 0.22426748477168645,
      "grad_norm": 0.10063958913087845,
      "learning_rate": 0.0009949101498300443,
      "loss": 1.6032,
      "step": 5044
    },
    {
      "epoch": 0.22435640923035882,
      "grad_norm": 0.09775926172733307,
      "learning_rate": 0.000994905122683721,
      "loss": 1.6077,
      "step": 5046
    },
    {
      "epoch": 0.22444533368903116,
      "grad_norm": 0.09834594279527664,
      "learning_rate": 0.0009949000930687289,
      "loss": 1.5963,
      "step": 5048
    },
    {
      "epoch": 0.22453425814770353,
      "grad_norm": 0.10165659338235855,
      "learning_rate": 0.0009948950609850928,
      "loss": 1.6073,
      "step": 5050
    },
    {
      "epoch": 0.22462318260637587,
      "grad_norm": 0.09759823232889175,
      "learning_rate": 0.000994890026432838,
      "loss": 1.602,
      "step": 5052
    },
    {
      "epoch": 0.22471210706504824,
      "grad_norm": 0.10388672351837158,
      "learning_rate": 0.0009948849894119895,
      "loss": 1.6052,
      "step": 5054
    },
    {
      "epoch": 0.2248010315237206,
      "grad_norm": 0.10322355479001999,
      "learning_rate": 0.0009948799499225726,
      "loss": 1.6042,
      "step": 5056
    },
    {
      "epoch": 0.22488995598239295,
      "grad_norm": 0.1011364683508873,
      "learning_rate": 0.0009948749079646125,
      "loss": 1.6039,
      "step": 5058
    },
    {
      "epoch": 0.22497888044106532,
      "grad_norm": 0.10648488998413086,
      "learning_rate": 0.0009948698635381342,
      "loss": 1.607,
      "step": 5060
    },
    {
      "epoch": 0.22506780489973768,
      "grad_norm": 0.10115467756986618,
      "learning_rate": 0.0009948648166431629,
      "loss": 1.6001,
      "step": 5062
    },
    {
      "epoch": 0.22515672935841002,
      "grad_norm": 0.1064210832118988,
      "learning_rate": 0.0009948597672797235,
      "loss": 1.6048,
      "step": 5064
    },
    {
      "epoch": 0.2252456538170824,
      "grad_norm": 0.10593225806951523,
      "learning_rate": 0.0009948547154478417,
      "loss": 1.6016,
      "step": 5066
    },
    {
      "epoch": 0.22533457827575476,
      "grad_norm": 0.10426605492830276,
      "learning_rate": 0.0009948496611475422,
      "loss": 1.5992,
      "step": 5068
    },
    {
      "epoch": 0.2254235027344271,
      "grad_norm": 0.10383313149213791,
      "learning_rate": 0.0009948446043788505,
      "loss": 1.6085,
      "step": 5070
    },
    {
      "epoch": 0.22551242719309947,
      "grad_norm": 0.10504084825515747,
      "learning_rate": 0.000994839545141792,
      "loss": 1.6052,
      "step": 5072
    },
    {
      "epoch": 0.2256013516517718,
      "grad_norm": 0.09993825107812881,
      "learning_rate": 0.0009948344834363916,
      "loss": 1.6049,
      "step": 5074
    },
    {
      "epoch": 0.22569027611044418,
      "grad_norm": 0.10596901178359985,
      "learning_rate": 0.0009948294192626745,
      "loss": 1.6018,
      "step": 5076
    },
    {
      "epoch": 0.22577920056911654,
      "grad_norm": 0.10218656808137894,
      "learning_rate": 0.0009948243526206664,
      "loss": 1.6093,
      "step": 5078
    },
    {
      "epoch": 0.22586812502778889,
      "grad_norm": 0.09963101893663406,
      "learning_rate": 0.0009948192835103921,
      "loss": 1.608,
      "step": 5080
    },
    {
      "epoch": 0.22595704948646125,
      "grad_norm": 0.09806007146835327,
      "learning_rate": 0.0009948142119318771,
      "loss": 1.6045,
      "step": 5082
    },
    {
      "epoch": 0.22604597394513362,
      "grad_norm": 0.10028783231973648,
      "learning_rate": 0.0009948091378851468,
      "loss": 1.6131,
      "step": 5084
    },
    {
      "epoch": 0.22613489840380596,
      "grad_norm": 0.10069727897644043,
      "learning_rate": 0.0009948040613702263,
      "loss": 1.6067,
      "step": 5086
    },
    {
      "epoch": 0.22622382286247833,
      "grad_norm": 0.1011575385928154,
      "learning_rate": 0.000994798982387141,
      "loss": 1.6084,
      "step": 5088
    },
    {
      "epoch": 0.22631274732115067,
      "grad_norm": 0.1032286211848259,
      "learning_rate": 0.0009947939009359163,
      "loss": 1.6047,
      "step": 5090
    },
    {
      "epoch": 0.22640167177982304,
      "grad_norm": 0.1022925153374672,
      "learning_rate": 0.0009947888170165774,
      "loss": 1.6037,
      "step": 5092
    },
    {
      "epoch": 0.2264905962384954,
      "grad_norm": 0.11113549023866653,
      "learning_rate": 0.0009947837306291496,
      "loss": 1.6072,
      "step": 5094
    },
    {
      "epoch": 0.22657952069716775,
      "grad_norm": 0.10652270913124084,
      "learning_rate": 0.0009947786417736587,
      "loss": 1.6041,
      "step": 5096
    },
    {
      "epoch": 0.22666844515584011,
      "grad_norm": 0.10547838360071182,
      "learning_rate": 0.0009947735504501297,
      "loss": 1.6035,
      "step": 5098
    },
    {
      "epoch": 0.22675736961451248,
      "grad_norm": 0.10703777521848679,
      "learning_rate": 0.000994768456658588,
      "loss": 1.6005,
      "step": 5100
    },
    {
      "epoch": 0.22684629407318482,
      "grad_norm": 0.10610830783843994,
      "learning_rate": 0.0009947633603990592,
      "loss": 1.6063,
      "step": 5102
    },
    {
      "epoch": 0.2269352185318572,
      "grad_norm": 0.10127437859773636,
      "learning_rate": 0.0009947582616715686,
      "loss": 1.605,
      "step": 5104
    },
    {
      "epoch": 0.22702414299052953,
      "grad_norm": 0.10472498089075089,
      "learning_rate": 0.0009947531604761416,
      "loss": 1.6021,
      "step": 5106
    },
    {
      "epoch": 0.2271130674492019,
      "grad_norm": 0.11169638484716415,
      "learning_rate": 0.000994748056812804,
      "loss": 1.5958,
      "step": 5108
    },
    {
      "epoch": 0.22720199190787427,
      "grad_norm": 0.10129303485155106,
      "learning_rate": 0.0009947429506815808,
      "loss": 1.5976,
      "step": 5110
    },
    {
      "epoch": 0.2272909163665466,
      "grad_norm": 0.10340652614831924,
      "learning_rate": 0.0009947378420824976,
      "loss": 1.6075,
      "step": 5112
    },
    {
      "epoch": 0.22737984082521898,
      "grad_norm": 0.0987604483962059,
      "learning_rate": 0.00099473273101558,
      "loss": 1.599,
      "step": 5114
    },
    {
      "epoch": 0.22746876528389134,
      "grad_norm": 0.10262270271778107,
      "learning_rate": 0.0009947276174808533,
      "loss": 1.602,
      "step": 5116
    },
    {
      "epoch": 0.22755768974256368,
      "grad_norm": 0.10149616748094559,
      "learning_rate": 0.0009947225014783432,
      "loss": 1.6048,
      "step": 5118
    },
    {
      "epoch": 0.22764661420123605,
      "grad_norm": 0.0976349264383316,
      "learning_rate": 0.0009947173830080751,
      "loss": 1.6016,
      "step": 5120
    },
    {
      "epoch": 0.22773553865990842,
      "grad_norm": 0.0950886532664299,
      "learning_rate": 0.0009947122620700747,
      "loss": 1.6017,
      "step": 5122
    },
    {
      "epoch": 0.22782446311858076,
      "grad_norm": 0.10552572458982468,
      "learning_rate": 0.0009947071386643672,
      "loss": 1.609,
      "step": 5124
    },
    {
      "epoch": 0.22791338757725313,
      "grad_norm": 0.10495138168334961,
      "learning_rate": 0.0009947020127909784,
      "loss": 1.6071,
      "step": 5126
    },
    {
      "epoch": 0.22800231203592547,
      "grad_norm": 0.10625366866588593,
      "learning_rate": 0.000994696884449934,
      "loss": 1.6039,
      "step": 5128
    },
    {
      "epoch": 0.22809123649459784,
      "grad_norm": 0.10566636174917221,
      "learning_rate": 0.0009946917536412594,
      "loss": 1.6031,
      "step": 5130
    },
    {
      "epoch": 0.2281801609532702,
      "grad_norm": 0.1036016196012497,
      "learning_rate": 0.0009946866203649804,
      "loss": 1.6067,
      "step": 5132
    },
    {
      "epoch": 0.22826908541194255,
      "grad_norm": 0.0977390855550766,
      "learning_rate": 0.0009946814846211222,
      "loss": 1.6106,
      "step": 5134
    },
    {
      "epoch": 0.22835800987061491,
      "grad_norm": 0.1010410338640213,
      "learning_rate": 0.0009946763464097105,
      "loss": 1.5996,
      "step": 5136
    },
    {
      "epoch": 0.22844693432928728,
      "grad_norm": 0.10956060141324997,
      "learning_rate": 0.000994671205730771,
      "loss": 1.602,
      "step": 5138
    },
    {
      "epoch": 0.22853585878795962,
      "grad_norm": 0.10265465825796127,
      "learning_rate": 0.0009946660625843298,
      "loss": 1.6062,
      "step": 5140
    },
    {
      "epoch": 0.228624783246632,
      "grad_norm": 0.10135070979595184,
      "learning_rate": 0.000994660916970412,
      "loss": 1.6026,
      "step": 5142
    },
    {
      "epoch": 0.22871370770530433,
      "grad_norm": 0.09793238341808319,
      "learning_rate": 0.0009946557688890435,
      "loss": 1.6002,
      "step": 5144
    },
    {
      "epoch": 0.2288026321639767,
      "grad_norm": 0.10020411014556885,
      "learning_rate": 0.0009946506183402498,
      "loss": 1.6111,
      "step": 5146
    },
    {
      "epoch": 0.22889155662264907,
      "grad_norm": 0.0976269543170929,
      "learning_rate": 0.0009946454653240568,
      "loss": 1.6027,
      "step": 5148
    },
    {
      "epoch": 0.2289804810813214,
      "grad_norm": 0.09884080290794373,
      "learning_rate": 0.00099464030984049,
      "loss": 1.6044,
      "step": 5150
    },
    {
      "epoch": 0.22906940553999378,
      "grad_norm": 0.10288077592849731,
      "learning_rate": 0.0009946351518895754,
      "loss": 1.6055,
      "step": 5152
    },
    {
      "epoch": 0.22915832999866614,
      "grad_norm": 0.10319656878709793,
      "learning_rate": 0.0009946299914713385,
      "loss": 1.6073,
      "step": 5154
    },
    {
      "epoch": 0.22924725445733848,
      "grad_norm": 0.10332750529050827,
      "learning_rate": 0.000994624828585805,
      "loss": 1.606,
      "step": 5156
    },
    {
      "epoch": 0.22933617891601085,
      "grad_norm": 0.10401324927806854,
      "learning_rate": 0.0009946196632330008,
      "loss": 1.6012,
      "step": 5158
    },
    {
      "epoch": 0.22942510337468322,
      "grad_norm": 0.10355940461158752,
      "learning_rate": 0.0009946144954129518,
      "loss": 1.6054,
      "step": 5160
    },
    {
      "epoch": 0.22951402783335556,
      "grad_norm": 0.10316723585128784,
      "learning_rate": 0.0009946093251256834,
      "loss": 1.609,
      "step": 5162
    },
    {
      "epoch": 0.22960295229202793,
      "grad_norm": 0.1029047816991806,
      "learning_rate": 0.0009946041523712214,
      "loss": 1.6074,
      "step": 5164
    },
    {
      "epoch": 0.22969187675070027,
      "grad_norm": 0.1026320829987526,
      "learning_rate": 0.0009945989771495921,
      "loss": 1.6014,
      "step": 5166
    },
    {
      "epoch": 0.22978080120937264,
      "grad_norm": 0.10016648471355438,
      "learning_rate": 0.0009945937994608209,
      "loss": 1.6034,
      "step": 5168
    },
    {
      "epoch": 0.229869725668045,
      "grad_norm": 0.10129445791244507,
      "learning_rate": 0.0009945886193049338,
      "loss": 1.6015,
      "step": 5170
    },
    {
      "epoch": 0.22995865012671735,
      "grad_norm": 0.10496019572019577,
      "learning_rate": 0.0009945834366819565,
      "loss": 1.6063,
      "step": 5172
    },
    {
      "epoch": 0.2300475745853897,
      "grad_norm": 0.10157696902751923,
      "learning_rate": 0.000994578251591915,
      "loss": 1.6045,
      "step": 5174
    },
    {
      "epoch": 0.23013649904406208,
      "grad_norm": 0.10173121839761734,
      "learning_rate": 0.0009945730640348348,
      "loss": 1.5999,
      "step": 5176
    },
    {
      "epoch": 0.23022542350273442,
      "grad_norm": 0.10079949349164963,
      "learning_rate": 0.0009945678740107423,
      "loss": 1.6074,
      "step": 5178
    },
    {
      "epoch": 0.2303143479614068,
      "grad_norm": 0.1033761128783226,
      "learning_rate": 0.000994562681519663,
      "loss": 1.6054,
      "step": 5180
    },
    {
      "epoch": 0.23040327242007913,
      "grad_norm": 0.1060091182589531,
      "learning_rate": 0.000994557486561623,
      "loss": 1.6007,
      "step": 5182
    },
    {
      "epoch": 0.2304921968787515,
      "grad_norm": 0.0948513001203537,
      "learning_rate": 0.0009945522891366483,
      "loss": 1.6028,
      "step": 5184
    },
    {
      "epoch": 0.23058112133742387,
      "grad_norm": 0.09779325127601624,
      "learning_rate": 0.0009945470892447647,
      "loss": 1.6045,
      "step": 5186
    },
    {
      "epoch": 0.2306700457960962,
      "grad_norm": 0.09960626810789108,
      "learning_rate": 0.000994541886885998,
      "loss": 1.6021,
      "step": 5188
    },
    {
      "epoch": 0.23075897025476858,
      "grad_norm": 0.1015731692314148,
      "learning_rate": 0.0009945366820603746,
      "loss": 1.6027,
      "step": 5190
    },
    {
      "epoch": 0.23084789471344094,
      "grad_norm": 0.10101474076509476,
      "learning_rate": 0.0009945314747679197,
      "loss": 1.6027,
      "step": 5192
    },
    {
      "epoch": 0.23093681917211328,
      "grad_norm": 0.10149352252483368,
      "learning_rate": 0.00099452626500866,
      "loss": 1.5992,
      "step": 5194
    },
    {
      "epoch": 0.23102574363078565,
      "grad_norm": 0.09866882860660553,
      "learning_rate": 0.0009945210527826209,
      "loss": 1.5996,
      "step": 5196
    },
    {
      "epoch": 0.231114668089458,
      "grad_norm": 0.10099692642688751,
      "learning_rate": 0.000994515838089829,
      "loss": 1.5983,
      "step": 5198
    },
    {
      "epoch": 0.23120359254813036,
      "grad_norm": 0.10036486387252808,
      "learning_rate": 0.0009945106209303099,
      "loss": 1.5955,
      "step": 5200
    },
    {
      "epoch": 0.23129251700680273,
      "grad_norm": 0.10122378170490265,
      "learning_rate": 0.0009945054013040897,
      "loss": 1.5993,
      "step": 5202
    },
    {
      "epoch": 0.23138144146547507,
      "grad_norm": 0.09628558903932571,
      "learning_rate": 0.0009945001792111944,
      "loss": 1.6037,
      "step": 5204
    },
    {
      "epoch": 0.23147036592414744,
      "grad_norm": 0.09720970690250397,
      "learning_rate": 0.0009944949546516503,
      "loss": 1.6031,
      "step": 5206
    },
    {
      "epoch": 0.2315592903828198,
      "grad_norm": 0.09955313801765442,
      "learning_rate": 0.0009944897276254832,
      "loss": 1.6018,
      "step": 5208
    },
    {
      "epoch": 0.23164821484149214,
      "grad_norm": 0.09602374583482742,
      "learning_rate": 0.0009944844981327194,
      "loss": 1.5969,
      "step": 5210
    },
    {
      "epoch": 0.2317371393001645,
      "grad_norm": 0.10124479979276657,
      "learning_rate": 0.0009944792661733847,
      "loss": 1.6058,
      "step": 5212
    },
    {
      "epoch": 0.23182606375883688,
      "grad_norm": 0.09888044744729996,
      "learning_rate": 0.0009944740317475054,
      "loss": 1.5966,
      "step": 5214
    },
    {
      "epoch": 0.23191498821750922,
      "grad_norm": 0.10130251944065094,
      "learning_rate": 0.0009944687948551074,
      "loss": 1.6004,
      "step": 5216
    },
    {
      "epoch": 0.2320039126761816,
      "grad_norm": 0.10554568469524384,
      "learning_rate": 0.000994463555496217,
      "loss": 1.6022,
      "step": 5218
    },
    {
      "epoch": 0.23209283713485393,
      "grad_norm": 0.10593781620264053,
      "learning_rate": 0.0009944583136708607,
      "loss": 1.6014,
      "step": 5220
    },
    {
      "epoch": 0.2321817615935263,
      "grad_norm": 0.1010642796754837,
      "learning_rate": 0.0009944530693790639,
      "loss": 1.6043,
      "step": 5222
    },
    {
      "epoch": 0.23227068605219867,
      "grad_norm": 0.0989823043346405,
      "learning_rate": 0.000994447822620853,
      "loss": 1.5977,
      "step": 5224
    },
    {
      "epoch": 0.232359610510871,
      "grad_norm": 0.10016317665576935,
      "learning_rate": 0.0009944425733962546,
      "loss": 1.6013,
      "step": 5226
    },
    {
      "epoch": 0.23244853496954337,
      "grad_norm": 0.10070092231035233,
      "learning_rate": 0.0009944373217052945,
      "loss": 1.6,
      "step": 5228
    },
    {
      "epoch": 0.23253745942821574,
      "grad_norm": 0.10165101289749146,
      "learning_rate": 0.0009944320675479988,
      "loss": 1.6034,
      "step": 5230
    },
    {
      "epoch": 0.23262638388688808,
      "grad_norm": 0.10388650000095367,
      "learning_rate": 0.000994426810924394,
      "loss": 1.5997,
      "step": 5232
    },
    {
      "epoch": 0.23271530834556045,
      "grad_norm": 0.10321874916553497,
      "learning_rate": 0.0009944215518345061,
      "loss": 1.5995,
      "step": 5234
    },
    {
      "epoch": 0.2328042328042328,
      "grad_norm": 0.09810288995504379,
      "learning_rate": 0.0009944162902783616,
      "loss": 1.6004,
      "step": 5236
    },
    {
      "epoch": 0.23289315726290516,
      "grad_norm": 0.09570595622062683,
      "learning_rate": 0.0009944110262559865,
      "loss": 1.6005,
      "step": 5238
    },
    {
      "epoch": 0.23298208172157753,
      "grad_norm": 0.10209929198026657,
      "learning_rate": 0.000994405759767407,
      "loss": 1.6032,
      "step": 5240
    },
    {
      "epoch": 0.23307100618024987,
      "grad_norm": 0.10126584023237228,
      "learning_rate": 0.0009944004908126495,
      "loss": 1.6095,
      "step": 5242
    },
    {
      "epoch": 0.23315993063892224,
      "grad_norm": 0.10170583426952362,
      "learning_rate": 0.0009943952193917406,
      "loss": 1.5949,
      "step": 5244
    },
    {
      "epoch": 0.2332488550975946,
      "grad_norm": 0.09847375005483627,
      "learning_rate": 0.000994389945504706,
      "loss": 1.6007,
      "step": 5246
    },
    {
      "epoch": 0.23333777955626694,
      "grad_norm": 0.09713829308748245,
      "learning_rate": 0.0009943846691515723,
      "loss": 1.6019,
      "step": 5248
    },
    {
      "epoch": 0.2334267040149393,
      "grad_norm": 0.09937822073698044,
      "learning_rate": 0.000994379390332366,
      "loss": 1.5981,
      "step": 5250
    },
    {
      "epoch": 0.23351562847361168,
      "grad_norm": 0.10339406877756119,
      "learning_rate": 0.000994374109047113,
      "loss": 1.6029,
      "step": 5252
    },
    {
      "epoch": 0.23360455293228402,
      "grad_norm": 0.10225261747837067,
      "learning_rate": 0.00099436882529584,
      "loss": 1.5984,
      "step": 5254
    },
    {
      "epoch": 0.2336934773909564,
      "grad_norm": 0.10872292518615723,
      "learning_rate": 0.0009943635390785733,
      "loss": 1.5941,
      "step": 5256
    },
    {
      "epoch": 0.23378240184962873,
      "grad_norm": 0.10657516866922379,
      "learning_rate": 0.000994358250395339,
      "loss": 1.5936,
      "step": 5258
    },
    {
      "epoch": 0.2338713263083011,
      "grad_norm": 0.1047828420996666,
      "learning_rate": 0.0009943529592461638,
      "loss": 1.5992,
      "step": 5260
    },
    {
      "epoch": 0.23396025076697347,
      "grad_norm": 0.1091296523809433,
      "learning_rate": 0.000994347665631074,
      "loss": 1.605,
      "step": 5262
    },
    {
      "epoch": 0.2340491752256458,
      "grad_norm": 0.1082061156630516,
      "learning_rate": 0.000994342369550096,
      "loss": 1.6055,
      "step": 5264
    },
    {
      "epoch": 0.23413809968431817,
      "grad_norm": 0.1011800467967987,
      "learning_rate": 0.0009943370710032563,
      "loss": 1.6019,
      "step": 5266
    },
    {
      "epoch": 0.23422702414299054,
      "grad_norm": 0.10550300031900406,
      "learning_rate": 0.000994331769990581,
      "loss": 1.6043,
      "step": 5268
    },
    {
      "epoch": 0.23431594860166288,
      "grad_norm": 0.10264909267425537,
      "learning_rate": 0.000994326466512097,
      "loss": 1.603,
      "step": 5270
    },
    {
      "epoch": 0.23440487306033525,
      "grad_norm": 0.10784979909658432,
      "learning_rate": 0.0009943211605678303,
      "loss": 1.6024,
      "step": 5272
    },
    {
      "epoch": 0.2344937975190076,
      "grad_norm": 0.10662925988435745,
      "learning_rate": 0.0009943158521578078,
      "loss": 1.5965,
      "step": 5274
    },
    {
      "epoch": 0.23458272197767996,
      "grad_norm": 0.10931091010570526,
      "learning_rate": 0.0009943105412820556,
      "loss": 1.6006,
      "step": 5276
    },
    {
      "epoch": 0.23467164643635233,
      "grad_norm": 0.09809394180774689,
      "learning_rate": 0.0009943052279406007,
      "loss": 1.5995,
      "step": 5278
    },
    {
      "epoch": 0.23476057089502467,
      "grad_norm": 0.10617434978485107,
      "learning_rate": 0.000994299912133469,
      "loss": 1.6007,
      "step": 5280
    },
    {
      "epoch": 0.23484949535369704,
      "grad_norm": 0.10415052622556686,
      "learning_rate": 0.0009942945938606874,
      "loss": 1.6003,
      "step": 5282
    },
    {
      "epoch": 0.2349384198123694,
      "grad_norm": 0.09706338495016098,
      "learning_rate": 0.0009942892731222821,
      "loss": 1.6022,
      "step": 5284
    },
    {
      "epoch": 0.23502734427104174,
      "grad_norm": 0.09842453896999359,
      "learning_rate": 0.00099428394991828,
      "loss": 1.593,
      "step": 5286
    },
    {
      "epoch": 0.2351162687297141,
      "grad_norm": 0.0980658233165741,
      "learning_rate": 0.0009942786242487075,
      "loss": 1.5997,
      "step": 5288
    },
    {
      "epoch": 0.23520519318838645,
      "grad_norm": 0.10015766322612762,
      "learning_rate": 0.0009942732961135913,
      "loss": 1.5994,
      "step": 5290
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.10355110466480255,
      "learning_rate": 0.0009942679655129576,
      "loss": 1.5988,
      "step": 5292
    },
    {
      "epoch": 0.2353830421057312,
      "grad_norm": 0.102814219892025,
      "learning_rate": 0.0009942626324468335,
      "loss": 1.5951,
      "step": 5294
    },
    {
      "epoch": 0.23547196656440353,
      "grad_norm": 0.10506147146224976,
      "learning_rate": 0.0009942572969152453,
      "loss": 1.5975,
      "step": 5296
    },
    {
      "epoch": 0.2355608910230759,
      "grad_norm": 0.10157327353954315,
      "learning_rate": 0.0009942519589182196,
      "loss": 1.6015,
      "step": 5298
    },
    {
      "epoch": 0.23564981548174826,
      "grad_norm": 0.10472837835550308,
      "learning_rate": 0.000994246618455783,
      "loss": 1.5976,
      "step": 5300
    },
    {
      "epoch": 0.2357387399404206,
      "grad_norm": 0.10771182179450989,
      "learning_rate": 0.0009942412755279623,
      "loss": 1.6041,
      "step": 5302
    },
    {
      "epoch": 0.23582766439909297,
      "grad_norm": 0.10492417216300964,
      "learning_rate": 0.000994235930134784,
      "loss": 1.5976,
      "step": 5304
    },
    {
      "epoch": 0.23591658885776534,
      "grad_norm": 0.1006433516740799,
      "learning_rate": 0.000994230582276275,
      "loss": 1.5996,
      "step": 5306
    },
    {
      "epoch": 0.23600551331643768,
      "grad_norm": 0.10901300609111786,
      "learning_rate": 0.0009942252319524619,
      "loss": 1.6019,
      "step": 5308
    },
    {
      "epoch": 0.23609443777511005,
      "grad_norm": 0.10461027175188065,
      "learning_rate": 0.000994219879163371,
      "loss": 1.5967,
      "step": 5310
    },
    {
      "epoch": 0.2361833622337824,
      "grad_norm": 0.09647329896688461,
      "learning_rate": 0.0009942145239090294,
      "loss": 1.5951,
      "step": 5312
    },
    {
      "epoch": 0.23627228669245476,
      "grad_norm": 0.10260302573442459,
      "learning_rate": 0.000994209166189464,
      "loss": 1.6025,
      "step": 5314
    },
    {
      "epoch": 0.23636121115112713,
      "grad_norm": 0.09995372593402863,
      "learning_rate": 0.0009942038060047008,
      "loss": 1.5971,
      "step": 5316
    },
    {
      "epoch": 0.23645013560979947,
      "grad_norm": 0.0978364571928978,
      "learning_rate": 0.0009941984433547674,
      "loss": 1.6077,
      "step": 5318
    },
    {
      "epoch": 0.23653906006847183,
      "grad_norm": 0.10295219719409943,
      "learning_rate": 0.0009941930782396899,
      "loss": 1.6031,
      "step": 5320
    },
    {
      "epoch": 0.2366279845271442,
      "grad_norm": 0.10604212433099747,
      "learning_rate": 0.0009941877106594953,
      "loss": 1.5958,
      "step": 5322
    },
    {
      "epoch": 0.23671690898581654,
      "grad_norm": 0.09841012209653854,
      "learning_rate": 0.0009941823406142104,
      "loss": 1.5927,
      "step": 5324
    },
    {
      "epoch": 0.2368058334444889,
      "grad_norm": 0.09888961166143417,
      "learning_rate": 0.0009941769681038619,
      "loss": 1.5987,
      "step": 5326
    },
    {
      "epoch": 0.23689475790316125,
      "grad_norm": 0.10190734267234802,
      "learning_rate": 0.0009941715931284767,
      "loss": 1.5938,
      "step": 5328
    },
    {
      "epoch": 0.23698368236183362,
      "grad_norm": 0.10010533779859543,
      "learning_rate": 0.0009941662156880816,
      "loss": 1.5994,
      "step": 5330
    },
    {
      "epoch": 0.237072606820506,
      "grad_norm": 0.09754836559295654,
      "learning_rate": 0.0009941608357827035,
      "loss": 1.5942,
      "step": 5332
    },
    {
      "epoch": 0.23716153127917833,
      "grad_norm": 0.09817097336053848,
      "learning_rate": 0.0009941554534123689,
      "loss": 1.598,
      "step": 5334
    },
    {
      "epoch": 0.2372504557378507,
      "grad_norm": 0.09796758741140366,
      "learning_rate": 0.000994150068577105,
      "loss": 1.598,
      "step": 5336
    },
    {
      "epoch": 0.23733938019652306,
      "grad_norm": 0.09705115854740143,
      "learning_rate": 0.0009941446812769385,
      "loss": 1.598,
      "step": 5338
    },
    {
      "epoch": 0.2374283046551954,
      "grad_norm": 0.0968238040804863,
      "learning_rate": 0.0009941392915118962,
      "loss": 1.6016,
      "step": 5340
    },
    {
      "epoch": 0.23751722911386777,
      "grad_norm": 0.09914400428533554,
      "learning_rate": 0.0009941338992820053,
      "loss": 1.5997,
      "step": 5342
    },
    {
      "epoch": 0.23760615357254014,
      "grad_norm": 0.09874872863292694,
      "learning_rate": 0.0009941285045872922,
      "loss": 1.6013,
      "step": 5344
    },
    {
      "epoch": 0.23769507803121248,
      "grad_norm": 0.10324931144714355,
      "learning_rate": 0.0009941231074277843,
      "loss": 1.6,
      "step": 5346
    },
    {
      "epoch": 0.23778400248988485,
      "grad_norm": 0.09846501797437668,
      "learning_rate": 0.0009941177078035084,
      "loss": 1.5945,
      "step": 5348
    },
    {
      "epoch": 0.2378729269485572,
      "grad_norm": 0.09790665656328201,
      "learning_rate": 0.0009941123057144912,
      "loss": 1.5987,
      "step": 5350
    },
    {
      "epoch": 0.23796185140722956,
      "grad_norm": 0.10351137071847916,
      "learning_rate": 0.00099410690116076,
      "loss": 1.5974,
      "step": 5352
    },
    {
      "epoch": 0.23805077586590193,
      "grad_norm": 0.1032329574227333,
      "learning_rate": 0.0009941014941423413,
      "loss": 1.5964,
      "step": 5354
    },
    {
      "epoch": 0.23813970032457427,
      "grad_norm": 0.0977087914943695,
      "learning_rate": 0.0009940960846592623,
      "loss": 1.5978,
      "step": 5356
    },
    {
      "epoch": 0.23822862478324663,
      "grad_norm": 0.10141148418188095,
      "learning_rate": 0.0009940906727115503,
      "loss": 1.6033,
      "step": 5358
    },
    {
      "epoch": 0.238317549241919,
      "grad_norm": 0.10696842521429062,
      "learning_rate": 0.0009940852582992319,
      "loss": 1.5956,
      "step": 5360
    },
    {
      "epoch": 0.23840647370059134,
      "grad_norm": 0.10355627536773682,
      "learning_rate": 0.0009940798414223343,
      "loss": 1.5973,
      "step": 5362
    },
    {
      "epoch": 0.2384953981592637,
      "grad_norm": 0.10255977511405945,
      "learning_rate": 0.000994074422080884,
      "loss": 1.597,
      "step": 5364
    },
    {
      "epoch": 0.23858432261793605,
      "grad_norm": 0.09989781677722931,
      "learning_rate": 0.000994069000274909,
      "loss": 1.5986,
      "step": 5366
    },
    {
      "epoch": 0.23867324707660842,
      "grad_norm": 0.10210977494716644,
      "learning_rate": 0.0009940635760044357,
      "loss": 1.5956,
      "step": 5368
    },
    {
      "epoch": 0.2387621715352808,
      "grad_norm": 0.09619224071502686,
      "learning_rate": 0.0009940581492694912,
      "loss": 1.5962,
      "step": 5370
    },
    {
      "epoch": 0.23885109599395313,
      "grad_norm": 0.09644219279289246,
      "learning_rate": 0.0009940527200701023,
      "loss": 1.6002,
      "step": 5372
    },
    {
      "epoch": 0.2389400204526255,
      "grad_norm": 0.102609783411026,
      "learning_rate": 0.000994047288406297,
      "loss": 1.5974,
      "step": 5374
    },
    {
      "epoch": 0.23902894491129786,
      "grad_norm": 0.10200612992048264,
      "learning_rate": 0.0009940418542781014,
      "loss": 1.5975,
      "step": 5376
    },
    {
      "epoch": 0.2391178693699702,
      "grad_norm": 0.09754066914319992,
      "learning_rate": 0.0009940364176855433,
      "loss": 1.6006,
      "step": 5378
    },
    {
      "epoch": 0.23920679382864257,
      "grad_norm": 0.10025352984666824,
      "learning_rate": 0.0009940309786286494,
      "loss": 1.5991,
      "step": 5380
    },
    {
      "epoch": 0.2392957182873149,
      "grad_norm": 0.10298909991979599,
      "learning_rate": 0.000994025537107447,
      "loss": 1.5975,
      "step": 5382
    },
    {
      "epoch": 0.23938464274598728,
      "grad_norm": 0.09619417786598206,
      "learning_rate": 0.0009940200931219632,
      "loss": 1.595,
      "step": 5384
    },
    {
      "epoch": 0.23947356720465965,
      "grad_norm": 0.10511010885238647,
      "learning_rate": 0.0009940146466722252,
      "loss": 1.5977,
      "step": 5386
    },
    {
      "epoch": 0.239562491663332,
      "grad_norm": 0.09961894154548645,
      "learning_rate": 0.0009940091977582602,
      "loss": 1.5973,
      "step": 5388
    },
    {
      "epoch": 0.23965141612200436,
      "grad_norm": 0.09826082736253738,
      "learning_rate": 0.0009940037463800951,
      "loss": 1.6016,
      "step": 5390
    },
    {
      "epoch": 0.23974034058067673,
      "grad_norm": 0.10174942016601562,
      "learning_rate": 0.0009939982925377575,
      "loss": 1.5996,
      "step": 5392
    },
    {
      "epoch": 0.23982926503934907,
      "grad_norm": 0.09913383424282074,
      "learning_rate": 0.0009939928362312745,
      "loss": 1.5986,
      "step": 5394
    },
    {
      "epoch": 0.23991818949802143,
      "grad_norm": 0.10252643376588821,
      "learning_rate": 0.000993987377460673,
      "loss": 1.601,
      "step": 5396
    },
    {
      "epoch": 0.2400071139566938,
      "grad_norm": 0.0993933230638504,
      "learning_rate": 0.0009939819162259806,
      "loss": 1.5984,
      "step": 5398
    },
    {
      "epoch": 0.24009603841536614,
      "grad_norm": 0.09531717747449875,
      "learning_rate": 0.0009939764525272244,
      "loss": 1.5911,
      "step": 5400
    },
    {
      "epoch": 0.2401849628740385,
      "grad_norm": 0.09904836863279343,
      "learning_rate": 0.0009939709863644316,
      "loss": 1.6022,
      "step": 5402
    },
    {
      "epoch": 0.24027388733271085,
      "grad_norm": 0.0987459197640419,
      "learning_rate": 0.0009939655177376298,
      "loss": 1.5894,
      "step": 5404
    },
    {
      "epoch": 0.24036281179138322,
      "grad_norm": 0.09736425429582596,
      "learning_rate": 0.0009939600466468456,
      "loss": 1.5912,
      "step": 5406
    },
    {
      "epoch": 0.2404517362500556,
      "grad_norm": 0.09566332399845123,
      "learning_rate": 0.000993954573092107,
      "loss": 1.5967,
      "step": 5408
    },
    {
      "epoch": 0.24054066070872793,
      "grad_norm": 0.10598226636648178,
      "learning_rate": 0.0009939490970734407,
      "loss": 1.5986,
      "step": 5410
    },
    {
      "epoch": 0.2406295851674003,
      "grad_norm": 0.10711424052715302,
      "learning_rate": 0.0009939436185908744,
      "loss": 1.599,
      "step": 5412
    },
    {
      "epoch": 0.24071850962607266,
      "grad_norm": 0.1163632795214653,
      "learning_rate": 0.0009939381376444353,
      "loss": 1.6,
      "step": 5414
    },
    {
      "epoch": 0.240807434084745,
      "grad_norm": 0.1008208766579628,
      "learning_rate": 0.0009939326542341511,
      "loss": 1.598,
      "step": 5416
    },
    {
      "epoch": 0.24089635854341737,
      "grad_norm": 0.10526162385940552,
      "learning_rate": 0.0009939271683600487,
      "loss": 1.5995,
      "step": 5418
    },
    {
      "epoch": 0.2409852830020897,
      "grad_norm": 0.10237567126750946,
      "learning_rate": 0.0009939216800221554,
      "loss": 1.5987,
      "step": 5420
    },
    {
      "epoch": 0.24107420746076208,
      "grad_norm": 0.09717054665088654,
      "learning_rate": 0.000993916189220499,
      "loss": 1.5884,
      "step": 5422
    },
    {
      "epoch": 0.24116313191943445,
      "grad_norm": 0.10505110770463943,
      "learning_rate": 0.0009939106959551065,
      "loss": 1.5971,
      "step": 5424
    },
    {
      "epoch": 0.2412520563781068,
      "grad_norm": 0.10448066890239716,
      "learning_rate": 0.0009939052002260055,
      "loss": 1.59,
      "step": 5426
    },
    {
      "epoch": 0.24134098083677916,
      "grad_norm": 0.10122029483318329,
      "learning_rate": 0.0009938997020332233,
      "loss": 1.5975,
      "step": 5428
    },
    {
      "epoch": 0.24142990529545152,
      "grad_norm": 0.10328763723373413,
      "learning_rate": 0.0009938942013767876,
      "loss": 1.5994,
      "step": 5430
    },
    {
      "epoch": 0.24151882975412386,
      "grad_norm": 0.10285595804452896,
      "learning_rate": 0.0009938886982567255,
      "loss": 1.6059,
      "step": 5432
    },
    {
      "epoch": 0.24160775421279623,
      "grad_norm": 0.10578134655952454,
      "learning_rate": 0.0009938831926730647,
      "loss": 1.5949,
      "step": 5434
    },
    {
      "epoch": 0.2416966786714686,
      "grad_norm": 0.09849389642477036,
      "learning_rate": 0.0009938776846258324,
      "loss": 1.598,
      "step": 5436
    },
    {
      "epoch": 0.24178560313014094,
      "grad_norm": 0.0992036834359169,
      "learning_rate": 0.0009938721741150564,
      "loss": 1.5966,
      "step": 5438
    },
    {
      "epoch": 0.2418745275888133,
      "grad_norm": 0.10037153214216232,
      "learning_rate": 0.0009938666611407638,
      "loss": 1.6067,
      "step": 5440
    },
    {
      "epoch": 0.24196345204748565,
      "grad_norm": 0.10501682758331299,
      "learning_rate": 0.0009938611457029826,
      "loss": 1.5944,
      "step": 5442
    },
    {
      "epoch": 0.24205237650615802,
      "grad_norm": 0.0952974334359169,
      "learning_rate": 0.00099385562780174,
      "loss": 1.5969,
      "step": 5444
    },
    {
      "epoch": 0.24214130096483039,
      "grad_norm": 0.1013774648308754,
      "learning_rate": 0.0009938501074370633,
      "loss": 1.6038,
      "step": 5446
    },
    {
      "epoch": 0.24223022542350273,
      "grad_norm": 0.09989754110574722,
      "learning_rate": 0.0009938445846089804,
      "loss": 1.6012,
      "step": 5448
    },
    {
      "epoch": 0.2423191498821751,
      "grad_norm": 0.10097597539424896,
      "learning_rate": 0.0009938390593175188,
      "loss": 1.5928,
      "step": 5450
    },
    {
      "epoch": 0.24240807434084746,
      "grad_norm": 0.09957058727741241,
      "learning_rate": 0.000993833531562706,
      "loss": 1.5999,
      "step": 5452
    },
    {
      "epoch": 0.2424969987995198,
      "grad_norm": 0.09687520563602448,
      "learning_rate": 0.0009938280013445696,
      "loss": 1.5957,
      "step": 5454
    },
    {
      "epoch": 0.24258592325819217,
      "grad_norm": 0.1019771546125412,
      "learning_rate": 0.000993822468663137,
      "loss": 1.6009,
      "step": 5456
    },
    {
      "epoch": 0.2426748477168645,
      "grad_norm": 0.09672028571367264,
      "learning_rate": 0.000993816933518436,
      "loss": 1.5961,
      "step": 5458
    },
    {
      "epoch": 0.24276377217553688,
      "grad_norm": 0.09733521938323975,
      "learning_rate": 0.000993811395910494,
      "loss": 1.5935,
      "step": 5460
    },
    {
      "epoch": 0.24285269663420925,
      "grad_norm": 0.09702073037624359,
      "learning_rate": 0.0009938058558393392,
      "loss": 1.5958,
      "step": 5462
    },
    {
      "epoch": 0.2429416210928816,
      "grad_norm": 0.10039489716291428,
      "learning_rate": 0.0009938003133049987,
      "loss": 1.5989,
      "step": 5464
    },
    {
      "epoch": 0.24303054555155396,
      "grad_norm": 0.0961323231458664,
      "learning_rate": 0.0009937947683075002,
      "loss": 1.5929,
      "step": 5466
    },
    {
      "epoch": 0.24311947001022632,
      "grad_norm": 0.0994456484913826,
      "learning_rate": 0.0009937892208468713,
      "loss": 1.5967,
      "step": 5468
    },
    {
      "epoch": 0.24320839446889866,
      "grad_norm": 0.09631854295730591,
      "learning_rate": 0.00099378367092314,
      "loss": 1.5909,
      "step": 5470
    },
    {
      "epoch": 0.24329731892757103,
      "grad_norm": 0.0983763337135315,
      "learning_rate": 0.0009937781185363338,
      "loss": 1.6006,
      "step": 5472
    },
    {
      "epoch": 0.24338624338624337,
      "grad_norm": 0.10499723255634308,
      "learning_rate": 0.0009937725636864803,
      "loss": 1.5919,
      "step": 5474
    },
    {
      "epoch": 0.24347516784491574,
      "grad_norm": 0.10987919569015503,
      "learning_rate": 0.0009937670063736072,
      "loss": 1.5988,
      "step": 5476
    },
    {
      "epoch": 0.2435640923035881,
      "grad_norm": 0.105495885014534,
      "learning_rate": 0.0009937614465977422,
      "loss": 1.5977,
      "step": 5478
    },
    {
      "epoch": 0.24365301676226045,
      "grad_norm": 0.09992179274559021,
      "learning_rate": 0.0009937558843589134,
      "loss": 1.6033,
      "step": 5480
    },
    {
      "epoch": 0.24374194122093282,
      "grad_norm": 0.10099846124649048,
      "learning_rate": 0.0009937503196571482,
      "loss": 1.5892,
      "step": 5482
    },
    {
      "epoch": 0.24383086567960519,
      "grad_norm": 0.10099624842405319,
      "learning_rate": 0.0009937447524924745,
      "loss": 1.5986,
      "step": 5484
    },
    {
      "epoch": 0.24391979013827753,
      "grad_norm": 0.10109474509954453,
      "learning_rate": 0.00099373918286492,
      "loss": 1.595,
      "step": 5486
    },
    {
      "epoch": 0.2440087145969499,
      "grad_norm": 0.09985104948282242,
      "learning_rate": 0.0009937336107745125,
      "loss": 1.5983,
      "step": 5488
    },
    {
      "epoch": 0.24409763905562226,
      "grad_norm": 0.09679286181926727,
      "learning_rate": 0.0009937280362212798,
      "loss": 1.5893,
      "step": 5490
    },
    {
      "epoch": 0.2441865635142946,
      "grad_norm": 0.09930288046598434,
      "learning_rate": 0.0009937224592052496,
      "loss": 1.598,
      "step": 5492
    },
    {
      "epoch": 0.24427548797296697,
      "grad_norm": 0.10350073873996735,
      "learning_rate": 0.00099371687972645,
      "loss": 1.5957,
      "step": 5494
    },
    {
      "epoch": 0.2443644124316393,
      "grad_norm": 0.10076453536748886,
      "learning_rate": 0.0009937112977849085,
      "loss": 1.598,
      "step": 5496
    },
    {
      "epoch": 0.24445333689031168,
      "grad_norm": 0.0963132232427597,
      "learning_rate": 0.0009937057133806533,
      "loss": 1.5974,
      "step": 5498
    },
    {
      "epoch": 0.24454226134898405,
      "grad_norm": 0.0963580459356308,
      "learning_rate": 0.0009937001265137118,
      "loss": 1.5958,
      "step": 5500
    },
    {
      "epoch": 0.24454226134898405,
      "eval_loss": 1.5716533660888672,
      "eval_runtime": 12.37,
      "eval_samples_per_second": 558.609,
      "eval_steps_per_second": 69.846,
      "step": 5500
    },
    {
      "epoch": 0.2446311858076564,
      "grad_norm": 0.0988180860877037,
      "learning_rate": 0.0009936945371841121,
      "loss": 1.5965,
      "step": 5502
    },
    {
      "epoch": 0.24472011026632876,
      "grad_norm": 0.10076916962862015,
      "learning_rate": 0.0009936889453918824,
      "loss": 1.5974,
      "step": 5504
    },
    {
      "epoch": 0.24480903472500112,
      "grad_norm": 0.09656892716884613,
      "learning_rate": 0.00099368335113705,
      "loss": 1.5992,
      "step": 5506
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 0.09603594988584518,
      "learning_rate": 0.0009936777544196432,
      "loss": 1.5994,
      "step": 5508
    },
    {
      "epoch": 0.24498688364234583,
      "grad_norm": 0.09735065698623657,
      "learning_rate": 0.00099367215523969,
      "loss": 1.5918,
      "step": 5510
    },
    {
      "epoch": 0.24507580810101817,
      "grad_norm": 0.09379937499761581,
      "learning_rate": 0.0009936665535972178,
      "loss": 1.5911,
      "step": 5512
    },
    {
      "epoch": 0.24516473255969054,
      "grad_norm": 0.09495367109775543,
      "learning_rate": 0.000993660949492255,
      "loss": 1.5949,
      "step": 5514
    },
    {
      "epoch": 0.2452536570183629,
      "grad_norm": 0.10689006745815277,
      "learning_rate": 0.0009936553429248296,
      "loss": 1.5944,
      "step": 5516
    },
    {
      "epoch": 0.24534258147703525,
      "grad_norm": 0.10062182694673538,
      "learning_rate": 0.0009936497338949692,
      "loss": 1.5903,
      "step": 5518
    },
    {
      "epoch": 0.24543150593570762,
      "grad_norm": 0.10268256068229675,
      "learning_rate": 0.0009936441224027022,
      "loss": 1.594,
      "step": 5520
    },
    {
      "epoch": 0.24552043039437998,
      "grad_norm": 0.09993576258420944,
      "learning_rate": 0.000993638508448056,
      "loss": 1.5991,
      "step": 5522
    },
    {
      "epoch": 0.24560935485305233,
      "grad_norm": 0.10125838965177536,
      "learning_rate": 0.0009936328920310594,
      "loss": 1.6004,
      "step": 5524
    },
    {
      "epoch": 0.2456982793117247,
      "grad_norm": 0.10417357832193375,
      "learning_rate": 0.00099362727315174,
      "loss": 1.5928,
      "step": 5526
    },
    {
      "epoch": 0.24578720377039706,
      "grad_norm": 0.09818510711193085,
      "learning_rate": 0.0009936216518101255,
      "loss": 1.5964,
      "step": 5528
    },
    {
      "epoch": 0.2458761282290694,
      "grad_norm": 0.0970878154039383,
      "learning_rate": 0.0009936160280062446,
      "loss": 1.5894,
      "step": 5530
    },
    {
      "epoch": 0.24596505268774177,
      "grad_norm": 0.1003979966044426,
      "learning_rate": 0.0009936104017401247,
      "loss": 1.5873,
      "step": 5532
    },
    {
      "epoch": 0.2460539771464141,
      "grad_norm": 0.09515174478292465,
      "learning_rate": 0.0009936047730117943,
      "loss": 1.5936,
      "step": 5534
    },
    {
      "epoch": 0.24614290160508648,
      "grad_norm": 0.10499396920204163,
      "learning_rate": 0.0009935991418212815,
      "loss": 1.5907,
      "step": 5536
    },
    {
      "epoch": 0.24623182606375885,
      "grad_norm": 0.09744211286306381,
      "learning_rate": 0.000993593508168614,
      "loss": 1.595,
      "step": 5538
    },
    {
      "epoch": 0.2463207505224312,
      "grad_norm": 0.10020717233419418,
      "learning_rate": 0.0009935878720538202,
      "loss": 1.5971,
      "step": 5540
    },
    {
      "epoch": 0.24640967498110355,
      "grad_norm": 0.09754054993391037,
      "learning_rate": 0.0009935822334769283,
      "loss": 1.5877,
      "step": 5542
    },
    {
      "epoch": 0.24649859943977592,
      "grad_norm": 0.09400993585586548,
      "learning_rate": 0.0009935765924379662,
      "loss": 1.5937,
      "step": 5544
    },
    {
      "epoch": 0.24658752389844826,
      "grad_norm": 0.10023734718561172,
      "learning_rate": 0.0009935709489369623,
      "loss": 1.593,
      "step": 5546
    },
    {
      "epoch": 0.24667644835712063,
      "grad_norm": 0.10244899243116379,
      "learning_rate": 0.0009935653029739444,
      "loss": 1.5898,
      "step": 5548
    },
    {
      "epoch": 0.24676537281579297,
      "grad_norm": 0.09722084552049637,
      "learning_rate": 0.0009935596545489409,
      "loss": 1.5888,
      "step": 5550
    },
    {
      "epoch": 0.24685429727446534,
      "grad_norm": 0.09749680012464523,
      "learning_rate": 0.0009935540036619798,
      "loss": 1.5938,
      "step": 5552
    },
    {
      "epoch": 0.2469432217331377,
      "grad_norm": 0.09648852795362473,
      "learning_rate": 0.0009935483503130896,
      "loss": 1.5966,
      "step": 5554
    },
    {
      "epoch": 0.24703214619181005,
      "grad_norm": 0.09504228085279465,
      "learning_rate": 0.0009935426945022982,
      "loss": 1.5899,
      "step": 5556
    },
    {
      "epoch": 0.24712107065048242,
      "grad_norm": 0.1025584489107132,
      "learning_rate": 0.000993537036229634,
      "loss": 1.5932,
      "step": 5558
    },
    {
      "epoch": 0.24720999510915478,
      "grad_norm": 0.096546970307827,
      "learning_rate": 0.0009935313754951248,
      "loss": 1.5959,
      "step": 5560
    },
    {
      "epoch": 0.24729891956782712,
      "grad_norm": 0.09677073359489441,
      "learning_rate": 0.0009935257122987995,
      "loss": 1.593,
      "step": 5562
    },
    {
      "epoch": 0.2473878440264995,
      "grad_norm": 0.09250880032777786,
      "learning_rate": 0.000993520046640686,
      "loss": 1.5935,
      "step": 5564
    },
    {
      "epoch": 0.24747676848517183,
      "grad_norm": 0.09518403559923172,
      "learning_rate": 0.0009935143785208126,
      "loss": 1.5942,
      "step": 5566
    },
    {
      "epoch": 0.2475656929438442,
      "grad_norm": 0.09718389809131622,
      "learning_rate": 0.0009935087079392074,
      "loss": 1.5959,
      "step": 5568
    },
    {
      "epoch": 0.24765461740251657,
      "grad_norm": 0.09807734936475754,
      "learning_rate": 0.0009935030348958989,
      "loss": 1.5914,
      "step": 5570
    },
    {
      "epoch": 0.2477435418611889,
      "grad_norm": 0.09537956118583679,
      "learning_rate": 0.0009934973593909156,
      "loss": 1.5932,
      "step": 5572
    },
    {
      "epoch": 0.24783246631986128,
      "grad_norm": 0.09631271660327911,
      "learning_rate": 0.000993491681424285,
      "loss": 1.5892,
      "step": 5574
    },
    {
      "epoch": 0.24792139077853365,
      "grad_norm": 0.10306434333324432,
      "learning_rate": 0.0009934860009960364,
      "loss": 1.5889,
      "step": 5576
    },
    {
      "epoch": 0.24801031523720599,
      "grad_norm": 0.09478336572647095,
      "learning_rate": 0.0009934803181061978,
      "loss": 1.5932,
      "step": 5578
    },
    {
      "epoch": 0.24809923969587835,
      "grad_norm": 0.0936090350151062,
      "learning_rate": 0.0009934746327547972,
      "loss": 1.5883,
      "step": 5580
    },
    {
      "epoch": 0.24818816415455072,
      "grad_norm": 0.09921901673078537,
      "learning_rate": 0.0009934689449418634,
      "loss": 1.5914,
      "step": 5582
    },
    {
      "epoch": 0.24827708861322306,
      "grad_norm": 0.09646906703710556,
      "learning_rate": 0.0009934632546674245,
      "loss": 1.5998,
      "step": 5584
    },
    {
      "epoch": 0.24836601307189543,
      "grad_norm": 0.09717948734760284,
      "learning_rate": 0.000993457561931509,
      "loss": 1.5946,
      "step": 5586
    },
    {
      "epoch": 0.24845493753056777,
      "grad_norm": 0.09641571342945099,
      "learning_rate": 0.0009934518667341452,
      "loss": 1.5949,
      "step": 5588
    },
    {
      "epoch": 0.24854386198924014,
      "grad_norm": 0.09554963558912277,
      "learning_rate": 0.0009934461690753616,
      "loss": 1.5885,
      "step": 5590
    },
    {
      "epoch": 0.2486327864479125,
      "grad_norm": 0.10000897943973541,
      "learning_rate": 0.0009934404689551867,
      "loss": 1.5928,
      "step": 5592
    },
    {
      "epoch": 0.24872171090658485,
      "grad_norm": 0.09571028500795364,
      "learning_rate": 0.0009934347663736486,
      "loss": 1.5956,
      "step": 5594
    },
    {
      "epoch": 0.24881063536525722,
      "grad_norm": 0.09674900025129318,
      "learning_rate": 0.0009934290613307763,
      "loss": 1.5952,
      "step": 5596
    },
    {
      "epoch": 0.24889955982392958,
      "grad_norm": 0.09690657258033752,
      "learning_rate": 0.0009934233538265978,
      "loss": 1.5927,
      "step": 5598
    },
    {
      "epoch": 0.24898848428260192,
      "grad_norm": 0.09681335836648941,
      "learning_rate": 0.0009934176438611418,
      "loss": 1.5968,
      "step": 5600
    },
    {
      "epoch": 0.2490774087412743,
      "grad_norm": 0.0922679454088211,
      "learning_rate": 0.0009934119314344363,
      "loss": 1.5888,
      "step": 5602
    },
    {
      "epoch": 0.24916633319994663,
      "grad_norm": 0.09346043318510056,
      "learning_rate": 0.0009934062165465106,
      "loss": 1.5937,
      "step": 5604
    },
    {
      "epoch": 0.249255257658619,
      "grad_norm": 0.09791743010282516,
      "learning_rate": 0.0009934004991973926,
      "loss": 1.5897,
      "step": 5606
    },
    {
      "epoch": 0.24934418211729137,
      "grad_norm": 0.09670942276716232,
      "learning_rate": 0.0009933947793871111,
      "loss": 1.5893,
      "step": 5608
    },
    {
      "epoch": 0.2494331065759637,
      "grad_norm": 0.09886854887008667,
      "learning_rate": 0.0009933890571156945,
      "loss": 1.5851,
      "step": 5610
    },
    {
      "epoch": 0.24952203103463608,
      "grad_norm": 0.10364110767841339,
      "learning_rate": 0.0009933833323831714,
      "loss": 1.591,
      "step": 5612
    },
    {
      "epoch": 0.24961095549330845,
      "grad_norm": 0.09943173825740814,
      "learning_rate": 0.0009933776051895705,
      "loss": 1.5873,
      "step": 5614
    },
    {
      "epoch": 0.24969987995198079,
      "grad_norm": 0.09556498378515244,
      "learning_rate": 0.00099337187553492,
      "loss": 1.59,
      "step": 5616
    },
    {
      "epoch": 0.24978880441065315,
      "grad_norm": 0.09815754741430283,
      "learning_rate": 0.0009933661434192488,
      "loss": 1.5908,
      "step": 5618
    },
    {
      "epoch": 0.24987772886932552,
      "grad_norm": 0.09476683288812637,
      "learning_rate": 0.0009933604088425852,
      "loss": 1.5901,
      "step": 5620
    },
    {
      "epoch": 0.24996665332799786,
      "grad_norm": 0.09704037010669708,
      "learning_rate": 0.000993354671804958,
      "loss": 1.5915,
      "step": 5622
    },
    {
      "epoch": 0.2500555777866702,
      "grad_norm": 0.09691561013460159,
      "learning_rate": 0.000993348932306396,
      "loss": 1.5928,
      "step": 5624
    },
    {
      "epoch": 0.25014450224534257,
      "grad_norm": 0.09649381786584854,
      "learning_rate": 0.0009933431903469275,
      "loss": 1.5923,
      "step": 5626
    },
    {
      "epoch": 0.25023342670401494,
      "grad_norm": 0.10152386128902435,
      "learning_rate": 0.0009933374459265812,
      "loss": 1.5896,
      "step": 5628
    },
    {
      "epoch": 0.2503223511626873,
      "grad_norm": 0.09325571358203888,
      "learning_rate": 0.000993331699045386,
      "loss": 1.5933,
      "step": 5630
    },
    {
      "epoch": 0.2504112756213597,
      "grad_norm": 0.09736236929893494,
      "learning_rate": 0.0009933259497033704,
      "loss": 1.5925,
      "step": 5632
    },
    {
      "epoch": 0.250500200080032,
      "grad_norm": 0.09810613095760345,
      "learning_rate": 0.0009933201979005631,
      "loss": 1.5907,
      "step": 5634
    },
    {
      "epoch": 0.25058912453870436,
      "grad_norm": 0.09613955765962601,
      "learning_rate": 0.0009933144436369925,
      "loss": 1.5983,
      "step": 5636
    },
    {
      "epoch": 0.2506780489973767,
      "grad_norm": 0.09884078800678253,
      "learning_rate": 0.0009933086869126878,
      "loss": 1.5894,
      "step": 5638
    },
    {
      "epoch": 0.2507669734560491,
      "grad_norm": 0.09554318338632584,
      "learning_rate": 0.0009933029277276774,
      "loss": 1.5948,
      "step": 5640
    },
    {
      "epoch": 0.25085589791472146,
      "grad_norm": 0.09753008186817169,
      "learning_rate": 0.0009932971660819901,
      "loss": 1.5868,
      "step": 5642
    },
    {
      "epoch": 0.2509448223733938,
      "grad_norm": 0.09506553411483765,
      "learning_rate": 0.0009932914019756547,
      "loss": 1.5923,
      "step": 5644
    },
    {
      "epoch": 0.25103374683206614,
      "grad_norm": 0.09722185879945755,
      "learning_rate": 0.0009932856354087,
      "loss": 1.5925,
      "step": 5646
    },
    {
      "epoch": 0.2511226712907385,
      "grad_norm": 0.10006613284349442,
      "learning_rate": 0.0009932798663811544,
      "loss": 1.5926,
      "step": 5648
    },
    {
      "epoch": 0.2512115957494109,
      "grad_norm": 0.09580423682928085,
      "learning_rate": 0.0009932740948930471,
      "loss": 1.5899,
      "step": 5650
    },
    {
      "epoch": 0.25130052020808324,
      "grad_norm": 0.09586209803819656,
      "learning_rate": 0.000993268320944407,
      "loss": 1.5942,
      "step": 5652
    },
    {
      "epoch": 0.2513894446667556,
      "grad_norm": 0.09892687201499939,
      "learning_rate": 0.0009932625445352623,
      "loss": 1.5944,
      "step": 5654
    },
    {
      "epoch": 0.2514783691254279,
      "grad_norm": 0.09688273817300797,
      "learning_rate": 0.000993256765665642,
      "loss": 1.5922,
      "step": 5656
    },
    {
      "epoch": 0.2515672935841003,
      "grad_norm": 0.0952548235654831,
      "learning_rate": 0.0009932509843355755,
      "loss": 1.5931,
      "step": 5658
    },
    {
      "epoch": 0.25165621804277266,
      "grad_norm": 0.09588255733251572,
      "learning_rate": 0.0009932452005450912,
      "loss": 1.5877,
      "step": 5660
    },
    {
      "epoch": 0.25174514250144503,
      "grad_norm": 0.09836259484291077,
      "learning_rate": 0.0009932394142942176,
      "loss": 1.5873,
      "step": 5662
    },
    {
      "epoch": 0.2518340669601174,
      "grad_norm": 0.09556706994771957,
      "learning_rate": 0.0009932336255829841,
      "loss": 1.5933,
      "step": 5664
    },
    {
      "epoch": 0.25192299141878977,
      "grad_norm": 0.10461506247520447,
      "learning_rate": 0.0009932278344114195,
      "loss": 1.5922,
      "step": 5666
    },
    {
      "epoch": 0.2520119158774621,
      "grad_norm": 0.09809567034244537,
      "learning_rate": 0.0009932220407795525,
      "loss": 1.5999,
      "step": 5668
    },
    {
      "epoch": 0.25210084033613445,
      "grad_norm": 0.09388605505228043,
      "learning_rate": 0.0009932162446874121,
      "loss": 1.5885,
      "step": 5670
    },
    {
      "epoch": 0.2521897647948068,
      "grad_norm": 0.09325406700372696,
      "learning_rate": 0.0009932104461350274,
      "loss": 1.5912,
      "step": 5672
    },
    {
      "epoch": 0.2522786892534792,
      "grad_norm": 0.0968768522143364,
      "learning_rate": 0.0009932046451224268,
      "loss": 1.5942,
      "step": 5674
    },
    {
      "epoch": 0.25236761371215155,
      "grad_norm": 0.0974501445889473,
      "learning_rate": 0.0009931988416496396,
      "loss": 1.5917,
      "step": 5676
    },
    {
      "epoch": 0.25245653817082386,
      "grad_norm": 0.09732449054718018,
      "learning_rate": 0.0009931930357166949,
      "loss": 1.5937,
      "step": 5678
    },
    {
      "epoch": 0.25254546262949623,
      "grad_norm": 0.10150519758462906,
      "learning_rate": 0.0009931872273236213,
      "loss": 1.5885,
      "step": 5680
    },
    {
      "epoch": 0.2526343870881686,
      "grad_norm": 0.1010938212275505,
      "learning_rate": 0.000993181416470448,
      "loss": 1.5845,
      "step": 5682
    },
    {
      "epoch": 0.25272331154684097,
      "grad_norm": 0.09250231832265854,
      "learning_rate": 0.000993175603157204,
      "loss": 1.5867,
      "step": 5684
    },
    {
      "epoch": 0.25281223600551334,
      "grad_norm": 0.09422440081834793,
      "learning_rate": 0.0009931697873839182,
      "loss": 1.5957,
      "step": 5686
    },
    {
      "epoch": 0.25290116046418565,
      "grad_norm": 0.09714167565107346,
      "learning_rate": 0.0009931639691506196,
      "loss": 1.5932,
      "step": 5688
    },
    {
      "epoch": 0.252990084922858,
      "grad_norm": 0.0974729061126709,
      "learning_rate": 0.0009931581484573374,
      "loss": 1.5889,
      "step": 5690
    },
    {
      "epoch": 0.2530790093815304,
      "grad_norm": 0.09551315754652023,
      "learning_rate": 0.0009931523253041002,
      "loss": 1.5846,
      "step": 5692
    },
    {
      "epoch": 0.25316793384020275,
      "grad_norm": 0.09637055546045303,
      "learning_rate": 0.0009931464996909376,
      "loss": 1.594,
      "step": 5694
    },
    {
      "epoch": 0.2532568582988751,
      "grad_norm": 0.09468736499547958,
      "learning_rate": 0.0009931406716178783,
      "loss": 1.5897,
      "step": 5696
    },
    {
      "epoch": 0.2533457827575475,
      "grad_norm": 0.09597856551408768,
      "learning_rate": 0.0009931348410849514,
      "loss": 1.5879,
      "step": 5698
    },
    {
      "epoch": 0.2534347072162198,
      "grad_norm": 0.09364350885152817,
      "learning_rate": 0.0009931290080921861,
      "loss": 1.5913,
      "step": 5700
    },
    {
      "epoch": 0.25352363167489217,
      "grad_norm": 0.09758356213569641,
      "learning_rate": 0.0009931231726396116,
      "loss": 1.5885,
      "step": 5702
    },
    {
      "epoch": 0.25361255613356454,
      "grad_norm": 0.09728454798460007,
      "learning_rate": 0.000993117334727257,
      "loss": 1.5935,
      "step": 5704
    },
    {
      "epoch": 0.2537014805922369,
      "grad_norm": 0.10622499138116837,
      "learning_rate": 0.0009931114943551509,
      "loss": 1.5899,
      "step": 5706
    },
    {
      "epoch": 0.2537904050509093,
      "grad_norm": 0.10362113267183304,
      "learning_rate": 0.000993105651523323,
      "loss": 1.59,
      "step": 5708
    },
    {
      "epoch": 0.2538793295095816,
      "grad_norm": 0.10291391611099243,
      "learning_rate": 0.0009930998062318021,
      "loss": 1.5932,
      "step": 5710
    },
    {
      "epoch": 0.25396825396825395,
      "grad_norm": 0.09546540677547455,
      "learning_rate": 0.0009930939584806176,
      "loss": 1.5853,
      "step": 5712
    },
    {
      "epoch": 0.2540571784269263,
      "grad_norm": 0.0974823534488678,
      "learning_rate": 0.0009930881082697988,
      "loss": 1.59,
      "step": 5714
    },
    {
      "epoch": 0.2541461028855987,
      "grad_norm": 0.09429586678743362,
      "learning_rate": 0.0009930822555993743,
      "loss": 1.5893,
      "step": 5716
    },
    {
      "epoch": 0.25423502734427106,
      "grad_norm": 0.09460654854774475,
      "learning_rate": 0.0009930764004693739,
      "loss": 1.5912,
      "step": 5718
    },
    {
      "epoch": 0.2543239518029434,
      "grad_norm": 0.09382274746894836,
      "learning_rate": 0.0009930705428798266,
      "loss": 1.5889,
      "step": 5720
    },
    {
      "epoch": 0.25441287626161574,
      "grad_norm": 0.09548698365688324,
      "learning_rate": 0.0009930646828307614,
      "loss": 1.5866,
      "step": 5722
    },
    {
      "epoch": 0.2545018007202881,
      "grad_norm": 0.0983201414346695,
      "learning_rate": 0.000993058820322208,
      "loss": 1.5947,
      "step": 5724
    },
    {
      "epoch": 0.2545907251789605,
      "grad_norm": 0.09615219384431839,
      "learning_rate": 0.000993052955354195,
      "loss": 1.595,
      "step": 5726
    },
    {
      "epoch": 0.25467964963763284,
      "grad_norm": 0.09847582876682281,
      "learning_rate": 0.0009930470879267522,
      "loss": 1.5901,
      "step": 5728
    },
    {
      "epoch": 0.2547685740963052,
      "grad_norm": 0.09659074246883392,
      "learning_rate": 0.0009930412180399087,
      "loss": 1.5912,
      "step": 5730
    },
    {
      "epoch": 0.2548574985549775,
      "grad_norm": 0.09810440987348557,
      "learning_rate": 0.000993035345693694,
      "loss": 1.5906,
      "step": 5732
    },
    {
      "epoch": 0.2549464230136499,
      "grad_norm": 0.1003175675868988,
      "learning_rate": 0.0009930294708881367,
      "loss": 1.5884,
      "step": 5734
    },
    {
      "epoch": 0.25503534747232226,
      "grad_norm": 0.10160094499588013,
      "learning_rate": 0.0009930235936232669,
      "loss": 1.5865,
      "step": 5736
    },
    {
      "epoch": 0.25512427193099463,
      "grad_norm": 0.09863349050283432,
      "learning_rate": 0.0009930177138991131,
      "loss": 1.5936,
      "step": 5738
    },
    {
      "epoch": 0.255213196389667,
      "grad_norm": 0.08960582315921783,
      "learning_rate": 0.0009930118317157054,
      "loss": 1.5874,
      "step": 5740
    },
    {
      "epoch": 0.2553021208483393,
      "grad_norm": 0.09829938411712646,
      "learning_rate": 0.000993005947073073,
      "loss": 1.587,
      "step": 5742
    },
    {
      "epoch": 0.2553910453070117,
      "grad_norm": 0.09991574287414551,
      "learning_rate": 0.000993000059971245,
      "loss": 1.5867,
      "step": 5744
    },
    {
      "epoch": 0.25547996976568405,
      "grad_norm": 0.09275034070014954,
      "learning_rate": 0.0009929941704102509,
      "loss": 1.5933,
      "step": 5746
    },
    {
      "epoch": 0.2555688942243564,
      "grad_norm": 0.09592308104038239,
      "learning_rate": 0.00099298827839012,
      "loss": 1.5872,
      "step": 5748
    },
    {
      "epoch": 0.2556578186830288,
      "grad_norm": 0.0935816764831543,
      "learning_rate": 0.0009929823839108817,
      "loss": 1.5868,
      "step": 5750
    },
    {
      "epoch": 0.25574674314170115,
      "grad_norm": 0.09557216614484787,
      "learning_rate": 0.0009929764869725656,
      "loss": 1.5898,
      "step": 5752
    },
    {
      "epoch": 0.25583566760037346,
      "grad_norm": 0.0934915617108345,
      "learning_rate": 0.0009929705875752007,
      "loss": 1.5913,
      "step": 5754
    },
    {
      "epoch": 0.25592459205904583,
      "grad_norm": 0.09909486025571823,
      "learning_rate": 0.0009929646857188168,
      "loss": 1.5847,
      "step": 5756
    },
    {
      "epoch": 0.2560135165177182,
      "grad_norm": 0.09650136530399323,
      "learning_rate": 0.0009929587814034434,
      "loss": 1.5949,
      "step": 5758
    },
    {
      "epoch": 0.25610244097639057,
      "grad_norm": 0.10222198814153671,
      "learning_rate": 0.0009929528746291094,
      "loss": 1.5915,
      "step": 5760
    },
    {
      "epoch": 0.25619136543506293,
      "grad_norm": 0.10196855664253235,
      "learning_rate": 0.000992946965395845,
      "loss": 1.5848,
      "step": 5762
    },
    {
      "epoch": 0.25628028989373525,
      "grad_norm": 0.09607810527086258,
      "learning_rate": 0.000992941053703679,
      "loss": 1.5936,
      "step": 5764
    },
    {
      "epoch": 0.2563692143524076,
      "grad_norm": 0.09293419867753983,
      "learning_rate": 0.0009929351395526413,
      "loss": 1.5874,
      "step": 5766
    },
    {
      "epoch": 0.25645813881108,
      "grad_norm": 0.1013997420668602,
      "learning_rate": 0.0009929292229427615,
      "loss": 1.5858,
      "step": 5768
    },
    {
      "epoch": 0.25654706326975235,
      "grad_norm": 0.10565856844186783,
      "learning_rate": 0.0009929233038740686,
      "loss": 1.591,
      "step": 5770
    },
    {
      "epoch": 0.2566359877284247,
      "grad_norm": 0.09518007189035416,
      "learning_rate": 0.0009929173823465926,
      "loss": 1.5874,
      "step": 5772
    },
    {
      "epoch": 0.2567249121870971,
      "grad_norm": 0.09767907112836838,
      "learning_rate": 0.000992911458360363,
      "loss": 1.5891,
      "step": 5774
    },
    {
      "epoch": 0.2568138366457694,
      "grad_norm": 0.09295496344566345,
      "learning_rate": 0.0009929055319154092,
      "loss": 1.5859,
      "step": 5776
    },
    {
      "epoch": 0.25690276110444177,
      "grad_norm": 0.0951056256890297,
      "learning_rate": 0.0009928996030117605,
      "loss": 1.5898,
      "step": 5778
    },
    {
      "epoch": 0.25699168556311414,
      "grad_norm": 0.09819343686103821,
      "learning_rate": 0.000992893671649447,
      "loss": 1.5859,
      "step": 5780
    },
    {
      "epoch": 0.2570806100217865,
      "grad_norm": 0.09449407458305359,
      "learning_rate": 0.000992887737828498,
      "loss": 1.5935,
      "step": 5782
    },
    {
      "epoch": 0.2571695344804589,
      "grad_norm": 0.09632644057273865,
      "learning_rate": 0.0009928818015489432,
      "loss": 1.594,
      "step": 5784
    },
    {
      "epoch": 0.2572584589391312,
      "grad_norm": 0.09808379411697388,
      "learning_rate": 0.000992875862810812,
      "loss": 1.5954,
      "step": 5786
    },
    {
      "epoch": 0.25734738339780355,
      "grad_norm": 0.09704041481018066,
      "learning_rate": 0.0009928699216141344,
      "loss": 1.5884,
      "step": 5788
    },
    {
      "epoch": 0.2574363078564759,
      "grad_norm": 0.09470924735069275,
      "learning_rate": 0.0009928639779589396,
      "loss": 1.5833,
      "step": 5790
    },
    {
      "epoch": 0.2575252323151483,
      "grad_norm": 0.09412840753793716,
      "learning_rate": 0.0009928580318452577,
      "loss": 1.5889,
      "step": 5792
    },
    {
      "epoch": 0.25761415677382066,
      "grad_norm": 0.09567838162183762,
      "learning_rate": 0.000992852083273118,
      "loss": 1.5891,
      "step": 5794
    },
    {
      "epoch": 0.25770308123249297,
      "grad_norm": 0.10092340409755707,
      "learning_rate": 0.0009928461322425502,
      "loss": 1.5908,
      "step": 5796
    },
    {
      "epoch": 0.25779200569116534,
      "grad_norm": 0.10121021419763565,
      "learning_rate": 0.0009928401787535842,
      "loss": 1.5911,
      "step": 5798
    },
    {
      "epoch": 0.2578809301498377,
      "grad_norm": 0.09260232001543045,
      "learning_rate": 0.0009928342228062497,
      "loss": 1.5844,
      "step": 5800
    },
    {
      "epoch": 0.2579698546085101,
      "grad_norm": 0.1004943922162056,
      "learning_rate": 0.0009928282644005762,
      "loss": 1.5827,
      "step": 5802
    },
    {
      "epoch": 0.25805877906718244,
      "grad_norm": 0.09976594150066376,
      "learning_rate": 0.0009928223035365934,
      "loss": 1.5867,
      "step": 5804
    },
    {
      "epoch": 0.2581477035258548,
      "grad_norm": 0.09819846600294113,
      "learning_rate": 0.0009928163402143312,
      "loss": 1.5836,
      "step": 5806
    },
    {
      "epoch": 0.2582366279845271,
      "grad_norm": 0.09671387076377869,
      "learning_rate": 0.0009928103744338192,
      "loss": 1.584,
      "step": 5808
    },
    {
      "epoch": 0.2583255524431995,
      "grad_norm": 0.09667721390724182,
      "learning_rate": 0.0009928044061950875,
      "loss": 1.5886,
      "step": 5810
    },
    {
      "epoch": 0.25841447690187186,
      "grad_norm": 0.09693595767021179,
      "learning_rate": 0.0009927984354981653,
      "loss": 1.5871,
      "step": 5812
    },
    {
      "epoch": 0.2585034013605442,
      "grad_norm": 0.08862930536270142,
      "learning_rate": 0.000992792462343083,
      "loss": 1.5902,
      "step": 5814
    },
    {
      "epoch": 0.2585923258192166,
      "grad_norm": 0.09225138276815414,
      "learning_rate": 0.00099278648672987,
      "loss": 1.5824,
      "step": 5816
    },
    {
      "epoch": 0.2586812502778889,
      "grad_norm": 0.09490638971328735,
      "learning_rate": 0.0009927805086585562,
      "loss": 1.5793,
      "step": 5818
    },
    {
      "epoch": 0.2587701747365613,
      "grad_norm": 0.09594158828258514,
      "learning_rate": 0.0009927745281291713,
      "loss": 1.583,
      "step": 5820
    },
    {
      "epoch": 0.25885909919523364,
      "grad_norm": 0.09334473311901093,
      "learning_rate": 0.0009927685451417453,
      "loss": 1.5852,
      "step": 5822
    },
    {
      "epoch": 0.258948023653906,
      "grad_norm": 0.09541244804859161,
      "learning_rate": 0.0009927625596963083,
      "loss": 1.59,
      "step": 5824
    },
    {
      "epoch": 0.2590369481125784,
      "grad_norm": 0.09415993094444275,
      "learning_rate": 0.0009927565717928895,
      "loss": 1.5866,
      "step": 5826
    },
    {
      "epoch": 0.25912587257125075,
      "grad_norm": 0.09244725853204727,
      "learning_rate": 0.0009927505814315194,
      "loss": 1.5902,
      "step": 5828
    },
    {
      "epoch": 0.25921479702992306,
      "grad_norm": 0.09592825919389725,
      "learning_rate": 0.0009927445886122276,
      "loss": 1.5915,
      "step": 5830
    },
    {
      "epoch": 0.25930372148859543,
      "grad_norm": 0.0941583588719368,
      "learning_rate": 0.0009927385933350436,
      "loss": 1.589,
      "step": 5832
    },
    {
      "epoch": 0.2593926459472678,
      "grad_norm": 0.09573867917060852,
      "learning_rate": 0.000992732595599998,
      "loss": 1.587,
      "step": 5834
    },
    {
      "epoch": 0.25948157040594017,
      "grad_norm": 0.09302788972854614,
      "learning_rate": 0.0009927265954071205,
      "loss": 1.589,
      "step": 5836
    },
    {
      "epoch": 0.25957049486461253,
      "grad_norm": 0.09388726204633713,
      "learning_rate": 0.0009927205927564408,
      "loss": 1.5858,
      "step": 5838
    },
    {
      "epoch": 0.25965941932328485,
      "grad_norm": 0.08946000784635544,
      "learning_rate": 0.000992714587647989,
      "loss": 1.5815,
      "step": 5840
    },
    {
      "epoch": 0.2597483437819572,
      "grad_norm": 0.09015568345785141,
      "learning_rate": 0.000992708580081795,
      "loss": 1.5907,
      "step": 5842
    },
    {
      "epoch": 0.2598372682406296,
      "grad_norm": 0.09112632274627686,
      "learning_rate": 0.0009927025700578892,
      "loss": 1.5878,
      "step": 5844
    },
    {
      "epoch": 0.25992619269930195,
      "grad_norm": 0.0954042598605156,
      "learning_rate": 0.0009926965575763008,
      "loss": 1.5831,
      "step": 5846
    },
    {
      "epoch": 0.2600151171579743,
      "grad_norm": 0.09618160128593445,
      "learning_rate": 0.0009926905426370604,
      "loss": 1.588,
      "step": 5848
    },
    {
      "epoch": 0.2601040416166467,
      "grad_norm": 0.09728830307722092,
      "learning_rate": 0.0009926845252401974,
      "loss": 1.5892,
      "step": 5850
    },
    {
      "epoch": 0.260192966075319,
      "grad_norm": 0.09168002754449844,
      "learning_rate": 0.0009926785053857425,
      "loss": 1.5904,
      "step": 5852
    },
    {
      "epoch": 0.26028189053399137,
      "grad_norm": 0.09341517090797424,
      "learning_rate": 0.0009926724830737256,
      "loss": 1.5891,
      "step": 5854
    },
    {
      "epoch": 0.26037081499266373,
      "grad_norm": 0.09883091598749161,
      "learning_rate": 0.0009926664583041764,
      "loss": 1.5898,
      "step": 5856
    },
    {
      "epoch": 0.2604597394513361,
      "grad_norm": 0.0963706523180008,
      "learning_rate": 0.000992660431077125,
      "loss": 1.5869,
      "step": 5858
    },
    {
      "epoch": 0.26054866391000847,
      "grad_norm": 0.09510736167430878,
      "learning_rate": 0.0009926544013926018,
      "loss": 1.5886,
      "step": 5860
    },
    {
      "epoch": 0.2606375883686808,
      "grad_norm": 0.09609393775463104,
      "learning_rate": 0.0009926483692506364,
      "loss": 1.59,
      "step": 5862
    },
    {
      "epoch": 0.26072651282735315,
      "grad_norm": 0.10332238674163818,
      "learning_rate": 0.0009926423346512596,
      "loss": 1.5906,
      "step": 5864
    },
    {
      "epoch": 0.2608154372860255,
      "grad_norm": 0.09607810527086258,
      "learning_rate": 0.0009926362975945008,
      "loss": 1.5821,
      "step": 5866
    },
    {
      "epoch": 0.2609043617446979,
      "grad_norm": 0.09439858794212341,
      "learning_rate": 0.0009926302580803904,
      "loss": 1.5893,
      "step": 5868
    },
    {
      "epoch": 0.26099328620337026,
      "grad_norm": 0.09515378624200821,
      "learning_rate": 0.0009926242161089583,
      "loss": 1.5872,
      "step": 5870
    },
    {
      "epoch": 0.26108221066204257,
      "grad_norm": 0.09659688174724579,
      "learning_rate": 0.0009926181716802351,
      "loss": 1.5856,
      "step": 5872
    },
    {
      "epoch": 0.26117113512071494,
      "grad_norm": 0.09571564942598343,
      "learning_rate": 0.0009926121247942506,
      "loss": 1.583,
      "step": 5874
    },
    {
      "epoch": 0.2612600595793873,
      "grad_norm": 0.09830082207918167,
      "learning_rate": 0.000992606075451035,
      "loss": 1.5874,
      "step": 5876
    },
    {
      "epoch": 0.2613489840380597,
      "grad_norm": 0.09110367298126221,
      "learning_rate": 0.0009926000236506187,
      "loss": 1.5869,
      "step": 5878
    },
    {
      "epoch": 0.26143790849673204,
      "grad_norm": 0.09880504012107849,
      "learning_rate": 0.0009925939693930315,
      "loss": 1.5823,
      "step": 5880
    },
    {
      "epoch": 0.2615268329554044,
      "grad_norm": 0.09947524964809418,
      "learning_rate": 0.0009925879126783039,
      "loss": 1.5878,
      "step": 5882
    },
    {
      "epoch": 0.2616157574140767,
      "grad_norm": 0.0937243178486824,
      "learning_rate": 0.0009925818535064662,
      "loss": 1.583,
      "step": 5884
    },
    {
      "epoch": 0.2617046818727491,
      "grad_norm": 0.0913049653172493,
      "learning_rate": 0.0009925757918775482,
      "loss": 1.5867,
      "step": 5886
    },
    {
      "epoch": 0.26179360633142146,
      "grad_norm": 0.09564990550279617,
      "learning_rate": 0.0009925697277915805,
      "loss": 1.5917,
      "step": 5888
    },
    {
      "epoch": 0.2618825307900938,
      "grad_norm": 0.09640397876501083,
      "learning_rate": 0.0009925636612485931,
      "loss": 1.5818,
      "step": 5890
    },
    {
      "epoch": 0.2619714552487662,
      "grad_norm": 0.09462813287973404,
      "learning_rate": 0.0009925575922486167,
      "loss": 1.5867,
      "step": 5892
    },
    {
      "epoch": 0.2620603797074385,
      "grad_norm": 0.10179334878921509,
      "learning_rate": 0.000992551520791681,
      "loss": 1.5877,
      "step": 5894
    },
    {
      "epoch": 0.2621493041661109,
      "grad_norm": 0.09697990119457245,
      "learning_rate": 0.0009925454468778167,
      "loss": 1.5843,
      "step": 5896
    },
    {
      "epoch": 0.26223822862478324,
      "grad_norm": 0.09205523133277893,
      "learning_rate": 0.0009925393705070538,
      "loss": 1.5882,
      "step": 5898
    },
    {
      "epoch": 0.2623271530834556,
      "grad_norm": 0.09468796104192734,
      "learning_rate": 0.0009925332916794229,
      "loss": 1.593,
      "step": 5900
    },
    {
      "epoch": 0.262416077542128,
      "grad_norm": 0.09811273962259293,
      "learning_rate": 0.000992527210394954,
      "loss": 1.5872,
      "step": 5902
    },
    {
      "epoch": 0.26250500200080035,
      "grad_norm": 0.09203442186117172,
      "learning_rate": 0.000992521126653678,
      "loss": 1.5857,
      "step": 5904
    },
    {
      "epoch": 0.26259392645947266,
      "grad_norm": 0.10104124248027802,
      "learning_rate": 0.0009925150404556245,
      "loss": 1.5929,
      "step": 5906
    },
    {
      "epoch": 0.262682850918145,
      "grad_norm": 0.10001523047685623,
      "learning_rate": 0.0009925089518008244,
      "loss": 1.5829,
      "step": 5908
    },
    {
      "epoch": 0.2627717753768174,
      "grad_norm": 0.10386830568313599,
      "learning_rate": 0.000992502860689308,
      "loss": 1.5874,
      "step": 5910
    },
    {
      "epoch": 0.26286069983548976,
      "grad_norm": 0.09889882802963257,
      "learning_rate": 0.0009924967671211052,
      "loss": 1.5876,
      "step": 5912
    },
    {
      "epoch": 0.26294962429416213,
      "grad_norm": 0.09799201786518097,
      "learning_rate": 0.000992490671096247,
      "loss": 1.5832,
      "step": 5914
    },
    {
      "epoch": 0.26303854875283444,
      "grad_norm": 0.10036963224411011,
      "learning_rate": 0.0009924845726147636,
      "loss": 1.5868,
      "step": 5916
    },
    {
      "epoch": 0.2631274732115068,
      "grad_norm": 0.09943078458309174,
      "learning_rate": 0.0009924784716766853,
      "loss": 1.5812,
      "step": 5918
    },
    {
      "epoch": 0.2632163976701792,
      "grad_norm": 0.0965849757194519,
      "learning_rate": 0.0009924723682820428,
      "loss": 1.5822,
      "step": 5920
    },
    {
      "epoch": 0.26330532212885155,
      "grad_norm": 0.10785552114248276,
      "learning_rate": 0.0009924662624308663,
      "loss": 1.5856,
      "step": 5922
    },
    {
      "epoch": 0.2633942465875239,
      "grad_norm": 0.09936713427305222,
      "learning_rate": 0.0009924601541231863,
      "loss": 1.5891,
      "step": 5924
    },
    {
      "epoch": 0.26348317104619623,
      "grad_norm": 0.1005997359752655,
      "learning_rate": 0.0009924540433590334,
      "loss": 1.5896,
      "step": 5926
    },
    {
      "epoch": 0.2635720955048686,
      "grad_norm": 0.0992506816983223,
      "learning_rate": 0.0009924479301384379,
      "loss": 1.5872,
      "step": 5928
    },
    {
      "epoch": 0.26366101996354097,
      "grad_norm": 0.09371292591094971,
      "learning_rate": 0.0009924418144614305,
      "loss": 1.5879,
      "step": 5930
    },
    {
      "epoch": 0.26374994442221333,
      "grad_norm": 0.09433597326278687,
      "learning_rate": 0.0009924356963280415,
      "loss": 1.5871,
      "step": 5932
    },
    {
      "epoch": 0.2638388688808857,
      "grad_norm": 0.08891277760267258,
      "learning_rate": 0.0009924295757383014,
      "loss": 1.583,
      "step": 5934
    },
    {
      "epoch": 0.26392779333955807,
      "grad_norm": 0.09156538546085358,
      "learning_rate": 0.000992423452692241,
      "loss": 1.5958,
      "step": 5936
    },
    {
      "epoch": 0.2640167177982304,
      "grad_norm": 0.09960367530584335,
      "learning_rate": 0.0009924173271898906,
      "loss": 1.5829,
      "step": 5938
    },
    {
      "epoch": 0.26410564225690275,
      "grad_norm": 0.09040109068155289,
      "learning_rate": 0.000992411199231281,
      "loss": 1.5869,
      "step": 5940
    },
    {
      "epoch": 0.2641945667155751,
      "grad_norm": 0.09623978286981583,
      "learning_rate": 0.0009924050688164424,
      "loss": 1.5814,
      "step": 5942
    },
    {
      "epoch": 0.2642834911742475,
      "grad_norm": 0.09259607642889023,
      "learning_rate": 0.0009923989359454056,
      "loss": 1.5886,
      "step": 5944
    },
    {
      "epoch": 0.26437241563291985,
      "grad_norm": 0.09638094156980515,
      "learning_rate": 0.0009923928006182012,
      "loss": 1.5863,
      "step": 5946
    },
    {
      "epoch": 0.26446134009159217,
      "grad_norm": 0.09382667392492294,
      "learning_rate": 0.0009923866628348597,
      "loss": 1.582,
      "step": 5948
    },
    {
      "epoch": 0.26455026455026454,
      "grad_norm": 0.09935595840215683,
      "learning_rate": 0.000992380522595412,
      "loss": 1.585,
      "step": 5950
    },
    {
      "epoch": 0.2646391890089369,
      "grad_norm": 0.0950867235660553,
      "learning_rate": 0.0009923743798998883,
      "loss": 1.5854,
      "step": 5952
    },
    {
      "epoch": 0.26472811346760927,
      "grad_norm": 0.09334602952003479,
      "learning_rate": 0.0009923682347483195,
      "loss": 1.5831,
      "step": 5954
    },
    {
      "epoch": 0.26481703792628164,
      "grad_norm": 0.09344839304685593,
      "learning_rate": 0.0009923620871407364,
      "loss": 1.5826,
      "step": 5956
    },
    {
      "epoch": 0.264905962384954,
      "grad_norm": 0.0958920270204544,
      "learning_rate": 0.0009923559370771694,
      "loss": 1.5838,
      "step": 5958
    },
    {
      "epoch": 0.2649948868436263,
      "grad_norm": 0.09009189903736115,
      "learning_rate": 0.0009923497845576493,
      "loss": 1.5878,
      "step": 5960
    },
    {
      "epoch": 0.2650838113022987,
      "grad_norm": 0.0952090173959732,
      "learning_rate": 0.0009923436295822066,
      "loss": 1.577,
      "step": 5962
    },
    {
      "epoch": 0.26517273576097106,
      "grad_norm": 0.08939055353403091,
      "learning_rate": 0.0009923374721508723,
      "loss": 1.578,
      "step": 5964
    },
    {
      "epoch": 0.2652616602196434,
      "grad_norm": 0.08977833390235901,
      "learning_rate": 0.0009923313122636769,
      "loss": 1.5773,
      "step": 5966
    },
    {
      "epoch": 0.2653505846783158,
      "grad_norm": 0.09042993932962418,
      "learning_rate": 0.000992325149920651,
      "loss": 1.5859,
      "step": 5968
    },
    {
      "epoch": 0.2654395091369881,
      "grad_norm": 0.0930049791932106,
      "learning_rate": 0.0009923189851218258,
      "loss": 1.5909,
      "step": 5970
    },
    {
      "epoch": 0.2655284335956605,
      "grad_norm": 0.09311951696872711,
      "learning_rate": 0.0009923128178672318,
      "loss": 1.5886,
      "step": 5972
    },
    {
      "epoch": 0.26561735805433284,
      "grad_norm": 0.09062470495700836,
      "learning_rate": 0.0009923066481568996,
      "loss": 1.5856,
      "step": 5974
    },
    {
      "epoch": 0.2657062825130052,
      "grad_norm": 0.09528020024299622,
      "learning_rate": 0.0009923004759908602,
      "loss": 1.5858,
      "step": 5976
    },
    {
      "epoch": 0.2657952069716776,
      "grad_norm": 0.09330100566148758,
      "learning_rate": 0.0009922943013691442,
      "loss": 1.5919,
      "step": 5978
    },
    {
      "epoch": 0.2658841314303499,
      "grad_norm": 0.09436290711164474,
      "learning_rate": 0.0009922881242917824,
      "loss": 1.5807,
      "step": 5980
    },
    {
      "epoch": 0.26597305588902226,
      "grad_norm": 0.09486601501703262,
      "learning_rate": 0.0009922819447588061,
      "loss": 1.5844,
      "step": 5982
    },
    {
      "epoch": 0.2660619803476946,
      "grad_norm": 0.09641608595848083,
      "learning_rate": 0.0009922757627702455,
      "loss": 1.5822,
      "step": 5984
    },
    {
      "epoch": 0.266150904806367,
      "grad_norm": 0.096797876060009,
      "learning_rate": 0.0009922695783261318,
      "loss": 1.5819,
      "step": 5986
    },
    {
      "epoch": 0.26623982926503936,
      "grad_norm": 0.09238893538713455,
      "learning_rate": 0.0009922633914264956,
      "loss": 1.5795,
      "step": 5988
    },
    {
      "epoch": 0.26632875372371173,
      "grad_norm": 0.09049165993928909,
      "learning_rate": 0.000992257202071368,
      "loss": 1.5835,
      "step": 5990
    },
    {
      "epoch": 0.26641767818238404,
      "grad_norm": 0.09518662840127945,
      "learning_rate": 0.0009922510102607796,
      "loss": 1.5844,
      "step": 5992
    },
    {
      "epoch": 0.2665066026410564,
      "grad_norm": 0.09077850729227066,
      "learning_rate": 0.0009922448159947618,
      "loss": 1.585,
      "step": 5994
    },
    {
      "epoch": 0.2665955270997288,
      "grad_norm": 0.09498831629753113,
      "learning_rate": 0.0009922386192733447,
      "loss": 1.5779,
      "step": 5996
    },
    {
      "epoch": 0.26668445155840115,
      "grad_norm": 0.09286242723464966,
      "learning_rate": 0.00099223242009656,
      "loss": 1.5788,
      "step": 5998
    },
    {
      "epoch": 0.2667733760170735,
      "grad_norm": 0.09513460099697113,
      "learning_rate": 0.000992226218464438,
      "loss": 1.5926,
      "step": 6000
    },
    {
      "epoch": 0.2667733760170735,
      "eval_loss": 1.5609192848205566,
      "eval_runtime": 12.357,
      "eval_samples_per_second": 559.198,
      "eval_steps_per_second": 69.92,
      "step": 6000
    },
    {
      "epoch": 0.26686230047574583,
      "grad_norm": 0.09536412358283997,
      "learning_rate": 0.0009922200143770103,
      "loss": 1.5798,
      "step": 6002
    },
    {
      "epoch": 0.2669512249344182,
      "grad_norm": 0.0923578068614006,
      "learning_rate": 0.000992213807834307,
      "loss": 1.5833,
      "step": 6004
    },
    {
      "epoch": 0.26704014939309056,
      "grad_norm": 0.0896889865398407,
      "learning_rate": 0.00099220759883636,
      "loss": 1.5824,
      "step": 6006
    },
    {
      "epoch": 0.26712907385176293,
      "grad_norm": 0.0965195745229721,
      "learning_rate": 0.0009922013873831994,
      "loss": 1.5842,
      "step": 6008
    },
    {
      "epoch": 0.2672179983104353,
      "grad_norm": 0.09298788011074066,
      "learning_rate": 0.0009921951734748566,
      "loss": 1.575,
      "step": 6010
    },
    {
      "epoch": 0.26730692276910767,
      "grad_norm": 0.0938933789730072,
      "learning_rate": 0.0009921889571113628,
      "loss": 1.5787,
      "step": 6012
    },
    {
      "epoch": 0.26739584722778,
      "grad_norm": 0.09369226545095444,
      "learning_rate": 0.0009921827382927486,
      "loss": 1.5808,
      "step": 6014
    },
    {
      "epoch": 0.26748477168645235,
      "grad_norm": 0.09126924723386765,
      "learning_rate": 0.0009921765170190454,
      "loss": 1.5752,
      "step": 6016
    },
    {
      "epoch": 0.2675736961451247,
      "grad_norm": 0.09294477850198746,
      "learning_rate": 0.0009921702932902839,
      "loss": 1.5761,
      "step": 6018
    },
    {
      "epoch": 0.2676626206037971,
      "grad_norm": 0.09301607310771942,
      "learning_rate": 0.0009921640671064953,
      "loss": 1.5772,
      "step": 6020
    },
    {
      "epoch": 0.26775154506246945,
      "grad_norm": 0.09419137239456177,
      "learning_rate": 0.0009921578384677106,
      "loss": 1.5782,
      "step": 6022
    },
    {
      "epoch": 0.26784046952114177,
      "grad_norm": 0.09414952993392944,
      "learning_rate": 0.000992151607373961,
      "loss": 1.5785,
      "step": 6024
    },
    {
      "epoch": 0.26792939397981413,
      "grad_norm": 0.09365058690309525,
      "learning_rate": 0.0009921453738252774,
      "loss": 1.584,
      "step": 6026
    },
    {
      "epoch": 0.2680183184384865,
      "grad_norm": 0.09278295934200287,
      "learning_rate": 0.000992139137821691,
      "loss": 1.5855,
      "step": 6028
    },
    {
      "epoch": 0.26810724289715887,
      "grad_norm": 0.09213846176862717,
      "learning_rate": 0.000992132899363233,
      "loss": 1.5798,
      "step": 6030
    },
    {
      "epoch": 0.26819616735583124,
      "grad_norm": 0.0945931226015091,
      "learning_rate": 0.0009921266584499344,
      "loss": 1.5792,
      "step": 6032
    },
    {
      "epoch": 0.2682850918145036,
      "grad_norm": 0.09439346194267273,
      "learning_rate": 0.000992120415081826,
      "loss": 1.5722,
      "step": 6034
    },
    {
      "epoch": 0.2683740162731759,
      "grad_norm": 0.09740027040243149,
      "learning_rate": 0.0009921141692589396,
      "loss": 1.582,
      "step": 6036
    },
    {
      "epoch": 0.2684629407318483,
      "grad_norm": 0.0935557633638382,
      "learning_rate": 0.0009921079209813063,
      "loss": 1.581,
      "step": 6038
    },
    {
      "epoch": 0.26855186519052066,
      "grad_norm": 0.09402929991483688,
      "learning_rate": 0.0009921016702489564,
      "loss": 1.5772,
      "step": 6040
    },
    {
      "epoch": 0.268640789649193,
      "grad_norm": 0.09664035588502884,
      "learning_rate": 0.000992095417061922,
      "loss": 1.5807,
      "step": 6042
    },
    {
      "epoch": 0.2687297141078654,
      "grad_norm": 0.09566997736692429,
      "learning_rate": 0.000992089161420234,
      "loss": 1.5822,
      "step": 6044
    },
    {
      "epoch": 0.2688186385665377,
      "grad_norm": 0.09569090604782104,
      "learning_rate": 0.0009920829033239235,
      "loss": 1.5799,
      "step": 6046
    },
    {
      "epoch": 0.2689075630252101,
      "grad_norm": 0.09623146057128906,
      "learning_rate": 0.0009920766427730218,
      "loss": 1.5803,
      "step": 6048
    },
    {
      "epoch": 0.26899648748388244,
      "grad_norm": 0.09390247613191605,
      "learning_rate": 0.0009920703797675603,
      "loss": 1.58,
      "step": 6050
    },
    {
      "epoch": 0.2690854119425548,
      "grad_norm": 0.09540102630853653,
      "learning_rate": 0.0009920641143075699,
      "loss": 1.5818,
      "step": 6052
    },
    {
      "epoch": 0.2691743364012272,
      "grad_norm": 0.09371043741703033,
      "learning_rate": 0.000992057846393082,
      "loss": 1.5885,
      "step": 6054
    },
    {
      "epoch": 0.2692632608598995,
      "grad_norm": 0.09126854687929153,
      "learning_rate": 0.000992051576024128,
      "loss": 1.5794,
      "step": 6056
    },
    {
      "epoch": 0.26935218531857186,
      "grad_norm": 0.09539595991373062,
      "learning_rate": 0.0009920453032007388,
      "loss": 1.5784,
      "step": 6058
    },
    {
      "epoch": 0.2694411097772442,
      "grad_norm": 0.09004338830709457,
      "learning_rate": 0.0009920390279229462,
      "loss": 1.5753,
      "step": 6060
    },
    {
      "epoch": 0.2695300342359166,
      "grad_norm": 0.09474993497133255,
      "learning_rate": 0.0009920327501907811,
      "loss": 1.5854,
      "step": 6062
    },
    {
      "epoch": 0.26961895869458896,
      "grad_norm": 0.10050513595342636,
      "learning_rate": 0.000992026470004275,
      "loss": 1.5815,
      "step": 6064
    },
    {
      "epoch": 0.26970788315326133,
      "grad_norm": 0.09140043705701828,
      "learning_rate": 0.0009920201873634593,
      "loss": 1.58,
      "step": 6066
    },
    {
      "epoch": 0.26979680761193364,
      "grad_norm": 0.09281838685274124,
      "learning_rate": 0.0009920139022683652,
      "loss": 1.5788,
      "step": 6068
    },
    {
      "epoch": 0.269885732070606,
      "grad_norm": 0.0929790586233139,
      "learning_rate": 0.0009920076147190242,
      "loss": 1.5826,
      "step": 6070
    },
    {
      "epoch": 0.2699746565292784,
      "grad_norm": 0.09214480966329575,
      "learning_rate": 0.0009920013247154674,
      "loss": 1.5772,
      "step": 6072
    },
    {
      "epoch": 0.27006358098795075,
      "grad_norm": 0.09295648336410522,
      "learning_rate": 0.0009919950322577263,
      "loss": 1.5797,
      "step": 6074
    },
    {
      "epoch": 0.2701525054466231,
      "grad_norm": 0.094325952231884,
      "learning_rate": 0.0009919887373458325,
      "loss": 1.586,
      "step": 6076
    },
    {
      "epoch": 0.2702414299052954,
      "grad_norm": 0.09574458748102188,
      "learning_rate": 0.0009919824399798169,
      "loss": 1.5852,
      "step": 6078
    },
    {
      "epoch": 0.2703303543639678,
      "grad_norm": 0.09337875992059708,
      "learning_rate": 0.0009919761401597114,
      "loss": 1.5748,
      "step": 6080
    },
    {
      "epoch": 0.27041927882264016,
      "grad_norm": 0.09532225131988525,
      "learning_rate": 0.0009919698378855473,
      "loss": 1.5696,
      "step": 6082
    },
    {
      "epoch": 0.27050820328131253,
      "grad_norm": 0.09588050097227097,
      "learning_rate": 0.000991963533157356,
      "loss": 1.585,
      "step": 6084
    },
    {
      "epoch": 0.2705971277399849,
      "grad_norm": 0.09399744123220444,
      "learning_rate": 0.0009919572259751688,
      "loss": 1.5727,
      "step": 6086
    },
    {
      "epoch": 0.27068605219865727,
      "grad_norm": 0.09603196382522583,
      "learning_rate": 0.0009919509163390174,
      "loss": 1.5792,
      "step": 6088
    },
    {
      "epoch": 0.2707749766573296,
      "grad_norm": 0.09013631939888,
      "learning_rate": 0.000991944604248933,
      "loss": 1.5795,
      "step": 6090
    },
    {
      "epoch": 0.27086390111600195,
      "grad_norm": 0.08968756347894669,
      "learning_rate": 0.0009919382897049475,
      "loss": 1.574,
      "step": 6092
    },
    {
      "epoch": 0.2709528255746743,
      "grad_norm": 0.090090312063694,
      "learning_rate": 0.0009919319727070919,
      "loss": 1.5763,
      "step": 6094
    },
    {
      "epoch": 0.2710417500333467,
      "grad_norm": 0.09772195667028427,
      "learning_rate": 0.0009919256532553982,
      "loss": 1.5812,
      "step": 6096
    },
    {
      "epoch": 0.27113067449201905,
      "grad_norm": 0.09192989021539688,
      "learning_rate": 0.0009919193313498976,
      "loss": 1.5723,
      "step": 6098
    },
    {
      "epoch": 0.27121959895069137,
      "grad_norm": 0.091171033680439,
      "learning_rate": 0.0009919130069906217,
      "loss": 1.5751,
      "step": 6100
    },
    {
      "epoch": 0.27130852340936373,
      "grad_norm": 0.0916176363825798,
      "learning_rate": 0.000991906680177602,
      "loss": 1.5787,
      "step": 6102
    },
    {
      "epoch": 0.2713974478680361,
      "grad_norm": 0.09237353503704071,
      "learning_rate": 0.0009919003509108702,
      "loss": 1.5746,
      "step": 6104
    },
    {
      "epoch": 0.27148637232670847,
      "grad_norm": 0.09654420614242554,
      "learning_rate": 0.0009918940191904577,
      "loss": 1.5804,
      "step": 6106
    },
    {
      "epoch": 0.27157529678538084,
      "grad_norm": 0.09240265935659409,
      "learning_rate": 0.0009918876850163966,
      "loss": 1.5769,
      "step": 6108
    },
    {
      "epoch": 0.27166422124405315,
      "grad_norm": 0.09678072482347488,
      "learning_rate": 0.0009918813483887176,
      "loss": 1.5787,
      "step": 6110
    },
    {
      "epoch": 0.2717531457027255,
      "grad_norm": 0.09163809567689896,
      "learning_rate": 0.000991875009307453,
      "loss": 1.581,
      "step": 6112
    },
    {
      "epoch": 0.2718420701613979,
      "grad_norm": 0.09179195761680603,
      "learning_rate": 0.000991868667772634,
      "loss": 1.5799,
      "step": 6114
    },
    {
      "epoch": 0.27193099462007025,
      "grad_norm": 0.09309026598930359,
      "learning_rate": 0.0009918623237842926,
      "loss": 1.5749,
      "step": 6116
    },
    {
      "epoch": 0.2720199190787426,
      "grad_norm": 0.09055127948522568,
      "learning_rate": 0.0009918559773424603,
      "loss": 1.5776,
      "step": 6118
    },
    {
      "epoch": 0.272108843537415,
      "grad_norm": 0.09597539901733398,
      "learning_rate": 0.0009918496284471688,
      "loss": 1.5733,
      "step": 6120
    },
    {
      "epoch": 0.2721977679960873,
      "grad_norm": 0.09350308775901794,
      "learning_rate": 0.0009918432770984495,
      "loss": 1.5775,
      "step": 6122
    },
    {
      "epoch": 0.27228669245475967,
      "grad_norm": 0.09235924482345581,
      "learning_rate": 0.0009918369232963344,
      "loss": 1.5705,
      "step": 6124
    },
    {
      "epoch": 0.27237561691343204,
      "grad_norm": 0.09772805869579315,
      "learning_rate": 0.0009918305670408552,
      "loss": 1.5795,
      "step": 6126
    },
    {
      "epoch": 0.2724645413721044,
      "grad_norm": 0.09387152642011642,
      "learning_rate": 0.0009918242083320433,
      "loss": 1.5857,
      "step": 6128
    },
    {
      "epoch": 0.2725534658307768,
      "grad_norm": 0.08818736672401428,
      "learning_rate": 0.0009918178471699309,
      "loss": 1.5767,
      "step": 6130
    },
    {
      "epoch": 0.2726423902894491,
      "grad_norm": 0.09092088788747787,
      "learning_rate": 0.000991811483554549,
      "loss": 1.5775,
      "step": 6132
    },
    {
      "epoch": 0.27273131474812146,
      "grad_norm": 0.09008492529392242,
      "learning_rate": 0.0009918051174859302,
      "loss": 1.5767,
      "step": 6134
    },
    {
      "epoch": 0.2728202392067938,
      "grad_norm": 0.0906415656208992,
      "learning_rate": 0.0009917987489641057,
      "loss": 1.5746,
      "step": 6136
    },
    {
      "epoch": 0.2729091636654662,
      "grad_norm": 0.09450238943099976,
      "learning_rate": 0.0009917923779891073,
      "loss": 1.5732,
      "step": 6138
    },
    {
      "epoch": 0.27299808812413856,
      "grad_norm": 0.08837480843067169,
      "learning_rate": 0.000991786004560967,
      "loss": 1.5756,
      "step": 6140
    },
    {
      "epoch": 0.27308701258281093,
      "grad_norm": 0.09205223619937897,
      "learning_rate": 0.0009917796286797163,
      "loss": 1.5763,
      "step": 6142
    },
    {
      "epoch": 0.27317593704148324,
      "grad_norm": 0.0925694927573204,
      "learning_rate": 0.0009917732503453873,
      "loss": 1.5766,
      "step": 6144
    },
    {
      "epoch": 0.2732648615001556,
      "grad_norm": 0.09327147156000137,
      "learning_rate": 0.0009917668695580118,
      "loss": 1.5742,
      "step": 6146
    },
    {
      "epoch": 0.273353785958828,
      "grad_norm": 0.09639228135347366,
      "learning_rate": 0.0009917604863176215,
      "loss": 1.5805,
      "step": 6148
    },
    {
      "epoch": 0.27344271041750035,
      "grad_norm": 0.09437435865402222,
      "learning_rate": 0.0009917541006242482,
      "loss": 1.5835,
      "step": 6150
    },
    {
      "epoch": 0.2735316348761727,
      "grad_norm": 0.09421025216579437,
      "learning_rate": 0.0009917477124779237,
      "loss": 1.5827,
      "step": 6152
    },
    {
      "epoch": 0.273620559334845,
      "grad_norm": 0.09246162325143814,
      "learning_rate": 0.0009917413218786802,
      "loss": 1.5771,
      "step": 6154
    },
    {
      "epoch": 0.2737094837935174,
      "grad_norm": 0.0891774520277977,
      "learning_rate": 0.0009917349288265493,
      "loss": 1.5819,
      "step": 6156
    },
    {
      "epoch": 0.27379840825218976,
      "grad_norm": 0.09381581842899323,
      "learning_rate": 0.0009917285333215628,
      "loss": 1.5758,
      "step": 6158
    },
    {
      "epoch": 0.27388733271086213,
      "grad_norm": 0.09442843496799469,
      "learning_rate": 0.000991722135363753,
      "loss": 1.5837,
      "step": 6160
    },
    {
      "epoch": 0.2739762571695345,
      "grad_norm": 0.08775163441896439,
      "learning_rate": 0.0009917157349531513,
      "loss": 1.581,
      "step": 6162
    },
    {
      "epoch": 0.2740651816282068,
      "grad_norm": 0.0929328128695488,
      "learning_rate": 0.00099170933208979,
      "loss": 1.5758,
      "step": 6164
    },
    {
      "epoch": 0.2741541060868792,
      "grad_norm": 0.09051446616649628,
      "learning_rate": 0.000991702926773701,
      "loss": 1.5774,
      "step": 6166
    },
    {
      "epoch": 0.27424303054555155,
      "grad_norm": 0.09004546701908112,
      "learning_rate": 0.0009916965190049161,
      "loss": 1.5744,
      "step": 6168
    },
    {
      "epoch": 0.2743319550042239,
      "grad_norm": 0.0947934165596962,
      "learning_rate": 0.0009916901087834676,
      "loss": 1.5806,
      "step": 6170
    },
    {
      "epoch": 0.2744208794628963,
      "grad_norm": 0.0934051126241684,
      "learning_rate": 0.0009916836961093869,
      "loss": 1.5796,
      "step": 6172
    },
    {
      "epoch": 0.27450980392156865,
      "grad_norm": 0.09276668727397919,
      "learning_rate": 0.0009916772809827063,
      "loss": 1.5775,
      "step": 6174
    },
    {
      "epoch": 0.27459872838024096,
      "grad_norm": 0.09658815711736679,
      "learning_rate": 0.0009916708634034582,
      "loss": 1.5745,
      "step": 6176
    },
    {
      "epoch": 0.27468765283891333,
      "grad_norm": 0.09510058909654617,
      "learning_rate": 0.0009916644433716738,
      "loss": 1.5706,
      "step": 6178
    },
    {
      "epoch": 0.2747765772975857,
      "grad_norm": 0.09232553839683533,
      "learning_rate": 0.0009916580208873859,
      "loss": 1.5735,
      "step": 6180
    },
    {
      "epoch": 0.27486550175625807,
      "grad_norm": 0.09260194003582001,
      "learning_rate": 0.0009916515959506257,
      "loss": 1.5781,
      "step": 6182
    },
    {
      "epoch": 0.27495442621493044,
      "grad_norm": 0.09699893742799759,
      "learning_rate": 0.000991645168561426,
      "loss": 1.5788,
      "step": 6184
    },
    {
      "epoch": 0.27504335067360275,
      "grad_norm": 0.0900624617934227,
      "learning_rate": 0.0009916387387198186,
      "loss": 1.5794,
      "step": 6186
    },
    {
      "epoch": 0.2751322751322751,
      "grad_norm": 0.08997295051813126,
      "learning_rate": 0.0009916323064258358,
      "loss": 1.5737,
      "step": 6188
    },
    {
      "epoch": 0.2752211995909475,
      "grad_norm": 0.0929020345211029,
      "learning_rate": 0.000991625871679509,
      "loss": 1.5808,
      "step": 6190
    },
    {
      "epoch": 0.27531012404961985,
      "grad_norm": 0.09494993090629578,
      "learning_rate": 0.000991619434480871,
      "loss": 1.5808,
      "step": 6192
    },
    {
      "epoch": 0.2753990485082922,
      "grad_norm": 0.08931610733270645,
      "learning_rate": 0.0009916129948299537,
      "loss": 1.5718,
      "step": 6194
    },
    {
      "epoch": 0.2754879729669646,
      "grad_norm": 0.09554082900285721,
      "learning_rate": 0.000991606552726789,
      "loss": 1.5794,
      "step": 6196
    },
    {
      "epoch": 0.2755768974256369,
      "grad_norm": 0.09193310141563416,
      "learning_rate": 0.0009916001081714091,
      "loss": 1.5806,
      "step": 6198
    },
    {
      "epoch": 0.27566582188430927,
      "grad_norm": 0.08917775005102158,
      "learning_rate": 0.0009915936611638466,
      "loss": 1.5737,
      "step": 6200
    },
    {
      "epoch": 0.27575474634298164,
      "grad_norm": 0.0933012068271637,
      "learning_rate": 0.0009915872117041333,
      "loss": 1.5803,
      "step": 6202
    },
    {
      "epoch": 0.275843670801654,
      "grad_norm": 0.09677395969629288,
      "learning_rate": 0.0009915807597923013,
      "loss": 1.5793,
      "step": 6204
    },
    {
      "epoch": 0.2759325952603264,
      "grad_norm": 0.09007348865270615,
      "learning_rate": 0.0009915743054283827,
      "loss": 1.5755,
      "step": 6206
    },
    {
      "epoch": 0.2760215197189987,
      "grad_norm": 0.09194600582122803,
      "learning_rate": 0.00099156784861241,
      "loss": 1.5735,
      "step": 6208
    },
    {
      "epoch": 0.27611044417767105,
      "grad_norm": 0.09006384760141373,
      "learning_rate": 0.0009915613893444154,
      "loss": 1.574,
      "step": 6210
    },
    {
      "epoch": 0.2761993686363434,
      "grad_norm": 0.09457103908061981,
      "learning_rate": 0.0009915549276244308,
      "loss": 1.577,
      "step": 6212
    },
    {
      "epoch": 0.2762882930950158,
      "grad_norm": 0.09670677036046982,
      "learning_rate": 0.0009915484634524887,
      "loss": 1.5838,
      "step": 6214
    },
    {
      "epoch": 0.27637721755368816,
      "grad_norm": 0.09744661301374435,
      "learning_rate": 0.0009915419968286212,
      "loss": 1.58,
      "step": 6216
    },
    {
      "epoch": 0.2764661420123605,
      "grad_norm": 0.09463924914598465,
      "learning_rate": 0.0009915355277528607,
      "loss": 1.5767,
      "step": 6218
    },
    {
      "epoch": 0.27655506647103284,
      "grad_norm": 0.09342074394226074,
      "learning_rate": 0.0009915290562252394,
      "loss": 1.5844,
      "step": 6220
    },
    {
      "epoch": 0.2766439909297052,
      "grad_norm": 0.08680648356676102,
      "learning_rate": 0.0009915225822457897,
      "loss": 1.5792,
      "step": 6222
    },
    {
      "epoch": 0.2767329153883776,
      "grad_norm": 0.09087889641523361,
      "learning_rate": 0.0009915161058145436,
      "loss": 1.5797,
      "step": 6224
    },
    {
      "epoch": 0.27682183984704994,
      "grad_norm": 0.0886339470744133,
      "learning_rate": 0.0009915096269315336,
      "loss": 1.5726,
      "step": 6226
    },
    {
      "epoch": 0.2769107643057223,
      "grad_norm": 0.08888717740774155,
      "learning_rate": 0.000991503145596792,
      "loss": 1.5751,
      "step": 6228
    },
    {
      "epoch": 0.2769996887643946,
      "grad_norm": 0.09028730541467667,
      "learning_rate": 0.0009914966618103512,
      "loss": 1.581,
      "step": 6230
    },
    {
      "epoch": 0.277088613223067,
      "grad_norm": 0.09174933284521103,
      "learning_rate": 0.0009914901755722434,
      "loss": 1.5828,
      "step": 6232
    },
    {
      "epoch": 0.27717753768173936,
      "grad_norm": 0.09008409827947617,
      "learning_rate": 0.000991483686882501,
      "loss": 1.5725,
      "step": 6234
    },
    {
      "epoch": 0.27726646214041173,
      "grad_norm": 0.09247057884931564,
      "learning_rate": 0.0009914771957411563,
      "loss": 1.584,
      "step": 6236
    },
    {
      "epoch": 0.2773553865990841,
      "grad_norm": 0.09128382802009583,
      "learning_rate": 0.000991470702148242,
      "loss": 1.5849,
      "step": 6238
    },
    {
      "epoch": 0.2774443110577564,
      "grad_norm": 0.09105904400348663,
      "learning_rate": 0.0009914642061037901,
      "loss": 1.5744,
      "step": 6240
    },
    {
      "epoch": 0.2775332355164288,
      "grad_norm": 0.08844447880983353,
      "learning_rate": 0.0009914577076078333,
      "loss": 1.5689,
      "step": 6242
    },
    {
      "epoch": 0.27762215997510115,
      "grad_norm": 0.08989200741052628,
      "learning_rate": 0.0009914512066604036,
      "loss": 1.5729,
      "step": 6244
    },
    {
      "epoch": 0.2777110844337735,
      "grad_norm": 0.09029513597488403,
      "learning_rate": 0.0009914447032615338,
      "loss": 1.573,
      "step": 6246
    },
    {
      "epoch": 0.2778000088924459,
      "grad_norm": 0.08931106328964233,
      "learning_rate": 0.0009914381974112565,
      "loss": 1.5832,
      "step": 6248
    },
    {
      "epoch": 0.27788893335111825,
      "grad_norm": 0.08635377138853073,
      "learning_rate": 0.0009914316891096037,
      "loss": 1.5818,
      "step": 6250
    },
    {
      "epoch": 0.27797785780979056,
      "grad_norm": 0.08634750545024872,
      "learning_rate": 0.0009914251783566081,
      "loss": 1.5733,
      "step": 6252
    },
    {
      "epoch": 0.27806678226846293,
      "grad_norm": 0.09367714077234268,
      "learning_rate": 0.0009914186651523021,
      "loss": 1.5813,
      "step": 6254
    },
    {
      "epoch": 0.2781557067271353,
      "grad_norm": 0.08927787095308304,
      "learning_rate": 0.0009914121494967183,
      "loss": 1.5758,
      "step": 6256
    },
    {
      "epoch": 0.27824463118580767,
      "grad_norm": 0.0868934839963913,
      "learning_rate": 0.000991405631389889,
      "loss": 1.5718,
      "step": 6258
    },
    {
      "epoch": 0.27833355564448004,
      "grad_norm": 0.09208554029464722,
      "learning_rate": 0.000991399110831847,
      "loss": 1.5722,
      "step": 6260
    },
    {
      "epoch": 0.27842248010315235,
      "grad_norm": 0.09598352760076523,
      "learning_rate": 0.0009913925878226246,
      "loss": 1.5769,
      "step": 6262
    },
    {
      "epoch": 0.2785114045618247,
      "grad_norm": 0.09622900187969208,
      "learning_rate": 0.0009913860623622544,
      "loss": 1.5752,
      "step": 6264
    },
    {
      "epoch": 0.2786003290204971,
      "grad_norm": 0.09845644980669022,
      "learning_rate": 0.0009913795344507687,
      "loss": 1.5793,
      "step": 6266
    },
    {
      "epoch": 0.27868925347916945,
      "grad_norm": 0.08853481709957123,
      "learning_rate": 0.0009913730040882007,
      "loss": 1.5745,
      "step": 6268
    },
    {
      "epoch": 0.2787781779378418,
      "grad_norm": 0.09148851037025452,
      "learning_rate": 0.0009913664712745825,
      "loss": 1.5824,
      "step": 6270
    },
    {
      "epoch": 0.2788671023965142,
      "grad_norm": 0.09253410995006561,
      "learning_rate": 0.0009913599360099469,
      "loss": 1.5755,
      "step": 6272
    },
    {
      "epoch": 0.2789560268551865,
      "grad_norm": 0.09554262459278107,
      "learning_rate": 0.000991353398294326,
      "loss": 1.5737,
      "step": 6274
    },
    {
      "epoch": 0.27904495131385887,
      "grad_norm": 0.09318879246711731,
      "learning_rate": 0.000991346858127753,
      "loss": 1.5677,
      "step": 6276
    },
    {
      "epoch": 0.27913387577253124,
      "grad_norm": 0.09248467534780502,
      "learning_rate": 0.0009913403155102603,
      "loss": 1.5769,
      "step": 6278
    },
    {
      "epoch": 0.2792228002312036,
      "grad_norm": 0.09165968000888824,
      "learning_rate": 0.0009913337704418805,
      "loss": 1.5741,
      "step": 6280
    },
    {
      "epoch": 0.279311724689876,
      "grad_norm": 0.09123395383358002,
      "learning_rate": 0.0009913272229226465,
      "loss": 1.5758,
      "step": 6282
    },
    {
      "epoch": 0.2794006491485483,
      "grad_norm": 0.08967551589012146,
      "learning_rate": 0.0009913206729525905,
      "loss": 1.5773,
      "step": 6284
    },
    {
      "epoch": 0.27948957360722065,
      "grad_norm": 0.08778689056634903,
      "learning_rate": 0.0009913141205317455,
      "loss": 1.5767,
      "step": 6286
    },
    {
      "epoch": 0.279578498065893,
      "grad_norm": 0.084915891289711,
      "learning_rate": 0.0009913075656601442,
      "loss": 1.5787,
      "step": 6288
    },
    {
      "epoch": 0.2796674225245654,
      "grad_norm": 0.0918944925069809,
      "learning_rate": 0.0009913010083378191,
      "loss": 1.5747,
      "step": 6290
    },
    {
      "epoch": 0.27975634698323776,
      "grad_norm": 0.09487152099609375,
      "learning_rate": 0.0009912944485648031,
      "loss": 1.5794,
      "step": 6292
    },
    {
      "epoch": 0.27984527144191007,
      "grad_norm": 0.08974961936473846,
      "learning_rate": 0.000991287886341129,
      "loss": 1.5725,
      "step": 6294
    },
    {
      "epoch": 0.27993419590058244,
      "grad_norm": 0.09361551702022552,
      "learning_rate": 0.0009912813216668291,
      "loss": 1.5786,
      "step": 6296
    },
    {
      "epoch": 0.2800231203592548,
      "grad_norm": 0.08920656889677048,
      "learning_rate": 0.0009912747545419364,
      "loss": 1.581,
      "step": 6298
    },
    {
      "epoch": 0.2801120448179272,
      "grad_norm": 0.08863282203674316,
      "learning_rate": 0.0009912681849664838,
      "loss": 1.5788,
      "step": 6300
    },
    {
      "epoch": 0.28020096927659954,
      "grad_norm": 0.08976864069700241,
      "learning_rate": 0.000991261612940504,
      "loss": 1.5735,
      "step": 6302
    },
    {
      "epoch": 0.2802898937352719,
      "grad_norm": 0.08996647596359253,
      "learning_rate": 0.0009912550384640296,
      "loss": 1.5788,
      "step": 6304
    },
    {
      "epoch": 0.2803788181939442,
      "grad_norm": 0.09197447448968887,
      "learning_rate": 0.0009912484615370936,
      "loss": 1.5751,
      "step": 6306
    },
    {
      "epoch": 0.2804677426526166,
      "grad_norm": 0.09102824330329895,
      "learning_rate": 0.0009912418821597289,
      "loss": 1.5789,
      "step": 6308
    },
    {
      "epoch": 0.28055666711128896,
      "grad_norm": 0.09098983556032181,
      "learning_rate": 0.0009912353003319679,
      "loss": 1.5816,
      "step": 6310
    },
    {
      "epoch": 0.28064559156996133,
      "grad_norm": 0.0941241905093193,
      "learning_rate": 0.0009912287160538436,
      "loss": 1.5775,
      "step": 6312
    },
    {
      "epoch": 0.2807345160286337,
      "grad_norm": 0.09087178856134415,
      "learning_rate": 0.000991222129325389,
      "loss": 1.5723,
      "step": 6314
    },
    {
      "epoch": 0.280823440487306,
      "grad_norm": 0.08986848592758179,
      "learning_rate": 0.0009912155401466372,
      "loss": 1.5755,
      "step": 6316
    },
    {
      "epoch": 0.2809123649459784,
      "grad_norm": 0.0928303524851799,
      "learning_rate": 0.0009912089485176203,
      "loss": 1.5737,
      "step": 6318
    },
    {
      "epoch": 0.28100128940465074,
      "grad_norm": 0.09010421484708786,
      "learning_rate": 0.000991202354438372,
      "loss": 1.5804,
      "step": 6320
    },
    {
      "epoch": 0.2810902138633231,
      "grad_norm": 0.09154298156499863,
      "learning_rate": 0.0009911957579089245,
      "loss": 1.5741,
      "step": 6322
    },
    {
      "epoch": 0.2811791383219955,
      "grad_norm": 0.09349411725997925,
      "learning_rate": 0.0009911891589293113,
      "loss": 1.571,
      "step": 6324
    },
    {
      "epoch": 0.28126806278066785,
      "grad_norm": 0.08908021450042725,
      "learning_rate": 0.000991182557499565,
      "loss": 1.5772,
      "step": 6326
    },
    {
      "epoch": 0.28135698723934016,
      "grad_norm": 0.09845002740621567,
      "learning_rate": 0.0009911759536197183,
      "loss": 1.5778,
      "step": 6328
    },
    {
      "epoch": 0.28144591169801253,
      "grad_norm": 0.08850311487913132,
      "learning_rate": 0.0009911693472898046,
      "loss": 1.5749,
      "step": 6330
    },
    {
      "epoch": 0.2815348361566849,
      "grad_norm": 0.09225910902023315,
      "learning_rate": 0.0009911627385098566,
      "loss": 1.577,
      "step": 6332
    },
    {
      "epoch": 0.28162376061535727,
      "grad_norm": 0.09028570353984833,
      "learning_rate": 0.0009911561272799075,
      "loss": 1.5768,
      "step": 6334
    },
    {
      "epoch": 0.28171268507402963,
      "grad_norm": 0.08955962210893631,
      "learning_rate": 0.00099114951359999,
      "loss": 1.5769,
      "step": 6336
    },
    {
      "epoch": 0.28180160953270195,
      "grad_norm": 0.08538378030061722,
      "learning_rate": 0.0009911428974701372,
      "loss": 1.5768,
      "step": 6338
    },
    {
      "epoch": 0.2818905339913743,
      "grad_norm": 0.08776732534170151,
      "learning_rate": 0.0009911362788903822,
      "loss": 1.5755,
      "step": 6340
    },
    {
      "epoch": 0.2819794584500467,
      "grad_norm": 0.08690127730369568,
      "learning_rate": 0.0009911296578607577,
      "loss": 1.5728,
      "step": 6342
    },
    {
      "epoch": 0.28206838290871905,
      "grad_norm": 0.08759032934904099,
      "learning_rate": 0.000991123034381297,
      "loss": 1.5728,
      "step": 6344
    },
    {
      "epoch": 0.2821573073673914,
      "grad_norm": 0.08740068972110748,
      "learning_rate": 0.0009911164084520333,
      "loss": 1.5709,
      "step": 6346
    },
    {
      "epoch": 0.28224623182606373,
      "grad_norm": 0.09269807487726212,
      "learning_rate": 0.0009911097800729993,
      "loss": 1.5711,
      "step": 6348
    },
    {
      "epoch": 0.2823351562847361,
      "grad_norm": 0.08954404294490814,
      "learning_rate": 0.0009911031492442282,
      "loss": 1.5742,
      "step": 6350
    },
    {
      "epoch": 0.28242408074340847,
      "grad_norm": 0.08641083538532257,
      "learning_rate": 0.000991096515965753,
      "loss": 1.5766,
      "step": 6352
    },
    {
      "epoch": 0.28251300520208084,
      "grad_norm": 0.0864805057644844,
      "learning_rate": 0.0009910898802376069,
      "loss": 1.5738,
      "step": 6354
    },
    {
      "epoch": 0.2826019296607532,
      "grad_norm": 0.09081622213125229,
      "learning_rate": 0.000991083242059823,
      "loss": 1.5746,
      "step": 6356
    },
    {
      "epoch": 0.28269085411942557,
      "grad_norm": 0.09132978320121765,
      "learning_rate": 0.0009910766014324345,
      "loss": 1.5765,
      "step": 6358
    },
    {
      "epoch": 0.2827797785780979,
      "grad_norm": 0.09193006902933121,
      "learning_rate": 0.000991069958355474,
      "loss": 1.5782,
      "step": 6360
    },
    {
      "epoch": 0.28286870303677025,
      "grad_norm": 0.09301292151212692,
      "learning_rate": 0.0009910633128289753,
      "loss": 1.5711,
      "step": 6362
    },
    {
      "epoch": 0.2829576274954426,
      "grad_norm": 0.09332393854856491,
      "learning_rate": 0.0009910566648529712,
      "loss": 1.5724,
      "step": 6364
    },
    {
      "epoch": 0.283046551954115,
      "grad_norm": 0.08995355665683746,
      "learning_rate": 0.000991050014427495,
      "loss": 1.5861,
      "step": 6366
    },
    {
      "epoch": 0.28313547641278736,
      "grad_norm": 0.09018755704164505,
      "learning_rate": 0.00099104336155258,
      "loss": 1.5695,
      "step": 6368
    },
    {
      "epoch": 0.28322440087145967,
      "grad_norm": 0.09240277856588364,
      "learning_rate": 0.000991036706228259,
      "loss": 1.5718,
      "step": 6370
    },
    {
      "epoch": 0.28331332533013204,
      "grad_norm": 0.08703695982694626,
      "learning_rate": 0.0009910300484545654,
      "loss": 1.5753,
      "step": 6372
    },
    {
      "epoch": 0.2834022497888044,
      "grad_norm": 0.09340234845876694,
      "learning_rate": 0.0009910233882315326,
      "loss": 1.578,
      "step": 6374
    },
    {
      "epoch": 0.2834911742474768,
      "grad_norm": 0.09236439317464828,
      "learning_rate": 0.0009910167255591934,
      "loss": 1.5762,
      "step": 6376
    },
    {
      "epoch": 0.28358009870614914,
      "grad_norm": 0.09320174157619476,
      "learning_rate": 0.0009910100604375812,
      "loss": 1.569,
      "step": 6378
    },
    {
      "epoch": 0.2836690231648215,
      "grad_norm": 0.08888036757707596,
      "learning_rate": 0.0009910033928667294,
      "loss": 1.5729,
      "step": 6380
    },
    {
      "epoch": 0.2837579476234938,
      "grad_norm": 0.0888710469007492,
      "learning_rate": 0.0009909967228466713,
      "loss": 1.5731,
      "step": 6382
    },
    {
      "epoch": 0.2838468720821662,
      "grad_norm": 0.09013516455888748,
      "learning_rate": 0.00099099005037744,
      "loss": 1.5704,
      "step": 6384
    },
    {
      "epoch": 0.28393579654083856,
      "grad_norm": 0.08720479905605316,
      "learning_rate": 0.0009909833754590688,
      "loss": 1.5738,
      "step": 6386
    },
    {
      "epoch": 0.2840247209995109,
      "grad_norm": 0.08885452896356583,
      "learning_rate": 0.000990976698091591,
      "loss": 1.5726,
      "step": 6388
    },
    {
      "epoch": 0.2841136454581833,
      "grad_norm": 0.09276578575372696,
      "learning_rate": 0.00099097001827504,
      "loss": 1.5756,
      "step": 6390
    },
    {
      "epoch": 0.2842025699168556,
      "grad_norm": 0.08871392160654068,
      "learning_rate": 0.000990963336009449,
      "loss": 1.5747,
      "step": 6392
    },
    {
      "epoch": 0.284291494375528,
      "grad_norm": 0.09305480867624283,
      "learning_rate": 0.0009909566512948513,
      "loss": 1.5691,
      "step": 6394
    },
    {
      "epoch": 0.28438041883420034,
      "grad_norm": 0.08993328362703323,
      "learning_rate": 0.0009909499641312805,
      "loss": 1.5752,
      "step": 6396
    },
    {
      "epoch": 0.2844693432928727,
      "grad_norm": 0.09120582789182663,
      "learning_rate": 0.0009909432745187697,
      "loss": 1.5774,
      "step": 6398
    },
    {
      "epoch": 0.2845582677515451,
      "grad_norm": 0.08618117868900299,
      "learning_rate": 0.0009909365824573523,
      "loss": 1.575,
      "step": 6400
    },
    {
      "epoch": 0.28464719221021745,
      "grad_norm": 0.08837292343378067,
      "learning_rate": 0.0009909298879470618,
      "loss": 1.5787,
      "step": 6402
    },
    {
      "epoch": 0.28473611666888976,
      "grad_norm": 0.08570041507482529,
      "learning_rate": 0.0009909231909879315,
      "loss": 1.5808,
      "step": 6404
    },
    {
      "epoch": 0.28482504112756213,
      "grad_norm": 0.09008973836898804,
      "learning_rate": 0.000990916491579995,
      "loss": 1.5763,
      "step": 6406
    },
    {
      "epoch": 0.2849139655862345,
      "grad_norm": 0.09252497553825378,
      "learning_rate": 0.0009909097897232854,
      "loss": 1.5776,
      "step": 6408
    },
    {
      "epoch": 0.28500289004490686,
      "grad_norm": 0.08563211560249329,
      "learning_rate": 0.0009909030854178362,
      "loss": 1.5674,
      "step": 6410
    },
    {
      "epoch": 0.28509181450357923,
      "grad_norm": 0.09084684401750565,
      "learning_rate": 0.0009908963786636811,
      "loss": 1.5732,
      "step": 6412
    },
    {
      "epoch": 0.28518073896225155,
      "grad_norm": 0.09400637447834015,
      "learning_rate": 0.0009908896694608533,
      "loss": 1.5712,
      "step": 6414
    },
    {
      "epoch": 0.2852696634209239,
      "grad_norm": 0.08952626585960388,
      "learning_rate": 0.0009908829578093863,
      "loss": 1.5744,
      "step": 6416
    },
    {
      "epoch": 0.2853585878795963,
      "grad_norm": 0.09080950170755386,
      "learning_rate": 0.0009908762437093137,
      "loss": 1.57,
      "step": 6418
    },
    {
      "epoch": 0.28544751233826865,
      "grad_norm": 0.08886551856994629,
      "learning_rate": 0.000990869527160669,
      "loss": 1.5763,
      "step": 6420
    },
    {
      "epoch": 0.285536436796941,
      "grad_norm": 0.09163545817136765,
      "learning_rate": 0.0009908628081634855,
      "loss": 1.577,
      "step": 6422
    },
    {
      "epoch": 0.28562536125561333,
      "grad_norm": 0.08795285224914551,
      "learning_rate": 0.0009908560867177966,
      "loss": 1.5714,
      "step": 6424
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.08879886567592621,
      "learning_rate": 0.0009908493628236364,
      "loss": 1.5738,
      "step": 6426
    },
    {
      "epoch": 0.28580321017295807,
      "grad_norm": 0.09103137999773026,
      "learning_rate": 0.000990842636481038,
      "loss": 1.5725,
      "step": 6428
    },
    {
      "epoch": 0.28589213463163043,
      "grad_norm": 0.08968726545572281,
      "learning_rate": 0.0009908359076900351,
      "loss": 1.5728,
      "step": 6430
    },
    {
      "epoch": 0.2859810590903028,
      "grad_norm": 0.08867494016885757,
      "learning_rate": 0.000990829176450661,
      "loss": 1.5695,
      "step": 6432
    },
    {
      "epoch": 0.28606998354897517,
      "grad_norm": 0.08909203112125397,
      "learning_rate": 0.0009908224427629497,
      "loss": 1.5785,
      "step": 6434
    },
    {
      "epoch": 0.2861589080076475,
      "grad_norm": 0.08877209573984146,
      "learning_rate": 0.0009908157066269346,
      "loss": 1.5675,
      "step": 6436
    },
    {
      "epoch": 0.28624783246631985,
      "grad_norm": 0.08952686190605164,
      "learning_rate": 0.0009908089680426492,
      "loss": 1.5712,
      "step": 6438
    },
    {
      "epoch": 0.2863367569249922,
      "grad_norm": 0.08946753293275833,
      "learning_rate": 0.0009908022270101273,
      "loss": 1.5706,
      "step": 6440
    },
    {
      "epoch": 0.2864256813836646,
      "grad_norm": 0.08814926445484161,
      "learning_rate": 0.0009907954835294023,
      "loss": 1.5738,
      "step": 6442
    },
    {
      "epoch": 0.28651460584233696,
      "grad_norm": 0.09422788769006729,
      "learning_rate": 0.000990788737600508,
      "loss": 1.5799,
      "step": 6444
    },
    {
      "epoch": 0.28660353030100927,
      "grad_norm": 0.09289488941431046,
      "learning_rate": 0.0009907819892234781,
      "loss": 1.5761,
      "step": 6446
    },
    {
      "epoch": 0.28669245475968164,
      "grad_norm": 0.08914397656917572,
      "learning_rate": 0.0009907752383983461,
      "loss": 1.569,
      "step": 6448
    },
    {
      "epoch": 0.286781379218354,
      "grad_norm": 0.08954908698797226,
      "learning_rate": 0.0009907684851251457,
      "loss": 1.5802,
      "step": 6450
    },
    {
      "epoch": 0.2868703036770264,
      "grad_norm": 0.09144933521747589,
      "learning_rate": 0.0009907617294039107,
      "loss": 1.5758,
      "step": 6452
    },
    {
      "epoch": 0.28695922813569874,
      "grad_norm": 0.09079254418611526,
      "learning_rate": 0.0009907549712346747,
      "loss": 1.5721,
      "step": 6454
    },
    {
      "epoch": 0.2870481525943711,
      "grad_norm": 0.08957768976688385,
      "learning_rate": 0.0009907482106174714,
      "loss": 1.5687,
      "step": 6456
    },
    {
      "epoch": 0.2871370770530434,
      "grad_norm": 0.09010159969329834,
      "learning_rate": 0.0009907414475523347,
      "loss": 1.5772,
      "step": 6458
    },
    {
      "epoch": 0.2872260015117158,
      "grad_norm": 0.09015689790248871,
      "learning_rate": 0.000990734682039298,
      "loss": 1.57,
      "step": 6460
    },
    {
      "epoch": 0.28731492597038816,
      "grad_norm": 0.0964498221874237,
      "learning_rate": 0.0009907279140783953,
      "loss": 1.5716,
      "step": 6462
    },
    {
      "epoch": 0.2874038504290605,
      "grad_norm": 0.09194839745759964,
      "learning_rate": 0.0009907211436696603,
      "loss": 1.5666,
      "step": 6464
    },
    {
      "epoch": 0.2874927748877329,
      "grad_norm": 0.0931999608874321,
      "learning_rate": 0.000990714370813127,
      "loss": 1.5766,
      "step": 6466
    },
    {
      "epoch": 0.2875816993464052,
      "grad_norm": 0.08518645912408829,
      "learning_rate": 0.0009907075955088286,
      "loss": 1.5778,
      "step": 6468
    },
    {
      "epoch": 0.2876706238050776,
      "grad_norm": 0.08883948624134064,
      "learning_rate": 0.0009907008177567997,
      "loss": 1.5709,
      "step": 6470
    },
    {
      "epoch": 0.28775954826374994,
      "grad_norm": 0.08934309333562851,
      "learning_rate": 0.0009906940375570734,
      "loss": 1.5703,
      "step": 6472
    },
    {
      "epoch": 0.2878484727224223,
      "grad_norm": 0.09022992104291916,
      "learning_rate": 0.0009906872549096837,
      "loss": 1.572,
      "step": 6474
    },
    {
      "epoch": 0.2879373971810947,
      "grad_norm": 0.09257543832063675,
      "learning_rate": 0.0009906804698146647,
      "loss": 1.5726,
      "step": 6476
    },
    {
      "epoch": 0.288026321639767,
      "grad_norm": 0.09016820788383484,
      "learning_rate": 0.0009906736822720501,
      "loss": 1.5735,
      "step": 6478
    },
    {
      "epoch": 0.28811524609843936,
      "grad_norm": 0.08995562046766281,
      "learning_rate": 0.0009906668922818735,
      "loss": 1.5661,
      "step": 6480
    },
    {
      "epoch": 0.2882041705571117,
      "grad_norm": 0.09062064439058304,
      "learning_rate": 0.0009906600998441694,
      "loss": 1.5656,
      "step": 6482
    },
    {
      "epoch": 0.2882930950157841,
      "grad_norm": 0.08658911287784576,
      "learning_rate": 0.0009906533049589708,
      "loss": 1.5682,
      "step": 6484
    },
    {
      "epoch": 0.28838201947445646,
      "grad_norm": 0.08712375164031982,
      "learning_rate": 0.0009906465076263123,
      "loss": 1.5734,
      "step": 6486
    },
    {
      "epoch": 0.28847094393312883,
      "grad_norm": 0.0906602293252945,
      "learning_rate": 0.0009906397078462276,
      "loss": 1.5742,
      "step": 6488
    },
    {
      "epoch": 0.28855986839180114,
      "grad_norm": 0.08672540634870529,
      "learning_rate": 0.0009906329056187505,
      "loss": 1.5789,
      "step": 6490
    },
    {
      "epoch": 0.2886487928504735,
      "grad_norm": 0.08817194402217865,
      "learning_rate": 0.0009906261009439152,
      "loss": 1.5728,
      "step": 6492
    },
    {
      "epoch": 0.2887377173091459,
      "grad_norm": 0.09028277546167374,
      "learning_rate": 0.0009906192938217552,
      "loss": 1.5735,
      "step": 6494
    },
    {
      "epoch": 0.28882664176781825,
      "grad_norm": 0.09235137701034546,
      "learning_rate": 0.000990612484252305,
      "loss": 1.5749,
      "step": 6496
    },
    {
      "epoch": 0.2889155662264906,
      "grad_norm": 0.09245535731315613,
      "learning_rate": 0.000990605672235598,
      "loss": 1.5751,
      "step": 6498
    },
    {
      "epoch": 0.28900449068516293,
      "grad_norm": 0.08902133256196976,
      "learning_rate": 0.0009905988577716687,
      "loss": 1.5744,
      "step": 6500
    },
    {
      "epoch": 0.28900449068516293,
      "eval_loss": 1.5468125343322754,
      "eval_runtime": 12.4163,
      "eval_samples_per_second": 556.526,
      "eval_steps_per_second": 69.586,
      "step": 6500
    },
    {
      "epoch": 0.2890934151438353,
      "grad_norm": 0.09113017469644547,
      "learning_rate": 0.0009905920408605508,
      "loss": 1.5807,
      "step": 6502
    },
    {
      "epoch": 0.28918233960250767,
      "grad_norm": 0.09066907316446304,
      "learning_rate": 0.0009905852215022783,
      "loss": 1.5756,
      "step": 6504
    },
    {
      "epoch": 0.28927126406118003,
      "grad_norm": 0.09358642995357513,
      "learning_rate": 0.0009905783996968854,
      "loss": 1.5779,
      "step": 6506
    },
    {
      "epoch": 0.2893601885198524,
      "grad_norm": 0.08834748715162277,
      "learning_rate": 0.000990571575444406,
      "loss": 1.5675,
      "step": 6508
    },
    {
      "epoch": 0.28944911297852477,
      "grad_norm": 0.08833714574575424,
      "learning_rate": 0.000990564748744874,
      "loss": 1.5699,
      "step": 6510
    },
    {
      "epoch": 0.2895380374371971,
      "grad_norm": 0.09438148140907288,
      "learning_rate": 0.000990557919598324,
      "loss": 1.5703,
      "step": 6512
    },
    {
      "epoch": 0.28962696189586945,
      "grad_norm": 0.08776549994945526,
      "learning_rate": 0.000990551088004789,
      "loss": 1.5659,
      "step": 6514
    },
    {
      "epoch": 0.2897158863545418,
      "grad_norm": 0.09066025912761688,
      "learning_rate": 0.0009905442539643043,
      "loss": 1.5698,
      "step": 6516
    },
    {
      "epoch": 0.2898048108132142,
      "grad_norm": 0.08665291219949722,
      "learning_rate": 0.0009905374174769032,
      "loss": 1.5707,
      "step": 6518
    },
    {
      "epoch": 0.28989373527188655,
      "grad_norm": 0.08850061148405075,
      "learning_rate": 0.0009905305785426202,
      "loss": 1.5764,
      "step": 6520
    },
    {
      "epoch": 0.28998265973055887,
      "grad_norm": 0.09048718959093094,
      "learning_rate": 0.000990523737161489,
      "loss": 1.5785,
      "step": 6522
    },
    {
      "epoch": 0.29007158418923124,
      "grad_norm": 0.08610177785158157,
      "learning_rate": 0.0009905168933335442,
      "loss": 1.5681,
      "step": 6524
    },
    {
      "epoch": 0.2901605086479036,
      "grad_norm": 0.08931626379489899,
      "learning_rate": 0.0009905100470588197,
      "loss": 1.5733,
      "step": 6526
    },
    {
      "epoch": 0.29024943310657597,
      "grad_norm": 0.08849899470806122,
      "learning_rate": 0.0009905031983373496,
      "loss": 1.5729,
      "step": 6528
    },
    {
      "epoch": 0.29033835756524834,
      "grad_norm": 0.08723746240139008,
      "learning_rate": 0.0009904963471691682,
      "loss": 1.5738,
      "step": 6530
    },
    {
      "epoch": 0.29042728202392065,
      "grad_norm": 0.08764983713626862,
      "learning_rate": 0.0009904894935543094,
      "loss": 1.5785,
      "step": 6532
    },
    {
      "epoch": 0.290516206482593,
      "grad_norm": 0.09550359100103378,
      "learning_rate": 0.0009904826374928077,
      "loss": 1.5718,
      "step": 6534
    },
    {
      "epoch": 0.2906051309412654,
      "grad_norm": 0.09002628922462463,
      "learning_rate": 0.0009904757789846973,
      "loss": 1.5796,
      "step": 6536
    },
    {
      "epoch": 0.29069405539993776,
      "grad_norm": 0.09034377336502075,
      "learning_rate": 0.0009904689180300122,
      "loss": 1.5693,
      "step": 6538
    },
    {
      "epoch": 0.2907829798586101,
      "grad_norm": 0.09026948362588882,
      "learning_rate": 0.0009904620546287866,
      "loss": 1.578,
      "step": 6540
    },
    {
      "epoch": 0.2908719043172825,
      "grad_norm": 0.09430602192878723,
      "learning_rate": 0.0009904551887810551,
      "loss": 1.574,
      "step": 6542
    },
    {
      "epoch": 0.2909608287759548,
      "grad_norm": 0.08742611855268478,
      "learning_rate": 0.0009904483204868516,
      "loss": 1.5757,
      "step": 6544
    },
    {
      "epoch": 0.2910497532346272,
      "grad_norm": 0.08668093383312225,
      "learning_rate": 0.0009904414497462104,
      "loss": 1.5713,
      "step": 6546
    },
    {
      "epoch": 0.29113867769329954,
      "grad_norm": 0.0894765630364418,
      "learning_rate": 0.000990434576559166,
      "loss": 1.5737,
      "step": 6548
    },
    {
      "epoch": 0.2912276021519719,
      "grad_norm": 0.09396469593048096,
      "learning_rate": 0.0009904277009257524,
      "loss": 1.5765,
      "step": 6550
    },
    {
      "epoch": 0.2913165266106443,
      "grad_norm": 0.08750537782907486,
      "learning_rate": 0.0009904208228460041,
      "loss": 1.5662,
      "step": 6552
    },
    {
      "epoch": 0.2914054510693166,
      "grad_norm": 0.08443306386470795,
      "learning_rate": 0.0009904139423199552,
      "loss": 1.5776,
      "step": 6554
    },
    {
      "epoch": 0.29149437552798896,
      "grad_norm": 0.08997094631195068,
      "learning_rate": 0.00099040705934764,
      "loss": 1.5678,
      "step": 6556
    },
    {
      "epoch": 0.2915832999866613,
      "grad_norm": 0.08915294706821442,
      "learning_rate": 0.0009904001739290934,
      "loss": 1.5724,
      "step": 6558
    },
    {
      "epoch": 0.2916722244453337,
      "grad_norm": 0.08931171149015427,
      "learning_rate": 0.0009903932860643493,
      "loss": 1.5644,
      "step": 6560
    },
    {
      "epoch": 0.29176114890400606,
      "grad_norm": 0.09019617736339569,
      "learning_rate": 0.0009903863957534417,
      "loss": 1.5755,
      "step": 6562
    },
    {
      "epoch": 0.29185007336267843,
      "grad_norm": 0.08960139006376266,
      "learning_rate": 0.0009903795029964058,
      "loss": 1.5696,
      "step": 6564
    },
    {
      "epoch": 0.29193899782135074,
      "grad_norm": 0.08686326444149017,
      "learning_rate": 0.0009903726077932753,
      "loss": 1.5728,
      "step": 6566
    },
    {
      "epoch": 0.2920279222800231,
      "grad_norm": 0.08807472884654999,
      "learning_rate": 0.0009903657101440848,
      "loss": 1.5748,
      "step": 6568
    },
    {
      "epoch": 0.2921168467386955,
      "grad_norm": 0.0936129167675972,
      "learning_rate": 0.0009903588100488687,
      "loss": 1.5737,
      "step": 6570
    },
    {
      "epoch": 0.29220577119736785,
      "grad_norm": 0.09177185595035553,
      "learning_rate": 0.0009903519075076617,
      "loss": 1.5696,
      "step": 6572
    },
    {
      "epoch": 0.2922946956560402,
      "grad_norm": 0.08630893379449844,
      "learning_rate": 0.0009903450025204978,
      "loss": 1.5701,
      "step": 6574
    },
    {
      "epoch": 0.29238362011471253,
      "grad_norm": 0.08818253129720688,
      "learning_rate": 0.0009903380950874116,
      "loss": 1.5675,
      "step": 6576
    },
    {
      "epoch": 0.2924725445733849,
      "grad_norm": 0.08376295864582062,
      "learning_rate": 0.0009903311852084379,
      "loss": 1.5677,
      "step": 6578
    },
    {
      "epoch": 0.29256146903205726,
      "grad_norm": 0.08859193325042725,
      "learning_rate": 0.0009903242728836106,
      "loss": 1.5702,
      "step": 6580
    },
    {
      "epoch": 0.29265039349072963,
      "grad_norm": 0.0904025286436081,
      "learning_rate": 0.0009903173581129645,
      "loss": 1.5719,
      "step": 6582
    },
    {
      "epoch": 0.292739317949402,
      "grad_norm": 0.08744067698717117,
      "learning_rate": 0.0009903104408965338,
      "loss": 1.57,
      "step": 6584
    },
    {
      "epoch": 0.29282824240807437,
      "grad_norm": 0.08911757916212082,
      "learning_rate": 0.0009903035212343535,
      "loss": 1.5734,
      "step": 6586
    },
    {
      "epoch": 0.2929171668667467,
      "grad_norm": 0.0855589210987091,
      "learning_rate": 0.000990296599126458,
      "loss": 1.5706,
      "step": 6588
    },
    {
      "epoch": 0.29300609132541905,
      "grad_norm": 0.0880928561091423,
      "learning_rate": 0.0009902896745728814,
      "loss": 1.5733,
      "step": 6590
    },
    {
      "epoch": 0.2930950157840914,
      "grad_norm": 0.08391828835010529,
      "learning_rate": 0.0009902827475736585,
      "loss": 1.5713,
      "step": 6592
    },
    {
      "epoch": 0.2931839402427638,
      "grad_norm": 0.08916763961315155,
      "learning_rate": 0.000990275818128824,
      "loss": 1.575,
      "step": 6594
    },
    {
      "epoch": 0.29327286470143615,
      "grad_norm": 0.08984120190143585,
      "learning_rate": 0.0009902688862384122,
      "loss": 1.5738,
      "step": 6596
    },
    {
      "epoch": 0.29336178916010847,
      "grad_norm": 0.08982984721660614,
      "learning_rate": 0.000990261951902458,
      "loss": 1.5709,
      "step": 6598
    },
    {
      "epoch": 0.29345071361878083,
      "grad_norm": 0.08733770996332169,
      "learning_rate": 0.0009902550151209956,
      "loss": 1.5711,
      "step": 6600
    },
    {
      "epoch": 0.2935396380774532,
      "grad_norm": 0.08438307046890259,
      "learning_rate": 0.00099024807589406,
      "loss": 1.5683,
      "step": 6602
    },
    {
      "epoch": 0.29362856253612557,
      "grad_norm": 0.08910389244556427,
      "learning_rate": 0.0009902411342216854,
      "loss": 1.5732,
      "step": 6604
    },
    {
      "epoch": 0.29371748699479794,
      "grad_norm": 0.09192346036434174,
      "learning_rate": 0.000990234190103907,
      "loss": 1.5696,
      "step": 6606
    },
    {
      "epoch": 0.29380641145347025,
      "grad_norm": 0.0915895402431488,
      "learning_rate": 0.0009902272435407587,
      "loss": 1.5671,
      "step": 6608
    },
    {
      "epoch": 0.2938953359121426,
      "grad_norm": 0.09286682307720184,
      "learning_rate": 0.0009902202945322757,
      "loss": 1.5757,
      "step": 6610
    },
    {
      "epoch": 0.293984260370815,
      "grad_norm": 0.0949014201760292,
      "learning_rate": 0.0009902133430784925,
      "loss": 1.5736,
      "step": 6612
    },
    {
      "epoch": 0.29407318482948736,
      "grad_norm": 0.09041435271501541,
      "learning_rate": 0.0009902063891794439,
      "loss": 1.5739,
      "step": 6614
    },
    {
      "epoch": 0.2941621092881597,
      "grad_norm": 0.09175708889961243,
      "learning_rate": 0.0009901994328351642,
      "loss": 1.5704,
      "step": 6616
    },
    {
      "epoch": 0.2942510337468321,
      "grad_norm": 0.09223156422376633,
      "learning_rate": 0.0009901924740456885,
      "loss": 1.5742,
      "step": 6618
    },
    {
      "epoch": 0.2943399582055044,
      "grad_norm": 0.08808215707540512,
      "learning_rate": 0.0009901855128110514,
      "loss": 1.5751,
      "step": 6620
    },
    {
      "epoch": 0.29442888266417677,
      "grad_norm": 0.08614702522754669,
      "learning_rate": 0.0009901785491312875,
      "loss": 1.5766,
      "step": 6622
    },
    {
      "epoch": 0.29451780712284914,
      "grad_norm": 0.08723152428865433,
      "learning_rate": 0.0009901715830064317,
      "loss": 1.5684,
      "step": 6624
    },
    {
      "epoch": 0.2946067315815215,
      "grad_norm": 0.08894877880811691,
      "learning_rate": 0.0009901646144365188,
      "loss": 1.5734,
      "step": 6626
    },
    {
      "epoch": 0.2946956560401939,
      "grad_norm": 0.09144090861082077,
      "learning_rate": 0.0009901576434215833,
      "loss": 1.574,
      "step": 6628
    },
    {
      "epoch": 0.2947845804988662,
      "grad_norm": 0.08594503253698349,
      "learning_rate": 0.0009901506699616601,
      "loss": 1.5716,
      "step": 6630
    },
    {
      "epoch": 0.29487350495753856,
      "grad_norm": 0.09015947580337524,
      "learning_rate": 0.000990143694056784,
      "loss": 1.5693,
      "step": 6632
    },
    {
      "epoch": 0.2949624294162109,
      "grad_norm": 0.08714131265878677,
      "learning_rate": 0.0009901367157069897,
      "loss": 1.5629,
      "step": 6634
    },
    {
      "epoch": 0.2950513538748833,
      "grad_norm": 0.08628025650978088,
      "learning_rate": 0.0009901297349123124,
      "loss": 1.5687,
      "step": 6636
    },
    {
      "epoch": 0.29514027833355566,
      "grad_norm": 0.09013169258832932,
      "learning_rate": 0.0009901227516727865,
      "loss": 1.5728,
      "step": 6638
    },
    {
      "epoch": 0.29522920279222803,
      "grad_norm": 0.09108193218708038,
      "learning_rate": 0.0009901157659884467,
      "loss": 1.5688,
      "step": 6640
    },
    {
      "epoch": 0.29531812725090034,
      "grad_norm": 0.08838926255702972,
      "learning_rate": 0.0009901087778593286,
      "loss": 1.5676,
      "step": 6642
    },
    {
      "epoch": 0.2954070517095727,
      "grad_norm": 0.085484080016613,
      "learning_rate": 0.0009901017872854662,
      "loss": 1.5715,
      "step": 6644
    },
    {
      "epoch": 0.2954959761682451,
      "grad_norm": 0.08966375142335892,
      "learning_rate": 0.0009900947942668948,
      "loss": 1.5704,
      "step": 6646
    },
    {
      "epoch": 0.29558490062691745,
      "grad_norm": 0.08792408555746078,
      "learning_rate": 0.0009900877988036493,
      "loss": 1.5707,
      "step": 6648
    },
    {
      "epoch": 0.2956738250855898,
      "grad_norm": 0.08903691917657852,
      "learning_rate": 0.0009900808008957642,
      "loss": 1.5769,
      "step": 6650
    },
    {
      "epoch": 0.2957627495442621,
      "grad_norm": 0.09390727430582047,
      "learning_rate": 0.000990073800543275,
      "loss": 1.5715,
      "step": 6652
    },
    {
      "epoch": 0.2958516740029345,
      "grad_norm": 0.09213802218437195,
      "learning_rate": 0.0009900667977462163,
      "loss": 1.5754,
      "step": 6654
    },
    {
      "epoch": 0.29594059846160686,
      "grad_norm": 0.0865757092833519,
      "learning_rate": 0.000990059792504623,
      "loss": 1.5704,
      "step": 6656
    },
    {
      "epoch": 0.29602952292027923,
      "grad_norm": 0.08747228235006332,
      "learning_rate": 0.0009900527848185302,
      "loss": 1.571,
      "step": 6658
    },
    {
      "epoch": 0.2961184473789516,
      "grad_norm": 0.08495338261127472,
      "learning_rate": 0.0009900457746879728,
      "loss": 1.5669,
      "step": 6660
    },
    {
      "epoch": 0.2962073718376239,
      "grad_norm": 0.0894813984632492,
      "learning_rate": 0.0009900387621129855,
      "loss": 1.5695,
      "step": 6662
    },
    {
      "epoch": 0.2962962962962963,
      "grad_norm": 0.08558983355760574,
      "learning_rate": 0.0009900317470936037,
      "loss": 1.5744,
      "step": 6664
    },
    {
      "epoch": 0.29638522075496865,
      "grad_norm": 0.08508554100990295,
      "learning_rate": 0.0009900247296298621,
      "loss": 1.5674,
      "step": 6666
    },
    {
      "epoch": 0.296474145213641,
      "grad_norm": 0.08840614557266235,
      "learning_rate": 0.000990017709721796,
      "loss": 1.569,
      "step": 6668
    },
    {
      "epoch": 0.2965630696723134,
      "grad_norm": 0.08688345551490784,
      "learning_rate": 0.0009900106873694402,
      "loss": 1.5724,
      "step": 6670
    },
    {
      "epoch": 0.29665199413098575,
      "grad_norm": 0.0900869369506836,
      "learning_rate": 0.0009900036625728294,
      "loss": 1.5707,
      "step": 6672
    },
    {
      "epoch": 0.29674091858965806,
      "grad_norm": 0.08745554089546204,
      "learning_rate": 0.0009899966353319994,
      "loss": 1.5689,
      "step": 6674
    },
    {
      "epoch": 0.29682984304833043,
      "grad_norm": 0.08753221482038498,
      "learning_rate": 0.0009899896056469845,
      "loss": 1.5709,
      "step": 6676
    },
    {
      "epoch": 0.2969187675070028,
      "grad_norm": 0.09187465906143188,
      "learning_rate": 0.0009899825735178204,
      "loss": 1.5717,
      "step": 6678
    },
    {
      "epoch": 0.29700769196567517,
      "grad_norm": 0.08736233413219452,
      "learning_rate": 0.000989975538944542,
      "loss": 1.5741,
      "step": 6680
    },
    {
      "epoch": 0.29709661642434754,
      "grad_norm": 0.08750621974468231,
      "learning_rate": 0.000989968501927184,
      "loss": 1.5702,
      "step": 6682
    },
    {
      "epoch": 0.29718554088301985,
      "grad_norm": 0.08598420023918152,
      "learning_rate": 0.0009899614624657817,
      "loss": 1.5622,
      "step": 6684
    },
    {
      "epoch": 0.2972744653416922,
      "grad_norm": 0.08676007390022278,
      "learning_rate": 0.0009899544205603705,
      "loss": 1.5741,
      "step": 6686
    },
    {
      "epoch": 0.2973633898003646,
      "grad_norm": 0.08978342264890671,
      "learning_rate": 0.0009899473762109855,
      "loss": 1.5686,
      "step": 6688
    },
    {
      "epoch": 0.29745231425903695,
      "grad_norm": 0.08682754635810852,
      "learning_rate": 0.0009899403294176613,
      "loss": 1.5657,
      "step": 6690
    },
    {
      "epoch": 0.2975412387177093,
      "grad_norm": 0.08936390280723572,
      "learning_rate": 0.0009899332801804335,
      "loss": 1.5699,
      "step": 6692
    },
    {
      "epoch": 0.2976301631763817,
      "grad_norm": 0.09042416512966156,
      "learning_rate": 0.0009899262284993372,
      "loss": 1.5695,
      "step": 6694
    },
    {
      "epoch": 0.297719087635054,
      "grad_norm": 0.08619032055139542,
      "learning_rate": 0.0009899191743744076,
      "loss": 1.5632,
      "step": 6696
    },
    {
      "epoch": 0.29780801209372637,
      "grad_norm": 0.0877993255853653,
      "learning_rate": 0.0009899121178056798,
      "loss": 1.5744,
      "step": 6698
    },
    {
      "epoch": 0.29789693655239874,
      "grad_norm": 0.09133929759263992,
      "learning_rate": 0.000989905058793189,
      "loss": 1.5672,
      "step": 6700
    },
    {
      "epoch": 0.2979858610110711,
      "grad_norm": 0.08733952045440674,
      "learning_rate": 0.0009898979973369706,
      "loss": 1.5676,
      "step": 6702
    },
    {
      "epoch": 0.2980747854697435,
      "grad_norm": 0.08532746136188507,
      "learning_rate": 0.0009898909334370594,
      "loss": 1.5743,
      "step": 6704
    },
    {
      "epoch": 0.2981637099284158,
      "grad_norm": 0.08905241638422012,
      "learning_rate": 0.0009898838670934912,
      "loss": 1.5686,
      "step": 6706
    },
    {
      "epoch": 0.29825263438708816,
      "grad_norm": 0.09203900396823883,
      "learning_rate": 0.0009898767983063008,
      "loss": 1.576,
      "step": 6708
    },
    {
      "epoch": 0.2983415588457605,
      "grad_norm": 0.20671740174293518,
      "learning_rate": 0.0009898697270755237,
      "loss": 1.5761,
      "step": 6710
    },
    {
      "epoch": 0.2984304833044329,
      "grad_norm": 0.22460798919200897,
      "learning_rate": 0.000989862653401195,
      "loss": 1.5757,
      "step": 6712
    },
    {
      "epoch": 0.29851940776310526,
      "grad_norm": 0.11507432907819748,
      "learning_rate": 0.00098985557728335,
      "loss": 1.5788,
      "step": 6714
    },
    {
      "epoch": 0.2986083322217776,
      "grad_norm": 0.09615212678909302,
      "learning_rate": 0.0009898484987220242,
      "loss": 1.5698,
      "step": 6716
    },
    {
      "epoch": 0.29869725668044994,
      "grad_norm": 0.14628173410892487,
      "learning_rate": 0.0009898414177172527,
      "loss": 1.5708,
      "step": 6718
    },
    {
      "epoch": 0.2987861811391223,
      "grad_norm": 0.10161300748586655,
      "learning_rate": 0.000989834334269071,
      "loss": 1.5803,
      "step": 6720
    },
    {
      "epoch": 0.2988751055977947,
      "grad_norm": 0.09993571043014526,
      "learning_rate": 0.0009898272483775143,
      "loss": 1.5728,
      "step": 6722
    },
    {
      "epoch": 0.29896403005646704,
      "grad_norm": 0.0945572704076767,
      "learning_rate": 0.0009898201600426178,
      "loss": 1.5717,
      "step": 6724
    },
    {
      "epoch": 0.2990529545151394,
      "grad_norm": 0.10044775158166885,
      "learning_rate": 0.0009898130692644173,
      "loss": 1.5734,
      "step": 6726
    },
    {
      "epoch": 0.2991418789738117,
      "grad_norm": 0.09027940034866333,
      "learning_rate": 0.0009898059760429477,
      "loss": 1.5767,
      "step": 6728
    },
    {
      "epoch": 0.2992308034324841,
      "grad_norm": 0.09142522513866425,
      "learning_rate": 0.0009897988803782448,
      "loss": 1.5732,
      "step": 6730
    },
    {
      "epoch": 0.29931972789115646,
      "grad_norm": 0.1005646288394928,
      "learning_rate": 0.0009897917822703436,
      "loss": 1.5717,
      "step": 6732
    },
    {
      "epoch": 0.29940865234982883,
      "grad_norm": 0.09144769608974457,
      "learning_rate": 0.0009897846817192796,
      "loss": 1.5635,
      "step": 6734
    },
    {
      "epoch": 0.2994975768085012,
      "grad_norm": 0.09377957135438919,
      "learning_rate": 0.0009897775787250887,
      "loss": 1.5698,
      "step": 6736
    },
    {
      "epoch": 0.2995865012671735,
      "grad_norm": 0.09546922147274017,
      "learning_rate": 0.0009897704732878057,
      "loss": 1.5698,
      "step": 6738
    },
    {
      "epoch": 0.2996754257258459,
      "grad_norm": 0.08886546641588211,
      "learning_rate": 0.0009897633654074663,
      "loss": 1.5649,
      "step": 6740
    },
    {
      "epoch": 0.29976435018451825,
      "grad_norm": 0.0911850556731224,
      "learning_rate": 0.0009897562550841058,
      "loss": 1.5702,
      "step": 6742
    },
    {
      "epoch": 0.2998532746431906,
      "grad_norm": 0.09427014738321304,
      "learning_rate": 0.0009897491423177598,
      "loss": 1.5766,
      "step": 6744
    },
    {
      "epoch": 0.299942199101863,
      "grad_norm": 0.1038573682308197,
      "learning_rate": 0.000989742027108464,
      "loss": 1.574,
      "step": 6746
    },
    {
      "epoch": 0.30003112356053535,
      "grad_norm": 0.08610797673463821,
      "learning_rate": 0.0009897349094562535,
      "loss": 1.5718,
      "step": 6748
    },
    {
      "epoch": 0.30012004801920766,
      "grad_norm": 0.09136275202035904,
      "learning_rate": 0.0009897277893611642,
      "loss": 1.5716,
      "step": 6750
    },
    {
      "epoch": 0.30020897247788003,
      "grad_norm": 0.09047017991542816,
      "learning_rate": 0.0009897206668232313,
      "loss": 1.5739,
      "step": 6752
    },
    {
      "epoch": 0.3002978969365524,
      "grad_norm": 0.08960193395614624,
      "learning_rate": 0.0009897135418424903,
      "loss": 1.5674,
      "step": 6754
    },
    {
      "epoch": 0.30038682139522477,
      "grad_norm": 0.08961663395166397,
      "learning_rate": 0.000989706414418977,
      "loss": 1.5759,
      "step": 6756
    },
    {
      "epoch": 0.30047574585389714,
      "grad_norm": 0.08908514678478241,
      "learning_rate": 0.0009896992845527266,
      "loss": 1.5714,
      "step": 6758
    },
    {
      "epoch": 0.30056467031256945,
      "grad_norm": 0.08930535614490509,
      "learning_rate": 0.000989692152243775,
      "loss": 1.5694,
      "step": 6760
    },
    {
      "epoch": 0.3006535947712418,
      "grad_norm": 0.09298436343669891,
      "learning_rate": 0.0009896850174921578,
      "loss": 1.5753,
      "step": 6762
    },
    {
      "epoch": 0.3007425192299142,
      "grad_norm": 0.08789312839508057,
      "learning_rate": 0.0009896778802979103,
      "loss": 1.5727,
      "step": 6764
    },
    {
      "epoch": 0.30083144368858655,
      "grad_norm": 0.08950556069612503,
      "learning_rate": 0.0009896707406610683,
      "loss": 1.5732,
      "step": 6766
    },
    {
      "epoch": 0.3009203681472589,
      "grad_norm": 0.0896010622382164,
      "learning_rate": 0.0009896635985816673,
      "loss": 1.5746,
      "step": 6768
    },
    {
      "epoch": 0.3010092926059313,
      "grad_norm": 0.0874442532658577,
      "learning_rate": 0.0009896564540597431,
      "loss": 1.5708,
      "step": 6770
    },
    {
      "epoch": 0.3010982170646036,
      "grad_norm": 0.08776126801967621,
      "learning_rate": 0.0009896493070953311,
      "loss": 1.5758,
      "step": 6772
    },
    {
      "epoch": 0.30118714152327597,
      "grad_norm": 0.08940175920724869,
      "learning_rate": 0.0009896421576884673,
      "loss": 1.5719,
      "step": 6774
    },
    {
      "epoch": 0.30127606598194834,
      "grad_norm": 0.08825140446424484,
      "learning_rate": 0.000989635005839187,
      "loss": 1.5693,
      "step": 6776
    },
    {
      "epoch": 0.3013649904406207,
      "grad_norm": 0.08824607729911804,
      "learning_rate": 0.000989627851547526,
      "loss": 1.5658,
      "step": 6778
    },
    {
      "epoch": 0.3014539148992931,
      "grad_norm": 0.08684536069631577,
      "learning_rate": 0.00098962069481352,
      "loss": 1.5636,
      "step": 6780
    },
    {
      "epoch": 0.3015428393579654,
      "grad_norm": 0.08771684765815735,
      "learning_rate": 0.0009896135356372046,
      "loss": 1.573,
      "step": 6782
    },
    {
      "epoch": 0.30163176381663775,
      "grad_norm": 0.08880012482404709,
      "learning_rate": 0.0009896063740186159,
      "loss": 1.5673,
      "step": 6784
    },
    {
      "epoch": 0.3017206882753101,
      "grad_norm": 0.08657423406839371,
      "learning_rate": 0.000989599209957789,
      "loss": 1.5675,
      "step": 6786
    },
    {
      "epoch": 0.3018096127339825,
      "grad_norm": 0.0867740735411644,
      "learning_rate": 0.0009895920434547603,
      "loss": 1.5677,
      "step": 6788
    },
    {
      "epoch": 0.30189853719265486,
      "grad_norm": 0.08525516837835312,
      "learning_rate": 0.000989584874509565,
      "loss": 1.5654,
      "step": 6790
    },
    {
      "epoch": 0.30198746165132717,
      "grad_norm": 0.08830083906650543,
      "learning_rate": 0.0009895777031222392,
      "loss": 1.5672,
      "step": 6792
    },
    {
      "epoch": 0.30207638610999954,
      "grad_norm": 0.0843716561794281,
      "learning_rate": 0.0009895705292928185,
      "loss": 1.5728,
      "step": 6794
    },
    {
      "epoch": 0.3021653105686719,
      "grad_norm": 0.0873875841498375,
      "learning_rate": 0.0009895633530213387,
      "loss": 1.5615,
      "step": 6796
    },
    {
      "epoch": 0.3022542350273443,
      "grad_norm": 0.08721845597028732,
      "learning_rate": 0.0009895561743078357,
      "loss": 1.5626,
      "step": 6798
    },
    {
      "epoch": 0.30234315948601664,
      "grad_norm": 0.09410282224416733,
      "learning_rate": 0.0009895489931523451,
      "loss": 1.5711,
      "step": 6800
    },
    {
      "epoch": 0.302432083944689,
      "grad_norm": 0.0908014252781868,
      "learning_rate": 0.000989541809554903,
      "loss": 1.5703,
      "step": 6802
    },
    {
      "epoch": 0.3025210084033613,
      "grad_norm": 0.08996500074863434,
      "learning_rate": 0.0009895346235155452,
      "loss": 1.5712,
      "step": 6804
    },
    {
      "epoch": 0.3026099328620337,
      "grad_norm": 0.08390094339847565,
      "learning_rate": 0.000989527435034307,
      "loss": 1.5733,
      "step": 6806
    },
    {
      "epoch": 0.30269885732070606,
      "grad_norm": 0.08549515902996063,
      "learning_rate": 0.0009895202441112252,
      "loss": 1.5645,
      "step": 6808
    },
    {
      "epoch": 0.30278778177937843,
      "grad_norm": 0.08361940085887909,
      "learning_rate": 0.000989513050746335,
      "loss": 1.569,
      "step": 6810
    },
    {
      "epoch": 0.3028767062380508,
      "grad_norm": 0.08554235100746155,
      "learning_rate": 0.0009895058549396723,
      "loss": 1.5686,
      "step": 6812
    },
    {
      "epoch": 0.3029656306967231,
      "grad_norm": 0.08416851609945297,
      "learning_rate": 0.0009894986566912733,
      "loss": 1.5664,
      "step": 6814
    },
    {
      "epoch": 0.3030545551553955,
      "grad_norm": 0.08411978930234909,
      "learning_rate": 0.0009894914560011736,
      "loss": 1.5648,
      "step": 6816
    },
    {
      "epoch": 0.30314347961406785,
      "grad_norm": 0.08546824753284454,
      "learning_rate": 0.0009894842528694095,
      "loss": 1.5651,
      "step": 6818
    },
    {
      "epoch": 0.3032324040727402,
      "grad_norm": 0.08174564689397812,
      "learning_rate": 0.0009894770472960164,
      "loss": 1.5734,
      "step": 6820
    },
    {
      "epoch": 0.3033213285314126,
      "grad_norm": 0.08649101853370667,
      "learning_rate": 0.0009894698392810307,
      "loss": 1.5737,
      "step": 6822
    },
    {
      "epoch": 0.30341025299008495,
      "grad_norm": 0.0850704163312912,
      "learning_rate": 0.0009894626288244883,
      "loss": 1.5716,
      "step": 6824
    },
    {
      "epoch": 0.30349917744875726,
      "grad_norm": 0.08794053643941879,
      "learning_rate": 0.0009894554159264248,
      "loss": 1.569,
      "step": 6826
    },
    {
      "epoch": 0.30358810190742963,
      "grad_norm": 0.08536579459905624,
      "learning_rate": 0.0009894482005868767,
      "loss": 1.5673,
      "step": 6828
    },
    {
      "epoch": 0.303677026366102,
      "grad_norm": 0.08639626950025558,
      "learning_rate": 0.0009894409828058795,
      "loss": 1.5668,
      "step": 6830
    },
    {
      "epoch": 0.30376595082477437,
      "grad_norm": 0.0900898426771164,
      "learning_rate": 0.0009894337625834695,
      "loss": 1.566,
      "step": 6832
    },
    {
      "epoch": 0.30385487528344673,
      "grad_norm": 0.07988198101520538,
      "learning_rate": 0.0009894265399196828,
      "loss": 1.5659,
      "step": 6834
    },
    {
      "epoch": 0.30394379974211905,
      "grad_norm": 0.08728215098381042,
      "learning_rate": 0.0009894193148145551,
      "loss": 1.5667,
      "step": 6836
    },
    {
      "epoch": 0.3040327242007914,
      "grad_norm": 0.08969441801309586,
      "learning_rate": 0.000989412087268123,
      "loss": 1.5712,
      "step": 6838
    },
    {
      "epoch": 0.3041216486594638,
      "grad_norm": 0.0852469727396965,
      "learning_rate": 0.0009894048572804217,
      "loss": 1.576,
      "step": 6840
    },
    {
      "epoch": 0.30421057311813615,
      "grad_norm": 0.08066057413816452,
      "learning_rate": 0.000989397624851488,
      "loss": 1.5643,
      "step": 6842
    },
    {
      "epoch": 0.3042994975768085,
      "grad_norm": 0.08737246692180634,
      "learning_rate": 0.0009893903899813575,
      "loss": 1.5672,
      "step": 6844
    },
    {
      "epoch": 0.30438842203548083,
      "grad_norm": 0.08560658246278763,
      "learning_rate": 0.0009893831526700666,
      "loss": 1.5696,
      "step": 6846
    },
    {
      "epoch": 0.3044773464941532,
      "grad_norm": 0.0847938284277916,
      "learning_rate": 0.0009893759129176514,
      "loss": 1.5641,
      "step": 6848
    },
    {
      "epoch": 0.30456627095282557,
      "grad_norm": 0.08661984652280807,
      "learning_rate": 0.000989368670724148,
      "loss": 1.5687,
      "step": 6850
    },
    {
      "epoch": 0.30465519541149794,
      "grad_norm": 0.08374079316854477,
      "learning_rate": 0.000989361426089592,
      "loss": 1.5665,
      "step": 6852
    },
    {
      "epoch": 0.3047441198701703,
      "grad_norm": 0.08264697343111038,
      "learning_rate": 0.0009893541790140202,
      "loss": 1.5669,
      "step": 6854
    },
    {
      "epoch": 0.3048330443288427,
      "grad_norm": 0.08502277731895447,
      "learning_rate": 0.0009893469294974687,
      "loss": 1.5616,
      "step": 6856
    },
    {
      "epoch": 0.304921968787515,
      "grad_norm": 0.08124475181102753,
      "learning_rate": 0.0009893396775399732,
      "loss": 1.5685,
      "step": 6858
    },
    {
      "epoch": 0.30501089324618735,
      "grad_norm": 0.08523332327604294,
      "learning_rate": 0.0009893324231415703,
      "loss": 1.5616,
      "step": 6860
    },
    {
      "epoch": 0.3050998177048597,
      "grad_norm": 0.08559540659189224,
      "learning_rate": 0.0009893251663022963,
      "loss": 1.5699,
      "step": 6862
    },
    {
      "epoch": 0.3051887421635321,
      "grad_norm": 0.08699463307857513,
      "learning_rate": 0.000989317907022187,
      "loss": 1.5725,
      "step": 6864
    },
    {
      "epoch": 0.30527766662220446,
      "grad_norm": 0.09094923734664917,
      "learning_rate": 0.0009893106453012785,
      "loss": 1.5698,
      "step": 6866
    },
    {
      "epoch": 0.30536659108087677,
      "grad_norm": 0.08431901037693024,
      "learning_rate": 0.0009893033811396075,
      "loss": 1.5723,
      "step": 6868
    },
    {
      "epoch": 0.30545551553954914,
      "grad_norm": 0.08522892743349075,
      "learning_rate": 0.00098929611453721,
      "loss": 1.5645,
      "step": 6870
    },
    {
      "epoch": 0.3055444399982215,
      "grad_norm": 0.08593329787254333,
      "learning_rate": 0.000989288845494122,
      "loss": 1.5708,
      "step": 6872
    },
    {
      "epoch": 0.3056333644568939,
      "grad_norm": 0.08536642789840698,
      "learning_rate": 0.0009892815740103805,
      "loss": 1.5676,
      "step": 6874
    },
    {
      "epoch": 0.30572228891556624,
      "grad_norm": 0.08635646104812622,
      "learning_rate": 0.000989274300086021,
      "loss": 1.5669,
      "step": 6876
    },
    {
      "epoch": 0.3058112133742386,
      "grad_norm": 0.09484677016735077,
      "learning_rate": 0.00098926702372108,
      "loss": 1.5701,
      "step": 6878
    },
    {
      "epoch": 0.3059001378329109,
      "grad_norm": 0.08951367437839508,
      "learning_rate": 0.000989259744915594,
      "loss": 1.5668,
      "step": 6880
    },
    {
      "epoch": 0.3059890622915833,
      "grad_norm": 0.0868958830833435,
      "learning_rate": 0.0009892524636695992,
      "loss": 1.5653,
      "step": 6882
    },
    {
      "epoch": 0.30607798675025566,
      "grad_norm": 0.08445209264755249,
      "learning_rate": 0.0009892451799831318,
      "loss": 1.5696,
      "step": 6884
    },
    {
      "epoch": 0.306166911208928,
      "grad_norm": 0.08534303307533264,
      "learning_rate": 0.0009892378938562284,
      "loss": 1.5702,
      "step": 6886
    },
    {
      "epoch": 0.3062558356676004,
      "grad_norm": 0.09028670191764832,
      "learning_rate": 0.0009892306052889251,
      "loss": 1.5702,
      "step": 6888
    },
    {
      "epoch": 0.3063447601262727,
      "grad_norm": 0.08568151295185089,
      "learning_rate": 0.0009892233142812584,
      "loss": 1.5752,
      "step": 6890
    },
    {
      "epoch": 0.3064336845849451,
      "grad_norm": 0.08130050450563431,
      "learning_rate": 0.0009892160208332644,
      "loss": 1.5648,
      "step": 6892
    },
    {
      "epoch": 0.30652260904361744,
      "grad_norm": 0.08191067725419998,
      "learning_rate": 0.0009892087249449798,
      "loss": 1.5686,
      "step": 6894
    },
    {
      "epoch": 0.3066115335022898,
      "grad_norm": 0.08696281164884567,
      "learning_rate": 0.0009892014266164408,
      "loss": 1.5645,
      "step": 6896
    },
    {
      "epoch": 0.3067004579609622,
      "grad_norm": 0.09006158262491226,
      "learning_rate": 0.000989194125847684,
      "loss": 1.5665,
      "step": 6898
    },
    {
      "epoch": 0.3067893824196345,
      "grad_norm": 0.08667782694101334,
      "learning_rate": 0.0009891868226387455,
      "loss": 1.5627,
      "step": 6900
    },
    {
      "epoch": 0.30687830687830686,
      "grad_norm": 0.08659591525793076,
      "learning_rate": 0.0009891795169896622,
      "loss": 1.5606,
      "step": 6902
    },
    {
      "epoch": 0.30696723133697923,
      "grad_norm": 0.0922919362783432,
      "learning_rate": 0.0009891722089004702,
      "loss": 1.5638,
      "step": 6904
    },
    {
      "epoch": 0.3070561557956516,
      "grad_norm": 0.09142550826072693,
      "learning_rate": 0.0009891648983712058,
      "loss": 1.5736,
      "step": 6906
    },
    {
      "epoch": 0.30714508025432397,
      "grad_norm": 0.085682712495327,
      "learning_rate": 0.0009891575854019058,
      "loss": 1.5656,
      "step": 6908
    },
    {
      "epoch": 0.30723400471299633,
      "grad_norm": 0.0815911665558815,
      "learning_rate": 0.0009891502699926067,
      "loss": 1.5646,
      "step": 6910
    },
    {
      "epoch": 0.30732292917166865,
      "grad_norm": 0.08682998269796371,
      "learning_rate": 0.0009891429521433448,
      "loss": 1.5631,
      "step": 6912
    },
    {
      "epoch": 0.307411853630341,
      "grad_norm": 0.0891093984246254,
      "learning_rate": 0.0009891356318541566,
      "loss": 1.5678,
      "step": 6914
    },
    {
      "epoch": 0.3075007780890134,
      "grad_norm": 0.08850381523370743,
      "learning_rate": 0.000989128309125079,
      "loss": 1.5682,
      "step": 6916
    },
    {
      "epoch": 0.30758970254768575,
      "grad_norm": 0.08838870376348495,
      "learning_rate": 0.0009891209839561477,
      "loss": 1.5721,
      "step": 6918
    },
    {
      "epoch": 0.3076786270063581,
      "grad_norm": 0.08760442584753036,
      "learning_rate": 0.0009891136563474,
      "loss": 1.5692,
      "step": 6920
    },
    {
      "epoch": 0.30776755146503043,
      "grad_norm": 0.08644498139619827,
      "learning_rate": 0.0009891063262988723,
      "loss": 1.566,
      "step": 6922
    },
    {
      "epoch": 0.3078564759237028,
      "grad_norm": 0.08855952322483063,
      "learning_rate": 0.0009890989938106008,
      "loss": 1.567,
      "step": 6924
    },
    {
      "epoch": 0.30794540038237517,
      "grad_norm": 0.08846352994441986,
      "learning_rate": 0.0009890916588826224,
      "loss": 1.5695,
      "step": 6926
    },
    {
      "epoch": 0.30803432484104754,
      "grad_norm": 0.08989361673593521,
      "learning_rate": 0.0009890843215149736,
      "loss": 1.568,
      "step": 6928
    },
    {
      "epoch": 0.3081232492997199,
      "grad_norm": 0.0872996598482132,
      "learning_rate": 0.0009890769817076912,
      "loss": 1.5655,
      "step": 6930
    },
    {
      "epoch": 0.30821217375839227,
      "grad_norm": 0.08718729019165039,
      "learning_rate": 0.0009890696394608115,
      "loss": 1.5768,
      "step": 6932
    },
    {
      "epoch": 0.3083010982170646,
      "grad_norm": 0.08712806552648544,
      "learning_rate": 0.0009890622947743713,
      "loss": 1.5647,
      "step": 6934
    },
    {
      "epoch": 0.30839002267573695,
      "grad_norm": 0.0897906944155693,
      "learning_rate": 0.0009890549476484073,
      "loss": 1.5751,
      "step": 6936
    },
    {
      "epoch": 0.3084789471344093,
      "grad_norm": 0.1027601882815361,
      "learning_rate": 0.0009890475980829558,
      "loss": 1.5664,
      "step": 6938
    },
    {
      "epoch": 0.3085678715930817,
      "grad_norm": 0.11154686659574509,
      "learning_rate": 0.0009890402460780538,
      "loss": 1.5664,
      "step": 6940
    },
    {
      "epoch": 0.30865679605175406,
      "grad_norm": 0.19219696521759033,
      "learning_rate": 0.000989032891633738,
      "loss": 1.5769,
      "step": 6942
    },
    {
      "epoch": 0.30874572051042637,
      "grad_norm": 0.10125477612018585,
      "learning_rate": 0.0009890255347500447,
      "loss": 1.5801,
      "step": 6944
    },
    {
      "epoch": 0.30883464496909874,
      "grad_norm": 0.10321390628814697,
      "learning_rate": 0.0009890181754270112,
      "loss": 1.5713,
      "step": 6946
    },
    {
      "epoch": 0.3089235694277711,
      "grad_norm": 0.09801045060157776,
      "learning_rate": 0.0009890108136646736,
      "loss": 1.5664,
      "step": 6948
    },
    {
      "epoch": 0.3090124938864435,
      "grad_norm": 0.09143440425395966,
      "learning_rate": 0.0009890034494630688,
      "loss": 1.5759,
      "step": 6950
    },
    {
      "epoch": 0.30910141834511584,
      "grad_norm": 0.09067144989967346,
      "learning_rate": 0.000988996082822234,
      "loss": 1.5677,
      "step": 6952
    },
    {
      "epoch": 0.3091903428037882,
      "grad_norm": 0.09082882851362228,
      "learning_rate": 0.0009889887137422054,
      "loss": 1.5746,
      "step": 6954
    },
    {
      "epoch": 0.3092792672624605,
      "grad_norm": 0.087799571454525,
      "learning_rate": 0.0009889813422230198,
      "loss": 1.5667,
      "step": 6956
    },
    {
      "epoch": 0.3093681917211329,
      "grad_norm": 0.08629163354635239,
      "learning_rate": 0.000988973968264714,
      "loss": 1.5645,
      "step": 6958
    },
    {
      "epoch": 0.30945711617980526,
      "grad_norm": 0.08560052514076233,
      "learning_rate": 0.000988966591867325,
      "loss": 1.5647,
      "step": 6960
    },
    {
      "epoch": 0.3095460406384776,
      "grad_norm": 0.08497489988803864,
      "learning_rate": 0.0009889592130308898,
      "loss": 1.5611,
      "step": 6962
    },
    {
      "epoch": 0.30963496509715,
      "grad_norm": 0.08397437632083893,
      "learning_rate": 0.0009889518317554445,
      "loss": 1.5667,
      "step": 6964
    },
    {
      "epoch": 0.3097238895558223,
      "grad_norm": 0.08864140510559082,
      "learning_rate": 0.0009889444480410267,
      "loss": 1.5644,
      "step": 6966
    },
    {
      "epoch": 0.3098128140144947,
      "grad_norm": 0.0851002037525177,
      "learning_rate": 0.0009889370618876723,
      "loss": 1.5669,
      "step": 6968
    },
    {
      "epoch": 0.30990173847316704,
      "grad_norm": 0.08869633823633194,
      "learning_rate": 0.000988929673295419,
      "loss": 1.5676,
      "step": 6970
    },
    {
      "epoch": 0.3099906629318394,
      "grad_norm": 0.08750467002391815,
      "learning_rate": 0.0009889222822643032,
      "loss": 1.5641,
      "step": 6972
    },
    {
      "epoch": 0.3100795873905118,
      "grad_norm": 0.08245990425348282,
      "learning_rate": 0.000988914888794362,
      "loss": 1.5615,
      "step": 6974
    },
    {
      "epoch": 0.3101685118491841,
      "grad_norm": 0.08736477047204971,
      "learning_rate": 0.0009889074928856323,
      "loss": 1.5645,
      "step": 6976
    },
    {
      "epoch": 0.31025743630785646,
      "grad_norm": 0.09274744242429733,
      "learning_rate": 0.0009889000945381506,
      "loss": 1.5684,
      "step": 6978
    },
    {
      "epoch": 0.31034636076652883,
      "grad_norm": 0.08163709938526154,
      "learning_rate": 0.0009888926937519542,
      "loss": 1.5637,
      "step": 6980
    },
    {
      "epoch": 0.3104352852252012,
      "grad_norm": 0.08836798369884491,
      "learning_rate": 0.00098888529052708,
      "loss": 1.565,
      "step": 6982
    },
    {
      "epoch": 0.31052420968387356,
      "grad_norm": 0.08659311383962631,
      "learning_rate": 0.0009888778848635646,
      "loss": 1.5653,
      "step": 6984
    },
    {
      "epoch": 0.31061313414254593,
      "grad_norm": 0.08404355496168137,
      "learning_rate": 0.0009888704767614452,
      "loss": 1.5669,
      "step": 6986
    },
    {
      "epoch": 0.31070205860121825,
      "grad_norm": 0.08402011543512344,
      "learning_rate": 0.000988863066220759,
      "loss": 1.5588,
      "step": 6988
    },
    {
      "epoch": 0.3107909830598906,
      "grad_norm": 0.08973820507526398,
      "learning_rate": 0.0009888556532415423,
      "loss": 1.5668,
      "step": 6990
    },
    {
      "epoch": 0.310879907518563,
      "grad_norm": 0.08140117675065994,
      "learning_rate": 0.0009888482378238325,
      "loss": 1.5649,
      "step": 6992
    },
    {
      "epoch": 0.31096883197723535,
      "grad_norm": 0.08229751884937286,
      "learning_rate": 0.0009888408199676666,
      "loss": 1.5577,
      "step": 6994
    },
    {
      "epoch": 0.3110577564359077,
      "grad_norm": 0.08518966287374496,
      "learning_rate": 0.0009888333996730817,
      "loss": 1.5675,
      "step": 6996
    },
    {
      "epoch": 0.31114668089458003,
      "grad_norm": 0.0844472348690033,
      "learning_rate": 0.0009888259769401148,
      "loss": 1.5685,
      "step": 6998
    },
    {
      "epoch": 0.3112356053532524,
      "grad_norm": 0.08895467966794968,
      "learning_rate": 0.0009888185517688023,
      "loss": 1.5693,
      "step": 7000
    },
    {
      "epoch": 0.3112356053532524,
      "eval_loss": 1.5398365259170532,
      "eval_runtime": 12.9215,
      "eval_samples_per_second": 534.768,
      "eval_steps_per_second": 66.865,
      "step": 7000
    },
    {
      "epoch": 0.31132452981192477,
      "grad_norm": 0.08610174804925919,
      "learning_rate": 0.0009888111241591819,
      "loss": 1.5637,
      "step": 7002
    },
    {
      "epoch": 0.31141345427059713,
      "grad_norm": 0.08628767728805542,
      "learning_rate": 0.0009888036941112907,
      "loss": 1.5625,
      "step": 7004
    },
    {
      "epoch": 0.3115023787292695,
      "grad_norm": 0.08559640496969223,
      "learning_rate": 0.0009887962616251654,
      "loss": 1.5633,
      "step": 7006
    },
    {
      "epoch": 0.31159130318794187,
      "grad_norm": 0.08582499623298645,
      "learning_rate": 0.0009887888267008432,
      "loss": 1.5718,
      "step": 7008
    },
    {
      "epoch": 0.3116802276466142,
      "grad_norm": 0.08453387767076492,
      "learning_rate": 0.0009887813893383612,
      "loss": 1.5644,
      "step": 7010
    },
    {
      "epoch": 0.31176915210528655,
      "grad_norm": 0.0880834087729454,
      "learning_rate": 0.0009887739495377564,
      "loss": 1.5675,
      "step": 7012
    },
    {
      "epoch": 0.3118580765639589,
      "grad_norm": 0.08896942436695099,
      "learning_rate": 0.0009887665072990661,
      "loss": 1.5601,
      "step": 7014
    },
    {
      "epoch": 0.3119470010226313,
      "grad_norm": 0.08615525811910629,
      "learning_rate": 0.0009887590626223273,
      "loss": 1.5647,
      "step": 7016
    },
    {
      "epoch": 0.31203592548130366,
      "grad_norm": 0.08899851888418198,
      "learning_rate": 0.0009887516155075772,
      "loss": 1.5663,
      "step": 7018
    },
    {
      "epoch": 0.31212484993997597,
      "grad_norm": 0.08589426428079605,
      "learning_rate": 0.000988744165954853,
      "loss": 1.5629,
      "step": 7020
    },
    {
      "epoch": 0.31221377439864834,
      "grad_norm": 0.0895439013838768,
      "learning_rate": 0.0009887367139641916,
      "loss": 1.5655,
      "step": 7022
    },
    {
      "epoch": 0.3123026988573207,
      "grad_norm": 0.08594179153442383,
      "learning_rate": 0.0009887292595356304,
      "loss": 1.5574,
      "step": 7024
    },
    {
      "epoch": 0.31239162331599307,
      "grad_norm": 0.08816447108983994,
      "learning_rate": 0.0009887218026692066,
      "loss": 1.5686,
      "step": 7026
    },
    {
      "epoch": 0.31248054777466544,
      "grad_norm": 0.08560863137245178,
      "learning_rate": 0.0009887143433649573,
      "loss": 1.5662,
      "step": 7028
    },
    {
      "epoch": 0.31256947223333775,
      "grad_norm": 0.0848335549235344,
      "learning_rate": 0.0009887068816229196,
      "loss": 1.5647,
      "step": 7030
    },
    {
      "epoch": 0.3126583966920101,
      "grad_norm": 0.08963989466428757,
      "learning_rate": 0.000988699417443131,
      "loss": 1.5681,
      "step": 7032
    },
    {
      "epoch": 0.3127473211506825,
      "grad_norm": 0.08423878997564316,
      "learning_rate": 0.0009886919508256286,
      "loss": 1.5634,
      "step": 7034
    },
    {
      "epoch": 0.31283624560935486,
      "grad_norm": 0.08075986057519913,
      "learning_rate": 0.0009886844817704496,
      "loss": 1.5645,
      "step": 7036
    },
    {
      "epoch": 0.3129251700680272,
      "grad_norm": 0.07971926033496857,
      "learning_rate": 0.0009886770102776313,
      "loss": 1.5615,
      "step": 7038
    },
    {
      "epoch": 0.3130140945266996,
      "grad_norm": 0.08668291568756104,
      "learning_rate": 0.0009886695363472107,
      "loss": 1.5657,
      "step": 7040
    },
    {
      "epoch": 0.3131030189853719,
      "grad_norm": 0.08376525342464447,
      "learning_rate": 0.0009886620599792258,
      "loss": 1.5634,
      "step": 7042
    },
    {
      "epoch": 0.3131919434440443,
      "grad_norm": 0.0836341604590416,
      "learning_rate": 0.0009886545811737132,
      "loss": 1.5597,
      "step": 7044
    },
    {
      "epoch": 0.31328086790271664,
      "grad_norm": 0.08614721894264221,
      "learning_rate": 0.0009886470999307102,
      "loss": 1.5616,
      "step": 7046
    },
    {
      "epoch": 0.313369792361389,
      "grad_norm": 0.08370938897132874,
      "learning_rate": 0.0009886396162502547,
      "loss": 1.5695,
      "step": 7048
    },
    {
      "epoch": 0.3134587168200614,
      "grad_norm": 0.08297300338745117,
      "learning_rate": 0.0009886321301323835,
      "loss": 1.5669,
      "step": 7050
    },
    {
      "epoch": 0.3135476412787337,
      "grad_norm": 0.07920370995998383,
      "learning_rate": 0.0009886246415771343,
      "loss": 1.5671,
      "step": 7052
    },
    {
      "epoch": 0.31363656573740606,
      "grad_norm": 0.08777054399251938,
      "learning_rate": 0.000988617150584544,
      "loss": 1.5684,
      "step": 7054
    },
    {
      "epoch": 0.3137254901960784,
      "grad_norm": 0.08385943621397018,
      "learning_rate": 0.0009886096571546505,
      "loss": 1.5594,
      "step": 7056
    },
    {
      "epoch": 0.3138144146547508,
      "grad_norm": 0.08605042845010757,
      "learning_rate": 0.0009886021612874908,
      "loss": 1.5674,
      "step": 7058
    },
    {
      "epoch": 0.31390333911342316,
      "grad_norm": 0.08706225454807281,
      "learning_rate": 0.0009885946629831024,
      "loss": 1.5579,
      "step": 7060
    },
    {
      "epoch": 0.31399226357209553,
      "grad_norm": 0.08403518795967102,
      "learning_rate": 0.000988587162241523,
      "loss": 1.5624,
      "step": 7062
    },
    {
      "epoch": 0.31408118803076784,
      "grad_norm": 0.08342042565345764,
      "learning_rate": 0.0009885796590627894,
      "loss": 1.5651,
      "step": 7064
    },
    {
      "epoch": 0.3141701124894402,
      "grad_norm": 0.08213895559310913,
      "learning_rate": 0.0009885721534469395,
      "loss": 1.5653,
      "step": 7066
    },
    {
      "epoch": 0.3142590369481126,
      "grad_norm": 0.08500027656555176,
      "learning_rate": 0.0009885646453940105,
      "loss": 1.5693,
      "step": 7068
    },
    {
      "epoch": 0.31434796140678495,
      "grad_norm": 0.08024073392152786,
      "learning_rate": 0.0009885571349040401,
      "loss": 1.565,
      "step": 7070
    },
    {
      "epoch": 0.3144368858654573,
      "grad_norm": 0.08580642938613892,
      "learning_rate": 0.0009885496219770656,
      "loss": 1.5634,
      "step": 7072
    },
    {
      "epoch": 0.31452581032412963,
      "grad_norm": 0.08305441588163376,
      "learning_rate": 0.0009885421066131244,
      "loss": 1.5678,
      "step": 7074
    },
    {
      "epoch": 0.314614734782802,
      "grad_norm": 0.0872965082526207,
      "learning_rate": 0.000988534588812254,
      "loss": 1.5619,
      "step": 7076
    },
    {
      "epoch": 0.31470365924147436,
      "grad_norm": 0.08917077630758286,
      "learning_rate": 0.0009885270685744921,
      "loss": 1.5585,
      "step": 7078
    },
    {
      "epoch": 0.31479258370014673,
      "grad_norm": 0.08678824454545975,
      "learning_rate": 0.000988519545899876,
      "loss": 1.5599,
      "step": 7080
    },
    {
      "epoch": 0.3148815081588191,
      "grad_norm": 0.08195692300796509,
      "learning_rate": 0.0009885120207884434,
      "loss": 1.565,
      "step": 7082
    },
    {
      "epoch": 0.3149704326174914,
      "grad_norm": 0.08080308139324188,
      "learning_rate": 0.000988504493240232,
      "loss": 1.5663,
      "step": 7084
    },
    {
      "epoch": 0.3150593570761638,
      "grad_norm": 0.08524754643440247,
      "learning_rate": 0.0009884969632552787,
      "loss": 1.5669,
      "step": 7086
    },
    {
      "epoch": 0.31514828153483615,
      "grad_norm": 0.08281772583723068,
      "learning_rate": 0.0009884894308336215,
      "loss": 1.5625,
      "step": 7088
    },
    {
      "epoch": 0.3152372059935085,
      "grad_norm": 0.08267832547426224,
      "learning_rate": 0.0009884818959752981,
      "loss": 1.5649,
      "step": 7090
    },
    {
      "epoch": 0.3153261304521809,
      "grad_norm": 0.08353657275438309,
      "learning_rate": 0.0009884743586803457,
      "loss": 1.5639,
      "step": 7092
    },
    {
      "epoch": 0.31541505491085325,
      "grad_norm": 0.08342524617910385,
      "learning_rate": 0.0009884668189488023,
      "loss": 1.5646,
      "step": 7094
    },
    {
      "epoch": 0.31550397936952557,
      "grad_norm": 0.08173046261072159,
      "learning_rate": 0.0009884592767807053,
      "loss": 1.5661,
      "step": 7096
    },
    {
      "epoch": 0.31559290382819793,
      "grad_norm": 0.08680406212806702,
      "learning_rate": 0.0009884517321760923,
      "loss": 1.5619,
      "step": 7098
    },
    {
      "epoch": 0.3156818282868703,
      "grad_norm": 0.08208227157592773,
      "learning_rate": 0.000988444185135001,
      "loss": 1.5639,
      "step": 7100
    },
    {
      "epoch": 0.31577075274554267,
      "grad_norm": 0.0871683731675148,
      "learning_rate": 0.0009884366356574692,
      "loss": 1.5657,
      "step": 7102
    },
    {
      "epoch": 0.31585967720421504,
      "grad_norm": 0.08410225808620453,
      "learning_rate": 0.0009884290837435341,
      "loss": 1.5616,
      "step": 7104
    },
    {
      "epoch": 0.31594860166288735,
      "grad_norm": 0.08295659720897675,
      "learning_rate": 0.0009884215293932339,
      "loss": 1.5653,
      "step": 7106
    },
    {
      "epoch": 0.3160375261215597,
      "grad_norm": 0.08682281523942947,
      "learning_rate": 0.000988413972606606,
      "loss": 1.5636,
      "step": 7108
    },
    {
      "epoch": 0.3161264505802321,
      "grad_norm": 0.08395165205001831,
      "learning_rate": 0.000988406413383688,
      "loss": 1.5628,
      "step": 7110
    },
    {
      "epoch": 0.31621537503890446,
      "grad_norm": 0.08330830186605453,
      "learning_rate": 0.0009883988517245177,
      "loss": 1.5585,
      "step": 7112
    },
    {
      "epoch": 0.3163042994975768,
      "grad_norm": 0.08235388249158859,
      "learning_rate": 0.0009883912876291329,
      "loss": 1.5712,
      "step": 7114
    },
    {
      "epoch": 0.3163932239562492,
      "grad_norm": 0.08317794650793076,
      "learning_rate": 0.0009883837210975714,
      "loss": 1.5658,
      "step": 7116
    },
    {
      "epoch": 0.3164821484149215,
      "grad_norm": 0.08337473124265671,
      "learning_rate": 0.0009883761521298708,
      "loss": 1.5615,
      "step": 7118
    },
    {
      "epoch": 0.3165710728735939,
      "grad_norm": 0.08434969931840897,
      "learning_rate": 0.000988368580726069,
      "loss": 1.5617,
      "step": 7120
    },
    {
      "epoch": 0.31665999733226624,
      "grad_norm": 0.08485856652259827,
      "learning_rate": 0.0009883610068862035,
      "loss": 1.5657,
      "step": 7122
    },
    {
      "epoch": 0.3167489217909386,
      "grad_norm": 0.08327708393335342,
      "learning_rate": 0.0009883534306103121,
      "loss": 1.5656,
      "step": 7124
    },
    {
      "epoch": 0.316837846249611,
      "grad_norm": 0.08332667499780655,
      "learning_rate": 0.000988345851898433,
      "loss": 1.5645,
      "step": 7126
    },
    {
      "epoch": 0.3169267707082833,
      "grad_norm": 0.08230622857809067,
      "learning_rate": 0.0009883382707506036,
      "loss": 1.5623,
      "step": 7128
    },
    {
      "epoch": 0.31701569516695566,
      "grad_norm": 0.08090732246637344,
      "learning_rate": 0.0009883306871668618,
      "loss": 1.5695,
      "step": 7130
    },
    {
      "epoch": 0.317104619625628,
      "grad_norm": 0.08214423060417175,
      "learning_rate": 0.0009883231011472455,
      "loss": 1.5633,
      "step": 7132
    },
    {
      "epoch": 0.3171935440843004,
      "grad_norm": 0.08126901835203171,
      "learning_rate": 0.0009883155126917925,
      "loss": 1.5634,
      "step": 7134
    },
    {
      "epoch": 0.31728246854297276,
      "grad_norm": 0.08316788822412491,
      "learning_rate": 0.0009883079218005407,
      "loss": 1.5626,
      "step": 7136
    },
    {
      "epoch": 0.31737139300164513,
      "grad_norm": 0.0886952131986618,
      "learning_rate": 0.000988300328473528,
      "loss": 1.5624,
      "step": 7138
    },
    {
      "epoch": 0.31746031746031744,
      "grad_norm": 0.0825994685292244,
      "learning_rate": 0.0009882927327107922,
      "loss": 1.5671,
      "step": 7140
    },
    {
      "epoch": 0.3175492419189898,
      "grad_norm": 0.09127391129732132,
      "learning_rate": 0.000988285134512371,
      "loss": 1.5592,
      "step": 7142
    },
    {
      "epoch": 0.3176381663776622,
      "grad_norm": 0.08798851072788239,
      "learning_rate": 0.0009882775338783028,
      "loss": 1.5618,
      "step": 7144
    },
    {
      "epoch": 0.31772709083633455,
      "grad_norm": 0.09108838438987732,
      "learning_rate": 0.000988269930808625,
      "loss": 1.5644,
      "step": 7146
    },
    {
      "epoch": 0.3178160152950069,
      "grad_norm": 0.08250788599252701,
      "learning_rate": 0.0009882623253033758,
      "loss": 1.5662,
      "step": 7148
    },
    {
      "epoch": 0.3179049397536792,
      "grad_norm": 0.08027481287717819,
      "learning_rate": 0.000988254717362593,
      "loss": 1.5601,
      "step": 7150
    },
    {
      "epoch": 0.3179938642123516,
      "grad_norm": 0.08304357528686523,
      "learning_rate": 0.0009882471069863146,
      "loss": 1.5578,
      "step": 7152
    },
    {
      "epoch": 0.31808278867102396,
      "grad_norm": 0.08454793691635132,
      "learning_rate": 0.0009882394941745788,
      "loss": 1.5567,
      "step": 7154
    },
    {
      "epoch": 0.31817171312969633,
      "grad_norm": 0.08314481377601624,
      "learning_rate": 0.0009882318789274233,
      "loss": 1.5692,
      "step": 7156
    },
    {
      "epoch": 0.3182606375883687,
      "grad_norm": 0.08372924476861954,
      "learning_rate": 0.000988224261244886,
      "loss": 1.5597,
      "step": 7158
    },
    {
      "epoch": 0.318349562047041,
      "grad_norm": 0.08202824741601944,
      "learning_rate": 0.000988216641127005,
      "loss": 1.5669,
      "step": 7160
    },
    {
      "epoch": 0.3184384865057134,
      "grad_norm": 0.08950933814048767,
      "learning_rate": 0.0009882090185738186,
      "loss": 1.5682,
      "step": 7162
    },
    {
      "epoch": 0.31852741096438575,
      "grad_norm": 0.08493774384260178,
      "learning_rate": 0.0009882013935853644,
      "loss": 1.5685,
      "step": 7164
    },
    {
      "epoch": 0.3186163354230581,
      "grad_norm": 0.08362516015768051,
      "learning_rate": 0.0009881937661616806,
      "loss": 1.5685,
      "step": 7166
    },
    {
      "epoch": 0.3187052598817305,
      "grad_norm": 0.08802633732557297,
      "learning_rate": 0.0009881861363028053,
      "loss": 1.5684,
      "step": 7168
    },
    {
      "epoch": 0.31879418434040285,
      "grad_norm": 0.08594125509262085,
      "learning_rate": 0.0009881785040087765,
      "loss": 1.5581,
      "step": 7170
    },
    {
      "epoch": 0.31888310879907517,
      "grad_norm": 0.08420670032501221,
      "learning_rate": 0.0009881708692796325,
      "loss": 1.5626,
      "step": 7172
    },
    {
      "epoch": 0.31897203325774753,
      "grad_norm": 0.08396308124065399,
      "learning_rate": 0.000988163232115411,
      "loss": 1.5624,
      "step": 7174
    },
    {
      "epoch": 0.3190609577164199,
      "grad_norm": 0.08624763786792755,
      "learning_rate": 0.00098815559251615,
      "loss": 1.5667,
      "step": 7176
    },
    {
      "epoch": 0.31914988217509227,
      "grad_norm": 0.0823800191283226,
      "learning_rate": 0.0009881479504818883,
      "loss": 1.5654,
      "step": 7178
    },
    {
      "epoch": 0.31923880663376464,
      "grad_norm": 0.08368480205535889,
      "learning_rate": 0.0009881403060126635,
      "loss": 1.5673,
      "step": 7180
    },
    {
      "epoch": 0.31932773109243695,
      "grad_norm": 0.08373897522687912,
      "learning_rate": 0.0009881326591085135,
      "loss": 1.565,
      "step": 7182
    },
    {
      "epoch": 0.3194166555511093,
      "grad_norm": 0.08619046956300735,
      "learning_rate": 0.0009881250097694772,
      "loss": 1.5597,
      "step": 7184
    },
    {
      "epoch": 0.3195055800097817,
      "grad_norm": 0.08758384734392166,
      "learning_rate": 0.000988117357995592,
      "loss": 1.5618,
      "step": 7186
    },
    {
      "epoch": 0.31959450446845405,
      "grad_norm": 0.08243545889854431,
      "learning_rate": 0.0009881097037868966,
      "loss": 1.5549,
      "step": 7188
    },
    {
      "epoch": 0.3196834289271264,
      "grad_norm": 0.07898540049791336,
      "learning_rate": 0.000988102047143429,
      "loss": 1.5597,
      "step": 7190
    },
    {
      "epoch": 0.3197723533857988,
      "grad_norm": 0.08437183499336243,
      "learning_rate": 0.0009880943880652274,
      "loss": 1.5637,
      "step": 7192
    },
    {
      "epoch": 0.3198612778444711,
      "grad_norm": 0.08325666189193726,
      "learning_rate": 0.0009880867265523296,
      "loss": 1.5613,
      "step": 7194
    },
    {
      "epoch": 0.31995020230314347,
      "grad_norm": 0.08307388424873352,
      "learning_rate": 0.0009880790626047747,
      "loss": 1.5627,
      "step": 7196
    },
    {
      "epoch": 0.32003912676181584,
      "grad_norm": 0.08156055212020874,
      "learning_rate": 0.0009880713962226,
      "loss": 1.5666,
      "step": 7198
    },
    {
      "epoch": 0.3201280512204882,
      "grad_norm": 0.08196540176868439,
      "learning_rate": 0.0009880637274058443,
      "loss": 1.5627,
      "step": 7200
    },
    {
      "epoch": 0.3202169756791606,
      "grad_norm": 0.08538369089365005,
      "learning_rate": 0.0009880560561545458,
      "loss": 1.5644,
      "step": 7202
    },
    {
      "epoch": 0.3203059001378329,
      "grad_norm": 0.08306334167718887,
      "learning_rate": 0.0009880483824687427,
      "loss": 1.5582,
      "step": 7204
    },
    {
      "epoch": 0.32039482459650526,
      "grad_norm": 0.08345794677734375,
      "learning_rate": 0.0009880407063484731,
      "loss": 1.5538,
      "step": 7206
    },
    {
      "epoch": 0.3204837490551776,
      "grad_norm": 0.08639951795339584,
      "learning_rate": 0.0009880330277937757,
      "loss": 1.5601,
      "step": 7208
    },
    {
      "epoch": 0.32057267351385,
      "grad_norm": 0.08315160125494003,
      "learning_rate": 0.0009880253468046883,
      "loss": 1.5591,
      "step": 7210
    },
    {
      "epoch": 0.32066159797252236,
      "grad_norm": 0.0834464579820633,
      "learning_rate": 0.0009880176633812493,
      "loss": 1.5663,
      "step": 7212
    },
    {
      "epoch": 0.3207505224311947,
      "grad_norm": 0.08205483853816986,
      "learning_rate": 0.0009880099775234977,
      "loss": 1.5662,
      "step": 7214
    },
    {
      "epoch": 0.32083944688986704,
      "grad_norm": 0.08744161576032639,
      "learning_rate": 0.000988002289231471,
      "loss": 1.5548,
      "step": 7216
    },
    {
      "epoch": 0.3209283713485394,
      "grad_norm": 0.08676169067621231,
      "learning_rate": 0.000987994598505208,
      "loss": 1.5595,
      "step": 7218
    },
    {
      "epoch": 0.3210172958072118,
      "grad_norm": 0.08752325177192688,
      "learning_rate": 0.0009879869053447468,
      "loss": 1.5624,
      "step": 7220
    },
    {
      "epoch": 0.32110622026588415,
      "grad_norm": 0.08283118903636932,
      "learning_rate": 0.0009879792097501258,
      "loss": 1.5675,
      "step": 7222
    },
    {
      "epoch": 0.3211951447245565,
      "grad_norm": 0.08117381483316422,
      "learning_rate": 0.0009879715117213838,
      "loss": 1.5604,
      "step": 7224
    },
    {
      "epoch": 0.3212840691832288,
      "grad_norm": 0.08399328589439392,
      "learning_rate": 0.0009879638112585587,
      "loss": 1.5532,
      "step": 7226
    },
    {
      "epoch": 0.3213729936419012,
      "grad_norm": 0.0844864770770073,
      "learning_rate": 0.0009879561083616893,
      "loss": 1.5662,
      "step": 7228
    },
    {
      "epoch": 0.32146191810057356,
      "grad_norm": 0.08030157536268234,
      "learning_rate": 0.0009879484030308136,
      "loss": 1.5637,
      "step": 7230
    },
    {
      "epoch": 0.32155084255924593,
      "grad_norm": 0.08214158564805984,
      "learning_rate": 0.0009879406952659705,
      "loss": 1.5642,
      "step": 7232
    },
    {
      "epoch": 0.3216397670179183,
      "grad_norm": 0.08249089121818542,
      "learning_rate": 0.0009879329850671981,
      "loss": 1.5614,
      "step": 7234
    },
    {
      "epoch": 0.3217286914765906,
      "grad_norm": 0.0858563482761383,
      "learning_rate": 0.000987925272434535,
      "loss": 1.5587,
      "step": 7236
    },
    {
      "epoch": 0.321817615935263,
