{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.993178717598909,
  "eval_steps": 500,
  "global_step": 915,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.005457025920873124,
      "grad_norm": 8.4054853913818,
      "learning_rate": 8.695652173913044e-07,
      "loss": 1.4358,
      "step": 1
    },
    {
      "epoch": 0.010914051841746248,
      "grad_norm": 8.27769761646866,
      "learning_rate": 1.7391304347826088e-06,
      "loss": 1.4226,
      "step": 2
    },
    {
      "epoch": 0.01637107776261937,
      "grad_norm": 8.322721166007087,
      "learning_rate": 2.6086956521739132e-06,
      "loss": 1.4269,
      "step": 3
    },
    {
      "epoch": 0.021828103683492497,
      "grad_norm": 7.801471051907299,
      "learning_rate": 3.4782608695652175e-06,
      "loss": 1.4294,
      "step": 4
    },
    {
      "epoch": 0.027285129604365622,
      "grad_norm": 6.35431414486227,
      "learning_rate": 4.347826086956522e-06,
      "loss": 1.3629,
      "step": 5
    },
    {
      "epoch": 0.03274215552523874,
      "grad_norm": 3.4381748863553043,
      "learning_rate": 5.2173913043478265e-06,
      "loss": 1.3331,
      "step": 6
    },
    {
      "epoch": 0.03819918144611187,
      "grad_norm": 2.674477084182652,
      "learning_rate": 6.086956521739132e-06,
      "loss": 1.3014,
      "step": 7
    },
    {
      "epoch": 0.04365620736698499,
      "grad_norm": 6.6546035127733765,
      "learning_rate": 6.956521739130435e-06,
      "loss": 1.3118,
      "step": 8
    },
    {
      "epoch": 0.04911323328785812,
      "grad_norm": 6.905025069917795,
      "learning_rate": 7.82608695652174e-06,
      "loss": 1.3125,
      "step": 9
    },
    {
      "epoch": 0.054570259208731244,
      "grad_norm": 6.814662162479713,
      "learning_rate": 8.695652173913044e-06,
      "loss": 1.3129,
      "step": 10
    },
    {
      "epoch": 0.06002728512960437,
      "grad_norm": 6.340208147957234,
      "learning_rate": 9.565217391304349e-06,
      "loss": 1.2766,
      "step": 11
    },
    {
      "epoch": 0.06548431105047749,
      "grad_norm": 5.393237093948397,
      "learning_rate": 1.0434782608695653e-05,
      "loss": 1.2624,
      "step": 12
    },
    {
      "epoch": 0.07094133697135062,
      "grad_norm": 3.425534423837231,
      "learning_rate": 1.1304347826086957e-05,
      "loss": 1.2328,
      "step": 13
    },
    {
      "epoch": 0.07639836289222374,
      "grad_norm": 2.2647521966523274,
      "learning_rate": 1.2173913043478263e-05,
      "loss": 1.2134,
      "step": 14
    },
    {
      "epoch": 0.08185538881309687,
      "grad_norm": 2.2313353605470545,
      "learning_rate": 1.3043478260869566e-05,
      "loss": 1.1939,
      "step": 15
    },
    {
      "epoch": 0.08731241473396999,
      "grad_norm": 2.361898565307054,
      "learning_rate": 1.391304347826087e-05,
      "loss": 1.18,
      "step": 16
    },
    {
      "epoch": 0.0927694406548431,
      "grad_norm": 2.33791134918371,
      "learning_rate": 1.4782608695652174e-05,
      "loss": 1.1511,
      "step": 17
    },
    {
      "epoch": 0.09822646657571624,
      "grad_norm": 1.6798499615830327,
      "learning_rate": 1.565217391304348e-05,
      "loss": 1.1595,
      "step": 18
    },
    {
      "epoch": 0.10368349249658936,
      "grad_norm": 1.6970897953125714,
      "learning_rate": 1.6521739130434785e-05,
      "loss": 1.1375,
      "step": 19
    },
    {
      "epoch": 0.10914051841746249,
      "grad_norm": 1.6395457480905744,
      "learning_rate": 1.739130434782609e-05,
      "loss": 1.1345,
      "step": 20
    },
    {
      "epoch": 0.1145975443383356,
      "grad_norm": 1.5035713589688318,
      "learning_rate": 1.8260869565217393e-05,
      "loss": 1.1337,
      "step": 21
    },
    {
      "epoch": 0.12005457025920874,
      "grad_norm": 1.225373073362722,
      "learning_rate": 1.9130434782608697e-05,
      "loss": 1.1186,
      "step": 22
    },
    {
      "epoch": 0.12551159618008187,
      "grad_norm": 1.5356355968160549,
      "learning_rate": 2e-05,
      "loss": 1.1286,
      "step": 23
    },
    {
      "epoch": 0.13096862210095497,
      "grad_norm": 1.1025414498974069,
      "learning_rate": 2.0869565217391306e-05,
      "loss": 1.0892,
      "step": 24
    },
    {
      "epoch": 0.1364256480218281,
      "grad_norm": 1.3350107601336392,
      "learning_rate": 2.173913043478261e-05,
      "loss": 1.0761,
      "step": 25
    },
    {
      "epoch": 0.14188267394270124,
      "grad_norm": 1.1929582575416837,
      "learning_rate": 2.2608695652173914e-05,
      "loss": 1.0883,
      "step": 26
    },
    {
      "epoch": 0.14733969986357434,
      "grad_norm": 0.935495351823663,
      "learning_rate": 2.3478260869565222e-05,
      "loss": 1.0597,
      "step": 27
    },
    {
      "epoch": 0.15279672578444747,
      "grad_norm": 1.5455709498780492,
      "learning_rate": 2.4347826086956526e-05,
      "loss": 1.0595,
      "step": 28
    },
    {
      "epoch": 0.1582537517053206,
      "grad_norm": 1.1083733884089981,
      "learning_rate": 2.521739130434783e-05,
      "loss": 1.0629,
      "step": 29
    },
    {
      "epoch": 0.16371077762619374,
      "grad_norm": 1.7349180323187294,
      "learning_rate": 2.608695652173913e-05,
      "loss": 1.0855,
      "step": 30
    },
    {
      "epoch": 0.16916780354706684,
      "grad_norm": 1.0670211570486026,
      "learning_rate": 2.6956521739130436e-05,
      "loss": 1.0503,
      "step": 31
    },
    {
      "epoch": 0.17462482946793997,
      "grad_norm": 1.6288820867791205,
      "learning_rate": 2.782608695652174e-05,
      "loss": 1.0613,
      "step": 32
    },
    {
      "epoch": 0.1800818553888131,
      "grad_norm": 1.329624751880889,
      "learning_rate": 2.8695652173913044e-05,
      "loss": 1.0592,
      "step": 33
    },
    {
      "epoch": 0.1855388813096862,
      "grad_norm": 1.1836294011167645,
      "learning_rate": 2.956521739130435e-05,
      "loss": 1.0546,
      "step": 34
    },
    {
      "epoch": 0.19099590723055934,
      "grad_norm": 1.4068311305836099,
      "learning_rate": 3.0434782608695656e-05,
      "loss": 1.0647,
      "step": 35
    },
    {
      "epoch": 0.19645293315143247,
      "grad_norm": 1.3722669810117456,
      "learning_rate": 3.130434782608696e-05,
      "loss": 1.0565,
      "step": 36
    },
    {
      "epoch": 0.2019099590723056,
      "grad_norm": 2.5377362025037993,
      "learning_rate": 3.2173913043478265e-05,
      "loss": 1.0801,
      "step": 37
    },
    {
      "epoch": 0.2073669849931787,
      "grad_norm": 1.3833891330993688,
      "learning_rate": 3.304347826086957e-05,
      "loss": 1.058,
      "step": 38
    },
    {
      "epoch": 0.21282401091405184,
      "grad_norm": 3.3467609300165893,
      "learning_rate": 3.391304347826087e-05,
      "loss": 1.039,
      "step": 39
    },
    {
      "epoch": 0.21828103683492497,
      "grad_norm": 2.5867789286474343,
      "learning_rate": 3.478260869565218e-05,
      "loss": 1.0357,
      "step": 40
    },
    {
      "epoch": 0.22373806275579808,
      "grad_norm": 2.620000651253596,
      "learning_rate": 3.565217391304348e-05,
      "loss": 1.0629,
      "step": 41
    },
    {
      "epoch": 0.2291950886766712,
      "grad_norm": 2.4068260667323664,
      "learning_rate": 3.6521739130434786e-05,
      "loss": 1.0411,
      "step": 42
    },
    {
      "epoch": 0.23465211459754434,
      "grad_norm": 2.2613496975518754,
      "learning_rate": 3.739130434782609e-05,
      "loss": 1.0615,
      "step": 43
    },
    {
      "epoch": 0.24010914051841747,
      "grad_norm": 2.4302337424475295,
      "learning_rate": 3.8260869565217395e-05,
      "loss": 1.042,
      "step": 44
    },
    {
      "epoch": 0.24556616643929058,
      "grad_norm": 2.0626105805304604,
      "learning_rate": 3.91304347826087e-05,
      "loss": 1.007,
      "step": 45
    },
    {
      "epoch": 0.25102319236016374,
      "grad_norm": 2.0977725511989327,
      "learning_rate": 4e-05,
      "loss": 1.03,
      "step": 46
    },
    {
      "epoch": 0.25648021828103684,
      "grad_norm": 2.1512078484152757,
      "learning_rate": 4.0869565217391314e-05,
      "loss": 1.0346,
      "step": 47
    },
    {
      "epoch": 0.26193724420190995,
      "grad_norm": 1.5000259392405852,
      "learning_rate": 4.173913043478261e-05,
      "loss": 1.0324,
      "step": 48
    },
    {
      "epoch": 0.2673942701227831,
      "grad_norm": 1.6825832607122888,
      "learning_rate": 4.2608695652173916e-05,
      "loss": 1.0265,
      "step": 49
    },
    {
      "epoch": 0.2728512960436562,
      "grad_norm": 2.3652304527471,
      "learning_rate": 4.347826086956522e-05,
      "loss": 1.0355,
      "step": 50
    },
    {
      "epoch": 0.2783083219645293,
      "grad_norm": 1.4430495759345763,
      "learning_rate": 4.4347826086956525e-05,
      "loss": 1.0204,
      "step": 51
    },
    {
      "epoch": 0.2837653478854025,
      "grad_norm": 1.9471116948950629,
      "learning_rate": 4.521739130434783e-05,
      "loss": 1.0233,
      "step": 52
    },
    {
      "epoch": 0.2892223738062756,
      "grad_norm": 0.9851562319686198,
      "learning_rate": 4.608695652173913e-05,
      "loss": 1.0224,
      "step": 53
    },
    {
      "epoch": 0.2946793997271487,
      "grad_norm": 2.289997829572797,
      "learning_rate": 4.6956521739130444e-05,
      "loss": 1.0494,
      "step": 54
    },
    {
      "epoch": 0.30013642564802184,
      "grad_norm": 1.9221832471449953,
      "learning_rate": 4.782608695652174e-05,
      "loss": 1.0078,
      "step": 55
    },
    {
      "epoch": 0.30559345156889495,
      "grad_norm": 2.0382214762471964,
      "learning_rate": 4.869565217391305e-05,
      "loss": 1.0475,
      "step": 56
    },
    {
      "epoch": 0.31105047748976805,
      "grad_norm": 2.518003982101406,
      "learning_rate": 4.956521739130435e-05,
      "loss": 1.0277,
      "step": 57
    },
    {
      "epoch": 0.3165075034106412,
      "grad_norm": 1.810486050762184,
      "learning_rate": 5.043478260869566e-05,
      "loss": 1.0167,
      "step": 58
    },
    {
      "epoch": 0.3219645293315143,
      "grad_norm": 2.693889814467265,
      "learning_rate": 5.130434782608696e-05,
      "loss": 1.0236,
      "step": 59
    },
    {
      "epoch": 0.3274215552523875,
      "grad_norm": 1.859820799030674,
      "learning_rate": 5.217391304347826e-05,
      "loss": 1.0109,
      "step": 60
    },
    {
      "epoch": 0.3328785811732606,
      "grad_norm": 2.1889623213157674,
      "learning_rate": 5.304347826086957e-05,
      "loss": 1.0222,
      "step": 61
    },
    {
      "epoch": 0.3383356070941337,
      "grad_norm": 2.300198913554314,
      "learning_rate": 5.391304347826087e-05,
      "loss": 1.0365,
      "step": 62
    },
    {
      "epoch": 0.34379263301500684,
      "grad_norm": 1.5430289249817781,
      "learning_rate": 5.478260869565218e-05,
      "loss": 1.0128,
      "step": 63
    },
    {
      "epoch": 0.34924965893587995,
      "grad_norm": 3.477710986140807,
      "learning_rate": 5.565217391304348e-05,
      "loss": 1.0071,
      "step": 64
    },
    {
      "epoch": 0.35470668485675305,
      "grad_norm": 2.7818030874339033,
      "learning_rate": 5.652173913043479e-05,
      "loss": 1.0202,
      "step": 65
    },
    {
      "epoch": 0.3601637107776262,
      "grad_norm": 2.6548687966990943,
      "learning_rate": 5.739130434782609e-05,
      "loss": 1.022,
      "step": 66
    },
    {
      "epoch": 0.3656207366984993,
      "grad_norm": 2.9183382843347623,
      "learning_rate": 5.82608695652174e-05,
      "loss": 1.0245,
      "step": 67
    },
    {
      "epoch": 0.3710777626193724,
      "grad_norm": 1.4627910570537404,
      "learning_rate": 5.91304347826087e-05,
      "loss": 0.9863,
      "step": 68
    },
    {
      "epoch": 0.3765347885402456,
      "grad_norm": 2.295615927686121,
      "learning_rate": 6.000000000000001e-05,
      "loss": 1.0038,
      "step": 69
    },
    {
      "epoch": 0.3819918144611187,
      "grad_norm": 2.463602586043397,
      "learning_rate": 6.086956521739131e-05,
      "loss": 1.0182,
      "step": 70
    },
    {
      "epoch": 0.3874488403819918,
      "grad_norm": 2.388407526596309,
      "learning_rate": 6.173913043478262e-05,
      "loss": 1.0142,
      "step": 71
    },
    {
      "epoch": 0.39290586630286495,
      "grad_norm": 2.5037634590498357,
      "learning_rate": 6.260869565217392e-05,
      "loss": 1.0078,
      "step": 72
    },
    {
      "epoch": 0.39836289222373805,
      "grad_norm": 1.3685081163007409,
      "learning_rate": 6.347826086956523e-05,
      "loss": 1.0144,
      "step": 73
    },
    {
      "epoch": 0.4038199181446112,
      "grad_norm": 2.987414238598592,
      "learning_rate": 6.434782608695653e-05,
      "loss": 1.0227,
      "step": 74
    },
    {
      "epoch": 0.4092769440654843,
      "grad_norm": 2.285557770071163,
      "learning_rate": 6.521739130434783e-05,
      "loss": 1.0216,
      "step": 75
    },
    {
      "epoch": 0.4147339699863574,
      "grad_norm": 2.1674983433636137,
      "learning_rate": 6.608695652173914e-05,
      "loss": 1.0075,
      "step": 76
    },
    {
      "epoch": 0.4201909959072306,
      "grad_norm": 2.3056726088264905,
      "learning_rate": 6.695652173913044e-05,
      "loss": 0.9964,
      "step": 77
    },
    {
      "epoch": 0.4256480218281037,
      "grad_norm": 2.6041661067718525,
      "learning_rate": 6.782608695652175e-05,
      "loss": 0.9956,
      "step": 78
    },
    {
      "epoch": 0.4311050477489768,
      "grad_norm": 1.837082219111084,
      "learning_rate": 6.869565217391305e-05,
      "loss": 1.0209,
      "step": 79
    },
    {
      "epoch": 0.43656207366984995,
      "grad_norm": 2.7126777057295275,
      "learning_rate": 6.956521739130436e-05,
      "loss": 1.0037,
      "step": 80
    },
    {
      "epoch": 0.44201909959072305,
      "grad_norm": 2.782292772347231,
      "learning_rate": 7.043478260869566e-05,
      "loss": 1.0018,
      "step": 81
    },
    {
      "epoch": 0.44747612551159616,
      "grad_norm": 1.8194383009699067,
      "learning_rate": 7.130434782608696e-05,
      "loss": 1.0102,
      "step": 82
    },
    {
      "epoch": 0.4529331514324693,
      "grad_norm": 2.3506780994607177,
      "learning_rate": 7.217391304347827e-05,
      "loss": 1.0011,
      "step": 83
    },
    {
      "epoch": 0.4583901773533424,
      "grad_norm": 2.588474724733464,
      "learning_rate": 7.304347826086957e-05,
      "loss": 0.9918,
      "step": 84
    },
    {
      "epoch": 0.4638472032742155,
      "grad_norm": 2.966355401640028,
      "learning_rate": 7.391304347826088e-05,
      "loss": 1.0082,
      "step": 85
    },
    {
      "epoch": 0.4693042291950887,
      "grad_norm": 1.63650415845399,
      "learning_rate": 7.478260869565218e-05,
      "loss": 0.9965,
      "step": 86
    },
    {
      "epoch": 0.4747612551159618,
      "grad_norm": 3.039157506943887,
      "learning_rate": 7.565217391304349e-05,
      "loss": 1.0186,
      "step": 87
    },
    {
      "epoch": 0.48021828103683495,
      "grad_norm": 2.5849399835417897,
      "learning_rate": 7.652173913043479e-05,
      "loss": 1.0188,
      "step": 88
    },
    {
      "epoch": 0.48567530695770805,
      "grad_norm": 1.5912811102421822,
      "learning_rate": 7.73913043478261e-05,
      "loss": 0.9933,
      "step": 89
    },
    {
      "epoch": 0.49113233287858116,
      "grad_norm": 2.3088953829836547,
      "learning_rate": 7.82608695652174e-05,
      "loss": 1.0237,
      "step": 90
    },
    {
      "epoch": 0.4965893587994543,
      "grad_norm": 2.965499963162432,
      "learning_rate": 7.91304347826087e-05,
      "loss": 0.9983,
      "step": 91
    },
    {
      "epoch": 0.5020463847203275,
      "grad_norm": 1.7790041385019335,
      "learning_rate": 8e-05,
      "loss": 1.0253,
      "step": 92
    },
    {
      "epoch": 0.5075034106412005,
      "grad_norm": 3.3482881631202175,
      "learning_rate": 7.999970857316263e-05,
      "loss": 1.0017,
      "step": 93
    },
    {
      "epoch": 0.5129604365620737,
      "grad_norm": 2.0737327995304757,
      "learning_rate": 7.999883429689698e-05,
      "loss": 1.0161,
      "step": 94
    },
    {
      "epoch": 0.5184174624829468,
      "grad_norm": 3.4718599185205252,
      "learning_rate": 7.999737718394245e-05,
      "loss": 1.0227,
      "step": 95
    },
    {
      "epoch": 0.5238744884038199,
      "grad_norm": 2.6905459489144494,
      "learning_rate": 7.999533725553111e-05,
      "loss": 1.0238,
      "step": 96
    },
    {
      "epoch": 0.529331514324693,
      "grad_norm": 2.643496365226572,
      "learning_rate": 7.999271454138746e-05,
      "loss": 0.9963,
      "step": 97
    },
    {
      "epoch": 0.5347885402455662,
      "grad_norm": 2.398472468783254,
      "learning_rate": 7.998950907972797e-05,
      "loss": 1.0067,
      "step": 98
    },
    {
      "epoch": 0.5402455661664393,
      "grad_norm": 2.4862794999589632,
      "learning_rate": 7.998572091726051e-05,
      "loss": 1.0016,
      "step": 99
    },
    {
      "epoch": 0.5457025920873124,
      "grad_norm": 2.703979830017154,
      "learning_rate": 7.99813501091837e-05,
      "loss": 0.9997,
      "step": 100
    },
    {
      "epoch": 0.5511596180081856,
      "grad_norm": 2.638325870561522,
      "learning_rate": 7.997639671918607e-05,
      "loss": 0.9666,
      "step": 101
    },
    {
      "epoch": 0.5566166439290586,
      "grad_norm": 1.923660396652347,
      "learning_rate": 7.997086081944518e-05,
      "loss": 0.9901,
      "step": 102
    },
    {
      "epoch": 0.5620736698499318,
      "grad_norm": 1.8412763920684536,
      "learning_rate": 7.996474249062649e-05,
      "loss": 0.9837,
      "step": 103
    },
    {
      "epoch": 0.567530695770805,
      "grad_norm": 3.299280808379093,
      "learning_rate": 7.995804182188227e-05,
      "loss": 0.9911,
      "step": 104
    },
    {
      "epoch": 0.572987721691678,
      "grad_norm": 2.370087367079061,
      "learning_rate": 7.995075891085025e-05,
      "loss": 0.9738,
      "step": 105
    },
    {
      "epoch": 0.5784447476125512,
      "grad_norm": 1.6382462163124434,
      "learning_rate": 7.994289386365223e-05,
      "loss": 0.9915,
      "step": 106
    },
    {
      "epoch": 0.5839017735334243,
      "grad_norm": 2.1387676364417407,
      "learning_rate": 7.99344467948925e-05,
      "loss": 0.9986,
      "step": 107
    },
    {
      "epoch": 0.5893587994542974,
      "grad_norm": 4.201439600148248,
      "learning_rate": 7.992541782765617e-05,
      "loss": 0.9959,
      "step": 108
    },
    {
      "epoch": 0.5948158253751705,
      "grad_norm": 1.9732356205001438,
      "learning_rate": 7.991580709350743e-05,
      "loss": 1.0058,
      "step": 109
    },
    {
      "epoch": 0.6002728512960437,
      "grad_norm": 5.603033692402054,
      "learning_rate": 7.990561473248756e-05,
      "loss": 1.0192,
      "step": 110
    },
    {
      "epoch": 0.6057298772169167,
      "grad_norm": 4.217157974337054,
      "learning_rate": 7.989484089311294e-05,
      "loss": 1.0206,
      "step": 111
    },
    {
      "epoch": 0.6111869031377899,
      "grad_norm": 4.733955047783259,
      "learning_rate": 7.988348573237286e-05,
      "loss": 1.032,
      "step": 112
    },
    {
      "epoch": 0.616643929058663,
      "grad_norm": 4.030897341866623,
      "learning_rate": 7.987154941572726e-05,
      "loss": 1.0136,
      "step": 113
    },
    {
      "epoch": 0.6221009549795361,
      "grad_norm": 3.2852614632490287,
      "learning_rate": 7.985903211710429e-05,
      "loss": 0.984,
      "step": 114
    },
    {
      "epoch": 0.6275579809004093,
      "grad_norm": 2.8477242486115157,
