{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "eval_steps": 500, "global_step": 18636, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "grad_norm": 44.63691711425781, "learning_rate": 1.0695187165775401e-07, "loss": 2.6868, "step": 1 }, { "epoch": 0.0, "grad_norm": 36.0067024230957, "learning_rate": 2.1390374331550802e-07, "loss": 2.711, "step": 2 }, { "epoch": 0.0, "grad_norm": 37.341617584228516, "learning_rate": 3.208556149732621e-07, "loss": 2.8987, "step": 3 }, { "epoch": 0.0, "grad_norm": 35.09183120727539, "learning_rate": 4.2780748663101604e-07, "loss": 2.7732, "step": 4 }, { "epoch": 0.0, "grad_norm": 35.931148529052734, "learning_rate": 5.347593582887701e-07, "loss": 2.533, "step": 5 }, { "epoch": 0.0, "grad_norm": 33.08338928222656, "learning_rate": 6.417112299465242e-07, "loss": 2.597, "step": 6 }, { "epoch": 0.0, "grad_norm": 35.615570068359375, "learning_rate": 7.486631016042781e-07, "loss": 2.7733, "step": 7 }, { "epoch": 0.0, "grad_norm": 31.061260223388672, "learning_rate": 8.556149732620321e-07, "loss": 2.6151, "step": 8 }, { "epoch": 0.0, "grad_norm": 30.769086837768555, "learning_rate": 9.625668449197862e-07, "loss": 2.6306, "step": 9 }, { "epoch": 0.0, "grad_norm": 29.25857162475586, "learning_rate": 1.0695187165775401e-06, "loss": 2.4867, "step": 10 }, { "epoch": 0.0, "grad_norm": 18.36466407775879, "learning_rate": 1.1764705882352942e-06, "loss": 2.2905, "step": 11 }, { "epoch": 0.0, "grad_norm": 22.19921112060547, "learning_rate": 1.2834224598930483e-06, "loss": 2.5767, "step": 12 }, { "epoch": 0.0, "grad_norm": 23.324115753173828, "learning_rate": 1.3903743315508022e-06, "loss": 2.319, "step": 13 }, { "epoch": 0.0, "grad_norm": 18.079172134399414, "learning_rate": 1.4973262032085562e-06, "loss": 2.3506, "step": 14 }, { "epoch": 0.0, "grad_norm": 21.01787757873535, "learning_rate": 1.6042780748663103e-06, "loss": 2.1042, "step": 15 }, { "epoch": 0.0, "grad_norm": 21.856233596801758, "learning_rate": 1.7112299465240642e-06, "loss": 2.2163, "step": 16 }, { "epoch": 0.0, "grad_norm": 26.98659896850586, "learning_rate": 1.8181818181818183e-06, "loss": 2.3461, "step": 17 }, { "epoch": 0.0, "grad_norm": 23.997295379638672, "learning_rate": 1.9251336898395724e-06, "loss": 2.1442, "step": 18 }, { "epoch": 0.0, "grad_norm": 23.490922927856445, "learning_rate": 2.0320855614973265e-06, "loss": 2.3537, "step": 19 }, { "epoch": 0.0, "grad_norm": 17.00440788269043, "learning_rate": 2.1390374331550802e-06, "loss": 2.1197, "step": 20 }, { "epoch": 0.0, "grad_norm": 17.1279239654541, "learning_rate": 2.2459893048128343e-06, "loss": 2.2151, "step": 21 }, { "epoch": 0.0, "grad_norm": 24.662363052368164, "learning_rate": 2.3529411764705885e-06, "loss": 2.0372, "step": 22 }, { "epoch": 0.0, "grad_norm": 21.704465866088867, "learning_rate": 2.4598930481283426e-06, "loss": 2.0636, "step": 23 }, { "epoch": 0.0, "grad_norm": 17.163585662841797, "learning_rate": 2.5668449197860967e-06, "loss": 2.0514, "step": 24 }, { "epoch": 0.0, "grad_norm": 16.522436141967773, "learning_rate": 2.673796791443851e-06, "loss": 2.0283, "step": 25 }, { "epoch": 0.0, "grad_norm": 12.984575271606445, "learning_rate": 2.7807486631016045e-06, "loss": 2.1394, "step": 26 }, { "epoch": 0.0, "grad_norm": 13.666719436645508, "learning_rate": 2.8877005347593586e-06, "loss": 1.9508, "step": 27 }, { "epoch": 0.0, "grad_norm": 11.294673919677734, "learning_rate": 2.9946524064171123e-06, "loss": 1.9075, "step": 28 }, { "epoch": 0.0, "grad_norm": 12.251426696777344, "learning_rate": 3.101604278074867e-06, "loss": 2.0909, "step": 29 }, { "epoch": 0.0, "grad_norm": 14.695710182189941, "learning_rate": 3.2085561497326205e-06, "loss": 1.9844, "step": 30 }, { "epoch": 0.0, "grad_norm": 10.683698654174805, "learning_rate": 3.3155080213903747e-06, "loss": 2.147, "step": 31 }, { "epoch": 0.01, "grad_norm": 14.114949226379395, "learning_rate": 3.4224598930481284e-06, "loss": 1.9331, "step": 32 }, { "epoch": 0.01, "grad_norm": 12.795902252197266, "learning_rate": 3.529411764705883e-06, "loss": 2.0854, "step": 33 }, { "epoch": 0.01, "grad_norm": 9.970728874206543, "learning_rate": 3.6363636363636366e-06, "loss": 1.8108, "step": 34 }, { "epoch": 0.01, "grad_norm": 12.814021110534668, "learning_rate": 3.7433155080213907e-06, "loss": 2.3158, "step": 35 }, { "epoch": 0.01, "grad_norm": 13.256742477416992, "learning_rate": 3.850267379679145e-06, "loss": 2.0202, "step": 36 }, { "epoch": 0.01, "grad_norm": 13.048867225646973, "learning_rate": 3.957219251336899e-06, "loss": 1.8974, "step": 37 }, { "epoch": 0.01, "grad_norm": 13.777260780334473, "learning_rate": 4.064171122994653e-06, "loss": 1.9419, "step": 38 }, { "epoch": 0.01, "grad_norm": 10.576271057128906, "learning_rate": 4.171122994652407e-06, "loss": 1.9857, "step": 39 }, { "epoch": 0.01, "grad_norm": 13.487005233764648, "learning_rate": 4.2780748663101604e-06, "loss": 1.9898, "step": 40 }, { "epoch": 0.01, "grad_norm": 11.693231582641602, "learning_rate": 4.385026737967915e-06, "loss": 2.0688, "step": 41 }, { "epoch": 0.01, "grad_norm": 11.535014152526855, "learning_rate": 4.491978609625669e-06, "loss": 1.8813, "step": 42 }, { "epoch": 0.01, "grad_norm": 9.708033561706543, "learning_rate": 4.598930481283423e-06, "loss": 1.9555, "step": 43 }, { "epoch": 0.01, "grad_norm": 10.739907264709473, "learning_rate": 4.705882352941177e-06, "loss": 1.9181, "step": 44 }, { "epoch": 0.01, "grad_norm": 11.725189208984375, "learning_rate": 4.812834224598931e-06, "loss": 2.142, "step": 45 }, { "epoch": 0.01, "grad_norm": 10.724346160888672, "learning_rate": 4.919786096256685e-06, "loss": 2.0172, "step": 46 }, { "epoch": 0.01, "grad_norm": 13.440933227539062, "learning_rate": 5.026737967914439e-06, "loss": 1.8154, "step": 47 }, { "epoch": 0.01, "grad_norm": 10.922639846801758, "learning_rate": 5.133689839572193e-06, "loss": 2.0259, "step": 48 }, { "epoch": 0.01, "grad_norm": 11.972490310668945, "learning_rate": 5.240641711229947e-06, "loss": 2.0184, "step": 49 }, { "epoch": 0.01, "grad_norm": 13.31084156036377, "learning_rate": 5.347593582887702e-06, "loss": 2.0606, "step": 50 }, { "epoch": 0.01, "grad_norm": 14.009495735168457, "learning_rate": 5.4545454545454545e-06, "loss": 1.918, "step": 51 }, { "epoch": 0.01, "grad_norm": 12.032031059265137, "learning_rate": 5.561497326203209e-06, "loss": 2.1064, "step": 52 }, { "epoch": 0.01, "grad_norm": 11.633992195129395, "learning_rate": 5.6684491978609635e-06, "loss": 1.829, "step": 53 }, { "epoch": 0.01, "grad_norm": 12.172515869140625, "learning_rate": 5.775401069518717e-06, "loss": 1.8735, "step": 54 }, { "epoch": 0.01, "grad_norm": 10.599992752075195, "learning_rate": 5.882352941176471e-06, "loss": 1.9931, "step": 55 }, { "epoch": 0.01, "grad_norm": 11.375288009643555, "learning_rate": 5.989304812834225e-06, "loss": 2.0912, "step": 56 }, { "epoch": 0.01, "grad_norm": 12.809346199035645, "learning_rate": 6.096256684491979e-06, "loss": 2.0183, "step": 57 }, { "epoch": 0.01, "grad_norm": 10.441542625427246, "learning_rate": 6.203208556149734e-06, "loss": 1.8861, "step": 58 }, { "epoch": 0.01, "grad_norm": 12.52021598815918, "learning_rate": 6.3101604278074865e-06, "loss": 2.0271, "step": 59 }, { "epoch": 0.01, "grad_norm": 12.20396614074707, "learning_rate": 6.417112299465241e-06, "loss": 1.9133, "step": 60 }, { "epoch": 0.01, "grad_norm": 10.227400779724121, "learning_rate": 6.524064171122996e-06, "loss": 1.8695, "step": 61 }, { "epoch": 0.01, "grad_norm": 11.501531600952148, "learning_rate": 6.631016042780749e-06, "loss": 1.8956, "step": 62 }, { "epoch": 0.01, "grad_norm": 12.04661750793457, "learning_rate": 6.737967914438504e-06, "loss": 1.8162, "step": 63 }, { "epoch": 0.01, "grad_norm": 11.101337432861328, "learning_rate": 6.844919786096257e-06, "loss": 1.9691, "step": 64 }, { "epoch": 0.01, "grad_norm": 12.107738494873047, "learning_rate": 6.951871657754011e-06, "loss": 1.9308, "step": 65 }, { "epoch": 0.01, "grad_norm": 12.799836158752441, "learning_rate": 7.058823529411766e-06, "loss": 1.9958, "step": 66 }, { "epoch": 0.01, "grad_norm": 12.854655265808105, "learning_rate": 7.1657754010695195e-06, "loss": 2.123, "step": 67 }, { "epoch": 0.01, "grad_norm": 10.8805570602417, "learning_rate": 7.272727272727273e-06, "loss": 1.9131, "step": 68 }, { "epoch": 0.01, "grad_norm": 14.97570514678955, "learning_rate": 7.379679144385027e-06, "loss": 1.9453, "step": 69 }, { "epoch": 0.01, "grad_norm": 11.656270980834961, "learning_rate": 7.486631016042781e-06, "loss": 1.9848, "step": 70 }, { "epoch": 0.01, "grad_norm": 12.01232624053955, "learning_rate": 7.593582887700536e-06, "loss": 1.9999, "step": 71 }, { "epoch": 0.01, "grad_norm": 13.459454536437988, "learning_rate": 7.70053475935829e-06, "loss": 1.9119, "step": 72 }, { "epoch": 0.01, "grad_norm": 11.674161911010742, "learning_rate": 7.807486631016043e-06, "loss": 1.8346, "step": 73 }, { "epoch": 0.01, "grad_norm": 11.570338249206543, "learning_rate": 7.914438502673799e-06, "loss": 1.8871, "step": 74 }, { "epoch": 0.01, "grad_norm": 13.990143775939941, "learning_rate": 8.02139037433155e-06, "loss": 1.9189, "step": 75 }, { "epoch": 0.01, "grad_norm": 14.802459716796875, "learning_rate": 8.128342245989306e-06, "loss": 1.9734, "step": 76 }, { "epoch": 0.01, "grad_norm": 12.477425575256348, "learning_rate": 8.23529411764706e-06, "loss": 2.0989, "step": 77 }, { "epoch": 0.01, "grad_norm": 13.244380950927734, "learning_rate": 8.342245989304813e-06, "loss": 1.9535, "step": 78 }, { "epoch": 0.01, "grad_norm": 15.053714752197266, "learning_rate": 8.449197860962567e-06, "loss": 1.9334, "step": 79 }, { "epoch": 0.01, "grad_norm": 12.626097679138184, "learning_rate": 8.556149732620321e-06, "loss": 1.935, "step": 80 }, { "epoch": 0.01, "grad_norm": 13.302371978759766, "learning_rate": 8.663101604278076e-06, "loss": 2.2032, "step": 81 }, { "epoch": 0.01, "grad_norm": 13.190651893615723, "learning_rate": 8.77005347593583e-06, "loss": 1.7957, "step": 82 }, { "epoch": 0.01, "grad_norm": 13.229026794433594, "learning_rate": 8.877005347593584e-06, "loss": 1.9089, "step": 83 }, { "epoch": 0.01, "grad_norm": 11.914862632751465, "learning_rate": 8.983957219251337e-06, "loss": 1.8117, "step": 84 }, { "epoch": 0.01, "grad_norm": 16.635679244995117, "learning_rate": 9.090909090909091e-06, "loss": 2.0111, "step": 85 }, { "epoch": 0.01, "grad_norm": 13.707334518432617, "learning_rate": 9.197860962566846e-06, "loss": 1.8986, "step": 86 }, { "epoch": 0.01, "grad_norm": 11.104602813720703, "learning_rate": 9.3048128342246e-06, "loss": 1.8556, "step": 87 }, { "epoch": 0.01, "grad_norm": 15.605399131774902, "learning_rate": 9.411764705882354e-06, "loss": 1.7362, "step": 88 }, { "epoch": 0.01, "grad_norm": 13.368927001953125, "learning_rate": 9.518716577540108e-06, "loss": 1.9554, "step": 89 }, { "epoch": 0.01, "grad_norm": 16.498302459716797, "learning_rate": 9.625668449197861e-06, "loss": 1.8877, "step": 90 }, { "epoch": 0.01, "grad_norm": 13.428933143615723, "learning_rate": 9.732620320855617e-06, "loss": 1.9237, "step": 91 }, { "epoch": 0.01, "grad_norm": 18.570425033569336, "learning_rate": 9.83957219251337e-06, "loss": 1.9216, "step": 92 }, { "epoch": 0.01, "grad_norm": 12.990293502807617, "learning_rate": 9.946524064171124e-06, "loss": 2.0676, "step": 93 }, { "epoch": 0.02, "grad_norm": 10.910529136657715, "learning_rate": 1.0053475935828878e-05, "loss": 1.7468, "step": 94 }, { "epoch": 0.02, "grad_norm": 13.541019439697266, "learning_rate": 1.0160427807486633e-05, "loss": 1.8191, "step": 95 }, { "epoch": 0.02, "grad_norm": 13.767446517944336, "learning_rate": 1.0267379679144387e-05, "loss": 1.9911, "step": 96 }, { "epoch": 0.02, "grad_norm": 12.26842212677002, "learning_rate": 1.0374331550802139e-05, "loss": 1.9492, "step": 97 }, { "epoch": 0.02, "grad_norm": 13.516999244689941, "learning_rate": 1.0481283422459894e-05, "loss": 2.0408, "step": 98 }, { "epoch": 0.02, "grad_norm": 11.88028335571289, "learning_rate": 1.0588235294117648e-05, "loss": 1.9005, "step": 99 }, { "epoch": 0.02, "grad_norm": 15.450510025024414, "learning_rate": 1.0695187165775403e-05, "loss": 2.0488, "step": 100 }, { "epoch": 0.02, "grad_norm": 14.716365814208984, "learning_rate": 1.0802139037433157e-05, "loss": 2.0103, "step": 101 }, { "epoch": 0.02, "grad_norm": 10.32108211517334, "learning_rate": 1.0909090909090909e-05, "loss": 1.9501, "step": 102 }, { "epoch": 0.02, "grad_norm": 10.909945487976074, "learning_rate": 1.1016042780748664e-05, "loss": 1.8484, "step": 103 }, { "epoch": 0.02, "grad_norm": 12.217025756835938, "learning_rate": 1.1122994652406418e-05, "loss": 1.7171, "step": 104 }, { "epoch": 0.02, "grad_norm": 11.380084037780762, "learning_rate": 1.1229946524064172e-05, "loss": 1.9766, "step": 105 }, { "epoch": 0.02, "grad_norm": 11.365348815917969, "learning_rate": 1.1336898395721927e-05, "loss": 1.8776, "step": 106 }, { "epoch": 0.02, "grad_norm": 11.926719665527344, "learning_rate": 1.1443850267379679e-05, "loss": 1.8909, "step": 107 }, { "epoch": 0.02, "grad_norm": 11.596673965454102, "learning_rate": 1.1550802139037434e-05, "loss": 2.0632, "step": 108 }, { "epoch": 0.02, "grad_norm": 10.11067008972168, "learning_rate": 1.1657754010695188e-05, "loss": 2.0485, "step": 109 }, { "epoch": 0.02, "grad_norm": 12.637042045593262, "learning_rate": 1.1764705882352942e-05, "loss": 1.9886, "step": 110 }, { "epoch": 0.02, "grad_norm": 10.708076477050781, "learning_rate": 1.1871657754010697e-05, "loss": 2.0109, "step": 111 }, { "epoch": 0.02, "grad_norm": 12.536702156066895, "learning_rate": 1.197860962566845e-05, "loss": 1.8694, "step": 112 }, { "epoch": 0.02, "grad_norm": 12.00039005279541, "learning_rate": 1.2085561497326203e-05, "loss": 1.9811, "step": 113 }, { "epoch": 0.02, "grad_norm": 10.211718559265137, "learning_rate": 1.2192513368983958e-05, "loss": 1.9113, "step": 114 }, { "epoch": 0.02, "grad_norm": 11.412898063659668, "learning_rate": 1.2299465240641712e-05, "loss": 2.0212, "step": 115 }, { "epoch": 0.02, "grad_norm": 12.235512733459473, "learning_rate": 1.2406417112299467e-05, "loss": 2.0333, "step": 116 }, { "epoch": 0.02, "grad_norm": 10.052408218383789, "learning_rate": 1.251336898395722e-05, "loss": 1.8429, "step": 117 }, { "epoch": 0.02, "grad_norm": 11.330059051513672, "learning_rate": 1.2620320855614973e-05, "loss": 2.0779, "step": 118 }, { "epoch": 0.02, "grad_norm": 11.376105308532715, "learning_rate": 1.2727272727272728e-05, "loss": 1.9569, "step": 119 }, { "epoch": 0.02, "grad_norm": 10.46397590637207, "learning_rate": 1.2834224598930482e-05, "loss": 1.966, "step": 120 }, { "epoch": 0.02, "grad_norm": 9.899772644042969, "learning_rate": 1.2941176470588238e-05, "loss": 1.9313, "step": 121 }, { "epoch": 0.02, "grad_norm": 11.327566146850586, "learning_rate": 1.3048128342245991e-05, "loss": 2.112, "step": 122 }, { "epoch": 0.02, "grad_norm": 10.688664436340332, "learning_rate": 1.3155080213903743e-05, "loss": 1.896, "step": 123 }, { "epoch": 0.02, "grad_norm": 11.84473705291748, "learning_rate": 1.3262032085561499e-05, "loss": 1.8922, "step": 124 }, { "epoch": 0.02, "grad_norm": 12.897757530212402, "learning_rate": 1.3368983957219252e-05, "loss": 2.1142, "step": 125 }, { "epoch": 0.02, "grad_norm": 12.044815063476562, "learning_rate": 1.3475935828877008e-05, "loss": 2.0517, "step": 126 }, { "epoch": 0.02, "grad_norm": 11.477570533752441, "learning_rate": 1.3582887700534761e-05, "loss": 1.8117, "step": 127 }, { "epoch": 0.02, "grad_norm": 9.955909729003906, "learning_rate": 1.3689839572192513e-05, "loss": 1.853, "step": 128 }, { "epoch": 0.02, "grad_norm": 11.939245223999023, "learning_rate": 1.3796791443850269e-05, "loss": 1.9396, "step": 129 }, { "epoch": 0.02, "grad_norm": 10.718303680419922, "learning_rate": 1.3903743315508022e-05, "loss": 1.8371, "step": 130 }, { "epoch": 0.02, "grad_norm": 13.616842269897461, "learning_rate": 1.4010695187165778e-05, "loss": 1.9804, "step": 131 }, { "epoch": 0.02, "grad_norm": 11.35547924041748, "learning_rate": 1.4117647058823532e-05, "loss": 2.0716, "step": 132 }, { "epoch": 0.02, "grad_norm": 9.926712989807129, "learning_rate": 1.4224598930481284e-05, "loss": 1.8981, "step": 133 }, { "epoch": 0.02, "grad_norm": 10.517195701599121, "learning_rate": 1.4331550802139039e-05, "loss": 2.0158, "step": 134 }, { "epoch": 0.02, "grad_norm": 11.402612686157227, "learning_rate": 1.4438502673796793e-05, "loss": 1.8539, "step": 135 }, { "epoch": 0.02, "grad_norm": 9.772167205810547, "learning_rate": 1.4545454545454546e-05, "loss": 1.9602, "step": 136 }, { "epoch": 0.02, "grad_norm": 11.135686874389648, "learning_rate": 1.4652406417112302e-05, "loss": 2.058, "step": 137 }, { "epoch": 0.02, "grad_norm": 12.15201187133789, "learning_rate": 1.4759358288770054e-05, "loss": 1.7949, "step": 138 }, { "epoch": 0.02, "grad_norm": 10.518499374389648, "learning_rate": 1.4866310160427807e-05, "loss": 1.9903, "step": 139 }, { "epoch": 0.02, "grad_norm": 10.60933780670166, "learning_rate": 1.4973262032085563e-05, "loss": 1.9291, "step": 140 }, { "epoch": 0.02, "grad_norm": 11.67188835144043, "learning_rate": 1.5080213903743316e-05, "loss": 2.0427, "step": 141 }, { "epoch": 0.02, "grad_norm": 12.441561698913574, "learning_rate": 1.5187165775401072e-05, "loss": 1.8618, "step": 142 }, { "epoch": 0.02, "grad_norm": 8.59052848815918, "learning_rate": 1.5294117647058822e-05, "loss": 1.8315, "step": 143 }, { "epoch": 0.02, "grad_norm": 9.600971221923828, "learning_rate": 1.540106951871658e-05, "loss": 1.8358, "step": 144 }, { "epoch": 0.02, "grad_norm": 10.16581916809082, "learning_rate": 1.5508021390374333e-05, "loss": 2.0162, "step": 145 }, { "epoch": 0.02, "grad_norm": 10.901065826416016, "learning_rate": 1.5614973262032087e-05, "loss": 1.8596, "step": 146 }, { "epoch": 0.02, "grad_norm": 9.497127532958984, "learning_rate": 1.572192513368984e-05, "loss": 1.9845, "step": 147 }, { "epoch": 0.02, "grad_norm": 10.089396476745605, "learning_rate": 1.5828877005347597e-05, "loss": 1.9639, "step": 148 }, { "epoch": 0.02, "grad_norm": 11.236745834350586, "learning_rate": 1.5935828877005348e-05, "loss": 2.0153, "step": 149 }, { "epoch": 0.02, "grad_norm": 12.315768241882324, "learning_rate": 1.60427807486631e-05, "loss": 2.1208, "step": 150 }, { "epoch": 0.02, "grad_norm": 11.593655586242676, "learning_rate": 1.614973262032086e-05, "loss": 1.9484, "step": 151 }, { "epoch": 0.02, "grad_norm": 9.006234169006348, "learning_rate": 1.6256684491978612e-05, "loss": 1.9488, "step": 152 }, { "epoch": 0.02, "grad_norm": 8.930652618408203, "learning_rate": 1.6363636363636366e-05, "loss": 1.8737, "step": 153 }, { "epoch": 0.02, "grad_norm": 10.18005084991455, "learning_rate": 1.647058823529412e-05, "loss": 2.1562, "step": 154 }, { "epoch": 0.02, "grad_norm": 10.79998779296875, "learning_rate": 1.6577540106951873e-05, "loss": 1.9431, "step": 155 }, { "epoch": 0.03, "grad_norm": 9.749403953552246, "learning_rate": 1.6684491978609627e-05, "loss": 1.9495, "step": 156 }, { "epoch": 0.03, "grad_norm": 9.75282096862793, "learning_rate": 1.679144385026738e-05, "loss": 2.0049, "step": 157 }, { "epoch": 0.03, "grad_norm": 9.932050704956055, "learning_rate": 1.6898395721925134e-05, "loss": 2.0635, "step": 158 }, { "epoch": 0.03, "grad_norm": 12.430822372436523, "learning_rate": 1.7005347593582888e-05, "loss": 1.893, "step": 159 }, { "epoch": 0.03, "grad_norm": 21.40743064880371, "learning_rate": 1.7112299465240642e-05, "loss": 1.9139, "step": 160 }, { "epoch": 0.03, "grad_norm": 35.722129821777344, "learning_rate": 1.7219251336898395e-05, "loss": 2.3321, "step": 161 }, { "epoch": 0.03, "grad_norm": 16.273971557617188, "learning_rate": 1.7326203208556153e-05, "loss": 2.1609, "step": 162 }, { "epoch": 0.03, "grad_norm": 15.158163070678711, "learning_rate": 1.7433155080213906e-05, "loss": 1.7769, "step": 163 }, { "epoch": 0.03, "grad_norm": 10.210084915161133, "learning_rate": 1.754010695187166e-05, "loss": 2.0348, "step": 164 }, { "epoch": 0.03, "grad_norm": 12.014740943908691, "learning_rate": 1.7647058823529414e-05, "loss": 1.9914, "step": 165 }, { "epoch": 0.03, "grad_norm": 11.81379222869873, "learning_rate": 1.7754010695187167e-05, "loss": 2.0319, "step": 166 }, { "epoch": 0.03, "grad_norm": 11.027584075927734, "learning_rate": 1.786096256684492e-05, "loss": 1.99, "step": 167 }, { "epoch": 0.03, "grad_norm": 12.567806243896484, "learning_rate": 1.7967914438502675e-05, "loss": 2.1734, "step": 168 }, { "epoch": 0.03, "grad_norm": 10.91496467590332, "learning_rate": 1.807486631016043e-05, "loss": 2.0696, "step": 169 }, { "epoch": 0.03, "grad_norm": 9.668896675109863, "learning_rate": 1.8181818181818182e-05, "loss": 2.0032, "step": 170 }, { "epoch": 0.03, "grad_norm": 10.977431297302246, "learning_rate": 1.8288770053475936e-05, "loss": 1.8887, "step": 171 }, { "epoch": 0.03, "grad_norm": 9.254753112792969, "learning_rate": 1.8395721925133693e-05, "loss": 1.9735, "step": 172 }, { "epoch": 0.03, "grad_norm": 12.16599178314209, "learning_rate": 1.8502673796791447e-05, "loss": 2.3243, "step": 173 }, { "epoch": 0.03, "grad_norm": 10.125239372253418, "learning_rate": 1.86096256684492e-05, "loss": 1.9803, "step": 174 }, { "epoch": 0.03, "grad_norm": 9.142389297485352, "learning_rate": 1.8716577540106954e-05, "loss": 2.0306, "step": 175 }, { "epoch": 0.03, "grad_norm": 8.698577880859375, "learning_rate": 1.8823529411764708e-05, "loss": 1.9568, "step": 176 }, { "epoch": 0.03, "grad_norm": 10.644491195678711, "learning_rate": 1.893048128342246e-05, "loss": 2.0299, "step": 177 }, { "epoch": 0.03, "grad_norm": 10.544769287109375, "learning_rate": 1.9037433155080215e-05, "loss": 1.9606, "step": 178 }, { "epoch": 0.03, "grad_norm": 9.659712791442871, "learning_rate": 1.9144385026737972e-05, "loss": 2.1088, "step": 179 }, { "epoch": 0.03, "grad_norm": 11.61441707611084, "learning_rate": 1.9251336898395722e-05, "loss": 2.0271, "step": 180 }, { "epoch": 0.03, "grad_norm": 9.993927955627441, "learning_rate": 1.9358288770053476e-05, "loss": 1.9861, "step": 181 }, { "epoch": 0.03, "grad_norm": 15.27168083190918, "learning_rate": 1.9465240641711233e-05, "loss": 1.8246, "step": 182 }, { "epoch": 0.03, "grad_norm": 13.195595741271973, "learning_rate": 1.9572192513368987e-05, "loss": 1.9966, "step": 183 }, { "epoch": 0.03, "grad_norm": 9.861351013183594, "learning_rate": 1.967914438502674e-05, "loss": 1.7295, "step": 184 }, { "epoch": 0.03, "grad_norm": 11.325094223022461, "learning_rate": 1.9786096256684494e-05, "loss": 2.0516, "step": 185 }, { "epoch": 0.03, "grad_norm": 9.599010467529297, "learning_rate": 1.9893048128342248e-05, "loss": 2.1291, "step": 186 }, { "epoch": 0.03, "grad_norm": 11.839276313781738, "learning_rate": 2e-05, "loss": 2.0306, "step": 187 }, { "epoch": 0.03, "grad_norm": 10.814986228942871, "learning_rate": 1.9999998640573795e-05, "loss": 2.0451, "step": 188 }, { "epoch": 0.03, "grad_norm": 12.093156814575195, "learning_rate": 1.999999456229555e-05, "loss": 1.8765, "step": 189 }, { "epoch": 0.03, "grad_norm": 12.045069694519043, "learning_rate": 1.9999987765166372e-05, "loss": 1.6989, "step": 190 }, { "epoch": 0.03, "grad_norm": 10.564959526062012, "learning_rate": 1.9999978249188113e-05, "loss": 2.0626, "step": 191 }, { "epoch": 0.03, "grad_norm": 11.577120780944824, "learning_rate": 1.9999966014363353e-05, "loss": 2.017, "step": 192 }, { "epoch": 0.03, "grad_norm": 10.69091796875, "learning_rate": 1.9999951060695427e-05, "loss": 1.7902, "step": 193 }, { "epoch": 0.03, "grad_norm": 10.30361270904541, "learning_rate": 1.999993338818839e-05, "loss": 2.1949, "step": 194 }, { "epoch": 0.03, "grad_norm": 11.8187255859375, "learning_rate": 1.9999912996847062e-05, "loss": 2.0871, "step": 195 }, { "epoch": 0.03, "grad_norm": 9.428878784179688, "learning_rate": 1.9999889886676974e-05, "loss": 2.0844, "step": 196 }, { "epoch": 0.03, "grad_norm": 11.041136741638184, "learning_rate": 1.9999864057684414e-05, "loss": 2.0277, "step": 197 }, { "epoch": 0.03, "grad_norm": 10.993311882019043, "learning_rate": 1.9999835509876406e-05, "loss": 2.1339, "step": 198 }, { "epoch": 0.03, "grad_norm": 9.714010238647461, "learning_rate": 1.999980424326071e-05, "loss": 1.9492, "step": 199 }, { "epoch": 0.03, "grad_norm": 9.192052841186523, "learning_rate": 1.9999770257845824e-05, "loss": 1.8637, "step": 200 }, { "epoch": 0.03, "grad_norm": 15.065098762512207, "learning_rate": 1.9999733553640997e-05, "loss": 2.2376, "step": 201 }, { "epoch": 0.03, "grad_norm": 9.977725982666016, "learning_rate": 1.99996941306562e-05, "loss": 2.0133, "step": 202 }, { "epoch": 0.03, "grad_norm": 10.887410163879395, "learning_rate": 1.999965198890215e-05, "loss": 2.162, "step": 203 }, { "epoch": 0.03, "grad_norm": 11.465036392211914, "learning_rate": 1.9999607128390313e-05, "loss": 1.9935, "step": 204 }, { "epoch": 0.03, "grad_norm": 13.521655082702637, "learning_rate": 1.9999559549132885e-05, "loss": 2.1109, "step": 205 }, { "epoch": 0.03, "grad_norm": 20.360754013061523, "learning_rate": 1.9999509251142794e-05, "loss": 2.0602, "step": 206 }, { "epoch": 0.03, "grad_norm": 16.762718200683594, "learning_rate": 1.999945623443372e-05, "loss": 2.2379, "step": 207 }, { "epoch": 0.03, "grad_norm": 16.801115036010742, "learning_rate": 1.999940049902008e-05, "loss": 2.0681, "step": 208 }, { "epoch": 0.03, "grad_norm": 11.889473915100098, "learning_rate": 1.9999342044917024e-05, "loss": 2.1233, "step": 209 }, { "epoch": 0.03, "grad_norm": 14.037755012512207, "learning_rate": 1.9999280872140442e-05, "loss": 2.1166, "step": 210 }, { "epoch": 0.03, "grad_norm": 9.300267219543457, "learning_rate": 1.9999216980706974e-05, "loss": 2.1527, "step": 211 }, { "epoch": 0.03, "grad_norm": 11.220389366149902, "learning_rate": 1.9999150370633987e-05, "loss": 2.1892, "step": 212 }, { "epoch": 0.03, "grad_norm": 10.553252220153809, "learning_rate": 1.9999081041939594e-05, "loss": 1.9629, "step": 213 }, { "epoch": 0.03, "grad_norm": 11.263272285461426, "learning_rate": 1.999900899464264e-05, "loss": 2.0378, "step": 214 }, { "epoch": 0.03, "grad_norm": 9.766071319580078, "learning_rate": 1.9998934228762714e-05, "loss": 2.224, "step": 215 }, { "epoch": 0.03, "grad_norm": 10.724212646484375, "learning_rate": 1.999885674432015e-05, "loss": 2.0347, "step": 216 }, { "epoch": 0.03, "grad_norm": 12.171621322631836, "learning_rate": 1.9998776541336002e-05, "loss": 1.935, "step": 217 }, { "epoch": 0.04, "grad_norm": 10.975146293640137, "learning_rate": 1.9998693619832087e-05, "loss": 1.8327, "step": 218 }, { "epoch": 0.04, "grad_norm": 9.847110748291016, "learning_rate": 1.999860797983095e-05, "loss": 1.9814, "step": 219 }, { "epoch": 0.04, "grad_norm": 9.50926399230957, "learning_rate": 1.999851962135587e-05, "loss": 1.9532, "step": 220 }, { "epoch": 0.04, "grad_norm": 11.191120147705078, "learning_rate": 1.9998428544430876e-05, "loss": 2.0356, "step": 221 }, { "epoch": 0.04, "grad_norm": 11.014785766601562, "learning_rate": 1.9998334749080723e-05, "loss": 2.0763, "step": 222 }, { "epoch": 0.04, "grad_norm": 15.525216102600098, "learning_rate": 1.999823823533092e-05, "loss": 1.9672, "step": 223 }, { "epoch": 0.04, "grad_norm": 12.781306266784668, "learning_rate": 1.9998139003207704e-05, "loss": 2.0868, "step": 224 }, { "epoch": 0.04, "grad_norm": 12.24928092956543, "learning_rate": 1.9998037052738055e-05, "loss": 2.1438, "step": 225 }, { "epoch": 0.04, "grad_norm": 12.256040573120117, "learning_rate": 1.9997932383949692e-05, "loss": 1.7282, "step": 226 }, { "epoch": 0.04, "grad_norm": 12.000025749206543, "learning_rate": 1.9997824996871072e-05, "loss": 1.8686, "step": 227 }, { "epoch": 0.04, "grad_norm": 8.496598243713379, "learning_rate": 1.9997714891531394e-05, "loss": 1.8924, "step": 228 }, { "epoch": 0.04, "grad_norm": 10.139318466186523, "learning_rate": 1.9997602067960593e-05, "loss": 2.0439, "step": 229 }, { "epoch": 0.04, "grad_norm": 15.80923080444336, "learning_rate": 1.9997486526189343e-05, "loss": 2.0934, "step": 230 }, { "epoch": 0.04, "grad_norm": 9.459860801696777, "learning_rate": 1.9997368266249064e-05, "loss": 2.0005, "step": 231 }, { "epoch": 0.04, "grad_norm": 12.109635353088379, "learning_rate": 1.99972472881719e-05, "loss": 2.0453, "step": 232 }, { "epoch": 0.04, "grad_norm": 10.437554359436035, "learning_rate": 1.999712359199075e-05, "loss": 1.915, "step": 233 }, { "epoch": 0.04, "grad_norm": 12.782486915588379, "learning_rate": 1.999699717773924e-05, "loss": 1.9672, "step": 234 }, { "epoch": 0.04, "grad_norm": 9.617237091064453, "learning_rate": 1.999686804545174e-05, "loss": 2.0296, "step": 235 }, { "epoch": 0.04, "grad_norm": 12.224469184875488, "learning_rate": 1.999673619516337e-05, "loss": 1.9169, "step": 236 }, { "epoch": 0.04, "grad_norm": 12.650546073913574, "learning_rate": 1.9996601626909962e-05, "loss": 2.0043, "step": 237 }, { "epoch": 0.04, "grad_norm": 12.657430648803711, "learning_rate": 1.9996464340728116e-05, "loss": 2.0395, "step": 238 }, { "epoch": 0.04, "grad_norm": 10.872553825378418, "learning_rate": 1.999632433665515e-05, "loss": 1.9531, "step": 239 }, { "epoch": 0.04, "grad_norm": 9.720267295837402, "learning_rate": 1.999618161472914e-05, "loss": 1.9684, "step": 240 }, { "epoch": 0.04, "grad_norm": 13.341397285461426, "learning_rate": 1.9996036174988877e-05, "loss": 1.9343, "step": 241 }, { "epoch": 0.04, "grad_norm": 11.776782035827637, "learning_rate": 1.9995888017473907e-05, "loss": 1.9597, "step": 242 }, { "epoch": 0.04, "grad_norm": 10.44223403930664, "learning_rate": 1.999573714222452e-05, "loss": 2.0767, "step": 243 }, { "epoch": 0.04, "grad_norm": 10.733047485351562, "learning_rate": 1.9995583549281722e-05, "loss": 1.9439, "step": 244 }, { "epoch": 0.04, "grad_norm": 13.397958755493164, "learning_rate": 1.999542723868729e-05, "loss": 2.0957, "step": 245 }, { "epoch": 0.04, "grad_norm": 13.132685661315918, "learning_rate": 1.999526821048371e-05, "loss": 2.1412, "step": 246 }, { "epoch": 0.04, "grad_norm": 12.34654712677002, "learning_rate": 1.9995106464714226e-05, "loss": 2.005, "step": 247 }, { "epoch": 0.04, "grad_norm": 10.84194564819336, "learning_rate": 1.9994942001422807e-05, "loss": 1.933, "step": 248 }, { "epoch": 0.04, "grad_norm": 10.143826484680176, "learning_rate": 1.9994774820654176e-05, "loss": 2.1562, "step": 249 }, { "epoch": 0.04, "grad_norm": 13.250110626220703, "learning_rate": 1.9994604922453785e-05, "loss": 1.9662, "step": 250 }, { "epoch": 0.04, "grad_norm": 12.293492317199707, "learning_rate": 1.9994432306867826e-05, "loss": 1.8468, "step": 251 }, { "epoch": 0.04, "grad_norm": 7.878584384918213, "learning_rate": 1.999425697394323e-05, "loss": 1.8498, "step": 252 }, { "epoch": 0.04, "grad_norm": 10.573433876037598, "learning_rate": 1.9994078923727666e-05, "loss": 2.013, "step": 253 }, { "epoch": 0.04, "grad_norm": 11.056818008422852, "learning_rate": 1.9993898156269545e-05, "loss": 2.0266, "step": 254 }, { "epoch": 0.04, "grad_norm": 13.916166305541992, "learning_rate": 1.9993714671618014e-05, "loss": 2.2129, "step": 255 }, { "epoch": 0.04, "grad_norm": 8.506329536437988, "learning_rate": 1.9993528469822966e-05, "loss": 1.9849, "step": 256 }, { "epoch": 0.04, "grad_norm": 14.011886596679688, "learning_rate": 1.9993339550935017e-05, "loss": 1.9593, "step": 257 }, { "epoch": 0.04, "grad_norm": 11.832563400268555, "learning_rate": 1.9993147915005534e-05, "loss": 2.1882, "step": 258 }, { "epoch": 0.04, "grad_norm": 11.307072639465332, "learning_rate": 1.9992953562086627e-05, "loss": 2.0329, "step": 259 }, { "epoch": 0.04, "grad_norm": 10.480134963989258, "learning_rate": 1.9992756492231126e-05, "loss": 1.9967, "step": 260 }, { "epoch": 0.04, "grad_norm": 8.337092399597168, "learning_rate": 1.9992556705492622e-05, "loss": 2.0634, "step": 261 }, { "epoch": 0.04, "grad_norm": 8.860726356506348, "learning_rate": 1.9992354201925427e-05, "loss": 2.066, "step": 262 }, { "epoch": 0.04, "grad_norm": 11.490599632263184, "learning_rate": 1.9992148981584604e-05, "loss": 2.1096, "step": 263 }, { "epoch": 0.04, "grad_norm": 10.177706718444824, "learning_rate": 1.9991941044525947e-05, "loss": 2.0078, "step": 264 }, { "epoch": 0.04, "grad_norm": 10.478736877441406, "learning_rate": 1.9991730390805988e-05, "loss": 1.8575, "step": 265 }, { "epoch": 0.04, "grad_norm": 8.179291725158691, "learning_rate": 1.9991517020482005e-05, "loss": 1.9345, "step": 266 }, { "epoch": 0.04, "grad_norm": 9.46813678741455, "learning_rate": 1.999130093361201e-05, "loss": 2.1468, "step": 267 }, { "epoch": 0.04, "grad_norm": 10.848602294921875, "learning_rate": 1.999108213025475e-05, "loss": 1.8952, "step": 268 }, { "epoch": 0.04, "grad_norm": 10.179422378540039, "learning_rate": 1.999086061046972e-05, "loss": 1.9582, "step": 269 }, { "epoch": 0.04, "grad_norm": 10.828667640686035, "learning_rate": 1.9990636374317144e-05, "loss": 2.1238, "step": 270 }, { "epoch": 0.04, "grad_norm": 15.101571083068848, "learning_rate": 1.9990409421857985e-05, "loss": 1.9801, "step": 271 }, { "epoch": 0.04, "grad_norm": 10.32287311553955, "learning_rate": 1.9990179753153954e-05, "loss": 1.8861, "step": 272 }, { "epoch": 0.04, "grad_norm": 11.67353343963623, "learning_rate": 1.9989947368267498e-05, "loss": 2.0031, "step": 273 }, { "epoch": 0.04, "grad_norm": 12.079906463623047, "learning_rate": 1.998971226726179e-05, "loss": 1.9407, "step": 274 }, { "epoch": 0.04, "grad_norm": 14.955655097961426, "learning_rate": 1.9989474450200754e-05, "loss": 1.9799, "step": 275 }, { "epoch": 0.04, "grad_norm": 10.835915565490723, "learning_rate": 1.9989233917149052e-05, "loss": 2.0959, "step": 276 }, { "epoch": 0.04, "grad_norm": 12.209071159362793, "learning_rate": 1.9988990668172078e-05, "loss": 1.9142, "step": 277 }, { "epoch": 0.04, "grad_norm": 18.511388778686523, "learning_rate": 1.9988744703335964e-05, "loss": 2.0283, "step": 278 }, { "epoch": 0.04, "grad_norm": 8.210430145263672, "learning_rate": 1.9988496022707594e-05, "loss": 1.9103, "step": 279 }, { "epoch": 0.05, "grad_norm": 9.00277328491211, "learning_rate": 1.9988244626354576e-05, "loss": 2.0527, "step": 280 }, { "epoch": 0.05, "grad_norm": 9.377263069152832, "learning_rate": 1.9987990514345258e-05, "loss": 2.0873, "step": 281 }, { "epoch": 0.05, "grad_norm": 10.656028747558594, "learning_rate": 1.998773368674873e-05, "loss": 2.0448, "step": 282 }, { "epoch": 0.05, "grad_norm": 9.439637184143066, "learning_rate": 1.9987474143634823e-05, "loss": 1.9602, "step": 283 }, { "epoch": 0.05, "grad_norm": 9.287220001220703, "learning_rate": 1.9987211885074102e-05, "loss": 1.7768, "step": 284 }, { "epoch": 0.05, "grad_norm": 10.28527545928955, "learning_rate": 1.998694691113787e-05, "loss": 2.1092, "step": 285 }, { "epoch": 0.05, "grad_norm": 9.357176780700684, "learning_rate": 1.998667922189817e-05, "loss": 1.9776, "step": 286 }, { "epoch": 0.05, "grad_norm": 10.653043746948242, "learning_rate": 1.998640881742778e-05, "loss": 1.9734, "step": 287 }, { "epoch": 0.05, "grad_norm": 9.578195571899414, "learning_rate": 1.9986135697800227e-05, "loss": 2.057, "step": 288 }, { "epoch": 0.05, "grad_norm": 11.707184791564941, "learning_rate": 1.998585986308976e-05, "loss": 1.7914, "step": 289 }, { "epoch": 0.05, "grad_norm": 10.261824607849121, "learning_rate": 1.9985581313371377e-05, "loss": 2.069, "step": 290 }, { "epoch": 0.05, "grad_norm": 10.40029239654541, "learning_rate": 1.9985300048720813e-05, "loss": 2.1049, "step": 291 }, { "epoch": 0.05, "grad_norm": 7.943147659301758, "learning_rate": 1.9985016069214538e-05, "loss": 1.8589, "step": 292 }, { "epoch": 0.05, "grad_norm": 10.280961036682129, "learning_rate": 1.9984729374929764e-05, "loss": 2.2107, "step": 293 }, { "epoch": 0.05, "grad_norm": 9.960143089294434, "learning_rate": 1.9984439965944435e-05, "loss": 2.0913, "step": 294 }, { "epoch": 0.05, "grad_norm": 9.713751792907715, "learning_rate": 1.998414784233724e-05, "loss": 2.0008, "step": 295 }, { "epoch": 0.05, "grad_norm": 7.937620639801025, "learning_rate": 1.99838530041876e-05, "loss": 1.9546, "step": 296 }, { "epoch": 0.05, "grad_norm": 8.533315658569336, "learning_rate": 1.9983555451575683e-05, "loss": 2.0636, "step": 297 }, { "epoch": 0.05, "grad_norm": 10.53643798828125, "learning_rate": 1.9983255184582387e-05, "loss": 2.1762, "step": 298 }, { "epoch": 0.05, "grad_norm": 10.535935401916504, "learning_rate": 1.9982952203289344e-05, "loss": 2.097, "step": 299 }, { "epoch": 0.05, "grad_norm": 9.861007690429688, "learning_rate": 1.998264650777894e-05, "loss": 2.0742, "step": 300 }, { "epoch": 0.05, "grad_norm": 9.799873352050781, "learning_rate": 1.998233809813428e-05, "loss": 2.0527, "step": 301 }, { "epoch": 0.05, "grad_norm": 12.733719825744629, "learning_rate": 1.9982026974439222e-05, "loss": 2.0356, "step": 302 }, { "epoch": 0.05, "grad_norm": 8.998753547668457, "learning_rate": 1.9981713136778356e-05, "loss": 1.9516, "step": 303 }, { "epoch": 0.05, "grad_norm": 11.560572624206543, "learning_rate": 1.998139658523701e-05, "loss": 2.0211, "step": 304 }, { "epoch": 0.05, "grad_norm": 8.11561107635498, "learning_rate": 1.998107731990124e-05, "loss": 1.8196, "step": 305 }, { "epoch": 0.05, "grad_norm": 8.214024543762207, "learning_rate": 1.998075534085786e-05, "loss": 1.752, "step": 306 }, { "epoch": 0.05, "grad_norm": 7.97877311706543, "learning_rate": 1.998043064819441e-05, "loss": 2.0104, "step": 307 }, { "epoch": 0.05, "grad_norm": 8.350578308105469, "learning_rate": 1.9980103241999167e-05, "loss": 1.9227, "step": 308 }, { "epoch": 0.05, "grad_norm": 10.708866119384766, "learning_rate": 1.997977312236115e-05, "loss": 2.1351, "step": 309 }, { "epoch": 0.05, "grad_norm": 9.15312671661377, "learning_rate": 1.997944028937011e-05, "loss": 1.9693, "step": 310 }, { "epoch": 0.05, "grad_norm": 9.02000904083252, "learning_rate": 1.9979104743116544e-05, "loss": 2.082, "step": 311 }, { "epoch": 0.05, "grad_norm": 10.922595024108887, "learning_rate": 1.997876648369168e-05, "loss": 1.9695, "step": 312 }, { "epoch": 0.05, "grad_norm": 8.940016746520996, "learning_rate": 1.9978425511187484e-05, "loss": 1.9138, "step": 313 }, { "epoch": 0.05, "grad_norm": 9.045475006103516, "learning_rate": 1.9978081825696664e-05, "loss": 2.017, "step": 314 }, { "epoch": 0.05, "grad_norm": 10.44865894317627, "learning_rate": 1.9977735427312658e-05, "loss": 1.888, "step": 315 }, { "epoch": 0.05, "grad_norm": 9.269403457641602, "learning_rate": 1.997738631612965e-05, "loss": 1.8336, "step": 316 }, { "epoch": 0.05, "grad_norm": 9.702951431274414, "learning_rate": 1.9977034492242567e-05, "loss": 1.8682, "step": 317 }, { "epoch": 0.05, "grad_norm": 9.888885498046875, "learning_rate": 1.9976679955747048e-05, "loss": 1.8147, "step": 318 }, { "epoch": 0.05, "grad_norm": 8.33469009399414, "learning_rate": 1.99763227067395e-05, "loss": 1.9586, "step": 319 }, { "epoch": 0.05, "grad_norm": 8.822064399719238, "learning_rate": 1.9975962745317046e-05, "loss": 1.8282, "step": 320 }, { "epoch": 0.05, "grad_norm": 9.641444206237793, "learning_rate": 1.9975600071577556e-05, "loss": 1.8174, "step": 321 }, { "epoch": 0.05, "grad_norm": 9.885940551757812, "learning_rate": 1.9975234685619636e-05, "loss": 2.0624, "step": 322 }, { "epoch": 0.05, "grad_norm": 9.520670890808105, "learning_rate": 1.9974866587542633e-05, "loss": 1.9877, "step": 323 }, { "epoch": 0.05, "grad_norm": 10.13728141784668, "learning_rate": 1.997449577744662e-05, "loss": 1.9444, "step": 324 }, { "epoch": 0.05, "grad_norm": 8.748649597167969, "learning_rate": 1.9974122255432417e-05, "loss": 1.9425, "step": 325 }, { "epoch": 0.05, "grad_norm": 9.880401611328125, "learning_rate": 1.9973746021601584e-05, "loss": 1.9406, "step": 326 }, { "epoch": 0.05, "grad_norm": 9.746261596679688, "learning_rate": 1.9973367076056407e-05, "loss": 2.0288, "step": 327 }, { "epoch": 0.05, "grad_norm": 8.241989135742188, "learning_rate": 1.9972985418899923e-05, "loss": 2.0629, "step": 328 }, { "epoch": 0.05, "grad_norm": 8.545827865600586, "learning_rate": 1.9972601050235893e-05, "loss": 1.9042, "step": 329 }, { "epoch": 0.05, "grad_norm": 10.589705467224121, "learning_rate": 1.9972213970168818e-05, "loss": 1.8327, "step": 330 }, { "epoch": 0.05, "grad_norm": 8.685239791870117, "learning_rate": 1.997182417880395e-05, "loss": 1.9396, "step": 331 }, { "epoch": 0.05, "grad_norm": 9.410160064697266, "learning_rate": 1.997143167624726e-05, "loss": 1.9669, "step": 332 }, { "epoch": 0.05, "grad_norm": 10.096940040588379, "learning_rate": 1.9971036462605466e-05, "loss": 2.1067, "step": 333 }, { "epoch": 0.05, "grad_norm": 10.328001022338867, "learning_rate": 1.997063853798602e-05, "loss": 1.9271, "step": 334 }, { "epoch": 0.05, "grad_norm": 9.861626625061035, "learning_rate": 1.997023790249711e-05, "loss": 2.1299, "step": 335 }, { "epoch": 0.05, "grad_norm": 11.99864673614502, "learning_rate": 1.9969834556247664e-05, "loss": 1.8536, "step": 336 }, { "epoch": 0.05, "grad_norm": 8.87740707397461, "learning_rate": 1.996942849934735e-05, "loss": 1.8067, "step": 337 }, { "epoch": 0.05, "grad_norm": 69.6119384765625, "learning_rate": 1.9969019731906564e-05, "loss": 2.1449, "step": 338 }, { "epoch": 0.05, "grad_norm": 150.921142578125, "learning_rate": 1.9968608254036445e-05, "loss": 2.9168, "step": 339 }, { "epoch": 0.05, "grad_norm": 11.05732536315918, "learning_rate": 1.996819406584887e-05, "loss": 2.0055, "step": 340 }, { "epoch": 0.05, "grad_norm": 8.179447174072266, "learning_rate": 1.9967777167456444e-05, "loss": 2.0281, "step": 341 }, { "epoch": 0.06, "grad_norm": 13.080718994140625, "learning_rate": 1.9967357558972526e-05, "loss": 2.0484, "step": 342 }, { "epoch": 0.06, "grad_norm": 10.030044555664062, "learning_rate": 1.9966935240511196e-05, "loss": 1.92, "step": 343 }, { "epoch": 0.06, "grad_norm": 9.182323455810547, "learning_rate": 1.9966510212187273e-05, "loss": 1.7103, "step": 344 }, { "epoch": 0.06, "grad_norm": 9.08924674987793, "learning_rate": 1.996608247411632e-05, "loss": 2.0843, "step": 345 }, { "epoch": 0.06, "grad_norm": 15.328361511230469, "learning_rate": 1.9965652026414627e-05, "loss": 1.9767, "step": 346 }, { "epoch": 0.06, "grad_norm": 8.970272064208984, "learning_rate": 1.9965218869199236e-05, "loss": 1.7557, "step": 347 }, { "epoch": 0.06, "grad_norm": 12.303771018981934, "learning_rate": 1.9964783002587912e-05, "loss": 2.0003, "step": 348 }, { "epoch": 0.06, "grad_norm": 8.317943572998047, "learning_rate": 1.9964344426699154e-05, "loss": 2.0051, "step": 349 }, { "epoch": 0.06, "grad_norm": 10.0634765625, "learning_rate": 1.9963903141652215e-05, "loss": 1.8355, "step": 350 }, { "epoch": 0.06, "grad_norm": 9.863419532775879, "learning_rate": 1.9963459147567065e-05, "loss": 2.1609, "step": 351 }, { "epoch": 0.06, "grad_norm": 10.860800743103027, "learning_rate": 1.9963012444564427e-05, "loss": 1.9775, "step": 352 }, { "epoch": 0.06, "grad_norm": 11.227287292480469, "learning_rate": 1.9962563032765745e-05, "loss": 1.8197, "step": 353 }, { "epoch": 0.06, "grad_norm": 8.828988075256348, "learning_rate": 1.9962110912293214e-05, "loss": 2.2156, "step": 354 }, { "epoch": 0.06, "grad_norm": 10.03002643585205, "learning_rate": 1.9961656083269754e-05, "loss": 1.9437, "step": 355 }, { "epoch": 0.06, "grad_norm": 9.11108112335205, "learning_rate": 1.9961198545819035e-05, "loss": 2.0819, "step": 356 }, { "epoch": 0.06, "grad_norm": 9.181781768798828, "learning_rate": 1.9960738300065444e-05, "loss": 2.0, "step": 357 }, { "epoch": 0.06, "grad_norm": 8.62926959991455, "learning_rate": 1.996027534613412e-05, "loss": 1.9257, "step": 358 }, { "epoch": 0.06, "grad_norm": 11.265356063842773, "learning_rate": 1.9959809684150934e-05, "loss": 1.9025, "step": 359 }, { "epoch": 0.06, "grad_norm": 10.882258415222168, "learning_rate": 1.9959341314242487e-05, "loss": 1.8455, "step": 360 }, { "epoch": 0.06, "grad_norm": 10.78380012512207, "learning_rate": 1.9958870236536133e-05, "loss": 2.0018, "step": 361 }, { "epoch": 0.06, "grad_norm": 12.970155715942383, "learning_rate": 1.9958396451159937e-05, "loss": 1.908, "step": 362 }, { "epoch": 0.06, "grad_norm": 11.603653907775879, "learning_rate": 1.9957919958242728e-05, "loss": 1.6851, "step": 363 }, { "epoch": 0.06, "grad_norm": 9.196601867675781, "learning_rate": 1.995744075791405e-05, "loss": 2.0649, "step": 364 }, { "epoch": 0.06, "grad_norm": 10.706315040588379, "learning_rate": 1.995695885030419e-05, "loss": 1.9625, "step": 365 }, { "epoch": 0.06, "grad_norm": 17.09244728088379, "learning_rate": 1.9956474235544172e-05, "loss": 1.8491, "step": 366 }, { "epoch": 0.06, "grad_norm": 9.862274169921875, "learning_rate": 1.995598691376576e-05, "loss": 1.9881, "step": 367 }, { "epoch": 0.06, "grad_norm": 8.7662353515625, "learning_rate": 1.9955496885101442e-05, "loss": 2.0058, "step": 368 }, { "epoch": 0.06, "grad_norm": 18.489601135253906, "learning_rate": 1.995500414968446e-05, "loss": 2.1862, "step": 369 }, { "epoch": 0.06, "grad_norm": 8.650236129760742, "learning_rate": 1.995450870764877e-05, "loss": 1.9185, "step": 370 }, { "epoch": 0.06, "grad_norm": 11.590608596801758, "learning_rate": 1.995401055912908e-05, "loss": 2.0079, "step": 371 }, { "epoch": 0.06, "grad_norm": 11.862245559692383, "learning_rate": 1.9953509704260834e-05, "loss": 2.0289, "step": 372 }, { "epoch": 0.06, "grad_norm": 25.758907318115234, "learning_rate": 1.99530061431802e-05, "loss": 1.898, "step": 373 }, { "epoch": 0.06, "grad_norm": 10.375194549560547, "learning_rate": 1.9952499876024093e-05, "loss": 1.8644, "step": 374 }, { "epoch": 0.06, "grad_norm": 10.469151496887207, "learning_rate": 1.995199090293016e-05, "loss": 1.9798, "step": 375 }, { "epoch": 0.06, "grad_norm": 9.36967658996582, "learning_rate": 1.9951479224036782e-05, "loss": 1.9413, "step": 376 }, { "epoch": 0.06, "grad_norm": 11.178262710571289, "learning_rate": 1.9950964839483072e-05, "loss": 1.8158, "step": 377 }, { "epoch": 0.06, "grad_norm": 12.748632431030273, "learning_rate": 1.9950447749408893e-05, "loss": 1.9222, "step": 378 }, { "epoch": 0.06, "grad_norm": 12.480185508728027, "learning_rate": 1.994992795395483e-05, "loss": 1.9466, "step": 379 }, { "epoch": 0.06, "grad_norm": 14.964760780334473, "learning_rate": 1.99494054532622e-05, "loss": 2.0263, "step": 380 }, { "epoch": 0.06, "grad_norm": 9.589376449584961, "learning_rate": 1.9948880247473076e-05, "loss": 1.9666, "step": 381 }, { "epoch": 0.06, "grad_norm": 14.44439697265625, "learning_rate": 1.994835233673025e-05, "loss": 1.8451, "step": 382 }, { "epoch": 0.06, "grad_norm": 23.258020401000977, "learning_rate": 1.9947821721177244e-05, "loss": 2.2127, "step": 383 }, { "epoch": 0.06, "grad_norm": 9.570511817932129, "learning_rate": 1.9947288400958338e-05, "loss": 1.9247, "step": 384 }, { "epoch": 0.06, "grad_norm": 13.872358322143555, "learning_rate": 1.9946752376218525e-05, "loss": 1.9342, "step": 385 }, { "epoch": 0.06, "grad_norm": 13.975397109985352, "learning_rate": 1.9946213647103544e-05, "loss": 1.9779, "step": 386 }, { "epoch": 0.06, "grad_norm": 14.989208221435547, "learning_rate": 1.9945672213759872e-05, "loss": 2.0076, "step": 387 }, { "epoch": 0.06, "grad_norm": 8.976994514465332, "learning_rate": 1.9945128076334707e-05, "loss": 1.8301, "step": 388 }, { "epoch": 0.06, "grad_norm": 12.683465957641602, "learning_rate": 1.9944581234976003e-05, "loss": 2.05, "step": 389 }, { "epoch": 0.06, "grad_norm": 19.585596084594727, "learning_rate": 1.994403168983243e-05, "loss": 1.8825, "step": 390 }, { "epoch": 0.06, "grad_norm": 10.950916290283203, "learning_rate": 1.9943479441053405e-05, "loss": 1.8083, "step": 391 }, { "epoch": 0.06, "grad_norm": 9.473930358886719, "learning_rate": 1.9942924488789077e-05, "loss": 1.8135, "step": 392 }, { "epoch": 0.06, "grad_norm": 17.25933074951172, "learning_rate": 1.9942366833190326e-05, "loss": 2.1475, "step": 393 }, { "epoch": 0.06, "grad_norm": 13.47601318359375, "learning_rate": 1.9941806474408774e-05, "loss": 1.8962, "step": 394 }, { "epoch": 0.06, "grad_norm": 10.4471435546875, "learning_rate": 1.994124341259677e-05, "loss": 2.057, "step": 395 }, { "epoch": 0.06, "grad_norm": 10.009971618652344, "learning_rate": 1.9940677647907408e-05, "loss": 1.998, "step": 396 }, { "epoch": 0.06, "grad_norm": 21.681259155273438, "learning_rate": 1.9940109180494505e-05, "loss": 2.006, "step": 397 }, { "epoch": 0.06, "grad_norm": 9.105794906616211, "learning_rate": 1.9939538010512623e-05, "loss": 1.9974, "step": 398 }, { "epoch": 0.06, "grad_norm": 8.01227855682373, "learning_rate": 1.9938964138117055e-05, "loss": 1.8157, "step": 399 }, { "epoch": 0.06, "grad_norm": 9.23107624053955, "learning_rate": 1.9938387563463823e-05, "loss": 2.0233, "step": 400 }, { "epoch": 0.06, "grad_norm": 13.386942863464355, "learning_rate": 1.9937808286709695e-05, "loss": 2.0162, "step": 401 }, { "epoch": 0.06, "grad_norm": 7.762180328369141, "learning_rate": 1.9937226308012165e-05, "loss": 1.9207, "step": 402 }, { "epoch": 0.06, "grad_norm": 12.036272048950195, "learning_rate": 1.9936641627529465e-05, "loss": 1.9387, "step": 403 }, { "epoch": 0.07, "grad_norm": 9.953751564025879, "learning_rate": 1.993605424542056e-05, "loss": 1.8492, "step": 404 }, { "epoch": 0.07, "grad_norm": 23.044994354248047, "learning_rate": 1.9935464161845154e-05, "loss": 1.989, "step": 405 }, { "epoch": 0.07, "grad_norm": 9.80964183807373, "learning_rate": 1.993487137696368e-05, "loss": 2.0347, "step": 406 }, { "epoch": 0.07, "grad_norm": 12.442686080932617, "learning_rate": 1.9934275890937306e-05, "loss": 1.976, "step": 407 }, { "epoch": 0.07, "grad_norm": 12.540276527404785, "learning_rate": 1.993367770392794e-05, "loss": 2.018, "step": 408 }, { "epoch": 0.07, "grad_norm": 12.692622184753418, "learning_rate": 1.993307681609821e-05, "loss": 1.9199, "step": 409 }, { "epoch": 0.07, "grad_norm": 11.172554016113281, "learning_rate": 1.99324732276115e-05, "loss": 2.2011, "step": 410 }, { "epoch": 0.07, "grad_norm": 11.112929344177246, "learning_rate": 1.9931866938631915e-05, "loss": 1.8975, "step": 411 }, { "epoch": 0.07, "grad_norm": 16.3880615234375, "learning_rate": 1.993125794932429e-05, "loss": 2.0766, "step": 412 }, { "epoch": 0.07, "grad_norm": 10.42484188079834, "learning_rate": 1.9930646259854205e-05, "loss": 2.0528, "step": 413 }, { "epoch": 0.07, "grad_norm": 7.840005397796631, "learning_rate": 1.993003187038797e-05, "loss": 1.8675, "step": 414 }, { "epoch": 0.07, "grad_norm": 10.859256744384766, "learning_rate": 1.9929414781092622e-05, "loss": 2.0514, "step": 415 }, { "epoch": 0.07, "grad_norm": 8.951383590698242, "learning_rate": 1.9928794992135944e-05, "loss": 2.0608, "step": 416 }, { "epoch": 0.07, "grad_norm": 7.936229705810547, "learning_rate": 1.992817250368645e-05, "loss": 1.8386, "step": 417 }, { "epoch": 0.07, "grad_norm": 8.919425964355469, "learning_rate": 1.992754731591338e-05, "loss": 1.8916, "step": 418 }, { "epoch": 0.07, "grad_norm": 7.550927639007568, "learning_rate": 1.9926919428986717e-05, "loss": 1.8659, "step": 419 }, { "epoch": 0.07, "grad_norm": 8.721219062805176, "learning_rate": 1.9926288843077168e-05, "loss": 1.982, "step": 420 }, { "epoch": 0.07, "grad_norm": 9.753211975097656, "learning_rate": 1.992565555835619e-05, "loss": 2.0045, "step": 421 }, { "epoch": 0.07, "grad_norm": 9.299637794494629, "learning_rate": 1.992501957499595e-05, "loss": 1.8046, "step": 422 }, { "epoch": 0.07, "grad_norm": 9.123473167419434, "learning_rate": 1.992438089316938e-05, "loss": 1.9152, "step": 423 }, { "epoch": 0.07, "grad_norm": 8.633929252624512, "learning_rate": 1.9923739513050113e-05, "loss": 1.9441, "step": 424 }, { "epoch": 0.07, "grad_norm": 10.620311737060547, "learning_rate": 1.992309543481254e-05, "loss": 2.0983, "step": 425 }, { "epoch": 0.07, "grad_norm": 12.553050994873047, "learning_rate": 1.992244865863177e-05, "loss": 1.9374, "step": 426 }, { "epoch": 0.07, "grad_norm": 8.529656410217285, "learning_rate": 1.9921799184683656e-05, "loss": 2.0625, "step": 427 }, { "epoch": 0.07, "grad_norm": 8.471674919128418, "learning_rate": 1.9921147013144782e-05, "loss": 1.9164, "step": 428 }, { "epoch": 0.07, "grad_norm": 7.987317085266113, "learning_rate": 1.9920492144192456e-05, "loss": 2.0753, "step": 429 }, { "epoch": 0.07, "grad_norm": 11.498106956481934, "learning_rate": 1.9919834578004734e-05, "loss": 1.9596, "step": 430 }, { "epoch": 0.07, "grad_norm": 7.861117362976074, "learning_rate": 1.99191743147604e-05, "loss": 1.8934, "step": 431 }, { "epoch": 0.07, "grad_norm": 9.270795822143555, "learning_rate": 1.9918511354638962e-05, "loss": 2.0483, "step": 432 }, { "epoch": 0.07, "grad_norm": 8.849078178405762, "learning_rate": 1.9917845697820678e-05, "loss": 2.1058, "step": 433 }, { "epoch": 0.07, "grad_norm": 8.990449905395508, "learning_rate": 1.991717734448652e-05, "loss": 1.9204, "step": 434 }, { "epoch": 0.07, "grad_norm": 7.873619556427002, "learning_rate": 1.9916506294818218e-05, "loss": 2.0607, "step": 435 }, { "epoch": 0.07, "grad_norm": 7.681189060211182, "learning_rate": 1.9915832548998205e-05, "loss": 1.8211, "step": 436 }, { "epoch": 0.07, "grad_norm": 8.70447063446045, "learning_rate": 1.9915156107209673e-05, "loss": 1.9698, "step": 437 }, { "epoch": 0.07, "grad_norm": 9.657299995422363, "learning_rate": 1.9914476969636533e-05, "loss": 1.8728, "step": 438 }, { "epoch": 0.07, "grad_norm": 7.9970598220825195, "learning_rate": 1.991379513646343e-05, "loss": 2.011, "step": 439 }, { "epoch": 0.07, "grad_norm": 8.777530670166016, "learning_rate": 1.9913110607875747e-05, "loss": 1.9877, "step": 440 }, { "epoch": 0.07, "grad_norm": 10.469786643981934, "learning_rate": 1.99124233840596e-05, "loss": 2.0943, "step": 441 }, { "epoch": 0.07, "grad_norm": 9.84034252166748, "learning_rate": 1.991173346520183e-05, "loss": 1.9518, "step": 442 }, { "epoch": 0.07, "grad_norm": 8.243837356567383, "learning_rate": 1.9911040851490017e-05, "loss": 2.1217, "step": 443 }, { "epoch": 0.07, "grad_norm": 14.462882995605469, "learning_rate": 1.991034554311248e-05, "loss": 1.9676, "step": 444 }, { "epoch": 0.07, "grad_norm": 8.050661087036133, "learning_rate": 1.9909647540258246e-05, "loss": 1.8984, "step": 445 }, { "epoch": 0.07, "grad_norm": 13.141295433044434, "learning_rate": 1.9908946843117108e-05, "loss": 2.12, "step": 446 }, { "epoch": 0.07, "grad_norm": 9.762130737304688, "learning_rate": 1.9908243451879566e-05, "loss": 1.873, "step": 447 }, { "epoch": 0.07, "grad_norm": 8.004844665527344, "learning_rate": 1.990753736673687e-05, "loss": 1.8351, "step": 448 }, { "epoch": 0.07, "grad_norm": 9.111143112182617, "learning_rate": 1.9906828587880985e-05, "loss": 1.8587, "step": 449 }, { "epoch": 0.07, "grad_norm": 8.517598152160645, "learning_rate": 1.990611711550462e-05, "loss": 1.9611, "step": 450 }, { "epoch": 0.07, "grad_norm": 9.20211410522461, "learning_rate": 1.990540294980122e-05, "loss": 1.9797, "step": 451 }, { "epoch": 0.07, "grad_norm": 9.830244064331055, "learning_rate": 1.990468609096495e-05, "loss": 1.7628, "step": 452 }, { "epoch": 0.07, "grad_norm": 8.46324348449707, "learning_rate": 1.9903966539190716e-05, "loss": 2.0453, "step": 453 }, { "epoch": 0.07, "grad_norm": 9.128844261169434, "learning_rate": 1.990324429467415e-05, "loss": 1.8744, "step": 454 }, { "epoch": 0.07, "grad_norm": 8.958897590637207, "learning_rate": 1.990251935761162e-05, "loss": 2.0463, "step": 455 }, { "epoch": 0.07, "grad_norm": 9.714156150817871, "learning_rate": 1.9901791728200234e-05, "loss": 1.7815, "step": 456 }, { "epoch": 0.07, "grad_norm": 8.719221115112305, "learning_rate": 1.9901061406637816e-05, "loss": 1.9675, "step": 457 }, { "epoch": 0.07, "grad_norm": 9.446579933166504, "learning_rate": 1.990032839312293e-05, "loss": 1.8681, "step": 458 }, { "epoch": 0.07, "grad_norm": 9.240344047546387, "learning_rate": 1.989959268785487e-05, "loss": 1.9021, "step": 459 }, { "epoch": 0.07, "grad_norm": 8.705483436584473, "learning_rate": 1.9898854291033668e-05, "loss": 1.8183, "step": 460 }, { "epoch": 0.07, "grad_norm": 7.580947399139404, "learning_rate": 1.9898113202860082e-05, "loss": 1.7868, "step": 461 }, { "epoch": 0.07, "grad_norm": 9.489875793457031, "learning_rate": 1.98973694235356e-05, "loss": 1.823, "step": 462 }, { "epoch": 0.07, "grad_norm": 7.888040542602539, "learning_rate": 1.9896622953262448e-05, "loss": 1.9402, "step": 463 }, { "epoch": 0.07, "grad_norm": 9.30180549621582, "learning_rate": 1.989587379224358e-05, "loss": 1.9806, "step": 464 }, { "epoch": 0.07, "grad_norm": 7.79681396484375, "learning_rate": 1.9895121940682678e-05, "loss": 1.6853, "step": 465 }, { "epoch": 0.08, "grad_norm": 7.780069351196289, "learning_rate": 1.9894367398784165e-05, "loss": 1.8024, "step": 466 }, { "epoch": 0.08, "grad_norm": 7.469237804412842, "learning_rate": 1.9893610166753187e-05, "loss": 1.8598, "step": 467 }, { "epoch": 0.08, "grad_norm": 8.480820655822754, "learning_rate": 1.9892850244795622e-05, "loss": 2.0746, "step": 468 }, { "epoch": 0.08, "grad_norm": 14.931609153747559, "learning_rate": 1.9892087633118087e-05, "loss": 1.9474, "step": 469 }, { "epoch": 0.08, "grad_norm": 10.264354705810547, "learning_rate": 1.9891322331927916e-05, "loss": 1.9338, "step": 470 }, { "epoch": 0.08, "grad_norm": 14.30178451538086, "learning_rate": 1.9890554341433194e-05, "loss": 2.0271, "step": 471 }, { "epoch": 0.08, "grad_norm": 14.327902793884277, "learning_rate": 1.988978366184272e-05, "loss": 1.8952, "step": 472 }, { "epoch": 0.08, "grad_norm": 9.082472801208496, "learning_rate": 1.988901029336603e-05, "loss": 1.7627, "step": 473 }, { "epoch": 0.08, "grad_norm": 12.554320335388184, "learning_rate": 1.9888234236213396e-05, "loss": 1.9399, "step": 474 }, { "epoch": 0.08, "grad_norm": 12.551857948303223, "learning_rate": 1.9887455490595807e-05, "loss": 2.1078, "step": 475 }, { "epoch": 0.08, "grad_norm": 7.849806308746338, "learning_rate": 1.9886674056725006e-05, "loss": 1.8513, "step": 476 }, { "epoch": 0.08, "grad_norm": 8.90108585357666, "learning_rate": 1.9885889934813444e-05, "loss": 1.9735, "step": 477 }, { "epoch": 0.08, "grad_norm": 12.745051383972168, "learning_rate": 1.9885103125074313e-05, "loss": 2.0311, "step": 478 }, { "epoch": 0.08, "grad_norm": 9.700244903564453, "learning_rate": 1.9884313627721536e-05, "loss": 1.8754, "step": 479 }, { "epoch": 0.08, "grad_norm": 9.543883323669434, "learning_rate": 1.9883521442969765e-05, "loss": 1.9064, "step": 480 }, { "epoch": 0.08, "grad_norm": 9.099937438964844, "learning_rate": 1.9882726571034386e-05, "loss": 2.0598, "step": 481 }, { "epoch": 0.08, "grad_norm": 9.018400192260742, "learning_rate": 1.988192901213151e-05, "loss": 1.8761, "step": 482 }, { "epoch": 0.08, "grad_norm": 9.36060619354248, "learning_rate": 1.9881128766477985e-05, "loss": 1.9701, "step": 483 }, { "epoch": 0.08, "grad_norm": 6.962739944458008, "learning_rate": 1.9880325834291385e-05, "loss": 1.6236, "step": 484 }, { "epoch": 0.08, "grad_norm": 7.664331912994385, "learning_rate": 1.987952021579001e-05, "loss": 1.9048, "step": 485 }, { "epoch": 0.08, "grad_norm": 8.826683044433594, "learning_rate": 1.98787119111929e-05, "loss": 1.7464, "step": 486 }, { "epoch": 0.08, "grad_norm": 8.72780704498291, "learning_rate": 1.9877900920719825e-05, "loss": 1.8408, "step": 487 }, { "epoch": 0.08, "grad_norm": 8.122699737548828, "learning_rate": 1.9877087244591274e-05, "loss": 1.9556, "step": 488 }, { "epoch": 0.08, "grad_norm": 8.41059398651123, "learning_rate": 1.987627088302848e-05, "loss": 1.9461, "step": 489 }, { "epoch": 0.08, "grad_norm": 8.604735374450684, "learning_rate": 1.9875451836253395e-05, "loss": 1.862, "step": 490 }, { "epoch": 0.08, "grad_norm": 7.891695976257324, "learning_rate": 1.9874630104488707e-05, "loss": 1.935, "step": 491 }, { "epoch": 0.08, "grad_norm": 11.39697265625, "learning_rate": 1.987380568795783e-05, "loss": 1.8836, "step": 492 }, { "epoch": 0.08, "grad_norm": 9.474388122558594, "learning_rate": 1.987297858688492e-05, "loss": 1.9662, "step": 493 }, { "epoch": 0.08, "grad_norm": 9.542935371398926, "learning_rate": 1.987214880149484e-05, "loss": 1.8987, "step": 494 }, { "epoch": 0.08, "grad_norm": 8.565422058105469, "learning_rate": 1.987131633201321e-05, "loss": 2.0584, "step": 495 }, { "epoch": 0.08, "grad_norm": 8.152287483215332, "learning_rate": 1.9870481178666357e-05, "loss": 1.7567, "step": 496 }, { "epoch": 0.08, "grad_norm": 8.868548393249512, "learning_rate": 1.986964334168135e-05, "loss": 2.0581, "step": 497 }, { "epoch": 0.08, "grad_norm": 8.114066123962402, "learning_rate": 1.9868802821285982e-05, "loss": 1.814, "step": 498 }, { "epoch": 0.08, "grad_norm": 10.002664566040039, "learning_rate": 1.9867959617708787e-05, "loss": 1.9592, "step": 499 }, { "epoch": 0.08, "grad_norm": 7.425230503082275, "learning_rate": 1.9867113731179003e-05, "loss": 1.8367, "step": 500 }, { "epoch": 0.08, "grad_norm": 9.044759750366211, "learning_rate": 1.9866265161926632e-05, "loss": 2.0676, "step": 501 }, { "epoch": 0.08, "grad_norm": 7.525268077850342, "learning_rate": 1.9865413910182377e-05, "loss": 1.9612, "step": 502 }, { "epoch": 0.08, "grad_norm": 10.268218040466309, "learning_rate": 1.9864559976177683e-05, "loss": 2.178, "step": 503 }, { "epoch": 0.08, "grad_norm": 8.143590927124023, "learning_rate": 1.986370336014472e-05, "loss": 1.8027, "step": 504 }, { "epoch": 0.08, "grad_norm": 14.03950023651123, "learning_rate": 1.9862844062316396e-05, "loss": 2.0165, "step": 505 }, { "epoch": 0.08, "grad_norm": 10.413082122802734, "learning_rate": 1.9861982082926333e-05, "loss": 2.1622, "step": 506 }, { "epoch": 0.08, "grad_norm": 12.920186996459961, "learning_rate": 1.9861117422208897e-05, "loss": 1.9713, "step": 507 }, { "epoch": 0.08, "grad_norm": 9.000523567199707, "learning_rate": 1.9860250080399175e-05, "loss": 1.916, "step": 508 }, { "epoch": 0.08, "grad_norm": 9.76311206817627, "learning_rate": 1.985938005773298e-05, "loss": 1.9795, "step": 509 }, { "epoch": 0.08, "grad_norm": 10.876928329467773, "learning_rate": 1.9858507354446865e-05, "loss": 2.0234, "step": 510 }, { "epoch": 0.08, "grad_norm": 14.350929260253906, "learning_rate": 1.9857631970778097e-05, "loss": 1.9193, "step": 511 }, { "epoch": 0.08, "grad_norm": 9.342623710632324, "learning_rate": 1.9856753906964686e-05, "loss": 1.9507, "step": 512 }, { "epoch": 0.08, "grad_norm": 12.097273826599121, "learning_rate": 1.9855873163245367e-05, "loss": 1.8892, "step": 513 }, { "epoch": 0.08, "grad_norm": 8.723252296447754, "learning_rate": 1.9854989739859595e-05, "loss": 2.0184, "step": 514 }, { "epoch": 0.08, "grad_norm": 8.700166702270508, "learning_rate": 1.9854103637047562e-05, "loss": 1.9448, "step": 515 }, { "epoch": 0.08, "grad_norm": 8.794655799865723, "learning_rate": 1.985321485505019e-05, "loss": 1.8267, "step": 516 }, { "epoch": 0.08, "grad_norm": 9.388087272644043, "learning_rate": 1.985232339410912e-05, "loss": 2.0954, "step": 517 }, { "epoch": 0.08, "grad_norm": 8.357881546020508, "learning_rate": 1.9851429254466726e-05, "loss": 1.6799, "step": 518 }, { "epoch": 0.08, "grad_norm": 7.894590377807617, "learning_rate": 1.985053243636612e-05, "loss": 1.9584, "step": 519 }, { "epoch": 0.08, "grad_norm": 8.662221908569336, "learning_rate": 1.9849632940051125e-05, "loss": 1.9727, "step": 520 }, { "epoch": 0.08, "grad_norm": 7.5137739181518555, "learning_rate": 1.9848730765766303e-05, "loss": 2.0411, "step": 521 }, { "epoch": 0.08, "grad_norm": 11.324359893798828, "learning_rate": 1.9847825913756946e-05, "loss": 2.0682, "step": 522 }, { "epoch": 0.08, "grad_norm": 7.790539741516113, "learning_rate": 1.9846918384269064e-05, "loss": 2.0098, "step": 523 }, { "epoch": 0.08, "grad_norm": 8.409256935119629, "learning_rate": 1.9846008177549406e-05, "loss": 2.0711, "step": 524 }, { "epoch": 0.08, "grad_norm": 10.319316864013672, "learning_rate": 1.9845095293845444e-05, "loss": 1.8838, "step": 525 }, { "epoch": 0.08, "grad_norm": 9.316356658935547, "learning_rate": 1.984417973340537e-05, "loss": 2.0394, "step": 526 }, { "epoch": 0.08, "grad_norm": 8.963420867919922, "learning_rate": 1.984326149647812e-05, "loss": 2.1003, "step": 527 }, { "epoch": 0.08, "grad_norm": 9.208054542541504, "learning_rate": 1.9842340583313343e-05, "loss": 2.1128, "step": 528 }, { "epoch": 0.09, "grad_norm": 10.655410766601562, "learning_rate": 1.9841416994161425e-05, "loss": 2.0834, "step": 529 }, { "epoch": 0.09, "grad_norm": 8.812787055969238, "learning_rate": 1.9840490729273477e-05, "loss": 2.1495, "step": 530 }, { "epoch": 0.09, "grad_norm": 8.516873359680176, "learning_rate": 1.9839561788901332e-05, "loss": 1.8857, "step": 531 }, { "epoch": 0.09, "grad_norm": 8.341115951538086, "learning_rate": 1.9838630173297563e-05, "loss": 2.0509, "step": 532 }, { "epoch": 0.09, "grad_norm": 8.538514137268066, "learning_rate": 1.9837695882715455e-05, "loss": 2.086, "step": 533 }, { "epoch": 0.09, "grad_norm": 9.637955665588379, "learning_rate": 1.983675891740903e-05, "loss": 1.8613, "step": 534 }, { "epoch": 0.09, "grad_norm": 8.290384292602539, "learning_rate": 1.983581927763304e-05, "loss": 1.9445, "step": 535 }, { "epoch": 0.09, "grad_norm": 9.085714340209961, "learning_rate": 1.983487696364295e-05, "loss": 1.8701, "step": 536 }, { "epoch": 0.09, "grad_norm": 10.747737884521484, "learning_rate": 1.983393197569497e-05, "loss": 2.0433, "step": 537 }, { "epoch": 0.09, "grad_norm": 8.779898643493652, "learning_rate": 1.9832984314046025e-05, "loss": 1.7439, "step": 538 }, { "epoch": 0.09, "grad_norm": 10.047880172729492, "learning_rate": 1.9832033978953768e-05, "loss": 2.006, "step": 539 }, { "epoch": 0.09, "grad_norm": 10.01353931427002, "learning_rate": 1.9831080970676582e-05, "loss": 1.7927, "step": 540 }, { "epoch": 0.09, "grad_norm": 9.968310356140137, "learning_rate": 1.9830125289473578e-05, "loss": 1.924, "step": 541 }, { "epoch": 0.09, "grad_norm": 9.646020889282227, "learning_rate": 1.9829166935604587e-05, "loss": 2.0606, "step": 542 }, { "epoch": 0.09, "grad_norm": 14.281914710998535, "learning_rate": 1.982820590933018e-05, "loss": 1.9633, "step": 543 }, { "epoch": 0.09, "grad_norm": 9.089286804199219, "learning_rate": 1.9827242210911638e-05, "loss": 1.8565, "step": 544 }, { "epoch": 0.09, "grad_norm": 9.427074432373047, "learning_rate": 1.982627584061098e-05, "loss": 1.8261, "step": 545 }, { "epoch": 0.09, "grad_norm": 9.892484664916992, "learning_rate": 1.9825306798690945e-05, "loss": 1.9347, "step": 546 }, { "epoch": 0.09, "grad_norm": 12.309922218322754, "learning_rate": 1.9824335085415e-05, "loss": 2.0097, "step": 547 }, { "epoch": 0.09, "grad_norm": 8.441412925720215, "learning_rate": 1.982336070104735e-05, "loss": 2.017, "step": 548 }, { "epoch": 0.09, "grad_norm": 7.73372745513916, "learning_rate": 1.9822383645852904e-05, "loss": 2.0008, "step": 549 }, { "epoch": 0.09, "grad_norm": 8.566071510314941, "learning_rate": 1.9821403920097313e-05, "loss": 1.9773, "step": 550 }, { "epoch": 0.09, "grad_norm": 13.697305679321289, "learning_rate": 1.982042152404695e-05, "loss": 1.9484, "step": 551 }, { "epoch": 0.09, "grad_norm": 7.132179260253906, "learning_rate": 1.9819436457968912e-05, "loss": 1.9519, "step": 552 }, { "epoch": 0.09, "grad_norm": 11.187341690063477, "learning_rate": 1.981844872213103e-05, "loss": 1.7251, "step": 553 }, { "epoch": 0.09, "grad_norm": 13.244024276733398, "learning_rate": 1.981745831680185e-05, "loss": 1.777, "step": 554 }, { "epoch": 0.09, "grad_norm": 9.554508209228516, "learning_rate": 1.9816465242250644e-05, "loss": 1.6826, "step": 555 }, { "epoch": 0.09, "grad_norm": 8.58526611328125, "learning_rate": 1.9815469498747425e-05, "loss": 1.8621, "step": 556 }, { "epoch": 0.09, "grad_norm": 8.072663307189941, "learning_rate": 1.9814471086562914e-05, "loss": 1.7117, "step": 557 }, { "epoch": 0.09, "grad_norm": 13.939444541931152, "learning_rate": 1.9813470005968564e-05, "loss": 1.8088, "step": 558 }, { "epoch": 0.09, "grad_norm": 10.591938972473145, "learning_rate": 1.981246625723656e-05, "loss": 2.029, "step": 559 }, { "epoch": 0.09, "grad_norm": 9.228619575500488, "learning_rate": 1.98114598406398e-05, "loss": 1.9791, "step": 560 }, { "epoch": 0.09, "grad_norm": 8.268694877624512, "learning_rate": 1.981045075645192e-05, "loss": 1.911, "step": 561 }, { "epoch": 0.09, "grad_norm": 9.049271583557129, "learning_rate": 1.980943900494727e-05, "loss": 1.8543, "step": 562 }, { "epoch": 0.09, "grad_norm": 10.180779457092285, "learning_rate": 1.9808424586400927e-05, "loss": 1.8508, "step": 563 }, { "epoch": 0.09, "grad_norm": 8.681690216064453, "learning_rate": 1.980740750108871e-05, "loss": 1.9953, "step": 564 }, { "epoch": 0.09, "grad_norm": 6.7981276512146, "learning_rate": 1.9806387749287132e-05, "loss": 1.9576, "step": 565 }, { "epoch": 0.09, "grad_norm": 9.764735221862793, "learning_rate": 1.980536533127346e-05, "loss": 2.0074, "step": 566 }, { "epoch": 0.09, "grad_norm": 8.091477394104004, "learning_rate": 1.9804340247325674e-05, "loss": 1.826, "step": 567 }, { "epoch": 0.09, "grad_norm": 8.341599464416504, "learning_rate": 1.9803312497722476e-05, "loss": 1.9058, "step": 568 }, { "epoch": 0.09, "grad_norm": 9.815007209777832, "learning_rate": 1.9802282082743294e-05, "loss": 1.9273, "step": 569 }, { "epoch": 0.09, "grad_norm": 10.88758659362793, "learning_rate": 1.980124900266829e-05, "loss": 1.8598, "step": 570 }, { "epoch": 0.09, "grad_norm": 8.688088417053223, "learning_rate": 1.9800213257778333e-05, "loss": 2.0036, "step": 571 }, { "epoch": 0.09, "grad_norm": 10.988228797912598, "learning_rate": 1.9799174848355038e-05, "loss": 1.7277, "step": 572 }, { "epoch": 0.09, "grad_norm": 9.267723083496094, "learning_rate": 1.979813377468072e-05, "loss": 1.8695, "step": 573 }, { "epoch": 0.09, "grad_norm": 8.477424621582031, "learning_rate": 1.9797090037038447e-05, "loss": 1.82, "step": 574 }, { "epoch": 0.09, "grad_norm": 10.81506061553955, "learning_rate": 1.9796043635711982e-05, "loss": 1.86, "step": 575 }, { "epoch": 0.09, "grad_norm": 9.127056121826172, "learning_rate": 1.9794994570985838e-05, "loss": 1.9109, "step": 576 }, { "epoch": 0.09, "grad_norm": 7.41264533996582, "learning_rate": 1.9793942843145227e-05, "loss": 1.9275, "step": 577 }, { "epoch": 0.09, "grad_norm": 8.991133689880371, "learning_rate": 1.9792888452476112e-05, "loss": 1.9054, "step": 578 }, { "epoch": 0.09, "grad_norm": 9.084615707397461, "learning_rate": 1.979183139926516e-05, "loss": 1.9532, "step": 579 }, { "epoch": 0.09, "grad_norm": 7.6576313972473145, "learning_rate": 1.9790771683799762e-05, "loss": 1.8677, "step": 580 }, { "epoch": 0.09, "grad_norm": 8.97928524017334, "learning_rate": 1.978970930636805e-05, "loss": 1.8612, "step": 581 }, { "epoch": 0.09, "grad_norm": 7.855606555938721, "learning_rate": 1.978864426725886e-05, "loss": 1.8871, "step": 582 }, { "epoch": 0.09, "grad_norm": 9.002568244934082, "learning_rate": 1.9787576566761767e-05, "loss": 1.9758, "step": 583 }, { "epoch": 0.09, "grad_norm": 12.047402381896973, "learning_rate": 1.9786506205167058e-05, "loss": 2.0807, "step": 584 }, { "epoch": 0.09, "grad_norm": 8.68134593963623, "learning_rate": 1.978543318276575e-05, "loss": 1.7091, "step": 585 }, { "epoch": 0.09, "grad_norm": 8.20746898651123, "learning_rate": 1.9784357499849584e-05, "loss": 1.8503, "step": 586 }, { "epoch": 0.09, "grad_norm": 17.29816246032715, "learning_rate": 1.9783279156711022e-05, "loss": 2.212, "step": 587 }, { "epoch": 0.09, "grad_norm": 9.21575927734375, "learning_rate": 1.978219815364325e-05, "loss": 1.9211, "step": 588 }, { "epoch": 0.09, "grad_norm": 7.853124618530273, "learning_rate": 1.978111449094017e-05, "loss": 1.8314, "step": 589 }, { "epoch": 0.09, "grad_norm": 10.698988914489746, "learning_rate": 1.9780028168896425e-05, "loss": 1.9503, "step": 590 }, { "epoch": 0.1, "grad_norm": 16.079877853393555, "learning_rate": 1.977893918780736e-05, "loss": 1.883, "step": 591 }, { "epoch": 0.1, "grad_norm": 9.01424789428711, "learning_rate": 1.9777847547969057e-05, "loss": 2.031, "step": 592 }, { "epoch": 0.1, "grad_norm": 9.561965942382812, "learning_rate": 1.9776753249678318e-05, "loss": 1.7809, "step": 593 }, { "epoch": 0.1, "grad_norm": 13.892199516296387, "learning_rate": 1.9775656293232663e-05, "loss": 1.8577, "step": 594 }, { "epoch": 0.1, "grad_norm": 11.893144607543945, "learning_rate": 1.9774556678930343e-05, "loss": 1.9447, "step": 595 }, { "epoch": 0.1, "grad_norm": 7.980258464813232, "learning_rate": 1.9773454407070327e-05, "loss": 1.8569, "step": 596 }, { "epoch": 0.1, "grad_norm": 12.01597785949707, "learning_rate": 1.9772349477952297e-05, "loss": 1.8295, "step": 597 }, { "epoch": 0.1, "grad_norm": 13.773883819580078, "learning_rate": 1.9771241891876678e-05, "loss": 2.0759, "step": 598 }, { "epoch": 0.1, "grad_norm": 11.299097061157227, "learning_rate": 1.9770131649144604e-05, "loss": 1.8386, "step": 599 }, { "epoch": 0.1, "grad_norm": 9.351110458374023, "learning_rate": 1.9769018750057926e-05, "loss": 2.0076, "step": 600 }, { "epoch": 0.1, "grad_norm": 12.494894981384277, "learning_rate": 1.9767903194919234e-05, "loss": 1.803, "step": 601 }, { "epoch": 0.1, "grad_norm": 16.919883728027344, "learning_rate": 1.976678498403183e-05, "loss": 1.9602, "step": 602 }, { "epoch": 0.1, "grad_norm": 8.340797424316406, "learning_rate": 1.9765664117699735e-05, "loss": 1.9548, "step": 603 }, { "epoch": 0.1, "grad_norm": 15.473407745361328, "learning_rate": 1.9764540596227694e-05, "loss": 1.9663, "step": 604 }, { "epoch": 0.1, "grad_norm": 12.245420455932617, "learning_rate": 1.9763414419921184e-05, "loss": 1.9881, "step": 605 }, { "epoch": 0.1, "grad_norm": 10.641395568847656, "learning_rate": 1.976228558908639e-05, "loss": 1.8331, "step": 606 }, { "epoch": 0.1, "grad_norm": 9.95263957977295, "learning_rate": 1.9761154104030226e-05, "loss": 2.0493, "step": 607 }, { "epoch": 0.1, "grad_norm": 17.189661026000977, "learning_rate": 1.9760019965060323e-05, "loss": 1.8751, "step": 608 }, { "epoch": 0.1, "grad_norm": 9.976832389831543, "learning_rate": 1.975888317248504e-05, "loss": 1.9252, "step": 609 }, { "epoch": 0.1, "grad_norm": 9.717578887939453, "learning_rate": 1.9757743726613454e-05, "loss": 1.868, "step": 610 }, { "epoch": 0.1, "grad_norm": 10.502005577087402, "learning_rate": 1.9756601627755364e-05, "loss": 1.8526, "step": 611 }, { "epoch": 0.1, "grad_norm": 8.601312637329102, "learning_rate": 1.975545687622129e-05, "loss": 1.9336, "step": 612 }, { "epoch": 0.1, "grad_norm": 8.187820434570312, "learning_rate": 1.975430947232247e-05, "loss": 1.8356, "step": 613 }, { "epoch": 0.1, "grad_norm": 8.676957130432129, "learning_rate": 1.9753159416370867e-05, "loss": 2.1244, "step": 614 }, { "epoch": 0.1, "grad_norm": 10.822765350341797, "learning_rate": 1.975200670867917e-05, "loss": 1.8389, "step": 615 }, { "epoch": 0.1, "grad_norm": 8.355901718139648, "learning_rate": 1.9750851349560777e-05, "loss": 1.8611, "step": 616 }, { "epoch": 0.1, "grad_norm": 7.219264984130859, "learning_rate": 1.974969333932981e-05, "loss": 1.9861, "step": 617 }, { "epoch": 0.1, "grad_norm": 9.160279273986816, "learning_rate": 1.9748532678301125e-05, "loss": 1.9423, "step": 618 }, { "epoch": 0.1, "grad_norm": 11.32370376586914, "learning_rate": 1.974736936679028e-05, "loss": 2.1825, "step": 619 }, { "epoch": 0.1, "grad_norm": 8.126993179321289, "learning_rate": 1.9746203405113564e-05, "loss": 1.8259, "step": 620 }, { "epoch": 0.1, "grad_norm": 8.603165626525879, "learning_rate": 1.974503479358799e-05, "loss": 1.9595, "step": 621 }, { "epoch": 0.1, "grad_norm": 8.267172813415527, "learning_rate": 1.974386353253128e-05, "loss": 1.9281, "step": 622 }, { "epoch": 0.1, "grad_norm": 7.720278263092041, "learning_rate": 1.9742689622261885e-05, "loss": 1.9888, "step": 623 }, { "epoch": 0.1, "grad_norm": 7.697701454162598, "learning_rate": 1.9741513063098977e-05, "loss": 1.9351, "step": 624 }, { "epoch": 0.1, "grad_norm": 7.449602127075195, "learning_rate": 1.9740333855362437e-05, "loss": 1.9435, "step": 625 }, { "epoch": 0.1, "grad_norm": 8.163806915283203, "learning_rate": 1.973915199937288e-05, "loss": 1.8944, "step": 626 }, { "epoch": 0.1, "grad_norm": 10.113236427307129, "learning_rate": 1.9737967495451637e-05, "loss": 1.8564, "step": 627 }, { "epoch": 0.1, "grad_norm": 10.481661796569824, "learning_rate": 1.9736780343920753e-05, "loss": 2.0504, "step": 628 }, { "epoch": 0.1, "grad_norm": 9.520278930664062, "learning_rate": 1.9735590545103e-05, "loss": 2.0007, "step": 629 }, { "epoch": 0.1, "grad_norm": 12.512636184692383, "learning_rate": 1.9734398099321864e-05, "loss": 1.9081, "step": 630 }, { "epoch": 0.1, "grad_norm": 8.088497161865234, "learning_rate": 1.9733203006901554e-05, "loss": 1.9163, "step": 631 }, { "epoch": 0.1, "grad_norm": 9.715938568115234, "learning_rate": 1.9732005268166996e-05, "loss": 1.9094, "step": 632 }, { "epoch": 0.1, "grad_norm": 8.537522315979004, "learning_rate": 1.9730804883443843e-05, "loss": 2.0733, "step": 633 }, { "epoch": 0.1, "grad_norm": 12.150190353393555, "learning_rate": 1.972960185305846e-05, "loss": 2.1171, "step": 634 }, { "epoch": 0.1, "grad_norm": 11.53976058959961, "learning_rate": 1.972839617733793e-05, "loss": 2.0025, "step": 635 }, { "epoch": 0.1, "grad_norm": 6.871995449066162, "learning_rate": 1.9727187856610062e-05, "loss": 1.9262, "step": 636 }, { "epoch": 0.1, "grad_norm": 8.542471885681152, "learning_rate": 1.972597689120338e-05, "loss": 1.8614, "step": 637 }, { "epoch": 0.1, "grad_norm": 17.742006301879883, "learning_rate": 1.972476328144712e-05, "loss": 1.9967, "step": 638 }, { "epoch": 0.1, "grad_norm": 8.740364074707031, "learning_rate": 1.9723547027671256e-05, "loss": 1.9672, "step": 639 }, { "epoch": 0.1, "grad_norm": 9.744230270385742, "learning_rate": 1.9722328130206466e-05, "loss": 2.026, "step": 640 }, { "epoch": 0.1, "grad_norm": 11.60556411743164, "learning_rate": 1.972110658938415e-05, "loss": 1.8833, "step": 641 }, { "epoch": 0.1, "grad_norm": 11.98351001739502, "learning_rate": 1.9719882405536422e-05, "loss": 1.8038, "step": 642 }, { "epoch": 0.1, "grad_norm": 7.419538497924805, "learning_rate": 1.9718655578996125e-05, "loss": 1.8617, "step": 643 }, { "epoch": 0.1, "grad_norm": 9.932061195373535, "learning_rate": 1.9717426110096817e-05, "loss": 1.9813, "step": 644 }, { "epoch": 0.1, "grad_norm": 10.620423316955566, "learning_rate": 1.9716193999172764e-05, "loss": 1.8347, "step": 645 }, { "epoch": 0.1, "grad_norm": 8.535698890686035, "learning_rate": 1.9714959246558963e-05, "loss": 1.9354, "step": 646 }, { "epoch": 0.1, "grad_norm": 9.001425743103027, "learning_rate": 1.971372185259113e-05, "loss": 2.0003, "step": 647 }, { "epoch": 0.1, "grad_norm": 9.69876480102539, "learning_rate": 1.9712481817605687e-05, "loss": 2.0348, "step": 648 }, { "epoch": 0.1, "grad_norm": 9.921782493591309, "learning_rate": 1.971123914193979e-05, "loss": 1.972, "step": 649 }, { "epoch": 0.1, "grad_norm": 10.080999374389648, "learning_rate": 1.970999382593129e-05, "loss": 1.9466, "step": 650 }, { "epoch": 0.1, "grad_norm": 7.335169315338135, "learning_rate": 1.970874586991878e-05, "loss": 1.7702, "step": 651 }, { "epoch": 0.1, "grad_norm": 10.166914939880371, "learning_rate": 1.970749527424156e-05, "loss": 1.8204, "step": 652 }, { "epoch": 0.11, "grad_norm": 10.038626670837402, "learning_rate": 1.970624203923965e-05, "loss": 1.7992, "step": 653 }, { "epoch": 0.11, "grad_norm": 9.835798263549805, "learning_rate": 1.9704986165253784e-05, "loss": 1.8156, "step": 654 }, { "epoch": 0.11, "grad_norm": 8.230291366577148, "learning_rate": 1.9703727652625414e-05, "loss": 1.9564, "step": 655 }, { "epoch": 0.11, "grad_norm": 7.4532928466796875, "learning_rate": 1.970246650169671e-05, "loss": 1.975, "step": 656 }, { "epoch": 0.11, "grad_norm": 8.5479097366333, "learning_rate": 1.9701202712810564e-05, "loss": 1.8031, "step": 657 }, { "epoch": 0.11, "grad_norm": 9.45980167388916, "learning_rate": 1.9699936286310583e-05, "loss": 1.936, "step": 658 }, { "epoch": 0.11, "grad_norm": 7.6901702880859375, "learning_rate": 1.9698667222541083e-05, "loss": 1.8138, "step": 659 }, { "epoch": 0.11, "grad_norm": 9.391766548156738, "learning_rate": 1.969739552184711e-05, "loss": 1.8996, "step": 660 }, { "epoch": 0.11, "grad_norm": 8.4002046585083, "learning_rate": 1.9696121184574415e-05, "loss": 1.7739, "step": 661 }, { "epoch": 0.11, "grad_norm": 11.909111022949219, "learning_rate": 1.9694844211069477e-05, "loss": 1.9151, "step": 662 }, { "epoch": 0.11, "grad_norm": 9.340831756591797, "learning_rate": 1.9693564601679485e-05, "loss": 1.7436, "step": 663 }, { "epoch": 0.11, "grad_norm": 7.9699225425720215, "learning_rate": 1.9692282356752343e-05, "loss": 1.7347, "step": 664 }, { "epoch": 0.11, "grad_norm": 10.665328025817871, "learning_rate": 1.9690997476636678e-05, "loss": 1.7018, "step": 665 }, { "epoch": 0.11, "grad_norm": 10.52219009399414, "learning_rate": 1.9689709961681828e-05, "loss": 1.97, "step": 666 }, { "epoch": 0.11, "grad_norm": 10.570534706115723, "learning_rate": 1.9688419812237848e-05, "loss": 1.9848, "step": 667 }, { "epoch": 0.11, "grad_norm": 7.403249263763428, "learning_rate": 1.9687127028655515e-05, "loss": 1.8379, "step": 668 }, { "epoch": 0.11, "grad_norm": 12.492889404296875, "learning_rate": 1.9685831611286312e-05, "loss": 2.0044, "step": 669 }, { "epoch": 0.11, "grad_norm": 13.114709854125977, "learning_rate": 1.968453356048245e-05, "loss": 2.0188, "step": 670 }, { "epoch": 0.11, "grad_norm": 8.375856399536133, "learning_rate": 1.9683232876596846e-05, "loss": 2.0332, "step": 671 }, { "epoch": 0.11, "grad_norm": 13.001858711242676, "learning_rate": 1.9681929559983132e-05, "loss": 1.8128, "step": 672 }, { "epoch": 0.11, "grad_norm": 13.554468154907227, "learning_rate": 1.9680623610995672e-05, "loss": 1.9021, "step": 673 }, { "epoch": 0.11, "grad_norm": 8.947842597961426, "learning_rate": 1.9679315029989528e-05, "loss": 1.7216, "step": 674 }, { "epoch": 0.11, "grad_norm": 8.904035568237305, "learning_rate": 1.967800381732048e-05, "loss": 1.8132, "step": 675 }, { "epoch": 0.11, "grad_norm": 10.921860694885254, "learning_rate": 1.9676689973345035e-05, "loss": 1.9412, "step": 676 }, { "epoch": 0.11, "grad_norm": 9.393473625183105, "learning_rate": 1.9675373498420404e-05, "loss": 1.7089, "step": 677 }, { "epoch": 0.11, "grad_norm": 10.866217613220215, "learning_rate": 1.9674054392904516e-05, "loss": 1.9371, "step": 678 }, { "epoch": 0.11, "grad_norm": 8.387203216552734, "learning_rate": 1.967273265715602e-05, "loss": 1.8983, "step": 679 }, { "epoch": 0.11, "grad_norm": 8.611442565917969, "learning_rate": 1.967140829153427e-05, "loss": 1.8968, "step": 680 }, { "epoch": 0.11, "grad_norm": 8.629012107849121, "learning_rate": 1.967008129639935e-05, "loss": 1.9912, "step": 681 }, { "epoch": 0.11, "grad_norm": 9.056966781616211, "learning_rate": 1.9668751672112044e-05, "loss": 1.8513, "step": 682 }, { "epoch": 0.11, "grad_norm": 9.120896339416504, "learning_rate": 1.966741941903386e-05, "loss": 1.8989, "step": 683 }, { "epoch": 0.11, "grad_norm": 9.260690689086914, "learning_rate": 1.9666084537527018e-05, "loss": 1.6848, "step": 684 }, { "epoch": 0.11, "grad_norm": 8.508567810058594, "learning_rate": 1.966474702795445e-05, "loss": 1.6113, "step": 685 }, { "epoch": 0.11, "grad_norm": 7.16511344909668, "learning_rate": 1.9663406890679808e-05, "loss": 1.7732, "step": 686 }, { "epoch": 0.11, "grad_norm": 11.877750396728516, "learning_rate": 1.966206412606745e-05, "loss": 1.8677, "step": 687 }, { "epoch": 0.11, "grad_norm": 8.79915714263916, "learning_rate": 1.9660718734482464e-05, "loss": 1.914, "step": 688 }, { "epoch": 0.11, "grad_norm": 9.607056617736816, "learning_rate": 1.9659370716290635e-05, "loss": 1.9648, "step": 689 }, { "epoch": 0.11, "grad_norm": 14.735142707824707, "learning_rate": 1.965802007185847e-05, "loss": 1.8569, "step": 690 }, { "epoch": 0.11, "grad_norm": 9.858439445495605, "learning_rate": 1.965666680155319e-05, "loss": 2.0916, "step": 691 }, { "epoch": 0.11, "grad_norm": 12.390646934509277, "learning_rate": 1.9655310905742727e-05, "loss": 1.736, "step": 692 }, { "epoch": 0.11, "grad_norm": 7.830264091491699, "learning_rate": 1.9653952384795735e-05, "loss": 1.9819, "step": 693 }, { "epoch": 0.11, "grad_norm": 10.934769630432129, "learning_rate": 1.965259123908157e-05, "loss": 2.0495, "step": 694 }, { "epoch": 0.11, "grad_norm": 14.617569923400879, "learning_rate": 1.965122746897031e-05, "loss": 2.0442, "step": 695 }, { "epoch": 0.11, "grad_norm": 9.410893440246582, "learning_rate": 1.9649861074832745e-05, "loss": 1.9556, "step": 696 }, { "epoch": 0.11, "grad_norm": 6.869527816772461, "learning_rate": 1.9648492057040372e-05, "loss": 1.9457, "step": 697 }, { "epoch": 0.11, "grad_norm": 13.391074180603027, "learning_rate": 1.9647120415965414e-05, "loss": 2.1628, "step": 698 }, { "epoch": 0.11, "grad_norm": 14.018997192382812, "learning_rate": 1.9645746151980794e-05, "loss": 1.9614, "step": 699 }, { "epoch": 0.11, "grad_norm": 7.843084335327148, "learning_rate": 1.9644369265460156e-05, "loss": 1.7797, "step": 700 }, { "epoch": 0.11, "grad_norm": 10.793905258178711, "learning_rate": 1.9642989756777858e-05, "loss": 1.713, "step": 701 }, { "epoch": 0.11, "grad_norm": 10.845542907714844, "learning_rate": 1.9641607626308966e-05, "loss": 1.9489, "step": 702 }, { "epoch": 0.11, "grad_norm": 9.173194885253906, "learning_rate": 1.964022287442926e-05, "loss": 2.0238, "step": 703 }, { "epoch": 0.11, "grad_norm": 8.731972694396973, "learning_rate": 1.9638835501515233e-05, "loss": 1.9609, "step": 704 }, { "epoch": 0.11, "grad_norm": 8.310286521911621, "learning_rate": 1.9637445507944096e-05, "loss": 1.9306, "step": 705 }, { "epoch": 0.11, "grad_norm": 8.123601913452148, "learning_rate": 1.963605289409376e-05, "loss": 1.841, "step": 706 }, { "epoch": 0.11, "grad_norm": 8.14062213897705, "learning_rate": 1.9634657660342863e-05, "loss": 1.635, "step": 707 }, { "epoch": 0.11, "grad_norm": 10.467580795288086, "learning_rate": 1.9633259807070744e-05, "loss": 1.8012, "step": 708 }, { "epoch": 0.11, "grad_norm": 9.101387977600098, "learning_rate": 1.9631859334657465e-05, "loss": 2.1036, "step": 709 }, { "epoch": 0.11, "grad_norm": 8.089895248413086, "learning_rate": 1.9630456243483782e-05, "loss": 1.8556, "step": 710 }, { "epoch": 0.11, "grad_norm": 8.328288078308105, "learning_rate": 1.9629050533931187e-05, "loss": 1.7363, "step": 711 }, { "epoch": 0.11, "grad_norm": 8.447441101074219, "learning_rate": 1.9627642206381864e-05, "loss": 1.7635, "step": 712 }, { "epoch": 0.11, "grad_norm": 9.396626472473145, "learning_rate": 1.9626231261218723e-05, "loss": 2.0777, "step": 713 }, { "epoch": 0.11, "grad_norm": 8.959794044494629, "learning_rate": 1.9624817698825372e-05, "loss": 1.8753, "step": 714 }, { "epoch": 0.12, "grad_norm": 9.205187797546387, "learning_rate": 1.9623401519586143e-05, "loss": 1.964, "step": 715 }, { "epoch": 0.12, "grad_norm": 9.171693801879883, "learning_rate": 1.9621982723886073e-05, "loss": 1.7152, "step": 716 }, { "epoch": 0.12, "grad_norm": 8.459547996520996, "learning_rate": 1.962056131211091e-05, "loss": 1.9853, "step": 717 }, { "epoch": 0.12, "grad_norm": 8.175357818603516, "learning_rate": 1.9619137284647114e-05, "loss": 1.7674, "step": 718 }, { "epoch": 0.12, "grad_norm": 7.671604633331299, "learning_rate": 1.961771064188186e-05, "loss": 1.8992, "step": 719 }, { "epoch": 0.12, "grad_norm": 8.327662467956543, "learning_rate": 1.961628138420303e-05, "loss": 1.8373, "step": 720 }, { "epoch": 0.12, "grad_norm": 7.177297115325928, "learning_rate": 1.9614849511999217e-05, "loss": 1.8047, "step": 721 }, { "epoch": 0.12, "grad_norm": 12.107439041137695, "learning_rate": 1.961341502565973e-05, "loss": 1.9924, "step": 722 }, { "epoch": 0.12, "grad_norm": 11.040671348571777, "learning_rate": 1.961197792557458e-05, "loss": 1.8879, "step": 723 }, { "epoch": 0.12, "grad_norm": 10.032115936279297, "learning_rate": 1.9610538212134494e-05, "loss": 2.0092, "step": 724 }, { "epoch": 0.12, "grad_norm": 12.196745872497559, "learning_rate": 1.9609095885730913e-05, "loss": 1.8914, "step": 725 }, { "epoch": 0.12, "grad_norm": 13.464482307434082, "learning_rate": 1.960765094675598e-05, "loss": 1.8135, "step": 726 }, { "epoch": 0.12, "grad_norm": 9.176491737365723, "learning_rate": 1.960620339560255e-05, "loss": 1.7822, "step": 727 }, { "epoch": 0.12, "grad_norm": 12.16747760772705, "learning_rate": 1.96047532326642e-05, "loss": 1.8433, "step": 728 }, { "epoch": 0.12, "grad_norm": 9.520370483398438, "learning_rate": 1.96033004583352e-05, "loss": 1.882, "step": 729 }, { "epoch": 0.12, "grad_norm": 9.27993392944336, "learning_rate": 1.960184507301054e-05, "loss": 1.9486, "step": 730 }, { "epoch": 0.12, "grad_norm": 9.89235782623291, "learning_rate": 1.9600387077085918e-05, "loss": 1.8209, "step": 731 }, { "epoch": 0.12, "grad_norm": 13.287773132324219, "learning_rate": 1.959892647095774e-05, "loss": 1.9463, "step": 732 }, { "epoch": 0.12, "grad_norm": 12.676133155822754, "learning_rate": 1.9597463255023127e-05, "loss": 1.8582, "step": 733 }, { "epoch": 0.12, "grad_norm": 10.17635726928711, "learning_rate": 1.9595997429679904e-05, "loss": 1.916, "step": 734 }, { "epoch": 0.12, "grad_norm": 12.76540470123291, "learning_rate": 1.9594528995326604e-05, "loss": 1.9906, "step": 735 }, { "epoch": 0.12, "grad_norm": 8.52157211303711, "learning_rate": 1.9593057952362476e-05, "loss": 1.8644, "step": 736 }, { "epoch": 0.12, "grad_norm": 8.743395805358887, "learning_rate": 1.9591584301187477e-05, "loss": 1.9449, "step": 737 }, { "epoch": 0.12, "grad_norm": 11.511383056640625, "learning_rate": 1.959010804220227e-05, "loss": 1.9063, "step": 738 }, { "epoch": 0.12, "grad_norm": 11.870858192443848, "learning_rate": 1.9588629175808224e-05, "loss": 1.905, "step": 739 }, { "epoch": 0.12, "grad_norm": 8.020773887634277, "learning_rate": 1.958714770240742e-05, "loss": 2.0446, "step": 740 }, { "epoch": 0.12, "grad_norm": 7.760090351104736, "learning_rate": 1.958566362240266e-05, "loss": 1.7785, "step": 741 }, { "epoch": 0.12, "grad_norm": 10.804485321044922, "learning_rate": 1.958417693619743e-05, "loss": 1.7129, "step": 742 }, { "epoch": 0.12, "grad_norm": 9.163077354431152, "learning_rate": 1.9582687644195942e-05, "loss": 2.0705, "step": 743 }, { "epoch": 0.12, "grad_norm": 7.873098373413086, "learning_rate": 1.958119574680312e-05, "loss": 1.736, "step": 744 }, { "epoch": 0.12, "grad_norm": 9.573057174682617, "learning_rate": 1.957970124442458e-05, "loss": 1.9894, "step": 745 }, { "epoch": 0.12, "grad_norm": 8.843155860900879, "learning_rate": 1.957820413746666e-05, "loss": 1.9403, "step": 746 }, { "epoch": 0.12, "grad_norm": 9.63378620147705, "learning_rate": 1.9576704426336395e-05, "loss": 2.09, "step": 747 }, { "epoch": 0.12, "grad_norm": 8.940881729125977, "learning_rate": 1.9575202111441544e-05, "loss": 1.9347, "step": 748 }, { "epoch": 0.12, "grad_norm": 10.372617721557617, "learning_rate": 1.9573697193190553e-05, "loss": 1.7979, "step": 749 }, { "epoch": 0.12, "grad_norm": 8.253705024719238, "learning_rate": 1.95721896719926e-05, "loss": 1.7716, "step": 750 }, { "epoch": 0.12, "grad_norm": 8.920816421508789, "learning_rate": 1.9570679548257547e-05, "loss": 2.0811, "step": 751 }, { "epoch": 0.12, "grad_norm": 8.125235557556152, "learning_rate": 1.956916682239598e-05, "loss": 1.8967, "step": 752 }, { "epoch": 0.12, "grad_norm": 9.426767349243164, "learning_rate": 1.9567651494819184e-05, "loss": 2.082, "step": 753 }, { "epoch": 0.12, "grad_norm": 7.489150047302246, "learning_rate": 1.9566133565939157e-05, "loss": 2.0041, "step": 754 }, { "epoch": 0.12, "grad_norm": 7.165369987487793, "learning_rate": 1.95646130361686e-05, "loss": 1.9758, "step": 755 }, { "epoch": 0.12, "grad_norm": 8.526873588562012, "learning_rate": 1.9563089905920916e-05, "loss": 1.8965, "step": 756 }, { "epoch": 0.12, "grad_norm": 10.800965309143066, "learning_rate": 1.9561564175610232e-05, "loss": 1.9045, "step": 757 }, { "epoch": 0.12, "grad_norm": 7.679756164550781, "learning_rate": 1.956003584565137e-05, "loss": 1.8796, "step": 758 }, { "epoch": 0.12, "grad_norm": 7.917092323303223, "learning_rate": 1.9558504916459857e-05, "loss": 1.856, "step": 759 }, { "epoch": 0.12, "grad_norm": 9.39007568359375, "learning_rate": 1.9556971388451932e-05, "loss": 1.8499, "step": 760 }, { "epoch": 0.12, "grad_norm": 9.076805114746094, "learning_rate": 1.9555435262044535e-05, "loss": 2.0699, "step": 761 }, { "epoch": 0.12, "grad_norm": 9.079532623291016, "learning_rate": 1.9553896537655317e-05, "loss": 1.8161, "step": 762 }, { "epoch": 0.12, "grad_norm": 7.889411449432373, "learning_rate": 1.955235521570264e-05, "loss": 1.8786, "step": 763 }, { "epoch": 0.12, "grad_norm": 8.758649826049805, "learning_rate": 1.955081129660556e-05, "loss": 1.915, "step": 764 }, { "epoch": 0.12, "grad_norm": 8.949542999267578, "learning_rate": 1.954926478078385e-05, "loss": 1.9239, "step": 765 }, { "epoch": 0.12, "grad_norm": 9.330246925354004, "learning_rate": 1.9547715668657983e-05, "loss": 1.9425, "step": 766 }, { "epoch": 0.12, "grad_norm": 7.474038124084473, "learning_rate": 1.954616396064914e-05, "loss": 1.9035, "step": 767 }, { "epoch": 0.12, "grad_norm": 10.93132495880127, "learning_rate": 1.954460965717921e-05, "loss": 1.9152, "step": 768 }, { "epoch": 0.12, "grad_norm": 7.494057655334473, "learning_rate": 1.954305275867078e-05, "loss": 1.7497, "step": 769 }, { "epoch": 0.12, "grad_norm": 7.993947982788086, "learning_rate": 1.954149326554715e-05, "loss": 1.8933, "step": 770 }, { "epoch": 0.12, "grad_norm": 8.042375564575195, "learning_rate": 1.9539931178232328e-05, "loss": 1.9218, "step": 771 }, { "epoch": 0.12, "grad_norm": 7.545215606689453, "learning_rate": 1.9538366497151012e-05, "loss": 1.8096, "step": 772 }, { "epoch": 0.12, "grad_norm": 7.863738536834717, "learning_rate": 1.9536799222728628e-05, "loss": 2.0154, "step": 773 }, { "epoch": 0.12, "grad_norm": 7.315031051635742, "learning_rate": 1.9535229355391286e-05, "loss": 1.8496, "step": 774 }, { "epoch": 0.12, "grad_norm": 9.098179817199707, "learning_rate": 1.9533656895565813e-05, "loss": 1.6593, "step": 775 }, { "epoch": 0.12, "grad_norm": 10.530941009521484, "learning_rate": 1.9532081843679734e-05, "loss": 1.7566, "step": 776 }, { "epoch": 0.13, "grad_norm": 8.448867797851562, "learning_rate": 1.953050420016129e-05, "loss": 1.712, "step": 777 }, { "epoch": 0.13, "grad_norm": 8.754901885986328, "learning_rate": 1.9528923965439414e-05, "loss": 1.705, "step": 778 }, { "epoch": 0.13, "grad_norm": 8.162744522094727, "learning_rate": 1.952734113994375e-05, "loss": 1.9533, "step": 779 }, { "epoch": 0.13, "grad_norm": 13.836995124816895, "learning_rate": 1.952575572410464e-05, "loss": 1.9845, "step": 780 }, { "epoch": 0.13, "grad_norm": 10.173358917236328, "learning_rate": 1.952416771835314e-05, "loss": 2.0309, "step": 781 }, { "epoch": 0.13, "grad_norm": 7.2666544914245605, "learning_rate": 1.9522577123121005e-05, "loss": 2.058, "step": 782 }, { "epoch": 0.13, "grad_norm": 16.6480655670166, "learning_rate": 1.9520983938840695e-05, "loss": 1.9874, "step": 783 }, { "epoch": 0.13, "grad_norm": 10.507447242736816, "learning_rate": 1.951938816594537e-05, "loss": 1.8549, "step": 784 }, { "epoch": 0.13, "grad_norm": 8.08291244506836, "learning_rate": 1.9517789804868902e-05, "loss": 1.947, "step": 785 }, { "epoch": 0.13, "grad_norm": 9.837760925292969, "learning_rate": 1.9516188856045854e-05, "loss": 2.047, "step": 786 }, { "epoch": 0.13, "grad_norm": 9.818593978881836, "learning_rate": 1.951458531991151e-05, "loss": 1.9956, "step": 787 }, { "epoch": 0.13, "grad_norm": 9.336548805236816, "learning_rate": 1.951297919690184e-05, "loss": 1.8343, "step": 788 }, { "epoch": 0.13, "grad_norm": 9.977380752563477, "learning_rate": 1.9511370487453528e-05, "loss": 1.8984, "step": 789 }, { "epoch": 0.13, "grad_norm": 9.501103401184082, "learning_rate": 1.950975919200396e-05, "loss": 2.066, "step": 790 }, { "epoch": 0.13, "grad_norm": 8.311579704284668, "learning_rate": 1.9508145310991223e-05, "loss": 1.857, "step": 791 }, { "epoch": 0.13, "grad_norm": 9.418643951416016, "learning_rate": 1.95065288448541e-05, "loss": 1.8864, "step": 792 }, { "epoch": 0.13, "grad_norm": 11.081645965576172, "learning_rate": 1.95049097940321e-05, "loss": 1.8237, "step": 793 }, { "epoch": 0.13, "grad_norm": 8.833626747131348, "learning_rate": 1.9503288158965405e-05, "loss": 2.0222, "step": 794 }, { "epoch": 0.13, "grad_norm": 7.923919200897217, "learning_rate": 1.9501663940094918e-05, "loss": 1.8113, "step": 795 }, { "epoch": 0.13, "grad_norm": 10.206025123596191, "learning_rate": 1.9500037137862243e-05, "loss": 1.9886, "step": 796 }, { "epoch": 0.13, "grad_norm": 9.485471725463867, "learning_rate": 1.949840775270968e-05, "loss": 1.9434, "step": 797 }, { "epoch": 0.13, "grad_norm": 9.501248359680176, "learning_rate": 1.9496775785080233e-05, "loss": 1.9163, "step": 798 }, { "epoch": 0.13, "grad_norm": 8.113144874572754, "learning_rate": 1.9495141235417617e-05, "loss": 1.8229, "step": 799 }, { "epoch": 0.13, "grad_norm": 8.545166015625, "learning_rate": 1.9493504104166235e-05, "loss": 1.8162, "step": 800 }, { "epoch": 0.13, "grad_norm": 10.538179397583008, "learning_rate": 1.9491864391771204e-05, "loss": 1.9042, "step": 801 }, { "epoch": 0.13, "grad_norm": 8.548885345458984, "learning_rate": 1.949022209867833e-05, "loss": 2.017, "step": 802 }, { "epoch": 0.13, "grad_norm": 9.039024353027344, "learning_rate": 1.9488577225334137e-05, "loss": 1.839, "step": 803 }, { "epoch": 0.13, "grad_norm": 8.235259056091309, "learning_rate": 1.9486929772185837e-05, "loss": 2.0145, "step": 804 }, { "epoch": 0.13, "grad_norm": 8.389944076538086, "learning_rate": 1.948527973968135e-05, "loss": 1.857, "step": 805 }, { "epoch": 0.13, "grad_norm": 7.415011882781982, "learning_rate": 1.9483627128269294e-05, "loss": 1.7525, "step": 806 }, { "epoch": 0.13, "grad_norm": 9.848081588745117, "learning_rate": 1.9481971938398994e-05, "loss": 1.8072, "step": 807 }, { "epoch": 0.13, "grad_norm": 9.925073623657227, "learning_rate": 1.9480314170520464e-05, "loss": 1.7555, "step": 808 }, { "epoch": 0.13, "grad_norm": 8.4024076461792, "learning_rate": 1.947865382508443e-05, "loss": 1.9445, "step": 809 }, { "epoch": 0.13, "grad_norm": 7.757234573364258, "learning_rate": 1.947699090254232e-05, "loss": 1.8245, "step": 810 }, { "epoch": 0.13, "grad_norm": 14.36515998840332, "learning_rate": 1.9475325403346252e-05, "loss": 2.2769, "step": 811 }, { "epoch": 0.13, "grad_norm": 9.607050895690918, "learning_rate": 1.9473657327949055e-05, "loss": 2.0324, "step": 812 }, { "epoch": 0.13, "grad_norm": 8.187308311462402, "learning_rate": 1.9471986676804252e-05, "loss": 1.8542, "step": 813 }, { "epoch": 0.13, "grad_norm": 7.434091567993164, "learning_rate": 1.9470313450366064e-05, "loss": 1.7881, "step": 814 }, { "epoch": 0.13, "grad_norm": 10.762941360473633, "learning_rate": 1.9468637649089425e-05, "loss": 2.0338, "step": 815 }, { "epoch": 0.13, "grad_norm": 8.162681579589844, "learning_rate": 1.9466959273429957e-05, "loss": 1.8977, "step": 816 }, { "epoch": 0.13, "grad_norm": 7.697405815124512, "learning_rate": 1.946527832384398e-05, "loss": 1.884, "step": 817 }, { "epoch": 0.13, "grad_norm": 8.317020416259766, "learning_rate": 1.946359480078853e-05, "loss": 1.7329, "step": 818 }, { "epoch": 0.13, "grad_norm": 8.48955249786377, "learning_rate": 1.9461908704721325e-05, "loss": 1.9181, "step": 819 }, { "epoch": 0.13, "grad_norm": 8.46105670928955, "learning_rate": 1.946022003610079e-05, "loss": 1.7785, "step": 820 }, { "epoch": 0.13, "grad_norm": 9.182477951049805, "learning_rate": 1.9458528795386052e-05, "loss": 1.7395, "step": 821 }, { "epoch": 0.13, "grad_norm": 9.289976119995117, "learning_rate": 1.945683498303693e-05, "loss": 1.6425, "step": 822 }, { "epoch": 0.13, "grad_norm": 10.86405086517334, "learning_rate": 1.9455138599513952e-05, "loss": 1.8792, "step": 823 }, { "epoch": 0.13, "grad_norm": 8.369268417358398, "learning_rate": 1.9453439645278335e-05, "loss": 1.7076, "step": 824 }, { "epoch": 0.13, "grad_norm": 10.495741844177246, "learning_rate": 1.9451738120792007e-05, "loss": 1.8407, "step": 825 }, { "epoch": 0.13, "grad_norm": 8.548133850097656, "learning_rate": 1.9450034026517577e-05, "loss": 1.8066, "step": 826 }, { "epoch": 0.13, "grad_norm": 8.318772315979004, "learning_rate": 1.9448327362918366e-05, "loss": 1.8347, "step": 827 }, { "epoch": 0.13, "grad_norm": 7.562827110290527, "learning_rate": 1.9446618130458395e-05, "loss": 1.9553, "step": 828 }, { "epoch": 0.13, "grad_norm": 10.465840339660645, "learning_rate": 1.9444906329602377e-05, "loss": 1.7942, "step": 829 }, { "epoch": 0.13, "grad_norm": 9.992239952087402, "learning_rate": 1.9443191960815726e-05, "loss": 2.008, "step": 830 }, { "epoch": 0.13, "grad_norm": 15.241172790527344, "learning_rate": 1.9441475024564554e-05, "loss": 1.9404, "step": 831 }, { "epoch": 0.13, "grad_norm": 8.733748435974121, "learning_rate": 1.9439755521315667e-05, "loss": 1.9775, "step": 832 }, { "epoch": 0.13, "grad_norm": 8.122631072998047, "learning_rate": 1.9438033451536574e-05, "loss": 1.7068, "step": 833 }, { "epoch": 0.13, "grad_norm": 12.876418113708496, "learning_rate": 1.9436308815695484e-05, "loss": 1.9697, "step": 834 }, { "epoch": 0.13, "grad_norm": 8.738603591918945, "learning_rate": 1.9434581614261293e-05, "loss": 1.8777, "step": 835 }, { "epoch": 0.13, "grad_norm": 7.7032575607299805, "learning_rate": 1.943285184770361e-05, "loss": 2.034, "step": 836 }, { "epoch": 0.13, "grad_norm": 8.313472747802734, "learning_rate": 1.9431119516492725e-05, "loss": 1.9587, "step": 837 }, { "epoch": 0.13, "grad_norm": 8.703226089477539, "learning_rate": 1.942938462109964e-05, "loss": 1.9867, "step": 838 }, { "epoch": 0.14, "grad_norm": 9.704682350158691, "learning_rate": 1.9427647161996046e-05, "loss": 1.9645, "step": 839 }, { "epoch": 0.14, "grad_norm": 10.227317810058594, "learning_rate": 1.942590713965433e-05, "loss": 1.7411, "step": 840 }, { "epoch": 0.14, "grad_norm": 9.338072776794434, "learning_rate": 1.942416455454758e-05, "loss": 1.6725, "step": 841 }, { "epoch": 0.14, "grad_norm": 7.598278045654297, "learning_rate": 1.9422419407149572e-05, "loss": 1.9977, "step": 842 }, { "epoch": 0.14, "grad_norm": 8.540964126586914, "learning_rate": 1.9420671697934802e-05, "loss": 1.9157, "step": 843 }, { "epoch": 0.14, "grad_norm": 10.322505950927734, "learning_rate": 1.941892142737843e-05, "loss": 1.6554, "step": 844 }, { "epoch": 0.14, "grad_norm": 9.783913612365723, "learning_rate": 1.941716859595634e-05, "loss": 1.7499, "step": 845 }, { "epoch": 0.14, "grad_norm": 8.812719345092773, "learning_rate": 1.9415413204145094e-05, "loss": 1.9239, "step": 846 }, { "epoch": 0.14, "grad_norm": 8.231749534606934, "learning_rate": 1.9413655252421958e-05, "loss": 2.101, "step": 847 }, { "epoch": 0.14, "grad_norm": 10.67330265045166, "learning_rate": 1.94118947412649e-05, "loss": 1.7097, "step": 848 }, { "epoch": 0.14, "grad_norm": 8.024706840515137, "learning_rate": 1.9410131671152567e-05, "loss": 2.0081, "step": 849 }, { "epoch": 0.14, "grad_norm": 7.1970343589782715, "learning_rate": 1.9408366042564316e-05, "loss": 1.8906, "step": 850 }, { "epoch": 0.14, "grad_norm": 8.355569839477539, "learning_rate": 1.9406597855980197e-05, "loss": 2.0641, "step": 851 }, { "epoch": 0.14, "grad_norm": 8.432036399841309, "learning_rate": 1.9404827111880955e-05, "loss": 1.9409, "step": 852 }, { "epoch": 0.14, "grad_norm": 9.234162330627441, "learning_rate": 1.9403053810748026e-05, "loss": 1.9358, "step": 853 }, { "epoch": 0.14, "grad_norm": 9.237397193908691, "learning_rate": 1.9401277953063546e-05, "loss": 1.877, "step": 854 }, { "epoch": 0.14, "grad_norm": 8.711994171142578, "learning_rate": 1.9399499539310343e-05, "loss": 1.8859, "step": 855 }, { "epoch": 0.14, "grad_norm": 35.93690490722656, "learning_rate": 1.939771856997194e-05, "loss": 1.7197, "step": 856 }, { "epoch": 0.14, "grad_norm": 8.632305145263672, "learning_rate": 1.9395935045532562e-05, "loss": 1.7785, "step": 857 }, { "epoch": 0.14, "grad_norm": 8.223588943481445, "learning_rate": 1.9394148966477118e-05, "loss": 1.8323, "step": 858 }, { "epoch": 0.14, "grad_norm": 6.490678310394287, "learning_rate": 1.9392360333291217e-05, "loss": 1.7533, "step": 859 }, { "epoch": 0.14, "grad_norm": 11.044794082641602, "learning_rate": 1.9390569146461167e-05, "loss": 1.8644, "step": 860 }, { "epoch": 0.14, "grad_norm": 9.696097373962402, "learning_rate": 1.9388775406473955e-05, "loss": 1.7544, "step": 861 }, { "epoch": 0.14, "grad_norm": 9.618382453918457, "learning_rate": 1.9386979113817283e-05, "loss": 1.7975, "step": 862 }, { "epoch": 0.14, "grad_norm": 9.140779495239258, "learning_rate": 1.938518026897953e-05, "loss": 1.8617, "step": 863 }, { "epoch": 0.14, "grad_norm": 8.355779647827148, "learning_rate": 1.9383378872449778e-05, "loss": 1.7769, "step": 864 }, { "epoch": 0.14, "grad_norm": 8.469406127929688, "learning_rate": 1.9381574924717805e-05, "loss": 1.8661, "step": 865 }, { "epoch": 0.14, "grad_norm": 7.520668983459473, "learning_rate": 1.9379768426274067e-05, "loss": 1.9598, "step": 866 }, { "epoch": 0.14, "grad_norm": 17.910768508911133, "learning_rate": 1.937795937760973e-05, "loss": 1.7358, "step": 867 }, { "epoch": 0.14, "grad_norm": 8.54953670501709, "learning_rate": 1.9376147779216644e-05, "loss": 2.0158, "step": 868 }, { "epoch": 0.14, "grad_norm": 7.78688907623291, "learning_rate": 1.9374333631587365e-05, "loss": 2.0564, "step": 869 }, { "epoch": 0.14, "grad_norm": 18.75167465209961, "learning_rate": 1.9372516935215123e-05, "loss": 1.8499, "step": 870 }, { "epoch": 0.14, "grad_norm": 9.291059494018555, "learning_rate": 1.937069769059386e-05, "loss": 1.9022, "step": 871 }, { "epoch": 0.14, "grad_norm": 8.766959190368652, "learning_rate": 1.936887589821819e-05, "loss": 1.9489, "step": 872 }, { "epoch": 0.14, "grad_norm": 12.430251121520996, "learning_rate": 1.9367051558583445e-05, "loss": 1.9467, "step": 873 }, { "epoch": 0.14, "grad_norm": 12.322244644165039, "learning_rate": 1.9365224672185626e-05, "loss": 1.9404, "step": 874 }, { "epoch": 0.14, "grad_norm": 6.9517645835876465, "learning_rate": 1.9363395239521437e-05, "loss": 1.8945, "step": 875 }, { "epoch": 0.14, "grad_norm": 19.317882537841797, "learning_rate": 1.936156326108828e-05, "loss": 2.058, "step": 876 }, { "epoch": 0.14, "grad_norm": 8.687305450439453, "learning_rate": 1.935972873738424e-05, "loss": 1.825, "step": 877 }, { "epoch": 0.14, "grad_norm": 8.844247817993164, "learning_rate": 1.9357891668908097e-05, "loss": 2.0982, "step": 878 }, { "epoch": 0.14, "grad_norm": 7.811300754547119, "learning_rate": 1.935605205615932e-05, "loss": 1.8504, "step": 879 }, { "epoch": 0.14, "grad_norm": 12.569132804870605, "learning_rate": 1.9354209899638074e-05, "loss": 1.9573, "step": 880 }, { "epoch": 0.14, "grad_norm": 8.134172439575195, "learning_rate": 1.9352365199845215e-05, "loss": 1.7321, "step": 881 }, { "epoch": 0.14, "grad_norm": 7.210638523101807, "learning_rate": 1.9350517957282292e-05, "loss": 1.8937, "step": 882 }, { "epoch": 0.14, "grad_norm": 8.676950454711914, "learning_rate": 1.934866817245154e-05, "loss": 1.9877, "step": 883 }, { "epoch": 0.14, "grad_norm": 10.266532897949219, "learning_rate": 1.9346815845855887e-05, "loss": 1.7316, "step": 884 }, { "epoch": 0.14, "grad_norm": 9.031750679016113, "learning_rate": 1.9344960977998958e-05, "loss": 1.9415, "step": 885 }, { "epoch": 0.14, "grad_norm": 7.5076799392700195, "learning_rate": 1.9343103569385057e-05, "loss": 1.7133, "step": 886 }, { "epoch": 0.14, "grad_norm": 8.674040794372559, "learning_rate": 