      "grad_norm": 0.08245126157999039,
      "learning_rate": 0.0009879175573680196,
      "loss": 1.5641,
      "step": 7238
    },
    {
      "epoch": 0.32190654039393535,
      "grad_norm": 0.0817423164844513,
      "learning_rate": 0.0009879098398676905,
      "loss": 1.5613,
      "step": 7240
    },
    {
      "epoch": 0.3219954648526077,
      "grad_norm": 0.08229317516088486,
      "learning_rate": 0.0009879021199335862,
      "loss": 1.558,
      "step": 7242
    },
    {
      "epoch": 0.3220843893112801,
      "grad_norm": 0.0869976207613945,
      "learning_rate": 0.000987894397565745,
      "loss": 1.556,
      "step": 7244
    },
    {
      "epoch": 0.32217331376995245,
      "grad_norm": 0.08708877116441727,
      "learning_rate": 0.0009878866727642056,
      "loss": 1.5642,
      "step": 7246
    },
    {
      "epoch": 0.32226223822862476,
      "grad_norm": 0.08197760581970215,
      "learning_rate": 0.0009878789455290064,
      "loss": 1.5577,
      "step": 7248
    },
    {
      "epoch": 0.32235116268729713,
      "grad_norm": 0.08504566550254822,
      "learning_rate": 0.0009878712158601863,
      "loss": 1.5633,
      "step": 7250
    },
    {
      "epoch": 0.3224400871459695,
      "grad_norm": 0.0837409570813179,
      "learning_rate": 0.0009878634837577832,
      "loss": 1.5607,
      "step": 7252
    },
    {
      "epoch": 0.32252901160464187,
      "grad_norm": 0.08278852701187134,
      "learning_rate": 0.0009878557492218364,
      "loss": 1.5618,
      "step": 7254
    },
    {
      "epoch": 0.32261793606331424,
      "grad_norm": 0.07882179319858551,
      "learning_rate": 0.0009878480122523842,
      "loss": 1.5579,
      "step": 7256
    },
    {
      "epoch": 0.32270686052198655,
      "grad_norm": 0.08126863837242126,
      "learning_rate": 0.0009878402728494648,
      "loss": 1.5578,
      "step": 7258
    },
    {
      "epoch": 0.3227957849806589,
      "grad_norm": 0.08077924698591232,
      "learning_rate": 0.0009878325310131173,
      "loss": 1.5598,
      "step": 7260
    },
    {
      "epoch": 0.3228847094393313,
      "grad_norm": 0.08375802636146545,
      "learning_rate": 0.0009878247867433803,
      "loss": 1.5595,
      "step": 7262
    },
    {
      "epoch": 0.32297363389800365,
      "grad_norm": 0.08182679116725922,
      "learning_rate": 0.000987817040040292,
      "loss": 1.5558,
      "step": 7264
    },
    {
      "epoch": 0.323062558356676,
      "grad_norm": 0.08209866285324097,
      "learning_rate": 0.0009878092909038916,
      "loss": 1.5582,
      "step": 7266
    },
    {
      "epoch": 0.32315148281534833,
      "grad_norm": 0.08449520170688629,
      "learning_rate": 0.0009878015393342172,
      "loss": 1.56,
      "step": 7268
    },
    {
      "epoch": 0.3232404072740207,
      "grad_norm": 0.07824712991714478,
      "learning_rate": 0.000987793785331308,
      "loss": 1.5512,
      "step": 7270
    },
    {
      "epoch": 0.32332933173269307,
      "grad_norm": 0.0843387246131897,
      "learning_rate": 0.0009877860288952022,
      "loss": 1.5599,
      "step": 7272
    },
    {
      "epoch": 0.32341825619136544,
      "grad_norm": 0.08296310156583786,
      "learning_rate": 0.000987778270025939,
      "loss": 1.5652,
      "step": 7274
    },
    {
      "epoch": 0.3235071806500378,
      "grad_norm": 0.08399280905723572,
      "learning_rate": 0.0009877705087235566,
      "loss": 1.5624,
      "step": 7276
    },
    {
      "epoch": 0.3235961051087102,
      "grad_norm": 0.08151458948850632,
      "learning_rate": 0.000987762744988094,
      "loss": 1.5661,
      "step": 7278
    },
    {
      "epoch": 0.3236850295673825,
      "grad_norm": 0.08649469912052155,
      "learning_rate": 0.00098775497881959,
      "loss": 1.5541,
      "step": 7280
    },
    {
      "epoch": 0.32377395402605486,
      "grad_norm": 0.0848914235830307,
      "learning_rate": 0.0009877472102180831,
      "loss": 1.5622,
      "step": 7282
    },
    {
      "epoch": 0.3238628784847272,
      "grad_norm": 0.08599822223186493,
      "learning_rate": 0.0009877394391836123,
      "loss": 1.5586,
      "step": 7284
    },
    {
      "epoch": 0.3239518029433996,
      "grad_norm": 0.09074999392032623,
      "learning_rate": 0.000987731665716216,
      "loss": 1.5584,
      "step": 7286
    },
    {
      "epoch": 0.32404072740207196,
      "grad_norm": 0.08198599517345428,
      "learning_rate": 0.0009877238898159332,
      "loss": 1.5623,
      "step": 7288
    },
    {
      "epoch": 0.3241296518607443,
      "grad_norm": 0.08142637461423874,
      "learning_rate": 0.0009877161114828026,
      "loss": 1.5591,
      "step": 7290
    },
    {
      "epoch": 0.32421857631941664,
      "grad_norm": 0.08540283888578415,
      "learning_rate": 0.0009877083307168633,
      "loss": 1.562,
      "step": 7292
    },
    {
      "epoch": 0.324307500778089,
      "grad_norm": 0.08285167068243027,
      "learning_rate": 0.0009877005475181539,
      "loss": 1.5581,
      "step": 7294
    },
    {
      "epoch": 0.3243964252367614,
      "grad_norm": 0.08604173362255096,
      "learning_rate": 0.000987692761886713,
      "loss": 1.5646,
      "step": 7296
    },
    {
      "epoch": 0.32448534969543374,
      "grad_norm": 0.08079639822244644,
      "learning_rate": 0.0009876849738225798,
      "loss": 1.5611,
      "step": 7298
    },
    {
      "epoch": 0.3245742741541061,
      "grad_norm": 0.08307557553052902,
      "learning_rate": 0.000987677183325793,
      "loss": 1.5594,
      "step": 7300
    },
    {
      "epoch": 0.3246631986127784,
      "grad_norm": 0.08180294185876846,
      "learning_rate": 0.0009876693903963913,
      "loss": 1.5631,
      "step": 7302
    },
    {
      "epoch": 0.3247521230714508,
      "grad_norm": 0.08266652375459671,
      "learning_rate": 0.000987661595034414,
      "loss": 1.559,
      "step": 7304
    },
    {
      "epoch": 0.32484104753012316,
      "grad_norm": 0.07898559421300888,
      "learning_rate": 0.0009876537972398995,
      "loss": 1.5604,
      "step": 7306
    },
    {
      "epoch": 0.32492997198879553,
      "grad_norm": 0.0870475322008133,
      "learning_rate": 0.000987645997012887,
      "loss": 1.5617,
      "step": 7308
    },
    {
      "epoch": 0.3250188964474679,
      "grad_norm": 0.08088637888431549,
      "learning_rate": 0.0009876381943534154,
      "loss": 1.5604,
      "step": 7310
    },
    {
      "epoch": 0.3251078209061402,
      "grad_norm": 0.08533639460802078,
      "learning_rate": 0.0009876303892615234,
      "loss": 1.5555,
      "step": 7312
    },
    {
      "epoch": 0.3251967453648126,
      "grad_norm": 0.08240604400634766,
      "learning_rate": 0.0009876225817372502,
      "loss": 1.5609,
      "step": 7314
    },
    {
      "epoch": 0.32528566982348495,
      "grad_norm": 0.08196206390857697,
      "learning_rate": 0.0009876147717806346,
      "loss": 1.5579,
      "step": 7316
    },
    {
      "epoch": 0.3253745942821573,
      "grad_norm": 0.08041080087423325,
      "learning_rate": 0.0009876069593917154,
      "loss": 1.5621,
      "step": 7318
    },
    {
      "epoch": 0.3254635187408297,
      "grad_norm": 0.07944405823945999,
      "learning_rate": 0.000987599144570532,
      "loss": 1.5583,
      "step": 7320
    },
    {
      "epoch": 0.32555244319950205,
      "grad_norm": 0.0833417996764183,
      "learning_rate": 0.000987591327317123,
      "loss": 1.5588,
      "step": 7322
    },
    {
      "epoch": 0.32564136765817436,
      "grad_norm": 0.08267171680927277,
      "learning_rate": 0.0009875835076315273,
      "loss": 1.5576,
      "step": 7324
    },
    {
      "epoch": 0.32573029211684673,
      "grad_norm": 0.08149762451648712,
      "learning_rate": 0.0009875756855137845,
      "loss": 1.5535,
      "step": 7326
    },
    {
      "epoch": 0.3258192165755191,
      "grad_norm": 0.0837164968252182,
      "learning_rate": 0.0009875678609639331,
      "loss": 1.5554,
      "step": 7328
    },
    {
      "epoch": 0.32590814103419147,
      "grad_norm": 0.08050676435232162,
      "learning_rate": 0.0009875600339820123,
      "loss": 1.561,
      "step": 7330
    },
    {
      "epoch": 0.32599706549286384,
      "grad_norm": 0.08078650385141373,
      "learning_rate": 0.000987552204568061,
      "loss": 1.5543,
      "step": 7332
    },
    {
      "epoch": 0.32608598995153615,
      "grad_norm": 0.08506888151168823,
      "learning_rate": 0.0009875443727221185,
      "loss": 1.5606,
      "step": 7334
    },
    {
      "epoch": 0.3261749144102085,
      "grad_norm": 0.08433322608470917,
      "learning_rate": 0.0009875365384442237,
      "loss": 1.5563,
      "step": 7336
    },
    {
      "epoch": 0.3262638388688809,
      "grad_norm": 0.08262262493371964,
      "learning_rate": 0.000987528701734416,
      "loss": 1.565,
      "step": 7338
    },
    {
      "epoch": 0.32635276332755325,
      "grad_norm": 0.08206149190664291,
      "learning_rate": 0.0009875208625927338,
      "loss": 1.5562,
      "step": 7340
    },
    {
      "epoch": 0.3264416877862256,
      "grad_norm": 0.08632387220859528,
      "learning_rate": 0.0009875130210192167,
      "loss": 1.5566,
      "step": 7342
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 0.08040108531713486,
      "learning_rate": 0.0009875051770139038,
      "loss": 1.5591,
      "step": 7344
    },
    {
      "epoch": 0.3266195367035703,
      "grad_norm": 0.08796191215515137,
      "learning_rate": 0.0009874973305768343,
      "loss": 1.5635,
      "step": 7346
    },
    {
      "epoch": 0.32670846116224267,
      "grad_norm": 0.08850188553333282,
      "learning_rate": 0.000987489481708047,
      "loss": 1.5547,
      "step": 7348
    },
    {
      "epoch": 0.32679738562091504,
      "grad_norm": 0.08623901754617691,
      "learning_rate": 0.000987481630407581,
      "loss": 1.5583,
      "step": 7350
    },
    {
      "epoch": 0.3268863100795874,
      "grad_norm": 0.08186235278844833,
      "learning_rate": 0.0009874737766754763,
      "loss": 1.5568,
      "step": 7352
    },
    {
      "epoch": 0.3269752345382598,
      "grad_norm": 0.08264327794313431,
      "learning_rate": 0.000987465920511771,
      "loss": 1.5589,
      "step": 7354
    },
    {
      "epoch": 0.3270641589969321,
      "grad_norm": 0.08318621665239334,
      "learning_rate": 0.000987458061916505,
      "loss": 1.5578,
      "step": 7356
    },
    {
      "epoch": 0.32715308345560445,
      "grad_norm": 0.08507546037435532,
      "learning_rate": 0.0009874502008897174,
      "loss": 1.5628,
      "step": 7358
    },
    {
      "epoch": 0.3272420079142768,
      "grad_norm": 0.08034215867519379,
      "learning_rate": 0.000987442337431447,
      "loss": 1.5597,
      "step": 7360
    },
    {
      "epoch": 0.3273309323729492,
      "grad_norm": 0.08105006068944931,
      "learning_rate": 0.0009874344715417333,
      "loss": 1.5562,
      "step": 7362
    },
    {
      "epoch": 0.32741985683162156,
      "grad_norm": 0.07907504588365555,
      "learning_rate": 0.0009874266032206156,
      "loss": 1.5649,
      "step": 7364
    },
    {
      "epoch": 0.32750878129029387,
      "grad_norm": 0.08351552486419678,
      "learning_rate": 0.0009874187324681331,
      "loss": 1.563,
      "step": 7366
    },
    {
      "epoch": 0.32759770574896624,
      "grad_norm": 0.07940589636564255,
      "learning_rate": 0.0009874108592843253,
      "loss": 1.5597,
      "step": 7368
    },
    {
      "epoch": 0.3276866302076386,
      "grad_norm": 0.0827498808503151,
      "learning_rate": 0.000987402983669231,
      "loss": 1.5544,
      "step": 7370
    },
    {
      "epoch": 0.327775554666311,
      "grad_norm": 0.08331042528152466,
      "learning_rate": 0.0009873951056228898,
      "loss": 1.56,
      "step": 7372
    },
    {
      "epoch": 0.32786447912498334,
      "grad_norm": 0.08180607855319977,
      "learning_rate": 0.000987387225145341,
      "loss": 1.5565,
      "step": 7374
    },
    {
      "epoch": 0.3279534035836557,
      "grad_norm": 0.08340569585561752,
      "learning_rate": 0.0009873793422366234,
      "loss": 1.5608,
      "step": 7376
    },
    {
      "epoch": 0.328042328042328,
      "grad_norm": 0.08175882697105408,
      "learning_rate": 0.000987371456896777,
      "loss": 1.5607,
      "step": 7378
    },
    {
      "epoch": 0.3281312525010004,
      "grad_norm": 0.0852338895201683,
      "learning_rate": 0.000987363569125841,
      "loss": 1.5584,
      "step": 7380
    },
    {
      "epoch": 0.32822017695967276,
      "grad_norm": 0.0846090242266655,
      "learning_rate": 0.0009873556789238545,
      "loss": 1.5586,
      "step": 7382
    },
    {
      "epoch": 0.32830910141834513,
      "grad_norm": 0.07967184484004974,
      "learning_rate": 0.000987347786290857,
      "loss": 1.5576,
      "step": 7384
    },
    {
      "epoch": 0.3283980258770175,
      "grad_norm": 0.08349616080522537,
      "learning_rate": 0.000987339891226888,
      "loss": 1.5628,
      "step": 7386
    },
    {
      "epoch": 0.3284869503356898,
      "grad_norm": 0.08276060223579407,
      "learning_rate": 0.0009873319937319864,
      "loss": 1.5597,
      "step": 7388
    },
    {
      "epoch": 0.3285758747943622,
      "grad_norm": 0.08043144643306732,
      "learning_rate": 0.0009873240938061921,
      "loss": 1.5537,
      "step": 7390
    },
    {
      "epoch": 0.32866479925303455,
      "grad_norm": 0.08394316583871841,
      "learning_rate": 0.0009873161914495444,
      "loss": 1.5573,
      "step": 7392
    },
    {
      "epoch": 0.3287537237117069,
      "grad_norm": 0.08426385372877121,
      "learning_rate": 0.0009873082866620824,
      "loss": 1.558,
      "step": 7394
    },
    {
      "epoch": 0.3288426481703793,
      "grad_norm": 0.08154802769422531,
      "learning_rate": 0.000987300379443846,
      "loss": 1.5572,
      "step": 7396
    },
    {
      "epoch": 0.3289315726290516,
      "grad_norm": 0.08731735497713089,
      "learning_rate": 0.0009872924697948745,
      "loss": 1.5567,
      "step": 7398
    },
    {
      "epoch": 0.32902049708772396,
      "grad_norm": 0.08999709039926529,
      "learning_rate": 0.000987284557715207,
      "loss": 1.556,
      "step": 7400
    },
    {
      "epoch": 0.32910942154639633,
      "grad_norm": 0.08462511003017426,
      "learning_rate": 0.0009872766432048835,
      "loss": 1.5585,
      "step": 7402
    },
    {
      "epoch": 0.3291983460050687,
      "grad_norm": 0.08666058629751205,
      "learning_rate": 0.000987268726263943,
      "loss": 1.5567,
      "step": 7404
    },
    {
      "epoch": 0.32928727046374107,
      "grad_norm": 0.08502695709466934,
      "learning_rate": 0.0009872608068924253,
      "loss": 1.5566,
      "step": 7406
    },
    {
      "epoch": 0.32937619492241343,
      "grad_norm": 0.08488810807466507,
      "learning_rate": 0.0009872528850903698,
      "loss": 1.5529,
      "step": 7408
    },
    {
      "epoch": 0.32946511938108575,
      "grad_norm": 0.08286993950605392,
      "learning_rate": 0.000987244960857816,
      "loss": 1.5639,
      "step": 7410
    },
    {
      "epoch": 0.3295540438397581,
      "grad_norm": 0.0867101177573204,
      "learning_rate": 0.0009872370341948036,
      "loss": 1.559,
      "step": 7412
    },
    {
      "epoch": 0.3296429682984305,
      "grad_norm": 0.0828615054488182,
      "learning_rate": 0.0009872291051013719,
      "loss": 1.5587,
      "step": 7414
    },
    {
      "epoch": 0.32973189275710285,
      "grad_norm": 0.08587847650051117,
      "learning_rate": 0.0009872211735775605,
      "loss": 1.5599,
      "step": 7416
    },
    {
      "epoch": 0.3298208172157752,
      "grad_norm": 0.08348828554153442,
      "learning_rate": 0.000987213239623409,
      "loss": 1.5556,
      "step": 7418
    },
    {
      "epoch": 0.32990974167444753,
      "grad_norm": 0.08008415997028351,
      "learning_rate": 0.0009872053032389573,
      "loss": 1.5642,
      "step": 7420
    },
    {
      "epoch": 0.3299986661331199,
      "grad_norm": 0.0780973806977272,
      "learning_rate": 0.0009871973644242442,
      "loss": 1.5563,
      "step": 7422
    },
    {
      "epoch": 0.33008759059179227,
      "grad_norm": 0.08216983824968338,
      "learning_rate": 0.00098718942317931,
      "loss": 1.5608,
      "step": 7424
    },
    {
      "epoch": 0.33017651505046464,
      "grad_norm": 0.08794374763965607,
      "learning_rate": 0.0009871814795041941,
      "loss": 1.5622,
      "step": 7426
    },
    {
      "epoch": 0.330265439509137,
      "grad_norm": 0.09090631455183029,
      "learning_rate": 0.0009871735333989362,
      "loss": 1.5562,
      "step": 7428
    },
    {
      "epoch": 0.3303543639678094,
      "grad_norm": 0.08172343671321869,
      "learning_rate": 0.0009871655848635757,
      "loss": 1.5569,
      "step": 7430
    },
    {
      "epoch": 0.3304432884264817,
      "grad_norm": 0.08407451957464218,
      "learning_rate": 0.0009871576338981525,
      "loss": 1.5586,
      "step": 7432
    },
    {
      "epoch": 0.33053221288515405,
      "grad_norm": 0.08738183975219727,
      "learning_rate": 0.000987149680502706,
      "loss": 1.5631,
      "step": 7434
    },
    {
      "epoch": 0.3306211373438264,
      "grad_norm": 0.09222719818353653,
      "learning_rate": 0.000987141724677276,
      "loss": 1.5581,
      "step": 7436
    },
    {
      "epoch": 0.3307100618024988,
      "grad_norm": 0.08091361075639725,
      "learning_rate": 0.0009871337664219024,
      "loss": 1.5561,
      "step": 7438
    },
    {
      "epoch": 0.33079898626117116,
      "grad_norm": 0.09107191115617752,
      "learning_rate": 0.0009871258057366247,
      "loss": 1.5559,
      "step": 7440
    },
    {
      "epoch": 0.33088791071984347,
      "grad_norm": 0.08714121580123901,
      "learning_rate": 0.0009871178426214826,
      "loss": 1.5662,
      "step": 7442
    },
    {
      "epoch": 0.33097683517851584,
      "grad_norm": 0.0870809406042099,
      "learning_rate": 0.000987109877076516,
      "loss": 1.5584,
      "step": 7444
    },
    {
      "epoch": 0.3310657596371882,
      "grad_norm": 0.0869300365447998,
      "learning_rate": 0.0009871019091017641,
      "loss": 1.5596,
      "step": 7446
    },
    {
      "epoch": 0.3311546840958606,
      "grad_norm": 0.08757723122835159,
      "learning_rate": 0.0009870939386972674,
      "loss": 1.558,
      "step": 7448
    },
    {
      "epoch": 0.33124360855453294,
      "grad_norm": 0.08233436197042465,
      "learning_rate": 0.000987085965863065,
      "loss": 1.5573,
      "step": 7450
    },
    {
      "epoch": 0.33133253301320525,
      "grad_norm": 0.08261076360940933,
      "learning_rate": 0.0009870779905991971,
      "loss": 1.5639,
      "step": 7452
    },
    {
      "epoch": 0.3314214574718776,
      "grad_norm": 0.08498179167509079,
      "learning_rate": 0.0009870700129057032,
      "loss": 1.5584,
      "step": 7454
    },
    {
      "epoch": 0.33151038193055,
      "grad_norm": 0.08367268741130829,
      "learning_rate": 0.0009870620327826233,
      "loss": 1.5633,
      "step": 7456
    },
    {
      "epoch": 0.33159930638922236,
      "grad_norm": 0.08087131381034851,
      "learning_rate": 0.0009870540502299973,
      "loss": 1.5592,
      "step": 7458
    },
    {
      "epoch": 0.3316882308478947,
      "grad_norm": 0.08614420145750046,
      "learning_rate": 0.0009870460652478645,
      "loss": 1.5625,
      "step": 7460
    },
    {
      "epoch": 0.3317771553065671,
      "grad_norm": 0.08858582377433777,
      "learning_rate": 0.0009870380778362654,
      "loss": 1.5612,
      "step": 7462
    },
    {
      "epoch": 0.3318660797652394,
      "grad_norm": 0.08249551802873611,
      "learning_rate": 0.0009870300879952394,
      "loss": 1.5645,
      "step": 7464
    },
    {
      "epoch": 0.3319550042239118,
      "grad_norm": 0.08597932755947113,
      "learning_rate": 0.0009870220957248264,
      "loss": 1.5571,
      "step": 7466
    },
    {
      "epoch": 0.33204392868258414,
      "grad_norm": 0.08219864219427109,
      "learning_rate": 0.0009870141010250666,
      "loss": 1.5608,
      "step": 7468
    },
    {
      "epoch": 0.3321328531412565,
      "grad_norm": 0.08492123335599899,
      "learning_rate": 0.0009870061038959994,
      "loss": 1.5632,
      "step": 7470
    },
    {
      "epoch": 0.3322217775999289,
      "grad_norm": 0.08149886131286621,
      "learning_rate": 0.0009869981043376648,
      "loss": 1.5541,
      "step": 7472
    },
    {
      "epoch": 0.3323107020586012,
      "grad_norm": 0.08603468537330627,
      "learning_rate": 0.000986990102350103,
      "loss": 1.5528,
      "step": 7474
    },
    {
      "epoch": 0.33239962651727356,
      "grad_norm": 0.0839407667517662,
      "learning_rate": 0.0009869820979333539,
      "loss": 1.5573,
      "step": 7476
    },
    {
      "epoch": 0.33248855097594593,
      "grad_norm": 0.07948873937129974,
      "learning_rate": 0.0009869740910874569,
      "loss": 1.5646,
      "step": 7478
    },
    {
      "epoch": 0.3325774754346183,
      "grad_norm": 0.08331003040075302,
      "learning_rate": 0.0009869660818124526,
      "loss": 1.5538,
      "step": 7480
    },
    {
      "epoch": 0.33266639989329067,
      "grad_norm": 0.08100567013025284,
      "learning_rate": 0.0009869580701083805,
      "loss": 1.5628,
      "step": 7482
    },
    {
      "epoch": 0.33275532435196303,
      "grad_norm": 0.08268828690052032,
      "learning_rate": 0.0009869500559752807,
      "loss": 1.5591,
      "step": 7484
    },
    {
      "epoch": 0.33284424881063535,
      "grad_norm": 0.08128658682107925,
      "learning_rate": 0.0009869420394131932,
      "loss": 1.559,
      "step": 7486
    },
    {
      "epoch": 0.3329331732693077,
      "grad_norm": 0.07909125089645386,
      "learning_rate": 0.0009869340204221582,
      "loss": 1.5581,
      "step": 7488
    },
    {
      "epoch": 0.3330220977279801,
      "grad_norm": 0.08491683751344681,
      "learning_rate": 0.0009869259990022152,
      "loss": 1.5523,
      "step": 7490
    },
    {
      "epoch": 0.33311102218665245,
      "grad_norm": 0.08070467412471771,
      "learning_rate": 0.0009869179751534046,
      "loss": 1.5558,
      "step": 7492
    },
    {
      "epoch": 0.3331999466453248,
      "grad_norm": 0.08046550303697586,
      "learning_rate": 0.0009869099488757662,
      "loss": 1.5511,
      "step": 7494
    },
    {
      "epoch": 0.33328887110399713,
      "grad_norm": 0.08419077098369598,
      "learning_rate": 0.0009869019201693403,
      "loss": 1.5632,
      "step": 7496
    },
    {
      "epoch": 0.3333777955626695,
      "grad_norm": 0.07918644696474075,
      "learning_rate": 0.0009868938890341668,
      "loss": 1.5552,
      "step": 7498
    },
    {
      "epoch": 0.33346672002134187,
      "grad_norm": 0.08392981439828873,
      "learning_rate": 0.0009868858554702856,
      "loss": 1.5588,
      "step": 7500
    },
    {
      "epoch": 0.33346672002134187,
      "eval_loss": 1.5330219268798828,
      "eval_runtime": 12.4092,
      "eval_samples_per_second": 556.847,
      "eval_steps_per_second": 69.626,
      "step": 7500
    },
    {
      "epoch": 0.33355564448001424,
      "grad_norm": 0.09008117020130157,
      "learning_rate": 0.0009868778194777371,
      "loss": 1.5609,
      "step": 7502
    },
    {
      "epoch": 0.3336445689386866,
      "grad_norm": 0.08291777223348618,
      "learning_rate": 0.0009868697810565613,
      "loss": 1.5649,
      "step": 7504
    },
    {
      "epoch": 0.33373349339735897,
      "grad_norm": 0.08472921699285507,
      "learning_rate": 0.0009868617402067979,
      "loss": 1.5604,
      "step": 7506
    },
    {
      "epoch": 0.3338224178560313,
      "grad_norm": 0.08425983041524887,
      "learning_rate": 0.0009868536969284876,
      "loss": 1.5625,
      "step": 7508
    },
    {
      "epoch": 0.33391134231470365,
      "grad_norm": 0.08368197828531265,
      "learning_rate": 0.0009868456512216702,
      "loss": 1.5517,
      "step": 7510
    },
    {
      "epoch": 0.334000266773376,
      "grad_norm": 0.07804959267377853,
      "learning_rate": 0.0009868376030863857,
      "loss": 1.5606,
      "step": 7512
    },
    {
      "epoch": 0.3340891912320484,
      "grad_norm": 0.08549796044826508,
      "learning_rate": 0.0009868295525226746,
      "loss": 1.5547,
      "step": 7514
    },
    {
      "epoch": 0.33417811569072076,
      "grad_norm": 0.08111416548490524,
      "learning_rate": 0.0009868214995305768,
      "loss": 1.5567,
      "step": 7516
    },
    {
      "epoch": 0.33426704014939307,
      "grad_norm": 0.07972585409879684,
      "learning_rate": 0.0009868134441101326,
      "loss": 1.5515,
      "step": 7518
    },
    {
      "epoch": 0.33435596460806544,
      "grad_norm": 0.0834115669131279,
      "learning_rate": 0.0009868053862613821,
      "loss": 1.5563,
      "step": 7520
    },
    {
      "epoch": 0.3344448890667378,
      "grad_norm": 0.08134639263153076,
      "learning_rate": 0.0009867973259843656,
      "loss": 1.5539,
      "step": 7522
    },
    {
      "epoch": 0.3345338135254102,
      "grad_norm": 0.07953674346208572,
      "learning_rate": 0.0009867892632791232,
      "loss": 1.5566,
      "step": 7524
    },
    {
      "epoch": 0.33462273798408254,
      "grad_norm": 0.07886262983083725,
      "learning_rate": 0.000986781198145695,
      "loss": 1.5549,
      "step": 7526
    },
    {
      "epoch": 0.33471166244275485,
      "grad_norm": 0.08340603113174438,
      "learning_rate": 0.0009867731305841217,
      "loss": 1.5609,
      "step": 7528
    },
    {
      "epoch": 0.3348005869014272,
      "grad_norm": 0.08290226757526398,
      "learning_rate": 0.000986765060594443,
      "loss": 1.5561,
      "step": 7530
    },
    {
      "epoch": 0.3348895113600996,
      "grad_norm": 0.08421459048986435,
      "learning_rate": 0.0009867569881766994,
      "loss": 1.5549,
      "step": 7532
    },
    {
      "epoch": 0.33497843581877196,
      "grad_norm": 0.08067211508750916,
      "learning_rate": 0.0009867489133309312,
      "loss": 1.5613,
      "step": 7534
    },
    {
      "epoch": 0.3350673602774443,
      "grad_norm": 0.08332887291908264,
      "learning_rate": 0.0009867408360571786,
      "loss": 1.5533,
      "step": 7536
    },
    {
      "epoch": 0.3351562847361167,
      "grad_norm": 0.08278394490480423,
      "learning_rate": 0.0009867327563554819,
      "loss": 1.5563,
      "step": 7538
    },
    {
      "epoch": 0.335245209194789,
      "grad_norm": 0.08533370494842529,
      "learning_rate": 0.0009867246742258813,
      "loss": 1.558,
      "step": 7540
    },
    {
      "epoch": 0.3353341336534614,
      "grad_norm": 0.08526033163070679,
      "learning_rate": 0.0009867165896684173,
      "loss": 1.5586,
      "step": 7542
    },
    {
      "epoch": 0.33542305811213374,
      "grad_norm": 0.0807008370757103,
      "learning_rate": 0.0009867085026831304,
      "loss": 1.563,
      "step": 7544
    },
    {
      "epoch": 0.3355119825708061,
      "grad_norm": 0.08265620470046997,
      "learning_rate": 0.0009867004132700606,
      "loss": 1.5492,
      "step": 7546
    },
    {
      "epoch": 0.3356009070294785,
      "grad_norm": 0.08214646577835083,
      "learning_rate": 0.0009866923214292482,
      "loss": 1.5571,
      "step": 7548
    },
    {
      "epoch": 0.3356898314881508,
      "grad_norm": 0.0798763632774353,
      "learning_rate": 0.0009866842271607336,
      "loss": 1.5533,
      "step": 7550
    },
    {
      "epoch": 0.33577875594682316,
      "grad_norm": 0.07971591502428055,
      "learning_rate": 0.0009866761304645575,
      "loss": 1.5582,
      "step": 7552
    },
    {
      "epoch": 0.33586768040549553,
      "grad_norm": 0.08346015959978104,
      "learning_rate": 0.00098666803134076,
      "loss": 1.5546,
      "step": 7554
    },
    {
      "epoch": 0.3359566048641679,
      "grad_norm": 0.08312725275754929,
      "learning_rate": 0.0009866599297893817,
      "loss": 1.5637,
      "step": 7556
    },
    {
      "epoch": 0.33604552932284026,
      "grad_norm": 0.07956881821155548,
      "learning_rate": 0.0009866518258104628,
      "loss": 1.5593,
      "step": 7558
    },
    {
      "epoch": 0.33613445378151263,
      "grad_norm": 0.08073802292346954,
      "learning_rate": 0.0009866437194040439,
      "loss": 1.5635,
      "step": 7560
    },
    {
      "epoch": 0.33622337824018494,
      "grad_norm": 0.08115088939666748,
      "learning_rate": 0.0009866356105701652,
      "loss": 1.5571,
      "step": 7562
    },
    {
      "epoch": 0.3363123026988573,
      "grad_norm": 0.08693643659353256,
      "learning_rate": 0.0009866274993088674,
      "loss": 1.5552,
      "step": 7564
    },
    {
      "epoch": 0.3364012271575297,
      "grad_norm": 0.0820782408118248,
      "learning_rate": 0.0009866193856201907,
      "loss": 1.5554,
      "step": 7566
    },
    {
      "epoch": 0.33649015161620205,
      "grad_norm": 0.08132241666316986,
      "learning_rate": 0.000986611269504176,
      "loss": 1.5558,
      "step": 7568
    },
    {
      "epoch": 0.3365790760748744,
      "grad_norm": 0.08526284992694855,
      "learning_rate": 0.0009866031509608633,
      "loss": 1.5552,
      "step": 7570
    },
    {
      "epoch": 0.33666800053354673,
      "grad_norm": 0.08132081478834152,
      "learning_rate": 0.0009865950299902935,
      "loss": 1.5619,
      "step": 7572
    },
    {
      "epoch": 0.3367569249922191,
      "grad_norm": 0.08150200545787811,
      "learning_rate": 0.0009865869065925068,
      "loss": 1.5563,
      "step": 7574
    },
    {
      "epoch": 0.33684584945089147,
      "grad_norm": 0.0802740678191185,
      "learning_rate": 0.0009865787807675438,
      "loss": 1.5508,
      "step": 7576
    },
    {
      "epoch": 0.33693477390956383,
      "grad_norm": 0.08373408019542694,
      "learning_rate": 0.0009865706525154453,
      "loss": 1.5592,
      "step": 7578
    },
    {
      "epoch": 0.3370236983682362,
      "grad_norm": 0.08197604864835739,
      "learning_rate": 0.0009865625218362514,
      "loss": 1.5523,
      "step": 7580
    },
    {
      "epoch": 0.3371126228269085,
      "grad_norm": 0.08425270020961761,
      "learning_rate": 0.0009865543887300028,
      "loss": 1.557,
      "step": 7582
    },
    {
      "epoch": 0.3372015472855809,
      "grad_norm": 0.08232254534959793,
      "learning_rate": 0.00098654625319674,
      "loss": 1.5567,
      "step": 7584
    },
    {
      "epoch": 0.33729047174425325,
      "grad_norm": 0.08390845358371735,
      "learning_rate": 0.0009865381152365043,
      "loss": 1.558,
      "step": 7586
    },
    {
      "epoch": 0.3373793962029256,
      "grad_norm": 0.08453506231307983,
      "learning_rate": 0.0009865299748493352,
      "loss": 1.5561,
      "step": 7588
    },
    {
      "epoch": 0.337468320661598,
      "grad_norm": 0.08450014144182205,
      "learning_rate": 0.0009865218320352742,
      "loss": 1.5515,
      "step": 7590
    },
    {
      "epoch": 0.33755724512027035,
      "grad_norm": 0.083348348736763,
      "learning_rate": 0.0009865136867943615,
      "loss": 1.5583,
      "step": 7592
    },
    {
      "epoch": 0.33764616957894267,
      "grad_norm": 0.0832839161157608,
      "learning_rate": 0.0009865055391266374,
      "loss": 1.5618,
      "step": 7594
    },
    {
      "epoch": 0.33773509403761504,
      "grad_norm": 0.07885057479143143,
      "learning_rate": 0.0009864973890321431,
      "loss": 1.5528,
      "step": 7596
    },
    {
      "epoch": 0.3378240184962874,
      "grad_norm": 0.0822066068649292,
      "learning_rate": 0.0009864892365109193,
      "loss": 1.5585,
      "step": 7598
    },
    {
      "epoch": 0.33791294295495977,
      "grad_norm": 0.08152958005666733,
      "learning_rate": 0.0009864810815630061,
      "loss": 1.559,
      "step": 7600
    },
    {
      "epoch": 0.33800186741363214,
      "grad_norm": 0.08215297013521194,
      "learning_rate": 0.0009864729241884447,
      "loss": 1.5569,
      "step": 7602
    },
    {
      "epoch": 0.33809079187230445,
      "grad_norm": 0.08144047111272812,
      "learning_rate": 0.0009864647643872757,
      "loss": 1.558,
      "step": 7604
    },
    {
      "epoch": 0.3381797163309768,
      "grad_norm": 0.08268721401691437,
      "learning_rate": 0.0009864566021595396,
      "loss": 1.5643,
      "step": 7606
    },
    {
      "epoch": 0.3382686407896492,
      "grad_norm": 0.08536458760499954,
      "learning_rate": 0.000986448437505277,
      "loss": 1.5528,
      "step": 7608
    },
    {
      "epoch": 0.33835756524832156,
      "grad_norm": 0.07940022647380829,
      "learning_rate": 0.0009864402704245292,
      "loss": 1.5514,
      "step": 7610
    },
    {
      "epoch": 0.3384464897069939,
      "grad_norm": 0.0803755670785904,
      "learning_rate": 0.0009864321009173365,
      "loss": 1.5582,
      "step": 7612
    },
    {
      "epoch": 0.3385354141656663,
      "grad_norm": 0.08060748130083084,
      "learning_rate": 0.0009864239289837395,
      "loss": 1.559,
      "step": 7614
    },
    {
      "epoch": 0.3386243386243386,
      "grad_norm": 0.07905225455760956,
      "learning_rate": 0.0009864157546237795,
      "loss": 1.5599,
      "step": 7616
    },
    {
      "epoch": 0.338713263083011,
      "grad_norm": 0.08046825975179672,
      "learning_rate": 0.0009864075778374967,
      "loss": 1.5617,
      "step": 7618
    },
    {
      "epoch": 0.33880218754168334,
      "grad_norm": 0.08096635341644287,
      "learning_rate": 0.0009863993986249325,
      "loss": 1.5552,
      "step": 7620
    },
    {
      "epoch": 0.3388911120003557,
      "grad_norm": 0.08172665536403656,
      "learning_rate": 0.000986391216986127,
      "loss": 1.5593,
      "step": 7622
    },
    {
      "epoch": 0.3389800364590281,
      "grad_norm": 0.0854748860001564,
      "learning_rate": 0.0009863830329211217,
      "loss": 1.5591,
      "step": 7624
    },
    {
      "epoch": 0.3390689609177004,
      "grad_norm": 0.0805579125881195,
      "learning_rate": 0.0009863748464299569,
      "loss": 1.5578,
      "step": 7626
    },
    {
      "epoch": 0.33915788537637276,
      "grad_norm": 0.07932641357183456,
      "learning_rate": 0.0009863666575126737,
      "loss": 1.5491,
      "step": 7628
    },
    {
      "epoch": 0.3392468098350451,
      "grad_norm": 0.0846443846821785,
      "learning_rate": 0.000986358466169313,
      "loss": 1.5591,
      "step": 7630
    },
    {
      "epoch": 0.3393357342937175,
      "grad_norm": 0.08422157168388367,
      "learning_rate": 0.0009863502723999153,
      "loss": 1.5564,
      "step": 7632
    },
    {
      "epoch": 0.33942465875238986,
      "grad_norm": 0.08966944366693497,
      "learning_rate": 0.0009863420762045217,
      "loss": 1.5575,
      "step": 7634
    },
    {
      "epoch": 0.3395135832110622,
      "grad_norm": 0.08580625057220459,
      "learning_rate": 0.000986333877583173,
      "loss": 1.5568,
      "step": 7636
    },
    {
      "epoch": 0.33960250766973454,
      "grad_norm": 0.08479603379964828,
      "learning_rate": 0.0009863256765359108,
      "loss": 1.5607,
      "step": 7638
    },
    {
      "epoch": 0.3396914321284069,
      "grad_norm": 0.0841260626912117,
      "learning_rate": 0.000986317473062775,
      "loss": 1.5547,
      "step": 7640
    },
    {
      "epoch": 0.3397803565870793,
      "grad_norm": 0.08255398273468018,
      "learning_rate": 0.0009863092671638068,
      "loss": 1.5613,
      "step": 7642
    },
    {
      "epoch": 0.33986928104575165,
      "grad_norm": 0.0812351405620575,
      "learning_rate": 0.0009863010588390473,
      "loss": 1.5575,
      "step": 7644
    },
    {
      "epoch": 0.339958205504424,
      "grad_norm": 0.0809001475572586,
      "learning_rate": 0.0009862928480885375,
      "loss": 1.5592,
      "step": 7646
    },
    {
      "epoch": 0.34004712996309633,
      "grad_norm": 0.0820452943444252,
      "learning_rate": 0.0009862846349123183,
      "loss": 1.5551,
      "step": 7648
    },
    {
      "epoch": 0.3401360544217687,
      "grad_norm": 0.08290841430425644,
      "learning_rate": 0.0009862764193104303,
      "loss": 1.5607,
      "step": 7650
    },
    {
      "epoch": 0.34022497888044106,
      "grad_norm": 0.08015654981136322,
      "learning_rate": 0.0009862682012829152,
      "loss": 1.5555,
      "step": 7652
    },
    {
      "epoch": 0.34031390333911343,
      "grad_norm": 0.08518238365650177,
      "learning_rate": 0.0009862599808298134,
      "loss": 1.5524,
      "step": 7654
    },
    {
      "epoch": 0.3404028277977858,
      "grad_norm": 0.08613689988851547,
      "learning_rate": 0.000986251757951166,
      "loss": 1.556,
      "step": 7656
    },
    {
      "epoch": 0.3404917522564581,
      "grad_norm": 0.08661675453186035,
      "learning_rate": 0.0009862435326470143,
      "loss": 1.5564,
      "step": 7658
    },
    {
      "epoch": 0.3405806767151305,
      "grad_norm": 0.07869388163089752,
      "learning_rate": 0.000986235304917399,
      "loss": 1.5577,
      "step": 7660
    },
    {
      "epoch": 0.34066960117380285,
      "grad_norm": 0.08099169284105301,
      "learning_rate": 0.0009862270747623616,
      "loss": 1.5534,
      "step": 7662
    },
    {
      "epoch": 0.3407585256324752,
      "grad_norm": 0.08418874442577362,
      "learning_rate": 0.0009862188421819425,
      "loss": 1.5575,
      "step": 7664
    },
    {
      "epoch": 0.3408474500911476,
      "grad_norm": 0.08251459896564484,
      "learning_rate": 0.0009862106071761834,
      "loss": 1.5578,
      "step": 7666
    },
    {
      "epoch": 0.34093637454981995,
      "grad_norm": 0.08254079520702362,
      "learning_rate": 0.0009862023697451248,
      "loss": 1.5507,
      "step": 7668
    },
    {
      "epoch": 0.34102529900849227,
      "grad_norm": 0.08215059340000153,
      "learning_rate": 0.0009861941298888082,
      "loss": 1.5562,
      "step": 7670
    },
    {
      "epoch": 0.34111422346716463,
      "grad_norm": 0.08373280614614487,
      "learning_rate": 0.0009861858876072747,
      "loss": 1.5594,
      "step": 7672
    },
    {
      "epoch": 0.341203147925837,
      "grad_norm": 0.07917515188455582,
      "learning_rate": 0.000986177642900565,
      "loss": 1.5542,
      "step": 7674
    },
    {
      "epoch": 0.34129207238450937,
      "grad_norm": 0.08267196267843246,
      "learning_rate": 0.0009861693957687208,
      "loss": 1.5612,
      "step": 7676
    },
    {
      "epoch": 0.34138099684318174,
      "grad_norm": 0.08618869632482529,
      "learning_rate": 0.0009861611462117829,
      "loss": 1.5604,
      "step": 7678
    },
    {
      "epoch": 0.34146992130185405,
      "grad_norm": 0.08248917013406754,
      "learning_rate": 0.0009861528942297923,
      "loss": 1.5624,
      "step": 7680
    },
    {
      "epoch": 0.3415588457605264,
      "grad_norm": 0.08377496898174286,
      "learning_rate": 0.0009861446398227904,
      "loss": 1.5614,
      "step": 7682
    },
    {
      "epoch": 0.3416477702191988,
      "grad_norm": 0.08078130334615707,
      "learning_rate": 0.0009861363829908186,
      "loss": 1.5528,
      "step": 7684
    },
    {
      "epoch": 0.34173669467787116,
      "grad_norm": 0.07990232855081558,
      "learning_rate": 0.0009861281237339176,
      "loss": 1.5554,
      "step": 7686
    },
    {
      "epoch": 0.3418256191365435,
      "grad_norm": 0.07614683359861374,
      "learning_rate": 0.0009861198620521288,
      "loss": 1.5562,
      "step": 7688
    },
    {
      "epoch": 0.3419145435952159,
      "grad_norm": 0.08139005303382874,
      "learning_rate": 0.0009861115979454935,
      "loss": 1.5544,
      "step": 7690
    },
    {
      "epoch": 0.3420034680538882,
      "grad_norm": 0.08052778244018555,
      "learning_rate": 0.000986103331414053,
      "loss": 1.5592,
      "step": 7692
    },
    {
      "epoch": 0.3420923925125606,
      "grad_norm": 0.08632178604602814,
      "learning_rate": 0.0009860950624578481,
      "loss": 1.5574,
      "step": 7694
    },
    {
      "epoch": 0.34218131697123294,
      "grad_norm": 0.0841991975903511,
      "learning_rate": 0.0009860867910769207,
      "loss": 1.5576,
      "step": 7696
    },
    {
      "epoch": 0.3422702414299053,
      "grad_norm": 0.08402567356824875,
      "learning_rate": 0.0009860785172713115,
      "loss": 1.5525,
      "step": 7698
    },
    {
      "epoch": 0.3423591658885777,
      "grad_norm": 0.08191527426242828,
      "learning_rate": 0.000986070241041062,
      "loss": 1.5613,
      "step": 7700
    },
    {
      "epoch": 0.34244809034725,
      "grad_norm": 0.08203962445259094,
      "learning_rate": 0.0009860619623862136,
      "loss": 1.5565,
      "step": 7702
    },
    {
      "epoch": 0.34253701480592236,
      "grad_norm": 0.07740230858325958,
      "learning_rate": 0.0009860536813068072,
      "loss": 1.5574,
      "step": 7704
    },
    {
      "epoch": 0.3426259392645947,
      "grad_norm": 0.07994545996189117,
      "learning_rate": 0.0009860453978028844,
      "loss": 1.5573,
      "step": 7706
    },
    {
      "epoch": 0.3427148637232671,
      "grad_norm": 0.08103641122579575,
      "learning_rate": 0.0009860371118744866,
      "loss": 1.5566,
      "step": 7708
    },
    {
      "epoch": 0.34280378818193946,
      "grad_norm": 0.08051252365112305,
      "learning_rate": 0.0009860288235216549,
      "loss": 1.5603,
      "step": 7710
    },
    {
      "epoch": 0.3428927126406118,
      "grad_norm": 0.08033636957406998,
      "learning_rate": 0.0009860205327444308,
      "loss": 1.5545,
      "step": 7712
    },
    {
      "epoch": 0.34298163709928414,
      "grad_norm": 0.08459188044071198,
      "learning_rate": 0.0009860122395428555,
      "loss": 1.5595,
      "step": 7714
    },
    {
      "epoch": 0.3430705615579565,
      "grad_norm": 0.08079506456851959,
      "learning_rate": 0.0009860039439169705,
      "loss": 1.5565,
      "step": 7716
    },
    {
      "epoch": 0.3431594860166289,
      "grad_norm": 0.08086486160755157,
      "learning_rate": 0.0009859956458668174,
      "loss": 1.5585,
      "step": 7718
    },
    {
      "epoch": 0.34324841047530125,
      "grad_norm": 0.0799533799290657,
      "learning_rate": 0.000985987345392437,
      "loss": 1.5492,
      "step": 7720
    },
    {
      "epoch": 0.3433373349339736,
      "grad_norm": 0.07945334911346436,
      "learning_rate": 0.0009859790424938714,
      "loss": 1.5486,
      "step": 7722
    },
    {
      "epoch": 0.3434262593926459,
      "grad_norm": 0.07963025569915771,
      "learning_rate": 0.0009859707371711614,
      "loss": 1.5606,
      "step": 7724
    },
    {
      "epoch": 0.3435151838513183,
      "grad_norm": 0.08368024230003357,
      "learning_rate": 0.0009859624294243487,
      "loss": 1.5599,
      "step": 7726
    },
    {
      "epoch": 0.34360410830999066,
      "grad_norm": 0.07972510904073715,
      "learning_rate": 0.000985954119253475,
      "loss": 1.5461,
      "step": 7728
    },
    {
      "epoch": 0.34369303276866303,
      "grad_norm": 0.08038385957479477,
      "learning_rate": 0.000985945806658581,
      "loss": 1.5568,
      "step": 7730
    },
    {
      "epoch": 0.3437819572273354,
      "grad_norm": 0.08230438828468323,
      "learning_rate": 0.0009859374916397093,
      "loss": 1.5553,
      "step": 7732
    },
    {
      "epoch": 0.3438708816860077,
      "grad_norm": 0.08172792196273804,
      "learning_rate": 0.0009859291741969004,
      "loss": 1.5544,
      "step": 7734
    },
    {
      "epoch": 0.3439598061446801,
      "grad_norm": 0.08643286675214767,
      "learning_rate": 0.0009859208543301962,
      "loss": 1.5551,
      "step": 7736
    },
    {
      "epoch": 0.34404873060335245,
      "grad_norm": 0.08269612491130829,
      "learning_rate": 0.0009859125320396379,
      "loss": 1.5538,
      "step": 7738
    },
    {
      "epoch": 0.3441376550620248,
      "grad_norm": 0.08148770779371262,
      "learning_rate": 0.0009859042073252673,
      "loss": 1.5511,
      "step": 7740
    },
    {
      "epoch": 0.3442265795206972,
      "grad_norm": 0.0830964595079422,
      "learning_rate": 0.0009858958801871257,
      "loss": 1.559,
      "step": 7742
    },
    {
      "epoch": 0.34431550397936955,
      "grad_norm": 0.08528967946767807,
      "learning_rate": 0.0009858875506252552,
      "loss": 1.5578,
      "step": 7744
    },
    {
      "epoch": 0.34440442843804187,
      "grad_norm": 0.08497735857963562,
      "learning_rate": 0.0009858792186396967,
      "loss": 1.5565,
      "step": 7746
    },
    {
      "epoch": 0.34449335289671423,
      "grad_norm": 0.07968219369649887,
      "learning_rate": 0.0009858708842304922,
      "loss": 1.5606,
      "step": 7748
    },
    {
      "epoch": 0.3445822773553866,
      "grad_norm": 0.08317109197378159,
      "learning_rate": 0.0009858625473976828,
      "loss": 1.5534,
      "step": 7750
    },
    {
      "epoch": 0.34467120181405897,
      "grad_norm": 0.07973136752843857,
      "learning_rate": 0.0009858542081413106,
      "loss": 1.5536,
      "step": 7752
    },
    {
      "epoch": 0.34476012627273134,
      "grad_norm": 0.07634979486465454,
      "learning_rate": 0.0009858458664614169,
      "loss": 1.5492,
      "step": 7754
    },
    {
      "epoch": 0.34484905073140365,
      "grad_norm": 0.07837677747011185,
      "learning_rate": 0.0009858375223580434,
      "loss": 1.5548,
      "step": 7756
    },
    {
      "epoch": 0.344937975190076,
      "grad_norm": 0.07845926284790039,
      "learning_rate": 0.0009858291758312317,
      "loss": 1.5537,
      "step": 7758
    },
    {
      "epoch": 0.3450268996487484,
      "grad_norm": 0.08067924529314041,
      "learning_rate": 0.0009858208268810232,
      "loss": 1.5546,
      "step": 7760
    },
    {
      "epoch": 0.34511582410742075,
      "grad_norm": 0.07944106310606003,
      "learning_rate": 0.00098581247550746,
      "loss": 1.5523,
      "step": 7762
    },
    {
      "epoch": 0.3452047485660931,
      "grad_norm": 0.08254885673522949,
      "learning_rate": 0.0009858041217105835,
      "loss": 1.5549,
      "step": 7764
    },
    {
      "epoch": 0.34529367302476544,
      "grad_norm": 0.08257626742124557,
      "learning_rate": 0.0009857957654904352,
      "loss": 1.5568,
      "step": 7766
    },
    {
      "epoch": 0.3453825974834378,
      "grad_norm": 0.07757585495710373,
      "learning_rate": 0.0009857874068470575,
      "loss": 1.5542,
      "step": 7768
    },
    {
      "epoch": 0.34547152194211017,
      "grad_norm": 0.07840435206890106,
      "learning_rate": 0.000985779045780491,
      "loss": 1.5541,
      "step": 7770
    },
    {
      "epoch": 0.34556044640078254,
      "grad_norm": 0.08048916608095169,
      "learning_rate": 0.0009857706822907783,
      "loss": 1.5509,
      "step": 7772
    },
    {
      "epoch": 0.3456493708594549,
      "grad_norm": 0.08014550805091858,
      "learning_rate": 0.0009857623163779606,
      "loss": 1.5515,
      "step": 7774
    },
    {
      "epoch": 0.3457382953181273,
      "grad_norm": 0.08095970749855042,
      "learning_rate": 0.00098575394804208,
      "loss": 1.5545,
      "step": 7776
    },
    {
      "epoch": 0.3458272197767996,
      "grad_norm": 0.0833306685090065,
      "learning_rate": 0.000985745577283178,
      "loss": 1.5512,
      "step": 7778
    },
    {
      "epoch": 0.34591614423547196,
      "grad_norm": 0.08204789459705353,
      "learning_rate": 0.0009857372041012966,
      "loss": 1.5512,
      "step": 7780
    },
    {
      "epoch": 0.3460050686941443,
      "grad_norm": 0.08149214088916779,
      "learning_rate": 0.000985728828496477,
      "loss": 1.5549,
      "step": 7782
    },
    {
      "epoch": 0.3460939931528167,
      "grad_norm": 0.08480332046747208,
      "learning_rate": 0.0009857204504687617,
      "loss": 1.5529,
      "step": 7784
    },
    {
      "epoch": 0.34618291761148906,
      "grad_norm": 0.0808916836977005,
      "learning_rate": 0.0009857120700181921,
      "loss": 1.5539,
      "step": 7786
    },
    {
      "epoch": 0.3462718420701614,
      "grad_norm": 0.08434953540563583,
      "learning_rate": 0.0009857036871448101,
      "loss": 1.5492,
      "step": 7788
    },
    {
      "epoch": 0.34636076652883374,
      "grad_norm": 0.077853262424469,
      "learning_rate": 0.0009856953018486576,
      "loss": 1.5522,
      "step": 7790
    },
    {
      "epoch": 0.3464496909875061,
      "grad_norm": 0.08755781501531601,
      "learning_rate": 0.0009856869141297762,
      "loss": 1.5561,
      "step": 7792
    },
    {
      "epoch": 0.3465386154461785,
      "grad_norm": 0.0840546116232872,
      "learning_rate": 0.0009856785239882079,
      "loss": 1.5522,
      "step": 7794
    },
    {
      "epoch": 0.34662753990485085,
      "grad_norm": 0.08388349413871765,
      "learning_rate": 0.0009856701314239942,
      "loss": 1.5519,
      "step": 7796
    },
    {
      "epoch": 0.3467164643635232,
      "grad_norm": 0.07955602556467056,
      "learning_rate": 0.0009856617364371776,
      "loss": 1.5584,
      "step": 7798
    },
    {
      "epoch": 0.3468053888221955,
      "grad_norm": 0.08307410031557083,
      "learning_rate": 0.0009856533390277995,
      "loss": 1.5527,
      "step": 7800
    },
    {
      "epoch": 0.3468943132808679,
      "grad_norm": 0.08166033774614334,
      "learning_rate": 0.000985644939195902,
      "loss": 1.5575,
      "step": 7802
    },
    {
      "epoch": 0.34698323773954026,
      "grad_norm": 0.08017229288816452,
      "learning_rate": 0.0009856365369415269,
      "loss": 1.5536,
      "step": 7804
    },
    {
      "epoch": 0.34707216219821263,
      "grad_norm": 0.08459261059761047,
      "learning_rate": 0.000985628132264716,
      "loss": 1.5624,
      "step": 7806
    },
    {
      "epoch": 0.347161086656885,
      "grad_norm": 0.07887223362922668,
      "learning_rate": 0.0009856197251655117,
      "loss": 1.5504,
      "step": 7808
    },
    {
      "epoch": 0.3472500111155573,
      "grad_norm": 0.08019403368234634,
      "learning_rate": 0.0009856113156439554,
      "loss": 1.5524,
      "step": 7810
    },
    {
      "epoch": 0.3473389355742297,
      "grad_norm": 0.0818062275648117,
      "learning_rate": 0.0009856029037000893,
      "loss": 1.5556,
      "step": 7812
    },
    {
      "epoch": 0.34742786003290205,
      "grad_norm": 0.07944361120462418,
      "learning_rate": 0.0009855944893339553,
      "loss": 1.5543,
      "step": 7814
    },
    {
      "epoch": 0.3475167844915744,
      "grad_norm": 0.07994072139263153,
      "learning_rate": 0.0009855860725455953,
      "loss": 1.5454,
      "step": 7816
    },
    {
      "epoch": 0.3476057089502468,
      "grad_norm": 0.07912899553775787,
      "learning_rate": 0.0009855776533350513,
      "loss": 1.5526,
      "step": 7818
    },
    {
      "epoch": 0.3476946334089191,
      "grad_norm": 0.08150925487279892,
      "learning_rate": 0.0009855692317023654,
      "loss": 1.5507,
      "step": 7820
    },
    {
      "epoch": 0.34778355786759146,
      "grad_norm": 0.0804237350821495,
      "learning_rate": 0.0009855608076475798,
      "loss": 1.5562,
      "step": 7822
    },
    {
      "epoch": 0.34787248232626383,
      "grad_norm": 0.07889696210622787,
      "learning_rate": 0.000985552381170736,
      "loss": 1.5572,
      "step": 7824
    },
    {
      "epoch": 0.3479614067849362,
      "grad_norm": 0.07692207396030426,
      "learning_rate": 0.0009855439522718766,
      "loss": 1.5534,
      "step": 7826
    },
    {
      "epoch": 0.34805033124360857,
      "grad_norm": 0.07672964036464691,
      "learning_rate": 0.0009855355209510432,
      "loss": 1.553,
      "step": 7828
    },
    {
      "epoch": 0.34813925570228094,
      "grad_norm": 0.07981939613819122,
      "learning_rate": 0.000985527087208278,
      "loss": 1.5549,
      "step": 7830
    },
    {
      "epoch": 0.34822818016095325,
      "grad_norm": 0.07698217779397964,
      "learning_rate": 0.0009855186510436232,
      "loss": 1.5567,
      "step": 7832
    },
    {
      "epoch": 0.3483171046196256,
      "grad_norm": 0.07928860932588577,
      "learning_rate": 0.000985510212457121,
      "loss": 1.5513,
      "step": 7834
    },
    {
      "epoch": 0.348406029078298,
      "grad_norm": 0.08210493624210358,
      "learning_rate": 0.0009855017714488128,
      "loss": 1.5503,
      "step": 7836
    },
    {
      "epoch": 0.34849495353697035,
      "grad_norm": 0.08063147217035294,
      "learning_rate": 0.0009854933280187416,
      "loss": 1.5549,
      "step": 7838
    },
    {
      "epoch": 0.3485838779956427,
      "grad_norm": 0.07979092001914978,
      "learning_rate": 0.0009854848821669487,
      "loss": 1.5482,
      "step": 7840
    },
    {
      "epoch": 0.34867280245431503,
      "grad_norm": 0.0807296484708786,
      "learning_rate": 0.0009854764338934768,
      "loss": 1.5546,
      "step": 7842
    },
    {
      "epoch": 0.3487617269129874,
      "grad_norm": 0.082137331366539,
      "learning_rate": 0.0009854679831983678,
      "loss": 1.5572,
      "step": 7844
    },
    {
      "epoch": 0.34885065137165977,
      "grad_norm": 0.0817376971244812,
      "learning_rate": 0.000985459530081664,
      "loss": 1.5512,
      "step": 7846
    },
    {
      "epoch": 0.34893957583033214,
      "grad_norm": 0.08156748861074448,
      "learning_rate": 0.0009854510745434075,
      "loss": 1.5563,
      "step": 7848
    },
    {
      "epoch": 0.3490285002890045,
      "grad_norm": 0.08468926697969437,
      "learning_rate": 0.0009854426165836403,
      "loss": 1.5537,
      "step": 7850
    },
    {
      "epoch": 0.3491174247476769,
      "grad_norm": 0.0817328691482544,
      "learning_rate": 0.000985434156202405,
      "loss": 1.559,
      "step": 7852
    },
    {
      "epoch": 0.3492063492063492,
      "grad_norm": 0.07793274521827698,
      "learning_rate": 0.0009854256933997435,
      "loss": 1.5527,
      "step": 7854
    },
    {
      "epoch": 0.34929527366502156,
      "grad_norm": 0.07918530702590942,
      "learning_rate": 0.0009854172281756979,
      "loss": 1.554,
      "step": 7856
    },
    {
      "epoch": 0.3493841981236939,
      "grad_norm": 0.08301304280757904,
      "learning_rate": 0.0009854087605303105,
      "loss": 1.554,
      "step": 7858
    },
    {
      "epoch": 0.3494731225823663,
      "grad_norm": 0.07774464040994644,
      "learning_rate": 0.0009854002904636238,
      "loss": 1.5517,
      "step": 7860
    },
    {
      "epoch": 0.34956204704103866,
      "grad_norm": 0.08034802228212357,
      "learning_rate": 0.00098539181797568,
      "loss": 1.5544,
      "step": 7862
    },
    {
      "epoch": 0.34965097149971097,
      "grad_norm": 0.07757385820150375,
      "learning_rate": 0.0009853833430665212,
      "loss": 1.5548,
      "step": 7864
    },
    {
      "epoch": 0.34973989595838334,
      "grad_norm": 0.08288813382387161,
      "learning_rate": 0.0009853748657361896,
      "loss": 1.5527,
      "step": 7866
    },
    {
      "epoch": 0.3498288204170557,
      "grad_norm": 0.08583445101976395,
      "learning_rate": 0.0009853663859847276,
      "loss": 1.5544,
      "step": 7868
    },
    {
      "epoch": 0.3499177448757281,
      "grad_norm": 0.07990463078022003,
      "learning_rate": 0.0009853579038121775,
      "loss": 1.5537,
      "step": 7870
    },
    {
      "epoch": 0.35000666933440044,
      "grad_norm": 0.08092455565929413,
      "learning_rate": 0.0009853494192185817,
      "loss": 1.5494,
      "step": 7872
    },
    {
      "epoch": 0.3500955937930728,
      "grad_norm": 0.07774668186903,
      "learning_rate": 0.0009853409322039823,
      "loss": 1.5512,
      "step": 7874
    },
    {
      "epoch": 0.3501845182517451,
      "grad_norm": 0.08283083885908127,
      "learning_rate": 0.000985332442768422,
      "loss": 1.5538,
      "step": 7876
    },
    {
      "epoch": 0.3502734427104175,
      "grad_norm": 0.08365800231695175,
      "learning_rate": 0.0009853239509119427,
      "loss": 1.5525,
      "step": 7878
    },
    {
      "epoch": 0.35036236716908986,
      "grad_norm": 0.077533058822155,
      "learning_rate": 0.000985315456634587,
      "loss": 1.5596,
      "step": 7880
    },
    {
      "epoch": 0.35045129162776223,
      "grad_norm": 0.08021321892738342,
      "learning_rate": 0.0009853069599363973,
      "loss": 1.5468,
      "step": 7882
    },
    {
      "epoch": 0.3505402160864346,
      "grad_norm": 0.11996711045503616,
      "learning_rate": 0.0009852984608174158,
      "loss": 1.5533,
      "step": 7884
    },
    {
      "epoch": 0.3506291405451069,
      "grad_norm": 0.3232218623161316,
      "learning_rate": 0.0009852899592776852,
      "loss": 1.6783,
      "step": 7886
    },
    {
      "epoch": 0.3507180650037793,
      "grad_norm": 4.166322231292725,
      "learning_rate": 0.0009852814553172476,
      "loss": 2.4033,
      "step": 7888
    },
    {
      "epoch": 0.35080698946245165,
      "grad_norm": 105.5245132446289,
      "learning_rate": 0.0009852729489361457,
      "loss": 3.5532,
      "step": 7890
    },
    {
      "epoch": 0.350895913921124,
      "grad_norm": 5.243870258331299,
      "learning_rate": 0.0009852644401344218,
      "loss": 7.1315,
      "step": 7892
    },
    {
      "epoch": 0.3509848383797964,
      "grad_norm": 6.443427085876465,
      "learning_rate": 0.0009852559289121182,
      "loss": 7.392,
      "step": 7894
    },
    {
      "epoch": 0.3510737628384687,
      "grad_norm": 11.5606050491333,
      "learning_rate": 0.0009852474152692774,
      "loss": 7.5255,
      "step": 7896
    },
    {
      "epoch": 0.35116268729714106,
      "grad_norm": 2.048408031463623,
      "learning_rate": 0.000985238899205942,
      "loss": 7.2354,
      "step": 7898
    },
    {
      "epoch": 0.35125161175581343,
      "grad_norm": 0.9983798265457153,
      "learning_rate": 0.0009852303807221545,
      "loss": 6.5717,
      "step": 7900
    },
    {
      "epoch": 0.3513405362144858,
      "grad_norm": 6.880504608154297,
      "learning_rate": 0.0009852218598179572,
      "loss": 6.7333,
      "step": 7902
    },
    {
      "epoch": 0.35142946067315817,
      "grad_norm": 0.685417890548706,
      "learning_rate": 0.0009852133364933928,
      "loss": 6.3766,
      "step": 7904
    },
    {
      "epoch": 0.35151838513183054,
      "grad_norm": 1.0315533876419067,
      "learning_rate": 0.0009852048107485039,
      "loss": 6.1477,
      "step": 7906
    },
    {
      "epoch": 0.35160730959050285,
      "grad_norm": 0.4206119179725647,
      "learning_rate": 0.0009851962825833324,
      "loss": 6.0279,
      "step": 7908
    },
    {
      "epoch": 0.3516962340491752,
      "grad_norm": 0.34929385781288147,
      "learning_rate": 0.0009851877519979218,
      "loss": 5.9475,
      "step": 7910
    },
    {
      "epoch": 0.3517851585078476,
      "grad_norm": 0.33057817816734314,
      "learning_rate": 0.000985179218992314,
      "loss": 5.8579,
      "step": 7912
    },
    {
      "epoch": 0.35187408296651995,
      "grad_norm": 0.2894933819770813,
      "learning_rate": 0.0009851706835665515,
      "loss": 5.8023,
      "step": 7914
    },
    {
      "epoch": 0.3519630074251923,
      "grad_norm": 0.21461133658885956,
      "learning_rate": 0.0009851621457206774,
      "loss": 5.7603,
      "step": 7916
    },
    {
      "epoch": 0.35205193188386463,
      "grad_norm": 0.21909675002098083,
      "learning_rate": 0.0009851536054547338,
      "loss": 5.7207,
      "step": 7918
    },
    {
      "epoch": 0.352140856342537,
      "grad_norm": 0.3142562508583069,
      "learning_rate": 0.0009851450627687635,
      "loss": 5.6821,
      "step": 7920
    },
    {
      "epoch": 0.35222978080120937,
      "grad_norm": 1.3621059656143188,
      "learning_rate": 0.000985136517662809,
      "loss": 5.6821,
      "step": 7922
    },
    {
      "epoch": 0.35231870525988174,
      "grad_norm": 0.7668919563293457,
      "learning_rate": 0.0009851279701369134,
      "loss": 5.6241,
      "step": 7924
    },
    {
      "epoch": 0.3524076297185541,
      "grad_norm": 0.3618534207344055,
      "learning_rate": 0.0009851194201911187,
      "loss": 5.5482,
      "step": 7926
    },
    {
      "epoch": 0.3524965541772265,
      "grad_norm": 0.2870240807533264,
      "learning_rate": 0.0009851108678254677,
      "loss": 5.5032,
      "step": 7928
    },
    {
      "epoch": 0.3525854786358988,
      "grad_norm": 0.485879510641098,
      "learning_rate": 0.0009851023130400035,
      "loss": 5.4567,
      "step": 7930
    },
    {
      "epoch": 0.35267440309457115,
      "grad_norm": 0.2687022387981415,
      "learning_rate": 0.000985093755834768,
      "loss": 5.3988,
      "step": 7932
    },
    {
      "epoch": 0.3527633275532435,
      "grad_norm": 0.3306155204772949,
      "learning_rate": 0.0009850851962098046,
      "loss": 5.3285,
      "step": 7934
    },
    {
      "epoch": 0.3528522520119159,
      "grad_norm": 0.7935633063316345,
      "learning_rate": 0.0009850766341651556,
      "loss": 5.2442,
      "step": 7936
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 1.1152472496032715,
      "learning_rate": 0.000985068069700864,
      "loss": 5.2263,
      "step": 7938
    },
    {
      "epoch": 0.35303010092926057,
      "grad_norm": 1.460869550704956,
      "learning_rate": 0.0009850595028169722,
      "loss": 5.2365,
      "step": 7940
    },
    {
      "epoch": 0.35311902538793294,
      "grad_norm": 1.1066856384277344,
      "learning_rate": 0.0009850509335135231,
      "loss": 5.1598,
      "step": 7942
    },
    {
      "epoch": 0.3532079498466053,
      "grad_norm": 0.8395753502845764,
      "learning_rate": 0.0009850423617905595,
      "loss": 5.0372,
      "step": 7944
    },
    {
      "epoch": 0.3532968743052777,
      "grad_norm": 0.550400972366333,
      "learning_rate": 0.0009850337876481242,
      "loss": 4.9278,
      "step": 7946
    },
    {
      "epoch": 0.35338579876395004,
      "grad_norm": 0.7241742014884949,
      "learning_rate": 0.0009850252110862596,
      "loss": 4.8471,
      "step": 7948
    },
    {
      "epoch": 0.35347472322262236,
      "grad_norm": 1.1229575872421265,
      "learning_rate": 0.000985016632105009,
      "loss": 4.7441,
      "step": 7950
    },
    {
      "epoch": 0.3535636476812947,
      "grad_norm": 0.5880857110023499,
      "learning_rate": 0.0009850080507044147,
      "loss": 4.6106,
      "step": 7952
    },
    {
      "epoch": 0.3536525721399671,
      "grad_norm": 0.6695611476898193,
      "learning_rate": 0.0009849994668845196,
      "loss": 4.4359,
      "step": 7954
    },
    {
      "epoch": 0.35374149659863946,
      "grad_norm": 0.4951477646827698,
      "learning_rate": 0.000984990880645367,
      "loss": 4.3121,
      "step": 7956
    },
    {
      "epoch": 0.35383042105731183,
      "grad_norm": 1.49424147605896,
      "learning_rate": 0.0009849822919869994,
      "loss": 4.3655,
      "step": 7958
    },
    {
      "epoch": 0.3539193455159842,
      "grad_norm": 0.6731409430503845,
      "learning_rate": 0.0009849737009094595,
      "loss": 4.1954,
      "step": 7960
    },
    {
      "epoch": 0.3540082699746565,
      "grad_norm": 0.657111644744873,
      "learning_rate": 0.00098496510741279,
      "loss": 3.9788,
      "step": 7962
    },
    {
      "epoch": 0.3540971944333289,
      "grad_norm": 1.2278788089752197,
      "learning_rate": 0.0009849565114970346,
      "loss": 3.8825,
      "step": 7964
    },
    {
      "epoch": 0.35418611889200124,
      "grad_norm": 1.1738439798355103,
      "learning_rate": 0.0009849479131622352,
      "loss": 3.7363,
      "step": 7966
    },
    {
      "epoch": 0.3542750433506736,
      "grad_norm": 0.9582828283309937,
      "learning_rate": 0.000984939312408435,
      "loss": 3.5467,
      "step": 7968
    },
    {
      "epoch": 0.354363967809346,
      "grad_norm": 0.9599164724349976,
      "learning_rate": 0.0009849307092356773,
      "loss": 3.3789,
      "step": 7970
    },
    {
      "epoch": 0.3544528922680183,
      "grad_norm": 0.8798990249633789,
      "learning_rate": 0.0009849221036440049,
      "loss": 3.201,
      "step": 7972
    },
    {
      "epoch": 0.35454181672669066,
      "grad_norm": 0.837184488773346,
      "learning_rate": 0.0009849134956334603,
      "loss": 3.0279,
      "step": 7974
    },
    {
      "epoch": 0.35463074118536303,
      "grad_norm": 0.869229793548584,
      "learning_rate": 0.0009849048852040867,
      "loss": 2.9193,
      "step": 7976
    },
    {
      "epoch": 0.3547196656440354,
      "grad_norm": 1.0861220359802246,
      "learning_rate": 0.000984896272355927,
      "loss": 2.7289,
      "step": 7978
    },
    {
      "epoch": 0.35480859010270777,
      "grad_norm": 0.8610190749168396,
      "learning_rate": 0.000984887657089024,
      "loss": 2.5584,
      "step": 7980
    },
    {
      "epoch": 0.35489751456138013,
      "grad_norm": 0.6836897134780884,
      "learning_rate": 0.0009848790394034214,
      "loss": 2.4055,
      "step": 7982
    },
    {
      "epoch": 0.35498643902005245,
      "grad_norm": 0.7401456832885742,
      "learning_rate": 0.0009848704192991613,
      "loss": 2.2528,
      "step": 7984
    },
    {
      "epoch": 0.3550753634787248,
      "grad_norm": 0.9028481245040894,
      "learning_rate": 0.0009848617967762872,
      "loss": 2.3112,
      "step": 7986
    },
    {
      "epoch": 0.3551642879373972,
      "grad_norm": 0.3118695318698883,
      "learning_rate": 0.000984853171834842,
      "loss": 2.1296,
      "step": 7988
    },
    {
      "epoch": 0.35525321239606955,
      "grad_norm": 0.30334171652793884,
      "learning_rate": 0.0009848445444748685,
      "loss": 2.0048,
      "step": 7990
    },
    {
      "epoch": 0.3553421368547419,
      "grad_norm": 0.19824688136577606,
      "learning_rate": 0.0009848359146964103,
      "loss": 1.9316,
      "step": 7992
    },
    {
      "epoch": 0.35543106131341423,
      "grad_norm": 0.2128942757844925,
      "learning_rate": 0.0009848272824995096,
      "loss": 1.8792,
      "step": 7994
    },
    {
      "epoch": 0.3555199857720866,
      "grad_norm": 0.15974895656108856,
      "learning_rate": 0.0009848186478842104,
      "loss": 1.8306,
      "step": 7996
    },
    {
      "epoch": 0.35560891023075897,
      "grad_norm": 0.13069583475589752,
      "learning_rate": 0.000984810010850555,
      "loss": 1.7945,
      "step": 7998
    },
    {
      "epoch": 0.35569783468943134,
      "grad_norm": 0.13678595423698425,
      "learning_rate": 0.000984801371398587,
      "loss": 1.7696,
      "step": 8000
    },
    {
      "epoch": 0.35569783468943134,
      "eval_loss": 1.6807708740234375,
      "eval_runtime": 12.3592,
      "eval_samples_per_second": 559.097,
      "eval_steps_per_second": 69.907,
      "step": 8000
    },
    {
      "epoch": 0.3557867591481037,
      "grad_norm": 0.12552303075790405,
      "learning_rate": 0.0009847927295283491,
      "loss": 1.7432,
      "step": 8002
    },
    {
      "epoch": 0.355875683606776,
      "grad_norm": 0.12398079037666321,
      "learning_rate": 0.0009847840852398846,
      "loss": 1.7169,
      "step": 8004
    },
    {
      "epoch": 0.3559646080654484,
      "grad_norm": 0.11141873896121979,
      "learning_rate": 0.0009847754385332369,
      "loss": 1.6993,
      "step": 8006
    },
    {
      "epoch": 0.35605353252412075,
      "grad_norm": 0.1098860651254654,
      "learning_rate": 0.0009847667894084485,
      "loss": 1.6952,
      "step": 8008
    },
    {
      "epoch": 0.3561424569827931,
      "grad_norm": 0.10826461762189865,
      "learning_rate": 0.000984758137865563,
      "loss": 1.6817,
      "step": 8010
    },
    {
      "epoch": 0.3562313814414655,
      "grad_norm": 0.10351142287254333,
      "learning_rate": 0.0009847494839046234,
      "loss": 1.6737,
      "step": 8012
    },
    {
      "epoch": 0.35632030590013786,
      "grad_norm": 0.09795521944761276,
      "learning_rate": 0.000984740827525673,
      "loss": 1.6679,
      "step": 8014
    },
    {
      "epoch": 0.35640923035881017,
      "grad_norm": 0.0955873504281044,
      "learning_rate": 0.000984732168728755,
      "loss": 1.6526,
      "step": 8016
    },
    {
      "epoch": 0.35649815481748254,
      "grad_norm": 0.09694499522447586,
      "learning_rate": 0.0009847235075139123,
      "loss": 1.6522,
      "step": 8018
    },
    {
      "epoch": 0.3565870792761549,
      "grad_norm": 0.09016109257936478,
      "learning_rate": 0.0009847148438811883,
      "loss": 1.6439,
      "step": 8020
    },
    {
      "epoch": 0.3566760037348273,
      "grad_norm": 0.08890122175216675,
      "learning_rate": 0.0009847061778306262,
      "loss": 1.6345,
      "step": 8022
    },
    {
      "epoch": 0.35676492819349964,
      "grad_norm": 0.09018497169017792,
      "learning_rate": 0.0009846975093622692,
      "loss": 1.6403,
      "step": 8024
    },
    {
      "epoch": 0.35685385265217195,
      "grad_norm": 0.0861065536737442,
      "learning_rate": 0.0009846888384761607,
      "loss": 1.6304,
      "step": 8026
    },
    {
      "epoch": 0.3569427771108443,
      "grad_norm": 0.08618265390396118,
      "learning_rate": 0.0009846801651723436,
      "loss": 1.6258,
      "step": 8028
    },
    {
      "epoch": 0.3570317015695167,
      "grad_norm": 0.0936831384897232,
      "learning_rate": 0.0009846714894508617,
      "loss": 1.6244,
      "step": 8030
    },
    {
      "epoch": 0.35712062602818906,
      "grad_norm": 0.08654095232486725,
      "learning_rate": 0.0009846628113117577,
      "loss": 1.6206,
      "step": 8032
    },
    {
      "epoch": 0.3572095504868614,
      "grad_norm": 0.09224793314933777,
      "learning_rate": 0.0009846541307550753,
      "loss": 1.6189,
      "step": 8034
    },
    {
      "epoch": 0.3572984749455338,
      "grad_norm": 0.08937020599842072,
      "learning_rate": 0.0009846454477808575,
      "loss": 1.6198,
      "step": 8036
    },
    {
      "epoch": 0.3573873994042061,
      "grad_norm": 0.09124995023012161,
      "learning_rate": 0.0009846367623891478,
      "loss": 1.6108,
      "step": 8038
    },
    {
      "epoch": 0.3574763238628785,
      "grad_norm": 0.08880987763404846,
      "learning_rate": 0.0009846280745799895,
      "loss": 1.6126,
      "step": 8040
    },
    {
      "epoch": 0.35756524832155084,
      "grad_norm": 0.0875345841050148,
      "learning_rate": 0.000984619384353426,
      "loss": 1.6134,
      "step": 8042
    },
    {
      "epoch": 0.3576541727802232,
      "grad_norm": 0.08219029754400253,
      "learning_rate": 0.0009846106917095006,
      "loss": 1.6066,
      "step": 8044
    },
    {
      "epoch": 0.3577430972388956,
      "grad_norm": 0.08467015624046326,
      "learning_rate": 0.0009846019966482564,
      "loss": 1.6128,
      "step": 8046
    },
    {
      "epoch": 0.3578320216975679,
      "grad_norm": 0.08793745934963226,
      "learning_rate": 0.0009845932991697373,
      "loss": 1.6093,
      "step": 8048
    },
    {
      "epoch": 0.35792094615624026,
      "grad_norm": 0.08105975389480591,
      "learning_rate": 0.000984584599273986,
      "loss": 1.602,
      "step": 8050
    },
    {
      "epoch": 0.35800987061491263,
      "grad_norm": 0.08754699677228928,
      "learning_rate": 0.0009845758969610466,
      "loss": 1.596,
      "step": 8052
    },
    {
      "epoch": 0.358098795073585,
      "grad_norm": 0.08423270285129547,
      "learning_rate": 0.000984567192230962,
      "loss": 1.5994,
      "step": 8054
    },
    {
      "epoch": 0.35818771953225736,
      "grad_norm": 0.08522899448871613,
      "learning_rate": 0.000984558485083776,
      "loss": 1.6003,
      "step": 8056
    },
    {
      "epoch": 0.35827664399092973,
      "grad_norm": 0.08278319239616394,
      "learning_rate": 0.0009845497755195318,
      "loss": 1.6026,
      "step": 8058
    },
    {
      "epoch": 0.35836556844960205,
      "grad_norm": 0.08185625076293945,
      "learning_rate": 0.0009845410635382728,
      "loss": 1.5918,
      "step": 8060
    },
    {
      "epoch": 0.3584544929082744,
      "grad_norm": 0.08351411670446396,
      "learning_rate": 0.0009845323491400426,
      "loss": 1.6026,
      "step": 8062
    },
    {
      "epoch": 0.3585434173669468,
      "grad_norm": 0.08093686401844025,
      "learning_rate": 0.0009845236323248845,
      "loss": 1.6011,
      "step": 8064
    },
    {
      "epoch": 0.35863234182561915,
      "grad_norm": 0.08070489019155502,
      "learning_rate": 0.0009845149130928421,
      "loss": 1.5963,
      "step": 8066
    },
    {
      "epoch": 0.3587212662842915,
      "grad_norm": 0.08184243738651276,
      "learning_rate": 0.0009845061914439592,
      "loss": 1.5887,
      "step": 8068
    },
    {
      "epoch": 0.35881019074296383,
      "grad_norm": 0.08118681609630585,
      "learning_rate": 0.0009844974673782786,
      "loss": 1.5864,
      "step": 8070
    },
    {
      "epoch": 0.3588991152016362,
      "grad_norm": 0.08214083313941956,
      "learning_rate": 0.0009844887408958444,
      "loss": 1.5933,
      "step": 8072
    },
    {
      "epoch": 0.35898803966030857,
      "grad_norm": 0.07901991903781891,
      "learning_rate": 0.0009844800119967,
      "loss": 1.5918,
      "step": 8074
    },
    {
      "epoch": 0.35907696411898093,
      "grad_norm": 0.0798892080783844,
      "learning_rate": 0.000984471280680889,
      "loss": 1.5866,
      "step": 8076
    },
    {
      "epoch": 0.3591658885776533,
      "grad_norm": 0.08176968991756439,
      "learning_rate": 0.0009844625469484546,
      "loss": 1.5841,
      "step": 8078
    },
    {
      "epoch": 0.3592548130363256,
      "grad_norm": 0.0794360563158989,
      "learning_rate": 0.0009844538107994405,
      "loss": 1.589,
      "step": 8080
    },
    {
      "epoch": 0.359343737494998,
      "grad_norm": 0.08003082126379013,
      "learning_rate": 0.0009844450722338906,
      "loss": 1.5884,
      "step": 8082
    },
    {
      "epoch": 0.35943266195367035,
      "grad_norm": 0.08419719338417053,
      "learning_rate": 0.000984436331251848,
      "loss": 1.5887,
      "step": 8084
    },
    {
      "epoch": 0.3595215864123427,
      "grad_norm": 0.07977907359600067,
      "learning_rate": 0.0009844275878533569,
      "loss": 1.5834,
      "step": 8086
    },
    {
      "epoch": 0.3596105108710151,
      "grad_norm": 0.08345074206590652,
      "learning_rate": 0.0009844188420384604,
      "loss": 1.584,
      "step": 8088
    },
    {
      "epoch": 0.35969943532968746,
      "grad_norm": 0.08102136850357056,
      "learning_rate": 0.0009844100938072023,
      "loss": 1.5883,
      "step": 8090
    },
    {
      "epoch": 0.35978835978835977,
      "grad_norm": 0.08167117834091187,
      "learning_rate": 0.0009844013431596263,
      "loss": 1.5822,
      "step": 8092
    },
    {
      "epoch": 0.35987728424703214,
      "grad_norm": 0.0804857537150383,
      "learning_rate": 0.000984392590095776,
      "loss": 1.5835,
      "step": 8094
    },
    {
      "epoch": 0.3599662087057045,
      "grad_norm": 0.08060716092586517,
      "learning_rate": 0.000984383834615695,
      "loss": 1.5853,
      "step": 8096
    },
    {
      "epoch": 0.3600551331643769,
      "grad_norm": 0.08844839036464691,
      "learning_rate": 0.000984375076719427,
      "loss": 1.582,
      "step": 8098
    },
    {
      "epoch": 0.36014405762304924,
      "grad_norm": 0.07595451176166534,
      "learning_rate": 0.0009843663164070158,
      "loss": 1.5848,
      "step": 8100
    },
    {
      "epoch": 0.36023298208172155,
      "grad_norm": 0.08353257924318314,
      "learning_rate": 0.000984357553678505,
      "loss": 1.5796,
      "step": 8102
    },
    {
      "epoch": 0.3603219065403939,
      "grad_norm": 0.08055184781551361,
      "learning_rate": 0.0009843487885339384,
      "loss": 1.5814,
      "step": 8104
    },
    {
      "epoch": 0.3604108309990663,
      "grad_norm": 0.07868727296590805,
      "learning_rate": 0.0009843400209733593,
      "loss": 1.5861,
      "step": 8106
    },
    {
      "epoch": 0.36049975545773866,
      "grad_norm": 0.08025962114334106,
      "learning_rate": 0.000984331250996812,
      "loss": 1.5845,
      "step": 8108
    },
    {
      "epoch": 0.360588679916411,
      "grad_norm": 0.07819036394357681,
      "learning_rate": 0.0009843224786043401,
      "loss": 1.5807,
      "step": 8110
    },
    {
      "epoch": 0.3606776043750834,
      "grad_norm": 0.07715727388858795,
      "learning_rate": 0.0009843137037959873,
      "loss": 1.5842,
      "step": 8112
    },
    {
      "epoch": 0.3607665288337557,
      "grad_norm": 0.08035559952259064,
      "learning_rate": 0.0009843049265717975,
      "loss": 1.5746,
      "step": 8114
    },
    {
      "epoch": 0.3608554532924281,
      "grad_norm": 0.08042401075363159,
      "learning_rate": 0.000984296146931814,
      "loss": 1.5779,
      "step": 8116
    },
    {
      "epoch": 0.36094437775110044,
      "grad_norm": 0.0793127790093422,
      "learning_rate": 0.000984287364876081,
      "loss": 1.5806,
      "step": 8118
    },
    {
      "epoch": 0.3610333022097728,
      "grad_norm": 0.07924724370241165,
      "learning_rate": 0.0009842785804046424,
      "loss": 1.5812,
      "step": 8120
    },
    {
      "epoch": 0.3611222266684452,
      "grad_norm": 0.08291906118392944,
      "learning_rate": 0.0009842697935175417,
      "loss": 1.5734,
      "step": 8122
    },
    {
      "epoch": 0.3612111511271175,
      "grad_norm": 0.08052452653646469,
      "learning_rate": 0.0009842610042148229,
      "loss": 1.575,
      "step": 8124
    },
    {
      "epoch": 0.36130007558578986,
      "grad_norm": 0.07844741642475128,
      "learning_rate": 0.0009842522124965298,
      "loss": 1.5802,
      "step": 8126
    },
    {
      "epoch": 0.3613890000444622,
      "grad_norm": 0.08405445516109467,
      "learning_rate": 0.0009842434183627063,
      "loss": 1.571,
      "step": 8128
    },
    {
      "epoch": 0.3614779245031346,
      "grad_norm": 0.07999279350042343,
      "learning_rate": 0.0009842346218133964,
      "loss": 1.5703,
      "step": 8130
    },
    {
      "epoch": 0.36156684896180696,
      "grad_norm": 0.08422758430242538,
      "learning_rate": 0.0009842258228486436,
      "loss": 1.5774,
      "step": 8132
    },
    {
      "epoch": 0.3616557734204793,
      "grad_norm": 0.07723760604858398,
      "learning_rate": 0.0009842170214684923,
      "loss": 1.5756,
      "step": 8134
    },
    {
      "epoch": 0.36174469787915164,
      "grad_norm": 0.07770794630050659,
      "learning_rate": 0.0009842082176729858,
      "loss": 1.5674,
      "step": 8136
    },
    {
      "epoch": 0.361833622337824,
      "grad_norm": 0.07809870690107346,
      "learning_rate": 0.0009841994114621686,
      "loss": 1.5741,
      "step": 8138
    },
    {
      "epoch": 0.3619225467964964,
      "grad_norm": 0.08160432428121567,
      "learning_rate": 0.0009841906028360842,
      "loss": 1.5726,
      "step": 8140
    },
    {
      "epoch": 0.36201147125516875,
      "grad_norm": 0.08070631325244904,
      "learning_rate": 0.0009841817917947767,
      "loss": 1.5736,
      "step": 8142
    },
    {
      "epoch": 0.3621003957138411,
      "grad_norm": 0.07993360608816147,
      "learning_rate": 0.0009841729783382903,
      "loss": 1.5757,
      "step": 8144
    },
    {
      "epoch": 0.36218932017251343,
      "grad_norm": 0.07653278112411499,
      "learning_rate": 0.0009841641624666684,
      "loss": 1.5737,
      "step": 8146
    },
    {
      "epoch": 0.3622782446311858,
      "grad_norm": 0.07840976864099503,
      "learning_rate": 0.0009841553441799554,
      "loss": 1.5702,
      "step": 8148
    },
    {
      "epoch": 0.36236716908985817,
      "grad_norm": 0.07674574106931686,
      "learning_rate": 0.0009841465234781952,
      "loss": 1.5761,
      "step": 8150
    },
    {
      "epoch": 0.36245609354853053,
      "grad_norm": 0.08122869580984116,
      "learning_rate": 0.0009841377003614318,
      "loss": 1.5696,
      "step": 8152
    },
    {
      "epoch": 0.3625450180072029,
      "grad_norm": 0.07849126309156418,
      "learning_rate": 0.0009841288748297092,
      "loss": 1.5692,
      "step": 8154
    },
    {
      "epoch": 0.3626339424658752,
      "grad_norm": 0.0775504857301712,
      "learning_rate": 0.0009841200468830714,
      "loss": 1.5748,
      "step": 8156
    },
    {
      "epoch": 0.3627228669245476,
      "grad_norm": 0.07808183133602142,
      "learning_rate": 0.0009841112165215624,
      "loss": 1.5702,
      "step": 8158
    },
    {
      "epoch": 0.36281179138321995,
      "grad_norm": 0.0813973993062973,
      "learning_rate": 0.0009841023837452263,
      "loss": 1.5721,
      "step": 8160
    },
    {
      "epoch": 0.3629007158418923,
      "grad_norm": 0.07763282209634781,
      "learning_rate": 0.0009840935485541072,
      "loss": 1.5666,
      "step": 8162
    },
    {
      "epoch": 0.3629896403005647,
      "grad_norm": 0.07707773894071579,
      "learning_rate": 0.000984084710948249,
      "loss": 1.5699,
      "step": 8164
    },
    {
      "epoch": 0.36307856475923705,
      "grad_norm": 0.07617053389549255,
      "learning_rate": 0.0009840758709276961,
      "loss": 1.5731,
      "step": 8166
    },
    {
      "epoch": 0.36316748921790937,
      "grad_norm": 0.07609335333108902,
      "learning_rate": 0.0009840670284924921,
      "loss": 1.5734,
      "step": 8168
    },
    {
      "epoch": 0.36325641367658174,
      "grad_norm": 0.07503112405538559,
      "learning_rate": 0.0009840581836426817,
      "loss": 1.5689,
      "step": 8170
    },
    {
      "epoch": 0.3633453381352541,
      "grad_norm": 0.07643923163414001,
      "learning_rate": 0.0009840493363783085,
      "loss": 1.5717,
      "step": 8172
    },
    {
      "epoch": 0.36343426259392647,
      "grad_norm": 0.07574044167995453,
      "learning_rate": 0.0009840404866994171,
      "loss": 1.569,
      "step": 8174
    },
    {
      "epoch": 0.36352318705259884,
      "grad_norm": 0.07417095452547073,
      "learning_rate": 0.0009840316346060513,
      "loss": 1.569,
      "step": 8176
    },
    {
      "epoch": 0.36361211151127115,
      "grad_norm": 0.07547099888324738,
      "learning_rate": 0.0009840227800982553,
      "loss": 1.5751,
      "step": 8178
    },
    {
      "epoch": 0.3637010359699435,
      "grad_norm": 0.07645308971405029,
      "learning_rate": 0.0009840139231760735,
      "loss": 1.5648,
      "step": 8180
    },
    {
      "epoch": 0.3637899604286159,
      "grad_norm": 0.07473346590995789,
      "learning_rate": 0.0009840050638395498,
      "loss": 1.5661,
      "step": 8182
    },
    {
      "epoch": 0.36387888488728826,
      "grad_norm": 0.07807636260986328,
      "learning_rate": 0.0009839962020887286,
      "loss": 1.5724,
      "step": 8184
    },
    {
      "epoch": 0.3639678093459606,
      "grad_norm": 0.07300268113613129,
      "learning_rate": 0.0009839873379236537,
      "loss": 1.571,
      "step": 8186
    },
    {
      "epoch": 0.36405673380463294,
      "grad_norm": 0.07703215628862381,
      "learning_rate": 0.00098397847134437,
      "loss": 1.5722,
      "step": 8188
    },
    {
      "epoch": 0.3641456582633053,
      "grad_norm": 0.07669734954833984,
      "learning_rate": 0.000983969602350921,
      "loss": 1.5676,
      "step": 8190
    },
    {
      "epoch": 0.3642345827219777,
      "grad_norm": 0.07603123784065247,
      "learning_rate": 0.0009839607309433515,
      "loss": 1.5688,
      "step": 8192
    },
    {
      "epoch": 0.36432350718065004,
      "grad_norm": 0.07759499549865723,
      "learning_rate": 0.0009839518571217055,
      "loss": 1.5643,
      "step": 8194
    },
    {
      "epoch": 0.3644124316393224,
      "grad_norm": 0.07855424284934998,
      "learning_rate": 0.0009839429808860274,
      "loss": 1.5641,
      "step": 8196
    },
    {
      "epoch": 0.3645013560979948,
      "grad_norm": 0.07806675881147385,
      "learning_rate": 0.0009839341022363612,
      "loss": 1.572,
      "step": 8198
    },
    {
      "epoch": 0.3645902805566671,
      "grad_norm": 0.08267048001289368,
      "learning_rate": 0.0009839252211727514,
      "loss": 1.5711,
      "step": 8200
    },
    {
      "epoch": 0.36467920501533946,
      "grad_norm": 0.07525893300771713,
      "learning_rate": 0.0009839163376952425,
      "loss": 1.5687,
      "step": 8202
    },
    {
      "epoch": 0.3647681294740118,
      "grad_norm": 0.07745590060949326,
      "learning_rate": 0.0009839074518038784,
      "loss": 1.5629,
      "step": 8204
    },
    {
      "epoch": 0.3648570539326842,
      "grad_norm": 0.07969274371862411,
      "learning_rate": 0.0009838985634987036,
      "loss": 1.5702,
      "step": 8206
    },
    {
      "epoch": 0.36494597839135656,
      "grad_norm": 0.0816626101732254,
      "learning_rate": 0.0009838896727797624,
      "loss": 1.5599,
      "step": 8208
    },
    {
      "epoch": 0.3650349028500289,
      "grad_norm": 0.07653602957725525,
      "learning_rate": 0.0009838807796470992,
      "loss": 1.5618,
      "step": 8210
    },
    {
      "epoch": 0.36512382730870124,
      "grad_norm": 0.08516646176576614,
      "learning_rate": 0.0009838718841007586,
      "loss": 1.5642,
      "step": 8212
    },
    {
      "epoch": 0.3652127517673736,
      "grad_norm": 0.07569476962089539,
      "learning_rate": 0.0009838629861407843,
      "loss": 1.5597,
      "step": 8214
    },
    {
      "epoch": 0.365301676226046,
      "grad_norm": 0.07699557393789291,
      "learning_rate": 0.0009838540857672214,
      "loss": 1.5649,
      "step": 8216
    },
    {
      "epoch": 0.36539060068471835,
      "grad_norm": 0.07943487912416458,
      "learning_rate": 0.000983845182980114,
      "loss": 1.5606,
      "step": 8218
    },
    {
      "epoch": 0.3654795251433907,
      "grad_norm": 0.07496616989374161,
      "learning_rate": 0.0009838362777795065,
      "loss": 1.5649,
      "step": 8220
    },
    {
      "epoch": 0.36556844960206303,
      "grad_norm": 0.07536393404006958,
      "learning_rate": 0.0009838273701654433,
      "loss": 1.5696,
      "step": 8222
    },
    {
      "epoch": 0.3656573740607354,
      "grad_norm": 0.07574569433927536,
      "learning_rate": 0.0009838184601379688,
      "loss": 1.5623,
      "step": 8224
    },
    {
      "epoch": 0.36574629851940776,
      "grad_norm": 0.07682286202907562,
      "learning_rate": 0.0009838095476971274,
      "loss": 1.5676,
      "step": 8226
    },
    {
      "epoch": 0.36583522297808013,
      "grad_norm": 0.08097031712532043,
      "learning_rate": 0.000983800632842964,
      "loss": 1.5641,
      "step": 8228
    },
    {
      "epoch": 0.3659241474367525,
      "grad_norm": 0.07592246681451797,
      "learning_rate": 0.0009837917155755226,
      "loss": 1.5562,
      "step": 8230
    },
    {
      "epoch": 0.3660130718954248,
      "grad_norm": 0.07768161594867706,
      "learning_rate": 0.0009837827958948477,
      "loss": 1.5638,
      "step": 8232
    },
    {
      "epoch": 0.3661019963540972,
      "grad_norm": 0.07560420036315918,
      "learning_rate": 0.0009837738738009841,
      "loss": 1.5625,
      "step": 8234
    },
    {
      "epoch": 0.36619092081276955,
      "grad_norm": 0.07644996047019958,
      "learning_rate": 0.000983764949293976,
      "loss": 1.5602,
      "step": 8236
    },
    {
      "epoch": 0.3662798452714419,
      "grad_norm": 0.07451751828193665,
      "learning_rate": 0.0009837560223738679,
      "loss": 1.5608,
      "step": 8238
    },
    {
      "epoch": 0.3663687697301143,
      "grad_norm": 0.0761575847864151,
      "learning_rate": 0.0009837470930407046,
      "loss": 1.5542,
      "step": 8240
    },
    {
      "epoch": 0.36645769418878665,
      "grad_norm": 0.0770341232419014,
      "learning_rate": 0.0009837381612945305,
      "loss": 1.5584,
      "step": 8242
    },
    {
      "epoch": 0.36654661864745897,
      "grad_norm": 0.07663694024085999,
      "learning_rate": 0.0009837292271353902,
      "loss": 1.5646,
      "step": 8244
    },
    {
      "epoch": 0.36663554310613133,
      "grad_norm": 0.0755605474114418,
      "learning_rate": 0.0009837202905633282,
      "loss": 1.5621,
      "step": 8246
    },
    {
      "epoch": 0.3667244675648037,
      "grad_norm": 0.07471001893281937,
      "learning_rate": 0.0009837113515783892,
      "loss": 1.5641,
      "step": 8248
    },
    {
      "epoch": 0.36681339202347607,
      "grad_norm": 0.07854770869016647,
      "learning_rate": 0.0009837024101806173,
      "loss": 1.5628,
      "step": 8250
    },
    {
      "epoch": 0.36690231648214844,
      "grad_norm": 0.07697580009698868,
      "learning_rate": 0.000983693466370058,
      "loss": 1.561,
      "step": 8252
    },
    {
      "epoch": 0.36699124094082075,
      "grad_norm": 0.07174623012542725,
      "learning_rate": 0.000983684520146755,
      "loss": 1.5571,
      "step": 8254
    },
    {
      "epoch": 0.3670801653994931,
      "grad_norm": 0.0765925943851471,
      "learning_rate": 0.0009836755715107535,
      "loss": 1.5634,
      "step": 8256
    },
    {
      "epoch": 0.3671690898581655,
      "grad_norm": 0.07561610639095306,
      "learning_rate": 0.000983666620462098,
      "loss": 1.5588,
      "step": 8258
    },
    {
      "epoch": 0.36725801431683786,
      "grad_norm": 0.07434527575969696,
      "learning_rate": 0.000983657667000833,
      "loss": 1.5617,
      "step": 8260
    },
    {
      "epoch": 0.3673469387755102,
      "grad_norm": 0.07405678927898407,
      "learning_rate": 0.0009836487111270034,
      "loss": 1.5646,
      "step": 8262
    },
    {
      "epoch": 0.36743586323418254,
      "grad_norm": 0.07639020681381226,
      "learning_rate": 0.0009836397528406534,
      "loss": 1.5586,
      "step": 8264
    },
    {
      "epoch": 0.3675247876928549,
      "grad_norm": 0.07487154006958008,
      "learning_rate": 0.0009836307921418284,
      "loss": 1.5547,
      "step": 8266
    },
    {
      "epoch": 0.36761371215152727,
      "grad_norm": 0.07496780157089233,
      "learning_rate": 0.0009836218290305728,
      "loss": 1.5611,
      "step": 8268
    },
    {
      "epoch": 0.36770263661019964,
      "grad_norm": 0.07398813217878342,
      "learning_rate": 0.000983612863506931,
      "loss": 1.558,
      "step": 8270
    },
    {
      "epoch": 0.367791561068872,
      "grad_norm": 0.07573290914297104,
      "learning_rate": 0.000983603895570948,
      "loss": 1.5595,
      "step": 8272
    },
    {
      "epoch": 0.3678804855275444,
      "grad_norm": 0.07165679335594177,
      "learning_rate": 0.0009835949252226686,
      "loss": 1.5649,
      "step": 8274
    },
    {
      "epoch": 0.3679694099862167,
      "grad_norm": 0.07558693736791611,
      "learning_rate": 0.0009835859524621373,
      "loss": 1.5568,
      "step": 8276
    },
    {
      "epoch": 0.36805833444488906,
      "grad_norm": 0.0720970556139946,
      "learning_rate": 0.000983576977289399,
      "loss": 1.5583,
      "step": 8278
    },
    {
      "epoch": 0.3681472589035614,
      "grad_norm": 0.07370097190141678,
      "learning_rate": 0.0009835679997044985,
      "loss": 1.5612,
      "step": 8280
    },
    {
      "epoch": 0.3682361833622338,
      "grad_norm": 0.07401144504547119,
      "learning_rate": 0.0009835590197074806,
      "loss": 1.5596,
      "step": 8282
    },
    {
      "epoch": 0.36832510782090616,
      "grad_norm": 0.07251115888357162,
      "learning_rate": 0.0009835500372983902,
      "loss": 1.5569,
      "step": 8284
    },
    {
      "epoch": 0.3684140322795785,
      "grad_norm": 0.07301607728004456,
      "learning_rate": 0.0009835410524772718,
      "loss": 1.5591,
      "step": 8286
    },
    {
      "epoch": 0.36850295673825084,
      "grad_norm": 0.07856374233961105,
      "learning_rate": 0.0009835320652441702,
      "loss": 1.5658,
      "step": 8288
    },
    {
      "epoch": 0.3685918811969232,
      "grad_norm": 0.07428453117609024,
      "learning_rate": 0.0009835230755991305,
      "loss": 1.5631,
      "step": 8290
    },
    {
      "epoch": 0.3686808056555956,
      "grad_norm": 0.07632127404212952,
      "learning_rate": 0.0009835140835421977,
      "loss": 1.5659,
      "step": 8292
    },
    {
      "epoch": 0.36876973011426795,
      "grad_norm": 0.07355497777462006,
      "learning_rate": 0.0009835050890734161,
      "loss": 1.5582,
      "step": 8294
    },
    {
      "epoch": 0.3688586545729403,
      "grad_norm": 0.07494645565748215,
      "learning_rate": 0.0009834960921928308,
      "loss": 1.5584,
      "step": 8296
    },
    {
      "epoch": 0.3689475790316126,
      "grad_norm": 0.07737577706575394,
      "learning_rate": 0.000983487092900487,
      "loss": 1.5562,
      "step": 8298
    },
    {
      "epoch": 0.369036503490285,
      "grad_norm": 0.07639828324317932,
      "learning_rate": 0.000983478091196429,
      "loss": 1.5646,
      "step": 8300
    },
    {
      "epoch": 0.36912542794895736,
      "grad_norm": 0.07329989224672318,
      "learning_rate": 0.0009834690870807023,
      "loss": 1.5579,
      "step": 8302
    },
    {
      "epoch": 0.36921435240762973,
      "grad_norm": 0.07461026310920715,
      "learning_rate": 0.0009834600805533516,
      "loss": 1.5522,
      "step": 8304
    },
    {
      "epoch": 0.3693032768663021,
      "grad_norm": 0.07623513787984848,
      "learning_rate": 0.0009834510716144214,
      "loss": 1.5643,
      "step": 8306
    },
    {
      "epoch": 0.3693922013249744,
      "grad_norm": 0.07590184360742569,
      "learning_rate": 0.0009834420602639574,
      "loss": 1.5595,
      "step": 8308
    },
    {
      "epoch": 0.3694811257836468,
      "grad_norm": 0.07395032793283463,
      "learning_rate": 0.0009834330465020038,
      "loss": 1.5598,
      "step": 8310
    },
    {
      "epoch": 0.36957005024231915,
      "grad_norm": 0.07140004634857178,
      "learning_rate": 0.0009834240303286063,
      "loss": 1.5628,
      "step": 8312
    },
    {
      "epoch": 0.3696589747009915,
      "grad_norm": 0.0748644694685936,
      "learning_rate": 0.0009834150117438091,
      "loss": 1.5601,
      "step": 8314
    },
    {
      "epoch": 0.3697478991596639,
      "grad_norm": 0.07483812421560287,
      "learning_rate": 0.0009834059907476579,
      "loss": 1.5533,
      "step": 8316
    },
    {
      "epoch": 0.3698368236183362,
      "grad_norm": 0.07460799068212509,
      "learning_rate": 0.0009833969673401972,
      "loss": 1.5583,
      "step": 8318
    },
    {
      "epoch": 0.36992574807700856,
      "grad_norm": 0.0767635703086853,
      "learning_rate": 0.0009833879415214722,
      "loss": 1.5656,
      "step": 8320
    },
    {
      "epoch": 0.37001467253568093,
      "grad_norm": 0.07315431535243988,
      "learning_rate": 0.0009833789132915282,
      "loss": 1.5562,
      "step": 8322
    },
    {
      "epoch": 0.3701035969943533,
      "grad_norm": 0.07698900997638702,
      "learning_rate": 0.0009833698826504096,
      "loss": 1.5598,
      "step": 8324
    },
    {
      "epoch": 0.37019252145302567,
      "grad_norm": 0.07581165432929993,
      "learning_rate": 0.0009833608495981618,
      "loss": 1.5574,
      "step": 8326
    },
    {
      "epoch": 0.37028144591169804,
      "grad_norm": 0.07427161186933517,
      "learning_rate": 0.0009833518141348298,
      "loss": 1.5555,
      "step": 8328
    },
    {
      "epoch": 0.37037037037037035,
      "grad_norm": 0.07862924039363861,
      "learning_rate": 0.000983342776260459,
      "loss": 1.5601,
      "step": 8330
    },
    {
      "epoch": 0.3704592948290427,
      "grad_norm": 0.07485569268465042,
      "learning_rate": 0.000983333735975094,
      "loss": 1.5526,
      "step": 8332
    },
    {
      "epoch": 0.3705482192877151,
      "grad_norm": 0.07545112073421478,
      "learning_rate": 0.0009833246932787801,
      "loss": 1.5548,
      "step": 8334
    },
    {
      "epoch": 0.37063714374638745,
      "grad_norm": 0.07797663658857346,
      "learning_rate": 0.0009833156481715624,
      "loss": 1.5603,
      "step": 8336
    },
    {
      "epoch": 0.3707260682050598,
      "grad_norm": 0.07360795885324478,
      "learning_rate": 0.0009833066006534861,
      "loss": 1.5535,
      "step": 8338
    },
    {
      "epoch": 0.37081499266373213,
      "grad_norm": 0.07508852332830429,
      "learning_rate": 0.0009832975507245963,
      "loss": 1.562,
      "step": 8340
    },
    {
      "epoch": 0.3709039171224045,
      "grad_norm": 0.07691474258899689,
      "learning_rate": 0.000983288498384938,
      "loss": 1.5559,
      "step": 8342
    },
    {
      "epoch": 0.37099284158107687,
      "grad_norm": 0.07170414924621582,
      "learning_rate": 0.0009832794436345565,
      "loss": 1.5595,
      "step": 8344
    },
    {
      "epoch": 0.37108176603974924,
      "grad_norm": 0.07313521206378937,
      "learning_rate": 0.000983270386473497,
      "loss": 1.5568,
      "step": 8346
    },
    {
      "epoch": 0.3711706904984216,
      "grad_norm": 0.07443317025899887,
      "learning_rate": 0.0009832613269018043,
      "loss": 1.5551,
      "step": 8348
    },
    {
      "epoch": 0.371259614957094,
      "grad_norm": 0.0727194994688034,
      "learning_rate": 0.000983252264919524,
      "loss": 1.5558,
      "step": 8350
    },
    {
      "epoch": 0.3713485394157663,
      "grad_norm": 0.07491002231836319,
      "learning_rate": 0.0009832432005267013,
      "loss": 1.5553,
      "step": 8352
    },
    {
      "epoch": 0.37143746387443866,
      "grad_norm": 0.07130132615566254,
      "learning_rate": 0.0009832341337233813,
      "loss": 1.5515,
      "step": 8354
    },
    {
      "epoch": 0.371526388333111,
      "grad_norm": 0.0775056853890419,
      "learning_rate": 0.0009832250645096091,
      "loss": 1.5612,
      "step": 8356
    },
    {
      "epoch": 0.3716153127917834,
      "grad_norm": 0.0759596973657608,
      "learning_rate": 0.00098321599288543,
      "loss": 1.554,
      "step": 8358
    },
    {
      "epoch": 0.37170423725045576,
      "grad_norm": 0.0767047107219696,
      "learning_rate": 0.0009832069188508896,
      "loss": 1.5613,
      "step": 8360
    },
    {
      "epoch": 0.3717931617091281,
      "grad_norm": 0.07511959224939346,
      "learning_rate": 0.0009831978424060328,
      "loss": 1.5553,
      "step": 8362
    },
    {
      "epoch": 0.37188208616780044,
      "grad_norm": 0.07674729824066162,
      "learning_rate": 0.0009831887635509049,
      "loss": 1.5505,
      "step": 8364
    },
    {
      "epoch": 0.3719710106264728,
      "grad_norm": 0.07360707968473434,
      "learning_rate": 0.0009831796822855512,
      "loss": 1.556,
      "step": 8366
    },
    {
      "epoch": 0.3720599350851452,
      "grad_norm": 0.0746261477470398,
      "learning_rate": 0.000983170598610017,
      "loss": 1.5553,
      "step": 8368
    },
    {
      "epoch": 0.37214885954381755,
      "grad_norm": 0.07465831935405731,
      "learning_rate": 0.0009831615125243478,
      "loss": 1.55,
      "step": 8370
    },
    {
      "epoch": 0.37223778400248986,
      "grad_norm": 0.07530109584331512,
      "learning_rate": 0.0009831524240285887,
      "loss": 1.5556,
      "step": 8372
    },
    {
      "epoch": 0.3723267084611622,
      "grad_norm": 0.07392653822898865,
      "learning_rate": 0.0009831433331227851,
      "loss": 1.5582,
      "step": 8374
    },
    {
      "epoch": 0.3724156329198346,
      "grad_norm": 0.0778929591178894,
      "learning_rate": 0.0009831342398069823,
      "loss": 1.5604,
      "step": 8376
    },
    {
      "epoch": 0.37250455737850696,
      "grad_norm": 0.07217849791049957,
      "learning_rate": 0.000983125144081226,
      "loss": 1.5528,
      "step": 8378
    },
    {
      "epoch": 0.37259348183717933,
      "grad_norm": 0.07852315157651901,
      "learning_rate": 0.000983116045945561,
      "loss": 1.56,
      "step": 8380
    },
    {
      "epoch": 0.3726824062958517,
      "grad_norm": 0.07284464687108994,
      "learning_rate": 0.000983106945400033,
      "loss": 1.5594,
      "step": 8382
    },
    {
      "epoch": 0.372771330754524,
      "grad_norm": 0.07296974211931229,
      "learning_rate": 0.0009830978424446874,
      "loss": 1.5558,
      "step": 8384
    },
    {
      "epoch": 0.3728602552131964,
      "grad_norm": 0.07269974797964096,
      "learning_rate": 0.0009830887370795695,
      "loss": 1.5563,
      "step": 8386
    },
    {
      "epoch": 0.37294917967186875,
      "grad_norm": 0.07310895621776581,
      "learning_rate": 0.0009830796293047249,
      "loss": 1.559,
      "step": 8388
    },
    {
      "epoch": 0.3730381041305411,
      "grad_norm": 0.07368705421686172,
      "learning_rate": 0.000983070519120199,
      "loss": 1.5569,
      "step": 8390
    },
    {
      "epoch": 0.3731270285892135,
      "grad_norm": 0.07440350204706192,
      "learning_rate": 0.000983061406526037,
      "loss": 1.5542,
      "step": 8392
    },
    {
      "epoch": 0.3732159530478858,
      "grad_norm": 0.07380738109350204,
      "learning_rate": 0.0009830522915222845,
      "loss": 1.5558,
      "step": 8394
    },
    {
      "epoch": 0.37330487750655816,
      "grad_norm": 0.0736425444483757,
      "learning_rate": 0.000983043174108987,
      "loss": 1.5493,
      "step": 8396
    },
    {
      "epoch": 0.37339380196523053,
      "grad_norm": 0.07320426404476166,
      "learning_rate": 0.0009830340542861902,
      "loss": 1.5516,
      "step": 8398
    },
    {
      "epoch": 0.3734827264239029,
      "grad_norm": 0.07398249953985214,
      "learning_rate": 0.0009830249320539391,
      "loss": 1.5563,
      "step": 8400
    },
    {
      "epoch": 0.37357165088257527,
      "grad_norm": 0.0731641873717308,
      "learning_rate": 0.0009830158074122795,
      "loss": 1.5547,
      "step": 8402
    },
    {
      "epoch": 0.37366057534124764,
      "grad_norm": 0.07476949691772461,
      "learning_rate": 0.000983006680361257,
      "loss": 1.5563,
      "step": 8404
    },
    {
      "epoch": 0.37374949979991995,
      "grad_norm": 0.07096338272094727,
      "learning_rate": 0.000982997550900917,
      "loss": 1.5579,
      "step": 8406
    },
    {
      "epoch": 0.3738384242585923,
      "grad_norm": 0.0731477290391922,
      "learning_rate": 0.0009829884190313048,
      "loss": 1.5571,
      "step": 8408
    },
    {
      "epoch": 0.3739273487172647,
      "grad_norm": 0.07443249970674515,
      "learning_rate": 0.0009829792847524663,
      "loss": 1.5524,
      "step": 8410
    },
    {
      "epoch": 0.37401627317593705,
      "grad_norm": 0.07522628456354141,
      "learning_rate": 0.000982970148064447,
      "loss": 1.5515,
      "step": 8412
    },
    {
      "epoch": 0.3741051976346094,
      "grad_norm": 0.08042244613170624,
      "learning_rate": 0.0009829610089672925,
      "loss": 1.552,
      "step": 8414
    },
    {
      "epoch": 0.37419412209328173,
      "grad_norm": 0.0764075294137001,
      "learning_rate": 0.0009829518674610482,
      "loss": 1.5509,
      "step": 8416
    },
    {
      "epoch": 0.3742830465519541,
      "grad_norm": 0.07679541409015656,
      "learning_rate": 0.0009829427235457599,
      "loss": 1.5558,
      "step": 8418
    },
    {
      "epoch": 0.37437197101062647,
      "grad_norm": 0.07601416856050491,
      "learning_rate": 0.0009829335772214732,
      "loss": 1.5551,
      "step": 8420
    },
    {
      "epoch": 0.37446089546929884,
      "grad_norm": 0.07198473066091537,
      "learning_rate": 0.0009829244284882336,
      "loss": 1.5515,
      "step": 8422
    },
    {
      "epoch": 0.3745498199279712,
      "grad_norm": 0.0754043385386467,
      "learning_rate": 0.0009829152773460869,
      "loss": 1.5522,
      "step": 8424
    },
    {
      "epoch": 0.3746387443866436,
      "grad_norm": 0.07835698872804642,
      "learning_rate": 0.0009829061237950783,
      "loss": 1.5558,
      "step": 8426
    },
    {
      "epoch": 0.3747276688453159,
      "grad_norm": 0.07373791188001633,
      "learning_rate": 0.000982896967835254,
      "loss": 1.5485,
      "step": 8428
    },
    {
      "epoch": 0.37481659330398825,
      "grad_norm": 0.07449112087488174,
      "learning_rate": 0.0009828878094666595,
      "loss": 1.5524,
      "step": 8430
    },
    {
      "epoch": 0.3749055177626606,
      "grad_norm": 0.07234968990087509,
      "learning_rate": 0.0009828786486893405,
      "loss": 1.5579,
      "step": 8432
    },
    {
      "epoch": 0.374994442221333,
      "grad_norm": 0.07672835886478424,
      "learning_rate": 0.0009828694855033425,
      "loss": 1.5522,
      "step": 8434
    },
    {
      "epoch": 0.37508336668000536,
      "grad_norm": 0.07505165040493011,
      "learning_rate": 0.0009828603199087114,
      "loss": 1.5515,
      "step": 8436
    },
    {
      "epoch": 0.37517229113867767,
      "grad_norm": 0.07347539067268372,
      "learning_rate": 0.0009828511519054928,
      "loss": 1.552,
      "step": 8438
    },
    {
      "epoch": 0.37526121559735004,
      "grad_norm": 0.07479852437973022,
      "learning_rate": 0.0009828419814937326,
      "loss": 1.5485,
      "step": 8440
    },
    {
      "epoch": 0.3753501400560224,
      "grad_norm": 0.0721561387181282,
      "learning_rate": 0.0009828328086734766,
      "loss": 1.5542,
      "step": 8442
    },
    {
      "epoch": 0.3754390645146948,
      "grad_norm": 0.07447963207960129,
      "learning_rate": 0.0009828236334447702,
      "loss": 1.5502,
      "step": 8444
    },
    {
      "epoch": 0.37552798897336714,
      "grad_norm": 0.07521109282970428,
      "learning_rate": 0.0009828144558076595,
      "loss": 1.5478,
      "step": 8446
    },
    {
      "epoch": 0.37561691343203946,
      "grad_norm": 0.08029814064502716,
      "learning_rate": 0.00098280527576219,
      "loss": 1.5552,
      "step": 8448
    },
    {
      "epoch": 0.3757058378907118,
      "grad_norm": 0.07279892265796661,
      "learning_rate": 0.0009827960933084076,
      "loss": 1.549,
      "step": 8450
    },
    {
      "epoch": 0.3757947623493842,
      "grad_norm": 0.07226986438035965,
      "learning_rate": 0.0009827869084463583,
      "loss": 1.5512,
      "step": 8452
    },
    {
      "epoch": 0.37588368680805656,
      "grad_norm": 0.07428599148988724,
      "learning_rate": 0.0009827777211760878,
      "loss": 1.5563,
      "step": 8454
    },
    {
      "epoch": 0.37597261126672893,
      "grad_norm": 0.07335691154003143,
      "learning_rate": 0.0009827685314976419,
      "loss": 1.5522,
      "step": 8456
    },
    {
      "epoch": 0.3760615357254013,
      "grad_norm": 0.0702577456831932,
      "learning_rate": 0.000982759339411066,
      "loss": 1.5493,
      "step": 8458
    },
    {
      "epoch": 0.3761504601840736,
      "grad_norm": 0.07333244383335114,
      "learning_rate": 0.000982750144916407,
      "loss": 1.5525,
      "step": 8460
    },
    {
      "epoch": 0.376239384642746,
      "grad_norm": 0.07639877498149872,
      "learning_rate": 0.0009827409480137098,
      "loss": 1.5517,
      "step": 8462
    },
    {
      "epoch": 0.37632830910141835,
      "grad_norm": 0.07263181358575821,
      "learning_rate": 0.0009827317487030207,
      "loss": 1.549,
      "step": 8464
    },
    {
      "epoch": 0.3764172335600907,
      "grad_norm": 0.07310689985752106,
      "learning_rate": 0.0009827225469843854,
      "loss": 1.5462,
      "step": 8466
    },
    {
      "epoch": 0.3765061580187631,
      "grad_norm": 0.07368572801351547,
      "learning_rate": 0.00098271334285785,
      "loss": 1.5532,
      "step": 8468
    },
    {
      "epoch": 0.3765950824774354,
      "grad_norm": 0.07141252607107162,
      "learning_rate": 0.0009827041363234604,
      "loss": 1.5565,
      "step": 8470
    },
    {
      "epoch": 0.37668400693610776,
      "grad_norm": 0.07508658617734909,
      "learning_rate": 0.0009826949273812622,
      "loss": 1.5514,
      "step": 8472
    },
    {
      "epoch": 0.37677293139478013,
      "grad_norm": 0.07459234446287155,
      "learning_rate": 0.0009826857160313018,
      "loss": 1.5596,
      "step": 8474
    },
    {
      "epoch": 0.3768618558534525,
      "grad_norm": 0.07409926503896713,
      "learning_rate": 0.0009826765022736249,
      "loss": 1.5514,
      "step": 8476
    },
    {
      "epoch": 0.37695078031212487,
      "grad_norm": 0.07309357076883316,
      "learning_rate": 0.0009826672861082775,
      "loss": 1.5485,
      "step": 8478
    },
    {
      "epoch": 0.37703970477079723,
      "grad_norm": 0.07403724640607834,
      "learning_rate": 0.0009826580675353054,
      "loss": 1.5558,
      "step": 8480
    },
    {
      "epoch": 0.37712862922946955,
      "grad_norm": 0.07526498287916183,
      "learning_rate": 0.000982648846554755,
      "loss": 1.5511,
      "step": 8482
    },
    {
      "epoch": 0.3772175536881419,
      "grad_norm": 0.07380390167236328,
      "learning_rate": 0.0009826396231666717,
      "loss": 1.5511,
      "step": 8484
    },
    {
      "epoch": 0.3773064781468143,
      "grad_norm": 0.07699756324291229,
      "learning_rate": 0.0009826303973711021,
      "loss": 1.5507,
      "step": 8486
    },
    {
      "epoch": 0.37739540260548665,
      "grad_norm": 0.07383593916893005,
      "learning_rate": 0.000982621169168092,
      "loss": 1.5523,
      "step": 8488
    },
    {
      "epoch": 0.377484327064159,
      "grad_norm": 0.07610490173101425,
      "learning_rate": 0.0009826119385576873,
      "loss": 1.5494,
      "step": 8490
    },
    {
      "epoch": 0.37757325152283133,
      "grad_norm": 0.07537056505680084,
      "learning_rate": 0.000982602705539934,
      "loss": 1.5563,
      "step": 8492
    },
    {
      "epoch": 0.3776621759815037,
      "grad_norm": 0.07339107990264893,
      "learning_rate": 0.0009825934701148784,
      "loss": 1.5511,
      "step": 8494
    },
    {
      "epoch": 0.37775110044017607,
      "grad_norm": 0.0728975385427475,
      "learning_rate": 0.0009825842322825668,
      "loss": 1.5491,
      "step": 8496
    },
    {
      "epoch": 0.37784002489884844,
      "grad_norm": 0.0715377926826477,
      "learning_rate": 0.0009825749920430444,
      "loss": 1.5493,
      "step": 8498
    },
    {
      "epoch": 0.3779289493575208,
      "grad_norm": 0.07521523535251617,
      "learning_rate": 0.000982565749396358,
      "loss": 1.5483,
      "step": 8500
    },
    {
      "epoch": 0.3779289493575208,
      "eval_loss": 1.525511622428894,
      "eval_runtime": 12.3728,
      "eval_samples_per_second": 558.484,
      "eval_steps_per_second": 69.831,
      "step": 8500
    },
    {
      "epoch": 0.3780178738161931,
      "grad_norm": 0.07613959908485413,
      "learning_rate": 0.0009825565043425537,
      "loss": 1.5473,
      "step": 8502
    },
    {
      "epoch": 0.3781067982748655,
      "grad_norm": 0.07284019142389297,
      "learning_rate": 0.0009825472568816775,
      "loss": 1.5514,
      "step": 8504
    },
    {
      "epoch": 0.37819572273353785,
      "grad_norm": 0.07196670770645142,
      "learning_rate": 0.0009825380070137752,
      "loss": 1.557,
      "step": 8506
    },
    {
      "epoch": 0.3782846471922102,
      "grad_norm": 0.07932493090629578,
      "learning_rate": 0.0009825287547388934,
      "loss": 1.5504,
      "step": 8508
    },
    {
      "epoch": 0.3783735716508826,
      "grad_norm": 0.07043527066707611,
      "learning_rate": 0.000982519500057078,
      "loss": 1.5411,
      "step": 8510
    },
    {
      "epoch": 0.37846249610955496,
      "grad_norm": 0.07196062058210373,
      "learning_rate": 0.0009825102429683755,
      "loss": 1.5507,
      "step": 8512
    },
    {
      "epoch": 0.37855142056822727,
      "grad_norm": 0.07716414332389832,
      "learning_rate": 0.0009825009834728315,
      "loss": 1.5461,
      "step": 8514
    },
    {
      "epoch": 0.37864034502689964,
      "grad_norm": 0.06990046054124832,
      "learning_rate": 0.0009824917215704926,
      "loss": 1.5522,
      "step": 8516
    },
    {
      "epoch": 0.378729269485572,
      "grad_norm": 0.07274769991636276,
      "learning_rate": 0.000982482457261405,
      "loss": 1.5507,
      "step": 8518
    },
    {
      "epoch": 0.3788181939442444,
      "grad_norm": 0.07060352712869644,
      "learning_rate": 0.0009824731905456149,
      "loss": 1.5491,
      "step": 8520
    },
    {
      "epoch": 0.37890711840291674,
      "grad_norm": 0.07348603010177612,
      "learning_rate": 0.000982463921423168,
      "loss": 1.5473,
      "step": 8522
    },
    {
      "epoch": 0.37899604286158906,
      "grad_norm": 0.07370968163013458,
      "learning_rate": 0.0009824546498941114,
      "loss": 1.5564,
      "step": 8524
    },
    {
      "epoch": 0.3790849673202614,
      "grad_norm": 0.07592252641916275,
      "learning_rate": 0.0009824453759584907,
      "loss": 1.5535,
      "step": 8526
    },
    {
      "epoch": 0.3791738917789338,
      "grad_norm": 0.07057470828294754,
      "learning_rate": 0.0009824360996163524,
      "loss": 1.5567,
      "step": 8528
    },
    {
      "epoch": 0.37926281623760616,
      "grad_norm": 0.07211954891681671,
      "learning_rate": 0.000982426820867743,
      "loss": 1.5548,
      "step": 8530
    },
    {
      "epoch": 0.3793517406962785,
      "grad_norm": 0.06937380880117416,
      "learning_rate": 0.0009824175397127083,
      "loss": 1.5457,
      "step": 8532
    },
    {
      "epoch": 0.3794406651549509,
      "grad_norm": 0.07182077318429947,
      "learning_rate": 0.0009824082561512948,
      "loss": 1.5589,
      "step": 8534
    },
    {
      "epoch": 0.3795295896136232,
      "grad_norm": 0.07062894105911255,
      "learning_rate": 0.000982398970183549,
      "loss": 1.5499,
      "step": 8536
    },
    {
      "epoch": 0.3796185140722956,
      "grad_norm": 0.07044274359941483,
      "learning_rate": 0.000982389681809517,
      "loss": 1.5555,
      "step": 8538
    },
    {
      "epoch": 0.37970743853096794,
      "grad_norm": 0.0717586874961853,
      "learning_rate": 0.0009823803910292452,
      "loss": 1.5452,
      "step": 8540
    },
    {
      "epoch": 0.3797963629896403,
      "grad_norm": 0.07334475219249725,
      "learning_rate": 0.0009823710978427798,
      "loss": 1.5576,
      "step": 8542
    },
    {
      "epoch": 0.3798852874483127,
      "grad_norm": 0.07173279672861099,
      "learning_rate": 0.0009823618022501675,
      "loss": 1.5451,
      "step": 8544
    },
    {
      "epoch": 0.379974211906985,
      "grad_norm": 0.07294796407222748,
      "learning_rate": 0.0009823525042514543,
      "loss": 1.5547,
      "step": 8546
    },
    {
      "epoch": 0.38006313636565736,
      "grad_norm": 0.0728803277015686,
      "learning_rate": 0.0009823432038466868,
      "loss": 1.5461,
      "step": 8548
    },
    {
      "epoch": 0.38015206082432973,
      "grad_norm": 0.070876345038414,
      "learning_rate": 0.0009823339010359114,
      "loss": 1.5525,
      "step": 8550
    },
    {
      "epoch": 0.3802409852830021,
      "grad_norm": 0.07284954190254211,
      "learning_rate": 0.0009823245958191745,
      "loss": 1.5486,
      "step": 8552
    },
    {
      "epoch": 0.38032990974167447,
      "grad_norm": 0.0750528946518898,
      "learning_rate": 0.0009823152881965223,
      "loss": 1.5507,
      "step": 8554
    },
    {
      "epoch": 0.3804188342003468,
      "grad_norm": 0.07205671817064285,
      "learning_rate": 0.0009823059781680014,
      "loss": 1.5513,
      "step": 8556
    },
    {
      "epoch": 0.38050775865901915,
      "grad_norm": 0.07429669797420502,
      "learning_rate": 0.0009822966657336582,
      "loss": 1.5503,
      "step": 8558
    },
    {
      "epoch": 0.3805966831176915,
      "grad_norm": 0.0736270323395729,
      "learning_rate": 0.0009822873508935394,
      "loss": 1.5513,
      "step": 8560
    },
    {
      "epoch": 0.3806856075763639,
      "grad_norm": 0.0691053494811058,
      "learning_rate": 0.0009822780336476908,
      "loss": 1.5482,
      "step": 8562
    },
    {
      "epoch": 0.38077453203503625,
      "grad_norm": 0.0720055103302002,
      "learning_rate": 0.0009822687139961596,
      "loss": 1.5469,
      "step": 8564
    },
    {
      "epoch": 0.3808634564937086,
      "grad_norm": 0.07141195237636566,
      "learning_rate": 0.0009822593919389922,
      "loss": 1.5539,
      "step": 8566
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 0.07182452082633972,
      "learning_rate": 0.0009822500674762345,
      "loss": 1.5476,
      "step": 8568
    },
    {
      "epoch": 0.3810413054110533,
      "grad_norm": 0.07438525557518005,
      "learning_rate": 0.0009822407406079336,
      "loss": 1.5486,
      "step": 8570
    },
    {
      "epoch": 0.38113022986972567,
      "grad_norm": 0.07334297895431519,
      "learning_rate": 0.0009822314113341359,
      "loss": 1.549,
      "step": 8572
    },
    {
      "epoch": 0.38121915432839804,
      "grad_norm": 0.07467876374721527,
      "learning_rate": 0.0009822220796548878,
      "loss": 1.5524,
      "step": 8574
    },
    {
      "epoch": 0.3813080787870704,
      "grad_norm": 0.0723448172211647,
      "learning_rate": 0.0009822127455702359,
      "loss": 1.5459,
      "step": 8576
    },
    {
      "epoch": 0.3813970032457427,
      "grad_norm": 0.07537683844566345,
      "learning_rate": 0.0009822034090802268,
      "loss": 1.5465,
      "step": 8578
    },
    {
      "epoch": 0.3814859277044151,
      "grad_norm": 0.07435756176710129,
      "learning_rate": 0.0009821940701849072,
      "loss": 1.5497,
      "step": 8580
    },
    {
      "epoch": 0.38157485216308745,
      "grad_norm": 0.07319629937410355,
      "learning_rate": 0.0009821847288843232,
      "loss": 1.5454,
      "step": 8582
    },
    {
      "epoch": 0.3816637766217598,
      "grad_norm": 0.07365921884775162,
      "learning_rate": 0.000982175385178522,
      "loss": 1.5457,
      "step": 8584
    },
    {
      "epoch": 0.3817527010804322,
      "grad_norm": 0.06942031532526016,
      "learning_rate": 0.00098216603906755,
      "loss": 1.5484,
      "step": 8586
    },
    {
      "epoch": 0.38184162553910456,
      "grad_norm": 0.07433439046144485,
      "learning_rate": 0.0009821566905514536,
      "loss": 1.5524,
      "step": 8588
    },
    {
      "epoch": 0.38193054999777687,
      "grad_norm": 0.07098887860774994,
      "learning_rate": 0.0009821473396302797,
      "loss": 1.5419,
      "step": 8590
    },
    {
      "epoch": 0.38201947445644924,
      "grad_norm": 0.07336100190877914,
      "learning_rate": 0.0009821379863040748,
      "loss": 1.5489,
      "step": 8592
    },
    {
      "epoch": 0.3821083989151216,
      "grad_norm": 0.07159991562366486,
      "learning_rate": 0.0009821286305728854,
      "loss": 1.5495,
      "step": 8594
    },
    {
      "epoch": 0.382197323373794,
      "grad_norm": 0.07283421605825424,
      "learning_rate": 0.0009821192724367585,
      "loss": 1.5474,
      "step": 8596
    },
    {
      "epoch": 0.38228624783246634,
      "grad_norm": 0.07494281232357025,
      "learning_rate": 0.0009821099118957407,
      "loss": 1.5481,
      "step": 8598
    },
    {
      "epoch": 0.38237517229113865,
      "grad_norm": 0.07011394947767258,
      "learning_rate": 0.0009821005489498787,
      "loss": 1.544,
      "step": 8600
    },
    {
      "epoch": 0.382464096749811,
      "grad_norm": 0.07218565046787262,
      "learning_rate": 0.000982091183599219,
      "loss": 1.5442,
      "step": 8602
    },
    {
      "epoch": 0.3825530212084834,
      "grad_norm": 0.07564288377761841,
      "learning_rate": 0.0009820818158438084,
      "loss": 1.5573,
      "step": 8604
    },
    {
      "epoch": 0.38264194566715576,
      "grad_norm": 0.07179772108793259,
      "learning_rate": 0.0009820724456836938,
      "loss": 1.5556,
      "step": 8606
    },
    {
      "epoch": 0.3827308701258281,
      "grad_norm": 0.07555480301380157,
      "learning_rate": 0.0009820630731189218,
      "loss": 1.5486,
      "step": 8608
    },
    {
      "epoch": 0.3828197945845005,
      "grad_norm": 0.07220586389303207,
      "learning_rate": 0.0009820536981495391,
      "loss": 1.5496,
      "step": 8610
    },
    {
      "epoch": 0.3829087190431728,
      "grad_norm": 0.07378540188074112,
      "learning_rate": 0.0009820443207755928,
      "loss": 1.5498,
      "step": 8612
    },
    {
      "epoch": 0.3829976435018452,
      "grad_norm": 0.07111547887325287,
      "learning_rate": 0.000982034940997129,
      "loss": 1.5494,
      "step": 8614
    },
    {
      "epoch": 0.38308656796051754,
      "grad_norm": 0.07289877533912659,
      "learning_rate": 0.000982025558814195,
      "loss": 1.5496,
      "step": 8616
    },
    {
      "epoch": 0.3831754924191899,
      "grad_norm": 0.07126215100288391,
      "learning_rate": 0.0009820161742268379,
      "loss": 1.5444,
      "step": 8618
    },
    {
      "epoch": 0.3832644168778623,
      "grad_norm": 0.07101909816265106,
      "learning_rate": 0.0009820067872351038,
      "loss": 1.5456,
      "step": 8620
    },
    {
      "epoch": 0.3833533413365346,
      "grad_norm": 0.0723530650138855,
      "learning_rate": 0.0009819973978390398,
      "loss": 1.5452,
      "step": 8622
    },
    {
      "epoch": 0.38344226579520696,
      "grad_norm": 0.07396049052476883,
      "learning_rate": 0.0009819880060386928,
      "loss": 1.5498,
      "step": 8624
    },
    {
      "epoch": 0.38353119025387933,
      "grad_norm": 0.07335137575864792,
      "learning_rate": 0.0009819786118341097,
      "loss": 1.554,
      "step": 8626
    },
    {
      "epoch": 0.3836201147125517,
      "grad_norm": 0.07116411626338959,
      "learning_rate": 0.0009819692152253372,
      "loss": 1.5461,
      "step": 8628
    },
    {
      "epoch": 0.38370903917122406,
      "grad_norm": 0.07290813326835632,
      "learning_rate": 0.0009819598162124224,
      "loss": 1.5431,
      "step": 8630
    },
    {
      "epoch": 0.3837979636298964,
      "grad_norm": 0.07139547169208527,
      "learning_rate": 0.0009819504147954119,
      "loss": 1.5439,
      "step": 8632
    },
    {
      "epoch": 0.38388688808856875,
      "grad_norm": 0.06995617598295212,
      "learning_rate": 0.0009819410109743528,
      "loss": 1.5455,
      "step": 8634
    },
    {
      "epoch": 0.3839758125472411,
      "grad_norm": 0.07103978097438812,
      "learning_rate": 0.0009819316047492918,
      "loss": 1.5439,
      "step": 8636
    },
    {
      "epoch": 0.3840647370059135,
      "grad_norm": 0.07140542566776276,
      "learning_rate": 0.000981922196120276,
      "loss": 1.5404,
      "step": 8638
    },
    {
      "epoch": 0.38415366146458585,
      "grad_norm": 0.0695871040225029,
      "learning_rate": 0.0009819127850873525,
      "loss": 1.5443,
      "step": 8640
    },
    {
      "epoch": 0.3842425859232582,
      "grad_norm": 0.07270847260951996,
      "learning_rate": 0.0009819033716505679,
      "loss": 1.5423,
      "step": 8642
    },
    {
      "epoch": 0.38433151038193053,
      "grad_norm": 0.07431674748659134,
      "learning_rate": 0.000981893955809969,
      "loss": 1.5434,
      "step": 8644
    },
    {
      "epoch": 0.3844204348406029,
      "grad_norm": 0.07491683959960938,
      "learning_rate": 0.0009818845375656035,
      "loss": 1.5443,
      "step": 8646
    },
    {
      "epoch": 0.38450935929927527,
      "grad_norm": 0.0724981278181076,
      "learning_rate": 0.0009818751169175177,
      "loss": 1.5455,
      "step": 8648
    },
    {
      "epoch": 0.38459828375794763,
      "grad_norm": 0.07452180981636047,
      "learning_rate": 0.0009818656938657589,
      "loss": 1.55,
      "step": 8650
    },
    {
      "epoch": 0.38468720821662,
      "grad_norm": 0.07273737341165543,
      "learning_rate": 0.000981856268410374,
      "loss": 1.5422,
      "step": 8652
    },
    {
      "epoch": 0.3847761326752923,
      "grad_norm": 0.07237744331359863,
      "learning_rate": 0.0009818468405514101,
      "loss": 1.5468,
      "step": 8654
    },
    {
      "epoch": 0.3848650571339647,
      "grad_norm": 0.07318416237831116,
      "learning_rate": 0.0009818374102889141,
      "loss": 1.5456,
      "step": 8656
    },
    {
      "epoch": 0.38495398159263705,
      "grad_norm": 0.0707939937710762,
      "learning_rate": 0.0009818279776229333,
      "loss": 1.5468,
      "step": 8658
    },
    {
      "epoch": 0.3850429060513094,
      "grad_norm": 0.07714451104402542,
      "learning_rate": 0.0009818185425535145,
      "loss": 1.55,
      "step": 8660
    },
    {
      "epoch": 0.3851318305099818,
      "grad_norm": 0.07308610528707504,
      "learning_rate": 0.0009818091050807047,
      "loss": 1.5422,
      "step": 8662
    },
    {
      "epoch": 0.38522075496865416,
      "grad_norm": 0.07573612779378891,
      "learning_rate": 0.0009817996652045512,
      "loss": 1.5528,
      "step": 8664
    },
    {
      "epoch": 0.38530967942732647,
      "grad_norm": 0.07352116703987122,
      "learning_rate": 0.0009817902229251009,
      "loss": 1.5453,
      "step": 8666
    },
    {
      "epoch": 0.38539860388599884,
      "grad_norm": 0.0721651017665863,
      "learning_rate": 0.000981780778242401,
      "loss": 1.5484,
      "step": 8668
    },
    {
      "epoch": 0.3854875283446712,
      "grad_norm": 0.07210839539766312,
      "learning_rate": 0.0009817713311564987,
      "loss": 1.5465,
      "step": 8670
    },
    {
      "epoch": 0.3855764528033436,
      "grad_norm": 0.07044050097465515,
      "learning_rate": 0.000981761881667441,
      "loss": 1.5464,
      "step": 8672
    },
    {
      "epoch": 0.38566537726201594,
      "grad_norm": 0.07243029773235321,
      "learning_rate": 0.000981752429775275,
      "loss": 1.5452,
      "step": 8674
    },
    {
      "epoch": 0.38575430172068825,
      "grad_norm": 0.07179372757673264,
      "learning_rate": 0.000981742975480048,
      "loss": 1.5457,
      "step": 8676
    },
    {
      "epoch": 0.3858432261793606,
      "grad_norm": 0.07337214797735214,
      "learning_rate": 0.000981733518781807,
      "loss": 1.5459,
      "step": 8678
    },
    {
      "epoch": 0.385932150638033,
      "grad_norm": 0.0745837390422821,
      "learning_rate": 0.0009817240596805994,
      "loss": 1.5494,
      "step": 8680
    },
    {
      "epoch": 0.38602107509670536,
      "grad_norm": 0.072391077876091,
      "learning_rate": 0.0009817145981764722,
      "loss": 1.5427,
      "step": 8682
    },
    {
      "epoch": 0.3861099995553777,
      "grad_norm": 0.07101153582334518,
      "learning_rate": 0.0009817051342694725,
      "loss": 1.5478,
      "step": 8684
    },
    {
      "epoch": 0.38619892401405004,
      "grad_norm": 0.07185351103544235,
      "learning_rate": 0.0009816956679596475,
      "loss": 1.5428,
      "step": 8686
    },
    {
      "epoch": 0.3862878484727224,
      "grad_norm": 0.07106965035200119,
      "learning_rate": 0.000981686199247045,
      "loss": 1.5515,
      "step": 8688
    },
    {
      "epoch": 0.3863767729313948,
      "grad_norm": 0.07175727933645248,
      "learning_rate": 0.0009816767281317113,
      "loss": 1.5422,
      "step": 8690
    },
    {
      "epoch": 0.38646569739006714,
      "grad_norm": 0.0721508041024208,
      "learning_rate": 0.0009816672546136944,
      "loss": 1.5444,
      "step": 8692
    },
    {
      "epoch": 0.3865546218487395,
      "grad_norm": 0.07354867458343506,
      "learning_rate": 0.0009816577786930414,
      "loss": 1.5476,
      "step": 8694
    },
    {
      "epoch": 0.3866435463074119,
      "grad_norm": 0.07081717252731323,
      "learning_rate": 0.000981648300369799,
      "loss": 1.5489,
      "step": 8696
    },
    {
      "epoch": 0.3867324707660842,
      "grad_norm": 0.07528056204319,
      "learning_rate": 0.0009816388196440154,
      "loss": 1.5507,
      "step": 8698
    },
    {
      "epoch": 0.38682139522475656,
      "grad_norm": 0.07244568318128586,
      "learning_rate": 0.000981629336515737,
      "loss": 1.5481,
      "step": 8700
    },
    {
      "epoch": 0.3869103196834289,
      "grad_norm": 0.07286922633647919,
      "learning_rate": 0.0009816198509850118,
      "loss": 1.5445,
      "step": 8702
    },
    {
      "epoch": 0.3869992441421013,
      "grad_norm": 0.06972247362136841,
      "learning_rate": 0.000981610363051887,
      "loss": 1.5486,
      "step": 8704
    },
    {
      "epoch": 0.38708816860077366,
      "grad_norm": 0.07090491056442261,
      "learning_rate": 0.0009816008727164093,
      "loss": 1.5433,
      "step": 8706
    },
    {
      "epoch": 0.387177093059446,
      "grad_norm": 0.0725286453962326,
      "learning_rate": 0.0009815913799786269,
      "loss": 1.5452,
      "step": 8708
    },
    {
      "epoch": 0.38726601751811834,
      "grad_norm": 0.07091175764799118,
      "learning_rate": 0.0009815818848385865,
      "loss": 1.5429,
      "step": 8710
    },
    {
      "epoch": 0.3873549419767907,
      "grad_norm": 0.07452746480703354,
      "learning_rate": 0.0009815723872963358,
      "loss": 1.5374,
      "step": 8712
    },
    {
      "epoch": 0.3874438664354631,
      "grad_norm": 0.07199020683765411,
      "learning_rate": 0.0009815628873519223,
      "loss": 1.5389,
      "step": 8714
    },
    {
      "epoch": 0.38753279089413545,
      "grad_norm": 0.07434893399477005,
      "learning_rate": 0.0009815533850053928,
      "loss": 1.5465,
      "step": 8716
    },
    {
      "epoch": 0.3876217153528078,
      "grad_norm": 0.0721900537610054,
      "learning_rate": 0.0009815438802567955,
      "loss": 1.5464,
      "step": 8718
    },
    {
      "epoch": 0.38771063981148013,
      "grad_norm": 0.0768774077296257,
      "learning_rate": 0.000981534373106177,
      "loss": 1.5477,
      "step": 8720
    },
    {
      "epoch": 0.3877995642701525,
      "grad_norm": 0.07451849430799484,
      "learning_rate": 0.0009815248635535853,
      "loss": 1.5439,
      "step": 8722
    },
    {
      "epoch": 0.38788848872882487,
      "grad_norm": 0.0778982937335968,
      "learning_rate": 0.0009815153515990679,
      "loss": 1.5425,
      "step": 8724
    },
    {
      "epoch": 0.38797741318749723,
      "grad_norm": 0.07124931365251541,
      "learning_rate": 0.0009815058372426716,
      "loss": 1.5497,
      "step": 8726
    },
    {
      "epoch": 0.3880663376461696,
      "grad_norm": 0.0718124508857727,
      "learning_rate": 0.0009814963204844444,
      "loss": 1.5436,
      "step": 8728
    },
    {
      "epoch": 0.3881552621048419,
      "grad_norm": 0.07384263724088669,
      "learning_rate": 0.0009814868013244336,
      "loss": 1.5444,
      "step": 8730
    },
    {
      "epoch": 0.3882441865635143,
      "grad_norm": 0.07269671559333801,
      "learning_rate": 0.0009814772797626867,
      "loss": 1.5482,
      "step": 8732
    },
    {
      "epoch": 0.38833311102218665,
      "grad_norm": 0.0724627822637558,
      "learning_rate": 0.0009814677557992513,
      "loss": 1.5495,
      "step": 8734
    },
    {
      "epoch": 0.388422035480859,
      "grad_norm": 0.07544320821762085,
      "learning_rate": 0.0009814582294341747,
      "loss": 1.5459,
      "step": 8736
    },
    {
      "epoch": 0.3885109599395314,
      "grad_norm": 0.0732409730553627,
      "learning_rate": 0.0009814487006675047,
      "loss": 1.543,
      "step": 8738
    },
    {
      "epoch": 0.3885998843982037,
      "grad_norm": 0.071528859436512,
      "learning_rate": 0.0009814391694992885,
      "loss": 1.5489,
      "step": 8740
    },
    {
      "epoch": 0.38868880885687607,
      "grad_norm": 0.07471916079521179,
      "learning_rate": 0.000981429635929574,
      "loss": 1.5427,
      "step": 8742
    },
    {
      "epoch": 0.38877773331554843,
      "grad_norm": 0.07275721430778503,
      "learning_rate": 0.0009814200999584085,
      "loss": 1.5449,
      "step": 8744
    },
    {
      "epoch": 0.3888666577742208,
      "grad_norm": 0.07258014380931854,
      "learning_rate": 0.0009814105615858395,
      "loss": 1.543,
      "step": 8746
    },
    {
      "epoch": 0.38895558223289317,
      "grad_norm": 0.07144364714622498,
      "learning_rate": 0.0009814010208119147,
      "loss": 1.5451,
      "step": 8748
    },
    {
      "epoch": 0.38904450669156554,
      "grad_norm": 0.07227019965648651,
      "learning_rate": 0.0009813914776366816,
      "loss": 1.5423,
      "step": 8750
    },
    {
      "epoch": 0.38913343115023785,
      "grad_norm": 0.0736045092344284,
      "learning_rate": 0.0009813819320601883,
      "loss": 1.5435,
      "step": 8752
    },
    {
      "epoch": 0.3892223556089102,
      "grad_norm": 0.07181594520807266,
      "learning_rate": 0.0009813723840824819,
      "loss": 1.5413,
      "step": 8754
    },
    {
      "epoch": 0.3893112800675826,
      "grad_norm": 0.07333064079284668,
      "learning_rate": 0.0009813628337036098,
      "loss": 1.5514,
      "step": 8756
    },
    {
      "epoch": 0.38940020452625496,
      "grad_norm": 0.07221698760986328,
      "learning_rate": 0.0009813532809236202,
      "loss": 1.5435,
      "step": 8758
    },
    {
      "epoch": 0.3894891289849273,
      "grad_norm": 0.07169820368289948,
      "learning_rate": 0.0009813437257425606,
      "loss": 1.5474,
      "step": 8760
    },
    {
      "epoch": 0.38957805344359964,
      "grad_norm": 0.07131782174110413,
      "learning_rate": 0.0009813341681604785,
      "loss": 1.5431,
      "step": 8762
    },
    {
      "epoch": 0.389666977902272,
      "grad_norm": 0.07381466031074524,
      "learning_rate": 0.0009813246081774218,
      "loss": 1.551,
      "step": 8764
    },
    {
      "epoch": 0.3897559023609444,
      "grad_norm": 0.07252717018127441,
      "learning_rate": 0.000981315045793438,
      "loss": 1.542,
      "step": 8766
    },
    {
      "epoch": 0.38984482681961674,
      "grad_norm": 0.07131364196538925,
      "learning_rate": 0.0009813054810085748,
      "loss": 1.5416,
      "step": 8768
    },
    {
      "epoch": 0.3899337512782891,
      "grad_norm": 0.07210860401391983,
      "learning_rate": 0.00098129591382288,
      "loss": 1.5414,
      "step": 8770
    },
    {
      "epoch": 0.3900226757369615,
      "grad_norm": 0.07035349309444427,
      "learning_rate": 0.0009812863442364014,
      "loss": 1.5424,
      "step": 8772
    },
    {
      "epoch": 0.3901116001956338,
      "grad_norm": 0.07026588171720505,
      "learning_rate": 0.0009812767722491864,
      "loss": 1.5484,
      "step": 8774
    },
    {
      "epoch": 0.39020052465430616,
      "grad_norm": 0.07101976126432419,
      "learning_rate": 0.000981267197861283,
      "loss": 1.5419,
      "step": 8776
    },
    {
      "epoch": 0.3902894491129785,
      "grad_norm": 0.07162394374608994,
      "learning_rate": 0.0009812576210727392,
      "loss": 1.5482,
      "step": 8778
    },
    {
      "epoch": 0.3903783735716509,
      "grad_norm": 0.07118743658065796,
      "learning_rate": 0.0009812480418836024,
      "loss": 1.5401,
      "step": 8780
    },
    {
      "epoch": 0.39046729803032326,
      "grad_norm": 0.0704973116517067,
      "learning_rate": 0.0009812384602939203,
      "loss": 1.5413,
      "step": 8782
    },
    {
      "epoch": 0.3905562224889956,
      "grad_norm": 0.07269991189241409,
      "learning_rate": 0.000981228876303741,
      "loss": 1.5438,
      "step": 8784
    },
    {
      "epoch": 0.39064514694766794,
      "grad_norm": 0.07165592908859253,
      "learning_rate": 0.0009812192899131122,
      "loss": 1.548,
      "step": 8786
    },
    {
      "epoch": 0.3907340714063403,
      "grad_norm": 0.07121030986309052,
      "learning_rate": 0.0009812097011220816,
      "loss": 1.5459,
      "step": 8788
    },
    {
      "epoch": 0.3908229958650127,
      "grad_norm": 0.07146286964416504,
      "learning_rate": 0.0009812001099306974,
      "loss": 1.539,
      "step": 8790
    },
    {
      "epoch": 0.39091192032368505,
      "grad_norm": 0.07126247137784958,
      "learning_rate": 0.0009811905163390068,
      "loss": 1.5442,
      "step": 8792
    },
    {
      "epoch": 0.3910008447823574,
      "grad_norm": 0.06753192096948624,
      "learning_rate": 0.0009811809203470582,
      "loss": 1.546,
      "step": 8794
    },
    {
      "epoch": 0.3910897692410297,
      "grad_norm": 0.0699109360575676,
      "learning_rate": 0.0009811713219548992,
      "loss": 1.542,
      "step": 8796
    },
    {
      "epoch": 0.3911786936997021,
      "grad_norm": 0.0715608224272728,
      "learning_rate": 0.0009811617211625781,
      "loss": 1.5429,
      "step": 8798
    },
    {
      "epoch": 0.39126761815837446,
      "grad_norm": 0.07075057178735733,
      "learning_rate": 0.0009811521179701422,
      "loss": 1.5433,
      "step": 8800
    },
    {
      "epoch": 0.39135654261704683,
      "grad_norm": 0.07018981128931046,
      "learning_rate": 0.0009811425123776398,
      "loss": 1.5432,
      "step": 8802
    },
    {
      "epoch": 0.3914454670757192,
      "grad_norm": 0.07095321267843246,
      "learning_rate": 0.0009811329043851185,
      "loss": 1.5474,
      "step": 8804
    },
    {
      "epoch": 0.3915343915343915,
      "grad_norm": 0.07263996452093124,
      "learning_rate": 0.0009811232939926267,
      "loss": 1.5394,
      "step": 8806
    },
    {
      "epoch": 0.3916233159930639,
      "grad_norm": 0.07189413160085678,
      "learning_rate": 0.0009811136812002117,
      "loss": 1.5441,
      "step": 8808
    },
    {
      "epoch": 0.39171224045173625,
      "grad_norm": 0.0719655454158783,
      "learning_rate": 0.000981104066007922,
      "loss": 1.545,
      "step": 8810
    },
    {
      "epoch": 0.3918011649104086,
      "grad_norm": 0.0707225650548935,
      "learning_rate": 0.0009810944484158052,
      "loss": 1.5435,
      "step": 8812
    },
    {
      "epoch": 0.391890089369081,
      "grad_norm": 0.07110429555177689,
      "learning_rate": 0.0009810848284239097,
      "loss": 1.5402,
      "step": 8814
    },
    {
      "epoch": 0.3919790138277533,
      "grad_norm": 0.07303628325462341,
      "learning_rate": 0.000981075206032283,
      "loss": 1.5408,
      "step": 8816
    },
    {
      "epoch": 0.39206793828642567,
      "grad_norm": 0.07266764342784882,
      "learning_rate": 0.0009810655812409735,
      "loss": 1.5399,
      "step": 8818
    },
    {
      "epoch": 0.39215686274509803,
      "grad_norm": 0.07289233058691025,
      "learning_rate": 0.0009810559540500289,
      "loss": 1.5437,
      "step": 8820
    },
    {
      "epoch": 0.3922457872037704,
      "grad_norm": 0.07148449867963791,
      "learning_rate": 0.0009810463244594974,
      "loss": 1.5445,
      "step": 8822
    },
    {
      "epoch": 0.39233471166244277,
      "grad_norm": 0.0700794905424118,
      "learning_rate": 0.0009810366924694269,
      "loss": 1.5361,
      "step": 8824
    },
    {
      "epoch": 0.39242363612111514,
      "grad_norm": 0.07210569828748703,
      "learning_rate": 0.0009810270580798656,
      "loss": 1.5488,
      "step": 8826
    },
    {
      "epoch": 0.39251256057978745,
      "grad_norm": 0.07074844092130661,
      "learning_rate": 0.0009810174212908614,
      "loss": 1.5454,
      "step": 8828
    },
    {
      "epoch": 0.3926014850384598,
      "grad_norm": 0.07008279860019684,
      "learning_rate": 0.0009810077821024625,
      "loss": 1.5463,
      "step": 8830
    },
    {
      "epoch": 0.3926904094971322,
      "grad_norm": 0.07124333083629608,
      "learning_rate": 0.0009809981405147171,
      "loss": 1.5415,
      "step": 8832
    },
    {
      "epoch": 0.39277933395580455,
      "grad_norm": 0.06977679580450058,
      "learning_rate": 0.000980988496527673,
      "loss": 1.5459,
      "step": 8834
    },
    {
      "epoch": 0.3928682584144769,
      "grad_norm": 0.0721663162112236,
      "learning_rate": 0.0009809788501413783,
      "loss": 1.5433,
      "step": 8836
    },
    {
      "epoch": 0.39295718287314924,
      "grad_norm": 0.07385817915201187,
      "learning_rate": 0.0009809692013558816,
      "loss": 1.545,
      "step": 8838
    },
    {
      "epoch": 0.3930461073318216,
      "grad_norm": 0.07491130381822586,
      "learning_rate": 0.0009809595501712302,
      "loss": 1.5419,
      "step": 8840
    },
    {
      "epoch": 0.39313503179049397,
      "grad_norm": 0.07294177263975143,
      "learning_rate": 0.0009809498965874732,
      "loss": 1.5461,
      "step": 8842
    },
    {
      "epoch": 0.39322395624916634,
      "grad_norm": 0.07299856841564178,
      "learning_rate": 0.000980940240604658,
      "loss": 1.5455,
      "step": 8844
    },
    {
      "epoch": 0.3933128807078387,
      "grad_norm": 0.07228664308786392,
      "learning_rate": 0.0009809305822228332,
      "loss": 1.5422,
      "step": 8846
    },
    {
      "epoch": 0.3934018051665111,
      "grad_norm": 0.07752872258424759,
      "learning_rate": 0.0009809209214420467,
      "loss": 1.5404,
      "step": 8848
    },
    {
      "epoch": 0.3934907296251834,
      "grad_norm": 0.07412002980709076,
      "learning_rate": 0.000980911258262347,
      "loss": 1.5446,
      "step": 8850
    },
    {
      "epoch": 0.39357965408385576,
      "grad_norm": 0.06998312473297119,
      "learning_rate": 0.0009809015926837818,
      "loss": 1.5454,
      "step": 8852
    },
    {
      "epoch": 0.3936685785425281,
      "grad_norm": 0.0724993646144867,
      "learning_rate": 0.0009808919247064,
      "loss": 1.5369,
      "step": 8854
    },
    {
      "epoch": 0.3937575030012005,
      "grad_norm": 0.0712260976433754,
      "learning_rate": 0.0009808822543302493,
      "loss": 1.544,
      "step": 8856
    },
    {
      "epoch": 0.39384642745987286,
      "grad_norm": 0.07351220399141312,
      "learning_rate": 0.0009808725815553781,
      "loss": 1.5481,
      "step": 8858
    },
    {
      "epoch": 0.3939353519185452,
      "grad_norm": 0.07386364787817001,
      "learning_rate": 0.0009808629063818347,
      "loss": 1.5355,
      "step": 8860
    },
    {
      "epoch": 0.39402427637721754,
      "grad_norm": 0.0700744241476059,
      "learning_rate": 0.000980853228809667,
      "loss": 1.5344,
      "step": 8862
    },
    {
      "epoch": 0.3941132008358899,
      "grad_norm": 0.06920643895864487,
      "learning_rate": 0.0009808435488389239,
      "loss": 1.5445,
      "step": 8864
    },
    {
      "epoch": 0.3942021252945623,
      "grad_norm": 0.07501189410686493,
      "learning_rate": 0.0009808338664696532,
      "loss": 1.545,
      "step": 8866
    },
    {
      "epoch": 0.39429104975323465,
      "grad_norm": 0.07265200465917587,
      "learning_rate": 0.0009808241817019035,
      "loss": 1.5478,
      "step": 8868
    },
    {
      "epoch": 0.39437997421190696,
      "grad_norm": 0.07012397050857544,
      "learning_rate": 0.000980814494535723,
      "loss": 1.5415,
      "step": 8870
    },
    {
      "epoch": 0.3944688986705793,
      "grad_norm": 0.07346207648515701,
      "learning_rate": 0.0009808048049711597,
      "loss": 1.5456,
      "step": 8872
    },
    {
      "epoch": 0.3945578231292517,
      "grad_norm": 0.06875230371952057,
      "learning_rate": 0.0009807951130082625,
      "loss": 1.5495,
      "step": 8874
    },
    {
      "epoch": 0.39464674758792406,
      "grad_norm": 0.07028649747371674,
      "learning_rate": 0.0009807854186470793,
      "loss": 1.5382,
      "step": 8876
    },
    {
      "epoch": 0.39473567204659643,
      "grad_norm": 0.07126179337501526,
      "learning_rate": 0.0009807757218876587,
      "loss": 1.5412,
      "step": 8878
    },
    {
      "epoch": 0.3948245965052688,
      "grad_norm": 0.07412000000476837,
      "learning_rate": 0.000980766022730049,
      "loss": 1.5441,
      "step": 8880
    },
    {
      "epoch": 0.3949135209639411,
      "grad_norm": 0.06931138783693314,
      "learning_rate": 0.0009807563211742987,
      "loss": 1.5408,
      "step": 8882
    },
    {
      "epoch": 0.3950024454226135,
      "grad_norm": 0.06960853189229965,
      "learning_rate": 0.0009807466172204557,
      "loss": 1.5467,
      "step": 8884
    },
    {
      "epoch": 0.39509136988128585,
      "grad_norm": 0.07029549777507782,
      "learning_rate": 0.000980736910868569,
      "loss": 1.5485,
      "step": 8886
    },
    {
      "epoch": 0.3951802943399582,
      "grad_norm": 0.070205919444561,
      "learning_rate": 0.0009807272021186868,
      "loss": 1.544,
      "step": 8888
    },
    {
      "epoch": 0.3952692187986306,
      "grad_norm": 0.07060658931732178,
      "learning_rate": 0.0009807174909708576,
      "loss": 1.542,
      "step": 8890
    },
    {
      "epoch": 0.3953581432573029,
      "grad_norm": 0.070548877120018,
      "learning_rate": 0.0009807077774251296,
      "loss": 1.5451,
      "step": 8892
    },
    {
      "epoch": 0.39544706771597526,
      "grad_norm": 0.06973063200712204,
      "learning_rate": 0.0009806980614815515,
      "loss": 1.5422,
      "step": 8894
    },
    {
      "epoch": 0.39553599217464763,
      "grad_norm": 0.07283268868923187,
      "learning_rate": 0.0009806883431401718,
      "loss": 1.5433,
      "step": 8896
    },
    {
      "epoch": 0.39562491663332,
      "grad_norm": 0.07797042280435562,
      "learning_rate": 0.0009806786224010387,
      "loss": 1.549,
      "step": 8898
    },
    {
      "epoch": 0.39571384109199237,
      "grad_norm": 0.07383356988430023,
      "learning_rate": 0.000980668899264201,
      "loss": 1.5419,
      "step": 8900
    },
    {
      "epoch": 0.39580276555066474,
      "grad_norm": 0.08115527033805847,
      "learning_rate": 0.0009806591737297069,
      "loss": 1.5418,
      "step": 8902
    },
    {
      "epoch": 0.39589169000933705,
      "grad_norm": 0.0753684863448143,
      "learning_rate": 0.0009806494457976049,
      "loss": 1.5422,
      "step": 8904
    },
    {
      "epoch": 0.3959806144680094,
      "grad_norm": 0.07579310983419418,
      "learning_rate": 0.0009806397154679437,
      "loss": 1.5431,
      "step": 8906
    },
    {
      "epoch": 0.3960695389266818,
      "grad_norm": 0.0755925178527832,
      "learning_rate": 0.0009806299827407722,
      "loss": 1.5424,
      "step": 8908
    },
    {
      "epoch": 0.39615846338535415,
      "grad_norm": 0.07460242509841919,
      "learning_rate": 0.0009806202476161383,
      "loss": 1.5464,
      "step": 8910
    },
    {
      "epoch": 0.3962473878440265,
      "grad_norm": 0.07479722797870636,
      "learning_rate": 0.000980610510094091,
      "loss": 1.5461,
      "step": 8912
    },
    {
      "epoch": 0.39633631230269883,
      "grad_norm": 0.0737161934375763,
      "learning_rate": 0.0009806007701746785,
      "loss": 1.5415,
      "step": 8914
    },
    {
      "epoch": 0.3964252367613712,
      "grad_norm": 0.07179709523916245,
      "learning_rate": 0.0009805910278579495,
      "loss": 1.5388,
      "step": 8916
    },
    {
      "epoch": 0.39651416122004357,
      "grad_norm": 0.07436472177505493,
      "learning_rate": 0.0009805812831439528,
      "loss": 1.5408,
      "step": 8918
    },
    {
      "epoch": 0.39660308567871594,
      "grad_norm": 0.07385652512311935,
      "learning_rate": 0.000980571536032737,
      "loss": 1.5375,
      "step": 8920
    },
    {
      "epoch": 0.3966920101373883,
      "grad_norm": 0.07151640951633453,
      "learning_rate": 0.0009805617865243504,
      "loss": 1.5407,
      "step": 8922
    },
    {
      "epoch": 0.3967809345960606,
      "grad_norm": 0.07092677056789398,
      "learning_rate": 0.000980552034618842,
      "loss": 1.5415,
      "step": 8924
    },
    {
      "epoch": 0.396869859054733,
      "grad_norm": 0.0716884657740593,
      "learning_rate": 0.0009805422803162603,
      "loss": 1.5341,
      "step": 8926
    },
    {
      "epoch": 0.39695878351340536,
      "grad_norm": 0.07016546279191971,
      "learning_rate": 0.0009805325236166538,
      "loss": 1.5385,
      "step": 8928
    },
    {
      "epoch": 0.3970477079720777,
      "grad_norm": 0.07406767457723618,
      "learning_rate": 0.0009805227645200713,
      "loss": 1.5411,
      "step": 8930
    },
    {
      "epoch": 0.3971366324307501,
      "grad_norm": 0.0695648342370987,
      "learning_rate": 0.0009805130030265616,
      "loss": 1.5376,
      "step": 8932
    },
    {
      "epoch": 0.39722555688942246,
      "grad_norm": 0.07370419800281525,
      "learning_rate": 0.0009805032391361733,
      "loss": 1.5464,
      "step": 8934
    },
    {
      "epoch": 0.3973144813480948,
      "grad_norm": 0.07572710514068604,
      "learning_rate": 0.0009804934728489551,
      "loss": 1.5397,
      "step": 8936
    },
    {
      "epoch": 0.39740340580676714,
      "grad_norm": 0.07438923418521881,
      "learning_rate": 0.0009804837041649556,
      "loss": 1.5395,
      "step": 8938
    },
    {
      "epoch": 0.3974923302654395,
      "grad_norm": 0.07321964204311371,
      "learning_rate": 0.0009804739330842235,
      "loss": 1.5405,
      "step": 8940
    },
    {
      "epoch": 0.3975812547241119,
      "grad_norm": 0.07280636578798294,
      "learning_rate": 0.000980464159606808,
      "loss": 1.5442,
      "step": 8942
    },
    {
      "epoch": 0.39767017918278424,
      "grad_norm": 0.07473626732826233,
      "learning_rate": 0.0009804543837327573,
      "loss": 1.5427,
      "step": 8944
    },
    {
      "epoch": 0.39775910364145656,
      "grad_norm": 0.06971806287765503,
      "learning_rate": 0.0009804446054621203,
      "loss": 1.5431,
      "step": 8946
    },
    {
      "epoch": 0.3978480281001289,
      "grad_norm": 0.07413081079721451,
      "learning_rate": 0.0009804348247949462,
      "loss": 1.5433,
      "step": 8948
    },
    {
      "epoch": 0.3979369525588013,
      "grad_norm": 0.07259391993284225,
      "learning_rate": 0.0009804250417312832,
      "loss": 1.5446,
      "step": 8950
    },
    {
      "epoch": 0.39802587701747366,
      "grad_norm": 0.0727652907371521,
      "learning_rate": 0.0009804152562711804,
      "loss": 1.5387,
      "step": 8952
    },
    {
      "epoch": 0.39811480147614603,
      "grad_norm": 0.07275709509849548,
      "learning_rate": 0.0009804054684146865,
      "loss": 1.5379,
      "step": 8954
    },
    {
      "epoch": 0.3982037259348184,
      "grad_norm": 0.07379709184169769,
      "learning_rate": 0.0009803956781618505,
      "loss": 1.5357,
      "step": 8956
    },
    {
      "epoch": 0.3982926503934907,
      "grad_norm": 0.06894747912883759,
      "learning_rate": 0.000980385885512721,
      "loss": 1.544,
      "step": 8958
    },
    {
      "epoch": 0.3983815748521631,
      "grad_norm": 0.06997144967317581,
      "learning_rate": 0.000980376090467347,
      "loss": 1.5365,
      "step": 8960
    },
    {
      "epoch": 0.39847049931083545,
      "grad_norm": 0.07101274281740189,
      "learning_rate": 0.0009803662930257773,
      "loss": 1.5355,
      "step": 8962
    },
    {
      "epoch": 0.3985594237695078,
      "grad_norm": 0.07142171263694763,
      "learning_rate": 0.0009803564931880606,
      "loss": 1.5441,
      "step": 8964
    },
    {
      "epoch": 0.3986483482281802,
      "grad_norm": 0.06851017475128174,
      "learning_rate": 0.0009803466909542463,
      "loss": 1.5381,
      "step": 8966
    },
    {
      "epoch": 0.3987372726868525,
      "grad_norm": 0.06920833140611649,
      "learning_rate": 0.0009803368863243829,
      "loss": 1.5397,
      "step": 8968
    },
    {
      "epoch": 0.39882619714552486,
      "grad_norm": 0.07191641628742218,
      "learning_rate": 0.0009803270792985194,
      "loss": 1.541,
      "step": 8970
    },
    {
      "epoch": 0.39891512160419723,
      "grad_norm": 0.07115970551967621,
      "learning_rate": 0.0009803172698767046,
      "loss": 1.5363,
      "step": 8972
    },
    {
      "epoch": 0.3990040460628696,
      "grad_norm": 0.06930098682641983,
      "learning_rate": 0.0009803074580589874,
      "loss": 1.5372,
      "step": 8974
    },
    {
      "epoch": 0.39909297052154197,
      "grad_norm": 0.06866835802793503,
      "learning_rate": 0.0009802976438454173,
      "loss": 1.5429,
      "step": 8976
    },
    {
      "epoch": 0.39918189498021434,
      "grad_norm": 0.06998659670352936,
      "learning_rate": 0.0009802878272360423,
      "loss": 1.537,
      "step": 8978
    },
    {
      "epoch": 0.39927081943888665,
      "grad_norm": 0.07126189023256302,
      "learning_rate": 0.0009802780082309123,
      "loss": 1.5386,
      "step": 8980
    },
    {
      "epoch": 0.399359743897559,
      "grad_norm": 0.0688776820898056,
      "learning_rate": 0.0009802681868300758,
      "loss": 1.5404,
      "step": 8982
    },
    {
      "epoch": 0.3994486683562314,
      "grad_norm": 0.07415543496608734,
      "learning_rate": 0.0009802583630335818,
      "loss": 1.5461,
      "step": 8984
    },
    {
      "epoch": 0.39953759281490375,
      "grad_norm": 0.06990732252597809,
      "learning_rate": 0.0009802485368414792,
      "loss": 1.54,
      "step": 8986
    },
    {
      "epoch": 0.3996265172735761,
      "grad_norm": 0.06985145807266235,
      "learning_rate": 0.0009802387082538174,
      "loss": 1.5392,
      "step": 8988
    },
    {
      "epoch": 0.39971544173224843,
      "grad_norm": 0.07134158909320831,
      "learning_rate": 0.000980228877270645,
      "loss": 1.5432,
      "step": 8990
    },
    {
      "epoch": 0.3998043661909208,
      "grad_norm": 0.07347475737333298,
      "learning_rate": 0.0009802190438920116,
      "loss": 1.5434,
      "step": 8992
    },
    {
      "epoch": 0.39989329064959317,
      "grad_norm": 0.07028983533382416,
      "learning_rate": 0.0009802092081179657,
      "loss": 1.5385,
      "step": 8994
    },
    {
      "epoch": 0.39998221510826554,
      "grad_norm": 0.07134208083152771,
      "learning_rate": 0.0009801993699485566,
      "loss": 1.5446,
      "step": 8996
    },
    {
      "epoch": 0.4000711395669379,
      "grad_norm": 0.07063104212284088,
      "learning_rate": 0.000980189529383833,
      "loss": 1.5452,
      "step": 8998
    },
    {
      "epoch": 0.4001600640256102,
      "grad_norm": 0.06996823847293854,
      "learning_rate": 0.0009801796864238447,
      "loss": 1.5371,
      "step": 9000
    },
    {
      "epoch": 0.4001600640256102,
      "eval_loss": 1.51895010471344,
      "eval_runtime": 12.3629,
      "eval_samples_per_second": 558.929,
      "eval_steps_per_second": 69.886,
      "step": 9000
    },
    {
      "epoch": 0.4002489884842826,
      "grad_norm": 0.07135413587093353,
      "learning_rate": 0.0009801698410686404,
      "loss": 1.5427,
      "step": 9002
    },
    {
      "epoch": 0.40033791294295495,
      "grad_norm": 0.07339084148406982,
      "learning_rate": 0.0009801599933182692,
      "loss": 1.5352,
      "step": 9004
    },
    {
      "epoch": 0.4004268374016273,
      "grad_norm": 0.06984297931194305,
      "learning_rate": 0.00098015014317278,
      "loss": 1.5358,
      "step": 9006
    },
    {
      "epoch": 0.4005157618602997,
      "grad_norm": 0.07473129779100418,
      "learning_rate": 0.0009801402906322225,
      "loss": 1.536,
      "step": 9008
    },
    {
      "epoch": 0.40060468631897206,
      "grad_norm": 0.07112877070903778,
      "learning_rate": 0.0009801304356966455,
      "loss": 1.5474,
      "step": 9010
    },
    {
      "epoch": 0.40069361077764437,
      "grad_norm": 0.07204887270927429,
      "learning_rate": 0.000980120578366098,
      "loss": 1.5418,
      "step": 9012
    },
    {
      "epoch": 0.40078253523631674,
      "grad_norm": 0.06817831844091415,
      "learning_rate": 0.0009801107186406296,
      "loss": 1.531,
      "step": 9014
    },
    {
      "epoch": 0.4008714596949891,
      "grad_norm": 0.07230226695537567,
      "learning_rate": 0.000980100856520289,
      "loss": 1.5388,
      "step": 9016
    },
    {
      "epoch": 0.4009603841536615,
      "grad_norm": 0.07334619015455246,
      "learning_rate": 0.0009800909920051257,
      "loss": 1.5435,
      "step": 9018
    },
    {
      "epoch": 0.40104930861233384,
      "grad_norm": 0.06880249083042145,
      "learning_rate": 0.0009800811250951888,
      "loss": 1.5336,
      "step": 9020
    },
    {
      "epoch": 0.40113823307100616,
      "grad_norm": 0.07309377938508987,
      "learning_rate": 0.0009800712557905278,
      "loss": 1.5394,
      "step": 9022
    },
    {
      "epoch": 0.4012271575296785,
      "grad_norm": 0.07203828543424606,
      "learning_rate": 0.0009800613840911913,
      "loss": 1.5421,
      "step": 9024
    },
    {
      "epoch": 0.4013160819883509,
      "grad_norm": 0.07213805615901947,
      "learning_rate": 0.0009800515099972294,
      "loss": 1.5456,
      "step": 9026
    },
    {
      "epoch": 0.40140500644702326,
      "grad_norm": 0.07065927237272263,
      "learning_rate": 0.0009800416335086907,
      "loss": 1.5389,
      "step": 9028
    },
    {
      "epoch": 0.40149393090569563,
      "grad_norm": 0.0714295282959938,
      "learning_rate": 0.0009800317546256245,
      "loss": 1.5386,
      "step": 9030
    },
    {
      "epoch": 0.401582855364368,
      "grad_norm": 0.07336141169071198,
      "learning_rate": 0.0009800218733480802,
      "loss": 1.54,
      "step": 9032
    },
    {
      "epoch": 0.4016717798230403,
      "grad_norm": 0.0692136138677597,
      "learning_rate": 0.0009800119896761074,
      "loss": 1.5393,
      "step": 9034
    },
    {
      "epoch": 0.4017607042817127,
      "grad_norm": 0.07176699489355087,
      "learning_rate": 0.0009800021036097549,
      "loss": 1.5424,
      "step": 9036
    },
    {
      "epoch": 0.40184962874038505,
      "grad_norm": 0.06777803599834442,
      "learning_rate": 0.0009799922151490722,
      "loss": 1.5379,
      "step": 9038
    },
    {
      "epoch": 0.4019385531990574,
      "grad_norm": 0.07473397254943848,
      "learning_rate": 0.0009799823242941088,
      "loss": 1.54,
      "step": 9040
    },
    {
      "epoch": 0.4020274776577298,
      "grad_norm": 0.07176635414361954,
      "learning_rate": 0.000979972431044914,
      "loss": 1.5363,
      "step": 9042
    },
    {
      "epoch": 0.4021164021164021,
      "grad_norm": 0.0704076886177063,
      "learning_rate": 0.0009799625354015367,
      "loss": 1.5381,
      "step": 9044
    },
    {
      "epoch": 0.40220532657507446,
      "grad_norm": 0.07321424782276154,
      "learning_rate": 0.0009799526373640267,
      "loss": 1.5393,
      "step": 9046
    },
    {
      "epoch": 0.40229425103374683,
      "grad_norm": 0.0706743597984314,
      "learning_rate": 0.0009799427369324334,
      "loss": 1.5404,
      "step": 9048
    },
    {
      "epoch": 0.4023831754924192,
      "grad_norm": 0.07315555214881897,
      "learning_rate": 0.000979932834106806,
      "loss": 1.5373,
      "step": 9050
    },
    {
      "epoch": 0.40247209995109157,
      "grad_norm": 0.07187056541442871,
      "learning_rate": 0.000979922928887194,
      "loss": 1.5369,
      "step": 9052
    },
    {
      "epoch": 0.4025610244097639,
      "grad_norm": 0.06762688606977463,
      "learning_rate": 0.0009799130212736467,
      "loss": 1.5424,
      "step": 9054
    },
    {
      "epoch": 0.40264994886843625,
      "grad_norm": 0.07067506015300751,
      "learning_rate": 0.0009799031112662138,
      "loss": 1.5421,
      "step": 9056
    },
    {
      "epoch": 0.4027388733271086,
      "grad_norm": 0.07152686268091202,
      "learning_rate": 0.0009798931988649442,
      "loss": 1.5434,
      "step": 9058
    },
    {
      "epoch": 0.402827797785781,
      "grad_norm": 0.07224275916814804,
      "learning_rate": 0.0009798832840698878,
      "loss": 1.5396,
      "step": 9060
    },
    {
      "epoch": 0.40291672224445335,
      "grad_norm": 0.06946686655282974,
      "learning_rate": 0.000979873366881094,
      "loss": 1.5462,
      "step": 9062
    },
    {
      "epoch": 0.4030056467031257,
      "grad_norm": 0.07184547185897827,
      "learning_rate": 0.000979863447298612,
      "loss": 1.5358,
      "step": 9064
    },
    {
      "epoch": 0.40309457116179803,
      "grad_norm": 0.06754238903522491,
      "learning_rate": 0.0009798535253224916,
      "loss": 1.5369,
      "step": 9066
    },
    {
      "epoch": 0.4031834956204704,
      "grad_norm": 0.06945636868476868,
      "learning_rate": 0.0009798436009527823,
      "loss": 1.5443,
      "step": 9068
    },
    {
      "epoch": 0.40327242007914277,
      "grad_norm": 0.07276376336812973,
      "learning_rate": 0.0009798336741895332,
      "loss": 1.5391,
      "step": 9070
    },
    {
      "epoch": 0.40336134453781514,
      "grad_norm": 0.07267561554908752,
      "learning_rate": 0.0009798237450327942,
      "loss": 1.5445,
      "step": 9072
    },
    {
      "epoch": 0.4034502689964875,
      "grad_norm": 0.07162649929523468,
      "learning_rate": 0.000979813813482615,
      "loss": 1.5416,
      "step": 9074
    },
    {
      "epoch": 0.4035391934551598,
      "grad_norm": 0.06982056051492691,
      "learning_rate": 0.0009798038795390445,
      "loss": 1.5402,
      "step": 9076
    },
    {
      "epoch": 0.4036281179138322,
      "grad_norm": 0.06865835189819336,
      "learning_rate": 0.0009797939432021326,
      "loss": 1.5426,
      "step": 9078
    },
    {
      "epoch": 0.40371704237250455,
      "grad_norm": 0.07144743204116821,
      "learning_rate": 0.000979784004471929,
      "loss": 1.5405,
      "step": 9080
    },
    {
      "epoch": 0.4038059668311769,
      "grad_norm": 0.0699494332075119,
      "learning_rate": 0.000979774063348483,
      "loss": 1.5377,
      "step": 9082
    },
    {
      "epoch": 0.4038948912898493,
      "grad_norm": 0.06973431259393692,
      "learning_rate": 0.0009797641198318445,
      "loss": 1.5392,
      "step": 9084
    },
    {
      "epoch": 0.40398381574852166,
      "grad_norm": 0.07053232192993164,
      "learning_rate": 0.000979754173922063,
      "loss": 1.537,
      "step": 9086
    },
    {
      "epoch": 0.40407274020719397,
      "grad_norm": 0.06677805632352829,
      "learning_rate": 0.0009797442256191877,
      "loss": 1.5489,
      "step": 9088
    },
    {
      "epoch": 0.40416166466586634,
      "grad_norm": 0.07138942182064056,
      "learning_rate": 0.000979734274923269,
      "loss": 1.5407,
      "step": 9090
    },
    {
      "epoch": 0.4042505891245387,
      "grad_norm": 0.07154888659715652,
      "learning_rate": 0.0009797243218343558,
      "loss": 1.5373,
      "step": 9092
    },
    {
      "epoch": 0.4043395135832111,
      "grad_norm": 0.07199876755475998,
      "learning_rate": 0.000979714366352498,
      "loss": 1.5344,
      "step": 9094
    },
    {
      "epoch": 0.40442843804188344,
      "grad_norm": 0.07113402336835861,
      "learning_rate": 0.0009797044084777456,
      "loss": 1.54,
      "step": 9096
    },
    {
      "epoch": 0.40451736250055575,
      "grad_norm": 0.07240010052919388,
      "learning_rate": 0.0009796944482101477,
      "loss": 1.5366,
      "step": 9098
    },
    {
      "epoch": 0.4046062869592281,
      "grad_norm": 0.07278742641210556,
      "learning_rate": 0.0009796844855497545,
      "loss": 1.5375,
      "step": 9100
    },
    {
      "epoch": 0.4046952114179005,
      "grad_norm": 0.06928924471139908,
      "learning_rate": 0.0009796745204966152,
      "loss": 1.5358,
      "step": 9102
    },
    {
      "epoch": 0.40478413587657286,
      "grad_norm": 0.06925825029611588,
      "learning_rate": 0.0009796645530507802,
      "loss": 1.544,
      "step": 9104
    },
    {
      "epoch": 0.4048730603352452,
      "grad_norm": 0.07367236167192459,
      "learning_rate": 0.0009796545832122985,
      "loss": 1.5375,
      "step": 9106
    },
    {
      "epoch": 0.40496198479391754,
      "grad_norm": 0.07212896645069122,
      "learning_rate": 0.00097964461098122,
      "loss": 1.5415,
      "step": 9108
    },
    {
      "epoch": 0.4050509092525899,
      "grad_norm": 0.07600012421607971,
      "learning_rate": 0.0009796346363575947,
      "loss": 1.5381,
      "step": 9110
    },
    {
      "epoch": 0.4051398337112623,
      "grad_norm": 0.07662640511989594,
      "learning_rate": 0.0009796246593414724,
      "loss": 1.5339,
      "step": 9112
    },
    {
      "epoch": 0.40522875816993464,
      "grad_norm": 0.070067398250103,
      "learning_rate": 0.0009796146799329025,
      "loss": 1.5423,
      "step": 9114
    },
    {
      "epoch": 0.405317682628607,
      "grad_norm": 0.07174143195152283,
      "learning_rate": 0.0009796046981319349,
      "loss": 1.5416,
      "step": 9116
    },
    {
      "epoch": 0.4054066070872794,
      "grad_norm": 0.06986463814973831,
      "learning_rate": 0.0009795947139386195,
      "loss": 1.5367,
      "step": 9118
    },
    {
      "epoch": 0.4054955315459517,
      "grad_norm": 0.0711868554353714,
      "learning_rate": 0.000979584727353006,
      "loss": 1.5423,
      "step": 9120
    },
    {
      "epoch": 0.40558445600462406,
      "grad_norm": 0.07185269147157669,
      "learning_rate": 0.0009795747383751446,
      "loss": 1.5374,
      "step": 9122
    },
    {
      "epoch": 0.40567338046329643,
      "grad_norm": 0.07167672365903854,
      "learning_rate": 0.0009795647470050846,
      "loss": 1.5397,
      "step": 9124
    },
    {
      "epoch": 0.4057623049219688,
      "grad_norm": 0.07179324328899384,
      "learning_rate": 0.000979554753242876,
      "loss": 1.5418,
      "step": 9126
    },
    {
      "epoch": 0.40585122938064117,
      "grad_norm": 0.07255962491035461,
      "learning_rate": 0.0009795447570885686,
      "loss": 1.5377,
      "step": 9128
    },
    {
      "epoch": 0.4059401538393135,
      "grad_norm": 0.0690963864326477,
      "learning_rate": 0.0009795347585422123,
      "loss": 1.5444,
      "step": 9130
    },
    {
      "epoch": 0.40602907829798585,
      "grad_norm": 0.07536199688911438,
      "learning_rate": 0.0009795247576038573,
      "loss": 1.5363,
      "step": 9132
    },
    {
      "epoch": 0.4061180027566582,
      "grad_norm": 0.06956053525209427,
      "learning_rate": 0.000979514754273553,
      "loss": 1.541,
      "step": 9134
    },
    {
      "epoch": 0.4062069272153306,
      "grad_norm": 0.07050742208957672,
      "learning_rate": 0.0009795047485513498,
      "loss": 1.5427,
      "step": 9136
    },
    {
      "epoch": 0.40629585167400295,
      "grad_norm": 0.0698935016989708,
      "learning_rate": 0.000979494740437297,
      "loss": 1.5354,
      "step": 9138
    },
    {
      "epoch": 0.4063847761326753,
      "grad_norm": 0.0688849613070488,
      "learning_rate": 0.0009794847299314448,
      "loss": 1.5354,
      "step": 9140
    },
    {
      "epoch": 0.40647370059134763,
      "grad_norm": 0.0699022114276886,
      "learning_rate": 0.0009794747170338435,
      "loss": 1.5383,
      "step": 9142
    },
    {
      "epoch": 0.40656262505002,
      "grad_norm": 0.06858205050230026,
      "learning_rate": 0.0009794647017445425,
      "loss": 1.5436,
      "step": 9144
    },
    {
      "epoch": 0.40665154950869237,
      "grad_norm": 0.06846922636032104,
      "learning_rate": 0.000979454684063592,
      "loss": 1.5442,
      "step": 9146
    },
    {
      "epoch": 0.40674047396736474,
      "grad_norm": 0.07031798362731934,
      "learning_rate": 0.000979444663991042,
      "loss": 1.5362,
      "step": 9148
    },
    {
      "epoch": 0.4068293984260371,
      "grad_norm": 0.07397054880857468,
      "learning_rate": 0.0009794346415269424,
      "loss": 1.5395,
      "step": 9150
    },
    {
      "epoch": 0.4069183228847094,
      "grad_norm": 0.0680195763707161,
      "learning_rate": 0.0009794246166713433,
      "loss": 1.5375,
      "step": 9152
    },
    {
      "epoch": 0.4070072473433818,
      "grad_norm": 0.06949716061353683,
      "learning_rate": 0.0009794145894242946,
      "loss": 1.5439,
      "step": 9154
    },
    {
      "epoch": 0.40709617180205415,
      "grad_norm": 0.07266386598348618,
      "learning_rate": 0.0009794045597858463,
      "loss": 1.5396,
      "step": 9156
    },
    {
      "epoch": 0.4071850962607265,
      "grad_norm": 0.07391185313463211,
      "learning_rate": 0.0009793945277560485,
      "loss": 1.5412,
      "step": 9158
    },
    {
      "epoch": 0.4072740207193989,
      "grad_norm": 0.07154107093811035,
      "learning_rate": 0.0009793844933349513,
      "loss": 1.5426,
      "step": 9160
    },
    {
      "epoch": 0.4073629451780712,
      "grad_norm": 0.06954929977655411,
      "learning_rate": 0.0009793744565226045,
      "loss": 1.5383,
      "step": 9162
    },
    {
      "epoch": 0.40745186963674357,
      "grad_norm": 0.06934426724910736,
      "learning_rate": 0.0009793644173190584,
      "loss": 1.5397,
      "step": 9164
    },
    {
      "epoch": 0.40754079409541594,
      "grad_norm": 0.07320673018693924,
      "learning_rate": 0.000979354375724363,
      "loss": 1.5324,
      "step": 9166
    },
    {
      "epoch": 0.4076297185540883,
      "grad_norm": 0.07140939682722092,
      "learning_rate": 0.0009793443317385685,
      "loss": 1.5336,
      "step": 9168
    },
    {
      "epoch": 0.4077186430127607,
      "grad_norm": 0.07228023558855057,
      "learning_rate": 0.0009793342853617248,
      "loss": 1.5398,
      "step": 9170
    },
    {
      "epoch": 0.40780756747143304,
      "grad_norm": 0.07187279313802719,
      "learning_rate": 0.0009793242365938822,
      "loss": 1.5452,
      "step": 9172
    },
    {
      "epoch": 0.40789649193010535,
      "grad_norm": 0.06871677190065384,
      "learning_rate": 0.000979314185435091,
      "loss": 1.5341,
      "step": 9174
    },
    {
      "epoch": 0.4079854163887777,
      "grad_norm": 0.06969435513019562,
      "learning_rate": 0.0009793041318854007,
      "loss": 1.5386,
      "step": 9176
    },
    {
      "epoch": 0.4080743408474501,
      "grad_norm": 0.07604380697011948,
      "learning_rate": 0.0009792940759448619,
      "loss": 1.5462,
      "step": 9178
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 0.06866500526666641,
      "learning_rate": 0.0009792840176135248,
      "loss": 1.5392,
      "step": 9180
    },
    {
      "epoch": 0.4082521897647948,
      "grad_norm": 0.06986360996961594,
      "learning_rate": 0.0009792739568914393,
      "loss": 1.5381,
      "step": 9182
    },
    {
      "epoch": 0.40834111422346714,
      "grad_norm": 0.06875632703304291,
      "learning_rate": 0.0009792638937786559,
      "loss": 1.5407,
      "step": 9184
    },
    {
      "epoch": 0.4084300386821395,
      "grad_norm": 0.06942211091518402,
      "learning_rate": 0.0009792538282752245,
      "loss": 1.5391,
      "step": 9186
    },
    {
      "epoch": 0.4085189631408119,
      "grad_norm": 0.06980740278959274,
      "learning_rate": 0.0009792437603811954,
      "loss": 1.5366,
      "step": 9188
    },
    {
      "epoch": 0.40860788759948424,
      "grad_norm": 0.0708228349685669,
      "learning_rate": 0.000979233690096619,
      "loss": 1.5413,
      "step": 9190
    },
    {
      "epoch": 0.4086968120581566,
      "grad_norm": 0.07046905905008316,
      "learning_rate": 0.0009792236174215455,
      "loss": 1.5366,
      "step": 9192
    },
    {
      "epoch": 0.408785736516829,
      "grad_norm": 0.0711686909198761,
      "learning_rate": 0.000979213542356025,
      "loss": 1.5439,
      "step": 9194
    },
    {
      "epoch": 0.4088746609755013,
      "grad_norm": 0.06964288651943207,
      "learning_rate": 0.0009792034649001079,
      "loss": 1.5353,
      "step": 9196
    },
    {
      "epoch": 0.40896358543417366,
      "grad_norm": 0.07251102477312088,
      "learning_rate": 0.0009791933850538442,
      "loss": 1.5396,
      "step": 9198
    },
    {
      "epoch": 0.40905250989284603,
      "grad_norm": 0.07063327729701996,
      "learning_rate": 0.0009791833028172843,
      "loss": 1.5388,
      "step": 9200
    },
    {
      "epoch": 0.4091414343515184,
      "grad_norm": 0.07538381218910217,
      "learning_rate": 0.0009791732181904788,
      "loss": 1.5417,
      "step": 9202
    },
    {
      "epoch": 0.40923035881019076,
      "grad_norm": 0.07273292541503906,
      "learning_rate": 0.0009791631311734774,
      "loss": 1.5364,
      "step": 9204
    },
    {
      "epoch": 0.4093192832688631,
      "grad_norm": 0.07258638739585876,
      "learning_rate": 0.0009791530417663308,
      "loss": 1.5409,
      "step": 9206
    },
    {
      "epoch": 0.40940820772753544,
      "grad_norm": 0.07772589474916458,
      "learning_rate": 0.0009791429499690896,
      "loss": 1.5398,
      "step": 9208
    },
    {
      "epoch": 0.4094971321862078,
      "grad_norm": 0.07376066595315933,
      "learning_rate": 0.0009791328557818035,
      "loss": 1.5427,
      "step": 9210
    },
    {
      "epoch": 0.4095860566448802,
      "grad_norm": 0.07310586422681808,
      "learning_rate": 0.0009791227592045235,
      "loss": 1.5431,
      "step": 9212
    },
    {
      "epoch": 0.40967498110355255,
      "grad_norm": 0.07371335476636887,
      "learning_rate": 0.0009791126602372996,
      "loss": 1.5384,
      "step": 9214
    },
    {
      "epoch": 0.4097639055622249,
      "grad_norm": 0.0725250244140625,
      "learning_rate": 0.000979102558880182,
      "loss": 1.5362,
      "step": 9216
    },
    {
      "epoch": 0.40985283002089723,
      "grad_norm": 0.07021638751029968,
      "learning_rate": 0.0009790924551332215,
      "loss": 1.5344,
      "step": 9218
    },
    {
      "epoch": 0.4099417544795696,
      "grad_norm": 0.06901945173740387,
      "learning_rate": 0.0009790823489964683,
      "loss": 1.5343,
      "step": 9220
    },
    {
      "epoch": 0.41003067893824197,
      "grad_norm": 0.06901168823242188,
      "learning_rate": 0.0009790722404699726,
      "loss": 1.5369,
      "step": 9222
    },
    {
      "epoch": 0.41011960339691433,
      "grad_norm": 0.06912058591842651,
      "learning_rate": 0.0009790621295537852,
      "loss": 1.5356,
      "step": 9224
    },
    {
      "epoch": 0.4102085278555867,
      "grad_norm": 0.07027976959943771,
      "learning_rate": 0.0009790520162479563,
      "loss": 1.5343,
      "step": 9226
    },
    {
      "epoch": 0.410297452314259,
      "grad_norm": 0.07126928120851517,
      "learning_rate": 0.0009790419005525366,
      "loss": 1.5394,
      "step": 9228
    },
    {
      "epoch": 0.4103863767729314,
      "grad_norm": 0.0708712786436081,
      "learning_rate": 0.0009790317824675762,
      "loss": 1.5438,
      "step": 9230
    },
    {
      "epoch": 0.41047530123160375,
      "grad_norm": 0.07011223584413528,
      "learning_rate": 0.000979021661993126,
      "loss": 1.5359,
      "step": 9232
    },
    {
      "epoch": 0.4105642256902761,
      "grad_norm": 0.07202787697315216,
      "learning_rate": 0.000979011539129236,
      "loss": 1.5422,
      "step": 9234
    },
    {
      "epoch": 0.4106531501489485,
      "grad_norm": 0.06934972107410431,
      "learning_rate": 0.0009790014138759571,
      "loss": 1.5399,
      "step": 9236
    },
    {
      "epoch": 0.4107420746076208,
      "grad_norm": 0.07186301797628403,
      "learning_rate": 0.0009789912862333394,
      "loss": 1.5426,
      "step": 9238
    },
    {
      "epoch": 0.41083099906629317,
      "grad_norm": 0.07302137464284897,
      "learning_rate": 0.0009789811562014338,
      "loss": 1.5408,
      "step": 9240
    },
    {
      "epoch": 0.41091992352496554,
      "grad_norm": 0.07044612616300583,
      "learning_rate": 0.0009789710237802908,
      "loss": 1.545,
      "step": 9242
    },
    {
      "epoch": 0.4110088479836379,
      "grad_norm": 0.07422881573438644,
      "learning_rate": 0.0009789608889699607,
      "loss": 1.5366,
      "step": 9244
    },
    {
      "epoch": 0.41109777244231027,
      "grad_norm": 0.07070881128311157,
      "learning_rate": 0.0009789507517704943,
      "loss": 1.5344,
      "step": 9246
    },
    {
      "epoch": 0.41118669690098264,
      "grad_norm": 0.07254374772310257,
      "learning_rate": 0.0009789406121819418,
      "loss": 1.5384,
      "step": 9248
    },
    {
      "epoch": 0.41127562135965495,
      "grad_norm": 0.07279791682958603,
      "learning_rate": 0.0009789304702043542,
      "loss": 1.5434,
      "step": 9250
    },
    {
      "epoch": 0.4113645458183273,
      "grad_norm": 0.07429590076208115,
      "learning_rate": 0.000978920325837782,
      "loss": 1.5449,
      "step": 9252
    },
    {
      "epoch": 0.4114534702769997,
      "grad_norm": 0.07568711042404175,
      "learning_rate": 0.0009789101790822756,
      "loss": 1.541,
      "step": 9254
    },
    {
      "epoch": 0.41154239473567206,
      "grad_norm": 0.07262945920228958,
      "learning_rate": 0.0009789000299378857,
      "loss": 1.5372,
      "step": 9256
    },
    {
      "epoch": 0.4116313191943444,
      "grad_norm": 0.07350386679172516,
      "learning_rate": 0.000978889878404663,
      "loss": 1.5409,
      "step": 9258
    },
    {
      "epoch": 0.41172024365301674,
      "grad_norm": 0.0701087936758995,
      "learning_rate": 0.000978879724482658,
      "loss": 1.5446,
      "step": 9260
    },
    {
      "epoch": 0.4118091681116891,
      "grad_norm": 0.06851620972156525,
      "learning_rate": 0.0009788695681719217,
      "loss": 1.5366,
      "step": 9262
    },
    {
      "epoch": 0.4118980925703615,
      "grad_norm": 0.06817908585071564,
      "learning_rate": 0.0009788594094725042,
      "loss": 1.5356,
      "step": 9264
    },
    {
      "epoch": 0.41198701702903384,
      "grad_norm": 0.06984448432922363,
      "learning_rate": 0.0009788492483844566,
      "loss": 1.5404,
      "step": 9266
    },
    {
      "epoch": 0.4120759414877062,
      "grad_norm": 0.07095382362604141,
      "learning_rate": 0.0009788390849078295,
      "loss": 1.5423,
      "step": 9268
    },
    {
      "epoch": 0.4121648659463786,
      "grad_norm": 0.06866953521966934,
      "learning_rate": 0.0009788289190426735,
      "loss": 1.5337,
      "step": 9270
    },
    {
      "epoch": 0.4122537904050509,
      "grad_norm": 0.06914438307285309,
      "learning_rate": 0.0009788187507890395,
      "loss": 1.538,
      "step": 9272
    },
    {
      "epoch": 0.41234271486372326,
      "grad_norm": 0.06855052709579468,
      "learning_rate": 0.0009788085801469778,
      "loss": 1.5351,
      "step": 9274
    },
    {
      "epoch": 0.4124316393223956,
      "grad_norm": 0.0681350976228714,
      "learning_rate": 0.0009787984071165398,
      "loss": 1.5395,
      "step": 9276
    },
    {
      "epoch": 0.412520563781068,
      "grad_norm": 0.07035652548074722,
      "learning_rate": 0.0009787882316977757,
      "loss": 1.5417,
      "step": 9278
    },
    {
      "epoch": 0.41260948823974036,
      "grad_norm": 0.06886482238769531,
      "learning_rate": 0.0009787780538907365,
      "loss": 1.537,
      "step": 9280
    },
    {
      "epoch": 0.4126984126984127,
      "grad_norm": 0.07588198781013489,
      "learning_rate": 0.000978767873695473,
      "loss": 1.5403,
      "step": 9282
    },
    {
      "epoch": 0.41278733715708504,
      "grad_norm": 0.069828562438488,
      "learning_rate": 0.0009787576911120355,
      "loss": 1.544,
      "step": 9284
    },
    {
      "epoch": 0.4128762616157574,
      "grad_norm": 0.07009178400039673,
      "learning_rate": 0.0009787475061404753,
      "loss": 1.5365,
      "step": 9286
    },
    {
      "epoch": 0.4129651860744298,
      "grad_norm": 0.06862553209066391,
      "learning_rate": 0.0009787373187808432,
      "loss": 1.538,
      "step": 9288
    },
    {
      "epoch": 0.41305411053310215,
      "grad_norm": 0.07022769749164581,
      "learning_rate": 0.00097872712903319,
      "loss": 1.543,
      "step": 9290
    },
    {
      "epoch": 0.41314303499177446,
      "grad_norm": 0.07054140418767929,
      "learning_rate": 0.000978716936897566,
      "loss": 1.5383,
      "step": 9292
    },
    {
      "epoch": 0.41323195945044683,
      "grad_norm": 0.07116847485303879,
      "learning_rate": 0.0009787067423740227,
      "loss": 1.5386,
      "step": 9294
    },
    {
      "epoch": 0.4133208839091192,
      "grad_norm": 0.06990710645914078,
      "learning_rate": 0.0009786965454626107,
      "loss": 1.5343,
      "step": 9296
    },
    {
      "epoch": 0.41340980836779156,
      "grad_norm": 0.06872306019067764,
      "learning_rate": 0.0009786863461633808,
      "loss": 1.5424,
      "step": 9298
    },
    {
      "epoch": 0.41349873282646393,
      "grad_norm": 0.06872355192899704,
      "learning_rate": 0.0009786761444763841,
      "loss": 1.541,
      "step": 9300
    },
    {
      "epoch": 0.4135876572851363,
      "grad_norm": 0.07212715595960617,
      "learning_rate": 0.0009786659404016711,
      "loss": 1.5376,
      "step": 9302
    },
    {
      "epoch": 0.4136765817438086,
      "grad_norm": 0.06772853434085846,
      "learning_rate": 0.0009786557339392932,
      "loss": 1.5369,
      "step": 9304
    },
    {
      "epoch": 0.413765506202481,
      "grad_norm": 0.06942387670278549,
      "learning_rate": 0.000978645525089301,
      "loss": 1.5344,
      "step": 9306
    },
    {
      "epoch": 0.41385443066115335,
      "grad_norm": 0.07053601741790771,
      "learning_rate": 0.0009786353138517454,
      "loss": 1.5401,
      "step": 9308
    },
    {
      "epoch": 0.4139433551198257,
      "grad_norm": 0.07274100184440613,
      "learning_rate": 0.0009786251002266773,
      "loss": 1.5324,
      "step": 9310
    },
    {
      "epoch": 0.4140322795784981,
      "grad_norm": 0.07127835601568222,
      "learning_rate": 0.000978614884214148,
      "loss": 1.5427,
      "step": 9312
    },
    {
      "epoch": 0.4141212040371704,
      "grad_norm": 0.0735345184803009,
      "learning_rate": 0.0009786046658142081,
      "loss": 1.5359,
      "step": 9314
    },
    {
      "epoch": 0.41421012849584277,
      "grad_norm": 0.07226788252592087,
      "learning_rate": 0.0009785944450269087,
      "loss": 1.5428,
      "step": 9316
    },
    {
      "epoch": 0.41429905295451513,
      "grad_norm": 0.0691119134426117,
      "learning_rate": 0.0009785842218523006,
      "loss": 1.5388,
      "step": 9318
    },
    {
      "epoch": 0.4143879774131875,
      "grad_norm": 0.06924230605363846,
      "learning_rate": 0.000978573996290435,
      "loss": 1.5377,
      "step": 9320
    },
    {
      "epoch": 0.41447690187185987,
      "grad_norm": 0.07052425295114517,
      "learning_rate": 0.000978563768341363,
      "loss": 1.5378,
      "step": 9322
    },
    {
      "epoch": 0.41456582633053224,
      "grad_norm": 0.07418438792228699,
      "learning_rate": 0.0009785535380051355,
      "loss": 1.5331,
      "step": 9324
    },
    {
      "epoch": 0.41465475078920455,
      "grad_norm": 0.07372714579105377,
      "learning_rate": 0.0009785433052818034,
      "loss": 1.5363,
      "step": 9326
    },
    {
      "epoch": 0.4147436752478769,
      "grad_norm": 0.07146646827459335,
      "learning_rate": 0.000978533070171418,
      "loss": 1.5329,
      "step": 9328
    },
    {
      "epoch": 0.4148325997065493,
      "grad_norm": 0.07278136909008026,
      "learning_rate": 0.00097852283267403,
      "loss": 1.5412,
      "step": 9330
    },
    {
      "epoch": 0.41492152416522166,
      "grad_norm": 0.07172350585460663,
      "learning_rate": 0.0009785125927896908,
      "loss": 1.5373,
      "step": 9332
    },
    {
      "epoch": 0.415010448623894,
      "grad_norm": 0.07292774319648743,
      "learning_rate": 0.0009785023505184513,
      "loss": 1.5368,
      "step": 9334
    },
    {
      "epoch": 0.41509937308256634,
      "grad_norm": 0.07089949399232864,
      "learning_rate": 0.0009784921058603629,
      "loss": 1.5319,
      "step": 9336
    },
    {
      "epoch": 0.4151882975412387,
      "grad_norm": 0.0746246799826622,
      "learning_rate": 0.0009784818588154762,
      "loss": 1.541,
      "step": 9338
    },
    {
      "epoch": 0.4152772219999111,
      "grad_norm": 0.07343286275863647,
      "learning_rate": 0.0009784716093838425,
      "loss": 1.5374,
      "step": 9340
    },
    {
      "epoch": 0.41536614645858344,
      "grad_norm": 0.06831394135951996,
      "learning_rate": 0.0009784613575655131,
      "loss": 1.532,
      "step": 9342
    },
    {
      "epoch": 0.4154550709172558,
      "grad_norm": 0.07356590032577515,
      "learning_rate": 0.000978451103360539,
      "loss": 1.5349,
      "step": 9344
    },
    {
      "epoch": 0.4155439953759281,
      "grad_norm": 0.07031702995300293,
      "learning_rate": 0.0009784408467689717,
      "loss": 1.5351,
      "step": 9346
    },
    {
      "epoch": 0.4156329198346005,
      "grad_norm": 0.07025104761123657,
      "learning_rate": 0.0009784305877908615,
      "loss": 1.5372,
      "step": 9348
    },
    {
      "epoch": 0.41572184429327286,
      "grad_norm": 0.06900499761104584,
      "learning_rate": 0.0009784203264262604,
      "loss": 1.5366,
      "step": 9350
    },
    {
      "epoch": 0.4158107687519452,
      "grad_norm": 0.06960804015398026,
      "learning_rate": 0.0009784100626752193,
      "loss": 1.5422,
      "step": 9352
    },
    {
      "epoch": 0.4158996932106176,
      "grad_norm": 0.07161404937505722,
      "learning_rate": 0.0009783997965377893,
      "loss": 1.5398,
      "step": 9354
    },
    {
      "epoch": 0.41598861766928996,
      "grad_norm": 0.07421243190765381,
      "learning_rate": 0.000978389528014022,
      "loss": 1.5351,
      "step": 9356
    },
    {
      "epoch": 0.4160775421279623,
      "grad_norm": 0.07687795907258987,
      "learning_rate": 0.000978379257103968,
      "loss": 1.5411,
      "step": 9358
    },
    {
      "epoch": 0.41616646658663464,
      "grad_norm": 0.07376468926668167,
      "learning_rate": 0.0009783689838076788,
      "loss": 1.5342,
      "step": 9360
    },
    {
      "epoch": 0.416255391045307,
      "grad_norm": 0.07067352533340454,
      "learning_rate": 0.000978358708125206,
      "loss": 1.5381,
      "step": 9362
    },
    {
      "epoch": 0.4163443155039794,
      "grad_norm": 0.06940723210573196,
      "learning_rate": 0.0009783484300566003,
      "loss": 1.541,
      "step": 9364
    },
    {
      "epoch": 0.41643323996265175,
      "grad_norm": 0.06866216659545898,
      "learning_rate": 0.0009783381496019136,
      "loss": 1.5412,
      "step": 9366
    },
    {
      "epoch": 0.41652216442132406,
      "grad_norm": 0.06930825859308243,
      "learning_rate": 0.0009783278667611964,
      "loss": 1.5347,
      "step": 9368
    },
    {
      "epoch": 0.4166110888799964,
      "grad_norm": 0.06734399497509003,
      "learning_rate": 0.0009783175815345005,
      "loss": 1.537,
      "step": 9370
    },
    {
      "epoch": 0.4167000133386688,
      "grad_norm": 0.06903259456157684,
      "learning_rate": 0.0009783072939218772,
      "loss": 1.5386,
      "step": 9372
    },
    {
      "epoch": 0.41678893779734116,
      "grad_norm": 0.0670986995100975,
      "learning_rate": 0.0009782970039233776,
      "loss": 1.5329,
      "step": 9374
    },
    {
      "epoch": 0.41687786225601353,
      "grad_norm": 0.07043662667274475,
      "learning_rate": 0.0009782867115390532,
      "loss": 1.5364,
      "step": 9376
    },
    {
      "epoch": 0.4169667867146859,
      "grad_norm": 0.06715195626020432,
      "learning_rate": 0.0009782764167689554,
      "loss": 1.5325,
      "step": 9378
    },
    {
      "epoch": 0.4170557111733582,
      "grad_norm": 0.06987708806991577,
      "learning_rate": 0.0009782661196131354,
      "loss": 1.5398,
      "step": 9380
    },
    {
      "epoch": 0.4171446356320306,
      "grad_norm": 0.06837180256843567,
      "learning_rate": 0.0009782558200716446,
      "loss": 1.5339,
      "step": 9382
    },
    {
      "epoch": 0.41723356009070295,
      "grad_norm": 0.0699625238776207,
      "learning_rate": 0.0009782455181445344,
      "loss": 1.5387,
      "step": 9384
    },
    {
      "epoch": 0.4173224845493753,
      "grad_norm": 0.07226444780826569,
      "learning_rate": 0.0009782352138318561,
      "loss": 1.5419,
      "step": 9386
    },
    {
      "epoch": 0.4174114090080477,
      "grad_norm": 0.06939677894115448,
      "learning_rate": 0.0009782249071336611,
      "loss": 1.5394,
      "step": 9388
    },
    {
      "epoch": 0.41750033346672,
      "grad_norm": 0.06957529485225677,
      "learning_rate": 0.000978214598050001,
      "loss": 1.5342,
      "step": 9390
    },
    {
      "epoch": 0.41758925792539237,
      "grad_norm": 0.06821033358573914,
      "learning_rate": 0.000978204286580927,
      "loss": 1.535,
      "step": 9392
    },
    {
      "epoch": 0.41767818238406473,
      "grad_norm": 0.06692170351743698,
      "learning_rate": 0.000978193972726491,
      "loss": 1.5296,
      "step": 9394
    },
    {
      "epoch": 0.4177671068427371,
      "grad_norm": 0.06882653385400772,
      "learning_rate": 0.0009781836564867437,
      "loss": 1.5383,
      "step": 9396
    },
    {
      "epoch": 0.41785603130140947,
      "grad_norm": 0.07427268475294113,
      "learning_rate": 0.000978173337861737,
      "loss": 1.5384,
      "step": 9398
    },
    {
      "epoch": 0.41794495576008184,
      "grad_norm": 0.07215935736894608,
      "learning_rate": 0.0009781630168515223,
      "loss": 1.5401,
      "step": 9400
    },
    {
      "epoch": 0.41803388021875415,
      "grad_norm": 0.07148215919733047,
      "learning_rate": 0.0009781526934561514,
      "loss": 1.5448,
      "step": 9402
    },
    {
      "epoch": 0.4181228046774265,
      "grad_norm": 0.06973208487033844,
      "learning_rate": 0.0009781423676756753,
      "loss": 1.5379,
      "step": 9404
    },
    {
      "epoch": 0.4182117291360989,
      "grad_norm": 0.07168926298618317,
      "learning_rate": 0.0009781320395101456,
      "loss": 1.5365,
      "step": 9406
    },
    {
      "epoch": 0.41830065359477125,
      "grad_norm": 0.0705469623208046,
      "learning_rate": 0.0009781217089596139,
      "loss": 1.5359,
      "step": 9408
    },
    {
      "epoch": 0.4183895780534436,
      "grad_norm": 0.06955131143331528,
      "learning_rate": 0.000978111376024132,
      "loss": 1.5371,
      "step": 9410
    },
    {
      "epoch": 0.41847850251211594,
      "grad_norm": 0.07158604264259338,
      "learning_rate": 0.000978101040703751,
      "loss": 1.54,
      "step": 9412
    },
    {
      "epoch": 0.4185674269707883,
      "grad_norm": 0.07020861655473709,
      "learning_rate": 0.0009780907029985227,
      "loss": 1.5381,
      "step": 9414
    },
    {
      "epoch": 0.41865635142946067,
      "grad_norm": 0.07165636867284775,
      "learning_rate": 0.0009780803629084987,
      "loss": 1.5382,
      "step": 9416
    },
    {
      "epoch": 0.41874527588813304,
      "grad_norm": 0.07206343859434128,
      "learning_rate": 0.0009780700204337304,
      "loss": 1.5348,
      "step": 9418
    },
    {
      "epoch": 0.4188342003468054,
      "grad_norm": 0.06763043999671936,
      "learning_rate": 0.0009780596755742694,
      "loss": 1.5375,
      "step": 9420
    },
    {
      "epoch": 0.4189231248054777,
      "grad_norm": 0.07210266590118408,
      "learning_rate": 0.0009780493283301675,
      "loss": 1.5403,
      "step": 9422
    },
    {
      "epoch": 0.4190120492641501,
      "grad_norm": 0.07414006441831589,
      "learning_rate": 0.000978038978701476,
      "loss": 1.537,
      "step": 9424
    },
    {
      "epoch": 0.41910097372282246,
      "grad_norm": 0.07020355015993118,
      "learning_rate": 0.000978028626688247,
      "loss": 1.5321,
      "step": 9426
    },
    {
      "epoch": 0.4191898981814948,
      "grad_norm": 0.07178431004285812,
      "learning_rate": 0.0009780182722905317,
      "loss": 1.5328,
      "step": 9428
    },
    {
      "epoch": 0.4192788226401672,
      "grad_norm": 0.07044417411088943,
      "learning_rate": 0.0009780079155083821,
      "loss": 1.5356,
      "step": 9430
    },
    {
      "epoch": 0.41936774709883956,
      "grad_norm": 0.07170826941728592,
      "learning_rate": 0.0009779975563418495,
      "loss": 1.5315,
      "step": 9432