      "learning_rate": 7.984593401889778e-05,
      "loss": 0.9861,
      "step": 115
    },
    {
      "epoch": 0.6330150068212824,
      "grad_norm": 2.979343225271941,
      "learning_rate": 7.98322553119646e-05,
      "loss": 1.0057,
      "step": 116
    },
    {
      "epoch": 0.6384720327421555,
      "grad_norm": 2.4230735574970295,
      "learning_rate": 7.981799619562186e-05,
      "loss": 0.9798,
      "step": 117
    },
    {
      "epoch": 0.6439290586630286,
      "grad_norm": 1.4903910249137124,
      "learning_rate": 7.980315687764404e-05,
      "loss": 0.9832,
      "step": 118
    },
    {
      "epoch": 0.6493860845839018,
      "grad_norm": 2.851390699017869,
      "learning_rate": 7.978773757425987e-05,
      "loss": 0.9859,
      "step": 119
    },
    {
      "epoch": 0.654843110504775,
      "grad_norm": 2.449200648456771,
      "learning_rate": 7.977173851014935e-05,
      "loss": 1.0016,
      "step": 120
    },
    {
      "epoch": 0.660300136425648,
      "grad_norm": 1.3908196441719074,
      "learning_rate": 7.975515991844028e-05,
      "loss": 0.9925,
      "step": 121
    },
    {
      "epoch": 0.6657571623465212,
      "grad_norm": 2.709126199988088,
      "learning_rate": 7.973800204070497e-05,
      "loss": 1.0183,
      "step": 122
    },
    {
      "epoch": 0.6712141882673943,
      "grad_norm": 2.4950773390337284,
      "learning_rate": 7.972026512695677e-05,
      "loss": 0.9983,
      "step": 123
    },
    {
      "epoch": 0.6766712141882674,
      "grad_norm": 1.8156715188883938,
      "learning_rate": 7.970194943564626e-05,
      "loss": 0.9966,
      "step": 124
    },
    {
      "epoch": 0.6821282401091405,
      "grad_norm": 2.0124460248386837,
      "learning_rate": 7.968305523365768e-05,
      "loss": 0.9861,
      "step": 125
    },
    {
      "epoch": 0.6875852660300137,
      "grad_norm": 2.327548310536662,
      "learning_rate": 7.96635827963049e-05,
      "loss": 1.005,
      "step": 126
    },
    {
      "epoch": 0.6930422919508867,
      "grad_norm": 2.0384540649729046,
      "learning_rate": 7.964353240732744e-05,
      "loss": 0.9643,
      "step": 127
    },
    {
      "epoch": 0.6984993178717599,
      "grad_norm": 1.5857232283507334,
      "learning_rate": 7.96229043588864e-05,
      "loss": 0.9942,
      "step": 128
    },
    {
      "epoch": 0.703956343792633,
      "grad_norm": 2.6642249621882947,
      "learning_rate": 7.960169895156011e-05,
      "loss": 1.0059,
      "step": 129
    },
    {
      "epoch": 0.7094133697135061,
      "grad_norm": 2.003197328585847,
      "learning_rate": 7.95799164943398e-05,
      "loss": 0.9703,
      "step": 130
    },
    {
      "epoch": 0.7148703956343793,
      "grad_norm": 2.080102961922397,
      "learning_rate": 7.955755730462512e-05,
      "loss": 0.9901,
      "step": 131
    },
    {
      "epoch": 0.7203274215552524,
      "grad_norm": 2.035255856600717,
      "learning_rate": 7.953462170821948e-05,
      "loss": 0.9782,
      "step": 132
    },
    {
      "epoch": 0.7257844474761255,
      "grad_norm": 2.352960888257511,
      "learning_rate": 7.951111003932526e-05,
      "loss": 0.9849,
      "step": 133
    },
    {
      "epoch": 0.7312414733969986,
      "grad_norm": 2.073525444432574,
      "learning_rate": 7.948702264053904e-05,
      "loss": 0.9678,
      "step": 134
    },
    {
      "epoch": 0.7366984993178718,
      "grad_norm": 2.3421982402930444,
      "learning_rate": 7.946235986284655e-05,
      "loss": 0.9783,
      "step": 135
    },
    {
      "epoch": 0.7421555252387448,
      "grad_norm": 1.5736240970851025,
      "learning_rate": 7.943712206561755e-05,
      "loss": 0.9841,
      "step": 136
    },
    {
      "epoch": 0.747612551159618,
      "grad_norm": 2.489956249226116,
      "learning_rate": 7.94113096166006e-05,
      "loss": 0.9639,
      "step": 137
    },
    {
      "epoch": 0.7530695770804912,
      "grad_norm": 1.5389665449534187,
      "learning_rate": 7.938492289191775e-05,
      "loss": 0.9749,
      "step": 138
    },
    {
      "epoch": 0.7585266030013642,
      "grad_norm": 2.8469902237174884,
      "learning_rate": 7.935796227605896e-05,
      "loss": 0.9704,
      "step": 139
    },
    {
      "epoch": 0.7639836289222374,
      "grad_norm": 2.0927134145359885,
      "learning_rate": 7.933042816187659e-05,
      "loss": 0.9851,
      "step": 140
    },
    {
      "epoch": 0.7694406548431105,
      "grad_norm": 1.9936042857918566,
      "learning_rate": 7.930232095057961e-05,
      "loss": 1.0029,
      "step": 141
    },
    {
      "epoch": 0.7748976807639836,
      "grad_norm": 2.386887860120376,
      "learning_rate": 7.927364105172784e-05,
      "loss": 0.9846,
      "step": 142
    },
    {
      "epoch": 0.7803547066848567,
      "grad_norm": 2.3622753734620368,
      "learning_rate": 7.924438888322588e-05,
      "loss": 0.9793,
      "step": 143
    },
    {
      "epoch": 0.7858117326057299,
      "grad_norm": 2.1818780607623753,
      "learning_rate": 7.921456487131706e-05,
      "loss": 0.9788,
      "step": 144
    },
    {
      "epoch": 0.791268758526603,
      "grad_norm": 1.182780599547151,
      "learning_rate": 7.918416945057726e-05,
      "loss": 0.9643,
      "step": 145
    },
    {
      "epoch": 0.7967257844474761,
      "grad_norm": 2.8211946276144255,
      "learning_rate": 7.915320306390856e-05,
      "loss": 1.0119,
      "step": 146
    },
    {
      "epoch": 0.8021828103683493,
      "grad_norm": 2.388157316898997,
      "learning_rate": 7.912166616253276e-05,
      "loss": 0.9619,
      "step": 147
    },
    {
      "epoch": 0.8076398362892224,
      "grad_norm": 1.5731846927644901,
      "learning_rate": 7.908955920598482e-05,
      "loss": 0.9627,
      "step": 148
    },
    {
      "epoch": 0.8130968622100955,
      "grad_norm": 2.546269463790635,
      "learning_rate": 7.90568826621062e-05,
      "loss": 1.0079,
      "step": 149
    },
    {
      "epoch": 0.8185538881309686,
      "grad_norm": 1.671363635943377,
      "learning_rate": 7.902363700703797e-05,
      "loss": 0.9845,
      "step": 150
    },
    {
      "epoch": 0.8240109140518418,
      "grad_norm": 2.1689468730120343,
      "learning_rate": 7.898982272521396e-05,
      "loss": 0.9834,
      "step": 151
    },
    {
      "epoch": 0.8294679399727148,
      "grad_norm": 2.5028825478335253,
      "learning_rate": 7.895544030935361e-05,
      "loss": 0.9785,
      "step": 152
    },
    {
      "epoch": 0.834924965893588,
      "grad_norm": 1.7649827216268308,
      "learning_rate": 7.892049026045487e-05,
      "loss": 0.9635,
      "step": 153
    },
    {
      "epoch": 0.8403819918144612,
      "grad_norm": 2.0516340222455143,
      "learning_rate": 7.888497308778685e-05,
      "loss": 0.9655,
      "step": 154
    },
    {
      "epoch": 0.8458390177353342,
      "grad_norm": 1.643525463034892,
      "learning_rate": 7.88488893088824e-05,
      "loss": 0.9771,
      "step": 155
    },
    {
      "epoch": 0.8512960436562074,
      "grad_norm": 2.5710664190674977,
      "learning_rate": 7.88122394495306e-05,
      "loss": 0.9734,
      "step": 156
    },
    {
      "epoch": 0.8567530695770805,
      "grad_norm": 1.8164577100205415,
      "learning_rate": 7.877502404376911e-05,
      "loss": 0.9765,
      "step": 157
    },
    {
      "epoch": 0.8622100954979536,
      "grad_norm": 2.675088927773194,
      "learning_rate": 7.87372436338763e-05,
      "loss": 0.973,
      "step": 158
    },
    {
      "epoch": 0.8676671214188267,
      "grad_norm": 1.6058702069668531,
      "learning_rate": 7.869889877036344e-05,
      "loss": 0.9927,
      "step": 159
    },
    {
      "epoch": 0.8731241473396999,
      "grad_norm": 2.726714878756649,
      "learning_rate": 7.865999001196666e-05,
      "loss": 0.9591,
      "step": 160
    },
    {
      "epoch": 0.878581173260573,
      "grad_norm": 2.294598196915522,
      "learning_rate": 7.862051792563878e-05,
      "loss": 0.9958,
      "step": 161
    },
    {
      "epoch": 0.8840381991814461,
      "grad_norm": 2.1238876211739766,
      "learning_rate": 7.858048308654104e-05,
      "loss": 0.9747,
      "step": 162
    },
    {
      "epoch": 0.8894952251023193,
      "grad_norm": 2.02064179400817,
      "learning_rate": 7.853988607803479e-05,
      "loss": 0.9917,
      "step": 163
    },
    {
      "epoch": 0.8949522510231923,
      "grad_norm": 1.3894273755079185,
      "learning_rate": 7.849872749167291e-05,
      "loss": 0.9781,
      "step": 164
    },
    {
      "epoch": 0.9004092769440655,
      "grad_norm": 1.648978217855913,
      "learning_rate": 7.845700792719124e-05,
      "loss": 0.9698,
      "step": 165
    },
    {
      "epoch": 0.9058663028649386,
      "grad_norm": 3.51030441484047,
      "learning_rate": 7.84147279924998e-05,
      "loss": 0.9863,
      "step": 166
    },
    {
      "epoch": 0.9113233287858117,
      "grad_norm": 1.5500687413665548,
      "learning_rate": 7.837188830367401e-05,
      "loss": 0.9986,
      "step": 167
    },
    {
      "epoch": 0.9167803547066848,
      "grad_norm": 5.248314809323723,
      "learning_rate": 7.832848948494559e-05,
      "loss": 1.0028,
      "step": 168
    },
    {
      "epoch": 0.922237380627558,
      "grad_norm": 3.4816630300759988,
      "learning_rate": 7.828453216869356e-05,
      "loss": 1.0078,
      "step": 169
    },
    {
      "epoch": 0.927694406548431,
      "grad_norm": 5.427302474462801,
      "learning_rate": 7.824001699543502e-05,
      "loss": 1.0043,
      "step": 170
    },
    {
      "epoch": 0.9331514324693042,
      "grad_norm": 4.826695970132574,
      "learning_rate": 7.819494461381577e-05,
      "loss": 1.0106,
      "step": 171
    },
    {
      "epoch": 0.9386084583901774,
      "grad_norm": 2.4392637209839765,
      "learning_rate": 7.81493156806009e-05,
      "loss": 0.9935,
      "step": 172
    },
    {
      "epoch": 0.9440654843110505,
      "grad_norm": 3.8244216268633036,
      "learning_rate": 7.81031308606652e-05,
      "loss": 0.976,
      "step": 173
    },
    {
      "epoch": 0.9495225102319236,
      "grad_norm": 2.8457649655988115,
      "learning_rate": 7.805639082698342e-05,
      "loss": 0.9786,
      "step": 174
    },
    {
      "epoch": 0.9549795361527967,
      "grad_norm": 2.731706315180723,
      "learning_rate": 7.800909626062064e-05,
      "loss": 0.9771,
      "step": 175
    },
    {
      "epoch": 0.9604365620736699,
      "grad_norm": 2.202992780626566,
      "learning_rate": 7.796124785072212e-05,
      "loss": 0.9915,
      "step": 176
    },
    {
      "epoch": 0.965893587994543,
      "grad_norm": 1.9587247755808166,
      "learning_rate": 7.791284629450338e-05,
      "loss": 0.9702,
      "step": 177
    },
    {
      "epoch": 0.9713506139154161,
      "grad_norm": 2.1379267827418538,
      "learning_rate": 7.78638922972401e-05,
      "loss": 0.9713,
      "step": 178
    },
    {
      "epoch": 0.9768076398362893,
      "grad_norm": 1.9276330849235372,
      "learning_rate": 7.781438657225764e-05,
      "loss": 0.9825,
      "step": 179
    },
    {
      "epoch": 0.9822646657571623,
      "grad_norm": 2.2740646257729353,
      "learning_rate": 7.77643298409209e-05,
      "loss": 0.9546,
      "step": 180
    },
    {
      "epoch": 0.9877216916780355,
      "grad_norm": 1.6849230748486481,
      "learning_rate": 7.77137228326236e-05,
      "loss": 0.9878,
      "step": 181
    },
    {
      "epoch": 0.9931787175989086,
      "grad_norm": 1.930950153364107,
      "learning_rate": 7.766256628477775e-05,
      "loss": 0.9614,
      "step": 182
    },
    {
      "epoch": 0.9986357435197817,
      "grad_norm": 1.651071282168761,
      "learning_rate": 7.761086094280291e-05,
      "loss": 0.9834,
      "step": 183
    },
    {
      "epoch": 1.004092769440655,
      "grad_norm": 3.4237344018894738,
      "learning_rate": 7.755860756011531e-05,
      "loss": 1.6698,
      "step": 184
    },
    {
      "epoch": 1.009549795361528,
      "grad_norm": 2.550843838303363,
      "learning_rate": 7.750580689811683e-05,
      "loss": 0.9341,
      "step": 185
    },
    {
      "epoch": 1.015006821282401,
      "grad_norm": 1.2237286076246516,
      "learning_rate": 7.745245972618396e-05,
      "loss": 0.969,
      "step": 186
    },
    {
      "epoch": 1.0204638472032743,
      "grad_norm": 1.7619044634166372,
      "learning_rate": 7.739856682165661e-05,
      "loss": 0.9697,
      "step": 187
    },
    {
      "epoch": 1.0259208731241474,
      "grad_norm": 1.84237685056023,
      "learning_rate": 7.73441289698267e-05,
      "loss": 0.9706,
      "step": 188
    },
    {
      "epoch": 1.0313778990450204,
      "grad_norm": 2.246735296743444,
      "learning_rate": 7.728914696392677e-05,
      "loss": 0.9599,
      "step": 189
    },
    {
      "epoch": 1.0368349249658937,
      "grad_norm": 1.5477004028342645,
      "learning_rate": 7.723362160511844e-05,
      "loss": 0.9341,
      "step": 190
    },
    {
      "epoch": 1.0422919508867667,
      "grad_norm": 1.628646582504066,
      "learning_rate": 7.717755370248069e-05,
      "loss": 0.9311,
      "step": 191
    },
    {
      "epoch": 1.0477489768076398,
      "grad_norm": 2.7608821191707436,
      "learning_rate": 7.712094407299809e-05,
      "loss": 0.9521,
      "step": 192
    },
    {
      "epoch": 1.053206002728513,
      "grad_norm": 1.6043767223562262,
      "learning_rate": 7.706379354154891e-05,
      "loss": 0.9591,
      "step": 193
    },
    {
      "epoch": 1.058663028649386,
      "grad_norm": 2.6819244792153336,
      "learning_rate": 7.700610294089309e-05,
      "loss": 0.9326,
      "step": 194
    },
    {
      "epoch": 1.0641200545702592,
      "grad_norm": 2.170173544833822,
      "learning_rate": 7.694787311166009e-05,
      "loss": 0.976,
      "step": 195
    },
    {
      "epoch": 1.0695770804911324,
      "grad_norm": 3.105411956285718,
      "learning_rate": 7.688910490233664e-05,
      "loss": 0.9667,
      "step": 196
    },
    {
      "epoch": 1.0750341064120055,
      "grad_norm": 2.0189173881999083,
      "learning_rate": 7.682979916925445e-05,
      "loss": 0.9432,
      "step": 197
    },
    {
      "epoch": 1.0804911323328785,
      "grad_norm": 3.0432618104084512,
      "learning_rate": 7.676995677657759e-05,
      "loss": 0.9494,
      "step": 198
    },
    {
      "epoch": 1.0859481582537518,
      "grad_norm": 2.6892073422610805,
      "learning_rate": 7.670957859629003e-05,
      "loss": 0.9705,
      "step": 199
    },
    {
      "epoch": 1.0914051841746248,
      "grad_norm": 1.7496055102061392,
      "learning_rate": 7.664866550818289e-05,
      "loss": 0.9446,
      "step": 200
    },
    {
      "epoch": 1.096862210095498,
      "grad_norm": 2.1862992841348547,
      "learning_rate": 7.65872183998416e-05,
      "loss": 0.9442,
      "step": 201
    },
    {
      "epoch": 1.1023192360163712,
      "grad_norm": 2.311563125962559,
      "learning_rate": 7.652523816663298e-05,
      "loss": 0.9488,
      "step": 202
    },
    {
      "epoch": 1.1077762619372442,
      "grad_norm": 1.5707872490013692,
      "learning_rate": 7.646272571169221e-05,
      "loss": 0.9428,
      "step": 203
    },
    {
      "epoch": 1.1132332878581173,
      "grad_norm": 1.1293404204912825,
      "learning_rate": 7.639968194590961e-05,
      "loss": 0.9377,
      "step": 204
    },
    {
      "epoch": 1.1186903137789905,
      "grad_norm": 2.0853381432635087,
      "learning_rate": 7.633610778791746e-05,
      "loss": 0.9435,
      "step": 205
    },
    {
      "epoch": 1.1241473396998636,
      "grad_norm": 1.6237281426907448,
      "learning_rate": 7.627200416407656e-05,
      "loss": 0.9545,
      "step": 206
    },
    {
      "epoch": 1.1296043656207366,
      "grad_norm": 2.338186851948375,
      "learning_rate": 7.620737200846271e-05,
      "loss": 0.9547,
      "step": 207
    },
    {
      "epoch": 1.13506139154161,
      "grad_norm": 1.6338755462437418,
      "learning_rate": 7.614221226285317e-05,
      "loss": 0.9571,
      "step": 208
    },
    {
      "epoch": 1.140518417462483,
      "grad_norm": 2.4727023861577915,
      "learning_rate": 7.607652587671284e-05,
      "loss": 0.9449,
      "step": 209
    },
    {
      "epoch": 1.145975443383356,
      "grad_norm": 1.6214090487676778,
      "learning_rate": 7.601031380718053e-05,
      "loss": 0.9336,
      "step": 210
    },
    {
      "epoch": 1.1514324693042293,
      "grad_norm": 2.8758801504575318,
      "learning_rate": 7.594357701905493e-05,
      "loss": 0.9478,
      "step": 211
    },
    {
      "epoch": 1.1568894952251023,
      "grad_norm": 2.343741537607968,
      "learning_rate": 7.58763164847806e-05,
      "loss": 0.9568,
      "step": 212
    },
    {
      "epoch": 1.1623465211459754,
      "grad_norm": 2.1342068216692374,
      "learning_rate": 7.580853318443378e-05,
      "loss": 0.9506,
      "step": 213
    },
    {
      "epoch": 1.1678035470668486,
      "grad_norm": 1.9405401625810694,
      "learning_rate": 7.574022810570811e-05,
      "loss": 0.9598,
      "step": 214
    },
    {
      "epoch": 1.1732605729877217,
      "grad_norm": 2.047006811241721,
      "learning_rate": 7.567140224390026e-05,
      "loss": 0.9398,
      "step": 215
    },
    {
      "epoch": 1.1787175989085947,
      "grad_norm": 1.2649054358403522,
      "learning_rate": 7.560205660189536e-05,
      "loss": 0.9496,
      "step": 216
    },
    {
      "epoch": 1.184174624829468,
      "grad_norm": 2.2873286146173024,
      "learning_rate": 7.553219219015247e-05,
      "loss": 0.9356,
      "step": 217
    },
    {
      "epoch": 1.189631650750341,
      "grad_norm": 1.7493032316806663,
      "learning_rate": 7.546181002668985e-05,
      "loss": 0.9433,
      "step": 218
    },
    {
      "epoch": 1.195088676671214,
      "grad_norm": 2.0223831026311516,
      "learning_rate": 7.539091113707003e-05,
      "loss": 0.9499,
      "step": 219
    },
    {
      "epoch": 1.2005457025920874,
      "grad_norm": 1.8819082575345663,
      "learning_rate": 7.5319496554385e-05,
      "loss": 0.9168,
      "step": 220
    },
    {
      "epoch": 1.2060027285129604,
      "grad_norm": 1.4387903068149017,
      "learning_rate": 7.524756731924103e-05,
      "loss": 0.9594,
      "step": 221
    },
    {
      "epoch": 1.2114597544338335,
      "grad_norm": 1.7254879140852797,
      "learning_rate": 7.517512447974361e-05,
      "loss": 0.9591,
      "step": 222
    },
    {
      "epoch": 1.2169167803547067,
      "grad_norm": 1.6398981256156826,
      "learning_rate": 7.510216909148214e-05,
      "loss": 0.9597,
      "step": 223
    },
    {
      "epoch": 1.2223738062755798,
      "grad_norm": 1.9307517212835448,
      "learning_rate": 7.502870221751448e-05,
      "loss": 0.9531,
      "step": 224
    },
    {
      "epoch": 1.2278308321964528,
      "grad_norm": 2.901172104154794,
      "learning_rate": 7.495472492835159e-05,
      "loss": 0.9514,
      "step": 225
    },
    {
      "epoch": 1.233287858117326,
      "grad_norm": 1.2148442619506596,
      "learning_rate": 7.488023830194185e-05,
      "loss": 0.9525,
      "step": 226
    },
    {
      "epoch": 1.2387448840381992,
      "grad_norm": 3.0744826253657154,
      "learning_rate": 7.480524342365535e-05,
      "loss": 0.9334,
      "step": 227
    },
    {
      "epoch": 1.2442019099590724,
      "grad_norm": 2.512018245011184,
      "learning_rate": 7.472974138626809e-05,
      "loss": 0.9748,
      "step": 228
    },
    {
      "epoch": 1.2496589358799455,
      "grad_norm": 2.0167717510810075,
      "learning_rate": 7.465373328994608e-05,
      "loss": 0.9561,
      "step": 229
    },
    {
      "epoch": 1.2551159618008185,
      "grad_norm": 2.69272126540707,