1.934124362051919e-05, "loss": 1.6515, "step": 887 }, { "epoch": 0.14, "grad_norm": 16.119592666625977, "learning_rate": 1.9339381131907053e-05, "loss": 1.8419, "step": 888 }, { "epoch": 0.14, "grad_norm": 11.052323341369629, "learning_rate": 1.9337516104055022e-05, "loss": 1.6704, "step": 889 }, { "epoch": 0.14, "grad_norm": 8.385467529296875, "learning_rate": 1.9335648537470177e-05, "loss": 1.8814, "step": 890 }, { "epoch": 0.14, "grad_norm": 8.906031608581543, "learning_rate": 1.933377843266028e-05, "loss": 2.0001, "step": 891 }, { "epoch": 0.14, "grad_norm": 9.579660415649414, "learning_rate": 1.9331905790133784e-05, "loss": 2.1433, "step": 892 }, { "epoch": 0.14, "grad_norm": 10.098712921142578, "learning_rate": 1.933003061039983e-05, "loss": 1.8289, "step": 893 }, { "epoch": 0.14, "grad_norm": 7.781762599945068, "learning_rate": 1.932815289396826e-05, "loss": 1.9012, "step": 894 }, { "epoch": 0.14, "grad_norm": 7.2602949142456055, "learning_rate": 1.932627264134959e-05, "loss": 1.872, "step": 895 }, { "epoch": 0.14, "grad_norm": 7.19596529006958, "learning_rate": 1.9324389853055034e-05, "loss": 1.7974, "step": 896 }, { "epoch": 0.14, "grad_norm": 9.633811950683594, "learning_rate": 1.9322504529596492e-05, "loss": 1.9771, "step": 897 }, { "epoch": 0.14, "grad_norm": 9.863388061523438, "learning_rate": 1.9320616671486565e-05, "loss": 1.8692, "step": 898 }, { "epoch": 0.14, "grad_norm": 7.537428379058838, "learning_rate": 1.9318726279238526e-05, "loss": 1.8881, "step": 899 }, { "epoch": 0.14, "grad_norm": 9.47636604309082, "learning_rate": 1.9316833353366348e-05, "loss": 1.8317, "step": 900 }, { "epoch": 0.15, "grad_norm": 8.940594673156738, "learning_rate": 1.9314937894384683e-05, "loss": 1.8488, "step": 901 }, { "epoch": 0.15, "grad_norm": 8.52361011505127, "learning_rate": 1.9313039902808887e-05, "loss": 1.8501, "step": 902 }, { "epoch": 0.15, "grad_norm": 7.629300594329834, "learning_rate": 1.9311139379154992e-05, "loss": 1.9057, "step": 903 }, { "epoch": 0.15, "grad_norm": 8.705419540405273, "learning_rate": 1.9309236323939724e-05, "loss": 1.8065, "step": 904 }, { "epoch": 0.15, "grad_norm": 10.996777534484863, "learning_rate": 1.930733073768049e-05, "loss": 1.9482, "step": 905 }, { "epoch": 0.15, "grad_norm": 7.986311912536621, "learning_rate": 1.93054226208954e-05, "loss": 1.9112, "step": 906 }, { "epoch": 0.15, "grad_norm": 9.520552635192871, "learning_rate": 1.9303511974103237e-05, "loss": 1.827, "step": 907 }, { "epoch": 0.15, "grad_norm": 9.581088066101074, "learning_rate": 1.9301598797823477e-05, "loss": 1.8467, "step": 908 }, { "epoch": 0.15, "grad_norm": 8.614437103271484, "learning_rate": 1.9299683092576287e-05, "loss": 1.74, "step": 909 }, { "epoch": 0.15, "grad_norm": 8.929459571838379, "learning_rate": 1.9297764858882516e-05, "loss": 2.0064, "step": 910 }, { "epoch": 0.15, "grad_norm": 9.63720989227295, "learning_rate": 1.9295844097263705e-05, "loss": 1.7836, "step": 911 }, { "epoch": 0.15, "grad_norm": 10.155715942382812, "learning_rate": 1.9293920808242084e-05, "loss": 1.698, "step": 912 }, { "epoch": 0.15, "grad_norm": 8.999330520629883, "learning_rate": 1.9291994992340563e-05, "loss": 2.0505, "step": 913 }, { "epoch": 0.15, "grad_norm": 15.33728313446045, "learning_rate": 1.9290066650082744e-05, "loss": 1.8431, "step": 914 }, { "epoch": 0.15, "grad_norm": 11.229108810424805, "learning_rate": 1.9288135781992914e-05, "loss": 1.8555, "step": 915 }, { "epoch": 0.15, "grad_norm": 9.554210662841797, "learning_rate": 1.9286202388596047e-05, "loss": 1.966, "step": 916 }, { "epoch": 0.15, "grad_norm": 10.996556282043457, "learning_rate": 1.9284266470417807e-05, "loss": 1.9366, "step": 917 }, { "epoch": 0.15, "grad_norm": 11.67075252532959, "learning_rate": 1.928232802798454e-05, "loss": 1.8311, "step": 918 }, { "epoch": 0.15, "grad_norm": 12.488362312316895, "learning_rate": 1.9280387061823282e-05, "loss": 1.8581, "step": 919 }, { "epoch": 0.15, "grad_norm": 8.032261848449707, "learning_rate": 1.9278443572461748e-05, "loss": 1.8219, "step": 920 }, { "epoch": 0.15, "grad_norm": 9.013344764709473, "learning_rate": 1.9276497560428348e-05, "loss": 1.8478, "step": 921 }, { "epoch": 0.15, "grad_norm": 17.04233169555664, "learning_rate": 1.927454902625217e-05, "loss": 1.9776, "step": 922 }, { "epoch": 0.15, "grad_norm": 12.850141525268555, "learning_rate": 1.9272597970463e-05, "loss": 1.9369, "step": 923 }, { "epoch": 0.15, "grad_norm": 7.383266448974609, "learning_rate": 1.9270644393591293e-05, "loss": 1.6743, "step": 924 }, { "epoch": 0.15, "grad_norm": 12.233861923217773, "learning_rate": 1.92686882961682e-05, "loss": 1.9948, "step": 925 }, { "epoch": 0.15, "grad_norm": 14.770657539367676, "learning_rate": 1.9266729678725555e-05, "loss": 1.943, "step": 926 }, { "epoch": 0.15, "grad_norm": 12.013991355895996, "learning_rate": 1.9264768541795878e-05, "loss": 1.938, "step": 927 }, { "epoch": 0.15, "grad_norm": 11.066890716552734, "learning_rate": 1.9262804885912375e-05, "loss": 1.8116, "step": 928 }, { "epoch": 0.15, "grad_norm": 20.628267288208008, "learning_rate": 1.926083871160893e-05, "loss": 1.8343, "step": 929 }, { "epoch": 0.15, "grad_norm": 7.630223274230957, "learning_rate": 1.9258870019420125e-05, "loss": 1.9672, "step": 930 }, { "epoch": 0.15, "grad_norm": 8.111309051513672, "learning_rate": 1.925689880988121e-05, "loss": 1.7943, "step": 931 }, { "epoch": 0.15, "grad_norm": 11.538825035095215, "learning_rate": 1.925492508352813e-05, "loss": 1.8896, "step": 932 }, { "epoch": 0.15, "grad_norm": 8.944169044494629, "learning_rate": 1.9252948840897513e-05, "loss": 1.7187, "step": 933 }, { "epoch": 0.15, "grad_norm": 8.548186302185059, "learning_rate": 1.925097008252667e-05, "loss": 1.7947, "step": 934 }, { "epoch": 0.15, "grad_norm": 7.543152809143066, "learning_rate": 1.92489888089536e-05, "loss": 1.7314, "step": 935 }, { "epoch": 0.15, "grad_norm": 8.9002046585083, "learning_rate": 1.9247005020716977e-05, "loss": 1.7818, "step": 936 }, { "epoch": 0.15, "grad_norm": 8.872665405273438, "learning_rate": 1.924501871835616e-05, "loss": 1.7465, "step": 937 }, { "epoch": 0.15, "grad_norm": 9.695121765136719, "learning_rate": 1.9243029902411207e-05, "loss": 1.722, "step": 938 }, { "epoch": 0.15, "grad_norm": 12.94925308227539, "learning_rate": 1.9241038573422838e-05, "loss": 1.8694, "step": 939 }, { "epoch": 0.15, "grad_norm": 7.712648868560791, "learning_rate": 1.923904473193247e-05, "loss": 1.8439, "step": 940 }, { "epoch": 0.15, "grad_norm": 9.096835136413574, "learning_rate": 1.92370483784822e-05, "loss": 1.9539, "step": 941 }, { "epoch": 0.15, "grad_norm": 11.519232749938965, "learning_rate": 1.9235049513614804e-05, "loss": 2.019, "step": 942 }, { "epoch": 0.15, "grad_norm": 11.555889129638672, "learning_rate": 1.9233048137873743e-05, "loss": 1.9432, "step": 943 }, { "epoch": 0.15, "grad_norm": 7.885487079620361, "learning_rate": 1.9231044251803162e-05, "loss": 1.7464, "step": 944 }, { "epoch": 0.15, "grad_norm": 7.264519691467285, "learning_rate": 1.9229037855947897e-05, "loss": 1.834, "step": 945 }, { "epoch": 0.15, "grad_norm": 6.7612762451171875, "learning_rate": 1.9227028950853444e-05, "loss": 1.9335, "step": 946 }, { "epoch": 0.15, "grad_norm": 10.236712455749512, "learning_rate": 1.9225017537066003e-05, "loss": 1.9964, "step": 947 }, { "epoch": 0.15, "grad_norm": 9.016097068786621, "learning_rate": 1.9223003615132443e-05, "loss": 1.7499, "step": 948 }, { "epoch": 0.15, "grad_norm": 8.155854225158691, "learning_rate": 1.9220987185600327e-05, "loss": 1.8959, "step": 949 }, { "epoch": 0.15, "grad_norm": 9.075858116149902, "learning_rate": 1.9218968249017884e-05, "loss": 2.0517, "step": 950 }, { "epoch": 0.15, "grad_norm": 8.00126838684082, "learning_rate": 1.921694680593404e-05, "loss": 1.723, "step": 951 }, { "epoch": 0.15, "grad_norm": 9.530396461486816, "learning_rate": 1.921492285689839e-05, "loss": 1.9925, "step": 952 }, { "epoch": 0.15, "grad_norm": 8.148945808410645, "learning_rate": 1.9212896402461213e-05, "loss": 1.9308, "step": 953 }, { "epoch": 0.15, "grad_norm": 7.553034782409668, "learning_rate": 1.9210867443173485e-05, "loss": 2.0574, "step": 954 }, { "epoch": 0.15, "grad_norm": 8.0021390914917, "learning_rate": 1.9208835979586838e-05, "loss": 1.778, "step": 955 }, { "epoch": 0.15, "grad_norm": 8.1503324508667, "learning_rate": 1.92068020122536e-05, "loss": 1.8011, "step": 956 }, { "epoch": 0.15, "grad_norm": 11.299881935119629, "learning_rate": 1.9204765541726783e-05, "loss": 1.857, "step": 957 }, { "epoch": 0.15, "grad_norm": 7.939301490783691, "learning_rate": 1.9202726568560064e-05, "loss": 1.879, "step": 958 }, { "epoch": 0.15, "grad_norm": 9.01579761505127, "learning_rate": 1.9200685093307814e-05, "loss": 2.0496, "step": 959 }, { "epoch": 0.15, "grad_norm": 7.657436370849609, "learning_rate": 1.9198641116525084e-05, "loss": 1.865, "step": 960 }, { "epoch": 0.15, "grad_norm": 7.477259159088135, "learning_rate": 1.9196594638767592e-05, "loss": 2.0299, "step": 961 }, { "epoch": 0.15, "grad_norm": 8.370911598205566, "learning_rate": 1.9194545660591753e-05, "loss": 1.7485, "step": 962 }, { "epoch": 0.16, "grad_norm": 8.375798225402832, "learning_rate": 1.9192494182554648e-05, "loss": 1.7435, "step": 963 }, { "epoch": 0.16, "grad_norm": 8.511798858642578, "learning_rate": 1.9190440205214048e-05, "loss": 1.7858, "step": 964 }, { "epoch": 0.16, "grad_norm": 9.367132186889648, "learning_rate": 1.9188383729128398e-05, "loss": 1.8288, "step": 965 }, { "epoch": 0.16, "grad_norm": 9.509806632995605, "learning_rate": 1.918632475485682e-05, "loss": 1.8999, "step": 966 }, { "epoch": 0.16, "grad_norm": 7.653421878814697, "learning_rate": 1.9184263282959126e-05, "loss": 1.9967, "step": 967 }, { "epoch": 0.16, "grad_norm": 10.807646751403809, "learning_rate": 1.9182199313995796e-05, "loss": 1.8075, "step": 968 }, { "epoch": 0.16, "grad_norm": 9.801875114440918, "learning_rate": 1.918013284852799e-05, "loss": 1.8167, "step": 969 }, { "epoch": 0.16, "grad_norm": 8.339674949645996, "learning_rate": 1.9178063887117552e-05, "loss": 2.0063, "step": 970 }, { "epoch": 0.16, "grad_norm": 9.029494285583496, "learning_rate": 1.9175992430327004e-05, "loss": 1.7797, "step": 971 }, { "epoch": 0.16, "grad_norm": 8.978803634643555, "learning_rate": 1.917391847871954e-05, "loss": 1.8509, "step": 972 }, { "epoch": 0.16, "grad_norm": 7.4726881980896, "learning_rate": 1.917184203285904e-05, "loss": 1.8126, "step": 973 }, { "epoch": 0.16, "grad_norm": 7.745903491973877, "learning_rate": 1.916976309331006e-05, "loss": 1.9996, "step": 974 }, { "epoch": 0.16, "grad_norm": 8.463911056518555, "learning_rate": 1.9167681660637832e-05, "loss": 1.7141, "step": 975 }, { "epoch": 0.16, "grad_norm": 7.884868621826172, "learning_rate": 1.9165597735408262e-05, "loss": 1.9027, "step": 976 }, { "epoch": 0.16, "grad_norm": 8.492218971252441, "learning_rate": 1.9163511318187944e-05, "loss": 1.8395, "step": 977 }, { "epoch": 0.16, "grad_norm": 7.654863357543945, "learning_rate": 1.9161422409544143e-05, "loss": 2.0243, "step": 978 }, { "epoch": 0.16, "grad_norm": 9.279779434204102, "learning_rate": 1.9159331010044804e-05, "loss": 2.0064, "step": 979 }, { "epoch": 0.16, "grad_norm": 7.860520362854004, "learning_rate": 1.9157237120258544e-05, "loss": 1.834, "step": 980 }, { "epoch": 0.16, "grad_norm": 6.833890438079834, "learning_rate": 1.915514074075466e-05, "loss": 1.9903, "step": 981 }, { "epoch": 0.16, "grad_norm": 7.247964382171631, "learning_rate": 1.915304187210313e-05, "loss": 1.899, "step": 982 }, { "epoch": 0.16, "grad_norm": 8.652059555053711, "learning_rate": 1.915094051487461e-05, "loss": 2.0296, "step": 983 }, { "epoch": 0.16, "grad_norm": 7.0891947746276855, "learning_rate": 1.9148836669640416e-05, "loss": 2.0028, "step": 984 }, { "epoch": 0.16, "grad_norm": 7.409886360168457, "learning_rate": 1.9146730336972562e-05, "loss": 1.7571, "step": 985 }, { "epoch": 0.16, "grad_norm": 8.45512866973877, "learning_rate": 1.9144621517443727e-05, "loss": 1.7954, "step": 986 }, { "epoch": 0.16, "grad_norm": 7.808953285217285, "learning_rate": 1.9142510211627264e-05, "loss": 1.7739, "step": 987 }, { "epoch": 0.16, "grad_norm": 7.422319412231445, "learning_rate": 1.9140396420097208e-05, "loss": 1.9692, "step": 988 }, { "epoch": 0.16, "grad_norm": 8.728447914123535, "learning_rate": 1.913828014342827e-05, "loss": 1.9991, "step": 989 }, { "epoch": 0.16, "grad_norm": 7.644820213317871, "learning_rate": 1.9136161382195833e-05, "loss": 1.5299, "step": 990 }, { "epoch": 0.16, "grad_norm": 8.437708854675293, "learning_rate": 1.9134040136975954e-05, "loss": 1.8121, "step": 991 }, { "epoch": 0.16, "grad_norm": 11.1105375289917, "learning_rate": 1.9131916408345374e-05, "loss": 1.9201, "step": 992 }, { "epoch": 0.16, "grad_norm": 8.781968116760254, "learning_rate": 1.91297901968815e-05, "loss": 1.9352, "step": 993 }, { "epoch": 0.16, "grad_norm": 7.180488586425781, "learning_rate": 1.9127661503162416e-05, "loss": 1.8484, "step": 994 }, { "epoch": 0.16, "grad_norm": 9.944459915161133, "learning_rate": 1.9125530327766887e-05, "loss": 1.9639, "step": 995 }, { "epoch": 0.16, "grad_norm": 10.401421546936035, "learning_rate": 1.9123396671274342e-05, "loss": 1.9607, "step": 996 }, { "epoch": 0.16, "grad_norm": 9.415070533752441, "learning_rate": 1.9121260534264893e-05, "loss": 2.0139, "step": 997 }, { "epoch": 0.16, "grad_norm": 8.400236129760742, "learning_rate": 1.9119121917319327e-05, "loss": 1.6943, "step": 998 }, { "epoch": 0.16, "grad_norm": 10.461196899414062, "learning_rate": 1.91169808210191e-05, "loss": 1.8591, "step": 999 }, { "epoch": 0.16, "grad_norm": 12.403242111206055, "learning_rate": 1.911483724594635e-05, "loss": 1.7677, "step": 1000 }, { "epoch": 0.16, "grad_norm": 7.1081671714782715, "learning_rate": 1.911269119268387e-05, "loss": 1.7545, "step": 1001 }, { "epoch": 0.16, "grad_norm": 9.161412239074707, "learning_rate": 1.9110542661815157e-05, "loss": 1.6596, "step": 1002 }, { "epoch": 0.16, "grad_norm": 11.382349014282227, "learning_rate": 1.9108391653924353e-05, "loss": 1.9862, "step": 1003 }, { "epoch": 0.16, "grad_norm": 8.625030517578125, "learning_rate": 1.910623816959629e-05, "loss": 1.8579, "step": 1004 }, { "epoch": 0.16, "grad_norm": 7.800964832305908, "learning_rate": 1.9104082209416465e-05, "loss": 2.0283, "step": 1005 }, { "epoch": 0.16, "grad_norm": 7.537708282470703, "learning_rate": 1.9101923773971057e-05, "loss": 1.8161, "step": 1006 }, { "epoch": 0.16, "grad_norm": 7.248372554779053, "learning_rate": 1.9099762863846906e-05, "loss": 1.8716, "step": 1007 }, { "epoch": 0.16, "grad_norm": 8.14974594116211, "learning_rate": 1.9097599479631538e-05, "loss": 1.8797, "step": 1008 }, { "epoch": 0.16, "grad_norm": 8.1490478515625, "learning_rate": 1.9095433621913142e-05, "loss": 1.6707, "step": 1009 }, { "epoch": 0.16, "grad_norm": 9.071755409240723, "learning_rate": 1.9093265291280583e-05, "loss": 1.7, "step": 1010 }, { "epoch": 0.16, "grad_norm": 9.805309295654297, "learning_rate": 1.9091094488323403e-05, "loss": 1.6821, "step": 1011 }, { "epoch": 0.16, "grad_norm": 8.26132583618164, "learning_rate": 1.9088921213631803e-05, "loss": 1.8798, "step": 1012 }, { "epoch": 0.16, "grad_norm": 9.329475402832031, "learning_rate": 1.9086745467796667e-05, "loss": 1.9003, "step": 1013 }, { "epoch": 0.16, "grad_norm": 7.833208084106445, "learning_rate": 1.908456725140955e-05, "loss": 2.0601, "step": 1014 }, { "epoch": 0.16, "grad_norm": 8.489057540893555, "learning_rate": 1.908238656506268e-05, "loss": 1.8781, "step": 1015 }, { "epoch": 0.16, "grad_norm": 7.780120372772217, "learning_rate": 1.9080203409348945e-05, "loss": 1.8607, "step": 1016 }, { "epoch": 0.16, "grad_norm": 7.153567790985107, "learning_rate": 1.907801778486192e-05, "loss": 1.9844, "step": 1017 }, { "epoch": 0.16, "grad_norm": 8.437777519226074, "learning_rate": 1.907582969219584e-05, "loss": 1.6905, "step": 1018 }, { "epoch": 0.16, "grad_norm": 8.4360933303833, "learning_rate": 1.907363913194562e-05, "loss": 1.7258, "step": 1019 }, { "epoch": 0.16, "grad_norm": 8.72169303894043, "learning_rate": 1.9071446104706836e-05, "loss": 1.8002, "step": 1020 }, { "epoch": 0.16, "grad_norm": 8.218103408813477, "learning_rate": 1.906925061107574e-05, "loss": 1.8154, "step": 1021 }, { "epoch": 0.16, "grad_norm": 8.457624435424805, "learning_rate": 1.9067052651649254e-05, "loss": 1.8574, "step": 1022 }, { "epoch": 0.16, "grad_norm": 8.611739158630371, "learning_rate": 1.9064852227024975e-05, "loss": 1.7173, "step": 1023 }, { "epoch": 0.16, "grad_norm": 10.178143501281738, "learning_rate": 1.906264933780116e-05, "loss": 2.087, "step": 1024 }, { "epoch": 0.17, "grad_norm": 8.422913551330566, "learning_rate": 1.906044398457675e-05, "loss": 1.8631, "step": 1025 }, { "epoch": 0.17, "grad_norm": 9.495409965515137, "learning_rate": 1.9058236167951345e-05, "loss": 1.9131, "step": 1026 }, { "epoch": 0.17, "grad_norm": 12.798405647277832, "learning_rate": 1.9056025888525212e-05, "loss": 1.6164, "step": 1027 }, { "epoch": 0.17, "grad_norm": 10.323179244995117, "learning_rate": 1.90538131468993e-05, "loss": 1.7338, "step": 1028 }, { "epoch": 0.17, "grad_norm": 7.91552734375, "learning_rate": 1.9051597943675216e-05, "loss": 1.9481, "step": 1029 }, { "epoch": 0.17, "grad_norm": 12.031129837036133, "learning_rate": 1.9049380279455244e-05, "loss": 2.2015, "step": 1030 }, { "epoch": 0.17, "grad_norm": 16.167816162109375, "learning_rate": 1.9047160154842336e-05, "loss": 1.9502, "step": 1031 }, { "epoch": 0.17, "grad_norm": 8.645329475402832, "learning_rate": 1.9044937570440107e-05, "loss": 1.7998, "step": 1032 }, { "epoch": 0.17, "grad_norm": 8.740265846252441, "learning_rate": 1.9042712526852846e-05, "loss": 1.7387, "step": 1033 }, { "epoch": 0.17, "grad_norm": 13.905311584472656, "learning_rate": 1.904048502468551e-05, "loss": 1.6309, "step": 1034 }, { "epoch": 0.17, "grad_norm": 9.577743530273438, "learning_rate": 1.9038255064543725e-05, "loss": 1.9253, "step": 1035 }, { "epoch": 0.17, "grad_norm": 6.800482749938965, "learning_rate": 1.9036022647033783e-05, "loss": 1.9217, "step": 1036 }, { "epoch": 0.17, "grad_norm": 11.06932258605957, "learning_rate": 1.9033787772762647e-05, "loss": 1.8254, "step": 1037 }, { "epoch": 0.17, "grad_norm": 7.324239730834961, "learning_rate": 1.903155044233794e-05, "loss": 1.7942, "step": 1038 }, { "epoch": 0.17, "grad_norm": 9.583963394165039, "learning_rate": 1.902931065636797e-05, "loss": 1.8617, "step": 1039 }, { "epoch": 0.17, "grad_norm": 8.366186141967773, "learning_rate": 1.9027068415461697e-05, "loss": 1.9665, "step": 1040 }, { "epoch": 0.17, "grad_norm": 9.17573356628418, "learning_rate": 1.9024823720228748e-05, "loss": 1.8136, "step": 1041 }, { "epoch": 0.17, "grad_norm": 8.732978820800781, "learning_rate": 1.9022576571279424e-05, "loss": 1.7224, "step": 1042 }, { "epoch": 0.17, "grad_norm": 7.856827735900879, "learning_rate": 1.90203269692247e-05, "loss": 1.7386, "step": 1043 }, { "epoch": 0.17, "grad_norm": 7.926042556762695, "learning_rate": 1.9018074914676205e-05, "loss": 1.8014, "step": 1044 }, { "epoch": 0.17, "grad_norm": 8.464417457580566, "learning_rate": 1.901582040824623e-05, "loss": 1.8903, "step": 1045 }, { "epoch": 0.17, "grad_norm": 7.787516117095947, "learning_rate": 1.9013563450547757e-05, "loss": 1.6843, "step": 1046 }, { "epoch": 0.17, "grad_norm": 9.30486011505127, "learning_rate": 1.901130404219441e-05, "loss": 1.7068, "step": 1047 }, { "epoch": 0.17, "grad_norm": 7.992535591125488, "learning_rate": 1.9009042183800493e-05, "loss": 1.8785, "step": 1048 }, { "epoch": 0.17, "grad_norm": 7.476897716522217, "learning_rate": 1.9006777875980972e-05, "loss": 1.8653, "step": 1049 }, { "epoch": 0.17, "grad_norm": 8.392167091369629, "learning_rate": 1.9004511119351473e-05, "loss": 1.9077, "step": 1050 }, { "epoch": 0.17, "grad_norm": 8.793010711669922, "learning_rate": 1.90022419145283e-05, "loss": 1.9748, "step": 1051 }, { "epoch": 0.17, "grad_norm": 10.93588638305664, "learning_rate": 1.8999970262128415e-05, "loss": 1.8988, "step": 1052 }, { "epoch": 0.17, "grad_norm": 11.520087242126465, "learning_rate": 1.8997696162769448e-05, "loss": 1.9394, "step": 1053 }, { "epoch": 0.17, "grad_norm": 10.096417427062988, "learning_rate": 1.8995419617069685e-05, "loss": 1.9874, "step": 1054 }, { "epoch": 0.17, "grad_norm": 14.494647979736328, "learning_rate": 1.89931406256481e-05, "loss": 1.9498, "step": 1055 }, { "epoch": 0.17, "grad_norm": 14.42122745513916, "learning_rate": 1.8990859189124303e-05, "loss": 1.7591, "step": 1056 }, { "epoch": 0.17, "grad_norm": 7.745213031768799, "learning_rate": 1.8988575308118587e-05, "loss": 2.023, "step": 1057 }, { "epoch": 0.17, "grad_norm": 7.39893913269043, "learning_rate": 1.8986288983251907e-05, "loss": 1.9068, "step": 1058 }, { "epoch": 0.17, "grad_norm": 7.781610012054443, "learning_rate": 1.8984000215145884e-05, "loss": 2.0081, "step": 1059 }, { "epoch": 0.17, "grad_norm": 9.01050090789795, "learning_rate": 1.8981709004422796e-05, "loss": 1.7957, "step": 1060 }, { "epoch": 0.17, "grad_norm": 11.027290344238281, "learning_rate": 1.897941535170559e-05, "loss": 1.8908, "step": 1061 }, { "epoch": 0.17, "grad_norm": 9.408148765563965, "learning_rate": 1.8977119257617878e-05, "loss": 1.8547, "step": 1062 }, { "epoch": 0.17, "grad_norm": 8.88369369506836, "learning_rate": 1.8974820722783932e-05, "loss": 1.917, "step": 1063 }, { "epoch": 0.17, "grad_norm": 11.365843772888184, "learning_rate": 1.897251974782869e-05, "loss": 1.7886, "step": 1064 }, { "epoch": 0.17, "grad_norm": 10.05237865447998, "learning_rate": 1.8970216333377752e-05, "loss": 2.0122, "step": 1065 }, { "epoch": 0.17, "grad_norm": 8.539021492004395, "learning_rate": 1.896791048005739e-05, "loss": 1.8387, "step": 1066 }, { "epoch": 0.17, "grad_norm": 7.872187614440918, "learning_rate": 1.896560218849452e-05, "loss": 1.8878, "step": 1067 }, { "epoch": 0.17, "grad_norm": 9.908541679382324, "learning_rate": 1.8963291459316742e-05, "loss": 1.8094, "step": 1068 }, { "epoch": 0.17, "grad_norm": 9.685304641723633, "learning_rate": 1.8960978293152303e-05, "loss": 2.0076, "step": 1069 }, { "epoch": 0.17, "grad_norm": 8.328533172607422, "learning_rate": 1.8958662690630124e-05, "loss": 2.1351, "step": 1070 }, { "epoch": 0.17, "grad_norm": 8.220758438110352, "learning_rate": 1.8956344652379776e-05, "loss": 1.8427, "step": 1071 }, { "epoch": 0.17, "grad_norm": 8.899036407470703, "learning_rate": 1.8954024179031504e-05, "loss": 1.9207, "step": 1072 }, { "epoch": 0.17, "grad_norm": 12.22178840637207, "learning_rate": 1.8951701271216212e-05, "loss": 1.8378, "step": 1073 }, { "epoch": 0.17, "grad_norm": 10.279688835144043, "learning_rate": 1.8949375929565463e-05, "loss": 1.7299, "step": 1074 }, { "epoch": 0.17, "grad_norm": 11.1865816116333, "learning_rate": 1.8947048154711477e-05, "loss": 1.8443, "step": 1075 }, { "epoch": 0.17, "grad_norm": 7.291100978851318, "learning_rate": 1.8944717947287152e-05, "loss": 1.8962, "step": 1076 }, { "epoch": 0.17, "grad_norm": 7.68377161026001, "learning_rate": 1.894238530792603e-05, "loss": 1.6583, "step": 1077 }, { "epoch": 0.17, "grad_norm": 8.011282920837402, "learning_rate": 1.8940050237262323e-05, "loss": 1.6925, "step": 1078 }, { "epoch": 0.17, "grad_norm": 11.067156791687012, "learning_rate": 1.8937712735930908e-05, "loss": 1.7611, "step": 1079 }, { "epoch": 0.17, "grad_norm": 7.954333782196045, "learning_rate": 1.89353728045673e-05, "loss": 1.8559, "step": 1080 }, { "epoch": 0.17, "grad_norm": 9.231598854064941, "learning_rate": 1.8933030443807716e-05, "loss": 1.9683, "step": 1081 }, { "epoch": 0.17, "grad_norm": 8.363301277160645, "learning_rate": 1.8930685654288988e-05, "loss": 1.9443, "step": 1082 }, { "epoch": 0.17, "grad_norm": 9.549668312072754, "learning_rate": 1.8928338436648642e-05, "loss": 1.8712, "step": 1083 }, { "epoch": 0.17, "grad_norm": 9.45719051361084, "learning_rate": 1.8925988791524847e-05, "loss": 1.821, "step": 1084 }, { "epoch": 0.17, "grad_norm": 8.852667808532715, "learning_rate": 1.8923636719556438e-05, "loss": 1.9172, "step": 1085 }, { "epoch": 0.17, "grad_norm": 10.27222728729248, "learning_rate": 1.892128222138291e-05, "loss": 1.8383, "step": 1086 }, { "epoch": 0.17, "grad_norm": 7.24739408493042, "learning_rate": 1.8918925297644418e-05, "loss": 1.8822, "step": 1087 }, { "epoch": 0.18, "grad_norm": 8.522356033325195, "learning_rate": 1.8916565948981764e-05, "loss": 1.8317, "step": 1088 }, { "epoch": 0.18, "grad_norm": 8.057518005371094, "learning_rate": 1.8914204176036435e-05, "loss": 1.8399, "step": 1089 }, { "epoch": 0.18, "grad_norm": 8.382121086120605, "learning_rate": 1.891183997945055e-05, "loss": 1.7933, "step": 1090 }, { "epoch": 0.18, "grad_norm": 8.079960823059082, "learning_rate": 1.8909473359866906e-05, "loss": 1.811, "step": 1091 }, { "epoch": 0.18, "grad_norm": 8.977620124816895, "learning_rate": 1.890710431792895e-05, "loss": 1.8075, "step": 1092 }, { "epoch": 0.18, "grad_norm": 8.5299711227417, "learning_rate": 1.890473285428079e-05, "loss": 1.9176, "step": 1093 }, { "epoch": 0.18, "grad_norm": 7.757791042327881, "learning_rate": 1.8902358969567195e-05, "loss": 1.6677, "step": 1094 }, { "epoch": 0.18, "grad_norm": 12.493413925170898, "learning_rate": 1.8899982664433582e-05, "loss": 1.8365, "step": 1095 }, { "epoch": 0.18, "grad_norm": 12.79686164855957, "learning_rate": 1.8897603939526033e-05, "loss": 1.7759, "step": 1096 }, { "epoch": 0.18, "grad_norm": 7.438371658325195, "learning_rate": 1.8895222795491303e-05, "loss": 1.935, "step": 1097 }, { "epoch": 0.18, "grad_norm": 9.07620620727539, "learning_rate": 1.889283923297677e-05, "loss": 1.9255, "step": 1098 }, { "epoch": 0.18, "grad_norm": 16.688980102539062, "learning_rate": 1.8890453252630503e-05, "loss": 1.8514, "step": 1099 }, { "epoch": 0.18, "grad_norm": 11.872942924499512, "learning_rate": 1.8888064855101208e-05, "loss": 1.8059, "step": 1100 }, { "epoch": 0.18, "grad_norm": 7.55921745300293, "learning_rate": 1.888567404103826e-05, "loss": 1.5566, "step": 1101 }, { "epoch": 0.18, "grad_norm": 11.290399551391602, "learning_rate": 1.8883280811091684e-05, "loss": 1.9089, "step": 1102 }, { "epoch": 0.18, "grad_norm": 13.526169776916504, "learning_rate": 1.888088516591216e-05, "loss": 1.806, "step": 1103 }, { "epoch": 0.18, "grad_norm": 8.592461585998535, "learning_rate": 1.8878487106151036e-05, "loss": 2.0735, "step": 1104 }, { "epoch": 0.18, "grad_norm": 7.831098556518555, "learning_rate": 1.8876086632460303e-05, "loss": 1.8737, "step": 1105 }, { "epoch": 0.18, "grad_norm": 7.897984981536865, "learning_rate": 1.8873683745492618e-05, "loss": 1.8092, "step": 1106 }, { "epoch": 0.18, "grad_norm": 9.711540222167969, "learning_rate": 1.887127844590129e-05, "loss": 1.8741, "step": 1107 }, { "epoch": 0.18, "grad_norm": 7.459097385406494, "learning_rate": 1.8868870734340282e-05, "loss": 1.6816, "step": 1108 }, { "epoch": 0.18, "grad_norm": 8.077688217163086, "learning_rate": 1.8866460611464218e-05, "loss": 1.7472, "step": 1109 }, { "epoch": 0.18, "grad_norm": 9.420668601989746, "learning_rate": 1.8864048077928372e-05, "loss": 1.7468, "step": 1110 }, { "epoch": 0.18, "grad_norm": 7.738216400146484, "learning_rate": 1.8861633134388678e-05, "loss": 1.8612, "step": 1111 }, { "epoch": 0.18, "grad_norm": 10.091425895690918, "learning_rate": 1.8859215781501727e-05, "loss": 2.0403, "step": 1112 }, { "epoch": 0.18, "grad_norm": 9.071244239807129, "learning_rate": 1.8856796019924754e-05, "loss": 1.7337, "step": 1113 }, { "epoch": 0.18, "grad_norm": 8.75747013092041, "learning_rate": 1.885437385031566e-05, "loss": 1.8034, "step": 1114 }, { "epoch": 0.18, "grad_norm": 7.45496129989624, "learning_rate": 1.8851949273332998e-05, "loss": 1.8929, "step": 1115 }, { "epoch": 0.18, "grad_norm": 7.868702411651611, "learning_rate": 1.8849522289635974e-05, "loss": 1.8811, "step": 1116 }, { "epoch": 0.18, "grad_norm": 7.750369548797607, "learning_rate": 1.8847092899884448e-05, "loss": 1.8574, "step": 1117 }, { "epoch": 0.18, "grad_norm": 8.459769248962402, "learning_rate": 1.884466110473894e-05, "loss": 1.9304, "step": 1118 }, { "epoch": 0.18, "grad_norm": 9.177490234375, "learning_rate": 1.8842226904860617e-05, "loss": 1.662, "step": 1119 }, { "epoch": 0.18, "grad_norm": 12.100234031677246, "learning_rate": 1.88397903009113e-05, "loss": 1.893, "step": 1120 }, { "epoch": 0.18, "grad_norm": 7.937577247619629, "learning_rate": 1.8837351293553464e-05, "loss": 1.7596, "step": 1121 }, { "epoch": 0.18, "grad_norm": 8.785045623779297, "learning_rate": 1.883490988345024e-05, "loss": 1.8397, "step": 1122 }, { "epoch": 0.18, "grad_norm": 8.651057243347168, "learning_rate": 1.8832466071265415e-05, "loss": 1.9863, "step": 1123 }, { "epoch": 0.18, "grad_norm": 10.987274169921875, "learning_rate": 1.8830019857663427e-05, "loss": 1.8625, "step": 1124 }, { "epoch": 0.18, "grad_norm": 10.665714263916016, "learning_rate": 1.8827571243309357e-05, "loss": 2.0116, "step": 1125 }, { "epoch": 0.18, "grad_norm": 7.789214611053467, "learning_rate": 1.8825120228868955e-05, "loss": 1.6571, "step": 1126 }, { "epoch": 0.18, "grad_norm": 17.24508285522461, "learning_rate": 1.882266681500861e-05, "loss": 1.9, "step": 1127 }, { "epoch": 0.18, "grad_norm": 20.04245948791504, "learning_rate": 1.8820211002395372e-05, "loss": 1.993, "step": 1128 }, { "epoch": 0.18, "grad_norm": 7.9108357429504395, "learning_rate": 1.881775279169694e-05, "loss": 1.9831, "step": 1129 }, { "epoch": 0.18, "grad_norm": 9.382389068603516, "learning_rate": 1.8815292183581663e-05, "loss": 1.7935, "step": 1130 }, { "epoch": 0.18, "grad_norm": 15.047473907470703, "learning_rate": 1.8812829178718548e-05, "loss": 1.9884, "step": 1131 }, { "epoch": 0.18, "grad_norm": 14.686062812805176, "learning_rate": 1.8810363777777247e-05, "loss": 2.0001, "step": 1132 }, { "epoch": 0.18, "grad_norm": 7.411862373352051, "learning_rate": 1.880789598142807e-05, "loss": 1.6611, "step": 1133 }, { "epoch": 0.18, "grad_norm": 9.605759620666504, "learning_rate": 1.8805425790341967e-05, "loss": 1.6624, "step": 1134 }, { "epoch": 0.18, "grad_norm": 12.850325584411621, "learning_rate": 1.8802953205190554e-05, "loss": 1.7763, "step": 1135 }, { "epoch": 0.18, "grad_norm": 10.299335479736328, "learning_rate": 1.8800478226646085e-05, "loss": 1.9257, "step": 1136 }, { "epoch": 0.18, "grad_norm": 7.585279941558838, "learning_rate": 1.8798000855381472e-05, "loss": 1.6464, "step": 1137 }, { "epoch": 0.18, "grad_norm": 7.9145588874816895, "learning_rate": 1.8795521092070273e-05, "loss": 1.6355, "step": 1138 }, { "epoch": 0.18, "grad_norm": 8.303479194641113, "learning_rate": 1.8793038937386706e-05, "loss": 1.6809, "step": 1139 }, { "epoch": 0.18, "grad_norm": 8.987263679504395, "learning_rate": 1.879055439200563e-05, "loss": 1.7165, "step": 1140 }, { "epoch": 0.18, "grad_norm": 8.199256896972656, "learning_rate": 1.878806745660255e-05, "loss": 1.7855, "step": 1141 }, { "epoch": 0.18, "grad_norm": 7.6216230392456055, "learning_rate": 1.8785578131853636e-05, "loss": 1.9038, "step": 1142 }, { "epoch": 0.18, "grad_norm": 8.966546058654785, "learning_rate": 1.878308641843569e-05, "loss": 1.8324, "step": 1143 }, { "epoch": 0.18, "grad_norm": 7.387916564941406, "learning_rate": 1.8780592317026176e-05, "loss": 1.7275, "step": 1144 }, { "epoch": 0.18, "grad_norm": 8.810188293457031, "learning_rate": 1.8778095828303208e-05, "loss": 1.8202, "step": 1145 }, { "epoch": 0.18, "grad_norm": 7.425343990325928, "learning_rate": 1.8775596952945533e-05, "loss": 1.9572, "step": 1146 }, { "epoch": 0.18, "grad_norm": 7.430014133453369, "learning_rate": 1.877309569163257e-05, "loss": 1.9022, "step": 1147 }, { "epoch": 0.18, "grad_norm": 7.226318836212158, "learning_rate": 1.8770592045044373e-05, "loss": 1.8055, "step": 1148 }, { "epoch": 0.18, "grad_norm": 8.73915958404541, "learning_rate": 1.8768086013861636e-05, "loss": 1.8731, "step": 1149 }, { "epoch": 0.19, "grad_norm": 7.326907157897949, "learning_rate": 1.8765577598765726e-05, "loss": 1.7516, "step": 1150 }, { "epoch": 0.19, "grad_norm": 7.86537504196167, "learning_rate": 1.8763066800438638e-05, "loss": 1.6984, "step": 1151 }, { "epoch": 0.19, "grad_norm": 9.028179168701172, "learning_rate": 1.876055361956302e-05, "loss": 1.8161, "step": 1152 }, { "epoch": 0.19, "grad_norm": 7.8999199867248535, "learning_rate": 1.8758038056822168e-05, "loss": 1.7366, "step": 1153 }, { "epoch": 0.19, "grad_norm": 10.201797485351562, "learning_rate": 1.8755520112900027e-05, "loss": 1.6035, "step": 1154 }, { "epoch": 0.19, "grad_norm": 8.998786926269531, "learning_rate": 1.875299978848119e-05, "loss": 2.0588, "step": 1155 }, { "epoch": 0.19, "grad_norm": 7.084962368011475, "learning_rate": 1.87504770842509e-05, "loss": 1.7889, "step": 1156 }, { "epoch": 0.19, "grad_norm": 9.414756774902344, "learning_rate": 1.8747952000895035e-05, "loss": 1.8373, "step": 1157 }, { "epoch": 0.19, "grad_norm": 8.241851806640625, "learning_rate": 1.8745424539100134e-05, "loss": 1.7871, "step": 1158 }, { "epoch": 0.19, "grad_norm": 8.193439483642578, "learning_rate": 1.8742894699553374e-05, "loss": 1.9808, "step": 1159 }, { "epoch": 0.19, "grad_norm": 7.779728889465332, "learning_rate": 1.874036248294258e-05, "loss": 1.9375, "step": 1160 }, { "epoch": 0.19, "grad_norm": 10.160680770874023, "learning_rate": 1.8737827889956226e-05, "loss": 1.7184, "step": 1161 }, { "epoch": 0.19, "grad_norm": 7.1348419189453125, "learning_rate": 1.873529092128343e-05, "loss": 1.7758, "step": 1162 }, { "epoch": 0.19, "grad_norm": 10.168661117553711, "learning_rate": 1.8732751577613956e-05, "loss": 1.772, "step": 1163 }, { "epoch": 0.19, "grad_norm": 8.545838356018066, "learning_rate": 1.8730209859638217e-05, "loss": 1.8077, "step": 1164 }, { "epoch": 0.19, "grad_norm": 7.975241184234619, "learning_rate": 1.8727665768047265e-05, "loss": 1.8409, "step": 1165 }, { "epoch": 0.19, "grad_norm": 11.245660781860352, "learning_rate": 1.8725119303532803e-05, "loss": 2.0519, "step": 1166 }, { "epoch": 0.19, "grad_norm": 7.309485912322998, "learning_rate": 1.8722570466787173e-05, "loss": 2.0645, "step": 1167 }, { "epoch": 0.19, "grad_norm": 7.660451889038086, "learning_rate": 1.8720019258503372e-05, "loss": 1.7688, "step": 1168 }, { "epoch": 0.19, "grad_norm": 8.004311561584473, "learning_rate": 1.8717465679375033e-05, "loss": 1.7155, "step": 1169 }, { "epoch": 0.19, "grad_norm": 8.498406410217285, "learning_rate": 1.8714909730096434e-05, "loss": 1.989, "step": 1170 }, { "epoch": 0.19, "grad_norm": 10.250463485717773, "learning_rate": 1.8712351411362506e-05, "loss": 1.7651, "step": 1171 }, { "epoch": 0.19, "grad_norm": 7.3253655433654785, "learning_rate": 1.870979072386881e-05, "loss": 1.7233, "step": 1172 }, { "epoch": 0.19, "grad_norm": 9.079808235168457, "learning_rate": 1.870722766831157e-05, "loss": 1.7269, "step": 1173 }, { "epoch": 0.19, "grad_norm": 8.811264038085938, "learning_rate": 1.8704662245387634e-05, "loss": 1.6998, "step": 1174 }, { "epoch": 0.19, "grad_norm": 13.090521812438965, "learning_rate": 1.8702094455794502e-05, "loss": 2.0901, "step": 1175 }, { "epoch": 0.19, "grad_norm": 7.3043599128723145, "learning_rate": 1.8699524300230325e-05, "loss": 1.8213, "step": 1176 }, { "epoch": 0.19, "grad_norm": 7.6223554611206055, "learning_rate": 1.8696951779393887e-05, "loss": 1.7845, "step": 1177 }, { "epoch": 0.19, "grad_norm": 8.946743965148926, "learning_rate": 1.8694376893984617e-05, "loss": 1.882, "step": 1178 }, { "epoch": 0.19, "grad_norm": 7.5937299728393555, "learning_rate": 1.869179964470259e-05, "loss": 1.9594, "step": 1179 }, { "epoch": 0.19, "grad_norm": 7.248358726501465, "learning_rate": 1.868922003224852e-05, "loss": 1.7409, "step": 1180 }, { "epoch": 0.19, "grad_norm": 8.747432708740234, "learning_rate": 1.868663805732377e-05, "loss": 1.8346, "step": 1181 }, { "epoch": 0.19, "grad_norm": 8.477384567260742, "learning_rate": 1.868405372063034e-05, "loss": 1.8913, "step": 1182 }, { "epoch": 0.19, "grad_norm": 8.202040672302246, "learning_rate": 1.8681467022870866e-05, "loss": 1.7145, "step": 1183 }, { "epoch": 0.19, "grad_norm": 8.309621810913086, "learning_rate": 1.8678877964748642e-05, "loss": 1.7697, "step": 1184 }, { "epoch": 0.19, "grad_norm": 7.261283874511719, "learning_rate": 1.8676286546967586e-05, "loss": 1.8457, "step": 1185 }, { "epoch": 0.19, "grad_norm": 9.670111656188965, "learning_rate": 1.8673692770232275e-05, "loss": 1.8084, "step": 1186 }, { "epoch": 0.19, "grad_norm": 7.934786796569824, "learning_rate": 1.8671096635247914e-05, "loss": 1.8139, "step": 1187 }, { "epoch": 0.19, "grad_norm": 9.010838508605957, "learning_rate": 1.8668498142720356e-05, "loss": 1.9006, "step": 1188 }, { "epoch": 0.19, "grad_norm": 7.555320739746094, "learning_rate": 1.8665897293356087e-05, "loss": 1.8155, "step": 1189 }, { "epoch": 0.19, "grad_norm": 7.900635242462158, "learning_rate": 1.8663294087862248e-05, "loss": 1.7857, "step": 1190 }, { "epoch": 0.19, "grad_norm": 8.197748184204102, "learning_rate": 1.8660688526946606e-05, "loss": 1.8125, "step": 1191 }, { "epoch": 0.19, "grad_norm": 8.864831924438477, "learning_rate": 1.8658080611317575e-05, "loss": 1.7295, "step": 1192 }, { "epoch": 0.19, "grad_norm": 8.989450454711914, "learning_rate": 1.865547034168421e-05, "loss": 1.8286, "step": 1193 }, { "epoch": 0.19, "grad_norm": 7.081507205963135, "learning_rate": 1.8652857718756205e-05, "loss": 1.9573, "step": 1194 }, { "epoch": 0.19, "grad_norm": 8.260982513427734, "learning_rate": 1.8650242743243898e-05, "loss": 1.8488, "step": 1195 }, { "epoch": 0.19, "grad_norm": 8.395746231079102, "learning_rate": 1.864762541585825e-05, "loss": 1.7763, "step": 1196 }, { "epoch": 0.19, "grad_norm": 8.848099708557129, "learning_rate": 1.864500573731089e-05, "loss": 2.0427, "step": 1197 }, { "epoch": 0.19, "grad_norm": 8.716891288757324, "learning_rate": 1.8642383708314056e-05, "loss": 1.5517, "step": 1198 }, { "epoch": 0.19, "grad_norm": 8.478108406066895, "learning_rate": 1.8639759329580647e-05, "loss": 1.8814, "step": 1199 }, { "epoch": 0.19, "grad_norm": 7.754581451416016, "learning_rate": 1.863713260182419e-05, "loss": 1.7648, "step": 1200 }, { "epoch": 0.19, "grad_norm": 7.283964157104492, "learning_rate": 1.8634503525758853e-05, "loss": 2.0295, "step": 1201 }, { "epoch": 0.19, "grad_norm": 7.582803249359131, "learning_rate": 1.863187210209944e-05, "loss": 1.826, "step": 1202 }, { "epoch": 0.19, "grad_norm": 7.253533840179443, "learning_rate": 1.8629238331561406e-05, "loss": 1.7635, "step": 1203 }, { "epoch": 0.19, "grad_norm": 8.610451698303223, "learning_rate": 1.8626602214860828e-05, "loss": 1.7909, "step": 1204 }, { "epoch": 0.19, "grad_norm": 7.019696235656738, "learning_rate": 1.8623963752714425e-05, "loss": 1.6975, "step": 1205 }, { "epoch": 0.19, "grad_norm": 11.295740127563477, "learning_rate": 1.862132294583956e-05, "loss": 1.7398, "step": 1206 }, { "epoch": 0.19, "grad_norm": 9.834456443786621, "learning_rate": 1.8618679794954225e-05, "loss": 1.7681, "step": 1207 }, { "epoch": 0.19, "grad_norm": 7.193734645843506, "learning_rate": 1.8616034300777062e-05, "loss": 1.8195, "step": 1208 }, { "epoch": 0.19, "grad_norm": 8.787298202514648, "learning_rate": 1.861338646402733e-05, "loss": 1.9744, "step": 1209 }, { "epoch": 0.19, "grad_norm": 10.10808277130127, "learning_rate": 1.861073628542495e-05, "loss": 2.0695, "step": 1210 }, { "epoch": 0.19, "grad_norm": 8.083991050720215, "learning_rate": 1.8608083765690455e-05, "loss": 2.012, "step": 1211 }, { "epoch": 0.2, "grad_norm": 10.210931777954102, "learning_rate": 1.860542890554503e-05, "loss": 1.939, "step": 1212 }, { "epoch": 0.2, "grad_norm": 9.301657676696777, "learning_rate": 1.8602771705710496e-05, "loss": 1.8556, "step": 1213 }, { "epoch": 0.2, "grad_norm": 7.979488372802734, "learning_rate": 1.8600112166909304e-05, "loss": 1.8816, "step": 1214 }, { "epoch": 0.2, "grad_norm": 7.07103157043457, "learning_rate": 1.8597450289864537e-05, "loss": 1.7361, "step": 1215 }, { "epoch": 0.2, "grad_norm": 7.95102071762085, "learning_rate": 1.859478607529993e-05, "loss": 1.84, "step": 1216 }, { "epoch": 0.2, "grad_norm": 8.556746482849121, "learning_rate": 1.859211952393984e-05, "loss": 1.8251, "step": 1217 }, { "epoch": 0.2, "grad_norm": 8.720152854919434, "learning_rate": 1.858945063650926e-05, "loss": 1.9682, "step": 1218 }, { "epoch": 0.2, "grad_norm": 7.720849514007568, "learning_rate": 1.8586779413733828e-05, "loss": 1.6436, "step": 1219 }, { "epoch": 0.2, "grad_norm": 7.061589241027832, "learning_rate": 1.8584105856339797e-05, "loss": 1.8257, "step": 1220 }, { "epoch": 0.2, "grad_norm": 12.165461540222168, "learning_rate": 1.8581429965054084e-05, "loss": 1.9535, "step": 1221 }, { "epoch": 0.2, "grad_norm": 8.315881729125977, "learning_rate": 1.857875174060421e-05, "loss": 1.8075, "step": 1222 }, { "epoch": 0.2, "grad_norm": 9.760275840759277, "learning_rate": 1.8576071183718357e-05, "loss": 1.9215, "step": 1223 }, { "epoch": 0.2, "grad_norm": 6.867757797241211, "learning_rate": 1.857338829512532e-05, "loss": 1.7133, "step": 1224 }, { "epoch": 0.2, "grad_norm": 7.054280757904053, "learning_rate": 1.8570703075554542e-05, "loss": 1.6229, "step": 1225 }, { "epoch": 0.2, "grad_norm": 8.03372573852539, "learning_rate": 1.8568015525736095e-05, "loss": 1.788, "step": 1226 }, { "epoch": 0.2, "grad_norm": 8.707822799682617, "learning_rate": 1.856532564640068e-05, "loss": 1.8362, "step": 1227 }, { "epoch": 0.2, "grad_norm": 10.989639282226562, "learning_rate": 1.8562633438279632e-05, "loss": 2.0062, "step": 1228 }, { "epoch": 0.2, "grad_norm": 7.881641864776611, "learning_rate": 1.855993890210493e-05, "loss": 1.7644, "step": 1229 }, { "epoch": 0.2, "grad_norm": 7.970564365386963, "learning_rate": 1.855724203860918e-05, "loss": 1.8442, "step": 1230 }, { "epoch": 0.2, "grad_norm": 8.521865844726562, "learning_rate": 1.8554542848525614e-05, "loss": 1.872, "step": 1231 }, { "epoch": 0.2, "grad_norm": 8.449475288391113, "learning_rate": 1.8551841332588103e-05, "loss": 1.921, "step": 1232 }, { "epoch": 0.2, "grad_norm": 7.771749496459961, "learning_rate": 1.8549137491531152e-05, "loss": 1.9067, "step": 1233 }, { "epoch": 0.2, "grad_norm": 9.615324974060059, "learning_rate": 1.854643132608989e-05, "loss": 1.7215, "step": 1234 }, { "epoch": 0.2, "grad_norm": 7.533430099487305, "learning_rate": 1.854372283700009e-05, "loss": 1.9169, "step": 1235 }, { "epoch": 0.2, "grad_norm": 7.737631797790527, "learning_rate": 1.8541012024998145e-05, "loss": 2.1487, "step": 1236 }, { "epoch": 0.2, "grad_norm": 9.898319244384766, "learning_rate": 1.853829889082109e-05, "loss": 1.8491, "step": 1237 }, { "epoch": 0.2, "grad_norm": 7.1347808837890625, "learning_rate": 1.8535583435206585e-05, "loss": 1.8283, "step": 1238 }, { "epoch": 0.2, "grad_norm": 8.820145606994629, "learning_rate": 1.8532865658892914e-05, "loss": 1.7278, "step": 1239 }, { "epoch": 0.2, "grad_norm": 9.642117500305176, "learning_rate": 1.853014556261901e-05, "loss": 1.7004, "step": 1240 }, { "epoch": 0.2, "grad_norm": 8.561197280883789, "learning_rate": 1.852742314712442e-05, "loss": 2.0487, "step": 1241 }, { "epoch": 0.2, "grad_norm": 7.9799017906188965, "learning_rate": 1.852469841314934e-05, "loss": 1.6252, "step": 1242 }, { "epoch": 0.2, "grad_norm": 9.894368171691895, "learning_rate": 1.8521971361434573e-05, "loss": 1.971, "step": 1243 }, { "epoch": 0.2, "grad_norm": 11.120004653930664, "learning_rate": 1.8519241992721567e-05, "loss": 1.8323, "step": 1244 }, { "epoch": 0.2, "grad_norm": 7.981195449829102, "learning_rate": 1.85165103077524e-05, "loss": 2.0118, "step": 1245 }, { "epoch": 0.2, "grad_norm": 7.430525779724121, "learning_rate": 1.8513776307269777e-05, "loss": 1.6772, "step": 1246 }, { "epoch": 0.2, "grad_norm": 9.886370658874512, "learning_rate": 1.8511039992017025e-05, "loss": 1.9834, "step": 1247 }, { "epoch": 0.2, "grad_norm": 9.45969009399414, "learning_rate": 1.850830136273812e-05, "loss": 1.631, "step": 1248 }, { "epoch": 0.2, "grad_norm": 8.674372673034668, "learning_rate": 1.8505560420177648e-05, "loss": 1.8695, "step": 1249 }, { "epoch": 0.2, "grad_norm": 8.159158706665039, "learning_rate": 1.8502817165080828e-05, "loss": 1.5737, "step": 1250 }, { "epoch": 0.2, "grad_norm": 12.10987663269043, "learning_rate": 1.8500071598193512e-05, "loss": 1.9596, "step": 1251 }, { "epoch": 0.2, "grad_norm": 10.180919647216797, "learning_rate": 1.8497323720262184e-05, "loss": 1.9142, "step": 1252 }, { "epoch": 0.2, "grad_norm": 7.314725875854492, "learning_rate": 1.8494573532033947e-05, "loss": 1.578, "step": 1253 }, { "epoch": 0.2, "grad_norm": 8.193748474121094, "learning_rate": 1.849182103425654e-05, "loss": 1.8035, "step": 1254 }, { "epoch": 0.2, "grad_norm": 7.796138286590576, "learning_rate": 1.8489066227678322e-05, "loss": 1.7477, "step": 1255 }, { "epoch": 0.2, "grad_norm": 8.194533348083496, "learning_rate": 1.8486309113048288e-05, "loss": 1.7683, "step": 1256 }, { "epoch": 0.2, "grad_norm": 8.529131889343262, "learning_rate": 1.8483549691116053e-05, "loss": 1.7396, "step": 1257 }, { "epoch": 0.2, "grad_norm": 7.564024448394775, "learning_rate": 1.848078796263187e-05, "loss": 1.7819, "step": 1258 }, { "epoch": 0.2, "grad_norm": 9.09427261352539, "learning_rate": 1.8478023928346604e-05, "loss": 1.9268, "step": 1259 }, { "epoch": 0.2, "grad_norm": 11.1843843460083, "learning_rate": 1.8475257589011754e-05, "loss": 1.8905, "step": 1260 }, { "epoch": 0.2, "grad_norm": 7.873772621154785, "learning_rate": 1.847248894537946e-05, "loss": 1.8162, "step": 1261 }, { "epoch": 0.2, "grad_norm": 14.368592262268066, "learning_rate": 1.8469717998202464e-05, "loss": 1.7666, "step": 1262 }, { "epoch": 0.2, "grad_norm": 8.578486442565918, "learning_rate": 1.8466944748234148e-05, "loss": 1.8481, "step": 1263 }, { "epoch": 0.2, "grad_norm": 7.919323444366455, "learning_rate": 1.8464169196228518e-05, "loss": 1.607, "step": 1264 }, { "epoch": 0.2, "grad_norm": 8.683323860168457, "learning_rate": 1.846139134294021e-05, "loss": 1.8998, "step": 1265 }, { "epoch": 0.2, "grad_norm": 6.919980525970459, "learning_rate": 1.8458611189124473e-05, "loss": 1.8613, "step": 1266 }, { "epoch": 0.2, "grad_norm": 9.115490913391113, "learning_rate": 1.8455828735537194e-05, "loss": 1.8335, "step": 1267 }, { "epoch": 0.2, "grad_norm": 7.6121439933776855, "learning_rate": 1.8453043982934886e-05, "loss": 1.9959, "step": 1268 }, { "epoch": 0.2, "grad_norm": 10.027979850769043, "learning_rate": 1.8450256932074674e-05, "loss": 1.605, "step": 1269 }, { "epoch": 0.2, "grad_norm": 8.034314155578613, "learning_rate": 1.844746758371432e-05, "loss": 1.8119, "step": 1270 }, { "epoch": 0.2, "grad_norm": 12.048789024353027, "learning_rate": 1.8444675938612203e-05, "loss": 1.8008, "step": 1271 }, { "epoch": 0.2, "grad_norm": 8.489983558654785, "learning_rate": 1.8441881997527336e-05, "loss": 1.9254, "step": 1272 }, { "epoch": 0.2, "grad_norm": 9.083150863647461, "learning_rate": 1.8439085761219345e-05, "loss": 2.0765, "step": 1273 }, { "epoch": 0.21, "grad_norm": 7.749396800994873, "learning_rate": 1.8436287230448493e-05, "loss": 1.6754, "step": 1274 }, { "epoch": 0.21, "grad_norm": 11.900274276733398, "learning_rate": 1.8433486405975652e-05, "loss": 1.7634, "step": 1275 }, { "epoch": 0.21, "grad_norm": 8.176797866821289, "learning_rate": 1.8430683288562322e-05, "loss": 1.7089, "step": 1276 }, { "epoch": 0.21, "grad_norm": 8.296685218811035, "learning_rate": 1.842787787897064e-05, "loss": 1.9378, "step": 1277 }, { "epoch": 0.21, "grad_norm": 8.023030281066895, "learning_rate": 1.8425070177963346e-05, "loss": 1.7682, "step": 1278 }, { "epoch": 0.21, "grad_norm": 10.130829811096191, "learning_rate": 1.8422260186303818e-05, "loss": 1.9535, "step": 1279 }, { "epoch": 0.21, "grad_norm": 8.522187232971191, "learning_rate": 1.841944790475605e-05, "loss": 1.7973, "step": 1280 }, { "epoch": 0.21, "grad_norm": 9.679072380065918, "learning_rate": 1.8416633334084654e-05, "loss": 1.8193, "step": 1281 }, { "epoch": 0.21, "grad_norm": 8.095109939575195, "learning_rate": 1.841381647505488e-05, "loss": 1.8876, "step": 1282 }, { "epoch": 0.21, "grad_norm": 10.104700088500977, "learning_rate": 1.8410997328432586e-05, "loss": 1.751, "step": 1283 }, { "epoch": 0.21, "grad_norm": 9.805227279663086, "learning_rate": 1.8408175894984254e-05, "loss": 1.8591, "step": 1284 }, { "epoch": 0.21, "grad_norm": 9.25502872467041, "learning_rate": 1.840535217547699e-05, "loss": 1.7717, "step": 1285 }, { "epoch": 0.21, "grad_norm": 7.533921718597412, "learning_rate": 1.840252617067853e-05, "loss": 1.9073, "step": 1286 }, { "epoch": 0.21, "grad_norm": 11.94642162322998, "learning_rate": 1.8399697881357214e-05, "loss": 1.7527, "step": 1287 }, { "epoch": 0.21, "grad_norm": 7.692166805267334, "learning_rate": 1.8396867308282015e-05, "loss": 1.9876, "step": 1288 }, { "epoch": 0.21, "grad_norm": 9.117722511291504, "learning_rate": 1.8394034452222522e-05, "loss": 1.9055, "step": 1289 }, { "epoch": 0.21, "grad_norm": 9.927703857421875, "learning_rate": 1.8391199313948952e-05, "loss": 1.7268, "step": 1290 }, { "epoch": 0.21, "grad_norm": 10.902655601501465, "learning_rate": 1.8388361894232132e-05, "loss": 1.8988, "step": 1291 }, { "epoch": 0.21, "grad_norm": 8.521817207336426, "learning_rate": 1.8385522193843516e-05, "loss": 1.4875, "step": 1292 }, { "epoch": 0.21, "grad_norm": 7.50910758972168, "learning_rate": 1.838268021355518e-05, "loss": 1.7963, "step": 1293 }, { "epoch": 0.21, "grad_norm": 8.528274536132812, "learning_rate": 1.837983595413981e-05, "loss": 1.9592, "step": 1294 }, { "epoch": 0.21, "grad_norm": 7.905785083770752, "learning_rate": 1.8376989416370724e-05, "loss": 1.9065, "step": 1295 }, { "epoch": 0.21, "grad_norm": 8.554696083068848, "learning_rate": 1.8374140601021854e-05, "loss": 1.7116, "step": 1296 }, { "epoch": 0.21, "grad_norm": 7.8974127769470215, "learning_rate": 1.8371289508867743e-05, "loss": 1.8736, "step": 1297 }, { "epoch": 0.21, "grad_norm": 9.470829963684082, "learning_rate": 1.836843614068357e-05, "loss": 1.8049, "step": 1298 }, { "epoch": 0.21, "grad_norm": 8.255562782287598, "learning_rate": 1.836558049724512e-05, "loss": 1.8343, "step": 1299 }, { "epoch": 0.21, "grad_norm": 9.269615173339844, "learning_rate": 1.8362722579328797e-05, "loss": 1.7299, "step": 1300 }, { "epoch": 0.21, "grad_norm": 9.339948654174805, "learning_rate": 1.8359862387711635e-05, "loss": 1.9384, "step": 1301 }, { "epoch": 0.21, "grad_norm": 7.4764790534973145, "learning_rate": 1.835699992317127e-05, "loss": 1.8128, "step": 1302 }, { "epoch": 0.21, "grad_norm": 9.269831657409668, "learning_rate": 1.8354135186485965e-05, "loss": 1.8453, "step": 1303 }, { "epoch": 0.21, "grad_norm": 8.966657638549805, "learning_rate": 1.83512681784346e-05, "loss": 2.1305, "step": 1304 }, { "epoch": 0.21, "grad_norm": 9.08002758026123, "learning_rate": 1.8348398899796676e-05, "loss": 2.1936, "step": 1305 }, { "epoch": 0.21, "grad_norm": 7.097823619842529, "learning_rate": 1.8345527351352305e-05, "loss": 1.756, "step": 1306 }, { "epoch": 0.21, "grad_norm": 9.40312671661377, "learning_rate": 1.834265353388222e-05, "loss": 1.8136, "step": 1307 }, { "epoch": 0.21, "grad_norm": 8.88334846496582, "learning_rate": 1.833977744816777e-05, "loss": 2.0288, "step": 1308 }, { "epoch": 0.21, "grad_norm": 7.334694862365723, "learning_rate": 1.833689909499091e-05, "loss": 1.6028, "step": 1309 }, { "epoch": 0.21, "grad_norm": 7.796982765197754, "learning_rate": 1.8334018475134238e-05, "loss": 1.7179, "step": 1310 }, { "epoch": 0.21, "grad_norm": 7.6836957931518555, "learning_rate": 1.833113558938094e-05, "loss": 1.8184, "step": 1311 }, { "epoch": 0.21, "grad_norm": 7.916046142578125, "learning_rate": 1.8328250438514837e-05, "loss": 1.9794, "step": 1312 }, { "epoch": 0.21, "grad_norm": 11.210794448852539, "learning_rate": 1.8325363023320353e-05, "loss": 1.8968, "step": 1313 }, { "epoch": 0.21, "grad_norm": 8.618979454040527, "learning_rate": 1.832247334458254e-05, "loss": 1.8934, "step": 1314 }, { "epoch": 0.21, "grad_norm": 7.751104831695557, "learning_rate": 1.8319581403087053e-05, "loss": 2.0342, "step": 1315 }, { "epoch": 0.21, "grad_norm": 8.686817169189453, "learning_rate": 1.8316687199620175e-05, "loss": 2.009, "step": 1316 }, { "epoch": 0.21, "grad_norm": 8.541163444519043, "learning_rate": 1.831379073496879e-05, "loss": 1.9103, "step": 1317 }, { "epoch": 0.21, "grad_norm": 9.199752807617188, "learning_rate": 1.831089200992041e-05, "loss": 1.8129, "step": 1318 }, { "epoch": 0.21, "grad_norm": 9.81206226348877, "learning_rate": 1.830799102526315e-05, "loss": 1.751, "step": 1319 }, { "epoch": 0.21, "grad_norm": 6.634701728820801, "learning_rate": 1.8305087781785745e-05, "loss": 1.7804, "step": 1320 }, { "epoch": 0.21, "grad_norm": 8.057719230651855, "learning_rate": 1.830218228027755e-05, "loss": 1.7931, "step": 1321 }, { "epoch": 0.21, "grad_norm": 9.194022178649902, "learning_rate": 1.8299274521528528e-05, "loss": 1.7816, "step": 1322 }, { "epoch": 0.21, "grad_norm": 7.514299392700195, "learning_rate": 1.8296364506329246e-05, "loss": 1.9195, "step": 1323 }, { "epoch": 0.21, "grad_norm": 8.05358600616455, "learning_rate": 1.8293452235470906e-05, "loss": 1.7508, "step": 1324 }, { "epoch": 0.21, "grad_norm": 8.369933128356934, "learning_rate": 1.8290537709745303e-05, "loss": 1.7723, "step": 1325 }, { "epoch": 0.21, "grad_norm": 10.120149612426758, "learning_rate": 1.8287620929944858e-05, "loss": 1.8659, "step": 1326 }, { "epoch": 0.21, "grad_norm": 7.357173919677734, "learning_rate": 1.82847018968626e-05, "loss": 1.9415, "step": 1327 }, { "epoch": 0.21, "grad_norm": 7.798434734344482, "learning_rate": 1.8281780611292165e-05, "loss": 1.75, "step": 1328 }, { "epoch": 0.21, "grad_norm": 11.043705940246582, "learning_rate": 1.8278857074027814e-05, "loss": 1.8723, "step": 1329 }, { "epoch": 0.21, "grad_norm": 10.642631530761719, "learning_rate": 1.8275931285864418e-05, "loss": 1.7833, "step": 1330 }, { "epoch": 0.21, "grad_norm": 8.426919937133789, "learning_rate": 1.8273003247597442e-05, "loss": 1.8105, "step": 1331 }, { "epoch": 0.21, "grad_norm": 17.741287231445312, "learning_rate": 1.827007296002299e-05, "loss": 2.1757, "step": 1332 }, { "epoch": 0.21, "grad_norm": 10.50174331665039, "learning_rate": 1.8267140423937753e-05, "loss": 1.6868, "step": 1333 }, { "epoch": 0.21, "grad_norm": 8.069438934326172, "learning_rate": 1.8264205640139052e-05, "loss": 1.7844, "step": 1334 }, { "epoch": 0.21, "grad_norm": 7.463778972625732, "learning_rate": 1.826126860942481e-05, "loss": 1.7491, "step": 1335 }, { "epoch": 0.22, "grad_norm": 8.813199043273926, "learning_rate": 1.825832933259356e-05, "loss": 1.6364, "step": 1336 }, { "epoch": 0.22, "grad_norm": 12.492267608642578, "learning_rate": 1.8255387810444447e-05, "loss": 1.8942, "step": 1337 }, { "epoch": 0.22, "grad_norm": 10.353422164916992, "learning_rate": 1.825244404377723e-05, "loss": 1.6062, "step": 1338 }, { "epoch": 0.22, "grad_norm": 10.436239242553711, "learning_rate": 1.8249498033392275e-05, "loss": 1.7787, "step": 1339 }, { "epoch": 0.22, "grad_norm": 9.240364074707031, "learning_rate": 1.8246549780090557e-05, "loss": 1.9947, "step": 1340 }, { "epoch": 0.22, "grad_norm": 17.19481086730957, "learning_rate": 1.824359928467367e-05, "loss": 1.8218, "step": 1341 }, { "epoch": 0.22, "grad_norm": 14.504600524902344, "learning_rate": 1.82406465479438e-05, "loss": 1.7018, "step": 1342 }, { "epoch": 0.22, "grad_norm": 9.210302352905273, "learning_rate": 1.8237691570703758e-05, "loss": 1.737, "step": 1343 }, { "epoch": 0.22, "grad_norm": 9.754006385803223, "learning_rate": 1.823473435375696e-05, "loss": 1.8558, "step": 1344 }, { "epoch": 0.22, "grad_norm": 9.77143383026123, "learning_rate": 1.8231774897907426e-05, "loss": 2.0714, "step": 1345 }, { "epoch": 0.22, "grad_norm": 7.1131157875061035, "learning_rate": 1.822881320395979e-05, "loss": 1.7879, "step": 1346 }, { "epoch": 0.22, "grad_norm": 8.273902893066406, "learning_rate": 1.822584927271929e-05, "loss": 1.8276, "step": 1347 }, { "epoch": 0.22, "grad_norm": 7.225007057189941, "learning_rate": 1.8222883104991782e-05, "loss": 1.6709, "step": 1348 }, { "epoch": 0.22, "grad_norm": 9.202335357666016, "learning_rate": 1.8219914701583715e-05, "loss": 2.089, "step": 1349 }, { "epoch": 0.22, "grad_norm": 11.476844787597656, "learning_rate": 1.8216944063302162e-05, "loss": 1.8693, "step": 1350 }, { "epoch": 0.22, "grad_norm": 10.235746383666992, "learning_rate": 1.8213971190954792e-05, "loss": 1.762, "step": 1351 }, { "epoch": 0.22, "grad_norm": 9.071858406066895, "learning_rate": 1.821099608534988e-05, "loss": 1.9227, "step": 1352 }, { "epoch": 0.22, "grad_norm": 8.609461784362793, "learning_rate": 1.8208018747296324e-05, "loss": 1.9591, "step": 1353 }, { "epoch": 0.22, "grad_norm": 10.443500518798828, "learning_rate": 1.8205039177603608e-05, "loss": 1.7592, "step": 1354 }, { "epoch": 0.22, "grad_norm": 8.754007339477539, "learning_rate": 1.820205737708184e-05, "loss": 1.859, "step": 1355 }, { "epoch": 0.22, "grad_norm": 10.27514362335205, "learning_rate": 1.8199073346541727e-05, "loss": 1.8525, "step": 1356 }, { "epoch": 0.22, "grad_norm": 7.937394618988037, "learning_rate": 1.8196087086794576e-05, "loss": 1.9053, "step": 1357 }, { "epoch": 0.22, "grad_norm": 9.699028968811035, "learning_rate": 1.8193098598652318e-05, "loss": 1.8974, "step": 1358 }, { "epoch": 0.22, "grad_norm": 7.327558994293213, "learning_rate": 1.819010788292747e-05, "loss": 2.0025, "step": 1359 }, { "epoch": 0.22, "grad_norm": 10.876296997070312, "learning_rate": 1.8187114940433165e-05, "loss": 1.8693, "step": 1360 }, { "epoch": 0.22, "grad_norm": 9.249028205871582, "learning_rate": 1.818411977198314e-05, "loss": 1.8728, "step": 1361 }, { "epoch": 0.22, "grad_norm": 14.072016716003418, "learning_rate": 1.818112237839174e-05, "loss": 1.9115, "step": 1362 }, { "epoch": 0.22, "grad_norm": 9.951681137084961, "learning_rate": 1.817812276047391e-05, "loss": 1.914, "step": 1363 }, { "epoch": 0.22, "grad_norm": 7.81296968460083, "learning_rate": 1.8175120919045204e-05, "loss": 1.9772, "step": 1364 }, { "epoch": 0.22, "grad_norm": 7.917456150054932, "learning_rate": 1.8172116854921774e-05, "loss": 1.7403, "step": 1365 }, { "epoch": 0.22, "grad_norm": 11.059257507324219, "learning_rate": 1.8169110568920385e-05, "loss": 1.7025, "step": 1366 }, { "epoch": 0.22, "grad_norm": 8.865398406982422, "learning_rate": 1.81661020618584e-05, "loss": 1.8379, "step": 1367 }, { "epoch": 0.22, "grad_norm": 9.933769226074219, "learning_rate": 1.8163091334553787e-05, "loss": 1.6208, "step": 1368 }, { "epoch": 0.22, "grad_norm": 8.046675682067871, "learning_rate": 1.816007838782512e-05, "loss": 1.6587, "step": 1369 }, { "epoch": 0.22, "grad_norm": 8.186602592468262, "learning_rate": 1.8157063222491572e-05, "loss": 1.8559, "step": 1370 }, { "epoch": 0.22, "grad_norm": 8.196744918823242, "learning_rate": 1.8154045839372926e-05, "loss": 1.9499, "step": 1371 }, { "epoch": 0.22, "grad_norm": 12.990235328674316, "learning_rate": 1.8151026239289556e-05, "loss": 1.8764, "step": 1372 }, { "epoch": 0.22, "grad_norm": 7.839725971221924, "learning_rate": 1.8148004423062458e-05, "loss": 1.9665, "step": 1373 }, { "epoch": 0.22, "grad_norm": 9.224679946899414, "learning_rate": 1.8144980391513213e-05, "loss": 1.8822, "step": 1374 }, { "epoch": 0.22, "grad_norm": 7.2393903732299805, "learning_rate": 1.8141954145464008e-05, "loss": 1.8577, "step": 1375 }, { "epoch": 0.22, "grad_norm": 12.863110542297363, "learning_rate": 1.813892568573764e-05, "loss": 1.7993, "step": 1376 }, { "epoch": 0.22, "grad_norm": 8.873236656188965, "learning_rate": 1.81358950131575e-05, "loss": 1.8154, "step": 1377 }, { "epoch": 0.22, "grad_norm": 8.94175910949707, "learning_rate": 1.8132862128547583e-05, "loss": 1.9526, "step": 1378 }, { "epoch": 0.22, "grad_norm": 14.120600700378418, "learning_rate": 1.8129827032732485e-05, "loss": 1.7775, "step": 1379 }, { "epoch": 0.22, "grad_norm": 8.562819480895996, "learning_rate": 1.812678972653741e-05, "loss": 1.9494, "step": 1380 }, { "epoch": 0.22, "grad_norm": 10.010725021362305, "learning_rate": 1.8123750210788148e-05, "loss": 1.6063, "step": 1381 }, { "epoch": 0.22, "grad_norm": 7.26815938949585, "learning_rate": 1.8120708486311098e-05, "loss": 1.8032, "step": 1382 }, { "epoch": 0.22, "grad_norm": 7.198014736175537, "learning_rate": 1.811766455393327e-05, "loss": 1.783, "step": 1383 }, { "epoch": 0.22, "grad_norm": 9.155001640319824, "learning_rate": 1.8114618414482254e-05, "loss": 1.8909, "step": 1384 }, { "epoch": 0.22, "grad_norm": 10.430132865905762, "learning_rate": 1.8111570068786254e-05, "loss": 1.8547, "step": 1385 }, { "epoch": 0.22, "grad_norm": 6.872344970703125, "learning_rate": 1.8108519517674075e-05, "loss": 1.6896, "step": 1386 }, { "epoch": 0.22, "grad_norm": 7.739284515380859, "learning_rate": 1.810546676197511e-05, "loss": 1.7815, "step": 1387 }, { "epoch": 0.22, "grad_norm": 8.201643943786621, "learning_rate": 1.810241180251936e-05, "loss": 1.8579, "step": 1388 }, { "epoch": 0.22, "grad_norm": 7.831824779510498, "learning_rate": 1.8099354640137427e-05, "loss": 1.6803, "step": 1389 }, { "epoch": 0.22, "grad_norm": 7.082134246826172, "learning_rate": 1.8096295275660504e-05, "loss": 1.9189, "step": 1390 }, { "epoch": 0.22, "grad_norm": 7.954312801361084, "learning_rate": 1.8093233709920392e-05, "loss": 1.5716, "step": 1391 }, { "epoch": 0.22, "grad_norm": 8.924363136291504, "learning_rate": 1.8090169943749477e-05, "loss": 1.8657, "step": 1392 }, { "epoch": 0.22, "grad_norm": 9.077749252319336, "learning_rate": 1.8087103977980758e-05, "loss": 1.7657, "step": 1393 }, { "epoch": 0.22, "grad_norm": 8.1124906539917, "learning_rate": 1.8084035813447824e-05, "loss": 2.0261, "step": 1394 }, { "epoch": 0.22, "grad_norm": 7.316488265991211, "learning_rate": 1.8080965450984867e-05, "loss": 1.8977, "step": 1395 }, { "epoch": 0.22, "grad_norm": 8.296265602111816, "learning_rate": 1.807789289142667e-05, "loss": 1.7847, "step": 1396 }, { "epoch": 0.22, "grad_norm": 7.228989124298096, "learning_rate": 1.8074818135608617e-05, "loss": 1.8093, "step": 1397 }, { "epoch": 0.23, "grad_norm": 8.73561954498291, "learning_rate": 1.8071741184366686e-05, "loss": 1.7382, "step": 1398 }, { "epoch": 0.23, "grad_norm": 6.866471290588379, "learning_rate": 1.8068662038537463e-05, "loss": 1.7335, "step": 1399 }, { "epoch": 0.23, "grad_norm": 7.113195896148682, "learning_rate": 1.806558069895811e-05, "loss": 1.8548, "step": 1400 }, { "epoch": 0.23, "grad_norm": 11.464706420898438, "learning_rate": 1.806249716646641e-05, "loss": 1.8974, "step": 1401 }, { "epoch": 0.23, "grad_norm": 7.172045707702637, "learning_rate": 1.8059411441900723e-05, "loss": 1.8336, "step": 1402 }, { "epoch": 0.23, "grad_norm": 7.422493934631348, "learning_rate": 1.8056323526100015e-05, "loss": 1.917, "step": 1403 }, { "epoch": 0.23, "grad_norm": 8.455967903137207, "learning_rate": 1.8053233419903843e-05, "loss": 1.6722, "step": 1404 }, { "epoch": 0.23, "grad_norm": 10.011442184448242, "learning_rate": 1.805014112415236e-05, "loss": 1.8077, "step": 1405 }, { "epoch": 0.23, "grad_norm": 9.260128021240234, "learning_rate": 1.804704663968632e-05, "loss": 1.7274, "step": 1406 }, { "epoch": 0.23, "grad_norm": 7.648123264312744, "learning_rate": 1.804394996734706e-05, "loss": 1.6599, "step": 1407 }, { "epoch": 0.23, "grad_norm": 7.756584167480469, "learning_rate": 1.804085110797653e-05, "loss": 1.7963, "step": 1408 }, { "epoch": 0.23, "grad_norm": 8.109429359436035, "learning_rate": 1.8037750062417253e-05, "loss": 1.7461, "step": 1409 }, { "epoch": 0.23, "grad_norm": 7.966662883758545, "learning_rate": 1.8034646831512367e-05, "loss": 1.9081, "step": 1410 }, { "epoch": 0.23, "grad_norm": 7.272926330566406, "learning_rate": 1.803154141610559e-05, "loss": 1.7642, "step": 1411 }, { "epoch": 0.23, "grad_norm": 11.738434791564941, "learning_rate": 1.8028433817041237e-05, "loss": 1.8333, "step": 1412 }, { "epoch": 0.23, "grad_norm": 8.684165000915527, "learning_rate": 1.8025324035164225e-05, "loss": 1.7315, "step": 1413 }, { "epoch": 0.23, "grad_norm": 10.010868072509766, "learning_rate": 1.802221207132005e-05, "loss": 1.9076, "step": 1414 }, { "epoch": 0.23, "grad_norm": 12.349165916442871, "learning_rate": 1.801909792635481e-05, "loss": 1.8082, "step": 1415 }, { "epoch": 0.23, "grad_norm": 7.566521644592285, "learning_rate": 1.8015981601115203e-05, "loss": 1.6072, "step": 1416 }, { "epoch": 0.23, "grad_norm": 10.811861991882324, "learning_rate": 1.8012863096448505e-05, "loss": 1.9542, "step": 1417 }, { "epoch": 0.23, "grad_norm": 7.650590896606445, "learning_rate": 1.800974241320259e-05, "loss": 1.6704, "step": 1418 }, { "epoch": 0.23, "grad_norm": 11.581965446472168, "learning_rate": 1.800661955222593e-05, "loss": 1.8109, "step": 1419 }, { "epoch": 0.23, "grad_norm": 8.093099594116211, "learning_rate": 1.8003494514367586e-05, "loss": 1.7991, "step": 1420 }, { "epoch": 0.23, "grad_norm": 11.913718223571777, "learning_rate": 1.80003673004772e-05, "loss": 1.7871, "step": 1421 }, { "epoch": 0.23, "grad_norm": 7.63661527633667, "learning_rate": 1.799723791140503e-05, "loss": 1.7646, "step": 1422 }, { "epoch": 0.23, "grad_norm": 7.512375354766846, "learning_rate": 1.7994106348001898e-05, "loss": 1.6548, "step": 1423 }, { "epoch": 0.23, "grad_norm": 8.328794479370117, "learning_rate": 1.7990972611119236e-05, "loss": 1.463, "step": 1424 }, { "epoch": 0.23, "grad_norm": 8.493249893188477, "learning_rate": 1.798783670160906e-05, "loss": 1.8202, "step": 1425 }, { "epoch": 0.23, "grad_norm": 8.35335922241211, "learning_rate": 1.7984698620323978e-05, "loss": 1.9553, "step": 1426 }, { "epoch": 0.23, "grad_norm": 7.105823040008545, "learning_rate": 1.7981558368117188e-05, "loss": 1.6175, "step": 1427 }, { "epoch": 0.23, "grad_norm": 9.643932342529297, "learning_rate": 1.7978415945842473e-05, "loss": 1.8868, "step": 1428 }, { "epoch": 0.23, "grad_norm": 9.525238037109375, "learning_rate": 1.797527135435422e-05, "loss": 1.7314, "step": 1429 }, { "epoch": 0.23, "grad_norm": 8.257509231567383, "learning_rate": 1.797212459450739e-05, "loss": 1.8769, "step": 1430 }, { "epoch": 0.23, "grad_norm": 8.125737190246582, "learning_rate": 1.796897566715754e-05, "loss": 1.6836, "step": 1431 }, { "epoch": 0.23, "grad_norm": 10.427423477172852, "learning_rate": 1.7965824573160825e-05, "loss": 1.5189, "step": 1432 }, { "epoch": 0.23, "grad_norm": 7.612471103668213, "learning_rate": 1.7962671313373975e-05, "loss": 1.7275, "step": 1433 }, { "epoch": 0.23, "grad_norm": 10.858115196228027, "learning_rate": 1.7959515888654317e-05, "loss": 1.8261, "step": 1434 }, { "epoch": 0.23, "grad_norm": 8.874879837036133, "learning_rate": 1.7956358299859758e-05, "loss": 1.9106, "step": 1435 }, { "epoch": 0.23, "grad_norm": 7.795431613922119, "learning_rate": 1.7953198547848812e-05, "loss": 1.5169, "step": 1436 }, { "epoch": 0.23, "grad_norm": 9.648048400878906, "learning_rate": 1.7950036633480557e-05, "loss": 1.8556, "step": 1437 }, { "epoch": 0.23, "grad_norm": 7.76369571685791, "learning_rate": 1.7946872557614677e-05, "loss": 1.6789, "step": 1438 }, { "epoch": 0.23, "grad_norm": 9.746540069580078, "learning_rate": 1.794370632111144e-05, "loss": 1.9739, "step": 1439 }, { "epoch": 0.23, "grad_norm": 10.373836517333984, "learning_rate": 1.794053792483169e-05, "loss": 1.971, "step": 1440 }, { "epoch": 0.23, "grad_norm": 8.705808639526367, "learning_rate": 1.7937367369636872e-05, "loss": 1.8406, "step": 1441 }, { "epoch": 0.23, "grad_norm": 10.463668823242188, "learning_rate": 1.7934194656389018e-05, "loss": 1.7053, "step": 1442 }, { "epoch": 0.23, "grad_norm": 7.657758712768555, "learning_rate": 1.7931019785950735e-05, "loss": 1.968, "step": 1443 }, { "epoch": 0.23, "grad_norm": 6.578880786895752, "learning_rate": 1.7927842759185225e-05, "loss": 1.6039, "step": 1444 }, { "epoch": 0.23, "grad_norm": 8.379292488098145, "learning_rate": 1.7924663576956275e-05, "loss": 1.8366, "step": 1445 }, { "epoch": 0.23, "grad_norm": 9.00742244720459, "learning_rate": 1.7921482240128262e-05, "loss": 1.706, "step": 1446 }, { "epoch": 0.23, "grad_norm": 7.367388725280762, "learning_rate": 1.791829874956614e-05, "loss": 1.6409, "step": 1447 }, { "epoch": 0.23, "grad_norm": 7.896468162536621, "learning_rate": 1.7915113106135454e-05, "loss": 1.6673, "step": 1448 }, { "epoch": 0.23, "grad_norm": 7.64634895324707, "learning_rate": 1.7911925310702333e-05, "loss": 1.8589, "step": 1449 }, { "epoch": 0.23, "grad_norm": 9.058424949645996, "learning_rate": 1.790873536413349e-05, "loss": 1.9106, "step": 1450 }, { "epoch": 0.23, "grad_norm": 9.559104919433594, "learning_rate": 1.790554326729623e-05, "loss": 1.7819, "step": 1451 }, { "epoch": 0.23, "grad_norm": 7.826646327972412, "learning_rate": 1.790234902105843e-05, "loss": 1.8543, "step": 1452 }, { "epoch": 0.23, "grad_norm": 12.396729469299316, "learning_rate": 1.7899152626288563e-05, "loss": 1.7169, "step": 1453 }, { "epoch": 0.23, "grad_norm": 8.773192405700684, "learning_rate": 1.789595408385568e-05, "loss": 1.6313, "step": 1454 }, { "epoch": 0.23, "grad_norm": 7.82925271987915, "learning_rate": 1.7892753394629415e-05, "loss": 1.9365, "step": 1455 }, { "epoch": 0.23, "grad_norm": 11.006667137145996, "learning_rate": 1.7889550559479992e-05, "loss": 1.8688, "step": 1456 }, { "epoch": 0.23, "grad_norm": 9.412535667419434, "learning_rate": 1.7886345579278218e-05, "loss": 1.5818, "step": 1457 }, { "epoch": 0.23, "grad_norm": 7.688195705413818, "learning_rate": 1.7883138454895472e-05, "loss": 1.7684, "step": 1458 }, { "epoch": 0.23, "grad_norm": 9.537877082824707, "learning_rate": 1.7879929187203728e-05, "loss": 1.9148, "step": 1459 }, { "epoch": 0.24, "grad_norm": 7.922815799713135, "learning_rate": 1.7876717777075538e-05, "loss": 1.6838, "step": 1460 }, { "epoch": 0.24, "grad_norm": 9.185300827026367, "learning_rate": 1.7873504225384034e-05, "loss": 1.7073, "step": 1461 }, { "epoch": 0.24, "grad_norm": 8.183757781982422, "learning_rate": 1.787028853300294e-05, "loss": 1.7307, "step": 1462 }, { "epoch": 0.24, "grad_norm": 8.292150497436523, "learning_rate": 1.786707070080655e-05, "loss": 1.6267, "step": 1463 }, { "epoch": 0.24, "grad_norm": 7.186605930328369, "learning_rate": 1.7863850729669746e-05, "loss": 1.7409, "step": 1464 }, { "epoch": 0.24, "grad_norm": 8.339845657348633, "learning_rate": 1.786062862046799e-05, "loss": 2.0212, "step": 1465 }, { "epoch": 0.24, "grad_norm": 12.036404609680176, "learning_rate": 1.785740437407733e-05, "loss": 1.7669, "step": 1466 }, { "epoch": 0.24, "grad_norm": 8.558846473693848, "learning_rate": 1.7854177991374385e-05, "loss": 1.7792, "step": 1467 }, { "epoch": 0.24, "grad_norm": 7.301778316497803, "learning_rate": 1.7850949473236365e-05, "loss": 1.8033, "step": 1468 }, { "epoch": 0.24, "grad_norm": 14.099976539611816, "learning_rate": 1.7847718820541054e-05, "loss": 1.8103, "step": 1469 }, { "epoch": 0.24, "grad_norm": 8.271821022033691, "learning_rate": 1.784448603416682e-05, "loss": 1.8644, "step": 1470 }, { "epoch": 0.24, "grad_norm": 6.64155387878418, "learning_rate": 1.784125111499261e-05, "loss": 1.7129, "step": 1471 }, { "epoch": 0.24, "grad_norm": 9.89072036743164, "learning_rate": 1.7838014063897952e-05, "loss": 1.9418, "step": 1472 }, { "epoch": 0.24, "grad_norm": 9.908866882324219, "learning_rate": 1.783477488176295e-05, "loss": 1.5312, "step": 1473 }, { "epoch": 0.24, "grad_norm": 8.769590377807617, "learning_rate": 1.783153356946829e-05, "loss": 1.8762, "step": 1474 }, { "epoch": 0.24, "grad_norm": 7.187405109405518, "learning_rate": 1.782829012789524e-05, "loss": 2.0378, "step": 1475 }, { "epoch": 0.24, "grad_norm": 8.71367359161377, "learning_rate": 1.782504455792564e-05, "loss": 1.9179, "step": 1476 }, { "epoch": 0.24, "grad_norm": 9.689894676208496, "learning_rate": 1.7821796860441914e-05, "loss": 1.9584, "step": 1477 }, { "epoch": 0.24, "grad_norm": 6.04748010635376, "learning_rate": 1.7818547036327065e-05, "loss": 1.7044, "step": 1478 }, { "epoch": 0.24, "grad_norm": 7.2961554527282715, "learning_rate": 1.781529508646467e-05, "loss": 1.743, "step": 1479 }, { "epoch": 0.24, "grad_norm": 9.996938705444336, "learning_rate": 1.7812041011738888e-05, "loss": 1.705, "step": 1480 }, { "epoch": 0.24, "grad_norm": 8.013317108154297, "learning_rate": 1.7808784813034452e-05, "loss": 1.6304, "step": 1481 }, { "epoch": 0.24, "grad_norm": 7.982450008392334, "learning_rate": 1.7805526491236675e-05, "loss": 1.8225, "step": 1482 }, { "epoch": 0.24, "grad_norm": 6.324021339416504, "learning_rate": 1.7802266047231446e-05, "loss": 1.7933, "step": 1483 }, { "epoch": 0.24, "grad_norm": 9.175995826721191, "learning_rate": 1.779900348190523e-05, "loss": 1.6459, "step": 1484 }, { "epoch": 0.24, "grad_norm": 9.249527931213379, "learning_rate": 1.779573879614508e-05, "loss": 1.6654, "step": 1485 }, { "epoch": 0.24, "grad_norm": 6.811366081237793, "learning_rate": 1.77924719908386e-05, "loss": 1.7309, "step": 1486 }, { "epoch": 0.24, "grad_norm": 9.685307502746582, "learning_rate": 1.7789203066874e-05, "loss": 2.1105, "step": 1487 }, { "epoch": 0.24, "grad_norm": 7.770510673522949, "learning_rate": 1.7785932025140042e-05, "loss": 1.8277, "step": 1488 }, { "epoch": 0.24, "grad_norm": 9.158452033996582, "learning_rate": 1.7782658866526083e-05, "loss": 1.5587, "step": 1489 }, { "epoch": 0.24, "grad_norm": 8.23528003692627, "learning_rate": 1.7779383591922038e-05, "loss": 1.8002, "step": 1490 }, { "epoch": 0.24, "grad_norm": 8.447554588317871, "learning_rate": 1.7776106202218414e-05, "loss": 1.6538, "step": 1491 }, { "epoch": 0.24, "grad_norm": 9.205560684204102, "learning_rate": 1.777282669830628e-05, "loss": 1.822, "step": 1492 }, { "epoch": 0.24, "grad_norm": 7.900252342224121, "learning_rate": 1.7769545081077285e-05, "loss": 1.7032, "step": 1493 }, { "epoch": 0.24, "grad_norm": 8.631847381591797, "learning_rate": 1.7766261351423653e-05, "loss": 1.8421, "step": 1494 }, { "epoch": 0.24, "grad_norm": 7.55910062789917, "learning_rate": 1.7762975510238184e-05, "loss": 1.7971, "step": 1495 }, { "epoch": 0.24, "grad_norm": 7.988583564758301, "learning_rate": 1.7759687558414246e-05, "loss": 1.7641, "step": 1496 }, { "epoch": 0.24, "grad_norm": 12.479720115661621, "learning_rate": 1.7756397496845783e-05, "loss": 2.0045, "step": 1497 }, { "epoch": 0.24, "grad_norm": 7.433629989624023, "learning_rate": 1.775310532642732e-05, "loss": 1.8642, "step": 1498 }, { "epoch": 0.24, "grad_norm": 11.38666820526123, "learning_rate": 1.7749811048053945e-05, "loss": 1.7662, "step": 1499 }, { "epoch": 0.24, "grad_norm": 10.99250602722168, "learning_rate": 1.7746514662621327e-05, "loss": 1.5927, "step": 1500 }, { "epoch": 0.24, "grad_norm": 8.572783470153809, "learning_rate": 1.77432161710257e-05, "loss": 2.0373, "step": 1501 }, { "epoch": 0.24, "grad_norm": 8.77432632446289, "learning_rate": 1.7739915574163882e-05, "loss": 1.8045, "step": 1502 }, { "epoch": 0.24, "grad_norm": 10.696587562561035, "learning_rate": 1.773661287293325e-05, "loss": 1.9462, "step": 1503 }, { "epoch": 0.24, "grad_norm": 8.78055477142334, "learning_rate": 1.7733308068231764e-05, "loss": 1.8736, "step": 1504 }, { "epoch": 0.24, "grad_norm": 6.747052192687988, "learning_rate": 1.7730001160957948e-05, "loss": 1.547, "step": 1505 }, { "epoch": 0.24, "grad_norm": 7.950844764709473, "learning_rate": 1.7726692152010903e-05, "loss": 1.8448, "step": 1506 }, { "epoch": 0.24, "grad_norm": 7.7995781898498535, "learning_rate": 1.7723381042290304e-05, "loss": 1.8084, "step": 1507 }, { "epoch": 0.24, "grad_norm": 7.728322505950928, "learning_rate": 1.7720067832696382e-05, "loss": 1.7945, "step": 1508 }, { "epoch": 0.24, "grad_norm": 11.93130111694336, "learning_rate": 1.7716752524129962e-05, "loss": 1.7903, "step": 1509 }, { "epoch": 0.24, "grad_norm": 7.18508768081665, "learning_rate": 1.7713435117492423e-05, "loss": 1.6411, "step": 1510 }, { "epoch": 0.24, "grad_norm": 7.201450347900391, "learning_rate": 1.7710115613685714e-05, "loss": 1.85, "step": 1511 }, { "epoch": 0.24, "grad_norm": 8.254265785217285, "learning_rate": 1.7706794013612367e-05, "loss": 1.7784, "step": 1512 }, { "epoch": 0.24, "grad_norm": 9.026972770690918, "learning_rate": 1.7703470318175467e-05, "loss": 1.7473, "step": 1513 }, { "epoch": 0.24, "grad_norm": 8.55180549621582, "learning_rate": 1.7700144528278688e-05, "loss": 1.8279, "step": 1514 }, { "epoch": 0.24, "grad_norm": 8.587038040161133, "learning_rate": 1.7696816644826254e-05, "loss": 1.7885, "step": 1515 }, { "epoch": 0.24, "grad_norm": 7.129663944244385, "learning_rate": 1.7693486668722972e-05, "loss": 1.5705, "step": 1516 }, { "epoch": 0.24, "grad_norm": 9.190773010253906, "learning_rate": 1.7690154600874215e-05, "loss": 1.6154, "step": 1517 }, { "epoch": 0.24, "grad_norm": 7.933110237121582, "learning_rate": 1.7686820442185916e-05, "loss": 1.7903, "step": 1518 }, { "epoch": 0.24, "grad_norm": 8.084612846374512, "learning_rate": 1.7683484193564596e-05, "loss": 1.6712, "step": 1519 }, { "epoch": 0.24, "grad_norm": 8.16672134399414, "learning_rate": 1.768014585591732e-05, "loss": 1.8155, "step": 1520 }, { "epoch": 0.24, "grad_norm": 11.657179832458496, "learning_rate": 1.7676805430151734e-05, "loss": 1.7615, "step": 1521 }, { "epoch": 0.25, "grad_norm": 7.255586624145508, "learning_rate": 1.7673462917176056e-05, "loss": 1.8453, "step": 1522 }, { "epoch": 0.25, "grad_norm": 9.249971389770508, "learning_rate": 1.7670118317899066e-05, "loss": 1.7222, "step": 1523 }, { "epoch": 0.25, "grad_norm": 10.090414047241211, "learning_rate": 1.7666771633230104e-05, "loss": 1.8553, "step": 1524 }, { "epoch": 0.25, "grad_norm": 7.219358921051025, "learning_rate": 1.766342286407909e-05, "loss": 1.625, "step": 1525 }, { "epoch": 0.25, "grad_norm": 8.66917610168457, "learning_rate": 1.7660072011356507e-05, "loss": 1.9554, "step": 1526 }, { "epoch": 0.25, "grad_norm": 7.926044940948486, "learning_rate": 1.7656719075973396e-05, "loss": 1.7948, "step": 1527 }, { "epoch": 0.25, "grad_norm": 7.918349742889404, "learning_rate": 1.765336405884137e-05, "loss": 1.7933, "step": 1528 }, { "epoch": 0.25, "grad_norm": 7.066356658935547, "learning_rate": 1.7650006960872615e-05, "loss": 1.7029, "step": 1529 }, { "epoch": 0.25, "grad_norm": 9.082940101623535, "learning_rate": 1.7646647782979875e-05, "loss": 1.6491, "step": 1530 }, { "epoch": 0.25, "grad_norm": 9.211152076721191, "learning_rate": 1.764328652607646e-05, "loss": 1.9159, "step": 1531 }, { "epoch": 0.25, "grad_norm": 9.38532829284668, "learning_rate": 1.7639923191076243e-05, "loss": 1.82, "step": 1532 }, { "epoch": 0.25, "grad_norm": 9.43415641784668, "learning_rate": 1.7636557778893672e-05, "loss": 1.9584, "step": 1533 }, { "epoch": 0.25, "grad_norm": 8.79750919342041, "learning_rate": 1.7633190290443744e-05, "loss": 1.658, "step": 1534 }, { "epoch": 0.25, "grad_norm": 8.925724983215332, "learning_rate": 1.7629820726642037e-05, "loss": 1.788, "step": 1535 }, { "epoch": 0.25, "grad_norm": 16.143098831176758, "learning_rate": 1.762644908840468e-05, "loss": 1.6089, "step": 1536 }, { "epoch": 0.25, "grad_norm": 8.659502983093262, "learning_rate": 1.7623075376648374e-05, "loss": 1.7561, "step": 1537 }, { "epoch": 0.25, "grad_norm": 9.831155776977539, "learning_rate": 1.7619699592290388e-05, "loss": 1.95, "step": 1538 }, { "epoch": 0.25, "grad_norm": 12.182497024536133, "learning_rate": 1.761632173624854e-05, "loss": 1.6859, "step": 1539 }, { "epoch": 0.25, "grad_norm": 8.406906127929688, "learning_rate": 1.7612941809441216e-05, "loss": 1.9374, "step": 1540 }, { "epoch": 0.25, "grad_norm": 8.408170700073242, "learning_rate": 1.7609559812787374e-05, "loss": 1.7754, "step": 1541 }, { "epoch": 0.25, "grad_norm": 10.06235408782959, "learning_rate": 1.7606175747206532e-05, "loss": 1.6698, "step": 1542 }, { "epoch": 0.25, "grad_norm": 13.81275749206543, "learning_rate": 1.760278961361876e-05, "loss": 1.905, "step": 1543 }, { "epoch": 0.25, "grad_norm": 7.745458126068115, "learning_rate": 1.7599401412944705e-05, "loss": 1.8106, "step": 1544 }, { "epoch": 0.25, "grad_norm": 9.471488952636719, "learning_rate": 1.7596011146105562e-05, "loss": 1.8299, "step": 1545 }, { "epoch": 0.25, "grad_norm": 8.19392204284668, "learning_rate": 1.7592618814023105e-05, "loss": 1.7661, "step": 1546 }, { "epoch": 0.25, "grad_norm": 9.254386901855469, "learning_rate": 1.758922441761964e-05, "loss": 1.5818, "step": 1547 }, { "epoch": 0.25, "grad_norm": 8.223718643188477, "learning_rate": 1.7585827957818075e-05, "loss": 1.8385, "step": 1548 }, { "epoch": 0.25, "grad_norm": 8.769546508789062, "learning_rate": 1.7582429435541843e-05, "loss": 1.8186, "step": 1549 }, { "epoch": 0.25, "grad_norm": 9.346748352050781, "learning_rate": 1.7579028851714958e-05, "loss": 1.7163, "step": 1550 }, { "epoch": 0.25, "grad_norm": 10.181279182434082, "learning_rate": 1.7575626207261985e-05, "loss": 2.109, "step": 1551 }, { "epoch": 0.25, "grad_norm": 8.625001907348633, "learning_rate": 1.7572221503108055e-05, "loss": 1.9642, "step": 1552 }, { "epoch": 0.25, "grad_norm": 7.4860711097717285, "learning_rate": 1.7568814740178855e-05, "loss": 1.7984, "step": 1553 }, { "epoch": 0.25, "grad_norm": 7.41444730758667, "learning_rate": 1.756540591940064e-05, "loss": 1.8296, "step": 1554 }, { "epoch": 0.25, "grad_norm": 8.380542755126953, "learning_rate": 1.756199504170021e-05, "loss": 2.0458, "step": 1555 }, { "epoch": 0.25, "grad_norm": 8.60954475402832, "learning_rate": 1.7558582108004936e-05, "loss": 1.6379, "step": 1556 }, { "epoch": 0.25, "grad_norm": 8.657482147216797, "learning_rate": 1.755516711924274e-05, "loss": 1.7622, "step": 1557 }, { "epoch": 0.25, "grad_norm": 7.430540084838867, "learning_rate": 1.7551750076342114e-05, "loss": 1.6865, "step": 1558 }, { "epoch": 0.25, "grad_norm": 13.032295227050781, "learning_rate": 1.7548330980232102e-05, "loss": 1.7302, "step": 1559 }, { "epoch": 0.25, "grad_norm": 7.528234481811523, "learning_rate": 1.75449098318423e-05, "loss": 1.7058, "step": 1560 }, { "epoch": 0.25, "grad_norm": 9.367335319519043, "learning_rate": 1.7541486632102867e-05, "loss": 1.8167, "step": 1561 }, { "epoch": 0.25, "grad_norm": 9.139829635620117, "learning_rate": 1.7538061381944524e-05, "loss": 1.875, "step": 1562 }, { "epoch": 0.25, "grad_norm": 9.27096939086914, "learning_rate": 1.7534634082298544e-05, "loss": 2.0107, "step": 1563 }, { "epoch": 0.25, "grad_norm": 7.978421688079834, "learning_rate": 1.7531204734096764e-05, "loss": 1.5226, "step": 1564 }, { "epoch": 0.25, "grad_norm": 8.285502433776855, "learning_rate": 1.752777333827157e-05, "loss": 2.0335, "step": 1565 }, { "epoch": 0.25, "grad_norm": 8.845949172973633, "learning_rate": 1.7524339895755903e-05, "loss": 1.6937, "step": 1566 }, { "epoch": 0.25, "grad_norm": 10.688368797302246, "learning_rate": 1.7520904407483276e-05, "loss": 1.8147, "step": 1567 }, { "epoch": 0.25, "grad_norm": 7.951508522033691, "learning_rate": 1.7517466874387737e-05, "loss": 1.6642, "step": 1568 }, { "epoch": 0.25, "grad_norm": 7.517422676086426, "learning_rate": 1.7514027297403906e-05, "loss": 1.8066, "step": 1569 }, { "epoch": 0.25, "grad_norm": 7.745699882507324, "learning_rate": 1.751058567746695e-05, "loss": 1.7188, "step": 1570 }, { "epoch": 0.25, "grad_norm": 9.360696792602539, "learning_rate": 1.75071420155126e-05, "loss": 1.8099, "step": 1571 }, { "epoch": 0.25, "grad_norm": 7.762989044189453, "learning_rate": 1.7503696312477138e-05, "loss": 1.5885, "step": 1572 }, { "epoch": 0.25, "grad_norm": 9.120820045471191, "learning_rate": 1.7500248569297387e-05, "loss": 1.727, "step": 1573 }, { "epoch": 0.25, "grad_norm": 8.038259506225586, "learning_rate": 1.749679878691075e-05, "loss": 1.7945, "step": 1574 }, { "epoch": 0.25, "grad_norm": 8.540531158447266, "learning_rate": 1.7493346966255165e-05, "loss": 1.6889, "step": 1575 }, { "epoch": 0.25, "grad_norm": 7.6534271240234375, "learning_rate": 1.748989310826914e-05, "loss": 1.7988, "step": 1576 }, { "epoch": 0.25, "grad_norm": 8.817056655883789, "learning_rate": 1.7486437213891717e-05, "loss": 1.8887, "step": 1577 }, { "epoch": 0.25, "grad_norm": 8.131871223449707, "learning_rate": 1.7482979284062512e-05, "loss": 1.6636, "step": 1578 }, { "epoch": 0.25, "grad_norm": 8.240620613098145, "learning_rate": 1.7479519319721676e-05, "loss": 1.5946, "step": 1579 }, { "epoch": 0.25, "grad_norm": 8.997210502624512, "learning_rate": 1.747605732180993e-05, "loss": 1.7525, "step": 1580 }, { "epoch": 0.25, "grad_norm": 7.777488708496094, "learning_rate": 1.7472593291268535e-05, "loss": 1.823, "step": 1581 }, { "epoch": 0.25, "grad_norm": 7.288871765136719, "learning_rate": 1.746912722903931e-05, "loss": 1.567, "step": 1582 }, { "epoch": 0.25, "grad_norm": 7.847954273223877, "learning_rate": 1.7465659136064634e-05, "loss": 1.8387, "step": 1583 }, { "epoch": 0.25, "grad_norm": 7.747968673706055, "learning_rate": 1.746218901328742e-05, "loss": 1.7347, "step": 1584 }, { "epoch": 0.26, "grad_norm": 8.724864959716797, "learning_rate": 1.7458716861651147e-05, "loss": 1.8102, "step": 1585 }, { "epoch": 0.26, "grad_norm": 9.719925880432129, "learning_rate": 1.7455242682099847e-05, "loss": 1.6469, "step": 1586 }, { "epoch": 0.26, "grad_norm": 11.705299377441406, "learning_rate": 1.745176647557809e-05, "loss": 1.7228, "step": 1587 }, { "epoch": 0.26, "grad_norm": 11.001118659973145, "learning_rate": 1.744828824303101e-05, "loss": 1.828, "step": 1588 }, { "epoch": 0.26, "grad_norm": 8.127153396606445, "learning_rate": 1.7444807985404285e-05, "loss": 1.5067, "step": 1589 }, { "epoch": 0.26, "grad_norm": 8.422666549682617, "learning_rate": 1.7441325703644144e-05, "loss": 1.7303, "step": 1590 }, { "epoch": 0.26, "grad_norm": 9.777226448059082, "learning_rate": 1.7437841398697378e-05, "loss": 1.8935, "step": 1591 }, { "epoch": 0.26, "grad_norm": 7.805936336517334, "learning_rate": 1.7434355071511303e-05, "loss": 1.8577, "step": 1592 }, { "epoch": 0.26, "grad_norm": 6.6895647048950195, "learning_rate": 1.7430866723033812e-05, "loss": 1.6675, "step": 1593 }, { "epoch": 0.26, "grad_norm": 7.444868087768555, "learning_rate": 1.742737635421333e-05, "loss": 1.8491, "step": 1594 }, { "epoch": 0.26, "grad_norm": 8.39752197265625, "learning_rate": 1.7423883965998834e-05, "loss": 1.6183, "step": 1595 }, { "epoch": 0.26, "grad_norm": 7.621873378753662, "learning_rate": 1.7420389559339864e-05, "loss": 1.6267, "step": 1596 }, { "epoch": 0.26, "grad_norm": 8.558167457580566, "learning_rate": 1.7416893135186482e-05, "loss": 1.9692, "step": 1597 }, { "epoch": 0.26, "grad_norm": 7.834481239318848, "learning_rate": 1.741339469448933e-05, "loss": 1.7452, "step": 1598 }, { "epoch": 0.26, "grad_norm": 11.504303932189941, "learning_rate": 1.740989423819957e-05, "loss": 1.8025, "step": 1599 }, { "epoch": 0.26, "grad_norm": 9.49265193939209, "learning_rate": 1.7406391767268927e-05, "loss": 1.7944, "step": 1600 }, { "epoch": 0.26, "grad_norm": 10.120572090148926, "learning_rate": 1.740288728264968e-05, "loss": 1.9109, "step": 1601 }, { "epoch": 0.26, "grad_norm": 6.7907843589782715, "learning_rate": 1.7399380785294634e-05, "loss": 1.7478, "step": 1602 }, { "epoch": 0.26, "grad_norm": 7.249929904937744, "learning_rate": 1.7395872276157163e-05, "loss": 1.8786, "step": 1603 }, { "epoch": 0.26, "grad_norm": 7.960564136505127, "learning_rate": 1.7392361756191174e-05, "loss": 1.6404, "step": 1604 }, { "epoch": 0.26, "grad_norm": 8.161141395568848, "learning_rate": 1.7388849226351126e-05, "loss": 1.587, "step": 1605 }, { "epoch": 0.26, "grad_norm": 7.500371932983398, "learning_rate": 1.7385334687592026e-05, "loss": 1.801, "step": 1606 }, { "epoch": 0.26, "grad_norm": 7.072832107543945, "learning_rate": 1.738181814086943e-05, "loss": 1.6566, "step": 1607 }, { "epoch": 0.26, "grad_norm": 8.892938613891602, "learning_rate": 1.7378299587139424e-05, "loss": 1.7956, "step": 1608 }, { "epoch": 0.26, "grad_norm": 8.231260299682617, "learning_rate": 1.7374779027358654e-05, "loss": 1.6122, "step": 1609 }, { "epoch": 0.26, "grad_norm": 8.123595237731934, "learning_rate": 1.7371256462484316e-05, "loss": 1.7395, "step": 1610 }, { "epoch": 0.26, "grad_norm": 8.563151359558105, "learning_rate": 1.7367731893474132e-05, "loss": 1.9484, "step": 1611 }, { "epoch": 0.26, "grad_norm": 8.84708309173584, "learning_rate": 1.7364205321286393e-05, "loss": 1.6432, "step": 1612 }, { "epoch": 0.26, "grad_norm": 7.8324103355407715, "learning_rate": 1.7360676746879913e-05, "loss": 1.7528, "step": 1613 }, { "epoch": 0.26, "grad_norm": 11.43135929107666, "learning_rate": 1.735714617121406e-05, "loss": 1.7989, "step": 1614 }, { "epoch": 0.26, "grad_norm": 9.191909790039062, "learning_rate": 1.735361359524875e-05, "loss": 1.7208, "step": 1615 }, { "epoch": 0.26, "grad_norm": 7.670129299163818, "learning_rate": 1.7350079019944432e-05, "loss": 1.6323, "step": 1616 }, { "epoch": 0.26, "grad_norm": 11.177983283996582, "learning_rate": 1.734654244626211e-05, "loss": 1.7995, "step": 1617 }, { "epoch": 0.26, "grad_norm": 8.766654014587402, "learning_rate": 1.7343003875163323e-05, "loss": 1.9916, "step": 1618 }, { "epoch": 0.26, "grad_norm": 11.868131637573242, "learning_rate": 1.7339463307610157e-05, "loss": 1.4658, "step": 1619 }, { "epoch": 0.26, "grad_norm": 8.40217113494873, "learning_rate": 1.7335920744565243e-05, "loss": 1.7803, "step": 1620 }, { "epoch": 0.26, "grad_norm": 8.748295783996582, "learning_rate": 1.7332376186991747e-05, "loss": 1.843, "step": 1621 }, { "epoch": 0.26, "grad_norm": 9.556873321533203, "learning_rate": 1.7328829635853388e-05, "loss": 1.5322, "step": 1622 }, { "epoch": 0.26, "grad_norm": 8.309516906738281, "learning_rate": 1.7325281092114413e-05, "loss": 1.769, "step": 1623 }, { "epoch": 0.26, "grad_norm": 11.072903633117676, "learning_rate": 1.7321730556739624e-05, "loss": 1.4221, "step": 1624 }, { "epoch": 0.26, "grad_norm": 7.978371620178223, "learning_rate": 1.731817803069436e-05, "loss": 1.9747, "step": 1625 }, { "epoch": 0.26, "grad_norm": 8.563511848449707, "learning_rate": 1.7314623514944493e-05, "loss": 1.6417, "step": 1626 }, { "epoch": 0.26, "grad_norm": 11.534858703613281, "learning_rate": 1.7311067010456453e-05, "loss": 1.887, "step": 1627 }, { "epoch": 0.26, "grad_norm": 8.187942504882812, "learning_rate": 1.7307508518197195e-05, "loss": 1.9159, "step": 1628 }, { "epoch": 0.26, "grad_norm": 7.344173908233643, "learning_rate": 1.7303948039134223e-05, "loss": 1.844, "step": 1629 }, { "epoch": 0.26, "grad_norm": 9.145931243896484, "learning_rate": 1.7300385574235575e-05, "loss": 1.5591, "step": 1630 }, { "epoch": 0.26, "grad_norm": 7.983305931091309, "learning_rate": 1.729682112446984e-05, "loss": 1.7994, "step": 1631 }, { "epoch": 0.26, "grad_norm": 8.945396423339844, "learning_rate": 1.7293254690806133e-05, "loss": 1.7342, "step": 1632 }, { "epoch": 0.26, "grad_norm": 9.07824993133545, "learning_rate": 1.7289686274214116e-05, "loss": 1.7247, "step": 1633 }, { "epoch": 0.26, "grad_norm": 9.736594200134277, "learning_rate": 1.728611587566399e-05, "loss": 1.5972, "step": 1634 }, { "epoch": 0.26, "grad_norm": 8.846030235290527, "learning_rate": 1.7282543496126493e-05, "loss": 1.8372, "step": 1635 }, { "epoch": 0.26, "grad_norm": 7.530782699584961, "learning_rate": 1.7278969136572903e-05, "loss": 1.6738, "step": 1636 }, { "epoch": 0.26, "grad_norm": 9.456826210021973, "learning_rate": 1.7275392797975034e-05, "loss": 1.9285, "step": 1637 }, { "epoch": 0.26, "grad_norm": 8.476786613464355, "learning_rate": 1.7271814481305243e-05, "loss": 1.8544, "step": 1638 }, { "epoch": 0.26, "grad_norm": 9.578093528747559, "learning_rate": 1.7268234187536413e-05, "loss": 1.7616, "step": 1639 }, { "epoch": 0.26, "grad_norm": 8.133233070373535, "learning_rate": 1.7264651917641983e-05, "loss": 1.6563, "step": 1640 }, { "epoch": 0.26, "grad_norm": 9.321555137634277, "learning_rate": 1.7261067672595917e-05, "loss": 1.8373, "step": 1641 }, { "epoch": 0.26, "grad_norm": 16.07085609436035, "learning_rate": 1.7257481453372715e-05, "loss": 1.8147, "step": 1642 }, { "epoch": 0.26, "grad_norm": 10.066702842712402, "learning_rate": 1.7253893260947417e-05, "loss": 1.719, "step": 1643 }, { "epoch": 0.26, "grad_norm": 7.176656246185303, "learning_rate": 1.7250303096295603e-05, "loss": 1.8101, "step": 1644 }, { "epoch": 0.26, "grad_norm": 11.053303718566895, "learning_rate": 1.7246710960393385e-05, "loss": 1.8316, "step": 1645 }, { "epoch": 0.26, "grad_norm": 10.489372253417969, "learning_rate": 1.724311685421741e-05, "loss": 1.6223, "step": 1646 }, { "epoch": 0.27, "grad_norm": 10.030912399291992, "learning_rate": 1.7239520778744864e-05, "loss": 1.7484, "step": 1647 }, { "epoch": 0.27, "grad_norm": 9.563247680664062, "learning_rate": 1.7235922734953465e-05, "loss": 1.7455, "step": 1648 }, { "epoch": 0.27, "grad_norm": 8.823653221130371, "learning_rate": 1.723232272382147e-05, "loss": 1.898, "step": 1649 }, { "epoch": 0.27, "grad_norm": 8.472105979919434, "learning_rate": 1.722872074632767e-05, "loss": 1.6997, "step": 1650 }, { "epoch": 0.27, "grad_norm": 10.960319519042969, "learning_rate": 1.7225116803451384e-05, "loss": 1.8773, "step": 1651 }, { "epoch": 0.27, "grad_norm": 10.043546676635742, "learning_rate": 1.7221510896172478e-05, "loss": 2.0199, "step": 1652 }, { "epoch": 0.27, "grad_norm": 7.546969413757324, "learning_rate": 1.721790302547134e-05, "loss": 1.6628, "step": 1653 }, { "epoch": 0.27, "grad_norm": 10.268216133117676, "learning_rate": 1.7214293192328896e-05, "loss": 1.7944, "step": 1654 }, { "epoch": 0.27, "grad_norm": 8.41356372833252, "learning_rate": 1.721068139772661e-05, "loss": 1.7041, "step": 1655 }, { "epoch": 0.27, "grad_norm": 9.968941688537598, "learning_rate": 1.7207067642646475e-05, "loss": 1.7817, "step": 1656 }, { "epoch": 0.27, "grad_norm": 9.211338996887207, "learning_rate": 1.7203451928071017e-05, "loss": 1.6333, "step": 1657 }, { "epoch": 0.27, "grad_norm": 8.969942092895508, "learning_rate": 1.719983425498329e-05, "loss": 1.5943, "step": 1658 }, { "epoch": 0.27, "grad_norm": 10.063603401184082, "learning_rate": 1.71962146243669e-05, "loss": 1.7693, "step": 1659 }, { "epoch": 0.27, "grad_norm": 10.514070510864258, "learning_rate": 1.719259303720596e-05, "loss": 2.0175, "step": 1660 }, { "epoch": 0.27, "grad_norm": 10.60177230834961, "learning_rate": 1.7188969494485125e-05, "loss": 1.9496, "step": 1661 }, { "epoch": 0.27, "grad_norm": 7.62278413772583, "learning_rate": 1.718534399718959e-05, "loss": 1.8388, "step": 1662 }, { "epoch": 0.27, "grad_norm": 12.824540138244629, "learning_rate": 1.718171654630507e-05, "loss": 1.7049, "step": 1663 }, { "epoch": 0.27, "grad_norm": 11.342214584350586, "learning_rate": 1.717808714281781e-05, "loss": 1.7403, "step": 1664 }, { "epoch": 0.27, "grad_norm": 11.055103302001953, "learning_rate": 1.7174455787714603e-05, "loss": 1.8482, "step": 1665 }, { "epoch": 0.27, "grad_norm": 9.387517929077148, "learning_rate": 1.7170822481982755e-05, "loss": 1.8947, "step": 1666 }, { "epoch": 0.27, "grad_norm": 12.311555862426758, "learning_rate": 1.7167187226610104e-05, "loss": 1.7004, "step": 1667 }, { "epoch": 0.27, "grad_norm": 13.265679359436035, "learning_rate": 1.716355002258503e-05, "loss": 1.834, "step": 1668 }, { "epoch": 0.27, "grad_norm": 9.705690383911133, "learning_rate": 1.715991087089643e-05, "loss": 1.8189, "step": 1669 }, { "epoch": 0.27, "grad_norm": 7.79991340637207, "learning_rate": 1.715626977253374e-05, "loss": 1.7862, "step": 1670 }, { "epoch": 0.27, "grad_norm": 8.716813087463379, "learning_rate": 1.7152626728486912e-05, "loss": 1.7523, "step": 1671 }, { "epoch": 0.27, "grad_norm": 7.07530403137207, "learning_rate": 1.7148981739746446e-05, "loss": 1.7312, "step": 1672 }, { "epoch": 0.27, "grad_norm": 8.061025619506836, "learning_rate": 1.7145334807303358e-05, "loss": 1.9636, "step": 1673 }, { "epoch": 0.27, "grad_norm": 9.812567710876465, "learning_rate": 1.714168593214919e-05, "loss": 1.7087, "step": 1674 }, { "epoch": 0.27, "grad_norm": 7.226221561431885, "learning_rate": 1.7138035115276027e-05, "loss": 1.9924, "step": 1675 }, { "epoch": 0.27, "grad_norm": 9.580814361572266, "learning_rate": 1.713438235767646e-05, "loss": 1.7623, "step": 1676 }, { "epoch": 0.27, "grad_norm": 8.231114387512207, "learning_rate": 1.7130727660343625e-05, "loss": 1.7584, "step": 1677 }, { "epoch": 0.27, "grad_norm": 9.071943283081055, "learning_rate": 1.712707102427119e-05, "loss": 1.8328, "step": 1678 }, { "epoch": 0.27, "grad_norm": 9.583913803100586, "learning_rate": 1.7123412450453326e-05, "loss": 1.6948, "step": 1679 }, { "epoch": 0.27, "grad_norm": 9.900665283203125, "learning_rate": 1.711975193988475e-05, "loss": 1.7432, "step": 1680 }, { "epoch": 0.27, "grad_norm": 8.52386474609375, "learning_rate": 1.7116089493560705e-05, "loss": 1.8952, "step": 1681 }, { "epoch": 0.27, "grad_norm": 7.934510707855225, "learning_rate": 1.711242511247695e-05, "loss": 1.6124, "step": 1682 }, { "epoch": 0.27, "grad_norm": 8.90293025970459, "learning_rate": 1.7108758797629777e-05, "loss": 1.8019, "step": 1683 }, { "epoch": 0.27, "grad_norm": 8.955644607543945, "learning_rate": 1.710509055001601e-05, "loss": 1.6591, "step": 1684 }, { "epoch": 0.27, "grad_norm": 8.998136520385742, "learning_rate": 1.7101420370632985e-05, "loss": 1.7767, "step": 1685 }, { "epoch": 0.27, "grad_norm": 8.653695106506348, "learning_rate": 1.709774826047857e-05, "loss": 1.5586, "step": 1686 }, { "epoch": 0.27, "grad_norm": 7.344781398773193, "learning_rate": 1.709407422055116e-05, "loss": 1.6694, "step": 1687 }, { "epoch": 0.27, "grad_norm": 7.357583999633789, "learning_rate": 1.7090398251849668e-05, "loss": 1.8847, "step": 1688 }, { "epoch": 0.27, "grad_norm": 9.919429779052734, "learning_rate": 1.7086720355373542e-05, "loss": 1.7182, "step": 1689 }, { "epoch": 0.27, "grad_norm": 11.091747283935547, "learning_rate": 1.7083040532122738e-05, "loss": 1.6673, "step": 1690 }, { "epoch": 0.27, "grad_norm": 8.375865936279297, "learning_rate": 1.7079358783097758e-05, "loss": 1.8673, "step": 1691 }, { "epoch": 0.27, "grad_norm": 8.049495697021484, "learning_rate": 1.707567510929961e-05, "loss": 1.9495, "step": 1692 }, { "epoch": 0.27, "grad_norm": 7.945570945739746, "learning_rate": 1.7071989511729824e-05, "loss": 1.816, "step": 1693 }, { "epoch": 0.27, "grad_norm": 10.571952819824219, "learning_rate": 1.7068301991390465e-05, "loss": 1.637, "step": 1694 }, { "epoch": 0.27, "grad_norm": 7.8968682289123535, "learning_rate": 1.7064612549284117e-05, "loss": 1.6531, "step": 1695 }, { "epoch": 0.27, "grad_norm": 8.753498077392578, "learning_rate": 1.7060921186413885e-05, "loss": 1.9392, "step": 1696 }, { "epoch": 0.27, "grad_norm": 8.78654670715332, "learning_rate": 1.7057227903783396e-05, "loss": 2.0282, "step": 1697 }, { "epoch": 0.27, "grad_norm": 7.073278427124023, "learning_rate": 1.7053532702396792e-05, "loss": 1.6184, "step": 1698 }, { "epoch": 0.27, "grad_norm": 7.502873420715332, "learning_rate": 1.7049835583258752e-05, "loss": 1.5245, "step": 1699 }, { "epoch": 0.27, "grad_norm": 8.9607572555542, "learning_rate": 1.7046136547374467e-05, "loss": 1.6287, "step": 1700 }, { "epoch": 0.27, "grad_norm": 8.083274841308594, "learning_rate": 1.7042435595749646e-05, "loss": 1.814, "step": 1701 }, { "epoch": 0.27, "grad_norm": 8.42752456665039, "learning_rate": 1.7038732729390524e-05, "loss": 1.6098, "step": 1702 }, { "epoch": 0.27, "grad_norm": 9.211321830749512, "learning_rate": 1.703502794930386e-05, "loss": 1.668, "step": 1703 }, { "epoch": 0.27, "grad_norm": 8.76557731628418, "learning_rate": 1.7031321256496926e-05, "loss": 1.8815, "step": 1704 }, { "epoch": 0.27, "grad_norm": 10.792926788330078, "learning_rate": 1.7027612651977518e-05, "loss": 1.8606, "step": 1705 }, { "epoch": 0.27, "grad_norm": 13.260427474975586, "learning_rate": 1.702390213675395e-05, "loss": 1.6416, "step": 1706 }, { "epoch": 0.27, "grad_norm": 8.511242866516113, "learning_rate": 1.7020189711835057e-05, "loss": 1.8417, "step": 1707 }, { "epoch": 0.27, "grad_norm": 7.9208292961120605, "learning_rate": 1.701647537823019e-05, "loss": 1.6901, "step": 1708 }, { "epoch": 0.28, "grad_norm": 11.380436897277832, "learning_rate": 1.7012759136949227e-05, "loss": 1.7941, "step": 1709 }, { "epoch": 0.28, "grad_norm": 9.725756645202637, "learning_rate": 1.7009040989002556e-05, "loss": 1.7091, "step": 1710 }, { "epoch": 0.28, "grad_norm": 8.449630737304688, "learning_rate": 1.7005320935401082e-05, "loss": 1.8245, "step": 1711 }, { "epoch": 0.28, "grad_norm": 9.708877563476562, "learning_rate": 1.700159897715624e-05, "loss": 1.5874, "step": 1712 }, { "epoch": 0.28, "grad_norm": 7.98975944519043, "learning_rate": 1.699787511527997e-05, "loss": 1.8253, "step": 1713 }, { "epoch": 0.28, "grad_norm": 9.884098052978516, "learning_rate": 1.699414935078474e-05, "loss": 1.6312, "step": 1714 }, { "epoch": 0.28, "grad_norm": 9.63912296295166, "learning_rate": 1.6990421684683528e-05, "loss": 1.4099, "step": 1715 }, { "epoch": 0.28, "grad_norm": 9.823244094848633, "learning_rate": 1.6986692117989832e-05, "loss": 1.9272, "step": 1716 }, { "epoch": 0.28, "grad_norm": 7.700746059417725, "learning_rate": 1.6982960651717663e-05, "loss": 1.6889, "step": 1717 }, { "epoch": 0.28, "grad_norm": 7.8306708335876465, "learning_rate": 1.6979227286881557e-05, "loss": 1.7505, "step": 1718 }, { "epoch": 0.28, "grad_norm": 8.864763259887695, "learning_rate": 1.6975492024496557e-05, "loss": 1.7195, "step": 1719 }, { "epoch": 0.28, "grad_norm": 7.513396739959717, "learning_rate": 1.6971754865578225e-05, "loss": 1.7661, "step": 1720 }, { "epoch": 0.28, "grad_norm": 9.28993034362793, "learning_rate": 1.6968015811142645e-05, "loss": 1.738, "step": 1721 }, { "epoch": 0.28, "grad_norm": 8.560503005981445, "learning_rate": 1.6964274862206402e-05, "loss": 1.6649, "step": 1722 }, { "epoch": 0.28, "grad_norm": 10.045729637145996, "learning_rate": 1.696053201978661e-05, "loss": 1.4437, "step": 1723 }, { "epoch": 0.28, "grad_norm": 8.505106925964355, "learning_rate": 1.6956787284900895e-05, "loss": 1.8712, "step": 1724 }, { "epoch": 0.28, "grad_norm": 10.204209327697754, "learning_rate": 1.695304065856739e-05, "loss": 1.8712, "step": 1725 }, { "epoch": 0.28, "grad_norm": 8.393092155456543, "learning_rate": 1.6949292141804752e-05, "loss": 1.9602, "step": 1726 }, { "epoch": 0.28, "grad_norm": 8.881821632385254, "learning_rate": 1.6945541735632144e-05, "loss": 1.7456, "step": 1727 }, { "epoch": 0.28, "grad_norm": 7.598594665527344, "learning_rate": 1.6941789441069247e-05, "loss": 1.8522, "step": 1728 }, { "epoch": 0.28, "grad_norm": 10.343785285949707, "learning_rate": 1.6938035259136254e-05, "loss": 1.7257, "step": 1729 }, { "epoch": 0.28, "grad_norm": 6.624359607696533, "learning_rate": 1.6934279190853874e-05, "loss": 1.6481, "step": 1730 }, { "epoch": 0.28, "grad_norm": 7.564788341522217, "learning_rate": 1.693052123724332e-05, "loss": 1.865, "step": 1731 }, { "epoch": 0.28, "grad_norm": 9.1671142578125, "learning_rate": 1.6926761399326336e-05, "loss": 1.7412, "step": 1732 }, { "epoch": 0.28, "grad_norm": 8.292716026306152, "learning_rate": 1.6922999678125156e-05, "loss": 1.8355, "step": 1733 }, { "epoch": 0.28, "grad_norm": 7.170383453369141, "learning_rate": 1.691923607466254e-05, "loss": 1.9177, "step": 1734 }, { "epoch": 0.28, "grad_norm": 7.867366790771484, "learning_rate": 1.6915470589961757e-05, "loss": 1.7122, "step": 1735 }, { "epoch": 0.28, "grad_norm": 7.115339756011963, "learning_rate": 1.6911703225046582e-05, "loss": 1.7363, "step": 1736 }, { "epoch": 0.28, "grad_norm": 7.937867641448975, "learning_rate": 1.6907933980941312e-05, "loss": 1.649, "step": 1737 }, { "epoch": 0.28, "grad_norm": 12.72653579711914, "learning_rate": 1.690416285867075e-05, "loss": 1.7393, "step": 1738 }, { "epoch": 0.28, "grad_norm": 8.804252624511719, "learning_rate": 1.69003898592602e-05, "loss": 1.8646, "step": 1739 }, { "epoch": 0.28, "grad_norm": 11.693686485290527, "learning_rate": 1.6896614983735495e-05, "loss": 1.5781, "step": 1740 }, { "epoch": 0.28, "grad_norm": 8.505775451660156, "learning_rate": 1.6892838233122958e-05, "loss": 1.712, "step": 1741 }, { "epoch": 0.28, "grad_norm": 8.927864074707031, "learning_rate": 1.688905960844944e-05, "loss": 1.7336, "step": 1742 }, { "epoch": 0.28, "grad_norm": 9.791263580322266, "learning_rate": 1.688527911074229e-05, "loss": 1.8481, "step": 1743 }, { "epoch": 0.28, "grad_norm": 10.220260620117188, "learning_rate": 1.688149674102937e-05, "loss": 1.8832, "step": 1744 }, { "epoch": 0.28, "grad_norm": 8.914925575256348, "learning_rate": 1.6877712500339044e-05, "loss": 1.819, "step": 1745 }, { "epoch": 0.28, "grad_norm": 8.080343246459961, "learning_rate": 1.68739263897002e-05, "loss": 1.8469, "step": 1746 }, { "epoch": 0.28, "grad_norm": 11.160319328308105, "learning_rate": 1.6870138410142227e-05, "loss": 1.6365, "step": 1747 }, { "epoch": 0.28, "grad_norm": 10.987247467041016, "learning_rate": 1.6866348562695015e-05, "loss": 1.5877, "step": 1748 }, { "epoch": 0.28, "grad_norm": 6.676432132720947, "learning_rate": 1.686255684838897e-05, "loss": 1.4993, "step": 1749 }, { "epoch": 0.28, "grad_norm": 13.783452033996582, "learning_rate": 1.6858763268254996e-05, "loss": 1.8999, "step": 1750 }, { "epoch": 0.28, "grad_norm": 8.845227241516113, "learning_rate": 1.6854967823324522e-05, "loss": 1.6778, "step": 1751 }, { "epoch": 0.28, "grad_norm": 8.03012752532959, "learning_rate": 1.685117051462947e-05, "loss": 1.6238, "step": 1752 }, { "epoch": 0.28, "grad_norm": 10.172728538513184, "learning_rate": 1.684737134320227e-05, "loss": 1.9309, "step": 1753 }, { "epoch": 0.28, "grad_norm": 7.727676868438721, "learning_rate": 1.6843570310075862e-05, "loss": 1.8016, "step": 1754 }, { "epoch": 0.28, "grad_norm": 7.744712829589844, "learning_rate": 1.683976741628369e-05, "loss": 1.9645, "step": 1755 }, { "epoch": 0.28, "grad_norm": 7.260523796081543, "learning_rate": 1.683596266285971e-05, "loss": 1.6748, "step": 1756 }, { "epoch": 0.28, "grad_norm": 7.817650318145752, "learning_rate": 1.6832156050838366e-05, "loss": 1.8583, "step": 1757 }, { "epoch": 0.28, "grad_norm": 7.301516532897949, "learning_rate": 1.6828347581254633e-05, "loss": 1.8007, "step": 1758 }, { "epoch": 0.28, "grad_norm": 9.454742431640625, "learning_rate": 1.682453725514397e-05, "loss": 1.8574, "step": 1759 }, { "epoch": 0.28, "grad_norm": 9.29407787322998, "learning_rate": 1.682072507354235e-05, "loss": 1.7157, "step": 1760 }, { "epoch": 0.28, "grad_norm": 7.8621506690979, "learning_rate": 1.681691103748625e-05, "loss": 1.714, "step": 1761 }, { "epoch": 0.28, "grad_norm": 11.239619255065918, "learning_rate": 1.681309514801265e-05, "loss": 1.7653, "step": 1762 }, { "epoch": 0.28, "grad_norm": 10.159780502319336, "learning_rate": 1.6809277406159036e-05, "loss": 1.8948, "step": 1763 }, { "epoch": 0.28, "grad_norm": 10.253511428833008, "learning_rate": 1.680545781296339e-05, "loss": 1.716, "step": 1764 }, { "epoch": 0.28, "grad_norm": 8.099055290222168, "learning_rate": 1.6801636369464208e-05, "loss": 1.8062, "step": 1765 }, { "epoch": 0.28, "grad_norm": 10.813438415527344, "learning_rate": 1.679781307670048e-05, "loss": 1.6198, "step": 1766 }, { "epoch": 0.28, "grad_norm": 11.306281089782715, "learning_rate": 1.6793987935711706e-05, "loss": 1.7324, "step": 1767 }, { "epoch": 0.28, "grad_norm": 9.504295349121094, "learning_rate": 1.6790160947537887e-05, "loss": 1.7337, "step": 1768 }, { "epoch": 