    },
    {
      "epoch": 0.4194566715575119,
      "grad_norm": 0.07005254924297333,
      "learning_rate": 0.0009779871947909857,
      "loss": 1.5367,
      "step": 9434
    },
    {
      "epoch": 0.41954559601618424,
      "grad_norm": 0.06897489726543427,
      "learning_rate": 0.0009779768308558427,
      "loss": 1.5291,
      "step": 9436
    },
    {
      "epoch": 0.4196345204748566,
      "grad_norm": 0.07095732539892197,
      "learning_rate": 0.0009779664645364716,
      "loss": 1.5301,
      "step": 9438
    },
    {
      "epoch": 0.419723444933529,
      "grad_norm": 0.07314879447221756,
      "learning_rate": 0.0009779560958329246,
      "loss": 1.5323,
      "step": 9440
    },
    {
      "epoch": 0.41981236939220135,
      "grad_norm": 0.07073177397251129,
      "learning_rate": 0.0009779457247452532,
      "loss": 1.5375,
      "step": 9442
    },
    {
      "epoch": 0.41990129385087366,
      "grad_norm": 0.0689346119761467,
      "learning_rate": 0.0009779353512735093,
      "loss": 1.5345,
      "step": 9444
    },
    {
      "epoch": 0.419990218309546,
      "grad_norm": 0.06825485080480576,
      "learning_rate": 0.0009779249754177444,
      "loss": 1.5362,
      "step": 9446
    },
    {
      "epoch": 0.4200791427682184,
      "grad_norm": 0.07014141231775284,
      "learning_rate": 0.0009779145971780103,
      "loss": 1.5383,
      "step": 9448
    },
    {
      "epoch": 0.42016806722689076,
      "grad_norm": 0.06948299705982208,
      "learning_rate": 0.0009779042165543592,
      "loss": 1.5405,
      "step": 9450
    },
    {
      "epoch": 0.42025699168556313,
      "grad_norm": 0.06821645051240921,
      "learning_rate": 0.0009778938335468423,
      "loss": 1.5361,
      "step": 9452
    },
    {
      "epoch": 0.4203459161442355,
      "grad_norm": 0.0685497298836708,
      "learning_rate": 0.0009778834481555118,
      "loss": 1.536,
      "step": 9454
    },
    {
      "epoch": 0.4204348406029078,
      "grad_norm": 0.07217854261398315,
      "learning_rate": 0.0009778730603804192,
      "loss": 1.5344,
      "step": 9456
    },
    {
      "epoch": 0.4205237650615802,
      "grad_norm": 0.06981514394283295,
      "learning_rate": 0.0009778626702216164,
      "loss": 1.5345,
      "step": 9458
    },
    {
      "epoch": 0.42061268952025255,
      "grad_norm": 0.07042837888002396,
      "learning_rate": 0.0009778522776791553,
      "loss": 1.5362,
      "step": 9460
    },
    {
      "epoch": 0.4207016139789249,
      "grad_norm": 0.07101742923259735,
      "learning_rate": 0.0009778418827530878,
      "loss": 1.5426,
      "step": 9462
    },
    {
      "epoch": 0.4207905384375973,
      "grad_norm": 0.06968886405229568,
      "learning_rate": 0.0009778314854434656,
      "loss": 1.5348,
      "step": 9464
    },
    {
      "epoch": 0.4208794628962696,
      "grad_norm": 0.06759744137525558,
      "learning_rate": 0.0009778210857503407,
      "loss": 1.5297,
      "step": 9466
    },
    {
      "epoch": 0.42096838735494196,
      "grad_norm": 0.0696602612733841,
      "learning_rate": 0.0009778106836737647,
      "loss": 1.5383,
      "step": 9468
    },
    {
      "epoch": 0.42105731181361433,
      "grad_norm": 0.06992914527654648,
      "learning_rate": 0.00097780027921379,
      "loss": 1.5319,
      "step": 9470
    },
    {
      "epoch": 0.4211462362722867,
      "grad_norm": 0.07088860124349594,
      "learning_rate": 0.0009777898723704681,
      "loss": 1.54,
      "step": 9472
    },
    {
      "epoch": 0.42123516073095907,
      "grad_norm": 0.0724070593714714,
      "learning_rate": 0.000977779463143851,
      "loss": 1.5378,
      "step": 9474
    },
    {
      "epoch": 0.4213240851896314,
      "grad_norm": 0.07188228517770767,
      "learning_rate": 0.0009777690515339905,
      "loss": 1.5392,
      "step": 9476
    },
    {
      "epoch": 0.42141300964830375,
      "grad_norm": 0.06922028958797455,
      "learning_rate": 0.0009777586375409389,
      "loss": 1.527,
      "step": 9478
    },
    {
      "epoch": 0.4215019341069761,
      "grad_norm": 0.07125406712293625,
      "learning_rate": 0.0009777482211647476,
      "loss": 1.5363,
      "step": 9480
    },
    {
      "epoch": 0.4215908585656485,
      "grad_norm": 0.06816605478525162,
      "learning_rate": 0.0009777378024054693,
      "loss": 1.5298,
      "step": 9482
    },
    {
      "epoch": 0.42167978302432085,
      "grad_norm": 0.06775356084108353,
      "learning_rate": 0.0009777273812631552,
      "loss": 1.5418,
      "step": 9484
    },
    {
      "epoch": 0.4217687074829932,
      "grad_norm": 0.07036928087472916,
      "learning_rate": 0.0009777169577378578,
      "loss": 1.537,
      "step": 9486
    },
    {
      "epoch": 0.42185763194166553,
      "grad_norm": 0.07150757312774658,
      "learning_rate": 0.0009777065318296288,
      "loss": 1.5352,
      "step": 9488
    },
    {
      "epoch": 0.4219465564003379,
      "grad_norm": 0.06876358389854431,
      "learning_rate": 0.0009776961035385203,
      "loss": 1.5388,
      "step": 9490
    },
    {
      "epoch": 0.42203548085901027,
      "grad_norm": 0.07003464549779892,
      "learning_rate": 0.0009776856728645844,
      "loss": 1.537,
      "step": 9492
    },
    {
      "epoch": 0.42212440531768264,
      "grad_norm": 0.070342518389225,
      "learning_rate": 0.0009776752398078731,
      "loss": 1.5322,
      "step": 9494
    },
    {
      "epoch": 0.422213329776355,
      "grad_norm": 0.06741419434547424,
      "learning_rate": 0.0009776648043684384,
      "loss": 1.5398,
      "step": 9496
    },
    {
      "epoch": 0.4223022542350273,
      "grad_norm": 0.07096916437149048,
      "learning_rate": 0.0009776543665463325,
      "loss": 1.5313,
      "step": 9498
    },
    {
      "epoch": 0.4223911786936997,
      "grad_norm": 0.06838645040988922,
      "learning_rate": 0.000977643926341607,
      "loss": 1.5372,
      "step": 9500
    },
    {
      "epoch": 0.4223911786936997,
      "eval_loss": 1.5147486925125122,
      "eval_runtime": 12.692,
      "eval_samples_per_second": 544.436,
      "eval_steps_per_second": 68.074,
      "step": 9500
    },
    {
      "epoch": 0.42248010315237206,
      "grad_norm": 0.07152796536684036,
      "learning_rate": 0.0009776334837543147,
      "loss": 1.5332,
      "step": 9502
    },
    {
      "epoch": 0.4225690276110444,
      "grad_norm": 0.0700407326221466,
      "learning_rate": 0.000977623038784507,
      "loss": 1.5358,
      "step": 9504
    },
    {
      "epoch": 0.4226579520697168,
      "grad_norm": 0.06791914999485016,
      "learning_rate": 0.0009776125914322364,
      "loss": 1.5323,
      "step": 9506
    },
    {
      "epoch": 0.42274687652838916,
      "grad_norm": 0.07237772643566132,
      "learning_rate": 0.0009776021416975549,
      "loss": 1.5339,
      "step": 9508
    },
    {
      "epoch": 0.42283580098706147,
      "grad_norm": 0.07078398764133453,
      "learning_rate": 0.0009775916895805145,
      "loss": 1.5327,
      "step": 9510
    },
    {
      "epoch": 0.42292472544573384,
      "grad_norm": 0.07085014879703522,
      "learning_rate": 0.0009775812350811678,
      "loss": 1.5297,
      "step": 9512
    },
    {
      "epoch": 0.4230136499044062,
      "grad_norm": 0.07237362116575241,
      "learning_rate": 0.0009775707781995665,
      "loss": 1.5296,
      "step": 9514
    },
    {
      "epoch": 0.4231025743630786,
      "grad_norm": 0.06783927977085114,
      "learning_rate": 0.0009775603189357627,
      "loss": 1.5367,
      "step": 9516
    },
    {
      "epoch": 0.42319149882175094,
      "grad_norm": 0.06836774200201035,
      "learning_rate": 0.0009775498572898089,
      "loss": 1.5336,
      "step": 9518
    },
    {
      "epoch": 0.42328042328042326,
      "grad_norm": 0.0725974515080452,
      "learning_rate": 0.0009775393932617573,
      "loss": 1.5319,
      "step": 9520
    },
    {
      "epoch": 0.4233693477390956,
      "grad_norm": 0.07181134074926376,
      "learning_rate": 0.0009775289268516597,
      "loss": 1.5352,
      "step": 9522
    },
    {
      "epoch": 0.423458272197768,
      "grad_norm": 0.07021788507699966,
      "learning_rate": 0.0009775184580595687,
      "loss": 1.5393,
      "step": 9524
    },
    {
      "epoch": 0.42354719665644036,
      "grad_norm": 0.06836768984794617,
      "learning_rate": 0.0009775079868855363,
      "loss": 1.5354,
      "step": 9526
    },
    {
      "epoch": 0.42363612111511273,
      "grad_norm": 0.06834634393453598,
      "learning_rate": 0.0009774975133296148,
      "loss": 1.5393,
      "step": 9528
    },
    {
      "epoch": 0.42372504557378504,
      "grad_norm": 0.0703490823507309,
      "learning_rate": 0.0009774870373918565,
      "loss": 1.5319,
      "step": 9530
    },
    {
      "epoch": 0.4238139700324574,
      "grad_norm": 0.06841664761304855,
      "learning_rate": 0.0009774765590723133,
      "loss": 1.5413,
      "step": 9532
    },
    {
      "epoch": 0.4239028944911298,
      "grad_norm": 0.06665299087762833,
      "learning_rate": 0.0009774660783710381,
      "loss": 1.5359,
      "step": 9534
    },
    {
      "epoch": 0.42399181894980215,
      "grad_norm": 0.06913217157125473,
      "learning_rate": 0.0009774555952880828,
      "loss": 1.5409,
      "step": 9536
    },
    {
      "epoch": 0.4240807434084745,
      "grad_norm": 0.06786402314901352,
      "learning_rate": 0.0009774451098234999,
      "loss": 1.5291,
      "step": 9538
    },
    {
      "epoch": 0.4241696678671469,
      "grad_norm": 0.06822808086872101,
      "learning_rate": 0.000977434621977341,
      "loss": 1.5398,
      "step": 9540
    },
    {
      "epoch": 0.4242585923258192,
      "grad_norm": 0.06818651407957077,
      "learning_rate": 0.0009774241317496593,
      "loss": 1.5413,
      "step": 9542
    },
    {
      "epoch": 0.42434751678449156,
      "grad_norm": 0.06693682074546814,
      "learning_rate": 0.000977413639140507,
      "loss": 1.5302,
      "step": 9544
    },
    {
      "epoch": 0.42443644124316393,
      "grad_norm": 0.06880820542573929,
      "learning_rate": 0.0009774031441499359,
      "loss": 1.5323,
      "step": 9546
    },
    {
      "epoch": 0.4245253657018363,
      "grad_norm": 0.0655490830540657,
      "learning_rate": 0.0009773926467779987,
      "loss": 1.5412,
      "step": 9548
    },
    {
      "epoch": 0.42461429016050867,
      "grad_norm": 0.06712329387664795,
      "learning_rate": 0.0009773821470247478,
      "loss": 1.5346,
      "step": 9550
    },
    {
      "epoch": 0.424703214619181,
      "grad_norm": 0.06615650653839111,
      "learning_rate": 0.0009773716448902355,
      "loss": 1.5341,
      "step": 9552
    },
    {
      "epoch": 0.42479213907785335,
      "grad_norm": 0.06936550885438919,
      "learning_rate": 0.0009773611403745143,
      "loss": 1.5341,
      "step": 9554
    },
    {
      "epoch": 0.4248810635365257,
      "grad_norm": 0.06751050800085068,
      "learning_rate": 0.0009773506334776363,
      "loss": 1.5379,
      "step": 9556
    },
    {
      "epoch": 0.4249699879951981,
      "grad_norm": 0.0678553357720375,
      "learning_rate": 0.0009773401241996542,
      "loss": 1.5381,
      "step": 9558
    },
    {
      "epoch": 0.42505891245387045,
      "grad_norm": 0.06892528384923935,
      "learning_rate": 0.0009773296125406203,
      "loss": 1.5373,
      "step": 9560
    },
    {
      "epoch": 0.4251478369125428,
      "grad_norm": 0.06836491823196411,
      "learning_rate": 0.0009773190985005872,
      "loss": 1.5374,
      "step": 9562
    },
    {
      "epoch": 0.42523676137121513,
      "grad_norm": 0.07044228166341782,
      "learning_rate": 0.000977308582079607,
      "loss": 1.5313,
      "step": 9564
    },
    {
      "epoch": 0.4253256858298875,
      "grad_norm": 0.06915424019098282,
      "learning_rate": 0.0009772980632777324,
      "loss": 1.5336,
      "step": 9566
    },
    {
      "epoch": 0.42541461028855987,
      "grad_norm": 0.0704522505402565,
      "learning_rate": 0.0009772875420950159,
      "loss": 1.5408,
      "step": 9568
    },
    {
      "epoch": 0.42550353474723224,
      "grad_norm": 0.06899117678403854,
      "learning_rate": 0.0009772770185315098,
      "loss": 1.538,
      "step": 9570
    },
    {
      "epoch": 0.4255924592059046,
      "grad_norm": 0.06853268295526505,
      "learning_rate": 0.000977266492587267,
      "loss": 1.5374,
      "step": 9572
    },
    {
      "epoch": 0.4256813836645769,
      "grad_norm": 0.07276134192943573,
      "learning_rate": 0.0009772559642623395,
      "loss": 1.5343,
      "step": 9574
    },
    {
      "epoch": 0.4257703081232493,
      "grad_norm": 0.06734218448400497,
      "learning_rate": 0.00097724543355678,
      "loss": 1.5361,
      "step": 9576
    },
    {
      "epoch": 0.42585923258192165,
      "grad_norm": 0.07109152525663376,
      "learning_rate": 0.0009772349004706412,
      "loss": 1.5397,
      "step": 9578
    },
    {
      "epoch": 0.425948157040594,
      "grad_norm": 0.06941290199756622,
      "learning_rate": 0.0009772243650039755,
      "loss": 1.5369,
      "step": 9580
    },
    {
      "epoch": 0.4260370814992664,
      "grad_norm": 0.07232563197612762,
      "learning_rate": 0.0009772138271568352,
      "loss": 1.5375,
      "step": 9582
    },
    {
      "epoch": 0.42612600595793876,
      "grad_norm": 0.07067085057497025,
      "learning_rate": 0.0009772032869292734,
      "loss": 1.5339,
      "step": 9584
    },
    {
      "epoch": 0.42621493041661107,
      "grad_norm": 0.06817600876092911,
      "learning_rate": 0.0009771927443213422,
      "loss": 1.5303,
      "step": 9586
    },
    {
      "epoch": 0.42630385487528344,
      "grad_norm": 0.06987394392490387,
      "learning_rate": 0.0009771821993330944,
      "loss": 1.5308,
      "step": 9588
    },
    {
      "epoch": 0.4263927793339558,
      "grad_norm": 0.06813080608844757,
      "learning_rate": 0.0009771716519645826,
      "loss": 1.5327,
      "step": 9590
    },
    {
      "epoch": 0.4264817037926282,
      "grad_norm": 0.06885334104299545,
      "learning_rate": 0.0009771611022158593,
      "loss": 1.5369,
      "step": 9592
    },
    {
      "epoch": 0.42657062825130054,
      "grad_norm": 0.06869365274906158,
      "learning_rate": 0.0009771505500869775,
      "loss": 1.541,
      "step": 9594
    },
    {
      "epoch": 0.42665955270997286,
      "grad_norm": 0.06896496564149857,
      "learning_rate": 0.0009771399955779893,
      "loss": 1.5353,
      "step": 9596
    },
    {
      "epoch": 0.4267484771686452,
      "grad_norm": 0.07151202112436295,
      "learning_rate": 0.0009771294386889478,
      "loss": 1.5341,
      "step": 9598
    },
    {
      "epoch": 0.4268374016273176,
      "grad_norm": 0.06725707650184631,
      "learning_rate": 0.0009771188794199053,
      "loss": 1.538,
      "step": 9600
    },
    {
      "epoch": 0.42692632608598996,
      "grad_norm": 0.07007326185703278,
      "learning_rate": 0.0009771083177709146,
      "loss": 1.5396,
      "step": 9602
    },
    {
      "epoch": 0.42701525054466233,
      "grad_norm": 0.06781750172376633,
      "learning_rate": 0.0009770977537420288,
      "loss": 1.5337,
      "step": 9604
    },
    {
      "epoch": 0.42710417500333464,
      "grad_norm": 0.07008969038724899,
      "learning_rate": 0.0009770871873332997,
      "loss": 1.5296,
      "step": 9606
    },
    {
      "epoch": 0.427193099462007,
      "grad_norm": 0.07271461933851242,
      "learning_rate": 0.0009770766185447808,
      "loss": 1.5362,
      "step": 9608
    },
    {
      "epoch": 0.4272820239206794,
      "grad_norm": 0.07218462973833084,
      "learning_rate": 0.0009770660473765245,
      "loss": 1.5313,
      "step": 9610
    },
    {
      "epoch": 0.42737094837935174,
      "grad_norm": 0.0696045383810997,
      "learning_rate": 0.0009770554738285835,
      "loss": 1.535,
      "step": 9612
    },
    {
      "epoch": 0.4274598728380241,
      "grad_norm": 0.0716555267572403,
      "learning_rate": 0.0009770448979010105,
      "loss": 1.5347,
      "step": 9614
    },
    {
      "epoch": 0.4275487972966965,
      "grad_norm": 0.06784503161907196,
      "learning_rate": 0.0009770343195938586,
      "loss": 1.5328,
      "step": 9616
    },
    {
      "epoch": 0.4276377217553688,
      "grad_norm": 0.0706615075469017,
      "learning_rate": 0.0009770237389071803,
      "loss": 1.5365,
      "step": 9618
    },
    {
      "epoch": 0.42772664621404116,
      "grad_norm": 0.07164894044399261,
      "learning_rate": 0.0009770131558410283,
      "loss": 1.5398,
      "step": 9620
    },
    {
      "epoch": 0.42781557067271353,
      "grad_norm": 0.07363253831863403,
      "learning_rate": 0.0009770025703954555,
      "loss": 1.5344,
      "step": 9622
    },
    {
      "epoch": 0.4279044951313859,
      "grad_norm": 0.07175491750240326,
      "learning_rate": 0.0009769919825705147,
      "loss": 1.5304,
      "step": 9624
    },
    {
      "epoch": 0.42799341959005827,
      "grad_norm": 0.07303240895271301,
      "learning_rate": 0.0009769813923662589,
      "loss": 1.5402,
      "step": 9626
    },
    {
      "epoch": 0.4280823440487306,
      "grad_norm": 0.06884127110242844,
      "learning_rate": 0.0009769707997827404,
      "loss": 1.5313,
      "step": 9628
    },
    {
      "epoch": 0.42817126850740295,
      "grad_norm": 0.06890305131673813,
      "learning_rate": 0.0009769602048200128,
      "loss": 1.5369,
      "step": 9630
    },
    {
      "epoch": 0.4282601929660753,
      "grad_norm": 0.07173550128936768,
      "learning_rate": 0.0009769496074781283,
      "loss": 1.5283,
      "step": 9632
    },
    {
      "epoch": 0.4283491174247477,
      "grad_norm": 0.068869449198246,
      "learning_rate": 0.0009769390077571398,
      "loss": 1.5363,
      "step": 9634
    },
    {
      "epoch": 0.42843804188342005,
      "grad_norm": 0.0685199499130249,
      "learning_rate": 0.0009769284056571005,
      "loss": 1.5358,
      "step": 9636
    },
    {
      "epoch": 0.4285269663420924,
      "grad_norm": 0.06906317174434662,
      "learning_rate": 0.0009769178011780632,
      "loss": 1.5338,
      "step": 9638
    },
    {
      "epoch": 0.42861589080076473,
      "grad_norm": 0.06904531270265579,
      "learning_rate": 0.0009769071943200808,
      "loss": 1.5367,
      "step": 9640
    },
    {
      "epoch": 0.4287048152594371,
      "grad_norm": 0.0690319761633873,
      "learning_rate": 0.0009768965850832062,
      "loss": 1.5332,
      "step": 9642
    },
    {
      "epoch": 0.42879373971810947,
      "grad_norm": 0.07049769908189774,
      "learning_rate": 0.0009768859734674922,
      "loss": 1.5366,
      "step": 9644
    },
    {
      "epoch": 0.42888266417678184,
      "grad_norm": 0.07084295898675919,
      "learning_rate": 0.0009768753594729918,
      "loss": 1.536,
      "step": 9646
    },
    {
      "epoch": 0.4289715886354542,
      "grad_norm": 0.06893621385097504,
      "learning_rate": 0.0009768647430997578,
      "loss": 1.5285,
      "step": 9648
    },
    {
      "epoch": 0.4290605130941265,
      "grad_norm": 0.07064125686883926,
      "learning_rate": 0.0009768541243478435,
      "loss": 1.53,
      "step": 9650
    },
    {
      "epoch": 0.4291494375527989,
      "grad_norm": 0.07157714664936066,
      "learning_rate": 0.0009768435032173016,
      "loss": 1.5316,
      "step": 9652
    },
    {
      "epoch": 0.42923836201147125,
      "grad_norm": 0.07000470906496048,
      "learning_rate": 0.0009768328797081852,
      "loss": 1.5395,
      "step": 9654
    },
    {
      "epoch": 0.4293272864701436,
      "grad_norm": 0.06912733614444733,
      "learning_rate": 0.000976822253820547,
      "loss": 1.5349,
      "step": 9656
    },
    {
      "epoch": 0.429416210928816,
      "grad_norm": 0.06659382581710815,
      "learning_rate": 0.0009768116255544407,
      "loss": 1.5309,
      "step": 9658
    },
    {
      "epoch": 0.4295051353874883,
      "grad_norm": 0.06706222891807556,
      "learning_rate": 0.0009768009949099184,
      "loss": 1.5337,
      "step": 9660
    },
    {
      "epoch": 0.42959405984616067,
      "grad_norm": 0.07012798637151718,
      "learning_rate": 0.0009767903618870337,
      "loss": 1.5429,
      "step": 9662
    },
    {
      "epoch": 0.42968298430483304,
      "grad_norm": 0.06715415418148041,
      "learning_rate": 0.0009767797264858397,
      "loss": 1.5405,
      "step": 9664
    },
    {
      "epoch": 0.4297719087635054,
      "grad_norm": 0.0690038651227951,
      "learning_rate": 0.0009767690887063894,
      "loss": 1.5364,
      "step": 9666
    },
    {
      "epoch": 0.4298608332221778,
      "grad_norm": 0.06869196146726608,
      "learning_rate": 0.0009767584485487356,
      "loss": 1.5336,
      "step": 9668
    },
    {
      "epoch": 0.42994975768085014,
      "grad_norm": 0.07014153897762299,
      "learning_rate": 0.0009767478060129313,
      "loss": 1.5349,
      "step": 9670
    },
    {
      "epoch": 0.43003868213952245,
      "grad_norm": 0.06871045380830765,
      "learning_rate": 0.00097673716109903,
      "loss": 1.5348,
      "step": 9672
    },
    {
      "epoch": 0.4301276065981948,
      "grad_norm": 0.06930448859930038,
      "learning_rate": 0.0009767265138070846,
      "loss": 1.5296,
      "step": 9674
    },
    {
      "epoch": 0.4302165310568672,
      "grad_norm": 0.07054897397756577,
      "learning_rate": 0.0009767158641371483,
      "loss": 1.5402,
      "step": 9676
    },
    {
      "epoch": 0.43030545551553956,
      "grad_norm": 0.07136155664920807,
      "learning_rate": 0.0009767052120892741,
      "loss": 1.5364,
      "step": 9678
    },
    {
      "epoch": 0.4303943799742119,
      "grad_norm": 0.06936744600534439,
      "learning_rate": 0.0009766945576635151,
      "loss": 1.5287,
      "step": 9680
    },
    {
      "epoch": 0.43048330443288424,
      "grad_norm": 0.07328307628631592,
      "learning_rate": 0.0009766839008599245,
      "loss": 1.5266,
      "step": 9682
    },
    {
      "epoch": 0.4305722288915566,
      "grad_norm": 0.06779733300209045,
      "learning_rate": 0.0009766732416785556,
      "loss": 1.5374,
      "step": 9684
    },
    {
      "epoch": 0.430661153350229,
      "grad_norm": 0.07060229033231735,
      "learning_rate": 0.0009766625801194613,
      "loss": 1.5293,
      "step": 9686
    },
    {
      "epoch": 0.43075007780890134,
      "grad_norm": 0.06920021772384644,
      "learning_rate": 0.000976651916182695,
      "loss": 1.5357,
      "step": 9688
    },
    {
      "epoch": 0.4308390022675737,
      "grad_norm": 0.0714344009757042,
      "learning_rate": 0.0009766412498683097,
      "loss": 1.536,
      "step": 9690
    },
    {
      "epoch": 0.4309279267262461,
      "grad_norm": 0.06918825954198837,
      "learning_rate": 0.000976630581176359,
      "loss": 1.5312,
      "step": 9692
    },
    {
      "epoch": 0.4310168511849184,
      "grad_norm": 0.07354065775871277,
      "learning_rate": 0.0009766199101068956,
      "loss": 1.5333,
      "step": 9694
    },
    {
      "epoch": 0.43110577564359076,
      "grad_norm": 0.07036007940769196,
      "learning_rate": 0.0009766092366599731,
      "loss": 1.5393,
      "step": 9696
    },
    {
      "epoch": 0.43119470010226313,
      "grad_norm": 0.07394345849752426,
      "learning_rate": 0.0009765985608356446,
      "loss": 1.5397,
      "step": 9698
    },
    {
      "epoch": 0.4312836245609355,
      "grad_norm": 0.07076235115528107,
      "learning_rate": 0.0009765878826339634,
      "loss": 1.5343,
      "step": 9700
    },
    {
      "epoch": 0.43137254901960786,
      "grad_norm": 0.07146693021059036,
      "learning_rate": 0.0009765772020549827,
      "loss": 1.5359,
      "step": 9702
    },
    {
      "epoch": 0.4314614734782802,
      "grad_norm": 0.06926552951335907,
      "learning_rate": 0.0009765665190987558,
      "loss": 1.5294,
      "step": 9704
    },
    {
      "epoch": 0.43155039793695255,
      "grad_norm": 0.07245678454637527,
      "learning_rate": 0.000976555833765336,
      "loss": 1.5303,
      "step": 9706
    },
    {
      "epoch": 0.4316393223956249,
      "grad_norm": 0.06824632734060287,
      "learning_rate": 0.0009765451460547766,
      "loss": 1.5339,
      "step": 9708
    },
    {
      "epoch": 0.4317282468542973,
      "grad_norm": 0.06656771898269653,
      "learning_rate": 0.0009765344559671307,
      "loss": 1.5326,
      "step": 9710
    },
    {
      "epoch": 0.43181717131296965,
      "grad_norm": 0.0666504055261612,
      "learning_rate": 0.0009765237635024522,
      "loss": 1.5307,
      "step": 9712
    },
    {
      "epoch": 0.43190609577164196,
      "grad_norm": 0.06956163048744202,
      "learning_rate": 0.0009765130686607938,
      "loss": 1.5284,
      "step": 9714
    },
    {
      "epoch": 0.43199502023031433,
      "grad_norm": 0.06977500021457672,
      "learning_rate": 0.0009765023714422092,
      "loss": 1.5358,
      "step": 9716
    },
    {
      "epoch": 0.4320839446889867,
      "grad_norm": 0.06824065744876862,
      "learning_rate": 0.0009764916718467517,
      "loss": 1.5359,
      "step": 9718
    },
    {
      "epoch": 0.43217286914765907,
      "grad_norm": 0.06885942071676254,
      "learning_rate": 0.0009764809698744746,
      "loss": 1.5375,
      "step": 9720
    },
    {
      "epoch": 0.43226179360633143,
      "grad_norm": 0.06814990937709808,
      "learning_rate": 0.0009764702655254314,
      "loss": 1.5334,
      "step": 9722
    },
    {
      "epoch": 0.4323507180650038,
      "grad_norm": 0.07028649002313614,
      "learning_rate": 0.0009764595587996754,
      "loss": 1.5322,
      "step": 9724
    },
    {
      "epoch": 0.4324396425236761,
      "grad_norm": 0.06551536917686462,
      "learning_rate": 0.00097644884969726,
      "loss": 1.5346,
      "step": 9726
    },
    {
      "epoch": 0.4325285669823485,
      "grad_norm": 0.06867454200983047,
      "learning_rate": 0.0009764381382182387,
      "loss": 1.5386,
      "step": 9728
    },
    {
      "epoch": 0.43261749144102085,
      "grad_norm": 0.065100759267807,
      "learning_rate": 0.0009764274243626649,
      "loss": 1.5314,
      "step": 9730
    },
    {
      "epoch": 0.4327064158996932,
      "grad_norm": 0.0707404837012291,
      "learning_rate": 0.000976416708130592,
      "loss": 1.5359,
      "step": 9732
    },
    {
      "epoch": 0.4327953403583656,
      "grad_norm": 0.0689748004078865,
      "learning_rate": 0.0009764059895220734,
      "loss": 1.5353,
      "step": 9734
    },
    {
      "epoch": 0.4328842648170379,
      "grad_norm": 0.06782509386539459,
      "learning_rate": 0.0009763952685371627,
      "loss": 1.5366,
      "step": 9736
    },
    {
      "epoch": 0.43297318927571027,
      "grad_norm": 0.07140477746725082,
      "learning_rate": 0.0009763845451759133,
      "loss": 1.5278,
      "step": 9738
    },
    {
      "epoch": 0.43306211373438264,
      "grad_norm": 0.06992009282112122,
      "learning_rate": 0.0009763738194383787,
      "loss": 1.533,
      "step": 9740
    },
    {
      "epoch": 0.433151038193055,
      "grad_norm": 0.07231522351503372,
      "learning_rate": 0.0009763630913246124,
      "loss": 1.5334,
      "step": 9742
    },
    {
      "epoch": 0.4332399626517274,
      "grad_norm": 0.07116207480430603,
      "learning_rate": 0.000976352360834668,
      "loss": 1.5301,
      "step": 9744
    },
    {
      "epoch": 0.43332888711039974,
      "grad_norm": 0.06921789795160294,
      "learning_rate": 0.000976341627968599,
      "loss": 1.5333,
      "step": 9746
    },
    {
      "epoch": 0.43341781156907205,
      "grad_norm": 0.06795500218868256,
      "learning_rate": 0.0009763308927264588,
      "loss": 1.5253,
      "step": 9748
    },
    {
      "epoch": 0.4335067360277444,
      "grad_norm": 0.06932450830936432,
      "learning_rate": 0.0009763201551083011,
      "loss": 1.5293,
      "step": 9750
    },
    {
      "epoch": 0.4335956604864168,
      "grad_norm": 0.07011115550994873,
      "learning_rate": 0.0009763094151141793,
      "loss": 1.5288,
      "step": 9752
    },
    {
      "epoch": 0.43368458494508916,
      "grad_norm": 0.06874189525842667,
      "learning_rate": 0.0009762986727441472,
      "loss": 1.5296,
      "step": 9754
    },
    {
      "epoch": 0.4337735094037615,
      "grad_norm": 0.06724654138088226,
      "learning_rate": 0.0009762879279982582,
      "loss": 1.5354,
      "step": 9756
    },
    {
      "epoch": 0.43386243386243384,
      "grad_norm": 0.06827713549137115,
      "learning_rate": 0.0009762771808765659,
      "loss": 1.5306,
      "step": 9758
    },
    {
      "epoch": 0.4339513583211062,
      "grad_norm": 0.06977761536836624,
      "learning_rate": 0.0009762664313791241,
      "loss": 1.5323,
      "step": 9760
    },
    {
      "epoch": 0.4340402827797786,
      "grad_norm": 0.0717058926820755,
      "learning_rate": 0.0009762556795059862,
      "loss": 1.5324,
      "step": 9762
    },
    {
      "epoch": 0.43412920723845094,
      "grad_norm": 0.06921355426311493,
      "learning_rate": 0.0009762449252572058,
      "loss": 1.5318,
      "step": 9764
    },
    {
      "epoch": 0.4342181316971233,
      "grad_norm": 0.0721471756696701,
      "learning_rate": 0.0009762341686328368,
      "loss": 1.5354,
      "step": 9766
    },
    {
      "epoch": 0.4343070561557957,
      "grad_norm": 0.06916612386703491,
      "learning_rate": 0.0009762234096329327,
      "loss": 1.531,
      "step": 9768
    },
    {
      "epoch": 0.434395980614468,
      "grad_norm": 0.07326947152614594,
      "learning_rate": 0.0009762126482575473,
      "loss": 1.5348,
      "step": 9770
    },
    {
      "epoch": 0.43448490507314036,
      "grad_norm": 0.07012344151735306,
      "learning_rate": 0.000976201884506734,
      "loss": 1.5335,
      "step": 9772
    },
    {
      "epoch": 0.4345738295318127,
      "grad_norm": 0.07134021073579788,
      "learning_rate": 0.0009761911183805466,
      "loss": 1.534,
      "step": 9774
    },
    {
      "epoch": 0.4346627539904851,
      "grad_norm": 0.06965865939855576,
      "learning_rate": 0.0009761803498790389,
      "loss": 1.5304,
      "step": 9776
    },
    {
      "epoch": 0.43475167844915746,
      "grad_norm": 0.06988412886857986,
      "learning_rate": 0.0009761695790022647,
      "loss": 1.5289,
      "step": 9778
    },
    {
      "epoch": 0.4348406029078298,
      "grad_norm": 0.0701071098446846,
      "learning_rate": 0.0009761588057502775,
      "loss": 1.5378,
      "step": 9780
    },
    {
      "epoch": 0.43492952736650214,
      "grad_norm": 0.07131971418857574,
      "learning_rate": 0.0009761480301231311,
      "loss": 1.5371,
      "step": 9782
    },
    {
      "epoch": 0.4350184518251745,
      "grad_norm": 0.06863526254892349,
      "learning_rate": 0.0009761372521208795,
      "loss": 1.5313,
      "step": 9784
    },
    {
      "epoch": 0.4351073762838469,
      "grad_norm": 0.06877754628658295,
      "learning_rate": 0.0009761264717435761,
      "loss": 1.5346,
      "step": 9786
    },
    {
      "epoch": 0.43519630074251925,
      "grad_norm": 0.06825356185436249,
      "learning_rate": 0.0009761156889912748,
      "loss": 1.5348,
      "step": 9788
    },
    {
      "epoch": 0.43528522520119156,
      "grad_norm": 0.06907788664102554,
      "learning_rate": 0.0009761049038640295,
      "loss": 1.5285,
      "step": 9790
    },
    {
      "epoch": 0.43537414965986393,
      "grad_norm": 0.06589856743812561,
      "learning_rate": 0.0009760941163618937,
      "loss": 1.5279,
      "step": 9792
    },
    {
      "epoch": 0.4354630741185363,
      "grad_norm": 0.06974221765995026,
      "learning_rate": 0.0009760833264849215,
      "loss": 1.5368,
      "step": 9794
    },
    {
      "epoch": 0.43555199857720867,
      "grad_norm": 0.06818997859954834,
      "learning_rate": 0.0009760725342331668,
      "loss": 1.5287,
      "step": 9796
    },
    {
      "epoch": 0.43564092303588103,
      "grad_norm": 0.0677684098482132,
      "learning_rate": 0.0009760617396066834,
      "loss": 1.5346,
      "step": 9798
    },
    {
      "epoch": 0.4357298474945534,
      "grad_norm": 0.06733290106058121,
      "learning_rate": 0.0009760509426055247,
      "loss": 1.5316,
      "step": 9800
    },
    {
      "epoch": 0.4358187719532257,
      "grad_norm": 0.06740870326757431,
      "learning_rate": 0.000976040143229745,
      "loss": 1.5319,
      "step": 9802
    },
    {
      "epoch": 0.4359076964118981,
      "grad_norm": 0.06795347481966019,
      "learning_rate": 0.0009760293414793979,
      "loss": 1.5353,
      "step": 9804
    },
    {
      "epoch": 0.43599662087057045,
      "grad_norm": 0.0665903314948082,
      "learning_rate": 0.0009760185373545376,
      "loss": 1.5337,
      "step": 9806
    },
    {
      "epoch": 0.4360855453292428,
      "grad_norm": 0.06611596792936325,
      "learning_rate": 0.0009760077308552178,
      "loss": 1.5304,
      "step": 9808
    },
    {
      "epoch": 0.4361744697879152,
      "grad_norm": 0.06896202266216278,
      "learning_rate": 0.0009759969219814924,
      "loss": 1.53,
      "step": 9810
    },
    {
      "epoch": 0.4362633942465875,
      "grad_norm": 0.06539107859134674,
      "learning_rate": 0.0009759861107334154,
      "loss": 1.5292,
      "step": 9812
    },
    {
      "epoch": 0.43635231870525987,
      "grad_norm": 0.07421990483999252,
      "learning_rate": 0.0009759752971110407,
      "loss": 1.5396,
      "step": 9814
    },
    {
      "epoch": 0.43644124316393224,
      "grad_norm": 0.06868709623813629,
      "learning_rate": 0.000975964481114422,
      "loss": 1.5387,
      "step": 9816
    },
    {
      "epoch": 0.4365301676226046,
      "grad_norm": 0.07227539271116257,
      "learning_rate": 0.0009759536627436137,
      "loss": 1.54,
      "step": 9818
    },
    {
      "epoch": 0.43661909208127697,
      "grad_norm": 0.06839174777269363,
      "learning_rate": 0.0009759428419986693,
      "loss": 1.5323,
      "step": 9820
    },
    {
      "epoch": 0.43670801653994934,
      "grad_norm": 0.07100740075111389,
      "learning_rate": 0.0009759320188796432,
      "loss": 1.5304,
      "step": 9822
    },
    {
      "epoch": 0.43679694099862165,
      "grad_norm": 0.07037629187107086,
      "learning_rate": 0.000975921193386589,
      "loss": 1.5327,
      "step": 9824
    },
    {
      "epoch": 0.436885865457294,
      "grad_norm": 0.07322507351636887,
      "learning_rate": 0.000975910365519561,
      "loss": 1.5359,
      "step": 9826
    },
    {
      "epoch": 0.4369747899159664,
      "grad_norm": 0.06714385747909546,
      "learning_rate": 0.0009758995352786131,
      "loss": 1.5287,
      "step": 9828
    },
    {
      "epoch": 0.43706371437463876,
      "grad_norm": 0.06822887063026428,
      "learning_rate": 0.0009758887026637994,
      "loss": 1.5342,
      "step": 9830
    },
    {
      "epoch": 0.4371526388333111,
      "grad_norm": 0.0696975588798523,
      "learning_rate": 0.0009758778676751737,
      "loss": 1.5314,
      "step": 9832
    },
    {
      "epoch": 0.43724156329198344,
      "grad_norm": 0.07337722182273865,
      "learning_rate": 0.0009758670303127903,
      "loss": 1.5383,
      "step": 9834
    },
    {
      "epoch": 0.4373304877506558,
      "grad_norm": 0.07148087024688721,
      "learning_rate": 0.0009758561905767032,
      "loss": 1.5315,
      "step": 9836
    },
    {
      "epoch": 0.4374194122093282,
      "grad_norm": 0.06952288746833801,
      "learning_rate": 0.0009758453484669663,
      "loss": 1.5327,
      "step": 9838
    },
    {
      "epoch": 0.43750833666800054,
      "grad_norm": 0.07030414789915085,
      "learning_rate": 0.000975834503983634,
      "loss": 1.5458,
      "step": 9840
    },
    {
      "epoch": 0.4375972611266729,
      "grad_norm": 0.06890011578798294,
      "learning_rate": 0.00097582365712676,
      "loss": 1.5377,
      "step": 9842
    },
    {
      "epoch": 0.4376861855853452,
      "grad_norm": 0.06955260038375854,
      "learning_rate": 0.0009758128078963988,
      "loss": 1.5374,
      "step": 9844
    },
    {
      "epoch": 0.4377751100440176,
      "grad_norm": 0.0680069625377655,
      "learning_rate": 0.0009758019562926041,
      "loss": 1.5341,
      "step": 9846
    },
    {
      "epoch": 0.43786403450268996,
      "grad_norm": 0.06949877738952637,
      "learning_rate": 0.0009757911023154305,
      "loss": 1.5361,
      "step": 9848
    },
    {
      "epoch": 0.4379529589613623,
      "grad_norm": 0.0744439959526062,
      "learning_rate": 0.0009757802459649319,
      "loss": 1.5328,
      "step": 9850
    },
    {
      "epoch": 0.4380418834200347,
      "grad_norm": 0.07167305797338486,
      "learning_rate": 0.0009757693872411622,
      "loss": 1.5318,
      "step": 9852
    },
    {
      "epoch": 0.43813080787870706,
      "grad_norm": 0.07135527580976486,
      "learning_rate": 0.0009757585261441762,
      "loss": 1.5315,
      "step": 9854
    },
    {
      "epoch": 0.4382197323373794,
      "grad_norm": 0.0703761875629425,
      "learning_rate": 0.0009757476626740274,
      "loss": 1.5339,
      "step": 9856
    },
    {
      "epoch": 0.43830865679605174,
      "grad_norm": 0.06967216730117798,
      "learning_rate": 0.0009757367968307705,
      "loss": 1.5276,
      "step": 9858
    },
    {
      "epoch": 0.4383975812547241,
      "grad_norm": 0.06713514029979706,
      "learning_rate": 0.0009757259286144593,
      "loss": 1.5282,
      "step": 9860
    },
    {
      "epoch": 0.4384865057133965,
      "grad_norm": 0.07148092240095139,
      "learning_rate": 0.0009757150580251481,
      "loss": 1.5397,
      "step": 9862
    },
    {
      "epoch": 0.43857543017206885,
      "grad_norm": 0.07046201080083847,
      "learning_rate": 0.0009757041850628915,
      "loss": 1.5359,
      "step": 9864
    },
    {
      "epoch": 0.43866435463074116,
      "grad_norm": 0.06848791241645813,
      "learning_rate": 0.0009756933097277434,
      "loss": 1.5323,
      "step": 9866
    },
    {
      "epoch": 0.43875327908941353,
      "grad_norm": 0.06691039353609085,
      "learning_rate": 0.000975682432019758,
      "loss": 1.5337,
      "step": 9868
    },
    {
      "epoch": 0.4388422035480859,
      "grad_norm": 0.06878488510847092,
      "learning_rate": 0.0009756715519389899,
      "loss": 1.5321,
      "step": 9870
    },
    {
      "epoch": 0.43893112800675826,
      "grad_norm": 0.06844276189804077,
      "learning_rate": 0.0009756606694854928,
      "loss": 1.5288,
      "step": 9872
    },
    {
      "epoch": 0.43902005246543063,
      "grad_norm": 0.07109802216291428,
      "learning_rate": 0.0009756497846593215,
      "loss": 1.5309,
      "step": 9874
    },
    {
      "epoch": 0.439108976924103,
      "grad_norm": 0.06382738053798676,
      "learning_rate": 0.0009756388974605302,
      "loss": 1.5289,
      "step": 9876
    },
    {
      "epoch": 0.4391979013827753,
      "grad_norm": 0.06882934272289276,
      "learning_rate": 0.000975628007889173,
      "loss": 1.5332,
      "step": 9878
    },
    {
      "epoch": 0.4392868258414477,
      "grad_norm": 0.06806602329015732,
      "learning_rate": 0.0009756171159453045,
      "loss": 1.5313,
      "step": 9880
    },
    {
      "epoch": 0.43937575030012005,
      "grad_norm": 0.06839878112077713,
      "learning_rate": 0.0009756062216289787,
      "loss": 1.5379,
      "step": 9882
    },
    {
      "epoch": 0.4394646747587924,
      "grad_norm": 0.0689515769481659,
      "learning_rate": 0.0009755953249402503,
      "loss": 1.5344,
      "step": 9884
    },
    {
      "epoch": 0.4395535992174648,
      "grad_norm": 0.06822851300239563,
      "learning_rate": 0.0009755844258791733,
      "loss": 1.5353,
      "step": 9886
    },
    {
      "epoch": 0.4396425236761371,
      "grad_norm": 0.06764476001262665,
      "learning_rate": 0.0009755735244458024,
      "loss": 1.5303,
      "step": 9888
    },
    {
      "epoch": 0.43973144813480947,
      "grad_norm": 0.06747211515903473,
      "learning_rate": 0.0009755626206401917,
      "loss": 1.5273,
      "step": 9890
    },
    {
      "epoch": 0.43982037259348183,
      "grad_norm": 0.06862806528806686,
      "learning_rate": 0.0009755517144623958,
      "loss": 1.5317,
      "step": 9892
    },
    {
      "epoch": 0.4399092970521542,
      "grad_norm": 0.06655488908290863,
      "learning_rate": 0.0009755408059124688,
      "loss": 1.5291,
      "step": 9894
    },
    {
      "epoch": 0.43999822151082657,
      "grad_norm": 0.06658562272787094,
      "learning_rate": 0.0009755298949904655,
      "loss": 1.5282,
      "step": 9896
    },
    {
      "epoch": 0.4400871459694989,
      "grad_norm": 0.06817793101072311,
      "learning_rate": 0.0009755189816964402,
      "loss": 1.5301,
      "step": 9898
    },
    {
      "epoch": 0.44017607042817125,
      "grad_norm": 0.07138610631227493,
      "learning_rate": 0.0009755080660304472,
      "loss": 1.5329,
      "step": 9900
    },
    {
      "epoch": 0.4402649948868436,
      "grad_norm": 0.06737557798624039,
      "learning_rate": 0.0009754971479925409,
      "loss": 1.5357,
      "step": 9902
    },
    {
      "epoch": 0.440353919345516,
      "grad_norm": 0.06973376125097275,
      "learning_rate": 0.0009754862275827762,
      "loss": 1.5267,
      "step": 9904
    },
    {
      "epoch": 0.44044284380418836,
      "grad_norm": 0.06866409629583359,
      "learning_rate": 0.000975475304801207,
      "loss": 1.5338,
      "step": 9906
    },
    {
      "epoch": 0.4405317682628607,
      "grad_norm": 0.06903880089521408,
      "learning_rate": 0.0009754643796478882,
      "loss": 1.5376,
      "step": 9908
    },
    {
      "epoch": 0.44062069272153304,
      "grad_norm": 0.06750325113534927,
      "learning_rate": 0.0009754534521228742,
      "loss": 1.5331,
      "step": 9910
    },
    {
      "epoch": 0.4407096171802054,
      "grad_norm": 0.0681682899594307,
      "learning_rate": 0.0009754425222262193,
      "loss": 1.535,
      "step": 9912
    },
    {
      "epoch": 0.44079854163887777,
      "grad_norm": 0.06802817434072495,
      "learning_rate": 0.0009754315899579783,
      "loss": 1.53,
      "step": 9914
    },
    {
      "epoch": 0.44088746609755014,
      "grad_norm": 0.06679469347000122,
      "learning_rate": 0.0009754206553182057,
      "loss": 1.5295,
      "step": 9916
    },
    {
      "epoch": 0.4409763905562225,
      "grad_norm": 0.06789480149745941,
      "learning_rate": 0.0009754097183069557,
      "loss": 1.5327,
      "step": 9918
    },
    {
      "epoch": 0.4410653150148948,
      "grad_norm": 0.07148660719394684,
      "learning_rate": 0.0009753987789242833,
      "loss": 1.5376,
      "step": 9920
    },
    {
      "epoch": 0.4411542394735672,
      "grad_norm": 0.07011796534061432,
      "learning_rate": 0.0009753878371702427,
      "loss": 1.5291,
      "step": 9922
    },
    {
      "epoch": 0.44124316393223956,
      "grad_norm": 0.07079441100358963,
      "learning_rate": 0.0009753768930448888,
      "loss": 1.5317,
      "step": 9924
    },
    {
      "epoch": 0.4413320883909119,
      "grad_norm": 0.07116356492042542,
      "learning_rate": 0.000975365946548276,
      "loss": 1.5337,
      "step": 9926
    },
    {
      "epoch": 0.4414210128495843,
      "grad_norm": 0.07054295390844345,
      "learning_rate": 0.000975354997680459,
      "loss": 1.5309,
      "step": 9928
    },
    {
      "epoch": 0.44150993730825666,
      "grad_norm": 0.06715717166662216,
      "learning_rate": 0.0009753440464414924,
      "loss": 1.5324,
      "step": 9930
    },
    {
      "epoch": 0.441598861766929,
      "grad_norm": 0.06757137924432755,
      "learning_rate": 0.0009753330928314308,
      "loss": 1.5335,
      "step": 9932
    },
    {
      "epoch": 0.44168778622560134,
      "grad_norm": 0.06650960445404053,
      "learning_rate": 0.0009753221368503286,
      "loss": 1.5276,
      "step": 9934
    },
    {
      "epoch": 0.4417767106842737,
      "grad_norm": 0.06757502257823944,
      "learning_rate": 0.0009753111784982409,
      "loss": 1.5326,
      "step": 9936
    },
    {
      "epoch": 0.4418656351429461,
      "grad_norm": 0.07040387392044067,
      "learning_rate": 0.0009753002177752221,
      "loss": 1.5321,
      "step": 9938
    },
    {
      "epoch": 0.44195455960161845,
      "grad_norm": 0.06907001882791519,
      "learning_rate": 0.000975289254681327,
      "loss": 1.5358,
      "step": 9940
    },
    {
      "epoch": 0.44204348406029076,
      "grad_norm": 0.06722556799650192,
      "learning_rate": 0.0009752782892166101,
      "loss": 1.5388,
      "step": 9942
    },
    {
      "epoch": 0.4421324085189631,
      "grad_norm": 0.06660650670528412,
      "learning_rate": 0.0009752673213811263,
      "loss": 1.5229,
      "step": 9944
    },
    {
      "epoch": 0.4422213329776355,
      "grad_norm": 0.0699448511004448,
      "learning_rate": 0.00097525635117493,
      "loss": 1.5271,
      "step": 9946
    },
    {
      "epoch": 0.44231025743630786,
      "grad_norm": 0.07192565500736237,
      "learning_rate": 0.0009752453785980763,
      "loss": 1.5317,
      "step": 9948
    },
    {
      "epoch": 0.44239918189498023,
      "grad_norm": 0.06789202988147736,
      "learning_rate": 0.0009752344036506197,
      "loss": 1.5254,
      "step": 9950
    },
    {
      "epoch": 0.4424881063536526,
      "grad_norm": 0.06763622164726257,
      "learning_rate": 0.000975223426332615,
      "loss": 1.5357,
      "step": 9952
    },
    {
      "epoch": 0.4425770308123249,
      "grad_norm": 0.0687379539012909,
      "learning_rate": 0.000975212446644117,
      "loss": 1.5316,
      "step": 9954
    },
    {
      "epoch": 0.4426659552709973,
      "grad_norm": 0.06635449826717377,
      "learning_rate": 0.0009752014645851805,
      "loss": 1.5354,
      "step": 9956
    },
    {
      "epoch": 0.44275487972966965,
      "grad_norm": 0.07067475467920303,
      "learning_rate": 0.0009751904801558602,
      "loss": 1.5283,
      "step": 9958
    },
    {
      "epoch": 0.442843804188342,
      "grad_norm": 0.07415599375963211,
      "learning_rate": 0.0009751794933562108,
      "loss": 1.534,
      "step": 9960
    },
    {
      "epoch": 0.4429327286470144,
      "grad_norm": 0.06655219942331314,
      "learning_rate": 0.0009751685041862872,
      "loss": 1.5341,
      "step": 9962
    },
    {
      "epoch": 0.4430216531056867,
      "grad_norm": 0.07229764014482498,
      "learning_rate": 0.0009751575126461443,
      "loss": 1.535,
      "step": 9964
    },
    {
      "epoch": 0.44311057756435906,
      "grad_norm": 0.06633423268795013,
      "learning_rate": 0.0009751465187358368,
      "loss": 1.5361,
      "step": 9966
    },
    {
      "epoch": 0.44319950202303143,
      "grad_norm": 0.07215137034654617,
      "learning_rate": 0.0009751355224554197,
      "loss": 1.5333,
      "step": 9968
    },
    {
      "epoch": 0.4432884264817038,
      "grad_norm": 0.06949516385793686,
      "learning_rate": 0.0009751245238049477,
      "loss": 1.53,
      "step": 9970
    },
    {
      "epoch": 0.44337735094037617,
      "grad_norm": 0.06816526502370834,
      "learning_rate": 0.0009751135227844758,
      "loss": 1.5338,
      "step": 9972
    },
    {
      "epoch": 0.4434662753990485,
      "grad_norm": 0.06912938505411148,
      "learning_rate": 0.0009751025193940586,
      "loss": 1.5347,
      "step": 9974
    },
    {
      "epoch": 0.44355519985772085,
      "grad_norm": 0.07131484895944595,
      "learning_rate": 0.0009750915136337513,
      "loss": 1.5317,
      "step": 9976
    },
    {
      "epoch": 0.4436441243163932,
      "grad_norm": 0.06642764806747437,
      "learning_rate": 0.0009750805055036086,
      "loss": 1.5271,
      "step": 9978
    },
    {
      "epoch": 0.4437330487750656,
      "grad_norm": 0.0668521523475647,
      "learning_rate": 0.0009750694950036855,
      "loss": 1.5312,
      "step": 9980
    },
    {
      "epoch": 0.44382197323373795,
      "grad_norm": 0.06732647120952606,
      "learning_rate": 0.0009750584821340369,
      "loss": 1.536,
      "step": 9982
    },
    {
      "epoch": 0.4439108976924103,
      "grad_norm": 0.06798292696475983,
      "learning_rate": 0.0009750474668947178,
      "loss": 1.5334,
      "step": 9984
    },
    {
      "epoch": 0.44399982215108263,
      "grad_norm": 0.06467396020889282,
      "learning_rate": 0.0009750364492857829,
      "loss": 1.532,
      "step": 9986
    },
    {
      "epoch": 0.444088746609755,
      "grad_norm": 0.06865821778774261,
      "learning_rate": 0.0009750254293072876,
      "loss": 1.5302,
      "step": 9988
    },
    {
      "epoch": 0.44417767106842737,
      "grad_norm": 0.06773124635219574,
      "learning_rate": 0.0009750144069592863,
      "loss": 1.5351,
      "step": 9990
    },
    {
      "epoch": 0.44426659552709974,
      "grad_norm": 0.06710581481456757,
      "learning_rate": 0.0009750033822418345,
      "loss": 1.5284,
      "step": 9992
    },
    {
      "epoch": 0.4443555199857721,
      "grad_norm": 0.06517042219638824,
      "learning_rate": 0.000974992355154987,
      "loss": 1.5325,
      "step": 9994
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.06951684504747391,
      "learning_rate": 0.0009749813256987987,
      "loss": 1.5345,
      "step": 9996
    },
    {
      "epoch": 0.4445333689031168,
      "grad_norm": 0.06696436554193497,
      "learning_rate": 0.0009749702938733247,
      "loss": 1.5336,
      "step": 9998
    },
    {
      "epoch": 0.44462229336178916,
      "grad_norm": 0.06807401776313782,
      "learning_rate": 0.00097495925967862,
      "loss": 1.5254,
      "step": 10000
    },
    {
      "epoch": 0.44462229336178916,
      "eval_loss": 1.5108540058135986,
      "eval_runtime": 12.3873,
      "eval_samples_per_second": 557.83,
      "eval_steps_per_second": 69.749,
      "step": 10000
    },
    {
      "epoch": 0.4447112178204615,
      "grad_norm": 0.07122799009084702,
      "learning_rate": 0.0006260291775788085,
      "loss": 1.5277,
      "step": 10002
    },
    {
      "epoch": 0.4448001422791339,
      "grad_norm": 0.06492965668439865,
      "learning_rate": 0.0006258877037424691,
      "loss": 1.5261,
      "step": 10004
    },
    {
      "epoch": 0.44488906673780626,
      "grad_norm": 0.06462486833333969,
      "learning_rate": 0.0006257462191446941,
      "loss": 1.5268,
      "step": 10006
    },
    {
      "epoch": 0.4449779911964786,
      "grad_norm": 0.06561804562807083,
      "learning_rate": 0.0006256047237975782,
      "loss": 1.5237,
      "step": 10008
    },
    {
      "epoch": 0.44506691565515094,
      "grad_norm": 0.06555454432964325,
      "learning_rate": 0.0006254632177132169,
      "loss": 1.5244,
      "step": 10010
    },
    {
      "epoch": 0.4451558401138233,
      "grad_norm": 0.06676924228668213,
      "learning_rate": 0.0006253217009037068,
      "loss": 1.5251,
      "step": 10012
    },
    {
      "epoch": 0.4452447645724957,
      "grad_norm": 0.06575804203748703,
      "learning_rate": 0.0006251801733811455,
      "loss": 1.5222,
      "step": 10014
    },
    {
      "epoch": 0.44533368903116805,
      "grad_norm": 0.06349354237318039,
      "learning_rate": 0.0006250386351576314,
      "loss": 1.5211,
      "step": 10016
    },
    {
      "epoch": 0.44542261348984036,
      "grad_norm": 0.06419903039932251,
      "learning_rate": 0.0006248970862452637,
      "loss": 1.5216,
      "step": 10018
    },
    {
      "epoch": 0.4455115379485127,
      "grad_norm": 0.06334590911865234,
      "learning_rate": 0.0006247555266561425,
      "loss": 1.5178,
      "step": 10020
    },
    {
      "epoch": 0.4456004624071851,
      "grad_norm": 0.06277309358119965,
      "learning_rate": 0.0006246139564023693,
      "loss": 1.5201,
      "step": 10022
    },
    {
      "epoch": 0.44568938686585746,
      "grad_norm": 0.06414992362260818,
      "learning_rate": 0.0006244723754960459,
      "loss": 1.5213,
      "step": 10024
    },
    {
      "epoch": 0.44577831132452983,
      "grad_norm": 0.06549572199583054,
      "learning_rate": 0.0006243307839492752,
      "loss": 1.5174,
      "step": 10026
    },
    {
      "epoch": 0.44586723578320214,
      "grad_norm": 0.0640779510140419,
      "learning_rate": 0.0006241891817741613,
      "loss": 1.5167,
      "step": 10028
    },
    {
      "epoch": 0.4459561602418745,
      "grad_norm": 0.06481896340847015,
      "learning_rate": 0.0006240475689828086,
      "loss": 1.5147,
      "step": 10030
    },
    {
      "epoch": 0.4460450847005469,
      "grad_norm": 0.06547030061483383,
      "learning_rate": 0.0006239059455873235,
      "loss": 1.5214,
      "step": 10032
    },
    {
      "epoch": 0.44613400915921925,
      "grad_norm": 0.061458081007003784,
      "learning_rate": 0.0006237643115998119,
      "loss": 1.5146,
      "step": 10034
    },
    {
      "epoch": 0.4462229336178916,
      "grad_norm": 0.06149187311530113,
      "learning_rate": 0.0006236226670323816,
      "loss": 1.5123,
      "step": 10036
    },
    {
      "epoch": 0.446311858076564,
      "grad_norm": 0.06510701775550842,
      "learning_rate": 0.0006234810118971408,
      "loss": 1.5149,
      "step": 10038
    },
    {
      "epoch": 0.4464007825352363,
      "grad_norm": 0.061700914055109024,
      "learning_rate": 0.0006233393462061989,
      "loss": 1.5157,
      "step": 10040
    },
    {
      "epoch": 0.44648970699390866,
      "grad_norm": 0.06388545781373978,
      "learning_rate": 0.0006231976699716664,
      "loss": 1.5221,
      "step": 10042
    },
    {
      "epoch": 0.44657863145258103,
      "grad_norm": 0.06275127828121185,
      "learning_rate": 0.0006230559832056539,
      "loss": 1.5054,
      "step": 10044
    },
    {
      "epoch": 0.4466675559112534,
      "grad_norm": 0.06381896138191223,
      "learning_rate": 0.0006229142859202739,
      "loss": 1.5148,
      "step": 10046
    },
    {
      "epoch": 0.44675648036992577,
      "grad_norm": 0.06251713633537292,
      "learning_rate": 0.0006227725781276389,
      "loss": 1.5142,
      "step": 10048
    },
    {
      "epoch": 0.4468454048285981,
      "grad_norm": 0.06425853818655014,
      "learning_rate": 0.000622630859839863,
      "loss": 1.5165,
      "step": 10050
    },
    {
      "epoch": 0.44693432928727045,
      "grad_norm": 0.06350740045309067,
      "learning_rate": 0.0006224891310690606,
      "loss": 1.5139,
      "step": 10052
    },
    {
      "epoch": 0.4470232537459428,
      "grad_norm": 0.06478399783372879,
      "learning_rate": 0.0006223473918273477,
      "loss": 1.5146,
      "step": 10054
    },
    {
      "epoch": 0.4471121782046152,
      "grad_norm": 0.06344152241945267,
      "learning_rate": 0.0006222056421268405,
      "loss": 1.5167,
      "step": 10056
    },
    {
      "epoch": 0.44720110266328755,
      "grad_norm": 0.06286389380693436,
      "learning_rate": 0.0006220638819796565,
      "loss": 1.5089,
      "step": 10058
    },
    {
      "epoch": 0.4472900271219599,
      "grad_norm": 0.06424534320831299,
      "learning_rate": 0.0006219221113979138,
      "loss": 1.5131,
      "step": 10060
    },
    {
      "epoch": 0.44737895158063223,
      "grad_norm": 0.061146993190050125,
      "learning_rate": 0.0006217803303937319,
      "loss": 1.5113,
      "step": 10062
    },
    {
      "epoch": 0.4474678760393046,
      "grad_norm": 0.06360126286745071,
      "learning_rate": 0.0006216385389792306,
      "loss": 1.5176,
      "step": 10064
    },
    {
      "epoch": 0.44755680049797697,
      "grad_norm": 0.06215475872159004,
      "learning_rate": 0.0006214967371665309,
      "loss": 1.5132,
      "step": 10066
    },
    {
      "epoch": 0.44764572495664934,
      "grad_norm": 0.06413223594427109,
      "learning_rate": 0.0006213549249677548,
      "loss": 1.5158,
      "step": 10068
    },
    {
      "epoch": 0.4477346494153217,
      "grad_norm": 0.06347139924764633,
      "learning_rate": 0.000621213102395025,
      "loss": 1.5089,
      "step": 10070
    },
    {
      "epoch": 0.447823573873994,
      "grad_norm": 0.06259430944919586,
      "learning_rate": 0.0006210712694604647,
      "loss": 1.5121,
      "step": 10072
    },
    {
      "epoch": 0.4479124983326664,
      "grad_norm": 0.06153608858585358,
      "learning_rate": 0.0006209294261761989,
      "loss": 1.5149,
      "step": 10074
    },
    {
      "epoch": 0.44800142279133875,
      "grad_norm": 0.06388126313686371,
      "learning_rate": 0.000620787572554353,
      "loss": 1.5123,
      "step": 10076
    },
    {
      "epoch": 0.4480903472500111,
      "grad_norm": 0.06113690882921219,
      "learning_rate": 0.0006206457086070531,
      "loss": 1.5124,
      "step": 10078
    },
    {
      "epoch": 0.4481792717086835,
      "grad_norm": 0.06311555206775665,
      "learning_rate": 0.000620503834346426,
      "loss": 1.5159,
      "step": 10080
    },
    {
      "epoch": 0.4482681961673558,
      "grad_norm": 0.06283438205718994,
      "learning_rate": 0.0006203619497846005,
      "loss": 1.5146,
      "step": 10082
    },
    {
      "epoch": 0.44835712062602817,
      "grad_norm": 0.06206725165247917,
      "learning_rate": 0.0006202200549337048,
      "loss": 1.5096,
      "step": 10084
    },
    {
      "epoch": 0.44844604508470054,
      "grad_norm": 0.062284741550683975,
      "learning_rate": 0.0006200781498058695,
      "loss": 1.5088,
      "step": 10086
    },
    {
      "epoch": 0.4485349695433729,
      "grad_norm": 0.06355787068605423,
      "learning_rate": 0.0006199362344132243,
      "loss": 1.5075,
      "step": 10088
    },
    {
      "epoch": 0.4486238940020453,
      "grad_norm": 0.06461120396852493,
      "learning_rate": 0.0006197943087679013,
      "loss": 1.5123,
      "step": 10090
    },
    {
      "epoch": 0.44871281846071764,
      "grad_norm": 0.06290825456380844,
      "learning_rate": 0.0006196523728820329,
      "loss": 1.5114,
      "step": 10092
    },
    {
      "epoch": 0.44880174291938996,
      "grad_norm": 0.06232891604304314,
      "learning_rate": 0.0006195104267677525,
      "loss": 1.5073,
      "step": 10094
    },
    {
      "epoch": 0.4488906673780623,
      "grad_norm": 0.06516814976930618,
      "learning_rate": 0.0006193684704371941,
      "loss": 1.519,
      "step": 10096
    },
    {
      "epoch": 0.4489795918367347,
      "grad_norm": 0.06370670348405838,
      "learning_rate": 0.0006192265039024928,
      "loss": 1.5129,
      "step": 10098
    },
    {
      "epoch": 0.44906851629540706,
      "grad_norm": 0.0628281906247139,
      "learning_rate": 0.0006190845271757846,
      "loss": 1.5132,
      "step": 10100
    },
    {
      "epoch": 0.44915744075407943,
      "grad_norm": 0.06411248445510864,
      "learning_rate": 0.0006189425402692061,
      "loss": 1.5139,
      "step": 10102
    },
    {
      "epoch": 0.44924636521275174,
      "grad_norm": 0.06423705816268921,
      "learning_rate": 0.0006188005431948953,
      "loss": 1.5087,
      "step": 10104
    },
    {
      "epoch": 0.4493352896714241,
      "grad_norm": 0.06506875157356262,
      "learning_rate": 0.0006186585359649903,
      "loss": 1.5149,
      "step": 10106
    },
    {
      "epoch": 0.4494242141300965,
      "grad_norm": 0.06523150950670242,
      "learning_rate": 0.0006185165185916308,
      "loss": 1.5154,
      "step": 10108
    },
    {
      "epoch": 0.44951313858876885,
      "grad_norm": 0.06264711171388626,
      "learning_rate": 0.0006183744910869571,
      "loss": 1.511,
      "step": 10110
    },
    {
      "epoch": 0.4496020630474412,
      "grad_norm": 0.06609227508306503,
      "learning_rate": 0.0006182324534631102,
      "loss": 1.5134,
      "step": 10112
    },
    {
      "epoch": 0.4496909875061136,
      "grad_norm": 0.06463871151208878,
      "learning_rate": 0.0006180904057322321,
      "loss": 1.5089,
      "step": 10114
    },
    {
      "epoch": 0.4497799119647859,
      "grad_norm": 0.06578138470649719,
      "learning_rate": 0.0006179483479064657,
      "loss": 1.5116,
      "step": 10116
    },
    {
      "epoch": 0.44986883642345826,
      "grad_norm": 0.0638619214296341,
      "learning_rate": 0.0006178062799979548,
      "loss": 1.5139,
      "step": 10118
    },
    {
      "epoch": 0.44995776088213063,
      "grad_norm": 0.06302259117364883,
      "learning_rate": 0.0006176642020188439,
      "loss": 1.5093,
      "step": 10120
    },
    {
      "epoch": 0.450046685340803,
      "grad_norm": 0.06259556859731674,
      "learning_rate": 0.0006175221139812784,
      "loss": 1.51,
      "step": 10122
    },
    {
      "epoch": 0.45013560979947537,
      "grad_norm": 0.06390849500894547,
      "learning_rate": 0.0006173800158974048,
      "loss": 1.5142,
      "step": 10124
    },
    {
      "epoch": 0.4502245342581477,
      "grad_norm": 0.06265709549188614,
      "learning_rate": 0.0006172379077793702,
      "loss": 1.5049,
      "step": 10126
    },
    {
      "epoch": 0.45031345871682005,
      "grad_norm": 0.061674658209085464,
      "learning_rate": 0.0006170957896393225,
      "loss": 1.5133,
      "step": 10128
    },
    {
      "epoch": 0.4504023831754924,
      "grad_norm": 0.06264129281044006,
      "learning_rate": 0.0006169536614894107,
      "loss": 1.5137,
      "step": 10130
    },
    {
      "epoch": 0.4504913076341648,
      "grad_norm": 0.06164534017443657,
      "learning_rate": 0.0006168115233417846,
      "loss": 1.5115,
      "step": 10132
    },
    {
      "epoch": 0.45058023209283715,
      "grad_norm": 0.06206073611974716,
      "learning_rate": 0.0006166693752085946,
      "loss": 1.5174,
      "step": 10134
    },
    {
      "epoch": 0.4506691565515095,
      "grad_norm": 0.06333418935537338,
      "learning_rate": 0.0006165272171019923,
      "loss": 1.5157,
      "step": 10136
    },
    {
      "epoch": 0.45075808101018183,
      "grad_norm": 0.06554005295038223,
      "learning_rate": 0.0006163850490341298,
      "loss": 1.5043,
      "step": 10138
    },
    {
      "epoch": 0.4508470054688542,
      "grad_norm": 0.06292907893657684,
      "learning_rate": 0.0006162428710171608,
      "loss": 1.5117,
      "step": 10140
    },
    {
      "epoch": 0.45093592992752657,
      "grad_norm": 0.06288234144449234,
      "learning_rate": 0.0006161006830632386,
      "loss": 1.5138,
      "step": 10142
    },
    {
      "epoch": 0.45102485438619894,
      "grad_norm": 0.06246441975235939,
      "learning_rate": 0.0006159584851845184,
      "loss": 1.5104,
      "step": 10144
    },
    {
      "epoch": 0.4511137788448713,
      "grad_norm": 0.062472037971019745,
      "learning_rate": 0.0006158162773931559,
      "loss": 1.5149,
      "step": 10146
    },
    {
      "epoch": 0.4512027033035436,
      "grad_norm": 0.06365121155977249,
      "learning_rate": 0.0006156740597013079,
      "loss": 1.5114,
      "step": 10148
    },
    {
      "epoch": 0.451291627762216,
      "grad_norm": 0.06463265419006348,
      "learning_rate": 0.0006155318321211312,
      "loss": 1.5189,
      "step": 10150
    },
    {
      "epoch": 0.45138055222088835,
      "grad_norm": 0.06455686688423157,
      "learning_rate": 0.0006153895946647845,
      "loss": 1.5085,
      "step": 10152
    },
    {
      "epoch": 0.4514694766795607,
      "grad_norm": 0.06434636563062668,
      "learning_rate": 0.0006152473473444265,
      "loss": 1.5093,
      "step": 10154
    },
    {
      "epoch": 0.4515584011382331,
      "grad_norm": 0.06507983803749084,
      "learning_rate": 0.0006151050901722177,
      "loss": 1.5084,
      "step": 10156
    },
    {
      "epoch": 0.4516473255969054,
      "grad_norm": 0.0638333186507225,
      "learning_rate": 0.0006149628231603184,
      "loss": 1.5092,
      "step": 10158
    },
    {
      "epoch": 0.45173625005557777,
      "grad_norm": 0.0641988068819046,
      "learning_rate": 0.0006148205463208902,
      "loss": 1.5138,
      "step": 10160
    },
    {
      "epoch": 0.45182517451425014,
      "grad_norm": 0.06309515982866287,
      "learning_rate": 0.000614678259666096,
      "loss": 1.5057,
      "step": 10162
    },
    {
      "epoch": 0.4519140989729225,
      "grad_norm": 0.06427089869976044,
      "learning_rate": 0.0006145359632080987,
      "loss": 1.5137,
      "step": 10164
    },
    {
      "epoch": 0.4520030234315949,
      "grad_norm": 0.06765829026699066,
      "learning_rate": 0.0006143936569590624,
      "loss": 1.5093,
      "step": 10166
    },
    {
      "epoch": 0.45209194789026724,
      "grad_norm": 0.0644116923213005,
      "learning_rate": 0.0006142513409311522,
      "loss": 1.5123,
      "step": 10168
    },
    {
      "epoch": 0.45218087234893956,
      "grad_norm": 0.06532014906406403,
      "learning_rate": 0.000614109015136534,
      "loss": 1.5055,
      "step": 10170
    },
    {
      "epoch": 0.4522697968076119,
      "grad_norm": 0.06273654848337173,
      "learning_rate": 0.0006139666795873743,
      "loss": 1.5117,
      "step": 10172
    },
    {
      "epoch": 0.4523587212662843,
      "grad_norm": 0.0660717710852623,
      "learning_rate": 0.0006138243342958405,
      "loss": 1.5065,
      "step": 10174
    },
    {
      "epoch": 0.45244764572495666,
      "grad_norm": 0.06457041203975677,
      "learning_rate": 0.0006136819792741011,
      "loss": 1.5117,
      "step": 10176
    },
    {
      "epoch": 0.452536570183629,
      "grad_norm": 0.06340939551591873,
      "learning_rate": 0.000613539614534325,
      "loss": 1.507,
      "step": 10178
    },
    {
      "epoch": 0.45262549464230134,
      "grad_norm": 0.06222638487815857,
      "learning_rate": 0.0006133972400886825,
      "loss": 1.5065,
      "step": 10180
    },
    {
      "epoch": 0.4527144191009737,
      "grad_norm": 0.06319104135036469,
      "learning_rate": 0.0006132548559493441,
      "loss": 1.5134,
      "step": 10182
    },
    {
      "epoch": 0.4528033435596461,
      "grad_norm": 0.06425220519304276,
      "learning_rate": 0.0006131124621284815,
      "loss": 1.509,
      "step": 10184
    },
    {
      "epoch": 0.45289226801831844,
      "grad_norm": 0.06511306017637253,
      "learning_rate": 0.0006129700586382671,
      "loss": 1.5106,
      "step": 10186
    },
    {
      "epoch": 0.4529811924769908,
      "grad_norm": 0.06421872228384018,
      "learning_rate": 0.0006128276454908742,
      "loss": 1.5101,
      "step": 10188
    },
    {
      "epoch": 0.4530701169356632,
      "grad_norm": 0.06392832100391388,
      "learning_rate": 0.0006126852226984771,
      "loss": 1.5032,
      "step": 10190
    },
    {
      "epoch": 0.4531590413943355,
      "grad_norm": 0.06504455208778381,
      "learning_rate": 0.0006125427902732506,
      "loss": 1.5128,
      "step": 10192
    },
    {
      "epoch": 0.45324796585300786,
      "grad_norm": 0.06360786408185959,
      "learning_rate": 0.0006124003482273704,
      "loss": 1.5038,
      "step": 10194
    },
    {
      "epoch": 0.45333689031168023,
      "grad_norm": 0.06323164701461792,
      "learning_rate": 0.000612257896573013,
      "loss": 1.5068,
      "step": 10196
    },
    {
      "epoch": 0.4534258147703526,
      "grad_norm": 0.06563688069581985,
      "learning_rate": 0.000612115435322356,
      "loss": 1.5096,
      "step": 10198
    },
    {
      "epoch": 0.45351473922902497,
      "grad_norm": 0.06295774132013321,
      "learning_rate": 0.0006119729644875774,
      "loss": 1.5103,
      "step": 10200
    },
    {
      "epoch": 0.4536036636876973,
      "grad_norm": 0.06223352998495102,
      "learning_rate": 0.0006118304840808565,
      "loss": 1.5064,
      "step": 10202
    },
    {
      "epoch": 0.45369258814636965,
      "grad_norm": 0.06273999065160751,
      "learning_rate": 0.0006116879941143728,
      "loss": 1.5096,
      "step": 10204
    },
    {
      "epoch": 0.453781512605042,
      "grad_norm": 0.06447035819292068,
      "learning_rate": 0.0006115454946003074,
      "loss": 1.5094,
      "step": 10206
    },
    {
      "epoch": 0.4538704370637144,
      "grad_norm": 0.06401224434375763,
      "learning_rate": 0.0006114029855508413,
      "loss": 1.5158,
      "step": 10208
    },
    {
      "epoch": 0.45395936152238675,
      "grad_norm": 0.06190183758735657,
      "learning_rate": 0.0006112604669781572,
      "loss": 1.5101,
      "step": 10210
    },
    {
      "epoch": 0.45404828598105906,
      "grad_norm": 0.06359678506851196,
      "learning_rate": 0.0006111179388944381,
      "loss": 1.5114,
      "step": 10212
    },
    {
      "epoch": 0.45413721043973143,
      "grad_norm": 0.06362069398164749,
      "learning_rate": 0.0006109754013118678,
      "loss": 1.5103,
      "step": 10214
    },
    {
      "epoch": 0.4542261348984038,
      "grad_norm": 0.06391779333353043,
      "learning_rate": 0.0006108328542426312,
      "loss": 1.5183,
      "step": 10216
    },
    {
      "epoch": 0.45431505935707617,
      "grad_norm": 0.06379327178001404,
      "learning_rate": 0.0006106902976989139,
      "loss": 1.5107,
      "step": 10218
    },
    {
      "epoch": 0.45440398381574854,
      "grad_norm": 0.06601396203041077,
      "learning_rate": 0.0006105477316929021,
      "loss": 1.5121,
      "step": 10220
    },
    {
      "epoch": 0.4544929082744209,
      "grad_norm": 0.0628276839852333,
      "learning_rate": 0.0006104051562367829,
      "loss": 1.5103,
      "step": 10222
    },
    {
      "epoch": 0.4545818327330932,
      "grad_norm": 0.06758300215005875,
      "learning_rate": 0.0006102625713427446,
      "loss": 1.5088,
      "step": 10224
    },
    {
      "epoch": 0.4546707571917656,
      "grad_norm": 0.06313972175121307,
      "learning_rate": 0.0006101199770229758,
      "loss": 1.5101,
      "step": 10226
    },
    {
      "epoch": 0.45475968165043795,
      "grad_norm": 0.06574665755033493,
      "learning_rate": 0.0006099773732896658,
      "loss": 1.5103,
      "step": 10228
    },
    {
      "epoch": 0.4548486061091103,
      "grad_norm": 0.06595081835985184,
      "learning_rate": 0.0006098347601550055,
      "loss": 1.5099,
      "step": 10230
    },
    {
      "epoch": 0.4549375305677827,
      "grad_norm": 0.06619875133037567,
      "learning_rate": 0.0006096921376311857,
      "loss": 1.5086,
      "step": 10232
    },
    {
      "epoch": 0.455026455026455,
      "grad_norm": 0.06375854462385178,
      "learning_rate": 0.0006095495057303988,
      "loss": 1.5088,
      "step": 10234
    },
    {
      "epoch": 0.45511537948512737,
      "grad_norm": 0.06172388046979904,
      "learning_rate": 0.0006094068644648373,
      "loss": 1.5109,
      "step": 10236
    },
    {
      "epoch": 0.45520430394379974,
      "grad_norm": 0.06474345177412033,
      "learning_rate": 0.0006092642138466948,
      "loss": 1.5136,
      "step": 10238
    },
    {
      "epoch": 0.4552932284024721,
      "grad_norm": 0.06749559938907623,
      "learning_rate": 0.0006091215538881658,
      "loss": 1.5045,
      "step": 10240
    },
    {
      "epoch": 0.4553821528611445,
      "grad_norm": 0.062397945672273636,
      "learning_rate": 0.0006089788846014457,
      "loss": 1.5162,
      "step": 10242
    },
    {
      "epoch": 0.45547107731981684,
      "grad_norm": 0.06550050526857376,
      "learning_rate": 0.0006088362059987301,
      "loss": 1.5102,
      "step": 10244
    },
    {
      "epoch": 0.45556000177848915,
      "grad_norm": 0.0633438304066658,
      "learning_rate": 0.0006086935180922159,
      "loss": 1.506,
      "step": 10246
    },
    {
      "epoch": 0.4556489262371615,
      "grad_norm": 0.06535664945840836,
      "learning_rate": 0.000608550820894101,
      "loss": 1.5092,
      "step": 10248
    },
    {
      "epoch": 0.4557378506958339,
      "grad_norm": 0.0659709945321083,
      "learning_rate": 0.0006084081144165835,
      "loss": 1.518,
      "step": 10250
    },
    {
      "epoch": 0.45582677515450626,
      "grad_norm": 0.06421976536512375,
      "learning_rate": 0.0006082653986718626,
      "loss": 1.5138,
      "step": 10252
    },
    {
      "epoch": 0.4559156996131786,
      "grad_norm": 0.06513391435146332,
      "learning_rate": 0.0006081226736721383,
      "loss": 1.5079,
      "step": 10254
    },
    {
      "epoch": 0.45600462407185094,
      "grad_norm": 0.06458871066570282,
      "learning_rate": 0.0006079799394296115,
      "loss": 1.5101,
      "step": 10256
    },
    {
      "epoch": 0.4560935485305233,
      "grad_norm": 0.06421530246734619,
      "learning_rate": 0.0006078371959564833,
      "loss": 1.5049,
      "step": 10258
    },
    {
      "epoch": 0.4561824729891957,
      "grad_norm": 0.0646352618932724,
      "learning_rate": 0.0006076944432649567,
      "loss": 1.504,
      "step": 10260
    },
    {
      "epoch": 0.45627139744786804,
      "grad_norm": 0.06522826105356216,
      "learning_rate": 0.0006075516813672342,
      "loss": 1.5081,
      "step": 10262
    },
    {
      "epoch": 0.4563603219065404,
      "grad_norm": 0.06355036050081253,
      "learning_rate": 0.0006074089102755204,
      "loss": 1.5131,
      "step": 10264
    },
    {
      "epoch": 0.4564492463652127,
      "grad_norm": 0.06306944042444229,
      "learning_rate": 0.0006072661300020193,
      "loss": 1.5033,
      "step": 10266
    },
    {
      "epoch": 0.4565381708238851,
      "grad_norm": 0.0642002746462822,
      "learning_rate": 0.0006071233405589368,
      "loss": 1.5074,
      "step": 10268
    },
    {
      "epoch": 0.45662709528255746,
      "grad_norm": 0.06410963833332062,
      "learning_rate": 0.0006069805419584791,
      "loss": 1.5062,
      "step": 10270
    },
    {
      "epoch": 0.45671601974122983,
      "grad_norm": 0.06610678881406784,
      "learning_rate": 0.0006068377342128532,
      "loss": 1.5125,
      "step": 10272
    },
    {
      "epoch": 0.4568049441999022,
      "grad_norm": 0.06421998143196106,
      "learning_rate": 0.000606694917334267,
      "loss": 1.5103,
      "step": 10274
    },
    {
      "epoch": 0.45689386865857456,
      "grad_norm": 0.0632401779294014,
      "learning_rate": 0.000606552091334929,
      "loss": 1.5144,
      "step": 10276
    },
    {
      "epoch": 0.4569827931172469,
      "grad_norm": 0.06443547457456589,
      "learning_rate": 0.0006064092562270487,
      "loss": 1.5061,
      "step": 10278
    },
    {
      "epoch": 0.45707171757591925,
      "grad_norm": 0.06429964303970337,
      "learning_rate": 0.0006062664120228363,
      "loss": 1.5122,
      "step": 10280
    },
    {
      "epoch": 0.4571606420345916,
      "grad_norm": 0.06369158625602722,
      "learning_rate": 0.0006061235587345025,
      "loss": 1.5081,
      "step": 10282
    },
    {
      "epoch": 0.457249566493264,
      "grad_norm": 0.062136199325323105,
      "learning_rate": 0.0006059806963742595,
      "loss": 1.5106,
      "step": 10284
    },
    {
      "epoch": 0.45733849095193635,
      "grad_norm": 0.06280120462179184,
      "learning_rate": 0.0006058378249543193,
      "loss": 1.5096,
      "step": 10286
    },
    {
      "epoch": 0.45742741541060866,
      "grad_norm": 0.06295254826545715,
      "learning_rate": 0.0006056949444868956,
      "loss": 1.5038,
      "step": 10288
    },
    {
      "epoch": 0.45751633986928103,
      "grad_norm": 0.0640212669968605,
      "learning_rate": 0.0006055520549842022,
      "loss": 1.5059,
      "step": 10290
    },
    {
      "epoch": 0.4576052643279534,
      "grad_norm": 0.06613568961620331,
      "learning_rate": 0.000605409156458454,
      "loss": 1.5122,
      "step": 10292
    },
    {
      "epoch": 0.45769418878662577,
      "grad_norm": 0.06338582932949066,
      "learning_rate": 0.0006052662489218665,
      "loss": 1.5069,
      "step": 10294
    },
    {
      "epoch": 0.45778311324529813,
      "grad_norm": 0.06454145908355713,
      "learning_rate": 0.0006051233323866563,
      "loss": 1.506,
      "step": 10296
    },
    {
      "epoch": 0.4578720377039705,
      "grad_norm": 0.06602118164300919,
      "learning_rate": 0.0006049804068650403,
      "loss": 1.5075,
      "step": 10298
    },
    {
      "epoch": 0.4579609621626428,
      "grad_norm": 0.06418369710445404,
      "learning_rate": 0.0006048374723692365,
      "loss": 1.508,
      "step": 10300
    },
    {
      "epoch": 0.4580498866213152,
      "grad_norm": 0.063139408826828,
      "learning_rate": 0.0006046945289114634,
      "loss": 1.5068,
      "step": 10302
    },
    {
      "epoch": 0.45813881107998755,
      "grad_norm": 0.06343439221382141,
      "learning_rate": 0.0006045515765039408,
      "loss": 1.5078,
      "step": 10304
    },
    {
      "epoch": 0.4582277355386599,
      "grad_norm": 0.06396979093551636,
      "learning_rate": 0.0006044086151588886,
      "loss": 1.5098,
      "step": 10306
    },
    {
      "epoch": 0.4583166599973323,
      "grad_norm": 0.06393872201442719,
      "learning_rate": 0.0006042656448885279,
      "loss": 1.507,
      "step": 10308
    },
    {
      "epoch": 0.4584055844560046,
      "grad_norm": 0.06459838151931763,
      "learning_rate": 0.0006041226657050804,
      "loss": 1.5107,
      "step": 10310
    },
    {
      "epoch": 0.45849450891467697,
      "grad_norm": 0.06489048898220062,
      "learning_rate": 0.0006039796776207686,
      "loss": 1.5094,
      "step": 10312
    },
    {
      "epoch": 0.45858343337334934,
      "grad_norm": 0.06409396976232529,
      "learning_rate": 0.0006038366806478157,
      "loss": 1.5047,
      "step": 10314
    },
    {
      "epoch": 0.4586723578320217,
      "grad_norm": 0.06495364010334015,
      "learning_rate": 0.0006036936747984456,
      "loss": 1.4996,
      "step": 10316
    },
    {
      "epoch": 0.4587612822906941,
      "grad_norm": 0.06571628898382187,
      "learning_rate": 0.0006035506600848835,
      "loss": 1.5113,
      "step": 10318
    },
    {
      "epoch": 0.45885020674936644,
      "grad_norm": 0.06504350155591965,
      "learning_rate": 0.0006034076365193545,
      "loss": 1.4964,
      "step": 10320
    },
    {
      "epoch": 0.45893913120803875,
      "grad_norm": 0.06698577105998993,
      "learning_rate": 0.0006032646041140849,
      "loss": 1.5051,
      "step": 10322
    },
    {
      "epoch": 0.4590280556667111,
      "grad_norm": 0.06189613789319992,
      "learning_rate": 0.0006031215628813021,
      "loss": 1.5097,
      "step": 10324
    },
    {
      "epoch": 0.4591169801253835,
      "grad_norm": 0.06384344398975372,
      "learning_rate": 0.0006029785128332336,
      "loss": 1.512,
      "step": 10326
    },
    {
      "epoch": 0.45920590458405586,
      "grad_norm": 0.06703834980726242,
      "learning_rate": 0.0006028354539821079,
      "loss": 1.5041,
      "step": 10328
    },
    {
      "epoch": 0.4592948290427282,
      "grad_norm": 0.06500376015901566,
      "learning_rate": 0.0006026923863401545,
      "loss": 1.5061,
      "step": 10330
    },
    {
      "epoch": 0.45938375350140054,
      "grad_norm": 0.06535261869430542,
      "learning_rate": 0.0006025493099196033,
      "loss": 1.5052,
      "step": 10332
    },
    {
      "epoch": 0.4594726779600729,
      "grad_norm": 0.062189262360334396,
      "learning_rate": 0.0006024062247326854,
      "loss": 1.5009,
      "step": 10334
    },
    {
      "epoch": 0.4595616024187453,
      "grad_norm": 0.06319093704223633,
      "learning_rate": 0.0006022631307916318,
      "loss": 1.5052,
      "step": 10336
    },
    {
      "epoch": 0.45965052687741764,
      "grad_norm": 0.0611843504011631,
      "learning_rate": 0.0006021200281086753,
      "loss": 1.5051,
      "step": 10338
    },
    {
      "epoch": 0.45973945133609,
      "grad_norm": 0.06422754377126694,
      "learning_rate": 0.0006019769166960485,
      "loss": 1.5042,
      "step": 10340
    },
    {
      "epoch": 0.4598283757947623,
      "grad_norm": 0.06372174620628357,
      "learning_rate": 0.0006018337965659859,
      "loss": 1.513,
      "step": 10342
    },
    {
      "epoch": 0.4599173002534347,
      "grad_norm": 0.06303902715444565,
      "learning_rate": 0.0006016906677307213,
      "loss": 1.5053,
      "step": 10344
    },
    {
      "epoch": 0.46000622471210706,
      "grad_norm": 0.0631803497672081,
      "learning_rate": 0.0006015475302024904,
      "loss": 1.5102,
      "step": 10346
    },
    {
      "epoch": 0.4600951491707794,
      "grad_norm": 0.06362450867891312,
      "learning_rate": 0.0006014043839935291,
      "loss": 1.5115,
      "step": 10348
    },
    {
      "epoch": 0.4601840736294518,
      "grad_norm": 0.06407329440116882,
      "learning_rate": 0.0006012612291160743,
      "loss": 1.5065,
      "step": 10350
    },
    {
      "epoch": 0.46027299808812416,
      "grad_norm": 0.061808858066797256,
      "learning_rate": 0.0006011180655823632,
      "loss": 1.5125,
      "step": 10352
    },
    {
      "epoch": 0.4603619225467965,
      "grad_norm": 0.0652756616473198,
      "learning_rate": 0.0006009748934046343,
      "loss": 1.5064,
      "step": 10354
    },
    {
      "epoch": 0.46045084700546884,
      "grad_norm": 0.0631522387266159,
      "learning_rate": 0.0006008317125951265,
      "loss": 1.504,
      "step": 10356
    },
    {
      "epoch": 0.4605397714641412,
      "grad_norm": 0.0668981596827507,
      "learning_rate": 0.0006006885231660796,
      "loss": 1.5145,
      "step": 10358
    },
    {
      "epoch": 0.4606286959228136,
      "grad_norm": 0.06222165748476982,
      "learning_rate": 0.0006005453251297341,
      "loss": 1.5067,
      "step": 10360
    },
    {
      "epoch": 0.46071762038148595,
      "grad_norm": 0.06397759169340134,
      "learning_rate": 0.0006004021184983309,
      "loss": 1.5067,
      "step": 10362
    },
    {
      "epoch": 0.46080654484015826,
      "grad_norm": 0.060776520520448685,
      "learning_rate": 0.0006002589032841122,
      "loss": 1.5025,
      "step": 10364
    },
    {
      "epoch": 0.46089546929883063,
      "grad_norm": 0.06292383372783661,
      "learning_rate": 0.0006001156794993208,
      "loss": 1.4995,
      "step": 10366
    },
    {
      "epoch": 0.460984393757503,
      "grad_norm": 0.06334930658340454,
      "learning_rate": 0.0005999724471561999,
      "loss": 1.5038,
      "step": 10368
    },
    {
      "epoch": 0.46107331821617537,
      "grad_norm": 0.062282536178827286,
      "learning_rate": 0.0005998292062669935,
      "loss": 1.5019,
      "step": 10370
    },
    {
      "epoch": 0.46116224267484773,
      "grad_norm": 0.06332574039697647,
      "learning_rate": 0.0005996859568439468,
      "loss": 1.5022,
      "step": 10372
    },
    {
      "epoch": 0.4612511671335201,
      "grad_norm": 0.06254351884126663,
      "learning_rate": 0.000599542698899305,
      "loss": 1.513,
      "step": 10374
    },
    {
      "epoch": 0.4613400915921924,
      "grad_norm": 0.06307613104581833,
      "learning_rate": 0.0005993994324453147,
      "loss": 1.5088,
      "step": 10376
    },
    {
      "epoch": 0.4614290160508648,
      "grad_norm": 0.062051285058259964,
      "learning_rate": 0.0005992561574942229,
      "loss": 1.4984,
      "step": 10378
    },
    {
      "epoch": 0.46151794050953715,
      "grad_norm": 0.06389043480157852,
      "learning_rate": 0.0005991128740582773,
      "loss": 1.5078,
      "step": 10380
    },
    {
      "epoch": 0.4616068649682095,
      "grad_norm": 0.06286501884460449,
      "learning_rate": 0.0005989695821497266,
      "loss": 1.5044,
      "step": 10382
    },
    {
      "epoch": 0.4616957894268819,
      "grad_norm": 0.0667327269911766,
      "learning_rate": 0.0005988262817808198,
      "loss": 1.5093,
      "step": 10384
    },
    {
      "epoch": 0.4617847138855542,
      "grad_norm": 0.06277990341186523,
      "learning_rate": 0.0005986829729638069,
      "loss": 1.5009,
      "step": 10386
    },
    {
      "epoch": 0.46187363834422657,
      "grad_norm": 0.06250549107789993,
      "learning_rate": 0.0005985396557109386,
      "loss": 1.509,
      "step": 10388
    },
    {
      "epoch": 0.46196256280289894,
      "grad_norm": 0.06798355281352997,
      "learning_rate": 0.0005983963300344662,
      "loss": 1.5055,
      "step": 10390
    },
    {
      "epoch": 0.4620514872615713,
      "grad_norm": 0.06521153450012207,
      "learning_rate": 0.000598252995946642,
      "loss": 1.5108,
      "step": 10392
    },
    {
      "epoch": 0.46214041172024367,
      "grad_norm": 0.06334955245256424,
      "learning_rate": 0.0005981096534597186,
      "loss": 1.5062,
      "step": 10394
    },
    {
      "epoch": 0.462229336178916,
      "grad_norm": 0.06671233475208282,
      "learning_rate": 0.0005979663025859499,
      "loss": 1.508,
      "step": 10396
    },
    {
      "epoch": 0.46231826063758835,
      "grad_norm": 0.06669692695140839,
      "learning_rate": 0.0005978229433375897,
      "loss": 1.5016,
      "step": 10398
    },
    {
      "epoch": 0.4624071850962607,
      "grad_norm": 0.06515517830848694,
      "learning_rate": 0.0005976795757268933,
      "loss": 1.5029,
      "step": 10400
    },
    {
      "epoch": 0.4624961095549331,
      "grad_norm": 0.06504440307617188,
      "learning_rate": 0.0005975361997661162,
      "loss": 1.5052,
      "step": 10402
    },
    {
      "epoch": 0.46258503401360546,
      "grad_norm": 0.06195230036973953,
      "learning_rate": 0.0005973928154675151,
      "loss": 1.5033,
      "step": 10404
    },
    {
      "epoch": 0.4626739584722778,
      "grad_norm": 0.0632019117474556,
      "learning_rate": 0.0005972494228433468,
      "loss": 1.5057,
      "step": 10406
    },
    {
      "epoch": 0.46276288293095014,
      "grad_norm": 0.06470786780118942,
      "learning_rate": 0.0005971060219058694,
      "loss": 1.5045,
      "step": 10408
    },
    {
      "epoch": 0.4628518073896225,
      "grad_norm": 0.06392474472522736,
      "learning_rate": 0.000596962612667341,
      "loss": 1.506,
      "step": 10410
    },
    {
      "epoch": 0.4629407318482949,
      "grad_norm": 0.06552916020154953,
      "learning_rate": 0.0005968191951400215,
      "loss": 1.5066,
      "step": 10412
    },
    {
      "epoch": 0.46302965630696724,
      "grad_norm": 0.06476908922195435,
      "learning_rate": 0.0005966757693361705,
      "loss": 1.5028,
      "step": 10414
    },
    {
      "epoch": 0.4631185807656396,
      "grad_norm": 0.06308183819055557,
      "learning_rate": 0.0005965323352680486,
      "loss": 1.5073,
      "step": 10416
    },
    {
      "epoch": 0.4632075052243119,
      "grad_norm": 0.06603847444057465,
      "learning_rate": 0.0005963888929479173,
      "loss": 1.516,
      "step": 10418
    },
    {
      "epoch": 0.4632964296829843,
      "grad_norm": 0.06707409769296646,
      "learning_rate": 0.0005962454423880387,
      "loss": 1.5026,
      "step": 10420
    },
    {
      "epoch": 0.46338535414165666,
      "grad_norm": 0.06557576358318329,
      "learning_rate": 0.0005961019836006755,
      "loss": 1.5101,
      "step": 10422
    },
    {
      "epoch": 0.463474278600329,
      "grad_norm": 0.06399369239807129,
      "learning_rate": 0.0005959585165980912,
      "loss": 1.5085,
      "step": 10424
    },
    {
      "epoch": 0.4635632030590014,
      "grad_norm": 0.06383886933326721,
      "learning_rate": 0.0005958150413925501,
      "loss": 1.5059,
      "step": 10426
    },
    {
      "epoch": 0.46365212751767376,
      "grad_norm": 0.0634358674287796,
      "learning_rate": 0.000595671557996317,
      "loss": 1.5054,
      "step": 10428
    },
    {
      "epoch": 0.4637410519763461,
      "grad_norm": 0.06545853614807129,
      "learning_rate": 0.0005955280664216575,
      "loss": 1.513,
      "step": 10430
    },
    {
      "epoch": 0.46382997643501844,
      "grad_norm": 0.06597691774368286,
      "learning_rate": 0.0005953845666808378,
      "loss": 1.5063,
      "step": 10432
    },
    {
      "epoch": 0.4639189008936908,
      "grad_norm": 0.06610873341560364,
      "learning_rate": 0.0005952410587861251,
      "loss": 1.5096,
      "step": 10434
    },
    {
      "epoch": 0.4640078253523632,
      "grad_norm": 0.06371376663446426,
      "learning_rate": 0.0005950975427497871,
      "loss": 1.5077,
      "step": 10436
    },
    {
      "epoch": 0.46409674981103555,
      "grad_norm": 0.06398597359657288,
      "learning_rate": 0.0005949540185840919,
      "loss": 1.5092,
      "step": 10438
    },
    {
      "epoch": 0.46418567426970786,
      "grad_norm": 0.06583784520626068,
      "learning_rate": 0.0005948104863013089,
      "loss": 1.5122,
      "step": 10440
    },
    {
      "epoch": 0.46427459872838023,
      "grad_norm": 0.06568284332752228,
      "learning_rate": 0.0005946669459137075,
      "loss": 1.5076,
      "step": 10442
    },
    {
      "epoch": 0.4643635231870526,
      "grad_norm": 0.06291704624891281,
      "learning_rate": 0.0005945233974335585,
      "loss": 1.5103,
      "step": 10444
    },
    {
      "epoch": 0.46445244764572496,
      "grad_norm": 0.06666183471679688,
      "learning_rate": 0.0005943798408731329,
      "loss": 1.5065,
      "step": 10446
    },
    {
      "epoch": 0.46454137210439733,
      "grad_norm": 0.06522826105356216,
      "learning_rate": 0.0005942362762447026,
      "loss": 1.5054,
      "step": 10448
    },
    {
      "epoch": 0.46463029656306964,
      "grad_norm": 0.06623510271310806,
      "learning_rate": 0.0005940927035605403,
      "loss": 1.5098,
      "step": 10450
    },
    {
      "epoch": 0.464719221021742,
      "grad_norm": 0.06403439491987228,
      "learning_rate": 0.0005939491228329187,
      "loss": 1.5,
      "step": 10452
    },
    {
      "epoch": 0.4648081454804144,
      "grad_norm": 0.0637822076678276,
      "learning_rate": 0.0005938055340741123,
      "loss": 1.5119,
      "step": 10454
    },
    {
      "epoch": 0.46489706993908675,
      "grad_norm": 0.0661793127655983,
      "learning_rate": 0.0005936619372963953,
      "loss": 1.5033,
      "step": 10456
    },
    {
      "epoch": 0.4649859943977591,
      "grad_norm": 0.06596322357654572,
      "learning_rate": 0.0005935183325120433,
      "loss": 1.5056,
      "step": 10458
    },
    {
      "epoch": 0.4650749188564315,
      "grad_norm": 0.06263718008995056,
      "learning_rate": 0.000593374719733332,
      "loss": 1.5065,
      "step": 10460
    },
    {
      "epoch": 0.4651638433151038,
      "grad_norm": 0.06337321549654007,
      "learning_rate": 0.0005932310989725382,
      "loss": 1.5115,
      "step": 10462
    },
    {
      "epoch": 0.46525276777377617,
      "grad_norm": 0.0630149319767952,
      "learning_rate": 0.0005930874702419392,
      "loss": 1.5084,
      "step": 10464
    },
    {
      "epoch": 0.46534169223244853,
      "grad_norm": 0.06514789909124374,
      "learning_rate": 0.0005929438335538131,
      "loss": 1.5035,
      "step": 10466
    },
    {
      "epoch": 0.4654306166911209,
      "grad_norm": 0.061315376311540604,
      "learning_rate": 0.0005928001889204385,
      "loss": 1.5073,
      "step": 10468
    },
    {
      "epoch": 0.46551954114979327,
      "grad_norm": 0.06364026665687561,
      "learning_rate": 0.0005926565363540947,
      "loss": 1.5027,
      "step": 10470
    },
    {
      "epoch": 0.4656084656084656,
      "grad_norm": 0.06301172822713852,
      "learning_rate": 0.0005925128758670619,
      "loss": 1.5064,
      "step": 10472
    },
    {
      "epoch": 0.46569739006713795,
      "grad_norm": 0.06264245510101318,
      "learning_rate": 0.0005923692074716209,
      "loss": 1.5051,
      "step": 10474
    },
    {
      "epoch": 0.4657863145258103,
      "grad_norm": 0.06521310657262802,
      "learning_rate": 0.0005922255311800529,
      "loss": 1.5008,
      "step": 10476
    },
    {
      "epoch": 0.4658752389844827,
      "grad_norm": 0.06265758723020554,
      "learning_rate": 0.0005920818470046399,
      "loss": 1.5011,
      "step": 10478
    },
    {
      "epoch": 0.46596416344315505,
      "grad_norm": 0.06298228353261948,
      "learning_rate": 0.000591938154957665,
      "loss": 1.5046,
      "step": 10480
    },
    {
      "epoch": 0.4660530879018274,
      "grad_norm": 0.06504108011722565,
      "learning_rate": 0.0005917944550514114,
      "loss": 1.5035,
      "step": 10482
    },
    {
      "epoch": 0.46614201236049974,
      "grad_norm": 0.0628167986869812,
      "learning_rate": 0.0005916507472981632,
      "loss": 1.5018,
      "step": 10484
    },
    {
      "epoch": 0.4662309368191721,
      "grad_norm": 0.06394867599010468,
      "learning_rate": 0.0005915070317102053,
      "loss": 1.5055,
      "step": 10486
    },
    {
      "epoch": 0.46631986127784447,
      "grad_norm": 0.06346921622753143,
      "learning_rate": 0.0005913633082998231,
      "loss": 1.5132,
      "step": 10488
    },
    {
      "epoch": 0.46640878573651684,
      "grad_norm": 0.06185340881347656,
      "learning_rate": 0.0005912195770793028,
      "loss": 1.5051,
      "step": 10490
    },
    {
      "epoch": 0.4664977101951892,
      "grad_norm": 0.06247565150260925,
      "learning_rate": 0.0005910758380609308,
      "loss": 1.5057,
      "step": 10492
    },
    {
      "epoch": 0.4665866346538615,
      "grad_norm": 0.061248041689395905,
      "learning_rate": 0.000590932091256995,
      "loss": 1.4972,
      "step": 10494
    },
    {
      "epoch": 0.4666755591125339,
      "grad_norm": 0.062036920338869095,
      "learning_rate": 0.0005907883366797832,
      "loss": 1.5055,
      "step": 10496
    },
    {
      "epoch": 0.46676448357120626,
      "grad_norm": 0.0646434798836708,
      "learning_rate": 0.0005906445743415845,
      "loss": 1.5068,
      "step": 10498
    },
    {
      "epoch": 0.4668534080298786,
      "grad_norm": 0.06403973698616028,
      "learning_rate": 0.0005905008042546878,
      "loss": 1.5053,
      "step": 10500
    },
    {
      "epoch": 0.4668534080298786,
      "eval_loss": 1.4856921434402466,
      "eval_runtime": 12.4102,
      "eval_samples_per_second": 556.799,
      "eval_steps_per_second": 69.62,
      "step": 10500
    },
    {
      "epoch": 0.466942332488551,
      "grad_norm": 0.06568461656570435,
      "learning_rate": 0.0005903570264313837,
      "loss": 1.5073,
      "step": 10502
    },
    {
      "epoch": 0.46703125694722336,
      "grad_norm": 0.06413400918245316,
      "learning_rate": 0.0005902132408839626,
      "loss": 1.5047,
      "step": 10504
    },
    {
      "epoch": 0.4671201814058957,
      "grad_norm": 0.06379328668117523,
      "learning_rate": 0.0005900694476247164,
      "loss": 1.5017,
      "step": 10506
    },
    {
      "epoch": 0.46720910586456804,
      "grad_norm": 0.06230514496564865,
      "learning_rate": 0.0005899256466659369,
      "loss": 1.5085,
      "step": 10508
    },
    {
      "epoch": 0.4672980303232404,
      "grad_norm": 0.06407498568296432,
      "learning_rate": 0.0005897818380199165,
      "loss": 1.5085,
      "step": 10510
    },
    {
      "epoch": 0.4673869547819128,
      "grad_norm": 0.06471338123083115,
      "learning_rate": 0.0005896380216989495,
      "loss": 1.5057,
      "step": 10512
    },
    {
      "epoch": 0.46747587924058515,
      "grad_norm": 0.06504790484905243,
      "learning_rate": 0.0005894941977153289,
      "loss": 1.5093,
      "step": 10514
    },
    {
      "epoch": 0.46756480369925746,
      "grad_norm": 0.06497068703174591,
      "learning_rate": 0.0005893503660813499,
      "loss": 1.5046,
      "step": 10516
    },
    {
      "epoch": 0.4676537281579298,
      "grad_norm": 0.06451025605201721,
      "learning_rate": 0.000589206526809308,
      "loss": 1.504,
      "step": 10518
    },
    {
      "epoch": 0.4677426526166022,
      "grad_norm": 0.062198251485824585,
      "learning_rate": 0.0005890626799114991,
      "loss": 1.5068,
      "step": 10520
    },
    {
      "epoch": 0.46783157707527456,
      "grad_norm": 0.0641157478094101,
      "learning_rate": 0.0005889188254002198,
      "loss": 1.5027,
      "step": 10522
    },
    {
      "epoch": 0.46792050153394693,
      "grad_norm": 0.06388696283102036,
      "learning_rate": 0.0005887749632877673,
      "loss": 1.5103,
      "step": 10524
    },
    {
      "epoch": 0.46800942599261924,
      "grad_norm": 0.06389610469341278,
      "learning_rate": 0.0005886310935864399,
      "loss": 1.5049,
      "step": 10526
    },
    {
      "epoch": 0.4680983504512916,
      "grad_norm": 0.06436503678560257,
      "learning_rate": 0.0005884872163085359,
      "loss": 1.5033,
      "step": 10528
    },
    {
      "epoch": 0.468187274909964,
      "grad_norm": 0.06388412415981293,
      "learning_rate": 0.0005883433314663549,
      "loss": 1.5082,
      "step": 10530
    },
    {
      "epoch": 0.46827619936863635,
      "grad_norm": 0.06415783613920212,
      "learning_rate": 0.0005881994390721964,
      "loss": 1.5031,
      "step": 10532
    },
    {
      "epoch": 0.4683651238273087,
      "grad_norm": 0.06618549674749374,
      "learning_rate": 0.0005880555391383613,
      "loss": 1.4983,
      "step": 10534
    },
    {
      "epoch": 0.4684540482859811,
      "grad_norm": 0.06373777985572815,
      "learning_rate": 0.0005879116316771507,
      "loss": 1.5058,
      "step": 10536
    },
    {
      "epoch": 0.4685429727446534,
      "grad_norm": 0.06715410202741623,
      "learning_rate": 0.0005877677167008663,
      "loss": 1.5013,
      "step": 10538
    },
    {
      "epoch": 0.46863189720332576,
      "grad_norm": 0.0642244890332222,
      "learning_rate": 0.0005876237942218107,
      "loss": 1.5073,
      "step": 10540
    },
    {
      "epoch": 0.46872082166199813,
      "grad_norm": 0.06601016223430634,
      "learning_rate": 0.0005874798642522869,
      "loss": 1.4981,
      "step": 10542
    },
    {
      "epoch": 0.4688097461206705,
      "grad_norm": 0.06458527594804764,
      "learning_rate": 0.0005873359268045991,
      "loss": 1.5065,
      "step": 10544
    },
    {
      "epoch": 0.46889867057934287,
      "grad_norm": 0.06326668709516525,
      "learning_rate": 0.0005871919818910511,
      "loss": 1.5067,
      "step": 10546
    },
    {
      "epoch": 0.4689875950380152,
      "grad_norm": 0.06331662088632584,
      "learning_rate": 0.0005870480295239486,
      "loss": 1.5053,
      "step": 10548
    },
    {
      "epoch": 0.46907651949668755,
      "grad_norm": 0.0645529255270958,
      "learning_rate": 0.0005869040697155966,
      "loss": 1.5113,
      "step": 10550
    },
    {
      "epoch": 0.4691654439553599,
      "grad_norm": 0.0635153278708458,
      "learning_rate": 0.0005867601024783021,
      "loss": 1.503,
      "step": 10552
    },
    {
      "epoch": 0.4692543684140323,
      "grad_norm": 0.06458617746829987,
      "learning_rate": 0.0005866161278243713,
      "loss": 1.5021,
      "step": 10554
    },
    {
      "epoch": 0.46934329287270465,
      "grad_norm": 0.0654539167881012,
      "learning_rate": 0.0005864721457661124,
      "loss": 1.5025,
      "step": 10556
    },
    {
      "epoch": 0.469432217331377,
      "grad_norm": 0.06510742008686066,
      "learning_rate": 0.0005863281563158332,
      "loss": 1.5045,
      "step": 10558
    },
    {
      "epoch": 0.46952114179004933,
      "grad_norm": 0.0627632662653923,
      "learning_rate": 0.000586184159485843,
      "loss": 1.5048,
      "step": 10560
    },
    {
      "epoch": 0.4696100662487217,
      "grad_norm": 0.06467565894126892,
      "learning_rate": 0.000586040155288451,
      "loss": 1.5027,
      "step": 10562
    },
    {
      "epoch": 0.46969899070739407,
      "grad_norm": 0.06286469101905823,
      "learning_rate": 0.0005858961437359674,
      "loss": 1.4974,
      "step": 10564
    },
    {
      "epoch": 0.46978791516606644,
      "grad_norm": 0.06415469199419022,
      "learning_rate": 0.0005857521248407027,
      "loss": 1.5036,
      "step": 10566
    },
    {
      "epoch": 0.4698768396247388,
      "grad_norm": 0.06374349445104599,
      "learning_rate": 0.0005856080986149687,
      "loss": 1.4962,
      "step": 10568
    },
    {
      "epoch": 0.4699657640834111,
      "grad_norm": 0.0641632080078125,
      "learning_rate": 0.0005854640650710771,
      "loss": 1.5011,
      "step": 10570
    },
    {
      "epoch": 0.4700546885420835,
      "grad_norm": 0.06440167129039764,
      "learning_rate": 0.0005853200242213405,
      "loss": 1.5109,
      "step": 10572
    },
    {
      "epoch": 0.47014361300075586,
      "grad_norm": 0.06403553485870361,
      "learning_rate": 0.0005851759760780724,
      "loss": 1.5112,
      "step": 10574
    },
    {
      "epoch": 0.4702325374594282,
      "grad_norm": 0.06359109282493591,
      "learning_rate": 0.0005850319206535863,
      "loss": 1.5012,
      "step": 10576
    },
    {
      "epoch": 0.4703214619181006,
      "grad_norm": 0.06405449658632278,
      "learning_rate": 0.0005848878579601971,
      "loss": 1.5061,
      "step": 10578
    },
    {
      "epoch": 0.4704103863767729,
      "grad_norm": 0.06662164628505707,
      "learning_rate": 0.0005847437880102196,
      "loss": 1.5053,
      "step": 10580
    },
    {
      "epoch": 0.4704993108354453,
      "grad_norm": 0.06323253363370895,
      "learning_rate": 0.0005845997108159697,
      "loss": 1.5118,
      "step": 10582
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 0.06517712771892548,
      "learning_rate": 0.0005844556263897637,
      "loss": 1.5091,
      "step": 10584
    },
    {
      "epoch": 0.47067715975279,
      "grad_norm": 0.0638645812869072,
      "learning_rate": 0.0005843115347439184,
      "loss": 1.5093,
      "step": 10586
    },
    {
      "epoch": 0.4707660842114624,
      "grad_norm": 0.06771344691514969,
      "learning_rate": 0.0005841674358907517,
      "loss": 1.512,
      "step": 10588
    },
    {
      "epoch": 0.47085500867013474,
      "grad_norm": 0.06667336076498032,
      "learning_rate": 0.0005840233298425818,
      "loss": 1.5051,
      "step": 10590
    },
    {
      "epoch": 0.47094393312880706,
      "grad_norm": 0.06689155846834183,
      "learning_rate": 0.0005838792166117273,
      "loss": 1.5107,
      "step": 10592
    },
    {
      "epoch": 0.4710328575874794,
      "grad_norm": 0.06521068513393402,
      "learning_rate": 0.0005837350962105076,
      "loss": 1.508,
      "step": 10594
    },
    {
      "epoch": 0.4711217820461518,
      "grad_norm": 0.06260724365711212,
      "learning_rate": 0.0005835909686512429,
      "loss": 1.4977,
      "step": 10596
    },
    {
      "epoch": 0.47121070650482416,
      "grad_norm": 0.06671774387359619,
      "learning_rate": 0.0005834468339462539,
      "loss": 1.5029,
      "step": 10598
    },
    {
      "epoch": 0.47129963096349653,
      "grad_norm": 0.0636298730969429,
      "learning_rate": 0.0005833026921078616,
      "loss": 1.5078,
      "step": 10600
    },
    {
      "epoch": 0.47138855542216884,
      "grad_norm": 0.0641683042049408,
      "learning_rate": 0.0005831585431483883,
      "loss": 1.4961,
      "step": 10602
    },
    {
      "epoch": 0.4714774798808412,
      "grad_norm": 0.06526412814855576,
      "learning_rate": 0.0005830143870801562,
      "loss": 1.5074,
      "step": 10604
    },
    {
      "epoch": 0.4715664043395136,
      "grad_norm": 0.06388209015130997,
      "learning_rate": 0.0005828702239154886,
      "loss": 1.5082,
      "step": 10606
    },
    {
      "epoch": 0.47165532879818595,
      "grad_norm": 0.0643208771944046,
      "learning_rate": 0.0005827260536667089,
      "loss": 1.5034,
      "step": 10608
    },
    {
      "epoch": 0.4717442532568583,
      "grad_norm": 0.0622149333357811,
      "learning_rate": 0.0005825818763461416,
      "loss": 1.5036,
      "step": 10610
    },
    {
      "epoch": 0.4718331777155307,
      "grad_norm": 0.06327205151319504,
      "learning_rate": 0.0005824376919661114,
      "loss": 1.5018,
      "step": 10612
    },
    {
      "epoch": 0.471922102174203,
      "grad_norm": 0.06469540297985077,
      "learning_rate": 0.0005822935005389443,
      "loss": 1.5091,
      "step": 10614
    },
    {
      "epoch": 0.47201102663287536,
      "grad_norm": 0.06485003978013992,
      "learning_rate": 0.000582149302076966,
      "loss": 1.5048,
      "step": 10616
    },
    {
      "epoch": 0.47209995109154773,
      "grad_norm": 0.06667590141296387,
      "learning_rate": 0.0005820050965925032,
      "loss": 1.5053,
      "step": 10618
    },
    {
      "epoch": 0.4721888755502201,
      "grad_norm": 0.0658281147480011,
      "learning_rate": 0.0005818608840978837,
      "loss": 1.5085,
      "step": 10620
    },
    {
      "epoch": 0.47227780000889247,
      "grad_norm": 0.06506504863500595,
      "learning_rate": 0.0005817166646054348,
      "loss": 1.5044,
      "step": 10622
    },
    {
      "epoch": 0.4723667244675648,
      "grad_norm": 0.06641970574855804,
      "learning_rate": 0.0005815724381274854,
      "loss": 1.507,
      "step": 10624
    },
    {
      "epoch": 0.47245564892623715,
      "grad_norm": 0.06248960644006729,
      "learning_rate": 0.0005814282046763643,
      "loss": 1.505,
      "step": 10626
    },
    {
      "epoch": 0.4725445733849095,
      "grad_norm": 0.06606055051088333,
      "learning_rate": 0.0005812839642644017,
      "loss": 1.5084,
      "step": 10628
    },
    {
      "epoch": 0.4726334978435819,
      "grad_norm": 0.06419503688812256,
      "learning_rate": 0.0005811397169039277,
      "loss": 1.5038,
      "step": 10630
    },
    {
      "epoch": 0.47272242230225425,
      "grad_norm": 0.06354578584432602,
      "learning_rate": 0.0005809954626072728,
      "loss": 1.5025,
      "step": 10632
    },
    {
      "epoch": 0.47281134676092657,
      "grad_norm": 0.06533468514680862,
      "learning_rate": 0.000580851201386769,
      "loss": 1.5118,
      "step": 10634
    },
    {
      "epoch": 0.47290027121959893,
      "grad_norm": 0.06488578021526337,
      "learning_rate": 0.0005807069332547482,
      "loss": 1.5051,
      "step": 10636
    },
    {
      "epoch": 0.4729891956782713,
      "grad_norm": 0.061041004955768585,
      "learning_rate": 0.000580562658223543,
      "loss": 1.5046,
      "step": 10638
    },
    {
      "epoch": 0.47307812013694367,
      "grad_norm": 0.06147739291191101,
      "learning_rate": 0.0005804183763054869,
      "loss": 1.5003,
      "step": 10640
    },
    {
      "epoch": 0.47316704459561604,
      "grad_norm": 0.06590231508016586,
      "learning_rate": 0.0005802740875129135,
      "loss": 1.5015,
      "step": 10642
    },
    {
      "epoch": 0.4732559690542884,
      "grad_norm": 0.06333669275045395,
      "learning_rate": 0.0005801297918581574,
      "loss": 1.5054,
      "step": 10644
    },
    {
      "epoch": 0.4733448935129607,
      "grad_norm": 0.06394929438829422,
      "learning_rate": 0.0005799854893535535,
      "loss": 1.4967,
      "step": 10646
    },
    {
      "epoch": 0.4734338179716331,
      "grad_norm": 0.06518962234258652,
      "learning_rate": 0.0005798411800114375,
      "loss": 1.5029,
      "step": 10648
    },
    {
      "epoch": 0.47352274243030545,
      "grad_norm": 0.06461624801158905,
      "learning_rate": 0.0005796968638441455,
      "loss": 1.5038,
      "step": 10650
    },
    {
      "epoch": 0.4736116668889778,
      "grad_norm": 0.06398481875658035,
      "learning_rate": 0.0005795525408640146,
      "loss": 1.501,
      "step": 10652
    },
    {
      "epoch": 0.4737005913476502,
      "grad_norm": 0.06461846828460693,
      "learning_rate": 0.0005794082110833817,
      "loss": 1.5096,
      "step": 10654
    },
    {
      "epoch": 0.4737895158063225,
      "grad_norm": 0.0636335164308548,
      "learning_rate": 0.0005792638745145851,
      "loss": 1.504,
      "step": 10656
    },
    {
      "epoch": 0.47387844026499487,
      "grad_norm": 0.0646030604839325,
      "learning_rate": 0.0005791195311699631,
      "loss": 1.5081,
      "step": 10658
    },
    {
      "epoch": 0.47396736472366724,
      "grad_norm": 0.06314581632614136,
      "learning_rate": 0.0005789751810618551,
      "loss": 1.5029,
      "step": 10660
    },
    {
      "epoch": 0.4740562891823396,
      "grad_norm": 0.06447373330593109,
      "learning_rate": 0.0005788308242026004,
      "loss": 1.5029,
      "step": 10662
    },
    {
      "epoch": 0.474145213641012,
      "grad_norm": 0.06374399363994598,
      "learning_rate": 0.0005786864606045396,
      "loss": 1.5069,
      "step": 10664
    },
    {
      "epoch": 0.47423413809968434,
      "grad_norm": 0.06367414444684982,
      "learning_rate": 0.0005785420902800131,
      "loss": 1.4976,
      "step": 10666
    },
    {
      "epoch": 0.47432306255835666,
      "grad_norm": 0.06507609784603119,
      "learning_rate": 0.0005783977132413629,
      "loss": 1.5004,
      "step": 10668
    },
    {
      "epoch": 0.474411987017029,
      "grad_norm": 0.06401117891073227,
      "learning_rate": 0.0005782533295009307,
      "loss": 1.5017,
      "step": 10670
    },
    {
      "epoch": 0.4745009114757014,
      "grad_norm": 0.06449148803949356,
      "learning_rate": 0.0005781089390710588,
      "loss": 1.5078,
      "step": 10672
    },
    {
      "epoch": 0.47458983593437376,
      "grad_norm": 0.06876698136329651,
      "learning_rate": 0.0005779645419640907,
      "loss": 1.5033,
      "step": 10674
    },
    {
      "epoch": 0.47467876039304613,
      "grad_norm": 0.06464815884828568,
      "learning_rate": 0.00057782013819237,
      "loss": 1.5032,
      "step": 10676
    },
    {
      "epoch": 0.47476768485171844,
      "grad_norm": 0.06513024866580963,
      "learning_rate": 0.000577675727768241,
      "loss": 1.5039,
      "step": 10678
    },
    {
      "epoch": 0.4748566093103908,
      "grad_norm": 0.06479975581169128,
      "learning_rate": 0.0005775313107040483,
      "loss": 1.5033,
      "step": 10680
    },
    {
      "epoch": 0.4749455337690632,
      "grad_norm": 0.06562959402799606,
      "learning_rate": 0.0005773868870121377,
      "loss": 1.5035,
      "step": 10682
    },
    {
      "epoch": 0.47503445822773555,
      "grad_norm": 0.06198382005095482,
      "learning_rate": 0.0005772424567048549,
      "loss": 1.502,
      "step": 10684
    },
    {
      "epoch": 0.4751233826864079,
      "grad_norm": 0.0658661350607872,
      "learning_rate": 0.0005770980197945464,
      "loss": 1.5029,
      "step": 10686
    },
    {
      "epoch": 0.4752123071450803,
      "grad_norm": 0.06593360006809235,
      "learning_rate": 0.0005769535762935595,
      "loss": 1.5024,
      "step": 10688
    },
    {
      "epoch": 0.4753012316037526,
      "grad_norm": 0.06711971014738083,
      "learning_rate": 0.0005768091262142416,
      "loss": 1.5028,
      "step": 10690
    },
    {
      "epoch": 0.47539015606242496,
      "grad_norm": 0.06662214547395706,
      "learning_rate": 0.0005766646695689415,
      "loss": 1.4979,
      "step": 10692
    },
    {
      "epoch": 0.47547908052109733,
      "grad_norm": 0.062251124531030655,
      "learning_rate": 0.0005765202063700072,
      "loss": 1.5074,
      "step": 10694
    },
    {
      "epoch": 0.4755680049797697,
      "grad_norm": 0.0647764578461647,
      "learning_rate": 0.0005763757366297886,
      "loss": 1.5038,
      "step": 10696
    },
    {
      "epoch": 0.47565692943844207,
      "grad_norm": 0.06302308291196823,
      "learning_rate": 0.0005762312603606355,
      "loss": 1.4976,
      "step": 10698
    },
    {
      "epoch": 0.4757458538971144,
      "grad_norm": 0.06450463831424713,
      "learning_rate": 0.0005760867775748983,
      "loss": 1.502,
      "step": 10700
    },
    {
      "epoch": 0.47583477835578675,
      "grad_norm": 0.06613396853208542,
      "learning_rate": 0.000575942288284928,
      "loss": 1.4991,
      "step": 10702
    },
    {
      "epoch": 0.4759237028144591,
      "grad_norm": 0.06332564353942871,
      "learning_rate": 0.0005757977925030763,
      "loss": 1.5066,
      "step": 10704
    },
    {
      "epoch": 0.4760126272731315,
      "grad_norm": 0.06272386014461517,
      "learning_rate": 0.0005756532902416952,
      "loss": 1.5016,
      "step": 10706
    },
    {
      "epoch": 0.47610155173180385,
      "grad_norm": 0.06387588381767273,
      "learning_rate": 0.0005755087815131375,
      "loss": 1.501,
      "step": 10708
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 0.0668317899107933,
      "learning_rate": 0.0005753642663297564,
      "loss": 1.5068,
      "step": 10710
    },
    {
      "epoch": 0.47627940064914853,
      "grad_norm": 0.06176994368433952,
      "learning_rate": 0.0005752197447039056,
      "loss": 1.5082,
      "step": 10712
    },
    {
      "epoch": 0.4763683251078209,
      "grad_norm": 0.06459102034568787,
      "learning_rate": 0.0005750752166479397,
      "loss": 1.5063,
      "step": 10714
    },
    {
      "epoch": 0.47645724956649327,
      "grad_norm": 0.0622757263481617,
      "learning_rate": 0.0005749306821742132,
      "loss": 1.5021,
      "step": 10716
    },
    {
      "epoch": 0.47654617402516564,
      "grad_norm": 0.06583724915981293,
      "learning_rate": 0.0005747861412950821,
      "loss": 1.5038,
      "step": 10718
    },
    {
      "epoch": 0.476635098483838,
      "grad_norm": 0.0620274692773819,
      "learning_rate": 0.0005746415940229018,
      "loss": 1.5039,
      "step": 10720
    },
    {
      "epoch": 0.4767240229425103,
      "grad_norm": 0.06492220610380173,
      "learning_rate": 0.0005744970403700292,
      "loss": 1.5058,
      "step": 10722
    },
    {
      "epoch": 0.4768129474011827,
      "grad_norm": 0.06156022846698761,
      "learning_rate": 0.0005743524803488214,
      "loss": 1.5004,
      "step": 10724
    },
    {
      "epoch": 0.47690187185985505,
      "grad_norm": 0.06396898627281189,
      "learning_rate": 0.0005742079139716358,
      "loss": 1.502,
      "step": 10726
    },
    {
      "epoch": 0.4769907963185274,
      "grad_norm": 0.06404668837785721,
      "learning_rate": 0.0005740633412508307,
      "loss": 1.5032,
      "step": 10728
    },
    {
      "epoch": 0.4770797207771998,
      "grad_norm": 0.06537303328514099,
      "learning_rate": 0.0005739187621987648,
      "loss": 1.5036,
      "step": 10730
    },
    {
      "epoch": 0.4771686452358721,
      "grad_norm": 0.0657230019569397,
      "learning_rate": 0.0005737741768277974,
      "loss": 1.505,
      "step": 10732
    },
    {
      "epoch": 0.47725756969454447,
      "grad_norm": 0.06339870393276215,
      "learning_rate": 0.0005736295851502882,
      "loss": 1.5034,
      "step": 10734
    },
    {
      "epoch": 0.47734649415321684,
      "grad_norm": 0.06523014605045319,
      "learning_rate": 0.0005734849871785976,
      "loss": 1.5058,
      "step": 10736
    },
    {
      "epoch": 0.4774354186118892,
      "grad_norm": 0.06503810733556747,
      "learning_rate": 0.0005733403829250865,
      "loss": 1.5031,
      "step": 10738
    },
    {
      "epoch": 0.4775243430705616,
      "grad_norm": 0.0636671707034111,
      "learning_rate": 0.0005731957724021163,
      "loss": 1.5059,
      "step": 10740
    },
    {
      "epoch": 0.47761326752923394,
      "grad_norm": 0.06321322917938232,
      "learning_rate": 0.0005730511556220488,
      "loss": 1.503,
      "step": 10742
    },
    {
      "epoch": 0.47770219198790626,
      "grad_norm": 0.06607749313116074,
      "learning_rate": 0.0005729065325972467,
      "loss": 1.5087,
      "step": 10744
    },
    {
      "epoch": 0.4777911164465786,
      "grad_norm": 0.06520562618970871,
      "learning_rate": 0.0005727619033400729,
      "loss": 1.5022,
      "step": 10746
    },
    {
      "epoch": 0.477880040905251,
      "grad_norm": 0.06609973311424255,
      "learning_rate": 0.0005726172678628907,
      "loss": 1.5012,
      "step": 10748
    },
    {
      "epoch": 0.47796896536392336,
      "grad_norm": 0.06533106416463852,
      "learning_rate": 0.0005724726261780648,
      "loss": 1.4998,
      "step": 10750
    },
    {
      "epoch": 0.4780578898225957,
      "grad_norm": 0.06694469600915909,
      "learning_rate": 0.0005723279782979591,
      "loss": 1.5035,
      "step": 10752
    },
    {
      "epoch": 0.47814681428126804,
      "grad_norm": 0.06440529227256775,
      "learning_rate": 0.0005721833242349393,
      "loss": 1.5098,
      "step": 10754
    },
    {
      "epoch": 0.4782357387399404,
      "grad_norm": 0.06357917189598083,
      "learning_rate": 0.0005720386640013707,
      "loss": 1.5016,
      "step": 10756
    },
    {
      "epoch": 0.4783246631986128,
      "grad_norm": 0.06445612013339996,
      "learning_rate": 0.0005718939976096198,
      "loss": 1.5005,
      "step": 10758
    },
    {
      "epoch": 0.47841358765728514,
      "grad_norm": 0.062472015619277954,
      "learning_rate": 0.000571749325072053,
      "loss": 1.5029,
      "step": 10760
    },
    {
      "epoch": 0.4785025121159575,
      "grad_norm": 0.06607341021299362,
      "learning_rate": 0.0005716046464010378,
      "loss": 1.5072,
      "step": 10762
    },
    {
      "epoch": 0.4785914365746298,
      "grad_norm": 0.06257757544517517,
      "learning_rate": 0.0005714599616089419,
      "loss": 1.5015,
      "step": 10764
    },
    {
      "epoch": 0.4786803610333022,
      "grad_norm": 0.06556899100542068,
      "learning_rate": 0.0005713152707081335,
      "loss": 1.5052,
      "step": 10766
    },
    {
      "epoch": 0.47876928549197456,
      "grad_norm": 0.06689021736383438,
      "learning_rate": 0.0005711705737109816,
      "loss": 1.5016,
      "step": 10768
    },
    {
      "epoch": 0.47885820995064693,
      "grad_norm": 0.06304630637168884,
      "learning_rate": 0.0005710258706298553,
      "loss": 1.5011,
      "step": 10770
    },
    {
      "epoch": 0.4789471344093193,
      "grad_norm": 0.06447041779756546,
      "learning_rate": 0.0005708811614771245,
      "loss": 1.5073,
      "step": 10772
    },
    {
      "epoch": 0.47903605886799167,
      "grad_norm": 0.06245449185371399,
      "learning_rate": 0.0005707364462651598,
      "loss": 1.4974,
      "step": 10774
    },
    {
      "epoch": 0.479124983326664,
      "grad_norm": 0.063130222260952,
      "learning_rate": 0.000570591725006332,
      "loss": 1.4997,
      "step": 10776
    },
    {
      "epoch": 0.47921390778533635,
      "grad_norm": 0.06639711558818817,
      "learning_rate": 0.0005704469977130123,
      "loss": 1.5042,
      "step": 10778
    },
    {
      "epoch": 0.4793028322440087,
      "grad_norm": 0.06467228382825851,
      "learning_rate": 0.0005703022643975728,
      "loss": 1.5027,
      "step": 10780
    },
    {
      "epoch": 0.4793917567026811,
      "grad_norm": 0.06834383308887482,
      "learning_rate": 0.0005701575250723859,
      "loss": 1.4984,
      "step": 10782
    },
    {
      "epoch": 0.47948068116135345,
      "grad_norm": 0.0679764524102211,
      "learning_rate": 0.0005700127797498248,
      "loss": 1.5074,
      "step": 10784
    },
    {
      "epoch": 0.47956960562002576,
      "grad_norm": 0.06350118666887283,
      "learning_rate": 0.0005698680284422626,
      "loss": 1.4981,
      "step": 10786
    },
    {
      "epoch": 0.47965853007869813,
      "grad_norm": 0.06375767290592194,
      "learning_rate": 0.0005697232711620733,
      "loss": 1.5007,
      "step": 10788
    },
    {
      "epoch": 0.4797474545373705,
      "grad_norm": 0.0636902004480362,
      "learning_rate": 0.0005695785079216318,
      "loss": 1.5074,
      "step": 10790
    },
    {
      "epoch": 0.47983637899604287,
      "grad_norm": 0.06637261062860489,
      "learning_rate": 0.0005694337387333127,
      "loss": 1.5031,
      "step": 10792
    },
    {
      "epoch": 0.47992530345471524,
      "grad_norm": 0.06588244438171387,
      "learning_rate": 0.0005692889636094917,
      "loss": 1.5034,
      "step": 10794
    },
    {
      "epoch": 0.4800142279133876,
      "grad_norm": 0.06770049780607224,
      "learning_rate": 0.0005691441825625446,
      "loss": 1.5026,
      "step": 10796
    },
    {
      "epoch": 0.4801031523720599,
      "grad_norm": 0.06415392458438873,
      "learning_rate": 0.0005689993956048481,
      "loss": 1.5048,
      "step": 10798
    },
    {
      "epoch": 0.4801920768307323,
      "grad_norm": 0.06577417999505997,
      "learning_rate": 0.0005688546027487792,
      "loss": 1.5023,
      "step": 10800
    },
    {
      "epoch": 0.48028100128940465,
      "grad_norm": 0.06189163774251938,
      "learning_rate": 0.0005687098040067153,
      "loss": 1.5017,
      "step": 10802
    },
    {
      "epoch": 0.480369925748077,
      "grad_norm": 0.06363365054130554,
      "learning_rate": 0.0005685649993910348,
      "loss": 1.5037,
      "step": 10804
    },
    {
      "epoch": 0.4804588502067494,
      "grad_norm": 0.06573474407196045,
      "learning_rate": 0.0005684201889141158,
      "loss": 1.5027,
      "step": 10806
    },
    {
      "epoch": 0.4805477746654217,
      "grad_norm": 0.06577087193727493,
      "learning_rate": 0.0005682753725883378,
      "loss": 1.5003,
      "step": 10808
    },
    {
      "epoch": 0.48063669912409407,
      "grad_norm": 0.062234729528427124,
      "learning_rate": 0.0005681305504260798,
      "loss": 1.5009,
      "step": 10810
    },
    {
      "epoch": 0.48072562358276644,
      "grad_norm": 0.06282081454992294,
      "learning_rate": 0.0005679857224397222,
      "loss": 1.5035,
      "step": 10812
    },
    {
      "epoch": 0.4808145480414388,
      "grad_norm": 0.06254705041646957,
      "learning_rate": 0.0005678408886416454,
      "loss": 1.5049,
      "step": 10814
    },
    {
      "epoch": 0.4809034725001112,
      "grad_norm": 0.06438016891479492,
      "learning_rate": 0.0005676960490442305,
      "loss": 1.4978,
      "step": 10816
    },
    {
      "epoch": 0.4809923969587835,
      "grad_norm": 0.06419461965560913,
      "learning_rate": 0.0005675512036598592,
      "loss": 1.5005,
      "step": 10818
    },
    {
      "epoch": 0.48108132141745585,
      "grad_norm": 0.06625373661518097,
      "learning_rate": 0.000567406352500913,
      "loss": 1.4987,
      "step": 10820
    },
    {
      "epoch": 0.4811702458761282,
      "grad_norm": 0.06367786973714828,
      "learning_rate": 0.0005672614955797749,
      "loss": 1.5023,
      "step": 10822
    },
    {
      "epoch": 0.4812591703348006,
      "grad_norm": 0.06512659043073654,
      "learning_rate": 0.0005671166329088278,
      "loss": 1.5038,
      "step": 10824
    },
    {
      "epoch": 0.48134809479347296,
      "grad_norm": 0.06565241515636444,
      "learning_rate": 0.0005669717645004551,
      "loss": 1.5081,
      "step": 10826
    },
    {
      "epoch": 0.4814370192521453,
      "grad_norm": 0.06338914483785629,
      "learning_rate": 0.0005668268903670407,
      "loss": 1.5058,
      "step": 10828
    },
    {
      "epoch": 0.48152594371081764,
      "grad_norm": 0.06349223107099533,
      "learning_rate": 0.0005666820105209694,
      "loss": 1.4945,
      "step": 10830
    },
    {
      "epoch": 0.48161486816949,
      "grad_norm": 0.06533645838499069,
      "learning_rate": 0.0005665371249746259,
      "loss": 1.503,
      "step": 10832
    },
    {
      "epoch": 0.4817037926281624,
      "grad_norm": 0.07015854865312576,
      "learning_rate": 0.0005663922337403957,
      "loss": 1.5032,
      "step": 10834
    },
    {
      "epoch": 0.48179271708683474,
      "grad_norm": 0.06523853540420532,
      "learning_rate": 0.0005662473368306649,
      "loss": 1.5043,
      "step": 10836
    },
    {
      "epoch": 0.4818816415455071,
      "grad_norm": 0.06229966878890991,
      "learning_rate": 0.0005661024342578197,
      "loss": 1.5009,
      "step": 10838
    },
    {
      "epoch": 0.4819705660041794,
      "grad_norm": 0.06518285721540451,
      "learning_rate": 0.0005659575260342473,
      "loss": 1.5044,
      "step": 10840
    },
    {
      "epoch": 0.4820594904628518,
      "grad_norm": 0.06520578265190125,
      "learning_rate": 0.0005658126121723346,
      "loss": 1.5082,
      "step": 10842
    },
    {
      "epoch": 0.48214841492152416,
      "grad_norm": 0.06587293744087219,
      "learning_rate": 0.00056566769268447,
      "loss": 1.5047,
      "step": 10844
    },
    {
      "epoch": 0.48223733938019653,
      "grad_norm": 0.06606019288301468,
      "learning_rate": 0.0005655227675830416,
      "loss": 1.4998,
      "step": 10846
    },
    {
      "epoch": 0.4823262638388689,
      "grad_norm": 0.06386158615350723,
      "learning_rate": 0.0005653778368804381,
      "loss": 1.5081,
      "step": 10848
    },
    {
      "epoch": 0.48241518829754126,
      "grad_norm": 0.06474409252405167,
      "learning_rate": 0.0005652329005890492,
      "loss": 1.5008,
      "step": 10850
    },
    {
      "epoch": 0.4825041127562136,
      "grad_norm": 0.06231878697872162,
      "learning_rate": 0.0005650879587212645,
      "loss": 1.4979,
      "step": 10852
    },
    {
      "epoch": 0.48259303721488594,
      "grad_norm": 0.06352438777685165,
      "learning_rate": 0.0005649430112894743,
      "loss": 1.5012,
      "step": 10854
    },
    {
      "epoch": 0.4826819616735583,
      "grad_norm": 0.06429651379585266,
      "learning_rate": 0.000564798058306069,
      "loss": 1.4951,
      "step": 10856
    },
    {
      "epoch": 0.4827708861322307,
      "grad_norm": 0.06389694660902023,
      "learning_rate": 0.0005646530997834403,
      "loss": 1.4992,
      "step": 10858
    },
    {
      "epoch": 0.48285981059090305,
      "grad_norm": 0.06426005810499191,
      "learning_rate": 0.0005645081357339797,
      "loss": 1.4968,
      "step": 10860
    },
    {
      "epoch": 0.48294873504957536,
      "grad_norm": 0.062285613268613815,
      "learning_rate": 0.0005643631661700796,
      "loss": 1.4996,
      "step": 10862
    },
    {
      "epoch": 0.48303765950824773,
      "grad_norm": 0.06469501554965973,
      "learning_rate": 0.0005642181911041321,
      "loss": 1.4993,
      "step": 10864
    },
    {
      "epoch": 0.4831265839669201,
      "grad_norm": 0.06305573880672455,
      "learning_rate": 0.0005640732105485308,
      "loss": 1.5,
      "step": 10866
    },
    {
      "epoch": 0.48321550842559247,
      "grad_norm": 0.06352468580007553,
      "learning_rate": 0.000563928224515669,
      "loss": 1.5065,
      "step": 10868
    },
    {
      "epoch": 0.48330443288426483,
      "grad_norm": 0.06422882527112961,
      "learning_rate": 0.0005637832330179409,
      "loss": 1.5032,
      "step": 10870
    },
    {
      "epoch": 0.4833933573429372,
      "grad_norm": 0.06555035710334778,
      "learning_rate": 0.000563638236067741,
      "loss": 1.5019,
      "step": 10872
    },
    {
      "epoch": 0.4834822818016095,
      "grad_norm": 0.0639529749751091,
      "learning_rate": 0.0005634932336774641,
      "loss": 1.4982,
      "step": 10874
    },
    {
      "epoch": 0.4835712062602819,
      "grad_norm": 0.06562841683626175,
      "learning_rate": 0.0005633482258595059,
      "loss": 1.5008,
      "step": 10876
    },
    {
      "epoch": 0.48366013071895425,
      "grad_norm": 0.06432850658893585,
      "learning_rate": 0.0005632032126262622,
      "loss": 1.507,
      "step": 10878
    },
    {
      "epoch": 0.4837490551776266,
      "grad_norm": 0.06400411576032639,
      "learning_rate": 0.0005630581939901294,
      "loss": 1.5074,
      "step": 10880
    },
    {
      "epoch": 0.483837979636299,
      "grad_norm": 0.06370353698730469,
      "learning_rate": 0.0005629131699635041,
      "loss": 1.4988,
      "step": 10882
    },
    {
      "epoch": 0.4839269040949713,
      "grad_norm": 0.06292074173688889,
      "learning_rate": 0.0005627681405587839,
      "loss": 1.5069,
      "step": 10884
    },
    {
      "epoch": 0.48401582855364367,
      "grad_norm": 0.0631115660071373,
      "learning_rate": 0.0005626231057883664,
      "loss": 1.503,
      "step": 10886
    },
    {
      "epoch": 0.48410475301231604,
      "grad_norm": 0.06345234811306,
      "learning_rate": 0.0005624780656646499,
      "loss": 1.5007,
      "step": 10888
    },
    {
      "epoch": 0.4841936774709884,
      "grad_norm": 0.06285906583070755,
      "learning_rate": 0.000562333020200033,
      "loss": 1.5008,
      "step": 10890
    },
    {
      "epoch": 0.48428260192966077,
      "grad_norm": 0.06470154970884323,
      "learning_rate": 0.0005621879694069148,
      "loss": 1.5029,
      "step": 10892
    },
    {
      "epoch": 0.4843715263883331,
      "grad_norm": 0.06552711874246597,
      "learning_rate": 0.000562042913297695,
      "loss": 1.5065,
      "step": 10894
    },
    {
      "epoch": 0.48446045084700545,
      "grad_norm": 0.0655001625418663,
      "learning_rate": 0.0005618978518847733,
      "loss": 1.5033,
      "step": 10896
    },
    {
      "epoch": 0.4845493753056778,
      "grad_norm": 0.06529513746500015,
      "learning_rate": 0.0005617527851805507,
      "loss": 1.4996,
      "step": 10898
    },
    {
      "epoch": 0.4846382997643502,
      "grad_norm": 0.06527028232812881,
      "learning_rate": 0.0005616077131974279,
      "loss": 1.5027,
      "step": 10900
    },
    {
      "epoch": 0.48472722422302256,
      "grad_norm": 0.06357049942016602,
      "learning_rate": 0.0005614626359478062,
      "loss": 1.5025,
      "step": 10902
    },
    {
      "epoch": 0.4848161486816949,
      "grad_norm": 0.06293467432260513,
      "learning_rate": 0.0005613175534440875,
      "loss": 1.5036,
      "step": 10904
    },
    {
      "epoch": 0.48490507314036724,
      "grad_norm": 0.06417332589626312,
      "learning_rate": 0.0005611724656986741,
      "loss": 1.4987,
      "step": 10906
    },
    {
      "epoch": 0.4849939975990396,
      "grad_norm": 0.06176379695534706,
      "learning_rate": 0.0005610273727239688,
      "loss": 1.5024,
      "step": 10908
    },
    {
      "epoch": 0.485082922057712,
      "grad_norm": 0.06424736231565475,
      "learning_rate": 0.0005608822745323748,
      "loss": 1.5012,
      "step": 10910
    },
    {
      "epoch": 0.48517184651638434,
      "grad_norm": 0.06530416756868362,
      "learning_rate": 0.0005607371711362956,
      "loss": 1.4997,
      "step": 10912
    },
    {
      "epoch": 0.4852607709750567,
      "grad_norm": 0.06718974560499191,
      "learning_rate": 0.0005605920625481353,
      "loss": 1.5028,
      "step": 10914
    },
    {
      "epoch": 0.485349695433729,
      "grad_norm": 0.06613506376743317,
      "learning_rate": 0.0005604469487802987,
      "loss": 1.4958,
      "step": 10916
    },
    {
      "epoch": 0.4854386198924014,
      "grad_norm": 0.06338746100664139,
      "learning_rate": 0.0005603018298451903,
      "loss": 1.5018,
      "step": 10918
    },
    {
      "epoch": 0.48552754435107376,
      "grad_norm": 0.06582844257354736,
      "learning_rate": 0.0005601567057552158,
      "loss": 1.4997,
      "step": 10920
    },
    {
      "epoch": 0.4856164688097461,
      "grad_norm": 0.06286223977804184,
      "learning_rate": 0.000560011576522781,
      "loss": 1.4998,
      "step": 10922
    },
    {
      "epoch": 0.4857053932684185,
      "grad_norm": 0.06453262269496918,
      "learning_rate": 0.0005598664421602921,
      "loss": 1.4966,
      "step": 10924
    },
    {
      "epoch": 0.48579431772709086,
      "grad_norm": 0.06320101022720337,
      "learning_rate": 0.0005597213026801561,
      "loss": 1.5054,
      "step": 10926
    },
    {
      "epoch": 0.4858832421857632,
      "grad_norm": 0.06558766961097717,
      "learning_rate": 0.0005595761580947798,
      "loss": 1.5011,
      "step": 10928
    },
    {
      "epoch": 0.48597216664443554,
      "grad_norm": 0.06365552544593811,
      "learning_rate": 0.000559431008416571,
      "loss": 1.5008,
      "step": 10930
    },
    {
      "epoch": 0.4860610911031079,
      "grad_norm": 0.06486231088638306,
      "learning_rate": 0.0005592858536579377,
      "loss": 1.5008,
      "step": 10932
    },
    {
      "epoch": 0.4861500155617803,
      "grad_norm": 0.063105009496212,
      "learning_rate": 0.0005591406938312885,
      "loss": 1.5011,
      "step": 10934
    },
    {
      "epoch": 0.48623894002045265,
      "grad_norm": 0.06280262023210526,
      "learning_rate": 0.000558995528949032,
      "loss": 1.4973,
      "step": 10936
    },
    {
      "epoch": 0.48632786447912496,
      "grad_norm": 0.0647190734744072,
      "learning_rate": 0.0005588503590235777,
      "loss": 1.4982,
      "step": 10938
    },
    {
      "epoch": 0.48641678893779733,
      "grad_norm": 0.06445516645908356,
      "learning_rate": 0.0005587051840673355,
      "loss": 1.5,
      "step": 10940
    },
    {
      "epoch": 0.4865057133964697,
      "grad_norm": 0.06466984003782272,
      "learning_rate": 0.0005585600040927154,
      "loss": 1.4962,
      "step": 10942
    },
    {
      "epoch": 0.48659463785514206,
      "grad_norm": 0.06264392286539078,
      "learning_rate": 0.0005584148191121279,
      "loss": 1.4962,
      "step": 10944
    },
    {
      "epoch": 0.48668356231381443,
      "grad_norm": 0.06432507932186127,
      "learning_rate": 0.0005582696291379843,
      "loss": 1.4995,
      "step": 10946
    },
    {
      "epoch": 0.48677248677248675,
      "grad_norm": 0.0634993389248848,
      "learning_rate": 0.0005581244341826963,
      "loss": 1.5032,
      "step": 10948
    },
    {
      "epoch": 0.4868614112311591,
      "grad_norm": 0.06333111226558685,
      "learning_rate": 0.0005579792342586753,
      "loss": 1.5059,
      "step": 10950
    },
    {
      "epoch": 0.4869503356898315,
      "grad_norm": 0.06281735002994537,
      "learning_rate": 0.0005578340293783339,
      "loss": 1.5028,
      "step": 10952
    },
    {
      "epoch": 0.48703926014850385,
      "grad_norm": 0.06381440162658691,
      "learning_rate": 0.0005576888195540848,
      "loss": 1.5039,
      "step": 10954
    },
    {
      "epoch": 0.4871281846071762,
      "grad_norm": 0.06332245469093323,
      "learning_rate": 0.000557543604798341,
      "loss": 1.5009,
      "step": 10956
    },
    {
      "epoch": 0.4872171090658486,
      "grad_norm": 0.06438940018415451,
      "learning_rate": 0.0005573983851235165,
      "loss": 1.5028,
      "step": 10958
    },
    {
      "epoch": 0.4873060335245209,
      "grad_norm": 0.06432240456342697,
      "learning_rate": 0.000557253160542025,
      "loss": 1.4928,
      "step": 10960
    },
    {
      "epoch": 0.48739495798319327,
      "grad_norm": 0.06432882696390152,
      "learning_rate": 0.0005571079310662811,
      "loss": 1.5016,
      "step": 10962
    },
    {
      "epoch": 0.48748388244186563,
      "grad_norm": 0.06522729992866516,
      "learning_rate": 0.0005569626967086995,
      "loss": 1.5001,
      "step": 10964
    },
    {
      "epoch": 0.487572806900538,
      "grad_norm": 0.0628109946846962,
      "learning_rate": 0.0005568174574816957,
      "loss": 1.505,
      "step": 10966
    },
    {
      "epoch": 0.48766173135921037,
      "grad_norm": 0.06767455488443375,
      "learning_rate": 0.0005566722133976851,
      "loss": 1.5021,
      "step": 10968
    },
    {
      "epoch": 0.4877506558178827,
      "grad_norm": 0.06257057934999466,
      "learning_rate": 0.000556526964469084,
      "loss": 1.4986,
      "step": 10970
    },
    {
      "epoch": 0.48783958027655505,
      "grad_norm": 0.06524206697940826,
      "learning_rate": 0.0005563817107083088,
      "loss": 1.4929,
      "step": 10972
    },
    {
      "epoch": 0.4879285047352274,
      "grad_norm": 0.06345758587121964,
      "learning_rate": 0.0005562364521277766,
      "loss": 1.4987,
      "step": 10974
    },
    {
      "epoch": 0.4880174291938998,
      "grad_norm": 0.0644221380352974,
      "learning_rate": 0.0005560911887399047,
      "loss": 1.4985,
      "step": 10976
    },
    {
      "epoch": 0.48810635365257216,
      "grad_norm": 0.06262285262346268,
      "learning_rate": 0.0005559459205571106,
      "loss": 1.5021,
      "step": 10978
    },
    {
      "epoch": 0.4881952781112445,
      "grad_norm": 0.06352121382951736,
      "learning_rate": 0.0005558006475918128,
      "loss": 1.5005,
      "step": 10980
    },
    {
      "epoch": 0.48828420256991684,
      "grad_norm": 0.06309682130813599,
      "learning_rate": 0.0005556553698564297,
      "loss": 1.5002,
      "step": 10982
    },
    {
      "epoch": 0.4883731270285892,
      "grad_norm": 0.061160314828157425,
      "learning_rate": 0.0005555100873633804,
      "loss": 1.5016,
      "step": 10984
    },
    {
      "epoch": 0.4884620514872616,
      "grad_norm": 0.06492006033658981,
      "learning_rate": 0.0005553648001250842,
      "loss": 1.5017,
      "step": 10986
    },
    {
      "epoch": 0.48855097594593394,
      "grad_norm": 0.06279385089874268,
      "learning_rate": 0.0005552195081539608,
      "loss": 1.5062,
      "step": 10988
    },
    {
      "epoch": 0.4886399004046063,
      "grad_norm": 0.06334224343299866,
      "learning_rate": 0.0005550742114624305,
      "loss": 1.5027,
      "step": 10990
    },
    {
      "epoch": 0.4887288248632786,
      "grad_norm": 0.06543200463056564,
      "learning_rate": 0.000554928910062914,
      "loss": 1.5028,
      "step": 10992
    },
    {
      "epoch": 0.488817749321951,
      "grad_norm": 0.06258574873209,
      "learning_rate": 0.0005547836039678321,
      "loss": 1.5023,
      "step": 10994
    },
    {
      "epoch": 0.48890667378062336,
      "grad_norm": 0.06658318638801575,
      "learning_rate": 0.0005546382931896065,
      "loss": 1.4961,
      "step": 10996
    },
    {
      "epoch": 0.4889955982392957,
      "grad_norm": 0.062175750732421875,
      "learning_rate": 0.0005544929777406586,
      "loss": 1.5022,
      "step": 10998
    },
    {
      "epoch": 0.4890845226979681,
      "grad_norm": 0.06402657181024551,
      "learning_rate": 0.0005543476576334109,
      "loss": 1.4992,
      "step": 11000
    },
    {
      "epoch": 0.4890845226979681,
      "eval_loss": 1.4820914268493652,
      "eval_runtime": 13.0255,
      "eval_samples_per_second": 530.5,
      "eval_steps_per_second": 66.332,
      "step": 11000
    },
    {
      "epoch": 0.4891734471566404,
      "grad_norm": 0.06335395574569702,
      "learning_rate": 0.000554202332880286,
      "loss": 1.504,
      "step": 11002
    },
    {
      "epoch": 0.4892623716153128,
      "grad_norm": 0.06110154837369919,
      "learning_rate": 0.0005540570034937066,
      "loss": 1.5059,
      "step": 11004
    },
    {
      "epoch": 0.48935129607398514,
      "grad_norm": 0.06317976117134094,
      "learning_rate": 0.0005539116694860965,
      "loss": 1.4984,
      "step": 11006
    },
    {
      "epoch": 0.4894402205326575,
      "grad_norm": 0.06273622810840607,
      "learning_rate": 0.0005537663308698792,
      "loss": 1.4975,
      "step": 11008
    },
    {
      "epoch": 0.4895291449913299,
      "grad_norm": 0.06430622190237045,
      "learning_rate": 0.0005536209876574792,
      "loss": 1.5068,
      "step": 11010
    },
    {
      "epoch": 0.48961806945000225,
      "grad_norm": 0.06317053735256195,
      "learning_rate": 0.0005534756398613206,
      "loss": 1.4996,
      "step": 11012
    },
    {
      "epoch": 0.48970699390867456,
      "grad_norm": 0.063200943171978,
      "learning_rate": 0.0005533302874938289,
      "loss": 1.4999,
      "step": 11014
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 0.06409791111946106,
      "learning_rate": 0.0005531849305674292,
      "loss": 1.4959,
      "step": 11016
    },
    {
      "epoch": 0.4898848428260193,
      "grad_norm": 0.06358983367681503,
      "learning_rate": 0.000553039569094547,
      "loss": 1.501,
      "step": 11018
    },
    {
      "epoch": 0.48997376728469166,
      "grad_norm": 0.06684956699609756,
      "learning_rate": 0.0005528942030876089,
      "loss": 1.4998,
      "step": 11020
    },
    {
      "epoch": 0.49006269174336403,
      "grad_norm": 0.06377758830785751,
      "learning_rate": 0.0005527488325590411,
      "loss": 1.4958,
      "step": 11022
    },
    {
      "epoch": 0.49015161620203634,
      "grad_norm": 0.06464217603206635,
      "learning_rate": 0.0005526034575212708,
      "loss": 1.5029,
      "step": 11024
    },
    {
      "epoch": 0.4902405406607087,
      "grad_norm": 0.06466019153594971,
      "learning_rate": 0.000552458077986725,
      "loss": 1.499,
      "step": 11026
    },
    {
      "epoch": 0.4903294651193811,
      "grad_norm": 0.06569449603557587,
      "learning_rate": 0.0005523126939678316,
      "loss": 1.5023,
      "step": 11028
    },
    {
      "epoch": 0.49041838957805345,
      "grad_norm": 0.06310504674911499,
      "learning_rate": 0.0005521673054770185,
      "loss": 1.4994,
      "step": 11030
    },
    {
      "epoch": 0.4905073140367258,
      "grad_norm": 0.06455674022436142,
      "learning_rate": 0.0005520219125267144,
      "loss": 1.5022,
      "step": 11032
    },
    {
      "epoch": 0.4905962384953982,
      "grad_norm": 0.06450628489255905,
      "learning_rate": 0.0005518765151293478,
      "loss": 1.5005,
      "step": 11034
    },
    {
      "epoch": 0.4906851629540705,
      "grad_norm": 0.06352430582046509,
      "learning_rate": 0.0005517311132973481,
      "loss": 1.4988,
      "step": 11036
    },
    {
      "epoch": 0.49077408741274287,
      "grad_norm": 0.06309404969215393,
      "learning_rate": 0.0005515857070431448,
      "loss": 1.4998,
      "step": 11038
    },
    {
      "epoch": 0.49086301187141523,
      "grad_norm": 0.0660061240196228,
      "learning_rate": 0.000551440296379168,
      "loss": 1.5007,
      "step": 11040
    },
    {
      "epoch": 0.4909519363300876,
      "grad_norm": 0.06624384969472885,
      "learning_rate": 0.0005512948813178482,
      "loss": 1.4957,
      "step": 11042
    },
    {
      "epoch": 0.49104086078875997,
      "grad_norm": 0.06285151839256287,
      "learning_rate": 0.0005511494618716155,
      "loss": 1.5028,
      "step": 11044
    },
    {
      "epoch": 0.4911297852474323,
      "grad_norm": 0.06388713419437408,
      "learning_rate": 0.0005510040380529017,
      "loss": 1.5095,
      "step": 11046
    },
    {
      "epoch": 0.49121870970610465,
      "grad_norm": 0.06444618105888367,
      "learning_rate": 0.0005508586098741378,
      "loss": 1.4973,
      "step": 11048
    },
    {
      "epoch": 0.491307634164777,
      "grad_norm": 0.06315982341766357,
      "learning_rate": 0.0005507131773477558,
      "loss": 1.4944,
      "step": 11050
    },
    {
      "epoch": 0.4913965586234494,
      "grad_norm": 0.06305846571922302,
      "learning_rate": 0.000550567740486188,
      "loss": 1.5036,
      "step": 11052
    },
    {
      "epoch": 0.49148548308212175,
      "grad_norm": 0.06245909631252289,
      "learning_rate": 0.0005504222993018668,
      "loss": 1.4955,
      "step": 11054
    },
    {
      "epoch": 0.4915744075407941,
      "grad_norm": 0.0626964196562767,
      "learning_rate": 0.0005502768538072254,
      "loss": 1.4963,
      "step": 11056
    },
    {
      "epoch": 0.49166333199946644,
      "grad_norm": 0.06086695194244385,
      "learning_rate": 0.0005501314040146967,
      "loss": 1.4928,
      "step": 11058
    },
    {
      "epoch": 0.4917522564581388,
      "grad_norm": 0.06215178221464157,
      "learning_rate": 0.0005499859499367149,
      "loss": 1.4947,
      "step": 11060
    },
    {
      "epoch": 0.49184118091681117,
      "grad_norm": 0.06229309365153313,
      "learning_rate": 0.0005498404915857137,
      "loss": 1.4948,
      "step": 11062
    },
    {
      "epoch": 0.49193010537548354,
      "grad_norm": 0.06522365659475327,
      "learning_rate": 0.0005496950289741278,
      "loss": 1.5071,
      "step": 11064
    },
    {
      "epoch": 0.4920190298341559,
      "grad_norm": 0.06355579197406769,
      "learning_rate": 0.0005495495621143917,
      "loss": 1.5063,
      "step": 11066
    },
    {
      "epoch": 0.4921079542928282,
      "grad_norm": 0.06263679265975952,
      "learning_rate": 0.0005494040910189407,
      "loss": 1.4971,
      "step": 11068
    },
    {
      "epoch": 0.4921968787515006,
      "grad_norm": 0.06291796267032623,
      "learning_rate": 0.0005492586157002102,
      "loss": 1.5011,
      "step": 11070
    },
    {
      "epoch": 0.49228580321017296,
      "grad_norm": 0.0636981651186943,
      "learning_rate": 0.0005491131361706363,
      "loss": 1.4985,
      "step": 11072
    },
    {
      "epoch": 0.4923747276688453,
      "grad_norm": 0.061627645045518875,
      "learning_rate": 0.0005489676524426551,
      "loss": 1.5001,
      "step": 11074
    },
    {
      "epoch": 0.4924636521275177,
      "grad_norm": 0.06285221129655838,
      "learning_rate": 0.000548822164528703,
      "loss": 1.4997,
      "step": 11076
    },
    {
      "epoch": 0.49255257658619,
      "grad_norm": 0.06427198648452759,
      "learning_rate": 0.0005486766724412172,
      "loss": 1.4987,
      "step": 11078
    },
    {
      "epoch": 0.4926415010448624,
      "grad_norm": 0.06529872864484787,
      "learning_rate": 0.0005485311761926349,
      "loss": 1.5027,
      "step": 11080
    },
    {
      "epoch": 0.49273042550353474,
      "grad_norm": 0.06436862051486969,
      "learning_rate": 0.000548385675795394,
      "loss": 1.5018,
      "step": 11082
    },
    {
      "epoch": 0.4928193499622071,
      "grad_norm": 0.06122404336929321,
      "learning_rate": 0.000548240171261932,
      "loss": 1.5029,
      "step": 11084
    },
    {
      "epoch": 0.4929082744208795,
      "grad_norm": 0.0629076138138771,
      "learning_rate": 0.0005480946626046879,
      "loss": 1.5,
      "step": 11086
    },
    {
      "epoch": 0.49299719887955185,
      "grad_norm": 0.06222783774137497,
      "learning_rate": 0.0005479491498360997,
      "loss": 1.4984,
      "step": 11088
    },
    {
      "epoch": 0.49308612333822416,
      "grad_norm": 0.06436508893966675,
      "learning_rate": 0.0005478036329686071,
      "loss": 1.4938,
      "step": 11090
    },
    {
      "epoch": 0.4931750477968965,
      "grad_norm": 0.06390917301177979,
      "learning_rate": 0.0005476581120146493,
      "loss": 1.5018,
      "step": 11092
    },
    {
      "epoch": 0.4932639722555689,
      "grad_norm": 0.06670954078435898,
      "learning_rate": 0.0005475125869866661,
      "loss": 1.5033,
      "step": 11094
    },
    {
      "epoch": 0.49335289671424126,
      "grad_norm": 0.0621398501098156,
      "learning_rate": 0.0005473670578970975,
      "loss": 1.4996,
      "step": 11096
    },
    {
      "epoch": 0.49344182117291363,
      "grad_norm": 0.06525732576847076,
      "learning_rate": 0.000547221524758384,
      "loss": 1.4983,
      "step": 11098
    },
    {
      "epoch": 0.49353074563158594,
      "grad_norm": 0.06158006191253662,
      "learning_rate": 0.0005470759875829665,
      "loss": 1.4993,
      "step": 11100
    },
    {
      "epoch": 0.4936196700902583,
      "grad_norm": 0.06350939720869064,
      "learning_rate": 0.0005469304463832862,
      "loss": 1.5011,
      "step": 11102
    },
    {
      "epoch": 0.4937085945489307,
      "grad_norm": 0.06323396414518356,
      "learning_rate": 0.0005467849011717845,
      "loss": 1.5057,
      "step": 11104
    },
    {
      "epoch": 0.49379751900760305,
      "grad_norm": 0.06293822079896927,
      "learning_rate": 0.0005466393519609032,
      "loss": 1.5004,
      "step": 11106
    },
    {
      "epoch": 0.4938864434662754,
      "grad_norm": 0.06489180028438568,
      "learning_rate": 0.0005464937987630845,
      "loss": 1.5013,
      "step": 11108
    },
    {
      "epoch": 0.4939753679249478,
      "grad_norm": 0.06281109154224396,
      "learning_rate": 0.0005463482415907713,
      "loss": 1.4972,
      "step": 11110
    },
    {
      "epoch": 0.4940642923836201,
      "grad_norm": 0.06400121748447418,
      "learning_rate": 0.0005462026804564058,
      "loss": 1.5008,
      "step": 11112
    },
    {
      "epoch": 0.49415321684229246,
      "grad_norm": 0.06272722035646439,
      "learning_rate": 0.0005460571153724318,
      "loss": 1.4962,
      "step": 11114
    },
    {
      "epoch": 0.49424214130096483,
      "grad_norm": 0.06551103293895721,
      "learning_rate": 0.0005459115463512925,
      "loss": 1.5093,
      "step": 11116
    },
    {
      "epoch": 0.4943310657596372,
      "grad_norm": 0.06318585574626923,
      "learning_rate": 0.000545765973405432,
      "loss": 1.493,
      "step": 11118
    },
    {
      "epoch": 0.49441999021830957,
      "grad_norm": 0.06164968013763428,
      "learning_rate": 0.0005456203965472944,
      "loss": 1.4923,
      "step": 11120
    },
    {
      "epoch": 0.4945089146769819,
      "grad_norm": 0.06324370205402374,
      "learning_rate": 0.0005454748157893241,
      "loss": 1.5067,
      "step": 11122
    },
    {
      "epoch": 0.49459783913565425,
      "grad_norm": 0.0638519749045372,
      "learning_rate": 0.0005453292311439662,
      "loss": 1.4976,
      "step": 11124
    },
    {
      "epoch": 0.4946867635943266,
      "grad_norm": 0.0644693598151207,
      "learning_rate": 0.000545183642623666,
      "loss": 1.498,
      "step": 11126
    },
    {
      "epoch": 0.494775688052999,
      "grad_norm": 0.06370065361261368,
      "learning_rate": 0.0005450380502408688,
      "loss": 1.4954,
      "step": 11128
    },
    {
      "epoch": 0.49486461251167135,
      "grad_norm": 0.06507989764213562,
      "learning_rate": 0.0005448924540080206,
      "loss": 1.5037,
      "step": 11130
    },
    {
      "epoch": 0.49495353697034367,
      "grad_norm": 0.06500361114740372,
      "learning_rate": 0.0005447468539375675,
      "loss": 1.4998,
      "step": 11132
    },
    {
      "epoch": 0.49504246142901603,
      "grad_norm": 0.06104636937379837,
      "learning_rate": 0.0005446012500419564,
      "loss": 1.4931,
      "step": 11134
    },
    {
      "epoch": 0.4951313858876884,
      "grad_norm": 0.06581141799688339,
      "learning_rate": 0.0005444556423336337,
      "loss": 1.4986,
      "step": 11136
    },
    {
      "epoch": 0.49522031034636077,
      "grad_norm": 0.06393726170063019,
      "learning_rate": 0.0005443100308250467,
      "loss": 1.4976,
      "step": 11138
    },
    {
      "epoch": 0.49530923480503314,
      "grad_norm": 0.06342989206314087,
      "learning_rate": 0.0005441644155286432,
      "loss": 1.4973,
      "step": 11140
    },
    {
      "epoch": 0.4953981592637055,
      "grad_norm": 0.06588321924209595,
      "learning_rate": 0.0005440187964568708,
      "loss": 1.5007,
      "step": 11142
    },
    {
      "epoch": 0.4954870837223778,
      "grad_norm": 0.06536178290843964,
      "learning_rate": 0.0005438731736221776,
      "loss": 1.5027,
      "step": 11144
    },
    {
      "epoch": 0.4955760081810502,
      "grad_norm": 0.06368371844291687,
      "learning_rate": 0.0005437275470370122,
      "loss": 1.5003,
      "step": 11146
    },
    {
      "epoch": 0.49566493263972256,
      "grad_norm": 0.06588061898946762,
      "learning_rate": 0.0005435819167138234,
      "loss": 1.5015,
      "step": 11148
    },
    {
      "epoch": 0.4957538570983949,
      "grad_norm": 0.06427399069070816,
      "learning_rate": 0.0005434362826650603,
      "loss": 1.5013,
      "step": 11150
    },
    {
      "epoch": 0.4958427815570673,
      "grad_norm": 0.0646776631474495,
      "learning_rate": 0.0005432906449031723,
      "loss": 1.5026,
      "step": 11152
    },
    {
      "epoch": 0.4959317060157396,
      "grad_norm": 0.06432519853115082,
      "learning_rate": 0.0005431450034406092,
      "loss": 1.5005,
      "step": 11154
    },
    {
      "epoch": 0.49602063047441197,
      "grad_norm": 0.06441538780927658,
      "learning_rate": 0.0005429993582898212,
      "loss": 1.4945,
      "step": 11156
    },
    {
      "epoch": 0.49610955493308434,
      "grad_norm": 0.06345316767692566,
      "learning_rate": 0.0005428537094632585,
      "loss": 1.495,
      "step": 11158
    },
    {
      "epoch": 0.4961984793917567,
      "grad_norm": 0.06269050389528275,
      "learning_rate": 0.0005427080569733718,
      "loss": 1.494,
      "step": 11160
    },
    {
      "epoch": 0.4962874038504291,
      "grad_norm": 0.06417939811944962,
      "learning_rate": 0.0005425624008326122,
      "loss": 1.4962,
      "step": 11162
    },
    {
      "epoch": 0.49637632830910144,
      "grad_norm": 0.06371048837900162,
      "learning_rate": 0.0005424167410534311,
      "loss": 1.5057,
      "step": 11164
    },
    {
      "epoch": 0.49646525276777376,
      "grad_norm": 0.06224866583943367,
      "learning_rate": 0.0005422710776482798,
      "loss": 1.5009,
      "step": 11166
    },
    {
      "epoch": 0.4965541772264461,
      "grad_norm": 0.06159532442688942,
      "learning_rate": 0.0005421254106296108,
      "loss": 1.4936,
      "step": 11168
    },
    {
      "epoch": 0.4966431016851185,
      "grad_norm": 0.06550117582082748,
      "learning_rate": 0.0005419797400098758,
      "loss": 1.4971,
      "step": 11170
    },
    {
      "epoch": 0.49673202614379086,
      "grad_norm": 0.06186149641871452,
      "learning_rate": 0.0005418340658015279,
      "loss": 1.4973,
      "step": 11172
    },
    {
      "epoch": 0.49682095060246323,
      "grad_norm": 0.06517408788204193,
      "learning_rate": 0.0005416883880170195,
      "loss": 1.5007,
      "step": 11174
    },
    {
      "epoch": 0.49690987506113554,
      "grad_norm": 0.06351044774055481,
      "learning_rate": 0.000541542706668804,
      "loss": 1.5013,
      "step": 11176
    },
    {
      "epoch": 0.4969987995198079,
      "grad_norm": 0.06350798904895782,
      "learning_rate": 0.0005413970217693348,
      "loss": 1.5037,
      "step": 11178
    },
    {
      "epoch": 0.4970877239784803,
      "grad_norm": 0.06402333080768585,
      "learning_rate": 0.0005412513333310661,
      "loss": 1.4991,
      "step": 11180
    },
    {
      "epoch": 0.49717664843715265,
      "grad_norm": 0.06381136924028397,
      "learning_rate": 0.0005411056413664512,
      "loss": 1.4986,
      "step": 11182
    },
    {
      "epoch": 0.497265572895825,
      "grad_norm": 0.06557981669902802,
      "learning_rate": 0.0005409599458879452,
      "loss": 1.4981,
      "step": 11184
    },
    {
      "epoch": 0.4973544973544973,
      "grad_norm": 0.06369341164827347,
      "learning_rate": 0.0005408142469080023,
      "loss": 1.4985,
      "step": 11186
    },
    {
      "epoch": 0.4974434218131697,
      "grad_norm": 0.06300416588783264,
      "learning_rate": 0.0005406685444390779,
      "loss": 1.5046,
      "step": 11188
    },
    {
      "epoch": 0.49753234627184206,
      "grad_norm": 0.0626647025346756,
      "learning_rate": 0.0005405228384936272,
      "loss": 1.4958,
      "step": 11190
    },
    {
      "epoch": 0.49762127073051443,
      "grad_norm": 0.06517183035612106,
      "learning_rate": 0.0005403771290841053,
      "loss": 1.5032,
      "step": 11192
    },
    {
      "epoch": 0.4977101951891868,
      "grad_norm": 0.06442451477050781,
      "learning_rate": 0.0005402314162229688,
      "loss": 1.4977,
      "step": 11194
    },
    {
      "epoch": 0.49779911964785917,
      "grad_norm": 0.06301950663328171,
      "learning_rate": 0.0005400856999226735,
      "loss": 1.4991,
      "step": 11196
    },
    {
      "epoch": 0.4978880441065315,
      "grad_norm": 0.06560328602790833,
      "learning_rate": 0.000539939980195676,
      "loss": 1.5007,
      "step": 11198
    },
    {
      "epoch": 0.49797696856520385,
      "grad_norm": 0.06166763976216316,
      "learning_rate": 0.0005397942570544327,
      "loss": 1.4942,
      "step": 11200
    },
    {
      "epoch": 0.4980658930238762,
      "grad_norm": 0.06296875327825546,
      "learning_rate": 0.0005396485305114011,
      "loss": 1.4945,
      "step": 11202
    },
    {
      "epoch": 0.4981548174825486,
      "grad_norm": 0.06445491313934326,
      "learning_rate": 0.0005395028005790384,
      "loss": 1.5026,
      "step": 11204
    },
    {
      "epoch": 0.49824374194122095,
      "grad_norm": 0.06422359496355057,
      "learning_rate": 0.000539357067269802,
      "loss": 1.4957,
      "step": 11206
    },
    {
      "epoch": 0.49833266639989326,
      "grad_norm": 0.06359175592660904,
      "learning_rate": 0.0005392113305961503,
      "loss": 1.4982,
      "step": 11208
    },
    {
      "epoch": 0.49842159085856563,
      "grad_norm": 0.061649519950151443,
      "learning_rate": 0.0005390655905705413,
      "loss": 1.4979,
      "step": 11210
    },
    {
      "epoch": 0.498510515317238,
      "grad_norm": 0.06346156448125839,
      "learning_rate": 0.0005389198472054334,
      "loss": 1.5022,
      "step": 11212
    },
    {
      "epoch": 0.49859943977591037,
      "grad_norm": 0.06347343325614929,
      "learning_rate": 0.0005387741005132853,
      "loss": 1.4976,
      "step": 11214
    },
    {
      "epoch": 0.49868836423458274,
      "grad_norm": 0.06473079323768616,
      "learning_rate": 0.0005386283505065565,
      "loss": 1.4985,
      "step": 11216
    },
    {
      "epoch": 0.4987772886932551,
      "grad_norm": 0.0626574233174324,
      "learning_rate": 0.0005384825971977059,
      "loss": 1.4948,
      "step": 11218
    },
    {
      "epoch": 0.4988662131519274,
      "grad_norm": 0.06385411322116852,
      "learning_rate": 0.0005383368405991932,
      "loss": 1.5038,
      "step": 11220
    },
    {
      "epoch": 0.4989551376105998,
      "grad_norm": 0.0622217021882534,
      "learning_rate": 0.0005381910807234785,
      "loss": 1.4998,
      "step": 11222
    },
    {
      "epoch": 0.49904406206927215,
      "grad_norm": 0.06290595233440399,
      "learning_rate": 0.0005380453175830219,
      "loss": 1.4955,
      "step": 11224
    },
    {
      "epoch": 0.4991329865279445,
      "grad_norm": 0.06268318742513657,
      "learning_rate": 0.0005378995511902841,
      "loss": 1.4992,
      "step": 11226
    },
    {
      "epoch": 0.4992219109866169,
      "grad_norm": 0.06381423771381378,
      "learning_rate": 0.0005377537815577254,
      "loss": 1.4967,
      "step": 11228
    },
    {
      "epoch": 0.4993108354452892,
      "grad_norm": 0.06410021334886551,
      "learning_rate": 0.0005376080086978072,
      "loss": 1.4954,
      "step": 11230
    },
    {
      "epoch": 0.49939975990396157,
      "grad_norm": 0.06397207826375961,
      "learning_rate": 0.0005374622326229906,
      "loss": 1.5047,
      "step": 11232
    },
    {
      "epoch": 0.49948868436263394,
      "grad_norm": 0.06497202068567276,
      "learning_rate": 0.0005373164533457374,
      "loss": 1.5001,
      "step": 11234
    },
    {
      "epoch": 0.4995776088213063,
      "grad_norm": 0.06162360683083534,
      "learning_rate": 0.0005371706708785091,
      "loss": 1.4959,
      "step": 11236
    },
    {
      "epoch": 0.4996665332799787,
      "grad_norm": 0.06346426904201508,
      "learning_rate": 0.0005370248852337682,
      "loss": 1.4943,
      "step": 11238
    },
    {
      "epoch": 0.49975545773865104,
      "grad_norm": 0.06419111043214798,
      "learning_rate": 0.000536879096423977,
      "loss": 1.5007,
      "step": 11240
    },
    {
      "epoch": 0.49984438219732336,
      "grad_norm": 0.06275220960378647,
      "learning_rate": 0.0005367333044615979,
      "loss": 1.4958,
      "step": 11242
    },
    {
      "epoch": 0.4999333066559957,
      "grad_norm": 0.06255465000867844,
      "learning_rate": 0.0005365875093590944,
      "loss": 1.4995,
      "step": 11244
    },
    {
      "epoch": 0.500022231114668,
      "grad_norm": 0.06293238699436188,
      "learning_rate": 0.000536441711128929,
      "loss": 1.5021,
      "step": 11246
    },
    {
      "epoch": 0.5001111555733404,
      "grad_norm": 0.06618613004684448,
      "learning_rate": 0.0005362959097835658,
      "loss": 1.4973,
      "step": 11248
    },
    {
      "epoch": 0.5002000800320128,
      "grad_norm": 0.06275363266468048,
      "learning_rate": 0.0005361501053354681,
      "loss": 1.495,
      "step": 11250
    },
    {
      "epoch": 0.5002890044906851,
      "grad_norm": 0.06456315517425537,
      "learning_rate": 0.0005360042977971002,
      "loss": 1.496,
      "step": 11252
    },
    {
      "epoch": 0.5003779289493575,
      "grad_norm": 0.06540905684232712,
      "learning_rate": 0.0005358584871809262,
      "loss": 1.4952,
      "step": 11254
    },
    {
      "epoch": 0.5004668534080299,
      "grad_norm": 0.06401478499174118,
      "learning_rate": 0.0005357126734994108,
      "loss": 1.4914,
      "step": 11256
    },
    {
      "epoch": 0.5005557778667022,
      "grad_norm": 0.06477171927690506,
      "learning_rate": 0.0005355668567650186,
      "loss": 1.5019,
      "step": 11258
    },
    {
      "epoch": 0.5006447023253746,
      "grad_norm": 0.06348700076341629,
      "learning_rate": 0.0005354210369902147,
      "loss": 1.4993,
      "step": 11260
    },
    {
      "epoch": 0.500733626784047,
      "grad_norm": 0.06129096448421478,
      "learning_rate": 0.0005352752141874645,
      "loss": 1.4959,
      "step": 11262
    },
    {
      "epoch": 0.5008225512427193,
      "grad_norm": 0.06324250996112823,
      "learning_rate": 0.0005351293883692335,
      "loss": 1.4915,
      "step": 11264
    },
    {
      "epoch": 0.5009114757013917,
      "grad_norm": 0.0653425082564354,
      "learning_rate": 0.0005349835595479879,
      "loss": 1.4971,
      "step": 11266
    },
    {
      "epoch": 0.501000400160064,
      "grad_norm": 0.06187792494893074,
      "learning_rate": 0.000534837727736193,
      "loss": 1.4998,
      "step": 11268
    },
    {
      "epoch": 0.5010893246187363,
      "grad_norm": 0.06403960287570953,
      "learning_rate": 0.0005346918929463159,
      "loss": 1.5046,
      "step": 11270
    },
    {
      "epoch": 0.5011782490774087,
      "grad_norm": 0.06291434913873672,
      "learning_rate": 0.0005345460551908227,
      "loss": 1.4957,
      "step": 11272
    },
    {
      "epoch": 0.5012671735360811,
      "grad_norm": 0.06603266298770905,
      "learning_rate": 0.0005344002144821806,
      "loss": 1.4996,
      "step": 11274
    },
    {
      "epoch": 0.5013560979947534,
      "grad_norm": 0.06421004980802536,
      "learning_rate": 0.0005342543708328566,
      "loss": 1.5001,
      "step": 11276
    },
    {
      "epoch": 0.5014450224534258,
      "grad_norm": 0.06628622859716415,
      "learning_rate": 0.0005341085242553179,
      "loss": 1.4974,
      "step": 11278
    },
    {
      "epoch": 0.5015339469120982,
      "grad_norm": 0.06342728435993195,
      "learning_rate": 0.0005339626747620323,
      "loss": 1.4976,
      "step": 11280
    },
    {
      "epoch": 0.5016228713707706,
      "grad_norm": 0.0662439614534378,
      "learning_rate": 0.0005338168223654676,
      "loss": 1.4981,
      "step": 11282
    },
    {
      "epoch": 0.5017117958294429,
      "grad_norm": 0.06446612626314163,
      "learning_rate": 0.000533670967078092,
      "loss": 1.4992,
      "step": 11284
    },
    {
      "epoch": 0.5018007202881153,
      "grad_norm": 0.06358826905488968,
      "learning_rate": 0.0005335251089123734,
      "loss": 1.5075,
      "step": 11286
    },
    {
      "epoch": 0.5018896447467877,
      "grad_norm": 0.06295160949230194,
      "learning_rate": 0.0005333792478807812,
      "loss": 1.5024,
      "step": 11288
    },
    {
      "epoch": 0.5019785692054599,
      "grad_norm": 0.06541930884122849,
      "learning_rate": 0.0005332333839957834,
      "loss": 1.4928,
      "step": 11290
    },
    {
      "epoch": 0.5020674936641323,
      "grad_norm": 0.06382886320352554,
      "learning_rate": 0.0005330875172698498,
      "loss": 1.4973,
      "step": 11292
    },
    {
      "epoch": 0.5021564181228046,
      "grad_norm": 0.06370875984430313,
      "learning_rate": 0.0005329416477154492,
      "loss": 1.4947,
      "step": 11294
    },
    {
      "epoch": 0.502245342581477,
      "grad_norm": 0.06501110643148422,
      "learning_rate": 0.0005327957753450515,
      "loss": 1.4953,
      "step": 11296
    },
    {
      "epoch": 0.5023342670401494,
      "grad_norm": 0.06344249099493027,
      "learning_rate": 0.0005326499001711264,
      "loss": 1.4958,
      "step": 11298
    },
    {
      "epoch": 0.5024231914988218,
      "grad_norm": 0.0630435198545456,
      "learning_rate": 0.0005325040222061438,
      "loss": 1.5025,
      "step": 11300
    },
    {
      "epoch": 0.5025121159574941,
      "grad_norm": 0.06559957563877106,
      "learning_rate": 0.0005323581414625743,
      "loss": 1.4961,
      "step": 11302
    },
    {
      "epoch": 0.5026010404161665,
      "grad_norm": 0.06591141223907471,
      "learning_rate": 0.0005322122579528883,
      "loss": 1.4967,
      "step": 11304
    },
    {
      "epoch": 0.5026899648748389,
      "grad_norm": 0.06230854615569115,
      "learning_rate": 0.0005320663716895565,
      "loss": 1.4959,
      "step": 11306
    },
    {
      "epoch": 0.5027788893335112,
      "grad_norm": 0.06417876482009888,
      "learning_rate": 0.0005319204826850498,
      "loss": 1.5024,
      "step": 11308
    },
    {
      "epoch": 0.5028678137921836,
      "grad_norm": 0.06187916174530983,
      "learning_rate": 0.0005317745909518397,
      "loss": 1.4925,
      "step": 11310
    },
    {
      "epoch": 0.5029567382508559,
      "grad_norm": 0.06452244520187378,
      "learning_rate": 0.0005316286965023976,
      "loss": 1.5,
      "step": 11312
    },
    {
      "epoch": 0.5030456627095282,
      "grad_norm": 0.064858078956604,
      "learning_rate": 0.0005314827993491948,
      "loss": 1.4968,
      "step": 11314
    },
    {
      "epoch": 0.5031345871682006,
      "grad_norm": 0.0635191798210144,
      "learning_rate": 0.0005313368995047039,
      "loss": 1.4931,
      "step": 11316
    },
    {
      "epoch": 0.503223511626873,
      "grad_norm": 0.06416906416416168,
      "learning_rate": 0.0005311909969813966,
      "loss": 1.4993,
      "step": 11318
    },
    {
      "epoch": 0.5033124360855453,
      "grad_norm": 0.06357254087924957,
      "learning_rate": 0.0005310450917917456,
      "loss": 1.4931,
      "step": 11320
    },
    {
      "epoch": 0.5034013605442177,
      "grad_norm": 0.06390412896871567,
      "learning_rate": 0.0005308991839482231,
      "loss": 1.498,
      "step": 11322
    },
    {
      "epoch": 0.5034902850028901,
      "grad_norm": 0.06369391083717346,
      "learning_rate": 0.0005307532734633024,
      "loss": 1.5001,
      "step": 11324
    },
    {
      "epoch": 0.5035792094615624,
      "grad_norm": 0.06292715668678284,
      "learning_rate": 0.0005306073603494561,
      "loss": 1.499,
      "step": 11326
    },
    {
      "epoch": 0.5036681339202348,
      "grad_norm": 0.06275395303964615,
      "learning_rate": 0.0005304614446191581,
      "loss": 1.5005,
      "step": 11328
    },
    {
      "epoch": 0.5037570583789072,
      "grad_norm": 0.06449761241674423,
      "learning_rate": 0.0005303155262848813,
      "loss": 1.4961,
      "step": 11330
    },
    {
      "epoch": 0.5038459828375795,
      "grad_norm": 0.06431379169225693,
      "learning_rate": 0.0005301696053590999,
      "loss": 1.4974,
      "step": 11332
    },
    {
      "epoch": 0.5039349072962518,
      "grad_norm": 0.06470140814781189,
      "learning_rate": 0.0005300236818542876,
      "loss": 1.4957,
      "step": 11334
    },
    {
      "epoch": 0.5040238317549242,
      "grad_norm": 0.06437893211841583,
      "learning_rate": 0.0005298777557829186,
      "loss": 1.4951,
      "step": 11336
    },
    {
      "epoch": 0.5041127562135965,
      "grad_norm": 0.06438335031270981,
      "learning_rate": 0.0005297318271574675,
      "loss": 1.4902,
      "step": 11338
    },
    {
      "epoch": 0.5042016806722689,
      "grad_norm": 0.06443442404270172,
      "learning_rate": 0.0005295858959904086,
      "loss": 1.4999,
      "step": 11340
    },
    {
      "epoch": 0.5042906051309413,
      "grad_norm": 0.06223178282380104,
      "learning_rate": 0.0005294399622942172,
      "loss": 1.5001,
      "step": 11342
    },
    {
      "epoch": 0.5043795295896136,
      "grad_norm": 0.06361085921525955,
      "learning_rate": 0.0005292940260813678,
      "loss": 1.4937,
      "step": 11344
    },
    {
      "epoch": 0.504468454048286,
      "grad_norm": 0.06470072269439697,
      "learning_rate": 0.0005291480873643362,
      "loss": 1.5009,
      "step": 11346
    },
    {
      "epoch": 0.5045573785069584,
      "grad_norm": 0.05959364399313927,
      "learning_rate": 0.0005290021461555974,
      "loss": 1.4964,
      "step": 11348
    },
    {
      "epoch": 0.5046463029656307,
      "grad_norm": 0.06182974949479103,
      "learning_rate": 0.0005288562024676275,
      "loss": 1.4977,
      "step": 11350
    },
    {
      "epoch": 0.5047352274243031,
      "grad_norm": 0.06326933205127716,
      "learning_rate": 0.0005287102563129023,
      "loss": 1.5012,
      "step": 11352
    },
    {
      "epoch": 0.5048241518829754,
      "grad_norm": 0.06258763372898102,
      "learning_rate": 0.0005285643077038977,
      "loss": 1.498,
      "step": 11354
    },
    {
      "epoch": 0.5049130763416477,
      "grad_norm": 0.06452583521604538,
      "learning_rate": 0.0005284183566530902,
      "loss": 1.4936,
      "step": 11356
    },
    {
      "epoch": 0.5050020008003201,
      "grad_norm": 0.06566081196069717,
      "learning_rate": 0.0005282724031729565,
      "loss": 1.4988,
      "step": 11358
    },
    {
      "epoch": 0.5050909252589925,
      "grad_norm": 0.06276465952396393,
      "learning_rate": 0.0005281264472759732,
      "loss": 1.4944,
      "step": 11360
    },
    {
      "epoch": 0.5051798497176648,
      "grad_norm": 0.061030905693769455,
      "learning_rate": 0.0005279804889746169,
      "loss": 1.4979,
      "step": 11362
    },
    {
      "epoch": 0.5052687741763372,
      "grad_norm": 0.06379514187574387,
      "learning_rate": 0.0005278345282813655,
      "loss": 1.4952,
      "step": 11364
    },
    {
      "epoch": 0.5053576986350096,
      "grad_norm": 0.06288314610719681,
      "learning_rate": 0.0005276885652086958,
      "loss": 1.4891,
      "step": 11366
    },
    {
      "epoch": 0.5054466230936819,
      "grad_norm": 0.061497218906879425,
      "learning_rate": 0.0005275425997690854,
      "loss": 1.5001,
      "step": 11368
    },
    {
      "epoch": 0.5055355475523543,
      "grad_norm": 0.06392903625965118,
      "learning_rate": 0.0005273966319750123,
      "loss": 1.4953,
      "step": 11370
    },
    {
      "epoch": 0.5056244720110267,
      "grad_norm": 0.06301810592412949,
      "learning_rate": 0.0005272506618389544,
      "loss": 1.4944,
      "step": 11372
    },
    {
      "epoch": 0.505713396469699,
      "grad_norm": 0.06324661523103714,
      "learning_rate": 0.00052710468937339,
      "loss": 1.499,
      "step": 11374
    },
    {
      "epoch": 0.5058023209283713,
      "grad_norm": 0.06164475902915001,
      "learning_rate": 0.0005269587145907969,
      "loss": 1.4962,
      "step": 11376
    },
    {
      "epoch": 0.5058912453870437,
      "grad_norm": 0.060896288603544235,
      "learning_rate": 0.0005268127375036545,
      "loss": 1.5005,
      "step": 11378
    },
    {
      "epoch": 0.505980169845716,
      "grad_norm": 0.06393571943044662,
      "learning_rate": 0.0005266667581244408,
      "loss": 1.5005,
      "step": 11380
    },
    {
      "epoch": 0.5060690943043884,
      "grad_norm": 0.06255924701690674,
      "learning_rate": 0.0005265207764656353,
      "loss": 1.4963,
      "step": 11382
    },
    {
      "epoch": 0.5061580187630608,
      "grad_norm": 0.06395517289638519,
      "learning_rate": 0.0005263747925397168,
      "loss": 1.4931,
      "step": 11384
    },
    {
      "epoch": 0.5062469432217331,
      "grad_norm": 0.0635991245508194,
      "learning_rate": 0.0005262288063591649,
      "loss": 1.4972,
      "step": 11386
    },
    {
      "epoch": 0.5063358676804055,
      "grad_norm": 0.06476542353630066,
      "learning_rate": 0.0005260828179364591,
      "loss": 1.4942,
      "step": 11388
    },
    {
      "epoch": 0.5064247921390779,
      "grad_norm": 0.06474872678518295,
      "learning_rate": 0.000525936827284079,
      "loss": 1.4969,
      "step": 11390
    },
    {
      "epoch": 0.5065137165977502,
      "grad_norm": 0.06325345486402512,
      "learning_rate": 0.0005257908344145046,
      "loss": 1.4924,
      "step": 11392
    },
    {
      "epoch": 0.5066026410564226,
      "grad_norm": 0.06438945233821869,
      "learning_rate": 0.0005256448393402161,
      "loss": 1.4962,
      "step": 11394
    },
    {
      "epoch": 0.506691565515095,
      "grad_norm": 0.06563424319028854,
      "learning_rate": 0.0005254988420736935,
      "loss": 1.5001,
      "step": 11396
    },
    {
      "epoch": 0.5067804899737672,
      "grad_norm": 0.06430519372224808,
      "learning_rate": 0.0005253528426274178,
      "loss": 1.4977,
      "step": 11398
    },
    {
      "epoch": 0.5068694144324396,
      "grad_norm": 0.06316456943750381,
      "learning_rate": 0.0005252068410138693,
      "loss": 1.4993,
      "step": 11400
    },
    {
      "epoch": 0.506958338891112,
      "grad_norm": 0.06348346173763275,
      "learning_rate": 0.0005250608372455288,
      "loss": 1.4894,
      "step": 11402
    },
    {
      "epoch": 0.5070472633497843,
      "grad_norm": 0.06484797596931458,
      "learning_rate": 0.0005249148313348776,
      "loss": 1.4929,
      "step": 11404
    },
    {
      "epoch": 0.5071361878084567,
      "grad_norm": 0.06406370550394058,
      "learning_rate": 0.0005247688232943969,
      "loss": 1.4931,
      "step": 11406
    },
    {
      "epoch": 0.5072251122671291,
      "grad_norm": 0.0638013482093811,
      "learning_rate": 0.0005246228131365678,
      "loss": 1.4977,
      "step": 11408
    },
    {
      "epoch": 0.5073140367258014,
      "grad_norm": 0.06718415021896362,
      "learning_rate": 0.0005244768008738724,
      "loss": 1.498,
      "step": 11410
    },
    {
      "epoch": 0.5074029611844738,
      "grad_norm": 0.06131194531917572,
      "learning_rate": 0.0005243307865187921,
      "loss": 1.4967,
      "step": 11412
    },
    {
      "epoch": 0.5074918856431462,
      "grad_norm": 0.06349305063486099,
      "learning_rate": 0.0005241847700838089,
      "loss": 1.4966,
      "step": 11414
    },
    {
      "epoch": 0.5075808101018185,
      "grad_norm": 0.06464185565710068,
      "learning_rate": 0.0005240387515814049,
      "loss": 1.4924,
      "step": 11416
    },
    {
      "epoch": 0.5076697345604909,
      "grad_norm": 0.06342726945877075,
      "learning_rate": 0.0005238927310240627,
      "loss": 1.5043,
      "step": 11418
    },
    {
      "epoch": 0.5077586590191632,
      "grad_norm": 0.06681360304355621,
      "learning_rate": 0.0005237467084242644,
      "loss": 1.5018,
      "step": 11420
    },
    {
      "epoch": 0.5078475834778355,
      "grad_norm": 0.06572667509317398,
      "learning_rate": 0.0005236006837944928,
      "loss": 1.5001,
      "step": 11422
    },
    {
      "epoch": 0.5079365079365079,
      "grad_norm": 0.06486863642930984,
      "learning_rate": 0.0005234546571472307,
      "loss": 1.4965,
      "step": 11424
    },
    {
      "epoch": 0.5080254323951803,
      "grad_norm": 0.06413675844669342,
      "learning_rate": 0.0005233086284949612,
      "loss": 1.4973,
      "step": 11426
    },
    {
      "epoch": 0.5081143568538526,
      "grad_norm": 0.0645928606390953,
      "learning_rate": 0.0005231625978501676,
      "loss": 1.4959,
      "step": 11428
    },
    {
      "epoch": 0.508203281312525,
      "grad_norm": 0.06422816962003708,
      "learning_rate": 0.0005230165652253329,
      "loss": 1.497,
      "step": 11430
    },
    {
      "epoch": 0.5082922057711974,
      "grad_norm": 0.064136803150177,
      "learning_rate": 0.0005228705306329407,
      "loss": 1.4967,
      "step": 11432
    },
    {
      "epoch": 0.5083811302298697,
      "grad_norm": 0.06331904232501984,
      "learning_rate": 0.0005227244940854748,
      "loss": 1.4981,
      "step": 11434
    },
    {
      "epoch": 0.5084700546885421,
      "grad_norm": 0.06546802818775177,
      "learning_rate": 0.0005225784555954191,
      "loss": 1.4998,
      "step": 11436
    },
    {
      "epoch": 0.5085589791472145,
      "grad_norm": 0.06174863874912262,
      "learning_rate": 0.0005224324151752575,
      "loss": 1.4924,
      "step": 11438
    },
    {
      "epoch": 0.5086479036058869,
      "grad_norm": 0.06472094357013702,
      "learning_rate": 0.0005222863728374742,
      "loss": 1.4909,
      "step": 11440
    },
    {
      "epoch": 0.5087368280645591,
      "grad_norm": 0.0627531185746193,
      "learning_rate": 0.0005221403285945534,
      "loss": 1.4991,
      "step": 11442
    },
    {
      "epoch": 0.5088257525232315,
      "grad_norm": 0.06593529880046844,
      "learning_rate": 0.0005219942824589798,
      "loss": 1.4974,
      "step": 11444
    },
    {
      "epoch": 0.5089146769819038,
      "grad_norm": 0.06247802823781967,
      "learning_rate": 0.0005218482344432381,
      "loss": 1.4976,
      "step": 11446
    },
    {
      "epoch": 0.5090036014405762,
      "grad_norm": 0.06401393562555313,
      "learning_rate": 0.000521702184559813,
      "loss": 1.4989,
      "step": 11448
    },
    {
      "epoch": 0.5090925258992486,
      "grad_norm": 0.06516777724027634,
      "learning_rate": 0.0005215561328211895,
      "loss": 1.4973,
      "step": 11450
    },
    {
      "epoch": 0.509181450357921,
      "grad_norm": 0.06341985613107681,
      "learning_rate": 0.0005214100792398529,
      "loss": 1.4956,
      "step": 11452
    },
    {
      "epoch": 0.5092703748165933,
      "grad_norm": 0.06545495986938477,
      "learning_rate": 0.0005212640238282882,
      "loss": 1.4988,
      "step": 11454
    },
    {
      "epoch": 0.5093592992752657,
      "grad_norm": 0.06209327280521393,
      "learning_rate": 0.0005211179665989811,
      "loss": 1.4951,
      "step": 11456
    },
    {
      "epoch": 0.509448223733938,
      "grad_norm": 0.06250801682472229,
      "learning_rate": 0.000520971907564417,
      "loss": 1.4963,
      "step": 11458
    },
    {
      "epoch": 0.5095371481926104,
      "grad_norm": 0.060768019407987595,
      "learning_rate": 0.0005208258467370821,
      "loss": 1.4939,
      "step": 11460
    },
    {
      "epoch": 0.5096260726512827,
      "grad_norm": 0.06457893550395966,
      "learning_rate": 0.0005206797841294618,
      "loss": 1.5005,
      "step": 11462
    },
    {
      "epoch": 0.509714997109955,
      "grad_norm": 0.06338069587945938,
      "learning_rate": 0.0005205337197540425,
      "loss": 1.4988,
      "step": 11464
    },
    {
      "epoch": 0.5098039215686274,
      "grad_norm": 0.0637018084526062,
      "learning_rate": 0.0005203876536233104,
      "loss": 1.4965,
      "step": 11466
    },
    {
      "epoch": 0.5098928460272998,
      "grad_norm": 0.06639863550662994,
      "learning_rate": 0.0005202415857497519,
      "loss": 1.4953,
      "step": 11468
    },
    {
      "epoch": 0.5099817704859722,
      "grad_norm": 0.0653444156050682,
      "learning_rate": 0.0005200955161458533,
      "loss": 1.4941,
      "step": 11470
    },
    {
      "epoch": 0.5100706949446445,
      "grad_norm": 0.06342481076717377,
      "learning_rate": 0.0005199494448241014,
      "loss": 1.4917,
      "step": 11472
    },
    {
      "epoch": 0.5101596194033169,
      "grad_norm": 0.0635971799492836,
      "learning_rate": 0.0005198033717969833,
      "loss": 1.495,
      "step": 11474
    },
    {
      "epoch": 0.5102485438619893,
      "grad_norm": 0.06396380811929703,
      "learning_rate": 0.0005196572970769854,
      "loss": 1.5032,
      "step": 11476
    },
    {
      "epoch": 0.5103374683206616,
      "grad_norm": 0.0634695515036583,
      "learning_rate": 0.0005195112206765952,
      "loss": 1.4967,
      "step": 11478
    },
    {
      "epoch": 0.510426392779334,
      "grad_norm": 0.06354030221700668,
      "learning_rate": 0.0005193651426082999,
      "loss": 1.4968,
      "step": 11480
    },
    {
      "epoch": 0.5105153172380064,
      "grad_norm": 0.06678715348243713,
      "learning_rate": 0.000519219062884587,
      "loss": 1.4991,
      "step": 11482
    },
    {
      "epoch": 0.5106042416966786,
      "grad_norm": 0.06397784501314163,
      "learning_rate": 0.0005190729815179437,
      "loss": 1.4931,
      "step": 11484
    },
    {
      "epoch": 0.510693166155351,
      "grad_norm": 0.0642557144165039,
      "learning_rate": 0.0005189268985208579,
      "loss": 1.4963,
      "step": 11486
    },
    {
      "epoch": 0.5107820906140234,
      "grad_norm": 0.06316371262073517,
      "learning_rate": 0.0005187808139058175,
      "loss": 1.4913,
      "step": 11488
    },
    {
      "epoch": 0.5108710150726957,
      "grad_norm": 0.06363008916378021,
      "learning_rate": 0.0005186347276853105,
      "loss": 1.495,
      "step": 11490
    },
    {
      "epoch": 0.5109599395313681,
      "grad_norm": 0.06320468336343765,
      "learning_rate": 0.0005184886398718246,
      "loss": 1.495,
      "step": 11492
    },
    {
      "epoch": 0.5110488639900405,
      "grad_norm": 0.06314976513385773,
      "learning_rate": 0.0005183425504778484,
      "loss": 1.4978,
      "step": 11494
    },
    {
      "epoch": 0.5111377884487128,
      "grad_norm": 0.06445315480232239,
      "learning_rate": 0.00051819645951587,
      "loss": 1.4968,
      "step": 11496
    },
    {
      "epoch": 0.5112267129073852,
      "grad_norm": 0.06370304524898529,
      "learning_rate": 0.0005180503669983782,
      "loss": 1.4937,
      "step": 11498
    },
    {
      "epoch": 0.5113156373660576,
      "grad_norm": 0.06207090616226196,
      "learning_rate": 0.0005179042729378616,
      "loss": 1.4991,
      "step": 11500
    },
    {
      "epoch": 0.5113156373660576,
      "eval_loss": 1.4778088331222534,
      "eval_runtime": 12.4152,
      "eval_samples_per_second": 556.574,
      "eval_steps_per_second": 69.592,
      "step": 11500
    },
    {
      "epoch": 0.5114045618247299,
      "grad_norm": 0.06190875545144081,
      "learning_rate": 0.0005177581773468087,
      "loss": 1.4958,
      "step": 11502
    },
    {
      "epoch": 0.5114934862834023,
      "grad_norm": 0.06312815099954605,
      "learning_rate": 0.0005176120802377086,
      "loss": 1.494,
      "step": 11504
    },
    {
      "epoch": 0.5115824107420746,
      "grad_norm": 0.06356559693813324,
      "learning_rate": 0.0005174659816230503,
      "loss": 1.5014,
      "step": 11506
    },
    {
      "epoch": 0.5116713352007469,
      "grad_norm": 0.060818467289209366,
      "learning_rate": 0.0005173198815153228,
      "loss": 1.4924,
      "step": 11508
    },
    {
      "epoch": 0.5117602596594193,
      "grad_norm": 0.06228560581803322,
      "learning_rate": 0.0005171737799270154,
      "loss": 1.4924,
      "step": 11510
    },
    {
      "epoch": 0.5118491841180917,
      "grad_norm": 0.06337162107229233,
      "learning_rate": 0.0005170276768706179,
      "loss": 1.5018,
      "step": 11512
    },
    {
      "epoch": 0.511938108576764,
      "grad_norm": 0.06353859603404999,
      "learning_rate": 0.0005168815723586194,
      "loss": 1.5007,
      "step": 11514
    },
    {
      "epoch": 0.5120270330354364,
      "grad_norm": 0.06312459707260132,
      "learning_rate": 0.0005167354664035096,
      "loss": 1.4998,
      "step": 11516
    },
    {
      "epoch": 0.5121159574941088,
      "grad_norm": 0.06268506497144699,
      "learning_rate": 0.0005165893590177786,
      "loss": 1.5017,
      "step": 11518
    },
    {
      "epoch": 0.5122048819527811,
      "grad_norm": 0.07257755100727081,
      "learning_rate": 0.0005164432502139158,
      "loss": 1.4964,
      "step": 11520
    },
    {
      "epoch": 0.5122938064114535,
      "grad_norm": 0.06548047810792923,
      "learning_rate": 0.0005162971400044119,
      "loss": 1.5002,
      "step": 11522
    },
    {
      "epoch": 0.5123827308701259,
      "grad_norm": 0.06318143010139465,
      "learning_rate": 0.0005161510284017563,
      "loss": 1.493,
      "step": 11524
    },
    {
      "epoch": 0.5124716553287982,
      "grad_norm": 0.06401795148849487,
      "learning_rate": 0.0005160049154184399,
      "loss": 1.4945,
      "step": 11526
    },
    {
      "epoch": 0.5125605797874705,
      "grad_norm": 0.06641431897878647,
      "learning_rate": 0.0005158588010669526,
      "loss": 1.495,
      "step": 11528
    },
    {
      "epoch": 0.5126495042461429,
      "grad_norm": 0.06431353837251663,
      "learning_rate": 0.0005157126853597852,
      "loss": 1.4966,
      "step": 11530
    },
    {
      "epoch": 0.5127384287048152,
      "grad_norm": 0.06240619346499443,
      "learning_rate": 0.000515566568309428,
      "loss": 1.4989,
      "step": 11532
    },
    {
      "epoch": 0.5128273531634876,
      "grad_norm": 0.06395246088504791,
      "learning_rate": 0.0005154204499283723,
      "loss": 1.4936,
      "step": 11534
    },
    {
      "epoch": 0.51291627762216,
      "grad_norm": 0.06343571096658707,
      "learning_rate": 0.0005152743302291083,
      "loss": 1.4887,
      "step": 11536
    },
    {
      "epoch": 0.5130052020808323,
      "grad_norm": 0.06305166333913803,
      "learning_rate": 0.0005151282092241273,
      "loss": 1.4997,
      "step": 11538
    },
    {
      "epoch": 0.5130941265395047,
      "grad_norm": 0.06751389056444168,
      "learning_rate": 0.0005149820869259203,
      "loss": 1.4989,
      "step": 11540
    },
    {
      "epoch": 0.5131830509981771,
      "grad_norm": 0.06132427230477333,
      "learning_rate": 0.0005148359633469784,
      "loss": 1.4996,
      "step": 11542
    },
    {
      "epoch": 0.5132719754568494,
      "grad_norm": 0.06433816254138947,
      "learning_rate": 0.0005146898384997933,
      "loss": 1.4972,
      "step": 11544
    },
    {
      "epoch": 0.5133608999155218,
      "grad_norm": 0.06302385777235031,
      "learning_rate": 0.0005145437123968558,
      "loss": 1.4971,
      "step": 11546
    },
    {
      "epoch": 0.5134498243741942,
      "grad_norm": 0.06478093564510345,
      "learning_rate": 0.0005143975850506579,
      "loss": 1.4962,
      "step": 11548
    },
    {
      "epoch": 0.5135387488328664,
      "grad_norm": 0.06177389994263649,
      "learning_rate": 0.0005142514564736906,
      "loss": 1.4966,
      "step": 11550
    },
    {
      "epoch": 0.5136276732915388,
      "grad_norm": 0.06433558464050293,
      "learning_rate": 0.0005141053266784462,
      "loss": 1.4989,
      "step": 11552
    },
    {
      "epoch": 0.5137165977502112,
      "grad_norm": 0.0624937042593956,
      "learning_rate": 0.0005139591956774165,
      "loss": 1.4979,
      "step": 11554
    },
    {
      "epoch": 0.5138055222088835,
      "grad_norm": 0.062215663492679596,
      "learning_rate": 0.0005138130634830928,
      "loss": 1.4982,
      "step": 11556
    },
    {
      "epoch": 0.5138944466675559,
      "grad_norm": 0.06470473855733871,
      "learning_rate": 0.0005136669301079679,
      "loss": 1.4914,
      "step": 11558
    },
    {
      "epoch": 0.5139833711262283,
      "grad_norm": 0.06185575947165489,
      "learning_rate": 0.0005135207955645336,
      "loss": 1.4924,
      "step": 11560
    },
    {
      "epoch": 0.5140722955849006,
      "grad_norm": 0.06259468197822571,
      "learning_rate": 0.0005133746598652821,
      "loss": 1.4997,
      "step": 11562
    },
    {
      "epoch": 0.514161220043573,
      "grad_norm": 0.06266958266496658,
      "learning_rate": 0.0005132285230227055,
      "loss": 1.496,
      "step": 11564
    },
    {
      "epoch": 0.5142501445022454,
      "grad_norm": 0.06312844902276993,
      "learning_rate": 0.0005130823850492967,
      "loss": 1.4945,
      "step": 11566
    },
    {
      "epoch": 0.5143390689609177,
      "grad_norm": 0.06307828426361084,
      "learning_rate": 0.0005129362459575478,
      "loss": 1.4946,
      "step": 11568
    },
    {
      "epoch": 0.5144279934195901,
      "grad_norm": 0.06209829822182655,
      "learning_rate": 0.0005127901057599516,
      "loss": 1.4996,
      "step": 11570
    },
    {
      "epoch": 0.5145169178782624,
      "grad_norm": 0.06419774144887924,
      "learning_rate": 0.000512643964469001,
      "loss": 1.4965,
      "step": 11572
    },
    {
      "epoch": 0.5146058423369347,
      "grad_norm": 0.06341389566659927,
      "learning_rate": 0.0005124978220971884,
      "loss": 1.4957,
      "step": 11574
    },
    {
      "epoch": 0.5146947667956071,
      "grad_norm": 0.062224533408880234,
      "learning_rate": 0.0005123516786570071,
      "loss": 1.4992,
      "step": 11576
    },
    {
      "epoch": 0.5147836912542795,
      "grad_norm": 0.06227099150419235,
      "learning_rate": 0.0005122055341609496,
      "loss": 1.5002,
      "step": 11578
    },
    {
      "epoch": 0.5148726157129518,
      "grad_norm": 0.06264078617095947,
      "learning_rate": 0.0005120593886215095,
      "loss": 1.4982,
      "step": 11580
    },
    {
      "epoch": 0.5149615401716242,
      "grad_norm": 0.0614069402217865,
      "learning_rate": 0.0005119132420511795,
      "loss": 1.4952,
      "step": 11582
    },
    {
      "epoch": 0.5150504646302966,
      "grad_norm": 0.06539449840784073,
      "learning_rate": 0.0005117670944624533,
      "loss": 1.4953,
      "step": 11584
    },
    {
      "epoch": 0.515139389088969,
      "grad_norm": 0.06382063776254654,
      "learning_rate": 0.0005116209458678238,
      "loss": 1.4929,
      "step": 11586
    },
    {
      "epoch": 0.5152283135476413,
      "grad_norm": 0.06359395384788513,
      "learning_rate": 0.0005114747962797848,
      "loss": 1.4938,
      "step": 11588
    },
    {
      "epoch": 0.5153172380063137,
      "grad_norm": 0.06395386159420013,
      "learning_rate": 0.0005113286457108296,
      "loss": 1.4926,
      "step": 11590
    },
    {
      "epoch": 0.5154061624649859,
      "grad_norm": 0.06651069223880768,
      "learning_rate": 0.0005111824941734516,
      "loss": 1.5002,
      "step": 11592
    },
    {
      "epoch": 0.5154950869236583,
      "grad_norm": 0.06515168398618698,
      "learning_rate": 0.000511036341680145,
      "loss": 1.4955,
      "step": 11594
    },
    {
      "epoch": 0.5155840113823307,
      "grad_norm": 0.06810913234949112,
      "learning_rate": 0.0005108901882434031,
      "loss": 1.4953,
      "step": 11596
    },
    {
      "epoch": 0.515672935841003,
      "grad_norm": 0.06428667902946472,
      "learning_rate": 0.0005107440338757202,
      "loss": 1.4911,
      "step": 11598
    },
    {
      "epoch": 0.5157618602996754,
      "grad_norm": 0.06621834635734558,
      "learning_rate": 0.0005105978785895898,
      "loss": 1.5015,
      "step": 11600
    },
    {
      "epoch": 0.5158507847583478,
      "grad_norm": 0.06360407173633575,
      "learning_rate": 0.000510451722397506,
      "loss": 1.4948,
      "step": 11602
    },
    {
      "epoch": 0.5159397092170201,
      "grad_norm": 0.06574375927448273,
      "learning_rate": 0.0005103055653119627,
      "loss": 1.4965,
      "step": 11604
    },
    {
      "epoch": 0.5160286336756925,
      "grad_norm": 0.06324627995491028,
      "learning_rate": 0.0005101594073454545,
      "loss": 1.4963,
      "step": 11606
    },
    {
      "epoch": 0.5161175581343649,
      "grad_norm": 0.06388534605503082,
      "learning_rate": 0.0005100132485104754,
      "loss": 1.4984,
      "step": 11608
    },
    {
      "epoch": 0.5162064825930373,
      "grad_norm": 0.06630777567625046,
      "learning_rate": 0.0005098670888195196,
      "loss": 1.4928,
      "step": 11610
    },
    {
      "epoch": 0.5162954070517096,
      "grad_norm": 0.0635385662317276,
      "learning_rate": 0.0005097209282850817,
      "loss": 1.4886,
      "step": 11612
    },
    {
      "epoch": 0.5163843315103819,
      "grad_norm": 0.062710702419281,
      "learning_rate": 0.0005095747669196559,
      "loss": 1.4977,