      "learning_rate": 7.457722024222927e-05,
      "loss": 0.9535,
      "step": 230
    },
    {
      "epoch": 1.2605729877216918,
      "grad_norm": 2.1489193698787075,
      "learning_rate": 7.450020335801544e-05,
      "loss": 0.9497,
      "step": 231
    },
    {
      "epoch": 1.2660300136425648,
      "grad_norm": 1.674495703338951,
      "learning_rate": 7.442268375954394e-05,
      "loss": 0.9468,
      "step": 232
    },
    {
      "epoch": 1.271487039563438,
      "grad_norm": 2.861569881618082,
      "learning_rate": 7.434466257637933e-05,
      "loss": 0.953,
      "step": 233
    },
    {
      "epoch": 1.2769440654843112,
      "grad_norm": 1.505253232519278,
      "learning_rate": 7.426614094539496e-05,
      "loss": 0.9478,
      "step": 234
    },
    {
      "epoch": 1.2824010914051842,
      "grad_norm": 2.1966862404860317,
      "learning_rate": 7.418712001075635e-05,
      "loss": 0.9434,
      "step": 235
    },
    {
      "epoch": 1.2878581173260573,
      "grad_norm": 1.8142529829887761,
      "learning_rate": 7.410760092390456e-05,
      "loss": 0.9185,
      "step": 236
    },
    {
      "epoch": 1.2933151432469305,
      "grad_norm": 2.042423103810234,
      "learning_rate": 7.402758484353938e-05,
      "loss": 0.947,
      "step": 237
    },
    {
      "epoch": 1.2987721691678036,
      "grad_norm": 2.807776807250219,
      "learning_rate": 7.394707293560248e-05,
      "loss": 0.9389,
      "step": 238
    },
    {
      "epoch": 1.3042291950886766,
      "grad_norm": 1.843186901692036,
      "learning_rate": 7.386606637326038e-05,
      "loss": 0.9495,
      "step": 239
    },
    {
      "epoch": 1.30968622100955,
      "grad_norm": 3.4432614710350853,
      "learning_rate": 7.378456633688742e-05,
      "loss": 0.944,
      "step": 240
    },
    {
      "epoch": 1.315143246930423,
      "grad_norm": 2.9796135612528145,
      "learning_rate": 7.370257401404846e-05,
      "loss": 0.96,
      "step": 241
    },
    {
      "epoch": 1.320600272851296,
      "grad_norm": 2.4600116762336706,
      "learning_rate": 7.36200905994817e-05,
      "loss": 0.9522,
      "step": 242
    },
    {
      "epoch": 1.3260572987721693,
      "grad_norm": 2.5423218083756924,
      "learning_rate": 7.353711729508114e-05,
      "loss": 0.9332,
      "step": 243
    },
    {
      "epoch": 1.3315143246930423,
      "grad_norm": 1.5354349040627853,
      "learning_rate": 7.34536553098792e-05,
      "loss": 0.9649,
      "step": 244
    },
    {
      "epoch": 1.3369713506139154,
      "grad_norm": 3.3281556440961704,
      "learning_rate": 7.336970586002896e-05,
      "loss": 0.9727,
      "step": 245
    },
    {
      "epoch": 1.3424283765347886,
      "grad_norm": 1.9703109238860381,
      "learning_rate": 7.328527016878658e-05,
      "loss": 0.9457,
      "step": 246
    },
    {
      "epoch": 1.3478854024556617,
      "grad_norm": 3.5456743561386026,
      "learning_rate": 7.320034946649339e-05,
      "loss": 0.939,
      "step": 247
    },
    {
      "epoch": 1.3533424283765347,
      "grad_norm": 3.043515887992997,
      "learning_rate": 7.311494499055794e-05,
      "loss": 0.9647,
      "step": 248
    },
    {
      "epoch": 1.358799454297408,
      "grad_norm": 2.2225676170915265,
      "learning_rate": 7.302905798543809e-05,
      "loss": 0.9434,
      "step": 249
    },
    {
      "epoch": 1.364256480218281,
      "grad_norm": 2.316972441261721,
      "learning_rate": 7.294268970262271e-05,
      "loss": 0.9507,
      "step": 250
    },
    {
      "epoch": 1.369713506139154,
      "grad_norm": 1.3452402784796123,
      "learning_rate": 7.28558414006136e-05,
      "loss": 0.9401,
      "step": 251
    },
    {
      "epoch": 1.3751705320600274,
      "grad_norm": 3.2470762419612194,
      "learning_rate": 7.276851434490706e-05,
      "loss": 0.9578,
      "step": 252
    },
    {
      "epoch": 1.3806275579809004,
      "grad_norm": 2.7728684726943884,
      "learning_rate": 7.268070980797547e-05,
      "loss": 0.947,
      "step": 253
    },
    {
      "epoch": 1.3860845839017735,
      "grad_norm": 2.9439862066425593,
      "learning_rate": 7.259242906924876e-05,
      "loss": 0.9588,
      "step": 254
    },
    {
      "epoch": 1.3915416098226467,
      "grad_norm": 2.4925113769599463,
      "learning_rate": 7.250367341509573e-05,
      "loss": 0.9481,
      "step": 255
    },
    {
      "epoch": 1.3969986357435198,
      "grad_norm": 2.4424975656100707,
      "learning_rate": 7.241444413880539e-05,
      "loss": 0.9477,
      "step": 256
    },
    {
      "epoch": 1.4024556616643928,
      "grad_norm": 2.181544082143711,
      "learning_rate": 7.232474254056801e-05,
      "loss": 0.9412,
      "step": 257
    },
    {
      "epoch": 1.407912687585266,
      "grad_norm": 1.7636209175878623,
      "learning_rate": 7.223456992745626e-05,
      "loss": 0.9347,
      "step": 258
    },
    {
      "epoch": 1.4133697135061392,
      "grad_norm": 2.146267438897671,
      "learning_rate": 7.214392761340611e-05,
      "loss": 0.9383,
      "step": 259
    },
    {
      "epoch": 1.4188267394270122,
      "grad_norm": 1.8385114341486435,
      "learning_rate": 7.205281691919769e-05,
      "loss": 0.9619,
      "step": 260
    },
    {
      "epoch": 1.4242837653478855,
      "grad_norm": 1.0980340436389295,
      "learning_rate": 7.196123917243609e-05,
      "loss": 0.9473,
      "step": 261
    },
    {
      "epoch": 1.4297407912687585,
      "grad_norm": 1.9383750019873172,
      "learning_rate": 7.186919570753194e-05,
      "loss": 0.9375,
      "step": 262
    },
    {
      "epoch": 1.4351978171896316,
      "grad_norm": 1.6119364273709456,
      "learning_rate": 7.177668786568207e-05,
      "loss": 0.9494,
      "step": 263
    },
    {
      "epoch": 1.4406548431105048,
      "grad_norm": 2.1786353908626364,
      "learning_rate": 7.168371699484984e-05,
      "loss": 0.9239,
      "step": 264
    },
    {
      "epoch": 1.446111869031378,
      "grad_norm": 1.2435906317794803,
      "learning_rate": 7.159028444974562e-05,
      "loss": 0.9503,
      "step": 265
    },
    {
      "epoch": 1.451568894952251,
      "grad_norm": 1.2927889875300842,
      "learning_rate": 7.149639159180695e-05,
      "loss": 0.9302,
      "step": 266
    },
    {
      "epoch": 1.4570259208731242,
      "grad_norm": 1.3968086773426196,
      "learning_rate": 7.140203978917875e-05,
      "loss": 0.9501,
      "step": 267
    },
    {
      "epoch": 1.4624829467939973,
      "grad_norm": 1.8563809213966032,
      "learning_rate": 7.130723041669342e-05,
      "loss": 0.9433,
      "step": 268
    },
    {
      "epoch": 1.4679399727148703,
      "grad_norm": 1.8388340071411,
      "learning_rate": 7.121196485585071e-05,
      "loss": 0.9297,
      "step": 269
    },
    {
      "epoch": 1.4733969986357436,
      "grad_norm": 1.6925793007469925,
      "learning_rate": 7.111624449479769e-05,
      "loss": 0.9584,
      "step": 270
    },
    {
      "epoch": 1.4788540245566166,
      "grad_norm": 1.549989447828772,
      "learning_rate": 7.102007072830847e-05,
      "loss": 0.9519,
      "step": 271
    },
    {
      "epoch": 1.4843110504774897,
      "grad_norm": 1.4412184055184636,
      "learning_rate": 7.092344495776387e-05,
      "loss": 0.9317,
      "step": 272
    },
    {
      "epoch": 1.489768076398363,
      "grad_norm": 2.3051269764350955,
      "learning_rate": 7.082636859113104e-05,
      "loss": 0.9262,
      "step": 273
    },
    {
      "epoch": 1.495225102319236,
      "grad_norm": 1.4684364353103458,
      "learning_rate": 7.072884304294289e-05,
      "loss": 0.9479,
      "step": 274
    },
    {
      "epoch": 1.500682128240109,
      "grad_norm": 1.8345686209311367,
      "learning_rate": 7.063086973427754e-05,
      "loss": 0.9554,
      "step": 275
    },
    {
      "epoch": 1.5061391541609823,
      "grad_norm": 1.2709699073133172,
      "learning_rate": 7.053245009273753e-05,
      "loss": 0.9412,
      "step": 276
    },
    {
      "epoch": 1.5115961800818554,
      "grad_norm": 1.9685573107871919,
      "learning_rate": 7.043358555242914e-05,
      "loss": 0.9543,
      "step": 277
    },
    {
      "epoch": 1.5170532060027284,
      "grad_norm": 1.8665000478114042,
      "learning_rate": 7.033427755394137e-05,
      "loss": 0.9238,
      "step": 278
    },
    {
      "epoch": 1.5225102319236017,
      "grad_norm": 1.5383694260036942,
      "learning_rate": 7.023452754432503e-05,
      "loss": 0.951,
      "step": 279
    },
    {
      "epoch": 1.5279672578444747,
      "grad_norm": 1.6875275552125621,
      "learning_rate": 7.013433697707159e-05,
      "loss": 0.9611,
      "step": 280
    },
    {
      "epoch": 1.5334242837653478,
      "grad_norm": 1.939216526044695,
      "learning_rate": 7.003370731209207e-05,
      "loss": 0.9387,
      "step": 281
    },
    {
      "epoch": 1.538881309686221,
      "grad_norm": 1.3819701612894104,
      "learning_rate": 6.993264001569574e-05,
      "loss": 0.9421,
      "step": 282
    },
    {
      "epoch": 1.544338335607094,
      "grad_norm": 1.660535171681228,
      "learning_rate": 6.983113656056867e-05,
      "loss": 0.9286,
      "step": 283
    },
    {
      "epoch": 1.5497953615279672,
      "grad_norm": 2.3185193154328876,
      "learning_rate": 6.972919842575247e-05,
      "loss": 0.9312,
      "step": 284
    },
    {
      "epoch": 1.5552523874488404,
      "grad_norm": 0.9052415588070069,
      "learning_rate": 6.96268270966225e-05,
      "loss": 0.9056,
      "step": 285
    },
    {
      "epoch": 1.5607094133697135,
      "grad_norm": 1.4561558929000948,
      "learning_rate": 6.952402406486644e-05,
      "loss": 0.9381,
      "step": 286
    },
    {
      "epoch": 1.5661664392905865,
      "grad_norm": 1.543574571453614,
      "learning_rate": 6.942079082846239e-05,
      "loss": 0.9584,
      "step": 287
    },
    {
      "epoch": 1.5716234652114598,
      "grad_norm": 1.499614067677993,
      "learning_rate": 6.931712889165711e-05,
      "loss": 0.9384,
      "step": 288
    },
    {
      "epoch": 1.5770804911323328,
      "grad_norm": 1.6336048217930654,
      "learning_rate": 6.921303976494416e-05,
      "loss": 0.9529,
      "step": 289
    },
    {
      "epoch": 1.5825375170532059,
      "grad_norm": 2.0168602944752982,
      "learning_rate": 6.910852496504175e-05,
      "loss": 0.9137,
      "step": 290
    },
    {
      "epoch": 1.5879945429740792,
      "grad_norm": 1.2730565942717105,
      "learning_rate": 6.900358601487079e-05,
      "loss": 0.9349,
      "step": 291
    },
    {
      "epoch": 1.5934515688949522,
      "grad_norm": 1.9190133574494301,
      "learning_rate": 6.889822444353258e-05,
      "loss": 0.9431,
      "step": 292
    },
    {
      "epoch": 1.5989085948158253,
      "grad_norm": 1.6265733490246796,
      "learning_rate": 6.87924417862866e-05,
      "loss": 0.9532,
      "step": 293
    },
    {
      "epoch": 1.6043656207366985,
      "grad_norm": 1.6623018948743888,
      "learning_rate": 6.868623958452812e-05,
      "loss": 0.9471,
      "step": 294
    },
    {
      "epoch": 1.6098226466575716,
      "grad_norm": 0.8931957332544022,
      "learning_rate": 6.857961938576573e-05,
      "loss": 0.935,
      "step": 295
    },
    {
      "epoch": 1.6152796725784446,
      "grad_norm": 1.954298138706908,
      "learning_rate": 6.84725827435988e-05,
      "loss": 0.9281,
      "step": 296
    },
    {
      "epoch": 1.620736698499318,
      "grad_norm": 1.5596400465604372,
      "learning_rate": 6.83651312176948e-05,
      "loss": 0.9291,
      "step": 297
    },
    {
      "epoch": 1.626193724420191,
      "grad_norm": 1.748314687412271,
      "learning_rate": 6.825726637376669e-05,
      "loss": 0.9491,
      "step": 298
    },
    {
      "epoch": 1.631650750341064,
      "grad_norm": 1.5797658678155715,
      "learning_rate": 6.814898978354997e-05,
      "loss": 0.9409,
      "step": 299
    },
    {
      "epoch": 1.6371077762619373,
      "grad_norm": 1.433746877281443,
      "learning_rate": 6.804030302477985e-05,
      "loss": 0.9199,
      "step": 300
    },
    {
      "epoch": 1.6425648021828103,
      "grad_norm": 1.581155799885627,
      "learning_rate": 6.793120768116825e-05,
      "loss": 0.923,
      "step": 301
    },
    {
      "epoch": 1.6480218281036834,
      "grad_norm": 2.1317295688148445,
      "learning_rate": 6.782170534238073e-05,
      "loss": 0.9147,
      "step": 302
    },
    {
      "epoch": 1.6534788540245566,
      "grad_norm": 1.3648515934883898,
      "learning_rate": 6.771179760401329e-05,
      "loss": 0.9261,
      "step": 303
    },
    {
      "epoch": 1.65893587994543,
      "grad_norm": 1.3152685343318224,
      "learning_rate": 6.760148606756916e-05,
      "loss": 0.929,
      "step": 304
    },
    {
      "epoch": 1.6643929058663027,
      "grad_norm": 1.5395500820286134,
      "learning_rate": 6.749077234043545e-05,
      "loss": 0.9204,
      "step": 305
    },
    {
      "epoch": 1.669849931787176,
      "grad_norm": 1.2417107057753007,
      "learning_rate": 6.737965803585975e-05,
      "loss": 0.9459,
      "step": 306
    },
    {
      "epoch": 1.6753069577080493,
      "grad_norm": 2.5080349522146173,
      "learning_rate": 6.726814477292654e-05,
      "loss": 0.9374,
      "step": 307
    },
    {
      "epoch": 1.680763983628922,
      "grad_norm": 1.0409667518300874,
      "learning_rate": 6.715623417653372e-05,
      "loss": 0.9348,
      "step": 308
    },
    {
      "epoch": 1.6862210095497954,
      "grad_norm": 1.4159003070105012,
      "learning_rate": 6.704392787736884e-05,
      "loss": 0.9429,
      "step": 309
    },
    {
      "epoch": 1.6916780354706686,
      "grad_norm": 1.4352837307400224,
      "learning_rate": 6.69312275118854e-05,
      "loss": 0.9355,
      "step": 310
    },
    {
      "epoch": 1.6971350613915415,
      "grad_norm": 2.217803527040767,
      "learning_rate": 6.681813472227893e-05,
      "loss": 0.9133,
      "step": 311
    },
    {
      "epoch": 1.7025920873124147,
      "grad_norm": 1.6239667118361827,
      "learning_rate": 6.670465115646313e-05,
      "loss": 0.9605,
      "step": 312
    },
    {
      "epoch": 1.708049113233288,
      "grad_norm": 1.6735982665826579,
      "learning_rate": 6.659077846804586e-05,
      "loss": 0.9508,
      "step": 313
    },
    {
      "epoch": 1.7135061391541608,
      "grad_norm": 1.6595574263156625,
      "learning_rate": 6.647651831630496e-05,
      "loss": 0.9255,
      "step": 314
    },
    {
      "epoch": 1.718963165075034,
      "grad_norm": 2.1924142398412374,
      "learning_rate": 6.63618723661642e-05,
      "loss": 0.9512,
      "step": 315
    },
    {
      "epoch": 1.7244201909959074,
      "grad_norm": 1.1050184457511385,
      "learning_rate": 6.624684228816887e-05,
      "loss": 0.9412,
      "step": 316
    },
    {
      "epoch": 1.7298772169167802,
      "grad_norm": 2.013890918931931,
      "learning_rate": 6.61314297584616e-05,
      "loss": 0.9637,
      "step": 317
    },
    {
      "epoch": 1.7353342428376535,
      "grad_norm": 1.6263566377440777,
      "learning_rate": 6.601563645875779e-05,
      "loss": 0.9604,
      "step": 318
    },
    {
      "epoch": 1.7407912687585267,
      "grad_norm": 1.6785802235449294,
      "learning_rate": 6.58994640763212e-05,
      "loss": 0.9508,
      "step": 319
    },
    {
      "epoch": 1.7462482946793996,
      "grad_norm": 2.332156868426827,
      "learning_rate": 6.578291430393935e-05,
      "loss": 0.943,
      "step": 320
    },
    {
      "epoch": 1.7517053206002728,
      "grad_norm": 1.501192527788078,
      "learning_rate": 6.566598883989879e-05,
      "loss": 0.9405,
      "step": 321
    },
    {
      "epoch": 1.7571623465211461,
      "grad_norm": 3.091321379287327,
      "learning_rate": 6.554868938796044e-05,
      "loss": 0.9403,
      "step": 322
    },
    {
      "epoch": 1.762619372442019,
      "grad_norm": 2.994851374878646,
      "learning_rate": 6.543101765733473e-05,
      "loss": 0.9372,
      "step": 323
    },
    {
      "epoch": 1.7680763983628922,
      "grad_norm": 1.6077026093358928,
      "learning_rate": 6.531297536265668e-05,
      "loss": 0.9329,
      "step": 324
    },
    {
      "epoch": 1.7735334242837655,
      "grad_norm": 1.9574778352925133,
      "learning_rate": 6.519456422396089e-05,
      "loss": 0.9482,
      "step": 325
    },
    {
      "epoch": 1.7789904502046383,
      "grad_norm": 1.4810080659785045,
      "learning_rate": 6.507578596665655e-05,
      "loss": 0.9598,
      "step": 326
    },
    {
      "epoch": 1.7844474761255116,
      "grad_norm": 1.4703841544273384,
      "learning_rate": 6.495664232150227e-05,
      "loss": 0.9413,
      "step": 327
    },
    {
      "epoch": 1.7899045020463848,
      "grad_norm": 1.2925625121336821,
      "learning_rate": 6.483713502458082e-05,
      "loss": 0.9423,
      "step": 328
    },
    {
      "epoch": 1.795361527967258,
      "grad_norm": 1.4691567325910997,
      "learning_rate": 6.471726581727391e-05,
      "loss": 0.9513,
      "step": 329
    },
    {
      "epoch": 1.800818553888131,
      "grad_norm": 1.6492564208989655,
      "learning_rate": 6.459703644623669e-05,
      "loss": 0.9247,
      "step": 330
    },
    {
      "epoch": 1.8062755798090042,
      "grad_norm": 1.8176791992294896,
      "learning_rate": 6.447644866337246e-05,
      "loss": 0.9201,
      "step": 331
    },
    {
      "epoch": 1.8117326057298773,
      "grad_norm": 1.2443659972541816,
      "learning_rate": 6.435550422580703e-05,
      "loss": 0.9428,
      "step": 332
    },
    {
      "epoch": 1.8171896316507503,
      "grad_norm": 0.9585197685327218,
      "learning_rate": 6.423420489586312e-05,
      "loss": 0.9098,
      "step": 333
    },
    {
      "epoch": 1.8226466575716236,
      "grad_norm": 1.9263653384878547,
      "learning_rate": 6.411255244103476e-05,
      "loss": 0.9445,
      "step": 334
    },
    {
      "epoch": 1.8281036834924966,
      "grad_norm": 1.4002772834034205,
      "learning_rate": 6.399054863396147e-05,
      "loss": 0.9407,
      "step": 335
    },
    {
      "epoch": 1.8335607094133697,
      "grad_norm": 2.263455086974397,
      "learning_rate": 6.38681952524024e-05,
      "loss": 0.9391,
      "step": 336
    },
    {
      "epoch": 1.839017735334243,
      "grad_norm": 1.6097342527863918,
      "learning_rate": 6.374549407921052e-05,
      "loss": 0.9373,
      "step": 337
    },
    {
      "epoch": 1.844474761255116,
      "grad_norm": 1.9906900776578618,
      "learning_rate": 6.362244690230658e-05,
      "loss": 0.9317,
      "step": 338
    },
    {
      "epoch": 1.849931787175989,
      "grad_norm": 1.7450912010683954,
      "learning_rate": 6.349905551465302e-05,
      "loss": 0.9338,
      "step": 339
    },
    {
      "epoch": 1.8553888130968623,
      "grad_norm": 1.4836167911447387,
      "learning_rate": 6.3375321714228e-05,
      "loss": 0.946,
      "step": 340
    },
    {
      "epoch": 1.8608458390177354,
      "grad_norm": 1.2615365954555837,
      "learning_rate": 6.325124730399897e-05,
      "loss": 0.94,
      "step": 341
    },
    {
      "epoch": 1.8663028649386084,
      "grad_norm": 1.512046504536024,
      "learning_rate": 6.312683409189659e-05,
      "loss": 0.927,
      "step": 342
    },
    {
      "epoch": 1.8717598908594817,
      "grad_norm": 1.0177814524104347,
      "learning_rate": 6.300208389078834e-05,
      "loss": 0.9239,
      "step": 343
    },
    {
      "epoch": 1.8772169167803547,
      "grad_norm": 2.0413829776753576,
      "learning_rate": 6.2876998518452e-05,
      "loss": 0.9407,
      "step": 344
    },
    {
      "epoch": 1.8826739427012278,
      "grad_norm": 1.9502926154029763,