0.28, "grad_norm": 7.502370357513428, "learning_rate": 1.6786332113219523e-05, "loss": 1.7141, "step": 1769 }, { "epoch": 0.28, "grad_norm": 9.225921630859375, "learning_rate": 1.678250143379761e-05, "loss": 1.8671, "step": 1770 }, { "epoch": 0.29, "grad_norm": 12.905987739562988, "learning_rate": 1.6778668910313666e-05, "loss": 1.6996, "step": 1771 }, { "epoch": 0.29, "grad_norm": 13.157349586486816, "learning_rate": 1.6774834543809688e-05, "loss": 1.7326, "step": 1772 }, { "epoch": 0.29, "grad_norm": 8.624484062194824, "learning_rate": 1.6770998335328188e-05, "loss": 1.6501, "step": 1773 }, { "epoch": 0.29, "grad_norm": 6.517758846282959, "learning_rate": 1.6767160285912175e-05, "loss": 1.5783, "step": 1774 }, { "epoch": 0.29, "grad_norm": 8.907115936279297, "learning_rate": 1.6763320396605157e-05, "loss": 1.7392, "step": 1775 }, { "epoch": 0.29, "grad_norm": 9.995135307312012, "learning_rate": 1.6759478668451142e-05, "loss": 1.6275, "step": 1776 }, { "epoch": 0.29, "grad_norm": 8.091658592224121, "learning_rate": 1.6755635102494637e-05, "loss": 1.9574, "step": 1777 }, { "epoch": 0.29, "grad_norm": 8.759628295898438, "learning_rate": 1.6751789699780658e-05, "loss": 1.7152, "step": 1778 }, { "epoch": 0.29, "grad_norm": 9.095915794372559, "learning_rate": 1.6747942461354707e-05, "loss": 1.6655, "step": 1779 }, { "epoch": 0.29, "grad_norm": 10.139488220214844, "learning_rate": 1.6744093388262795e-05, "loss": 1.7474, "step": 1780 }, { "epoch": 0.29, "grad_norm": 8.83455753326416, "learning_rate": 1.6740242481551425e-05, "loss": 1.7526, "step": 1781 }, { "epoch": 0.29, "grad_norm": 8.608070373535156, "learning_rate": 1.6736389742267604e-05, "loss": 1.6027, "step": 1782 }, { "epoch": 0.29, "grad_norm": 7.036437034606934, "learning_rate": 1.6732535171458834e-05, "loss": 1.7151, "step": 1783 }, { "epoch": 0.29, "grad_norm": 7.847032070159912, "learning_rate": 1.6728678770173118e-05, "loss": 1.565, "step": 1784 }, { "epoch": 0.29, "grad_norm": 7.500700950622559, "learning_rate": 1.672482053945895e-05, "loss": 1.892, "step": 1785 }, { "epoch": 0.29, "grad_norm": 8.68311595916748, "learning_rate": 1.6720960480365332e-05, "loss": 1.6062, "step": 1786 }, { "epoch": 0.29, "grad_norm": 7.882802963256836, "learning_rate": 1.6717098593941753e-05, "loss": 1.6704, "step": 1787 }, { "epoch": 0.29, "grad_norm": 10.597650527954102, "learning_rate": 1.6713234881238202e-05, "loss": 1.6246, "step": 1788 }, { "epoch": 0.29, "grad_norm": 8.139127731323242, "learning_rate": 1.670936934330517e-05, "loss": 1.6158, "step": 1789 }, { "epoch": 0.29, "grad_norm": 6.728243350982666, "learning_rate": 1.6705501981193632e-05, "loss": 1.5093, "step": 1790 }, { "epoch": 0.29, "grad_norm": 7.786482810974121, "learning_rate": 1.6701632795955075e-05, "loss": 1.6651, "step": 1791 }, { "epoch": 0.29, "grad_norm": 9.615945816040039, "learning_rate": 1.6697761788641472e-05, "loss": 1.7172, "step": 1792 }, { "epoch": 0.29, "grad_norm": 9.514617919921875, "learning_rate": 1.6693888960305288e-05, "loss": 1.4393, "step": 1793 }, { "epoch": 0.29, "grad_norm": 8.587552070617676, "learning_rate": 1.669001431199949e-05, "loss": 1.8137, "step": 1794 }, { "epoch": 0.29, "grad_norm": 7.2962565422058105, "learning_rate": 1.668613784477754e-05, "loss": 1.8194, "step": 1795 }, { "epoch": 0.29, "grad_norm": 8.42656421661377, "learning_rate": 1.668225955969339e-05, "loss": 1.988, "step": 1796 }, { "epoch": 0.29, "grad_norm": 8.62259292602539, "learning_rate": 1.667837945780149e-05, "loss": 2.0877, "step": 1797 }, { "epoch": 0.29, "grad_norm": 7.914419651031494, "learning_rate": 1.667449754015678e-05, "loss": 1.9161, "step": 1798 }, { "epoch": 0.29, "grad_norm": 7.520977020263672, "learning_rate": 1.6670613807814697e-05, "loss": 1.877, "step": 1799 }, { "epoch": 0.29, "grad_norm": 7.657093048095703, "learning_rate": 1.6666728261831173e-05, "loss": 1.8901, "step": 1800 }, { "epoch": 0.29, "grad_norm": 10.645879745483398, "learning_rate": 1.6662840903262628e-05, "loss": 1.6806, "step": 1801 }, { "epoch": 0.29, "grad_norm": 7.532654285430908, "learning_rate": 1.665895173316598e-05, "loss": 1.6229, "step": 1802 }, { "epoch": 0.29, "grad_norm": 9.514966011047363, "learning_rate": 1.665506075259863e-05, "loss": 1.7064, "step": 1803 }, { "epoch": 0.29, "grad_norm": 7.777235507965088, "learning_rate": 1.6651167962618488e-05, "loss": 1.6141, "step": 1804 }, { "epoch": 0.29, "grad_norm": 7.956114292144775, "learning_rate": 1.664727336428394e-05, "loss": 1.7849, "step": 1805 }, { "epoch": 0.29, "grad_norm": 6.3481268882751465, "learning_rate": 1.664337695865387e-05, "loss": 1.5969, "step": 1806 }, { "epoch": 0.29, "grad_norm": 9.010835647583008, "learning_rate": 1.6639478746787654e-05, "loss": 1.9867, "step": 1807 }, { "epoch": 0.29, "grad_norm": 9.027135848999023, "learning_rate": 1.663557872974516e-05, "loss": 1.63, "step": 1808 }, { "epoch": 0.29, "grad_norm": 8.288732528686523, "learning_rate": 1.6631676908586743e-05, "loss": 1.6431, "step": 1809 }, { "epoch": 0.29, "grad_norm": 9.603647232055664, "learning_rate": 1.6627773284373253e-05, "loss": 1.8451, "step": 1810 }, { "epoch": 0.29, "grad_norm": 7.337112903594971, "learning_rate": 1.6623867858166028e-05, "loss": 1.6945, "step": 1811 }, { "epoch": 0.29, "grad_norm": 7.582287311553955, "learning_rate": 1.661996063102689e-05, "loss": 2.0918, "step": 1812 }, { "epoch": 0.29, "grad_norm": 7.9576544761657715, "learning_rate": 1.6616051604018162e-05, "loss": 1.579, "step": 1813 }, { "epoch": 0.29, "grad_norm": 8.19168472290039, "learning_rate": 1.661214077820265e-05, "loss": 1.8993, "step": 1814 }, { "epoch": 0.29, "grad_norm": 7.4783453941345215, "learning_rate": 1.6608228154643643e-05, "loss": 1.6586, "step": 1815 }, { "epoch": 0.29, "grad_norm": 8.501970291137695, "learning_rate": 1.6604313734404936e-05, "loss": 1.8799, "step": 1816 }, { "epoch": 0.29, "grad_norm": 8.2039213180542, "learning_rate": 1.66003975185508e-05, "loss": 1.8788, "step": 1817 }, { "epoch": 0.29, "grad_norm": 9.312494277954102, "learning_rate": 1.6596479508145987e-05, "loss": 1.7498, "step": 1818 }, { "epoch": 0.29, "grad_norm": 7.85659646987915, "learning_rate": 1.6592559704255758e-05, "loss": 1.5047, "step": 1819 }, { "epoch": 0.29, "grad_norm": 8.060044288635254, "learning_rate": 1.658863810794584e-05, "loss": 1.7054, "step": 1820 }, { "epoch": 0.29, "grad_norm": 7.382716178894043, "learning_rate": 1.658471472028247e-05, "loss": 1.7056, "step": 1821 }, { "epoch": 0.29, "grad_norm": 9.838153839111328, "learning_rate": 1.6580789542332345e-05, "loss": 1.6306, "step": 1822 }, { "epoch": 0.29, "grad_norm": 8.996524810791016, "learning_rate": 1.6576862575162673e-05, "loss": 1.8272, "step": 1823 }, { "epoch": 0.29, "grad_norm": 8.404590606689453, "learning_rate": 1.6572933819841132e-05, "loss": 1.5878, "step": 1824 }, { "epoch": 0.29, "grad_norm": 8.104220390319824, "learning_rate": 1.6569003277435896e-05, "loss": 1.5736, "step": 1825 }, { "epoch": 0.29, "grad_norm": 7.947280406951904, "learning_rate": 1.656507094901562e-05, "loss": 1.7293, "step": 1826 }, { "epoch": 0.29, "grad_norm": 9.748902320861816, "learning_rate": 1.6561136835649446e-05, "loss": 1.4349, "step": 1827 }, { "epoch": 0.29, "grad_norm": 6.779302597045898, "learning_rate": 1.6557200938407004e-05, "loss": 1.7166, "step": 1828 }, { "epoch": 0.29, "grad_norm": 10.25810718536377, "learning_rate": 1.6553263258358403e-05, "loss": 1.875, "step": 1829 }, { "epoch": 0.29, "grad_norm": 7.448399543762207, "learning_rate": 1.6549323796574243e-05, "loss": 1.8626, "step": 1830 }, { "epoch": 0.29, "grad_norm": 7.590145111083984, "learning_rate": 1.6545382554125603e-05, "loss": 1.5571, "step": 1831 }, { "epoch": 0.29, "grad_norm": 11.512560844421387, "learning_rate": 1.6541439532084048e-05, "loss": 1.9449, "step": 1832 }, { "epoch": 0.3, "grad_norm": 8.605287551879883, "learning_rate": 1.6537494731521633e-05, "loss": 1.5894, "step": 1833 }, { "epoch": 0.3, "grad_norm": 6.7282538414001465, "learning_rate": 1.6533548153510883e-05, "loss": 1.6053, "step": 1834 }, { "epoch": 0.3, "grad_norm": 7.649819374084473, "learning_rate": 1.652959979912482e-05, "loss": 1.5681, "step": 1835 }, { "epoch": 0.3, "grad_norm": 7.744753837585449, "learning_rate": 1.6525649669436942e-05, "loss": 1.3806, "step": 1836 }, { "epoch": 0.3, "grad_norm": 6.904871463775635, "learning_rate": 1.6521697765521232e-05, "loss": 1.7146, "step": 1837 }, { "epoch": 0.3, "grad_norm": 11.696518898010254, "learning_rate": 1.6517744088452154e-05, "loss": 1.6828, "step": 1838 }, { "epoch": 0.3, "grad_norm": 10.703245162963867, "learning_rate": 1.6513788639304654e-05, "loss": 1.6781, "step": 1839 }, { "epoch": 0.3, "grad_norm": 8.03940200805664, "learning_rate": 1.6509831419154156e-05, "loss": 1.8481, "step": 1840 }, { "epoch": 0.3, "grad_norm": 7.887338638305664, "learning_rate": 1.6505872429076575e-05, "loss": 1.7148, "step": 1841 }, { "epoch": 0.3, "grad_norm": 12.895943641662598, "learning_rate": 1.65019116701483e-05, "loss": 1.4896, "step": 1842 }, { "epoch": 0.3, "grad_norm": 9.552122116088867, "learning_rate": 1.6497949143446206e-05, "loss": 1.7752, "step": 1843 }, { "epoch": 0.3, "grad_norm": 7.523780822753906, "learning_rate": 1.649398485004764e-05, "loss": 1.8601, "step": 1844 }, { "epoch": 0.3, "grad_norm": 8.06678295135498, "learning_rate": 1.649001879103044e-05, "loss": 1.7107, "step": 1845 }, { "epoch": 0.3, "grad_norm": 8.691240310668945, "learning_rate": 1.6486050967472916e-05, "loss": 1.535, "step": 1846 }, { "epoch": 0.3, "grad_norm": 10.462372779846191, "learning_rate": 1.6482081380453858e-05, "loss": 1.6724, "step": 1847 }, { "epoch": 0.3, "grad_norm": 9.114336967468262, "learning_rate": 1.6478110031052548e-05, "loss": 1.8508, "step": 1848 }, { "epoch": 0.3, "grad_norm": 8.40785026550293, "learning_rate": 1.6474136920348728e-05, "loss": 1.8032, "step": 1849 }, { "epoch": 0.3, "grad_norm": 9.619051933288574, "learning_rate": 1.6470162049422627e-05, "loss": 1.6932, "step": 1850 }, { "epoch": 0.3, "grad_norm": 7.288489818572998, "learning_rate": 1.6466185419354964e-05, "loss": 1.6513, "step": 1851 }, { "epoch": 0.3, "grad_norm": 8.341829299926758, "learning_rate": 1.646220703122692e-05, "loss": 1.8015, "step": 1852 }, { "epoch": 0.3, "grad_norm": 7.930058479309082, "learning_rate": 1.6458226886120156e-05, "loss": 1.7866, "step": 1853 }, { "epoch": 0.3, "grad_norm": 8.598238945007324, "learning_rate": 1.6454244985116817e-05, "loss": 1.7067, "step": 1854 }, { "epoch": 0.3, "grad_norm": 9.12668228149414, "learning_rate": 1.6450261329299526e-05, "loss": 1.9134, "step": 1855 }, { "epoch": 0.3, "grad_norm": 13.238183975219727, "learning_rate": 1.644627591975138e-05, "loss": 1.7936, "step": 1856 }, { "epoch": 0.3, "grad_norm": 8.147065162658691, "learning_rate": 1.644228875755595e-05, "loss": 1.6722, "step": 1857 }, { "epoch": 0.3, "grad_norm": 9.224235534667969, "learning_rate": 1.643829984379729e-05, "loss": 1.5866, "step": 1858 }, { "epoch": 0.3, "grad_norm": 7.937987804412842, "learning_rate": 1.6434309179559927e-05, "loss": 1.8183, "step": 1859 }, { "epoch": 0.3, "grad_norm": 8.635107040405273, "learning_rate": 1.643031676592886e-05, "loss": 1.5803, "step": 1860 }, { "epoch": 0.3, "grad_norm": 8.70404052734375, "learning_rate": 1.642632260398957e-05, "loss": 1.8134, "step": 1861 }, { "epoch": 0.3, "grad_norm": 11.86419677734375, "learning_rate": 1.642232669482801e-05, "loss": 1.7977, "step": 1862 }, { "epoch": 0.3, "grad_norm": 8.82665729522705, "learning_rate": 1.6418329039530605e-05, "loss": 1.7837, "step": 1863 }, { "epoch": 0.3, "grad_norm": 20.981138229370117, "learning_rate": 1.6414329639184265e-05, "loss": 1.5042, "step": 1864 }, { "epoch": 0.3, "grad_norm": 11.983558654785156, "learning_rate": 1.6410328494876368e-05, "loss": 1.6579, "step": 1865 }, { "epoch": 0.3, "grad_norm": 7.375823497772217, "learning_rate": 1.640632560769476e-05, "loss": 1.8273, "step": 1866 }, { "epoch": 0.3, "grad_norm": 9.121716499328613, "learning_rate": 1.640232097872777e-05, "loss": 1.9729, "step": 1867 }, { "epoch": 0.3, "grad_norm": 6.783158302307129, "learning_rate": 1.6398314609064194e-05, "loss": 1.6164, "step": 1868 }, { "epoch": 0.3, "grad_norm": 10.430986404418945, "learning_rate": 1.6394306499793315e-05, "loss": 1.9002, "step": 1869 }, { "epoch": 0.3, "grad_norm": 10.690946578979492, "learning_rate": 1.639029665200487e-05, "loss": 1.5412, "step": 1870 }, { "epoch": 0.3, "grad_norm": 8.904211044311523, "learning_rate": 1.6386285066789075e-05, "loss": 1.7001, "step": 1871 }, { "epoch": 0.3, "grad_norm": 7.2518744468688965, "learning_rate": 1.638227174523663e-05, "loss": 1.5865, "step": 1872 }, { "epoch": 0.3, "grad_norm": 7.683706283569336, "learning_rate": 1.637825668843869e-05, "loss": 1.8614, "step": 1873 }, { "epoch": 0.3, "grad_norm": 7.809867858886719, "learning_rate": 1.63742398974869e-05, "loss": 1.5678, "step": 1874 }, { "epoch": 0.3, "grad_norm": 8.159370422363281, "learning_rate": 1.6370221373473352e-05, "loss": 1.961, "step": 1875 }, { "epoch": 0.3, "grad_norm": 7.128573894500732, "learning_rate": 1.6366201117490634e-05, "loss": 1.6812, "step": 1876 }, { "epoch": 0.3, "grad_norm": 9.28414249420166, "learning_rate": 1.636217913063179e-05, "loss": 1.6502, "step": 1877 }, { "epoch": 0.3, "grad_norm": 7.9496588706970215, "learning_rate": 1.6358155413990337e-05, "loss": 1.6285, "step": 1878 }, { "epoch": 0.3, "grad_norm": 11.191554069519043, "learning_rate": 1.635412996866027e-05, "loss": 1.8272, "step": 1879 }, { "epoch": 0.3, "grad_norm": 10.769219398498535, "learning_rate": 1.6350102795736046e-05, "loss": 1.8802, "step": 1880 }, { "epoch": 0.3, "grad_norm": 7.6863813400268555, "learning_rate": 1.634607389631259e-05, "loss": 1.6182, "step": 1881 }, { "epoch": 0.3, "grad_norm": 8.051215171813965, "learning_rate": 1.6342043271485306e-05, "loss": 1.7061, "step": 1882 }, { "epoch": 0.3, "grad_norm": 8.05473804473877, "learning_rate": 1.6338010922350056e-05, "loss": 1.7072, "step": 1883 }, { "epoch": 0.3, "grad_norm": 12.51197624206543, "learning_rate": 1.6333976850003178e-05, "loss": 1.7469, "step": 1884 }, { "epoch": 0.3, "grad_norm": 9.999958038330078, "learning_rate": 1.6329941055541478e-05, "loss": 1.8974, "step": 1885 }, { "epoch": 0.3, "grad_norm": 7.949757099151611, "learning_rate": 1.632590354006223e-05, "loss": 1.8337, "step": 1886 }, { "epoch": 0.3, "grad_norm": 8.959297180175781, "learning_rate": 1.6321864304663174e-05, "loss": 1.6876, "step": 1887 }, { "epoch": 0.3, "grad_norm": 10.441591262817383, "learning_rate": 1.6317823350442515e-05, "loss": 1.7074, "step": 1888 }, { "epoch": 0.3, "grad_norm": 9.538839340209961, "learning_rate": 1.6313780678498932e-05, "loss": 1.5964, "step": 1889 }, { "epoch": 0.3, "grad_norm": 9.627578735351562, "learning_rate": 1.630973628993157e-05, "loss": 1.827, "step": 1890 }, { "epoch": 0.3, "grad_norm": 9.034186363220215, "learning_rate": 1.6305690185840036e-05, "loss": 1.7906, "step": 1891 }, { "epoch": 0.3, "grad_norm": 8.69048023223877, "learning_rate": 1.6301642367324404e-05, "loss": 1.6369, "step": 1892 }, { "epoch": 0.3, "grad_norm": 12.427610397338867, "learning_rate": 1.6297592835485218e-05, "loss": 1.9184, "step": 1893 }, { "epoch": 0.3, "grad_norm": 11.033987045288086, "learning_rate": 1.6293541591423484e-05, "loss": 1.785, "step": 1894 }, { "epoch": 0.31, "grad_norm": 9.422246932983398, "learning_rate": 1.628948863624068e-05, "loss": 1.7842, "step": 1895 }, { "epoch": 0.31, "grad_norm": 7.918975353240967, "learning_rate": 1.6285433971038745e-05, "loss": 1.8812, "step": 1896 }, { "epoch": 0.31, "grad_norm": 10.754992485046387, "learning_rate": 1.628137759692008e-05, "loss": 1.7599, "step": 1897 }, { "epoch": 0.31, "grad_norm": 9.187499046325684, "learning_rate": 1.6277319514987548e-05, "loss": 1.7902, "step": 1898 }, { "epoch": 0.31, "grad_norm": 7.5873122215271, "learning_rate": 1.6273259726344488e-05, "loss": 1.6906, "step": 1899 }, { "epoch": 0.31, "grad_norm": 9.875038146972656, "learning_rate": 1.6269198232094696e-05, "loss": 1.8517, "step": 1900 }, { "epoch": 0.31, "grad_norm": 10.248043060302734, "learning_rate": 1.6265135033342432e-05, "loss": 1.7109, "step": 1901 }, { "epoch": 0.31, "grad_norm": 9.84097957611084, "learning_rate": 1.626107013119242e-05, "loss": 1.707, "step": 1902 }, { "epoch": 0.31, "grad_norm": 8.465648651123047, "learning_rate": 1.6257003526749846e-05, "loss": 1.7617, "step": 1903 }, { "epoch": 0.31, "grad_norm": 9.61074161529541, "learning_rate": 1.625293522112036e-05, "loss": 1.8447, "step": 1904 }, { "epoch": 0.31, "grad_norm": 9.137978553771973, "learning_rate": 1.624886521541007e-05, "loss": 1.8062, "step": 1905 }, { "epoch": 0.31, "grad_norm": 9.069786071777344, "learning_rate": 1.6244793510725564e-05, "loss": 1.714, "step": 1906 }, { "epoch": 0.31, "grad_norm": 7.5394134521484375, "learning_rate": 1.6240720108173863e-05, "loss": 1.8178, "step": 1907 }, { "epoch": 0.31, "grad_norm": 7.239002227783203, "learning_rate": 1.623664500886247e-05, "loss": 1.6114, "step": 1908 }, { "epoch": 0.31, "grad_norm": 7.820661544799805, "learning_rate": 1.623256821389935e-05, "loss": 1.543, "step": 1909 }, { "epoch": 0.31, "grad_norm": 7.916224956512451, "learning_rate": 1.6228489724392916e-05, "loss": 1.806, "step": 1910 }, { "epoch": 0.31, "grad_norm": 8.768199920654297, "learning_rate": 1.6224409541452054e-05, "loss": 1.6703, "step": 1911 }, { "epoch": 0.31, "grad_norm": 8.388323783874512, "learning_rate": 1.62203276661861e-05, "loss": 1.7094, "step": 1912 }, { "epoch": 0.31, "grad_norm": 9.090232849121094, "learning_rate": 1.6216244099704864e-05, "loss": 1.8171, "step": 1913 }, { "epoch": 0.31, "grad_norm": 6.151561260223389, "learning_rate": 1.6212158843118597e-05, "loss": 1.6822, "step": 1914 }, { "epoch": 0.31, "grad_norm": 9.590721130371094, "learning_rate": 1.620807189753803e-05, "loss": 1.624, "step": 1915 }, { "epoch": 0.31, "grad_norm": 14.991259574890137, "learning_rate": 1.620398326407434e-05, "loss": 1.9823, "step": 1916 }, { "epoch": 0.31, "grad_norm": 10.706489562988281, "learning_rate": 1.6199892943839162e-05, "loss": 1.559, "step": 1917 }, { "epoch": 0.31, "grad_norm": 8.805658340454102, "learning_rate": 1.6195800937944595e-05, "loss": 1.6086, "step": 1918 }, { "epoch": 0.31, "grad_norm": 12.179167747497559, "learning_rate": 1.61917072475032e-05, "loss": 1.8817, "step": 1919 }, { "epoch": 0.31, "grad_norm": 12.604114532470703, "learning_rate": 1.6187611873627987e-05, "loss": 1.6821, "step": 1920 }, { "epoch": 0.31, "grad_norm": 9.07976245880127, "learning_rate": 1.618351481743243e-05, "loss": 1.7044, "step": 1921 }, { "epoch": 0.31, "grad_norm": 7.645601272583008, "learning_rate": 1.6179416080030455e-05, "loss": 1.821, "step": 1922 }, { "epoch": 0.31, "grad_norm": 7.508134841918945, "learning_rate": 1.6175315662536447e-05, "loss": 1.8741, "step": 1923 }, { "epoch": 0.31, "grad_norm": 7.417752265930176, "learning_rate": 1.6171213566065255e-05, "loss": 1.6169, "step": 1924 }, { "epoch": 0.31, "grad_norm": 9.294188499450684, "learning_rate": 1.6167109791732173e-05, "loss": 1.5797, "step": 1925 }, { "epoch": 0.31, "grad_norm": 10.831299781799316, "learning_rate": 1.6163004340652957e-05, "loss": 1.5462, "step": 1926 }, { "epoch": 0.31, "grad_norm": 7.520143508911133, "learning_rate": 1.6158897213943826e-05, "loss": 1.8411, "step": 1927 }, { "epoch": 0.31, "grad_norm": 9.883987426757812, "learning_rate": 1.615478841272144e-05, "loss": 1.7154, "step": 1928 }, { "epoch": 0.31, "grad_norm": 8.657779693603516, "learning_rate": 1.6150677938102923e-05, "loss": 1.904, "step": 1929 }, { "epoch": 0.31, "grad_norm": 9.880422592163086, "learning_rate": 1.614656579120585e-05, "loss": 1.794, "step": 1930 }, { "epoch": 0.31, "grad_norm": 9.753754615783691, "learning_rate": 1.6142451973148255e-05, "loss": 1.6016, "step": 1931 }, { "epoch": 0.31, "grad_norm": 6.839356899261475, "learning_rate": 1.6138336485048626e-05, "loss": 1.7354, "step": 1932 }, { "epoch": 0.31, "grad_norm": 7.7022528648376465, "learning_rate": 1.6134219328025903e-05, "loss": 1.5885, "step": 1933 }, { "epoch": 0.31, "grad_norm": 7.754483699798584, "learning_rate": 1.613010050319948e-05, "loss": 1.7542, "step": 1934 }, { "epoch": 0.31, "grad_norm": 10.205183029174805, "learning_rate": 1.6125980011689203e-05, "loss": 1.9228, "step": 1935 }, { "epoch": 0.31, "grad_norm": 7.293696880340576, "learning_rate": 1.6121857854615374e-05, "loss": 1.6289, "step": 1936 }, { "epoch": 0.31, "grad_norm": 8.676684379577637, "learning_rate": 1.6117734033098744e-05, "loss": 1.687, "step": 1937 }, { "epoch": 0.31, "grad_norm": 7.279211521148682, "learning_rate": 1.6113608548260526e-05, "loss": 1.9035, "step": 1938 }, { "epoch": 0.31, "grad_norm": 7.96088981628418, "learning_rate": 1.6109481401222377e-05, "loss": 1.7022, "step": 1939 }, { "epoch": 0.31, "grad_norm": 8.226861953735352, "learning_rate": 1.6105352593106397e-05, "loss": 1.9127, "step": 1940 }, { "epoch": 0.31, "grad_norm": 7.2622809410095215, "learning_rate": 1.610122212503516e-05, "loss": 1.7875, "step": 1941 }, { "epoch": 0.31, "grad_norm": 8.813976287841797, "learning_rate": 1.6097089998131674e-05, "loss": 1.9056, "step": 1942 }, { "epoch": 0.31, "grad_norm": 8.967574119567871, "learning_rate": 1.6092956213519408e-05, "loss": 1.671, "step": 1943 }, { "epoch": 0.31, "grad_norm": 7.548727512359619, "learning_rate": 1.608882077232227e-05, "loss": 1.8138, "step": 1944 }, { "epoch": 0.31, "grad_norm": 7.44279146194458, "learning_rate": 1.6084683675664625e-05, "loss": 1.7144, "step": 1945 }, { "epoch": 0.31, "grad_norm": 8.737553596496582, "learning_rate": 1.6080544924671296e-05, "loss": 1.6288, "step": 1946 }, { "epoch": 0.31, "grad_norm": 9.70362377166748, "learning_rate": 1.6076404520467548e-05, "loss": 1.6652, "step": 1947 }, { "epoch": 0.31, "grad_norm": 8.973271369934082, "learning_rate": 1.6072262464179088e-05, "loss": 1.6323, "step": 1948 }, { "epoch": 0.31, "grad_norm": 8.615445137023926, "learning_rate": 1.6068118756932086e-05, "loss": 1.725, "step": 1949 }, { "epoch": 0.31, "grad_norm": 10.184289932250977, "learning_rate": 1.606397339985315e-05, "loss": 1.4799, "step": 1950 }, { "epoch": 0.31, "grad_norm": 8.336980819702148, "learning_rate": 1.6059826394069352e-05, "loss": 1.8205, "step": 1951 }, { "epoch": 0.31, "grad_norm": 8.228729248046875, "learning_rate": 1.6055677740708188e-05, "loss": 1.6335, "step": 1952 }, { "epoch": 0.31, "grad_norm": 9.560572624206543, "learning_rate": 1.6051527440897625e-05, "loss": 1.9183, "step": 1953 }, { "epoch": 0.31, "grad_norm": 7.496698379516602, "learning_rate": 1.6047375495766063e-05, "loss": 1.8228, "step": 1954 }, { "epoch": 0.31, "grad_norm": 7.897149085998535, "learning_rate": 1.6043221906442354e-05, "loss": 1.7107, "step": 1955 }, { "epoch": 0.31, "grad_norm": 8.542627334594727, "learning_rate": 1.6039066674055806e-05, "loss": 1.6742, "step": 1956 }, { "epoch": 0.32, "grad_norm": 7.678966999053955, "learning_rate": 1.6034909799736155e-05, "loss": 1.6748, "step": 1957 }, { "epoch": 0.32, "grad_norm": 7.27999210357666, "learning_rate": 1.6030751284613602e-05, "loss": 1.8126, "step": 1958 }, { "epoch": 0.32, "grad_norm": 7.1684770584106445, "learning_rate": 1.6026591129818782e-05, "loss": 1.7496, "step": 1959 }, { "epoch": 0.32, "grad_norm": 8.226146697998047, "learning_rate": 1.602242933648278e-05, "loss": 1.7866, "step": 1960 }, { "epoch": 0.32, "grad_norm": 7.492559432983398, "learning_rate": 1.6018265905737128e-05, "loss": 1.7278, "step": 1961 }, { "epoch": 0.32, "grad_norm": 8.061422348022461, "learning_rate": 1.6014100838713796e-05, "loss": 1.5251, "step": 1962 }, { "epoch": 0.32, "grad_norm": 7.263148784637451, "learning_rate": 1.6009934136545208e-05, "loss": 1.4672, "step": 1963 }, { "epoch": 0.32, "grad_norm": 7.474334239959717, "learning_rate": 1.6005765800364233e-05, "loss": 1.6741, "step": 1964 }, { "epoch": 0.32, "grad_norm": 6.739398002624512, "learning_rate": 1.6001595831304175e-05, "loss": 1.8325, "step": 1965 }, { "epoch": 0.32, "grad_norm": 7.009296417236328, "learning_rate": 1.5997424230498782e-05, "loss": 1.9309, "step": 1966 }, { "epoch": 0.32, "grad_norm": 9.029295921325684, "learning_rate": 1.599325099908226e-05, "loss": 1.766, "step": 1967 }, { "epoch": 0.32, "grad_norm": 8.20340633392334, "learning_rate": 1.5989076138189248e-05, "loss": 1.8811, "step": 1968 }, { "epoch": 0.32, "grad_norm": 8.005865097045898, "learning_rate": 1.5984899648954825e-05, "loss": 1.437, "step": 1969 }, { "epoch": 0.32, "grad_norm": 8.835622787475586, "learning_rate": 1.5980721532514518e-05, "loss": 1.7554, "step": 1970 }, { "epoch": 0.32, "grad_norm": 7.880939483642578, "learning_rate": 1.5976541790004297e-05, "loss": 2.029, "step": 1971 }, { "epoch": 0.32, "grad_norm": 7.275417327880859, "learning_rate": 1.597236042256057e-05, "loss": 1.7348, "step": 1972 }, { "epoch": 0.32, "grad_norm": 6.86923360824585, "learning_rate": 1.596817743132019e-05, "loss": 1.7067, "step": 1973 }, { "epoch": 0.32, "grad_norm": 9.909432411193848, "learning_rate": 1.5963992817420448e-05, "loss": 1.9487, "step": 1974 }, { "epoch": 0.32, "grad_norm": 8.909941673278809, "learning_rate": 1.5959806581999085e-05, "loss": 1.7606, "step": 1975 }, { "epoch": 0.32, "grad_norm": 7.994794845581055, "learning_rate": 1.595561872619427e-05, "loss": 1.6961, "step": 1976 }, { "epoch": 0.32, "grad_norm": 8.532252311706543, "learning_rate": 1.5951429251144626e-05, "loss": 1.8295, "step": 1977 }, { "epoch": 0.32, "grad_norm": 8.234542846679688, "learning_rate": 1.59472381579892e-05, "loss": 1.7241, "step": 1978 }, { "epoch": 0.32, "grad_norm": 8.174495697021484, "learning_rate": 1.59430454478675e-05, "loss": 1.8977, "step": 1979 }, { "epoch": 0.32, "grad_norm": 7.866890907287598, "learning_rate": 1.5938851121919453e-05, "loss": 1.6811, "step": 1980 }, { "epoch": 0.32, "grad_norm": 8.968592643737793, "learning_rate": 1.593465518128544e-05, "loss": 1.673, "step": 1981 }, { "epoch": 0.32, "grad_norm": 7.99185848236084, "learning_rate": 1.5930457627106272e-05, "loss": 1.5594, "step": 1982 }, { "epoch": 0.32, "grad_norm": 8.130263328552246, "learning_rate": 1.59262584605232e-05, "loss": 1.6852, "step": 1983 }, { "epoch": 0.32, "grad_norm": 8.6261568069458, "learning_rate": 1.5922057682677917e-05, "loss": 1.7254, "step": 1984 }, { "epoch": 0.32, "grad_norm": 7.3557634353637695, "learning_rate": 1.591785529471256e-05, "loss": 1.7284, "step": 1985 }, { "epoch": 0.32, "grad_norm": 8.28761100769043, "learning_rate": 1.5913651297769685e-05, "loss": 1.5994, "step": 1986 }, { "epoch": 0.32, "grad_norm": 8.476308822631836, "learning_rate": 1.59094456929923e-05, "loss": 1.8145, "step": 1987 }, { "epoch": 0.32, "grad_norm": 7.182476043701172, "learning_rate": 1.590523848152385e-05, "loss": 1.6076, "step": 1988 }, { "epoch": 0.32, "grad_norm": 9.080642700195312, "learning_rate": 1.5901029664508215e-05, "loss": 1.6273, "step": 1989 }, { "epoch": 0.32, "grad_norm": 8.31082820892334, "learning_rate": 1.5896819243089706e-05, "loss": 1.8925, "step": 1990 }, { "epoch": 0.32, "grad_norm": 9.597949981689453, "learning_rate": 1.5892607218413073e-05, "loss": 1.8469, "step": 1991 }, { "epoch": 0.32, "grad_norm": 7.270137310028076, "learning_rate": 1.5888393591623507e-05, "loss": 1.5361, "step": 1992 }, { "epoch": 0.32, "grad_norm": 7.863422393798828, "learning_rate": 1.5884178363866634e-05, "loss": 1.4811, "step": 1993 }, { "epoch": 0.32, "grad_norm": 7.43480920791626, "learning_rate": 1.5879961536288506e-05, "loss": 1.8442, "step": 1994 }, { "epoch": 0.32, "grad_norm": 8.248194694519043, "learning_rate": 1.5875743110035616e-05, "loss": 1.7988, "step": 1995 }, { "epoch": 0.32, "grad_norm": 8.713356018066406, "learning_rate": 1.5871523086254898e-05, "loss": 1.637, "step": 1996 }, { "epoch": 0.32, "grad_norm": 6.644055366516113, "learning_rate": 1.586730146609371e-05, "loss": 1.5922, "step": 1997 }, { "epoch": 0.32, "grad_norm": 7.581531047821045, "learning_rate": 1.5863078250699845e-05, "loss": 1.7323, "step": 1998 }, { "epoch": 0.32, "grad_norm": 8.46491527557373, "learning_rate": 1.585885344122154e-05, "loss": 1.5783, "step": 1999 }, { "epoch": 0.32, "grad_norm": 8.68262767791748, "learning_rate": 1.5854627038807454e-05, "loss": 1.8034, "step": 2000 }, { "epoch": 0.32, "grad_norm": 10.512350082397461, "learning_rate": 1.5850399044606687e-05, "loss": 1.8814, "step": 2001 }, { "epoch": 0.32, "grad_norm": 6.727400302886963, "learning_rate": 1.584616945976876e-05, "loss": 1.8373, "step": 2002 }, { "epoch": 0.32, "grad_norm": 7.481900691986084, "learning_rate": 1.5841938285443644e-05, "loss": 1.7578, "step": 2003 }, { "epoch": 0.32, "grad_norm": 6.470006465911865, "learning_rate": 1.583770552278173e-05, "loss": 1.8251, "step": 2004 }, { "epoch": 0.32, "grad_norm": 7.987014293670654, "learning_rate": 1.5833471172933842e-05, "loss": 1.7784, "step": 2005 }, { "epoch": 0.32, "grad_norm": 10.678996086120605, "learning_rate": 1.582923523705124e-05, "loss": 1.7122, "step": 2006 }, { "epoch": 0.32, "grad_norm": 8.217026710510254, "learning_rate": 1.5824997716285606e-05, "loss": 1.598, "step": 2007 }, { "epoch": 0.32, "grad_norm": 6.79136848449707, "learning_rate": 1.5820758611789066e-05, "loss": 1.8151, "step": 2008 }, { "epoch": 0.32, "grad_norm": 8.667113304138184, "learning_rate": 1.581651792471417e-05, "loss": 1.6284, "step": 2009 }, { "epoch": 0.32, "grad_norm": 11.433363914489746, "learning_rate": 1.5812275656213896e-05, "loss": 1.7207, "step": 2010 }, { "epoch": 0.32, "grad_norm": 7.387059211730957, "learning_rate": 1.580803180744165e-05, "loss": 1.92, "step": 2011 }, { "epoch": 0.32, "grad_norm": 7.251891613006592, "learning_rate": 1.580378637955128e-05, "loss": 1.6592, "step": 2012 }, { "epoch": 0.32, "grad_norm": 7.230704307556152, "learning_rate": 1.5799539373697052e-05, "loss": 1.7642, "step": 2013 }, { "epoch": 0.32, "grad_norm": 8.49705696105957, "learning_rate": 1.579529079103366e-05, "loss": 1.6935, "step": 2014 }, { "epoch": 0.32, "grad_norm": 7.402725696563721, "learning_rate": 1.579104063271624e-05, "loss": 1.7082, "step": 2015 }, { "epoch": 0.32, "grad_norm": 8.3992338180542, "learning_rate": 1.5786788899900332e-05, "loss": 1.7871, "step": 2016 }, { "epoch": 0.32, "grad_norm": 8.230449676513672, "learning_rate": 1.578253559374194e-05, "loss": 1.6889, "step": 2017 }, { "epoch": 0.32, "grad_norm": 7.194096565246582, "learning_rate": 1.5778280715397457e-05, "loss": 1.709, "step": 2018 }, { "epoch": 0.33, "grad_norm": 7.2667717933654785, "learning_rate": 1.5774024266023732e-05, "loss": 1.7624, "step": 2019 }, { "epoch": 0.33, "grad_norm": 10.099437713623047, "learning_rate": 1.5769766246778028e-05, "loss": 1.7053, "step": 2020 }, { "epoch": 0.33, "grad_norm": 7.769886016845703, "learning_rate": 1.576550665881804e-05, "loss": 2.0048, "step": 2021 }, { "epoch": 0.33, "grad_norm": 7.3910813331604, "learning_rate": 1.576124550330188e-05, "loss": 2.0582, "step": 2022 }, { "epoch": 0.33, "grad_norm": 7.10804557800293, "learning_rate": 1.57569827813881e-05, "loss": 1.7258, "step": 2023 }, { "epoch": 0.33, "grad_norm": 9.282371520996094, "learning_rate": 1.5752718494235666e-05, "loss": 1.6427, "step": 2024 }, { "epoch": 0.33, "grad_norm": 11.13532829284668, "learning_rate": 1.574845264300398e-05, "loss": 1.6077, "step": 2025 }, { "epoch": 0.33, "grad_norm": 6.797539710998535, "learning_rate": 1.574418522885286e-05, "loss": 1.6763, "step": 2026 }, { "epoch": 0.33, "grad_norm": 7.4723663330078125, "learning_rate": 1.5739916252942556e-05, "loss": 1.5757, "step": 2027 }, { "epoch": 0.33, "grad_norm": 8.171815872192383, "learning_rate": 1.5735645716433737e-05, "loss": 1.9289, "step": 2028 }, { "epoch": 0.33, "grad_norm": 8.688456535339355, "learning_rate": 1.5731373620487503e-05, "loss": 1.7553, "step": 2029 }, { "epoch": 0.33, "grad_norm": 8.375536918640137, "learning_rate": 1.5727099966265365e-05, "loss": 1.6448, "step": 2030 }, { "epoch": 0.33, "grad_norm": 10.568947792053223, "learning_rate": 1.572282475492928e-05, "loss": 1.5242, "step": 2031 }, { "epoch": 0.33, "grad_norm": 10.954500198364258, "learning_rate": 1.5718547987641603e-05, "loss": 1.647, "step": 2032 }, { "epoch": 0.33, "grad_norm": 8.975900650024414, "learning_rate": 1.571426966556513e-05, "loss": 1.7183, "step": 2033 }, { "epoch": 0.33, "grad_norm": 10.774421691894531, "learning_rate": 1.570998978986307e-05, "loss": 1.685, "step": 2034 }, { "epoch": 0.33, "grad_norm": 13.142233848571777, "learning_rate": 1.5705708361699053e-05, "loss": 1.6874, "step": 2035 }, { "epoch": 0.33, "grad_norm": 7.271115779876709, "learning_rate": 1.5701425382237153e-05, "loss": 1.9134, "step": 2036 }, { "epoch": 0.33, "grad_norm": 8.920947074890137, "learning_rate": 1.5697140852641835e-05, "loss": 1.7561, "step": 2037 }, { "epoch": 0.33, "grad_norm": 8.030533790588379, "learning_rate": 1.5692854774078004e-05, "loss": 1.7627, "step": 2038 }, { "epoch": 0.33, "grad_norm": 6.218264579772949, "learning_rate": 1.568856714771098e-05, "loss": 1.7091, "step": 2039 }, { "epoch": 0.33, "grad_norm": 7.636033058166504, "learning_rate": 1.5684277974706503e-05, "loss": 1.5469, "step": 2040 }, { "epoch": 0.33, "grad_norm": 8.745565414428711, "learning_rate": 1.5679987256230743e-05, "loss": 1.7634, "step": 2041 }, { "epoch": 0.33, "grad_norm": 8.606216430664062, "learning_rate": 1.567569499345028e-05, "loss": 1.8078, "step": 2042 }, { "epoch": 0.33, "grad_norm": 14.161050796508789, "learning_rate": 1.5671401187532112e-05, "loss": 1.7505, "step": 2043 }, { "epoch": 0.33, "grad_norm": 7.432353973388672, "learning_rate": 1.5667105839643667e-05, "loss": 1.5248, "step": 2044 }, { "epoch": 0.33, "grad_norm": 12.268610000610352, "learning_rate": 1.5662808950952784e-05, "loss": 1.6779, "step": 2045 }, { "epoch": 0.33, "grad_norm": 7.902520656585693, "learning_rate": 1.5658510522627724e-05, "loss": 1.6995, "step": 2046 }, { "epoch": 0.33, "grad_norm": 7.894237041473389, "learning_rate": 1.5654210555837166e-05, "loss": 1.6656, "step": 2047 }, { "epoch": 0.33, "grad_norm": 7.696222305297852, "learning_rate": 1.5649909051750212e-05, "loss": 1.7121, "step": 2048 }, { "epoch": 0.33, "grad_norm": 9.068448066711426, "learning_rate": 1.564560601153637e-05, "loss": 1.6468, "step": 2049 }, { "epoch": 0.33, "grad_norm": 8.91800594329834, "learning_rate": 1.564130143636558e-05, "loss": 1.4779, "step": 2050 }, { "epoch": 0.33, "grad_norm": 9.075319290161133, "learning_rate": 1.5636995327408188e-05, "loss": 1.6019, "step": 2051 }, { "epoch": 0.33, "grad_norm": 6.758200168609619, "learning_rate": 1.563268768583496e-05, "loss": 1.728, "step": 2052 }, { "epoch": 0.33, "grad_norm": 8.406549453735352, "learning_rate": 1.5628378512817085e-05, "loss": 1.8224, "step": 2053 }, { "epoch": 0.33, "grad_norm": 8.161641120910645, "learning_rate": 1.5624067809526164e-05, "loss": 1.6827, "step": 2054 }, { "epoch": 0.33, "grad_norm": 8.51504898071289, "learning_rate": 1.561975557713421e-05, "loss": 1.6884, "step": 2055 }, { "epoch": 0.33, "grad_norm": 10.643168449401855, "learning_rate": 1.5615441816813653e-05, "loss": 1.7814, "step": 2056 }, { "epoch": 0.33, "grad_norm": 7.554748058319092, "learning_rate": 1.5611126529737348e-05, "loss": 1.5767, "step": 2057 }, { "epoch": 0.33, "grad_norm": 10.011309623718262, "learning_rate": 1.560680971707855e-05, "loss": 1.6979, "step": 2058 }, { "epoch": 0.33, "grad_norm": 6.8463897705078125, "learning_rate": 1.5602491380010944e-05, "loss": 1.7552, "step": 2059 }, { "epoch": 0.33, "grad_norm": 13.44117546081543, "learning_rate": 1.5598171519708618e-05, "loss": 1.5318, "step": 2060 }, { "epoch": 0.33, "grad_norm": 8.644599914550781, "learning_rate": 1.5593850137346077e-05, "loss": 1.81, "step": 2061 }, { "epoch": 0.33, "grad_norm": 13.854301452636719, "learning_rate": 1.5589527234098247e-05, "loss": 1.6461, "step": 2062 }, { "epoch": 0.33, "grad_norm": 7.3202104568481445, "learning_rate": 1.5585202811140458e-05, "loss": 1.6995, "step": 2063 }, { "epoch": 0.33, "grad_norm": 8.716704368591309, "learning_rate": 1.5580876869648452e-05, "loss": 1.7238, "step": 2064 }, { "epoch": 0.33, "grad_norm": 10.325505256652832, "learning_rate": 1.5576549410798396e-05, "loss": 1.6624, "step": 2065 }, { "epoch": 0.33, "grad_norm": 10.217865943908691, "learning_rate": 1.5572220435766858e-05, "loss": 1.5416, "step": 2066 }, { "epoch": 0.33, "grad_norm": 9.109761238098145, "learning_rate": 1.5567889945730822e-05, "loss": 1.5887, "step": 2067 }, { "epoch": 0.33, "grad_norm": 9.59637451171875, "learning_rate": 1.556355794186769e-05, "loss": 1.5071, "step": 2068 }, { "epoch": 0.33, "grad_norm": 8.913079261779785, "learning_rate": 1.5559224425355264e-05, "loss": 1.7925, "step": 2069 }, { "epoch": 0.33, "grad_norm": 7.314448356628418, "learning_rate": 1.5554889397371767e-05, "loss": 1.4451, "step": 2070 }, { "epoch": 0.33, "grad_norm": 9.813277244567871, "learning_rate": 1.5550552859095827e-05, "loss": 1.7002, "step": 2071 }, { "epoch": 0.33, "grad_norm": 7.067071437835693, "learning_rate": 1.5546214811706485e-05, "loss": 1.5832, "step": 2072 }, { "epoch": 0.33, "grad_norm": 7.3300347328186035, "learning_rate": 1.554187525638319e-05, "loss": 1.8263, "step": 2073 }, { "epoch": 0.33, "grad_norm": 7.563827037811279, "learning_rate": 1.553753419430581e-05, "loss": 1.6943, "step": 2074 }, { "epoch": 0.33, "grad_norm": 9.001516342163086, "learning_rate": 1.5533191626654606e-05, "loss": 1.6043, "step": 2075 }, { "epoch": 0.33, "grad_norm": 8.808286666870117, "learning_rate": 1.5528847554610263e-05, "loss": 1.8076, "step": 2076 }, { "epoch": 0.33, "grad_norm": 9.225560188293457, "learning_rate": 1.5524501979353872e-05, "loss": 1.7221, "step": 2077 }, { "epoch": 0.33, "grad_norm": 7.297783851623535, "learning_rate": 1.552015490206693e-05, "loss": 1.7137, "step": 2078 }, { "epoch": 0.33, "grad_norm": 7.307004928588867, "learning_rate": 1.551580632393134e-05, "loss": 1.6143, "step": 2079 }, { "epoch": 0.33, "grad_norm": 7.89572286605835, "learning_rate": 1.551145624612942e-05, "loss": 1.6863, "step": 2080 }, { "epoch": 0.33, "grad_norm": 8.785066604614258, "learning_rate": 1.5507104669843887e-05, "loss": 1.8586, "step": 2081 }, { "epoch": 0.34, "grad_norm": 7.563694953918457, "learning_rate": 1.5502751596257875e-05, "loss": 1.6464, "step": 2082 }, { "epoch": 0.34, "grad_norm": 8.065217971801758, "learning_rate": 1.549839702655492e-05, "loss": 1.6407, "step": 2083 }, { "epoch": 0.34, "grad_norm": 7.006573677062988, "learning_rate": 1.5494040961918964e-05, "loss": 1.6783, "step": 2084 }, { "epoch": 0.34, "grad_norm": 7.542549133300781, "learning_rate": 1.5489683403534352e-05, "loss": 1.7182, "step": 2085 }, { "epoch": 0.34, "grad_norm": 7.9340901374816895, "learning_rate": 1.548532435258585e-05, "loss": 1.6092, "step": 2086 }, { "epoch": 0.34, "grad_norm": 8.985841751098633, "learning_rate": 1.5480963810258614e-05, "loss": 1.6276, "step": 2087 }, { "epoch": 0.34, "grad_norm": 7.833774089813232, "learning_rate": 1.547660177773821e-05, "loss": 1.4204, "step": 2088 }, { "epoch": 0.34, "grad_norm": 8.100363731384277, "learning_rate": 1.5472238256210612e-05, "loss": 1.7691, "step": 2089 }, { "epoch": 0.34, "grad_norm": 7.822362422943115, "learning_rate": 1.5467873246862197e-05, "loss": 1.9155, "step": 2090 }, { "epoch": 0.34, "grad_norm": 10.381099700927734, "learning_rate": 1.5463506750879747e-05, "loss": 1.7976, "step": 2091 }, { "epoch": 0.34, "grad_norm": 7.504476547241211, "learning_rate": 1.5459138769450448e-05, "loss": 1.76, "step": 2092 }, { "epoch": 0.34, "grad_norm": 9.416753768920898, "learning_rate": 1.5454769303761887e-05, "loss": 1.5641, "step": 2093 }, { "epoch": 0.34, "grad_norm": 8.967190742492676, "learning_rate": 1.5450398355002057e-05, "loss": 1.6888, "step": 2094 }, { "epoch": 0.34, "grad_norm": 7.686070442199707, "learning_rate": 1.544602592435936e-05, "loss": 1.689, "step": 2095 }, { "epoch": 0.34, "grad_norm": 10.032136917114258, "learning_rate": 1.5441652013022593e-05, "loss": 1.716, "step": 2096 }, { "epoch": 0.34, "grad_norm": 10.230822563171387, "learning_rate": 1.5437276622180953e-05, "loss": 1.7389, "step": 2097 }, { "epoch": 0.34, "grad_norm": 7.754800796508789, "learning_rate": 1.543289975302405e-05, "loss": 1.7257, "step": 2098 }, { "epoch": 0.34, "grad_norm": 6.925958156585693, "learning_rate": 1.5428521406741888e-05, "loss": 1.6891, "step": 2099 }, { "epoch": 0.34, "grad_norm": 7.839892387390137, "learning_rate": 1.5424141584524876e-05, "loss": 1.6937, "step": 2100 }, { "epoch": 0.34, "grad_norm": 10.216348648071289, "learning_rate": 1.541976028756382e-05, "loss": 1.6994, "step": 2101 }, { "epoch": 0.34, "grad_norm": 7.378163814544678, "learning_rate": 1.5415377517049933e-05, "loss": 1.6117, "step": 2102 }, { "epoch": 0.34, "grad_norm": 8.603313446044922, "learning_rate": 1.5410993274174827e-05, "loss": 1.8043, "step": 2103 }, { "epoch": 0.34, "grad_norm": 8.415882110595703, "learning_rate": 1.5406607560130506e-05, "loss": 1.798, "step": 2104 }, { "epoch": 0.34, "grad_norm": 8.355718612670898, "learning_rate": 1.5402220376109386e-05, "loss": 1.7733, "step": 2105 }, { "epoch": 0.34, "grad_norm": 10.17424488067627, "learning_rate": 1.5397831723304278e-05, "loss": 1.6802, "step": 2106 }, { "epoch": 0.34, "grad_norm": 7.448206901550293, "learning_rate": 1.5393441602908395e-05, "loss": 1.7494, "step": 2107 }, { "epoch": 0.34, "grad_norm": 9.086813926696777, "learning_rate": 1.538905001611534e-05, "loss": 1.8842, "step": 2108 }, { "epoch": 0.34, "grad_norm": 12.772475242614746, "learning_rate": 1.538465696411912e-05, "loss": 1.5866, "step": 2109 }, { "epoch": 0.34, "grad_norm": 8.74178409576416, "learning_rate": 1.538026244811414e-05, "loss": 1.7692, "step": 2110 }, { "epoch": 0.34, "grad_norm": 12.885123252868652, "learning_rate": 1.5375866469295213e-05, "loss": 1.8442, "step": 2111 }, { "epoch": 0.34, "grad_norm": 7.8096089363098145, "learning_rate": 1.5371469028857534e-05, "loss": 1.465, "step": 2112 }, { "epoch": 0.34, "grad_norm": 8.224424362182617, "learning_rate": 1.5367070127996702e-05, "loss": 1.5761, "step": 2113 }, { "epoch": 0.34, "grad_norm": 9.944413185119629, "learning_rate": 1.5362669767908716e-05, "loss": 1.6974, "step": 2114 }, { "epoch": 0.34, "grad_norm": 8.157191276550293, "learning_rate": 1.5358267949789968e-05, "loss": 1.8573, "step": 2115 }, { "epoch": 0.34, "grad_norm": 8.32761001586914, "learning_rate": 1.5353864674837248e-05, "loss": 1.7055, "step": 2116 }, { "epoch": 0.34, "grad_norm": 10.562308311462402, "learning_rate": 1.534945994424774e-05, "loss": 1.6192, "step": 2117 }, { "epoch": 0.34, "grad_norm": 9.23558235168457, "learning_rate": 1.5345053759219024e-05, "loss": 1.7442, "step": 2118 }, { "epoch": 0.34, "grad_norm": 7.139028072357178, "learning_rate": 1.5340646120949077e-05, "loss": 1.8949, "step": 2119 }, { "epoch": 0.34, "grad_norm": 6.149474620819092, "learning_rate": 1.5336237030636276e-05, "loss": 1.5671, "step": 2120 }, { "epoch": 0.34, "grad_norm": 9.99124526977539, "learning_rate": 1.533182648947938e-05, "loss": 1.4754, "step": 2121 }, { "epoch": 0.34, "grad_norm": 9.409079551696777, "learning_rate": 1.5327414498677556e-05, "loss": 1.5695, "step": 2122 }, { "epoch": 0.34, "grad_norm": 9.880412101745605, "learning_rate": 1.5323001059430357e-05, "loss": 1.9161, "step": 2123 }, { "epoch": 0.34, "grad_norm": 7.699712753295898, "learning_rate": 1.5318586172937732e-05, "loss": 1.7352, "step": 2124 }, { "epoch": 0.34, "grad_norm": 8.453593254089355, "learning_rate": 1.531416984040002e-05, "loss": 1.7128, "step": 2125 }, { "epoch": 0.34, "grad_norm": 7.570315837860107, "learning_rate": 1.5309752063017963e-05, "loss": 1.889, "step": 2126 }, { "epoch": 0.34, "grad_norm": 7.294005870819092, "learning_rate": 1.530533284199269e-05, "loss": 1.6716, "step": 2127 }, { "epoch": 0.34, "grad_norm": 10.335264205932617, "learning_rate": 1.5300912178525716e-05, "loss": 1.7573, "step": 2128 }, { "epoch": 0.34, "grad_norm": 7.714483261108398, "learning_rate": 1.5296490073818956e-05, "loss": 1.8626, "step": 2129 }, { "epoch": 0.34, "grad_norm": 7.540449619293213, "learning_rate": 1.5292066529074715e-05, "loss": 1.7561, "step": 2130 }, { "epoch": 0.34, "grad_norm": 7.053147792816162, "learning_rate": 1.528764154549569e-05, "loss": 1.5437, "step": 2131 }, { "epoch": 0.34, "grad_norm": 7.951387882232666, "learning_rate": 1.5283215124284972e-05, "loss": 1.8059, "step": 2132 }, { "epoch": 0.34, "grad_norm": 8.435399055480957, "learning_rate": 1.5278787266646034e-05, "loss": 1.602, "step": 2133 }, { "epoch": 0.34, "grad_norm": 13.08942699432373, "learning_rate": 1.5274357973782747e-05, "loss": 1.8015, "step": 2134 }, { "epoch": 0.34, "grad_norm": 8.003021240234375, "learning_rate": 1.5269927246899375e-05, "loss": 1.9772, "step": 2135 }, { "epoch": 0.34, "grad_norm": 10.33984375, "learning_rate": 1.526549508720056e-05, "loss": 1.6292, "step": 2136 }, { "epoch": 0.34, "grad_norm": 10.168055534362793, "learning_rate": 1.5261061495891345e-05, "loss": 1.7141, "step": 2137 }, { "epoch": 0.34, "grad_norm": 8.36406135559082, "learning_rate": 1.5256626474177157e-05, "loss": 1.5498, "step": 2138 }, { "epoch": 0.34, "grad_norm": 9.640942573547363, "learning_rate": 1.5252190023263811e-05, "loss": 1.6403, "step": 2139 }, { "epoch": 0.34, "grad_norm": 9.660639762878418, "learning_rate": 1.5247752144357516e-05, "loss": 1.693, "step": 2140 }, { "epoch": 0.34, "grad_norm": 16.295101165771484, "learning_rate": 1.5243312838664866e-05, "loss": 1.7646, "step": 2141 }, { "epoch": 0.34, "grad_norm": 8.36565113067627, "learning_rate": 1.5238872107392842e-05, "loss": 1.707, "step": 2142 }, { "epoch": 0.34, "grad_norm": 9.172396659851074, "learning_rate": 1.5234429951748805e-05, "loss": 1.5267, "step": 2143 }, { "epoch": 0.35, "grad_norm": 10.870437622070312, "learning_rate": 1.5229986372940526e-05, "loss": 1.6367, "step": 2144 }, { "epoch": 0.35, "grad_norm": 9.839536666870117, "learning_rate": 1.5225541372176142e-05, "loss": 1.6513, "step": 2145 }, { "epoch": 0.35, "grad_norm": 10.421070098876953, "learning_rate": 1.5221094950664178e-05, "loss": 1.4374, "step": 2146 }, { "epoch": 0.35, "grad_norm": 7.615229606628418, "learning_rate": 1.521664710961356e-05, "loss": 1.8762, "step": 2147 }, { "epoch": 0.35, "grad_norm": 7.256735801696777, "learning_rate": 1.5212197850233578e-05, "loss": 1.7606, "step": 2148 }, { "epoch": 0.35, "grad_norm": 9.234436988830566, "learning_rate": 1.5207747173733935e-05, "loss": 1.7934, "step": 2149 }, { "epoch": 0.35, "grad_norm": 10.437911033630371, "learning_rate": 1.5203295081324694e-05, "loss": 1.6701, "step": 2150 }, { "epoch": 0.35, "grad_norm": 7.426792621612549, "learning_rate": 1.5198841574216315e-05, "loss": 1.6129, "step": 2151 }, { "epoch": 0.35, "grad_norm": 7.617865085601807, "learning_rate": 1.519438665361964e-05, "loss": 1.6767, "step": 2152 }, { "epoch": 0.35, "grad_norm": 8.803878784179688, "learning_rate": 1.5189930320745899e-05, "loss": 1.6197, "step": 2153 }, { "epoch": 0.35, "grad_norm": 9.386178970336914, "learning_rate": 1.5185472576806703e-05, "loss": 1.6892, "step": 2154 }, { "epoch": 0.35, "grad_norm": 8.533537864685059, "learning_rate": 1.5181013423014044e-05, "loss": 1.6725, "step": 2155 }, { "epoch": 0.35, "grad_norm": 8.585227012634277, "learning_rate": 1.5176552860580302e-05, "loss": 1.6709, "step": 2156 }, { "epoch": 0.35, "grad_norm": 9.306947708129883, "learning_rate": 1.5172090890718237e-05, "loss": 1.6601, "step": 2157 }, { "epoch": 0.35, "grad_norm": 10.270577430725098, "learning_rate": 1.5167627514640994e-05, "loss": 1.5875, "step": 2158 }, { "epoch": 0.35, "grad_norm": 7.824845314025879, "learning_rate": 1.51631627335621e-05, "loss": 1.6937, "step": 2159 }, { "epoch": 0.35, "grad_norm": 6.994386196136475, "learning_rate": 1.5158696548695465e-05, "loss": 1.7512, "step": 2160 }, { "epoch": 0.35, "grad_norm": 14.297425270080566, "learning_rate": 1.515422896125537e-05, "loss": 1.8541, "step": 2161 }, { "epoch": 0.35, "grad_norm": 9.075075149536133, "learning_rate": 1.514975997245649e-05, "loss": 1.5954, "step": 2162 }, { "epoch": 0.35, "grad_norm": 7.600062847137451, "learning_rate": 1.5145289583513884e-05, "loss": 1.9197, "step": 2163 }, { "epoch": 0.35, "grad_norm": 7.0530619621276855, "learning_rate": 1.5140817795642976e-05, "loss": 1.6306, "step": 2164 }, { "epoch": 0.35, "grad_norm": 9.239937782287598, "learning_rate": 1.5136344610059586e-05, "loss": 1.8318, "step": 2165 }, { "epoch": 0.35, "grad_norm": 10.14870548248291, "learning_rate": 1.51318700279799e-05, "loss": 1.7072, "step": 2166 }, { "epoch": 0.35, "grad_norm": 7.551357269287109, "learning_rate": 1.5127394050620497e-05, "loss": 1.8685, "step": 2167 }, { "epoch": 0.35, "grad_norm": 7.268665790557861, "learning_rate": 1.5122916679198325e-05, "loss": 1.7463, "step": 2168 }, { "epoch": 0.35, "grad_norm": 6.9935503005981445, "learning_rate": 1.5118437914930719e-05, "loss": 1.6671, "step": 2169 }, { "epoch": 0.35, "grad_norm": 8.269112586975098, "learning_rate": 1.5113957759035388e-05, "loss": 1.8375, "step": 2170 }, { "epoch": 0.35, "grad_norm": 6.602900981903076, "learning_rate": 1.5109476212730415e-05, "loss": 1.6057, "step": 2171 }, { "epoch": 0.35, "grad_norm": 11.656549453735352, "learning_rate": 1.510499327723427e-05, "loss": 1.5939, "step": 2172 }, { "epoch": 0.35, "grad_norm": 6.929116249084473, "learning_rate": 1.51005089537658e-05, "loss": 1.6594, "step": 2173 }, { "epoch": 0.35, "grad_norm": 8.367817878723145, "learning_rate": 1.5096023243544223e-05, "loss": 1.5399, "step": 2174 }, { "epoch": 0.35, "grad_norm": 9.914570808410645, "learning_rate": 1.509153614778914e-05, "loss": 1.7419, "step": 2175 }, { "epoch": 0.35, "grad_norm": 9.803070068359375, "learning_rate": 1.5087047667720521e-05, "loss": 1.6941, "step": 2176 }, { "epoch": 0.35, "grad_norm": 7.945676803588867, "learning_rate": 1.5082557804558722e-05, "loss": 1.7329, "step": 2177 }, { "epoch": 0.35, "grad_norm": 11.422955513000488, "learning_rate": 1.5078066559524472e-05, "loss": 1.7405, "step": 2178 }, { "epoch": 0.35, "grad_norm": 7.648712158203125, "learning_rate": 1.5073573933838868e-05, "loss": 1.561, "step": 2179 }, { "epoch": 0.35, "grad_norm": 8.283636093139648, "learning_rate": 1.5069079928723394e-05, "loss": 1.5218, "step": 2180 }, { "epoch": 0.35, "grad_norm": 7.377476215362549, "learning_rate": 1.5064584545399901e-05, "loss": 1.759, "step": 2181 }, { "epoch": 0.35, "grad_norm": 8.612385749816895, "learning_rate": 1.5060087785090618e-05, "loss": 1.7706, "step": 2182 }, { "epoch": 0.35, "grad_norm": 8.549281120300293, "learning_rate": 1.5055589649018147e-05, "loss": 1.9166, "step": 2183 }, { "epoch": 0.35, "grad_norm": 8.119498252868652, "learning_rate": 1.5051090138405466e-05, "loss": 1.5604, "step": 2184 }, { "epoch": 0.35, "grad_norm": 7.497628688812256, "learning_rate": 1.504658925447593e-05, "loss": 1.7872, "step": 2185 }, { "epoch": 0.35, "grad_norm": 8.585989952087402, "learning_rate": 1.504208699845325e-05, "loss": 1.7395, "step": 2186 }, { "epoch": 0.35, "grad_norm": 8.487015724182129, "learning_rate": 1.5037583371561538e-05, "loss": 1.8085, "step": 2187 }, { "epoch": 0.35, "grad_norm": 7.596621990203857, "learning_rate": 1.5033078375025257e-05, "loss": 1.6529, "step": 2188 }, { "epoch": 0.35, "grad_norm": 7.740676403045654, "learning_rate": 1.5028572010069245e-05, "loss": 1.6993, "step": 2189 }, { "epoch": 0.35, "grad_norm": 9.391162872314453, "learning_rate": 1.5024064277918719e-05, "loss": 1.6978, "step": 2190 }, { "epoch": 0.35, "grad_norm": 7.582511901855469, "learning_rate": 1.501955517979927e-05, "loss": 1.9275, "step": 2191 }, { "epoch": 0.35, "grad_norm": 12.196687698364258, "learning_rate": 1.5015044716936848e-05, "loss": 1.6735, "step": 2192 }, { "epoch": 0.35, "grad_norm": 9.719663619995117, "learning_rate": 1.5010532890557787e-05, "loss": 1.6693, "step": 2193 }, { "epoch": 0.35, "grad_norm": 8.062163352966309, "learning_rate": 1.5006019701888783e-05, "loss": 1.6904, "step": 2194 }, { "epoch": 0.35, "grad_norm": 7.953310966491699, "learning_rate": 1.5001505152156903e-05, "loss": 1.485, "step": 2195 }, { "epoch": 0.35, "grad_norm": 8.113821029663086, "learning_rate": 1.4996989242589593e-05, "loss": 1.6429, "step": 2196 }, { "epoch": 0.35, "grad_norm": 12.139039993286133, "learning_rate": 1.4992471974414658e-05, "loss": 1.684, "step": 2197 }, { "epoch": 0.35, "grad_norm": 12.47142219543457, "learning_rate": 1.4987953348860273e-05, "loss": 1.7307, "step": 2198 }, { "epoch": 0.35, "grad_norm": 7.752344131469727, "learning_rate": 1.4983433367154994e-05, "loss": 1.6016, "step": 2199 }, { "epoch": 0.35, "grad_norm": 6.754793643951416, "learning_rate": 1.4978912030527727e-05, "loss": 1.4997, "step": 2200 }, { "epoch": 0.35, "grad_norm": 7.84208345413208, "learning_rate": 1.4974389340207766e-05, "loss": 1.7737, "step": 2201 }, { "epoch": 0.35, "grad_norm": 14.207002639770508, "learning_rate": 1.4969865297424764e-05, "loss": 1.9143, "step": 2202 }, { "epoch": 0.35, "grad_norm": 12.688886642456055, "learning_rate": 1.4965339903408735e-05, "loss": 1.6201, "step": 2203 }, { "epoch": 0.35, "grad_norm": 7.396913051605225, "learning_rate": 1.4960813159390069e-05, "loss": 1.6875, "step": 2204 }, { "epoch": 0.35, "grad_norm": 8.878341674804688, "learning_rate": 1.495628506659952e-05, "loss": 1.7724, "step": 2205 }, { "epoch": 0.36, "grad_norm": 8.288783073425293, "learning_rate": 1.4951755626268212e-05, "loss": 1.582, "step": 2206 }, { "epoch": 0.36, "grad_norm": 7.879135608673096, "learning_rate": 1.4947224839627637e-05, "loss": 1.8576, "step": 2207 }, { "epoch": 0.36, "grad_norm": 6.7551398277282715, "learning_rate": 1.4942692707909639e-05, "loss": 1.628, "step": 2208 }, { "epoch": 0.36, "grad_norm": 8.997371673583984, "learning_rate": 1.4938159232346441e-05, "loss": 1.5864, "step": 2209 }, { "epoch": 0.36, "grad_norm": 9.095409393310547, "learning_rate": 1.4933624414170633e-05, "loss": 1.7135, "step": 2210 }, { "epoch": 0.36, "grad_norm": 8.195283889770508, "learning_rate": 1.4929088254615161e-05, "loss": 1.5014, "step": 2211 }, { "epoch": 0.36, "grad_norm": 7.723085880279541, "learning_rate": 1.4924550754913341e-05, "loss": 1.6238, "step": 2212 }, { "epoch": 0.36, "grad_norm": 7.939207553863525, "learning_rate": 1.4920011916298856e-05, "loss": 1.7525, "step": 2213 }, { "epoch": 0.36, "grad_norm": 10.962862014770508, "learning_rate": 1.4915471740005739e-05, "loss": 1.7705, "step": 2214 }, { "epoch": 0.36, "grad_norm": 9.617756843566895, "learning_rate": 1.4910930227268405e-05, "loss": 1.471, "step": 2215 }, { "epoch": 0.36, "grad_norm": 7.230254173278809, "learning_rate": 1.4906387379321623e-05, "loss": 1.6648, "step": 2216 }, { "epoch": 0.36, "grad_norm": 8.00655460357666, "learning_rate": 1.4901843197400528e-05, "loss": 1.788, "step": 2217 }, { "epoch": 0.36, "grad_norm": 10.26345157623291, "learning_rate": 1.489729768274061e-05, "loss": 1.5338, "step": 2218 }, { "epoch": 0.36, "grad_norm": 7.5392327308654785, "learning_rate": 1.4892750836577734e-05, "loss": 1.5547, "step": 2219 }, { "epoch": 0.36, "grad_norm": 6.729666233062744, "learning_rate": 1.4888202660148117e-05, "loss": 1.8416, "step": 2220 }, { "epoch": 0.36, "grad_norm": 8.736335754394531, "learning_rate": 1.4883653154688343e-05, "loss": 1.6727, "step": 2221 }, { "epoch": 0.36, "grad_norm": 7.950988292694092, "learning_rate": 1.4879102321435351e-05, "loss": 1.7817, "step": 2222 }, { "epoch": 0.36, "grad_norm": 8.411725044250488, "learning_rate": 1.487455016162645e-05, "loss": 1.6719, "step": 2223 }, { "epoch": 0.36, "grad_norm": 7.182684421539307, "learning_rate": 1.4869996676499302e-05, "loss": 1.5763, "step": 2224 }, { "epoch": 0.36, "grad_norm": 9.026506423950195, "learning_rate": 1.4865441867291935e-05, "loss": 1.6767, "step": 2225 }, { "epoch": 0.36, "grad_norm": 8.843639373779297, "learning_rate": 1.486088573524273e-05, "loss": 1.5872, "step": 2226 }, { "epoch": 0.36, "grad_norm": 7.900869846343994, "learning_rate": 1.4856328281590442e-05, "loss": 1.5519, "step": 2227 }, { "epoch": 0.36, "grad_norm": 7.5866193771362305, "learning_rate": 1.4851769507574166e-05, "loss": 1.6667, "step": 2228 }, { "epoch": 0.36, "grad_norm": 8.90255355834961, "learning_rate": 1.4847209414433365e-05, "loss": 1.8407, "step": 2229 }, { "epoch": 0.36, "grad_norm": 7.424229621887207, "learning_rate": 1.4842648003407866e-05, "loss": 1.4818, "step": 2230 }, { "epoch": 0.36, "grad_norm": 7.836453914642334, "learning_rate": 1.4838085275737847e-05, "loss": 1.6541, "step": 2231 }, { "epoch": 0.36, "grad_norm": 7.233096122741699, "learning_rate": 1.4833521232663847e-05, "loss": 1.6184, "step": 2232 }, { "epoch": 0.36, "grad_norm": 14.097890853881836, "learning_rate": 1.482895587542676e-05, "loss": 1.6487, "step": 2233 }, { "epoch": 0.36, "grad_norm": 9.082054138183594, "learning_rate": 1.4824389205267847e-05, "loss": 1.7049, "step": 2234 }, { "epoch": 0.36, "grad_norm": 8.06987476348877, "learning_rate": 1.4819821223428708e-05, "loss": 1.7545, "step": 2235 }, { "epoch": 0.36, "grad_norm": 8.439476013183594, "learning_rate": 1.4815251931151313e-05, "loss": 1.6385, "step": 2236 }, { "epoch": 0.36, "grad_norm": 9.493224143981934, "learning_rate": 1.4810681329677988e-05, "loss": 1.6785, "step": 2237 }, { "epoch": 0.36, "grad_norm": 7.4895734786987305, "learning_rate": 1.4806109420251407e-05, "loss": 1.5997, "step": 2238 }, { "epoch": 0.36, "grad_norm": 9.684379577636719, "learning_rate": 1.480153620411461e-05, "loss": 1.7452, "step": 2239 }, { "epoch": 0.36, "grad_norm": 8.282444953918457, "learning_rate": 1.4796961682510986e-05, "loss": 1.6749, "step": 2240 }, { "epoch": 0.36, "grad_norm": 7.907978534698486, "learning_rate": 1.4792385856684278e-05, "loss": 1.7073, "step": 2241 }, { "epoch": 0.36, "grad_norm": 6.4789557456970215, "learning_rate": 1.4787808727878586e-05, "loss": 1.6851, "step": 2242 }, { "epoch": 0.36, "grad_norm": 10.286151885986328, "learning_rate": 1.478323029733836e-05, "loss": 1.9189, "step": 2243 }, { "epoch": 0.36, "grad_norm": 12.708471298217773, "learning_rate": 1.4778650566308418e-05, "loss": 1.8314, "step": 2244 }, { "epoch": 0.36, "grad_norm": 7.6604905128479, "learning_rate": 1.4774069536033913e-05, "loss": 1.7323, "step": 2245 }, { "epoch": 0.36, "grad_norm": 9.065228462219238, "learning_rate": 1.4769487207760362e-05, "loss": 1.8393, "step": 2246 }, { "epoch": 0.36, "grad_norm": 7.984690189361572, "learning_rate": 1.476490358273363e-05, "loss": 1.6702, "step": 2247 }, { "epoch": 0.36, "grad_norm": 9.60896110534668, "learning_rate": 1.4760318662199939e-05, "loss": 1.9215, "step": 2248 }, { "epoch": 0.36, "grad_norm": 12.761642456054688, "learning_rate": 1.4755732447405863e-05, "loss": 1.6824, "step": 2249 }, { "epoch": 0.36, "grad_norm": 9.719700813293457, "learning_rate": 1.4751144939598326e-05, "loss": 1.7932, "step": 2250 }, { "epoch": 0.36, "grad_norm": 7.581261157989502, "learning_rate": 1.4746556140024599e-05, "loss": 1.7122, "step": 2251 }, { "epoch": 0.36, "grad_norm": 9.910577774047852, "learning_rate": 1.474196604993231e-05, "loss": 1.7334, "step": 2252 }, { "epoch": 0.36, "grad_norm": 12.697086334228516, "learning_rate": 1.4737374670569443e-05, "loss": 1.7458, "step": 2253 }, { "epoch": 0.36, "grad_norm": 7.702571392059326, "learning_rate": 1.4732782003184318e-05, "loss": 1.7236, "step": 2254 }, { "epoch": 0.36, "grad_norm": 7.527020454406738, "learning_rate": 1.4728188049025623e-05, "loss": 1.5528, "step": 2255 }, { "epoch": 0.36, "grad_norm": 9.123132705688477, "learning_rate": 1.4723592809342374e-05, "loss": 1.709, "step": 2256 }, { "epoch": 0.36, "grad_norm": 7.19807767868042, "learning_rate": 1.4718996285383957e-05, "loss": 1.5297, "step": 2257 }, { "epoch": 0.36, "grad_norm": 6.265560626983643, "learning_rate": 1.4714398478400101e-05, "loss": 1.8957, "step": 2258 }, { "epoch": 0.36, "grad_norm": 8.30396556854248, "learning_rate": 1.4709799389640875e-05, "loss": 1.649, "step": 2259 }, { "epoch": 0.36, "grad_norm": 7.662035942077637, "learning_rate": 1.470519902035671e-05, "loss": 1.4619, "step": 2260 }, { "epoch": 0.36, "grad_norm": 11.007131576538086, "learning_rate": 1.4700597371798373e-05, "loss": 1.5896, "step": 2261 }, { "epoch": 0.36, "grad_norm": 8.404512405395508, "learning_rate": 1.4695994445216985e-05, "loss": 1.7081, "step": 2262 }, { "epoch": 0.36, "grad_norm": 8.432291984558105, "learning_rate": 1.4691390241864018e-05, "loss": 1.7054, "step": 2263 }, { "epoch": 0.36, "grad_norm": 8.600809097290039, "learning_rate": 1.4686784762991284e-05, "loss": 1.6701, "step": 2264 }, { "epoch": 0.36, "grad_norm": 10.492718696594238, "learning_rate": 1.4682178009850947e-05, "loss": 1.6781, "step": 2265 }, { "epoch": 0.36, "grad_norm": 11.049477577209473, "learning_rate": 1.467756998369551e-05, "loss": 1.8498, "step": 2266 }, { "epoch": 0.36, "grad_norm": 9.205427169799805, "learning_rate": 1.467296068577783e-05, "loss": 1.5646, "step": 2267 }, { "epoch": 0.37, "grad_norm": 7.183644771575928, "learning_rate": 1.466835011735111e-05, "loss": 1.6429, "step": 2268 }, { "epoch": 0.37, "grad_norm": 8.363006591796875, "learning_rate": 1.4663738279668894e-05, "loss": 1.5547, "step": 2269 }, { "epoch": 0.37, "grad_norm": 7.212450981140137, "learning_rate": 1.4659125173985069e-05, "loss": 1.6373, "step": 2270 }, { "epoch": 0.37, "grad_norm": 9.187090873718262, "learning_rate": 1.4654510801553874e-05, "loss": 1.3862, "step": 2271 }, { "epoch": 0.37, "grad_norm": 8.73009967803955, "learning_rate": 1.464989516362989e-05, "loss": 1.5521, "step": 2272 }, { "epoch": 0.37, "grad_norm": 7.769900321960449, "learning_rate": 1.4645278261468037e-05, "loss": 1.5124, "step": 2273 }, { "epoch": 0.37, "grad_norm": 8.567597389221191, "learning_rate": 1.4640660096323583e-05, "loss": 1.709, "step": 2274 }, { "epoch": 0.37, "grad_norm": 7.714412689208984, "learning_rate": 1.463604066945214e-05, "loss": 1.8387, "step": 2275 }, { "epoch": 0.37, "grad_norm": 8.760498046875, "learning_rate": 1.4631419982109665e-05, "loss": 1.8131, "step": 2276 }, { "epoch": 0.37, "grad_norm": 8.937557220458984, "learning_rate": 1.4626798035552453e-05, "loss": 1.7987, "step": 2277 }, { "epoch": 0.37, "grad_norm": 7.930296897888184, "learning_rate": 1.4622174831037138e-05, "loss": 1.5812, "step": 2278 }, { "epoch": 0.37, "grad_norm": 7.894315719604492, "learning_rate": 1.4617550369820706e-05, "loss": 1.7944, "step": 2279 }, { "epoch": 0.37, "grad_norm": 32.01365661621094, "learning_rate": 1.4612924653160478e-05, "loss": 1.6826, "step": 2280 }, { "epoch": 0.37, "grad_norm": 8.764009475708008, "learning_rate": 1.4608297682314122e-05, "loss": 1.8159, "step": 2281 }, { "epoch": 0.37, "grad_norm": 9.13406753540039, "learning_rate": 1.4603669458539638e-05, "loss": 1.566, "step": 2282 }, { "epoch": 0.37, "grad_norm": 8.086601257324219, "learning_rate": 1.4599039983095372e-05, "loss": 1.8096, "step": 2283 }, { "epoch": 0.37, "grad_norm": 8.122062683105469, "learning_rate": 1.4594409257240015e-05, "loss": 1.6189, "step": 2284 }, { "epoch": 0.37, "grad_norm": 9.114706039428711, "learning_rate": 1.4589777282232585e-05, "loss": 1.8034, "step": 2285 }, { "epoch": 0.37, "grad_norm": 9.620138168334961, "learning_rate": 1.4585144059332456e-05, "loss": 1.8001, "step": 2286 }, { "epoch": 0.37, "grad_norm": 8.709541320800781, "learning_rate": 1.458050958979933e-05, "loss": 1.8121, "step": 2287 }, { "epoch": 0.37, "grad_norm": 8.044870376586914, "learning_rate": 1.457587387489325e-05, "loss": 1.6143, "step": 2288 }, { "epoch": 0.37, "grad_norm": 9.759812355041504, "learning_rate": 1.4571236915874595e-05, "loss": 1.6827, "step": 2289 }, { "epoch": 0.37, "grad_norm": 7.391739845275879, "learning_rate": 1.456659871400409e-05, "loss": 1.5992, "step": 2290 }, { "epoch": 0.37, "grad_norm": 10.76621150970459, "learning_rate": 1.4561959270542796e-05, "loss": 1.7496, "step": 2291 }, { "epoch": 0.37, "grad_norm": 6.797476768493652, "learning_rate": 1.4557318586752107e-05, "loss": 1.4449, "step": 2292 }, { "epoch": 0.37, "grad_norm": 7.536102771759033, "learning_rate": 1.4552676663893755e-05, "loss": 1.638, "step": 2293 }, { "epoch": 0.37, "grad_norm": 10.152800559997559, "learning_rate": 1.4548033503229808e-05, "loss": 1.548, "step": 2294 }, { "epoch": 0.37, "grad_norm": 8.879152297973633, "learning_rate": 1.4543389106022676e-05, "loss": 1.8918, "step": 2295 }, { "epoch": 0.37, "grad_norm": 7.591149806976318, "learning_rate": 1.4538743473535104e-05, "loss": 1.5876, "step": 2296 }, { "epoch": 0.37, "grad_norm": 8.6647367477417, "learning_rate": 1.4534096607030167e-05, "loss": 1.7735, "step": 2297 }, { "epoch": 0.37, "grad_norm": 10.877237319946289, "learning_rate": 1.4529448507771285e-05, "loss": 1.7258, "step": 2298 }, { "epoch": 0.37, "grad_norm": 7.769505500793457, "learning_rate": 1.4524799177022197e-05, "loss": 1.6261, "step": 2299 }, { "epoch": 0.37, "grad_norm": 8.148695945739746, "learning_rate": 1.4520148616046998e-05, "loss": 1.8079, "step": 2300 }, { "epoch": 0.37, "grad_norm": 7.706264972686768, "learning_rate": 1.4515496826110104e-05, "loss": 1.4819, "step": 2301 }, { "epoch": 0.37, "grad_norm": 9.944128036499023, "learning_rate": 1.4510843808476268e-05, "loss": 1.7098, "step": 2302 }, { "epoch": 0.37, "grad_norm": 8.701708793640137, "learning_rate": 1.4506189564410574e-05, "loss": 1.7322, "step": 2303 }, { "epoch": 0.37, "grad_norm": 8.886385917663574, "learning_rate": 1.4501534095178442e-05, "loss": 1.8716, "step": 2304 }, { "epoch": 0.37, "grad_norm": 7.356377124786377, "learning_rate": 1.449687740204563e-05, "loss": 1.5686, "step": 2305 }, { "epoch": 0.37, "grad_norm": 8.536746978759766, "learning_rate": 1.4492219486278222e-05, "loss": 1.8634, "step": 2306 }, { "epoch": 0.37, "grad_norm": 13.344511985778809, "learning_rate": 1.4487560349142634e-05, "loss": 1.6862, "step": 2307 }, { "epoch": 0.37, "grad_norm": 9.387551307678223, "learning_rate": 1.4482899991905621e-05, "loss": 1.7231, "step": 2308 }, { "epoch": 0.37, "grad_norm": 9.880509376525879, "learning_rate": 1.4478238415834263e-05, "loss": 1.8816, "step": 2309 }, { "epoch": 0.37, "grad_norm": 8.385004043579102, "learning_rate": 1.4473575622195974e-05, "loss": 1.7751, "step": 2310 }, { "epoch": 0.37, "grad_norm": 8.311614990234375, "learning_rate": 1.4468911612258499e-05, "loss": 1.8948, "step": 2311 }, { "epoch": 0.37, "grad_norm": 7.420595645904541, "learning_rate": 1.4464246387289913e-05, "loss": 1.774, "step": 2312 }, { "epoch": 0.37, "grad_norm": 7.246115207672119, "learning_rate": 1.4459579948558622e-05, "loss": 1.6841, "step": 2313 }, { "epoch": 0.37, "grad_norm": 8.095364570617676, "learning_rate": 1.4454912297333361e-05, "loss": 1.4801, "step": 2314 }, { "epoch": 0.37, "grad_norm": 9.087023735046387, "learning_rate": 1.4450243434883197e-05, "loss": 1.7366, "step": 2315 }, { "epoch": 0.37, "grad_norm": 7.471587181091309, "learning_rate": 1.4445573362477523e-05, "loss": 1.5033, "step": 2316 }, { "epoch": 0.37, "grad_norm": 8.507797241210938, "learning_rate": 1.4440902081386066e-05, "loss": 1.6089, "step": 2317 }, { "epoch": 0.37, "grad_norm": 8.114120483398438, "learning_rate": 1.4436229592878873e-05, "loss": 1.702, "step": 2318 }, { "epoch": 0.37, "grad_norm": 7.127394199371338, "learning_rate": 1.443155589822633e-05, "loss": 1.4874, "step": 2319 }, { "epoch": 0.37, "grad_norm": 8.115497589111328, "learning_rate": 1.4426880998699144e-05, "loss": 1.4493, "step": 2320 }, { "epoch": 0.37, "grad_norm": 7.4448699951171875, "learning_rate": 1.442220489556835e-05, "loss": 1.4606, "step": 2321 }, { "epoch": 0.37, "grad_norm": 8.354195594787598, "learning_rate": 1.441752759010531e-05, "loss": 1.5719, "step": 2322 }, { "epoch": 0.37, "grad_norm": 8.287508010864258, "learning_rate": 1.441284908358172e-05, "loss": 1.4952, "step": 2323 }, { "epoch": 0.37, "grad_norm": 8.10088062286377, "learning_rate": 1.4408169377269592e-05, "loss": 1.6828, "step": 2324 }, { "epoch": 0.37, "grad_norm": 7.591000080108643, "learning_rate": 1.4403488472441272e-05, "loss": 1.7231, "step": 2325 }, { "epoch": 0.37, "grad_norm": 9.436606407165527, "learning_rate": 1.4398806370369425e-05, "loss": 1.5819, "step": 2326 }, { "epoch": 0.37, "grad_norm": 8.102751731872559, "learning_rate": 1.4394123072327049e-05, "loss": 1.8484, "step": 2327 }, { "epoch": 0.37, "grad_norm": 8.406691551208496, "learning_rate": 1.4389438579587458e-05, "loss": 1.8323, "step": 2328 }, { "epoch": 0.37, "grad_norm": 9.758280754089355, "learning_rate": 1.4384752893424307e-05, "loss": 1.6671, "step": 2329 }, { "epoch": 0.38, "grad_norm": 8.23262882232666, "learning_rate": 1.4380066015111557e-05, "loss": 1.592, "step": 2330 }, { "epoch": 0.38, "grad_norm": 8.159443855285645, "learning_rate": 1.4375377945923502e-05, "loss": 1.5132, "step": 2331 }, { "epoch": 0.38, "grad_norm": 10.022614479064941, "learning_rate": 1.437068868713476e-05, "loss": 1.5173, "step": 2332 }, { "epoch": 0.38, "grad_norm": 8.335766792297363, "learning_rate": 1.436599824002027e-05, "loss": 1.9255, "step": 2333 }, { "epoch": 0.38, "grad_norm": 6.430185317993164, "learning_rate": 1.4361306605855297e-05, "loss": 1.6043, "step": 2334 }, { "epoch": 0.38, "grad_norm": 7.731550216674805, "learning_rate": 1.4356613785915425e-05, "loss": 1.4938, "step": 2335 }, { "epoch": 0.38, "grad_norm": 7.280863285064697, "learning_rate": 1.4351919781476564e-05, "loss": 1.4198, "step": 2336 }, { "epoch": 0.38, "grad_norm": 9.04041862487793, "learning_rate": 1.4347224593814946e-05, "loss": 1.5394, "step": 2337 }, { "epoch": 0.38, "grad_norm": 7.44563627243042, "learning_rate": 1.434252822420712e-05, "loss": 1.8131, "step": 2338 }, { "epoch": 0.38, "grad_norm": 8.691362380981445, "learning_rate": 1.4337830673929963e-05, "loss": 1.7125, "step": 2339 }, { "epoch": 0.38, "grad_norm": 8.304543495178223, "learning_rate": 1.4333131944260666e-05, "loss": 1.9687, "step": 2340 }, { "epoch": 0.38, "grad_norm": 8.509227752685547, "learning_rate": 1.4328432036476748e-05, "loss": 1.8389, "step": 2341 }, { "epoch": 0.38, "grad_norm": 7.2592291831970215, "learning_rate": 1.4323730951856036e-05, "loss": 1.7576, "step": 2342 }, { "epoch": 0.38, "grad_norm": 9.584750175476074, "learning_rate": 1.4319028691676698e-05, "loss": 1.8748, "step": 2343 }, { "epoch": 0.38, "grad_norm": 9.119144439697266, "learning_rate": 1.43143252572172e-05, "loss": 1.6308, "step": 2344 }, { "epoch": 0.38, "grad_norm": 8.126891136169434, "learning_rate": 1.4309620649756345e-05, "loss": 1.7554, "step": 2345 }, { "epoch": 0.38, "grad_norm": 7.405935764312744, "learning_rate": 1.4304914870573239e-05, "loss": 1.6869, "step": 2346 }, { "epoch": 0.38, "grad_norm": 7.999362468719482, "learning_rate": 1.430020792094731e-05, "loss": 1.915, "step": 2347 }, { "epoch": 0.38, "grad_norm": 13.355512619018555, "learning_rate": 1.4295499802158319e-05, "loss": 1.6502, "step": 2348 }, { "epoch": 0.38, "grad_norm": 7.897162914276123, "learning_rate": 1.4290790515486328e-05, "loss": 1.5846, "step": 2349 }, { "epoch": 0.38, "grad_norm": 8.589869499206543, "learning_rate": 1.4286080062211726e-05, "loss": 1.6949, "step": 2350 }, { "epoch": 0.38, "grad_norm": 7.456057548522949, "learning_rate": 1.4281368443615211e-05, "loss": 1.7916, "step": 2351 }, { "epoch": 0.38, "grad_norm": 9.550138473510742, "learning_rate": 1.4276655660977804e-05, "loss": 1.4549, "step": 2352 }, { "epoch": 0.38, "grad_norm": 8.84347152709961, "learning_rate": 1.4271941715580842e-05, "loss": 1.7394, "step": 2353 }, { "epoch": 0.38, "grad_norm": 7.669593811035156, "learning_rate": 1.4267226608705978e-05, "loss": 1.6102, "step": 2354 }, { "epoch": 0.38, "grad_norm": 7.7482829093933105, "learning_rate": 1.4262510341635181e-05, "loss": 1.7031, "step": 2355 }, { "epoch": 0.38, "grad_norm": 10.256887435913086, "learning_rate": 1.4257792915650728e-05, "loss": 1.6061, "step": 2356 }, { "epoch": 0.38, "grad_norm": 10.106775283813477, "learning_rate": 1.4253074332035224e-05, "loss": 1.6472, "step": 2357 }, { "epoch": 0.38, "grad_norm": 7.828833103179932, "learning_rate": 1.424835459207158e-05, "loss": 1.599, "step": 2358 }, { "epoch": 0.38, "grad_norm": 6.954387664794922, "learning_rate": 1.4243633697043025e-05, "loss": 1.6643, "step": 2359 }, { "epoch": 0.38, "grad_norm": 8.70785903930664, "learning_rate": 1.4238911648233095e-05, "loss": 1.5019, "step": 2360 }, { "epoch": 0.38, "grad_norm": 13.364802360534668, "learning_rate": 1.4234188446925654e-05, "loss": 1.7772, "step": 2361 }, { "epoch": 0.38, "grad_norm": 8.30273151397705, "learning_rate": 1.4229464094404866e-05, "loss": 1.6791, "step": 2362 }, { "epoch": 0.38, "grad_norm": 10.748398780822754, "learning_rate": 1.4224738591955215e-05, "loss": 1.5105, "step": 2363 }, { "epoch": 0.38, "grad_norm": 10.126462936401367, "learning_rate": 1.4220011940861488e-05, "loss": 1.4405, "step": 2364 }, { "epoch": 0.38, "grad_norm": 9.866246223449707, "learning_rate": 1.4215284142408802e-05, "loss": 1.6112, "step": 2365 }, { "epoch": 0.38, "grad_norm": 11.029294967651367, "learning_rate": 1.4210555197882571e-05, "loss": 1.7354, "step": 2366 }, { "epoch": 0.38, "grad_norm": 8.227192878723145, "learning_rate": 1.4205825108568525e-05, "loss": 1.6488, "step": 2367 }, { "epoch": 0.38, "grad_norm": 8.255217552185059, "learning_rate": 1.4201093875752703e-05, "loss": 1.6017, "step": 2368 }, { "epoch": 0.38, "grad_norm": 15.324321746826172, "learning_rate": 1.4196361500721464e-05, "loss": 1.6936, "step": 2369 }, { "epoch": 0.38, "grad_norm": 8.610987663269043, "learning_rate": 1.4191627984761462e-05, "loss": 1.5404, "step": 2370 }, { "epoch": 0.38, "grad_norm": 8.52711009979248, "learning_rate": 1.4186893329159679e-05, "loss": 1.5256, "step": 2371 }, { "epoch": 0.38, "grad_norm": 8.849525451660156, "learning_rate": 1.4182157535203394e-05, "loss": 1.6757, "step": 2372 }, { "epoch": 0.38, "grad_norm": 10.370574951171875, "learning_rate": 1.4177420604180196e-05, "loss": 1.8949, "step": 2373 }, { "epoch": 0.38, "grad_norm": 8.577253341674805, "learning_rate": 1.4172682537377993e-05, "loss": 1.7102, "step": 2374 }, { "epoch": 0.38, "grad_norm": 11.849950790405273, "learning_rate": 1.4167943336084992e-05, "loss": 1.6338, "step": 2375 }, { "epoch": 0.38, "grad_norm": 9.64438533782959, "learning_rate": 1.4163203001589713e-05, "loss": 1.5198, "step": 2376 }, { "epoch": 0.38, "grad_norm": 7.604032039642334, "learning_rate": 1.4158461535180983e-05, "loss": 1.7854, "step": 2377 }, { "epoch": 0.38, "grad_norm": 7.231365203857422, "learning_rate": 1.4153718938147936e-05, "loss": 1.4625, "step": 2378 }, { "epoch": 0.38, "grad_norm": 10.127460479736328, "learning_rate": 1.4148975211780014e-05, "loss": 1.3934, "step": 2379 }, { "epoch": 0.38, "grad_norm": 12.409961700439453, "learning_rate": 1.4144230357366964e-05, "loss": 1.6031, "step": 2380 }, { "epoch": 0.38, "grad_norm": 12.239802360534668, "learning_rate": 1.4139484376198847e-05, "loss": 1.6859, "step": 2381 }, { "epoch": 0.38, "grad_norm": 8.532365798950195, "learning_rate": 1.4134737269566023e-05, "loss": 1.6878, "step": 2382 }, { "epoch": 0.38, "grad_norm": 12.20368766784668, "learning_rate": 1.412998903875916e-05, "loss": 1.6688, "step": 2383 }, { "epoch": 0.38, "grad_norm": 12.725234985351562, "learning_rate": 1.412523968506923e-05, "loss": 1.6718, "step": 2384 }, { "epoch": 0.38, "grad_norm": 13.203750610351562, "learning_rate": 1.4120489209787513e-05, "loss": 1.7881, "step": 2385 }, { "epoch": 0.38, "grad_norm": 7.875146389007568, "learning_rate": 1.4115737614205598e-05, "loss": 1.6756, "step": 2386 }, { "epoch": 0.38, "grad_norm": 10.885321617126465, "learning_rate": 1.4110984899615367e-05, "loss": 1.378, "step": 2387 }, { "epoch": 0.38, "grad_norm": 12.101174354553223, "learning_rate": 1.4106231067309018e-05, "loss": 1.6639, "step": 2388 }, { "epoch": 0.38, "grad_norm": 10.060187339782715, "learning_rate": 1.4101476118579042e-05, "loss": 1.8413, "step": 2389 }, { "epoch": 0.38, "grad_norm": 8.607748985290527, "learning_rate": 1.4096720054718245e-05, "loss": 1.6611, "step": 2390 }, { "epoch": 0.38, "grad_norm": 7.912643909454346, "learning_rate": 1.4091962877019729e-05, "loss": 1.6185, "step": 2391 }, { "epoch": 0.39, "grad_norm": 7.996652603149414, "learning_rate": 1.4087204586776901e-05, "loss": 1.681, "step": 2392 }, { "epoch": 0.39, "grad_norm": 12.838080406188965, "learning_rate": 1.4082445185283469e-05, "loss": 1.5898, "step": 2393 }, { "epoch": 0.39, "grad_norm": 12.905715942382812, "learning_rate": 1.4077684673833438e-05, "loss": 1.6907, "step": 2394 }, { "epoch": 0.39, "grad_norm": 8.123584747314453, "learning_rate": 1.407292305372113e-05, "loss": 1.6346, "step": 2395 }, { "epoch": 0.39, "grad_norm": 8.720720291137695, "learning_rate": 1.4068160326241158e-05, "loss": 1.5986, "step": 2396 }, { "epoch": 0.39, "grad_norm": 10.408102035522461, "learning_rate": 1.4063396492688434e-05, "loss": 1.8749, "step": 2397 }, { "epoch": 0.39, "grad_norm": 12.017396926879883, "learning_rate": 1.4058631554358177e-05, "loss": 1.7814, "step": 2398 }, { "epoch": 0.39, "grad_norm": 7.508077144622803, "learning_rate": 1.4053865512545896e-05, "loss": 1.6289, "step": 2399 }, { "epoch": 0.39, "grad_norm": 8.10350227355957, "learning_rate": 1.4049098368547419e-05, "loss": 1.8975, "step": 2400 }, { "epoch": 0.39, "grad_norm": 8.172501564025879, "learning_rate": 1.4044330123658854e-05, "loss": 1.6852, "step": 2401 }, { "epoch": 0.39, "grad_norm": 7.86643648147583, "learning_rate": 1.4039560779176617e-05, "loss": 1.4868, "step": 2402 }, { "epoch": 0.39, "grad_norm": 10.49643325805664, "learning_rate": 1.4034790336397427e-05, "loss": 1.6241, "step": 2403 }, { "epoch": 0.39, "grad_norm": 8.672849655151367, "learning_rate": 1.403001879661829e-05, "loss": 1.7966, "step": 2404 }, { "epoch": 0.39, "grad_norm": 8.60525131225586, "learning_rate": 1.4025246161136524e-05, "loss": 1.7435, "step": 2405 }, { "epoch": 0.39, "grad_norm": 7.450148105621338, "learning_rate": 1.4020472431249734e-05, "loss": 1.6475, "step": 2406 }, { "epoch": 0.39, "grad_norm": 9.515373229980469, "learning_rate": 1.4015697608255829e-05, "loss": 1.5272, "step": 2407 }, { "epoch": 0.39, "grad_norm": 10.54525089263916, "learning_rate": 1.401092169345301e-05, "loss": 1.7986, "step": 2408 }, { "epoch": 0.39, "grad_norm": 9.22631549835205, "learning_rate": 1.4006144688139782e-05, "loss": 1.814, "step": 2409 }, { "epoch": 0.39, "grad_norm": 8.630343437194824, "learning_rate": 1.4001366593614938e-05, "loss": 1.9157, "step": 2410 }, { "epoch": 0.39, "grad_norm": 9.949531555175781, "learning_rate": 1.3996587411177576e-05, "loss": 1.8653, "step": 2411 }, { "epoch": 0.39, "grad_norm": 7.359720706939697, "learning_rate": 1.3991807142127082e-05, "loss": 1.7286, "step": 2412 }, { "epoch": 0.39, "grad_norm": 11.190728187561035, "learning_rate": 1.3987025787763138e-05, "loss": 1.5402, "step": 2413 }, { "epoch": 0.39, "grad_norm": 8.203958511352539, "learning_rate": 1.3982243349385726e-05, "loss": 1.5904, "step": 2414 }, { "epoch": 0.39, "grad_norm": 7.968279838562012, "learning_rate": 1.3977459828295126e-05, "loss": 1.5529, "step": 2415 }, { "epoch": 0.39, "grad_norm": 7.677429676055908, "learning_rate": 1.3972675225791898e-05, "loss": 1.7326, "step": 2416 }, { "epoch": 0.39, "grad_norm": 6.646246433258057, "learning_rate": 1.3967889543176912e-05, "loss": 1.4081, "step": 2417 }, { "epoch": 0.39, "grad_norm": 10.398520469665527, "learning_rate": 1.3963102781751319e-05, "loss": 1.7368, "step": 2418 }, { "epoch": 0.39, "grad_norm": 8.615803718566895, "learning_rate": 1.3958314942816571e-05, "loss": 1.6525, "step": 2419 }, { "epoch": 0.39, "grad_norm": 7.29184103012085, "learning_rate": 1.3953526027674412e-05, "loss": 1.6395, "step": 2420 }, { "epoch": 0.39, "grad_norm": 8.068448066711426, "learning_rate": 1.3948736037626873e-05, "loss": 1.6717, "step": 2421 }, { "epoch": 0.39, "grad_norm": 7.606424331665039, "learning_rate": 1.3943944973976287e-05, "loss": 1.9657, "step": 2422 }, { "epoch": 0.39, "grad_norm": 8.875410079956055, "learning_rate": 1.393915283802527e-05, "loss": 1.5471, "step": 2423 }, { "epoch": 0.39, "grad_norm": 7.6394944190979, "learning_rate": 1.3934359631076737e-05, "loss": 1.7033, "step": 2424 }, { "epoch": 0.39, "grad_norm": 8.032715797424316, "learning_rate": 1.3929565354433885e-05, "loss": 1.4963, "step": 2425 }, { "epoch": 0.39, "grad_norm": 7.426381587982178, "learning_rate": 1.392477000940021e-05, "loss": 1.6909, "step": 2426 }, { "epoch": 0.39, "grad_norm": 8.138448715209961, "learning_rate": 1.3919973597279498e-05, "loss": 1.7059, "step": 2427 }, { "epoch": 0.39, "grad_norm": 8.37482738494873, "learning_rate": 1.3915176119375812e-05, "loss": 1.6601, "step": 2428 }, { "epoch": 0.39, "grad_norm": 9.358864784240723, "learning_rate": 1.391037757699353e-05, "loss": 1.6817, "step": 2429 }, { "epoch": 0.39, "grad_norm": 8.330999374389648, "learning_rate": 1.3905577971437296e-05, "loss": 1.7122, "step": 2430 }, { "epoch": 0.39, "grad_norm": 7.692313194274902, "learning_rate": 1.3900777304012054e-05, "loss": 1.5939, "step": 2431 }, { "epoch": 0.39, "grad_norm": 7.445542335510254, "learning_rate": 1.3895975576023033e-05, "loss": 1.5126, "step": 2432 }, { "epoch": 0.39, "grad_norm": 7.787473201751709, "learning_rate": 1.3891172788775756e-05, "loss": 1.6076, "step": 2433 }, { "epoch": 0.39, "grad_norm": 7.6419806480407715, "learning_rate": 1.3886368943576027e-05, "loss": 1.5599, "step": 2434 }, { "epoch": 0.39, "grad_norm": 9.954404830932617, "learning_rate": 1.388156404172994e-05, "loss": 1.7731, "step": 2435 }, { "epoch": 0.39, "grad_norm": 8.541272163391113, "learning_rate": 1.3876758084543877e-05, "loss": 1.5329, "step": 2436 }, { "epoch": 0.39, "grad_norm": 8.202839851379395, "learning_rate": 1.3871951073324508e-05, "loss": 1.8158, "step": 2437 }, { "epoch": 0.39, "grad_norm": 7.130127906799316, "learning_rate": 1.386714300937879e-05, "loss": 1.574, "step": 2438 }, { "epoch": 0.39, "grad_norm": 8.305237770080566, "learning_rate": 1.3862333894013961e-05, "loss": 1.6595, "step": 2439 }, { "epoch": 0.39, "grad_norm": 7.533670425415039, "learning_rate": 1.3857523728537555e-05, "loss": 1.5354, "step": 2440 }, { "epoch": 0.39, "grad_norm": 8.667236328125, "learning_rate": 1.3852712514257373e-05, "loss": 1.6035, "step": 2441 }, { "epoch": 0.39, "grad_norm": 7.192141532897949, "learning_rate": 1.3847900252481526e-05, "loss": 1.5566, "step": 2442 }, { "epoch": 0.39, "grad_norm": 8.762961387634277, "learning_rate": 1.3843086944518389e-05, "loss": 1.7449, "step": 2443 }, { "epoch": 0.39, "grad_norm": 10.30605697631836, "learning_rate": 1.3838272591676631e-05, "loss": 1.7584, "step": 2444 }, { "epoch": 0.39, "grad_norm": 8.924623489379883, "learning_rate": 1.3833457195265209e-05, "loss": 1.663, "step": 2445 }, { "epoch": 0.39, "grad_norm": 10.116806983947754, "learning_rate": 1.3828640756593348e-05, "loss": 1.532, "step": 2446 }, { "epoch": 0.39, "grad_norm": 7.673985004425049, "learning_rate": 1.3823823276970574e-05, "loss": 1.7325, "step": 2447 }, { "epoch": 0.39, "grad_norm": 8.679394721984863, "learning_rate": 1.3819004757706686e-05, "loss": 1.6008, "step": 2448 }, { "epoch": 0.39, "grad_norm": 8.052361488342285, "learning_rate": 1.3814185200111772e-05, "loss": 1.6711, "step": 2449 }, { "epoch": 0.39, "grad_norm": 7.607576847076416, "learning_rate": 1.3809364605496197e-05, "loss": 1.835, "step": 2450 }, { "epoch": 0.39, "grad_norm": 6.708164215087891, "learning_rate": 1.38045429751706e-05, "loss": 1.4109, "step": 2451 }, { "epoch": 0.39, "grad_norm": 10.559837341308594, "learning_rate": 1.3799720310445923e-05, "loss": 1.767, "step": 2452 }, { "epoch": 0.39, "grad_norm": 8.57465648651123, "learning_rate": 1.3794896612633375e-05, "loss": 1.6957, "step": 2453 }, { "epoch": 0.4, "grad_norm": 8.005013465881348, "learning_rate": 1.3790071883044447e-05, "loss": 1.6534, "step": 2454 }, { "epoch": 0.4, "grad_norm": 9.74360179901123, "learning_rate": 1.3785246122990912e-05, "loss": 1.7168, "step": 2455 }, { "epoch": 0.4, "grad_norm": 10.455878257751465, "learning_rate": 1.378041933378482e-05, "loss": 1.6732, "step": 2456 }, { "epoch": 0.4, "grad_norm": 8.234244346618652, "learning_rate": 1.3775591516738507e-05, "loss": 1.8004, "step": 2457 }, { "epoch": 0.4, "grad_norm": 8.77137279510498, "learning_rate": 1.3770762673164586e-05, "loss": 1.7529, "step": 2458 }, { "epoch": 0.4, "grad_norm": 7.125123023986816, "learning_rate": 1.3765932804375945e-05, "loss": 1.6374, "step": 2459 }, { "epoch": 0.4, "grad_norm": 10.220600128173828, "learning_rate": 1.3761101911685756e-05, "loss": 1.8067, "step": 2460 }, { "epoch": 0.4, "grad_norm": 7.6228203773498535, "learning_rate": 1.3756269996407468e-05, "loss": 1.5436, "step": 2461 }, { "epoch": 0.4, "grad_norm": 6.743691921234131, "learning_rate": 1.3751437059854809e-05, "loss": 1.5971, "step": 2462 }, { "epoch": 0.4, "grad_norm": 8.103483200073242, "learning_rate": 1.3746603103341775e-05, "loss": 1.6857, "step": 2463 }, { "epoch": 0.4, "grad_norm": 8.970428466796875, "learning_rate": 1.3741768128182658e-05, "loss": 1.5459, "step": 2464 }, { "epoch": 0.4, "grad_norm": 7.930623531341553, "learning_rate": 1.373693213569201e-05, "loss": 1.7025, "step": 2465 }, { "epoch": 0.4, "grad_norm": 9.139039039611816, "learning_rate": 1.3732095127184668e-05, "loss": 1.7127, "step": 2466 }, { "epoch": 0.4, "grad_norm": 8.43200397491455, "learning_rate": 1.3727257103975741e-05, "loss": 1.6124, "step": 2467 }, { "epoch": 0.4, "grad_norm": 7.521298408508301, "learning_rate": 1.372241806738062e-05, "loss": 1.5238, "step": 2468 }, { "epoch": 0.4, "grad_norm": 10.36978530883789, "learning_rate": 1.3717578018714965e-05, "loss": 1.7796, "step": 2469 }, { "epoch": 0.4, "grad_norm": 9.281477928161621, "learning_rate": 1.3712736959294715e-05, "loss": 1.4141, "step": 2470 }, { "epoch": 0.4, "grad_norm": 13.492668151855469, "learning_rate": 1.3707894890436083e-05, "loss": 1.77, "step": 2471 }, { "epoch": 0.4, "grad_norm": 7.433669090270996, "learning_rate": 1.3703051813455552e-05, "loss": 1.4941, "step": 2472 }, { "epoch": 0.4, "grad_norm": 7.002445220947266, "learning_rate": 1.3698207729669888e-05, "loss": 1.7721, "step": 2473 }, { "epoch": 0.4, "grad_norm": 8.733394622802734, "learning_rate": 1.3693362640396122e-05, "loss": 1.7093, "step": 2474 }, { "epoch": 0.4, "grad_norm": 12.425880432128906, "learning_rate": 1.3688516546951564e-05, "loss": 1.6772, "step": 2475 }, { "epoch": 0.4, "grad_norm": 7.612839221954346, "learning_rate": 1.3683669450653798e-05, "loss": 1.5187, "step": 2476 }, { "epoch": 0.4, "grad_norm": 7.501223087310791, "learning_rate": 1.3678821352820677e-05, "loss": 1.6899, "step": 2477 }, { "epoch": 0.4, "grad_norm": 10.324125289916992, "learning_rate": 1.3673972254770324e-05, "loss": 1.844, "step": 2478 }, { "epoch": 0.4, "grad_norm": 8.921401023864746, "learning_rate": 1.366912215782114e-05, "loss": 1.7649, "step": 2479 }, { "epoch": 0.4, "grad_norm": 7.345307350158691, "learning_rate": 1.3664271063291789e-05, "loss": 1.8137, "step": 2480 }, { "epoch": 0.4, "grad_norm": 7.726607799530029, "learning_rate": 1.3659418972501221e-05, "loss": 1.7328, "step": 2481 }, { "epoch": 0.4, "grad_norm": 12.637697219848633, "learning_rate": 1.3654565886768644e-05, "loss": 1.7143, "step": 2482 }, { "epoch": 0.4, "grad_norm": 8.227235794067383, "learning_rate": 1.3649711807413541e-05, "loss": 1.6307, "step": 2483 }, { "epoch": 0.4, "grad_norm": 6.884199619293213, "learning_rate": 1.3644856735755657e-05, "loss": 1.6252, "step": 2484 }, { "epoch": 0.4, "grad_norm": 8.337343215942383, "learning_rate": 1.3640000673115028e-05, "loss": 1.6959, "step": 2485 }, { "epoch": 0.4, "grad_norm": 7.590356349945068, "learning_rate": 1.3635143620811935e-05, "loss": 1.4929, "step": 2486 }, { "epoch": 0.4, "grad_norm": 14.35571575164795, "learning_rate": 1.3630285580166946e-05, "loss": 1.8481, "step": 2487 }, { "epoch": 0.4, "grad_norm": 7.413284778594971, "learning_rate": 1.3625426552500885e-05, "loss": 1.5427, "step": 2488 }, { "epoch": 0.4, "grad_norm": 6.877005577087402, "learning_rate": 1.362056653913485e-05, "loss": 1.6456, "step": 2489 }, { "epoch": 0.4, "grad_norm": 7.968170166015625, "learning_rate": 1.361570554139021e-05, "loss": 1.7126, "step": 2490 }, { "epoch": 0.4, "grad_norm": 11.87096881866455, "learning_rate": 1.3610843560588598e-05, "loss": 1.7304, "step": 2491 }, { "epoch": 0.4, "grad_norm": 8.668758392333984, "learning_rate": 1.3605980598051915e-05, "loss": 1.7324, "step": 2492 }, { "epoch": 0.4, "grad_norm": 8.67383098602295, "learning_rate": 1.3601116655102327e-05, "loss": 1.6917, "step": 2493 }, { "epoch": 0.4, "grad_norm": 7.115024566650391, "learning_rate": 1.3596251733062269e-05, "loss": 1.4674, "step": 2494 }, { "epoch": 0.4, "grad_norm": 9.846492767333984, "learning_rate": 1.3591385833254443e-05, "loss": 1.6584, "step": 2495 }, { "epoch": 0.4, "grad_norm": 8.988268852233887, "learning_rate": 1.3586518957001814e-05, "loss": 1.8742, "step": 2496 }, { "epoch": 0.4, "grad_norm": 8.095165252685547, "learning_rate": 1.358165110562761e-05, "loss": 1.6171, "step": 2497 }, { "epoch": 0.4, "grad_norm": 10.114738464355469, "learning_rate": 1.3576782280455336e-05, "loss": 1.6212, "step": 2498 }, { "epoch": 0.4, "grad_norm": 8.841646194458008, "learning_rate": 1.3571912482808749e-05, "loss": 1.8568, "step": 2499 }, { "epoch": 0.4, "grad_norm": 13.514656066894531, "learning_rate": 1.3567041714011877e-05, "loss": 1.5295, "step": 2500 }, { "epoch": 0.4, "grad_norm": 8.382092475891113, "learning_rate": 1.3562169975389009e-05, "loss": 1.6268, "step": 2501 }, { "epoch": 0.4, "grad_norm": 8.945272445678711, "learning_rate": 1.3557297268264696e-05, "loss": 1.626, "step": 2502 }, { "epoch": 0.4, "grad_norm": 7.8293375968933105, "learning_rate": 1.3552423593963757e-05, "loss": 1.599, "step": 2503 }, { "epoch": 0.4, "grad_norm": 7.847530841827393, "learning_rate": 1.3547548953811274e-05, "loss": 1.6909, "step": 2504 }, { "epoch": 0.4, "grad_norm": 8.924927711486816, "learning_rate": 1.3542673349132591e-05, "loss": 1.7467, "step": 2505 }, { "epoch": 0.4, "grad_norm": 8.891215324401855, "learning_rate": 1.3537796781253309e-05, "loss": 1.5207, "step": 2506 }, { "epoch": 0.4, "grad_norm": 13.045597076416016, "learning_rate": 1.3532919251499297e-05, "loss": 1.8295, "step": 2507 }, { "epoch": 0.4, "grad_norm": 8.741211891174316, "learning_rate": 1.3528040761196681e-05, "loss": 1.6907, "step": 2508 }, { "epoch": 0.4, "grad_norm": 9.930217742919922, "learning_rate": 1.3523161311671852e-05, "loss": 1.3462, "step": 2509 }, { "epoch": 0.4, "grad_norm": 13.52177619934082, "learning_rate": 1.3518280904251464e-05, "loss": 1.8132, "step": 2510 }, { "epoch": 0.4, "grad_norm": 9.461336135864258, "learning_rate": 1.3513399540262422e-05, "loss": 1.6658, "step": 2511 }, { "epoch": 0.4, "grad_norm": 6.261139392852783, "learning_rate": 1.3508517221031898e-05, "loss": 1.4418, "step": 2512 }, { "epoch": 0.4, "grad_norm": 9.905787467956543, "learning_rate": 1.3503633947887325e-05, "loss": 1.5543, "step": 2513 }, { "epoch": 0.4, "grad_norm": 7.043800354003906, "learning_rate": 1.3498749722156395e-05, "loss": 1.5196, "step": 2514 }, { "epoch": 0.4, "grad_norm": 9.436563491821289, "learning_rate": 1.3493864545167049e-05, "loss": 1.6271, "step": 2515 }, { "epoch": 0.41, "grad_norm": 7.78753137588501, "learning_rate": 1.34889784182475e-05, "loss": 1.5594, "step": 2516 }, { "epoch": 0.41, "grad_norm": 7.745561122894287, "learning_rate": 1.3484091342726213e-05, "loss": 1.5656, "step": 2517 }, { "epoch": 0.41, "grad_norm": 12.561630249023438, "learning_rate": 1.3479203319931911e-05, "loss": 1.4927, "step": 2518 }, { "epoch": 0.41, "grad_norm": 10.15830135345459, "learning_rate": 1.3474314351193577e-05, "loss": 1.7045, "step": 2519 }, { "epoch": 0.41, "grad_norm": 8.77077865600586, "learning_rate": 1.3469424437840449e-05, "loss": 1.7068, "step": 2520 }, { "epoch": 0.41, "grad_norm": 9.101191520690918, "learning_rate": 1.346453358120202e-05, "loss": 1.838, "step": 2521 }, { "epoch": 0.41, "grad_norm": 8.955343246459961, "learning_rate": 1.3459641782608041e-05, "loss": 1.7003, "step": 2522 }, { "epoch": 0.41, "grad_norm": 8.166020393371582, "learning_rate": 1.3454749043388527e-05, "loss": 1.5809, "step": 2523 }, { "epoch": 0.41, "grad_norm": 7.841664791107178, "learning_rate": 1.3449855364873733e-05, "loss": 1.5504, "step": 2524 }, { "epoch": 0.41, "grad_norm": 8.383543014526367, "learning_rate": 1.3444960748394185e-05, "loss": 1.7466, "step": 2525 }, { "epoch": 0.41, "grad_norm": 9.330923080444336, "learning_rate": 1.3440065195280652e-05, "loss": 1.7937, "step": 2526 }, { "epoch": 0.41, "grad_norm": 9.751694679260254, "learning_rate": 1.3435168706864162e-05, "loss": 1.6994, "step": 2527 }, { "epoch": 0.41, "grad_norm": 6.119568347930908, "learning_rate": 1.3430271284476e-05, "loss": 1.6367, "step": 2528 }, { "epoch": 0.41, "grad_norm": 8.701218605041504, "learning_rate": 1.3425372929447708e-05, "loss": 1.3921, "step": 2529 }, { "epoch": 0.41, "grad_norm": 10.599954605102539, "learning_rate": 1.3420473643111067e-05, "loss": 1.8743, "step": 2530 }, { "epoch": 0.41, "grad_norm": 8.413649559020996, "learning_rate": 1.3415573426798127e-05, "loss": 1.4953, "step": 2531 }, { "epoch": 0.41, "grad_norm": 7.017732620239258, "learning_rate": 1.3410672281841182e-05, "loss": 1.6082, "step": 2532 }, { "epoch": 0.41, "grad_norm": 7.267090797424316, "learning_rate": 1.3405770209572781e-05, "loss": 1.6095, "step": 2533 }, { "epoch": 0.41, "grad_norm": 6.7213335037231445, "learning_rate": 1.3400867211325727e-05, "loss": 1.4659, "step": 2534 }, { "epoch": 0.41, "grad_norm": 8.905498504638672, "learning_rate": 1.3395963288433071e-05, "loss": 1.5567, "step": 2535 }, { "epoch": 0.41, "grad_norm": 12.308056831359863, "learning_rate": 1.3391058442228117e-05, "loss": 1.3576, "step": 2536 }, { "epoch": 0.41, "grad_norm": 8.488703727722168, "learning_rate": 1.3386152674044421e-05, "loss": 1.5911, "step": 2537 }, { "epoch": 0.41, "grad_norm": 7.010749816894531, "learning_rate": 1.3381245985215791e-05, "loss": 1.5311, "step": 2538 }, { "epoch": 0.41, "grad_norm": 9.422577857971191, "learning_rate": 1.3376338377076281e-05, "loss": 1.5854, "step": 2539 }, { "epoch": 0.41, "grad_norm": 8.735539436340332, "learning_rate": 1.3371429850960198e-05, "loss": 1.7752, "step": 2540 }, { "epoch": 0.41, "grad_norm": 7.023444175720215, "learning_rate": 1.3366520408202095e-05, "loss": 1.4149, "step": 2541 }, { "epoch": 0.41, "grad_norm": 8.428343772888184, "learning_rate": 1.3361610050136781e-05, "loss": 1.6671, "step": 2542 }, { "epoch": 0.41, "grad_norm": 6.901571750640869, "learning_rate": 1.3356698778099309e-05, "loss": 1.6531, "step": 2543 }, { "epoch": 0.41, "grad_norm": 6.77885627746582, "learning_rate": 1.3351786593424977e-05, "loss": 1.3153, "step": 2544 }, { "epoch": 0.41, "grad_norm": 7.812315464019775, "learning_rate": 1.3346873497449342e-05, "loss": 1.6585, "step": 2545 }, { "epoch": 0.41, "grad_norm": 7.887238025665283, "learning_rate": 1.3341959491508201e-05, "loss": 1.6563, "step": 2546 }, { "epoch": 0.41, "grad_norm": 8.44024658203125, "learning_rate": 1.3337044576937596e-05, "loss": 1.6775, "step": 2547 }, { "epoch": 0.41, "grad_norm": 7.127779006958008, "learning_rate": 1.3332128755073823e-05, "loss": 1.6301, "step": 2548 }, { "epoch": 0.41, "grad_norm": 8.040322303771973, "learning_rate": 1.3327212027253418e-05, "loss": 1.6125, "step": 2549 }, { "epoch": 0.41, "grad_norm": 9.053622245788574, "learning_rate": 1.3322294394813171e-05, "loss": 1.8223, "step": 2550 }, { "epoch": 0.41, "grad_norm": 10.658293724060059, "learning_rate": 1.3317375859090111e-05, "loss": 1.5122, "step": 2551 }, { "epoch": 0.41, "grad_norm": 7.860474586486816, "learning_rate": 1.3312456421421519e-05, "loss": 1.4944, "step": 2552 }, { "epoch": 0.41, "grad_norm": 8.889870643615723, "learning_rate": 1.330753608314491e-05, "loss": 1.7863, "step": 2553 }, { "epoch": 0.41, "grad_norm": 6.989866256713867, "learning_rate": 1.3302614845598058e-05, "loss": 1.644, "step": 2554 }, { "epoch": 0.41, "grad_norm": 10.54224967956543, "learning_rate": 1.3297692710118973e-05, "loss": 1.5202, "step": 2555 }, { "epoch": 0.41, "grad_norm": 9.343140602111816, "learning_rate": 1.329276967804591e-05, "loss": 1.4255, "step": 2556 }, { "epoch": 0.41, "grad_norm": 7.223217487335205, "learning_rate": 1.328784575071737e-05, "loss": 1.7175, "step": 2557 }, { "epoch": 0.41, "grad_norm": 8.382204055786133, "learning_rate": 1.3282920929472093e-05, "loss": 1.7662, "step": 2558 }, { "epoch": 0.41, "grad_norm": 8.291699409484863, "learning_rate": 1.3277995215649071e-05, "loss": 1.4809, "step": 2559 }, { "epoch": 0.41, "grad_norm": 6.81038761138916, "learning_rate": 1.3273068610587528e-05, "loss": 1.6127, "step": 2560 }, { "epoch": 0.41, "grad_norm": 11.430822372436523, "learning_rate": 1.3268141115626938e-05, "loss": 1.5507, "step": 2561 }, { "epoch": 0.41, "grad_norm": 7.45315408706665, "learning_rate": 1.3263212732107014e-05, "loss": 1.71, "step": 2562 }, { "epoch": 0.41, "grad_norm": 10.070817947387695, "learning_rate": 1.325828346136771e-05, "loss": 1.7401, "step": 2563 }, { "epoch": 0.41, "grad_norm": 8.059075355529785, "learning_rate": 1.3253353304749218e-05, "loss": 1.5687, "step": 2564 }, { "epoch": 0.41, "grad_norm": 8.246824264526367, "learning_rate": 1.324842226359198e-05, "loss": 1.8108, "step": 2565 }, { "epoch": 0.41, "grad_norm": 8.136686325073242, "learning_rate": 1.3243490339236674e-05, "loss": 1.5988, "step": 2566 }, { "epoch": 0.41, "grad_norm": 9.943509101867676, "learning_rate": 1.3238557533024215e-05, "loss": 1.5376, "step": 2567 }, { "epoch": 0.41, "grad_norm": 7.9106903076171875, "learning_rate": 1.3233623846295759e-05, "loss": 1.5409, "step": 2568 }, { "epoch": 0.41, "grad_norm": 7.392731666564941, "learning_rate": 1.3228689280392706e-05, "loss": 1.8026, "step": 2569 }, { "epoch": 0.41, "grad_norm": 8.37217903137207, "learning_rate": 1.3223753836656685e-05, "loss": 1.4718, "step": 2570 }, { "epoch": 0.41, "grad_norm": 8.451284408569336, "learning_rate": 1.3218817516429579e-05, "loss": 1.5081, "step": 2571 }, { "epoch": 0.41, "grad_norm": 7.766127586364746, "learning_rate": 1.3213880321053498e-05, "loss": 1.723, "step": 2572 }, { "epoch": 0.41, "grad_norm": 10.289220809936523, "learning_rate": 1.3208942251870791e-05, "loss": 1.5119, "step": 2573 }, { "epoch": 0.41, "grad_norm": 7.753491401672363, "learning_rate": 1.3204003310224042e-05, "loss": 1.5469, "step": 2574 }, { "epoch": 0.41, "grad_norm": 6.9890007972717285, "learning_rate": 1.3199063497456084e-05, "loss": 1.7696, "step": 2575 }, { "epoch": 0.41, "grad_norm": 7.668457984924316, "learning_rate": 1.3194122814909975e-05, "loss": 1.6098, "step": 2576 }, { "epoch": 0.41, "grad_norm": 8.03205680847168, "learning_rate": 1.3189181263929019e-05, "loss": 1.6294, "step": 2577 }, { "epoch": 0.42, "grad_norm": 10.120912551879883, "learning_rate": 1.3184238845856743e-05, "loss": 1.6045, "step": 2578 }, { "epoch": 0.42, "grad_norm": 7.083071708679199, "learning_rate": 1.317929556203692e-05, "loss": 1.6149, "step": 2579 }, { "epoch": 0.42, "grad_norm": 6.890326976776123, "learning_rate": 1.3174351413813557e-05, "loss": 1.5625, "step": 2580 }, { "epoch": 0.42, "grad_norm": 8.437628746032715, "learning_rate": 1.3169406402530895e-05, "loss": 1.2192, "step": 2581 }, { "epoch": 0.42, "grad_norm": 7.207089900970459, "learning_rate": 1.3164460529533412e-05, "loss": 1.681, "step": 2582 }, { "epoch": 0.42, "grad_norm": 7.856481075286865, "learning_rate": 1.315951379616581e-05, "loss": 1.5063, "step": 2583 }, { "epoch": 0.42, "grad_norm": 7.733956336975098, "learning_rate": 1.315456620377304e-05, "loss": 1.64, "step": 2584 }, { "epoch": 0.42, "grad_norm": 10.579957962036133, "learning_rate": 1.3149617753700281e-05, "loss": 1.6192, "step": 2585 }, { "epoch": 0.42, "grad_norm": 8.51645278930664, "learning_rate": 1.3144668447292936e-05, "loss": 1.5424, "step": 2586 }, { "epoch": 0.42, "grad_norm": 8.417981147766113, "learning_rate": 1.3139718285896657e-05, "loss": 1.5724, "step": 2587 }, { "epoch": 0.42, "grad_norm": 7.829004287719727, "learning_rate": 1.313476727085731e-05, "loss": 1.6384, "step": 2588 }, { "epoch": 0.42, "grad_norm": 7.5569281578063965, "learning_rate": 1.3129815403521008e-05, "loss": 1.6305, "step": 2589 }, { "epoch": 0.42, "grad_norm": 8.247326850891113, "learning_rate": 1.3124862685234092e-05, "loss": 1.492, "step": 2590 }, { "epoch": 0.42, "grad_norm": 6.897265911102295, "learning_rate": 1.3119909117343133e-05, "loss": 1.5157, "step": 2591 }, { "epoch": 0.42, "grad_norm": 10.421944618225098, "learning_rate": 1.311495470119493e-05, "loss": 1.7656, "step": 2592 }, { "epoch": 0.42, "grad_norm": 8.306224822998047, "learning_rate": 1.3109999438136519e-05, "loss": 1.6246, "step": 2593 }, { "epoch": 0.42, "grad_norm": 7.345351219177246, "learning_rate": 1.310504332951516e-05, "loss": 1.628, "step": 2594 }, { "epoch": 0.42, "grad_norm": 8.880175590515137, "learning_rate": 1.3100086376678345e-05, "loss": 1.4273, "step": 2595 }, { "epoch": 0.42, "grad_norm": 8.291412353515625, "learning_rate": 1.3095128580973801e-05, "loss": 1.3142, "step": 2596 }, { "epoch": 0.42, "grad_norm": 8.146236419677734, "learning_rate": 1.3090169943749475e-05, "loss": 1.6667, "step": 2597 }, { "epoch": 0.42, "grad_norm": 7.89778995513916, "learning_rate": 1.3085210466353551e-05, "loss": 1.6509, "step": 2598 }, { "epoch": 0.42, "grad_norm": 6.840627193450928, "learning_rate": 1.3080250150134434e-05, "loss": 1.5663, "step": 2599 }, { "epoch": 0.42, "grad_norm": 8.030433654785156, "learning_rate": 1.3075288996440764e-05, "loss": 1.4981, "step": 2600 }, { "epoch": 0.42, "grad_norm": 9.901013374328613, "learning_rate": 1.3070327006621404e-05, "loss": 1.7364, "step": 2601 }, { "epoch": 0.42, "grad_norm": 8.974263191223145, "learning_rate": 1.3065364182025444e-05, "loss": 1.7895, "step": 2602 }, { "epoch": 0.42, "grad_norm": 11.59145450592041, "learning_rate": 1.3060400524002206e-05, "loss": 1.7768, "step": 2603 }, { "epoch": 0.42, "grad_norm": 11.204959869384766, "learning_rate": 1.3055436033901234e-05, "loss": 1.7343, "step": 2604 }, { "epoch": 0.42, "grad_norm": 9.381643295288086, "learning_rate": 1.3050470713072299e-05, "loss": 1.6588, "step": 2605 }, { "epoch": 0.42, "grad_norm": 11.283259391784668, "learning_rate": 1.30455045628654e-05, "loss": 1.744, "step": 2606 }, { "epoch": 0.42, "grad_norm": 7.196193695068359, "learning_rate": 1.3040537584630757e-05, "loss": 1.6689, "step": 2607 }, { "epoch": 0.42, "grad_norm": 7.054354190826416, "learning_rate": 1.3035569779718817e-05, "loss": 1.69, "step": 2608 }, { "epoch": 0.42, "grad_norm": 8.064269065856934, "learning_rate": 1.3030601149480263e-05, "loss": 1.6745, "step": 2609 }, { "epoch": 0.42, "grad_norm": 8.608505249023438, "learning_rate": 1.302563169526598e-05, "loss": 1.7374, "step": 2610 }, { "epoch": 0.42, "grad_norm": 8.783567428588867, "learning_rate": 1.3020661418427097e-05, "loss": 1.695, "step": 2611 }, { "epoch": 0.42, "grad_norm": 9.862377166748047, "learning_rate": 1.3015690320314952e-05, "loss": 1.5577, "step": 2612 }, { "epoch": 0.42, "grad_norm": 8.250218391418457, "learning_rate": 1.3010718402281122e-05, "loss": 1.4648, "step": 2613 }, { "epoch": 0.42, "grad_norm": 10.593067169189453, "learning_rate": 1.3005745665677394e-05, "loss": 1.6029, "step": 2614 }, { "epoch": 0.42, "grad_norm": 8.754542350769043, "learning_rate": 1.3000772111855779e-05, "loss": 1.6074, "step": 2615 }, { "epoch": 0.42, "grad_norm": 8.13202953338623, "learning_rate": 1.2995797742168517e-05, "loss": 1.5085, "step": 2616 }, { "epoch": 0.42, "grad_norm": 9.637017250061035, "learning_rate": 1.299082255796806e-05, "loss": 1.5494, "step": 2617 }, { "epoch": 0.42, "grad_norm": 7.677379608154297, "learning_rate": 1.2985846560607092e-05, "loss": 1.5372, "step": 2618 }, { "epoch": 0.42, "grad_norm": 7.657971382141113, "learning_rate": 1.2980869751438516e-05, "loss": 1.55, "step": 2619 }, { "epoch": 0.42, "grad_norm": 8.094873428344727, "learning_rate": 1.2975892131815445e-05, "loss": 1.5089, "step": 2620 }, { "epoch": 0.42, "grad_norm": 7.11915397644043, "learning_rate": 1.2970913703091226e-05, "loss": 1.672, "step": 2621 }, { "epoch": 0.42, "grad_norm": 7.343900203704834, "learning_rate": 1.296593446661942e-05, "loss": 1.4509, "step": 2622 }, { "epoch": 0.42, "grad_norm": 12.382451057434082, "learning_rate": 1.2960954423753801e-05, "loss": 1.5991, "step": 2623 }, { "epoch": 0.42, "grad_norm": 8.31522274017334, "learning_rate": 1.2955973575848382e-05, "loss": 1.6333, "step": 2624 }, { "epoch": 0.42, "grad_norm": 7.861814498901367, "learning_rate": 1.2950991924257372e-05, "loss": 1.6711, "step": 2625 }, { "epoch": 0.42, "grad_norm": 7.132401943206787, "learning_rate": 1.294600947033521e-05, "loss": 1.4708, "step": 2626 }, { "epoch": 0.42, "grad_norm": 7.79975700378418, "learning_rate": 1.2941026215436553e-05, "loss": 1.7439, "step": 2627 }, { "epoch": 0.42, "grad_norm": 7.525201797485352, "learning_rate": 1.2936042160916278e-05, "loss": 1.5866, "step": 2628 }, { "epoch": 0.42, "grad_norm": 8.352649688720703, "learning_rate": 1.2931057308129471e-05, "loss": 1.5163, "step": 2629 }, { "epoch": 0.42, "grad_norm": 8.6788911819458, "learning_rate": 1.2926071658431443e-05, "loss": 1.6445, "step": 2630 }, { "epoch": 0.42, "grad_norm": 9.21825885772705, "learning_rate": 1.2921085213177713e-05, "loss": 1.7158, "step": 2631 }, { "epoch": 0.42, "grad_norm": 8.004931449890137, "learning_rate": 1.2916097973724026e-05, "loss": 1.5951, "step": 2632 }, { "epoch": 0.42, "grad_norm": 7.26378870010376, "learning_rate": 1.2911109941426343e-05, "loss": 1.6091, "step": 2633 }, { "epoch": 0.42, "grad_norm": 11.37102222442627, "learning_rate": 1.2906121117640829e-05, "loss": 1.7866, "step": 2634 }, { "epoch": 0.42, "grad_norm": 10.108138084411621, "learning_rate": 1.2901131503723874e-05, "loss": 1.7016, "step": 2635 }, { "epoch": 0.42, "grad_norm": 7.380431652069092, "learning_rate": 1.289614110103208e-05, "loss": 1.6805, "step": 2636 }, { "epoch": 0.42, "grad_norm": 8.964704513549805, "learning_rate": 1.2891149910922267e-05, "loss": 1.6099, "step": 2637 }, { "epoch": 0.42, "grad_norm": 7.712954998016357, "learning_rate": 1.2886157934751462e-05, "loss": 1.6178, "step": 2638 }, { "epoch": 0.42, "grad_norm": 8.497761726379395, "learning_rate": 1.2881165173876913e-05, "loss": 1.7359, "step": 2639 }, { "epoch": 0.42, "grad_norm": 8.165399551391602, "learning_rate": 1.2876171629656076e-05, "loss": 1.675, "step": 2640 }, { "epoch": 0.43, "grad_norm": 9.148069381713867, "learning_rate": 1.2871177303446621e-05, "loss": 1.8256, "step": 2641 }, { "epoch": 0.43, "grad_norm": 7.125479698181152, "learning_rate": 1.2866182196606433e-05, "loss": 1.6217, "step": 2642 }, { "epoch": 0.43, "grad_norm": 7.374441146850586, "learning_rate": 1.2861186310493608e-05, "loss": 1.8115, "step": 2643 }, { "epoch": 0.43, "grad_norm": 8.355578422546387, "learning_rate": 1.2856189646466452e-05, "loss": 1.5186, "step": 2644 }, { "epoch": 0.43, "grad_norm": 8.605730056762695, "learning_rate": 1.2851192205883489e-05, "loss": 1.6113, "step": 2645 }, { "epoch": 0.43, "grad_norm": 8.727757453918457, "learning_rate": 1.2846193990103444e-05, "loss": 1.7839, "step": 2646 }, { "epoch": 0.43, "grad_norm": 7.360819339752197, "learning_rate": 1.2841195000485258e-05, "loss": 1.5108, "step": 2647 }, { "epoch": 0.43, "grad_norm": 7.903881549835205, "learning_rate": 1.2836195238388088e-05, "loss": 1.5589, "step": 2648 }, { "epoch": 0.43, "grad_norm": 