      "step": 11614
    },
    {
      "epoch": 0.5164732559690542,
      "grad_norm": 0.06373061239719391,
      "learning_rate": 0.0005094286047357368,
      "loss": 1.4934,
      "step": 11616
    },
    {
      "epoch": 0.5165621804277266,
      "grad_norm": 0.06358326971530914,
      "learning_rate": 0.0005092824417458191,
      "loss": 1.496,
      "step": 11618
    },
    {
      "epoch": 0.516651104886399,
      "grad_norm": 0.06176624819636345,
      "learning_rate": 0.0005091362779623972,
      "loss": 1.4983,
      "step": 11620
    },
    {
      "epoch": 0.5167400293450713,
      "grad_norm": 0.06667372584342957,
      "learning_rate": 0.0005089901133979663,
      "loss": 1.5,
      "step": 11622
    },
    {
      "epoch": 0.5168289538037437,
      "grad_norm": 0.0629948228597641,
      "learning_rate": 0.0005088439480650207,
      "loss": 1.5001,
      "step": 11624
    },
    {
      "epoch": 0.5169178782624161,
      "grad_norm": 0.06550941616296768,
      "learning_rate": 0.0005086977819760554,
      "loss": 1.4975,
      "step": 11626
    },
    {
      "epoch": 0.5170068027210885,
      "grad_norm": 0.06235315650701523,
      "learning_rate": 0.0005085516151435655,
      "loss": 1.4933,
      "step": 11628
    },
    {
      "epoch": 0.5170957271797608,
      "grad_norm": 0.06560411304235458,
      "learning_rate": 0.0005084054475800458,
      "loss": 1.4941,
      "step": 11630
    },
    {
      "epoch": 0.5171846516384332,
      "grad_norm": 0.06175919994711876,
      "learning_rate": 0.0005082592792979914,
      "loss": 1.494,
      "step": 11632
    },
    {
      "epoch": 0.5172735760971056,
      "grad_norm": 0.06700320541858673,
      "learning_rate": 0.0005081131103098974,
      "loss": 1.4909,
      "step": 11634
    },
    {
      "epoch": 0.5173625005557778,
      "grad_norm": 0.06476999819278717,
      "learning_rate": 0.0005079669406282588,
      "loss": 1.4967,
      "step": 11636
    },
    {
      "epoch": 0.5174514250144502,
      "grad_norm": 0.06396272778511047,
      "learning_rate": 0.0005078207702655714,
      "loss": 1.5023,
      "step": 11638
    },
    {
      "epoch": 0.5175403494731226,
      "grad_norm": 0.0631399154663086,
      "learning_rate": 0.0005076745992343296,
      "loss": 1.4899,
      "step": 11640
    },
    {
      "epoch": 0.5176292739317949,
      "grad_norm": 0.06382925063371658,
      "learning_rate": 0.0005075284275470296,
      "loss": 1.4977,
      "step": 11642
    },
    {
      "epoch": 0.5177181983904673,
      "grad_norm": 0.06469811499118805,
      "learning_rate": 0.0005073822552161664,
      "loss": 1.5019,
      "step": 11644
    },
    {
      "epoch": 0.5178071228491397,
      "grad_norm": 0.06676246225833893,
      "learning_rate": 0.0005072360822542354,
      "loss": 1.4926,
      "step": 11646
    },
    {
      "epoch": 0.517896047307812,
      "grad_norm": 0.0667065978050232,
      "learning_rate": 0.0005070899086737322,
      "loss": 1.4994,
      "step": 11648
    },
    {
      "epoch": 0.5179849717664844,
      "grad_norm": 0.06245999410748482,
      "learning_rate": 0.0005069437344871523,
      "loss": 1.4896,
      "step": 11650
    },
    {
      "epoch": 0.5180738962251568,
      "grad_norm": 0.06560548394918442,
      "learning_rate": 0.0005067975597069917,
      "loss": 1.497,
      "step": 11652
    },
    {
      "epoch": 0.5181628206838291,
      "grad_norm": 0.06353573501110077,
      "learning_rate": 0.0005066513843457454,
      "loss": 1.4976,
      "step": 11654
    },
    {
      "epoch": 0.5182517451425015,
      "grad_norm": 0.06430783122777939,
      "learning_rate": 0.0005065052084159096,
      "loss": 1.4974,
      "step": 11656
    },
    {
      "epoch": 0.5183406696011738,
      "grad_norm": 0.06370967626571655,
      "learning_rate": 0.0005063590319299799,
      "loss": 1.4924,
      "step": 11658
    },
    {
      "epoch": 0.5184295940598461,
      "grad_norm": 0.06251141428947449,
      "learning_rate": 0.0005062128549004521,
      "loss": 1.4985,
      "step": 11660
    },
    {
      "epoch": 0.5185185185185185,
      "grad_norm": 0.06270091235637665,
      "learning_rate": 0.0005060666773398221,
      "loss": 1.4939,
      "step": 11662
    },
    {
      "epoch": 0.5186074429771909,
      "grad_norm": 0.06116213649511337,
      "learning_rate": 0.0005059204992605857,
      "loss": 1.499,
      "step": 11664
    },
    {
      "epoch": 0.5186963674358632,
      "grad_norm": 0.06304262578487396,
      "learning_rate": 0.0005057743206752391,
      "loss": 1.4885,
      "step": 11666
    },
    {
      "epoch": 0.5187852918945356,
      "grad_norm": 0.062481291592121124,
      "learning_rate": 0.000505628141596278,
      "loss": 1.4923,
      "step": 11668
    },
    {
      "epoch": 0.518874216353208,
      "grad_norm": 0.06374236196279526,
      "learning_rate": 0.0005054819620361986,
      "loss": 1.5012,
      "step": 11670
    },
    {
      "epoch": 0.5189631408118803,
      "grad_norm": 0.06343623250722885,
      "learning_rate": 0.0005053357820074968,
      "loss": 1.4938,
      "step": 11672
    },
    {
      "epoch": 0.5190520652705527,
      "grad_norm": 0.06267695873975754,
      "learning_rate": 0.0005051896015226691,
      "loss": 1.4951,
      "step": 11674
    },
    {
      "epoch": 0.5191409897292251,
      "grad_norm": 0.06275571882724762,
      "learning_rate": 0.0005050434205942113,
      "loss": 1.4897,
      "step": 11676
    },
    {
      "epoch": 0.5192299141878974,
      "grad_norm": 0.06288079917430878,
      "learning_rate": 0.0005048972392346196,
      "loss": 1.4895,
      "step": 11678
    },
    {
      "epoch": 0.5193188386465697,
      "grad_norm": 0.06326977908611298,
      "learning_rate": 0.0005047510574563902,
      "loss": 1.4993,
      "step": 11680
    },
    {
      "epoch": 0.5194077631052421,
      "grad_norm": 0.06266149133443832,
      "learning_rate": 0.0005046048752720197,
      "loss": 1.4982,
      "step": 11682
    },
    {
      "epoch": 0.5194966875639144,
      "grad_norm": 0.06486205756664276,
      "learning_rate": 0.0005044586926940041,
      "loss": 1.496,
      "step": 11684
    },
    {
      "epoch": 0.5195856120225868,
      "grad_norm": 0.06361040472984314,
      "learning_rate": 0.0005043125097348398,
      "loss": 1.4952,
      "step": 11686
    },
    {
      "epoch": 0.5196745364812592,
      "grad_norm": 0.0649038627743721,
      "learning_rate": 0.0005041663264070231,
      "loss": 1.4968,
      "step": 11688
    },
    {
      "epoch": 0.5197634609399315,
      "grad_norm": 0.06310614198446274,
      "learning_rate": 0.0005040201427230504,
      "loss": 1.4945,
      "step": 11690
    },
    {
      "epoch": 0.5198523853986039,
      "grad_norm": 0.0667177364230156,
      "learning_rate": 0.0005038739586954186,
      "loss": 1.4936,
      "step": 11692
    },
    {
      "epoch": 0.5199413098572763,
      "grad_norm": 0.0625763013958931,
      "learning_rate": 0.0005037277743366232,
      "loss": 1.4931,
      "step": 11694
    },
    {
      "epoch": 0.5200302343159486,
      "grad_norm": 0.06521979719400406,
      "learning_rate": 0.0005035815896591616,
      "loss": 1.4938,
      "step": 11696
    },
    {
      "epoch": 0.520119158774621,
      "grad_norm": 0.06360069662332535,
      "learning_rate": 0.0005034354046755296,
      "loss": 1.494,
      "step": 11698
    },
    {
      "epoch": 0.5202080832332934,
      "grad_norm": 0.06382516026496887,
      "learning_rate": 0.0005032892193982243,
      "loss": 1.4954,
      "step": 11700
    },
    {
      "epoch": 0.5202970076919656,
      "grad_norm": 0.06371236592531204,
      "learning_rate": 0.000503143033839742,
      "loss": 1.4961,
      "step": 11702
    },
    {
      "epoch": 0.520385932150638,
      "grad_norm": 0.06548450887203217,
      "learning_rate": 0.0005029968480125792,
      "loss": 1.4937,
      "step": 11704
    },
    {
      "epoch": 0.5204748566093104,
      "grad_norm": 0.06387647241353989,
      "learning_rate": 0.0005028506619292328,
      "loss": 1.4924,
      "step": 11706
    },
    {
      "epoch": 0.5205637810679827,
      "grad_norm": 0.06516347825527191,
      "learning_rate": 0.0005027044756021992,
      "loss": 1.4932,
      "step": 11708
    },
    {
      "epoch": 0.5206527055266551,
      "grad_norm": 0.06408815830945969,
      "learning_rate": 0.0005025582890439751,
      "loss": 1.4929,
      "step": 11710
    },
    {
      "epoch": 0.5207416299853275,
      "grad_norm": 0.06484084576368332,
      "learning_rate": 0.0005024121022670572,
      "loss": 1.4895,
      "step": 11712
    },
    {
      "epoch": 0.5208305544439998,
      "grad_norm": 0.062469739466905594,
      "learning_rate": 0.0005022659152839421,
      "loss": 1.4925,
      "step": 11714
    },
    {
      "epoch": 0.5209194789026722,
      "grad_norm": 0.06194042041897774,
      "learning_rate": 0.0005021197281071265,
      "loss": 1.502,
      "step": 11716
    },
    {
      "epoch": 0.5210084033613446,
      "grad_norm": 0.06476829200983047,
      "learning_rate": 0.0005019735407491072,
      "loss": 1.5043,
      "step": 11718
    },
    {
      "epoch": 0.5210973278200169,
      "grad_norm": 0.06216420605778694,
      "learning_rate": 0.0005018273532223812,
      "loss": 1.4906,
      "step": 11720
    },
    {
      "epoch": 0.5211862522786892,
      "grad_norm": 0.06267587095499039,
      "learning_rate": 0.0005016811655394449,
      "loss": 1.4925,
      "step": 11722
    },
    {
      "epoch": 0.5212751767373616,
      "grad_norm": 0.06346725672483444,
      "learning_rate": 0.0005015349777127952,
      "loss": 1.4922,
      "step": 11724
    },
    {
      "epoch": 0.5213641011960339,
      "grad_norm": 0.06293226033449173,
      "learning_rate": 0.0005013887897549285,
      "loss": 1.4935,
      "step": 11726
    },
    {
      "epoch": 0.5214530256547063,
      "grad_norm": 0.061574727296829224,
      "learning_rate": 0.0005012426016783422,
      "loss": 1.4944,
      "step": 11728
    },
    {
      "epoch": 0.5215419501133787,
      "grad_norm": 0.062728650867939,
      "learning_rate": 0.0005010964134955329,
      "loss": 1.4908,
      "step": 11730
    },
    {
      "epoch": 0.521630874572051,
      "grad_norm": 0.06296613812446594,
      "learning_rate": 0.0005009502252189972,
      "loss": 1.4882,
      "step": 11732
    },
    {
      "epoch": 0.5217197990307234,
      "grad_norm": 0.0656297355890274,
      "learning_rate": 0.0005008040368612321,
      "loss": 1.4928,
      "step": 11734
    },
    {
      "epoch": 0.5218087234893958,
      "grad_norm": 0.06233159825205803,
      "learning_rate": 0.0005006578484347345,
      "loss": 1.5,
      "step": 11736
    },
    {
      "epoch": 0.5218976479480681,
      "grad_norm": 0.06443444639444351,
      "learning_rate": 0.000500511659952001,
      "loss": 1.4908,
      "step": 11738
    },
    {
      "epoch": 0.5219865724067405,
      "grad_norm": 0.06185967102646828,
      "learning_rate": 0.0005003654714255285,
      "loss": 1.4877,
      "step": 11740
    },
    {
      "epoch": 0.5220754968654129,
      "grad_norm": 0.0641622245311737,
      "learning_rate": 0.0005002192828678139,
      "loss": 1.4962,
      "step": 11742
    },
    {
      "epoch": 0.5221644213240851,
      "grad_norm": 0.06316876411437988,
      "learning_rate": 0.0005000730942913541,
      "loss": 1.4998,
      "step": 11744
    },
    {
      "epoch": 0.5222533457827575,
      "grad_norm": 0.06322462856769562,
      "learning_rate": 0.0004999269057086459,
      "loss": 1.4944,
      "step": 11746
    },
    {
      "epoch": 0.5223422702414299,
      "grad_norm": 0.06409668177366257,
      "learning_rate": 0.0004997807171321861,
      "loss": 1.4944,
      "step": 11748
    },
    {
      "epoch": 0.5224311947001022,
      "grad_norm": 0.06382996588945389,
      "learning_rate": 0.0004996345285744716,
      "loss": 1.4929,
      "step": 11750
    },
    {
      "epoch": 0.5225201191587746,
      "grad_norm": 0.06639515608549118,
      "learning_rate": 0.0004994883400479991,
      "loss": 1.493,
      "step": 11752
    },
    {
      "epoch": 0.522609043617447,
      "grad_norm": 0.06325390189886093,
      "learning_rate": 0.0004993421515652656,
      "loss": 1.4928,
      "step": 11754
    },
    {
      "epoch": 0.5226979680761193,
      "grad_norm": 0.06696011871099472,
      "learning_rate": 0.0004991959631387679,
      "loss": 1.4937,
      "step": 11756
    },
    {
      "epoch": 0.5227868925347917,
      "grad_norm": 0.06227093189954758,
      "learning_rate": 0.0004990497747810028,
      "loss": 1.4913,
      "step": 11758
    },
    {
      "epoch": 0.5228758169934641,
      "grad_norm": 0.0665513128042221,
      "learning_rate": 0.0004989035865044673,
      "loss": 1.4935,
      "step": 11760
    },
    {
      "epoch": 0.5229647414521365,
      "grad_norm": 0.0690368264913559,
      "learning_rate": 0.0004987573983216576,
      "loss": 1.4907,
      "step": 11762
    },
    {
      "epoch": 0.5230536659108088,
      "grad_norm": 0.06481185555458069,
      "learning_rate": 0.0004986112102450715,
      "loss": 1.4942,
      "step": 11764
    },
    {
      "epoch": 0.5231425903694811,
      "grad_norm": 0.06394998729228973,
      "learning_rate": 0.0004984650222872051,
      "loss": 1.4943,
      "step": 11766
    },
    {
      "epoch": 0.5232315148281534,
      "grad_norm": 0.06454624235630035,
      "learning_rate": 0.0004983188344605552,
      "loss": 1.4975,
      "step": 11768
    },
    {
      "epoch": 0.5233204392868258,
      "grad_norm": 0.06496671587228775,
      "learning_rate": 0.0004981726467776188,
      "loss": 1.4941,
      "step": 11770
    },
    {
      "epoch": 0.5234093637454982,
      "grad_norm": 0.0641636922955513,
      "learning_rate": 0.0004980264592508927,
      "loss": 1.4984,
      "step": 11772
    },
    {
      "epoch": 0.5234982882041705,
      "grad_norm": 0.06285618245601654,
      "learning_rate": 0.0004978802718928735,
      "loss": 1.4953,
      "step": 11774
    },
    {
      "epoch": 0.5235872126628429,
      "grad_norm": 0.06363879889249802,
      "learning_rate": 0.0004977340847160581,
      "loss": 1.4954,
      "step": 11776
    },
    {
      "epoch": 0.5236761371215153,
      "grad_norm": 0.06370127201080322,
      "learning_rate": 0.0004975878977329429,
      "loss": 1.4919,
      "step": 11778
    },
    {
      "epoch": 0.5237650615801877,
      "grad_norm": 0.06240037456154823,
      "learning_rate": 0.000497441710956025,
      "loss": 1.4935,
      "step": 11780
    },
    {
      "epoch": 0.52385398603886,
      "grad_norm": 0.06250346451997757,
      "learning_rate": 0.000497295524397801,
      "loss": 1.4893,
      "step": 11782
    },
    {
      "epoch": 0.5239429104975324,
      "grad_norm": 0.06271068751811981,
      "learning_rate": 0.0004971493380707671,
      "loss": 1.4963,
      "step": 11784
    },
    {
      "epoch": 0.5240318349562048,
      "grad_norm": 0.06360255181789398,
      "learning_rate": 0.0004970031519874209,
      "loss": 1.4935,
      "step": 11786
    },
    {
      "epoch": 0.524120759414877,
      "grad_norm": 0.06317956000566483,
      "learning_rate": 0.0004968569661602582,
      "loss": 1.4939,
      "step": 11788
    },
    {
      "epoch": 0.5242096838735494,
      "grad_norm": 0.06279599666595459,
      "learning_rate": 0.0004967107806017757,
      "loss": 1.4987,
      "step": 11790
    },
    {
      "epoch": 0.5242986083322217,
      "grad_norm": 0.0632399320602417,
      "learning_rate": 0.0004965645953244705,
      "loss": 1.4954,
      "step": 11792
    },
    {
      "epoch": 0.5243875327908941,
      "grad_norm": 0.06396782398223877,
      "learning_rate": 0.0004964184103408386,
      "loss": 1.4951,
      "step": 11794
    },
    {
      "epoch": 0.5244764572495665,
      "grad_norm": 0.061761051416397095,
      "learning_rate": 0.0004962722256633768,
      "loss": 1.502,
      "step": 11796
    },
    {
      "epoch": 0.5245653817082389,
      "grad_norm": 0.06565398722887039,
      "learning_rate": 0.0004961260413045818,
      "loss": 1.4963,
      "step": 11798
    },
    {
      "epoch": 0.5246543061669112,
      "grad_norm": 0.06184777244925499,
      "learning_rate": 0.0004959798572769495,
      "loss": 1.493,
      "step": 11800
    },
    {
      "epoch": 0.5247432306255836,
      "grad_norm": 0.06266294419765472,
      "learning_rate": 0.000495833673592977,
      "loss": 1.4999,
      "step": 11802
    },
    {
      "epoch": 0.524832155084256,
      "grad_norm": 0.06519590318202972,
      "learning_rate": 0.0004956874902651603,
      "loss": 1.4917,
      "step": 11804
    },
    {
      "epoch": 0.5249210795429283,
      "grad_norm": 0.06418920308351517,
      "learning_rate": 0.0004955413073059959,
      "loss": 1.4897,
      "step": 11806
    },
    {
      "epoch": 0.5250100040016007,
      "grad_norm": 0.06677830964326859,
      "learning_rate": 0.0004953951247279804,
      "loss": 1.4942,
      "step": 11808
    },
    {
      "epoch": 0.525098928460273,
      "grad_norm": 0.06483811140060425,
      "learning_rate": 0.0004952489425436097,
      "loss": 1.4902,
      "step": 11810
    },
    {
      "epoch": 0.5251878529189453,
      "grad_norm": 0.06273074448108673,
      "learning_rate": 0.0004951027607653805,
      "loss": 1.4937,
      "step": 11812
    },
    {
      "epoch": 0.5252767773776177,
      "grad_norm": 0.06601008027791977,
      "learning_rate": 0.0004949565794057889,
      "loss": 1.4903,
      "step": 11814
    },
    {
      "epoch": 0.52536570183629,
      "grad_norm": 0.06314726918935776,
      "learning_rate": 0.0004948103984773309,
      "loss": 1.4964,
      "step": 11816
    },
    {
      "epoch": 0.5254546262949624,
      "grad_norm": 0.06756903976202011,
      "learning_rate": 0.0004946642179925032,
      "loss": 1.4895,
      "step": 11818
    },
    {
      "epoch": 0.5255435507536348,
      "grad_norm": 0.06394748389720917,
      "learning_rate": 0.0004945180379638015,
      "loss": 1.4906,
      "step": 11820
    },
    {
      "epoch": 0.5256324752123072,
      "grad_norm": 0.06343099474906921,
      "learning_rate": 0.0004943718584037221,
      "loss": 1.4973,
      "step": 11822
    },
    {
      "epoch": 0.5257213996709795,
      "grad_norm": 0.06174701824784279,
      "learning_rate": 0.000494225679324761,
      "loss": 1.4924,
      "step": 11824
    },
    {
      "epoch": 0.5258103241296519,
      "grad_norm": 0.06279890984296799,
      "learning_rate": 0.0004940795007394143,
      "loss": 1.4946,
      "step": 11826
    },
    {
      "epoch": 0.5258992485883243,
      "grad_norm": 0.06350038945674896,
      "learning_rate": 0.000493933322660178,
      "loss": 1.4906,
      "step": 11828
    },
    {
      "epoch": 0.5259881730469965,
      "grad_norm": 0.06374242156744003,
      "learning_rate": 0.000493787145099548,
      "loss": 1.497,
      "step": 11830
    },
    {
      "epoch": 0.5260770975056689,
      "grad_norm": 0.062003325670957565,
      "learning_rate": 0.0004936409680700201,
      "loss": 1.4964,
      "step": 11832
    },
    {
      "epoch": 0.5261660219643413,
      "grad_norm": 0.06445759534835815,
      "learning_rate": 0.0004934947915840904,
      "loss": 1.4866,
      "step": 11834
    },
    {
      "epoch": 0.5262549464230136,
      "grad_norm": 0.06202596426010132,
      "learning_rate": 0.0004933486156542548,
      "loss": 1.4899,
      "step": 11836
    },
    {
      "epoch": 0.526343870881686,
      "grad_norm": 0.06291262060403824,
      "learning_rate": 0.0004932024402930084,
      "loss": 1.4951,
      "step": 11838
    },
    {
      "epoch": 0.5264327953403584,
      "grad_norm": 0.06197972595691681,
      "learning_rate": 0.0004930562655128477,
      "loss": 1.4858,
      "step": 11840
    },
    {
      "epoch": 0.5265217197990307,
      "grad_norm": 0.06182212010025978,
      "learning_rate": 0.0004929100913262679,
      "loss": 1.501,
      "step": 11842
    },
    {
      "epoch": 0.5266106442577031,
      "grad_norm": 0.06719043105840683,
      "learning_rate": 0.0004927639177457647,
      "loss": 1.5003,
      "step": 11844
    },
    {
      "epoch": 0.5266995687163755,
      "grad_norm": 0.06187283247709274,
      "learning_rate": 0.0004926177447838338,
      "loss": 1.4959,
      "step": 11846
    },
    {
      "epoch": 0.5267884931750478,
      "grad_norm": 0.06271807104349136,
      "learning_rate": 0.0004924715724529704,
      "loss": 1.495,
      "step": 11848
    },
    {
      "epoch": 0.5268774176337202,
      "grad_norm": 0.06492515653371811,
      "learning_rate": 0.0004923254007656704,
      "loss": 1.4968,
      "step": 11850
    },
    {
      "epoch": 0.5269663420923925,
      "grad_norm": 0.0634203776717186,
      "learning_rate": 0.0004921792297344289,
      "loss": 1.4964,
      "step": 11852
    },
    {
      "epoch": 0.5270552665510648,
      "grad_norm": 0.0632987841963768,
      "learning_rate": 0.0004920330593717411,
      "loss": 1.482,
      "step": 11854
    },
    {
      "epoch": 0.5271441910097372,
      "grad_norm": 0.0637960284948349,
      "learning_rate": 0.0004918868896901027,
      "loss": 1.4902,
      "step": 11856
    },
    {
      "epoch": 0.5272331154684096,
      "grad_norm": 0.0617951899766922,
      "learning_rate": 0.0004917407207020088,
      "loss": 1.4918,
      "step": 11858
    },
    {
      "epoch": 0.5273220399270819,
      "grad_norm": 0.06447748094797134,
      "learning_rate": 0.0004915945524199542,
      "loss": 1.4945,
      "step": 11860
    },
    {
      "epoch": 0.5274109643857543,
      "grad_norm": 0.06240464001893997,
      "learning_rate": 0.0004914483848564346,
      "loss": 1.4893,
      "step": 11862
    },
    {
      "epoch": 0.5274998888444267,
      "grad_norm": 0.06561367958784103,
      "learning_rate": 0.0004913022180239446,
      "loss": 1.4904,
      "step": 11864
    },
    {
      "epoch": 0.527588813303099,
      "grad_norm": 0.06380762904882431,
      "learning_rate": 0.0004911560519349794,
      "loss": 1.4985,
      "step": 11866
    },
    {
      "epoch": 0.5276777377617714,
      "grad_norm": 0.06274005025625229,
      "learning_rate": 0.0004910098866020339,
      "loss": 1.4938,
      "step": 11868
    },
    {
      "epoch": 0.5277666622204438,
      "grad_norm": 0.06453198194503784,
      "learning_rate": 0.0004908637220376027,
      "loss": 1.4965,
      "step": 11870
    },
    {
      "epoch": 0.5278555866791161,
      "grad_norm": 0.06285727769136429,
      "learning_rate": 0.000490717558254181,
      "loss": 1.4924,
      "step": 11872
    },
    {
      "epoch": 0.5279445111377884,
      "grad_norm": 0.06540695577859879,
      "learning_rate": 0.0004905713952642634,
      "loss": 1.4931,
      "step": 11874
    },
    {
      "epoch": 0.5280334355964608,
      "grad_norm": 0.06500522047281265,
      "learning_rate": 0.0004904252330803442,
      "loss": 1.4884,
      "step": 11876
    },
    {
      "epoch": 0.5281223600551331,
      "grad_norm": 0.06160023435950279,
      "learning_rate": 0.0004902790717149184,
      "loss": 1.494,
      "step": 11878
    },
    {
      "epoch": 0.5282112845138055,
      "grad_norm": 0.06481318175792694,
      "learning_rate": 0.0004901329111804805,
      "loss": 1.4887,
      "step": 11880
    },
    {
      "epoch": 0.5283002089724779,
      "grad_norm": 0.06199006363749504,
      "learning_rate": 0.0004899867514895247,
      "loss": 1.4942,
      "step": 11882
    },
    {
      "epoch": 0.5283891334311502,
      "grad_norm": 0.06205325201153755,
      "learning_rate": 0.0004898405926545455,
      "loss": 1.4907,
      "step": 11884
    },
    {
      "epoch": 0.5284780578898226,
      "grad_norm": 0.06204449385404587,
      "learning_rate": 0.0004896944346880372,
      "loss": 1.4962,
      "step": 11886
    },
    {
      "epoch": 0.528566982348495,
      "grad_norm": 0.0624365471303463,
      "learning_rate": 0.0004895482776024942,
      "loss": 1.4945,
      "step": 11888
    },
    {
      "epoch": 0.5286559068071673,
      "grad_norm": 0.06314221024513245,
      "learning_rate": 0.0004894021214104105,
      "loss": 1.4899,
      "step": 11890
    },
    {
      "epoch": 0.5287448312658397,
      "grad_norm": 0.06316719204187393,
      "learning_rate": 0.0004892559661242798,
      "loss": 1.4979,
      "step": 11892
    },
    {
      "epoch": 0.5288337557245121,
      "grad_norm": 0.0631944015622139,
      "learning_rate": 0.0004891098117565969,
      "loss": 1.491,
      "step": 11894
    },
    {
      "epoch": 0.5289226801831843,
      "grad_norm": 0.06547391414642334,
      "learning_rate": 0.0004889636583198549,
      "loss": 1.4915,
      "step": 11896
    },
    {
      "epoch": 0.5290116046418567,
      "grad_norm": 0.061480239033699036,
      "learning_rate": 0.0004888175058265484,
      "loss": 1.4923,
      "step": 11898
    },
    {
      "epoch": 0.5291005291005291,
      "grad_norm": 0.06591926515102386,
      "learning_rate": 0.0004886713542891707,
      "loss": 1.4942,
      "step": 11900
    },
    {
      "epoch": 0.5291894535592014,
      "grad_norm": 0.06440820544958115,
      "learning_rate": 0.0004885252037202154,
      "loss": 1.4928,
      "step": 11902
    },
    {
      "epoch": 0.5292783780178738,
      "grad_norm": 0.0646994411945343,
      "learning_rate": 0.0004883790541321762,
      "loss": 1.4951,
      "step": 11904
    },
    {
      "epoch": 0.5293673024765462,
      "grad_norm": 0.06326589733362198,
      "learning_rate": 0.0004882329055375469,
      "loss": 1.4959,
      "step": 11906
    },
    {
      "epoch": 0.5294562269352185,
      "grad_norm": 0.06294218450784683,
      "learning_rate": 0.0004880867579488205,
      "loss": 1.4862,
      "step": 11908
    },
    {
      "epoch": 0.5295451513938909,
      "grad_norm": 0.061898913234472275,
      "learning_rate": 0.0004879406113784906,
      "loss": 1.4971,
      "step": 11910
    },
    {
      "epoch": 0.5296340758525633,
      "grad_norm": 0.062190067023038864,
      "learning_rate": 0.00048779446583905055,
      "loss": 1.4951,
      "step": 11912
    },
    {
      "epoch": 0.5297230003112356,
      "grad_norm": 0.06890960782766342,
      "learning_rate": 0.000487648321342993,
      "loss": 1.494,
      "step": 11914
    },
    {
      "epoch": 0.529811924769908,
      "grad_norm": 0.06532379984855652,
      "learning_rate": 0.0004875021779028117,
      "loss": 1.4943,
      "step": 11916
    },
    {
      "epoch": 0.5299008492285803,
      "grad_norm": 0.0694059357047081,
      "learning_rate": 0.000487356035530999,
      "loss": 1.4941,
      "step": 11918
    },
    {
      "epoch": 0.5299897736872526,
      "grad_norm": 0.060807377099990845,
      "learning_rate": 0.0004872098942400484,
      "loss": 1.4945,
      "step": 11920
    },
    {
      "epoch": 0.530078698145925,
      "grad_norm": 0.0658683255314827,
      "learning_rate": 0.00048706375404245227,
      "loss": 1.4913,
      "step": 11922
    },
    {
      "epoch": 0.5301676226045974,
      "grad_norm": 0.062365368008613586,
      "learning_rate": 0.00048691761495070344,
      "loss": 1.4966,
      "step": 11924
    },
    {
      "epoch": 0.5302565470632697,
      "grad_norm": 0.0646410584449768,
      "learning_rate": 0.00048677147697729457,
      "loss": 1.4989,
      "step": 11926
    },
    {
      "epoch": 0.5303454715219421,
      "grad_norm": 0.06210246682167053,
      "learning_rate": 0.00048662534013471816,
      "loss": 1.4891,
      "step": 11928
    },
    {
      "epoch": 0.5304343959806145,
      "grad_norm": 0.06435447186231613,
      "learning_rate": 0.0004864792044354665,
      "loss": 1.4881,
      "step": 11930
    },
    {
      "epoch": 0.5305233204392868,
      "grad_norm": 0.06157999858260155,
      "learning_rate": 0.00048633306989203214,
      "loss": 1.4902,
      "step": 11932
    },
    {
      "epoch": 0.5306122448979592,
      "grad_norm": 0.06224975362420082,
      "learning_rate": 0.0004861869365169071,
      "loss": 1.4926,
      "step": 11934
    },
    {
      "epoch": 0.5307011693566316,
      "grad_norm": 0.06169992312788963,
      "learning_rate": 0.00048604080432258366,
      "loss": 1.493,
      "step": 11936
    },
    {
      "epoch": 0.530790093815304,
      "grad_norm": 0.06507695466279984,
      "learning_rate": 0.00048589467332155385,
      "loss": 1.4962,
      "step": 11938
    },
    {
      "epoch": 0.5308790182739762,
      "grad_norm": 0.06133843585848808,
      "learning_rate": 0.0004857485435263094,
      "loss": 1.4874,
      "step": 11940
    },
    {
      "epoch": 0.5309679427326486,
      "grad_norm": 0.06246451288461685,
      "learning_rate": 0.0004856024149493423,
      "loss": 1.4917,
      "step": 11942
    },
    {
      "epoch": 0.531056867191321,
      "grad_norm": 0.06303883343935013,
      "learning_rate": 0.0004854562876031444,
      "loss": 1.4862,
      "step": 11944
    },
    {
      "epoch": 0.5311457916499933,
      "grad_norm": 0.06277139484882355,
      "learning_rate": 0.0004853101615002068,
      "loss": 1.4925,
      "step": 11946
    },
    {
      "epoch": 0.5312347161086657,
      "grad_norm": 0.06367526948451996,
      "learning_rate": 0.0004851640366530216,
      "loss": 1.4969,
      "step": 11948
    },
    {
      "epoch": 0.531323640567338,
      "grad_norm": 0.06328226625919342,
      "learning_rate": 0.0004850179130740796,
      "loss": 1.4958,
      "step": 11950
    },
    {
      "epoch": 0.5314125650260104,
      "grad_norm": 0.06432175636291504,
      "learning_rate": 0.0004848717907758728,
      "loss": 1.5002,
      "step": 11952
    },
    {
      "epoch": 0.5315014894846828,
      "grad_norm": 0.06231130287051201,
      "learning_rate": 0.0004847256697708919,
      "loss": 1.4921,
      "step": 11954
    },
    {
      "epoch": 0.5315904139433552,
      "grad_norm": 0.06357510387897491,
      "learning_rate": 0.0004845795500716279,
      "loss": 1.487,
      "step": 11956
    },
    {
      "epoch": 0.5316793384020275,
      "grad_norm": 0.0623379684984684,
      "learning_rate": 0.000484433431690572,
      "loss": 1.4889,
      "step": 11958
    },
    {
      "epoch": 0.5317682628606998,
      "grad_norm": 0.060883667320013046,
      "learning_rate": 0.00048428731464021505,
      "loss": 1.4919,
      "step": 11960
    },
    {
      "epoch": 0.5318571873193721,
      "grad_norm": 0.06377879530191422,
      "learning_rate": 0.0004841411989330475,
      "loss": 1.4888,
      "step": 11962
    },
    {
      "epoch": 0.5319461117780445,
      "grad_norm": 0.06436724215745926,
      "learning_rate": 0.00048399508458156027,
      "loss": 1.4935,
      "step": 11964
    },
    {
      "epoch": 0.5320350362367169,
      "grad_norm": 0.06252797693014145,
      "learning_rate": 0.00048384897159824364,
      "loss": 1.4916,
      "step": 11966
    },
    {
      "epoch": 0.5321239606953893,
      "grad_norm": 0.0638202354311943,
      "learning_rate": 0.0004837028599955882,
      "loss": 1.486,
      "step": 11968
    },
    {
      "epoch": 0.5322128851540616,
      "grad_norm": 0.062130406498909,
      "learning_rate": 0.0004835567497860842,
      "loss": 1.4902,
      "step": 11970
    },
    {
      "epoch": 0.532301809612734,
      "grad_norm": 0.06190851330757141,
      "learning_rate": 0.00048341064098222144,
      "loss": 1.4931,
      "step": 11972
    },
    {
      "epoch": 0.5323907340714064,
      "grad_norm": 0.06283093988895416,
      "learning_rate": 0.0004832645335964904,
      "loss": 1.4941,
      "step": 11974
    },
    {
      "epoch": 0.5324796585300787,
      "grad_norm": 0.06219896674156189,
      "learning_rate": 0.0004831184276413808,
      "loss": 1.4913,
      "step": 11976
    },
    {
      "epoch": 0.5325685829887511,
      "grad_norm": 0.06502451747655869,
      "learning_rate": 0.0004829723231293822,
      "loss": 1.4953,
      "step": 11978
    },
    {
      "epoch": 0.5326575074474235,
      "grad_norm": 0.06261445581912994,
      "learning_rate": 0.0004828262200729846,
      "loss": 1.4897,
      "step": 11980
    },
    {
      "epoch": 0.5327464319060957,
      "grad_norm": 0.06267430633306503,
      "learning_rate": 0.0004826801184846774,
      "loss": 1.4912,
      "step": 11982
    },
    {
      "epoch": 0.5328353563647681,
      "grad_norm": 0.06361742317676544,
      "learning_rate": 0.0004825340183769499,
      "loss": 1.4931,
      "step": 11984
    },
    {
      "epoch": 0.5329242808234405,
      "grad_norm": 0.062050919979810715,
      "learning_rate": 0.0004823879197622916,
      "loss": 1.5026,
      "step": 11986
    },
    {
      "epoch": 0.5330132052821128,
      "grad_norm": 0.06152861937880516,
      "learning_rate": 0.00048224182265319134,
      "loss": 1.4903,
      "step": 11988
    },
    {
      "epoch": 0.5331021297407852,
      "grad_norm": 0.061182718724012375,
      "learning_rate": 0.0004820957270621385,
      "loss": 1.4921,
      "step": 11990
    },
    {
      "epoch": 0.5331910541994576,
      "grad_norm": 0.06338909268379211,
      "learning_rate": 0.00048194963300162184,
      "loss": 1.4925,
      "step": 11992
    },
    {
      "epoch": 0.5332799786581299,
      "grad_norm": 0.06249266117811203,
      "learning_rate": 0.0004818035404841299,
      "loss": 1.4878,
      "step": 11994
    },
    {
      "epoch": 0.5333689031168023,
      "grad_norm": 0.06300443410873413,
      "learning_rate": 0.0004816574495221517,
      "loss": 1.4991,
      "step": 11996
    },
    {
      "epoch": 0.5334578275754747,
      "grad_norm": 0.06762851029634476,
      "learning_rate": 0.0004815113601281756,
      "loss": 1.4926,
      "step": 11998
    },
    {
      "epoch": 0.533546752034147,
      "grad_norm": 0.062195733189582825,
      "learning_rate": 0.0004813652723146896,
      "loss": 1.4881,
      "step": 12000
    },
    {
      "epoch": 0.533546752034147,
      "eval_loss": 1.474120020866394,
      "eval_runtime": 12.3992,
      "eval_samples_per_second": 557.296,
      "eval_steps_per_second": 69.682,
      "step": 12000
    },
    {
      "epoch": 0.5336356764928194,
      "grad_norm": 0.06374084204435349,
      "learning_rate": 0.0004812191860941826,
      "loss": 1.4982,
      "step": 12002
    },
    {
      "epoch": 0.5337246009514917,
      "grad_norm": 0.06320127099752426,
      "learning_rate": 0.00048107310147914203,
      "loss": 1.4914,
      "step": 12004
    },
    {
      "epoch": 0.533813525410164,
      "grad_norm": 0.06292546540498734,
      "learning_rate": 0.00048092701848205646,
      "loss": 1.4917,
      "step": 12006
    },
    {
      "epoch": 0.5339024498688364,
      "grad_norm": 0.06197582930326462,
      "learning_rate": 0.00048078093711541325,
      "loss": 1.4799,
      "step": 12008
    },
    {
      "epoch": 0.5339913743275088,
      "grad_norm": 0.06429006904363632,
      "learning_rate": 0.00048063485739170014,
      "loss": 1.4908,
      "step": 12010
    },
    {
      "epoch": 0.5340802987861811,
      "grad_norm": 0.06294738501310349,
      "learning_rate": 0.0004804887793234049,
      "loss": 1.4883,
      "step": 12012
    },
    {
      "epoch": 0.5341692232448535,
      "grad_norm": 0.06511865556240082,
      "learning_rate": 0.00048034270292301475,
      "loss": 1.4928,
      "step": 12014
    },
    {
      "epoch": 0.5342581477035259,
      "grad_norm": 0.0640450119972229,
      "learning_rate": 0.0004801966282030169,
      "loss": 1.4937,
      "step": 12016
    },
    {
      "epoch": 0.5343470721621982,
      "grad_norm": 0.062151722609996796,
      "learning_rate": 0.0004800505551758986,
      "loss": 1.49,
      "step": 12018
    },
    {
      "epoch": 0.5344359966208706,
      "grad_norm": 0.06464537233114243,
      "learning_rate": 0.0004799044838541467,
      "loss": 1.4898,
      "step": 12020
    },
    {
      "epoch": 0.534524921079543,
      "grad_norm": 0.06477969139814377,
      "learning_rate": 0.0004797584142502482,
      "loss": 1.4911,
      "step": 12022
    },
    {
      "epoch": 0.5346138455382153,
      "grad_norm": 0.06363675743341446,
      "learning_rate": 0.00047961234637668973,
      "loss": 1.493,
      "step": 12024
    },
    {
      "epoch": 0.5347027699968876,
      "grad_norm": 0.06466624140739441,
      "learning_rate": 0.0004794662802459574,
      "loss": 1.4926,
      "step": 12026
    },
    {
      "epoch": 0.53479169445556,
      "grad_norm": 0.06489887833595276,
      "learning_rate": 0.00047932021587053833,
      "loss": 1.4876,
      "step": 12028
    },
    {
      "epoch": 0.5348806189142323,
      "grad_norm": 0.06438994407653809,
      "learning_rate": 0.0004791741532629181,
      "loss": 1.4938,
      "step": 12030
    },
    {
      "epoch": 0.5349695433729047,
      "grad_norm": 0.06246186047792435,
      "learning_rate": 0.000479028092435583,
      "loss": 1.4969,
      "step": 12032
    },
    {
      "epoch": 0.5350584678315771,
      "grad_norm": 0.06295336037874222,
      "learning_rate": 0.0004788820334010191,
      "loss": 1.4918,
      "step": 12034
    },
    {
      "epoch": 0.5351473922902494,
      "grad_norm": 0.062076494097709656,
      "learning_rate": 0.000478735976171712,
      "loss": 1.4924,
      "step": 12036
    },
    {
      "epoch": 0.5352363167489218,
      "grad_norm": 0.061363715678453445,
      "learning_rate": 0.0004785899207601473,
      "loss": 1.4896,
      "step": 12038
    },
    {
      "epoch": 0.5353252412075942,
      "grad_norm": 0.06081344187259674,
      "learning_rate": 0.0004784438671788106,
      "loss": 1.4919,
      "step": 12040
    },
    {
      "epoch": 0.5354141656662665,
      "grad_norm": 0.06194039806723595,
      "learning_rate": 0.00047829781544018703,
      "loss": 1.4896,
      "step": 12042
    },
    {
      "epoch": 0.5355030901249389,
      "grad_norm": 0.06131895259022713,
      "learning_rate": 0.00047815176555676193,
      "loss": 1.4981,
      "step": 12044
    },
    {
      "epoch": 0.5355920145836113,
      "grad_norm": 0.060021188110113144,
      "learning_rate": 0.00047800571754102024,
      "loss": 1.4876,
      "step": 12046
    },
    {
      "epoch": 0.5356809390422835,
      "grad_norm": 0.06131913512945175,
      "learning_rate": 0.00047785967140544656,
      "loss": 1.4898,
      "step": 12048
    },
    {
      "epoch": 0.5357698635009559,
      "grad_norm": 0.061648186296224594,
      "learning_rate": 0.00047771362716252585,
      "loss": 1.4876,
      "step": 12050
    },
    {
      "epoch": 0.5358587879596283,
      "grad_norm": 0.061754439026117325,
      "learning_rate": 0.0004775675848247427,
      "loss": 1.4927,
      "step": 12052
    },
    {
      "epoch": 0.5359477124183006,
      "grad_norm": 0.06522786617279053,
      "learning_rate": 0.00047742154440458083,
      "loss": 1.4918,
      "step": 12054
    },
    {
      "epoch": 0.536036636876973,
      "grad_norm": 0.06362829357385635,
      "learning_rate": 0.0004772755059145252,
      "loss": 1.4909,
      "step": 12056
    },
    {
      "epoch": 0.5361255613356454,
      "grad_norm": 0.06584829092025757,
      "learning_rate": 0.0004771294693670592,
      "loss": 1.4919,
      "step": 12058
    },
    {
      "epoch": 0.5362144857943177,
      "grad_norm": 0.06313332915306091,
      "learning_rate": 0.0004769834347746672,
      "loss": 1.4916,
      "step": 12060
    },
    {
      "epoch": 0.5363034102529901,
      "grad_norm": 0.06333259493112564,
      "learning_rate": 0.00047683740214983256,
      "loss": 1.487,
      "step": 12062
    },
    {
      "epoch": 0.5363923347116625,
      "grad_norm": 0.06263703107833862,
      "learning_rate": 0.00047669137150503876,
      "loss": 1.485,
      "step": 12064
    },
    {
      "epoch": 0.5364812591703348,
      "grad_norm": 0.06329838186502457,
      "learning_rate": 0.0004765453428527693,
      "loss": 1.4955,
      "step": 12066
    },
    {
      "epoch": 0.5365701836290072,
      "grad_norm": 0.06317045539617538,
      "learning_rate": 0.0004763993162055074,
      "loss": 1.4871,
      "step": 12068
    },
    {
      "epoch": 0.5366591080876795,
      "grad_norm": 0.06274587661027908,
      "learning_rate": 0.0004762532915757357,
      "loss": 1.4957,
      "step": 12070
    },
    {
      "epoch": 0.5367480325463518,
      "grad_norm": 0.06153559684753418,
      "learning_rate": 0.00047610726897593747,
      "loss": 1.491,
      "step": 12072
    },
    {
      "epoch": 0.5368369570050242,
      "grad_norm": 0.061654843389987946,
      "learning_rate": 0.00047596124841859505,
      "loss": 1.4954,
      "step": 12074
    },
    {
      "epoch": 0.5369258814636966,
      "grad_norm": 0.06349113583564758,
      "learning_rate": 0.0004758152299161912,
      "loss": 1.4941,
      "step": 12076
    },
    {
      "epoch": 0.5370148059223689,
      "grad_norm": 0.06033482402563095,
      "learning_rate": 0.0004756692134812081,
      "loss": 1.4873,
      "step": 12078
    },
    {
      "epoch": 0.5371037303810413,
      "grad_norm": 0.06259752810001373,
      "learning_rate": 0.00047552319912612764,
      "loss": 1.4953,
      "step": 12080
    },
    {
      "epoch": 0.5371926548397137,
      "grad_norm": 0.06465926766395569,
      "learning_rate": 0.00047537718686343225,
      "loss": 1.4911,
      "step": 12082
    },
    {
      "epoch": 0.537281579298386,
      "grad_norm": 0.06145221367478371,
      "learning_rate": 0.00047523117670560333,
      "loss": 1.4896,
      "step": 12084
    },
    {
      "epoch": 0.5373705037570584,
      "grad_norm": 0.06425707787275314,
      "learning_rate": 0.0004750851686651225,
      "loss": 1.4916,
      "step": 12086
    },
    {
      "epoch": 0.5374594282157308,
      "grad_norm": 0.06472969055175781,
      "learning_rate": 0.0004749391627544713,
      "loss": 1.4974,
      "step": 12088
    },
    {
      "epoch": 0.537548352674403,
      "grad_norm": 0.0711335614323616,
      "learning_rate": 0.0004747931589861308,
      "loss": 1.4928,
      "step": 12090
    },
    {
      "epoch": 0.5376372771330754,
      "grad_norm": 0.06172974407672882,
      "learning_rate": 0.0004746471573725823,
      "loss": 1.4879,
      "step": 12092
    },
    {
      "epoch": 0.5377262015917478,
      "grad_norm": 0.062273457646369934,
      "learning_rate": 0.0004745011579263065,
      "loss": 1.4867,
      "step": 12094
    },
    {
      "epoch": 0.5378151260504201,
      "grad_norm": 0.06254924833774567,
      "learning_rate": 0.00047435516065978396,
      "loss": 1.4943,
      "step": 12096
    },
    {
      "epoch": 0.5379040505090925,
      "grad_norm": 0.06490933150053024,
      "learning_rate": 0.0004742091655854954,
      "loss": 1.4962,
      "step": 12098
    },
    {
      "epoch": 0.5379929749677649,
      "grad_norm": 0.06268811225891113,
      "learning_rate": 0.0004740631727159211,
      "loss": 1.4896,
      "step": 12100
    },
    {
      "epoch": 0.5380818994264372,
      "grad_norm": 0.06337209790945053,
      "learning_rate": 0.00047391718206354097,
      "loss": 1.4944,
      "step": 12102
    },
    {
      "epoch": 0.5381708238851096,
      "grad_norm": 0.06212884932756424,
      "learning_rate": 0.0004737711936408351,
      "loss": 1.4923,
      "step": 12104
    },
    {
      "epoch": 0.538259748343782,
      "grad_norm": 0.06264811009168625,
      "learning_rate": 0.00047362520746028336,
      "loss": 1.4862,
      "step": 12106
    },
    {
      "epoch": 0.5383486728024544,
      "grad_norm": 0.06359831243753433,
      "learning_rate": 0.00047347922353436475,
      "loss": 1.4975,
      "step": 12108
    },
    {
      "epoch": 0.5384375972611267,
      "grad_norm": 0.06291768699884415,
      "learning_rate": 0.00047333324187555934,
      "loss": 1.4913,
      "step": 12110
    },
    {
      "epoch": 0.538526521719799,
      "grad_norm": 0.06421522051095963,
      "learning_rate": 0.0004731872624963456,
      "loss": 1.493,
      "step": 12112
    },
    {
      "epoch": 0.5386154461784713,
      "grad_norm": 0.06364668905735016,
      "learning_rate": 0.0004730412854092031,
      "loss": 1.4885,
      "step": 12114
    },
    {
      "epoch": 0.5387043706371437,
      "grad_norm": 0.06267435848712921,
      "learning_rate": 0.0004728953106266103,
      "loss": 1.4859,
      "step": 12116
    },
    {
      "epoch": 0.5387932950958161,
      "grad_norm": 0.06271584331989288,
      "learning_rate": 0.0004727493381610456,
      "loss": 1.488,
      "step": 12118
    },
    {
      "epoch": 0.5388822195544885,
      "grad_norm": 0.06468139588832855,
      "learning_rate": 0.0004726033680249877,
      "loss": 1.4919,
      "step": 12120
    },
    {
      "epoch": 0.5389711440131608,
      "grad_norm": 0.06396504491567612,
      "learning_rate": 0.0004724574002309147,
      "loss": 1.4947,
      "step": 12122
    },
    {
      "epoch": 0.5390600684718332,
      "grad_norm": 0.06503921747207642,
      "learning_rate": 0.00047231143479130436,
      "loss": 1.4909,
      "step": 12124
    },
    {
      "epoch": 0.5391489929305056,
      "grad_norm": 0.06351353228092194,
      "learning_rate": 0.00047216547171863467,
      "loss": 1.492,
      "step": 12126
    },
    {
      "epoch": 0.5392379173891779,
      "grad_norm": 0.06411804258823395,
      "learning_rate": 0.000472019511025383,
      "loss": 1.4905,
      "step": 12128
    },
    {
      "epoch": 0.5393268418478503,
      "grad_norm": 0.06250099092721939,
      "learning_rate": 0.000471873552724027,
      "loss": 1.4884,
      "step": 12130
    },
    {
      "epoch": 0.5394157663065227,
      "grad_norm": 0.06506571918725967,
      "learning_rate": 0.00047172759682704375,
      "loss": 1.4917,
      "step": 12132
    },
    {
      "epoch": 0.5395046907651949,
      "grad_norm": 0.06468180567026138,
      "learning_rate": 0.0004715816433469097,
      "loss": 1.493,
      "step": 12134
    },
    {
      "epoch": 0.5395936152238673,
      "grad_norm": 0.06447789818048477,
      "learning_rate": 0.00047143569229610247,
      "loss": 1.4964,
      "step": 12136
    },
    {
      "epoch": 0.5396825396825397,
      "grad_norm": 0.06448780745267868,
      "learning_rate": 0.00047128974368709797,
      "loss": 1.4869,
      "step": 12138
    },
    {
      "epoch": 0.539771464141212,
      "grad_norm": 0.06413755565881729,
      "learning_rate": 0.00047114379753237265,
      "loss": 1.4866,
      "step": 12140
    },
    {
      "epoch": 0.5398603885998844,
      "grad_norm": 0.0629299134016037,
      "learning_rate": 0.0004709978538444027,
      "loss": 1.4883,
      "step": 12142
    },
    {
      "epoch": 0.5399493130585568,
      "grad_norm": 0.06269446760416031,
      "learning_rate": 0.000470851912635664,
      "loss": 1.4944,
      "step": 12144
    },
    {
      "epoch": 0.5400382375172291,
      "grad_norm": 0.0642932578921318,
      "learning_rate": 0.0004707059739186323,
      "loss": 1.4922,
      "step": 12146
    },
    {
      "epoch": 0.5401271619759015,
      "grad_norm": 0.06270511448383331,
      "learning_rate": 0.000470560037705783,
      "loss": 1.4956,
      "step": 12148
    },
    {
      "epoch": 0.5402160864345739,
      "grad_norm": 0.06323356926441193,
      "learning_rate": 0.0004704141040095914,
      "loss": 1.4927,
      "step": 12150
    },
    {
      "epoch": 0.5403050108932462,
      "grad_norm": 0.06155591458082199,
      "learning_rate": 0.00047026817284253264,
      "loss": 1.492,
      "step": 12152
    },
    {
      "epoch": 0.5403939353519186,
      "grad_norm": 0.06190359592437744,
      "learning_rate": 0.0004701222442170815,
      "loss": 1.4937,
      "step": 12154
    },
    {
      "epoch": 0.5404828598105909,
      "grad_norm": 0.06423845887184143,
      "learning_rate": 0.00046997631814571245,
      "loss": 1.4927,
      "step": 12156
    },
    {
      "epoch": 0.5405717842692632,
      "grad_norm": 0.06188628450036049,
      "learning_rate": 0.0004698303946409002,
      "loss": 1.4844,
      "step": 12158
    },
    {
      "epoch": 0.5406607087279356,
      "grad_norm": 0.06377921998500824,
      "learning_rate": 0.00046968447371511883,
      "loss": 1.4897,
      "step": 12160
    },
    {
      "epoch": 0.540749633186608,
      "grad_norm": 0.06156056001782417,
      "learning_rate": 0.00046953855538084193,
      "loss": 1.4852,
      "step": 12162
    },
    {
      "epoch": 0.5408385576452803,
      "grad_norm": 0.06201445311307907,
      "learning_rate": 0.0004693926396505439,
      "loss": 1.4835,
      "step": 12164
    },
    {
      "epoch": 0.5409274821039527,
      "grad_norm": 0.0629076287150383,
      "learning_rate": 0.0004692467265366976,
      "loss": 1.4976,
      "step": 12166
    },
    {
      "epoch": 0.5410164065626251,
      "grad_norm": 0.062005601823329926,
      "learning_rate": 0.00046910081605177695,
      "loss": 1.4922,
      "step": 12168
    },
    {
      "epoch": 0.5411053310212974,
      "grad_norm": 0.06382474303245544,
      "learning_rate": 0.0004689549082082546,
      "loss": 1.4909,
      "step": 12170
    },
    {
      "epoch": 0.5411942554799698,
      "grad_norm": 0.061615731567144394,
      "learning_rate": 0.00046880900301860337,
      "loss": 1.4913,
      "step": 12172
    },
    {
      "epoch": 0.5412831799386422,
      "grad_norm": 0.061972636729478836,
      "learning_rate": 0.00046866310049529613,
      "loss": 1.485,
      "step": 12174
    },
    {
      "epoch": 0.5413721043973145,
      "grad_norm": 0.06166384369134903,
      "learning_rate": 0.0004685172006508052,
      "loss": 1.4903,
      "step": 12176
    },
    {
      "epoch": 0.5414610288559868,
      "grad_norm": 0.06348898261785507,
      "learning_rate": 0.00046837130349760257,
      "loss": 1.4945,
      "step": 12178
    },
    {
      "epoch": 0.5415499533146592,
      "grad_norm": 0.062341634184122086,
      "learning_rate": 0.0004682254090481604,
      "loss": 1.4892,
      "step": 12180
    },
    {
      "epoch": 0.5416388777733315,
      "grad_norm": 0.062497496604919434,
      "learning_rate": 0.0004680795173149502,
      "loss": 1.4911,
      "step": 12182
    },
    {
      "epoch": 0.5417278022320039,
      "grad_norm": 0.06050288304686546,
      "learning_rate": 0.0004679336283104436,
      "loss": 1.4854,
      "step": 12184
    },
    {
      "epoch": 0.5418167266906763,
      "grad_norm": 0.061905790120363235,
      "learning_rate": 0.00046778774204711196,
      "loss": 1.494,
      "step": 12186
    },
    {
      "epoch": 0.5419056511493486,
      "grad_norm": 0.0627816841006279,
      "learning_rate": 0.00046764185853742564,
      "loss": 1.4943,
      "step": 12188
    },
    {
      "epoch": 0.541994575608021,
      "grad_norm": 0.06469935178756714,
      "learning_rate": 0.0004674959777938562,
      "loss": 1.4926,
      "step": 12190
    },
    {
      "epoch": 0.5420835000666934,
      "grad_norm": 0.06236010789871216,
      "learning_rate": 0.0004673500998288738,
      "loss": 1.4896,
      "step": 12192
    },
    {
      "epoch": 0.5421724245253657,
      "grad_norm": 0.06338387727737427,
      "learning_rate": 0.00046720422465494854,
      "loss": 1.4871,
      "step": 12194
    },
    {
      "epoch": 0.5422613489840381,
      "grad_norm": 0.06159103289246559,
      "learning_rate": 0.00046705835228455083,
      "loss": 1.4935,
      "step": 12196
    },
    {
      "epoch": 0.5423502734427104,
      "grad_norm": 0.06336644291877747,
      "learning_rate": 0.0004669124827301503,
      "loss": 1.4883,
      "step": 12198
    },
    {
      "epoch": 0.5424391979013827,
      "grad_norm": 0.061430975794792175,
      "learning_rate": 0.00046676661600421657,
      "loss": 1.4915,
      "step": 12200
    },
    {
      "epoch": 0.5425281223600551,
      "grad_norm": 0.06243472546339035,
      "learning_rate": 0.000466620752119219,
      "loss": 1.491,
      "step": 12202
    },
    {
      "epoch": 0.5426170468187275,
      "grad_norm": 0.06282027810811996,
      "learning_rate": 0.0004664748910876265,
      "loss": 1.4878,
      "step": 12204
    },
    {
      "epoch": 0.5427059712773998,
      "grad_norm": 0.06300549954175949,
      "learning_rate": 0.00046632903292190823,
      "loss": 1.493,
      "step": 12206
    },
    {
      "epoch": 0.5427948957360722,
      "grad_norm": 0.062161967158317566,
      "learning_rate": 0.00046618317763453263,
      "loss": 1.4932,
      "step": 12208
    },
    {
      "epoch": 0.5428838201947446,
      "grad_norm": 0.062358465045690536,
      "learning_rate": 0.0004660373252379677,
      "loss": 1.4947,
      "step": 12210
    },
    {
      "epoch": 0.5429727446534169,
      "grad_norm": 0.06448014080524445,
      "learning_rate": 0.00046589147574468227,
      "loss": 1.4914,
      "step": 12212
    },
    {
      "epoch": 0.5430616691120893,
      "grad_norm": 0.06277332454919815,
      "learning_rate": 0.00046574562916714345,
      "loss": 1.4838,
      "step": 12214
    },
    {
      "epoch": 0.5431505935707617,
      "grad_norm": 0.06348789483308792,
      "learning_rate": 0.0004655997855178195,
      "loss": 1.4943,
      "step": 12216
    },
    {
      "epoch": 0.543239518029434,
      "grad_norm": 0.0647101104259491,
      "learning_rate": 0.0004654539448091774,
      "loss": 1.4911,
      "step": 12218
    },
    {
      "epoch": 0.5433284424881063,
      "grad_norm": 0.06389828771352768,
      "learning_rate": 0.0004653081070536843,
      "loss": 1.4876,
      "step": 12220
    },
    {
      "epoch": 0.5434173669467787,
      "grad_norm": 0.06637488305568695,
      "learning_rate": 0.00046516227226380707,
      "loss": 1.4933,
      "step": 12222
    },
    {
      "epoch": 0.543506291405451,
      "grad_norm": 0.06405807286500931,
      "learning_rate": 0.00046501644045201244,
      "loss": 1.4877,
      "step": 12224
    },
    {
      "epoch": 0.5435952158641234,
      "grad_norm": 0.06238565593957901,
      "learning_rate": 0.0004648706116307664,
      "loss": 1.49,
      "step": 12226
    },
    {
      "epoch": 0.5436841403227958,
      "grad_norm": 0.06455307453870773,
      "learning_rate": 0.0004647247858125355,
      "loss": 1.4907,
      "step": 12228
    },
    {
      "epoch": 0.5437730647814681,
      "grad_norm": 0.06793267279863358,
      "learning_rate": 0.0004645789630097854,
      "loss": 1.4835,
      "step": 12230
    },
    {
      "epoch": 0.5438619892401405,
      "grad_norm": 0.06355708837509155,
      "learning_rate": 0.0004644331432349815,
      "loss": 1.4914,
      "step": 12232
    },
    {
      "epoch": 0.5439509136988129,
      "grad_norm": 0.06357616931200027,
      "learning_rate": 0.0004642873265005893,
      "loss": 1.4844,
      "step": 12234
    },
    {
      "epoch": 0.5440398381574852,
      "grad_norm": 0.06270436942577362,
      "learning_rate": 0.0004641415128190737,
      "loss": 1.4898,
      "step": 12236
    },
    {
      "epoch": 0.5441287626161576,
      "grad_norm": 0.06114961579442024,
      "learning_rate": 0.0004639957022028998,
      "loss": 1.4918,
      "step": 12238
    },
    {
      "epoch": 0.54421768707483,
      "grad_norm": 0.06168103963136673,
      "learning_rate": 0.00046384989466453203,
      "loss": 1.4879,
      "step": 12240
    },
    {
      "epoch": 0.5443066115335022,
      "grad_norm": 0.06130845472216606,
      "learning_rate": 0.0004637040902164342,
      "loss": 1.4863,
      "step": 12242
    },
    {
      "epoch": 0.5443955359921746,
      "grad_norm": 0.060740306973457336,
      "learning_rate": 0.00046355828887107103,
      "loss": 1.4925,
      "step": 12244
    },
    {
      "epoch": 0.544484460450847,
      "grad_norm": 0.06276367604732513,
      "learning_rate": 0.00046341249064090585,
      "loss": 1.4919,
      "step": 12246
    },
    {
      "epoch": 0.5445733849095193,
      "grad_norm": 0.06420931220054626,
      "learning_rate": 0.00046326669553840204,
      "loss": 1.4946,
      "step": 12248
    },
    {
      "epoch": 0.5446623093681917,
      "grad_norm": 0.06452270597219467,
      "learning_rate": 0.00046312090357602316,
      "loss": 1.4963,
      "step": 12250
    },
    {
      "epoch": 0.5447512338268641,
      "grad_norm": 0.06321154534816742,
      "learning_rate": 0.0004629751147662318,
      "loss": 1.4855,
      "step": 12252
    },
    {
      "epoch": 0.5448401582855364,
      "grad_norm": 0.06343185901641846,
      "learning_rate": 0.0004628293291214909,
      "loss": 1.4905,
      "step": 12254
    },
    {
      "epoch": 0.5449290827442088,
      "grad_norm": 0.062312331050634384,
      "learning_rate": 0.0004626835466542628,
      "loss": 1.4856,
      "step": 12256
    },
    {
      "epoch": 0.5450180072028812,
      "grad_norm": 0.0644093006849289,
      "learning_rate": 0.0004625377673770094,
      "loss": 1.4834,
      "step": 12258
    },
    {
      "epoch": 0.5451069316615536,
      "grad_norm": 0.06491252779960632,
      "learning_rate": 0.0004623919913021929,
      "loss": 1.4909,
      "step": 12260
    },
    {
      "epoch": 0.5451958561202259,
      "grad_norm": 0.06439858675003052,
      "learning_rate": 0.00046224621844227477,
      "loss": 1.4882,
      "step": 12262
    },
    {
      "epoch": 0.5452847805788982,
      "grad_norm": 0.06259090453386307,
      "learning_rate": 0.00046210044880971596,
      "loss": 1.487,
      "step": 12264
    },
    {
      "epoch": 0.5453737050375705,
      "grad_norm": 0.06402361392974854,
      "learning_rate": 0.0004619546824169781,
      "loss": 1.4849,
      "step": 12266
    },
    {
      "epoch": 0.5454626294962429,
      "grad_norm": 0.06298209726810455,
      "learning_rate": 0.0004618089192765214,
      "loss": 1.4888,
      "step": 12268
    },
    {
      "epoch": 0.5455515539549153,
      "grad_norm": 0.06361270695924759,
      "learning_rate": 0.0004616631594008069,
      "loss": 1.4922,
      "step": 12270
    },
    {
      "epoch": 0.5456404784135876,
      "grad_norm": 0.06214560195803642,
      "learning_rate": 0.00046151740280229433,
      "loss": 1.4865,
      "step": 12272
    },
    {
      "epoch": 0.54572940287226,
      "grad_norm": 0.06411103904247284,
      "learning_rate": 0.0004613716494934437,
      "loss": 1.4955,
      "step": 12274
    },
    {
      "epoch": 0.5458183273309324,
      "grad_norm": 0.06292589753866196,
      "learning_rate": 0.0004612258994867147,
      "loss": 1.4852,
      "step": 12276
    },
    {
      "epoch": 0.5459072517896048,
      "grad_norm": 0.06273365765810013,
      "learning_rate": 0.00046108015279456676,
      "loss": 1.4856,
      "step": 12278
    },
    {
      "epoch": 0.5459961762482771,
      "grad_norm": 0.06191820278763771,
      "learning_rate": 0.00046093440942945874,
      "loss": 1.4907,
      "step": 12280
    },
    {
      "epoch": 0.5460851007069495,
      "grad_norm": 0.06268199533224106,
      "learning_rate": 0.00046078866940384974,
      "loss": 1.4926,
      "step": 12282
    },
    {
      "epoch": 0.5461740251656219,
      "grad_norm": 0.06177908927202225,
      "learning_rate": 0.00046064293273019797,
      "loss": 1.4907,
      "step": 12284
    },
    {
      "epoch": 0.5462629496242941,
      "grad_norm": 0.06444894522428513,
      "learning_rate": 0.0004604971994209617,
      "loss": 1.4875,
      "step": 12286
    },
    {
      "epoch": 0.5463518740829665,
      "grad_norm": 0.06176155060529709,
      "learning_rate": 0.000460351469488599,
      "loss": 1.4836,
      "step": 12288
    },
    {
      "epoch": 0.5464407985416389,
      "grad_norm": 0.06493420898914337,
      "learning_rate": 0.0004602057429455673,
      "loss": 1.4828,
      "step": 12290
    },
    {
      "epoch": 0.5465297230003112,
      "grad_norm": 0.0612507201731205,
      "learning_rate": 0.00046006001980432423,
      "loss": 1.487,
      "step": 12292
    },
    {
      "epoch": 0.5466186474589836,
      "grad_norm": 0.06377555429935455,
      "learning_rate": 0.0004599143000773267,
      "loss": 1.483,
      "step": 12294
    },
    {
      "epoch": 0.546707571917656,
      "grad_norm": 0.06083401292562485,
      "learning_rate": 0.00045976858377703116,
      "loss": 1.4897,
      "step": 12296
    },
    {
      "epoch": 0.5467964963763283,
      "grad_norm": 0.06468698382377625,
      "learning_rate": 0.0004596228709158947,
      "loss": 1.4917,
      "step": 12298
    },
    {
      "epoch": 0.5468854208350007,
      "grad_norm": 0.06334157288074493,
      "learning_rate": 0.0004594771615063731,
      "loss": 1.4881,
      "step": 12300
    },
    {
      "epoch": 0.5469743452936731,
      "grad_norm": 0.06522412598133087,
      "learning_rate": 0.00045933145556092215,
      "loss": 1.4908,
      "step": 12302
    },
    {
      "epoch": 0.5470632697523454,
      "grad_norm": 0.06641488522291183,
      "learning_rate": 0.0004591857530919977,
      "loss": 1.4941,
      "step": 12304
    },
    {
      "epoch": 0.5471521942110178,
      "grad_norm": 0.06378214806318283,
      "learning_rate": 0.0004590400541120549,
      "loss": 1.4886,
      "step": 12306
    },
    {
      "epoch": 0.54724111866969,
      "grad_norm": 0.06377306580543518,
      "learning_rate": 0.00045889435863354886,
      "loss": 1.4878,
      "step": 12308
    },
    {
      "epoch": 0.5473300431283624,
      "grad_norm": 0.062183529138565063,
      "learning_rate": 0.0004587486666689342,
      "loss": 1.4899,
      "step": 12310
    },
    {
      "epoch": 0.5474189675870348,
      "grad_norm": 0.062393732368946075,
      "learning_rate": 0.0004586029782306651,
      "loss": 1.4875,
      "step": 12312
    },
    {
      "epoch": 0.5475078920457072,
      "grad_norm": 0.06595607101917267,
      "learning_rate": 0.0004584572933311961,
      "loss": 1.4902,
      "step": 12314
    },
    {
      "epoch": 0.5475968165043795,
      "grad_norm": 0.06217242404818535,
      "learning_rate": 0.0004583116119829807,
      "loss": 1.4901,
      "step": 12316
    },
    {
      "epoch": 0.5476857409630519,
      "grad_norm": 0.06375078856945038,
      "learning_rate": 0.0004581659341984722,
      "loss": 1.4894,
      "step": 12318
    },
    {
      "epoch": 0.5477746654217243,
      "grad_norm": 0.06128295138478279,
      "learning_rate": 0.00045802025999012433,
      "loss": 1.4846,
      "step": 12320
    },
    {
      "epoch": 0.5478635898803966,
      "grad_norm": 0.06317851692438126,
      "learning_rate": 0.00045787458937038923,
      "loss": 1.4912,
      "step": 12322
    },
    {
      "epoch": 0.547952514339069,
      "grad_norm": 0.06150876730680466,
      "learning_rate": 0.00045772892235172025,
      "loss": 1.4873,
      "step": 12324
    },
    {
      "epoch": 0.5480414387977414,
      "grad_norm": 0.061902642250061035,
      "learning_rate": 0.00045758325894656917,
      "loss": 1.4865,
      "step": 12326
    },
    {
      "epoch": 0.5481303632564136,
      "grad_norm": 0.06313225626945496,
      "learning_rate": 0.0004574375991673879,
      "loss": 1.4819,
      "step": 12328
    },
    {
      "epoch": 0.548219287715086,
      "grad_norm": 0.06134433671832085,
      "learning_rate": 0.0004572919430266284,
      "loss": 1.4875,
      "step": 12330
    },
    {
      "epoch": 0.5483082121737584,
      "grad_norm": 0.06121361255645752,
      "learning_rate": 0.0004571462905367417,
      "loss": 1.4894,
      "step": 12332
    },
    {
      "epoch": 0.5483971366324307,
      "grad_norm": 0.06105126067996025,
      "learning_rate": 0.0004570006417101789,
      "loss": 1.4887,
      "step": 12334
    },
    {
      "epoch": 0.5484860610911031,
      "grad_norm": 0.061685241758823395,
      "learning_rate": 0.0004568549965593908,
      "loss": 1.4945,
      "step": 12336
    },
    {
      "epoch": 0.5485749855497755,
      "grad_norm": 0.06117040663957596,
      "learning_rate": 0.0004567093550968277,
      "loss": 1.4852,
      "step": 12338
    },
    {
      "epoch": 0.5486639100084478,
      "grad_norm": 0.06458423286676407,
      "learning_rate": 0.00045656371733493977,
      "loss": 1.4911,
      "step": 12340
    },
    {
      "epoch": 0.5487528344671202,
      "grad_norm": 0.06184089928865433,
      "learning_rate": 0.00045641808328617667,
      "loss": 1.4882,
      "step": 12342
    },
    {
      "epoch": 0.5488417589257926,
      "grad_norm": 0.06287356466054916,
      "learning_rate": 0.0004562724529629878,
      "loss": 1.4925,
      "step": 12344
    },
    {
      "epoch": 0.5489306833844649,
      "grad_norm": 0.06341803073883057,
      "learning_rate": 0.00045612682637782246,
      "loss": 1.487,
      "step": 12346
    },
    {
      "epoch": 0.5490196078431373,
      "grad_norm": 0.06103647127747536,
      "learning_rate": 0.00045598120354312945,
      "loss": 1.4941,
      "step": 12348
    },
    {
      "epoch": 0.5491085323018096,
      "grad_norm": 0.061173129826784134,
      "learning_rate": 0.0004558355844713568,
      "loss": 1.4914,
      "step": 12350
    },
    {
      "epoch": 0.5491974567604819,
      "grad_norm": 0.06562421470880508,
      "learning_rate": 0.00045568996917495335,
      "loss": 1.4899,
      "step": 12352
    },
    {
      "epoch": 0.5492863812191543,
      "grad_norm": 0.06319348514080048,
      "learning_rate": 0.00045554435766636647,
      "loss": 1.4836,
      "step": 12354
    },
    {
      "epoch": 0.5493753056778267,
      "grad_norm": 0.06338098645210266,
      "learning_rate": 0.0004553987499580438,
      "loss": 1.4881,
      "step": 12356
    },
    {
      "epoch": 0.549464230136499,
      "grad_norm": 0.06358245015144348,
      "learning_rate": 0.0004552531460624325,
      "loss": 1.4929,
      "step": 12358
    },
    {
      "epoch": 0.5495531545951714,
      "grad_norm": 0.06360200047492981,
      "learning_rate": 0.00045510754599197944,
      "loss": 1.4937,
      "step": 12360
    },
    {
      "epoch": 0.5496420790538438,
      "grad_norm": 0.06298113614320755,
      "learning_rate": 0.0004549619497591313,
      "loss": 1.4883,
      "step": 12362
    },
    {
      "epoch": 0.5497310035125161,
      "grad_norm": 0.0645737498998642,
      "learning_rate": 0.0004548163573763341,
      "loss": 1.4853,
      "step": 12364
    },
    {
      "epoch": 0.5498199279711885,
      "grad_norm": 0.06510405987501144,
      "learning_rate": 0.00045467076885603375,
      "loss": 1.4872,
      "step": 12366
    },
    {
      "epoch": 0.5499088524298609,
      "grad_norm": 0.06831739097833633,
      "learning_rate": 0.00045452518421067596,
      "loss": 1.4967,
      "step": 12368
    },
    {
      "epoch": 0.5499977768885332,
      "grad_norm": 0.06525734066963196,
      "learning_rate": 0.0004543796034527059,
      "loss": 1.487,
      "step": 12370
    },
    {
      "epoch": 0.5500867013472055,
      "grad_norm": 0.06312596797943115,
      "learning_rate": 0.00045423402659456805,
      "loss": 1.4896,
      "step": 12372
    },
    {
      "epoch": 0.5501756258058779,
      "grad_norm": 0.06690060347318649,
      "learning_rate": 0.0004540884536487076,
      "loss": 1.4905,
      "step": 12374
    },
    {
      "epoch": 0.5502645502645502,
      "grad_norm": 0.06204893812537193,
      "learning_rate": 0.0004539428846275682,
      "loss": 1.4871,
      "step": 12376
    },
    {
      "epoch": 0.5503534747232226,
      "grad_norm": 0.06456583738327026,
      "learning_rate": 0.00045379731954359425,
      "loss": 1.4915,
      "step": 12378
    },
    {
      "epoch": 0.550442399181895,
      "grad_norm": 0.06286507844924927,
      "learning_rate": 0.00045365175840922893,
      "loss": 1.4849,
      "step": 12380
    },
    {
      "epoch": 0.5505313236405673,
      "grad_norm": 0.06443874537944794,
      "learning_rate": 0.00045350620123691543,
      "loss": 1.4932,
      "step": 12382
    },
    {
      "epoch": 0.5506202480992397,
      "grad_norm": 0.06251480430364609,
      "learning_rate": 0.00045336064803909687,
      "loss": 1.4924,
      "step": 12384
    },
    {
      "epoch": 0.5507091725579121,
      "grad_norm": 0.06327617913484573,
      "learning_rate": 0.00045321509882821566,
      "loss": 1.4907,
      "step": 12386
    },
    {
      "epoch": 0.5507980970165844,
      "grad_norm": 0.06285477429628372,
      "learning_rate": 0.0004530695536167139,
      "loss": 1.4884,
      "step": 12388
    },
    {
      "epoch": 0.5508870214752568,
      "grad_norm": 0.06266319751739502,
      "learning_rate": 0.00045292401241703355,
      "loss": 1.487,
      "step": 12390
    },
    {
      "epoch": 0.5509759459339292,
      "grad_norm": 0.06416985392570496,
      "learning_rate": 0.000452778475241616,
      "loss": 1.491,
      "step": 12392
    },
    {
      "epoch": 0.5510648703926014,
      "grad_norm": 0.06277859956026077,
      "learning_rate": 0.00045263294210290265,
      "loss": 1.4911,
      "step": 12394
    },
    {
      "epoch": 0.5511537948512738,
      "grad_norm": 0.06268072873353958,
      "learning_rate": 0.0004524874130133341,
      "loss": 1.4863,
      "step": 12396
    },
    {
      "epoch": 0.5512427193099462,
      "grad_norm": 0.06368980556726456,
      "learning_rate": 0.0004523418879853507,
      "loss": 1.4886,
      "step": 12398
    },
    {
      "epoch": 0.5513316437686185,
      "grad_norm": 0.06306835263967514,
      "learning_rate": 0.0004521963670313929,
      "loss": 1.4822,
      "step": 12400
    },
    {
      "epoch": 0.5514205682272909,
      "grad_norm": 0.06334751844406128,
      "learning_rate": 0.0004520508501639004,
      "loss": 1.4873,
      "step": 12402
    },
    {
      "epoch": 0.5515094926859633,
      "grad_norm": 0.06404084712266922,
      "learning_rate": 0.0004519053373953122,
      "loss": 1.4901,
      "step": 12404
    },
    {
      "epoch": 0.5515984171446356,
      "grad_norm": 0.06552892923355103,
      "learning_rate": 0.0004517598287380681,
      "loss": 1.4888,
      "step": 12406
    },
    {
      "epoch": 0.551687341603308,
      "grad_norm": 0.06249921768903732,
      "learning_rate": 0.000451614324204606,
      "loss": 1.487,
      "step": 12408
    },
    {
      "epoch": 0.5517762660619804,
      "grad_norm": 0.06481725722551346,
      "learning_rate": 0.0004514688238073651,
      "loss": 1.4913,
      "step": 12410
    },
    {
      "epoch": 0.5518651905206527,
      "grad_norm": 0.06446146219968796,
      "learning_rate": 0.0004513233275587829,
      "loss": 1.4864,
      "step": 12412
    },
    {
      "epoch": 0.5519541149793251,
      "grad_norm": 0.0644034743309021,
      "learning_rate": 0.00045117783547129705,
      "loss": 1.4838,
      "step": 12414
    },
    {
      "epoch": 0.5520430394379974,
      "grad_norm": 0.06331180036067963,
      "learning_rate": 0.0004510323475573451,
      "loss": 1.4922,
      "step": 12416
    },
    {
      "epoch": 0.5521319638966697,
      "grad_norm": 0.06320036947727203,
      "learning_rate": 0.0004508868638293638,
      "loss": 1.4861,
      "step": 12418
    },
    {
      "epoch": 0.5522208883553421,
      "grad_norm": 0.06392395496368408,
      "learning_rate": 0.00045074138429978974,
      "loss": 1.4894,
      "step": 12420
    },
    {
      "epoch": 0.5523098128140145,
      "grad_norm": 0.06227342411875725,
      "learning_rate": 0.00045059590898105935,
      "loss": 1.4881,
      "step": 12422
    },
    {
      "epoch": 0.5523987372726868,
      "grad_norm": 0.06373913586139679,
      "learning_rate": 0.0004504504378856085,
      "loss": 1.4876,
      "step": 12424
    },
    {
      "epoch": 0.5524876617313592,
      "grad_norm": 0.06222604960203171,
      "learning_rate": 0.00045030497102587223,
      "loss": 1.4854,
      "step": 12426
    },
    {
      "epoch": 0.5525765861900316,
      "grad_norm": 0.06317233294248581,
      "learning_rate": 0.00045015950841428636,
      "loss": 1.4906,
      "step": 12428
    },
    {
      "epoch": 0.552665510648704,
      "grad_norm": 0.06286189705133438,
      "learning_rate": 0.00045001405006328505,
      "loss": 1.487,
      "step": 12430
    },
    {
      "epoch": 0.5527544351073763,
      "grad_norm": 0.06075560674071312,
      "learning_rate": 0.00044986859598530334,
      "loss": 1.4837,
      "step": 12432
    },
    {
      "epoch": 0.5528433595660487,
      "grad_norm": 0.0642814040184021,
      "learning_rate": 0.0004497231461927748,
      "loss": 1.4896,
      "step": 12434
    },
    {
      "epoch": 0.552932284024721,
      "grad_norm": 0.061715368181467056,
      "learning_rate": 0.00044957770069813325,
      "loss": 1.4897,
      "step": 12436
    },
    {
      "epoch": 0.5530212084833933,
      "grad_norm": 0.06312602758407593,
      "learning_rate": 0.0004494322595138122,
      "loss": 1.4886,
      "step": 12438
    },
    {
      "epoch": 0.5531101329420657,
      "grad_norm": 0.05953274294734001,
      "learning_rate": 0.00044928682265224436,
      "loss": 1.4871,
      "step": 12440
    },
    {
      "epoch": 0.553199057400738,
      "grad_norm": 0.06225527450442314,
      "learning_rate": 0.0004491413901258623,
      "loss": 1.4847,
      "step": 12442
    },
    {
      "epoch": 0.5532879818594104,
      "grad_norm": 0.06316383928060532,
      "learning_rate": 0.00044899596194709847,
      "loss": 1.4923,
      "step": 12444
    },
    {
      "epoch": 0.5533769063180828,
      "grad_norm": 0.06311739236116409,
      "learning_rate": 0.00044885053812838445,
      "loss": 1.4821,
      "step": 12446
    },
    {
      "epoch": 0.5534658307767552,
      "grad_norm": 0.06298642605543137,
      "learning_rate": 0.000448705118682152,
      "loss": 1.4868,
      "step": 12448
    },
    {
      "epoch": 0.5535547552354275,
      "grad_norm": 0.06279413402080536,
      "learning_rate": 0.00044855970362083197,
      "loss": 1.4835,
      "step": 12450
    },
    {
      "epoch": 0.5536436796940999,
      "grad_norm": 0.06067779287695885,
      "learning_rate": 0.0004484142929568552,
      "loss": 1.4835,
      "step": 12452
    },
    {
      "epoch": 0.5537326041527723,
      "grad_norm": 0.06174848973751068,
      "learning_rate": 0.000448268886702652,
      "loss": 1.4821,
      "step": 12454
    },
    {
      "epoch": 0.5538215286114446,
      "grad_norm": 0.06377837806940079,
      "learning_rate": 0.0004481234848706524,
      "loss": 1.4846,
      "step": 12456
    },
    {
      "epoch": 0.5539104530701169,
      "grad_norm": 0.061777420341968536,
      "learning_rate": 0.0004479780874732857,
      "loss": 1.4896,
      "step": 12458
    },
    {
      "epoch": 0.5539993775287892,
      "grad_norm": 0.0632123202085495,
      "learning_rate": 0.0004478326945229816,
      "loss": 1.4873,
      "step": 12460
    },
    {
      "epoch": 0.5540883019874616,
      "grad_norm": 0.06222397834062576,
      "learning_rate": 0.00044768730603216835,
      "loss": 1.4761,
      "step": 12462
    },
    {
      "epoch": 0.554177226446134,
      "grad_norm": 0.06313126534223557,
      "learning_rate": 0.0004475419220132751,
      "loss": 1.4866,
      "step": 12464
    },
    {
      "epoch": 0.5542661509048064,
      "grad_norm": 0.06050386652350426,
      "learning_rate": 0.00044739654247872933,
      "loss": 1.4873,
      "step": 12466
    },
    {
      "epoch": 0.5543550753634787,
      "grad_norm": 0.06345865875482559,
      "learning_rate": 0.0004472511674409589,
      "loss": 1.4869,
      "step": 12468
    },
    {
      "epoch": 0.5544439998221511,
      "grad_norm": 0.06424275785684586,
      "learning_rate": 0.00044710579691239116,
      "loss": 1.4879,
      "step": 12470
    },
    {
      "epoch": 0.5545329242808235,
      "grad_norm": 0.06436572223901749,
      "learning_rate": 0.0004469604309054531,
      "loss": 1.4837,
      "step": 12472
    },
    {
      "epoch": 0.5546218487394958,
      "grad_norm": 0.06335853785276413,
      "learning_rate": 0.00044681506943257095,
      "loss": 1.4916,
      "step": 12474
    },
    {
      "epoch": 0.5547107731981682,
      "grad_norm": 0.06440746039152145,
      "learning_rate": 0.00044666971250617116,
      "loss": 1.4837,
      "step": 12476
    },
    {
      "epoch": 0.5547996976568406,
      "grad_norm": 0.06688931584358215,
      "learning_rate": 0.0004465243601386795,
      "loss": 1.4906,
      "step": 12478
    },
    {
      "epoch": 0.5548886221155128,
      "grad_norm": 0.06155811622738838,
      "learning_rate": 0.0004463790123425209,
      "loss": 1.4918,
      "step": 12480
    },
    {
      "epoch": 0.5549775465741852,
      "grad_norm": 0.06473089009523392,
      "learning_rate": 0.00044623366913012087,
      "loss": 1.4893,
      "step": 12482
    },
    {
      "epoch": 0.5550664710328576,
      "grad_norm": 0.0628729984164238,
      "learning_rate": 0.00044608833051390344,
      "loss": 1.4888,
      "step": 12484
    },
    {
      "epoch": 0.5551553954915299,
      "grad_norm": 0.06283288449048996,
      "learning_rate": 0.0004459429965062935,
      "loss": 1.4896,
      "step": 12486
    },
    {
      "epoch": 0.5552443199502023,
      "grad_norm": 0.061963021755218506,
      "learning_rate": 0.00044579766711971423,
      "loss": 1.4856,
      "step": 12488
    },
    {
      "epoch": 0.5553332444088747,
      "grad_norm": 0.06432387232780457,
      "learning_rate": 0.0004456523423665892,
      "loss": 1.4847,
      "step": 12490
    },
    {
      "epoch": 0.555422168867547,
      "grad_norm": 0.06147773563861847,
      "learning_rate": 0.00044550702225934155,
      "loss": 1.4901,
      "step": 12492
    },
    {
      "epoch": 0.5555110933262194,
      "grad_norm": 0.0623924657702446,
      "learning_rate": 0.0004453617068103937,
      "loss": 1.4929,
      "step": 12494
    },
    {
      "epoch": 0.5556000177848918,
      "grad_norm": 0.06333832442760468,
      "learning_rate": 0.0004452163960321679,
      "loss": 1.4814,
      "step": 12496
    },
    {
      "epoch": 0.5556889422435641,
      "grad_norm": 0.06210675835609436,
      "learning_rate": 0.0004450710899370861,
      "loss": 1.4911,
      "step": 12498
    },
    {
      "epoch": 0.5557778667022365,
      "grad_norm": 0.06140570342540741,
      "learning_rate": 0.0004449257885375695,
      "loss": 1.4867,
      "step": 12500
    },
    {
      "epoch": 0.5557778667022365,
      "eval_loss": 1.47047758102417,
      "eval_runtime": 12.3971,
      "eval_samples_per_second": 557.39,
      "eval_steps_per_second": 69.694,
      "step": 12500
    },
    {
      "epoch": 0.5558667911609088,
      "grad_norm": 0.06514092534780502,
      "learning_rate": 0.00044478049184603925,
      "loss": 1.4915,
      "step": 12502
    },
    {
      "epoch": 0.5559557156195811,
      "grad_norm": 0.06373412162065506,
      "learning_rate": 0.00044463519987491607,
      "loss": 1.4812,
      "step": 12504
    },
    {
      "epoch": 0.5560446400782535,
      "grad_norm": 0.06132040545344353,
      "learning_rate": 0.0004444899126366196,
      "loss": 1.4855,
      "step": 12506
    },
    {
      "epoch": 0.5561335645369259,
      "grad_norm": 0.062154147773981094,
      "learning_rate": 0.0004443446301435704,
      "loss": 1.486,
      "step": 12508
    },
    {
      "epoch": 0.5562224889955982,
      "grad_norm": 0.0628192275762558,
      "learning_rate": 0.0004441993524081873,
      "loss": 1.4891,
      "step": 12510
    },
    {
      "epoch": 0.5563114134542706,
      "grad_norm": 0.06253869086503983,
      "learning_rate": 0.00044405407944288943,
      "loss": 1.4888,
      "step": 12512
    },
    {
      "epoch": 0.556400337912943,
      "grad_norm": 0.06273260712623596,
      "learning_rate": 0.0004439088112600955,
      "loss": 1.4872,
      "step": 12514
    },
    {
      "epoch": 0.5564892623716153,
      "grad_norm": 0.060649313032627106,
      "learning_rate": 0.00044376354787222346,
      "loss": 1.4841,
      "step": 12516
    },
    {
      "epoch": 0.5565781868302877,
      "grad_norm": 0.06059176102280617,
      "learning_rate": 0.00044361828929169125,
      "loss": 1.4883,
      "step": 12518
    },
    {
      "epoch": 0.5566671112889601,
      "grad_norm": 0.06078102067112923,
      "learning_rate": 0.00044347303553091616,
      "loss": 1.4899,
      "step": 12520
    },
    {
      "epoch": 0.5567560357476324,
      "grad_norm": 0.0650026723742485,
      "learning_rate": 0.000443327786602315,
      "loss": 1.4901,
      "step": 12522
    },
    {
      "epoch": 0.5568449602063047,
      "grad_norm": 0.06332740187644958,
      "learning_rate": 0.00044318254251830445,
      "loss": 1.4882,
      "step": 12524
    },
    {
      "epoch": 0.5569338846649771,
      "grad_norm": 0.06303809583187103,
      "learning_rate": 0.0004430373032913006,
      "loss": 1.4893,
      "step": 12526
    },
    {
      "epoch": 0.5570228091236494,
      "grad_norm": 0.061641015112400055,
      "learning_rate": 0.00044289206893371894,
      "loss": 1.4913,
      "step": 12528
    },
    {
      "epoch": 0.5571117335823218,
      "grad_norm": 0.062258537858724594,
      "learning_rate": 0.000442746839457975,
      "loss": 1.4891,
      "step": 12530
    },
    {
      "epoch": 0.5572006580409942,
      "grad_norm": 0.061097778379917145,
      "learning_rate": 0.0004426016148764835,
      "loss": 1.4916,
      "step": 12532
    },
    {
      "epoch": 0.5572895824996665,
      "grad_norm": 0.06280693411827087,
      "learning_rate": 0.0004424563952016589,
      "loss": 1.4904,
      "step": 12534
    },
    {
      "epoch": 0.5573785069583389,
      "grad_norm": 0.061958327889442444,
      "learning_rate": 0.00044231118044591545,
      "loss": 1.4892,
      "step": 12536
    },
    {
      "epoch": 0.5574674314170113,
      "grad_norm": 0.062417879700660706,
      "learning_rate": 0.0004421659706216661,
      "loss": 1.4855,
      "step": 12538
    },
    {
      "epoch": 0.5575563558756836,
      "grad_norm": 0.0647241473197937,
      "learning_rate": 0.00044202076574132484,
      "loss": 1.4862,
      "step": 12540
    },
    {
      "epoch": 0.557645280334356,
      "grad_norm": 0.06341023743152618,
      "learning_rate": 0.0004418755658173039,
      "loss": 1.4923,
      "step": 12542
    },
    {
      "epoch": 0.5577342047930284,
      "grad_norm": 0.062425922602415085,
      "learning_rate": 0.0004417303708620156,
      "loss": 1.4892,
      "step": 12544
    },
    {
      "epoch": 0.5578231292517006,
      "grad_norm": 0.0629575252532959,
      "learning_rate": 0.00044158518088787215,
      "loss": 1.489,
      "step": 12546
    },
    {
      "epoch": 0.557912053710373,
      "grad_norm": 0.060888003557920456,
      "learning_rate": 0.0004414399959072848,
      "loss": 1.4942,
      "step": 12548
    },
    {
      "epoch": 0.5580009781690454,
      "grad_norm": 0.0610843189060688,
      "learning_rate": 0.00044129481593266465,
      "loss": 1.4922,
      "step": 12550
    },
    {
      "epoch": 0.5580899026277177,
      "grad_norm": 0.06121043115854263,
      "learning_rate": 0.00044114964097642233,
      "loss": 1.4876,
      "step": 12552
    },
    {
      "epoch": 0.5581788270863901,
      "grad_norm": 0.062700554728508,
      "learning_rate": 0.00044100447105096803,
      "loss": 1.4922,
      "step": 12554
    },
    {
      "epoch": 0.5582677515450625,
      "grad_norm": 0.06433205306529999,
      "learning_rate": 0.00044085930616871163,
      "loss": 1.4865,
      "step": 12556
    },
    {
      "epoch": 0.5583566760037348,
      "grad_norm": 0.06155223399400711,
      "learning_rate": 0.0004407141463420624,
      "loss": 1.4894,
      "step": 12558
    },
    {
      "epoch": 0.5584456004624072,
      "grad_norm": 0.0634516254067421,
      "learning_rate": 0.00044056899158342893,
      "loss": 1.4841,
      "step": 12560
    },
    {
      "epoch": 0.5585345249210796,
      "grad_norm": 0.06104006990790367,
      "learning_rate": 0.0004404238419052203,
      "loss": 1.4868,
      "step": 12562
    },
    {
      "epoch": 0.558623449379752,
      "grad_norm": 0.06177747994661331,
      "learning_rate": 0.0004402786973198441,
      "loss": 1.4813,
      "step": 12564
    },
    {
      "epoch": 0.5587123738384242,
      "grad_norm": 0.06346866488456726,
      "learning_rate": 0.00044013355783970785,
      "loss": 1.4911,
      "step": 12566
    },
    {
      "epoch": 0.5588012982970966,
      "grad_norm": 0.06526315212249756,
      "learning_rate": 0.0004399884234772191,
      "loss": 1.4937,
      "step": 12568
    },
    {
      "epoch": 0.5588902227557689,
      "grad_norm": 0.0635879635810852,
      "learning_rate": 0.00043984329424478423,
      "loss": 1.4778,
      "step": 12570
    },
    {
      "epoch": 0.5589791472144413,
      "grad_norm": 0.06199366971850395,
      "learning_rate": 0.00043969817015480976,
      "loss": 1.4854,
      "step": 12572
    },
    {
      "epoch": 0.5590680716731137,
      "grad_norm": 0.06245632842183113,
      "learning_rate": 0.0004395530512197015,
      "loss": 1.4814,
      "step": 12574
    },
    {
      "epoch": 0.559156996131786,
      "grad_norm": 0.0628722682595253,
      "learning_rate": 0.0004394079374518647,
      "loss": 1.4936,
      "step": 12576
    },
    {
      "epoch": 0.5592459205904584,
      "grad_norm": 0.06258208304643631,
      "learning_rate": 0.00043926282886370445,
      "loss": 1.485,
      "step": 12578
    },
    {
      "epoch": 0.5593348450491308,
      "grad_norm": 0.06096351519227028,
      "learning_rate": 0.0004391177254676253,
      "loss": 1.4896,
      "step": 12580
    },
    {
      "epoch": 0.5594237695078031,
      "grad_norm": 0.06193548068404198,
      "learning_rate": 0.0004389726272760312,
      "loss": 1.4913,
      "step": 12582
    },
    {
      "epoch": 0.5595126939664755,
      "grad_norm": 0.06140163913369179,
      "learning_rate": 0.0004388275343013259,
      "loss": 1.4898,
      "step": 12584
    },
    {
      "epoch": 0.5596016184251479,
      "grad_norm": 0.06241277977824211,
      "learning_rate": 0.0004386824465559125,
      "loss": 1.4877,
      "step": 12586
    },
    {
      "epoch": 0.5596905428838201,
      "grad_norm": 0.06165548413991928,
      "learning_rate": 0.00043853736405219385,
      "loss": 1.482,
      "step": 12588
    },
    {
      "epoch": 0.5597794673424925,
      "grad_norm": 0.06252114474773407,
      "learning_rate": 0.0004383922868025723,
      "loss": 1.4912,
      "step": 12590
    },
    {
      "epoch": 0.5598683918011649,
      "grad_norm": 0.061745136976242065,
      "learning_rate": 0.00043824721481944924,
      "loss": 1.4839,
      "step": 12592
    },
    {
      "epoch": 0.5599573162598372,
      "grad_norm": 0.06240737810730934,
      "learning_rate": 0.00043810214811522674,
      "loss": 1.4879,
      "step": 12594
    },
    {
      "epoch": 0.5600462407185096,
      "grad_norm": 0.06228557974100113,
      "learning_rate": 0.00043795708670230527,
      "loss": 1.4882,
      "step": 12596
    },
    {
      "epoch": 0.560135165177182,
      "grad_norm": 0.06238147243857384,
      "learning_rate": 0.00043781203059308535,
      "loss": 1.4856,
      "step": 12598
    },
    {
      "epoch": 0.5602240896358543,
      "grad_norm": 0.06147747486829758,
      "learning_rate": 0.00043766697979996715,
      "loss": 1.4861,
      "step": 12600
    },
    {
      "epoch": 0.5603130140945267,
      "grad_norm": 0.0619262270629406,
      "learning_rate": 0.00043752193433535026,
      "loss": 1.4849,
      "step": 12602
    },
    {
      "epoch": 0.5604019385531991,
      "grad_norm": 0.06108873710036278,
      "learning_rate": 0.0004373768942116337,
      "loss": 1.4901,
      "step": 12604
    },
    {
      "epoch": 0.5604908630118715,
      "grad_norm": 0.06347453594207764,
      "learning_rate": 0.0004372318594412162,
      "loss": 1.4785,
      "step": 12606
    },
    {
      "epoch": 0.5605797874705438,
      "grad_norm": 0.060936614871025085,
      "learning_rate": 0.0004370868300364959,
      "loss": 1.488,
      "step": 12608
    },
    {
      "epoch": 0.5606687119292161,
      "grad_norm": 0.060470595955848694,
      "learning_rate": 0.00043694180600987076,
      "loss": 1.4801,
      "step": 12610
    },
    {
      "epoch": 0.5607576363878884,
      "grad_norm": 0.06342191249132156,
      "learning_rate": 0.000436796787373738,
      "loss": 1.4921,
      "step": 12612
    },
    {
      "epoch": 0.5608465608465608,
      "grad_norm": 0.06200144812464714,
      "learning_rate": 0.00043665177414049404,
      "loss": 1.4875,
      "step": 12614
    },
    {
      "epoch": 0.5609354853052332,
      "grad_norm": 0.06348490715026855,
      "learning_rate": 0.000436506766322536,
      "loss": 1.4866,
      "step": 12616
    },
    {
      "epoch": 0.5610244097639056,
      "grad_norm": 0.06057412549853325,
      "learning_rate": 0.0004363617639322592,
      "loss": 1.4826,
      "step": 12618
    },
    {
      "epoch": 0.5611133342225779,
      "grad_norm": 0.059886761009693146,
      "learning_rate": 0.0004362167669820592,
      "loss": 1.4859,
      "step": 12620
    },
    {
      "epoch": 0.5612022586812503,
      "grad_norm": 0.06247553601861,
      "learning_rate": 0.00043607177548433105,
      "loss": 1.4826,
      "step": 12622
    },
    {
      "epoch": 0.5612911831399227,
      "grad_norm": 0.06425140798091888,
      "learning_rate": 0.0004359267894514693,
      "loss": 1.4877,
      "step": 12624
    },
    {
      "epoch": 0.561380107598595,
      "grad_norm": 0.06306110322475433,
      "learning_rate": 0.0004357818088958679,
      "loss": 1.4864,
      "step": 12626
    },
    {
      "epoch": 0.5614690320572674,
      "grad_norm": 0.06385751813650131,
      "learning_rate": 0.00043563683382992057,
      "loss": 1.494,
      "step": 12628
    },
    {
      "epoch": 0.5615579565159398,
      "grad_norm": 0.0631057620048523,
      "learning_rate": 0.0004354918642660203,
      "loss": 1.4885,
      "step": 12630
    },
    {
      "epoch": 0.561646880974612,
      "grad_norm": 0.06637078523635864,
      "learning_rate": 0.0004353469002165597,
      "loss": 1.4927,
      "step": 12632
    },
    {
      "epoch": 0.5617358054332844,
      "grad_norm": 0.06537287682294846,
      "learning_rate": 0.0004352019416939311,
      "loss": 1.4901,
      "step": 12634
    },
    {
      "epoch": 0.5618247298919568,
      "grad_norm": 0.06411311775445938,
      "learning_rate": 0.00043505698871052585,
      "loss": 1.4912,
      "step": 12636
    },
    {
      "epoch": 0.5619136543506291,
      "grad_norm": 0.06516771763563156,
      "learning_rate": 0.00043491204127873556,
      "loss": 1.481,
      "step": 12638
    },
    {
      "epoch": 0.5620025788093015,
      "grad_norm": 0.06372273713350296,
      "learning_rate": 0.00043476709941095074,
      "loss": 1.4882,
      "step": 12640
    },
    {
      "epoch": 0.5620915032679739,
      "grad_norm": 0.06318344920873642,
      "learning_rate": 0.0004346221631195618,
      "loss": 1.4841,
      "step": 12642
    },
    {
      "epoch": 0.5621804277266462,
      "grad_norm": 0.06348215788602829,
      "learning_rate": 0.0004344772324169586,
      "loss": 1.4878,
      "step": 12644
    },
    {
      "epoch": 0.5622693521853186,
      "grad_norm": 0.0606432780623436,
      "learning_rate": 0.00043433230731553,
      "loss": 1.494,
      "step": 12646
    },
    {
      "epoch": 0.562358276643991,
      "grad_norm": 0.06193801760673523,
      "learning_rate": 0.00043418738782766554,
      "loss": 1.4876,
      "step": 12648
    },
    {
      "epoch": 0.5624472011026633,
      "grad_norm": 0.062378112226724625,
      "learning_rate": 0.00043404247396575303,
      "loss": 1.4922,
      "step": 12650
    },
    {
      "epoch": 0.5625361255613357,
      "grad_norm": 0.06375192105770111,
      "learning_rate": 0.00043389756574218035,
      "loss": 1.4903,
      "step": 12652
    },
    {
      "epoch": 0.562625050020008,
      "grad_norm": 0.06285295635461807,
      "learning_rate": 0.0004337526631693353,
      "loss": 1.4871,
      "step": 12654
    },
    {
      "epoch": 0.5627139744786803,
      "grad_norm": 0.0635739341378212,
      "learning_rate": 0.0004336077662596043,
      "loss": 1.4837,
      "step": 12656
    },
    {
      "epoch": 0.5628028989373527,
      "grad_norm": 0.06320179253816605,
      "learning_rate": 0.00043346287502537417,
      "loss": 1.4951,
      "step": 12658
    },
    {
      "epoch": 0.5628918233960251,
      "grad_norm": 0.06251043826341629,
      "learning_rate": 0.0004333179894790307,
      "loss": 1.4853,
      "step": 12660
    },
    {
      "epoch": 0.5629807478546974,
      "grad_norm": 0.06125814467668533,
      "learning_rate": 0.00043317310963295925,
      "loss": 1.4879,
      "step": 12662
    },
    {
      "epoch": 0.5630696723133698,
      "grad_norm": 0.06437008082866669,
      "learning_rate": 0.000433028235499545,
      "loss": 1.4902,
      "step": 12664
    },
    {
      "epoch": 0.5631585967720422,
      "grad_norm": 0.0625307485461235,
      "learning_rate": 0.0004328833670911724,
      "loss": 1.4882,
      "step": 12666
    },
    {
      "epoch": 0.5632475212307145,
      "grad_norm": 0.06289555132389069,
      "learning_rate": 0.00043273850442022505,
      "loss": 1.4853,
      "step": 12668
    },
    {
      "epoch": 0.5633364456893869,
      "grad_norm": 0.06292538344860077,
      "learning_rate": 0.00043259364749908706,
      "loss": 1.4879,
      "step": 12670
    },
    {
      "epoch": 0.5634253701480593,
      "grad_norm": 0.06323841959238052,
      "learning_rate": 0.00043244879634014106,
      "loss": 1.4838,
      "step": 12672
    },
    {
      "epoch": 0.5635142946067316,
      "grad_norm": 0.06537654995918274,
      "learning_rate": 0.0004323039509557695,
      "loss": 1.4864,
      "step": 12674
    },
    {
      "epoch": 0.5636032190654039,
      "grad_norm": 0.06365048885345459,
      "learning_rate": 0.00043215911135835466,
      "loss": 1.4812,
      "step": 12676
    },
    {
      "epoch": 0.5636921435240763,
      "grad_norm": 0.06092437356710434,
      "learning_rate": 0.0004320142775602778,
      "loss": 1.4825,
      "step": 12678
    },
    {
      "epoch": 0.5637810679827486,
      "grad_norm": 0.06534268707036972,
      "learning_rate": 0.0004318694495739203,
      "loss": 1.4839,
      "step": 12680
    },
    {
      "epoch": 0.563869992441421,
      "grad_norm": 0.06497418135404587,
      "learning_rate": 0.0004317246274116624,
      "loss": 1.486,
      "step": 12682
    },
    {
      "epoch": 0.5639589169000934,
      "grad_norm": 0.06123911216855049,
      "learning_rate": 0.0004315798110858842,
      "loss": 1.486,
      "step": 12684
    },
    {
      "epoch": 0.5640478413587657,
      "grad_norm": 0.062008924782276154,
      "learning_rate": 0.0004314350006089653,
      "loss": 1.481,
      "step": 12686
    },
    {
      "epoch": 0.5641367658174381,
      "grad_norm": 0.06340475380420685,
      "learning_rate": 0.0004312901959932847,
      "loss": 1.4898,
      "step": 12688
    },
    {
      "epoch": 0.5642256902761105,
      "grad_norm": 0.062276601791381836,
      "learning_rate": 0.00043114539725122087,
      "loss": 1.4852,
      "step": 12690
    },
    {
      "epoch": 0.5643146147347828,
      "grad_norm": 0.06282736361026764,
      "learning_rate": 0.00043100060439515205,
      "loss": 1.4893,
      "step": 12692
    },
    {
      "epoch": 0.5644035391934552,
      "grad_norm": 0.06258688867092133,
      "learning_rate": 0.00043085581743745546,
      "loss": 1.4856,
      "step": 12694
    },
    {
      "epoch": 0.5644924636521275,
      "grad_norm": 0.0631018653512001,
      "learning_rate": 0.00043071103639050846,
      "loss": 1.4859,
      "step": 12696
    },
    {
      "epoch": 0.5645813881107998,
      "grad_norm": 0.06103930622339249,
      "learning_rate": 0.0004305662612666875,
      "loss": 1.4892,
      "step": 12698
    },
    {
      "epoch": 0.5646703125694722,
      "grad_norm": 0.06183280423283577,
      "learning_rate": 0.00043042149207836824,
      "loss": 1.4848,
      "step": 12700
    },
    {
      "epoch": 0.5647592370281446,
      "grad_norm": 0.06280186027288437,
      "learning_rate": 0.0004302767288379267,
      "loss": 1.4846,
      "step": 12702
    },
    {
      "epoch": 0.5648481614868169,
      "grad_norm": 0.061180781573057175,
      "learning_rate": 0.0004301319715577376,
      "loss": 1.4802,
      "step": 12704
    },
    {
      "epoch": 0.5649370859454893,
      "grad_norm": 0.06304103881120682,
      "learning_rate": 0.00042998722025017536,
      "loss": 1.4853,
      "step": 12706
    },
    {
      "epoch": 0.5650260104041617,
      "grad_norm": 0.06135190278291702,
      "learning_rate": 0.0004298424749276141,
      "loss": 1.489,
      "step": 12708
    },
    {
      "epoch": 0.565114934862834,
      "grad_norm": 0.061154115945100784,
      "learning_rate": 0.0004296977356024272,
      "loss": 1.487,
      "step": 12710
    },
    {
      "epoch": 0.5652038593215064,
      "grad_norm": 0.06199873611330986,
      "learning_rate": 0.0004295530022869878,
      "loss": 1.4856,
      "step": 12712
    },
    {
      "epoch": 0.5652927837801788,
      "grad_norm": 0.06099383533000946,
      "learning_rate": 0.0004294082749936682,
      "loss": 1.4791,
      "step": 12714
    },
    {
      "epoch": 0.5653817082388511,
      "grad_norm": 0.061755064874887466,
      "learning_rate": 0.0004292635537348402,
      "loss": 1.4841,
      "step": 12716
    },
    {
      "epoch": 0.5654706326975234,
      "grad_norm": 0.0626377984881401,
      "learning_rate": 0.00042911883852287546,
      "loss": 1.4911,
      "step": 12718
    },
    {
      "epoch": 0.5655595571561958,
      "grad_norm": 0.06231605261564255,
      "learning_rate": 0.00042897412937014496,
      "loss": 1.4779,
      "step": 12720
    },
    {
      "epoch": 0.5656484816148681,
      "grad_norm": 0.06421330571174622,
      "learning_rate": 0.0004288294262890185,
      "loss": 1.4914,
      "step": 12722
    },
    {
      "epoch": 0.5657374060735405,
      "grad_norm": 0.06046907603740692,
      "learning_rate": 0.0004286847292918666,
      "loss": 1.4845,
      "step": 12724
    },
    {
      "epoch": 0.5658263305322129,
      "grad_norm": 0.062285587191581726,
      "learning_rate": 0.00042854003839105806,
      "loss": 1.4837,
      "step": 12726
    },
    {
      "epoch": 0.5659152549908852,
      "grad_norm": 0.0627601146697998,
      "learning_rate": 0.00042839535359896227,
      "loss": 1.4861,
      "step": 12728
    },
    {
      "epoch": 0.5660041794495576,
      "grad_norm": 0.06236814707517624,
      "learning_rate": 0.00042825067492794706,
      "loss": 1.4876,
      "step": 12730
    },
    {
      "epoch": 0.56609310390823,
      "grad_norm": 0.06265702098608017,
      "learning_rate": 0.0004281060023903803,
      "loss": 1.4797,
      "step": 12732
    },
    {
      "epoch": 0.5661820283669023,
      "grad_norm": 0.06238122656941414,
      "learning_rate": 0.0004279613359986293,
      "loss": 1.4885,
      "step": 12734
    },
    {
      "epoch": 0.5662709528255747,
      "grad_norm": 0.06290338933467865,
      "learning_rate": 0.00042781667576506085,
      "loss": 1.4882,
      "step": 12736
    },
    {
      "epoch": 0.5663598772842471,
      "grad_norm": 0.06228485703468323,
      "learning_rate": 0.0004276720217020409,
      "loss": 1.4914,
      "step": 12738
    },
    {
      "epoch": 0.5664488017429193,
      "grad_norm": 0.06285703182220459,
      "learning_rate": 0.0004275273738219354,
      "loss": 1.4851,
      "step": 12740
    },
    {
      "epoch": 0.5665377262015917,
      "grad_norm": 0.06270929425954819,
      "learning_rate": 0.00042738273213710937,
      "loss": 1.4811,
      "step": 12742
    },
    {
      "epoch": 0.5666266506602641,
      "grad_norm": 0.06521732360124588,
      "learning_rate": 0.00042723809665992733,
      "loss": 1.4876,
      "step": 12744
    },
    {
      "epoch": 0.5667155751189364,
      "grad_norm": 0.06187422201037407,
      "learning_rate": 0.00042709346740275346,
      "loss": 1.4874,
      "step": 12746
    },
    {
      "epoch": 0.5668044995776088,
      "grad_norm": 0.06465815752744675,
      "learning_rate": 0.0004269488443779512,
      "loss": 1.4846,
      "step": 12748
    },
    {
      "epoch": 0.5668934240362812,
      "grad_norm": 0.0634174719452858,
      "learning_rate": 0.00042680422759788377,
      "loss": 1.4852,
      "step": 12750
    },
    {
      "epoch": 0.5669823484949535,
      "grad_norm": 0.0651431754231453,
      "learning_rate": 0.0004266596170749137,
      "loss": 1.4861,
      "step": 12752
    },
    {
      "epoch": 0.5670712729536259,
      "grad_norm": 0.06035947799682617,
      "learning_rate": 0.0004265150128214024,
      "loss": 1.4858,
      "step": 12754
    },
    {
      "epoch": 0.5671601974122983,
      "grad_norm": 0.06423404812812805,
      "learning_rate": 0.0004263704148497119,
      "loss": 1.4907,
      "step": 12756
    },
    {
      "epoch": 0.5672491218709707,
      "grad_norm": 0.06168157607316971,
      "learning_rate": 0.0004262258231722028,
      "loss": 1.4827,
      "step": 12758
    },
    {
      "epoch": 0.567338046329643,
      "grad_norm": 0.06296733021736145,
      "learning_rate": 0.0004260812378012353,
      "loss": 1.4828,
      "step": 12760
    },
    {
      "epoch": 0.5674269707883153,
      "grad_norm": 0.06025325134396553,
      "learning_rate": 0.00042593665874916944,
      "loss": 1.4839,
      "step": 12762
    },
    {
      "epoch": 0.5675158952469876,
      "grad_norm": 0.062312737107276917,
      "learning_rate": 0.0004257920860283642,
      "loss": 1.4904,
      "step": 12764
    },
    {
      "epoch": 0.56760481970566,
      "grad_norm": 0.06261510401964188,
      "learning_rate": 0.0004256475196511787,
      "loss": 1.4859,
      "step": 12766
    },
    {
      "epoch": 0.5676937441643324,
      "grad_norm": 0.06052310764789581,
      "learning_rate": 0.00042550295962997083,
      "loss": 1.4845,
      "step": 12768
    },
    {
      "epoch": 0.5677826686230047,
      "grad_norm": 0.061324592679739,
      "learning_rate": 0.00042535840597709816,
      "loss": 1.4843,
      "step": 12770
    },
    {
      "epoch": 0.5678715930816771,
      "grad_norm": 0.06264898926019669,
      "learning_rate": 0.000425213858704918,
      "loss": 1.4873,
      "step": 12772
    },
    {
      "epoch": 0.5679605175403495,
      "grad_norm": 0.06049567088484764,
      "learning_rate": 0.0004250693178257868,
      "loss": 1.4887,
      "step": 12774
    },
    {
      "epoch": 0.5680494419990219,
      "grad_norm": 0.06124699115753174,
      "learning_rate": 0.00042492478335206033,
      "loss": 1.486,
      "step": 12776
    },
    {
      "epoch": 0.5681383664576942,
      "grad_norm": 0.06057760491967201,
      "learning_rate": 0.00042478025529609447,
      "loss": 1.4847,
      "step": 12778
    },
    {
      "epoch": 0.5682272909163666,
      "grad_norm": 0.0630040317773819,
      "learning_rate": 0.0004246357336702436,
      "loss": 1.4834,
      "step": 12780
    },
    {
      "epoch": 0.568316215375039,
      "grad_norm": 0.061218783259391785,
      "learning_rate": 0.0004244912184868626,
      "loss": 1.4846,
      "step": 12782
    },
    {
      "epoch": 0.5684051398337112,
      "grad_norm": 0.0640929564833641,
      "learning_rate": 0.0004243467097583049,
      "loss": 1.4862,
      "step": 12784
    },
    {
      "epoch": 0.5684940642923836,
      "grad_norm": 0.061070337891578674,
      "learning_rate": 0.00042420220749692377,
      "loss": 1.4816,
      "step": 12786
    },
    {
      "epoch": 0.568582988751056,
      "grad_norm": 0.06318749487400055,
      "learning_rate": 0.00042405771171507206,
      "loss": 1.4888,
      "step": 12788
    },
    {
      "epoch": 0.5686719132097283,
      "grad_norm": 0.06423699110746384,
      "learning_rate": 0.0004239132224251018,
      "loss": 1.4879,
      "step": 12790
    },
    {
      "epoch": 0.5687608376684007,
      "grad_norm": 0.06260624527931213,
      "learning_rate": 0.0004237687396393645,
      "loss": 1.491,
      "step": 12792
    },
    {
      "epoch": 0.568849762127073,
      "grad_norm": 0.062127046287059784,
      "learning_rate": 0.00042362426337021143,
      "loss": 1.483,
      "step": 12794
    },
    {
      "epoch": 0.5689386865857454,
      "grad_norm": 0.06342507153749466,
      "learning_rate": 0.00042347979362999293,
      "loss": 1.486,
      "step": 12796
    },
    {
      "epoch": 0.5690276110444178,
      "grad_norm": 0.06286569684743881,
      "learning_rate": 0.00042333533043105865,
      "loss": 1.4895,
      "step": 12798
    },
    {
      "epoch": 0.5691165355030902,
      "grad_norm": 0.06699131429195404,
      "learning_rate": 0.0004231908737857584,
      "loss": 1.4929,
      "step": 12800
    },
    {
      "epoch": 0.5692054599617625,
      "grad_norm": 0.06216110289096832,
      "learning_rate": 0.00042304642370644054,
      "loss": 1.4828,
      "step": 12802
    },
    {
      "epoch": 0.5692943844204349,
      "grad_norm": 0.06298429518938065,
      "learning_rate": 0.00042290198020545375,
      "loss": 1.482,
      "step": 12804
    },
    {
      "epoch": 0.5693833088791072,
      "grad_norm": 0.06303789466619492,
      "learning_rate": 0.0004227575432951453,
      "loss": 1.4873,
      "step": 12806
    },
    {
      "epoch": 0.5694722333377795,
      "grad_norm": 0.06614295393228531,
      "learning_rate": 0.0004226131129878624,
      "loss": 1.4907,
      "step": 12808
    },
    {
      "epoch": 0.5695611577964519,
      "grad_norm": 0.06474742293357849,
      "learning_rate": 0.0004224686892959517,
      "loss": 1.4831,
      "step": 12810
    },
    {
      "epoch": 0.5696500822551243,
      "grad_norm": 0.06322574615478516,
      "learning_rate": 0.0004223242722317592,
      "loss": 1.4893,
      "step": 12812
    },
    {
      "epoch": 0.5697390067137966,
      "grad_norm": 0.060625430196523666,
      "learning_rate": 0.00042217986180763,
      "loss": 1.477,
      "step": 12814
    },
    {
      "epoch": 0.569827931172469,
      "grad_norm": 0.06206146627664566,
      "learning_rate": 0.00042203545803590935,
      "loss": 1.4866,
      "step": 12816
    },
    {
      "epoch": 0.5699168556311414,
      "grad_norm": 0.06130099669098854,
      "learning_rate": 0.0004218910609289412,
      "loss": 1.485,
      "step": 12818
    },
    {
      "epoch": 0.5700057800898137,
      "grad_norm": 0.06254599988460541,
      "learning_rate": 0.00042174667049906943,
      "loss": 1.4766,
      "step": 12820
    },
    {
      "epoch": 0.5700947045484861,
      "grad_norm": 0.06300479918718338,
      "learning_rate": 0.00042160228675863716,
      "loss": 1.4921,
      "step": 12822
    },
    {
      "epoch": 0.5701836290071585,
      "grad_norm": 0.06208550184965134,
      "learning_rate": 0.0004214579097199868,
      "loss": 1.4773,
      "step": 12824
    },
    {
      "epoch": 0.5702725534658307,
      "grad_norm": 0.06384430080652237,
      "learning_rate": 0.0004213135393954605,
      "loss": 1.4851,
      "step": 12826
    },
    {
      "epoch": 0.5703614779245031,
      "grad_norm": 0.06255348771810532,
      "learning_rate": 0.0004211691757973998,
      "loss": 1.4875,
      "step": 12828
    },
    {
      "epoch": 0.5704504023831755,
      "grad_norm": 0.06358494609594345,
      "learning_rate": 0.000421024818938145,
      "loss": 1.4814,
      "step": 12830
    },
    {
      "epoch": 0.5705393268418478,
      "grad_norm": 0.0655054897069931,
      "learning_rate": 0.000420880468830037,
      "loss": 1.4862,
      "step": 12832
    },
    {
      "epoch": 0.5706282513005202,
      "grad_norm": 0.06272534281015396,
      "learning_rate": 0.0004207361254854149,
      "loss": 1.4882,
      "step": 12834
    },
    {
      "epoch": 0.5707171757591926,
      "grad_norm": 0.06250984966754913,
      "learning_rate": 0.00042059178891661845,
      "loss": 1.4822,
      "step": 12836
    },
    {
      "epoch": 0.5708061002178649,
      "grad_norm": 0.06336545944213867,
      "learning_rate": 0.0004204474591359856,
      "loss": 1.4828,
      "step": 12838
    },
    {
      "epoch": 0.5708950246765373,
      "grad_norm": 0.06069128215312958,
      "learning_rate": 0.00042030313615585446,
      "loss": 1.4854,
      "step": 12840
    },
    {
      "epoch": 0.5709839491352097,
      "grad_norm": 0.06559595465660095,
      "learning_rate": 0.00042015881998856263,
      "loss": 1.4875,
      "step": 12842
    },
    {
      "epoch": 0.571072873593882,
      "grad_norm": 0.06050030142068863,
      "learning_rate": 0.0004200145106464467,
      "loss": 1.4861,
      "step": 12844
    },
    {
      "epoch": 0.5711617980525544,
      "grad_norm": 0.06325476616621017,
      "learning_rate": 0.0004198702081418427,
      "loss": 1.4888,
      "step": 12846
    },
    {
      "epoch": 0.5712507225112267,
      "grad_norm": 0.06051871180534363,
      "learning_rate": 0.00041972591248708657,
      "loss": 1.483,
      "step": 12848
    },
    {
      "epoch": 0.571339646969899,
      "grad_norm": 0.06114453822374344,
      "learning_rate": 0.0004195816236945131,
      "loss": 1.4873,
      "step": 12850
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.06214149668812752,
      "learning_rate": 0.000419437341776457,
      "loss": 1.4859,
      "step": 12852
    },
    {
      "epoch": 0.5715174958872438,
      "grad_norm": 0.060995914041996,
      "learning_rate": 0.000419293066745252,
      "loss": 1.4919,
      "step": 12854
    },
    {
      "epoch": 0.5716064203459161,
      "grad_norm": 0.06173815205693245,
      "learning_rate": 0.000419148798613231,
      "loss": 1.4881,
      "step": 12856
    },
    {
      "epoch": 0.5716953448045885,
      "grad_norm": 0.060767561197280884,
      "learning_rate": 0.0004190045373927273,
      "loss": 1.488,
      "step": 12858
    },
    {
      "epoch": 0.5717842692632609,
      "grad_norm": 0.061665672808885574,
      "learning_rate": 0.0004188602830960726,
      "loss": 1.4862,
      "step": 12860
    },
    {
      "epoch": 0.5718731937219332,
      "grad_norm": 0.06136852502822876,
      "learning_rate": 0.00041871603573559837,
      "loss": 1.4881,
      "step": 12862
    },
    {
      "epoch": 0.5719621181806056,
      "grad_norm": 0.0626179501414299,
      "learning_rate": 0.0004185717953236357,
      "loss": 1.4832,
      "step": 12864
    },
    {
      "epoch": 0.572051042639278,
      "grad_norm": 0.06289484351873398,
      "learning_rate": 0.0004184275618725148,
      "loss": 1.4861,
      "step": 12866
    },
    {
      "epoch": 0.5721399670979503,
      "grad_norm": 0.06334191560745239,
      "learning_rate": 0.0004182833353945653,
      "loss": 1.4887,
      "step": 12868
    },
    {
      "epoch": 0.5722288915566226,
      "grad_norm": 0.061536043882369995,
      "learning_rate": 0.0004181391159021165,
      "loss": 1.4828,
      "step": 12870
    },
    {
      "epoch": 0.572317816015295,
      "grad_norm": 0.06327944248914719,
      "learning_rate": 0.0004179949034074967,
      "loss": 1.4843,
      "step": 12872
    },
    {
      "epoch": 0.5724067404739673,
      "grad_norm": 0.061356909573078156,
      "learning_rate": 0.0004178506979230341,
      "loss": 1.4833,
      "step": 12874
    },
    {
      "epoch": 0.5724956649326397,
      "grad_norm": 0.0625733733177185,
      "learning_rate": 0.00041770649946105585,
      "loss": 1.4834,
      "step": 12876
    },
    {
      "epoch": 0.5725845893913121,
      "grad_norm": 0.06169121339917183,
      "learning_rate": 0.0004175623080338885,
      "loss": 1.48,
      "step": 12878
    },
    {
      "epoch": 0.5726735138499844,
      "grad_norm": 0.06390050053596497,
      "learning_rate": 0.0004174181236538585,
      "loss": 1.4884,
      "step": 12880
    },
    {
      "epoch": 0.5727624383086568,
      "grad_norm": 0.0622803270816803,
      "learning_rate": 0.00041727394633329137,
      "loss": 1.4838,
      "step": 12882
    },
    {
      "epoch": 0.5728513627673292,
      "grad_norm": 0.062278926372528076,
      "learning_rate": 0.0004171297760845115,
      "loss": 1.4837,
      "step": 12884
    },
    {
      "epoch": 0.5729402872260015,
      "grad_norm": 0.06416445970535278,
      "learning_rate": 0.00041698561291984387,
      "loss": 1.485,
      "step": 12886
    },
    {
      "epoch": 0.5730292116846739,
      "grad_norm": 0.06131880730390549,
      "learning_rate": 0.0004168414568516116,
      "loss": 1.4871,
      "step": 12888
    },
    {
      "epoch": 0.5731181361433463,
      "grad_norm": 0.06132744997739792,
      "learning_rate": 0.0004166973078921384,
      "loss": 1.4895,
      "step": 12890
    },
    {
      "epoch": 0.5732070606020185,
      "grad_norm": 0.06064356490969658,
      "learning_rate": 0.0004165531660537463,
      "loss": 1.4859,
      "step": 12892
    },
    {
      "epoch": 0.5732959850606909,
      "grad_norm": 0.060645733028650284,
      "learning_rate": 0.00041640903134875716,
      "loss": 1.4801,
      "step": 12894
    },
    {
      "epoch": 0.5733849095193633,
      "grad_norm": 0.06099729612469673,
      "learning_rate": 0.00041626490378949253,
      "loss": 1.4751,
      "step": 12896
    },
    {
      "epoch": 0.5734738339780356,
      "grad_norm": 0.06112310662865639,
      "learning_rate": 0.00041612078338827295,
      "loss": 1.4874,
      "step": 12898
    },
    {
      "epoch": 0.573562758436708,
      "grad_norm": 0.06212225928902626,
      "learning_rate": 0.00041597667015741836,
      "loss": 1.4863,
      "step": 12900
    },
    {
      "epoch": 0.5736516828953804,
      "grad_norm": 0.06187918782234192,
      "learning_rate": 0.0004158325641092483,
      "loss": 1.4783,
      "step": 12902
    },
    {
      "epoch": 0.5737406073540527,
      "grad_norm": 0.06254956126213074,
      "learning_rate": 0.00041568846525608154,
      "loss": 1.4861,
      "step": 12904
    },
    {
      "epoch": 0.5738295318127251,
      "grad_norm": 0.06175340339541435,
      "learning_rate": 0.00041554437361023645,
      "loss": 1.4867,
      "step": 12906
    },
    {
      "epoch": 0.5739184562713975,
      "grad_norm": 0.06396617740392685,
      "learning_rate": 0.0004154002891840306,
      "loss": 1.4828,
      "step": 12908
    },
    {
      "epoch": 0.5740073807300698,
      "grad_norm": 0.0628194510936737,
      "learning_rate": 0.0004152562119897804,
      "loss": 1.4834,
      "step": 12910
    },
    {
      "epoch": 0.5740963051887422,
      "grad_norm": 0.06112290546298027,
      "learning_rate": 0.0004151121420398031,
      "loss": 1.4838,
      "step": 12912
    },
    {
      "epoch": 0.5741852296474145,
      "grad_norm": 0.06214891001582146,
      "learning_rate": 0.00041496807934641376,
      "loss": 1.4872,
      "step": 12914
    },
    {
      "epoch": 0.5742741541060868,
      "grad_norm": 0.06411154568195343,
      "learning_rate": 0.00041482402392192775,
      "loss": 1.4844,
      "step": 12916
    },
    {
      "epoch": 0.5743630785647592,
      "grad_norm": 0.06268089264631271,
      "learning_rate": 0.0004146799757786596,
      "loss": 1.4916,
      "step": 12918
    },
    {
      "epoch": 0.5744520030234316,
      "grad_norm": 0.06198188662528992,
      "learning_rate": 0.00041453593492892307,
      "loss": 1.4868,
      "step": 12920
    },
    {
      "epoch": 0.574540927482104,
      "grad_norm": 0.06695974618196487,
      "learning_rate": 0.00041439190138503135,
      "loss": 1.4801,
      "step": 12922
    },
    {
      "epoch": 0.5746298519407763,
      "grad_norm": 0.06644386053085327,
      "learning_rate": 0.0004142478751592973,
      "loss": 1.494,
      "step": 12924
    },
    {
      "epoch": 0.5747187763994487,
      "grad_norm": 0.06115036457777023,
      "learning_rate": 0.0004141038562640327,
      "loss": 1.4766,
      "step": 12926
    },
    {
      "epoch": 0.574807700858121,
      "grad_norm": 0.062477122992277145,
      "learning_rate": 0.000413959844711549,
      "loss": 1.4887,
      "step": 12928
    },
    {
      "epoch": 0.5748966253167934,
      "grad_norm": 0.06162472814321518,
      "learning_rate": 0.0004138158405141571,
      "loss": 1.4835,
      "step": 12930
    },
    {
      "epoch": 0.5749855497754658,
      "grad_norm": 0.06390776485204697,
      "learning_rate": 0.0004136718436841667,
      "loss": 1.4821,
      "step": 12932
    },
    {
      "epoch": 0.575074474234138,
      "grad_norm": 0.06126069650053978,
      "learning_rate": 0.00041352785423388774,
      "loss": 1.4833,
      "step": 12934
    },
    {
      "epoch": 0.5751633986928104,
      "grad_norm": 0.06258382648229599,
      "learning_rate": 0.0004133838721756289,
      "loss": 1.4808,
      "step": 12936
    },
    {
      "epoch": 0.5752523231514828,
      "grad_norm": 0.061541568487882614,
      "learning_rate": 0.0004132398975216981,
      "loss": 1.4852,
      "step": 12938
    },
    {
      "epoch": 0.5753412476101551,
      "grad_norm": 0.06053704768419266,
      "learning_rate": 0.00041309593028440354,
      "loss": 1.4843,
      "step": 12940
    },
    {
      "epoch": 0.5754301720688275,
      "grad_norm": 0.060443904250860214,
      "learning_rate": 0.0004129519704760514,
      "loss": 1.484,
      "step": 12942
    },
    {
      "epoch": 0.5755190965274999,
      "grad_norm": 0.06167231872677803,
      "learning_rate": 0.0004128080181089489,
      "loss": 1.4891,
      "step": 12944
    },
    {
      "epoch": 0.5756080209861723,
      "grad_norm": 0.06110317260026932,
      "learning_rate": 0.00041266407319540105,
      "loss": 1.4837,
      "step": 12946
    },
    {
      "epoch": 0.5756969454448446,
      "grad_norm": 0.06018367037177086,
      "learning_rate": 0.000412520135747713,
      "loss": 1.4861,
      "step": 12948
    },
    {
      "epoch": 0.575785869903517,
      "grad_norm": 0.06103086471557617,
      "learning_rate": 0.0004123762057781894,
      "loss": 1.4848,
      "step": 12950
    },
    {
      "epoch": 0.5758747943621894,
      "grad_norm": 0.06104036420583725,
      "learning_rate": 0.0004122322832991339,
      "loss": 1.4899,
      "step": 12952
    },
    {
      "epoch": 0.5759637188208617,
      "grad_norm": 0.061243943870067596,
      "learning_rate": 0.0004120883683228494,
      "loss": 1.4895,
      "step": 12954
    },
    {
      "epoch": 0.576052643279534,
      "grad_norm": 0.062343720346689224,
      "learning_rate": 0.0004119444608616388,
      "loss": 1.4847,
      "step": 12956
    },
    {
      "epoch": 0.5761415677382064,
      "grad_norm": 0.061404332518577576,
      "learning_rate": 0.00041180056092780354,
      "loss": 1.4812,
      "step": 12958
    },
    {
      "epoch": 0.5762304921968787,
      "grad_norm": 0.06235596537590027,
      "learning_rate": 0.0004116566685336452,
      "loss": 1.4852,
      "step": 12960
    },
    {
      "epoch": 0.5763194166555511,
      "grad_norm": 0.061843257397413254,
      "learning_rate": 0.0004115127836914642,
      "loss": 1.49,
      "step": 12962
    },
    {
      "epoch": 0.5764083411142235,
      "grad_norm": 0.05963189899921417,
      "learning_rate": 0.00041136890641356005,
      "loss": 1.487,
      "step": 12964
    },
    {
      "epoch": 0.5764972655728958,
      "grad_norm": 0.06309180706739426,
      "learning_rate": 0.0004112250367122328,
      "loss": 1.4832,
      "step": 12966
    },
    {
      "epoch": 0.5765861900315682,
      "grad_norm": 0.05972014740109444,
      "learning_rate": 0.0004110811745997804,
      "loss": 1.4868,
      "step": 12968
    },
    {
      "epoch": 0.5766751144902406,
      "grad_norm": 0.06337624043226242,
      "learning_rate": 0.00041093732008850106,
      "loss": 1.4848,
      "step": 12970
    },
    {
      "epoch": 0.5767640389489129,
      "grad_norm": 0.06255938112735748,
      "learning_rate": 0.0004107934731906921,
      "loss": 1.4831,
      "step": 12972
    },
    {
      "epoch": 0.5768529634075853,
      "grad_norm": 0.06092438846826553,
      "learning_rate": 0.0004106496339186501,
      "loss": 1.4789,
      "step": 12974
    },
    {
      "epoch": 0.5769418878662577,
      "grad_norm": 0.062436189502477646,
      "learning_rate": 0.0004105058022846713,
      "loss": 1.4888,
      "step": 12976
    },
    {
      "epoch": 0.5770308123249299,
      "grad_norm": 0.0622539147734642,
      "learning_rate": 0.00041036197830105083,
      "loss": 1.4801,
      "step": 12978
    },
    {
      "epoch": 0.5771197367836023,
      "grad_norm": 0.06234341859817505,
      "learning_rate": 0.0004102181619800834,
      "loss": 1.4812,
      "step": 12980
    },
    {
      "epoch": 0.5772086612422747,
      "grad_norm": 0.06356412917375565,
      "learning_rate": 0.00041007435333406325,
      "loss": 1.4836,
      "step": 12982
    },
    {
      "epoch": 0.577297585700947,
      "grad_norm": 0.06135694682598114,
      "learning_rate": 0.00040993055237528367,
      "loss": 1.4868,
      "step": 12984
    },
    {
      "epoch": 0.5773865101596194,
      "grad_norm": 0.06356058269739151,
      "learning_rate": 0.00040978675911603727,
      "loss": 1.4841,
      "step": 12986
    },
    {
      "epoch": 0.5774754346182918,
      "grad_norm": 0.060618288815021515,
      "learning_rate": 0.00040964297356861634,
      "loss": 1.4792,
      "step": 12988
    },
    {
      "epoch": 0.5775643590769641,
      "grad_norm": 0.06027473136782646,
      "learning_rate": 0.0004094991957453124,
      "loss": 1.4804,
      "step": 12990
    },
    {
      "epoch": 0.5776532835356365,
      "grad_norm": 0.06143961101770401,
      "learning_rate": 0.00040935542565841566,
      "loss": 1.4767,
      "step": 12992
    },
    {
      "epoch": 0.5777422079943089,
      "grad_norm": 0.06066564470529556,
      "learning_rate": 0.00040921166332021693,
      "loss": 1.4848,
      "step": 12994
    },
    {
      "epoch": 0.5778311324529812,
      "grad_norm": 0.060524292290210724,
      "learning_rate": 0.000409067908743005,
      "loss": 1.4874,
      "step": 12996
    },
    {
      "epoch": 0.5779200569116536,
      "grad_norm": 0.0610857717692852,
      "learning_rate": 0.0004089241619390692,
      "loss": 1.4846,
      "step": 12998
    },
    {
      "epoch": 0.5780089813703259,
      "grad_norm": 0.06156258285045624,
      "learning_rate": 0.0004087804229206974,
      "loss": 1.4799,
      "step": 13000
    },
    {
      "epoch": 0.5780089813703259,
      "eval_loss": 1.4671134948730469,
      "eval_runtime": 14.09,
      "eval_samples_per_second": 490.418,
      "eval_steps_per_second": 61.32,
      "step": 13000
    },
    {
      "epoch": 0.5780979058289982,
      "grad_norm": 0.0617251992225647,
      "learning_rate": 0.00040863669170017694,
      "loss": 1.4861,
      "step": 13002
    },
    {
      "epoch": 0.5781868302876706,
      "grad_norm": 0.06050444394350052,
      "learning_rate": 0.0004084929682897947,
      "loss": 1.4846,
      "step": 13004
    },
    {
      "epoch": 0.578275754746343,
      "grad_norm": 0.06325755268335342,
      "learning_rate": 0.00040834925270183685,
      "loss": 1.4795,
      "step": 13006
    },
    {
      "epoch": 0.5783646792050153,
      "grad_norm": 0.06123512238264084,
      "learning_rate": 0.0004082055449485886,
      "loss": 1.484,
      "step": 13008
    },
    {
      "epoch": 0.5784536036636877,
      "grad_norm": 0.061540085822343826,
      "learning_rate": 0.00040806184504233513,
      "loss": 1.4841,
      "step": 13010
    },
    {
      "epoch": 0.5785425281223601,
      "grad_norm": 0.06179589033126831,
      "learning_rate": 0.00040791815299536004,
      "loss": 1.4848,
      "step": 13012
    },
    {
      "epoch": 0.5786314525810324,
      "grad_norm": 0.06283790618181229,
      "learning_rate": 0.0004077744688199473,
      "loss": 1.4888,
      "step": 13014
    },
    {
      "epoch": 0.5787203770397048,
      "grad_norm": 0.06377764791250229,
      "learning_rate": 0.0004076307925283794,
      "loss": 1.4902,
      "step": 13016
    },
    {
      "epoch": 0.5788093014983772,
      "grad_norm": 0.06321964412927628,
      "learning_rate": 0.0004074871241329381,
      "loss": 1.4781,
      "step": 13018
    },
    {
      "epoch": 0.5788982259570495,
      "grad_norm": 0.06205172836780548,
      "learning_rate": 0.0004073434636459054,
      "loss": 1.4869,
      "step": 13020
    },
    {
      "epoch": 0.5789871504157218,
      "grad_norm": 0.06128274276852608,
      "learning_rate": 0.0004071998110795617,
      "loss": 1.4815,
      "step": 13022
    },
    {
      "epoch": 0.5790760748743942,
      "grad_norm": 0.061635617166757584,
      "learning_rate": 0.000407056166446187,
      "loss": 1.4746,
      "step": 13024
    },
    {
      "epoch": 0.5791649993330665,
      "grad_norm": 0.061317794024944305,
      "learning_rate": 0.0004069125297580609,
      "loss": 1.4782,
      "step": 13026
    },
    {
      "epoch": 0.5792539237917389,
      "grad_norm": 0.06503716111183167,
      "learning_rate": 0.0004067689010274618,
      "loss": 1.4814,
      "step": 13028
    },
    {
      "epoch": 0.5793428482504113,
      "grad_norm": 0.06180577725172043,
      "learning_rate": 0.0004066252802666681,
      "loss": 1.4876,
      "step": 13030
    },
    {
      "epoch": 0.5794317727090836,
      "grad_norm": 0.06261365860700607,
      "learning_rate": 0.0004064816674879568,
      "loss": 1.4816,
      "step": 13032
    },
    {
      "epoch": 0.579520697167756,
      "grad_norm": 0.062196627259254456,
      "learning_rate": 0.0004063380627036047,
      "loss": 1.4831,
      "step": 13034
    },
    {
      "epoch": 0.5796096216264284,
      "grad_norm": 0.06358969211578369,
      "learning_rate": 0.00040619446592588784,
      "loss": 1.4869,
      "step": 13036
    },
    {
      "epoch": 0.5796985460851007,
      "grad_norm": 0.061073388904333115,
      "learning_rate": 0.0004060508771670814,
      "loss": 1.4873,
      "step": 13038
    },
    {
      "epoch": 0.5797874705437731,
      "grad_norm": 0.061576247215270996,
      "learning_rate": 0.0004059072964394599,
      "loss": 1.4804,
      "step": 13040
    },
    {
      "epoch": 0.5798763950024455,
      "grad_norm": 0.06221195310354233,
      "learning_rate": 0.0004057637237552974,
      "loss": 1.487,
      "step": 13042
    },
    {
      "epoch": 0.5799653194611177,
      "grad_norm": 0.0631050392985344,
      "learning_rate": 0.00040562015912686705,
      "loss": 1.4875,
      "step": 13044
    },
    {
      "epoch": 0.5800542439197901,
      "grad_norm": 0.062043990939855576,
      "learning_rate": 0.00040547660256644145,
      "loss": 1.4873,
      "step": 13046
    },
    {
      "epoch": 0.5801431683784625,
      "grad_norm": 0.06122414022684097,
      "learning_rate": 0.00040533305408629264,
      "loss": 1.4815,
      "step": 13048
    },
    {
      "epoch": 0.5802320928371348,
      "grad_norm": 0.06327483803033829,
      "learning_rate": 0.0004051895136986912,
      "loss": 1.4902,
      "step": 13050
    },
    {
      "epoch": 0.5803210172958072,
      "grad_norm": 0.06129778176546097,
      "learning_rate": 0.0004050459814159082,
      "loss": 1.477,
      "step": 13052
    },
    {
      "epoch": 0.5804099417544796,
      "grad_norm": 0.0637560561299324,
      "learning_rate": 0.000404902457250213,
      "loss": 1.4809,
      "step": 13054
    },
    {
      "epoch": 0.5804988662131519,
      "grad_norm": 0.06117291748523712,
      "learning_rate": 0.00040475894121387487,
      "loss": 1.4849,
      "step": 13056
    },
    {
      "epoch": 0.5805877906718243,
      "grad_norm": 0.06376662105321884,
      "learning_rate": 0.00040461543331916217,
      "loss": 1.4824,
      "step": 13058
    },
    {
      "epoch": 0.5806767151304967,
      "grad_norm": 0.06137915700674057,
      "learning_rate": 0.0004044719335783426,
      "loss": 1.484,
      "step": 13060
    },
    {
      "epoch": 0.580765639589169,
      "grad_norm": 0.06189562380313873,
      "learning_rate": 0.000404328442003683,
      "loss": 1.4871,
      "step": 13062
    },
    {
      "epoch": 0.5808545640478413,
      "grad_norm": 0.06184488162398338,
      "learning_rate": 0.00040418495860744996,
      "loss": 1.4856,
      "step": 13064
    },
    {
      "epoch": 0.5809434885065137,
      "grad_norm": 0.06308455020189285,
      "learning_rate": 0.0004040414834019088,
      "loss": 1.487,
      "step": 13066
    },
    {
      "epoch": 0.581032412965186,
      "grad_norm": 0.06119653582572937,
      "learning_rate": 0.0004038980163993245,
      "loss": 1.4843,
      "step": 13068
    },
    {
      "epoch": 0.5811213374238584,
      "grad_norm": 0.06232026219367981,
      "learning_rate": 0.0004037545576119615,
      "loss": 1.482,
      "step": 13070
    },
    {
      "epoch": 0.5812102618825308,
      "grad_norm": 0.06109381094574928,
      "learning_rate": 0.0004036111070520827,
      "loss": 1.4848,
      "step": 13072
    },
    {
      "epoch": 0.5812991863412031,
      "grad_norm": 0.0632401630282402,
      "learning_rate": 0.0004034676647319515,
      "loss": 1.4796,
      "step": 13074
    },
    {
      "epoch": 0.5813881107998755,
      "grad_norm": 0.06033066287636757,
      "learning_rate": 0.0004033242306638297,
      "loss": 1.48,
      "step": 13076
    },
    {
      "epoch": 0.5814770352585479,
      "grad_norm": 0.05948418378829956,
      "learning_rate": 0.0004031808048599785,
      "loss": 1.4845,
      "step": 13078
    },
    {
      "epoch": 0.5815659597172202,
      "grad_norm": 0.062302976846694946,
      "learning_rate": 0.00040303738733265897,
      "loss": 1.4862,
      "step": 13080
    },
    {
      "epoch": 0.5816548841758926,
      "grad_norm": 0.061018794775009155,
      "learning_rate": 0.00040289397809413073,
      "loss": 1.4819,
      "step": 13082
    },
    {
      "epoch": 0.581743808634565,
      "grad_norm": 0.061782609671354294,
      "learning_rate": 0.0004027505771566533,
      "loss": 1.4811,
      "step": 13084
    },
    {
      "epoch": 0.5818327330932372,
      "grad_norm": 0.05988406762480736,
      "learning_rate": 0.0004026071845324851,
      "loss": 1.4856,
      "step": 13086
    },
    {
      "epoch": 0.5819216575519096,
      "grad_norm": 0.06019321829080582,
      "learning_rate": 0.0004024638002338838,
      "loss": 1.4828,
      "step": 13088
    },
    {
      "epoch": 0.582010582010582,
      "grad_norm": 0.06038999930024147,
      "learning_rate": 0.00040232042427310684,
      "loss": 1.4912,
      "step": 13090
    },
    {
      "epoch": 0.5820995064692543,
      "grad_norm": 0.06270384043455124,
      "learning_rate": 0.00040217705666241053,
      "loss": 1.4808,
      "step": 13092
    },
    {
      "epoch": 0.5821884309279267,
      "grad_norm": 0.06618236005306244,
      "learning_rate": 0.0004020336974140502,
      "loss": 1.4852,
      "step": 13094
    },
    {
      "epoch": 0.5822773553865991,
      "grad_norm": 0.060933563858270645,
      "learning_rate": 0.0004018903465402815,
      "loss": 1.4846,
      "step": 13096
    },
    {
      "epoch": 0.5823662798452715,
      "grad_norm": 0.06165250390768051,
      "learning_rate": 0.000401747004053358,
      "loss": 1.4827,
      "step": 13098
    },
    {
      "epoch": 0.5824552043039438,
      "grad_norm": 0.06271658837795258,
      "learning_rate": 0.0004016036699655339,
      "loss": 1.4864,
      "step": 13100
    },
    {
      "epoch": 0.5825441287626162,
      "grad_norm": 0.0629393607378006,
      "learning_rate": 0.0004014603442890616,
      "loss": 1.476,
      "step": 13102
    },
    {
      "epoch": 0.5826330532212886,
      "grad_norm": 0.06081317737698555,
      "learning_rate": 0.0004013170270361932,
      "loss": 1.4821,
      "step": 13104
    },
    {
      "epoch": 0.5827219776799609,
      "grad_norm": 0.06229792535305023,
      "learning_rate": 0.0004011737182191803,
      "loss": 1.4831,
      "step": 13106
    },
    {
      "epoch": 0.5828109021386332,
      "grad_norm": 0.0620463564991951,
      "learning_rate": 0.0004010304178502735,
      "loss": 1.4862,
      "step": 13108
    },
    {
      "epoch": 0.5828998265973055,
      "grad_norm": 0.06133623793721199,
      "learning_rate": 0.00040088712594172265,
      "loss": 1.4845,
      "step": 13110
    },
    {
      "epoch": 0.5829887510559779,
      "grad_norm": 0.06249558553099632,
      "learning_rate": 0.0004007438425057771,
      "loss": 1.4844,
      "step": 13112
    },
    {
      "epoch": 0.5830776755146503,
      "grad_norm": 0.06116337701678276,
      "learning_rate": 0.00040060056755468536,
      "loss": 1.487,
      "step": 13114
    },
    {
      "epoch": 0.5831665999733227,
      "grad_norm": 0.06192251667380333,
      "learning_rate": 0.000400457301100695,
      "loss": 1.4818,
      "step": 13116
    },
    {
      "epoch": 0.583255524431995,
      "grad_norm": 0.060988899320364,
      "learning_rate": 0.00040031404315605336,
      "loss": 1.4796,
      "step": 13118
    },
    {
      "epoch": 0.5833444488906674,
      "grad_norm": 0.06281070411205292,
      "learning_rate": 0.00040017079373300644,
      "loss": 1.4803,
      "step": 13120
    },
    {
      "epoch": 0.5834333733493398,
      "grad_norm": 0.06199118122458458,
      "learning_rate": 0.0004000275528438002,
      "loss": 1.4867,
      "step": 13122
    },
    {
      "epoch": 0.5835222978080121,
      "grad_norm": 0.06208673492074013,
      "learning_rate": 0.00039988432050067936,
      "loss": 1.4852,
      "step": 13124
    },
    {
      "epoch": 0.5836112222666845,
      "grad_norm": 0.06448842585086823,
      "learning_rate": 0.00039974109671588764,
      "loss": 1.4811,
      "step": 13126
    },
    {
      "epoch": 0.5837001467253569,
      "grad_norm": 0.06038080155849457,
      "learning_rate": 0.0003995978815016692,
      "loss": 1.4787,
      "step": 13128
    },
    {
      "epoch": 0.5837890711840291,
      "grad_norm": 0.06375820189714432,
      "learning_rate": 0.00039945467487026616,
      "loss": 1.4855,
      "step": 13130
    },
    {
      "epoch": 0.5838779956427015,
      "grad_norm": 0.06250657886266708,
      "learning_rate": 0.0003993114768339205,
      "loss": 1.4816,
      "step": 13132
    },
    {
      "epoch": 0.5839669201013739,
      "grad_norm": 0.061796385794878006,
      "learning_rate": 0.0003991682874048736,
      "loss": 1.4791,
      "step": 13134
    },
    {
      "epoch": 0.5840558445600462,
      "grad_norm": 0.06170998141169548,
      "learning_rate": 0.0003990251065953658,
      "loss": 1.4829,
      "step": 13136
    },
    {
      "epoch": 0.5841447690187186,
      "grad_norm": 0.060505084693431854,
      "learning_rate": 0.0003988819344176369,
      "loss": 1.4766,
      "step": 13138
    },
    {
      "epoch": 0.584233693477391,
      "grad_norm": 0.06287679821252823,
      "learning_rate": 0.0003987387708839259,
      "loss": 1.4804,
      "step": 13140
    },
    {
      "epoch": 0.5843226179360633,
      "grad_norm": 0.061412662267684937,
      "learning_rate": 0.0003985956160064709,
      "loss": 1.4776,
      "step": 13142
    },
    {
      "epoch": 0.5844115423947357,
      "grad_norm": 0.06276515871286392,
      "learning_rate": 0.0003984524697975096,
      "loss": 1.4818,
      "step": 13144
    },
    {
      "epoch": 0.5845004668534081,
      "grad_norm": 0.06057889387011528,
      "learning_rate": 0.00039830933226927883,
      "loss": 1.4876,
      "step": 13146
    },
    {
      "epoch": 0.5845893913120804,
      "grad_norm": 0.061848003417253494,
      "learning_rate": 0.00039816620343401414,
      "loss": 1.4808,
      "step": 13148
    },
    {
      "epoch": 0.5846783157707528,
      "grad_norm": 0.06087247282266617,
      "learning_rate": 0.00039802308330395147,
      "loss": 1.4855,
      "step": 13150
    },
    {
      "epoch": 0.5847672402294251,
      "grad_norm": 0.06292174756526947,
      "learning_rate": 0.00039787997189132476,
      "loss": 1.4838,
      "step": 13152
    },
    {
      "epoch": 0.5848561646880974,
      "grad_norm": 0.06331959366798401,
      "learning_rate": 0.0003977368692083683,
      "loss": 1.482,
      "step": 13154
    },
    {
      "epoch": 0.5849450891467698,
      "grad_norm": 0.06192371994256973,
      "learning_rate": 0.0003975937752673149,
      "loss": 1.4817,
      "step": 13156
    },
    {
      "epoch": 0.5850340136054422,
      "grad_norm": 0.06208168342709541,
      "learning_rate": 0.0003974506900803967,
      "loss": 1.4868,
      "step": 13158
    },
    {
      "epoch": 0.5851229380641145,
      "grad_norm": 0.0600164569914341,
      "learning_rate": 0.0003973076136598456,
      "loss": 1.4877,
      "step": 13160
    },
    {
      "epoch": 0.5852118625227869,
      "grad_norm": 0.06281714886426926,
      "learning_rate": 0.0003971645460178922,
      "loss": 1.4813,
      "step": 13162
    },
    {
      "epoch": 0.5853007869814593,
      "grad_norm": 0.061320312321186066,
      "learning_rate": 0.0003970214871667665,
      "loss": 1.4813,
      "step": 13164
    },
    {
      "epoch": 0.5853897114401316,
      "grad_norm": 0.06139989197254181,
      "learning_rate": 0.000396878437118698,
      "loss": 1.4865,
      "step": 13166
    },
    {
      "epoch": 0.585478635898804,
      "grad_norm": 0.060984056442976,
      "learning_rate": 0.000396735395885915,
      "loss": 1.4816,
      "step": 13168
    },
    {
      "epoch": 0.5855675603574764,
      "grad_norm": 0.06185859069228172,
      "learning_rate": 0.00039659236348064557,
      "loss": 1.4809,
      "step": 13170
    },
    {
      "epoch": 0.5856564848161487,
      "grad_norm": 0.06174325570464134,
      "learning_rate": 0.0003964493399151166,
      "loss": 1.4774,
      "step": 13172
    },
    {
      "epoch": 0.585745409274821,
      "grad_norm": 0.06207715719938278,
      "learning_rate": 0.0003963063252015543,
      "loss": 1.4763,
      "step": 13174
    },
    {
      "epoch": 0.5858343337334934,
      "grad_norm": 0.06227646395564079,
      "learning_rate": 0.00039616331935218434,
      "loss": 1.4841,
      "step": 13176
    },
    {
      "epoch": 0.5859232581921657,
      "grad_norm": 0.06518425047397614,
      "learning_rate": 0.00039602032237923157,
      "loss": 1.4859,
      "step": 13178
    },
    {
      "epoch": 0.5860121826508381,
      "grad_norm": 0.06016527861356735,
      "learning_rate": 0.0003958773342949196,
      "loss": 1.4804,
      "step": 13180
    },
    {
      "epoch": 0.5861011071095105,
      "grad_norm": 0.060563359409570694,
      "learning_rate": 0.0003957343551114722,
      "loss": 1.4884,
      "step": 13182
    },
    {
      "epoch": 0.5861900315681828,
      "grad_norm": 0.060045305639505386,
      "learning_rate": 0.00039559138484111155,
      "loss": 1.4795,
      "step": 13184
    },
    {
      "epoch": 0.5862789560268552,
      "grad_norm": 0.06347208470106125,
      "learning_rate": 0.0003954484234960593,
      "loss": 1.4833,
      "step": 13186
    },
    {
      "epoch": 0.5863678804855276,
      "grad_norm": 0.0632077008485794,
      "learning_rate": 0.00039530547108853665,
      "loss": 1.4842,
      "step": 13188
    },
    {
      "epoch": 0.5864568049441999,
      "grad_norm": 0.061214666813611984,
      "learning_rate": 0.0003951625276307636,
      "loss": 1.4823,
      "step": 13190
    },
    {
      "epoch": 0.5865457294028723,
      "grad_norm": 0.06085453927516937,
      "learning_rate": 0.00039501959313495986,
      "loss": 1.4797,
      "step": 13192
    },
    {
      "epoch": 0.5866346538615446,
      "grad_norm": 0.06207851693034172,
      "learning_rate": 0.00039487666761334384,
      "loss": 1.4758,
      "step": 13194
    },
    {
      "epoch": 0.5867235783202169,
      "grad_norm": 0.061899829655885696,
      "learning_rate": 0.0003947337510781335,
      "loss": 1.4801,
      "step": 13196
    },
    {
      "epoch": 0.5868125027788893,
      "grad_norm": 0.06145618110895157,
      "learning_rate": 0.0003945908435415461,
      "loss": 1.4831,
      "step": 13198
    },
    {
      "epoch": 0.5869014272375617,
      "grad_norm": 0.06099288538098335,
      "learning_rate": 0.000394447945015798,
      "loss": 1.4886,
      "step": 13200
    },
    {
      "epoch": 0.586990351696234,
      "grad_norm": 0.06336424499750137,
      "learning_rate": 0.0003943050555131044,
      "loss": 1.4824,
      "step": 13202
    },
    {
      "epoch": 0.5870792761549064,
      "grad_norm": 0.062214065343141556,
      "learning_rate": 0.00039416217504568073,
      "loss": 1.4845,
      "step": 13204
    },
    {
      "epoch": 0.5871682006135788,
      "grad_norm": 0.0627097636461258,
      "learning_rate": 0.0003940193036257405,
      "loss": 1.4835,
      "step": 13206
    },
    {
      "epoch": 0.5872571250722511,
      "grad_norm": 0.06262244284152985,
      "learning_rate": 0.0003938764412654975,
      "loss": 1.4757,
      "step": 13208
    },
    {
      "epoch": 0.5873460495309235,
      "grad_norm": 0.06010544300079346,
      "learning_rate": 0.0003937335879771639,
      "loss": 1.4811,
      "step": 13210
    },
    {
      "epoch": 0.5874349739895959,
      "grad_norm": 0.06312010437250137,
      "learning_rate": 0.0003935907437729514,
      "loss": 1.4795,
      "step": 13212
    },
    {
      "epoch": 0.5875238984482682,
      "grad_norm": 0.06187771260738373,
      "learning_rate": 0.00039344790866507114,
      "loss": 1.4758,
      "step": 13214
    },
    {
      "epoch": 0.5876128229069405,
      "grad_norm": 0.060956425964832306,
      "learning_rate": 0.0003933050826657332,
      "loss": 1.478,
      "step": 13216
    },
    {
      "epoch": 0.5877017473656129,
      "grad_norm": 0.062204815447330475,
      "learning_rate": 0.00039316226578714686,
      "loss": 1.487,
      "step": 13218
    },
    {
      "epoch": 0.5877906718242852,
      "grad_norm": 0.06293026357889175,
      "learning_rate": 0.000393019458041521,
      "loss": 1.4837,
      "step": 13220
    },
    {
      "epoch": 0.5878795962829576,
      "grad_norm": 0.06233726814389229,
      "learning_rate": 0.00039287665944106324,
      "loss": 1.4822,
      "step": 13222
    },
    {
      "epoch": 0.58796852074163,
      "grad_norm": 0.06225262209773064,
      "learning_rate": 0.0003927338699979807,
      "loss": 1.4815,
      "step": 13224
    },
    {
      "epoch": 0.5880574452003023,
      "grad_norm": 0.06425213813781738,
      "learning_rate": 0.00039259108972447977,
      "loss": 1.4861,
      "step": 13226
    },
    {
      "epoch": 0.5881463696589747,
      "grad_norm": 0.06262674182653427,
      "learning_rate": 0.0003924483186327656,
      "loss": 1.4801,
      "step": 13228
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.06313468515872955,
      "learning_rate": 0.0003923055567350434,
      "loss": 1.4816,
      "step": 13230
    },
    {
      "epoch": 0.5883242185763194,
      "grad_norm": 0.06385188549757004,
      "learning_rate": 0.00039216280404351676,
      "loss": 1.4834,
      "step": 13232
    },
    {
      "epoch": 0.5884131430349918,
      "grad_norm": 0.06282700598239899,
      "learning_rate": 0.00039202006057038863,
      "loss": 1.4805,
      "step": 13234
    },
    {
      "epoch": 0.5885020674936642,
      "grad_norm": 0.061191339045763016,
      "learning_rate": 0.0003918773263278618,
      "loss": 1.4793,
      "step": 13236
    },
    {
      "epoch": 0.5885909919523364,
      "grad_norm": 0.059706203639507294,
      "learning_rate": 0.0003917346013281376,
      "loss": 1.482,
      "step": 13238
    },
    {
      "epoch": 0.5886799164110088,
      "grad_norm": 0.06191098317503929,
      "learning_rate": 0.00039159188558341666,
      "loss": 1.4848,
      "step": 13240
    },
    {
      "epoch": 0.5887688408696812,
      "grad_norm": 0.0614837221801281,
      "learning_rate": 0.0003914491791058991,
      "loss": 1.4783,
      "step": 13242
    },
    {
      "epoch": 0.5888577653283535,
      "grad_norm": 0.06151347979903221,
      "learning_rate": 0.0003913064819077841,
      "loss": 1.4825,
      "step": 13244
    },
    {
      "epoch": 0.5889466897870259,
      "grad_norm": 0.06056448817253113,
      "learning_rate": 0.00039116379400127,
      "loss": 1.4811,
      "step": 13246
    },
    {
      "epoch": 0.5890356142456983,
      "grad_norm": 0.06178190931677818,
      "learning_rate": 0.00039102111539855445,
      "loss": 1.4794,
      "step": 13248
    },
    {
      "epoch": 0.5891245387043706,
      "grad_norm": 0.061939485371112823,
      "learning_rate": 0.0003908784461118341,
      "loss": 1.4831,
      "step": 13250
    },
    {
      "epoch": 0.589213463163043,
      "grad_norm": 0.0606502890586853,
      "learning_rate": 0.0003907357861533052,
      "loss": 1.4822,
      "step": 13252
    },
    {
      "epoch": 0.5893023876217154,
      "grad_norm": 0.06153089553117752,
      "learning_rate": 0.0003905931355351629,
      "loss": 1.4805,
      "step": 13254
    },
    {
      "epoch": 0.5893913120803878,
      "grad_norm": 0.05961450934410095,
      "learning_rate": 0.00039045049426960124,
      "loss": 1.4787,
      "step": 13256
    },
    {
      "epoch": 0.5894802365390601,
      "grad_norm": 0.06326043605804443,
      "learning_rate": 0.0003903078623688143,
      "loss": 1.4821,
      "step": 13258
    },
    {
      "epoch": 0.5895691609977324,
      "grad_norm": 0.06015404686331749,
      "learning_rate": 0.00039016523984499455,
      "loss": 1.4795,
      "step": 13260
    },
    {
      "epoch": 0.5896580854564047,
      "grad_norm": 0.06200879067182541,
      "learning_rate": 0.0003900226267103343,
      "loss": 1.4836,
      "step": 13262
    },
    {
      "epoch": 0.5897470099150771,
      "grad_norm": 0.060929592698812485,
      "learning_rate": 0.0003898800229770245,
      "loss": 1.4835,
      "step": 13264
    },
    {
      "epoch": 0.5898359343737495,
      "grad_norm": 0.061469994485378265,
      "learning_rate": 0.0003897374286572555,
      "loss": 1.4779,
      "step": 13266
    },
    {
      "epoch": 0.5899248588324219,
      "grad_norm": 0.06109357252717018,
      "learning_rate": 0.0003895948437632171,
      "loss": 1.4852,
      "step": 13268
    },
    {
      "epoch": 0.5900137832910942,
      "grad_norm": 0.06095615401864052,
      "learning_rate": 0.0003894522683070981,
      "loss": 1.4847,
      "step": 13270
    },
    {
      "epoch": 0.5901027077497666,
      "grad_norm": 0.06231928989291191,
      "learning_rate": 0.00038930970230108616,
      "loss": 1.4761,
      "step": 13272
    },
    {
      "epoch": 0.590191632208439,
      "grad_norm": 0.06076860427856445,
      "learning_rate": 0.0003891671457573688,
      "loss": 1.4803,
      "step": 13274
    },
    {
      "epoch": 0.5902805566671113,
      "grad_norm": 0.06351371854543686,
      "learning_rate": 0.00038902459868813213,
      "loss": 1.4779,
      "step": 13276
    },
    {
      "epoch": 0.5903694811257837,
      "grad_norm": 0.0635959729552269,
      "learning_rate": 0.00038888206110556197,
      "loss": 1.4837,
      "step": 13278
    },
    {
      "epoch": 0.5904584055844561,
      "grad_norm": 0.06213785707950592,
      "learning_rate": 0.00038873953302184284,
      "loss": 1.481,
      "step": 13280
    },
    {
      "epoch": 0.5905473300431283,
      "grad_norm": 0.060719020664691925,
      "learning_rate": 0.0003885970144491587,
      "loss": 1.4851,
      "step": 13282
    },
    {
      "epoch": 0.5906362545018007,
      "grad_norm": 0.06250161677598953,
      "learning_rate": 0.00038845450539969274,
      "loss": 1.4812,
      "step": 13284
    },
    {
      "epoch": 0.590725178960473,
      "grad_norm": 0.061430674046278,
      "learning_rate": 0.00038831200588562736,
      "loss": 1.481,
      "step": 13286
    },
    {
      "epoch": 0.5908141034191454,
      "grad_norm": 0.06262683868408203,
      "learning_rate": 0.00038816951591914356,
      "loss": 1.4801,
      "step": 13288
    },
    {
      "epoch": 0.5909030278778178,
      "grad_norm": 0.062282636761665344,
      "learning_rate": 0.00038802703551242275,
      "loss": 1.4776,
      "step": 13290
    },
    {
      "epoch": 0.5909919523364902,
      "grad_norm": 0.06258291751146317,
      "learning_rate": 0.000387884564677644,
      "loss": 1.4808,
      "step": 13292
    },
    {
      "epoch": 0.5910808767951625,
      "grad_norm": 0.06155584007501602,
      "learning_rate": 0.00038774210342698714,
      "loss": 1.4778,
      "step": 13294
    },
    {
      "epoch": 0.5911698012538349,
      "grad_norm": 0.062455207109451294,
      "learning_rate": 0.0003875996517726298,
      "loss": 1.4812,
      "step": 13296
    },
    {
      "epoch": 0.5912587257125073,
      "grad_norm": 0.06287974864244461,
      "learning_rate": 0.0003874572097267495,
      "loss": 1.4872,
      "step": 13298
    },
    {
      "epoch": 0.5913476501711796,
      "grad_norm": 0.060571350157260895,
      "learning_rate": 0.00038731477730152294,
      "loss": 1.4832,
      "step": 13300
    },
    {
      "epoch": 0.5914365746298519,
      "grad_norm": 0.061406493186950684,
      "learning_rate": 0.0003871723545091258,
      "loss": 1.4813,
      "step": 13302
    },
    {
      "epoch": 0.5915254990885243,
      "grad_norm": 0.060636457055807114,
      "learning_rate": 0.000387029941361733,
      "loss": 1.485,
      "step": 13304
    },
    {
      "epoch": 0.5916144235471966,
      "grad_norm": 0.06447170674800873,
      "learning_rate": 0.00038688753787151863,
      "loss": 1.4819,
      "step": 13306
    },
    {
      "epoch": 0.591703348005869,
      "grad_norm": 0.06106022372841835,
      "learning_rate": 0.0003867451440506562,
      "loss": 1.4842,
      "step": 13308
    },
    {
      "epoch": 0.5917922724645414,
      "grad_norm": 0.062031909823417664,
      "learning_rate": 0.00038660275991131757,
      "loss": 1.4892,
      "step": 13310
    },
    {
      "epoch": 0.5918811969232137,
      "grad_norm": 0.05990834906697273,
      "learning_rate": 0.0003864603854656751,
      "loss": 1.4825,
      "step": 13312
    },
    {
      "epoch": 0.5919701213818861,
      "grad_norm": 0.062491413205862045,
      "learning_rate": 0.0003863180207258989,
      "loss": 1.4884,
      "step": 13314
    },
    {
      "epoch": 0.5920590458405585,
      "grad_norm": 0.06330056488513947,
      "learning_rate": 0.00038617566570415965,
      "loss": 1.4825,
      "step": 13316
    },
    {
      "epoch": 0.5921479702992308,
      "grad_norm": 0.060776226222515106,
      "learning_rate": 0.0003860333204126259,
      "loss": 1.4821,
      "step": 13318
    },
    {
      "epoch": 0.5922368947579032,
      "grad_norm": 0.05994359776377678,
      "learning_rate": 0.00038589098486346607,
      "loss": 1.4898,
      "step": 13320
    },
    {
      "epoch": 0.5923258192165756,
      "grad_norm": 0.0633685290813446,
      "learning_rate": 0.00038574865906884783,
      "loss": 1.4819,
      "step": 13322
    },
    {
      "epoch": 0.5924147436752478,
      "grad_norm": 0.06285269558429718,
      "learning_rate": 0.0003856063430409377,
      "loss": 1.4866,
      "step": 13324
    },
    {
      "epoch": 0.5925036681339202,
      "grad_norm": 0.06075499951839447,
      "learning_rate": 0.00038546403679190147,
      "loss": 1.4801,
      "step": 13326
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 0.062371302396059036,
      "learning_rate": 0.0003853217403339041,
      "loss": 1.4821,
      "step": 13328
    },
    {
      "epoch": 0.5926815170512649,
      "grad_norm": 0.06306248158216476,
      "learning_rate": 0.00038517945367910967,
      "loss": 1.4912,
      "step": 13330
    },
    {
      "epoch": 0.5927704415099373,
      "grad_norm": 0.0632634311914444,
      "learning_rate": 0.00038503717683968164,
      "loss": 1.4819,
      "step": 13332
    },
    {
      "epoch": 0.5928593659686097,
      "grad_norm": 0.06258311122655869,
      "learning_rate": 0.0003848949098277824,
      "loss": 1.4828,
      "step": 13334
    },
    {
      "epoch": 0.592948290427282,
      "grad_norm": 0.06177235394716263,
      "learning_rate": 0.00038475265265557334,
      "loss": 1.4772,
      "step": 13336
    },
    {
      "epoch": 0.5930372148859544,
      "grad_norm": 0.0634724497795105,
      "learning_rate": 0.0003846104053352156,
      "loss": 1.4817,
      "step": 13338
    },
    {
      "epoch": 0.5931261393446268,
      "grad_norm": 0.06223030388355255,
      "learning_rate": 0.00038446816787886896,
      "loss": 1.4796,
      "step": 13340
    },
    {
      "epoch": 0.5932150638032991,
      "grad_norm": 0.06202046200633049,
      "learning_rate": 0.0003843259402986922,
      "loss": 1.4812,
      "step": 13342
    },
    {
      "epoch": 0.5933039882619715,
      "grad_norm": 0.06367559731006622,
      "learning_rate": 0.00038418372260684406,
      "loss": 1.4762,
      "step": 13344
    },
    {
      "epoch": 0.5933929127206438,
      "grad_norm": 0.060448892414569855,
      "learning_rate": 0.00038404151481548143,
      "loss": 1.4802,
      "step": 13346
    },
    {
      "epoch": 0.5934818371793161,
      "grad_norm": 0.06097383797168732,
      "learning_rate": 0.0003838993169367615,
      "loss": 1.4802,
      "step": 13348
    },
    {
      "epoch": 0.5935707616379885,
      "grad_norm": 0.061548784375190735,
      "learning_rate": 0.0003837571289828394,
      "loss": 1.4821,
      "step": 13350
    },
    {
      "epoch": 0.5936596860966609,
      "grad_norm": 0.060715217143297195,
      "learning_rate": 0.0003836149509658701,
      "loss": 1.4833,
      "step": 13352
    },
    {
      "epoch": 0.5937486105553332,
      "grad_norm": 0.062499675899744034,
      "learning_rate": 0.0003834727828980078,
      "loss": 1.4768,
      "step": 13354
    },
    {
      "epoch": 0.5938375350140056,
      "grad_norm": 0.06051585450768471,
      "learning_rate": 0.00038333062479140555,
      "loss": 1.4806,
      "step": 13356
    },
    {
      "epoch": 0.593926459472678,
      "grad_norm": 0.06072830781340599,
      "learning_rate": 0.00038318847665821547,
      "loss": 1.4808,
      "step": 13358
    },
    {
      "epoch": 0.5940153839313503,
      "grad_norm": 0.06030082702636719,
      "learning_rate": 0.00038304633851058936,
      "loss": 1.4784,
      "step": 13360
    },
    {
      "epoch": 0.5941043083900227,
      "grad_norm": 0.06220033019781113,
      "learning_rate": 0.0003829042103606777,
      "loss": 1.4826,
      "step": 13362
    },
    {
      "epoch": 0.5941932328486951,
      "grad_norm": 0.0611334852874279,
      "learning_rate": 0.0003827620922206299,
      "loss": 1.4782,
      "step": 13364
    },
    {
      "epoch": 0.5942821573073674,
      "grad_norm": 0.06260346621274948,
      "learning_rate": 0.00038261998410259537,
      "loss": 1.4794,
      "step": 13366
    },
    {
      "epoch": 0.5943710817660397,
      "grad_norm": 0.06368494778871536,
      "learning_rate": 0.00038247788601872154,
      "loss": 1.4802,
      "step": 13368
    },
    {
      "epoch": 0.5944600062247121,
      "grad_norm": 0.06044352427124977,
      "learning_rate": 0.00038233579798115623,
      "loss": 1.4727,
      "step": 13370
    },
    {
      "epoch": 0.5945489306833844,
      "grad_norm": 0.06313683092594147,
      "learning_rate": 0.00038219372000204536,
      "loss": 1.4744,
      "step": 13372
    },
    {
      "epoch": 0.5946378551420568,
      "grad_norm": 0.06014512851834297,
      "learning_rate": 0.0003820516520935344,
      "loss": 1.4856,
      "step": 13374
    },
    {
      "epoch": 0.5947267796007292,
      "grad_norm": 0.06175994500517845,
      "learning_rate": 0.00038190959426776807,
      "loss": 1.4838,
      "step": 13376
    },
    {
      "epoch": 0.5948157040594015,
      "grad_norm": 0.06352592259645462,
      "learning_rate": 0.00038176754653689,
      "loss": 1.4735,
      "step": 13378
    },
    {
      "epoch": 0.5949046285180739,
      "grad_norm": 0.060022275894880295,
      "learning_rate": 0.00038162550891304297,
      "loss": 1.4773,
      "step": 13380
    },
    {
      "epoch": 0.5949935529767463,
      "grad_norm": 0.06236466392874718,
      "learning_rate": 0.00038148348140836923,
      "loss": 1.4835,
      "step": 13382
    },
    {
      "epoch": 0.5950824774354186,
      "grad_norm": 0.06078318506479263,
      "learning_rate": 0.0003813414640350097,
      "loss": 1.4867,
      "step": 13384
    },
    {
      "epoch": 0.595171401894091,
      "grad_norm": 0.06466275453567505,
      "learning_rate": 0.0003811994568051048,
      "loss": 1.4751,
      "step": 13386
    },
    {
      "epoch": 0.5952603263527634,
      "grad_norm": 0.062311556190252304,
      "learning_rate": 0.00038105745973079404,
      "loss": 1.4843,
      "step": 13388
    },
    {
      "epoch": 0.5953492508114356,
      "grad_norm": 0.06244157627224922,
      "learning_rate": 0.0003809154728242154,
      "loss": 1.4692,
      "step": 13390
    },
    {
      "epoch": 0.595438175270108,
      "grad_norm": 0.0595109798014164,
      "learning_rate": 0.00038077349609750725,
      "loss": 1.4827,
      "step": 13392
    },
    {
      "epoch": 0.5955270997287804,
      "grad_norm": 0.061755720525979996,
      "learning_rate": 0.000380631529562806,
      "loss": 1.4748,
      "step": 13394
    },
    {
      "epoch": 0.5956160241874527,
      "grad_norm": 0.06010547652840614,
      "learning_rate": 0.00038048957323224755,
      "loss": 1.4804,
      "step": 13396
    },
    {
      "epoch": 0.5957049486461251,
      "grad_norm": 0.0607173889875412,
      "learning_rate": 0.0003803476271179671,
      "loss": 1.48,
      "step": 13398
    },
    {
      "epoch": 0.5957938731047975,
      "grad_norm": 0.061135951429605484,
      "learning_rate": 0.0003802056912320987,
      "loss": 1.4828,
      "step": 13400
    },
    {
      "epoch": 0.5958827975634698,
      "grad_norm": 0.06174211576581001,
      "learning_rate": 0.00038006376558677587,
      "loss": 1.4777,
      "step": 13402
    },
    {
      "epoch": 0.5959717220221422,
      "grad_norm": 0.062369443476200104,
      "learning_rate": 0.00037992185019413083,
      "loss": 1.4799,
      "step": 13404
    },
    {
      "epoch": 0.5960606464808146,
      "grad_norm": 0.06260187923908234,
      "learning_rate": 0.0003797799450662951,
      "loss": 1.4755,
      "step": 13406
    },
    {
      "epoch": 0.596149570939487,
      "grad_norm": 0.06085886061191559,
      "learning_rate": 0.0003796380502153996,
      "loss": 1.4821,
      "step": 13408
    },
    {
      "epoch": 0.5962384953981593,
      "grad_norm": 0.061523739248514175,
      "learning_rate": 0.000379496165653574,
      "loss": 1.4822,
      "step": 13410
    },
    {
      "epoch": 0.5963274198568316,
      "grad_norm": 0.06035231426358223,
      "learning_rate": 0.0003793542913929471,
      "loss": 1.4808,
      "step": 13412
    },
    {
      "epoch": 0.5964163443155039,
      "grad_norm": 0.061069756746292114,
      "learning_rate": 0.0003792124274456471,
      "loss": 1.4753,
      "step": 13414
    },
    {
      "epoch": 0.5965052687741763,
      "grad_norm": 0.062064070254564285,
      "learning_rate": 0.000379070573823801,
      "loss": 1.482,
      "step": 13416
    },
    {
      "epoch": 0.5965941932328487,
      "grad_norm": 0.06175654008984566,
      "learning_rate": 0.0003789287305395352,
      "loss": 1.4805,
      "step": 13418
    },
    {
      "epoch": 0.596683117691521,
      "grad_norm": 0.06208856776356697,
      "learning_rate": 0.0003787868976049753,
      "loss": 1.4824,
      "step": 13420
    },
    {
      "epoch": 0.5967720421501934,
      "grad_norm": 0.062433741986751556,
      "learning_rate": 0.0003786450750322452,
      "loss": 1.4762,
      "step": 13422
    },
    {
      "epoch": 0.5968609666088658,
      "grad_norm": 0.061507366597652435,
      "learning_rate": 0.0003785032628334692,
      "loss": 1.482,
      "step": 13424
    },
    {
      "epoch": 0.5969498910675382,
      "grad_norm": 0.061356283724308014,
      "learning_rate": 0.0003783614610207696,
      "loss": 1.4781,
      "step": 13426
    },
    {
      "epoch": 0.5970388155262105,
      "grad_norm": 0.06329002976417542,
      "learning_rate": 0.00037821966960626817,
      "loss": 1.4757,
      "step": 13428
    },
    {
      "epoch": 0.5971277399848829,
      "grad_norm": 0.06240663677453995,
      "learning_rate": 0.0003780778886020863,
      "loss": 1.4779,
      "step": 13430
    },
    {
      "epoch": 0.5972166644435551,
      "grad_norm": 0.06056615337729454,
      "learning_rate": 0.0003779361180203437,
      "loss": 1.4722,
      "step": 13432
    },
    {
      "epoch": 0.5973055889022275,
      "grad_norm": 0.060851022601127625,
      "learning_rate": 0.0003777943578731596,
      "loss": 1.4827,
      "step": 13434
    },
    {
      "epoch": 0.5973945133608999,
      "grad_norm": 0.06265364587306976,
      "learning_rate": 0.0003776526081726524,
      "loss": 1.4797,
      "step": 13436
    },
    {
      "epoch": 0.5974834378195722,
      "grad_norm": 0.06204288452863693,
      "learning_rate": 0.0003775108689309393,
      "loss": 1.4828,
      "step": 13438
    },
    {
      "epoch": 0.5975723622782446,
      "grad_norm": 0.06286075711250305,
      "learning_rate": 0.00037736914016013714,
      "loss": 1.4838,
      "step": 13440
    },
    {
      "epoch": 0.597661286736917,
      "grad_norm": 0.059687890112400055,
      "learning_rate": 0.00037722742187236126,
      "loss": 1.4833,
      "step": 13442
    },
    {
      "epoch": 0.5977502111955894,
      "grad_norm": 0.060804031789302826,
      "learning_rate": 0.00037708571407972616,
      "loss": 1.4799,
      "step": 13444
    },
    {
      "epoch": 0.5978391356542617,
      "grad_norm": 0.061151038855314255,
      "learning_rate": 0.0003769440167943462,
      "loss": 1.4771,
      "step": 13446
    },
    {
      "epoch": 0.5979280601129341,
      "grad_norm": 0.06135578826069832,
      "learning_rate": 0.0003768023300283338,
      "loss": 1.4824,
      "step": 13448
    },
    {
      "epoch": 0.5980169845716065,
      "grad_norm": 0.06151726096868515,
      "learning_rate": 0.0003766606537938011,
      "loss": 1.4837,
      "step": 13450
    },
    {
      "epoch": 0.5981059090302788,
      "grad_norm": 0.06048937514424324,
      "learning_rate": 0.00037651898810285936,
      "loss": 1.4808,
      "step": 13452
    },
    {
      "epoch": 0.5981948334889511,
      "grad_norm": 0.06177039071917534,
      "learning_rate": 0.00037637733296761854,
      "loss": 1.4743,
      "step": 13454
    },
    {
      "epoch": 0.5982837579476235,
      "grad_norm": 0.06174348294734955,
      "learning_rate": 0.0003762356884001882,
      "loss": 1.4838,
      "step": 13456
    },
    {
      "epoch": 0.5983726824062958,
      "grad_norm": 0.06009867042303085,
      "learning_rate": 0.0003760940544126766,
      "loss": 1.4775,
      "step": 13458
    },
    {
      "epoch": 0.5984616068649682,
      "grad_norm": 0.06166113540530205,
      "learning_rate": 0.0003759524310171912,
      "loss": 1.4841,
      "step": 13460
    },
    {
      "epoch": 0.5985505313236406,
      "grad_norm": 0.061972811818122864,
      "learning_rate": 0.0003758108182258388,
      "loss": 1.4817,
      "step": 13462
    },
    {
      "epoch": 0.5986394557823129,
      "grad_norm": 0.059086523950099945,
      "learning_rate": 0.0003756692160507249,
      "loss": 1.4769,
      "step": 13464
    },
    {
      "epoch": 0.5987283802409853,
      "grad_norm": 0.06444913148880005,
      "learning_rate": 0.0003755276245039542,
      "loss": 1.4782,
      "step": 13466
    },
    {
      "epoch": 0.5988173046996577,
      "grad_norm": 0.06151975691318512,
      "learning_rate": 0.00037538604359763084,
      "loss": 1.4785,
      "step": 13468
    },
    {
      "epoch": 0.59890622915833,
      "grad_norm": 0.06152950972318649,
      "learning_rate": 0.0003752444733438575,
      "loss": 1.4749,
      "step": 13470
    },
    {
      "epoch": 0.5989951536170024,
      "grad_norm": 0.060316842049360275,
      "learning_rate": 0.0003751029137547364,
      "loss": 1.4807,
      "step": 13472
    },
    {
      "epoch": 0.5990840780756748,
      "grad_norm": 0.06243583559989929,
      "learning_rate": 0.00037496136484236887,
      "loss": 1.4782,
      "step": 13474
    },
    {
      "epoch": 0.599173002534347,
      "grad_norm": 0.06194531172513962,
      "learning_rate": 0.0003748198266188545,
      "loss": 1.481,
      "step": 13476
    },
    {
      "epoch": 0.5992619269930194,
      "grad_norm": 0.061740942299366,
      "learning_rate": 0.0003746782990962933,
      "loss": 1.4746,
      "step": 13478
    },
    {
      "epoch": 0.5993508514516918,
      "grad_norm": 0.06074299290776253,
      "learning_rate": 0.00037453678228678333,
      "loss": 1.4771,
      "step": 13480
    },
    {
      "epoch": 0.5994397759103641,
      "grad_norm": 0.060749128460884094,
      "learning_rate": 0.00037439527620242196,
      "loss": 1.4847,
      "step": 13482
    },
    {
      "epoch": 0.5995287003690365,
      "grad_norm": 0.06093888357281685,
      "learning_rate": 0.00037425378085530604,
      "loss": 1.4798,
      "step": 13484
    },
    {
      "epoch": 0.5996176248277089,
      "grad_norm": 0.061810996383428574,
      "learning_rate": 0.0003741122962575309,
      "loss": 1.4813,
      "step": 13486
    },
    {
      "epoch": 0.5997065492863812,
      "grad_norm": 0.06119661033153534,
      "learning_rate": 0.00037397082242119164,
      "loss": 1.4694,
      "step": 13488
    },
    {
      "epoch": 0.5997954737450536,
      "grad_norm": 0.0609118789434433,
      "learning_rate": 0.0003738293593583818,
      "loss": 1.4833,
      "step": 13490
    },
    {
      "epoch": 0.599884398203726,
      "grad_norm": 0.06090276688337326,
      "learning_rate": 0.00037368790708119427,
      "loss": 1.4821,
      "step": 13492
    },
    {
      "epoch": 0.5999733226623983,
      "grad_norm": 0.06049749627709389,
      "learning_rate": 0.0003735464656017212,
      "loss": 1.4806,
      "step": 13494
    },
    {
      "epoch": 0.6000622471210707,
      "grad_norm": 0.06154240667819977,
      "learning_rate": 0.00037340503493205355,
      "loss": 1.4845,
      "step": 13496
    },
    {
      "epoch": 0.600151171579743,
      "grad_norm": 0.060583166778087616,
      "learning_rate": 0.0003732636150842811,
      "loss": 1.4793,
      "step": 13498
    },
    {
      "epoch": 0.6002400960384153,
      "grad_norm": 0.06009206175804138,
      "learning_rate": 0.00037312220607049377,
      "loss": 1.4784,
      "step": 13500
    },
    {
      "epoch": 0.6002400960384153,
      "eval_loss": 1.4636259078979492,
      "eval_runtime": 12.3839,
      "eval_samples_per_second": 557.984,
      "eval_steps_per_second": 69.768,
      "step": 13500
    },
    {
      "epoch": 0.6003290204970877,
      "grad_norm": 0.061223968863487244,
      "learning_rate": 0.0003729808079027792,
      "loss": 1.4799,
      "step": 13502
    },
    {
      "epoch": 0.6004179449557601,
      "grad_norm": 0.06168459728360176,
      "learning_rate": 0.0003728394205932248,
      "loss": 1.4809,
      "step": 13504
    },
    {
      "epoch": 0.6005068694144324,
      "grad_norm": 0.062344878911972046,
      "learning_rate": 0.00037269804415391726,
      "loss": 1.4763,
      "step": 13506
    },
    {
      "epoch": 0.6005957938731048,
      "grad_norm": 0.061719466000795364,
      "learning_rate": 0.00037255667859694173,
      "loss": 1.4798,
      "step": 13508
    },
    {
      "epoch": 0.6006847183317772,
      "grad_norm": 0.06244780495762825,
      "learning_rate": 0.00037241532393438313,
      "loss": 1.4775,
      "step": 13510
    },
    {
      "epoch": 0.6007736427904495,
      "grad_norm": 0.06072971224784851,
      "learning_rate": 0.00037227398017832485,
      "loss": 1.4808,
      "step": 13512
    },
    {
      "epoch": 0.6008625672491219,
      "grad_norm": 0.06084993854165077,
      "learning_rate": 0.0003721326473408495,
      "loss": 1.4764,
      "step": 13514
    },
    {
      "epoch": 0.6009514917077943,
      "grad_norm": 0.06384997069835663,
      "learning_rate": 0.00037199132543403916,
      "loss": 1.4758,
      "step": 13516
    },
    {
      "epoch": 0.6010404161664666,
      "grad_norm": 0.05961621552705765,
      "learning_rate": 0.0003718500144699743,
      "loss": 1.4846,
      "step": 13518
    },
    {
      "epoch": 0.6011293406251389,
      "grad_norm": 0.05951659008860588,
      "learning_rate": 0.0003717087144607348,
      "loss": 1.4813,
      "step": 13520
    },
    {
      "epoch": 0.6012182650838113,
      "grad_norm": 0.06181299686431885,
      "learning_rate": 0.00037156742541840005,
      "loss": 1.4786,
      "step": 13522
    },
    {
      "epoch": 0.6013071895424836,
      "grad_norm": 0.06189553067088127,
      "learning_rate": 0.00037142614735504755,
      "loss": 1.4674,
      "step": 13524
    },
    {
      "epoch": 0.601396114001156,
      "grad_norm": 0.06160914897918701,
      "learning_rate": 0.00037128488028275467,
      "loss": 1.4807,
      "step": 13526
    },
    {
      "epoch": 0.6014850384598284,
      "grad_norm": 0.05913588032126427,
      "learning_rate": 0.0003711436242135975,
      "loss": 1.4799,
      "step": 13528
    },
    {
      "epoch": 0.6015739629185007,
      "grad_norm": 0.060983993113040924,
      "learning_rate": 0.00037100237915965094,
      "loss": 1.4758,
      "step": 13530
    },
    {
      "epoch": 0.6016628873771731,
      "grad_norm": 0.06069378927350044,
      "learning_rate": 0.0003708611451329898,
      "loss": 1.4809,
      "step": 13532
    },
    {
      "epoch": 0.6017518118358455,
      "grad_norm": 0.061977144330739975,
      "learning_rate": 0.00037071992214568694,
      "loss": 1.4809,
      "step": 13534
    },
    {
      "epoch": 0.6018407362945178,
      "grad_norm": 0.062056753784418106,
      "learning_rate": 0.00037057871020981483,
      "loss": 1.4725,
      "step": 13536
    },
    {
      "epoch": 0.6019296607531902,
      "grad_norm": 0.0617976188659668,
      "learning_rate": 0.00037043750933744493,
      "loss": 1.4793,
      "step": 13538
    },
    {
      "epoch": 0.6020185852118626,
      "grad_norm": 0.06159581243991852,
      "learning_rate": 0.0003702963195406476,
      "loss": 1.4774,
      "step": 13540
    },
    {
      "epoch": 0.6021075096705348,
      "grad_norm": 0.0624043270945549,
      "learning_rate": 0.0003701551408314926,
      "loss": 1.4865,
      "step": 13542
    },
    {
      "epoch": 0.6021964341292072,
      "grad_norm": 0.06266829371452332,
      "learning_rate": 0.00037001397322204834,
      "loss": 1.4791,
      "step": 13544
    },
    {
      "epoch": 0.6022853585878796,
      "grad_norm": 0.061347655951976776,
      "learning_rate": 0.0003698728167243824,
      "loss": 1.4767,
      "step": 13546
    },
    {
      "epoch": 0.6023742830465519,
      "grad_norm": 0.05982671678066254,
      "learning_rate": 0.0003697316713505616,
      "loss": 1.4754,
      "step": 13548
    },
    {
      "epoch": 0.6024632075052243,
      "grad_norm": 0.06235219165682793,
      "learning_rate": 0.00036959053711265176,
      "loss": 1.4841,
      "step": 13550
    },
    {
      "epoch": 0.6025521319638967,
      "grad_norm": 0.06052573025226593,
      "learning_rate": 0.00036944941402271714,
      "loss": 1.4776,
      "step": 13552
    },
    {
      "epoch": 0.602641056422569,
      "grad_norm": 0.06316308677196503,
      "learning_rate": 0.00036930830209282227,
      "loss": 1.486,
      "step": 13554
    },
    {
      "epoch": 0.6027299808812414,
      "grad_norm": 0.06305599957704544,
      "learning_rate": 0.0003691672013350296,
      "loss": 1.477,
      "step": 13556
    },
    {
      "epoch": 0.6028189053399138,
      "grad_norm": 0.061367884278297424,
      "learning_rate": 0.00036902611176140097,
      "loss": 1.4807,
      "step": 13558
    },
    {
      "epoch": 0.6029078297985861,
      "grad_norm": 0.06356421858072281,
      "learning_rate": 0.00036888503338399766,
      "loss": 1.4872,
      "step": 13560
    },
    {
      "epoch": 0.6029967542572584,
      "grad_norm": 0.06272092461585999,
      "learning_rate": 0.0003687439662148793,
      "loss": 1.4741,
      "step": 13562
    },
    {
      "epoch": 0.6030856787159308,
      "grad_norm": 0.060393061488866806,
      "learning_rate": 0.0003686029102661053,
      "loss": 1.4814,
      "step": 13564
    },
    {
      "epoch": 0.6031746031746031,
      "grad_norm": 0.06250613927841187,
      "learning_rate": 0.0003684618655497336,
      "loss": 1.4819,
      "step": 13566
    },
    {
      "epoch": 0.6032635276332755,
      "grad_norm": 0.0645843967795372,
      "learning_rate": 0.00036832083207782114,
      "loss": 1.482,
      "step": 13568
    },
    {
      "epoch": 0.6033524520919479,
      "grad_norm": 0.06406543403863907,
      "learning_rate": 0.0003681798098624244,
      "loss": 1.4779,
      "step": 13570
    },
    {
      "epoch": 0.6034413765506202,
      "grad_norm": 0.06072346493601799,
      "learning_rate": 0.0003680387989155984,
      "loss": 1.4713,
      "step": 13572
    },
    {
      "epoch": 0.6035303010092926,
      "grad_norm": 0.06097434088587761,
      "learning_rate": 0.00036789779924939725,
      "loss": 1.4776,
      "step": 13574
    },
    {
      "epoch": 0.603619225467965,
      "grad_norm": 0.06126450002193451,
      "learning_rate": 0.00036775681087587454,
      "loss": 1.4791,
      "step": 13576
    },
    {
      "epoch": 0.6037081499266373,
      "grad_norm": 0.061803415417671204,
      "learning_rate": 0.0003676158338070823,
      "loss": 1.4737,
      "step": 13578
    },
    {
      "epoch": 0.6037970743853097,
      "grad_norm": 0.058394886553287506,
      "learning_rate": 0.0003674748680550721,
      "loss": 1.4767,
      "step": 13580
    },
    {
      "epoch": 0.6038859988439821,
      "grad_norm": 0.061498358845710754,
      "learning_rate": 0.0003673339136318943,
      "loss": 1.4785,
      "step": 13582
    },
    {
      "epoch": 0.6039749233026543,
      "grad_norm": 0.06015564873814583,
      "learning_rate": 0.0003671929705495979,
      "loss": 1.4798,
      "step": 13584
    },
    {
      "epoch": 0.6040638477613267,
      "grad_norm": 0.06150843948125839,
      "learning_rate": 0.0003670520388202319,
      "loss": 1.4846,
      "step": 13586
    },
    {
      "epoch": 0.6041527722199991,
      "grad_norm": 0.06126067414879799,
      "learning_rate": 0.0003669111184558435,
      "loss": 1.4828,
      "step": 13588
    },
    {
      "epoch": 0.6042416966786714,
      "grad_norm": 0.06068529933691025,
      "learning_rate": 0.00036677020946847895,
      "loss": 1.4826,
      "step": 13590
    },
    {
      "epoch": 0.6043306211373438,
      "grad_norm": 0.06273922324180603,
      "learning_rate": 0.00036662931187018423,
      "loss": 1.4812,
      "step": 13592
    },
    {
      "epoch": 0.6044195455960162,
      "grad_norm": 0.06314177811145782,
      "learning_rate": 0.00036648842567300356,
      "loss": 1.4776,
      "step": 13594
    },
    {
      "epoch": 0.6045084700546886,
      "grad_norm": 0.0634959414601326,
      "learning_rate": 0.00036634755088898064,
      "loss": 1.482,
      "step": 13596
    },
    {
      "epoch": 0.6045973945133609,
      "grad_norm": 0.06375706940889359,
      "learning_rate": 0.0003662066875301582,
      "loss": 1.4779,
      "step": 13598
    },
    {
      "epoch": 0.6046863189720333,
      "grad_norm": 0.0621272437274456,
      "learning_rate": 0.00036606583560857756,
      "loss": 1.4859,
      "step": 13600
    },
    {
      "epoch": 0.6047752434307057,
      "grad_norm": 0.0640636533498764,
      "learning_rate": 0.00036592499513627953,
      "loss": 1.4766,
      "step": 13602
    },
    {
      "epoch": 0.604864167889378,
      "grad_norm": 0.06891714781522751,
      "learning_rate": 0.00036578416612530396,
      "loss": 1.4821,
      "step": 13604
    },
    {
      "epoch": 0.6049530923480503,
      "grad_norm": 0.065804123878479,
      "learning_rate": 0.00036564334858768906,
      "loss": 1.4774,
      "step": 13606
    },
    {
      "epoch": 0.6050420168067226,
      "grad_norm": 0.06563190370798111,
      "learning_rate": 0.0003655025425354731,
      "loss": 1.4788,
      "step": 13608
    },
    {
      "epoch": 0.605130941265395,
      "grad_norm": 0.06128527969121933,
      "learning_rate": 0.00036536174798069217,
      "loss": 1.4766,
      "step": 13610
    },
    {
      "epoch": 0.6052198657240674,
      "grad_norm": 0.06118803471326828,
      "learning_rate": 0.0003652209649353827,
      "loss": 1.4826,
      "step": 13612
    },
    {
      "epoch": 0.6053087901827398,
      "grad_norm": 0.06206532567739487,
      "learning_rate": 0.00036508019341157896,
      "loss": 1.4774,
      "step": 13614
    },
    {
      "epoch": 0.6053977146414121,
      "grad_norm": 0.062083590775728226,
      "learning_rate": 0.00036493943342131473,
      "loss": 1.4813,
      "step": 13616
    },
    {
      "epoch": 0.6054866391000845,
      "grad_norm": 0.06275159120559692,
      "learning_rate": 0.0003647986849766231,
      "loss": 1.4792,
      "step": 13618
    },
    {
      "epoch": 0.6055755635587569,
      "grad_norm": 0.06044034659862518,
      "learning_rate": 0.0003646579480895357,
      "loss": 1.4774,
      "step": 13620
    },
    {
      "epoch": 0.6056644880174292,
      "grad_norm": 0.06259037554264069,
      "learning_rate": 0.00036451722277208314,
      "loss": 1.482,
      "step": 13622
    },
    {
      "epoch": 0.6057534124761016,
      "grad_norm": 0.0622219480574131,
      "learning_rate": 0.0003643765090362956,
      "loss": 1.4756,
      "step": 13624
    },
    {
      "epoch": 0.605842336934774,
      "grad_norm": 0.060978252440690994,
      "learning_rate": 0.0003642358068942017,
      "loss": 1.4794,
      "step": 13626
    },
    {
      "epoch": 0.6059312613934462,
      "grad_norm": 0.059897031635046005,
      "learning_rate": 0.00036409511635782923,
      "loss": 1.4817,
      "step": 13628
    },
    {
      "epoch": 0.6060201858521186,
      "grad_norm": 0.06172829493880272,
      "learning_rate": 0.00036395443743920524,
      "loss": 1.4788,
      "step": 13630
    },
    {
      "epoch": 0.606109110310791,
      "grad_norm": 0.06160893663764,
      "learning_rate": 0.0003638137701503554,
      "loss": 1.4734,
      "step": 13632
    },
    {
      "epoch": 0.6061980347694633,
      "grad_norm": 0.06013627350330353,
      "learning_rate": 0.0003636731145033047,
      "loss": 1.4766,
      "step": 13634
    },
    {
      "epoch": 0.6062869592281357,
      "grad_norm": 0.06238262727856636,
      "learning_rate": 0.0003635324705100771,
      "loss": 1.483,
      "step": 13636
    },
    {
      "epoch": 0.6063758836868081,
      "grad_norm": 0.06085238978266716,
      "learning_rate": 0.0003633918381826949,
      "loss": 1.479,
      "step": 13638
    },
    {
      "epoch": 0.6064648081454804,
      "grad_norm": 0.06417615711688995,
      "learning_rate": 0.0003632512175331808,
      "loss": 1.4799,
      "step": 13640
    },
    {
      "epoch": 0.6065537326041528,
      "grad_norm": 0.06320105493068695,
      "learning_rate": 0.0003631106085735551,
      "loss": 1.4779,
      "step": 13642
    },
    {
      "epoch": 0.6066426570628252,
      "grad_norm": 0.062168557196855545,
      "learning_rate": 0.00036297001131583776,
      "loss": 1.4769,
      "step": 13644
    },
    {
      "epoch": 0.6067315815214975,
      "grad_norm": 0.06036188453435898,
      "learning_rate": 0.00036282942577204783,
      "loss": 1.4753,
      "step": 13646
    },
    {
      "epoch": 0.6068205059801699,
      "grad_norm": 0.0608220137655735,
      "learning_rate": 0.00036268885195420294,
      "loss": 1.4779,
      "step": 13648
    },
    {
      "epoch": 0.6069094304388422,
      "grad_norm": 0.060881808400154114,
      "learning_rate": 0.0003625482898743202,
      "loss": 1.4739,
      "step": 13650
    },
    {
      "epoch": 0.6069983548975145,
      "grad_norm": 0.060534775257110596,
      "learning_rate": 0.0003624077395444154,
      "loss": 1.4814,
      "step": 13652
    },
    {
      "epoch": 0.6070872793561869,
      "grad_norm": 0.06239446997642517,
      "learning_rate": 0.00036226720097650323,
      "loss": 1.4804,
      "step": 13654
    },
    {
      "epoch": 0.6071762038148593,
      "grad_norm": 0.06479929387569427,
      "learning_rate": 0.0003621266741825978,
      "loss": 1.4788,
      "step": 13656
    },
    {
      "epoch": 0.6072651282735316,
      "grad_norm": 0.06032795459032059,
      "learning_rate": 0.000361986159174712,
      "loss": 1.4804,
      "step": 13658
    },
    {
      "epoch": 0.607354052732204,
      "grad_norm": 0.06258971989154816,
      "learning_rate": 0.0003618456559648572,
      "loss": 1.4784,
      "step": 13660
    },
    {
      "epoch": 0.6074429771908764,
      "grad_norm": 0.060947444289922714,
      "learning_rate": 0.0003617051645650448,
      "loss": 1.4767,
      "step": 13662
    },
    {
      "epoch": 0.6075319016495487,
      "grad_norm": 0.0637674406170845,
      "learning_rate": 0.0003615646849872841,
      "loss": 1.4773,
      "step": 13664
    },
    {
      "epoch": 0.6076208261082211,
      "grad_norm": 0.05858977884054184,
      "learning_rate": 0.00036142421724358456,
      "loss": 1.4761,
      "step": 13666
    },
    {
      "epoch": 0.6077097505668935,
      "grad_norm": 0.060713764280080795,
      "learning_rate": 0.0003612837613459534,
      "loss": 1.4735,
      "step": 13668
    },
    {
      "epoch": 0.6077986750255657,
      "grad_norm": 0.05963486433029175,
      "learning_rate": 0.0003611433173063976,
      "loss": 1.4801,
      "step": 13670
    },
    {
      "epoch": 0.6078875994842381,
      "grad_norm": 0.06113218143582344,
      "learning_rate": 0.00036100288513692304,
      "loss": 1.4722,
      "step": 13672
    },
    {
      "epoch": 0.6079765239429105,
      "grad_norm": 0.06065460667014122,
      "learning_rate": 0.00036086246484953445,
      "loss": 1.4768,
      "step": 13674
    },
    {
      "epoch": 0.6080654484015828,
      "grad_norm": 0.060706645250320435,
      "learning_rate": 0.00036072205645623536,
      "loss": 1.4783,
      "step": 13676
    },
    {
      "epoch": 0.6081543728602552,
      "grad_norm": 0.06271646171808243,
      "learning_rate": 0.00036058165996902876,
      "loss": 1.479,
      "step": 13678
    },
    {
      "epoch": 0.6082432973189276,
      "grad_norm": 0.06224247068166733,
      "learning_rate": 0.0003604412753999163,
      "loss": 1.4785,
      "step": 13680
    },
    {
      "epoch": 0.6083322217775999,
      "grad_norm": 0.06208979710936546,
      "learning_rate": 0.0003603009027608985,
      "loss": 1.4811,
      "step": 13682
    },
    {
      "epoch": 0.6084211462362723,
      "grad_norm": 0.06194361299276352,
      "learning_rate": 0.0003601605420639753,
      "loss": 1.4771,
      "step": 13684
    },
    {
      "epoch": 0.6085100706949447,
      "grad_norm": 0.06201132759451866,
      "learning_rate": 0.00036002019332114513,
      "loss": 1.4754,
      "step": 13686
    },
    {
      "epoch": 0.608598995153617,
      "grad_norm": 0.06057362258434296,
      "learning_rate": 0.0003598798565444058,
      "loss": 1.4791,
      "step": 13688
    },
    {
      "epoch": 0.6086879196122894,
      "grad_norm": 0.062178537249565125,
      "learning_rate": 0.0003597395317457538,
      "loss": 1.476,
      "step": 13690
    },
    {
      "epoch": 0.6087768440709617,
      "grad_norm": 0.06185256317257881,
      "learning_rate": 0.0003595992189371846,
      "loss": 1.4752,
      "step": 13692
    },
    {
      "epoch": 0.608865768529634,
      "grad_norm": 0.06281464546918869,
      "learning_rate": 0.0003594589181306931,
      "loss": 1.478,
      "step": 13694
    },
    {
      "epoch": 0.6089546929883064,
      "grad_norm": 0.060524832457304,
      "learning_rate": 0.00035931862933827254,
      "loss": 1.4829,
      "step": 13696
    },
    {
      "epoch": 0.6090436174469788,
      "grad_norm": 0.06258238852024078,
      "learning_rate": 0.0003591783525719154,
      "loss": 1.4736,
      "step": 13698
    },
    {
      "epoch": 0.6091325419056511,
      "grad_norm": 0.06145402789115906,
      "learning_rate": 0.00035903808784361335,
      "loss": 1.4815,
      "step": 13700
    },
    {
      "epoch": 0.6092214663643235,
      "grad_norm": 0.06079664081335068,
      "learning_rate": 0.0003588978351653567,
      "loss": 1.4756,
      "step": 13702
    },
    {
      "epoch": 0.6093103908229959,
      "grad_norm": 0.061961762607097626,
      "learning_rate": 0.000358757594549135,
      "loss": 1.4807,
      "step": 13704
    },
    {
      "epoch": 0.6093993152816682,
      "grad_norm": 0.06455682963132858,
      "learning_rate": 0.0003586173660069365,
      "loss": 1.4786,
      "step": 13706
    },
    {
      "epoch": 0.6094882397403406,
      "grad_norm": 0.06444698572158813,
      "learning_rate": 0.0003584771495507486,
      "loss": 1.4827,
      "step": 13708
    },
    {
      "epoch": 0.609577164199013,
      "grad_norm": 0.06271133571863174,
      "learning_rate": 0.00035833694519255764,
      "loss": 1.4749,
      "step": 13710
    },
    {
      "epoch": 0.6096660886576853,
      "grad_norm": 0.06196901574730873,
      "learning_rate": 0.00035819675294434904,
      "loss": 1.4734,
      "step": 13712
    },
    {
      "epoch": 0.6097550131163576,
      "grad_norm": 0.06276630610227585,
      "learning_rate": 0.0003580565728181066,
      "loss": 1.4832,
      "step": 13714
    },
    {
      "epoch": 0.60984393757503,
      "grad_norm": 0.06265407800674438,
      "learning_rate": 0.0003579164048258141,
      "loss": 1.4774,
      "step": 13716
    },
    {
      "epoch": 0.6099328620337023,
      "grad_norm": 0.06209501251578331,
      "learning_rate": 0.00035777624897945324,
      "loss": 1.4789,
      "step": 13718
    },
    {
      "epoch": 0.6100217864923747,
      "grad_norm": 0.06323768943548203,
      "learning_rate": 0.00035763610529100567,
      "loss": 1.4795,
      "step": 13720
    },
    {
      "epoch": 0.6101107109510471,
      "grad_norm": 0.06337448954582214,
      "learning_rate": 0.000357495973772451,
      "loss": 1.4814,
      "step": 13722
    },
    {
      "epoch": 0.6101996354097194,
      "grad_norm": 0.06219518184661865,
      "learning_rate": 0.00035735585443576846,
      "loss": 1.4698,
      "step": 13724
    },
    {
      "epoch": 0.6102885598683918,
      "grad_norm": 0.06129265949130058,
      "learning_rate": 0.00035721574729293616,
      "loss": 1.4807,
      "step": 13726
    },
    {
      "epoch": 0.6103774843270642,
      "grad_norm": 0.0608796663582325,
      "learning_rate": 0.00035707565235593104,
      "loss": 1.4743,
      "step": 13728
    },
    {
      "epoch": 0.6104664087857365,
      "grad_norm": 0.06377924978733063,
      "learning_rate": 0.0003569355696367289,
      "loss": 1.4794,
      "step": 13730
    },
    {
      "epoch": 0.6105553332444089,
      "grad_norm": 0.06089383736252785,
      "learning_rate": 0.0003567954991473048,
      "loss": 1.4803,
      "step": 13732
    },
    {
      "epoch": 0.6106442577030813,
      "grad_norm": 0.0606466569006443,
      "learning_rate": 0.00035665544089963244,
      "loss": 1.4818,
      "step": 13734
    },
    {
      "epoch": 0.6107331821617535,
      "grad_norm": 0.06015947833657265,
      "learning_rate": 0.0003565153949056849,
      "loss": 1.4785,
      "step": 13736
    },
    {
      "epoch": 0.6108221066204259,
      "grad_norm": 0.06213843822479248,
      "learning_rate": 0.00035637536117743364,
      "loss": 1.4752,
      "step": 13738
    },
    {
      "epoch": 0.6109110310790983,
      "grad_norm": 0.061907023191452026,
      "learning_rate": 0.0003562353397268492,
      "loss": 1.4774,
      "step": 13740
    },
    {
      "epoch": 0.6109999555377706,
      "grad_norm": 0.06074616312980652,
      "learning_rate": 0.0003560953305659018,
      "loss": 1.4726,
      "step": 13742
    },
    {
      "epoch": 0.611088879996443,
      "grad_norm": 0.06233498081564903,
      "learning_rate": 0.0003559553337065596,
      "loss": 1.4772,
      "step": 13744
    },
    {
      "epoch": 0.6111778044551154,
      "grad_norm": 0.06261789798736572,
      "learning_rate": 0.00035581534916079017,
      "loss": 1.4796,
      "step": 13746
    },
    {
      "epoch": 0.6112667289137877,
      "grad_norm": 0.06077142059803009,
      "learning_rate": 0.0003556753769405601,
      "loss": 1.4757,
      "step": 13748
    },
    {
      "epoch": 0.6113556533724601,
      "grad_norm": 0.05928738787770271,
      "learning_rate": 0.0003555354170578349,
      "loss": 1.4729,
      "step": 13750
    },
    {
      "epoch": 0.6114445778311325,
      "grad_norm": 0.06241046264767647,
      "learning_rate": 0.00035539546952457877,
      "loss": 1.4816,
      "step": 13752
    },
    {
      "epoch": 0.6115335022898049,
      "grad_norm": 0.06255488097667694,
      "learning_rate": 0.00035525553435275536,
      "loss": 1.4826,
      "step": 13754
    },
    {
      "epoch": 0.6116224267484772,
      "grad_norm": 0.06241544336080551,
      "learning_rate": 0.0003551156115543265,
      "loss": 1.4832,
      "step": 13756
    },
    {
      "epoch": 0.6117113512071495,
      "grad_norm": 0.0616549476981163,
      "learning_rate": 0.00035497570114125383,
      "loss": 1.481,
      "step": 13758
    },
    {
      "epoch": 0.6118002756658218,
      "grad_norm": 0.06216934695839882,
      "learning_rate": 0.0003548358031254973,
      "loss": 1.4804,
      "step": 13760
    },
    {
      "epoch": 0.6118892001244942,
      "grad_norm": 0.06092384457588196,
      "learning_rate": 0.000354695917519016,
      "loss": 1.48,
      "step": 13762
    },
    {
      "epoch": 0.6119781245831666,
      "grad_norm": 0.06199733540415764,
      "learning_rate": 0.000354556044333768,
      "loss": 1.4815,
      "step": 13764
    },
    {
      "epoch": 0.612067049041839,
      "grad_norm": 0.060262758284807205,
      "learning_rate": 0.00035441618358171046,
      "loss": 1.4774,
      "step": 13766
    },
    {
      "epoch": 0.6121559735005113,
      "grad_norm": 0.06175270304083824,
      "learning_rate": 0.00035427633527479875,
      "loss": 1.4805,
      "step": 13768
    },
    {
      "epoch": 0.6122448979591837,
      "grad_norm": 0.06414217501878738,
      "learning_rate": 0.00035413649942498836,
      "loss": 1.4717,
      "step": 13770
    },
    {
      "epoch": 0.612333822417856,
      "grad_norm": 0.06109687313437462,
      "learning_rate": 0.00035399667604423256,
      "loss": 1.4784,
      "step": 13772
    },
    {
      "epoch": 0.6124227468765284,
      "grad_norm": 0.06094755604863167,
      "learning_rate": 0.0003538568651444847,
      "loss": 1.481,
      "step": 13774
    },
    {
      "epoch": 0.6125116713352008,
      "grad_norm": 0.059903766959905624,
      "learning_rate": 0.00035371706673769575,
      "loss": 1.4742,
      "step": 13776
    },
    {
      "epoch": 0.6126005957938732,
      "grad_norm": 0.06042702868580818,
      "learning_rate": 0.0003535772808358166,
      "loss": 1.4756,
      "step": 13778
    },
    {
      "epoch": 0.6126895202525454,
      "grad_norm": 0.05981525033712387,
      "learning_rate": 0.00035343750745079683,
      "loss": 1.482,
      "step": 13780
    },
    {
      "epoch": 0.6127784447112178,
      "grad_norm": 0.060655247420072556,
      "learning_rate": 0.0003532977465945848,
      "loss": 1.4782,
      "step": 13782
    },
    {
      "epoch": 0.6128673691698902,
      "grad_norm": 0.06158040091395378,
      "learning_rate": 0.00035315799827912776,
      "loss": 1.4821,
      "step": 13784
    },
    {
      "epoch": 0.6129562936285625,
      "grad_norm": 0.06122865527868271,
      "learning_rate": 0.0003530182625163723,
      "loss": 1.4724,
      "step": 13786
    },
    {
      "epoch": 0.6130452180872349,
      "grad_norm": 0.061763033270835876,
      "learning_rate": 0.0003528785393182633,
      "loss": 1.4795,
      "step": 13788
    },
    {
      "epoch": 0.6131341425459073,
      "grad_norm": 0.06082763522863388,
      "learning_rate": 0.0003527388286967453,
      "loss": 1.4738,
      "step": 13790
    },
    {
      "epoch": 0.6132230670045796,
      "grad_norm": 0.061021216213703156,
      "learning_rate": 0.0003525991306637612,
      "loss": 1.4728,
      "step": 13792
    },
    {
      "epoch": 0.613311991463252,
      "grad_norm": 0.06356395781040192,