      "learning_rate": 6.275157979754932e-05,
      "loss": 0.9476,
      "step": 345
    },
    {
      "epoch": 1.888130968622101,
      "grad_norm": 1.1523138049064319,
      "learning_rate": 6.262582955559936e-05,
      "loss": 0.9483,
      "step": 346
    },
    {
      "epoch": 1.893587994542974,
      "grad_norm": 1.4261842624847951,
      "learning_rate": 6.249974962495187e-05,
      "loss": 0.9441,
      "step": 347
    },
    {
      "epoch": 1.8990450204638472,
      "grad_norm": 1.2008850401868307,
      "learning_rate": 6.237334184276066e-05,
      "loss": 0.9143,
      "step": 348
    },
    {
      "epoch": 1.9045020463847204,
      "grad_norm": 2.2130995474869377,
      "learning_rate": 6.224660805095669e-05,
      "loss": 0.9554,
      "step": 349
    },
    {
      "epoch": 1.9099590723055935,
      "grad_norm": 1.2762628629624744,
      "learning_rate": 6.21195500962214e-05,
      "loss": 0.9392,
      "step": 350
    },
    {
      "epoch": 1.9154160982264665,
      "grad_norm": 1.5320175021839157,
      "learning_rate": 6.199216982995966e-05,
      "loss": 0.9311,
      "step": 351
    },
    {
      "epoch": 1.9208731241473398,
      "grad_norm": 1.0910833731453622,
      "learning_rate": 6.18644691082729e-05,
      "loss": 0.9125,
      "step": 352
    },
    {
      "epoch": 1.9263301500682128,
      "grad_norm": 2.3128739475758846,
      "learning_rate": 6.173644979193199e-05,
      "loss": 0.9354,
      "step": 353
    },
    {
      "epoch": 1.931787175989086,
      "grad_norm": 1.525750875087295,
      "learning_rate": 6.160811374635015e-05,
      "loss": 0.9308,
      "step": 354
    },
    {
      "epoch": 1.9372442019099592,
      "grad_norm": 1.1778808648494785,
      "learning_rate": 6.147946284155576e-05,
      "loss": 0.9388,
      "step": 355
    },
    {
      "epoch": 1.9427012278308322,
      "grad_norm": 3.1414998446163396,
      "learning_rate": 6.135049895216516e-05,
      "loss": 0.9208,
      "step": 356
    },
    {
      "epoch": 1.9481582537517053,
      "grad_norm": 2.4226620501686384,
      "learning_rate": 6.122122395735525e-05,
      "loss": 0.9219,
      "step": 357
    },
    {
      "epoch": 1.9536152796725785,
      "grad_norm": 3.2604962642078203,
      "learning_rate": 6.10916397408362e-05,
      "loss": 0.9344,
      "step": 358
    },
    {
      "epoch": 1.9590723055934516,
      "grad_norm": 2.0869000164131326,
      "learning_rate": 6.096174819082389e-05,
      "loss": 0.9495,
      "step": 359
    },
    {
      "epoch": 1.9645293315143246,
      "grad_norm": 3.808496465576956,
      "learning_rate": 6.0831551200012534e-05,
      "loss": 0.9567,
      "step": 360
    },
    {
      "epoch": 1.969986357435198,
      "grad_norm": 3.241680278099208,
      "learning_rate": 6.070105066554698e-05,
      "loss": 0.9117,
      "step": 361
    },
    {
      "epoch": 1.975443383356071,
      "grad_norm": 2.0073822332535984,
      "learning_rate": 6.057024848899515e-05,
      "loss": 0.9246,
      "step": 362
    },
    {
      "epoch": 1.980900409276944,
      "grad_norm": 1.8404335145090698,
      "learning_rate": 6.0439146576320256e-05,
      "loss": 0.9528,
      "step": 363
    },
    {
      "epoch": 1.9863574351978173,
      "grad_norm": 2.4786795274720053,
      "learning_rate": 6.0307746837853087e-05,
      "loss": 0.9327,
      "step": 364
    },
    {
      "epoch": 1.9918144611186903,
      "grad_norm": 1.5081344365906557,
      "learning_rate": 6.017605118826417e-05,
      "loss": 0.9451,
      "step": 365
    },
    {
      "epoch": 1.9972714870395634,
      "grad_norm": 2.5201008232251625,
      "learning_rate": 6.0044061546535815e-05,
      "loss": 0.931,
      "step": 366
    },
    {
      "epoch": 2.0027285129604366,
      "grad_norm": 2.517633847206527,
      "learning_rate": 5.991177983593423e-05,
      "loss": 1.6346,
      "step": 367
    },
    {
      "epoch": 2.00818553888131,
      "grad_norm": 2.525593797167957,
      "learning_rate": 5.9779207983981435e-05,
      "loss": 0.9096,
      "step": 368
    },
    {
      "epoch": 2.0136425648021827,
      "grad_norm": 1.6371289597658272,
      "learning_rate": 5.964634792242723e-05,
      "loss": 0.924,
      "step": 369
    },
    {
      "epoch": 2.019099590723056,
      "grad_norm": 2.3442082909219586,
      "learning_rate": 5.951320158722095e-05,
      "loss": 0.8972,
      "step": 370
    },
    {
      "epoch": 2.0245566166439293,
      "grad_norm": 1.8551133091360161,
      "learning_rate": 5.937977091848341e-05,
      "loss": 0.9081,
      "step": 371
    },
    {
      "epoch": 2.030013642564802,
      "grad_norm": 1.6933903655893108,
      "learning_rate": 5.924605786047847e-05,
      "loss": 0.9227,
      "step": 372
    },
    {
      "epoch": 2.0354706684856754,
      "grad_norm": 1.9844111089596346,
      "learning_rate": 5.911206436158482e-05,
      "loss": 0.9123,
      "step": 373
    },
    {
      "epoch": 2.0409276944065486,
      "grad_norm": 1.0962915420790558,
      "learning_rate": 5.897779237426753e-05,
      "loss": 0.9033,
      "step": 374
    },
    {
      "epoch": 2.0463847203274215,
      "grad_norm": 2.2324391296180934,
      "learning_rate": 5.884324385504966e-05,
      "loss": 0.9129,
      "step": 375
    },
    {
      "epoch": 2.0518417462482947,
      "grad_norm": 1.2807887920016723,
      "learning_rate": 5.870842076448364e-05,
      "loss": 0.9066,
      "step": 376
    },
    {
      "epoch": 2.057298772169168,
      "grad_norm": 1.6112376316686408,
      "learning_rate": 5.857332506712285e-05,
      "loss": 0.9257,
      "step": 377
    },
    {
      "epoch": 2.062755798090041,
      "grad_norm": 1.6753107815242307,
      "learning_rate": 5.843795873149285e-05,
      "loss": 0.933,
      "step": 378
    },
    {
      "epoch": 2.068212824010914,
      "grad_norm": 1.3373949314164593,
      "learning_rate": 5.830232373006283e-05,
      "loss": 0.9306,
      "step": 379
    },
    {
      "epoch": 2.0736698499317874,
      "grad_norm": 1.5344827214716636,
      "learning_rate": 5.816642203921674e-05,
      "loss": 0.8968,
      "step": 380
    },
    {
      "epoch": 2.07912687585266,
      "grad_norm": 1.177900589075632,
      "learning_rate": 5.8030255639224584e-05,
      "loss": 0.885,
      "step": 381
    },
    {
      "epoch": 2.0845839017735335,
      "grad_norm": 1.5511949126090052,
      "learning_rate": 5.789382651421354e-05,
      "loss": 0.9086,
      "step": 382
    },
    {
      "epoch": 2.0900409276944067,
      "grad_norm": 1.3198466527101826,
      "learning_rate": 5.775713665213899e-05,
      "loss": 0.9153,
      "step": 383
    },
    {
      "epoch": 2.0954979536152796,
      "grad_norm": 1.3654261455944285,
      "learning_rate": 5.7620188044755684e-05,
      "loss": 0.9245,
      "step": 384
    },
    {
      "epoch": 2.100954979536153,
      "grad_norm": 1.4121475396535077,
      "learning_rate": 5.748298268758859e-05,
      "loss": 0.9045,
      "step": 385
    },
    {
      "epoch": 2.106412005457026,
      "grad_norm": 1.2226144369638559,
      "learning_rate": 5.734552257990386e-05,
      "loss": 0.9309,
      "step": 386
    },
    {
      "epoch": 2.111869031377899,
      "grad_norm": 1.1754628781979755,
      "learning_rate": 5.720780972467974e-05,
      "loss": 0.9064,
      "step": 387
    },
    {
      "epoch": 2.117326057298772,
      "grad_norm": 1.6352507288221967,
      "learning_rate": 5.7069846128577304e-05,
      "loss": 0.9032,
      "step": 388
    },
    {
      "epoch": 2.1227830832196455,
      "grad_norm": 1.2127070473615575,
      "learning_rate": 5.693163380191127e-05,
      "loss": 0.8939,
      "step": 389
    },
    {
      "epoch": 2.1282401091405183,
      "grad_norm": 1.2698085787934459,
      "learning_rate": 5.679317475862073e-05,
      "loss": 0.9115,
      "step": 390
    },
    {
      "epoch": 2.1336971350613916,
      "grad_norm": 1.0503429136187135,
      "learning_rate": 5.665447101623971e-05,
      "loss": 0.89,
      "step": 391
    },
    {
      "epoch": 2.139154160982265,
      "grad_norm": 1.2859700958716116,
      "learning_rate": 5.6515524595867855e-05,
      "loss": 0.9136,
      "step": 392
    },
    {
      "epoch": 2.1446111869031377,
      "grad_norm": 1.8523331576107729,
      "learning_rate": 5.6376337522141e-05,
      "loss": 0.8926,
      "step": 393
    },
    {
      "epoch": 2.150068212824011,
      "grad_norm": 0.9691735187094292,
      "learning_rate": 5.623691182320153e-05,
      "loss": 0.9126,
      "step": 394
    },
    {
      "epoch": 2.155525238744884,
      "grad_norm": 1.2395540253844441,
      "learning_rate": 5.6097249530669004e-05,
      "loss": 0.9189,
      "step": 395
    },
    {
      "epoch": 2.160982264665757,
      "grad_norm": 0.9624385840728625,
      "learning_rate": 5.5957352679610417e-05,
      "loss": 0.9054,
      "step": 396
    },
    {
      "epoch": 2.1664392905866303,
      "grad_norm": 1.512630856479847,
      "learning_rate": 5.58172233085106e-05,
      "loss": 0.8978,
      "step": 397
    },
    {
      "epoch": 2.1718963165075036,
      "grad_norm": 1.3726882811029215,
      "learning_rate": 5.567686345924256e-05,
      "loss": 0.9254,
      "step": 398
    },
    {
      "epoch": 2.1773533424283764,
      "grad_norm": 1.514528048915799,
      "learning_rate": 5.5536275177037615e-05,
      "loss": 0.8953,
      "step": 399
    },
    {
      "epoch": 2.1828103683492497,
      "grad_norm": 0.9854058808309464,
      "learning_rate": 5.53954605104557e-05,
      "loss": 0.8994,
      "step": 400
    },
    {
      "epoch": 2.188267394270123,
      "grad_norm": 1.3982634603651707,
      "learning_rate": 5.5254421511355466e-05,
      "loss": 0.9191,
      "step": 401
    },
    {
      "epoch": 2.193724420190996,
      "grad_norm": 0.945935420435088,
      "learning_rate": 5.511316023486437e-05,
      "loss": 0.8954,
      "step": 402
    },
    {
      "epoch": 2.199181446111869,
      "grad_norm": 1.7995542419322705,
      "learning_rate": 5.497167873934877e-05,
      "loss": 0.912,
      "step": 403
    },
    {
      "epoch": 2.2046384720327423,
      "grad_norm": 1.4321355586507285,
      "learning_rate": 5.482997908638393e-05,
      "loss": 0.8979,
      "step": 404
    },
    {
      "epoch": 2.210095497953615,
      "grad_norm": 0.9196538171557571,
      "learning_rate": 5.46880633407239e-05,
      "loss": 0.895,
      "step": 405
    },
    {
      "epoch": 2.2155525238744884,
      "grad_norm": 1.5548446371415476,
      "learning_rate": 5.454593357027154e-05,
      "loss": 0.913,
      "step": 406
    },
    {
      "epoch": 2.2210095497953617,
      "grad_norm": 1.657041068857267,
      "learning_rate": 5.440359184604834e-05,
      "loss": 0.9279,
      "step": 407
    },
    {
      "epoch": 2.2264665757162345,
      "grad_norm": 1.0626769214920089,
      "learning_rate": 5.42610402421642e-05,
      "loss": 0.893,
      "step": 408
    },
    {
      "epoch": 2.231923601637108,
      "grad_norm": 1.5408374648839536,
      "learning_rate": 5.411828083578729e-05,
      "loss": 0.9044,
      "step": 409
    },
    {
      "epoch": 2.237380627557981,
      "grad_norm": 1.05976554269391,
      "learning_rate": 5.397531570711373e-05,
      "loss": 0.8687,
      "step": 410
    },
    {
      "epoch": 2.242837653478854,
      "grad_norm": 1.8958984234334453,
      "learning_rate": 5.3832146939337267e-05,
      "loss": 0.9318,
      "step": 411
    },
    {
      "epoch": 2.248294679399727,
      "grad_norm": 1.3494319992439094,
      "learning_rate": 5.368877661861899e-05,
      "loss": 0.9235,
      "step": 412
    },
    {
      "epoch": 2.2537517053206004,
      "grad_norm": 1.4368846046733357,
      "learning_rate": 5.3545206834056816e-05,
      "loss": 0.9123,
      "step": 413
    },
    {
      "epoch": 2.2592087312414733,
      "grad_norm": 1.8256579203781316,
      "learning_rate": 5.340143967765519e-05,
      "loss": 0.9035,
      "step": 414
    },
    {
      "epoch": 2.2646657571623465,
      "grad_norm": 1.3166251617751332,
      "learning_rate": 5.325747724429448e-05,
      "loss": 0.9132,
      "step": 415
    },
    {
      "epoch": 2.27012278308322,
      "grad_norm": 1.5644479487633818,
      "learning_rate": 5.311332163170054e-05,
      "loss": 0.8882,
      "step": 416
    },
    {
      "epoch": 2.2755798090040926,
      "grad_norm": 1.7730325855876201,
      "learning_rate": 5.296897494041407e-05,
      "loss": 0.9096,
      "step": 417
    },
    {
      "epoch": 2.281036834924966,
      "grad_norm": 0.690579518867841,
      "learning_rate": 5.282443927376004e-05,
      "loss": 0.9008,
      "step": 418
    },
    {
      "epoch": 2.286493860845839,
      "grad_norm": 1.525525192362239,
      "learning_rate": 5.267971673781708e-05,
      "loss": 0.9091,
      "step": 419
    },
    {
      "epoch": 2.291950886766712,
      "grad_norm": 1.1450930632937755,
      "learning_rate": 5.253480944138675e-05,
      "loss": 0.9077,
      "step": 420
    },
    {
      "epoch": 2.2974079126875853,
      "grad_norm": 1.5853135303829244,
      "learning_rate": 5.238971949596277e-05,
      "loss": 0.9076,
      "step": 421
    },
    {
      "epoch": 2.3028649386084585,
      "grad_norm": 1.0964682564769086,
      "learning_rate": 5.2244449015700364e-05,
      "loss": 0.9065,
      "step": 422
    },
    {
      "epoch": 2.3083219645293314,
      "grad_norm": 0.9213215067007345,
      "learning_rate": 5.209900011738536e-05,
      "loss": 0.9075,
      "step": 423
    },
    {
      "epoch": 2.3137789904502046,
      "grad_norm": 0.9574566792724861,
      "learning_rate": 5.1953374920403367e-05,
      "loss": 0.8995,
      "step": 424
    },
    {
      "epoch": 2.319236016371078,
      "grad_norm": 1.4073999493333642,
      "learning_rate": 5.1807575546708925e-05,
      "loss": 0.9051,
      "step": 425
    },
    {
      "epoch": 2.3246930422919507,
      "grad_norm": 1.5239194175384891,
      "learning_rate": 5.166160412079456e-05,
      "loss": 0.8978,
      "step": 426
    },
    {
      "epoch": 2.330150068212824,
      "grad_norm": 0.5990721338816578,
      "learning_rate": 5.15154627696598e-05,
      "loss": 0.9289,
      "step": 427
    },
    {
      "epoch": 2.3356070941336973,
      "grad_norm": 0.9305075248801276,
      "learning_rate": 5.136915362278025e-05,
      "loss": 0.9002,
      "step": 428
    },
    {
      "epoch": 2.34106412005457,
      "grad_norm": 1.159552986565052,
      "learning_rate": 5.1222678812076514e-05,
      "loss": 0.8991,
      "step": 429
    },
    {
      "epoch": 2.3465211459754434,
      "grad_norm": 1.544608318550361,
      "learning_rate": 5.1076040471883105e-05,
      "loss": 0.9041,
      "step": 430
    },
    {
      "epoch": 2.3519781718963166,
      "grad_norm": 0.9906368417277753,
      "learning_rate": 5.092924073891745e-05,
      "loss": 0.9056,
      "step": 431
    },
    {
      "epoch": 2.3574351978171895,
      "grad_norm": 1.2503392254533339,
      "learning_rate": 5.078228175224861e-05,
      "loss": 0.9071,
      "step": 432
    },
    {
      "epoch": 2.3628922237380627,
      "grad_norm": 1.0117345420422286,
      "learning_rate": 5.063516565326622e-05,
      "loss": 0.9007,
      "step": 433
    },
    {
      "epoch": 2.368349249658936,
      "grad_norm": 1.5307989423567165,
      "learning_rate": 5.048789458564928e-05,
      "loss": 0.9082,
      "step": 434
    },
    {
      "epoch": 2.373806275579809,
      "grad_norm": 1.3272756265512895,
      "learning_rate": 5.034047069533485e-05,
      "loss": 0.9169,
      "step": 435
    },
    {
      "epoch": 2.379263301500682,
      "grad_norm": 0.9338875905847049,
      "learning_rate": 5.019289613048683e-05,
      "loss": 0.908,
      "step": 436
    },
    {
      "epoch": 2.3847203274215554,
      "grad_norm": 1.0158152693801568,
      "learning_rate": 5.004517304146467e-05,
      "loss": 0.8957,
      "step": 437
    },
    {
      "epoch": 2.390177353342428,
      "grad_norm": 1.535254150172697,
      "learning_rate": 4.989730358079198e-05,
      "loss": 0.8862,
      "step": 438
    },
    {
      "epoch": 2.3956343792633015,
      "grad_norm": 1.0474435875199142,
      "learning_rate": 4.9749289903125236e-05,
      "loss": 0.8882,
      "step": 439
    },
    {
      "epoch": 2.4010914051841747,
      "grad_norm": 1.5467304860312037,
      "learning_rate": 4.960113416522233e-05,
      "loss": 0.9022,
      "step": 440
    },
    {
      "epoch": 2.4065484311050476,
      "grad_norm": 0.7171663599148432,
      "learning_rate": 4.945283852591119e-05,
      "loss": 0.8952,
      "step": 441
    },
    {
      "epoch": 2.412005457025921,
      "grad_norm": 1.184417711535613,
      "learning_rate": 4.930440514605826e-05,
      "loss": 0.9065,
      "step": 442
    },
    {
      "epoch": 2.417462482946794,
      "grad_norm": 1.204723229806714,
      "learning_rate": 4.9155836188537045e-05,
      "loss": 0.909,
      "step": 443
    },
    {
      "epoch": 2.422919508867667,
      "grad_norm": 1.2380345749344617,
      "learning_rate": 4.9007133818196636e-05,
      "loss": 0.8951,
      "step": 444
    },
    {
      "epoch": 2.42837653478854,
      "grad_norm": 1.265693694540876,
      "learning_rate": 4.885830020183011e-05,
      "loss": 0.9181,
      "step": 445
    },
    {
      "epoch": 2.4338335607094135,
      "grad_norm": 1.099951391116176,
      "learning_rate": 4.870933750814296e-05,
      "loss": 0.9179,
      "step": 446
    },
    {
      "epoch": 2.4392905866302863,
      "grad_norm": 1.2414338650044854,
      "learning_rate": 4.8560247907721535e-05,
      "loss": 0.886,
      "step": 447
    },
    {
      "epoch": 2.4447476125511596,
      "grad_norm": 1.5704709003561375,
      "learning_rate": 4.841103357300137e-05,
      "loss": 0.9175,
      "step": 448
    },
    {
      "epoch": 2.450204638472033,
      "grad_norm": 0.534836475556195,
      "learning_rate": 4.826169667823553e-05,
      "loss": 0.9199,
      "step": 449
    },
    {
      "epoch": 2.4556616643929057,
      "grad_norm": 0.8776370257962468,
      "learning_rate": 4.811223939946299e-05,
      "loss": 0.8944,
      "step": 450
    },
    {
      "epoch": 2.461118690313779,
      "grad_norm": 1.9706453577764773,
      "learning_rate": 4.796266391447683e-05,
      "loss": 0.9127,
      "step": 451
    },
    {
      "epoch": 2.466575716234652,
      "grad_norm": 0.8249118962292379,
      "learning_rate": 4.7812972402792574e-05,
      "loss": 0.9089,
      "step": 452
    },
    {
      "epoch": 2.472032742155525,
      "grad_norm": 1.7629114368166088,
      "learning_rate": 4.766316704561645e-05,
      "loss": 0.9111,
      "step": 453
    },
    {
      "epoch": 2.4774897680763983,
      "grad_norm": 1.4811036485248286,
      "learning_rate": 4.75132500258135e-05,
      "loss": 0.9073,
      "step": 454
    },
    {
      "epoch": 2.4829467939972716,
      "grad_norm": 1.1161608577851434,
      "learning_rate": 4.736322352787586e-05,
      "loss": 0.9044,
      "step": 455
    },
    {
      "epoch": 2.488403819918145,
      "grad_norm": 1.2828229918284093,
      "learning_rate": 4.721308973789096e-05,
      "loss": 0.9,
      "step": 456
    },
    {
      "epoch": 2.4938608458390177,
      "grad_norm": 1.517713518373473,
      "learning_rate": 4.7062850843509564e-05,
      "loss": 0.9027,
      "step": 457
    },
    {
      "epoch": 2.499317871759891,
      "grad_norm": 1.3348391079469917,
      "learning_rate": 4.6912509033913945e-05,
      "loss": 0.9195,
      "step": 458
    },
    {
      "epoch": 2.504774897680764,
      "grad_norm": 1.245752982162094,
      "learning_rate": 4.6762066499786015e-05,
      "loss": 0.9243,
      "step": 459
    },
    {
      "epoch": 2.510231923601637,
      "grad_norm": 1.1304452292363005,