7.62420129776001, "learning_rate": 1.283119470517129e-05, "loss": 1.7516, "step": 2649 }, { "epoch": 0.43, "grad_norm": 6.41347599029541, "learning_rate": 1.2826193402194439e-05, "loss": 1.5166, "step": 2650 }, { "epoch": 0.43, "grad_norm": 7.933125019073486, "learning_rate": 1.282119133081731e-05, "loss": 1.4327, "step": 2651 }, { "epoch": 0.43, "grad_norm": 6.695756912231445, "learning_rate": 1.28161884923999e-05, "loss": 1.6652, "step": 2652 }, { "epoch": 0.43, "grad_norm": 6.837563991546631, "learning_rate": 1.2811184888302401e-05, "loss": 1.4798, "step": 2653 }, { "epoch": 0.43, "grad_norm": 9.927124977111816, "learning_rate": 1.2806180519885221e-05, "loss": 1.7345, "step": 2654 }, { "epoch": 0.43, "grad_norm": 9.028541564941406, "learning_rate": 1.2801175388508972e-05, "loss": 1.4809, "step": 2655 }, { "epoch": 0.43, "grad_norm": 7.295807361602783, "learning_rate": 1.2796169495534482e-05, "loss": 1.6434, "step": 2656 }, { "epoch": 0.43, "grad_norm": 11.099503517150879, "learning_rate": 1.2791162842322771e-05, "loss": 1.8687, "step": 2657 }, { "epoch": 0.43, "grad_norm": 8.32163143157959, "learning_rate": 1.2786155430235081e-05, "loss": 1.649, "step": 2658 }, { "epoch": 0.43, "grad_norm": 10.912809371948242, "learning_rate": 1.2781147260632847e-05, "loss": 1.5179, "step": 2659 }, { "epoch": 0.43, "grad_norm": 6.84681510925293, "learning_rate": 1.2776138334877718e-05, "loss": 1.7235, "step": 2660 }, { "epoch": 0.43, "grad_norm": 7.915419578552246, "learning_rate": 1.2771128654331558e-05, "loss": 1.905, "step": 2661 }, { "epoch": 0.43, "grad_norm": 9.461021423339844, "learning_rate": 1.276611822035641e-05, "loss": 1.5769, "step": 2662 }, { "epoch": 0.43, "grad_norm": 10.549513816833496, "learning_rate": 1.2761107034314541e-05, "loss": 1.5712, "step": 2663 }, { "epoch": 0.43, "grad_norm": 10.122342109680176, "learning_rate": 1.275609509756842e-05, "loss": 1.6454, "step": 2664 }, { "epoch": 0.43, "grad_norm": 7.659687042236328, "learning_rate": 1.2751082411480721e-05, "loss": 1.7967, "step": 2665 }, { "epoch": 0.43, "grad_norm": 8.681502342224121, "learning_rate": 1.2746068977414319e-05, "loss": 1.6031, "step": 2666 }, { "epoch": 0.43, "grad_norm": 10.137786865234375, "learning_rate": 1.2741054796732294e-05, "loss": 1.6599, "step": 2667 }, { "epoch": 0.43, "grad_norm": 7.526918411254883, "learning_rate": 1.2736039870797918e-05, "loss": 1.5132, "step": 2668 }, { "epoch": 0.43, "grad_norm": 8.953261375427246, "learning_rate": 1.2731024200974684e-05, "loss": 1.7361, "step": 2669 }, { "epoch": 0.43, "grad_norm": 7.789882659912109, "learning_rate": 1.272600778862628e-05, "loss": 1.7502, "step": 2670 }, { "epoch": 0.43, "grad_norm": 7.795870304107666, "learning_rate": 1.2720990635116588e-05, "loss": 1.6779, "step": 2671 }, { "epoch": 0.43, "grad_norm": 8.696989059448242, "learning_rate": 1.2715972741809705e-05, "loss": 1.5368, "step": 2672 }, { "epoch": 0.43, "grad_norm": 7.989012718200684, "learning_rate": 1.2710954110069917e-05, "loss": 1.4721, "step": 2673 }, { "epoch": 0.43, "grad_norm": 8.274422645568848, "learning_rate": 1.2705934741261713e-05, "loss": 1.6649, "step": 2674 }, { "epoch": 0.43, "grad_norm": 11.641751289367676, "learning_rate": 1.2700914636749795e-05, "loss": 1.6879, "step": 2675 }, { "epoch": 0.43, "grad_norm": 7.763103008270264, "learning_rate": 1.2695893797899049e-05, "loss": 1.8107, "step": 2676 }, { "epoch": 0.43, "grad_norm": 7.459493637084961, "learning_rate": 1.2690872226074568e-05, "loss": 1.681, "step": 2677 }, { "epoch": 0.43, "grad_norm": 11.161229133605957, "learning_rate": 1.2685849922641643e-05, "loss": 1.5535, "step": 2678 }, { "epoch": 0.43, "grad_norm": 7.1336140632629395, "learning_rate": 1.2680826888965762e-05, "loss": 1.4561, "step": 2679 }, { "epoch": 0.43, "grad_norm": 7.283381462097168, "learning_rate": 1.267580312641262e-05, "loss": 1.3937, "step": 2680 }, { "epoch": 0.43, "grad_norm": 7.995090961456299, "learning_rate": 1.2670778636348098e-05, "loss": 1.7427, "step": 2681 }, { "epoch": 0.43, "grad_norm": 7.385272026062012, "learning_rate": 1.2665753420138282e-05, "loss": 1.613, "step": 2682 }, { "epoch": 0.43, "grad_norm": 9.40786361694336, "learning_rate": 1.2660727479149455e-05, "loss": 1.5305, "step": 2683 }, { "epoch": 0.43, "grad_norm": 8.782303810119629, "learning_rate": 1.2655700814748098e-05, "loss": 1.4892, "step": 2684 }, { "epoch": 0.43, "grad_norm": 9.826786041259766, "learning_rate": 1.2650673428300883e-05, "loss": 1.5858, "step": 2685 }, { "epoch": 0.43, "grad_norm": 7.871799468994141, "learning_rate": 1.2645645321174686e-05, "loss": 1.6318, "step": 2686 }, { "epoch": 0.43, "grad_norm": 8.923142433166504, "learning_rate": 1.264061649473657e-05, "loss": 1.8032, "step": 2687 }, { "epoch": 0.43, "grad_norm": 8.869832992553711, "learning_rate": 1.2635586950353806e-05, "loss": 1.5263, "step": 2688 }, { "epoch": 0.43, "grad_norm": 10.37035083770752, "learning_rate": 1.2630556689393846e-05, "loss": 1.7688, "step": 2689 }, { "epoch": 0.43, "grad_norm": 7.602029800415039, "learning_rate": 1.2625525713224347e-05, "loss": 1.5844, "step": 2690 }, { "epoch": 0.43, "grad_norm": 10.48419189453125, "learning_rate": 1.2620494023213157e-05, "loss": 1.8313, "step": 2691 }, { "epoch": 0.43, "grad_norm": 8.463376998901367, "learning_rate": 1.2615461620728317e-05, "loss": 1.4727, "step": 2692 }, { "epoch": 0.43, "grad_norm": 9.533076286315918, "learning_rate": 1.2610428507138065e-05, "loss": 1.6654, "step": 2693 }, { "epoch": 0.43, "grad_norm": 10.526233673095703, "learning_rate": 1.260539468381083e-05, "loss": 1.6411, "step": 2694 }, { "epoch": 0.43, "grad_norm": 10.092452049255371, "learning_rate": 1.2600360152115231e-05, "loss": 1.6876, "step": 2695 }, { "epoch": 0.43, "grad_norm": 9.170245170593262, "learning_rate": 1.2595324913420086e-05, "loss": 1.7544, "step": 2696 }, { "epoch": 0.43, "grad_norm": 9.18564224243164, "learning_rate": 1.2590288969094404e-05, "loss": 1.6054, "step": 2697 }, { "epoch": 0.43, "grad_norm": 12.15224552154541, "learning_rate": 1.2585252320507379e-05, "loss": 1.6947, "step": 2698 }, { "epoch": 0.43, "grad_norm": 6.508998870849609, "learning_rate": 1.2580214969028404e-05, "loss": 1.7964, "step": 2699 }, { "epoch": 0.43, "grad_norm": 8.109219551086426, "learning_rate": 1.257517691602706e-05, "loss": 1.647, "step": 2700 }, { "epoch": 0.43, "grad_norm": 10.149828910827637, "learning_rate": 1.2570138162873118e-05, "loss": 1.4542, "step": 2701 }, { "epoch": 0.43, "grad_norm": 8.673192977905273, "learning_rate": 1.2565098710936546e-05, "loss": 1.7551, "step": 2702 }, { "epoch": 0.44, "grad_norm": 7.413625717163086, "learning_rate": 1.2560058561587488e-05, "loss": 1.7347, "step": 2703 }, { "epoch": 0.44, "grad_norm": 7.705195903778076, "learning_rate": 1.2555017716196293e-05, "loss": 1.2298, "step": 2704 }, { "epoch": 0.44, "grad_norm": 7.629835605621338, "learning_rate": 1.2549976176133494e-05, "loss": 1.8087, "step": 2705 }, { "epoch": 0.44, "grad_norm": 8.6569242477417, "learning_rate": 1.2544933942769804e-05, "loss": 1.6476, "step": 2706 }, { "epoch": 0.44, "grad_norm": 8.419171333312988, "learning_rate": 1.2539891017476133e-05, "loss": 1.6885, "step": 2707 }, { "epoch": 0.44, "grad_norm": 8.508397102355957, "learning_rate": 1.2534847401623585e-05, "loss": 1.3531, "step": 2708 }, { "epoch": 0.44, "grad_norm": 8.437565803527832, "learning_rate": 1.252980309658344e-05, "loss": 1.5547, "step": 2709 }, { "epoch": 0.44, "grad_norm": 7.89519739151001, "learning_rate": 1.2524758103727171e-05, "loss": 1.7203, "step": 2710 }, { "epoch": 0.44, "grad_norm": 7.240360260009766, "learning_rate": 1.2519712424426434e-05, "loss": 1.6111, "step": 2711 }, { "epoch": 0.44, "grad_norm": 8.423141479492188, "learning_rate": 1.2514666060053075e-05, "loss": 1.8355, "step": 2712 }, { "epoch": 0.44, "grad_norm": 9.308478355407715, "learning_rate": 1.2509619011979131e-05, "loss": 1.517, "step": 2713 }, { "epoch": 0.44, "grad_norm": 8.22494888305664, "learning_rate": 1.2504571281576818e-05, "loss": 1.7017, "step": 2714 }, { "epoch": 0.44, "grad_norm": 8.377262115478516, "learning_rate": 1.249952287021854e-05, "loss": 1.7755, "step": 2715 }, { "epoch": 0.44, "grad_norm": 7.817478179931641, "learning_rate": 1.249447377927688e-05, "loss": 1.4817, "step": 2716 }, { "epoch": 0.44, "grad_norm": 7.379669189453125, "learning_rate": 1.2489424010124615e-05, "loss": 1.5991, "step": 2717 }, { "epoch": 0.44, "grad_norm": 7.945923805236816, "learning_rate": 1.2484373564134703e-05, "loss": 1.7189, "step": 2718 }, { "epoch": 0.44, "grad_norm": 7.223389148712158, "learning_rate": 1.247932244268029e-05, "loss": 1.545, "step": 2719 }, { "epoch": 0.44, "grad_norm": 6.979830741882324, "learning_rate": 1.2474270647134696e-05, "loss": 1.6027, "step": 2720 }, { "epoch": 0.44, "grad_norm": 7.107755661010742, "learning_rate": 1.2469218178871427e-05, "loss": 1.7325, "step": 2721 }, { "epoch": 0.44, "grad_norm": 7.800538539886475, "learning_rate": 1.246416503926418e-05, "loss": 1.6352, "step": 2722 }, { "epoch": 0.44, "grad_norm": 7.754364490509033, "learning_rate": 1.2459111229686828e-05, "loss": 1.6736, "step": 2723 }, { "epoch": 0.44, "grad_norm": 8.740925788879395, "learning_rate": 1.2454056751513428e-05, "loss": 1.5868, "step": 2724 }, { "epoch": 0.44, "grad_norm": 7.233051300048828, "learning_rate": 1.2449001606118217e-05, "loss": 1.5136, "step": 2725 }, { "epoch": 0.44, "grad_norm": 8.097946166992188, "learning_rate": 1.2443945794875609e-05, "loss": 1.433, "step": 2726 }, { "epoch": 0.44, "grad_norm": 8.453789710998535, "learning_rate": 1.2438889319160214e-05, "loss": 1.6482, "step": 2727 }, { "epoch": 0.44, "grad_norm": 7.4984259605407715, "learning_rate": 1.2433832180346808e-05, "loss": 1.7505, "step": 2728 }, { "epoch": 0.44, "grad_norm": 7.874278545379639, "learning_rate": 1.242877437981035e-05, "loss": 1.6025, "step": 2729 }, { "epoch": 0.44, "grad_norm": 8.102807998657227, "learning_rate": 1.2423715918925989e-05, "loss": 1.5226, "step": 2730 }, { "epoch": 0.44, "grad_norm": 9.569923400878906, "learning_rate": 1.241865679906904e-05, "loss": 1.4878, "step": 2731 }, { "epoch": 0.44, "grad_norm": 8.801226615905762, "learning_rate": 1.2413597021615003e-05, "loss": 1.5122, "step": 2732 }, { "epoch": 0.44, "grad_norm": 9.159744262695312, "learning_rate": 1.2408536587939557e-05, "loss": 1.6215, "step": 2733 }, { "epoch": 0.44, "grad_norm": 7.4337053298950195, "learning_rate": 1.2403475499418563e-05, "loss": 1.8389, "step": 2734 }, { "epoch": 0.44, "grad_norm": 8.182426452636719, "learning_rate": 1.2398413757428051e-05, "loss": 1.6444, "step": 2735 }, { "epoch": 0.44, "grad_norm": 9.206244468688965, "learning_rate": 1.2393351363344239e-05, "loss": 1.813, "step": 2736 }, { "epoch": 0.44, "grad_norm": 7.158106803894043, "learning_rate": 1.2388288318543513e-05, "loss": 1.6657, "step": 2737 }, { "epoch": 0.44, "grad_norm": 8.038784980773926, "learning_rate": 1.238322462440244e-05, "loss": 1.6506, "step": 2738 }, { "epoch": 0.44, "grad_norm": 7.8130621910095215, "learning_rate": 1.2378160282297769e-05, "loss": 1.4887, "step": 2739 }, { "epoch": 0.44, "grad_norm": 9.391538619995117, "learning_rate": 1.2373095293606417e-05, "loss": 1.7017, "step": 2740 }, { "epoch": 0.44, "grad_norm": 8.516375541687012, "learning_rate": 1.2368029659705477e-05, "loss": 1.7262, "step": 2741 }, { "epoch": 0.44, "grad_norm": 7.690047264099121, "learning_rate": 1.2362963381972222e-05, "loss": 1.4544, "step": 2742 }, { "epoch": 0.44, "grad_norm": 7.13681697845459, "learning_rate": 1.2357896461784097e-05, "loss": 1.6578, "step": 2743 }, { "epoch": 0.44, "grad_norm": 7.768351078033447, "learning_rate": 1.2352828900518727e-05, "loss": 1.3353, "step": 2744 }, { "epoch": 0.44, "grad_norm": 8.4710111618042, "learning_rate": 1.23477606995539e-05, "loss": 1.6492, "step": 2745 }, { "epoch": 0.44, "grad_norm": 7.496953010559082, "learning_rate": 1.2342691860267596e-05, "loss": 1.6821, "step": 2746 }, { "epoch": 0.44, "grad_norm": 9.085906028747559, "learning_rate": 1.2337622384037949e-05, "loss": 1.6672, "step": 2747 }, { "epoch": 0.44, "grad_norm": 8.03682804107666, "learning_rate": 1.2332552272243274e-05, "loss": 1.6814, "step": 2748 }, { "epoch": 0.44, "grad_norm": 7.959461212158203, "learning_rate": 1.2327481526262065e-05, "loss": 1.5482, "step": 2749 }, { "epoch": 0.44, "grad_norm": 6.299744129180908, "learning_rate": 1.2322410147472976e-05, "loss": 1.3663, "step": 2750 }, { "epoch": 0.44, "grad_norm": 8.593952178955078, "learning_rate": 1.231733813725485e-05, "loss": 1.6159, "step": 2751 }, { "epoch": 0.44, "grad_norm": 10.072769165039062, "learning_rate": 1.2312265496986683e-05, "loss": 1.6911, "step": 2752 }, { "epoch": 0.44, "grad_norm": 8.170872688293457, "learning_rate": 1.2307192228047656e-05, "loss": 1.8592, "step": 2753 }, { "epoch": 0.44, "grad_norm": 7.897317886352539, "learning_rate": 1.230211833181711e-05, "loss": 1.6368, "step": 2754 }, { "epoch": 0.44, "grad_norm": 6.989278316497803, "learning_rate": 1.2297043809674566e-05, "loss": 1.6331, "step": 2755 }, { "epoch": 0.44, "grad_norm": 8.939229011535645, "learning_rate": 1.2291968662999718e-05, "loss": 1.7791, "step": 2756 }, { "epoch": 0.44, "grad_norm": 8.005184173583984, "learning_rate": 1.2286892893172415e-05, "loss": 1.6347, "step": 2757 }, { "epoch": 0.44, "grad_norm": 8.308435440063477, "learning_rate": 1.2281816501572686e-05, "loss": 1.7084, "step": 2758 }, { "epoch": 0.44, "grad_norm": 7.213654041290283, "learning_rate": 1.2276739489580724e-05, "loss": 1.5383, "step": 2759 }, { "epoch": 0.44, "grad_norm": 6.873128414154053, "learning_rate": 1.2271661858576902e-05, "loss": 1.6966, "step": 2760 }, { "epoch": 0.44, "grad_norm": 7.828890800476074, "learning_rate": 1.2266583609941746e-05, "loss": 1.4967, "step": 2761 }, { "epoch": 0.44, "grad_norm": 9.659419059753418, "learning_rate": 1.2261504745055963e-05, "loss": 1.6047, "step": 2762 }, { "epoch": 0.44, "grad_norm": 7.832325458526611, "learning_rate": 1.2256425265300413e-05, "loss": 1.7609, "step": 2763 }, { "epoch": 0.44, "grad_norm": 7.264443874359131, "learning_rate": 1.2251345172056134e-05, "loss": 1.6176, "step": 2764 }, { "epoch": 0.45, "grad_norm": 8.602680206298828, "learning_rate": 1.2246264466704334e-05, "loss": 1.5978, "step": 2765 }, { "epoch": 0.45, "grad_norm": 7.120947360992432, "learning_rate": 1.2241183150626377e-05, "loss": 1.8364, "step": 2766 }, { "epoch": 0.45, "grad_norm": 7.9309773445129395, "learning_rate": 1.2236101225203803e-05, "loss": 1.584, "step": 2767 }, { "epoch": 0.45, "grad_norm": 8.630187034606934, "learning_rate": 1.2231018691818301e-05, "loss": 1.6873, "step": 2768 }, { "epoch": 0.45, "grad_norm": 8.534361839294434, "learning_rate": 1.2225935551851747e-05, "loss": 1.507, "step": 2769 }, { "epoch": 0.45, "grad_norm": 7.400129795074463, "learning_rate": 1.2220851806686169e-05, "loss": 1.5115, "step": 2770 }, { "epoch": 0.45, "grad_norm": 8.761417388916016, "learning_rate": 1.2215767457703764e-05, "loss": 1.6069, "step": 2771 }, { "epoch": 0.45, "grad_norm": 7.953553199768066, "learning_rate": 1.221068250628689e-05, "loss": 1.8045, "step": 2772 }, { "epoch": 0.45, "grad_norm": 7.965190887451172, "learning_rate": 1.2205596953818069e-05, "loss": 1.568, "step": 2773 }, { "epoch": 0.45, "grad_norm": 11.350908279418945, "learning_rate": 1.2200510801679988e-05, "loss": 1.6967, "step": 2774 }, { "epoch": 0.45, "grad_norm": 9.240769386291504, "learning_rate": 1.2195424051255498e-05, "loss": 1.5946, "step": 2775 }, { "epoch": 0.45, "grad_norm": 7.504601001739502, "learning_rate": 1.219033670392761e-05, "loss": 1.6188, "step": 2776 }, { "epoch": 0.45, "grad_norm": 7.697402000427246, "learning_rate": 1.2185248761079497e-05, "loss": 1.6837, "step": 2777 }, { "epoch": 0.45, "grad_norm": 6.343660831451416, "learning_rate": 1.2180160224094502e-05, "loss": 1.5015, "step": 2778 }, { "epoch": 0.45, "grad_norm": 10.224660873413086, "learning_rate": 1.2175071094356116e-05, "loss": 1.6098, "step": 2779 }, { "epoch": 0.45, "grad_norm": 9.135364532470703, "learning_rate": 1.2169981373248002e-05, "loss": 1.7214, "step": 2780 }, { "epoch": 0.45, "grad_norm": 7.655171871185303, "learning_rate": 1.2164891062153983e-05, "loss": 1.5776, "step": 2781 }, { "epoch": 0.45, "grad_norm": 7.815268516540527, "learning_rate": 1.2159800162458031e-05, "loss": 1.5445, "step": 2782 }, { "epoch": 0.45, "grad_norm": 8.032096862792969, "learning_rate": 1.2154708675544294e-05, "loss": 1.7651, "step": 2783 }, { "epoch": 0.45, "grad_norm": 8.256999015808105, "learning_rate": 1.2149616602797067e-05, "loss": 1.6946, "step": 2784 }, { "epoch": 0.45, "grad_norm": 8.340679168701172, "learning_rate": 1.2144523945600814e-05, "loss": 1.6395, "step": 2785 }, { "epoch": 0.45, "grad_norm": 7.510686874389648, "learning_rate": 1.2139430705340152e-05, "loss": 1.5564, "step": 2786 }, { "epoch": 0.45, "grad_norm": 8.405211448669434, "learning_rate": 1.2134336883399855e-05, "loss": 1.7441, "step": 2787 }, { "epoch": 0.45, "grad_norm": 11.651546478271484, "learning_rate": 1.212924248116486e-05, "loss": 1.352, "step": 2788 }, { "epoch": 0.45, "grad_norm": 6.7624053955078125, "learning_rate": 1.212414750002026e-05, "loss": 1.3282, "step": 2789 }, { "epoch": 0.45, "grad_norm": 8.107758522033691, "learning_rate": 1.2119051941351306e-05, "loss": 1.6175, "step": 2790 }, { "epoch": 0.45, "grad_norm": 7.809074878692627, "learning_rate": 1.2113955806543403e-05, "loss": 1.4806, "step": 2791 }, { "epoch": 0.45, "grad_norm": 8.233942985534668, "learning_rate": 1.2108859096982118e-05, "loss": 1.6062, "step": 2792 }, { "epoch": 0.45, "grad_norm": 8.864030838012695, "learning_rate": 1.2103761814053168e-05, "loss": 1.6595, "step": 2793 }, { "epoch": 0.45, "grad_norm": 11.585865020751953, "learning_rate": 1.2098663959142429e-05, "loss": 1.7422, "step": 2794 }, { "epoch": 0.45, "grad_norm": 9.458026885986328, "learning_rate": 1.2093565533635934e-05, "loss": 1.6295, "step": 2795 }, { "epoch": 0.45, "grad_norm": 7.125947952270508, "learning_rate": 1.208846653891987e-05, "loss": 1.4033, "step": 2796 }, { "epoch": 0.45, "grad_norm": 7.726698398590088, "learning_rate": 1.2083366976380574e-05, "loss": 1.7083, "step": 2797 }, { "epoch": 0.45, "grad_norm": 11.575249671936035, "learning_rate": 1.207826684740455e-05, "loss": 1.5281, "step": 2798 }, { "epoch": 0.45, "grad_norm": 9.453253746032715, "learning_rate": 1.2073166153378444e-05, "loss": 1.6675, "step": 2799 }, { "epoch": 0.45, "grad_norm": 10.610207557678223, "learning_rate": 1.2068064895689058e-05, "loss": 1.5028, "step": 2800 }, { "epoch": 0.45, "grad_norm": 7.363902568817139, "learning_rate": 1.2062963075723348e-05, "loss": 1.9467, "step": 2801 }, { "epoch": 0.45, "grad_norm": 9.854195594787598, "learning_rate": 1.2057860694868422e-05, "loss": 1.4489, "step": 2802 }, { "epoch": 0.45, "grad_norm": 10.19770336151123, "learning_rate": 1.2052757754511548e-05, "loss": 1.6693, "step": 2803 }, { "epoch": 0.45, "grad_norm": 7.693286895751953, "learning_rate": 1.2047654256040139e-05, "loss": 1.484, "step": 2804 }, { "epoch": 0.45, "grad_norm": 8.429178237915039, "learning_rate": 1.2042550200841758e-05, "loss": 1.6863, "step": 2805 }, { "epoch": 0.45, "grad_norm": 7.541811466217041, "learning_rate": 1.2037445590304124e-05, "loss": 1.4266, "step": 2806 }, { "epoch": 0.45, "grad_norm": 8.948667526245117, "learning_rate": 1.2032340425815102e-05, "loss": 1.7123, "step": 2807 }, { "epoch": 0.45, "grad_norm": 6.678216934204102, "learning_rate": 1.2027234708762713e-05, "loss": 1.5965, "step": 2808 }, { "epoch": 0.45, "grad_norm": 7.485115051269531, "learning_rate": 1.2022128440535132e-05, "loss": 1.5294, "step": 2809 }, { "epoch": 0.45, "grad_norm": 6.985195636749268, "learning_rate": 1.2017021622520669e-05, "loss": 1.5657, "step": 2810 }, { "epoch": 0.45, "grad_norm": 7.691072940826416, "learning_rate": 1.2011914256107794e-05, "loss": 1.5961, "step": 2811 }, { "epoch": 0.45, "grad_norm": 7.196331977844238, "learning_rate": 1.2006806342685127e-05, "loss": 1.5248, "step": 2812 }, { "epoch": 0.45, "grad_norm": 8.280518531799316, "learning_rate": 1.2001697883641436e-05, "loss": 1.5099, "step": 2813 }, { "epoch": 0.45, "grad_norm": 8.27878475189209, "learning_rate": 1.1996588880365634e-05, "loss": 1.5209, "step": 2814 }, { "epoch": 0.45, "grad_norm": 7.438783645629883, "learning_rate": 1.1991479334246783e-05, "loss": 1.4147, "step": 2815 }, { "epoch": 0.45, "grad_norm": 9.163198471069336, "learning_rate": 1.198636924667409e-05, "loss": 1.4841, "step": 2816 }, { "epoch": 0.45, "grad_norm": 6.624181270599365, "learning_rate": 1.1981258619036918e-05, "loss": 1.5193, "step": 2817 }, { "epoch": 0.45, "grad_norm": 9.298868179321289, "learning_rate": 1.1976147452724769e-05, "loss": 1.5951, "step": 2818 }, { "epoch": 0.45, "grad_norm": 9.682733535766602, "learning_rate": 1.197103574912729e-05, "loss": 1.4508, "step": 2819 }, { "epoch": 0.45, "grad_norm": 8.240623474121094, "learning_rate": 1.1965923509634286e-05, "loss": 1.3237, "step": 2820 }, { "epoch": 0.45, "grad_norm": 8.69334602355957, "learning_rate": 1.1960810735635694e-05, "loss": 1.5783, "step": 2821 }, { "epoch": 0.45, "grad_norm": 8.07140827178955, "learning_rate": 1.1955697428521601e-05, "loss": 1.4376, "step": 2822 }, { "epoch": 0.45, "grad_norm": 11.482633590698242, "learning_rate": 1.1950583589682241e-05, "loss": 1.6156, "step": 2823 }, { "epoch": 0.45, "grad_norm": 8.131508827209473, "learning_rate": 1.1945469220507992e-05, "loss": 1.5902, "step": 2824 }, { "epoch": 0.45, "grad_norm": 8.286796569824219, "learning_rate": 1.1940354322389375e-05, "loss": 1.6425, "step": 2825 }, { "epoch": 0.45, "grad_norm": 7.727294921875, "learning_rate": 1.1935238896717058e-05, "loss": 1.4414, "step": 2826 }, { "epoch": 0.46, "grad_norm": 8.323892593383789, "learning_rate": 1.1930122944881846e-05, "loss": 1.6445, "step": 2827 }, { "epoch": 0.46, "grad_norm": 9.159804344177246, "learning_rate": 1.1925006468274689e-05, "loss": 1.656, "step": 2828 }, { "epoch": 0.46, "grad_norm": 10.971695899963379, "learning_rate": 1.1919889468286687e-05, "loss": 1.6465, "step": 2829 }, { "epoch": 0.46, "grad_norm": 7.6622772216796875, "learning_rate": 1.1914771946309073e-05, "loss": 1.5301, "step": 2830 }, { "epoch": 0.46, "grad_norm": 9.17999267578125, "learning_rate": 1.1909653903733226e-05, "loss": 1.748, "step": 2831 }, { "epoch": 0.46, "grad_norm": 7.7408599853515625, "learning_rate": 1.1904535341950672e-05, "loss": 1.486, "step": 2832 }, { "epoch": 0.46, "grad_norm": 7.91771936416626, "learning_rate": 1.1899416262353064e-05, "loss": 1.4437, "step": 2833 }, { "epoch": 0.46, "grad_norm": 8.833075523376465, "learning_rate": 1.1894296666332206e-05, "loss": 1.7999, "step": 2834 }, { "epoch": 0.46, "grad_norm": 7.651315689086914, "learning_rate": 1.1889176555280042e-05, "loss": 1.7112, "step": 2835 }, { "epoch": 0.46, "grad_norm": 7.544612407684326, "learning_rate": 1.1884055930588659e-05, "loss": 1.4818, "step": 2836 }, { "epoch": 0.46, "grad_norm": 8.171263694763184, "learning_rate": 1.1878934793650273e-05, "loss": 1.6111, "step": 2837 }, { "epoch": 0.46, "grad_norm": 7.800546646118164, "learning_rate": 1.187381314585725e-05, "loss": 1.7294, "step": 2838 }, { "epoch": 0.46, "grad_norm": 7.420600891113281, "learning_rate": 1.1868690988602085e-05, "loss": 1.4582, "step": 2839 }, { "epoch": 0.46, "grad_norm": 7.328935623168945, "learning_rate": 1.1863568323277419e-05, "loss": 1.5917, "step": 2840 }, { "epoch": 0.46, "grad_norm": 8.006163597106934, "learning_rate": 1.1858445151276035e-05, "loss": 1.5885, "step": 2841 }, { "epoch": 0.46, "grad_norm": 8.114283561706543, "learning_rate": 1.185332147399084e-05, "loss": 1.6752, "step": 2842 }, { "epoch": 0.46, "grad_norm": 9.93197250366211, "learning_rate": 1.184819729281489e-05, "loss": 1.5655, "step": 2843 }, { "epoch": 0.46, "grad_norm": 7.1501007080078125, "learning_rate": 1.1843072609141373e-05, "loss": 1.5508, "step": 2844 }, { "epoch": 0.46, "grad_norm": 9.09634780883789, "learning_rate": 1.1837947424363614e-05, "loss": 1.5792, "step": 2845 }, { "epoch": 0.46, "grad_norm": 8.428784370422363, "learning_rate": 1.1832821739875082e-05, "loss": 1.525, "step": 2846 }, { "epoch": 0.46, "grad_norm": 8.646236419677734, "learning_rate": 1.1827695557069365e-05, "loss": 1.5717, "step": 2847 }, { "epoch": 0.46, "grad_norm": 7.965404033660889, "learning_rate": 1.1822568877340201e-05, "loss": 1.4341, "step": 2848 }, { "epoch": 0.46, "grad_norm": 7.6117377281188965, "learning_rate": 1.1817441702081458e-05, "loss": 1.6664, "step": 2849 }, { "epoch": 0.46, "grad_norm": 7.677888870239258, "learning_rate": 1.1812314032687142e-05, "loss": 1.6328, "step": 2850 }, { "epoch": 0.46, "grad_norm": 8.54433536529541, "learning_rate": 1.180718587055139e-05, "loss": 1.7226, "step": 2851 }, { "epoch": 0.46, "grad_norm": 6.533612251281738, "learning_rate": 1.180205721706847e-05, "loss": 1.4765, "step": 2852 }, { "epoch": 0.46, "grad_norm": 7.769745826721191, "learning_rate": 1.1796928073632788e-05, "loss": 1.5345, "step": 2853 }, { "epoch": 0.46, "grad_norm": 7.320694446563721, "learning_rate": 1.1791798441638881e-05, "loss": 1.8014, "step": 2854 }, { "epoch": 0.46, "grad_norm": 6.882086277008057, "learning_rate": 1.1786668322481428e-05, "loss": 1.315, "step": 2855 }, { "epoch": 0.46, "grad_norm": 7.166660308837891, "learning_rate": 1.1781537717555227e-05, "loss": 1.3146, "step": 2856 }, { "epoch": 0.46, "grad_norm": 8.25988483428955, "learning_rate": 1.1776406628255214e-05, "loss": 1.4241, "step": 2857 }, { "epoch": 0.46, "grad_norm": 8.296093940734863, "learning_rate": 1.1771275055976457e-05, "loss": 1.6694, "step": 2858 }, { "epoch": 0.46, "grad_norm": 7.513505935668945, "learning_rate": 1.176614300211415e-05, "loss": 1.6114, "step": 2859 }, { "epoch": 0.46, "grad_norm": 6.69553804397583, "learning_rate": 1.1761010468063629e-05, "loss": 1.6354, "step": 2860 }, { "epoch": 0.46, "grad_norm": 7.116929054260254, "learning_rate": 1.1755877455220355e-05, "loss": 1.5972, "step": 2861 }, { "epoch": 0.46, "grad_norm": 6.005568981170654, "learning_rate": 1.1750743964979919e-05, "loss": 1.3824, "step": 2862 }, { "epoch": 0.46, "grad_norm": 7.152434349060059, "learning_rate": 1.174560999873803e-05, "loss": 1.49, "step": 2863 }, { "epoch": 0.46, "grad_norm": 7.989598751068115, "learning_rate": 1.174047555789055e-05, "loss": 1.1925, "step": 2864 }, { "epoch": 0.46, "grad_norm": 8.603377342224121, "learning_rate": 1.1735340643833452e-05, "loss": 1.5517, "step": 2865 }, { "epoch": 0.46, "grad_norm": 8.84826374053955, "learning_rate": 1.1730205257962848e-05, "loss": 1.4741, "step": 2866 }, { "epoch": 0.46, "grad_norm": 7.572869777679443, "learning_rate": 1.172506940167497e-05, "loss": 1.6906, "step": 2867 }, { "epoch": 0.46, "grad_norm": 8.509078979492188, "learning_rate": 1.1719933076366176e-05, "loss": 1.658, "step": 2868 }, { "epoch": 0.46, "grad_norm": 8.264411926269531, "learning_rate": 1.171479628343297e-05, "loss": 1.7017, "step": 2869 }, { "epoch": 0.46, "grad_norm": 7.512705326080322, "learning_rate": 1.170965902427196e-05, "loss": 1.6665, "step": 2870 }, { "epoch": 0.46, "grad_norm": 7.52984094619751, "learning_rate": 1.1704521300279894e-05, "loss": 1.4851, "step": 2871 }, { "epoch": 0.46, "grad_norm": 7.555070400238037, "learning_rate": 1.1699383112853644e-05, "loss": 1.603, "step": 2872 }, { "epoch": 0.46, "grad_norm": 6.663818836212158, "learning_rate": 1.1694244463390207e-05, "loss": 1.5675, "step": 2873 }, { "epoch": 0.46, "grad_norm": 7.778237819671631, "learning_rate": 1.1689105353286705e-05, "loss": 1.7182, "step": 2874 }, { "epoch": 0.46, "grad_norm": 6.484940528869629, "learning_rate": 1.1683965783940388e-05, "loss": 1.5689, "step": 2875 }, { "epoch": 0.46, "grad_norm": 8.01427936553955, "learning_rate": 1.1678825756748626e-05, "loss": 1.6066, "step": 2876 }, { "epoch": 0.46, "grad_norm": 8.296075820922852, "learning_rate": 1.1673685273108922e-05, "loss": 1.4375, "step": 2877 }, { "epoch": 0.46, "grad_norm": 11.331954956054688, "learning_rate": 1.166854433441889e-05, "loss": 1.6575, "step": 2878 }, { "epoch": 0.46, "grad_norm": 7.204132556915283, "learning_rate": 1.1663402942076283e-05, "loss": 1.4216, "step": 2879 }, { "epoch": 0.46, "grad_norm": 7.80161190032959, "learning_rate": 1.1658261097478964e-05, "loss": 1.484, "step": 2880 }, { "epoch": 0.46, "grad_norm": 8.285881042480469, "learning_rate": 1.1653118802024928e-05, "loss": 1.5433, "step": 2881 }, { "epoch": 0.46, "grad_norm": 8.542397499084473, "learning_rate": 1.1647976057112286e-05, "loss": 1.7089, "step": 2882 }, { "epoch": 0.46, "grad_norm": 7.370809555053711, "learning_rate": 1.164283286413928e-05, "loss": 1.4463, "step": 2883 }, { "epoch": 0.46, "grad_norm": 8.689858436584473, "learning_rate": 1.163768922450426e-05, "loss": 1.5412, "step": 2884 }, { "epoch": 0.46, "grad_norm": 8.785224914550781, "learning_rate": 1.1632545139605716e-05, "loss": 1.6022, "step": 2885 }, { "epoch": 0.46, "grad_norm": 8.42178726196289, "learning_rate": 1.1627400610842238e-05, "loss": 1.3427, "step": 2886 }, { "epoch": 0.46, "grad_norm": 6.964816570281982, "learning_rate": 1.1622255639612553e-05, "loss": 1.4743, "step": 2887 }, { "epoch": 0.46, "grad_norm": 11.040409088134766, "learning_rate": 1.161711022731551e-05, "loss": 1.4407, "step": 2888 }, { "epoch": 0.47, "grad_norm": 9.110222816467285, "learning_rate": 1.1611964375350055e-05, "loss": 1.7154, "step": 2889 }, { "epoch": 0.47, "grad_norm": 8.680368423461914, "learning_rate": 1.1606818085115281e-05, "loss": 1.8368, "step": 2890 }, { "epoch": 0.47, "grad_norm": 7.819875240325928, "learning_rate": 1.1601671358010383e-05, "loss": 1.7154, "step": 2891 }, { "epoch": 0.47, "grad_norm": 8.178199768066406, "learning_rate": 1.1596524195434679e-05, "loss": 1.5611, "step": 2892 }, { "epoch": 0.47, "grad_norm": 12.03365421295166, "learning_rate": 1.1591376598787614e-05, "loss": 1.5127, "step": 2893 }, { "epoch": 0.47, "grad_norm": 9.8196439743042, "learning_rate": 1.158622856946874e-05, "loss": 1.6906, "step": 2894 }, { "epoch": 0.47, "grad_norm": 8.735026359558105, "learning_rate": 1.1581080108877727e-05, "loss": 1.5394, "step": 2895 }, { "epoch": 0.47, "grad_norm": 7.711101531982422, "learning_rate": 1.1575931218414368e-05, "loss": 1.5974, "step": 2896 }, { "epoch": 0.47, "grad_norm": 6.548882007598877, "learning_rate": 1.1570781899478565e-05, "loss": 1.4458, "step": 2897 }, { "epoch": 0.47, "grad_norm": 10.778955459594727, "learning_rate": 1.1565632153470351e-05, "loss": 1.3244, "step": 2898 }, { "epoch": 0.47, "grad_norm": 10.957237243652344, "learning_rate": 1.1560481981789864e-05, "loss": 1.679, "step": 2899 }, { "epoch": 0.47, "grad_norm": 10.269322395324707, "learning_rate": 1.1555331385837354e-05, "loss": 1.5677, "step": 2900 }, { "epoch": 0.47, "grad_norm": 8.677872657775879, "learning_rate": 1.1550180367013193e-05, "loss": 1.6219, "step": 2901 }, { "epoch": 0.47, "grad_norm": 8.243786811828613, "learning_rate": 1.1545028926717874e-05, "loss": 1.779, "step": 2902 }, { "epoch": 0.47, "grad_norm": 12.026450157165527, "learning_rate": 1.153987706635199e-05, "loss": 1.5681, "step": 2903 }, { "epoch": 0.47, "grad_norm": 9.6666841506958, "learning_rate": 1.1534724787316263e-05, "loss": 1.3461, "step": 2904 }, { "epoch": 0.47, "grad_norm": 8.136616706848145, "learning_rate": 1.1529572091011514e-05, "loss": 1.4317, "step": 2905 }, { "epoch": 0.47, "grad_norm": 8.201972961425781, "learning_rate": 1.1524418978838686e-05, "loss": 1.5543, "step": 2906 }, { "epoch": 0.47, "grad_norm": 8.042191505432129, "learning_rate": 1.1519265452198838e-05, "loss": 1.7227, "step": 2907 }, { "epoch": 0.47, "grad_norm": 7.739882469177246, "learning_rate": 1.1514111512493136e-05, "loss": 1.3969, "step": 2908 }, { "epoch": 0.47, "grad_norm": 8.80219841003418, "learning_rate": 1.1508957161122863e-05, "loss": 1.7693, "step": 2909 }, { "epoch": 0.47, "grad_norm": 8.025158882141113, "learning_rate": 1.150380239948941e-05, "loss": 1.4418, "step": 2910 }, { "epoch": 0.47, "grad_norm": 10.769341468811035, "learning_rate": 1.1498647228994274e-05, "loss": 1.7149, "step": 2911 }, { "epoch": 0.47, "grad_norm": 11.494485855102539, "learning_rate": 1.1493491651039077e-05, "loss": 2.0554, "step": 2912 }, { "epoch": 0.47, "grad_norm": 9.811553955078125, "learning_rate": 1.1488335667025543e-05, "loss": 1.8304, "step": 2913 }, { "epoch": 0.47, "grad_norm": 8.22568416595459, "learning_rate": 1.1483179278355507e-05, "loss": 1.3964, "step": 2914 }, { "epoch": 0.47, "grad_norm": 6.008050441741943, "learning_rate": 1.1478022486430916e-05, "loss": 1.2628, "step": 2915 }, { "epoch": 0.47, "grad_norm": 9.291084289550781, "learning_rate": 1.1472865292653823e-05, "loss": 1.5417, "step": 2916 }, { "epoch": 0.47, "grad_norm": 13.278042793273926, "learning_rate": 1.1467707698426398e-05, "loss": 1.507, "step": 2917 }, { "epoch": 0.47, "grad_norm": 10.217933654785156, "learning_rate": 1.146254970515091e-05, "loss": 1.4337, "step": 2918 }, { "epoch": 0.47, "grad_norm": 9.459331512451172, "learning_rate": 1.1457391314229742e-05, "loss": 1.6354, "step": 2919 }, { "epoch": 0.47, "grad_norm": 7.6650190353393555, "learning_rate": 1.1452232527065389e-05, "loss": 1.4068, "step": 2920 }, { "epoch": 0.47, "grad_norm": 7.079159736633301, "learning_rate": 1.1447073345060442e-05, "loss": 1.5502, "step": 2921 }, { "epoch": 0.47, "grad_norm": 10.034412384033203, "learning_rate": 1.1441913769617609e-05, "loss": 1.621, "step": 2922 }, { "epoch": 0.47, "grad_norm": 9.83951187133789, "learning_rate": 1.1436753802139706e-05, "loss": 1.5453, "step": 2923 }, { "epoch": 0.47, "grad_norm": 7.954181671142578, "learning_rate": 1.1431593444029646e-05, "loss": 1.2841, "step": 2924 }, { "epoch": 0.47, "grad_norm": 8.796581268310547, "learning_rate": 1.1426432696690461e-05, "loss": 1.5697, "step": 2925 }, { "epoch": 0.47, "grad_norm": 7.864656448364258, "learning_rate": 1.1421271561525277e-05, "loss": 1.6208, "step": 2926 }, { "epoch": 0.47, "grad_norm": 7.411020278930664, "learning_rate": 1.1416110039937332e-05, "loss": 1.3641, "step": 2927 }, { "epoch": 0.47, "grad_norm": 8.513081550598145, "learning_rate": 1.1410948133329966e-05, "loss": 1.4265, "step": 2928 }, { "epoch": 0.47, "grad_norm": 7.709059715270996, "learning_rate": 1.1405785843106628e-05, "loss": 1.5225, "step": 2929 }, { "epoch": 0.47, "grad_norm": 6.758538722991943, "learning_rate": 1.1400623170670866e-05, "loss": 1.4606, "step": 2930 }, { "epoch": 0.47, "grad_norm": 6.831662654876709, "learning_rate": 1.1395460117426338e-05, "loss": 1.7332, "step": 2931 }, { "epoch": 0.47, "grad_norm": 8.258992195129395, "learning_rate": 1.13902966847768e-05, "loss": 1.5579, "step": 2932 }, { "epoch": 0.47, "grad_norm": 8.319601058959961, "learning_rate": 1.138513287412611e-05, "loss": 1.3865, "step": 2933 }, { "epoch": 0.47, "grad_norm": 7.415413856506348, "learning_rate": 1.1379968686878238e-05, "loss": 1.4947, "step": 2934 }, { "epoch": 0.47, "grad_norm": 8.049775123596191, "learning_rate": 1.1374804124437242e-05, "loss": 1.4782, "step": 2935 }, { "epoch": 0.47, "grad_norm": 7.934499263763428, "learning_rate": 1.13696391882073e-05, "loss": 1.4506, "step": 2936 }, { "epoch": 0.47, "grad_norm": 6.612964153289795, "learning_rate": 1.1364473879592674e-05, "loss": 1.4246, "step": 2937 }, { "epoch": 0.47, "grad_norm": 8.419999122619629, "learning_rate": 1.135930819999774e-05, "loss": 1.5363, "step": 2938 }, { "epoch": 0.47, "grad_norm": 8.475457191467285, "learning_rate": 1.1354142150826968e-05, "loss": 1.6742, "step": 2939 }, { "epoch": 0.47, "grad_norm": 8.215331077575684, "learning_rate": 1.134897573348493e-05, "loss": 1.4235, "step": 2940 }, { "epoch": 0.47, "grad_norm": 8.006080627441406, "learning_rate": 1.1343808949376303e-05, "loss": 1.4637, "step": 2941 }, { "epoch": 0.47, "grad_norm": 8.846148490905762, "learning_rate": 1.1338641799905854e-05, "loss": 1.6678, "step": 2942 }, { "epoch": 0.47, "grad_norm": 7.846101760864258, "learning_rate": 1.1333474286478457e-05, "loss": 1.4306, "step": 2943 }, { "epoch": 0.47, "grad_norm": 10.321693420410156, "learning_rate": 1.1328306410499078e-05, "loss": 1.5498, "step": 2944 }, { "epoch": 0.47, "grad_norm": 7.754867076873779, "learning_rate": 1.1323138173372795e-05, "loss": 1.5698, "step": 2945 }, { "epoch": 0.47, "grad_norm": 8.015265464782715, "learning_rate": 1.1317969576504769e-05, "loss": 1.5809, "step": 2946 }, { "epoch": 0.47, "grad_norm": 7.636668682098389, "learning_rate": 1.1312800621300269e-05, "loss": 1.7536, "step": 2947 }, { "epoch": 0.47, "grad_norm": 8.198262214660645, "learning_rate": 1.1307631309164656e-05, "loss": 1.7472, "step": 2948 }, { "epoch": 0.47, "grad_norm": 8.925783157348633, "learning_rate": 1.1302461641503383e-05, "loss": 1.5434, "step": 2949 }, { "epoch": 0.47, "grad_norm": 7.924588680267334, "learning_rate": 1.1297291619722018e-05, "loss": 1.3864, "step": 2950 }, { "epoch": 0.48, "grad_norm": 11.365374565124512, "learning_rate": 1.1292121245226208e-05, "loss": 1.4998, "step": 2951 }, { "epoch": 0.48, "grad_norm": 9.10915470123291, "learning_rate": 1.1286950519421702e-05, "loss": 1.5288, "step": 2952 }, { "epoch": 0.48, "grad_norm": 8.235992431640625, "learning_rate": 1.1281779443714342e-05, "loss": 1.5187, "step": 2953 }, { "epoch": 0.48, "grad_norm": 7.574215888977051, "learning_rate": 1.127660801951007e-05, "loss": 1.4308, "step": 2954 }, { "epoch": 0.48, "grad_norm": 7.6596903800964355, "learning_rate": 1.127143624821492e-05, "loss": 1.7118, "step": 2955 }, { "epoch": 0.48, "grad_norm": 9.63509750366211, "learning_rate": 1.1266264131235018e-05, "loss": 1.8452, "step": 2956 }, { "epoch": 0.48, "grad_norm": 7.057692050933838, "learning_rate": 1.126109166997659e-05, "loss": 1.5701, "step": 2957 }, { "epoch": 0.48, "grad_norm": 7.9865641593933105, "learning_rate": 1.1255918865845945e-05, "loss": 1.3615, "step": 2958 }, { "epoch": 0.48, "grad_norm": 9.973479270935059, "learning_rate": 1.1250745720249498e-05, "loss": 1.5949, "step": 2959 }, { "epoch": 0.48, "grad_norm": 7.340075492858887, "learning_rate": 1.1245572234593752e-05, "loss": 1.3346, "step": 2960 }, { "epoch": 0.48, "grad_norm": 7.0013203620910645, "learning_rate": 1.1240398410285297e-05, "loss": 1.4896, "step": 2961 }, { "epoch": 0.48, "grad_norm": 9.740717887878418, "learning_rate": 1.1235224248730821e-05, "loss": 1.5686, "step": 2962 }, { "epoch": 0.48, "grad_norm": 7.501969337463379, "learning_rate": 1.1230049751337103e-05, "loss": 1.5064, "step": 2963 }, { "epoch": 0.48, "grad_norm": 6.818426132202148, "learning_rate": 1.122487491951101e-05, "loss": 1.5219, "step": 2964 }, { "epoch": 0.48, "grad_norm": 7.932427406311035, "learning_rate": 1.1219699754659506e-05, "loss": 1.4595, "step": 2965 }, { "epoch": 0.48, "grad_norm": 7.704092502593994, "learning_rate": 1.1214524258189638e-05, "loss": 1.4772, "step": 2966 }, { "epoch": 0.48, "grad_norm": 9.679253578186035, "learning_rate": 1.1209348431508554e-05, "loss": 1.5141, "step": 2967 }, { "epoch": 0.48, "grad_norm": 10.115635871887207, "learning_rate": 1.1204172276023477e-05, "loss": 1.6406, "step": 2968 }, { "epoch": 0.48, "grad_norm": 8.205681800842285, "learning_rate": 1.119899579314173e-05, "loss": 1.4925, "step": 2969 }, { "epoch": 0.48, "grad_norm": 8.89778995513916, "learning_rate": 1.1193818984270723e-05, "loss": 1.4173, "step": 2970 }, { "epoch": 0.48, "grad_norm": 8.004404067993164, "learning_rate": 1.1188641850817957e-05, "loss": 1.4519, "step": 2971 }, { "epoch": 0.48, "grad_norm": 7.382018089294434, "learning_rate": 1.1183464394191011e-05, "loss": 1.3981, "step": 2972 }, { "epoch": 0.48, "grad_norm": 7.448096752166748, "learning_rate": 1.1178286615797565e-05, "loss": 1.6059, "step": 2973 }, { "epoch": 0.48, "grad_norm": 7.207790374755859, "learning_rate": 1.1173108517045377e-05, "loss": 1.4493, "step": 2974 }, { "epoch": 0.48, "grad_norm": 7.838238716125488, "learning_rate": 1.1167930099342298e-05, "loss": 1.5037, "step": 2975 }, { "epoch": 0.48, "grad_norm": 8.049110412597656, "learning_rate": 1.1162751364096262e-05, "loss": 1.7173, "step": 2976 }, { "epoch": 0.48, "grad_norm": 7.315575122833252, "learning_rate": 1.1157572312715292e-05, "loss": 1.3415, "step": 2977 }, { "epoch": 0.48, "grad_norm": 8.534832000732422, "learning_rate": 1.1152392946607493e-05, "loss": 1.3971, "step": 2978 }, { "epoch": 0.48, "grad_norm": 8.040056228637695, "learning_rate": 1.1147213267181062e-05, "loss": 1.5175, "step": 2979 }, { "epoch": 0.48, "grad_norm": 7.150942325592041, "learning_rate": 1.1142033275844275e-05, "loss": 1.4423, "step": 2980 }, { "epoch": 0.48, "grad_norm": 7.6376447677612305, "learning_rate": 1.1136852974005494e-05, "loss": 1.5839, "step": 2981 }, { "epoch": 0.48, "grad_norm": 8.680776596069336, "learning_rate": 1.1131672363073167e-05, "loss": 1.5602, "step": 2982 }, { "epoch": 0.48, "grad_norm": 8.700601577758789, "learning_rate": 1.1126491444455829e-05, "loss": 1.4482, "step": 2983 }, { "epoch": 0.48, "grad_norm": 8.450233459472656, "learning_rate": 1.1121310219562096e-05, "loss": 1.7518, "step": 2984 }, { "epoch": 0.48, "grad_norm": 8.834112167358398, "learning_rate": 1.111612868980066e-05, "loss": 1.6255, "step": 2985 }, { "epoch": 0.48, "grad_norm": 7.545492172241211, "learning_rate": 1.1110946856580308e-05, "loss": 1.4677, "step": 2986 }, { "epoch": 0.48, "grad_norm": 7.976753234863281, "learning_rate": 1.11057647213099e-05, "loss": 1.7689, "step": 2987 }, { "epoch": 0.48, "grad_norm": 6.699664115905762, "learning_rate": 1.1100582285398384e-05, "loss": 1.5763, "step": 2988 }, { "epoch": 0.48, "grad_norm": 8.306933403015137, "learning_rate": 1.1095399550254791e-05, "loss": 1.4904, "step": 2989 }, { "epoch": 0.48, "grad_norm": 8.458754539489746, "learning_rate": 1.1090216517288226e-05, "loss": 1.6187, "step": 2990 }, { "epoch": 0.48, "grad_norm": 6.974617958068848, "learning_rate": 1.108503318790788e-05, "loss": 1.7114, "step": 2991 }, { "epoch": 0.48, "grad_norm": 7.525434494018555, "learning_rate": 1.1079849563523021e-05, "loss": 1.609, "step": 2992 }, { "epoch": 0.48, "grad_norm": 8.014806747436523, "learning_rate": 1.1074665645543008e-05, "loss": 1.4491, "step": 2993 }, { "epoch": 0.48, "grad_norm": 8.249286651611328, "learning_rate": 1.1069481435377267e-05, "loss": 1.5401, "step": 2994 }, { "epoch": 0.48, "grad_norm": 8.092357635498047, "learning_rate": 1.1064296934435307e-05, "loss": 1.5702, "step": 2995 }, { "epoch": 0.48, "grad_norm": 8.358024597167969, "learning_rate": 1.1059112144126715e-05, "loss": 1.281, "step": 2996 }, { "epoch": 0.48, "grad_norm": 9.967625617980957, "learning_rate": 1.1053927065861165e-05, "loss": 1.6586, "step": 2997 }, { "epoch": 0.48, "grad_norm": 7.687284469604492, "learning_rate": 1.1048741701048401e-05, "loss": 1.6128, "step": 2998 }, { "epoch": 0.48, "grad_norm": 7.61221170425415, "learning_rate": 1.104355605109825e-05, "loss": 1.9046, "step": 2999 }, { "epoch": 0.48, "grad_norm": 10.2304105758667, "learning_rate": 1.1038370117420607e-05, "loss": 1.5703, "step": 3000 }, { "epoch": 0.48, "grad_norm": 8.161955833435059, "learning_rate": 1.1033183901425452e-05, "loss": 1.4058, "step": 3001 }, { "epoch": 0.48, "grad_norm": 10.45551872253418, "learning_rate": 1.1027997404522848e-05, "loss": 1.7887, "step": 3002 }, { "epoch": 0.48, "grad_norm": 7.034506320953369, "learning_rate": 1.1022810628122918e-05, "loss": 1.3433, "step": 3003 }, { "epoch": 0.48, "grad_norm": 9.10641860961914, "learning_rate": 1.1017623573635876e-05, "loss": 1.6221, "step": 3004 }, { "epoch": 0.48, "grad_norm": 10.249463081359863, "learning_rate": 1.1012436242472002e-05, "loss": 1.4903, "step": 3005 }, { "epoch": 0.48, "grad_norm": 7.376263618469238, "learning_rate": 1.1007248636041658e-05, "loss": 1.4672, "step": 3006 }, { "epoch": 0.48, "grad_norm": 8.281874656677246, "learning_rate": 1.1002060755755276e-05, "loss": 1.4794, "step": 3007 }, { "epoch": 0.48, "grad_norm": 8.479479789733887, "learning_rate": 1.0996872603023362e-05, "loss": 1.7101, "step": 3008 }, { "epoch": 0.48, "grad_norm": 6.518590450286865, "learning_rate": 1.0991684179256502e-05, "loss": 1.5801, "step": 3009 }, { "epoch": 0.48, "grad_norm": 7.446047782897949, "learning_rate": 1.0986495485865348e-05, "loss": 1.6696, "step": 3010 }, { "epoch": 0.48, "grad_norm": 8.3037691116333, "learning_rate": 1.098130652426063e-05, "loss": 1.6478, "step": 3011 }, { "epoch": 0.48, "grad_norm": 7.329252243041992, "learning_rate": 1.0976117295853155e-05, "loss": 1.4254, "step": 3012 }, { "epoch": 0.49, "grad_norm": 7.729955196380615, "learning_rate": 1.0970927802053791e-05, "loss": 1.7251, "step": 3013 }, { "epoch": 0.49, "grad_norm": 9.48712158203125, "learning_rate": 1.0965738044273489e-05, "loss": 1.4688, "step": 3014 }, { "epoch": 0.49, "grad_norm": 9.34497356414795, "learning_rate": 1.0960548023923266e-05, "loss": 1.6604, "step": 3015 }, { "epoch": 0.49, "grad_norm": 10.415933609008789, "learning_rate": 1.095535774241421e-05, "loss": 1.5888, "step": 3016 }, { "epoch": 0.49, "grad_norm": 8.207254409790039, "learning_rate": 1.0950167201157484e-05, "loss": 1.4197, "step": 3017 }, { "epoch": 0.49, "grad_norm": 6.960496425628662, "learning_rate": 1.0944976401564323e-05, "loss": 1.6175, "step": 3018 }, { "epoch": 0.49, "grad_norm": 9.270720481872559, "learning_rate": 1.093978534504602e-05, "loss": 1.4898, "step": 3019 }, { "epoch": 0.49, "grad_norm": 8.054098129272461, "learning_rate": 1.0934594033013957e-05, "loss": 1.7172, "step": 3020 }, { "epoch": 0.49, "grad_norm": 7.676660537719727, "learning_rate": 1.0929402466879568e-05, "loss": 1.5556, "step": 3021 }, { "epoch": 0.49, "grad_norm": 8.388954162597656, "learning_rate": 1.0924210648054366e-05, "loss": 1.5665, "step": 3022 }, { "epoch": 0.49, "grad_norm": 8.356572151184082, "learning_rate": 1.0919018577949927e-05, "loss": 1.5907, "step": 3023 }, { "epoch": 0.49, "grad_norm": 7.530580043792725, "learning_rate": 1.0913826257977902e-05, "loss": 1.4803, "step": 3024 }, { "epoch": 0.49, "grad_norm": 8.5408935546875, "learning_rate": 1.0908633689550003e-05, "loss": 1.5082, "step": 3025 }, { "epoch": 0.49, "grad_norm": 10.290842056274414, "learning_rate": 1.0903440874078017e-05, "loss": 1.5943, "step": 3026 }, { "epoch": 0.49, "grad_norm": 8.693343162536621, "learning_rate": 1.089824781297379e-05, "loss": 1.5385, "step": 3027 }, { "epoch": 0.49, "grad_norm": 7.78187370300293, "learning_rate": 1.089305450764924e-05, "loss": 1.5192, "step": 3028 }, { "epoch": 0.49, "grad_norm": 8.712772369384766, "learning_rate": 1.088786095951635e-05, "loss": 1.5577, "step": 3029 }, { "epoch": 0.49, "grad_norm": 7.912591934204102, "learning_rate": 1.0882667169987166e-05, "loss": 1.6178, "step": 3030 }, { "epoch": 0.49, "grad_norm": 7.79211950302124, "learning_rate": 1.0877473140473812e-05, "loss": 1.4763, "step": 3031 }, { "epoch": 0.49, "grad_norm": 6.7855963706970215, "learning_rate": 1.0872278872388457e-05, "loss": 1.5536, "step": 3032 }, { "epoch": 0.49, "grad_norm": 6.659775733947754, "learning_rate": 1.0867084367143352e-05, "loss": 1.4544, "step": 3033 }, { "epoch": 0.49, "grad_norm": 9.2325439453125, "learning_rate": 1.0861889626150804e-05, "loss": 1.4387, "step": 3034 }, { "epoch": 0.49, "grad_norm": 6.918796539306641, "learning_rate": 1.0856694650823187e-05, "loss": 1.4554, "step": 3035 }, { "epoch": 0.49, "grad_norm": 8.03100299835205, "learning_rate": 1.0851499442572941e-05, "loss": 1.3869, "step": 3036 }, { "epoch": 0.49, "grad_norm": 8.518181800842285, "learning_rate": 1.0846304002812564e-05, "loss": 1.5925, "step": 3037 }, { "epoch": 0.49, "grad_norm": 7.43514347076416, "learning_rate": 1.0841108332954617e-05, "loss": 1.4099, "step": 3038 }, { "epoch": 0.49, "grad_norm": 8.481008529663086, "learning_rate": 1.0835912434411727e-05, "loss": 1.7187, "step": 3039 }, { "epoch": 0.49, "grad_norm": 9.191075325012207, "learning_rate": 1.0830716308596585e-05, "loss": 1.5592, "step": 3040 }, { "epoch": 0.49, "grad_norm": 8.686460494995117, "learning_rate": 1.0825519956921939e-05, "loss": 1.5722, "step": 3041 }, { "epoch": 0.49, "grad_norm": 7.652161121368408, "learning_rate": 1.0820323380800602e-05, "loss": 1.4239, "step": 3042 }, { "epoch": 0.49, "grad_norm": 6.956836223602295, "learning_rate": 1.0815126581645441e-05, "loss": 1.7265, "step": 3043 }, { "epoch": 0.49, "grad_norm": 9.359244346618652, "learning_rate": 1.0809929560869397e-05, "loss": 1.7364, "step": 3044 }, { "epoch": 0.49, "grad_norm": 7.701906681060791, "learning_rate": 1.0804732319885456e-05, "loss": 1.421, "step": 3045 }, { "epoch": 0.49, "grad_norm": 7.731686115264893, "learning_rate": 1.0799534860106676e-05, "loss": 1.7216, "step": 3046 }, { "epoch": 0.49, "grad_norm": 8.929662704467773, "learning_rate": 1.0794337182946171e-05, "loss": 1.734, "step": 3047 }, { "epoch": 0.49, "grad_norm": 7.783422946929932, "learning_rate": 1.0789139289817103e-05, "loss": 1.5107, "step": 3048 }, { "epoch": 0.49, "grad_norm": 7.003215789794922, "learning_rate": 1.0783941182132713e-05, "loss": 1.6019, "step": 3049 }, { "epoch": 0.49, "grad_norm": 7.543262004852295, "learning_rate": 1.0778742861306283e-05, "loss": 1.5197, "step": 3050 }, { "epoch": 0.49, "grad_norm": 7.7893500328063965, "learning_rate": 1.0773544328751166e-05, "loss": 1.8551, "step": 3051 }, { "epoch": 0.49, "grad_norm": 6.9833664894104, "learning_rate": 1.0768345585880765e-05, "loss": 1.2711, "step": 3052 }, { "epoch": 0.49, "grad_norm": 8.103517532348633, "learning_rate": 1.076314663410853e-05, "loss": 1.7105, "step": 3053 }, { "epoch": 0.49, "grad_norm": 6.968966960906982, "learning_rate": 1.0757947474847994e-05, "loss": 1.4169, "step": 3054 }, { "epoch": 0.49, "grad_norm": 8.186918258666992, "learning_rate": 1.0752748109512728e-05, "loss": 1.5577, "step": 3055 }, { "epoch": 0.49, "grad_norm": 7.461410045623779, "learning_rate": 1.0747548539516356e-05, "loss": 1.3856, "step": 3056 }, { "epoch": 0.49, "grad_norm": 8.607996940612793, "learning_rate": 1.0742348766272572e-05, "loss": 1.7187, "step": 3057 }, { "epoch": 0.49, "grad_norm": 7.159780979156494, "learning_rate": 1.0737148791195116e-05, "loss": 1.3205, "step": 3058 }, { "epoch": 0.49, "grad_norm": 9.888628005981445, "learning_rate": 1.0731948615697783e-05, "loss": 1.6345, "step": 3059 }, { "epoch": 0.49, "grad_norm": 8.846323013305664, "learning_rate": 1.0726748241194422e-05, "loss": 1.8338, "step": 3060 }, { "epoch": 0.49, "grad_norm": 8.600767135620117, "learning_rate": 1.0721547669098943e-05, "loss": 1.3759, "step": 3061 }, { "epoch": 0.49, "grad_norm": 8.292858123779297, "learning_rate": 1.0716346900825298e-05, "loss": 1.4923, "step": 3062 }, { "epoch": 0.49, "grad_norm": 8.276534080505371, "learning_rate": 1.0711145937787506e-05, "loss": 1.8205, "step": 3063 }, { "epoch": 0.49, "grad_norm": 8.012430191040039, "learning_rate": 1.070594478139963e-05, "loss": 1.4454, "step": 3064 }, { "epoch": 0.49, "grad_norm": 6.983343124389648, "learning_rate": 1.0700743433075784e-05, "loss": 1.5892, "step": 3065 }, { "epoch": 0.49, "grad_norm": 7.970432758331299, "learning_rate": 1.0695541894230143e-05, "loss": 1.475, "step": 3066 }, { "epoch": 0.49, "grad_norm": 7.81063985824585, "learning_rate": 1.0690340166276927e-05, "loss": 1.679, "step": 3067 }, { "epoch": 0.49, "grad_norm": 8.747967720031738, "learning_rate": 1.0685138250630408e-05, "loss": 1.5481, "step": 3068 }, { "epoch": 0.49, "grad_norm": 8.722548484802246, "learning_rate": 1.067993614870491e-05, "loss": 1.5412, "step": 3069 }, { "epoch": 0.49, "grad_norm": 7.895411491394043, "learning_rate": 1.0674733861914808e-05, "loss": 1.6314, "step": 3070 }, { "epoch": 0.49, "grad_norm": 6.787002086639404, "learning_rate": 1.0669531391674528e-05, "loss": 1.3343, "step": 3071 }, { "epoch": 0.49, "grad_norm": 9.15383243560791, "learning_rate": 1.0664328739398542e-05, "loss": 1.4441, "step": 3072 }, { "epoch": 0.49, "grad_norm": 6.895575046539307, "learning_rate": 1.065912590650138e-05, "loss": 1.432, "step": 3073 }, { "epoch": 0.49, "grad_norm": 7.825084686279297, "learning_rate": 1.065392289439761e-05, "loss": 1.8444, "step": 3074 }, { "epoch": 0.5, "grad_norm": 8.158355712890625, "learning_rate": 1.0648719704501858e-05, "loss": 1.6702, "step": 3075 }, { "epoch": 0.5, "grad_norm": 7.76922082901001, "learning_rate": 1.0643516338228792e-05, "loss": 1.6116, "step": 3076 }, { "epoch": 0.5, "grad_norm": 8.132482528686523, "learning_rate": 1.0638312796993128e-05, "loss": 1.4969, "step": 3077 }, { "epoch": 0.5, "grad_norm": 8.268080711364746, "learning_rate": 1.0633109082209638e-05, "loss": 1.8323, "step": 3078 }, { "epoch": 0.5, "grad_norm": 7.700231552124023, "learning_rate": 1.0627905195293135e-05, "loss": 1.6427, "step": 3079 }, { "epoch": 0.5, "grad_norm": 10.421917915344238, "learning_rate": 1.0622701137658475e-05, "loss": 1.7604, "step": 3080 }, { "epoch": 0.5, "grad_norm": 6.967203617095947, "learning_rate": 1.0617496910720565e-05, "loss": 1.5214, "step": 3081 }, { "epoch": 0.5, "grad_norm": 8.550183296203613, "learning_rate": 1.0612292515894358e-05, "loss": 1.5445, "step": 3082 }, { "epoch": 0.5, "grad_norm": 7.545625686645508, "learning_rate": 1.0607087954594854e-05, "loss": 1.6168, "step": 3083 }, { "epoch": 0.5, "grad_norm": 8.15576457977295, "learning_rate": 1.0601883228237098e-05, "loss": 1.4802, "step": 3084 }, { "epoch": 0.5, "grad_norm": 6.734289169311523, "learning_rate": 1.0596678338236173e-05, "loss": 1.6876, "step": 3085 }, { "epoch": 0.5, "grad_norm": 9.18223762512207, "learning_rate": 1.059147328600721e-05, "loss": 1.504, "step": 3086 }, { "epoch": 0.5, "grad_norm": 8.1949462890625, "learning_rate": 1.0586268072965395e-05, "loss": 1.4337, "step": 3087 }, { "epoch": 0.5, "grad_norm": 9.883711814880371, "learning_rate": 1.0581062700525941e-05, "loss": 1.5178, "step": 3088 }, { "epoch": 0.5, "grad_norm": 8.165512084960938, "learning_rate": 1.057585717010412e-05, "loss": 1.6663, "step": 3089 }, { "epoch": 0.5, "grad_norm": 9.440312385559082, "learning_rate": 1.057065148311523e-05, "loss": 1.7027, "step": 3090 }, { "epoch": 0.5, "grad_norm": 7.8010969161987305, "learning_rate": 1.056544564097462e-05, "loss": 1.4317, "step": 3091 }, { "epoch": 0.5, "grad_norm": 7.187862396240234, "learning_rate": 1.056023964509769e-05, "loss": 1.5888, "step": 3092 }, { "epoch": 0.5, "grad_norm": 6.889472484588623, "learning_rate": 1.0555033496899868e-05, "loss": 1.2558, "step": 3093 }, { "epoch": 0.5, "grad_norm": 8.160175323486328, "learning_rate": 1.054982719779663e-05, "loss": 1.4255, "step": 3094 }, { "epoch": 0.5, "grad_norm": 8.269631385803223, "learning_rate": 1.0544620749203492e-05, "loss": 1.5636, "step": 3095 }, { "epoch": 0.5, "grad_norm": 8.740200996398926, "learning_rate": 1.053941415253601e-05, "loss": 1.5454, "step": 3096 }, { "epoch": 0.5, "grad_norm": 7.5935378074646, "learning_rate": 1.0534207409209783e-05, "loss": 1.5856, "step": 3097 }, { "epoch": 0.5, "grad_norm": 7.795093059539795, "learning_rate": 1.0529000520640443e-05, "loss": 1.5139, "step": 3098 }, { "epoch": 0.5, "grad_norm": 8.565417289733887, "learning_rate": 1.052379348824367e-05, "loss": 1.6816, "step": 3099 }, { "epoch": 0.5, "grad_norm": 8.575765609741211, "learning_rate": 1.051858631343518e-05, "loss": 1.4713, "step": 3100 }, { "epoch": 0.5, "grad_norm": 7.014543056488037, "learning_rate": 1.0513378997630723e-05, "loss": 1.5114, "step": 3101 }, { "epoch": 0.5, "grad_norm": 7.890753269195557, "learning_rate": 1.0508171542246093e-05, "loss": 1.6853, "step": 3102 }, { "epoch": 0.5, "grad_norm": 7.145125389099121, "learning_rate": 1.0502963948697123e-05, "loss": 1.7303, "step": 3103 }, { "epoch": 0.5, "grad_norm": 9.117389678955078, "learning_rate": 1.0497756218399677e-05, "loss": 1.348, "step": 3104 }, { "epoch": 0.5, "grad_norm": 7.25550651550293, "learning_rate": 1.0492548352769662e-05, "loss": 1.4847, "step": 3105 }, { "epoch": 0.5, "grad_norm": 7.634192943572998, "learning_rate": 1.0487340353223018e-05, "loss": 1.527, "step": 3106 }, { "epoch": 0.5, "grad_norm": 7.8409833908081055, "learning_rate": 1.0482132221175727e-05, "loss": 1.5717, "step": 3107 }, { "epoch": 0.5, "grad_norm": 8.771851539611816, "learning_rate": 1.04769239580438e-05, "loss": 1.4437, "step": 3108 }, { "epoch": 0.5, "grad_norm": 7.17968225479126, "learning_rate": 1.0471715565243288e-05, "loss": 1.6573, "step": 3109 }, { "epoch": 0.5, "grad_norm": 7.26973819732666, "learning_rate": 1.0466507044190274e-05, "loss": 1.278, "step": 3110 }, { "epoch": 0.5, "grad_norm": 8.357109069824219, "learning_rate": 1.0461298396300884e-05, "loss": 1.5254, "step": 3111 }, { "epoch": 0.5, "grad_norm": 7.013434410095215, "learning_rate": 1.0456089622991264e-05, "loss": 1.4626, "step": 3112 }, { "epoch": 0.5, "grad_norm": 7.5604634284973145, "learning_rate": 1.0450880725677607e-05, "loss": 1.512, "step": 3113 }, { "epoch": 0.5, "grad_norm": 7.370473384857178, "learning_rate": 1.0445671705776137e-05, "loss": 1.4526, "step": 3114 }, { "epoch": 0.5, "grad_norm": 8.236638069152832, "learning_rate": 1.0440462564703107e-05, "loss": 1.5339, "step": 3115 }, { "epoch": 0.5, "grad_norm": 8.641007423400879, "learning_rate": 1.0435253303874808e-05, "loss": 1.9597, "step": 3116 }, { "epoch": 0.5, "grad_norm": 7.1637420654296875, "learning_rate": 1.0430043924707559e-05, "loss": 1.713, "step": 3117 }, { "epoch": 0.5, "grad_norm": 8.245264053344727, "learning_rate": 1.0424834428617715e-05, "loss": 1.6781, "step": 3118 }, { "epoch": 0.5, "grad_norm": 7.742191791534424, "learning_rate": 1.0419624817021657e-05, "loss": 1.6874, "step": 3119 }, { "epoch": 0.5, "grad_norm": 7.893298625946045, "learning_rate": 1.0414415091335804e-05, "loss": 1.3745, "step": 3120 }, { "epoch": 0.5, "grad_norm": 7.0894389152526855, "learning_rate": 1.0409205252976609e-05, "loss": 1.4768, "step": 3121 }, { "epoch": 0.5, "grad_norm": 8.319713592529297, "learning_rate": 1.040399530336054e-05, "loss": 1.5596, "step": 3122 }, { "epoch": 0.5, "grad_norm": 7.656250953674316, "learning_rate": 1.0398785243904113e-05, "loss": 1.5091, "step": 3123 }, { "epoch": 0.5, "grad_norm": 6.852848052978516, "learning_rate": 1.0393575076023863e-05, "loss": 1.6681, "step": 3124 }, { "epoch": 0.5, "grad_norm": 8.325231552124023, "learning_rate": 1.0388364801136357e-05, "loss": 1.7054, "step": 3125 }, { "epoch": 0.5, "grad_norm": 8.539438247680664, "learning_rate": 1.03831544206582e-05, "loss": 1.4311, "step": 3126 }, { "epoch": 0.5, "grad_norm": 7.639702796936035, "learning_rate": 1.0377943936006005e-05, "loss": 1.4607, "step": 3127 }, { "epoch": 0.5, "grad_norm": 7.496586322784424, "learning_rate": 1.0372733348596433e-05, "loss": 1.4375, "step": 3128 }, { "epoch": 0.5, "grad_norm": 8.123563766479492, "learning_rate": 1.0367522659846163e-05, "loss": 1.4449, "step": 3129 }, { "epoch": 0.5, "grad_norm": 8.311209678649902, "learning_rate": 1.0362311871171908e-05, "loss": 1.5703, "step": 3130 }, { "epoch": 0.5, "grad_norm": 8.947867393493652, "learning_rate": 1.0357100983990405e-05, "loss": 1.4686, "step": 3131 }, { "epoch": 0.5, "grad_norm": 7.376493453979492, "learning_rate": 1.0351889999718413e-05, "loss": 1.4476, "step": 3132 }, { "epoch": 0.5, "grad_norm": 6.744120121002197, "learning_rate": 1.0346678919772723e-05, "loss": 1.6001, "step": 3133 }, { "epoch": 0.5, "grad_norm": 7.840961933135986, "learning_rate": 1.034146774557015e-05, "loss": 1.4409, "step": 3134 }, { "epoch": 0.5, "grad_norm": 8.29100227355957, "learning_rate": 1.0336256478527538e-05, "loss": 1.6743, "step": 3135 }, { "epoch": 0.5, "grad_norm": 7.500393867492676, "learning_rate": 1.0331045120061753e-05, "loss": 1.7151, "step": 3136 }, { "epoch": 0.5, "grad_norm": 8.944930076599121, "learning_rate": 1.0325833671589687e-05, "loss": 1.424, "step": 3137 }, { "epoch": 0.51, "grad_norm": 7.571225166320801, "learning_rate": 1.032062213452825e-05, "loss": 1.4654, "step": 3138 }, { "epoch": 0.51, "grad_norm": 8.874655723571777, "learning_rate": 1.031541051029439e-05, "loss": 1.6864, "step": 3139 }, { "epoch": 0.51, "grad_norm": 9.339982032775879, "learning_rate": 1.0310198800305066e-05, "loss": 1.6195, "step": 3140 }, { "epoch": 0.51, "grad_norm": 11.11643123626709, "learning_rate": 1.0304987005977269e-05, "loss": 1.5353, "step": 3141 }, { "epoch": 0.51, "grad_norm": 7.370965480804443, "learning_rate": 1.0299775128728007e-05, "loss": 1.4166, "step": 3142 }, { "epoch": 0.51, "grad_norm": 7.057020664215088, "learning_rate": 1.0294563169974308e-05, "loss": 1.2531, "step": 3143 }, { "epoch": 0.51, "grad_norm": 7.454875469207764, "learning_rate": 1.0289351131133232e-05, "loss": 1.4494, "step": 3144 }, { "epoch": 0.51, "grad_norm": 8.979438781738281, "learning_rate": 1.0284139013621854e-05, "loss": 1.6187, "step": 3145 }, { "epoch": 0.51, "grad_norm": 7.947530269622803, "learning_rate": 1.027892681885727e-05, "loss": 1.7888, "step": 3146 }, { "epoch": 0.51, "grad_norm": 7.800020217895508, "learning_rate": 1.0273714548256603e-05, "loss": 1.6018, "step": 3147 }, { "epoch": 0.51, "grad_norm": 8.703261375427246, "learning_rate": 1.0268502203236993e-05, "loss": 1.2688, "step": 3148 }, { "epoch": 0.51, "grad_norm": 9.840922355651855, "learning_rate": 1.026328978521559e-05, "loss": 1.6113, "step": 3149 }, { "epoch": 0.51, "grad_norm": 7.885308265686035, "learning_rate": 1.0258077295609585e-05, "loss": 1.535, "step": 3150 }, { "epoch": 0.51, "grad_norm": 8.294088363647461, "learning_rate": 1.0252864735836172e-05, "loss": 1.4225, "step": 3151 }, { "epoch": 0.51, "grad_norm": 7.010716438293457, "learning_rate": 1.0247652107312566e-05, "loss": 1.2307, "step": 3152 }, { "epoch": 0.51, "grad_norm": 7.535408020019531, "learning_rate": 1.024243941145601e-05, "loss": 1.6231, "step": 3153 }, { "epoch": 0.51, "grad_norm": 8.599109649658203, "learning_rate": 1.0237226649683756e-05, "loss": 1.7051, "step": 3154 }, { "epoch": 0.51, "grad_norm": 7.8232550621032715, "learning_rate": 1.0232013823413078e-05, "loss": 1.6586, "step": 3155 }, { "epoch": 0.51, "grad_norm": 7.768212795257568, "learning_rate": 1.0226800934061264e-05, "loss": 1.3961, "step": 3156 }, { "epoch": 0.51, "grad_norm": 11.291053771972656, "learning_rate": 1.0221587983045624e-05, "loss": 1.5081, "step": 3157 }, { "epoch": 0.51, "grad_norm": 7.291332244873047, "learning_rate": 1.0216374971783482e-05, "loss": 1.4521, "step": 3158 }, { "epoch": 0.51, "grad_norm": 7.413994312286377, "learning_rate": 1.0211161901692176e-05, "loss": 1.5906, "step": 3159 }, { "epoch": 0.51, "grad_norm": 9.099884986877441, "learning_rate": 1.0205948774189069e-05, "loss": 1.3235, "step": 3160 }, { "epoch": 0.51, "grad_norm": 9.1386137008667, "learning_rate": 1.020073559069153e-05, "loss": 1.5966, "step": 3161 }, { "epoch": 0.51, "grad_norm": 7.108361721038818, "learning_rate": 1.0195522352616942e-05, "loss": 1.1708, "step": 3162 }, { "epoch": 0.51, "grad_norm": 7.354628086090088, "learning_rate": 1.0190309061382717e-05, "loss": 1.4779, "step": 3163 }, { "epoch": 0.51, "grad_norm": 6.758542537689209, "learning_rate": 1.0185095718406266e-05, "loss": 1.5244, "step": 3164 }, { "epoch": 0.51, "grad_norm": 8.254945755004883, "learning_rate": 1.0179882325105021e-05, "loss": 1.7134, "step": 3165 }, { "epoch": 0.51, "grad_norm": 8.478456497192383, "learning_rate": 1.0174668882896426e-05, "loss": 1.4521, "step": 3166 }, { "epoch": 0.51, "grad_norm": 7.708477020263672, "learning_rate": 1.0169455393197937e-05, "loss": 1.6766, "step": 3167 }, { "epoch": 0.51, "grad_norm": 9.316699981689453, "learning_rate": 1.0164241857427033e-05, "loss": 1.3434, "step": 3168 }, { "epoch": 0.51, "grad_norm": 7.432011127471924, "learning_rate": 1.0159028277001189e-05, "loss": 1.5507, "step": 3169 }, { "epoch": 0.51, "grad_norm": 9.949041366577148, "learning_rate": 1.0153814653337904e-05, "loss": 1.4982, "step": 3170 }, { "epoch": 0.51, "grad_norm": 8.2584810256958, "learning_rate": 1.0148600987854685e-05, "loss": 1.4302, "step": 3171 }, { "epoch": 0.51, "grad_norm": 7.159143447875977, "learning_rate": 1.0143387281969048e-05, "loss": 1.2646, "step": 3172 }, { "epoch": 0.51, "grad_norm": 8.295003890991211, "learning_rate": 1.0138173537098525e-05, "loss": 1.6925, "step": 3173 }, { "epoch": 0.51, "grad_norm": 9.508624076843262, "learning_rate": 1.0132959754660663e-05, "loss": 1.6059, "step": 3174 }, { "epoch": 0.51, "grad_norm": 10.047199249267578, "learning_rate": 1.0127745936073001e-05, "loss": 1.6753, "step": 3175 }, { "epoch": 0.51, "grad_norm": 7.141935348510742, "learning_rate": 1.0122532082753106e-05, "loss": 1.4358, "step": 3176 }, { "epoch": 0.51, "grad_norm": 8.289459228515625, "learning_rate": 1.0117318196118544e-05, "loss": 1.5829, "step": 3177 }, { "epoch": 0.51, "grad_norm": 7.625728607177734, "learning_rate": 1.0112104277586898e-05, "loss": 1.411, "step": 3178 }, { "epoch": 0.51, "grad_norm": 9.369670867919922, "learning_rate": 1.0106890328575756e-05, "loss": 1.4075, "step": 3179 }, { "epoch": 0.51, "grad_norm": 10.035778045654297, "learning_rate": 1.0101676350502707e-05, "loss": 1.4269, "step": 3180 }, { "epoch": 0.51, "grad_norm": 6.48426628112793, "learning_rate": 1.0096462344785358e-05, "loss": 1.4762, "step": 3181 }, { "epoch": 0.51, "grad_norm": 7.6061625480651855, "learning_rate": 1.0091248312841325e-05, "loss": 1.5037, "step": 3182 }, { "epoch": 0.51, "grad_norm": 8.231671333312988, "learning_rate": 1.008603425608822e-05, "loss": 1.3894, "step": 3183 }, { "epoch": 0.51, "grad_norm": 9.499314308166504, "learning_rate": 1.0080820175943673e-05, "loss": 1.4763, "step": 3184 }, { "epoch": 0.51, "grad_norm": 8.304411888122559, "learning_rate": 1.0075606073825313e-05, "loss": 1.5156, "step": 3185 }, { "epoch": 0.51, "grad_norm": 7.196983814239502, "learning_rate": 1.0070391951150774e-05, "loss": 1.3562, "step": 3186 }, { "epoch": 0.51, "grad_norm": 6.391162395477295, "learning_rate": 1.0065177809337703e-05, "loss": 1.2605, "step": 3187 }, { "epoch": 0.51, "grad_norm": 7.341288089752197, "learning_rate": 1.005996364980375e-05, "loss": 1.293, "step": 3188 }, { "epoch": 0.51, "grad_norm": 8.896180152893066, "learning_rate": 1.0054749473966565e-05, "loss": 1.7582, "step": 3189 }, { "epoch": 0.51, "grad_norm": 8.631267547607422, "learning_rate": 1.0049535283243806e-05, "loss": 1.5119, "step": 3190 }, { "epoch": 0.51, "grad_norm": 10.517976760864258, "learning_rate": 1.0044321079053131e-05, "loss": 1.5387, "step": 3191 }, { "epoch": 0.51, "grad_norm": 6.668485164642334, "learning_rate": 1.0039106862812214e-05, "loss": 1.5152, "step": 3192 }, { "epoch": 0.51, "grad_norm": 8.006475448608398, "learning_rate": 1.0033892635938715e-05, "loss": 1.4291, "step": 3193 }, { "epoch": 0.51, "grad_norm": 8.647627830505371, "learning_rate": 1.0028678399850311e-05, "loss": 1.5028, "step": 3194 }, { "epoch": 0.51, "grad_norm": 7.926056385040283, "learning_rate": 1.0023464155964671e-05, "loss": 1.5426, "step": 3195 }, { "epoch": 0.51, "grad_norm": 8.90684986114502, "learning_rate": 1.0018249905699475e-05, "loss": 1.6054, "step": 3196 }, { "epoch": 0.51, "grad_norm": 7.310891628265381, "learning_rate": 1.0013035650472399e-05, "loss": 1.5247, "step": 3197 }, { "epoch": 0.51, "grad_norm": 11.596370697021484, "learning_rate": 1.000782139170112e-05, "loss": 1.7354, "step": 3198 }, { "epoch": 0.51, "grad_norm": 9.102843284606934, "learning_rate": 1.0002607130803321e-05, "loss": 1.441, "step": 3199 }, { "epoch": 0.52, "grad_norm": 8.034178733825684, "learning_rate": 9.99739286919668e-06, "loss": 1.4513, "step": 3200 }, { "epoch": 0.52, "grad_norm": 7.654232501983643, "learning_rate": 9.992178608298882e-06, "loss": 1.5514, "step": 3201 }, { "epoch": 0.52, "grad_norm": 8.77811336517334, "learning_rate": 9.986964349527606e-06, "loss": 1.4763, "step": 3202 }, { "epoch": 0.52, "grad_norm": 8.835768699645996, "learning_rate": 9.98175009430053e-06, "loss": 1.3442, "step": 3203 }, { "epoch": 0.52, "grad_norm": 8.18421459197998, "learning_rate": 9.976535844035332e-06, "loss": 1.6046, "step": 3204 }, { "epoch": 0.52, "grad_norm": 6.9323410987854, "learning_rate": 9.971321600149692e-06, "loss": 1.497, "step": 3205 }, { "epoch": 0.52, "grad_norm": 7.566946506500244, "learning_rate": 9.966107364061285e-06, "loss": 1.6614, "step": 3206 }, { "epoch": 0.52, "grad_norm": 7.2143378257751465, "learning_rate": 9.96089313718779e-06, "loss": 1.4717, "step": 3207 }, { "epoch": 0.52, "grad_norm": 7.604838848114014, "learning_rate": 9.955678920946874e-06, "loss": 1.4907, "step": 3208 }, { "epoch": 0.52, "grad_norm": 8.668562889099121, "learning_rate": 9.950464716756199e-06, "loss": 1.4491, "step": 3209 }, { "epoch": 0.52, "grad_norm": 11.527807235717773, "learning_rate": 9.945250526033438e-06, "loss": 1.642, "step": 3210 }, { "epoch": 0.52, "grad_norm": 10.456160545349121, "learning_rate": 9.940036350196251e-06, "loss": 1.589, "step": 3211 }, { "epoch": 0.52, "grad_norm": 10.889249801635742, "learning_rate": 9.934822190662299e-06, "loss": 1.8124, "step": 3212 }, { "epoch": 0.52, "grad_norm": 8.463492393493652, "learning_rate": 9.92960804884923e-06, "loss": 1.504, "step": 3213 }, { "epoch": 0.52, "grad_norm": 8.693138122558594, "learning_rate": 9.92439392617469e-06, "loss": 1.4339, "step": 3214 }, { "epoch": 0.52, "grad_norm": 7.911741733551025, "learning_rate": 9.91917982405633e-06, "loss": 1.477, "step": 3215 }, { "epoch": 0.52, "grad_norm": 8.911535263061523, "learning_rate": 9.91396574391178e-06, "loss": 1.6765, "step": 3216 }, { "epoch": 0.52, "grad_norm": 9.019137382507324, "learning_rate": 9.908751687158679e-06, "loss": 1.3206, "step": 3217 }, { "epoch": 0.52, "grad_norm": 8.22321605682373, "learning_rate": 9.903537655214644e-06, "loss": 1.3703, "step": 3218 }, { "epoch": 0.52, "grad_norm": 8.769412994384766, "learning_rate": 9.898323649497294e-06, "loss": 1.2911, "step": 3219 }, { "epoch": 0.52, "grad_norm": 7.477945327758789, "learning_rate": 9.893109671424249e-06, "loss": 1.3874, "step": 3220 }, { "epoch": 0.52, "grad_norm": 9.519975662231445, "learning_rate": 9.887895722413102e-06, "loss": 1.4983, "step": 3221 }, { "epoch": 0.52, "grad_norm": 10.191900253295898, "learning_rate": 9.88268180388146e-06, "loss": 1.196, "step": 3222 }, { "epoch": 0.52, "grad_norm": 6.551916599273682, "learning_rate": 9.877467917246896e-06, "loss": 1.4163, "step": 3223 }, { "epoch": 0.52, "grad_norm": 8.872427940368652, "learning_rate": 9.872254063927e-06, "loss": 1.5239, "step": 3224 }, { "epoch": 0.52, "grad_norm": 9.317096710205078, "learning_rate": 9.867040245339342e-06, "loss": 1.6797, "step": 3225 }, { "epoch": 0.52, "grad_norm": 7.416871547698975, "learning_rate": 9.861826462901473e-06, "loss": 1.3926, "step": 3226 }, { "epoch": 0.52, "grad_norm": 7.454830169677734, "learning_rate": 9.856612718030955e-06, "loss": 1.5334, "step": 3227 }, { "epoch": 0.52, "grad_norm": 8.10920524597168, "learning_rate": 9.851399012145318e-06, "loss": 1.4137, "step": 3228 }, { "epoch": 0.52, "grad_norm": 7.120640754699707, "learning_rate": 9.8461853466621e-06, "loss": 1.2989, "step": 3229 }, { "epoch": 0.52, "grad_norm": 8.329580307006836, "learning_rate": 9.840971722998816e-06, "loss": 1.4201, "step": 3230 }, { "epoch": 0.52, "grad_norm": 8.704961776733398, "learning_rate": 9.83575814257297e-06, "loss": 1.8238, "step": 3231 }, { "epoch": 0.52, "grad_norm": 9.269977569580078, "learning_rate": 9.830544606802066e-06, "loss": 1.3801, "step": 3232 }, { "epoch": 0.52, "grad_norm": 7.781318187713623, "learning_rate": 9.825331117103578e-06, "loss": 1.4, "step": 3233 }, { "epoch": 0.52, "grad_norm": 7.241171836853027, "learning_rate": 9.820117674894982e-06, "loss": 1.5078, "step": 3234 }, { "epoch": 0.52, "grad_norm": 7.787428855895996, "learning_rate": 9.81490428159374e-06, "loss": 1.5031, "step": 3235 }, { "epoch": 0.52, "grad_norm": 7.686794757843018, "learning_rate": 9.809690938617285e-06, "loss": 1.4064, "step": 3236 }, { "epoch": 0.52, "grad_norm": 7.159696102142334, "learning_rate": 9.804477647383061e-06, "loss": 1.3379, "step": 3237 }, { "epoch": 0.52, "grad_norm": 9.598908424377441, "learning_rate": 9.799264409308473e-06, "loss": 1.8199, "step": 3238 }, { "epoch": 0.52, "grad_norm": 8.705483436584473, "learning_rate": 9.794051225810933e-06, "loss": 1.6171, "step": 3239 }, { "epoch": 0.52, "grad_norm": 7.99992561340332, "learning_rate": 9.788838098307827e-06, "loss": 1.6464, "step": 3240 }, { "epoch": 0.52, "grad_norm": 7.419693946838379, "learning_rate": 9.783625028216521e-06, "loss": 1.4043, "step": 3241 }, { "epoch": 0.52, "grad_norm": 7.316072463989258, "learning_rate": 9.77841201695438e-06, "loss": 1.7103, "step": 3242 }, { "epoch": 0.52, "grad_norm": 7.648674011230469, "learning_rate": 9.773199065938737e-06, "loss": 1.5971, "step": 3243 }, { "epoch": 0.52, "grad_norm": 6.9208478927612305, "learning_rate": 9.767986176586925e-06, "loss": 1.5143, "step": 3244 }, { "epoch": 0.52, "grad_norm": 7.0013651847839355, "learning_rate": 9.762773350316247e-06, "loss": 1.494, "step": 3245 }, { "epoch": 0.52, "grad_norm": 7.893321514129639, "learning_rate": 9.757560588543991e-06, "loss": 1.5992, "step": 3246 }, { "epoch": 0.52, "grad_norm": 10.810249328613281, "learning_rate": 9.752347892687435e-06, "loss": 1.6584, "step": 3247 }, { "epoch": 0.52, "grad_norm": 7.362729549407959, "learning_rate": 9.747135264163829e-06, "loss": 1.6127, "step": 3248 }, { "epoch": 0.52, "grad_norm": 8.826040267944336, "learning_rate": 9.741922704390418e-06, "loss": 1.5329, "step": 3249 }, { "epoch": 0.52, "grad_norm": 7.574654579162598, "learning_rate": 9.736710214784415e-06, "loss": 1.7354, "step": 3250 }, { "epoch": 0.52, "grad_norm": 6.770334243774414, "learning_rate": 9.731497796763012e-06, "loss": 1.3192, "step": 3251 }, { "epoch": 0.52, "grad_norm": 8.360241889953613, "learning_rate": 9.726285451743399e-06, "loss": 1.6765, "step": 3252 }, { "epoch": 0.52, "grad_norm": 7.585373401641846, "learning_rate": 9.721073181142728e-06, "loss": 1.6341, "step": 3253 }, { "epoch": 0.52, "grad_norm": 12.438690185546875, "learning_rate": 9.71586098637815e-06, "loss": 1.7055, "step": 3254 }, { "epoch": 0.52, "grad_norm": 7.255761623382568, "learning_rate": 9.710648868866773e-06, "loss": 1.5684, "step": 3255 }, { "epoch": 0.52, "grad_norm": 7.352438449859619, "learning_rate": 9.705436830025695e-06, "loss": 1.4555, "step": 3256 }, { "epoch": 0.52, "grad_norm": 7.996696949005127, "learning_rate": 9.700224871271998e-06, "loss": 1.5533, "step": 3257 }, { "epoch": 0.52, "grad_norm": 8.0927734375, "learning_rate": 9.695012994022733e-06, "loss": 1.7883, "step": 3258 }, { "epoch": 0.52, "grad_norm": 7.831493377685547, "learning_rate": 9.689801199694935e-06, "loss": 1.4812, "step": 3259 }, { "epoch": 0.52, "grad_norm": 7.823376655578613, "learning_rate": 9.68458948970561e-06, "loss": 1.4938, "step": 3260 }, { "epoch": 0.52, "grad_norm": 9.839712142944336, "learning_rate": 9.679377865471751e-06, "loss": 1.4581, "step": 3261 }, { "epoch": 0.53, "grad_norm": 7.682333469390869, "learning_rate": 9.674166328410318e-06, "loss": 1.5739, "step": 3262 }, { "epoch": 0.53, "grad_norm": 9.64484977722168, "learning_rate": 9.668954879938249e-06, "loss": 1.7193, "step": 3263 }, { "epoch": 0.53, "grad_norm": 8.185096740722656, "learning_rate": 9.663743521472466e-06, "loss": 1.618, "step": 3264 }, { "epoch": 0.53, "grad_norm": 8.385231971740723, "learning_rate": 9.65853225442985e-06, "loss": 1.5987, "step": 3265 }, { "epoch": 0.53, "grad_norm": 7.821916103363037, "learning_rate": 9.65332108022728e-06, "loss": 1.2176, "step": 3266 }, { "epoch": 0.53, "grad_norm": 6.659529685974121, "learning_rate": 9.648110000281592e-06, "loss": 1.4948, "step": 3267 }, { "epoch": 0.53, "grad_norm": 7.969651222229004, "learning_rate": 9.642899016009596e-06, "loss": 1.6102, "step": 3268 }, { "epoch": 0.53, "grad_norm": 6.855320930480957, "learning_rate": 9.637688128828095e-06, "loss": 1.5449, "step": 3269 }, { "epoch": 0.53, "grad_norm": 7.094454288482666, "learning_rate": 9.632477340153837e-06, "loss": 1.4902, "step": 3270 }, { "epoch": 0.53, "grad_norm": 7.028907299041748, "learning_rate": 9.62726665140357e-06, "loss": 1.4178, "step": 3271 }, { "epoch": 0.53, "grad_norm": 7.631662368774414, "learning_rate": 9.622056063994e-06, "loss": 1.4622, "step": 3272 }, { "epoch": 0.53, "grad_norm": 7.680933475494385, "learning_rate": 9.616845579341803e-06, "loss": 1.5628, "step": 3273 }, { "epoch": 0.53, "grad_norm": 8.0335693359375, "learning_rate": 9.611635198863645e-06, "loss": 1.5273, "step": 3274 }, { "epoch": 0.53, "grad_norm": 6.2591681480407715, "learning_rate": 9.606424923976138e-06, "loss": 1.3135, "step": 3275 }, { "epoch": 0.53, "grad_norm": 7.828134059906006, "learning_rate": 9.601214756095889e-06, "loss": 1.6307, "step": 3276 }, { "epoch": 0.53, "grad_norm": 8.060234069824219, "learning_rate": 9.596004696639464e-06, "loss": 1.438, "step": 3277 }, { "epoch": 0.53, "grad_norm": 9.23891830444336, "learning_rate": 9.590794747023394e-06, "loss": 1.5748, "step": 3278 }, { "epoch": 0.53, "grad_norm": 7.401759624481201, "learning_rate": 9.5855849086642e-06, "loss": 1.4338, "step": 3279 }, { "epoch": 0.53, "grad_norm": 10.066822052001953, "learning_rate": 9.580375182978347e-06, "loss": 1.6356, "step": 3280 }, { "epoch": 0.53, "grad_norm": 8.238266944885254, "learning_rate": 9.57516557138229e-06, "loss": 1.3664, "step": 3281 }, { "epoch": 0.53, "grad_norm": 8.246448516845703, "learning_rate": 9.569956075292446e-06, "loss": 1.4919, "step": 3282 }, { "epoch": 0.53, "grad_norm": 6.98387336730957, "learning_rate": 9.564746696125194e-06, "loss": 1.5651, "step": 3283 }, { "epoch": 0.53, "grad_norm": 8.836977005004883, "learning_rate": 9.559537435296898e-06, "loss": 1.4546, "step": 3284 }, { "epoch": 0.53, "grad_norm": 10.100865364074707, "learning_rate": 9.554328294223865e-06, "loss": 1.5746, "step": 3285 }, { "epoch": 0.53, "grad_norm": 10.453890800476074, "learning_rate": 9.549119274322394e-06, "loss": 1.7702, "step": 3286 }, { "epoch": 0.53, "grad_norm": 8.019362449645996, "learning_rate": 9.543910377008741e-06, "loss": 1.7043, "step": 3287 }, { "epoch": 0.53, "grad_norm": 6.947462558746338, "learning_rate": 9.538701603699121e-06, "loss": 1.6473, "step": 3288 }, { "epoch": 0.53, "grad_norm": 7.585561275482178, "learning_rate": 9.53349295580973e-06, "loss": 1.4384, "step": 3289 }, { "epoch": 0.53, "grad_norm": 7.234795093536377, "learning_rate": 9.528284434756715e-06, "loss": 1.4213, "step": 3290 }, { "epoch": 0.53, "grad_norm": 10.747798919677734, "learning_rate": 9.523076041956203e-06, "loss": 1.4854, "step": 3291 }, { "epoch": 0.53, "grad_norm": 9.56531810760498, "learning_rate": 9.517867778824278e-06, "loss": 1.5908, "step": 3292 }, { "epoch": 0.53, "grad_norm": 7.936947822570801, "learning_rate": 9.512659646776983e-06, "loss": 1.7111, "step": 3293 }, { "epoch": 0.53, "grad_norm": 7.98375940322876, "learning_rate": 9.507451647230341e-06, "loss": 1.4259, "step": 3294 }, { "epoch": 0.53, "grad_norm": 9.104769706726074, "learning_rate": 9.502243781600324e-06, "loss": 1.6083, "step": 3295 }, { "epoch": 0.53, "grad_norm": 11.112497329711914, "learning_rate": 9.49703605130288e-06, "loss": 1.3295, "step": 3296 }, { "epoch": 0.53, "grad_norm": 7.328771591186523, "learning_rate": 9.491828457753907e-06, "loss": 1.5588, "step": 3297 }, { "epoch": 0.53, "grad_norm": 8.999688148498535, "learning_rate": 9.48662100236928e-06, "loss": 1.7205, "step": 3298 }, { "epoch": 0.53, "grad_norm": 6.906674385070801, "learning_rate": 9.481413686564825e-06, "loss": 1.3789, "step": 3299 }, { "epoch": 0.53, "grad_norm": 7.713040828704834, "learning_rate": 9.476206511756331e-06, "loss": 1.3999, "step": 3300 }, { "epoch": 0.53, "grad_norm": 8.258370399475098, "learning_rate": 9.470999479359559e-06, "loss": 1.6131, "step": 3301 }, { "epoch": 0.53, "grad_norm": 7.3598761558532715, "learning_rate": 9.465792590790219e-06, "loss": 1.3583, "step": 3302 }, { "epoch": 0.53, "grad_norm": 8.697495460510254, "learning_rate": 9.460585847463992e-06, "loss": 1.5398, "step": 3303 }, { "epoch": 0.53, "grad_norm": 9.594842910766602, "learning_rate": 9.45537925079651e-06, "loss": 1.6208, "step": 3304 }, { "epoch": 0.53, "grad_norm": 8.352524757385254, "learning_rate": 9.450172802203371e-06, "loss": 1.3764, "step": 3305 }, { "epoch": 0.53, "grad_norm": 8.508323669433594, "learning_rate": 9.444966503100134e-06, "loss": 1.4912, "step": 3306 }, { "epoch": 0.53, "grad_norm": 6.90737247467041, "learning_rate": 9.43976035490231e-06, "loss": 1.5296, "step": 3307 }, { "epoch": 0.53, "grad_norm": 5.9465179443359375, "learning_rate": 9.434554359025383e-06, "loss": 1.2637, "step": 3308 }, { "epoch": 0.53, "grad_norm": 7.8563008308410645, "learning_rate": 9.429348516884776e-06, "loss": 1.5093, "step": 3309 }, { "epoch": 0.53, "grad_norm": 6.535040378570557, "learning_rate": 9.424142829895883e-06, "loss": 1.4485, "step": 3310 }, { "epoch": 0.53, "grad_norm": 6.5127644538879395, "learning_rate": 9.41893729947406e-06, "loss": 1.431, "step": 3311 }, { "epoch": 0.53, "grad_norm": 7.469045162200928, "learning_rate": 9.413731927034607e-06, "loss": 1.4932, "step": 3312 }, { "epoch": 0.53, "grad_norm": 7.513950824737549, "learning_rate": 9.408526713992792e-06, "loss": 1.6215, "step": 3313 }, { "epoch": 0.53, "grad_norm": 7.956136226654053, "learning_rate": 9.403321661763834e-06, "loss": 1.4465, "step": 3314 }, { "epoch": 0.53, "grad_norm": 11.456660270690918, "learning_rate": 9.398116771762905e-06, "loss": 1.7467, "step": 3315 }, { "epoch": 0.53, "grad_norm": 7.551541328430176, "learning_rate": 9.392912045405148e-06, "loss": 1.2558, "step": 3316 }, { "epoch": 0.53, "grad_norm": 6.923919200897217, "learning_rate": 9.387707484105642e-06, "loss": 1.5954, "step": 3317 }, { "epoch": 0.53, "grad_norm": 8.604029655456543, "learning_rate": 9.382503089279436e-06, "loss": 1.6692, "step": 3318 }, { "epoch": 0.53, "grad_norm": 7.862287998199463, "learning_rate": 9.37729886234153e-06, "loss": 1.5766, "step": 3319 }, { "epoch": 0.53, "grad_norm": 8.436521530151367, "learning_rate": 9.372094804706867e-06, "loss": 1.4983, "step": 3320 }, { "epoch": 0.53, "grad_norm": 7.53306245803833, "learning_rate": 9.366890917790364e-06, "loss": 1.5154, "step": 3321 }, { "epoch": 0.53, "grad_norm": 7.707698822021484, "learning_rate": 9.361687203006872e-06, "loss": 1.6187, "step": 3322 }, { "epoch": 0.53, "grad_norm": 7.572998046875, "learning_rate": 9.35648366177121e-06, "loss": 1.4675, "step": 3323 }, { "epoch": 0.54, "grad_norm": 7.8621439933776855, "learning_rate": 9.351280295498147e-06, "loss": 1.3821, "step": 3324 }, { "epoch": 0.54, "grad_norm": 8.215958595275879, "learning_rate": 9.346077105602392e-06, "loss": 1.7122, "step": 3325 }, { "epoch": 0.54, "grad_norm": 7.695422649383545, "learning_rate": 9.340874093498623e-06, "loss": 1.5581, "step": 3326 }, { "epoch": 0.54, "grad_norm": 7.451794624328613, "learning_rate": 9.335671260601458e-06, "loss": 1.4078, "step": 3327 }, { "epoch": 0.54, "grad_norm": 6.783291816711426, "learning_rate": 9.330468608325475e-06, "loss": 1.5403, "step": 3328 }, { "epoch": 0.54, "grad_norm": 7.930881500244141, "learning_rate": 9.325266138085197e-06, "loss": 1.3291, "step": 3329 }, { "epoch": 0.54, "grad_norm": 8.181793212890625, "learning_rate": 9.320063851295093e-06, "loss": 1.5968, "step": 3330 }, { "epoch": 0.54, "grad_norm": 6.808318138122559, "learning_rate": 9.314861749369597e-06, "loss": 1.5353, "step": 3331 }, { "epoch": 0.54, "grad_norm": 6.964217662811279, "learning_rate": 9.309659833723076e-06, "loss": 1.5149, "step": 3332 }, { "epoch": 0.54, "grad_norm": 7.383213043212891, "learning_rate": 9.304458105769858e-06, "loss": 1.5597, "step": 3333 }, { "epoch": 0.54, "grad_norm": 8.099119186401367, "learning_rate": 9.299256566924216e-06, "loss": 1.4682, "step": 3334 }, { "epoch": 0.54, "grad_norm": 8.143658638000488, "learning_rate": 9.294055218600374e-06, "loss": 1.5815, "step": 3335 }, { "epoch": 0.54, "grad_norm": 9.686481475830078, "learning_rate": 9.288854062212499e-06, "loss": 1.3274, "step": 3336 }, { "epoch": 0.54, "grad_norm": 7.760528564453125, "learning_rate": 9.283653099174704e-06, "loss": 1.4034, "step": 3337 }, { "epoch": 0.54, "grad_norm": 8.562058448791504, "learning_rate": 9.278452330901062e-06, "loss": 1.3907, "step": 3338 }, { "epoch": 0.54, "grad_norm": 8.34398365020752, "learning_rate": 9.27325175880558e-06, "loss": 1.4377, "step": 3339 }, { "epoch": 0.54, "grad_norm": 8.244750022888184, "learning_rate": 9.26805138430222e-06, "loss": 1.3614, "step": 3340 }, { "epoch": 0.54, "grad_norm": 7.921983242034912, "learning_rate": 9.262851208804887e-06, "loss": 1.566, "step": 3341 }, { "epoch": 0.54, "grad_norm": 7.889716148376465, "learning_rate": 9.25765123372743e-06, "loss": 1.4438, "step": 3342 }, { "epoch": 0.54, "grad_norm": 7.801502704620361, "learning_rate": 9.252451460483645e-06, "loss": 1.3667, "step": 3343 }, { "epoch": 0.54, "grad_norm": 12.709646224975586, "learning_rate": 9.247251890487274e-06, "loss": 1.8421, "step": 3344 }, { "epoch": 0.54, "grad_norm": 9.963732719421387, "learning_rate": 9.242052525152007e-06, "loss": 1.4969, "step": 3345 }, { "epoch": 0.54, "grad_norm": 7.906045913696289, "learning_rate": 9.236853365891471e-06, "loss": 1.3029, "step": 3346 }, { "epoch": 0.54, "grad_norm": 10.032047271728516, "learning_rate": 9.23165441411924e-06, "loss": 1.5143, "step": 3347 }, { "epoch": 0.54, "grad_norm": 10.587409019470215, "learning_rate": 9.226455671248837e-06, "loss": 1.408, "step": 3348 }, { "epoch": 0.54, "grad_norm": 8.154187202453613, "learning_rate": 9.221257138693717e-06, "loss": 1.509, "step": 3349 }, { "epoch": 0.54, "grad_norm": 7.978038311004639, "learning_rate": 9.21605881786729e-06, "loss": 1.5153, "step": 3350 }, { "epoch": 0.54, "grad_norm": 8.594575881958008, "learning_rate": 9.2108607101829e-06, "loss": 1.6168, "step": 3351 }, { "epoch": 0.54, "grad_norm": 7.88539981842041, "learning_rate": 9.205662817053834e-06, "loss": 1.2273, "step": 3352 }, { "epoch": 0.54, "grad_norm": 8.54265308380127, "learning_rate": 9.200465139893326e-06, "loss": 1.6214, "step": 3353 }, { "epoch": 0.54, "grad_norm": 7.9464521408081055, "learning_rate": 9.195267680114544e-06, "loss": 1.3377, "step": 3354 }, { "epoch": 0.54, "grad_norm": 7.45587682723999, "learning_rate": 9.190070439130608e-06, "loss": 1.4103, "step": 3355 }, { "epoch": 0.54, "grad_norm": 7.203526020050049, "learning_rate": 9.18487341835456e-06, "loss": 1.2846, "step": 3356 }, { "epoch": 0.54, "grad_norm": 7.961169242858887, "learning_rate": 9.1796766191994e-06, "loss": 1.6131, "step": 3357 }, { "epoch": 0.54, "grad_norm": 7.909389972686768, "learning_rate": 9.174480043078065e-06, "loss": 1.4951, "step": 3358 }, { "epoch": 0.54, "grad_norm": 8.28486442565918, "learning_rate": 9.169283691403417e-06, "loss": 1.4512, "step": 3359 }, { "epoch": 0.54, "grad_norm": 7.720513343811035, "learning_rate": 9.164087565588277e-06, "loss": 1.392, "step": 3360 }, { "epoch": 0.54, "grad_norm": 6.819819450378418, "learning_rate": 9.158891667045388e-06, "loss": 1.4707, "step": 3361 }, { "epoch": 0.54, "grad_norm": 7.958302021026611, "learning_rate": 9.15369599718744e-06, "loss": 1.6429, "step": 3362 }, { "epoch": 0.54, "grad_norm": 7.512327194213867, "learning_rate": 9.148500557427062e-06, "loss": 1.4853, "step": 3363 }, { "epoch": 0.54, "grad_norm": 7.332778453826904, "learning_rate": 9.143305349176813e-06, "loss": 1.4247, "step": 3364 }, { "epoch": 0.54, "grad_norm": 7.474925994873047, "learning_rate": 9.1381103738492e-06, "loss": 1.3961, "step": 3365 }, { "epoch": 0.54, "grad_norm": 7.277134895324707, "learning_rate": 9.132915632856651e-06, "loss": 1.1538, "step": 3366 }, { "epoch": 0.54, "grad_norm": 8.614189147949219, "learning_rate": 9.127721127611544e-06, "loss": 1.3355, "step": 3367 }, { "epoch": 0.54, "grad_norm": 9.037017822265625, "learning_rate": 9.122526859526193e-06, "loss": 1.6165, "step": 3368 }, { "epoch": 0.54, "grad_norm": 10.007755279541016, "learning_rate": 9.117332830012832e-06, "loss": 1.5135, "step": 3369 }, { "epoch": 0.54, "grad_norm": 8.2009916305542, "learning_rate": 9.112139040483652e-06, "loss": 1.4756, "step": 3370 }, { "epoch": 0.54, "grad_norm": 12.683374404907227, "learning_rate": 9.106945492350761e-06, "loss": 1.7168, "step": 3371 }, { "epoch": 0.54, "grad_norm": 8.472737312316895, "learning_rate": 9.101752187026212e-06, "loss": 1.4302, "step": 3372 }, { "epoch": 0.54, "grad_norm": 7.933819770812988, "learning_rate": 9.096559125921988e-06, "loss": 1.3406, "step": 3373 }, { "epoch": 0.54, "grad_norm": 9.560172080993652, "learning_rate": 9.091366310449999e-06, "loss": 1.5334, "step": 3374 }, { "epoch": 0.54, "grad_norm": 9.472702980041504, "learning_rate": 9.086173742022101e-06, "loss": 1.3707, "step": 3375 }, { "epoch": 0.54, "grad_norm": 8.748501777648926, "learning_rate": 9.080981422050073e-06, "loss": 1.4438, "step": 3376 }, { "epoch": 0.54, "grad_norm": 7.272838592529297, "learning_rate": 9.075789351945639e-06, "loss": 1.3563, "step": 3377 }, { "epoch": 0.54, "grad_norm": 8.597646713256836, "learning_rate": 9.070597533120437e-06, "loss": 1.3766, "step": 3378 }, { "epoch": 0.54, "grad_norm": 7.479740619659424, "learning_rate": 9.065405966986047e-06, "loss": 1.6186, "step": 3379 }, { "epoch": 0.54, "grad_norm": 7.521187782287598, "learning_rate": 9.060214654953981e-06, "loss": 1.3379, "step": 3380 }, { "epoch": 0.54, "grad_norm": 10.640103340148926, "learning_rate": 9.055023598435679e-06, "loss": 1.5899, "step": 3381 }, { "epoch": 0.54, "grad_norm": 7.835803985595703, "learning_rate": 9.049832798842519e-06, "loss": 1.3011, "step": 3382 }, { "epoch": 0.54, "grad_norm": 7.081650257110596, "learning_rate": 9.044642257585795e-06, "loss": 1.4304, "step": 3383 }, { "epoch": 0.54, "grad_norm": 8.615711212158203, "learning_rate": 9.03945197607674e-06, "loss": 1.5122, "step": 3384 }, { "epoch": 0.54, "grad_norm": 8.092183113098145, "learning_rate": 9.034261955726515e-06, "loss": 1.3116, "step": 3385 }, { "epoch": 0.55, "grad_norm": 9.982396125793457, "learning_rate": 9.02907219794621e-06, "loss": 1.496, "step": 3386 }, { "epoch": 0.55, "grad_norm": 10.72201919555664, "learning_rate": 9.023882704146848e-06, "loss": 1.5917, "step": 3387 }, { "epoch": 0.55, "grad_norm": 8.807870864868164, "learning_rate": 9.018693475739373e-06, "loss": 1.6269, "step": 3388 }, { "epoch": 0.55, "grad_norm": 7.002970218658447, "learning_rate": 9.013504514134656e-06, "loss": 1.3753, "step": 3389 }, { "epoch": 0.55, "grad_norm": 7.2870049476623535, "learning_rate": 9.008315820743504e-06, "loss": 1.5986, "step": 3390 }, { "epoch": 0.55, "grad_norm": 8.215996742248535, "learning_rate": 9.003127396976638e-06, "loss": 1.757, "step": 3391 }, { "epoch": 0.55, "grad_norm": 6.621988773345947, "learning_rate": 8.997939244244728e-06, "loss": 1.7553, "step": 3392 }, { "epoch": 0.55, "grad_norm": 6.969228744506836, "learning_rate": 8.992751363958345e-06, "loss": 1.513, "step": 3393 }, { "epoch": 0.55, "grad_norm": 7.273033618927002, "learning_rate": 8.987563757528e-06, "loss": 1.5376, "step": 3394 }, { "epoch": 0.55, "grad_norm": 7.693988800048828, "learning_rate": 8.982376426364127e-06, "loss": 1.3655, "step": 3395 }, { "epoch": 0.55, "grad_norm": 7.301080703735352, "learning_rate": 8.977189371877082e-06, "loss": 1.3626, "step": 3396 }, { "epoch": 0.55, "grad_norm": 7.888976097106934, "learning_rate": 8.972002595477156e-06, "loss": 1.6183, "step": 3397 }, { "epoch": 0.55, "grad_norm": 8.343218803405762, "learning_rate": 8.96681609857455e-06, "loss": 1.7455, "step": 3398 }, { "epoch": 0.55, "grad_norm": 7.18881893157959, "learning_rate": 8.961629882579396e-06, "loss": 1.4503, "step": 3399 }, { "epoch": 0.55, "grad_norm": 7.575653076171875, "learning_rate": 8.956443948901755e-06, "loss": 1.4526, "step": 3400 }, { "epoch": 0.55, "grad_norm": 7.38035774230957, "learning_rate": 8.951258298951599e-06, "loss": 1.4566, "step": 3401 }, { "epoch": 0.55, "grad_norm": 7.3210859298706055, "learning_rate": 8.946072934138836e-06, "loss": 1.6171, "step": 3402 }, { "epoch": 0.55, "grad_norm": 7.593682765960693, "learning_rate": 8.940887855873287e-06, "loss": 1.4283, "step": 3403 }, { "epoch": 0.55, "grad_norm": 7.526524066925049, "learning_rate": 8.935703065564695e-06, "loss": 1.6011, "step": 3404 }, { "epoch": 0.55, "grad_norm": 9.1856050491333, "learning_rate": 8.930518564622738e-06, "loss": 1.533, "step": 3405 }, { "epoch": 0.55, "grad_norm": 5.893603324890137, "learning_rate": 8.925334354456992e-06, "loss": 1.4105, "step": 3406 }, { "epoch": 0.55, "grad_norm": 8.337667465209961, "learning_rate": 8.92015043647698e-06, "loss": 1.5233, "step": 3407 }, { "epoch": 0.55, "grad_norm": 9.031383514404297, "learning_rate": 8.914966812092122e-06, "loss": 1.455, "step": 3408 }, { "epoch": 0.55, "grad_norm": 7.655910968780518, "learning_rate": 8.909783482711777e-06, "loss": 1.4268, "step": 3409 }, { "epoch": 0.55, "grad_norm": 6.8177666664123535, "learning_rate": 8.904600449745212e-06, "loss": 1.4382, "step": 3410 }, { "epoch": 0.55, "grad_norm": 6.727724552154541, "learning_rate": 8.899417714601617e-06, "loss": 1.5574, "step": 3411 }, { "epoch": 0.55, "grad_norm": 7.96099853515625, "learning_rate": 8.894235278690104e-06, "loss": 1.6111, "step": 3412 }, { "epoch": 0.55, "grad_norm": 7.877175807952881, "learning_rate": 8.889053143419694e-06, "loss": 1.531, "step": 3413 }, { "epoch": 0.55, "grad_norm": 8.536805152893066, "learning_rate": 8.883871310199343e-06, "loss": 1.4705, "step": 3414 }, { "epoch": 0.55, "grad_norm": 7.268801212310791, "learning_rate": 8.87868978043791e-06, "loss": 1.3819, "step": 3415 }, { "epoch": 0.55, "grad_norm": 7.92653751373291, "learning_rate": 8.873508555544171e-06, "loss": 1.4156, "step": 3416 }, { "epoch": 0.55, "grad_norm": 8.663615226745605, "learning_rate": 8.868327636926836e-06, "loss": 1.6906, "step": 3417 }, { "epoch": 0.55, "grad_norm": 7.495673179626465, "learning_rate": 8.863147025994508e-06, "loss": 1.474, "step": 3418 }, { "epoch": 0.55, "grad_norm": 7.795389652252197, "learning_rate": 8.857966724155728e-06, "loss": 1.5486, "step": 3419 }, { "epoch": 0.55, "grad_norm": 8.26573657989502, "learning_rate": 8.852786732818943e-06, "loss": 1.4154, "step": 3420 }, { "epoch": 0.55, "grad_norm": 7.595717430114746, "learning_rate": 8.847607053392509e-06, "loss": 1.5704, "step": 3421 }, { "epoch": 0.55, "grad_norm": 8.725198745727539, "learning_rate": 8.842427687284711e-06, "loss": 1.3977, "step": 3422 }, { "epoch": 0.55, "grad_norm": 8.419879913330078, "learning_rate": 8.83724863590374e-06, "loss": 1.5325, "step": 3423 }, { "epoch": 0.55, "grad_norm": 6.87568998336792, "learning_rate": 8.832069900657705e-06, "loss": 1.3924, "step": 3424 }, { "epoch": 0.55, "grad_norm": 7.1130690574646, "learning_rate": 8.826891482954628e-06, "loss": 1.5345, "step": 3425 }, { "epoch": 0.55, "grad_norm": 8.349703788757324, "learning_rate": 8.821713384202438e-06, "loss": 1.6227, "step": 3426 }, { "epoch": 0.55, "grad_norm": 7.051556587219238, "learning_rate": 8.816535605808992e-06, "loss": 1.4672, "step": 3427 }, { "epoch": 0.55, "grad_norm": 6.873019218444824, "learning_rate": 8.811358149182045e-06, "loss": 1.3837, "step": 3428 }, { "epoch": 0.55, "grad_norm": 8.061370849609375, "learning_rate": 8.806181015729278e-06, "loss": 1.3644, "step": 3429 }, { "epoch": 0.55, "grad_norm": 6.9856462478637695, "learning_rate": 8.801004206858275e-06, "loss": 1.498, "step": 3430 }, { "epoch": 0.55, "grad_norm": 8.342188835144043, "learning_rate": 8.795827723976526e-06, "loss": 1.5117, "step": 3431 }, { "epoch": 0.55, "grad_norm": 7.899488925933838, "learning_rate": 8.79065156849145e-06, "loss": 1.5727, "step": 3432 }, { "epoch": 0.55, "grad_norm": 7.922801971435547, "learning_rate": 8.785475741810361e-06, "loss": 1.2411, "step": 3433 }, { "epoch": 0.55, "grad_norm": 8.194945335388184, "learning_rate": 8.780300245340497e-06, "loss": 1.4912, "step": 3434 }, { "epoch": 0.55, "grad_norm": 8.406078338623047, "learning_rate": 8.775125080488996e-06, "loss": 1.3225, "step": 3435 }, { "epoch": 0.55, "grad_norm": 7.155364990234375, "learning_rate": 8.769950248662902e-06, "loss": 1.3989, "step": 3436 }, { "epoch": 0.55, "grad_norm": 7.692113399505615, "learning_rate": 8.764775751269184e-06, "loss": 1.4847, "step": 3437 }, { "epoch": 0.55, "grad_norm": 7.13429069519043, "learning_rate": 8.759601589714705e-06, "loss": 1.4831, "step": 3438 }, { "epoch": 0.55, "grad_norm": 7.5369873046875, "learning_rate": 8.754427765406252e-06, "loss": 1.5148, "step": 3439 }, { "epoch": 0.55, "grad_norm": 9.224095344543457, "learning_rate": 8.749254279750505e-06, "loss": 1.8208, "step": 3440 }, { "epoch": 0.55, "grad_norm": 7.261047840118408, "learning_rate": 8.744081134154057e-06, "loss": 1.5313, "step": 3441 }, { "epoch": 0.55, "grad_norm": 8.851920127868652, "learning_rate": 8.738908330023415e-06, "loss": 1.3584, "step": 3442 }, { "epoch": 0.55, "grad_norm": 7.582375526428223, "learning_rate": 8.733735868764982e-06, "loss": 1.6307, "step": 3443 }, { "epoch": 0.55, "grad_norm": 7.782669544219971, "learning_rate": 8.728563751785084e-06, "loss": 1.2256, "step": 3444 }, { "epoch": 0.55, "grad_norm": 8.155216217041016, "learning_rate": 8.723391980489931e-06, "loss": 1.39, "step": 3445 }, { "epoch": 0.55, "grad_norm": 9.23241138458252, "learning_rate": 8.71822055628566e-06, "loss": 1.5066, "step": 3446 }, { "epoch": 0.55, "grad_norm": 8.558978080749512, "learning_rate": 8.713049480578302e-06, "loss": 1.6577, "step": 3447 }, { "epoch": 0.56, "grad_norm": 7.456271648406982, "learning_rate": 8.707878754773793e-06, "loss": 1.4259, "step": 3448 }, { "epoch": 0.56, "grad_norm": 11.829216003417969, "learning_rate": 8.702708380277984e-06, "loss": 1.4925, "step": 3449 }, { "epoch": 0.56, "grad_norm": 8.481657981872559, "learning_rate": 8.697538358496617e-06, "loss": 1.3524, "step": 3450 }, { "epoch": 0.56, "grad_norm": 8.563260078430176, "learning_rate": 8.69236869083535e-06, "loss": 1.5448, "step": 3451 }, { "epoch": 0.56, "grad_norm": 8.39290714263916, "learning_rate": 8.687199378699736e-06, "loss": 1.4022, "step": 3452 }, { "epoch": 0.56, "grad_norm": 6.964887619018555, "learning_rate": 8.682030423495233e-06, "loss": 1.225, "step": 3453 }, { "epoch": 0.56, "grad_norm": 8.805925369262695, "learning_rate": 8.67686182662721e-06, "loss": 1.7645, "step": 3454 }, { "epoch": 0.56, "grad_norm": 8.316703796386719, "learning_rate": 8.671693589500922e-06, "loss": 1.3832, "step": 3455 }, { "epoch": 0.56, "grad_norm": 9.80985164642334, "learning_rate": 8.666525713521548e-06, "loss": 1.2965, "step": 3456 }, { "epoch": 0.56, "grad_norm": 9.072491645812988, "learning_rate": 8.661358200094153e-06, "loss": 1.845, "step": 3457 }, { "epoch": 0.56, "grad_norm": 8.123779296875, "learning_rate": 8.6561910506237e-06, "loss": 1.4682, "step": 3458 }, { "epoch": 0.56, "grad_norm": 7.9592461585998535, "learning_rate": 8.651024266515073e-06, "loss": 1.4848, "step": 3459 }, { "epoch": 0.56, "grad_norm": 8.40068531036377, "learning_rate": 8.645857849173034e-06, "loss": 1.5822, "step": 3460 }, { "epoch": 0.56, "grad_norm": 7.0547308921813965, "learning_rate": 8.640691800002262e-06, "loss": 1.4326, "step": 3461 }, { "epoch": 0.56, "grad_norm": 7.5240092277526855, "learning_rate": 8.63552612040733e-06, "loss": 1.3456, "step": 3462 }, { "epoch": 0.56, "grad_norm": 7.492598056793213, "learning_rate": 8.630360811792702e-06, "loss": 1.4433, "step": 3463 }, { "epoch": 0.56, "grad_norm": 7.948658466339111, "learning_rate": 8.625195875562761e-06, "loss": 1.5159, "step": 3464 }, { "epoch": 0.56, "grad_norm": 8.76047420501709, "learning_rate": 8.620031313121767e-06, "loss": 1.4946, "step": 3465 }, { "epoch": 0.56, "grad_norm": 8.041772842407227, "learning_rate": 8.614867125873891e-06, "loss": 1.605, "step": 3466 }, { "epoch": 0.56, "grad_norm": 7.962422847747803, "learning_rate": 8.609703315223206e-06, "loss": 1.3911, "step": 3467 }, { "epoch": 0.56, "grad_norm": 8.067803382873535, "learning_rate": 8.604539882573664e-06, "loss": 1.4609, "step": 3468 }, { "epoch": 0.56, "grad_norm": 7.301872253417969, "learning_rate": 8.599376829329137e-06, "loss": 1.4457, "step": 3469 }, { "epoch": 0.56, "grad_norm": 8.130311012268066, "learning_rate": 8.594214156893373e-06, "loss": 1.4933, "step": 3470 }, { "epoch": 0.56, "grad_norm": 7.996828079223633, "learning_rate": 8.589051866670035e-06, "loss": 1.6525, "step": 3471 }, { "epoch": 0.56, "grad_norm": 7.3521270751953125, "learning_rate": 8.583889960062673e-06, "loss": 1.5406, "step": 3472 }, { "epoch": 0.56, "grad_norm": 9.632269859313965, "learning_rate": 8.578728438474726e-06, "loss": 1.6648, "step": 3473 }, { "epoch": 0.56, "grad_norm": 8.052114486694336, "learning_rate": 8.573567303309542e-06, "loss": 1.5089, "step": 3474 }, { "epoch": 0.56, "grad_norm": 8.002729415893555, "learning_rate": 8.568406555970355e-06, "loss": 1.6238, "step": 3475 }, { "epoch": 0.56, "grad_norm": 11.383553504943848, "learning_rate": 8.563246197860297e-06, "loss": 1.4395, "step": 3476 }, { "epoch": 0.56, "grad_norm": 7.244509220123291, "learning_rate": 8.558086230382396e-06, "loss": 1.3327, "step": 3477 }, { "epoch": 0.56, "grad_norm": 9.337241172790527, "learning_rate": 8.552926654939561e-06, "loss": 1.381, "step": 3478 }, { "epoch": 0.56, "grad_norm": 10.576663970947266, "learning_rate": 8.547767472934616e-06, "loss": 1.4396, "step": 3479 }, { "epoch": 0.56, "grad_norm": 8.153815269470215, "learning_rate": 8.542608685770256e-06, "loss": 1.2883, "step": 3480 }, { "epoch": 0.56, "grad_norm": 9.346476554870605, "learning_rate": 8.537450294849093e-06, "loss": 1.5778, "step": 3481 }, { "epoch": 0.56, "grad_norm": 9.323090553283691, "learning_rate": 8.532292301573603e-06, "loss": 1.732, "step": 3482 }, { "epoch": 0.56, "grad_norm": 8.55916690826416, "learning_rate": 8.527134707346179e-06, "loss": 1.6262, "step": 3483 }, { "epoch": 0.56, "grad_norm": 8.538518905639648, "learning_rate": 8.521977513569088e-06, "loss": 1.4607, "step": 3484 }, { "epoch": 0.56, "grad_norm": 8.418392181396484, "learning_rate": 8.516820721644494e-06, "loss": 1.5755, "step": 3485 }, { "epoch": 0.56, "grad_norm": 6.827256679534912, "learning_rate": 8.51166433297446e-06, "loss": 1.4485, "step": 3486 }, { "epoch": 0.56, "grad_norm": 10.750079154968262, "learning_rate": 8.506508348960924e-06, "loss": 1.4916, "step": 3487 }, { "epoch": 0.56, "grad_norm": 7.40295934677124, "learning_rate": 8.50135277100573e-06, "loss": 1.5004, "step": 3488 }, { "epoch": 0.56, "grad_norm": 7.473851203918457, "learning_rate": 8.496197600510596e-06, "loss": 1.641, "step": 3489 }, { "epoch": 0.56, "grad_norm": 7.362739562988281, "learning_rate": 8.491042838877138e-06, "loss": 1.691, "step": 3490 }, { "epoch": 0.56, "grad_norm": 7.229111671447754, "learning_rate": 8.485888487506865e-06, "loss": 1.632, "step": 3491 }, { "epoch": 0.56, "grad_norm": 6.9163432121276855, "learning_rate": 8.480734547801162e-06, "loss": 1.6718, "step": 3492 }, { "epoch": 0.56, "grad_norm": 7.309288501739502, "learning_rate": 8.475581021161318e-06, "loss": 1.2953, "step": 3493 }, { "epoch": 0.56, "grad_norm": 7.134573459625244, "learning_rate": 8.470427908988493e-06, "loss": 1.4538, "step": 3494 }, { "epoch": 0.56, "grad_norm": 6.879514694213867, "learning_rate": 8.46527521268374e-06, "loss": 1.456, "step": 3495 }, { "epoch": 0.56, "grad_norm": 7.271092891693115, "learning_rate": 8.460122933648013e-06, "loss": 1.2805, "step": 3496 }, { "epoch": 0.56, "grad_norm": 7.94778299331665, "learning_rate": 8.454971073282128e-06, "loss": 1.3286, "step": 3497 }, { "epoch": 0.56, "grad_norm": 7.9106764793396, "learning_rate": 8.449819632986809e-06, "loss": 1.5224, "step": 3498 }, { "epoch": 0.56, "grad_norm": 6.631632328033447, "learning_rate": 8.444668614162651e-06, "loss": 1.2857, "step": 3499 }, { "epoch": 0.56, "grad_norm": 8.605866432189941, "learning_rate": 8.439518018210138e-06, "loss": 1.7704, "step": 3500 }, { "epoch": 0.56, "grad_norm": 9.387898445129395, "learning_rate": 8.434367846529652e-06, "loss": 1.5966, "step": 3501 }, { "epoch": 0.56, "grad_norm": 7.188185214996338, "learning_rate": 8.429218100521435e-06, "loss": 1.479, "step": 3502 }, { "epoch": 0.56, "grad_norm": 10.7774658203125, "learning_rate": 8.424068781585636e-06, "loss": 1.137, "step": 3503 }, { "epoch": 0.56, "grad_norm": 8.045463562011719, "learning_rate": 8.41891989112228e-06, "loss": 1.4356, "step": 3504 }, { "epoch": 0.56, "grad_norm": 9.669730186462402, "learning_rate": 8.413771430531263e-06, "loss": 1.6749, "step": 3505 }, { "epoch": 0.56, "grad_norm": 8.696136474609375, "learning_rate": 8.40862340121239e-06, "loss": 1.4503, "step": 3506 }, { "epoch": 0.56, "grad_norm": 7.2378740310668945, "learning_rate": 8.403475804565319e-06, "loss": 1.1731, "step": 3507 }, { "epoch": 0.56, "grad_norm": 8.077529907226562, "learning_rate": 8.398328641989619e-06, "loss": 1.2796, "step": 3508 }, { "epoch": 0.56, "grad_norm": 8.449429512023926, "learning_rate": 8.393181914884726e-06, "loss": 1.5116, "step": 3509 }, { "epoch": 0.57, "grad_norm": 8.561786651611328, "learning_rate": 8.388035624649947e-06, "loss": 1.4621, "step": 3510 }, { "epoch": 0.57, "grad_norm": 8.32929801940918, "learning_rate": 8.382889772684497e-06, "loss": 1.554, "step": 3511 }, { "epoch": 0.57, "grad_norm": 7.145478248596191, "learning_rate": 8.377744360387447e-06, "loss": 1.4013, "step": 3512 }, { "epoch": 0.57, "grad_norm": 7.927107810974121, "learning_rate": 8.372599389157764e-06, "loss": 1.3487, "step": 3513 }, { "epoch": 0.57, "grad_norm": 6.690553188323975, "learning_rate": 8.36745486039429e-06, "loss": 1.3548, "step": 3514 }, { "epoch": 0.57, "grad_norm": 7.960625648498535, "learning_rate": 8.362310775495741e-06, "loss": 1.5634, "step": 3515 }, { "epoch": 0.57, "grad_norm": 7.891608238220215, "learning_rate": 8.357167135860727e-06, "loss": 1.6702, "step": 3516 }, { "epoch": 0.57, "grad_norm": 7.751349925994873, "learning_rate": 8.352023942887716e-06, "loss": 1.723, "step": 3517 }, { "epoch": 0.57, "grad_norm": 7.31533145904541, "learning_rate": 8.346881197975077e-06, "loss": 1.5604, "step": 3518 }, { "epoch": 0.57, "grad_norm": 8.570663452148438, "learning_rate": 8.341738902521041e-06, "loss": 1.406, "step": 3519 }, { "epoch": 0.57, "grad_norm": 7.903316020965576, "learning_rate": 8.33659705792372e-06, "loss": 1.5629, "step": 3520 }, { "epoch": 0.57, "grad_norm": 7.590131759643555, "learning_rate": 8.331455665581114e-06, "loss": 1.3335, "step": 3521 }, { "epoch": 0.57, "grad_norm": 8.069557189941406, "learning_rate": 8.326314726891083e-06, "loss": 1.6717, "step": 3522 }, { "epoch": 0.57, "grad_norm": 9.184011459350586, "learning_rate": 8.321174243251376e-06, "loss": 1.5523, "step": 3523 }, { "epoch": 0.57, "grad_norm": 7.13422155380249, "learning_rate": 8.316034216059613e-06, "loss": 1.4842, "step": 3524 }, { "epoch": 0.57, "grad_norm": 7.3754401206970215, "learning_rate": 8.310894646713298e-06, "loss": 1.4048, "step": 3525 }, { "epoch": 0.57, "grad_norm": 8.663140296936035, "learning_rate": 8.305755536609797e-06, "loss": 1.5008, "step": 3526 }, { "epoch": 0.57, "grad_norm": 7.610958099365234, "learning_rate": 8.30061688714636e-06, "loss": 1.401, "step": 3527 }, { "epoch": 0.57, "grad_norm": 8.9475679397583, "learning_rate": 8.295478699720108e-06, "loss": 1.4988, "step": 3528 }, { "epoch": 0.57, "grad_norm": 7.797390937805176, "learning_rate": 8.290340975728042e-06, "loss": 1.4164, "step": 3529 }, { "epoch": 0.57, "grad_norm": 7.3067307472229, "learning_rate": 8.285203716567034e-06, "loss": 1.216, "step": 3530 }, { "epoch": 0.57, "grad_norm": 7.453532695770264, "learning_rate": 8.280066923633826e-06, "loss": 1.3399, "step": 3531 }, { "epoch": 0.57, "grad_norm": 8.3779296875, "learning_rate": 8.274930598325032e-06, "loss": 1.5141, "step": 3532 }, { "epoch": 0.57, "grad_norm": 10.968270301818848, "learning_rate": 8.269794742037156e-06, "loss": 1.703, "step": 3533 }, { "epoch": 0.57, "grad_norm": 7.952436447143555, "learning_rate": 8.264659356166548e-06, "loss": 1.4265, "step": 3534 }, { "epoch": 0.57, "grad_norm": 7.138411045074463, "learning_rate": 8.259524442109452e-06, "loss": 1.5499, "step": 3535 }, { "epoch": 0.57, "grad_norm": 6.843843460083008, "learning_rate": 8.254390001261973e-06, "loss": 1.3864, "step": 3536 }, { "epoch": 0.57, "grad_norm": 8.269532203674316, "learning_rate": 8.249256035020086e-06, "loss": 1.5113, "step": 3537 }, { "epoch": 0.57, "grad_norm": 8.301777839660645, "learning_rate": 8.244122544779646e-06, "loss": 1.4753, "step": 3538 }, { "epoch": 0.57, "grad_norm": 7.918750762939453, "learning_rate": 8.238989531936369e-06, "loss": 1.4667, "step": 3539 }, { "epoch": 0.57, "grad_norm": 8.001060485839844, "learning_rate": 8.233856997885853e-06, "loss": 1.339, "step": 3540 }, { "epoch": 0.57, "grad_norm": 7.5221991539001465, "learning_rate": 8.22872494402355e-06, "loss": 1.3335, "step": 3541 }, { "epoch": 0.57, "grad_norm": 8.310895919799805, "learning_rate": 8.22359337174479e-06, "loss": 1.3686, "step": 3542 }, { "epoch": 0.57, "grad_norm": 8.33996295928955, "learning_rate": 8.218462282444777e-06, "loss": 1.332, "step": 3543 }, { "epoch": 0.57, "grad_norm": 6.993942737579346, "learning_rate": 8.213331677518572e-06, "loss": 1.5425, "step": 3544 }, { "epoch": 0.57, "grad_norm": 9.177495002746582, "learning_rate": 8.20820155836112e-06, "loss": 1.5367, "step": 3545 }, { "epoch": 0.57, "grad_norm": 8.622085571289062, "learning_rate": 8.203071926367217e-06, "loss": 1.4836, "step": 3546 }, { "epoch": 0.57, "grad_norm": 8.309316635131836, "learning_rate": 8.197942782931533e-06, "loss": 1.5381, "step": 3547 }, { "epoch": 0.57, "grad_norm": 8.438750267028809, "learning_rate": 8.192814129448615e-06, "loss": 1.3645, "step": 3548 }, { "epoch": 0.57, "grad_norm": 9.321581840515137, "learning_rate": 8.18768596731286e-06, "loss": 1.4714, "step": 3549 }, { "epoch": 0.57, "grad_norm": 8.720611572265625, "learning_rate": 8.182558297918545e-06, "loss": 1.4382, "step": 3550 }, { "epoch": 0.57, "grad_norm": 8.018508911132812, "learning_rate": 8.177431122659802e-06, "loss": 1.5915, "step": 3551 }, { "epoch": 0.57, "grad_norm": 7.431460380554199, "learning_rate": 8.172304442930637e-06, "loss": 1.1884, "step": 3552 }, { "epoch": 0.57, "grad_norm": 9.104870796203613, "learning_rate": 8.167178260124924e-06, "loss": 1.4679, "step": 3553 }, { "epoch": 0.57, "grad_norm": 9.412195205688477, "learning_rate": 8.162052575636387e-06, "loss": 1.6617, "step": 3554 }, { "epoch": 0.57, "grad_norm": 9.245584487915039, "learning_rate": 8.156927390858629e-06, "loss": 1.4265, "step": 3555 }, { "epoch": 0.57, "grad_norm": 7.884559154510498, "learning_rate": 8.151802707185115e-06, "loss": 1.538, "step": 3556 }, { "epoch": 0.57, "grad_norm": 8.760272979736328, "learning_rate": 8.146678526009164e-06, "loss": 1.1203, "step": 3557 }, { "epoch": 0.57, "grad_norm": 8.798954963684082, "learning_rate": 8.14155484872397e-06, "loss": 1.5508, "step": 3558 }, { "epoch": 0.57, "grad_norm": 6.663790225982666, "learning_rate": 8.136431676722582e-06, "loss": 1.3643, "step": 3559 }, { "epoch": 0.57, "grad_norm": 10.045855522155762, "learning_rate": 8.131309011397918e-06, "loss": 1.5058, "step": 3560 }, { "epoch": 0.57, "grad_norm": 8.670689582824707, "learning_rate": 8.126186854142752e-06, "loss": 1.4165, "step": 3561 }, { "epoch": 0.57, "grad_norm": 10.132168769836426, "learning_rate": 8.12106520634973e-06, "loss": 1.3829, "step": 3562 }, { "epoch": 0.57, "grad_norm": 8.382247924804688, "learning_rate": 8.115944069411345e-06, "loss": 1.7739, "step": 3563 }, { "epoch": 0.57, "grad_norm": 8.271489143371582, "learning_rate": 8.110823444719958e-06, "loss": 1.4765, "step": 3564 }, { "epoch": 0.57, "grad_norm": 7.576735019683838, "learning_rate": 8.105703333667797e-06, "loss": 1.4179, "step": 3565 }, { "epoch": 0.57, "grad_norm": 7.298073768615723, "learning_rate": 8.100583737646938e-06, "loss": 1.5096, "step": 3566 }, { "epoch": 0.57, "grad_norm": 7.719305515289307, "learning_rate": 8.095464658049333e-06, "loss": 1.3841, "step": 3567 }, { "epoch": 0.57, "grad_norm": 7.957104682922363, "learning_rate": 8.090346096266777e-06, "loss": 1.5262, "step": 3568 }, { "epoch": 0.57, "grad_norm": 9.510543823242188, "learning_rate": 8.08522805369093e-06, "loss": 1.7603, "step": 3569 }, { "epoch": 0.57, "grad_norm": 10.00693416595459, "learning_rate": 8.080110531713316e-06, "loss": 1.3906, "step": 3570 }, { "epoch": 0.57, "grad_norm": 7.457469463348389, "learning_rate": 8.074993531725311e-06, "loss": 1.4699, "step": 3571 }, { "epoch": 0.58, "grad_norm": 7.748441219329834, "learning_rate": 8.069877055118158e-06, "loss": 1.4527, "step": 3572 }, { "epoch": 0.58, "grad_norm": 8.689763069152832, "learning_rate": 8.064761103282948e-06, "loss": 1.5672, "step": 3573 }, { "epoch": 0.58, "grad_norm": 9.806768417358398, "learning_rate": 8.059645677610626e-06, "loss": 1.89, "step": 3574 }, { "epoch": 0.58, "grad_norm": 7.961200714111328, "learning_rate": 8.05453077949201e-06, "loss": 1.3961, "step": 3575 }, { "epoch": 0.58, "grad_norm": 6.425591468811035, "learning_rate": 8.049416410317759e-06, "loss": 1.3942, "step": 3576 }, { "epoch": 0.58, "grad_norm": 8.17427921295166, "learning_rate": 8.044302571478402e-06, "loss": 1.4354, "step": 3577 }, { "epoch": 0.58, "grad_norm": 8.012580871582031, "learning_rate": 8.03918926436431e-06, "loss": 1.5704, "step": 3578 }, { "epoch": 0.58, "grad_norm": 8.246516227722168, "learning_rate": 8.034076490365716e-06, "loss": 1.4366, "step": 3579 }, { "epoch": 0.58, "grad_norm": 7.473476409912109, "learning_rate": 8.028964250872711e-06, "loss": 1.4288, "step": 3580 }, { "epoch": 0.58, "grad_norm": 8.868585586547852, "learning_rate": 8.023852547275233e-06, "loss": 1.5812, "step": 3581 }, { "epoch": 0.58, "grad_norm": 5.843895435333252, "learning_rate": 8.018741380963086e-06, "loss": 1.5251, "step": 3582 }, { "epoch": 0.58, "grad_norm": 7.98234748840332, "learning_rate": 8.013630753325914e-06, "loss": 1.6446, "step": 3583 }, { "epoch": 0.58, "grad_norm": 8.1402006149292, "learning_rate": 8.00852066575322e-06, "loss": 1.2435, "step": 3584 }, { "epoch": 0.58, "grad_norm": 6.727962493896484, "learning_rate": 8.00341111963437e-06, "loss": 1.3466, "step": 3585 }, { "epoch": 0.58, "grad_norm": 7.940356731414795, "learning_rate": 7.998302116358566e-06, "loss": 1.3931, "step": 3586 }, { "epoch": 0.58, "grad_norm": 9.087540626525879, "learning_rate": 7.993193657314874e-06, "loss": 1.4544, "step": 3587 }, { "epoch": 0.58, "grad_norm": 8.085029602050781, "learning_rate": 7.98808574389221e-06, "loss": 1.5111, "step": 3588 }, { "epoch": 0.58, "grad_norm": 6.835176467895508, "learning_rate": 7.982978377479335e-06, "loss": 1.2004, "step": 3589 }, { "epoch": 0.58, "grad_norm": 9.551239013671875, "learning_rate": 7.977871559464873e-06, "loss": 1.5062, "step": 3590 }, { "epoch": 0.58, "grad_norm": 8.375282287597656, "learning_rate": 7.972765291237288e-06, "loss": 1.5583, "step": 3591 }, { "epoch": 0.58, "grad_norm": 7.810132026672363, "learning_rate": 7.967659574184901e-06, "loss": 1.301, "step": 3592 }, { "epoch": 0.58, "grad_norm": 10.095498085021973, "learning_rate": 7.962554409695882e-06, "loss": 1.4763, "step": 3593 }, { "epoch": 0.58, "grad_norm": 9.837014198303223, "learning_rate": 7.957449799158244e-06, "loss": 1.339, "step": 3594 }, { "epoch": 0.58, "grad_norm": 7.33799409866333, "learning_rate": 7.952345743959864e-06, "loss": 1.2366, "step": 3595 }, { "epoch": 0.58, "grad_norm": 8.548219680786133, "learning_rate": 7.947242245488451e-06, "loss": 1.3949, "step": 3596 }, { "epoch": 0.58, "grad_norm": 8.438170433044434, "learning_rate": 7.94213930513158e-06, "loss": 1.5206, "step": 3597 }, { "epoch": 0.58, "grad_norm": 8.263792037963867, "learning_rate": 7.937036924276655e-06, "loss": 1.292, "step": 3598 }, { "epoch": 0.58, "grad_norm": 9.954191207885742, "learning_rate": 7.931935104310945e-06, "loss": 1.4021, "step": 3599 }, { "epoch": 0.58, "grad_norm": 10.20656967163086, "learning_rate": 7.92683384662156e-06, "loss": 1.4903, "step": 3600 }, { "epoch": 0.58, "grad_norm": 9.000204086303711, "learning_rate": 7.92173315259545e-06, "loss": 1.5113, "step": 3601 }, { "epoch": 0.58, "grad_norm": 8.569355010986328, "learning_rate": 7.916633023619427e-06, "loss": 1.3735, "step": 3602 }, { "epoch": 0.58, "grad_norm": 8.22984504699707, "learning_rate": 7.911533461080132e-06, "loss": 1.3347, "step": 3603 }, { "epoch": 0.58, "grad_norm": 7.668810844421387, "learning_rate": 7.906434466364068e-06, "loss": 1.4635, "step": 3604 }, { "epoch": 0.58, "grad_norm": 9.02048110961914, "learning_rate": 7.901336040857576e-06, "loss": 1.5356, "step": 3605 }, { "epoch": 0.58, "grad_norm": 9.122057914733887, "learning_rate": 7.896238185946837e-06, "loss": 1.5958, "step": 3606 }, { "epoch": 0.58, "grad_norm": 8.725517272949219, "learning_rate": 7.891140903017885e-06, "loss": 1.2373, "step": 3607 }, { "epoch": 0.58, "grad_norm": 7.116713523864746, "learning_rate": 7.886044193456598e-06, "loss": 1.4294, "step": 3608 }, { "epoch": 0.58, "grad_norm": 12.527607917785645, "learning_rate": 7.880948058648696e-06, "loss": 1.5011, "step": 3609 }, { "epoch": 0.58, "grad_norm": 7.1247878074646, "learning_rate": 7.875852499979744e-06, "loss": 1.4201, "step": 3610 }, { "epoch": 0.58, "grad_norm": 9.110069274902344, "learning_rate": 7.870757518835143e-06, "loss": 1.4968, "step": 3611 }, { "epoch": 0.58, "grad_norm": 8.126364707946777, "learning_rate": 7.865663116600149e-06, "loss": 1.602, "step": 3612 }, { "epoch": 0.58, "grad_norm": 9.099882125854492, "learning_rate": 7.86056929465985e-06, "loss": 1.5052, "step": 3613 }, { "epoch": 0.58, "grad_norm": 8.363934516906738, "learning_rate": 7.855476054399187e-06, "loss": 1.6512, "step": 3614 }, { "epoch": 0.58, "grad_norm": 7.682722091674805, "learning_rate": 7.850383397202936e-06, "loss": 1.4784, "step": 3615 }, { "epoch": 0.58, "grad_norm": 8.926868438720703, "learning_rate": 7.845291324455707e-06, "loss": 1.4289, "step": 3616 }, { "epoch": 0.58, "grad_norm": 8.679990768432617, "learning_rate": 7.84019983754197e-06, "loss": 1.5369, "step": 3617 }, { "epoch": 0.58, "grad_norm": 8.436100959777832, "learning_rate": 7.835108937846018e-06, "loss": 1.2599, "step": 3618 }, { "epoch": 0.58, "grad_norm": 7.435399532318115, "learning_rate": 7.830018626752e-06, "loss": 1.5568, "step": 3619 }, { "epoch": 0.58, "grad_norm": 7.341434478759766, "learning_rate": 7.824928905643887e-06, "loss": 1.3899, "step": 3620 }, { "epoch": 0.58, "grad_norm": 7.509795665740967, "learning_rate": 7.819839775905501e-06, "loss": 1.3682, "step": 3621 }, { "epoch": 0.58, "grad_norm": 7.589110374450684, "learning_rate": 7.814751238920504e-06, "loss": 1.2251, "step": 3622 }, { "epoch": 0.58, "grad_norm": 9.088395118713379, "learning_rate": 7.809663296072392e-06, "loss": 1.3926, "step": 3623 }, { "epoch": 0.58, "grad_norm": 6.470757961273193, "learning_rate": 7.804575948744506e-06, "loss": 1.2799, "step": 3624 }, { "epoch": 0.58, "grad_norm": 10.005070686340332, "learning_rate": 7.799489198320019e-06, "loss": 1.4967, "step": 3625 }, { "epoch": 0.58, "grad_norm": 8.315825462341309, "learning_rate": 7.794403046181935e-06, "loss": 1.5476, "step": 3626 }, { "epoch": 0.58, "grad_norm": 7.2451066970825195, "learning_rate": 7.789317493713113e-06, "loss": 1.5263, "step": 3627 }, { "epoch": 0.58, "grad_norm": 7.699170112609863, "learning_rate": 7.784232542296236e-06, "loss": 1.439, "step": 3628 }, { "epoch": 0.58, "grad_norm": 7.903811454772949, "learning_rate": 7.779148193313833e-06, "loss": 1.3568, "step": 3629 }, { "epoch": 0.58, "grad_norm": 10.45702838897705, "learning_rate": 7.774064448148255e-06, "loss": 1.5608, "step": 3630 }, { "epoch": 0.58, "grad_norm": 8.532240867614746, "learning_rate": 7.768981308181702e-06, "loss": 1.3652, "step": 3631 }, { "epoch": 0.58, "grad_norm": 8.790132522583008, "learning_rate": 7.763898774796204e-06, "loss": 1.3017, "step": 3632 }, { "epoch": 0.58, "grad_norm": 8.346508979797363, "learning_rate": 7.758816849373624e-06, "loss": 1.4963, "step": 3633 }, { "epoch": 0.58, "grad_norm": 7.1818013191223145, "learning_rate": 7.75373553329567e-06, "loss": 1.3473, "step": 3634 }, { "epoch": 0.59, "grad_norm": 8.336629867553711, "learning_rate": 7.748654827943866e-06, "loss": 1.3066, "step": 3635 }, { "epoch": 0.59, "grad_norm": 6.79742431640625, "learning_rate": 7.743574734699589e-06, "loss": 1.3694, "step": 3636 }, { "epoch": 0.59, "grad_norm": 7.359267234802246, "learning_rate": 7.738495254944042e-06, "loss": 1.3192, "step": 3637 }, { "epoch": 0.59, "grad_norm": 9.406243324279785, "learning_rate": 7.733416390058254e-06, "loss": 1.4933, "step": 3638 }, { "epoch": 0.59, "grad_norm": 9.556588172912598, "learning_rate": 7.7283381414231e-06, "loss": 1.5262, "step": 3639 }, { "epoch": 0.59, "grad_norm": 7.081025123596191, "learning_rate": 7.723260510419276e-06, "loss": 1.5212, "step": 3640 }, { "epoch": 0.59, "grad_norm": 7.488142967224121, "learning_rate": 7.718183498427317e-06, "loss": 1.5403, "step": 3641 }, { "epoch": 0.59, "grad_norm": 6.533266544342041, "learning_rate": 7.71310710682759e-06, "loss": 1.3964, "step": 3642 }, { "epoch": 0.59, "grad_norm": 7.95514440536499, "learning_rate": 7.708031337000284e-06, "loss": 1.4501, "step": 3643 }, { "epoch": 0.59, "grad_norm": 8.677912712097168, "learning_rate": 7.702956190325435e-06, "loss": 1.361, "step": 3644 }, { "epoch": 0.59, "grad_norm": 9.601069450378418, "learning_rate": 7.697881668182892e-06, "loss": 1.2833, "step": 3645 }, { "epoch": 0.59, "grad_norm": 7.915455341339111, "learning_rate": 7.69280777195235e-06, "loss": 1.4113, "step": 3646 }, { "epoch": 0.59, "grad_norm": 9.900712013244629, "learning_rate": 7.687734503013322e-06, "loss": 1.3405, "step": 3647 }, { "epoch": 0.59, "grad_norm": 6.788515567779541, "learning_rate": 7.682661862745153e-06, "loss": 1.2487, "step": 3648 }, { "epoch": 0.59, "grad_norm": 8.070281982421875, "learning_rate": 7.677589852527027e-06, "loss": 1.4381, "step": 3649 }, { "epoch": 0.59, "grad_norm": 7.800512790679932, "learning_rate": 7.672518473737937e-06, "loss": 1.4856, "step": 3650 }, { "epoch": 0.59, "grad_norm": 12.218361854553223, "learning_rate": 7.667447727756728e-06, "loss": 1.6682, "step": 3651 }, { "epoch": 0.59, "grad_norm": 9.391643524169922, "learning_rate": 7.662377615962056e-06, "loss": 1.4488, "step": 3652 }, { "epoch": 0.59, "grad_norm": 8.793267250061035, "learning_rate": 7.657308139732405e-06, "loss": 1.4952, "step": 3653 }, { "epoch": 0.59, "grad_norm": 8.639135360717773, "learning_rate": 7.652239300446101e-06, "loss": 1.2948, "step": 3654 }, { "epoch": 0.59, "grad_norm": 9.723383903503418, "learning_rate": 7.647171099481274e-06, "loss": 1.6526, "step": 3655 }, { "epoch": 0.59, "grad_norm": 6.9471564292907715, "learning_rate": 7.642103538215904e-06, "loss": 1.6466, "step": 3656 }, { "epoch": 0.59, "grad_norm": 8.389240264892578, "learning_rate": 7.637036618027783e-06, "loss": 1.5604, "step": 3657 }, { "epoch": 0.59, "grad_norm": 7.633325099945068, "learning_rate": 7.631970340294526e-06, "loss": 1.5639, "step": 3658 }, { "epoch": 0.59, "grad_norm": 9.513532638549805, "learning_rate": 7.626904706393587e-06, "loss": 1.5098, "step": 3659 }, { "epoch": 0.59, "grad_norm": 11.162376403808594, "learning_rate": 7.621839717702233e-06, "loss": 1.231, "step": 3660 }, { "epoch": 0.59, "grad_norm": 10.26362133026123, "learning_rate": 7.616775375597561e-06, "loss": 1.695, "step": 3661 }, { "epoch": 0.59, "grad_norm": 7.5393452644348145, "learning_rate": 7.611711681456493e-06, "loss": 1.3755, "step": 3662 }, { "epoch": 0.59, "grad_norm": 10.669122695922852, "learning_rate": 7.606648636655764e-06, "loss": 1.2184, "step": 3663 }, { "epoch": 0.59, "grad_norm": 6.922031879425049, "learning_rate": 7.6015862425719514e-06, "loss": 1.3388, "step": 3664 }, { "epoch": 0.59, "grad_norm": 8.702180862426758, "learning_rate": 7.596524500581438e-06, "loss": 1.5047, "step": 3665 }, { "epoch": 0.59, "grad_norm": 7.656057357788086, "learning_rate": 7.591463412060445e-06, "loss": 1.3279, "step": 3666 }, { "epoch": 0.59, "grad_norm": 8.313104629516602, "learning_rate": 7.586402978385002e-06, "loss": 1.243, "step": 3667 }, { "epoch": 0.59, "grad_norm": 7.425998687744141, "learning_rate": 7.581343200930964e-06, "loss": 1.6251, "step": 3668 }, { "epoch": 0.59, "grad_norm": 7.9854326248168945, "learning_rate": 7.5762840810740145e-06, "loss": 1.2769, "step": 3669 }, { "epoch": 0.59, "grad_norm": 8.235331535339355, "learning_rate": 7.571225620189648e-06, "loss": 1.5208, "step": 3670 }, { "epoch": 0.59, "grad_norm": 8.484843254089355, "learning_rate": 7.566167819653195e-06, "loss": 1.7377, "step": 3671 }, { "epoch": 0.59, "grad_norm": 8.587316513061523, "learning_rate": 7.5611106808397875e-06, "loss": 1.7074, "step": 3672 }, { "epoch": 0.59, "grad_norm": 7.2339558601379395, "learning_rate": 7.556054205124392e-06, "loss": 1.5568, "step": 3673 }, { "epoch": 0.59, "grad_norm": 7.378410339355469, "learning_rate": 7.550998393881787e-06, "loss": 1.3216, "step": 3674 }, { "epoch": 0.59, "grad_norm": 7.956549167633057, "learning_rate": 7.545943248486573e-06, "loss": 1.5845, "step": 3675 }, { "epoch": 0.59, "grad_norm": 6.680653095245361, "learning_rate": 7.540888770313174e-06, "loss": 1.426, "step": 3676 }, { "epoch": 0.59, "grad_norm": 6.735803604125977, "learning_rate": 7.53583496073582e-06, "loss": 1.4678, "step": 3677 }, { "epoch": 0.59, "grad_norm": 7.503078937530518, "learning_rate": 7.530781821128575e-06, "loss": 1.4084, "step": 3678 }, { "epoch": 0.59, "grad_norm": 7.971983909606934, "learning_rate": 7.525729352865309e-06, "loss": 1.1434, "step": 3679 }, { "epoch": 0.59, "grad_norm": 7.315600872039795, "learning_rate": 7.520677557319711e-06, "loss": 1.382, "step": 3680 }, { "epoch": 0.59, "grad_norm": 9.312154769897461, "learning_rate": 7.5156264358652975e-06, "loss": 1.5358, "step": 3681 }, { "epoch": 0.59, "grad_norm": 8.317914009094238, "learning_rate": 7.510575989875386e-06, "loss": 1.5073, "step": 3682 }, { "epoch": 0.59, "grad_norm": 7.735452175140381, "learning_rate": 7.505526220723125e-06, "loss": 1.4589, "step": 3683 }, { "epoch": 0.59, "grad_norm": 8.492888450622559, "learning_rate": 7.500477129781466e-06, "loss": 1.4099, "step": 3684 }, { "epoch": 0.59, "grad_norm": 7.605221271514893, "learning_rate": 7.495428718423183e-06, "loss": 1.3534, "step": 3685 }, { "epoch": 0.59, "grad_norm": 7.494222164154053, "learning_rate": 7.490380988020872e-06, "loss": 1.3263, "step": 3686 }, { "epoch": 0.59, "grad_norm": 7.593504428863525, "learning_rate": 7.485333939946926e-06, "loss": 1.4424, "step": 3687 }, { "epoch": 0.59, "grad_norm": 7.671426296234131, "learning_rate": 7.480287575573571e-06, "loss": 1.4576, "step": 3688 }, { "epoch": 0.59, "grad_norm": 7.124802589416504, "learning_rate": 7.475241896272836e-06, "loss": 1.4005, "step": 3689 }, { "epoch": 0.59, "grad_norm": 8.741043090820312, "learning_rate": 7.470196903416562e-06, "loss": 1.507, "step": 3690 }, { "epoch": 0.59, "grad_norm": 8.192656517028809, "learning_rate": 7.465152598376419e-06, "loss": 1.3705, "step": 3691 }, { "epoch": 0.59, "grad_norm": 7.4763288497924805, "learning_rate": 7.460108982523868e-06, "loss": 1.4653, "step": 3692 }, { "epoch": 0.59, "grad_norm": 8.961297035217285, "learning_rate": 7.455066057230201e-06, "loss": 1.6343, "step": 3693 }, { "epoch": 0.59, "grad_norm": 8.122835159301758, "learning_rate": 7.4500238238665125e-06, "loss": 1.6264, "step": 3694 }, { "epoch": 0.59, "grad_norm": 8.184385299682617, "learning_rate": 7.444982283803708e-06, "loss": 1.3435, "step": 3695 }, { "epoch": 0.59, "grad_norm": 7.5678277015686035, "learning_rate": 7.4399414384125155e-06, "loss": 1.3101, "step": 3696 }, { "epoch": 0.6, "grad_norm": 6.968812465667725, "learning_rate": 7.4349012890634585e-06, "loss": 1.4115, "step": 3697 }, { "epoch": 0.6, "grad_norm": 8.62995433807373, "learning_rate": 7.429861837126884e-06, "loss": 1.3337, "step": 3698 }, { "epoch": 0.6, "grad_norm": 8.221208572387695, "learning_rate": 7.424823083972945e-06, "loss": 1.5938, "step": 3699 }, { "epoch": 0.6, "grad_norm": 9.839801788330078, "learning_rate": 7.4197850309716e-06, "loss": 1.4102, "step": 3700 }, { "epoch": 0.6, "grad_norm": 7.985947608947754, "learning_rate": 7.414747679492627e-06, "loss": 1.5301, "step": 3701 }, { "epoch": 0.6, "grad_norm": 7.772754669189453, "learning_rate": 7.4097110309056e-06, "loss": 1.4145, "step": 3702 }, { "epoch": 0.6, "grad_norm": 6.743189811706543, "learning_rate": 7.404675086579915e-06, "loss": 1.2016, "step": 3703 }, { "epoch": 0.6, "grad_norm": 9.338519096374512, "learning_rate": 7.399639847884773e-06, "loss": 1.4602, "step": 3704 }, { "epoch": 0.6, "grad_norm": 8.024715423583984, "learning_rate": 7.394605316189173e-06, "loss": 1.3388, "step": 3705 }, { "epoch": 0.6, "grad_norm": 7.069015026092529, "learning_rate": 7.389571492861939e-06, "loss": 1.1881, "step": 3706 }, { "epoch": 0.6, "grad_norm": 7.454481601715088, "learning_rate": 7.384538379271684e-06, "loss": 1.5789, "step": 3707 }, { "epoch": 0.6, "grad_norm": 8.172996520996094, "learning_rate": 7.379505976786845e-06, "loss": 1.429, "step": 3708 }, { "epoch": 0.6, "grad_norm": 7.860394477844238, "learning_rate": 7.374474286775652e-06, "loss": 1.3337, "step": 3709 }, { "epoch": 0.6, "grad_norm": 8.360976219177246, "learning_rate": 7.369443310606157e-06, "loss": 1.6721, "step": 3710 }, { "epoch": 0.6, "grad_norm": 7.551934242248535, "learning_rate": 7.364413049646198e-06, "loss": 1.6447, "step": 3711 }, { "epoch": 0.6, "grad_norm": 10.649571418762207, "learning_rate": 7.359383505263431e-06, "loss": 1.201, "step": 3712 }, { "epoch": 0.6, "grad_norm": 7.406446933746338, "learning_rate": 7.354354678825317e-06, "loss": 1.4606, "step": 3713 }, { "epoch": 0.6, "grad_norm": 8.857792854309082, "learning_rate": 7.349326571699117e-06, "loss": 1.4785, "step": 3714 }, { "epoch": 0.6, "grad_norm": 7.383384704589844, "learning_rate": 7.344299185251904e-06, "loss": 1.5538, "step": 3715 }, { "epoch": 0.6, "grad_norm": 6.917562961578369, "learning_rate": 7.339272520850547e-06, "loss": 1.3393, "step": 3716 }, { "epoch": 0.6, "grad_norm": 9.604183197021484, "learning_rate": 7.334246579861719e-06, "loss": 1.8413, "step": 3717 }, { "epoch": 0.6, "grad_norm": 7.129518032073975, "learning_rate": 7.329221363651906e-06, "loss": 1.3989, "step": 3718 }, { "epoch": 0.6, "grad_norm": 10.317983627319336, "learning_rate": 7.324196873587381e-06, "loss": 1.5172, "step": 3719 }, { "epoch": 0.6, "grad_norm": 8.409467697143555, "learning_rate": 7.31917311103424e-06, "loss": 1.2906, "step": 3720 }, { "epoch": 0.6, "grad_norm": 8.6055326461792, "learning_rate": 7.314150077358362e-06, "loss": 1.5703, "step": 3721 }, { "epoch": 0.6, "grad_norm": 8.729107856750488, "learning_rate": 7.309127773925434e-06, "loss": 1.4318, "step": 3722 }, { "epoch": 0.6, "grad_norm": 8.08101749420166, "learning_rate": 7.304106202100955e-06, "loss": 1.6424, "step": 3723 }, { "epoch": 0.6, "grad_norm": 7.5381598472595215, "learning_rate": 7.2990853632502056e-06, "loss": 1.496, "step": 3724 }, { "epoch": 0.6, "grad_norm": 8.565125465393066, "learning_rate": 7.294065258738288e-06, "loss": 1.5688, "step": 3725 }, { "epoch": 0.6, "grad_norm": 7.6320109367370605, "learning_rate": 7.289045889930089e-06, "loss": 1.5932, "step": 3726 }, { "epoch": 0.6, "grad_norm": 7.130731582641602, "learning_rate": 7.284027258190298e-06, "loss": 1.1721, "step": 3727 }, { "epoch": 0.6, "grad_norm": 9.605298042297363, "learning_rate": 7.279009364883414e-06, "loss": 1.6222, "step": 3728 }, { "epoch": 0.6, "grad_norm": 8.691862106323242, "learning_rate": 7.273992211373722e-06, "loss": 1.7376, "step": 3729 }, { "epoch": 0.6, "grad_norm": 7.673044204711914, "learning_rate": 7.268975799025319e-06, "loss": 1.3976, "step": 3730 }, { "epoch": 0.6, "grad_norm": 7.273626804351807, "learning_rate": 7.263960129202086e-06, "loss": 1.3087, "step": 3731 }, { "epoch": 0.6, "grad_norm": 8.712641716003418, "learning_rate": 7.258945203267711e-06, "loss": 1.4251, "step": 3732 }, { "epoch": 0.6, "grad_norm": 6.1485419273376465, "learning_rate": 7.253931022585684e-06, "loss": 1.4666, "step": 3733 }, { "epoch": 0.6, "grad_norm": 7.435717582702637, "learning_rate": 7.248917588519279e-06, "loss": 1.4164, "step": 3734 }, { "epoch": 0.6, "grad_norm": 6.661858081817627, "learning_rate": 7.243904902431582e-06, "loss": 1.4294, "step": 3735 }, { "epoch": 0.6, "grad_norm": 9.938680648803711, "learning_rate": 7.238892965685465e-06, "loss": 1.615, "step": 3736 }, { "epoch": 0.6, "grad_norm": 7.881718635559082, "learning_rate": 7.233881779643595e-06, "loss": 1.5729, "step": 3737 }, { "epoch": 0.6, "grad_norm": 8.687936782836914, "learning_rate": 7.228871345668449e-06, "loss": 1.5925, "step": 3738 }, { "epoch": 0.6, "grad_norm": 6.179525852203369, "learning_rate": 7.22386166512228e-06, "loss": 1.0753, "step": 3739 }, { "epoch": 0.6, "grad_norm": 8.07955551147461, "learning_rate": 7.218852739367155e-06, "loss": 1.4878, "step": 3740 }, { "epoch": 0.6, "grad_norm": 8.786494255065918, "learning_rate": 7.2138445697649255e-06, "loss": 1.2792, "step": 3741 }, { "epoch": 0.6, "grad_norm": 8.426606178283691, "learning_rate": 7.208837157677231e-06, "loss": 1.4981, "step": 3742 }, { "epoch": 0.6, "grad_norm": 7.887040615081787, "learning_rate": 7.2038305044655235e-06, "loss": 1.0492, "step": 3743 }, { "epoch": 0.6, "grad_norm": 6.594301223754883, "learning_rate": 7.198824611491028e-06, "loss": 1.3886, "step": 3744 }, { "epoch": 0.6, "grad_norm": 9.624049186706543, "learning_rate": 7.193819480114782e-06, "loss": 1.4885, "step": 3745 }, { "epoch": 0.6, "grad_norm": 8.700055122375488, "learning_rate": 7.1888151116976e-06, "loss": 1.3186, "step": 3746 }, { "epoch": 0.6, "grad_norm": 8.667389869689941, "learning_rate": 7.1838115076001034e-06, "loss": 1.4487, "step": 3747 }, { "epoch": 0.6, "grad_norm": 7.879819869995117, "learning_rate": 7.178808669182694e-06, "loss": 1.3285, "step": 3748 }, { "epoch": 0.6, "grad_norm": 8.442161560058594, "learning_rate": 7.173806597805566e-06, "loss": 1.2387, "step": 3749 }, { "epoch": 0.6, "grad_norm": 7.855050086975098, "learning_rate": 7.168805294828713e-06, "loss": 1.4765, "step": 3750 }, { "epoch": 0.6, "grad_norm": 8.193105697631836, "learning_rate": 7.163804761611912e-06, "loss": 1.35, "step": 3751 }, { "epoch": 0.6, "grad_norm": 10.272113800048828, "learning_rate": 7.158804999514744e-06, "loss": 1.8146, "step": 3752 }, { "epoch": 0.6, "grad_norm": 8.281051635742188, "learning_rate": 7.1538060098965625e-06, "loss": 1.3514, "step": 3753 }, { "epoch": 0.6, "grad_norm": 8.567418098449707, "learning_rate": 7.148807794116515e-06, "loss": 1.301, "step": 3754 }, { "epoch": 0.6, "grad_norm": 7.871013641357422, "learning_rate": 7.143810353533549e-06, "loss": 1.2227, "step": 3755 }, { "epoch": 0.6, "grad_norm": 8.332205772399902, "learning_rate": 7.138813689506392e-06, "loss": 1.4087, "step": 3756 }, { "epoch": 0.6, "grad_norm": 8.609530448913574, "learning_rate": 7.13381780339357e-06, "loss": 1.3853, "step": 3757 }, { "epoch": 0.6, "grad_norm": 8.909804344177246, "learning_rate": 7.128822696553383e-06, "loss": 1.4456, "step": 3758 }, { "epoch": 0.61, "grad_norm": 6.875055313110352, "learning_rate": 7.123828370343928e-06, "loss": 1.3758, "step": 3759 }, { "epoch": 0.61, "grad_norm": 8.471609115600586, "learning_rate": 7.11883482612309e-06, "loss": 1.6615, "step": 3760 }, { "epoch": 0.61, "grad_norm": 9.498099327087402, "learning_rate": 7.113842065248537e-06, "loss": 1.496, "step": 3761 }, { "epoch": 0.61, "grad_norm": 8.35847282409668, "learning_rate": 7.108850089077736e-06, "loss": 1.3196, "step": 3762 }, { "epoch": 0.61, "grad_norm": 8.818148612976074, "learning_rate": 7.103858898967921e-06, "loss": 1.5936, "step": 3763 }, { "epoch": 0.61, "grad_norm": 6.849881649017334, "learning_rate": 7.09886849627613e-06, "loss": 1.1719, "step": 3764 }, { "epoch": 0.61, "grad_norm": 6.38147497177124, "learning_rate": 7.093878882359175e-06, "loss": 1.4011, "step": 3765 }, { "epoch": 0.61, "grad_norm": 6.921889305114746, "learning_rate": 7.088890058573658e-06, "loss": 1.4331, "step": 3766 }, { "epoch": 0.61, "grad_norm": 7.160775661468506, "learning_rate": 7.083902026275975e-06, "loss": 1.4091, "step": 3767 }, { "epoch": 0.61, "grad_norm": 8.941510200500488, "learning_rate": 7.07891478682229e-06, "loss": 1.5776, "step": 3768 }, { "epoch": 0.61, "grad_norm": 8.164468765258789, "learning_rate": 7.0739283415685585e-06, "loss": 1.3777, "step": 3769 }, { "epoch": 0.61, "grad_norm": 8.765122413635254, "learning_rate": 7.0689426918705315e-06, "loss": 1.2757, "step": 3770 }, { "epoch": 0.61, "grad_norm": 10.14555835723877, "learning_rate": 7.063957839083721e-06, "loss": 1.522, "step": 3771 }, { "epoch": 0.61, "grad_norm": 7.2635111808776855, "learning_rate": 7.0589737845634474e-06, "loss": 1.3924, "step": 3772 }, { "epoch": 0.61, "grad_norm": 8.019185066223145, "learning_rate": 7.053990529664792e-06, "loss": 1.5551, "step": 3773 }, { "epoch": 0.61, "grad_norm": 7.710824489593506, "learning_rate": 7.04900807574263e-06, "loss": 1.4042, "step": 3774 }, { "epoch": 0.61, "grad_norm": 8.232827186584473, "learning_rate": 7.044026424151622e-06, "loss": 1.2499, "step": 3775 }, { "epoch": 0.61, "grad_norm": 8.793900489807129, "learning_rate": 7.039045576246197e-06, "loss": 1.424, "step": 3776 }, { "epoch": 0.61, "grad_norm": 9.808868408203125, "learning_rate": 7.034065533380585e-06, "loss": 1.6168, "step": 3777 }, { "epoch": 0.61, "grad_norm": 10.24260139465332, "learning_rate": 7.029086296908778e-06, "loss": 1.1656, "step": 3778 }, { "epoch": 0.61, "grad_norm": 10.316784858703613, "learning_rate": 7.024107868184558e-06, "loss": 1.6373, "step": 3779 }, { "epoch": 0.61, "grad_norm": 9.952693939208984, "learning_rate": 7.0191302485614896e-06, "loss": 1.439, "step": 3780 }, { "epoch": 0.61, "grad_norm": 7.709405422210693, "learning_rate": 7.014153439392909e-06, "loss": 1.4629, "step": 3781 }, { "epoch": 0.61, "grad_norm": 8.696826934814453, "learning_rate": 7.009177442031945e-06, "loss": 1.386, "step": 3782 }, { "epoch": 0.61, "grad_norm": 8.154853820800781, "learning_rate": 7.00420225783149e-06, "loss": 1.6683, "step": 3783 }, { "epoch": 0.61, "grad_norm": 7.835132122039795, "learning_rate": 6.9992278881442245e-06, "loss": 1.427, "step": 3784 }, { "epoch": 0.61, "grad_norm": 8.28396987915039, "learning_rate": 6.994254334322611e-06, "loss": 1.4178, "step": 3785 }, { "epoch": 0.61, "grad_norm": 6.741214752197266, "learning_rate": 6.989281597718879e-06, "loss": 1.3169, "step": 3786 }, { "epoch": 0.61, "grad_norm": 8.961577415466309, "learning_rate": 6.984309679685049e-06, "loss": 1.339, "step": 3787 }, { "epoch": 0.61, "grad_norm": 8.589962005615234, "learning_rate": 6.9793385815729055e-06, "loss": 1.1875, "step": 3788 }, { "epoch": 0.61, "grad_norm": 7.672857284545898, "learning_rate": 6.974368304734022e-06, "loss": 1.4047, "step": 3789 }, { "epoch": 0.61, "grad_norm": 7.259000301361084, "learning_rate": 6.9693988505197415e-06, "loss": 1.3935, "step": 3790 }, { "epoch": 0.61, "grad_norm": 7.21064567565918, "learning_rate": 6.964430220281182e-06, "loss": 1.4623, "step": 3791 }, { "epoch": 0.61, "grad_norm": 7.224093437194824, "learning_rate": 6.959462415369247e-06, "loss": 1.4403, "step": 3792 }, { "epoch": 0.61, "grad_norm": 8.828861236572266, "learning_rate": 6.954495437134603e-06, "loss": 1.5117, "step": 3793 }, { "epoch": 0.61, "grad_norm": 8.307425498962402, "learning_rate": 6.949529286927704e-06, "loss": 1.6029, "step": 3794 }, { "epoch": 0.61, "grad_norm": 8.244769096374512, "learning_rate": 6.944563966098772e-06, "loss": 1.2422, "step": 3795 }, { "epoch": 0.61, "grad_norm": 6.926759243011475, "learning_rate": 6.939599475997797e-06, "loss": 1.3772, "step": 3796 }, { "epoch": 0.61, "grad_norm": 8.935966491699219, "learning_rate": 6.9346358179745574e-06, "loss": 1.5196, "step": 3797 }, { "epoch": 0.61, "grad_norm": 8.069478034973145, "learning_rate": 6.929672993378598e-06, "loss": 1.2548, "step": 3798 }, { "epoch": 0.61, "grad_norm": 7.922458648681641, "learning_rate": 6.924711003559239e-06, "loss": 1.3066, "step": 3799 }, { "epoch": 0.61, "grad_norm": 8.110422134399414, "learning_rate": 6.91974984986557e-06, "loss": 1.6285, "step": 3800 }, { "epoch": 0.61, "grad_norm": 8.19252872467041, "learning_rate": 6.9147895336464525e-06, "loss": 1.2768, "step": 3801 }, { "epoch": 0.61, "grad_norm": 6.5496978759765625, "learning_rate": 6.909830056250527e-06, "loss": 1.1131, "step": 3802 }, { "epoch": 0.61, "grad_norm": 8.208222389221191, "learning_rate": 6.9048714190262e-06, "loss": 1.4649, "step": 3803 }, { "epoch": 0.61, "grad_norm": 9.87598991394043, "learning_rate": 6.899913623321657e-06, "loss": 1.4034, "step": 3804 }, { "epoch": 0.61, "grad_norm": 8.570489883422852, "learning_rate": 6.894956670484847e-06, "loss": 1.7864, "step": 3805 }, { "epoch": 0.61, "grad_norm": 7.817288398742676, "learning_rate": 6.890000561863486e-06, "loss": 1.416, "step": 3806 }, { "epoch": 0.61, "grad_norm": 8.434532165527344, "learning_rate": 6.885045298805073e-06, "loss": 1.3794, "step": 3807 }, { "epoch": 0.61, "grad_norm": 8.684596061706543, "learning_rate": 6.880090882656869e-06, "loss": 1.2412, "step": 3808 }, { "epoch": 0.61, "grad_norm": 8.150996208190918, "learning_rate": 6.875137314765911e-06, "loss": 1.3002, "step": 3809 }, { "epoch": 0.61, "grad_norm": 10.452916145324707, "learning_rate": 6.8701845964789975e-06, "loss": 1.3958, "step": 3810 }, { "epoch": 0.61, "grad_norm": 6.732020378112793, "learning_rate": 6.865232729142695e-06, "loss": 1.4579, "step": 3811 }, { "epoch": 0.61, "grad_norm": 7.2015862464904785, "learning_rate": 6.86028171410335e-06, "loss": 1.5153, "step": 3812 }, { "epoch": 0.61, "grad_norm": 8.106274604797363, "learning_rate": 6.855331552707065e-06, "loss": 1.4635, "step": 3813 }, { "epoch": 0.61, "grad_norm": 8.774352073669434, "learning_rate": 6.8503822462997225e-06, "loss": 1.4682, "step": 3814 }, { "epoch": 0.61, "grad_norm": 7.476193428039551, "learning_rate": 6.8454337962269615e-06, "loss": 1.1695, "step": 3815 }, { "epoch": 0.61, "grad_norm": 6.973299026489258, "learning_rate": 6.840486203834193e-06, "loss": 1.4856, "step": 3816 }, { "epoch": 0.61, "grad_norm": 9.022244453430176, "learning_rate": 6.8355394704665935e-06, "loss": 1.5131, "step": 3817 }, { "epoch": 0.61, "grad_norm": 7.269217491149902, "learning_rate": 6.830593597469106e-06, "loss": 1.4774, "step": 3818 }, { "epoch": 0.61, "grad_norm": 7.8718156814575195, "learning_rate": 6.8256485861864465e-06, "loss": 1.3184, "step": 3819 }, { "epoch": 0.61, "grad_norm": 8.210745811462402, "learning_rate": 6.820704437963084e-06, "loss": 1.2666, "step": 3820 }, { "epoch": 0.62, "grad_norm": 8.260676383972168, "learning_rate": 6.81576115414326e-06, "loss": 1.728, "step": 3821 }, { "epoch": 0.62, "grad_norm": 7.872150897979736, "learning_rate": 6.810818736070985e-06, "loss": 1.5526, "step": 3822 }, { "epoch": 0.62, "grad_norm": 8.553590774536133, "learning_rate": 6.805877185090024e-06, "loss": 1.5004, "step": 3823 }, { "epoch": 0.62, "grad_norm": 7.939025402069092, "learning_rate": 6.800936502543918e-06, "loss": 1.6511, "step": 3824 }, { "epoch": 0.62, "grad_norm": 7.29510498046875, "learning_rate": 6.795996689775959e-06, "loss": 1.5045, "step": 3825 }, { "epoch": 0.62, "grad_norm": 8.491726875305176, "learning_rate": 6.791057748129213e-06, "loss": 1.4896, "step": 3826 }, { "epoch": 0.62, "grad_norm": 7.495156764984131, "learning_rate": 6.786119678946506e-06, "loss": 1.5451, "step": 3827 }, { "epoch": 0.62, "grad_norm": 8.01884651184082, "learning_rate": 6.781182483570421e-06, "loss": 1.2439, "step": 3828 }, { "epoch": 0.62, "grad_norm": 7.2430267333984375, "learning_rate": 6.776246163343317e-06, "loss": 1.6557, "step": 3829 }, { "epoch": 0.62, "grad_norm": 7.075965881347656, "learning_rate": 6.771310719607298e-06, "loss": 1.4053, "step": 3830 }, { "epoch": 0.62, "grad_norm": 7.0770955085754395, "learning_rate": 6.766376153704244e-06, "loss": 1.4854, "step": 3831 }, { "epoch": 0.62, "grad_norm": 7.26938009262085, "learning_rate": 6.761442466975792e-06, "loss": 1.2685, "step": 3832 }, { "epoch": 0.62, "grad_norm": 7.828185558319092, "learning_rate": 6.756509660763329e-06, "loss": 1.1689, "step": 3833 }, { "epoch": 0.62, "grad_norm": 7.669249534606934, "learning_rate": 6.751577736408024e-06, "loss": 1.5443, "step": 3834 }, { "epoch": 0.62, "grad_norm": 7.252708911895752, "learning_rate": 6.746646695250785e-06, "loss": 1.4005, "step": 3835 }, { "epoch": 0.62, "grad_norm": 7.414449214935303, "learning_rate": 6.741716538632295e-06, "loss": 1.4956, "step": 3836 }, { "epoch": 0.62, "grad_norm": 7.622738838195801, "learning_rate": 6.736787267892991e-06, "loss": 1.5193, "step": 3837 }, { "epoch": 0.62, "grad_norm": 8.263288497924805, "learning_rate": 6.731858884373064e-06, "loss": 1.3696, "step": 3838 }, { "epoch": 0.62, "grad_norm": 8.546059608459473, "learning_rate": 6.726931389412476e-06, "loss": 1.4633, "step": 3839 }, { "epoch": 0.62, "grad_norm": 7.304541110992432, "learning_rate": 6.7220047843509314e-06, "loss": 1.2411, "step": 3840 }, { "epoch": 0.62, "grad_norm": 10.117681503295898, "learning_rate": 6.7170790705279075e-06, "loss": 1.2358, "step": 3841 }, { "epoch": 0.62, "grad_norm": 7.918576240539551, "learning_rate": 6.7121542492826365e-06, "loss": 1.4756, "step": 3842 }, { "epoch": 0.62, "grad_norm": 8.0985746383667, "learning_rate": 6.707230321954092e-06, "loss": 1.3184, "step": 3843 }, { "epoch": 0.62, "grad_norm": 6.916600704193115, "learning_rate": 6.70230728988103e-06, "loss": 1.7025, "step": 3844 }, { "epoch": 0.62, "grad_norm": 9.205869674682617, "learning_rate": 6.697385154401943e-06, "loss": 1.5169, "step": 3845 }, { "epoch": 0.62, "grad_norm": 7.904818058013916, "learning_rate": 6.692463916855092e-06, "loss": 1.2714, "step": 3846 }, { "epoch": 0.62, "grad_norm": 7.800785064697266, "learning_rate": 6.687543578578488e-06, "loss": 1.4947, "step": 3847 }, { "epoch": 0.62, "grad_norm": 8.60533332824707, "learning_rate": 6.6826241409098905e-06, "loss": 1.5362, "step": 3848 }, { "epoch": 0.62, "grad_norm": 7.758907318115234, "learning_rate": 6.67770560518683e-06, "loss": 1.5861, "step": 3849 }, { "epoch": 0.62, "grad_norm": 9.625874519348145, "learning_rate": 6.6727879727465815e-06, "loss": 1.3012, "step": 3850 }, { "epoch": 0.62, "grad_norm": 7.275128364562988, "learning_rate": 6.667871244926181e-06, "loss": 1.2729, "step": 3851 }, { "epoch": 0.62, "grad_norm": 6.996921539306641, "learning_rate": 6.6629554230624096e-06, "loss": 1.2623, "step": 3852 }, { "epoch": 0.62, "grad_norm": 8.101728439331055, "learning_rate": 6.658040508491802e-06, "loss": 1.3141, "step": 3853 }, { "epoch": 0.62, "grad_norm": 7.819444179534912, "learning_rate": 6.65312650255066e-06, "loss": 1.2978, "step": 3854 }, { "epoch": 0.62, "grad_norm": 8.168121337890625, "learning_rate": 6.6482134065750226e-06, "loss": 1.4071, "step": 3855 }, { "epoch": 0.62, "grad_norm": 7.207710266113281, "learning_rate": 6.6433012219006955e-06, "loss": 1.1378, "step": 3856 }, { "epoch": 0.62, "grad_norm": 8.779569625854492, "learning_rate": 6.6383899498632245e-06, "loss": 1.6027, "step": 3857 }, { "epoch": 0.62, "grad_norm": 7.816015243530273, "learning_rate": 6.633479591797908e-06, "loss": 1.7478, "step": 3858 }, { "epoch": 0.62, "grad_norm": 8.834107398986816, "learning_rate": 6.628570149039806e-06, "loss": 1.5251, "step": 3859 }, { "epoch": 0.62, "grad_norm": 7.328250885009766, "learning_rate": 6.6236616229237205e-06, "loss": 1.3761, "step": 3860 }, { "epoch": 0.62, "grad_norm": 7.6692585945129395, "learning_rate": 6.618754014784212e-06, "loss": 1.2709, "step": 3861 }, { "epoch": 0.62, "grad_norm": 8.171813011169434, "learning_rate": 6.613847325955578e-06, "loss": 1.4021, "step": 3862 }, { "epoch": 0.62, "grad_norm": 7.092027187347412, "learning_rate": 6.608941557771886e-06, "loss": 1.2896, "step": 3863 }, { "epoch": 0.62, "grad_norm": 8.677645683288574, "learning_rate": 6.604036711566932e-06, "loss": 1.5034, "step": 3864 }, { "epoch": 0.62, "grad_norm": 8.431353569030762, "learning_rate": 6.599132788674274e-06, "loss": 1.3741, "step": 3865 }, { "epoch": 0.62, "grad_norm": 9.683629989624023, "learning_rate": 6.594229790427222e-06, "loss": 1.2594, "step": 3866 }, { "epoch": 0.62, "grad_norm": 9.170758247375488, "learning_rate": 6.58932771815882e-06, "loss": 1.4941, "step": 3867 }, { "epoch": 0.62, "grad_norm": 7.770731449127197, "learning_rate": 6.584426573201874e-06, "loss": 1.4991, "step": 3868 }, { "epoch": 0.62, "grad_norm": 9.457159042358398, "learning_rate": 6.579526356888937e-06, "loss": 1.6573, "step": 3869 }, { "epoch": 0.62, "grad_norm": 7.943587303161621, "learning_rate": 6.574627070552295e-06, "loss": 1.3053, "step": 3870 }, { "epoch": 0.62, "grad_norm": 11.259455680847168, "learning_rate": 6.569728715524002e-06, "loss": 1.603, "step": 3871 }, { "epoch": 0.62, "grad_norm": 7.528128147125244, "learning_rate": 6.5648312931358405e-06, "loss": 1.3613, "step": 3872 }, { "epoch": 0.62, "grad_norm": 7.863809585571289, "learning_rate": 6.559934804719352e-06, "loss": 1.3245, "step": 3873 }, { "epoch": 0.62, "grad_norm": 8.754083633422852, "learning_rate": 6.55503925160582e-06, "loss": 1.3895, "step": 3874 }, { "epoch": 0.62, "grad_norm": 8.476705551147461, "learning_rate": 6.550144635126268e-06, "loss": 1.4154, "step": 3875 }, { "epoch": 0.62, "grad_norm": 7.935164451599121, "learning_rate": 6.5452509566114775e-06, "loss": 1.4974, "step": 3876 }, { "epoch": 0.62, "grad_norm": 8.588483810424805, "learning_rate": 6.540358217391959e-06, "loss": 1.5053, "step": 3877 }, { "epoch": 0.62, "grad_norm": 9.352851867675781, "learning_rate": 6.535466418797982e-06, "loss": 1.4631, "step": 3878 }, { "epoch": 0.62, "grad_norm": 7.497502326965332, "learning_rate": 6.5305755621595555e-06, "loss": 1.4078, "step": 3879 }, { "epoch": 0.62, "grad_norm": 6.958573341369629, "learning_rate": 6.525685648806425e-06, "loss": 1.2837, "step": 3880 }, { "epoch": 0.62, "grad_norm": 7.748814582824707, "learning_rate": 6.520796680068092e-06, "loss": 1.7343, "step": 3881 }, { "epoch": 0.62, "grad_norm": 8.520401954650879, "learning_rate": 6.515908657273788e-06, "loss": 1.4927, "step": 3882 }, { "epoch": 0.63, "grad_norm": 7.412539958953857, "learning_rate": 6.511021581752503e-06, "loss": 1.2423, "step": 3883 }, { "epoch": 0.63, "grad_norm": 7.633443832397461, "learning_rate": 6.506135454832956e-06, "loss": 1.3253, "step": 3884 }, { "epoch": 0.63, "grad_norm": 7.792604923248291, "learning_rate": 6.501250277843609e-06, "loss": 1.5575, "step": 3885 }, { "epoch": 0.63, "grad_norm": 7.921741962432861, "learning_rate": 6.4963660521126785e-06, "loss": 1.5766, "step": 3886 }, { "epoch": 0.63, "grad_norm": 8.073250770568848, "learning_rate": 6.491482778968103e-06, "loss": 1.4513, "step": 3887 }, { "epoch": 0.63, "grad_norm": 6.89336633682251, "learning_rate": 6.486600459737582e-06, "loss": 1.2365, "step": 3888 }, { "epoch": 0.63, "grad_norm": 8.771644592285156, "learning_rate": 6.481719095748542e-06, "loss": 1.4985, "step": 3889 }, { "epoch": 0.63, "grad_norm": 7.55137825012207, "learning_rate": 6.47683868832815e-06, "loss": 1.3246, "step": 3890 }, { "epoch": 0.63, "grad_norm": 7.533205986022949, "learning_rate": 6.471959238803325e-06, "loss": 1.4891, "step": 3891 }, { "epoch": 0.63, "grad_norm": 7.138176441192627, "learning_rate": 6.467080748500707e-06, "loss": 1.5801, "step": 3892 }, { "epoch": 0.63, "grad_norm": 8.507805824279785, "learning_rate": 6.4622032187466944e-06, "loss": 1.4669, "step": 3893 }, { "epoch": 0.63, "grad_norm": 6.952922344207764, "learning_rate": 6.457326650867415e-06, "loss": 1.5496, "step": 3894 }, { "epoch": 0.63, "grad_norm": 8.488314628601074, "learning_rate": 6.4524510461887276e-06, "loss": 1.2276, "step": 3895 }, { "epoch": 0.63, "grad_norm": 8.254371643066406, "learning_rate": 6.4475764060362455e-06, "loss": 1.5763, "step": 3896 }, { "epoch": 0.63, "grad_norm": 8.913294792175293, "learning_rate": 6.442702731735307e-06, "loss": 1.5575, "step": 3897 }, { "epoch": 0.63, "grad_norm": 7.935380935668945, "learning_rate": 6.437830024610996e-06, "loss": 1.2888, "step": 3898 }, { "epoch": 0.63, "grad_norm": 7.719210624694824, "learning_rate": 6.432958285988124e-06, "loss": 1.1938, "step": 3899 }, { "epoch": 0.63, "grad_norm": 8.075078964233398, "learning_rate": 6.428087517191252e-06, "loss": 1.1901, "step": 3900 }, { "epoch": 0.63, "grad_norm": 8.275275230407715, "learning_rate": 6.423217719544665e-06, "loss": 1.5065, "step": 3901 }, { "epoch": 0.63, "grad_norm": 6.9720892906188965, "learning_rate": 6.4183488943723884e-06, "loss": 1.4821, "step": 3902 }, { "epoch": 0.63, "grad_norm": 9.410685539245605, "learning_rate": 6.413481042998191e-06, "loss": 1.3313, "step": 3903 }, { "epoch": 0.63, "grad_norm": 8.423079490661621, "learning_rate": 6.408614166745558e-06, "loss": 1.2333, "step": 3904 }, { "epoch": 0.63, "grad_norm": 7.7469329833984375, "learning_rate": 6.403748266937734e-06, "loss": 1.4847, "step": 3905 }, { "epoch": 0.63, "grad_norm": 7.686495780944824, "learning_rate": 6.398883344897676e-06, "loss": 1.3628, "step": 3906 }, { "epoch": 0.63, "grad_norm": 9.175394058227539, "learning_rate": 6.394019401948086e-06, "loss": 1.4591, "step": 3907 }, { "epoch": 0.63, "grad_norm": 6.904367446899414, "learning_rate": 6.389156439411404e-06, "loss": 1.4477, "step": 3908 }, { "epoch": 0.63, "grad_norm": 7.794161319732666, "learning_rate": 6.38429445860979e-06, "loss": 1.6463, "step": 3909 }, { "epoch": 0.63, "grad_norm": 8.001272201538086, "learning_rate": 6.379433460865153e-06, "loss": 1.52, "step": 3910 }, { "epoch": 0.63, "grad_norm": 7.062386512756348, "learning_rate": 6.374573447499119e-06, "loss": 1.4784, "step": 3911 }, { "epoch": 0.63, "grad_norm": 6.925318717956543, "learning_rate": 6.369714419833056e-06, "loss": 1.3637, "step": 3912 }, { "epoch": 0.63, "grad_norm": 10.212793350219727, "learning_rate": 6.364856379188067e-06, "loss": 1.5476, "step": 3913 }, { "epoch": 0.63, "grad_norm": 7.547763824462891, "learning_rate": 6.359999326884974e-06, "loss": 1.5141, "step": 3914 }, { "epoch": 0.63, "grad_norm": 7.424927234649658, "learning_rate": 6.355143264244343e-06, "loss": 1.5037, "step": 3915 }, { "epoch": 0.63, "grad_norm": 9.072516441345215, "learning_rate": 6.350288192586465e-06, "loss": 1.3035, "step": 3916 }, { "epoch": 0.63, "grad_norm": 6.551723957061768, "learning_rate": 6.345434113231357e-06, "loss": 1.4463, "step": 3917 }, { "epoch": 0.63, "grad_norm": 7.324949264526367, "learning_rate": 6.340581027498782e-06, "loss": 1.2521, "step": 3918 }, { "epoch": 0.63, "grad_norm": 9.009096145629883, "learning_rate": 6.335728936708211e-06, "loss": 1.3817, "step": 3919 }, { "epoch": 0.63, "grad_norm": 8.130229949951172, "learning_rate": 6.330877842178864e-06, "loss": 1.2159, "step": 3920 }, { "epoch": 0.63, "grad_norm": 9.218814849853516, "learning_rate": 6.32602774522968e-06, "loss": 1.4599, "step": 3921 }, { "epoch": 0.63, "grad_norm": 7.877289772033691, "learning_rate": 6.321178647179324e-06, "loss": 1.2989, "step": 3922 }, { "epoch": 0.63, "grad_norm": 6.845953941345215, "learning_rate": 6.316330549346205e-06, "loss": 1.407, "step": 3923 }, { "epoch": 0.63, "grad_norm": 7.610520839691162, "learning_rate": 6.3114834530484355e-06, "loss": 1.5011, "step": 3924 }, { "epoch": 0.63, "grad_norm": 9.098103523254395, "learning_rate": 6.30663735960388e-06, "loss": 1.5655, "step": 3925 }, { "epoch": 0.63, "grad_norm": 8.556034088134766, "learning_rate": 6.301792270330118e-06, "loss": 1.2794, "step": 3926 }, { "epoch": 0.63, "grad_norm": 9.141718864440918, "learning_rate": 6.2969481865444515e-06, "loss": 1.5492, "step": 3927 }, { "epoch": 0.63, "grad_norm": 7.976918697357178, "learning_rate": 6.292105109563924e-06, "loss": 1.3126, "step": 3928 }, { "epoch": 0.63, "grad_norm": 8.729336738586426, "learning_rate": 6.287263040705287e-06, "loss": 1.5355, "step": 3929 }, { "epoch": 0.63, "grad_norm": 7.980291843414307, "learning_rate": 6.282421981285036e-06, "loss": 1.1463, "step": 3930 }, { "epoch": 0.63, "grad_norm": 7.976552486419678, "learning_rate": 6.2775819326193835e-06, "loss": 1.3399, "step": 3931 }, { "epoch": 0.63, "grad_norm": 7.542834281921387, "learning_rate": 6.27274289602426e-06, "loss": 1.2578, "step": 3932 }, { "epoch": 0.63, "grad_norm": 7.607019424438477, "learning_rate": 6.267904872815337e-06, "loss": 1.4542, "step": 3933 }, { "epoch": 0.63, "grad_norm": 9.156915664672852, "learning_rate": 6.263067864307992e-06, "loss": 1.3648, "step": 3934 }, { "epoch": 0.63, "grad_norm": 6.884602069854736, "learning_rate": 6.258231871817344e-06, "loss": 1.3821, "step": 3935 }, { "epoch": 0.63, "grad_norm": 7.43351936340332, "learning_rate": 6.253396896658224e-06, "loss": 1.2724, "step": 3936 }, { "epoch": 0.63, "grad_norm": 6.85175085067749, "learning_rate": 6.2485629401451954e-06, "loss": 1.3587, "step": 3937 }, { "epoch": 0.63, "grad_norm": 8.841726303100586, "learning_rate": 6.2437300035925365e-06, "loss": 1.5536, "step": 3938 }, { "epoch": 0.63, "grad_norm": 10.230990409851074, "learning_rate": 6.238898088314246e-06, "loss": 1.4932, "step": 3939 }, { "epoch": 0.63, "grad_norm": 7.618183612823486, "learning_rate": 6.234067195624057e-06, "loss": 1.4725, "step": 3940 }, { "epoch": 0.63, "grad_norm": 8.483302116394043, "learning_rate": 6.2292373268354156e-06, "loss": 1.6844, "step": 3941 }, { "epoch": 0.63, "grad_norm": 6.848047733306885, "learning_rate": 6.224408483261495e-06, "loss": 1.5764, "step": 3942 }, { "epoch": 0.63, "grad_norm": 8.447484016418457, "learning_rate": 6.219580666215185e-06, "loss": 1.2727, "step": 3943 }, { "epoch": 0.63, "grad_norm": 8.354989051818848, "learning_rate": 6.214753877009092e-06, "loss": 1.2424, "step": 3944 }, { "epoch": 0.64, "grad_norm": 7.659246444702148, "learning_rate": 6.209928116955556e-06, "loss": 1.39, "step": 3945 }, { "epoch": 0.64, "grad_norm": 8.047863006591797, "learning_rate": 6.2051033873666244e-06, "loss": 1.5121, "step": 3946 }, { "epoch": 0.64, "grad_norm": 9.038667678833008, "learning_rate": 6.200279689554078e-06, "loss": 1.4061, "step": 3947 }, { "epoch": 0.64, "grad_norm": 8.877473831176758, "learning_rate": 6.195457024829402e-06, "loss": 1.3414, "step": 3948 }, { "epoch": 0.64, "grad_norm": 9.84945011138916, "learning_rate": 6.1906353945038075e-06, "loss": 1.5833, "step": 3949 }, { "epoch": 0.64, "grad_norm": 8.045188903808594, "learning_rate": 6.185814799888231e-06, "loss": 1.5392, "step": 3950 }, { "epoch": 0.64, "grad_norm": 7.062826156616211, "learning_rate": 6.180995242293313e-06, "loss": 1.3421, "step": 3951 }, { "epoch": 0.64, "grad_norm": 8.1379976272583, "learning_rate": 6.176176723029428e-06, "loss": 1.2697, "step": 3952 }, { "epoch": 0.64, "grad_norm": 8.892139434814453, "learning_rate": 6.171359243406655e-06, "loss": 1.5067, "step": 3953 }, { "epoch": 0.64, "grad_norm": 7.775486469268799, "learning_rate": 6.166542804734794e-06, "loss": 1.4801, "step": 3954 }, { "epoch": 0.64, "grad_norm": 7.657649993896484, "learning_rate": 6.16172740832337e-06, "loss": 1.2419, "step": 3955 }, { "epoch": 0.64, "grad_norm": 8.356221199035645, "learning_rate": 6.156913055481613e-06, "loss": 1.289, "step": 3956 }, { "epoch": 0.64, "grad_norm": 7.753824710845947, "learning_rate": 6.152099747518477e-06, "loss": 1.4383, "step": 3957 }, { "epoch": 0.64, "grad_norm": 7.558091163635254, "learning_rate": 6.147287485742629e-06, "loss": 1.2559, "step": 3958 }, { "epoch": 0.64, "grad_norm": 7.388925075531006, "learning_rate": 6.142476271462449e-06, "loss": 1.2091, "step": 3959 }, { "epoch": 0.64, "grad_norm": 10.534942626953125, "learning_rate": 6.137666105986041e-06, "loss": 1.3558, "step": 3960 }, { "epoch": 0.64, "grad_norm": 9.552285194396973, "learning_rate": 6.132856990621211e-06, "loss": 1.5517, "step": 3961 }, { "epoch": 0.64, "grad_norm": 7.478733539581299, "learning_rate": 6.128048926675494e-06, "loss": 1.3361, "step": 3962 }, { "epoch": 0.64, "grad_norm": 7.337597846984863, "learning_rate": 6.123241915456126e-06, "loss": 1.4604, "step": 3963 }, { "epoch": 0.64, "grad_norm": 8.877425193786621, "learning_rate": 6.118435958270063e-06, "loss": 0.9975, "step": 3964 }, { "epoch": 0.64, "grad_norm": 7.740479469299316, "learning_rate": 6.113631056423978e-06, "loss": 1.3572, "step": 3965 }, { "epoch": 0.64, "grad_norm": 8.629721641540527, "learning_rate": 6.1088272112242465e-06, "loss": 1.6303, "step": 3966 }, { "epoch": 0.64, "grad_norm": 9.38941764831543, "learning_rate": 6.104024423976971e-06, "loss": 1.5041, "step": 3967 }, { "epoch": 0.64, "grad_norm": 8.504534721374512, "learning_rate": 6.099222695987951e-06, "loss": 1.4467, "step": 3968 }, { "epoch": 0.64, "grad_norm": 8.138043403625488, "learning_rate": 6.094422028562706e-06, "loss": 1.4681, "step": 3969 }, { "epoch": 0.64, "grad_norm": 8.717535972595215, "learning_rate": 6.089622423006474e-06, "loss": 1.4796, "step": 3970 }, { "epoch": 0.64, "grad_norm": 6.9260149002075195, "learning_rate": 6.084823880624189e-06, "loss": 1.2605, "step": 3971 }, { "epoch": 0.64, "grad_norm": 8.40572738647461, "learning_rate": 6.080026402720507e-06, "loss": 1.289, "step": 3972 }, { "epoch": 0.64, "grad_norm": 8.277386665344238, "learning_rate": 6.075229990599792e-06, "loss": 1.2791, "step": 3973 }, { "epoch": 0.64, "grad_norm": 7.431147575378418, "learning_rate": 6.070434645566117e-06, "loss": 1.1293, "step": 3974 }, { "epoch": 0.64, "grad_norm": 8.163564682006836, "learning_rate": 6.0656403689232675e-06, "loss": 1.3434, "step": 3975 }, { "epoch": 0.64, "grad_norm": 6.983323574066162, "learning_rate": 6.060847161974731e-06, "loss": 1.2836, "step": 3976 }, { "epoch": 0.64, "grad_norm": 7.840671539306641, "learning_rate": 6.056055026023715e-06, "loss": 1.214, "step": 3977 }, { "epoch": 0.64, "grad_norm": 8.023370742797852, "learning_rate": 6.051263962373126e-06, "loss": 1.1966, "step": 3978 }, { "epoch": 0.64, "grad_norm": 9.033288955688477, "learning_rate": 6.046473972325592e-06, "loss": 1.3893, "step": 3979 }, { "epoch": 0.64, "grad_norm": 7.890898704528809, "learning_rate": 6.041685057183434e-06, "loss": 1.2532, "step": 3980 }, { "epoch": 0.64, "grad_norm": 10.021934509277344, "learning_rate": 6.036897218248685e-06, "loss": 1.2941, "step": 3981 }, { "epoch": 0.64, "grad_norm": 7.131109714508057, "learning_rate": 6.0321104568230925e-06, "loss": 1.2638, "step": 3982 }, { "epoch": 0.64, "grad_norm": 9.557040214538574, "learning_rate": 6.027324774208102e-06, "loss": 1.5798, "step": 3983 }, { "epoch": 0.64, "grad_norm": 7.381633758544922, "learning_rate": 6.0225401717048784e-06, "loss": 1.4774, "step": 3984 }, { "epoch": 0.64, "grad_norm": 8.212599754333496, "learning_rate": 6.017756650614277e-06, "loss": 1.2405, "step": 3985 }, { "epoch": 0.64, "grad_norm": 8.071678161621094, "learning_rate": 6.012974212236866e-06, "loss": 1.3224, "step": 3986 }, { "epoch": 0.64, "grad_norm": 7.482896327972412, "learning_rate": 6.0081928578729235e-06, "loss": 1.7192, "step": 3987 }, { "epoch": 0.64, "grad_norm": 7.999155044555664, "learning_rate": 6.003412588822426e-06, "loss": 1.4187, "step": 3988 }, { "epoch": 0.64, "grad_norm": 8.61369800567627, "learning_rate": 5.998633406385064e-06, "loss": 1.2641, "step": 3989 }, { "epoch": 0.64, "grad_norm": 7.651284217834473, "learning_rate": 5.9938553118602235e-06, "loss": 1.3587, "step": 3990 }, { "epoch": 0.64, "grad_norm": 9.162314414978027, "learning_rate": 5.989078306546992e-06, "loss": 1.4294, "step": 3991 }, { "epoch": 0.64, "grad_norm": 7.677321434020996, "learning_rate": 5.984302391744175e-06, "loss": 1.3617, "step": 3992 }, { "epoch": 0.64, "grad_norm": 6.384744644165039, "learning_rate": 5.979527568750267e-06, "loss": 1.2246, "step": 3993 }, { "epoch": 0.64, "grad_norm": 8.05327320098877, "learning_rate": 5.974753838863479e-06, "loss": 1.3433, "step": 3994 }, { "epoch": 0.64, "grad_norm": 7.3184494972229, "learning_rate": 5.969981203381715e-06, "loss": 1.2809, "step": 3995 }, { "epoch": 0.64, "grad_norm": 7.541521072387695, "learning_rate": 5.965209663602578e-06, "loss": 1.2829, "step": 3996 }, { "epoch": 0.64, "grad_norm": 12.143475532531738, "learning_rate": 5.9604392208233864e-06, "loss": 1.5295, "step": 3997 }, { "epoch": 0.64, "grad_norm": 8.671375274658203, "learning_rate": 5.955669876341148e-06, "loss": 1.3807, "step": 3998 }, { "epoch": 0.64, "grad_norm": 8.537759780883789, "learning_rate": 5.9509016314525855e-06, "loss": 1.4305, "step": 3999 }, { "epoch": 0.64, "grad_norm": 8.20478343963623, "learning_rate": 5.946134487454106e-06, "loss": 1.2276, "step": 4000 }, { "epoch": 0.64, "grad_norm": 6.830082416534424, "learning_rate": 5.941368445641827e-06, "loss": 1.3967, "step": 4001 }, { "epoch": 0.64, "grad_norm": 7.792994499206543, "learning_rate": 5.936603507311569e-06, "loss": 1.3398, "step": 4002 }, { "epoch": 0.64, "grad_norm": 8.015504837036133, "learning_rate": 5.931839673758843e-06, "loss": 1.4242, "step": 4003 }, { "epoch": 0.64, "grad_norm": 6.639220714569092, "learning_rate": 5.927076946278872e-06, "loss": 1.2845, "step": 4004 }, { "epoch": 0.64, "grad_norm": 8.1324462890625, "learning_rate": 5.922315326166565e-06, "loss": 1.4235, "step": 4005 }, { "epoch": 0.64, "grad_norm": 7.566966533660889, "learning_rate": 5.917554814716535e-06, "loss": 1.397, "step": 4006 }, { "epoch": 0.65, "grad_norm": 9.816786766052246, "learning_rate": 5.912795413223104e-06, "loss": 1.5338, "step": 4007 }, { "epoch": 0.65, "grad_norm": 8.480690956115723, "learning_rate": 5.908037122980273e-06, "loss": 1.42, "step": 4008 }, { "epoch": 0.65, "grad_norm": 9.360276222229004, "learning_rate": 5.9032799452817565e-06, "loss": 1.4533, "step": 4009 }, { "epoch": 0.65, "grad_norm": 7.652730941772461, "learning_rate": 5.8985238814209585e-06, "loss": 1.4944, "step": 4010 }, { "epoch": 0.65, "grad_norm": 6.3500142097473145, "learning_rate": 5.893768932690984e-06, "loss": 1.4019, "step": 4011 }, { "epoch": 0.65, "grad_norm": 7.965738773345947, "learning_rate": 5.889015100384636e-06, "loss": 1.5535, "step": 4012 }, { "epoch": 0.65, "grad_norm": 9.143726348876953, "learning_rate": 5.884262385794403e-06, "loss": 1.5071, "step": 4013 }, { "epoch": 0.65, "grad_norm": 9.754758834838867, "learning_rate": 5.879510790212488e-06, "loss": 1.5109, "step": 4014 }, { "epoch": 0.65, "grad_norm": 6.024707794189453, "learning_rate": 5.874760314930772e-06, "loss": 1.431, "step": 4015 }, { "epoch": 0.65, "grad_norm": 7.38146448135376, "learning_rate": 5.870010961240843e-06, "loss": 1.4121, "step": 4016 }, { "epoch": 0.65, "grad_norm": 8.57936954498291, "learning_rate": 5.86526273043398e-06, "loss": 1.5113, "step": 4017 }, { "epoch": 0.65, "grad_norm": 8.415308952331543, "learning_rate": 5.860515623801155e-06, "loss": 1.4698, "step": 4018 }, { "epoch": 0.65, "grad_norm": 8.191534042358398, "learning_rate": 5.855769642633039e-06, "loss": 1.6057, "step": 4019 }, { "epoch": 0.65, "grad_norm": 7.375767230987549, "learning_rate": 5.85102478821999e-06, "loss": 1.3794, "step": 4020 }, { "epoch": 0.65, "grad_norm": 8.125309944152832, "learning_rate": 5.846281061852067e-06, "loss": 1.335, "step": 4021 }, { "epoch": 0.65, "grad_norm": 8.103516578674316, "learning_rate": 5.841538464819021e-06, "loss": 1.3423, "step": 4022 }, { "epoch": 0.65, "grad_norm": 7.869524955749512, "learning_rate": 5.836796998410286e-06, "loss": 1.4174, "step": 4023 }, { "epoch": 0.65, "grad_norm": 7.435299873352051, "learning_rate": 5.832056663915011e-06, "loss": 1.4453, "step": 4024 }, { "epoch": 0.65, "grad_norm": 8.260722160339355, "learning_rate": 5.827317462622006e-06, "loss": 1.4378, "step": 4025 }, { "epoch": 0.65, "grad_norm": 7.188547134399414, "learning_rate": 5.8225793958198075e-06, "loss": 1.3547, "step": 4026 }, { "epoch": 0.65, "grad_norm": 8.049945831298828, "learning_rate": 5.817842464796611e-06, "loss": 1.5008, "step": 4027 }, { "epoch": 0.65, "grad_norm": 10.064682006835938, "learning_rate": 5.81310667084032e-06, "loss": 1.508, "step": 4028 }, { "epoch": 0.65, "grad_norm": 7.843758583068848, "learning_rate": 5.808372015238541e-06, "loss": 1.2907, "step": 4029 }, { "epoch": 0.65, "grad_norm": 9.395615577697754, "learning_rate": 5.803638499278538e-06, "loss": 1.3932, "step": 4030 }, { "epoch": 0.65, "grad_norm": 6.900972843170166, "learning_rate": 5.798906124247301e-06, "loss": 1.3579, "step": 4031 }, { "epoch": 0.65, "grad_norm": 6.961080551147461, "learning_rate": 5.794174891431481e-06, "loss": 1.1139, "step": 4032 }, { "epoch": 0.65, "grad_norm": 7.642001152038574, "learning_rate": 5.78944480211743e-06, "loss": 1.2687, "step": 4033 }, { "epoch": 0.65, "grad_norm": 10.029869079589844, "learning_rate": 5.784715857591201e-06, "loss": 1.3321, "step": 4034 }, { "epoch": 0.65, "grad_norm": 7.744554042816162, "learning_rate": 5.779988059138511e-06, "loss": 1.4264, "step": 4035 }, { "epoch": 0.65, "grad_norm": 8.364697456359863, "learning_rate": 5.775261408044791e-06, "loss": 1.3114, "step": 4036 }, { "epoch": 0.65, "grad_norm": 8.34190559387207, "learning_rate": 5.770535905595138e-06, "loss": 1.512, "step": 4037 }, { "epoch": 0.65, "grad_norm": 7.520856857299805, "learning_rate": 5.7658115530743456e-06, "loss": 1.2016, "step": 4038 }, { "epoch": 0.65, "grad_norm": 12.749138832092285, "learning_rate": 5.761088351766908e-06, "loss": 1.5874, "step": 4039 }, { "epoch": 0.65, "grad_norm": 7.930093765258789, "learning_rate": 5.756366302956977e-06, "loss": 1.4954, "step": 4040 }, { "epoch": 0.65, "grad_norm": 7.888177871704102, "learning_rate": 5.751645407928421e-06, "loss": 1.3837, "step": 4041 }, { "epoch": 0.65, "grad_norm": 9.980905532836914, "learning_rate": 5.746925667964781e-06, "loss": 1.6304, "step": 4042 }, { "epoch": 0.65, "grad_norm": 8.1597900390625, "learning_rate": 5.742207084349274e-06, "loss": 1.4418, "step": 4043 }, { "epoch": 0.65, "grad_norm": 7.172753810882568, "learning_rate": 5.737489658364826e-06, "loss": 1.2893, "step": 4044 }, { "epoch": 0.65, "grad_norm": 8.8717679977417, "learning_rate": 5.732773391294022e-06, "loss": 1.373, "step": 4045 }, { "epoch": 0.65, "grad_norm": 7.653341770172119, "learning_rate": 5.728058284419158e-06, "loss": 1.5303, "step": 4046 }, { "epoch": 0.65, "grad_norm": 7.603499412536621, "learning_rate": 5.723344339022196e-06, "loss": 1.4148, "step": 4047 }, { "epoch": 0.65, "grad_norm": 7.577644348144531, "learning_rate": 5.7186315563847904e-06, "loss": 1.2199, "step": 4048 }, { "epoch": 0.65, "grad_norm": 6.842846393585205, "learning_rate": 5.71391993778828e-06, "loss": 1.3345, "step": 4049 }, { "epoch": 0.65, "grad_norm": 6.999796390533447, "learning_rate": 5.709209484513672e-06, "loss": 1.468, "step": 4050 }, { "epoch": 0.65, "grad_norm": 6.869265079498291, "learning_rate": 5.7045001978416805e-06, "loss": 1.3543, "step": 4051 }, { "epoch": 0.65, "grad_norm": 8.629371643066406, "learning_rate": 5.6997920790526895e-06, "loss": 1.4003, "step": 4052 }, { "epoch": 0.65, "grad_norm": 8.807666778564453, "learning_rate": 5.695085129426765e-06, "loss": 1.3938, "step": 4053 }, { "epoch": 0.65, "grad_norm": 8.761788368225098, "learning_rate": 5.690379350243661e-06, "loss": 1.3355, "step": 4054 }, { "epoch": 0.65, "grad_norm": 6.788712024688721, "learning_rate": 5.685674742782799e-06, "loss": 1.5363, "step": 4055 }, { "epoch": 0.65, "grad_norm": 8.032292366027832, "learning_rate": 5.680971308323302e-06, "loss": 1.3843, "step": 4056 }, { "epoch": 0.65, "grad_norm": 9.750985145568848, "learning_rate": 5.676269048143963e-06, "loss": 1.4948, "step": 4057 }, { "epoch": 0.65, "grad_norm": 8.674107551574707, "learning_rate": 5.671567963523257e-06, "loss": 1.346, "step": 4058 }, { "epoch": 0.65, "grad_norm": 8.656786918640137, "learning_rate": 5.666868055739341e-06, "loss": 1.4961, "step": 4059 }, { "epoch": 0.65, "grad_norm": 8.171919822692871, "learning_rate": 5.662169326070039e-06, "loss": 1.5173, "step": 4060 }, { "epoch": 0.65, "grad_norm": 7.176979064941406, "learning_rate": 5.657471775792882e-06, "loss": 1.3332, "step": 4061 }, { "epoch": 0.65, "grad_norm": 8.563918113708496, "learning_rate": 5.652775406185056e-06, "loss": 1.6542, "step": 4062 }, { "epoch": 0.65, "grad_norm": 7.606847763061523, "learning_rate": 5.648080218523438e-06, "loss": 1.4079, "step": 4063 }, { "epoch": 0.65, "grad_norm": 7.857030868530273, "learning_rate": 5.643386214084577e-06, "loss": 1.359, "step": 4064 }, { "epoch": 0.65, "grad_norm": 8.648712158203125, "learning_rate": 5.6386933941447066e-06, "loss": 1.3351, "step": 4065 }, { "epoch": 0.65, "grad_norm": 7.789175510406494, "learning_rate": 5.634001759979732e-06, "loss": 1.2901, "step": 4066 }, { "epoch": 0.65, "grad_norm": 8.145665168762207, "learning_rate": 5.629311312865243e-06, "loss": 1.3991, "step": 4067 }, { "epoch": 0.65, "grad_norm": 6.461203575134277, "learning_rate": 5.624622054076501e-06, "loss": 1.4935, "step": 4068 }, { "epoch": 0.66, "grad_norm": 8.984085083007812, "learning_rate": 5.619933984888446e-06, "loss": 1.3672, "step": 4069 }, { "epoch": 0.66, "grad_norm": 8.897239685058594, "learning_rate": 5.615247106575695e-06, "loss": 1.4418, "step": 4070 }, { "epoch": 0.66, "grad_norm": 7.966395854949951, "learning_rate": 5.610561420412541e-06, "loss": 1.1389, "step": 4071 }, { "epoch": 0.66, "grad_norm": 7.812715530395508, "learning_rate": 5.605876927672956e-06, "loss": 1.199, "step": 4072 }, { "epoch": 0.66, "grad_norm": 7.29452657699585, "learning_rate": 5.601193629630578e-06, "loss": 1.4276, "step": 4073 }, { "epoch": 0.66, "grad_norm": 7.6616950035095215, "learning_rate": 5.596511527558732e-06, "loss": 1.3409, "step": 4074 }, { "epoch": 0.66, "grad_norm": 7.212934970855713, "learning_rate": 5.59183062273041e-06, "loss": 1.2211, "step": 4075 }, { "epoch": 0.66, "grad_norm": 9.442789077758789, "learning_rate": 5.5871509164182825e-06, "loss": 1.4178, "step": 4076 }, { "epoch": 0.66, "grad_norm": 9.44754695892334, "learning_rate": 5.58247240989469e-06, "loss": 1.5861, "step": 4077 }, { "epoch": 0.66, "grad_norm": 7.406143665313721, "learning_rate": 5.5777951044316524e-06, "loss": 1.442, "step": 4078 }, { "epoch": 0.66, "grad_norm": 8.367552757263184, "learning_rate": 5.573119001300858e-06, "loss": 1.3584, "step": 4079 }, { "epoch": 0.66, "grad_norm": 8.196030616760254, "learning_rate": 5.568444101773671e-06, "loss": 1.2519, "step": 4080 }, { "epoch": 0.66, "grad_norm": 7.85637092590332, "learning_rate": 5.563770407121128e-06, "loss": 1.1653, "step": 4081 }, { "epoch": 0.66, "grad_norm": 8.25697135925293, "learning_rate": 5.559097918613938e-06, "loss": 1.2158, "step": 4082 }, { "epoch": 0.66, "grad_norm": 9.035626411437988, "learning_rate": 5.554426637522478e-06, "loss": 1.306, "step": 4083 }, { "epoch": 0.66, "grad_norm": 7.958425045013428, "learning_rate": 5.549756565116806e-06, "loss": 1.4673, "step": 4084 }, { "epoch": 0.66, "grad_norm": 9.408998489379883, "learning_rate": 5.545087702666641e-06, "loss": 1.3421, "step": 4085 }, { "epoch": 0.66, "grad_norm": 7.929865837097168, "learning_rate": 5.540420051441382e-06, "loss": 1.1163, "step": 4086 }, { "epoch": 0.66, "grad_norm": 11.114076614379883, "learning_rate": 5.535753612710091e-06, "loss": 1.4319, "step": 4087 }, { "epoch": 0.66, "grad_norm": 7.473315715789795, "learning_rate": 5.5310883877415034e-06, "loss": 1.5478, "step": 4088 }, { "epoch": 0.66, "grad_norm": 7.085943222045898, "learning_rate": 5.526424377804028e-06, "loss": 1.1677, "step": 4089 }, { "epoch": 0.66, "grad_norm": 9.409343719482422, "learning_rate": 5.521761584165739e-06, "loss": 1.3542, "step": 4090 }, { "epoch": 0.66, "grad_norm": 6.920526504516602, "learning_rate": 5.517100008094381e-06, "loss": 1.172, "step": 4091 }, { "epoch": 0.66, "grad_norm": 8.554563522338867, "learning_rate": 5.512439650857367e-06, "loss": 1.5631, "step": 4092 }, { "epoch": 0.66, "grad_norm": 9.373334884643555, "learning_rate": 5.507780513721781e-06, "loss": 1.5507, "step": 4093 }, { "epoch": 0.66, "grad_norm": 7.248819351196289, "learning_rate": 5.503122597954369e-06, "loss": 1.3652, "step": 4094 }, { "epoch": 0.66, "grad_norm": 8.940778732299805, "learning_rate": 5.498465904821561e-06, "loss": 1.6355, "step": 4095 }, { "epoch": 0.66, "grad_norm": 10.744329452514648, "learning_rate": 5.49381043558943e-06, "loss": 1.5005, "step": 4096 }, { "epoch": 0.66, "grad_norm": 7.17780065536499, "learning_rate": 5.4891561915237365e-06, "loss": 1.5104, "step": 4097 }, { "epoch": 0.66, "grad_norm": 8.721900939941406, "learning_rate": 5.484503173889899e-06, "loss": 1.4181, "step": 4098 }, { "epoch": 0.66, "grad_norm": 8.336848258972168, "learning_rate": 5.4798513839529995e-06, "loss": 1.476, "step": 4099 }, { "epoch": 0.66, "grad_norm": 8.242539405822754, "learning_rate": 5.475200822977805e-06, "loss": 1.3137, "step": 4100 }, { "epoch": 0.66, "grad_norm": 9.026008605957031, "learning_rate": 5.470551492228721e-06, "loss": 1.356, "step": 4101 }, { "epoch": 0.66, "grad_norm": 7.921029090881348, "learning_rate": 5.465903392969836e-06, "loss": 1.2311, "step": 4102 }, { "epoch": 0.66, "grad_norm": 7.928692817687988, "learning_rate": 5.4612565264649e-06, "loss": 1.2101, "step": 4103 }, { "epoch": 0.66, "grad_norm": 7.384766101837158, "learning_rate": 5.4566108939773236e-06, "loss": 1.3488, "step": 4104 }, { "epoch": 0.66, "grad_norm": 8.664082527160645, "learning_rate": 5.451966496770196e-06, "loss": 1.4556, "step": 4105 }, { "epoch": 0.66, "grad_norm": 8.311960220336914, "learning_rate": 5.447323336106251e-06, "loss": 1.1785, "step": 4106 }, { "epoch": 0.66, "grad_norm": 8.971458435058594, "learning_rate": 5.4426814132478976e-06, "loss": 1.1979, "step": 4107 }, { "epoch": 0.66, "grad_norm": 9.41463565826416, "learning_rate": 5.4380407294572065e-06, "loss": 1.5321, "step": 4108 }, { "epoch": 0.66, "grad_norm": 7.855958938598633, "learning_rate": 5.433401285995907e-06, "loss": 1.0481, "step": 4109 }, { "epoch": 0.66, "grad_norm": 9.350293159484863, "learning_rate": 5.428763084125408e-06, "loss": 1.302, "step": 4110 }, { "epoch": 0.66, "grad_norm": 8.656627655029297, "learning_rate": 5.424126125106757e-06, "loss": 1.2268, "step": 4111 }, { "epoch": 0.66, "grad_norm": 7.167073726654053, "learning_rate": 5.419490410200675e-06, "loss": 1.398, "step": 4112 }, { "epoch": 0.66, "grad_norm": 7.721221446990967, "learning_rate": 5.414855940667547e-06, "loss": 1.2982, "step": 4113 }, { "epoch": 0.66, "grad_norm": 7.79058313369751, "learning_rate": 5.410222717767415e-06, "loss": 1.2024, "step": 4114 }, { "epoch": 0.66, "grad_norm": 8.328873634338379, "learning_rate": 5.405590742759992e-06, "loss": 1.3364, "step": 4115 }, { "epoch": 0.66, "grad_norm": 7.889769077301025, "learning_rate": 5.400960016904633e-06, "loss": 1.2672, "step": 4116 }, { "epoch": 0.66, "grad_norm": 7.0915117263793945, "learning_rate": 5.396330541460368e-06, "loss": 1.2764, "step": 4117 }, { "epoch": 0.66, "grad_norm": 8.033404350280762, "learning_rate": 5.391702317685884e-06, "loss": 1.2939, "step": 4118 }, { "epoch": 0.66, "grad_norm": 8.277716636657715, "learning_rate": 5.387075346839522e-06, "loss": 1.3496, "step": 4119 }, { "epoch": 0.66, "grad_norm": 8.728103637695312, "learning_rate": 5.382449630179299e-06, "loss": 1.3593, "step": 4120 }, { "epoch": 0.66, "grad_norm": 7.926110744476318, "learning_rate": 5.377825168962867e-06, "loss": 1.1343, "step": 4121 }, { "epoch": 0.66, "grad_norm": 8.662793159484863, "learning_rate": 5.37320196444755e-06, "loss": 1.0958, "step": 4122 }, { "epoch": 0.66, "grad_norm": 7.185258865356445, "learning_rate": 5.368580017890339e-06, "loss": 1.515, "step": 4123 }, { "epoch": 0.66, "grad_norm": 7.564103603363037, "learning_rate": 5.363959330547859e-06, "loss": 1.0422, "step": 4124 }, { "epoch": 0.66, "grad_norm": 10.157236099243164, "learning_rate": 5.359339903676421e-06, "loss": 1.1359, "step": 4125 }, { "epoch": 0.66, "grad_norm": 7.281752109527588, "learning_rate": 5.354721738531965e-06, "loss": 1.4876, "step": 4126 }, { "epoch": 0.66, "grad_norm": 8.353995323181152, "learning_rate": 5.350104836370111e-06, "loss": 1.3437, "step": 4127 }, { "epoch": 0.66, "grad_norm": 10.518166542053223, "learning_rate": 5.34548919844613e-06, "loss": 1.3539, "step": 4128 }, { "epoch": 0.66, "grad_norm": 8.329898834228516, "learning_rate": 5.340874826014931e-06, "loss": 1.4207, "step": 4129 }, { "epoch": 0.66, "grad_norm": 9.236017227172852, "learning_rate": 5.336261720331112e-06, "loss": 1.4344, "step": 4130 }, { "epoch": 0.67, "grad_norm": 8.266944885253906, "learning_rate": 5.33164988264889e-06, "loss": 1.2631, "step": 4131 }, { "epoch": 0.67, "grad_norm": 8.473219871520996, "learning_rate": 5.32703931422217e-06, "loss": 1.2527, "step": 4132 }, { "epoch": 0.67, "grad_norm": 9.078926086425781, "learning_rate": 5.322430016304496e-06, "loss": 1.4315, "step": 4133 }, { "epoch": 0.67, "grad_norm": 9.104469299316406, "learning_rate": 5.3178219901490554e-06, "loss": 1.5368, "step": 4134 }, { "epoch": 0.67, "grad_norm": 7.540671348571777, "learning_rate": 5.31321523700872e-06, "loss": 1.4344, "step": 4135 }, { "epoch": 0.67, "grad_norm": 7.580283164978027, "learning_rate": 5.308609758135983e-06, "loss": 1.4789, "step": 4136 }, { "epoch": 0.67, "grad_norm": 7.481764316558838, "learning_rate": 5.304005554783015e-06, "loss": 1.3065, "step": 4137 }, { "epoch": 0.67, "grad_norm": 8.919994354248047, "learning_rate": 5.299402628201633e-06, "loss": 1.3338, "step": 4138 }, { "epoch": 0.67, "grad_norm": 8.641339302062988, "learning_rate": 5.294800979643293e-06, "loss": 1.3846, "step": 4139 }, { "epoch": 0.67, "grad_norm": 6.751736640930176, "learning_rate": 5.2902006103591285e-06, "loss": 1.1943, "step": 4140 }, { "epoch": 0.67, "grad_norm": 7.553556442260742, "learning_rate": 5.2856015215999015e-06, "loss": 1.3604, "step": 4141 }, { "epoch": 0.67, "grad_norm": 8.687135696411133, "learning_rate": 5.281003714616042e-06, "loss": 1.4664, "step": 4142 }, { "epoch": 0.67, "grad_norm": 7.833780765533447, "learning_rate": 5.2764071906576305e-06, "loss": 1.4692, "step": 4143 }, { "epoch": 0.67, "grad_norm": 7.642091274261475, "learning_rate": 5.271811950974381e-06, "loss": 1.4666, "step": 4144 }, { "epoch": 0.67, "grad_norm": 7.7136406898498535, "learning_rate": 5.267217996815682e-06, "loss": 1.2983, "step": 4145 }, { "epoch": 0.67, "grad_norm": 11.213031768798828, "learning_rate": 5.26262532943056e-06, "loss": 1.343, "step": 4146 }, { "epoch": 0.67, "grad_norm": 8.259572982788086, "learning_rate": 5.258033950067691e-06, "loss": 1.3392, "step": 4147 }, { "epoch": 0.67, "grad_norm": 8.4909086227417, "learning_rate": 5.253443859975407e-06, "loss": 1.4368, "step": 4148 }, { "epoch": 0.67, "grad_norm": 8.866713523864746, "learning_rate": 5.2488550604016764e-06, "loss": 1.5904, "step": 4149 }, { "epoch": 0.67, "grad_norm": 7.534059047698975, "learning_rate": 5.244267552594138e-06, "loss": 1.1676, "step": 4150 }, { "epoch": 0.67, "grad_norm": 8.452345848083496, "learning_rate": 5.239681337800061e-06, "loss": 1.3225, "step": 4151 }, { "epoch": 0.67, "grad_norm": 8.881861686706543, "learning_rate": 5.2350964172663725e-06, "loss": 1.353, "step": 4152 }, { "epoch": 0.67, "grad_norm": 8.618027687072754, "learning_rate": 5.230512792239643e-06, "loss": 1.3337, "step": 4153 }, { "epoch": 0.67, "grad_norm": 8.064921379089355, "learning_rate": 5.225930463966089e-06, "loss": 1.2942, "step": 4154 }, { "epoch": 0.67, "grad_norm": 7.479403495788574, "learning_rate": 5.221349433691584e-06, "loss": 1.1987, "step": 4155 }, { "epoch": 0.67, "grad_norm": 8.400806427001953, "learning_rate": 5.21676970266164e-06, "loss": 1.3071, "step": 4156 }, { "epoch": 0.67, "grad_norm": 7.676933288574219, "learning_rate": 5.2121912721214175e-06, "loss": 1.5632, "step": 4157 }, { "epoch": 0.67, "grad_norm": 7.959653854370117, "learning_rate": 5.207614143315729e-06, "loss": 1.2627, "step": 4158 }, { "epoch": 0.67, "grad_norm": 8.73315715789795, "learning_rate": 5.2030383174890175e-06, "loss": 1.1814, "step": 4159 }, { "epoch": 0.67, "grad_norm": 8.252829551696777, "learning_rate": 5.198463795885391e-06, "loss": 1.3284, "step": 4160 }, { "epoch": 0.67, "grad_norm": 8.07414722442627, "learning_rate": 5.193890579748595e-06, "loss": 1.4513, "step": 4161 }, { "epoch": 0.67, "grad_norm": 9.453277587890625, "learning_rate": 5.189318670322016e-06, "loss": 1.3917, "step": 4162 }, { "epoch": 0.67, "grad_norm": 7.424759387969971, "learning_rate": 5.1847480688486906e-06, "loss": 1.3873, "step": 4163 }, { "epoch": 0.67, "grad_norm": 7.861403465270996, "learning_rate": 5.180178776571297e-06, "loss": 1.5777, "step": 4164 }, { "epoch": 0.67, "grad_norm": 8.655701637268066, "learning_rate": 5.1756107947321575e-06, "loss": 1.3951, "step": 4165 }, { "epoch": 0.67, "grad_norm": 7.471983909606934, "learning_rate": 5.171044124573239e-06, "loss": 1.3734, "step": 4166 }, { "epoch": 0.67, "grad_norm": 8.079511642456055, "learning_rate": 5.166478767336154e-06, "loss": 1.3786, "step": 4167 }, { "epoch": 0.67, "grad_norm": 7.678864479064941, "learning_rate": 5.161914724262155e-06, "loss": 1.3251, "step": 4168 }, { "epoch": 0.67, "grad_norm": 7.567127704620361, "learning_rate": 5.157351996592137e-06, "loss": 1.5068, "step": 4169 }, { "epoch": 0.67, "grad_norm": 8.19806957244873, "learning_rate": 5.152790585566637e-06, "loss": 1.4049, "step": 4170 }, { "epoch": 0.67, "grad_norm": 8.618162155151367, "learning_rate": 5.148230492425838e-06, "loss": 1.3297, "step": 4171 }, { "epoch": 0.67, "grad_norm": 7.735912799835205, "learning_rate": 5.143671718409562e-06, "loss": 1.6144, "step": 4172 }, { "epoch": 0.67, "grad_norm": 8.948659896850586, "learning_rate": 5.13911426475727e-06, "loss": 1.4076, "step": 4173 }, { "epoch": 0.67, "grad_norm": 7.005580425262451, "learning_rate": 5.134558132708067e-06, "loss": 1.5194, "step": 4174 }, { "epoch": 0.67, "grad_norm": 6.785656452178955, "learning_rate": 5.130003323500701e-06, "loss": 1.2267, "step": 4175 }, { "epoch": 0.67, "grad_norm": 7.9047369956970215, "learning_rate": 5.125449838373554e-06, "loss": 1.5706, "step": 4176 }, { "epoch": 0.67, "grad_norm": 8.58803653717041, "learning_rate": 5.120897678564652e-06, "loss": 1.2092, "step": 4177 }, { "epoch": 0.67, "grad_norm": 7.724248886108398, "learning_rate": 5.1163468453116616e-06, "loss": 1.3887, "step": 4178 }, { "epoch": 0.67, "grad_norm": 6.667989730834961, "learning_rate": 5.111797339851886e-06, "loss": 1.2417, "step": 4179 }, { "epoch": 0.67, "grad_norm": 9.799051284790039, "learning_rate": 5.1072491634222685e-06, "loss": 1.5691, "step": 4180 }, { "epoch": 0.67, "grad_norm": 9.666064262390137, "learning_rate": 5.102702317259391e-06, "loss": 1.4006, "step": 4181 }, { "epoch": 0.67, "grad_norm": 8.068839073181152, "learning_rate": 5.098156802599477e-06, "loss": 1.3551, "step": 4182 }, { "epoch": 0.67, "grad_norm": 8.844252586364746, "learning_rate": 5.093612620678379e-06, "loss": 1.5971, "step": 4183 }, { "epoch": 0.67, "grad_norm": 7.294127464294434, "learning_rate": 5.089069772731598e-06, "loss": 1.2914, "step": 4184 }, { "epoch": 0.67, "grad_norm": 8.221867561340332, "learning_rate": 5.0845282599942645e-06, "loss": 1.3706, "step": 4185 }, { "epoch": 0.67, "grad_norm": 8.069480895996094, "learning_rate": 5.079988083701151e-06, "loss": 1.2175, "step": 4186 }, { "epoch": 0.67, "grad_norm": 9.135502815246582, "learning_rate": 5.075449245086661e-06, "loss": 1.2086, "step": 4187 }, { "epoch": 0.67, "grad_norm": 6.66193962097168, "learning_rate": 5.070911745384842e-06, "loss": 1.4524, "step": 4188 }, { "epoch": 0.67, "grad_norm": 7.994013786315918, "learning_rate": 5.066375585829371e-06, "loss": 1.4219, "step": 4189 }, { "epoch": 0.67, "grad_norm": 6.8658766746521, "learning_rate": 5.061840767653562e-06, "loss": 1.4952, "step": 4190 }, { "epoch": 0.67, "grad_norm": 7.002236843109131, "learning_rate": 5.057307292090366e-06, "loss": 1.5109, "step": 4191 }, { "epoch": 0.67, "grad_norm": 7.890289783477783, "learning_rate": 5.05277516037237e-06, "loss": 1.3613, "step": 4192 }, { "epoch": 0.67, "grad_norm": 8.233662605285645, "learning_rate": 5.04824437373179e-06, "loss": 1.3578, "step": 4193 }, { "epoch": 0.68, "grad_norm": 7.560840129852295, "learning_rate": 5.0437149334004835e-06, "loss": 1.3343, "step": 4194 }, { "epoch": 0.68, "grad_norm": 8.412328720092773, "learning_rate": 5.039186840609935e-06, "loss": 1.5347, "step": 4195 }, { "epoch": 0.68, "grad_norm": 7.834179401397705, "learning_rate": 5.034660096591269e-06, "loss": 1.3093, "step": 4196 }, { "epoch": 0.68, "grad_norm": 7.316079616546631, "learning_rate": 5.03013470257524e-06, "loss": 1.2441, "step": 4197 }, { "epoch": 0.68, "grad_norm": 7.591643333435059, "learning_rate": 5.02561065979223e-06, "loss": 1.3601, "step": 4198 }, { "epoch": 0.68, "grad_norm": 8.003488540649414, "learning_rate": 5.021087969472273e-06, "loss": 1.5071, "step": 4199 }, { "epoch": 0.68, "grad_norm": 7.773194313049316, "learning_rate": 5.016566632845008e-06, "loss": 1.5646, "step": 4200 }, { "epoch": 0.68, "grad_norm": 7.608649253845215, "learning_rate": 5.01204665113973e-06, "loss": 1.1707, "step": 4201 }, { "epoch": 0.68, "grad_norm": 7.701330184936523, "learning_rate": 5.0075280255853474e-06, "loss": 1.4062, "step": 4202 }, { "epoch": 0.68, "grad_norm": 7.331657409667969, "learning_rate": 5.003010757410407e-06, "loss": 1.4057, "step": 4203 }, { "epoch": 0.68, "grad_norm": 7.878892421722412, "learning_rate": 4.9984948478430995e-06, "loss": 1.2511, "step": 4204 }, { "epoch": 0.68, "grad_norm": 7.884768486022949, "learning_rate": 4.993980298111217e-06, "loss": 1.5138, "step": 4205 }, { "epoch": 0.68, "grad_norm": 9.43010425567627, "learning_rate": 4.989467109442217e-06, "loss": 1.538, "step": 4206 }, { "epoch": 0.68, "grad_norm": 8.013144493103027, "learning_rate": 4.984955283063154e-06, "loss": 1.4686, "step": 4207 }, { "epoch": 0.68, "grad_norm": 13.133132934570312, "learning_rate": 4.980444820200731e-06, "loss": 1.2628, "step": 4208 }, { "epoch": 0.68, "grad_norm": 8.377537727355957, "learning_rate": 4.975935722081283e-06, "loss": 1.4637, "step": 4209 }, { "epoch": 0.68, "grad_norm": 7.537102222442627, "learning_rate": 4.971427989930756e-06, "loss": 1.3737, "step": 4210 }, { "epoch": 0.68, "grad_norm": 8.808133125305176, "learning_rate": 4.96692162497475e-06, "loss": 1.5497, "step": 4211 }, { "epoch": 0.68, "grad_norm": 9.997045516967773, "learning_rate": 4.962416628438466e-06, "loss": 1.4781, "step": 4212 }, { "epoch": 0.68, "grad_norm": 8.718497276306152, "learning_rate": 4.957913001546748e-06, "loss": 1.4755, "step": 4213 }, { "epoch": 0.68, "grad_norm": 10.089585304260254, "learning_rate": 4.9534107455240764e-06, "loss": 1.1625, "step": 4214 }, { "epoch": 0.68, "grad_norm": 10.538310050964355, "learning_rate": 4.948909861594533e-06, "loss": 1.3488, "step": 4215 }, { "epoch": 0.68, "grad_norm": 7.8751749992370605, "learning_rate": 4.944410350981856e-06, "loss": 1.3415, "step": 4216 }, { "epoch": 0.68, "grad_norm": 8.336969375610352, "learning_rate": 4.939912214909387e-06, "loss": 1.3492, "step": 4217 }, { "epoch": 0.68, "grad_norm": 7.970378875732422, "learning_rate": 4.9354154546001006e-06, "loss": 1.3334, "step": 4218 }, { "epoch": 0.68, "grad_norm": 10.041936874389648, "learning_rate": 4.9309200712766115e-06, "loss": 1.5186, "step": 4219 }, { "epoch": 0.68, "grad_norm": 9.320059776306152, "learning_rate": 4.926426066161133e-06, "loss": 1.1714, "step": 4220 }, { "epoch": 0.68, "grad_norm": 8.69690227508545, "learning_rate": 4.921933440475531e-06, "loss": 1.4501, "step": 4221 }, { "epoch": 0.68, "grad_norm": 8.919734001159668, "learning_rate": 4.917442195441282e-06, "loss": 1.2356, "step": 4222 }, { "epoch": 0.68, "grad_norm": 9.403884887695312, "learning_rate": 4.912952332279479e-06, "loss": 1.349, "step": 4223 }, { "epoch": 0.68, "grad_norm": 8.238199234008789, "learning_rate": 4.9084638522108666e-06, "loss": 1.4876, "step": 4224 }, { "epoch": 0.68, "grad_norm": 7.607275009155273, "learning_rate": 4.903976756455777e-06, "loss": 1.245, "step": 4225 }, { "epoch": 0.68, "grad_norm": 8.176192283630371, "learning_rate": 4.8994910462342006e-06, "loss": 1.3372, "step": 4226 }, { "epoch": 0.68, "grad_norm": 9.57279109954834, "learning_rate": 4.895006722765734e-06, "loss": 1.5398, "step": 4227 }, { "epoch": 0.68, "grad_norm": 8.68606948852539, "learning_rate": 4.890523787269588e-06, "loss": 1.3852, "step": 4228 }, { "epoch": 0.68, "grad_norm": 8.472431182861328, "learning_rate": 4.88604224096462e-06, "loss": 1.4421, "step": 4229 }, { "epoch": 0.68, "grad_norm": 9.018608093261719, "learning_rate": 4.881562085069283e-06, "loss": 1.4301, "step": 4230 }, { "epoch": 0.68, "grad_norm": 7.447325229644775, "learning_rate": 4.877083320801675e-06, "loss": 1.1003, "step": 4231 }, { "epoch": 0.68, "grad_norm": 8.548877716064453, "learning_rate": 4.872605949379509e-06, "loss": 1.5574, "step": 4232 }, { "epoch": 0.68, "grad_norm": 7.453362941741943, "learning_rate": 4.8681299720201e-06, "loss": 1.4317, "step": 4233 }, { "epoch": 0.68, "grad_norm": 6.902675151824951, "learning_rate": 4.86365538994042e-06, "loss": 1.0754, "step": 4234 }, { "epoch": 0.68, "grad_norm": 7.976073741912842, "learning_rate": 4.859182204357024e-06, "loss": 1.5569, "step": 4235 }, { "epoch": 0.68, "grad_norm": 7.67423677444458, "learning_rate": 4.854710416486117e-06, "loss": 1.4562, "step": 4236 }, { "epoch": 0.68, "grad_norm": 8.119131088256836, "learning_rate": 4.850240027543509e-06, "loss": 1.3116, "step": 4237 }, { "epoch": 0.68, "grad_norm": 8.699703216552734, "learning_rate": 4.845771038744632e-06, "loss": 1.2447, "step": 4238 }, { "epoch": 0.68, "grad_norm": 9.25011920928955, "learning_rate": 4.8413034513045435e-06, "loss": 1.4809, "step": 4239 }, { "epoch": 0.68, "grad_norm": 9.476512908935547, "learning_rate": 4.8368372664379e-06, "loss": 1.2205, "step": 4240 }, { "epoch": 0.68, "grad_norm": 9.218154907226562, "learning_rate": 4.832372485359006e-06, "loss": 1.3942, "step": 4241 }, { "epoch": 0.68, "grad_norm": 7.240344524383545, "learning_rate": 4.827909109281763e-06, "loss": 1.4478, "step": 4242 }, { "epoch": 0.68, "grad_norm": 8.312174797058105, "learning_rate": 4.823447139419699e-06, "loss": 1.4451, "step": 4243 }, { "epoch": 0.68, "grad_norm": 7.75718355178833, "learning_rate": 4.818986576985962e-06, "loss": 1.3699, "step": 4244 }, { "epoch": 0.68, "grad_norm": 7.340720176696777, "learning_rate": 4.814527423193299e-06, "loss": 1.4049, "step": 4245 }, { "epoch": 0.68, "grad_norm": 7.610511779785156, "learning_rate": 4.810069679254102e-06, "loss": 1.1527, "step": 4246 }, { "epoch": 0.68, "grad_norm": 8.021452903747559, "learning_rate": 4.805613346380362e-06, "loss": 1.2688, "step": 4247 }, { "epoch": 0.68, "grad_norm": 7.930950164794922, "learning_rate": 4.801158425783689e-06, "loss": 1.5332, "step": 4248 }, { "epoch": 0.68, "grad_norm": 9.592677116394043, "learning_rate": 4.796704918675309e-06, "loss": 1.4884, "step": 4249 }, { "epoch": 0.68, "grad_norm": 7.3128981590271, "learning_rate": 4.792252826266067e-06, "loss": 1.0898, "step": 4250 }, { "epoch": 0.68, "grad_norm": 7.398237228393555, "learning_rate": 4.787802149766421e-06, "loss": 1.2885, "step": 4251 }, { "epoch": 0.68, "grad_norm": 7.300411701202393, "learning_rate": 4.783352890386443e-06, "loss": 1.4503, "step": 4252 }, { "epoch": 0.68, "grad_norm": 6.8651533126831055, "learning_rate": 4.778905049335822e-06, "loss": 1.274, "step": 4253 }, { "epoch": 0.68, "grad_norm": 9.093786239624023, "learning_rate": 4.774458627823861e-06, "loss": 1.5352, "step": 4254 }, { "epoch": 0.68, "grad_norm": 8.582598686218262, "learning_rate": 4.770013627059476e-06, "loss": 1.5297, "step": 4255 }, { "epoch": 0.69, "grad_norm": 8.600749969482422, "learning_rate": 4.765570048251195e-06, "loss": 1.3357, "step": 4256 }, { "epoch": 0.69, "grad_norm": 8.600008964538574, "learning_rate": 4.7611278926071625e-06, "loss": 1.4271, "step": 4257 }, { "epoch": 0.69, "grad_norm": 9.079302787780762, "learning_rate": 4.756687161335137e-06, "loss": 1.4452, "step": 4258 }, { "epoch": 0.69, "grad_norm": 8.263915061950684, "learning_rate": 4.752247855642486e-06, "loss": 1.4536, "step": 4259 }, { "epoch": 0.69, "grad_norm": 8.172942161560059, "learning_rate": 4.747809976736192e-06, "loss": 1.2082, "step": 4260 }, { "epoch": 0.69, "grad_norm": 8.156038284301758, "learning_rate": 4.743373525822847e-06, "loss": 1.4495, "step": 4261 }, { "epoch": 0.69, "grad_norm": 8.097326278686523, "learning_rate": 4.738938504108659e-06, "loss": 1.3157, "step": 4262 }, { "epoch": 0.69, "grad_norm": 8.874666213989258, "learning_rate": 4.7345049127994434e-06, "loss": 1.3442, "step": 4263 }, { "epoch": 0.69, "grad_norm": 8.187589645385742, "learning_rate": 4.73007275310063e-06, "loss": 1.4161, "step": 4264 }, { "epoch": 0.69, "grad_norm": 8.457716941833496, "learning_rate": 4.725642026217255e-06, "loss": 1.5523, "step": 4265 }, { "epoch": 0.69, "grad_norm": 8.021641731262207, "learning_rate": 4.721212733353969e-06, "loss": 1.3807, "step": 4266 }, { "epoch": 0.69, "grad_norm": 8.15222454071045, "learning_rate": 4.716784875715031e-06, "loss": 1.34, "step": 4267 }, { "epoch": 0.69, "grad_norm": 8.57267951965332, "learning_rate": 4.712358454504311e-06, "loss": 1.4454, "step": 4268 }, { "epoch": 0.69, "grad_norm": 6.755398750305176, "learning_rate": 4.707933470925288e-06, "loss": 1.3128, "step": 4269 }, { "epoch": 0.69, "grad_norm": 7.2559285163879395, "learning_rate": 4.703509926181048e-06, "loss": 1.5387, "step": 4270 }, { "epoch": 0.69, "grad_norm": 9.35583782196045, "learning_rate": 4.699087821474288e-06, "loss": 1.2939, "step": 4271 }, { "epoch": 0.69, "grad_norm": 8.106098175048828, "learning_rate": 4.694667158007313e-06, "loss": 1.2256, "step": 4272 }, { "epoch": 0.69, "grad_norm": 8.00272274017334, "learning_rate": 4.690247936982037e-06, "loss": 1.3534, "step": 4273 }, { "epoch": 0.69, "grad_norm": 8.946736335754395, "learning_rate": 4.68583015959998e-06, "loss": 1.1627, "step": 4274 }, { "epoch": 0.69, "grad_norm": 6.803450584411621, "learning_rate": 4.681413827062271e-06, "loss": 1.4459, "step": 4275 }, { "epoch": 0.69, "grad_norm": 8.527734756469727, "learning_rate": 4.676998940569647e-06, "loss": 1.5042, "step": 4276 }, { "epoch": 0.69, "grad_norm": 7.658292770385742, "learning_rate": 4.672585501322447e-06, "loss": 1.0801, "step": 4277 }, { "epoch": 0.69, "grad_norm": 7.657285213470459, "learning_rate": 4.668173510520622e-06, "loss": 1.5623, "step": 4278 }, { "epoch": 0.69, "grad_norm": 8.321723937988281, "learning_rate": 4.663762969363725e-06, "loss": 1.3906, "step": 4279 }, { "epoch": 0.69, "grad_norm": 8.904733657836914, "learning_rate": 4.659353879050926e-06, "loss": 1.3616, "step": 4280 }, { "epoch": 0.69, "grad_norm": 7.808715343475342, "learning_rate": 4.654946240780981e-06, "loss": 1.4594, "step": 4281 }, { "epoch": 0.69, "grad_norm": 7.720008850097656, "learning_rate": 4.650540055752266e-06, "loss": 1.1924, "step": 4282 }, { "epoch": 0.69, "grad_norm": 7.794734954833984, "learning_rate": 4.646135325162757e-06, "loss": 1.2084, "step": 4283 }, { "epoch": 0.69, "grad_norm": 6.992603302001953, "learning_rate": 4.641732050210032e-06, "loss": 1.1487, "step": 4284 }, { "epoch": 0.69, "grad_norm": 7.390432357788086, "learning_rate": 4.637330232091287e-06, "loss": 1.2865, "step": 4285 }, { "epoch": 0.69, "grad_norm": 8.053306579589844, "learning_rate": 4.632929872003301e-06, "loss": 1.2136, "step": 4286 }, { "epoch": 0.69, "grad_norm": 7.345661640167236, "learning_rate": 4.6285309711424706e-06, "loss": 1.2827, "step": 4287 }, { "epoch": 0.69, "grad_norm": 8.124411582946777, "learning_rate": 4.6241335307047905e-06, "loss": 1.4144, "step": 4288 }, { "epoch": 0.69, "grad_norm": 7.720739364624023, "learning_rate": 4.619737551885859e-06, "loss": 1.4738, "step": 4289 }, { "epoch": 0.69, "grad_norm": 8.3056001663208, "learning_rate": 4.615343035880887e-06, "loss": 1.3551, "step": 4290 }, { "epoch": 0.69, "grad_norm": 10.453058242797852, "learning_rate": 4.610949983884667e-06, "loss": 1.5839, "step": 4291 }, { "epoch": 0.69, "grad_norm": 7.988091468811035, "learning_rate": 4.606558397091611e-06, "loss": 1.157, "step": 4292 }, { "epoch": 0.69, "grad_norm": 9.069911003112793, "learning_rate": 4.6021682766957235e-06, "loss": 1.3551, "step": 4293 }, { "epoch": 0.69, "grad_norm": 7.656840801239014, "learning_rate": 4.5977796238906125e-06, "loss": 1.5397, "step": 4294 }, { "epoch": 0.69, "grad_norm": 8.352067947387695, "learning_rate": 4.593392439869497e-06, "loss": 1.3164, "step": 4295 }, { "epoch": 0.69, "grad_norm": 7.425095558166504, "learning_rate": 4.589006725825178e-06, "loss": 1.3407, "step": 4296 }, { "epoch": 0.69, "grad_norm": 7.576709747314453, "learning_rate": 4.5846224829500696e-06, "loss": 1.1717, "step": 4297 }, { "epoch": 0.69, "grad_norm": 7.7982048988342285, "learning_rate": 4.580239712436183e-06, "loss": 1.451, "step": 4298 }, { "epoch": 0.69, "grad_norm": 7.352289199829102, "learning_rate": 4.575858415475124e-06, "loss": 1.3575, "step": 4299 }, { "epoch": 0.69, "grad_norm": 7.583439826965332, "learning_rate": 4.5714785932581154e-06, "loss": 1.2366, "step": 4300 }, { "epoch": 0.69, "grad_norm": 9.953203201293945, "learning_rate": 4.567100246975953e-06, "loss": 1.612, "step": 4301 }, { "epoch": 0.69, "grad_norm": 7.9307355880737305, "learning_rate": 4.562723377819046e-06, "loss": 1.3769, "step": 4302 }, { "epoch": 0.69, "grad_norm": 8.708372116088867, "learning_rate": 4.558347986977412e-06, "loss": 1.6002, "step": 4303 }, { "epoch": 0.69, "grad_norm": 7.868404388427734, "learning_rate": 4.553974075640639e-06, "loss": 1.4321, "step": 4304 }, { "epoch": 0.69, "grad_norm": 9.705328941345215, "learning_rate": 4.5496016449979455e-06, "loss": 1.5794, "step": 4305 }, { "epoch": 0.69, "grad_norm": 6.8156304359436035, "learning_rate": 4.5452306962381176e-06, "loss": 1.3284, "step": 4306 }, { "epoch": 0.69, "grad_norm": 8.596562385559082, "learning_rate": 4.5408612305495535e-06, "loss": 1.4356, "step": 4307 }, { "epoch": 0.69, "grad_norm": 7.520486354827881, "learning_rate": 4.5364932491202576e-06, "loss": 1.2104, "step": 4308 }, { "epoch": 0.69, "grad_norm": 8.683833122253418, "learning_rate": 4.532126753137802e-06, "loss": 1.3054, "step": 4309 }, { "epoch": 0.69, "grad_norm": 7.0151262283325195, "learning_rate": 4.527761743789391e-06, "loss": 1.3838, "step": 4310 }, { "epoch": 0.69, "grad_norm": 7.417459011077881, "learning_rate": 4.52339822226179e-06, "loss": 1.2662, "step": 4311 }, { "epoch": 0.69, "grad_norm": 6.994936466217041, "learning_rate": 4.519036189741386e-06, "loss": 1.2036, "step": 4312 }, { "epoch": 0.69, "grad_norm": 8.771870613098145, "learning_rate": 4.514675647414153e-06, "loss": 1.2807, "step": 4313 }, { "epoch": 0.69, "grad_norm": 8.728093147277832, "learning_rate": 4.510316596465647e-06, "loss": 1.4672, "step": 4314 }, { "epoch": 0.69, "grad_norm": 9.74901008605957, "learning_rate": 4.505959038081043e-06, "loss": 1.4041, "step": 4315 }, { "epoch": 0.69, "grad_norm": 8.018800735473633, "learning_rate": 4.501602973445082e-06, "loss": 1.2582, "step": 4316 }, { "epoch": 0.69, "grad_norm": 8.982218742370605, "learning_rate": 4.497248403742126e-06, "loss": 1.2428, "step": 4317 }, { "epoch": 0.7, "grad_norm": 10.257996559143066, "learning_rate": 4.492895330156119e-06, "loss": 1.3674, "step": 4318 }, { "epoch": 0.7, "grad_norm": 8.287434577941895, "learning_rate": 4.488543753870583e-06, "loss": 1.0984, "step": 4319 }, { "epoch": 0.7, "grad_norm": 7.569572448730469, "learning_rate": 4.484193676068665e-06, "loss": 1.3234, "step": 4320 }, { "epoch": 0.7, "grad_norm": 8.348264694213867, "learning_rate": 4.479845097933072e-06, "loss": 1.3077, "step": 4321 }, { "epoch": 0.7, "grad_norm": 8.018633842468262, "learning_rate": 4.475498020646128e-06, "loss": 1.2709, "step": 4322 }, { "epoch": 0.7, "grad_norm": 7.665100574493408, "learning_rate": 4.47115244538974e-06, "loss": 1.4126, "step": 4323 }, { "epoch": 0.7, "grad_norm": 8.250825881958008, "learning_rate": 4.466808373345396e-06, "loss": 1.4182, "step": 4324 }, { "epoch": 0.7, "grad_norm": 9.294035911560059, "learning_rate": 4.462465805694194e-06, "loss": 1.2372, "step": 4325 }, { "epoch": 0.7, "grad_norm": 7.448025703430176, "learning_rate": 4.4581247436168105e-06, "loss": 1.0548, "step": 4326 }, { "epoch": 0.7, "grad_norm": 10.640067100524902, "learning_rate": 4.453785188293517e-06, "loss": 1.3879, "step": 4327 }, { "epoch": 0.7, "grad_norm": 8.80362319946289, "learning_rate": 4.4494471409041785e-06, "loss": 1.3031, "step": 4328 }, { "epoch": 0.7, "grad_norm": 8.121140480041504, "learning_rate": 4.4451106026282335e-06, "loss": 1.2336, "step": 4329 }, { "epoch": 0.7, "grad_norm": 7.922790050506592, "learning_rate": 4.440775574644736e-06, "loss": 1.4297, "step": 4330 }, { "epoch": 0.7, "grad_norm": 8.438445091247559, "learning_rate": 4.436442058132311e-06, "loss": 1.48, "step": 4331 }, { "epoch": 0.7, "grad_norm": 13.358110427856445, "learning_rate": 4.4321100542691775e-06, "loss": 1.3473, "step": 4332 }, { "epoch": 0.7, "grad_norm": 8.099849700927734, "learning_rate": 4.427779564233148e-06, "loss": 1.602, "step": 4333 }, { "epoch": 0.7, "grad_norm": 8.79731273651123, "learning_rate": 4.423450589201607e-06, "loss": 1.3735, "step": 4334 }, { "epoch": 0.7, "grad_norm": 11.224164009094238, "learning_rate": 4.4191231303515504e-06, "loss": 1.4039, "step": 4335 }, { "epoch": 0.7, "grad_norm": 7.188910961151123, "learning_rate": 4.414797188859546e-06, "loss": 1.0648, "step": 4336 }, { "epoch": 0.7, "grad_norm": 7.1417694091796875, "learning_rate": 4.410472765901755e-06, "loss": 1.3494, "step": 4337 }, { "epoch": 0.7, "grad_norm": 9.124141693115234, "learning_rate": 4.406149862653928e-06, "loss": 1.3514, "step": 4338 }, { "epoch": 0.7, "grad_norm": 8.514199256896973, "learning_rate": 4.401828480291384e-06, "loss": 1.2805, "step": 4339 }, { "epoch": 0.7, "grad_norm": 9.456389427185059, "learning_rate": 4.397508619989058e-06, "loss": 1.3794, "step": 4340 }, { "epoch": 0.7, "grad_norm": 7.26619291305542, "learning_rate": 4.393190282921452e-06, "loss": 1.1958, "step": 4341 }, { "epoch": 0.7, "grad_norm": 8.843596458435059, "learning_rate": 4.388873470262657e-06, "loss": 1.2704, "step": 4342 }, { "epoch": 0.7, "grad_norm": 9.255080223083496, "learning_rate": 4.384558183186354e-06, "loss": 1.4615, "step": 4343 }, { "epoch": 0.7, "grad_norm": 7.921366214752197, "learning_rate": 4.380244422865795e-06, "loss": 1.2259, "step": 4344 }, { "epoch": 0.7, "grad_norm": 8.872203826904297, "learning_rate": 4.3759321904738395e-06, "loss": 1.6071, "step": 4345 }, { "epoch": 0.7, "grad_norm": 9.278539657592773, "learning_rate": 4.371621487182916e-06, "loss": 1.4332, "step": 4346 }, { "epoch": 0.7, "grad_norm": 8.55385684967041, "learning_rate": 4.367312314165042e-06, "loss": 1.4011, "step": 4347 }, { "epoch": 0.7, "grad_norm": 9.458539962768555, "learning_rate": 4.36300467259182e-06, "loss": 1.3582, "step": 4348 }, { "epoch": 0.7, "grad_norm": 8.400250434875488, "learning_rate": 4.358698563634424e-06, "loss": 1.2398, "step": 4349 }, { "epoch": 0.7, "grad_norm": 8.653449058532715, "learning_rate": 4.354393988463632e-06, "loss": 1.2917, "step": 4350 }, { "epoch": 0.7, "grad_norm": 7.903552532196045, "learning_rate": 4.350090948249791e-06, "loss": 1.1391, "step": 4351 }, { "epoch": 0.7, "grad_norm": 10.48428726196289, "learning_rate": 4.345789444162835e-06, "loss": 1.592, "step": 4352 }, { "epoch": 0.7, "grad_norm": 8.300301551818848, "learning_rate": 4.341489477372278e-06, "loss": 1.2066, "step": 4353 }, { "epoch": 0.7, "grad_norm": 8.320935249328613, "learning_rate": 4.337191049047219e-06, "loss": 1.3601, "step": 4354 }, { "epoch": 0.7, "grad_norm": 7.2961649894714355, "learning_rate": 4.332894160356337e-06, "loss": 1.315, "step": 4355 }, { "epoch": 0.7, "grad_norm": 8.369087219238281, "learning_rate": 4.3285988124678916e-06, "loss": 1.3775, "step": 4356 }, { "epoch": 0.7, "grad_norm": 8.26858901977539, "learning_rate": 4.324305006549725e-06, "loss": 1.1933, "step": 4357 }, { "epoch": 0.7, "grad_norm": 8.04784107208252, "learning_rate": 4.320012743769259e-06, "loss": 1.1625, "step": 4358 }, { "epoch": 0.7, "grad_norm": 9.62668514251709, "learning_rate": 4.3157220252934975e-06, "loss": 1.2989, "step": 4359 }, { "epoch": 0.7, "grad_norm": 9.189210891723633, "learning_rate": 4.311432852289024e-06, "loss": 1.5111, "step": 4360 }, { "epoch": 0.7, "grad_norm": 7.281806468963623, "learning_rate": 4.3071452259219995e-06, "loss": 1.1809, "step": 4361 }, { "epoch": 0.7, "grad_norm": 7.634232521057129, "learning_rate": 4.302859147358168e-06, "loss": 1.2788, "step": 4362 }, { "epoch": 0.7, "grad_norm": 8.429532051086426, "learning_rate": 4.298574617762851e-06, "loss": 1.1638, "step": 4363 }, { "epoch": 0.7, "grad_norm": 8.243349075317383, "learning_rate": 4.294291638300947e-06, "loss": 1.4897, "step": 4364 }, { "epoch": 0.7, "grad_norm": 8.15941333770752, "learning_rate": 4.290010210136936e-06, "loss": 1.3811, "step": 4365 }, { "epoch": 0.7, "grad_norm": 7.800127983093262, "learning_rate": 4.2857303344348765e-06, "loss": 1.3605, "step": 4366 }, { "epoch": 0.7, "grad_norm": 7.760676860809326, "learning_rate": 4.281452012358401e-06, "loss": 1.4752, "step": 4367 }, { "epoch": 0.7, "grad_norm": 9.277925491333008, "learning_rate": 4.277175245070725e-06, "loss": 1.5054, "step": 4368 }, { "epoch": 0.7, "grad_norm": 8.486105918884277, "learning_rate": 4.272900033734635e-06, "loss": 1.4292, "step": 4369 }, { "epoch": 0.7, "grad_norm": 7.828988075256348, "learning_rate": 4.268626379512502e-06, "loss": 1.4418, "step": 4370 }, { "epoch": 0.7, "grad_norm": 7.854024410247803, "learning_rate": 4.264354283566266e-06, "loss": 1.4075, "step": 4371 }, { "epoch": 0.7, "grad_norm": 7.859005451202393, "learning_rate": 4.260083747057446e-06, "loss": 1.1478, "step": 4372 }, { "epoch": 0.7, "grad_norm": 9.793822288513184, "learning_rate": 4.255814771147142e-06, "loss": 1.2616, "step": 4373 }, { "epoch": 0.7, "grad_norm": 8.604144096374512, "learning_rate": 4.2515473569960245e-06, "loss": 1.3375, "step": 4374 }, { "epoch": 0.7, "grad_norm": 9.501699447631836, "learning_rate": 4.247281505764337e-06, "loss": 1.334, "step": 4375 }, { "epoch": 0.7, "grad_norm": 8.406108856201172, "learning_rate": 4.243017218611905e-06, "loss": 1.3435, "step": 4376 }, { "epoch": 0.7, "grad_norm": 9.083771705627441, "learning_rate": 4.238754496698125e-06, "loss": 1.4077, "step": 4377 }, { "epoch": 0.7, "grad_norm": 7.543388843536377, "learning_rate": 4.234493341181966e-06, "loss": 1.3244, "step": 4378 }, { "epoch": 0.7, "grad_norm": 8.078675270080566, "learning_rate": 4.230233753221975e-06, "loss": 1.3808, "step": 4379 }, { "epoch": 0.71, "grad_norm": 9.348905563354492, "learning_rate": 4.22597573397627e-06, "loss": 1.7228, "step": 4380 }, { "epoch": 0.71, "grad_norm": 7.864398002624512, "learning_rate": 4.221719284602544e-06, "loss": 1.2839, "step": 4381 }, { "epoch": 0.71, "grad_norm": 8.05014705657959, "learning_rate": 4.217464406258065e-06, "loss": 1.1747, "step": 4382 }, { "epoch": 0.71, "grad_norm": 8.834342956542969, "learning_rate": 4.213211100099664e-06, "loss": 1.5716, "step": 4383 }, { "epoch": 0.71, "grad_norm": 8.26952075958252, "learning_rate": 4.208959367283766e-06, "loss": 1.2405, "step": 4384 }, { "epoch": 0.71, "grad_norm": 8.490957260131836, "learning_rate": 4.204709208966342e-06, "loss": 1.5872, "step": 4385 }, { "epoch": 0.71, "grad_norm": 8.227259635925293, "learning_rate": 4.200460626302952e-06, "loss": 1.2514, "step": 4386 }, { "epoch": 0.71, "grad_norm": 8.108805656433105, "learning_rate": 4.196213620448724e-06, "loss": 1.185, "step": 4387 }, { "epoch": 0.71, "grad_norm": 7.525509834289551, "learning_rate": 4.1919681925583485e-06, "loss": 1.199, "step": 4388 }, { "epoch": 0.71, "grad_norm": 9.108528137207031, "learning_rate": 4.187724343786109e-06, "loss": 1.3881, "step": 4389 }, { "epoch": 0.71, "grad_norm": 8.176563262939453, "learning_rate": 4.18348207528583e-06, "loss": 1.0689, "step": 4390 }, { "epoch": 0.71, "grad_norm": 8.907876968383789, "learning_rate": 4.179241388210936e-06, "loss": 1.2966, "step": 4391 }, { "epoch": 0.71, "grad_norm": 6.483551025390625, "learning_rate": 4.175002283714397e-06, "loss": 1.1249, "step": 4392 }, { "epoch": 0.71, "grad_norm": 6.4934611320495605, "learning_rate": 4.170764762948762e-06, "loss": 1.2121, "step": 4393 }, { "epoch": 0.71, "grad_norm": 7.190771579742432, "learning_rate": 4.166528827066162e-06, "loss": 1.7348, "step": 4394 }, { "epoch": 0.71, "grad_norm": 8.607808113098145, "learning_rate": 4.1622944772182695e-06, "loss": 1.3445, "step": 4395 }, { "epoch": 0.71, "grad_norm": 7.616454124450684, "learning_rate": 4.158061714556358e-06, "loss": 1.4652, "step": 4396 }, { "epoch": 0.71, "grad_norm": 8.175337791442871, "learning_rate": 4.153830540231243e-06, "loss": 1.482, "step": 4397 }, { "epoch": 0.71, "grad_norm": 9.122979164123535, "learning_rate": 4.149600955393316e-06, "loss": 1.1737, "step": 4398 }, { "epoch": 0.71, "grad_norm": 8.819297790527344, "learning_rate": 4.14537296119255e-06, "loss": 1.1549, "step": 4399 }, { "epoch": 0.71, "grad_norm": 8.168839454650879, "learning_rate": 4.14114655877846e-06, "loss": 1.1622, "step": 4400 }, { "epoch": 0.71, "grad_norm": 7.442788124084473, "learning_rate": 4.136921749300159e-06, "loss": 1.2472, "step": 4401 }, { "epoch": 0.71, "grad_norm": 8.018179893493652, "learning_rate": 4.132698533906296e-06, "loss": 1.1876, "step": 4402 }, { "epoch": 0.71, "grad_norm": 8.584211349487305, "learning_rate": 4.128476913745103e-06, "loss": 1.3077, "step": 4403 }, { "epoch": 0.71, "grad_norm": 8.193556785583496, "learning_rate": 4.124256889964388e-06, "loss": 1.4252, "step": 4404 }, { "epoch": 0.71, "grad_norm": 9.666267395019531, "learning_rate": 4.1200384637114965e-06, "loss": 1.2863, "step": 4405 }, { "epoch": 0.71, "grad_norm": 9.19701099395752, "learning_rate": 4.115821636133368e-06, "loss": 1.2163, "step": 4406 }, { "epoch": 0.71, "grad_norm": 8.687453269958496, "learning_rate": 4.111606408376496e-06, "loss": 1.1414, "step": 4407 }, { "epoch": 0.71, "grad_norm": 8.088698387145996, "learning_rate": 4.1073927815869276e-06, "loss": 1.1426, "step": 4408 }, { "epoch": 0.71, "grad_norm": 7.995748043060303, "learning_rate": 4.1031807569102996e-06, "loss": 1.1834, "step": 4409 }, { "epoch": 0.71, "grad_norm": 8.240252494812012, "learning_rate": 4.098970335491786e-06, "loss": 1.3942, "step": 4410 }, { "epoch": 0.71, "grad_norm": 8.640181541442871, "learning_rate": 4.094761518476149e-06, "loss": 1.3922, "step": 4411 }, { "epoch": 0.71, "grad_norm": 8.372856140136719, "learning_rate": 4.0905543070077036e-06, "loss": 1.1078, "step": 4412 }, { "epoch": 0.71, "grad_norm": 8.15580940246582, "learning_rate": 4.086348702230318e-06, "loss": 1.3106, "step": 4413 }, { "epoch": 0.71, "grad_norm": 7.458577632904053, "learning_rate": 4.0821447052874475e-06, "loss": 1.2484, "step": 4414 }, { "epoch": 0.71, "grad_norm": 8.493745803833008, "learning_rate": 4.077942317322082e-06, "loss": 1.3917, "step": 4415 }, { "epoch": 0.71, "grad_norm": 7.854189872741699, "learning_rate": 4.073741539476802e-06, "loss": 1.2923, "step": 4416 }, { "epoch": 0.71, "grad_norm": 7.321229457855225, "learning_rate": 4.069542372893735e-06, "loss": 1.5006, "step": 4417 }, { "epoch": 0.71, "grad_norm": 7.352191925048828, "learning_rate": 4.065344818714562e-06, "loss": 1.3217, "step": 4418 }, { "epoch": 0.71, "grad_norm": 8.02314281463623, "learning_rate": 4.06114887808055e-06, "loss": 1.2907, "step": 4419 }, { "epoch": 0.71, "grad_norm": 8.100839614868164, "learning_rate": 4.0569545521325e-06, "loss": 1.337, "step": 4420 }, { "epoch": 0.71, "grad_norm": 7.8414812088012695, "learning_rate": 4.052761842010797e-06, "loss": 1.419, "step": 4421 }, { "epoch": 0.71, "grad_norm": 7.774778366088867, "learning_rate": 4.0485707488553795e-06, "loss": 1.5698, "step": 4422 }, { "epoch": 0.71, "grad_norm": 8.340944290161133, "learning_rate": 4.04438127380573e-06, "loss": 1.4631, "step": 4423 }, { "epoch": 0.71, "grad_norm": 8.38426399230957, "learning_rate": 4.040193418000919e-06, "loss": 1.4035, "step": 4424 }, { "epoch": 0.71, "grad_norm": 7.321808815002441, "learning_rate": 4.036007182579552e-06, "loss": 1.2924, "step": 4425 }, { "epoch": 0.71, "grad_norm": 9.190796852111816, "learning_rate": 4.031822568679813e-06, "loss": 1.4043, "step": 4426 }, { "epoch": 0.71, "grad_norm": 8.598217964172363, "learning_rate": 4.027639577439433e-06, "loss": 1.5386, "step": 4427 }, { "epoch": 0.71, "grad_norm": 8.197354316711426, "learning_rate": 4.023458209995705e-06, "loss": 1.0974, "step": 4428 }, { "epoch": 0.71, "grad_norm": 9.095519065856934, "learning_rate": 4.019278467485483e-06, "loss": 1.6224, "step": 4429 }, { "epoch": 0.71, "grad_norm": 8.226521492004395, "learning_rate": 4.0151003510451765e-06, "loss": 1.276, "step": 4430 }, { "epoch": 0.71, "grad_norm": 7.920708656311035, "learning_rate": 4.010923861810754e-06, "loss": 1.1929, "step": 4431 }, { "epoch": 0.71, "grad_norm": 8.137483596801758, "learning_rate": 4.00674900091774e-06, "loss": 1.1784, "step": 4432 }, { "epoch": 0.71, "grad_norm": 8.6702299118042, "learning_rate": 4.002575769501218e-06, "loss": 1.0478, "step": 4433 }, { "epoch": 0.71, "grad_norm": 8.853342056274414, "learning_rate": 3.9984041686958295e-06, "loss": 1.3941, "step": 4434 }, { "epoch": 0.71, "grad_norm": 8.559539794921875, "learning_rate": 3.994234199635769e-06, "loss": 1.0717, "step": 4435 }, { "epoch": 0.71, "grad_norm": 9.620400428771973, "learning_rate": 3.990065863454792e-06, "loss": 1.2952, "step": 4436 }, { "epoch": 0.71, "grad_norm": 9.772709846496582, "learning_rate": 3.985899161286205e-06, "loss": 1.5028, "step": 4437 }, { "epoch": 0.71, "grad_norm": 7.87382173538208, "learning_rate": 3.9817340942628756e-06, "loss": 1.3888, "step": 4438 }, { "epoch": 0.71, "grad_norm": 8.844759941101074, "learning_rate": 3.977570663517222e-06, "loss": 1.2843, "step": 4439 }, { "epoch": 0.71, "grad_norm": 6.939152240753174, "learning_rate": 3.973408870181219e-06, "loss": 1.0717, "step": 4440 }, { "epoch": 0.71, "grad_norm": 7.378833770751953, "learning_rate": 3.969248715386399e-06, "loss": 1.3257, "step": 4441 }, { "epoch": 0.72, "grad_norm": 7.675825119018555, "learning_rate": 3.965090200263844e-06, "loss": 1.18, "step": 4442 }, { "epoch": 0.72, "grad_norm": 9.197813034057617, "learning_rate": 3.960933325944196e-06, "loss": 1.2698, "step": 4443 }, { "epoch": 0.72, "grad_norm": 8.232952117919922, "learning_rate": 3.956778093557646e-06, "loss": 1.2952, "step": 4444 }, { "epoch": 0.72, "grad_norm": 8.914319038391113, "learning_rate": 3.95262450423394e-06, "loss": 1.4729, "step": 4445 }, { "epoch": 0.72, "grad_norm": 10.020853042602539, "learning_rate": 3.94847255910238e-06, "loss": 1.4616, "step": 4446 }, { "epoch": 0.72, "grad_norm": 8.76673698425293, "learning_rate": 3.944322259291815e-06, "loss": 1.1926, "step": 4447 }, { "epoch": 0.72, "grad_norm": 9.007305145263672, "learning_rate": 3.940173605930653e-06, "loss": 1.3703, "step": 4448 }, { "epoch": 0.72, "grad_norm": 9.572609901428223, "learning_rate": 3.936026600146851e-06, "loss": 1.246, "step": 4449 }, { "epoch": 0.72, "grad_norm": 10.7872953414917, "learning_rate": 3.931881243067918e-06, "loss": 1.2275, "step": 4450 }, { "epoch": 0.72, "grad_norm": 8.70161247253418, "learning_rate": 3.927737535820915e-06, "loss": 1.1404, "step": 4451 }, { "epoch": 0.72, "grad_norm": 7.869673728942871, "learning_rate": 3.923595479532456e-06, "loss": 1.4174, "step": 4452 }, { "epoch": 0.72, "grad_norm": 8.380003929138184, "learning_rate": 3.919455075328706e-06, "loss": 1.139, "step": 4453 }, { "epoch": 0.72, "grad_norm": 7.309077739715576, "learning_rate": 3.915316324335377e-06, "loss": 1.4449, "step": 4454 }, { "epoch": 0.72, "grad_norm": 8.405803680419922, "learning_rate": 3.911179227677735e-06, "loss": 1.3036, "step": 4455 }, { "epoch": 0.72, "grad_norm": 8.609174728393555, "learning_rate": 3.9070437864805974e-06, "loss": 1.1939, "step": 4456 }, { "epoch": 0.72, "grad_norm": 9.845320701599121, "learning_rate": 3.902910001868329e-06, "loss": 1.3533, "step": 4457 }, { "epoch": 0.72, "grad_norm": 7.637335300445557, "learning_rate": 3.898777874964843e-06, "loss": 1.4171, "step": 4458 }, { "epoch": 0.72, "grad_norm": 9.87376594543457, "learning_rate": 3.894647406893605e-06, "loss": 1.5094, "step": 4459 }, { "epoch": 0.72, "grad_norm": 7.171720027923584, "learning_rate": 3.89051859877763e-06, "loss": 1.2162, "step": 4460 }, { "epoch": 0.72, "grad_norm": 9.17357349395752, "learning_rate": 3.886391451739476e-06, "loss": 1.3753, "step": 4461 }, { "epoch": 0.72, "grad_norm": 8.154457092285156, "learning_rate": 3.882265966901257e-06, "loss": 1.0807, "step": 4462 }, { "epoch": 0.72, "grad_norm": 8.438146591186523, "learning_rate": 3.87814214538463e-06, "loss": 1.0144, "step": 4463 }, { "epoch": 0.72, "grad_norm": 9.035799026489258, "learning_rate": 3.874019988310797e-06, "loss": 1.2196, "step": 4464 }, { "epoch": 0.72, "grad_norm": 7.397160530090332, "learning_rate": 3.869899496800524e-06, "loss": 1.3397, "step": 4465 }, { "epoch": 0.72, "grad_norm": 9.442837715148926, "learning_rate": 3.865780671974101e-06, "loss": 1.4511, "step": 4466 }, { "epoch": 0.72, "grad_norm": 6.585517883300781, "learning_rate": 3.861663514951377e-06, "loss": 1.3079, "step": 4467 }, { "epoch": 0.72, "grad_norm": 9.814226150512695, "learning_rate": 3.857548026851749e-06, "loss": 1.4756, "step": 4468 }, { "epoch": 0.72, "grad_norm": 7.710859775543213, "learning_rate": 3.853434208794151e-06, "loss": 1.241, "step": 4469 }, { "epoch": 0.72, "grad_norm": 8.233975410461426, "learning_rate": 3.8493220618970824e-06, "loss": 0.9161, "step": 4470 }, { "epoch": 0.72, "grad_norm": 8.935392379760742, "learning_rate": 3.845211587278565e-06, "loss": 1.4411, "step": 4471 }, { "epoch": 0.72, "grad_norm": 7.066764831542969, "learning_rate": 3.841102786056178e-06, "loss": 1.0666, "step": 4472 }, { "epoch": 0.72, "grad_norm": 7.365680694580078, "learning_rate": 3.836995659347045e-06, "loss": 1.1367, "step": 4473 }, { "epoch": 0.72, "grad_norm": 7.341396808624268, "learning_rate": 3.832890208267828e-06, "loss": 1.1551, "step": 4474 }, { "epoch": 0.72, "grad_norm": 9.535139083862305, "learning_rate": 3.82878643393475e-06, "loss": 1.358, "step": 4475 }, { "epoch": 0.72, "grad_norm": 7.778215408325195, "learning_rate": 3.824684337463557e-06, "loss": 1.1583, "step": 4476 }, { "epoch": 0.72, "grad_norm": 8.041041374206543, "learning_rate": 3.820583919969551e-06, "loss": 1.1832, "step": 4477 }, { "epoch": 0.72, "grad_norm": 8.312314987182617, "learning_rate": 3.816485182567576e-06, "loss": 1.3959, "step": 4478 }, { "epoch": 0.72, "grad_norm": 9.921216011047363, "learning_rate": 3.812388126372013e-06, "loss": 1.3599, "step": 4479 }, { "epoch": 0.72, "grad_norm": 8.100531578063965, "learning_rate": 3.808292752496804e-06, "loss": 1.4312, "step": 4480 }, { "epoch": 0.72, "grad_norm": 8.52425479888916, "learning_rate": 3.804199062055408e-06, "loss": 1.2738, "step": 4481 }, { "epoch": 0.72, "grad_norm": 9.319398880004883, "learning_rate": 3.800107056160843e-06, "loss": 1.2425, "step": 4482 }, { "epoch": 0.72, "grad_norm": 9.511603355407715, "learning_rate": 3.796016735925666e-06, "loss": 1.241, "step": 4483 }, { "epoch": 0.72, "grad_norm": 9.246519088745117, "learning_rate": 3.7919281024619702e-06, "loss": 1.2755, "step": 4484 }, { "epoch": 0.72, "grad_norm": 10.610244750976562, "learning_rate": 3.7878411568814057e-06, "loss": 1.273, "step": 4485 }, { "epoch": 0.72, "grad_norm": 6.9863152503967285, "learning_rate": 3.7837559002951416e-06, "loss": 1.2512, "step": 4486 }, { "epoch": 0.72, "grad_norm": 7.401047706604004, "learning_rate": 3.7796723338138995e-06, "loss": 1.4503, "step": 4487 }, { "epoch": 0.72, "grad_norm": 6.655360698699951, "learning_rate": 3.7755904585479518e-06, "loss": 1.2293, "step": 4488 }, { "epoch": 0.72, "grad_norm": 9.679678916931152, "learning_rate": 3.7715102756070854e-06, "loss": 1.3988, "step": 4489 }, { "epoch": 0.72, "grad_norm": 10.630749702453613, "learning_rate": 3.7674317861006546e-06, "loss": 1.516, "step": 4490 }, { "epoch": 0.72, "grad_norm": 11.001835823059082, "learning_rate": 3.7633549911375333e-06, "loss": 1.3621, "step": 4491 }, { "epoch": 0.72, "grad_norm": 8.834798812866211, "learning_rate": 3.7592798918261385e-06, "loss": 1.1617, "step": 4492 }, { "epoch": 0.72, "grad_norm": 8.776956558227539, "learning_rate": 3.7552064892744423e-06, "loss": 1.3073, "step": 4493 }, { "epoch": 0.72, "grad_norm": 7.958617687225342, "learning_rate": 3.751134784589927e-06, "loss": 1.2097, "step": 4494 }, { "epoch": 0.72, "grad_norm": 7.529139041900635, "learning_rate": 3.7470647788796446e-06, "loss": 1.3844, "step": 4495 }, { "epoch": 0.72, "grad_norm": 8.759448051452637, "learning_rate": 3.742996473250159e-06, "loss": 0.9368, "step": 4496 }, { "epoch": 0.72, "grad_norm": 8.940445899963379, "learning_rate": 3.7389298688075805e-06, "loss": 1.4233, "step": 4497 }, { "epoch": 0.72, "grad_norm": 7.305788040161133, "learning_rate": 3.734864966657572e-06, "loss": 1.2553, "step": 4498 }, { "epoch": 0.72, "grad_norm": 10.253430366516113, "learning_rate": 3.7308017679053042e-06, "loss": 1.5433, "step": 4499 }, { "epoch": 0.72, "grad_norm": 7.2945556640625, "learning_rate": 3.726740273655516e-06, "loss": 1.0124, "step": 4500 }, { "epoch": 0.72, "grad_norm": 8.19592571258545, "learning_rate": 3.7226804850124533e-06, "loss": 1.4768, "step": 4501 }, { "epoch": 0.72, "grad_norm": 8.931976318359375, "learning_rate": 3.7186224030799234e-06, "loss": 1.3881, "step": 4502 }, { "epoch": 0.72, "grad_norm": 8.221543312072754, "learning_rate": 3.7145660289612596e-06, "loss": 1.2166, "step": 4503 }, { "epoch": 0.73, "grad_norm": 7.8065385818481445, "learning_rate": 3.710511363759318e-06, "loss": 1.4809, "step": 4504 }, { "epoch": 0.73, "grad_norm": 7.4096503257751465, "learning_rate": 3.706458408576519e-06, "loss": 1.298, "step": 4505 }, { "epoch": 0.73, "grad_norm": 8.689431190490723, "learning_rate": 3.702407164514783e-06, "loss": 1.227, "step": 4506 }, { "epoch": 0.73, "grad_norm": 7.711462497711182, "learning_rate": 3.698357632675598e-06, "loss": 1.0155, "step": 4507 }, { "epoch": 0.73, "grad_norm": 9.744192123413086, "learning_rate": 3.69430981415997e-06, "loss": 1.3181, "step": 4508 }, { "epoch": 0.73, "grad_norm": 9.714954376220703, "learning_rate": 3.6902637100684313e-06, "loss": 1.56, "step": 4509 }, { "epoch": 0.73, "grad_norm": 7.832899570465088, "learning_rate": 3.686219321501068e-06, "loss": 1.4433, "step": 4510 }, { "epoch": 0.73, "grad_norm": 9.230016708374023, "learning_rate": 3.682176649557486e-06, "loss": 1.3839, "step": 4511 }, { "epoch": 0.73, "grad_norm": 6.2311787605285645, "learning_rate": 3.6781356953368286e-06, "loss": 1.0579, "step": 4512 }, { "epoch": 0.73, "grad_norm": 8.86666202545166, "learning_rate": 3.674096459937775e-06, "loss": 1.1415, "step": 4513 }, { "epoch": 0.73, "grad_norm": 7.476680278778076, "learning_rate": 3.670058944458522e-06, "loss": 1.2365, "step": 4514 }, { "epoch": 0.73, "grad_norm": 7.359304904937744, "learning_rate": 3.6660231499968237e-06, "loss": 1.439, "step": 4515 }, { "epoch": 0.73, "grad_norm": 9.475521087646484, "learning_rate": 3.6619890776499465e-06, "loss": 1.4988, "step": 4516 }, { "epoch": 0.73, "grad_norm": 8.151078224182129, "learning_rate": 3.6579567285146976e-06, "loss": 1.2474, "step": 4517 }, { "epoch": 0.73, "grad_norm": 7.250974178314209, "learning_rate": 3.6539261036874153e-06, "loss": 1.1411, "step": 4518 }, { "epoch": 0.73, "grad_norm": 8.769679069519043, "learning_rate": 3.649897204263957e-06, "loss": 1.4612, "step": 4519 }, { "epoch": 0.73, "grad_norm": 6.270603179931641, "learning_rate": 3.64587003133973e-06, "loss": 1.1501, "step": 4520 }, { "epoch": 0.73, "grad_norm": 7.099504470825195, "learning_rate": 3.641844586009663e-06, "loss": 1.3402, "step": 4521 }, { "epoch": 0.73, "grad_norm": 7.973742485046387, "learning_rate": 3.6378208693682115e-06, "loss": 1.226, "step": 4522 }, { "epoch": 0.73, "grad_norm": 7.430667877197266, "learning_rate": 3.6337988825093707e-06, "loss": 1.131, "step": 4523 }, { "epoch": 0.73, "grad_norm": 9.652567863464355, "learning_rate": 3.629778626526649e-06, "loss": 1.5588, "step": 4524 }, { "epoch": 0.73, "grad_norm": 6.910123348236084, "learning_rate": 3.625760102513103e-06, "loss": 1.1123, "step": 4525 }, { "epoch": 0.73, "grad_norm": 7.946967124938965, "learning_rate": 3.621743311561309e-06, "loss": 1.3022, "step": 4526 }, { "epoch": 0.73, "grad_norm": 8.582806587219238, "learning_rate": 3.6177282547633706e-06, "loss": 1.498, "step": 4527 }, { "epoch": 0.73, "grad_norm": 8.852519035339355, "learning_rate": 3.613714933210929e-06, "loss": 1.2491, "step": 4528 }, { "epoch": 0.73, "grad_norm": 10.526849746704102, "learning_rate": 3.6097033479951336e-06, "loss": 1.729, "step": 4529 }, { "epoch": 0.73, "grad_norm": 7.776992321014404, "learning_rate": 3.6056935002066885e-06, "loss": 1.1463, "step": 4530 }, { "epoch": 0.73, "grad_norm": 9.909778594970703, "learning_rate": 3.601685390935806e-06, "loss": 1.5989, "step": 4531 }, { "epoch": 0.73, "grad_norm": 8.139925003051758, "learning_rate": 3.5976790212722346e-06, "loss": 1.2278, "step": 4532 }, { "epoch": 0.73, "grad_norm": 9.331317901611328, "learning_rate": 3.5936743923052443e-06, "loss": 1.4181, "step": 4533 }, { "epoch": 0.73, "grad_norm": 9.964926719665527, "learning_rate": 3.5896715051236363e-06, "loss": 1.4398, "step": 4534 }, { "epoch": 0.73, "grad_norm": 8.489023208618164, "learning_rate": 3.5856703608157362e-06, "loss": 1.4041, "step": 4535 }, { "epoch": 0.73, "grad_norm": 6.885572910308838, "learning_rate": 3.5816709604693965e-06, "loss": 0.9169, "step": 4536 }, { "epoch": 0.73, "grad_norm": 6.961298942565918, "learning_rate": 3.5776733051719935e-06, "loss": 1.1425, "step": 4537 }, { "epoch": 0.73, "grad_norm": 7.103003978729248, "learning_rate": 3.5736773960104333e-06, "loss": 1.3711, "step": 4538 }, { "epoch": 0.73, "grad_norm": 8.898727416992188, "learning_rate": 3.569683234071143e-06, "loss": 1.5713, "step": 4539 }, { "epoch": 0.73, "grad_norm": 7.781208038330078, "learning_rate": 3.5656908204400753e-06, "loss": 1.481, "step": 4540 }, { "epoch": 0.73, "grad_norm": 7.252180576324463, "learning_rate": 3.561700156202711e-06, "loss": 1.3226, "step": 4541 }, { "epoch": 0.73, "grad_norm": 9.133667945861816, "learning_rate": 3.557711242444051e-06, "loss": 1.3159, "step": 4542 }, { "epoch": 0.73, "grad_norm": 8.84311294555664, "learning_rate": 3.553724080248623e-06, "loss": 1.2235, "step": 4543 }, { "epoch": 0.73, "grad_norm": 7.68709659576416, "learning_rate": 3.5497386707004767e-06, "loss": 1.3938, "step": 4544 }, { "epoch": 0.73, "grad_norm": 8.22840690612793, "learning_rate": 3.5457550148831865e-06, "loss": 1.5469, "step": 4545 }, { "epoch": 0.73, "grad_norm": 8.61273193359375, "learning_rate": 3.5417731138798494e-06, "loss": 1.3431, "step": 4546 }, { "epoch": 0.73, "grad_norm": 7.7836174964904785, "learning_rate": 3.5377929687730863e-06, "loss": 1.525, "step": 4547 }, { "epoch": 0.73, "grad_norm": 8.36072063446045, "learning_rate": 3.5338145806450396e-06, "loss": 1.4871, "step": 4548 }, { "epoch": 0.73, "grad_norm": 7.783292770385742, "learning_rate": 3.529837950577374e-06, "loss": 1.2489, "step": 4549 }, { "epoch": 0.73, "grad_norm": 8.591495513916016, "learning_rate": 3.525863079651276e-06, "loss": 1.3045, "step": 4550 }, { "epoch": 0.73, "grad_norm": 8.671175956726074, "learning_rate": 3.521889968947456e-06, "loss": 1.2085, "step": 4551 }, { "epoch": 0.73, "grad_norm": 7.470466613769531, "learning_rate": 3.5179186195461424e-06, "loss": 1.33, "step": 4552 }, { "epoch": 0.73, "grad_norm": 8.393516540527344, "learning_rate": 3.5139490325270886e-06, "loss": 1.201, "step": 4553 }, { "epoch": 0.73, "grad_norm": 8.339495658874512, "learning_rate": 3.5099812089695638e-06, "loss": 1.3058, "step": 4554 }, { "epoch": 0.73, "grad_norm": 8.903800964355469, "learning_rate": 3.506015149952363e-06, "loss": 1.0674, "step": 4555 }, { "epoch": 0.73, "grad_norm": 8.197624206542969, "learning_rate": 3.502050856553798e-06, "loss": 1.4703, "step": 4556 }, { "epoch": 0.73, "grad_norm": 7.2010087966918945, "learning_rate": 3.4980883298517034e-06, "loss": 1.236, "step": 4557 }, { "epoch": 0.73, "grad_norm": 8.203733444213867, "learning_rate": 3.494127570923429e-06, "loss": 1.1886, "step": 4558 }, { "epoch": 0.73, "grad_norm": 8.431140899658203, "learning_rate": 3.4901685808458484e-06, "loss": 1.4726, "step": 4559 }, { "epoch": 0.73, "grad_norm": 8.893315315246582, "learning_rate": 3.4862113606953517e-06, "loss": 1.1762, "step": 4560 }, { "epoch": 0.73, "grad_norm": 7.636176109313965, "learning_rate": 3.4822559115478494e-06, "loss": 1.3723, "step": 4561 }, { "epoch": 0.73, "grad_norm": 7.675425052642822, "learning_rate": 3.4783022344787698e-06, "loss": 1.1681, "step": 4562 }, { "epoch": 0.73, "grad_norm": 7.123659133911133, "learning_rate": 3.47435033056306e-06, "loss": 1.2165, "step": 4563 }, { "epoch": 0.73, "grad_norm": 9.859236717224121, "learning_rate": 3.4704002008751813e-06, "loss": 1.3655, "step": 4564 }, { "epoch": 0.73, "grad_norm": 7.108338832855225, "learning_rate": 3.4664518464891194e-06, "loss": 1.1888, "step": 4565 }, { "epoch": 0.74, "grad_norm": 7.711860656738281, "learning_rate": 3.4625052684783724e-06, "loss": 1.1435, "step": 4566 }, { "epoch": 0.74, "grad_norm": 8.779023170471191, "learning_rate": 3.4585604679159544e-06, "loss": 1.4165, "step": 4567 }, { "epoch": 0.74, "grad_norm": 8.036394119262695, "learning_rate": 3.4546174458743976e-06, "loss": 1.3317, "step": 4568 }, { "epoch": 0.74, "grad_norm": 8.736040115356445, "learning_rate": 3.450676203425761e-06, "loss": 1.5262, "step": 4569 }, { "epoch": 0.74, "grad_norm": 7.775277614593506, "learning_rate": 3.446736741641601e-06, "loss": 1.1421, "step": 4570 }, { "epoch": 0.74, "grad_norm": 8.66751480102539, "learning_rate": 3.4427990615929995e-06, "loss": 1.5418, "step": 4571 }, { "epoch": 0.74, "grad_norm": 7.485045909881592, "learning_rate": 3.4388631643505565e-06, "loss": 1.2508, "step": 4572 }, { "epoch": 0.74, "grad_norm": 8.323970794677734, "learning_rate": 3.4349290509843802e-06, "loss": 1.1853, "step": 4573 }, { "epoch": 0.74, "grad_norm": 8.702556610107422, "learning_rate": 3.4309967225641083e-06, "loss": 1.3756, "step": 4574 }, { "epoch": 0.74, "grad_norm": 7.5157575607299805, "learning_rate": 3.427066180158869e-06, "loss": 1.4315, "step": 4575 }, { "epoch": 0.74, "grad_norm": 8.91662883758545, "learning_rate": 3.4231374248373316e-06, "loss": 1.0599, "step": 4576 }, { "epoch": 0.74, "grad_norm": 10.572585105895996, "learning_rate": 3.4192104576676576e-06, "loss": 1.4329, "step": 4577 }, { "epoch": 0.74, "grad_norm": 11.194655418395996, "learning_rate": 3.4152852797175316e-06, "loss": 1.3714, "step": 4578 }, { "epoch": 0.74, "grad_norm": 7.7991108894348145, "learning_rate": 3.41136189205416e-06, "loss": 1.3024, "step": 4579 }, { "epoch": 0.74, "grad_norm": 8.490649223327637, "learning_rate": 3.4074402957442433e-06, "loss": 1.3125, "step": 4580 }, { "epoch": 0.74, "grad_norm": 7.53840446472168, "learning_rate": 3.4035204918540167e-06, "loss": 1.3359, "step": 4581 }, { "epoch": 0.74, "grad_norm": 8.209517478942871, "learning_rate": 3.3996024814492066e-06, "loss": 1.1829, "step": 4582 }, { "epoch": 0.74, "grad_norm": 7.743111610412598, "learning_rate": 3.395686265595065e-06, "loss": 1.1419, "step": 4583 }, { "epoch": 0.74, "grad_norm": 9.723570823669434, "learning_rate": 3.391771845356361e-06, "loss": 1.2014, "step": 4584 }, { "epoch": 0.74, "grad_norm": 10.369216918945312, "learning_rate": 3.387859221797354e-06, "loss": 1.282, "step": 4585 }, { "epoch": 0.74, "grad_norm": 8.353949546813965, "learning_rate": 3.3839483959818444e-06, "loss": 1.5944, "step": 4586 }, { "epoch": 0.74, "grad_norm": 8.48044490814209, "learning_rate": 3.380039368973115e-06, "loss": 1.2295, "step": 4587 }, { "epoch": 0.74, "grad_norm": 8.287620544433594, "learning_rate": 3.3761321418339744e-06, "loss": 1.2058, "step": 4588 }, { "epoch": 0.74, "grad_norm": 9.077815055847168, "learning_rate": 3.3722267156267507e-06, "loss": 1.553, "step": 4589 }, { "epoch": 0.74, "grad_norm": 10.0988130569458, "learning_rate": 3.368323091413256e-06, "loss": 1.1008, "step": 4590 }, { "epoch": 0.74, "grad_norm": 11.1289701461792, "learning_rate": 3.3644212702548394e-06, "loss": 1.356, "step": 4591 }, { "epoch": 0.74, "grad_norm": 7.93710470199585, "learning_rate": 3.360521253212349e-06, "loss": 1.5442, "step": 4592 }, { "epoch": 0.74, "grad_norm": 8.04394245147705, "learning_rate": 3.356623041346131e-06, "loss": 1.4329, "step": 4593 }, { "epoch": 0.74, "grad_norm": 8.459552764892578, "learning_rate": 3.352726635716065e-06, "loss": 1.2349, "step": 4594 }, { "epoch": 0.74, "grad_norm": 7.142966270446777, "learning_rate": 3.348832037381513e-06, "loss": 1.1708, "step": 4595 }, { "epoch": 0.74, "grad_norm": 7.614665985107422, "learning_rate": 3.3449392474013696e-06, "loss": 1.2559, "step": 4596 }, { "epoch": 0.74, "grad_norm": 7.046380519866943, "learning_rate": 3.341048266834026e-06, "loss": 1.097, "step": 4597 }, { "epoch": 0.74, "grad_norm": 9.508950233459473, "learning_rate": 3.3371590967373734e-06, "loss": 1.3192, "step": 4598 }, { "epoch": 0.74, "grad_norm": 6.845777988433838, "learning_rate": 3.3332717381688316e-06, "loss": 1.2967, "step": 4599 }, { "epoch": 0.74, "grad_norm": 8.138312339782715, "learning_rate": 3.3293861921853034e-06, "loss": 1.4022, "step": 4600 }, { "epoch": 0.74, "grad_norm": 8.732057571411133, "learning_rate": 3.3255024598432216e-06, "loss": 1.2094, "step": 4601 }, { "epoch": 0.74, "grad_norm": 8.70136833190918, "learning_rate": 3.321620542198516e-06, "loss": 1.2685, "step": 4602 }, { "epoch": 0.74, "grad_norm": 9.203987121582031, "learning_rate": 3.3177404403066117e-06, "loss": 1.1318, "step": 4603 }, { "epoch": 0.74, "grad_norm": 8.008842468261719, "learning_rate": 3.313862155222465e-06, "loss": 1.3173, "step": 4604 }, { "epoch": 0.74, "grad_norm": 7.958980560302734, "learning_rate": 3.309985688000511e-06, "loss": 1.4007, "step": 4605 }, { "epoch": 0.74, "grad_norm": 7.566827774047852, "learning_rate": 3.306111039694714e-06, "loss": 1.1655, "step": 4606 }, { "epoch": 0.74, "grad_norm": 9.003353118896484, "learning_rate": 3.302238211358534e-06, "loss": 1.2006, "step": 4607 }, { "epoch": 0.74, "grad_norm": 8.994793891906738, "learning_rate": 3.2983672040449254e-06, "loss": 1.0315, "step": 4608 }, { "epoch": 0.74, "grad_norm": 7.4852294921875, "learning_rate": 3.2944980188063713e-06, "loss": 1.101, "step": 4609 }, { "epoch": 0.74, "grad_norm": 8.417320251464844, "learning_rate": 3.2906306566948344e-06, "loss": 1.3751, "step": 4610 }, { "epoch": 0.74, "grad_norm": 7.943363666534424, "learning_rate": 3.2867651187617997e-06, "loss": 1.23, "step": 4611 }, { "epoch": 0.74, "grad_norm": 9.400644302368164, "learning_rate": 3.2829014060582498e-06, "loss": 1.2981, "step": 4612 }, { "epoch": 0.74, "grad_norm": 8.84731388092041, "learning_rate": 3.27903951963467e-06, "loss": 1.1972, "step": 4613 }, { "epoch": 0.74, "grad_norm": 7.589398384094238, "learning_rate": 3.2751794605410504e-06, "loss": 1.4473, "step": 4614 }, { "epoch": 0.74, "grad_norm": 7.730644226074219, "learning_rate": 3.271321229826884e-06, "loss": 1.1992, "step": 4615 }, { "epoch": 0.74, "grad_norm": 8.239459037780762, "learning_rate": 3.267464828541167e-06, "loss": 1.3931, "step": 4616 }, { "epoch": 0.74, "grad_norm": 8.417708396911621, "learning_rate": 3.263610257732397e-06, "loss": 1.3262, "step": 4617 }, { "epoch": 0.74, "grad_norm": 9.548640251159668, "learning_rate": 3.2597575184485763e-06, "loss": 1.299, "step": 4618 }, { "epoch": 0.74, "grad_norm": 7.124185562133789, "learning_rate": 3.2559066117372075e-06, "loss": 1.2676, "step": 4619 }, { "epoch": 0.74, "grad_norm": 9.088894844055176, "learning_rate": 3.2520575386452948e-06, "loss": 1.2903, "step": 4620 }, { "epoch": 0.74, "grad_norm": 6.508674621582031, "learning_rate": 3.2482103002193444e-06, "loss": 1.1491, "step": 4621 }, { "epoch": 0.74, "grad_norm": 8.69420337677002, "learning_rate": 3.2443648975053643e-06, "loss": 1.3947, "step": 4622 }, { "epoch": 0.74, "grad_norm": 9.77818489074707, "learning_rate": 3.2405213315488614e-06, "loss": 1.3383, "step": 4623 }, { "epoch": 0.74, "grad_norm": 8.527155876159668, "learning_rate": 3.2366796033948466e-06, "loss": 1.6232, "step": 4624 }, { "epoch": 0.74, "grad_norm": 9.195982933044434, "learning_rate": 3.2328397140878276e-06, "loss": 1.2722, "step": 4625 }, { "epoch": 0.74, "grad_norm": 7.9742608070373535, "learning_rate": 3.2290016646718136e-06, "loss": 1.1501, "step": 4626 }, { "epoch": 0.74, "grad_norm": 9.651236534118652, "learning_rate": 3.2251654561903144e-06, "loss": 1.2465, "step": 4627 }, { "epoch": 0.75, "grad_norm": 7.726917743682861, "learning_rate": 3.221331089686338e-06, "loss": 1.3884, "step": 4628 }, { "epoch": 0.75, "grad_norm": 8.61681842803955, "learning_rate": 3.2174985662023915e-06, "loss": 1.0028, "step": 4629 }, { "epoch": 0.75, "grad_norm": 9.279662132263184, "learning_rate": 3.213667886780483e-06, "loss": 1.408, "step": 4630 }, { "epoch": 0.75, "grad_norm": 8.781529426574707, "learning_rate": 3.2098390524621157e-06, "loss": 1.1611, "step": 4631 }, { "epoch": 0.75, "grad_norm": 9.163649559020996, "learning_rate": 3.2060120642882953e-06, "loss": 1.2068, "step": 4632 }, { "epoch": 0.75, "grad_norm": 6.886148929595947, "learning_rate": 3.2021869232995216e-06, "loss": 1.3875, "step": 4633 }, { "epoch": 0.75, "grad_norm": 7.051267147064209, "learning_rate": 3.1983636305357955e-06, "loss": 1.2749, "step": 4634 }, { "epoch": 0.75, "grad_norm": 8.694595336914062, "learning_rate": 3.194542187036612e-06, "loss": 1.346, "step": 4635 }, { "epoch": 0.75, "grad_norm": 7.980004787445068, "learning_rate": 3.190722593840967e-06, "loss": 1.3832, "step": 4636 }, { "epoch": 0.75, "grad_norm": 8.495509147644043, "learning_rate": 3.1869048519873514e-06, "loss": 1.2823, "step": 4637 }, { "epoch": 0.75, "grad_norm": 9.097369194030762, "learning_rate": 3.1830889625137505e-06, "loss": 1.0703, "step": 4638 }, { "epoch": 0.75, "grad_norm": 9.357893943786621, "learning_rate": 3.1792749264576516e-06, "loss": 1.579, "step": 4639 }, { "epoch": 0.75, "grad_norm": 8.656359672546387, "learning_rate": 3.1754627448560325e-06, "loss": 1.3652, "step": 4640 }, { "epoch": 0.75, "grad_norm": 8.384878158569336, "learning_rate": 3.17165241874537e-06, "loss": 1.3078, "step": 4641 }, { "epoch": 0.75, "grad_norm": 7.615793228149414, "learning_rate": 3.167843949161635e-06, "loss": 1.0741, "step": 4642 }, { "epoch": 0.75, "grad_norm": 7.734640121459961, "learning_rate": 3.1640373371402957e-06, "loss": 1.347, "step": 4643 }, { "epoch": 0.75, "grad_norm": 8.51125431060791, "learning_rate": 3.1602325837163116e-06, "loss": 1.1367, "step": 4644 }, { "epoch": 0.75, "grad_norm": 8.122054100036621, "learning_rate": 3.156429689924141e-06, "loss": 1.3261, "step": 4645 }, { "epoch": 0.75, "grad_norm": 7.690173625946045, "learning_rate": 3.1526286567977325e-06, "loss": 1.263, "step": 4646 }, { "epoch": 0.75, "grad_norm": 7.478602886199951, "learning_rate": 3.1488294853705337e-06, "loss": 1.1872, "step": 4647 }, { "epoch": 0.75, "grad_norm": 9.816329002380371, "learning_rate": 3.1450321766754798e-06, "loss": 1.3324, "step": 4648 }, { "epoch": 0.75, "grad_norm": 7.76926851272583, "learning_rate": 3.1412367317450053e-06, "loss": 1.1713, "step": 4649 }, { "epoch": 0.75, "grad_norm": 8.568570137023926, "learning_rate": 3.1374431516110358e-06, "loss": 1.2859, "step": 4650 }, { "epoch": 0.75, "grad_norm": 8.653106689453125, "learning_rate": 3.1336514373049887e-06, "loss": 1.2325, "step": 4651 }, { "epoch": 0.75, "grad_norm": 7.654730796813965, "learning_rate": 3.1298615898577756e-06, "loss": 1.2397, "step": 4652 }, { "epoch": 0.75, "grad_norm": 6.862995147705078, "learning_rate": 3.126073610299799e-06, "loss": 1.2072, "step": 4653 }, { "epoch": 0.75, "grad_norm": 7.639853477478027, "learning_rate": 3.1222874996609544e-06, "loss": 1.1592, "step": 4654 }, { "epoch": 0.75, "grad_norm": 10.665618896484375, "learning_rate": 3.1185032589706354e-06, "loss": 1.4375, "step": 4655 }, { "epoch": 0.75, "grad_norm": 8.961423873901367, "learning_rate": 3.1147208892577143e-06, "loss": 1.2685, "step": 4656 }, { "epoch": 0.75, "grad_norm": 7.751614570617676, "learning_rate": 3.110940391550562e-06, "loss": 1.0424, "step": 4657 }, { "epoch": 0.75, "grad_norm": 8.49979019165039, "learning_rate": 3.107161766877044e-06, "loss": 1.3356, "step": 4658 }, { "epoch": 0.75, "grad_norm": 7.849248886108398, "learning_rate": 3.103385016264505e-06, "loss": 1.3062, "step": 4659 }, { "epoch": 0.75, "grad_norm": 8.436721801757812, "learning_rate": 3.0996101407398014e-06, "loss": 1.3398, "step": 4660 }, { "epoch": 0.75, "grad_norm": 7.693558216094971, "learning_rate": 3.0958371413292544e-06, "loss": 1.2568, "step": 4661 }, { "epoch": 0.75, "grad_norm": 8.526638984680176, "learning_rate": 3.0920660190586893e-06, "loss": 1.1485, "step": 4662 }, { "epoch": 0.75, "grad_norm": 9.304497718811035, "learning_rate": 3.08829677495342e-06, "loss": 1.4023, "step": 4663 }, { "epoch": 0.75, "grad_norm": 9.184489250183105, "learning_rate": 3.084529410038245e-06, "loss": 1.3232, "step": 4664 }, { "epoch": 0.75, "grad_norm": 7.6314897537231445, "learning_rate": 3.0807639253374644e-06, "loss": 1.3754, "step": 4665 }, { "epoch": 0.75, "grad_norm": 7.406195163726807, "learning_rate": 3.077000321874848e-06, "loss": 1.3598, "step": 4666 }, { "epoch": 0.75, "grad_norm": 8.581576347351074, "learning_rate": 3.073238600673668e-06, "loss": 1.3625, "step": 4667 }, { "epoch": 0.75, "grad_norm": 8.112115859985352, "learning_rate": 3.0694787627566815e-06, "loss": 1.0975, "step": 4668 }, { "epoch": 0.75, "grad_norm": 8.193795204162598, "learning_rate": 3.0657208091461277e-06, "loss": 1.3814, "step": 4669 }, { "epoch": 0.75, "grad_norm": 7.608819484710693, "learning_rate": 3.0619647408637496e-06, "loss": 1.1781, "step": 4670 }, { "epoch": 0.75, "grad_norm": 7.194788455963135, "learning_rate": 3.0582105589307577e-06, "loss": 1.0812, "step": 4671 }, { "epoch": 0.75, "grad_norm": 7.175710201263428, "learning_rate": 3.0544582643678566e-06, "loss": 1.1315, "step": 4672 }, { "epoch": 0.75, "grad_norm": 11.47225284576416, "learning_rate": 3.050707858195252e-06, "loss": 1.5339, "step": 4673 }, { "epoch": 0.75, "grad_norm": 8.205732345581055, "learning_rate": 3.0469593414326094e-06, "loss": 1.3611, "step": 4674 }, { "epoch": 0.75, "grad_norm": 7.916146755218506, "learning_rate": 3.0432127150991088e-06, "loss": 1.1291, "step": 4675 }, { "epoch": 0.75, "grad_norm": 6.854058742523193, "learning_rate": 3.039467980213392e-06, "loss": 1.2074, "step": 4676 }, { "epoch": 0.75, "grad_norm": 7.826797962188721, "learning_rate": 3.0357251377935983e-06, "loss": 1.2258, "step": 4677 }, { "epoch": 0.75, "grad_norm": 8.217113494873047, "learning_rate": 3.031984188857361e-06, "loss": 1.4427, "step": 4678 }, { "epoch": 0.75, "grad_norm": 7.134827613830566, "learning_rate": 3.0282451344217747e-06, "loss": 1.4006, "step": 4679 }, { "epoch": 0.75, "grad_norm": 9.362237930297852, "learning_rate": 3.0245079755034466e-06, "loss": 1.385, "step": 4680 }, { "epoch": 0.75, "grad_norm": 8.086333274841309, "learning_rate": 3.0207727131184474e-06, "loss": 1.3191, "step": 4681 }, { "epoch": 0.75, "grad_norm": 8.339790344238281, "learning_rate": 3.017039348282337e-06, "loss": 1.3307, "step": 4682 }, { "epoch": 0.75, "grad_norm": 8.290328025817871, "learning_rate": 3.0133078820101725e-06, "loss": 1.101, "step": 4683 }, { "epoch": 0.75, "grad_norm": 8.478279113769531, "learning_rate": 3.0095783153164716e-06, "loss": 1.4541, "step": 4684 }, { "epoch": 0.75, "grad_norm": 7.636178016662598, "learning_rate": 3.0058506492152627e-06, "loss": 1.1679, "step": 4685 }, { "epoch": 0.75, "grad_norm": 9.531475067138672, "learning_rate": 3.0021248847200323e-06, "loss": 1.4016, "step": 4686 }, { "epoch": 0.75, "grad_norm": 7.475851535797119, "learning_rate": 2.998401022843761e-06, "loss": 1.3753, "step": 4687 }, { "epoch": 0.75, "grad_norm": 8.029245376586914, "learning_rate": 2.994679064598922e-06, "loss": 1.3782, "step": 4688 }, { "epoch": 0.75, "grad_norm": 7.554228782653809, "learning_rate": 2.9909590109974474e-06, "loss": 1.2823, "step": 4689 }, { "epoch": 0.75, "grad_norm": 8.572206497192383, "learning_rate": 2.9872408630507767e-06, "loss": 1.217, "step": 4690 }, { "epoch": 0.76, "grad_norm": 8.555697441101074, "learning_rate": 2.9835246217698087e-06, "loss": 1.3789, "step": 4691 }, { "epoch": 0.76, "grad_norm": 8.63428020477295, "learning_rate": 2.979810288164944e-06, "loss": 1.271, "step": 4692 }, { "epoch": 0.76, "grad_norm": 8.413121223449707, "learning_rate": 2.9760978632460535e-06, "loss": 1.1721, "step": 4693 }, { "epoch": 0.76, "grad_norm": 8.77429485321045, "learning_rate": 2.972387348022482e-06, "loss": 1.4504, "step": 4694 }, { "epoch": 0.76, "grad_norm": 8.626254081726074, "learning_rate": 2.9686787435030737e-06, "loss": 1.6058, "step": 4695 }, { "epoch": 0.76, "grad_norm": 8.99588680267334, "learning_rate": 2.96497205069614e-06, "loss": 1.1641, "step": 4696 }, { "epoch": 0.76, "grad_norm": 7.637059688568115, "learning_rate": 2.9612672706094758e-06, "loss": 1.2777, "step": 4697 }, { "epoch": 0.76, "grad_norm": 7.616948127746582, "learning_rate": 2.957564404250359e-06, "loss": 1.1742, "step": 4698 }, { "epoch": 0.76, "grad_norm": 8.596125602722168, "learning_rate": 2.953863452625535e-06, "loss": 1.1631, "step": 4699 }, { "epoch": 0.76, "grad_norm": 9.406505584716797, "learning_rate": 2.950164416741249e-06, "loss": 1.3799, "step": 4700 }, { "epoch": 0.76, "grad_norm": 8.435296058654785, "learning_rate": 2.946467297603208e-06, "loss": 1.4076, "step": 4701 }, { "epoch": 0.76, "grad_norm": 7.269887447357178, "learning_rate": 2.9427720962166075e-06, "loss": 1.2309, "step": 4702 }, { "epoch": 0.76, "grad_norm": 9.985013008117676, "learning_rate": 2.9390788135861193e-06, "loss": 1.2507, "step": 4703 }, { "epoch": 0.76, "grad_norm": 7.524306297302246, "learning_rate": 2.935387450715883e-06, "loss": 1.1531, "step": 4704 }, { "epoch": 0.76, "grad_norm": 7.377172470092773, "learning_rate": 2.9316980086095357e-06, "loss": 1.3102, "step": 4705 }, { "epoch": 0.76, "grad_norm": 8.391057014465332, "learning_rate": 2.9280104882701786e-06, "loss": 1.0942, "step": 4706 }, { "epoch": 0.76, "grad_norm": 7.558752536773682, "learning_rate": 2.924324890700395e-06, "loss": 1.6283, "step": 4707 }, { "epoch": 0.76, "grad_norm": 7.349853992462158, "learning_rate": 2.920641216902247e-06, "loss": 1.3382, "step": 4708 }, { "epoch": 0.76, "grad_norm": 7.692368984222412, "learning_rate": 2.9169594678772617e-06, "loss": 1.1902, "step": 4709 }, { "epoch": 0.76, "grad_norm": 7.760998249053955, "learning_rate": 2.913279644626461e-06, "loss": 1.2971, "step": 4710 }, { "epoch": 0.76, "grad_norm": 9.290623664855957, "learning_rate": 2.9096017481503336e-06, "loss": 1.3488, "step": 4711 }, { "epoch": 0.76, "grad_norm": 8.264223098754883, "learning_rate": 2.9059257794488428e-06, "loss": 1.1884, "step": 4712 }, { "epoch": 0.76, "grad_norm": 8.209938049316406, "learning_rate": 2.9022517395214344e-06, "loss": 1.6001, "step": 4713 }, { "epoch": 0.76, "grad_norm": 8.372669219970703, "learning_rate": 2.8985796293670166e-06, "loss": 1.383, "step": 4714 }, { "epoch": 0.76, "grad_norm": 8.601072311401367, "learning_rate": 2.8949094499839913e-06, "loss": 1.1707, "step": 4715 }, { "epoch": 0.76, "grad_norm": 8.33383846282959, "learning_rate": 2.891241202370223e-06, "loss": 1.3329, "step": 4716 }, { "epoch": 0.76, "grad_norm": 8.366466522216797, "learning_rate": 2.8875748875230523e-06, "loss": 1.1238, "step": 4717 }, { "epoch": 0.76, "grad_norm": 7.296361923217773, "learning_rate": 2.883910506439299e-06, "loss": 1.161, "step": 4718 }, { "epoch": 0.76, "grad_norm": 8.401975631713867, "learning_rate": 2.8802480601152526e-06, "loss": 1.4274, "step": 4719 }, { "epoch": 0.76, "grad_norm": 7.125290870666504, "learning_rate": 2.8765875495466778e-06, "loss": 1.2425, "step": 4720 }, { "epoch": 0.76, "grad_norm": 9.327248573303223, "learning_rate": 2.872928975728815e-06, "loss": 1.0755, "step": 4721 }, { "epoch": 0.76, "grad_norm": 8.254805564880371, "learning_rate": 2.8692723396563746e-06, "loss": 1.2991, "step": 4722 }, { "epoch": 0.76, "grad_norm": 8.766081809997559, "learning_rate": 2.865617642323543e-06, "loss": 1.3031, "step": 4723 }, { "epoch": 0.76, "grad_norm": 8.329731941223145, "learning_rate": 2.8619648847239777e-06, "loss": 1.195, "step": 4724 }, { "epoch": 0.76, "grad_norm": 8.139975547790527, "learning_rate": 2.8583140678508113e-06, "loss": 1.299, "step": 4725 }, { "epoch": 0.76, "grad_norm": 8.978482246398926, "learning_rate": 2.854665192696645e-06, "loss": 1.4672, "step": 4726 }, { "epoch": 0.76, "grad_norm": 8.92512321472168, "learning_rate": 2.851018260253555e-06, "loss": 1.4673, "step": 4727 }, { "epoch": 0.76, "grad_norm": 8.86961841583252, "learning_rate": 2.8473732715130895e-06, "loss": 1.2452, "step": 4728 }, { "epoch": 0.76, "grad_norm": 8.229256629943848, "learning_rate": 2.843730227466265e-06, "loss": 1.2522, "step": 4729 }, { "epoch": 0.76, "grad_norm": 8.97922420501709, "learning_rate": 2.8400891291035726e-06, "loss": 1.2561, "step": 4730 }, { "epoch": 0.76, "grad_norm": 9.134062767028809, "learning_rate": 2.8364499774149733e-06, "loss": 1.2655, "step": 4731 }, { "epoch": 0.76, "grad_norm": 8.09110164642334, "learning_rate": 2.832812773389898e-06, "loss": 1.131, "step": 4732 }, { "epoch": 0.76, "grad_norm": 8.46399211883545, "learning_rate": 2.8291775180172497e-06, "loss": 1.1146, "step": 4733 }, { "epoch": 0.76, "grad_norm": 8.287381172180176, "learning_rate": 2.8255442122854003e-06, "loss": 1.3394, "step": 4734 }, { "epoch": 0.76, "grad_norm": 9.026497840881348, "learning_rate": 2.8219128571821917e-06, "loss": 1.1594, "step": 4735 }, { "epoch": 0.76, "grad_norm": 7.429845809936523, "learning_rate": 2.8182834536949365e-06, "loss": 1.124, "step": 4736 }, { "epoch": 0.76, "grad_norm": 8.252425193786621, "learning_rate": 2.8146560028104155e-06, "loss": 1.1963, "step": 4737 }, { "epoch": 0.76, "grad_norm": 8.315139770507812, "learning_rate": 2.8110305055148777e-06, "loss": 1.2276, "step": 4738 }, { "epoch": 0.76, "grad_norm": 9.723699569702148, "learning_rate": 2.8074069627940447e-06, "loss": 1.263, "step": 4739 }, { "epoch": 0.76, "grad_norm": 7.469051837921143, "learning_rate": 2.8037853756331035e-06, "loss": 1.4189, "step": 4740 }, { "epoch": 0.76, "grad_norm": 9.178092002868652, "learning_rate": 2.80016574501671e-06, "loss": 1.3981, "step": 4741 }, { "epoch": 0.76, "grad_norm": 8.02131462097168, "learning_rate": 2.7965480719289874e-06, "loss": 1.3543, "step": 4742 }, { "epoch": 0.76, "grad_norm": 10.60721492767334, "learning_rate": 2.792932357353528e-06, "loss": 1.5094, "step": 4743 }, { "epoch": 0.76, "grad_norm": 9.356996536254883, "learning_rate": 2.789318602273393e-06, "loss": 1.4348, "step": 4744 }, { "epoch": 0.76, "grad_norm": 9.253150939941406, "learning_rate": 2.785706807671107e-06, "loss": 1.3021, "step": 4745 }, { "epoch": 0.76, "grad_norm": 9.047102928161621, "learning_rate": 2.7820969745286653e-06, "loss": 1.4008, "step": 4746 }, { "epoch": 0.76, "grad_norm": 8.558175086975098, "learning_rate": 2.7784891038275263e-06, "loss": 1.4468, "step": 4747 }, { "epoch": 0.76, "grad_norm": 9.833550453186035, "learning_rate": 2.7748831965486155e-06, "loss": 1.4499, "step": 4748 }, { "epoch": 0.76, "grad_norm": 7.966786861419678, "learning_rate": 2.771279253672334e-06, "loss": 1.2436, "step": 4749 }, { "epoch": 0.76, "grad_norm": 8.299736976623535, "learning_rate": 2.767677276178532e-06, "loss": 1.3157, "step": 4750 }, { "epoch": 0.76, "grad_norm": 7.449375629425049, "learning_rate": 2.7640772650465364e-06, "loss": 1.6255, "step": 4751 }, { "epoch": 0.76, "grad_norm": 7.987654685974121, "learning_rate": 2.760479221255139e-06, "loss": 1.1955, "step": 4752 }, { "epoch": 0.77, "grad_norm": 7.569361209869385, "learning_rate": 2.7568831457825894e-06, "loss": 1.2751, "step": 4753 }, { "epoch": 0.77, "grad_norm": 9.16832160949707, "learning_rate": 2.753289039606617e-06, "loss": 1.1305, "step": 4754 }, { "epoch": 0.77, "grad_norm": 7.41701078414917, "learning_rate": 2.7496969037043985e-06, "loss": 1.1681, "step": 4755 }, { "epoch": 0.77, "grad_norm": 8.045875549316406, "learning_rate": 2.7461067390525853e-06, "loss": 1.2472, "step": 4756 }, { "epoch": 0.77, "grad_norm": 7.730862617492676, "learning_rate": 2.7425185466272886e-06, "loss": 1.2511, "step": 4757 }, { "epoch": 0.77, "grad_norm": 7.820840835571289, "learning_rate": 2.7389323274040836e-06, "loss": 1.1238, "step": 4758 }, { "epoch": 0.77, "grad_norm": 7.684683322906494, "learning_rate": 2.7353480823580193e-06, "loss": 1.13, "step": 4759 }, { "epoch": 0.77, "grad_norm": 9.74135971069336, "learning_rate": 2.731765812463589e-06, "loss": 1.3907, "step": 4760 }, { "epoch": 0.77, "grad_norm": 8.57617473602295, "learning_rate": 2.7281855186947624e-06, "loss": 1.1876, "step": 4761 }, { "epoch": 0.77, "grad_norm": 7.6454010009765625, "learning_rate": 2.724607202024969e-06, "loss": 1.328, "step": 4762 }, { "epoch": 0.77, "grad_norm": 7.411135196685791, "learning_rate": 2.721030863427098e-06, "loss": 1.4033, "step": 4763 }, { "epoch": 0.77, "grad_norm": 8.321751594543457, "learning_rate": 2.71745650387351e-06, "loss": 1.3072, "step": 4764 }, { "epoch": 0.77, "grad_norm": 8.066367149353027, "learning_rate": 2.7138841243360103e-06, "loss": 1.0872, "step": 4765 }, { "epoch": 0.77, "grad_norm": 8.25729751586914, "learning_rate": 2.7103137257858867e-06, "loss": 1.2905, "step": 4766 }, { "epoch": 0.77, "grad_norm": 8.639188766479492, "learning_rate": 2.70674530919387e-06, "loss": 1.2952, "step": 4767 }, { "epoch": 0.77, "grad_norm": 7.995893478393555, "learning_rate": 2.7031788755301604e-06, "loss": 1.2017, "step": 4768 }, { "epoch": 0.77, "grad_norm": 8.654540061950684, "learning_rate": 2.699614425764426e-06, "loss": 1.2283, "step": 4769 }, { "epoch": 0.77, "grad_norm": 9.896026611328125, "learning_rate": 2.6960519608657788e-06, "loss": 1.3254, "step": 4770 }, { "epoch": 0.77, "grad_norm": 7.764041423797607, "learning_rate": 2.6924914818028093e-06, "loss": 1.1664, "step": 4771 }, { "epoch": 0.77, "grad_norm": 8.383490562438965, "learning_rate": 2.688932989543551e-06, "loss": 1.4097, "step": 4772 }, { "epoch": 0.77, "grad_norm": 8.201016426086426, "learning_rate": 2.6853764850555076e-06, "loss": 1.4037, "step": 4773 }, { "epoch": 0.77, "grad_norm": 9.490365982055664, "learning_rate": 2.681821969305647e-06, "loss": 1.5984, "step": 4774 }, { "epoch": 0.77, "grad_norm": 8.069356918334961, "learning_rate": 2.678269443260377e-06, "loss": 1.1382, "step": 4775 }, { "epoch": 0.77, "grad_norm": 7.425266265869141, "learning_rate": 2.6747189078855883e-06, "loss": 1.3131, "step": 4776 }, { "epoch": 0.77, "grad_norm": 8.240967750549316, "learning_rate": 2.6711703641466168e-06, "loss": 1.2524, "step": 4777 }, { "epoch": 0.77, "grad_norm": 8.054457664489746, "learning_rate": 2.667623813008252e-06, "loss": 1.2219, "step": 4778 }, { "epoch": 0.77, "grad_norm": 8.813188552856445, "learning_rate": 2.6640792554347607e-06, "loss": 1.226, "step": 4779 }, { "epoch": 0.77, "grad_norm": 7.27203369140625, "learning_rate": 2.6605366923898426e-06, "loss": 1.0846, "step": 4780 }, { "epoch": 0.77, "grad_norm": 8.157575607299805, "learning_rate": 2.656996124836678e-06, "loss": 1.3772, "step": 4781 }, { "epoch": 0.77, "grad_norm": 8.672717094421387, "learning_rate": 2.6534575537378946e-06, "loss": 1.3902, "step": 4782 }, { "epoch": 0.77, "grad_norm": 7.505105495452881, "learning_rate": 2.6499209800555693e-06, "loss": 1.2287, "step": 4783 }, { "epoch": 0.77, "grad_norm": 7.271894931793213, "learning_rate": 2.6463864047512555e-06, "loss": 1.0075, "step": 4784 }, { "epoch": 0.77, "grad_norm": 10.235411643981934, "learning_rate": 2.6428538287859406e-06, "loss": 1.4094, "step": 4785 }, { "epoch": 0.77, "grad_norm": 8.092668533325195, "learning_rate": 2.6393232531200884e-06, "loss": 1.2086, "step": 4786 }, { "epoch": 0.77, "grad_norm": 8.745756149291992, "learning_rate": 2.635794678713611e-06, "loss": 1.4206, "step": 4787 }, { "epoch": 0.77, "grad_norm": 6.925349235534668, "learning_rate": 2.632268106525866e-06, "loss": 1.1373, "step": 4788 }, { "epoch": 0.77, "grad_norm": 7.460227966308594, "learning_rate": 2.628743537515689e-06, "loss": 1.2719, "step": 4789 }, { "epoch": 0.77, "grad_norm": 7.5764970779418945, "learning_rate": 2.625220972641346e-06, "loss": 1.1515, "step": 4790 }, { "epoch": 0.77, "grad_norm": 8.386026382446289, "learning_rate": 2.6217004128605796e-06, "loss": 1.1423, "step": 4791 }, { "epoch": 0.77, "grad_norm": 7.903557777404785, "learning_rate": 2.6181818591305775e-06, "loss": 1.3126, "step": 4792 }, { "epoch": 0.77, "grad_norm": 8.61256217956543, "learning_rate": 2.6146653124079735e-06, "loss": 1.1672, "step": 4793 }, { "epoch": 0.77, "grad_norm": 7.750946521759033, "learning_rate": 2.6111507736488773e-06, "loss": 1.4077, "step": 4794 }, { "epoch": 0.77, "grad_norm": 10.298998832702637, "learning_rate": 2.6076382438088277e-06, "loss": 1.2267, "step": 4795 }, { "epoch": 0.77, "grad_norm": 7.24483585357666, "learning_rate": 2.6041277238428398e-06, "loss": 1.2978, "step": 4796 }, { "epoch": 0.77, "grad_norm": 12.580269813537598, "learning_rate": 2.60061921470537e-06, "loss": 1.4291, "step": 4797 }, { "epoch": 0.77, "grad_norm": 7.390072822570801, "learning_rate": 2.597112717350323e-06, "loss": 1.2291, "step": 4798 }, { "epoch": 0.77, "grad_norm": 9.731551170349121, "learning_rate": 2.593608232731073e-06, "loss": 1.566, "step": 4799 }, { "epoch": 0.77, "grad_norm": 7.723793983459473, "learning_rate": 2.590105761800432e-06, "loss": 1.2949, "step": 4800 }, { "epoch": 0.77, "grad_norm": 8.607007026672363, "learning_rate": 2.5866053055106732e-06, "loss": 1.503, "step": 4801 }, { "epoch": 0.77, "grad_norm": 7.774867057800293, "learning_rate": 2.583106864813518e-06, "loss": 1.2848, "step": 4802 }, { "epoch": 0.77, "grad_norm": 7.578629016876221, "learning_rate": 2.57961044066014e-06, "loss": 1.4391, "step": 4803 }, { "epoch": 0.77, "grad_norm": 9.29443645477295, "learning_rate": 2.576116034001166e-06, "loss": 1.3602, "step": 4804 }, { "epoch": 0.77, "grad_norm": 6.99123477935791, "learning_rate": 2.572623645786673e-06, "loss": 1.0369, "step": 4805 }, { "epoch": 0.77, "grad_norm": 8.361458778381348, "learning_rate": 2.5691332769661903e-06, "loss": 1.2392, "step": 4806 }, { "epoch": 0.77, "grad_norm": 8.51903247833252, "learning_rate": 2.5656449284886973e-06, "loss": 1.0431, "step": 4807 }, { "epoch": 0.77, "grad_norm": 8.519292831420898, "learning_rate": 2.5621586013026257e-06, "loss": 1.3057, "step": 4808 }, { "epoch": 0.77, "grad_norm": 7.5964531898498535, "learning_rate": 2.558674296355855e-06, "loss": 1.267, "step": 4809 }, { "epoch": 0.77, "grad_norm": 8.853778839111328, "learning_rate": 2.5551920145957167e-06, "loss": 1.3557, "step": 4810 }, { "epoch": 0.77, "grad_norm": 8.198176383972168, "learning_rate": 2.551711756968992e-06, "loss": 1.6451, "step": 4811 }, { "epoch": 0.77, "grad_norm": 9.995356559753418, "learning_rate": 2.5482335244219114e-06, "loss": 1.3022, "step": 4812 }, { "epoch": 0.77, "grad_norm": 8.177289009094238, "learning_rate": 2.544757317900155e-06, "loss": 1.4409, "step": 4813 }, { "epoch": 0.77, "grad_norm": 7.1001386642456055, "learning_rate": 2.5412831383488524e-06, "loss": 1.0968, "step": 4814 }, { "epoch": 0.78, "grad_norm": 9.130339622497559, "learning_rate": 2.537810986712582e-06, "loss": 1.4049, "step": 4815 }, { "epoch": 0.78, "grad_norm": 6.987699508666992, "learning_rate": 2.5343408639353693e-06, "loss": 1.2106, "step": 4816 }, { "epoch": 0.78, "grad_norm": 8.308408737182617, "learning_rate": 2.5308727709606907e-06, "loss": 1.3111, "step": 4817 }, { "epoch": 0.78, "grad_norm": 7.275054931640625, "learning_rate": 2.527406708731468e-06, "loss": 1.2428, "step": 4818 }, { "epoch": 0.78, "grad_norm": 7.8793439865112305, "learning_rate": 2.5239426781900743e-06, "loss": 1.5696, "step": 4819 }, { "epoch": 0.78, "grad_norm": 7.879265785217285, "learning_rate": 2.5204806802783266e-06, "loss": 1.368, "step": 4820 }, { "epoch": 0.78, "grad_norm": 6.950896263122559, "learning_rate": 2.5170207159374928e-06, "loss": 1.2287, "step": 4821 }, { "epoch": 0.78, "grad_norm": 8.107029914855957, "learning_rate": 2.5135627861082845e-06, "loss": 1.3883, "step": 4822 }, { "epoch": 0.78, "grad_norm": 8.336609840393066, "learning_rate": 2.5101068917308637e-06, "loss": 1.0332, "step": 4823 }, { "epoch": 0.78, "grad_norm": 9.211649894714355, "learning_rate": 2.506653033744835e-06, "loss": 1.4916, "step": 4824 }, { "epoch": 0.78, "grad_norm": 8.15517807006836, "learning_rate": 2.5032012130892524e-06, "loss": 1.4714, "step": 4825 }, { "epoch": 0.78, "grad_norm": 6.6738362312316895, "learning_rate": 2.499751430702615e-06, "loss": 1.0356, "step": 4826 }, { "epoch": 0.78, "grad_norm": 8.947218894958496, "learning_rate": 2.4963036875228676e-06, "loss": 1.4299, "step": 4827 }, { "epoch": 0.78, "grad_norm": 8.280454635620117, "learning_rate": 2.4928579844874e-06, "loss": 1.3235, "step": 4828 }, { "epoch": 0.78, "grad_norm": 8.508090019226074, "learning_rate": 2.489414322533049e-06, "loss": 1.4439, "step": 4829 }, { "epoch": 0.78, "grad_norm": 6.162621021270752, "learning_rate": 2.4859727025960955e-06, "loss": 1.0355, "step": 4830 }, { "epoch": 0.78, "grad_norm": 9.888999938964844, "learning_rate": 2.482533125612265e-06, "loss": 1.5899, "step": 4831 }, { "epoch": 0.78, "grad_norm": 8.556652069091797, "learning_rate": 2.479095592516727e-06, "loss": 1.1311, "step": 4832 }, { "epoch": 0.78, "grad_norm": 6.585271835327148, "learning_rate": 2.4756601042440975e-06, "loss": 1.0266, "step": 4833 }, { "epoch": 0.78, "grad_norm": 8.598305702209473, "learning_rate": 2.472226661728433e-06, "loss": 1.2858, "step": 4834 }, { "epoch": 0.78, "grad_norm": 8.399184226989746, "learning_rate": 2.468795265903238e-06, "loss": 1.2784, "step": 4835 }, { "epoch": 0.78, "grad_norm": 7.998427391052246, "learning_rate": 2.4653659177014564e-06, "loss": 1.4061, "step": 4836 }, { "epoch": 0.78, "grad_norm": 7.605798244476318, "learning_rate": 2.4619386180554783e-06, "loss": 1.3013, "step": 4837 }, { "epoch": 0.78, "grad_norm": 7.1872053146362305, "learning_rate": 2.4585133678971364e-06, "loss": 1.0683, "step": 4838 }, { "epoch": 0.78, "grad_norm": 6.620077133178711, "learning_rate": 2.455090168157702e-06, "loss": 1.0146, "step": 4839 }, { "epoch": 0.78, "grad_norm": 10.045721054077148, "learning_rate": 2.4516690197679016e-06, "loss": 1.43, "step": 4840 }, { "epoch": 0.78, "grad_norm": 9.32962417602539, "learning_rate": 2.448249923657887e-06, "loss": 1.3476, "step": 4841 }, { "epoch": 0.78, "grad_norm": 7.615710735321045, "learning_rate": 2.4448328807572607e-06, "loss": 1.1958, "step": 4842 }, { "epoch": 0.78, "grad_norm": 8.128190040588379, "learning_rate": 2.4414178919950683e-06, "loss": 1.2351, "step": 4843 }, { "epoch": 0.78, "grad_norm": 7.681344032287598, "learning_rate": 2.438004958299791e-06, "loss": 1.3532, "step": 4844 }, { "epoch": 0.78, "grad_norm": 8.698037147521973, "learning_rate": 2.434594080599364e-06, "loss": 1.1661, "step": 4845 }, { "epoch": 0.78, "grad_norm": 8.683711051940918, "learning_rate": 2.4311852598211473e-06, "loss": 1.1533, "step": 4846 }, { "epoch": 0.78, "grad_norm": 8.067079544067383, "learning_rate": 2.4277784968919483e-06, "loss": 1.3596, "step": 4847 }, { "epoch": 0.78, "grad_norm": 7.723513126373291, "learning_rate": 2.4243737927380196e-06, "loss": 1.2055, "step": 4848 }, { "epoch": 0.78, "grad_norm": 8.283839225769043, "learning_rate": 2.420971148285044e-06, "loss": 1.326, "step": 4849 }, { "epoch": 0.78, "grad_norm": 9.232552528381348, "learning_rate": 2.4175705644581604e-06, "loss": 1.4665, "step": 4850 }, { "epoch": 0.78, "grad_norm": 7.615983009338379, "learning_rate": 2.4141720421819293e-06, "loss": 1.1958, "step": 4851 }, { "epoch": 0.78, "grad_norm": 8.17577838897705, "learning_rate": 2.410775582380358e-06, "loss": 1.2092, "step": 4852 }, { "epoch": 0.78, "grad_norm": 9.116792678833008, "learning_rate": 2.4073811859769016e-06, "loss": 1.3402, "step": 4853 }, { "epoch": 0.78, "grad_norm": 8.886439323425293, "learning_rate": 2.4039888538944358e-06, "loss": 1.3833, "step": 4854 }, { "epoch": 0.78, "grad_norm": 8.090376853942871, "learning_rate": 2.4005985870552973e-06, "loss": 1.2128, "step": 4855 }, { "epoch": 0.78, "grad_norm": 8.540779113769531, "learning_rate": 2.3972103863812414e-06, "loss": 1.2823, "step": 4856 }, { "epoch": 0.78, "grad_norm": 9.774606704711914, "learning_rate": 2.3938242527934684e-06, "loss": 1.2986, "step": 4857 }, { "epoch": 0.78, "grad_norm": 8.098212242126465, "learning_rate": 2.390440187212627e-06, "loss": 1.3325, "step": 4858 }, { "epoch": 0.78, "grad_norm": 7.538463592529297, "learning_rate": 2.3870581905587852e-06, "loss": 1.3134, "step": 4859 }, { "epoch": 0.78, "grad_norm": 7.897462844848633, "learning_rate": 2.3836782637514666e-06, "loss": 1.2004, "step": 4860 }, { "epoch": 0.78, "grad_norm": 9.192386627197266, "learning_rate": 2.380300407709617e-06, "loss": 1.2087, "step": 4861 }, { "epoch": 0.78, "grad_norm": 9.753692626953125, "learning_rate": 2.3769246233516243e-06, "loss": 1.2313, "step": 4862 }, { "epoch": 0.78, "grad_norm": 10.031174659729004, "learning_rate": 2.3735509115953226e-06, "loss": 1.2998, "step": 4863 }, { "epoch": 0.78, "grad_norm": 8.136509895324707, "learning_rate": 2.3701792733579645e-06, "loss": 1.2555, "step": 4864 }, { "epoch": 0.78, "grad_norm": 7.617273807525635, "learning_rate": 2.366809709556259e-06, "loss": 1.4699, "step": 4865 }, { "epoch": 0.78, "grad_norm": 7.949397087097168, "learning_rate": 2.3634422211063334e-06, "loss": 1.2814, "step": 4866 }, { "epoch": 0.78, "grad_norm": 8.542193412780762, "learning_rate": 2.360076808923757e-06, "loss": 1.2958, "step": 4867 }, { "epoch": 0.78, "grad_norm": 7.525131702423096, "learning_rate": 2.356713473923544e-06, "loss": 1.4106, "step": 4868 }, { "epoch": 0.78, "grad_norm": 8.120185852050781, "learning_rate": 2.353352217020125e-06, "loss": 1.3466, "step": 4869 }, { "epoch": 0.78, "grad_norm": 8.042093276977539, "learning_rate": 2.349993039127387e-06, "loss": 1.2373, "step": 4870 }, { "epoch": 0.78, "grad_norm": 8.987281799316406, "learning_rate": 2.3466359411586327e-06, "loss": 1.2832, "step": 4871 }, { "epoch": 0.78, "grad_norm": 9.077136039733887, "learning_rate": 2.3432809240266073e-06, "loss": 1.198, "step": 4872 }, { "epoch": 0.78, "grad_norm": 10.245779037475586, "learning_rate": 2.339927988643499e-06, "loss": 1.346, "step": 4873 }, { "epoch": 0.78, "grad_norm": 9.191503524780273, "learning_rate": 2.336577135920911e-06, "loss": 1.5939, "step": 4874 }, { "epoch": 0.78, "grad_norm": 8.213257789611816, "learning_rate": 2.3332283667698997e-06, "loss": 1.2919, "step": 4875 }, { "epoch": 0.78, "grad_norm": 8.815072059631348, "learning_rate": 2.3298816821009373e-06, "loss": 1.1559, "step": 4876 }, { "epoch": 0.79, "grad_norm": 8.255131721496582, "learning_rate": 2.3265370828239442e-06, "loss": 1.1162, "step": 4877 }, { "epoch": 0.79, "grad_norm": 8.314403533935547, "learning_rate": 2.3231945698482693e-06, "loss": 1.458, "step": 4878 }, { "epoch": 0.79, "grad_norm": 7.822061538696289, "learning_rate": 2.3198541440826837e-06, "loss": 1.2762, "step": 4879 }, { "epoch": 0.79, "grad_norm": 7.877317428588867, "learning_rate": 2.3165158064354077e-06, "loss": 1.4816, "step": 4880 }, { "epoch": 0.79, "grad_norm": 6.9745001792907715, "learning_rate": 2.3131795578140838e-06, "loss": 1.4477, "step": 4881 }, { "epoch": 0.79, "grad_norm": 8.98854923248291, "learning_rate": 2.3098453991257873e-06, "loss": 1.0889, "step": 4882 }, { "epoch": 0.79, "grad_norm": 8.174982070922852, "learning_rate": 2.3065133312770315e-06, "loss": 1.259, "step": 4883 }, { "epoch": 0.79, "grad_norm": 8.507893562316895, "learning_rate": 2.303183355173748e-06, "loss": 1.1868, "step": 4884 }, { "epoch": 0.79, "grad_norm": 7.687399864196777, "learning_rate": 2.299855471721316e-06, "loss": 1.1623, "step": 4885 }, { "epoch": 0.79, "grad_norm": 8.461949348449707, "learning_rate": 2.296529681824534e-06, "loss": 1.3191, "step": 4886 }, { "epoch": 0.79, "grad_norm": 7.502328872680664, "learning_rate": 2.2932059863876364e-06, "loss": 1.3468, "step": 4887 }, { "epoch": 0.79, "grad_norm": 8.682802200317383, "learning_rate": 2.2898843863142905e-06, "loss": 1.2955, "step": 4888 }, { "epoch": 0.79, "grad_norm": 6.721014499664307, "learning_rate": 2.2865648825075804e-06, "loss": 1.1891, "step": 4889 }, { "epoch": 0.79, "grad_norm": 7.779499530792236, "learning_rate": 2.2832474758700383e-06, "loss": 1.3127, "step": 4890 }, { "epoch": 0.79, "grad_norm": 7.734772205352783, "learning_rate": 2.2799321673036167e-06, "loss": 1.3317, "step": 4891 }, { "epoch": 0.79, "grad_norm": 7.265625953674316, "learning_rate": 2.276618957709699e-06, "loss": 1.0694, "step": 4892 }, { "epoch": 0.79, "grad_norm": 7.848859786987305, "learning_rate": 2.2733078479891004e-06, "loss": 1.1738, "step": 4893 }, { "epoch": 0.79, "grad_norm": 8.063349723815918, "learning_rate": 2.2699988390420535e-06, "loss": 1.5466, "step": 4894 }, { "epoch": 0.79, "grad_norm": 7.605988025665283, "learning_rate": 2.266691931768239e-06, "loss": 1.106, "step": 4895 }, { "epoch": 0.79, "grad_norm": 7.969528675079346, "learning_rate": 2.263387127066752e-06, "loss": 1.0752, "step": 4896 }, { "epoch": 0.79, "grad_norm": 6.895132541656494, "learning_rate": 2.2600844258361197e-06, "loss": 1.376, "step": 4897 }, { "epoch": 0.79, "grad_norm": 8.512279510498047, "learning_rate": 2.256783828974304e-06, "loss": 1.2833, "step": 4898 }, { "epoch": 0.79, "grad_norm": 9.298395156860352, "learning_rate": 2.253485337378676e-06, "loss": 1.1118, "step": 4899 }, { "epoch": 0.79, "grad_norm": 7.738555908203125, "learning_rate": 2.2501889519460574e-06, "loss": 1.002, "step": 4900 }, { "epoch": 0.79, "grad_norm": 7.856217861175537, "learning_rate": 2.2468946735726825e-06, "loss": 1.2674, "step": 4901 }, { "epoch": 0.79, "grad_norm": 8.010135650634766, "learning_rate": 2.2436025031542196e-06, "loss": 1.4716, "step": 4902 }, { "epoch": 0.79, "grad_norm": 9.644628524780273, "learning_rate": 2.240312441585758e-06, "loss": 1.4605, "step": 4903 }, { "epoch": 0.79, "grad_norm": 6.730765342712402, "learning_rate": 2.23702448976182e-06, "loss": 0.9271, "step": 4904 }, { "epoch": 0.79, "grad_norm": 8.67628288269043, "learning_rate": 2.233738648576349e-06, "loss": 1.355, "step": 4905 }, { "epoch": 0.79, "grad_norm": 7.2683329582214355, "learning_rate": 2.2304549189227177e-06, "loss": 1.269, "step": 4906 }, { "epoch": 0.79, "grad_norm": 7.803089618682861, "learning_rate": 2.2271733016937226e-06, "loss": 1.1603, "step": 4907 }, { "epoch": 0.79, "grad_norm": 9.951798439025879, "learning_rate": 2.2238937977815876e-06, "loss": 1.236, "step": 4908 }, { "epoch": 0.79, "grad_norm": 9.318193435668945, "learning_rate": 2.2206164080779624e-06, "loss": 1.3939, "step": 4909 }, { "epoch": 0.79, "grad_norm": 10.017553329467773, "learning_rate": 2.2173411334739204e-06, "loss": 1.3925, "step": 4910 }, { "epoch": 0.79, "grad_norm": 8.80438232421875, "learning_rate": 2.21406797485996e-06, "loss": 1.4653, "step": 4911 }, { "epoch": 0.79, "grad_norm": 10.338260650634766, "learning_rate": 2.210796933126005e-06, "loss": 1.261, "step": 4912 }, { "epoch": 0.79, "grad_norm": 8.917123794555664, "learning_rate": 2.2075280091614024e-06, "loss": 1.3582, "step": 4913 }, { "epoch": 0.79, "grad_norm": 7.78106164932251, "learning_rate": 2.2042612038549262e-06, "loss": 1.1266, "step": 4914 }, { "epoch": 0.79, "grad_norm": 8.327191352844238, "learning_rate": 2.20099651809477e-06, "loss": 1.2451, "step": 4915 }, { "epoch": 0.79, "grad_norm": 9.692105293273926, "learning_rate": 2.197733952768557e-06, "loss": 1.4244, "step": 4916 }, { "epoch": 0.79, "grad_norm": 10.14787483215332, "learning_rate": 2.1944735087633283e-06, "loss": 1.0905, "step": 4917 }, { "epoch": 0.79, "grad_norm": 8.284232139587402, "learning_rate": 2.1912151869655508e-06, "loss": 1.2752, "step": 4918 }, { "epoch": 0.79, "grad_norm": 9.668457984924316, "learning_rate": 2.187958988261114e-06, "loss": 1.4377, "step": 4919 }, { "epoch": 0.79, "grad_norm": 10.072100639343262, "learning_rate": 2.1847049135353303e-06, "loss": 1.3115, "step": 4920 }, { "epoch": 0.79, "grad_norm": 9.126220703125, "learning_rate": 2.1814529636729354e-06, "loss": 1.2512, "step": 4921 }, { "epoch": 0.79, "grad_norm": 6.259712219238281, "learning_rate": 2.178203139558086e-06, "loss": 1.165, "step": 4922 }, { "epoch": 0.79, "grad_norm": 9.515402793884277, "learning_rate": 2.1749554420743613e-06, "loss": 1.2122, "step": 4923 }, { "epoch": 0.79, "grad_norm": 9.396657943725586, "learning_rate": 2.1717098721047624e-06, "loss": 1.4133, "step": 4924 }, { "epoch": 0.79, "grad_norm": 9.254739761352539, "learning_rate": 2.1684664305317114e-06, "loss": 1.2064, "step": 4925 }, { "epoch": 0.79, "grad_norm": 8.169998168945312, "learning_rate": 2.165225118237053e-06, "loss": 1.1025, "step": 4926 }, { "epoch": 0.79, "grad_norm": 7.286832809448242, "learning_rate": 2.1619859361020514e-06, "loss": 1.1007, "step": 4927 }, { "epoch": 0.79, "grad_norm": 8.927978515625, "learning_rate": 2.1587488850073924e-06, "loss": 1.4577, "step": 4928 }, { "epoch": 0.79, "grad_norm": 8.655777931213379, "learning_rate": 2.1555139658331824e-06, "loss": 1.3057, "step": 4929 }, { "epoch": 0.79, "grad_norm": 8.209604263305664, "learning_rate": 2.1522811794589493e-06, "loss": 1.3232, "step": 4930 }, { "epoch": 0.79, "grad_norm": 9.162894248962402, "learning_rate": 2.1490505267636385e-06, "loss": 1.1176, "step": 4931 }, { "epoch": 0.79, "grad_norm": 8.147102355957031, "learning_rate": 2.1458220086256186e-06, "loss": 1.0746, "step": 4932 }, { "epoch": 0.79, "grad_norm": 9.708927154541016, "learning_rate": 2.142595625922671e-06, "loss": 1.2236, "step": 4933 }, { "epoch": 0.79, "grad_norm": 14.698881149291992, "learning_rate": 2.1393713795320128e-06, "loss": 0.9992, "step": 4934 }, { "epoch": 0.79, "grad_norm": 6.417976379394531, "learning_rate": 2.136149270330257e-06, "loss": 1.2947, "step": 4935 }, { "epoch": 0.79, "grad_norm": 9.769158363342285, "learning_rate": 2.1329292991934526e-06, "loss": 1.3647, "step": 4936 }, { "epoch": 0.79, "grad_norm": 8.086031913757324, "learning_rate": 2.129711466997062e-06, "loss": 1.4678, "step": 4937 }, { "epoch": 0.79, "grad_norm": 6.498185634613037, "learning_rate": 2.126495774615964e-06, "loss": 1.1573, "step": 4938 }, { "epoch": 0.8, "grad_norm": 7.3241753578186035, "learning_rate": 2.1232822229244655e-06, "loss": 1.1077, "step": 4939 }, { "epoch": 0.8, "grad_norm": 9.122727394104004, "learning_rate": 2.1200708127962743e-06, "loss": 1.2541, "step": 4940 }, { "epoch": 0.8, "grad_norm": 8.460481643676758, "learning_rate": 2.1168615451045304e-06, "loss": 1.1303, "step": 4941 }, { "epoch": 0.8, "grad_norm": 8.589034080505371, "learning_rate": 2.1136544207217856e-06, "loss": 1.6231, "step": 4942 }, { "epoch": 0.8, "grad_norm": 9.02236557006836, "learning_rate": 2.110449440520006e-06, "loss": 1.2123, "step": 4943 }, { "epoch": 0.8, "grad_norm": 6.093847751617432, "learning_rate": 2.1072466053705874e-06, "loss": 1.084, "step": 4944 }, { "epoch": 0.8, "grad_norm": 9.890849113464355, "learning_rate": 2.104045916144325e-06, "loss": 1.5125, "step": 4945 }, { "epoch": 0.8, "grad_norm": 8.095799446105957, "learning_rate": 2.100847373711442e-06, "loss": 1.3592, "step": 4946 }, { "epoch": 0.8, "grad_norm": 8.24221420288086, "learning_rate": 2.0976509789415755e-06, "loss": 1.1222, "step": 4947 }, { "epoch": 0.8, "grad_norm": 8.933514595031738, "learning_rate": 2.0944567327037736e-06, "loss": 1.0265, "step": 4948 }, { "epoch": 0.8, "grad_norm": 10.764974594116211, "learning_rate": 2.0912646358665146e-06, "loss": 1.4281, "step": 4949 }, { "epoch": 0.8, "grad_norm": 8.045713424682617, "learning_rate": 2.0880746892976723e-06, "loss": 1.2285, "step": 4950 }, { "epoch": 0.8, "grad_norm": 8.94100570678711, "learning_rate": 2.0848868938645504e-06, "loss": 1.4196, "step": 4951 }, { "epoch": 0.8, "grad_norm": 8.40652847290039, "learning_rate": 2.0817012504338642e-06, "loss": 1.275, "step": 4952 }, { "epoch": 0.8, "grad_norm": 8.283065795898438, "learning_rate": 2.0785177598717386e-06, "loss": 1.214, "step": 4953 }, { "epoch": 0.8, "grad_norm": 8.155344009399414, "learning_rate": 2.0753364230437255e-06, "loss": 1.313, "step": 4954 }, { "epoch": 0.8, "grad_norm": 8.996586799621582, "learning_rate": 2.072157240814775e-06, "loss": 1.4267, "step": 4955 }, { "epoch": 0.8, "grad_norm": 8.785477638244629, "learning_rate": 2.0689802140492664e-06, "loss": 1.2194, "step": 4956 }, { "epoch": 0.8, "grad_norm": 8.195290565490723, "learning_rate": 2.065805343610987e-06, "loss": 1.3107, "step": 4957 }, { "epoch": 0.8, "grad_norm": 7.58887243270874, "learning_rate": 2.062632630363127e-06, "loss": 1.2567, "step": 4958 }, { "epoch": 0.8, "grad_norm": 8.776372909545898, "learning_rate": 2.059462075168315e-06, "loss": 1.2929, "step": 4959 }, { "epoch": 0.8, "grad_norm": 6.981733322143555, "learning_rate": 2.0562936788885637e-06, "loss": 1.1857, "step": 4960 }, { "epoch": 0.8, "grad_norm": 6.884397506713867, "learning_rate": 2.053127442385323e-06, "loss": 1.4581, "step": 4961 }, { "epoch": 0.8, "grad_norm": 10.294380187988281, "learning_rate": 2.049963366519446e-06, "loss": 1.2546, "step": 4962 }, { "epoch": 0.8, "grad_norm": 7.800389766693115, "learning_rate": 2.04680145215119e-06, "loss": 1.1757, "step": 4963 }, { "epoch": 0.8, "grad_norm": 7.856297016143799, "learning_rate": 2.043641700140243e-06, "loss": 1.2259, "step": 4964 }, { "epoch": 0.8, "grad_norm": 7.537790298461914, "learning_rate": 2.0404841113456854e-06, "loss": 1.1135, "step": 4965 }, { "epoch": 0.8, "grad_norm": 9.585786819458008, "learning_rate": 2.0373286866260256e-06, "loss": 1.4449, "step": 4966 }, { "epoch": 0.8, "grad_norm": 7.689888000488281, "learning_rate": 2.034175426839178e-06, "loss": 1.1838, "step": 4967 }, { "epoch": 0.8, "grad_norm": 7.3127899169921875, "learning_rate": 2.0310243328424597e-06, "loss": 1.1683, "step": 4968 }, { "epoch": 0.8, "grad_norm": 7.8288726806640625, "learning_rate": 2.0278754054926154e-06, "loss": 1.4831, "step": 4969 }, { "epoch": 0.8, "grad_norm": 8.409416198730469, "learning_rate": 2.0247286456457837e-06, "loss": 1.3826, "step": 4970 }, { "epoch": 0.8, "grad_norm": 10.43065071105957, "learning_rate": 2.021584054157528e-06, "loss": 1.3181, "step": 4971 }, { "epoch": 0.8, "grad_norm": 9.307397842407227, "learning_rate": 2.0184416318828183e-06, "loss": 1.1626, "step": 4972 }, { "epoch": 0.8, "grad_norm": 10.018138885498047, "learning_rate": 2.0153013796760234e-06, "loss": 1.3733, "step": 4973 }, { "epoch": 0.8, "grad_norm": 9.003579139709473, "learning_rate": 2.0121632983909434e-06, "loss": 1.0346, "step": 4974 }, { "epoch": 0.8, "grad_norm": 7.663578033447266, "learning_rate": 2.0090273888807643e-06, "loss": 0.9527, "step": 4975 }, { "epoch": 0.8, "grad_norm": 7.311576843261719, "learning_rate": 2.005893651998103e-06, "loss": 1.0218, "step": 4976 }, { "epoch": 0.8, "grad_norm": 8.831402778625488, "learning_rate": 2.002762088594975e-06, "loss": 1.2051, "step": 4977 }, { "epoch": 0.8, "grad_norm": 7.8158345222473145, "learning_rate": 1.999632699522799e-06, "loss": 1.3356, "step": 4978 }, { "epoch": 0.8, "grad_norm": 9.078507423400879, "learning_rate": 1.9965054856324174e-06, "loss": 1.3926, "step": 4979 }, { "epoch": 0.8, "grad_norm": 7.8506317138671875, "learning_rate": 1.9933804477740705e-06, "loss": 1.2046, "step": 4980 }, { "epoch": 0.8, "grad_norm": 8.509284019470215, "learning_rate": 1.9902575867974107e-06, "loss": 1.4182, "step": 4981 }, { "epoch": 0.8, "grad_norm": 8.441496849060059, "learning_rate": 1.9871369035515e-06, "loss": 1.4787, "step": 4982 }, { "epoch": 0.8, "grad_norm": 7.924543380737305, "learning_rate": 1.9840183988847983e-06, "loss": 1.3139, "step": 4983 }, { "epoch": 0.8, "grad_norm": 8.511385917663574, "learning_rate": 1.980902073645189e-06, "loss": 1.3088, "step": 4984 }, { "epoch": 0.8, "grad_norm": 7.458401203155518, "learning_rate": 1.9777879286799517e-06, "loss": 1.3478, "step": 4985 }, { "epoch": 0.8, "grad_norm": 7.610616683959961, "learning_rate": 1.974675964835777e-06, "loss": 1.0987, "step": 4986 }, { "epoch": 0.8, "grad_norm": 6.714752197265625, "learning_rate": 1.9715661829587653e-06, "loss": 1.0286, "step": 4987 }, { "epoch": 0.8, "grad_norm": 8.32361888885498, "learning_rate": 1.9684585838944114e-06, "loss": 1.2153, "step": 4988 }, { "epoch": 0.8, "grad_norm": 9.5364408493042, "learning_rate": 1.9653531684876337e-06, "loss": 1.3308, "step": 4989 }, { "epoch": 0.8, "grad_norm": 8.039544105529785, "learning_rate": 1.962249937582746e-06, "loss": 1.1071, "step": 4990 }, { "epoch": 0.8, "grad_norm": 8.405479431152344, "learning_rate": 1.9591488920234725e-06, "loss": 1.1407, "step": 4991 }, { "epoch": 0.8, "grad_norm": 7.37800407409668, "learning_rate": 1.95605003265294e-06, "loss": 1.2357, "step": 4992 }, { "epoch": 0.8, "grad_norm": 7.99930477142334, "learning_rate": 1.952953360313683e-06, "loss": 1.2609, "step": 4993 }, { "epoch": 0.8, "grad_norm": 8.821211814880371, "learning_rate": 1.9498588758476413e-06, "loss": 1.2247, "step": 4994 }, { "epoch": 0.8, "grad_norm": 7.683000564575195, "learning_rate": 1.9467665800961586e-06, "loss": 1.4378, "step": 4995 }, { "epoch": 0.8, "grad_norm": 7.703995227813721, "learning_rate": 1.943676473899987e-06, "loss": 1.2855, "step": 4996 }, { "epoch": 0.8, "grad_norm": 8.740386962890625, "learning_rate": 1.9405885580992777e-06, "loss": 1.3646, "step": 4997 }, { "epoch": 0.8, "grad_norm": 10.126374244689941, "learning_rate": 1.937502833533591e-06, "loss": 1.2641, "step": 4998 }, { "epoch": 0.8, "grad_norm": 8.223852157592773, "learning_rate": 1.93441930104189e-06, "loss": 1.0593, "step": 4999 }, { "epoch": 0.8, "grad_norm": 8.16995620727539, "learning_rate": 1.9313379614625405e-06, "loss": 1.2563, "step": 5000 }, { "epoch": 0.81, "grad_norm": 6.843425750732422, "learning_rate": 1.7166272877299937e-05, "loss": 1.4149, "step": 5001 }, { "epoch": 0.81, "grad_norm": 9.760581970214844, "learning_rate": 1.7165060596204065e-05, "loss": 1.0854, "step": 5002 }, { "epoch": 0.81, "grad_norm": 8.299047470092773, "learning_rate": 1.7163848098679448e-05, "loss": 1.214, "step": 5003 }, { "epoch": 0.81, "grad_norm": 6.708827018737793, "learning_rate": 1.716263538476271e-05, "loss": 1.1543, "step": 5004 }, { "epoch": 0.81, "grad_norm": 7.964986324310303, "learning_rate": 1.7161422454490474e-05, "loss": 1.4007, "step": 5005 }, { "epoch": 0.81, "grad_norm": 7.368851184844971, "learning_rate": 1.716020930789939e-05, "loss": 1.2785, "step": 5006 }, { "epoch": 0.81, "grad_norm": 8.415141105651855, "learning_rate": 1.7158995945026097e-05, "loss": 1.5284, "step": 5007 }, { "epoch": 0.81, "grad_norm": 10.118163108825684, "learning_rate": 1.7157782365907243e-05, "loss": 1.3271, "step": 5008 }, { "epoch": 0.81, "grad_norm": 9.67189884185791, "learning_rate": 1.715656857057949e-05, "loss": 1.2776, "step": 5009 }, { "epoch": 0.81, "grad_norm": 7.984910011291504, "learning_rate": 1.7155354559079502e-05, "loss": 1.2917, "step": 5010 }, { "epoch": 0.81, "grad_norm": 9.1172456741333, "learning_rate": 1.7154140331443943e-05, "loss": 1.3166, "step": 5011 }, { "epoch": 0.81, "grad_norm": 7.229843616485596, "learning_rate": 1.71529258877095e-05, "loss": 1.4991, "step": 5012 }, { "epoch": 0.81, "grad_norm": 9.355494499206543, "learning_rate": 1.715171122791285e-05, "loss": 1.4613, "step": 5013 }, { "epoch": 0.81, "grad_norm": 9.72788143157959, "learning_rate": 1.7150496352090685e-05, "loss": 1.3179, "step": 5014 }, { "epoch": 0.81, "grad_norm": 6.949715614318848, "learning_rate": 1.7149281260279703e-05, "loss": 0.9385, "step": 5015 }, { "epoch": 0.81, "grad_norm": 8.335970878601074, "learning_rate": 1.7148065952516603e-05, "loss": 1.4213, "step": 5016 }, { "epoch": 0.81, "grad_norm": 9.731266021728516, "learning_rate": 1.7146850428838104e-05, "loss": 1.4621, "step": 5017 }, { "epoch": 0.81, "grad_norm": 7.8637919425964355, "learning_rate": 1.714563468928091e-05, "loss": 1.3756, "step": 5018 }, { "epoch": 0.81, "grad_norm": 7.063701629638672, "learning_rate": 1.714441873388175e-05, "loss": 1.3102, "step": 5019 }, { "epoch": 0.81, "grad_norm": 9.323596000671387, "learning_rate": 1.7143202562677354e-05, "loss": 1.299, "step": 5020 }, { "epoch": 0.81, "grad_norm": 7.6661858558654785, "learning_rate": 1.7141986175704457e-05, "loss": 1.0934, "step": 5021 }, { "epoch": 0.81, "grad_norm": 7.335458755493164, "learning_rate": 1.7140769572999803e-05, "loss": 1.1316, "step": 5022 }, { "epoch": 0.81, "grad_norm": 8.199723243713379, "learning_rate": 1.7139552754600134e-05, "loss": 1.3434, "step": 5023 }, { "epoch": 0.81, "grad_norm": 7.962564945220947, "learning_rate": 1.7138335720542218e-05, "loss": 1.1576, "step": 5024 }, { "epoch": 0.81, "grad_norm": 7.423969268798828, "learning_rate": 1.7137118470862805e-05, "loss": 1.2825, "step": 5025 }, { "epoch": 0.81, "grad_norm": 7.973435878753662, "learning_rate": 1.7135901005598667e-05, "loss": 1.3612, "step": 5026 }, { "epoch": 0.81, "grad_norm": 8.033288955688477, "learning_rate": 1.7134683324786578e-05, "loss": 1.4768, "step": 5027 }, { "epoch": 0.81, "grad_norm": 9.181100845336914, "learning_rate": 1.7133465428463324e-05, "loss": 1.4368, "step": 5028 }, { "epoch": 0.81, "grad_norm": 9.857210159301758, "learning_rate": 1.713224731666569e-05, "loss": 1.3558, "step": 5029 }, { "epoch": 0.81, "grad_norm": 7.699789524078369, "learning_rate": 1.713102898943047e-05, "loss": 1.3958, "step": 5030 }, { "epoch": 0.81, "grad_norm": 9.01565170288086, "learning_rate": 1.7129810446794464e-05, "loss": 1.3613, "step": 5031 }, { "epoch": 0.81, "grad_norm": 8.74122428894043, "learning_rate": 1.7128591688794483e-05, "loss": 1.4081, "step": 5032 }, { "epoch": 0.81, "grad_norm": 12.62136173248291, "learning_rate": 1.712737271546734e-05, "loss": 1.4953, "step": 5033 }, { "epoch": 0.81, "grad_norm": 7.658019542694092, "learning_rate": 1.712615352684985e-05, "loss": 1.2188, "step": 5034 }, { "epoch": 0.81, "grad_norm": 8.704947471618652, "learning_rate": 1.7124934122978846e-05, "loss": 1.3421, "step": 5035 }, { "epoch": 0.81, "grad_norm": 8.53943157196045, "learning_rate": 1.712371450389116e-05, "loss": 1.2549, "step": 5036 }, { "epoch": 0.81, "grad_norm": 11.244009971618652, "learning_rate": 1.7122494669623632e-05, "loss": 1.5973, "step": 5037 }, { "epoch": 0.81, "grad_norm": 7.133210182189941, "learning_rate": 1.712127462021311e-05, "loss": 1.267, "step": 5038 }, { "epoch": 0.81, "grad_norm": 7.585564136505127, "learning_rate": 1.712005435569644e-05, "loss": 1.2785, "step": 5039 }, { "epoch": 0.81, "grad_norm": 8.625540733337402, "learning_rate": 1.711883387611049e-05, "loss": 1.3586, "step": 5040 }, { "epoch": 0.81, "grad_norm": 7.411342620849609, "learning_rate": 1.7117613181492122e-05, "loss": 1.468, "step": 5041 }, { "epoch": 0.81, "grad_norm": 7.01547908782959, "learning_rate": 1.7116392271878212e-05, "loss": 1.3017, "step": 5042 }, { "epoch": 0.81, "grad_norm": 8.330269813537598, "learning_rate": 1.7115171147305632e-05, "loss": 1.2627, "step": 5043 }, { "epoch": 0.81, "grad_norm": 9.058889389038086, "learning_rate": 1.7113949807811273e-05, "loss": 1.4324, "step": 5044 }, { "epoch": 0.81, "grad_norm": 7.468283653259277, "learning_rate": 1.7112728253432024e-05, "loss": 1.0961, "step": 5045 }, { "epoch": 0.81, "grad_norm": 7.07023811340332, "learning_rate": 1.7111506484204785e-05, "loss": 1.2285, "step": 5046 }, { "epoch": 0.81, "grad_norm": 8.841772079467773, "learning_rate": 1.711028450016646e-05, "loss": 1.4532, "step": 5047 }, { "epoch": 0.81, "grad_norm": 7.873715400695801, "learning_rate": 1.710906230135396e-05, "loss": 1.1544, "step": 5048 }, { "epoch": 0.81, "grad_norm": 7.403555393218994, "learning_rate": 1.710783988780421e-05, "loss": 1.3445, "step": 5049 }, { "epoch": 0.81, "grad_norm": 8.532718658447266, "learning_rate": 1.7106617259554124e-05, "loss": 1.5342, "step": 5050 }, { "epoch": 0.81, "grad_norm": 7.581603527069092, "learning_rate": 1.710539441664064e-05, "loss": 1.1946, "step": 5051 }, { "epoch": 0.81, "grad_norm": 8.641804695129395, "learning_rate": 1.7104171359100687e-05, "loss": 1.4498, "step": 5052 }, { "epoch": 0.81, "grad_norm": 9.496636390686035, "learning_rate": 1.7102948086971222e-05, "loss": 1.3742, "step": 5053 }, { "epoch": 0.81, "grad_norm": 8.941265106201172, "learning_rate": 1.710172460028918e-05, "loss": 1.2537, "step": 5054 }, { "epoch": 0.81, "grad_norm": 8.477838516235352, "learning_rate": 1.710050089909153e-05, "loss": 1.1703, "step": 5055 }, { "epoch": 0.81, "grad_norm": 7.083213806152344, "learning_rate": 1.709927698341523e-05, "loss": 1.4364, "step": 5056 }, { "epoch": 0.81, "grad_norm": 9.446223258972168, "learning_rate": 1.709805285329725e-05, "loss": 1.3815, "step": 5057 }, { "epoch": 0.81, "grad_norm": 8.438838005065918, "learning_rate": 1.709682850877457e-05, "loss": 1.3796, "step": 5058 }, { "epoch": 0.81, "grad_norm": 7.505289554595947, "learning_rate": 1.7095603949884164e-05, "loss": 1.1941, "step": 5059 }, { "epoch": 0.81, "grad_norm": 7.556071758270264, "learning_rate": 1.7094379176663032e-05, "loss": 1.2269, "step": 5060 }, { "epoch": 0.81, "grad_norm": 7.0700812339782715, "learning_rate": 1.709315418914816e-05, "loss": 1.5916, "step": 5061 }, { "epoch": 0.81, "grad_norm": 8.6182279586792, "learning_rate": 1.7091928987376557e-05, "loss": 1.2824, "step": 5062 }, { "epoch": 0.82, "grad_norm": 7.897914409637451, "learning_rate": 1.709070357138523e-05, "loss": 1.3323, "step": 5063 }, { "epoch": 0.82, "grad_norm": 7.120639324188232, "learning_rate": 1.708947794121119e-05, "loss": 1.2176, "step": 5064 }, { "epoch": 0.82, "grad_norm": 6.801907539367676, "learning_rate": 1.7088252096891468e-05, "loss": 1.1713, "step": 5065 }, { "epoch": 0.82, "grad_norm": 9.164345741271973, "learning_rate": 1.708702603846308e-05, "loss": 1.3076, "step": 5066 }, { "epoch": 0.82, "grad_norm": 7.378202438354492, "learning_rate": 1.7085799765963064e-05, "loss": 1.4247, "step": 5067 }, { "epoch": 0.82, "grad_norm": 7.608414173126221, "learning_rate": 1.708457327942847e-05, "loss": 1.2519, "step": 5068 }, { "epoch": 0.82, "grad_norm": 9.345022201538086, "learning_rate": 1.7083346578896335e-05, "loss": 1.4285, "step": 5069 }, { "epoch": 0.82, "grad_norm": 8.739882469177246, "learning_rate": 1.7082119664403715e-05, "loss": 1.5276, "step": 5070 }, { "epoch": 0.82, "grad_norm": 7.976317882537842, "learning_rate": 1.7080892535987676e-05, "loss": 1.2943, "step": 5071 }, { "epoch": 0.82, "grad_norm": 7.3925323486328125, "learning_rate": 1.7079665193685274e-05, "loss": 1.3235, "step": 5072 }, { "epoch": 0.82, "grad_norm": 7.896756649017334, "learning_rate": 1.7078437637533595e-05, "loss": 1.3191, "step": 5073 }, { "epoch": 0.82, "grad_norm": 8.260765075683594, "learning_rate": 1.707720986756971e-05, "loss": 1.4169, "step": 5074 }, { "epoch": 0.82, "grad_norm": 9.699515342712402, "learning_rate": 1.7075981883830706e-05, "loss": 1.1705, "step": 5075 }, { "epoch": 0.82, "grad_norm": 8.933488845825195, "learning_rate": 1.7074753686353683e-05, "loss": 1.2491, "step": 5076 }, { "epoch": 0.82, "grad_norm": 8.39721393585205, "learning_rate": 1.707352527517573e-05, "loss": 1.3959, "step": 5077 }, { "epoch": 0.82, "grad_norm": 8.38569450378418, "learning_rate": 1.707229665033396e-05, "loss": 1.4402, "step": 5078 }, { "epoch": 0.82, "grad_norm": 7.8911871910095215, "learning_rate": 1.7071067811865477e-05, "loss": 1.2414, "step": 5079 }, { "epoch": 0.82, "grad_norm": 7.50200080871582, "learning_rate": 1.7069838759807406e-05, "loss": 1.4276, "step": 5080 }, { "epoch": 0.82, "grad_norm": 7.375244140625, "learning_rate": 1.7068609494196873e-05, "loss": 1.5528, "step": 5081 }, { "epoch": 0.82, "grad_norm": 7.2177653312683105, "learning_rate": 1.7067380015071004e-05, "loss": 1.0517, "step": 5082 }, { "epoch": 0.82, "grad_norm": 7.874471664428711, "learning_rate": 1.7066150322466943e-05, "loss": 1.6364, "step": 5083 }, { "epoch": 0.82, "grad_norm": 6.9945783615112305, "learning_rate": 1.706492041642183e-05, "loss": 1.3155, "step": 5084 }, { "epoch": 0.82, "grad_norm": 9.544767379760742, "learning_rate": 1.7063690296972814e-05, "loss": 1.3251, "step": 5085 }, { "epoch": 0.82, "grad_norm": 8.898012161254883, "learning_rate": 1.7062459964157057e-05, "loss": 1.4122, "step": 5086 }, { "epoch": 0.82, "grad_norm": 7.610021114349365, "learning_rate": 1.7061229418011716e-05, "loss": 1.2435, "step": 5087 }, { "epoch": 0.82, "grad_norm": 7.681457996368408, "learning_rate": 1.7059998658573968e-05, "loss": 1.5043, "step": 5088 }, { "epoch": 0.82, "grad_norm": 8.400896072387695, "learning_rate": 1.7058767685880992e-05, "loss": 1.2783, "step": 5089 }, { "epoch": 0.82, "grad_norm": 8.621236801147461, "learning_rate": 1.7057536499969958e-05, "loss": 1.406, "step": 5090 }, { "epoch": 0.82, "grad_norm": 6.759722709655762, "learning_rate": 1.7056305100878065e-05, "loss": 1.3677, "step": 5091 }, { "epoch": 0.82, "grad_norm": 8.408862113952637, "learning_rate": 1.7055073488642507e-05, "loss": 1.3307, "step": 5092 }, { "epoch": 0.82, "grad_norm": 7.494323253631592, "learning_rate": 1.7053841663300488e-05, "loss": 0.9958, "step": 5093 }, { "epoch": 0.82, "grad_norm": 7.838232517242432, "learning_rate": 1.7052609624889214e-05, "loss": 1.4614, "step": 5094 }, { "epoch": 0.82, "grad_norm": 9.237481117248535, "learning_rate": 1.7051377373445898e-05, "loss": 1.3994, "step": 5095 }, { "epoch": 0.82, "grad_norm": 9.006864547729492, "learning_rate": 1.7050144909007768e-05, "loss": 1.389, "step": 5096 }, { "epoch": 0.82, "grad_norm": 7.863646507263184, "learning_rate": 1.7048912231612047e-05, "loss": 1.393, "step": 5097 }, { "epoch": 0.82, "grad_norm": 7.335413455963135, "learning_rate": 1.704767934129597e-05, "loss": 1.5394, "step": 5098 }, { "epoch": 0.82, "grad_norm": 8.1514310836792, "learning_rate": 1.704644623809678e-05, "loss": 1.3607, "step": 5099 }, { "epoch": 0.82, "grad_norm": 7.197454452514648, "learning_rate": 1.7045212922051725e-05, "loss": 1.3461, "step": 5100 }, { "epoch": 0.82, "grad_norm": 8.855135917663574, "learning_rate": 1.704397939319805e-05, "loss": 1.3359, "step": 5101 }, { "epoch": 0.82, "grad_norm": 9.499317169189453, "learning_rate": 1.704274565157303e-05, "loss": 1.2298, "step": 5102 }, { "epoch": 0.82, "grad_norm": 8.242332458496094, "learning_rate": 1.7041511697213916e-05, "loss": 1.3873, "step": 5103 }, { "epoch": 0.82, "grad_norm": 8.690832138061523, "learning_rate": 1.704027753015799e-05, "loss": 1.053, "step": 5104 }, { "epoch": 0.82, "grad_norm": 7.307956695556641, "learning_rate": 1.703904315044253e-05, "loss": 1.3167, "step": 5105 }, { "epoch": 0.82, "grad_norm": 8.20458984375, "learning_rate": 1.7037808558104825e-05, "loss": 1.19, "step": 5106 }, { "epoch": 0.82, "grad_norm": 8.07159423828125, "learning_rate": 1.703657375318216e-05, "loss": 1.6841, "step": 5107 }, { "epoch": 0.82, "grad_norm": 8.171089172363281, "learning_rate": 1.703533873571184e-05, "loss": 1.284, "step": 5108 }, { "epoch": 0.82, "grad_norm": 8.918216705322266, "learning_rate": 1.7034103505731167e-05, "loss": 1.6411, "step": 5109 }, { "epoch": 0.82, "grad_norm": 9.394146919250488, "learning_rate": 1.7032868063277454e-05, "loss": 1.2853, "step": 5110 }, { "epoch": 0.82, "grad_norm": 7.843118667602539, "learning_rate": 1.7031632408388013e-05, "loss": 0.9398, "step": 5111 }, { "epoch": 0.82, "grad_norm": 7.777726650238037, "learning_rate": 1.703039654110018e-05, "loss": 1.5702, "step": 5112 }, { "epoch": 0.82, "grad_norm": 7.315724849700928, "learning_rate": 1.7029160461451275e-05, "loss": 1.5032, "step": 5113 }, { "epoch": 0.82, "grad_norm": 8.935653686523438, "learning_rate": 1.702792416947864e-05, "loss": 1.2592, "step": 5114 }, { "epoch": 0.82, "grad_norm": 9.985260009765625, "learning_rate": 1.7026687665219617e-05, "loss": 1.3752, "step": 5115 }, { "epoch": 0.82, "grad_norm": 7.592688083648682, "learning_rate": 1.7025450948711556e-05, "loss": 1.2991, "step": 5116 }, { "epoch": 0.82, "grad_norm": 7.494613170623779, "learning_rate": 1.7024214019991818e-05, "loss": 1.2594, "step": 5117 }, { "epoch": 0.82, "grad_norm": 7.762071132659912, "learning_rate": 1.7022976879097763e-05, "loss": 1.2653, "step": 5118 }, { "epoch": 0.82, "grad_norm": 7.115687847137451, "learning_rate": 1.7021739526066756e-05, "loss": 1.2308, "step": 5119 }, { "epoch": 0.82, "grad_norm": 9.052783012390137, "learning_rate": 1.702050196093618e-05, "loss": 1.2847, "step": 5120 }, { "epoch": 0.82, "grad_norm": 7.256692409515381, "learning_rate": 1.7019264183743404e-05, "loss": 1.3337, "step": 5121 }, { "epoch": 0.82, "grad_norm": 7.191842079162598, "learning_rate": 1.7018026194525832e-05, "loss": 1.3412, "step": 5122 }, { "epoch": 0.82, "grad_norm": 6.6616411209106445, "learning_rate": 1.7016787993320854e-05, "loss": 1.4359, "step": 5123 }, { "epoch": 0.82, "grad_norm": 8.97773265838623, "learning_rate": 1.701554958016587e-05, "loss": 1.1459, "step": 5124 }, { "epoch": 0.83, "grad_norm": 9.04466724395752, "learning_rate": 1.701431095509828e-05, "loss": 1.2792, "step": 5125 }, { "epoch": 0.83, "grad_norm": 10.360319137573242, "learning_rate": 1.7013072118155514e-05, "loss": 1.5011, "step": 5126 }, { "epoch": 0.83, "grad_norm": 8.52453899383545, "learning_rate": 1.7011833069374982e-05, "loss": 1.2481, "step": 5127 }, { "epoch": 0.83, "grad_norm": 8.108281135559082, "learning_rate": 1.7010593808794106e-05, "loss": 1.6944, "step": 5128 }, { "epoch": 0.83, "grad_norm": 7.6916303634643555, "learning_rate": 1.7009354336450334e-05, "loss": 1.2505, "step": 5129 }, { "epoch": 0.83, "grad_norm": 7.0082268714904785, "learning_rate": 1.7008114652381092e-05, "loss": 1.0656, "step": 5130 }, { "epoch": 0.83, "grad_norm": 9.59801197052002, "learning_rate": 1.7006874756623836e-05, "loss": 1.5253, "step": 5131 }, { "epoch": 0.83, "grad_norm": 8.096464157104492, "learning_rate": 1.7005634649216017e-05, "loss": 1.321, "step": 5132 }, { "epoch": 0.83, "grad_norm": 10.34772777557373, "learning_rate": 1.7004394330195086e-05, "loss": 1.3725, "step": 5133 }, { "epoch": 0.83, "grad_norm": 7.184013366699219, "learning_rate": 1.7003153799598514e-05, "loss": 1.3673, "step": 5134 }, { "epoch": 0.83, "grad_norm": 9.131834030151367, "learning_rate": 1.7001913057463772e-05, "loss": 1.5907, "step": 5135 }, { "epoch": 0.83, "grad_norm": 8.453859329223633, "learning_rate": 1.7000672103828338e-05, "loss": 1.5184, "step": 5136 }, { "epoch": 0.83, "grad_norm": 8.664766311645508, "learning_rate": 1.6999430938729692e-05, "loss": 1.2555, "step": 5137 }, { "epoch": 0.83, "grad_norm": 8.178474426269531, "learning_rate": 1.6998189562205335e-05, "loss": 1.3266, "step": 5138 }, { "epoch": 0.83, "grad_norm": 7.753713607788086, "learning_rate": 1.6996947974292758e-05, "loss": 1.3945, "step": 5139 }, { "epoch": 0.83, "grad_norm": 6.96235990524292, "learning_rate": 1.6995706175029462e-05, "loss": 1.2249, "step": 5140 }, { "epoch": 0.83, "grad_norm": 8.225560188293457, "learning_rate": 1.699446416445296e-05, "loss": 1.1815, "step": 5141 }, { "epoch": 0.83, "grad_norm": 7.2064642906188965, "learning_rate": 1.6993221942600774e-05, "loss": 1.0979, "step": 5142 }, { "epoch": 0.83, "grad_norm": 8.362914085388184, "learning_rate": 1.6991979509510415e-05, "loss": 1.1876, "step": 5143 }, { "epoch": 0.83, "grad_norm": 8.049640655517578, "learning_rate": 1.699073686521942e-05, "loss": 1.4095, "step": 5144 }, { "epoch": 0.83, "grad_norm": 9.518401145935059, "learning_rate": 1.6989494009765318e-05, "loss": 1.3964, "step": 5145 }, { "epoch": 0.83, "grad_norm": 7.862864971160889, "learning_rate": 1.6988250943185662e-05, "loss": 1.2734, "step": 5146 }, { "epoch": 0.83, "grad_norm": 8.059334754943848, "learning_rate": 1.6987007665517992e-05, "loss": 1.2523, "step": 5147 }, { "epoch": 0.83, "grad_norm": 8.5786714553833, "learning_rate": 1.6985764176799863e-05, "loss": 1.3643, "step": 5148 }, { "epoch": 0.83, "grad_norm": 7.383233547210693, "learning_rate": 1.6984520477068835e-05, "loss": 1.2367, "step": 5149 }, { "epoch": 0.83, "grad_norm": 7.775006294250488, "learning_rate": 1.698327656636248e-05, "loss": 1.3343, "step": 5150 }, { "epoch": 0.83, "grad_norm": 8.042798042297363, "learning_rate": 1.698203244471837e-05, "loss": 1.1485, "step": 5151 }, { "epoch": 0.83, "grad_norm": 6.851081848144531, "learning_rate": 1.6980788112174083e-05, "loss": 1.2969, "step": 5152 }, { "epoch": 0.83, "grad_norm": 8.471253395080566, "learning_rate": 1.6979543568767205e-05, "loss": 1.2271, "step": 5153 }, { "epoch": 0.83, "grad_norm": 8.195878028869629, "learning_rate": 1.6978298814535333e-05, "loss": 1.3738, "step": 5154 }, { "epoch": 0.83, "grad_norm": 8.122747421264648, "learning_rate": 1.697705384951606e-05, "loss": 1.3668, "step": 5155 }, { "epoch": 0.83, "grad_norm": 7.644906044006348, "learning_rate": 1.6975808673747e-05, "loss": 1.4768, "step": 5156 }, { "epoch": 0.83, "grad_norm": 9.490708351135254, "learning_rate": 1.6974563287265756e-05, "loss": 1.348, "step": 5157 }, { "epoch": 0.83, "grad_norm": 8.339503288269043, "learning_rate": 1.6973317690109955e-05, "loss": 1.1025, "step": 5158 }, { "epoch": 0.83, "grad_norm": 7.816207408905029, "learning_rate": 1.6972071882317216e-05, "loss": 1.2224, "step": 5159 }, { "epoch": 0.83, "grad_norm": 8.650991439819336, "learning_rate": 1.697082586392517e-05, "loss": 1.3745, "step": 5160 }, { "epoch": 0.83, "grad_norm": 7.64881706237793, "learning_rate": 1.696957963497146e-05, "loss": 1.3156, "step": 5161 }, { "epoch": 0.83, "grad_norm": 9.104010581970215, "learning_rate": 1.696833319549372e-05, "loss": 1.366, "step": 5162 }, { "epoch": 0.83, "grad_norm": 6.747280120849609, "learning_rate": 1.696708654552961e-05, "loss": 1.197, "step": 5163 }, { "epoch": 0.83, "grad_norm": 8.802167892456055, "learning_rate": 1.6965839685116778e-05, "loss": 1.4309, "step": 5164 }, { "epoch": 0.83, "grad_norm": 7.557948112487793, "learning_rate": 1.6964592614292893e-05, "loss": 1.4369, "step": 5165 }, { "epoch": 0.83, "grad_norm": 7.980496883392334, "learning_rate": 1.6963345333095622e-05, "loss": 1.0943, "step": 5166 }, { "epoch": 0.83, "grad_norm": 8.369779586791992, "learning_rate": 1.696209784156264e-05, "loss": 1.32, "step": 5167 }, { "epoch": 0.83, "grad_norm": 8.134007453918457, "learning_rate": 1.696085013973163e-05, "loss": 1.467, "step": 5168 }, { "epoch": 0.83, "grad_norm": 8.078629493713379, "learning_rate": 1.6959602227640276e-05, "loss": 1.5587, "step": 5169 }, { "epoch": 0.83, "grad_norm": 8.300760269165039, "learning_rate": 1.695835410532628e-05, "loss": 1.3618, "step": 5170 }, { "epoch": 0.83, "grad_norm": 8.761281967163086, "learning_rate": 1.6957105772827343e-05, "loss": 1.3011, "step": 5171 }, { "epoch": 0.83, "grad_norm": 7.314205646514893, "learning_rate": 1.6955857230181163e-05, "loss": 1.4118, "step": 5172 }, { "epoch": 0.83, "grad_norm": 9.02027416229248, "learning_rate": 1.6954608477425463e-05, "loss": 1.3039, "step": 5173 }, { "epoch": 0.83, "grad_norm": 8.181673049926758, "learning_rate": 1.6953359514597954e-05, "loss": 1.0796, "step": 5174 }, { "epoch": 0.83, "grad_norm": 9.073678016662598, "learning_rate": 1.695211034173637e-05, "loss": 1.4832, "step": 5175 }, { "epoch": 0.83, "grad_norm": 8.694235801696777, "learning_rate": 1.6950860958878445e-05, "loss": 1.265, "step": 5176 }, { "epoch": 0.83, "grad_norm": 7.678819179534912, "learning_rate": 1.6949611366061908e-05, "loss": 1.6252, "step": 5177 }, { "epoch": 0.83, "grad_norm": 9.553631782531738, "learning_rate": 1.6948361563324515e-05, "loss": 1.4139, "step": 5178 }, { "epoch": 0.83, "grad_norm": 7.59065580368042, "learning_rate": 1.6947111550704013e-05, "loss": 1.4567, "step": 5179 }, { "epoch": 0.83, "grad_norm": 7.60117244720459, "learning_rate": 1.694586132823816e-05, "loss": 1.3058, "step": 5180 }, { "epoch": 0.83, "grad_norm": 9.01414966583252, "learning_rate": 1.6944610895964717e-05, "loss": 1.3785, "step": 5181 }, { "epoch": 0.83, "grad_norm": 7.975002765655518, "learning_rate": 1.6943360253921462e-05, "loss": 1.2691, "step": 5182 }, { "epoch": 0.83, "grad_norm": 7.3314337730407715, "learning_rate": 1.694210940214617e-05, "loss": 1.3345, "step": 5183 }, { "epoch": 0.83, "grad_norm": 7.472005367279053, "learning_rate": 1.6940858340676622e-05, "loss": 1.4284, "step": 5184 }, { "epoch": 0.83, "grad_norm": 7.768496036529541, "learning_rate": 1.6939607069550606e-05, "loss": 1.3095, "step": 5185 }, { "epoch": 0.83, "grad_norm": 7.940646648406982, "learning_rate": 1.6938355588805918e-05, "loss": 1.3498, "step": 5186 }, { "epoch": 0.83, "grad_norm": 6.985260009765625, "learning_rate": 1.693710389848037e-05, "loss": 1.0549, "step": 5187 }, { "epoch": 0.84, "grad_norm": 8.790406227111816, "learning_rate": 1.6935851998611757e-05, "loss": 1.2928, "step": 5188 }, { "epoch": 0.84, "grad_norm": 7.301360607147217, "learning_rate": 1.6934599889237908e-05, "loss": 1.2304, "step": 5189 }, { "epoch": 0.84, "grad_norm": 8.02729606628418, "learning_rate": 1.6933347570396633e-05, "loss": 1.2543, "step": 5190 }, { "epoch": 0.84, "grad_norm": 8.018486976623535, "learning_rate": 1.6932095042125764e-05, "loss": 1.3275, "step": 5191 }, { "epoch": 0.84, "grad_norm": 9.145834922790527, "learning_rate": 1.6930842304463135e-05, "loss": 1.4733, "step": 5192 }, { "epoch": 0.84, "grad_norm": 8.349222183227539, "learning_rate": 1.6929589357446587e-05, "loss": 1.4933, "step": 5193 }, { "epoch": 0.84, "grad_norm": 9.102684020996094, "learning_rate": 1.6928336201113967e-05, "loss": 1.3765, "step": 5194 }, { "epoch": 0.84, "grad_norm": 10.134744644165039, "learning_rate": 1.6927082835503125e-05, "loss": 1.5785, "step": 5195 }, { "epoch": 0.84, "grad_norm": 9.750080108642578, "learning_rate": 1.6925829260651925e-05, "loss": 1.3271, "step": 5196 }, { "epoch": 0.84, "grad_norm": 8.207449913024902, "learning_rate": 1.6924575476598225e-05, "loss": 1.0427, "step": 5197 }, { "epoch": 0.84, "grad_norm": 8.236926078796387, "learning_rate": 1.6923321483379906e-05, "loss": 1.4928, "step": 5198 }, { "epoch": 0.84, "grad_norm": 10.767271995544434, "learning_rate": 1.692206728103484e-05, "loss": 1.409, "step": 5199 }, { "epoch": 0.84, "grad_norm": 9.043925285339355, "learning_rate": 1.692081286960092e-05, "loss": 1.367, "step": 5200 }, { "epoch": 0.84, "grad_norm": 8.13421630859375, "learning_rate": 1.6919558249116023e-05, "loss": 1.3218, "step": 5201 }, { "epoch": 0.84, "grad_norm": 8.445618629455566, "learning_rate": 1.6918303419618057e-05, "loss": 1.0867, "step": 5202 }, { "epoch": 0.84, "grad_norm": 9.122893333435059, "learning_rate": 1.6917048381144923e-05, "loss": 1.333, "step": 5203 }, { "epoch": 0.84, "grad_norm": 6.85024881362915, "learning_rate": 1.6915793133734528e-05, "loss": 1.4231, "step": 5204 }, { "epoch": 0.84, "grad_norm": 9.09119701385498, "learning_rate": 1.691453767742479e-05, "loss": 1.4592, "step": 5205 }, { "epoch": 0.84, "grad_norm": 8.152901649475098, "learning_rate": 1.6913282012253636e-05, "loss": 1.2118, "step": 5206 }, { "epoch": 0.84, "grad_norm": 7.929128646850586, "learning_rate": 1.6912026138258987e-05, "loss": 1.3415, "step": 5207 }, { "epoch": 0.84, "grad_norm": 8.680595397949219, "learning_rate": 1.6910770055478783e-05, "loss": 1.4841, "step": 5208 }, { "epoch": 0.84, "grad_norm": 10.522771835327148, "learning_rate": 1.6909513763950965e-05, "loss": 1.4611, "step": 5209 }, { "epoch": 0.84, "grad_norm": 8.436812400817871, "learning_rate": 1.690825726371348e-05, "loss": 1.4752, "step": 5210 }, { "epoch": 0.84, "grad_norm": 6.8688178062438965, "learning_rate": 1.690700055480428e-05, "loss": 1.3396, "step": 5211 }, { "epoch": 0.84, "grad_norm": 9.146331787109375, "learning_rate": 1.6905743637261332e-05, "loss": 1.3302, "step": 5212 }, { "epoch": 0.84, "grad_norm": 9.658053398132324, "learning_rate": 1.6904486511122593e-05, "loss": 1.2388, "step": 5213 }, { "epoch": 0.84, "grad_norm": 8.058130264282227, "learning_rate": 1.6903229176426043e-05, "loss": 1.4414, "step": 5214 }, { "epoch": 0.84, "grad_norm": 10.10176944732666, "learning_rate": 1.6901971633209657e-05, "loss": 1.5873, "step": 5215 }, { "epoch": 0.84, "grad_norm": 7.770334243774414, "learning_rate": 1.6900713881511425e-05, "loss": 1.2377, "step": 5216 }, { "epoch": 0.84, "grad_norm": 7.658485412597656, "learning_rate": 1.6899455921369338e-05, "loss": 1.2364, "step": 5217 }, { "epoch": 0.84, "grad_norm": 9.540116310119629, "learning_rate": 1.6898197752821385e-05, "loss": 1.0487, "step": 5218 }, { "epoch": 0.84, "grad_norm": 8.0682373046875, "learning_rate": 1.6896939375905583e-05, "loss": 1.0667, "step": 5219 }, { "epoch": 0.84, "grad_norm": 7.1557488441467285, "learning_rate": 1.689568079065994e-05, "loss": 1.253, "step": 5220 }, { "epoch": 0.84, "grad_norm": 8.100905418395996, "learning_rate": 1.6894421997122465e-05, "loss": 1.2038, "step": 5221 }, { "epoch": 0.84, "grad_norm": 8.171610832214355, "learning_rate": 1.689316299533119e-05, "loss": 1.0159, "step": 5222 }, { "epoch": 0.84, "grad_norm": 8.78913688659668, "learning_rate": 1.6891903785324146e-05, "loss": 1.4554, "step": 5223 }, { "epoch": 0.84, "grad_norm": 8.376022338867188, "learning_rate": 1.6890644367139354e-05, "loss": 1.4305, "step": 5224 }, { "epoch": 0.84, "grad_norm": 7.899483680725098, "learning_rate": 1.6889384740814874e-05, "loss": 1.2759, "step": 5225 }, { "epoch": 0.84, "grad_norm": 8.505977630615234, "learning_rate": 1.688812490638874e-05, "loss": 1.5619, "step": 5226 }, { "epoch": 0.84, "grad_norm": 8.539198875427246, "learning_rate": 1.688686486389902e-05, "loss": 1.4479, "step": 5227 }, { "epoch": 0.84, "grad_norm": 9.368253707885742, "learning_rate": 1.6885604613383765e-05, "loss": 1.4343, "step": 5228 }, { "epoch": 0.84, "grad_norm": 8.418259620666504, "learning_rate": 1.688434415488105e-05, "loss": 1.2217, "step": 5229 }, { "epoch": 0.84, "grad_norm": 9.309258460998535, "learning_rate": 1.688308348842894e-05, "loss": 1.4361, "step": 5230 }, { "epoch": 0.84, "grad_norm": 9.01664924621582, "learning_rate": 1.6881822614065522e-05, "loss": 1.2386, "step": 5231 }, { "epoch": 0.84, "grad_norm": 8.686124801635742, "learning_rate": 1.688056153182888e-05, "loss": 1.4756, "step": 5232 }, { "epoch": 0.84, "grad_norm": 7.421427249908447, "learning_rate": 1.68793002417571e-05, "loss": 1.453, "step": 5233 }, { "epoch": 0.84, "grad_norm": 9.306814193725586, "learning_rate": 1.6878038743888292e-05, "loss": 1.3068, "step": 5234 }, { "epoch": 0.84, "grad_norm": 10.647151947021484, "learning_rate": 1.6876777038260557e-05, "loss": 1.3346, "step": 5235 }, { "epoch": 0.84, "grad_norm": 8.424510955810547, "learning_rate": 1.6875515124912e-05, "loss": 1.3578, "step": 5236 }, { "epoch": 0.84, "grad_norm": 8.256460189819336, "learning_rate": 1.6874253003880743e-05, "loss": 1.3721, "step": 5237 }, { "epoch": 0.84, "grad_norm": 7.848912715911865, "learning_rate": 1.6872990675204913e-05, "loss": 1.2898, "step": 5238 }, { "epoch": 0.84, "grad_norm": 8.985787391662598, "learning_rate": 1.6871728138922636e-05, "loss": 1.2586, "step": 5239 }, { "epoch": 0.84, "grad_norm": 9.200331687927246, "learning_rate": 1.6870465395072048e-05, "loss": 1.1401, "step": 5240 }, { "epoch": 0.84, "grad_norm": 7.29705810546875, "learning_rate": 1.6869202443691297e-05, "loss": 1.4263, "step": 5241 }, { "epoch": 0.84, "grad_norm": 8.641103744506836, "learning_rate": 1.6867939284818527e-05, "loss": 1.4706, "step": 5242 }, { "epoch": 0.84, "grad_norm": 7.772019386291504, "learning_rate": 1.6866675918491893e-05, "loss": 1.2204, "step": 5243 }, { "epoch": 0.84, "grad_norm": 8.930492401123047, "learning_rate": 1.6865412344749557e-05, "loss": 1.5035, "step": 5244 }, { "epoch": 0.84, "grad_norm": 7.152806758880615, "learning_rate": 1.6864148563629687e-05, "loss": 1.2813, "step": 5245 }, { "epoch": 0.84, "grad_norm": 8.227720260620117, "learning_rate": 1.6862884575170456e-05, "loss": 1.1698, "step": 5246 }, { "epoch": 0.84, "grad_norm": 7.603642463684082, "learning_rate": 1.6861620379410048e-05, "loss": 1.1176, "step": 5247 }, { "epoch": 0.84, "grad_norm": 8.186322212219238, "learning_rate": 1.6860355976386644e-05, "loss": 1.4986, "step": 5248 }, { "epoch": 0.84, "grad_norm": 7.635524272918701, "learning_rate": 1.685909136613844e-05, "loss": 1.3328, "step": 5249 }, { "epoch": 0.85, "grad_norm": 7.463228702545166, "learning_rate": 1.6857826548703634e-05, "loss": 1.2943, "step": 5250 }, { "epoch": 0.85, "grad_norm": 9.461745262145996, "learning_rate": 1.6856561524120434e-05, "loss": 1.4664, "step": 5251 }, { "epoch": 0.85, "grad_norm": 8.441994667053223, "learning_rate": 1.6855296292427046e-05, "loss": 1.3539, "step": 5252 }, { "epoch": 0.85, "grad_norm": 8.307439804077148, "learning_rate": 1.6854030853661695e-05, "loss": 1.551, "step": 5253 }, { "epoch": 0.85, "grad_norm": 7.420289516448975, "learning_rate": 1.6852765207862597e-05, "loss": 1.3925, "step": 5254 }, { "epoch": 0.85, "grad_norm": 8.785223960876465, "learning_rate": 1.685149935506799e-05, "loss": 1.4953, "step": 5255 }, { "epoch": 0.85, "grad_norm": 8.667328834533691, "learning_rate": 1.6850233295316103e-05, "loss": 1.3497, "step": 5256 }, { "epoch": 0.85, "grad_norm": 7.300899505615234, "learning_rate": 1.684896702864518e-05, "loss": 1.1081, "step": 5257 }, { "epoch": 0.85, "grad_norm": 6.728004455566406, "learning_rate": 1.684770055509348e-05, "loss": 1.1665, "step": 5258 }, { "epoch": 0.85, "grad_norm": 7.396978855133057, "learning_rate": 1.6846433874699248e-05, "loss": 1.3205, "step": 5259 }, { "epoch": 0.85, "grad_norm": 7.936194896697998, "learning_rate": 1.6845166987500747e-05, "loss": 1.4931, "step": 5260 }, { "epoch": 0.85, "grad_norm": 8.484394073486328, "learning_rate": 1.6843899893536246e-05, "loss": 1.3996, "step": 5261 }, { "epoch": 0.85, "grad_norm": 8.212181091308594, "learning_rate": 1.684263259284402e-05, "loss": 1.1573, "step": 5262 }, { "epoch": 0.85, "grad_norm": 7.540182113647461, "learning_rate": 1.6841365085462347e-05, "loss": 1.1858, "step": 5263 }, { "epoch": 0.85, "grad_norm": 6.639489650726318, "learning_rate": 1.6840097371429514e-05, "loss": 1.3915, "step": 5264 }, { "epoch": 0.85, "grad_norm": 8.019022941589355, "learning_rate": 1.6838829450783818e-05, "loss": 1.4564, "step": 5265 }, { "epoch": 0.85, "grad_norm": 6.332847595214844, "learning_rate": 1.6837561323563553e-05, "loss": 1.2275, "step": 5266 }, { "epoch": 0.85, "grad_norm": 8.744157791137695, "learning_rate": 1.6836292989807023e-05, "loss": 1.455, "step": 5267 }, { "epoch": 0.85, "grad_norm": 9.587684631347656, "learning_rate": 1.6835024449552545e-05, "loss": 1.1943, "step": 5268 }, { "epoch": 0.85, "grad_norm": 7.156374931335449, "learning_rate": 1.6833755702838434e-05, "loss": 1.2758, "step": 5269 }, { "epoch": 0.85, "grad_norm": 9.497736930847168, "learning_rate": 1.683248674970301e-05, "loss": 1.3454, "step": 5270 }, { "epoch": 0.85, "grad_norm": 7.294111728668213, "learning_rate": 1.683121759018461e-05, "loss": 1.4346, "step": 5271 }, { "epoch": 0.85, "grad_norm": 8.031635284423828, "learning_rate": 1.682994822432157e-05, "loss": 1.3254, "step": 5272 }, { "epoch": 0.85, "grad_norm": 10.061723709106445, "learning_rate": 1.6828678652152225e-05, "loss": 1.44, "step": 5273 }, { "epoch": 0.85, "grad_norm": 10.011369705200195, "learning_rate": 1.6827408873714932e-05, "loss": 1.4212, "step": 5274 }, { "epoch": 0.85, "grad_norm": 7.925987720489502, "learning_rate": 1.682613888904804e-05, "loss": 1.2413, "step": 5275 }, { "epoch": 0.85, "grad_norm": 8.413728713989258, "learning_rate": 1.6824868698189917e-05, "loss": 1.3172, "step": 5276 }, { "epoch": 0.85, "grad_norm": 10.264875411987305, "learning_rate": 1.682359830117892e-05, "loss": 1.5007, "step": 5277 }, { "epoch": 0.85, "grad_norm": 7.893741607666016, "learning_rate": 1.6822327698053435e-05, "loss": 1.4296, "step": 5278 }, { "epoch": 0.85, "grad_norm": 8.060005187988281, "learning_rate": 1.6821056888851838e-05, "loss": 1.3481, "step": 5279 }, { "epoch": 0.85, "grad_norm": 7.413450717926025, "learning_rate": 1.681978587361251e-05, "loss": 1.022, "step": 5280 }, { "epoch": 0.85, "grad_norm": 7.426199436187744, "learning_rate": 1.6818514652373846e-05, "loss": 1.3796, "step": 5281 }, { "epoch": 0.85, "grad_norm": 8.000792503356934, "learning_rate": 1.681724322517425e-05, "loss": 1.3396, "step": 5282 }, { "epoch": 0.85, "grad_norm": 10.762808799743652, "learning_rate": 1.6815971592052124e-05, "loss": 1.3385, "step": 5283 }, { "epoch": 0.85, "grad_norm": 7.81898307800293, "learning_rate": 1.6814699753045874e-05, "loss": 1.3196, "step": 5284 }, { "epoch": 0.85, "grad_norm": 6.872551918029785, "learning_rate": 1.681342770819392e-05, "loss": 1.3048, "step": 5285 }, { "epoch": 0.85, "grad_norm": 7.185919284820557, "learning_rate": 1.681215545753469e-05, "loss": 1.2633, "step": 5286 }, { "epoch": 0.85, "grad_norm": 8.462794303894043, "learning_rate": 1.6810883001106607e-05, "loss": 1.396, "step": 5287 }, { "epoch": 0.85, "grad_norm": 8.708723068237305, "learning_rate": 1.6809610338948114e-05, "loss": 1.3737, "step": 5288 }, { "epoch": 0.85, "grad_norm": 9.132670402526855, "learning_rate": 1.6808337471097652e-05, "loss": 1.4696, "step": 5289 }, { "epoch": 0.85, "grad_norm": 7.931397438049316, "learning_rate": 1.6807064397593662e-05, "loss": 1.4932, "step": 5290 }, { "epoch": 0.85, "grad_norm": 8.093805313110352, "learning_rate": 1.6805791118474606e-05, "loss": 1.2881, "step": 5291 }, { "epoch": 0.85, "grad_norm": 9.185807228088379, "learning_rate": 1.6804517633778943e-05, "loss": 1.5163, "step": 5292 }, { "epoch": 0.85, "grad_norm": 7.256712436676025, "learning_rate": 1.6803243943545138e-05, "loss": 1.3451, "step": 5293 }, { "epoch": 0.85, "grad_norm": 8.992241859436035, "learning_rate": 1.680197004781167e-05, "loss": 1.4407, "step": 5294 }, { "epoch": 0.85, "grad_norm": 8.267910957336426, "learning_rate": 1.6800695946617008e-05, "loss": 1.5139, "step": 5295 }, { "epoch": 0.85, "grad_norm": 7.4719390869140625, "learning_rate": 1.6799421639999648e-05, "loss": 1.531, "step": 5296 }, { "epoch": 0.85, "grad_norm": 13.10441780090332, "learning_rate": 1.6798147127998076e-05, "loss": 1.487, "step": 5297 }, { "epoch": 0.85, "grad_norm": 7.692919731140137, "learning_rate": 1.6796872410650794e-05, "loss": 1.3931, "step": 5298 }, { "epoch": 0.85, "grad_norm": 8.135744094848633, "learning_rate": 1.6795597487996302e-05, "loss": 1.4059, "step": 5299 }, { "epoch": 