      "learning_rate": 0.00035245944523125275,
      "loss": 1.4708,
      "step": 13794
    },
    {
      "epoch": 0.6134009159219244,
      "grad_norm": 0.05992809683084488,
      "learning_rate": 0.00035231977241116137,
      "loss": 1.476,
      "step": 13796
    },
    {
      "epoch": 0.6134898403805967,
      "grad_norm": 0.06261426210403442,
      "learning_rate": 0.00035218011221542657,
      "loss": 1.477,
      "step": 13798
    },
    {
      "epoch": 0.613578764839269,
      "grad_norm": 0.06266981363296509,
      "learning_rate": 0.00035204046465598705,
      "loss": 1.4794,
      "step": 13800
    },
    {
      "epoch": 0.6136676892979414,
      "grad_norm": 0.06202618405222893,
      "learning_rate": 0.0003519008297447808,
      "loss": 1.4745,
      "step": 13802
    },
    {
      "epoch": 0.6137566137566137,
      "grad_norm": 0.06045003980398178,
      "learning_rate": 0.00035176120749374416,
      "loss": 1.4759,
      "step": 13804
    },
    {
      "epoch": 0.6138455382152861,
      "grad_norm": 0.06169283390045166,
      "learning_rate": 0.00035162159791481286,
      "loss": 1.4783,
      "step": 13806
    },
    {
      "epoch": 0.6139344626739585,
      "grad_norm": 0.06302846968173981,
      "learning_rate": 0.00035148200101992126,
      "loss": 1.4796,
      "step": 13808
    },
    {
      "epoch": 0.6140233871326308,
      "grad_norm": 0.06206449121236801,
      "learning_rate": 0.00035134241682100267,
      "loss": 1.4747,
      "step": 13810
    },
    {
      "epoch": 0.6141123115913032,
      "grad_norm": 0.06087673828005791,
      "learning_rate": 0.00035120284532998944,
      "loss": 1.4813,
      "step": 13812
    },
    {
      "epoch": 0.6142012360499756,
      "grad_norm": 0.06068653240799904,
      "learning_rate": 0.0003510632865588127,
      "loss": 1.4734,
      "step": 13814
    },
    {
      "epoch": 0.6142901605086479,
      "grad_norm": 0.05969279631972313,
      "learning_rate": 0.00035092374051940247,
      "loss": 1.4742,
      "step": 13816
    },
    {
      "epoch": 0.6143790849673203,
      "grad_norm": 0.06131194904446602,
      "learning_rate": 0.000350784207223688,
      "loss": 1.4765,
      "step": 13818
    },
    {
      "epoch": 0.6144680094259927,
      "grad_norm": 0.061875831335783005,
      "learning_rate": 0.0003506446866835972,
      "loss": 1.4741,
      "step": 13820
    },
    {
      "epoch": 0.6145569338846649,
      "grad_norm": 0.061578866094350815,
      "learning_rate": 0.00035050517891105636,
      "loss": 1.4751,
      "step": 13822
    },
    {
      "epoch": 0.6146458583433373,
      "grad_norm": 0.06131550669670105,
      "learning_rate": 0.000350365683917992,
      "loss": 1.4812,
      "step": 13824
    },
    {
      "epoch": 0.6147347828020097,
      "grad_norm": 0.06422258168458939,
      "learning_rate": 0.00035022620171632813,
      "loss": 1.4829,
      "step": 13826
    },
    {
      "epoch": 0.614823707260682,
      "grad_norm": 0.059642113745212555,
      "learning_rate": 0.0003500867323179888,
      "loss": 1.4753,
      "step": 13828
    },
    {
      "epoch": 0.6149126317193544,
      "grad_norm": 0.06245987117290497,
      "learning_rate": 0.00034994727573489615,
      "loss": 1.4773,
      "step": 13830
    },
    {
      "epoch": 0.6150015561780268,
      "grad_norm": 0.06119692325592041,
      "learning_rate": 0.0003498078319789716,
      "loss": 1.4806,
      "step": 13832
    },
    {
      "epoch": 0.6150904806366991,
      "grad_norm": 0.05997243896126747,
      "learning_rate": 0.0003496684010621354,
      "loss": 1.4792,
      "step": 13834
    },
    {
      "epoch": 0.6151794050953715,
      "grad_norm": 0.061276134103536606,
      "learning_rate": 0.00034952898299630685,
      "loss": 1.4761,
      "step": 13836
    },
    {
      "epoch": 0.6152683295540439,
      "grad_norm": 0.06231103837490082,
      "learning_rate": 0.0003493895777934039,
      "loss": 1.4709,
      "step": 13838
    },
    {
      "epoch": 0.6153572540127162,
      "grad_norm": 0.06226308271288872,
      "learning_rate": 0.0003492501854653436,
      "loss": 1.4754,
      "step": 13840
    },
    {
      "epoch": 0.6154461784713886,
      "grad_norm": 0.06134946271777153,
      "learning_rate": 0.00034911080602404164,
      "loss": 1.4771,
      "step": 13842
    },
    {
      "epoch": 0.6155351029300609,
      "grad_norm": 0.06262792646884918,
      "learning_rate": 0.0003489714394814131,
      "loss": 1.4744,
      "step": 13844
    },
    {
      "epoch": 0.6156240273887332,
      "grad_norm": 0.06038961932063103,
      "learning_rate": 0.00034883208584937154,
      "loss": 1.4786,
      "step": 13846
    },
    {
      "epoch": 0.6157129518474056,
      "grad_norm": 0.06001369282603264,
      "learning_rate": 0.0003486927451398292,
      "loss": 1.4753,
      "step": 13848
    },
    {
      "epoch": 0.615801876306078,
      "grad_norm": 0.06046263128519058,
      "learning_rate": 0.000348553417364698,
      "loss": 1.4684,
      "step": 13850
    },
    {
      "epoch": 0.6158908007647503,
      "grad_norm": 0.06046475097537041,
      "learning_rate": 0.0003484141025358881,
      "loss": 1.4787,
      "step": 13852
    },
    {
      "epoch": 0.6159797252234227,
      "grad_norm": 0.06241816654801369,
      "learning_rate": 0.0003482748006653086,
      "loss": 1.4803,
      "step": 13854
    },
    {
      "epoch": 0.6160686496820951,
      "grad_norm": 0.06165211275219917,
      "learning_rate": 0.0003481355117648678,
      "loss": 1.4798,
      "step": 13856
    },
    {
      "epoch": 0.6161575741407674,
      "grad_norm": 0.061339523643255234,
      "learning_rate": 0.00034799623584647266,
      "loss": 1.4842,
      "step": 13858
    },
    {
      "epoch": 0.6162464985994398,
      "grad_norm": 0.060382258147001266,
      "learning_rate": 0.00034785697292202925,
      "loss": 1.476,
      "step": 13860
    },
    {
      "epoch": 0.6163354230581122,
      "grad_norm": 0.05996260046958923,
      "learning_rate": 0.0003477177230034422,
      "loss": 1.4739,
      "step": 13862
    },
    {
      "epoch": 0.6164243475167845,
      "grad_norm": 0.06047176197171211,
      "learning_rate": 0.0003475784861026152,
      "loss": 1.4773,
      "step": 13864
    },
    {
      "epoch": 0.6165132719754568,
      "grad_norm": 0.061495665460824966,
      "learning_rate": 0.0003474392622314511,
      "loss": 1.4768,
      "step": 13866
    },
    {
      "epoch": 0.6166021964341292,
      "grad_norm": 0.06136169657111168,
      "learning_rate": 0.0003473000514018511,
      "loss": 1.4779,
      "step": 13868
    },
    {
      "epoch": 0.6166911208928015,
      "grad_norm": 0.06489419937133789,
      "learning_rate": 0.0003471608536257155,
      "loss": 1.476,
      "step": 13870
    },
    {
      "epoch": 0.6167800453514739,
      "grad_norm": 0.06050872802734375,
      "learning_rate": 0.0003470216689149438,
      "loss": 1.4778,
      "step": 13872
    },
    {
      "epoch": 0.6168689698101463,
      "grad_norm": 0.06324207037687302,
      "learning_rate": 0.00034688249728143415,
      "loss": 1.4782,
      "step": 13874
    },
    {
      "epoch": 0.6169578942688186,
      "grad_norm": 0.059641290456056595,
      "learning_rate": 0.0003467433387370831,
      "loss": 1.4749,
      "step": 13876
    },
    {
      "epoch": 0.617046818727491,
      "grad_norm": 0.06075885146856308,
      "learning_rate": 0.000346604193293787,
      "loss": 1.4795,
      "step": 13878
    },
    {
      "epoch": 0.6171357431861634,
      "grad_norm": 0.062138065695762634,
      "learning_rate": 0.00034646506096344025,
      "loss": 1.4735,
      "step": 13880
    },
    {
      "epoch": 0.6172246676448357,
      "grad_norm": 0.06034298986196518,
      "learning_rate": 0.000346325941757937,
      "loss": 1.4779,
      "step": 13882
    },
    {
      "epoch": 0.6173135921035081,
      "grad_norm": 0.06234195455908775,
      "learning_rate": 0.00034618683568916935,
      "loss": 1.4729,
      "step": 13884
    },
    {
      "epoch": 0.6174025165621805,
      "grad_norm": 0.06217122822999954,
      "learning_rate": 0.0003460477427690287,
      "loss": 1.4708,
      "step": 13886
    },
    {
      "epoch": 0.6174914410208527,
      "grad_norm": 0.05944179743528366,
      "learning_rate": 0.0003459086630094056,
      "loss": 1.4764,
      "step": 13888
    },
    {
      "epoch": 0.6175803654795251,
      "grad_norm": 0.06370831280946732,
      "learning_rate": 0.000345769596422189,
      "loss": 1.4839,
      "step": 13890
    },
    {
      "epoch": 0.6176692899381975,
      "grad_norm": 0.06237664818763733,
      "learning_rate": 0.0003456305430192669,
      "loss": 1.4778,
      "step": 13892
    },
    {
      "epoch": 0.6177582143968698,
      "grad_norm": 0.06287901103496552,
      "learning_rate": 0.00034549150281252633,
      "loss": 1.4808,
      "step": 13894
    },
    {
      "epoch": 0.6178471388555422,
      "grad_norm": 0.0613352507352829,
      "learning_rate": 0.0003453524758138529,
      "loss": 1.4807,
      "step": 13896
    },
    {
      "epoch": 0.6179360633142146,
      "grad_norm": 0.061771195381879807,
      "learning_rate": 0.0003452134620351315,
      "loss": 1.474,
      "step": 13898
    },
    {
      "epoch": 0.618024987772887,
      "grad_norm": 0.0651378482580185,
      "learning_rate": 0.0003450744614882455,
      "loss": 1.4709,
      "step": 13900
    },
    {
      "epoch": 0.6181139122315593,
      "grad_norm": 0.062017977237701416,
      "learning_rate": 0.00034493547418507704,
      "loss": 1.4781,
      "step": 13902
    },
    {
      "epoch": 0.6182028366902317,
      "grad_norm": 0.06174427643418312,
      "learning_rate": 0.0003447965001375079,
      "loss": 1.4819,
      "step": 13904
    },
    {
      "epoch": 0.618291761148904,
      "grad_norm": 0.06139495223760605,
      "learning_rate": 0.0003446575393574177,
      "loss": 1.4749,
      "step": 13906
    },
    {
      "epoch": 0.6183806856075764,
      "grad_norm": 0.05985395982861519,
      "learning_rate": 0.00034451859185668555,
      "loss": 1.4709,
      "step": 13908
    },
    {
      "epoch": 0.6184696100662487,
      "grad_norm": 0.05860113725066185,
      "learning_rate": 0.0003443796576471895,
      "loss": 1.4727,
      "step": 13910
    },
    {
      "epoch": 0.618558534524921,
      "grad_norm": 0.05987655743956566,
      "learning_rate": 0.0003442407367408059,
      "loss": 1.4781,
      "step": 13912
    },
    {
      "epoch": 0.6186474589835934,
      "grad_norm": 0.05981361120939255,
      "learning_rate": 0.0003441018291494108,
      "loss": 1.4788,
      "step": 13914
    },
    {
      "epoch": 0.6187363834422658,
      "grad_norm": 0.061126016080379486,
      "learning_rate": 0.00034396293488487824,
      "loss": 1.4694,
      "step": 13916
    },
    {
      "epoch": 0.6188253079009381,
      "grad_norm": 0.05968809127807617,
      "learning_rate": 0.00034382405395908157,
      "loss": 1.4756,
      "step": 13918
    },
    {
      "epoch": 0.6189142323596105,
      "grad_norm": 0.05997713655233383,
      "learning_rate": 0.0003436851863838931,
      "loss": 1.4803,
      "step": 13920
    },
    {
      "epoch": 0.6190031568182829,
      "grad_norm": 0.06370899826288223,
      "learning_rate": 0.0003435463321711838,
      "loss": 1.4788,
      "step": 13922
    },
    {
      "epoch": 0.6190920812769553,
      "grad_norm": 0.06034357100725174,
      "learning_rate": 0.00034340749133282333,
      "loss": 1.4737,
      "step": 13924
    },
    {
      "epoch": 0.6191810057356276,
      "grad_norm": 0.0625743493437767,
      "learning_rate": 0.0003432686638806807,
      "loss": 1.4807,
      "step": 13926
    },
    {
      "epoch": 0.6192699301943,
      "grad_norm": 0.05968970060348511,
      "learning_rate": 0.0003431298498266233,
      "loss": 1.4737,
      "step": 13928
    },
    {
      "epoch": 0.6193588546529722,
      "grad_norm": 0.0622946061193943,
      "learning_rate": 0.00034299104918251777,
      "loss": 1.4767,
      "step": 13930
    },
    {
      "epoch": 0.6194477791116446,
      "grad_norm": 0.060472749173641205,
      "learning_rate": 0.0003428522619602293,
      "loss": 1.4727,
      "step": 13932
    },
    {
      "epoch": 0.619536703570317,
      "grad_norm": 0.06135667487978935,
      "learning_rate": 0.00034271348817162175,
      "loss": 1.4727,
      "step": 13934
    },
    {
      "epoch": 0.6196256280289894,
      "grad_norm": 0.06154874339699745,
      "learning_rate": 0.0003425747278285587,
      "loss": 1.4723,
      "step": 13936
    },
    {
      "epoch": 0.6197145524876617,
      "grad_norm": 0.061515383422374725,
      "learning_rate": 0.0003424359809429015,
      "loss": 1.4735,
      "step": 13938
    },
    {
      "epoch": 0.6198034769463341,
      "grad_norm": 0.06175931170582771,
      "learning_rate": 0.00034229724752651093,
      "loss": 1.4762,
      "step": 13940
    },
    {
      "epoch": 0.6198924014050065,
      "grad_norm": 0.06058594584465027,
      "learning_rate": 0.0003421585275912467,
      "loss": 1.4732,
      "step": 13942
    },
    {
      "epoch": 0.6199813258636788,
      "grad_norm": 0.06232278421521187,
      "learning_rate": 0.0003420198211489671,
      "loss": 1.4754,
      "step": 13944
    },
    {
      "epoch": 0.6200702503223512,
      "grad_norm": 0.061653584241867065,
      "learning_rate": 0.0003418811282115294,
      "loss": 1.4707,
      "step": 13946
    },
    {
      "epoch": 0.6201591747810236,
      "grad_norm": 0.061312127858400345,
      "learning_rate": 0.00034174244879078966,
      "loss": 1.4743,
      "step": 13948
    },
    {
      "epoch": 0.6202480992396959,
      "grad_norm": 0.06045181304216385,
      "learning_rate": 0.0003416037828986027,
      "loss": 1.4723,
      "step": 13950
    },
    {
      "epoch": 0.6203370236983682,
      "grad_norm": 0.06308136880397797,
      "learning_rate": 0.0003414651305468226,
      "loss": 1.4732,
      "step": 13952
    },
    {
      "epoch": 0.6204259481570406,
      "grad_norm": 0.060169022530317307,
      "learning_rate": 0.0003413264917473018,
      "loss": 1.4769,
      "step": 13954
    },
    {
      "epoch": 0.6205148726157129,
      "grad_norm": 0.0620846189558506,
      "learning_rate": 0.0003411878665118915,
      "loss": 1.4735,
      "step": 13956
    },
    {
      "epoch": 0.6206037970743853,
      "grad_norm": 0.0605727918446064,
      "learning_rate": 0.00034104925485244254,
      "loss": 1.4743,
      "step": 13958
    },
    {
      "epoch": 0.6206927215330577,
      "grad_norm": 0.06262971460819244,
      "learning_rate": 0.00034091065678080356,
      "loss": 1.4758,
      "step": 13960
    },
    {
      "epoch": 0.62078164599173,
      "grad_norm": 0.06452898681163788,
      "learning_rate": 0.0003407720723088227,
      "loss": 1.4752,
      "step": 13962
    },
    {
      "epoch": 0.6208705704504024,
      "grad_norm": 0.05955608934164047,
      "learning_rate": 0.00034063350144834684,
      "loss": 1.4758,
      "step": 13964
    },
    {
      "epoch": 0.6209594949090748,
      "grad_norm": 0.06136883795261383,
      "learning_rate": 0.0003404949442112215,
      "loss": 1.4781,
      "step": 13966
    },
    {
      "epoch": 0.6210484193677471,
      "grad_norm": 0.0601998008787632,
      "learning_rate": 0.0003403564006092914,
      "loss": 1.4796,
      "step": 13968
    },
    {
      "epoch": 0.6211373438264195,
      "grad_norm": 0.060754548758268356,
      "learning_rate": 0.0003402178706543996,
      "loss": 1.478,
      "step": 13970
    },
    {
      "epoch": 0.6212262682850919,
      "grad_norm": 0.06004665046930313,
      "learning_rate": 0.0003400793543583883,
      "loss": 1.4739,
      "step": 13972
    },
    {
      "epoch": 0.6213151927437641,
      "grad_norm": 0.06137534976005554,
      "learning_rate": 0.0003399408517330987,
      "loss": 1.4741,
      "step": 13974
    },
    {
      "epoch": 0.6214041172024365,
      "grad_norm": 0.06013936921954155,
      "learning_rate": 0.00033980236279037033,
      "loss": 1.4734,
      "step": 13976
    },
    {
      "epoch": 0.6214930416611089,
      "grad_norm": 0.06054317578673363,
      "learning_rate": 0.00033966388754204195,
      "loss": 1.4776,
      "step": 13978
    },
    {
      "epoch": 0.6215819661197812,
      "grad_norm": 0.06105973199009895,
      "learning_rate": 0.00033952542599995116,
      "loss": 1.4717,
      "step": 13980
    },
    {
      "epoch": 0.6216708905784536,
      "grad_norm": 0.05983678251504898,
      "learning_rate": 0.000339386978175934,
      "loss": 1.4794,
      "step": 13982
    },
    {
      "epoch": 0.621759815037126,
      "grad_norm": 0.06286296248435974,
      "learning_rate": 0.0003392485440818259,
      "loss": 1.4766,
      "step": 13984
    },
    {
      "epoch": 0.6218487394957983,
      "grad_norm": 0.05888976529240608,
      "learning_rate": 0.0003391101237294606,
      "loss": 1.4761,
      "step": 13986
    },
    {
      "epoch": 0.6219376639544707,
      "grad_norm": 0.061478178948163986,
      "learning_rate": 0.00033897171713067085,
      "loss": 1.4737,
      "step": 13988
    },
    {
      "epoch": 0.6220265884131431,
      "grad_norm": 0.06167341768741608,
      "learning_rate": 0.0003388333242972885,
      "loss": 1.4782,
      "step": 13990
    },
    {
      "epoch": 0.6221155128718154,
      "grad_norm": 0.059394970536231995,
      "learning_rate": 0.0003386949452411438,
      "loss": 1.473,
      "step": 13992
    },
    {
      "epoch": 0.6222044373304878,
      "grad_norm": 0.06255074590444565,
      "learning_rate": 0.00033855657997406585,
      "loss": 1.4779,
      "step": 13994
    },
    {
      "epoch": 0.6222933617891601,
      "grad_norm": 0.05998929962515831,
      "learning_rate": 0.0003384182285078831,
      "loss": 1.4753,
      "step": 13996
    },
    {
      "epoch": 0.6223822862478324,
      "grad_norm": 0.06142527610063553,
      "learning_rate": 0.0003382798908544221,
      "loss": 1.4754,
      "step": 13998
    },
    {
      "epoch": 0.6224712107065048,
      "grad_norm": 0.06191844493150711,
      "learning_rate": 0.00033814156702550877,
      "loss": 1.4789,
      "step": 14000
    },
    {
      "epoch": 0.6224712107065048,
      "eval_loss": 1.460575819015503,
      "eval_runtime": 12.4458,
      "eval_samples_per_second": 555.207,
      "eval_steps_per_second": 69.421,
      "step": 14000
    },
    {
      "epoch": 0.6225601351651772,
      "grad_norm": 0.061490852385759354,
      "learning_rate": 0.00033800325703296763,
      "loss": 1.4737,
      "step": 14002
    },
    {
      "epoch": 0.6226490596238495,
      "grad_norm": 0.06254355609416962,
      "learning_rate": 0.0003378649608886218,
      "loss": 1.4782,
      "step": 14004
    },
    {
      "epoch": 0.6227379840825219,
      "grad_norm": 0.06114278733730316,
      "learning_rate": 0.0003377266786042938,
      "loss": 1.4775,
      "step": 14006
    },
    {
      "epoch": 0.6228269085411943,
      "grad_norm": 0.06295786052942276,
      "learning_rate": 0.0003375884101918045,
      "loss": 1.4736,
      "step": 14008
    },
    {
      "epoch": 0.6229158329998666,
      "grad_norm": 0.0627705305814743,
      "learning_rate": 0.00033745015566297333,
      "loss": 1.4771,
      "step": 14010
    },
    {
      "epoch": 0.623004757458539,
      "grad_norm": 0.06055225431919098,
      "learning_rate": 0.0003373119150296195,
      "loss": 1.4748,
      "step": 14012
    },
    {
      "epoch": 0.6230936819172114,
      "grad_norm": 0.06257602572441101,
      "learning_rate": 0.00033717368830356,
      "loss": 1.4773,
      "step": 14014
    },
    {
      "epoch": 0.6231826063758837,
      "grad_norm": 0.06255099177360535,
      "learning_rate": 0.00033703547549661107,
      "loss": 1.4767,
      "step": 14016
    },
    {
      "epoch": 0.623271530834556,
      "grad_norm": 0.06176856905221939,
      "learning_rate": 0.000336897276620588,
      "loss": 1.4765,
      "step": 14018
    },
    {
      "epoch": 0.6233604552932284,
      "grad_norm": 0.060302965342998505,
      "learning_rate": 0.00033675909168730435,
      "loss": 1.4745,
      "step": 14020
    },
    {
      "epoch": 0.6234493797519007,
      "grad_norm": 0.05996936559677124,
      "learning_rate": 0.00033662092070857295,
      "loss": 1.4733,
      "step": 14022
    },
    {
      "epoch": 0.6235383042105731,
      "grad_norm": 0.060551125556230545,
      "learning_rate": 0.0003364827636962053,
      "loss": 1.4816,
      "step": 14024
    },
    {
      "epoch": 0.6236272286692455,
      "grad_norm": 0.06198984757065773,
      "learning_rate": 0.0003363446206620116,
      "loss": 1.4746,
      "step": 14026
    },
    {
      "epoch": 0.6237161531279178,
      "grad_norm": 0.060537002980709076,
      "learning_rate": 0.0003362064916178009,
      "loss": 1.475,
      "step": 14028
    },
    {
      "epoch": 0.6238050775865902,
      "grad_norm": 0.05999968200922012,
      "learning_rate": 0.0003360683765753813,
      "loss": 1.4746,
      "step": 14030
    },
    {
      "epoch": 0.6238940020452626,
      "grad_norm": 0.0614023394882679,
      "learning_rate": 0.00033593027554655895,
      "loss": 1.4797,
      "step": 14032
    },
    {
      "epoch": 0.6239829265039349,
      "grad_norm": 0.061192482709884644,
      "learning_rate": 0.00033579218854314,
      "loss": 1.4771,
      "step": 14034
    },
    {
      "epoch": 0.6240718509626073,
      "grad_norm": 0.059598758816719055,
      "learning_rate": 0.0003356541155769281,
      "loss": 1.4795,
      "step": 14036
    },
    {
      "epoch": 0.6241607754212796,
      "grad_norm": 0.061559103429317474,
      "learning_rate": 0.00033551605665972694,
      "loss": 1.4822,
      "step": 14038
    },
    {
      "epoch": 0.6242496998799519,
      "grad_norm": 0.06176281347870827,
      "learning_rate": 0.00033537801180333804,
      "loss": 1.4763,
      "step": 14040
    },
    {
      "epoch": 0.6243386243386243,
      "grad_norm": 0.060048338025808334,
      "learning_rate": 0.00033523998101956206,
      "loss": 1.4764,
      "step": 14042
    },
    {
      "epoch": 0.6244275487972967,
      "grad_norm": 0.06090456247329712,
      "learning_rate": 0.0003351019643201987,
      "loss": 1.4757,
      "step": 14044
    },
    {
      "epoch": 0.624516473255969,
      "grad_norm": 0.061848804354667664,
      "learning_rate": 0.00033496396171704603,
      "loss": 1.4783,
      "step": 14046
    },
    {
      "epoch": 0.6246053977146414,
      "grad_norm": 0.06011731177568436,
      "learning_rate": 0.0003348259732219011,
      "loss": 1.4753,
      "step": 14048
    },
    {
      "epoch": 0.6246943221733138,
      "grad_norm": 0.06071808561682701,
      "learning_rate": 0.00033468799884656006,
      "loss": 1.4786,
      "step": 14050
    },
    {
      "epoch": 0.6247832466319861,
      "grad_norm": 0.06180998310446739,
      "learning_rate": 0.0003345500386028172,
      "loss": 1.4742,
      "step": 14052
    },
    {
      "epoch": 0.6248721710906585,
      "grad_norm": 0.0607328861951828,
      "learning_rate": 0.00033441209250246627,
      "loss": 1.4755,
      "step": 14054
    },
    {
      "epoch": 0.6249610955493309,
      "grad_norm": 0.060531482100486755,
      "learning_rate": 0.00033427416055729946,
      "loss": 1.4702,
      "step": 14056
    },
    {
      "epoch": 0.6250500200080032,
      "grad_norm": 0.06345600634813309,
      "learning_rate": 0.0003341362427791076,
      "loss": 1.4709,
      "step": 14058
    },
    {
      "epoch": 0.6251389444666755,
      "grad_norm": 0.06127747520804405,
      "learning_rate": 0.00033399833917968077,
      "loss": 1.4732,
      "step": 14060
    },
    {
      "epoch": 0.6252278689253479,
      "grad_norm": 0.06288928538560867,
      "learning_rate": 0.0003338604497708075,
      "loss": 1.4748,
      "step": 14062
    },
    {
      "epoch": 0.6253167933840202,
      "grad_norm": 0.06117086485028267,
      "learning_rate": 0.0003337225745642749,
      "loss": 1.4751,
      "step": 14064
    },
    {
      "epoch": 0.6254057178426926,
      "grad_norm": 0.06286124885082245,
      "learning_rate": 0.0003335847135718696,
      "loss": 1.4751,
      "step": 14066
    },
    {
      "epoch": 0.625494642301365,
      "grad_norm": 0.061294663697481155,
      "learning_rate": 0.00033344686680537633,
      "loss": 1.4768,
      "step": 14068
    },
    {
      "epoch": 0.6255835667600373,
      "grad_norm": 0.05968429148197174,
      "learning_rate": 0.00033330903427657876,
      "loss": 1.4799,
      "step": 14070
    },
    {
      "epoch": 0.6256724912187097,
      "grad_norm": 0.06150318309664726,
      "learning_rate": 0.0003331712159972596,
      "loss": 1.4739,
      "step": 14072
    },
    {
      "epoch": 0.6257614156773821,
      "grad_norm": 0.0625072494149208,
      "learning_rate": 0.00033303341197919994,
      "loss": 1.4786,
      "step": 14074
    },
    {
      "epoch": 0.6258503401360545,
      "grad_norm": 0.06237860396504402,
      "learning_rate": 0.0003328956222341802,
      "loss": 1.4776,
      "step": 14076
    },
    {
      "epoch": 0.6259392645947268,
      "grad_norm": 0.0603688582777977,
      "learning_rate": 0.00033275784677397905,
      "loss": 1.4719,
      "step": 14078
    },
    {
      "epoch": 0.6260281890533992,
      "grad_norm": 0.06012796610593796,
      "learning_rate": 0.00033262008561037404,
      "loss": 1.4764,
      "step": 14080
    },
    {
      "epoch": 0.6261171135120714,
      "grad_norm": 0.06131792068481445,
      "learning_rate": 0.0003324823387551419,
      "loss": 1.473,
      "step": 14082
    },
    {
      "epoch": 0.6262060379707438,
      "grad_norm": 0.06047110632061958,
      "learning_rate": 0.0003323446062200577,
      "loss": 1.4762,
      "step": 14084
    },
    {
      "epoch": 0.6262949624294162,
      "grad_norm": 0.06241205334663391,
      "learning_rate": 0.0003322068880168951,
      "loss": 1.4782,
      "step": 14086
    },
    {
      "epoch": 0.6263838868880885,
      "grad_norm": 0.0601295605301857,
      "learning_rate": 0.0003320691841574275,
      "loss": 1.4751,
      "step": 14088
    },
    {
      "epoch": 0.6264728113467609,
      "grad_norm": 0.06591775268316269,
      "learning_rate": 0.00033193149465342574,
      "loss": 1.4799,
      "step": 14090
    },
    {
      "epoch": 0.6265617358054333,
      "grad_norm": 0.06100832670927048,
      "learning_rate": 0.0003317938195166608,
      "loss": 1.4682,
      "step": 14092
    },
    {
      "epoch": 0.6266506602641057,
      "grad_norm": 0.0626838430762291,
      "learning_rate": 0.0003316561587589013,
      "loss": 1.4759,
      "step": 14094
    },
    {
      "epoch": 0.626739584722778,
      "grad_norm": 0.060527194291353226,
      "learning_rate": 0.00033151851239191513,
      "loss": 1.4687,
      "step": 14096
    },
    {
      "epoch": 0.6268285091814504,
      "grad_norm": 0.06166524067521095,
      "learning_rate": 0.00033138088042746904,
      "loss": 1.4758,
      "step": 14098
    },
    {
      "epoch": 0.6269174336401228,
      "grad_norm": 0.06203189119696617,
      "learning_rate": 0.00033124326287732847,
      "loss": 1.4757,
      "step": 14100
    },
    {
      "epoch": 0.6270063580987951,
      "grad_norm": 0.06214692071080208,
      "learning_rate": 0.00033110565975325734,
      "loss": 1.4731,
      "step": 14102
    },
    {
      "epoch": 0.6270952825574674,
      "grad_norm": 0.06142725050449371,
      "learning_rate": 0.00033096807106701876,
      "loss": 1.4802,
      "step": 14104
    },
    {
      "epoch": 0.6271842070161397,
      "grad_norm": 0.06279194355010986,
      "learning_rate": 0.00033083049683037437,
      "loss": 1.4776,
      "step": 14106
    },
    {
      "epoch": 0.6272731314748121,
      "grad_norm": 0.06059839949011803,
      "learning_rate": 0.0003306929370550846,
      "loss": 1.4767,
      "step": 14108
    },
    {
      "epoch": 0.6273620559334845,
      "grad_norm": 0.06027147173881531,
      "learning_rate": 0.0003305553917529087,
      "loss": 1.4788,
      "step": 14110
    },
    {
      "epoch": 0.6274509803921569,
      "grad_norm": 0.06027664244174957,
      "learning_rate": 0.0003304178609356046,
      "loss": 1.4724,
      "step": 14112
    },
    {
      "epoch": 0.6275399048508292,
      "grad_norm": 0.0626441091299057,
      "learning_rate": 0.00033028034461492917,
      "loss": 1.4738,
      "step": 14114
    },
    {
      "epoch": 0.6276288293095016,
      "grad_norm": 0.060072094202041626,
      "learning_rate": 0.00033014284280263786,
      "loss": 1.4736,
      "step": 14116
    },
    {
      "epoch": 0.627717753768174,
      "grad_norm": 0.059987835586071014,
      "learning_rate": 0.0003300053555104846,
      "loss": 1.4746,
      "step": 14118
    },
    {
      "epoch": 0.6278066782268463,
      "grad_norm": 0.06026008725166321,
      "learning_rate": 0.000329867882750223,
      "loss": 1.4749,
      "step": 14120
    },
    {
      "epoch": 0.6278956026855187,
      "grad_norm": 0.060853175818920135,
      "learning_rate": 0.00032973042453360426,
      "loss": 1.4706,
      "step": 14122
    },
    {
      "epoch": 0.6279845271441911,
      "grad_norm": 0.06199175864458084,
      "learning_rate": 0.00032959298087237944,
      "loss": 1.4794,
      "step": 14124
    },
    {
      "epoch": 0.6280734516028633,
      "grad_norm": 0.058771684765815735,
      "learning_rate": 0.00032945555177829746,
      "loss": 1.4797,
      "step": 14126
    },
    {
      "epoch": 0.6281623760615357,
      "grad_norm": 0.05990970879793167,
      "learning_rate": 0.0003293181372631064,
      "loss": 1.4769,
      "step": 14128
    },
    {
      "epoch": 0.6282513005202081,
      "grad_norm": 0.059724360704422,
      "learning_rate": 0.00032918073733855327,
      "loss": 1.4791,
      "step": 14130
    },
    {
      "epoch": 0.6283402249788804,
      "grad_norm": 0.060740843415260315,
      "learning_rate": 0.0003290433520163835,
      "loss": 1.4779,
      "step": 14132
    },
    {
      "epoch": 0.6284291494375528,
      "grad_norm": 0.060699593275785446,
      "learning_rate": 0.00032890598130834117,
      "loss": 1.4727,
      "step": 14134
    },
    {
      "epoch": 0.6285180738962252,
      "grad_norm": 0.061070021241903305,
      "learning_rate": 0.0003287686252261697,
      "loss": 1.4799,
      "step": 14136
    },
    {
      "epoch": 0.6286069983548975,
      "grad_norm": 0.06295224279165268,
      "learning_rate": 0.0003286312837816109,
      "loss": 1.4739,
      "step": 14138
    },
    {
      "epoch": 0.6286959228135699,
      "grad_norm": 0.06164375692605972,
      "learning_rate": 0.00032849395698640485,
      "loss": 1.4738,
      "step": 14140
    },
    {
      "epoch": 0.6287848472722423,
      "grad_norm": 0.06068945676088333,
      "learning_rate": 0.0003283566448522914,
      "loss": 1.4754,
      "step": 14142
    },
    {
      "epoch": 0.6288737717309146,
      "grad_norm": 0.06089947745203972,
      "learning_rate": 0.0003282193473910081,
      "loss": 1.4776,
      "step": 14144
    },
    {
      "epoch": 0.628962696189587,
      "grad_norm": 0.06197715550661087,
      "learning_rate": 0.00032808206461429226,
      "loss": 1.4781,
      "step": 14146
    },
    {
      "epoch": 0.6290516206482593,
      "grad_norm": 0.06002449244260788,
      "learning_rate": 0.00032794479653387905,
      "loss": 1.4751,
      "step": 14148
    },
    {
      "epoch": 0.6291405451069316,
      "grad_norm": 0.061960477381944656,
      "learning_rate": 0.00032780754316150276,
      "loss": 1.4709,
      "step": 14150
    },
    {
      "epoch": 0.629229469565604,
      "grad_norm": 0.06243150681257248,
      "learning_rate": 0.0003276703045088966,
      "loss": 1.4719,
      "step": 14152
    },
    {
      "epoch": 0.6293183940242764,
      "grad_norm": 0.06339258700609207,
      "learning_rate": 0.0003275330805877923,
      "loss": 1.4753,
      "step": 14154
    },
    {
      "epoch": 0.6294073184829487,
      "grad_norm": 0.06161705031991005,
      "learning_rate": 0.00032739587140992015,
      "loss": 1.4731,
      "step": 14156
    },
    {
      "epoch": 0.6294962429416211,
      "grad_norm": 0.06279416382312775,
      "learning_rate": 0.0003272586769870097,
      "loss": 1.4711,
      "step": 14158
    },
    {
      "epoch": 0.6295851674002935,
      "grad_norm": 0.06209520995616913,
      "learning_rate": 0.0003271214973307887,
      "loss": 1.4711,
      "step": 14160
    },
    {
      "epoch": 0.6296740918589658,
      "grad_norm": 0.06087113171815872,
      "learning_rate": 0.000326984332452984,
      "loss": 1.4769,
      "step": 14162
    },
    {
      "epoch": 0.6297630163176382,
      "grad_norm": 0.06049448251724243,
      "learning_rate": 0.0003268471823653211,
      "loss": 1.4766,
      "step": 14164
    },
    {
      "epoch": 0.6298519407763106,
      "grad_norm": 0.06183711439371109,
      "learning_rate": 0.0003267100470795239,
      "loss": 1.4719,
      "step": 14166
    },
    {
      "epoch": 0.6299408652349828,
      "grad_norm": 0.06014047563076019,
      "learning_rate": 0.0003265729266073157,
      "loss": 1.4793,
      "step": 14168
    },
    {
      "epoch": 0.6300297896936552,
      "grad_norm": 0.06054830551147461,
      "learning_rate": 0.00032643582096041806,
      "loss": 1.4755,
      "step": 14170
    },
    {
      "epoch": 0.6301187141523276,
      "grad_norm": 0.05997813493013382,
      "learning_rate": 0.00032629873015055106,
      "loss": 1.4747,
      "step": 14172
    },
    {
      "epoch": 0.6302076386109999,
      "grad_norm": 0.06141214445233345,
      "learning_rate": 0.0003261616541894344,
      "loss": 1.4715,
      "step": 14174
    },
    {
      "epoch": 0.6302965630696723,
      "grad_norm": 0.060470618307590485,
      "learning_rate": 0.0003260245930887853,
      "loss": 1.4731,
      "step": 14176
    },
    {
      "epoch": 0.6303854875283447,
      "grad_norm": 0.06036084517836571,
      "learning_rate": 0.00032588754686032093,
      "loss": 1.4774,
      "step": 14178
    },
    {
      "epoch": 0.630474411987017,
      "grad_norm": 0.06185087189078331,
      "learning_rate": 0.0003257505155157563,
      "loss": 1.475,
      "step": 14180
    },
    {
      "epoch": 0.6305633364456894,
      "grad_norm": 0.0611613430082798,
      "learning_rate": 0.0003256134990668054,
      "loss": 1.4794,
      "step": 14182
    },
    {
      "epoch": 0.6306522609043618,
      "grad_norm": 0.060027047991752625,
      "learning_rate": 0.00032547649752518115,
      "loss": 1.4753,
      "step": 14184
    },
    {
      "epoch": 0.6307411853630341,
      "grad_norm": 0.06171073764562607,
      "learning_rate": 0.000325339510902595,
      "loss": 1.4729,
      "step": 14186
    },
    {
      "epoch": 0.6308301098217065,
      "grad_norm": 0.06028326600790024,
      "learning_rate": 0.00032520253921075707,
      "loss": 1.4784,
      "step": 14188
    },
    {
      "epoch": 0.6309190342803788,
      "grad_norm": 0.06145615503191948,
      "learning_rate": 0.0003250655824613765,
      "loss": 1.4791,
      "step": 14190
    },
    {
      "epoch": 0.6310079587390511,
      "grad_norm": 0.0612344890832901,
      "learning_rate": 0.00032492864066616095,
      "loss": 1.4732,
      "step": 14192
    },
    {
      "epoch": 0.6310968831977235,
      "grad_norm": 0.0602327398955822,
      "learning_rate": 0.00032479171383681637,
      "loss": 1.4745,
      "step": 14194
    },
    {
      "epoch": 0.6311858076563959,
      "grad_norm": 0.062162987887859344,
      "learning_rate": 0.0003246548019850486,
      "loss": 1.4775,
      "step": 14196
    },
    {
      "epoch": 0.6312747321150682,
      "grad_norm": 0.06073742359876633,
      "learning_rate": 0.00032451790512256075,
      "loss": 1.4777,
      "step": 14198
    },
    {
      "epoch": 0.6313636565737406,
      "grad_norm": 0.06158788129687309,
      "learning_rate": 0.0003243810232610559,
      "loss": 1.4747,
      "step": 14200
    },
    {
      "epoch": 0.631452581032413,
      "grad_norm": 0.06189347803592682,
      "learning_rate": 0.00032424415641223517,
      "loss": 1.4696,
      "step": 14202
    },
    {
      "epoch": 0.6315415054910853,
      "grad_norm": 0.06252620369195938,
      "learning_rate": 0.00032410730458779825,
      "loss": 1.472,
      "step": 14204
    },
    {
      "epoch": 0.6316304299497577,
      "grad_norm": 0.06133527681231499,
      "learning_rate": 0.0003239704677994442,
      "loss": 1.4734,
      "step": 14206
    },
    {
      "epoch": 0.6317193544084301,
      "grad_norm": 0.06081013008952141,
      "learning_rate": 0.00032383364605887034,
      "loss": 1.4778,
      "step": 14208
    },
    {
      "epoch": 0.6318082788671024,
      "grad_norm": 0.061107292771339417,
      "learning_rate": 0.00032369683937777264,
      "loss": 1.4686,
      "step": 14210
    },
    {
      "epoch": 0.6318972033257747,
      "grad_norm": 0.059868182986974716,
      "learning_rate": 0.0003235600477678461,
      "loss": 1.4787,
      "step": 14212
    },
    {
      "epoch": 0.6319861277844471,
      "grad_norm": 0.06001732498407364,
      "learning_rate": 0.00032342327124078416,
      "loss": 1.4724,
      "step": 14214
    },
    {
      "epoch": 0.6320750522431194,
      "grad_norm": 0.061543405055999756,
      "learning_rate": 0.00032328650980827925,
      "loss": 1.4688,
      "step": 14216
    },
    {
      "epoch": 0.6321639767017918,
      "grad_norm": 0.05957053229212761,
      "learning_rate": 0.0003231497634820222,
      "loss": 1.4732,
      "step": 14218
    },
    {
      "epoch": 0.6322529011604642,
      "grad_norm": 0.060270920395851135,
      "learning_rate": 0.0003230130322737027,
      "loss": 1.4665,
      "step": 14220
    },
    {
      "epoch": 0.6323418256191365,
      "grad_norm": 0.0641508549451828,
      "learning_rate": 0.00032287631619500913,
      "loss": 1.4746,
      "step": 14222
    },
    {
      "epoch": 0.6324307500778089,
      "grad_norm": 0.060125336050987244,
      "learning_rate": 0.00032273961525762886,
      "loss": 1.4736,
      "step": 14224
    },
    {
      "epoch": 0.6325196745364813,
      "grad_norm": 0.06322526186704636,
      "learning_rate": 0.00032260292947324697,
      "loss": 1.473,
      "step": 14226
    },
    {
      "epoch": 0.6326085989951536,
      "grad_norm": 0.06271185725927353,
      "learning_rate": 0.0003224662588535489,
      "loss": 1.4777,
      "step": 14228
    },
    {
      "epoch": 0.632697523453826,
      "grad_norm": 0.0621628575026989,
      "learning_rate": 0.00032232960341021697,
      "loss": 1.474,
      "step": 14230
    },
    {
      "epoch": 0.6327864479124984,
      "grad_norm": 0.06069038808345795,
      "learning_rate": 0.00032219296315493385,
      "loss": 1.4773,
      "step": 14232
    },
    {
      "epoch": 0.6328753723711706,
      "grad_norm": 0.06033967062830925,
      "learning_rate": 0.00032205633809937975,
      "loss": 1.4754,
      "step": 14234
    },
    {
      "epoch": 0.632964296829843,
      "grad_norm": 0.06296979635953903,
      "learning_rate": 0.00032191972825523383,
      "loss": 1.4727,
      "step": 14236
    },
    {
      "epoch": 0.6330532212885154,
      "grad_norm": 0.061179324984550476,
      "learning_rate": 0.00032178313363417455,
      "loss": 1.4744,
      "step": 14238
    },
    {
      "epoch": 0.6331421457471877,
      "grad_norm": 0.06097666174173355,
      "learning_rate": 0.0003216465542478783,
      "loss": 1.4738,
      "step": 14240
    },
    {
      "epoch": 0.6332310702058601,
      "grad_norm": 0.05940008535981178,
      "learning_rate": 0.0003215099901080205,
      "loss": 1.4743,
      "step": 14242
    },
    {
      "epoch": 0.6333199946645325,
      "grad_norm": 0.06126728653907776,
      "learning_rate": 0.0003213734412262754,
      "loss": 1.4739,
      "step": 14244
    },
    {
      "epoch": 0.6334089191232048,
      "grad_norm": 0.06011746823787689,
      "learning_rate": 0.00032123690761431565,
      "loss": 1.4811,
      "step": 14246
    },
    {
      "epoch": 0.6334978435818772,
      "grad_norm": 0.06307831406593323,
      "learning_rate": 0.0003211003892838129,
      "loss": 1.4803,
      "step": 14248
    },
    {
      "epoch": 0.6335867680405496,
      "grad_norm": 0.061354558914899826,
      "learning_rate": 0.00032096388624643736,
      "loss": 1.4758,
      "step": 14250
    },
    {
      "epoch": 0.633675692499222,
      "grad_norm": 0.05967186391353607,
      "learning_rate": 0.0003208273985138576,
      "loss": 1.4686,
      "step": 14252
    },
    {
      "epoch": 0.6337646169578943,
      "grad_norm": 0.06030327454209328,
      "learning_rate": 0.00032069092609774157,
      "loss": 1.4719,
      "step": 14254
    },
    {
      "epoch": 0.6338535414165666,
      "grad_norm": 0.061803799122571945,
      "learning_rate": 0.0003205544690097553,
      "loss": 1.472,
      "step": 14256
    },
    {
      "epoch": 0.633942465875239,
      "grad_norm": 0.05942728742957115,
      "learning_rate": 0.00032041802726156377,
      "loss": 1.4767,
      "step": 14258
    },
    {
      "epoch": 0.6340313903339113,
      "grad_norm": 0.06224941089749336,
      "learning_rate": 0.0003202816008648307,
      "loss": 1.4734,
      "step": 14260
    },
    {
      "epoch": 0.6341203147925837,
      "grad_norm": 0.06028662249445915,
      "learning_rate": 0.00032014518983121843,
      "loss": 1.4727,
      "step": 14262
    },
    {
      "epoch": 0.634209239251256,
      "grad_norm": 0.061462193727493286,
      "learning_rate": 0.00032000879417238784,
      "loss": 1.4763,
      "step": 14264
    },
    {
      "epoch": 0.6342981637099284,
      "grad_norm": 0.061825983226299286,
      "learning_rate": 0.00031987241389999876,
      "loss": 1.4754,
      "step": 14266
    },
    {
      "epoch": 0.6343870881686008,
      "grad_norm": 0.062059544026851654,
      "learning_rate": 0.0003197360490257094,
      "loss": 1.4714,
      "step": 14268
    },
    {
      "epoch": 0.6344760126272732,
      "grad_norm": 0.06092653051018715,
      "learning_rate": 0.00031959969956117716,
      "loss": 1.4772,
      "step": 14270
    },
    {
      "epoch": 0.6345649370859455,
      "grad_norm": 0.061005737632513046,
      "learning_rate": 0.0003194633655180575,
      "loss": 1.4744,
      "step": 14272
    },
    {
      "epoch": 0.6346538615446179,
      "grad_norm": 0.06450587511062622,
      "learning_rate": 0.00031932704690800493,
      "loss": 1.4779,
      "step": 14274
    },
    {
      "epoch": 0.6347427860032903,
      "grad_norm": 0.0617448128759861,
      "learning_rate": 0.00031919074374267277,
      "loss": 1.4732,
      "step": 14276
    },
    {
      "epoch": 0.6348317104619625,
      "grad_norm": 0.061758507043123245,
      "learning_rate": 0.0003190544560337126,
      "loss": 1.4744,
      "step": 14278
    },
    {
      "epoch": 0.6349206349206349,
      "grad_norm": 0.06205600127577782,
      "learning_rate": 0.0003189181837927747,
      "loss": 1.4701,
      "step": 14280
    },
    {
      "epoch": 0.6350095593793073,
      "grad_norm": 0.0619838610291481,
      "learning_rate": 0.0003187819270315087,
      "loss": 1.4782,
      "step": 14282
    },
    {
      "epoch": 0.6350984838379796,
      "grad_norm": 0.05949383229017258,
      "learning_rate": 0.000318645685761562,
      "loss": 1.4753,
      "step": 14284
    },
    {
      "epoch": 0.635187408296652,
      "grad_norm": 0.061724212020635605,
      "learning_rate": 0.00031850945999458135,
      "loss": 1.4738,
      "step": 14286
    },
    {
      "epoch": 0.6352763327553244,
      "grad_norm": 0.060357339680194855,
      "learning_rate": 0.00031837324974221193,
      "loss": 1.472,
      "step": 14288
    },
    {
      "epoch": 0.6353652572139967,
      "grad_norm": 0.06111651286482811,
      "learning_rate": 0.0003182370550160974,
      "loss": 1.4803,
      "step": 14290
    },
    {
      "epoch": 0.6354541816726691,
      "grad_norm": 0.05993794649839401,
      "learning_rate": 0.0003181008758278805,
      "loss": 1.4755,
      "step": 14292
    },
    {
      "epoch": 0.6355431061313415,
      "grad_norm": 0.06252172589302063,
      "learning_rate": 0.00031796471218920237,
      "loss": 1.4704,
      "step": 14294
    },
    {
      "epoch": 0.6356320305900138,
      "grad_norm": 0.0633573830127716,
      "learning_rate": 0.0003178285641117027,
      "loss": 1.4694,
      "step": 14296
    },
    {
      "epoch": 0.6357209550486861,
      "grad_norm": 0.06102222204208374,
      "learning_rate": 0.0003176924316070203,
      "loss": 1.4782,
      "step": 14298
    },
    {
      "epoch": 0.6358098795073585,
      "grad_norm": 0.06126133352518082,
      "learning_rate": 0.0003175563146867921,
      "loss": 1.4734,
      "step": 14300
    },
    {
      "epoch": 0.6358988039660308,
      "grad_norm": 0.06115219369530678,
      "learning_rate": 0.0003174202133626543,
      "loss": 1.4785,
      "step": 14302
    },
    {
      "epoch": 0.6359877284247032,
      "grad_norm": 0.05963146686553955,
      "learning_rate": 0.0003172841276462414,
      "loss": 1.4754,
      "step": 14304
    },
    {
      "epoch": 0.6360766528833756,
      "grad_norm": 0.06013455241918564,
      "learning_rate": 0.00031714805754918607,
      "loss": 1.4751,
      "step": 14306
    },
    {
      "epoch": 0.6361655773420479,
      "grad_norm": 0.060067515820264816,
      "learning_rate": 0.00031701200308312087,
      "loss": 1.4703,
      "step": 14308
    },
    {
      "epoch": 0.6362545018007203,
      "grad_norm": 0.06043084338307381,
      "learning_rate": 0.000316875964259676,
      "loss": 1.4663,
      "step": 14310
    },
    {
      "epoch": 0.6363434262593927,
      "grad_norm": 0.059209950268268585,
      "learning_rate": 0.0003167399410904807,
      "loss": 1.4715,
      "step": 14312
    },
    {
      "epoch": 0.636432350718065,
      "grad_norm": 0.06119099631905556,
      "learning_rate": 0.00031660393358716275,
      "loss": 1.4731,
      "step": 14314
    },
    {
      "epoch": 0.6365212751767374,
      "grad_norm": 0.059667252004146576,
      "learning_rate": 0.00031646794176134883,
      "loss": 1.4765,
      "step": 14316
    },
    {
      "epoch": 0.6366101996354098,
      "grad_norm": 0.062074486166238785,
      "learning_rate": 0.000316331965624664,
      "loss": 1.4751,
      "step": 14318
    },
    {
      "epoch": 0.636699124094082,
      "grad_norm": 0.05942871421575546,
      "learning_rate": 0.0003161960051887322,
      "loss": 1.4739,
      "step": 14320
    },
    {
      "epoch": 0.6367880485527544,
      "grad_norm": 0.0627705380320549,
      "learning_rate": 0.0003160600604651758,
      "loss": 1.4667,
      "step": 14322
    },
    {
      "epoch": 0.6368769730114268,
      "grad_norm": 0.05913541093468666,
      "learning_rate": 0.00031592413146561604,
      "loss": 1.4713,
      "step": 14324
    },
    {
      "epoch": 0.6369658974700991,
      "grad_norm": 0.06256476044654846,
      "learning_rate": 0.00031578821820167294,
      "loss": 1.4733,
      "step": 14326
    },
    {
      "epoch": 0.6370548219287715,
      "grad_norm": 0.06173428148031235,
      "learning_rate": 0.0003156523206849644,
      "loss": 1.4775,
      "step": 14328
    },
    {
      "epoch": 0.6371437463874439,
      "grad_norm": 0.0602475143969059,
      "learning_rate": 0.00031551643892710815,
      "loss": 1.4763,
      "step": 14330
    },
    {
      "epoch": 0.6372326708461162,
      "grad_norm": 0.06171022728085518,
      "learning_rate": 0.0003153805729397197,
      "loss": 1.474,
      "step": 14332
    },
    {
      "epoch": 0.6373215953047886,
      "grad_norm": 0.06058694049715996,
      "learning_rate": 0.0003152447227344132,
      "loss": 1.4785,
      "step": 14334
    },
    {
      "epoch": 0.637410519763461,
      "grad_norm": 0.06109078973531723,
      "learning_rate": 0.0003151088883228022,
      "loss": 1.469,
      "step": 14336
    },
    {
      "epoch": 0.6374994442221333,
      "grad_norm": 0.06186812371015549,
      "learning_rate": 0.00031497306971649806,
      "loss": 1.4757,
      "step": 14338
    },
    {
      "epoch": 0.6375883686808057,
      "grad_norm": 0.059616927057504654,
      "learning_rate": 0.00031483726692711143,
      "loss": 1.473,
      "step": 14340
    },
    {
      "epoch": 0.637677293139478,
      "grad_norm": 0.06053665280342102,
      "learning_rate": 0.0003147014799662513,
      "loss": 1.4735,
      "step": 14342
    },
    {
      "epoch": 0.6377662175981503,
      "grad_norm": 0.06036613509058952,
      "learning_rate": 0.000314565708845525,
      "loss": 1.4742,
      "step": 14344
    },
    {
      "epoch": 0.6378551420568227,
      "grad_norm": 0.05983301252126694,
      "learning_rate": 0.00031442995357653926,
      "loss": 1.4764,
      "step": 14346
    },
    {
      "epoch": 0.6379440665154951,
      "grad_norm": 0.060548748821020126,
      "learning_rate": 0.00031429421417089885,
      "loss": 1.4712,
      "step": 14348
    },
    {
      "epoch": 0.6380329909741674,
      "grad_norm": 0.059574976563453674,
      "learning_rate": 0.0003141584906402073,
      "loss": 1.4698,
      "step": 14350
    },
    {
      "epoch": 0.6381219154328398,
      "grad_norm": 0.06052010506391525,
      "learning_rate": 0.0003140227829960671,
      "loss": 1.4738,
      "step": 14352
    },
    {
      "epoch": 0.6382108398915122,
      "grad_norm": 0.059967853128910065,
      "learning_rate": 0.0003138870912500788,
      "loss": 1.4719,
      "step": 14354
    },
    {
      "epoch": 0.6382997643501845,
      "grad_norm": 0.06056461110711098,
      "learning_rate": 0.00031375141541384224,
      "loss": 1.4771,
      "step": 14356
    },
    {
      "epoch": 0.6383886888088569,
      "grad_norm": 0.062207210808992386,
      "learning_rate": 0.0003136157554989556,
      "loss": 1.4707,
      "step": 14358
    },
    {
      "epoch": 0.6384776132675293,
      "grad_norm": 0.05935196205973625,
      "learning_rate": 0.00031348011151701523,
      "loss": 1.4646,
      "step": 14360
    },
    {
      "epoch": 0.6385665377262016,
      "grad_norm": 0.06075610965490341,
      "learning_rate": 0.0003133444834796172,
      "loss": 1.4748,
      "step": 14362
    },
    {
      "epoch": 0.6386554621848739,
      "grad_norm": 0.06111859157681465,
      "learning_rate": 0.00031320887139835515,
      "loss": 1.4721,
      "step": 14364
    },
    {
      "epoch": 0.6387443866435463,
      "grad_norm": 0.05913927033543587,
      "learning_rate": 0.0003130732752848218,
      "loss": 1.4773,
      "step": 14366
    },
    {
      "epoch": 0.6388333111022186,
      "grad_norm": 0.06242065876722336,
      "learning_rate": 0.00031293769515060884,
      "loss": 1.4778,
      "step": 14368
    },
    {
      "epoch": 0.638922235560891,
      "grad_norm": 0.060579001903533936,
      "learning_rate": 0.0003128021310073059,
      "loss": 1.4749,
      "step": 14370
    },
    {
      "epoch": 0.6390111600195634,
      "grad_norm": 0.06134400889277458,
      "learning_rate": 0.0003126665828665018,
      "loss": 1.4755,
      "step": 14372
    },
    {
      "epoch": 0.6391000844782357,
      "grad_norm": 0.06002011522650719,
      "learning_rate": 0.00031253105073978385,
      "loss": 1.4783,
      "step": 14374
    },
    {
      "epoch": 0.6391890089369081,
      "grad_norm": 0.06118055060505867,
      "learning_rate": 0.0003123955346387377,
      "loss": 1.4711,
      "step": 14376
    },
    {
      "epoch": 0.6392779333955805,
      "grad_norm": 0.060449860990047455,
      "learning_rate": 0.0003122600345749481,
      "loss": 1.4773,
      "step": 14378
    },
    {
      "epoch": 0.6393668578542528,
      "grad_norm": 0.06100989878177643,
      "learning_rate": 0.00031212455055999814,
      "loss": 1.4668,
      "step": 14380
    },
    {
      "epoch": 0.6394557823129252,
      "grad_norm": 0.06005636230111122,
      "learning_rate": 0.0003119890826054693,
      "loss": 1.4761,
      "step": 14382
    },
    {
      "epoch": 0.6395447067715976,
      "grad_norm": 0.06204495206475258,
      "learning_rate": 0.00031185363072294247,
      "loss": 1.4733,
      "step": 14384
    },
    {
      "epoch": 0.6396336312302698,
      "grad_norm": 0.05984333157539368,
      "learning_rate": 0.0003117181949239963,
      "loss": 1.4742,
      "step": 14386
    },
    {
      "epoch": 0.6397225556889422,
      "grad_norm": 0.06164374575018883,
      "learning_rate": 0.0003115827752202086,
      "loss": 1.4814,
      "step": 14388
    },
    {
      "epoch": 0.6398114801476146,
      "grad_norm": 0.06006844714283943,
      "learning_rate": 0.00031144737162315555,
      "loss": 1.4765,
      "step": 14390
    },
    {
      "epoch": 0.6399004046062869,
      "grad_norm": 0.06163867563009262,
      "learning_rate": 0.0003113119841444121,
      "loss": 1.4774,
      "step": 14392
    },
    {
      "epoch": 0.6399893290649593,
      "grad_norm": 0.06025887280702591,
      "learning_rate": 0.00031117661279555175,
      "loss": 1.4702,
      "step": 14394
    },
    {
      "epoch": 0.6400782535236317,
      "grad_norm": 0.06053690239787102,
      "learning_rate": 0.00031104125758814674,
      "loss": 1.4726,
      "step": 14396
    },
    {
      "epoch": 0.640167177982304,
      "grad_norm": 0.06094555929303169,
      "learning_rate": 0.0003109059185337676,
      "loss": 1.4739,
      "step": 14398
    },
    {
      "epoch": 0.6402561024409764,
      "grad_norm": 0.060711413621902466,
      "learning_rate": 0.00031077059564398404,
      "loss": 1.4722,
      "step": 14400
    },
    {
      "epoch": 0.6403450268996488,
      "grad_norm": 0.05947254225611687,
      "learning_rate": 0.0003106352889303638,
      "loss": 1.4736,
      "step": 14402
    },
    {
      "epoch": 0.6404339513583212,
      "grad_norm": 0.05949154123663902,
      "learning_rate": 0.00031049999840447346,
      "loss": 1.4744,
      "step": 14404
    },
    {
      "epoch": 0.6405228758169934,
      "grad_norm": 0.05873184651136398,
      "learning_rate": 0.0003103647240778785,
      "loss": 1.4736,
      "step": 14406
    },
    {
      "epoch": 0.6406118002756658,
      "grad_norm": 0.05965226888656616,
      "learning_rate": 0.0003102294659621425,
      "loss": 1.4766,
      "step": 14408
    },
    {
      "epoch": 0.6407007247343381,
      "grad_norm": 0.06001347303390503,
      "learning_rate": 0.0003100942240688281,
      "loss": 1.4721,
      "step": 14410
    },
    {
      "epoch": 0.6407896491930105,
      "grad_norm": 0.06033129617571831,
      "learning_rate": 0.0003099589984094965,
      "loss": 1.477,
      "step": 14412
    },
    {
      "epoch": 0.6408785736516829,
      "grad_norm": 0.060869600623846054,
      "learning_rate": 0.00030982378899570685,
      "loss": 1.4678,
      "step": 14414
    },
    {
      "epoch": 0.6409674981103552,
      "grad_norm": 0.060569677501916885,
      "learning_rate": 0.0003096885958390181,
      "loss": 1.4698,
      "step": 14416
    },
    {
      "epoch": 0.6410564225690276,
      "grad_norm": 0.060459256172180176,
      "learning_rate": 0.0003095534189509867,
      "loss": 1.4784,
      "step": 14418
    },
    {
      "epoch": 0.6411453470277,
      "grad_norm": 0.06125481799244881,
      "learning_rate": 0.00030941825834316825,
      "loss": 1.4712,
      "step": 14420
    },
    {
      "epoch": 0.6412342714863724,
      "grad_norm": 0.060924872756004333,
      "learning_rate": 0.00030928311402711705,
      "loss": 1.475,
      "step": 14422
    },
    {
      "epoch": 0.6413231959450447,
      "grad_norm": 0.06078309938311577,
      "learning_rate": 0.00030914798601438566,
      "loss": 1.4719,
      "step": 14424
    },
    {
      "epoch": 0.6414121204037171,
      "grad_norm": 0.06124448776245117,
      "learning_rate": 0.0003090128743165256,
      "loss": 1.4706,
      "step": 14426
    },
    {
      "epoch": 0.6415010448623893,
      "grad_norm": 0.060657963156700134,
      "learning_rate": 0.0003088777789450866,
      "loss": 1.4778,
      "step": 14428
    },
    {
      "epoch": 0.6415899693210617,
      "grad_norm": 0.0634801834821701,
      "learning_rate": 0.0003087426999116173,
      "loss": 1.4766,
      "step": 14430
    },
    {
      "epoch": 0.6416788937797341,
      "grad_norm": 0.06083952635526657,
      "learning_rate": 0.00030860763722766486,
      "loss": 1.4667,
      "step": 14432
    },
    {
      "epoch": 0.6417678182384065,
      "grad_norm": 0.06175751984119415,
      "learning_rate": 0.0003084725909047752,
      "loss": 1.475,
      "step": 14434
    },
    {
      "epoch": 0.6418567426970788,
      "grad_norm": 0.06247575953602791,
      "learning_rate": 0.00030833756095449217,
      "loss": 1.4704,
      "step": 14436
    },
    {
      "epoch": 0.6419456671557512,
      "grad_norm": 0.06103740632534027,
      "learning_rate": 0.0003082025473883594,
      "loss": 1.4742,
      "step": 14438
    },
    {
      "epoch": 0.6420345916144236,
      "grad_norm": 0.06011580675840378,
      "learning_rate": 0.00030806755021791797,
      "loss": 1.4773,
      "step": 14440
    },
    {
      "epoch": 0.6421235160730959,
      "grad_norm": 0.05931413918733597,
      "learning_rate": 0.0003079325694547081,
      "loss": 1.4702,
      "step": 14442
    },
    {
      "epoch": 0.6422124405317683,
      "grad_norm": 0.05985075607895851,
      "learning_rate": 0.00030779760511026865,
      "loss": 1.4705,
      "step": 14444
    },
    {
      "epoch": 0.6423013649904407,
      "grad_norm": 0.06110270693898201,
      "learning_rate": 0.00030766265719613686,
      "loss": 1.4728,
      "step": 14446
    },
    {
      "epoch": 0.642390289449113,
      "grad_norm": 0.061616457998752594,
      "learning_rate": 0.00030752772572384884,
      "loss": 1.4699,
      "step": 14448
    },
    {
      "epoch": 0.6424792139077853,
      "grad_norm": 0.05985260009765625,
      "learning_rate": 0.000307392810704939,
      "loss": 1.4706,
      "step": 14450
    },
    {
      "epoch": 0.6425681383664577,
      "grad_norm": 0.06176939606666565,
      "learning_rate": 0.00030725791215094046,
      "loss": 1.4756,
      "step": 14452
    },
    {
      "epoch": 0.64265706282513,
      "grad_norm": 0.060520898550748825,
      "learning_rate": 0.000307123030073385,
      "loss": 1.4789,
      "step": 14454
    },
    {
      "epoch": 0.6427459872838024,
      "grad_norm": 0.060216981917619705,
      "learning_rate": 0.00030698816448380294,
      "loss": 1.4725,
      "step": 14456
    },
    {
      "epoch": 0.6428349117424748,
      "grad_norm": 0.06227912753820419,
      "learning_rate": 0.000306853315393723,
      "loss": 1.4763,
      "step": 14458
    },
    {
      "epoch": 0.6429238362011471,
      "grad_norm": 0.05990215763449669,
      "learning_rate": 0.000306718482814673,
      "loss": 1.4747,
      "step": 14460
    },
    {
      "epoch": 0.6430127606598195,
      "grad_norm": 0.060426462441682816,
      "learning_rate": 0.0003065836667581787,
      "loss": 1.4722,
      "step": 14462
    },
    {
      "epoch": 0.6431016851184919,
      "grad_norm": 0.061631377786397934,
      "learning_rate": 0.0003064488672357651,
      "loss": 1.4755,
      "step": 14464
    },
    {
      "epoch": 0.6431906095771642,
      "grad_norm": 0.061112042516469955,
      "learning_rate": 0.00030631408425895536,
      "loss": 1.4776,
      "step": 14466
    },
    {
      "epoch": 0.6432795340358366,
      "grad_norm": 0.06276817619800568,
      "learning_rate": 0.00030617931783927086,
      "loss": 1.4744,
      "step": 14468
    },
    {
      "epoch": 0.643368458494509,
      "grad_norm": 0.06161990389227867,
      "learning_rate": 0.0003060445679882328,
      "loss": 1.4732,
      "step": 14470
    },
    {
      "epoch": 0.6434573829531812,
      "grad_norm": 0.06073969602584839,
      "learning_rate": 0.0003059098347173597,
      "loss": 1.4772,
      "step": 14472
    },
    {
      "epoch": 0.6435463074118536,
      "grad_norm": 0.06146194413304329,
      "learning_rate": 0.0003057751180381691,
      "loss": 1.4648,
      "step": 14474
    },
    {
      "epoch": 0.643635231870526,
      "grad_norm": 0.06184494122862816,
      "learning_rate": 0.00030564041796217744,
      "loss": 1.4684,
      "step": 14476
    },
    {
      "epoch": 0.6437241563291983,
      "grad_norm": 0.06379452347755432,
      "learning_rate": 0.0003055057345008993,
      "loss": 1.4741,
      "step": 14478
    },
    {
      "epoch": 0.6438130807878707,
      "grad_norm": 0.06197437271475792,
      "learning_rate": 0.0003053710676658482,
      "loss": 1.4728,
      "step": 14480
    },
    {
      "epoch": 0.6439020052465431,
      "grad_norm": 0.06214113533496857,
      "learning_rate": 0.00030523641746853594,
      "loss": 1.4751,
      "step": 14482
    },
    {
      "epoch": 0.6439909297052154,
      "grad_norm": 0.061672694981098175,
      "learning_rate": 0.0003051017839204728,
      "loss": 1.4736,
      "step": 14484
    },
    {
      "epoch": 0.6440798541638878,
      "grad_norm": 0.06234847381711006,
      "learning_rate": 0.0003049671670331683,
      "loss": 1.4706,
      "step": 14486
    },
    {
      "epoch": 0.6441687786225602,
      "grad_norm": 0.06098306179046631,
      "learning_rate": 0.00030483256681812987,
      "loss": 1.4699,
      "step": 14488
    },
    {
      "epoch": 0.6442577030812325,
      "grad_norm": 0.060952410101890564,
      "learning_rate": 0.0003046979832868634,
      "loss": 1.4701,
      "step": 14490
    },
    {
      "epoch": 0.6443466275399049,
      "grad_norm": 0.061424147337675095,
      "learning_rate": 0.00030456341645087426,
      "loss": 1.4764,
      "step": 14492
    },
    {
      "epoch": 0.6444355519985772,
      "grad_norm": 0.06046923249959946,
      "learning_rate": 0.0003044288663216653,
      "loss": 1.4673,
      "step": 14494
    },
    {
      "epoch": 0.6445244764572495,
      "grad_norm": 0.06237080320715904,
      "learning_rate": 0.00030429433291073905,
      "loss": 1.4692,
      "step": 14496
    },
    {
      "epoch": 0.6446134009159219,
      "grad_norm": 0.0605134479701519,
      "learning_rate": 0.00030415981622959544,
      "loss": 1.4683,
      "step": 14498
    },
    {
      "epoch": 0.6447023253745943,
      "grad_norm": 0.061160653829574585,
      "learning_rate": 0.00030402531628973375,
      "loss": 1.4755,
      "step": 14500
    },
    {
      "epoch": 0.6447023253745943,
      "eval_loss": 1.45697820186615,
      "eval_runtime": 12.467,
      "eval_samples_per_second": 554.261,
      "eval_steps_per_second": 69.303,
      "step": 14500
    },
    {
      "epoch": 0.6447912498332666,
      "grad_norm": 0.06107184290885925,
      "learning_rate": 0.00030389083310265177,
      "loss": 1.4756,
      "step": 14502
    },
    {
      "epoch": 0.644880174291939,
      "grad_norm": 0.06288035959005356,
      "learning_rate": 0.0003037563666798456,
      "loss": 1.4804,
      "step": 14504
    },
    {
      "epoch": 0.6449690987506114,
      "grad_norm": 0.061453141272068024,
      "learning_rate": 0.0003036219170328099,
      "loss": 1.4762,
      "step": 14506
    },
    {
      "epoch": 0.6450580232092837,
      "grad_norm": 0.059680961072444916,
      "learning_rate": 0.0003034874841730382,
      "loss": 1.4776,
      "step": 14508
    },
    {
      "epoch": 0.6451469476679561,
      "grad_norm": 0.061881791800260544,
      "learning_rate": 0.00030335306811202245,
      "loss": 1.4725,
      "step": 14510
    },
    {
      "epoch": 0.6452358721266285,
      "grad_norm": 0.06059206277132034,
      "learning_rate": 0.00030321866886125285,
      "loss": 1.471,
      "step": 14512
    },
    {
      "epoch": 0.6453247965853008,
      "grad_norm": 0.0622367300093174,
      "learning_rate": 0.0003030842864322187,
      "loss": 1.4723,
      "step": 14514
    },
    {
      "epoch": 0.6454137210439731,
      "grad_norm": 0.06035193055868149,
      "learning_rate": 0.0003029499208364076,
      "loss": 1.472,
      "step": 14516
    },
    {
      "epoch": 0.6455026455026455,
      "grad_norm": 0.06035499647259712,
      "learning_rate": 0.00030281557208530557,
      "loss": 1.4712,
      "step": 14518
    },
    {
      "epoch": 0.6455915699613178,
      "grad_norm": 0.06048353761434555,
      "learning_rate": 0.00030268124019039753,
      "loss": 1.4763,
      "step": 14520
    },
    {
      "epoch": 0.6456804944199902,
      "grad_norm": 0.06177537515759468,
      "learning_rate": 0.0003025469251631664,
      "loss": 1.475,
      "step": 14522
    },
    {
      "epoch": 0.6457694188786626,
      "grad_norm": 0.061068419367074966,
      "learning_rate": 0.0003024126270150945,
      "loss": 1.471,
      "step": 14524
    },
    {
      "epoch": 0.6458583433373349,
      "grad_norm": 0.06005478650331497,
      "learning_rate": 0.0003022783457576619,
      "loss": 1.4684,
      "step": 14526
    },
    {
      "epoch": 0.6459472677960073,
      "grad_norm": 0.06285406649112701,
      "learning_rate": 0.0003021440814023475,
      "loss": 1.4752,
      "step": 14528
    },
    {
      "epoch": 0.6460361922546797,
      "grad_norm": 0.060045741498470306,
      "learning_rate": 0.00030200983396062907,
      "loss": 1.4733,
      "step": 14530
    },
    {
      "epoch": 0.646125116713352,
      "grad_norm": 0.059561315923929214,
      "learning_rate": 0.00030187560344398245,
      "loss": 1.4758,
      "step": 14532
    },
    {
      "epoch": 0.6462140411720244,
      "grad_norm": 0.061188627034425735,
      "learning_rate": 0.0003017413898638825,
      "loss": 1.4796,
      "step": 14534
    },
    {
      "epoch": 0.6463029656306967,
      "grad_norm": 0.06275120377540588,
      "learning_rate": 0.0003016071932318022,
      "loss": 1.4765,
      "step": 14536
    },
    {
      "epoch": 0.646391890089369,
      "grad_norm": 0.06224878132343292,
      "learning_rate": 0.0003014730135592132,
      "loss": 1.4726,
      "step": 14538
    },
    {
      "epoch": 0.6464808145480414,
      "grad_norm": 0.06063244864344597,
      "learning_rate": 0.00030133885085758595,
      "loss": 1.4701,
      "step": 14540
    },
    {
      "epoch": 0.6465697390067138,
      "grad_norm": 0.06237416714429855,
      "learning_rate": 0.00030120470513838937,
      "loss": 1.4742,
      "step": 14542
    },
    {
      "epoch": 0.6466586634653861,
      "grad_norm": 0.06257627159357071,
      "learning_rate": 0.00030107057641309034,
      "loss": 1.4741,
      "step": 14544
    },
    {
      "epoch": 0.6467475879240585,
      "grad_norm": 0.05979495123028755,
      "learning_rate": 0.0003009364646931554,
      "loss": 1.4755,
      "step": 14546
    },
    {
      "epoch": 0.6468365123827309,
      "grad_norm": 0.059078190475702286,
      "learning_rate": 0.00030080236999004833,
      "loss": 1.4745,
      "step": 14548
    },
    {
      "epoch": 0.6469254368414032,
      "grad_norm": 0.059303369373083115,
      "learning_rate": 0.00030066829231523287,
      "loss": 1.4728,
      "step": 14550
    },
    {
      "epoch": 0.6470143613000756,
      "grad_norm": 0.06229490414261818,
      "learning_rate": 0.00030053423168017014,
      "loss": 1.4669,
      "step": 14552
    },
    {
      "epoch": 0.647103285758748,
      "grad_norm": 0.06081756576895714,
      "learning_rate": 0.00030040018809632013,
      "loss": 1.4707,
      "step": 14554
    },
    {
      "epoch": 0.6471922102174203,
      "grad_norm": 0.05991298332810402,
      "learning_rate": 0.00030026616157514176,
      "loss": 1.4704,
      "step": 14556
    },
    {
      "epoch": 0.6472811346760926,
      "grad_norm": 0.0638953223824501,
      "learning_rate": 0.0003001321521280921,
      "loss": 1.4678,
      "step": 14558
    },
    {
      "epoch": 0.647370059134765,
      "grad_norm": 0.062413301318883896,
      "learning_rate": 0.0002999981597666268,
      "loss": 1.4718,
      "step": 14560
    },
    {
      "epoch": 0.6474589835934373,
      "grad_norm": 0.060770995914936066,
      "learning_rate": 0.0002998641845022003,
      "loss": 1.4711,
      "step": 14562
    },
    {
      "epoch": 0.6475479080521097,
      "grad_norm": 0.06420081853866577,
      "learning_rate": 0.00029973022634626514,
      "loss": 1.4737,
      "step": 14564
    },
    {
      "epoch": 0.6476368325107821,
      "grad_norm": 0.059697262942790985,
      "learning_rate": 0.00029959628531027287,
      "loss": 1.4752,
      "step": 14566
    },
    {
      "epoch": 0.6477257569694544,
      "grad_norm": 0.06121120974421501,
      "learning_rate": 0.0002994623614056733,
      "loss": 1.4698,
      "step": 14568
    },
    {
      "epoch": 0.6478146814281268,
      "grad_norm": 0.06134429946541786,
      "learning_rate": 0.00029932845464391476,
      "loss": 1.471,
      "step": 14570
    },
    {
      "epoch": 0.6479036058867992,
      "grad_norm": 0.06118486449122429,
      "learning_rate": 0.00029919456503644425,
      "loss": 1.4717,
      "step": 14572
    },
    {
      "epoch": 0.6479925303454716,
      "grad_norm": 0.06073017045855522,
      "learning_rate": 0.00029906069259470747,
      "loss": 1.4767,
      "step": 14574
    },
    {
      "epoch": 0.6480814548041439,
      "grad_norm": 0.06142178550362587,
      "learning_rate": 0.0002989268373301478,
      "loss": 1.4737,
      "step": 14576
    },
    {
      "epoch": 0.6481703792628163,
      "grad_norm": 0.05868706852197647,
      "learning_rate": 0.00029879299925420855,
      "loss": 1.4715,
      "step": 14578
    },
    {
      "epoch": 0.6482593037214885,
      "grad_norm": 0.06036018207669258,
      "learning_rate": 0.00029865917837833023,
      "loss": 1.4641,
      "step": 14580
    },
    {
      "epoch": 0.6483482281801609,
      "grad_norm": 0.06149023771286011,
      "learning_rate": 0.0002985253747139526,
      "loss": 1.4738,
      "step": 14582
    },
    {
      "epoch": 0.6484371526388333,
      "grad_norm": 0.0594891719520092,
      "learning_rate": 0.0002983915882725139,
      "loss": 1.4698,
      "step": 14584
    },
    {
      "epoch": 0.6485260770975056,
      "grad_norm": 0.06031309440732002,
      "learning_rate": 0.00029825781906545057,
      "loss": 1.4683,
      "step": 14586
    },
    {
      "epoch": 0.648615001556178,
      "grad_norm": 0.06182754039764404,
      "learning_rate": 0.000298124067104198,
      "loss": 1.4741,
      "step": 14588
    },
    {
      "epoch": 0.6487039260148504,
      "grad_norm": 0.06004558876156807,
      "learning_rate": 0.0002979903324001898,
      "loss": 1.4774,
      "step": 14590
    },
    {
      "epoch": 0.6487928504735228,
      "grad_norm": 0.05962319299578667,
      "learning_rate": 0.00029785661496485816,
      "loss": 1.4754,
      "step": 14592
    },
    {
      "epoch": 0.6488817749321951,
      "grad_norm": 0.06000960245728493,
      "learning_rate": 0.00029772291480963396,
      "loss": 1.4699,
      "step": 14594
    },
    {
      "epoch": 0.6489706993908675,
      "grad_norm": 0.059110336005687714,
      "learning_rate": 0.00029758923194594657,
      "loss": 1.4747,
      "step": 14596
    },
    {
      "epoch": 0.6490596238495399,
      "grad_norm": 0.0625082477927208,
      "learning_rate": 0.0002974555663852232,
      "loss": 1.4723,
      "step": 14598
    },
    {
      "epoch": 0.6491485483082122,
      "grad_norm": 0.06030767038464546,
      "learning_rate": 0.000297321918138891,
      "loss": 1.4739,
      "step": 14600
    },
    {
      "epoch": 0.6492374727668845,
      "grad_norm": 0.060511305928230286,
      "learning_rate": 0.00029718828721837407,
      "loss": 1.4659,
      "step": 14602
    },
    {
      "epoch": 0.6493263972255569,
      "grad_norm": 0.060783982276916504,
      "learning_rate": 0.0002970546736350964,
      "loss": 1.4677,
      "step": 14604
    },
    {
      "epoch": 0.6494153216842292,
      "grad_norm": 0.060023460537195206,
      "learning_rate": 0.0002969210774004796,
      "loss": 1.4695,
      "step": 14606
    },
    {
      "epoch": 0.6495042461429016,
      "grad_norm": 0.0627756267786026,
      "learning_rate": 0.0002967874985259438,
      "loss": 1.472,
      "step": 14608
    },
    {
      "epoch": 0.649593170601574,
      "grad_norm": 0.061621058732271194,
      "learning_rate": 0.00029665393702290834,
      "loss": 1.4801,
      "step": 14610
    },
    {
      "epoch": 0.6496820950602463,
      "grad_norm": 0.05914198234677315,
      "learning_rate": 0.0002965203929027904,
      "loss": 1.4706,
      "step": 14612
    },
    {
      "epoch": 0.6497710195189187,
      "grad_norm": 0.06379103660583496,
      "learning_rate": 0.0002963868661770059,
      "loss": 1.4757,
      "step": 14614
    },
    {
      "epoch": 0.6498599439775911,
      "grad_norm": 0.06029783561825752,
      "learning_rate": 0.00029625335685696955,
      "loss": 1.4718,
      "step": 14616
    },
    {
      "epoch": 0.6499488684362634,
      "grad_norm": 0.06065394729375839,
      "learning_rate": 0.00029611986495409393,
      "loss": 1.4682,
      "step": 14618
    },
    {
      "epoch": 0.6500377928949358,
      "grad_norm": 0.06273026019334793,
      "learning_rate": 0.0002959863904797908,
      "loss": 1.4757,
      "step": 14620
    },
    {
      "epoch": 0.6501267173536082,
      "grad_norm": 0.06018160283565521,
      "learning_rate": 0.00029585293344547026,
      "loss": 1.4719,
      "step": 14622
    },
    {
      "epoch": 0.6502156418122804,
      "grad_norm": 0.059954337775707245,
      "learning_rate": 0.0002957194938625403,
      "loss": 1.4712,
      "step": 14624
    },
    {
      "epoch": 0.6503045662709528,
      "grad_norm": 0.06017017737030983,
      "learning_rate": 0.0002955860717424085,
      "loss": 1.4692,
      "step": 14626
    },
    {
      "epoch": 0.6503934907296252,
      "grad_norm": 0.0594681017100811,
      "learning_rate": 0.00029545266709648003,
      "loss": 1.4721,
      "step": 14628
    },
    {
      "epoch": 0.6504824151882975,
      "grad_norm": 0.06060068681836128,
      "learning_rate": 0.00029531927993615894,
      "loss": 1.4698,
      "step": 14630
    },
    {
      "epoch": 0.6505713396469699,
      "grad_norm": 0.061522986739873886,
      "learning_rate": 0.0002951859102728478,
      "loss": 1.4723,
      "step": 14632
    },
    {
      "epoch": 0.6506602641056423,
      "grad_norm": 0.06037484481930733,
      "learning_rate": 0.0002950525581179477,
      "loss": 1.4712,
      "step": 14634
    },
    {
      "epoch": 0.6507491885643146,
      "grad_norm": 0.061230264604091644,
      "learning_rate": 0.000294919223482858,
      "loss": 1.4734,
      "step": 14636
    },
    {
      "epoch": 0.650838113022987,
      "grad_norm": 0.06225361302495003,
      "learning_rate": 0.00029478590637897696,
      "loss": 1.4777,
      "step": 14638
    },
    {
      "epoch": 0.6509270374816594,
      "grad_norm": 0.05939456820487976,
      "learning_rate": 0.0002946526068177009,
      "loss": 1.4755,
      "step": 14640
    },
    {
      "epoch": 0.6510159619403317,
      "grad_norm": 0.059683386236429214,
      "learning_rate": 0.000294519324810425,
      "loss": 1.4751,
      "step": 14642
    },
    {
      "epoch": 0.6511048863990041,
      "grad_norm": 0.062137801200151443,
      "learning_rate": 0.0002943860603685428,
      "loss": 1.4733,
      "step": 14644
    },
    {
      "epoch": 0.6511938108576764,
      "grad_norm": 0.05937831103801727,
      "learning_rate": 0.0002942528135034461,
      "loss": 1.4712,
      "step": 14646
    },
    {
      "epoch": 0.6512827353163487,
      "grad_norm": 0.06024874746799469,
      "learning_rate": 0.0002941195842265257,
      "loss": 1.468,
      "step": 14648
    },
    {
      "epoch": 0.6513716597750211,
      "grad_norm": 0.06173409894108772,
      "learning_rate": 0.0002939863725491706,
      "loss": 1.4691,
      "step": 14650
    },
    {
      "epoch": 0.6514605842336935,
      "grad_norm": 0.060418277978897095,
      "learning_rate": 0.00029385317848276807,
      "loss": 1.4726,
      "step": 14652
    },
    {
      "epoch": 0.6515495086923658,
      "grad_norm": 0.06066744774580002,
      "learning_rate": 0.0002937200020387045,
      "loss": 1.4757,
      "step": 14654
    },
    {
      "epoch": 0.6516384331510382,
      "grad_norm": 0.06171972304582596,
      "learning_rate": 0.00029358684322836396,
      "loss": 1.4771,
      "step": 14656
    },
    {
      "epoch": 0.6517273576097106,
      "grad_norm": 0.06183767318725586,
      "learning_rate": 0.00029345370206312996,
      "loss": 1.4738,
      "step": 14658
    },
    {
      "epoch": 0.6518162820683829,
      "grad_norm": 0.06188305467367172,
      "learning_rate": 0.0002933205785543836,
      "loss": 1.4689,
      "step": 14660
    },
    {
      "epoch": 0.6519052065270553,
      "grad_norm": 0.06321601569652557,
      "learning_rate": 0.00029318747271350493,
      "loss": 1.4732,
      "step": 14662
    },
    {
      "epoch": 0.6519941309857277,
      "grad_norm": 0.06155180558562279,
      "learning_rate": 0.0002930543845518726,
      "loss": 1.4707,
      "step": 14664
    },
    {
      "epoch": 0.6520830554443999,
      "grad_norm": 0.0595284141600132,
      "learning_rate": 0.00029292131408086345,
      "loss": 1.4722,
      "step": 14666
    },
    {
      "epoch": 0.6521719799030723,
      "grad_norm": 0.060346901416778564,
      "learning_rate": 0.0002927882613118528,
      "loss": 1.477,
      "step": 14668
    },
    {
      "epoch": 0.6522609043617447,
      "grad_norm": 0.061850935220718384,
      "learning_rate": 0.0002926552262562149,
      "loss": 1.4768,
      "step": 14670
    },
    {
      "epoch": 0.652349828820417,
      "grad_norm": 0.061827659606933594,
      "learning_rate": 0.0002925222089253218,
      "loss": 1.471,
      "step": 14672
    },
    {
      "epoch": 0.6524387532790894,
      "grad_norm": 0.061195094138383865,
      "learning_rate": 0.0002923892093305448,
      "loss": 1.4789,
      "step": 14674
    },
    {
      "epoch": 0.6525276777377618,
      "grad_norm": 0.0616355836391449,
      "learning_rate": 0.0002922562274832531,
      "loss": 1.4651,
      "step": 14676
    },
    {
      "epoch": 0.6526166021964341,
      "grad_norm": 0.05983952432870865,
      "learning_rate": 0.0002921232633948143,
      "loss": 1.4705,
      "step": 14678
    },
    {
      "epoch": 0.6527055266551065,
      "grad_norm": 0.060748714953660965,
      "learning_rate": 0.0002919903170765954,
      "loss": 1.4728,
      "step": 14680
    },
    {
      "epoch": 0.6527944511137789,
      "grad_norm": 0.06134764477610588,
      "learning_rate": 0.0002918573885399607,
      "loss": 1.4724,
      "step": 14682
    },
    {
      "epoch": 0.6528833755724512,
      "grad_norm": 0.061923548579216,
      "learning_rate": 0.00029172447779627367,
      "loss": 1.4699,
      "step": 14684
    },
    {
      "epoch": 0.6529723000311236,
      "grad_norm": 0.05965703725814819,
      "learning_rate": 0.00029159158485689594,
      "loss": 1.4746,
      "step": 14686
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 0.06172526627779007,
      "learning_rate": 0.0002914587097331881,
      "loss": 1.4738,
      "step": 14688
    },
    {
      "epoch": 0.6531501489484682,
      "grad_norm": 0.06159723922610283,
      "learning_rate": 0.000291325852436509,
      "loss": 1.4705,
      "step": 14690
    },
    {
      "epoch": 0.6532390734071406,
      "grad_norm": 0.059951264411211014,
      "learning_rate": 0.00029119301297821554,
      "loss": 1.4704,
      "step": 14692
    },
    {
      "epoch": 0.653327997865813,
      "grad_norm": 0.05888056382536888,
      "learning_rate": 0.0002910601913696631,
      "loss": 1.4688,
      "step": 14694
    },
    {
      "epoch": 0.6534169223244853,
      "grad_norm": 0.05950944125652313,
      "learning_rate": 0.0002909273876222067,
      "loss": 1.4749,
      "step": 14696
    },
    {
      "epoch": 0.6535058467831577,
      "grad_norm": 0.060837604105472565,
      "learning_rate": 0.00029079460174719865,
      "loss": 1.47,
      "step": 14698
    },
    {
      "epoch": 0.6535947712418301,
      "grad_norm": 0.059955399483442307,
      "learning_rate": 0.00029066183375598965,
      "loss": 1.4737,
      "step": 14700
    },
    {
      "epoch": 0.6536836957005024,
      "grad_norm": 0.06083989143371582,
      "learning_rate": 0.0002905290836599297,
      "loss": 1.4775,
      "step": 14702
    },
    {
      "epoch": 0.6537726201591748,
      "grad_norm": 0.0601428858935833,
      "learning_rate": 0.000290396351470367,
      "loss": 1.47,
      "step": 14704
    },
    {
      "epoch": 0.6538615446178472,
      "grad_norm": 0.06066612899303436,
      "learning_rate": 0.00029026363719864755,
      "loss": 1.4735,
      "step": 14706
    },
    {
      "epoch": 0.6539504690765195,
      "grad_norm": 0.060709454119205475,
      "learning_rate": 0.0002901309408561168,
      "loss": 1.471,
      "step": 14708
    },
    {
      "epoch": 0.6540393935351918,
      "grad_norm": 0.061341624706983566,
      "learning_rate": 0.00028999826245411803,
      "loss": 1.4721,
      "step": 14710
    },
    {
      "epoch": 0.6541283179938642,
      "grad_norm": 0.062061794102191925,
      "learning_rate": 0.00028986560200399347,
      "loss": 1.4718,
      "step": 14712
    },
    {
      "epoch": 0.6542172424525365,
      "grad_norm": 0.06041424348950386,
      "learning_rate": 0.000289732959517083,
      "loss": 1.4697,
      "step": 14714
    },
    {
      "epoch": 0.6543061669112089,
      "grad_norm": 0.0598766915500164,
      "learning_rate": 0.00028960033500472583,
      "loss": 1.4718,
      "step": 14716
    },
    {
      "epoch": 0.6543950913698813,
      "grad_norm": 0.061096884310245514,
      "learning_rate": 0.00028946772847825944,
      "loss": 1.4709,
      "step": 14718
    },
    {
      "epoch": 0.6544840158285536,
      "grad_norm": 0.06066969409584999,
      "learning_rate": 0.0002893351399490194,
      "loss": 1.473,
      "step": 14720
    },
    {
      "epoch": 0.654572940287226,
      "grad_norm": 0.061327192932367325,
      "learning_rate": 0.00028920256942833956,
      "loss": 1.4684,
      "step": 14722
    },
    {
      "epoch": 0.6546618647458984,
      "grad_norm": 0.06126781925559044,
      "learning_rate": 0.00028907001692755343,
      "loss": 1.4715,
      "step": 14724
    },
    {
      "epoch": 0.6547507892045707,
      "grad_norm": 0.062026720494031906,
      "learning_rate": 0.00028893748245799154,
      "loss": 1.4663,
      "step": 14726
    },
    {
      "epoch": 0.6548397136632431,
      "grad_norm": 0.06028749421238899,
      "learning_rate": 0.00028880496603098403,
      "loss": 1.4747,
      "step": 14728
    },
    {
      "epoch": 0.6549286381219155,
      "grad_norm": 0.06083076074719429,
      "learning_rate": 0.0002886724676578586,
      "loss": 1.4711,
      "step": 14730
    },
    {
      "epoch": 0.6550175625805877,
      "grad_norm": 0.06158638745546341,
      "learning_rate": 0.0002885399873499418,
      "loss": 1.4754,
      "step": 14732
    },
    {
      "epoch": 0.6551064870392601,
      "grad_norm": 0.06135280430316925,
      "learning_rate": 0.000288407525118559,
      "loss": 1.4757,
      "step": 14734
    },
    {
      "epoch": 0.6551954114979325,
      "grad_norm": 0.0598473995923996,
      "learning_rate": 0.0002882750809750332,
      "loss": 1.4694,
      "step": 14736
    },
    {
      "epoch": 0.6552843359566048,
      "grad_norm": 0.060813672840595245,
      "learning_rate": 0.0002881426549306866,
      "loss": 1.4696,
      "step": 14738
    },
    {
      "epoch": 0.6553732604152772,
      "grad_norm": 0.0595141276717186,
      "learning_rate": 0.00028801024699683954,
      "loss": 1.4734,
      "step": 14740
    },
    {
      "epoch": 0.6554621848739496,
      "grad_norm": 0.059074971824884415,
      "learning_rate": 0.0002878778571848107,
      "loss": 1.4677,
      "step": 14742
    },
    {
      "epoch": 0.655551109332622,
      "grad_norm": 0.06075437366962433,
      "learning_rate": 0.0002877454855059174,
      "loss": 1.4735,
      "step": 14744
    },
    {
      "epoch": 0.6556400337912943,
      "grad_norm": 0.06068753823637962,
      "learning_rate": 0.0002876131319714755,
      "loss": 1.469,
      "step": 14746
    },
    {
      "epoch": 0.6557289582499667,
      "grad_norm": 0.05928294360637665,
      "learning_rate": 0.0002874807965927989,
      "loss": 1.4735,
      "step": 14748
    },
    {
      "epoch": 0.655817882708639,
      "grad_norm": 0.060513000935316086,
      "learning_rate": 0.0002873484793812005,
      "loss": 1.472,
      "step": 14750
    },
    {
      "epoch": 0.6559068071673114,
      "grad_norm": 0.060318686068058014,
      "learning_rate": 0.000287216180347991,
      "loss": 1.4728,
      "step": 14752
    },
    {
      "epoch": 0.6559957316259837,
      "grad_norm": 0.0590246245265007,
      "learning_rate": 0.0002870838995044802,
      "loss": 1.4708,
      "step": 14754
    },
    {
      "epoch": 0.656084656084656,
      "grad_norm": 0.060487277805805206,
      "learning_rate": 0.0002869516368619761,
      "loss": 1.4684,
      "step": 14756
    },
    {
      "epoch": 0.6561735805433284,
      "grad_norm": 0.060591381043195724,
      "learning_rate": 0.0002868193924317848,
      "loss": 1.4745,
      "step": 14758
    },
    {
      "epoch": 0.6562625050020008,
      "grad_norm": 0.059523772448301315,
      "learning_rate": 0.00028668716622521126,
      "loss": 1.4743,
      "step": 14760
    },
    {
      "epoch": 0.6563514294606732,
      "grad_norm": 0.058641474694013596,
      "learning_rate": 0.0002865549582535589,
      "loss": 1.4712,
      "step": 14762
    },
    {
      "epoch": 0.6564403539193455,
      "grad_norm": 0.06107234209775925,
      "learning_rate": 0.0002864227685281293,
      "loss": 1.473,
      "step": 14764
    },
    {
      "epoch": 0.6565292783780179,
      "grad_norm": 0.059071581810712814,
      "learning_rate": 0.0002862905970602225,
      "loss": 1.4703,
      "step": 14766
    },
    {
      "epoch": 0.6566182028366903,
      "grad_norm": 0.06059209629893303,
      "learning_rate": 0.0002861584438611375,
      "loss": 1.4692,
      "step": 14768
    },
    {
      "epoch": 0.6567071272953626,
      "grad_norm": 0.06286212056875229,
      "learning_rate": 0.000286026308942171,
      "loss": 1.4715,
      "step": 14770
    },
    {
      "epoch": 0.656796051754035,
      "grad_norm": 0.0598897784948349,
      "learning_rate": 0.0002858941923146187,
      "loss": 1.4647,
      "step": 14772
    },
    {
      "epoch": 0.6568849762127073,
      "grad_norm": 0.05993928760290146,
      "learning_rate": 0.0002857620939897741,
      "loss": 1.4691,
      "step": 14774
    },
    {
      "epoch": 0.6569739006713796,
      "grad_norm": 0.05909041315317154,
      "learning_rate": 0.00028563001397892987,
      "loss": 1.4722,
      "step": 14776
    },
    {
      "epoch": 0.657062825130052,
      "grad_norm": 0.061181001365184784,
      "learning_rate": 0.0002854979522933769,
      "loss": 1.4705,
      "step": 14778
    },
    {
      "epoch": 0.6571517495887244,
      "grad_norm": 0.05939111486077309,
      "learning_rate": 0.0002853659089444041,
      "loss": 1.4646,
      "step": 14780
    },
    {
      "epoch": 0.6572406740473967,
      "grad_norm": 0.060132816433906555,
      "learning_rate": 0.0002852338839432993,
      "loss": 1.473,
      "step": 14782
    },
    {
      "epoch": 0.6573295985060691,
      "grad_norm": 0.060787100344896317,
      "learning_rate": 0.0002851018773013486,
      "loss": 1.4695,
      "step": 14784
    },
    {
      "epoch": 0.6574185229647415,
      "grad_norm": 0.06070558354258537,
      "learning_rate": 0.00028496988902983644,
      "loss": 1.4726,
      "step": 14786
    },
    {
      "epoch": 0.6575074474234138,
      "grad_norm": 0.060331154614686966,
      "learning_rate": 0.0002848379191400456,
      "loss": 1.4664,
      "step": 14788
    },
    {
      "epoch": 0.6575963718820862,
      "grad_norm": 0.059469323605298996,
      "learning_rate": 0.00028470596764325795,
      "loss": 1.4688,
      "step": 14790
    },
    {
      "epoch": 0.6576852963407586,
      "grad_norm": 0.06065972149372101,
      "learning_rate": 0.0002845740345507527,
      "loss": 1.4715,
      "step": 14792
    },
    {
      "epoch": 0.6577742207994309,
      "grad_norm": 0.062771275639534,
      "learning_rate": 0.00028444211987380835,
      "loss": 1.4711,
      "step": 14794
    },
    {
      "epoch": 0.6578631452581032,
      "grad_norm": 0.05880487337708473,
      "learning_rate": 0.0002843102236237015,
      "loss": 1.4718,
      "step": 14796
    },
    {
      "epoch": 0.6579520697167756,
      "grad_norm": 0.05996553599834442,
      "learning_rate": 0.0002841783458117075,
      "loss": 1.4745,
      "step": 14798
    },
    {
      "epoch": 0.6580409941754479,
      "grad_norm": 0.06091736629605293,
      "learning_rate": 0.0002840464864490995,
      "loss": 1.473,
      "step": 14800
    },
    {
      "epoch": 0.6581299186341203,
      "grad_norm": 0.059029486030340195,
      "learning_rate": 0.0002839146455471492,
      "loss": 1.4726,
      "step": 14802
    },
    {
      "epoch": 0.6582188430927927,
      "grad_norm": 0.06295252591371536,
      "learning_rate": 0.0002837828231171277,
      "loss": 1.4668,
      "step": 14804
    },
    {
      "epoch": 0.658307767551465,
      "grad_norm": 0.06021355465054512,
      "learning_rate": 0.00028365101917030315,
      "loss": 1.4625,
      "step": 14806
    },
    {
      "epoch": 0.6583966920101374,
      "grad_norm": 0.058032263070344925,
      "learning_rate": 0.0002835192337179428,
      "loss": 1.4654,
      "step": 14808
    },
    {
      "epoch": 0.6584856164688098,
      "grad_norm": 0.05962485074996948,
      "learning_rate": 0.0002833874667713123,
      "loss": 1.4726,
      "step": 14810
    },
    {
      "epoch": 0.6585745409274821,
      "grad_norm": 0.061198100447654724,
      "learning_rate": 0.00028325571834167585,
      "loss": 1.4708,
      "step": 14812
    },
    {
      "epoch": 0.6586634653861545,
      "grad_norm": 0.05982755497097969,
      "learning_rate": 0.00028312398844029585,
      "loss": 1.4727,
      "step": 14814
    },
    {
      "epoch": 0.6587523898448269,
      "grad_norm": 0.06041552871465683,
      "learning_rate": 0.0002829922770784329,
      "loss": 1.4748,
      "step": 14816
    },
    {
      "epoch": 0.6588413143034991,
      "grad_norm": 0.0609116330742836,
      "learning_rate": 0.0002828605842673464,
      "loss": 1.4678,
      "step": 14818
    },
    {
      "epoch": 0.6589302387621715,
      "grad_norm": 0.06071128696203232,
      "learning_rate": 0.0002827289100182943,
      "loss": 1.4696,
      "step": 14820
    },
    {
      "epoch": 0.6590191632208439,
      "grad_norm": 0.060914020985364914,
      "learning_rate": 0.0002825972543425324,
      "loss": 1.4741,
      "step": 14822
    },
    {
      "epoch": 0.6591080876795162,
      "grad_norm": 0.05891665443778038,
      "learning_rate": 0.0002824656172513149,
      "loss": 1.4643,
      "step": 14824
    },
    {
      "epoch": 0.6591970121381886,
      "grad_norm": 0.06179531663656235,
      "learning_rate": 0.0002823339987558955,
      "loss": 1.472,
      "step": 14826
    },
    {
      "epoch": 0.659285936596861,
      "grad_norm": 0.06368506699800491,
      "learning_rate": 0.000282202398867525,
      "loss": 1.4742,
      "step": 14828
    },
    {
      "epoch": 0.6593748610555333,
      "grad_norm": 0.061851486563682556,
      "learning_rate": 0.0002820708175974531,
      "loss": 1.4685,
      "step": 14830
    },
    {
      "epoch": 0.6594637855142057,
      "grad_norm": 0.06155168637633324,
      "learning_rate": 0.00028193925495692817,
      "loss": 1.474,
      "step": 14832
    },
    {
      "epoch": 0.6595527099728781,
      "grad_norm": 0.06134020909667015,
      "learning_rate": 0.0002818077109571967,
      "loss": 1.4658,
      "step": 14834
    },
    {
      "epoch": 0.6596416344315504,
      "grad_norm": 0.060675278306007385,
      "learning_rate": 0.0002816761856095037,
      "loss": 1.4737,
      "step": 14836
    },
    {
      "epoch": 0.6597305588902228,
      "grad_norm": 0.060609932988882065,
      "learning_rate": 0.0002815446789250924,
      "loss": 1.4669,
      "step": 14838
    },
    {
      "epoch": 0.6598194833488951,
      "grad_norm": 0.0593482069671154,
      "learning_rate": 0.00028141319091520454,
      "loss": 1.4744,
      "step": 14840
    },
    {
      "epoch": 0.6599084078075674,
      "grad_norm": 0.061323292553424835,
      "learning_rate": 0.00028128172159108065,
      "loss": 1.4715,
      "step": 14842
    },
    {
      "epoch": 0.6599973322662398,
      "grad_norm": 0.05969291925430298,
      "learning_rate": 0.0002811502709639589,
      "loss": 1.4666,
      "step": 14844
    },
    {
      "epoch": 0.6600862567249122,
      "grad_norm": 0.05991503968834877,
      "learning_rate": 0.00028101883904507643,
      "loss": 1.4695,
      "step": 14846
    },
    {
      "epoch": 0.6601751811835845,
      "grad_norm": 0.06149459630250931,
      "learning_rate": 0.00028088742584566874,
      "loss": 1.4683,
      "step": 14848
    },
    {
      "epoch": 0.6602641056422569,
      "grad_norm": 0.05918089300394058,
      "learning_rate": 0.0002807560313769694,
      "loss": 1.4754,
      "step": 14850
    },
    {
      "epoch": 0.6603530301009293,
      "grad_norm": 0.06039484962821007,
      "learning_rate": 0.00028062465565021054,
      "loss": 1.4686,
      "step": 14852
    },
    {
      "epoch": 0.6604419545596016,
      "grad_norm": 0.060235604643821716,
      "learning_rate": 0.00028049329867662313,
      "loss": 1.465,
      "step": 14854
    },
    {
      "epoch": 0.660530879018274,
      "grad_norm": 0.05938028544187546,
      "learning_rate": 0.0002803619604674357,
      "loss": 1.4735,
      "step": 14856
    },
    {
      "epoch": 0.6606198034769464,
      "grad_norm": 0.05908706784248352,
      "learning_rate": 0.00028023064103387596,
      "loss": 1.4733,
      "step": 14858
    },
    {
      "epoch": 0.6607087279356187,
      "grad_norm": 0.06113129109144211,
      "learning_rate": 0.00028009934038716937,
      "loss": 1.4708,
      "step": 14860
    },
    {
      "epoch": 0.660797652394291,
      "grad_norm": 0.0593777671456337,
      "learning_rate": 0.00027996805853854024,
      "loss": 1.4708,
      "step": 14862
    },
    {
      "epoch": 0.6608865768529634,
      "grad_norm": 0.05915198102593422,
      "learning_rate": 0.00027983679549921126,
      "loss": 1.4649,
      "step": 14864
    },
    {
      "epoch": 0.6609755013116357,
      "grad_norm": 0.061543058604002,
      "learning_rate": 0.0002797055512804031,
      "loss": 1.474,
      "step": 14866
    },
    {
      "epoch": 0.6610644257703081,
      "grad_norm": 0.06172054633498192,
      "learning_rate": 0.0002795743258933352,
      "loss": 1.4728,
      "step": 14868
    },
    {
      "epoch": 0.6611533502289805,
      "grad_norm": 0.06032607704401016,
      "learning_rate": 0.0002794431193492254,
      "loss": 1.4695,
      "step": 14870
    },
    {
      "epoch": 0.6612422746876528,
      "grad_norm": 0.06093483418226242,
      "learning_rate": 0.0002793119316592897,
      "loss": 1.4651,
      "step": 14872
    },
    {
      "epoch": 0.6613311991463252,
      "grad_norm": 0.06096861511468887,
      "learning_rate": 0.0002791807628347425,
      "loss": 1.4723,
      "step": 14874
    },
    {
      "epoch": 0.6614201236049976,
      "grad_norm": 0.06007368490099907,
      "learning_rate": 0.00027904961288679705,
      "loss": 1.4758,
      "step": 14876
    },
    {
      "epoch": 0.66150904806367,
      "grad_norm": 0.06006438657641411,
      "learning_rate": 0.00027891848182666414,
      "loss": 1.4672,
      "step": 14878
    },
    {
      "epoch": 0.6615979725223423,
      "grad_norm": 0.05987627059221268,
      "learning_rate": 0.00027878736966555393,
      "loss": 1.4656,
      "step": 14880
    },
    {
      "epoch": 0.6616868969810147,
      "grad_norm": 0.06110451743006706,
      "learning_rate": 0.0002786562764146738,
      "loss": 1.4748,
      "step": 14882
    },
    {
      "epoch": 0.6617758214396869,
      "grad_norm": 0.060226909816265106,
      "learning_rate": 0.00027852520208523104,
      "loss": 1.4724,
      "step": 14884
    },
    {
      "epoch": 0.6618647458983593,
      "grad_norm": 0.06197149679064751,
      "learning_rate": 0.00027839414668842996,
      "loss": 1.4728,
      "step": 14886
    },
    {
      "epoch": 0.6619536703570317,
      "grad_norm": 0.059934891760349274,
      "learning_rate": 0.0002782631102354736,
      "loss": 1.4699,
      "step": 14888
    },
    {
      "epoch": 0.662042594815704,
      "grad_norm": 0.0607532262802124,
      "learning_rate": 0.0002781320927375637,
      "loss": 1.4705,
      "step": 14890
    },
    {
      "epoch": 0.6621315192743764,
      "grad_norm": 0.060573361814022064,
      "learning_rate": 0.0002780010942059005,
      "loss": 1.4681,
      "step": 14892
    },
    {
      "epoch": 0.6622204437330488,
      "grad_norm": 0.061003051698207855,
      "learning_rate": 0.0002778701146516819,
      "loss": 1.4741,
      "step": 14894
    },
    {
      "epoch": 0.6623093681917211,
      "grad_norm": 0.06043396145105362,
      "learning_rate": 0.0002777391540861048,
      "loss": 1.4689,
      "step": 14896
    },
    {
      "epoch": 0.6623982926503935,
      "grad_norm": 0.06064370647072792,
      "learning_rate": 0.0002776082125203645,
      "loss": 1.4736,
      "step": 14898
    },
    {
      "epoch": 0.6624872171090659,
      "grad_norm": 0.061219021677970886,
      "learning_rate": 0.000277477289965654,
      "loss": 1.467,
      "step": 14900
    },
    {
      "epoch": 0.6625761415677383,
      "grad_norm": 0.06040164828300476,
      "learning_rate": 0.00027734638643316534,
      "loss": 1.4677,
      "step": 14902
    },
    {
      "epoch": 0.6626650660264105,
      "grad_norm": 0.05900850147008896,
      "learning_rate": 0.00027721550193408884,
      "loss": 1.4704,
      "step": 14904
    },
    {
      "epoch": 0.6627539904850829,
      "grad_norm": 0.060347117483615875,
      "learning_rate": 0.00027708463647961324,
      "loss": 1.4718,
      "step": 14906
    },
    {
      "epoch": 0.6628429149437552,
      "grad_norm": 0.060297053307294846,
      "learning_rate": 0.0002769537900809252,
      "loss": 1.4679,
      "step": 14908
    },
    {
      "epoch": 0.6629318394024276,
      "grad_norm": 0.060264792293310165,
      "learning_rate": 0.0002768229627492097,
      "loss": 1.4718,
      "step": 14910
    },
    {
      "epoch": 0.6630207638611,
      "grad_norm": 0.06151338666677475,
      "learning_rate": 0.00027669215449565134,
      "loss": 1.466,
      "step": 14912
    },
    {
      "epoch": 0.6631096883197724,
      "grad_norm": 0.05946485698223114,
      "learning_rate": 0.0002765613653314316,
      "loss": 1.4713,
      "step": 14914
    },
    {
      "epoch": 0.6631986127784447,
      "grad_norm": 0.059629566967487335,
      "learning_rate": 0.0002764305952677309,
      "loss": 1.4705,
      "step": 14916
    },
    {
      "epoch": 0.6632875372371171,
      "grad_norm": 0.061307553201913834,
      "learning_rate": 0.0002762998443157281,
      "loss": 1.4723,
      "step": 14918
    },
    {
      "epoch": 0.6633764616957895,
      "grad_norm": 0.05890590697526932,
      "learning_rate": 0.0002761691124866005,
      "loss": 1.4718,
      "step": 14920
    },
    {
      "epoch": 0.6634653861544618,
      "grad_norm": 0.060055799782276154,
      "learning_rate": 0.00027603839979152366,
      "loss": 1.4676,
      "step": 14922
    },
    {
      "epoch": 0.6635543106131342,
      "grad_norm": 0.05887698009610176,
      "learning_rate": 0.00027590770624167114,
      "loss": 1.4651,
      "step": 14924
    },
    {
      "epoch": 0.6636432350718064,
      "grad_norm": 0.060526877641677856,
      "learning_rate": 0.0002757770318482156,
      "loss": 1.469,
      "step": 14926
    },
    {
      "epoch": 0.6637321595304788,
      "grad_norm": 0.06122924014925957,
      "learning_rate": 0.00027564637662232756,
      "loss": 1.4692,
      "step": 14928
    },
    {
      "epoch": 0.6638210839891512,
      "grad_norm": 0.061400964856147766,
      "learning_rate": 0.000275515740575176,
      "loss": 1.4705,
      "step": 14930
    },
    {
      "epoch": 0.6639100084478236,
      "grad_norm": 0.05991072207689285,
      "learning_rate": 0.0002753851237179278,
      "loss": 1.4702,
      "step": 14932
    },
    {
      "epoch": 0.6639989329064959,
      "grad_norm": 0.05974270775914192,
      "learning_rate": 0.0002752545260617494,
      "loss": 1.4725,
      "step": 14934
    },
    {
      "epoch": 0.6640878573651683,
      "grad_norm": 0.06016663461923599,
      "learning_rate": 0.00027512394761780443,
      "loss": 1.4658,
      "step": 14936
    },
    {
      "epoch": 0.6641767818238407,
      "grad_norm": 0.060294173657894135,
      "learning_rate": 0.0002749933883972555,
      "loss": 1.4701,
      "step": 14938
    },
    {
      "epoch": 0.664265706282513,
      "grad_norm": 0.06034474819898605,
      "learning_rate": 0.0002748628484112632,
      "loss": 1.4747,
      "step": 14940
    },
    {
      "epoch": 0.6643546307411854,
      "grad_norm": 0.0617527961730957,
      "learning_rate": 0.0002747323276709868,
      "loss": 1.4662,
      "step": 14942
    },
    {
      "epoch": 0.6644435551998578,
      "grad_norm": 0.05941234156489372,
      "learning_rate": 0.0002746018261875839,
      "loss": 1.4704,
      "step": 14944
    },
    {
      "epoch": 0.6645324796585301,
      "grad_norm": 0.06100812181830406,
      "learning_rate": 0.00027447134397220996,
      "loss": 1.4689,
      "step": 14946
    },
    {
      "epoch": 0.6646214041172024,
      "grad_norm": 0.06031123921275139,
      "learning_rate": 0.0002743408810360194,
      "loss": 1.4724,
      "step": 14948
    },
    {
      "epoch": 0.6647103285758748,
      "grad_norm": 0.05882007256150246,
      "learning_rate": 0.00027421043739016504,
      "loss": 1.4708,
      "step": 14950
    },
    {
      "epoch": 0.6647992530345471,
      "grad_norm": 0.06024463474750519,
      "learning_rate": 0.0002740800130457973,
      "loss": 1.465,
      "step": 14952
    },
    {
      "epoch": 0.6648881774932195,
      "grad_norm": 0.060123246163129807,
      "learning_rate": 0.00027394960801406564,
      "loss": 1.4727,
      "step": 14954
    },
    {
      "epoch": 0.6649771019518919,
      "grad_norm": 0.059828802943229675,
      "learning_rate": 0.0002738192223061178,
      "loss": 1.4666,
      "step": 14956
    },
    {
      "epoch": 0.6650660264105642,
      "grad_norm": 0.06216980516910553,
      "learning_rate": 0.0002736888559330994,
      "loss": 1.4677,
      "step": 14958
    },
    {
      "epoch": 0.6651549508692366,
      "grad_norm": 0.06128519028425217,
      "learning_rate": 0.00027355850890615504,
      "loss": 1.4701,
      "step": 14960
    },
    {
      "epoch": 0.665243875327909,
      "grad_norm": 0.062251608818769455,
      "learning_rate": 0.0002734281812364273,
      "loss": 1.4673,
      "step": 14962
    },
    {
      "epoch": 0.6653327997865813,
      "grad_norm": 0.06012072041630745,
      "learning_rate": 0.0002732978729350569,
      "loss": 1.4749,
      "step": 14964
    },
    {
      "epoch": 0.6654217242452537,
      "grad_norm": 0.06006190553307533,
      "learning_rate": 0.0002731675840131836,
      "loss": 1.4654,
      "step": 14966
    },
    {
      "epoch": 0.6655106487039261,
      "grad_norm": 0.06113257631659508,
      "learning_rate": 0.00027303731448194467,
      "loss": 1.4711,
      "step": 14968
    },
    {
      "epoch": 0.6655995731625983,
      "grad_norm": 0.06147268787026405,
      "learning_rate": 0.0002729070643524763,
      "loss": 1.4706,
      "step": 14970
    },
    {
      "epoch": 0.6656884976212707,
      "grad_norm": 0.06147456541657448,
      "learning_rate": 0.000272776833635913,
      "loss": 1.4646,
      "step": 14972
    },
    {
      "epoch": 0.6657774220799431,
      "grad_norm": 0.06183717027306557,
      "learning_rate": 0.0002726466223433871,
      "loss": 1.473,
      "step": 14974
    },
    {
      "epoch": 0.6658663465386154,
      "grad_norm": 0.06242965906858444,
      "learning_rate": 0.0002725164304860298,
      "loss": 1.4701,
      "step": 14976
    },
    {
      "epoch": 0.6659552709972878,
      "grad_norm": 0.06013220548629761,
      "learning_rate": 0.00027238625807497074,
      "loss": 1.4705,
      "step": 14978
    },
    {
      "epoch": 0.6660441954559602,
      "grad_norm": 0.06084521859884262,
      "learning_rate": 0.00027225610512133727,
      "loss": 1.4733,
      "step": 14980
    },
    {
      "epoch": 0.6661331199146325,
      "grad_norm": 0.06399312615394592,
      "learning_rate": 0.00027212597163625546,
      "loss": 1.4703,
      "step": 14982
    },
    {
      "epoch": 0.6662220443733049,
      "grad_norm": 0.06068024784326553,
      "learning_rate": 0.00027199585763085005,
      "loss": 1.4684,
      "step": 14984
    },
    {
      "epoch": 0.6663109688319773,
      "grad_norm": 0.05974745377898216,
      "learning_rate": 0.00027186576311624336,
      "loss": 1.4707,
      "step": 14986
    },
    {
      "epoch": 0.6663998932906496,
      "grad_norm": 0.061118923127651215,
      "learning_rate": 0.0002717356881035567,
      "loss": 1.4704,
      "step": 14988
    },
    {
      "epoch": 0.666488817749322,
      "grad_norm": 0.06090380996465683,
      "learning_rate": 0.000271605632603909,
      "loss": 1.4672,
      "step": 14990
    },
    {
      "epoch": 0.6665777422079943,
      "grad_norm": 0.060469526797533035,
      "learning_rate": 0.0002714755966284187,
      "loss": 1.4701,
      "step": 14992
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.05882229283452034,
      "learning_rate": 0.0002713455801882015,
      "loss": 1.4697,
      "step": 14994
    },
    {
      "epoch": 0.666755591125339,
      "grad_norm": 0.06075397878885269,
      "learning_rate": 0.00027121558329437157,
      "loss": 1.4723,
      "step": 14996
    },
    {
      "epoch": 0.6668445155840114,
      "grad_norm": 0.05918700248003006,
      "learning_rate": 0.0002710856059580418,
      "loss": 1.4664,
      "step": 14998
    },
    {
      "epoch": 0.6669334400426837,
      "grad_norm": 0.06100982427597046,
      "learning_rate": 0.0002709556481903234,
      "loss": 1.4677,
      "step": 15000
    },
    {
      "epoch": 0.6669334400426837,
      "eval_loss": 1.4535506963729858,
      "eval_runtime": 12.4672,
      "eval_samples_per_second": 554.252,
      "eval_steps_per_second": 69.302,
      "step": 15000
    },
    {
      "epoch": 0.6670223645013561,
      "grad_norm": 0.05966784805059433,
      "learning_rate": 0.00027082571000232525,
      "loss": 1.4677,
      "step": 15002
    },
    {
      "epoch": 0.6671112889600285,
      "grad_norm": 0.05990458279848099,
      "learning_rate": 0.0002706957914051555,
      "loss": 1.4647,
      "step": 15004
    },
    {
      "epoch": 0.6672002134187008,
      "grad_norm": 0.05977652221918106,
      "learning_rate": 0.00027056589240991995,
      "loss": 1.4705,
      "step": 15006
    },
    {
      "epoch": 0.6672891378773732,
      "grad_norm": 0.05973276495933533,
      "learning_rate": 0.00027043601302772325,
      "loss": 1.4662,
      "step": 15008
    },
    {
      "epoch": 0.6673780623360456,
      "grad_norm": 0.05999859794974327,
      "learning_rate": 0.00027030615326966757,
      "loss": 1.4668,
      "step": 15010
    },
    {
      "epoch": 0.6674669867947179,
      "grad_norm": 0.05877234786748886,
      "learning_rate": 0.00027017631314685416,
      "loss": 1.4632,
      "step": 15012
    },
    {
      "epoch": 0.6675559112533902,
      "grad_norm": 0.060439884662628174,
      "learning_rate": 0.0002700464926703825,
      "loss": 1.4734,
      "step": 15014
    },
    {
      "epoch": 0.6676448357120626,
      "grad_norm": 0.05888623371720314,
      "learning_rate": 0.00026991669185135,
      "loss": 1.4699,
      "step": 15016
    },
    {
      "epoch": 0.6677337601707349,
      "grad_norm": 0.05988156795501709,
      "learning_rate": 0.0002697869107008522,
      "loss": 1.4681,
      "step": 15018
    },
    {
      "epoch": 0.6678226846294073,
      "grad_norm": 0.05992647632956505,
      "learning_rate": 0.0002696571492299842,
      "loss": 1.4652,
      "step": 15020
    },
    {
      "epoch": 0.6679116090880797,
      "grad_norm": 0.06108422204852104,
      "learning_rate": 0.00026952740744983823,
      "loss": 1.4671,
      "step": 15022
    },
    {
      "epoch": 0.668000533546752,
      "grad_norm": 0.06095761060714722,
      "learning_rate": 0.0002693976853715049,
      "loss": 1.471,
      "step": 15024
    },
    {
      "epoch": 0.6680894580054244,
      "grad_norm": 0.061448004096746445,
      "learning_rate": 0.00026926798300607363,
      "loss": 1.4727,
      "step": 15026
    },
    {
      "epoch": 0.6681783824640968,
      "grad_norm": 0.05938434228301048,
      "learning_rate": 0.00026913830036463197,
      "loss": 1.4692,
      "step": 15028
    },
    {
      "epoch": 0.6682673069227691,
      "grad_norm": 0.06313709169626236,
      "learning_rate": 0.000269008637458266,
      "loss": 1.4659,
      "step": 15030
    },
    {
      "epoch": 0.6683562313814415,
      "grad_norm": 0.06244370713829994,
      "learning_rate": 0.00026887899429805953,
      "loss": 1.4704,
      "step": 15032
    },
    {
      "epoch": 0.6684451558401138,
      "grad_norm": 0.06154157221317291,
      "learning_rate": 0.000268749370895095,
      "loss": 1.4732,
      "step": 15034
    },
    {
      "epoch": 0.6685340802987861,
      "grad_norm": 0.06056886166334152,
      "learning_rate": 0.0002686197672604536,
      "loss": 1.4795,
      "step": 15036
    },
    {
      "epoch": 0.6686230047574585,
      "grad_norm": 0.060194727033376694,
      "learning_rate": 0.00026849018340521417,
      "loss": 1.4705,
      "step": 15038
    },
    {
      "epoch": 0.6687119292161309,
      "grad_norm": 0.05959151312708855,
      "learning_rate": 0.00026836061934045364,
      "loss": 1.4729,
      "step": 15040
    },
    {
      "epoch": 0.6688008536748032,
      "grad_norm": 0.06189076974987984,
      "learning_rate": 0.0002682310750772486,
      "loss": 1.471,
      "step": 15042
    },
    {
      "epoch": 0.6688897781334756,
      "grad_norm": 0.06022195145487785,
      "learning_rate": 0.00026810155062667243,
      "loss": 1.4695,
      "step": 15044
    },
    {
      "epoch": 0.668978702592148,
      "grad_norm": 0.059737030416727066,
      "learning_rate": 0.00026797204599979783,
      "loss": 1.4644,
      "step": 15046
    },
    {
      "epoch": 0.6690676270508203,
      "grad_norm": 0.05960986763238907,
      "learning_rate": 0.00026784256120769513,
      "loss": 1.4697,
      "step": 15048
    },
    {
      "epoch": 0.6691565515094927,
      "grad_norm": 0.060930218547582626,
      "learning_rate": 0.00026771309626143333,
      "loss": 1.4694,
      "step": 15050
    },
    {
      "epoch": 0.6692454759681651,
      "grad_norm": 0.05909661948680878,
      "learning_rate": 0.0002675836511720799,
      "loss": 1.4655,
      "step": 15052
    },
    {
      "epoch": 0.6693344004268375,
      "grad_norm": 0.06040756776928902,
      "learning_rate": 0.0002674542259507,
      "loss": 1.4687,
      "step": 15054
    },
    {
      "epoch": 0.6694233248855097,
      "grad_norm": 0.05938173085451126,
      "learning_rate": 0.0002673248206083577,
      "loss": 1.4736,
      "step": 15056
    },
    {
      "epoch": 0.6695122493441821,
      "grad_norm": 0.05951856076717377,
      "learning_rate": 0.0002671954351561152,
      "loss": 1.4709,
      "step": 15058
    },
    {
      "epoch": 0.6696011738028544,
      "grad_norm": 0.060165997594594955,
      "learning_rate": 0.0002670660696050328,
      "loss": 1.4699,
      "step": 15060
    },
    {
      "epoch": 0.6696900982615268,
      "grad_norm": 0.060211215168237686,
      "learning_rate": 0.0002669367239661692,
      "loss": 1.4756,
      "step": 15062
    },
    {
      "epoch": 0.6697790227201992,
      "grad_norm": 0.06052954122424126,
      "learning_rate": 0.00026680739825058164,
      "loss": 1.4744,
      "step": 15064
    },
    {
      "epoch": 0.6698679471788715,
      "grad_norm": 0.05990380421280861,
      "learning_rate": 0.0002666780924693252,
      "loss": 1.4674,
      "step": 15066
    },
    {
      "epoch": 0.6699568716375439,
      "grad_norm": 0.05932345241308212,
      "learning_rate": 0.00026654880663345385,
      "loss": 1.4717,
      "step": 15068
    },
    {
      "epoch": 0.6700457960962163,
      "grad_norm": 0.06290200352668762,
      "learning_rate": 0.00026641954075401903,
      "loss": 1.4679,
      "step": 15070
    },
    {
      "epoch": 0.6701347205548887,
      "grad_norm": 0.05994856730103493,
      "learning_rate": 0.00026629029484207125,
      "loss": 1.4655,
      "step": 15072
    },
    {
      "epoch": 0.670223645013561,
      "grad_norm": 0.06112265959382057,
      "learning_rate": 0.0002661610689086592,
      "loss": 1.4684,
      "step": 15074
    },
    {
      "epoch": 0.6703125694722334,
      "grad_norm": 0.06035906448960304,
      "learning_rate": 0.0002660318629648293,
      "loss": 1.4712,
      "step": 15076
    },
    {
      "epoch": 0.6704014939309056,
      "grad_norm": 0.060564734041690826,
      "learning_rate": 0.00026590267702162674,
      "loss": 1.4769,
      "step": 15078
    },
    {
      "epoch": 0.670490418389578,
      "grad_norm": 0.06007889285683632,
      "learning_rate": 0.0002657735110900953,
      "loss": 1.4714,
      "step": 15080
    },
    {
      "epoch": 0.6705793428482504,
      "grad_norm": 0.06008253991603851,
      "learning_rate": 0.000265644365181276,
      "loss": 1.4692,
      "step": 15082
    },
    {
      "epoch": 0.6706682673069227,
      "grad_norm": 0.05926406383514404,
      "learning_rate": 0.00026551523930620926,
      "loss": 1.471,
      "step": 15084
    },
    {
      "epoch": 0.6707571917655951,
      "grad_norm": 0.059321608394384384,
      "learning_rate": 0.0002653861334759333,
      "loss": 1.4656,
      "step": 15086
    },
    {
      "epoch": 0.6708461162242675,
      "grad_norm": 0.060134414583444595,
      "learning_rate": 0.0002652570477014844,
      "loss": 1.464,
      "step": 15088
    },
    {
      "epoch": 0.6709350406829399,
      "grad_norm": 0.0598330982029438,
      "learning_rate": 0.0002651279819938976,
      "loss": 1.468,
      "step": 15090
    },
    {
      "epoch": 0.6710239651416122,
      "grad_norm": 0.05979737266898155,
      "learning_rate": 0.0002649989363642061,
      "loss": 1.4636,
      "step": 15092
    },
    {
      "epoch": 0.6711128896002846,
      "grad_norm": 0.061708901077508926,
      "learning_rate": 0.0002648699108234409,
      "loss": 1.4689,
      "step": 15094
    },
    {
      "epoch": 0.671201814058957,
      "grad_norm": 0.059828344732522964,
      "learning_rate": 0.00026474090538263216,
      "loss": 1.4688,
      "step": 15096
    },
    {
      "epoch": 0.6712907385176293,
      "grad_norm": 0.060239676386117935,
      "learning_rate": 0.00026461192005280715,
      "loss": 1.4696,
      "step": 15098
    },
    {
      "epoch": 0.6713796629763016,
      "grad_norm": 0.06201807036995888,
      "learning_rate": 0.00026448295484499297,
      "loss": 1.4712,
      "step": 15100
    },
    {
      "epoch": 0.671468587434974,
      "grad_norm": 0.059881266206502914,
      "learning_rate": 0.0002643540097702136,
      "loss": 1.4662,
      "step": 15102
    },
    {
      "epoch": 0.6715575118936463,
      "grad_norm": 0.06092246621847153,
      "learning_rate": 0.0002642250848394918,
      "loss": 1.4675,
      "step": 15104
    },
    {
      "epoch": 0.6716464363523187,
      "grad_norm": 0.06052365154027939,
      "learning_rate": 0.00026409618006384883,
      "loss": 1.47,
      "step": 15106
    },
    {
      "epoch": 0.6717353608109911,
      "grad_norm": 0.06088147312402725,
      "learning_rate": 0.00026396729545430413,
      "loss": 1.4677,
      "step": 15108
    },
    {
      "epoch": 0.6718242852696634,
      "grad_norm": 0.05907963961362839,
      "learning_rate": 0.000263838431021875,
      "loss": 1.4646,
      "step": 15110
    },
    {
      "epoch": 0.6719132097283358,
      "grad_norm": 0.060525842010974884,
      "learning_rate": 0.00026370958677757753,
      "loss": 1.4652,
      "step": 15112
    },
    {
      "epoch": 0.6720021341870082,
      "grad_norm": 0.06022028997540474,
      "learning_rate": 0.0002635807627324258,
      "loss": 1.4651,
      "step": 15114
    },
    {
      "epoch": 0.6720910586456805,
      "grad_norm": 0.05875980481505394,
      "learning_rate": 0.0002634519588974326,
      "loss": 1.4643,
      "step": 15116
    },
    {
      "epoch": 0.6721799831043529,
      "grad_norm": 0.06109503656625748,
      "learning_rate": 0.00026332317528360836,
      "loss": 1.468,
      "step": 15118
    },
    {
      "epoch": 0.6722689075630253,
      "grad_norm": 0.05978439748287201,
      "learning_rate": 0.0002631944119019617,
      "loss": 1.4701,
      "step": 15120
    },
    {
      "epoch": 0.6723578320216975,
      "grad_norm": 0.059926699846982956,
      "learning_rate": 0.0002630656687635007,
      "loss": 1.4698,
      "step": 15122
    },
    {
      "epoch": 0.6724467564803699,
      "grad_norm": 0.06024140492081642,
      "learning_rate": 0.00026293694587923043,
      "loss": 1.4632,
      "step": 15124
    },
    {
      "epoch": 0.6725356809390423,
      "grad_norm": 0.05961621552705765,
      "learning_rate": 0.0002628082432601546,
      "loss": 1.4681,
      "step": 15126
    },
    {
      "epoch": 0.6726246053977146,
      "grad_norm": 0.05957215651869774,
      "learning_rate": 0.00026267956091727534,
      "loss": 1.4639,
      "step": 15128
    },
    {
      "epoch": 0.672713529856387,
      "grad_norm": 0.06190573051571846,
      "learning_rate": 0.00026255089886159313,
      "loss": 1.4712,
      "step": 15130
    },
    {
      "epoch": 0.6728024543150594,
      "grad_norm": 0.060655590146780014,
      "learning_rate": 0.0002624222571041066,
      "loss": 1.4714,
      "step": 15132
    },
    {
      "epoch": 0.6728913787737317,
      "grad_norm": 0.06115647032856941,
      "learning_rate": 0.0002622936356558123,
      "loss": 1.4678,
      "step": 15134
    },
    {
      "epoch": 0.6729803032324041,
      "grad_norm": 0.06026843562722206,
      "learning_rate": 0.00026216503452770564,
      "loss": 1.4616,
      "step": 15136
    },
    {
      "epoch": 0.6730692276910765,
      "grad_norm": 0.058876439929008484,
      "learning_rate": 0.0002620364537307801,
      "loss": 1.4623,
      "step": 15138
    },
    {
      "epoch": 0.6731581521497488,
      "grad_norm": 0.06088225543498993,
      "learning_rate": 0.00026190789327602696,
      "loss": 1.4688,
      "step": 15140
    },
    {
      "epoch": 0.6732470766084211,
      "grad_norm": 0.059225499629974365,
      "learning_rate": 0.0002617793531744363,
      "loss": 1.4674,
      "step": 15142
    },
    {
      "epoch": 0.6733360010670935,
      "grad_norm": 0.05934342369437218,
      "learning_rate": 0.0002616508334369966,
      "loss": 1.4642,
      "step": 15144
    },
    {
      "epoch": 0.6734249255257658,
      "grad_norm": 0.059114716947078705,
      "learning_rate": 0.000261522334074694,
      "loss": 1.4685,
      "step": 15146
    },
    {
      "epoch": 0.6735138499844382,
      "grad_norm": 0.05941977724432945,
      "learning_rate": 0.00026139385509851284,
      "loss": 1.4729,
      "step": 15148
    },
    {
      "epoch": 0.6736027744431106,
      "grad_norm": 0.06069924682378769,
      "learning_rate": 0.0002612653965194368,
      "loss": 1.469,
      "step": 15150
    },
    {
      "epoch": 0.6736916989017829,
      "grad_norm": 0.06208193302154541,
      "learning_rate": 0.0002611369583484466,
      "loss": 1.4714,
      "step": 15152
    },
    {
      "epoch": 0.6737806233604553,
      "grad_norm": 0.058625586330890656,
      "learning_rate": 0.000261008540596522,
      "loss": 1.4672,
      "step": 15154
    },
    {
      "epoch": 0.6738695478191277,
      "grad_norm": 0.06107405945658684,
      "learning_rate": 0.00026088014327464037,
      "loss": 1.4675,
      "step": 15156
    },
    {
      "epoch": 0.6739584722778,
      "grad_norm": 0.06048043444752693,
      "learning_rate": 0.00026075176639377785,
      "loss": 1.4662,
      "step": 15158
    },
    {
      "epoch": 0.6740473967364724,
      "grad_norm": 0.060831520706415176,
      "learning_rate": 0.00026062340996490884,
      "loss": 1.4712,
      "step": 15160
    },
    {
      "epoch": 0.6741363211951448,
      "grad_norm": 0.06307994574308395,
      "learning_rate": 0.0002604950739990055,
      "loss": 1.4725,
      "step": 15162
    },
    {
      "epoch": 0.674225245653817,
      "grad_norm": 0.06147250533103943,
      "learning_rate": 0.00026036675850703864,
      "loss": 1.4684,
      "step": 15164
    },
    {
      "epoch": 0.6743141701124894,
      "grad_norm": 0.05897139012813568,
      "learning_rate": 0.0002602384634999775,
      "loss": 1.4698,
      "step": 15166
    },
    {
      "epoch": 0.6744030945711618,
      "grad_norm": 0.06173459067940712,
      "learning_rate": 0.0002601101889887889,
      "loss": 1.472,
      "step": 15168
    },
    {
      "epoch": 0.6744920190298341,
      "grad_norm": 0.05954406037926674,
      "learning_rate": 0.00025998193498443856,
      "loss": 1.4712,
      "step": 15170
    },
    {
      "epoch": 0.6745809434885065,
      "grad_norm": 0.06112904101610184,
      "learning_rate": 0.0002598537014978903,
      "loss": 1.4697,
      "step": 15172
    },
    {
      "epoch": 0.6746698679471789,
      "grad_norm": 0.06085880100727081,
      "learning_rate": 0.00025972548854010576,
      "loss": 1.468,
      "step": 15174
    },
    {
      "epoch": 0.6747587924058512,
      "grad_norm": 0.061143457889556885,
      "learning_rate": 0.0002595972961220455,
      "loss": 1.4706,
      "step": 15176
    },
    {
      "epoch": 0.6748477168645236,
      "grad_norm": 0.060030248016119,
      "learning_rate": 0.0002594691242546675,
      "loss": 1.4674,
      "step": 15178
    },
    {
      "epoch": 0.674936641323196,
      "grad_norm": 0.05988433212041855,
      "learning_rate": 0.0002593409729489289,
      "loss": 1.4637,
      "step": 15180
    },
    {
      "epoch": 0.6750255657818683,
      "grad_norm": 0.06074235960841179,
      "learning_rate": 0.00025921284221578456,
      "loss": 1.4677,
      "step": 15182
    },
    {
      "epoch": 0.6751144902405407,
      "grad_norm": 0.06032219156622887,
      "learning_rate": 0.00025908473206618743,
      "loss": 1.4667,
      "step": 15184
    },
    {
      "epoch": 0.675203414699213,
      "grad_norm": 0.059529777616262436,
      "learning_rate": 0.00025895664251108906,
      "loss": 1.4663,
      "step": 15186
    },
    {
      "epoch": 0.6752923391578853,
      "grad_norm": 0.05997844040393829,
      "learning_rate": 0.00025882857356143935,
      "loss": 1.4708,
      "step": 15188
    },
    {
      "epoch": 0.6753812636165577,
      "grad_norm": 0.059802982956171036,
      "learning_rate": 0.00025870052522818576,
      "loss": 1.4654,
      "step": 15190
    },
    {
      "epoch": 0.6754701880752301,
      "grad_norm": 0.05979980528354645,
      "learning_rate": 0.00025857249752227467,
      "loss": 1.4636,
      "step": 15192
    },
    {
      "epoch": 0.6755591125339024,
      "grad_norm": 0.06097504124045372,
      "learning_rate": 0.0002584444904546506,
      "loss": 1.4648,
      "step": 15194
    },
    {
      "epoch": 0.6756480369925748,
      "grad_norm": 0.06036361679434776,
      "learning_rate": 0.00025831650403625585,
      "loss": 1.4674,
      "step": 15196
    },
    {
      "epoch": 0.6757369614512472,
      "grad_norm": 0.06019364297389984,
      "learning_rate": 0.0002581885382780313,
      "loss": 1.4697,
      "step": 15198
    },
    {
      "epoch": 0.6758258859099195,
      "grad_norm": 0.06247372180223465,
      "learning_rate": 0.0002580605931909164,
      "loss": 1.4712,
      "step": 15200
    },
    {
      "epoch": 0.6759148103685919,
      "grad_norm": 0.05914263427257538,
      "learning_rate": 0.000257932668785848,
      "loss": 1.4626,
      "step": 15202
    },
    {
      "epoch": 0.6760037348272643,
      "grad_norm": 0.06002722680568695,
      "learning_rate": 0.000257804765073762,
      "loss": 1.4704,
      "step": 15204
    },
    {
      "epoch": 0.6760926592859366,
      "grad_norm": 0.06022915244102478,
      "learning_rate": 0.0002576768820655916,
      "loss": 1.4735,
      "step": 15206
    },
    {
      "epoch": 0.6761815837446089,
      "grad_norm": 0.05858427286148071,
      "learning_rate": 0.0002575490197722696,
      "loss": 1.4651,
      "step": 15208
    },
    {
      "epoch": 0.6762705082032813,
      "grad_norm": 0.06035829707980156,
      "learning_rate": 0.00025742117820472585,
      "loss": 1.4657,
      "step": 15210
    },
    {
      "epoch": 0.6763594326619536,
      "grad_norm": 0.05977611243724823,
      "learning_rate": 0.0002572933573738886,
      "loss": 1.4648,
      "step": 15212
    },
    {
      "epoch": 0.676448357120626,
      "grad_norm": 0.06203882768750191,
      "learning_rate": 0.0002571655572906847,
      "loss": 1.4681,
      "step": 15214
    },
    {
      "epoch": 0.6765372815792984,
      "grad_norm": 0.0590372197329998,
      "learning_rate": 0.00025703777796603935,
      "loss": 1.4712,
      "step": 15216
    },
    {
      "epoch": 0.6766262060379707,
      "grad_norm": 0.06090830639004707,
      "learning_rate": 0.00025691001941087515,
      "loss": 1.4722,
      "step": 15218
    },
    {
      "epoch": 0.6767151304966431,
      "grad_norm": 0.060008611530065536,
      "learning_rate": 0.00025678228163611384,
      "loss": 1.467,
      "step": 15220
    },
    {
      "epoch": 0.6768040549553155,
      "grad_norm": 0.060110870748758316,
      "learning_rate": 0.0002566545646526749,
      "loss": 1.4608,
      "step": 15222
    },
    {
      "epoch": 0.6768929794139878,
      "grad_norm": 0.05973638966679573,
      "learning_rate": 0.00025652686847147635,
      "loss": 1.4685,
      "step": 15224
    },
    {
      "epoch": 0.6769819038726602,
      "grad_norm": 0.05918769910931587,
      "learning_rate": 0.000256399193103434,
      "loss": 1.468,
      "step": 15226
    },
    {
      "epoch": 0.6770708283313326,
      "grad_norm": 0.060597680509090424,
      "learning_rate": 0.0002562715385594618,
      "loss": 1.4648,
      "step": 15228
    },
    {
      "epoch": 0.6771597527900048,
      "grad_norm": 0.06197728216648102,
      "learning_rate": 0.00025614390485047294,
      "loss": 1.4742,
      "step": 15230
    },
    {
      "epoch": 0.6772486772486772,
      "grad_norm": 0.06051739677786827,
      "learning_rate": 0.00025601629198737777,
      "loss": 1.4697,
      "step": 15232
    },
    {
      "epoch": 0.6773376017073496,
      "grad_norm": 0.06158318743109703,
      "learning_rate": 0.0002558886999810849,
      "loss": 1.4675,
      "step": 15234
    },
    {
      "epoch": 0.677426526166022,
      "grad_norm": 0.05960030108690262,
      "learning_rate": 0.00025576112884250174,
      "loss": 1.4682,
      "step": 15236
    },
    {
      "epoch": 0.6775154506246943,
      "grad_norm": 0.059879280626773834,
      "learning_rate": 0.0002556335785825336,
      "loss": 1.4612,
      "step": 15238
    },
    {
      "epoch": 0.6776043750833667,
      "grad_norm": 0.06018189340829849,
      "learning_rate": 0.0002555060492120843,
      "loss": 1.4737,
      "step": 15240
    },
    {
      "epoch": 0.677693299542039,
      "grad_norm": 0.058622147887945175,
      "learning_rate": 0.00025537854074205514,
      "loss": 1.4675,
      "step": 15242
    },
    {
      "epoch": 0.6777822240007114,
      "grad_norm": 0.06036924943327904,
      "learning_rate": 0.0002552510531833463,
      "loss": 1.4663,
      "step": 15244
    },
    {
      "epoch": 0.6778711484593838,
      "grad_norm": 0.060898203402757645,
      "learning_rate": 0.0002551235865468562,
      "loss": 1.4729,
      "step": 15246
    },
    {
      "epoch": 0.6779600729180562,
      "grad_norm": 0.0606052428483963,
      "learning_rate": 0.0002549961408434809,
      "loss": 1.4669,
      "step": 15248
    },
    {
      "epoch": 0.6780489973767285,
      "grad_norm": 0.061549894511699677,
      "learning_rate": 0.0002548687160841152,
      "loss": 1.4674,
      "step": 15250
    },
    {
      "epoch": 0.6781379218354008,
      "grad_norm": 0.061437010765075684,
      "learning_rate": 0.000254741312279652,
      "loss": 1.4741,
      "step": 15252
    },
    {
      "epoch": 0.6782268462940731,
      "grad_norm": 0.06180231273174286,
      "learning_rate": 0.0002546139294409822,
      "loss": 1.4658,
      "step": 15254
    },
    {
      "epoch": 0.6783157707527455,
      "grad_norm": 0.06152801215648651,
      "learning_rate": 0.000254486567578995,
      "loss": 1.4682,
      "step": 15256
    },
    {
      "epoch": 0.6784046952114179,
      "grad_norm": 0.061020851135253906,
      "learning_rate": 0.00025435922670457813,
      "loss": 1.4651,
      "step": 15258
    },
    {
      "epoch": 0.6784936196700903,
      "grad_norm": 0.060287658125162125,
      "learning_rate": 0.00025423190682861686,
      "loss": 1.4621,
      "step": 15260
    },
    {
      "epoch": 0.6785825441287626,
      "grad_norm": 0.060152024030685425,
      "learning_rate": 0.0002541046079619955,
      "loss": 1.4688,
      "step": 15262
    },
    {
      "epoch": 0.678671468587435,
      "grad_norm": 0.06008097529411316,
      "learning_rate": 0.0002539773301155956,
      "loss": 1.4642,
      "step": 15264
    },
    {
      "epoch": 0.6787603930461074,
      "grad_norm": 0.060999296605587006,
      "learning_rate": 0.00025385007330029787,
      "loss": 1.4673,
      "step": 15266
    },
    {
      "epoch": 0.6788493175047797,
      "grad_norm": 0.06069813296198845,
      "learning_rate": 0.00025372283752698066,
      "loss": 1.4636,
      "step": 15268
    },
    {
      "epoch": 0.6789382419634521,
      "grad_norm": 0.060081273317337036,
      "learning_rate": 0.0002535956228065205,
      "loss": 1.4647,
      "step": 15270
    },
    {
      "epoch": 0.6790271664221244,
      "grad_norm": 0.061477821320295334,
      "learning_rate": 0.00025346842914979233,
      "loss": 1.4675,
      "step": 15272
    },
    {
      "epoch": 0.6791160908807967,
      "grad_norm": 0.058304693549871445,
      "learning_rate": 0.0002533412565676695,
      "loss": 1.4641,
      "step": 15274
    },
    {
      "epoch": 0.6792050153394691,
      "grad_norm": 0.0618242472410202,
      "learning_rate": 0.0002532141050710228,
      "loss": 1.4667,
      "step": 15276
    },
    {
      "epoch": 0.6792939397981415,
      "grad_norm": 0.060133617371320724,
      "learning_rate": 0.0002530869746707219,
      "loss": 1.464,
      "step": 15278
    },
    {
      "epoch": 0.6793828642568138,
      "grad_norm": 0.060968294739723206,
      "learning_rate": 0.00025295986537763487,
      "loss": 1.4707,
      "step": 15280
    },
    {
      "epoch": 0.6794717887154862,
      "grad_norm": 0.06188968941569328,
      "learning_rate": 0.000252832777202627,
      "loss": 1.4677,
      "step": 15282
    },
    {
      "epoch": 0.6795607131741586,
      "grad_norm": 0.06084349378943443,
      "learning_rate": 0.0002527057101565626,
      "loss": 1.4682,
      "step": 15284
    },
    {
      "epoch": 0.6796496376328309,
      "grad_norm": 0.06000750884413719,
      "learning_rate": 0.00025257866425030384,
      "loss": 1.4678,
      "step": 15286
    },
    {
      "epoch": 0.6797385620915033,
      "grad_norm": 0.05905289202928543,
      "learning_rate": 0.00025245163949471115,
      "loss": 1.4669,
      "step": 15288
    },
    {
      "epoch": 0.6798274865501757,
      "grad_norm": 0.060218751430511475,
      "learning_rate": 0.00025232463590064345,
      "loss": 1.4648,
      "step": 15290
    },
    {
      "epoch": 0.679916411008848,
      "grad_norm": 0.058075301349163055,
      "learning_rate": 0.0002521976534789571,
      "loss": 1.4685,
      "step": 15292
    },
    {
      "epoch": 0.6800053354675203,
      "grad_norm": 0.06031598150730133,
      "learning_rate": 0.0002520706922405074,
      "loss": 1.4672,
      "step": 15294
    },
    {
      "epoch": 0.6800942599261927,
      "grad_norm": 0.059682924300432205,
      "learning_rate": 0.00025194375219614773,
      "loss": 1.4666,
      "step": 15296
    },
    {
      "epoch": 0.680183184384865,
      "grad_norm": 0.05814158171415329,
      "learning_rate": 0.00025181683335672903,
      "loss": 1.4643,
      "step": 15298
    },
    {
      "epoch": 0.6802721088435374,
      "grad_norm": 0.06083943694829941,
      "learning_rate": 0.0002516899357331012,
      "loss": 1.4697,
      "step": 15300
    },
    {
      "epoch": 0.6803610333022098,
      "grad_norm": 0.06108289211988449,
      "learning_rate": 0.00025156305933611204,
      "loss": 1.4643,
      "step": 15302
    },
    {
      "epoch": 0.6804499577608821,
      "grad_norm": 0.05812264606356621,
      "learning_rate": 0.00025143620417660716,
      "loss": 1.4682,
      "step": 15304
    },
    {
      "epoch": 0.6805388822195545,
      "grad_norm": 0.06007948890328407,
      "learning_rate": 0.00025130937026543107,
      "loss": 1.4659,
      "step": 15306
    },
    {
      "epoch": 0.6806278066782269,
      "grad_norm": 0.062025126069784164,
      "learning_rate": 0.0002511825576134259,
      "loss": 1.4717,
      "step": 15308
    },
    {
      "epoch": 0.6807167311368992,
      "grad_norm": 0.05901065096259117,
      "learning_rate": 0.00025105576623143234,
      "loss": 1.4637,
      "step": 15310
    },
    {
      "epoch": 0.6808056555955716,
      "grad_norm": 0.06109865382313728,
      "learning_rate": 0.00025092899613028897,
      "loss": 1.4713,
      "step": 15312
    },
    {
      "epoch": 0.680894580054244,
      "grad_norm": 0.06021618843078613,
      "learning_rate": 0.0002508022473208323,
      "loss": 1.4658,
      "step": 15314
    },
    {
      "epoch": 0.6809835045129162,
      "grad_norm": 0.0613398440182209,
      "learning_rate": 0.00025067551981389804,
      "loss": 1.4682,
      "step": 15316
    },
    {
      "epoch": 0.6810724289715886,
      "grad_norm": 0.0613284669816494,
      "learning_rate": 0.00025054881362031914,
      "loss": 1.4665,
      "step": 15318
    },
    {
      "epoch": 0.681161353430261,
      "grad_norm": 0.060059648007154465,
      "learning_rate": 0.0002504221287509267,
      "loss": 1.4659,
      "step": 15320
    },
    {
      "epoch": 0.6812502778889333,
      "grad_norm": 0.05834935978055,
      "learning_rate": 0.00025029546521655055,
      "loss": 1.4698,
      "step": 15322
    },
    {
      "epoch": 0.6813392023476057,
      "grad_norm": 0.059527888894081116,
      "learning_rate": 0.00025016882302801845,
      "loss": 1.4657,
      "step": 15324
    },
    {
      "epoch": 0.6814281268062781,
      "grad_norm": 0.06185444816946983,
      "learning_rate": 0.00025004220219615654,
      "loss": 1.4731,
      "step": 15326
    },
    {
      "epoch": 0.6815170512649504,
      "grad_norm": 0.058973994106054306,
      "learning_rate": 0.0002499156027317885,
      "loss": 1.4675,
      "step": 15328
    },
    {
      "epoch": 0.6816059757236228,
      "grad_norm": 0.060899462550878525,
      "learning_rate": 0.00024978902464573683,
      "loss": 1.4688,
      "step": 15330
    },
    {
      "epoch": 0.6816949001822952,
      "grad_norm": 0.06071661040186882,
      "learning_rate": 0.0002496624679488222,
      "loss": 1.4631,
      "step": 15332
    },
    {
      "epoch": 0.6817838246409675,
      "grad_norm": 0.059728339314460754,
      "learning_rate": 0.00024953593265186297,
      "loss": 1.4672,
      "step": 15334
    },
    {
      "epoch": 0.6818727490996399,
      "grad_norm": 0.058348335325717926,
      "learning_rate": 0.00024940941876567564,
      "loss": 1.4617,
      "step": 15336
    },
    {
      "epoch": 0.6819616735583122,
      "grad_norm": 0.05913185328245163,
      "learning_rate": 0.00024928292630107586,
      "loss": 1.4696,
      "step": 15338
    },
    {
      "epoch": 0.6820505980169845,
      "grad_norm": 0.062302857637405396,
      "learning_rate": 0.00024915645526887636,
      "loss": 1.47,
      "step": 15340
    },
    {
      "epoch": 0.6821395224756569,
      "grad_norm": 0.062192488461732864,
      "learning_rate": 0.0002490300056798884,
      "loss": 1.4651,
      "step": 15342
    },
    {
      "epoch": 0.6822284469343293,
      "grad_norm": 0.061173710972070694,
      "learning_rate": 0.00024890357754492147,
      "loss": 1.4652,
      "step": 15344
    },
    {
      "epoch": 0.6823173713930016,
      "grad_norm": 0.06143614277243614,
      "learning_rate": 0.00024877717087478334,
      "loss": 1.4693,
      "step": 15346
    },
    {
      "epoch": 0.682406295851674,
      "grad_norm": 0.05982348322868347,
      "learning_rate": 0.00024865078568027995,
      "loss": 1.4709,
      "step": 15348
    },
    {
      "epoch": 0.6824952203103464,
      "grad_norm": 0.05981658026576042,
      "learning_rate": 0.0002485244219722148,
      "loss": 1.467,
      "step": 15350
    },
    {
      "epoch": 0.6825841447690187,
      "grad_norm": 0.05860120430588722,
      "learning_rate": 0.0002483980797613903,
      "loss": 1.4644,
      "step": 15352
    },
    {
      "epoch": 0.6826730692276911,
      "grad_norm": 0.05877653881907463,
      "learning_rate": 0.00024827175905860687,
      "loss": 1.4707,
      "step": 15354
    },
    {
      "epoch": 0.6827619936863635,
      "grad_norm": 0.061559949070215225,
      "learning_rate": 0.00024814545987466284,
      "loss": 1.4666,
      "step": 15356
    },
    {
      "epoch": 0.6828509181450358,
      "grad_norm": 0.05921272560954094,
      "learning_rate": 0.00024801918222035446,
      "loss": 1.4688,
      "step": 15358
    },
    {
      "epoch": 0.6829398426037081,
      "grad_norm": 0.061051640659570694,
      "learning_rate": 0.0002478929261064772,
      "loss": 1.4603,
      "step": 15360
    },
    {
      "epoch": 0.6830287670623805,
      "grad_norm": 0.059043582528829575,
      "learning_rate": 0.00024776669154382336,
      "loss": 1.463,
      "step": 15362
    },
    {
      "epoch": 0.6831176915210528,
      "grad_norm": 0.059850361198186874,
      "learning_rate": 0.0002476404785431845,
      "loss": 1.4664,
      "step": 15364
    },
    {
      "epoch": 0.6832066159797252,
      "grad_norm": 0.06030626222491264,
      "learning_rate": 0.00024751428711534946,
      "loss": 1.4634,
      "step": 15366
    },
    {
      "epoch": 0.6832955404383976,
      "grad_norm": 0.058735646307468414,
      "learning_rate": 0.00024738811727110583,
      "loss": 1.4638,
      "step": 15368
    },
    {
      "epoch": 0.6833844648970699,
      "grad_norm": 0.05930692330002785,
      "learning_rate": 0.0002472619690212393,
      "loss": 1.4693,
      "step": 15370
    },
    {
      "epoch": 0.6834733893557423,
      "grad_norm": 0.0614553838968277,
      "learning_rate": 0.0002471358423765333,
      "loss": 1.4653,
      "step": 15372
    },
    {
      "epoch": 0.6835623138144147,
      "grad_norm": 0.06112890690565109,
      "learning_rate": 0.00024700973734776983,
      "loss": 1.4676,
      "step": 15374
    },
    {
      "epoch": 0.683651238273087,
      "grad_norm": 0.059310752898454666,
      "learning_rate": 0.0002468836539457291,
      "loss": 1.4685,
      "step": 15376
    },
    {
      "epoch": 0.6837401627317594,
      "grad_norm": 0.06061893329024315,
      "learning_rate": 0.00024675759218118886,
      "loss": 1.468,
      "step": 15378
    },
    {
      "epoch": 0.6838290871904318,
      "grad_norm": 0.060217492282390594,
      "learning_rate": 0.00024663155206492565,
      "loss": 1.4595,
      "step": 15380
    },
    {
      "epoch": 0.683918011649104,
      "grad_norm": 0.059973131865262985,
      "learning_rate": 0.00024650553360771417,
      "loss": 1.4674,
      "step": 15382
    },
    {
      "epoch": 0.6840069361077764,
      "grad_norm": 0.059515953063964844,
      "learning_rate": 0.00024637953682032655,
      "loss": 1.4681,
      "step": 15384
    },
    {
      "epoch": 0.6840958605664488,
      "grad_norm": 0.05842442810535431,
      "learning_rate": 0.00024625356171353375,
      "loss": 1.4665,
      "step": 15386
    },
    {
      "epoch": 0.6841847850251211,
      "grad_norm": 0.06066687032580376,
      "learning_rate": 0.000246127608298105,
      "loss": 1.4712,
      "step": 15388
    },
    {
      "epoch": 0.6842737094837935,
      "grad_norm": 0.061743155121803284,
      "learning_rate": 0.0002460016765848068,
      "loss": 1.4652,
      "step": 15390
    },
    {
      "epoch": 0.6843626339424659,
      "grad_norm": 0.0611640103161335,
      "learning_rate": 0.00024587576658440484,
      "loss": 1.4622,
      "step": 15392
    },
    {
      "epoch": 0.6844515584011382,
      "grad_norm": 0.059736188501119614,
      "learning_rate": 0.0002457498783076621,
      "loss": 1.4646,
      "step": 15394
    },
    {
      "epoch": 0.6845404828598106,
      "grad_norm": 0.060952868312597275,
      "learning_rate": 0.00024562401176534017,
      "loss": 1.4665,
      "step": 15396
    },
    {
      "epoch": 0.684629407318483,
      "grad_norm": 0.060799382627010345,
      "learning_rate": 0.0002454981669681989,
      "loss": 1.4682,
      "step": 15398
    },
    {
      "epoch": 0.6847183317771554,
      "grad_norm": 0.05983178690075874,
      "learning_rate": 0.00024537234392699566,
      "loss": 1.4653,
      "step": 15400
    },
    {
      "epoch": 0.6848072562358276,
      "grad_norm": 0.05988869071006775,
      "learning_rate": 0.0002452465426524866,
      "loss": 1.4679,
      "step": 15402
    },
    {
      "epoch": 0.6848961806945,
      "grad_norm": 0.059868037700653076,
      "learning_rate": 0.00024512076315542594,
      "loss": 1.4633,
      "step": 15404
    },
    {
      "epoch": 0.6849851051531723,
      "grad_norm": 0.0584200955927372,
      "learning_rate": 0.00024499500544656537,
      "loss": 1.4682,
      "step": 15406
    },
    {
      "epoch": 0.6850740296118447,
      "grad_norm": 0.059604521840810776,
      "learning_rate": 0.0002448692695366556,
      "loss": 1.4674,
      "step": 15408
    },
    {
      "epoch": 0.6851629540705171,
      "grad_norm": 0.05988859757781029,
      "learning_rate": 0.00024474355543644517,
      "loss": 1.4599,
      "step": 15410
    },
    {
      "epoch": 0.6852518785291895,
      "grad_norm": 0.06065924093127251,
      "learning_rate": 0.00024461786315668033,
      "loss": 1.4673,
      "step": 15412
    },
    {
      "epoch": 0.6853408029878618,
      "grad_norm": 0.060476917773485184,
      "learning_rate": 0.0002444921927081061,
      "loss": 1.4695,
      "step": 15414
    },
    {
      "epoch": 0.6854297274465342,
      "grad_norm": 0.060220107436180115,
      "learning_rate": 0.00024436654410146493,
      "loss": 1.4639,
      "step": 15416
    },
    {
      "epoch": 0.6855186519052066,
      "grad_norm": 0.059693094342947006,
      "learning_rate": 0.0002442409173474984,
      "loss": 1.4677,
      "step": 15418
    },
    {
      "epoch": 0.6856075763638789,
      "grad_norm": 0.06192352995276451,
      "learning_rate": 0.00024411531245694545,
      "loss": 1.4658,
      "step": 15420
    },
    {
      "epoch": 0.6856965008225513,
      "grad_norm": 0.05954059958457947,
      "learning_rate": 0.00024398972944054297,
      "loss": 1.4665,
      "step": 15422
    },
    {
      "epoch": 0.6857854252812235,
      "grad_norm": 0.05900617316365242,
      "learning_rate": 0.00024386416830902668,
      "loss": 1.4659,
      "step": 15424
    },
    {
      "epoch": 0.6858743497398959,
      "grad_norm": 0.060082681477069855,
      "learning_rate": 0.00024373862907313017,
      "loss": 1.466,
      "step": 15426
    },
    {
      "epoch": 0.6859632741985683,
      "grad_norm": 0.06016187369823456,
      "learning_rate": 0.0002436131117435848,
      "loss": 1.4625,
      "step": 15428
    },
    {
      "epoch": 0.6860521986572407,
      "grad_norm": 0.0597849078476429,
      "learning_rate": 0.0002434876163311205,
      "loss": 1.4654,
      "step": 15430
    },
    {
      "epoch": 0.686141123115913,
      "grad_norm": 0.060573283582925797,
      "learning_rate": 0.00024336214284646513,
      "loss": 1.4613,
      "step": 15432
    },
    {
      "epoch": 0.6862300475745854,
      "grad_norm": 0.05994643270969391,
      "learning_rate": 0.00024323669130034498,
      "loss": 1.4657,
      "step": 15434
    },
    {
      "epoch": 0.6863189720332578,
      "grad_norm": 0.060550957918167114,
      "learning_rate": 0.0002431112617034838,
      "loss": 1.4726,
      "step": 15436
    },
    {
      "epoch": 0.6864078964919301,
      "grad_norm": 0.06180041283369064,
      "learning_rate": 0.000242985854066604,
      "loss": 1.4693,
      "step": 15438
    },
    {
      "epoch": 0.6864968209506025,
      "grad_norm": 0.05975709855556488,
      "learning_rate": 0.00024286046840042618,
      "loss": 1.4725,
      "step": 15440
    },
    {
      "epoch": 0.6865857454092749,
      "grad_norm": 0.06140446290373802,
      "learning_rate": 0.00024273510471566878,
      "loss": 1.4673,
      "step": 15442
    },
    {
      "epoch": 0.6866746698679472,
      "grad_norm": 0.06379693001508713,
      "learning_rate": 0.00024260976302304792,
      "loss": 1.4751,
      "step": 15444
    },
    {
      "epoch": 0.6867635943266195,
      "grad_norm": 0.05907401442527771,
      "learning_rate": 0.00024248444333327924,
      "loss": 1.4691,
      "step": 15446
    },
    {
      "epoch": 0.6868525187852919,
      "grad_norm": 0.05944070219993591,
      "learning_rate": 0.00024235914565707496,
      "loss": 1.4702,
      "step": 15448
    },
    {
      "epoch": 0.6869414432439642,
      "grad_norm": 0.061419710516929626,
      "learning_rate": 0.00024223387000514647,
      "loss": 1.4651,
      "step": 15450
    },
    {
      "epoch": 0.6870303677026366,
      "grad_norm": 0.05813660845160484,
      "learning_rate": 0.00024210861638820252,
      "loss": 1.4564,
      "step": 15452
    },
    {
      "epoch": 0.687119292161309,
      "grad_norm": 0.05996819585561752,
      "learning_rate": 0.00024198338481695053,
      "loss": 1.4653,
      "step": 15454
    },
    {
      "epoch": 0.6872082166199813,
      "grad_norm": 0.060112178325653076,
      "learning_rate": 0.00024185817530209604,
      "loss": 1.4666,
      "step": 15456
    },
    {
      "epoch": 0.6872971410786537,
      "grad_norm": 0.05905492231249809,
      "learning_rate": 0.00024173298785434217,
      "loss": 1.4681,
      "step": 15458
    },
    {
      "epoch": 0.6873860655373261,
      "grad_norm": 0.06022144481539726,
      "learning_rate": 0.00024160782248439056,
      "loss": 1.465,
      "step": 15460
    },
    {
      "epoch": 0.6874749899959984,
      "grad_norm": 0.05972209945321083,
      "learning_rate": 0.00024148267920294126,
      "loss": 1.4695,
      "step": 15462
    },
    {
      "epoch": 0.6875639144546708,
      "grad_norm": 0.060659270733594894,
      "learning_rate": 0.00024135755802069176,
      "loss": 1.4673,
      "step": 15464
    },
    {
      "epoch": 0.6876528389133432,
      "grad_norm": 0.060310158878564835,
      "learning_rate": 0.00024123245894833756,
      "loss": 1.4646,
      "step": 15466
    },
    {
      "epoch": 0.6877417633720154,
      "grad_norm": 0.06041952967643738,
      "learning_rate": 0.00024110738199657357,
      "loss": 1.4709,
      "step": 15468
    },
    {
      "epoch": 0.6878306878306878,
      "grad_norm": 0.06220192089676857,
      "learning_rate": 0.00024098232717609125,
      "loss": 1.4675,
      "step": 15470
    },
    {
      "epoch": 0.6879196122893602,
      "grad_norm": 0.060216281563043594,
      "learning_rate": 0.00024085729449758127,
      "loss": 1.4641,
      "step": 15472
    },
    {
      "epoch": 0.6880085367480325,
      "grad_norm": 0.06173881143331528,
      "learning_rate": 0.00024073228397173159,
      "loss": 1.4647,
      "step": 15474
    },
    {
      "epoch": 0.6880974612067049,
      "grad_norm": 0.06371451169252396,
      "learning_rate": 0.0002406072956092288,
      "loss": 1.4726,
      "step": 15476
    },
    {
      "epoch": 0.6881863856653773,
      "grad_norm": 0.060280535370111465,
      "learning_rate": 0.00024048232942075766,
      "loss": 1.4632,
      "step": 15478
    },
    {
      "epoch": 0.6882753101240496,
      "grad_norm": 0.06460434943437576,
      "learning_rate": 0.00024035738541700047,
      "loss": 1.4655,
      "step": 15480
    },
    {
      "epoch": 0.688364234582722,
      "grad_norm": 0.060472600162029266,
      "learning_rate": 0.00024023246360863821,
      "loss": 1.4621,
      "step": 15482
    },
    {
      "epoch": 0.6884531590413944,
      "grad_norm": 0.060170937329530716,
      "learning_rate": 0.00024010756400634993,
      "loss": 1.4619,
      "step": 15484
    },
    {
      "epoch": 0.6885420835000667,
      "grad_norm": 0.06201303005218506,
      "learning_rate": 0.00023998268662081212,
      "loss": 1.467,
      "step": 15486
    },
    {
      "epoch": 0.6886310079587391,
      "grad_norm": 0.06021099537611008,
      "learning_rate": 0.00023985783146270018,
      "loss": 1.465,
      "step": 15488
    },
    {
      "epoch": 0.6887199324174114,
      "grad_norm": 0.05890370160341263,
      "learning_rate": 0.00023973299854268743,
      "loss": 1.4657,
      "step": 15490
    },
    {
      "epoch": 0.6888088568760837,
      "grad_norm": 0.06088082119822502,
      "learning_rate": 0.00023960818787144468,
      "loss": 1.4669,
      "step": 15492
    },
    {
      "epoch": 0.6888977813347561,
      "grad_norm": 0.06058606877923012,
      "learning_rate": 0.00023948339945964155,
      "loss": 1.4647,
      "step": 15494
    },
    {
      "epoch": 0.6889867057934285,
      "grad_norm": 0.06058656796813011,
      "learning_rate": 0.00023935863331794567,
      "loss": 1.4713,
      "step": 15496
    },
    {
      "epoch": 0.6890756302521008,
      "grad_norm": 0.06112731248140335,
      "learning_rate": 0.0002392338894570223,
      "loss": 1.4691,
      "step": 15498
    },
    {
      "epoch": 0.6891645547107732,
      "grad_norm": 0.0599660649895668,
      "learning_rate": 0.00023910916788753533,
      "loss": 1.4686,
      "step": 15500
    },
    {
      "epoch": 0.6891645547107732,
      "eval_loss": 1.4509516954421997,
      "eval_runtime": 12.4424,
      "eval_samples_per_second": 555.36,
      "eval_steps_per_second": 69.44,
      "step": 15500
    },
    {
      "epoch": 0.6892534791694456,
      "grad_norm": 0.059089768677949905,
      "learning_rate": 0.00023898446862014605,
      "loss": 1.4649,
      "step": 15502
    },
    {
      "epoch": 0.6893424036281179,
      "grad_norm": 0.05946287512779236,
      "learning_rate": 0.00023885979166551503,
      "loss": 1.4652,
      "step": 15504
    },
    {
      "epoch": 0.6894313280867903,
      "grad_norm": 0.059581682085990906,
      "learning_rate": 0.00023873513703429983,
      "loss": 1.4743,
      "step": 15506
    },
    {
      "epoch": 0.6895202525454627,
      "grad_norm": 0.0592639334499836,
      "learning_rate": 0.00023861050473715628,
      "loss": 1.4693,
      "step": 15508
    },
    {
      "epoch": 0.6896091770041349,
      "grad_norm": 0.06175572797656059,
      "learning_rate": 0.0002384858947847387,
      "loss": 1.4681,
      "step": 15510
    },
    {
      "epoch": 0.6896981014628073,
      "grad_norm": 0.05914521589875221,
      "learning_rate": 0.00023836130718769943,
      "loss": 1.4664,
      "step": 15512
    },
    {
      "epoch": 0.6897870259214797,
      "grad_norm": 0.060408368706703186,
      "learning_rate": 0.00023823674195668849,
      "loss": 1.4636,
      "step": 15514
    },
    {
      "epoch": 0.689875950380152,
      "grad_norm": 0.05882785841822624,
      "learning_rate": 0.00023811219910235438,
      "loss": 1.4699,
      "step": 15516
    },
    {
      "epoch": 0.6899648748388244,
      "grad_norm": 0.061456430703401566,
      "learning_rate": 0.00023798767863534377,
      "loss": 1.4662,
      "step": 15518
    },
    {
      "epoch": 0.6900537992974968,
      "grad_norm": 0.059389494359493256,
      "learning_rate": 0.00023786318056630092,
      "loss": 1.4615,
      "step": 15520
    },
    {
      "epoch": 0.6901427237561691,
      "grad_norm": 0.06065116822719574,
      "learning_rate": 0.00023773870490586868,
      "loss": 1.4686,
      "step": 15522
    },
    {
      "epoch": 0.6902316482148415,
      "grad_norm": 0.06044527515769005,
      "learning_rate": 0.00023761425166468736,
      "loss": 1.4611,
      "step": 15524
    },
    {
      "epoch": 0.6903205726735139,
      "grad_norm": 0.060601379722356796,
      "learning_rate": 0.00023748982085339647,
      "loss": 1.4673,
      "step": 15526
    },
    {
      "epoch": 0.6904094971321862,
      "grad_norm": 0.05913134291768074,
      "learning_rate": 0.0002373654124826326,
      "loss": 1.4657,
      "step": 15528
    },
    {
      "epoch": 0.6904984215908586,
      "grad_norm": 0.06059577316045761,
      "learning_rate": 0.00023724102656303044,
      "loss": 1.4623,
      "step": 15530
    },
    {
      "epoch": 0.6905873460495309,
      "grad_norm": 0.05980486422777176,
      "learning_rate": 0.00023711666310522324,
      "loss": 1.4603,
      "step": 15532
    },
    {
      "epoch": 0.6906762705082032,
      "grad_norm": 0.05991894379258156,
      "learning_rate": 0.0002369923221198424,
      "loss": 1.467,
      "step": 15534
    },
    {
      "epoch": 0.6907651949668756,
      "grad_norm": 0.06080161780118942,
      "learning_rate": 0.0002368680036175167,
      "loss": 1.4608,
      "step": 15536
    },
    {
      "epoch": 0.690854119425548,
      "grad_norm": 0.06057791784405708,
      "learning_rate": 0.0002367437076088736,
      "loss": 1.4728,
      "step": 15538
    },
    {
      "epoch": 0.6909430438842203,
      "grad_norm": 0.05937578156590462,
      "learning_rate": 0.00023661943410453856,
      "loss": 1.4685,
      "step": 15540
    },
    {
      "epoch": 0.6910319683428927,
      "grad_norm": 0.06065383553504944,
      "learning_rate": 0.00023649518311513512,
      "loss": 1.4729,
      "step": 15542
    },
    {
      "epoch": 0.6911208928015651,
      "grad_norm": 0.062062423676252365,
      "learning_rate": 0.0002363709546512845,
      "loss": 1.4643,
      "step": 15544
    },
    {
      "epoch": 0.6912098172602374,
      "grad_norm": 0.06103227660059929,
      "learning_rate": 0.0002362467487236064,
      "loss": 1.4647,
      "step": 15546
    },
    {
      "epoch": 0.6912987417189098,
      "grad_norm": 0.05918658524751663,
      "learning_rate": 0.00023612256534271875,
      "loss": 1.4664,
      "step": 15548
    },
    {
      "epoch": 0.6913876661775822,
      "grad_norm": 0.05857004597783089,
      "learning_rate": 0.00023599840451923716,
      "loss": 1.4638,
      "step": 15550
    },
    {
      "epoch": 0.6914765906362546,
      "grad_norm": 0.06153101101517677,
      "learning_rate": 0.0002358742662637749,
      "loss": 1.4681,
      "step": 15552
    },
    {
      "epoch": 0.6915655150949268,
      "grad_norm": 0.06063772737979889,
      "learning_rate": 0.0002357501505869447,
      "loss": 1.4705,
      "step": 15554
    },
    {
      "epoch": 0.6916544395535992,
      "grad_norm": 0.059646911919116974,
      "learning_rate": 0.000235626057499356,
      "loss": 1.4624,
      "step": 15556
    },
    {
      "epoch": 0.6917433640122715,
      "grad_norm": 0.060654833912849426,
      "learning_rate": 0.00023550198701161718,
      "loss": 1.4647,
      "step": 15558
    },
    {
      "epoch": 0.6918322884709439,
      "grad_norm": 0.06024772301316261,
      "learning_rate": 0.00023537793913433385,
      "loss": 1.463,
      "step": 15560
    },
    {
      "epoch": 0.6919212129296163,
      "grad_norm": 0.060860976576805115,
      "learning_rate": 0.00023525391387811047,
      "loss": 1.4673,
      "step": 15562
    },
    {
      "epoch": 0.6920101373882886,
      "grad_norm": 0.058908648788928986,
      "learning_rate": 0.00023512991125354944,
      "loss": 1.4642,
      "step": 15564
    },
    {
      "epoch": 0.692099061846961,
      "grad_norm": 0.05974110960960388,
      "learning_rate": 0.0002350059312712506,
      "loss": 1.4639,
      "step": 15566
    },
    {
      "epoch": 0.6921879863056334,
      "grad_norm": 0.059412941336631775,
      "learning_rate": 0.0002348819739418126,
      "loss": 1.4672,
      "step": 15568
    },
    {
      "epoch": 0.6922769107643058,
      "grad_norm": 0.06173228472471237,
      "learning_rate": 0.00023475803927583194,
      "loss": 1.4662,
      "step": 15570
    },
    {
      "epoch": 0.6923658352229781,
      "grad_norm": 0.0602116584777832,
      "learning_rate": 0.00023463412728390282,
      "loss": 1.4638,
      "step": 15572
    },
    {
      "epoch": 0.6924547596816505,
      "grad_norm": 0.061301339417696,
      "learning_rate": 0.0002345102379766179,
      "loss": 1.4684,
      "step": 15574
    },
    {
      "epoch": 0.6925436841403227,
      "grad_norm": 0.06103100627660751,
      "learning_rate": 0.00023438637136456803,
      "loss": 1.4651,
      "step": 15576
    },
    {
      "epoch": 0.6926326085989951,
      "grad_norm": 0.06038039177656174,
      "learning_rate": 0.00023426252745834144,
      "loss": 1.4656,
      "step": 15578
    },
    {
      "epoch": 0.6927215330576675,
      "grad_norm": 0.05906345322728157,
      "learning_rate": 0.0002341387062685253,
      "loss": 1.4673,
      "step": 15580
    },
    {
      "epoch": 0.6928104575163399,
      "grad_norm": 0.06025807186961174,
      "learning_rate": 0.0002340149078057039,
      "loss": 1.461,
      "step": 15582
    },
    {
      "epoch": 0.6928993819750122,
      "grad_norm": 0.06042305752635002,
      "learning_rate": 0.00023389113208046038,
      "loss": 1.4679,
      "step": 15584
    },
    {
      "epoch": 0.6929883064336846,
      "grad_norm": 0.05965936928987503,
      "learning_rate": 0.00023376737910337582,
      "loss": 1.4676,
      "step": 15586
    },
    {
      "epoch": 0.693077230892357,
      "grad_norm": 0.06160576641559601,
      "learning_rate": 0.0002336436488850287,
      "loss": 1.4613,
      "step": 15588
    },
    {
      "epoch": 0.6931661553510293,
      "grad_norm": 0.06115745007991791,
      "learning_rate": 0.0002335199414359963,
      "loss": 1.4667,
      "step": 15590
    },
    {
      "epoch": 0.6932550798097017,
      "grad_norm": 0.06310083717107773,
      "learning_rate": 0.00023339625676685376,
      "loss": 1.4616,
      "step": 15592
    },
    {
      "epoch": 0.6933440042683741,
      "grad_norm": 0.06115524098277092,
      "learning_rate": 0.00023327259488817398,
      "loss": 1.4633,
      "step": 15594
    },
    {
      "epoch": 0.6934329287270464,
      "grad_norm": 0.058892738074064255,
      "learning_rate": 0.0002331489558105282,
      "loss": 1.4674,
      "step": 15596
    },
    {
      "epoch": 0.6935218531857187,
      "grad_norm": 0.058157991617918015,
      "learning_rate": 0.00023302533954448586,
      "loss": 1.4577,
      "step": 15598
    },
    {
      "epoch": 0.693610777644391,
      "grad_norm": 0.06278180330991745,
      "learning_rate": 0.00023290174610061376,
      "loss": 1.4603,
      "step": 15600
    },
    {
      "epoch": 0.6936997021030634,
      "grad_norm": 0.06078719347715378,
      "learning_rate": 0.0002327781754894775,
      "loss": 1.4642,
      "step": 15602
    },
    {
      "epoch": 0.6937886265617358,
      "grad_norm": 0.06112992390990257,
      "learning_rate": 0.0002326546277216406,
      "loss": 1.4734,
      "step": 15604
    },
    {
      "epoch": 0.6938775510204082,
      "grad_norm": 0.0634625107049942,
      "learning_rate": 0.0002325311028076641,
      "loss": 1.4633,
      "step": 15606
    },
    {
      "epoch": 0.6939664754790805,
      "grad_norm": 0.06064048036932945,
      "learning_rate": 0.0002324076007581078,
      "loss": 1.4626,
      "step": 15608
    },
    {
      "epoch": 0.6940553999377529,
      "grad_norm": 0.059506576508283615,
      "learning_rate": 0.00023228412158352862,
      "loss": 1.4643,
      "step": 15610
    },
    {
      "epoch": 0.6941443243964253,
      "grad_norm": 0.06171632558107376,
      "learning_rate": 0.00023216066529448294,
      "loss": 1.4637,
      "step": 15612
    },
    {
      "epoch": 0.6942332488550976,
      "grad_norm": 0.06034686788916588,
      "learning_rate": 0.00023203723190152393,
      "loss": 1.4683,
      "step": 15614
    },
    {
      "epoch": 0.69432217331377,
      "grad_norm": 0.061040595173835754,
      "learning_rate": 0.00023191382141520296,
      "loss": 1.4659,
      "step": 15616
    },
    {
      "epoch": 0.6944110977724424,
      "grad_norm": 0.061496566981077194,
      "learning_rate": 0.00023179043384607002,
      "loss": 1.4674,
      "step": 15618
    },
    {
      "epoch": 0.6945000222311146,
      "grad_norm": 0.058754466474056244,
      "learning_rate": 0.0002316670692046729,
      "loss": 1.4616,
      "step": 15620
    },
    {
      "epoch": 0.694588946689787,
      "grad_norm": 0.06135370582342148,
      "learning_rate": 0.000231543727501557,
      "loss": 1.4669,
      "step": 15622
    },
    {
      "epoch": 0.6946778711484594,
      "grad_norm": 0.058612678200006485,
      "learning_rate": 0.0002314204087472664,
      "loss": 1.4647,
      "step": 15624
    },
    {
      "epoch": 0.6947667956071317,
      "grad_norm": 0.06040282920002937,
      "learning_rate": 0.00023129711295234278,
      "loss": 1.4625,
      "step": 15626
    },
    {
      "epoch": 0.6948557200658041,
      "grad_norm": 0.06024818494915962,
      "learning_rate": 0.00023117384012732629,
      "loss": 1.4604,
      "step": 15628
    },
    {
      "epoch": 0.6949446445244765,
      "grad_norm": 0.060510408133268356,
      "learning_rate": 0.00023105059028275465,
      "loss": 1.4619,
      "step": 15630
    },
    {
      "epoch": 0.6950335689831488,
      "grad_norm": 0.06075219810009003,
      "learning_rate": 0.00023092736342916338,
      "loss": 1.4687,
      "step": 15632
    },
    {
      "epoch": 0.6951224934418212,
      "grad_norm": 0.0619949996471405,
      "learning_rate": 0.00023080415957708723,
      "loss": 1.4668,
      "step": 15634
    },
    {
      "epoch": 0.6952114179004936,
      "grad_norm": 0.0615934357047081,
      "learning_rate": 0.0002306809787370578,
      "loss": 1.468,
      "step": 15636
    },
    {
      "epoch": 0.6953003423591659,
      "grad_norm": 0.0594145767390728,
      "learning_rate": 0.00023055782091960497,
      "loss": 1.465,
      "step": 15638
    },
    {
      "epoch": 0.6953892668178382,
      "grad_norm": 0.05926194414496422,
      "learning_rate": 0.00023043468613525702,
      "loss": 1.4656,
      "step": 15640
    },
    {
      "epoch": 0.6954781912765106,
      "grad_norm": 0.06039900332689285,
      "learning_rate": 0.00023031157439454004,
      "loss": 1.471,
      "step": 15642
    },
    {
      "epoch": 0.6955671157351829,
      "grad_norm": 0.05915983021259308,
      "learning_rate": 0.00023018848570797835,
      "loss": 1.4628,
      "step": 15644
    },
    {
      "epoch": 0.6956560401938553,
      "grad_norm": 0.061098966747522354,
      "learning_rate": 0.0002300654200860937,
      "loss": 1.4688,
      "step": 15646
    },
    {
      "epoch": 0.6957449646525277,
      "grad_norm": 0.06001387909054756,
      "learning_rate": 0.00022994237753940655,
      "loss": 1.462,
      "step": 15648
    },
    {
      "epoch": 0.6958338891112,
      "grad_norm": 0.06392589956521988,
      "learning_rate": 0.00022981935807843523,
      "loss": 1.4688,
      "step": 15650
    },
    {
      "epoch": 0.6959228135698724,
      "grad_norm": 0.05922695994377136,
      "learning_rate": 0.00022969636171369585,
      "loss": 1.4647,
      "step": 15652
    },
    {
      "epoch": 0.6960117380285448,
      "grad_norm": 0.06134214252233505,
      "learning_rate": 0.00022957338845570232,
      "loss": 1.4697,
      "step": 15654
    },
    {
      "epoch": 0.6961006624872171,
      "grad_norm": 0.058561984449625015,
      "learning_rate": 0.0002294504383149676,
      "loss": 1.4655,
      "step": 15656
    },
    {
      "epoch": 0.6961895869458895,
      "grad_norm": 0.05878908187150955,
      "learning_rate": 0.0002293275113020017,
      "loss": 1.4722,
      "step": 15658
    },
    {
      "epoch": 0.6962785114045619,
      "grad_norm": 0.061014749109745026,
      "learning_rate": 0.00022920460742731268,
      "loss": 1.4645,
      "step": 15660
    },
    {
      "epoch": 0.6963674358632341,
      "grad_norm": 0.06033896282315254,
      "learning_rate": 0.0002290817267014073,
      "loss": 1.4719,
      "step": 15662
    },
    {
      "epoch": 0.6964563603219065,
      "grad_norm": 0.059221964329481125,
      "learning_rate": 0.00022895886913478974,
      "loss": 1.4656,
      "step": 15664
    },
    {
      "epoch": 0.6965452847805789,
      "grad_norm": 0.062421657145023346,
      "learning_rate": 0.00022883603473796273,
      "loss": 1.4704,
      "step": 15666
    },
    {
      "epoch": 0.6966342092392512,
      "grad_norm": 0.06081133335828781,
      "learning_rate": 0.0002287132235214262,
      "loss": 1.4621,
      "step": 15668
    },
    {
      "epoch": 0.6967231336979236,
      "grad_norm": 0.05905033275485039,
      "learning_rate": 0.00022859043549567888,
      "loss": 1.4667,
      "step": 15670
    },
    {
      "epoch": 0.696812058156596,
      "grad_norm": 0.05814428627490997,
      "learning_rate": 0.00022846767067121737,
      "loss": 1.4651,
      "step": 15672
    },
    {
      "epoch": 0.6969009826152683,
      "grad_norm": 0.06016310676932335,
      "learning_rate": 0.0002283449290585358,
      "loss": 1.4621,
      "step": 15674
    },
    {
      "epoch": 0.6969899070739407,
      "grad_norm": 0.05994448438286781,
      "learning_rate": 0.00022822221066812686,
      "loss": 1.4716,
      "step": 15676
    },
    {
      "epoch": 0.6970788315326131,
      "grad_norm": 0.062008995562791824,
      "learning_rate": 0.00022809951551048125,
      "loss": 1.4699,
      "step": 15678
    },
    {
      "epoch": 0.6971677559912854,
      "grad_norm": 0.057963281869888306,
      "learning_rate": 0.00022797684359608702,
      "loss": 1.4699,
      "step": 15680
    },
    {
      "epoch": 0.6972566804499578,
      "grad_norm": 0.06139107421040535,
      "learning_rate": 0.00022785419493543101,
      "loss": 1.4601,
      "step": 15682
    },
    {
      "epoch": 0.6973456049086301,
      "grad_norm": 0.05990011617541313,
      "learning_rate": 0.00022773156953899786,
      "loss": 1.4648,
      "step": 15684
    },
    {
      "epoch": 0.6974345293673024,
      "grad_norm": 0.059842146933078766,
      "learning_rate": 0.00022760896741726988,
      "loss": 1.4699,
      "step": 15686
    },
    {
      "epoch": 0.6975234538259748,
      "grad_norm": 0.06125682219862938,
      "learning_rate": 0.0002274863885807279,
      "loss": 1.4593,
      "step": 15688
    },
    {
      "epoch": 0.6976123782846472,
      "grad_norm": 0.06033066660165787,
      "learning_rate": 0.00022736383303985014,
      "loss": 1.4641,
      "step": 15690
    },
    {
      "epoch": 0.6977013027433195,
      "grad_norm": 0.06027790158987045,
      "learning_rate": 0.00022724130080511347,
      "loss": 1.4641,
      "step": 15692
    },
    {
      "epoch": 0.6977902272019919,
      "grad_norm": 0.05997991934418678,
      "learning_rate": 0.0002271187918869926,
      "loss": 1.4673,
      "step": 15694
    },
    {
      "epoch": 0.6978791516606643,
      "grad_norm": 0.06118232384324074,
      "learning_rate": 0.00022699630629595969,
      "loss": 1.4696,
      "step": 15696
    },
    {
      "epoch": 0.6979680761193366,
      "grad_norm": 0.06090354546904564,
      "learning_rate": 0.0002268738440424856,
      "loss": 1.4667,
      "step": 15698
    },
    {
      "epoch": 0.698057000578009,
      "grad_norm": 0.059883639216423035,
      "learning_rate": 0.00022675140513703923,
      "loss": 1.4629,
      "step": 15700
    },
    {
      "epoch": 0.6981459250366814,
      "grad_norm": 0.06094551458954811,
      "learning_rate": 0.00022662898959008665,
      "loss": 1.4616,
      "step": 15702
    },
    {
      "epoch": 0.6982348494953537,
      "grad_norm": 0.06020454317331314,
      "learning_rate": 0.0002265065974120928,
      "loss": 1.4652,
      "step": 15704
    },
    {
      "epoch": 0.698323773954026,
      "grad_norm": 0.06263810396194458,
      "learning_rate": 0.00022638422861352036,
      "loss": 1.461,
      "step": 15706
    },
    {
      "epoch": 0.6984126984126984,
      "grad_norm": 0.059346914291381836,
      "learning_rate": 0.00022626188320482975,
      "loss": 1.468,
      "step": 15708
    },
    {
      "epoch": 0.6985016228713707,
      "grad_norm": 0.059429168701171875,
      "learning_rate": 0.00022613956119647981,
      "loss": 1.4603,
      "step": 15710
    },
    {
      "epoch": 0.6985905473300431,
      "grad_norm": 0.0603039413690567,
      "learning_rate": 0.0002260172625989269,
      "loss": 1.4664,
      "step": 15712
    },
    {
      "epoch": 0.6986794717887155,
      "grad_norm": 0.060589831322431564,
      "learning_rate": 0.00022589498742262576,
      "loss": 1.4685,
      "step": 15714
    },
    {
      "epoch": 0.6987683962473878,
      "grad_norm": 0.05926838517189026,
      "learning_rate": 0.0002257727356780292,
      "loss": 1.4665,
      "step": 15716
    },
    {
      "epoch": 0.6988573207060602,
      "grad_norm": 0.0597996711730957,
      "learning_rate": 0.00022565050737558752,
      "loss": 1.4617,
      "step": 15718
    },
    {
      "epoch": 0.6989462451647326,
      "grad_norm": 0.06095963716506958,
      "learning_rate": 0.00022552830252574953,
      "loss": 1.4685,
      "step": 15720
    },
    {
      "epoch": 0.699035169623405,
      "grad_norm": 0.05980850011110306,
      "learning_rate": 0.00022540612113896192,
      "loss": 1.4661,
      "step": 15722
    },
    {
      "epoch": 0.6991240940820773,
      "grad_norm": 0.060585636645555496,
      "learning_rate": 0.00022528396322566896,
      "loss": 1.4626,
      "step": 15724
    },
    {
      "epoch": 0.6992130185407497,
      "grad_norm": 0.060612861067056656,
      "learning_rate": 0.00022516182879631352,
      "loss": 1.4685,
      "step": 15726
    },
    {
      "epoch": 0.6993019429994219,
      "grad_norm": 0.05883822590112686,
      "learning_rate": 0.0002250397178613363,
      "loss": 1.4628,
      "step": 15728
    },
    {
      "epoch": 0.6993908674580943,
      "grad_norm": 0.06021643057465553,
      "learning_rate": 0.0002249176304311755,
      "loss": 1.4612,
      "step": 15730
    },
    {
      "epoch": 0.6994797919167667,
      "grad_norm": 0.058791451156139374,
      "learning_rate": 0.00022479556651626792,
      "loss": 1.4669,
      "step": 15732
    },
    {
      "epoch": 0.699568716375439,
      "grad_norm": 0.06060413271188736,
      "learning_rate": 0.00022467352612704812,
      "loss": 1.464,
      "step": 15734
    },
    {
      "epoch": 0.6996576408341114,
      "grad_norm": 0.05910290777683258,
      "learning_rate": 0.0002245515092739488,
      "loss": 1.4708,
      "step": 15736
    },
    {
      "epoch": 0.6997465652927838,
      "grad_norm": 0.05985668674111366,
      "learning_rate": 0.0002244295159674003,
      "loss": 1.4653,
      "step": 15738
    },
    {
      "epoch": 0.6998354897514562,
      "grad_norm": 0.05992359668016434,
      "learning_rate": 0.00022430754621783085,
      "loss": 1.4624,
      "step": 15740
    },
    {
      "epoch": 0.6999244142101285,
      "grad_norm": 0.058332495391368866,
      "learning_rate": 0.00022418560003566762,
      "loss": 1.4663,
      "step": 15742
    },
    {
      "epoch": 0.7000133386688009,
      "grad_norm": 0.061433497816324234,
      "learning_rate": 0.00022406367743133475,
      "loss": 1.4643,
      "step": 15744
    },
    {
      "epoch": 0.7001022631274733,
      "grad_norm": 0.06030712649226189,
      "learning_rate": 0.0002239417784152546,
      "loss": 1.4649,
      "step": 15746
    },
    {
      "epoch": 0.7001911875861456,
      "grad_norm": 0.060619745403528214,
      "learning_rate": 0.00022381990299784777,
      "loss": 1.4661,
      "step": 15748
    },
    {
      "epoch": 0.7002801120448179,
      "grad_norm": 0.06033237650990486,
      "learning_rate": 0.00022369805118953274,
      "loss": 1.4677,
      "step": 15750
    },
    {
      "epoch": 0.7003690365034902,
      "grad_norm": 0.06091267988085747,
      "learning_rate": 0.0002235762230007261,
      "loss": 1.4652,
      "step": 15752
    },
    {
      "epoch": 0.7004579609621626,
      "grad_norm": 0.06021000072360039,
      "learning_rate": 0.00022345441844184188,
      "loss": 1.4668,
      "step": 15754
    },
    {
      "epoch": 0.700546885420835,
      "grad_norm": 0.05979836732149124,
      "learning_rate": 0.00022333263752329264,
      "loss": 1.4638,
      "step": 15756
    },
    {
      "epoch": 0.7006358098795074,
      "grad_norm": 0.05988521873950958,
      "learning_rate": 0.00022321088025548902,
      "loss": 1.4593,
      "step": 15758
    },
    {
      "epoch": 0.7007247343381797,
      "grad_norm": 0.061469241976737976,
      "learning_rate": 0.00022308914664883911,
      "loss": 1.4653,
      "step": 15760
    },
    {
      "epoch": 0.7008136587968521,
      "grad_norm": 0.05871487781405449,
      "learning_rate": 0.00022296743671374887,
      "loss": 1.4624,
      "step": 15762
    },
    {
      "epoch": 0.7009025832555245,
      "grad_norm": 0.05979468673467636,
      "learning_rate": 0.00022284575046062333,
      "loss": 1.4673,
      "step": 15764
    },
    {
      "epoch": 0.7009915077141968,
      "grad_norm": 0.06295553594827652,
      "learning_rate": 0.00022272408789986427,
      "loss": 1.4617,
      "step": 15766
    },
    {
      "epoch": 0.7010804321728692,
      "grad_norm": 0.0636453926563263,
      "learning_rate": 0.00022260244904187226,
      "loss": 1.4601,
      "step": 15768
    },
    {
      "epoch": 0.7011693566315415,
      "grad_norm": 0.06022004410624504,
      "learning_rate": 0.00022248083389704516,
      "loss": 1.4608,
      "step": 15770
    },
    {
      "epoch": 0.7012582810902138,
      "grad_norm": 0.059936974197626114,
      "learning_rate": 0.00022235924247577938,
      "loss": 1.4611,
      "step": 15772
    },
    {
      "epoch": 0.7013472055488862,
      "grad_norm": 0.059660959988832474,
      "learning_rate": 0.00022223767478846924,
      "loss": 1.4592,
      "step": 15774
    },
    {
      "epoch": 0.7014361300075586,
      "grad_norm": 0.06002466008067131,
      "learning_rate": 0.0002221161308455066,
      "loss": 1.4679,
      "step": 15776
    },
    {
      "epoch": 0.7015250544662309,
      "grad_norm": 0.06080063059926033,
      "learning_rate": 0.0002219946106572816,
      "loss": 1.4663,
      "step": 15778
    },
    {
      "epoch": 0.7016139789249033,
      "grad_norm": 0.059667252004146576,
      "learning_rate": 0.00022187311423418267,
      "loss": 1.4672,
      "step": 15780
    },
    {
      "epoch": 0.7017029033835757,
      "grad_norm": 0.05923038348555565,
      "learning_rate": 0.0002217516415865954,
      "loss": 1.4637,
      "step": 15782
    },
    {
      "epoch": 0.701791827842248,
      "grad_norm": 0.0604298934340477,
      "learning_rate": 0.00022163019272490397,
      "loss": 1.4652,
      "step": 15784
    },
    {
      "epoch": 0.7018807523009204,
      "grad_norm": 0.05981077253818512,
      "learning_rate": 0.00022150876765949062,
      "loss": 1.4664,
      "step": 15786
    },
    {
      "epoch": 0.7019696767595928,
      "grad_norm": 0.058474358171224594,
      "learning_rate": 0.00022138736640073487,
      "loss": 1.4659,
      "step": 15788
    },
    {
      "epoch": 0.7020586012182651,
      "grad_norm": 0.05936760455369949,
      "learning_rate": 0.00022126598895901484,
      "loss": 1.4668,
      "step": 15790
    },
    {
      "epoch": 0.7021475256769374,
      "grad_norm": 0.060833852738142014,
      "learning_rate": 0.00022114463534470664,
      "loss": 1.4602,
      "step": 15792
    },
    {
      "epoch": 0.7022364501356098,
      "grad_norm": 0.06116551160812378,
      "learning_rate": 0.00022102330556818367,
      "loss": 1.4691,
      "step": 15794
    },
    {
      "epoch": 0.7023253745942821,
      "grad_norm": 0.06091858446598053,
      "learning_rate": 0.00022090199963981815,
      "loss": 1.4575,
      "step": 15796
    },
    {
      "epoch": 0.7024142990529545,
      "grad_norm": 0.05815834179520607,
      "learning_rate": 0.00022078071756997947,
      "loss": 1.4642,
      "step": 15798
    },
    {
      "epoch": 0.7025032235116269,
      "grad_norm": 0.05963579937815666,
      "learning_rate": 0.00022065945936903554,
      "loss": 1.4623,
      "step": 15800
    },
    {
      "epoch": 0.7025921479702992,
      "grad_norm": 0.05970877781510353,
      "learning_rate": 0.00022053822504735222,
      "loss": 1.4654,
      "step": 15802
    },
    {
      "epoch": 0.7026810724289716,
      "grad_norm": 0.059889744967222214,
      "learning_rate": 0.00022041701461529274,
      "loss": 1.464,
      "step": 15804
    },
    {
      "epoch": 0.702769996887644,
      "grad_norm": 0.0590536892414093,
      "learning_rate": 0.00022029582808321896,
      "loss": 1.4651,
      "step": 15806
    },
    {
      "epoch": 0.7028589213463163,
      "grad_norm": 0.05903364717960358,
      "learning_rate": 0.0002201746654614906,
      "loss": 1.4673,
      "step": 15808
    },
    {
      "epoch": 0.7029478458049887,
      "grad_norm": 0.05853195860981941,
      "learning_rate": 0.0002200535267604648,
      "loss": 1.4652,
      "step": 15810
    },
    {
      "epoch": 0.7030367702636611,
      "grad_norm": 0.0594969168305397,
      "learning_rate": 0.00021993241199049724,
      "loss": 1.4621,
      "step": 15812
    },
    {
      "epoch": 0.7031256947223333,
      "grad_norm": 0.061632584780454636,
      "learning_rate": 0.00021981132116194146,
      "loss": 1.4674,
      "step": 15814
    },
    {
      "epoch": 0.7032146191810057,
      "grad_norm": 0.05890651419758797,
      "learning_rate": 0.00021969025428514855,
      "loss": 1.464,
      "step": 15816
    },
    {
      "epoch": 0.7033035436396781,
      "grad_norm": 0.05904078111052513,
      "learning_rate": 0.00021956921137046814,
      "loss": 1.4609,
      "step": 15818
    },
    {
      "epoch": 0.7033924680983504,
      "grad_norm": 0.05928374454379082,
      "learning_rate": 0.00021944819242824705,
      "loss": 1.4711,
      "step": 15820
    },
    {
      "epoch": 0.7034813925570228,
      "grad_norm": 0.05989319831132889,
      "learning_rate": 0.00021932719746883112,
      "loss": 1.4646,
      "step": 15822
    },
    {
      "epoch": 0.7035703170156952,
      "grad_norm": 0.05920826643705368,
      "learning_rate": 0.0002192062265025633,
      "loss": 1.4723,
      "step": 15824
    },
    {
      "epoch": 0.7036592414743675,
      "grad_norm": 0.0592825673520565,
      "learning_rate": 0.0002190852795397844,
      "loss": 1.464,
      "step": 15826
    },
    {
      "epoch": 0.7037481659330399,
      "grad_norm": 0.05995682626962662,
      "learning_rate": 0.0002189643565908338,
      "loss": 1.4645,
      "step": 15828
    },
    {
      "epoch": 0.7038370903917123,
      "grad_norm": 0.05902191996574402,
      "learning_rate": 0.00021884345766604867,
      "loss": 1.4597,
      "step": 15830
    },
    {
      "epoch": 0.7039260148503846,
      "grad_norm": 0.060309432446956635,
      "learning_rate": 0.00021872258277576363,
      "loss": 1.4737,
      "step": 15832
    },
    {
      "epoch": 0.704014939309057,
      "grad_norm": 0.06095599755644798,
      "learning_rate": 0.00021860173193031173,
      "loss": 1.4623,
      "step": 15834
    },
    {
      "epoch": 0.7041038637677293,
      "grad_norm": 0.060994766652584076,
      "learning_rate": 0.00021848090514002406,
      "loss": 1.461,
      "step": 15836
    },
    {
      "epoch": 0.7041927882264016,
      "grad_norm": 0.05787515640258789,
      "learning_rate": 0.000218360102415229,
      "loss": 1.4625,
      "step": 15838
    },
    {
      "epoch": 0.704281712685074,
      "grad_norm": 0.05939411744475365,
      "learning_rate": 0.00021823932376625362,
      "loss": 1.4617,
      "step": 15840
    },
    {
      "epoch": 0.7043706371437464,
      "grad_norm": 0.059615593403577805,
      "learning_rate": 0.0002181185692034225,
      "loss": 1.4611,
      "step": 15842
    },
    {
      "epoch": 0.7044595616024187,
      "grad_norm": 0.059437982738018036,
      "learning_rate": 0.00021799783873705842,
      "loss": 1.4641,
      "step": 15844
    },
    {
      "epoch": 0.7045484860610911,
      "grad_norm": 0.06425923109054565,
      "learning_rate": 0.00021787713237748192,
      "loss": 1.4658,
      "step": 15846
    },
    {
      "epoch": 0.7046374105197635,
      "grad_norm": 0.06056644767522812,
      "learning_rate": 0.000217756450135011,
      "loss": 1.4645,
      "step": 15848
    },
    {
      "epoch": 0.7047263349784358,
      "grad_norm": 0.06096633896231651,
      "learning_rate": 0.00021763579201996286,
      "loss": 1.4702,
      "step": 15850
    },
    {
      "epoch": 0.7048152594371082,
      "grad_norm": 0.06033441051840782,
      "learning_rate": 0.0002175151580426516,
      "loss": 1.466,
      "step": 15852
    },
    {
      "epoch": 0.7049041838957806,
      "grad_norm": 0.0599847212433815,
      "learning_rate": 0.00021739454821338933,
      "loss": 1.4668,
      "step": 15854
    },
    {
      "epoch": 0.704993108354453,
      "grad_norm": 0.06081864982843399,
      "learning_rate": 0.00021727396254248645,
      "loss": 1.4669,
      "step": 15856
    },
    {
      "epoch": 0.7050820328131252,
      "grad_norm": 0.06006255000829697,
      "learning_rate": 0.00021715340104025123,
      "loss": 1.46,
      "step": 15858
    },
    {
      "epoch": 0.7051709572717976,
      "grad_norm": 0.059641413390636444,
      "learning_rate": 0.0002170328637169899,
      "loss": 1.4586,
      "step": 15860
    },
    {
      "epoch": 0.7052598817304699,
      "grad_norm": 0.06192189082503319,
      "learning_rate": 0.0002169123505830062,
      "loss": 1.4649,
      "step": 15862
    },
    {
      "epoch": 0.7053488061891423,
      "grad_norm": 0.05896773561835289,
      "learning_rate": 0.00021679186164860227,
      "loss": 1.4648,
      "step": 15864
    },
    {
      "epoch": 0.7054377306478147,
      "grad_norm": 0.060805827379226685,
      "learning_rate": 0.00021667139692407828,
      "loss": 1.4672,
      "step": 15866
    },
    {
      "epoch": 0.705526655106487,
      "grad_norm": 0.06141723319888115,
      "learning_rate": 0.00021655095641973188,
      "loss": 1.458,
      "step": 15868
    },
    {
      "epoch": 0.7056155795651594,
      "grad_norm": 0.061006996780633926,
      "learning_rate": 0.00021643054014585845,
      "loss": 1.4693,
      "step": 15870
    },
    {
      "epoch": 0.7057045040238318,
      "grad_norm": 0.05978891998529434,
      "learning_rate": 0.00021631014811275247,
      "loss": 1.4703,
      "step": 15872
    },
    {
      "epoch": 0.7057934284825041,
      "grad_norm": 0.06066666170954704,
      "learning_rate": 0.00021618978033070501,
      "loss": 1.4592,
      "step": 15874
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.05984902381896973,
      "learning_rate": 0.00021606943681000607,
      "loss": 1.4589,
      "step": 15876
    },
    {
      "epoch": 0.7059712773998488,
      "grad_norm": 0.05905771628022194,
      "learning_rate": 0.0002159491175609427,
      "loss": 1.464,
      "step": 15878
    },
    {
      "epoch": 0.7060602018585211,
      "grad_norm": 0.06035059690475464,
      "learning_rate": 0.00021582882259380055,
      "loss": 1.4608,
      "step": 15880
    },
    {
      "epoch": 0.7061491263171935,
      "grad_norm": 0.061551641672849655,
      "learning_rate": 0.00021570855191886313,
      "loss": 1.4607,
      "step": 15882
    },
    {
      "epoch": 0.7062380507758659,
      "grad_norm": 0.05943434312939644,
      "learning_rate": 0.0002155883055464113,
      "loss": 1.4696,
      "step": 15884
    },
    {
      "epoch": 0.7063269752345382,
      "grad_norm": 0.05969441309571266,
      "learning_rate": 0.00021546808348672452,
      "loss": 1.4641,
      "step": 15886
    },
    {
      "epoch": 0.7064158996932106,
      "grad_norm": 0.05943189188838005,
      "learning_rate": 0.00021534788575008,
      "loss": 1.4646,
      "step": 15888
    },
    {
      "epoch": 0.706504824151883,
      "grad_norm": 0.05974946543574333,
      "learning_rate": 0.00021522771234675249,
      "loss": 1.4665,
      "step": 15890
    },
    {
      "epoch": 0.7065937486105553,
      "grad_norm": 0.05903458595275879,
      "learning_rate": 0.00021510756328701504,
      "loss": 1.4597,
      "step": 15892
    },
    {
      "epoch": 0.7066826730692277,
      "grad_norm": 0.061366867274045944,
      "learning_rate": 0.00021498743858113878,
      "loss": 1.4645,
      "step": 15894
    },
    {
      "epoch": 0.7067715975279001,
      "grad_norm": 0.05912630632519722,
      "learning_rate": 0.00021486733823939202,
      "loss": 1.4645,
      "step": 15896
    },
    {
      "epoch": 0.7068605219865725,
      "grad_norm": 0.06042192131280899,
      "learning_rate": 0.00021474726227204183,
      "loss": 1.4685,
      "step": 15898
    },
    {
      "epoch": 0.7069494464452447,
      "grad_norm": 0.05886279419064522,
      "learning_rate": 0.00021462721068935282,
      "loss": 1.4635,
      "step": 15900
    },
    {
      "epoch": 0.7070383709039171,
      "grad_norm": 0.058296773582696915,
      "learning_rate": 0.0002145071835015873,
      "loss": 1.4642,
      "step": 15902
    },
    {
      "epoch": 0.7071272953625894,
      "grad_norm": 0.05788257345557213,
      "learning_rate": 0.00021438718071900605,
      "loss": 1.4653,
      "step": 15904
    },
    {
      "epoch": 0.7072162198212618,
      "grad_norm": 0.05922654643654823,
      "learning_rate": 0.00021426720235186703,
      "loss": 1.4671,
      "step": 15906
    },
    {
      "epoch": 0.7073051442799342,
      "grad_norm": 0.059532783925533295,
      "learning_rate": 0.00021414724841042676,
      "loss": 1.4659,
      "step": 15908
    },
    {
      "epoch": 0.7073940687386066,
      "grad_norm": 0.05968281626701355,
      "learning_rate": 0.00021402731890493958,
      "loss": 1.4614,
      "step": 15910
    },
    {
      "epoch": 0.7074829931972789,
      "grad_norm": 0.06003040820360184,
      "learning_rate": 0.00021390741384565726,
      "loss": 1.466,
      "step": 15912
    },
    {
      "epoch": 0.7075719176559513,
      "grad_norm": 0.059364233165979385,
      "learning_rate": 0.00021378753324283002,
      "loss": 1.4609,
      "step": 15914
    },
    {
      "epoch": 0.7076608421146237,
      "grad_norm": 0.058379728347063065,
      "learning_rate": 0.0002136676771067058,
      "loss": 1.4637,
      "step": 15916
    },
    {
      "epoch": 0.707749766573296,
      "grad_norm": 0.059285663068294525,
      "learning_rate": 0.00021354784544753026,
      "loss": 1.4653,
      "step": 15918
    },
    {
      "epoch": 0.7078386910319684,
      "grad_norm": 0.061500970274209976,
      "learning_rate": 0.00021342803827554725,
      "loss": 1.464,
      "step": 15920
    },
    {
      "epoch": 0.7079276154906406,
      "grad_norm": 0.05940185859799385,
      "learning_rate": 0.00021330825560099854,
      "loss": 1.4641,
      "step": 15922
    },
    {
      "epoch": 0.708016539949313,
      "grad_norm": 0.05833735316991806,
      "learning_rate": 0.00021318849743412344,
      "loss": 1.465,
      "step": 15924
    },
    {
      "epoch": 0.7081054644079854,
      "grad_norm": 0.05976291745901108,
      "learning_rate": 0.00021306876378515967,
      "loss": 1.4637,
      "step": 15926
    },
    {
      "epoch": 0.7081943888666578,
      "grad_norm": 0.058585334569215775,
      "learning_rate": 0.000212949054664342,
      "loss": 1.4653,
      "step": 15928
    },
    {
      "epoch": 0.7082833133253301,
      "grad_norm": 0.061199333518743515,
      "learning_rate": 0.00021282937008190457,
      "loss": 1.465,
      "step": 15930
    },
    {
      "epoch": 0.7083722377840025,
      "grad_norm": 0.05798893794417381,
      "learning_rate": 0.00021270971004807809,
      "loss": 1.4624,
      "step": 15932
    },
    {
      "epoch": 0.7084611622426749,
      "grad_norm": 0.060919709503650665,
      "learning_rate": 0.00021259007457309149,
      "loss": 1.4684,
      "step": 15934
    },
    {
      "epoch": 0.7085500867013472,
      "grad_norm": 0.05985601246356964,
      "learning_rate": 0.00021247046366717183,
      "loss": 1.4633,
      "step": 15936
    },
    {
      "epoch": 0.7086390111600196,
      "grad_norm": 0.059532877057790756,
      "learning_rate": 0.00021235087734054425,
      "loss": 1.4666,
      "step": 15938
    },
    {
      "epoch": 0.708727935618692,
      "grad_norm": 0.060997124761343,
      "learning_rate": 0.00021223131560343112,
      "loss": 1.4664,
      "step": 15940
    },
    {
      "epoch": 0.7088168600773643,
      "grad_norm": 0.061860326677560806,
      "learning_rate": 0.00021211177846605327,
      "loss": 1.4628,
      "step": 15942
    },
    {
      "epoch": 0.7089057845360366,
      "grad_norm": 0.05831504985690117,
      "learning_rate": 0.0002119922659386293,
      "loss": 1.465,
      "step": 15944
    },
    {
      "epoch": 0.708994708994709,
      "grad_norm": 0.05970292538404465,
      "learning_rate": 0.00021187277803137584,
      "loss": 1.4616,
      "step": 15946
    },
    {
      "epoch": 0.7090836334533813,
      "grad_norm": 0.058645959943532944,
      "learning_rate": 0.00021175331475450703,
      "loss": 1.463,
      "step": 15948
    },
    {
      "epoch": 0.7091725579120537,
      "grad_norm": 0.05951127037405968,
      "learning_rate": 0.00021163387611823482,
      "loss": 1.4609,
      "step": 15950
    },
    {
      "epoch": 0.7092614823707261,
      "grad_norm": 0.05968938767910004,
      "learning_rate": 0.00021151446213276998,
      "loss": 1.4673,
      "step": 15952
    },
    {
      "epoch": 0.7093504068293984,
      "grad_norm": 0.0600113719701767,
      "learning_rate": 0.0002113950728083202,
      "loss": 1.4608,
      "step": 15954
    },
    {
      "epoch": 0.7094393312880708,
      "grad_norm": 0.06004900485277176,
      "learning_rate": 0.00021127570815509134,
      "loss": 1.462,
      "step": 15956
    },
    {
      "epoch": 0.7095282557467432,
      "grad_norm": 0.059999480843544006,
      "learning_rate": 0.00021115636818328727,
      "loss": 1.4652,
      "step": 15958
    },
    {
      "epoch": 0.7096171802054155,
      "grad_norm": 0.06003829836845398,
      "learning_rate": 0.0002110370529031097,
      "loss": 1.4673,
      "step": 15960
    },
    {
      "epoch": 0.7097061046640879,
      "grad_norm": 0.058980792760849,
      "learning_rate": 0.00021091776232475852,
      "loss": 1.4609,
      "step": 15962
    },
    {
      "epoch": 0.7097950291227603,
      "grad_norm": 0.05989384278655052,
      "learning_rate": 0.00021079849645843073,
      "loss": 1.4604,
      "step": 15964
    },
    {
      "epoch": 0.7098839535814325,
      "grad_norm": 0.0600026473402977,
      "learning_rate": 0.00021067925531432198,
      "loss": 1.4649,
      "step": 15966
    },
    {
      "epoch": 0.7099728780401049,
      "grad_norm": 0.05990760773420334,
      "learning_rate": 0.0002105600389026256,
      "loss": 1.4647,
      "step": 15968
    },
    {
      "epoch": 0.7100618024987773,