      "learning_rate": 4.661152543327539e-05,
      "loss": 0.9053,
      "step": 460
    },
    {
      "epoch": 2.5156889495225103,
      "grad_norm": 1.5152505332881556,
      "learning_rate": 4.64608880279674e-05,
      "loss": 0.9025,
      "step": 461
    },
    {
      "epoch": 2.5211459754433836,
      "grad_norm": 0.6895652709575827,
      "learning_rate": 4.631015647885118e-05,
      "loss": 0.877,
      "step": 462
    },
    {
      "epoch": 2.5266030013642564,
      "grad_norm": 1.0031211901568315,
      "learning_rate": 4.6159332982287654e-05,
      "loss": 0.9165,
      "step": 463
    },
    {
      "epoch": 2.5320600272851297,
      "grad_norm": 0.9856933305068003,
      "learning_rate": 4.6008419735977567e-05,
      "loss": 0.8947,
      "step": 464
    },
    {
      "epoch": 2.5375170532060025,
      "grad_norm": 1.4074395011531378,
      "learning_rate": 4.585741893892941e-05,
      "loss": 0.9031,
      "step": 465
    },
    {
      "epoch": 2.542974079126876,
      "grad_norm": 1.0947772923264532,
      "learning_rate": 4.570633279142743e-05,
      "loss": 0.895,
      "step": 466
    },
    {
      "epoch": 2.548431105047749,
      "grad_norm": 0.9954166963630334,
      "learning_rate": 4.5555163494999535e-05,
      "loss": 0.9155,
      "step": 467
    },
    {
      "epoch": 2.5538881309686223,
      "grad_norm": 1.0252150819212769,
      "learning_rate": 4.5403913252385206e-05,
      "loss": 0.8998,
      "step": 468
    },
    {
      "epoch": 2.559345156889495,
      "grad_norm": 0.5854437882911242,
      "learning_rate": 4.525258426750346e-05,
      "loss": 0.9013,
      "step": 469
    },
    {
      "epoch": 2.5648021828103684,
      "grad_norm": 0.7840815488236171,
      "learning_rate": 4.510117874542064e-05,
      "loss": 0.8883,
      "step": 470
    },
    {
      "epoch": 2.5702592087312413,
      "grad_norm": 1.0631188620110363,
      "learning_rate": 4.494969889231839e-05,
      "loss": 0.89,
      "step": 471
    },
    {
      "epoch": 2.5757162346521145,
      "grad_norm": 0.6218999930771435,
      "learning_rate": 4.4798146915461446e-05,
      "loss": 0.8793,
      "step": 472
    },
    {
      "epoch": 2.581173260572988,
      "grad_norm": 0.8064002493767205,
      "learning_rate": 4.464652502316545e-05,
      "loss": 0.9076,
      "step": 473
    },
    {
      "epoch": 2.586630286493861,
      "grad_norm": 0.8565103993043336,
      "learning_rate": 4.4494835424764846e-05,
      "loss": 0.8928,
      "step": 474
    },
    {
      "epoch": 2.592087312414734,
      "grad_norm": 0.8227008226865807,
      "learning_rate": 4.434308033058062e-05,
      "loss": 0.9115,
      "step": 475
    },
    {
      "epoch": 2.597544338335607,
      "grad_norm": 0.552330723413834,
      "learning_rate": 4.4191261951888137e-05,
      "loss": 0.905,
      "step": 476
    },
    {
      "epoch": 2.60300136425648,
      "grad_norm": 0.9160367599695006,
      "learning_rate": 4.4039382500884884e-05,
      "loss": 0.8907,
      "step": 477
    },
    {
      "epoch": 2.6084583901773533,
      "grad_norm": 1.2743846931315213,
      "learning_rate": 4.3887444190658276e-05,
      "loss": 0.8979,
      "step": 478
    },
    {
      "epoch": 2.6139154160982265,
      "grad_norm": 1.0391301799339259,
      "learning_rate": 4.373544923515337e-05,
      "loss": 0.9025,
      "step": 479
    },
    {
      "epoch": 2.6193724420191,
      "grad_norm": 1.3713321504465736,
      "learning_rate": 4.358339984914063e-05,
      "loss": 0.8801,
      "step": 480
    },
    {
      "epoch": 2.6248294679399726,
      "grad_norm": 0.8066646654611648,
      "learning_rate": 4.343129824818363e-05,
      "loss": 0.9041,
      "step": 481
    },
    {
      "epoch": 2.630286493860846,
      "grad_norm": 0.7832853640483513,
      "learning_rate": 4.32791466486068e-05,
      "loss": 0.9203,
      "step": 482
    },
    {
      "epoch": 2.6357435197817187,
      "grad_norm": 0.7057444529833893,
      "learning_rate": 4.312694726746312e-05,
      "loss": 0.9049,
      "step": 483
    },
    {
      "epoch": 2.641200545702592,
      "grad_norm": 0.6493100641693694,
      "learning_rate": 4.29747023225018e-05,
      "loss": 0.8992,
      "step": 484
    },
    {
      "epoch": 2.6466575716234653,
      "grad_norm": 0.6608005917805958,
      "learning_rate": 4.282241403213598e-05,
      "loss": 0.8936,
      "step": 485
    },
    {
      "epoch": 2.6521145975443385,
      "grad_norm": 0.9967757502926118,
      "learning_rate": 4.267008461541041e-05,
      "loss": 0.8957,
      "step": 486
    },
    {
      "epoch": 2.6575716234652114,
      "grad_norm": 1.4601123062706318,
      "learning_rate": 4.251771629196908e-05,
      "loss": 0.8933,
      "step": 487
    },
    {
      "epoch": 2.6630286493860846,
      "grad_norm": 0.8449635915962943,
      "learning_rate": 4.2365311282022936e-05,
      "loss": 0.897,
      "step": 488
    },
    {
      "epoch": 2.6684856753069575,
      "grad_norm": 0.796727650357652,
      "learning_rate": 4.221287180631747e-05,
      "loss": 0.8943,
      "step": 489
    },
    {
      "epoch": 2.6739427012278307,
      "grad_norm": 1.064925467642507,
      "learning_rate": 4.20604000861004e-05,
      "loss": 0.9028,
      "step": 490
    },
    {
      "epoch": 2.679399727148704,
      "grad_norm": 1.8132030562855181,
      "learning_rate": 4.190789834308929e-05,
      "loss": 0.8899,
      "step": 491
    },
    {
      "epoch": 2.6848567530695773,
      "grad_norm": 0.6460897975291272,
      "learning_rate": 4.175536879943919e-05,
      "loss": 0.888,
      "step": 492
    },
    {
      "epoch": 2.69031377899045,
      "grad_norm": 2.32226453518839,
      "learning_rate": 4.160281367771019e-05,
      "loss": 0.9041,
      "step": 493
    },
    {
      "epoch": 2.6957708049113234,
      "grad_norm": 1.074926776476922,
      "learning_rate": 4.1450235200835145e-05,
      "loss": 0.9112,
      "step": 494
    },
    {
      "epoch": 2.701227830832196,
      "grad_norm": 2.988265988173717,
      "learning_rate": 4.1297635592087196e-05,
      "loss": 0.9137,
      "step": 495
    },
    {
      "epoch": 2.7066848567530695,
      "grad_norm": 2.748929837066849,
      "learning_rate": 4.114501707504741e-05,
      "loss": 0.8964,
      "step": 496
    },
    {
      "epoch": 2.7121418826739427,
      "grad_norm": 1.3745356070302501,
      "learning_rate": 4.099238187357239e-05,
      "loss": 0.9038,
      "step": 497
    },
    {
      "epoch": 2.717598908594816,
      "grad_norm": 1.5677671965351183,
      "learning_rate": 4.083973221176182e-05,
      "loss": 0.9294,
      "step": 498
    },
    {
      "epoch": 2.723055934515689,
      "grad_norm": 1.573723869725014,
      "learning_rate": 4.0687070313926114e-05,
      "loss": 0.919,
      "step": 499
    },
    {
      "epoch": 2.728512960436562,
      "grad_norm": 0.9328025105721132,
      "learning_rate": 4.053439840455398e-05,
      "loss": 0.8881,
      "step": 500
    },
    {
      "epoch": 2.733969986357435,
      "grad_norm": 1.3306299787288607,
      "learning_rate": 4.038171870827999e-05,
      "loss": 0.9194,
      "step": 501
    },
    {
      "epoch": 2.739427012278308,
      "grad_norm": 1.0631582399329886,
      "learning_rate": 4.0229033449852206e-05,
      "loss": 0.9342,
      "step": 502
    },
    {
      "epoch": 2.7448840381991815,
      "grad_norm": 1.4745520296021233,
      "learning_rate": 4.0076344854099724e-05,
      "loss": 0.9026,
      "step": 503
    },
    {
      "epoch": 2.7503410641200547,
      "grad_norm": 1.0276507362237068,
      "learning_rate": 3.9923655145900275e-05,
      "loss": 0.9064,
      "step": 504
    },
    {
      "epoch": 2.7557980900409276,
      "grad_norm": 1.4421360481508985,
      "learning_rate": 3.97709665501478e-05,
      "loss": 0.9256,
      "step": 505
    },
    {
      "epoch": 2.761255115961801,
      "grad_norm": 1.324734470431852,
      "learning_rate": 3.9618281291720025e-05,
      "loss": 0.884,
      "step": 506
    },
    {
      "epoch": 2.7667121418826737,
      "grad_norm": 0.7620830957680698,
      "learning_rate": 3.9465601595446035e-05,
      "loss": 0.9004,
      "step": 507
    },
    {
      "epoch": 2.772169167803547,
      "grad_norm": 1.4595376240477347,
      "learning_rate": 3.93129296860739e-05,
      "loss": 0.911,
      "step": 508
    },
    {
      "epoch": 2.77762619372442,
      "grad_norm": 1.0958345521160193,
      "learning_rate": 3.9160267788238196e-05,
      "loss": 0.9113,
      "step": 509
    },
    {
      "epoch": 2.7830832196452935,
      "grad_norm": 0.9499614564643925,
      "learning_rate": 3.9007618126427614e-05,
      "loss": 0.8727,
      "step": 510
    },
    {
      "epoch": 2.7885402455661663,
      "grad_norm": 1.6986030669586536,
      "learning_rate": 3.885498292495259e-05,
      "loss": 0.9059,
      "step": 511
    },
    {
      "epoch": 2.7939972714870396,
      "grad_norm": 0.8253935525812467,
      "learning_rate": 3.870236440791281e-05,
      "loss": 0.8994,
      "step": 512
    },
    {
      "epoch": 2.799454297407913,
      "grad_norm": 1.62017400053124,
      "learning_rate": 3.854976479916486e-05,
      "loss": 0.8818,
      "step": 513
    },
    {
      "epoch": 2.8049113233287857,
      "grad_norm": 1.0295253405729237,
      "learning_rate": 3.839718632228982e-05,
      "loss": 0.8977,
      "step": 514
    },
    {
      "epoch": 2.810368349249659,
      "grad_norm": 1.3655909260152348,
      "learning_rate": 3.8244631200560835e-05,
      "loss": 0.9078,
      "step": 515
    },
    {
      "epoch": 2.815825375170532,
      "grad_norm": 1.0954541627416656,
      "learning_rate": 3.809210165691072e-05,
      "loss": 0.8977,
      "step": 516
    },
    {
      "epoch": 2.821282401091405,
      "grad_norm": 1.1586044724363185,
      "learning_rate": 3.79395999138996e-05,
      "loss": 0.9056,
      "step": 517
    },
    {
      "epoch": 2.8267394270122783,
      "grad_norm": 0.8829661993056394,
      "learning_rate": 3.7787128193682546e-05,
      "loss": 0.8965,
      "step": 518
    },
    {
      "epoch": 2.8321964529331516,
      "grad_norm": 0.8815798999675472,
      "learning_rate": 3.763468871797708e-05,
      "loss": 0.9197,
      "step": 519
    },
    {
      "epoch": 2.8376534788540244,
      "grad_norm": 1.2538953598040594,
      "learning_rate": 3.7482283708030936e-05,
      "loss": 0.8774,
      "step": 520
    },
    {
      "epoch": 2.8431105047748977,
      "grad_norm": 0.728098507307315,
      "learning_rate": 3.7329915384589615e-05,
      "loss": 0.9103,
      "step": 521
    },
    {
      "epoch": 2.848567530695771,
      "grad_norm": 1.1109117113371303,
      "learning_rate": 3.717758596786404e-05,
      "loss": 0.9196,
      "step": 522
    },
    {
      "epoch": 2.854024556616644,
      "grad_norm": 1.0718036954886652,
      "learning_rate": 3.702529767749821e-05,
      "loss": 0.8963,
      "step": 523
    },
    {
      "epoch": 2.859481582537517,
      "grad_norm": 1.048803705132213,
      "learning_rate": 3.687305273253689e-05,
      "loss": 0.8935,
      "step": 524
    },
    {
      "epoch": 2.8649386084583903,
      "grad_norm": 0.8887966210033448,
      "learning_rate": 3.672085335139321e-05,
      "loss": 0.9014,
      "step": 525
    },
    {
      "epoch": 2.870395634379263,
      "grad_norm": 0.7017332784967888,
      "learning_rate": 3.656870175181638e-05,
      "loss": 0.9019,
      "step": 526
    },
    {
      "epoch": 2.8758526603001364,
      "grad_norm": 0.8896432018913878,
      "learning_rate": 3.6416600150859386e-05,
      "loss": 0.9178,
      "step": 527
    },
    {
      "epoch": 2.8813096862210097,
      "grad_norm": 0.7645003604259587,
      "learning_rate": 3.626455076484665e-05,
      "loss": 0.8967,
      "step": 528
    },
    {
      "epoch": 2.8867667121418825,
      "grad_norm": 1.2198348096547293,
      "learning_rate": 3.611255580934173e-05,
      "loss": 0.8818,
      "step": 529
    },
    {
      "epoch": 2.892223738062756,
      "grad_norm": 0.6151717260376284,
      "learning_rate": 3.596061749911512e-05,
      "loss": 0.8922,
      "step": 530
    },
    {
      "epoch": 2.897680763983629,
      "grad_norm": 0.815800617981215,
      "learning_rate": 3.5808738048111884e-05,
      "loss": 0.9042,
      "step": 531
    },
    {
      "epoch": 2.903137789904502,
      "grad_norm": 0.8465415157806591,
      "learning_rate": 3.565691966941939e-05,
      "loss": 0.9033,
      "step": 532
    },
    {
      "epoch": 2.908594815825375,
      "grad_norm": 0.9997567059531529,
      "learning_rate": 3.550516457523517e-05,
      "loss": 0.8865,
      "step": 533
    },
    {
      "epoch": 2.9140518417462484,
      "grad_norm": 0.9200304101564798,
      "learning_rate": 3.535347497683456e-05,
      "loss": 0.8817,
      "step": 534
    },
    {
      "epoch": 2.9195088676671213,
      "grad_norm": 0.8027379053061545,
      "learning_rate": 3.5201853084538554e-05,
      "loss": 0.9084,
      "step": 535
    },
    {
      "epoch": 2.9249658935879945,
      "grad_norm": 0.5469780406914913,
      "learning_rate": 3.505030110768161e-05,
      "loss": 0.8802,
      "step": 536
    },
    {
      "epoch": 2.930422919508868,
      "grad_norm": 0.7413900149381596,
      "learning_rate": 3.4898821254579374e-05,
      "loss": 0.903,
      "step": 537
    },
    {
      "epoch": 2.9358799454297406,
      "grad_norm": 0.8762727718053043,
      "learning_rate": 3.4747415732496556e-05,
      "loss": 0.9188,
      "step": 538
    },
    {
      "epoch": 2.941336971350614,
      "grad_norm": 0.9396772677381491,
      "learning_rate": 3.459608674761481e-05,
      "loss": 0.901,
      "step": 539
    },
    {
      "epoch": 2.946793997271487,
      "grad_norm": 0.6680089017701383,
      "learning_rate": 3.4444836505000485e-05,
      "loss": 0.8923,
      "step": 540
    },
    {
      "epoch": 2.9522510231923604,
      "grad_norm": 0.47043991279916797,
      "learning_rate": 3.429366720857259e-05,
      "loss": 0.88,
      "step": 541
    },
    {
      "epoch": 2.9577080491132333,
      "grad_norm": 0.4955656876316972,
      "learning_rate": 3.4142581061070596e-05,
      "loss": 0.8939,
      "step": 542
    },
    {
      "epoch": 2.9631650750341065,
      "grad_norm": 0.8318305087447663,
      "learning_rate": 3.399158026402245e-05,
      "loss": 0.9006,
      "step": 543
    },
    {
      "epoch": 2.9686221009549794,
      "grad_norm": 0.9724940321029891,
      "learning_rate": 3.384066701771235e-05,
      "loss": 0.8856,
      "step": 544
    },
    {
      "epoch": 2.9740791268758526,
      "grad_norm": 0.6009045145969995,
      "learning_rate": 3.368984352114883e-05,
      "loss": 0.9143,
      "step": 545
    },
    {
      "epoch": 2.979536152796726,
      "grad_norm": 0.39363538472622567,
      "learning_rate": 3.353911197203262e-05,
      "loss": 0.8902,
      "step": 546
    },
    {
      "epoch": 2.984993178717599,
      "grad_norm": 0.5286633463839729,
      "learning_rate": 3.338847456672463e-05,
      "loss": 0.8895,
      "step": 547
    },
    {
      "epoch": 2.990450204638472,
      "grad_norm": 0.7531125160359047,
      "learning_rate": 3.323793350021399e-05,
      "loss": 0.9012,
      "step": 548
    },
    {
      "epoch": 2.9959072305593453,
      "grad_norm": 0.7249830120473333,
      "learning_rate": 3.3087490966086075e-05,
      "loss": 0.8936,
      "step": 549
    },
    {
      "epoch": 3.001364256480218,
      "grad_norm": 0.7613973707952983,
      "learning_rate": 3.293714915649045e-05,
      "loss": 1.5762,
      "step": 550
    },
    {
      "epoch": 3.0068212824010914,
      "grad_norm": 0.5390667663982144,
      "learning_rate": 3.278691026210904e-05,
      "loss": 0.86,
      "step": 551
    },
    {
      "epoch": 3.0122783083219646,
      "grad_norm": 0.4441457511204983,
      "learning_rate": 3.2636776472124144e-05,
      "loss": 0.8687,
      "step": 552
    },
    {
      "epoch": 3.0177353342428375,
      "grad_norm": 0.547857567525407,
      "learning_rate": 3.248674997418652e-05,
      "loss": 0.8613,
      "step": 553
    },
    {
      "epoch": 3.0231923601637107,
      "grad_norm": 0.5080127949107659,
      "learning_rate": 3.233683295438356e-05,
      "loss": 0.8634,
      "step": 554
    },
    {
      "epoch": 3.028649386084584,
      "grad_norm": 0.4627457990776784,
      "learning_rate": 3.2187027597207426e-05,
      "loss": 0.8707,
      "step": 555
    },
    {
      "epoch": 3.034106412005457,
      "grad_norm": 0.4908817778767444,
      "learning_rate": 3.2037336085523186e-05,
      "loss": 0.8597,
      "step": 556
    },
    {
      "epoch": 3.03956343792633,
      "grad_norm": 0.5198441754829152,
      "learning_rate": 3.188776060053702e-05,
      "loss": 0.8431,
      "step": 557
    },
    {
      "epoch": 3.0450204638472034,
      "grad_norm": 0.4818792863535078,
      "learning_rate": 3.173830332176448e-05,
      "loss": 0.8809,
      "step": 558
    },
    {
      "epoch": 3.050477489768076,
      "grad_norm": 0.44582198493157366,
      "learning_rate": 3.158896642699865e-05,
      "loss": 0.8648,
      "step": 559
    },
    {
      "epoch": 3.0559345156889495,
      "grad_norm": 0.33617908042386857,
      "learning_rate": 3.1439752092278485e-05,
      "loss": 0.8884,
      "step": 560
    },
    {
      "epoch": 3.0613915416098227,
      "grad_norm": 0.5488599072908693,
      "learning_rate": 3.129066249185704e-05,
      "loss": 0.8681,
      "step": 561
    },
    {
      "epoch": 3.0668485675306956,
      "grad_norm": 0.573743615074981,
      "learning_rate": 3.11416997981699e-05,
      "loss": 0.8716,
      "step": 562
    },
    {
      "epoch": 3.072305593451569,
      "grad_norm": 0.3440741022604367,
      "learning_rate": 3.099286618180337e-05,
      "loss": 0.8685,
      "step": 563
    },
    {
      "epoch": 3.077762619372442,
      "grad_norm": 0.48263781958172847,
      "learning_rate": 3.084416381146297e-05,
      "loss": 0.8612,
      "step": 564
    },
    {
      "epoch": 3.083219645293315,
      "grad_norm": 0.5341856968582221,
      "learning_rate": 3.069559485394176e-05,
      "loss": 0.8906,
      "step": 565
    },
    {
      "epoch": 3.088676671214188,
      "grad_norm": 0.44955890178043545,
      "learning_rate": 3.054716147408883e-05,
      "loss": 0.8647,
      "step": 566
    },
    {
      "epoch": 3.0941336971350615,
      "grad_norm": 0.24528391692821006,
      "learning_rate": 3.039886583477767e-05,
      "loss": 0.8734,
      "step": 567
    },
    {
      "epoch": 3.0995907230559343,
      "grad_norm": 0.3811275418491481,
      "learning_rate": 3.0250710096874778e-05,
      "loss": 0.8571,
      "step": 568
    },
    {
      "epoch": 3.1050477489768076,
      "grad_norm": 0.4556061152291196,
      "learning_rate": 3.0102696419208032e-05,
      "loss": 0.8425,
      "step": 569
    },
    {
      "epoch": 3.110504774897681,
      "grad_norm": 0.396338506458531,
      "learning_rate": 2.995482695853535e-05,
      "loss": 0.876,
      "step": 570
    },
    {
      "epoch": 3.1159618008185537,
      "grad_norm": 0.2967118646166294,
      "learning_rate": 2.9807103869513184e-05,
      "loss": 0.8728,
      "step": 571
    },
    {
      "epoch": 3.121418826739427,
      "grad_norm": 0.3675099822893278,
      "learning_rate": 2.965952930466517e-05,
      "loss": 0.8933,
      "step": 572
    },
    {
      "epoch": 3.1268758526603,
      "grad_norm": 0.45762346450524527,
      "learning_rate": 2.9512105414350722e-05,
      "loss": 0.8888,
      "step": 573
    },
    {
      "epoch": 3.132332878581173,
      "grad_norm": 0.4221166157533506,
      "learning_rate": 2.9364834346733784e-05,
      "loss": 0.8791,
      "step": 574
    },
    {
      "epoch": 3.1377899045020463,