0.85, "grad_norm": 8.041363716125488, "learning_rate": 1.6794322360073113e-05, "loss": 1.1497, "step": 5300 }, { "epoch": 0.85, "grad_norm": 8.362360954284668, "learning_rate": 1.6793047026919746e-05, "loss": 1.3028, "step": 5301 }, { "epoch": 0.85, "grad_norm": 7.99965238571167, "learning_rate": 1.679177148857472e-05, "loss": 1.1983, "step": 5302 }, { "epoch": 0.85, "grad_norm": 8.31478500366211, "learning_rate": 1.6790495745076568e-05, "loss": 1.2749, "step": 5303 }, { "epoch": 0.85, "grad_norm": 7.78708028793335, "learning_rate": 1.6789219796463817e-05, "loss": 1.0974, "step": 5304 }, { "epoch": 0.85, "grad_norm": 8.626265525817871, "learning_rate": 1.6787943642775018e-05, "loss": 1.3755, "step": 5305 }, { "epoch": 0.85, "grad_norm": 8.99659538269043, "learning_rate": 1.6786667284048714e-05, "loss": 1.3006, "step": 5306 }, { "epoch": 0.85, "grad_norm": 6.8231964111328125, "learning_rate": 1.678539072032346e-05, "loss": 1.1533, "step": 5307 }, { "epoch": 0.85, "grad_norm": 9.705451965332031, "learning_rate": 1.6784113951637813e-05, "loss": 1.7273, "step": 5308 }, { "epoch": 0.85, "grad_norm": 7.945984840393066, "learning_rate": 1.6782836978030346e-05, "loss": 1.145, "step": 5309 }, { "epoch": 0.85, "grad_norm": 8.716338157653809, "learning_rate": 1.6781559799539623e-05, "loss": 1.3104, "step": 5310 }, { "epoch": 0.85, "grad_norm": 9.569693565368652, "learning_rate": 1.678028241620423e-05, "loss": 1.5834, "step": 5311 }, { "epoch": 0.86, "grad_norm": 9.992087364196777, "learning_rate": 1.6779004828062747e-05, "loss": 1.5993, "step": 5312 }, { "epoch": 0.86, "grad_norm": 8.241342544555664, "learning_rate": 1.677772703515377e-05, "loss": 1.2879, "step": 5313 }, { "epoch": 0.86, "grad_norm": 7.327584743499756, "learning_rate": 1.6776449037515887e-05, "loss": 1.5907, "step": 5314 }, { "epoch": 0.86, "grad_norm": 9.555501937866211, "learning_rate": 1.6775170835187713e-05, "loss": 1.6091, "step": 5315 }, { "epoch": 0.86, "grad_norm": 8.561091423034668, "learning_rate": 1.677389242820785e-05, "loss": 1.3461, "step": 5316 }, { "epoch": 0.86, "grad_norm": 7.767305850982666, "learning_rate": 1.6772613816614915e-05, "loss": 1.4683, "step": 5317 }, { "epoch": 0.86, "grad_norm": 7.688977241516113, "learning_rate": 1.6771335000447527e-05, "loss": 1.4726, "step": 5318 }, { "epoch": 0.86, "grad_norm": 8.042080879211426, "learning_rate": 1.6770055979744325e-05, "loss": 1.3659, "step": 5319 }, { "epoch": 0.86, "grad_norm": 8.478531837463379, "learning_rate": 1.676877675454393e-05, "loss": 1.3803, "step": 5320 }, { "epoch": 0.86, "grad_norm": 6.600549221038818, "learning_rate": 1.6767497324884993e-05, "loss": 1.1235, "step": 5321 }, { "epoch": 0.86, "grad_norm": 8.363143920898438, "learning_rate": 1.6766217690806152e-05, "loss": 1.3983, "step": 5322 }, { "epoch": 0.86, "grad_norm": 10.98414134979248, "learning_rate": 1.6764937852346067e-05, "loss": 1.3451, "step": 5323 }, { "epoch": 0.86, "grad_norm": 7.958385944366455, "learning_rate": 1.6763657809543387e-05, "loss": 1.5367, "step": 5324 }, { "epoch": 0.86, "grad_norm": 9.114265441894531, "learning_rate": 1.676237756243679e-05, "loss": 1.3026, "step": 5325 }, { "epoch": 0.86, "grad_norm": 8.649622917175293, "learning_rate": 1.6761097111064942e-05, "loss": 1.3661, "step": 5326 }, { "epoch": 0.86, "grad_norm": 7.827535629272461, "learning_rate": 1.6759816455466515e-05, "loss": 1.2212, "step": 5327 }, { "epoch": 0.86, "grad_norm": 9.324599266052246, "learning_rate": 1.6758535595680197e-05, "loss": 1.3876, "step": 5328 }, { "epoch": 0.86, "grad_norm": 9.062389373779297, "learning_rate": 1.6757254531744683e-05, "loss": 1.5194, "step": 5329 }, { "epoch": 0.86, "grad_norm": 7.355959892272949, "learning_rate": 1.675597326369866e-05, "loss": 1.2785, "step": 5330 }, { "epoch": 0.86, "grad_norm": 7.106095790863037, "learning_rate": 1.6754691791580837e-05, "loss": 1.377, "step": 5331 }, { "epoch": 0.86, "grad_norm": 7.941001892089844, "learning_rate": 1.6753410115429915e-05, "loss": 1.2468, "step": 5332 }, { "epoch": 0.86, "grad_norm": 8.889610290527344, "learning_rate": 1.6752128235284616e-05, "loss": 1.3842, "step": 5333 }, { "epoch": 0.86, "grad_norm": 8.702665328979492, "learning_rate": 1.675084615118366e-05, "loss": 1.3885, "step": 5334 }, { "epoch": 0.86, "grad_norm": 7.991218090057373, "learning_rate": 1.674956386316577e-05, "loss": 1.1534, "step": 5335 }, { "epoch": 0.86, "grad_norm": 7.660171031951904, "learning_rate": 1.674828137126968e-05, "loss": 1.3564, "step": 5336 }, { "epoch": 0.86, "grad_norm": 6.816803455352783, "learning_rate": 1.6746998675534127e-05, "loss": 1.4803, "step": 5337 }, { "epoch": 0.86, "grad_norm": 7.903443813323975, "learning_rate": 1.6745715775997856e-05, "loss": 1.1583, "step": 5338 }, { "epoch": 0.86, "grad_norm": 7.409320831298828, "learning_rate": 1.6744432672699627e-05, "loss": 1.1581, "step": 5339 }, { "epoch": 0.86, "grad_norm": 8.357007026672363, "learning_rate": 1.6743149365678194e-05, "loss": 1.1533, "step": 5340 }, { "epoch": 0.86, "grad_norm": 8.345328330993652, "learning_rate": 1.6741865854972313e-05, "loss": 1.3183, "step": 5341 }, { "epoch": 0.86, "grad_norm": 7.543519020080566, "learning_rate": 1.6740582140620762e-05, "loss": 1.6364, "step": 5342 }, { "epoch": 0.86, "grad_norm": 8.23012638092041, "learning_rate": 1.673929822266231e-05, "loss": 1.4437, "step": 5343 }, { "epoch": 0.86, "grad_norm": 6.949639320373535, "learning_rate": 1.673801410113575e-05, "loss": 1.3106, "step": 5344 }, { "epoch": 0.86, "grad_norm": 8.700206756591797, "learning_rate": 1.673672977607986e-05, "loss": 1.2312, "step": 5345 }, { "epoch": 0.86, "grad_norm": 8.36051082611084, "learning_rate": 1.673544524753344e-05, "loss": 1.2592, "step": 5346 }, { "epoch": 0.86, "grad_norm": 8.168911933898926, "learning_rate": 1.673416051553529e-05, "loss": 1.2383, "step": 5347 }, { "epoch": 0.86, "grad_norm": 9.43035888671875, "learning_rate": 1.6732875580124214e-05, "loss": 1.1473, "step": 5348 }, { "epoch": 0.86, "grad_norm": 9.34891414642334, "learning_rate": 1.6731590441339026e-05, "loss": 1.1784, "step": 5349 }, { "epoch": 0.86, "grad_norm": 8.081225395202637, "learning_rate": 1.6730305099218548e-05, "loss": 1.1226, "step": 5350 }, { "epoch": 0.86, "grad_norm": 8.036750793457031, "learning_rate": 1.6729019553801604e-05, "loss": 1.3447, "step": 5351 }, { "epoch": 0.86, "grad_norm": 8.507996559143066, "learning_rate": 1.672773380512702e-05, "loss": 1.3686, "step": 5352 }, { "epoch": 0.86, "grad_norm": 8.797309875488281, "learning_rate": 1.6726447853233643e-05, "loss": 1.2587, "step": 5353 }, { "epoch": 0.86, "grad_norm": 8.15790843963623, "learning_rate": 1.6725161698160307e-05, "loss": 1.4442, "step": 5354 }, { "epoch": 0.86, "grad_norm": 8.614487648010254, "learning_rate": 1.672387533994587e-05, "loss": 1.2523, "step": 5355 }, { "epoch": 0.86, "grad_norm": 9.47329044342041, "learning_rate": 1.6722588778629184e-05, "loss": 1.3762, "step": 5356 }, { "epoch": 0.86, "grad_norm": 9.650856971740723, "learning_rate": 1.6721302014249108e-05, "loss": 1.5592, "step": 5357 }, { "epoch": 0.86, "grad_norm": 7.203831672668457, "learning_rate": 1.6720015046844517e-05, "loss": 1.2703, "step": 5358 }, { "epoch": 0.86, "grad_norm": 7.809701442718506, "learning_rate": 1.6718727876454278e-05, "loss": 1.4319, "step": 5359 }, { "epoch": 0.86, "grad_norm": 7.4983110427856445, "learning_rate": 1.6717440503117275e-05, "loss": 1.2193, "step": 5360 }, { "epoch": 0.86, "grad_norm": 7.834787368774414, "learning_rate": 1.67161529268724e-05, "loss": 1.4073, "step": 5361 }, { "epoch": 0.86, "grad_norm": 11.72396469116211, "learning_rate": 1.6714865147758537e-05, "loss": 1.3476, "step": 5362 }, { "epoch": 0.86, "grad_norm": 7.383223056793213, "learning_rate": 1.6713577165814586e-05, "loss": 1.3874, "step": 5363 }, { "epoch": 0.86, "grad_norm": 7.386955738067627, "learning_rate": 1.671228898107946e-05, "loss": 1.4103, "step": 5364 }, { "epoch": 0.86, "grad_norm": 7.522303104400635, "learning_rate": 1.6711000593592063e-05, "loss": 1.4294, "step": 5365 }, { "epoch": 0.86, "grad_norm": 8.206954956054688, "learning_rate": 1.670971200339131e-05, "loss": 1.3865, "step": 5366 }, { "epoch": 0.86, "grad_norm": 8.99955940246582, "learning_rate": 1.6708423210516128e-05, "loss": 1.3083, "step": 5367 }, { "epoch": 0.86, "grad_norm": 8.847046852111816, "learning_rate": 1.6707134215005444e-05, "loss": 1.3109, "step": 5368 }, { "epoch": 0.86, "grad_norm": 8.146958351135254, "learning_rate": 1.67058450168982e-05, "loss": 1.2221, "step": 5369 }, { "epoch": 0.86, "grad_norm": 10.006171226501465, "learning_rate": 1.6704555616233335e-05, "loss": 1.5928, "step": 5370 }, { "epoch": 0.86, "grad_norm": 7.313013076782227, "learning_rate": 1.6703266013049793e-05, "loss": 1.4089, "step": 5371 }, { "epoch": 0.86, "grad_norm": 8.292561531066895, "learning_rate": 1.670197620738653e-05, "loss": 1.4915, "step": 5372 }, { "epoch": 0.86, "grad_norm": 7.7174482345581055, "learning_rate": 1.6700686199282508e-05, "loss": 1.1695, "step": 5373 }, { "epoch": 0.87, "grad_norm": 7.43696928024292, "learning_rate": 1.669939598877669e-05, "loss": 1.2857, "step": 5374 }, { "epoch": 0.87, "grad_norm": 10.063515663146973, "learning_rate": 1.6698105575908047e-05, "loss": 1.3234, "step": 5375 }, { "epoch": 0.87, "grad_norm": 7.69944953918457, "learning_rate": 1.6696814960715564e-05, "loss": 1.4127, "step": 5376 }, { "epoch": 0.87, "grad_norm": 7.8174943923950195, "learning_rate": 1.669552414323822e-05, "loss": 1.3124, "step": 5377 }, { "epoch": 0.87, "grad_norm": 9.719568252563477, "learning_rate": 1.6694233123515006e-05, "loss": 1.5271, "step": 5378 }, { "epoch": 0.87, "grad_norm": 7.7722554206848145, "learning_rate": 1.669294190158492e-05, "loss": 1.1988, "step": 5379 }, { "epoch": 0.87, "grad_norm": 7.00101375579834, "learning_rate": 1.6691650477486968e-05, "loss": 1.1832, "step": 5380 }, { "epoch": 0.87, "grad_norm": 7.440938949584961, "learning_rate": 1.6690358851260154e-05, "loss": 1.2794, "step": 5381 }, { "epoch": 0.87, "grad_norm": 7.297031879425049, "learning_rate": 1.6689067022943492e-05, "loss": 1.2453, "step": 5382 }, { "epoch": 0.87, "grad_norm": 7.763930797576904, "learning_rate": 1.6687774992576008e-05, "loss": 1.5009, "step": 5383 }, { "epoch": 0.87, "grad_norm": 8.205896377563477, "learning_rate": 1.6686482760196726e-05, "loss": 1.2249, "step": 5384 }, { "epoch": 0.87, "grad_norm": 8.079437255859375, "learning_rate": 1.668519032584468e-05, "loss": 1.4855, "step": 5385 }, { "epoch": 0.87, "grad_norm": 6.899654865264893, "learning_rate": 1.6683897689558914e-05, "loss": 1.1362, "step": 5386 }, { "epoch": 0.87, "grad_norm": 6.678747653961182, "learning_rate": 1.6682604851378466e-05, "loss": 1.3767, "step": 5387 }, { "epoch": 0.87, "grad_norm": 7.793663501739502, "learning_rate": 1.668131181134239e-05, "loss": 1.3059, "step": 5388 }, { "epoch": 0.87, "grad_norm": 7.8888044357299805, "learning_rate": 1.6680018569489743e-05, "loss": 1.3616, "step": 5389 }, { "epoch": 0.87, "grad_norm": 9.831718444824219, "learning_rate": 1.6678725125859594e-05, "loss": 1.4692, "step": 5390 }, { "epoch": 0.87, "grad_norm": 8.133204460144043, "learning_rate": 1.667743148049101e-05, "loss": 1.349, "step": 5391 }, { "epoch": 0.87, "grad_norm": 7.408848762512207, "learning_rate": 1.667613763342306e-05, "loss": 1.6065, "step": 5392 }, { "epoch": 0.87, "grad_norm": 8.541091918945312, "learning_rate": 1.667484358469484e-05, "loss": 1.351, "step": 5393 }, { "epoch": 0.87, "grad_norm": 8.48468017578125, "learning_rate": 1.667354933434543e-05, "loss": 1.3499, "step": 5394 }, { "epoch": 0.87, "grad_norm": 7.659640789031982, "learning_rate": 1.667225488241392e-05, "loss": 1.317, "step": 5395 }, { "epoch": 0.87, "grad_norm": 8.437503814697266, "learning_rate": 1.667096022893942e-05, "loss": 1.319, "step": 5396 }, { "epoch": 0.87, "grad_norm": 6.921021461486816, "learning_rate": 1.6669665373961033e-05, "loss": 1.4165, "step": 5397 }, { "epoch": 0.87, "grad_norm": 7.3032331466674805, "learning_rate": 1.666837031751787e-05, "loss": 1.4102, "step": 5398 }, { "epoch": 0.87, "grad_norm": 9.187978744506836, "learning_rate": 1.6667075059649046e-05, "loss": 1.2243, "step": 5399 }, { "epoch": 0.87, "grad_norm": 10.971760749816895, "learning_rate": 1.666577960039369e-05, "loss": 1.5661, "step": 5400 }, { "epoch": 0.87, "grad_norm": 9.705690383911133, "learning_rate": 1.6664483939790935e-05, "loss": 1.2395, "step": 5401 }, { "epoch": 0.87, "grad_norm": 8.789350509643555, "learning_rate": 1.6663188077879916e-05, "loss": 1.2767, "step": 5402 }, { "epoch": 0.87, "grad_norm": 9.101371765136719, "learning_rate": 1.6661892014699774e-05, "loss": 1.3308, "step": 5403 }, { "epoch": 0.87, "grad_norm": 7.917412757873535, "learning_rate": 1.666059575028966e-05, "loss": 1.5032, "step": 5404 }, { "epoch": 0.87, "grad_norm": 7.206613540649414, "learning_rate": 1.6659299284688728e-05, "loss": 1.3465, "step": 5405 }, { "epoch": 0.87, "grad_norm": 8.316781997680664, "learning_rate": 1.665800261793614e-05, "loss": 1.2946, "step": 5406 }, { "epoch": 0.87, "grad_norm": 8.876728057861328, "learning_rate": 1.6656705750071066e-05, "loss": 1.3996, "step": 5407 }, { "epoch": 0.87, "grad_norm": 8.11590576171875, "learning_rate": 1.6655408681132673e-05, "loss": 1.3194, "step": 5408 }, { "epoch": 0.87, "grad_norm": 9.116893768310547, "learning_rate": 1.6654111411160144e-05, "loss": 1.7148, "step": 5409 }, { "epoch": 0.87, "grad_norm": 8.262332916259766, "learning_rate": 1.6652813940192666e-05, "loss": 1.3359, "step": 5410 }, { "epoch": 0.87, "grad_norm": 9.321624755859375, "learning_rate": 1.665151626826943e-05, "loss": 1.2716, "step": 5411 }, { "epoch": 0.87, "grad_norm": 8.26662540435791, "learning_rate": 1.665021839542963e-05, "loss": 1.261, "step": 5412 }, { "epoch": 0.87, "grad_norm": 8.188505172729492, "learning_rate": 1.6648920321712475e-05, "loss": 1.274, "step": 5413 }, { "epoch": 0.87, "grad_norm": 7.700925350189209, "learning_rate": 1.664762204715717e-05, "loss": 1.4708, "step": 5414 }, { "epoch": 0.87, "grad_norm": 7.352231502532959, "learning_rate": 1.6646323571802937e-05, "loss": 1.3939, "step": 5415 }, { "epoch": 0.87, "grad_norm": 8.18486499786377, "learning_rate": 1.664502489568899e-05, "loss": 1.1846, "step": 5416 }, { "epoch": 0.87, "grad_norm": 8.428362846374512, "learning_rate": 1.6643726018854563e-05, "loss": 1.3525, "step": 5417 }, { "epoch": 0.87, "grad_norm": 9.269770622253418, "learning_rate": 1.6642426941338885e-05, "loss": 1.2762, "step": 5418 }, { "epoch": 0.87, "grad_norm": 7.390449047088623, "learning_rate": 1.66411276631812e-05, "loss": 1.4175, "step": 5419 }, { "epoch": 0.87, "grad_norm": 9.091629028320312, "learning_rate": 1.6639828184420753e-05, "loss": 1.3212, "step": 5420 }, { "epoch": 0.87, "grad_norm": 7.665863990783691, "learning_rate": 1.6638528505096802e-05, "loss": 1.2035, "step": 5421 }, { "epoch": 0.87, "grad_norm": 9.291253089904785, "learning_rate": 1.6637228625248595e-05, "loss": 1.0714, "step": 5422 }, { "epoch": 0.87, "grad_norm": 9.980329513549805, "learning_rate": 1.66359285449154e-05, "loss": 1.4653, "step": 5423 }, { "epoch": 0.87, "grad_norm": 8.656777381896973, "learning_rate": 1.6634628264136493e-05, "loss": 1.1239, "step": 5424 }, { "epoch": 0.87, "grad_norm": 7.133738994598389, "learning_rate": 1.6633327782951143e-05, "loss": 1.311, "step": 5425 }, { "epoch": 0.87, "grad_norm": 8.42536735534668, "learning_rate": 1.6632027101398632e-05, "loss": 1.2544, "step": 5426 }, { "epoch": 0.87, "grad_norm": 9.032977104187012, "learning_rate": 1.663072621951826e-05, "loss": 1.2623, "step": 5427 }, { "epoch": 0.87, "grad_norm": 8.213433265686035, "learning_rate": 1.6629425137349305e-05, "loss": 1.4197, "step": 5428 }, { "epoch": 0.87, "grad_norm": 8.68863296508789, "learning_rate": 1.662812385493108e-05, "loss": 1.2398, "step": 5429 }, { "epoch": 0.87, "grad_norm": 8.754018783569336, "learning_rate": 1.662682237230289e-05, "loss": 1.5516, "step": 5430 }, { "epoch": 0.87, "grad_norm": 9.992372512817383, "learning_rate": 1.6625520689504043e-05, "loss": 1.3944, "step": 5431 }, { "epoch": 0.87, "grad_norm": 7.190135478973389, "learning_rate": 1.662421880657386e-05, "loss": 1.1063, "step": 5432 }, { "epoch": 0.87, "grad_norm": 8.52702808380127, "learning_rate": 1.662291672355167e-05, "loss": 1.2742, "step": 5433 }, { "epoch": 0.87, "grad_norm": 9.103296279907227, "learning_rate": 1.66216144404768e-05, "loss": 1.5427, "step": 5434 }, { "epoch": 0.87, "grad_norm": 8.059375762939453, "learning_rate": 1.6620311957388585e-05, "loss": 1.3773, "step": 5435 }, { "epoch": 0.88, "grad_norm": 7.806466579437256, "learning_rate": 1.6619009274326368e-05, "loss": 1.2889, "step": 5436 }, { "epoch": 0.88, "grad_norm": 7.777608871459961, "learning_rate": 1.6617706391329504e-05, "loss": 1.4639, "step": 5437 }, { "epoch": 0.88, "grad_norm": 8.510432243347168, "learning_rate": 1.661640330843734e-05, "loss": 1.4856, "step": 5438 }, { "epoch": 0.88, "grad_norm": 7.38323974609375, "learning_rate": 1.6615100025689245e-05, "loss": 1.3576, "step": 5439 }, { "epoch": 0.88, "grad_norm": 8.13664722442627, "learning_rate": 1.6613796543124582e-05, "loss": 1.4783, "step": 5440 }, { "epoch": 0.88, "grad_norm": 7.532087326049805, "learning_rate": 1.6612492860782727e-05, "loss": 1.0716, "step": 5441 }, { "epoch": 0.88, "grad_norm": 8.343868255615234, "learning_rate": 1.661118897870305e-05, "loss": 1.4469, "step": 5442 }, { "epoch": 0.88, "grad_norm": 7.4716057777404785, "learning_rate": 1.6609884896924948e-05, "loss": 1.1548, "step": 5443 }, { "epoch": 0.88, "grad_norm": 8.022143363952637, "learning_rate": 1.6608580615487805e-05, "loss": 1.1231, "step": 5444 }, { "epoch": 0.88, "grad_norm": 8.011892318725586, "learning_rate": 1.6607276134431024e-05, "loss": 1.3028, "step": 5445 }, { "epoch": 0.88, "grad_norm": 7.321710586547852, "learning_rate": 1.6605971453794002e-05, "loss": 1.5014, "step": 5446 }, { "epoch": 0.88, "grad_norm": 8.313247680664062, "learning_rate": 1.660466657361615e-05, "loss": 1.2595, "step": 5447 }, { "epoch": 0.88, "grad_norm": 9.585269927978516, "learning_rate": 1.6603361493936886e-05, "loss": 1.4697, "step": 5448 }, { "epoch": 0.88, "grad_norm": 9.065361022949219, "learning_rate": 1.6602056214795634e-05, "loss": 1.2797, "step": 5449 }, { "epoch": 0.88, "grad_norm": 8.426358222961426, "learning_rate": 1.660075073623181e-05, "loss": 1.159, "step": 5450 }, { "epoch": 0.88, "grad_norm": 7.7412824630737305, "learning_rate": 1.659944505828486e-05, "loss": 1.2486, "step": 5451 }, { "epoch": 0.88, "grad_norm": 9.30050277709961, "learning_rate": 1.659813918099422e-05, "loss": 1.4069, "step": 5452 }, { "epoch": 0.88, "grad_norm": 8.699951171875, "learning_rate": 1.6596833104399332e-05, "loss": 1.229, "step": 5453 }, { "epoch": 0.88, "grad_norm": 7.744914531707764, "learning_rate": 1.6595526828539648e-05, "loss": 1.0182, "step": 5454 }, { "epoch": 0.88, "grad_norm": 8.277987480163574, "learning_rate": 1.6594220353454627e-05, "loss": 1.4763, "step": 5455 }, { "epoch": 0.88, "grad_norm": 9.755577087402344, "learning_rate": 1.6592913679183733e-05, "loss": 1.3392, "step": 5456 }, { "epoch": 0.88, "grad_norm": 8.728372573852539, "learning_rate": 1.6591606805766436e-05, "loss": 1.3871, "step": 5457 }, { "epoch": 0.88, "grad_norm": 10.644112586975098, "learning_rate": 1.659029973324221e-05, "loss": 1.381, "step": 5458 }, { "epoch": 0.88, "grad_norm": 7.495260238647461, "learning_rate": 1.6588992461650535e-05, "loss": 1.2621, "step": 5459 }, { "epoch": 0.88, "grad_norm": 7.706284523010254, "learning_rate": 1.6587684991030903e-05, "loss": 1.2387, "step": 5460 }, { "epoch": 0.88, "grad_norm": 8.63148021697998, "learning_rate": 1.6586377321422804e-05, "loss": 1.2988, "step": 5461 }, { "epoch": 0.88, "grad_norm": 7.509593963623047, "learning_rate": 1.6585069452865742e-05, "loss": 1.382, "step": 5462 }, { "epoch": 0.88, "grad_norm": 7.959890842437744, "learning_rate": 1.658376138539922e-05, "loss": 1.0886, "step": 5463 }, { "epoch": 0.88, "grad_norm": 8.183489799499512, "learning_rate": 1.6582453119062746e-05, "loss": 1.5022, "step": 5464 }, { "epoch": 0.88, "grad_norm": 7.333704471588135, "learning_rate": 1.6581144653895845e-05, "loss": 1.1452, "step": 5465 }, { "epoch": 0.88, "grad_norm": 9.76994514465332, "learning_rate": 1.6579835989938033e-05, "loss": 1.4328, "step": 5466 }, { "epoch": 0.88, "grad_norm": 6.875327110290527, "learning_rate": 1.6578527127228846e-05, "loss": 1.2298, "step": 5467 }, { "epoch": 0.88, "grad_norm": 9.370908737182617, "learning_rate": 1.657721806580782e-05, "loss": 1.3121, "step": 5468 }, { "epoch": 0.88, "grad_norm": 8.388289451599121, "learning_rate": 1.6575908805714487e-05, "loss": 1.2192, "step": 5469 }, { "epoch": 0.88, "grad_norm": 8.569122314453125, "learning_rate": 1.6574599346988405e-05, "loss": 1.7111, "step": 5470 }, { "epoch": 0.88, "grad_norm": 6.693025588989258, "learning_rate": 1.6573289689669123e-05, "loss": 1.2118, "step": 5471 }, { "epoch": 0.88, "grad_norm": 8.327848434448242, "learning_rate": 1.6571979833796203e-05, "loss": 1.0354, "step": 5472 }, { "epoch": 0.88, "grad_norm": 7.431266784667969, "learning_rate": 1.657066977940921e-05, "loss": 1.3683, "step": 5473 }, { "epoch": 0.88, "grad_norm": 7.9666852951049805, "learning_rate": 1.6569359526547714e-05, "loss": 1.2385, "step": 5474 }, { "epoch": 0.88, "grad_norm": 8.603178024291992, "learning_rate": 1.6568049075251297e-05, "loss": 1.2321, "step": 5475 }, { "epoch": 0.88, "grad_norm": 6.873382091522217, "learning_rate": 1.6566738425559536e-05, "loss": 1.0045, "step": 5476 }, { "epoch": 0.88, "grad_norm": 8.014849662780762, "learning_rate": 1.6565427577512023e-05, "loss": 1.279, "step": 5477 }, { "epoch": 0.88, "grad_norm": 8.629374504089355, "learning_rate": 1.656411653114836e-05, "loss": 1.4215, "step": 5478 }, { "epoch": 0.88, "grad_norm": 8.381563186645508, "learning_rate": 1.6562805286508136e-05, "loss": 1.3298, "step": 5479 }, { "epoch": 0.88, "grad_norm": 8.105045318603516, "learning_rate": 1.6561493843630973e-05, "loss": 1.2313, "step": 5480 }, { "epoch": 0.88, "grad_norm": 8.323155403137207, "learning_rate": 1.6560182202556473e-05, "loss": 1.2683, "step": 5481 }, { "epoch": 0.88, "grad_norm": 8.928693771362305, "learning_rate": 1.6558870363324266e-05, "loss": 1.2095, "step": 5482 }, { "epoch": 0.88, "grad_norm": 9.30085563659668, "learning_rate": 1.6557558325973966e-05, "loss": 1.3061, "step": 5483 }, { "epoch": 0.88, "grad_norm": 8.507628440856934, "learning_rate": 1.655624609054521e-05, "loss": 1.4322, "step": 5484 }, { "epoch": 0.88, "grad_norm": 8.45837688446045, "learning_rate": 1.655493365707764e-05, "loss": 1.7007, "step": 5485 }, { "epoch": 0.88, "grad_norm": 8.94965648651123, "learning_rate": 1.6553621025610896e-05, "loss": 1.374, "step": 5486 }, { "epoch": 0.88, "grad_norm": 8.263503074645996, "learning_rate": 1.6552308196184624e-05, "loss": 1.3995, "step": 5487 }, { "epoch": 0.88, "grad_norm": 9.26261043548584, "learning_rate": 1.6550995168838485e-05, "loss": 1.3083, "step": 5488 }, { "epoch": 0.88, "grad_norm": 8.487092018127441, "learning_rate": 1.654968194361214e-05, "loss": 1.1707, "step": 5489 }, { "epoch": 0.88, "grad_norm": 8.660940170288086, "learning_rate": 1.6548368520545248e-05, "loss": 1.4186, "step": 5490 }, { "epoch": 0.88, "grad_norm": 7.839488506317139, "learning_rate": 1.6547054899677492e-05, "loss": 1.4456, "step": 5491 }, { "epoch": 0.88, "grad_norm": 9.907487869262695, "learning_rate": 1.654574108104855e-05, "loss": 1.0498, "step": 5492 }, { "epoch": 0.88, "grad_norm": 8.506675720214844, "learning_rate": 1.6544427064698102e-05, "loss": 0.956, "step": 5493 }, { "epoch": 0.88, "grad_norm": 8.100667953491211, "learning_rate": 1.6543112850665848e-05, "loss": 1.5329, "step": 5494 }, { "epoch": 0.88, "grad_norm": 8.319348335266113, "learning_rate": 1.6541798438991477e-05, "loss": 1.4032, "step": 5495 }, { "epoch": 0.88, "grad_norm": 9.597932815551758, "learning_rate": 1.6540483829714697e-05, "loss": 1.334, "step": 5496 }, { "epoch": 0.88, "grad_norm": 7.8210368156433105, "learning_rate": 1.6539169022875213e-05, "loss": 1.2774, "step": 5497 }, { "epoch": 0.89, "grad_norm": 7.793323993682861, "learning_rate": 1.6537854018512746e-05, "loss": 1.4051, "step": 5498 }, { "epoch": 0.89, "grad_norm": 8.630598068237305, "learning_rate": 1.6536538816667013e-05, "loss": 1.5098, "step": 5499 }, { "epoch": 0.89, "grad_norm": 8.864593505859375, "learning_rate": 1.6535223417377745e-05, "loss": 1.4881, "step": 5500 }, { "epoch": 0.89, "grad_norm": 9.302775382995605, "learning_rate": 1.653390782068467e-05, "loss": 1.3674, "step": 5501 }, { "epoch": 0.89, "grad_norm": 9.972875595092773, "learning_rate": 1.653259202662753e-05, "loss": 1.2621, "step": 5502 }, { "epoch": 0.89, "grad_norm": 7.706853866577148, "learning_rate": 1.6531276035246074e-05, "loss": 1.472, "step": 5503 }, { "epoch": 0.89, "grad_norm": 8.419069290161133, "learning_rate": 1.6529959846580043e-05, "loss": 1.0708, "step": 5504 }, { "epoch": 0.89, "grad_norm": 7.689968109130859, "learning_rate": 1.6528643460669203e-05, "loss": 1.2894, "step": 5505 }, { "epoch": 0.89, "grad_norm": 9.007800102233887, "learning_rate": 1.6527326877553307e-05, "loss": 1.3209, "step": 5506 }, { "epoch": 0.89, "grad_norm": 7.574196815490723, "learning_rate": 1.652601009727214e-05, "loss": 1.629, "step": 5507 }, { "epoch": 0.89, "grad_norm": 9.4451322555542, "learning_rate": 1.6524693119865464e-05, "loss": 1.402, "step": 5508 }, { "epoch": 0.89, "grad_norm": 7.787986755371094, "learning_rate": 1.6523375945373062e-05, "loss": 1.229, "step": 5509 }, { "epoch": 0.89, "grad_norm": 8.803349494934082, "learning_rate": 1.652205857383472e-05, "loss": 0.9803, "step": 5510 }, { "epoch": 0.89, "grad_norm": 10.153093338012695, "learning_rate": 1.6520741005290234e-05, "loss": 1.3853, "step": 5511 }, { "epoch": 0.89, "grad_norm": 8.744359016418457, "learning_rate": 1.65194232397794e-05, "loss": 1.2965, "step": 5512 }, { "epoch": 0.89, "grad_norm": 9.121849060058594, "learning_rate": 1.651810527734203e-05, "loss": 1.4573, "step": 5513 }, { "epoch": 0.89, "grad_norm": 8.79183578491211, "learning_rate": 1.6516787118017922e-05, "loss": 1.4239, "step": 5514 }, { "epoch": 0.89, "grad_norm": 10.069229125976562, "learning_rate": 1.6515468761846902e-05, "loss": 1.5534, "step": 5515 }, { "epoch": 0.89, "grad_norm": 9.978963851928711, "learning_rate": 1.6514150208868786e-05, "loss": 1.2794, "step": 5516 }, { "epoch": 0.89, "grad_norm": 8.920015335083008, "learning_rate": 1.651283145912341e-05, "loss": 1.2361, "step": 5517 }, { "epoch": 0.89, "grad_norm": 8.050314903259277, "learning_rate": 1.65115125126506e-05, "loss": 1.4156, "step": 5518 }, { "epoch": 0.89, "grad_norm": 7.407353401184082, "learning_rate": 1.6510193369490202e-05, "loss": 1.3003, "step": 5519 }, { "epoch": 0.89, "grad_norm": 8.65638542175293, "learning_rate": 1.650887402968206e-05, "loss": 1.4582, "step": 5520 }, { "epoch": 0.89, "grad_norm": 7.783069133758545, "learning_rate": 1.650755449326603e-05, "loss": 1.3175, "step": 5521 }, { "epoch": 0.89, "grad_norm": 9.35927963256836, "learning_rate": 1.650623476028196e-05, "loss": 1.4492, "step": 5522 }, { "epoch": 0.89, "grad_norm": 8.587445259094238, "learning_rate": 1.650491483076973e-05, "loss": 1.261, "step": 5523 }, { "epoch": 0.89, "grad_norm": 8.786480903625488, "learning_rate": 1.6503594704769192e-05, "loss": 1.1961, "step": 5524 }, { "epoch": 0.89, "grad_norm": 8.394922256469727, "learning_rate": 1.6502274382320235e-05, "loss": 1.4086, "step": 5525 }, { "epoch": 0.89, "grad_norm": 7.689829349517822, "learning_rate": 1.6500953863462735e-05, "loss": 1.3111, "step": 5526 }, { "epoch": 0.89, "grad_norm": 8.673139572143555, "learning_rate": 1.649963314823658e-05, "loss": 1.2682, "step": 5527 }, { "epoch": 0.89, "grad_norm": 8.136582374572754, "learning_rate": 1.6498312236681667e-05, "loss": 1.3747, "step": 5528 }, { "epoch": 0.89, "grad_norm": 7.499693870544434, "learning_rate": 1.6496991128837892e-05, "loss": 1.1174, "step": 5529 }, { "epoch": 0.89, "grad_norm": 9.87630558013916, "learning_rate": 1.6495669824745163e-05, "loss": 1.4479, "step": 5530 }, { "epoch": 0.89, "grad_norm": 9.06910514831543, "learning_rate": 1.649434832444339e-05, "loss": 1.4956, "step": 5531 }, { "epoch": 0.89, "grad_norm": 8.974688529968262, "learning_rate": 1.6493026627972494e-05, "loss": 1.1472, "step": 5532 }, { "epoch": 0.89, "grad_norm": 7.3878254890441895, "learning_rate": 1.6491704735372393e-05, "loss": 1.3248, "step": 5533 }, { "epoch": 0.89, "grad_norm": 7.982567310333252, "learning_rate": 1.6490382646683016e-05, "loss": 1.2904, "step": 5534 }, { "epoch": 0.89, "grad_norm": 7.490289211273193, "learning_rate": 1.6489060361944305e-05, "loss": 1.2547, "step": 5535 }, { "epoch": 0.89, "grad_norm": 9.96505355834961, "learning_rate": 1.6487737881196194e-05, "loss": 1.56, "step": 5536 }, { "epoch": 0.89, "grad_norm": 7.0218706130981445, "learning_rate": 1.6486415204478632e-05, "loss": 0.9319, "step": 5537 }, { "epoch": 0.89, "grad_norm": 7.999757766723633, "learning_rate": 1.6485092331831575e-05, "loss": 1.2447, "step": 5538 }, { "epoch": 0.89, "grad_norm": 9.122136116027832, "learning_rate": 1.648376926329498e-05, "loss": 1.3811, "step": 5539 }, { "epoch": 0.89, "grad_norm": 6.938902378082275, "learning_rate": 1.648244599890881e-05, "loss": 0.9999, "step": 5540 }, { "epoch": 0.89, "grad_norm": 8.6776704788208, "learning_rate": 1.6481122538713033e-05, "loss": 1.3237, "step": 5541 }, { "epoch": 0.89, "grad_norm": 7.966042995452881, "learning_rate": 1.6479798882747636e-05, "loss": 1.2599, "step": 5542 }, { "epoch": 0.89, "grad_norm": 10.105895042419434, "learning_rate": 1.6478475031052594e-05, "loss": 1.7039, "step": 5543 }, { "epoch": 0.89, "grad_norm": 7.17905855178833, "learning_rate": 1.64771509836679e-05, "loss": 1.1385, "step": 5544 }, { "epoch": 0.89, "grad_norm": 7.573371887207031, "learning_rate": 1.647582674063354e-05, "loss": 1.4393, "step": 5545 }, { "epoch": 0.89, "grad_norm": 8.377226829528809, "learning_rate": 1.647450230198952e-05, "loss": 1.555, "step": 5546 }, { "epoch": 0.89, "grad_norm": 7.252344608306885, "learning_rate": 1.6473177667775846e-05, "loss": 1.2442, "step": 5547 }, { "epoch": 0.89, "grad_norm": 7.996689796447754, "learning_rate": 1.647185283803253e-05, "loss": 1.3955, "step": 5548 }, { "epoch": 0.89, "grad_norm": 8.34496021270752, "learning_rate": 1.647052781279959e-05, "loss": 1.3276, "step": 5549 }, { "epoch": 0.89, "grad_norm": 7.971853733062744, "learning_rate": 1.6469202592117047e-05, "loss": 1.1709, "step": 5550 }, { "epoch": 0.89, "grad_norm": 9.290556907653809, "learning_rate": 1.6467877176024935e-05, "loss": 1.4204, "step": 5551 }, { "epoch": 0.89, "grad_norm": 7.649341106414795, "learning_rate": 1.6466551564563287e-05, "loss": 1.2731, "step": 5552 }, { "epoch": 0.89, "grad_norm": 8.504068374633789, "learning_rate": 1.6465225757772146e-05, "loss": 1.3222, "step": 5553 }, { "epoch": 0.89, "grad_norm": 6.323020935058594, "learning_rate": 1.6463899755691558e-05, "loss": 1.2379, "step": 5554 }, { "epoch": 0.89, "grad_norm": 8.956599235534668, "learning_rate": 1.6462573558361578e-05, "loss": 1.3698, "step": 5555 }, { "epoch": 0.89, "grad_norm": 7.355197906494141, "learning_rate": 1.6461247165822267e-05, "loss": 1.0876, "step": 5556 }, { "epoch": 0.89, "grad_norm": 8.6325101852417, "learning_rate": 1.6459920578113684e-05, "loss": 1.3278, "step": 5557 }, { "epoch": 0.89, "grad_norm": 9.025325775146484, "learning_rate": 1.6458593795275906e-05, "loss": 1.3738, "step": 5558 }, { "epoch": 0.89, "grad_norm": 9.332959175109863, "learning_rate": 1.645726681734901e-05, "loss": 1.0769, "step": 5559 }, { "epoch": 0.9, "grad_norm": 9.289353370666504, "learning_rate": 1.6455939644373074e-05, "loss": 1.4375, "step": 5560 }, { "epoch": 0.9, "grad_norm": 8.983880996704102, "learning_rate": 1.645461227638819e-05, "loss": 1.1549, "step": 5561 }, { "epoch": 0.9, "grad_norm": 9.533864974975586, "learning_rate": 1.645328471343445e-05, "loss": 1.3969, "step": 5562 }, { "epoch": 0.9, "grad_norm": 8.339910507202148, "learning_rate": 1.645195695555196e-05, "loss": 1.1675, "step": 5563 }, { "epoch": 0.9, "grad_norm": 8.98707389831543, "learning_rate": 1.6450629002780823e-05, "loss": 1.4265, "step": 5564 }, { "epoch": 0.9, "grad_norm": 8.69410514831543, "learning_rate": 1.644930085516115e-05, "loss": 1.3613, "step": 5565 }, { "epoch": 0.9, "grad_norm": 8.08252239227295, "learning_rate": 1.644797251273306e-05, "loss": 1.4705, "step": 5566 }, { "epoch": 0.9, "grad_norm": 8.079635620117188, "learning_rate": 1.644664397553668e-05, "loss": 1.3457, "step": 5567 }, { "epoch": 0.9, "grad_norm": 7.257567882537842, "learning_rate": 1.6445315243612135e-05, "loss": 1.2515, "step": 5568 }, { "epoch": 0.9, "grad_norm": 8.065139770507812, "learning_rate": 1.6443986316999565e-05, "loss": 1.2258, "step": 5569 }, { "epoch": 0.9, "grad_norm": 8.066864013671875, "learning_rate": 1.6442657195739112e-05, "loss": 1.2864, "step": 5570 }, { "epoch": 0.9, "grad_norm": 8.211201667785645, "learning_rate": 1.6441327879870916e-05, "loss": 1.3842, "step": 5571 }, { "epoch": 0.9, "grad_norm": 8.041800498962402, "learning_rate": 1.643999836943514e-05, "loss": 1.3062, "step": 5572 }, { "epoch": 0.9, "grad_norm": 8.177300453186035, "learning_rate": 1.643866866447194e-05, "loss": 1.4922, "step": 5573 }, { "epoch": 0.9, "grad_norm": 8.208232879638672, "learning_rate": 1.6437338765021483e-05, "loss": 1.2393, "step": 5574 }, { "epoch": 0.9, "grad_norm": 8.174016952514648, "learning_rate": 1.643600867112394e-05, "loss": 1.2739, "step": 5575 }, { "epoch": 0.9, "grad_norm": 7.467746734619141, "learning_rate": 1.6434678382819478e-05, "loss": 0.8938, "step": 5576 }, { "epoch": 0.9, "grad_norm": 8.222021102905273, "learning_rate": 1.6433347900148292e-05, "loss": 1.106, "step": 5577 }, { "epoch": 0.9, "grad_norm": 8.13613224029541, "learning_rate": 1.6432017223150567e-05, "loss": 1.4418, "step": 5578 }, { "epoch": 0.9, "grad_norm": 7.8440117835998535, "learning_rate": 1.6430686351866495e-05, "loss": 1.1953, "step": 5579 }, { "epoch": 0.9, "grad_norm": 9.486592292785645, "learning_rate": 1.6429355286336285e-05, "loss": 1.4357, "step": 5580 }, { "epoch": 0.9, "grad_norm": 9.444430351257324, "learning_rate": 1.6428024026600133e-05, "loss": 1.2624, "step": 5581 }, { "epoch": 0.9, "grad_norm": 9.055819511413574, "learning_rate": 1.6426692572698252e-05, "loss": 1.3424, "step": 5582 }, { "epoch": 0.9, "grad_norm": 8.557695388793945, "learning_rate": 1.6425360924670867e-05, "loss": 1.4078, "step": 5583 }, { "epoch": 0.9, "grad_norm": 8.304722785949707, "learning_rate": 1.6424029082558197e-05, "loss": 1.3184, "step": 5584 }, { "epoch": 0.9, "grad_norm": 8.174351692199707, "learning_rate": 1.6422697046400475e-05, "loss": 1.2868, "step": 5585 }, { "epoch": 0.9, "grad_norm": 8.27103328704834, "learning_rate": 1.6421364816237933e-05, "loss": 1.5, "step": 5586 }, { "epoch": 0.9, "grad_norm": 9.401922225952148, "learning_rate": 1.6420032392110816e-05, "loss": 1.0931, "step": 5587 }, { "epoch": 0.9, "grad_norm": 8.068258285522461, "learning_rate": 1.6418699774059365e-05, "loss": 1.2203, "step": 5588 }, { "epoch": 0.9, "grad_norm": 7.098601818084717, "learning_rate": 1.641736696212384e-05, "loss": 1.4371, "step": 5589 }, { "epoch": 0.9, "grad_norm": 8.054268836975098, "learning_rate": 1.64160339563445e-05, "loss": 1.5007, "step": 5590 }, { "epoch": 0.9, "grad_norm": 6.8919758796691895, "learning_rate": 1.6414700756761607e-05, "loss": 1.4471, "step": 5591 }, { "epoch": 0.9, "grad_norm": 7.901702404022217, "learning_rate": 1.6413367363415435e-05, "loss": 1.3752, "step": 5592 }, { "epoch": 0.9, "grad_norm": 8.09146785736084, "learning_rate": 1.6412033776346254e-05, "loss": 1.1778, "step": 5593 }, { "epoch": 0.9, "grad_norm": 8.116337776184082, "learning_rate": 1.6410699995594353e-05, "loss": 1.1904, "step": 5594 }, { "epoch": 0.9, "grad_norm": 7.750445365905762, "learning_rate": 1.6409366021200018e-05, "loss": 1.2224, "step": 5595 }, { "epoch": 0.9, "grad_norm": 8.13184928894043, "learning_rate": 1.6408031853203543e-05, "loss": 1.3784, "step": 5596 }, { "epoch": 0.9, "grad_norm": 9.878854751586914, "learning_rate": 1.640669749164523e-05, "loss": 1.2041, "step": 5597 }, { "epoch": 0.9, "grad_norm": 6.785094261169434, "learning_rate": 1.6405362936565383e-05, "loss": 1.2344, "step": 5598 }, { "epoch": 0.9, "grad_norm": 8.582082748413086, "learning_rate": 1.6404028188004317e-05, "loss": 1.4746, "step": 5599 }, { "epoch": 0.9, "grad_norm": 8.570526123046875, "learning_rate": 1.6402693246002338e-05, "loss": 1.3065, "step": 5600 }, { "epoch": 0.9, "grad_norm": 10.013943672180176, "learning_rate": 1.6401358110599782e-05, "loss": 1.4602, "step": 5601 }, { "epoch": 0.9, "grad_norm": 7.863314628601074, "learning_rate": 1.6400022781836976e-05, "loss": 1.271, "step": 5602 }, { "epoch": 0.9, "grad_norm": 9.204883575439453, "learning_rate": 1.6398687259754255e-05, "loss": 1.6063, "step": 5603 }, { "epoch": 0.9, "grad_norm": 8.840970993041992, "learning_rate": 1.6397351544391954e-05, "loss": 1.4454, "step": 5604 }, { "epoch": 0.9, "grad_norm": 7.959878444671631, "learning_rate": 1.6396015635790426e-05, "loss": 1.1645, "step": 5605 }, { "epoch": 0.9, "grad_norm": 7.9460015296936035, "learning_rate": 1.6394679533990024e-05, "loss": 1.3075, "step": 5606 }, { "epoch": 0.9, "grad_norm": 9.234745979309082, "learning_rate": 1.6393343239031104e-05, "loss": 1.2972, "step": 5607 }, { "epoch": 0.9, "grad_norm": 7.4295654296875, "learning_rate": 1.6392006750954032e-05, "loss": 1.0678, "step": 5608 }, { "epoch": 0.9, "grad_norm": 10.215948104858398, "learning_rate": 1.639067006979917e-05, "loss": 1.5429, "step": 5609 }, { "epoch": 0.9, "grad_norm": 8.612393379211426, "learning_rate": 1.6389333195606907e-05, "loss": 1.3015, "step": 5610 }, { "epoch": 0.9, "grad_norm": 8.51307487487793, "learning_rate": 1.6387996128417617e-05, "loss": 1.2427, "step": 5611 }, { "epoch": 0.9, "grad_norm": 8.269895553588867, "learning_rate": 1.6386658868271687e-05, "loss": 1.4883, "step": 5612 }, { "epoch": 0.9, "grad_norm": 8.035043716430664, "learning_rate": 1.6385321415209514e-05, "loss": 1.4494, "step": 5613 }, { "epoch": 0.9, "grad_norm": 7.800591468811035, "learning_rate": 1.6383983769271495e-05, "loss": 1.3982, "step": 5614 }, { "epoch": 0.9, "grad_norm": 7.4921112060546875, "learning_rate": 1.6382645930498037e-05, "loss": 1.1703, "step": 5615 }, { "epoch": 0.9, "grad_norm": 7.971167087554932, "learning_rate": 1.638130789892955e-05, "loss": 1.496, "step": 5616 }, { "epoch": 0.9, "grad_norm": 9.221023559570312, "learning_rate": 1.6379969674606452e-05, "loss": 1.4583, "step": 5617 }, { "epoch": 0.9, "grad_norm": 7.281923294067383, "learning_rate": 1.6378631257569164e-05, "loss": 1.315, "step": 5618 }, { "epoch": 0.9, "grad_norm": 7.95815896987915, "learning_rate": 1.6377292647858115e-05, "loss": 1.2716, "step": 5619 }, { "epoch": 0.9, "grad_norm": 7.606445789337158, "learning_rate": 1.6375953845513735e-05, "loss": 1.0773, "step": 5620 }, { "epoch": 0.9, "grad_norm": 8.348993301391602, "learning_rate": 1.637461485057647e-05, "loss": 1.1257, "step": 5621 }, { "epoch": 0.91, "grad_norm": 7.244253158569336, "learning_rate": 1.6373275663086766e-05, "loss": 1.1769, "step": 5622 }, { "epoch": 0.91, "grad_norm": 9.650140762329102, "learning_rate": 1.637193628308507e-05, "loss": 1.5079, "step": 5623 }, { "epoch": 0.91, "grad_norm": 7.977808952331543, "learning_rate": 1.637059671061184e-05, "loss": 1.4687, "step": 5624 }, { "epoch": 0.91, "grad_norm": 7.632570743560791, "learning_rate": 1.6369256945707545e-05, "loss": 1.3857, "step": 5625 }, { "epoch": 0.91, "grad_norm": 7.109489440917969, "learning_rate": 1.636791698841265e-05, "loss": 1.2511, "step": 5626 }, { "epoch": 0.91, "grad_norm": 9.232776641845703, "learning_rate": 1.6366576838767628e-05, "loss": 1.2169, "step": 5627 }, { "epoch": 0.91, "grad_norm": 9.196794509887695, "learning_rate": 1.6365236496812964e-05, "loss": 1.1827, "step": 5628 }, { "epoch": 0.91, "grad_norm": 9.621728897094727, "learning_rate": 1.636389596258914e-05, "loss": 1.2959, "step": 5629 }, { "epoch": 0.91, "grad_norm": 7.506319999694824, "learning_rate": 1.6362555236136653e-05, "loss": 1.236, "step": 5630 }, { "epoch": 0.91, "grad_norm": 7.889614105224609, "learning_rate": 1.6361214317496e-05, "loss": 1.4417, "step": 5631 }, { "epoch": 0.91, "grad_norm": 8.305524826049805, "learning_rate": 1.635987320670768e-05, "loss": 1.1733, "step": 5632 }, { "epoch": 0.91, "grad_norm": 6.558561325073242, "learning_rate": 1.635853190381221e-05, "loss": 1.0136, "step": 5633 }, { "epoch": 0.91, "grad_norm": 8.629351615905762, "learning_rate": 1.63571904088501e-05, "loss": 1.3487, "step": 5634 }, { "epoch": 0.91, "grad_norm": 7.662482738494873, "learning_rate": 1.6355848721861876e-05, "loss": 1.324, "step": 5635 }, { "epoch": 0.91, "grad_norm": 9.39193058013916, "learning_rate": 1.6354506842888063e-05, "loss": 1.4529, "step": 5636 }, { "epoch": 0.91, "grad_norm": 7.3063483238220215, "learning_rate": 1.6353164771969195e-05, "loss": 1.3014, "step": 5637 }, { "epoch": 0.91, "grad_norm": 7.611579895019531, "learning_rate": 1.6351822509145808e-05, "loss": 1.5351, "step": 5638 }, { "epoch": 0.91, "grad_norm": 7.093709468841553, "learning_rate": 1.6350480054458446e-05, "loss": 1.41, "step": 5639 }, { "epoch": 0.91, "grad_norm": 9.4401216506958, "learning_rate": 1.6349137407947663e-05, "loss": 1.4804, "step": 5640 }, { "epoch": 0.91, "grad_norm": 8.84472942352295, "learning_rate": 1.6347794569654014e-05, "loss": 1.4154, "step": 5641 }, { "epoch": 0.91, "grad_norm": 6.862771987915039, "learning_rate": 1.634645153961806e-05, "loss": 1.2519, "step": 5642 }, { "epoch": 0.91, "grad_norm": 8.885904312133789, "learning_rate": 1.6345108317880368e-05, "loss": 1.1737, "step": 5643 }, { "epoch": 0.91, "grad_norm": 7.919705390930176, "learning_rate": 1.6343764904481516e-05, "loss": 1.3608, "step": 5644 }, { "epoch": 0.91, "grad_norm": 9.305794715881348, "learning_rate": 1.634242129946208e-05, "loss": 1.4116, "step": 5645 }, { "epoch": 0.91, "grad_norm": 9.4457368850708, "learning_rate": 1.6341077502862646e-05, "loss": 1.3477, "step": 5646 }, { "epoch": 0.91, "grad_norm": 8.437028884887695, "learning_rate": 1.63397335147238e-05, "loss": 1.1293, "step": 5647 }, { "epoch": 0.91, "grad_norm": 8.313302040100098, "learning_rate": 1.6338389335086145e-05, "loss": 1.5311, "step": 5648 }, { "epoch": 0.91, "grad_norm": 9.810463905334473, "learning_rate": 1.6337044963990284e-05, "loss": 1.4001, "step": 5649 }, { "epoch": 0.91, "grad_norm": 10.602775573730469, "learning_rate": 1.633570040147682e-05, "loss": 1.3724, "step": 5650 }, { "epoch": 0.91, "grad_norm": 7.039052963256836, "learning_rate": 1.633435564758637e-05, "loss": 1.2968, "step": 5651 }, { "epoch": 0.91, "grad_norm": 8.10632038116455, "learning_rate": 1.6333010702359552e-05, "loss": 1.3891, "step": 5652 }, { "epoch": 0.91, "grad_norm": 8.18994426727295, "learning_rate": 1.6331665565836995e-05, "loss": 1.3312, "step": 5653 }, { "epoch": 0.91, "grad_norm": 9.146692276000977, "learning_rate": 1.6330320238059328e-05, "loss": 1.0002, "step": 5654 }, { "epoch": 0.91, "grad_norm": 7.870633602142334, "learning_rate": 1.632897471906719e-05, "loss": 1.3866, "step": 5655 }, { "epoch": 0.91, "grad_norm": 8.485367774963379, "learning_rate": 1.6327629008901224e-05, "loss": 1.0995, "step": 5656 }, { "epoch": 0.91, "grad_norm": 9.314573287963867, "learning_rate": 1.632628310760207e-05, "loss": 1.2531, "step": 5657 }, { "epoch": 0.91, "grad_norm": 7.171536922454834, "learning_rate": 1.6324937015210396e-05, "loss": 0.8854, "step": 5658 }, { "epoch": 0.91, "grad_norm": 7.602930068969727, "learning_rate": 1.6323590731766854e-05, "loss": 1.1799, "step": 5659 }, { "epoch": 0.91, "grad_norm": 8.850112915039062, "learning_rate": 1.632224425731211e-05, "loss": 1.4482, "step": 5660 }, { "epoch": 0.91, "grad_norm": 7.160019397735596, "learning_rate": 1.632089759188684e-05, "loss": 1.3867, "step": 5661 }, { "epoch": 0.91, "grad_norm": 7.242702007293701, "learning_rate": 1.6319550735531717e-05, "loss": 1.2411, "step": 5662 }, { "epoch": 0.91, "grad_norm": 8.761207580566406, "learning_rate": 1.631820368828743e-05, "loss": 1.2552, "step": 5663 }, { "epoch": 0.91, "grad_norm": 9.859394073486328, "learning_rate": 1.631685645019466e-05, "loss": 1.4619, "step": 5664 }, { "epoch": 0.91, "grad_norm": 8.21652889251709, "learning_rate": 1.631550902129411e-05, "loss": 1.5303, "step": 5665 }, { "epoch": 0.91, "grad_norm": 10.260065078735352, "learning_rate": 1.6314161401626476e-05, "loss": 1.4697, "step": 5666 }, { "epoch": 0.91, "grad_norm": 8.052861213684082, "learning_rate": 1.631281359123247e-05, "loss": 1.5015, "step": 5667 }, { "epoch": 0.91, "grad_norm": 8.262983322143555, "learning_rate": 1.6311465590152793e-05, "loss": 1.338, "step": 5668 }, { "epoch": 0.91, "grad_norm": 7.441346645355225, "learning_rate": 1.6310117398428172e-05, "loss": 1.3058, "step": 5669 }, { "epoch": 0.91, "grad_norm": 8.26574993133545, "learning_rate": 1.630876901609933e-05, "loss": 1.199, "step": 5670 }, { "epoch": 0.91, "grad_norm": 9.202828407287598, "learning_rate": 1.6307420443206988e-05, "loss": 1.2164, "step": 5671 }, { "epoch": 0.91, "grad_norm": 8.254900932312012, "learning_rate": 1.6306071679791893e-05, "loss": 1.2764, "step": 5672 }, { "epoch": 0.91, "grad_norm": 7.026710033416748, "learning_rate": 1.6304722725894778e-05, "loss": 1.104, "step": 5673 }, { "epoch": 0.91, "grad_norm": 8.549535751342773, "learning_rate": 1.6303373581556396e-05, "loss": 1.3737, "step": 5674 }, { "epoch": 0.91, "grad_norm": 9.866323471069336, "learning_rate": 1.6302024246817496e-05, "loss": 1.4659, "step": 5675 }, { "epoch": 0.91, "grad_norm": 8.792805671691895, "learning_rate": 1.630067472171883e-05, "loss": 1.552, "step": 5676 }, { "epoch": 0.91, "grad_norm": 7.450841426849365, "learning_rate": 1.6299325006301173e-05, "loss": 1.1817, "step": 5677 }, { "epoch": 0.91, "grad_norm": 9.759871482849121, "learning_rate": 1.629797510060529e-05, "loss": 1.5026, "step": 5678 }, { "epoch": 0.91, "grad_norm": 8.703347206115723, "learning_rate": 1.6296625004671957e-05, "loss": 1.3683, "step": 5679 }, { "epoch": 0.91, "grad_norm": 8.054075241088867, "learning_rate": 1.6295274718541947e-05, "loss": 1.3727, "step": 5680 }, { "epoch": 0.91, "grad_norm": 8.774043083190918, "learning_rate": 1.629392424225606e-05, "loss": 1.1967, "step": 5681 }, { "epoch": 0.91, "grad_norm": 8.7804536819458, "learning_rate": 1.6292573575855078e-05, "loss": 1.2115, "step": 5682 }, { "epoch": 0.91, "grad_norm": 7.792965412139893, "learning_rate": 1.6291222719379808e-05, "loss": 1.3887, "step": 5683 }, { "epoch": 0.92, "grad_norm": 8.91433048248291, "learning_rate": 1.6289871672871053e-05, "loss": 1.4565, "step": 5684 }, { "epoch": 0.92, "grad_norm": 9.271598815917969, "learning_rate": 1.6288520436369616e-05, "loss": 1.2866, "step": 5685 }, { "epoch": 0.92, "grad_norm": 7.427525043487549, "learning_rate": 1.6287169009916314e-05, "loss": 1.211, "step": 5686 }, { "epoch": 0.92, "grad_norm": 8.454215049743652, "learning_rate": 1.6285817393551977e-05, "loss": 1.4356, "step": 5687 }, { "epoch": 0.92, "grad_norm": 8.364550590515137, "learning_rate": 1.6284465587317423e-05, "loss": 1.2924, "step": 5688 }, { "epoch": 0.92, "grad_norm": 8.287510871887207, "learning_rate": 1.628311359125349e-05, "loss": 1.1402, "step": 5689 }, { "epoch": 0.92, "grad_norm": 7.266632080078125, "learning_rate": 1.6281761405401014e-05, "loss": 1.1405, "step": 5690 }, { "epoch": 0.92, "grad_norm": 8.739418029785156, "learning_rate": 1.628040902980084e-05, "loss": 1.2266, "step": 5691 }, { "epoch": 0.92, "grad_norm": 8.677289009094238, "learning_rate": 1.6279056464493818e-05, "loss": 1.4259, "step": 5692 }, { "epoch": 0.92, "grad_norm": 7.304751873016357, "learning_rate": 1.62777037095208e-05, "loss": 1.3448, "step": 5693 }, { "epoch": 0.92, "grad_norm": 9.882668495178223, "learning_rate": 1.6276350764922656e-05, "loss": 1.2783, "step": 5694 }, { "epoch": 0.92, "grad_norm": 8.054383277893066, "learning_rate": 1.6274997630740246e-05, "loss": 1.4746, "step": 5695 }, { "epoch": 0.92, "grad_norm": 10.314072608947754, "learning_rate": 1.6273644307014444e-05, "loss": 1.2843, "step": 5696 }, { "epoch": 0.92, "grad_norm": 7.980288982391357, "learning_rate": 1.6272290793786135e-05, "loss": 1.2962, "step": 5697 }, { "epoch": 0.92, "grad_norm": 8.82234001159668, "learning_rate": 1.6270937091096196e-05, "loss": 1.1342, "step": 5698 }, { "epoch": 0.92, "grad_norm": 8.08882999420166, "learning_rate": 1.626958319898552e-05, "loss": 1.3422, "step": 5699 }, { "epoch": 0.92, "grad_norm": 8.359796524047852, "learning_rate": 1.6268229117494997e-05, "loss": 1.3235, "step": 5700 }, { "epoch": 0.92, "grad_norm": 8.526086807250977, "learning_rate": 1.6266874846665542e-05, "loss": 1.5516, "step": 5701 }, { "epoch": 0.92, "grad_norm": 7.961855888366699, "learning_rate": 1.626552038653805e-05, "loss": 1.109, "step": 5702 }, { "epoch": 0.92, "grad_norm": 7.837517738342285, "learning_rate": 1.6264165737153438e-05, "loss": 1.276, "step": 5703 }, { "epoch": 0.92, "grad_norm": 8.015397071838379, "learning_rate": 1.6262810898552625e-05, "loss": 1.4833, "step": 5704 }, { "epoch": 0.92, "grad_norm": 9.160676956176758, "learning_rate": 1.6261455870776538e-05, "loss": 1.2649, "step": 5705 }, { "epoch": 0.92, "grad_norm": 8.215250968933105, "learning_rate": 1.6260100653866105e-05, "loss": 1.0236, "step": 5706 }, { "epoch": 0.92, "grad_norm": 8.151759147644043, "learning_rate": 1.6258745247862258e-05, "loss": 1.3215, "step": 5707 }, { "epoch": 0.92, "grad_norm": 7.997508525848389, "learning_rate": 1.6257389652805942e-05, "loss": 1.263, "step": 5708 }, { "epoch": 0.92, "grad_norm": 8.633212089538574, "learning_rate": 1.6256033868738107e-05, "loss": 1.4072, "step": 5709 }, { "epoch": 0.92, "grad_norm": 6.913481712341309, "learning_rate": 1.6254677895699702e-05, "loss": 1.4825, "step": 5710 }, { "epoch": 0.92, "grad_norm": 8.75634479522705, "learning_rate": 1.6253321733731686e-05, "loss": 1.3501, "step": 5711 }, { "epoch": 0.92, "grad_norm": 6.813934326171875, "learning_rate": 1.6251965382875027e-05, "loss": 1.2306, "step": 5712 }, { "epoch": 0.92, "grad_norm": 7.327389240264893, "learning_rate": 1.625060884317069e-05, "loss": 1.2618, "step": 5713 }, { "epoch": 0.92, "grad_norm": 8.90805721282959, "learning_rate": 1.6249252114659654e-05, "loss": 1.285, "step": 5714 }, { "epoch": 0.92, "grad_norm": 7.035309791564941, "learning_rate": 1.62478951973829e-05, "loss": 1.2214, "step": 5715 }, { "epoch": 0.92, "grad_norm": 8.34249496459961, "learning_rate": 1.6246538091381413e-05, "loss": 1.2221, "step": 5716 }, { "epoch": 0.92, "grad_norm": 12.49238109588623, "learning_rate": 1.6245180796696192e-05, "loss": 1.5009, "step": 5717 }, { "epoch": 0.92, "grad_norm": 7.388978958129883, "learning_rate": 1.624382331336823e-05, "loss": 1.1546, "step": 5718 }, { "epoch": 0.92, "grad_norm": 7.606466770172119, "learning_rate": 1.624246564143853e-05, "loss": 1.2534, "step": 5719 }, { "epoch": 0.92, "grad_norm": 8.0596342086792, "learning_rate": 1.624110778094811e-05, "loss": 1.4926, "step": 5720 }, { "epoch": 0.92, "grad_norm": 8.072484970092773, "learning_rate": 1.6239749731937972e-05, "loss": 1.1888, "step": 5721 }, { "epoch": 0.92, "grad_norm": 8.609823226928711, "learning_rate": 1.6238391494449156e-05, "loss": 1.3495, "step": 5722 }, { "epoch": 0.92, "grad_norm": 8.88032341003418, "learning_rate": 1.6237033068522673e-05, "loss": 1.3468, "step": 5723 }, { "epoch": 0.92, "grad_norm": 7.3695573806762695, "learning_rate": 1.623567445419956e-05, "loss": 1.4336, "step": 5724 }, { "epoch": 0.92, "grad_norm": 8.96261978149414, "learning_rate": 1.6234315651520862e-05, "loss": 1.2607, "step": 5725 }, { "epoch": 0.92, "grad_norm": 8.18705940246582, "learning_rate": 1.6232956660527617e-05, "loss": 1.2499, "step": 5726 }, { "epoch": 0.92, "grad_norm": 7.499212265014648, "learning_rate": 1.6231597481260875e-05, "loss": 1.178, "step": 5727 }, { "epoch": 0.92, "grad_norm": 8.340384483337402, "learning_rate": 1.62302381137617e-05, "loss": 1.2388, "step": 5728 }, { "epoch": 0.92, "grad_norm": 7.594801902770996, "learning_rate": 1.6228878558071136e-05, "loss": 1.2065, "step": 5729 }, { "epoch": 0.92, "grad_norm": 7.297041893005371, "learning_rate": 1.6227518814230268e-05, "loss": 1.1141, "step": 5730 }, { "epoch": 0.92, "grad_norm": 10.293205261230469, "learning_rate": 1.6226158882280156e-05, "loss": 1.4305, "step": 5731 }, { "epoch": 0.92, "grad_norm": 7.493716716766357, "learning_rate": 1.6224798762261885e-05, "loss": 1.1057, "step": 5732 }, { "epoch": 0.92, "grad_norm": 8.327473640441895, "learning_rate": 1.6223438454216534e-05, "loss": 1.5253, "step": 5733 }, { "epoch": 0.92, "grad_norm": 7.794563293457031, "learning_rate": 1.6222077958185197e-05, "loss": 1.3756, "step": 5734 }, { "epoch": 0.92, "grad_norm": 7.101304054260254, "learning_rate": 1.622071727420897e-05, "loss": 1.5449, "step": 5735 }, { "epoch": 0.92, "grad_norm": 8.432894706726074, "learning_rate": 1.6219356402328953e-05, "loss": 1.3422, "step": 5736 }, { "epoch": 0.92, "grad_norm": 8.506773948669434, "learning_rate": 1.621799534258625e-05, "loss": 1.3684, "step": 5737 }, { "epoch": 0.92, "grad_norm": 8.581953048706055, "learning_rate": 1.621663409502197e-05, "loss": 1.3346, "step": 5738 }, { "epoch": 0.92, "grad_norm": 7.663697719573975, "learning_rate": 1.6215272659677242e-05, "loss": 1.4363, "step": 5739 }, { "epoch": 0.92, "grad_norm": 7.294276714324951, "learning_rate": 1.6213911036593183e-05, "loss": 1.2244, "step": 5740 }, { "epoch": 0.92, "grad_norm": 8.254663467407227, "learning_rate": 1.6212549225810922e-05, "loss": 1.3005, "step": 5741 }, { "epoch": 0.92, "grad_norm": 9.255257606506348, "learning_rate": 1.621118722737159e-05, "loss": 1.221, "step": 5742 }, { "epoch": 0.92, "grad_norm": 7.730784893035889, "learning_rate": 1.620982504131634e-05, "loss": 1.2463, "step": 5743 }, { "epoch": 0.92, "grad_norm": 8.29914665222168, "learning_rate": 1.6208462667686306e-05, "loss": 1.2845, "step": 5744 }, { "epoch": 0.92, "grad_norm": 8.463778495788574, "learning_rate": 1.620710010652265e-05, "loss": 1.1792, "step": 5745 }, { "epoch": 0.92, "grad_norm": 9.29736614227295, "learning_rate": 1.620573735786652e-05, "loss": 1.3499, "step": 5746 }, { "epoch": 0.93, "grad_norm": 8.703298568725586, "learning_rate": 1.6204374421759086e-05, "loss": 1.2242, "step": 5747 }, { "epoch": 0.93, "grad_norm": 8.521716117858887, "learning_rate": 1.6203011298241514e-05, "loss": 1.3945, "step": 5748 }, { "epoch": 0.93, "grad_norm": 6.871730327606201, "learning_rate": 1.6201647987354983e-05, "loss": 1.3567, "step": 5749 }, { "epoch": 0.93, "grad_norm": 8.148078918457031, "learning_rate": 1.620028448914067e-05, "loss": 1.381, "step": 5750 }, { "epoch": 0.93, "grad_norm": 7.718022346496582, "learning_rate": 1.6198920803639757e-05, "loss": 1.3564, "step": 5751 }, { "epoch": 0.93, "grad_norm": 7.552688121795654, "learning_rate": 1.6197556930893443e-05, "loss": 1.1858, "step": 5752 }, { "epoch": 0.93, "grad_norm": 6.9899468421936035, "learning_rate": 1.6196192870942923e-05, "loss": 1.3115, "step": 5753 }, { "epoch": 0.93, "grad_norm": 8.499974250793457, "learning_rate": 1.6194828623829394e-05, "loss": 1.2732, "step": 5754 }, { "epoch": 0.93, "grad_norm": 6.992260456085205, "learning_rate": 1.6193464189594076e-05, "loss": 1.2998, "step": 5755 }, { "epoch": 0.93, "grad_norm": 9.785601615905762, "learning_rate": 1.6192099568278176e-05, "loss": 1.4417, "step": 5756 }, { "epoch": 0.93, "grad_norm": 9.04407787322998, "learning_rate": 1.6190734759922914e-05, "loss": 1.3352, "step": 5757 }, { "epoch": 0.93, "grad_norm": 8.057098388671875, "learning_rate": 1.6189369764569516e-05, "loss": 1.4687, "step": 5758 }, { "epoch": 0.93, "grad_norm": 7.714899063110352, "learning_rate": 1.6188004582259216e-05, "loss": 1.3709, "step": 5759 }, { "epoch": 0.93, "grad_norm": 8.24851131439209, "learning_rate": 1.6186639213033247e-05, "loss": 1.3143, "step": 5760 }, { "epoch": 0.93, "grad_norm": 8.183902740478516, "learning_rate": 1.618527365693285e-05, "loss": 1.3635, "step": 5761 }, { "epoch": 0.93, "grad_norm": 8.170514106750488, "learning_rate": 1.6183907913999282e-05, "loss": 1.3126, "step": 5762 }, { "epoch": 0.93, "grad_norm": 7.502264022827148, "learning_rate": 1.6182541984273784e-05, "loss": 0.951, "step": 5763 }, { "epoch": 0.93, "grad_norm": 7.832949161529541, "learning_rate": 1.618117586779763e-05, "loss": 1.4819, "step": 5764 }, { "epoch": 0.93, "grad_norm": 9.008597373962402, "learning_rate": 1.6179809564612076e-05, "loss": 1.2157, "step": 5765 }, { "epoch": 0.93, "grad_norm": 8.070618629455566, "learning_rate": 1.617844307475839e-05, "loss": 1.4372, "step": 5766 }, { "epoch": 0.93, "grad_norm": 9.727049827575684, "learning_rate": 1.617707639827786e-05, "loss": 1.3163, "step": 5767 }, { "epoch": 0.93, "grad_norm": 8.310452461242676, "learning_rate": 1.6175709535211754e-05, "loss": 1.2335, "step": 5768 }, { "epoch": 0.93, "grad_norm": 7.055182933807373, "learning_rate": 1.617434248560137e-05, "loss": 1.0898, "step": 5769 }, { "epoch": 0.93, "grad_norm": 7.601905345916748, "learning_rate": 1.6172975249487995e-05, "loss": 1.1946, "step": 5770 }, { "epoch": 0.93, "grad_norm": 9.018762588500977, "learning_rate": 1.6171607826912935e-05, "loss": 1.4007, "step": 5771 }, { "epoch": 0.93, "grad_norm": 7.471350193023682, "learning_rate": 1.6170240217917487e-05, "loss": 1.5044, "step": 5772 }, { "epoch": 0.93, "grad_norm": 8.135873794555664, "learning_rate": 1.6168872422542966e-05, "loss": 1.4034, "step": 5773 }, { "epoch": 0.93, "grad_norm": 6.782045841217041, "learning_rate": 1.6167504440830682e-05, "loss": 1.4211, "step": 5774 }, { "epoch": 0.93, "grad_norm": 8.727498054504395, "learning_rate": 1.6166136272821966e-05, "loss": 1.3101, "step": 5775 }, { "epoch": 0.93, "grad_norm": 7.362107276916504, "learning_rate": 1.616476791855814e-05, "loss": 1.5387, "step": 5776 }, { "epoch": 0.93, "grad_norm": 7.219125747680664, "learning_rate": 1.6163399378080534e-05, "loss": 1.143, "step": 5777 }, { "epoch": 0.93, "grad_norm": 9.132499694824219, "learning_rate": 1.6162030651430486e-05, "loss": 1.3163, "step": 5778 }, { "epoch": 0.93, "grad_norm": 8.378528594970703, "learning_rate": 1.6160661738649345e-05, "loss": 1.2029, "step": 5779 }, { "epoch": 0.93, "grad_norm": 8.940356254577637, "learning_rate": 1.615929263977846e-05, "loss": 1.3694, "step": 5780 }, { "epoch": 0.93, "grad_norm": 8.268366813659668, "learning_rate": 1.615792335485918e-05, "loss": 1.183, "step": 5781 }, { "epoch": 0.93, "grad_norm": 7.83988094329834, "learning_rate": 1.6156553883932876e-05, "loss": 1.2846, "step": 5782 }, { "epoch": 0.93, "grad_norm": 7.2827959060668945, "learning_rate": 1.6155184227040907e-05, "loss": 1.0827, "step": 5783 }, { "epoch": 0.93, "grad_norm": 8.262015342712402, "learning_rate": 1.6153814384224644e-05, "loss": 1.4746, "step": 5784 }, { "epoch": 0.93, "grad_norm": 7.0347981452941895, "learning_rate": 1.6152444355525466e-05, "loss": 1.2233, "step": 5785 }, { "epoch": 0.93, "grad_norm": 8.448637962341309, "learning_rate": 1.615107414098476e-05, "loss": 1.2084, "step": 5786 }, { "epoch": 0.93, "grad_norm": 9.645307540893555, "learning_rate": 1.6149703740643912e-05, "loss": 1.0581, "step": 5787 }, { "epoch": 0.93, "grad_norm": 10.175827026367188, "learning_rate": 1.6148333154544315e-05, "loss": 1.4322, "step": 5788 }, { "epoch": 0.93, "grad_norm": 8.498770713806152, "learning_rate": 1.6146962382727374e-05, "loss": 1.3852, "step": 5789 }, { "epoch": 0.93, "grad_norm": 8.488215446472168, "learning_rate": 1.614559142523449e-05, "loss": 1.1019, "step": 5790 }, { "epoch": 0.93, "grad_norm": 8.437700271606445, "learning_rate": 1.614422028210708e-05, "loss": 1.3037, "step": 5791 }, { "epoch": 0.93, "grad_norm": 7.9342145919799805, "learning_rate": 1.6142848953386552e-05, "loss": 1.1896, "step": 5792 }, { "epoch": 0.93, "grad_norm": 8.51221752166748, "learning_rate": 1.6141477439114336e-05, "loss": 1.3476, "step": 5793 }, { "epoch": 0.93, "grad_norm": 8.382949829101562, "learning_rate": 1.6140105739331856e-05, "loss": 1.4256, "step": 5794 }, { "epoch": 0.93, "grad_norm": 9.401302337646484, "learning_rate": 1.6138733854080545e-05, "loss": 1.1742, "step": 5795 }, { "epoch": 0.93, "grad_norm": 8.07795524597168, "learning_rate": 1.613736178340185e-05, "loss": 1.3565, "step": 5796 }, { "epoch": 0.93, "grad_norm": 8.896900177001953, "learning_rate": 1.6135989527337206e-05, "loss": 1.4272, "step": 5797 }, { "epoch": 0.93, "grad_norm": 7.939786911010742, "learning_rate": 1.613461708592807e-05, "loss": 1.372, "step": 5798 }, { "epoch": 0.93, "grad_norm": 8.55784797668457, "learning_rate": 1.6133244459215894e-05, "loss": 1.6089, "step": 5799 }, { "epoch": 0.93, "grad_norm": 8.321921348571777, "learning_rate": 1.613187164724215e-05, "loss": 1.4605, "step": 5800 }, { "epoch": 0.93, "grad_norm": 8.596748352050781, "learning_rate": 1.613049865004829e-05, "loss": 1.423, "step": 5801 }, { "epoch": 0.93, "grad_norm": 7.468973159790039, "learning_rate": 1.6129125467675794e-05, "loss": 1.4017, "step": 5802 }, { "epoch": 0.93, "grad_norm": 10.740744590759277, "learning_rate": 1.6127752100166146e-05, "loss": 1.4204, "step": 5803 }, { "epoch": 0.93, "grad_norm": 7.619846820831299, "learning_rate": 1.612637854756082e-05, "loss": 1.2281, "step": 5804 }, { "epoch": 0.93, "grad_norm": 7.693716049194336, "learning_rate": 1.6125004809901313e-05, "loss": 1.3954, "step": 5805 }, { "epoch": 0.93, "grad_norm": 9.549215316772461, "learning_rate": 1.6123630887229117e-05, "loss": 1.3485, "step": 5806 }, { "epoch": 0.93, "grad_norm": 9.088991165161133, "learning_rate": 1.6122256779585736e-05, "loss": 1.6559, "step": 5807 }, { "epoch": 0.93, "grad_norm": 8.037068367004395, "learning_rate": 1.6120882487012673e-05, "loss": 1.3066, "step": 5808 }, { "epoch": 0.94, "grad_norm": 9.093257904052734, "learning_rate": 1.6119508009551437e-05, "loss": 1.3246, "step": 5809 }, { "epoch": 0.94, "grad_norm": 9.534400939941406, "learning_rate": 1.6118133347243557e-05, "loss": 1.2703, "step": 5810 }, { "epoch": 0.94, "grad_norm": 10.771445274353027, "learning_rate": 1.6116758500130546e-05, "loss": 1.296, "step": 5811 }, { "epoch": 0.94, "grad_norm": 9.253551483154297, "learning_rate": 1.6115383468253937e-05, "loss": 1.2713, "step": 5812 }, { "epoch": 0.94, "grad_norm": 8.480095863342285, "learning_rate": 1.611400825165526e-05, "loss": 1.4443, "step": 5813 }, { "epoch": 0.94, "grad_norm": 8.019623756408691, "learning_rate": 1.611263285037606e-05, "loss": 1.3184, "step": 5814 }, { "epoch": 0.94, "grad_norm": 7.668603897094727, "learning_rate": 1.6111257264457884e-05, "loss": 1.1454, "step": 5815 }, { "epoch": 0.94, "grad_norm": 8.305614471435547, "learning_rate": 1.610988149394228e-05, "loss": 1.2314, "step": 5816 }, { "epoch": 0.94, "grad_norm": 7.397609233856201, "learning_rate": 1.61085055388708e-05, "loss": 1.3581, "step": 5817 }, { "epoch": 0.94, "grad_norm": 8.378396987915039, "learning_rate": 1.6107129399285016e-05, "loss": 1.2573, "step": 5818 }, { "epoch": 0.94, "grad_norm": 7.910336494445801, "learning_rate": 1.6105753075226486e-05, "loss": 1.1552, "step": 5819 }, { "epoch": 0.94, "grad_norm": 8.944500923156738, "learning_rate": 1.6104376566736794e-05, "loss": 1.3085, "step": 5820 }, { "epoch": 0.94, "grad_norm": 6.98545503616333, "learning_rate": 1.610299987385751e-05, "loss": 1.1841, "step": 5821 }, { "epoch": 0.94, "grad_norm": 8.330595016479492, "learning_rate": 1.6101622996630223e-05, "loss": 1.3316, "step": 5822 }, { "epoch": 0.94, "grad_norm": 7.697747230529785, "learning_rate": 1.6100245935096522e-05, "loss": 1.2698, "step": 5823 }, { "epoch": 0.94, "grad_norm": 9.17243480682373, "learning_rate": 1.6098868689298003e-05, "loss": 1.4781, "step": 5824 }, { "epoch": 0.94, "grad_norm": 7.319958209991455, "learning_rate": 1.6097491259276265e-05, "loss": 1.4883, "step": 5825 }, { "epoch": 0.94, "grad_norm": 7.605210781097412, "learning_rate": 1.609611364507292e-05, "loss": 1.3294, "step": 5826 }, { "epoch": 0.94, "grad_norm": 8.300721168518066, "learning_rate": 1.6094735846729573e-05, "loss": 1.2205, "step": 5827 }, { "epoch": 0.94, "grad_norm": 7.4087748527526855, "learning_rate": 1.609335786428785e-05, "loss": 1.3564, "step": 5828 }, { "epoch": 0.94, "grad_norm": 8.210562705993652, "learning_rate": 1.6091979697789368e-05, "loss": 1.2933, "step": 5829 }, { "epoch": 0.94, "grad_norm": 10.011859893798828, "learning_rate": 1.609060134727576e-05, "loss": 1.3065, "step": 5830 }, { "epoch": 0.94, "grad_norm": 8.922436714172363, "learning_rate": 1.6089222812788657e-05, "loss": 1.5174, "step": 5831 }, { "epoch": 0.94, "grad_norm": 9.261383056640625, "learning_rate": 1.6087844094369708e-05, "loss": 1.6311, "step": 5832 }, { "epoch": 0.94, "grad_norm": 10.51838493347168, "learning_rate": 1.6086465192060547e-05, "loss": 1.3029, "step": 5833 }, { "epoch": 0.94, "grad_norm": 9.447335243225098, "learning_rate": 1.608508610590283e-05, "loss": 1.3571, "step": 5834 }, { "epoch": 0.94, "grad_norm": 8.904068946838379, "learning_rate": 1.6083706835938218e-05, "loss": 1.3177, "step": 5835 }, { "epoch": 0.94, "grad_norm": 8.962750434875488, "learning_rate": 1.6082327382208368e-05, "loss": 1.3013, "step": 5836 }, { "epoch": 0.94, "grad_norm": 7.997025966644287, "learning_rate": 1.6080947744754948e-05, "loss": 1.0776, "step": 5837 }, { "epoch": 0.94, "grad_norm": 9.438255310058594, "learning_rate": 1.6079567923619638e-05, "loss": 1.2142, "step": 5838 }, { "epoch": 0.94, "grad_norm": 9.234829902648926, "learning_rate": 1.607818791884411e-05, "loss": 1.2306, "step": 5839 }, { "epoch": 0.94, "grad_norm": 6.573995113372803, "learning_rate": 1.607680773047005e-05, "loss": 1.3654, "step": 5840 }, { "epoch": 0.94, "grad_norm": 9.08246898651123, "learning_rate": 1.607542735853915e-05, "loss": 1.3147, "step": 5841 }, { "epoch": 0.94, "grad_norm": 9.025193214416504, "learning_rate": 1.6074046803093108e-05, "loss": 1.3574, "step": 5842 }, { "epoch": 0.94, "grad_norm": 9.53503131866455, "learning_rate": 1.6072666064173618e-05, "loss": 1.4389, "step": 5843 }, { "epoch": 0.94, "grad_norm": 7.207415580749512, "learning_rate": 1.607128514182239e-05, "loss": 1.167, "step": 5844 }, { "epoch": 0.94, "grad_norm": 7.86125373840332, "learning_rate": 1.6069904036081138e-05, "loss": 1.2094, "step": 5845 }, { "epoch": 0.94, "grad_norm": 8.865315437316895, "learning_rate": 1.606852274699158e-05, "loss": 1.4059, "step": 5846 }, { "epoch": 0.94, "grad_norm": 7.997402191162109, "learning_rate": 1.606714127459544e-05, "loss": 1.4021, "step": 5847 }, { "epoch": 0.94, "grad_norm": 7.1129536628723145, "learning_rate": 1.6065759618934444e-05, "loss": 1.1939, "step": 5848 }, { "epoch": 0.94, "grad_norm": 7.315049648284912, "learning_rate": 1.6064377780050326e-05, "loss": 1.1095, "step": 5849 }, { "epoch": 0.94, "grad_norm": 7.621871471405029, "learning_rate": 1.6062995757984832e-05, "loss": 1.2164, "step": 5850 }, { "epoch": 0.94, "grad_norm": 7.483527183532715, "learning_rate": 1.60616135527797e-05, "loss": 1.2813, "step": 5851 }, { "epoch": 0.94, "grad_norm": 7.541978359222412, "learning_rate": 1.6060231164476686e-05, "loss": 1.4738, "step": 5852 }, { "epoch": 0.94, "grad_norm": 7.275294780731201, "learning_rate": 1.6058848593117543e-05, "loss": 0.9952, "step": 5853 }, { "epoch": 0.94, "grad_norm": 8.212891578674316, "learning_rate": 1.605746583874404e-05, "loss": 1.2929, "step": 5854 }, { "epoch": 0.94, "grad_norm": 7.969964504241943, "learning_rate": 1.6056082901397935e-05, "loss": 1.3, "step": 5855 }, { "epoch": 0.94, "grad_norm": 7.658880710601807, "learning_rate": 1.6054699781121005e-05, "loss": 1.3501, "step": 5856 }, { "epoch": 0.94, "grad_norm": 8.404703140258789, "learning_rate": 1.6053316477955035e-05, "loss": 1.4129, "step": 5857 }, { "epoch": 0.94, "grad_norm": 9.3441743850708, "learning_rate": 1.60519329919418e-05, "loss": 1.4795, "step": 5858 }, { "epoch": 0.94, "grad_norm": 9.8771390914917, "learning_rate": 1.6050549323123095e-05, "loss": 1.5291, "step": 5859 }, { "epoch": 0.94, "grad_norm": 8.387494087219238, "learning_rate": 1.6049165471540717e-05, "loss": 1.3056, "step": 5860 }, { "epoch": 0.94, "grad_norm": 9.256875038146973, "learning_rate": 1.604778143723646e-05, "loss": 1.2228, "step": 5861 }, { "epoch": 0.94, "grad_norm": 9.131057739257812, "learning_rate": 1.6046397220252135e-05, "loss": 1.3069, "step": 5862 }, { "epoch": 0.94, "grad_norm": 8.829757690429688, "learning_rate": 1.6045012820629554e-05, "loss": 1.1647, "step": 5863 }, { "epoch": 0.94, "grad_norm": 10.788813591003418, "learning_rate": 1.6043628238410535e-05, "loss": 1.4067, "step": 5864 }, { "epoch": 0.94, "grad_norm": 8.546467781066895, "learning_rate": 1.6042243473636895e-05, "loss": 1.3068, "step": 5865 }, { "epoch": 0.94, "grad_norm": 8.320292472839355, "learning_rate": 1.6040858526350467e-05, "loss": 1.1501, "step": 5866 }, { "epoch": 0.94, "grad_norm": 7.679238319396973, "learning_rate": 1.6039473396593085e-05, "loss": 1.3696, "step": 5867 }, { "epoch": 0.94, "grad_norm": 9.556317329406738, "learning_rate": 1.6038088084406588e-05, "loss": 1.2872, "step": 5868 }, { "epoch": 0.94, "grad_norm": 8.554826736450195, "learning_rate": 1.6036702589832823e-05, "loss": 1.3667, "step": 5869 }, { "epoch": 0.94, "grad_norm": 9.207618713378906, "learning_rate": 1.6035316912913638e-05, "loss": 1.4216, "step": 5870 }, { "epoch": 0.95, "grad_norm": 9.126968383789062, "learning_rate": 1.6033931053690888e-05, "loss": 1.2616, "step": 5871 }, { "epoch": 0.95, "grad_norm": 8.219630241394043, "learning_rate": 1.6032545012206435e-05, "loss": 1.1022, "step": 5872 }, { "epoch": 0.95, "grad_norm": 8.729851722717285, "learning_rate": 1.6031158788502146e-05, "loss": 1.205, "step": 5873 }, { "epoch": 0.95, "grad_norm": 7.583488464355469, "learning_rate": 1.6029772382619894e-05, "loss": 1.2537, "step": 5874 }, { "epoch": 0.95, "grad_norm": 6.897061824798584, "learning_rate": 1.6028385794601557e-05, "loss": 1.3077, "step": 5875 }, { "epoch": 0.95, "grad_norm": 9.074992179870605, "learning_rate": 1.602699902448902e-05, "loss": 1.1431, "step": 5876 }, { "epoch": 0.95, "grad_norm": 7.616514682769775, "learning_rate": 1.602561207232417e-05, "loss": 1.5913, "step": 5877 }, { "epoch": 0.95, "grad_norm": 7.776228904724121, "learning_rate": 1.6024224938148902e-05, "loss": 1.3997, "step": 5878 }, { "epoch": 0.95, "grad_norm": 8.789432525634766, "learning_rate": 1.6022837622005116e-05, "loss": 1.2574, "step": 5879 }, { "epoch": 0.95, "grad_norm": 7.407373905181885, "learning_rate": 1.6021450123934716e-05, "loss": 1.2976, "step": 5880 }, { "epoch": 0.95, "grad_norm": 6.861759185791016, "learning_rate": 1.6020062443979616e-05, "loss": 1.0418, "step": 5881 }, { "epoch": 0.95, "grad_norm": 8.933060646057129, "learning_rate": 1.601867458218173e-05, "loss": 1.3716, "step": 5882 }, { "epoch": 0.95, "grad_norm": 8.34827709197998, "learning_rate": 1.601728653858298e-05, "loss": 1.3341, "step": 5883 }, { "epoch": 0.95, "grad_norm": 8.413507461547852, "learning_rate": 1.6015898313225294e-05, "loss": 1.2455, "step": 5884 }, { "epoch": 0.95, "grad_norm": 9.616353988647461, "learning_rate": 1.6014509906150604e-05, "loss": 1.4405, "step": 5885 }, { "epoch": 0.95, "grad_norm": 8.862931251525879, "learning_rate": 1.6013121317400852e-05, "loss": 1.086, "step": 5886 }, { "epoch": 0.95, "grad_norm": 8.864651679992676, "learning_rate": 1.601173254701798e-05, "loss": 1.4089, "step": 5887 }, { "epoch": 0.95, "grad_norm": 8.430753707885742, "learning_rate": 1.601034359504393e-05, "loss": 1.6435, "step": 5888 }, { "epoch": 0.95, "grad_norm": 9.526534080505371, "learning_rate": 1.600895446152067e-05, "loss": 1.4878, "step": 5889 }, { "epoch": 0.95, "grad_norm": 7.021139621734619, "learning_rate": 1.600756514649015e-05, "loss": 1.2977, "step": 5890 }, { "epoch": 0.95, "grad_norm": 9.702686309814453, "learning_rate": 1.600617564999434e-05, "loss": 1.3226, "step": 5891 }, { "epoch": 0.95, "grad_norm": 7.238156795501709, "learning_rate": 1.6004785972075212e-05, "loss": 1.0398, "step": 5892 }, { "epoch": 0.95, "grad_norm": 10.549025535583496, "learning_rate": 1.600339611277474e-05, "loss": 1.5076, "step": 5893 }, { "epoch": 0.95, "grad_norm": 9.316636085510254, "learning_rate": 1.600200607213491e-05, "loss": 1.4528, "step": 5894 }, { "epoch": 0.95, "grad_norm": 8.638714790344238, "learning_rate": 1.6000615850197708e-05, "loss": 1.5681, "step": 5895 }, { "epoch": 0.95, "grad_norm": 6.786914825439453, "learning_rate": 1.5999225447005127e-05, "loss": 1.296, "step": 5896 }, { "epoch": 0.95, "grad_norm": 8.635381698608398, "learning_rate": 1.5997834862599164e-05, "loss": 1.2258, "step": 5897 }, { "epoch": 0.95, "grad_norm": 8.069222450256348, "learning_rate": 1.5996444097021826e-05, "loss": 1.485, "step": 5898 }, { "epoch": 0.95, "grad_norm": 10.312408447265625, "learning_rate": 1.5995053150315124e-05, "loss": 1.5247, "step": 5899 }, { "epoch": 0.95, "grad_norm": 8.276908874511719, "learning_rate": 1.5993662022521066e-05, "loss": 1.1214, "step": 5900 }, { "epoch": 0.95, "grad_norm": 7.816133499145508, "learning_rate": 1.5992270713681678e-05, "loss": 1.3575, "step": 5901 }, { "epoch": 0.95, "grad_norm": 7.358316421508789, "learning_rate": 1.599087922383899e-05, "loss": 1.3451, "step": 5902 }, { "epoch": 0.95, "grad_norm": 9.466357231140137, "learning_rate": 1.5989487553035022e-05, "loss": 1.1994, "step": 5903 }, { "epoch": 0.95, "grad_norm": 7.943315029144287, "learning_rate": 1.5988095701311825e-05, "loss": 1.1476, "step": 5904 }, { "epoch": 0.95, "grad_norm": 7.733428955078125, "learning_rate": 1.5986703668711426e-05, "loss": 1.4147, "step": 5905 }, { "epoch": 0.95, "grad_norm": 7.600646018981934, "learning_rate": 1.5985311455275887e-05, "loss": 1.1234, "step": 5906 }, { "epoch": 0.95, "grad_norm": 7.768596172332764, "learning_rate": 1.5983919061047253e-05, "loss": 1.5287, "step": 5907 }, { "epoch": 0.95, "grad_norm": 7.039245128631592, "learning_rate": 1.5982526486067586e-05, "loss": 1.2655, "step": 5908 }, { "epoch": 0.95, "grad_norm": 7.299607276916504, "learning_rate": 1.5981133730378952e-05, "loss": 1.2414, "step": 5909 }, { "epoch": 0.95, "grad_norm": 8.213630676269531, "learning_rate": 1.5979740794023415e-05, "loss": 1.3521, "step": 5910 }, { "epoch": 0.95, "grad_norm": 7.61501407623291, "learning_rate": 1.5978347677043055e-05, "loss": 1.0472, "step": 5911 }, { "epoch": 0.95, "grad_norm": 7.322629451751709, "learning_rate": 1.597695437947995e-05, "loss": 1.3249, "step": 5912 }, { "epoch": 0.95, "grad_norm": 10.200058937072754, "learning_rate": 1.597556090137619e-05, "loss": 1.4616, "step": 5913 }, { "epoch": 0.95, "grad_norm": 7.1402740478515625, "learning_rate": 1.5974167242773863e-05, "loss": 1.2138, "step": 5914 }, { "epoch": 0.95, "grad_norm": 8.283209800720215, "learning_rate": 1.5972773403715063e-05, "loss": 1.0949, "step": 5915 }, { "epoch": 0.95, "grad_norm": 10.157167434692383, "learning_rate": 1.59713793842419e-05, "loss": 1.2992, "step": 5916 }, { "epoch": 0.95, "grad_norm": 8.9927339553833, "learning_rate": 1.596998518439648e-05, "loss": 1.4775, "step": 5917 }, { "epoch": 0.95, "grad_norm": 8.874640464782715, "learning_rate": 1.596859080422091e-05, "loss": 1.1534, "step": 5918 }, { "epoch": 0.95, "grad_norm": 9.182759284973145, "learning_rate": 1.596719624375732e-05, "loss": 1.0918, "step": 5919 }, { "epoch": 0.95, "grad_norm": 8.789820671081543, "learning_rate": 1.5965801503047824e-05, "loss": 1.3098, "step": 5920 }, { "epoch": 0.95, "grad_norm": 9.560174942016602, "learning_rate": 1.5964406582134556e-05, "loss": 1.3727, "step": 5921 }, { "epoch": 0.95, "grad_norm": 10.306095123291016, "learning_rate": 1.5963011481059653e-05, "loss": 1.591, "step": 5922 }, { "epoch": 0.95, "grad_norm": 10.100249290466309, "learning_rate": 1.5961616199865247e-05, "loss": 1.5628, "step": 5923 }, { "epoch": 0.95, "grad_norm": 8.580833435058594, "learning_rate": 1.59602207385935e-05, "loss": 1.3727, "step": 5924 }, { "epoch": 0.95, "grad_norm": 8.561400413513184, "learning_rate": 1.5958825097286547e-05, "loss": 1.3379, "step": 5925 }, { "epoch": 0.95, "grad_norm": 9.237763404846191, "learning_rate": 1.5957429275986553e-05, "loss": 1.4866, "step": 5926 }, { "epoch": 0.95, "grad_norm": 9.90740966796875, "learning_rate": 1.595603327473568e-05, "loss": 1.6187, "step": 5927 }, { "epoch": 0.95, "grad_norm": 8.59878158569336, "learning_rate": 1.5954637093576096e-05, "loss": 1.4851, "step": 5928 }, { "epoch": 0.95, "grad_norm": 7.7404375076293945, "learning_rate": 1.5953240732549974e-05, "loss": 1.2724, "step": 5929 }, { "epoch": 0.95, "grad_norm": 8.832649230957031, "learning_rate": 1.595184419169949e-05, "loss": 1.0635, "step": 5930 }, { "epoch": 0.95, "grad_norm": 7.795347690582275, "learning_rate": 1.5950447471066827e-05, "loss": 1.2169, "step": 5931 }, { "epoch": 0.95, "grad_norm": 7.644548416137695, "learning_rate": 1.5949050570694184e-05, "loss": 1.2077, "step": 5932 }, { "epoch": 0.96, "grad_norm": 8.461718559265137, "learning_rate": 1.5947653490623744e-05, "loss": 1.272, "step": 5933 }, { "epoch": 0.96, "grad_norm": 8.6852388381958, "learning_rate": 1.5946256230897715e-05, "loss": 1.3265, "step": 5934 }, { "epoch": 0.96, "grad_norm": 10.279351234436035, "learning_rate": 1.5944858791558302e-05, "loss": 1.3163, "step": 5935 }, { "epoch": 0.96, "grad_norm": 10.529058456420898, "learning_rate": 1.594346117264771e-05, "loss": 1.5728, "step": 5936 }, { "epoch": 0.96, "grad_norm": 9.522162437438965, "learning_rate": 1.5942063374208165e-05, "loss": 1.2089, "step": 5937 }, { "epoch": 0.96, "grad_norm": 10.993865013122559, "learning_rate": 1.5940665396281883e-05, "loss": 1.3755, "step": 5938 }, { "epoch": 0.96, "grad_norm": 8.431547164916992, "learning_rate": 1.5939267238911096e-05, "loss": 1.2836, "step": 5939 }, { "epoch": 0.96, "grad_norm": 7.423593044281006, "learning_rate": 1.5937868902138035e-05, "loss": 1.3392, "step": 5940 }, { "epoch": 0.96, "grad_norm": 8.02773380279541, "learning_rate": 1.5936470386004932e-05, "loss": 1.4749, "step": 5941 }, { "epoch": 0.96, "grad_norm": 8.51757526397705, "learning_rate": 1.593507169055404e-05, "loss": 1.2843, "step": 5942 }, { "epoch": 0.96, "grad_norm": 8.1514253616333, "learning_rate": 1.5933672815827603e-05, "loss": 1.252, "step": 5943 }, { "epoch": 0.96, "grad_norm": 9.131309509277344, "learning_rate": 1.5932273761867877e-05, "loss": 1.4223, "step": 5944 }, { "epoch": 0.96, "grad_norm": 7.5753068923950195, "learning_rate": 1.593087452871712e-05, "loss": 1.4006, "step": 5945 }, { "epoch": 0.96, "grad_norm": 8.71946907043457, "learning_rate": 1.59294751164176e-05, "loss": 1.4665, "step": 5946 }, { "epoch": 0.96, "grad_norm": 7.988361835479736, "learning_rate": 1.592807552501159e-05, "loss": 1.4413, "step": 5947 }, { "epoch": 0.96, "grad_norm": 9.09757137298584, "learning_rate": 1.592667575454136e-05, "loss": 1.5464, "step": 5948 }, { "epoch": 0.96, "grad_norm": 9.013455390930176, "learning_rate": 1.59252758050492e-05, "loss": 1.3307, "step": 5949 }, { "epoch": 0.96, "grad_norm": 9.06326675415039, "learning_rate": 1.592387567657739e-05, "loss": 1.3131, "step": 5950 }, { "epoch": 0.96, "grad_norm": 7.13545560836792, "learning_rate": 1.592247536916822e-05, "loss": 1.1331, "step": 5951 }, { "epoch": 0.96, "grad_norm": 6.533105373382568, "learning_rate": 1.5921074882863997e-05, "loss": 0.9952, "step": 5952 }, { "epoch": 0.96, "grad_norm": 6.430059909820557, "learning_rate": 1.591967421770702e-05, "loss": 1.3385, "step": 5953 }, { "epoch": 0.96, "grad_norm": 7.138305187225342, "learning_rate": 1.5918273373739596e-05, "loss": 1.3108, "step": 5954 }, { "epoch": 0.96, "grad_norm": 10.232789993286133, "learning_rate": 1.591687235100404e-05, "loss": 1.253, "step": 5955 }, { "epoch": 0.96, "grad_norm": 9.25677490234375, "learning_rate": 1.591547114954267e-05, "loss": 1.4531, "step": 5956 }, { "epoch": 0.96, "grad_norm": 8.592665672302246, "learning_rate": 1.5914069769397816e-05, "loss": 1.3343, "step": 5957 }, { "epoch": 0.96, "grad_norm": 7.288995742797852, "learning_rate": 1.5912668210611805e-05, "loss": 1.4819, "step": 5958 }, { "epoch": 0.96, "grad_norm": 8.124587059020996, "learning_rate": 1.591126647322697e-05, "loss": 1.2439, "step": 5959 }, { "epoch": 0.96, "grad_norm": 7.247494697570801, "learning_rate": 1.5909864557285656e-05, "loss": 1.1363, "step": 5960 }, { "epoch": 0.96, "grad_norm": 8.123369216918945, "learning_rate": 1.590846246283021e-05, "loss": 1.3119, "step": 5961 }, { "epoch": 0.96, "grad_norm": 8.291203498840332, "learning_rate": 1.5907060189902984e-05, "loss": 1.1204, "step": 5962 }, { "epoch": 0.96, "grad_norm": 8.919044494628906, "learning_rate": 1.590565773854633e-05, "loss": 1.3482, "step": 5963 }, { "epoch": 0.96, "grad_norm": 9.148136138916016, "learning_rate": 1.5904255108802612e-05, "loss": 1.6096, "step": 5964 }, { "epoch": 0.96, "grad_norm": 9.833090782165527, "learning_rate": 1.5902852300714205e-05, "loss": 1.2821, "step": 5965 }, { "epoch": 0.96, "grad_norm": 8.342939376831055, "learning_rate": 1.5901449314323475e-05, "loss": 1.3172, "step": 5966 }, { "epoch": 0.96, "grad_norm": 8.544074058532715, "learning_rate": 1.59000461496728e-05, "loss": 1.3867, "step": 5967 }, { "epoch": 0.96, "grad_norm": 8.512044906616211, "learning_rate": 1.5898642806804573e-05, "loss": 1.5119, "step": 5968 }, { "epoch": 0.96, "grad_norm": 8.058045387268066, "learning_rate": 1.5897239285761175e-05, "loss": 1.4858, "step": 5969 }, { "epoch": 0.96, "grad_norm": 9.192418098449707, "learning_rate": 1.5895835586585004e-05, "loss": 1.3712, "step": 5970 }, { "epoch": 0.96, "grad_norm": 10.699989318847656, "learning_rate": 1.589443170931846e-05, "loss": 1.153, "step": 5971 }, { "epoch": 0.96, "grad_norm": 7.549127578735352, "learning_rate": 1.589302765400395e-05, "loss": 1.0064, "step": 5972 }, { "epoch": 0.96, "grad_norm": 9.036659240722656, "learning_rate": 1.5891623420683885e-05, "loss": 1.2757, "step": 5973 }, { "epoch": 0.96, "grad_norm": 10.170267105102539, "learning_rate": 1.5890219009400678e-05, "loss": 1.4196, "step": 5974 }, { "epoch": 0.96, "grad_norm": 9.71047306060791, "learning_rate": 1.5888814420196753e-05, "loss": 1.1457, "step": 5975 }, { "epoch": 0.96, "grad_norm": 10.485345840454102, "learning_rate": 1.5887409653114537e-05, "loss": 1.2402, "step": 5976 }, { "epoch": 0.96, "grad_norm": 8.170108795166016, "learning_rate": 1.5886004708196466e-05, "loss": 1.1806, "step": 5977 }, { "epoch": 0.96, "grad_norm": 8.438278198242188, "learning_rate": 1.5884599585484973e-05, "loss": 1.3564, "step": 5978 }, { "epoch": 0.96, "grad_norm": 8.762463569641113, "learning_rate": 1.58831942850225e-05, "loss": 1.3764, "step": 5979 }, { "epoch": 0.96, "grad_norm": 9.65591812133789, "learning_rate": 1.5881788806851505e-05, "loss": 1.3169, "step": 5980 }, { "epoch": 0.96, "grad_norm": 7.055756092071533, "learning_rate": 1.5880383151014433e-05, "loss": 1.2398, "step": 5981 }, { "epoch": 0.96, "grad_norm": 9.011860847473145, "learning_rate": 1.587897731755375e-05, "loss": 1.2058, "step": 5982 }, { "epoch": 0.96, "grad_norm": 6.520601272583008, "learning_rate": 1.5877571306511912e-05, "loss": 1.2092, "step": 5983 }, { "epoch": 0.96, "grad_norm": 9.501237869262695, "learning_rate": 1.5876165117931398e-05, "loss": 1.3674, "step": 5984 }, { "epoch": 0.96, "grad_norm": 8.508119583129883, "learning_rate": 1.5874758751854678e-05, "loss": 1.2494, "step": 5985 }, { "epoch": 0.96, "grad_norm": 8.546441078186035, "learning_rate": 1.5873352208324235e-05, "loss": 1.4931, "step": 5986 }, { "epoch": 0.96, "grad_norm": 8.661935806274414, "learning_rate": 1.5871945487382554e-05, "loss": 1.3276, "step": 5987 }, { "epoch": 0.96, "grad_norm": 10.216667175292969, "learning_rate": 1.587053858907213e-05, "loss": 1.3049, "step": 5988 }, { "epoch": 0.96, "grad_norm": 6.359475612640381, "learning_rate": 1.5869131513435453e-05, "loss": 1.2443, "step": 5989 }, { "epoch": 0.96, "grad_norm": 7.34089469909668, "learning_rate": 1.5867724260515035e-05, "loss": 1.1394, "step": 5990 }, { "epoch": 0.96, "grad_norm": 8.626664161682129, "learning_rate": 1.586631683035338e-05, "loss": 1.3123, "step": 5991 }, { "epoch": 0.96, "grad_norm": 7.195301055908203, "learning_rate": 1.5864909222992993e-05, "loss": 1.347, "step": 5992 }, { "epoch": 0.96, "grad_norm": 8.634967803955078, "learning_rate": 1.5863501438476405e-05, "loss": 1.3346, "step": 5993 }, { "epoch": 0.96, "grad_norm": 7.553064346313477, "learning_rate": 1.586209347684613e-05, "loss": 1.4615, "step": 5994 }, { "epoch": 0.97, "grad_norm": 8.332743644714355, "learning_rate": 1.58606853381447e-05, "loss": 1.2421, "step": 5995 }, { "epoch": 0.97, "grad_norm": 8.870962142944336, "learning_rate": 1.5859277022414653e-05, "loss": 1.4202, "step": 5996 }, { "epoch": 0.97, "grad_norm": 8.849425315856934, "learning_rate": 1.5857868529698523e-05, "loss": 1.2953, "step": 5997 }, { "epoch": 0.97, "grad_norm": 8.055314064025879, "learning_rate": 1.585645986003886e-05, "loss": 1.4089, "step": 5998 }, { "epoch": 0.97, "grad_norm": 7.658185005187988, "learning_rate": 1.5855051013478215e-05, "loss": 1.22, "step": 5999 }, { "epoch": 0.97, "grad_norm": 7.6992011070251465, "learning_rate": 1.5853641990059135e-05, "loss": 1.2603, "step": 6000 }, { "epoch": 0.97, "grad_norm": 8.442996978759766, "learning_rate": 1.585223278982419e-05, "loss": 1.0911, "step": 6001 }, { "epoch": 0.97, "grad_norm": 7.997025489807129, "learning_rate": 1.5850823412815944e-05, "loss": 1.3607, "step": 6002 }, { "epoch": 0.97, "grad_norm": 7.675576686859131, "learning_rate": 1.5849413859076973e-05, "loss": 1.2263, "step": 6003 }, { "epoch": 0.97, "grad_norm": 7.349827289581299, "learning_rate": 1.5848004128649847e-05, "loss": 1.3153, "step": 6004 }, { "epoch": 0.97, "grad_norm": 8.505477905273438, "learning_rate": 1.584659422157715e-05, "loss": 1.5208, "step": 6005 }, { "epoch": 0.97, "grad_norm": 7.845250606536865, "learning_rate": 1.584518413790147e-05, "loss": 1.1132, "step": 6006 }, { "epoch": 0.97, "grad_norm": 9.58854866027832, "learning_rate": 1.58437738776654e-05, "loss": 1.3151, "step": 6007 }, { "epoch": 0.97, "grad_norm": 8.7525053024292, "learning_rate": 1.5842363440911547e-05, "loss": 1.2825, "step": 6008 }, { "epoch": 0.97, "grad_norm": 7.431450366973877, "learning_rate": 1.58409528276825e-05, "loss": 1.2196, "step": 6009 }, { "epoch": 0.97, "grad_norm": 7.2545905113220215, "learning_rate": 1.5839542038020884e-05, "loss": 1.0826, "step": 6010 }, { "epoch": 0.97, "grad_norm": 8.142962455749512, "learning_rate": 1.5838131071969298e-05, "loss": 1.4485, "step": 6011 }, { "epoch": 0.97, "grad_norm": 8.959760665893555, "learning_rate": 1.5836719929570374e-05, "loss": 1.2706, "step": 6012 }, { "epoch": 0.97, "grad_norm": 7.888503551483154, "learning_rate": 1.583530861086673e-05, "loss": 1.4477, "step": 6013 }, { "epoch": 0.97, "grad_norm": 8.335185050964355, "learning_rate": 1.5833897115901e-05, "loss": 1.1634, "step": 6014 }, { "epoch": 0.97, "grad_norm": 9.680267333984375, "learning_rate": 1.5832485444715818e-05, "loss": 1.3716, "step": 6015 }, { "epoch": 0.97, "grad_norm": 7.443350315093994, "learning_rate": 1.5831073597353826e-05, "loss": 1.4152, "step": 6016 }, { "epoch": 0.97, "grad_norm": 8.72097110748291, "learning_rate": 1.582966157385767e-05, "loss": 1.4836, "step": 6017 }, { "epoch": 0.97, "grad_norm": 7.317540645599365, "learning_rate": 1.582824937427e-05, "loss": 1.0861, "step": 6018 }, { "epoch": 0.97, "grad_norm": 9.086274147033691, "learning_rate": 1.582683699863348e-05, "loss": 1.4951, "step": 6019 }, { "epoch": 0.97, "grad_norm": 7.222758769989014, "learning_rate": 1.5825424446990764e-05, "loss": 1.1911, "step": 6020 }, { "epoch": 0.97, "grad_norm": 7.4760050773620605, "learning_rate": 1.5824011719384523e-05, "loss": 1.4763, "step": 6021 }, { "epoch": 0.97, "grad_norm": 7.852446556091309, "learning_rate": 1.5822598815857433e-05, "loss": 1.2389, "step": 6022 }, { "epoch": 0.97, "grad_norm": 8.689176559448242, "learning_rate": 1.5821185736452165e-05, "loss": 1.225, "step": 6023 }, { "epoch": 0.97, "grad_norm": 7.81779146194458, "learning_rate": 1.5819772481211412e-05, "loss": 1.2583, "step": 6024 }, { "epoch": 0.97, "grad_norm": 8.307738304138184, "learning_rate": 1.5818359050177854e-05, "loss": 1.1165, "step": 6025 }, { "epoch": 0.97, "grad_norm": 8.9359712600708, "learning_rate": 1.5816945443394193e-05, "loss": 1.6626, "step": 6026 }, { "epoch": 0.97, "grad_norm": 9.727700233459473, "learning_rate": 1.581553166090312e-05, "loss": 1.6843, "step": 6027 }, { "epoch": 0.97, "grad_norm": 8.805941581726074, "learning_rate": 1.581411770274735e-05, "loss": 1.4079, "step": 6028 }, { "epoch": 0.97, "grad_norm": 8.784010887145996, "learning_rate": 1.5812703568969588e-05, "loss": 1.1859, "step": 6029 }, { "epoch": 0.97, "grad_norm": 7.546133518218994, "learning_rate": 1.5811289259612544e-05, "loss": 1.1031, "step": 6030 }, { "epoch": 0.97, "grad_norm": 8.715738296508789, "learning_rate": 1.580987477471895e-05, "loss": 1.3647, "step": 6031 }, { "epoch": 0.97, "grad_norm": 7.693239688873291, "learning_rate": 1.5808460114331525e-05, "loss": 1.4717, "step": 6032 }, { "epoch": 0.97, "grad_norm": 9.503229141235352, "learning_rate": 1.5807045278493004e-05, "loss": 1.3804, "step": 6033 }, { "epoch": 0.97, "grad_norm": 7.176292896270752, "learning_rate": 1.580563026724612e-05, "loss": 1.2036, "step": 6034 }, { "epoch": 0.97, "grad_norm": 8.233052253723145, "learning_rate": 1.5804215080633614e-05, "loss": 1.428, "step": 6035 }, { "epoch": 0.97, "grad_norm": 7.950324058532715, "learning_rate": 1.580279971869824e-05, "loss": 1.3382, "step": 6036 }, { "epoch": 0.97, "grad_norm": 8.752302169799805, "learning_rate": 1.5801384181482745e-05, "loss": 1.2408, "step": 6037 }, { "epoch": 0.97, "grad_norm": 8.010551452636719, "learning_rate": 1.579996846902989e-05, "loss": 1.2874, "step": 6038 }, { "epoch": 0.97, "grad_norm": 9.052477836608887, "learning_rate": 1.5798552581382432e-05, "loss": 1.154, "step": 6039 }, { "epoch": 0.97, "grad_norm": 8.50432300567627, "learning_rate": 1.579713651858315e-05, "loss": 1.3344, "step": 6040 }, { "epoch": 0.97, "grad_norm": 8.194808006286621, "learning_rate": 1.579572028067481e-05, "loss": 1.3732, "step": 6041 }, { "epoch": 0.97, "grad_norm": 8.735888481140137, "learning_rate": 1.5794303867700192e-05, "loss": 1.2742, "step": 6042 }, { "epoch": 0.97, "grad_norm": 7.5740580558776855, "learning_rate": 1.5792887279702084e-05, "loss": 1.1841, "step": 6043 }, { "epoch": 0.97, "grad_norm": 7.950502395629883, "learning_rate": 1.579147051672327e-05, "loss": 1.3504, "step": 6044 }, { "epoch": 0.97, "grad_norm": 7.800952434539795, "learning_rate": 1.579005357880655e-05, "loss": 1.1933, "step": 6045 }, { "epoch": 0.97, "grad_norm": 8.666797637939453, "learning_rate": 1.578863646599472e-05, "loss": 1.2048, "step": 6046 }, { "epoch": 0.97, "grad_norm": 7.746561050415039, "learning_rate": 1.578721917833059e-05, "loss": 1.2535, "step": 6047 }, { "epoch": 0.97, "grad_norm": 8.73088550567627, "learning_rate": 1.5785801715856966e-05, "loss": 1.2844, "step": 6048 }, { "epoch": 0.97, "grad_norm": 8.955778121948242, "learning_rate": 1.578438407861667e-05, "loss": 1.6026, "step": 6049 }, { "epoch": 0.97, "grad_norm": 7.858487129211426, "learning_rate": 1.5782966266652516e-05, "loss": 1.2795, "step": 6050 }, { "epoch": 0.97, "grad_norm": 9.287834167480469, "learning_rate": 1.5781548280007336e-05, "loss": 1.5551, "step": 6051 }, { "epoch": 0.97, "grad_norm": 9.405862808227539, "learning_rate": 1.578013011872396e-05, "loss": 1.6018, "step": 6052 }, { "epoch": 0.97, "grad_norm": 8.995418548583984, "learning_rate": 1.577871178284523e-05, "loss": 1.3945, "step": 6053 }, { "epoch": 0.97, "grad_norm": 8.74168586730957, "learning_rate": 1.5777293272413978e-05, "loss": 1.3204, "step": 6054 }, { "epoch": 0.97, "grad_norm": 10.617864608764648, "learning_rate": 1.5775874587473058e-05, "loss": 1.0543, "step": 6055 }, { "epoch": 0.97, "grad_norm": 8.449090957641602, "learning_rate": 1.5774455728065326e-05, "loss": 1.729, "step": 6056 }, { "epoch": 0.98, "grad_norm": 8.836997032165527, "learning_rate": 1.5773036694233633e-05, "loss": 1.5307, "step": 6057 }, { "epoch": 0.98, "grad_norm": 6.999104976654053, "learning_rate": 1.577161748602085e-05, "loss": 1.2353, "step": 6058 }, { "epoch": 0.98, "grad_norm": 10.584830284118652, "learning_rate": 1.577019810346984e-05, "loss": 1.0229, "step": 6059 }, { "epoch": 0.98, "grad_norm": 9.142830848693848, "learning_rate": 1.5768778546623484e-05, "loss": 1.2525, "step": 6060 }, { "epoch": 0.98, "grad_norm": 7.7660627365112305, "learning_rate": 1.576735881552465e-05, "loss": 1.3789, "step": 6061 }, { "epoch": 0.98, "grad_norm": 8.464756965637207, "learning_rate": 1.5765938910216233e-05, "loss": 1.162, "step": 6062 }, { "epoch": 0.98, "grad_norm": 8.751483917236328, "learning_rate": 1.576451883074112e-05, "loss": 1.4048, "step": 6063 }, { "epoch": 0.98, "grad_norm": 7.146045207977295, "learning_rate": 1.57630985771422e-05, "loss": 1.2807, "step": 6064 }, { "epoch": 0.98, "grad_norm": 10.32752799987793, "learning_rate": 1.576167814946238e-05, "loss": 1.3861, "step": 6065 }, { "epoch": 0.98, "grad_norm": 8.305596351623535, "learning_rate": 1.5760257547744566e-05, "loss": 1.1634, "step": 6066 }, { "epoch": 0.98, "grad_norm": 9.991872787475586, "learning_rate": 1.5758836772031667e-05, "loss": 1.5674, "step": 6067 }, { "epoch": 0.98, "grad_norm": 6.892634868621826, "learning_rate": 1.5757415822366596e-05, "loss": 1.2616, "step": 6068 }, { "epoch": 0.98, "grad_norm": 8.79352855682373, "learning_rate": 1.575599469879228e-05, "loss": 1.1658, "step": 6069 }, { "epoch": 0.98, "grad_norm": 8.197813034057617, "learning_rate": 1.5754573401351643e-05, "loss": 1.0319, "step": 6070 }, { "epoch": 0.98, "grad_norm": 7.885430812835693, "learning_rate": 1.5753151930087617e-05, "loss": 1.2936, "step": 6071 }, { "epoch": 0.98, "grad_norm": 7.264811992645264, "learning_rate": 1.575173028504314e-05, "loss": 1.1135, "step": 6072 }, { "epoch": 0.98, "grad_norm": 8.961605072021484, "learning_rate": 1.5750308466261154e-05, "loss": 1.4082, "step": 6073 }, { "epoch": 0.98, "grad_norm": 8.4212064743042, "learning_rate": 1.5748886473784605e-05, "loss": 1.2757, "step": 6074 }, { "epoch": 0.98, "grad_norm": 7.151626110076904, "learning_rate": 1.5747464307656447e-05, "loss": 1.2083, "step": 6075 }, { "epoch": 0.98, "grad_norm": 10.430174827575684, "learning_rate": 1.574604196791964e-05, "loss": 1.4037, "step": 6076 }, { "epoch": 0.98, "grad_norm": 9.585426330566406, "learning_rate": 1.5744619454617142e-05, "loss": 1.3143, "step": 6077 }, { "epoch": 0.98, "grad_norm": 7.86748743057251, "learning_rate": 1.574319676779193e-05, "loss": 1.0705, "step": 6078 }, { "epoch": 0.98, "grad_norm": 8.353686332702637, "learning_rate": 1.5741773907486968e-05, "loss": 1.189, "step": 6079 }, { "epoch": 0.98, "grad_norm": 8.487744331359863, "learning_rate": 1.5740350873745242e-05, "loss": 1.4787, "step": 6080 }, { "epoch": 0.98, "grad_norm": 11.349625587463379, "learning_rate": 1.5738927666609735e-05, "loss": 1.3751, "step": 6081 }, { "epoch": 0.98, "grad_norm": 6.762343406677246, "learning_rate": 1.5737504286123437e-05, "loss": 1.1968, "step": 6082 }, { "epoch": 0.98, "grad_norm": 9.749469757080078, "learning_rate": 1.5736080732329343e-05, "loss": 1.3961, "step": 6083 }, { "epoch": 0.98, "grad_norm": 8.778112411499023, "learning_rate": 1.573465700527045e-05, "loss": 1.4647, "step": 6084 }, { "epoch": 0.98, "grad_norm": 10.743188858032227, "learning_rate": 1.573323310498976e-05, "loss": 1.3304, "step": 6085 }, { "epoch": 0.98, "grad_norm": 9.49962329864502, "learning_rate": 1.57318090315303e-05, "loss": 1.1946, "step": 6086 }, { "epoch": 0.98, "grad_norm": 7.959069728851318, "learning_rate": 1.5730384784935065e-05, "loss": 1.3197, "step": 6087 }, { "epoch": 0.98, "grad_norm": 7.9659881591796875, "learning_rate": 1.5728960365247087e-05, "loss": 1.2015, "step": 6088 }, { "epoch": 0.98, "grad_norm": 7.827226638793945, "learning_rate": 1.5727535772509393e-05, "loss": 1.4437, "step": 6089 }, { "epoch": 0.98, "grad_norm": 9.025078773498535, "learning_rate": 1.5726111006765013e-05, "loss": 1.5435, "step": 6090 }, { "epoch": 0.98, "grad_norm": 8.921255111694336, "learning_rate": 1.5724686068056984e-05, "loss": 1.2386, "step": 6091 }, { "epoch": 0.98, "grad_norm": 9.497029304504395, "learning_rate": 1.5723260956428343e-05, "loss": 1.3599, "step": 6092 }, { "epoch": 0.98, "grad_norm": 7.470432281494141, "learning_rate": 1.572183567192214e-05, "loss": 1.1233, "step": 6093 }, { "epoch": 0.98, "grad_norm": 8.365455627441406, "learning_rate": 1.5720410214581435e-05, "loss": 1.1005, "step": 6094 }, { "epoch": 0.98, "grad_norm": 7.178290843963623, "learning_rate": 1.571898458444927e-05, "loss": 1.0325, "step": 6095 }, { "epoch": 0.98, "grad_norm": 8.287542343139648, "learning_rate": 1.5717558781568724e-05, "loss": 1.2906, "step": 6096 }, { "epoch": 0.98, "grad_norm": 8.332866668701172, "learning_rate": 1.5716132805982852e-05, "loss": 1.2382, "step": 6097 }, { "epoch": 0.98, "grad_norm": 7.201393127441406, "learning_rate": 1.5714706657734735e-05, "loss": 1.4844, "step": 6098 }, { "epoch": 0.98, "grad_norm": 7.49866247177124, "learning_rate": 1.571328033686745e-05, "loss": 1.44, "step": 6099 }, { "epoch": 0.98, "grad_norm": 8.747177124023438, "learning_rate": 1.5711853843424078e-05, "loss": 1.0183, "step": 6100 }, { "epoch": 0.98, "grad_norm": 8.276290893554688, "learning_rate": 1.5710427177447705e-05, "loss": 1.3314, "step": 6101 }, { "epoch": 0.98, "grad_norm": 8.085708618164062, "learning_rate": 1.570900033898144e-05, "loss": 1.2314, "step": 6102 }, { "epoch": 0.98, "grad_norm": 8.29267692565918, "learning_rate": 1.5707573328068364e-05, "loss": 1.4979, "step": 6103 }, { "epoch": 0.98, "grad_norm": 7.833959579467773, "learning_rate": 1.570614614475159e-05, "loss": 1.2772, "step": 6104 }, { "epoch": 0.98, "grad_norm": 8.300676345825195, "learning_rate": 1.5704718789074228e-05, "loss": 1.2971, "step": 6105 }, { "epoch": 0.98, "grad_norm": 9.04997444152832, "learning_rate": 1.5703291261079395e-05, "loss": 1.5197, "step": 6106 }, { "epoch": 0.98, "grad_norm": 8.893218994140625, "learning_rate": 1.5701863560810206e-05, "loss": 1.2002, "step": 6107 }, { "epoch": 0.98, "grad_norm": 8.600454330444336, "learning_rate": 1.5700435688309787e-05, "loss": 1.4607, "step": 6108 }, { "epoch": 0.98, "grad_norm": 8.761054992675781, "learning_rate": 1.5699007643621267e-05, "loss": 1.267, "step": 6109 }, { "epoch": 0.98, "grad_norm": 7.293472766876221, "learning_rate": 1.569757942678779e-05, "loss": 1.0914, "step": 6110 }, { "epoch": 0.98, "grad_norm": 7.230249881744385, "learning_rate": 1.5696151037852486e-05, "loss": 1.1206, "step": 6111 }, { "epoch": 0.98, "grad_norm": 7.938395023345947, "learning_rate": 1.5694722476858513e-05, "loss": 1.4306, "step": 6112 }, { "epoch": 0.98, "grad_norm": 9.97689437866211, "learning_rate": 1.569329374384901e-05, "loss": 1.4702, "step": 6113 }, { "epoch": 0.98, "grad_norm": 9.872200965881348, "learning_rate": 1.5691864838867143e-05, "loss": 1.6216, "step": 6114 }, { "epoch": 0.98, "grad_norm": 7.792153358459473, "learning_rate": 1.5690435761956067e-05, "loss": 1.1887, "step": 6115 }, { "epoch": 0.98, "grad_norm": 7.4501142501831055, "learning_rate": 1.5689006513158954e-05, "loss": 1.1142, "step": 6116 }, { "epoch": 0.98, "grad_norm": 8.904143333435059, "learning_rate": 1.5687577092518975e-05, "loss": 1.2952, "step": 6117 }, { "epoch": 0.98, "grad_norm": 9.157203674316406, "learning_rate": 1.56861475000793e-05, "loss": 1.5029, "step": 6118 }, { "epoch": 0.99, "grad_norm": 8.586822509765625, "learning_rate": 1.5684717735883125e-05, "loss": 1.4329, "step": 6119 }, { "epoch": 0.99, "grad_norm": 9.272732734680176, "learning_rate": 1.568328779997363e-05, "loss": 1.167, "step": 6120 }, { "epoch": 0.99, "grad_norm": 8.923848152160645, "learning_rate": 1.5681857692394003e-05, "loss": 1.3712, "step": 6121 }, { "epoch": 0.99, "grad_norm": 9.1453275680542, "learning_rate": 1.5680427413187453e-05, "loss": 1.5464, "step": 6122 }, { "epoch": 0.99, "grad_norm": 7.510891437530518, "learning_rate": 1.5678996962397174e-05, "loss": 1.5499, "step": 6123 }, { "epoch": 0.99, "grad_norm": 7.6573357582092285, "learning_rate": 1.567756634006638e-05, "loss": 1.3152, "step": 6124 }, { "epoch": 0.99, "grad_norm": 7.690390586853027, "learning_rate": 1.567613554623828e-05, "loss": 1.6404, "step": 6125 }, { "epoch": 0.99, "grad_norm": 7.705830097198486, "learning_rate": 1.5674704580956092e-05, "loss": 1.1949, "step": 6126 }, { "epoch": 0.99, "grad_norm": 7.937863826751709, "learning_rate": 1.567327344426305e-05, "loss": 1.2045, "step": 6127 }, { "epoch": 0.99, "grad_norm": 9.052300453186035, "learning_rate": 1.567184213620237e-05, "loss": 1.2221, "step": 6128 }, { "epoch": 0.99, "grad_norm": 9.319783210754395, "learning_rate": 1.5670410656817295e-05, "loss": 1.3905, "step": 6129 }, { "epoch": 0.99, "grad_norm": 8.647828102111816, "learning_rate": 1.566897900615106e-05, "loss": 1.3514, "step": 6130 }, { "epoch": 0.99, "grad_norm": 7.3512115478515625, "learning_rate": 1.5667547184246916e-05, "loss": 1.2952, "step": 6131 }, { "epoch": 0.99, "grad_norm": 9.173408508300781, "learning_rate": 1.5666115191148102e-05, "loss": 1.4036, "step": 6132 }, { "epoch": 0.99, "grad_norm": 7.106276035308838, "learning_rate": 1.5664683026897883e-05, "loss": 1.3533, "step": 6133 }, { "epoch": 0.99, "grad_norm": 8.506546020507812, "learning_rate": 1.5663250691539514e-05, "loss": 1.2491, "step": 6134 }, { "epoch": 0.99, "grad_norm": 7.751543998718262, "learning_rate": 1.5661818185116265e-05, "loss": 1.4074, "step": 6135 }, { "epoch": 0.99, "grad_norm": 9.106209754943848, "learning_rate": 1.5660385507671398e-05, "loss": 1.5338, "step": 6136 }, { "epoch": 0.99, "grad_norm": 7.743923187255859, "learning_rate": 1.5658952659248195e-05, "loss": 1.3372, "step": 6137 }, { "epoch": 0.99, "grad_norm": 8.235795974731445, "learning_rate": 1.5657519639889938e-05, "loss": 1.1598, "step": 6138 }, { "epoch": 0.99, "grad_norm": 8.221792221069336, "learning_rate": 1.5656086449639907e-05, "loss": 1.4472, "step": 6139 }, { "epoch": 0.99, "grad_norm": 7.861452579498291, "learning_rate": 1.5654653088541397e-05, "loss": 1.3732, "step": 6140 }, { "epoch": 0.99, "grad_norm": 8.13797664642334, "learning_rate": 1.565321955663771e-05, "loss": 1.2701, "step": 6141 }, { "epoch": 0.99, "grad_norm": 7.896544933319092, "learning_rate": 1.5651785853972133e-05, "loss": 1.3372, "step": 6142 }, { "epoch": 0.99, "grad_norm": 8.733686447143555, "learning_rate": 1.5650351980587987e-05, "loss": 1.2799, "step": 6143 }, { "epoch": 0.99, "grad_norm": 8.757316589355469, "learning_rate": 1.5648917936528577e-05, "loss": 1.3336, "step": 6144 }, { "epoch": 0.99, "grad_norm": 8.687844276428223, "learning_rate": 1.5647483721837218e-05, "loss": 1.2168, "step": 6145 }, { "epoch": 0.99, "grad_norm": 8.432954788208008, "learning_rate": 1.5646049336557235e-05, "loss": 1.3106, "step": 6146 }, { "epoch": 0.99, "grad_norm": 8.33088493347168, "learning_rate": 1.5644614780731956e-05, "loss": 1.4304, "step": 6147 }, { "epoch": 0.99, "grad_norm": 6.992467880249023, "learning_rate": 1.564318005440471e-05, "loss": 0.9686, "step": 6148 }, { "epoch": 0.99, "grad_norm": 9.34458065032959, "learning_rate": 1.5641745157618838e-05, "loss": 1.3441, "step": 6149 }, { "epoch": 0.99, "grad_norm": 8.75549602508545, "learning_rate": 1.5640310090417688e-05, "loss": 1.1525, "step": 6150 }, { "epoch": 0.99, "grad_norm": 8.43448257446289, "learning_rate": 1.5638874852844595e-05, "loss": 1.3293, "step": 6151 }, { "epoch": 0.99, "grad_norm": 9.050600051879883, "learning_rate": 1.5637439444942918e-05, "loss": 1.4789, "step": 6152 }, { "epoch": 0.99, "grad_norm": 9.328062057495117, "learning_rate": 1.5636003866756017e-05, "loss": 1.5289, "step": 6153 }, { "epoch": 0.99, "grad_norm": 7.827425956726074, "learning_rate": 1.5634568118327257e-05, "loss": 1.0823, "step": 6154 }, { "epoch": 0.99, "grad_norm": 8.198657035827637, "learning_rate": 1.5633132199699996e-05, "loss": 1.2878, "step": 6155 }, { "epoch": 0.99, "grad_norm": 8.741039276123047, "learning_rate": 1.563169611091762e-05, "loss": 1.3886, "step": 6156 }, { "epoch": 0.99, "grad_norm": 8.570011138916016, "learning_rate": 1.5630259852023502e-05, "loss": 1.3276, "step": 6157 }, { "epoch": 0.99, "grad_norm": 9.51602554321289, "learning_rate": 1.5628823423061022e-05, "loss": 1.1675, "step": 6158 }, { "epoch": 0.99, "grad_norm": 8.38430404663086, "learning_rate": 1.5627386824073578e-05, "loss": 1.1047, "step": 6159 }, { "epoch": 0.99, "grad_norm": 10.046683311462402, "learning_rate": 1.5625950055104557e-05, "loss": 1.2931, "step": 6160 }, { "epoch": 0.99, "grad_norm": 8.18657398223877, "learning_rate": 1.562451311619736e-05, "loss": 1.2874, "step": 6161 }, { "epoch": 0.99, "grad_norm": 7.365925312042236, "learning_rate": 1.562307600739539e-05, "loss": 1.1761, "step": 6162 }, { "epoch": 0.99, "grad_norm": 7.2673115730285645, "learning_rate": 1.5621638728742062e-05, "loss": 1.382, "step": 6163 }, { "epoch": 0.99, "grad_norm": 9.719547271728516, "learning_rate": 1.562020128028079e-05, "loss": 1.3924, "step": 6164 }, { "epoch": 0.99, "grad_norm": 7.124171257019043, "learning_rate": 1.5618763662054984e-05, "loss": 1.1041, "step": 6165 }, { "epoch": 0.99, "grad_norm": 8.269244194030762, "learning_rate": 1.561732587410808e-05, "loss": 1.3028, "step": 6166 }, { "epoch": 0.99, "grad_norm": 7.444097995758057, "learning_rate": 1.56158879164835e-05, "loss": 1.2001, "step": 6167 }, { "epoch": 0.99, "grad_norm": 10.629716873168945, "learning_rate": 1.5614449789224685e-05, "loss": 1.2777, "step": 6168 }, { "epoch": 0.99, "grad_norm": 8.087823867797852, "learning_rate": 1.5613011492375073e-05, "loss": 1.0782, "step": 6169 }, { "epoch": 0.99, "grad_norm": 9.159987449645996, "learning_rate": 1.5611573025978112e-05, "loss": 1.1421, "step": 6170 }, { "epoch": 0.99, "grad_norm": 8.831388473510742, "learning_rate": 1.5610134390077244e-05, "loss": 1.495, "step": 6171 }, { "epoch": 0.99, "grad_norm": 7.492094993591309, "learning_rate": 1.5608695584715935e-05, "loss": 1.2725, "step": 6172 }, { "epoch": 0.99, "grad_norm": 9.858691215515137, "learning_rate": 1.560725660993764e-05, "loss": 1.4707, "step": 6173 }, { "epoch": 0.99, "grad_norm": 8.558564186096191, "learning_rate": 1.5605817465785825e-05, "loss": 1.3941, "step": 6174 }, { "epoch": 0.99, "grad_norm": 8.82918930053711, "learning_rate": 1.560437815230396e-05, "loss": 1.2231, "step": 6175 }, { "epoch": 0.99, "grad_norm": 7.8784308433532715, "learning_rate": 1.560293866953553e-05, "loss": 1.5217, "step": 6176 }, { "epoch": 0.99, "grad_norm": 9.081093788146973, "learning_rate": 1.5601499017524e-05, "loss": 1.2993, "step": 6177 }, { "epoch": 0.99, "grad_norm": 7.946300506591797, "learning_rate": 1.5600059196312877e-05, "loss": 1.3193, "step": 6178 }, { "epoch": 0.99, "grad_norm": 7.358333110809326, "learning_rate": 1.559861920594563e-05, "loss": 1.2477, "step": 6179 }, { "epoch": 0.99, "grad_norm": 7.811799049377441, "learning_rate": 1.5597179046465774e-05, "loss": 1.1689, "step": 6180 }, { "epoch": 1.0, "grad_norm": 8.131181716918945, "learning_rate": 