      "grad_norm": 0.0623035728931427,
      "learning_rate": 0.00021044084723353246,
      "loss": 1.4708,
      "step": 15970
    },
    {
      "epoch": 0.7101507269574496,
      "grad_norm": 0.059693653136491776,
      "learning_rate": 0.00021032168031723174,
      "loss": 1.4644,
      "step": 15972
    },
    {
      "epoch": 0.710239651416122,
      "grad_norm": 0.05945916846394539,
      "learning_rate": 0.00021020253816391056,
      "loss": 1.4599,
      "step": 15974
    },
    {
      "epoch": 0.7103285758747944,
      "grad_norm": 0.060173675417900085,
      "learning_rate": 0.00021008342078375353,
      "loss": 1.4667,
      "step": 15976
    },
    {
      "epoch": 0.7104175003334667,
      "grad_norm": 0.0596684031188488,
      "learning_rate": 0.00020996432818694288,
      "loss": 1.4599,
      "step": 15978
    },
    {
      "epoch": 0.7105064247921391,
      "grad_norm": 0.06060690060257912,
      "learning_rate": 0.00020984526038366003,
      "loss": 1.4643,
      "step": 15980
    },
    {
      "epoch": 0.7105953492508115,
      "grad_norm": 0.05914461240172386,
      "learning_rate": 0.00020972621738408277,
      "loss": 1.4646,
      "step": 15982
    },
    {
      "epoch": 0.7106842737094838,
      "grad_norm": 0.06105831637978554,
      "learning_rate": 0.00020960719919838787,
      "loss": 1.4666,
      "step": 15984
    },
    {
      "epoch": 0.7107731981681562,
      "grad_norm": 0.06017794460058212,
      "learning_rate": 0.0002094882058367491,
      "loss": 1.4646,
      "step": 15986
    },
    {
      "epoch": 0.7108621226268285,
      "grad_norm": 0.059006720781326294,
      "learning_rate": 0.0002093692373093387,
      "loss": 1.4621,
      "step": 15988
    },
    {
      "epoch": 0.7109510470855008,
      "grad_norm": 0.05977842956781387,
      "learning_rate": 0.0002092502936263269,
      "loss": 1.4631,
      "step": 15990
    },
    {
      "epoch": 0.7110399715441732,
      "grad_norm": 0.06044214591383934,
      "learning_rate": 0.00020913137479788109,
      "loss": 1.4614,
      "step": 15992
    },
    {
      "epoch": 0.7111288960028456,
      "grad_norm": 0.05967364087700844,
      "learning_rate": 0.0002090124808341672,
      "loss": 1.4622,
      "step": 15994
    },
    {
      "epoch": 0.7112178204615179,
      "grad_norm": 0.05973359942436218,
      "learning_rate": 0.00020889361174534894,
      "loss": 1.459,
      "step": 15996
    },
    {
      "epoch": 0.7113067449201903,
      "grad_norm": 0.05971769616007805,
      "learning_rate": 0.00020877476754158742,
      "loss": 1.4605,
      "step": 15998
    },
    {
      "epoch": 0.7113956693788627,
      "grad_norm": 0.060989778488874435,
      "learning_rate": 0.0002086559482330422,
      "loss": 1.46,
      "step": 16000
    },
    {
      "epoch": 0.7113956693788627,
      "eval_loss": 1.4480788707733154,
      "eval_runtime": 12.4236,
      "eval_samples_per_second": 556.2,
      "eval_steps_per_second": 69.545,
      "step": 16000
    },
    {
      "epoch": 0.711484593837535,
      "grad_norm": 0.05998692661523819,
      "learning_rate": 0.00020853715382987054,
      "loss": 1.4688,
      "step": 16002
    },
    {
      "epoch": 0.7115735182962074,
      "grad_norm": 0.06045999005436897,
      "learning_rate": 0.00020841838434222732,
      "loss": 1.468,
      "step": 16004
    },
    {
      "epoch": 0.7116624427548798,
      "grad_norm": 0.05892810970544815,
      "learning_rate": 0.00020829963978026566,
      "loss": 1.4649,
      "step": 16006
    },
    {
      "epoch": 0.711751367213552,
      "grad_norm": 0.0596071220934391,
      "learning_rate": 0.00020818092015413616,
      "loss": 1.4559,
      "step": 16008
    },
    {
      "epoch": 0.7118402916722244,
      "grad_norm": 0.061141762882471085,
      "learning_rate": 0.00020806222547398756,
      "loss": 1.4602,
      "step": 16010
    },
    {
      "epoch": 0.7119292161308968,
      "grad_norm": 0.059475455433130264,
      "learning_rate": 0.0002079435557499667,
      "loss": 1.4639,
      "step": 16012
    },
    {
      "epoch": 0.7120181405895691,
      "grad_norm": 0.06107206270098686,
      "learning_rate": 0.00020782491099221752,
      "loss": 1.4668,
      "step": 16014
    },
    {
      "epoch": 0.7121070650482415,
      "grad_norm": 0.060450684279203415,
      "learning_rate": 0.00020770629121088252,
      "loss": 1.4543,
      "step": 16016
    },
    {
      "epoch": 0.7121959895069139,
      "grad_norm": 0.059023890644311905,
      "learning_rate": 0.00020758769641610202,
      "loss": 1.4608,
      "step": 16018
    },
    {
      "epoch": 0.7122849139655862,
      "grad_norm": 0.06070045754313469,
      "learning_rate": 0.0002074691266180137,
      "loss": 1.4574,
      "step": 16020
    },
    {
      "epoch": 0.7123738384242586,
      "grad_norm": 0.0602857731282711,
      "learning_rate": 0.0002073505818267536,
      "loss": 1.4639,
      "step": 16022
    },
    {
      "epoch": 0.712462762882931,
      "grad_norm": 0.05997920036315918,
      "learning_rate": 0.00020723206205245555,
      "loss": 1.4654,
      "step": 16024
    },
    {
      "epoch": 0.7125516873416033,
      "grad_norm": 0.06003386527299881,
      "learning_rate": 0.00020711356730525082,
      "loss": 1.4625,
      "step": 16026
    },
    {
      "epoch": 0.7126406118002757,
      "grad_norm": 0.05993438884615898,
      "learning_rate": 0.00020699509759526913,
      "loss": 1.4609,
      "step": 16028
    },
    {
      "epoch": 0.712729536258948,
      "grad_norm": 0.059374984353780746,
      "learning_rate": 0.0002068766529326379,
      "loss": 1.4637,
      "step": 16030
    },
    {
      "epoch": 0.7128184607176203,
      "grad_norm": 0.05907906964421272,
      "learning_rate": 0.00020675823332748196,
      "loss": 1.4645,
      "step": 16032
    },
    {
      "epoch": 0.7129073851762927,
      "grad_norm": 0.05863776430487633,
      "learning_rate": 0.00020663983878992464,
      "loss": 1.4627,
      "step": 16034
    },
    {
      "epoch": 0.7129963096349651,
      "grad_norm": 0.058749303221702576,
      "learning_rate": 0.00020652146933008636,
      "loss": 1.4612,
      "step": 16036
    },
    {
      "epoch": 0.7130852340936374,
      "grad_norm": 0.0615452416241169,
      "learning_rate": 0.00020640312495808654,
      "loss": 1.4709,
      "step": 16038
    },
    {
      "epoch": 0.7131741585523098,
      "grad_norm": 0.05932776257395744,
      "learning_rate": 0.0002062848056840415,
      "loss": 1.4652,
      "step": 16040
    },
    {
      "epoch": 0.7132630830109822,
      "grad_norm": 0.06059432402253151,
      "learning_rate": 0.00020616651151806543,
      "loss": 1.4626,
      "step": 16042
    },
    {
      "epoch": 0.7133520074696545,
      "grad_norm": 0.06045154482126236,
      "learning_rate": 0.00020604824247027077,
      "loss": 1.464,
      "step": 16044
    },
    {
      "epoch": 0.7134409319283269,
      "grad_norm": 0.06008205935359001,
      "learning_rate": 0.000205929998550768,
      "loss": 1.467,
      "step": 16046
    },
    {
      "epoch": 0.7135298563869993,
      "grad_norm": 0.059912290424108505,
      "learning_rate": 0.00020581177976966464,
      "loss": 1.4664,
      "step": 16048
    },
    {
      "epoch": 0.7136187808456717,
      "grad_norm": 0.061931006610393524,
      "learning_rate": 0.00020569358613706684,
      "loss": 1.4685,
      "step": 16050
    },
    {
      "epoch": 0.7137077053043439,
      "grad_norm": 0.059048622846603394,
      "learning_rate": 0.00020557541766307825,
      "loss": 1.4646,
      "step": 16052
    },
    {
      "epoch": 0.7137966297630163,
      "grad_norm": 0.058656416833400726,
      "learning_rate": 0.00020545727435780065,
      "loss": 1.4562,
      "step": 16054
    },
    {
      "epoch": 0.7138855542216886,
      "grad_norm": 0.05939547345042229,
      "learning_rate": 0.0002053391562313332,
      "loss": 1.4609,
      "step": 16056
    },
    {
      "epoch": 0.713974478680361,
      "grad_norm": 0.060639627277851105,
      "learning_rate": 0.00020522106329377287,
      "loss": 1.4673,
      "step": 16058
    },
    {
      "epoch": 0.7140634031390334,
      "grad_norm": 0.06143512949347496,
      "learning_rate": 0.00020510299555521544,
      "loss": 1.4704,
      "step": 16060
    },
    {
      "epoch": 0.7141523275977057,
      "grad_norm": 0.059509847313165665,
      "learning_rate": 0.00020498495302575354,
      "loss": 1.4638,
      "step": 16062
    },
    {
      "epoch": 0.7142412520563781,
      "grad_norm": 0.05967727676033974,
      "learning_rate": 0.00020486693571547776,
      "loss": 1.4603,
      "step": 16064
    },
    {
      "epoch": 0.7143301765150505,
      "grad_norm": 0.06043224409222603,
      "learning_rate": 0.00020474894363447694,
      "loss": 1.4602,
      "step": 16066
    },
    {
      "epoch": 0.7144191009737229,
      "grad_norm": 0.05880916118621826,
      "learning_rate": 0.00020463097679283748,
      "loss": 1.4666,
      "step": 16068
    },
    {
      "epoch": 0.7145080254323952,
      "grad_norm": 0.06067218258976936,
      "learning_rate": 0.00020451303520064402,
      "loss": 1.4597,
      "step": 16070
    },
    {
      "epoch": 0.7145969498910676,
      "grad_norm": 0.06095737963914871,
      "learning_rate": 0.0002043951188679783,
      "loss": 1.4666,
      "step": 16072
    },
    {
      "epoch": 0.7146858743497398,
      "grad_norm": 0.06045563891530037,
      "learning_rate": 0.00020427722780492049,
      "loss": 1.4595,
      "step": 16074
    },
    {
      "epoch": 0.7147747988084122,
      "grad_norm": 0.05901022255420685,
      "learning_rate": 0.00020415936202154862,
      "loss": 1.4609,
      "step": 16076
    },
    {
      "epoch": 0.7148637232670846,
      "grad_norm": 0.06187785044312477,
      "learning_rate": 0.00020404152152793804,
      "loss": 1.4616,
      "step": 16078
    },
    {
      "epoch": 0.714952647725757,
      "grad_norm": 0.06016907840967178,
      "learning_rate": 0.00020392370633416247,
      "loss": 1.4627,
      "step": 16080
    },
    {
      "epoch": 0.7150415721844293,
      "grad_norm": 0.05958845466375351,
      "learning_rate": 0.00020380591645029338,
      "loss": 1.4614,
      "step": 16082
    },
    {
      "epoch": 0.7151304966431017,
      "grad_norm": 0.06001560017466545,
      "learning_rate": 0.00020368815188639966,
      "loss": 1.4599,
      "step": 16084
    },
    {
      "epoch": 0.715219421101774,
      "grad_norm": 0.05985542759299278,
      "learning_rate": 0.00020357041265254856,
      "loss": 1.4627,
      "step": 16086
    },
    {
      "epoch": 0.7153083455604464,
      "grad_norm": 0.05955475568771362,
      "learning_rate": 0.00020345269875880507,
      "loss": 1.4571,
      "step": 16088
    },
    {
      "epoch": 0.7153972700191188,
      "grad_norm": 0.059921056032180786,
      "learning_rate": 0.00020333501021523154,
      "loss": 1.464,
      "step": 16090
    },
    {
      "epoch": 0.7154861944777912,
      "grad_norm": 0.06080423295497894,
      "learning_rate": 0.00020321734703188893,
      "loss": 1.4646,
      "step": 16092
    },
    {
      "epoch": 0.7155751189364635,
      "grad_norm": 0.05900348350405693,
      "learning_rate": 0.00020309970921883515,
      "loss": 1.4668,
      "step": 16094
    },
    {
      "epoch": 0.7156640433951358,
      "grad_norm": 0.05854041501879692,
      "learning_rate": 0.0002029820967861266,
      "loss": 1.4641,
      "step": 16096
    },
    {
      "epoch": 0.7157529678538082,
      "grad_norm": 0.06043776124715805,
      "learning_rate": 0.0002028645097438176,
      "loss": 1.4555,
      "step": 16098
    },
    {
      "epoch": 0.7158418923124805,
      "grad_norm": 0.06004379689693451,
      "learning_rate": 0.00020274694810195948,
      "loss": 1.4663,
      "step": 16100
    },
    {
      "epoch": 0.7159308167711529,
      "grad_norm": 0.059593565762043,
      "learning_rate": 0.0002026294118706022,
      "loss": 1.4625,
      "step": 16102
    },
    {
      "epoch": 0.7160197412298253,
      "grad_norm": 0.05828922986984253,
      "learning_rate": 0.00020251190105979346,
      "loss": 1.4581,
      "step": 16104
    },
    {
      "epoch": 0.7161086656884976,
      "grad_norm": 0.060029514133930206,
      "learning_rate": 0.00020239441567957822,
      "loss": 1.4621,
      "step": 16106
    },
    {
      "epoch": 0.71619759014717,
      "grad_norm": 0.05973885953426361,
      "learning_rate": 0.00020227695573999982,
      "loss": 1.4635,
      "step": 16108
    },
    {
      "epoch": 0.7162865146058424,
      "grad_norm": 0.05934428423643112,
      "learning_rate": 0.00020215952125109948,
      "loss": 1.4608,
      "step": 16110
    },
    {
      "epoch": 0.7163754390645147,
      "grad_norm": 0.06081746146082878,
      "learning_rate": 0.00020204211222291557,
      "loss": 1.4604,
      "step": 16112
    },
    {
      "epoch": 0.7164643635231871,
      "grad_norm": 0.05820313096046448,
      "learning_rate": 0.00020192472866548524,
      "loss": 1.4615,
      "step": 16114
    },
    {
      "epoch": 0.7165532879818595,
      "grad_norm": 0.05866072699427605,
      "learning_rate": 0.0002018073705888424,
      "loss": 1.4571,
      "step": 16116
    },
    {
      "epoch": 0.7166422124405317,
      "grad_norm": 0.05988702178001404,
      "learning_rate": 0.00020169003800301965,
      "loss": 1.4649,
      "step": 16118
    },
    {
      "epoch": 0.7167311368992041,
      "grad_norm": 0.059877146035432816,
      "learning_rate": 0.00020157273091804724,
      "loss": 1.463,
      "step": 16120
    },
    {
      "epoch": 0.7168200613578765,
      "grad_norm": 0.06032383814454079,
      "learning_rate": 0.00020145544934395271,
      "loss": 1.4683,
      "step": 16122
    },
    {
      "epoch": 0.7169089858165488,
      "grad_norm": 0.05933660268783569,
      "learning_rate": 0.00020133819329076207,
      "loss": 1.4627,
      "step": 16124
    },
    {
      "epoch": 0.7169979102752212,
      "grad_norm": 0.05780529975891113,
      "learning_rate": 0.00020122096276849899,
      "loss": 1.4619,
      "step": 16126
    },
    {
      "epoch": 0.7170868347338936,
      "grad_norm": 0.06042304262518883,
      "learning_rate": 0.00020110375778718449,
      "loss": 1.4691,
      "step": 16128
    },
    {
      "epoch": 0.7171757591925659,
      "grad_norm": 0.05824956297874451,
      "learning_rate": 0.00020098657835683798,
      "loss": 1.4589,
      "step": 16130
    },
    {
      "epoch": 0.7172646836512383,
      "grad_norm": 0.060649555176496506,
      "learning_rate": 0.00020086942448747664,
      "loss": 1.4593,
      "step": 16132
    },
    {
      "epoch": 0.7173536081099107,
      "grad_norm": 0.06005752086639404,
      "learning_rate": 0.00020075229618911493,
      "loss": 1.4662,
      "step": 16134
    },
    {
      "epoch": 0.717442532568583,
      "grad_norm": 0.06136421114206314,
      "learning_rate": 0.0002006351934717657,
      "loss": 1.4603,
      "step": 16136
    },
    {
      "epoch": 0.7175314570272553,
      "grad_norm": 0.06038883700966835,
      "learning_rate": 0.00020051811634543937,
      "loss": 1.4623,
      "step": 16138
    },
    {
      "epoch": 0.7176203814859277,
      "grad_norm": 0.05902881920337677,
      "learning_rate": 0.00020040106482014443,
      "loss": 1.4677,
      "step": 16140
    },
    {
      "epoch": 0.7177093059446,
      "grad_norm": 0.05924457684159279,
      "learning_rate": 0.00020028403890588677,
      "loss": 1.4614,
      "step": 16142
    },
    {
      "epoch": 0.7177982304032724,
      "grad_norm": 0.05954696983098984,
      "learning_rate": 0.00020016703861266987,
      "loss": 1.4647,
      "step": 16144
    },
    {
      "epoch": 0.7178871548619448,
      "grad_norm": 0.05851093307137489,
      "learning_rate": 0.0002000500639504962,
      "loss": 1.4644,
      "step": 16146
    },
    {
      "epoch": 0.7179760793206171,
      "grad_norm": 0.06012677401304245,
      "learning_rate": 0.0001999331149293649,
      "loss": 1.4596,
      "step": 16148
    },
    {
      "epoch": 0.7180650037792895,
      "grad_norm": 0.059234343469142914,
      "learning_rate": 0.0001998161915592731,
      "loss": 1.4601,
      "step": 16150
    },
    {
      "epoch": 0.7181539282379619,
      "grad_norm": 0.06154603511095047,
      "learning_rate": 0.00019969929385021612,
      "loss": 1.4609,
      "step": 16152
    },
    {
      "epoch": 0.7182428526966342,
      "grad_norm": 0.06049703061580658,
      "learning_rate": 0.0001995824218121871,
      "loss": 1.4565,
      "step": 16154
    },
    {
      "epoch": 0.7183317771553066,
      "grad_norm": 0.059687089174985886,
      "learning_rate": 0.00019946557545517636,
      "loss": 1.4608,
      "step": 16156
    },
    {
      "epoch": 0.718420701613979,
      "grad_norm": 0.06014818698167801,
      "learning_rate": 0.00019934875478917264,
      "loss": 1.4606,
      "step": 16158
    },
    {
      "epoch": 0.7185096260726512,
      "grad_norm": 0.061506614089012146,
      "learning_rate": 0.00019923195982416225,
      "loss": 1.4618,
      "step": 16160
    },
    {
      "epoch": 0.7185985505313236,
      "grad_norm": 0.059169650077819824,
      "learning_rate": 0.00019911519057012962,
      "loss": 1.4604,
      "step": 16162
    },
    {
      "epoch": 0.718687474989996,
      "grad_norm": 0.05947177857160568,
      "learning_rate": 0.00019899844703705644,
      "loss": 1.4603,
      "step": 16164
    },
    {
      "epoch": 0.7187763994486683,
      "grad_norm": 0.05948949605226517,
      "learning_rate": 0.0001988817292349221,
      "loss": 1.463,
      "step": 16166
    },
    {
      "epoch": 0.7188653239073407,
      "grad_norm": 0.05881119892001152,
      "learning_rate": 0.00019876503717370487,
      "loss": 1.4641,
      "step": 16168
    },
    {
      "epoch": 0.7189542483660131,
      "grad_norm": 0.061151519417762756,
      "learning_rate": 0.00019864837086337977,
      "loss": 1.4674,
      "step": 16170
    },
    {
      "epoch": 0.7190431728246854,
      "grad_norm": 0.06064696982502937,
      "learning_rate": 0.0001985317303139198,
      "loss": 1.4673,
      "step": 16172
    },
    {
      "epoch": 0.7191320972833578,
      "grad_norm": 0.062243394553661346,
      "learning_rate": 0.000198415115535296,
      "loss": 1.4636,
      "step": 16174
    },
    {
      "epoch": 0.7192210217420302,
      "grad_norm": 0.05873696506023407,
      "learning_rate": 0.0001982985265374772,
      "loss": 1.464,
      "step": 16176
    },
    {
      "epoch": 0.7193099462007025,
      "grad_norm": 0.05942462012171745,
      "learning_rate": 0.0001981819633304301,
      "loss": 1.4635,
      "step": 16178
    },
    {
      "epoch": 0.7193988706593749,
      "grad_norm": 0.0588180348277092,
      "learning_rate": 0.00019806542592411863,
      "loss": 1.4619,
      "step": 16180
    },
    {
      "epoch": 0.7194877951180472,
      "grad_norm": 0.059856392443180084,
      "learning_rate": 0.00019794891432850514,
      "loss": 1.4649,
      "step": 16182
    },
    {
      "epoch": 0.7195767195767195,
      "grad_norm": 0.05942876264452934,
      "learning_rate": 0.00019783242855354977,
      "loss": 1.4572,
      "step": 16184
    },
    {
      "epoch": 0.7196656440353919,
      "grad_norm": 0.060655154287815094,
      "learning_rate": 0.00019771596860920982,
      "loss": 1.4618,
      "step": 16186
    },
    {
      "epoch": 0.7197545684940643,
      "grad_norm": 0.06054103747010231,
      "learning_rate": 0.00019759953450544098,
      "loss": 1.4609,
      "step": 16188
    },
    {
      "epoch": 0.7198434929527366,
      "grad_norm": 0.05870714411139488,
      "learning_rate": 0.00019748312625219672,
      "loss": 1.461,
      "step": 16190
    },
    {
      "epoch": 0.719932417411409,
      "grad_norm": 0.05910949036478996,
      "learning_rate": 0.0001973667438594278,
      "loss": 1.4573,
      "step": 16192
    },
    {
      "epoch": 0.7200213418700814,
      "grad_norm": 0.06156644597649574,
      "learning_rate": 0.00019725038733708333,
      "loss": 1.4639,
      "step": 16194
    },
    {
      "epoch": 0.7201102663287537,
      "grad_norm": 0.05833414942026138,
      "learning_rate": 0.00019713405669511007,
      "loss": 1.4545,
      "step": 16196
    },
    {
      "epoch": 0.7201991907874261,
      "grad_norm": 0.05948293209075928,
      "learning_rate": 0.00019701775194345218,
      "loss": 1.457,
      "step": 16198
    },
    {
      "epoch": 0.7202881152460985,
      "grad_norm": 0.05885034427046776,
      "learning_rate": 0.0001969014730920522,
      "loss": 1.4604,
      "step": 16200
    },
    {
      "epoch": 0.7203770397047708,
      "grad_norm": 0.060151953250169754,
      "learning_rate": 0.00019678522015084983,
      "loss": 1.4617,
      "step": 16202
    },
    {
      "epoch": 0.7204659641634431,
      "grad_norm": 0.06010283902287483,
      "learning_rate": 0.0001966689931297831,
      "loss": 1.4669,
      "step": 16204
    },
    {
      "epoch": 0.7205548886221155,
      "grad_norm": 0.060459204018116,
      "learning_rate": 0.00019655279203878768,
      "loss": 1.4593,
      "step": 16206
    },
    {
      "epoch": 0.7206438130807878,
      "grad_norm": 0.06011422351002693,
      "learning_rate": 0.00019643661688779673,
      "loss": 1.4639,
      "step": 16208
    },
    {
      "epoch": 0.7207327375394602,
      "grad_norm": 0.06095078960061073,
      "learning_rate": 0.0001963204676867415,
      "loss": 1.4649,
      "step": 16210
    },
    {
      "epoch": 0.7208216619981326,
      "grad_norm": 0.058982133865356445,
      "learning_rate": 0.0001962043444455512,
      "loss": 1.4607,
      "step": 16212
    },
    {
      "epoch": 0.720910586456805,
      "grad_norm": 0.05980036035180092,
      "learning_rate": 0.00019608824717415207,
      "loss": 1.4636,
      "step": 16214
    },
    {
      "epoch": 0.7209995109154773,
      "grad_norm": 0.06017722934484482,
      "learning_rate": 0.0001959721758824689,
      "loss": 1.4604,
      "step": 16216
    },
    {
      "epoch": 0.7210884353741497,
      "grad_norm": 0.05889592692255974,
      "learning_rate": 0.0001958561305804241,
      "loss": 1.462,
      "step": 16218
    },
    {
      "epoch": 0.721177359832822,
      "grad_norm": 0.060982596129179,
      "learning_rate": 0.00019574011127793746,
      "loss": 1.4658,
      "step": 16220
    },
    {
      "epoch": 0.7212662842914944,
      "grad_norm": 0.05975702032446861,
      "learning_rate": 0.00019562411798492701,
      "loss": 1.4623,
      "step": 16222
    },
    {
      "epoch": 0.7213552087501668,
      "grad_norm": 0.05935823172330856,
      "learning_rate": 0.00019550815071130823,
      "loss": 1.4642,
      "step": 16224
    },
    {
      "epoch": 0.721444133208839,
      "grad_norm": 0.05940350890159607,
      "learning_rate": 0.0001953922094669945,
      "loss": 1.4637,
      "step": 16226
    },
    {
      "epoch": 0.7215330576675114,
      "grad_norm": 0.058843355625867844,
      "learning_rate": 0.0001952762942618973,
      "loss": 1.4622,
      "step": 16228
    },
    {
      "epoch": 0.7216219821261838,
      "grad_norm": 0.05933767557144165,
      "learning_rate": 0.00019516040510592515,
      "loss": 1.4674,
      "step": 16230
    },
    {
      "epoch": 0.7217109065848561,
      "grad_norm": 0.059428151696920395,
      "learning_rate": 0.0001950445420089849,
      "loss": 1.4658,
      "step": 16232
    },
    {
      "epoch": 0.7217998310435285,
      "grad_norm": 0.05923604965209961,
      "learning_rate": 0.0001949287049809813,
      "loss": 1.4643,
      "step": 16234
    },
    {
      "epoch": 0.7218887555022009,
      "grad_norm": 0.05947105586528778,
      "learning_rate": 0.00019481289403181623,
      "loss": 1.4598,
      "step": 16236
    },
    {
      "epoch": 0.7219776799608733,
      "grad_norm": 0.05970315635204315,
      "learning_rate": 0.00019469710917138994,
      "loss": 1.4593,
      "step": 16238
    },
    {
      "epoch": 0.7220666044195456,
      "grad_norm": 0.05884910002350807,
      "learning_rate": 0.00019458135040960036,
      "loss": 1.4637,
      "step": 16240
    },
    {
      "epoch": 0.722155528878218,
      "grad_norm": 0.05918232724070549,
      "learning_rate": 0.00019446561775634264,
      "loss": 1.458,
      "step": 16242
    },
    {
      "epoch": 0.7222444533368904,
      "grad_norm": 0.05901836231350899,
      "learning_rate": 0.00019434991122151057,
      "loss": 1.4594,
      "step": 16244
    },
    {
      "epoch": 0.7223333777955626,
      "grad_norm": 0.05842548981308937,
      "learning_rate": 0.0001942342308149947,
      "loss": 1.466,
      "step": 16246
    },
    {
      "epoch": 0.722422302254235,
      "grad_norm": 0.05925069749355316,
      "learning_rate": 0.00019411857654668457,
      "loss": 1.4643,
      "step": 16248
    },
    {
      "epoch": 0.7225112267129074,
      "grad_norm": 0.05993708595633507,
      "learning_rate": 0.0001940029484264665,
      "loss": 1.4551,
      "step": 16250
    },
    {
      "epoch": 0.7226001511715797,
      "grad_norm": 0.060077860951423645,
      "learning_rate": 0.00019388734646422472,
      "loss": 1.4578,
      "step": 16252
    },
    {
      "epoch": 0.7226890756302521,
      "grad_norm": 0.05832253396511078,
      "learning_rate": 0.00019377177066984158,
      "loss": 1.4647,
      "step": 16254
    },
    {
      "epoch": 0.7227780000889245,
      "grad_norm": 0.06057137995958328,
      "learning_rate": 0.00019365622105319713,
      "loss": 1.4583,
      "step": 16256
    },
    {
      "epoch": 0.7228669245475968,
      "grad_norm": 0.060661908239126205,
      "learning_rate": 0.00019354069762416877,
      "loss": 1.4647,
      "step": 16258
    },
    {
      "epoch": 0.7229558490062692,
      "grad_norm": 0.058103371411561966,
      "learning_rate": 0.00019342520039263207,
      "loss": 1.4619,
      "step": 16260
    },
    {
      "epoch": 0.7230447734649416,
      "grad_norm": 0.058487582951784134,
      "learning_rate": 0.0001933097293684603,
      "loss": 1.4574,
      "step": 16262
    },
    {
      "epoch": 0.7231336979236139,
      "grad_norm": 0.062292974442243576,
      "learning_rate": 0.00019319428456152454,
      "loss": 1.4686,
      "step": 16264
    },
    {
      "epoch": 0.7232226223822863,
      "grad_norm": 0.05761106684803963,
      "learning_rate": 0.0001930788659816932,
      "loss": 1.4576,
      "step": 16266
    },
    {
      "epoch": 0.7233115468409586,
      "grad_norm": 0.058435920625925064,
      "learning_rate": 0.00019296347363883303,
      "loss": 1.4618,
      "step": 16268
    },
    {
      "epoch": 0.7234004712996309,
      "grad_norm": 0.06008351221680641,
      "learning_rate": 0.00019284810754280834,
      "loss": 1.4556,
      "step": 16270
    },
    {
      "epoch": 0.7234893957583033,
      "grad_norm": 0.05883776769042015,
      "learning_rate": 0.00019273276770348098,
      "loss": 1.4617,
      "step": 16272
    },
    {
      "epoch": 0.7235783202169757,
      "grad_norm": 0.058516763150691986,
      "learning_rate": 0.00019261745413071035,
      "loss": 1.4609,
      "step": 16274
    },
    {
      "epoch": 0.723667244675648,
      "grad_norm": 0.0610274001955986,
      "learning_rate": 0.00019250216683435473,
      "loss": 1.4621,
      "step": 16276
    },
    {
      "epoch": 0.7237561691343204,
      "grad_norm": 0.06014120578765869,
      "learning_rate": 0.00019238690582426893,
      "loss": 1.466,
      "step": 16278
    },
    {
      "epoch": 0.7238450935929928,
      "grad_norm": 0.06050468608736992,
      "learning_rate": 0.0001922716711103059,
      "loss": 1.466,
      "step": 16280
    },
    {
      "epoch": 0.7239340180516651,
      "grad_norm": 0.05953977629542351,
      "learning_rate": 0.0001921564627023165,
      "loss": 1.4586,
      "step": 16282
    },
    {
      "epoch": 0.7240229425103375,
      "grad_norm": 0.059618160128593445,
      "learning_rate": 0.00019204128061014935,
      "loss": 1.4582,
      "step": 16284
    },
    {
      "epoch": 0.7241118669690099,
      "grad_norm": 0.06013248860836029,
      "learning_rate": 0.0001919261248436508,
      "loss": 1.4617,
      "step": 16286
    },
    {
      "epoch": 0.7242007914276822,
      "grad_norm": 0.05888787657022476,
      "learning_rate": 0.00019181099541266456,
      "loss": 1.4647,
      "step": 16288
    },
    {
      "epoch": 0.7242897158863545,
      "grad_norm": 0.05903095752000809,
      "learning_rate": 0.00019169589232703255,
      "loss": 1.4615,
      "step": 16290
    },
    {
      "epoch": 0.7243786403450269,
      "grad_norm": 0.059270888566970825,
      "learning_rate": 0.00019158081559659453,
      "loss": 1.4635,
      "step": 16292
    },
    {
      "epoch": 0.7244675648036992,
      "grad_norm": 0.06252747029066086,
      "learning_rate": 0.00019146576523118754,
      "loss": 1.4604,
      "step": 16294
    },
    {
      "epoch": 0.7245564892623716,
      "grad_norm": 0.060627248138189316,
      "learning_rate": 0.00019135074124064627,
      "loss": 1.4606,
      "step": 16296
    },
    {
      "epoch": 0.724645413721044,
      "grad_norm": 0.06071724370121956,
      "learning_rate": 0.00019123574363480417,
      "loss": 1.4627,
      "step": 16298
    },
    {
      "epoch": 0.7247343381797163,
      "grad_norm": 0.06156003847718239,
      "learning_rate": 0.0001911207724234912,
      "loss": 1.4647,
      "step": 16300
    },
    {
      "epoch": 0.7248232626383887,
      "grad_norm": 0.0580633170902729,
      "learning_rate": 0.0001910058276165359,
      "loss": 1.4643,
      "step": 16302
    },
    {
      "epoch": 0.7249121870970611,
      "grad_norm": 0.060547344386577606,
      "learning_rate": 0.00019089090922376405,
      "loss": 1.4638,
      "step": 16304
    },
    {
      "epoch": 0.7250011115557334,
      "grad_norm": 0.060422513633966446,
      "learning_rate": 0.00019077601725499944,
      "loss": 1.456,
      "step": 16306
    },
    {
      "epoch": 0.7250900360144058,
      "grad_norm": 0.059972185641527176,
      "learning_rate": 0.00019066115172006377,
      "loss": 1.4615,
      "step": 16308
    },
    {
      "epoch": 0.7251789604730782,
      "grad_norm": 0.05940213426947594,
      "learning_rate": 0.0001905463126287759,
      "loss": 1.4647,
      "step": 16310
    },
    {
      "epoch": 0.7252678849317504,
      "grad_norm": 0.06141269952058792,
      "learning_rate": 0.00019043149999095293,
      "loss": 1.4681,
      "step": 16312
    },
    {
      "epoch": 0.7253568093904228,
      "grad_norm": 0.06061433255672455,
      "learning_rate": 0.00019031671381640976,
      "loss": 1.4581,
      "step": 16314
    },
    {
      "epoch": 0.7254457338490952,
      "grad_norm": 0.05905766785144806,
      "learning_rate": 0.00019020195411495838,
      "loss": 1.459,
      "step": 16316
    },
    {
      "epoch": 0.7255346583077675,
      "grad_norm": 0.05946394428610802,
      "learning_rate": 0.0001900872208964093,
      "loss": 1.4631,
      "step": 16318
    },
    {
      "epoch": 0.7256235827664399,
      "grad_norm": 0.060677699744701385,
      "learning_rate": 0.00018997251417057044,
      "loss": 1.4648,
      "step": 16320
    },
    {
      "epoch": 0.7257125072251123,
      "grad_norm": 0.05936264991760254,
      "learning_rate": 0.00018985783394724715,
      "loss": 1.4605,
      "step": 16322
    },
    {
      "epoch": 0.7258014316837846,
      "grad_norm": 0.060269761830568314,
      "learning_rate": 0.00018974318023624293,
      "loss": 1.4642,
      "step": 16324
    },
    {
      "epoch": 0.725890356142457,
      "grad_norm": 0.05980468913912773,
      "learning_rate": 0.00018962855304735916,
      "loss": 1.4613,
      "step": 16326
    },
    {
      "epoch": 0.7259792806011294,
      "grad_norm": 0.05888427793979645,
      "learning_rate": 0.0001895139523903942,
      "loss": 1.4571,
      "step": 16328
    },
    {
      "epoch": 0.7260682050598017,
      "grad_norm": 0.05930918827652931,
      "learning_rate": 0.00018939937827514507,
      "loss": 1.4606,
      "step": 16330
    },
    {
      "epoch": 0.7261571295184741,
      "grad_norm": 0.06069708615541458,
      "learning_rate": 0.00018928483071140552,
      "loss": 1.4585,
      "step": 16332
    },
    {
      "epoch": 0.7262460539771464,
      "grad_norm": 0.05740377679467201,
      "learning_rate": 0.0001891703097089682,
      "loss": 1.4629,
      "step": 16334
    },
    {
      "epoch": 0.7263349784358187,
      "grad_norm": 0.05887899175286293,
      "learning_rate": 0.00018905581527762266,
      "loss": 1.4606,
      "step": 16336
    },
    {
      "epoch": 0.7264239028944911,
      "grad_norm": 0.06021280959248543,
      "learning_rate": 0.00018894134742715614,
      "loss": 1.4624,
      "step": 16338
    },
    {
      "epoch": 0.7265128273531635,
      "grad_norm": 0.06010861694812775,
      "learning_rate": 0.00018882690616735397,
      "loss": 1.4649,
      "step": 16340
    },
    {
      "epoch": 0.7266017518118358,
      "grad_norm": 0.05954209715127945,
      "learning_rate": 0.00018871249150799934,
      "loss": 1.4681,
      "step": 16342
    },
    {
      "epoch": 0.7266906762705082,
      "grad_norm": 0.058117982000112534,
      "learning_rate": 0.00018859810345887263,
      "loss": 1.4617,
      "step": 16344
    },
    {
      "epoch": 0.7267796007291806,
      "grad_norm": 0.05962996557354927,
      "learning_rate": 0.00018848374202975232,
      "loss": 1.4578,
      "step": 16346
    },
    {
      "epoch": 0.7268685251878529,
      "grad_norm": 0.06057076156139374,
      "learning_rate": 0.00018836940723041467,
      "loss": 1.4604,
      "step": 16348
    },
    {
      "epoch": 0.7269574496465253,
      "grad_norm": 0.06041696295142174,
      "learning_rate": 0.00018825509907063325,
      "loss": 1.4573,
      "step": 16350
    },
    {
      "epoch": 0.7270463741051977,
      "grad_norm": 0.059521712362766266,
      "learning_rate": 0.00018814081756017998,
      "loss": 1.4587,
      "step": 16352
    },
    {
      "epoch": 0.72713529856387,
      "grad_norm": 0.05939910560846329,
      "learning_rate": 0.00018802656270882357,
      "loss": 1.4637,
      "step": 16354
    },
    {
      "epoch": 0.7272242230225423,
      "grad_norm": 0.05934542044997215,
      "learning_rate": 0.00018791233452633178,
      "loss": 1.4619,
      "step": 16356
    },
    {
      "epoch": 0.7273131474812147,
      "grad_norm": 0.05895724520087242,
      "learning_rate": 0.00018779813302246894,
      "loss": 1.461,
      "step": 16358
    },
    {
      "epoch": 0.727402071939887,
      "grad_norm": 0.05960603058338165,
      "learning_rate": 0.00018768395820699735,
      "loss": 1.4599,
      "step": 16360
    },
    {
      "epoch": 0.7274909963985594,
      "grad_norm": 0.05965695157647133,
      "learning_rate": 0.00018756981008967734,
      "loss": 1.4609,
      "step": 16362
    },
    {
      "epoch": 0.7275799208572318,
      "grad_norm": 0.06047413870692253,
      "learning_rate": 0.00018745568868026698,
      "loss": 1.4629,
      "step": 16364
    },
    {
      "epoch": 0.7276688453159041,
      "grad_norm": 0.05957818776369095,
      "learning_rate": 0.00018734159398852148,
      "loss": 1.4595,
      "step": 16366
    },
    {
      "epoch": 0.7277577697745765,
      "grad_norm": 0.06116658076643944,
      "learning_rate": 0.0001872275260241944,
      "loss": 1.4633,
      "step": 16368
    },
    {
      "epoch": 0.7278466942332489,
      "grad_norm": 0.058435603976249695,
      "learning_rate": 0.00018711348479703673,
      "loss": 1.4567,
      "step": 16370
    },
    {
      "epoch": 0.7279356186919212,
      "grad_norm": 0.059727638959884644,
      "learning_rate": 0.00018699947031679736,
      "loss": 1.465,
      "step": 16372
    },
    {
      "epoch": 0.7280245431505936,
      "grad_norm": 0.06051727011799812,
      "learning_rate": 0.00018688548259322252,
      "loss": 1.4598,
      "step": 16374
    },
    {
      "epoch": 0.7281134676092659,
      "grad_norm": 0.05949892848730087,
      "learning_rate": 0.00018677152163605642,
      "loss": 1.4609,
      "step": 16376
    },
    {
      "epoch": 0.7282023920679382,
      "grad_norm": 0.060253631323575974,
      "learning_rate": 0.0001866575874550412,
      "loss": 1.4575,
      "step": 16378
    },
    {
      "epoch": 0.7282913165266106,
      "grad_norm": 0.05913444608449936,
      "learning_rate": 0.0001865436800599163,
      "loss": 1.4526,
      "step": 16380
    },
    {
      "epoch": 0.728380240985283,
      "grad_norm": 0.06042875722050667,
      "learning_rate": 0.00018642979946041855,
      "loss": 1.4702,
      "step": 16382
    },
    {
      "epoch": 0.7284691654439553,
      "grad_norm": 0.0590120404958725,
      "learning_rate": 0.0001863159456662838,
      "loss": 1.4635,
      "step": 16384
    },
    {
      "epoch": 0.7285580899026277,
      "grad_norm": 0.058893073350191116,
      "learning_rate": 0.00018620211868724417,
      "loss": 1.4607,
      "step": 16386
    },
    {
      "epoch": 0.7286470143613001,
      "grad_norm": 0.05858619511127472,
      "learning_rate": 0.00018608831853303054,
      "loss": 1.4628,
      "step": 16388
    },
    {
      "epoch": 0.7287359388199725,
      "grad_norm": 0.05882981792092323,
      "learning_rate": 0.0001859745452133706,
      "loss": 1.4651,
      "step": 16390
    },
    {
      "epoch": 0.7288248632786448,
      "grad_norm": 0.05942827835679054,
      "learning_rate": 0.0001858607987379904,
      "loss": 1.4644,
      "step": 16392
    },
    {
      "epoch": 0.7289137877373172,
      "grad_norm": 0.059885479509830475,
      "learning_rate": 0.00018574707911661365,
      "loss": 1.4578,
      "step": 16394
    },
    {
      "epoch": 0.7290027121959896,
      "grad_norm": 0.06129121780395508,
      "learning_rate": 0.00018563338635896125,
      "loss": 1.4568,
      "step": 16396
    },
    {
      "epoch": 0.7290916366546618,
      "grad_norm": 0.060367047786712646,
      "learning_rate": 0.0001855197204747524,
      "loss": 1.4601,
      "step": 16398
    },
    {
      "epoch": 0.7291805611133342,
      "grad_norm": 0.05986754596233368,
      "learning_rate": 0.00018540608147370386,
      "loss": 1.4617,
      "step": 16400
    },
    {
      "epoch": 0.7292694855720065,
      "grad_norm": 0.0603383369743824,
      "learning_rate": 0.0001852924693655297,
      "loss": 1.4597,
      "step": 16402
    },
    {
      "epoch": 0.7293584100306789,
      "grad_norm": 0.059178128838539124,
      "learning_rate": 0.00018517888415994205,
      "loss": 1.4629,
      "step": 16404
    },
    {
      "epoch": 0.7294473344893513,
      "grad_norm": 0.05977354943752289,
      "learning_rate": 0.00018506532586665098,
      "loss": 1.4619,
      "step": 16406
    },
    {
      "epoch": 0.7295362589480237,
      "grad_norm": 0.06036766618490219,
      "learning_rate": 0.00018495179449536358,
      "loss": 1.457,
      "step": 16408
    },
    {
      "epoch": 0.729625183406696,
      "grad_norm": 0.05909400433301926,
      "learning_rate": 0.0001848382900557853,
      "loss": 1.4589,
      "step": 16410
    },
    {
      "epoch": 0.7297141078653684,
      "grad_norm": 0.059279441833496094,
      "learning_rate": 0.0001847248125576187,
      "loss": 1.4622,
      "step": 16412
    },
    {
      "epoch": 0.7298030323240408,
      "grad_norm": 0.06124156713485718,
      "learning_rate": 0.0001846113620105645,
      "loss": 1.4588,
      "step": 16414
    },
    {
      "epoch": 0.7298919567827131,
      "grad_norm": 0.05898246914148331,
      "learning_rate": 0.0001844979384243211,
      "loss": 1.4576,
      "step": 16416
    },
    {
      "epoch": 0.7299808812413855,
      "grad_norm": 0.05937924608588219,
      "learning_rate": 0.00018438454180858417,
      "loss": 1.4656,
      "step": 16418
    },
    {
      "epoch": 0.7300698057000578,
      "grad_norm": 0.05933384969830513,
      "learning_rate": 0.0001842711721730474,
      "loss": 1.4565,
      "step": 16420
    },
    {
      "epoch": 0.7301587301587301,
      "grad_norm": 0.05890939012169838,
      "learning_rate": 0.00018415782952740245,
      "loss": 1.4622,
      "step": 16422
    },
    {
      "epoch": 0.7302476546174025,
      "grad_norm": 0.059906478971242905,
      "learning_rate": 0.00018404451388133786,
      "loss": 1.4655,
      "step": 16424
    },
    {
      "epoch": 0.7303365790760749,
      "grad_norm": 0.05846390128135681,
      "learning_rate": 0.00018393122524454065,
      "loss": 1.4616,
      "step": 16426
    },
    {
      "epoch": 0.7304255035347472,
      "grad_norm": 0.0599554218351841,
      "learning_rate": 0.00018381796362669533,
      "loss": 1.4652,
      "step": 16428
    },
    {
      "epoch": 0.7305144279934196,
      "grad_norm": 0.058733414858579636,
      "learning_rate": 0.0001837047290374837,
      "loss": 1.4564,
      "step": 16430
    },
    {
      "epoch": 0.730603352452092,
      "grad_norm": 0.06102525070309639,
      "learning_rate": 0.00018359152148658564,
      "loss": 1.4606,
      "step": 16432
    },
    {
      "epoch": 0.7306922769107643,
      "grad_norm": 0.05933784320950508,
      "learning_rate": 0.00018347834098367895,
      "loss": 1.4593,
      "step": 16434
    },
    {
      "epoch": 0.7307812013694367,
      "grad_norm": 0.06041508913040161,
      "learning_rate": 0.0001833651875384383,
      "loss": 1.4602,
      "step": 16436
    },
    {
      "epoch": 0.7308701258281091,
      "grad_norm": 0.05959257856011391,
      "learning_rate": 0.000183252061160537,
      "loss": 1.4654,
      "step": 16438
    },
    {
      "epoch": 0.7309590502867814,
      "grad_norm": 0.05934099480509758,
      "learning_rate": 0.00018313896185964497,
      "loss": 1.456,
      "step": 16440
    },
    {
      "epoch": 0.7310479747454537,
      "grad_norm": 0.06000977009534836,
      "learning_rate": 0.0001830258896454312,
      "loss": 1.4548,
      "step": 16442
    },
    {
      "epoch": 0.7311368992041261,
      "grad_norm": 0.05849349498748779,
      "learning_rate": 0.0001829128445275613,
      "loss": 1.4563,
      "step": 16444
    },
    {
      "epoch": 0.7312258236627984,
      "grad_norm": 0.060585957020521164,
      "learning_rate": 0.0001827998265156986,
      "loss": 1.4601,
      "step": 16446
    },
    {
      "epoch": 0.7313147481214708,
      "grad_norm": 0.059966158121824265,
      "learning_rate": 0.00018268683561950455,
      "loss": 1.462,
      "step": 16448
    },
    {
      "epoch": 0.7314036725801432,
      "grad_norm": 0.05956388637423515,
      "learning_rate": 0.0001825738718486384,
      "loss": 1.4581,
      "step": 16450
    },
    {
      "epoch": 0.7314925970388155,
      "grad_norm": 0.05950170010328293,
      "learning_rate": 0.00018246093521275632,
      "loss": 1.459,
      "step": 16452
    },
    {
      "epoch": 0.7315815214974879,
      "grad_norm": 0.05926720052957535,
      "learning_rate": 0.00018234802572151282,
      "loss": 1.4568,
      "step": 16454
    },
    {
      "epoch": 0.7316704459561603,
      "grad_norm": 0.05866430699825287,
      "learning_rate": 0.00018223514338455994,
      "loss": 1.4574,
      "step": 16456
    },
    {
      "epoch": 0.7317593704148326,
      "grad_norm": 0.05895557999610901,
      "learning_rate": 0.00018212228821154754,
      "loss": 1.4648,
      "step": 16458
    },
    {
      "epoch": 0.731848294873505,
      "grad_norm": 0.06050115451216698,
      "learning_rate": 0.0001820094602121228,
      "loss": 1.4635,
      "step": 16460
    },
    {
      "epoch": 0.7319372193321774,
      "grad_norm": 0.06033334881067276,
      "learning_rate": 0.0001818966593959303,
      "loss": 1.4609,
      "step": 16462
    },
    {
      "epoch": 0.7320261437908496,
      "grad_norm": 0.0586191788315773,
      "learning_rate": 0.00018178388577261363,
      "loss": 1.456,
      "step": 16464
    },
    {
      "epoch": 0.732115068249522,
      "grad_norm": 0.06057421490550041,
      "learning_rate": 0.00018167113935181273,
      "loss": 1.4593,
      "step": 16466
    },
    {
      "epoch": 0.7322039927081944,
      "grad_norm": 0.05998922884464264,
      "learning_rate": 0.00018155842014316547,
      "loss": 1.4598,
      "step": 16468
    },
    {
      "epoch": 0.7322929171668667,
      "grad_norm": 0.05810751020908356,
      "learning_rate": 0.0001814457281563078,
      "loss": 1.465,
      "step": 16470
    },
    {
      "epoch": 0.7323818416255391,
      "grad_norm": 0.05943676084280014,
      "learning_rate": 0.00018133306340087325,
      "loss": 1.4579,
      "step": 16472
    },
    {
      "epoch": 0.7324707660842115,
      "grad_norm": 0.05965890362858772,
      "learning_rate": 0.00018122042588649263,
      "loss": 1.4541,
      "step": 16474
    },
    {
      "epoch": 0.7325596905428838,
      "grad_norm": 0.05899742990732193,
      "learning_rate": 0.00018110781562279477,
      "loss": 1.4605,
      "step": 16476
    },
    {
      "epoch": 0.7326486150015562,
      "grad_norm": 0.06035918369889259,
      "learning_rate": 0.0001809952326194062,
      "loss": 1.4651,
      "step": 16478
    },
    {
      "epoch": 0.7327375394602286,
      "grad_norm": 0.05863313376903534,
      "learning_rate": 0.00018088267688595112,
      "loss": 1.4603,
      "step": 16480
    },
    {
      "epoch": 0.7328264639189009,
      "grad_norm": 0.0600309744477272,
      "learning_rate": 0.00018077014843205093,
      "loss": 1.4588,
      "step": 16482
    },
    {
      "epoch": 0.7329153883775733,
      "grad_norm": 0.05916277691721916,
      "learning_rate": 0.00018065764726732524,
      "loss": 1.4567,
      "step": 16484
    },
    {
      "epoch": 0.7330043128362456,
      "grad_norm": 0.05814970284700394,
      "learning_rate": 0.0001805451734013913,
      "loss": 1.4591,
      "step": 16486
    },
    {
      "epoch": 0.7330932372949179,
      "grad_norm": 0.05853576585650444,
      "learning_rate": 0.0001804327268438638,
      "loss": 1.4623,
      "step": 16488
    },
    {
      "epoch": 0.7331821617535903,
      "grad_norm": 0.061256907880306244,
      "learning_rate": 0.00018032030760435476,
      "loss": 1.4625,
      "step": 16490
    },
    {
      "epoch": 0.7332710862122627,
      "grad_norm": 0.05845688655972481,
      "learning_rate": 0.0001802079156924749,
      "loss": 1.4617,
      "step": 16492
    },
    {
      "epoch": 0.733360010670935,
      "grad_norm": 0.058402858674526215,
      "learning_rate": 0.00018009555111783154,
      "loss": 1.457,
      "step": 16494
    },
    {
      "epoch": 0.7334489351296074,
      "grad_norm": 0.0583033561706543,
      "learning_rate": 0.00017998321389003043,
      "loss": 1.4627,
      "step": 16496
    },
    {
      "epoch": 0.7335378595882798,
      "grad_norm": 0.05811697617173195,
      "learning_rate": 0.00017987090401867423,
      "loss": 1.4628,
      "step": 16498
    },
    {
      "epoch": 0.7336267840469521,
      "grad_norm": 0.060463517904281616,
      "learning_rate": 0.00017975862151336396,
      "loss": 1.4621,
      "step": 16500
    },
    {
      "epoch": 0.7336267840469521,
      "eval_loss": 1.4455397129058838,
      "eval_runtime": 12.4314,
      "eval_samples_per_second": 555.851,
      "eval_steps_per_second": 69.501,
      "step": 16500
    },
    {
      "epoch": 0.7337157085056245,
      "grad_norm": 0.06006486341357231,
      "learning_rate": 0.0001796463663836982,
      "loss": 1.4574,
      "step": 16502
    },
    {
      "epoch": 0.7338046329642969,
      "grad_norm": 0.05787288397550583,
      "learning_rate": 0.00017953413863927253,
      "loss": 1.4524,
      "step": 16504
    },
    {
      "epoch": 0.7338935574229691,
      "grad_norm": 0.058139074593782425,
      "learning_rate": 0.000179421938289681,
      "loss": 1.4588,
      "step": 16506
    },
    {
      "epoch": 0.7339824818816415,
      "grad_norm": 0.05989261716604233,
      "learning_rate": 0.00017930976534451503,
      "loss": 1.4555,
      "step": 16508
    },
    {
      "epoch": 0.7340714063403139,
      "grad_norm": 0.05919159948825836,
      "learning_rate": 0.00017919761981336347,
      "loss": 1.4618,
      "step": 16510
    },
    {
      "epoch": 0.7341603307989862,
      "grad_norm": 0.059001095592975616,
      "learning_rate": 0.00017908550170581307,
      "loss": 1.4559,
      "step": 16512
    },
    {
      "epoch": 0.7342492552576586,
      "grad_norm": 0.058799173682928085,
      "learning_rate": 0.00017897341103144837,
      "loss": 1.4628,
      "step": 16514
    },
    {
      "epoch": 0.734338179716331,
      "grad_norm": 0.06127883121371269,
      "learning_rate": 0.00017886134779985103,
      "loss": 1.4597,
      "step": 16516
    },
    {
      "epoch": 0.7344271041750033,
      "grad_norm": 0.05968368798494339,
      "learning_rate": 0.00017874931202060108,
      "loss": 1.4596,
      "step": 16518
    },
    {
      "epoch": 0.7345160286336757,
      "grad_norm": 0.05886247009038925,
      "learning_rate": 0.00017863730370327546,
      "loss": 1.4568,
      "step": 16520
    },
    {
      "epoch": 0.7346049530923481,
      "grad_norm": 0.057910215109586716,
      "learning_rate": 0.00017852532285744931,
      "loss": 1.4605,
      "step": 16522
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 0.06086508929729462,
      "learning_rate": 0.0001784133694926955,
      "loss": 1.4616,
      "step": 16524
    },
    {
      "epoch": 0.7347828020096928,
      "grad_norm": 0.05869476869702339,
      "learning_rate": 0.0001783014436185838,
      "loss": 1.456,
      "step": 16526
    },
    {
      "epoch": 0.7348717264683651,
      "grad_norm": 0.05936345458030701,
      "learning_rate": 0.0001781895452446824,
      "loss": 1.4626,
      "step": 16528
    },
    {
      "epoch": 0.7349606509270374,
      "grad_norm": 0.05925047770142555,
      "learning_rate": 0.000178077674380557,
      "loss": 1.4621,
      "step": 16530
    },
    {
      "epoch": 0.7350495753857098,
      "grad_norm": 0.05956713482737541,
      "learning_rate": 0.0001779658310357705,
      "loss": 1.4588,
      "step": 16532
    },
    {
      "epoch": 0.7351384998443822,
      "grad_norm": 0.06348378211259842,
      "learning_rate": 0.00017785401521988391,
      "loss": 1.4613,
      "step": 16534
    },
    {
      "epoch": 0.7352274243030545,
      "grad_norm": 0.059166379272937775,
      "learning_rate": 0.00017774222694245595,
      "loss": 1.4652,
      "step": 16536
    },
    {
      "epoch": 0.7353163487617269,
      "grad_norm": 0.0596497543156147,
      "learning_rate": 0.00017763046621304242,
      "loss": 1.4576,
      "step": 16538
    },
    {
      "epoch": 0.7354052732203993,
      "grad_norm": 0.0589081346988678,
      "learning_rate": 0.00017751873304119742,
      "loss": 1.4604,
      "step": 16540
    },
    {
      "epoch": 0.7354941976790716,
      "grad_norm": 0.06031009927392006,
      "learning_rate": 0.00017740702743647207,
      "loss": 1.4592,
      "step": 16542
    },
    {
      "epoch": 0.735583122137744,
      "grad_norm": 0.06029221788048744,
      "learning_rate": 0.00017729534940841564,
      "loss": 1.4599,
      "step": 16544
    },
    {
      "epoch": 0.7356720465964164,
      "grad_norm": 0.059007350355386734,
      "learning_rate": 0.00017718369896657505,
      "loss": 1.4602,
      "step": 16546
    },
    {
      "epoch": 0.7357609710550888,
      "grad_norm": 0.058588165789842606,
      "learning_rate": 0.0001770720761204943,
      "loss": 1.4562,
      "step": 16548
    },
    {
      "epoch": 0.735849895513761,
      "grad_norm": 0.05922502651810646,
      "learning_rate": 0.00017696048087971556,
      "loss": 1.4607,
      "step": 16550
    },
    {
      "epoch": 0.7359388199724334,
      "grad_norm": 0.06027013808488846,
      "learning_rate": 0.00017684891325377866,
      "loss": 1.4597,
      "step": 16552
    },
    {
      "epoch": 0.7360277444311057,
      "grad_norm": 0.05908496677875519,
      "learning_rate": 0.0001767373732522206,
      "loss": 1.462,
      "step": 16554
    },
    {
      "epoch": 0.7361166688897781,
      "grad_norm": 0.05901119112968445,
      "learning_rate": 0.00017662586088457643,
      "loss": 1.4629,
      "step": 16556
    },
    {
      "epoch": 0.7362055933484505,
      "grad_norm": 0.0609251968562603,
      "learning_rate": 0.0001765143761603789,
      "loss": 1.4652,
      "step": 16558
    },
    {
      "epoch": 0.7362945178071229,
      "grad_norm": 0.06070336699485779,
      "learning_rate": 0.0001764029190891579,
      "loss": 1.4599,
      "step": 16560
    },
    {
      "epoch": 0.7363834422657952,
      "grad_norm": 0.06053367629647255,
      "learning_rate": 0.0001762914896804414,
      "loss": 1.4651,
      "step": 16562
    },
    {
      "epoch": 0.7364723667244676,
      "grad_norm": 0.061270855367183685,
      "learning_rate": 0.0001761800879437549,
      "loss": 1.4535,
      "step": 16564
    },
    {
      "epoch": 0.73656129118314,
      "grad_norm": 0.05957317352294922,
      "learning_rate": 0.00017606871388862173,
      "loss": 1.4577,
      "step": 16566
    },
    {
      "epoch": 0.7366502156418123,
      "grad_norm": 0.05948825180530548,
      "learning_rate": 0.00017595736752456237,
      "loss": 1.4614,
      "step": 16568
    },
    {
      "epoch": 0.7367391401004847,
      "grad_norm": 0.060310911387205124,
      "learning_rate": 0.0001758460488610949,
      "loss": 1.4581,
      "step": 16570
    },
    {
      "epoch": 0.736828064559157,
      "grad_norm": 0.06115229055285454,
      "learning_rate": 0.000175734757907736,
      "loss": 1.4584,
      "step": 16572
    },
    {
      "epoch": 0.7369169890178293,
      "grad_norm": 0.059415318071842194,
      "learning_rate": 0.00017562349467399902,
      "loss": 1.4579,
      "step": 16574
    },
    {
      "epoch": 0.7370059134765017,
      "grad_norm": 0.059758272022008896,
      "learning_rate": 0.000175512259169395,
      "loss": 1.4589,
      "step": 16576
    },
    {
      "epoch": 0.737094837935174,
      "grad_norm": 0.059409987181425095,
      "learning_rate": 0.0001754010514034331,
      "loss": 1.4623,
      "step": 16578
    },
    {
      "epoch": 0.7371837623938464,
      "grad_norm": 0.06052602455019951,
      "learning_rate": 0.0001752898713856197,
      "loss": 1.4563,
      "step": 16580
    },
    {
      "epoch": 0.7372726868525188,
      "grad_norm": 0.05821981281042099,
      "learning_rate": 0.0001751787191254593,
      "loss": 1.4576,
      "step": 16582
    },
    {
      "epoch": 0.7373616113111912,
      "grad_norm": 0.06081145256757736,
      "learning_rate": 0.0001750675946324532,
      "loss": 1.4663,
      "step": 16584
    },
    {
      "epoch": 0.7374505357698635,
      "grad_norm": 0.06057305634021759,
      "learning_rate": 0.00017495649791610102,
      "loss": 1.4634,
      "step": 16586
    },
    {
      "epoch": 0.7375394602285359,
      "grad_norm": 0.05856640636920929,
      "learning_rate": 0.00017484542898590005,
      "loss": 1.4573,
      "step": 16588
    },
    {
      "epoch": 0.7376283846872083,
      "grad_norm": 0.0607236847281456,
      "learning_rate": 0.00017473438785134472,
      "loss": 1.455,
      "step": 16590
    },
    {
      "epoch": 0.7377173091458806,
      "grad_norm": 0.060658324509859085,
      "learning_rate": 0.00017462337452192695,
      "loss": 1.4624,
      "step": 16592
    },
    {
      "epoch": 0.7378062336045529,
      "grad_norm": 0.06017633154988289,
      "learning_rate": 0.00017451238900713733,
      "loss": 1.4547,
      "step": 16594
    },
    {
      "epoch": 0.7378951580632253,
      "grad_norm": 0.05868356674909592,
      "learning_rate": 0.00017440143131646307,
      "loss": 1.4602,
      "step": 16596
    },
    {
      "epoch": 0.7379840825218976,
      "grad_norm": 0.05967675894498825,
      "learning_rate": 0.0001742905014593892,
      "loss": 1.4572,
      "step": 16598
    },
    {
      "epoch": 0.73807300698057,
      "grad_norm": 0.06043080985546112,
      "learning_rate": 0.00017417959944539862,
      "loss": 1.4637,
      "step": 16600
    },
    {
      "epoch": 0.7381619314392424,
      "grad_norm": 0.059699367731809616,
      "learning_rate": 0.00017406872528397173,
      "loss": 1.4584,
      "step": 16602
    },
    {
      "epoch": 0.7382508558979147,
      "grad_norm": 0.059518732130527496,
      "learning_rate": 0.00017395787898458664,
      "loss": 1.4556,
      "step": 16604
    },
    {
      "epoch": 0.7383397803565871,
      "grad_norm": 0.059061914682388306,
      "learning_rate": 0.0001738470605567188,
      "loss": 1.4565,
      "step": 16606
    },
    {
      "epoch": 0.7384287048152595,
      "grad_norm": 0.058894891291856766,
      "learning_rate": 0.00017373627000984144,
      "loss": 1.4545,
      "step": 16608
    },
    {
      "epoch": 0.7385176292739318,
      "grad_norm": 0.05930294841527939,
      "learning_rate": 0.00017362550735342574,
      "loss": 1.459,
      "step": 16610
    },
    {
      "epoch": 0.7386065537326042,
      "grad_norm": 0.06055053696036339,
      "learning_rate": 0.00017351477259693976,
      "loss": 1.4617,
      "step": 16612
    },
    {
      "epoch": 0.7386954781912765,
      "grad_norm": 0.05868672579526901,
      "learning_rate": 0.00017340406574984984,
      "loss": 1.4601,
      "step": 16614
    },
    {
      "epoch": 0.7387844026499488,
      "grad_norm": 0.05979511886835098,
      "learning_rate": 0.00017329338682161978,
      "loss": 1.4574,
      "step": 16616
    },
    {
      "epoch": 0.7388733271086212,
      "grad_norm": 0.05988418683409691,
      "learning_rate": 0.00017318273582171058,
      "loss": 1.459,
      "step": 16618
    },
    {
      "epoch": 0.7389622515672936,
      "grad_norm": 0.05983160063624382,
      "learning_rate": 0.0001730721127595814,
      "loss": 1.4637,
      "step": 16620
    },
    {
      "epoch": 0.7390511760259659,
      "grad_norm": 0.05880555137991905,
      "learning_rate": 0.00017296151764468887,
      "loss": 1.4618,
      "step": 16622
    },
    {
      "epoch": 0.7391401004846383,
      "grad_norm": 0.058000050485134125,
      "learning_rate": 0.00017285095048648685,
      "loss": 1.4601,
      "step": 16624
    },
    {
      "epoch": 0.7392290249433107,
      "grad_norm": 0.058923520147800446,
      "learning_rate": 0.00017274041129442753,
      "loss": 1.46,
      "step": 16626
    },
    {
      "epoch": 0.739317949401983,
      "grad_norm": 0.059510573744773865,
      "learning_rate": 0.00017262990007795975,
      "loss": 1.4588,
      "step": 16628
    },
    {
      "epoch": 0.7394068738606554,
      "grad_norm": 0.05785347521305084,
      "learning_rate": 0.0001725194168465309,
      "loss": 1.4559,
      "step": 16630
    },
    {
      "epoch": 0.7394957983193278,
      "grad_norm": 0.05921796336770058,
      "learning_rate": 0.00017240896160958559,
      "loss": 1.4559,
      "step": 16632
    },
    {
      "epoch": 0.7395847227780001,
      "grad_norm": 0.059610430151224136,
      "learning_rate": 0.00017229853437656568,
      "loss": 1.4579,
      "step": 16634
    },
    {
      "epoch": 0.7396736472366724,
      "grad_norm": 0.058959461748600006,
      "learning_rate": 0.00017218813515691117,
      "loss": 1.4584,
      "step": 16636
    },
    {
      "epoch": 0.7397625716953448,
      "grad_norm": 0.05981831252574921,
      "learning_rate": 0.00017207776396005976,
      "loss": 1.4602,
      "step": 16638
    },
    {
      "epoch": 0.7398514961540171,
      "grad_norm": 0.060530826449394226,
      "learning_rate": 0.0001719674207954459,
      "loss": 1.4639,
      "step": 16640
    },
    {
      "epoch": 0.7399404206126895,
      "grad_norm": 0.05915441736578941,
      "learning_rate": 0.00017185710567250257,
      "loss": 1.4624,
      "step": 16642
    },
    {
      "epoch": 0.7400293450713619,
      "grad_norm": 0.05957896262407303,
      "learning_rate": 0.0001717468186006601,
      "loss": 1.4577,
      "step": 16644
    },
    {
      "epoch": 0.7401182695300342,
      "grad_norm": 0.05835951864719391,
      "learning_rate": 0.000171636559589346,
      "loss": 1.4562,
      "step": 16646
    },
    {
      "epoch": 0.7402071939887066,
      "grad_norm": 0.06005425006151199,
      "learning_rate": 0.00017152632864798594,
      "loss": 1.4611,
      "step": 16648
    },
    {
      "epoch": 0.740296118447379,
      "grad_norm": 0.06107817590236664,
      "learning_rate": 0.00017141612578600252,
      "loss": 1.4582,
      "step": 16650
    },
    {
      "epoch": 0.7403850429060513,
      "grad_norm": 0.05926118418574333,
      "learning_rate": 0.00017130595101281704,
      "loss": 1.4605,
      "step": 16652
    },
    {
      "epoch": 0.7404739673647237,
      "grad_norm": 0.06142883375287056,
      "learning_rate": 0.00017119580433784737,
      "loss": 1.469,
      "step": 16654
    },
    {
      "epoch": 0.7405628918233961,
      "grad_norm": 0.059701576828956604,
      "learning_rate": 0.0001710856857705092,
      "loss": 1.455,
      "step": 16656
    },
    {
      "epoch": 0.7406518162820683,
      "grad_norm": 0.059347156435251236,
      "learning_rate": 0.00017097559532021605,
      "loss": 1.4636,
      "step": 16658
    },
    {
      "epoch": 0.7407407407407407,
      "grad_norm": 0.05961744487285614,
      "learning_rate": 0.00017086553299637913,
      "loss": 1.4567,
      "step": 16660
    },
    {
      "epoch": 0.7408296651994131,
      "grad_norm": 0.05953764170408249,
      "learning_rate": 0.0001707554988084068,
      "loss": 1.4541,
      "step": 16662
    },
    {
      "epoch": 0.7409185896580854,
      "grad_norm": 0.060986440628767014,
      "learning_rate": 0.00017064549276570528,
      "loss": 1.467,
      "step": 16664
    },
    {
      "epoch": 0.7410075141167578,
      "grad_norm": 0.05836610123515129,
      "learning_rate": 0.00017053551487767867,
      "loss": 1.4574,
      "step": 16666
    },
    {
      "epoch": 0.7410964385754302,
      "grad_norm": 0.06008746847510338,
      "learning_rate": 0.00017042556515372797,
      "loss": 1.4545,
      "step": 16668
    },
    {
      "epoch": 0.7411853630341025,
      "grad_norm": 0.06051292270421982,
      "learning_rate": 0.00017031564360325236,
      "loss": 1.4583,
      "step": 16670
    },
    {
      "epoch": 0.7412742874927749,
      "grad_norm": 0.060724154114723206,
      "learning_rate": 0.00017020575023564844,
      "loss": 1.4589,
      "step": 16672
    },
    {
      "epoch": 0.7413632119514473,
      "grad_norm": 0.05805688351392746,
      "learning_rate": 0.00017009588506031043,
      "loss": 1.4577,
      "step": 16674
    },
    {
      "epoch": 0.7414521364101196,
      "grad_norm": 0.05907231569290161,
      "learning_rate": 0.0001699860480866301,
      "loss": 1.4653,
      "step": 16676
    },
    {
      "epoch": 0.741541060868792,
      "grad_norm": 0.060612257570028305,
      "learning_rate": 0.00016987623932399631,
      "loss": 1.4621,
      "step": 16678
    },
    {
      "epoch": 0.7416299853274643,
      "grad_norm": 0.059990622103214264,
      "learning_rate": 0.00016976645878179675,
      "loss": 1.4614,
      "step": 16680
    },
    {
      "epoch": 0.7417189097861366,
      "grad_norm": 0.058527737855911255,
      "learning_rate": 0.00016965670646941556,
      "loss": 1.4589,
      "step": 16682
    },
    {
      "epoch": 0.741807834244809,
      "grad_norm": 0.059194039553403854,
      "learning_rate": 0.00016954698239623472,
      "loss": 1.4593,
      "step": 16684
    },
    {
      "epoch": 0.7418967587034814,
      "grad_norm": 0.06000838056206703,
      "learning_rate": 0.0001694372865716341,
      "loss": 1.455,
      "step": 16686
    },
    {
      "epoch": 0.7419856831621537,
      "grad_norm": 0.05965815484523773,
      "learning_rate": 0.00016932761900499094,
      "loss": 1.4623,
      "step": 16688
    },
    {
      "epoch": 0.7420746076208261,
      "grad_norm": 0.05837101861834526,
      "learning_rate": 0.00016921797970568031,
      "loss": 1.4601,
      "step": 16690
    },
    {
      "epoch": 0.7421635320794985,
      "grad_norm": 0.06096314638853073,
      "learning_rate": 0.00016910836868307434,
      "loss": 1.4579,
      "step": 16692
    },
    {
      "epoch": 0.7422524565381708,
      "grad_norm": 0.05966602638363838,
      "learning_rate": 0.00016899878594654315,
      "loss": 1.4562,
      "step": 16694
    },
    {
      "epoch": 0.7423413809968432,
      "grad_norm": 0.059270136058330536,
      "learning_rate": 0.0001688892315054546,
      "loss": 1.4612,
      "step": 16696
    },
    {
      "epoch": 0.7424303054555156,
      "grad_norm": 0.05927656963467598,
      "learning_rate": 0.00016877970536917375,
      "loss": 1.455,
      "step": 16698
    },
    {
      "epoch": 0.742519229914188,
      "grad_norm": 0.06073769927024841,
      "learning_rate": 0.00016867020754706292,
      "loss": 1.4615,
      "step": 16700
    },
    {
      "epoch": 0.7426081543728602,
      "grad_norm": 0.05787941813468933,
      "learning_rate": 0.00016856073804848328,
      "loss": 1.4577,
      "step": 16702
    },
    {
      "epoch": 0.7426970788315326,
      "grad_norm": 0.06029796600341797,
      "learning_rate": 0.00016845129688279216,
      "loss": 1.4592,
      "step": 16704
    },
    {
      "epoch": 0.7427860032902049,
      "grad_norm": 0.05889935791492462,
      "learning_rate": 0.00016834188405934542,
      "loss": 1.4618,
      "step": 16706
    },
    {
      "epoch": 0.7428749277488773,
      "grad_norm": 0.058539580553770065,
      "learning_rate": 0.00016823249958749581,
      "loss": 1.4616,
      "step": 16708
    },
    {
      "epoch": 0.7429638522075497,
      "grad_norm": 0.06004004552960396,
      "learning_rate": 0.00016812314347659424,
      "loss": 1.4536,
      "step": 16710
    },
    {
      "epoch": 0.743052776666222,
      "grad_norm": 0.060982439666986465,
      "learning_rate": 0.00016801381573598907,
      "loss": 1.4549,
      "step": 16712
    },
    {
      "epoch": 0.7431417011248944,
      "grad_norm": 0.05949532985687256,
      "learning_rate": 0.00016790451637502574,
      "loss": 1.4583,
      "step": 16714
    },
    {
      "epoch": 0.7432306255835668,
      "grad_norm": 0.05871553719043732,
      "learning_rate": 0.00016779524540304786,
      "loss": 1.4598,
      "step": 16716
    },
    {
      "epoch": 0.7433195500422392,
      "grad_norm": 0.059430256485939026,
      "learning_rate": 0.00016768600282939655,
      "loss": 1.4618,
      "step": 16718
    },
    {
      "epoch": 0.7434084745009115,
      "grad_norm": 0.059652287513017654,
      "learning_rate": 0.00016757678866341003,
      "loss": 1.4654,
      "step": 16720
    },
    {
      "epoch": 0.7434973989595839,
      "grad_norm": 0.05944843962788582,
      "learning_rate": 0.00016746760291442452,
      "loss": 1.4606,
      "step": 16722
    },
    {
      "epoch": 0.7435863234182561,
      "grad_norm": 0.059113163501024246,
      "learning_rate": 0.00016735844559177392,
      "loss": 1.4584,
      "step": 16724
    },
    {
      "epoch": 0.7436752478769285,
      "grad_norm": 0.05857875198125839,
      "learning_rate": 0.00016724931670478911,
      "loss": 1.4563,
      "step": 16726
    },
    {
      "epoch": 0.7437641723356009,
      "grad_norm": 0.06093486398458481,
      "learning_rate": 0.00016714021626279912,
      "loss": 1.4632,
      "step": 16728
    },
    {
      "epoch": 0.7438530967942732,
      "grad_norm": 0.05900067463517189,
      "learning_rate": 0.0001670311442751305,
      "loss": 1.465,
      "step": 16730
    },
    {
      "epoch": 0.7439420212529456,
      "grad_norm": 0.05907110497355461,
      "learning_rate": 0.0001669221007511068,
      "loss": 1.4637,
      "step": 16732
    },
    {
      "epoch": 0.744030945711618,
      "grad_norm": 0.05955798551440239,
      "learning_rate": 0.00016681308570004992,
      "loss": 1.4578,
      "step": 16734
    },
    {
      "epoch": 0.7441198701702904,
      "grad_norm": 0.05999176949262619,
      "learning_rate": 0.00016670409913127865,
      "loss": 1.456,
      "step": 16736
    },
    {
      "epoch": 0.7442087946289627,
      "grad_norm": 0.05871756374835968,
      "learning_rate": 0.00016659514105410973,
      "loss": 1.4558,
      "step": 16738
    },
    {
      "epoch": 0.7442977190876351,
      "grad_norm": 0.05930541455745697,
      "learning_rate": 0.00016648621147785764,
      "loss": 1.4695,
      "step": 16740
    },
    {
      "epoch": 0.7443866435463075,
      "grad_norm": 0.0586344413459301,
      "learning_rate": 0.00016637731041183374,
      "loss": 1.4612,
      "step": 16742
    },
    {
      "epoch": 0.7444755680049797,
      "grad_norm": 0.05983453243970871,
      "learning_rate": 0.00016626843786534756,
      "loss": 1.4609,
      "step": 16744
    },
    {
      "epoch": 0.7445644924636521,
      "grad_norm": 0.058784399181604385,
      "learning_rate": 0.00016615959384770623,
      "loss": 1.4539,
      "step": 16746
    },
    {
      "epoch": 0.7446534169223245,
      "grad_norm": 0.05940399691462517,
      "learning_rate": 0.0001660507783682138,
      "loss": 1.4582,
      "step": 16748
    },
    {
      "epoch": 0.7447423413809968,
      "grad_norm": 0.05869759991765022,
      "learning_rate": 0.00016594199143617251,
      "loss": 1.4617,
      "step": 16750
    },
    {
      "epoch": 0.7448312658396692,
      "grad_norm": 0.059345047920942307,
      "learning_rate": 0.0001658332330608821,
      "loss": 1.4588,
      "step": 16752
    },
    {
      "epoch": 0.7449201902983416,
      "grad_norm": 0.059553761035203934,
      "learning_rate": 0.0001657245032516394,
      "loss": 1.4596,
      "step": 16754
    },
    {
      "epoch": 0.7450091147570139,
      "grad_norm": 0.05867783725261688,
      "learning_rate": 0.00016561580201773947,
      "loss": 1.4613,
      "step": 16756
    },
    {
      "epoch": 0.7450980392156863,
      "grad_norm": 0.057739853858947754,
      "learning_rate": 0.00016550712936847395,
      "loss": 1.4595,
      "step": 16758
    },
    {
      "epoch": 0.7451869636743587,
      "grad_norm": 0.0593368336558342,
      "learning_rate": 0.00016539848531313346,
      "loss": 1.4607,
      "step": 16760
    },
    {
      "epoch": 0.745275888133031,
      "grad_norm": 0.058571264147758484,
      "learning_rate": 0.00016528986986100497,
      "loss": 1.4567,
      "step": 16762
    },
    {
      "epoch": 0.7453648125917034,
      "grad_norm": 0.06027856841683388,
      "learning_rate": 0.00016518128302137324,
      "loss": 1.4572,
      "step": 16764
    },
    {
      "epoch": 0.7454537370503757,
      "grad_norm": 0.060379598289728165,
      "learning_rate": 0.000165072724803521,
      "loss": 1.4616,
      "step": 16766
    },
    {
      "epoch": 0.745542661509048,
      "grad_norm": 0.06043410673737526,
      "learning_rate": 0.00016496419521672834,
      "loss": 1.463,
      "step": 16768
    },
    {
      "epoch": 0.7456315859677204,
      "grad_norm": 0.06077970936894417,
      "learning_rate": 0.00016485569427027253,
      "loss": 1.4633,
      "step": 16770
    },
    {
      "epoch": 0.7457205104263928,
      "grad_norm": 0.05859680846333504,
      "learning_rate": 0.00016474722197342895,
      "loss": 1.4503,
      "step": 16772
    },
    {
      "epoch": 0.7458094348850651,
      "grad_norm": 0.06022398918867111,
      "learning_rate": 0.00016463877833547026,
      "loss": 1.4622,
      "step": 16774
    },
    {
      "epoch": 0.7458983593437375,
      "grad_norm": 0.05970166251063347,
      "learning_rate": 0.00016453036336566685,
      "loss": 1.4621,
      "step": 16776
    },
    {
      "epoch": 0.7459872838024099,
      "grad_norm": 0.05986715480685234,
      "learning_rate": 0.00016442197707328615,
      "loss": 1.4628,
      "step": 16778
    },
    {
      "epoch": 0.7460762082610822,
      "grad_norm": 0.05800659954547882,
      "learning_rate": 0.0001643136194675937,
      "loss": 1.4602,
      "step": 16780
    },
    {
      "epoch": 0.7461651327197546,
      "grad_norm": 0.05899570882320404,
      "learning_rate": 0.00016420529055785255,
      "loss": 1.4611,
      "step": 16782
    },
    {
      "epoch": 0.746254057178427,
      "grad_norm": 0.05969346687197685,
      "learning_rate": 0.00016409699035332297,
      "loss": 1.4599,
      "step": 16784
    },
    {
      "epoch": 0.7463429816370993,
      "grad_norm": 0.0593080073595047,
      "learning_rate": 0.00016398871886326262,
      "loss": 1.4559,
      "step": 16786
    },
    {
      "epoch": 0.7464319060957716,
      "grad_norm": 0.05878398194909096,
      "learning_rate": 0.00016388047609692763,
      "loss": 1.4562,
      "step": 16788
    },
    {
      "epoch": 0.746520830554444,
      "grad_norm": 0.05969779193401337,
      "learning_rate": 0.00016377226206357076,
      "loss": 1.4649,
      "step": 16790
    },
    {
      "epoch": 0.7466097550131163,
      "grad_norm": 0.059288714081048965,
      "learning_rate": 0.00016366407677244249,
      "loss": 1.4623,
      "step": 16792
    },
    {
      "epoch": 0.7466986794717887,
      "grad_norm": 0.06077198311686516,
      "learning_rate": 0.00016355592023279113,
      "loss": 1.459,
      "step": 16794
    },
    {
      "epoch": 0.7467876039304611,
      "grad_norm": 0.060099512338638306,
      "learning_rate": 0.00016344779245386237,
      "loss": 1.4619,
      "step": 16796
    },
    {
      "epoch": 0.7468765283891334,
      "grad_norm": 0.06000978872179985,
      "learning_rate": 0.00016333969344489962,
      "loss": 1.4646,
      "step": 16798
    },
    {
      "epoch": 0.7469654528478058,
      "grad_norm": 0.05997542664408684,
      "learning_rate": 0.00016323162321514335,
      "loss": 1.4584,
      "step": 16800
    },
    {
      "epoch": 0.7470543773064782,
      "grad_norm": 0.059203166514635086,
      "learning_rate": 0.00016312358177383195,
      "loss": 1.461,
      "step": 16802
    },
    {
      "epoch": 0.7471433017651505,
      "grad_norm": 0.059819675981998444,
      "learning_rate": 0.00016301556913020154,
      "loss": 1.4598,
      "step": 16804
    },
    {
      "epoch": 0.7472322262238229,
      "grad_norm": 0.05942592769861221,
      "learning_rate": 0.0001629075852934853,
      "loss": 1.4564,
      "step": 16806
    },
    {
      "epoch": 0.7473211506824953,
      "grad_norm": 0.05992693081498146,
      "learning_rate": 0.0001627996302729139,
      "loss": 1.4544,
      "step": 16808
    },
    {
      "epoch": 0.7474100751411675,
      "grad_norm": 0.06033800169825554,
      "learning_rate": 0.00016269170407771645,
      "loss": 1.4539,
      "step": 16810
    },
    {
      "epoch": 0.7474989995998399,
      "grad_norm": 0.05810123309493065,
      "learning_rate": 0.00016258380671711837,
      "loss": 1.4565,
      "step": 16812
    },
    {
      "epoch": 0.7475879240585123,
      "grad_norm": 0.059875525534152985,
      "learning_rate": 0.00016247593820034362,
      "loss": 1.4554,
      "step": 16814
    },
    {
      "epoch": 0.7476768485171846,
      "grad_norm": 0.05907121300697327,
      "learning_rate": 0.00016236809853661295,
      "loss": 1.4597,
      "step": 16816
    },
    {
      "epoch": 0.747765772975857,
      "grad_norm": 0.05930914729833603,
      "learning_rate": 0.00016226028773514513,
      "loss": 1.4576,
      "step": 16818
    },
    {
      "epoch": 0.7478546974345294,
      "grad_norm": 0.0589991994202137,
      "learning_rate": 0.0001621525058051564,
      "loss": 1.4528,
      "step": 16820
    },
    {
      "epoch": 0.7479436218932017,
      "grad_norm": 0.05885922908782959,
      "learning_rate": 0.00016204475275586016,
      "loss": 1.4642,
      "step": 16822
    },
    {
      "epoch": 0.7480325463518741,
      "grad_norm": 0.058898232877254486,
      "learning_rate": 0.0001619370285964678,
      "loss": 1.4609,
      "step": 16824
    },
    {
      "epoch": 0.7481214708105465,
      "grad_norm": 0.058432307094335556,
      "learning_rate": 0.0001618293333361882,
      "loss": 1.4588,
      "step": 16826
    },
    {
      "epoch": 0.7482103952692188,
      "grad_norm": 0.059765320271253586,
      "learning_rate": 0.0001617216669842273,
      "loss": 1.4541,
      "step": 16828
    },
    {
      "epoch": 0.7482993197278912,
      "grad_norm": 0.05895643308758736,
      "learning_rate": 0.000161614029549789,
      "loss": 1.4602,
      "step": 16830
    },
    {
      "epoch": 0.7483882441865635,
      "grad_norm": 0.05961558222770691,
      "learning_rate": 0.00016150642104207486,
      "loss": 1.4598,
      "step": 16832
    },
    {
      "epoch": 0.7484771686452358,
      "grad_norm": 0.06014033779501915,
      "learning_rate": 0.0001613988414702834,
      "loss": 1.4579,
      "step": 16834
    },
    {
      "epoch": 0.7485660931039082,
      "grad_norm": 0.06016629934310913,
      "learning_rate": 0.00016129129084361128,
      "loss": 1.4579,
      "step": 16836
    },
    {
      "epoch": 0.7486550175625806,
      "grad_norm": 0.05935119092464447,
      "learning_rate": 0.0001611837691712521,
      "loss": 1.4623,
      "step": 16838
    },
    {
      "epoch": 0.7487439420212529,
      "grad_norm": 0.060207169502973557,
      "learning_rate": 0.00016107627646239748,
      "loss": 1.4623,
      "step": 16840
    },
    {
      "epoch": 0.7488328664799253,
      "grad_norm": 0.05937547609210014,
      "learning_rate": 0.00016096881272623643,
      "loss": 1.4519,
      "step": 16842
    },
    {
      "epoch": 0.7489217909385977,
      "grad_norm": 0.059106145054101944,
      "learning_rate": 0.0001608613779719552,
      "loss": 1.4604,
      "step": 16844
    },
    {
      "epoch": 0.74901071539727,
      "grad_norm": 0.05962870270013809,
      "learning_rate": 0.00016075397220873794,
      "loss": 1.4623,
      "step": 16846
    },
    {
      "epoch": 0.7490996398559424,
      "grad_norm": 0.059200629591941833,
      "learning_rate": 0.0001606465954457663,
      "loss": 1.46,
      "step": 16848
    },
    {
      "epoch": 0.7491885643146148,
      "grad_norm": 0.05879664793610573,
      "learning_rate": 0.000160539247692219,
      "loss": 1.4581,
      "step": 16850
    },
    {
      "epoch": 0.7492774887732871,
      "grad_norm": 0.05944131687283516,
      "learning_rate": 0.0001604319289572727,
      "loss": 1.459,
      "step": 16852
    },
    {
      "epoch": 0.7493664132319594,
      "grad_norm": 0.05925022438168526,
      "learning_rate": 0.00016032463925010176,
      "loss": 1.4585,
      "step": 16854
    },
    {
      "epoch": 0.7494553376906318,
      "grad_norm": 0.06039819493889809,
      "learning_rate": 0.0001602173785798774,
      "loss": 1.466,
      "step": 16856
    },
    {
      "epoch": 0.7495442621493041,
      "grad_norm": 0.05966150015592575,
      "learning_rate": 0.0001601101469557689,
      "loss": 1.4556,
      "step": 16858
    },
    {
      "epoch": 0.7496331866079765,
      "grad_norm": 0.05846158042550087,
      "learning_rate": 0.000160002944386943,
      "loss": 1.4595,
      "step": 16860
    },
    {
      "epoch": 0.7497221110666489,
      "grad_norm": 0.05919952318072319,
      "learning_rate": 0.00015989577088256363,
      "loss": 1.4562,
      "step": 16862
    },
    {
      "epoch": 0.7498110355253212,
      "grad_norm": 0.05901890993118286,
      "learning_rate": 0.0001597886264517927,
      "loss": 1.4583,
      "step": 16864
    },
    {
      "epoch": 0.7498999599839936,
      "grad_norm": 0.058764733374118805,
      "learning_rate": 0.00015968151110378888,
      "loss": 1.4553,
      "step": 16866
    },
    {
      "epoch": 0.749988884442666,
      "grad_norm": 0.05931887403130531,
      "learning_rate": 0.00015957442484770956,
      "loss": 1.4591,
      "step": 16868
    },
    {
      "epoch": 0.7500778089013383,
      "grad_norm": 0.05878930538892746,
      "learning_rate": 0.00015946736769270863,
      "loss": 1.4572,
      "step": 16870
    },
    {
      "epoch": 0.7501667333600107,
      "grad_norm": 0.060362525284290314,
      "learning_rate": 0.00015936033964793767,
      "loss": 1.4579,
      "step": 16872
    },
    {
      "epoch": 0.750255657818683,
      "grad_norm": 0.059850580990314484,
      "learning_rate": 0.00015925334072254595,
      "loss": 1.4622,
      "step": 16874
    },
    {
      "epoch": 0.7503445822773553,
      "grad_norm": 0.05898622050881386,
      "learning_rate": 0.00015914637092568045,
      "loss": 1.4627,
      "step": 16876
    },
    {
      "epoch": 0.7504335067360277,
      "grad_norm": 0.05974709987640381,
      "learning_rate": 0.00015903943026648515,
      "loss": 1.4618,
      "step": 16878
    },
    {
      "epoch": 0.7505224311947001,
      "grad_norm": 0.05845978111028671,
      "learning_rate": 0.0001589325187541019,
      "loss": 1.4622,
      "step": 16880
    },
    {
      "epoch": 0.7506113556533724,
      "grad_norm": 0.059004079550504684,
      "learning_rate": 0.00015882563639766994,
      "loss": 1.4616,
      "step": 16882
    },
    {
      "epoch": 0.7507002801120448,
      "grad_norm": 0.06059623137116432,
      "learning_rate": 0.0001587187832063263,
      "loss": 1.4603,
      "step": 16884
    },
    {
      "epoch": 0.7507892045707172,
      "grad_norm": 0.05917629599571228,
      "learning_rate": 0.00015861195918920502,
      "loss": 1.4617,
      "step": 16886
    },
    {
      "epoch": 0.7508781290293896,
      "grad_norm": 0.0598984956741333,
      "learning_rate": 0.0001585051643554376,
      "loss": 1.4587,
      "step": 16888
    },
    {
      "epoch": 0.7509670534880619,
      "grad_norm": 0.061087653040885925,
      "learning_rate": 0.000158398398714154,
      "loss": 1.4605,
      "step": 16890
    },
    {
      "epoch": 0.7510559779467343,
      "grad_norm": 0.0595669187605381,
      "learning_rate": 0.00015829166227448066,
      "loss": 1.4579,
      "step": 16892
    },
    {
      "epoch": 0.7511449024054067,
      "grad_norm": 0.05967925488948822,
      "learning_rate": 0.0001581849550455417,
      "loss": 1.4628,
      "step": 16894
    },
    {
      "epoch": 0.7512338268640789,
      "grad_norm": 0.060816604644060135,
      "learning_rate": 0.0001580782770364591,
      "loss": 1.4586,
      "step": 16896
    },
    {
      "epoch": 0.7513227513227513,
      "grad_norm": 0.060024797916412354,
      "learning_rate": 0.00015797162825635218,
      "loss": 1.4575,
      "step": 16898
    },
    {
      "epoch": 0.7514116757814236,
      "grad_norm": 0.058512914925813675,
      "learning_rate": 0.00015786500871433796,
      "loss": 1.4558,
      "step": 16900
    },
    {
      "epoch": 0.751500600240096,
      "grad_norm": 0.05933115631341934,
      "learning_rate": 0.00015775841841953036,
      "loss": 1.4552,
      "step": 16902
    },
    {
      "epoch": 0.7515895246987684,
      "grad_norm": 0.059899043291807175,
      "learning_rate": 0.00015765185738104133,
      "loss": 1.4588,
      "step": 16904
    },
    {
      "epoch": 0.7516784491574408,
      "grad_norm": 0.05907595157623291,
      "learning_rate": 0.0001575453256079804,
      "loss": 1.4609,
      "step": 16906
    },
    {
      "epoch": 0.7517673736161131,
      "grad_norm": 0.05995360016822815,
      "learning_rate": 0.000157438823109454,
      "loss": 1.4613,
      "step": 16908
    },
    {
      "epoch": 0.7518562980747855,
      "grad_norm": 0.05959051102399826,
      "learning_rate": 0.00015733234989456663,
      "loss": 1.4605,
      "step": 16910
    },
    {
      "epoch": 0.7519452225334579,
      "grad_norm": 0.06011039391160011,
      "learning_rate": 0.00015722590597242025,
      "loss": 1.4554,
      "step": 16912
    },
    {
      "epoch": 0.7520341469921302,
      "grad_norm": 0.05920354649424553,
      "learning_rate": 0.00015711949135211394,
      "loss": 1.4523,
      "step": 16914
    },
    {
      "epoch": 0.7521230714508026,
      "grad_norm": 0.059563346207141876,
      "learning_rate": 0.00015701310604274426,
      "loss": 1.4578,
      "step": 16916
    },
    {
      "epoch": 0.7522119959094749,
      "grad_norm": 0.05948107689619064,
      "learning_rate": 0.0001569067500534061,
      "loss": 1.4571,
      "step": 16918
    },
    {
      "epoch": 0.7523009203681472,
      "grad_norm": 0.058655984699726105,
      "learning_rate": 0.00015680042339319072,
      "loss": 1.4562,
      "step": 16920
    },
    {
      "epoch": 0.7523898448268196,
      "grad_norm": 0.05892255902290344,
      "learning_rate": 0.00015669412607118778,
      "loss": 1.4593,
      "step": 16922
    },
    {
      "epoch": 0.752478769285492,
      "grad_norm": 0.05882655829191208,
      "learning_rate": 0.00015658785809648362,
      "loss": 1.4542,
      "step": 16924
    },
    {
      "epoch": 0.7525676937441643,
      "grad_norm": 0.058582112193107605,
      "learning_rate": 0.00015648161947816276,
      "loss": 1.4666,
      "step": 16926
    },
    {
      "epoch": 0.7526566182028367,
      "grad_norm": 0.05810071900486946,
      "learning_rate": 0.00015637541022530706,
      "loss": 1.4571,
      "step": 16928
    },
    {
      "epoch": 0.7527455426615091,
      "grad_norm": 0.0615081861615181,
      "learning_rate": 0.00015626923034699542,
      "loss": 1.4623,
      "step": 16930
    },
    {
      "epoch": 0.7528344671201814,
      "grad_norm": 0.0594865120947361,
      "learning_rate": 0.00015616307985230477,
      "loss": 1.4576,
      "step": 16932
    },
    {
      "epoch": 0.7529233915788538,
      "grad_norm": 0.05900559946894646,
      "learning_rate": 0.0001560569587503094,
      "loss": 1.4514,
      "step": 16934
    },
    {
      "epoch": 0.7530123160375262,
      "grad_norm": 0.06001661345362663,
      "learning_rate": 0.00015595086705008076,
      "loss": 1.4621,
      "step": 16936
    },
    {
      "epoch": 0.7531012404961985,
      "grad_norm": 0.05865577980875969,
      "learning_rate": 0.00015584480476068825,
      "loss": 1.4546,
      "step": 16938
    },
    {
      "epoch": 0.7531901649548708,
      "grad_norm": 0.05930325761437416,
      "learning_rate": 0.00015573877189119855,
      "loss": 1.462,
      "step": 16940
    },
    {
      "epoch": 0.7532790894135432,
      "grad_norm": 0.05888302996754646,
      "learning_rate": 0.00015563276845067565,
      "loss": 1.4552,
      "step": 16942
    },
    {
      "epoch": 0.7533680138722155,
      "grad_norm": 0.06017635017633438,
      "learning_rate": 0.00015552679444818135,
      "loss": 1.4547,
      "step": 16944
    },
    {
      "epoch": 0.7534569383308879,
      "grad_norm": 0.06248828396201134,
      "learning_rate": 0.00015542084989277462,
      "loss": 1.4556,
      "step": 16946
    },
    {
      "epoch": 0.7535458627895603,
      "grad_norm": 0.058105915784835815,
      "learning_rate": 0.0001553149347935121,
      "loss": 1.4615,
      "step": 16948
    },
    {
      "epoch": 0.7536347872482326,
      "grad_norm": 0.059730805456638336,
      "learning_rate": 0.00015520904915944817,
      "loss": 1.4619,
      "step": 16950
    },
    {
      "epoch": 0.753723711706905,
      "grad_norm": 0.059080708771944046,
      "learning_rate": 0.00015510319299963393,
      "loss": 1.4627,
      "step": 16952
    },
    {
      "epoch": 0.7538126361655774,
      "grad_norm": 0.06086655706167221,
      "learning_rate": 0.00015499736632311868,
      "loss": 1.4642,
      "step": 16954
    },
    {
      "epoch": 0.7539015606242497,
      "grad_norm": 0.05988055095076561,
      "learning_rate": 0.0001548915691389491,
      "loss": 1.4542,
      "step": 16956
    },
    {
      "epoch": 0.7539904850829221,
      "grad_norm": 0.05994189530611038,
      "learning_rate": 0.00015478580145616888,
      "loss": 1.4542,
      "step": 16958
    },
    {
      "epoch": 0.7540794095415945,
      "grad_norm": 0.06018702685832977,
      "learning_rate": 0.00015468006328381968,
      "loss": 1.4618,
      "step": 16960
    },
    {
      "epoch": 0.7541683340002667,
      "grad_norm": 0.0580444373190403,
      "learning_rate": 0.00015457435463094054,
      "loss": 1.4637,
      "step": 16962
    },
    {
      "epoch": 0.7542572584589391,
      "grad_norm": 0.06005840376019478,
      "learning_rate": 0.00015446867550656767,
      "loss": 1.4638,
      "step": 16964
    },
    {
      "epoch": 0.7543461829176115,
      "grad_norm": 0.061004381626844406,
      "learning_rate": 0.00015436302591973516,
      "loss": 1.4629,
      "step": 16966
    },
    {
      "epoch": 0.7544351073762838,
      "grad_norm": 0.05958648771047592,
      "learning_rate": 0.00015425740587947435,
      "loss": 1.4598,
      "step": 16968
    },
    {
      "epoch": 0.7545240318349562,
      "grad_norm": 0.059469547122716904,
      "learning_rate": 0.00015415181539481427,
      "loss": 1.4588,
      "step": 16970
    },
    {
      "epoch": 0.7546129562936286,
      "grad_norm": 0.05835345759987831,
      "learning_rate": 0.00015404625447478117,
      "loss": 1.454,
      "step": 16972
    },
    {
      "epoch": 0.7547018807523009,
      "grad_norm": 0.058979202061891556,
      "learning_rate": 0.0001539407231283984,
      "loss": 1.4601,
      "step": 16974
    },
    {
      "epoch": 0.7547908052109733,
      "grad_norm": 0.060271427035331726,
      "learning_rate": 0.00015383522136468803,
      "loss": 1.4585,
      "step": 16976
    },
    {
      "epoch": 0.7548797296696457,
      "grad_norm": 0.05911035090684891,
      "learning_rate": 0.0001537297491926684,
      "loss": 1.4615,
      "step": 16978
    },
    {
      "epoch": 0.754968654128318,
      "grad_norm": 0.059652071446180344,
      "learning_rate": 0.00015362430662135568,
      "loss": 1.4659,
      "step": 16980
    },
    {
      "epoch": 0.7550575785869903,
      "grad_norm": 0.05946535989642143,
      "learning_rate": 0.00015351889365976357,
      "loss": 1.4537,
      "step": 16982
    },
    {
      "epoch": 0.7551465030456627,
      "grad_norm": 0.06029531732201576,
      "learning_rate": 0.00015341351031690354,
      "loss": 1.4597,
      "step": 16984
    },
    {
      "epoch": 0.755235427504335,
      "grad_norm": 0.05937948077917099,
      "learning_rate": 0.00015330815660178382,
      "loss": 1.4593,
      "step": 16986
    },
    {
      "epoch": 0.7553243519630074,
      "grad_norm": 0.05890772491693497,
      "learning_rate": 0.0001532028325234107,
      "loss": 1.4543,
      "step": 16988
    },
    {
      "epoch": 0.7554132764216798,
      "grad_norm": 0.06144456937909126,
      "learning_rate": 0.00015309753809078774,
      "loss": 1.4574,
      "step": 16990
    },
    {
      "epoch": 0.7555022008803521,
      "grad_norm": 0.059851065278053284,
      "learning_rate": 0.00015299227331291615,
      "loss": 1.4601,
      "step": 16992
    },
    {
      "epoch": 0.7555911253390245,
      "grad_norm": 0.05949746072292328,
      "learning_rate": 0.00015288703819879424,
      "loss": 1.4574,
      "step": 16994
    },
    {
      "epoch": 0.7556800497976969,
      "grad_norm": 0.05919702351093292,
      "learning_rate": 0.00015278183275741763,
      "loss": 1.4577,
      "step": 16996
    },
    {
      "epoch": 0.7557689742563692,
      "grad_norm": 0.05892585590481758,
      "learning_rate": 0.0001526766569977805,
      "loss": 1.4593,
      "step": 16998
    },
    {
      "epoch": 0.7558578987150416,
      "grad_norm": 0.06064224988222122,
      "learning_rate": 0.0001525715109288733,
      "loss": 1.4542,
      "step": 17000
    },
    {
      "epoch": 0.7558578987150416,
      "eval_loss": 1.443333387374878,
      "eval_runtime": 12.4431,
      "eval_samples_per_second": 555.329,
      "eval_steps_per_second": 69.436,
      "step": 17000
    },
    {
      "epoch": 0.755946823173714,
      "grad_norm": 0.05818698927760124,
      "learning_rate": 0.0001524663945596842,
      "loss": 1.4525,
      "step": 17002
    },
    {
      "epoch": 0.7560357476323862,
      "grad_norm": 0.05946521461009979,
      "learning_rate": 0.00015236130789919927,
      "loss": 1.4631,
      "step": 17004
    },
    {
      "epoch": 0.7561246720910586,
      "grad_norm": 0.059593699872493744,
      "learning_rate": 0.0001522562509564017,
      "loss": 1.4585,
      "step": 17006
    },
    {
      "epoch": 0.756213596549731,
      "grad_norm": 0.05993930995464325,
      "learning_rate": 0.00015215122374027246,
      "loss": 1.4616,
      "step": 17008
    },
    {
      "epoch": 0.7563025210084033,
      "grad_norm": 0.05795668810606003,
      "learning_rate": 0.0001520462262597893,
      "loss": 1.4631,
      "step": 17010
    },
    {
      "epoch": 0.7563914454670757,
      "grad_norm": 0.05875843018293381,
      "learning_rate": 0.0001519412585239282,
      "loss": 1.451,
      "step": 17012
    },
    {
      "epoch": 0.7564803699257481,
      "grad_norm": 0.057727470993995667,
      "learning_rate": 0.0001518363205416623,
      "loss": 1.4582,
      "step": 17014
    },
    {
      "epoch": 0.7565692943844204,
      "grad_norm": 0.058522436767816544,
      "learning_rate": 0.0001517314123219618,
      "loss": 1.4639,
      "step": 17016
    },
    {
      "epoch": 0.7566582188430928,
      "grad_norm": 0.058987364172935486,
      "learning_rate": 0.000151626533873795,
      "loss": 1.4567,
      "step": 17018
    },
    {
      "epoch": 0.7567471433017652,
      "grad_norm": 0.05821183696389198,
      "learning_rate": 0.00015152168520612746,
      "loss": 1.4583,
      "step": 17020
    },
    {
      "epoch": 0.7568360677604375,
      "grad_norm": 0.06063270568847656,
      "learning_rate": 0.00015141686632792184,
      "loss": 1.4641,
      "step": 17022
    },
    {
      "epoch": 0.7569249922191099,
      "grad_norm": 0.0597229078412056,
      "learning_rate": 0.00015131207724813862,
      "loss": 1.4509,
      "step": 17024
    },
    {
      "epoch": 0.7570139166777822,
      "grad_norm": 0.05786067619919777,
      "learning_rate": 0.0001512073179757358,
      "loss": 1.4589,
      "step": 17026
    },
    {
      "epoch": 0.7571028411364545,
      "grad_norm": 0.062379490584135056,
      "learning_rate": 0.00015110258851966835,
      "loss": 1.457,
      "step": 17028
    },
    {
      "epoch": 0.7571917655951269,
      "grad_norm": 0.058465585112571716,
      "learning_rate": 0.00015099788888888938,
      "loss": 1.4601,
      "step": 17030
    },
    {
      "epoch": 0.7572806900537993,
      "grad_norm": 0.058587927371263504,
      "learning_rate": 0.00015089321909234865,
      "loss": 1.4605,
      "step": 17032
    },
    {
      "epoch": 0.7573696145124716,
      "grad_norm": 0.06187847629189491,
      "learning_rate": 0.00015078857913899407,
      "loss": 1.4613,
      "step": 17034
    },
    {
      "epoch": 0.757458538971144,
      "grad_norm": 0.05908993259072304,
      "learning_rate": 0.00015068396903777082,
      "loss": 1.4529,
      "step": 17036
    },
    {
      "epoch": 0.7575474634298164,
      "grad_norm": 0.05782734975218773,
      "learning_rate": 0.00015057938879762113,
      "loss": 1.4518,
      "step": 17038
    },
    {
      "epoch": 0.7576363878884887,
      "grad_norm": 0.06174134463071823,
      "learning_rate": 0.0001504748384274851,
      "loss": 1.4595,
      "step": 17040
    },
    {
      "epoch": 0.7577253123471611,
      "grad_norm": 0.06178918853402138,
      "learning_rate": 0.00015037031793630035,
      "loss": 1.4626,
      "step": 17042
    },
    {
      "epoch": 0.7578142368058335,
      "grad_norm": 0.05829121544957161,
      "learning_rate": 0.00015026582733300144,
      "loss": 1.4541,
      "step": 17044
    },
    {
      "epoch": 0.7579031612645059,
      "grad_norm": 0.06056341528892517,
      "learning_rate": 0.00015016136662652086,
      "loss": 1.4642,
      "step": 17046
    },
    {
      "epoch": 0.7579920857231781,
      "grad_norm": 0.05929940566420555,
      "learning_rate": 0.0001500569358257885,
      "loss": 1.4596,
      "step": 17048
    },
    {
      "epoch": 0.7580810101818505,
      "grad_norm": 0.05996108427643776,
      "learning_rate": 0.0001499525349397312,
      "loss": 1.4619,
      "step": 17050
    },
    {
      "epoch": 0.7581699346405228,
      "grad_norm": 0.05888962745666504,
      "learning_rate": 0.000149848163977274,
      "loss": 1.4585,
      "step": 17052
    },
    {
      "epoch": 0.7582588590991952,
      "grad_norm": 0.05805379897356033,
      "learning_rate": 0.00014974382294733873,
      "loss": 1.4582,
      "step": 17054
    },
    {
      "epoch": 0.7583477835578676,
      "grad_norm": 0.06010428071022034,
      "learning_rate": 0.00014963951185884494,
      "loss": 1.461,
      "step": 17056
    },
    {
      "epoch": 0.75843670801654,
      "grad_norm": 0.06018170714378357,
      "learning_rate": 0.00014953523072070985,
      "loss": 1.4593,
      "step": 17058
    },
    {
      "epoch": 0.7585256324752123,
      "grad_norm": 0.058989543467760086,
      "learning_rate": 0.00014943097954184752,
      "loss": 1.4541,
      "step": 17060
    },
    {
      "epoch": 0.7586145569338847,
      "grad_norm": 0.060051459819078445,
      "learning_rate": 0.00014932675833116994,
      "loss": 1.4549,
      "step": 17062
    },
    {
      "epoch": 0.758703481392557,
      "grad_norm": 0.06050692871212959,
      "learning_rate": 0.00014922256709758662,
      "loss": 1.4664,
      "step": 17064
    },
    {
      "epoch": 0.7587924058512294,
      "grad_norm": 0.05857968330383301,
      "learning_rate": 0.00014911840585000392,
      "loss": 1.4577,
      "step": 17066
    },
    {
      "epoch": 0.7588813303099018,
      "grad_norm": 0.06070830672979355,
      "learning_rate": 0.00014901427459732613,
      "loss": 1.4657,
      "step": 17068
    },
    {
      "epoch": 0.758970254768574,
      "grad_norm": 0.059025511145591736,
      "learning_rate": 0.00014891017334845515,
      "loss": 1.4577,
      "step": 17070
    },
    {
      "epoch": 0.7590591792272464,
      "grad_norm": 0.060259878635406494,
      "learning_rate": 0.00014880610211228951,
      "loss": 1.4623,
      "step": 17072
    },
    {
      "epoch": 0.7591481036859188,
      "grad_norm": 0.05835847184062004,
      "learning_rate": 0.00014870206089772592,
      "loss": 1.4571,
      "step": 17074
    },
    {
      "epoch": 0.7592370281445912,
      "grad_norm": 0.058684930205345154,
      "learning_rate": 0.00014859804971365832,
      "loss": 1.458,
      "step": 17076
    },
    {
      "epoch": 0.7593259526032635,
      "grad_norm": 0.06079687178134918,
      "learning_rate": 0.00014849406856897818,
      "loss": 1.4562,
      "step": 17078
    },
    {
      "epoch": 0.7594148770619359,
      "grad_norm": 0.059008531272411346,
      "learning_rate": 0.00014839011747257407,
      "loss": 1.4605,
      "step": 17080
    },
    {
      "epoch": 0.7595038015206083,
      "grad_norm": 0.058418843895196915,
      "learning_rate": 0.00014828619643333192,
      "loss": 1.4533,
      "step": 17082
    },
    {
      "epoch": 0.7595927259792806,
      "grad_norm": 0.060562197118997574,
      "learning_rate": 0.00014818230546013596,
      "loss": 1.4563,
      "step": 17084
    },
    {
      "epoch": 0.759681650437953,
      "grad_norm": 0.05741146579384804,
      "learning_rate": 0.00014807844456186692,
      "loss": 1.4577,
      "step": 17086
    },
    {
      "epoch": 0.7597705748966254,
      "grad_norm": 0.058450762182474136,
      "learning_rate": 0.00014797461374740312,
      "loss": 1.4551,
      "step": 17088
    },
    {
      "epoch": 0.7598594993552977,
      "grad_norm": 0.05948064476251602,
      "learning_rate": 0.0001478708130256207,
      "loss": 1.4559,
      "step": 17090
    },
    {
      "epoch": 0.75994842381397,
      "grad_norm": 0.05894921347498894,
      "learning_rate": 0.00014776704240539296,
      "loss": 1.455,
      "step": 17092
    },
    {
      "epoch": 0.7600373482726424,
      "grad_norm": 0.05906720459461212,
      "learning_rate": 0.00014766330189559085,
      "loss": 1.4564,
      "step": 17094
    },
    {
      "epoch": 0.7601262727313147,
      "grad_norm": 0.059796903282403946,
      "learning_rate": 0.00014755959150508223,
      "loss": 1.4582,
      "step": 17096
    },
    {
      "epoch": 0.7602151971899871,
      "grad_norm": 0.060287073254585266,
      "learning_rate": 0.00014745591124273289,
      "loss": 1.4608,
      "step": 17098
    },
    {
      "epoch": 0.7603041216486595,
      "grad_norm": 0.059052977710962296,
      "learning_rate": 0.00014735226111740603,
      "loss": 1.456,
      "step": 17100
    },
    {
      "epoch": 0.7603930461073318,
      "grad_norm": 0.05815219506621361,
      "learning_rate": 0.0001472486411379619,
      "loss": 1.4511,
      "step": 17102
    },
    {
      "epoch": 0.7604819705660042,
      "grad_norm": 0.05960388854146004,
      "learning_rate": 0.00014714505131325817,
      "loss": 1.46,
      "step": 17104
    },
    {
      "epoch": 0.7605708950246766,
      "grad_norm": 0.060582149773836136,
      "learning_rate": 0.00014704149165215076,
      "loss": 1.4646,
      "step": 17106
    },
    {
      "epoch": 0.7606598194833489,
      "grad_norm": 0.058946821838617325,
      "learning_rate": 0.0001469379621634921,
      "loss": 1.4587,
      "step": 17108
    },
    {
      "epoch": 0.7607487439420213,
      "grad_norm": 0.06056873872876167,
      "learning_rate": 0.00014683446285613212,
      "loss": 1.4572,
      "step": 17110
    },
    {
      "epoch": 0.7608376684006936,
      "grad_norm": 0.05924323946237564,
      "learning_rate": 0.00014673099373891863,
      "loss": 1.4545,
      "step": 17112
    },
    {
      "epoch": 0.7609265928593659,
      "grad_norm": 0.05958441272377968,
      "learning_rate": 0.0001466275548206966,
      "loss": 1.4507,
      "step": 17114
    },
    {
      "epoch": 0.7610155173180383,
      "grad_norm": 0.05905461683869362,
      "learning_rate": 0.00014652414611030862,
      "loss": 1.4613,
      "step": 17116
    },
    {
      "epoch": 0.7611044417767107,
      "grad_norm": 0.05967036262154579,
      "learning_rate": 0.00014642076761659417,
      "loss": 1.458,
      "step": 17118
    },
    {
      "epoch": 0.761193366235383,
      "grad_norm": 0.05881161987781525,
      "learning_rate": 0.00014631741934839065,
      "loss": 1.459,
      "step": 17120
    },
    {
      "epoch": 0.7612822906940554,
      "grad_norm": 0.06017891317605972,
      "learning_rate": 0.00014621410131453294,
      "loss": 1.4535,
      "step": 17122
    },
    {
      "epoch": 0.7613712151527278,
      "grad_norm": 0.05896136537194252,
      "learning_rate": 0.00014611081352385296,
      "loss": 1.4567,
      "step": 17124
    },
    {
      "epoch": 0.7614601396114001,
      "grad_norm": 0.05887780338525772,
      "learning_rate": 0.00014600755598517985,
      "loss": 1.4566,
      "step": 17126
    },
    {
      "epoch": 0.7615490640700725,
      "grad_norm": 0.06204426288604736,
      "learning_rate": 0.00014590432870734116,
      "loss": 1.4587,
      "step": 17128
    },
    {
      "epoch": 0.7616379885287449,
      "grad_norm": 0.05948631837964058,
      "learning_rate": 0.0001458011316991607,
      "loss": 1.452,
      "step": 17130
    },
    {
      "epoch": 0.7617269129874172,
      "grad_norm": 0.05867043510079384,
      "learning_rate": 0.00014569796496946064,
      "loss": 1.4582,
      "step": 17132
    },
    {
      "epoch": 0.7618158374460895,
      "grad_norm": 0.0589086189866066,
      "learning_rate": 0.00014559482852705968,
      "loss": 1.4573,
      "step": 17134
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 0.0606837160885334,
      "learning_rate": 0.00014549172238077456,
      "loss": 1.4536,
      "step": 17136
    },
    {
      "epoch": 0.7619936863634342,
      "grad_norm": 0.05865649878978729,
      "learning_rate": 0.00014538864653941953,
      "loss": 1.4604,
      "step": 17138
    },
    {
      "epoch": 0.7620826108221066,
      "grad_norm": 0.058400411158800125,
      "learning_rate": 0.0001452856010118055,
      "loss": 1.4619,
      "step": 17140
    },
    {
      "epoch": 0.762171535280779,
      "grad_norm": 0.05969015881419182,
      "learning_rate": 0.0001451825858067415,
      "loss": 1.4589,
      "step": 17142
    },
    {
      "epoch": 0.7622604597394513,
      "grad_norm": 0.059678588062524796,
      "learning_rate": 0.0001450796009330338,
      "loss": 1.454,
      "step": 17144
    },
    {
      "epoch": 0.7623493841981237,
      "grad_norm": 0.058471839874982834,
      "learning_rate": 0.00014497664639948578,
      "loss": 1.4564,
      "step": 17146
    },
    {
      "epoch": 0.7624383086567961,
      "grad_norm": 0.059258636087179184,
      "learning_rate": 0.0001448737222148986,
      "loss": 1.455,
      "step": 17148
    },
    {
      "epoch": 0.7625272331154684,
      "grad_norm": 0.06029491126537323,
      "learning_rate": 0.00014477082838807072,
      "loss": 1.4602,
      "step": 17150
    },
    {
      "epoch": 0.7626161575741408,
      "grad_norm": 0.05870896950364113,
      "learning_rate": 0.0001446679649277977,
      "loss": 1.4544,
      "step": 17152
    },
    {
      "epoch": 0.7627050820328132,
      "grad_norm": 0.05867839604616165,
      "learning_rate": 0.000144565131842873,
      "loss": 1.452,
      "step": 17154
    },
    {
      "epoch": 0.7627940064914854,
      "grad_norm": 0.058226361870765686,
      "learning_rate": 0.00014446232914208735,
      "loss": 1.4586,
      "step": 17156
    },
    {
      "epoch": 0.7628829309501578,
      "grad_norm": 0.05763303488492966,
      "learning_rate": 0.0001443595568342284,
      "loss": 1.4539,
      "step": 17158
    },
    {
      "epoch": 0.7629718554088302,
      "grad_norm": 0.05882842093706131,
      "learning_rate": 0.00014425681492808196,
      "loss": 1.4542,
      "step": 17160
    },
    {
      "epoch": 0.7630607798675025,
      "grad_norm": 0.05935533344745636,
      "learning_rate": 0.00014415410343243036,
      "loss": 1.4603,
      "step": 17162
    },
    {
      "epoch": 0.7631497043261749,
      "grad_norm": 0.060341961681842804,
      "learning_rate": 0.0001440514223560545,
      "loss": 1.4578,
      "step": 17164
    },
    {
      "epoch": 0.7632386287848473,
      "grad_norm": 0.05959847569465637,
      "learning_rate": 0.00014394877170773168,
      "loss": 1.4586,
      "step": 17166
    },
    {
      "epoch": 0.7633275532435196,
      "grad_norm": 0.058998953551054,
      "learning_rate": 0.00014384615149623674,
      "loss": 1.4553,
      "step": 17168
    },
    {
      "epoch": 0.763416477702192,
      "grad_norm": 0.058729879558086395,
      "learning_rate": 0.0001437435617303423,
      "loss": 1.4607,
      "step": 17170
    },
    {
      "epoch": 0.7635054021608644,
      "grad_norm": 0.05923215672373772,
      "learning_rate": 0.00014364100241881834,
      "loss": 1.4544,
      "step": 17172
    },
    {
      "epoch": 0.7635943266195367,
      "grad_norm": 0.05920840799808502,
      "learning_rate": 0.00014353847357043175,
      "loss": 1.4641,
      "step": 17174
    },
    {
      "epoch": 0.7636832510782091,
      "grad_norm": 0.058832135051488876,
      "learning_rate": 0.0001434359751939473,
      "loss": 1.454,
      "step": 17176
    },
    {
      "epoch": 0.7637721755368814,
      "grad_norm": 0.05877342447638512,
      "learning_rate": 0.00014333350729812722,
      "loss": 1.4617,
      "step": 17178
    },
    {
      "epoch": 0.7638610999955537,
      "grad_norm": 0.05918002128601074,
      "learning_rate": 0.00014323106989173052,
      "loss": 1.4572,
      "step": 17180
    },
    {
      "epoch": 0.7639500244542261,
      "grad_norm": 0.058956753462553024,
      "learning_rate": 0.00014312866298351435,
      "loss": 1.4561,
      "step": 17182
    },
    {
      "epoch": 0.7640389489128985,
      "grad_norm": 0.05958571657538414,
      "learning_rate": 0.00014302628658223243,
      "loss": 1.4537,
      "step": 17184
    },
    {
      "epoch": 0.7641278733715708,
      "grad_norm": 0.058351002633571625,
      "learning_rate": 0.000142923940696637,
      "loss": 1.4526,
      "step": 17186
    },
    {
      "epoch": 0.7642167978302432,
      "grad_norm": 0.0581650584936142,
      "learning_rate": 0.00014282162533547666,
      "loss": 1.4606,
      "step": 17188
    },
    {
      "epoch": 0.7643057222889156,
      "grad_norm": 0.0594148151576519,
      "learning_rate": 0.00014271934050749774,
      "loss": 1.4577,
      "step": 17190
    },
    {
      "epoch": 0.764394646747588,
      "grad_norm": 0.058780916035175323,
      "learning_rate": 0.00014261708622144404,
      "loss": 1.4568,
      "step": 17192
    },
    {
      "epoch": 0.7644835712062603,
      "grad_norm": 0.06111384555697441,
      "learning_rate": 0.00014251486248605688,
      "loss": 1.4585,
      "step": 17194
    },
    {
      "epoch": 0.7645724956649327,
      "grad_norm": 0.060085564851760864,
      "learning_rate": 0.0001424126693100745,
      "loss": 1.4593,
      "step": 17196
    },
    {
      "epoch": 0.764661420123605,
      "grad_norm": 0.05976768955588341,
      "learning_rate": 0.000142310506702233,
      "loss": 1.4553,
      "step": 17198
    },
    {
      "epoch": 0.7647503445822773,
      "grad_norm": 0.057454586029052734,
      "learning_rate": 0.0001422083746712657,
      "loss": 1.4578,
      "step": 17200
    },
    {
      "epoch": 0.7648392690409497,
      "grad_norm": 0.05781802162528038,
      "learning_rate": 0.0001421062732259034,
      "loss": 1.4557,
      "step": 17202
    },
    {
      "epoch": 0.764928193499622,
      "grad_norm": 0.058845892548561096,
      "learning_rate": 0.00014200420237487388,
      "loss": 1.4594,
      "step": 17204
    },
    {
      "epoch": 0.7650171179582944,
      "grad_norm": 0.05837889760732651,
      "learning_rate": 0.00014190216212690282,
      "loss": 1.4635,
      "step": 17206
    },
    {
      "epoch": 0.7651060424169668,
      "grad_norm": 0.057773832231760025,
      "learning_rate": 0.00014180015249071315,
      "loss": 1.4523,
      "step": 17208
    },
    {
      "epoch": 0.7651949668756391,
      "grad_norm": 0.05930472910404205,
      "learning_rate": 0.000141698173475025,
      "loss": 1.4551,
      "step": 17210
    },
    {
      "epoch": 0.7652838913343115,
      "grad_norm": 0.058624934405088425,
      "learning_rate": 0.00014159622508855564,
      "loss": 1.4633,
      "step": 17212
    },
    {
      "epoch": 0.7653728157929839,
      "grad_norm": 0.0588277168571949,
      "learning_rate": 0.0001414943073400207,
      "loss": 1.4488,
      "step": 17214
    },
    {
      "epoch": 0.7654617402516563,
      "grad_norm": 0.06121628358960152,
      "learning_rate": 0.00014139242023813214,
      "loss": 1.4541,
      "step": 17216
    },
    {
      "epoch": 0.7655506647103286,
      "grad_norm": 0.06059684976935387,
      "learning_rate": 0.00014129056379159994,
      "loss": 1.458,
      "step": 17218
    },
    {
      "epoch": 0.765639589169001,
      "grad_norm": 0.059569355100393295,
      "learning_rate": 0.00014118873800913096,
      "loss": 1.4564,
      "step": 17220
    },
    {
      "epoch": 0.7657285136276732,
      "grad_norm": 0.058082085102796555,
      "learning_rate": 0.0001410869428994299,
      "loss": 1.4604,
      "step": 17222
    },
    {
      "epoch": 0.7658174380863456,
      "grad_norm": 0.05903531610965729,
      "learning_rate": 0.00014098517847119886,
      "loss": 1.4629,
      "step": 17224
    },
    {
      "epoch": 0.765906362545018,
      "grad_norm": 0.057636991143226624,
      "learning_rate": 0.00014088344473313664,
      "loss": 1.4602,
      "step": 17226
    },
    {
      "epoch": 0.7659952870036904,
      "grad_norm": 0.05922068655490875,
      "learning_rate": 0.0001407817416939402,
      "loss": 1.4566,
      "step": 17228
    },
    {
      "epoch": 0.7660842114623627,
      "grad_norm": 0.05783383175730705,
      "learning_rate": 0.00014068006936230365,
      "loss": 1.4545,
      "step": 17230
    },
    {
      "epoch": 0.7661731359210351,
      "grad_norm": 0.05895695090293884,
      "learning_rate": 0.0001405784277469182,
      "loss": 1.4631,
      "step": 17232
    },
    {
      "epoch": 0.7662620603797075,
      "grad_norm": 0.05768261104822159,
      "learning_rate": 0.0001404768168564724,
      "loss": 1.4533,
      "step": 17234
    },
    {
      "epoch": 0.7663509848383798,
      "grad_norm": 0.0601133368909359,
      "learning_rate": 0.00014037523669965297,
      "loss": 1.4535,
      "step": 17236
    },
    {
      "epoch": 0.7664399092970522,
      "grad_norm": 0.058861732482910156,
      "learning_rate": 0.00014027368728514294,
      "loss": 1.4581,
      "step": 17238
    },
    {
      "epoch": 0.7665288337557246,
      "grad_norm": 0.05894797295331955,
      "learning_rate": 0.00014017216862162357,
      "loss": 1.4537,
      "step": 17240
    },
    {
      "epoch": 0.7666177582143968,
      "grad_norm": 0.05880829319357872,
      "learning_rate": 0.00014007068071777274,
      "loss": 1.4577,
      "step": 17242
    },
    {
      "epoch": 0.7667066826730692,
      "grad_norm": 0.06195038929581642,
      "learning_rate": 0.00013996922358226632,
      "loss": 1.459,
      "step": 17244
    },
    {
      "epoch": 0.7667956071317416,
      "grad_norm": 0.058604203164577484,
      "learning_rate": 0.0001398677972237774,
      "loss": 1.4592,
      "step": 17246
    },
    {
      "epoch": 0.7668845315904139,
      "grad_norm": 0.059475913643836975,
      "learning_rate": 0.00013976640165097609,
      "loss": 1.4559,
      "step": 17248
    },
    {
      "epoch": 0.7669734560490863,
      "grad_norm": 0.058169715106487274,
      "learning_rate": 0.00013966503687253023,
      "loss": 1.4575,
      "step": 17250
    },
    {
      "epoch": 0.7670623805077587,
      "grad_norm": 0.06019798666238785,
      "learning_rate": 0.00013956370289710518,
      "loss": 1.4533,
      "step": 17252
    },
    {
      "epoch": 0.767151304966431,
      "grad_norm": 0.05990200489759445,
      "learning_rate": 0.000139462399733363,
      "loss": 1.4578,
      "step": 17254
    },
    {
      "epoch": 0.7672402294251034,
      "grad_norm": 0.061555687338113785,
      "learning_rate": 0.00013936112738996377,
      "loss": 1.4562,
      "step": 17256
    },
    {
      "epoch": 0.7673291538837758,
      "grad_norm": 0.05985216051340103,
      "learning_rate": 0.00013925988587556477,
      "loss": 1.4601,
      "step": 17258
    },
    {
      "epoch": 0.7674180783424481,
      "grad_norm": 0.058907974511384964,
      "learning_rate": 0.00013915867519882026,
      "loss": 1.4597,
      "step": 17260
    },
    {
      "epoch": 0.7675070028011205,
      "grad_norm": 0.059296950697898865,
      "learning_rate": 0.0001390574953683824,
      "loss": 1.4566,
      "step": 17262
    },
    {
      "epoch": 0.7675959272597928,
      "grad_norm": 0.058819953352212906,
      "learning_rate": 0.00013895634639290067,
      "loss": 1.4568,
      "step": 17264
    },
    {
      "epoch": 0.7676848517184651,
      "grad_norm": 0.05849660560488701,
      "learning_rate": 0.00013885522828102132,
      "loss": 1.4597,
      "step": 17266
    },
    {
      "epoch": 0.7677737761771375,
      "grad_norm": 0.05969657003879547,
      "learning_rate": 0.00013875414104138874,
      "loss": 1.4551,
      "step": 17268
    },
    {
      "epoch": 0.7678627006358099,
      "grad_norm": 0.059172701090574265,
      "learning_rate": 0.00013865308468264382,
      "loss": 1.4562,
      "step": 17270
    },
    {
      "epoch": 0.7679516250944822,
      "grad_norm": 0.059122659265995026,
      "learning_rate": 0.0001385520592134259,
      "loss": 1.4559,
      "step": 17272
    },
    {
      "epoch": 0.7680405495531546,
      "grad_norm": 0.05972140282392502,
      "learning_rate": 0.00013845106464237085,
      "loss": 1.4559,
      "step": 17274
    },
    {
      "epoch": 0.768129474011827,
      "grad_norm": 0.058670710772275925,
      "learning_rate": 0.00013835010097811191,
      "loss": 1.4532,
      "step": 17276
    },
    {
      "epoch": 0.7682183984704993,
      "grad_norm": 0.05994569510221481,
      "learning_rate": 0.00013824916822928013,
      "loss": 1.4585,
      "step": 17278
    },
    {
      "epoch": 0.7683073229291717,
      "grad_norm": 0.060833681374788284,
      "learning_rate": 0.00013814826640450373,
      "loss": 1.4545,
      "step": 17280
    },
    {
      "epoch": 0.7683962473878441,
      "grad_norm": 0.058985013514757156,
      "learning_rate": 0.00013804739551240804,
      "loss": 1.4596,
      "step": 17282
    },
    {
      "epoch": 0.7684851718465164,
      "grad_norm": 0.05949314683675766,
      "learning_rate": 0.000137946555561616,
      "loss": 1.4517,
      "step": 17284
    },
    {
      "epoch": 0.7685740963051887,
      "grad_norm": 0.0588274747133255,
      "learning_rate": 0.00013784574656074795,
      "loss": 1.4546,
      "step": 17286
    },
    {
      "epoch": 0.7686630207638611,
      "grad_norm": 0.06145096197724342,
      "learning_rate": 0.00013774496851842155,
      "loss": 1.4592,
      "step": 17288
    },
    {
      "epoch": 0.7687519452225334,
      "grad_norm": 0.058883629739284515,
      "learning_rate": 0.00013764422144325173,
      "loss": 1.4589,
      "step": 17290
    },
    {
      "epoch": 0.7688408696812058,
      "grad_norm": 0.06038152053952217,
      "learning_rate": 0.00013754350534385035,
      "loss": 1.4592,
      "step": 17292
    },
    {
      "epoch": 0.7689297941398782,
      "grad_norm": 0.05828770250082016,
      "learning_rate": 0.00013744282022882775,
      "loss": 1.456,
      "step": 17294
    },
    {
      "epoch": 0.7690187185985505,
      "grad_norm": 0.06063667684793472,
      "learning_rate": 0.00013734216610679055,
      "loss": 1.4467,
      "step": 17296
    },
    {
      "epoch": 0.7691076430572229,
      "grad_norm": 0.05999766290187836,
      "learning_rate": 0.0001372415429863431,
      "loss": 1.4586,
      "step": 17298
    },
    {
      "epoch": 0.7691965675158953,
      "grad_norm": 0.05993552505970001,
      "learning_rate": 0.00013714095087608718,
      "loss": 1.4532,
      "step": 17300
    },
    {
      "epoch": 0.7692854919745676,
      "grad_norm": 0.058838147670030594,
      "learning_rate": 0.00013704038978462195,
      "loss": 1.4539,
      "step": 17302
    },
    {
      "epoch": 0.76937441643324,
      "grad_norm": 0.059014104306697845,
      "learning_rate": 0.00013693985972054363,
      "loss": 1.4509,
      "step": 17304
    },
    {
      "epoch": 0.7694633408919124,
      "grad_norm": 0.05862824246287346,
      "learning_rate": 0.00013683936069244603,
      "loss": 1.4539,
      "step": 17306
    },
    {
      "epoch": 0.7695522653505846,
      "grad_norm": 0.059112753719091415,
      "learning_rate": 0.0001367388927089202,
      "loss": 1.4498,
      "step": 17308
    },
    {
      "epoch": 0.769641189809257,
      "grad_norm": 0.05892875790596008,
      "learning_rate": 0.0001366384557785549,
      "loss": 1.461,
      "step": 17310
    },
    {
      "epoch": 0.7697301142679294,
      "grad_norm": 0.061551161110401154,
      "learning_rate": 0.00013653804990993552,
      "loss": 1.4524,
      "step": 17312
    },
    {
      "epoch": 0.7698190387266017,
      "grad_norm": 0.05997699499130249,
      "learning_rate": 0.0001364376751116453,
      "loss": 1.456,
      "step": 17314
    },
    {
      "epoch": 0.7699079631852741,
      "grad_norm": 0.05965949594974518,
      "learning_rate": 0.00013633733139226494,
      "loss": 1.4535,
      "step": 17316
    },
    {
      "epoch": 0.7699968876439465,
      "grad_norm": 0.059972114861011505,
      "learning_rate": 0.00013623701876037215,
      "loss": 1.4613,
      "step": 17318
    },
    {
      "epoch": 0.7700858121026188,
      "grad_norm": 0.06004555895924568,
      "learning_rate": 0.00013613673722454172,
      "loss": 1.4589,
      "step": 17320
    },
    {
      "epoch": 0.7701747365612912,
      "grad_norm": 0.05914655700325966,
      "learning_rate": 0.00013603648679334679,
      "loss": 1.4538,
      "step": 17322
    },
    {
      "epoch": 0.7702636610199636,
      "grad_norm": 0.058920904994010925,
      "learning_rate": 0.00013593626747535674,
      "loss": 1.4505,
      "step": 17324
    },
    {
      "epoch": 0.7703525854786359,
      "grad_norm": 0.05806851387023926,
      "learning_rate": 0.00013583607927913915,
      "loss": 1.4613,
      "step": 17326
    },
    {
      "epoch": 0.7704415099373083,
      "grad_norm": 0.06060624495148659,
      "learning_rate": 0.0001357359222132581,
      "loss": 1.4561,
      "step": 17328
    },
    {
      "epoch": 0.7705304343959806,
      "grad_norm": 0.060192741453647614,
      "learning_rate": 0.0001356357962862757,
      "loss": 1.4562,
      "step": 17330
    },
    {
      "epoch": 0.7706193588546529,
      "grad_norm": 0.05914744362235069,
      "learning_rate": 0.00013553570150675126,
      "loss": 1.4527,
      "step": 17332
    },
    {
      "epoch": 0.7707082833133253,
      "grad_norm": 0.05795003101229668,
      "learning_rate": 0.00013543563788324104,
      "loss": 1.4546,
      "step": 17334
    },
    {
      "epoch": 0.7707972077719977,
      "grad_norm": 0.05958685278892517,
      "learning_rate": 0.00013533560542429912,
      "loss": 1.4505,
      "step": 17336
    },
    {
      "epoch": 0.77088613223067,
      "grad_norm": 0.05854284018278122,
      "learning_rate": 0.0001352356041384768,
      "loss": 1.4587,
      "step": 17338
    },
    {
      "epoch": 0.7709750566893424,
      "grad_norm": 0.05816273018717766,
      "learning_rate": 0.00013513563403432232,
      "loss": 1.4561,
      "step": 17340
    },
    {
      "epoch": 0.7710639811480148,
      "grad_norm": 0.060084130614995956,
      "learning_rate": 0.00013503569512038184,
      "loss": 1.4549,
      "step": 17342
    },
    {
      "epoch": 0.7711529056066871,
      "grad_norm": 0.06050108000636101,
      "learning_rate": 0.00013493578740519856,
      "loss": 1.4592,
      "step": 17344
    },
    {
      "epoch": 0.7712418300653595,
      "grad_norm": 0.05803850293159485,
      "learning_rate": 0.00013483591089731285,
      "loss": 1.4543,
      "step": 17346
    },
    {
      "epoch": 0.7713307545240319,
      "grad_norm": 0.05967843905091286,
      "learning_rate": 0.0001347360656052628,
      "loss": 1.454,
      "step": 17348
    },
    {
      "epoch": 0.7714196789827041,
      "grad_norm": 0.05886238068342209,
      "learning_rate": 0.00013463625153758342,
      "loss": 1.4569,
      "step": 17350
    },
    {
      "epoch": 0.7715086034413765,
      "grad_norm": 0.0592324323952198,
      "learning_rate": 0.00013453646870280735,
      "loss": 1.4549,
      "step": 17352
    },
    {
      "epoch": 0.7715975279000489,
      "grad_norm": 0.059191394597291946,
      "learning_rate": 0.0001344367171094646,
      "loss": 1.4547,
      "step": 17354
    },
    {
      "epoch": 0.7716864523587212,
      "grad_norm": 0.05849238485097885,
      "learning_rate": 0.00013433699676608214,
      "loss": 1.4564,
      "step": 17356
    },
    {
      "epoch": 0.7717753768173936,
      "grad_norm": 0.058643490076065063,
      "learning_rate": 0.00013423730768118464,
      "loss": 1.4555,
      "step": 17358
    },
    {
      "epoch": 0.771864301276066,
      "grad_norm": 0.05843536928296089,
      "learning_rate": 0.00013413764986329408,
      "loss": 1.4584,
      "step": 17360
    },
    {
      "epoch": 0.7719532257347383,
      "grad_norm": 0.05824931710958481,
      "learning_rate": 0.00013403802332092934,
      "loss": 1.4561,
      "step": 17362
    },
    {
      "epoch": 0.7720421501934107,
      "grad_norm": 0.059431836009025574,
      "learning_rate": 0.00013393842806260715,
      "loss": 1.4552,
      "step": 17364
    },
    {
      "epoch": 0.7721310746520831,
      "grad_norm": 0.06101321056485176,
      "learning_rate": 0.00013383886409684142,
      "loss": 1.4558,
      "step": 17366
    },
    {
      "epoch": 0.7722199991107555,
      "grad_norm": 0.05848054587841034,
      "learning_rate": 0.00013373933143214306,
      "loss": 1.4496,
      "step": 17368
    },
    {
      "epoch": 0.7723089235694278,
      "grad_norm": 0.05772422254085541,
      "learning_rate": 0.00013363983007702075,
      "loss": 1.4613,
      "step": 17370
    },
    {
      "epoch": 0.7723978480281001,
      "grad_norm": 0.05888652801513672,
      "learning_rate": 0.00013354036003998033,
      "loss": 1.4597,
      "step": 17372
    },
    {
      "epoch": 0.7724867724867724,
      "grad_norm": 0.058828819543123245,
      "learning_rate": 0.0001334409213295248,
      "loss": 1.4565,
      "step": 17374
    },
    {
      "epoch": 0.7725756969454448,
      "grad_norm": 0.060525476932525635,
      "learning_rate": 0.00013334151395415477,
      "loss": 1.4589,
      "step": 17376
    },
    {
      "epoch": 0.7726646214041172,
      "grad_norm": 0.06004347652196884,
      "learning_rate": 0.00013324213792236772,
      "loss": 1.4552,
      "step": 17378
    },
    {
      "epoch": 0.7727535458627895,
      "grad_norm": 0.05766339972615242,
      "learning_rate": 0.00013314279324265921,
      "loss": 1.454,
      "step": 17380
    },
    {
      "epoch": 0.7728424703214619,
      "grad_norm": 0.05979818478226662,
      "learning_rate": 0.00013304347992352145,
      "loss": 1.4531,
      "step": 17382
    },
    {
      "epoch": 0.7729313947801343,
      "grad_norm": 0.05923622101545334,
      "learning_rate": 0.00013294419797344397,
      "loss": 1.4577,
      "step": 17384
    },
    {
      "epoch": 0.7730203192388067,
      "grad_norm": 0.059375032782554626,
      "learning_rate": 0.00013284494740091403,
      "loss": 1.458,
      "step": 17386
    },
    {
      "epoch": 0.773109243697479,
      "grad_norm": 0.05873361974954605,
      "learning_rate": 0.0001327457282144161,
      "loss": 1.4583,
      "step": 17388
    },
    {
      "epoch": 0.7731981681561514,
      "grad_norm": 0.05951083451509476,
      "learning_rate": 0.00013264654042243168,
      "loss": 1.4531,
      "step": 17390
    },
    {
      "epoch": 0.7732870926148238,
      "grad_norm": 0.059443507343530655,
      "learning_rate": 0.0001325473840334398,
      "loss": 1.4564,
      "step": 17392
    },
    {
      "epoch": 0.773376017073496,
      "grad_norm": 0.05906612053513527,
      "learning_rate": 0.0001324482590559169,
      "loss": 1.4523,
      "step": 17394
    },
    {
      "epoch": 0.7734649415321684,
      "grad_norm": 0.05907057970762253,
      "learning_rate": 0.00013234916549833664,
      "loss": 1.4547,
      "step": 17396
    },
    {
      "epoch": 0.7735538659908407,
      "grad_norm": 0.058966733515262604,
      "learning_rate": 0.00013225010336916998,
      "loss": 1.4513,
      "step": 17398
    },
    {
      "epoch": 0.7736427904495131,
      "grad_norm": 0.06092256307601929,
      "learning_rate": 0.00013215107267688475,
      "loss": 1.4487,
      "step": 17400
    },
    {
      "epoch": 0.7737317149081855,
      "grad_norm": 0.060688234865665436,
      "learning_rate": 0.0001320520734299472,
      "loss": 1.4537,
      "step": 17402
    },
    {
      "epoch": 0.7738206393668579,
      "grad_norm": 0.0587213858962059,
      "learning_rate": 0.0001319531056368199,
      "loss": 1.4579,
      "step": 17404
    },
    {
      "epoch": 0.7739095638255302,
      "grad_norm": 0.05942368134856224,
      "learning_rate": 0.0001318541693059629,
      "loss": 1.4543,
      "step": 17406
    },
    {
      "epoch": 0.7739984882842026,
      "grad_norm": 0.06065099313855171,
      "learning_rate": 0.0001317552644458339,
      "loss": 1.4541,
      "step": 17408
    },
    {
      "epoch": 0.774087412742875,
      "grad_norm": 0.05897912383079529,
      "learning_rate": 0.00013165639106488768,
      "loss": 1.4523,
      "step": 17410
    },
    {
      "epoch": 0.7741763372015473,
      "grad_norm": 0.05976055935025215,
      "learning_rate": 0.00013155754917157654,
      "loss": 1.4512,
      "step": 17412
    },
    {
      "epoch": 0.7742652616602197,
      "grad_norm": 0.05808144807815552,
      "learning_rate": 0.00013145873877434965,
      "loss": 1.454,
      "step": 17414
    },
    {
      "epoch": 0.774354186118892,
      "grad_norm": 0.06036299839615822,
      "learning_rate": 0.00013135995988165384,
      "loss": 1.4538,
      "step": 17416
    },
    {
      "epoch": 0.7744431105775643,
      "grad_norm": 0.058639515191316605,
      "learning_rate": 0.00013126121250193334,
      "loss": 1.4528,
      "step": 17418
    },
    {
      "epoch": 0.7745320350362367,
      "grad_norm": 0.05818388611078262,
      "learning_rate": 0.00013116249664362945,
      "loss": 1.4578,
      "step": 17420
    },
    {
      "epoch": 0.7746209594949091,
      "grad_norm": 0.05927572399377823,
      "learning_rate": 0.00013106381231518037,
      "loss": 1.4547,
      "step": 17422
    },
    {
      "epoch": 0.7747098839535814,
      "grad_norm": 0.05805305764079094,
      "learning_rate": 0.0001309651595250228,
      "loss": 1.4548,
      "step": 17424
    },
    {
      "epoch": 0.7747988084122538,
      "grad_norm": 0.05744117870926857,
      "learning_rate": 0.00013086653828158973,
      "loss": 1.4596,
      "step": 17426
    },
    {
      "epoch": 0.7748877328709262,
      "grad_norm": 0.05873503535985947,
      "learning_rate": 0.0001307679485933115,
      "loss": 1.455,
      "step": 17428
    },
    {
      "epoch": 0.7749766573295985,
      "grad_norm": 0.05853577330708504,
      "learning_rate": 0.00013066939046861613,
      "loss": 1.4544,
      "step": 17430
    },
    {
      "epoch": 0.7750655817882709,
      "grad_norm": 0.05812995135784149,
      "learning_rate": 0.00013057086391592889,
      "loss": 1.4523,
      "step": 17432
    },
    {
      "epoch": 0.7751545062469433,
      "grad_norm": 0.05899371579289436,
      "learning_rate": 0.00013047236894367237,
      "loss": 1.4546,
      "step": 17434
    },
    {
      "epoch": 0.7752434307056156,
      "grad_norm": 0.05924930423498154,
      "learning_rate": 0.00013037390556026608,
      "loss": 1.4557,
      "step": 17436
    },
    {
      "epoch": 0.7753323551642879,
      "grad_norm": 0.059978023171424866,
      "learning_rate": 0.0001302754737741272,
      "loss": 1.4537,
      "step": 17438
    },
    {
      "epoch": 0.7754212796229603,
      "grad_norm": 0.05814290791749954,
      "learning_rate": 0.00013017707359367026,
      "loss": 1.4523,
      "step": 17440
    },
    {
      "epoch": 0.7755102040816326,
      "grad_norm": 0.05937810242176056,
      "learning_rate": 0.00013007870502730674,
      "loss": 1.451,
      "step": 17442
    },
    {
      "epoch": 0.775599128540305,
      "grad_norm": 0.058999329805374146,
      "learning_rate": 0.00012998036808344572,
      "loss": 1.4548,
      "step": 17444
    },
    {
      "epoch": 0.7756880529989774,
      "grad_norm": 0.05975823104381561,
      "learning_rate": 0.00012988206277049353,
      "loss": 1.4563,
      "step": 17446
    },
    {
      "epoch": 0.7757769774576497,
      "grad_norm": 0.058464065194129944,
      "learning_rate": 0.00012978378909685357,
      "loss": 1.448,
      "step": 17448
    },
    {
      "epoch": 0.7758659019163221,
      "grad_norm": 0.06097313389182091,
      "learning_rate": 0.00012968554707092682,
      "loss": 1.4552,
      "step": 17450
    },
    {
      "epoch": 0.7759548263749945,
      "grad_norm": 0.060069650411605835,
      "learning_rate": 0.00012958733670111156,
      "loss": 1.4555,
      "step": 17452
    },
    {
      "epoch": 0.7760437508336668,
      "grad_norm": 0.05999191105365753,
      "learning_rate": 0.00012948915799580296,
      "loss": 1.456,
      "step": 17454
    },
    {
      "epoch": 0.7761326752923392,
      "grad_norm": 0.059085212647914886,
      "learning_rate": 0.00012939101096339406,
      "loss": 1.4568,
      "step": 17456
    },
    {
      "epoch": 0.7762215997510116,
      "grad_norm": 0.06041676923632622,
      "learning_rate": 0.00012929289561227465,
      "loss": 1.4599,
      "step": 17458
    },
    {
      "epoch": 0.7763105242096838,
      "grad_norm": 0.059452079236507416,
      "learning_rate": 0.00012919481195083215,
      "loss": 1.4529,
      "step": 17460
    },
    {
      "epoch": 0.7763994486683562,
      "grad_norm": 0.05959062650799751,
      "learning_rate": 0.0001290967599874514,
      "loss": 1.4565,
      "step": 17462
    },
    {
      "epoch": 0.7764883731270286,
      "grad_norm": 0.059135422110557556,
      "learning_rate": 0.0001289987397305139,
      "loss": 1.4584,
      "step": 17464
    },
    {
      "epoch": 0.7765772975857009,
      "grad_norm": 0.060259103775024414,
      "learning_rate": 0.00012890075118839918,
      "loss": 1.451,
      "step": 17466
    },
    {
      "epoch": 0.7766662220443733,
      "grad_norm": 0.05845576897263527,
      "learning_rate": 0.00012880279436948368,
      "loss": 1.4531,
      "step": 17468
    },
    {
      "epoch": 0.7767551465030457,
      "grad_norm": 0.05925283208489418,
      "learning_rate": 0.00012870486928214103,
      "loss": 1.4575,
      "step": 17470
    },
    {
      "epoch": 0.776844070961718,
      "grad_norm": 0.060372382402420044,
      "learning_rate": 0.0001286069759347424,
      "loss": 1.4531,
      "step": 17472
    },
    {
      "epoch": 0.7769329954203904,
      "grad_norm": 0.059411074966192245,
      "learning_rate": 0.00012850911433565627,
      "loss": 1.4571,
      "step": 17474
    },
    {
      "epoch": 0.7770219198790628,
      "grad_norm": 0.05941460654139519,
      "learning_rate": 0.000128411284493248,
      "loss": 1.4575,
      "step": 17476
    },
    {
      "epoch": 0.7771108443377351,
      "grad_norm": 0.05880104377865791,
      "learning_rate": 0.00012831348641588087,
      "loss": 1.4566,
      "step": 17478
    },
    {
      "epoch": 0.7771997687964074,
      "grad_norm": 0.05832483246922493,
      "learning_rate": 0.0001282157201119145,
      "loss": 1.4555,
      "step": 17480
    },
    {
      "epoch": 0.7772886932550798,
      "grad_norm": 0.059995248913764954,
      "learning_rate": 0.00012811798558970717,
      "loss": 1.4586,
      "step": 17482
    },
    {
      "epoch": 0.7773776177137521,
      "grad_norm": 0.0595800019800663,
      "learning_rate": 0.00012802028285761318,
      "loss": 1.4615,
      "step": 17484
    },
    {
      "epoch": 0.7774665421724245,
      "grad_norm": 0.06008690595626831,
      "learning_rate": 0.0001279226119239846,
      "loss": 1.4572,
      "step": 17486
    },
    {
      "epoch": 0.7775554666310969,
      "grad_norm": 0.058767497539520264,
      "learning_rate": 0.00012782497279717082,
      "loss": 1.4608,
      "step": 17488
    },
    {
      "epoch": 0.7776443910897692,
      "grad_norm": 0.058635737746953964,
      "learning_rate": 0.00012772736548551862,
      "loss": 1.4545,
      "step": 17490
    },
    {
      "epoch": 0.7777333155484416,
      "grad_norm": 0.059888698160648346,
      "learning_rate": 0.00012762978999737162,
      "loss": 1.451,
      "step": 17492
    },
    {
      "epoch": 0.777822240007114,
      "grad_norm": 0.060602057725191116,
      "learning_rate": 0.00012753224634107118,
      "loss": 1.4643,
      "step": 17494
    },
    {
      "epoch": 0.7779111644657863,
      "grad_norm": 0.0588313527405262,
      "learning_rate": 0.00012743473452495586,
      "loss": 1.4517,
      "step": 17496
    },
    {
      "epoch": 0.7780000889244587,
      "grad_norm": 0.05848833918571472,
      "learning_rate": 0.0001273372545573611,
      "loss": 1.4569,
      "step": 17498
    },
    {
      "epoch": 0.7780890133831311,
      "grad_norm": 0.05902280658483505,
      "learning_rate": 0.00012723980644662014,
      "loss": 1.4577,
      "step": 17500
    },
    {
      "epoch": 0.7780890133831311,
      "eval_loss": 1.441103219985962,
      "eval_runtime": 12.4299,
      "eval_samples_per_second": 555.918,
      "eval_steps_per_second": 69.51,
      "step": 17500
    },
    {
      "epoch": 0.7781779378418033,
      "grad_norm": 0.05878506973385811,
      "learning_rate": 0.00012714239020106323,
      "loss": 1.446,
      "step": 17502
    },
    {
      "epoch": 0.7782668623004757,
      "grad_norm": 0.05805620923638344,
      "learning_rate": 0.00012704500582901807,
      "loss": 1.448,
      "step": 17504
    },
    {
      "epoch": 0.7783557867591481,
      "grad_norm": 0.06127835810184479,
      "learning_rate": 0.00012694765333880935,
      "loss": 1.4584,
      "step": 17506
    },
    {
      "epoch": 0.7784447112178204,
      "grad_norm": 0.059799838811159134,
      "learning_rate": 0.00012685033273875896,
      "loss": 1.455,
      "step": 17508
    },
    {
      "epoch": 0.7785336356764928,
      "grad_norm": 0.0602266825735569,
      "learning_rate": 0.0001267530440371868,
      "loss": 1.4574,
      "step": 17510
    },
    {
      "epoch": 0.7786225601351652,
      "grad_norm": 0.05929147079586983,
      "learning_rate": 0.00012665578724240927,
      "loss": 1.4532,
      "step": 17512
    },
    {
      "epoch": 0.7787114845938375,
      "grad_norm": 0.059397101402282715,
      "learning_rate": 0.00012655856236274023,
      "loss": 1.4546,
      "step": 17514
    },
    {
      "epoch": 0.7788004090525099,
      "grad_norm": 0.059210579842329025,
      "learning_rate": 0.00012646136940649093,
      "loss": 1.4494,
      "step": 17516
    },
    {
      "epoch": 0.7788893335111823,
      "grad_norm": 0.059511348605155945,
      "learning_rate": 0.00012636420838196995,
      "loss": 1.4598,
      "step": 17518
    },
    {
      "epoch": 0.7789782579698546,
      "grad_norm": 0.05979214981198311,
      "learning_rate": 0.0001262670792974831,
      "loss": 1.4606,
      "step": 17520
    },
    {
      "epoch": 0.779067182428527,
      "grad_norm": 0.059690359979867935,
      "learning_rate": 0.00012616998216133314,
      "loss": 1.4603,
      "step": 17522
    },
    {
      "epoch": 0.7791561068871993,
      "grad_norm": 0.059829000383615494,
      "learning_rate": 0.00012607291698182055,
      "loss": 1.4563,
      "step": 17524
    },
    {
      "epoch": 0.7792450313458716,
      "grad_norm": 0.06074198707938194,
      "learning_rate": 0.000125975883767243,
      "loss": 1.4569,
      "step": 17526
    },
    {
      "epoch": 0.779333955804544,
      "grad_norm": 0.058196790516376495,
      "learning_rate": 0.0001258788825258952,
      "loss": 1.4577,
      "step": 17528
    },
    {
      "epoch": 0.7794228802632164,
      "grad_norm": 0.059000782668590546,
      "learning_rate": 0.0001257819132660689,
      "loss": 1.4581,
      "step": 17530
    },
    {
      "epoch": 0.7795118047218887,
      "grad_norm": 0.058559540659189224,
      "learning_rate": 0.00012568497599605412,
      "loss": 1.4561,
      "step": 17532
    },
    {
      "epoch": 0.7796007291805611,
      "grad_norm": 0.057577475905418396,
      "learning_rate": 0.0001255880707241369,
      "loss": 1.455,
      "step": 17534
    },
    {
      "epoch": 0.7796896536392335,
      "grad_norm": 0.06146826967597008,
      "learning_rate": 0.00012549119745860162,
      "loss": 1.4585,
      "step": 17536
    },
    {
      "epoch": 0.7797785780979059,
      "grad_norm": 0.05853557214140892,
      "learning_rate": 0.00012539435620772898,
      "loss": 1.4611,
      "step": 17538
    },
    {
      "epoch": 0.7798675025565782,
      "grad_norm": 0.0580923855304718,
      "learning_rate": 0.00012529754697979763,
      "loss": 1.4624,
      "step": 17540
    },
    {
      "epoch": 0.7799564270152506,
      "grad_norm": 0.058282919228076935,
      "learning_rate": 0.00012520076978308337,
      "loss": 1.4516,
      "step": 17542
    },
    {
      "epoch": 0.780045351473923,
      "grad_norm": 0.06060004606842995,
      "learning_rate": 0.00012510402462585885,
      "loss": 1.4515,
      "step": 17544
    },
    {
      "epoch": 0.7801342759325952,
      "grad_norm": 0.05811069533228874,
      "learning_rate": 0.0001250073115163944,
      "loss": 1.4587,
      "step": 17546
    },
    {
      "epoch": 0.7802232003912676,
      "grad_norm": 0.05835281312465668,
      "learning_rate": 0.00012491063046295763,
      "loss": 1.4563,
      "step": 17548
    },
    {
      "epoch": 0.78031212484994,
      "grad_norm": 0.059804681688547134,
      "learning_rate": 0.00012481398147381307,
      "loss": 1.4531,
      "step": 17550
    },
    {
      "epoch": 0.7804010493086123,
      "grad_norm": 0.05878927931189537,
      "learning_rate": 0.0001247173645572227,
      "loss": 1.4627,
      "step": 17552
    },
    {
      "epoch": 0.7804899737672847,
      "grad_norm": 0.05973034352064133,
      "learning_rate": 0.00012462077972144597,
      "loss": 1.457,
      "step": 17554
    },
    {
      "epoch": 0.780578898225957,
      "grad_norm": 0.057263296097517014,
      "learning_rate": 0.0001245242269747391,
      "loss": 1.4613,
      "step": 17556
    },
    {
      "epoch": 0.7806678226846294,
      "grad_norm": 0.05798415094614029,
      "learning_rate": 0.00012442770632535598,
      "loss": 1.4565,
      "step": 17558
    },
    {
      "epoch": 0.7807567471433018,
      "grad_norm": 0.05924572795629501,
      "learning_rate": 0.00012433121778154777,
      "loss": 1.4583,
      "step": 17560
    },
    {
      "epoch": 0.7808456716019742,
      "grad_norm": 0.05844467505812645,
      "learning_rate": 0.00012423476135156242,
      "loss": 1.4604,
      "step": 17562
    },
    {
      "epoch": 0.7809345960606465,
      "grad_norm": 0.05907626450061798,
      "learning_rate": 0.00012413833704364574,
      "loss": 1.4532,
      "step": 17564
    },
    {
      "epoch": 0.7810235205193189,
      "grad_norm": 0.059602320194244385,
      "learning_rate": 0.00012404194486604025,
      "loss": 1.455,
      "step": 17566
    },
    {
      "epoch": 0.7811124449779911,
      "grad_norm": 0.05900027975440025,
      "learning_rate": 0.00012394558482698613,
      "loss": 1.4606,
      "step": 17568
    },
    {
      "epoch": 0.7812013694366635,
      "grad_norm": 0.058760497719049454,
      "learning_rate": 0.00012384925693472077,
      "loss": 1.4548,
      "step": 17570
    },
    {
      "epoch": 0.7812902938953359,
      "grad_norm": 0.05956209450960159,
      "learning_rate": 0.0001237529611974784,
      "loss": 1.4485,
      "step": 17572
    },
    {
      "epoch": 0.7813792183540083,
      "grad_norm": 0.057924702763557434,
      "learning_rate": 0.00012365669762349102,
      "loss": 1.4541,
      "step": 17574
    },
    {
      "epoch": 0.7814681428126806,
      "grad_norm": 0.059304650872945786,
      "learning_rate": 0.00012356046622098772,
      "loss": 1.4603,
      "step": 17576
    },
    {
      "epoch": 0.781557067271353,
      "grad_norm": 0.05885639041662216,
      "learning_rate": 0.00012346426699819457,
      "loss": 1.4611,
      "step": 17578
    },
    {
      "epoch": 0.7816459917300254,
      "grad_norm": 0.059227973222732544,
      "learning_rate": 0.0001233680999633352,
      "loss": 1.4572,
      "step": 17580
    },
    {
      "epoch": 0.7817349161886977,
      "grad_norm": 0.059174761176109314,
      "learning_rate": 0.00012327196512463057,
      "loss": 1.4501,
      "step": 17582
    },
    {
      "epoch": 0.7818238406473701,
      "grad_norm": 0.05834069103002548,
      "learning_rate": 0.0001231758624902984,
      "loss": 1.4517,
      "step": 17584
    },
    {
      "epoch": 0.7819127651060425,
      "grad_norm": 0.05900967866182327,
      "learning_rate": 0.0001230797920685542,
      "loss": 1.4536,
      "step": 17586
    },
    {
      "epoch": 0.7820016895647148,
      "grad_norm": 0.0585876889526844,
      "learning_rate": 0.0001229837538676102,
      "loss": 1.4607,
      "step": 17588
    },
    {
      "epoch": 0.7820906140233871,
      "grad_norm": 0.0589335672557354,
      "learning_rate": 0.0001228877478956766,
      "loss": 1.4593,
      "step": 17590
    },
    {
      "epoch": 0.7821795384820595,
      "grad_norm": 0.059638556092977524,
      "learning_rate": 0.00012279177416096022,
      "loss": 1.454,
      "step": 17592
    },
    {
      "epoch": 0.7822684629407318,
      "grad_norm": 0.05850829929113388,
      "learning_rate": 0.00012269583267166513,
      "loss": 1.4535,
      "step": 17594
    },
    {
      "epoch": 0.7823573873994042,
      "grad_norm": 0.05961374565958977,
      "learning_rate": 0.00012259992343599302,
      "loss": 1.4504,
      "step": 17596
    },
    {
      "epoch": 0.7824463118580766,
      "grad_norm": 0.05841527134180069,
      "learning_rate": 0.0001225040464621427,
      "loss": 1.4527,
      "step": 17598
    },
    {
      "epoch": 0.7825352363167489,
      "grad_norm": 0.059410110116004944,
      "learning_rate": 0.00012240820175830997,
      "loss": 1.4515,
      "step": 17600
    },
    {
      "epoch": 0.7826241607754213,
      "grad_norm": 0.05802929401397705,
      "learning_rate": 0.00012231238933268812,
      "loss": 1.4523,
      "step": 17602
    },
    {
      "epoch": 0.7827130852340937,
      "grad_norm": 0.05958373472094536,
      "learning_rate": 0.0001222166091934676,
      "loss": 1.4586,
      "step": 17604
    },
    {
      "epoch": 0.782802009692766,
      "grad_norm": 0.059802230447530746,
      "learning_rate": 0.0001221208613488364,
      "loss": 1.4544,
      "step": 17606
    },
    {
      "epoch": 0.7828909341514384,
      "grad_norm": 0.05877742916345596,
      "learning_rate": 0.00012202514580697904,
      "loss": 1.4551,
      "step": 17608
    },
    {
      "epoch": 0.7829798586101107,
      "grad_norm": 0.05887303501367569,
      "learning_rate": 0.0001219294625760779,
      "loss": 1.4622,
      "step": 17610
    },
    {
      "epoch": 0.783068783068783,
      "grad_norm": 0.05912398546934128,
      "learning_rate": 0.00012183381166431251,
      "loss": 1.453,
      "step": 17612
    },
    {
      "epoch": 0.7831577075274554,
      "grad_norm": 0.05929466336965561,
      "learning_rate": 0.00012173819307985945,
      "loss": 1.4483,
      "step": 17614
    },
    {
      "epoch": 0.7832466319861278,
      "grad_norm": 0.05975263938307762,
      "learning_rate": 0.00012164260683089234,
      "loss": 1.4542,
      "step": 17616
    },
    {
      "epoch": 0.7833355564448001,
      "grad_norm": 0.059671733528375626,
      "learning_rate": 0.00012154705292558283,
      "loss": 1.4588,
      "step": 17618
    },
    {
      "epoch": 0.7834244809034725,
      "grad_norm": 0.06014909967780113,
      "learning_rate": 0.00012145153137209896,
      "loss": 1.451,
      "step": 17620
    },
    {
      "epoch": 0.7835134053621449,
      "grad_norm": 0.05790025368332863,
      "learning_rate": 0.00012135604217860635,
      "loss": 1.4554,
      "step": 17622
    },
    {
      "epoch": 0.7836023298208172,
      "grad_norm": 0.05892461538314819,
      "learning_rate": 0.00012126058535326784,
      "loss": 1.46,
      "step": 17624
    },
    {
      "epoch": 0.7836912542794896,
      "grad_norm": 0.05957599729299545,
      "learning_rate": 0.00012116516090424357,
      "loss": 1.4584,
      "step": 17626
    },
    {
      "epoch": 0.783780178738162,
      "grad_norm": 0.059336572885513306,
      "learning_rate": 0.0001210697688396909,
      "loss": 1.4471,
      "step": 17628
    },
    {
      "epoch": 0.7838691031968343,
      "grad_norm": 0.05946873500943184,
      "learning_rate": 0.0001209744091677642,
      "loss": 1.456,
      "step": 17630
    },
    {
      "epoch": 0.7839580276555066,
      "grad_norm": 0.05893925949931145,
      "learning_rate": 0.00012087908189661528,
      "loss": 1.4565,
      "step": 17632
    },
    {
      "epoch": 0.784046952114179,
      "grad_norm": 0.05950813367962837,
      "learning_rate": 0.0001207837870343933,
      "loss": 1.4495,
      "step": 17634
    },
    {
      "epoch": 0.7841358765728513,
      "grad_norm": 0.058802422136068344,
      "learning_rate": 0.00012068852458924439,
      "loss": 1.4563,
      "step": 17636
    },
    {
      "epoch": 0.7842248010315237,
      "grad_norm": 0.05822760984301567,
      "learning_rate": 0.00012059329456931167,
      "loss": 1.4588,
      "step": 17638
    },
    {
      "epoch": 0.7843137254901961,
      "grad_norm": 0.059016041457653046,
      "learning_rate": 0.00012049809698273645,
      "loss": 1.4554,
      "step": 17640
    },
    {
      "epoch": 0.7844026499488684,
      "grad_norm": 0.05806576833128929,
      "learning_rate": 0.0001204029318376561,
      "loss": 1.4568,
      "step": 17642
    },
    {
      "epoch": 0.7844915744075408,
      "grad_norm": 0.058557651937007904,
      "learning_rate": 0.00012030779914220613,
      "loss": 1.4526,
      "step": 17644
    },
    {
      "epoch": 0.7845804988662132,
      "grad_norm": 0.05871611833572388,
      "learning_rate": 0.00012021269890451863,
      "loss": 1.4518,
      "step": 17646
    },
    {
      "epoch": 0.7846694233248855,
      "grad_norm": 0.05869081988930702,
      "learning_rate": 0.00012011763113272329,
      "loss": 1.452,
      "step": 17648
    },
    {
      "epoch": 0.7847583477835579,
      "grad_norm": 0.059065330773591995,
      "learning_rate": 0.00012002259583494701,
      "loss": 1.4584,
      "step": 17650
    },
    {
      "epoch": 0.7848472722422303,
      "grad_norm": 0.05952310189604759,
      "learning_rate": 0.00011992759301931366,
      "loss": 1.4574,
      "step": 17652
    },
    {
      "epoch": 0.7849361967009025,
      "grad_norm": 0.05919057875871658,
      "learning_rate": 0.00011983262269394457,
      "loss": 1.4492,
      "step": 17654
    },
    {
      "epoch": 0.7850251211595749,
      "grad_norm": 0.05960650369524956,
      "learning_rate": 0.00011973768486695835,
      "loss": 1.4556,
      "step": 17656
    },
    {
      "epoch": 0.7851140456182473,
      "grad_norm": 0.0592602901160717,
      "learning_rate": 0.00011964277954647046,
      "loss": 1.4503,
      "step": 17658
    },
    {
      "epoch": 0.7852029700769196,
      "grad_norm": 0.05901474878191948,
      "learning_rate": 0.00011954790674059401,
      "loss": 1.4545,
      "step": 17660
    },
    {
      "epoch": 0.785291894535592,
      "grad_norm": 0.05881018191576004,
      "learning_rate": 0.0001194530664574392,
      "loss": 1.4524,
      "step": 17662
    },
    {
      "epoch": 0.7853808189942644,
      "grad_norm": 0.05886894464492798,
      "learning_rate": 0.00011935825870511308,
      "loss": 1.4614,
      "step": 17664
    },
    {
      "epoch": 0.7854697434529367,
      "grad_norm": 0.059833452105522156,
      "learning_rate": 0.00011926348349172051,
      "loss": 1.458,
      "step": 17666
    },
    {
      "epoch": 0.7855586679116091,
      "grad_norm": 0.06001525744795799,
      "learning_rate": 0.00011916874082536333,
      "loss": 1.4564,
      "step": 17668
    },
    {
      "epoch": 0.7856475923702815,
      "grad_norm": 0.06033634394407272,
      "learning_rate": 0.00011907403071414036,
      "loss": 1.4544,
      "step": 17670
    },
    {
      "epoch": 0.7857365168289538,
      "grad_norm": 0.058471135795116425,
      "learning_rate": 0.00011897935316614799,
      "loss": 1.4526,
      "step": 17672
    },
    {
      "epoch": 0.7858254412876262,
      "grad_norm": 0.06029105186462402,
      "learning_rate": 0.00011888470818947955,
      "loss": 1.4565,
      "step": 17674
    },
    {
      "epoch": 0.7859143657462985,
      "grad_norm": 0.059264685958623886,
      "learning_rate": 0.00011879009579222577,
      "loss": 1.4518,
      "step": 17676
    },
    {
      "epoch": 0.7860032902049708,
      "grad_norm": 0.059138696640729904,
      "learning_rate": 0.00011869551598247463,
      "loss": 1.4484,
      "step": 17678
    },
    {
      "epoch": 0.7860922146636432,
      "grad_norm": 0.05941601097583771,
      "learning_rate": 0.000118600968768311,
      "loss": 1.4567,
      "step": 17680
    },
    {
      "epoch": 0.7861811391223156,
      "grad_norm": 0.05975447967648506,
      "learning_rate": 0.00011850645415781736,
      "loss": 1.4537,
      "step": 17682
    },
    {
      "epoch": 0.7862700635809879,
      "grad_norm": 0.05916404351592064,
      "learning_rate": 0.00011841197215907335,
      "loss": 1.4555,
      "step": 17684
    },
    {
      "epoch": 0.7863589880396603,
      "grad_norm": 0.061422303318977356,
      "learning_rate": 0.00011831752278015545,
      "loss": 1.4601,
      "step": 17686
    },
    {
      "epoch": 0.7864479124983327,
      "grad_norm": 0.05970718339085579,
      "learning_rate": 0.00011822310602913766,
      "loss": 1.4566,
      "step": 17688
    },
    {
      "epoch": 0.786536836957005,
      "grad_norm": 0.05843975022435188,
      "learning_rate": 0.00011812872191409135,
      "loss": 1.454,
      "step": 17690
    },
    {
      "epoch": 0.7866257614156774,
      "grad_norm": 0.05985402688384056,
      "learning_rate": 0.00011803437044308463,
      "loss": 1.4532,
      "step": 17692
    },
    {
      "epoch": 0.7867146858743498,
      "grad_norm": 0.059144627302885056,
      "learning_rate": 0.00011794005162418336,
      "loss": 1.4588,
      "step": 17694
    },
    {
      "epoch": 0.7868036103330222,
      "grad_norm": 0.058817509561777115,
      "learning_rate": 0.00011784576546544984,
      "loss": 1.4517,
      "step": 17696
    },
    {
      "epoch": 0.7868925347916944,
      "grad_norm": 0.059203263372182846,
      "learning_rate": 0.00011775151197494466,
      "loss": 1.4543,
      "step": 17698
    },
    {
      "epoch": 0.7869814592503668,
      "grad_norm": 0.05954177677631378,
      "learning_rate": 0.00011765729116072476,
      "loss": 1.4546,
      "step": 17700
    },
    {
      "epoch": 0.7870703837090391,
      "grad_norm": 0.0586751364171505,
      "learning_rate": 0.0001175631030308444,
      "loss": 1.4472,
      "step": 17702
    },
    {
      "epoch": 0.7871593081677115,
      "grad_norm": 0.058443281799554825,
      "learning_rate": 0.00011746894759335535,
      "loss": 1.4547,
      "step": 17704
    },
    {
      "epoch": 0.7872482326263839,
      "grad_norm": 0.05910266563296318,
      "learning_rate": 0.00011737482485630657,
      "loss": 1.4556,
      "step": 17706
    },
    {
      "epoch": 0.7873371570850562,
      "grad_norm": 0.05853736400604248,
      "learning_rate": 0.00011728073482774377,
      "loss": 1.4566,
      "step": 17708
    },
    {
      "epoch": 0.7874260815437286,
      "grad_norm": 0.0591680109500885,
      "learning_rate": 0.00011718667751571032,
      "loss": 1.4587,
      "step": 17710
    },
    {
      "epoch": 0.787515006002401,
      "grad_norm": 0.05998816341161728,
      "learning_rate": 0.00011709265292824672,
      "loss": 1.4581,
      "step": 17712
    },
    {
      "epoch": 0.7876039304610734,
      "grad_norm": 0.0575900636613369,
      "learning_rate": 0.00011699866107339068,
      "loss": 1.4542,
      "step": 17714
    },
    {
      "epoch": 0.7876928549197457,
      "grad_norm": 0.05857378616929054,
      "learning_rate": 0.00011690470195917696,
      "loss": 1.4543,
      "step": 17716
    },
    {
      "epoch": 0.787781779378418,
      "grad_norm": 0.05935800448060036,
      "learning_rate": 0.00011681077559363729,
      "loss": 1.4565,
      "step": 17718
    },
    {
      "epoch": 0.7878707038370903,
      "grad_norm": 0.058298259973526,
      "learning_rate": 0.00011671688198480141,
      "loss": 1.455,
      "step": 17720
    },
    {
      "epoch": 0.7879596282957627,
      "grad_norm": 0.058936238288879395,
      "learning_rate": 0.00011662302114069556,
      "loss": 1.4562,
      "step": 17722
    },
    {
      "epoch": 0.7880485527544351,
      "grad_norm": 0.05869641527533531,
      "learning_rate": 0.00011652919306934324,
      "loss": 1.454,
      "step": 17724
    },
    {
      "epoch": 0.7881374772131075,
      "grad_norm": 0.058170247822999954,
      "learning_rate": 0.00011643539777876532,
      "loss": 1.4549,
      "step": 17726
    },
    {
      "epoch": 0.7882264016717798,
      "grad_norm": 0.05906420946121216,
      "learning_rate": 0.00011634163527697999,
      "loss": 1.4512,
      "step": 17728
    },
    {
      "epoch": 0.7883153261304522,
      "grad_norm": 0.05926912650465965,
      "learning_rate": 0.00011624790557200254,
      "loss": 1.4524,
      "step": 17730
    },
    {
      "epoch": 0.7884042505891246,
      "grad_norm": 0.05841286852955818,
      "learning_rate": 0.00011615420867184506,
      "loss": 1.4534,
      "step": 17732
    },
    {
      "epoch": 0.7884931750477969,
      "grad_norm": 0.059124939143657684,
      "learning_rate": 0.00011606054458451742,
      "loss": 1.4579,
      "step": 17734
    },
    {
      "epoch": 0.7885820995064693,
      "grad_norm": 0.05994617938995361,
      "learning_rate": 0.00011596691331802656,
      "loss": 1.451,
      "step": 17736
    },
    {
      "epoch": 0.7886710239651417,
      "grad_norm": 0.058965619653463364,
      "learning_rate": 0.00011587331488037611,
      "loss": 1.4557,
      "step": 17738
    },
    {
      "epoch": 0.7887599484238139,
      "grad_norm": 0.059728700667619705,
      "learning_rate": 0.00011577974927956758,
      "loss": 1.4493,
      "step": 17740
    },
    {
      "epoch": 0.7888488728824863,
      "grad_norm": 0.061226364225149155,
      "learning_rate": 0.00011568621652359935,
      "loss": 1.4536,
      "step": 17742
    },
    {
      "epoch": 0.7889377973411587,
      "grad_norm": 0.06103841960430145,
      "learning_rate": 0.000115592716620467,
      "loss": 1.4539,
      "step": 17744
    },
    {
      "epoch": 0.789026721799831,
      "grad_norm": 0.05864092707633972,
      "learning_rate": 0.00011549924957816293,
      "loss": 1.4571,
      "step": 17746
    },
    {
      "epoch": 0.7891156462585034,
      "grad_norm": 0.059012021869421005,
      "learning_rate": 0.00011540581540467782,
      "loss": 1.4519,
      "step": 17748
    },
    {
      "epoch": 0.7892045707171758,
      "grad_norm": 0.0590151883661747,
      "learning_rate": 0.00011531241410799825,
      "loss": 1.4522,
      "step": 17750
    },
    {
      "epoch": 0.7892934951758481,
      "grad_norm": 0.058573827147483826,
      "learning_rate": 0.00011521904569610892,
      "loss": 1.4585,
      "step": 17752
    },
    {
      "epoch": 0.7893824196345205,
      "grad_norm": 0.058176856487989426,
      "learning_rate": 0.00011512571017699114,
      "loss": 1.4557,
      "step": 17754
    },
    {
      "epoch": 0.7894713440931929,
      "grad_norm": 0.05986884981393814,
      "learning_rate": 0.00011503240755862371,
      "loss": 1.4505,
      "step": 17756
    },
    {
      "epoch": 0.7895602685518652,
      "grad_norm": 0.059089165180921555,
      "learning_rate": 0.0001149391378489827,
      "loss": 1.4538,
      "step": 17758
    },
    {
      "epoch": 0.7896491930105376,
      "grad_norm": 0.05954886972904205,
      "learning_rate": 0.00011484590105604098,
      "loss": 1.4562,
      "step": 17760
    },
    {
      "epoch": 0.7897381174692099,
      "grad_norm": 0.05822194367647171,
      "learning_rate": 0.00011475269718776898,
      "loss": 1.4533,
      "step": 17762
    },
    {
      "epoch": 0.7898270419278822,
      "grad_norm": 0.06061677634716034,
      "learning_rate": 0.00011465952625213428,
      "loss": 1.4525,
      "step": 17764
    },
    {
      "epoch": 0.7899159663865546,
      "grad_norm": 0.05945700407028198,
      "learning_rate": 0.00011456638825710125,
      "loss": 1.4522,
      "step": 17766
    },
    {
      "epoch": 0.790004890845227,
      "grad_norm": 0.060659851878881454,
      "learning_rate": 0.00011447328321063188,
      "loss": 1.4567,
      "step": 17768
    },
    {
      "epoch": 0.7900938153038993,
      "grad_norm": 0.05868120491504669,
      "learning_rate": 0.00011438021112068541,
      "loss": 1.4555,
      "step": 17770
    },
    {
      "epoch": 0.7901827397625717,
      "grad_norm": 0.05930672958493233,
      "learning_rate": 0.00011428717199521771,
      "loss": 1.4523,
      "step": 17772
    },
    {
      "epoch": 0.7902716642212441,
      "grad_norm": 0.058497123420238495,
      "learning_rate": 0.0001141941658421825,
      "loss": 1.4578,
      "step": 17774
    },
    {
      "epoch": 0.7903605886799164,
      "grad_norm": 0.05888426676392555,
      "learning_rate": 0.00011410119266953001,
      "loss": 1.4535,
      "step": 17776
    },
    {
      "epoch": 0.7904495131385888,
      "grad_norm": 0.05962875857949257,
      "learning_rate": 0.00011400825248520818,
      "loss": 1.4571,
      "step": 17778
    },
    {
      "epoch": 0.7905384375972612,
      "grad_norm": 0.060313645750284195,
      "learning_rate": 0.00011391534529716208,
      "loss": 1.4503,
      "step": 17780
    },
    {
      "epoch": 0.7906273620559335,
      "grad_norm": 0.059360720217227936,
      "learning_rate": 0.00011382247111333355,
      "loss": 1.4556,
      "step": 17782
    },
    {
      "epoch": 0.7907162865146058,
      "grad_norm": 0.058396752923727036,
      "learning_rate": 0.00011372962994166203,
      "loss": 1.4519,
      "step": 17784
    },
    {
      "epoch": 0.7908052109732782,
      "grad_norm": 0.058448676019907,
      "learning_rate": 0.00011363682179008406,
      "loss": 1.4496,
      "step": 17786
    },
    {
      "epoch": 0.7908941354319505,
      "grad_norm": 0.06033730506896973,
      "learning_rate": 0.00011354404666653311,
      "loss": 1.4524,
      "step": 17788
    },
    {
      "epoch": 0.7909830598906229,
      "grad_norm": 0.059229444712400436,
      "learning_rate": 0.00011345130457894015,
      "loss": 1.4524,
      "step": 17790
    },
    {
      "epoch": 0.7910719843492953,
      "grad_norm": 0.0583866648375988,
      "learning_rate": 0.0001133585955352332,
      "loss": 1.4583,
      "step": 17792
    },
    {
      "epoch": 0.7911609088079676,
      "grad_norm": 0.05953531712293625,
      "learning_rate": 0.00011326591954333732,
      "loss": 1.4556,
      "step": 17794
    },
    {
      "epoch": 0.79124983326664,
      "grad_norm": 0.05893312767148018,
      "learning_rate": 0.00011317327661117488,
      "loss": 1.4497,
      "step": 17796
    },
    {
      "epoch": 0.7913387577253124,
      "grad_norm": 0.057053301483392715,
      "learning_rate": 0.0001130806667466654,
      "loss": 1.4489,
      "step": 17798
    },
    {
      "epoch": 0.7914276821839847,
      "grad_norm": 0.05844390019774437,
      "learning_rate": 0.00011298808995772581,
      "loss": 1.4558,
      "step": 17800
    },
    {
      "epoch": 0.7915166066426571,
      "grad_norm": 0.05953795462846756,
      "learning_rate": 0.00011289554625226977,
      "loss": 1.4548,
      "step": 17802
    },
    {
      "epoch": 0.7916055311013295,
      "grad_norm": 0.05917130783200264,
      "learning_rate": 0.00011280303563820815,
      "loss": 1.4518,
      "step": 17804
    },
    {
      "epoch": 0.7916944555600017,
      "grad_norm": 0.0593201145529747,
      "learning_rate": 0.00011271055812344955,
      "loss": 1.4529,
      "step": 17806
    },
    {
      "epoch": 0.7917833800186741,
      "grad_norm": 0.0592842660844326,
      "learning_rate": 0.00011261811371589919,
      "loss": 1.4521,
      "step": 17808
    },
    {
      "epoch": 0.7918723044773465,
      "grad_norm": 0.059808555990457535,
      "learning_rate": 0.0001125257024234595,
      "loss": 1.4525,
      "step": 17810
    },
    {
      "epoch": 0.7919612289360188,
      "grad_norm": 0.05857331305742264,
      "learning_rate": 0.0001124333242540303,
      "loss": 1.4547,
      "step": 17812
    },
    {
      "epoch": 0.7920501533946912,
      "grad_norm": 0.05948074907064438,
      "learning_rate": 0.00011234097921550862,
      "loss": 1.4527,
      "step": 17814
    },
    {
      "epoch": 0.7921390778533636,
      "grad_norm": 0.05830054357647896,
      "learning_rate": 0.00011224866731578831,
      "loss": 1.4561,
      "step": 17816
    },
    {
      "epoch": 0.7922280023120359,
      "grad_norm": 0.057090599089860916,
      "learning_rate": 0.00011215638856276061,
      "loss": 1.4556,
      "step": 17818
    },
    {
      "epoch": 0.7923169267707083,
      "grad_norm": 0.05876989662647247,
      "learning_rate": 0.000112064142964314,
      "loss": 1.4605,
      "step": 17820
    },
    {
      "epoch": 0.7924058512293807,
      "grad_norm": 0.05962676927447319,
      "learning_rate": 0.00011197193052833421,
      "loss": 1.4564,
      "step": 17822
    },
    {
      "epoch": 0.792494775688053,
      "grad_norm": 0.058956604450941086,
      "learning_rate": 0.00011187975126270372,
      "loss": 1.4526,
      "step": 17824
    },
    {
      "epoch": 0.7925837001467254,
      "grad_norm": 0.06046443060040474,
      "learning_rate": 0.00011178760517530224,
      "loss": 1.4493,
      "step": 17826
    },
    {
      "epoch": 0.7926726246053977,
      "grad_norm": 0.059528037905693054,
      "learning_rate": 0.00011169549227400733,
      "loss": 1.4591,
      "step": 17828
    },
    {
      "epoch": 0.79276154906407,
      "grad_norm": 0.060843151062726974,
      "learning_rate": 0.00011160341256669299,
      "loss": 1.4522,
      "step": 17830
    },
    {
      "epoch": 0.7928504735227424,
      "grad_norm": 0.06023893877863884,
      "learning_rate": 0.00011151136606123036,
      "loss": 1.4516,
      "step": 17832
    },
    {
      "epoch": 0.7929393979814148,
      "grad_norm": 0.05774928256869316,
      "learning_rate": 0.0001114193527654882,
      "loss": 1.4534,
      "step": 17834
    },
    {
      "epoch": 0.7930283224400871,
      "grad_norm": 0.05887966603040695,
      "learning_rate": 0.00011132737268733217,
      "loss": 1.4559,
      "step": 17836
    },
    {
      "epoch": 0.7931172468987595,
      "grad_norm": 0.05917837843298912,
      "learning_rate": 0.00011123542583462531,
      "loss": 1.4621,
      "step": 17838
    },
    {
      "epoch": 0.7932061713574319,
      "grad_norm": 0.05823491886258125,
      "learning_rate": 0.00011114351221522728,
      "loss": 1.4513,
      "step": 17840
    },
    {
      "epoch": 0.7932950958161042,
      "grad_norm": 0.059828367084264755,
      "learning_rate": 0.00011105163183699552,
      "loss": 1.4576,
      "step": 17842
    },
    {
      "epoch": 0.7933840202747766,
      "grad_norm": 0.05933848023414612,
      "learning_rate": 0.00011095978470778445,
      "loss": 1.4588,
      "step": 17844
    },
    {
      "epoch": 0.793472944733449,
      "grad_norm": 0.05880282446742058,
      "learning_rate": 0.00011086797083544526,
      "loss": 1.46,
      "step": 17846
    },
    {
      "epoch": 0.7935618691921212,
      "grad_norm": 0.05894262716174126,
      "learning_rate": 0.00011077619022782676,
      "loss": 1.4547,
      "step": 17848
    },
    {
      "epoch": 0.7936507936507936,
      "grad_norm": 0.05894016847014427,
      "learning_rate": 0.00011068444289277496,
      "loss": 1.4554,
      "step": 17850
    },
    {
      "epoch": 0.793739718109466,
      "grad_norm": 0.06030754745006561,
      "learning_rate": 0.00011059272883813248,
      "loss": 1.4538,
      "step": 17852
    },
    {
      "epoch": 0.7938286425681383,
      "grad_norm": 0.05975908041000366,
      "learning_rate": 0.0001105010480717396,
      "loss": 1.4545,
      "step": 17854
    },
    {
      "epoch": 0.7939175670268107,
      "grad_norm": 0.05732753127813339,
      "learning_rate": 0.00011040940060143373,
      "loss": 1.4432,
      "step": 17856
    },
    {
      "epoch": 0.7940064914854831,
      "grad_norm": 0.05890670418739319,
      "learning_rate": 0.00011031778643504902,
      "loss": 1.4489,
      "step": 17858
    },
    {
      "epoch": 0.7940954159441554,
      "grad_norm": 0.05884822458028793,
      "learning_rate": 0.00011022620558041735,
      "loss": 1.4521,
      "step": 17860
    },
    {
      "epoch": 0.7941843404028278,
      "grad_norm": 0.058066993951797485,
      "learning_rate": 0.00011013465804536715,
      "loss": 1.457,
      "step": 17862
    },
    {
      "epoch": 0.7942732648615002,
      "grad_norm": 0.05799723044037819,
      "learning_rate": 0.00011004314383772446,
      "loss": 1.4489,
      "step": 17864
    },
    {
      "epoch": 0.7943621893201726,
      "grad_norm": 0.058095499873161316,
      "learning_rate": 0.00010995166296531251,
      "loss": 1.446,
      "step": 17866
    },
    {
      "epoch": 0.7944511137788449,
      "grad_norm": 0.05935432016849518,
      "learning_rate": 0.0001098602154359511,
      "loss": 1.4607,
      "step": 17868
    },
    {
      "epoch": 0.7945400382375172,
      "grad_norm": 0.06020105257630348,
      "learning_rate": 0.0001097688012574578,
      "loss": 1.4538,
      "step": 17870
    },
    {
      "epoch": 0.7946289626961895,
      "grad_norm": 0.05931893363595009,
      "learning_rate": 0.00010967742043764717,
      "loss": 1.454,
      "step": 17872
    },
    {
      "epoch": 0.7947178871548619,
      "grad_norm": 0.059080854058265686,
      "learning_rate": 0.00010958607298433059,
      "loss": 1.454,
      "step": 17874
    },
    {
      "epoch": 0.7948068116135343,
      "grad_norm": 0.05938325077295303,
      "learning_rate": 0.00010949475890531701,
      "loss": 1.4534,
      "step": 17876
    },
    {
      "epoch": 0.7948957360722066,
      "grad_norm": 0.059726350009441376,
      "learning_rate": 0.00010940347820841251,
      "loss": 1.4464,
      "step": 17878
    },
    {
      "epoch": 0.794984660530879,
      "grad_norm": 0.05838613212108612,
      "learning_rate": 0.00010931223090141984,
      "loss": 1.4567,
      "step": 17880
    },
    {
      "epoch": 0.7950735849895514,
      "grad_norm": 0.05989585816860199,
      "learning_rate": 0.00010922101699213948,
      "loss": 1.4549,
      "step": 17882
    },
    {
      "epoch": 0.7951625094482238,
      "grad_norm": 0.05945861339569092,
      "learning_rate": 0.00010912983648836861,
      "loss": 1.4555,
      "step": 17884
    },
    {
      "epoch": 0.7952514339068961,
      "grad_norm": 0.05932047963142395,
      "learning_rate": 0.0001090386893979018,
      "loss": 1.4556,
      "step": 17886
    },
    {
      "epoch": 0.7953403583655685,
      "grad_norm": 0.059849612414836884,
      "learning_rate": 0.00010894757572853086,
      "loss": 1.4606,
      "step": 17888
    },
    {
      "epoch": 0.7954292828242409,
      "grad_norm": 0.05838996544480324,
      "learning_rate": 0.00010885649548804433,
      "loss": 1.4567,
      "step": 17890
    },
    {
      "epoch": 0.7955182072829131,
      "grad_norm": 0.05834618955850601,
      "learning_rate": 0.00010876544868422827,
      "loss": 1.4574,
      "step": 17892
    },
    {
      "epoch": 0.7956071317415855,
      "grad_norm": 0.05943123623728752,
      "learning_rate": 0.00010867443532486587,
      "loss": 1.4577,
      "step": 17894
    },
    {
      "epoch": 0.7956960562002579,
      "grad_norm": 0.05874134600162506,
      "learning_rate": 0.00010858345541773717,
      "loss": 1.4508,
      "step": 17896
    },
    {
      "epoch": 0.7957849806589302,
      "grad_norm": 0.0590997152030468,
      "learning_rate": 0.00010849250897061963,
      "loss": 1.4554,
      "step": 17898
    },
    {
      "epoch": 0.7958739051176026,
      "grad_norm": 0.05840156599879265,
      "learning_rate": 0.00010840159599128785,
      "loss": 1.4579,
      "step": 17900
    },
    {
      "epoch": 0.795962829576275,
      "grad_norm": 0.05960099771618843,
      "learning_rate": 0.00010831071648751322,
      "loss": 1.4622,
      "step": 17902
    },
    {
      "epoch": 0.7960517540349473,
      "grad_norm": 0.05900590494275093,
      "learning_rate": 0.00010821987046706472,
      "loss": 1.4528,
      "step": 17904
    },
    {
      "epoch": 0.7961406784936197,
      "grad_norm": 0.05795317143201828,
      "learning_rate": 0.00010812905793770816,
      "loss": 1.4526,
      "step": 17906
    },
    {
      "epoch": 0.7962296029522921,
      "grad_norm": 0.06108953431248665,
      "learning_rate": 0.00010803827890720675,
      "loss": 1.4571,
      "step": 17908
    },
    {
      "epoch": 0.7963185274109644,
      "grad_norm": 0.059297602623701096,
      "learning_rate": 0.00010794753338332064,
      "loss": 1.4518,
      "step": 17910
    },
    {
      "epoch": 0.7964074518696368,
      "grad_norm": 0.059192463755607605,
      "learning_rate": 0.00010785682137380681,
      "loss": 1.4542,
      "step": 17912
    },
    {
      "epoch": 0.796496376328309,
      "grad_norm": 0.05878617987036705,
      "learning_rate": 0.00010776614288642028,
      "loss": 1.4563,
      "step": 17914
    },
    {
      "epoch": 0.7965853007869814,
      "grad_norm": 0.05914062634110451,
      "learning_rate": 0.00010767549792891235,
      "loss": 1.4566,
      "step": 17916
    },
    {
      "epoch": 0.7966742252456538,
      "grad_norm": 0.06022977828979492,
      "learning_rate": 0.00010758488650903165,
      "loss": 1.4577,
      "step": 17918
    },
    {
      "epoch": 0.7967631497043262,
      "grad_norm": 0.05821533128619194,
      "learning_rate": 0.0001074943086345242,
      "loss": 1.4544,
      "step": 17920
    },
    {
      "epoch": 0.7968520741629985,
      "grad_norm": 0.0580059252679348,
      "learning_rate": 0.00010740376431313293,
      "loss": 1.4557,
      "step": 17922
    },
    {
      "epoch": 0.7969409986216709,
      "grad_norm": 0.05985068529844284,
      "learning_rate": 0.00010731325355259813,
      "loss": 1.4548,
      "step": 17924
    },
    {
      "epoch": 0.7970299230803433,
      "grad_norm": 0.05859328433871269,
      "learning_rate": 0.0001072227763606568,
      "loss": 1.4509,
      "step": 17926
    },
    {
      "epoch": 0.7971188475390156,
      "grad_norm": 0.0583089180290699,
      "learning_rate": 0.00010713233274504348,
      "loss": 1.4518,
      "step": 17928
    },
    {
      "epoch": 0.797207771997688,
      "grad_norm": 0.05798826739192009,
      "learning_rate": 0.00010704192271348978,
      "loss": 1.4551,
      "step": 17930
    },
    {
      "epoch": 0.7972966964563604,
      "grad_norm": 0.05828414484858513,
      "learning_rate": 0.00010695154627372428,
      "loss": 1.4539,
      "step": 17932
    },
    {
      "epoch": 0.7973856209150327,
      "grad_norm": 0.05891380086541176,
      "learning_rate": 0.00010686120343347238,
      "loss": 1.4568,
      "step": 17934
    },
    {
      "epoch": 0.797474545373705,
      "grad_norm": 0.059466563165187836,
      "learning_rate": 0.00010677089420045765,
      "loss": 1.4491,
      "step": 17936
    },
    {
      "epoch": 0.7975634698323774,
      "grad_norm": 0.059041835367679596,
      "learning_rate": 0.00010668061858239974,
      "loss": 1.4512,
      "step": 17938
    },
    {
      "epoch": 0.7976523942910497,
      "grad_norm": 0.05893898755311966,
      "learning_rate": 0.00010659037658701576,
      "loss": 1.4581,
      "step": 17940
    },
    {
      "epoch": 0.7977413187497221,
      "grad_norm": 0.05860443785786629,
      "learning_rate": 0.00010650016822202008,
      "loss": 1.4536,
      "step": 17942
    },
    {
      "epoch": 0.7978302432083945,
      "grad_norm": 0.058257170021533966,
      "learning_rate": 0.00010640999349512409,
      "loss": 1.4592,
      "step": 17944
    },
    {
      "epoch": 0.7979191676670668,
      "grad_norm": 0.05957287922501564,
      "learning_rate": 0.00010631985241403653,
      "loss": 1.4529,
      "step": 17946
    },
    {
      "epoch": 0.7980080921257392,
      "grad_norm": 0.05837190896272659,
      "learning_rate": 0.00010622974498646271,
      "loss": 1.4488,
      "step": 17948
    },
    {
      "epoch": 0.7980970165844116,
      "grad_norm": 0.05990975350141525,
      "learning_rate": 0.00010613967122010553,
      "loss": 1.4598,
      "step": 17950
    },
    {
      "epoch": 0.7981859410430839,
      "grad_norm": 0.05874193459749222,
      "learning_rate": 0.00010604963112266513,
      "loss": 1.4528,
      "step": 17952
    },
    {
      "epoch": 0.7982748655017563,
      "grad_norm": 0.05808878690004349,
      "learning_rate": 0.00010595962470183823,
      "loss": 1.4524,
      "step": 17954
    },
    {
      "epoch": 0.7983637899604287,
      "grad_norm": 0.05945536866784096,
      "learning_rate": 0.00010586965196531905,
      "loss": 1.4542,
      "step": 17956
    },
    {
      "epoch": 0.7984527144191009,
      "grad_norm": 0.06018710881471634,
      "learning_rate": 0.00010577971292079907,
      "loss": 1.4543,
      "step": 17958
    },
    {
      "epoch": 0.7985416388777733,
      "grad_norm": 0.05900532007217407,
      "learning_rate": 0.00010568980757596636,
      "loss": 1.4469,
      "step": 17960
    },
    {
      "epoch": 0.7986305633364457,
      "grad_norm": 0.0572827085852623,
      "learning_rate": 0.00010559993593850654,
      "loss": 1.4496,
      "step": 17962
    },
    {
      "epoch": 0.798719487795118,
      "grad_norm": 0.06311970204114914,
      "learning_rate": 0.0001055100980161024,
      "loss": 1.4553,
      "step": 17964
    },
    {
      "epoch": 0.7988084122537904,
      "grad_norm": 0.058263469487428665,
      "learning_rate": 0.00010542029381643343,
      "loss": 1.4536,
      "step": 17966
    },
    {
      "epoch": 0.7988973367124628,
      "grad_norm": 0.06007849797606468,
      "learning_rate": 0.00010533052334717675,
      "loss": 1.4567,
      "step": 17968
    },
    {
      "epoch": 0.7989862611711351,
      "grad_norm": 0.05845961719751358,
      "learning_rate": 0.000105240786616006,
      "loss": 1.4598,
      "step": 17970
    },
    {
      "epoch": 0.7990751856298075,
      "grad_norm": 0.058356136083602905,
      "learning_rate": 0.0001051510836305925,
      "loss": 1.4548,
      "step": 17972
    },
    {
      "epoch": 0.7991641100884799,
      "grad_norm": 0.05997835844755173,
      "learning_rate": 0.0001050614143986045,
      "loss": 1.4535,
      "step": 17974
    },
    {
      "epoch": 0.7992530345471522,
      "grad_norm": 0.06147778779268265,
      "learning_rate": 0.00010497177892770715,
      "loss": 1.4545,
      "step": 17976
    },
    {
      "epoch": 0.7993419590058245,
      "grad_norm": 0.058420371264219284,
      "learning_rate": 0.00010488217722556292,
      "loss": 1.4469,
      "step": 17978
    },
    {
      "epoch": 0.7994308834644969,
      "grad_norm": 0.05859336629509926,
      "learning_rate": 0.00010479260929983159,
      "loss": 1.4563,
      "step": 17980
    },
    {
      "epoch": 0.7995198079231692,
      "grad_norm": 0.059518374502658844,
      "learning_rate": 0.00010470307515816951,
      "loss": 1.4526,
      "step": 17982
    },
    {
      "epoch": 0.7996087323818416,
      "grad_norm": 0.058449771255254745,
      "learning_rate": 0.00010461357480823052,
      "loss": 1.4492,
      "step": 17984
    },
    {
      "epoch": 0.799697656840514,
      "grad_norm": 0.06011267751455307,
      "learning_rate": 0.00010452410825766579,
      "loss": 1.4624,
      "step": 17986
    },
    {
      "epoch": 0.7997865812991863,
      "grad_norm": 0.05873945727944374,
      "learning_rate": 0.00010443467551412289,
      "loss": 1.4558,
      "step": 17988
    },
    {
      "epoch": 0.7998755057578587,
      "grad_norm": 0.06056896224617958,
      "learning_rate": 0.00010434527658524729,
      "loss": 1.4524,
      "step": 17990
    },
    {
      "epoch": 0.7999644302165311,
      "grad_norm": 0.05866500735282898,
      "learning_rate": 0.00010425591147868086,
      "loss": 1.4497,
      "step": 17992
    },
    {
      "epoch": 0.8000533546752034,
      "grad_norm": 0.0592012032866478,
      "learning_rate": 0.00010416658020206316,
      "loss": 1.4537,
      "step": 17994
    },
    {
      "epoch": 0.8001422791338758,
      "grad_norm": 0.06068159639835358,
      "learning_rate": 0.0001040772827630307,
      "loss": 1.4533,
      "step": 17996
    },
    {
      "epoch": 0.8002312035925482,
      "grad_norm": 0.060801684856414795,
      "learning_rate": 0.00010398801916921669,
      "loss": 1.4512,
      "step": 17998
    },
    {
      "epoch": 0.8003201280512204,
      "grad_norm": 0.05898714438080788,
      "learning_rate": 0.00010389878942825204,
      "loss": 1.451,
      "step": 18000
    },
    {
      "epoch": 0.8003201280512204,
      "eval_loss": 1.4389121532440186,
      "eval_runtime": 12.4447,
      "eval_samples_per_second": 555.256,
      "eval_steps_per_second": 69.427,
      "step": 18000
    },
    {
      "epoch": 0.8004090525098928,
      "grad_norm": 0.05900314450263977,
      "learning_rate": 0.00010380959354776454,
      "loss": 1.4508,
      "step": 18002
    },
    {
      "epoch": 0.8004979769685652,
      "grad_norm": 0.05843342840671539,
      "learning_rate": 0.0001037204315353788,
      "loss": 1.4576,
      "step": 18004
    },
    {
      "epoch": 0.8005869014272375,
      "grad_norm": 0.059655383229255676,
      "learning_rate": 0.0001036313033987169,
      "loss": 1.4518,
      "step": 18006
    },
    {
      "epoch": 0.8006758258859099,
      "grad_norm": 0.05997476726770401,
      "learning_rate": 0.00010354220914539808,
      "loss": 1.4529,
      "step": 18008
    },
    {
      "epoch": 0.8007647503445823,
      "grad_norm": 0.05772737041115761,
      "learning_rate": 0.00010345314878303824,
      "loss": 1.4528,
      "step": 18010
    },
    {
      "epoch": 0.8008536748032546,
      "grad_norm": 0.05794164165854454,
      "learning_rate": 0.00010336412231925074,
      "loss": 1.4483,
      "step": 18012
    },
    {
      "epoch": 0.800942599261927,
      "grad_norm": 0.05926957726478577,
      "learning_rate": 0.000103275129761646,
      "loss": 1.4539,
      "step": 18014
    },
    {
      "epoch": 0.8010315237205994,
      "grad_norm": 0.05818663910031319,
      "learning_rate": 0.00010318617111783157,
      "loss": 1.4481,
      "step": 18016
    },
    {
      "epoch": 0.8011204481792717,
      "grad_norm": 0.060195665806531906,
      "learning_rate": 0.00010309724639541196,
      "loss": 1.4547,
      "step": 18018
    },
    {
      "epoch": 0.8012093726379441,
      "grad_norm": 0.06009439751505852,
      "learning_rate": 0.00010300835560198873,
      "loss": 1.4512,
      "step": 18020
    },
    {
      "epoch": 0.8012982970966164,
      "grad_norm": 0.05936377868056297,
      "learning_rate": 0.00010291949874516071,
      "loss": 1.4537,
      "step": 18022
    },
    {
      "epoch": 0.8013872215552887,
      "grad_norm": 0.05843667313456535,
      "learning_rate": 0.00010283067583252393,
      "loss": 1.4567,
      "step": 18024
    },
    {
      "epoch": 0.8014761460139611,
      "grad_norm": 0.05879830941557884,
      "learning_rate": 0.00010274188687167112,
      "loss": 1.4563,
      "step": 18026
    },
    {
      "epoch": 0.8015650704726335,
      "grad_norm": 0.058328744024038315,
      "learning_rate": 0.00010265313187019254,
      "loss": 1.4555,
      "step": 18028
    },
    {
      "epoch": 0.8016539949313058,
      "grad_norm": 0.05817698314785957,
      "learning_rate": 0.00010256441083567524,
      "loss": 1.4539,
      "step": 18030
    },
    {
      "epoch": 0.8017429193899782,
      "grad_norm": 0.06012728065252304,
      "learning_rate": 0.00010247572377570369,
      "loss": 1.454,
      "step": 18032
    },
    {
      "epoch": 0.8018318438486506,
      "grad_norm": 0.05919637903571129,
      "learning_rate": 0.00010238707069785897,
      "loss": 1.4564,
      "step": 18034
    },
    {
      "epoch": 0.801920768307323,
      "grad_norm": 0.058458033949136734,
      "learning_rate": 0.00010229845160971967,
      "loss": 1.4531,
      "step": 18036
    },
    {
      "epoch": 0.8020096927659953,
      "grad_norm": 0.059698980301618576,
      "learning_rate": 0.00010220986651886149,
      "loss": 1.4494,
      "step": 18038
    },
    {
      "epoch": 0.8020986172246677,
      "grad_norm": 0.05928831920027733,
      "learning_rate": 0.00010212131543285691,
      "loss": 1.4512,
      "step": 18040
    },
    {
      "epoch": 0.80218754168334,
      "grad_norm": 0.05956317111849785,
      "learning_rate": 0.00010203279835927536,
      "loss": 1.4557,
      "step": 18042
    },
    {
      "epoch": 0.8022764661420123,
      "grad_norm": 0.05859542265534401,
      "learning_rate": 0.00010194431530568432,
      "loss": 1.4568,
      "step": 18044
    },
    {
      "epoch": 0.8023653906006847,
      "grad_norm": 0.060236856341362,
      "learning_rate": 0.00010185586627964727,
      "loss": 1.4524,
      "step": 18046
    },
    {
      "epoch": 0.802454315059357,
      "grad_norm": 0.05890626460313797,
      "learning_rate": 0.00010176745128872544,
      "loss": 1.448,
      "step": 18048
    },
    {
      "epoch": 0.8025432395180294,
      "grad_norm": 0.05818956717848778,
      "learning_rate": 0.00010167907034047669,
      "loss": 1.4551,
      "step": 18050
    },
    {
      "epoch": 0.8026321639767018,
      "grad_norm": 0.05896204337477684,
      "learning_rate": 0.00010159072344245634,
      "loss": 1.4493,
      "step": 18052
    },
    {
      "epoch": 0.8027210884353742,
      "grad_norm": 0.0573628768324852,
      "learning_rate": 0.00010150241060221682,
      "loss": 1.4554,
      "step": 18054
    },
    {
      "epoch": 0.8028100128940465,
      "grad_norm": 0.0591515451669693,
      "learning_rate": 0.00010141413182730724,
      "loss": 1.4532,
      "step": 18056
    },
    {
      "epoch": 0.8028989373527189,
      "grad_norm": 0.0614108145236969,
      "learning_rate": 0.00010132588712527418,
      "loss": 1.4555,
      "step": 18058
    },
    {
      "epoch": 0.8029878618113913,
      "grad_norm": 0.05975984036922455,
      "learning_rate": 0.00010123767650366134,
      "loss": 1.4472,
      "step": 18060
    },
    {
      "epoch": 0.8030767862700636,
      "grad_norm": 0.05843626707792282,
      "learning_rate": 0.00010114949997000916,
      "loss": 1.4553,
      "step": 18062
    },
    {
      "epoch": 0.803165710728736,
      "grad_norm": 0.05943736433982849,
      "learning_rate": 0.0001010613575318552,
      "loss": 1.4515,
      "step": 18064
    },
    {
      "epoch": 0.8032546351874083,
      "grad_norm": 0.05971449986100197,
      "learning_rate": 0.00010097324919673467,
      "loss": 1.4504,
      "step": 18066
    },
    {
      "epoch": 0.8033435596460806,
      "grad_norm": 0.05878691002726555,
      "learning_rate": 0.00010088517497217914,
      "loss": 1.4576,
      "step": 18068
    },
    {
      "epoch": 0.803432484104753,
      "grad_norm": 0.059320129454135895,
      "learning_rate": 0.00010079713486571784,
      "loss": 1.4577,
      "step": 18070
    },
    {
      "epoch": 0.8035214085634254,
      "grad_norm": 0.05824529007077217,
      "learning_rate": 0.00010070912888487649,
      "loss": 1.4574,
      "step": 18072
    },
    {
      "epoch": 0.8036103330220977,
      "grad_norm": 0.05816292017698288,
      "learning_rate": 0.0001006211570371785,
      "loss": 1.453,
      "step": 18074
    },
    {
      "epoch": 0.8036992574807701,
      "grad_norm": 0.059266943484544754,
      "learning_rate": 0.00010053321933014409,
      "loss": 1.4515,
      "step": 18076
    },
    {
      "epoch": 0.8037881819394425,
      "grad_norm": 0.05980881303548813,
      "learning_rate": 0.0001004453157712904,
      "loss": 1.4508,
      "step": 18078
    },
    {
      "epoch": 0.8038771063981148,
      "grad_norm": 0.05897355452179909,
      "learning_rate": 0.00010035744636813187,
      "loss": 1.4602,
      "step": 18080
    },
    {
      "epoch": 0.8039660308567872,
      "grad_norm": 0.060474496334791183,
      "learning_rate": 0.00010026961112818011,
      "loss": 1.4631,
      "step": 18082
    },
    {
      "epoch": 0.8040549553154596,
      "grad_norm": 0.05896021053195,
      "learning_rate": 0.00010018181005894345,
      "loss": 1.4544,
      "step": 18084
    },
    {
      "epoch": 0.8041438797741318,
      "grad_norm": 0.05860500782728195,
      "learning_rate": 0.00010009404316792753,
      "loss": 1.4476,
      "step": 18086
    },
    {
      "epoch": 0.8042328042328042,
      "grad_norm": 0.06036067008972168,
      "learning_rate": 0.00010000631046263537,
      "loss": 1.4552,
      "step": 18088
    },
    {
      "epoch": 0.8043217286914766,
      "grad_norm": 0.061282288283109665,
      "learning_rate": 9.991861195056629e-05,
      "loss": 1.4558,
      "step": 18090
    },
    {
      "epoch": 0.8044106531501489,
      "grad_norm": 0.058544036000967026,
      "learning_rate": 9.98309476392174e-05,
      "loss": 1.4548,
      "step": 18092
    },
    {
      "epoch": 0.8044995776088213,
      "grad_norm": 0.05853986367583275,
      "learning_rate": 9.974331753608274e-05,
      "loss": 1.4542,
      "step": 18094
    },
    {
      "epoch": 0.8045885020674937,
      "grad_norm": 0.058233581483364105,
      "learning_rate": 9.965572164865299e-05,
      "loss": 1.4565,
      "step": 18096
    },
    {
      "epoch": 0.804677426526166,
      "grad_norm": 0.05811081454157829,
      "learning_rate": 9.956815998441659e-05,
      "loss": 1.4523,
      "step": 18098
    },
    {
      "epoch": 0.8047663509848384,
      "grad_norm": 0.0591614656150341,
      "learning_rate": 9.948063255085821e-05,
      "loss": 1.4454,
      "step": 18100
    },
    {
      "epoch": 0.8048552754435108,
      "grad_norm": 0.0593455545604229,
      "learning_rate": 9.939313935546069e-05,
      "loss": 1.4566,
      "step": 18102
    },
    {
      "epoch": 0.8049441999021831,
      "grad_norm": 0.059242065995931625,
      "learning_rate": 9.930568040570304e-05,
      "loss": 1.453,
      "step": 18104
    },
    {
      "epoch": 0.8050331243608555,
      "grad_norm": 0.05771373212337494,
      "learning_rate": 9.921825570906145e-05,
      "loss": 1.4504,
      "step": 18106
    },
    {
      "epoch": 0.8051220488195278,
      "grad_norm": 0.05904533714056015,
      "learning_rate": 9.913086527300962e-05,
      "loss": 1.4539,
      "step": 18108
    },
    {
      "epoch": 0.8052109732782001,
      "grad_norm": 0.057896826416254044,
      "learning_rate": 9.904350910501808e-05,
      "loss": 1.4482,
      "step": 18110
    },
    {
      "epoch": 0.8052998977368725,
      "grad_norm": 0.05723195523023605,
      "learning_rate": 9.895618721255422e-05,
      "loss": 1.4536,
      "step": 18112
    },
    {
      "epoch": 0.8053888221955449,
      "grad_norm": 0.05811459571123123,
      "learning_rate": 9.886889960308281e-05,
      "loss": 1.4508,
      "step": 18114
    },
    {
      "epoch": 0.8054777466542172,
      "grad_norm": 0.058095820248126984,
      "learning_rate": 9.878164628406578e-05,
      "loss": 1.4489,
      "step": 18116
    },
    {
      "epoch": 0.8055666711128896,
      "grad_norm": 0.059355780482292175,
      "learning_rate": 9.869442726296157e-05,
      "loss": 1.456,
      "step": 18118
    },
    {
      "epoch": 0.805655595571562,
      "grad_norm": 0.059321098029613495,
      "learning_rate": 9.860724254722631e-05,
      "loss": 1.4501,
      "step": 18120
    },
    {
      "epoch": 0.8057445200302343,
      "grad_norm": 0.05816931650042534,
      "learning_rate": 9.852009214431262e-05,
      "loss": 1.4581,
      "step": 18122
    },
    {
      "epoch": 0.8058334444889067,
      "grad_norm": 0.05967382341623306,
      "learning_rate": 9.843297606167095e-05,
      "loss": 1.4558,
      "step": 18124
    },
    {
      "epoch": 0.8059223689475791,
      "grad_norm": 0.05909135937690735,
      "learning_rate": 9.834589430674817e-05,
      "loss": 1.4519,
      "step": 18126
    },
    {
      "epoch": 0.8060112934062514,
      "grad_norm": 0.05855505168437958,
      "learning_rate": 9.825884688698822e-05,
      "loss": 1.4511,
      "step": 18128
    },
    {
      "epoch": 0.8061002178649237,
      "grad_norm": 0.0576552115380764,
      "learning_rate": 9.817183380983246e-05,
      "loss": 1.4466,
      "step": 18130
    },
    {
      "epoch": 0.8061891423235961,
      "grad_norm": 0.059187132865190506,
      "learning_rate": 9.808485508271925e-05,
      "loss": 1.4462,
      "step": 18132
    },
    {
      "epoch": 0.8062780667822684,
      "grad_norm": 0.059183891862630844,
      "learning_rate": 9.799791071308373e-05,
      "loss": 1.4586,
      "step": 18134
    },
    {
      "epoch": 0.8063669912409408,
      "grad_norm": 0.05837311968207359,
      "learning_rate": 9.791100070835834e-05,
      "loss": 1.4525,
      "step": 18136
    },
    {
      "epoch": 0.8064559156996132,
      "grad_norm": 0.059481870383024216,
      "learning_rate": 9.782412507597255e-05,
      "loss": 1.4545,
      "step": 18138