      "grad_norm": 0.2995025167192171,
      "learning_rate": 2.9217718247751407e-05,
      "loss": 0.8767,
      "step": 575
    },
    {
      "epoch": 3.1432469304229196,
      "grad_norm": 0.4571280036227124,
      "learning_rate": 2.907075926108257e-05,
      "loss": 0.879,
      "step": 576
    },
    {
      "epoch": 3.148703956343793,
      "grad_norm": 0.5184416527023764,
      "learning_rate": 2.892395952811691e-05,
      "loss": 0.8824,
      "step": 577
    },
    {
      "epoch": 3.1541609822646657,
      "grad_norm": 0.2985433643589337,
      "learning_rate": 2.8777321187923513e-05,
      "loss": 0.8598,
      "step": 578
    },
    {
      "epoch": 3.159618008185539,
      "grad_norm": 0.3051428245095018,
      "learning_rate": 2.8630846377219756e-05,
      "loss": 0.8637,
      "step": 579
    },
    {
      "epoch": 3.1650750341064118,
      "grad_norm": 0.3572619248836196,
      "learning_rate": 2.848453723034021e-05,
      "loss": 0.88,
      "step": 580
    },
    {
      "epoch": 3.170532060027285,
      "grad_norm": 0.3556049119216331,
      "learning_rate": 2.8338395879205447e-05,
      "loss": 0.8648,
      "step": 581
    },
    {
      "epoch": 3.1759890859481583,
      "grad_norm": 0.22986119097957075,
      "learning_rate": 2.819242445329108e-05,
      "loss": 0.8574,
      "step": 582
    },
    {
      "epoch": 3.1814461118690316,
      "grad_norm": 0.2812293245829613,
      "learning_rate": 2.8046625079596643e-05,
      "loss": 0.8635,
      "step": 583
    },
    {
      "epoch": 3.1869031377899044,
      "grad_norm": 0.36671153393330513,
      "learning_rate": 2.7900999882614663e-05,
      "loss": 0.8559,
      "step": 584
    },
    {
      "epoch": 3.1923601637107777,
      "grad_norm": 0.2537974452996025,
      "learning_rate": 2.7755550984299656e-05,
      "loss": 0.8718,
      "step": 585
    },
    {
      "epoch": 3.197817189631651,
      "grad_norm": 0.2168685243882707,
      "learning_rate": 2.7610280504037236e-05,
      "loss": 0.8677,
      "step": 586
    },
    {
      "epoch": 3.203274215552524,
      "grad_norm": 0.2751037843213298,
      "learning_rate": 2.746519055861326e-05,
      "loss": 0.8715,
      "step": 587
    },
    {
      "epoch": 3.208731241473397,
      "grad_norm": 0.25433874445888727,
      "learning_rate": 2.732028326218292e-05,
      "loss": 0.8526,
      "step": 588
    },
    {
      "epoch": 3.2141882673942703,
      "grad_norm": 0.23235048668899896,
      "learning_rate": 2.7175560726239972e-05,
      "loss": 0.8542,
      "step": 589
    },
    {
      "epoch": 3.219645293315143,
      "grad_norm": 0.2449996067653389,
      "learning_rate": 2.7031025059585957e-05,
      "loss": 0.8805,
      "step": 590
    },
    {
      "epoch": 3.2251023192360164,
      "grad_norm": 0.797386710016043,
      "learning_rate": 2.6886678368299478e-05,
      "loss": 0.879,
      "step": 591
    },
    {
      "epoch": 3.2305593451568897,
      "grad_norm": 0.2212327644878727,
      "learning_rate": 2.674252275570552e-05,
      "loss": 0.8581,
      "step": 592
    },
    {
      "epoch": 3.2360163710777625,
      "grad_norm": 0.2762951579133119,
      "learning_rate": 2.659856032234482e-05,
      "loss": 0.8672,
      "step": 593
    },
    {
      "epoch": 3.241473396998636,
      "grad_norm": 0.22565321140932834,
      "learning_rate": 2.6454793165943187e-05,
      "loss": 0.8635,
      "step": 594
    },
    {
      "epoch": 3.246930422919509,
      "grad_norm": 0.2521628451989612,
      "learning_rate": 2.631122338138103e-05,
      "loss": 0.8596,
      "step": 595
    },
    {
      "epoch": 3.252387448840382,
      "grad_norm": 0.2570062440080345,
      "learning_rate": 2.6167853060662743e-05,
      "loss": 0.8886,
      "step": 596
    },
    {
      "epoch": 3.257844474761255,
      "grad_norm": 0.2607472825748654,
      "learning_rate": 2.6024684292886288e-05,
      "loss": 0.8508,
      "step": 597
    },
    {
      "epoch": 3.2633015006821284,
      "grad_norm": 0.23914118744624313,
      "learning_rate": 2.5881719164212714e-05,
      "loss": 0.8578,
      "step": 598
    },
    {
      "epoch": 3.2687585266030013,
      "grad_norm": 0.3034320915776727,
      "learning_rate": 2.5738959757835807e-05,
      "loss": 0.8555,
      "step": 599
    },
    {
      "epoch": 3.2742155525238745,
      "grad_norm": 0.20652047060980766,
      "learning_rate": 2.559640815395167e-05,
      "loss": 0.892,
      "step": 600
    },
    {
      "epoch": 3.279672578444748,
      "grad_norm": 0.28264748490488173,
      "learning_rate": 2.5454066429728465e-05,
      "loss": 0.8568,
      "step": 601
    },
    {
      "epoch": 3.2851296043656206,
      "grad_norm": 0.20253926721485932,
      "learning_rate": 2.531193665927612e-05,
      "loss": 0.8494,
      "step": 602
    },
    {
      "epoch": 3.290586630286494,
      "grad_norm": 0.2776465457759516,
      "learning_rate": 2.5170020913616092e-05,
      "loss": 0.8653,
      "step": 603
    },
    {
      "epoch": 3.296043656207367,
      "grad_norm": 0.2444576656506593,
      "learning_rate": 2.5028321260651235e-05,
      "loss": 0.8653,
      "step": 604
    },
    {
      "epoch": 3.30150068212824,
      "grad_norm": 0.2442116426325601,
      "learning_rate": 2.4886839765135643e-05,
      "loss": 0.8519,
      "step": 605
    },
    {
      "epoch": 3.3069577080491133,
      "grad_norm": 0.2959584392164269,
      "learning_rate": 2.4745578488644544e-05,
      "loss": 0.8746,
      "step": 606
    },
    {
      "epoch": 3.3124147339699865,
      "grad_norm": 0.2021198154975386,
      "learning_rate": 2.460453948954431e-05,
      "loss": 0.8553,
      "step": 607
    },
    {
      "epoch": 3.3178717598908594,
      "grad_norm": 0.24343202473760756,
      "learning_rate": 2.44637248229624e-05,
      "loss": 0.8804,
      "step": 608
    },
    {
      "epoch": 3.3233287858117326,
      "grad_norm": 0.2043418483873985,
      "learning_rate": 2.432313654075746e-05,
      "loss": 0.8563,
      "step": 609
    },
    {
      "epoch": 3.328785811732606,
      "grad_norm": 0.24604327736427975,
      "learning_rate": 2.4182776691489407e-05,
      "loss": 0.8845,
      "step": 610
    },
    {
      "epoch": 3.3342428376534787,
      "grad_norm": 0.2359424714772047,
      "learning_rate": 2.4042647320389597e-05,
      "loss": 0.87,
      "step": 611
    },
    {
      "epoch": 3.339699863574352,
      "grad_norm": 0.21890191711148535,
      "learning_rate": 2.3902750469331003e-05,
      "loss": 0.8768,
      "step": 612
    },
    {
      "epoch": 3.3451568894952253,
      "grad_norm": 0.26622346979171607,
      "learning_rate": 2.3763088176798473e-05,
      "loss": 0.8673,
      "step": 613
    },
    {
      "epoch": 3.350613915416098,
      "grad_norm": 0.2335026323983004,
      "learning_rate": 2.3623662477859016e-05,
      "loss": 0.8764,
      "step": 614
    },
    {
      "epoch": 3.3560709413369714,
      "grad_norm": 0.3242878727490004,
      "learning_rate": 2.348447540413215e-05,
      "loss": 0.8594,
      "step": 615
    },
    {
      "epoch": 3.3615279672578446,
      "grad_norm": 0.30589131325231633,
      "learning_rate": 2.3345528983760306e-05,
      "loss": 0.887,
      "step": 616
    },
    {
      "epoch": 3.3669849931787175,
      "grad_norm": 0.3363507012203662,
      "learning_rate": 2.3206825241379275e-05,
      "loss": 0.8726,
      "step": 617
    },
    {
      "epoch": 3.3724420190995907,
      "grad_norm": 0.26053908816844135,
      "learning_rate": 2.3068366198088725e-05,
      "loss": 0.8585,
      "step": 618
    },
    {
      "epoch": 3.377899045020464,
      "grad_norm": 0.34640941236780975,
      "learning_rate": 2.293015387142271e-05,
      "loss": 0.8686,
      "step": 619
    },
    {
      "epoch": 3.383356070941337,
      "grad_norm": 0.22519137087127883,
      "learning_rate": 2.2792190275320268e-05,
      "loss": 0.8524,
      "step": 620
    },
    {
      "epoch": 3.38881309686221,
      "grad_norm": 0.27989008535136384,
      "learning_rate": 2.265447742009615e-05,
      "loss": 0.8757,
      "step": 621
    },
    {
      "epoch": 3.3942701227830834,
      "grad_norm": 0.22353901032718793,
      "learning_rate": 2.251701731241143e-05,
      "loss": 0.8682,
      "step": 622
    },
    {
      "epoch": 3.399727148703956,
      "grad_norm": 0.2451781215581771,
      "learning_rate": 2.237981195524432e-05,
      "loss": 0.8658,
      "step": 623
    },
    {
      "epoch": 3.4051841746248295,
      "grad_norm": 0.21236674174467807,
      "learning_rate": 2.2242863347861018e-05,
      "loss": 0.857,
      "step": 624
    },
    {
      "epoch": 3.4106412005457027,
      "grad_norm": 0.24488702827145759,
      "learning_rate": 2.210617348578649e-05,
      "loss": 0.8751,
      "step": 625
    },
    {
      "epoch": 3.4160982264665756,
      "grad_norm": 0.19931948544158734,
      "learning_rate": 2.196974436077543e-05,
      "loss": 0.8721,
      "step": 626
    },
    {
      "epoch": 3.421555252387449,
      "grad_norm": 0.20656952281941016,
      "learning_rate": 2.1833577960783266e-05,
      "loss": 0.8837,
      "step": 627
    },
    {
      "epoch": 3.427012278308322,
      "grad_norm": 0.2111569456766622,
      "learning_rate": 2.1697676269937178e-05,
      "loss": 0.8578,
      "step": 628
    },
    {
      "epoch": 3.432469304229195,
      "grad_norm": 0.20607917856291608,
      "learning_rate": 2.1562041268507157e-05,
      "loss": 0.8572,
      "step": 629
    },
    {
      "epoch": 3.437926330150068,
      "grad_norm": 0.2091976197169965,
      "learning_rate": 2.1426674932877156e-05,
      "loss": 0.8574,
      "step": 630
    },
    {
      "epoch": 3.4433833560709415,
      "grad_norm": 0.18994556197856097,
      "learning_rate": 2.1291579235516363e-05,
      "loss": 0.8732,
      "step": 631
    },
    {
      "epoch": 3.4488403819918143,
      "grad_norm": 0.20762295552640858,
      "learning_rate": 2.115675614495036e-05,
      "loss": 0.8643,
      "step": 632
    },
    {
      "epoch": 3.4542974079126876,
      "grad_norm": 0.21448026859846708,
      "learning_rate": 2.1022207625732477e-05,
      "loss": 0.8554,
      "step": 633
    },
    {
      "epoch": 3.459754433833561,
      "grad_norm": 0.22780231585554378,
      "learning_rate": 2.0887935638415203e-05,
      "loss": 0.8538,
      "step": 634
    },
    {
      "epoch": 3.4652114597544337,
      "grad_norm": 0.21671493285699542,
      "learning_rate": 2.075394213952155e-05,
      "loss": 0.8899,
      "step": 635
    },
    {
      "epoch": 3.470668485675307,
      "grad_norm": 0.25593975320430495,
      "learning_rate": 2.0620229081516596e-05,
      "loss": 0.87,
      "step": 636
    },
    {
      "epoch": 3.47612551159618,
      "grad_norm": 0.26516912586558916,
      "learning_rate": 2.0486798412779053e-05,
      "loss": 0.8776,
      "step": 637
    },
    {
      "epoch": 3.481582537517053,
      "grad_norm": 0.30391402267040696,
      "learning_rate": 2.035365207757279e-05,
      "loss": 0.8562,
      "step": 638
    },
    {
      "epoch": 3.4870395634379263,
      "grad_norm": 0.2728528801053457,
      "learning_rate": 2.022079201601856e-05,
      "loss": 0.8689,
      "step": 639
    },
    {
      "epoch": 3.4924965893587996,
      "grad_norm": 0.3399534342641818,
      "learning_rate": 2.008822016406578e-05,
      "loss": 0.8664,
      "step": 640
    },
    {
      "epoch": 3.4979536152796724,
      "grad_norm": 0.23908020280596304,
      "learning_rate": 1.9955938453464198e-05,
      "loss": 0.8514,
      "step": 641
    },
    {
      "epoch": 3.5034106412005457,
      "grad_norm": 0.2747275787837572,
      "learning_rate": 1.9823948811735834e-05,
      "loss": 0.864,
      "step": 642
    },
    {
      "epoch": 3.508867667121419,
      "grad_norm": 0.2675522923940259,
      "learning_rate": 1.9692253162146913e-05,
      "loss": 0.8726,
      "step": 643
    },
    {
      "epoch": 3.5143246930422922,
      "grad_norm": 0.21623449651605725,
      "learning_rate": 1.9560853423679754e-05,
      "loss": 0.8738,
      "step": 644
    },
    {
      "epoch": 3.519781718963165,
      "grad_norm": 0.2815341384705888,
      "learning_rate": 1.9429751511004858e-05,
      "loss": 0.8616,
      "step": 645
    },
    {
      "epoch": 3.5252387448840383,
      "grad_norm": 0.23139814338333467,
      "learning_rate": 1.9298949334453026e-05,
      "loss": 0.892,
      "step": 646
    },
    {
      "epoch": 3.530695770804911,
      "grad_norm": 0.22944247960132097,
      "learning_rate": 1.916844879998748e-05,
      "loss": 0.8786,
      "step": 647
    },
    {
      "epoch": 3.5361527967257844,
      "grad_norm": 0.2976290320962648,
      "learning_rate": 1.9038251809176117e-05,
      "loss": 0.8858,
      "step": 648
    },
    {
      "epoch": 3.5416098226466577,
      "grad_norm": 0.17470790976276032,
      "learning_rate": 1.890836025916382e-05,
      "loss": 0.8743,
      "step": 649
    },
    {
      "epoch": 3.547066848567531,
      "grad_norm": 0.2972531009548509,
      "learning_rate": 1.8778776042644763e-05,
      "loss": 0.8656,
      "step": 650
    },
    {
      "epoch": 3.552523874488404,
      "grad_norm": 0.19583546204252775,
      "learning_rate": 1.864950104783484e-05,
      "loss": 0.8715,
      "step": 651
    },
    {
      "epoch": 3.557980900409277,
      "grad_norm": 0.3083259254483549,
      "learning_rate": 1.852053715844424e-05,
      "loss": 0.8813,
      "step": 652
    },
    {
      "epoch": 3.56343792633015,
      "grad_norm": 0.27570698182575126,
      "learning_rate": 1.839188625364986e-05,
      "loss": 0.8781,
      "step": 653
    },
    {
      "epoch": 3.568894952251023,
      "grad_norm": 0.2509442468585389,
      "learning_rate": 1.8263550208068018e-05,
      "loss": 0.8655,
      "step": 654
    },
    {
      "epoch": 3.5743519781718964,
      "grad_norm": 0.23809316521872612,
      "learning_rate": 1.813553089172709e-05,
      "loss": 0.852,
      "step": 655
    },
    {
      "epoch": 3.5798090040927697,
      "grad_norm": 0.26122110657564457,
      "learning_rate": 1.800783017004034e-05,
      "loss": 0.8493,
      "step": 656
    },
    {
      "epoch": 3.5852660300136425,
      "grad_norm": 0.2398312011858498,
      "learning_rate": 1.7880449903778608e-05,
      "loss": 0.8434,
      "step": 657
    },
    {
      "epoch": 3.590723055934516,
      "grad_norm": 0.2312681495260136,
      "learning_rate": 1.7753391949043318e-05,
      "loss": 0.875,
      "step": 658
    },
    {
      "epoch": 3.5961800818553886,
      "grad_norm": 0.2590426316177245,
      "learning_rate": 1.7626658157239358e-05,
      "loss": 0.8791,
      "step": 659
    },
    {
      "epoch": 3.601637107776262,
      "grad_norm": 0.2288407400611565,
      "learning_rate": 1.7500250375048136e-05,
      "loss": 0.8338,
      "step": 660
    },
    {
      "epoch": 3.607094133697135,
      "grad_norm": 0.26275325952125483,
      "learning_rate": 1.7374170444400645e-05,
      "loss": 0.8831,
      "step": 661
    },
    {
      "epoch": 3.6125511596180084,
      "grad_norm": 0.206477324415559,
      "learning_rate": 1.7248420202450687e-05,
      "loss": 0.8725,
      "step": 662
    },
    {
      "epoch": 3.6180081855388813,
      "grad_norm": 0.24242230587828761,
      "learning_rate": 1.7123001481548017e-05,
      "loss": 0.8669,
      "step": 663
    },
    {
      "epoch": 3.6234652114597545,
      "grad_norm": 0.1974706432396845,
      "learning_rate": 1.6997916109211674e-05,
      "loss": 0.8524,
      "step": 664
    },
    {
      "epoch": 3.6289222373806274,
      "grad_norm": 0.22518193266840958,
      "learning_rate": 1.6873165908103413e-05,
      "loss": 0.8776,
      "step": 665
    },
    {
      "epoch": 3.6343792633015006,
      "grad_norm": 0.21504454890010272,
      "learning_rate": 1.6748752696001047e-05,
      "loss": 0.8508,
      "step": 666
    },
    {
      "epoch": 3.639836289222374,
      "grad_norm": 0.19073320821411494,
      "learning_rate": 1.6624678285772006e-05,
      "loss": 0.8681,
      "step": 667
    },
    {
      "epoch": 3.645293315143247,
      "grad_norm": 0.21561726002963372,
      "learning_rate": 1.6500944485346972e-05,
      "loss": 0.8768,
      "step": 668
    },
    {
      "epoch": 3.65075034106412,
      "grad_norm": 0.20856126216642382,
      "learning_rate": 1.6377553097693435e-05,
      "loss": 0.8642,
      "step": 669
    },
    {
      "epoch": 3.6562073669849933,
      "grad_norm": 0.2503748509640339,
      "learning_rate": 1.625450592078949e-05,
      "loss": 0.8606,
      "step": 670
    },
    {
      "epoch": 3.661664392905866,
      "grad_norm": 0.2114606010902437,
      "learning_rate": 1.6131804747597616e-05,
      "loss": 0.8402,
      "step": 671
    },
    {
      "epoch": 3.6671214188267394,
      "grad_norm": 0.23321442279608323,
      "learning_rate": 1.600945136603855e-05,
      "loss": 0.8757,
      "step": 672
    },
    {
      "epoch": 3.6725784447476126,
      "grad_norm": 0.21972342779108284,
      "learning_rate": 1.5887447558965256e-05,
      "loss": 0.8768,
      "step": 673
    },
    {
      "epoch": 3.678035470668486,
      "grad_norm": 0.21435770379641797,
      "learning_rate": 1.5765795104136894e-05,
      "loss": 0.8776,
      "step": 674
    },
    {
      "epoch": 3.6834924965893587,
      "grad_norm": 0.26135419893907624,
      "learning_rate": 1.5644495774192998e-05,
      "loss": 0.8729,
      "step": 675
    },
    {
      "epoch": 3.688949522510232,
      "grad_norm": 0.21062873576314992,
      "learning_rate": 1.552355133662755e-05,
      "loss": 0.8789,
      "step": 676
    },
    {
      "epoch": 3.694406548431105,
      "grad_norm": 0.24113475613862131,
      "learning_rate": 1.540296355376332e-05,
      "loss": 0.8548,
      "step": 677
    },
    {
      "epoch": 3.699863574351978,
      "grad_norm": 0.23808416561771722,
      "learning_rate": 1.5282734182726113e-05,
      "loss": 0.8503,
      "step": 678
    },
    {
      "epoch": 3.7053206002728514,
      "grad_norm": 0.21962958512416478,
      "learning_rate": 1.516286497541919e-05,
      "loss": 0.87,
      "step": 679
    },
    {
      "epoch": 3.7107776261937246,
      "grad_norm": 0.2069921814306847,
      "learning_rate": 1.5043357678497734e-05,
      "loss": 0.8506,
      "step": 680
    },
    {
      "epoch": 3.7162346521145975,
      "grad_norm": 0.20788126394222645,
      "learning_rate": 1.4924214033343454e-05,
      "loss": 0.8746,
      "step": 681
    },
    {
      "epoch": 3.7216916780354707,
      "grad_norm": 0.20079691647476383,
      "learning_rate": 1.4805435776039123e-05,
      "loss": 0.8776,
      "step": 682
    },
    {
      "epoch": 3.7271487039563436,
      "grad_norm": 0.19394731600989945,
      "learning_rate": 1.4687024637343337e-05,
      "loss": 0.8732,
      "step": 683
    },
    {
      "epoch": 3.732605729877217,
      "grad_norm": 0.2298056829986313,
      "learning_rate": 1.4568982342665274e-05,
      "loss": 0.8782,
      "step": 684
    },
    {
      "epoch": 3.73806275579809,
      "grad_norm": 0.19743254239762598,
      "learning_rate": 1.4451310612039566e-05,
      "loss": 0.8615,
      "step": 685
    },
    {
      "epoch": 3.7435197817189634,
      "grad_norm": 0.21442008615184732,
      "learning_rate": 1.4334011160101216e-05,
      "loss": 0.8755,
      "step": 686
    },
    {
      "epoch": 3.748976807639836,
      "grad_norm": 0.22137254411181218,
      "learning_rate": 1.4217085696060667e-05,
      "loss": 0.8513,
      "step": 687
    },
    {
      "epoch": 3.7544338335607095,
      "grad_norm": 0.23777216991358147,
      "learning_rate": 1.4100535923678815e-05,
      "loss": 0.8649,
      "step": 688
    },
    {
      "epoch": 3.7598908594815823,
      "grad_norm": 0.23646741573904131,