    },
    {
      "epoch": 0.8065448401582855,
      "grad_norm": 0.05896589532494545,
      "learning_rate": 9.7737283823353e-05,
      "loss": 1.4523,
      "step": 18140
    },
    {
      "epoch": 0.8066337646169579,
      "grad_norm": 0.058587439358234406,
      "learning_rate": 9.765047695792306e-05,
      "loss": 1.4515,
      "step": 18142
    },
    {
      "epoch": 0.8067226890756303,
      "grad_norm": 0.058542508631944656,
      "learning_rate": 9.756370448710339e-05,
      "loss": 1.45,
      "step": 18144
    },
    {
      "epoch": 0.8068116135343026,
      "grad_norm": 0.05747520923614502,
      "learning_rate": 9.747696641831194e-05,
      "loss": 1.45,
      "step": 18146
    },
    {
      "epoch": 0.806900537992975,
      "grad_norm": 0.05875362455844879,
      "learning_rate": 9.739026275896323e-05,
      "loss": 1.451,
      "step": 18148
    },
    {
      "epoch": 0.8069894624516474,
      "grad_norm": 0.06102447956800461,
      "learning_rate": 9.730359351646884e-05,
      "loss": 1.4512,
      "step": 18150
    },
    {
      "epoch": 0.8070783869103196,
      "grad_norm": 0.059680186212062836,
      "learning_rate": 9.72169586982381e-05,
      "loss": 1.4508,
      "step": 18152
    },
    {
      "epoch": 0.807167311368992,
      "grad_norm": 0.05882233753800392,
      "learning_rate": 9.713035831167672e-05,
      "loss": 1.4519,
      "step": 18154
    },
    {
      "epoch": 0.8072562358276644,
      "grad_norm": 0.05874011665582657,
      "learning_rate": 9.704379236418776e-05,
      "loss": 1.4537,
      "step": 18156
    },
    {
      "epoch": 0.8073451602863367,
      "grad_norm": 0.059738658368587494,
      "learning_rate": 9.695726086317108e-05,
      "loss": 1.4502,
      "step": 18158
    },
    {
      "epoch": 0.8074340847450091,
      "grad_norm": 0.05960926041007042,
      "learning_rate": 9.687076381602388e-05,
      "loss": 1.4602,
      "step": 18160
    },
    {
      "epoch": 0.8075230092036815,
      "grad_norm": 0.060264989733695984,
      "learning_rate": 9.678430123014048e-05,
      "loss": 1.4487,
      "step": 18162
    },
    {
      "epoch": 0.8076119336623538,
      "grad_norm": 0.05822618305683136,
      "learning_rate": 9.669787311291178e-05,
      "loss": 1.4562,
      "step": 18164
    },
    {
      "epoch": 0.8077008581210262,
      "grad_norm": 0.05792885646224022,
      "learning_rate": 9.661147947172616e-05,
      "loss": 1.4564,
      "step": 18166
    },
    {
      "epoch": 0.8077897825796986,
      "grad_norm": 0.060065533965826035,
      "learning_rate": 9.652512031396916e-05,
      "loss": 1.4594,
      "step": 18168
    },
    {
      "epoch": 0.807878707038371,
      "grad_norm": 0.05816401168704033,
      "learning_rate": 9.643879564702269e-05,
      "loss": 1.4583,
      "step": 18170
    },
    {
      "epoch": 0.8079676314970433,
      "grad_norm": 0.05799930915236473,
      "learning_rate": 9.635250547826646e-05,
      "loss": 1.4538,
      "step": 18172
    },
    {
      "epoch": 0.8080565559557156,
      "grad_norm": 0.05812399461865425,
      "learning_rate": 9.626624981507703e-05,
      "loss": 1.4559,
      "step": 18174
    },
    {
      "epoch": 0.8081454804143879,
      "grad_norm": 0.05904964357614517,
      "learning_rate": 9.618002866482761e-05,
      "loss": 1.4522,
      "step": 18176
    },
    {
      "epoch": 0.8082344048730603,
      "grad_norm": 0.05795786529779434,
      "learning_rate": 9.609384203488903e-05,
      "loss": 1.4508,
      "step": 18178
    },
    {
      "epoch": 0.8083233293317327,
      "grad_norm": 0.0590299516916275,
      "learning_rate": 9.60076899326287e-05,
      "loss": 1.4599,
      "step": 18180
    },
    {
      "epoch": 0.808412253790405,
      "grad_norm": 0.05843953415751457,
      "learning_rate": 9.592157236541132e-05,
      "loss": 1.454,
      "step": 18182
    },
    {
      "epoch": 0.8085011782490774,
      "grad_norm": 0.05828475579619408,
      "learning_rate": 9.583548934059882e-05,
      "loss": 1.4551,
      "step": 18184
    },
    {
      "epoch": 0.8085901027077498,
      "grad_norm": 0.058298975229263306,
      "learning_rate": 9.574944086554965e-05,
      "loss": 1.4501,
      "step": 18186
    },
    {
      "epoch": 0.8086790271664221,
      "grad_norm": 0.05863676592707634,
      "learning_rate": 9.566342694761976e-05,
      "loss": 1.4562,
      "step": 18188
    },
    {
      "epoch": 0.8087679516250945,
      "grad_norm": 0.05812937393784523,
      "learning_rate": 9.557744759416209e-05,
      "loss": 1.4525,
      "step": 18190
    },
    {
      "epoch": 0.8088568760837669,
      "grad_norm": 0.05726495757699013,
      "learning_rate": 9.549150281252633e-05,
      "loss": 1.448,
      "step": 18192
    },
    {
      "epoch": 0.8089458005424393,
      "grad_norm": 0.06033691018819809,
      "learning_rate": 9.54055926100595e-05,
      "loss": 1.4607,
      "step": 18194
    },
    {
      "epoch": 0.8090347250011115,
      "grad_norm": 0.0588860847055912,
      "learning_rate": 9.531971699410574e-05,
      "loss": 1.4526,
      "step": 18196
    },
    {
      "epoch": 0.8091236494597839,
      "grad_norm": 0.058994125574827194,
      "learning_rate": 9.523387597200578e-05,
      "loss": 1.4491,
      "step": 18198
    },
    {
      "epoch": 0.8092125739184562,
      "grad_norm": 0.05951934680342674,
      "learning_rate": 9.514806955109789e-05,
      "loss": 1.4552,
      "step": 18200
    },
    {
      "epoch": 0.8093014983771286,
      "grad_norm": 0.05920296534895897,
      "learning_rate": 9.506229773871727e-05,
      "loss": 1.4548,
      "step": 18202
    },
    {
      "epoch": 0.809390422835801,
      "grad_norm": 0.05928421393036842,
      "learning_rate": 9.497656054219578e-05,
      "loss": 1.4561,
      "step": 18204
    },
    {
      "epoch": 0.8094793472944734,
      "grad_norm": 0.059079330414533615,
      "learning_rate": 9.489085796886293e-05,
      "loss": 1.4492,
      "step": 18206
    },
    {
      "epoch": 0.8095682717531457,
      "grad_norm": 0.058482855558395386,
      "learning_rate": 9.480519002604454e-05,
      "loss": 1.449,
      "step": 18208
    },
    {
      "epoch": 0.8096571962118181,
      "grad_norm": 0.05823582410812378,
      "learning_rate": 9.471955672106436e-05,
      "loss": 1.4564,
      "step": 18210
    },
    {
      "epoch": 0.8097461206704905,
      "grad_norm": 0.05962125584483147,
      "learning_rate": 9.463395806124254e-05,
      "loss": 1.4559,
      "step": 18212
    },
    {
      "epoch": 0.8098350451291628,
      "grad_norm": 0.05842282250523567,
      "learning_rate": 9.45483940538962e-05,
      "loss": 1.4502,
      "step": 18214
    },
    {
      "epoch": 0.8099239695878351,
      "grad_norm": 0.05836905539035797,
      "learning_rate": 9.446286470633997e-05,
      "loss": 1.4528,
      "step": 18216
    },
    {
      "epoch": 0.8100128940465074,
      "grad_norm": 0.05897628888487816,
      "learning_rate": 9.437737002588525e-05,
      "loss": 1.4538,
      "step": 18218
    },
    {
      "epoch": 0.8101018185051798,
      "grad_norm": 0.05997723713517189,
      "learning_rate": 9.429191001984039e-05,
      "loss": 1.4573,
      "step": 18220
    },
    {
      "epoch": 0.8101907429638522,
      "grad_norm": 0.05981513485312462,
      "learning_rate": 9.420648469551097e-05,
      "loss": 1.456,
      "step": 18222
    },
    {
      "epoch": 0.8102796674225246,
      "grad_norm": 0.059481166303157806,
      "learning_rate": 9.412109406019948e-05,
      "loss": 1.4506,
      "step": 18224
    },
    {
      "epoch": 0.8103685918811969,
      "grad_norm": 0.059796884655952454,
      "learning_rate": 9.403573812120564e-05,
      "loss": 1.4566,
      "step": 18226
    },
    {
      "epoch": 0.8104575163398693,
      "grad_norm": 0.05872061103582382,
      "learning_rate": 9.395041688582596e-05,
      "loss": 1.4532,
      "step": 18228
    },
    {
      "epoch": 0.8105464407985417,
      "grad_norm": 0.060998715460300446,
      "learning_rate": 9.38651303613538e-05,
      "loss": 1.4493,
      "step": 18230
    },
    {
      "epoch": 0.810635365257214,
      "grad_norm": 0.05913839489221573,
      "learning_rate": 9.37798785550803e-05,
      "loss": 1.4568,
      "step": 18232
    },
    {
      "epoch": 0.8107242897158864,
      "grad_norm": 0.05807075276970863,
      "learning_rate": 9.369466147429296e-05,
      "loss": 1.4499,
      "step": 18234
    },
    {
      "epoch": 0.8108132141745588,
      "grad_norm": 0.05896204710006714,
      "learning_rate": 9.360947912627632e-05,
      "loss": 1.4542,
      "step": 18236
    },
    {
      "epoch": 0.810902138633231,
      "grad_norm": 0.05949874222278595,
      "learning_rate": 9.352433151831236e-05,
      "loss": 1.4478,
      "step": 18238
    },
    {
      "epoch": 0.8109910630919034,
      "grad_norm": 0.059194955974817276,
      "learning_rate": 9.34392186576798e-05,
      "loss": 1.454,
      "step": 18240
    },
    {
      "epoch": 0.8110799875505758,
      "grad_norm": 0.05962388962507248,
      "learning_rate": 9.335414055165459e-05,
      "loss": 1.4458,
      "step": 18242
    },
    {
      "epoch": 0.8111689120092481,
      "grad_norm": 0.05830967053771019,
      "learning_rate": 9.326909720750937e-05,
      "loss": 1.4562,
      "step": 18244
    },
    {
      "epoch": 0.8112578364679205,
      "grad_norm": 0.05975262448191643,
      "learning_rate": 9.31840886325141e-05,
      "loss": 1.4476,
      "step": 18246
    },
    {
      "epoch": 0.8113467609265929,
      "grad_norm": 0.058683622628450394,
      "learning_rate": 9.309911483393586e-05,
      "loss": 1.4508,
      "step": 18248
    },
    {
      "epoch": 0.8114356853852652,
      "grad_norm": 0.05931386351585388,
      "learning_rate": 9.301417581903831e-05,
      "loss": 1.4536,
      "step": 18250
    },
    {
      "epoch": 0.8115246098439376,
      "grad_norm": 0.058767132461071014,
      "learning_rate": 9.292927159508258e-05,
      "loss": 1.4536,
      "step": 18252
    },
    {
      "epoch": 0.81161353430261,
      "grad_norm": 0.060294680297374725,
      "learning_rate": 9.284440216932666e-05,
      "loss": 1.4611,
      "step": 18254
    },
    {
      "epoch": 0.8117024587612823,
      "grad_norm": 0.05967991426587105,
      "learning_rate": 9.27595675490256e-05,
      "loss": 1.4506,
      "step": 18256
    },
    {
      "epoch": 0.8117913832199547,
      "grad_norm": 0.05928469076752663,
      "learning_rate": 9.267476774143102e-05,
      "loss": 1.4445,
      "step": 18258
    },
    {
      "epoch": 0.811880307678627,
      "grad_norm": 0.059566475450992584,
      "learning_rate": 9.259000275379264e-05,
      "loss": 1.4484,
      "step": 18260
    },
    {
      "epoch": 0.8119692321372993,
      "grad_norm": 0.06041361764073372,
      "learning_rate": 9.250527259335606e-05,
      "loss": 1.4577,
      "step": 18262
    },
    {
      "epoch": 0.8120581565959717,
      "grad_norm": 0.059527479112148285,
      "learning_rate": 9.242057726736469e-05,
      "loss": 1.4552,
      "step": 18264
    },
    {
      "epoch": 0.8121470810546441,
      "grad_norm": 0.058299604803323746,
      "learning_rate": 9.233591678305842e-05,
      "loss": 1.4522,
      "step": 18266
    },
    {
      "epoch": 0.8122360055133164,
      "grad_norm": 0.05863171070814133,
      "learning_rate": 9.225129114767445e-05,
      "loss": 1.4465,
      "step": 18268
    },
    {
      "epoch": 0.8123249299719888,
      "grad_norm": 0.05925234407186508,
      "learning_rate": 9.216670036844715e-05,
      "loss": 1.454,
      "step": 18270
    },
    {
      "epoch": 0.8124138544306612,
      "grad_norm": 0.05960720032453537,
      "learning_rate": 9.208214445260749e-05,
      "loss": 1.4497,
      "step": 18272
    },
    {
      "epoch": 0.8125027788893335,
      "grad_norm": 0.05938321352005005,
      "learning_rate": 9.199762340738365e-05,
      "loss": 1.4473,
      "step": 18274
    },
    {
      "epoch": 0.8125917033480059,
      "grad_norm": 0.058269064873456955,
      "learning_rate": 9.191313724000117e-05,
      "loss": 1.4525,
      "step": 18276
    },
    {
      "epoch": 0.8126806278066783,
      "grad_norm": 0.05836362764239311,
      "learning_rate": 9.182868595768196e-05,
      "loss": 1.4513,
      "step": 18278
    },
    {
      "epoch": 0.8127695522653506,
      "grad_norm": 0.058995332568883896,
      "learning_rate": 9.174426956764542e-05,
      "loss": 1.4535,
      "step": 18280
    },
    {
      "epoch": 0.8128584767240229,
      "grad_norm": 0.05960138887166977,
      "learning_rate": 9.165988807710801e-05,
      "loss": 1.4555,
      "step": 18282
    },
    {
      "epoch": 0.8129474011826953,
      "grad_norm": 0.06037072464823723,
      "learning_rate": 9.157554149328267e-05,
      "loss": 1.4468,
      "step": 18284
    },
    {
      "epoch": 0.8130363256413676,
      "grad_norm": 0.060436177998781204,
      "learning_rate": 9.149122982338009e-05,
      "loss": 1.452,
      "step": 18286
    },
    {
      "epoch": 0.81312525010004,
      "grad_norm": 0.058294836431741714,
      "learning_rate": 9.140695307460728e-05,
      "loss": 1.4551,
      "step": 18288
    },
    {
      "epoch": 0.8132141745587124,
      "grad_norm": 0.06048420071601868,
      "learning_rate": 9.132271125416874e-05,
      "loss": 1.4588,
      "step": 18290
    },
    {
      "epoch": 0.8133030990173847,
      "grad_norm": 0.060230907052755356,
      "learning_rate": 9.12385043692659e-05,
      "loss": 1.4543,
      "step": 18292
    },
    {
      "epoch": 0.8133920234760571,
      "grad_norm": 0.05797860398888588,
      "learning_rate": 9.115433242709697e-05,
      "loss": 1.4504,
      "step": 18294
    },
    {
      "epoch": 0.8134809479347295,
      "grad_norm": 0.05805959552526474,
      "learning_rate": 9.107019543485745e-05,
      "loss": 1.457,
      "step": 18296
    },
    {
      "epoch": 0.8135698723934018,
      "grad_norm": 0.05792992562055588,
      "learning_rate": 9.098609339973985e-05,
      "loss": 1.4444,
      "step": 18298
    },
    {
      "epoch": 0.8136587968520742,
      "grad_norm": 0.05891840532422066,
      "learning_rate": 9.09020263289333e-05,
      "loss": 1.4542,
      "step": 18300
    },
    {
      "epoch": 0.8137477213107466,
      "grad_norm": 0.05851350352168083,
      "learning_rate": 9.081799422962434e-05,
      "loss": 1.4516,
      "step": 18302
    },
    {
      "epoch": 0.8138366457694188,
      "grad_norm": 0.05911056697368622,
      "learning_rate": 9.073399710899661e-05,
      "loss": 1.4541,
      "step": 18304
    },
    {
      "epoch": 0.8139255702280912,
      "grad_norm": 0.05874814838171005,
      "learning_rate": 9.065003497423024e-05,
      "loss": 1.4522,
      "step": 18306
    },
    {
      "epoch": 0.8140144946867636,
      "grad_norm": 0.058131199330091476,
      "learning_rate": 9.056610783250286e-05,
      "loss": 1.4552,
      "step": 18308
    },
    {
      "epoch": 0.8141034191454359,
      "grad_norm": 0.058527685701847076,
      "learning_rate": 9.048221569098903e-05,
      "loss": 1.4523,
      "step": 18310
    },
    {
      "epoch": 0.8141923436041083,
      "grad_norm": 0.0587623156607151,
      "learning_rate": 9.03983585568599e-05,
      "loss": 1.4523,
      "step": 18312
    },
    {
      "epoch": 0.8142812680627807,
      "grad_norm": 0.05750168114900589,
      "learning_rate": 9.031453643728433e-05,
      "loss": 1.456,
      "step": 18314
    },
    {
      "epoch": 0.814370192521453,
      "grad_norm": 0.058666735887527466,
      "learning_rate": 9.023074933942743e-05,
      "loss": 1.4556,
      "step": 18316
    },
    {
      "epoch": 0.8144591169801254,
      "grad_norm": 0.05856854468584061,
      "learning_rate": 9.014699727045189e-05,
      "loss": 1.4595,
      "step": 18318
    },
    {
      "epoch": 0.8145480414387978,
      "grad_norm": 0.058630138635635376,
      "learning_rate": 9.006328023751736e-05,
      "loss": 1.452,
      "step": 18320
    },
    {
      "epoch": 0.8146369658974701,
      "grad_norm": 0.05786571651697159,
      "learning_rate": 8.997959824777996e-05,
      "loss": 1.4531,
      "step": 18322
    },
    {
      "epoch": 0.8147258903561424,
      "grad_norm": 0.0589289627969265,
      "learning_rate": 8.989595130839345e-05,
      "loss": 1.4475,
      "step": 18324
    },
    {
      "epoch": 0.8148148148148148,
      "grad_norm": 0.05925757810473442,
      "learning_rate": 8.981233942650841e-05,
      "loss": 1.4559,
      "step": 18326
    },
    {
      "epoch": 0.8149037392734871,
      "grad_norm": 0.05926579236984253,
      "learning_rate": 8.97287626092721e-05,
      "loss": 1.4516,
      "step": 18328
    },
    {
      "epoch": 0.8149926637321595,
      "grad_norm": 0.05869268253445625,
      "learning_rate": 8.96452208638292e-05,
      "loss": 1.4525,
      "step": 18330
    },
    {
      "epoch": 0.8150815881908319,
      "grad_norm": 0.05770133063197136,
      "learning_rate": 8.956171419732118e-05,
      "loss": 1.4493,
      "step": 18332
    },
    {
      "epoch": 0.8151705126495042,
      "grad_norm": 0.05826232582330704,
      "learning_rate": 8.947824261688669e-05,
      "loss": 1.4566,
      "step": 18334
    },
    {
      "epoch": 0.8152594371081766,
      "grad_norm": 0.05967103689908981,
      "learning_rate": 8.939480612966112e-05,
      "loss": 1.4492,
      "step": 18336
    },
    {
      "epoch": 0.815348361566849,
      "grad_norm": 0.05853937938809395,
      "learning_rate": 8.931140474277677e-05,
      "loss": 1.4557,
      "step": 18338
    },
    {
      "epoch": 0.8154372860255213,
      "grad_norm": 0.05865015089511871,
      "learning_rate": 8.922803846336363e-05,
      "loss": 1.4566,
      "step": 18340
    },
    {
      "epoch": 0.8155262104841937,
      "grad_norm": 0.0575866624712944,
      "learning_rate": 8.914470729854802e-05,
      "loss": 1.4522,
      "step": 18342
    },
    {
      "epoch": 0.8156151349428661,
      "grad_norm": 0.05788535624742508,
      "learning_rate": 8.90614112554532e-05,
      "loss": 1.4507,
      "step": 18344
    },
    {
      "epoch": 0.8157040594015383,
      "grad_norm": 0.058020517230033875,
      "learning_rate": 8.897815034119994e-05,
      "loss": 1.4498,
      "step": 18346
    },
    {
      "epoch": 0.8157929838602107,
      "grad_norm": 0.061876051127910614,
      "learning_rate": 8.889492456290571e-05,
      "loss": 1.4573,
      "step": 18348
    },
    {
      "epoch": 0.8158819083188831,
      "grad_norm": 0.05803605914115906,
      "learning_rate": 8.88117339276851e-05,
      "loss": 1.4507,
      "step": 18350
    },
    {
      "epoch": 0.8159708327775554,
      "grad_norm": 0.058719929307699203,
      "learning_rate": 8.872857844264942e-05,
      "loss": 1.4549,
      "step": 18352
    },
    {
      "epoch": 0.8160597572362278,
      "grad_norm": 0.0580810084939003,
      "learning_rate": 8.864545811490731e-05,
      "loss": 1.4587,
      "step": 18354
    },
    {
      "epoch": 0.8161486816949002,
      "grad_norm": 0.059138279408216476,
      "learning_rate": 8.856237295156427e-05,
      "loss": 1.4546,
      "step": 18356
    },
    {
      "epoch": 0.8162376061535725,
      "grad_norm": 0.05815105512738228,
      "learning_rate": 8.847932295972277e-05,
      "loss": 1.4511,
      "step": 18358
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 0.05784906446933746,
      "learning_rate": 8.839630814648203e-05,
      "loss": 1.4503,
      "step": 18360
    },
    {
      "epoch": 0.8164154550709173,
      "grad_norm": 0.06052152067422867,
      "learning_rate": 8.831332851893897e-05,
      "loss": 1.4514,
      "step": 18362
    },
    {
      "epoch": 0.8165043795295897,
      "grad_norm": 0.058384090662002563,
      "learning_rate": 8.823038408418671e-05,
      "loss": 1.4554,
      "step": 18364
    },
    {
      "epoch": 0.816593303988262,
      "grad_norm": 0.059240229427814484,
      "learning_rate": 8.814747484931595e-05,
      "loss": 1.4546,
      "step": 18366
    },
    {
      "epoch": 0.8166822284469343,
      "grad_norm": 0.059385642409324646,
      "learning_rate": 8.806460082141393e-05,
      "loss": 1.4566,
      "step": 18368
    },
    {
      "epoch": 0.8167711529056066,
      "grad_norm": 0.05777799338102341,
      "learning_rate": 8.798176200756513e-05,
      "loss": 1.453,
      "step": 18370
    },
    {
      "epoch": 0.816860077364279,
      "grad_norm": 0.05718129873275757,
      "learning_rate": 8.789895841485118e-05,
      "loss": 1.4475,
      "step": 18372
    },
    {
      "epoch": 0.8169490018229514,
      "grad_norm": 0.05819839611649513,
      "learning_rate": 8.781619005035019e-05,
      "loss": 1.4509,
      "step": 18374
    },
    {
      "epoch": 0.8170379262816237,
      "grad_norm": 0.05856141820549965,
      "learning_rate": 8.773345692113771e-05,
      "loss": 1.4562,
      "step": 18376
    },
    {
      "epoch": 0.8171268507402961,
      "grad_norm": 0.05731408670544624,
      "learning_rate": 8.76507590342862e-05,
      "loss": 1.4501,
      "step": 18378
    },
    {
      "epoch": 0.8172157751989685,
      "grad_norm": 0.05817363038659096,
      "learning_rate": 8.756809639686492e-05,
      "loss": 1.4516,
      "step": 18380
    },
    {
      "epoch": 0.8173046996576409,
      "grad_norm": 0.05989028140902519,
      "learning_rate": 8.748546901594028e-05,
      "loss": 1.4541,
      "step": 18382
    },
    {
      "epoch": 0.8173936241163132,
      "grad_norm": 0.058697037398815155,
      "learning_rate": 8.740287689857574e-05,
      "loss": 1.4507,
      "step": 18384
    },
    {
      "epoch": 0.8174825485749856,
      "grad_norm": 0.05862530320882797,
      "learning_rate": 8.732032005183144e-05,
      "loss": 1.4558,
      "step": 18386
    },
    {
      "epoch": 0.817571473033658,
      "grad_norm": 0.058638229966163635,
      "learning_rate": 8.72377984827648e-05,
      "loss": 1.4499,
      "step": 18388
    },
    {
      "epoch": 0.8176603974923302,
      "grad_norm": 0.05860773101449013,
      "learning_rate": 8.715531219843026e-05,
      "loss": 1.4585,
      "step": 18390
    },
    {
      "epoch": 0.8177493219510026,
      "grad_norm": 0.05907618626952171,
      "learning_rate": 8.707286120587882e-05,
      "loss": 1.4575,
      "step": 18392
    },
    {
      "epoch": 0.817838246409675,
      "grad_norm": 0.058161091059446335,
      "learning_rate": 8.699044551215907e-05,
      "loss": 1.4474,
      "step": 18394
    },
    {
      "epoch": 0.8179271708683473,
      "grad_norm": 0.05811753496527672,
      "learning_rate": 8.690806512431598e-05,
      "loss": 1.4469,
      "step": 18396
    },
    {
      "epoch": 0.8180160953270197,
      "grad_norm": 0.058913908898830414,
      "learning_rate": 8.682572004939188e-05,
      "loss": 1.4574,
      "step": 18398
    },
    {
      "epoch": 0.8181050197856921,
      "grad_norm": 0.0583617277443409,
      "learning_rate": 8.674341029442623e-05,
      "loss": 1.4558,
      "step": 18400
    },
    {
      "epoch": 0.8181939442443644,
      "grad_norm": 0.05864092707633972,
      "learning_rate": 8.666113586645485e-05,
      "loss": 1.4534,
      "step": 18402
    },
    {
      "epoch": 0.8182828687030368,
      "grad_norm": 0.057395271956920624,
      "learning_rate": 8.657889677251113e-05,
      "loss": 1.4517,
      "step": 18404
    },
    {
      "epoch": 0.8183717931617092,
      "grad_norm": 0.057702574878931046,
      "learning_rate": 8.649669301962532e-05,
      "loss": 1.46,
      "step": 18406
    },
    {
      "epoch": 0.8184607176203815,
      "grad_norm": 0.05797472596168518,
      "learning_rate": 8.64145246148243e-05,
      "loss": 1.4515,
      "step": 18408
    },
    {
      "epoch": 0.8185496420790539,
      "grad_norm": 0.058402277529239655,
      "learning_rate": 8.633239156513234e-05,
      "loss": 1.4523,
      "step": 18410
    },
    {
      "epoch": 0.8186385665377262,
      "grad_norm": 0.057839684188365936,
      "learning_rate": 8.625029387757066e-05,
      "loss": 1.4467,
      "step": 18412
    },
    {
      "epoch": 0.8187274909963985,
      "grad_norm": 0.0590512678027153,
      "learning_rate": 8.616823155915704e-05,
      "loss": 1.451,
      "step": 18414
    },
    {
      "epoch": 0.8188164154550709,
      "grad_norm": 0.05765726789832115,
      "learning_rate": 8.608620461690681e-05,
      "loss": 1.4472,
      "step": 18416
    },
    {
      "epoch": 0.8189053399137433,
      "grad_norm": 0.058935705572366714,
      "learning_rate": 8.600421305783168e-05,
      "loss": 1.4459,
      "step": 18418
    },
    {
      "epoch": 0.8189942643724156,
      "grad_norm": 0.05756261944770813,
      "learning_rate": 8.592225688894101e-05,
      "loss": 1.4481,
      "step": 18420
    },
    {
      "epoch": 0.819083188831088,
      "grad_norm": 0.05783841758966446,
      "learning_rate": 8.584033611724063e-05,
      "loss": 1.4493,
      "step": 18422
    },
    {
      "epoch": 0.8191721132897604,
      "grad_norm": 0.05876602604985237,
      "learning_rate": 8.575845074973337e-05,
      "loss": 1.4499,
      "step": 18424
    },
    {
      "epoch": 0.8192610377484327,
      "grad_norm": 0.05919130891561508,
      "learning_rate": 8.567660079341922e-05,
      "loss": 1.4478,
      "step": 18426
    },
    {
      "epoch": 0.8193499622071051,
      "grad_norm": 0.058414239436388016,
      "learning_rate": 8.559478625529521e-05,
      "loss": 1.4534,
      "step": 18428
    },
    {
      "epoch": 0.8194388866657775,
      "grad_norm": 0.05952535197138786,
      "learning_rate": 8.551300714235494e-05,
      "loss": 1.4541,
      "step": 18430
    },
    {
      "epoch": 0.8195278111244498,
      "grad_norm": 0.05861512944102287,
      "learning_rate": 8.543126346158947e-05,
      "loss": 1.4536,
      "step": 18432
    },
    {
      "epoch": 0.8196167355831221,
      "grad_norm": 0.059502918273210526,
      "learning_rate": 8.534955521998666e-05,
      "loss": 1.4562,
      "step": 18434
    },
    {
      "epoch": 0.8197056600417945,
      "grad_norm": 0.06032497435808182,
      "learning_rate": 8.526788242453098e-05,
      "loss": 1.4523,
      "step": 18436
    },
    {
      "epoch": 0.8197945845004668,
      "grad_norm": 0.05803174898028374,
      "learning_rate": 8.518624508220446e-05,
      "loss": 1.4503,
      "step": 18438
    },
    {
      "epoch": 0.8198835089591392,
      "grad_norm": 0.058706771582365036,
      "learning_rate": 8.510464319998568e-05,
      "loss": 1.4481,
      "step": 18440
    },
    {
      "epoch": 0.8199724334178116,
      "grad_norm": 0.059511151164770126,
      "learning_rate": 8.502307678485055e-05,
      "loss": 1.4465,
      "step": 18442
    },
    {
      "epoch": 0.8200613578764839,
      "grad_norm": 0.05901079997420311,
      "learning_rate": 8.494154584377156e-05,
      "loss": 1.4479,
      "step": 18444
    },
    {
      "epoch": 0.8201502823351563,
      "grad_norm": 0.05895798280835152,
      "learning_rate": 8.486005038371803e-05,
      "loss": 1.4492,
      "step": 18446
    },
    {
      "epoch": 0.8202392067938287,
      "grad_norm": 0.05760321766138077,
      "learning_rate": 8.477859041165714e-05,
      "loss": 1.4478,
      "step": 18448
    },
    {
      "epoch": 0.820328131252501,
      "grad_norm": 0.05806753411889076,
      "learning_rate": 8.469716593455218e-05,
      "loss": 1.4519,
      "step": 18450
    },
    {
      "epoch": 0.8204170557111734,
      "grad_norm": 0.06014678627252579,
      "learning_rate": 8.461577695936351e-05,
      "loss": 1.455,
      "step": 18452
    },
    {
      "epoch": 0.8205059801698457,
      "grad_norm": 0.05981629341840744,
      "learning_rate": 8.453442349304879e-05,
      "loss": 1.4504,
      "step": 18454
    },
    {
      "epoch": 0.820594904628518,
      "grad_norm": 0.05972183868288994,
      "learning_rate": 8.445310554256241e-05,
      "loss": 1.4487,
      "step": 18456
    },
    {
      "epoch": 0.8206838290871904,
      "grad_norm": 0.05913654714822769,
      "learning_rate": 8.437182311485597e-05,
      "loss": 1.4543,
      "step": 18458
    },
    {
      "epoch": 0.8207727535458628,
      "grad_norm": 0.05821114405989647,
      "learning_rate": 8.429057621687758e-05,
      "loss": 1.4494,
      "step": 18460
    },
    {
      "epoch": 0.8208616780045351,
      "grad_norm": 0.05849185958504677,
      "learning_rate": 8.420936485557268e-05,
      "loss": 1.45,
      "step": 18462
    },
    {
      "epoch": 0.8209506024632075,
      "grad_norm": 0.06026323139667511,
      "learning_rate": 8.412818903788377e-05,
      "loss": 1.4534,
      "step": 18464
    },
    {
      "epoch": 0.8210395269218799,
      "grad_norm": 0.058231696486473083,
      "learning_rate": 8.404704877074992e-05,
      "loss": 1.4521,
      "step": 18466
    },
    {
      "epoch": 0.8211284513805522,
      "grad_norm": 0.05949164554476738,
      "learning_rate": 8.396594406110713e-05,
      "loss": 1.4514,
      "step": 18468
    },
    {
      "epoch": 0.8212173758392246,
      "grad_norm": 0.05902651697397232,
      "learning_rate": 8.388487491588914e-05,
      "loss": 1.4489,
      "step": 18470
    },
    {
      "epoch": 0.821306300297897,
      "grad_norm": 0.05884255841374397,
      "learning_rate": 8.380384134202563e-05,
      "loss": 1.4476,
      "step": 18472
    },
    {
      "epoch": 0.8213952247565693,
      "grad_norm": 0.059297338128089905,
      "learning_rate": 8.372284334644404e-05,
      "loss": 1.4528,
      "step": 18474
    },
    {
      "epoch": 0.8214841492152416,
      "grad_norm": 0.05986732244491577,
      "learning_rate": 8.364188093606812e-05,
      "loss": 1.4511,
      "step": 18476
    },
    {
      "epoch": 0.821573073673914,
      "grad_norm": 0.06006040424108505,
      "learning_rate": 8.356095411781906e-05,
      "loss": 1.4573,
      "step": 18478
    },
    {
      "epoch": 0.8216619981325863,
      "grad_norm": 0.05927327647805214,
      "learning_rate": 8.348006289861493e-05,
      "loss": 1.4549,
      "step": 18480
    },
    {
      "epoch": 0.8217509225912587,
      "grad_norm": 0.05835961177945137,
      "learning_rate": 8.339920728537049e-05,
      "loss": 1.4518,
      "step": 18482
    },
    {
      "epoch": 0.8218398470499311,
      "grad_norm": 0.05824832245707512,
      "learning_rate": 8.331838728499768e-05,
      "loss": 1.4575,
      "step": 18484
    },
    {
      "epoch": 0.8219287715086034,
      "grad_norm": 0.05883040279150009,
      "learning_rate": 8.323760290440552e-05,
      "loss": 1.4548,
      "step": 18486
    },
    {
      "epoch": 0.8220176959672758,
      "grad_norm": 0.058843452483415604,
      "learning_rate": 8.31568541504995e-05,
      "loss": 1.4494,
      "step": 18488
    },
    {
      "epoch": 0.8221066204259482,
      "grad_norm": 0.05943866819143295,
      "learning_rate": 8.307614103018262e-05,
      "loss": 1.4532,
      "step": 18490
    },
    {
      "epoch": 0.8221955448846205,
      "grad_norm": 0.05836620554327965,
      "learning_rate": 8.299546355035465e-05,
      "loss": 1.4517,
      "step": 18492
    },
    {
      "epoch": 0.8222844693432929,
      "grad_norm": 0.05824948847293854,
      "learning_rate": 8.291482171791198e-05,
      "loss": 1.4466,
      "step": 18494
    },
    {
      "epoch": 0.8223733938019653,
      "grad_norm": 0.059235066175460815,
      "learning_rate": 8.28342155397484e-05,
      "loss": 1.4545,
      "step": 18496
    },
    {
      "epoch": 0.8224623182606375,
      "grad_norm": 0.058670178055763245,
      "learning_rate": 8.275364502275457e-05,
      "loss": 1.4537,
      "step": 18498
    },
    {
      "epoch": 0.8225512427193099,
      "grad_norm": 0.060267142951488495,
      "learning_rate": 8.26731101738178e-05,
      "loss": 1.4537,
      "step": 18500
    },
    {
      "epoch": 0.8225512427193099,
      "eval_loss": 1.4376198053359985,
      "eval_runtime": 12.4249,
      "eval_samples_per_second": 556.143,
      "eval_steps_per_second": 69.538,
      "step": 18500
    },
    {
      "epoch": 0.8226401671779823,
      "grad_norm": 0.058397773653268814,
      "learning_rate": 8.25926109998228e-05,
      "loss": 1.4565,
      "step": 18502
    },
    {
      "epoch": 0.8227290916366546,
      "grad_norm": 0.058602750301361084,
      "learning_rate": 8.251214750765073e-05,
      "loss": 1.4461,
      "step": 18504
    },
    {
      "epoch": 0.822818016095327,
      "grad_norm": 0.058638736605644226,
      "learning_rate": 8.243171970418012e-05,
      "loss": 1.4507,
      "step": 18506
    },
    {
      "epoch": 0.8229069405539994,
      "grad_norm": 0.05909786745905876,
      "learning_rate": 8.235132759628639e-05,
      "loss": 1.4492,
      "step": 18508
    },
    {
      "epoch": 0.8229958650126717,
      "grad_norm": 0.05795243754982948,
      "learning_rate": 8.22709711908416e-05,
      "loss": 1.451,
      "step": 18510
    },
    {
      "epoch": 0.8230847894713441,
      "grad_norm": 0.05899320915341377,
      "learning_rate": 8.219065049471503e-05,
      "loss": 1.4544,
      "step": 18512
    },
    {
      "epoch": 0.8231737139300165,
      "grad_norm": 0.05861734598875046,
      "learning_rate": 8.211036551477296e-05,
      "loss": 1.4559,
      "step": 18514
    },
    {
      "epoch": 0.8232626383886888,
      "grad_norm": 0.057675547897815704,
      "learning_rate": 8.203011625787838e-05,
      "loss": 1.4515,
      "step": 18516
    },
    {
      "epoch": 0.8233515628473612,
      "grad_norm": 0.05739520117640495,
      "learning_rate": 8.194990273089137e-05,
      "loss": 1.4536,
      "step": 18518
    },
    {
      "epoch": 0.8234404873060335,
      "grad_norm": 0.057842835783958435,
      "learning_rate": 8.186972494066907e-05,
      "loss": 1.4515,
      "step": 18520
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.05861688032746315,
      "learning_rate": 8.178958289406518e-05,
      "loss": 1.4518,
      "step": 18522
    },
    {
      "epoch": 0.8236183362233782,
      "grad_norm": 0.059118542820215225,
      "learning_rate": 8.170947659793093e-05,
      "loss": 1.4575,
      "step": 18524
    },
    {
      "epoch": 0.8237072606820506,
      "grad_norm": 0.057545505464076996,
      "learning_rate": 8.162940605911368e-05,
      "loss": 1.4463,
      "step": 18526
    },
    {
      "epoch": 0.823796185140723,
      "grad_norm": 0.05849096179008484,
      "learning_rate": 8.154937128445872e-05,
      "loss": 1.453,
      "step": 18528
    },
    {
      "epoch": 0.8238851095993953,
      "grad_norm": 0.05910954251885414,
      "learning_rate": 8.146937228080758e-05,
      "loss": 1.4583,
      "step": 18530
    },
    {
      "epoch": 0.8239740340580677,
      "grad_norm": 0.057638462632894516,
      "learning_rate": 8.13894090549988e-05,
      "loss": 1.4574,
      "step": 18532
    },
    {
      "epoch": 0.82406295851674,
      "grad_norm": 0.05978331342339516,
      "learning_rate": 8.130948161386808e-05,
      "loss": 1.4462,
      "step": 18534
    },
    {
      "epoch": 0.8241518829754124,
      "grad_norm": 0.058321740478277206,
      "learning_rate": 8.122958996424806e-05,
      "loss": 1.4468,
      "step": 18536
    },
    {
      "epoch": 0.8242408074340848,
      "grad_norm": 0.05788661167025566,
      "learning_rate": 8.11497341129681e-05,
      "loss": 1.4529,
      "step": 18538
    },
    {
      "epoch": 0.8243297318927572,
      "grad_norm": 0.0584048330783844,
      "learning_rate": 8.106991406685466e-05,
      "loss": 1.4579,
      "step": 18540
    },
    {
      "epoch": 0.8244186563514294,
      "grad_norm": 0.0584276020526886,
      "learning_rate": 8.099012983273107e-05,
      "loss": 1.4532,
      "step": 18542
    },
    {
      "epoch": 0.8245075808101018,
      "grad_norm": 0.059375640004873276,
      "learning_rate": 8.091038141741791e-05,
      "loss": 1.4545,
      "step": 18544
    },
    {
      "epoch": 0.8245965052687741,
      "grad_norm": 0.05816349759697914,
      "learning_rate": 8.083066882773205e-05,
      "loss": 1.4539,
      "step": 18546
    },
    {
      "epoch": 0.8246854297274465,
      "grad_norm": 0.06018855422735214,
      "learning_rate": 8.075099207048781e-05,
      "loss": 1.451,
      "step": 18548
    },
    {
      "epoch": 0.8247743541861189,
      "grad_norm": 0.05824248492717743,
      "learning_rate": 8.067135115249646e-05,
      "loss": 1.4551,
      "step": 18550
    },
    {
      "epoch": 0.8248632786447913,
      "grad_norm": 0.057999785989522934,
      "learning_rate": 8.059174608056597e-05,
      "loss": 1.4548,
      "step": 18552
    },
    {
      "epoch": 0.8249522031034636,
      "grad_norm": 0.058654844760894775,
      "learning_rate": 8.051217686150109e-05,
      "loss": 1.4579,
      "step": 18554
    },
    {
      "epoch": 0.825041127562136,
      "grad_norm": 0.059156231582164764,
      "learning_rate": 8.04326435021041e-05,
      "loss": 1.4535,
      "step": 18556
    },
    {
      "epoch": 0.8251300520208084,
      "grad_norm": 0.058144066482782364,
      "learning_rate": 8.035314600917365e-05,
      "loss": 1.4522,
      "step": 18558
    },
    {
      "epoch": 0.8252189764794807,
      "grad_norm": 0.057454727590084076,
      "learning_rate": 8.027368438950577e-05,
      "loss": 1.4521,
      "step": 18560
    },
    {
      "epoch": 0.8253079009381531,
      "grad_norm": 0.05804436281323433,
      "learning_rate": 8.019425864989283e-05,
      "loss": 1.4533,
      "step": 18562
    },
    {
      "epoch": 0.8253968253968254,
      "grad_norm": 0.0597645603120327,
      "learning_rate": 8.011486879712471e-05,
      "loss": 1.4472,
      "step": 18564
    },
    {
      "epoch": 0.8254857498554977,
      "grad_norm": 0.05968466028571129,
      "learning_rate": 8.00355148379881e-05,
      "loss": 1.4561,
      "step": 18566
    },
    {
      "epoch": 0.8255746743141701,
      "grad_norm": 0.05988946184515953,
      "learning_rate": 7.99561967792663e-05,
      "loss": 1.4529,
      "step": 18568
    },
    {
      "epoch": 0.8256635987728425,
      "grad_norm": 0.05921318754553795,
      "learning_rate": 7.987691462773983e-05,
      "loss": 1.4514,
      "step": 18570
    },
    {
      "epoch": 0.8257525232315148,
      "grad_norm": 0.05899718776345253,
      "learning_rate": 7.979766839018627e-05,
      "loss": 1.4482,
      "step": 18572
    },
    {
      "epoch": 0.8258414476901872,
      "grad_norm": 0.058463871479034424,
      "learning_rate": 7.971845807337979e-05,
      "loss": 1.4549,
      "step": 18574
    },
    {
      "epoch": 0.8259303721488596,
      "grad_norm": 0.059063419699668884,
      "learning_rate": 7.96392836840914e-05,
      "loss": 1.4508,
      "step": 18576
    },
    {
      "epoch": 0.8260192966075319,
      "grad_norm": 0.05859994888305664,
      "learning_rate": 7.956014522908972e-05,
      "loss": 1.4503,
      "step": 18578
    },
    {
      "epoch": 0.8261082210662043,
      "grad_norm": 0.05792148783802986,
      "learning_rate": 7.948104271513956e-05,
      "loss": 1.4562,
      "step": 18580
    },
    {
      "epoch": 0.8261971455248767,
      "grad_norm": 0.05839909240603447,
      "learning_rate": 7.940197614900314e-05,
      "loss": 1.4477,
      "step": 18582
    },
    {
      "epoch": 0.8262860699835489,
      "grad_norm": 0.05920722335577011,
      "learning_rate": 7.932294553743918e-05,
      "loss": 1.4532,
      "step": 18584
    },
    {
      "epoch": 0.8263749944422213,
      "grad_norm": 0.057403240352869034,
      "learning_rate": 7.924395088720371e-05,
      "loss": 1.4469,
      "step": 18586
    },
    {
      "epoch": 0.8264639189008937,
      "grad_norm": 0.058875951915979385,
      "learning_rate": 7.916499220504964e-05,
      "loss": 1.4541,
      "step": 18588
    },
    {
      "epoch": 0.826552843359566,
      "grad_norm": 0.05903790891170502,
      "learning_rate": 7.908606949772646e-05,
      "loss": 1.4532,
      "step": 18590
    },
    {
      "epoch": 0.8266417678182384,
      "grad_norm": 0.058501675724983215,
      "learning_rate": 7.900718277198099e-05,
      "loss": 1.4579,
      "step": 18592
    },
    {
      "epoch": 0.8267306922769108,
      "grad_norm": 0.06078590452671051,
      "learning_rate": 7.892833203455695e-05,
      "loss": 1.453,
      "step": 18594
    },
    {
      "epoch": 0.8268196167355831,
      "grad_norm": 0.05794868990778923,
      "learning_rate": 7.884951729219453e-05,
      "loss": 1.4482,
      "step": 18596
    },
    {
      "epoch": 0.8269085411942555,
      "grad_norm": 0.05853716656565666,
      "learning_rate": 7.87707385516313e-05,
      "loss": 1.454,
      "step": 18598
    },
    {
      "epoch": 0.8269974656529279,
      "grad_norm": 0.0581338033080101,
      "learning_rate": 7.86919958196018e-05,
      "loss": 1.455,
      "step": 18600
    },
    {
      "epoch": 0.8270863901116002,
      "grad_norm": 0.06062675267457962,
      "learning_rate": 7.8613289102837e-05,
      "loss": 1.4577,
      "step": 18602
    },
    {
      "epoch": 0.8271753145702726,
      "grad_norm": 0.05953441187739372,
      "learning_rate": 7.853461840806525e-05,
      "loss": 1.4478,
      "step": 18604
    },
    {
      "epoch": 0.8272642390289449,
      "grad_norm": 0.05867122486233711,
      "learning_rate": 7.845598374201174e-05,
      "loss": 1.4472,
      "step": 18606
    },
    {
      "epoch": 0.8273531634876172,
      "grad_norm": 0.05937234312295914,
      "learning_rate": 7.837738511139836e-05,
      "loss": 1.4522,
      "step": 18608
    },
    {
      "epoch": 0.8274420879462896,
      "grad_norm": 0.05736441910266876,
      "learning_rate": 7.829882252294423e-05,
      "loss": 1.4549,
      "step": 18610
    },
    {
      "epoch": 0.827531012404962,
      "grad_norm": 0.059804897755384445,
      "learning_rate": 7.8220295983365e-05,
      "loss": 1.4554,
      "step": 18612
    },
    {
      "epoch": 0.8276199368636343,
      "grad_norm": 0.058750297874212265,
      "learning_rate": 7.81418054993736e-05,
      "loss": 1.4478,
      "step": 18614
    },
    {
      "epoch": 0.8277088613223067,
      "grad_norm": 0.05871342122554779,
      "learning_rate": 7.806335107767981e-05,
      "loss": 1.446,
      "step": 18616
    },
    {
      "epoch": 0.8277977857809791,
      "grad_norm": 0.05835486575961113,
      "learning_rate": 7.798493272499013e-05,
      "loss": 1.444,
      "step": 18618
    },
    {
      "epoch": 0.8278867102396514,
      "grad_norm": 0.058702003210783005,
      "learning_rate": 7.790655044800815e-05,
      "loss": 1.4525,
      "step": 18620
    },
    {
      "epoch": 0.8279756346983238,
      "grad_norm": 0.058793142437934875,
      "learning_rate": 7.78282042534344e-05,
      "loss": 1.4532,
      "step": 18622
    },
    {
      "epoch": 0.8280645591569962,
      "grad_norm": 0.0574706606566906,
      "learning_rate": 7.774989414796612e-05,
      "loss": 1.4462,
      "step": 18624
    },
    {
      "epoch": 0.8281534836156685,
      "grad_norm": 0.0592108853161335,
      "learning_rate": 7.767162013829771e-05,
      "loss": 1.4489,
      "step": 18626
    },
    {
      "epoch": 0.8282424080743408,
      "grad_norm": 0.05900189280509949,
      "learning_rate": 7.759338223112044e-05,
      "loss": 1.4521,
      "step": 18628
    },
    {
      "epoch": 0.8283313325330132,
      "grad_norm": 0.058755598962306976,
      "learning_rate": 7.751518043312228e-05,
      "loss": 1.4563,
      "step": 18630
    },
    {
      "epoch": 0.8284202569916855,
      "grad_norm": 0.05981973186135292,
      "learning_rate": 7.743701475098835e-05,
      "loss": 1.4524,
      "step": 18632
    },
    {
      "epoch": 0.8285091814503579,
      "grad_norm": 0.058422692120075226,
      "learning_rate": 7.735888519140044e-05,
      "loss": 1.4499,
      "step": 18634
    },
    {
      "epoch": 0.8285981059090303,
      "grad_norm": 0.058538686484098434,
      "learning_rate": 7.728079176103769e-05,
      "loss": 1.4498,
      "step": 18636
    },
    {
      "epoch": 0.8286870303677026,
      "grad_norm": 0.05935661122202873,
      "learning_rate": 7.720273446657577e-05,
      "loss": 1.4577,
      "step": 18638
    },
    {
      "epoch": 0.828775954826375,
      "grad_norm": 0.0582767128944397,
      "learning_rate": 7.712471331468717e-05,
      "loss": 1.4507,
      "step": 18640
    },
    {
      "epoch": 0.8288648792850474,
      "grad_norm": 0.05873246118426323,
      "learning_rate": 7.704672831204168e-05,
      "loss": 1.458,
      "step": 18642
    },
    {
      "epoch": 0.8289538037437197,
      "grad_norm": 0.05824227258563042,
      "learning_rate": 7.696877946530584e-05,
      "loss": 1.4528,
      "step": 18644
    },
    {
      "epoch": 0.8290427282023921,
      "grad_norm": 0.05939396098256111,
      "learning_rate": 7.689086678114282e-05,
      "loss": 1.4556,
      "step": 18646
    },
    {
      "epoch": 0.8291316526610645,
      "grad_norm": 0.05786508694291115,
      "learning_rate": 7.681299026621307e-05,
      "loss": 1.4546,
      "step": 18648
    },
    {
      "epoch": 0.8292205771197367,
      "grad_norm": 0.059782709926366806,
      "learning_rate": 7.67351499271739e-05,
      "loss": 1.4482,
      "step": 18650
    },
    {
      "epoch": 0.8293095015784091,
      "grad_norm": 0.0591367743909359,
      "learning_rate": 7.665734577067951e-05,
      "loss": 1.4528,
      "step": 18652
    },
    {
      "epoch": 0.8293984260370815,
      "grad_norm": 0.05903034284710884,
      "learning_rate": 7.657957780338076e-05,
      "loss": 1.4533,
      "step": 18654
    },
    {
      "epoch": 0.8294873504957538,
      "grad_norm": 0.05796166509389877,
      "learning_rate": 7.650184603192544e-05,
      "loss": 1.4451,
      "step": 18656
    },
    {
      "epoch": 0.8295762749544262,
      "grad_norm": 0.05834323912858963,
      "learning_rate": 7.642415046295886e-05,
      "loss": 1.455,
      "step": 18658
    },
    {
      "epoch": 0.8296651994130986,
      "grad_norm": 0.05898818001151085,
      "learning_rate": 7.634649110312253e-05,
      "loss": 1.4545,
      "step": 18660
    },
    {
      "epoch": 0.8297541238717709,
      "grad_norm": 0.059623461216688156,
      "learning_rate": 7.626886795905497e-05,
      "loss": 1.4535,
      "step": 18662
    },
    {
      "epoch": 0.8298430483304433,
      "grad_norm": 0.05911053344607353,
      "learning_rate": 7.619128103739193e-05,
      "loss": 1.4496,
      "step": 18664
    },
    {
      "epoch": 0.8299319727891157,
      "grad_norm": 0.05885409191250801,
      "learning_rate": 7.611373034476587e-05,
      "loss": 1.452,
      "step": 18666
    },
    {
      "epoch": 0.830020897247788,
      "grad_norm": 0.05906490609049797,
      "learning_rate": 7.603621588780624e-05,
      "loss": 1.4513,
      "step": 18668
    },
    {
      "epoch": 0.8301098217064604,
      "grad_norm": 0.05751986429095268,
      "learning_rate": 7.59587376731391e-05,
      "loss": 1.4457,
      "step": 18670
    },
    {
      "epoch": 0.8301987461651327,
      "grad_norm": 0.05804996192455292,
      "learning_rate": 7.588129570738778e-05,
      "loss": 1.4479,
      "step": 18672
    },
    {
      "epoch": 0.830287670623805,
      "grad_norm": 0.060301125049591064,
      "learning_rate": 7.580388999717241e-05,
      "loss": 1.4481,
      "step": 18674
    },
    {
      "epoch": 0.8303765950824774,
      "grad_norm": 0.05793362483382225,
      "learning_rate": 7.572652054910984e-05,
      "loss": 1.4502,
      "step": 18676
    },
    {
      "epoch": 0.8304655195411498,
      "grad_norm": 0.059740450233221054,
      "learning_rate": 7.564918736981397e-05,
      "loss": 1.4541,
      "step": 18678
    },
    {
      "epoch": 0.8305544439998221,
      "grad_norm": 0.05822866037487984,
      "learning_rate": 7.557189046589575e-05,
      "loss": 1.4532,
      "step": 18680
    },
    {
      "epoch": 0.8306433684584945,
      "grad_norm": 0.05787384882569313,
      "learning_rate": 7.549462984396255e-05,
      "loss": 1.4513,
      "step": 18682
    },
    {
      "epoch": 0.8307322929171669,
      "grad_norm": 0.06009489670395851,
      "learning_rate": 7.541740551061916e-05,
      "loss": 1.453,
      "step": 18684
    },
    {
      "epoch": 0.8308212173758392,
      "grad_norm": 0.0584663450717926,
      "learning_rate": 7.534021747246717e-05,
      "loss": 1.4485,
      "step": 18686
    },
    {
      "epoch": 0.8309101418345116,
      "grad_norm": 0.05907716229557991,
      "learning_rate": 7.526306573610464e-05,
      "loss": 1.4501,
      "step": 18688
    },
    {
      "epoch": 0.830999066293184,
      "grad_norm": 0.058759596198797226,
      "learning_rate": 7.518595030812713e-05,
      "loss": 1.4551,
      "step": 18690
    },
    {
      "epoch": 0.8310879907518562,
      "grad_norm": 0.05805884301662445,
      "learning_rate": 7.51088711951266e-05,
      "loss": 1.4562,
      "step": 18692
    },
    {
      "epoch": 0.8311769152105286,
      "grad_norm": 0.05766487121582031,
      "learning_rate": 7.503182840369221e-05,
      "loss": 1.4534,
      "step": 18694
    },
    {
      "epoch": 0.831265839669201,
      "grad_norm": 0.05738608166575432,
      "learning_rate": 7.495482194040998e-05,
      "loss": 1.447,
      "step": 18696
    },
    {
      "epoch": 0.8313547641278733,
      "grad_norm": 0.058979667723178864,
      "learning_rate": 7.487785181186258e-05,
      "loss": 1.447,
      "step": 18698
    },
    {
      "epoch": 0.8314436885865457,
      "grad_norm": 0.05860074982047081,
      "learning_rate": 7.480091802462985e-05,
      "loss": 1.454,
      "step": 18700
    },
    {
      "epoch": 0.8315326130452181,
      "grad_norm": 0.05817517265677452,
      "learning_rate": 7.472402058528854e-05,
      "loss": 1.4527,
      "step": 18702
    },
    {
      "epoch": 0.8316215375038905,
      "grad_norm": 0.05893828719854355,
      "learning_rate": 7.464715950041202e-05,
      "loss": 1.4522,
      "step": 18704
    },
    {
      "epoch": 0.8317104619625628,
      "grad_norm": 0.058481235057115555,
      "learning_rate": 7.457033477657071e-05,
      "loss": 1.4537,
      "step": 18706
    },
    {
      "epoch": 0.8317993864212352,
      "grad_norm": 0.059741776436567307,
      "learning_rate": 7.449354642033212e-05,
      "loss": 1.455,
      "step": 18708
    },
    {
      "epoch": 0.8318883108799076,
      "grad_norm": 0.059183038771152496,
      "learning_rate": 7.441679443826021e-05,
      "loss": 1.454,
      "step": 18710
    },
    {
      "epoch": 0.8319772353385799,
      "grad_norm": 0.058914851397275925,
      "learning_rate": 7.434007883691635e-05,
      "loss": 1.4553,
      "step": 18712
    },
    {
      "epoch": 0.8320661597972522,
      "grad_norm": 0.05881735682487488,
      "learning_rate": 7.426339962285822e-05,
      "loss": 1.4568,
      "step": 18714
    },
    {
      "epoch": 0.8321550842559245,
      "grad_norm": 0.05866105481982231,
      "learning_rate": 7.418675680264087e-05,
      "loss": 1.4504,
      "step": 18716
    },
    {
      "epoch": 0.8322440087145969,
      "grad_norm": 0.06024124473333359,
      "learning_rate": 7.411015038281615e-05,
      "loss": 1.4505,
      "step": 18718
    },
    {
      "epoch": 0.8323329331732693,
      "grad_norm": 0.06013770401477814,
      "learning_rate": 7.403358036993252e-05,
      "loss": 1.4568,
      "step": 18720
    },
    {
      "epoch": 0.8324218576319417,
      "grad_norm": 0.058068759739398956,
      "learning_rate": 7.39570467705356e-05,
      "loss": 1.4542,
      "step": 18722
    },
    {
      "epoch": 0.832510782090614,
      "grad_norm": 0.05928467959165573,
      "learning_rate": 7.388054959116796e-05,
      "loss": 1.4483,
      "step": 18724
    },
    {
      "epoch": 0.8325997065492864,
      "grad_norm": 0.05854354798793793,
      "learning_rate": 7.380408883836864e-05,
      "loss": 1.4482,
      "step": 18726
    },
    {
      "epoch": 0.8326886310079588,
      "grad_norm": 0.05878078192472458,
      "learning_rate": 7.372766451867402e-05,
      "loss": 1.4453,
      "step": 18728
    },
    {
      "epoch": 0.8327775554666311,
      "grad_norm": 0.06043083965778351,
      "learning_rate": 7.365127663861726e-05,
      "loss": 1.4513,
      "step": 18730
    },
    {
      "epoch": 0.8328664799253035,
      "grad_norm": 0.05829674378037453,
      "learning_rate": 7.357492520472814e-05,
      "loss": 1.4538,
      "step": 18732
    },
    {
      "epoch": 0.8329554043839759,
      "grad_norm": 0.057941779494285583,
      "learning_rate": 7.349861022353355e-05,
      "loss": 1.4487,
      "step": 18734
    },
    {
      "epoch": 0.8330443288426481,
      "grad_norm": 0.05707291513681412,
      "learning_rate": 7.342233170155732e-05,
      "loss": 1.4506,
      "step": 18736
    },
    {
      "epoch": 0.8331332533013205,
      "grad_norm": 0.058747537434101105,
      "learning_rate": 7.334608964532019e-05,
      "loss": 1.4549,
      "step": 18738
    },
    {
      "epoch": 0.8332221777599929,
      "grad_norm": 0.05869237333536148,
      "learning_rate": 7.32698840613395e-05,
      "loss": 1.4558,
      "step": 18740
    },
    {
      "epoch": 0.8333111022186652,
      "grad_norm": 0.05862243101000786,
      "learning_rate": 7.319371495612942e-05,
      "loss": 1.4534,
      "step": 18742
    },
    {
      "epoch": 0.8334000266773376,
      "grad_norm": 0.059139467775821686,
      "learning_rate": 7.311758233620169e-05,
      "loss": 1.4485,
      "step": 18744
    },
    {
      "epoch": 0.83348895113601,
      "grad_norm": 0.05852329730987549,
      "learning_rate": 7.304148620806423e-05,
      "loss": 1.4489,
      "step": 18746
    },
    {
      "epoch": 0.8335778755946823,
      "grad_norm": 0.057609908282756805,
      "learning_rate": 7.296542657822197e-05,
      "loss": 1.4519,
      "step": 18748
    },
    {
      "epoch": 0.8336668000533547,
      "grad_norm": 0.059576258063316345,
      "learning_rate": 7.28894034531769e-05,
      "loss": 1.4488,
      "step": 18750
    },
    {
      "epoch": 0.8337557245120271,
      "grad_norm": 0.05912381783127785,
      "learning_rate": 7.281341683942799e-05,
      "loss": 1.4512,
      "step": 18752
    },
    {
      "epoch": 0.8338446489706994,
      "grad_norm": 0.05865836516022682,
      "learning_rate": 7.273746674347059e-05,
      "loss": 1.4459,
      "step": 18754
    },
    {
      "epoch": 0.8339335734293718,
      "grad_norm": 0.057585492730140686,
      "learning_rate": 7.266155317179746e-05,
      "loss": 1.4505,
      "step": 18756
    },
    {
      "epoch": 0.8340224978880441,
      "grad_norm": 0.057790596038103104,
      "learning_rate": 7.2585676130898e-05,
      "loss": 1.451,
      "step": 18758
    },
    {
      "epoch": 0.8341114223467164,
      "grad_norm": 0.06023659557104111,
      "learning_rate": 7.25098356272586e-05,
      "loss": 1.4554,
      "step": 18760
    },
    {
      "epoch": 0.8342003468053888,
      "grad_norm": 0.05900289863348007,
      "learning_rate": 7.243403166736229e-05,
      "loss": 1.4525,
      "step": 18762
    },
    {
      "epoch": 0.8342892712640612,
      "grad_norm": 0.059401899576187134,
      "learning_rate": 7.235826425768899e-05,
      "loss": 1.4531,
      "step": 18764
    },
    {
      "epoch": 0.8343781957227335,
      "grad_norm": 0.059451449662446976,
      "learning_rate": 7.228253340471608e-05,
      "loss": 1.4482,
      "step": 18766
    },
    {
      "epoch": 0.8344671201814059,
      "grad_norm": 0.058755919337272644,
      "learning_rate": 7.220683911491704e-05,
      "loss": 1.4484,
      "step": 18768
    },
    {
      "epoch": 0.8345560446400783,
      "grad_norm": 0.06049460545182228,
      "learning_rate": 7.213118139476255e-05,
      "loss": 1.4498,
      "step": 18770
    },
    {
      "epoch": 0.8346449690987506,
      "grad_norm": 0.058131422847509384,
      "learning_rate": 7.205556025072019e-05,
      "loss": 1.4494,
      "step": 18772
    },
    {
      "epoch": 0.834733893557423,
      "grad_norm": 0.057763516902923584,
      "learning_rate": 7.197997568925446e-05,
      "loss": 1.4524,
      "step": 18774
    },
    {
      "epoch": 0.8348228180160954,
      "grad_norm": 0.0582827590405941,
      "learning_rate": 7.19044277168267e-05,
      "loss": 1.4584,
      "step": 18776
    },
    {
      "epoch": 0.8349117424747677,
      "grad_norm": 0.05903775990009308,
      "learning_rate": 7.182891633989497e-05,
      "loss": 1.4507,
      "step": 18778
    },
    {
      "epoch": 0.83500066693344,
      "grad_norm": 0.05882865563035011,
      "learning_rate": 7.175344156491431e-05,
      "loss": 1.4472,
      "step": 18780
    },
    {
      "epoch": 0.8350895913921124,
      "grad_norm": 0.05889016017317772,
      "learning_rate": 7.16780033983368e-05,
      "loss": 1.4513,
      "step": 18782
    },
    {
      "epoch": 0.8351785158507847,
      "grad_norm": 0.05878911539912224,
      "learning_rate": 7.1602601846611e-05,
      "loss": 1.4541,
      "step": 18784
    },
    {
      "epoch": 0.8352674403094571,
      "grad_norm": 0.05862708389759064,
      "learning_rate": 7.152723691618268e-05,
      "loss": 1.4547,
      "step": 18786
    },
    {
      "epoch": 0.8353563647681295,
      "grad_norm": 0.057450249791145325,
      "learning_rate": 7.145190861349449e-05,
      "loss": 1.4471,
      "step": 18788
    },
    {
      "epoch": 0.8354452892268018,
      "grad_norm": 0.0574302151799202,
      "learning_rate": 7.137661694498559e-05,
      "loss": 1.4475,
      "step": 18790
    },
    {
      "epoch": 0.8355342136854742,
      "grad_norm": 0.05855082347989082,
      "learning_rate": 7.130136191709235e-05,
      "loss": 1.4506,
      "step": 18792
    },
    {
      "epoch": 0.8356231381441466,
      "grad_norm": 0.05884725600481033,
      "learning_rate": 7.122614353624807e-05,
      "loss": 1.4479,
      "step": 18794
    },
    {
      "epoch": 0.8357120626028189,
      "grad_norm": 0.0579058937728405,
      "learning_rate": 7.115096180888242e-05,
      "loss": 1.453,
      "step": 18796
    },
    {
      "epoch": 0.8358009870614913,
      "grad_norm": 0.0576477013528347,
      "learning_rate": 7.107581674142255e-05,
      "loss": 1.4533,
      "step": 18798
    },
    {
      "epoch": 0.8358899115201637,
      "grad_norm": 0.05866725370287895,
      "learning_rate": 7.100070834029198e-05,
      "loss": 1.4529,
      "step": 18800
    },
    {
      "epoch": 0.8359788359788359,
      "grad_norm": 0.059061404317617416,
      "learning_rate": 7.092563661191137e-05,
      "loss": 1.4497,
      "step": 18802
    },
    {
      "epoch": 0.8360677604375083,
      "grad_norm": 0.058392900973558426,
      "learning_rate": 7.085060156269835e-05,
      "loss": 1.4571,
      "step": 18804
    },
    {
      "epoch": 0.8361566848961807,
      "grad_norm": 0.05853873863816261,
      "learning_rate": 7.077560319906695e-05,
      "loss": 1.4481,
      "step": 18806
    },
    {
      "epoch": 0.836245609354853,
      "grad_norm": 0.058212365955114365,
      "learning_rate": 7.070064152742855e-05,
      "loss": 1.4575,
      "step": 18808
    },
    {
      "epoch": 0.8363345338135254,
      "grad_norm": 0.05963822454214096,
      "learning_rate": 7.062571655419132e-05,
      "loss": 1.449,
      "step": 18810
    },
    {
      "epoch": 0.8364234582721978,
      "grad_norm": 0.058400917798280716,
      "learning_rate": 7.055082828575987e-05,
      "loss": 1.4544,
      "step": 18812
    },
    {
      "epoch": 0.8365123827308701,
      "grad_norm": 0.058825619518756866,
      "learning_rate": 7.047597672853618e-05,
      "loss": 1.4532,
      "step": 18814
    },
    {
      "epoch": 0.8366013071895425,
      "grad_norm": 0.059138767421245575,
      "learning_rate": 7.040116188891893e-05,
      "loss": 1.4483,
      "step": 18816
    },
    {
      "epoch": 0.8366902316482149,
      "grad_norm": 0.058162230998277664,
      "learning_rate": 7.032638377330341e-05,
      "loss": 1.451,
      "step": 18818
    },
    {
      "epoch": 0.8367791561068872,
      "grad_norm": 0.05852394923567772,
      "learning_rate": 7.025164238808229e-05,
      "loss": 1.455,
      "step": 18820
    },
    {
      "epoch": 0.8368680805655595,
      "grad_norm": 0.058616820722818375,
      "learning_rate": 7.017693773964445e-05,
      "loss": 1.4485,
      "step": 18822
    },
    {
      "epoch": 0.8369570050242319,
      "grad_norm": 0.05944205820560455,
      "learning_rate": 7.010226983437617e-05,
      "loss": 1.4441,
      "step": 18824
    },
    {
      "epoch": 0.8370459294829042,
      "grad_norm": 0.058783505111932755,
      "learning_rate": 7.002763867866047e-05,
      "loss": 1.4517,
      "step": 18826
    },
    {
      "epoch": 0.8371348539415766,
      "grad_norm": 0.059438593685626984,
      "learning_rate": 6.995304427887688e-05,
      "loss": 1.4488,
      "step": 18828
    },
    {
      "epoch": 0.837223778400249,
      "grad_norm": 0.05818268656730652,
      "learning_rate": 6.987848664140229e-05,
      "loss": 1.4501,
      "step": 18830
    },
    {
      "epoch": 0.8373127028589213,
      "grad_norm": 0.05838819593191147,
      "learning_rate": 6.980396577261022e-05,
      "loss": 1.4559,
      "step": 18832
    },
    {
      "epoch": 0.8374016273175937,
      "grad_norm": 0.059087976813316345,
      "learning_rate": 6.972948167887088e-05,
      "loss": 1.4521,
      "step": 18834
    },
    {
      "epoch": 0.8374905517762661,
      "grad_norm": 0.05923278629779816,
      "learning_rate": 6.965503436655158e-05,
      "loss": 1.4543,
      "step": 18836
    },
    {
      "epoch": 0.8375794762349384,
      "grad_norm": 0.060685187578201294,
      "learning_rate": 6.958062384201652e-05,
      "loss": 1.4532,
      "step": 18838
    },
    {
      "epoch": 0.8376684006936108,
      "grad_norm": 0.058887116611003876,
      "learning_rate": 6.950625011162642e-05,
      "loss": 1.4534,
      "step": 18840
    },
    {
      "epoch": 0.8377573251522832,
      "grad_norm": 0.05865625664591789,
      "learning_rate": 6.943191318173919e-05,
      "loss": 1.448,
      "step": 18842
    },
    {
      "epoch": 0.8378462496109554,
      "grad_norm": 0.05893245339393616,
      "learning_rate": 6.935761305870946e-05,
      "loss": 1.4504,
      "step": 18844
    },
    {
      "epoch": 0.8379351740696278,
      "grad_norm": 0.05826593190431595,
      "learning_rate": 6.928334974888889e-05,
      "loss": 1.4481,
      "step": 18846
    },
    {
      "epoch": 0.8380240985283002,
      "grad_norm": 0.05828888714313507,
      "learning_rate": 6.920912325862566e-05,
      "loss": 1.445,
      "step": 18848
    },
    {
      "epoch": 0.8381130229869725,
      "grad_norm": 0.0586436428129673,
      "learning_rate": 6.913493359426475e-05,
      "loss": 1.4572,
      "step": 18850
    },
    {
      "epoch": 0.8382019474456449,
      "grad_norm": 0.05876269191503525,
      "learning_rate": 6.906078076214877e-05,
      "loss": 1.4558,
      "step": 18852
    },
    {
      "epoch": 0.8382908719043173,
      "grad_norm": 0.05837937071919441,
      "learning_rate": 6.898666476861631e-05,
      "loss": 1.4541,
      "step": 18854
    },
    {
      "epoch": 0.8383797963629896,
      "grad_norm": 0.058034468442201614,
      "learning_rate": 6.891258562000302e-05,
      "loss": 1.449,
      "step": 18856
    },
    {
      "epoch": 0.838468720821662,
      "grad_norm": 0.05731835216283798,
      "learning_rate": 6.883854332264167e-05,
      "loss": 1.4469,
      "step": 18858
    },
    {
      "epoch": 0.8385576452803344,
      "grad_norm": 0.05783449858427048,
      "learning_rate": 6.876453788286174e-05,
      "loss": 1.45,
      "step": 18860
    },
    {
      "epoch": 0.8386465697390068,
      "grad_norm": 0.058289241045713425,
      "learning_rate": 6.869056930698958e-05,
      "loss": 1.4564,
      "step": 18862
    },
    {
      "epoch": 0.8387354941976791,
      "grad_norm": 0.05786917731165886,
      "learning_rate": 6.861663760134818e-05,
      "loss": 1.4477,
      "step": 18864
    },
    {
      "epoch": 0.8388244186563514,
      "grad_norm": 0.05944148823618889,
      "learning_rate": 6.854274277225764e-05,
      "loss": 1.4545,
      "step": 18866
    },
    {
      "epoch": 0.8389133431150237,
      "grad_norm": 0.05793730542063713,
      "learning_rate": 6.846888482603492e-05,
      "loss": 1.4501,
      "step": 18868
    },
    {
      "epoch": 0.8390022675736961,
      "grad_norm": 0.05772128701210022,
      "learning_rate": 6.839506376899362e-05,
      "loss": 1.4506,
      "step": 18870
    },
    {
      "epoch": 0.8390911920323685,
      "grad_norm": 0.05777378007769585,
      "learning_rate": 6.832127960744405e-05,
      "loss": 1.4493,
      "step": 18872
    },
    {
      "epoch": 0.8391801164910409,
      "grad_norm": 0.058291662484407425,
      "learning_rate": 6.82475323476941e-05,
      "loss": 1.4474,
      "step": 18874
    },
    {
      "epoch": 0.8392690409497132,
      "grad_norm": 0.058231450617313385,
      "learning_rate": 6.81738219960476e-05,
      "loss": 1.4488,
      "step": 18876
    },
    {
      "epoch": 0.8393579654083856,
      "grad_norm": 0.057656459510326385,
      "learning_rate": 6.810014855880598e-05,
      "loss": 1.4476,
      "step": 18878
    },
    {
      "epoch": 0.839446889867058,
      "grad_norm": 0.057494621723890305,
      "learning_rate": 6.802651204226678e-05,
      "loss": 1.4471,
      "step": 18880
    },
    {
      "epoch": 0.8395358143257303,
      "grad_norm": 0.05767682567238808,
      "learning_rate": 6.795291245272506e-05,
      "loss": 1.4493,
      "step": 18882
    },
    {
      "epoch": 0.8396247387844027,
      "grad_norm": 0.05807575583457947,
      "learning_rate": 6.787934979647242e-05,
      "loss": 1.4505,
      "step": 18884
    },
    {
      "epoch": 0.8397136632430751,
      "grad_norm": 0.05979083850979805,
      "learning_rate": 6.780582407979712e-05,
      "loss": 1.4455,
      "step": 18886
    },
    {
      "epoch": 0.8398025877017473,
      "grad_norm": 0.05773860588669777,
      "learning_rate": 6.773233530898465e-05,
      "loss": 1.4442,
      "step": 18888
    },
    {
      "epoch": 0.8398915121604197,
      "grad_norm": 0.058765195310115814,
      "learning_rate": 6.765888349031718e-05,
      "loss": 1.4499,
      "step": 18890
    },
    {
      "epoch": 0.839980436619092,
      "grad_norm": 0.058183662593364716,
      "learning_rate": 6.758546863007353e-05,
      "loss": 1.4457,
      "step": 18892
    },
    {
      "epoch": 0.8400693610777644,
      "grad_norm": 0.058729760348796844,
      "learning_rate": 6.751209073452958e-05,
      "loss": 1.4453,
      "step": 18894
    },
    {
      "epoch": 0.8401582855364368,
      "grad_norm": 0.05806111916899681,
      "learning_rate": 6.743874980995813e-05,
      "loss": 1.4462,
      "step": 18896
    },
    {
      "epoch": 0.8402472099951092,
      "grad_norm": 0.05843853950500488,
      "learning_rate": 6.736544586262849e-05,
      "loss": 1.4506,
      "step": 18898
    },
    {
      "epoch": 0.8403361344537815,
      "grad_norm": 0.059424083679914474,
      "learning_rate": 6.729217889880712e-05,
      "loss": 1.4461,
      "step": 18900
    },
    {
      "epoch": 0.8404250589124539,
      "grad_norm": 0.058906249701976776,
      "learning_rate": 6.721894892475727e-05,
      "loss": 1.4515,
      "step": 18902
    },
    {
      "epoch": 0.8405139833711263,
      "grad_norm": 0.05890142172574997,
      "learning_rate": 6.714575594673871e-05,
      "loss": 1.4516,
      "step": 18904
    },
    {
      "epoch": 0.8406029078297986,
      "grad_norm": 0.05828310176730156,
      "learning_rate": 6.707259997100862e-05,
      "loss": 1.4501,
      "step": 18906
    },
    {
      "epoch": 0.840691832288471,
      "grad_norm": 0.0580887570977211,
      "learning_rate": 6.699948100382042e-05,
      "loss": 1.4513,
      "step": 18908
    },
    {
      "epoch": 0.8407807567471433,
      "grad_norm": 0.057612862437963486,
      "learning_rate": 6.692639905142477e-05,
      "loss": 1.4552,
      "step": 18910
    },
    {
      "epoch": 0.8408696812058156,
      "grad_norm": 0.06015538424253464,
      "learning_rate": 6.68533541200691e-05,
      "loss": 1.4516,
      "step": 18912
    },
    {
      "epoch": 0.840958605664488,
      "grad_norm": 0.06045129895210266,
      "learning_rate": 6.678034621599743e-05,
      "loss": 1.4497,
      "step": 18914
    },
    {
      "epoch": 0.8410475301231604,
      "grad_norm": 0.059288546442985535,
      "learning_rate": 6.670737534545097e-05,
      "loss": 1.4492,
      "step": 18916
    },
    {
      "epoch": 0.8411364545818327,
      "grad_norm": 0.05912403389811516,
      "learning_rate": 6.663444151466752e-05,
      "loss": 1.4524,
      "step": 18918
    },
    {
      "epoch": 0.8412253790405051,
      "grad_norm": 0.05908604711294174,
      "learning_rate": 6.656154472988174e-05,
      "loss": 1.4495,
      "step": 18920
    },
    {
      "epoch": 0.8413143034991775,
      "grad_norm": 0.05895715579390526,
      "learning_rate": 6.648868499732524e-05,
      "loss": 1.4498,
      "step": 18922
    },
    {
      "epoch": 0.8414032279578498,
      "grad_norm": 0.059338223189115524,
      "learning_rate": 6.641586232322644e-05,
      "loss": 1.4547,
      "step": 18924
    },
    {
      "epoch": 0.8414921524165222,
      "grad_norm": 0.05875730887055397,
      "learning_rate": 6.63430767138104e-05,
      "loss": 1.4521,
      "step": 18926
    },
    {
      "epoch": 0.8415810768751946,
      "grad_norm": 0.057817842811346054,
      "learning_rate": 6.627032817529926e-05,
      "loss": 1.4473,
      "step": 18928
    },
    {
      "epoch": 0.8416700013338669,
      "grad_norm": 0.0594828836619854,
      "learning_rate": 6.619761671391173e-05,
      "loss": 1.4581,
      "step": 18930
    },
    {
      "epoch": 0.8417589257925392,
      "grad_norm": 0.058293212205171585,
      "learning_rate": 6.61249423358638e-05,
      "loss": 1.4499,
      "step": 18932
    },
    {
      "epoch": 0.8418478502512116,
      "grad_norm": 0.05719943344593048,
      "learning_rate": 6.605230504736782e-05,
      "loss": 1.4485,
      "step": 18934
    },
    {
      "epoch": 0.8419367747098839,
      "grad_norm": 0.058276139199733734,
      "learning_rate": 6.597970485463306e-05,
      "loss": 1.4467,
      "step": 18936
    },
    {
      "epoch": 0.8420256991685563,
      "grad_norm": 0.057104870676994324,
      "learning_rate": 6.590714176386575e-05,
      "loss": 1.4477,
      "step": 18938
    },
    {
      "epoch": 0.8421146236272287,
      "grad_norm": 0.058028530329465866,
      "learning_rate": 6.58346157812691e-05,
      "loss": 1.4521,
      "step": 18940
    },
    {
      "epoch": 0.842203548085901,
      "grad_norm": 0.05872227996587753,
      "learning_rate": 6.576212691304267e-05,
      "loss": 1.4559,
      "step": 18942
    },
    {
      "epoch": 0.8422924725445734,
      "grad_norm": 0.05847456306219101,
      "learning_rate": 6.568967516538326e-05,
      "loss": 1.4534,
      "step": 18944
    },
    {
      "epoch": 0.8423813970032458,
      "grad_norm": 0.05918823927640915,
      "learning_rate": 6.561726054448447e-05,
      "loss": 1.4535,
      "step": 18946
    },
    {
      "epoch": 0.8424703214619181,
      "grad_norm": 0.05854053795337677,
      "learning_rate": 6.554488305653639e-05,
      "loss": 1.4515,
      "step": 18948
    },
    {
      "epoch": 0.8425592459205905,
      "grad_norm": 0.05881137028336525,
      "learning_rate": 6.54725427077264e-05,
      "loss": 1.4548,
      "step": 18950
    },
    {
      "epoch": 0.8426481703792628,
      "grad_norm": 0.058693382889032364,
      "learning_rate": 6.540023950423818e-05,
      "loss": 1.4498,
      "step": 18952
    },
    {
      "epoch": 0.8427370948379351,
      "grad_norm": 0.05968313291668892,
      "learning_rate": 6.532797345225288e-05,
      "loss": 1.4423,
      "step": 18954
    },
    {
      "epoch": 0.8428260192966075,
      "grad_norm": 0.05926063284277916,
      "learning_rate": 6.525574455794797e-05,
      "loss": 1.4563,
      "step": 18956
    },
    {
      "epoch": 0.8429149437552799,
      "grad_norm": 0.05805355682969093,
      "learning_rate": 6.518355282749777e-05,
      "loss": 1.4526,
      "step": 18958
    },
    {
      "epoch": 0.8430038682139522,
      "grad_norm": 0.05876849219202995,
      "learning_rate": 6.511139826707362e-05,
      "loss": 1.4477,
      "step": 18960
    },
    {
      "epoch": 0.8430927926726246,
      "grad_norm": 0.058109939098358154,
      "learning_rate": 6.503928088284377e-05,
      "loss": 1.4478,
      "step": 18962
    },
    {
      "epoch": 0.843181717131297,
      "grad_norm": 0.059168893843889236,
      "learning_rate": 6.496720068097283e-05,
      "loss": 1.4579,
      "step": 18964
    },
    {
      "epoch": 0.8432706415899693,
      "grad_norm": 0.05811697244644165,
      "learning_rate": 6.489515766762277e-05,
      "loss": 1.4523,
      "step": 18966
    },
    {
      "epoch": 0.8433595660486417,
      "grad_norm": 0.05842481181025505,
      "learning_rate": 6.482315184895205e-05,
      "loss": 1.4528,
      "step": 18968
    },
    {
      "epoch": 0.8434484905073141,
      "grad_norm": 0.059310369193553925,
      "learning_rate": 6.475118323111622e-05,
      "loss": 1.4513,
      "step": 18970
    },
    {
      "epoch": 0.8435374149659864,
      "grad_norm": 0.05840446799993515,
      "learning_rate": 6.467925182026718e-05,
      "loss": 1.4555,
      "step": 18972
    },
    {
      "epoch": 0.8436263394246587,
      "grad_norm": 0.05892022326588631,
      "learning_rate": 6.46073576225541e-05,
      "loss": 1.4511,
      "step": 18974
    },
    {
      "epoch": 0.8437152638833311,
      "grad_norm": 0.05745990201830864,
      "learning_rate": 6.45355006441229e-05,
      "loss": 1.4561,
      "step": 18976
    },
    {
      "epoch": 0.8438041883420034,
      "grad_norm": 0.058618173003196716,
      "learning_rate": 6.446368089111615e-05,
      "loss": 1.4544,
      "step": 18978
    },
    {
      "epoch": 0.8438931128006758,
      "grad_norm": 0.05895305797457695,
      "learning_rate": 6.439189836967307e-05,
      "loss": 1.4508,
      "step": 18980
    },
    {
      "epoch": 0.8439820372593482,
      "grad_norm": 0.058520711958408356,
      "learning_rate": 6.432015308593043e-05,
      "loss": 1.4521,
      "step": 18982
    },
    {
      "epoch": 0.8440709617180205,
      "grad_norm": 0.05948729068040848,
      "learning_rate": 6.424844504602095e-05,
      "loss": 1.4467,
      "step": 18984
    },
    {
      "epoch": 0.8441598861766929,
      "grad_norm": 0.058824650943279266,
      "learning_rate": 6.417677425607483e-05,
      "loss": 1.4447,
      "step": 18986
    },
    {
      "epoch": 0.8442488106353653,
      "grad_norm": 0.058834102004766464,
      "learning_rate": 6.410514072221852e-05,
      "loss": 1.4448,
      "step": 18988
    },
    {
      "epoch": 0.8443377350940376,
      "grad_norm": 0.05896035209298134,
      "learning_rate": 6.403354445057569e-05,
      "loss": 1.4515,
      "step": 18990
    },
    {
      "epoch": 0.84442665955271,
      "grad_norm": 0.0575827993452549,
      "learning_rate": 6.396198544726683e-05,
      "loss": 1.4469,
      "step": 18992
    },
    {
      "epoch": 0.8445155840113824,
      "grad_norm": 0.05748899281024933,
      "learning_rate": 6.38904637184089e-05,
      "loss": 1.4457,
      "step": 18994
    },
    {
      "epoch": 0.8446045084700546,
      "grad_norm": 0.057869840413331985,
      "learning_rate": 6.381897927011604e-05,
      "loss": 1.4436,
      "step": 18996
    },
    {
      "epoch": 0.844693432928727,
      "grad_norm": 0.05841681361198425,
      "learning_rate": 6.374753210849909e-05,
      "loss": 1.4493,
      "step": 18998
    },
    {
      "epoch": 0.8447823573873994,
      "grad_norm": 0.058826643973588943,
      "learning_rate": 6.367612223966551e-05,
      "loss": 1.4515,
      "step": 19000
    },
    {
      "epoch": 0.8447823573873994,
      "eval_loss": 1.4360281229019165,
      "eval_runtime": 12.4748,
      "eval_samples_per_second": 553.917,
      "eval_steps_per_second": 69.26,
      "step": 19000
    },
    {
      "epoch": 0.8448712818460717,
      "grad_norm": 0.059161681681871414,
      "learning_rate": 6.36047496697198e-05,
      "loss": 1.4485,
      "step": 19002
    },
    {
      "epoch": 0.8449602063047441,
      "grad_norm": 0.05967250093817711,
      "learning_rate": 6.353341440476334e-05,
      "loss": 1.4478,
      "step": 19004
    },
    {
      "epoch": 0.8450491307634165,
      "grad_norm": 0.05787471681833267,
      "learning_rate": 6.346211645089394e-05,
      "loss": 1.4505,
      "step": 19006
    },
    {
      "epoch": 0.8451380552220888,
      "grad_norm": 0.05900755897164345,
      "learning_rate": 6.339085581420672e-05,
      "loss": 1.4497,
      "step": 19008
    },
    {
      "epoch": 0.8452269796807612,
      "grad_norm": 0.05949883162975311,
      "learning_rate": 6.33196325007931e-05,
      "loss": 1.4547,
      "step": 19010
    },
    {
      "epoch": 0.8453159041394336,
      "grad_norm": 0.05793669447302818,
      "learning_rate": 6.324844651674171e-05,
      "loss": 1.4532,
      "step": 19012
    },
    {
      "epoch": 0.845404828598106,
      "grad_norm": 0.05827039107680321,
      "learning_rate": 6.317729786813786e-05,
      "loss": 1.455,
      "step": 19014
    },
    {
      "epoch": 0.8454937530567783,
      "grad_norm": 0.05869372561573982,
      "learning_rate": 6.310618656106354e-05,
      "loss": 1.4442,
      "step": 19016
    },
    {
      "epoch": 0.8455826775154506,
      "grad_norm": 0.05820537358522415,
      "learning_rate": 6.303511260159772e-05,
      "loss": 1.4527,
      "step": 19018
    },
    {
      "epoch": 0.8456716019741229,
      "grad_norm": 0.05905662849545479,
      "learning_rate": 6.296407599581622e-05,
      "loss": 1.4486,
      "step": 19020
    },
    {
      "epoch": 0.8457605264327953,
      "grad_norm": 0.05853661149740219,
      "learning_rate": 6.289307674979133e-05,
      "loss": 1.4539,
      "step": 19022
    },
    {
      "epoch": 0.8458494508914677,
      "grad_norm": 0.059646617621183395,
      "learning_rate": 6.282211486959255e-05,
      "loss": 1.4524,
      "step": 19024
    },
    {
      "epoch": 0.84593837535014,
      "grad_norm": 0.05904247611761093,
      "learning_rate": 6.275119036128602e-05,
      "loss": 1.4483,
      "step": 19026
    },
    {
      "epoch": 0.8460272998088124,
      "grad_norm": 0.05890771001577377,
      "learning_rate": 6.268030323093455e-05,
      "loss": 1.4495,
      "step": 19028
    },
    {
      "epoch": 0.8461162242674848,
      "grad_norm": 0.0586511567234993,
      "learning_rate": 6.260945348459801e-05,
      "loss": 1.4458,
      "step": 19030
    },
    {
      "epoch": 0.8462051487261572,
      "grad_norm": 0.057969868183135986,
      "learning_rate": 6.253864112833296e-05,
      "loss": 1.4514,
      "step": 19032
    },
    {
      "epoch": 0.8462940731848295,
      "grad_norm": 0.05988781899213791,
      "learning_rate": 6.246786616819261e-05,
      "loss": 1.45,
      "step": 19034
    },
    {
      "epoch": 0.8463829976435019,
      "grad_norm": 0.05796820670366287,
      "learning_rate": 6.239712861022728e-05,
      "loss": 1.4498,
      "step": 19036
    },
    {
      "epoch": 0.8464719221021743,
      "grad_norm": 0.059687115252017975,
      "learning_rate": 6.232642846048364e-05,
      "loss": 1.4522,
      "step": 19038
    },
    {
      "epoch": 0.8465608465608465,
      "grad_norm": 0.05780632048845291,
      "learning_rate": 6.225576572500591e-05,
      "loss": 1.4479,
      "step": 19040
    },
    {
      "epoch": 0.8466497710195189,
      "grad_norm": 0.05791749432682991,
      "learning_rate": 6.218514040983437e-05,
      "loss": 1.4547,
      "step": 19042
    },
    {
      "epoch": 0.8467386954781913,
      "grad_norm": 0.05878786742687225,
      "learning_rate": 6.211455252100635e-05,
      "loss": 1.4467,
      "step": 19044
    },
    {
      "epoch": 0.8468276199368636,
      "grad_norm": 0.058243002742528915,
      "learning_rate": 6.204400206455601e-05,
      "loss": 1.4504,
      "step": 19046
    },
    {
      "epoch": 0.846916544395536,
      "grad_norm": 0.0594029575586319,
      "learning_rate": 6.197348904651456e-05,
      "loss": 1.4464,
      "step": 19048
    },
    {
      "epoch": 0.8470054688542084,
      "grad_norm": 0.05850706622004509,
      "learning_rate": 6.190301347290944e-05,
      "loss": 1.4451,
      "step": 19050
    },
    {
      "epoch": 0.8470943933128807,
      "grad_norm": 0.059582531452178955,
      "learning_rate": 6.183257534976534e-05,
      "loss": 1.4492,
      "step": 19052
    },
    {
      "epoch": 0.8471833177715531,
      "grad_norm": 0.05976349487900734,
      "learning_rate": 6.176217468310368e-05,
      "loss": 1.4506,
      "step": 19054
    },
    {
      "epoch": 0.8472722422302255,
      "grad_norm": 0.05981292203068733,
      "learning_rate": 6.169181147894265e-05,
      "loss": 1.4549,
      "step": 19056
    },
    {
      "epoch": 0.8473611666888978,
      "grad_norm": 0.05852234736084938,
      "learning_rate": 6.162148574329717e-05,
      "loss": 1.4511,
      "step": 19058
    },
    {
      "epoch": 0.8474500911475701,
      "grad_norm": 0.05882321670651436,
      "learning_rate": 6.155119748217874e-05,
      "loss": 1.4511,
      "step": 19060
    },
    {
      "epoch": 0.8475390156062425,
      "grad_norm": 0.05775250121951103,
      "learning_rate": 6.148094670159632e-05,
      "loss": 1.4495,
      "step": 19062
    },
    {
      "epoch": 0.8476279400649148,
      "grad_norm": 0.058737363666296005,
      "learning_rate": 6.14107334075551e-05,
      "loss": 1.4487,
      "step": 19064
    },
    {
      "epoch": 0.8477168645235872,
      "grad_norm": 0.057975850999355316,
      "learning_rate": 6.134055760605706e-05,
      "loss": 1.4439,
      "step": 19066
    },
    {
      "epoch": 0.8478057889822596,
      "grad_norm": 0.05753524228930473,
      "learning_rate": 6.127041930310129e-05,
      "loss": 1.4506,
      "step": 19068
    },
    {
      "epoch": 0.8478947134409319,
      "grad_norm": 0.05863305553793907,
      "learning_rate": 6.12003185046836e-05,
      "loss": 1.4506,
      "step": 19070
    },
    {
      "epoch": 0.8479836378996043,
      "grad_norm": 0.058391932398080826,
      "learning_rate": 6.113025521679627e-05,
      "loss": 1.4492,
      "step": 19072
    },
    {
      "epoch": 0.8480725623582767,
      "grad_norm": 0.057830072939395905,
      "learning_rate": 6.106022944542878e-05,
      "loss": 1.4506,
      "step": 19074
    },
    {
      "epoch": 0.848161486816949,
      "grad_norm": 0.05919577181339264,
      "learning_rate": 6.0990241196567205e-05,
      "loss": 1.4546,
      "step": 19076
    },
    {
      "epoch": 0.8482504112756214,
      "grad_norm": 0.05908111110329628,
      "learning_rate": 6.092029047619463e-05,
      "loss": 1.454,
      "step": 19078
    },
    {
      "epoch": 0.8483393357342938,
      "grad_norm": 0.05926148593425751,
      "learning_rate": 6.0850377290290415e-05,
      "loss": 1.4512,
      "step": 19080
    },
    {
      "epoch": 0.848428260192966,
      "grad_norm": 0.05787745118141174,
      "learning_rate": 6.078050164483129e-05,
      "loss": 1.4463,
      "step": 19082
    },
    {
      "epoch": 0.8485171846516384,
      "grad_norm": 0.059534478932619095,
      "learning_rate": 6.071066354579047e-05,
      "loss": 1.4511,
      "step": 19084
    },
    {
      "epoch": 0.8486061091103108,
      "grad_norm": 0.058711905032396317,
      "learning_rate": 6.064086299913807e-05,
      "loss": 1.4487,
      "step": 19086
    },
    {
      "epoch": 0.8486950335689831,
      "grad_norm": 0.059324994683265686,
      "learning_rate": 6.057110001084071e-05,
      "loss": 1.4388,
      "step": 19088
    },
    {
      "epoch": 0.8487839580276555,
      "grad_norm": 0.05815599113702774,
      "learning_rate": 6.050137458686239e-05,
      "loss": 1.4477,
      "step": 19090
    },
    {
      "epoch": 0.8488728824863279,
      "grad_norm": 0.05865519493818283,
      "learning_rate": 6.0431686733163294e-05,
      "loss": 1.4548,
      "step": 19092
    },
    {
      "epoch": 0.8489618069450002,
      "grad_norm": 0.05810709670186043,
      "learning_rate": 6.036203645570082e-05,
      "loss": 1.4453,
      "step": 19094
    },
    {
      "epoch": 0.8490507314036726,
      "grad_norm": 0.058259494602680206,
      "learning_rate": 6.029242376042882e-05,
      "loss": 1.4519,
      "step": 19096
    },
    {
      "epoch": 0.849139655862345,
      "grad_norm": 0.05749960616230965,
      "learning_rate": 6.0222848653298135e-05,
      "loss": 1.4485,
      "step": 19098
    },
    {
      "epoch": 0.8492285803210173,
      "grad_norm": 0.05866747722029686,
      "learning_rate": 6.015331114025646e-05,
      "loss": 1.4471,
      "step": 19100
    },
    {
      "epoch": 0.8493175047796897,
      "grad_norm": 0.05797070264816284,
      "learning_rate": 6.008381122724804e-05,
      "loss": 1.4522,
      "step": 19102
    },
    {
      "epoch": 0.849406429238362,
      "grad_norm": 0.05796579271554947,
      "learning_rate": 6.001434892021407e-05,
      "loss": 1.4536,
      "step": 19104
    },
    {
      "epoch": 0.8494953536970343,
      "grad_norm": 0.05819402635097504,
      "learning_rate": 5.994492422509257e-05,
      "loss": 1.4491,
      "step": 19106
    },
    {
      "epoch": 0.8495842781557067,
      "grad_norm": 0.058709751814603806,
      "learning_rate": 5.9875537147818126e-05,
      "loss": 1.4468,
      "step": 19108
    },
    {
      "epoch": 0.8496732026143791,
      "grad_norm": 0.05779354274272919,
      "learning_rate": 5.9806187694322334e-05,
      "loss": 1.448,
      "step": 19110
    },
    {
      "epoch": 0.8497621270730514,
      "grad_norm": 0.058577634394168854,
      "learning_rate": 5.973687587053361e-05,
      "loss": 1.4512,
      "step": 19112
    },
    {
      "epoch": 0.8498510515317238,
      "grad_norm": 0.05885601416230202,
      "learning_rate": 5.966760168237684e-05,
      "loss": 1.4475,
      "step": 19114
    },
    {
      "epoch": 0.8499399759903962,
      "grad_norm": 0.05744828283786774,
      "learning_rate": 5.9598365135773994e-05,
      "loss": 1.4527,
      "step": 19116
    },
    {
      "epoch": 0.8500289004490685,
      "grad_norm": 0.05852948874235153,
      "learning_rate": 5.952916623664362e-05,
      "loss": 1.4536,
      "step": 19118
    },
    {
      "epoch": 0.8501178249077409,
      "grad_norm": 0.05883361026644707,
      "learning_rate": 5.946000499090121e-05,
      "loss": 1.455,
      "step": 19120
    },
    {
      "epoch": 0.8502067493664133,
      "grad_norm": 0.05857586860656738,
      "learning_rate": 5.939088140445903e-05,
      "loss": 1.4478,
      "step": 19122
    },
    {
      "epoch": 0.8502956738250856,
      "grad_norm": 0.0588909313082695,
      "learning_rate": 5.932179548322597e-05,
      "loss": 1.4553,
      "step": 19124
    },
    {
      "epoch": 0.8503845982837579,
      "grad_norm": 0.05820483714342117,
      "learning_rate": 5.925274723310781e-05,
      "loss": 1.4496,
      "step": 19126
    },
    {
      "epoch": 0.8504735227424303,
      "grad_norm": 0.0587213970720768,
      "learning_rate": 5.9183736660007204e-05,
      "loss": 1.4544,
      "step": 19128
    },
    {
      "epoch": 0.8505624472011026,
      "grad_norm": 0.05785829946398735,
      "learning_rate": 5.911476376982333e-05,
      "loss": 1.4548,
      "step": 19130
    },
    {
      "epoch": 0.850651371659775,
      "grad_norm": 0.05887683108448982,
      "learning_rate": 5.904582856845236e-05,
      "loss": 1.454,
      "step": 19132
    },
    {
      "epoch": 0.8507402961184474,
      "grad_norm": 0.05940135940909386,
      "learning_rate": 5.897693106178725e-05,
      "loss": 1.4566,
      "step": 19134
    },
    {
      "epoch": 0.8508292205771197,
      "grad_norm": 0.05797107517719269,
      "learning_rate": 5.8908071255717566e-05,
      "loss": 1.4525,
      "step": 19136
    },
    {
      "epoch": 0.8509181450357921,
      "grad_norm": 0.058985110372304916,
      "learning_rate": 5.8839249156129705e-05,
      "loss": 1.4494,
      "step": 19138
    },
    {
      "epoch": 0.8510070694944645,
      "grad_norm": 0.057915061712265015,
      "learning_rate": 5.8770464768907125e-05,
      "loss": 1.4463,
      "step": 19140
    },
    {
      "epoch": 0.8510959939531368,
      "grad_norm": 0.05887880176305771,
      "learning_rate": 5.870171809992952e-05,
      "loss": 1.4533,
      "step": 19142
    },
    {
      "epoch": 0.8511849184118092,
      "grad_norm": 0.057847775518894196,
      "learning_rate": 5.8633009155073915e-05,
      "loss": 1.4527,
      "step": 19144
    },
    {
      "epoch": 0.8512738428704816,
      "grad_norm": 0.05895482003688812,
      "learning_rate": 5.856433794021349e-05,
      "loss": 1.4462,
      "step": 19146
    },
    {
      "epoch": 0.8513627673291538,
      "grad_norm": 0.05755472183227539,
      "learning_rate": 5.849570446121905e-05,
      "loss": 1.4506,
      "step": 19148
    },
    {
      "epoch": 0.8514516917878262,
      "grad_norm": 0.05808274820446968,
      "learning_rate": 5.8427108723957403e-05,
      "loss": 1.454,
      "step": 19150
    },
    {
      "epoch": 0.8515406162464986,
      "grad_norm": 0.06018178537487984,
      "learning_rate": 5.835855073429236e-05,
      "loss": 1.4481,
      "step": 19152
    },
    {
      "epoch": 0.8516295407051709,
      "grad_norm": 0.05914274603128433,
      "learning_rate": 5.829003049808462e-05,
      "loss": 1.4529,
      "step": 19154
    },
    {
      "epoch": 0.8517184651638433,
      "grad_norm": 0.058473363518714905,
      "learning_rate": 5.8221548021191785e-05,
      "loss": 1.45,
      "step": 19156
    },
    {
      "epoch": 0.8518073896225157,
      "grad_norm": 0.05827651917934418,
      "learning_rate": 5.815310330946777e-05,
      "loss": 1.4485,
      "step": 19158
    },
    {
      "epoch": 0.851896314081188,
      "grad_norm": 0.05839327350258827,
      "learning_rate": 5.8084696368763565e-05,
      "loss": 1.4478,
      "step": 19160
    },
    {
      "epoch": 0.8519852385398604,
      "grad_norm": 0.059893328696489334,
      "learning_rate": 5.8016327204927054e-05,
      "loss": 1.4493,
      "step": 19162
    },
    {
      "epoch": 0.8520741629985328,
      "grad_norm": 0.05871112272143364,
      "learning_rate": 5.7947995823802724e-05,
      "loss": 1.4493,
      "step": 19164
    },
    {
      "epoch": 0.8521630874572051,
      "grad_norm": 0.0575588084757328,
      "learning_rate": 5.787970223123179e-05,
      "loss": 1.4478,
      "step": 19166
    },
    {
      "epoch": 0.8522520119158775,
      "grad_norm": 0.058595284819602966,
      "learning_rate": 5.781144643305203e-05,
      "loss": 1.4501,
      "step": 19168
    },
    {
      "epoch": 0.8523409363745498,
      "grad_norm": 0.05944250524044037,
      "learning_rate": 5.774322843509872e-05,
      "loss": 1.4566,
      "step": 19170
    },
    {
      "epoch": 0.8524298608332221,
      "grad_norm": 0.05856909975409508,
      "learning_rate": 5.767504824320319e-05,
      "loss": 1.4448,
      "step": 19172
    },
    {
      "epoch": 0.8525187852918945,
      "grad_norm": 0.0576825849711895,
      "learning_rate": 5.7606905863193725e-05,
      "loss": 1.4392,
      "step": 19174
    },
    {
      "epoch": 0.8526077097505669,
      "grad_norm": 0.05843258276581764,
      "learning_rate": 5.7538801300895507e-05,
      "loss": 1.4544,
      "step": 19176
    },
    {
      "epoch": 0.8526966342092392,
      "grad_norm": 0.05991073697805405,
      "learning_rate": 5.7470734562130364e-05,
      "loss": 1.4511,
      "step": 19178
    },
    {
      "epoch": 0.8527855586679116,
      "grad_norm": 0.05866760015487671,
      "learning_rate": 5.740270565271716e-05,
      "loss": 1.4554,
      "step": 19180
    },
    {
      "epoch": 0.852874483126584,
      "grad_norm": 0.05758367106318474,
      "learning_rate": 5.7334714578471004e-05,
      "loss": 1.4499,
      "step": 19182
    },
    {
      "epoch": 0.8529634075852564,
      "grad_norm": 0.058923859149217606,
      "learning_rate": 5.72667613452042e-05,
      "loss": 1.4569,
      "step": 19184
    },
    {
      "epoch": 0.8530523320439287,
      "grad_norm": 0.05793732777237892,
      "learning_rate": 5.719884595872582e-05,
      "loss": 1.4534,
      "step": 19186
    },
    {
      "epoch": 0.8531412565026011,
      "grad_norm": 0.05803541839122772,
      "learning_rate": 5.713096842484139e-05,
      "loss": 1.4461,
      "step": 19188
    },
    {
      "epoch": 0.8532301809612733,
      "grad_norm": 0.057840730994939804,
      "learning_rate": 5.7063128749353375e-05,
      "loss": 1.4472,
      "step": 19190
    },
    {
      "epoch": 0.8533191054199457,
      "grad_norm": 0.05936199426651001,
      "learning_rate": 5.69953269380612e-05,
      "loss": 1.4522,
      "step": 19192
    },
    {
      "epoch": 0.8534080298786181,
      "grad_norm": 0.059048447757959366,
      "learning_rate": 5.6927562996760715e-05,
      "loss": 1.4507,
      "step": 19194
    },
    {
      "epoch": 0.8534969543372904,
      "grad_norm": 0.057632967829704285,
      "learning_rate": 5.685983693124452e-05,
      "loss": 1.4467,
      "step": 19196
    },
    {
      "epoch": 0.8535858787959628,
      "grad_norm": 0.05861329287290573,
      "learning_rate": 5.679214874730249e-05,
      "loss": 1.4473,
      "step": 19198
    },
    {
      "epoch": 0.8536748032546352,
      "grad_norm": 0.05806206539273262,
      "learning_rate": 5.6724498450720695e-05,
      "loss": 1.4405,
      "step": 19200
    },
    {
      "epoch": 0.8537637277133076,
      "grad_norm": 0.05802812799811363,
      "learning_rate": 5.6656886047282305e-05,
      "loss": 1.4478,
      "step": 19202
    },
    {
      "epoch": 0.8538526521719799,
      "grad_norm": 0.058287039399147034,
      "learning_rate": 5.658931154276692e-05,
      "loss": 1.4474,
      "step": 19204
    },
    {
      "epoch": 0.8539415766306523,
      "grad_norm": 0.05866856500506401,
      "learning_rate": 5.652177494295124e-05,
      "loss": 1.4494,
      "step": 19206
    },
    {
      "epoch": 0.8540305010893247,
      "grad_norm": 0.05808721482753754,
      "learning_rate": 5.645427625360866e-05,
      "loss": 1.4472,
      "step": 19208
    },
    {
      "epoch": 0.854119425547997,
      "grad_norm": 0.05806644633412361,
      "learning_rate": 5.638681548050917e-05,
      "loss": 1.4471,
      "step": 19210
    },
    {
      "epoch": 0.8542083500066693,
      "grad_norm": 0.058878298848867416,
      "learning_rate": 5.6319392629419544e-05,
      "loss": 1.4508,
      "step": 19212
    },
    {
      "epoch": 0.8542972744653416,
      "grad_norm": 0.058544594794511795,
      "learning_rate": 5.6252007706103614e-05,
      "loss": 1.4456,
      "step": 19214
    },
    {
      "epoch": 0.854386198924014,
      "grad_norm": 0.059308234602212906,
      "learning_rate": 5.61846607163215e-05,
      "loss": 1.4529,
      "step": 19216
    },
    {
      "epoch": 0.8544751233826864,
      "grad_norm": 0.05778004601597786,
      "learning_rate": 5.6117351665830426e-05,
      "loss": 1.4538,
      "step": 19218
    },
    {
      "epoch": 0.8545640478413588,
      "grad_norm": 0.05761827155947685,
      "learning_rate": 5.605008056038435e-05,
      "loss": 1.4521,
      "step": 19220
    },
    {
      "epoch": 0.8546529723000311,
      "grad_norm": 0.058810003101825714,
      "learning_rate": 5.598284740573367e-05,
      "loss": 1.4511,
      "step": 19222
    },
    {
      "epoch": 0.8547418967587035,
      "grad_norm": 0.05870932340621948,
      "learning_rate": 5.5915652207626064e-05,
      "loss": 1.4459,
      "step": 19224
    },
    {
      "epoch": 0.8548308212173759,
      "grad_norm": 0.05845349654555321,
      "learning_rate": 5.584849497180544e-05,
      "loss": 1.4601,
      "step": 19226
    },
    {
      "epoch": 0.8549197456760482,
      "grad_norm": 0.05838220566511154,
      "learning_rate": 5.578137570401271e-05,
      "loss": 1.4566,
      "step": 19228
    },
    {
      "epoch": 0.8550086701347206,
      "grad_norm": 0.05828014388680458,
      "learning_rate": 5.5714294409985766e-05,
      "loss": 1.4478,
      "step": 19230
    },
    {
      "epoch": 0.855097594593393,
      "grad_norm": 0.058326125144958496,
      "learning_rate": 5.5647251095458707e-05,
      "loss": 1.4441,
      "step": 19232
    },
    {
      "epoch": 0.8551865190520652,
      "grad_norm": 0.05799197405576706,
      "learning_rate": 5.558024576616277e-05,
      "loss": 1.448,
      "step": 19234
    },
    {
      "epoch": 0.8552754435107376,
      "grad_norm": 0.058170199394226074,
      "learning_rate": 5.5513278427826105e-05,
      "loss": 1.451,
      "step": 19236
    },
    {
      "epoch": 0.85536436796941,
      "grad_norm": 0.05785010755062103,
      "learning_rate": 5.544634908617302e-05,
      "loss": 1.4515,
      "step": 19238
    },
    {
      "epoch": 0.8554532924280823,
      "grad_norm": 0.05972233787178993,
      "learning_rate": 5.5379457746925155e-05,
      "loss": 1.4452,
      "step": 19240
    },
    {
      "epoch": 0.8555422168867547,
      "grad_norm": 0.060089897364377975,
      "learning_rate": 5.5312604415800716e-05,
      "loss": 1.4574,
      "step": 19242
    },
    {
      "epoch": 0.8556311413454271,
      "grad_norm": 0.06087521091103554,
      "learning_rate": 5.5245789098514406e-05,
      "loss": 1.4495,
      "step": 19244
    },
    {
      "epoch": 0.8557200658040994,
      "grad_norm": 0.05851989984512329,
      "learning_rate": 5.517901180077811e-05,
      "loss": 1.4488,
      "step": 19246
    },
    {
      "epoch": 0.8558089902627718,
      "grad_norm": 0.05812680348753929,
      "learning_rate": 5.5112272528299925e-05,
      "loss": 1.4514,
      "step": 19248
    },
    {
      "epoch": 0.8558979147214442,
      "grad_norm": 0.058141183108091354,
      "learning_rate": 5.50455712867855e-05,
      "loss": 1.4451,
      "step": 19250
    },
    {
      "epoch": 0.8559868391801165,
      "grad_norm": 0.05929475277662277,
      "learning_rate": 5.497890808193645e-05,
      "loss": 1.4509,
      "step": 19252
    },
    {
      "epoch": 0.8560757636387889,
      "grad_norm": 0.05837177485227585,
      "learning_rate": 5.491228291945138e-05,
      "loss": 1.4526,
      "step": 19254
    },
    {
      "epoch": 0.8561646880974612,
      "grad_norm": 0.05776189640164375,
      "learning_rate": 5.4845695805025855e-05,
      "loss": 1.4493,
      "step": 19256
    },
    {
      "epoch": 0.8562536125561335,
      "grad_norm": 0.057521410286426544,
      "learning_rate": 5.4779146744352027e-05,
      "loss": 1.4415,
      "step": 19258
    },
    {
      "epoch": 0.8563425370148059,
      "grad_norm": 0.06062265858054161,
      "learning_rate": 5.4712635743118685e-05,
      "loss": 1.4444,
      "step": 19260
    },
    {
      "epoch": 0.8564314614734783,
      "grad_norm": 0.05821528658270836,
      "learning_rate": 5.464616280701151e-05,
      "loss": 1.4483,
      "step": 19262
    },
    {
      "epoch": 0.8565203859321506,
      "grad_norm": 0.05795830488204956,
      "learning_rate": 5.457972794171312e-05,
      "loss": 1.4485,
      "step": 19264
    },
    {
      "epoch": 0.856609310390823,
      "grad_norm": 0.057552520185709,
      "learning_rate": 5.451333115290236e-05,
      "loss": 1.4473,
      "step": 19266
    },
    {
      "epoch": 0.8566982348494954,
      "grad_norm": 0.05941027030348778,
      "learning_rate": 5.444697244625524e-05,
      "loss": 1.4486,
      "step": 19268
    },
    {
      "epoch": 0.8567871593081677,
      "grad_norm": 0.05889924615621567,
      "learning_rate": 5.4380651827444394e-05,
      "loss": 1.4476,
      "step": 19270
    },
    {
      "epoch": 0.8568760837668401,
      "grad_norm": 0.057859137654304504,
      "learning_rate": 5.431436930213934e-05,
      "loss": 1.4475,
      "step": 19272
    },
    {
      "epoch": 0.8569650082255125,
      "grad_norm": 0.058761779218912125,
      "learning_rate": 5.4248124876006054e-05,
      "loss": 1.4519,
      "step": 19274
    },
    {
      "epoch": 0.8570539326841848,
      "grad_norm": 0.05748540908098221,
      "learning_rate": 5.418191855470722e-05,
      "loss": 1.4431,
      "step": 19276
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.0578523650765419,
      "learning_rate": 5.411575034390287e-05,
      "loss": 1.45,
      "step": 19278
    },
    {
      "epoch": 0.8572317816015295,
      "grad_norm": 0.05837201699614525,
      "learning_rate": 5.404962024924909e-05,
      "loss": 1.4471,
      "step": 19280
    },
    {
      "epoch": 0.8573207060602018,
      "grad_norm": 0.05988498777151108,
      "learning_rate": 5.3983528276398966e-05,
      "loss": 1.4496,
      "step": 19282
    },
    {
      "epoch": 0.8574096305188742,
      "grad_norm": 0.058663882315158844,
      "learning_rate": 5.391747443100231e-05,
      "loss": 1.4462,
      "step": 19284
    },
    {
      "epoch": 0.8574985549775466,
      "grad_norm": 0.05759890750050545,
      "learning_rate": 5.385145871870578e-05,
      "loss": 1.4473,
      "step": 19286
    },
    {
      "epoch": 0.8575874794362189,
      "grad_norm": 0.05909644067287445,
      "learning_rate": 5.37854811451528e-05,
      "loss": 1.4538,
      "step": 19288
    },
    {
      "epoch": 0.8576764038948913,
      "grad_norm": 0.0579351969063282,
      "learning_rate": 5.3719541715983145e-05,
      "loss": 1.454,
      "step": 19290
    },
    {
      "epoch": 0.8577653283535637,
      "grad_norm": 0.05818416178226471,
      "learning_rate": 5.36536404368338e-05,
      "loss": 1.4482,
      "step": 19292
    },
    {
      "epoch": 0.857854252812236,
      "grad_norm": 0.05801917612552643,
      "learning_rate": 5.358777731333836e-05,
      "loss": 1.4485,
      "step": 19294
    },
    {
      "epoch": 0.8579431772709084,
      "grad_norm": 0.05826589837670326,
      "learning_rate": 5.3521952351126946e-05,
      "loss": 1.4471,
      "step": 19296
    },
    {
      "epoch": 0.8580321017295808,
      "grad_norm": 0.05808262154459953,
      "learning_rate": 5.3456165555826445e-05,
      "loss": 1.4514,
      "step": 19298
    },
    {
      "epoch": 0.858121026188253,
      "grad_norm": 0.058341771364212036,
      "learning_rate": 5.339041693306096e-05,
      "loss": 1.4544,
      "step": 19300
    },
    {
      "epoch": 0.8582099506469254,
      "grad_norm": 0.05911700800061226,
      "learning_rate": 5.332470648845067e-05,
      "loss": 1.4494,
      "step": 19302
    },
    {
      "epoch": 0.8582988751055978,
      "grad_norm": 0.0577431358397007,
      "learning_rate": 5.3259034227613026e-05,
      "loss": 1.4517,
      "step": 19304
    },
    {
      "epoch": 0.8583877995642701,
      "grad_norm": 0.059968046844005585,
      "learning_rate": 5.319340015616175e-05,
      "loss": 1.4513,
      "step": 19306
    },
    {
      "epoch": 0.8584767240229425,
      "grad_norm": 0.05771856755018234,
      "learning_rate": 5.312780427970765e-05,
      "loss": 1.4496,
      "step": 19308
    },
    {
      "epoch": 0.8585656484816149,
      "grad_norm": 0.05939215421676636,
      "learning_rate": 5.306224660385822e-05,
      "loss": 1.4509,
      "step": 19310
    },
    {
      "epoch": 0.8586545729402872,
      "grad_norm": 0.05893797054886818,
      "learning_rate": 5.299672713421744e-05,
      "loss": 1.4462,
      "step": 19312
    },
    {
      "epoch": 0.8587434973989596,
      "grad_norm": 0.05804156884551048,
      "learning_rate": 5.2931245876386314e-05,
      "loss": 1.4448,
      "step": 19314
    },
    {
      "epoch": 0.858832421857632,
      "grad_norm": 0.058172598481178284,
      "learning_rate": 5.286580283596254e-05,
      "loss": 1.4499,
      "step": 19316
    },
    {
      "epoch": 0.8589213463163043,
      "grad_norm": 0.05792252719402313,
      "learning_rate": 5.280039801854036e-05,
      "loss": 1.4474,
      "step": 19318
    },
    {
      "epoch": 0.8590102707749766,
      "grad_norm": 0.05791470780968666,
      "learning_rate": 5.273503142971081e-05,
      "loss": 1.4542,
      "step": 19320
    },
    {
      "epoch": 0.859099195233649,
      "grad_norm": 0.05813904106616974,
      "learning_rate": 5.266970307506197e-05,
      "loss": 1.4486,
      "step": 19322
    },
    {
      "epoch": 0.8591881196923213,
      "grad_norm": 0.057610899209976196,
      "learning_rate": 5.260441296017815e-05,
      "loss": 1.4461,
      "step": 19324
    },
    {
      "epoch": 0.8592770441509937,
      "grad_norm": 0.05858998745679855,
      "learning_rate": 5.2539161090640664e-05,
      "loss": 1.4518,
      "step": 19326
    },
    {
      "epoch": 0.8593659686096661,
      "grad_norm": 0.05812988802790642,
      "learning_rate": 5.247394747202772e-05,
      "loss": 1.4471,
      "step": 19328
    },
    {
      "epoch": 0.8594548930683384,
      "grad_norm": 0.05788353830575943,
      "learning_rate": 5.2408772109913796e-05,
      "loss": 1.4433,
      "step": 19330
    },
    {
      "epoch": 0.8595438175270108,
      "grad_norm": 0.057957880198955536,
      "learning_rate": 5.2343635009870604e-05,
      "loss": 1.4528,
      "step": 19332
    },
    {
      "epoch": 0.8596327419856832,
      "grad_norm": 0.05941891297698021,
      "learning_rate": 5.22785361774662e-05,
      "loss": 1.456,
      "step": 19334
    },
    {
      "epoch": 0.8597216664443555,
      "grad_norm": 0.059053994715213776,
      "learning_rate": 5.2213475618265505e-05,
      "loss": 1.4444,
      "step": 19336
    },
    {
      "epoch": 0.8598105909030279,
      "grad_norm": 0.05831526964902878,
      "learning_rate": 5.214845333783036e-05,
      "loss": 1.4422,
      "step": 19338
    },
    {
      "epoch": 0.8598995153617003,
      "grad_norm": 0.05919569730758667,
      "learning_rate": 5.208346934171898e-05,
      "loss": 1.4492,
      "step": 19340
    },
    {
      "epoch": 0.8599884398203725,
      "grad_norm": 0.057435665279626846,
      "learning_rate": 5.2018523635486536e-05,
      "loss": 1.4433,
      "step": 19342
    },
    {
      "epoch": 0.8600773642790449,
      "grad_norm": 0.059063155204057693,
      "learning_rate": 5.1953616224684916e-05,
      "loss": 1.4567,
      "step": 19344
    },
    {
      "epoch": 0.8601662887377173,
      "grad_norm": 0.0575404167175293,
      "learning_rate": 5.188874711486263e-05,
      "loss": 1.4542,
      "step": 19346
    },
    {
      "epoch": 0.8602552131963896,
      "grad_norm": 0.05753824859857559,
      "learning_rate": 5.182391631156497e-05,
      "loss": 1.4458,
      "step": 19348
    },
    {
      "epoch": 0.860344137655062,
      "grad_norm": 0.05860580503940582,
      "learning_rate": 5.1759123820334096e-05,
      "loss": 1.4482,
      "step": 19350
    },
    {
      "epoch": 0.8604330621137344,
      "grad_norm": 0.057950492948293686,
      "learning_rate": 5.16943696467086e-05,
      "loss": 1.44,
      "step": 19352
    },
    {
      "epoch": 0.8605219865724067,
      "grad_norm": 0.0587628073990345,
      "learning_rate": 5.162965379622403e-05,
      "loss": 1.4523,
      "step": 19354
    },
    {
      "epoch": 0.8606109110310791,
      "grad_norm": 0.05854136124253273,
      "learning_rate": 5.156497627441242e-05,
      "loss": 1.4526,
      "step": 19356
    },
    {
      "epoch": 0.8606998354897515,
      "grad_norm": 0.05751117691397667,
      "learning_rate": 5.150033708680296e-05,
      "loss": 1.4459,
      "step": 19358
    },
    {
      "epoch": 0.8607887599484239,
      "grad_norm": 0.0575980618596077,
      "learning_rate": 5.1435736238921226e-05,
      "loss": 1.4479,
      "step": 19360
    },
    {
      "epoch": 0.8608776844070962,
      "grad_norm": 0.05835995450615883,
      "learning_rate": 5.137117373628936e-05,
      "loss": 1.4543,
      "step": 19362
    },
    {
      "epoch": 0.8609666088657685,
      "grad_norm": 0.057942815124988556,
      "learning_rate": 5.1306649584426654e-05,
      "loss": 1.4561,
      "step": 19364
    },
    {
      "epoch": 0.8610555333244408,
      "grad_norm": 0.05824165791273117,
      "learning_rate": 5.124216378884894e-05,
      "loss": 1.4495,
      "step": 19366
    },
    {
      "epoch": 0.8611444577831132,
      "grad_norm": 0.05746667832136154,
      "learning_rate": 5.117771635506857e-05,
      "loss": 1.4532,
      "step": 19368
    },
    {
      "epoch": 0.8612333822417856,
      "grad_norm": 0.05848689749836922,
      "learning_rate": 5.111330728859492e-05,
      "loss": 1.4543,
      "step": 19370
    },
    {
      "epoch": 0.861322306700458,
      "grad_norm": 0.057919543236494064,
      "learning_rate": 5.104893659493392e-05,
      "loss": 1.4482,
      "step": 19372
    },
    {
      "epoch": 0.8614112311591303,
      "grad_norm": 0.058501534163951874,
      "learning_rate": 5.0984604279588456e-05,
      "loss": 1.4445,
      "step": 19374
    },
    {
      "epoch": 0.8615001556178027,
      "grad_norm": 0.05916326865553856,
      "learning_rate": 5.092031034805761e-05,
      "loss": 1.4471,
      "step": 19376
    },
    {
      "epoch": 0.8615890800764751,
      "grad_norm": 0.05860411003232002,
      "learning_rate": 5.085605480583766e-05,
      "loss": 1.4482,
      "step": 19378
    },
    {
      "epoch": 0.8616780045351474,
      "grad_norm": 0.059459760785102844,
      "learning_rate": 5.0791837658421544e-05,
      "loss": 1.4528,
      "step": 19380
    },
    {
      "epoch": 0.8617669289938198,
      "grad_norm": 0.059411801397800446,
      "learning_rate": 5.072765891129877e-05,
      "loss": 1.4544,
      "step": 19382
    },
    {
      "epoch": 0.8618558534524922,
      "grad_norm": 0.05691593512892723,
      "learning_rate": 5.066351856995538e-05,
      "loss": 1.4445,
      "step": 19384
    },
    {
      "epoch": 0.8619447779111644,
      "grad_norm": 0.05798202380537987,
      "learning_rate": 5.0599416639874776e-05,
      "loss": 1.4533,
      "step": 19386
    },
    {
      "epoch": 0.8620337023698368,
      "grad_norm": 0.058786388486623764,
      "learning_rate": 5.053535312653651e-05,
      "loss": 1.4485,
      "step": 19388
    },
    {
      "epoch": 0.8621226268285092,
      "grad_norm": 0.05767536908388138,
      "learning_rate": 5.0471328035416876e-05,
      "loss": 1.4481,
      "step": 19390
    },
    {
      "epoch": 0.8622115512871815,
      "grad_norm": 0.05754457041621208,
      "learning_rate": 5.04073413719891e-05,
      "loss": 1.4473,
      "step": 19392
    },
    {
      "epoch": 0.8623004757458539,
      "grad_norm": 0.05914188176393509,
      "learning_rate": 5.034339314172309e-05,
      "loss": 1.4438,
      "step": 19394
    },
    {
      "epoch": 0.8623894002045263,
      "grad_norm": 0.058261413127183914,
      "learning_rate": 5.027948335008553e-05,
      "loss": 1.453,
      "step": 19396
    },
    {
      "epoch": 0.8624783246631986,
      "grad_norm": 0.05782920494675636,
      "learning_rate": 5.021561200253943e-05,
      "loss": 1.4516,
      "step": 19398
    },
    {
      "epoch": 0.862567249121871,
      "grad_norm": 0.05883567035198212,
      "learning_rate": 5.015177910454499e-05,
      "loss": 1.4472,
      "step": 19400
    },
    {
      "epoch": 0.8626561735805434,
      "grad_norm": 0.05863523483276367,
      "learning_rate": 5.0087984661559005e-05,
      "loss": 1.4518,
      "step": 19402
    },
    {
      "epoch": 0.8627450980392157,
      "grad_norm": 0.05786546319723129,
      "learning_rate": 5.0024228679034775e-05,
      "loss": 1.4483,
      "step": 19404
    },
    {
      "epoch": 0.8628340224978881,
      "grad_norm": 0.058120548725128174,
      "learning_rate": 4.996051116242228e-05,
      "loss": 1.4442,
      "step": 19406
    },
    {
      "epoch": 0.8629229469565604,
      "grad_norm": 0.05752819404006004,
      "learning_rate": 4.989683211716872e-05,
      "loss": 1.4516,
      "step": 19408
    },
    {
      "epoch": 0.8630118714152327,
      "grad_norm": 0.05808071419596672,
      "learning_rate": 4.983319154871741e-05,
      "loss": 1.4465,
      "step": 19410
    },
    {
      "epoch": 0.8631007958739051,
      "grad_norm": 0.05908680707216263,
      "learning_rate": 4.9769589462508756e-05,
      "loss": 1.449,
      "step": 19412
    },
    {
      "epoch": 0.8631897203325775,
      "grad_norm": 0.0582328736782074,
      "learning_rate": 4.9706025863979654e-05,
      "loss": 1.4444,
      "step": 19414
    },
    {
      "epoch": 0.8632786447912498,
      "grad_norm": 0.058734990656375885,
      "learning_rate": 4.96425007585638e-05,
      "loss": 1.4558,
      "step": 19416
    },
    {
      "epoch": 0.8633675692499222,
      "grad_norm": 0.05865876004099846,
      "learning_rate": 4.95790141516918e-05,
      "loss": 1.4477,
      "step": 19418
    },
    {
      "epoch": 0.8634564937085946,
      "grad_norm": 0.05872330814599991,
      "learning_rate": 4.9515566048790485e-05,
      "loss": 1.4483,
      "step": 19420
    },
    {
      "epoch": 0.8635454181672669,
      "grad_norm": 0.05882033705711365,
      "learning_rate": 4.945215645528378e-05,
      "loss": 1.446,
      "step": 19422
    },
    {
      "epoch": 0.8636343426259393,
      "grad_norm": 0.05818769335746765,
      "learning_rate": 4.938878537659236e-05,
      "loss": 1.4462,
      "step": 19424
    },
    {
      "epoch": 0.8637232670846117,
      "grad_norm": 0.05744488164782524,
      "learning_rate": 4.932545281813322e-05,
      "loss": 1.448,
      "step": 19426
    },
    {
      "epoch": 0.8638121915432839,
      "grad_norm": 0.0585269033908844,
      "learning_rate": 4.9262158785320474e-05,
      "loss": 1.4513,
      "step": 19428
    },
    {
      "epoch": 0.8639011160019563,
      "grad_norm": 0.057957399636507034,
      "learning_rate": 4.91989032835648e-05,
      "loss": 1.4463,
      "step": 19430
    },
    {
      "epoch": 0.8639900404606287,
      "grad_norm": 0.058509595692157745,
      "learning_rate": 4.913568631827342e-05,
      "loss": 1.451,
      "step": 19432
    },
    {
      "epoch": 0.864078964919301,
      "grad_norm": 0.05889742821455002,
      "learning_rate": 4.9072507894850404e-05,
      "loss": 1.4461,
      "step": 19434
    },
    {
      "epoch": 0.8641678893779734,
      "grad_norm": 0.05817759409546852,
      "learning_rate": 4.9009368018696766e-05,
      "loss": 1.4507,
      "step": 19436
    },
    {
      "epoch": 0.8642568138366458,
      "grad_norm": 0.05786820873618126,
      "learning_rate": 4.8946266695209626e-05,
      "loss": 1.4559,
      "step": 19438
    },
    {
      "epoch": 0.8643457382953181,
      "grad_norm": 0.05808315426111221,
      "learning_rate": 4.8883203929783514e-05,
      "loss": 1.4505,
      "step": 19440
    },
    {
      "epoch": 0.8644346627539905,
      "grad_norm": 0.058381590992212296,
      "learning_rate": 4.8820179727808886e-05,
      "loss": 1.444,
      "step": 19442
    },
    {
      "epoch": 0.8645235872126629,
      "grad_norm": 0.05841744318604469,
      "learning_rate": 4.875719409467377e-05,
      "loss": 1.4468,
      "step": 19444
    },
    {
      "epoch": 0.8646125116713352,
      "grad_norm": 0.058670420199632645,
      "learning_rate": 4.869424703576231e-05,
      "loss": 1.4511,
      "step": 19446
    },
    {
      "epoch": 0.8647014361300076,
      "grad_norm": 0.05914199724793434,
      "learning_rate": 4.8631338556455354e-05,
      "loss": 1.4477,
      "step": 19448
    },
    {
      "epoch": 0.8647903605886799,
      "grad_norm": 0.05767590180039406,
      "learning_rate": 4.856846866213066e-05,
      "loss": 1.4442,
      "step": 19450
    },
    {
      "epoch": 0.8648792850473522,
      "grad_norm": 0.05876242741942406,
      "learning_rate": 4.8505637358162765e-05,
      "loss": 1.4517,
      "step": 19452
    },
    {
      "epoch": 0.8649682095060246,
      "grad_norm": 0.0582275465130806,
      "learning_rate": 4.8442844649922604e-05,
      "loss": 1.4491,
      "step": 19454
    },
    {
      "epoch": 0.865057133964697,
      "grad_norm": 0.058131519705057144,
      "learning_rate": 4.8380090542777974e-05,
      "loss": 1.4483,
      "step": 19456
    },
    {
      "epoch": 0.8651460584233693,
      "grad_norm": 0.05899100378155708,
      "learning_rate": 4.831737504209355e-05,
      "loss": 1.4476,
      "step": 19458
    },
    {
      "epoch": 0.8652349828820417,
      "grad_norm": 0.05828023701906204,
      "learning_rate": 4.8254698153230305e-05,
      "loss": 1.4483,
      "step": 19460
    },
    {
      "epoch": 0.8653239073407141,
      "grad_norm": 0.05877317860722542,
      "learning_rate": 4.8192059881546345e-05,
      "loss": 1.4492,
      "step": 19462
    },
    {
      "epoch": 0.8654128317993864,
      "grad_norm": 0.05795292928814888,
      "learning_rate": 4.812946023239595e-05,
      "loss": 1.448,
      "step": 19464
    },
    {
      "epoch": 0.8655017562580588,
      "grad_norm": 0.05792052298784256,
      "learning_rate": 4.806689921113083e-05,
      "loss": 1.4499,
      "step": 19466
    },
    {
      "epoch": 0.8655906807167312,
      "grad_norm": 0.05851710960268974,
      "learning_rate": 4.800437682309877e-05,
      "loss": 1.4556,
      "step": 19468
    },
    {
      "epoch": 0.8656796051754035,
      "grad_norm": 0.057633742690086365,
      "learning_rate": 4.794189307364433e-05,
      "loss": 1.4414,
      "step": 19470
    },
    {
      "epoch": 0.8657685296340758,
      "grad_norm": 0.058146603405475616,
      "learning_rate": 4.787944796810906e-05,
      "loss": 1.4447,
      "step": 19472
    },
    {
      "epoch": 0.8658574540927482,
      "grad_norm": 0.058894164860248566,
      "learning_rate": 4.7817041511831136e-05,
      "loss": 1.4552,
      "step": 19474
    },
    {
      "epoch": 0.8659463785514205,
      "grad_norm": 0.0587955042719841,
      "learning_rate": 4.775467371014508e-05,
      "loss": 1.4472,
      "step": 19476
    },
    {
      "epoch": 0.8660353030100929,
      "grad_norm": 0.05853498354554176,
      "learning_rate": 4.769234456838245e-05,
      "loss": 1.4491,
      "step": 19478
    },
    {
      "epoch": 0.8661242274687653,
      "grad_norm": 0.05728617683053017,
      "learning_rate": 4.763005409187154e-05,
      "loss": 1.4472,
      "step": 19480
    },
    {
      "epoch": 0.8662131519274376,
      "grad_norm": 0.058214664459228516,
      "learning_rate": 4.7567802285937155e-05,
      "loss": 1.4534,
      "step": 19482
    },
    {
      "epoch": 0.86630207638611,
      "grad_norm": 0.05796115845441818,
      "learning_rate": 4.750558915590081e-05,
      "loss": 1.4468,
      "step": 19484
    },
    {
      "epoch": 0.8663910008447824,
      "grad_norm": 0.05892049893736839,
      "learning_rate": 4.744341470708075e-05,
      "loss": 1.4474,
      "step": 19486
    },
    {
      "epoch": 0.8664799253034547,
      "grad_norm": 0.05804290249943733,
      "learning_rate": 4.738127894479205e-05,
      "loss": 1.4496,
      "step": 19488
    },
    {
      "epoch": 0.8665688497621271,
      "grad_norm": 0.05845332145690918,
      "learning_rate": 4.731918187434625e-05,
      "loss": 1.4493,
      "step": 19490
    },
    {
      "epoch": 0.8666577742207995,
      "grad_norm": 0.058075401932001114,
      "learning_rate": 4.725712350105149e-05,
      "loss": 1.4496,
      "step": 19492
    },
    {
      "epoch": 0.8667466986794717,
      "grad_norm": 0.057721007615327835,
      "learning_rate": 4.719510383021319e-05,
      "loss": 1.4517,
      "step": 19494
    },
    {
      "epoch": 0.8668356231381441,
      "grad_norm": 0.05866990238428116,
      "learning_rate": 4.7133122867132725e-05,
      "loss": 1.4498,
      "step": 19496
    },
    {
      "epoch": 0.8669245475968165,
      "grad_norm": 0.05763349309563637,
      "learning_rate": 4.70711806171088e-05,
      "loss": 1.4501,
      "step": 19498
    },
    {
      "epoch": 0.8670134720554888,
      "grad_norm": 0.058722514659166336,
      "learning_rate": 4.7009277085436176e-05,
      "loss": 1.4518,
      "step": 19500
    },
    {
      "epoch": 0.8670134720554888,
      "eval_loss": 1.4349604845046997,
      "eval_runtime": 12.4467,
      "eval_samples_per_second": 555.169,
      "eval_steps_per_second": 69.416,
      "step": 19500
    },
    {
      "epoch": 0.8671023965141612,
      "grad_norm": 0.058399032801389694,
      "learning_rate": 4.69474122774069e-05,
      "loss": 1.4443,
      "step": 19502
    },
    {
      "epoch": 0.8671913209728336,
      "grad_norm": 0.057792019098997116,
      "learning_rate": 4.6885586198309406e-05,
      "loss": 1.4442,
      "step": 19504
    },
    {
      "epoch": 0.867280245431506,
      "grad_norm": 0.05912178382277489,
      "learning_rate": 4.682379885342869e-05,
      "loss": 1.4518,
      "step": 19506
    },
    {
      "epoch": 0.8673691698901783,
      "grad_norm": 0.05834237486124039,
      "learning_rate": 4.676205024804675e-05,
      "loss": 1.4467,
      "step": 19508
    },
    {
      "epoch": 0.8674580943488507,
      "grad_norm": 0.057614706456661224,
      "learning_rate": 4.6700340387442255e-05,
      "loss": 1.4508,
      "step": 19510
    },
    {
      "epoch": 0.867547018807523,
      "grad_norm": 0.05703110247850418,
      "learning_rate": 4.663866927689025e-05,
      "loss": 1.4479,
      "step": 19512
    },
    {
      "epoch": 0.8676359432661954,
      "grad_norm": 0.05827274173498154,
      "learning_rate": 4.657703692166248e-05,
      "loss": 1.4489,
      "step": 19514
    },
    {
      "epoch": 0.8677248677248677,
      "grad_norm": 0.058673057705163956,
      "learning_rate": 4.6515443327028006e-05,
      "loss": 1.4462,
      "step": 19516
    },
    {
      "epoch": 0.86781379218354,
      "grad_norm": 0.057610198855400085,
      "learning_rate": 4.6453888498251715e-05,
      "loss": 1.45,
      "step": 19518
    },
    {
      "epoch": 0.8679027166422124,
      "grad_norm": 0.05858117714524269,
      "learning_rate": 4.639237244059591e-05,
      "loss": 1.4429,
      "step": 19520
    },
    {
      "epoch": 0.8679916411008848,
      "grad_norm": 0.057463061064481735,
      "learning_rate": 4.633089515931899e-05,
      "loss": 1.4498,
      "step": 19522
    },
    {
      "epoch": 0.8680805655595571,
      "grad_norm": 0.05761371925473213,
      "learning_rate": 4.6269456659676355e-05,
      "loss": 1.4479,
      "step": 19524
    },
    {
      "epoch": 0.8681694900182295,
      "grad_norm": 0.05824030563235283,
      "learning_rate": 4.6208056946920204e-05,
      "loss": 1.4496,
      "step": 19526
    },
    {
      "epoch": 0.8682584144769019,
      "grad_norm": 0.05864088982343674,
      "learning_rate": 4.6146696026299064e-05,
      "loss": 1.4461,
      "step": 19528
    },
    {
      "epoch": 0.8683473389355743,
      "grad_norm": 0.057611025869846344,
      "learning_rate": 4.608537390305845e-05,
      "loss": 1.4491,
      "step": 19530
    },
    {
      "epoch": 0.8684362633942466,
      "grad_norm": 0.05795488879084587,
      "learning_rate": 4.6024090582440446e-05,
      "loss": 1.446,
      "step": 19532
    },
    {
      "epoch": 0.868525187852919,
      "grad_norm": 0.05817627161741257,
      "learning_rate": 4.5962846069683705e-05,
      "loss": 1.4447,
      "step": 19534
    },
    {
      "epoch": 0.8686141123115914,
      "grad_norm": 0.058427758514881134,
      "learning_rate": 4.590164037002381e-05,
      "loss": 1.4526,
      "step": 19536
    },
    {
      "epoch": 0.8687030367702636,
      "grad_norm": 0.05801371484994888,
      "learning_rate": 4.5840473488692856e-05,
      "loss": 1.4533,
      "step": 19538
    },
    {
      "epoch": 0.868791961228936,
      "grad_norm": 0.058494169265031815,
      "learning_rate": 4.57793454309196e-05,
      "loss": 1.447,
      "step": 19540
    },
    {
      "epoch": 0.8688808856876084,
      "grad_norm": 0.058679576963186264,
      "learning_rate": 4.571825620192965e-05,
      "loss": 1.4489,
      "step": 19542
    },
    {
      "epoch": 0.8689698101462807,
      "grad_norm": 0.05957064405083656,
      "learning_rate": 4.565720580694505e-05,
      "loss": 1.446,
      "step": 19544
    },
    {
      "epoch": 0.8690587346049531,
      "grad_norm": 0.05911765247583389,
      "learning_rate": 4.559619425118472e-05,
      "loss": 1.4461,
      "step": 19546
    },
    {
      "epoch": 0.8691476590636255,
      "grad_norm": 0.05717119202017784,
      "learning_rate": 4.553522153986428e-05,
      "loss": 1.4417,
      "step": 19548
    },
    {
      "epoch": 0.8692365835222978,
      "grad_norm": 0.057892777025699615,
      "learning_rate": 4.547428767819578e-05,
      "loss": 1.4474,
      "step": 19550
    },
    {
      "epoch": 0.8693255079809702,
      "grad_norm": 0.05849381908774376,
      "learning_rate": 4.541339267138816e-05,
      "loss": 1.4522,
      "step": 19552
    },
    {
      "epoch": 0.8694144324396426,
      "grad_norm": 0.05854662507772446,
      "learning_rate": 4.535253652464716e-05,
      "loss": 1.443,
      "step": 19554
    },
    {
      "epoch": 0.8695033568983149,
      "grad_norm": 0.05722244083881378,
      "learning_rate": 4.529171924317482e-05,
      "loss": 1.4438,
      "step": 19556
    },
    {
      "epoch": 0.8695922813569872,
      "grad_norm": 0.05876478552818298,
      "learning_rate": 4.523094083217011e-05,
      "loss": 1.4451,
      "step": 19558
    },
    {
      "epoch": 0.8696812058156596,
      "grad_norm": 0.058361224830150604,
      "learning_rate": 4.5170201296828804e-05,
      "loss": 1.4514,
      "step": 19560
    },
    {
      "epoch": 0.8697701302743319,
      "grad_norm": 0.0587281808257103,
      "learning_rate": 4.510950064234298e-05,
      "loss": 1.4467,
      "step": 19562
    },
    {
      "epoch": 0.8698590547330043,
      "grad_norm": 0.05788736417889595,
      "learning_rate": 4.504883887390165e-05,
      "loss": 1.4514,
      "step": 19564
    },
    {
      "epoch": 0.8699479791916767,
      "grad_norm": 0.05839037150144577,
      "learning_rate": 4.498821599669051e-05,
      "loss": 1.4457,
      "step": 19566
    },
    {
      "epoch": 0.870036903650349,
      "grad_norm": 0.059564560651779175,
      "learning_rate": 4.49276320158919e-05,
      "loss": 1.4439,
      "step": 19568
    },
    {
      "epoch": 0.8701258281090214,
      "grad_norm": 0.05748181417584419,
      "learning_rate": 4.486708693668473e-05,
      "loss": 1.4528,
      "step": 19570
    },
    {
      "epoch": 0.8702147525676938,
      "grad_norm": 0.058979492634534836,
      "learning_rate": 4.480658076424454e-05,
      "loss": 1.443,
      "step": 19572
    },
    {
      "epoch": 0.8703036770263661,
      "grad_norm": 0.058040812611579895,
      "learning_rate": 4.474611350374397e-05,
      "loss": 1.4517,
      "step": 19574
    },
    {
      "epoch": 0.8703926014850385,
      "grad_norm": 0.05852758511900902,
      "learning_rate": 4.468568516035182e-05,
      "loss": 1.4503,
      "step": 19576
    },
    {
      "epoch": 0.8704815259437109,
      "grad_norm": 0.05870926007628441,
      "learning_rate": 4.462529573923374e-05,
      "loss": 1.4487,
      "step": 19578
    },
    {
      "epoch": 0.8705704504023831,
      "grad_norm": 0.05752955749630928,
      "learning_rate": 4.4564945245552155e-05,
      "loss": 1.4509,
      "step": 19580
    },
    {
      "epoch": 0.8706593748610555,
      "grad_norm": 0.05794141814112663,
      "learning_rate": 4.450463368446617e-05,
      "loss": 1.4525,
      "step": 19582
    },
    {
      "epoch": 0.8707482993197279,
      "grad_norm": 0.05781754106283188,
      "learning_rate": 4.4444361061131314e-05,
      "loss": 1.4514,
      "step": 19584
    },
    {
      "epoch": 0.8708372237784002,
      "grad_norm": 0.05853389576077461,
      "learning_rate": 4.438412738070008e-05,
      "loss": 1.4528,
      "step": 19586
    },
    {
      "epoch": 0.8709261482370726,
      "grad_norm": 0.05923375114798546,
      "learning_rate": 4.432393264832141e-05,
      "loss": 1.4523,
      "step": 19588
    },
    {
      "epoch": 0.871015072695745,
      "grad_norm": 0.058429352939128876,
      "learning_rate": 4.426377686914118e-05,
      "loss": 1.4478,
      "step": 19590
    },
    {
      "epoch": 0.8711039971544173,
      "grad_norm": 0.05943598598241806,
      "learning_rate": 4.420366004830173e-05,
      "loss": 1.4575,
      "step": 19592
    },
    {
      "epoch": 0.8711929216130897,
      "grad_norm": 0.05901796370744705,
      "learning_rate": 4.414358219094178e-05,
      "loss": 1.4495,
      "step": 19594
    },
    {
      "epoch": 0.8712818460717621,
      "grad_norm": 0.05831238627433777,
      "learning_rate": 4.408354330219755e-05,
      "loss": 1.4525,
      "step": 19596
    },
    {
      "epoch": 0.8713707705304344,
      "grad_norm": 0.059671416878700256,
      "learning_rate": 4.402354338720122e-05,
      "loss": 1.4544,
      "step": 19598
    },
    {
      "epoch": 0.8714596949891068,
      "grad_norm": 0.05828830599784851,
      "learning_rate": 4.3963582451081675e-05,
      "loss": 1.4521,
      "step": 19600
    },
    {
      "epoch": 0.8715486194477791,
      "grad_norm": 0.058360446244478226,
      "learning_rate": 4.3903660498964824e-05,
      "loss": 1.451,
      "step": 19602
    },
    {
      "epoch": 0.8716375439064514,
      "grad_norm": 0.0570736788213253,
      "learning_rate": 4.384377753597302e-05,
      "loss": 1.4486,
      "step": 19604
    },
    {
      "epoch": 0.8717264683651238,
      "grad_norm": 0.05778930336236954,
      "learning_rate": 4.3783933567225374e-05,
      "loss": 1.4499,
      "step": 19606
    },
    {
      "epoch": 0.8718153928237962,
      "grad_norm": 0.0583699494600296,
      "learning_rate": 4.372412859783753e-05,
      "loss": 1.4499,
      "step": 19608
    },
    {
      "epoch": 0.8719043172824685,
      "grad_norm": 0.05747527629137039,
      "learning_rate": 4.3664362632921894e-05,
      "loss": 1.4511,
      "step": 19610
    },
    {
      "epoch": 0.8719932417411409,
      "grad_norm": 0.058236028999090195,
      "learning_rate": 4.3604635677587653e-05,
      "loss": 1.4475,
      "step": 19612
    },
    {
      "epoch": 0.8720821661998133,
      "grad_norm": 0.057254135608673096,
      "learning_rate": 4.3544947736940286e-05,
      "loss": 1.4533,
      "step": 19614
    },
    {
      "epoch": 0.8721710906584856,
      "grad_norm": 0.05867518112063408,
      "learning_rate": 4.348529881608237e-05,
      "loss": 1.446,
      "step": 19616
    },
    {
      "epoch": 0.872260015117158,
      "grad_norm": 0.057885196059942245,
      "learning_rate": 4.3425688920113006e-05,
      "loss": 1.4482,
      "step": 19618
    },
    {
      "epoch": 0.8723489395758304,
      "grad_norm": 0.058469321578741074,
      "learning_rate": 4.3366118054127656e-05,
      "loss": 1.4529,
      "step": 19620
    },
    {
      "epoch": 0.8724378640345027,
      "grad_norm": 0.058005768805742264,
      "learning_rate": 4.330658622321892e-05,
      "loss": 1.4456,
      "step": 19622
    },
    {
      "epoch": 0.872526788493175,
      "grad_norm": 0.05836115777492523,
      "learning_rate": 4.324709343247585e-05,
      "loss": 1.4495,
      "step": 19624
    },
    {
      "epoch": 0.8726157129518474,
      "grad_norm": 0.058056388050317764,
      "learning_rate": 4.318763968698397e-05,
      "loss": 1.4488,
      "step": 19626
    },
    {
      "epoch": 0.8727046374105197,
      "grad_norm": 0.05839812010526657,
      "learning_rate": 4.312822499182584e-05,
      "loss": 1.4443,
      "step": 19628
    },
    {
      "epoch": 0.8727935618691921,
      "grad_norm": 0.05871996283531189,
      "learning_rate": 4.306884935208033e-05,
      "loss": 1.4433,
      "step": 19630
    },
    {
      "epoch": 0.8728824863278645,
      "grad_norm": 0.05736853554844856,
      "learning_rate": 4.300951277282322e-05,
      "loss": 1.4489,
      "step": 19632
    },
    {
      "epoch": 0.8729714107865368,
      "grad_norm": 0.058451537042856216,
      "learning_rate": 4.29502152591269e-05,
      "loss": 1.4537,
      "step": 19634
    },
    {
      "epoch": 0.8730603352452092,
      "grad_norm": 0.05809301510453224,
      "learning_rate": 4.289095681606031e-05,
      "loss": 1.4495,
      "step": 19636
    },
    {
      "epoch": 0.8731492597038816,
      "grad_norm": 0.0578899048268795,
      "learning_rate": 4.283173744868912e-05,
      "loss": 1.4448,
      "step": 19638
    },
    {
      "epoch": 0.8732381841625539,
      "grad_norm": 0.0587192140519619,
      "learning_rate": 4.277255716207573e-05,
      "loss": 1.4503,
      "step": 19640
    },
    {
      "epoch": 0.8733271086212263,
      "grad_norm": 0.05794324725866318,
      "learning_rate": 4.2713415961279034e-05,
      "loss": 1.4495,
      "step": 19642
    },
    {
      "epoch": 0.8734160330798987,
      "grad_norm": 0.057733118534088135,
      "learning_rate": 4.2654313851354765e-05,
      "loss": 1.448,
      "step": 19644
    },
    {
      "epoch": 0.8735049575385709,
      "grad_norm": 0.05751529708504677,
      "learning_rate": 4.259525083735527e-05,
      "loss": 1.4532,
      "step": 19646
    },
    {
      "epoch": 0.8735938819972433,
      "grad_norm": 0.058284077793359756,
      "learning_rate": 4.2536226924329344e-05,
      "loss": 1.4469,
      "step": 19648
    },
    {
      "epoch": 0.8736828064559157,
      "grad_norm": 0.05852330103516579,
      "learning_rate": 4.247724211732285e-05,
      "loss": 1.4482,
      "step": 19650
    },
    {
      "epoch": 0.873771730914588,
      "grad_norm": 0.05764244496822357,
      "learning_rate": 4.2418296421377786e-05,
      "loss": 1.4503,
      "step": 19652
    },
    {
      "epoch": 0.8738606553732604,
      "grad_norm": 0.05952933803200722,
      "learning_rate": 4.235938984153326e-05,
      "loss": 1.448,
      "step": 19654
    },
    {
      "epoch": 0.8739495798319328,
      "grad_norm": 0.058142997324466705,
      "learning_rate": 4.230052238282495e-05,
      "loss": 1.4516,
      "step": 19656
    },
    {
      "epoch": 0.8740385042906051,
      "grad_norm": 0.05889532342553139,
      "learning_rate": 4.224169405028494e-05,
      "loss": 1.4509,
      "step": 19658
    },
    {
      "epoch": 0.8741274287492775,
      "grad_norm": 0.058623868972063065,
      "learning_rate": 4.218290484894216e-05,
      "loss": 1.4537,
      "step": 19660
    },
    {
      "epoch": 0.8742163532079499,
      "grad_norm": 0.05701803043484688,
      "learning_rate": 4.2124154783822314e-05,
      "loss": 1.4467,
      "step": 19662
    },
    {
      "epoch": 0.8743052776666222,
      "grad_norm": 0.05911652743816376,
      "learning_rate": 4.2065443859947425e-05,
      "loss": 1.4449,
      "step": 19664
    },
    {
      "epoch": 0.8743942021252946,
      "grad_norm": 0.05890338122844696,
      "learning_rate": 4.200677208233644e-05,
      "loss": 1.4566,
      "step": 19666
    },
    {
      "epoch": 0.8744831265839669,
      "grad_norm": 0.05840785801410675,
      "learning_rate": 4.194813945600495e-05,
      "loss": 1.4515,
      "step": 19668
    },
    {
      "epoch": 0.8745720510426392,
      "grad_norm": 0.059223078191280365,
      "learning_rate": 4.188954598596506e-05,
      "loss": 1.4478,
      "step": 19670
    },
    {
      "epoch": 0.8746609755013116,
      "grad_norm": 0.05858706682920456,
      "learning_rate": 4.183099167722554e-05,
      "loss": 1.4449,
      "step": 19672
    },
    {
      "epoch": 0.874749899959984,
      "grad_norm": 0.057242508977651596,
      "learning_rate": 4.177247653479194e-05,
      "loss": 1.4504,
      "step": 19674
    },
    {
      "epoch": 0.8748388244186563,
      "grad_norm": 0.058398645371198654,
      "learning_rate": 4.171400056366648e-05,
      "loss": 1.4501,
      "step": 19676
    },
    {
      "epoch": 0.8749277488773287,
      "grad_norm": 0.05813181772828102,
      "learning_rate": 4.165556376884788e-05,
      "loss": 1.4488,
      "step": 19678
    },
    {
      "epoch": 0.8750166733360011,
      "grad_norm": 0.05748685821890831,
      "learning_rate": 4.1597166155331323e-05,
      "loss": 1.4515,
      "step": 19680
    },
    {
      "epoch": 0.8751055977946735,
      "grad_norm": 0.05673710256814957,
      "learning_rate": 4.153880772810936e-05,
      "loss": 1.4491,
      "step": 19682
    },
    {
      "epoch": 0.8751945222533458,
      "grad_norm": 0.05813974887132645,
      "learning_rate": 4.1480488492170396e-05,
      "loss": 1.4472,
      "step": 19684
    },
    {
      "epoch": 0.8752834467120182,
      "grad_norm": 0.05882754176855087,
      "learning_rate": 4.142220845249989e-05,
      "loss": 1.4506,
      "step": 19686
    },
    {
      "epoch": 0.8753723711706904,
      "grad_norm": 0.05838124826550484,
      "learning_rate": 4.13639676140799e-05,
      "loss": 1.4483,
      "step": 19688
    },
    {
      "epoch": 0.8754612956293628,
      "grad_norm": 0.05817935988306999,
      "learning_rate": 4.130576598188907e-05,
      "loss": 1.4489,
      "step": 19690
    },
    {
      "epoch": 0.8755502200880352,
      "grad_norm": 0.05672944337129593,
      "learning_rate": 4.124760356090285e-05,
      "loss": 1.4481,
      "step": 19692
    },
    {
      "epoch": 0.8756391445467075,
      "grad_norm": 0.05868963524699211,
      "learning_rate": 4.1189480356093045e-05,
      "loss": 1.4463,
      "step": 19694
    },
    {
      "epoch": 0.8757280690053799,
      "grad_norm": 0.058534860610961914,
      "learning_rate": 4.113139637242835e-05,
      "loss": 1.4481,
      "step": 19696
    },
    {
      "epoch": 0.8758169934640523,
      "grad_norm": 0.05786217749118805,
      "learning_rate": 4.107335161487419e-05,
      "loss": 1.4436,
      "step": 19698
    },
    {
      "epoch": 0.8759059179227247,
      "grad_norm": 0.057530727237463,
      "learning_rate": 4.101534608839236e-05,
      "loss": 1.4428,
      "step": 19700
    },
    {
      "epoch": 0.875994842381397,
      "grad_norm": 0.05776270851492882,
      "learning_rate": 4.095737979794123e-05,
      "loss": 1.4538,
      "step": 19702
    },
    {
      "epoch": 0.8760837668400694,
      "grad_norm": 0.05870446935296059,
      "learning_rate": 4.0899452748476394e-05,
      "loss": 1.4482,
      "step": 19704
    },
    {
      "epoch": 0.8761726912987418,
      "grad_norm": 0.05895115062594414,
      "learning_rate": 4.08415649449495e-05,
      "loss": 1.4544,
      "step": 19706
    },
    {
      "epoch": 0.8762616157574141,
      "grad_norm": 0.05859552323818207,
      "learning_rate": 4.0783716392309047e-05,
      "loss": 1.4508,
      "step": 19708
    },
    {
      "epoch": 0.8763505402160864,
      "grad_norm": 0.06085187941789627,
      "learning_rate": 4.072590709550017e-05,
      "loss": 1.4512,
      "step": 19710
    },
    {
      "epoch": 0.8764394646747588,
      "grad_norm": 0.057687316089868546,
      "learning_rate": 4.066813705946476e-05,
      "loss": 1.4437,
      "step": 19712
    },
    {
      "epoch": 0.8765283891334311,
      "grad_norm": 0.05841357260942459,
      "learning_rate": 4.061040628914125e-05,
      "loss": 1.449,
      "step": 19714
    },
    {
      "epoch": 0.8766173135921035,
      "grad_norm": 0.058153048157691956,
      "learning_rate": 4.05527147894646e-05,
      "loss": 1.4421,
      "step": 19716
    },
    {
      "epoch": 0.8767062380507759,
      "grad_norm": 0.058645881712436676,
      "learning_rate": 4.049506256536656e-05,
      "loss": 1.454,
      "step": 19718
    },
    {
      "epoch": 0.8767951625094482,
      "grad_norm": 0.057623352855443954,
      "learning_rate": 4.04374496217757e-05,
      "loss": 1.4492,
      "step": 19720
    },
    {
      "epoch": 0.8768840869681206,
      "grad_norm": 0.059648510068655014,
      "learning_rate": 4.0379875963616754e-05,
      "loss": 1.4486,
      "step": 19722
    },
    {
      "epoch": 0.876973011426793,
      "grad_norm": 0.05818026140332222,
      "learning_rate": 4.03223415958115e-05,
      "loss": 1.4503,
      "step": 19724
    },
    {
      "epoch": 0.8770619358854653,
      "grad_norm": 0.05832851305603981,
      "learning_rate": 4.0264846523278334e-05,
      "loss": 1.4549,
      "step": 19726
    },
    {
      "epoch": 0.8771508603441377,
      "grad_norm": 0.05934127792716026,
      "learning_rate": 4.020739075093194e-05,
      "loss": 1.449,
      "step": 19728
    },
    {
      "epoch": 0.8772397848028101,
      "grad_norm": 0.05757347121834755,
      "learning_rate": 4.0149974283684055e-05,
      "loss": 1.4422,
      "step": 19730
    },
    {
      "epoch": 0.8773287092614823,
      "grad_norm": 0.057523492723703384,
      "learning_rate": 4.009259712644292e-05,
      "loss": 1.4492,
      "step": 19732
    },
    {
      "epoch": 0.8774176337201547,
      "grad_norm": 0.0594845674932003,
      "learning_rate": 4.0035259284113276e-05,
      "loss": 1.4449,
      "step": 19734
    },
    {
      "epoch": 0.8775065581788271,
      "grad_norm": 0.05800360441207886,
      "learning_rate": 3.997796076159671e-05,
      "loss": 1.4463,
      "step": 19736
    },
    {
      "epoch": 0.8775954826374994,
      "grad_norm": 0.05898301303386688,
      "learning_rate": 3.992070156379124e-05,
      "loss": 1.4462,
      "step": 19738
    },
    {
      "epoch": 0.8776844070961718,
      "grad_norm": 0.05766226351261139,
      "learning_rate": 3.986348169559173e-05,
      "loss": 1.4474,
      "step": 19740
    },
    {
      "epoch": 0.8777733315548442,
      "grad_norm": 0.05817679688334465,
      "learning_rate": 3.98063011618896e-05,
      "loss": 1.4465,
      "step": 19742
    },
    {
      "epoch": 0.8778622560135165,
      "grad_norm": 0.058511506766080856,
      "learning_rate": 3.9749159967572834e-05,
      "loss": 1.4514,
      "step": 19744
    },
    {
      "epoch": 0.8779511804721889,
      "grad_norm": 0.058822281658649445,
      "learning_rate": 3.9692058117526076e-05,
      "loss": 1.4508,
      "step": 19746
    },
    {
      "epoch": 0.8780401049308613,
      "grad_norm": 0.058104876428842545,
      "learning_rate": 3.9634995616630806e-05,
      "loss": 1.4482,
      "step": 19748
    },
    {
      "epoch": 0.8781290293895336,
      "grad_norm": 0.05765243247151375,
      "learning_rate": 3.95779724697648e-05,
      "loss": 1.451,
      "step": 19750
    },
    {
      "epoch": 0.878217953848206,
      "grad_norm": 0.058590538799762726,
      "learning_rate": 3.952098868180271e-05,
      "loss": 1.4445,
      "step": 19752
    },
    {
      "epoch": 0.8783068783068783,
      "grad_norm": 0.05874716117978096,
      "learning_rate": 3.946404425761585e-05,
      "loss": 1.4548,
      "step": 19754
    },
    {
      "epoch": 0.8783958027655506,
      "grad_norm": 0.05839841812849045,
      "learning_rate": 3.940713920207195e-05,
      "loss": 1.4452,
      "step": 19756
    },
    {
      "epoch": 0.878484727224223,
      "grad_norm": 0.05867445841431618,
      "learning_rate": 3.9350273520035615e-05,
      "loss": 1.4459,
      "step": 19758
    },
    {
      "epoch": 0.8785736516828954,
      "grad_norm": 0.0578501895070076,
      "learning_rate": 3.929344721636774e-05,
      "loss": 1.4534,
      "step": 19760
    },
    {
      "epoch": 0.8786625761415677,
      "grad_norm": 0.05831950530409813,
      "learning_rate": 3.9236660295926485e-05,
      "loss": 1.4527,
      "step": 19762
    },
    {
      "epoch": 0.8787515006002401,
      "grad_norm": 0.05866897478699684,
      "learning_rate": 3.9179912763565975e-05,
      "loss": 1.4518,
      "step": 19764
    },
    {
      "epoch": 0.8788404250589125,
      "grad_norm": 0.059070635586977005,
      "learning_rate": 3.912320462413721e-05,
      "loss": 1.4476,
      "step": 19766
    },
    {
      "epoch": 0.8789293495175848,
      "grad_norm": 0.05861429125070572,
      "learning_rate": 3.906653588248793e-05,
      "loss": 1.4478,
      "step": 19768
    },
    {
      "epoch": 0.8790182739762572,
      "grad_norm": 0.057909827679395676,
      "learning_rate": 3.9009906543462536e-05,
      "loss": 1.4473,
      "step": 19770
    },
    {
      "epoch": 0.8791071984349296,
      "grad_norm": 0.057409241795539856,
      "learning_rate": 3.895331661190177e-05,
      "loss": 1.4447,
      "step": 19772
    },
    {
      "epoch": 0.8791961228936019,
      "grad_norm": 0.05790772661566734,
      "learning_rate": 3.8896766092643257e-05,
      "loss": 1.4461,
      "step": 19774
    },
    {
      "epoch": 0.8792850473522742,
      "grad_norm": 0.05888407304883003,
      "learning_rate": 3.884025499052124e-05,
      "loss": 1.446,
      "step": 19776
    },
    {
      "epoch": 0.8793739718109466,
      "grad_norm": 0.05903869867324829,
      "learning_rate": 3.878378331036647e-05,
      "loss": 1.4504,
      "step": 19778
    },
    {
      "epoch": 0.8794628962696189,
      "grad_norm": 0.057740140706300735,
      "learning_rate": 3.8727351057006365e-05,
      "loss": 1.447,
      "step": 19780
    },
    {
      "epoch": 0.8795518207282913,
      "grad_norm": 0.059584252536296844,
      "learning_rate": 3.867095823526506e-05,
      "loss": 1.4504,
      "step": 19782
    },
    {
      "epoch": 0.8796407451869637,
      "grad_norm": 0.05790328234434128,
      "learning_rate": 3.8614604849963375e-05,
      "loss": 1.4489,
      "step": 19784
    },
    {
      "epoch": 0.879729669645636,
      "grad_norm": 0.058795176446437836,
      "learning_rate": 3.85582909059185e-05,
      "loss": 1.4462,
      "step": 19786
    },
    {
      "epoch": 0.8798185941043084,
      "grad_norm": 0.057657308876514435,
      "learning_rate": 3.850201640794426e-05,
      "loss": 1.4487,
      "step": 19788
    },
    {
      "epoch": 0.8799075185629808,
      "grad_norm": 0.0583043098449707,
      "learning_rate": 3.8445781360851585e-05,
      "loss": 1.4394,
      "step": 19790
    },
    {
      "epoch": 0.8799964430216531,
      "grad_norm": 0.05816715955734253,
      "learning_rate": 3.838958576944751e-05,
      "loss": 1.4454,
      "step": 19792
    },
    {
      "epoch": 0.8800853674803255,
      "grad_norm": 0.05845462530851364,
      "learning_rate": 3.833342963853587e-05,
      "loss": 1.4484,
      "step": 19794
    },
    {
      "epoch": 0.8801742919389978,
      "grad_norm": 0.05977516621351242,
      "learning_rate": 3.8277312972917086e-05,
      "loss": 1.4492,
      "step": 19796
    },
    {
      "epoch": 0.8802632163976701,
      "grad_norm": 0.058808039873838425,
      "learning_rate": 3.8221235777388396e-05,
      "loss": 1.4487,
      "step": 19798
    },
    {
      "epoch": 0.8803521408563425,
      "grad_norm": 0.057808782905340195,
      "learning_rate": 3.816519805674351e-05,
      "loss": 1.4443,
      "step": 19800
    },
    {
      "epoch": 0.8804410653150149,
      "grad_norm": 0.05733674019575119,
      "learning_rate": 3.8109199815772656e-05,
      "loss": 1.4458,
      "step": 19802
    },
    {
      "epoch": 0.8805299897736872,
      "grad_norm": 0.05880142003297806,
      "learning_rate": 3.8053241059262886e-05,
      "loss": 1.4473,
      "step": 19804
    },
    {
      "epoch": 0.8806189142323596,
      "grad_norm": 0.0574544295668602,
      "learning_rate": 3.799732179199788e-05,
      "loss": 1.4467,
      "step": 19806
    },
    {
      "epoch": 0.880707838691032,
      "grad_norm": 0.05944616347551346,
      "learning_rate": 3.794144201875782e-05,
      "loss": 1.4481,
      "step": 19808
    },
    {
      "epoch": 0.8807967631497043,
      "grad_norm": 0.058573972433805466,
      "learning_rate": 3.788560174431932e-05,
      "loss": 1.4517,
      "step": 19810
    },
    {
      "epoch": 0.8808856876083767,
      "grad_norm": 0.05748533084988594,
      "learning_rate": 3.782980097345623e-05,
      "loss": 1.4506,
      "step": 19812
    },
    {
      "epoch": 0.8809746120670491,
      "grad_norm": 0.05795510113239288,
      "learning_rate": 3.77740397109384e-05,
      "loss": 1.4546,
      "step": 19814
    },
    {
      "epoch": 0.8810635365257214,
      "grad_norm": 0.057292576879262924,
      "learning_rate": 3.771831796153269e-05,
      "loss": 1.4429,
      "step": 19816
    },
    {
      "epoch": 0.8811524609843937,
      "grad_norm": 0.05805271118879318,
      "learning_rate": 3.7662635730002284e-05,
      "loss": 1.449,
      "step": 19818
    },
    {
      "epoch": 0.8812413854430661,
      "grad_norm": 0.05669526755809784,
      "learning_rate": 3.760699302110726e-05,
      "loss": 1.443,
      "step": 19820
    },
    {
      "epoch": 0.8813303099017384,
      "grad_norm": 0.05827340483665466,
      "learning_rate": 3.755138983960421e-05,
      "loss": 1.4438,
      "step": 19822
    },
    {
      "epoch": 0.8814192343604108,
      "grad_norm": 0.05783102661371231,
      "learning_rate": 3.7495826190246275e-05,
      "loss": 1.4488,
      "step": 19824
    },
    {
      "epoch": 0.8815081588190832,
      "grad_norm": 0.05988376960158348,
      "learning_rate": 3.744030207778326e-05,
      "loss": 1.4502,
      "step": 19826
    },
    {
      "epoch": 0.8815970832777555,
      "grad_norm": 0.057520363479852676,
      "learning_rate": 3.738481750696176e-05,
      "loss": 1.4514,
      "step": 19828
    },
    {
      "epoch": 0.8816860077364279,
      "grad_norm": 0.057553213089704514,
      "learning_rate": 3.732937248252472e-05,
      "loss": 1.4461,
      "step": 19830
    },
    {
      "epoch": 0.8817749321951003,
      "grad_norm": 0.058701615780591965,
      "learning_rate": 3.7273967009211706e-05,
      "loss": 1.4476,
      "step": 19832
    },
    {
      "epoch": 0.8818638566537726,
      "grad_norm": 0.0582745298743248,
      "learning_rate": 3.721860109175934e-05,
      "loss": 1.4501,
      "step": 19834
    },
    {
      "epoch": 0.881952781112445,
      "grad_norm": 0.057957906275987625,
      "learning_rate": 3.716327473490022e-05,
      "loss": 1.4523,
      "step": 19836
    },
    {
      "epoch": 0.8820417055711174,
      "grad_norm": 0.057991642504930496,
      "learning_rate": 3.710798794336418e-05,
      "loss": 1.4489,
      "step": 19838
    },
    {
      "epoch": 0.8821306300297896,
      "grad_norm": 0.05811452493071556,
      "learning_rate": 3.705274072187709e-05,
      "loss": 1.4473,
      "step": 19840
    },
    {
      "epoch": 0.882219554488462,
      "grad_norm": 0.058310460299253464,
      "learning_rate": 3.6997533075161905e-05,
      "loss": 1.4449,
      "step": 19842
    },
    {
      "epoch": 0.8823084789471344,
      "grad_norm": 0.05879312753677368,
      "learning_rate": 3.694236500793802e-05,
      "loss": 1.4473,
      "step": 19844
    },
    {
      "epoch": 0.8823974034058067,
      "grad_norm": 0.05760977044701576,
      "learning_rate": 3.6887236524921374e-05,
      "loss": 1.4455,
      "step": 19846
    },
    {
      "epoch": 0.8824863278644791,
      "grad_norm": 0.057671599090099335,
      "learning_rate": 3.68321476308246e-05,
      "loss": 1.4504,
      "step": 19848
    },
    {
      "epoch": 0.8825752523231515,
      "grad_norm": 0.058317407965660095,
      "learning_rate": 3.677709833035703e-05,
      "loss": 1.4492,
      "step": 19850
    },
    {
      "epoch": 0.8826641767818239,
      "grad_norm": 0.059517450630664825,
      "learning_rate": 3.6722088628224346e-05,
      "loss": 1.4431,
      "step": 19852
    },
    {
      "epoch": 0.8827531012404962,
      "grad_norm": 0.05849652737379074,
      "learning_rate": 3.6667118529129186e-05,
      "loss": 1.4487,
      "step": 19854
    },
    {
      "epoch": 0.8828420256991686,
      "grad_norm": 0.058554280549287796,
      "learning_rate": 3.661218803777061e-05,
      "loss": 1.4479,
      "step": 19856
    },
    {
      "epoch": 0.882930950157841,
      "grad_norm": 0.05854412168264389,
      "learning_rate": 3.6557297158844205e-05,
      "loss": 1.446,
      "step": 19858
    },
    {
      "epoch": 0.8830198746165133,
      "grad_norm": 0.05787697806954384,
      "learning_rate": 3.650244589704238e-05,
      "loss": 1.4494,
      "step": 19860
    },
    {
      "epoch": 0.8831087990751856,
      "grad_norm": 0.05933505296707153,
      "learning_rate": 3.6447634257054163e-05,
      "loss": 1.4462,
      "step": 19862
    },
    {
      "epoch": 0.883197723533858,
      "grad_norm": 0.059682779014110565,
      "learning_rate": 3.6392862243564864e-05,
      "loss": 1.4473,
      "step": 19864
    },
    {
      "epoch": 0.8832866479925303,
      "grad_norm": 0.05752909556031227,
      "learning_rate": 3.633812986125679e-05,
      "loss": 1.4441,
      "step": 19866
    },
    {
      "epoch": 0.8833755724512027,
      "grad_norm": 0.0576491579413414,
      "learning_rate": 3.6283437114808526e-05,
      "loss": 1.4428,
      "step": 19868
    },
    {
      "epoch": 0.883464496909875,
      "grad_norm": 0.05781255662441254,
      "learning_rate": 3.622878400889579e-05,
      "loss": 1.4531,
      "step": 19870
    },
    {
      "epoch": 0.8835534213685474,
      "grad_norm": 0.058144886046648026,
      "learning_rate": 3.617417054819039e-05,
      "loss": 1.4454,
      "step": 19872
    },
    {
      "epoch": 0.8836423458272198,
      "grad_norm": 0.05706218257546425,
      "learning_rate": 3.611959673736076e-05,
      "loss": 1.4469,
      "step": 19874
    },
    {
      "epoch": 0.8837312702858922,
      "grad_norm": 0.05844343453645706,
      "learning_rate": 3.6065062581072283e-05,
      "loss": 1.4444,
      "step": 19876
    },
    {
      "epoch": 0.8838201947445645,
      "grad_norm": 0.05753406509757042,
      "learning_rate": 3.6010568083986837e-05,
      "loss": 1.4454,
      "step": 19878
    },
    {
      "epoch": 0.8839091192032369,
      "grad_norm": 0.05792682617902756,
      "learning_rate": 3.595611325076264e-05,
      "loss": 1.4469,
      "step": 19880
    },
    {
      "epoch": 0.8839980436619093,
      "grad_norm": 0.059252262115478516,
      "learning_rate": 3.5901698086054926e-05,
      "loss": 1.4485,
      "step": 19882
    },
    {
      "epoch": 0.8840869681205815,
      "grad_norm": 0.05811452120542526,
      "learning_rate": 3.584732259451523e-05,
      "loss": 1.4531,
      "step": 19884
    },
    {
      "epoch": 0.8841758925792539,
      "grad_norm": 0.0585939846932888,
      "learning_rate": 3.579298678079196e-05,
      "loss": 1.4519,
      "step": 19886
    },
    {
      "epoch": 0.8842648170379263,
      "grad_norm": 0.058722611516714096,
      "learning_rate": 3.573869064952989e-05,
      "loss": 1.4502,
      "step": 19888
    },
    {
      "epoch": 0.8843537414965986,
      "grad_norm": 0.05796421319246292,
      "learning_rate": 3.5684434205370266e-05,
      "loss": 1.4456,
      "step": 19890
    },
    {
      "epoch": 0.884442665955271,
      "grad_norm": 0.058497413992881775,
      "learning_rate": 3.563021745295158e-05,
      "loss": 1.4431,
      "step": 19892
    },
    {
      "epoch": 0.8845315904139434,
      "grad_norm": 0.05818541720509529,
      "learning_rate": 3.557604039690826e-05,
      "loss": 1.4429,
      "step": 19894
    },
    {
      "epoch": 0.8846205148726157,
      "grad_norm": 0.05828235670924187,
      "learning_rate": 3.552190304187164e-05,
      "loss": 1.4481,
      "step": 19896
    },
    {
      "epoch": 0.8847094393312881,
      "grad_norm": 0.05807334557175636,
      "learning_rate": 3.546780539246958e-05,
      "loss": 1.4478,
      "step": 19898
    },
    {
      "epoch": 0.8847983637899605,
      "grad_norm": 0.056775208562612534,
      "learning_rate": 3.541374745332676e-05,
      "loss": 1.4516,
      "step": 19900
    },
    {
      "epoch": 0.8848872882486328,
      "grad_norm": 0.05986194312572479,
      "learning_rate": 3.535972922906411e-05,
      "loss": 1.4534,
      "step": 19902
    },
    {
      "epoch": 0.8849762127073052,
      "grad_norm": 0.056858133524656296,
      "learning_rate": 3.530575072429937e-05,
      "loss": 1.4486,
      "step": 19904
    },
    {
      "epoch": 0.8850651371659775,
      "grad_norm": 0.058056510984897614,
      "learning_rate": 3.5251811943646904e-05,
      "loss": 1.4485,
      "step": 19906
    },
    {
      "epoch": 0.8851540616246498,
      "grad_norm": 0.05905826762318611,
      "learning_rate": 3.5197912891717754e-05,
      "loss": 1.4528,
      "step": 19908
    },
    {
      "epoch": 0.8852429860833222,
      "grad_norm": 0.05918426811695099,
      "learning_rate": 3.514405357311923e-05,
      "loss": 1.4511,
      "step": 19910
    },
    {
      "epoch": 0.8853319105419946,
      "grad_norm": 0.05840429291129112,
      "learning_rate": 3.5090233992455536e-05,
      "loss": 1.4492,
      "step": 19912
    },
    {
      "epoch": 0.8854208350006669,
      "grad_norm": 0.058113958686590195,
      "learning_rate": 3.503645415432749e-05,
      "loss": 1.4486,
      "step": 19914
    },
    {
      "epoch": 0.8855097594593393,
      "grad_norm": 0.05794018507003784,
      "learning_rate": 3.498271406333242e-05,
      "loss": 1.4522,
      "step": 19916
    },
    {
      "epoch": 0.8855986839180117,
      "grad_norm": 0.058955464512109756,
      "learning_rate": 3.492901372406404e-05,
      "loss": 1.4498,
      "step": 19918
    },
    {
      "epoch": 0.885687608376684,
      "grad_norm": 0.05927339568734169,
      "learning_rate": 3.487535314111323e-05,
      "loss": 1.4484,
      "step": 19920
    },
    {
      "epoch": 0.8857765328353564,
      "grad_norm": 0.05773244425654411,
      "learning_rate": 3.4821732319066876e-05,
      "loss": 1.4503,
      "step": 19922
    },
    {
      "epoch": 0.8858654572940288,
      "grad_norm": 0.05872263014316559,
      "learning_rate": 3.476815126250893e-05,
      "loss": 1.4463,
      "step": 19924
    },
    {
      "epoch": 0.885954381752701,
      "grad_norm": 0.05775640159845352,
      "learning_rate": 3.4714609976019496e-05,
      "loss": 1.4479,
      "step": 19926
    },
    {
      "epoch": 0.8860433062113734,
      "grad_norm": 0.057550862431526184,
      "learning_rate": 3.4661108464175696e-05,
      "loss": 1.4462,
      "step": 19928
    },
    {
      "epoch": 0.8861322306700458,
      "grad_norm": 0.059137601405382156,
      "learning_rate": 3.460764673155109e-05,
      "loss": 1.4451,
      "step": 19930
    },
    {
      "epoch": 0.8862211551287181,
      "grad_norm": 0.057657770812511444,
      "learning_rate": 3.4554224782715636e-05,
      "loss": 1.4474,
      "step": 19932
    },
    {
      "epoch": 0.8863100795873905,
      "grad_norm": 0.05691711977124214,
      "learning_rate": 3.450084262223624e-05,
      "loss": 1.4476,
      "step": 19934
    },
    {
      "epoch": 0.8863990040460629,
      "grad_norm": 0.05885188281536102,
      "learning_rate": 3.4447500254676304e-05,
      "loss": 1.4472,
      "step": 19936
    },
    {
      "epoch": 0.8864879285047352,
      "grad_norm": 0.05741071701049805,
      "learning_rate": 3.4394197684595564e-05,
      "loss": 1.4481,
      "step": 19938
    },
    {
      "epoch": 0.8865768529634076,
      "grad_norm": 0.058238908648490906,
      "learning_rate": 3.434093491655066e-05,
      "loss": 1.4475,
      "step": 19940
    },
    {
      "epoch": 0.88666577742208,
      "grad_norm": 0.05798852816224098,
      "learning_rate": 3.428771195509483e-05,
      "loss": 1.4466,
      "step": 19942
    },
    {
      "epoch": 0.8867547018807523,
      "grad_norm": 0.05808843672275543,
      "learning_rate": 3.423452880477762e-05,
      "loss": 1.45,
      "step": 19944
    },
    {
      "epoch": 0.8868436263394247,
      "grad_norm": 0.059082068502902985,
      "learning_rate": 3.4181385470145546e-05,
      "loss": 1.4458,
      "step": 19946
    },
    {
      "epoch": 0.886932550798097,
      "grad_norm": 0.05787166953086853,
      "learning_rate": 3.41282819557413e-05,
      "loss": 1.449,
      "step": 19948
    },
    {
      "epoch": 0.8870214752567693,
      "grad_norm": 0.05777514725923538,
      "learning_rate": 3.40752182661046e-05,
      "loss": 1.4453,
      "step": 19950
    },
    {
      "epoch": 0.8871103997154417,
      "grad_norm": 0.059281278401613235,
      "learning_rate": 3.402219440577159e-05,
      "loss": 1.4491,
      "step": 19952
    },
    {
      "epoch": 0.8871993241741141,
      "grad_norm": 0.058436159044504166,
      "learning_rate": 3.3969210379274805e-05,
      "loss": 1.4448,
      "step": 19954
    },
    {
      "epoch": 0.8872882486327864,
      "grad_norm": 0.057971615344285965,
      "learning_rate": 3.391626619114363e-05,
      "loss": 1.4542,
      "step": 19956
    },
    {
      "epoch": 0.8873771730914588,
      "grad_norm": 0.05890185385942459,
      "learning_rate": 3.386336184590405e-05,
      "loss": 1.4479,
      "step": 19958
    },
    {
      "epoch": 0.8874660975501312,
      "grad_norm": 0.05890071764588356,
      "learning_rate": 3.381049734807845e-05,
      "loss": 1.4439,
      "step": 19960
    },
    {
      "epoch": 0.8875550220088035,
      "grad_norm": 0.05897240713238716,
      "learning_rate": 3.375767270218599e-05,
      "loss": 1.4473,
      "step": 19962
    },
    {
      "epoch": 0.8876439464674759,
      "grad_norm": 0.05797523632645607,
      "learning_rate": 3.370488791274234e-05,
      "loss": 1.4474,
      "step": 19964
    },
    {
      "epoch": 0.8877328709261483,
      "grad_norm": 0.05868737772107124,
      "learning_rate": 3.3652142984259783e-05,
      "loss": 1.446,
      "step": 19966
    },
    {
      "epoch": 0.8878217953848206,
      "grad_norm": 0.05914899334311485,
      "learning_rate": 3.359943792124709e-05,
      "loss": 1.4449,
      "step": 19968
    },
    {
      "epoch": 0.8879107198434929,
      "grad_norm": 0.05786535516381264,
      "learning_rate": 3.354677272820994e-05,
      "loss": 1.4519,
      "step": 19970
    },
    {
      "epoch": 0.8879996443021653,
      "grad_norm": 0.05873554199934006,
      "learning_rate": 3.349414740965012e-05,
      "loss": 1.4475,
      "step": 19972
    },
    {
      "epoch": 0.8880885687608376,
      "grad_norm": 0.05829008296132088,
      "learning_rate": 3.344156197006654e-05,
      "loss": 1.448,
      "step": 19974
    },
    {
      "epoch": 0.88817749321951,
      "grad_norm": 0.05718972161412239,
      "learning_rate": 3.338901641395409e-05,
      "loss": 1.4437,
      "step": 19976
    },
    {
      "epoch": 0.8882664176781824,
      "grad_norm": 0.058192022144794464,
      "learning_rate": 3.333651074580496e-05,
      "loss": 1.4443,
      "step": 19978
    },
    {
      "epoch": 0.8883553421368547,
      "grad_norm": 0.05977290868759155,
      "learning_rate": 3.328404497010745e-05,
      "loss": 1.4562,
      "step": 19980
    },
    {
      "epoch": 0.8884442665955271,
      "grad_norm": 0.057841334491968155,
      "learning_rate": 3.323161909134642e-05,
      "loss": 1.4467,
      "step": 19982
    },
    {
      "epoch": 0.8885331910541995,
      "grad_norm": 0.0573255680501461,
      "learning_rate": 3.317923311400356e-05,
      "loss": 1.4424,
      "step": 19984
    },
    {
      "epoch": 0.8886221155128718,
      "grad_norm": 0.057299330830574036,
      "learning_rate": 3.312688704255717e-05,
      "loss": 1.4463,
      "step": 19986
    },
    {
      "epoch": 0.8887110399715442,
      "grad_norm": 0.0575009286403656,
      "learning_rate": 3.3074580881481854e-05,
      "loss": 1.4464,
      "step": 19988
    },
    {
      "epoch": 0.8887999644302166,
      "grad_norm": 0.05899973586201668,
      "learning_rate": 3.302231463524902e-05,
      "loss": 1.4517,
      "step": 19990
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.057911746203899384,
      "learning_rate": 3.29700883083266e-05,
      "loss": 1.4506,
      "step": 19992
    },
    {
      "epoch": 0.8889778133475612,
      "grad_norm": 0.057607442140579224,
      "learning_rate": 3.2917901905179294e-05,
      "loss": 1.4492,
      "step": 19994
    },
    {
      "epoch": 0.8890667378062336,
      "grad_norm": 0.05750446021556854,
      "learning_rate": 3.2865755430268095e-05,
      "loss": 1.4517,
      "step": 19996
    },
    {
      "epoch": 0.8891556622649059,
      "grad_norm": 0.05804445222020149,
      "learning_rate": 3.281364888805055e-05,
      "loss": 1.4511,
      "step": 19998
    },
    {
      "epoch": 0.8892445867235783,
      "grad_norm": 0.05784785374999046,
      "learning_rate": 3.276158228298126e-05,
      "loss": 1.45,
      "step": 20000
    },
    {
      "epoch": 0.8892445867235783,
      "eval_loss": 1.4338102340698242,
      "eval_runtime": 12.4535,
      "eval_samples_per_second": 554.865,
      "eval_steps_per_second": 69.378,
      "step": 20000
    },
    {
      "epoch": 0.8893335111822507,
      "grad_norm": 0.05788677558302879,
      "learning_rate": 3.2709555619511e-05,
      "loss": 1.4518,
      "step": 20002
    },
    {
      "epoch": 0.889422435640923,
      "grad_norm": 0.058794233947992325,
      "learning_rate": 3.26575689020871e-05,
      "loss": 1.4477,
      "step": 20004
    },
    {
      "epoch": 0.8895113600995954,
      "grad_norm": 0.05749224126338959,
      "learning_rate": 3.260562213515372e-05,
      "loss": 1.4488,
      "step": 20006
    },
    {
      "epoch": 0.8896002845582678,
      "grad_norm": 0.05827973410487175,
      "learning_rate": 3.2553715323151486e-05,
      "loss": 1.4487,
      "step": 20008
    },
    {
      "epoch": 0.8896892090169402,
      "grad_norm": 0.05793258920311928,
      "learning_rate": 3.250184847051779e-05,
      "loss": 1.4495,
      "step": 20010
    },
    {
      "epoch": 0.8897781334756125,
      "grad_norm": 0.05844861641526222,
      "learning_rate": 3.245002158168614e-05,
      "loss": 1.452,
      "step": 20012
    },
    {
      "epoch": 0.8898670579342848,
      "grad_norm": 0.05809600278735161,
      "learning_rate": 3.23982346610871e-05,
      "loss": 1.4541,
      "step": 20014
    },
    {
      "epoch": 0.8899559823929571,
      "grad_norm": 0.05752575770020485,
      "learning_rate": 3.2346487713147684e-05,
      "loss": 1.4477,
      "step": 20016
    },
    {
      "epoch": 0.8900449068516295,
      "grad_norm": 0.05790848284959793,
      "learning_rate": 3.229478074229125e-05,
      "loss": 1.4473,
      "step": 20018
    },
    {
      "epoch": 0.8901338313103019,
      "grad_norm": 0.058815497905015945,
      "learning_rate": 3.224311375293815e-05,
      "loss": 1.4475,
      "step": 20020
    },
    {
      "epoch": 0.8902227557689742,
      "grad_norm": 0.05812389776110649,
      "learning_rate": 3.219148674950506e-05,
      "loss": 1.4455,
      "step": 20022
    },
    {
      "epoch": 0.8903116802276466,
      "grad_norm": 0.058545514941215515,
      "learning_rate": 3.2139899736405245e-05,
      "loss": 1.4428,
      "step": 20024
    },
    {
      "epoch": 0.890400604686319,
      "grad_norm": 0.05910952761769295,
      "learning_rate": 3.208835271804844e-05,
      "loss": 1.4495,
      "step": 20026
    },
    {
      "epoch": 0.8904895291449914,
      "grad_norm": 0.05743895471096039,
      "learning_rate": 3.2036845698841465e-05,
      "loss": 1.4403,
      "step": 20028
    },
    {
      "epoch": 0.8905784536036637,
      "grad_norm": 0.059019867330789566,
      "learning_rate": 3.1985378683187014e-05,
      "loss": 1.446,
      "step": 20030
    },
    {
      "epoch": 0.8906673780623361,
      "grad_norm": 0.05902063101530075,
      "learning_rate": 3.193395167548502e-05,
      "loss": 1.4457,
      "step": 20032
    },
    {
      "epoch": 0.8907563025210085,
      "grad_norm": 0.05790397897362709,
      "learning_rate": 3.18825646801314e-05,
      "loss": 1.45,
      "step": 20034
    },
    {
      "epoch": 0.8908452269796807,
      "grad_norm": 0.05794849991798401,
      "learning_rate": 3.183121770151909e-05,
      "loss": 1.4494,
      "step": 20036
    },
    {
      "epoch": 0.8909341514383531,
      "grad_norm": 0.058465514332056046,
      "learning_rate": 3.1779910744037574e-05,
      "loss": 1.4499,
      "step": 20038
    },
    {
      "epoch": 0.8910230758970255,
      "grad_norm": 0.05779614299535751,
      "learning_rate": 3.172864381207252e-05,
      "loss": 1.4433,
      "step": 20040
    },
    {
      "epoch": 0.8911120003556978,
      "grad_norm": 0.058368321508169174,
      "learning_rate": 3.167741691000664e-05,
      "loss": 1.4451,
      "step": 20042
    },
    {
      "epoch": 0.8912009248143702,
      "grad_norm": 0.05803350359201431,
      "learning_rate": 3.162623004221904e-05,
      "loss": 1.4465,
      "step": 20044
    },
    {
      "epoch": 0.8912898492730426,
      "grad_norm": 0.05906941741704941,
      "learning_rate": 3.1575083213085275e-05,
      "loss": 1.443,
      "step": 20046
    },
    {
      "epoch": 0.8913787737317149,
      "grad_norm": 0.058908261358737946,
      "learning_rate": 3.152397642697774e-05,
      "loss": 1.4541,
      "step": 20048
    },
    {
      "epoch": 0.8914676981903873,
      "grad_norm": 0.05919423699378967,
      "learning_rate": 3.147290968826522e-05,
      "loss": 1.4473,
      "step": 20050
    },
    {
      "epoch": 0.8915566226490597,
      "grad_norm": 0.057987507432699203,
      "learning_rate": 3.142188300131305e-05,
      "loss": 1.4466,
      "step": 20052
    },
    {
      "epoch": 0.891645547107732,
      "grad_norm": 0.057514190673828125,
      "learning_rate": 3.137089637048335e-05,
      "loss": 1.4462,
      "step": 20054
    },
    {
      "epoch": 0.8917344715664043,
      "grad_norm": 0.05833049863576889,
      "learning_rate": 3.131994980013453e-05,
      "loss": 1.447,
      "step": 20056
    },
    {
      "epoch": 0.8918233960250767,
      "grad_norm": 0.05808662623167038,
      "learning_rate": 3.126904329462182e-05,
      "loss": 1.4454,
      "step": 20058
    },
    {
      "epoch": 0.891912320483749,
      "grad_norm": 0.05755183845758438,
      "learning_rate": 3.1218176858296974e-05,
      "loss": 1.4474,
      "step": 20060
    },
    {
      "epoch": 0.8920012449424214,
      "grad_norm": 0.05792864039540291,
      "learning_rate": 3.1167350495508175e-05,
      "loss": 1.446,
      "step": 20062
    },
    {
      "epoch": 0.8920901694010938,
      "grad_norm": 0.05887740105390549,
      "learning_rate": 3.111656421060033e-05,
      "loss": 1.4454,
      "step": 20064
    },
    {
      "epoch": 0.8921790938597661,
      "grad_norm": 0.05969681963324547,
      "learning_rate": 3.106581800791497e-05,
      "loss": 1.453,
      "step": 20066
    },
    {
      "epoch": 0.8922680183184385,
      "grad_norm": 0.0589289627969265,
      "learning_rate": 3.10151118917899e-05,
      "loss": 1.4504,
      "step": 20068
    },
    {
      "epoch": 0.8923569427771109,
      "grad_norm": 0.05786190554499626,
      "learning_rate": 3.096444586655983e-05,
      "loss": 1.4524,
      "step": 20070
    },
    {
      "epoch": 0.8924458672357832,
      "grad_norm": 0.05816512554883957,
      "learning_rate": 3.091381993655601e-05,
      "loss": 1.4438,
      "step": 20072
    },
    {
      "epoch": 0.8925347916944556,
      "grad_norm": 0.057929810136556625,
      "learning_rate": 3.086323410610592e-05,
      "loss": 1.4494,
      "step": 20074
    },
    {
      "epoch": 0.892623716153128,
      "grad_norm": 0.05758269876241684,
      "learning_rate": 3.081268837953405e-05,
      "loss": 1.4496,
      "step": 20076
    },
    {
      "epoch": 0.8927126406118002,
      "grad_norm": 0.05764364078640938,
      "learning_rate": 3.076218276116116e-05,
      "loss": 1.4491,
      "step": 20078
    },
    {
      "epoch": 0.8928015650704726,
      "grad_norm": 0.057650718837976456,
      "learning_rate": 3.0711717255304864e-05,
      "loss": 1.4495,
      "step": 20080
    },
    {
      "epoch": 0.892890489529145,
      "grad_norm": 0.057383351027965546,
      "learning_rate": 3.066129186627909e-05,
      "loss": 1.4452,
      "step": 20082
    },
    {
      "epoch": 0.8929794139878173,
      "grad_norm": 0.05764487385749817,
      "learning_rate": 3.061090659839422e-05,
      "loss": 1.4497,
      "step": 20084
    },
    {
      "epoch": 0.8930683384464897,
      "grad_norm": 0.05808642879128456,
      "learning_rate": 3.05605614559577e-05,
      "loss": 1.4498,
      "step": 20086
    },
    {
      "epoch": 0.8931572629051621,
      "grad_norm": 0.05805491656064987,
      "learning_rate": 3.051025644327321e-05,
      "loss": 1.4483,
      "step": 20088
    },
    {
      "epoch": 0.8932461873638344,
      "grad_norm": 0.05781690031290054,
      "learning_rate": 3.0459991564640855e-05,
      "loss": 1.4429,
      "step": 20090
    },
    {
      "epoch": 0.8933351118225068,
      "grad_norm": 0.06007815897464752,
      "learning_rate": 3.040976682435759e-05,
      "loss": 1.4486,
      "step": 20092
    },
    {
      "epoch": 0.8934240362811792,
      "grad_norm": 0.058555953204631805,
      "learning_rate": 3.0359582226716975e-05,
      "loss": 1.4454,
      "step": 20094
    },
    {
      "epoch": 0.8935129607398515,
      "grad_norm": 0.05821076035499573,
      "learning_rate": 3.0309437776008865e-05,
      "loss": 1.4527,
      "step": 20096
    },
    {
      "epoch": 0.8936018851985239,
      "grad_norm": 0.058373384177684784,
      "learning_rate": 3.0259333476519825e-05,
      "loss": 1.4439,
      "step": 20098
    },
    {
      "epoch": 0.8936908096571962,
      "grad_norm": 0.05813969671726227,
      "learning_rate": 3.0209269332533094e-05,
      "loss": 1.4469,
      "step": 20100
    },
    {
      "epoch": 0.8937797341158685,
      "grad_norm": 0.058415379375219345,
      "learning_rate": 3.0159245348328368e-05,
      "loss": 1.447,
      "step": 20102
    },
    {
      "epoch": 0.8938686585745409,
      "grad_norm": 0.05789436027407646,
      "learning_rate": 3.010926152818183e-05,
      "loss": 1.4537,
      "step": 20104
    },
    {
      "epoch": 0.8939575830332133,
      "grad_norm": 0.058166682720184326,
      "learning_rate": 3.0059317876366286e-05,
      "loss": 1.4486,
      "step": 20106
    },
    {
      "epoch": 0.8940465074918856,
      "grad_norm": 0.05696212127804756,
      "learning_rate": 3.0009414397151326e-05,
      "loss": 1.4443,
      "step": 20108
    },
    {
      "epoch": 0.894135431950558,
      "grad_norm": 0.057195693254470825,
      "learning_rate": 2.995955109480275e-05,
      "loss": 1.4523,
      "step": 20110
    },
    {
      "epoch": 0.8942243564092304,
      "grad_norm": 0.058714404702186584,
      "learning_rate": 2.9909727973583156e-05,
      "loss": 1.4506,
      "step": 20112
    },
    {
      "epoch": 0.8943132808679027,
      "grad_norm": 0.05786760151386261,
      "learning_rate": 2.985994503775158e-05,
      "loss": 1.4493,
      "step": 20114
    },
    {
      "epoch": 0.8944022053265751,
      "grad_norm": 0.05894371122121811,
      "learning_rate": 2.981020229156378e-05,
      "loss": 1.4467,
      "step": 20116
    },
    {
      "epoch": 0.8944911297852475,
      "grad_norm": 0.05780460685491562,
      "learning_rate": 2.9760499739271974e-05,
      "loss": 1.4471,
      "step": 20118
    },
    {
      "epoch": 0.8945800542439198,
      "grad_norm": 0.058124516159296036,
      "learning_rate": 2.9710837385124924e-05,
      "loss": 1.4449,
      "step": 20120
    },
    {
      "epoch": 0.8946689787025921,
      "grad_norm": 0.057611338794231415,
      "learning_rate": 2.9661215233367956e-05,
      "loss": 1.4518,
      "step": 20122
    },
    {
      "epoch": 0.8947579031612645,
      "grad_norm": 0.05695406347513199,
      "learning_rate": 2.961163328824307e-05,
      "loss": 1.4398,
      "step": 20124
    },
    {
      "epoch": 0.8948468276199368,
      "grad_norm": 0.057146843522787094,
      "learning_rate": 2.9562091553988758e-05,
      "loss": 1.4436,
      "step": 20126
    },
    {
      "epoch": 0.8949357520786092,
      "grad_norm": 0.05892251059412956,
      "learning_rate": 2.9512590034839802e-05,
      "loss": 1.4539,
      "step": 20128
    },
    {
      "epoch": 0.8950246765372816,
      "grad_norm": 0.05868358537554741,
      "learning_rate": 2.946312873502821e-05,
      "loss": 1.4473,
      "step": 20130
    },
    {
      "epoch": 0.8951136009959539,
      "grad_norm": 0.058262553066015244,
      "learning_rate": 2.9413707658781876e-05,
      "loss": 1.4478,
      "step": 20132
    },
    {
      "epoch": 0.8952025254546263,
      "grad_norm": 0.05769951269030571,
      "learning_rate": 2.936432681032569e-05,
      "loss": 1.4514,
      "step": 20134
    },
    {
      "epoch": 0.8952914499132987,
      "grad_norm": 0.057862915098667145,
      "learning_rate": 2.9314986193880842e-05,
      "loss": 1.4489,
      "step": 20136
    },
    {
      "epoch": 0.895380374371971,
      "grad_norm": 0.05768333002924919,
      "learning_rate": 2.9265685813665178e-05,
      "loss": 1.4443,
      "step": 20138
    },
    {
      "epoch": 0.8954692988306434,
      "grad_norm": 0.05807989463210106,
      "learning_rate": 2.921642567389321e-05,
      "loss": 1.4404,
      "step": 20140
    },
    {
      "epoch": 0.8955582232893158,
      "grad_norm": 0.05821651592850685,
      "learning_rate": 2.9167205778775795e-05,
      "loss": 1.4466,
      "step": 20142
    },
    {
      "epoch": 0.895647147747988,
      "grad_norm": 0.058034975081682205,
      "learning_rate": 2.9118026132520513e-05,
      "loss": 1.443,
      "step": 20144
    },
    {
      "epoch": 0.8957360722066604,
      "grad_norm": 0.057850636541843414,
      "learning_rate": 2.9068886739331557e-05,
      "loss": 1.4407,
      "step": 20146
    },
    {
      "epoch": 0.8958249966653328,
      "grad_norm": 0.05788253992795944,
      "learning_rate": 2.90197876034094e-05,
      "loss": 1.4454,
      "step": 20148
    },
    {
      "epoch": 0.8959139211240051,
      "grad_norm": 0.05744844675064087,
      "learning_rate": 2.8970728728951347e-05,
      "loss": 1.4516,
      "step": 20150
    },
    {
      "epoch": 0.8960028455826775,
      "grad_norm": 0.058373648673295975,
      "learning_rate": 2.8921710120151212e-05,
      "loss": 1.4466,
      "step": 20152
    },
    {
      "epoch": 0.8960917700413499,
      "grad_norm": 0.05803242325782776,
      "learning_rate": 2.8872731781199257e-05,
      "loss": 1.4488,
      "step": 20154
    },
    {
      "epoch": 0.8961806945000222,
      "grad_norm": 0.057802025228738785,
      "learning_rate": 2.88237937162823e-05,
      "loss": 1.4517,
      "step": 20156
    },
    {
      "epoch": 0.8962696189586946,
      "grad_norm": 0.05781133100390434,
      "learning_rate": 2.877489592958399e-05,
      "loss": 1.4443,
      "step": 20158
    },
    {
      "epoch": 0.896358543417367,
      "grad_norm": 0.05783500894904137,
      "learning_rate": 2.8726038425284096e-05,
      "loss": 1.4474,
      "step": 20160
    },
    {
      "epoch": 0.8964474678760394,
      "grad_norm": 0.05782508850097656,
      "learning_rate": 2.867722120755939e-05,
      "loss": 1.4438,
      "step": 20162
    },
    {
      "epoch": 0.8965363923347116,
      "grad_norm": 0.05776134133338928,
      "learning_rate": 2.8628444280582756e-05,
      "loss": 1.4513,
      "step": 20164
    },
    {
      "epoch": 0.896625316793384,
      "grad_norm": 0.0572594478726387,
      "learning_rate": 2.857970764852397e-05,
      "loss": 1.4473,
      "step": 20166
    },
    {
      "epoch": 0.8967142412520563,
      "grad_norm": 0.05813061445951462,
      "learning_rate": 2.853101131554936e-05,
      "loss": 1.4434,
      "step": 20168
    },
    {
      "epoch": 0.8968031657107287,
      "grad_norm": 0.05757516622543335,
      "learning_rate": 2.8482355285821494e-05,
      "loss": 1.4455,
      "step": 20170
    },
    {
      "epoch": 0.8968920901694011,
      "grad_norm": 0.05779808387160301,
      "learning_rate": 2.8433739563499817e-05,
      "loss": 1.4454,
      "step": 20172
    },
    {
      "epoch": 0.8969810146280734,
      "grad_norm": 0.058449193835258484,
      "learning_rate": 2.8385164152740283e-05,
      "loss": 1.4442,
      "step": 20174
    },
    {
      "epoch": 0.8970699390867458,
      "grad_norm": 0.05776485055685043,
      "learning_rate": 2.833662905769524e-05,
      "loss": 1.444,
      "step": 20176
    },
    {
      "epoch": 0.8971588635454182,
      "grad_norm": 0.0574498176574707,
      "learning_rate": 2.8288134282513644e-05,
      "loss": 1.4476,
      "step": 20178
    },
    {
      "epoch": 0.8972477880040906,
      "grad_norm": 0.05783705785870552,
      "learning_rate": 2.8239679831341126e-05,
      "loss": 1.4464,
      "step": 20180
    },
    {
      "epoch": 0.8973367124627629,
      "grad_norm": 0.05792314559221268,
      "learning_rate": 2.819126570831976e-05,
      "loss": 1.4502,
      "step": 20182
    },
    {
      "epoch": 0.8974256369214353,
      "grad_norm": 0.057795315980911255,
      "learning_rate": 2.814289191758823e-05,
      "loss": 1.4469,
      "step": 20184
    },
    {
      "epoch": 0.8975145613801075,
      "grad_norm": 0.05749436840415001,
      "learning_rate": 2.8094558463281572e-05,
      "loss": 1.4488,
      "step": 20186
    },
    {
      "epoch": 0.8976034858387799,
      "grad_norm": 0.05765572935342789,
      "learning_rate": 2.804626534953181e-05,
      "loss": 1.4416,
      "step": 20188
    },
    {
      "epoch": 0.8976924102974523,
      "grad_norm": 0.05794264376163483,
      "learning_rate": 2.7998012580467137e-05,
      "loss": 1.446,
      "step": 20190
    },
    {
      "epoch": 0.8977813347561246,
      "grad_norm": 0.05829514190554619,
      "learning_rate": 2.7949800160212312e-05,
      "loss": 1.4488,
      "step": 20192
    },
    {
      "epoch": 0.897870259214797,
      "grad_norm": 0.05872653052210808,
      "learning_rate": 2.790162809288882e-05,
      "loss": 1.4526,
      "step": 20194
    },
    {
      "epoch": 0.8979591836734694,
      "grad_norm": 0.0570625476539135,
      "learning_rate": 2.7853496382614695e-05,
      "loss": 1.4446,
      "step": 20196
    },
    {
      "epoch": 0.8980481081321418,
      "grad_norm": 0.058061014860868454,
      "learning_rate": 2.7805405033504317e-05,
      "loss": 1.4446,
      "step": 20198
    },
    {
      "epoch": 0.8981370325908141,
      "grad_norm": 0.05758427456021309,
      "learning_rate": 2.7757354049668783e-05,
      "loss": 1.4426,
      "step": 20200
    },
    {
      "epoch": 0.8982259570494865,
      "grad_norm": 0.05844208970665932,
      "learning_rate": 2.770934343521575e-05,
      "loss": 1.4509,
      "step": 20202
    },
    {
      "epoch": 0.8983148815081589,
      "grad_norm": 0.05929677188396454,
      "learning_rate": 2.766137319424944e-05,
      "loss": 1.4505,
      "step": 20204
    },
    {
      "epoch": 0.8984038059668312,
      "grad_norm": 0.05805506557226181,
      "learning_rate": 2.761344333087046e-05,
      "loss": 1.4492,
      "step": 20206
    },
    {
      "epoch": 0.8984927304255035,
      "grad_norm": 0.05872786045074463,
      "learning_rate": 2.756555384917603e-05,
      "loss": 1.4478,
      "step": 20208
    },
    {
      "epoch": 0.8985816548841759,
      "grad_norm": 0.057989608496427536,
      "learning_rate": 2.751770475326004e-05,
      "loss": 1.4431,
      "step": 20210
    },
    {
      "epoch": 0.8986705793428482,
      "grad_norm": 0.05771687999367714,
      "learning_rate": 2.746989604721284e-05,
      "loss": 1.4469,
      "step": 20212
    },
    {
      "epoch": 0.8987595038015206,
      "grad_norm": 0.05826197937130928,
      "learning_rate": 2.7422127735121206e-05,
      "loss": 1.4516,
      "step": 20214
    },
    {
      "epoch": 0.898848428260193,
      "grad_norm": 0.058696992695331573,
      "learning_rate": 2.7374399821068762e-05,
      "loss": 1.4488,
      "step": 20216
    },
    {
      "epoch": 0.8989373527188653,
      "grad_norm": 0.05959450826048851,
      "learning_rate": 2.7326712309135414e-05,
      "loss": 1.4471,
      "step": 20218
    },
    {
      "epoch": 0.8990262771775377,
      "grad_norm": 0.05827097222208977,
      "learning_rate": 2.7279065203397678e-05,
      "loss": 1.4493,
      "step": 20220
    },
    {
      "epoch": 0.8991152016362101,
      "grad_norm": 0.05881284549832344,
      "learning_rate": 2.723145850792863e-05,
      "loss": 1.4558,
      "step": 20222
    },
    {
      "epoch": 0.8992041260948824,
      "grad_norm": 0.057856813073158264,
      "learning_rate": 2.71838922267979e-05,
      "loss": 1.4469,
      "step": 20224
    },
    {
      "epoch": 0.8992930505535548,
      "grad_norm": 0.05925833061337471,
      "learning_rate": 2.7136366364071797e-05,
      "loss": 1.4445,
      "step": 20226
    },
    {
      "epoch": 0.8993819750122272,
      "grad_norm": 0.05858567729592323,
      "learning_rate": 2.70888809238129e-05,
      "loss": 1.4451,
      "step": 20228
    },
    {
      "epoch": 0.8994708994708994,
      "grad_norm": 0.05886606499552727,
      "learning_rate": 2.704143591008046e-05,
      "loss": 1.4467,
      "step": 20230
    },
    {
      "epoch": 0.8995598239295718,
      "grad_norm": 0.05794280394911766,
      "learning_rate": 2.6994031326930456e-05,
      "loss": 1.4432,
      "step": 20232
    },
    {
      "epoch": 0.8996487483882442,
      "grad_norm": 0.05774116888642311,
      "learning_rate": 2.6946667178415097e-05,
      "loss": 1.4478,
      "step": 20234
    },
    {
      "epoch": 0.8997376728469165,
      "grad_norm": 0.05794171988964081,
      "learning_rate": 2.689934346858314e-05,
      "loss": 1.4489,
      "step": 20236
    },
    {
      "epoch": 0.8998265973055889,
      "grad_norm": 0.05861863121390343,
      "learning_rate": 2.68520602014804e-05,
      "loss": 1.4459,
      "step": 20238
    },
    {
      "epoch": 0.8999155217642613,
      "grad_norm": 0.057649560272693634,
      "learning_rate": 2.6804817381148537e-05,
      "loss": 1.4465,
      "step": 20240
    },
    {
      "epoch": 0.9000044462229336,
      "grad_norm": 0.05889919400215149,
      "learning_rate": 2.6757615011626212e-05,
      "loss": 1.4519,
      "step": 20242
    },
    {
      "epoch": 0.900093370681606,
      "grad_norm": 0.058443833142519,
      "learning_rate": 2.6710453096948473e-05,
      "loss": 1.4464,
      "step": 20244
    },
    {
      "epoch": 0.9001822951402784,
      "grad_norm": 0.0582452267408371,
      "learning_rate": 2.6663331641146872e-05,
      "loss": 1.4514,
      "step": 20246
    },
    {
      "epoch": 0.9002712195989507,
      "grad_norm": 0.058483488857746124,
      "learning_rate": 2.661625064824974e-05,
      "loss": 1.4484,
      "step": 20248
    },
    {
      "epoch": 0.9003601440576231,
      "grad_norm": 0.05783051252365112,
      "learning_rate": 2.656921012228153e-05,
      "loss": 1.4494,
      "step": 20250
    },
    {
      "epoch": 0.9004490685162954,
      "grad_norm": 0.05933910235762596,
      "learning_rate": 2.6522210067263573e-05,
      "loss": 1.4465,
      "step": 20252
    },
    {
      "epoch": 0.9005379929749677,
      "grad_norm": 0.0582958348095417,
      "learning_rate": 2.647525048721372e-05,
      "loss": 1.4489,
      "step": 20254
    },
    {
      "epoch": 0.9006269174336401,
      "grad_norm": 0.0582851879298687,
      "learning_rate": 2.6428331386146143e-05,
      "loss": 1.4492,
      "step": 20256
    },
    {
      "epoch": 0.9007158418923125,
      "grad_norm": 0.059080787003040314,
      "learning_rate": 2.6381452768071747e-05,
      "loss": 1.4483,
      "step": 20258
    },
    {
      "epoch": 0.9008047663509848,
      "grad_norm": 0.05819986015558243,
      "learning_rate": 2.6334614636998045e-05,
      "loss": 1.4498,
      "step": 20260
    },
    {
      "epoch": 0.9008936908096572,
      "grad_norm": 0.058520715683698654,
      "learning_rate": 2.6287816996928783e-05,
      "loss": 1.4497,
      "step": 20262
    },
    {
      "epoch": 0.9009826152683296,
      "grad_norm": 0.0575350746512413,
      "learning_rate": 2.624105985186448e-05,
      "loss": 1.4402,
      "step": 20264
    },
    {
      "epoch": 0.9010715397270019,
      "grad_norm": 0.05844678357243538,
      "learning_rate": 2.6194343205802272e-05,
      "loss": 1.4483,
      "step": 20266
    },
    {
      "epoch": 0.9011604641856743,
      "grad_norm": 0.05773119255900383,
      "learning_rate": 2.6147667062735468e-05,
      "loss": 1.443,
      "step": 20268
    },
    {
      "epoch": 0.9012493886443467,
      "grad_norm": 0.05865318328142166,
      "learning_rate": 2.610103142665443e-05,
      "loss": 1.4502,
      "step": 20270
    },
    {
      "epoch": 0.901338313103019,
      "grad_norm": 0.05893605947494507,
      "learning_rate": 2.6054436301545526e-05,
      "loss": 1.4491,
      "step": 20272
    },
    {
      "epoch": 0.9014272375616913,
      "grad_norm": 0.059134792536497116,
      "learning_rate": 2.600788169139201e-05,
      "loss": 1.443,
      "step": 20274
    },
    {
      "epoch": 0.9015161620203637,
      "grad_norm": 0.05810784548521042,
      "learning_rate": 2.5961367600173646e-05,
      "loss": 1.4503,
      "step": 20276
    },
    {
      "epoch": 0.901605086479036,
      "grad_norm": 0.05842362344264984,
      "learning_rate": 2.5914894031866533e-05,
      "loss": 1.4463,
      "step": 20278
    },
    {
      "epoch": 0.9016940109377084,
      "grad_norm": 0.05697064474225044,
      "learning_rate": 2.5868460990443542e-05,
      "loss": 1.4439,
      "step": 20280
    },
    {
      "epoch": 0.9017829353963808,
      "grad_norm": 0.05963601917028427,
      "learning_rate": 2.582206847987395e-05,
      "loss": 1.4428,
      "step": 20282
    },
    {
      "epoch": 0.9018718598550531,
      "grad_norm": 0.05725273862481117,
      "learning_rate": 2.5775716504123525e-05,
      "loss": 1.4458,
      "step": 20284
    },
    {
      "epoch": 0.9019607843137255,
      "grad_norm": 0.05789296701550484,
      "learning_rate": 2.572940506715471e-05,
      "loss": 1.4422,
      "step": 20286
    },
    {
      "epoch": 0.9020497087723979,
      "grad_norm": 0.05861978977918625,
      "learning_rate": 2.5683134172926446e-05,
      "loss": 1.451,
      "step": 20288
    },
    {
      "epoch": 0.9021386332310702,
      "grad_norm": 0.05937764793634415,
      "learning_rate": 2.5636903825394074e-05,
      "loss": 1.45,
      "step": 20290
    },
    {
      "epoch": 0.9022275576897426,
      "grad_norm": 0.05819464474916458,
      "learning_rate": 2.559071402850971e-05,
      "loss": 1.4512,
      "step": 20292
    },
    {
      "epoch": 0.9023164821484149,
      "grad_norm": 0.05875251069664955,
      "learning_rate": 2.554456478622158e-05,
      "loss": 1.4478,
      "step": 20294
    },
    {
      "epoch": 0.9024054066070872,
      "grad_norm": 0.05768872797489166,
      "learning_rate": 2.549845610247503e-05,
      "loss": 1.4479,
      "step": 20296
    },
    {
      "epoch": 0.9024943310657596,
      "grad_norm": 0.05741060525178909,
      "learning_rate": 2.5452387981211523e-05,
      "loss": 1.4453,
      "step": 20298
    },
    {
      "epoch": 0.902583255524432,
      "grad_norm": 0.05832884460687637,
      "learning_rate": 2.5406360426369123e-05,
      "loss": 1.4453,
      "step": 20300
    },
    {
      "epoch": 0.9026721799831043,
      "grad_norm": 0.05834457650780678,
      "learning_rate": 2.5360373441882467e-05,
      "loss": 1.4494,
      "step": 20302
    },
    {
      "epoch": 0.9027611044417767,
      "grad_norm": 0.05867211893200874,
      "learning_rate": 2.531442703168285e-05,
      "loss": 1.4489,
      "step": 20304
    },
    {
      "epoch": 0.9028500289004491,
      "grad_norm": 0.057918209582567215,
      "learning_rate": 2.5268521199697746e-05,
      "loss": 1.4454,
      "step": 20306
    },
    {
      "epoch": 0.9029389533591214,
      "grad_norm": 0.057650770992040634,
      "learning_rate": 2.5222655949851626e-05,
      "loss": 1.4484,
      "step": 20308
    },
    {
      "epoch": 0.9030278778177938,
      "grad_norm": 0.05866733193397522,
      "learning_rate": 2.517683128606507e-05,
      "loss": 1.4527,
      "step": 20310
    },
    {
      "epoch": 0.9031168022764662,
      "grad_norm": 0.057978980243206024,
      "learning_rate": 2.5131047212255566e-05,
      "loss": 1.448,
      "step": 20312
    },
    {
      "epoch": 0.9032057267351385,
      "grad_norm": 0.05772881582379341,
      "learning_rate": 2.508530373233675e-05,
      "loss": 1.4488,
      "step": 20314
    },
    {
      "epoch": 0.9032946511938108,
      "grad_norm": 0.057447269558906555,
      "learning_rate": 2.5039600850219058e-05,
      "loss": 1.4449,
      "step": 20316
    },
    {
      "epoch": 0.9033835756524832,
      "grad_norm": 0.057314515113830566,