      "learning_rate": 1.3984363541242218e-05,
      "loss": 0.8616,
      "step": 689
    },
    {
      "epoch": 3.7653478854024556,
      "grad_norm": 0.16903481368402593,
      "learning_rate": 1.3868570241538412e-05,
      "loss": 0.8729,
      "step": 690
    },
    {
      "epoch": 3.770804911323329,
      "grad_norm": 0.22814476574366072,
      "learning_rate": 1.375315771183114e-05,
      "loss": 0.8792,
      "step": 691
    },
    {
      "epoch": 3.776261937244202,
      "grad_norm": 0.2138508475050253,
      "learning_rate": 1.3638127633835808e-05,
      "loss": 0.8514,
      "step": 692
    },
    {
      "epoch": 3.781718963165075,
      "grad_norm": 0.1702228436710734,
      "learning_rate": 1.3523481683695034e-05,
      "loss": 0.8569,
      "step": 693
    },
    {
      "epoch": 3.787175989085948,
      "grad_norm": 0.275935108213432,
      "learning_rate": 1.3409221531954146e-05,
      "loss": 0.8642,
      "step": 694
    },
    {
      "epoch": 3.792633015006821,
      "grad_norm": 0.18178655458313908,
      "learning_rate": 1.329534884353687e-05,
      "loss": 0.8617,
      "step": 695
    },
    {
      "epoch": 3.7980900409276943,
      "grad_norm": 0.19502338070740763,
      "learning_rate": 1.3181865277721086e-05,
      "loss": 0.8957,
      "step": 696
    },
    {
      "epoch": 3.8035470668485676,
      "grad_norm": 0.24435912255988515,
      "learning_rate": 1.3068772488114619e-05,
      "loss": 0.8534,
      "step": 697
    },
    {
      "epoch": 3.809004092769441,
      "grad_norm": 0.1790945885669287,
      "learning_rate": 1.2956072122631173e-05,
      "loss": 0.8767,
      "step": 698
    },
    {
      "epoch": 3.8144611186903137,
      "grad_norm": 0.1799037891667036,
      "learning_rate": 1.2843765823466292e-05,
      "loss": 0.8817,
      "step": 699
    },
    {
      "epoch": 3.819918144611187,
      "grad_norm": 0.23900669879884437,
      "learning_rate": 1.2731855227073475e-05,
      "loss": 0.8577,
      "step": 700
    },
    {
      "epoch": 3.8253751705320598,
      "grad_norm": 0.1996247040670919,
      "learning_rate": 1.2620341964140264e-05,
      "loss": 0.8767,
      "step": 701
    },
    {
      "epoch": 3.830832196452933,
      "grad_norm": 0.19203450849102435,
      "learning_rate": 1.2509227659564548e-05,
      "loss": 0.8604,
      "step": 702
    },
    {
      "epoch": 3.8362892223738063,
      "grad_norm": 0.20665454749391476,
      "learning_rate": 1.2398513932430847e-05,
      "loss": 0.88,
      "step": 703
    },
    {
      "epoch": 3.8417462482946796,
      "grad_norm": 0.20396894732105844,
      "learning_rate": 1.2288202395986723e-05,
      "loss": 0.8705,
      "step": 704
    },
    {
      "epoch": 3.8472032742155524,
      "grad_norm": 0.18826995936772803,
      "learning_rate": 1.2178294657619274e-05,
      "loss": 0.8938,
      "step": 705
    },
    {
      "epoch": 3.8526603001364257,
      "grad_norm": 0.25405087801781573,
      "learning_rate": 1.2068792318831748e-05,
      "loss": 0.8786,
      "step": 706
    },
    {
      "epoch": 3.8581173260572985,
      "grad_norm": 0.17190779379675156,
      "learning_rate": 1.1959696975220156e-05,
      "loss": 0.8731,
      "step": 707
    },
    {
      "epoch": 3.863574351978172,
      "grad_norm": 0.2035464468290126,
      "learning_rate": 1.185101021645004e-05,
      "loss": 0.8544,
      "step": 708
    },
    {
      "epoch": 3.869031377899045,
      "grad_norm": 0.1775096388954675,
      "learning_rate": 1.1742733626233322e-05,
      "loss": 0.8738,
      "step": 709
    },
    {
      "epoch": 3.8744884038199183,
      "grad_norm": 0.1962472136724812,
      "learning_rate": 1.163486878230521e-05,
      "loss": 0.8655,
      "step": 710
    },
    {
      "epoch": 3.879945429740791,
      "grad_norm": 0.19182896784098635,
      "learning_rate": 1.1527417256401217e-05,
      "loss": 0.8625,
      "step": 711
    },
    {
      "epoch": 3.8854024556616644,
      "grad_norm": 0.1772961855031946,
      "learning_rate": 1.1420380614234277e-05,
      "loss": 0.8691,
      "step": 712
    },
    {
      "epoch": 3.8908594815825372,
      "grad_norm": 0.19619850458804647,
      "learning_rate": 1.1313760415471875e-05,
      "loss": 0.8613,
      "step": 713
    },
    {
      "epoch": 3.8963165075034105,
      "grad_norm": 0.2241487330551882,
      "learning_rate": 1.1207558213713398e-05,
      "loss": 0.8811,
      "step": 714
    },
    {
      "epoch": 3.901773533424284,
      "grad_norm": 0.158115197476845,
      "learning_rate": 1.1101775556467427e-05,
      "loss": 0.8665,
      "step": 715
    },
    {
      "epoch": 3.907230559345157,
      "grad_norm": 0.22420983108248752,
      "learning_rate": 1.0996413985129224e-05,
      "loss": 0.8894,
      "step": 716
    },
    {
      "epoch": 3.91268758526603,
      "grad_norm": 0.1610631974388593,
      "learning_rate": 1.0891475034958265e-05,
      "loss": 0.8623,
      "step": 717
    },
    {
      "epoch": 3.918144611186903,
      "grad_norm": 0.15967594778215696,
      "learning_rate": 1.0786960235055855e-05,
      "loss": 0.8882,
      "step": 718
    },
    {
      "epoch": 3.923601637107776,
      "grad_norm": 0.19319520783715666,
      "learning_rate": 1.0682871108342896e-05,
      "loss": 0.857,
      "step": 719
    },
    {
      "epoch": 3.9290586630286493,
      "grad_norm": 0.15711658830357056,
      "learning_rate": 1.0579209171537626e-05,
      "loss": 0.8576,
      "step": 720
    },
    {
      "epoch": 3.9345156889495225,
      "grad_norm": 0.1826574560537012,
      "learning_rate": 1.0475975935133573e-05,
      "loss": 0.8515,
      "step": 721
    },
    {
      "epoch": 3.939972714870396,
      "grad_norm": 0.16069667852451064,
      "learning_rate": 1.0373172903377507e-05,
      "loss": 0.8692,
      "step": 722
    },
    {
      "epoch": 3.9454297407912686,
      "grad_norm": 0.16866708527324803,
      "learning_rate": 1.0270801574247553e-05,
      "loss": 0.8788,
      "step": 723
    },
    {
      "epoch": 3.950886766712142,
      "grad_norm": 0.1530248706749601,
      "learning_rate": 1.0168863439431336e-05,
      "loss": 0.8653,
      "step": 724
    },
    {
      "epoch": 3.956343792633015,
      "grad_norm": 0.15769274523646723,
      "learning_rate": 1.0067359984304286e-05,
      "loss": 0.8676,
      "step": 725
    },
    {
      "epoch": 3.961800818553888,
      "grad_norm": 0.15424491068484994,
      "learning_rate": 9.966292687907928e-06,
      "loss": 0.853,
      "step": 726
    },
    {
      "epoch": 3.9672578444747613,
      "grad_norm": 0.15860068150396306,
      "learning_rate": 9.865663022928413e-06,
      "loss": 0.8605,
      "step": 727
    },
    {
      "epoch": 3.9727148703956345,
      "grad_norm": 0.1700340270255791,
      "learning_rate": 9.76547245567498e-06,
      "loss": 0.8682,
      "step": 728
    },
    {
      "epoch": 3.9781718963165074,
      "grad_norm": 0.15646496600104898,
      "learning_rate": 9.66572244605864e-06,
      "loss": 0.8555,
      "step": 729
    },
    {
      "epoch": 3.9836289222373806,
      "grad_norm": 0.1541837949594622,
      "learning_rate": 9.566414447570861e-06,
      "loss": 0.8697,
      "step": 730
    },
    {
      "epoch": 3.989085948158254,
      "grad_norm": 0.15359278717510694,
      "learning_rate": 9.467549907262476e-06,
      "loss": 0.8709,
      "step": 731
    },
    {
      "epoch": 3.9945429740791267,
      "grad_norm": 0.16227617763941107,
      "learning_rate": 9.36913026572248e-06,
      "loss": 0.8664,
      "step": 732
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.3013369962541664,
      "learning_rate": 9.271156957057123e-06,
      "loss": 1.5027,
      "step": 733
    },
    {
      "epoch": 4.005457025920873,
      "grad_norm": 0.17363692755284943,
      "learning_rate": 9.173631408868976e-06,
      "loss": 0.85,
      "step": 734
    },
    {
      "epoch": 4.0109140518417465,
      "grad_norm": 0.19633143503130812,
      "learning_rate": 9.076555042236142e-06,
      "loss": 0.8465,
      "step": 735
    },
    {
      "epoch": 4.01637107776262,
      "grad_norm": 0.19091321505385692,
      "learning_rate": 8.979929271691538e-06,
      "loss": 0.8522,
      "step": 736
    },
    {
      "epoch": 4.021828103683492,
      "grad_norm": 0.17707201863006375,
      "learning_rate": 8.88375550520232e-06,
      "loss": 0.8506,
      "step": 737
    },
    {
      "epoch": 4.0272851296043655,
      "grad_norm": 0.20661036944371583,
      "learning_rate": 8.788035144149299e-06,
      "loss": 0.8323,
      "step": 738
    },
    {
      "epoch": 4.032742155525239,
      "grad_norm": 0.1798715373187608,
      "learning_rate": 8.692769583306596e-06,
      "loss": 0.8603,
      "step": 739
    },
    {
      "epoch": 4.038199181446112,
      "grad_norm": 0.15777175533167295,
      "learning_rate": 8.597960210821261e-06,
      "loss": 0.8279,
      "step": 740
    },
    {
      "epoch": 4.043656207366985,
      "grad_norm": 0.22344847319218858,
      "learning_rate": 8.50360840819307e-06,
      "loss": 0.8403,
      "step": 741
    },
    {
      "epoch": 4.0491132332878585,
      "grad_norm": 0.1816715320324387,
      "learning_rate": 8.409715550254396e-06,
      "loss": 0.8376,
      "step": 742
    },
    {
      "epoch": 4.054570259208731,
      "grad_norm": 0.15853189298493875,
      "learning_rate": 8.316283005150159e-06,
      "loss": 0.8511,
      "step": 743
    },
    {
      "epoch": 4.060027285129604,
      "grad_norm": 0.21309295778042034,
      "learning_rate": 8.223312134317942e-06,
      "loss": 0.8586,
      "step": 744
    },
    {
      "epoch": 4.0654843110504775,
      "grad_norm": 0.1689534095697035,
      "learning_rate": 8.130804292468069e-06,
      "loss": 0.8117,
      "step": 745
    },
    {
      "epoch": 4.070941336971351,
      "grad_norm": 0.13977942461160103,
      "learning_rate": 8.038760827563936e-06,
      "loss": 0.8463,
      "step": 746
    },
    {
      "epoch": 4.076398362892224,
      "grad_norm": 0.16628039538402697,
      "learning_rate": 7.94718308080233e-06,
      "loss": 0.8631,
      "step": 747
    },
    {
      "epoch": 4.081855388813097,
      "grad_norm": 0.1799528768822378,
      "learning_rate": 7.856072386593915e-06,
      "loss": 0.8504,
      "step": 748
    },
    {
      "epoch": 4.08731241473397,
      "grad_norm": 0.15460807238366664,
      "learning_rate": 7.765430072543748e-06,
      "loss": 0.8358,
      "step": 749
    },
    {
      "epoch": 4.092769440654843,
      "grad_norm": 0.1615993268635049,
      "learning_rate": 7.675257459431988e-06,
      "loss": 0.8407,
      "step": 750
    },
    {
      "epoch": 4.098226466575716,
      "grad_norm": 0.15392647616267377,
      "learning_rate": 7.585555861194613e-06,
      "loss": 0.8334,
      "step": 751
    },
    {
      "epoch": 4.1036834924965895,
      "grad_norm": 0.14497560365020726,
      "learning_rate": 7.496326584904277e-06,
      "loss": 0.8362,
      "step": 752
    },
    {
      "epoch": 4.109140518417463,
      "grad_norm": 0.15870809306205555,
      "learning_rate": 7.4075709307512575e-06,
      "loss": 0.8534,
      "step": 753
    },
    {
      "epoch": 4.114597544338336,
      "grad_norm": 0.15582879109817865,
      "learning_rate": 7.31929019202454e-06,
      "loss": 0.8379,
      "step": 754
    },
    {
      "epoch": 4.120054570259208,
      "grad_norm": 0.14414796007354208,
      "learning_rate": 7.231485655092943e-06,
      "loss": 0.8479,
      "step": 755
    },
    {
      "epoch": 4.125511596180082,
      "grad_norm": 0.15940644110537328,
      "learning_rate": 7.144158599386402e-06,
      "loss": 0.8708,
      "step": 756
    },
    {
      "epoch": 4.130968622100955,
      "grad_norm": 0.15073810143549604,
      "learning_rate": 7.057310297377298e-06,
      "loss": 0.8466,
      "step": 757
    },
    {
      "epoch": 4.136425648021828,
      "grad_norm": 0.13150444462070968,
      "learning_rate": 6.97094201456193e-06,
      "loss": 0.8383,
      "step": 758
    },
    {
      "epoch": 4.1418826739427015,
      "grad_norm": 0.1682880467445867,
      "learning_rate": 6.8850550094420675e-06,
      "loss": 0.8492,
      "step": 759
    },
    {
      "epoch": 4.147339699863575,
      "grad_norm": 0.1282839298295238,
      "learning_rate": 6.799650533506632e-06,
      "loss": 0.86,
      "step": 760
    },
    {
      "epoch": 4.152796725784447,
      "grad_norm": 0.14656575634048025,
      "learning_rate": 6.714729831213422e-06,
      "loss": 0.8438,
      "step": 761
    },
    {
      "epoch": 4.15825375170532,
      "grad_norm": 0.1533256626816939,
      "learning_rate": 6.630294139971045e-06,
      "loss": 0.8403,
      "step": 762
    },
    {
      "epoch": 4.163710777626194,
      "grad_norm": 0.1380671788725022,
      "learning_rate": 6.546344690120809e-06,
      "loss": 0.8521,
      "step": 763
    },
    {
      "epoch": 4.169167803547067,
      "grad_norm": 0.1203348172009764,
      "learning_rate": 6.46288270491886e-06,
      "loss": 0.8558,
      "step": 764
    },
    {
      "epoch": 4.17462482946794,
      "grad_norm": 0.13733698891901175,
      "learning_rate": 6.3799094005183095e-06,
      "loss": 0.8369,
      "step": 765
    },
    {
      "epoch": 4.1800818553888135,
      "grad_norm": 0.13345236448780423,
      "learning_rate": 6.297425985951546e-06,
      "loss": 0.8195,
      "step": 766
    },
    {
      "epoch": 4.185538881309686,
      "grad_norm": 0.14321025830983808,
      "learning_rate": 6.215433663112596e-06,
      "loss": 0.8547,
      "step": 767
    },
    {
      "epoch": 4.190995907230559,
      "grad_norm": 0.13924098390685724,
      "learning_rate": 6.133933626739623e-06,
      "loss": 0.8255,
      "step": 768
    },
    {
      "epoch": 4.196452933151432,
      "grad_norm": 0.1261806757953622,
      "learning_rate": 6.052927064397534e-06,
      "loss": 0.8417,
      "step": 769
    },
    {
      "epoch": 4.201909959072306,
      "grad_norm": 0.14319746757365104,
      "learning_rate": 5.972415156460631e-06,
      "loss": 0.836,
      "step": 770
    },
    {
      "epoch": 4.207366984993179,
      "grad_norm": 0.13488872447248845,
      "learning_rate": 5.892399076095454e-06,
      "loss": 0.823,
      "step": 771
    },
    {
      "epoch": 4.212824010914052,
      "grad_norm": 0.14256043162591792,
      "learning_rate": 5.812879989243661e-06,
      "loss": 0.8292,
      "step": 772
    },
    {
      "epoch": 4.218281036834925,
      "grad_norm": 0.14297739682297367,
      "learning_rate": 5.733859054605044e-06,
      "loss": 0.8492,
      "step": 773
    },
    {
      "epoch": 4.223738062755798,
      "grad_norm": 0.13214429368901517,
      "learning_rate": 5.655337423620677e-06,
      "loss": 0.8485,
      "step": 774
    },
    {
      "epoch": 4.229195088676671,
      "grad_norm": 0.1309888866682516,
      "learning_rate": 5.5773162404560675e-06,
      "loss": 0.8587,
      "step": 775
    },
    {
      "epoch": 4.234652114597544,
      "grad_norm": 0.5572690678106634,
      "learning_rate": 5.499796641984572e-06,
      "loss": 0.8711,
      "step": 776
    },
    {
      "epoch": 4.240109140518418,
      "grad_norm": 0.142020973503765,
      "learning_rate": 5.422779757770742e-06,
      "loss": 0.8395,
      "step": 777
    },
    {
      "epoch": 4.245566166439291,
      "grad_norm": 0.1422911809797413,
      "learning_rate": 5.346266710053938e-06,
      "loss": 0.8371,
      "step": 778
    },
    {
      "epoch": 4.251023192360163,
      "grad_norm": 0.13566149323405943,
      "learning_rate": 5.270258613731929e-06,
      "loss": 0.837,
      "step": 779
    },
    {
      "epoch": 4.256480218281037,
      "grad_norm": 0.1403658624168453,
      "learning_rate": 5.194756576344664e-06,
      "loss": 0.8472,
      "step": 780
    },
    {
      "epoch": 4.26193724420191,
      "grad_norm": 0.12851416052393547,
      "learning_rate": 5.119761698058158e-06,
      "loss": 0.842,
      "step": 781
    },
    {
      "epoch": 4.267394270122783,
      "grad_norm": 0.1267181054298902,
      "learning_rate": 5.045275071648417e-06,
      "loss": 0.8486,
      "step": 782
    },
    {
      "epoch": 4.272851296043656,
      "grad_norm": 0.14230492360141342,
      "learning_rate": 4.971297782485534e-06,
      "loss": 0.8523,
      "step": 783
    },
    {
      "epoch": 4.27830832196453,
      "grad_norm": 0.1385288616652607,
      "learning_rate": 4.897830908517884e-06,
      "loss": 0.8636,
      "step": 784
    },
    {
      "epoch": 4.283765347885402,
      "grad_norm": 0.12011049268492519,
      "learning_rate": 4.824875520256398e-06,
      "loss": 0.8527,
      "step": 785
    },
    {
      "epoch": 4.289222373806275,
      "grad_norm": 0.12515350933731695,
      "learning_rate": 4.752432680758978e-06,
      "loss": 0.8446,
      "step": 786
    },
    {
      "epoch": 4.294679399727149,
      "grad_norm": 0.13980725339910127,
      "learning_rate": 4.6805034456150145e-06,
      "loss": 0.8463,
      "step": 787
    },
    {
      "epoch": 4.300136425648022,
      "grad_norm": 0.16408440175020803,
      "learning_rate": 4.609088862929971e-06,
      "loss": 0.8292,
      "step": 788
    },
    {
      "epoch": 4.305593451568895,
      "grad_norm": 0.12454830323243238,
      "learning_rate": 4.5381899733101606e-06,
      "loss": 0.8512,
      "step": 789
    },
    {
      "epoch": 4.311050477489768,
      "grad_norm": 0.13173108796782726,
      "learning_rate": 4.4678078098475374e-06,
      "loss": 0.8636,
      "step": 790
    },
    {
      "epoch": 4.316507503410641,
      "grad_norm": 0.1372653963988401,
      "learning_rate": 4.39794339810466e-06,
      "loss": 0.8546,
      "step": 791
    },
    {
      "epoch": 4.321964529331514,
      "grad_norm": 0.1454111561073045,
      "learning_rate": 4.328597756099764e-06,
      "loss": 0.8285,
      "step": 792
    },
    {
      "epoch": 4.327421555252387,
      "grad_norm": 0.13292312782936808,
      "learning_rate": 4.259771894291889e-06,
      "loss": 0.844,
      "step": 793
    },
    {
      "epoch": 4.332878581173261,
      "grad_norm": 0.13113926812021118,
      "learning_rate": 4.191466815566223e-06,
      "loss": 0.8624,
      "step": 794
    },
    {
      "epoch": 4.338335607094134,
      "grad_norm": 0.13723911665389374,
      "learning_rate": 4.123683515219403e-06,
      "loss": 0.866,
      "step": 795
    },
    {
      "epoch": 4.343792633015007,
      "grad_norm": 0.1274185802096606,
      "learning_rate": 4.056422980945076e-06,
      "loss": 0.835,
      "step": 796
    },
    {
      "epoch": 4.34924965893588,
      "grad_norm": 0.15079186440244974,
      "learning_rate": 3.98968619281948e-06,
      "loss": 0.855,
      "step": 797
    },
    {
      "epoch": 4.354706684856753,
      "grad_norm": 0.13975651293727326,
      "learning_rate": 3.923474123287165e-06,
      "loss": 0.838,
      "step": 798
    },
    {
      "epoch": 4.360163710777626,
      "grad_norm": 0.12596602187941267,
      "learning_rate": 3.857787737146841e-06,
      "loss": 0.8424,
      "step": 799
    },
    {
      "epoch": 4.365620736698499,
      "grad_norm": 0.13192216189243972,
      "learning_rate": 3.7926279915372877e-06,
      "loss": 0.8488,
      "step": 800
    },
    {
      "epoch": 4.371077762619373,
      "grad_norm": 0.11294983013886342,
      "learning_rate": 3.727995835923448e-06,
      "loss": 0.8377,
      "step": 801
    },
    {
      "epoch": 4.376534788540246,
      "grad_norm": 0.12208950440799905,
      "learning_rate": 3.663892212082547e-06,
      "loss": 0.8451,
      "step": 802
    },
    {
      "epoch": 4.381991814461118,
      "grad_norm": 0.12500010048360874,
      "learning_rate": 3.6003180540904014e-06,
      "loss": 0.8565,
      "step": 803