      "learning_rate": 2.4993938569809416e-05,
      "loss": 1.445,
      "step": 20318
    },
    {
      "epoch": 0.9034725001111555,
      "grad_norm": 0.058126822113990784,
      "learning_rate": 2.4948316895011204e-05,
      "loss": 1.4505,
      "step": 20320
    },
    {
      "epoch": 0.9035614245698279,
      "grad_norm": 0.05834449455142021,
      "learning_rate": 2.490273582972419e-05,
      "loss": 1.4474,
      "step": 20322
    },
    {
      "epoch": 0.9036503490285003,
      "grad_norm": 0.0587538406252861,
      "learning_rate": 2.4857195377845144e-05,
      "loss": 1.4403,
      "step": 20324
    },
    {
      "epoch": 0.9037392734871726,
      "grad_norm": 0.05739228054881096,
      "learning_rate": 2.4811695543266842e-05,
      "loss": 1.4439,
      "step": 20326
    },
    {
      "epoch": 0.903828197945845,
      "grad_norm": 0.057395629584789276,
      "learning_rate": 2.476623632987901e-05,
      "loss": 1.4444,
      "step": 20328
    },
    {
      "epoch": 0.9039171224045174,
      "grad_norm": 0.05809532478451729,
      "learning_rate": 2.472081774156748e-05,
      "loss": 1.4442,
      "step": 20330
    },
    {
      "epoch": 0.9040060468631897,
      "grad_norm": 0.05822272226214409,
      "learning_rate": 2.4675439782214924e-05,
      "loss": 1.4502,
      "step": 20332
    },
    {
      "epoch": 0.9040949713218621,
      "grad_norm": 0.059671252965927124,
      "learning_rate": 2.4630102455700577e-05,
      "loss": 1.444,
      "step": 20334
    },
    {
      "epoch": 0.9041838957805345,
      "grad_norm": 0.05677417665719986,
      "learning_rate": 2.4584805765899885e-05,
      "loss": 1.4455,
      "step": 20336
    },
    {
      "epoch": 0.9042728202392067,
      "grad_norm": 0.058038029819726944,
      "learning_rate": 2.4539549716685038e-05,
      "loss": 1.4463,
      "step": 20338
    },
    {
      "epoch": 0.9043617446978791,
      "grad_norm": 0.057455874979496,
      "learning_rate": 2.449433431192488e-05,
      "loss": 1.4421,
      "step": 20340
    },
    {
      "epoch": 0.9044506691565515,
      "grad_norm": 0.05858970433473587,
      "learning_rate": 2.444915955548449e-05,
      "loss": 1.4471,
      "step": 20342
    },
    {
      "epoch": 0.9045395936152238,
      "grad_norm": 0.05943557992577553,
      "learning_rate": 2.4404025451225497e-05,
      "loss": 1.4534,
      "step": 20344
    },
    {
      "epoch": 0.9046285180738962,
      "grad_norm": 0.058774013072252274,
      "learning_rate": 2.4358932003006427e-05,
      "loss": 1.447,
      "step": 20346
    },
    {
      "epoch": 0.9047174425325686,
      "grad_norm": 0.05879400297999382,
      "learning_rate": 2.4313879214681866e-05,
      "loss": 1.4474,
      "step": 20348
    },
    {
      "epoch": 0.904806366991241,
      "grad_norm": 0.05844084918498993,
      "learning_rate": 2.4268867090103232e-05,
      "loss": 1.4454,
      "step": 20350
    },
    {
      "epoch": 0.9048952914499133,
      "grad_norm": 0.05857951194047928,
      "learning_rate": 2.4223895633118276e-05,
      "loss": 1.4535,
      "step": 20352
    },
    {
      "epoch": 0.9049842159085857,
      "grad_norm": 0.05743425339460373,
      "learning_rate": 2.417896484757137e-05,
      "loss": 1.4417,
      "step": 20354
    },
    {
      "epoch": 0.905073140367258,
      "grad_norm": 0.056919701397418976,
      "learning_rate": 2.4134074737303558e-05,
      "loss": 1.4449,
      "step": 20356
    },
    {
      "epoch": 0.9051620648259304,
      "grad_norm": 0.05839109793305397,
      "learning_rate": 2.4089225306151986e-05,
      "loss": 1.4459,
      "step": 20358
    },
    {
      "epoch": 0.9052509892846027,
      "grad_norm": 0.05757851153612137,
      "learning_rate": 2.40444165579507e-05,
      "loss": 1.4408,
      "step": 20360
    },
    {
      "epoch": 0.905339913743275,
      "grad_norm": 0.057318564504384995,
      "learning_rate": 2.3999648496530247e-05,
      "loss": 1.4473,
      "step": 20362
    },
    {
      "epoch": 0.9054288382019474,
      "grad_norm": 0.05825290456414223,
      "learning_rate": 2.3954921125717454e-05,
      "loss": 1.4484,
      "step": 20364
    },
    {
      "epoch": 0.9055177626606198,
      "grad_norm": 0.057050399482250214,
      "learning_rate": 2.391023444933582e-05,
      "loss": 1.4491,
      "step": 20366
    },
    {
      "epoch": 0.9056066871192922,
      "grad_norm": 0.058133237063884735,
      "learning_rate": 2.3865588471205514e-05,
      "loss": 1.4488,
      "step": 20368
    },
    {
      "epoch": 0.9056956115779645,
      "grad_norm": 0.0579213872551918,
      "learning_rate": 2.3820983195142864e-05,
      "loss": 1.4432,
      "step": 20370
    },
    {
      "epoch": 0.9057845360366369,
      "grad_norm": 0.05713074654340744,
      "learning_rate": 2.377641862496105e-05,
      "loss": 1.4493,
      "step": 20372
    },
    {
      "epoch": 0.9058734604953093,
      "grad_norm": 0.05779387801885605,
      "learning_rate": 2.3731894764469687e-05,
      "loss": 1.4474,
      "step": 20374
    },
    {
      "epoch": 0.9059623849539816,
      "grad_norm": 0.05732658505439758,
      "learning_rate": 2.368741161747473e-05,
      "loss": 1.4462,
      "step": 20376
    },
    {
      "epoch": 0.906051309412654,
      "grad_norm": 0.058318544179201126,
      "learning_rate": 2.3642969187778916e-05,
      "loss": 1.4432,
      "step": 20378
    },
    {
      "epoch": 0.9061402338713264,
      "grad_norm": 0.057634204626083374,
      "learning_rate": 2.359856747918121e-05,
      "loss": 1.4426,
      "step": 20380
    },
    {
      "epoch": 0.9062291583299986,
      "grad_norm": 0.05897272378206253,
      "learning_rate": 2.3554206495477515e-05,
      "loss": 1.4453,
      "step": 20382
    },
    {
      "epoch": 0.906318082788671,
      "grad_norm": 0.05731482058763504,
      "learning_rate": 2.350988624045991e-05,
      "loss": 1.4424,
      "step": 20384
    },
    {
      "epoch": 0.9064070072473434,
      "grad_norm": 0.05799233168363571,
      "learning_rate": 2.3465606717916975e-05,
      "loss": 1.4518,
      "step": 20386
    },
    {
      "epoch": 0.9064959317060157,
      "grad_norm": 0.0577356293797493,
      "learning_rate": 2.3421367931633963e-05,
      "loss": 1.4431,
      "step": 20388
    },
    {
      "epoch": 0.9065848561646881,
      "grad_norm": 0.060062676668167114,
      "learning_rate": 2.3377169885392734e-05,
      "loss": 1.4531,
      "step": 20390
    },
    {
      "epoch": 0.9066737806233605,
      "grad_norm": 0.05856315419077873,
      "learning_rate": 2.333301258297138e-05,
      "loss": 1.4402,
      "step": 20392
    },
    {
      "epoch": 0.9067627050820328,
      "grad_norm": 0.05849577859044075,
      "learning_rate": 2.3288896028144657e-05,
      "loss": 1.4438,
      "step": 20394
    },
    {
      "epoch": 0.9068516295407052,
      "grad_norm": 0.058690641075372696,
      "learning_rate": 2.324482022468394e-05,
      "loss": 1.4547,
      "step": 20396
    },
    {
      "epoch": 0.9069405539993776,
      "grad_norm": 0.05806424096226692,
      "learning_rate": 2.3200785176357042e-05,
      "loss": 1.4501,
      "step": 20398
    },
    {
      "epoch": 0.9070294784580499,
      "grad_norm": 0.05833980441093445,
      "learning_rate": 2.3156790886928237e-05,
      "loss": 1.4476,
      "step": 20400
    },
    {
      "epoch": 0.9071184029167223,
      "grad_norm": 0.0578826479613781,
      "learning_rate": 2.3112837360158122e-05,
      "loss": 1.4482,
      "step": 20402
    },
    {
      "epoch": 0.9072073273753946,
      "grad_norm": 0.057997167110443115,
      "learning_rate": 2.3068924599804475e-05,
      "loss": 1.445,
      "step": 20404
    },
    {
      "epoch": 0.9072962518340669,
      "grad_norm": 0.057818666100502014,
      "learning_rate": 2.3025052609620843e-05,
      "loss": 1.4447,
      "step": 20406
    },
    {
      "epoch": 0.9073851762927393,
      "grad_norm": 0.0576237328350544,
      "learning_rate": 2.2981221393357622e-05,
      "loss": 1.4494,
      "step": 20408
    },
    {
      "epoch": 0.9074741007514117,
      "grad_norm": 0.05845033749938011,
      "learning_rate": 2.2937430954761752e-05,
      "loss": 1.4427,
      "step": 20410
    },
    {
      "epoch": 0.907563025210084,
      "grad_norm": 0.05697134882211685,
      "learning_rate": 2.289368129757674e-05,
      "loss": 1.4486,
      "step": 20412
    },
    {
      "epoch": 0.9076519496687564,
      "grad_norm": 0.05806869640946388,
      "learning_rate": 2.2849972425542265e-05,
      "loss": 1.4464,
      "step": 20414
    },
    {
      "epoch": 0.9077408741274288,
      "grad_norm": 0.05777909234166145,
      "learning_rate": 2.280630434239489e-05,
      "loss": 1.4493,
      "step": 20416
    },
    {
      "epoch": 0.9078297985861011,
      "grad_norm": 0.05883679911494255,
      "learning_rate": 2.2762677051867577e-05,
      "loss": 1.4485,
      "step": 20418
    },
    {
      "epoch": 0.9079187230447735,
      "grad_norm": 0.05816845968365669,
      "learning_rate": 2.2719090557689782e-05,
      "loss": 1.4434,
      "step": 20420
    },
    {
      "epoch": 0.9080076475034459,
      "grad_norm": 0.05854851007461548,
      "learning_rate": 2.267554486358747e-05,
      "loss": 1.4494,
      "step": 20422
    },
    {
      "epoch": 0.9080965719621181,
      "grad_norm": 0.056842558085918427,
      "learning_rate": 2.2632039973282947e-05,
      "loss": 1.4488,
      "step": 20424
    },
    {
      "epoch": 0.9081854964207905,
      "grad_norm": 0.05853312835097313,
      "learning_rate": 2.25885758904954e-05,
      "loss": 1.4544,
      "step": 20426
    },
    {
      "epoch": 0.9082744208794629,
      "grad_norm": 0.058467112481594086,
      "learning_rate": 2.2545152618940355e-05,
      "loss": 1.4481,
      "step": 20428
    },
    {
      "epoch": 0.9083633453381352,
      "grad_norm": 0.05814722180366516,
      "learning_rate": 2.2501770162329626e-05,
      "loss": 1.445,
      "step": 20430
    },
    {
      "epoch": 0.9084522697968076,
      "grad_norm": 0.05995301157236099,
      "learning_rate": 2.2458428524371855e-05,
      "loss": 1.4447,
      "step": 20432
    },
    {
      "epoch": 0.90854119425548,
      "grad_norm": 0.05826203152537346,
      "learning_rate": 2.2415127708772077e-05,
      "loss": 1.4434,
      "step": 20434
    },
    {
      "epoch": 0.9086301187141523,
      "grad_norm": 0.05822010338306427,
      "learning_rate": 2.2371867719231943e-05,
      "loss": 1.4443,
      "step": 20436
    },
    {
      "epoch": 0.9087190431728247,
      "grad_norm": 0.057935092598199844,
      "learning_rate": 2.2328648559449273e-05,
      "loss": 1.4474,
      "step": 20438
    },
    {
      "epoch": 0.9088079676314971,
      "grad_norm": 0.057203538715839386,
      "learning_rate": 2.2285470233118778e-05,
      "loss": 1.4479,
      "step": 20440
    },
    {
      "epoch": 0.9088968920901694,
      "grad_norm": 0.05857103317975998,
      "learning_rate": 2.2242332743931558e-05,
      "loss": 1.4451,
      "step": 20442
    },
    {
      "epoch": 0.9089858165488418,
      "grad_norm": 0.058484774082899094,
      "learning_rate": 2.219923609557506e-05,
      "loss": 1.4552,
      "step": 20444
    },
    {
      "epoch": 0.9090747410075141,
      "grad_norm": 0.058287981897592545,
      "learning_rate": 2.2156180291733497e-05,
      "loss": 1.4511,
      "step": 20446
    },
    {
      "epoch": 0.9091636654661864,
      "grad_norm": 0.057483308017253876,
      "learning_rate": 2.2113165336087482e-05,
      "loss": 1.449,
      "step": 20448
    },
    {
      "epoch": 0.9092525899248588,
      "grad_norm": 0.05801312252879143,
      "learning_rate": 2.207019123231402e-05,
      "loss": 1.4427,
      "step": 20450
    },
    {
      "epoch": 0.9093415143835312,
      "grad_norm": 0.05689322575926781,
      "learning_rate": 2.202725798408678e-05,
      "loss": 1.4482,
      "step": 20452
    },
    {
      "epoch": 0.9094304388422035,
      "grad_norm": 0.05802132561802864,
      "learning_rate": 2.1984365595075882e-05,
      "loss": 1.4516,
      "step": 20454
    },
    {
      "epoch": 0.9095193633008759,
      "grad_norm": 0.058080337941646576,
      "learning_rate": 2.1941514068947955e-05,
      "loss": 1.4435,
      "step": 20456
    },
    {
      "epoch": 0.9096082877595483,
      "grad_norm": 0.056655969470739365,
      "learning_rate": 2.1898703409366228e-05,
      "loss": 1.4453,
      "step": 20458
    },
    {
      "epoch": 0.9096972122182206,
      "grad_norm": 0.057819753885269165,
      "learning_rate": 2.1855933619990166e-05,
      "loss": 1.451,
      "step": 20460
    },
    {
      "epoch": 0.909786136676893,
      "grad_norm": 0.057411663234233856,
      "learning_rate": 2.181320470447601e-05,
      "loss": 1.4467,
      "step": 20462
    },
    {
      "epoch": 0.9098750611355654,
      "grad_norm": 0.05854671448469162,
      "learning_rate": 2.1770516666476502e-05,
      "loss": 1.4534,
      "step": 20464
    },
    {
      "epoch": 0.9099639855942377,
      "grad_norm": 0.05828188359737396,
      "learning_rate": 2.172786950964062e-05,
      "loss": 1.4449,
      "step": 20466
    },
    {
      "epoch": 0.91005291005291,
      "grad_norm": 0.05763471499085426,
      "learning_rate": 2.168526323761416e-05,
      "loss": 1.4474,
      "step": 20468
    },
    {
      "epoch": 0.9101418345115824,
      "grad_norm": 0.05832042172551155,
      "learning_rate": 2.1642697854039272e-05,
      "loss": 1.4473,
      "step": 20470
    },
    {
      "epoch": 0.9102307589702547,
      "grad_norm": 0.05753203481435776,
      "learning_rate": 2.1600173362554597e-05,
      "loss": 1.453,
      "step": 20472
    },
    {
      "epoch": 0.9103196834289271,
      "grad_norm": 0.05804789811372757,
      "learning_rate": 2.155768976679534e-05,
      "loss": 1.4507,
      "step": 20474
    },
    {
      "epoch": 0.9104086078875995,
      "grad_norm": 0.05770070105791092,
      "learning_rate": 2.151524707039326e-05,
      "loss": 1.449,
      "step": 20476
    },
    {
      "epoch": 0.9104975323462718,
      "grad_norm": 0.05669693648815155,
      "learning_rate": 2.1472845276976404e-05,
      "loss": 1.4436,
      "step": 20478
    },
    {
      "epoch": 0.9105864568049442,
      "grad_norm": 0.05826473981142044,
      "learning_rate": 2.1430484390169648e-05,
      "loss": 1.4424,
      "step": 20480
    },
    {
      "epoch": 0.9106753812636166,
      "grad_norm": 0.05782058462500572,
      "learning_rate": 2.1388164413593924e-05,
      "loss": 1.4464,
      "step": 20482
    },
    {
      "epoch": 0.910764305722289,
      "grad_norm": 0.05837662145495415,
      "learning_rate": 2.1345885350867123e-05,
      "loss": 1.4528,
      "step": 20484
    },
    {
      "epoch": 0.9108532301809613,
      "grad_norm": 0.058443278074264526,
      "learning_rate": 2.130364720560346e-05,
      "loss": 1.4433,
      "step": 20486
    },
    {
      "epoch": 0.9109421546396337,
      "grad_norm": 0.057444412261247635,
      "learning_rate": 2.126144998141355e-05,
      "loss": 1.451,
      "step": 20488
    },
    {
      "epoch": 0.9110310790983059,
      "grad_norm": 0.0574500598013401,
      "learning_rate": 2.121929368190456e-05,
      "loss": 1.4479,
      "step": 20490
    },
    {
      "epoch": 0.9111200035569783,
      "grad_norm": 0.05753646418452263,
      "learning_rate": 2.117717831068039e-05,
      "loss": 1.4474,
      "step": 20492
    },
    {
      "epoch": 0.9112089280156507,
      "grad_norm": 0.058596670627593994,
      "learning_rate": 2.1135103871341043e-05,
      "loss": 1.4481,
      "step": 20494
    },
    {
      "epoch": 0.911297852474323,
      "grad_norm": 0.05796939134597778,
      "learning_rate": 2.109307036748326e-05,
      "loss": 1.4488,
      "step": 20496
    },
    {
      "epoch": 0.9113867769329954,
      "grad_norm": 0.058441001921892166,
      "learning_rate": 2.1051077802700435e-05,
      "loss": 1.4437,
      "step": 20498
    },
    {
      "epoch": 0.9114757013916678,
      "grad_norm": 0.05807716026902199,
      "learning_rate": 2.100912618058204e-05,
      "loss": 1.4438,
      "step": 20500
    },
    {
      "epoch": 0.9114757013916678,
      "eval_loss": 1.433039903640747,
      "eval_runtime": 12.5537,
      "eval_samples_per_second": 550.436,
      "eval_steps_per_second": 68.824,
      "step": 20500
    },
    {
      "epoch": 0.9115646258503401,
      "grad_norm": 0.05796836316585541,
      "learning_rate": 2.0967215504714366e-05,
      "loss": 1.4506,
      "step": 20502
    },
    {
      "epoch": 0.9116535503090125,
      "grad_norm": 0.05753817781805992,
      "learning_rate": 2.092534577868016e-05,
      "loss": 1.4498,
      "step": 20504
    },
    {
      "epoch": 0.9117424747676849,
      "grad_norm": 0.05816453695297241,
      "learning_rate": 2.088351700605867e-05,
      "loss": 1.4505,
      "step": 20506
    },
    {
      "epoch": 0.9118313992263573,
      "grad_norm": 0.05715755373239517,
      "learning_rate": 2.084172919042554e-05,
      "loss": 1.4524,
      "step": 20508
    },
    {
      "epoch": 0.9119203236850296,
      "grad_norm": 0.05733273923397064,
      "learning_rate": 2.079998233535285e-05,
      "loss": 1.4492,
      "step": 20510
    },
    {
      "epoch": 0.9120092481437019,
      "grad_norm": 0.057934537529945374,
      "learning_rate": 2.075827644440953e-05,
      "loss": 1.4469,
      "step": 20512
    },
    {
      "epoch": 0.9120981726023742,
      "grad_norm": 0.056834906339645386,
      "learning_rate": 2.0716611521160776e-05,
      "loss": 1.4416,
      "step": 20514
    },
    {
      "epoch": 0.9121870970610466,
      "grad_norm": 0.05724672973155975,
      "learning_rate": 2.067498756916808e-05,
      "loss": 1.4451,
      "step": 20516
    },
    {
      "epoch": 0.912276021519719,
      "grad_norm": 0.05750434473156929,
      "learning_rate": 2.0633404591989757e-05,
      "loss": 1.4472,
      "step": 20518
    },
    {
      "epoch": 0.9123649459783914,
      "grad_norm": 0.05747924745082855,
      "learning_rate": 2.059186259318052e-05,
      "loss": 1.4487,
      "step": 20520
    },
    {
      "epoch": 0.9124538704370637,
      "grad_norm": 0.057014673948287964,
      "learning_rate": 2.0550361576291645e-05,
      "loss": 1.4518,
      "step": 20522
    },
    {
      "epoch": 0.9125427948957361,
      "grad_norm": 0.05780208110809326,
      "learning_rate": 2.050890154487062e-05,
      "loss": 1.4477,
      "step": 20524
    },
    {
      "epoch": 0.9126317193544085,
      "grad_norm": 0.0579492561519146,
      "learning_rate": 2.0467482502461775e-05,
      "loss": 1.4455,
      "step": 20526
    },
    {
      "epoch": 0.9127206438130808,
      "grad_norm": 0.058637332171201706,
      "learning_rate": 2.0426104452605786e-05,
      "loss": 1.4503,
      "step": 20528
    },
    {
      "epoch": 0.9128095682717532,
      "grad_norm": 0.05861702188849449,
      "learning_rate": 2.038476739883982e-05,
      "loss": 1.447,
      "step": 20530
    },
    {
      "epoch": 0.9128984927304254,
      "grad_norm": 0.05932488292455673,
      "learning_rate": 2.034347134469744e-05,
      "loss": 1.4504,
      "step": 20532
    },
    {
      "epoch": 0.9129874171890978,
      "grad_norm": 0.05963583663105965,
      "learning_rate": 2.0302216293708987e-05,
      "loss": 1.4429,
      "step": 20534
    },
    {
      "epoch": 0.9130763416477702,
      "grad_norm": 0.05832473933696747,
      "learning_rate": 2.0261002249401028e-05,
      "loss": 1.4492,
      "step": 20536
    },
    {
      "epoch": 0.9131652661064426,
      "grad_norm": 0.058115072548389435,
      "learning_rate": 2.021982921529675e-05,
      "loss": 1.4511,
      "step": 20538
    },
    {
      "epoch": 0.9132541905651149,
      "grad_norm": 0.057589709758758545,
      "learning_rate": 2.0178697194915717e-05,
      "loss": 1.4502,
      "step": 20540
    },
    {
      "epoch": 0.9133431150237873,
      "grad_norm": 0.05809624865651131,
      "learning_rate": 2.0137606191774183e-05,
      "loss": 1.4538,
      "step": 20542
    },
    {
      "epoch": 0.9134320394824597,
      "grad_norm": 0.058290865272283554,
      "learning_rate": 2.009655620938483e-05,
      "loss": 1.4441,
      "step": 20544
    },
    {
      "epoch": 0.913520963941132,
      "grad_norm": 0.05669715628027916,
      "learning_rate": 2.0055547251256633e-05,
      "loss": 1.444,
      "step": 20546
    },
    {
      "epoch": 0.9136098883998044,
      "grad_norm": 0.05922799929976463,
      "learning_rate": 2.0014579320895344e-05,
      "loss": 1.4514,
      "step": 20548
    },
    {
      "epoch": 0.9136988128584768,
      "grad_norm": 0.058014314621686935,
      "learning_rate": 1.9973652421803045e-05,
      "loss": 1.4445,
      "step": 20550
    },
    {
      "epoch": 0.9137877373171491,
      "grad_norm": 0.058082759380340576,
      "learning_rate": 1.9932766557478333e-05,
      "loss": 1.4509,
      "step": 20552
    },
    {
      "epoch": 0.9138766617758214,
      "grad_norm": 0.05879303067922592,
      "learning_rate": 1.9891921731416297e-05,
      "loss": 1.4449,
      "step": 20554
    },
    {
      "epoch": 0.9139655862344938,
      "grad_norm": 0.05874144658446312,
      "learning_rate": 1.9851117947108643e-05,
      "loss": 1.4441,
      "step": 20556
    },
    {
      "epoch": 0.9140545106931661,
      "grad_norm": 0.057445310056209564,
      "learning_rate": 1.9810355208043306e-05,
      "loss": 1.4476,
      "step": 20558
    },
    {
      "epoch": 0.9141434351518385,
      "grad_norm": 0.05829305201768875,
      "learning_rate": 1.976963351770489e-05,
      "loss": 1.4493,
      "step": 20560
    },
    {
      "epoch": 0.9142323596105109,
      "grad_norm": 0.059354912489652634,
      "learning_rate": 1.9728952879574604e-05,
      "loss": 1.4551,
      "step": 20562
    },
    {
      "epoch": 0.9143212840691832,
      "grad_norm": 0.057627931237220764,
      "learning_rate": 1.968831329712989e-05,
      "loss": 1.4429,
      "step": 20564
    },
    {
      "epoch": 0.9144102085278556,
      "grad_norm": 0.0584438256919384,
      "learning_rate": 1.9647714773844804e-05,
      "loss": 1.4498,
      "step": 20566
    },
    {
      "epoch": 0.914499132986528,
      "grad_norm": 0.05829082801938057,
      "learning_rate": 1.9607157313189904e-05,
      "loss": 1.4472,
      "step": 20568
    },
    {
      "epoch": 0.9145880574452003,
      "grad_norm": 0.057573914527893066,
      "learning_rate": 1.9566640918632194e-05,
      "loss": 1.4428,
      "step": 20570
    },
    {
      "epoch": 0.9146769819038727,
      "grad_norm": 0.05751367285847664,
      "learning_rate": 1.9526165593635235e-05,
      "loss": 1.4479,
      "step": 20572
    },
    {
      "epoch": 0.9147659063625451,
      "grad_norm": 0.05755585804581642,
      "learning_rate": 1.948573134165904e-05,
      "loss": 1.4433,
      "step": 20574
    },
    {
      "epoch": 0.9148548308212173,
      "grad_norm": 0.0580437146127224,
      "learning_rate": 1.9445338166160064e-05,
      "loss": 1.448,
      "step": 20576
    },
    {
      "epoch": 0.9149437552798897,
      "grad_norm": 0.057542815804481506,
      "learning_rate": 1.9404986070591322e-05,
      "loss": 1.4461,
      "step": 20578
    },
    {
      "epoch": 0.9150326797385621,
      "grad_norm": 0.05818972736597061,
      "learning_rate": 1.936467505840228e-05,
      "loss": 1.4475,
      "step": 20580
    },
    {
      "epoch": 0.9151216041972344,
      "grad_norm": 0.05846571922302246,
      "learning_rate": 1.9324405133038902e-05,
      "loss": 1.4432,
      "step": 20582
    },
    {
      "epoch": 0.9152105286559068,
      "grad_norm": 0.057439543306827545,
      "learning_rate": 1.9284176297943712e-05,
      "loss": 1.4487,
      "step": 20584
    },
    {
      "epoch": 0.9152994531145792,
      "grad_norm": 0.05826476961374283,
      "learning_rate": 1.9243988556555515e-05,
      "loss": 1.4475,
      "step": 20586
    },
    {
      "epoch": 0.9153883775732515,
      "grad_norm": 0.05997009575366974,
      "learning_rate": 1.920384191230984e-05,
      "loss": 1.447,
      "step": 20588
    },
    {
      "epoch": 0.9154773020319239,
      "grad_norm": 0.05828747898340225,
      "learning_rate": 1.9163736368638496e-05,
      "loss": 1.4409,
      "step": 20590
    },
    {
      "epoch": 0.9155662264905963,
      "grad_norm": 0.057352207601070404,
      "learning_rate": 1.912367192896991e-05,
      "loss": 1.4441,
      "step": 20592
    },
    {
      "epoch": 0.9156551509492686,
      "grad_norm": 0.05746382474899292,
      "learning_rate": 1.908364859672912e-05,
      "loss": 1.4532,
      "step": 20594
    },
    {
      "epoch": 0.915744075407941,
      "grad_norm": 0.05716506391763687,
      "learning_rate": 1.9043666375337276e-05,
      "loss": 1.4396,
      "step": 20596
    },
    {
      "epoch": 0.9158329998666133,
      "grad_norm": 0.05829397961497307,
      "learning_rate": 1.9003725268212367e-05,
      "loss": 1.4568,
      "step": 20598
    },
    {
      "epoch": 0.9159219243252856,
      "grad_norm": 0.058943524956703186,
      "learning_rate": 1.8963825278768776e-05,
      "loss": 1.4509,
      "step": 20600
    },
    {
      "epoch": 0.916010848783958,
      "grad_norm": 0.05764896050095558,
      "learning_rate": 1.892396641041716e-05,
      "loss": 1.4512,
      "step": 20602
    },
    {
      "epoch": 0.9160997732426304,
      "grad_norm": 0.05844266340136528,
      "learning_rate": 1.888414866656496e-05,
      "loss": 1.446,
      "step": 20604
    },
    {
      "epoch": 0.9161886977013027,
      "grad_norm": 0.0566147081553936,
      "learning_rate": 1.8844372050615953e-05,
      "loss": 1.4433,
      "step": 20606
    },
    {
      "epoch": 0.9162776221599751,
      "grad_norm": 0.06036548689007759,
      "learning_rate": 1.8804636565970422e-05,
      "loss": 1.4485,
      "step": 20608
    },
    {
      "epoch": 0.9163665466186475,
      "grad_norm": 0.05751558393239975,
      "learning_rate": 1.876494221602504e-05,
      "loss": 1.4479,
      "step": 20610
    },
    {
      "epoch": 0.9164554710773198,
      "grad_norm": 0.05685598403215408,
      "learning_rate": 1.8725289004173198e-05,
      "loss": 1.4439,
      "step": 20612
    },
    {
      "epoch": 0.9165443955359922,
      "grad_norm": 0.05743536353111267,
      "learning_rate": 1.8685676933804575e-05,
      "loss": 1.4509,
      "step": 20614
    },
    {
      "epoch": 0.9166333199946646,
      "grad_norm": 0.05938420072197914,
      "learning_rate": 1.8646106008305352e-05,
      "loss": 1.4504,
      "step": 20616
    },
    {
      "epoch": 0.9167222444533369,
      "grad_norm": 0.058147843927145004,
      "learning_rate": 1.860657623105816e-05,
      "loss": 1.4463,
      "step": 20618
    },
    {
      "epoch": 0.9168111689120092,
      "grad_norm": 0.059247348457574844,
      "learning_rate": 1.85670876054424e-05,
      "loss": 1.45,
      "step": 20620
    },
    {
      "epoch": 0.9169000933706816,
      "grad_norm": 0.05791396647691727,
      "learning_rate": 1.85276401348336e-05,
      "loss": 1.4464,
      "step": 20622
    },
    {
      "epoch": 0.9169890178293539,
      "grad_norm": 0.05942825973033905,
      "learning_rate": 1.8488233822603782e-05,
      "loss": 1.4504,
      "step": 20624
    },
    {
      "epoch": 0.9170779422880263,
      "grad_norm": 0.05869719013571739,
      "learning_rate": 1.8448868672121756e-05,
      "loss": 1.4469,
      "step": 20626
    },
    {
      "epoch": 0.9171668667466987,
      "grad_norm": 0.058402083814144135,
      "learning_rate": 1.8409544686752542e-05,
      "loss": 1.4417,
      "step": 20628
    },
    {
      "epoch": 0.917255791205371,
      "grad_norm": 0.05802400782704353,
      "learning_rate": 1.8370261869857798e-05,
      "loss": 1.4483,
      "step": 20630
    },
    {
      "epoch": 0.9173447156640434,
      "grad_norm": 0.058382607996463776,
      "learning_rate": 1.8331020224795493e-05,
      "loss": 1.445,
      "step": 20632
    },
    {
      "epoch": 0.9174336401227158,
      "grad_norm": 0.05774873495101929,
      "learning_rate": 1.8291819754920224e-05,
      "loss": 1.4541,
      "step": 20634
    },
    {
      "epoch": 0.9175225645813881,
      "grad_norm": 0.05814613774418831,
      "learning_rate": 1.8252660463583036e-05,
      "loss": 1.4459,
      "step": 20636
    },
    {
      "epoch": 0.9176114890400605,
      "grad_norm": 0.05833296850323677,
      "learning_rate": 1.8213542354131474e-05,
      "loss": 1.4463,
      "step": 20638
    },
    {
      "epoch": 0.9177004134987329,
      "grad_norm": 0.058397140353918076,
      "learning_rate": 1.8174465429909358e-05,
      "loss": 1.446,
      "step": 20640
    },
    {
      "epoch": 0.9177893379574051,
      "grad_norm": 0.05853003263473511,
      "learning_rate": 1.8135429694257354e-05,
      "loss": 1.4472,
      "step": 20642
    },
    {
      "epoch": 0.9178782624160775,
      "grad_norm": 0.05980772152543068,
      "learning_rate": 1.809643515051229e-05,
      "loss": 1.4465,
      "step": 20644
    },
    {
      "epoch": 0.9179671868747499,
      "grad_norm": 0.05789174139499664,
      "learning_rate": 1.8057481802007724e-05,
      "loss": 1.4434,
      "step": 20646
    },
    {
      "epoch": 0.9180561113334222,
      "grad_norm": 0.05812879651784897,
      "learning_rate": 1.801856965207338e-05,
      "loss": 1.4429,
      "step": 20648
    },
    {
      "epoch": 0.9181450357920946,
      "grad_norm": 0.05842747539281845,
      "learning_rate": 1.7979698704035708e-05,
      "loss": 1.4446,
      "step": 20650
    },
    {
      "epoch": 0.918233960250767,
      "grad_norm": 0.0583362840116024,
      "learning_rate": 1.794086896121766e-05,
      "loss": 1.452,
      "step": 20652
    },
    {
      "epoch": 0.9183228847094393,
      "grad_norm": 0.059007447212934494,
      "learning_rate": 1.7902080426938417e-05,
      "loss": 1.4548,
      "step": 20654
    },
    {
      "epoch": 0.9184118091681117,
      "grad_norm": 0.057736873626708984,
      "learning_rate": 1.7863333104513934e-05,
      "loss": 1.4478,
      "step": 20656
    },
    {
      "epoch": 0.9185007336267841,
      "grad_norm": 0.05727485939860344,
      "learning_rate": 1.7824626997256456e-05,
      "loss": 1.4414,
      "step": 20658
    },
    {
      "epoch": 0.9185896580854565,
      "grad_norm": 0.059946656227111816,
      "learning_rate": 1.7785962108474663e-05,
      "loss": 1.4492,
      "step": 20660
    },
    {
      "epoch": 0.9186785825441287,
      "grad_norm": 0.05915560573339462,
      "learning_rate": 1.7747338441473914e-05,
      "loss": 1.4452,
      "step": 20662
    },
    {
      "epoch": 0.9187675070028011,
      "grad_norm": 0.05715799331665039,
      "learning_rate": 1.770875599955596e-05,
      "loss": 1.4486,
      "step": 20664
    },
    {
      "epoch": 0.9188564314614734,
      "grad_norm": 0.05892267823219299,
      "learning_rate": 1.7670214786018823e-05,
      "loss": 1.4465,
      "step": 20666
    },
    {
      "epoch": 0.9189453559201458,
      "grad_norm": 0.05755788832902908,
      "learning_rate": 1.7631714804157316e-05,
      "loss": 1.4423,
      "step": 20668
    },
    {
      "epoch": 0.9190342803788182,
      "grad_norm": 0.0570998415350914,
      "learning_rate": 1.7593256057262642e-05,
      "loss": 1.448,
      "step": 20670
    },
    {
      "epoch": 0.9191232048374905,
      "grad_norm": 0.05825738236308098,
      "learning_rate": 1.755483854862222e-05,
      "loss": 1.4507,
      "step": 20672
    },
    {
      "epoch": 0.9192121292961629,
      "grad_norm": 0.057699088007211685,
      "learning_rate": 1.7516462281520372e-05,
      "loss": 1.4468,
      "step": 20674
    },
    {
      "epoch": 0.9193010537548353,
      "grad_norm": 0.05778341740369797,
      "learning_rate": 1.7478127259237474e-05,
      "loss": 1.4453,
      "step": 20676
    },
    {
      "epoch": 0.9193899782135077,
      "grad_norm": 0.05825795978307724,
      "learning_rate": 1.7439833485050682e-05,
      "loss": 1.441,
      "step": 20678
    },
    {
      "epoch": 0.91947890267218,
      "grad_norm": 0.05925915762782097,
      "learning_rate": 1.7401580962233597e-05,
      "loss": 1.4467,
      "step": 20680
    },
    {
      "epoch": 0.9195678271308524,
      "grad_norm": 0.059423577040433884,
      "learning_rate": 1.7363369694055997e-05,
      "loss": 1.4483,
      "step": 20682
    },
    {
      "epoch": 0.9196567515895246,
      "grad_norm": 0.05769523233175278,
      "learning_rate": 1.7325199683784486e-05,
      "loss": 1.4526,
      "step": 20684
    },
    {
      "epoch": 0.919745676048197,
      "grad_norm": 0.05748085305094719,
      "learning_rate": 1.7287070934682013e-05,
      "loss": 1.4489,
      "step": 20686
    },
    {
      "epoch": 0.9198346005068694,
      "grad_norm": 0.05770343542098999,
      "learning_rate": 1.7248983450007914e-05,
      "loss": 1.4407,
      "step": 20688
    },
    {
      "epoch": 0.9199235249655418,
      "grad_norm": 0.05809416621923447,
      "learning_rate": 1.7210937233018143e-05,
      "loss": 1.4469,
      "step": 20690
    },
    {
      "epoch": 0.9200124494242141,
      "grad_norm": 0.05900378152728081,
      "learning_rate": 1.7172932286965093e-05,
      "loss": 1.441,
      "step": 20692
    },
    {
      "epoch": 0.9201013738828865,
      "grad_norm": 0.05754346400499344,
      "learning_rate": 1.71349686150975e-05,
      "loss": 1.4483,
      "step": 20694
    },
    {
      "epoch": 0.9201902983415589,
      "grad_norm": 0.05826412886381149,
      "learning_rate": 1.709704622066077e-05,
      "loss": 1.4425,
      "step": 20696
    },
    {
      "epoch": 0.9202792228002312,
      "grad_norm": 0.05790907144546509,
      "learning_rate": 1.7059165106896478e-05,
      "loss": 1.4424,
      "step": 20698
    },
    {
      "epoch": 0.9203681472589036,
      "grad_norm": 0.05832275003194809,
      "learning_rate": 1.702132527704309e-05,
      "loss": 1.4435,
      "step": 20700
    },
    {
      "epoch": 0.920457071717576,
      "grad_norm": 0.05824443697929382,
      "learning_rate": 1.6983526734335287e-05,
      "loss": 1.4501,
      "step": 20702
    },
    {
      "epoch": 0.9205459961762483,
      "grad_norm": 0.06036430969834328,
      "learning_rate": 1.6945769482004104e-05,
      "loss": 1.4482,
      "step": 20704
    },
    {
      "epoch": 0.9206349206349206,
      "grad_norm": 0.05813805013895035,
      "learning_rate": 1.6908053523277344e-05,
      "loss": 1.4487,
      "step": 20706
    },
    {
      "epoch": 0.920723845093593,
      "grad_norm": 0.05797953903675079,
      "learning_rate": 1.68703788613791e-05,
      "loss": 1.4517,
      "step": 20708
    },
    {
      "epoch": 0.9208127695522653,
      "grad_norm": 0.05762023478746414,
      "learning_rate": 1.68327454995299e-05,
      "loss": 1.4477,
      "step": 20710
    },
    {
      "epoch": 0.9209016940109377,
      "grad_norm": 0.05841365456581116,
      "learning_rate": 1.6795153440946897e-05,
      "loss": 1.4423,
      "step": 20712
    },
    {
      "epoch": 0.9209906184696101,
      "grad_norm": 0.05898358300328255,
      "learning_rate": 1.6757602688843577e-05,
      "loss": 1.4517,
      "step": 20714
    },
    {
      "epoch": 0.9210795429282824,
      "grad_norm": 0.05752791091799736,
      "learning_rate": 1.672009324642998e-05,
      "loss": 1.441,
      "step": 20716
    },
    {
      "epoch": 0.9211684673869548,
      "grad_norm": 0.05729573592543602,
      "learning_rate": 1.6682625116912597e-05,
      "loss": 1.4516,
      "step": 20718
    },
    {
      "epoch": 0.9212573918456272,
      "grad_norm": 0.060360997915267944,
      "learning_rate": 1.66451983034942e-05,
      "loss": 1.451,
      "step": 20720
    },
    {
      "epoch": 0.9213463163042995,
      "grad_norm": 0.058285191655159,
      "learning_rate": 1.6607812809374455e-05,
      "loss": 1.449,
      "step": 20722
    },
    {
      "epoch": 0.9214352407629719,
      "grad_norm": 0.05837748199701309,
      "learning_rate": 1.6570468637749083e-05,
      "loss": 1.4424,
      "step": 20724
    },
    {
      "epoch": 0.9215241652216443,
      "grad_norm": 0.057116538286209106,
      "learning_rate": 1.653316579181041e-05,
      "loss": 1.4439,
      "step": 20726
    },
    {
      "epoch": 0.9216130896803165,
      "grad_norm": 0.05754510685801506,
      "learning_rate": 1.649590427474734e-05,
      "loss": 1.443,
      "step": 20728
    },
    {
      "epoch": 0.9217020141389889,
      "grad_norm": 0.05777478963136673,
      "learning_rate": 1.64586840897451e-05,
      "loss": 1.4462,
      "step": 20730
    },
    {
      "epoch": 0.9217909385976613,
      "grad_norm": 0.05793027579784393,
      "learning_rate": 1.6421505239985422e-05,
      "loss": 1.4449,
      "step": 20732
    },
    {
      "epoch": 0.9218798630563336,
      "grad_norm": 0.057621315121650696,
      "learning_rate": 1.6384367728646542e-05,
      "loss": 1.4483,
      "step": 20734
    },
    {
      "epoch": 0.921968787515006,
      "grad_norm": 0.05741515010595322,
      "learning_rate": 1.634727155890309e-05,
      "loss": 1.4445,
      "step": 20736
    },
    {
      "epoch": 0.9220577119736784,
      "grad_norm": 0.058283522725105286,
      "learning_rate": 1.6310216733926353e-05,
      "loss": 1.4464,
      "step": 20738
    },
    {
      "epoch": 0.9221466364323507,
      "grad_norm": 0.057298872619867325,
      "learning_rate": 1.627320325688375e-05,
      "loss": 1.4422,
      "step": 20740
    },
    {
      "epoch": 0.9222355608910231,
      "grad_norm": 0.05811677128076553,
      "learning_rate": 1.6236231130939473e-05,
      "loss": 1.4502,
      "step": 20742
    },
    {
      "epoch": 0.9223244853496955,
      "grad_norm": 0.056963641196489334,
      "learning_rate": 1.61993003592541e-05,
      "loss": 1.4417,
      "step": 20744
    },
    {
      "epoch": 0.9224134098083678,
      "grad_norm": 0.05848962813615799,
      "learning_rate": 1.6162410944984552e-05,
      "loss": 1.448,
      "step": 20746
    },
    {
      "epoch": 0.9225023342670402,
      "grad_norm": 0.05890442803502083,
      "learning_rate": 1.6125562891284252e-05,
      "loss": 1.4442,
      "step": 20748
    },
    {
      "epoch": 0.9225912587257125,
      "grad_norm": 0.05794266238808632,
      "learning_rate": 1.6088756201303235e-05,
      "loss": 1.4415,
      "step": 20750
    },
    {
      "epoch": 0.9226801831843848,
      "grad_norm": 0.05748625472187996,
      "learning_rate": 1.6051990878187874e-05,
      "loss": 1.4459,
      "step": 20752
    },
    {
      "epoch": 0.9227691076430572,
      "grad_norm": 0.05740351602435112,
      "learning_rate": 1.601526692508104e-05,
      "loss": 1.4488,
      "step": 20754
    },
    {
      "epoch": 0.9228580321017296,
      "grad_norm": 0.05720726400613785,
      "learning_rate": 1.5978584345122005e-05,
      "loss": 1.4466,
      "step": 20756
    },
    {
      "epoch": 0.9229469565604019,
      "grad_norm": 0.058184560388326645,
      "learning_rate": 1.5941943141446593e-05,
      "loss": 1.4507,
      "step": 20758
    },
    {
      "epoch": 0.9230358810190743,
      "grad_norm": 0.05791355296969414,
      "learning_rate": 1.5905343317187126e-05,
      "loss": 1.4435,
      "step": 20760
    },
    {
      "epoch": 0.9231248054777467,
      "grad_norm": 0.05806044861674309,
      "learning_rate": 1.5868784875472166e-05,
      "loss": 1.4496,
      "step": 20762
    },
    {
      "epoch": 0.923213729936419,
      "grad_norm": 0.05714624375104904,
      "learning_rate": 1.5832267819426982e-05,
      "loss": 1.4434,
      "step": 20764
    },
    {
      "epoch": 0.9233026543950914,
      "grad_norm": 0.057807646691799164,
      "learning_rate": 1.579579215217325e-05,
      "loss": 1.4504,
      "step": 20766
    },
    {
      "epoch": 0.9233915788537638,
      "grad_norm": 0.058163121342659,
      "learning_rate": 1.5759357876828974e-05,
      "loss": 1.4481,
      "step": 20768
    },
    {
      "epoch": 0.9234805033124361,
      "grad_norm": 0.059579599648714066,
      "learning_rate": 1.572296499650877e-05,
      "loss": 1.4458,
      "step": 20770
    },
    {
      "epoch": 0.9235694277711084,
      "grad_norm": 0.05729829519987106,
      "learning_rate": 1.568661351432371e-05,
      "loss": 1.4442,
      "step": 20772
    },
    {
      "epoch": 0.9236583522297808,
      "grad_norm": 0.05755337327718735,
      "learning_rate": 1.5650303433381195e-05,
      "loss": 1.445,
      "step": 20774
    },
    {
      "epoch": 0.9237472766884531,
      "grad_norm": 0.057855114340782166,
      "learning_rate": 1.5614034756785188e-05,
      "loss": 1.4457,
      "step": 20776
    },
    {
      "epoch": 0.9238362011471255,
      "grad_norm": 0.05798733979463577,
      "learning_rate": 1.5577807487636154e-05,
      "loss": 1.4439,
      "step": 20778
    },
    {
      "epoch": 0.9239251256057979,
      "grad_norm": 0.059019967913627625,
      "learning_rate": 1.5541621629030832e-05,
      "loss": 1.4456,
      "step": 20780
    },
    {
      "epoch": 0.9240140500644702,
      "grad_norm": 0.057757288217544556,
      "learning_rate": 1.55054771840627e-05,
      "loss": 1.4496,
      "step": 20782
    },
    {
      "epoch": 0.9241029745231426,
      "grad_norm": 0.057486891746520996,
      "learning_rate": 1.546937415582145e-05,
      "loss": 1.4503,
      "step": 20784
    },
    {
      "epoch": 0.924191898981815,
      "grad_norm": 0.05713411793112755,
      "learning_rate": 1.543331254739333e-05,
      "loss": 1.4441,
      "step": 20786
    },
    {
      "epoch": 0.9242808234404873,
      "grad_norm": 0.058376528322696686,
      "learning_rate": 1.539729236186116e-05,
      "loss": 1.4535,
      "step": 20788
    },
    {
      "epoch": 0.9243697478991597,
      "grad_norm": 0.05829921364784241,
      "learning_rate": 1.5361313602303916e-05,
      "loss": 1.4443,
      "step": 20790
    },
    {
      "epoch": 0.924458672357832,
      "grad_norm": 0.0577683262526989,
      "learning_rate": 1.5325376271797364e-05,
      "loss": 1.4487,
      "step": 20792
    },
    {
      "epoch": 0.9245475968165043,
      "grad_norm": 0.0584307499229908,
      "learning_rate": 1.5289480373413545e-05,
      "loss": 1.4399,
      "step": 20794
    },
    {
      "epoch": 0.9246365212751767,
      "grad_norm": 0.05748804286122322,
      "learning_rate": 1.5253625910221003e-05,
      "loss": 1.4444,
      "step": 20796
    },
    {
      "epoch": 0.9247254457338491,
      "grad_norm": 0.058221444487571716,
      "learning_rate": 1.5217812885284677e-05,
      "loss": 1.4441,
      "step": 20798
    },
    {
      "epoch": 0.9248143701925214,
      "grad_norm": 0.05775991082191467,
      "learning_rate": 1.5182041301666115e-05,
      "loss": 1.4454,
      "step": 20800
    },
    {
      "epoch": 0.9249032946511938,
      "grad_norm": 0.05797930434346199,
      "learning_rate": 1.5146311162423155e-05,
      "loss": 1.4487,
      "step": 20802
    },
    {
      "epoch": 0.9249922191098662,
      "grad_norm": 0.05848506838083267,
      "learning_rate": 1.5110622470610236e-05,
      "loss": 1.4563,
      "step": 20804
    },
    {
      "epoch": 0.9250811435685385,
      "grad_norm": 0.05729779601097107,
      "learning_rate": 1.5074975229278087e-05,
      "loss": 1.4438,
      "step": 20806
    },
    {
      "epoch": 0.9251700680272109,
      "grad_norm": 0.058297332376241684,
      "learning_rate": 1.5039369441474105e-05,
      "loss": 1.4454,
      "step": 20808
    },
    {
      "epoch": 0.9252589924858833,
      "grad_norm": 0.057348962873220444,
      "learning_rate": 1.5003805110241964e-05,
      "loss": 1.4499,
      "step": 20810
    },
    {
      "epoch": 0.9253479169445556,
      "grad_norm": 0.05848599225282669,
      "learning_rate": 1.4968282238621844e-05,
      "loss": 1.4407,
      "step": 20812
    },
    {
      "epoch": 0.9254368414032279,
      "grad_norm": 0.058535609394311905,
      "learning_rate": 1.4932800829650429e-05,
      "loss": 1.4499,
      "step": 20814
    },
    {
      "epoch": 0.9255257658619003,
      "grad_norm": 0.05695459619164467,
      "learning_rate": 1.48973608863609e-05,
      "loss": 1.4477,
      "step": 20816
    },
    {
      "epoch": 0.9256146903205726,
      "grad_norm": 0.057941921055316925,
      "learning_rate": 1.486196241178267e-05,
      "loss": 1.4571,
      "step": 20818
    },
    {
      "epoch": 0.925703614779245,
      "grad_norm": 0.05711508169770241,
      "learning_rate": 1.4826605408941818e-05,
      "loss": 1.4482,
      "step": 20820
    },
    {
      "epoch": 0.9257925392379174,
      "grad_norm": 0.05759931355714798,
      "learning_rate": 1.4791289880860814e-05,
      "loss": 1.4482,
      "step": 20822
    },
    {
      "epoch": 0.9258814636965897,
      "grad_norm": 0.05856931582093239,
      "learning_rate": 1.4756015830558634e-05,
      "loss": 1.4511,
      "step": 20824
    },
    {
      "epoch": 0.9259703881552621,
      "grad_norm": 0.05778467282652855,
      "learning_rate": 1.4720783261050641e-05,
      "loss": 1.4477,
      "step": 20826
    },
    {
      "epoch": 0.9260593126139345,
      "grad_norm": 0.05779944732785225,
      "learning_rate": 1.468559217534854e-05,
      "loss": 1.4449,
      "step": 20828
    },
    {
      "epoch": 0.9261482370726069,
      "grad_norm": 0.057623885571956635,
      "learning_rate": 1.4650442576460865e-05,
      "loss": 1.4483,
      "step": 20830
    },
    {
      "epoch": 0.9262371615312792,
      "grad_norm": 0.058261968195438385,
      "learning_rate": 1.4615334467392216e-05,
      "loss": 1.4486,
      "step": 20832
    },
    {
      "epoch": 0.9263260859899516,
      "grad_norm": 0.057926472276449203,
      "learning_rate": 1.4580267851143747e-05,
      "loss": 1.4472,
      "step": 20834
    },
    {
      "epoch": 0.9264150104486238,
      "grad_norm": 0.05730114132165909,
      "learning_rate": 1.4545242730713115e-05,
      "loss": 1.447,
      "step": 20836
    },
    {
      "epoch": 0.9265039349072962,
      "grad_norm": 0.05763164535164833,
      "learning_rate": 1.4510259109094481e-05,
      "loss": 1.4426,
      "step": 20838
    },
    {
      "epoch": 0.9265928593659686,
      "grad_norm": 0.058550890535116196,
      "learning_rate": 1.4475316989278453e-05,
      "loss": 1.4475,
      "step": 20840
    },
    {
      "epoch": 0.926681783824641,
      "grad_norm": 0.05741726979613304,
      "learning_rate": 1.444041637425192e-05,
      "loss": 1.4479,
      "step": 20842
    },
    {
      "epoch": 0.9267707082833133,
      "grad_norm": 0.05805116891860962,
      "learning_rate": 1.4405557266998382e-05,
      "loss": 1.4446,
      "step": 20844
    },
    {
      "epoch": 0.9268596327419857,
      "grad_norm": 0.05792229250073433,
      "learning_rate": 1.437073967049779e-05,
      "loss": 1.4407,
      "step": 20846
    },
    {
      "epoch": 0.926948557200658,
      "grad_norm": 0.0565202422440052,
      "learning_rate": 1.4335963587726431e-05,
      "loss": 1.4491,
      "step": 20848
    },
    {
      "epoch": 0.9270374816593304,
      "grad_norm": 0.05785936489701271,
      "learning_rate": 1.4301229021657147e-05,
      "loss": 1.4506,
      "step": 20850
    },
    {
      "epoch": 0.9271264061180028,
      "grad_norm": 0.05873408913612366,
      "learning_rate": 1.4266535975259287e-05,
      "loss": 1.4455,
      "step": 20852
    },
    {
      "epoch": 0.9272153305766752,
      "grad_norm": 0.05857241526246071,
      "learning_rate": 1.4231884451498478e-05,
      "loss": 1.4426,
      "step": 20854
    },
    {
      "epoch": 0.9273042550353475,
      "grad_norm": 0.05902234464883804,
      "learning_rate": 1.4197274453336795e-05,
      "loss": 1.4501,
      "step": 20856
    },
    {
      "epoch": 0.9273931794940198,
      "grad_norm": 0.05742455646395683,
      "learning_rate": 1.4162705983733038e-05,
      "loss": 1.4464,
      "step": 20858
    },
    {
      "epoch": 0.9274821039526921,
      "grad_norm": 0.05817548558115959,
      "learning_rate": 1.412817904564212e-05,
      "loss": 1.4495,
      "step": 20860
    },
    {
      "epoch": 0.9275710284113645,
      "grad_norm": 0.0579364076256752,
      "learning_rate": 1.4093693642015737e-05,
      "loss": 1.4484,
      "step": 20862
    },
    {
      "epoch": 0.9276599528700369,
      "grad_norm": 0.05716056376695633,
      "learning_rate": 1.4059249775801642e-05,
      "loss": 1.4411,
      "step": 20864
    },
    {
      "epoch": 0.9277488773287093,
      "grad_norm": 0.05732995644211769,
      "learning_rate": 1.4024847449944366e-05,
      "loss": 1.4498,
      "step": 20866
    },
    {
      "epoch": 0.9278378017873816,
      "grad_norm": 0.05792165920138359,
      "learning_rate": 1.3990486667384784e-05,
      "loss": 1.4499,
      "step": 20868
    },
    {
      "epoch": 0.927926726246054,
      "grad_norm": 0.05896197259426117,
      "learning_rate": 1.395616743106015e-05,
      "loss": 1.4476,
      "step": 20870
    },
    {
      "epoch": 0.9280156507047264,
      "grad_norm": 0.0585983581840992,
      "learning_rate": 1.3921889743904238e-05,
      "loss": 1.4504,
      "step": 20872
    },
    {
      "epoch": 0.9281045751633987,
      "grad_norm": 0.058123402297496796,
      "learning_rate": 1.3887653608847361e-05,
      "loss": 1.4508,
      "step": 20874
    },
    {
      "epoch": 0.9281934996220711,
      "grad_norm": 0.057394903153181076,
      "learning_rate": 1.3853459028815963e-05,
      "loss": 1.4494,
      "step": 20876
    },
    {
      "epoch": 0.9282824240807435,
      "grad_norm": 0.0583951398730278,
      "learning_rate": 1.3819306006733312e-05,
      "loss": 1.4535,
      "step": 20878
    },
    {
      "epoch": 0.9283713485394157,
      "grad_norm": 0.05813184753060341,
      "learning_rate": 1.3785194545518964e-05,
      "loss": 1.4472,
      "step": 20880
    },
    {
      "epoch": 0.9284602729980881,
      "grad_norm": 0.057534657418727875,
      "learning_rate": 1.3751124648088754e-05,
      "loss": 1.4378,
      "step": 20882
    },
    {
      "epoch": 0.9285491974567605,
      "grad_norm": 0.05735687166452408,
      "learning_rate": 1.371709631735535e-05,
      "loss": 1.4454,
      "step": 20884
    },
    {
      "epoch": 0.9286381219154328,
      "grad_norm": 0.05778350308537483,
      "learning_rate": 1.368310955622748e-05,
      "loss": 1.4487,
      "step": 20886
    },
    {
      "epoch": 0.9287270463741052,
      "grad_norm": 0.05846724286675453,
      "learning_rate": 1.364916436761049e-05,
      "loss": 1.4464,
      "step": 20888
    },
    {
      "epoch": 0.9288159708327776,
      "grad_norm": 0.057493094354867935,
      "learning_rate": 1.3615260754406278e-05,
      "loss": 1.4436,
      "step": 20890
    },
    {
      "epoch": 0.9289048952914499,
      "grad_norm": 0.05848955735564232,
      "learning_rate": 1.3581398719512972e-05,
      "loss": 1.4501,
      "step": 20892
    },
    {
      "epoch": 0.9289938197501223,
      "grad_norm": 0.05717703327536583,
      "learning_rate": 1.3547578265825255e-05,
      "loss": 1.443,
      "step": 20894
    },
    {
      "epoch": 0.9290827442087947,
      "grad_norm": 0.057882264256477356,
      "learning_rate": 1.3513799396234372e-05,
      "loss": 1.4442,
      "step": 20896
    },
    {
      "epoch": 0.929171668667467,
      "grad_norm": 0.057226452976465225,
      "learning_rate": 1.3480062113627679e-05,
      "loss": 1.4469,
      "step": 20898
    },
    {
      "epoch": 0.9292605931261393,
      "grad_norm": 0.057946473360061646,
      "learning_rate": 1.3446366420889366e-05,
      "loss": 1.4427,
      "step": 20900
    },
    {
      "epoch": 0.9293495175848117,
      "grad_norm": 0.05745415389537811,
      "learning_rate": 1.3412712320899856e-05,
      "loss": 1.4486,
      "step": 20902
    },
    {
      "epoch": 0.929438442043484,
      "grad_norm": 0.05738044157624245,
      "learning_rate": 1.33790998165359e-05,
      "loss": 1.4482,
      "step": 20904
    },
    {
      "epoch": 0.9295273665021564,
      "grad_norm": 0.05781969055533409,
      "learning_rate": 1.3345528910671033e-05,
      "loss": 1.4496,
      "step": 20906
    },
    {
      "epoch": 0.9296162909608288,
      "grad_norm": 0.058522261679172516,
      "learning_rate": 1.3311999606175018e-05,
      "loss": 1.4469,
      "step": 20908
    },
    {
      "epoch": 0.9297052154195011,
      "grad_norm": 0.058261334896087646,
      "learning_rate": 1.3278511905913948e-05,
      "loss": 1.4434,
      "step": 20910
    },
    {
      "epoch": 0.9297941398781735,
      "grad_norm": 0.05749499797821045,
      "learning_rate": 1.3245065812750645e-05,
      "loss": 1.4438,
      "step": 20912
    },
    {
      "epoch": 0.9298830643368459,
      "grad_norm": 0.058415595442056656,
      "learning_rate": 1.32116613295441e-05,
      "loss": 1.4472,
      "step": 20914
    },
    {
      "epoch": 0.9299719887955182,
      "grad_norm": 0.0577775239944458,
      "learning_rate": 1.3178298459150084e-05,
      "loss": 1.4461,
      "step": 20916
    },
    {
      "epoch": 0.9300609132541906,
      "grad_norm": 0.058564167469739914,
      "learning_rate": 1.3144977204420427e-05,
      "loss": 1.4459,
      "step": 20918
    },
    {
      "epoch": 0.930149837712863,
      "grad_norm": 0.05728235840797424,
      "learning_rate": 1.3111697568203574e-05,
      "loss": 1.4465,
      "step": 20920
    },
    {
      "epoch": 0.9302387621715352,
      "grad_norm": 0.05898916721343994,
      "learning_rate": 1.307845955334447e-05,
      "loss": 1.4465,
      "step": 20922
    },
    {
      "epoch": 0.9303276866302076,
      "grad_norm": 0.05832632631063461,
      "learning_rate": 1.304526316268445e-05,
      "loss": 1.4451,
      "step": 20924
    },
    {
      "epoch": 0.93041661108888,
      "grad_norm": 0.057864874601364136,
      "learning_rate": 1.301210839906125e-05,
      "loss": 1.4472,
      "step": 20926
    },
    {
      "epoch": 0.9305055355475523,
      "grad_norm": 0.05715201422572136,
      "learning_rate": 1.2978995265309158e-05,
      "loss": 1.4447,
      "step": 20928
    },
    {
      "epoch": 0.9305944600062247,
      "grad_norm": 0.05769526958465576,
      "learning_rate": 1.2945923764258738e-05,
      "loss": 1.4445,
      "step": 20930
    },
    {
      "epoch": 0.9306833844648971,
      "grad_norm": 0.05778975784778595,
      "learning_rate": 1.2912893898737177e-05,
      "loss": 1.4497,
      "step": 20932
    },
    {
      "epoch": 0.9307723089235694,
      "grad_norm": 0.05751507729291916,
      "learning_rate": 1.2879905671567938e-05,
      "loss": 1.4536,
      "step": 20934
    },
    {
      "epoch": 0.9308612333822418,
      "grad_norm": 0.05899496749043465,
      "learning_rate": 1.2846959085570985e-05,
      "loss": 1.4458,
      "step": 20936
    },
    {
      "epoch": 0.9309501578409142,
      "grad_norm": 0.05783265456557274,
      "learning_rate": 1.2814054143562848e-05,
      "loss": 1.4489,
      "step": 20938
    },
    {
      "epoch": 0.9310390822995865,
      "grad_norm": 0.05845189839601517,
      "learning_rate": 1.2781190848356328e-05,
      "loss": 1.4459,
      "step": 20940
    },
    {
      "epoch": 0.9311280067582589,
      "grad_norm": 0.05726044625043869,
      "learning_rate": 1.2748369202760678e-05,
      "loss": 1.4408,
      "step": 20942
    },
    {
      "epoch": 0.9312169312169312,
      "grad_norm": 0.058978818356990814,
      "learning_rate": 1.2715589209581713e-05,
      "loss": 1.4472,
      "step": 20944
    },
    {
      "epoch": 0.9313058556756035,
      "grad_norm": 0.05756903067231178,
      "learning_rate": 1.2682850871621521e-05,
      "loss": 1.4468,
      "step": 20946
    },
    {
      "epoch": 0.9313947801342759,
      "grad_norm": 0.05813246965408325,
      "learning_rate": 1.2650154191678864e-05,
      "loss": 1.4485,
      "step": 20948
    },
    {
      "epoch": 0.9314837045929483,
      "grad_norm": 0.05751319229602814,
      "learning_rate": 1.2617499172548619e-05,
      "loss": 1.449,
      "step": 20950
    },
    {
      "epoch": 0.9315726290516206,
      "grad_norm": 0.05843638256192207,
      "learning_rate": 1.2584885817022385e-05,
      "loss": 1.4567,
      "step": 20952
    },
    {
      "epoch": 0.931661553510293,
      "grad_norm": 0.05761188268661499,
      "learning_rate": 1.2552314127888153e-05,
      "loss": 1.4446,
      "step": 20954
    },
    {
      "epoch": 0.9317504779689654,
      "grad_norm": 0.05716591700911522,
      "learning_rate": 1.2519784107930144e-05,
      "loss": 1.4493,
      "step": 20956
    },
    {
      "epoch": 0.9318394024276377,
      "grad_norm": 0.05824635922908783,
      "learning_rate": 1.248729575992924e-05,
      "loss": 1.4445,
      "step": 20958
    },
    {
      "epoch": 0.9319283268863101,
      "grad_norm": 0.05802621319890022,
      "learning_rate": 1.2454849086662723e-05,
      "loss": 1.4477,
      "step": 20960
    },
    {
      "epoch": 0.9320172513449825,
      "grad_norm": 0.05747433751821518,
      "learning_rate": 1.242244409090426e-05,
      "loss": 1.4448,
      "step": 20962
    },
    {
      "epoch": 0.9321061758036548,
      "grad_norm": 0.05807986482977867,
      "learning_rate": 1.2390080775423917e-05,
      "loss": 1.4437,
      "step": 20964
    },
    {
      "epoch": 0.9321951002623271,
      "grad_norm": 0.05921353027224541,
      "learning_rate": 1.2357759142988367e-05,
      "loss": 1.4476,
      "step": 20966
    },
    {
      "epoch": 0.9322840247209995,
      "grad_norm": 0.05769893899559975,
      "learning_rate": 1.2325479196360457e-05,
      "loss": 1.4546,
      "step": 20968
    },
    {
      "epoch": 0.9323729491796718,
      "grad_norm": 0.058186814188957214,
      "learning_rate": 1.229324093829981e-05,
      "loss": 1.447,
      "step": 20970
    },
    {
      "epoch": 0.9324618736383442,
      "grad_norm": 0.057806309312582016,
      "learning_rate": 1.2261044371562058e-05,
      "loss": 1.4412,
      "step": 20972
    },
    {
      "epoch": 0.9325507980970166,
      "grad_norm": 0.05800100043416023,
      "learning_rate": 1.2228889498899664e-05,
      "loss": 1.4428,
      "step": 20974
    },
    {
      "epoch": 0.9326397225556889,
      "grad_norm": 0.057893089950084686,
      "learning_rate": 1.2196776323061376e-05,
      "loss": 1.4504,
      "step": 20976
    },
    {
      "epoch": 0.9327286470143613,
      "grad_norm": 0.05721002444624901,
      "learning_rate": 1.2164704846792275e-05,
      "loss": 1.4393,
      "step": 20978
    },
    {
      "epoch": 0.9328175714730337,
      "grad_norm": 0.05754590034484863,
      "learning_rate": 1.2132675072834054e-05,
      "loss": 1.4513,
      "step": 20980
    },
    {
      "epoch": 0.932906495931706,
      "grad_norm": 0.05840574577450752,
      "learning_rate": 1.2100687003924748e-05,
      "loss": 1.4451,
      "step": 20982
    },
    {
      "epoch": 0.9329954203903784,
      "grad_norm": 0.058282025158405304,
      "learning_rate": 1.2068740642798782e-05,
      "loss": 1.4451,
      "step": 20984
    },
    {
      "epoch": 0.9330843448490508,
      "grad_norm": 0.05762806534767151,
      "learning_rate": 1.2036835992187078e-05,
      "loss": 1.4521,
      "step": 20986
    },
    {
      "epoch": 0.933173269307723,
      "grad_norm": 0.058149222284555435,
      "learning_rate": 1.2004973054817125e-05,
      "loss": 1.4418,
      "step": 20988
    },
    {
      "epoch": 0.9332621937663954,
      "grad_norm": 0.057563796639442444,
      "learning_rate": 1.1973151833412466e-05,
      "loss": 1.4417,
      "step": 20990
    },
    {
      "epoch": 0.9333511182250678,
      "grad_norm": 0.05817318707704544,
      "learning_rate": 1.194137233069359e-05,
      "loss": 1.4438,
      "step": 20992
    },
    {
      "epoch": 0.9334400426837401,
      "grad_norm": 0.05729508027434349,
      "learning_rate": 1.1909634549376935e-05,
      "loss": 1.4476,
      "step": 20994
    },
    {
      "epoch": 0.9335289671424125,
      "grad_norm": 0.05712838098406792,
      "learning_rate": 1.187793849217561e-05,
      "loss": 1.4403,
      "step": 20996
    },
    {
      "epoch": 0.9336178916010849,
      "grad_norm": 0.05793168023228645,
      "learning_rate": 1.1846284161799281e-05,
      "loss": 1.4466,
      "step": 20998
    },
    {
      "epoch": 0.9337068160597572,
      "grad_norm": 0.057333990931510925,
      "learning_rate": 1.1814671560953783e-05,
      "loss": 1.4399,
      "step": 21000
    },
    {
      "epoch": 0.9337068160597572,
      "eval_loss": 1.4323413372039795,
      "eval_runtime": 12.424,
      "eval_samples_per_second": 556.179,
      "eval_steps_per_second": 69.543,
      "step": 21000
    },
    {
      "epoch": 0.9337957405184296,
      "grad_norm": 0.05756537243723869,
      "learning_rate": 1.1783100692341508e-05,
      "loss": 1.4479,
      "step": 21002
    },
    {
      "epoch": 0.933884664977102,
      "grad_norm": 0.0578317753970623,
      "learning_rate": 1.17515715586613e-05,
      "loss": 1.4463,
      "step": 21004
    },
    {
      "epoch": 0.9339735894357744,
      "grad_norm": 0.056406132876873016,
      "learning_rate": 1.172008416260839e-05,
      "loss": 1.4472,
      "step": 21006
    },
    {
      "epoch": 0.9340625138944467,
      "grad_norm": 0.0582900270819664,
      "learning_rate": 1.1688638506874405e-05,
      "loss": 1.452,
      "step": 21008
    },
    {
      "epoch": 0.934151438353119,
      "grad_norm": 0.05787406116724014,
      "learning_rate": 1.1657234594147636e-05,
      "loss": 1.4507,
      "step": 21010
    },
    {
      "epoch": 0.9342403628117913,
      "grad_norm": 0.05701528117060661,
      "learning_rate": 1.1625872427112439e-05,
      "loss": 1.4506,
      "step": 21012
    },
    {
      "epoch": 0.9343292872704637,
      "grad_norm": 0.05795227363705635,
      "learning_rate": 1.1594552008449888e-05,
      "loss": 1.4454,
      "step": 21014
    },
    {
      "epoch": 0.9344182117291361,
      "grad_norm": 0.05739319324493408,
      "learning_rate": 1.1563273340837289e-05,
      "loss": 1.451,
      "step": 21016
    },
    {
      "epoch": 0.9345071361878085,
      "grad_norm": 0.05824971944093704,
      "learning_rate": 1.1532036426948667e-05,
      "loss": 1.4479,
      "step": 21018
    },
    {
      "epoch": 0.9345960606464808,
      "grad_norm": 0.05758555978536606,
      "learning_rate": 1.1500841269454166e-05,
      "loss": 1.4407,
      "step": 21020
    },
    {
      "epoch": 0.9346849851051532,
      "grad_norm": 0.05768615007400513,
      "learning_rate": 1.1469687871020428e-05,
      "loss": 1.4472,
      "step": 21022
    },
    {
      "epoch": 0.9347739095638256,
      "grad_norm": 0.057626668363809586,
      "learning_rate": 1.143857623431066e-05,
      "loss": 1.4431,
      "step": 21024
    },
    {
      "epoch": 0.9348628340224979,
      "grad_norm": 0.057193391025066376,
      "learning_rate": 1.1407506361984509e-05,
      "loss": 1.4451,
      "step": 21026
    },
    {
      "epoch": 0.9349517584811703,
      "grad_norm": 0.05753851309418678,
      "learning_rate": 1.1376478256697797e-05,
      "loss": 1.4413,
      "step": 21028
    },
    {
      "epoch": 0.9350406829398425,
      "grad_norm": 0.05880241096019745,
      "learning_rate": 1.1345491921102957e-05,
      "loss": 1.4487,
      "step": 21030
    },
    {
      "epoch": 0.9351296073985149,
      "grad_norm": 0.057462744414806366,
      "learning_rate": 1.1314547357849037e-05,
      "loss": 1.4432,
      "step": 21032
    },
    {
      "epoch": 0.9352185318571873,
      "grad_norm": 0.05817759782075882,
      "learning_rate": 1.1283644569581087e-05,
      "loss": 1.4497,
      "step": 21034
    },
    {
      "epoch": 0.9353074563158597,
      "grad_norm": 0.058256879448890686,
      "learning_rate": 1.1252783558940882e-05,
      "loss": 1.4495,
      "step": 21036
    },
    {
      "epoch": 0.935396380774532,
      "grad_norm": 0.05921512097120285,
      "learning_rate": 1.1221964328566647e-05,
      "loss": 1.444,
      "step": 21038
    },
    {
      "epoch": 0.9354853052332044,
      "grad_norm": 0.05717106908559799,
      "learning_rate": 1.119118688109283e-05,
      "loss": 1.4431,
      "step": 21040
    },
    {
      "epoch": 0.9355742296918768,
      "grad_norm": 0.058333754539489746,
      "learning_rate": 1.1160451219150548e-05,
      "loss": 1.4481,
      "step": 21042
    },
    {
      "epoch": 0.9356631541505491,
      "grad_norm": 0.05773552879691124,
      "learning_rate": 1.112975734536703e-05,
      "loss": 1.4508,
      "step": 21044
    },
    {
      "epoch": 0.9357520786092215,
      "grad_norm": 0.058473654091358185,
      "learning_rate": 1.1099105262366294e-05,
      "loss": 1.452,
      "step": 21046
    },
    {
      "epoch": 0.9358410030678939,
      "grad_norm": 0.057561490684747696,
      "learning_rate": 1.1068494972768572e-05,
      "loss": 1.4452,
      "step": 21048
    },
    {
      "epoch": 0.9359299275265662,
      "grad_norm": 0.05912725254893303,
      "learning_rate": 1.1037926479190497e-05,
      "loss": 1.4493,
      "step": 21050
    },
    {
      "epoch": 0.9360188519852385,
      "grad_norm": 0.05738065019249916,
      "learning_rate": 1.1007399784245308e-05,
      "loss": 1.4438,
      "step": 21052
    },
    {
      "epoch": 0.9361077764439109,
      "grad_norm": 0.05802926421165466,
      "learning_rate": 1.0976914890542422e-05,
      "loss": 1.4481,
      "step": 21054
    },
    {
      "epoch": 0.9361967009025832,
      "grad_norm": 0.05792355164885521,
      "learning_rate": 1.0946471800688029e-05,
      "loss": 1.4466,
      "step": 21056
    },
    {
      "epoch": 0.9362856253612556,
      "grad_norm": 0.059775542467832565,
      "learning_rate": 1.0916070517284327e-05,
      "loss": 1.4437,
      "step": 21058
    },
    {
      "epoch": 0.936374549819928,
      "grad_norm": 0.058133359998464584,
      "learning_rate": 1.0885711042930235e-05,
      "loss": 1.4453,
      "step": 21060
    },
    {
      "epoch": 0.9364634742786003,
      "grad_norm": 0.05763556808233261,
      "learning_rate": 1.0855393380221068e-05,
      "loss": 1.4554,
      "step": 21062
    },
    {
      "epoch": 0.9365523987372727,
      "grad_norm": 0.058549582958221436,
      "learning_rate": 1.0825117531748474e-05,
      "loss": 1.447,
      "step": 21064
    },
    {
      "epoch": 0.9366413231959451,
      "grad_norm": 0.057294197380542755,
      "learning_rate": 1.0794883500100494e-05,
      "loss": 1.4436,
      "step": 21066
    },
    {
      "epoch": 0.9367302476546174,
      "grad_norm": 0.05782110616564751,
      "learning_rate": 1.0764691287861783e-05,
      "loss": 1.4539,
      "step": 21068
    },
    {
      "epoch": 0.9368191721132898,
      "grad_norm": 0.05737898498773575,
      "learning_rate": 1.0734540897613221e-05,
      "loss": 1.4546,
      "step": 21070
    },
    {
      "epoch": 0.9369080965719622,
      "grad_norm": 0.05745162069797516,
      "learning_rate": 1.0704432331932245e-05,
      "loss": 1.4459,
      "step": 21072
    },
    {
      "epoch": 0.9369970210306344,
      "grad_norm": 0.05851050093770027,
      "learning_rate": 1.0674365593392633e-05,
      "loss": 1.4427,
      "step": 21074
    },
    {
      "epoch": 0.9370859454893068,
      "grad_norm": 0.058115605264902115,
      "learning_rate": 1.0644340684564657e-05,
      "loss": 1.442,
      "step": 21076
    },
    {
      "epoch": 0.9371748699479792,
      "grad_norm": 0.05693921446800232,
      "learning_rate": 1.061435760801499e-05,
      "loss": 1.4476,
      "step": 21078
    },
    {
      "epoch": 0.9372637944066515,
      "grad_norm": 0.05739055946469307,
      "learning_rate": 1.0584416366306581e-05,
      "loss": 1.4492,
      "step": 21080
    },
    {
      "epoch": 0.9373527188653239,
      "grad_norm": 0.05716922506690025,
      "learning_rate": 1.0554516961999105e-05,
      "loss": 1.4438,
      "step": 21082
    },
    {
      "epoch": 0.9374416433239963,
      "grad_norm": 0.05674619972705841,
      "learning_rate": 1.0524659397648462e-05,
      "loss": 1.4482,
      "step": 21084
    },
    {
      "epoch": 0.9375305677826686,
      "grad_norm": 0.058578867465257645,
      "learning_rate": 1.0494843675806942e-05,
      "loss": 1.4419,
      "step": 21086
    },
    {
      "epoch": 0.937619492241341,
      "grad_norm": 0.05818881839513779,
      "learning_rate": 1.0465069799023397e-05,
      "loss": 1.4496,
      "step": 21088
    },
    {
      "epoch": 0.9377084167000134,
      "grad_norm": 0.058019597083330154,
      "learning_rate": 1.0435337769843012e-05,
      "loss": 1.4423,
      "step": 21090
    },
    {
      "epoch": 0.9377973411586857,
      "grad_norm": 0.05781329795718193,
      "learning_rate": 1.0405647590807366e-05,
      "loss": 1.4467,
      "step": 21092
    },
    {
      "epoch": 0.9378862656173581,
      "grad_norm": 0.05810556560754776,
      "learning_rate": 1.0375999264454483e-05,
      "loss": 1.4453,
      "step": 21094
    },
    {
      "epoch": 0.9379751900760304,
      "grad_norm": 0.05800723284482956,
      "learning_rate": 1.0346392793319005e-05,
      "loss": 1.442,
      "step": 21096
    },
    {
      "epoch": 0.9380641145347027,
      "grad_norm": 0.0583004355430603,
      "learning_rate": 1.031682817993157e-05,
      "loss": 1.447,
      "step": 21098
    },
    {
      "epoch": 0.9381530389933751,
      "grad_norm": 0.05934945121407509,
      "learning_rate": 1.0287305426819771e-05,
      "loss": 1.4457,
      "step": 21100
    },
    {
      "epoch": 0.9382419634520475,
      "grad_norm": 0.05703364312648773,
      "learning_rate": 1.0257824536507088e-05,
      "loss": 1.4435,
      "step": 21102
    },
    {
      "epoch": 0.9383308879107198,
      "grad_norm": 0.057916540652513504,
      "learning_rate": 1.0228385511513782e-05,
      "loss": 1.4447,
      "step": 21104
    },
    {
      "epoch": 0.9384198123693922,
      "grad_norm": 0.05742233991622925,
      "learning_rate": 1.0198988354356509e-05,
      "loss": 1.442,
      "step": 21106
    },
    {
      "epoch": 0.9385087368280646,
      "grad_norm": 0.05896897241473198,
      "learning_rate": 1.0169633067548145e-05,
      "loss": 1.4476,
      "step": 21108
    },
    {
      "epoch": 0.9385976612867369,
      "grad_norm": 0.05869178846478462,
      "learning_rate": 1.0140319653598185e-05,
      "loss": 1.4472,
      "step": 21110
    },
    {
      "epoch": 0.9386865857454093,
      "grad_norm": 0.05712299793958664,
      "learning_rate": 1.0111048115012456e-05,
      "loss": 1.4457,
      "step": 21112
    },
    {
      "epoch": 0.9387755102040817,
      "grad_norm": 0.05764850974082947,
      "learning_rate": 1.0081818454293124e-05,
      "loss": 1.4435,
      "step": 21114
    },
    {
      "epoch": 0.938864434662754,
      "grad_norm": 0.05720280855894089,
      "learning_rate": 1.0052630673939023e-05,
      "loss": 1.4491,
      "step": 21116
    },
    {
      "epoch": 0.9389533591214263,
      "grad_norm": 0.058444440364837646,
      "learning_rate": 1.0023484776445158e-05,
      "loss": 1.4496,
      "step": 21118
    },
    {
      "epoch": 0.9390422835800987,
      "grad_norm": 0.0576569139957428,
      "learning_rate": 9.994380764303091e-06,
      "loss": 1.4523,
      "step": 21120
    },
    {
      "epoch": 0.939131208038771,
      "grad_norm": 0.05701880529522896,
      "learning_rate": 9.96531864000072e-06,
      "loss": 1.4496,
      "step": 21122
    },
    {
      "epoch": 0.9392201324974434,
      "grad_norm": 0.05704936385154724,
      "learning_rate": 9.936298406022393e-06,
      "loss": 1.4445,
      "step": 21124
    },
    {
      "epoch": 0.9393090569561158,
      "grad_norm": 0.05782592296600342,
      "learning_rate": 9.907320064848956e-06,
      "loss": 1.4462,
      "step": 21126
    },
    {
      "epoch": 0.9393979814147881,
      "grad_norm": 0.05813451111316681,
      "learning_rate": 9.878383618957599e-06,
      "loss": 1.4475,
      "step": 21128
    },
    {
      "epoch": 0.9394869058734605,
      "grad_norm": 0.05756333842873573,
      "learning_rate": 9.849489070821893e-06,
      "loss": 1.4443,
      "step": 21130
    },
    {
      "epoch": 0.9395758303321329,
      "grad_norm": 0.05767510086297989,
      "learning_rate": 9.820636422911811e-06,
      "loss": 1.4527,
      "step": 21132
    },
    {
      "epoch": 0.9396647547908052,
      "grad_norm": 0.057892054319381714,
      "learning_rate": 9.791825677693988e-06,
      "loss": 1.4458,
      "step": 21134
    },
    {
      "epoch": 0.9397536792494776,
      "grad_norm": 0.057954948395490646,
      "learning_rate": 9.763056837631123e-06,
      "loss": 1.4412,
      "step": 21136
    },
    {
      "epoch": 0.93984260370815,
      "grad_norm": 0.05756092816591263,
      "learning_rate": 9.734329905182527e-06,
      "loss": 1.4492,
      "step": 21138
    },
    {
      "epoch": 0.9399315281668222,
      "grad_norm": 0.05734538286924362,
      "learning_rate": 9.705644882803954e-06,
      "loss": 1.4451,
      "step": 21140
    },
    {
      "epoch": 0.9400204526254946,
      "grad_norm": 0.057080693542957306,
      "learning_rate": 9.677001772947614e-06,
      "loss": 1.4529,
      "step": 21142
    },
    {
      "epoch": 0.940109377084167,
      "grad_norm": 0.057391393929719925,
      "learning_rate": 9.648400578061822e-06,
      "loss": 1.4459,
      "step": 21144
    },
    {
      "epoch": 0.9401983015428393,
      "grad_norm": 0.0573163703083992,
      "learning_rate": 9.619841300591736e-06,
      "loss": 1.4446,
      "step": 21146
    },
    {
      "epoch": 0.9402872260015117,
      "grad_norm": 0.058304496109485626,
      "learning_rate": 9.591323942978624e-06,
      "loss": 1.4483,
      "step": 21148
    },
    {
      "epoch": 0.9403761504601841,
      "grad_norm": 0.059254664927721024,
      "learning_rate": 9.562848507660316e-06,
      "loss": 1.4455,
      "step": 21150
    },
    {
      "epoch": 0.9404650749188564,
      "grad_norm": 0.05732493847608566,
      "learning_rate": 9.534414997070973e-06,
      "loss": 1.4498,
      "step": 21152
    },
    {
      "epoch": 0.9405539993775288,
      "grad_norm": 0.05808182805776596,
      "learning_rate": 9.506023413641263e-06,
      "loss": 1.4464,
      "step": 21154
    },
    {
      "epoch": 0.9406429238362012,
      "grad_norm": 0.05795228108763695,
      "learning_rate": 9.477673759798188e-06,
      "loss": 1.444,
      "step": 21156
    },
    {
      "epoch": 0.9407318482948736,
      "grad_norm": 0.0579368956387043,
      "learning_rate": 9.449366037965313e-06,
      "loss": 1.4475,
      "step": 21158
    },
    {
      "epoch": 0.9408207727535458,
      "grad_norm": 0.05821996554732323,
      "learning_rate": 9.421100250562309e-06,
      "loss": 1.4519,
      "step": 21160
    },
    {
      "epoch": 0.9409096972122182,
      "grad_norm": 0.058779843151569366,
      "learning_rate": 9.392876400005579e-06,
      "loss": 1.4436,
      "step": 21162
    },
    {
      "epoch": 0.9409986216708905,
      "grad_norm": 0.05720444768667221,
      "learning_rate": 9.364694488707858e-06,
      "loss": 1.4545,
      "step": 21164
    },
    {
      "epoch": 0.9410875461295629,
      "grad_norm": 0.057297226041555405,
      "learning_rate": 9.336554519078221e-06,
      "loss": 1.4425,
      "step": 21166
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.057402316480875015,
      "learning_rate": 9.308456493522133e-06,
      "loss": 1.4458,
      "step": 21168
    },
    {
      "epoch": 0.9412653950469076,
      "grad_norm": 0.05686284974217415,
      "learning_rate": 9.280400414441613e-06,
      "loss": 1.4437,
      "step": 21170
    },
    {
      "epoch": 0.94135431950558,
      "grad_norm": 0.05828374624252319,
      "learning_rate": 9.252386284235026e-06,
      "loss": 1.4463,
      "step": 21172
    },
    {
      "epoch": 0.9414432439642524,
      "grad_norm": 0.05804366618394852,
      "learning_rate": 9.224414105297064e-06,
      "loss": 1.4455,
      "step": 21174
    },
    {
      "epoch": 0.9415321684229248,
      "grad_norm": 0.058524325489997864,
      "learning_rate": 9.19648388001898e-06,
      "loss": 1.4406,
      "step": 21176
    },
    {
      "epoch": 0.9416210928815971,
      "grad_norm": 0.05712493881583214,
      "learning_rate": 9.168595610788365e-06,
      "loss": 1.445,
      "step": 21178
    },
    {
      "epoch": 0.9417100173402695,
      "grad_norm": 0.05792125314474106,
      "learning_rate": 9.140749299989205e-06,
      "loss": 1.4381,
      "step": 21180
    },
    {
      "epoch": 0.9417989417989417,
      "grad_norm": 0.05810127779841423,
      "learning_rate": 9.112944950001978e-06,
      "loss": 1.4472,
      "step": 21182
    },
    {
      "epoch": 0.9418878662576141,
      "grad_norm": 0.05826450139284134,
      "learning_rate": 9.085182563203453e-06,
      "loss": 1.4468,
      "step": 21184
    },
    {
      "epoch": 0.9419767907162865,
      "grad_norm": 0.0577828474342823,
      "learning_rate": 9.057462141966898e-06,
      "loss": 1.4437,
      "step": 21186
    },
    {
      "epoch": 0.9420657151749589,
      "grad_norm": 0.05752667412161827,
      "learning_rate": 9.02978368866203e-06,
      "loss": 1.4479,
      "step": 21188
    },
    {
      "epoch": 0.9421546396336312,
      "grad_norm": 0.05732182040810585,
      "learning_rate": 9.002147205654843e-06,
      "loss": 1.4476,
      "step": 21190
    },
    {
      "epoch": 0.9422435640923036,
      "grad_norm": 0.057963550090789795,
      "learning_rate": 8.974552695307948e-06,
      "loss": 1.4481,
      "step": 21192
    },
    {
      "epoch": 0.942332488550976,
      "grad_norm": 0.05983150750398636,
      "learning_rate": 8.947000159980067e-06,
      "loss": 1.4485,
      "step": 21194
    },
    {
      "epoch": 0.9424214130096483,
      "grad_norm": 0.05860733240842819,
      "learning_rate": 8.919489602026653e-06,
      "loss": 1.4393,
      "step": 21196
    },
    {
      "epoch": 0.9425103374683207,
      "grad_norm": 0.05784778296947479,
      "learning_rate": 8.892021023799435e-06,
      "loss": 1.4464,
      "step": 21198
    },
    {
      "epoch": 0.9425992619269931,
      "grad_norm": 0.05762090906500816,
      "learning_rate": 8.864594427646478e-06,
      "loss": 1.4431,
      "step": 21200
    },
    {
      "epoch": 0.9426881863856654,
      "grad_norm": 0.0579993836581707,
      "learning_rate": 8.837209815912295e-06,
      "loss": 1.4431,
      "step": 21202
    },
    {
      "epoch": 0.9427771108443377,
      "grad_norm": 0.0585738830268383,
      "learning_rate": 8.809867190938014e-06,
      "loss": 1.4496,
      "step": 21204
    },
    {
      "epoch": 0.94286603530301,
      "grad_norm": 0.05800570920109749,
      "learning_rate": 8.782566555060822e-06,
      "loss": 1.4428,
      "step": 21206
    },
    {
      "epoch": 0.9429549597616824,
      "grad_norm": 0.058509331196546555,
      "learning_rate": 8.755307910614574e-06,
      "loss": 1.4455,
      "step": 21208
    },
    {
      "epoch": 0.9430438842203548,
      "grad_norm": 0.05728688836097717,
      "learning_rate": 8.728091259929404e-06,
      "loss": 1.4459,
      "step": 21210
    },
    {
      "epoch": 0.9431328086790272,
      "grad_norm": 0.057143598794937134,
      "learning_rate": 8.70091660533201e-06,
      "loss": 1.4493,
      "step": 21212
    },
    {
      "epoch": 0.9432217331376995,
      "grad_norm": 0.05791422724723816,
      "learning_rate": 8.673783949145364e-06,
      "loss": 1.4482,
      "step": 21214
    },
    {
      "epoch": 0.9433106575963719,
      "grad_norm": 0.05727590247988701,
      "learning_rate": 8.646693293688834e-06,
      "loss": 1.4514,
      "step": 21216
    },
    {
      "epoch": 0.9433995820550443,
      "grad_norm": 0.05757603794336319,
      "learning_rate": 8.61964464127829e-06,
      "loss": 1.4526,
      "step": 21218
    },
    {
      "epoch": 0.9434885065137166,
      "grad_norm": 0.057514291256666183,
      "learning_rate": 8.592637994225994e-06,
      "loss": 1.4421,
      "step": 21220
    },
    {
      "epoch": 0.943577430972389,
      "grad_norm": 0.05774332210421562,
      "learning_rate": 8.565673354840543e-06,
      "loss": 1.4437,
      "step": 21222
    },
    {
      "epoch": 0.9436663554310614,
      "grad_norm": 0.05754677578806877,
      "learning_rate": 8.53875072542698e-06,
      "loss": 1.4513,
      "step": 21224
    },
    {
      "epoch": 0.9437552798897336,
      "grad_norm": 0.05838412791490555,
      "learning_rate": 8.511870108286857e-06,
      "loss": 1.449,
      "step": 21226
    },
    {
      "epoch": 0.943844204348406,
      "grad_norm": 0.05649451166391373,
      "learning_rate": 8.485031505718e-06,
      "loss": 1.4425,
      "step": 21228
    },
    {
      "epoch": 0.9439331288070784,
      "grad_norm": 0.05882545933127403,
      "learning_rate": 8.458234920014684e-06,
      "loss": 1.4436,
      "step": 21230
    },
    {
      "epoch": 0.9440220532657507,
      "grad_norm": 0.05716662108898163,
      "learning_rate": 8.431480353467524e-06,
      "loss": 1.4402,
      "step": 21232
    },
    {
      "epoch": 0.9441109777244231,
      "grad_norm": 0.05844126269221306,
      "learning_rate": 8.404767808363744e-06,
      "loss": 1.4457,
      "step": 21234
    },
    {
      "epoch": 0.9441999021830955,
      "grad_norm": 0.05845796316862106,
      "learning_rate": 8.378097286986852e-06,
      "loss": 1.4508,
      "step": 21236
    },
    {
      "epoch": 0.9442888266417678,
      "grad_norm": 0.059185273945331573,
      "learning_rate": 8.351468791616634e-06,
      "loss": 1.4452,
      "step": 21238
    },
    {
      "epoch": 0.9443777511004402,
      "grad_norm": 0.05794282630085945,
      "learning_rate": 8.324882324529548e-06,
      "loss": 1.4503,
      "step": 21240
    },
    {
      "epoch": 0.9444666755591126,
      "grad_norm": 0.05867988243699074,
      "learning_rate": 8.29833788799822e-06,
      "loss": 1.4503,
      "step": 21242
    },
    {
      "epoch": 0.9445556000177849,
      "grad_norm": 0.05754483863711357,
      "learning_rate": 8.271835484291835e-06,
      "loss": 1.4445,
      "step": 21244
    },
    {
      "epoch": 0.9446445244764573,
      "grad_norm": 0.05763980373740196,
      "learning_rate": 8.245375115675912e-06,
      "loss": 1.4464,
      "step": 21246
    },
    {
      "epoch": 0.9447334489351296,
      "grad_norm": 0.05797286331653595,
      "learning_rate": 8.218956784412479e-06,
      "loss": 1.448,
      "step": 21248
    },
    {
      "epoch": 0.9448223733938019,
      "grad_norm": 0.05742442235350609,
      "learning_rate": 8.192580492759783e-06,
      "loss": 1.4453,
      "step": 21250
    },
    {
      "epoch": 0.9449112978524743,
      "grad_norm": 0.05741652473807335,
      "learning_rate": 8.16624624297263e-06,
      "loss": 1.4499,
      "step": 21252
    },
    {
      "epoch": 0.9450002223111467,
      "grad_norm": 0.058388851583004,
      "learning_rate": 8.139954037302222e-06,
      "loss": 1.4424,
      "step": 21254
    },
    {
      "epoch": 0.945089146769819,
      "grad_norm": 0.058313120156526566,
      "learning_rate": 8.113703877996092e-06,
      "loss": 1.4426,
      "step": 21256
    },
    {
      "epoch": 0.9451780712284914,
      "grad_norm": 0.05811810865998268,
      "learning_rate": 8.087495767298226e-06,
      "loss": 1.441,
      "step": 21258
    },
    {
      "epoch": 0.9452669956871638,
      "grad_norm": 0.057785093784332275,
      "learning_rate": 8.061329707448939e-06,
      "loss": 1.4496,
      "step": 21260
    },
    {
      "epoch": 0.9453559201458361,
      "grad_norm": 0.05778678134083748,
      "learning_rate": 8.035205700685167e-06,
      "loss": 1.4433,
      "step": 21262
    },
    {
      "epoch": 0.9454448446045085,
      "grad_norm": 0.057463571429252625,
      "learning_rate": 8.00912374924001e-06,
      "loss": 1.4445,
      "step": 21264
    },
    {
      "epoch": 0.9455337690631809,
      "grad_norm": 0.058068107813596725,
      "learning_rate": 7.983083855343132e-06,
      "loss": 1.4452,
      "step": 21266
    },
    {
      "epoch": 0.9456226935218531,
      "grad_norm": 0.05680304020643234,
      "learning_rate": 7.957086021220417e-06,
      "loss": 1.4465,
      "step": 21268
    },
    {
      "epoch": 0.9457116179805255,
      "grad_norm": 0.057669758796691895,
      "learning_rate": 7.931130249094365e-06,
      "loss": 1.4493,
      "step": 21270
    },
    {
      "epoch": 0.9458005424391979,
      "grad_norm": 0.056808967143297195,
      "learning_rate": 7.905216541183869e-06,
      "loss": 1.4448,
      "step": 21272
    },
    {
      "epoch": 0.9458894668978702,
      "grad_norm": 0.05728383734822273,
      "learning_rate": 7.879344899703932e-06,
      "loss": 1.4457,
      "step": 21274
    },
    {
      "epoch": 0.9459783913565426,
      "grad_norm": 0.05765995383262634,
      "learning_rate": 7.853515326866345e-06,
      "loss": 1.4424,
      "step": 21276
    },
    {
      "epoch": 0.946067315815215,
      "grad_norm": 0.0581035315990448,
      "learning_rate": 7.827727824879116e-06,
      "loss": 1.443,
      "step": 21278
    },
    {
      "epoch": 0.9461562402738873,
      "grad_norm": 0.05767706036567688,
      "learning_rate": 7.801982395946649e-06,
      "loss": 1.4423,
      "step": 21280
    },
    {
      "epoch": 0.9462451647325597,
      "grad_norm": 0.057066384702920914,
      "learning_rate": 7.776279042269685e-06,
      "loss": 1.4482,
      "step": 21282
    },
    {
      "epoch": 0.9463340891912321,
      "grad_norm": 0.058075472712516785,
      "learning_rate": 7.750617766045688e-06,
      "loss": 1.4452,
      "step": 21284
    },
    {
      "epoch": 0.9464230136499044,
      "grad_norm": 0.05739322304725647,
      "learning_rate": 7.724998569468066e-06,
      "loss": 1.4477,
      "step": 21286
    },
    {
      "epoch": 0.9465119381085768,
      "grad_norm": 0.05727776512503624,
      "learning_rate": 7.699421454726963e-06,
      "loss": 1.447,
      "step": 21288
    },
    {
      "epoch": 0.9466008625672491,
      "grad_norm": 0.057564686983823776,
      "learning_rate": 7.67388642400879e-06,
      "loss": 1.4455,
      "step": 21290
    },
    {
      "epoch": 0.9466897870259214,
      "grad_norm": 0.05791555717587471,
      "learning_rate": 7.648393479496419e-06,
      "loss": 1.4429,
      "step": 21292
    },
    {
      "epoch": 0.9467787114845938,
      "grad_norm": 0.057518370449543,
      "learning_rate": 7.622942623369156e-06,
      "loss": 1.4435,
      "step": 21294
    },
    {
      "epoch": 0.9468676359432662,
      "grad_norm": 0.05689391866326332,
      "learning_rate": 7.597533857802541e-06,
      "loss": 1.4461,
      "step": 21296
    },
    {
      "epoch": 0.9469565604019385,
      "grad_norm": 0.058041494339704514,
      "learning_rate": 7.57216718496867e-06,
      "loss": 1.4456,
      "step": 21298
    },
    {
      "epoch": 0.9470454848606109,
      "grad_norm": 0.057841625064611435,
      "learning_rate": 7.5468426070360865e-06,
      "loss": 1.4501,
      "step": 21300
    },
    {
      "epoch": 0.9471344093192833,
      "grad_norm": 0.057104241102933884,
      "learning_rate": 7.521560126169502e-06,
      "loss": 1.4433,
      "step": 21302
    },
    {
      "epoch": 0.9472233337779556,
      "grad_norm": 0.058944664895534515,
      "learning_rate": 7.4963197445302445e-06,
      "loss": 1.4469,
      "step": 21304
    },
    {
      "epoch": 0.947312258236628,
      "grad_norm": 0.05764506757259369,
      "learning_rate": 7.471121464275976e-06,
      "loss": 1.4485,
      "step": 21306
    },
    {
      "epoch": 0.9474011826953004,
      "grad_norm": 0.05761958658695221,
      "learning_rate": 7.445965287560752e-06,
      "loss": 1.4491,
      "step": 21308
    },
    {
      "epoch": 0.9474901071539727,
      "grad_norm": 0.0575152151286602,
      "learning_rate": 7.420851216535019e-06,
      "loss": 1.4428,
      "step": 21310
    },
    {
      "epoch": 0.947579031612645,
      "grad_norm": 0.058228038251399994,
      "learning_rate": 7.3957792533456695e-06,
      "loss": 1.4463,
      "step": 21312
    },
    {
      "epoch": 0.9476679560713174,
      "grad_norm": 0.057448986917734146,
      "learning_rate": 7.3707494001358785e-06,
      "loss": 1.4482,
      "step": 21314
    },
    {
      "epoch": 0.9477568805299897,
      "grad_norm": 0.05910767614841461,
      "learning_rate": 7.3457616590454916e-06,
      "loss": 1.445,
      "step": 21316
    },
    {
      "epoch": 0.9478458049886621,
      "grad_norm": 0.057866428047418594,
      "learning_rate": 7.320816032210353e-06,
      "loss": 1.4464,
      "step": 21318
    },
    {
      "epoch": 0.9479347294473345,
      "grad_norm": 0.05850965157151222,
      "learning_rate": 7.295912521763038e-06,
      "loss": 1.4437,
      "step": 21320
    },
    {
      "epoch": 0.9480236539060068,
      "grad_norm": 0.057498205453157425,
      "learning_rate": 7.271051129832451e-06,
      "loss": 1.449,
      "step": 21322
    },
    {
      "epoch": 0.9481125783646792,
      "grad_norm": 0.057792823761701584,
      "learning_rate": 7.246231858543784e-06,
      "loss": 1.4393,
      "step": 21324
    },
    {
      "epoch": 0.9482015028233516,
      "grad_norm": 0.058098308742046356,
      "learning_rate": 7.221454710018671e-06,
      "loss": 1.4419,
      "step": 21326
    },
    {
      "epoch": 0.948290427282024,
      "grad_norm": 0.058741819113492966,
      "learning_rate": 7.196719686375308e-06,
      "loss": 1.4463,
      "step": 21328
    },
    {
      "epoch": 0.9483793517406963,
      "grad_norm": 0.0579305998980999,
      "learning_rate": 7.1720267897279475e-06,
      "loss": 1.4438,
      "step": 21330
    },
    {
      "epoch": 0.9484682761993687,
      "grad_norm": 0.05758350342512131,
      "learning_rate": 7.147376022187624e-06,
      "loss": 1.4539,
      "step": 21332
    },
    {
      "epoch": 0.9485572006580409,
      "grad_norm": 0.05874762684106827,
      "learning_rate": 7.122767385861484e-06,
      "loss": 1.4462,
      "step": 21334
    },
    {
      "epoch": 0.9486461251167133,
      "grad_norm": 0.05720106512308121,
      "learning_rate": 7.09820088285329e-06,
      "loss": 1.4421,
      "step": 21336
    },
    {
      "epoch": 0.9487350495753857,
      "grad_norm": 0.057676248252391815,
      "learning_rate": 7.073676515263028e-06,
      "loss": 1.4495,
      "step": 21338
    },
    {
      "epoch": 0.948823974034058,
      "grad_norm": 0.05915232002735138,
      "learning_rate": 7.049194285187077e-06,
      "loss": 1.4458,
      "step": 21340
    },
    {
      "epoch": 0.9489128984927304,
      "grad_norm": 0.05813376605510712,
      "learning_rate": 7.02475419471843e-06,
      "loss": 1.4398,
      "step": 21342
    },
    {
      "epoch": 0.9490018229514028,
      "grad_norm": 0.057365693151950836,
      "learning_rate": 7.000356245946249e-06,
      "loss": 1.4464,
      "step": 21344
    },
    {
      "epoch": 0.9490907474100752,
      "grad_norm": 0.05754502862691879,
      "learning_rate": 6.976000440956198e-06,
      "loss": 1.4493,
      "step": 21346
    },
    {
      "epoch": 0.9491796718687475,
      "grad_norm": 0.058054231107234955,
      "learning_rate": 6.9516867818302796e-06,
      "loss": 1.4469,
      "step": 21348
    },
    {
      "epoch": 0.9492685963274199,
      "grad_norm": 0.05699177458882332,
      "learning_rate": 6.927415270647053e-06,
      "loss": 1.4509,
      "step": 21350
    },
    {
      "epoch": 0.9493575207860923,
      "grad_norm": 0.058153219521045685,
      "learning_rate": 6.903185909481191e-06,
      "loss": 1.4432,
      "step": 21352
    },
    {
      "epoch": 0.9494464452447646,
      "grad_norm": 0.05815225467085838,
      "learning_rate": 6.878998700403982e-06,
      "loss": 1.439,
      "step": 21354
    },
    {
      "epoch": 0.9495353697034369,
      "grad_norm": 0.05740995332598686,
      "learning_rate": 6.854853645483106e-06,
      "loss": 1.4472,
      "step": 21356
    },
    {
      "epoch": 0.9496242941621093,
      "grad_norm": 0.05770250782370567,
      "learning_rate": 6.830750746782633e-06,
      "loss": 1.4437,
      "step": 21358
    },
    {
      "epoch": 0.9497132186207816,
      "grad_norm": 0.05776236578822136,
      "learning_rate": 6.806690006362859e-06,
      "loss": 1.4447,
      "step": 21360
    },
    {
      "epoch": 0.949802143079454,
      "grad_norm": 0.057613179087638855,
      "learning_rate": 6.7826714262806396e-06,
      "loss": 1.4448,
      "step": 21362
    },
    {
      "epoch": 0.9498910675381264,
      "grad_norm": 0.0582854337990284,
      "learning_rate": 6.758695008589221e-06,
      "loss": 1.4439,
      "step": 21364
    },
    {
      "epoch": 0.9499799919967987,
      "grad_norm": 0.057772018015384674,
      "learning_rate": 6.734760755338243e-06,
      "loss": 1.4434,
      "step": 21366
    },
    {
      "epoch": 0.9500689164554711,
      "grad_norm": 0.057995546609163284,
      "learning_rate": 6.7108686685735665e-06,
      "loss": 1.444,
      "step": 21368
    },
    {
      "epoch": 0.9501578409141435,
      "grad_norm": 0.05806703120470047,
      "learning_rate": 6.687018750337726e-06,
      "loss": 1.4412,
      "step": 21370
    },
    {
      "epoch": 0.9502467653728158,
      "grad_norm": 0.056987687945365906,
      "learning_rate": 6.663211002669534e-06,
      "loss": 1.4426,
      "step": 21372
    },
    {
      "epoch": 0.9503356898314882,
      "grad_norm": 0.05830015987157822,
      "learning_rate": 6.639445427604085e-06,
      "loss": 1.4377,
      "step": 21374
    },
    {
      "epoch": 0.9504246142901606,
      "grad_norm": 0.057094842195510864,
      "learning_rate": 6.615722027173032e-06,
      "loss": 1.4469,
      "step": 21376
    },
    {
      "epoch": 0.9505135387488328,
      "grad_norm": 0.057993218302726746,
      "learning_rate": 6.592040803404309e-06,
      "loss": 1.4415,
      "step": 21378
    },
    {
      "epoch": 0.9506024632075052,
      "grad_norm": 0.05756436288356781,
      "learning_rate": 6.5684017583223506e-06,
      "loss": 1.4454,
      "step": 21380
    },
    {
      "epoch": 0.9506913876661776,
      "grad_norm": 0.05718113109469414,
      "learning_rate": 6.544804893947876e-06,
      "loss": 1.4446,
      "step": 21382
    },
    {
      "epoch": 0.9507803121248499,
      "grad_norm": 0.05671248957514763,
      "learning_rate": 6.521250212298046e-06,
      "loss": 1.4501,
      "step": 21384
    },
    {
      "epoch": 0.9508692365835223,
      "grad_norm": 0.05659397318959236,
      "learning_rate": 6.497737715386476e-06,
      "loss": 1.4458,
      "step": 21386
    },
    {
      "epoch": 0.9509581610421947,
      "grad_norm": 0.05791900306940079,
      "learning_rate": 6.474267405223e-06,
      "loss": 1.4472,
      "step": 21388
    },
    {
      "epoch": 0.951047085500867,
      "grad_norm": 0.05804454907774925,
      "learning_rate": 6.450839283814125e-06,
      "loss": 1.4465,
      "step": 21390
    },
    {
      "epoch": 0.9511360099595394,
      "grad_norm": 0.05747028812766075,
      "learning_rate": 6.4274533531624715e-06,
      "loss": 1.4468,
      "step": 21392
    },
    {
      "epoch": 0.9512249344182118,
      "grad_norm": 0.058244530111551285,
      "learning_rate": 6.404109615267218e-06,
      "loss": 1.4506,
      "step": 21394
    },
    {
      "epoch": 0.9513138588768841,
      "grad_norm": 0.057229701429605484,
      "learning_rate": 6.380808072123934e-06,
      "loss": 1.4488,
      "step": 21396
    },
    {
      "epoch": 0.9514027833355564,
      "grad_norm": 0.05814434215426445,
      "learning_rate": 6.357548725724416e-06,
      "loss": 1.4449,
      "step": 21398
    },
    {
      "epoch": 0.9514917077942288,
      "grad_norm": 0.057511039078235626,
      "learning_rate": 6.334331578057018e-06,
      "loss": 1.4408,
      "step": 21400
    },
    {
      "epoch": 0.9515806322529011,
      "grad_norm": 0.05873566493391991,
      "learning_rate": 6.31115663110654e-06,
      "loss": 1.449,
      "step": 21402
    },
    {
      "epoch": 0.9516695567115735,
      "grad_norm": 0.05769232288002968,
      "learning_rate": 6.288023886854011e-06,
      "loss": 1.4451,
      "step": 21404
    },
    {
      "epoch": 0.9517584811702459,
      "grad_norm": 0.05827988311648369,
      "learning_rate": 6.264933347276847e-06,
      "loss": 1.4442,
      "step": 21406
    },
    {
      "epoch": 0.9518474056289182,
      "grad_norm": 0.05725441128015518,
      "learning_rate": 6.2418850143490805e-06,
      "loss": 1.4455,
      "step": 21408
    },
    {
      "epoch": 0.9519363300875906,
      "grad_norm": 0.05901114642620087,
      "learning_rate": 6.218878890040858e-06,
      "loss": 1.4456,
      "step": 21410
    },
    {
      "epoch": 0.952025254546263,
      "grad_norm": 0.05790838971734047,
      "learning_rate": 6.195914976318884e-06,
      "loss": 1.4424,
      "step": 21412
    },
    {
      "epoch": 0.9521141790049353,
      "grad_norm": 0.057518161833286285,
      "learning_rate": 6.17299327514631e-06,
      "loss": 1.45,
      "step": 21414
    },
    {
      "epoch": 0.9522031034636077,
      "grad_norm": 0.05830460786819458,
      "learning_rate": 6.150113788482403e-06,
      "loss": 1.451,
      "step": 21416
    },
    {
      "epoch": 0.9522920279222801,
      "grad_norm": 0.057020124047994614,
      "learning_rate": 6.127276518283153e-06,
      "loss": 1.4477,
      "step": 21418
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 0.05747748538851738,
      "learning_rate": 6.104481466500667e-06,
      "loss": 1.4502,
      "step": 21420
    },
    {
      "epoch": 0.9524698768396247,
      "grad_norm": 0.05771753937005997,
      "learning_rate": 6.081728635083661e-06,
      "loss": 1.4461,
      "step": 21422
    },
    {
      "epoch": 0.9525588012982971,
      "grad_norm": 0.057178083807229996,
      "learning_rate": 6.059018025977137e-06,
      "loss": 1.4498,
      "step": 21424
    },
    {
      "epoch": 0.9526477257569694,
      "grad_norm": 0.05818738043308258,
      "learning_rate": 6.036349641122429e-06,
      "loss": 1.4535,
      "step": 21426
    },
    {
      "epoch": 0.9527366502156418,
      "grad_norm": 0.057868827134370804,
      "learning_rate": 6.013723482457434e-06,
      "loss": 1.4438,
      "step": 21428
    },
    {
      "epoch": 0.9528255746743142,
      "grad_norm": 0.05839633569121361,
      "learning_rate": 5.9911395519162695e-06,
      "loss": 1.4444,
      "step": 21430
    },
    {
      "epoch": 0.9529144991329865,
      "grad_norm": 0.05750703439116478,
      "learning_rate": 5.968597851429502e-06,
      "loss": 1.4501,
      "step": 21432
    },
    {
      "epoch": 0.9530034235916589,
      "grad_norm": 0.05813831835985184,
      "learning_rate": 5.946098382924148e-06,
      "loss": 1.4435,
      "step": 21434
    },
    {
      "epoch": 0.9530923480503313,
      "grad_norm": 0.05906336382031441,
      "learning_rate": 5.9236411483235e-06,
      "loss": 1.4442,
      "step": 21436
    },
    {
      "epoch": 0.9531812725090036,
      "grad_norm": 0.05761095881462097,
      "learning_rate": 5.901226149547356e-06,
      "loss": 1.4428,
      "step": 21438
    },
    {
      "epoch": 0.953270196967676,
      "grad_norm": 0.058021754026412964,
      "learning_rate": 5.8788533885117956e-06,
      "loss": 1.4458,
      "step": 21440
    },
    {
      "epoch": 0.9533591214263483,
      "grad_norm": 0.057688046246767044,
      "learning_rate": 5.856522867129343e-06,
      "loss": 1.4438,
      "step": 21442
    },
    {
      "epoch": 0.9534480458850206,
      "grad_norm": 0.05762839689850807,
      "learning_rate": 5.834234587309028e-06,
      "loss": 1.4459,
      "step": 21444
    },
    {
      "epoch": 0.953536970343693,
      "grad_norm": 0.05795801803469658,
      "learning_rate": 5.811988550955993e-06,
      "loss": 1.4487,
      "step": 21446
    },
    {
      "epoch": 0.9536258948023654,
      "grad_norm": 0.0580856017768383,
      "learning_rate": 5.789784759971994e-06,
      "loss": 1.4503,
      "step": 21448
    },
    {
      "epoch": 0.9537148192610377,
      "grad_norm": 0.05720821022987366,
      "learning_rate": 5.767623216255125e-06,
      "loss": 1.4477,
      "step": 21450
    },
    {
      "epoch": 0.9538037437197101,
      "grad_norm": 0.057731300592422485,
      "learning_rate": 5.745503921699868e-06,
      "loss": 1.4478,
      "step": 21452
    },
    {
      "epoch": 0.9538926681783825,
      "grad_norm": 0.05763096734881401,
      "learning_rate": 5.7234268781969915e-06,
      "loss": 1.4447,
      "step": 21454
    },
    {
      "epoch": 0.9539815926370548,
      "grad_norm": 0.057338956743478775,
      "learning_rate": 5.701392087633761e-06,
      "loss": 1.445,
      "step": 21456
    },
    {
      "epoch": 0.9540705170957272,
      "grad_norm": 0.057891324162483215,
      "learning_rate": 5.679399551893893e-06,
      "loss": 1.4418,
      "step": 21458
    },
    {
      "epoch": 0.9541594415543996,
      "grad_norm": 0.05737922713160515,
      "learning_rate": 5.657449272857385e-06,
      "loss": 1.4444,
      "step": 21460
    },
    {
      "epoch": 0.954248366013072,
      "grad_norm": 0.05823858082294464,
      "learning_rate": 5.635541252400511e-06,
      "loss": 1.4367,
      "step": 21462
    },
    {
      "epoch": 0.9543372904717442,
      "grad_norm": 0.05750085040926933,
      "learning_rate": 5.61367549239622e-06,
      "loss": 1.4491,
      "step": 21464
    },
    {
      "epoch": 0.9544262149304166,
      "grad_norm": 0.05752246081829071,
      "learning_rate": 5.591851994713681e-06,
      "loss": 1.4429,
      "step": 21466
    },
    {
      "epoch": 0.9545151393890889,
      "grad_norm": 0.057769421488046646,
      "learning_rate": 5.5700707612184044e-06,
      "loss": 1.4498,
      "step": 21468
    },
    {
      "epoch": 0.9546040638477613,
      "grad_norm": 0.057797472923994064,
      "learning_rate": 5.548331793772288e-06,
      "loss": 1.4463,
      "step": 21470
    },
    {
      "epoch": 0.9546929883064337,
      "grad_norm": 0.057377446442842484,
      "learning_rate": 5.52663509423379e-06,
      "loss": 1.4473,
      "step": 21472
    },
    {
      "epoch": 0.954781912765106,
      "grad_norm": 0.05764816701412201,
      "learning_rate": 5.504980664457593e-06,
      "loss": 1.4454,
      "step": 21474
    },
    {
      "epoch": 0.9548708372237784,
      "grad_norm": 0.05828756093978882,
      "learning_rate": 5.483368506294828e-06,
      "loss": 1.4446,
      "step": 21476
    },
    {
      "epoch": 0.9549597616824508,
      "grad_norm": 0.058148372918367386,
      "learning_rate": 5.461798621593017e-06,
      "loss": 1.441,
      "step": 21478
    },
    {
      "epoch": 0.9550486861411231,
      "grad_norm": 0.05808459222316742,
      "learning_rate": 5.440271012195963e-06,
      "loss": 1.4473,
      "step": 21480
    },
    {
      "epoch": 0.9551376105997955,
      "grad_norm": 0.05690108239650726,
      "learning_rate": 5.4187856799440275e-06,
      "loss": 1.4451,
      "step": 21482
    },
    {
      "epoch": 0.9552265350584679,
      "grad_norm": 0.05813979357481003,
      "learning_rate": 5.3973426266737955e-06,
      "loss": 1.4465,
      "step": 21484
    },
    {
      "epoch": 0.9553154595171401,
      "grad_norm": 0.057522471994161606,
      "learning_rate": 5.3759418542184134e-06,
      "loss": 1.4487,
      "step": 21486
    },
    {
      "epoch": 0.9554043839758125,
      "grad_norm": 0.05860830470919609,
      "learning_rate": 5.354583364407251e-06,
      "loss": 1.4444,
      "step": 21488
    },
    {
      "epoch": 0.9554933084344849,
      "grad_norm": 0.05663066357374191,
      "learning_rate": 5.333267159066124e-06,
      "loss": 1.444,
      "step": 21490
    },
    {
      "epoch": 0.9555822328931572,
      "grad_norm": 0.058597080409526825,
      "learning_rate": 5.311993240017243e-06,
      "loss": 1.4472,
      "step": 21492
    },
    {
      "epoch": 0.9556711573518296,
      "grad_norm": 0.057328660041093826,
      "learning_rate": 5.290761609079209e-06,
      "loss": 1.447,
      "step": 21494
    },
    {
      "epoch": 0.955760081810502,
      "grad_norm": 0.0580228753387928,
      "learning_rate": 5.269572268066958e-06,
      "loss": 1.4468,
      "step": 21496
    },
    {
      "epoch": 0.9558490062691744,
      "grad_norm": 0.05755842849612236,
      "learning_rate": 5.248425218791874e-06,
      "loss": 1.4492,
      "step": 21498
    },
    {
      "epoch": 0.9559379307278467,
      "grad_norm": 0.05705071613192558,
      "learning_rate": 5.227320463061791e-06,
      "loss": 1.4469,
      "step": 21500
    },
    {
      "epoch": 0.9559379307278467,
      "eval_loss": 1.432187557220459,
      "eval_runtime": 12.4558,
      "eval_samples_per_second": 554.762,
      "eval_steps_per_second": 69.365,
      "step": 21500
    },
    {
      "epoch": 0.9560268551865191,
      "grad_norm": 0.057521793991327286,
      "learning_rate": 5.206258002680653e-06,
      "loss": 1.4425,
      "step": 21502
    },
    {
      "epoch": 0.9561157796451915,
      "grad_norm": 0.05715570226311684,
      "learning_rate": 5.18523783944913e-06,
      "loss": 1.4504,
      "step": 21504
    },
    {
      "epoch": 0.9562047041038638,
      "grad_norm": 0.05779632553458214,
      "learning_rate": 5.164259975164009e-06,
      "loss": 1.4429,
      "step": 21506
    },
    {
      "epoch": 0.9562936285625361,
      "grad_norm": 0.059060875326395035,
      "learning_rate": 5.143324411618577e-06,
      "loss": 1.4463,
      "step": 21508
    },
    {
      "epoch": 0.9563825530212084,
      "grad_norm": 0.05711085721850395,
      "learning_rate": 5.122431150602624e-06,
      "loss": 1.4455,
      "step": 21510
    },
    {
      "epoch": 0.9564714774798808,
      "grad_norm": 0.05797237157821655,
      "learning_rate": 5.101580193902055e-06,
      "loss": 1.4448,
      "step": 21512
    },
    {
      "epoch": 0.9565604019385532,
      "grad_norm": 0.057503245770931244,
      "learning_rate": 5.080771543299389e-06,
      "loss": 1.4502,
      "step": 21514
    },
    {
      "epoch": 0.9566493263972256,
      "grad_norm": 0.05919162929058075,
      "learning_rate": 5.060005200573425e-06,
      "loss": 1.4475,
      "step": 21516
    },
    {
      "epoch": 0.9567382508558979,
      "grad_norm": 0.05886862426996231,
      "learning_rate": 5.039281167499299e-06,
      "loss": 1.4526,
      "step": 21518
    },
    {
      "epoch": 0.9568271753145703,
      "grad_norm": 0.058441340923309326,
      "learning_rate": 5.018599445848648e-06,
      "loss": 1.4441,
      "step": 21520
    },
    {
      "epoch": 0.9569160997732427,
      "grad_norm": 0.05824016034603119,
      "learning_rate": 4.997960037389449e-06,
      "loss": 1.4449,
      "step": 21522
    },
    {
      "epoch": 0.957005024231915,
      "grad_norm": 0.057525213807821274,
      "learning_rate": 4.97736294388601e-06,
      "loss": 1.4477,
      "step": 21524
    },
    {
      "epoch": 0.9570939486905874,
      "grad_norm": 0.058176856487989426,
      "learning_rate": 4.956808167099092e-06,
      "loss": 1.4471,
      "step": 21526
    },
    {
      "epoch": 0.9571828731492596,
      "grad_norm": 0.057681165635585785,
      "learning_rate": 4.936295708785732e-06,
      "loss": 1.4497,
      "step": 21528
    },
    {
      "epoch": 0.957271797607932,
      "grad_norm": 0.05783999338746071,
      "learning_rate": 4.915825570699584e-06,
      "loss": 1.4408,
      "step": 21530
    },
    {
      "epoch": 0.9573607220666044,
      "grad_norm": 0.05777215585112572,
      "learning_rate": 4.895397754590414e-06,
      "loss": 1.4461,
      "step": 21532
    },
    {
      "epoch": 0.9574496465252768,
      "grad_norm": 0.058080073446035385,
      "learning_rate": 4.875012262204492e-06,
      "loss": 1.451,
      "step": 21534
    },
    {
      "epoch": 0.9575385709839491,
      "grad_norm": 0.05700400471687317,
      "learning_rate": 4.854669095284425e-06,
      "loss": 1.4511,
      "step": 21536
    },
    {
      "epoch": 0.9576274954426215,
      "grad_norm": 0.05668550357222557,
      "learning_rate": 4.834368255569322e-06,
      "loss": 1.4435,
      "step": 21538
    },
    {
      "epoch": 0.9577164199012939,
      "grad_norm": 0.05787520855665207,
      "learning_rate": 4.81410974479457e-06,
      "loss": 1.4419,
      "step": 21540
    },
    {
      "epoch": 0.9578053443599662,
      "grad_norm": 0.05773087963461876,
      "learning_rate": 4.793893564691843e-06,
      "loss": 1.4458,
      "step": 21542
    },
    {
      "epoch": 0.9578942688186386,
      "grad_norm": 0.057870473712682724,
      "learning_rate": 4.773719716989477e-06,
      "loss": 1.4454,
      "step": 21544
    },
    {
      "epoch": 0.957983193277311,
      "grad_norm": 0.05800451338291168,
      "learning_rate": 4.753588203411929e-06,
      "loss": 1.4423,
      "step": 21546
    },
    {
      "epoch": 0.9580721177359833,
      "grad_norm": 0.05686897784471512,
      "learning_rate": 4.7334990256800995e-06,
      "loss": 1.4462,
      "step": 21548
    },
    {
      "epoch": 0.9581610421946556,
      "grad_norm": 0.056949593126773834,
      "learning_rate": 4.7134521855113355e-06,
      "loss": 1.4418,
      "step": 21550
    },
    {
      "epoch": 0.958249966653328,
      "grad_norm": 0.05743638426065445,
      "learning_rate": 4.693447684619379e-06,
      "loss": 1.4471,
      "step": 21552
    },
    {
      "epoch": 0.9583388911120003,
      "grad_norm": 0.05902915075421333,
      "learning_rate": 4.673485524714305e-06,
      "loss": 1.4437,
      "step": 21554
    },
    {
      "epoch": 0.9584278155706727,
      "grad_norm": 0.058458685874938965,
      "learning_rate": 4.653565707502416e-06,
      "loss": 1.4422,
      "step": 21556
    },
    {
      "epoch": 0.9585167400293451,
      "grad_norm": 0.05795848369598389,
      "learning_rate": 4.633688234686739e-06,
      "loss": 1.4426,
      "step": 21558
    },
    {
      "epoch": 0.9586056644880174,
      "grad_norm": 0.058213118463754654,
      "learning_rate": 4.613853107966359e-06,
      "loss": 1.4477,
      "step": 21560
    },
    {
      "epoch": 0.9586945889466898,
      "grad_norm": 0.05738683417439461,
      "learning_rate": 4.59406032903692e-06,
      "loss": 1.4455,
      "step": 21562
    },
    {
      "epoch": 0.9587835134053622,
      "grad_norm": 0.057969436049461365,
      "learning_rate": 4.5743098995903455e-06,
      "loss": 1.4482,
      "step": 21564
    },
    {
      "epoch": 0.9588724378640345,
      "grad_norm": 0.05753438174724579,
      "learning_rate": 4.554601821315063e-06,
      "loss": 1.4418,
      "step": 21566
    },
    {
      "epoch": 0.9589613623227069,
      "grad_norm": 0.056868597865104675,
      "learning_rate": 4.534936095895781e-06,
      "loss": 1.449,
      "step": 21568
    },
    {
      "epoch": 0.9590502867813793,
      "grad_norm": 0.057507824152708054,
      "learning_rate": 4.515312725013598e-06,
      "loss": 1.4416,
      "step": 21570
    },
    {
      "epoch": 0.9591392112400515,
      "grad_norm": 0.05689457431435585,
      "learning_rate": 4.495731710346007e-06,
      "loss": 1.4416,
      "step": 21572
    },
    {
      "epoch": 0.9592281356987239,
      "grad_norm": 0.0581645630300045,
      "learning_rate": 4.47619305356689e-06,
      "loss": 1.446,
      "step": 21574
    },
    {
      "epoch": 0.9593170601573963,
      "grad_norm": 0.058234803378582,
      "learning_rate": 4.4566967563465234e-06,
      "loss": 1.4436,
      "step": 21576
    },
    {
      "epoch": 0.9594059846160686,
      "grad_norm": 0.05846071243286133,
      "learning_rate": 4.4372428203514615e-06,
      "loss": 1.4399,
      "step": 21578
    },
    {
      "epoch": 0.959494909074741,
      "grad_norm": 0.057688742876052856,
      "learning_rate": 4.417831247244819e-06,
      "loss": 1.4489,
      "step": 21580
    },
    {
      "epoch": 0.9595838335334134,
      "grad_norm": 0.058130089193582535,
      "learning_rate": 4.398462038685824e-06,
      "loss": 1.4435,
      "step": 21582
    },
    {
      "epoch": 0.9596727579920857,
      "grad_norm": 0.05822238698601723,
      "learning_rate": 4.3791351963304304e-06,
      "loss": 1.4503,
      "step": 21584
    },
    {
      "epoch": 0.9597616824507581,
      "grad_norm": 0.05791193246841431,
      "learning_rate": 4.3598507218306495e-06,
      "loss": 1.4482,
      "step": 21586
    },
    {
      "epoch": 0.9598506069094305,
      "grad_norm": 0.05749112367630005,
      "learning_rate": 4.340608616835051e-06,
      "loss": 1.4458,
      "step": 21588
    },
    {
      "epoch": 0.9599395313681028,
      "grad_norm": 0.0575961209833622,
      "learning_rate": 4.321408882988542e-06,
      "loss": 1.4476,
      "step": 21590
    },
    {
      "epoch": 0.9600284558267752,
      "grad_norm": 0.05781961977481842,
      "learning_rate": 4.302251521932366e-06,
      "loss": 1.4448,
      "step": 21592
    },
    {
      "epoch": 0.9601173802854475,
      "grad_norm": 0.05863596498966217,
      "learning_rate": 4.283136535304155e-06,
      "loss": 1.4491,
      "step": 21594
    },
    {
      "epoch": 0.9602063047441198,
      "grad_norm": 0.057458702474832535,
      "learning_rate": 4.264063924738104e-06,
      "loss": 1.4479,
      "step": 21596
    },
    {
      "epoch": 0.9602952292027922,
      "grad_norm": 0.057083528488874435,
      "learning_rate": 4.2450336918644085e-06,
      "loss": 1.4482,
      "step": 21598
    },
    {
      "epoch": 0.9603841536614646,
      "grad_norm": 0.05720750615000725,
      "learning_rate": 4.226045838309989e-06,
      "loss": 1.4473,
      "step": 21600
    },
    {
      "epoch": 0.9604730781201369,
      "grad_norm": 0.05795248597860336,
      "learning_rate": 4.207100365697936e-06,
      "loss": 1.4519,
      "step": 21602
    },
    {
      "epoch": 0.9605620025788093,
      "grad_norm": 0.057493992149829865,
      "learning_rate": 4.188197275647898e-06,
      "loss": 1.4457,
      "step": 21604
    },
    {
      "epoch": 0.9606509270374817,
      "grad_norm": 0.0582110695540905,
      "learning_rate": 4.169336569775695e-06,
      "loss": 1.4448,
      "step": 21606
    },
    {
      "epoch": 0.960739851496154,
      "grad_norm": 0.05797712132334709,
      "learning_rate": 4.150518249693647e-06,
      "loss": 1.4529,
      "step": 21608
    },
    {
      "epoch": 0.9608287759548264,
      "grad_norm": 0.0583646297454834,
      "learning_rate": 4.1317423170104675e-06,
      "loss": 1.4422,
      "step": 21610
    },
    {
      "epoch": 0.9609177004134988,
      "grad_norm": 0.05701034516096115,
      "learning_rate": 4.113008773331151e-06,
      "loss": 1.4434,
      "step": 21612
    },
    {
      "epoch": 0.9610066248721711,
      "grad_norm": 0.05788358673453331,
      "learning_rate": 4.094317620257138e-06,
      "loss": 1.4412,
      "step": 21614
    },
    {
      "epoch": 0.9610955493308434,
      "grad_norm": 0.05787377804517746,
      "learning_rate": 4.075668859386261e-06,
      "loss": 1.4417,
      "step": 21616
    },
    {
      "epoch": 0.9611844737895158,
      "grad_norm": 0.05681522190570831,
      "learning_rate": 4.057062492312691e-06,
      "loss": 1.4483,
      "step": 21618
    },
    {
      "epoch": 0.9612733982481881,
      "grad_norm": 0.05852191522717476,
      "learning_rate": 4.038498520626932e-06,
      "loss": 1.4491,
      "step": 21620
    },
    {
      "epoch": 0.9613623227068605,
      "grad_norm": 0.05827983468770981,
      "learning_rate": 4.0199769459159376e-06,
      "loss": 1.4491,
      "step": 21622
    },
    {
      "epoch": 0.9614512471655329,
      "grad_norm": 0.05857324227690697,
      "learning_rate": 4.001497769763107e-06,
      "loss": 1.4505,
      "step": 21624
    },
    {
      "epoch": 0.9615401716242052,
      "grad_norm": 0.058631282299757004,
      "learning_rate": 3.983060993747956e-06,
      "loss": 1.4475,
      "step": 21626
    },
    {
      "epoch": 0.9616290960828776,
      "grad_norm": 0.056906893849372864,
      "learning_rate": 3.964666619446666e-06,
      "loss": 1.4458,
      "step": 21628
    },
    {
      "epoch": 0.96171802054155,
      "grad_norm": 0.057671476155519485,
      "learning_rate": 3.946314648431648e-06,
      "loss": 1.4407,
      "step": 21630
    },
    {
      "epoch": 0.9618069450002223,
      "grad_norm": 0.057756319642066956,
      "learning_rate": 3.928005082271646e-06,
      "loss": 1.4427,
      "step": 21632
    },
    {
      "epoch": 0.9618958694588947,
      "grad_norm": 0.05779024213552475,
      "learning_rate": 3.909737922531909e-06,
      "loss": 1.4454,
      "step": 21634
    },
    {
      "epoch": 0.961984793917567,
      "grad_norm": 0.05701223015785217,
      "learning_rate": 3.891513170773964e-06,
      "loss": 1.4496,
      "step": 21636
    },
    {
      "epoch": 0.9620737183762393,
      "grad_norm": 0.05769617483019829,
      "learning_rate": 3.873330828555788e-06,
      "loss": 1.443,
      "step": 21638
    },
    {
      "epoch": 0.9621626428349117,
      "grad_norm": 0.05806345120072365,
      "learning_rate": 3.855190897431637e-06,
      "loss": 1.4428,
      "step": 21640
    },
    {
      "epoch": 0.9622515672935841,
      "grad_norm": 0.0573996864259243,
      "learning_rate": 3.837093378952217e-06,
      "loss": 1.4492,
      "step": 21642
    },
    {
      "epoch": 0.9623404917522564,
      "grad_norm": 0.05671732872724533,
      "learning_rate": 3.81903827466451e-06,
      "loss": 1.4445,
      "step": 21644
    },
    {
      "epoch": 0.9624294162109288,
      "grad_norm": 0.057457756251096725,
      "learning_rate": 3.801025586112061e-06,
      "loss": 1.4476,
      "step": 21646
    },
    {
      "epoch": 0.9625183406696012,
      "grad_norm": 0.05706573650240898,
      "learning_rate": 3.783055314834638e-06,
      "loss": 1.4462,
      "step": 21648
    },
    {
      "epoch": 0.9626072651282735,
      "grad_norm": 0.05761359632015228,
      "learning_rate": 3.765127462368345e-06,
      "loss": 1.4483,
      "step": 21650
    },
    {
      "epoch": 0.9626961895869459,
      "grad_norm": 0.057819437235593796,
      "learning_rate": 3.7472420302458455e-06,
      "loss": 1.4464,
      "step": 21652
    },
    {
      "epoch": 0.9627851140456183,
      "grad_norm": 0.05833171680569649,
      "learning_rate": 3.729399019996027e-06,
      "loss": 1.4449,
      "step": 21654
    },
    {
      "epoch": 0.9628740385042907,
      "grad_norm": 0.057252366095781326,
      "learning_rate": 3.71159843314417e-06,
      "loss": 1.4462,
      "step": 21656
    },
    {
      "epoch": 0.9629629629629629,
      "grad_norm": 0.058833926916122437,
      "learning_rate": 3.69384027121189e-06,
      "loss": 1.4441,
      "step": 21658
    },
    {
      "epoch": 0.9630518874216353,
      "grad_norm": 0.05726613849401474,
      "learning_rate": 3.676124535717307e-06,
      "loss": 1.4473,
      "step": 21660
    },
    {
      "epoch": 0.9631408118803076,
      "grad_norm": 0.05816185846924782,
      "learning_rate": 3.658451228174875e-06,
      "loss": 1.4407,
      "step": 21662
    },
    {
      "epoch": 0.96322973633898,
      "grad_norm": 0.05848447605967522,
      "learning_rate": 3.640820350095331e-06,
      "loss": 1.4431,
      "step": 21664
    },
    {
      "epoch": 0.9633186607976524,
      "grad_norm": 0.05794060230255127,
      "learning_rate": 3.6232319029858017e-06,
      "loss": 1.4511,
      "step": 21666
    },
    {
      "epoch": 0.9634075852563247,
      "grad_norm": 0.05679847672581673,
      "learning_rate": 3.6056858883499187e-06,
      "loss": 1.4469,
      "step": 21668
    },
    {
      "epoch": 0.9634965097149971,
      "grad_norm": 0.05763591080904007,
      "learning_rate": 3.588182307687482e-06,
      "loss": 1.4494,
      "step": 21670
    },
    {
      "epoch": 0.9635854341736695,
      "grad_norm": 0.05757317319512367,
      "learning_rate": 3.5707211624949055e-06,
      "loss": 1.4486,
      "step": 21672
    },
    {
      "epoch": 0.9636743586323419,
      "grad_norm": 0.057650867849588394,
      "learning_rate": 3.5533024542647174e-06,
      "loss": 1.4514,
      "step": 21674
    },
    {
      "epoch": 0.9637632830910142,
      "grad_norm": 0.057070985436439514,
      "learning_rate": 3.53592618448606e-06,
      "loss": 1.4482,
      "step": 21676
    },
    {
      "epoch": 0.9638522075496866,
      "grad_norm": 0.05852733552455902,
      "learning_rate": 3.5185923546442457e-06,
      "loss": 1.448,
      "step": 21678
    },
    {
      "epoch": 0.9639411320083588,
      "grad_norm": 0.058048900216817856,
      "learning_rate": 3.501300966221088e-06,
      "loss": 1.4493,
      "step": 21680
    },
    {
      "epoch": 0.9640300564670312,
      "grad_norm": 0.05750441551208496,
      "learning_rate": 3.4840520206947392e-06,
      "loss": 1.4466,
      "step": 21682
    },
    {
      "epoch": 0.9641189809257036,
      "grad_norm": 0.05819224938750267,
      "learning_rate": 3.4668455195396854e-06,
      "loss": 1.4456,
      "step": 21684
    },
    {
      "epoch": 0.964207905384376,
      "grad_norm": 0.057015545666217804,
      "learning_rate": 3.449681464226806e-06,
      "loss": 1.4428,
      "step": 21686
    },
    {
      "epoch": 0.9642968298430483,
      "grad_norm": 0.056825630366802216,
      "learning_rate": 3.4325598562234274e-06,
      "loss": 1.447,
      "step": 21688
    },
    {
      "epoch": 0.9643857543017207,
      "grad_norm": 0.05699056014418602,
      "learning_rate": 3.415480696993101e-06,
      "loss": 1.4432,
      "step": 21690
    },
    {
      "epoch": 0.9644746787603931,
      "grad_norm": 0.057919371873140335,
      "learning_rate": 3.3984439879958807e-06,
      "loss": 1.4484,
      "step": 21692
    },
    {
      "epoch": 0.9645636032190654,
      "grad_norm": 0.05731486156582832,
      "learning_rate": 3.381449730688102e-06,
      "loss": 1.4495,
      "step": 21694
    },
    {
      "epoch": 0.9646525276777378,
      "grad_norm": 0.05739450827240944,
      "learning_rate": 3.3644979265225474e-06,
      "loss": 1.449,
      "step": 21696
    },
    {
      "epoch": 0.9647414521364102,
      "grad_norm": 0.058767009526491165,
      "learning_rate": 3.3475885769482796e-06,
      "loss": 1.4444,
      "step": 21698
    },
    {
      "epoch": 0.9648303765950825,
      "grad_norm": 0.05814867466688156,
      "learning_rate": 3.330721683410809e-06,
      "loss": 1.4522,
      "step": 21700
    },
    {
      "epoch": 0.9649193010537548,
      "grad_norm": 0.05803316831588745,
      "learning_rate": 3.313897247352038e-06,
      "loss": 1.4435,
      "step": 21702
    },
    {
      "epoch": 0.9650082255124272,
      "grad_norm": 0.05720685422420502,
      "learning_rate": 3.297115270210149e-06,
      "loss": 1.4459,
      "step": 21704
    },
    {
      "epoch": 0.9650971499710995,
      "grad_norm": 0.05773816257715225,
      "learning_rate": 3.2803757534197175e-06,
      "loss": 1.4459,
      "step": 21706
    },
    {
      "epoch": 0.9651860744297719,
      "grad_norm": 0.05913296714425087,
      "learning_rate": 3.263678698411765e-06,
      "loss": 1.4439,
      "step": 21708
    },
    {
      "epoch": 0.9652749988884443,
      "grad_norm": 0.057658907026052475,
      "learning_rate": 3.2470241066136495e-06,
      "loss": 1.4482,
      "step": 21710
    },
    {
      "epoch": 0.9653639233471166,
      "grad_norm": 0.05754270777106285,
      "learning_rate": 3.230411979448955e-06,
      "loss": 1.4479,
      "step": 21712
    },
    {
      "epoch": 0.965452847805789,
      "grad_norm": 0.05791141465306282,
      "learning_rate": 3.2138423183378785e-06,
      "loss": 1.4476,
      "step": 21714
    },
    {
      "epoch": 0.9655417722644614,
      "grad_norm": 0.057228460907936096,
      "learning_rate": 3.1973151246967868e-06,
      "loss": 1.4455,
      "step": 21716
    },
    {
      "epoch": 0.9656306967231337,
      "grad_norm": 0.05856222286820412,
      "learning_rate": 3.18083039993855e-06,
      "loss": 1.4421,
      "step": 21718
    },
    {
      "epoch": 0.9657196211818061,
      "grad_norm": 0.05858870595693588,
      "learning_rate": 3.1643881454723744e-06,
      "loss": 1.4433,
      "step": 21720
    },
    {
      "epoch": 0.9658085456404785,
      "grad_norm": 0.057810429483652115,
      "learning_rate": 3.1479883627037464e-06,
      "loss": 1.4414,
      "step": 21722
    },
    {
      "epoch": 0.9658974700991507,
      "grad_norm": 0.05786995589733124,
      "learning_rate": 3.131631053034656e-06,
      "loss": 1.4479,
      "step": 21724
    },
    {
      "epoch": 0.9659863945578231,
      "grad_norm": 0.05765226110816002,
      "learning_rate": 3.115316217863373e-06,
      "loss": 1.444,
      "step": 21726
    },
    {
      "epoch": 0.9660753190164955,
      "grad_norm": 0.05825477093458176,
      "learning_rate": 3.0990438585845048e-06,
      "loss": 1.4505,
      "step": 21728
    },
    {
      "epoch": 0.9661642434751678,
      "grad_norm": 0.057248134166002274,
      "learning_rate": 3.082813976589216e-06,
      "loss": 1.4484,
      "step": 21730
    },
    {
      "epoch": 0.9662531679338402,
      "grad_norm": 0.057661522179841995,
      "learning_rate": 3.0666265732647856e-06,
      "loss": 1.4497,
      "step": 21732
    },
    {
      "epoch": 0.9663420923925126,
      "grad_norm": 0.05723179504275322,
      "learning_rate": 3.0504816499950516e-06,
      "loss": 1.4442,
      "step": 21734
    },
    {
      "epoch": 0.9664310168511849,
      "grad_norm": 0.05735337361693382,
      "learning_rate": 3.034379208160076e-06,
      "loss": 1.4482,
      "step": 21736
    },
    {
      "epoch": 0.9665199413098573,
      "grad_norm": 0.05809729918837547,
      "learning_rate": 3.018319249136481e-06,
      "loss": 1.4436,
      "step": 21738
    },
    {
      "epoch": 0.9666088657685297,
      "grad_norm": 0.05790387839078903,
      "learning_rate": 3.002301774297056e-06,
      "loss": 1.4404,
      "step": 21740
    },
    {
      "epoch": 0.966697790227202,
      "grad_norm": 0.05688699334859848,
      "learning_rate": 2.9863267850110955e-06,
      "loss": 1.449,
      "step": 21742
    },
    {
      "epoch": 0.9667867146858744,
      "grad_norm": 0.056998882442712784,
      "learning_rate": 2.9703942826441734e-06,
      "loss": 1.4472,
      "step": 21744
    },
    {
      "epoch": 0.9668756391445467,
      "grad_norm": 0.058484215289354324,
      "learning_rate": 2.9545042685583112e-06,
      "loss": 1.4486,
      "step": 21746
    },
    {
      "epoch": 0.966964563603219,
      "grad_norm": 0.05785754323005676,
      "learning_rate": 2.938656744111812e-06,
      "loss": 1.4462,
      "step": 21748
    },
    {
      "epoch": 0.9670534880618914,
      "grad_norm": 0.05740215256810188,
      "learning_rate": 2.9228517106594244e-06,
      "loss": 1.4409,
      "step": 21750
    },
    {
      "epoch": 0.9671424125205638,
      "grad_norm": 0.057988543063402176,
      "learning_rate": 2.9070891695521796e-06,
      "loss": 1.4468,
      "step": 21752
    },
    {
      "epoch": 0.9672313369792361,
      "grad_norm": 0.05775817483663559,
      "learning_rate": 2.8913691221376102e-06,
      "loss": 1.4534,
      "step": 21754
    },
    {
      "epoch": 0.9673202614379085,
      "grad_norm": 0.05813591554760933,
      "learning_rate": 2.8756915697594756e-06,
      "loss": 1.4463,
      "step": 21756
    },
    {
      "epoch": 0.9674091858965809,
      "grad_norm": 0.05779729038476944,
      "learning_rate": 2.86005651375798e-06,
      "loss": 1.4475,
      "step": 21758
    },
    {
      "epoch": 0.9674981103552532,
      "grad_norm": 0.05689067393541336,
      "learning_rate": 2.844463955469723e-06,
      "loss": 1.4441,
      "step": 21760
    },
    {
      "epoch": 0.9675870348139256,
      "grad_norm": 0.05785556882619858,
      "learning_rate": 2.828913896227525e-06,
      "loss": 1.446,
      "step": 21762
    },
    {
      "epoch": 0.967675959272598,
      "grad_norm": 0.0574394054710865,
      "learning_rate": 2.8134063373607687e-06,
      "loss": 1.4418,
      "step": 21764
    },
    {
      "epoch": 0.9677648837312702,
      "grad_norm": 0.05662890523672104,
      "learning_rate": 2.79794128019506e-06,
      "loss": 1.4432,
      "step": 21766
    },
    {
      "epoch": 0.9678538081899426,
      "grad_norm": 0.0582660436630249,
      "learning_rate": 2.7825187260523966e-06,
      "loss": 1.4476,
      "step": 21768
    },
    {
      "epoch": 0.967942732648615,
      "grad_norm": 0.05858684331178665,
      "learning_rate": 2.767138676251224e-06,
      "loss": 1.4451,
      "step": 21770
    },
    {
      "epoch": 0.9680316571072873,
      "grad_norm": 0.056393761187791824,
      "learning_rate": 2.7518011321062687e-06,
      "loss": 1.444,
      "step": 21772
    },
    {
      "epoch": 0.9681205815659597,
      "grad_norm": 0.058356158435344696,
      "learning_rate": 2.736506094928648e-06,
      "loss": 1.4424,
      "step": 21774
    },
    {
      "epoch": 0.9682095060246321,
      "grad_norm": 0.05726497247815132,
      "learning_rate": 2.7212535660258718e-06,
      "loss": 1.4445,
      "step": 21776
    },
    {
      "epoch": 0.9682984304833044,
      "grad_norm": 0.057576339691877365,
      "learning_rate": 2.7060435467017865e-06,
      "loss": 1.4466,
      "step": 21778
    },
    {
      "epoch": 0.9683873549419768,
      "grad_norm": 0.05778598040342331,
      "learning_rate": 2.6908760382565735e-06,
      "loss": 1.448,
      "step": 21780
    },
    {
      "epoch": 0.9684762794006492,
      "grad_norm": 0.056812673807144165,
      "learning_rate": 2.6757510419868624e-06,
      "loss": 1.4487,
      "step": 21782
    },
    {
      "epoch": 0.9685652038593215,
      "grad_norm": 0.05817960575222969,
      "learning_rate": 2.660668559185564e-06,
      "loss": 1.4498,
      "step": 21784
    },
    {
      "epoch": 0.9686541283179939,
      "grad_norm": 0.05673222243785858,
      "learning_rate": 2.6456285911420353e-06,
      "loss": 1.4489,
      "step": 21786
    },
    {
      "epoch": 0.9687430527766662,
      "grad_norm": 0.058771610260009766,
      "learning_rate": 2.63063113914197e-06,
      "loss": 1.4485,
      "step": 21788
    },
    {
      "epoch": 0.9688319772353385,
      "grad_norm": 0.05711560696363449,
      "learning_rate": 2.615676204467343e-06,
      "loss": 1.4452,
      "step": 21790
    },
    {
      "epoch": 0.9689209016940109,
      "grad_norm": 0.057385995984077454,
      "learning_rate": 2.6007637883966872e-06,
      "loss": 1.4476,
      "step": 21792
    },
    {
      "epoch": 0.9690098261526833,
      "grad_norm": 0.05783890560269356,
      "learning_rate": 2.5858938922046495e-06,
      "loss": 1.4458,
      "step": 21794
    },
    {
      "epoch": 0.9690987506113556,
      "grad_norm": 0.05787234753370285,
      "learning_rate": 2.571066517162435e-06,
      "loss": 1.4456,
      "step": 21796
    },
    {
      "epoch": 0.969187675070028,
      "grad_norm": 0.05799291655421257,
      "learning_rate": 2.556281664537585e-06,
      "loss": 1.4469,
      "step": 21798
    },
    {
      "epoch": 0.9692765995287004,
      "grad_norm": 0.05841020122170448,
      "learning_rate": 2.541539335593923e-06,
      "loss": 1.4422,
      "step": 21800
    },
    {
      "epoch": 0.9693655239873727,
      "grad_norm": 0.057031866163015366,
      "learning_rate": 2.526839531591718e-06,
      "loss": 1.4509,
      "step": 21802
    },
    {
      "epoch": 0.9694544484460451,
      "grad_norm": 0.05724103003740311,
      "learning_rate": 2.5121822537875206e-06,
      "loss": 1.4438,
      "step": 21804
    },
    {
      "epoch": 0.9695433729047175,
      "grad_norm": 0.057726915925741196,
      "learning_rate": 2.4975675034343835e-06,
      "loss": 1.4444,
      "step": 21806
    },
    {
      "epoch": 0.9696322973633899,
      "grad_norm": 0.058166489005088806,
      "learning_rate": 2.482995281781586e-06,
      "loss": 1.4455,
      "step": 21808
    },
    {
      "epoch": 0.9697212218220621,
      "grad_norm": 0.058684688061475754,
      "learning_rate": 2.4684655900748532e-06,
      "loss": 1.4486,
      "step": 21810
    },
    {
      "epoch": 0.9698101462807345,
      "grad_norm": 0.05730533227324486,
      "learning_rate": 2.453978429556247e-06,
      "loss": 1.4456,
      "step": 21812
    },
    {
      "epoch": 0.9698990707394068,
      "grad_norm": 0.057590242475271225,
      "learning_rate": 2.439533801464111e-06,
      "loss": 1.4445,
      "step": 21814
    },
    {
      "epoch": 0.9699879951980792,
      "grad_norm": 0.057827651500701904,
      "learning_rate": 2.425131707033346e-06,
      "loss": 1.4461,
      "step": 21816
    },
    {
      "epoch": 0.9700769196567516,
      "grad_norm": 0.05834771320223808,
      "learning_rate": 2.4107721474950774e-06,
      "loss": 1.4415,
      "step": 21818
    },
    {
      "epoch": 0.970165844115424,
      "grad_norm": 0.057678621262311935,
      "learning_rate": 2.3964551240767684e-06,
      "loss": 1.4477,
      "step": 21820
    },
    {
      "epoch": 0.9702547685740963,
      "grad_norm": 0.05750925838947296,
      "learning_rate": 2.382180638002385e-06,
      "loss": 1.4441,
      "step": 21822
    },
    {
      "epoch": 0.9703436930327687,
      "grad_norm": 0.056883055716753006,
      "learning_rate": 2.3679486904921164e-06,
      "loss": 1.4452,
      "step": 21824
    },
    {
      "epoch": 0.970432617491441,
      "grad_norm": 0.057405877858400345,
      "learning_rate": 2.353759282762602e-06,
      "loss": 1.4487,
      "step": 21826
    },
    {
      "epoch": 0.9705215419501134,
      "grad_norm": 0.056650012731552124,
      "learning_rate": 2.339612416026815e-06,
      "loss": 1.4426,
      "step": 21828
    },
    {
      "epoch": 0.9706104664087858,
      "grad_norm": 0.05729759484529495,
      "learning_rate": 2.3255080914940106e-06,
      "loss": 1.4483,
      "step": 21830
    },
    {
      "epoch": 0.970699390867458,
      "grad_norm": 0.05801773816347122,
      "learning_rate": 2.3114463103700023e-06,
      "loss": 1.4429,
      "step": 21832
    },
    {
      "epoch": 0.9707883153261304,
      "grad_norm": 0.057637542486190796,
      "learning_rate": 2.2974270738567726e-06,
      "loss": 1.4442,
      "step": 21834
    },
    {
      "epoch": 0.9708772397848028,
      "grad_norm": 0.058352191001176834,
      "learning_rate": 2.2834503831528076e-06,
      "loss": 1.4448,
      "step": 21836
    },
    {
      "epoch": 0.9709661642434751,
      "grad_norm": 0.057397257536649704,
      "learning_rate": 2.2695162394528735e-06,
      "loss": 1.4414,
      "step": 21838
    },
    {
      "epoch": 0.9710550887021475,
      "grad_norm": 0.05693025141954422,
      "learning_rate": 2.2556246439481286e-06,
      "loss": 1.4481,
      "step": 21840
    },
    {
      "epoch": 0.9711440131608199,
      "grad_norm": 0.05720305070281029,
      "learning_rate": 2.2417755978260123e-06,
      "loss": 1.452,
      "step": 21842
    },
    {
      "epoch": 0.9712329376194923,
      "grad_norm": 0.057424396276474,
      "learning_rate": 2.2279691022705216e-06,
      "loss": 1.4438,
      "step": 21844
    },
    {
      "epoch": 0.9713218620781646,
      "grad_norm": 0.057667020708322525,
      "learning_rate": 2.2142051584618237e-06,
      "loss": 1.4478,
      "step": 21846
    },
    {
      "epoch": 0.971410786536837,
      "grad_norm": 0.058313172310590744,
      "learning_rate": 2.200483767576589e-06,
      "loss": 1.4476,
      "step": 21848
    },
    {
      "epoch": 0.9714997109955094,
      "grad_norm": 0.057696230709552765,
      "learning_rate": 2.1868049307877113e-06,
      "loss": 1.4433,
      "step": 21850
    },
    {
      "epoch": 0.9715886354541817,
      "grad_norm": 0.057298045605421066,
      "learning_rate": 2.1731686492644787e-06,
      "loss": 1.447,
      "step": 21852
    },
    {
      "epoch": 0.971677559912854,
      "grad_norm": 0.05838226154446602,
      "learning_rate": 2.1595749241726805e-06,
      "loss": 1.4426,
      "step": 21854
    },
    {
      "epoch": 0.9717664843715264,
      "grad_norm": 0.057391196489334106,
      "learning_rate": 2.146023756674331e-06,
      "loss": 1.4436,
      "step": 21856
    },
    {
      "epoch": 0.9718554088301987,
      "grad_norm": 0.05735941603779793,
      "learning_rate": 2.132515147927838e-06,
      "loss": 1.4449,
      "step": 21858
    },
    {
      "epoch": 0.9719443332888711,
      "grad_norm": 0.05849364399909973,
      "learning_rate": 2.1190490990879994e-06,
      "loss": 1.4509,
      "step": 21860
    },
    {
      "epoch": 0.9720332577475435,
      "grad_norm": 0.05746670067310333,
      "learning_rate": 2.10562561130595e-06,
      "loss": 1.4453,
      "step": 21862
    },
    {
      "epoch": 0.9721221822062158,
      "grad_norm": 0.05694587528705597,
      "learning_rate": 2.0922446857291055e-06,
      "loss": 1.4459,
      "step": 21864
    },
    {
      "epoch": 0.9722111066648882,
      "grad_norm": 0.05823846161365509,
      "learning_rate": 2.0789063235014396e-06,
      "loss": 1.4386,
      "step": 21866
    },
    {
      "epoch": 0.9723000311235606,
      "grad_norm": 0.05770362168550491,
      "learning_rate": 2.065610525763095e-06,
      "loss": 1.4468,
      "step": 21868
    },
    {
      "epoch": 0.9723889555822329,
      "grad_norm": 0.05771014466881752,
      "learning_rate": 2.052357293650775e-06,
      "loss": 1.4437,
      "step": 21870
    },
    {
      "epoch": 0.9724778800409053,
      "grad_norm": 0.058271683752536774,
      "learning_rate": 2.039146628297295e-06,
      "loss": 1.446,
      "step": 21872
    },
    {
      "epoch": 0.9725668044995777,
      "grad_norm": 0.057467687875032425,
      "learning_rate": 2.0259785308319176e-06,
      "loss": 1.4474,
      "step": 21874
    },
    {
      "epoch": 0.9726557289582499,
      "grad_norm": 0.058236099779605865,
      "learning_rate": 2.012853002380466e-06,
      "loss": 1.4466,
      "step": 21876
    },
    {
      "epoch": 0.9727446534169223,
      "grad_norm": 0.057283271104097366,
      "learning_rate": 1.9997700440649302e-06,
      "loss": 1.4443,
      "step": 21878
    },
    {
      "epoch": 0.9728335778755947,
      "grad_norm": 0.057595402002334595,
      "learning_rate": 1.9867296570036387e-06,
      "loss": 1.4457,
      "step": 21880
    },
    {
      "epoch": 0.972922502334267,
      "grad_norm": 0.05784250423312187,
      "learning_rate": 1.973731842311366e-06,
      "loss": 1.4421,
      "step": 21882
    },
    {
      "epoch": 0.9730114267929394,
      "grad_norm": 0.057085875421762466,
      "learning_rate": 1.9607766010991677e-06,
      "loss": 1.4472,
      "step": 21884
    },
    {
      "epoch": 0.9731003512516118,
      "grad_norm": 0.05725502967834473,
      "learning_rate": 1.9478639344746585e-06,
      "loss": 1.4512,
      "step": 21886
    },
    {
      "epoch": 0.9731892757102841,
      "grad_norm": 0.057032741606235504,
      "learning_rate": 1.934993843541566e-06,
      "loss": 1.4412,
      "step": 21888
    },
    {
      "epoch": 0.9732782001689565,
      "grad_norm": 0.056812822818756104,
      "learning_rate": 1.9221663294000657e-06,
      "loss": 1.4454,
      "step": 21890
    },
    {
      "epoch": 0.9733671246276289,
      "grad_norm": 0.057856421917676926,
      "learning_rate": 1.9093813931467807e-06,
      "loss": 1.4446,
      "step": 21892
    },
    {
      "epoch": 0.9734560490863012,
      "grad_norm": 0.05846314877271652,
      "learning_rate": 1.8966390358746145e-06,
      "loss": 1.4441,
      "step": 21894
    },
    {
      "epoch": 0.9735449735449735,
      "grad_norm": 0.05669744685292244,
      "learning_rate": 1.883939258672751e-06,
      "loss": 1.4504,
      "step": 21896
    },
    {
      "epoch": 0.9736338980036459,
      "grad_norm": 0.05733288824558258,
      "learning_rate": 1.8712820626268778e-06,
      "loss": 1.453,
      "step": 21898
    },
    {
      "epoch": 0.9737228224623182,
      "grad_norm": 0.05750066787004471,
      "learning_rate": 1.8586674488190736e-06,
      "loss": 1.4435,
      "step": 21900
    },
    {
      "epoch": 0.9738117469209906,
      "grad_norm": 0.05767687410116196,
      "learning_rate": 1.8460954183275314e-06,
      "loss": 1.4426,
      "step": 21902
    },
    {
      "epoch": 0.973900671379663,
      "grad_norm": 0.05756646394729614,
      "learning_rate": 1.8335659722271136e-06,
      "loss": 1.4368,
      "step": 21904
    },
    {
      "epoch": 0.9739895958383353,
      "grad_norm": 0.057719554752111435,
      "learning_rate": 1.8210791115887971e-06,
      "loss": 1.4422,
      "step": 21906
    },
    {
      "epoch": 0.9740785202970077,
      "grad_norm": 0.05758875980973244,
      "learning_rate": 1.8086348374800055e-06,
      "loss": 1.4432,
      "step": 21908
    },
    {
      "epoch": 0.9741674447556801,
      "grad_norm": 0.0580807626247406,
      "learning_rate": 1.7962331509646103e-06,
      "loss": 1.4417,
      "step": 21910
    },
    {
      "epoch": 0.9742563692143524,
      "grad_norm": 0.057357918471097946,
      "learning_rate": 1.7838740531027076e-06,
      "loss": 1.4442,
      "step": 21912
    },
    {
      "epoch": 0.9743452936730248,
      "grad_norm": 0.05980123206973076,
      "learning_rate": 1.7715575449508413e-06,
      "loss": 1.4484,
      "step": 21914
    },
    {
      "epoch": 0.9744342181316972,
      "grad_norm": 0.05732397362589836,
      "learning_rate": 1.7592836275618361e-06,
      "loss": 1.4435,
      "step": 21916
    },
    {
      "epoch": 0.9745231425903694,
      "grad_norm": 0.05790344625711441,
      "learning_rate": 1.7470523019849084e-06,
      "loss": 1.446,
      "step": 21918
    },
    {
      "epoch": 0.9746120670490418,
      "grad_norm": 0.0583200603723526,
      "learning_rate": 1.734863569265721e-06,
      "loss": 1.4472,
      "step": 21920
    },
    {
      "epoch": 0.9747009915077142,
      "grad_norm": 0.057706188410520554,
      "learning_rate": 1.7227174304461635e-06,
      "loss": 1.4404,
      "step": 21922
    },
    {
      "epoch": 0.9747899159663865,
      "grad_norm": 0.05923660844564438,
      "learning_rate": 1.7106138865645716e-06,
      "loss": 1.4471,
      "step": 21924
    },
    {
      "epoch": 0.9748788404250589,
      "grad_norm": 0.057506807148456573,
      "learning_rate": 1.698552938655562e-06,
      "loss": 1.4454,
      "step": 21926
    },
    {
      "epoch": 0.9749677648837313,
      "grad_norm": 0.057615265250205994,
      "learning_rate": 1.6865345877502546e-06,
      "loss": 1.4465,
      "step": 21928
    },
    {
      "epoch": 0.9750566893424036,
      "grad_norm": 0.05670270696282387,
      "learning_rate": 1.6745588348758833e-06,
      "loss": 1.4418,
      "step": 21930
    },
    {
      "epoch": 0.975145613801076,
      "grad_norm": 0.057712372392416,
      "learning_rate": 1.662625681056351e-06,
      "loss": 1.4462,
      "step": 21932
    },
    {
      "epoch": 0.9752345382597484,
      "grad_norm": 0.05813715234398842,
      "learning_rate": 1.65073512731162e-06,
      "loss": 1.446,
      "step": 21934
    },
    {
      "epoch": 0.9753234627184207,
      "grad_norm": 0.05662925913929939,
      "learning_rate": 1.6388871746582102e-06,
      "loss": 1.4411,
      "step": 21936
    },
    {
      "epoch": 0.9754123871770931,
      "grad_norm": 0.056773193180561066,
      "learning_rate": 1.627081824108978e-06,
      "loss": 1.4469,
      "step": 21938
    },
    {
      "epoch": 0.9755013116357654,
      "grad_norm": 0.057524967938661575,
      "learning_rate": 1.6153190766730053e-06,
      "loss": 1.4463,
      "step": 21940
    },
    {
      "epoch": 0.9755902360944377,
      "grad_norm": 0.05839113891124725,
      "learning_rate": 1.6035989333558765e-06,
      "loss": 1.4537,
      "step": 21942
    },
    {
      "epoch": 0.9756791605531101,
      "grad_norm": 0.058153193444013596,
      "learning_rate": 1.5919213951594569e-06,
      "loss": 1.4476,
      "step": 21944
    },
    {
      "epoch": 0.9757680850117825,
      "grad_norm": 0.05714964494109154,
      "learning_rate": 1.5802864630820591e-06,
      "loss": 1.4424,
      "step": 21946
    },
    {
      "epoch": 0.9758570094704548,
      "grad_norm": 0.0575794093310833,
      "learning_rate": 1.5686941381182206e-06,
      "loss": 1.4466,
      "step": 21948
    },
    {
      "epoch": 0.9759459339291272,
      "grad_norm": 0.057474132627248764,
      "learning_rate": 1.5571444212588715e-06,
      "loss": 1.4455,
      "step": 21950
    },
    {
      "epoch": 0.9760348583877996,
      "grad_norm": 0.05650651827454567,
      "learning_rate": 1.5456373134914437e-06,
      "loss": 1.449,
      "step": 21952
    },
    {
      "epoch": 0.9761237828464719,
      "grad_norm": 0.05746529996395111,
      "learning_rate": 1.534172815799484e-06,
      "loss": 1.4398,
      "step": 21954
    },
    {
      "epoch": 0.9762127073051443,
      "grad_norm": 0.05808810144662857,
      "learning_rate": 1.522750929163208e-06,
      "loss": 1.439,
      "step": 21956
    },
    {
      "epoch": 0.9763016317638167,
      "grad_norm": 0.05777526646852493,
      "learning_rate": 1.5113716545588352e-06,
      "loss": 1.4447,
      "step": 21958
    },
    {
      "epoch": 0.976390556222489,
      "grad_norm": 0.057585395872592926,
      "learning_rate": 1.5000349929591982e-06,
      "loss": 1.4507,
      "step": 21960
    },
    {
      "epoch": 0.9764794806811613,
      "grad_norm": 0.05766819417476654,
      "learning_rate": 1.4887409453333555e-06,
      "loss": 1.4515,
      "step": 21962
    },
    {
      "epoch": 0.9765684051398337,
      "grad_norm": 0.05719029903411865,
      "learning_rate": 1.4774895126468125e-06,
      "loss": 1.4468,
      "step": 21964
    },
    {
      "epoch": 0.976657329598506,
      "grad_norm": 0.05869666114449501,
      "learning_rate": 1.4662806958614105e-06,
      "loss": 1.4479,
      "step": 21966
    },
    {
      "epoch": 0.9767462540571784,
      "grad_norm": 0.057645734399557114,
      "learning_rate": 1.4551144959352724e-06,
      "loss": 1.4539,
      "step": 21968
    },
    {
      "epoch": 0.9768351785158508,
      "grad_norm": 0.058074142783880234,
      "learning_rate": 1.443990913822968e-06,
      "loss": 1.4441,
      "step": 21970
    },
    {
      "epoch": 0.9769241029745231,
      "grad_norm": 0.057652901858091354,
      "learning_rate": 1.432909950475403e-06,
      "loss": 1.4502,
      "step": 21972
    },
    {
      "epoch": 0.9770130274331955,
      "grad_norm": 0.056621208786964417,
      "learning_rate": 1.4218716068398196e-06,
      "loss": 1.4418,
      "step": 21974
    },
    {
      "epoch": 0.9771019518918679,
      "grad_norm": 0.05749661475419998,
      "learning_rate": 1.4108758838597969e-06,
      "loss": 1.4433,
      "step": 21976
    },
    {
      "epoch": 0.9771908763505402,
      "grad_norm": 0.05827660858631134,
      "learning_rate": 1.3999227824753047e-06,
      "loss": 1.4444,
      "step": 21978
    },
    {
      "epoch": 0.9772798008092126,
      "grad_norm": 0.057929717004299164,
      "learning_rate": 1.3890123036227054e-06,
      "loss": 1.4509,
      "step": 21980
    },
    {
      "epoch": 0.977368725267885,
      "grad_norm": 0.058121420443058014,
      "learning_rate": 1.3781444482345863e-06,
      "loss": 1.4477,
      "step": 21982
    },
    {
      "epoch": 0.9774576497265572,
      "grad_norm": 0.057763341814279556,
      "learning_rate": 1.3673192172400927e-06,
      "loss": 1.4427,
      "step": 21984
    },
    {
      "epoch": 0.9775465741852296,
      "grad_norm": 0.05874239280819893,
      "learning_rate": 1.3565366115645405e-06,
      "loss": 1.4478,
      "step": 21986
    },
    {
      "epoch": 0.977635498643902,
      "grad_norm": 0.05708703398704529,
      "learning_rate": 1.3457966321296921e-06,
      "loss": 1.4451,
      "step": 21988
    },
    {
      "epoch": 0.9777244231025743,
      "grad_norm": 0.05787283182144165,
      "learning_rate": 1.335099279853591e-06,
      "loss": 1.4457,
      "step": 21990
    },
    {
      "epoch": 0.9778133475612467,
      "grad_norm": 0.05789671465754509,
      "learning_rate": 1.3244445556507834e-06,
      "loss": 1.445,
      "step": 21992
    },
    {
      "epoch": 0.9779022720199191,
      "grad_norm": 0.0576728992164135,
      "learning_rate": 1.3138324604320961e-06,
      "loss": 1.4467,
      "step": 21994
    },
    {
      "epoch": 0.9779911964785915,
      "grad_norm": 0.05753236636519432,
      "learning_rate": 1.3032629951045817e-06,
      "loss": 1.4464,
      "step": 21996
    },
    {
      "epoch": 0.9780801209372638,
      "grad_norm": 0.05750956013798714,
      "learning_rate": 1.2927361605718502e-06,
      "loss": 1.4436,
      "step": 21998
    },
    {
      "epoch": 0.9781690453959362,
      "grad_norm": 0.05757934972643852,
      "learning_rate": 1.2822519577337932e-06,
      "loss": 1.4457,
      "step": 22000
    },
    {
      "epoch": 0.9781690453959362,
      "eval_loss": 1.4320533275604248,
      "eval_runtime": 12.4698,
      "eval_samples_per_second": 554.141,
      "eval_steps_per_second": 69.288,
      "step": 22000
    },
    {
      "epoch": 0.9782579698546086,
      "grad_norm": 0.057963717728853226,
      "learning_rate": 1.2718103874865827e-06,
      "loss": 1.4517,
      "step": 22002
    },
    {
      "epoch": 0.9783468943132808,
      "grad_norm": 0.05782179906964302,
      "learning_rate": 1.261411450722838e-06,
      "loss": 1.4397,
      "step": 22004
    },
    {
      "epoch": 0.9784358187719532,
      "grad_norm": 0.05823444202542305,
      "learning_rate": 1.2510551483315146e-06,
      "loss": 1.4472,
      "step": 22006
    },
    {
      "epoch": 0.9785247432306255,
      "grad_norm": 0.057919129729270935,
      "learning_rate": 1.2407414811979601e-06,
      "loss": 1.4441,
      "step": 22008
    },
    {
      "epoch": 0.9786136676892979,
      "grad_norm": 0.058388952165842056,
      "learning_rate": 1.2304704502037467e-06,
      "loss": 1.4497,
      "step": 22010
    },
    {
      "epoch": 0.9787025921479703,
      "grad_norm": 0.057490069419145584,
      "learning_rate": 1.2202420562268946e-06,
      "loss": 1.4493,
      "step": 22012
    },
    {
      "epoch": 0.9787915166066427,
      "grad_norm": 0.05721807852387428,
      "learning_rate": 1.2100563001418708e-06,
      "loss": 1.4464,
      "step": 22014
    },
    {
      "epoch": 0.978880441065315,
      "grad_norm": 0.057543035596609116,
      "learning_rate": 1.1999131828192567e-06,
      "loss": 1.4467,
      "step": 22016
    },
    {
      "epoch": 0.9789693655239874,
      "grad_norm": 0.057324040681123734,
      "learning_rate": 1.1898127051262476e-06,
      "loss": 1.4435,
      "step": 22018
    },
    {
      "epoch": 0.9790582899826598,
      "grad_norm": 0.05752379819750786,
      "learning_rate": 1.1797548679262638e-06,
      "loss": 1.4421,
      "step": 22020
    },
    {
      "epoch": 0.9791472144413321,
      "grad_norm": 0.05681290104985237,
      "learning_rate": 1.1697396720790065e-06,
      "loss": 1.4482,
      "step": 22022
    },
    {
      "epoch": 0.9792361389000045,
      "grad_norm": 0.05741885304450989,
      "learning_rate": 1.15976711844068e-06,
      "loss": 1.4459,
      "step": 22024
    },
    {
      "epoch": 0.9793250633586768,
      "grad_norm": 0.05764324218034744,
      "learning_rate": 1.1498372078638243e-06,
      "loss": 1.4434,
      "step": 22026
    },
    {
      "epoch": 0.9794139878173491,
      "grad_norm": 0.05784038081765175,
      "learning_rate": 1.1399499411972048e-06,
      "loss": 1.4504,
      "step": 22028
    },
    {
      "epoch": 0.9795029122760215,
      "grad_norm": 0.05776010826230049,
      "learning_rate": 1.13010531928609e-06,
      "loss": 1.4494,
      "step": 22030
    },
    {
      "epoch": 0.9795918367346939,
      "grad_norm": 0.05751819536089897,
      "learning_rate": 1.1203033429719734e-06,
      "loss": 1.4483,
      "step": 22032
    },
    {
      "epoch": 0.9796807611933662,
      "grad_norm": 0.05655774101614952,
      "learning_rate": 1.1105440130929067e-06,
      "loss": 1.4474,
      "step": 22034
    },
    {
      "epoch": 0.9797696856520386,
      "grad_norm": 0.0575784407556057,
      "learning_rate": 1.1008273304830008e-06,
      "loss": 1.4497,
      "step": 22036
    },
    {
      "epoch": 0.979858610110711,
      "grad_norm": 0.05941782891750336,
      "learning_rate": 1.0911532959729797e-06,
      "loss": 1.45,
      "step": 22038
    },
    {
      "epoch": 0.9799475345693833,
      "grad_norm": 0.05766258016228676,
      "learning_rate": 1.0815219103897933e-06,
      "loss": 1.4442,
      "step": 22040
    },
    {
      "epoch": 0.9800364590280557,
      "grad_norm": 0.056954920291900635,
      "learning_rate": 1.0719331745567828e-06,
      "loss": 1.4441,
      "step": 22042
    },
    {
      "epoch": 0.9801253834867281,
      "grad_norm": 0.05810870602726936,
      "learning_rate": 1.0623870892936261e-06,
      "loss": 1.4414,
      "step": 22044
    },
    {
      "epoch": 0.9802143079454004,
      "grad_norm": 0.05758174881339073,
      "learning_rate": 1.0528836554163368e-06,
      "loss": 1.4461,
      "step": 22046
    },
    {
      "epoch": 0.9803032324040727,
      "grad_norm": 0.05731727182865143,
      "learning_rate": 1.0434228737373764e-06,
      "loss": 1.4426,
      "step": 22048
    },
    {
      "epoch": 0.9803921568627451,
      "grad_norm": 0.05725093558430672,
      "learning_rate": 1.034004745065431e-06,
      "loss": 1.4474,
      "step": 22050
    },
    {
      "epoch": 0.9804810813214174,
      "grad_norm": 0.0574093721807003,
      "learning_rate": 1.0246292702056348e-06,
      "loss": 1.4473,
      "step": 22052
    },
    {
      "epoch": 0.9805700057800898,
      "grad_norm": 0.057217229157686234,
      "learning_rate": 1.0152964499594575e-06,
      "loss": 1.4416,
      "step": 22054
    },
    {
      "epoch": 0.9806589302387622,
      "grad_norm": 0.05822160094976425,
      "learning_rate": 1.0060062851247053e-06,
      "loss": 1.4464,
      "step": 22056
    },
    {
      "epoch": 0.9807478546974345,
      "grad_norm": 0.057691022753715515,
      "learning_rate": 9.967587764955211e-07,
      "loss": 1.4492,
      "step": 22058
    },
    {
      "epoch": 0.9808367791561069,
      "grad_norm": 0.056380316615104675,
      "learning_rate": 9.875539248624388e-07,
      "loss": 1.4399,
      "step": 22060
    },
    {
      "epoch": 0.9809257036147793,
      "grad_norm": 0.058779843151569366,
      "learning_rate": 9.783917310122737e-07,
      "loss": 1.4478,
      "step": 22062
    },
    {
      "epoch": 0.9810146280734516,
      "grad_norm": 0.058194488286972046,
      "learning_rate": 9.692721957283435e-07,
      "loss": 1.4477,
      "step": 22064
    },
    {
      "epoch": 0.981103552532124,
      "grad_norm": 0.05809153616428375,
      "learning_rate": 9.601953197901913e-07,
      "loss": 1.4381,
      "step": 22066
    },
    {
      "epoch": 0.9811924769907964,
      "grad_norm": 0.05672544613480568,
      "learning_rate": 9.511611039737522e-07,
      "loss": 1.4442,
      "step": 22068
    },
    {
      "epoch": 0.9812814014494686,
      "grad_norm": 0.058622874319553375,
      "learning_rate": 9.421695490512416e-07,
      "loss": 1.4394,
      "step": 22070
    },
    {
      "epoch": 0.981370325908141,
      "grad_norm": 0.0575464628636837,
      "learning_rate": 9.332206557914336e-07,
      "loss": 1.447,
      "step": 22072
    },
    {
      "epoch": 0.9814592503668134,
      "grad_norm": 0.057317234575748444,
      "learning_rate": 9.243144249591606e-07,
      "loss": 1.4467,
      "step": 22074
    },
    {
      "epoch": 0.9815481748254857,
      "grad_norm": 0.05799569934606552,
      "learning_rate": 9.154508573158693e-07,
      "loss": 1.4454,
      "step": 22076
    },
    {
      "epoch": 0.9816370992841581,
      "grad_norm": 0.05822969973087311,
      "learning_rate": 9.066299536192313e-07,
      "loss": 1.4448,
      "step": 22078
    },
    {
      "epoch": 0.9817260237428305,
      "grad_norm": 0.058205485343933105,
      "learning_rate": 8.9785171462331e-07,
      "loss": 1.455,
      "step": 22080
    },
    {
      "epoch": 0.9818149482015028,
      "grad_norm": 0.057623036205768585,
      "learning_rate": 8.891161410785053e-07,
      "loss": 1.4465,
      "step": 22082
    },
    {
      "epoch": 0.9819038726601752,
      "grad_norm": 0.056971605867147446,
      "learning_rate": 8.804232337315532e-07,
      "loss": 1.4439,
      "step": 22084
    },
    {
      "epoch": 0.9819927971188476,
      "grad_norm": 0.056623686105012894,
      "learning_rate": 8.717729933255814e-07,
      "loss": 1.4419,
      "step": 22086
    },
    {
      "epoch": 0.9820817215775199,
      "grad_norm": 0.05784169211983681,
      "learning_rate": 8.631654206000539e-07,
      "loss": 1.4461,
      "step": 22088
    },
    {
      "epoch": 0.9821706460361923,
      "grad_norm": 0.05700985714793205,
      "learning_rate": 8.546005162907156e-07,
      "loss": 1.4379,
      "step": 22090
    },
    {
      "epoch": 0.9822595704948646,
      "grad_norm": 0.057645298540592194,
      "learning_rate": 8.46078281129814e-07,
      "loss": 1.4479,
      "step": 22092
    },
    {
      "epoch": 0.9823484949535369,
      "grad_norm": 0.057107895612716675,
      "learning_rate": 8.375987158458775e-07,
      "loss": 1.448,
      "step": 22094
    },
    {
      "epoch": 0.9824374194122093,
      "grad_norm": 0.05772538483142853,
      "learning_rate": 8.291618211637153e-07,
      "loss": 1.4449,
      "step": 22096
    },
    {
      "epoch": 0.9825263438708817,
      "grad_norm": 0.05685894191265106,
      "learning_rate": 8.207675978045281e-07,
      "loss": 1.4485,
      "step": 22098
    },
    {
      "epoch": 0.982615268329554,
      "grad_norm": 0.058153510093688965,
      "learning_rate": 8.124160464859642e-07,
      "loss": 1.4477,
      "step": 22100
    },
    {
      "epoch": 0.9827041927882264,
      "grad_norm": 0.057993143796920776,
      "learning_rate": 8.041071679219525e-07,
      "loss": 1.4428,
      "step": 22102
    },
    {
      "epoch": 0.9827931172468988,
      "grad_norm": 0.0569118969142437,
      "learning_rate": 7.95840962822758e-07,
      "loss": 1.4464,
      "step": 22104
    },
    {
      "epoch": 0.9828820417055711,
      "grad_norm": 0.057352546602487564,
      "learning_rate": 7.876174318949824e-07,
      "loss": 1.4476,
      "step": 22106
    },
    {
      "epoch": 0.9829709661642435,
      "grad_norm": 0.05711887776851654,
      "learning_rate": 7.794365758416188e-07,
      "loss": 1.443,
      "step": 22108
    },
    {
      "epoch": 0.9830598906229159,
      "grad_norm": 0.058069389313459396,
      "learning_rate": 7.712983953619967e-07,
      "loss": 1.4485,
      "step": 22110
    },
    {
      "epoch": 0.9831488150815882,
      "grad_norm": 0.05813687667250633,
      "learning_rate": 7.632028911518374e-07,
      "loss": 1.4526,
      "step": 22112
    },
    {
      "epoch": 0.9832377395402605,
      "grad_norm": 0.05850476771593094,
      "learning_rate": 7.551500639031427e-07,
      "loss": 1.4463,
      "step": 22114
    },
    {
      "epoch": 0.9833266639989329,
      "grad_norm": 0.0575152225792408,
      "learning_rate": 7.471399143043067e-07,
      "loss": 1.4549,
      "step": 22116
    },
    {
      "epoch": 0.9834155884576052,
      "grad_norm": 0.05797769874334335,
      "learning_rate": 7.391724430401148e-07,
      "loss": 1.449,
      "step": 22118
    },
    {
      "epoch": 0.9835045129162776,
      "grad_norm": 0.056744713336229324,
      "learning_rate": 7.312476507916332e-07,
      "loss": 1.4456,
      "step": 22120
    },
    {
      "epoch": 0.98359343737495,
      "grad_norm": 0.057250816375017166,
      "learning_rate": 7.233655382363202e-07,
      "loss": 1.4396,
      "step": 22122
    },
    {
      "epoch": 0.9836823618336223,
      "grad_norm": 0.05758194625377655,
      "learning_rate": 7.155261060479701e-07,
      "loss": 1.4521,
      "step": 22124
    },
    {
      "epoch": 0.9837712862922947,
      "grad_norm": 0.05834776535630226,
      "learning_rate": 7.077293548966579e-07,
      "loss": 1.4438,
      "step": 22126
    },
    {
      "epoch": 0.9838602107509671,
      "grad_norm": 0.0584811232984066,
      "learning_rate": 6.999752854490171e-07,
      "loss": 1.4453,
      "step": 22128
    },
    {
      "epoch": 0.9839491352096394,
      "grad_norm": 0.05721158906817436,
      "learning_rate": 6.922638983677954e-07,
      "loss": 1.4502,
      "step": 22130
    },
    {
      "epoch": 0.9840380596683118,
      "grad_norm": 0.05716519057750702,
      "learning_rate": 6.84595194312243e-07,
      "loss": 1.4425,
      "step": 22132
    },
    {
      "epoch": 0.9841269841269841,
      "grad_norm": 0.05842083320021629,
      "learning_rate": 6.769691739378913e-07,
      "loss": 1.4459,
      "step": 22134
    },
    {
      "epoch": 0.9842159085856564,
      "grad_norm": 0.05762701854109764,
      "learning_rate": 6.693858378967188e-07,
      "loss": 1.4475,
      "step": 22136
    },
    {
      "epoch": 0.9843048330443288,
      "grad_norm": 0.0569646880030632,
      "learning_rate": 6.618451868368736e-07,
      "loss": 1.4453,
      "step": 22138
    },
    {
      "epoch": 0.9843937575030012,
      "grad_norm": 0.056726690381765366,
      "learning_rate": 6.543472214030066e-07,
      "loss": 1.4471,
      "step": 22140
    },
    {
      "epoch": 0.9844826819616735,
      "grad_norm": 0.056243959814310074,
      "learning_rate": 6.468919422361052e-07,
      "loss": 1.4397,
      "step": 22142
    },
    {
      "epoch": 0.9845716064203459,
      "grad_norm": 0.058017753064632416,
      "learning_rate": 6.394793499734375e-07,
      "loss": 1.4452,
      "step": 22144
    },
    {
      "epoch": 0.9846605308790183,
      "grad_norm": 0.057117003947496414,
      "learning_rate": 6.321094452487186e-07,
      "loss": 1.4457,
      "step": 22146
    },
    {
      "epoch": 0.9847494553376906,
      "grad_norm": 0.059118859469890594,
      "learning_rate": 6.247822286918892e-07,
      "loss": 1.4479,
      "step": 22148
    },
    {
      "epoch": 0.984838379796363,
      "grad_norm": 0.05744896084070206,
      "learning_rate": 6.174977009293925e-07,
      "loss": 1.4502,
      "step": 22150
    },
    {
      "epoch": 0.9849273042550354,
      "grad_norm": 0.05781065300107002,
      "learning_rate": 6.102558625838417e-07,
      "loss": 1.4416,
      "step": 22152
    },
    {
      "epoch": 0.9850162287137078,
      "grad_norm": 0.057693760842084885,
      "learning_rate": 6.030567142744081e-07,
      "loss": 1.4452,
      "step": 22154
    },
    {
      "epoch": 0.98510515317238,
      "grad_norm": 0.05738704651594162,
      "learning_rate": 5.959002566164328e-07,
      "loss": 1.4488,
      "step": 22156
    },
    {
      "epoch": 0.9851940776310524,
      "grad_norm": 0.05740581080317497,
      "learning_rate": 5.887864902217044e-07,
      "loss": 1.4462,
      "step": 22158
    },
    {
      "epoch": 0.9852830020897247,
      "grad_norm": 0.057811297476291656,
      "learning_rate": 5.817154156983473e-07,
      "loss": 1.4463,
      "step": 22160
    },
    {
      "epoch": 0.9853719265483971,
      "grad_norm": 0.0571342296898365,
      "learning_rate": 5.746870336508225e-07,
      "loss": 1.4455,
      "step": 22162
    },
    {
      "epoch": 0.9854608510070695,
      "grad_norm": 0.05744681507349014,
      "learning_rate": 5.677013446799828e-07,
      "loss": 1.4499,
      "step": 22164
    },
    {
      "epoch": 0.9855497754657419,
      "grad_norm": 0.05767764896154404,
      "learning_rate": 5.60758349382906e-07,
      "loss": 1.4425,
      "step": 22166
    },
    {
      "epoch": 0.9856386999244142,
      "grad_norm": 0.05785730853676796,
      "learning_rate": 5.53858048353173e-07,
      "loss": 1.4449,
      "step": 22168
    },
    {
      "epoch": 0.9857276243830866,
      "grad_norm": 0.058226075023412704,
      "learning_rate": 5.470004421806452e-07,
      "loss": 1.4434,
      "step": 22170
    },
    {
      "epoch": 0.985816548841759,
      "grad_norm": 0.057991281151771545,
      "learning_rate": 5.401855314515758e-07,
      "loss": 1.445,
      "step": 22172
    },
    {
      "epoch": 0.9859054733004313,
      "grad_norm": 0.05784721300005913,
      "learning_rate": 5.334133167484434e-07,
      "loss": 1.448,
      "step": 22174
    },
    {
      "epoch": 0.9859943977591037,
      "grad_norm": 0.05827952176332474,
      "learning_rate": 5.266837986502294e-07,
      "loss": 1.4506,
      "step": 22176
    },
    {
      "epoch": 0.986083322217776,
      "grad_norm": 0.05748755857348442,
      "learning_rate": 5.199969777321955e-07,
      "loss": 1.4483,
      "step": 22178
    },
    {
      "epoch": 0.9861722466764483,
      "grad_norm": 0.057960059493780136,
      "learning_rate": 5.13352854565996e-07,
      "loss": 1.4474,
      "step": 22180
    },
    {
      "epoch": 0.9862611711351207,
      "grad_norm": 0.05742307007312775,
      "learning_rate": 5.067514297195098e-07,
      "loss": 1.4512,
      "step": 22182
    },
    {
      "epoch": 0.986350095593793,
      "grad_norm": 0.05780716612935066,
      "learning_rate": 5.001927037571186e-07,
      "loss": 1.4459,
      "step": 22184
    },
    {
      "epoch": 0.9864390200524654,
      "grad_norm": 0.0577070526778698,
      "learning_rate": 4.936766772394851e-07,
      "loss": 1.4409,
      "step": 22186
    },
    {
      "epoch": 0.9865279445111378,
      "grad_norm": 0.057526715099811554,
      "learning_rate": 4.872033507236084e-07,
      "loss": 1.4471,
      "step": 22188
    },
    {
      "epoch": 0.9866168689698102,
      "grad_norm": 0.05801134556531906,
      "learning_rate": 4.80772724762879e-07,
      "loss": 1.4386,
      "step": 22190
    },
    {
      "epoch": 0.9867057934284825,
      "grad_norm": 0.05754433944821358,
      "learning_rate": 4.743847999070239e-07,
      "loss": 1.4458,
      "step": 22192
    },
    {
      "epoch": 0.9867947178871549,
      "grad_norm": 0.05766000971198082,
      "learning_rate": 4.680395767021062e-07,
      "loss": 1.4399,
      "step": 22194
    },
    {
      "epoch": 0.9868836423458273,
      "grad_norm": 0.0569644533097744,
      "learning_rate": 4.617370556904699e-07,
      "loss": 1.4417,
      "step": 22196
    },
    {
      "epoch": 0.9869725668044996,
      "grad_norm": 0.05827400088310242,
      "learning_rate": 4.554772374110172e-07,
      "loss": 1.4557,
      "step": 22198
    },
    {
      "epoch": 0.9870614912631719,
      "grad_norm": 0.05730379745364189,
      "learning_rate": 4.4926012239870916e-07,
      "loss": 1.4435,
      "step": 22200
    },
    {
      "epoch": 0.9871504157218443,
      "grad_norm": 0.05722169205546379,
      "learning_rate": 4.4308571118517605e-07,
      "loss": 1.4427,
      "step": 22202
    },
    {
      "epoch": 0.9872393401805166,
      "grad_norm": 0.05816062167286873,
      "learning_rate": 4.369540042981068e-07,
      "loss": 1.4448,
      "step": 22204
    },
    {
      "epoch": 0.987328264639189,
      "grad_norm": 0.05683821812272072,
      "learning_rate": 4.3086500226169334e-07,
      "loss": 1.445,
      "step": 22206
    },
    {
      "epoch": 0.9874171890978614,
      "grad_norm": 0.05709170922636986,
      "learning_rate": 4.248187055965191e-07,
      "loss": 1.4453,
      "step": 22208
    },
    {
      "epoch": 0.9875061135565337,
      "grad_norm": 0.05717954784631729,
      "learning_rate": 4.1881511481939304e-07,
      "loss": 1.4421,
      "step": 22210
    },
    {
      "epoch": 0.9875950380152061,
      "grad_norm": 0.05784517154097557,
      "learning_rate": 4.1285423044351565e-07,
      "loss": 1.4496,
      "step": 22212
    },
    {
      "epoch": 0.9876839624738785,
      "grad_norm": 0.057573914527893066,
      "learning_rate": 4.0693605297842384e-07,
      "loss": 1.4458,
      "step": 22214
    },
    {
      "epoch": 0.9877728869325508,
      "grad_norm": 0.056905996054410934,
      "learning_rate": 4.0106058293015723e-07,
      "loss": 1.4495,
      "step": 22216
    },
    {
      "epoch": 0.9878618113912232,
      "grad_norm": 0.05849708244204521,
      "learning_rate": 3.952278208008697e-07,
      "loss": 1.4523,
      "step": 22218
    },
    {
      "epoch": 0.9879507358498956,
      "grad_norm": 0.0573749914765358,
      "learning_rate": 3.8943776708916247e-07,
      "loss": 1.4409,
      "step": 22220
    },
    {
      "epoch": 0.9880396603085678,
      "grad_norm": 0.05719359219074249,
      "learning_rate": 3.836904222900284e-07,
      "loss": 1.4437,
      "step": 22222
    },
    {
      "epoch": 0.9881285847672402,
      "grad_norm": 0.058678288012742996,
      "learning_rate": 3.779857868947967e-07,
      "loss": 1.4416,
      "step": 22224
    },
    {
      "epoch": 0.9882175092259126,
      "grad_norm": 0.058080919086933136,
      "learning_rate": 3.723238613910773e-07,
      "loss": 1.4414,
      "step": 22226
    },
    {
      "epoch": 0.9883064336845849,
      "grad_norm": 0.05763470381498337,
      "learning_rate": 3.6670464626292754e-07,
      "loss": 1.4479,
      "step": 22228
    },
    {
      "epoch": 0.9883953581432573,
      "grad_norm": 0.0569768100976944,
      "learning_rate": 3.611281419906853e-07,
      "loss": 1.4533,
      "step": 22230
    },
    {
      "epoch": 0.9884842826019297,
      "grad_norm": 0.05751827359199524,
      "learning_rate": 3.5559434905102497e-07,
      "loss": 1.4468,
      "step": 22232
    },
    {
      "epoch": 0.988573207060602,
      "grad_norm": 0.057300738990306854,
      "learning_rate": 3.501032679170124e-07,
      "loss": 1.4462,
      "step": 22234
    },
    {
      "epoch": 0.9886621315192744,
      "grad_norm": 0.057894255965948105,
      "learning_rate": 3.4465489905810555e-07,
      "loss": 1.444,
      "step": 22236
    },
    {
      "epoch": 0.9887510559779468,
      "grad_norm": 0.05746915191411972,
      "learning_rate": 3.392492429399874e-07,
      "loss": 1.4461,
      "step": 22238
    },
    {
      "epoch": 0.9888399804366191,
      "grad_norm": 0.05837760120630264,
      "learning_rate": 3.3388630002473273e-07,
      "loss": 1.4482,
      "step": 22240
    },
    {
      "epoch": 0.9889289048952915,
      "grad_norm": 0.05805863440036774,
      "learning_rate": 3.2856607077086374e-07,
      "loss": 1.4444,
      "step": 22242
    },
    {
      "epoch": 0.9890178293539638,
      "grad_norm": 0.056360069662332535,
      "learning_rate": 3.2328855563318326e-07,
      "loss": 1.4449,
      "step": 22244
    },
    {
      "epoch": 0.9891067538126361,
      "grad_norm": 0.05757332220673561,
      "learning_rate": 3.180537550627749e-07,
      "loss": 1.4413,
      "step": 22246
    },
    {
      "epoch": 0.9891956782713085,
      "grad_norm": 0.05767189711332321,
      "learning_rate": 3.1286166950711405e-07,
      "loss": 1.4476,
      "step": 22248
    },
    {
      "epoch": 0.9892846027299809,
      "grad_norm": 0.058686595410108566,
      "learning_rate": 3.0771229941012337e-07,
      "loss": 1.446,
      "step": 22250
    },
    {
      "epoch": 0.9893735271886532,
      "grad_norm": 0.05774107575416565,
      "learning_rate": 3.026056452119508e-07,
      "loss": 1.4448,
      "step": 22252
    },
    {
      "epoch": 0.9894624516473256,
      "grad_norm": 0.05755819380283356,
      "learning_rate": 2.975417073491915e-07,
      "loss": 1.4493,
      "step": 22254
    },
    {
      "epoch": 0.989551376105998,
      "grad_norm": 0.057851407676935196,
      "learning_rate": 2.9252048625461047e-07,
      "loss": 1.4421,
      "step": 22256
    },
    {
      "epoch": 0.9896403005646703,
      "grad_norm": 0.05904214829206467,
      "learning_rate": 2.8754198235758643e-07,
      "loss": 1.4491,
      "step": 22258
    },
    {
      "epoch": 0.9897292250233427,
      "grad_norm": 0.0583941750228405,
      "learning_rate": 2.826061960836124e-07,
      "loss": 1.4429,
      "step": 22260
    },
    {
      "epoch": 0.9898181494820151,
      "grad_norm": 0.05874110758304596,
      "learning_rate": 2.7771312785462853e-07,
      "loss": 1.4454,
      "step": 22262
    },
    {
      "epoch": 0.9899070739406873,
      "grad_norm": 0.05823364108800888,
      "learning_rate": 2.7286277808891145e-07,
      "loss": 1.45,
      "step": 22264
    },
    {
      "epoch": 0.9899959983993597,
      "grad_norm": 0.056944265961647034,
      "learning_rate": 2.6805514720112946e-07,
      "loss": 1.4472,
      "step": 22266
    },
    {
      "epoch": 0.9900849228580321,
      "grad_norm": 0.05789085850119591,
      "learning_rate": 2.632902356022315e-07,
      "loss": 1.4496,
      "step": 22268
    },
    {
      "epoch": 0.9901738473167044,
      "grad_norm": 0.058085713535547256,
      "learning_rate": 2.5856804369955854e-07,
      "loss": 1.4492,
      "step": 22270
    },
    {
      "epoch": 0.9902627717753768,
      "grad_norm": 0.05807514116168022,
      "learning_rate": 2.538885718967876e-07,
      "loss": 1.4439,
      "step": 22272
    },
    {
      "epoch": 0.9903516962340492,
      "grad_norm": 0.057060327380895615,
      "learning_rate": 2.49251820593932e-07,
      "loss": 1.4408,
      "step": 22274
    },
    {
      "epoch": 0.9904406206927215,
      "grad_norm": 0.057622525840997696,
      "learning_rate": 2.44657790187397e-07,
      "loss": 1.4477,
      "step": 22276
    },
    {
      "epoch": 0.9905295451513939,
      "grad_norm": 0.056390151381492615,
      "learning_rate": 2.401064810698128e-07,
      "loss": 1.4408,
      "step": 22278
    },
    {
      "epoch": 0.9906184696100663,
      "grad_norm": 0.05661400407552719,
      "learning_rate": 2.355978936303127e-07,
      "loss": 1.4452,
      "step": 22280
    },
    {
      "epoch": 0.9907073940687386,
      "grad_norm": 0.057079821825027466,
      "learning_rate": 2.3113202825425505e-07,
      "loss": 1.4395,
      "step": 22282
    },
    {
      "epoch": 0.990796318527411,
      "grad_norm": 0.05696665868163109,
      "learning_rate": 2.267088853235011e-07,
      "loss": 1.447,
      "step": 22284
    },
    {
      "epoch": 0.9908852429860833,
      "grad_norm": 0.056844066828489304,
      "learning_rate": 2.2232846521608175e-07,
      "loss": 1.4413,
      "step": 22286
    },
    {
      "epoch": 0.9909741674447556,
      "grad_norm": 0.05780512094497681,
      "learning_rate": 2.1799076830647525e-07,
      "loss": 1.4417,
      "step": 22288
    },
    {
      "epoch": 0.991063091903428,
      "grad_norm": 0.05735721439123154,
      "learning_rate": 2.1369579496549607e-07,
      "loss": 1.4478,
      "step": 22290
    },
    {
      "epoch": 0.9911520163621004,
      "grad_norm": 0.05718429014086723,
      "learning_rate": 2.0944354556023947e-07,
      "loss": 1.4469,
      "step": 22292
    },
    {
      "epoch": 0.9912409408207727,
      "grad_norm": 0.05797044187784195,
      "learning_rate": 2.0523402045430352e-07,
      "loss": 1.4446,
      "step": 22294
    },
    {
      "epoch": 0.9913298652794451,
      "grad_norm": 0.05726039782166481,
      "learning_rate": 2.0106722000745592e-07,
      "loss": 1.4418,
      "step": 22296
    },
    {
      "epoch": 0.9914187897381175,
      "grad_norm": 0.05672242492437363,
      "learning_rate": 1.969431445759673e-07,
      "loss": 1.4446,
      "step": 22298
    },
    {
      "epoch": 0.9915077141967898,
      "grad_norm": 0.057257115840911865,
      "learning_rate": 1.9286179451227792e-07,
      "loss": 1.448,
      "step": 22300
    },
    {
      "epoch": 0.9915966386554622,
      "grad_norm": 0.05845504254102707,
      "learning_rate": 1.8882317016538642e-07,
      "loss": 1.4445,
      "step": 22302
    },
    {
      "epoch": 0.9916855631141346,
      "grad_norm": 0.058910876512527466,
      "learning_rate": 1.8482727188046112e-07,
      "loss": 1.4505,
      "step": 22304
    },
    {
      "epoch": 0.991774487572807,
      "grad_norm": 0.05711057409644127,
      "learning_rate": 1.8087409999911763e-07,
      "loss": 1.4427,
      "step": 22306
    },
    {
      "epoch": 0.9918634120314792,
      "grad_norm": 0.05767577514052391,
      "learning_rate": 1.7696365485930787e-07,
      "loss": 1.4478,
      "step": 22308
    },
    {
      "epoch": 0.9919523364901516,
      "grad_norm": 0.057249389588832855,
      "learning_rate": 1.7309593679526447e-07,
      "loss": 1.4401,
      "step": 22310
    },
    {
      "epoch": 0.9920412609488239,
      "grad_norm": 0.05789843201637268,
      "learning_rate": 1.6927094613766736e-07,
      "loss": 1.4432,
      "step": 22312
    },
    {
      "epoch": 0.9921301854074963,
      "grad_norm": 0.05697624757885933,
      "learning_rate": 1.654886832134772e-07,
      "loss": 1.4426,
      "step": 22314
    },
    {
      "epoch": 0.9922191098661687,
      "grad_norm": 0.058805204927921295,
      "learning_rate": 1.6174914834599097e-07,
      "loss": 1.4463,
      "step": 22316
    },
    {
      "epoch": 0.992308034324841,
      "grad_norm": 0.05726942420005798,
      "learning_rate": 1.5805234185495287e-07,
      "loss": 1.4472,
      "step": 22318
    },
    {
      "epoch": 0.9923969587835134,
      "grad_norm": 0.05799218639731407,
      "learning_rate": 1.5439826405627688e-07,
      "loss": 1.4422,
      "step": 22320
    },
    {
      "epoch": 0.9924858832421858,
      "grad_norm": 0.05797122046351433,
      "learning_rate": 1.5078691526243526e-07,
      "loss": 1.4429,
      "step": 22322
    },
    {
      "epoch": 0.9925748077008582,
      "grad_norm": 0.057014547288417816,
      "learning_rate": 1.4721829578207002e-07,
      "loss": 1.4429,
      "step": 22324
    },
    {
      "epoch": 0.9926637321595305,
      "grad_norm": 0.05727580189704895,
      "learning_rate": 1.4369240592021493e-07,
      "loss": 1.4485,
      "step": 22326
    },
    {
      "epoch": 0.9927526566182029,
      "grad_norm": 0.05705159902572632,
      "learning_rate": 1.4020924597840656e-07,
      "loss": 1.4497,
      "step": 22328
    },
    {
      "epoch": 0.9928415810768751,
      "grad_norm": 0.05742578208446503,
      "learning_rate": 1.3676881625424021e-07,
      "loss": 1.4469,
      "step": 22330
    },
    {
      "epoch": 0.9929305055355475,
      "grad_norm": 0.058664653450250626,
      "learning_rate": 1.3337111704198047e-07,
      "loss": 1.4475,
      "step": 22332
    },
    {
      "epoch": 0.9930194299942199,
      "grad_norm": 0.05729004368185997,
      "learning_rate": 1.3001614863195066e-07,
      "loss": 1.4456,
      "step": 22334
    },
    {
      "epoch": 0.9931083544528923,
      "grad_norm": 0.05801570415496826,
      "learning_rate": 1.2670391131103242e-07,
      "loss": 1.4544,
      "step": 22336
    },
    {
      "epoch": 0.9931972789115646,
      "grad_norm": 0.05761876702308655,
      "learning_rate": 1.2343440536227714e-07,
      "loss": 1.4461,
      "step": 22338
    },
    {
      "epoch": 0.993286203370237,
      "grad_norm": 0.05685674026608467,
      "learning_rate": 1.2020763106529443e-07,
      "loss": 1.4403,
      "step": 22340
    },
    {
      "epoch": 0.9933751278289094,
      "grad_norm": 0.05733815208077431,
      "learning_rate": 1.170235886958082e-07,
      "loss": 1.4474,
      "step": 22342
    },
    {
      "epoch": 0.9934640522875817,
      "grad_norm": 0.057545486837625504,
      "learning_rate": 1.1388227852610066e-07,
      "loss": 1.445,
      "step": 22344
    },
    {
      "epoch": 0.9935529767462541,
      "grad_norm": 0.0574163943529129,
      "learning_rate": 1.1078370082467926e-07,
      "loss": 1.4422,
      "step": 22346
    },
    {
      "epoch": 0.9936419012049265,
      "grad_norm": 0.05813299119472504,
      "learning_rate": 1.0772785585633215e-07,
      "loss": 1.4507,
      "step": 22348
    },
    {
      "epoch": 0.9937308256635988,
      "grad_norm": 0.05734492093324661,
      "learning_rate": 1.0471474388240587e-07,
      "loss": 1.4422,
      "step": 22350
    },
    {
      "epoch": 0.9938197501222711,
      "grad_norm": 0.05803303048014641,
      "learning_rate": 1.0174436516047214e-07,
      "loss": 1.4482,
      "step": 22352
    },
    {
      "epoch": 0.9939086745809435,
      "grad_norm": 0.058048054575920105,
      "learning_rate": 9.881671994432795e-08,
      "loss": 1.4479,
      "step": 22354
    },
    {
      "epoch": 0.9939975990396158,
      "grad_norm": 0.057730190455913544,
      "learning_rate": 9.59318084843841e-08,
      "loss": 1.4538,
      "step": 22356
    },
    {
      "epoch": 0.9940865234982882,
      "grad_norm": 0.05775808170437813,
      "learning_rate": 9.308963102716561e-08,
      "loss": 1.442,
      "step": 22358
    },
    {
      "epoch": 0.9941754479569606,
      "grad_norm": 0.05732406675815582,
      "learning_rate": 9.029018781570031e-08,
      "loss": 1.447,
      "step": 22360
    },
    {
      "epoch": 0.9942643724156329,
      "grad_norm": 0.05907835438847542,
      "learning_rate": 8.753347908924126e-08,
      "loss": 1.4477,
      "step": 22362
    },
    {
      "epoch": 0.9943532968743053,
      "grad_norm": 0.05664451792836189,
      "learning_rate": 8.481950508343328e-08,
      "loss": 1.4409,
      "step": 22364
    },
    {
      "epoch": 0.9944422213329777,
      "grad_norm": 0.05807731673121452,
      "learning_rate": 8.214826603031301e-08,
      "loss": 1.4443,
      "step": 22366
    },
    {
      "epoch": 0.99453114579165,
      "grad_norm": 0.0581711083650589,
      "learning_rate": 7.951976215825329e-08,
      "loss": 1.4511,
      "step": 22368
    },
    {
      "epoch": 0.9946200702503224,
      "grad_norm": 0.057496510446071625,
      "learning_rate": 7.693399369190779e-08,
      "loss": 1.4406,
      "step": 22370
    },
    {
      "epoch": 0.9947089947089947,
      "grad_norm": 0.05667787045240402,
      "learning_rate": 7.43909608523774e-08,
      "loss": 1.4445,
      "step": 22372
    },
    {
      "epoch": 0.994797919167667,
      "grad_norm": 0.05935773253440857,
      "learning_rate": 7.189066385693276e-08,
      "loss": 1.4474,
      "step": 22374
    },
    {
      "epoch": 0.9948868436263394,
      "grad_norm": 0.05779668688774109,
      "learning_rate": 6.943310291945836e-08,
      "loss": 1.4459,
      "step": 22376
    },
    {
      "epoch": 0.9949757680850118,
      "grad_norm": 0.056512489914894104,
      "learning_rate": 6.701827824989737e-08,
      "loss": 1.4443,
      "step": 22378
    },
    {
      "epoch": 0.9950646925436841,
      "grad_norm": 0.058668848127126694,
      "learning_rate": 6.464619005480676e-08,
      "loss": 1.4499,
      "step": 22380
    },
    {
      "epoch": 0.9951536170023565,
      "grad_norm": 0.05988642945885658,
      "learning_rate": 6.231683853685776e-08,
      "loss": 1.4459,
      "step": 22382
    },
    {
      "epoch": 0.9952425414610289,
      "grad_norm": 0.057425398379564285,
      "learning_rate": 6.003022389522439e-08,
      "loss": 1.4456,
      "step": 22384
    },
    {
      "epoch": 0.9953314659197012,
      "grad_norm": 0.05751774087548256,
      "learning_rate": 5.778634632536139e-08,
      "loss": 1.4431,
      "step": 22386
    },
    {
      "epoch": 0.9954203903783736,
      "grad_norm": 0.05865611135959625,
      "learning_rate": 5.558520601917083e-08,
      "loss": 1.4444,
      "step": 22388
    },
    {
      "epoch": 0.995509314837046,
      "grad_norm": 0.05733845382928848,
      "learning_rate": 5.342680316466897e-08,
      "loss": 1.4468,
      "step": 22390
    },
    {
      "epoch": 0.9955982392957183,
      "grad_norm": 0.059069231152534485,
      "learning_rate": 5.131113794643039e-08,
      "loss": 1.4488,
      "step": 22392
    },
    {
      "epoch": 0.9956871637543906,
      "grad_norm": 0.05756256729364395,
      "learning_rate": 4.923821054536592e-08,
      "loss": 1.4397,
      "step": 22394
    },
    {
      "epoch": 0.995776088213063,
      "grad_norm": 0.057630039751529694,
      "learning_rate": 4.7208021138611666e-08,
      "loss": 1.4437,
      "step": 22396
    },
    {
      "epoch": 0.9958650126717353,
      "grad_norm": 0.057779572904109955,
      "learning_rate": 4.5220569899750984e-08,
      "loss": 1.4464,
      "step": 22398
    },
    {
      "epoch": 0.9959539371304077,
      "grad_norm": 0.05961634963750839,
      "learning_rate": 4.327585699859249e-08,
      "loss": 1.4437,
      "step": 22400
    },
    {
      "epoch": 0.9960428615890801,
      "grad_norm": 0.05674600601196289,
      "learning_rate": 4.137388260155861e-08,
      "loss": 1.4464,
      "step": 22402
    },
    {
      "epoch": 0.9961317860477524,
      "grad_norm": 0.056886181235313416,
      "learning_rate": 3.951464687107498e-08,
      "loss": 1.445,
      "step": 22404
    },
    {
      "epoch": 0.9962207105064248,
      "grad_norm": 0.0578632578253746,
      "learning_rate": 3.769814996612553e-08,
      "loss": 1.4484,
      "step": 22406
    },
    {
      "epoch": 0.9963096349650972,
      "grad_norm": 0.057221461087465286,
      "learning_rate": 3.592439204197495e-08,
      "loss": 1.445,
      "step": 22408
    },
    {
      "epoch": 0.9963985594237695,
      "grad_norm": 0.058185044676065445,
      "learning_rate": 3.4193373250335224e-08,
      "loss": 1.4412,
      "step": 22410
    },
    {
      "epoch": 0.9964874838824419,
      "grad_norm": 0.05843387171626091,
      "learning_rate": 3.250509373908805e-08,
      "loss": 1.4423,
      "step": 22412
    },
    {
      "epoch": 0.9965764083411143,
      "grad_norm": 0.05737845599651337,
      "learning_rate": 3.085955365261794e-08,
      "loss": 1.4465,
      "step": 22414
    },
    {
      "epoch": 0.9966653327997865,
      "grad_norm": 0.05840914696455002,
      "learning_rate": 2.925675313159015e-08,
      "loss": 1.4438,
      "step": 22416
    },
    {
      "epoch": 0.9967542572584589,
      "grad_norm": 0.05688779801130295,
      "learning_rate": 2.769669231295069e-08,
      "loss": 1.4438,
      "step": 22418
    },
    {
      "epoch": 0.9968431817171313,
      "grad_norm": 0.05774452164769173,
      "learning_rate": 2.617937133009285e-08,
      "loss": 1.4466,
      "step": 22420
    },
    {
      "epoch": 0.9969321061758036,
      "grad_norm": 0.05797778069972992,
      "learning_rate": 2.4704790312746194e-08,
      "loss": 1.4443,
      "step": 22422
    },
    {
      "epoch": 0.997021030634476,
      "grad_norm": 0.057229943573474884,
      "learning_rate": 2.327294938697655e-08,
      "loss": 1.4496,
      "step": 22424
    },
    {
      "epoch": 0.9971099550931484,
      "grad_norm": 0.05771101638674736,
      "learning_rate": 2.188384867513049e-08,
      "loss": 1.4412,
      "step": 22426
    },
    {
      "epoch": 0.9971988795518207,
      "grad_norm": 0.057632721960544586,
      "learning_rate": 2.053748829600188e-08,
      "loss": 1.4515,
      "step": 22428
    },
    {
      "epoch": 0.9972878040104931,
      "grad_norm": 0.05861422419548035,
      "learning_rate": 1.9233868364665342e-08,
      "loss": 1.4463,
      "step": 22430
    },
    {
      "epoch": 0.9973767284691655,
      "grad_norm": 0.05753085017204285,
      "learning_rate": 1.797298899258726e-08,
      "loss": 1.4453,
      "step": 22432
    },
    {
      "epoch": 0.9974656529278378,
      "grad_norm": 0.057691846042871475,
      "learning_rate": 1.6754850287459267e-08,
      "loss": 1.4435,
      "step": 22434
    },
    {
      "epoch": 0.9975545773865102,
      "grad_norm": 0.05841954052448273,
      "learning_rate": 1.5579452353531308e-08,
      "loss": 1.4526,
      "step": 22436
    },
    {
      "epoch": 0.9976435018451825,
      "grad_norm": 0.05724509432911873,
      "learning_rate": 1.4446795291223059e-08,
      "loss": 1.4428,
      "step": 22438
    },
    {
      "epoch": 0.9977324263038548,
      "grad_norm": 0.05789647623896599,
      "learning_rate": 1.3356879197401473e-08,
      "loss": 1.445,
      "step": 22440
    },
    {
      "epoch": 0.9978213507625272,
      "grad_norm": 0.058276426047086716,
      "learning_rate": 1.2309704165158753e-08,
      "loss": 1.4454,
      "step": 22442
    },
    {
      "epoch": 0.9979102752211996,
      "grad_norm": 0.057980820536613464,
      "learning_rate": 1.1305270284089897e-08,
      "loss": 1.4448,
      "step": 22444
    },
    {
      "epoch": 0.9979991996798719,
      "grad_norm": 0.05673612281680107,
      "learning_rate": 1.0343577639959634e-08,
      "loss": 1.446,
      "step": 22446
    },
    {
      "epoch": 0.9980881241385443,
      "grad_norm": 0.05783254653215408,
      "learning_rate": 9.424626315091e-09,
      "loss": 1.4437,
      "step": 22448
    },
    {
      "epoch": 0.9981770485972167,
      "grad_norm": 0.058315496891736984,
      "learning_rate": 8.548416388032277e-09,
      "loss": 1.4446,
      "step": 22450
    },
    {
      "epoch": 0.998265973055889,
      "grad_norm": 0.05717792361974716,
      "learning_rate": 7.714947933556982e-09,
      "loss": 1.4445,
      "step": 22452
    },
    {
      "epoch": 0.9983548975145614,
      "grad_norm": 0.05846971645951271,
      "learning_rate": 6.924221023052457e-09,
      "loss": 1.4477,
      "step": 22454
    },
    {
      "epoch": 0.9984438219732338,
      "grad_norm": 0.05766315013170242,
      "learning_rate": 6.176235724075774e-09,
      "loss": 1.4465,
      "step": 22456
    },
    {
      "epoch": 0.9985327464319061,
      "grad_norm": 0.058049630373716354,
      "learning_rate": 5.470992100520267e-09,
      "loss": 1.4475,
      "step": 22458
    },
    {
      "epoch": 0.9986216708905784,
      "grad_norm": 0.057937148958444595,
      "learning_rate": 4.808490212726557e-09,
      "loss": 1.4404,
      "step": 22460
    },
    {
      "epoch": 0.9987105953492508,
      "grad_norm": 0.05762643739581108,
      "learning_rate": 4.18873011731602e-09,
      "loss": 1.446,
      "step": 22462
    },
    {
      "epoch": 0.9987995198079231,
      "grad_norm": 0.05884932354092598,
      "learning_rate": 3.611711867246292e-09,
      "loss": 1.4505,
      "step": 22464
    },
    {
      "epoch": 0.9988884442665955,
      "grad_norm": 0.05755854770541191,
      "learning_rate": 3.0774355119222996e-09,
      "loss": 1.4517,
      "step": 22466
    },
    {
      "epoch": 0.9989773687252679,
      "grad_norm": 0.05807171016931534,
      "learning_rate": 2.585901096863186e-09,
      "loss": 1.4417,
      "step": 22468
    },
    {
      "epoch": 0.9990662931839402,
      "grad_norm": 0.057942621409893036,
      "learning_rate": 2.1371086642574255e-09,
      "loss": 1.4481,
      "step": 22470
    },
    {
      "epoch": 0.9991552176426126,
      "grad_norm": 0.05796192213892937,
      "learning_rate": 1.7310582523522023e-09,
      "loss": 1.4428,
      "step": 22472
    },
    {
      "epoch": 0.999244142101285,
      "grad_norm": 0.05767205357551575,
      "learning_rate": 1.3677498958974966e-09,
      "loss": 1.448,
      "step": 22474
    },
    {
      "epoch": 0.9993330665599574,
      "grad_norm": 0.05703622102737427,
      "learning_rate": 1.047183625924042e-09,
      "loss": 1.4464,
      "step": 22476
    },
    {
      "epoch": 0.9994219910186297,
      "grad_norm": 0.05870012938976288,
      "learning_rate": 7.693594699098582e-10,
      "loss": 1.4521,
      "step": 22478
    },
    {
      "epoch": 0.9995109154773021,
      "grad_norm": 0.057245586067438126,
      "learning_rate": 5.342774515026961e-10,
      "loss": 1.4448,
      "step": 22480
    },
    {
      "epoch": 0.9995998399359743,
      "grad_norm": 0.05662744492292404,
      "learning_rate": 3.419375909086142e-10,
      "loss": 1.4334,
      "step": 22482
    },
    {
      "epoch": 0.9996887643946467,
      "grad_norm": 0.05766800418496132,
      "learning_rate": 1.923399044478913e-10,
      "loss": 1.4482,
      "step": 22484
    },
    {
      "epoch": 0.9997776888533191,
      "grad_norm": 0.058209143579006195,
      "learning_rate": 8.548440505462551e-11,
      "loss": 1.4435,
      "step": 22486
    },
    {
      "epoch": 0.9998666133119914,
      "grad_norm": 0.05729079991579056,
      "learning_rate": 2.1371101721623376e-11,
      "loss": 1.4459,
      "step": 22488
    },
    {
      "epoch": 0.9999555377706638,
      "grad_norm": 0.057013168931007385,
      "learning_rate": 0.0,
      "loss": 1.4473,
      "step": 22490
    }
  ],
  "logging_steps": 2,
  "max_steps": 22491,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.812454399322358e+19,
  "train_batch_size": 768,
  "trial_name": null,
  "trial_params": null
}