    },
    {
      "epoch": 4.387448840381992,
      "grad_norm": 0.11635114649417948,
      "learning_rate": 3.5372742883078083e-06,
      "loss": 0.8577,
      "step": 804
    },
    {
      "epoch": 4.392905866302865,
      "grad_norm": 0.13168309738770168,
      "learning_rate": 3.4747618333670263e-06,
      "loss": 0.8469,
      "step": 805
    },
    {
      "epoch": 4.398362892223738,
      "grad_norm": 0.12825374722074376,
      "learning_rate": 3.4127816001583968e-06,
      "loss": 0.8563,
      "step": 806
    },
    {
      "epoch": 4.403819918144611,
      "grad_norm": 0.11335046162731692,
      "learning_rate": 3.351334491817113e-06,
      "loss": 0.8449,
      "step": 807
    },
    {
      "epoch": 4.409276944065485,
      "grad_norm": 0.12437993837114146,
      "learning_rate": 3.290421403709978e-06,
      "loss": 0.8403,
      "step": 808
    },
    {
      "epoch": 4.414733969986357,
      "grad_norm": 0.12277805128760136,
      "learning_rate": 3.230043223422432e-06,
      "loss": 0.8476,
      "step": 809
    },
    {
      "epoch": 4.42019099590723,
      "grad_norm": 0.1229213914667106,
      "learning_rate": 3.170200830745569e-06,
      "loss": 0.8395,
      "step": 810
    },
    {
      "epoch": 4.425648021828104,
      "grad_norm": 0.10562773517470789,
      "learning_rate": 3.1108950976633624e-06,
      "loss": 0.854,
      "step": 811
    },
    {
      "epoch": 4.431105047748977,
      "grad_norm": 0.11479598409158459,
      "learning_rate": 3.052126888339917e-06,
      "loss": 0.8356,
      "step": 812
    },
    {
      "epoch": 4.43656207366985,
      "grad_norm": 0.11116219748454473,
      "learning_rate": 2.993897059106914e-06,
      "loss": 0.8485,
      "step": 813
    },
    {
      "epoch": 4.442019099590723,
      "grad_norm": 0.12390938866072493,
      "learning_rate": 2.936206458451092e-06,
      "loss": 0.8421,
      "step": 814
    },
    {
      "epoch": 4.447476125511596,
      "grad_norm": 0.11225533367333312,
      "learning_rate": 2.879055927001919e-06,
      "loss": 0.8326,
      "step": 815
    },
    {
      "epoch": 4.452933151432469,
      "grad_norm": 0.12120537299560043,
      "learning_rate": 2.8224462975193235e-06,
      "loss": 0.8268,
      "step": 816
    },
    {
      "epoch": 4.458390177353342,
      "grad_norm": 0.11161258091057706,
      "learning_rate": 2.7663783948815725e-06,
      "loss": 0.8404,
      "step": 817
    },
    {
      "epoch": 4.463847203274216,
      "grad_norm": 0.1115528748397204,
      "learning_rate": 2.7108530360732354e-06,
      "loss": 0.8712,
      "step": 818
    },
    {
      "epoch": 4.469304229195089,
      "grad_norm": 0.10731054795986365,
      "learning_rate": 2.655871030173307e-06,
      "loss": 0.832,
      "step": 819
    },
    {
      "epoch": 4.474761255115962,
      "grad_norm": 0.1151101918058789,
      "learning_rate": 2.6014331783433957e-06,
      "loss": 0.8151,
      "step": 820
    },
    {
      "epoch": 4.480218281036835,
      "grad_norm": 0.12150762806194712,
      "learning_rate": 2.5475402738160426e-06,
      "loss": 0.8709,
      "step": 821
    },
    {
      "epoch": 4.485675306957708,
      "grad_norm": 0.11481783034848463,
      "learning_rate": 2.4941931018831822e-06,
      "loss": 0.8589,
      "step": 822
    },
    {
      "epoch": 4.491132332878581,
      "grad_norm": 0.12049475973057963,
      "learning_rate": 2.4413924398846957e-06,
      "loss": 0.8455,
      "step": 823
    },
    {
      "epoch": 4.496589358799454,
      "grad_norm": 0.1263474250088142,
      "learning_rate": 2.3891390571970875e-06,
      "loss": 0.8728,
      "step": 824
    },
    {
      "epoch": 4.502046384720328,
      "grad_norm": 0.11596607365643856,
      "learning_rate": 2.3374337152222503e-06,
      "loss": 0.8586,
      "step": 825
    },
    {
      "epoch": 4.507503410641201,
      "grad_norm": 0.11715195084129398,
      "learning_rate": 2.2862771673764116e-06,
      "loss": 0.8267,
      "step": 826
    },
    {
      "epoch": 4.512960436562073,
      "grad_norm": 0.10916905965796952,
      "learning_rate": 2.2356701590791107e-06,
      "loss": 0.8377,
      "step": 827
    },
    {
      "epoch": 4.5184174624829465,
      "grad_norm": 0.10989084515422318,
      "learning_rate": 2.1856134277423634e-06,
      "loss": 0.8456,
      "step": 828
    },
    {
      "epoch": 4.52387448840382,
      "grad_norm": 0.11415331936192948,
      "learning_rate": 2.13610770275992e-06,
      "loss": 0.8428,
      "step": 829
    },
    {
      "epoch": 4.529331514324693,
      "grad_norm": 0.11187464286160874,
      "learning_rate": 2.087153705496623e-06,
      "loss": 0.855,
      "step": 830
    },
    {
      "epoch": 4.534788540245566,
      "grad_norm": 0.11068155097698111,
      "learning_rate": 2.038752149277898e-06,
      "loss": 0.8439,
      "step": 831
    },
    {
      "epoch": 4.54024556616644,
      "grad_norm": 0.11113729825886462,
      "learning_rate": 1.9909037393793708e-06,
      "loss": 0.8448,
      "step": 832
    },
    {
      "epoch": 4.545702592087313,
      "grad_norm": 0.13136302301656613,
      "learning_rate": 1.943609173016583e-06,
      "loss": 0.8413,
      "step": 833
    },
    {
      "epoch": 4.551159618008185,
      "grad_norm": 0.11768000783406421,
      "learning_rate": 1.896869139334827e-06,
      "loss": 0.8622,
      "step": 834
    },
    {
      "epoch": 4.5566166439290585,
      "grad_norm": 0.12259054296868203,
      "learning_rate": 1.8506843193991076e-06,
      "loss": 0.8287,
      "step": 835
    },
    {
      "epoch": 4.562073669849932,
      "grad_norm": 0.12126631078010071,
      "learning_rate": 1.8050553861842335e-06,
      "loss": 0.8568,
      "step": 836
    },
    {
      "epoch": 4.567530695770805,
      "grad_norm": 0.12030422351425824,
      "learning_rate": 1.7599830045649825e-06,
      "loss": 0.851,
      "step": 837
    },
    {
      "epoch": 4.572987721691678,
      "grad_norm": 0.10681554191695464,
      "learning_rate": 1.7154678313064455e-06,
      "loss": 0.8517,
      "step": 838
    },
    {
      "epoch": 4.578444747612551,
      "grad_norm": 0.10201295749034411,
      "learning_rate": 1.671510515054422e-06,
      "loss": 0.8482,
      "step": 839
    },
    {
      "epoch": 4.583901773533424,
      "grad_norm": 0.10882756711091733,
      "learning_rate": 1.6281116963259957e-06,
      "loss": 0.8454,
      "step": 840
    },
    {
      "epoch": 4.589358799454297,
      "grad_norm": 0.11080725448199538,
      "learning_rate": 1.585272007500196e-06,
      "loss": 0.8228,
      "step": 841
    },
    {
      "epoch": 4.5948158253751705,
      "grad_norm": 0.1064247093738944,
      "learning_rate": 1.5429920728087688e-06,
      "loss": 0.852,
      "step": 842
    },
    {
      "epoch": 4.600272851296044,
      "grad_norm": 0.10624759120004654,
      "learning_rate": 1.5012725083270963e-06,
      "loss": 0.851,
      "step": 843
    },
    {
      "epoch": 4.605729877216917,
      "grad_norm": 0.11230799021650528,
      "learning_rate": 1.4601139219652205e-06,
      "loss": 0.8757,
      "step": 844
    },
    {
      "epoch": 4.61118690313779,
      "grad_norm": 0.11562162612177647,
      "learning_rate": 1.4195169134589715e-06,
      "loss": 0.8608,
      "step": 845
    },
    {
      "epoch": 4.616643929058663,
      "grad_norm": 0.11226676876853169,
      "learning_rate": 1.3794820743612404e-06,
      "loss": 0.858,
      "step": 846
    },
    {
      "epoch": 4.622100954979536,
      "grad_norm": 0.11074079216642702,
      "learning_rate": 1.340009988033346e-06,
      "loss": 0.8457,
      "step": 847
    },
    {
      "epoch": 4.627557980900409,
      "grad_norm": 0.11782568400682915,
      "learning_rate": 1.3011012296365633e-06,
      "loss": 0.8486,
      "step": 848
    },
    {
      "epoch": 4.6330150068212825,
      "grad_norm": 0.1151989040836682,
      "learning_rate": 1.2627563661237097e-06,
      "loss": 0.8394,
      "step": 849
    },
    {
      "epoch": 4.638472032742156,
      "grad_norm": 0.10221115562948982,
      "learning_rate": 1.2249759562308916e-06,
      "loss": 0.8371,
      "step": 850
    },
    {
      "epoch": 4.643929058663028,
      "grad_norm": 0.10670215373834935,
      "learning_rate": 1.1877605504693945e-06,
      "loss": 0.8535,
      "step": 851
    },
    {
      "epoch": 4.6493860845839015,
      "grad_norm": 0.09994602009397384,
      "learning_rate": 1.151110691117605e-06,
      "loss": 0.8488,
      "step": 852
    },
    {
      "epoch": 4.654843110504775,
      "grad_norm": 0.10378931622905237,
      "learning_rate": 1.1150269122131596e-06,
      "loss": 0.8287,
      "step": 853
    },
    {
      "epoch": 4.660300136425648,
      "grad_norm": 0.11214724803482708,
      "learning_rate": 1.0795097395451371e-06,
      "loss": 0.8516,
      "step": 854
    },
    {
      "epoch": 4.665757162346521,
      "grad_norm": 0.10833228711670279,
      "learning_rate": 1.044559690646394e-06,
      "loss": 0.8346,
      "step": 855
    },
    {
      "epoch": 4.6712141882673945,
      "grad_norm": 0.1066533070122156,
      "learning_rate": 1.0101772747860462e-06,
      "loss": 0.8482,
      "step": 856
    },
    {
      "epoch": 4.676671214188268,
      "grad_norm": 0.1020151272866244,
      "learning_rate": 9.763629929620345e-07,
      "loss": 0.8365,
      "step": 857
    },
    {
      "epoch": 4.68212824010914,
      "grad_norm": 0.10837508125099268,
      "learning_rate": 9.431173378938108e-07,
      "loss": 0.8399,
      "step": 858
    },
    {
      "epoch": 4.6875852660300135,
      "grad_norm": 0.10123855993633649,
      "learning_rate": 9.104407940151882e-07,
      "loss": 0.8531,
      "step": 859
    },
    {
      "epoch": 4.693042291950887,
      "grad_norm": 0.09594740104370002,
      "learning_rate": 8.783338374672534e-07,
      "loss": 0.8498,
      "step": 860
    },
    {
      "epoch": 4.69849931787176,
      "grad_norm": 0.10314744523546084,
      "learning_rate": 8.467969360914519e-07,
      "loss": 0.8417,
      "step": 861
    },
    {
      "epoch": 4.703956343792633,
      "grad_norm": 0.1070045235427388,
      "learning_rate": 8.158305494227492e-07,
      "loss": 0.8422,
      "step": 862
    },
    {
      "epoch": 4.709413369713506,
      "grad_norm": 0.11029818790553916,
      "learning_rate": 7.854351286829565e-07,
      "loss": 0.8342,
      "step": 863
    },
    {
      "epoch": 4.714870395634379,
      "grad_norm": 0.10651305096899819,
      "learning_rate": 7.556111167741354e-07,
      "loss": 0.8238,
      "step": 864
    },
    {
      "epoch": 4.720327421555252,
      "grad_norm": 0.11712154625311387,
      "learning_rate": 7.263589482721678e-07,
      "loss": 0.8367,
      "step": 865
    },
    {
      "epoch": 4.7257844474761255,
      "grad_norm": 0.11677316573492764,
      "learning_rate": 6.976790494203967e-07,
      "loss": 0.8289,
      "step": 866
    },
    {
      "epoch": 4.731241473396999,
      "grad_norm": 0.10787652603448866,
      "learning_rate": 6.695718381234306e-07,
      "loss": 0.8726,
      "step": 867
    },
    {
      "epoch": 4.736698499317872,
      "grad_norm": 0.10325574905197629,
      "learning_rate": 6.420377239410514e-07,
      "loss": 0.8545,
      "step": 868
    },
    {
      "epoch": 4.742155525238745,
      "grad_norm": 0.10313792569802335,
      "learning_rate": 6.150771080822581e-07,
      "loss": 0.824,
      "step": 869
    },
    {
      "epoch": 4.747612551159618,
      "grad_norm": 0.10775368857971664,
      "learning_rate": 5.886903833994018e-07,
      "loss": 0.8471,
      "step": 870
    },
    {
      "epoch": 4.753069577080491,
      "grad_norm": 0.11948172384943638,
      "learning_rate": 5.628779343824642e-07,
      "loss": 0.8506,
      "step": 871
    },
    {
      "epoch": 4.758526603001364,
      "grad_norm": 0.10870113816849278,
      "learning_rate": 5.376401371534589e-07,
      "loss": 0.8288,
      "step": 872
    },
    {
      "epoch": 4.7639836289222375,
      "grad_norm": 0.10292503580108092,
      "learning_rate": 5.129773594609688e-07,
      "loss": 0.8541,
      "step": 873
    },
    {
      "epoch": 4.769440654843111,
      "grad_norm": 0.11208161020367986,
      "learning_rate": 4.888899606747544e-07,
      "loss": 0.8348,
      "step": 874
    },
    {
      "epoch": 4.774897680763983,
      "grad_norm": 0.10001712675478809,
      "learning_rate": 4.6537829178053165e-07,
      "loss": 0.8376,
      "step": 875
    },
    {
      "epoch": 4.780354706684856,
      "grad_norm": 0.10211251792319674,
      "learning_rate": 4.424426953748784e-07,
      "loss": 0.8347,
      "step": 876
    },
    {
      "epoch": 4.78581173260573,
      "grad_norm": 0.1007729458690365,
      "learning_rate": 4.200835056602026e-07,
      "loss": 0.82,
      "step": 877
    },
    {
      "epoch": 4.791268758526603,
      "grad_norm": 0.09603334227951416,
      "learning_rate": 3.983010484399019e-07,
      "loss": 0.8404,
      "step": 878
    },
    {
      "epoch": 4.796725784447476,
      "grad_norm": 0.09858339692794088,
      "learning_rate": 3.7709564111361176e-07,
      "loss": 0.8478,
      "step": 879
    },
    {
      "epoch": 4.8021828103683495,
      "grad_norm": 0.1018617706221292,
      "learning_rate": 3.564675926725647e-07,
      "loss": 0.8304,
      "step": 880
    },
    {
      "epoch": 4.807639836289223,
      "grad_norm": 0.11128093657030838,
      "learning_rate": 3.364172036951141e-07,
      "loss": 0.8509,
      "step": 881
    },
    {
      "epoch": 4.813096862210095,
      "grad_norm": 0.10484688329338818,
      "learning_rate": 3.169447663423242e-07,
      "loss": 0.845,
      "step": 882
    },
    {
      "epoch": 4.818553888130968,
      "grad_norm": 0.09768509450771364,
      "learning_rate": 2.9805056435374677e-07,
      "loss": 0.8298,
      "step": 883
    },
    {
      "epoch": 4.824010914051842,
      "grad_norm": 0.10582684400822324,
      "learning_rate": 2.79734873043247e-07,
      "loss": 0.8591,
      "step": 884
    },
    {
      "epoch": 4.829467939972715,
      "grad_norm": 0.09840698310186516,
      "learning_rate": 2.619979592950328e-07,
      "loss": 0.8289,
      "step": 885
    },
    {
      "epoch": 4.834924965893588,
      "grad_norm": 0.0984641962271483,
      "learning_rate": 2.44840081559734e-07,
      "loss": 0.8257,
      "step": 886
    },
    {
      "epoch": 4.8403819918144615,
      "grad_norm": 0.09623214439665646,
      "learning_rate": 2.2826148985065855e-07,
      "loss": 0.8317,
      "step": 887
    },
    {
      "epoch": 4.845839017735334,
      "grad_norm": 0.0979824677211241,
      "learning_rate": 2.12262425740124e-07,
      "loss": 0.8601,
      "step": 888
    },
    {
      "epoch": 4.851296043656207,
      "grad_norm": 0.0973877467431622,
      "learning_rate": 1.9684312235597635e-07,
      "loss": 0.8434,
      "step": 889
    },
    {
      "epoch": 4.85675306957708,
      "grad_norm": 0.10398288588066054,
      "learning_rate": 1.820038043781436e-07,
      "loss": 0.8389,
      "step": 890
    },
    {
      "epoch": 4.862210095497954,
      "grad_norm": 0.10107521480254324,
      "learning_rate": 1.6774468803540723e-07,
      "loss": 0.8467,
      "step": 891
    },
    {
      "epoch": 4.867667121418827,
      "grad_norm": 0.09869803265596083,
      "learning_rate": 1.5406598110222714e-07,
      "loss": 0.8658,
      "step": 892
    },
    {
      "epoch": 4.8731241473397,
      "grad_norm": 0.09751231833043027,
      "learning_rate": 1.4096788289571727e-07,
      "loss": 0.8295,
      "step": 893
    },
    {
      "epoch": 4.878581173260573,
      "grad_norm": 0.10249837791894234,
      "learning_rate": 1.2845058427274126e-07,
      "loss": 0.8336,
      "step": 894
    },
    {
      "epoch": 4.884038199181446,
      "grad_norm": 0.1026402652640808,
      "learning_rate": 1.1651426762714135e-07,
      "loss": 0.847,
      "step": 895
    },
    {
      "epoch": 4.889495225102319,
      "grad_norm": 0.09511533539202549,
      "learning_rate": 1.0515910688706942e-07,
      "loss": 0.8365,
      "step": 896
    },
    {
      "epoch": 4.894952251023192,
      "grad_norm": 0.09998622276585635,
      "learning_rate": 9.43852675124468e-08,
      "loss": 0.8445,
      "step": 897
    },
    {
      "epoch": 4.900409276944066,
      "grad_norm": 0.1110312354933706,
      "learning_rate": 8.419290649257505e-08,
      "loss": 0.842,
      "step": 898
    },
    {
      "epoch": 4.905866302864939,
      "grad_norm": 0.10338515861192228,
      "learning_rate": 7.458217234383114e-08,
      "loss": 0.8529,
      "step": 899
    },
    {
      "epoch": 4.911323328785811,
      "grad_norm": 0.09839456714368296,
      "learning_rate": 6.555320510750474e-08,
      "loss": 0.866,
      "step": 900
    },
    {
      "epoch": 4.916780354706685,
      "grad_norm": 0.09788726743721791,
      "learning_rate": 5.710613634777318e-08,
      "loss": 0.857,
      "step": 901
    },
    {
      "epoch": 4.922237380627558,
      "grad_norm": 0.10202064595622327,
      "learning_rate": 4.92410891497519e-08,
      "loss": 0.8388,
      "step": 902
    },
    {
      "epoch": 4.927694406548431,
      "grad_norm": 0.09501921671184542,
      "learning_rate": 4.195817811774028e-08,
      "loss": 0.8476,
      "step": 903
    },
    {
      "epoch": 4.933151432469304,
      "grad_norm": 0.09497532005633476,
      "learning_rate": 3.5257509373520793e-08,
      "loss": 0.8456,
      "step": 904
    },
    {
      "epoch": 4.938608458390178,
      "grad_norm": 0.09984936254065943,
      "learning_rate": 2.9139180554826894e-08,
      "loss": 0.837,
      "step": 905
    },
    {
      "epoch": 4.94406548431105,
      "grad_norm": 0.09792411355955151,
      "learning_rate": 2.3603280813926376e-08,
      "loss": 0.8349,
      "step": 906
    },
    {
      "epoch": 4.949522510231923,
      "grad_norm": 0.10968982169077339,
      "learning_rate": 1.864989081630242e-08,
      "loss": 0.833,
      "step": 907
    },
    {
      "epoch": 4.954979536152797,
      "grad_norm": 0.10153023843053058,
      "learning_rate": 1.427908273949008e-08,
      "loss": 0.8522,
      "step": 908
    },
    {
      "epoch": 4.96043656207367,
      "grad_norm": 0.10063425319130553,
      "learning_rate": 1.049092027203713e-08,
      "loss": 0.8565,
      "step": 909
    },
    {
      "epoch": 4.965893587994543,
      "grad_norm": 0.10080262727182424,
      "learning_rate": 7.285458612544815e-09,
      "loss": 0.8523,
      "step": 910
    },
    {
      "epoch": 4.971350613915416,
      "grad_norm": 0.10291513147053918,
      "learning_rate": 4.662744468899583e-09,
      "loss": 0.8401,
      "step": 911
    },
    {
      "epoch": 4.97680763983629,
      "grad_norm": 0.10762944685776475,
      "learning_rate": 2.6228160575580976e-09,
      "loss": 0.8275,
      "step": 912
    },
    {
      "epoch": 4.982264665757162,
      "grad_norm": 0.5698686611905339,
      "learning_rate": 1.1657031030232192e-09,
      "loss": 0.8362,
      "step": 913
    },
    {
      "epoch": 4.987721691678035,
      "grad_norm": 0.099846122094202,
      "learning_rate": 2.914268373732654e-10,
      "loss": 0.8625,
      "step": 914
    },
    {
      "epoch": 4.993178717598909,
      "grad_norm": 0.09803245510137683,
      "learning_rate": 0.0,
      "loss": 0.8503,
      "step": 915
    },
    {
      "epoch": 4.993178717598909,
      "step": 915,
      "total_flos": 1.883960626772548e+19,
      "train_loss": 0.9225219208686078,
      "train_runtime": 49142.0532,
      "train_samples_per_second": 9.54,
      "train_steps_per_second": 0.019
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 915,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.883960626772548e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}