1.55957387179168e-05, "loss": 1.3797, "step": 6181 }, { "epoch": 1.0, "grad_norm": 8.102826118469238, "learning_rate": 1.559429822034222e-05, "loss": 1.3123, "step": 6182 }, { "epoch": 1.0, "grad_norm": 7.052562236785889, "learning_rate": 1.559285755378554e-05, "loss": 1.3685, "step": 6183 }, { "epoch": 1.0, "grad_norm": 9.431912422180176, "learning_rate": 1.5591416718290283e-05, "loss": 1.3254, "step": 6184 }, { "epoch": 1.0, "grad_norm": 8.149179458618164, "learning_rate": 1.558997571389997e-05, "loss": 1.4806, "step": 6185 }, { "epoch": 1.0, "grad_norm": 9.690969467163086, "learning_rate": 1.5588534540658126e-05, "loss": 1.428, "step": 6186 }, { "epoch": 1.0, "grad_norm": 9.38239574432373, "learning_rate": 1.5587093198608283e-05, "loss": 1.329, "step": 6187 }, { "epoch": 1.0, "grad_norm": 9.34152889251709, "learning_rate": 1.5585651687793987e-05, "loss": 1.1584, "step": 6188 }, { "epoch": 1.0, "grad_norm": 8.073591232299805, "learning_rate": 1.5584210008258764e-05, "loss": 0.9112, "step": 6189 }, { "epoch": 1.0, "grad_norm": 8.471647262573242, "learning_rate": 1.5582768160046176e-05, "loss": 1.3161, "step": 6190 }, { "epoch": 1.0, "grad_norm": 9.066520690917969, "learning_rate": 1.558132614319977e-05, "loss": 1.4513, "step": 6191 }, { "epoch": 1.0, "grad_norm": 7.6435370445251465, "learning_rate": 1.55798839577631e-05, "loss": 1.4249, "step": 6192 }, { "epoch": 1.0, "grad_norm": 8.768693923950195, "learning_rate": 1.557844160377974e-05, "loss": 1.0881, "step": 6193 }, { "epoch": 1.0, "grad_norm": 8.62700366973877, "learning_rate": 1.5576999081293246e-05, "loss": 1.4844, "step": 6194 }, { "epoch": 1.0, "grad_norm": 8.246460914611816, "learning_rate": 1.5575556390347195e-05, "loss": 1.4302, "step": 6195 }, { "epoch": 1.0, "grad_norm": 8.618663787841797, "learning_rate": 1.5574113530985167e-05, "loss": 1.1228, "step": 6196 }, { "epoch": 1.0, "grad_norm": 9.01761245727539, "learning_rate": 1.5572670503250747e-05, "loss": 1.6575, "step": 6197 }, { "epoch": 1.0, "grad_norm": 8.168876647949219, "learning_rate": 1.557122730718752e-05, "loss": 1.5732, "step": 6198 }, { "epoch": 1.0, "grad_norm": 6.948616981506348, "learning_rate": 1.5569783942839077e-05, "loss": 1.2443, "step": 6199 }, { "epoch": 1.0, "grad_norm": 8.440849304199219, "learning_rate": 1.556834041024902e-05, "loss": 1.3168, "step": 6200 }, { "epoch": 1.0, "grad_norm": 7.691679954528809, "learning_rate": 1.5566896709460953e-05, "loss": 1.2695, "step": 6201 }, { "epoch": 1.0, "grad_norm": 7.4523844718933105, "learning_rate": 1.5565452840518482e-05, "loss": 1.1165, "step": 6202 }, { "epoch": 1.0, "grad_norm": 7.759124755859375, "learning_rate": 1.5564008803465225e-05, "loss": 1.2179, "step": 6203 }, { "epoch": 1.0, "grad_norm": 10.270981788635254, "learning_rate": 1.5562564598344793e-05, "loss": 1.2483, "step": 6204 }, { "epoch": 1.0, "grad_norm": 7.500125408172607, "learning_rate": 1.556112022520082e-05, "loss": 1.1023, "step": 6205 }, { "epoch": 1.0, "grad_norm": 7.586414337158203, "learning_rate": 1.5559675684076928e-05, "loss": 1.1089, "step": 6206 }, { "epoch": 1.0, "grad_norm": 7.992264747619629, "learning_rate": 1.5558230975016754e-05, "loss": 1.391, "step": 6207 }, { "epoch": 1.0, "grad_norm": 7.945639133453369, "learning_rate": 1.5556786098063933e-05, "loss": 1.1494, "step": 6208 }, { "epoch": 1.0, "grad_norm": 8.93952465057373, "learning_rate": 1.555534105326211e-05, "loss": 1.5675, "step": 6209 }, { "epoch": 1.0, "grad_norm": 7.248538494110107, "learning_rate": 1.555389584065494e-05, "loss": 1.2955, "step": 6210 }, { "epoch": 1.0, "grad_norm": 6.966527462005615, "learning_rate": 1.5552450460286076e-05, "loss": 1.3668, "step": 6211 }, { "epoch": 1.0, "grad_norm": 9.039973258972168, "learning_rate": 1.5551004912199168e-05, "loss": 1.331, "step": 6212 }, { "epoch": 1.0, "grad_norm": 7.329585552215576, "learning_rate": 1.554955919643789e-05, "loss": 1.2213, "step": 6213 }, { "epoch": 1.0, "grad_norm": 8.171490669250488, "learning_rate": 1.5548113313045908e-05, "loss": 1.2775, "step": 6214 }, { "epoch": 1.0, "grad_norm": 7.881073474884033, "learning_rate": 1.5546667262066903e-05, "loss": 1.2556, "step": 6215 }, { "epoch": 1.0, "grad_norm": 6.24595308303833, "learning_rate": 1.554522104354454e-05, "loss": 1.0164, "step": 6216 }, { "epoch": 1.0, "grad_norm": 6.287655830383301, "learning_rate": 1.554377465752252e-05, "loss": 1.0485, "step": 6217 }, { "epoch": 1.0, "grad_norm": 7.464818477630615, "learning_rate": 1.5542328104044518e-05, "loss": 1.2922, "step": 6218 }, { "epoch": 1.0, "grad_norm": 8.184073448181152, "learning_rate": 1.554088138315424e-05, "loss": 0.8805, "step": 6219 }, { "epoch": 1.0, "grad_norm": 8.11931037902832, "learning_rate": 1.553943449489538e-05, "loss": 1.1354, "step": 6220 }, { "epoch": 1.0, "grad_norm": 6.961741924285889, "learning_rate": 1.553798743931165e-05, "loss": 1.1307, "step": 6221 }, { "epoch": 1.0, "grad_norm": 6.537869453430176, "learning_rate": 1.553654021644675e-05, "loss": 0.8585, "step": 6222 }, { "epoch": 1.0, "grad_norm": 7.6847944259643555, "learning_rate": 1.55350928263444e-05, "loss": 1.0307, "step": 6223 }, { "epoch": 1.0, "grad_norm": 8.229745864868164, "learning_rate": 1.553364526904832e-05, "loss": 1.1379, "step": 6224 }, { "epoch": 1.0, "grad_norm": 7.378057479858398, "learning_rate": 1.553219754460223e-05, "loss": 1.1104, "step": 6225 }, { "epoch": 1.0, "grad_norm": 8.183778762817383, "learning_rate": 1.5530749653049874e-05, "loss": 1.0591, "step": 6226 }, { "epoch": 1.0, "grad_norm": 7.211236000061035, "learning_rate": 1.5529301594434973e-05, "loss": 0.8448, "step": 6227 }, { "epoch": 1.0, "grad_norm": 6.853067874908447, "learning_rate": 1.5527853368801275e-05, "loss": 1.0292, "step": 6228 }, { "epoch": 1.0, "grad_norm": 6.98396110534668, "learning_rate": 1.552640497619252e-05, "loss": 1.1931, "step": 6229 }, { "epoch": 1.0, "grad_norm": 7.579531192779541, "learning_rate": 1.552495641665246e-05, "loss": 0.993, "step": 6230 }, { "epoch": 1.0, "grad_norm": 7.892803192138672, "learning_rate": 1.5523507690224852e-05, "loss": 1.254, "step": 6231 }, { "epoch": 1.0, "grad_norm": 8.476266860961914, "learning_rate": 1.552205879695346e-05, "loss": 1.1616, "step": 6232 }, { "epoch": 1.0, "grad_norm": 7.408792495727539, "learning_rate": 1.5520609736882038e-05, "loss": 1.1727, "step": 6233 }, { "epoch": 1.0, "grad_norm": 8.642122268676758, "learning_rate": 1.551916051005437e-05, "loss": 1.1739, "step": 6234 }, { "epoch": 1.0, "grad_norm": 8.59482479095459, "learning_rate": 1.5517711116514226e-05, "loss": 1.1006, "step": 6235 }, { "epoch": 1.0, "grad_norm": 8.330899238586426, "learning_rate": 1.551626155630538e-05, "loss": 1.2225, "step": 6236 }, { "epoch": 1.0, "grad_norm": 7.588085174560547, "learning_rate": 1.551481182947163e-05, "loss": 1.1515, "step": 6237 }, { "epoch": 1.0, "grad_norm": 8.806741714477539, "learning_rate": 1.5513361936056753e-05, "loss": 1.0556, "step": 6238 }, { "epoch": 1.0, "grad_norm": 8.033977508544922, "learning_rate": 1.5511911876104558e-05, "loss": 0.9884, "step": 6239 }, { "epoch": 1.0, "grad_norm": 9.624595642089844, "learning_rate": 1.5510461649658834e-05, "loss": 1.1687, "step": 6240 }, { "epoch": 1.0, "grad_norm": 8.971270561218262, "learning_rate": 1.5509011256763395e-05, "loss": 1.1357, "step": 6241 }, { "epoch": 1.0, "grad_norm": 8.056976318359375, "learning_rate": 1.550756069746205e-05, "loss": 1.1038, "step": 6242 }, { "epoch": 1.0, "grad_norm": 9.034976959228516, "learning_rate": 1.5506109971798612e-05, "loss": 0.984, "step": 6243 }, { "epoch": 1.01, "grad_norm": 7.919661998748779, "learning_rate": 1.5504659079816903e-05, "loss": 1.0476, "step": 6244 }, { "epoch": 1.01, "grad_norm": 7.733551025390625, "learning_rate": 1.5503208021560752e-05, "loss": 0.9025, "step": 6245 }, { "epoch": 1.01, "grad_norm": 7.35228157043457, "learning_rate": 1.5501756797073986e-05, "loss": 1.0089, "step": 6246 }, { "epoch": 1.01, "grad_norm": 7.42291259765625, "learning_rate": 1.5500305406400443e-05, "loss": 0.9745, "step": 6247 }, { "epoch": 1.01, "grad_norm": 8.643035888671875, "learning_rate": 1.5498853849583963e-05, "loss": 1.0866, "step": 6248 }, { "epoch": 1.01, "grad_norm": 8.416054725646973, "learning_rate": 1.5497402126668387e-05, "loss": 1.0567, "step": 6249 }, { "epoch": 1.01, "grad_norm": 9.187230110168457, "learning_rate": 1.5495950237697576e-05, "loss": 1.2411, "step": 6250 }, { "epoch": 1.01, "grad_norm": 9.186338424682617, "learning_rate": 1.5494498182715378e-05, "loss": 1.1051, "step": 6251 }, { "epoch": 1.01, "grad_norm": 9.436699867248535, "learning_rate": 1.549304596176566e-05, "loss": 1.2847, "step": 6252 }, { "epoch": 1.01, "grad_norm": 8.088770866394043, "learning_rate": 1.5491593574892283e-05, "loss": 0.9243, "step": 6253 }, { "epoch": 1.01, "grad_norm": 8.510902404785156, "learning_rate": 1.549014102213912e-05, "loss": 1.201, "step": 6254 }, { "epoch": 1.01, "grad_norm": 9.532636642456055, "learning_rate": 1.5488688303550045e-05, "loss": 1.0795, "step": 6255 }, { "epoch": 1.01, "grad_norm": 8.254289627075195, "learning_rate": 1.5487235419168944e-05, "loss": 1.0093, "step": 6256 }, { "epoch": 1.01, "grad_norm": 7.559506416320801, "learning_rate": 1.5485782369039696e-05, "loss": 1.1141, "step": 6257 }, { "epoch": 1.01, "grad_norm": 8.43625259399414, "learning_rate": 1.54843291532062e-05, "loss": 1.0889, "step": 6258 }, { "epoch": 1.01, "grad_norm": 9.911977767944336, "learning_rate": 1.5482875771712344e-05, "loss": 1.3179, "step": 6259 }, { "epoch": 1.01, "grad_norm": 7.261898040771484, "learning_rate": 1.5481422224602032e-05, "loss": 0.9252, "step": 6260 }, { "epoch": 1.01, "grad_norm": 9.782508850097656, "learning_rate": 1.547996851191917e-05, "loss": 1.2504, "step": 6261 }, { "epoch": 1.01, "grad_norm": 7.507898807525635, "learning_rate": 1.5478514633707675e-05, "loss": 1.0182, "step": 6262 }, { "epoch": 1.01, "grad_norm": 8.468903541564941, "learning_rate": 1.5477060590011457e-05, "loss": 0.9147, "step": 6263 }, { "epoch": 1.01, "grad_norm": 7.78025484085083, "learning_rate": 1.5475606380874437e-05, "loss": 0.9998, "step": 6264 }, { "epoch": 1.01, "grad_norm": 7.293120861053467, "learning_rate": 1.547415200634054e-05, "loss": 0.8365, "step": 6265 }, { "epoch": 1.01, "grad_norm": 8.960081100463867, "learning_rate": 1.5472697466453702e-05, "loss": 1.198, "step": 6266 }, { "epoch": 1.01, "grad_norm": 7.557275295257568, "learning_rate": 1.547124276125785e-05, "loss": 1.0565, "step": 6267 }, { "epoch": 1.01, "grad_norm": 9.275749206542969, "learning_rate": 1.5469787890796937e-05, "loss": 1.0693, "step": 6268 }, { "epoch": 1.01, "grad_norm": 8.198647499084473, "learning_rate": 1.5468332855114902e-05, "loss": 1.2074, "step": 6269 }, { "epoch": 1.01, "grad_norm": 7.012026786804199, "learning_rate": 1.5466877654255695e-05, "loss": 1.2558, "step": 6270 }, { "epoch": 1.01, "grad_norm": 11.397287368774414, "learning_rate": 1.5465422288263275e-05, "loss": 1.2583, "step": 6271 }, { "epoch": 1.01, "grad_norm": 11.030876159667969, "learning_rate": 1.5463966757181603e-05, "loss": 1.126, "step": 6272 }, { "epoch": 1.01, "grad_norm": 9.322731018066406, "learning_rate": 1.5462511061054643e-05, "loss": 1.097, "step": 6273 }, { "epoch": 1.01, "grad_norm": 9.237082481384277, "learning_rate": 1.546105519992637e-05, "loss": 1.2398, "step": 6274 }, { "epoch": 1.01, "grad_norm": 9.791245460510254, "learning_rate": 1.5459599173840754e-05, "loss": 1.3203, "step": 6275 }, { "epoch": 1.01, "grad_norm": 9.962615966796875, "learning_rate": 1.5458142982841777e-05, "loss": 1.2877, "step": 6276 }, { "epoch": 1.01, "grad_norm": 9.37236213684082, "learning_rate": 1.545668662697343e-05, "loss": 1.1021, "step": 6277 }, { "epoch": 1.01, "grad_norm": 7.16204309463501, "learning_rate": 1.54552301062797e-05, "loss": 0.8392, "step": 6278 }, { "epoch": 1.01, "grad_norm": 8.337480545043945, "learning_rate": 1.545377342080458e-05, "loss": 1.2005, "step": 6279 }, { "epoch": 1.01, "grad_norm": 9.619913101196289, "learning_rate": 1.5452316570592074e-05, "loss": 0.9916, "step": 6280 }, { "epoch": 1.01, "grad_norm": 9.478014945983887, "learning_rate": 1.5450859555686193e-05, "loss": 1.1421, "step": 6281 }, { "epoch": 1.01, "grad_norm": 9.088417053222656, "learning_rate": 1.5449402376130937e-05, "loss": 1.2911, "step": 6282 }, { "epoch": 1.01, "grad_norm": 8.266681671142578, "learning_rate": 1.544794503197033e-05, "loss": 1.2709, "step": 6283 }, { "epoch": 1.01, "grad_norm": 9.61153793334961, "learning_rate": 1.5446487523248392e-05, "loss": 1.0172, "step": 6284 }, { "epoch": 1.01, "grad_norm": 7.422783374786377, "learning_rate": 1.5445029850009148e-05, "loss": 1.1314, "step": 6285 }, { "epoch": 1.01, "grad_norm": 8.913312911987305, "learning_rate": 1.5443572012296624e-05, "loss": 1.0094, "step": 6286 }, { "epoch": 1.01, "grad_norm": 8.812931060791016, "learning_rate": 1.5442114010154862e-05, "loss": 1.0007, "step": 6287 }, { "epoch": 1.01, "grad_norm": 8.751644134521484, "learning_rate": 1.5440655843627895e-05, "loss": 1.1163, "step": 6288 }, { "epoch": 1.01, "grad_norm": 9.003067016601562, "learning_rate": 1.5439197512759778e-05, "loss": 1.0694, "step": 6289 }, { "epoch": 1.01, "grad_norm": 7.405996799468994, "learning_rate": 1.5437739017594554e-05, "loss": 0.9285, "step": 6290 }, { "epoch": 1.01, "grad_norm": 8.185877799987793, "learning_rate": 1.5436280358176286e-05, "loss": 1.1392, "step": 6291 }, { "epoch": 1.01, "grad_norm": 9.331515312194824, "learning_rate": 1.5434821534549026e-05, "loss": 1.1668, "step": 6292 }, { "epoch": 1.01, "grad_norm": 9.265183448791504, "learning_rate": 1.5433362546756843e-05, "loss": 1.1161, "step": 6293 }, { "epoch": 1.01, "grad_norm": 10.524515151977539, "learning_rate": 1.543190339484381e-05, "loss": 1.328, "step": 6294 }, { "epoch": 1.01, "grad_norm": 8.724212646484375, "learning_rate": 1.5430444078853998e-05, "loss": 1.2152, "step": 6295 }, { "epoch": 1.01, "grad_norm": 7.47344446182251, "learning_rate": 1.542898459883149e-05, "loss": 1.075, "step": 6296 }, { "epoch": 1.01, "grad_norm": 8.739662170410156, "learning_rate": 1.5427524954820373e-05, "loss": 1.1866, "step": 6297 }, { "epoch": 1.01, "grad_norm": 8.153454780578613, "learning_rate": 1.5426065146864734e-05, "loss": 1.0555, "step": 6298 }, { "epoch": 1.01, "grad_norm": 10.373564720153809, "learning_rate": 1.5424605175008664e-05, "loss": 1.1541, "step": 6299 }, { "epoch": 1.01, "grad_norm": 8.703590393066406, "learning_rate": 1.5423145039296272e-05, "loss": 1.1761, "step": 6300 }, { "epoch": 1.01, "grad_norm": 6.991188049316406, "learning_rate": 1.542168473977166e-05, "loss": 0.9404, "step": 6301 }, { "epoch": 1.01, "grad_norm": 8.517215728759766, "learning_rate": 1.542022427647893e-05, "loss": 0.9659, "step": 6302 }, { "epoch": 1.01, "grad_norm": 7.62423849105835, "learning_rate": 1.5418763649462212e-05, "loss": 0.9504, "step": 6303 }, { "epoch": 1.01, "grad_norm": 10.426161766052246, "learning_rate": 1.5417302858765615e-05, "loss": 1.3207, "step": 6304 }, { "epoch": 1.01, "grad_norm": 8.763517379760742, "learning_rate": 1.5415841904433268e-05, "loss": 1.3562, "step": 6305 }, { "epoch": 1.02, "grad_norm": 8.052791595458984, "learning_rate": 1.5414380786509297e-05, "loss": 1.1106, "step": 6306 }, { "epoch": 1.02, "grad_norm": 8.734039306640625, "learning_rate": 1.5412919505037843e-05, "loss": 1.0948, "step": 6307 }, { "epoch": 1.02, "grad_norm": 7.553258419036865, "learning_rate": 1.5411458060063037e-05, "loss": 0.8692, "step": 6308 }, { "epoch": 1.02, "grad_norm": 8.544700622558594, "learning_rate": 1.5409996451629032e-05, "loss": 1.0129, "step": 6309 }, { "epoch": 1.02, "grad_norm": 8.101072311401367, "learning_rate": 1.5408534679779973e-05, "loss": 1.0595, "step": 6310 }, { "epoch": 1.02, "grad_norm": 6.826074123382568, "learning_rate": 1.5407072744560013e-05, "loss": 1.0057, "step": 6311 }, { "epoch": 1.02, "grad_norm": 8.463802337646484, "learning_rate": 1.5405610646013317e-05, "loss": 0.901, "step": 6312 }, { "epoch": 1.02, "grad_norm": 7.788783073425293, "learning_rate": 1.5404148384184043e-05, "loss": 0.8571, "step": 6313 }, { "epoch": 1.02, "grad_norm": 11.156245231628418, "learning_rate": 1.5402685959116368e-05, "loss": 1.4613, "step": 6314 }, { "epoch": 1.02, "grad_norm": 7.8648762702941895, "learning_rate": 1.540122337085446e-05, "loss": 0.821, "step": 6315 }, { "epoch": 1.02, "grad_norm": 7.964131832122803, "learning_rate": 1.5399760619442498e-05, "loss": 0.8616, "step": 6316 }, { "epoch": 1.02, "grad_norm": 8.46975326538086, "learning_rate": 1.539829770492467e-05, "loss": 1.249, "step": 6317 }, { "epoch": 1.02, "grad_norm": 8.888932228088379, "learning_rate": 1.5396834627345162e-05, "loss": 1.2336, "step": 6318 }, { "epoch": 1.02, "grad_norm": 9.15582275390625, "learning_rate": 1.5395371386748168e-05, "loss": 1.2877, "step": 6319 }, { "epoch": 1.02, "grad_norm": 10.317804336547852, "learning_rate": 1.5393907983177885e-05, "loss": 1.321, "step": 6320 }, { "epoch": 1.02, "grad_norm": 9.649770736694336, "learning_rate": 1.539244441667852e-05, "loss": 1.1234, "step": 6321 }, { "epoch": 1.02, "grad_norm": 8.967284202575684, "learning_rate": 1.5390980687294284e-05, "loss": 1.164, "step": 6322 }, { "epoch": 1.02, "grad_norm": 9.172301292419434, "learning_rate": 1.538951679506939e-05, "loss": 1.2021, "step": 6323 }, { "epoch": 1.02, "grad_norm": 8.2002534866333, "learning_rate": 1.5388052740048052e-05, "loss": 1.0055, "step": 6324 }, { "epoch": 1.02, "grad_norm": 8.780803680419922, "learning_rate": 1.538658852227449e-05, "loss": 1.2455, "step": 6325 }, { "epoch": 1.02, "grad_norm": 7.964524745941162, "learning_rate": 1.5385124141792948e-05, "loss": 1.1339, "step": 6326 }, { "epoch": 1.02, "grad_norm": 8.411117553710938, "learning_rate": 1.538365959864764e-05, "loss": 1.1706, "step": 6327 }, { "epoch": 1.02, "grad_norm": 9.04819393157959, "learning_rate": 1.5382194892882817e-05, "loss": 1.4317, "step": 6328 }, { "epoch": 1.02, "grad_norm": 8.26449203491211, "learning_rate": 1.5380730024542717e-05, "loss": 1.1625, "step": 6329 }, { "epoch": 1.02, "grad_norm": 10.151373863220215, "learning_rate": 1.5379264993671594e-05, "loss": 1.2551, "step": 6330 }, { "epoch": 1.02, "grad_norm": 10.013193130493164, "learning_rate": 1.537779980031369e-05, "loss": 1.4207, "step": 6331 }, { "epoch": 1.02, "grad_norm": 6.6665358543396, "learning_rate": 1.5376334444513275e-05, "loss": 1.0181, "step": 6332 }, { "epoch": 1.02, "grad_norm": 7.491965293884277, "learning_rate": 1.5374868926314604e-05, "loss": 1.0262, "step": 6333 }, { "epoch": 1.02, "grad_norm": 7.970547199249268, "learning_rate": 1.5373403245761943e-05, "loss": 1.125, "step": 6334 }, { "epoch": 1.02, "grad_norm": 8.774348258972168, "learning_rate": 1.5371937402899572e-05, "loss": 1.1302, "step": 6335 }, { "epoch": 1.02, "grad_norm": 9.861520767211914, "learning_rate": 1.5370471397771763e-05, "loss": 1.0714, "step": 6336 }, { "epoch": 1.02, "grad_norm": 7.914842128753662, "learning_rate": 1.5369005230422803e-05, "loss": 0.8281, "step": 6337 }, { "epoch": 1.02, "grad_norm": 7.73818302154541, "learning_rate": 1.536753890089697e-05, "loss": 1.2844, "step": 6338 }, { "epoch": 1.02, "grad_norm": 7.9127044677734375, "learning_rate": 1.536607240923857e-05, "loss": 0.986, "step": 6339 }, { "epoch": 1.02, "grad_norm": 8.102127075195312, "learning_rate": 1.5364605755491886e-05, "loss": 1.0494, "step": 6340 }, { "epoch": 1.02, "grad_norm": 7.7245683670043945, "learning_rate": 1.5363138939701227e-05, "loss": 1.159, "step": 6341 }, { "epoch": 1.02, "grad_norm": 9.172994613647461, "learning_rate": 1.53616719619109e-05, "loss": 1.0838, "step": 6342 }, { "epoch": 1.02, "grad_norm": 8.477018356323242, "learning_rate": 1.5360204822165215e-05, "loss": 1.1858, "step": 6343 }, { "epoch": 1.02, "grad_norm": 9.303365707397461, "learning_rate": 1.5358737520508492e-05, "loss": 1.3797, "step": 6344 }, { "epoch": 1.02, "grad_norm": 7.408660888671875, "learning_rate": 1.535727005698505e-05, "loss": 0.9761, "step": 6345 }, { "epoch": 1.02, "grad_norm": 7.455320358276367, "learning_rate": 1.5355802431639213e-05, "loss": 0.9969, "step": 6346 }, { "epoch": 1.02, "grad_norm": 11.457661628723145, "learning_rate": 1.5354334644515314e-05, "loss": 0.9135, "step": 6347 }, { "epoch": 1.02, "grad_norm": 7.425030708312988, "learning_rate": 1.535286669565769e-05, "loss": 0.8545, "step": 6348 }, { "epoch": 1.02, "grad_norm": 7.625483989715576, "learning_rate": 1.5351398585110686e-05, "loss": 0.9172, "step": 6349 }, { "epoch": 1.02, "grad_norm": 8.451168060302734, "learning_rate": 1.534993031291864e-05, "loss": 0.9875, "step": 6350 }, { "epoch": 1.02, "grad_norm": 6.897863388061523, "learning_rate": 1.5348461879125905e-05, "loss": 0.9324, "step": 6351 }, { "epoch": 1.02, "grad_norm": 10.642303466796875, "learning_rate": 1.5346993283776835e-05, "loss": 1.1313, "step": 6352 }, { "epoch": 1.02, "grad_norm": 8.496404647827148, "learning_rate": 1.53455245269158e-05, "loss": 0.9712, "step": 6353 }, { "epoch": 1.02, "grad_norm": 8.982364654541016, "learning_rate": 1.534405560858716e-05, "loss": 0.9509, "step": 6354 }, { "epoch": 1.02, "grad_norm": 7.463940143585205, "learning_rate": 1.5342586528835283e-05, "loss": 0.914, "step": 6355 }, { "epoch": 1.02, "grad_norm": 7.333802700042725, "learning_rate": 1.5341117287704547e-05, "loss": 0.8211, "step": 6356 }, { "epoch": 1.02, "grad_norm": 9.375747680664062, "learning_rate": 1.5339647885239327e-05, "loss": 1.1593, "step": 6357 }, { "epoch": 1.02, "grad_norm": 8.649879455566406, "learning_rate": 1.5338178321484015e-05, "loss": 1.111, "step": 6358 }, { "epoch": 1.02, "grad_norm": 11.429696083068848, "learning_rate": 1.5336708596482997e-05, "loss": 1.2182, "step": 6359 }, { "epoch": 1.02, "grad_norm": 11.631509780883789, "learning_rate": 1.533523871028067e-05, "loss": 1.0566, "step": 6360 }, { "epoch": 1.02, "grad_norm": 9.099239349365234, "learning_rate": 1.5333768662921427e-05, "loss": 1.2574, "step": 6361 }, { "epoch": 1.02, "grad_norm": 8.735146522521973, "learning_rate": 1.533229845444968e-05, "loss": 0.9727, "step": 6362 }, { "epoch": 1.02, "grad_norm": 7.871514797210693, "learning_rate": 1.5330828084909838e-05, "loss": 0.9787, "step": 6363 }, { "epoch": 1.02, "grad_norm": 8.389548301696777, "learning_rate": 1.532935755434631e-05, "loss": 0.7647, "step": 6364 }, { "epoch": 1.02, "grad_norm": 9.757450103759766, "learning_rate": 1.532788686280352e-05, "loss": 1.1342, "step": 6365 }, { "epoch": 1.02, "grad_norm": 10.301313400268555, "learning_rate": 1.532641601032589e-05, "loss": 1.0338, "step": 6366 }, { "epoch": 1.02, "grad_norm": 9.225231170654297, "learning_rate": 1.5324944996957847e-05, "loss": 1.2084, "step": 6367 }, { "epoch": 1.03, "grad_norm": 7.630700588226318, "learning_rate": 1.5323473822743826e-05, "loss": 1.0641, "step": 6368 }, { "epoch": 1.03, "grad_norm": 7.963260650634766, "learning_rate": 1.5322002487728268e-05, "loss": 1.0102, "step": 6369 }, { "epoch": 1.03, "grad_norm": 8.180593490600586, "learning_rate": 1.532053099195561e-05, "loss": 0.9633, "step": 6370 }, { "epoch": 1.03, "grad_norm": 10.331149101257324, "learning_rate": 1.5319059335470307e-05, "loss": 0.929, "step": 6371 }, { "epoch": 1.03, "grad_norm": 8.16218090057373, "learning_rate": 1.531758751831681e-05, "loss": 1.1102, "step": 6372 }, { "epoch": 1.03, "grad_norm": 8.476302146911621, "learning_rate": 1.5316115540539576e-05, "loss": 1.0336, "step": 6373 }, { "epoch": 1.03, "grad_norm": 7.147061824798584, "learning_rate": 1.5314643402183066e-05, "loss": 0.8086, "step": 6374 }, { "epoch": 1.03, "grad_norm": 7.679186820983887, "learning_rate": 1.531317110329175e-05, "loss": 1.1696, "step": 6375 }, { "epoch": 1.03, "grad_norm": 9.327970504760742, "learning_rate": 1.5311698643910102e-05, "loss": 0.9558, "step": 6376 }, { "epoch": 1.03, "grad_norm": 10.206941604614258, "learning_rate": 1.5310226024082596e-05, "loss": 1.1744, "step": 6377 }, { "epoch": 1.03, "grad_norm": 7.77450704574585, "learning_rate": 1.5308753243853714e-05, "loss": 1.0623, "step": 6378 }, { "epoch": 1.03, "grad_norm": 9.59387493133545, "learning_rate": 1.5307280303267946e-05, "loss": 0.9457, "step": 6379 }, { "epoch": 1.03, "grad_norm": 8.462964057922363, "learning_rate": 1.5305807202369782e-05, "loss": 0.8364, "step": 6380 }, { "epoch": 1.03, "grad_norm": 10.907856941223145, "learning_rate": 1.530433394120372e-05, "loss": 1.1471, "step": 6381 }, { "epoch": 1.03, "grad_norm": 7.622507572174072, "learning_rate": 1.5302860519814256e-05, "loss": 1.238, "step": 6382 }, { "epoch": 1.03, "grad_norm": 9.21577262878418, "learning_rate": 1.5301386938245904e-05, "loss": 1.0571, "step": 6383 }, { "epoch": 1.03, "grad_norm": 7.771770000457764, "learning_rate": 1.5299913196543175e-05, "loss": 0.9152, "step": 6384 }, { "epoch": 1.03, "grad_norm": 10.581716537475586, "learning_rate": 1.5298439294750578e-05, "loss": 0.9364, "step": 6385 }, { "epoch": 1.03, "grad_norm": 10.279641151428223, "learning_rate": 1.529696523291264e-05, "loss": 1.0391, "step": 6386 }, { "epoch": 1.03, "grad_norm": 7.862023830413818, "learning_rate": 1.5295491011073883e-05, "loss": 0.767, "step": 6387 }, { "epoch": 1.03, "grad_norm": 8.812541007995605, "learning_rate": 1.5294016629278842e-05, "loss": 1.1306, "step": 6388 }, { "epoch": 1.03, "grad_norm": 9.672611236572266, "learning_rate": 1.5292542087572044e-05, "loss": 1.3012, "step": 6389 }, { "epoch": 1.03, "grad_norm": 10.573513984680176, "learning_rate": 1.5291067385998037e-05, "loss": 1.0057, "step": 6390 }, { "epoch": 1.03, "grad_norm": 8.767925262451172, "learning_rate": 1.5289592524601363e-05, "loss": 1.1442, "step": 6391 }, { "epoch": 1.03, "grad_norm": 8.197075843811035, "learning_rate": 1.5288117503426575e-05, "loss": 1.0768, "step": 6392 }, { "epoch": 1.03, "grad_norm": 7.676432132720947, "learning_rate": 1.5286642322518225e-05, "loss": 0.9089, "step": 6393 }, { "epoch": 1.03, "grad_norm": 7.742837429046631, "learning_rate": 1.528516698192087e-05, "loss": 0.9755, "step": 6394 }, { "epoch": 1.03, "grad_norm": 10.392925262451172, "learning_rate": 1.5283691481679078e-05, "loss": 1.1077, "step": 6395 }, { "epoch": 1.03, "grad_norm": 7.936995506286621, "learning_rate": 1.5282215821837422e-05, "loss": 1.2814, "step": 6396 }, { "epoch": 1.03, "grad_norm": 8.064528465270996, "learning_rate": 1.5280740002440468e-05, "loss": 0.9758, "step": 6397 }, { "epoch": 1.03, "grad_norm": 7.634825229644775, "learning_rate": 1.5279264023532797e-05, "loss": 0.9987, "step": 6398 }, { "epoch": 1.03, "grad_norm": 7.722126007080078, "learning_rate": 1.5277787885158994e-05, "loss": 0.909, "step": 6399 }, { "epoch": 1.03, "grad_norm": 9.1707124710083, "learning_rate": 1.5276311587363652e-05, "loss": 0.9624, "step": 6400 }, { "epoch": 1.03, "grad_norm": 8.070160865783691, "learning_rate": 1.527483513019135e-05, "loss": 1.1561, "step": 6401 }, { "epoch": 1.03, "grad_norm": 8.054512977600098, "learning_rate": 1.52733585136867e-05, "loss": 1.0347, "step": 6402 }, { "epoch": 1.03, "grad_norm": 7.680056095123291, "learning_rate": 1.5271881737894297e-05, "loss": 0.8855, "step": 6403 }, { "epoch": 1.03, "grad_norm": 7.574863910675049, "learning_rate": 1.527040480285876e-05, "loss": 1.141, "step": 6404 }, { "epoch": 1.03, "grad_norm": 10.238926887512207, "learning_rate": 1.5268927708624687e-05, "loss": 0.9241, "step": 6405 }, { "epoch": 1.03, "grad_norm": 9.024621963500977, "learning_rate": 1.5267450455236705e-05, "loss": 1.0755, "step": 6406 }, { "epoch": 1.03, "grad_norm": 9.395217895507812, "learning_rate": 1.5265973042739428e-05, "loss": 0.9964, "step": 6407 }, { "epoch": 1.03, "grad_norm": 8.708480834960938, "learning_rate": 1.526449547117749e-05, "loss": 0.8698, "step": 6408 }, { "epoch": 1.03, "grad_norm": 8.710713386535645, "learning_rate": 1.526301774059552e-05, "loss": 1.0867, "step": 6409 }, { "epoch": 1.03, "grad_norm": 8.052865982055664, "learning_rate": 1.526153985103816e-05, "loss": 0.7895, "step": 6410 }, { "epoch": 1.03, "grad_norm": 8.411518096923828, "learning_rate": 1.526006180255004e-05, "loss": 1.0141, "step": 6411 }, { "epoch": 1.03, "grad_norm": 9.303936958312988, "learning_rate": 1.5258583595175813e-05, "loss": 1.0884, "step": 6412 }, { "epoch": 1.03, "grad_norm": 8.400490760803223, "learning_rate": 1.525710522896013e-05, "loss": 1.0064, "step": 6413 }, { "epoch": 1.03, "grad_norm": 9.451638221740723, "learning_rate": 1.5255626703947645e-05, "loss": 1.2133, "step": 6414 }, { "epoch": 1.03, "grad_norm": 9.607084274291992, "learning_rate": 1.5254148020183021e-05, "loss": 1.2906, "step": 6415 }, { "epoch": 1.03, "grad_norm": 10.527120590209961, "learning_rate": 1.5252669177710923e-05, "loss": 1.1393, "step": 6416 }, { "epoch": 1.03, "grad_norm": 9.549798011779785, "learning_rate": 1.525119017657602e-05, "loss": 1.2523, "step": 6417 }, { "epoch": 1.03, "grad_norm": 9.251750946044922, "learning_rate": 1.5249711016822985e-05, "loss": 0.9677, "step": 6418 }, { "epoch": 1.03, "grad_norm": 8.500478744506836, "learning_rate": 1.5248231698496501e-05, "loss": 1.1166, "step": 6419 }, { "epoch": 1.03, "grad_norm": 8.639172554016113, "learning_rate": 1.5246752221641253e-05, "loss": 1.1309, "step": 6420 }, { "epoch": 1.03, "grad_norm": 8.511662483215332, "learning_rate": 1.5245272586301925e-05, "loss": 1.2794, "step": 6421 }, { "epoch": 1.03, "grad_norm": 10.083131790161133, "learning_rate": 1.5243792792523214e-05, "loss": 1.2038, "step": 6422 }, { "epoch": 1.03, "grad_norm": 9.237825393676758, "learning_rate": 1.5242312840349823e-05, "loss": 0.9506, "step": 6423 }, { "epoch": 1.03, "grad_norm": 7.636457920074463, "learning_rate": 1.5240832729826446e-05, "loss": 0.8457, "step": 6424 }, { "epoch": 1.03, "grad_norm": 9.684635162353516, "learning_rate": 1.52393524609978e-05, "loss": 1.1471, "step": 6425 }, { "epoch": 1.03, "grad_norm": 7.786375522613525, "learning_rate": 1.5237872033908598e-05, "loss": 1.0385, "step": 6426 }, { "epoch": 1.03, "grad_norm": 10.180091857910156, "learning_rate": 1.5236391448603554e-05, "loss": 1.2358, "step": 6427 }, { "epoch": 1.03, "grad_norm": 10.187891006469727, "learning_rate": 1.5234910705127391e-05, "loss": 1.2991, "step": 6428 }, { "epoch": 1.03, "grad_norm": 9.784804344177246, "learning_rate": 1.5233429803524837e-05, "loss": 1.2841, "step": 6429 }, { "epoch": 1.04, "grad_norm": 9.03783893585205, "learning_rate": 1.5231948743840628e-05, "loss": 1.2056, "step": 6430 }, { "epoch": 1.04, "grad_norm": 8.746915817260742, "learning_rate": 1.5230467526119495e-05, "loss": 1.1715, "step": 6431 }, { "epoch": 1.04, "grad_norm": 7.432458400726318, "learning_rate": 1.5228986150406181e-05, "loss": 0.8792, "step": 6432 }, { "epoch": 1.04, "grad_norm": 7.383057594299316, "learning_rate": 1.5227504616745438e-05, "loss": 0.851, "step": 6433 }, { "epoch": 1.04, "grad_norm": 9.181208610534668, "learning_rate": 1.5226022925182014e-05, "loss": 1.2061, "step": 6434 }, { "epoch": 1.04, "grad_norm": 12.315799713134766, "learning_rate": 1.5224541075760667e-05, "loss": 1.168, "step": 6435 }, { "epoch": 1.04, "grad_norm": 9.19285774230957, "learning_rate": 1.5223059068526154e-05, "loss": 0.9552, "step": 6436 }, { "epoch": 1.04, "grad_norm": 7.23040771484375, "learning_rate": 1.5221576903523243e-05, "loss": 1.0325, "step": 6437 }, { "epoch": 1.04, "grad_norm": 9.876721382141113, "learning_rate": 1.5220094580796701e-05, "loss": 1.2184, "step": 6438 }, { "epoch": 1.04, "grad_norm": 8.047661781311035, "learning_rate": 1.5218612100391312e-05, "loss": 0.9729, "step": 6439 }, { "epoch": 1.04, "grad_norm": 8.593487739562988, "learning_rate": 1.5217129462351848e-05, "loss": 0.9837, "step": 6440 }, { "epoch": 1.04, "grad_norm": 7.894111633300781, "learning_rate": 1.5215646666723097e-05, "loss": 1.1361, "step": 6441 }, { "epoch": 1.04, "grad_norm": 10.439593315124512, "learning_rate": 1.5214163713549841e-05, "loss": 1.167, "step": 6442 }, { "epoch": 1.04, "grad_norm": 7.676239967346191, "learning_rate": 1.5212680602876887e-05, "loss": 0.9385, "step": 6443 }, { "epoch": 1.04, "grad_norm": 8.23311710357666, "learning_rate": 1.5211197334749028e-05, "loss": 0.8855, "step": 6444 }, { "epoch": 1.04, "grad_norm": 8.89449405670166, "learning_rate": 1.520971390921107e-05, "loss": 0.9835, "step": 6445 }, { "epoch": 1.04, "grad_norm": 10.983988761901855, "learning_rate": 1.5208230326307813e-05, "loss": 1.0, "step": 6446 }, { "epoch": 1.04, "grad_norm": 8.390898704528809, "learning_rate": 1.5206746586084084e-05, "loss": 1.1109, "step": 6447 }, { "epoch": 1.04, "grad_norm": 7.111846446990967, "learning_rate": 1.520526268858469e-05, "loss": 0.7509, "step": 6448 }, { "epoch": 1.04, "grad_norm": 8.522960662841797, "learning_rate": 1.5203778633854458e-05, "loss": 1.05, "step": 6449 }, { "epoch": 1.04, "grad_norm": 9.171714782714844, "learning_rate": 1.5202294421938213e-05, "loss": 1.0858, "step": 6450 }, { "epoch": 1.04, "grad_norm": 8.266109466552734, "learning_rate": 1.5200810052880793e-05, "loss": 0.8594, "step": 6451 }, { "epoch": 1.04, "grad_norm": 9.380932807922363, "learning_rate": 1.5199325526727027e-05, "loss": 1.1268, "step": 6452 }, { "epoch": 1.04, "grad_norm": 9.0874605178833, "learning_rate": 1.5197840843521765e-05, "loss": 1.0737, "step": 6453 }, { "epoch": 1.04, "grad_norm": 9.50327205657959, "learning_rate": 1.5196356003309846e-05, "loss": 1.1308, "step": 6454 }, { "epoch": 1.04, "grad_norm": 8.414498329162598, "learning_rate": 1.519487100613613e-05, "loss": 1.0531, "step": 6455 }, { "epoch": 1.04, "grad_norm": 7.592790126800537, "learning_rate": 1.5193385852045466e-05, "loss": 1.1324, "step": 6456 }, { "epoch": 1.04, "grad_norm": 7.6879143714904785, "learning_rate": 1.5191900541082713e-05, "loss": 1.0066, "step": 6457 }, { "epoch": 1.04, "grad_norm": 7.701913833618164, "learning_rate": 1.5190415073292743e-05, "loss": 0.9732, "step": 6458 }, { "epoch": 1.04, "grad_norm": 9.806794166564941, "learning_rate": 1.5188929448720426e-05, "loss": 1.1646, "step": 6459 }, { "epoch": 1.04, "grad_norm": 8.653604507446289, "learning_rate": 1.5187443667410633e-05, "loss": 1.3109, "step": 6460 }, { "epoch": 1.04, "grad_norm": 9.494762420654297, "learning_rate": 1.5185957729408246e-05, "loss": 1.2339, "step": 6461 }, { "epoch": 1.04, "grad_norm": 8.89987850189209, "learning_rate": 1.5184471634758145e-05, "loss": 1.2101, "step": 6462 }, { "epoch": 1.04, "grad_norm": 8.43492317199707, "learning_rate": 1.5182985383505227e-05, "loss": 0.9478, "step": 6463 }, { "epoch": 1.04, "grad_norm": 7.318111896514893, "learning_rate": 1.5181498975694377e-05, "loss": 0.9614, "step": 6464 }, { "epoch": 1.04, "grad_norm": 8.785943031311035, "learning_rate": 1.5180012411370499e-05, "loss": 0.952, "step": 6465 }, { "epoch": 1.04, "grad_norm": 9.557389259338379, "learning_rate": 1.51785256905785e-05, "loss": 0.9345, "step": 6466 }, { "epoch": 1.04, "grad_norm": 8.785365104675293, "learning_rate": 1.517703881336328e-05, "loss": 1.0644, "step": 6467 }, { "epoch": 1.04, "grad_norm": 9.003337860107422, "learning_rate": 1.5175551779769758e-05, "loss": 1.154, "step": 6468 }, { "epoch": 1.04, "grad_norm": 8.971991539001465, "learning_rate": 1.5174064589842847e-05, "loss": 1.1901, "step": 6469 }, { "epoch": 1.04, "grad_norm": 11.459334373474121, "learning_rate": 1.5172577243627473e-05, "loss": 1.2358, "step": 6470 }, { "epoch": 1.04, "grad_norm": 9.701204299926758, "learning_rate": 1.5171089741168562e-05, "loss": 0.999, "step": 6471 }, { "epoch": 1.04, "grad_norm": 8.353025436401367, "learning_rate": 1.5169602082511041e-05, "loss": 1.1629, "step": 6472 }, { "epoch": 1.04, "grad_norm": 7.0844221115112305, "learning_rate": 1.5168114267699853e-05, "loss": 0.9312, "step": 6473 }, { "epoch": 1.04, "grad_norm": 9.931441307067871, "learning_rate": 1.5166626296779936e-05, "loss": 1.1454, "step": 6474 }, { "epoch": 1.04, "grad_norm": 11.187477111816406, "learning_rate": 1.5165138169796239e-05, "loss": 1.0154, "step": 6475 }, { "epoch": 1.04, "grad_norm": 8.224852561950684, "learning_rate": 1.5163649886793708e-05, "loss": 0.7918, "step": 6476 }, { "epoch": 1.04, "grad_norm": 11.8853759765625, "learning_rate": 1.5162161447817299e-05, "loss": 1.4043, "step": 6477 }, { "epoch": 1.04, "grad_norm": 9.269043922424316, "learning_rate": 1.5160672852911975e-05, "loss": 1.3459, "step": 6478 }, { "epoch": 1.04, "grad_norm": 7.818443298339844, "learning_rate": 1.5159184102122697e-05, "loss": 0.9379, "step": 6479 }, { "epoch": 1.04, "grad_norm": 9.132213592529297, "learning_rate": 1.5157695195494437e-05, "loss": 0.9626, "step": 6480 }, { "epoch": 1.04, "grad_norm": 10.66551685333252, "learning_rate": 1.5156206133072168e-05, "loss": 0.9761, "step": 6481 }, { "epoch": 1.04, "grad_norm": 8.433810234069824, "learning_rate": 1.5154716914900869e-05, "loss": 1.2691, "step": 6482 }, { "epoch": 1.04, "grad_norm": 8.46733570098877, "learning_rate": 1.5153227541025524e-05, "loss": 1.1202, "step": 6483 }, { "epoch": 1.04, "grad_norm": 9.1643705368042, "learning_rate": 1.515173801149112e-05, "loss": 1.1224, "step": 6484 }, { "epoch": 1.04, "grad_norm": 11.623916625976562, "learning_rate": 1.515024832634265e-05, "loss": 1.3398, "step": 6485 }, { "epoch": 1.04, "grad_norm": 9.531122207641602, "learning_rate": 1.5148758485625119e-05, "loss": 1.1014, "step": 6486 }, { "epoch": 1.04, "grad_norm": 8.255443572998047, "learning_rate": 1.5147268489383518e-05, "loss": 1.1848, "step": 6487 }, { "epoch": 1.04, "grad_norm": 8.574347496032715, "learning_rate": 1.514577833766286e-05, "loss": 0.9525, "step": 6488 }, { "epoch": 1.04, "grad_norm": 9.75300121307373, "learning_rate": 1.5144288030508154e-05, "loss": 1.4412, "step": 6489 }, { "epoch": 1.04, "grad_norm": 8.51225471496582, "learning_rate": 1.5142797567964418e-05, "loss": 1.1349, "step": 6490 }, { "epoch": 1.04, "grad_norm": 8.515158653259277, "learning_rate": 1.5141306950076674e-05, "loss": 1.0238, "step": 6491 }, { "epoch": 1.05, "grad_norm": 6.9927778244018555, "learning_rate": 1.5139816176889945e-05, "loss": 0.8172, "step": 6492 }, { "epoch": 1.05, "grad_norm": 7.738452911376953, "learning_rate": 1.5138325248449263e-05, "loss": 0.8808, "step": 6493 }, { "epoch": 1.05, "grad_norm": 9.524491310119629, "learning_rate": 1.5136834164799667e-05, "loss": 0.9708, "step": 6494 }, { "epoch": 1.05, "grad_norm": 7.436611175537109, "learning_rate": 1.5135342925986191e-05, "loss": 0.8964, "step": 6495 }, { "epoch": 1.05, "grad_norm": 8.373017311096191, "learning_rate": 1.5133851532053883e-05, "loss": 1.1111, "step": 6496 }, { "epoch": 1.05, "grad_norm": 7.925312519073486, "learning_rate": 1.5132359983047789e-05, "loss": 1.0878, "step": 6497 }, { "epoch": 1.05, "grad_norm": 6.587380409240723, "learning_rate": 1.513086827901297e-05, "loss": 1.0199, "step": 6498 }, { "epoch": 1.05, "grad_norm": 8.522022247314453, "learning_rate": 1.5129376419994476e-05, "loss": 1.1486, "step": 6499 }, { "epoch": 1.05, "grad_norm": 10.214077949523926, "learning_rate": 1.5127884406037372e-05, "loss": 1.2327, "step": 6500 }, { "epoch": 1.05, "grad_norm": 7.392748832702637, "learning_rate": 1.5126392237186731e-05, "loss": 0.8511, "step": 6501 }, { "epoch": 1.05, "grad_norm": 9.432777404785156, "learning_rate": 1.5124899913487622e-05, "loss": 1.1709, "step": 6502 }, { "epoch": 1.05, "grad_norm": 8.037459373474121, "learning_rate": 1.5123407434985123e-05, "loss": 1.3526, "step": 6503 }, { "epoch": 1.05, "grad_norm": 10.47724437713623, "learning_rate": 1.5121914801724313e-05, "loss": 1.263, "step": 6504 }, { "epoch": 1.05, "grad_norm": 6.764521598815918, "learning_rate": 1.5120422013750286e-05, "loss": 0.9961, "step": 6505 }, { "epoch": 1.05, "grad_norm": 11.33834171295166, "learning_rate": 1.5118929071108127e-05, "loss": 1.2413, "step": 6506 }, { "epoch": 1.05, "grad_norm": 8.57413387298584, "learning_rate": 1.5117435973842934e-05, "loss": 1.1114, "step": 6507 }, { "epoch": 1.05, "grad_norm": 7.811498641967773, "learning_rate": 1.5115942721999807e-05, "loss": 0.8453, "step": 6508 }, { "epoch": 1.05, "grad_norm": 9.614519119262695, "learning_rate": 1.511444931562385e-05, "loss": 1.1493, "step": 6509 }, { "epoch": 1.05, "grad_norm": 7.357575416564941, "learning_rate": 1.5112955754760181e-05, "loss": 0.8092, "step": 6510 }, { "epoch": 1.05, "grad_norm": 8.518487930297852, "learning_rate": 1.5111462039453906e-05, "loss": 1.0252, "step": 6511 }, { "epoch": 1.05, "grad_norm": 10.675176620483398, "learning_rate": 1.5109968169750144e-05, "loss": 1.4417, "step": 6512 }, { "epoch": 1.05, "grad_norm": 10.743783950805664, "learning_rate": 1.5108474145694026e-05, "loss": 1.0062, "step": 6513 }, { "epoch": 1.05, "grad_norm": 8.57036018371582, "learning_rate": 1.5106979967330674e-05, "loss": 0.7947, "step": 6514 }, { "epoch": 1.05, "grad_norm": 9.607552528381348, "learning_rate": 1.5105485634705225e-05, "loss": 1.2023, "step": 6515 }, { "epoch": 1.05, "grad_norm": 9.890274047851562, "learning_rate": 1.5103991147862815e-05, "loss": 0.978, "step": 6516 }, { "epoch": 1.05, "grad_norm": 8.678827285766602, "learning_rate": 1.5102496506848594e-05, "loss": 1.3003, "step": 6517 }, { "epoch": 1.05, "grad_norm": 9.206110954284668, "learning_rate": 1.5101001711707697e-05, "loss": 1.2281, "step": 6518 }, { "epoch": 1.05, "grad_norm": 10.814441680908203, "learning_rate": 1.5099506762485283e-05, "loss": 1.2853, "step": 6519 }, { "epoch": 1.05, "grad_norm": 9.228198051452637, "learning_rate": 1.5098011659226509e-05, "loss": 1.1177, "step": 6520 }, { "epoch": 1.05, "grad_norm": 6.923428535461426, "learning_rate": 1.5096516401976531e-05, "loss": 0.8196, "step": 6521 }, { "epoch": 1.05, "grad_norm": 9.295330047607422, "learning_rate": 1.5095020990780525e-05, "loss": 0.9859, "step": 6522 }, { "epoch": 1.05, "grad_norm": 10.694509506225586, "learning_rate": 1.5093525425683649e-05, "loss": 1.1694, "step": 6523 }, { "epoch": 1.05, "grad_norm": 8.621025085449219, "learning_rate": 1.5092029706731089e-05, "loss": 0.9503, "step": 6524 }, { "epoch": 1.05, "grad_norm": 8.150275230407715, "learning_rate": 1.5090533833968022e-05, "loss": 1.073, "step": 6525 }, { "epoch": 1.05, "grad_norm": 8.398383140563965, "learning_rate": 1.5089037807439626e-05, "loss": 1.1473, "step": 6526 }, { "epoch": 1.05, "grad_norm": 9.212102890014648, "learning_rate": 1.50875416271911e-05, "loss": 1.1125, "step": 6527 }, { "epoch": 1.05, "grad_norm": 7.477639198303223, "learning_rate": 1.508604529326763e-05, "loss": 0.9801, "step": 6528 }, { "epoch": 1.05, "grad_norm": 9.414258003234863, "learning_rate": 1.5084548805714422e-05, "loss": 1.4635, "step": 6529 }, { "epoch": 1.05, "grad_norm": 9.720305442810059, "learning_rate": 1.508305216457667e-05, "loss": 0.9439, "step": 6530 }, { "epoch": 1.05, "grad_norm": 8.776970863342285, "learning_rate": 1.5081555369899588e-05, "loss": 1.0525, "step": 6531 }, { "epoch": 1.05, "grad_norm": 8.832648277282715, "learning_rate": 1.5080058421728385e-05, "loss": 1.2287, "step": 6532 }, { "epoch": 1.05, "grad_norm": 9.59503173828125, "learning_rate": 1.5078561320108282e-05, "loss": 1.0766, "step": 6533 }, { "epoch": 1.05, "grad_norm": 10.132771492004395, "learning_rate": 1.5077064065084499e-05, "loss": 1.1894, "step": 6534 }, { "epoch": 1.05, "grad_norm": 8.048397064208984, "learning_rate": 1.507556665670226e-05, "loss": 1.3465, "step": 6535 }, { "epoch": 1.05, "grad_norm": 9.471837043762207, "learning_rate": 1.5074069095006799e-05, "loss": 0.9955, "step": 6536 }, { "epoch": 1.05, "grad_norm": 9.33910846710205, "learning_rate": 1.507257138004335e-05, "loss": 1.0127, "step": 6537 }, { "epoch": 1.05, "grad_norm": 8.242992401123047, "learning_rate": 1.5071073511857153e-05, "loss": 1.1471, "step": 6538 }, { "epoch": 1.05, "grad_norm": 10.373634338378906, "learning_rate": 1.5069575490493453e-05, "loss": 1.1694, "step": 6539 }, { "epoch": 1.05, "grad_norm": 9.784109115600586, "learning_rate": 1.50680773159975e-05, "loss": 0.9998, "step": 6540 }, { "epoch": 1.05, "grad_norm": 9.734086036682129, "learning_rate": 1.506657898841455e-05, "loss": 0.9866, "step": 6541 }, { "epoch": 1.05, "grad_norm": 11.432289123535156, "learning_rate": 1.5065080507789853e-05, "loss": 0.9817, "step": 6542 }, { "epoch": 1.05, "grad_norm": 8.980244636535645, "learning_rate": 1.5063581874168685e-05, "loss": 1.0583, "step": 6543 }, { "epoch": 1.05, "grad_norm": 9.811476707458496, "learning_rate": 1.5062083087596305e-05, "loss": 1.0392, "step": 6544 }, { "epoch": 1.05, "grad_norm": 9.712773323059082, "learning_rate": 1.5060584148117986e-05, "loss": 1.0113, "step": 6545 }, { "epoch": 1.05, "grad_norm": 8.55257511138916, "learning_rate": 1.5059085055779009e-05, "loss": 1.1849, "step": 6546 }, { "epoch": 1.05, "grad_norm": 8.211897850036621, "learning_rate": 1.5057585810624653e-05, "loss": 1.1155, "step": 6547 }, { "epoch": 1.05, "grad_norm": 8.834423065185547, "learning_rate": 1.5056086412700207e-05, "loss": 1.2896, "step": 6548 }, { "epoch": 1.05, "grad_norm": 10.611686706542969, "learning_rate": 1.505458686205096e-05, "loss": 1.107, "step": 6549 }, { "epoch": 1.05, "grad_norm": 7.891209602355957, "learning_rate": 1.5053087158722207e-05, "loss": 1.0496, "step": 6550 }, { "epoch": 1.05, "grad_norm": 8.109119415283203, "learning_rate": 1.505158730275925e-05, "loss": 0.8834, "step": 6551 }, { "epoch": 1.05, "grad_norm": 9.735189437866211, "learning_rate": 1.5050087294207393e-05, "loss": 1.0476, "step": 6552 }, { "epoch": 1.05, "grad_norm": 7.542333602905273, "learning_rate": 1.5048587133111947e-05, "loss": 0.7926, "step": 6553 }, { "epoch": 1.06, "grad_norm": 9.141173362731934, "learning_rate": 1.5047086819518222e-05, "loss": 1.2263, "step": 6554 }, { "epoch": 1.06, "grad_norm": 9.003769874572754, "learning_rate": 1.504558635347154e-05, "loss": 0.9237, "step": 6555 }, { "epoch": 1.06, "grad_norm": 11.200748443603516, "learning_rate": 1.5044085735017227e-05, "loss": 1.1445, "step": 6556 }, { "epoch": 1.06, "grad_norm": 7.111963272094727, "learning_rate": 1.5042584964200604e-05, "loss": 0.7733, "step": 6557 }, { "epoch": 1.06, "grad_norm": 8.044981002807617, "learning_rate": 1.5041084041067007e-05, "loss": 1.0315, "step": 6558 }, { "epoch": 1.06, "grad_norm": 9.199095726013184, "learning_rate": 1.5039582965661775e-05, "loss": 1.0745, "step": 6559 }, { "epoch": 1.06, "grad_norm": 8.159137725830078, "learning_rate": 1.5038081738030243e-05, "loss": 1.1705, "step": 6560 }, { "epoch": 1.06, "grad_norm": 9.64909839630127, "learning_rate": 1.5036580358217767e-05, "loss": 1.2643, "step": 6561 }, { "epoch": 1.06, "grad_norm": 8.302068710327148, "learning_rate": 1.503507882626969e-05, "loss": 0.862, "step": 6562 }, { "epoch": 1.06, "grad_norm": 9.605777740478516, "learning_rate": 1.5033577142231372e-05, "loss": 0.8098, "step": 6563 }, { "epoch": 1.06, "grad_norm": 7.272134304046631, "learning_rate": 1.5032075306148171e-05, "loss": 1.1944, "step": 6564 }, { "epoch": 1.06, "grad_norm": 8.494803428649902, "learning_rate": 1.5030573318065452e-05, "loss": 1.0986, "step": 6565 }, { "epoch": 1.06, "grad_norm": 8.413108825683594, "learning_rate": 1.5029071178028582e-05, "loss": 1.0792, "step": 6566 }, { "epoch": 1.06, "grad_norm": 10.181402206420898, "learning_rate": 1.5027568886082939e-05, "loss": 1.1836, "step": 6567 }, { "epoch": 1.06, "grad_norm": 11.085859298706055, "learning_rate": 1.5026066442273902e-05, "loss": 1.0467, "step": 6568 }, { "epoch": 1.06, "grad_norm": 8.665338516235352, "learning_rate": 1.5024563846646846e-05, "loss": 1.1153, "step": 6569 }, { "epoch": 1.06, "grad_norm": 10.091863632202148, "learning_rate": 1.5023061099247169e-05, "loss": 1.0424, "step": 6570 }, { "epoch": 1.06, "grad_norm": 8.87381649017334, "learning_rate": 1.5021558200120257e-05, "loss": 1.035, "step": 6571 }, { "epoch": 1.06, "grad_norm": 8.474468231201172, "learning_rate": 1.5020055149311508e-05, "loss": 0.8519, "step": 6572 }, { "epoch": 1.06, "grad_norm": 10.710922241210938, "learning_rate": 1.5018551946866322e-05, "loss": 1.024, "step": 6573 }, { "epoch": 1.06, "grad_norm": 9.547966003417969, "learning_rate": 1.5017048592830103e-05, "loss": 0.9377, "step": 6574 }, { "epoch": 1.06, "grad_norm": 9.105335235595703, "learning_rate": 1.5015545087248268e-05, "loss": 1.2372, "step": 6575 }, { "epoch": 1.06, "grad_norm": 10.413628578186035, "learning_rate": 1.5014041430166236e-05, "loss": 1.0976, "step": 6576 }, { "epoch": 1.06, "grad_norm": 10.803670883178711, "learning_rate": 1.501253762162941e-05, "loss": 1.0662, "step": 6577 }, { "epoch": 1.06, "grad_norm": 8.948770523071289, "learning_rate": 1.501103366168323e-05, "loss": 1.0302, "step": 6578 }, { "epoch": 1.06, "grad_norm": 9.242288589477539, "learning_rate": 1.5009529550373114e-05, "loss": 1.1192, "step": 6579 }, { "epoch": 1.06, "grad_norm": 9.709328651428223, "learning_rate": 1.5008025287744503e-05, "loss": 1.1884, "step": 6580 }, { "epoch": 1.06, "grad_norm": 9.503530502319336, "learning_rate": 1.5006520873842832e-05, "loss": 1.0835, "step": 6581 }, { "epoch": 1.06, "grad_norm": 8.730644226074219, "learning_rate": 1.5005016308713543e-05, "loss": 0.8815, "step": 6582 }, { "epoch": 1.06, "grad_norm": 9.703890800476074, "learning_rate": 1.500351159240208e-05, "loss": 1.1484, "step": 6583 }, { "epoch": 1.06, "grad_norm": 11.411490440368652, "learning_rate": 1.5002006724953901e-05, "loss": 1.1386, "step": 6584 }, { "epoch": 1.06, "grad_norm": 10.782094955444336, "learning_rate": 1.5000501706414459e-05, "loss": 1.2652, "step": 6585 }, { "epoch": 1.06, "grad_norm": 7.5385260581970215, "learning_rate": 1.4998996536829218e-05, "loss": 0.9203, "step": 6586 }, { "epoch": 1.06, "grad_norm": 9.923107147216797, "learning_rate": 1.4997491216243636e-05, "loss": 1.057, "step": 6587 }, { "epoch": 1.06, "grad_norm": 9.803120613098145, "learning_rate": 1.4995985744703192e-05, "loss": 1.3552, "step": 6588 }, { "epoch": 1.06, "grad_norm": 7.338006496429443, "learning_rate": 1.4994480122253355e-05, "loss": 0.8198, "step": 6589 }, { "epoch": 1.06, "grad_norm": 8.12337589263916, "learning_rate": 1.4992974348939604e-05, "loss": 1.2258, "step": 6590 }, { "epoch": 1.06, "grad_norm": 10.104512214660645, "learning_rate": 1.4991468424807422e-05, "loss": 1.1445, "step": 6591 }, { "epoch": 1.06, "grad_norm": 10.198009490966797, "learning_rate": 1.4989962349902302e-05, "loss": 1.0802, "step": 6592 }, { "epoch": 1.06, "grad_norm": 8.57174015045166, "learning_rate": 1.4988456124269732e-05, "loss": 1.1681, "step": 6593 }, { "epoch": 1.06, "grad_norm": 8.814406394958496, "learning_rate": 1.4986949747955208e-05, "loss": 0.9553, "step": 6594 }, { "epoch": 1.06, "grad_norm": 7.748596668243408, "learning_rate": 1.4985443221004238e-05, "loss": 0.9797, "step": 6595 }, { "epoch": 1.06, "grad_norm": 9.126605987548828, "learning_rate": 1.4983936543462323e-05, "loss": 1.1899, "step": 6596 }, { "epoch": 1.06, "grad_norm": 9.32706069946289, "learning_rate": 1.4982429715374976e-05, "loss": 1.3701, "step": 6597 }, { "epoch": 1.06, "grad_norm": 7.315613269805908, "learning_rate": 1.4980922736787714e-05, "loss": 1.0875, "step": 6598 }, { "epoch": 1.06, "grad_norm": 11.02521800994873, "learning_rate": 1.497941560774605e-05, "loss": 1.1954, "step": 6599 }, { "epoch": 1.06, "grad_norm": 9.332379341125488, "learning_rate": 1.4977908328295521e-05, "loss": 0.9268, "step": 6600 }, { "epoch": 1.06, "grad_norm": 9.19748306274414, "learning_rate": 1.4976400898481645e-05, "loss": 1.1175, "step": 6601 }, { "epoch": 1.06, "grad_norm": 9.513754844665527, "learning_rate": 1.497489331834996e-05, "loss": 1.2585, "step": 6602 }, { "epoch": 1.06, "grad_norm": 8.86513614654541, "learning_rate": 1.4973385587946001e-05, "loss": 1.2594, "step": 6603 }, { "epoch": 1.06, "grad_norm": 7.861668109893799, "learning_rate": 1.4971877707315315e-05, "loss": 0.9439, "step": 6604 }, { "epoch": 1.06, "grad_norm": 7.633935451507568, "learning_rate": 1.4970369676503446e-05, "loss": 0.995, "step": 6605 }, { "epoch": 1.06, "grad_norm": 7.040165424346924, "learning_rate": 1.4968861495555948e-05, "loss": 1.0607, "step": 6606 }, { "epoch": 1.06, "grad_norm": 8.915794372558594, "learning_rate": 1.4967353164518379e-05, "loss": 1.0126, "step": 6607 }, { "epoch": 1.06, "grad_norm": 10.401593208312988, "learning_rate": 1.4965844683436298e-05, "loss": 1.13, "step": 6608 }, { "epoch": 1.06, "grad_norm": 9.44517707824707, "learning_rate": 1.4964336052355264e-05, "loss": 1.1068, "step": 6609 }, { "epoch": 1.06, "grad_norm": 8.810553550720215, "learning_rate": 1.4962827271320857e-05, "loss": 0.9608, "step": 6610 }, { "epoch": 1.06, "grad_norm": 7.929388523101807, "learning_rate": 1.4961318340378648e-05, "loss": 0.9168, "step": 6611 }, { "epoch": 1.06, "grad_norm": 8.014939308166504, "learning_rate": 1.4959809259574214e-05, "loss": 0.9163, "step": 6612 }, { "epoch": 1.06, "grad_norm": 8.457052230834961, "learning_rate": 1.4958300028953137e-05, "loss": 0.938, "step": 6613 }, { "epoch": 1.06, "grad_norm": 8.914114952087402, "learning_rate": 1.495679064856101e-05, "loss": 1.3699, "step": 6614 }, { "epoch": 1.06, "grad_norm": 8.379371643066406, "learning_rate": 1.4955281118443421e-05, "loss": 1.2849, "step": 6615 }, { "epoch": 1.07, "grad_norm": 6.930414199829102, "learning_rate": 1.4953771438645973e-05, "loss": 0.795, "step": 6616 }, { "epoch": 1.07, "grad_norm": 7.152333736419678, "learning_rate": 1.4952261609214259e-05, "loss": 1.0501, "step": 6617 }, { "epoch": 1.07, "grad_norm": 9.822359085083008, "learning_rate": 1.4950751630193892e-05, "loss": 0.9501, "step": 6618 }, { "epoch": 1.07, "grad_norm": 9.927939414978027, "learning_rate": 1.4949241501630482e-05, "loss": 1.1995, "step": 6619 }, { "epoch": 1.07, "grad_norm": 7.781224250793457, "learning_rate": 1.4947731223569641e-05, "loss": 1.1586, "step": 6620 }, { "epoch": 1.07, "grad_norm": 9.991540908813477, "learning_rate": 1.4946220796056992e-05, "loss": 1.029, "step": 6621 }, { "epoch": 1.07, "grad_norm": 10.312621116638184, "learning_rate": 1.4944710219138154e-05, "loss": 1.0815, "step": 6622 }, { "epoch": 1.07, "grad_norm": 7.164866924285889, "learning_rate": 1.4943199492858762e-05, "loss": 0.9578, "step": 6623 }, { "epoch": 1.07, "grad_norm": 9.500629425048828, "learning_rate": 1.4941688617264447e-05, "loss": 1.0052, "step": 6624 }, { "epoch": 1.07, "grad_norm": 7.616687774658203, "learning_rate": 1.4940177592400842e-05, "loss": 1.1244, "step": 6625 }, { "epoch": 1.07, "grad_norm": 9.305146217346191, "learning_rate": 1.4938666418313594e-05, "loss": 1.2353, "step": 6626 }, { "epoch": 1.07, "grad_norm": 8.231767654418945, "learning_rate": 1.4937155095048352e-05, "loss": 1.0235, "step": 6627 }, { "epoch": 1.07, "grad_norm": 8.408809661865234, "learning_rate": 1.4935643622650764e-05, "loss": 1.1775, "step": 6628 }, { "epoch": 1.07, "grad_norm": 7.709610939025879, "learning_rate": 1.4934132001166485e-05, "loss": 0.7895, "step": 6629 }, { "epoch": 1.07, "grad_norm": 9.287121772766113, "learning_rate": 1.4932620230641174e-05, "loss": 1.0171, "step": 6630 }, { "epoch": 1.07, "grad_norm": 9.083958625793457, "learning_rate": 1.4931108311120501e-05, "loss": 1.1679, "step": 6631 }, { "epoch": 1.07, "grad_norm": 9.256385803222656, "learning_rate": 1.4929596242650132e-05, "loss": 0.8941, "step": 6632 }, { "epoch": 1.07, "grad_norm": 11.166678428649902, "learning_rate": 1.4928084025275739e-05, "loss": 1.0815, "step": 6633 }, { "epoch": 1.07, "grad_norm": 11.254502296447754, "learning_rate": 1.4926571659043003e-05, "loss": 1.1153, "step": 6634 }, { "epoch": 1.07, "grad_norm": 9.236501693725586, "learning_rate": 1.4925059143997605e-05, "loss": 1.1347, "step": 6635 }, { "epoch": 1.07, "grad_norm": 9.623753547668457, "learning_rate": 1.4923546480185234e-05, "loss": 1.2372, "step": 6636 }, { "epoch": 1.07, "grad_norm": 10.308332443237305, "learning_rate": 1.4922033667651582e-05, "loss": 1.3318, "step": 6637 }, { "epoch": 1.07, "grad_norm": 8.065389633178711, "learning_rate": 1.4920520706442341e-05, "loss": 0.9086, "step": 6638 }, { "epoch": 1.07, "grad_norm": 10.514959335327148, "learning_rate": 1.4919007596603216e-05, "loss": 1.0372, "step": 6639 }, { "epoch": 1.07, "grad_norm": 7.225949764251709, "learning_rate": 1.4917494338179913e-05, "loss": 0.8207, "step": 6640 }, { "epoch": 1.07, "grad_norm": 8.959695816040039, "learning_rate": 1.4915980931218137e-05, "loss": 1.1186, "step": 6641 }, { "epoch": 1.07, "grad_norm": 8.780253410339355, "learning_rate": 1.4914467375763604e-05, "loss": 1.0291, "step": 6642 }, { "epoch": 1.07, "grad_norm": 9.110628128051758, "learning_rate": 1.4912953671862037e-05, "loss": 1.3695, "step": 6643 }, { "epoch": 1.07, "grad_norm": 8.877312660217285, "learning_rate": 1.4911439819559153e-05, "loss": 0.9779, "step": 6644 }, { "epoch": 1.07, "grad_norm": 9.170419692993164, "learning_rate": 1.4909925818900679e-05, "loss": 1.1052, "step": 6645 }, { "epoch": 1.07, "grad_norm": 10.527152061462402, "learning_rate": 1.4908411669932354e-05, "loss": 1.2112, "step": 6646 }, { "epoch": 1.07, "grad_norm": 8.979619979858398, "learning_rate": 1.490689737269991e-05, "loss": 1.0685, "step": 6647 }, { "epoch": 1.07, "grad_norm": 7.020381927490234, "learning_rate": 1.490538292724909e-05, "loss": 0.8912, "step": 6648 }, { "epoch": 1.07, "grad_norm": 8.723040580749512, "learning_rate": 1.4903868333625638e-05, "loss": 0.9695, "step": 6649 }, { "epoch": 1.07, "grad_norm": 8.729066848754883, "learning_rate": 1.4902353591875305e-05, "loss": 1.0035, "step": 6650 }, { "epoch": 1.07, "grad_norm": 8.838642120361328, "learning_rate": 1.4900838702043845e-05, "loss": 1.1797, "step": 6651 }, { "epoch": 1.07, "grad_norm": 8.770442962646484, "learning_rate": 1.4899323664177014e-05, "loss": 1.2702, "step": 6652 }, { "epoch": 1.07, "grad_norm": 9.95851993560791, "learning_rate": 1.4897808478320581e-05, "loss": 1.1862, "step": 6653 }, { "epoch": 1.07, "grad_norm": 8.17101001739502, "learning_rate": 1.4896293144520308e-05, "loss": 1.0581, "step": 6654 }, { "epoch": 1.07, "grad_norm": 8.28370475769043, "learning_rate": 1.4894777662821972e-05, "loss": 1.3032, "step": 6655 }, { "epoch": 1.07, "grad_norm": 8.54781436920166, "learning_rate": 1.489326203327135e-05, "loss": 0.8899, "step": 6656 }, { "epoch": 1.07, "grad_norm": 8.09907054901123, "learning_rate": 1.4891746255914221e-05, "loss": 0.9563, "step": 6657 }, { "epoch": 1.07, "grad_norm": 9.498059272766113, "learning_rate": 1.4890230330796373e-05, "loss": 1.0472, "step": 6658 }, { "epoch": 1.07, "grad_norm": 9.207850456237793, "learning_rate": 1.4888714257963593e-05, "loss": 0.9133, "step": 6659 }, { "epoch": 1.07, "grad_norm": 8.725301742553711, "learning_rate": 1.4887198037461677e-05, "loss": 1.0515, "step": 6660 }, { "epoch": 1.07, "grad_norm": 8.96410083770752, "learning_rate": 1.4885681669336423e-05, "loss": 1.1813, "step": 6661 }, { "epoch": 1.07, "grad_norm": 6.933183193206787, "learning_rate": 1.4884165153633641e-05, "loss": 1.0029, "step": 6662 }, { "epoch": 1.07, "grad_norm": 8.670900344848633, "learning_rate": 1.488264849039913e-05, "loss": 0.8166, "step": 6663 }, { "epoch": 1.07, "grad_norm": 7.530620574951172, "learning_rate": 1.4881131679678707e-05, "loss": 1.0443, "step": 6664 }, { "epoch": 1.07, "grad_norm": 11.436890602111816, "learning_rate": 1.4879614721518188e-05, "loss": 1.0481, "step": 6665 }, { "epoch": 1.07, "grad_norm": 9.521754264831543, "learning_rate": 1.4878097615963399e-05, "loss": 1.2073, "step": 6666 }, { "epoch": 1.07, "grad_norm": 7.0435872077941895, "learning_rate": 1.4876580363060157e-05, "loss": 0.9037, "step": 6667 }, { "epoch": 1.07, "grad_norm": 8.941455841064453, "learning_rate": 1.4875062962854302e-05, "loss": 1.2463, "step": 6668 }, { "epoch": 1.07, "grad_norm": 10.715716361999512, "learning_rate": 1.4873545415391661e-05, "loss": 1.2803, "step": 6669 }, { "epoch": 1.07, "grad_norm": 9.776942253112793, "learning_rate": 1.4872027720718078e-05, "loss": 1.1851, "step": 6670 }, { "epoch": 1.07, "grad_norm": 7.491165637969971, "learning_rate": 1.4870509878879396e-05, "loss": 1.2362, "step": 6671 }, { "epoch": 1.07, "grad_norm": 7.6067214012146, "learning_rate": 1.4868991889921461e-05, "loss": 0.9213, "step": 6672 }, { "epoch": 1.07, "grad_norm": 8.24029541015625, "learning_rate": 1.4867473753890127e-05, "loss": 0.9206, "step": 6673 }, { "epoch": 1.07, "grad_norm": 10.27090835571289, "learning_rate": 1.4865955470831251e-05, "loss": 0.9778, "step": 6674 }, { "epoch": 1.07, "grad_norm": 8.462859153747559, "learning_rate": 1.4864437040790693e-05, "loss": 1.1401, "step": 6675 }, { "epoch": 1.07, "grad_norm": 9.753171920776367, "learning_rate": 1.4862918463814319e-05, "loss": 1.0659, "step": 6676 }, { "epoch": 1.07, "grad_norm": 8.974089622497559, "learning_rate": 1.4861399739948003e-05, "loss": 1.2097, "step": 6677 }, { "epoch": 1.08, "grad_norm": 7.6170878410339355, "learning_rate": 1.485988086923762e-05, "loss": 0.93, "step": 6678 }, { "epoch": 1.08, "grad_norm": 8.526952743530273, "learning_rate": 1.4858361851729042e-05, "loss": 0.8886, "step": 6679 }, { "epoch": 1.08, "grad_norm": 7.542149543762207, "learning_rate": 1.4856842687468155e-05, "loss": 0.9055, "step": 6680 }, { "epoch": 1.08, "grad_norm": 7.27910041809082, "learning_rate": 1.4855323376500854e-05, "loss": 1.0376, "step": 6681 }, { "epoch": 1.08, "grad_norm": 8.993293762207031, "learning_rate": 1.4853803918873027e-05, "loss": 1.2877, "step": 6682 }, { "epoch": 1.08, "grad_norm": 8.702302932739258, "learning_rate": 1.4852284314630567e-05, "loss": 1.3146, "step": 6683 }, { "epoch": 1.08, "grad_norm": 9.269773483276367, "learning_rate": 1.485076456381938e-05, "loss": 1.2187, "step": 6684 }, { "epoch": 1.08, "grad_norm": 9.251751899719238, "learning_rate": 1.4849244666485371e-05, "loss": 1.0566, "step": 6685 }, { "epoch": 1.08, "grad_norm": 9.172491073608398, "learning_rate": 1.484772462267445e-05, "loss": 0.9859, "step": 6686 }, { "epoch": 1.08, "grad_norm": 10.329610824584961, "learning_rate": 1.4846204432432533e-05, "loss": 1.1311, "step": 6687 }, { "epoch": 1.08, "grad_norm": 10.329875946044922, "learning_rate": 1.4844684095805537e-05, "loss": 1.1332, "step": 6688 }, { "epoch": 1.08, "grad_norm": 10.99521541595459, "learning_rate": 1.4843163612839387e-05, "loss": 1.2428, "step": 6689 }, { "epoch": 1.08, "grad_norm": 8.79784870147705, "learning_rate": 1.4841642983580007e-05, "loss": 0.949, "step": 6690 }, { "epoch": 1.08, "grad_norm": 7.710679531097412, "learning_rate": 1.4840122208073336e-05, "loss": 0.8053, "step": 6691 }, { "epoch": 1.08, "grad_norm": 8.02370548248291, "learning_rate": 1.4838601286365304e-05, "loss": 0.9887, "step": 6692 }, { "epoch": 1.08, "grad_norm": 8.74355411529541, "learning_rate": 1.4837080218501857e-05, "loss": 0.9827, "step": 6693 }, { "epoch": 1.08, "grad_norm": 8.916412353515625, "learning_rate": 1.4835559004528938e-05, "loss": 1.2674, "step": 6694 }, { "epoch": 1.08, "grad_norm": 10.787611961364746, "learning_rate": 1.4834037644492498e-05, "loss": 1.1814, "step": 6695 }, { "epoch": 1.08, "grad_norm": 9.903899192810059, "learning_rate": 1.4832516138438488e-05, "loss": 1.2706, "step": 6696 }, { "epoch": 1.08, "grad_norm": 9.843253135681152, "learning_rate": 1.4830994486412877e-05, "loss": 1.0986, "step": 6697 }, { "epoch": 1.08, "grad_norm": 9.686861991882324, "learning_rate": 1.4829472688461616e-05, "loss": 1.2772, "step": 6698 }, { "epoch": 1.08, "grad_norm": 11.414002418518066, "learning_rate": 1.4827950744630681e-05, "loss": 1.2657, "step": 6699 }, { "epoch": 1.08, "grad_norm": 9.432555198669434, "learning_rate": 1.4826428654966039e-05, "loss": 1.1105, "step": 6700 }, { "epoch": 1.08, "grad_norm": 7.669124126434326, "learning_rate": 1.4824906419513671e-05, "loss": 0.834, "step": 6701 }, { "epoch": 1.08, "grad_norm": 9.223856925964355, "learning_rate": 1.4823384038319553e-05, "loss": 1.1645, "step": 6702 }, { "epoch": 1.08, "grad_norm": 8.472288131713867, "learning_rate": 1.4821861511429672e-05, "loss": 1.251, "step": 6703 }, { "epoch": 1.08, "grad_norm": 8.231522560119629, "learning_rate": 1.4820338838890018e-05, "loss": 1.2361, "step": 6704 }, { "epoch": 1.08, "grad_norm": 7.5912346839904785, "learning_rate": 1.4818816020746586e-05, "loss": 1.0451, "step": 6705 }, { "epoch": 1.08, "grad_norm": 8.301799774169922, "learning_rate": 1.4817293057045372e-05, "loss": 1.2253, "step": 6706 }, { "epoch": 1.08, "grad_norm": 9.352800369262695, "learning_rate": 1.4815769947832382e-05, "loss": 1.1534, "step": 6707 }, { "epoch": 1.08, "grad_norm": 9.11325740814209, "learning_rate": 1.4814246693153622e-05, "loss": 1.2074, "step": 6708 }, { "epoch": 1.08, "grad_norm": 9.594727516174316, "learning_rate": 1.4812723293055105e-05, "loss": 1.1897, "step": 6709 }, { "epoch": 1.08, "grad_norm": 9.712100982666016, "learning_rate": 1.4811199747582841e-05, "loss": 1.2836, "step": 6710 }, { "epoch": 1.08, "grad_norm": 10.400674819946289, "learning_rate": 1.4809676056782859e-05, "loss": 0.9486, "step": 6711 }, { "epoch": 1.08, "grad_norm": 8.026827812194824, "learning_rate": 1.4808152220701175e-05, "loss": 1.1448, "step": 6712 }, { "epoch": 1.08, "grad_norm": 8.807637214660645, "learning_rate": 1.4806628239383827e-05, "loss": 1.2468, "step": 6713 }, { "epoch": 1.08, "grad_norm": 8.872626304626465, "learning_rate": 1.4805104112876843e-05, "loss": 1.0662, "step": 6714 }, { "epoch": 1.08, "grad_norm": 10.29439926147461, "learning_rate": 1.480357984122626e-05, "loss": 1.3965, "step": 6715 }, { "epoch": 1.08, "grad_norm": 7.825582504272461, "learning_rate": 1.480205542447812e-05, "loss": 1.0517, "step": 6716 }, { "epoch": 1.08, "grad_norm": 9.273719787597656, "learning_rate": 1.480053086267848e-05, "loss": 1.1828, "step": 6717 }, { "epoch": 1.08, "grad_norm": 8.811699867248535, "learning_rate": 1.479900615587338e-05, "loss": 1.0292, "step": 6718 }, { "epoch": 1.08, "grad_norm": 9.040218353271484, "learning_rate": 1.4797481304108879e-05, "loss": 1.1404, "step": 6719 }, { "epoch": 1.08, "grad_norm": 8.481401443481445, "learning_rate": 1.4795956307431033e-05, "loss": 1.1387, "step": 6720 }, { "epoch": 1.08, "grad_norm": 9.016995429992676, "learning_rate": 1.4794431165885916e-05, "loss": 1.3028, "step": 6721 }, { "epoch": 1.08, "grad_norm": 8.587762832641602, "learning_rate": 1.4792905879519587e-05, "loss": 1.1907, "step": 6722 }, { "epoch": 1.08, "grad_norm": 7.50217342376709, "learning_rate": 1.4791380448378125e-05, "loss": 1.0756, "step": 6723 }, { "epoch": 1.08, "grad_norm": 8.729488372802734, "learning_rate": 1.4789854872507605e-05, "loss": 1.159, "step": 6724 }, { "epoch": 1.08, "grad_norm": 9.657990455627441, "learning_rate": 1.4788329151954109e-05, "loss": 1.0676, "step": 6725 }, { "epoch": 1.08, "grad_norm": 8.367527961730957, "learning_rate": 1.4786803286763718e-05, "loss": 1.3496, "step": 6726 }, { "epoch": 1.08, "grad_norm": 9.317319869995117, "learning_rate": 1.4785277276982531e-05, "loss": 1.2356, "step": 6727 }, { "epoch": 1.08, "grad_norm": 9.529886245727539, "learning_rate": 1.478375112265664e-05, "loss": 1.1153, "step": 6728 }, { "epoch": 1.08, "grad_norm": 7.440575122833252, "learning_rate": 1.4782224823832143e-05, "loss": 0.9994, "step": 6729 }, { "epoch": 1.08, "grad_norm": 7.739589214324951, "learning_rate": 1.4780698380555142e-05, "loss": 1.1418, "step": 6730 }, { "epoch": 1.08, "grad_norm": 8.566594123840332, "learning_rate": 1.477917179287175e-05, "loss": 1.2301, "step": 6731 }, { "epoch": 1.08, "grad_norm": 9.787210464477539, "learning_rate": 1.4777645060828074e-05, "loss": 1.3594, "step": 6732 }, { "epoch": 1.08, "grad_norm": 8.025842666625977, "learning_rate": 1.4776118184470237e-05, "loss": 1.2312, "step": 6733 }, { "epoch": 1.08, "grad_norm": 8.810503005981445, "learning_rate": 1.4774591163844349e-05, "loss": 1.1562, "step": 6734 }, { "epoch": 1.08, "grad_norm": 8.941957473754883, "learning_rate": 1.4773063998996546e-05, "loss": 1.2038, "step": 6735 }, { "epoch": 1.08, "grad_norm": 7.9635090827941895, "learning_rate": 1.4771536689972953e-05, "loss": 1.3951, "step": 6736 }, { "epoch": 1.08, "grad_norm": 7.420888900756836, "learning_rate": 1.4770009236819706e-05, "loss": 0.9091, "step": 6737 }, { "epoch": 1.08, "grad_norm": 7.905752658843994, "learning_rate": 1.4768481639582943e-05, "loss": 1.0334, "step": 6738 }, { "epoch": 1.08, "grad_norm": 8.760319709777832, "learning_rate": 1.4766953898308808e-05, "loss": 1.2899, "step": 6739 }, { "epoch": 1.08, "grad_norm": 9.199676513671875, "learning_rate": 1.4765426013043444e-05, "loss": 1.3265, "step": 6740 }, { "epoch": 1.09, "grad_norm": 7.121110439300537, "learning_rate": 1.4763897983833006e-05, "loss": 1.0487, "step": 6741 }, { "epoch": 1.09, "grad_norm": 9.66496467590332, "learning_rate": 1.476236981072365e-05, "loss": 1.1307, "step": 6742 }, { "epoch": 1.09, "grad_norm": 9.226988792419434, "learning_rate": 1.4760841493761532e-05, "loss": 1.094, "step": 6743 }, { "epoch": 1.09, "grad_norm": 9.047324180603027, "learning_rate": 1.4759313032992821e-05, "loss": 1.0505, "step": 6744 }, { "epoch": 1.09, "grad_norm": 8.50602912902832, "learning_rate": 1.4757784428463686e-05, "loss": 1.1079, "step": 6745 }, { "epoch": 1.09, "grad_norm": 8.225408554077148, "learning_rate": 1.4756255680220294e-05, "loss": 1.17, "step": 6746 }, { "epoch": 1.09, "grad_norm": 8.51710319519043, "learning_rate": 1.4754726788308833e-05, "loss": 1.0533, "step": 6747 }, { "epoch": 1.09, "grad_norm": 7.380760192871094, "learning_rate": 1.4753197752775478e-05, "loss": 0.8968, "step": 6748 }, { "epoch": 1.09, "grad_norm": 8.767114639282227, "learning_rate": 1.4751668573666415e-05, "loss": 0.871, "step": 6749 }, { "epoch": 1.09, "grad_norm": 8.581366539001465, "learning_rate": 1.4750139251027838e-05, "loss": 1.0147, "step": 6750 }, { "epoch": 1.09, "grad_norm": 8.089919090270996, "learning_rate": 1.4748609784905938e-05, "loss": 1.3249, "step": 6751 }, { "epoch": 1.09, "grad_norm": 9.598626136779785, "learning_rate": 1.474708017534692e-05, "loss": 0.9505, "step": 6752 }, { "epoch": 1.09, "grad_norm": 7.614184379577637, "learning_rate": 1.4745550422396979e-05, "loss": 0.823, "step": 6753 }, { "epoch": 1.09, "grad_norm": 9.605498313903809, "learning_rate": 1.4744020526102329e-05, "loss": 1.3995, "step": 6754 }, { "epoch": 1.09, "grad_norm": 8.895020484924316, "learning_rate": 1.4742490486509181e-05, "loss": 1.3622, "step": 6755 }, { "epoch": 1.09, "grad_norm": 7.467238903045654, "learning_rate": 1.4740960303663752e-05, "loss": 0.8893, "step": 6756 }, { "epoch": 1.09, "grad_norm": 8.823079109191895, "learning_rate": 1.4739429977612262e-05, "loss": 1.0322, "step": 6757 }, { "epoch": 1.09, "grad_norm": 8.093932151794434, "learning_rate": 1.4737899508400937e-05, "loss": 1.0417, "step": 6758 }, { "epoch": 1.09, "grad_norm": 7.391561031341553, "learning_rate": 1.4736368896076005e-05, "loss": 0.7463, "step": 6759 }, { "epoch": 1.09, "grad_norm": 7.990946292877197, "learning_rate": 1.4734838140683704e-05, "loss": 1.0599, "step": 6760 }, { "epoch": 1.09, "grad_norm": 9.314956665039062, "learning_rate": 1.4733307242270267e-05, "loss": 1.3639, "step": 6761 }, { "epoch": 1.09, "grad_norm": 7.734868049621582, "learning_rate": 1.473177620088194e-05, "loss": 0.7536, "step": 6762 }, { "epoch": 1.09, "grad_norm": 7.9001688957214355, "learning_rate": 1.4730245016564967e-05, "loss": 1.057, "step": 6763 }, { "epoch": 1.09, "grad_norm": 9.35260009765625, "learning_rate": 1.4728713689365605e-05, "loss": 1.2679, "step": 6764 }, { "epoch": 1.09, "grad_norm": 8.488699913024902, "learning_rate": 1.4727182219330102e-05, "loss": 0.9287, "step": 6765 }, { "epoch": 1.09, "grad_norm": 8.301129341125488, "learning_rate": 1.4725650606504722e-05, "loss": 0.9355, "step": 6766 }, { "epoch": 1.09, "grad_norm": 8.737679481506348, "learning_rate": 1.4724118850935726e-05, "loss": 1.0673, "step": 6767 }, { "epoch": 1.09, "grad_norm": 9.748590469360352, "learning_rate": 1.4722586952669384e-05, "loss": 1.0648, "step": 6768 }, { "epoch": 1.09, "grad_norm": 10.05780029296875, "learning_rate": 1.4721054911751973e-05, "loss": 1.3451, "step": 6769 }, { "epoch": 1.09, "grad_norm": 9.59890365600586, "learning_rate": 1.4719522728229763e-05, "loss": 1.1017, "step": 6770 }, { "epoch": 1.09, "grad_norm": 8.52358341217041, "learning_rate": 1.471799040214904e-05, "loss": 0.9698, "step": 6771 }, { "epoch": 1.09, "grad_norm": 9.360042572021484, "learning_rate": 1.471645793355609e-05, "loss": 0.8107, "step": 6772 }, { "epoch": 1.09, "grad_norm": 10.758094787597656, "learning_rate": 1.47149253224972e-05, "loss": 1.2898, "step": 6773 }, { "epoch": 1.09, "grad_norm": 7.842085838317871, "learning_rate": 1.4713392569018665e-05, "loss": 0.9175, "step": 6774 }, { "epoch": 1.09, "grad_norm": 9.769180297851562, "learning_rate": 1.4711859673166782e-05, "loss": 1.0292, "step": 6775 }, { "epoch": 1.09, "grad_norm": 8.134821891784668, "learning_rate": 1.471032663498786e-05, "loss": 0.9636, "step": 6776 }, { "epoch": 1.09, "grad_norm": 10.577072143554688, "learning_rate": 1.4708793454528198e-05, "loss": 1.3688, "step": 6777 }, { "epoch": 1.09, "grad_norm": 9.375459671020508, "learning_rate": 1.4707260131834113e-05, "loss": 1.0219, "step": 6778 }, { "epoch": 1.09, "grad_norm": 9.945821762084961, "learning_rate": 1.470572666695192e-05, "loss": 1.2144, "step": 6779 }, { "epoch": 1.09, "grad_norm": 8.185449600219727, "learning_rate": 1.4704193059927938e-05, "loss": 1.0261, "step": 6780 }, { "epoch": 1.09, "grad_norm": 8.294548034667969, "learning_rate": 1.4702659310808491e-05, "loss": 1.1552, "step": 6781 }, { "epoch": 1.09, "grad_norm": 10.530014991760254, "learning_rate": 1.470112541963991e-05, "loss": 1.0629, "step": 6782 }, { "epoch": 1.09, "grad_norm": 10.445577621459961, "learning_rate": 1.4699591386468524e-05, "loss": 1.1889, "step": 6783 }, { "epoch": 1.09, "grad_norm": 8.409510612487793, "learning_rate": 1.4698057211340673e-05, "loss": 0.8957, "step": 6784 }, { "epoch": 1.09, "grad_norm": 8.99513053894043, "learning_rate": 1.4696522894302697e-05, "loss": 0.9489, "step": 6785 }, { "epoch": 1.09, "grad_norm": 9.331947326660156, "learning_rate": 1.4694988435400945e-05, "loss": 1.0658, "step": 6786 }, { "epoch": 1.09, "grad_norm": 7.428487777709961, "learning_rate": 1.469345383468176e-05, "loss": 0.8637, "step": 6787 }, { "epoch": 1.09, "grad_norm": 12.035600662231445, "learning_rate": 1.4691919092191504e-05, "loss": 1.1411, "step": 6788 }, { "epoch": 1.09, "grad_norm": 9.094148635864258, "learning_rate": 1.4690384207976533e-05, "loss": 1.0455, "step": 6789 }, { "epoch": 1.09, "grad_norm": 7.406406879425049, "learning_rate": 1.468884918208321e-05, "loss": 1.0186, "step": 6790 }, { "epoch": 1.09, "grad_norm": 8.704253196716309, "learning_rate": 1.46873140145579e-05, "loss": 1.131, "step": 6791 }, { "epoch": 1.09, "grad_norm": 9.15419864654541, "learning_rate": 1.4685778705446978e-05, "loss": 0.9762, "step": 6792 }, { "epoch": 1.09, "grad_norm": 9.001925468444824, "learning_rate": 1.4684243254796818e-05, "loss": 1.0928, "step": 6793 }, { "epoch": 1.09, "grad_norm": 8.947901725769043, "learning_rate": 1.4682707662653797e-05, "loss": 0.9539, "step": 6794 }, { "epoch": 1.09, "grad_norm": 7.130845069885254, "learning_rate": 1.4681171929064307e-05, "loss": 0.835, "step": 6795 }, { "epoch": 1.09, "grad_norm": 8.048507690429688, "learning_rate": 1.4679636054074728e-05, "loss": 1.2127, "step": 6796 }, { "epoch": 1.09, "grad_norm": 10.658090591430664, "learning_rate": 1.4678100037731455e-05, "loss": 1.0427, "step": 6797 }, { "epoch": 1.09, "grad_norm": 9.319153785705566, "learning_rate": 1.467656388008089e-05, "loss": 1.1882, "step": 6798 }, { "epoch": 1.09, "grad_norm": 8.274429321289062, "learning_rate": 1.4675027581169432e-05, "loss": 1.075, "step": 6799 }, { "epoch": 1.09, "grad_norm": 8.629864692687988, "learning_rate": 1.4673491141043483e-05, "loss": 1.0025, "step": 6800 }, { "epoch": 1.09, "grad_norm": 9.698718070983887, "learning_rate": 1.4671954559749457e-05, "loss": 1.0476, "step": 6801 }, { "epoch": 1.09, "grad_norm": 9.703452110290527, "learning_rate": 1.4670417837333765e-05, "loss": 1.3588, "step": 6802 }, { "epoch": 1.1, "grad_norm": 9.278498649597168, "learning_rate": 1.4668880973842834e-05, "loss": 0.9407, "step": 6803 }, { "epoch": 1.1, "grad_norm": 8.61411190032959, "learning_rate": 1.4667343969323073e-05, "loss": 1.2738, "step": 6804 }, { "epoch": 1.1, "grad_norm": 8.826215744018555, "learning_rate": 1.4665806823820918e-05, "loss": 0.9992, "step": 6805 }, { "epoch": 1.1, "grad_norm": 10.618721008300781, "learning_rate": 1.4664269537382797e-05, "loss": 1.0536, "step": 6806 }, { "epoch": 1.1, "grad_norm": 8.183314323425293, "learning_rate": 1.4662732110055149e-05, "loss": 0.8443, "step": 6807 }, { "epoch": 1.1, "grad_norm": 10.026043891906738, "learning_rate": 1.4661194541884412e-05, "loss": 1.2406, "step": 6808 }, { "epoch": 1.1, "grad_norm": 8.118824005126953, "learning_rate": 1.4659656832917026e-05, "loss": 0.8339, "step": 6809 }, { "epoch": 1.1, "grad_norm": 9.279589653015137, "learning_rate": 1.4658118983199443e-05, "loss": 1.138, "step": 6810 }, { "epoch": 1.1, "grad_norm": 8.224940299987793, "learning_rate": 1.465658099277812e-05, "loss": 1.1906, "step": 6811 }, { "epoch": 1.1, "grad_norm": 9.987420082092285, "learning_rate": 1.4655042861699505e-05, "loss": 1.0823, "step": 6812 }, { "epoch": 1.1, "grad_norm": 8.166669845581055, "learning_rate": 1.4653504590010061e-05, "loss": 0.9499, "step": 6813 }, { "epoch": 1.1, "grad_norm": 8.132464408874512, "learning_rate": 1.4651966177756259e-05, "loss": 1.01, "step": 6814 }, { "epoch": 1.1, "grad_norm": 9.332783699035645, "learning_rate": 1.4650427624984565e-05, "loss": 1.247, "step": 6815 }, { "epoch": 1.1, "grad_norm": 10.810738563537598, "learning_rate": 1.4648888931741448e-05, "loss": 1.0141, "step": 6816 }, { "epoch": 1.1, "grad_norm": 8.577277183532715, "learning_rate": 1.464735009807339e-05, "loss": 1.0122, "step": 6817 }, { "epoch": 1.1, "grad_norm": 8.504525184631348, "learning_rate": 1.4645811124026877e-05, "loss": 1.2241, "step": 6818 }, { "epoch": 1.1, "grad_norm": 10.034585952758789, "learning_rate": 1.4644272009648391e-05, "loss": 1.0014, "step": 6819 }, { "epoch": 1.1, "grad_norm": 10.394457817077637, "learning_rate": 1.4642732754984424e-05, "loss": 1.3202, "step": 6820 }, { "epoch": 1.1, "grad_norm": 10.317615509033203, "learning_rate": 1.4641193360081467e-05, "loss": 1.0899, "step": 6821 }, { "epoch": 1.1, "grad_norm": 8.903745651245117, "learning_rate": 1.4639653824986026e-05, "loss": 1.0378, "step": 6822 }, { "epoch": 1.1, "grad_norm": 8.012726783752441, "learning_rate": 1.46381141497446e-05, "loss": 0.968, "step": 6823 }, { "epoch": 1.1, "grad_norm": 8.7350492477417, "learning_rate": 1.4636574334403698e-05, "loss": 1.1251, "step": 6824 }, { "epoch": 1.1, "grad_norm": 9.830194473266602, "learning_rate": 1.4635034379009831e-05, "loss": 1.1691, "step": 6825 }, { "epoch": 1.1, "grad_norm": 8.144437789916992, "learning_rate": 1.4633494283609518e-05, "loss": 1.1519, "step": 6826 }, { "epoch": 1.1, "grad_norm": 8.51024055480957, "learning_rate": 1.4631954048249272e-05, "loss": 1.0526, "step": 6827 }, { "epoch": 1.1, "grad_norm": 8.44455337524414, "learning_rate": 1.4630413672975627e-05, "loss": 1.0382, "step": 6828 }, { "epoch": 1.1, "grad_norm": 9.07689094543457, "learning_rate": 1.4628873157835106e-05, "loss": 1.0677, "step": 6829 }, { "epoch": 1.1, "grad_norm": 7.65854549407959, "learning_rate": 1.4627332502874242e-05, "loss": 0.8872, "step": 6830 }, { "epoch": 1.1, "grad_norm": 8.306804656982422, "learning_rate": 1.4625791708139576e-05, "loss": 1.3094, "step": 6831 }, { "epoch": 1.1, "grad_norm": 10.544929504394531, "learning_rate": 1.4624250773677646e-05, "loss": 1.0966, "step": 6832 }, { "epoch": 1.1, "grad_norm": 9.965076446533203, "learning_rate": 1.4622709699534998e-05, "loss": 1.061, "step": 6833 }, { "epoch": 1.1, "grad_norm": 8.633857727050781, "learning_rate": 1.4621168485758183e-05, "loss": 1.045, "step": 6834 }, { "epoch": 1.1, "grad_norm": 9.884926795959473, "learning_rate": 1.4619627132393754e-05, "loss": 1.2708, "step": 6835 }, { "epoch": 1.1, "grad_norm": 8.684552192687988, "learning_rate": 1.4618085639488271e-05, "loss": 1.1173, "step": 6836 }, { "epoch": 1.1, "grad_norm": 9.619451522827148, "learning_rate": 1.4616544007088292e-05, "loss": 0.934, "step": 6837 }, { "epoch": 1.1, "grad_norm": 10.663020133972168, "learning_rate": 1.4615002235240393e-05, "loss": 1.2308, "step": 6838 }, { "epoch": 1.1, "grad_norm": 8.678337097167969, "learning_rate": 1.4613460323991136e-05, "loss": 1.2378, "step": 6839 }, { "epoch": 1.1, "grad_norm": 9.535491943359375, "learning_rate": 1.4611918273387098e-05, "loss": 1.3116, "step": 6840 }, { "epoch": 1.1, "grad_norm": 7.4964189529418945, "learning_rate": 1.4610376083474859e-05, "loss": 0.949, "step": 6841 }, { "epoch": 1.1, "grad_norm": 8.216999053955078, "learning_rate": 1.4608833754301008e-05, "loss": 1.206, "step": 6842 }, { "epoch": 1.1, "grad_norm": 7.849591255187988, "learning_rate": 1.4607291285912126e-05, "loss": 1.1459, "step": 6843 }, { "epoch": 1.1, "grad_norm": 6.938811779022217, "learning_rate": 1.4605748678354806e-05, "loss": 0.8317, "step": 6844 }, { "epoch": 1.1, "grad_norm": 9.245490074157715, "learning_rate": 1.4604205931675645e-05, "loss": 0.8238, "step": 6845 }, { "epoch": 1.1, "grad_norm": 9.131637573242188, "learning_rate": 1.4602663045921246e-05, "loss": 0.9491, "step": 6846 }, { "epoch": 1.1, "grad_norm": 7.212952136993408, "learning_rate": 1.4601120021138208e-05, "loss": 0.9649, "step": 6847 }, { "epoch": 1.1, "grad_norm": 7.153547763824463, "learning_rate": 1.4599576857373144e-05, "loss": 1.1321, "step": 6848 }, { "epoch": 1.1, "grad_norm": 9.016457557678223, "learning_rate": 1.4598033554672666e-05, "loss": 1.2999, "step": 6849 }, { "epoch": 1.1, "grad_norm": 8.402009010314941, "learning_rate": 1.4596490113083395e-05, "loss": 1.2003, "step": 6850 }, { "epoch": 1.1, "grad_norm": 10.276604652404785, "learning_rate": 1.4594946532651946e-05, "loss": 1.1306, "step": 6851 }, { "epoch": 1.1, "grad_norm": 10.254349708557129, "learning_rate": 1.4593402813424947e-05, "loss": 1.2366, "step": 6852 }, { "epoch": 1.1, "grad_norm": 9.326251983642578, "learning_rate": 1.4591858955449028e-05, "loss": 1.5047, "step": 6853 }, { "epoch": 1.1, "grad_norm": 8.747230529785156, "learning_rate": 1.4590314958770826e-05, "loss": 1.2102, "step": 6854 }, { "epoch": 1.1, "grad_norm": 7.872401714324951, "learning_rate": 1.4588770823436972e-05, "loss": 0.9667, "step": 6855 }, { "epoch": 1.1, "grad_norm": 8.14645767211914, "learning_rate": 1.4587226549494116e-05, "loss": 0.9927, "step": 6856 }, { "epoch": 1.1, "grad_norm": 10.289106369018555, "learning_rate": 1.45856821369889e-05, "loss": 0.8462, "step": 6857 }, { "epoch": 1.1, "grad_norm": 7.236433506011963, "learning_rate": 1.4584137585967972e-05, "loss": 1.0065, "step": 6858 }, { "epoch": 1.1, "grad_norm": 10.852615356445312, "learning_rate": 1.4582592896477994e-05, "loss": 1.1232, "step": 6859 }, { "epoch": 1.1, "grad_norm": 9.586128234863281, "learning_rate": 1.4581048068565622e-05, "loss": 1.0657, "step": 6860 }, { "epoch": 1.1, "grad_norm": 8.641332626342773, "learning_rate": 1.4579503102277519e-05, "loss": 1.0328, "step": 6861 }, { "epoch": 1.1, "grad_norm": 8.04345417022705, "learning_rate": 1.4577957997660355e-05, "loss": 0.8545, "step": 6862 }, { "epoch": 1.1, "grad_norm": 9.583654403686523, "learning_rate": 1.4576412754760796e-05, "loss": 0.9683, "step": 6863 }, { "epoch": 1.1, "grad_norm": 9.34390640258789, "learning_rate": 1.4574867373625521e-05, "loss": 1.0503, "step": 6864 }, { "epoch": 1.11, "grad_norm": 8.247819900512695, "learning_rate": 1.457332185430121e-05, "loss": 1.1815, "step": 6865 }, { "epoch": 1.11, "grad_norm": 9.632878303527832, "learning_rate": 1.457177619683455e-05, "loss": 1.2429, "step": 6866 }, { "epoch": 1.11, "grad_norm": 8.700615882873535, "learning_rate": 1.4570230401272225e-05, "loss": 1.069, "step": 6867 }, { "epoch": 1.11, "grad_norm": 11.247464179992676, "learning_rate": 1.4568684467660925e-05, "loss": 1.1259, "step": 6868 }, { "epoch": 1.11, "grad_norm": 8.810776710510254, "learning_rate": 1.4567138396047355e-05, "loss": 1.0723, "step": 6869 }, { "epoch": 1.11, "grad_norm": 8.925978660583496, "learning_rate": 1.4565592186478213e-05, "loss": 1.2133, "step": 6870 }, { "epoch": 1.11, "grad_norm": 7.708931922912598, "learning_rate": 1.4564045839000198e-05, "loss": 1.044, "step": 6871 }, { "epoch": 1.11, "grad_norm": 7.978443622589111, "learning_rate": 1.4562499353660024e-05, "loss": 0.8616, "step": 6872 }, { "epoch": 1.11, "grad_norm": 8.116116523742676, "learning_rate": 1.456095273050441e-05, "loss": 0.9766, "step": 6873 }, { "epoch": 1.11, "grad_norm": 9.857054710388184, "learning_rate": 1.455940596958006e-05, "loss": 1.0005, "step": 6874 }, { "epoch": 1.11, "grad_norm": 9.444952011108398, "learning_rate": 1.4557859070933706e-05, "loss": 1.2066, "step": 6875 }, { "epoch": 1.11, "grad_norm": 8.489911079406738, "learning_rate": 1.4556312034612073e-05, "loss": 0.875, "step": 6876 }, { "epoch": 1.11, "grad_norm": 9.021191596984863, "learning_rate": 1.4554764860661885e-05, "loss": 0.984, "step": 6877 }, { "epoch": 1.11, "grad_norm": 8.651473045349121, "learning_rate": 1.4553217549129884e-05, "loss": 1.0734, "step": 6878 }, { "epoch": 1.11, "grad_norm": 8.740994453430176, "learning_rate": 1.4551670100062802e-05, "loss": 1.0891, "step": 6879 }, { "epoch": 1.11, "grad_norm": 8.185216903686523, "learning_rate": 1.4550122513507383e-05, "loss": 1.1257, "step": 6880 }, { "epoch": 1.11, "grad_norm": 8.226668357849121, "learning_rate": 1.4548574789510376e-05, "loss": 0.9932, "step": 6881 }, { "epoch": 1.11, "grad_norm": 9.164557456970215, "learning_rate": 1.4547026928118533e-05, "loss": 1.2708, "step": 6882 }, { "epoch": 1.11, "grad_norm": 10.951507568359375, "learning_rate": 1.4545478929378602e-05, "loss": 1.4195, "step": 6883 }, { "epoch": 1.11, "grad_norm": 8.299050331115723, "learning_rate": 1.4543930793337347e-05, "loss": 1.1702, "step": 6884 }, { "epoch": 1.11, "grad_norm": 8.832427978515625, "learning_rate": 1.4542382520041533e-05, "loss": 1.1447, "step": 6885 }, { "epoch": 1.11, "grad_norm": 8.347783088684082, "learning_rate": 1.4540834109537922e-05, "loss": 1.0887, "step": 6886 }, { "epoch": 1.11, "grad_norm": 10.141631126403809, "learning_rate": 1.4539285561873287e-05, "loss": 1.1572, "step": 6887 }, { "epoch": 1.11, "grad_norm": 9.411890029907227, "learning_rate": 1.4537736877094404e-05, "loss": 1.1597, "step": 6888 }, { "epoch": 1.11, "grad_norm": 9.270453453063965, "learning_rate": 1.4536188055248056e-05, "loss": 1.0552, "step": 6889 }, { "epoch": 1.11, "grad_norm": 8.172608375549316, "learning_rate": 1.4534639096381028e-05, "loss": 1.1388, "step": 6890 }, { "epoch": 1.11, "grad_norm": 10.524155616760254, "learning_rate": 1.4533090000540098e-05, "loss": 1.1895, "step": 6891 }, { "epoch": 1.11, "grad_norm": 9.71420955657959, "learning_rate": 1.4531540767772067e-05, "loss": 1.2657, "step": 6892 }, { "epoch": 1.11, "grad_norm": 8.598657608032227, "learning_rate": 1.4529991398123732e-05, "loss": 1.0306, "step": 6893 }, { "epoch": 1.11, "grad_norm": 8.688070297241211, "learning_rate": 1.4528441891641888e-05, "loss": 1.2487, "step": 6894 }, { "epoch": 1.11, "grad_norm": 7.396509170532227, "learning_rate": 1.4526892248373341e-05, "loss": 1.243, "step": 6895 }, { "epoch": 1.11, "grad_norm": 9.599038124084473, "learning_rate": 1.4525342468364903e-05, "loss": 0.9292, "step": 6896 }, { "epoch": 1.11, "grad_norm": 9.194232940673828, "learning_rate": 1.4523792551663385e-05, "loss": 1.2411, "step": 6897 }, { "epoch": 1.11, "grad_norm": 10.44398307800293, "learning_rate": 1.4522242498315601e-05, "loss": 1.1079, "step": 6898 }, { "epoch": 1.11, "grad_norm": 7.70261812210083, "learning_rate": 1.4520692308368377e-05, "loss": 0.77, "step": 6899 }, { "epoch": 1.11, "grad_norm": 10.603708267211914, "learning_rate": 1.4519141981868534e-05, "loss": 1.1277, "step": 6900 }, { "epoch": 1.11, "grad_norm": 11.870927810668945, "learning_rate": 1.4517591518862905e-05, "loss": 1.1079, "step": 6901 }, { "epoch": 1.11, "grad_norm": 9.006753921508789, "learning_rate": 1.4516040919398322e-05, "loss": 1.0632, "step": 6902 }, { "epoch": 1.11, "grad_norm": 10.932101249694824, "learning_rate": 1.4514490183521622e-05, "loss": 1.139, "step": 6903 }, { "epoch": 1.11, "grad_norm": 10.492619514465332, "learning_rate": 1.4512939311279647e-05, "loss": 1.063, "step": 6904 }, { "epoch": 1.11, "grad_norm": 7.560225009918213, "learning_rate": 1.4511388302719245e-05, "loss": 1.009, "step": 6905 }, { "epoch": 1.11, "grad_norm": 8.716146469116211, "learning_rate": 1.450983715788726e-05, "loss": 1.1172, "step": 6906 }, { "epoch": 1.11, "grad_norm": 9.844162940979004, "learning_rate": 1.4508285876830552e-05, "loss": 1.0618, "step": 6907 }, { "epoch": 1.11, "grad_norm": 9.20811939239502, "learning_rate": 1.4506734459595978e-05, "loss": 1.2154, "step": 6908 }, { "epoch": 1.11, "grad_norm": 8.365092277526855, "learning_rate": 1.45051829062304e-05, "loss": 1.2005, "step": 6909 }, { "epoch": 1.11, "grad_norm": 8.974395751953125, "learning_rate": 1.4503631216780682e-05, "loss": 1.2461, "step": 6910 }, { "epoch": 1.11, "grad_norm": 8.100406646728516, "learning_rate": 1.4502079391293698e-05, "loss": 1.1325, "step": 6911 }, { "epoch": 1.11, "grad_norm": 8.61938190460205, "learning_rate": 1.4500527429816322e-05, "loss": 1.2527, "step": 6912 }, { "epoch": 1.11, "grad_norm": 8.224977493286133, "learning_rate": 1.4498975332395433e-05, "loss": 1.0583, "step": 6913 }, { "epoch": 1.11, "grad_norm": 8.105415344238281, "learning_rate": 1.449742309907791e-05, "loss": 0.8214, "step": 6914 }, { "epoch": 1.11, "grad_norm": 9.516505241394043, "learning_rate": 1.4495870729910641e-05, "loss": 1.1338, "step": 6915 }, { "epoch": 1.11, "grad_norm": 8.665658950805664, "learning_rate": 1.4494318224940523e-05, "loss": 1.2749, "step": 6916 }, { "epoch": 1.11, "grad_norm": 9.541457176208496, "learning_rate": 1.4492765584214445e-05, "loss": 1.0755, "step": 6917 }, { "epoch": 1.11, "grad_norm": 7.769615650177002, "learning_rate": 1.4491212807779312e-05, "loss": 1.1372, "step": 6918 }, { "epoch": 1.11, "grad_norm": 10.563132286071777, "learning_rate": 1.4489659895682016e-05, "loss": 0.9522, "step": 6919 }, { "epoch": 1.11, "grad_norm": 9.393345832824707, "learning_rate": 1.4488106847969479e-05, "loss": 1.0581, "step": 6920 }, { "epoch": 1.11, "grad_norm": 10.720019340515137, "learning_rate": 1.4486553664688603e-05, "loss": 1.1911, "step": 6921 }, { "epoch": 1.11, "grad_norm": 9.070898056030273, "learning_rate": 1.4485000345886305e-05, "loss": 1.0273, "step": 6922 }, { "epoch": 1.11, "grad_norm": 9.04123592376709, "learning_rate": 1.4483446891609508e-05, "loss": 1.0327, "step": 6923 }, { "epoch": 1.11, "grad_norm": 9.056990623474121, "learning_rate": 1.4481893301905135e-05, "loss": 1.1299, "step": 6924 }, { "epoch": 1.11, "grad_norm": 8.829272270202637, "learning_rate": 1.4480339576820109e-05, "loss": 1.2052, "step": 6925 }, { "epoch": 1.11, "grad_norm": 10.103577613830566, "learning_rate": 1.447878571640137e-05, "loss": 1.1284, "step": 6926 }, { "epoch": 1.12, "grad_norm": 10.942769050598145, "learning_rate": 1.4477231720695846e-05, "loss": 0.9676, "step": 6927 }, { "epoch": 1.12, "grad_norm": 10.85904312133789, "learning_rate": 1.4475677589750482e-05, "loss": 0.9965, "step": 6928 }, { "epoch": 1.12, "grad_norm": 9.314205169677734, "learning_rate": 1.4474123323612221e-05, "loss": 1.1516, "step": 6929 }, { "epoch": 1.12, "grad_norm": 8.853012084960938, "learning_rate": 1.4472568922328013e-05, "loss": 0.9275, "step": 6930 }, { "epoch": 1.12, "grad_norm": 8.879229545593262, "learning_rate": 1.447101438594481e-05, "loss": 1.2269, "step": 6931 }, { "epoch": 1.12, "grad_norm": 7.955592632293701, "learning_rate": 1.4469459714509571e-05, "loss": 0.7724, "step": 6932 }, { "epoch": 1.12, "grad_norm": 10.232807159423828, "learning_rate": 1.4467904908069249e-05, "loss": 1.1652, "step": 6933 }, { "epoch": 1.12, "grad_norm": 8.641507148742676, "learning_rate": 1.4466349966670811e-05, "loss": 1.0764, "step": 6934 }, { "epoch": 1.12, "grad_norm": 7.369453430175781, "learning_rate": 1.4464794890361231e-05, "loss": 0.8545, "step": 6935 }, { "epoch": 1.12, "grad_norm": 8.880725860595703, "learning_rate": 1.446323967918748e-05, "loss": 1.0925, "step": 6936 }, { "epoch": 1.12, "grad_norm": 9.599390029907227, "learning_rate": 1.4461684333196533e-05, "loss": 1.0886, "step": 6937 }, { "epoch": 1.12, "grad_norm": 8.540738105773926, "learning_rate": 1.446012885243537e-05, "loss": 1.1314, "step": 6938 }, { "epoch": 1.12, "grad_norm": 7.9197187423706055, "learning_rate": 1.4458573236950974e-05, "loss": 1.2054, "step": 6939 }, { "epoch": 1.12, "grad_norm": 10.487117767333984, "learning_rate": 1.4457017486790345e-05, "loss": 1.227, "step": 6940 }, { "epoch": 1.12, "grad_norm": 9.991941452026367, "learning_rate": 1.4455461602000465e-05, "loss": 0.9153, "step": 6941 }, { "epoch": 1.12, "grad_norm": 10.437474250793457, "learning_rate": 1.4453905582628336e-05, "loss": 1.3077, "step": 6942 }, { "epoch": 1.12, "grad_norm": 8.104569435119629, "learning_rate": 1.4452349428720957e-05, "loss": 0.857, "step": 6943 }, { "epoch": 1.12, "grad_norm": 10.240492820739746, "learning_rate": 1.445079314032534e-05, "loss": 0.9709, "step": 6944 }, { "epoch": 1.12, "grad_norm": 8.598649024963379, "learning_rate": 1.4449236717488485e-05, "loss": 1.0491, "step": 6945 }, { "epoch": 1.12, "grad_norm": 8.739226341247559, "learning_rate": 1.4447680160257406e-05, "loss": 1.1417, "step": 6946 }, { "epoch": 1.12, "grad_norm": 11.015270233154297, "learning_rate": 1.444612346867913e-05, "loss": 0.9847, "step": 6947 }, { "epoch": 1.12, "grad_norm": 9.130002975463867, "learning_rate": 1.4444566642800675e-05, "loss": 1.2591, "step": 6948 }, { "epoch": 1.12, "grad_norm": 7.777892112731934, "learning_rate": 1.4443009682669058e-05, "loss": 0.8383, "step": 6949 }, { "epoch": 1.12, "grad_norm": 8.372814178466797, "learning_rate": 1.4441452588331319e-05, "loss": 1.03, "step": 6950 }, { "epoch": 1.12, "grad_norm": 7.43764066696167, "learning_rate": 1.443989535983449e-05, "loss": 0.6868, "step": 6951 }, { "epoch": 1.12, "grad_norm": 11.231075286865234, "learning_rate": 1.4438337997225604e-05, "loss": 1.1665, "step": 6952 }, { "epoch": 1.12, "grad_norm": 9.576780319213867, "learning_rate": 1.4436780500551709e-05, "loss": 1.1988, "step": 6953 }, { "epoch": 1.12, "grad_norm": 7.26950216293335, "learning_rate": 1.4435222869859847e-05, "loss": 1.0999, "step": 6954 }, { "epoch": 1.12, "grad_norm": 9.357141494750977, "learning_rate": 1.4433665105197065e-05, "loss": 1.2034, "step": 6955 }, { "epoch": 1.12, "grad_norm": 10.990184783935547, "learning_rate": 1.4432107206610426e-05, "loss": 0.8962, "step": 6956 }, { "epoch": 1.12, "grad_norm": 8.299980163574219, "learning_rate": 1.4430549174146981e-05, "loss": 0.9125, "step": 6957 }, { "epoch": 1.12, "grad_norm": 9.157110214233398, "learning_rate": 1.4428991007853793e-05, "loss": 1.0229, "step": 6958 }, { "epoch": 1.12, "grad_norm": 9.044456481933594, "learning_rate": 1.4427432707777933e-05, "loss": 1.116, "step": 6959 }, { "epoch": 1.12, "grad_norm": 10.0826416015625, "learning_rate": 1.4425874273966465e-05, "loss": 1.1765, "step": 6960 }, { "epoch": 1.12, "grad_norm": 9.325160026550293, "learning_rate": 1.4424315706466465e-05, "loss": 1.0218, "step": 6961 }, { "epoch": 1.12, "grad_norm": 9.91696834564209, "learning_rate": 1.4422757005325014e-05, "loss": 1.1438, "step": 6962 }, { "epoch": 1.12, "grad_norm": 9.103812217712402, "learning_rate": 1.4421198170589193e-05, "loss": 1.0423, "step": 6963 }, { "epoch": 1.12, "grad_norm": 10.845067024230957, "learning_rate": 1.4419639202306086e-05, "loss": 1.5152, "step": 6964 }, { "epoch": 1.12, "grad_norm": 10.107439994812012, "learning_rate": 1.4418080100522787e-05, "loss": 1.148, "step": 6965 }, { "epoch": 1.12, "grad_norm": 9.235445022583008, "learning_rate": 1.4416520865286385e-05, "loss": 1.4352, "step": 6966 }, { "epoch": 1.12, "grad_norm": 9.253796577453613, "learning_rate": 1.4414961496643987e-05, "loss": 1.1028, "step": 6967 }, { "epoch": 1.12, "grad_norm": 9.374094009399414, "learning_rate": 1.4413401994642688e-05, "loss": 1.2152, "step": 6968 }, { "epoch": 1.12, "grad_norm": 9.455805778503418, "learning_rate": 1.4411842359329596e-05, "loss": 1.0938, "step": 6969 }, { "epoch": 1.12, "grad_norm": 8.711380004882812, "learning_rate": 1.4410282590751824e-05, "loss": 1.0648, "step": 6970 }, { "epoch": 1.12, "grad_norm": 9.808292388916016, "learning_rate": 1.4408722688956487e-05, "loss": 1.1216, "step": 6971 }, { "epoch": 1.12, "grad_norm": 8.300434112548828, "learning_rate": 1.44071626539907e-05, "loss": 0.8562, "step": 6972 }, { "epoch": 1.12, "grad_norm": 8.905372619628906, "learning_rate": 1.4405602485901587e-05, "loss": 1.0511, "step": 6973 }, { "epoch": 1.12, "grad_norm": 8.74999713897705, "learning_rate": 1.4404042184736276e-05, "loss": 0.9461, "step": 6974 }, { "epoch": 1.12, "grad_norm": 7.117004871368408, "learning_rate": 1.4402481750541898e-05, "loss": 0.988, "step": 6975 }, { "epoch": 1.12, "grad_norm": 7.399832248687744, "learning_rate": 1.4400921183365584e-05, "loss": 0.9632, "step": 6976 }, { "epoch": 1.12, "grad_norm": 9.671278953552246, "learning_rate": 1.4399360483254477e-05, "loss": 1.1343, "step": 6977 }, { "epoch": 1.12, "grad_norm": 9.336069107055664, "learning_rate": 1.4397799650255715e-05, "loss": 1.0411, "step": 6978 }, { "epoch": 1.12, "grad_norm": 9.540372848510742, "learning_rate": 1.4396238684416448e-05, "loss": 1.2101, "step": 6979 }, { "epoch": 1.12, "grad_norm": 10.545689582824707, "learning_rate": 1.4394677585783827e-05, "loss": 1.2454, "step": 6980 }, { "epoch": 1.12, "grad_norm": 8.738402366638184, "learning_rate": 1.4393116354405007e-05, "loss": 1.2727, "step": 6981 }, { "epoch": 1.12, "grad_norm": 9.363627433776855, "learning_rate": 1.4391554990327146e-05, "loss": 1.1754, "step": 6982 }, { "epoch": 1.12, "grad_norm": 10.478337287902832, "learning_rate": 1.4389993493597407e-05, "loss": 1.2939, "step": 6983 }, { "epoch": 1.12, "grad_norm": 9.746554374694824, "learning_rate": 1.4388431864262955e-05, "loss": 0.7991, "step": 6984 }, { "epoch": 1.12, "grad_norm": 9.626298904418945, "learning_rate": 1.4386870102370963e-05, "loss": 1.1182, "step": 6985 }, { "epoch": 1.12, "grad_norm": 7.718388557434082, "learning_rate": 1.4385308207968604e-05, "loss": 1.1787, "step": 6986 }, { "epoch": 1.12, "grad_norm": 8.475801467895508, "learning_rate": 1.4383746181103059e-05, "loss": 1.1371, "step": 6987 }, { "epoch": 1.12, "grad_norm": 8.594618797302246, "learning_rate": 1.4382184021821507e-05, "loss": 1.0201, "step": 6988 }, { "epoch": 1.13, "grad_norm": 8.993834495544434, "learning_rate": 1.4380621730171139e-05, "loss": 1.1585, "step": 6989 }, { "epoch": 1.13, "grad_norm": 11.416581153869629, "learning_rate": 1.4379059306199143e-05, "loss": 1.2946, "step": 6990 }, { "epoch": 1.13, "grad_norm": 9.813149452209473, "learning_rate": 1.4377496749952717e-05, "loss": 1.2528, "step": 6991 }, { "epoch": 1.13, "grad_norm": 8.922343254089355, "learning_rate": 1.4375934061479056e-05, "loss": 0.9555, "step": 6992 }, { "epoch": 1.13, "grad_norm": 8.80932903289795, "learning_rate": 1.4374371240825365e-05, "loss": 1.0282, "step": 6993 }, { "epoch": 1.13, "grad_norm": 8.778782844543457, "learning_rate": 1.4372808288038849e-05, "loss": 1.1922, "step": 6994 }, { "epoch": 1.13, "grad_norm": 8.379358291625977, "learning_rate": 1.4371245203166721e-05, "loss": 1.238, "step": 6995 }, { "epoch": 1.13, "grad_norm": 9.365262985229492, "learning_rate": 1.4369681986256194e-05, "loss": 1.0662, "step": 6996 }, { "epoch": 1.13, "grad_norm": 8.985633850097656, "learning_rate": 1.4368118637354486e-05, "loss": 1.1204, "step": 6997 }, { "epoch": 1.13, "grad_norm": 8.817281723022461, "learning_rate": 1.4366555156508823e-05, "loss": 1.0093, "step": 6998 }, { "epoch": 1.13, "grad_norm": 10.489444732666016, "learning_rate": 1.4364991543766432e-05, "loss": 1.1829, "step": 6999 }, { "epoch": 1.13, "grad_norm": 8.370086669921875, "learning_rate": 1.4363427799174536e-05, "loss": 1.0212, "step": 7000 }, { "epoch": 1.13, "grad_norm": 8.531774520874023, "learning_rate": 1.4361863922780379e-05, "loss": 0.7907, "step": 7001 }, { "epoch": 1.13, "grad_norm": 9.379974365234375, "learning_rate": 1.4360299914631195e-05, "loss": 1.2782, "step": 7002 }, { "epoch": 1.13, "grad_norm": 8.14751148223877, "learning_rate": 1.4358735774774228e-05, "loss": 0.8384, "step": 7003 }, { "epoch": 1.13, "grad_norm": 9.558321952819824, "learning_rate": 1.4357171503256722e-05, "loss": 1.1324, "step": 7004 }, { "epoch": 1.13, "grad_norm": 7.651862144470215, "learning_rate": 1.435560710012593e-05, "loss": 1.0687, "step": 7005 }, { "epoch": 1.13, "grad_norm": 12.751455307006836, "learning_rate": 1.4354042565429107e-05, "loss": 1.1329, "step": 7006 }, { "epoch": 1.13, "grad_norm": 8.022077560424805, "learning_rate": 1.4352477899213513e-05, "loss": 1.0316, "step": 7007 }, { "epoch": 1.13, "grad_norm": 8.449755668640137, "learning_rate": 1.4350913101526405e-05, "loss": 1.1469, "step": 7008 }, { "epoch": 1.13, "grad_norm": 8.490344047546387, "learning_rate": 1.4349348172415054e-05, "loss": 1.0266, "step": 7009 }, { "epoch": 1.13, "grad_norm": 7.662446022033691, "learning_rate": 1.4347783111926727e-05, "loss": 0.9439, "step": 7010 }, { "epoch": 1.13, "grad_norm": 10.369890213012695, "learning_rate": 1.4346217920108702e-05, "loss": 1.3667, "step": 7011 }, { "epoch": 1.13, "grad_norm": 6.871692657470703, "learning_rate": 1.4344652597008258e-05, "loss": 1.095, "step": 7012 }, { "epoch": 1.13, "grad_norm": 8.7824068069458, "learning_rate": 1.4343087142672674e-05, "loss": 1.12, "step": 7013 }, { "epoch": 1.13, "grad_norm": 7.72653341293335, "learning_rate": 1.4341521557149239e-05, "loss": 0.9861, "step": 7014 }, { "epoch": 1.13, "grad_norm": 9.140451431274414, "learning_rate": 1.4339955840485239e-05, "loss": 1.1094, "step": 7015 }, { "epoch": 1.13, "grad_norm": 8.990036010742188, "learning_rate": 1.4338389992727972e-05, "loss": 1.2573, "step": 7016 }, { "epoch": 1.13, "grad_norm": 9.747600555419922, "learning_rate": 1.4336824013924736e-05, "loss": 1.2643, "step": 7017 }, { "epoch": 1.13, "grad_norm": 10.713885307312012, "learning_rate": 1.4335257904122833e-05, "loss": 1.0434, "step": 7018 }, { "epoch": 1.13, "grad_norm": 7.857950687408447, "learning_rate": 1.4333691663369569e-05, "loss": 1.0677, "step": 7019 }, { "epoch": 1.13, "grad_norm": 11.015944480895996, "learning_rate": 1.433212529171225e-05, "loss": 1.0397, "step": 7020 }, { "epoch": 1.13, "grad_norm": 7.279433727264404, "learning_rate": 1.4330558789198196e-05, "loss": 1.1554, "step": 7021 }, { "epoch": 1.13, "grad_norm": 9.815882682800293, "learning_rate": 1.4328992155874724e-05, "loss": 1.1737, "step": 7022 }, { "epoch": 1.13, "grad_norm": 8.973029136657715, "learning_rate": 1.4327425391789153e-05, "loss": 1.2037, "step": 7023 }, { "epoch": 1.13, "grad_norm": 10.777312278747559, "learning_rate": 1.432585849698881e-05, "loss": 1.2157, "step": 7024 }, { "epoch": 1.13, "grad_norm": 8.1882905960083, "learning_rate": 1.4324291471521027e-05, "loss": 1.1457, "step": 7025 }, { "epoch": 1.13, "grad_norm": 8.25704288482666, "learning_rate": 1.4322724315433138e-05, "loss": 0.9026, "step": 7026 }, { "epoch": 1.13, "grad_norm": 8.083586692810059, "learning_rate": 1.4321157028772476e-05, "loss": 0.7806, "step": 7027 }, { "epoch": 1.13, "grad_norm": 9.736982345581055, "learning_rate": 1.4319589611586386e-05, "loss": 0.7509, "step": 7028 }, { "epoch": 1.13, "grad_norm": 8.818502426147461, "learning_rate": 1.4318022063922215e-05, "loss": 1.0084, "step": 7029 }, { "epoch": 1.13, "grad_norm": 8.032567977905273, "learning_rate": 1.4316454385827307e-05, "loss": 0.8613, "step": 7030 }, { "epoch": 1.13, "grad_norm": 9.380465507507324, "learning_rate": 1.4314886577349021e-05, "loss": 1.0944, "step": 7031 }, { "epoch": 1.13, "grad_norm": 9.033512115478516, "learning_rate": 1.4313318638534711e-05, "loss": 1.3986, "step": 7032 }, { "epoch": 1.13, "grad_norm": 7.876359462738037, "learning_rate": 1.4311750569431742e-05, "loss": 1.3225, "step": 7033 }, { "epoch": 1.13, "grad_norm": 8.463457107543945, "learning_rate": 1.4310182370087478e-05, "loss": 0.9801, "step": 7034 }, { "epoch": 1.13, "grad_norm": 9.233412742614746, "learning_rate": 1.4308614040549285e-05, "loss": 1.0783, "step": 7035 }, { "epoch": 1.13, "grad_norm": 7.940191745758057, "learning_rate": 1.430704558086454e-05, "loss": 0.8868, "step": 7036 }, { "epoch": 1.13, "grad_norm": 7.818748950958252, "learning_rate": 1.4305476991080619e-05, "loss": 1.0328, "step": 7037 }, { "epoch": 1.13, "grad_norm": 9.634127616882324, "learning_rate": 1.4303908271244904e-05, "loss": 1.006, "step": 7038 }, { "epoch": 1.13, "grad_norm": 7.644046306610107, "learning_rate": 1.4302339421404777e-05, "loss": 1.0547, "step": 7039 }, { "epoch": 1.13, "grad_norm": 11.156831741333008, "learning_rate": 1.4300770441607625e-05, "loss": 1.0279, "step": 7040 }, { "epoch": 1.13, "grad_norm": 11.190803527832031, "learning_rate": 1.429920133190085e-05, "loss": 1.0248, "step": 7041 }, { "epoch": 1.13, "grad_norm": 8.648496627807617, "learning_rate": 1.429763209233184e-05, "loss": 0.959, "step": 7042 }, { "epoch": 1.13, "grad_norm": 7.231069087982178, "learning_rate": 1.4296062722947999e-05, "loss": 1.0555, "step": 7043 }, { "epoch": 1.13, "grad_norm": 8.262846946716309, "learning_rate": 1.4294493223796733e-05, "loss": 0.9943, "step": 7044 }, { "epoch": 1.13, "grad_norm": 11.186288833618164, "learning_rate": 1.4292923594925447e-05, "loss": 1.1056, "step": 7045 }, { "epoch": 1.13, "grad_norm": 8.753095626831055, "learning_rate": 1.4291353836381559e-05, "loss": 1.0627, "step": 7046 }, { "epoch": 1.13, "grad_norm": 9.234111785888672, "learning_rate": 1.4289783948212476e-05, "loss": 1.1597, "step": 7047 }, { "epoch": 1.13, "grad_norm": 9.817110061645508, "learning_rate": 1.4288213930465629e-05, "loss": 1.0807, "step": 7048 }, { "epoch": 1.13, "grad_norm": 9.919234275817871, "learning_rate": 1.4286643783188434e-05, "loss": 1.1505, "step": 7049 }, { "epoch": 1.13, "grad_norm": 9.008974075317383, "learning_rate": 1.4285073506428322e-05, "loss": 1.194, "step": 7050 }, { "epoch": 1.14, "grad_norm": 10.918331146240234, "learning_rate": 1.4283503100232725e-05, "loss": 1.0973, "step": 7051 }, { "epoch": 1.14, "grad_norm": 8.1054105758667, "learning_rate": 1.4281932564649078e-05, "loss": 1.0092, "step": 7052 }, { "epoch": 1.14, "grad_norm": 8.367398262023926, "learning_rate": 1.4280361899724827e-05, "loss": 1.1952, "step": 7053 }, { "epoch": 1.14, "grad_norm": 11.612688064575195, "learning_rate": 1.4278791105507406e-05, "loss": 1.1596, "step": 7054 }, { "epoch": 1.14, "grad_norm": 8.824880599975586, "learning_rate": 1.4277220182044268e-05, "loss": 0.7993, "step": 7055 }, { "epoch": 1.14, "grad_norm": 9.205427169799805, "learning_rate": 1.4275649129382863e-05, "loss": 1.081, "step": 7056 }, { "epoch": 1.14, "grad_norm": 9.339435577392578, "learning_rate": 1.4274077947570649e-05, "loss": 0.8461, "step": 7057 }, { "epoch": 1.14, "grad_norm": 8.742775917053223, "learning_rate": 1.4272506636655082e-05, "loss": 0.9938, "step": 7058 }, { "epoch": 1.14, "grad_norm": 8.95098876953125, "learning_rate": 1.4270935196683627e-05, "loss": 0.9096, "step": 7059 }, { "epoch": 1.14, "grad_norm": 9.549283981323242, "learning_rate": 1.4269363627703747e-05, "loss": 1.0723, "step": 7060 }, { "epoch": 1.14, "grad_norm": 8.337060928344727, "learning_rate": 1.4267791929762922e-05, "loss": 1.0465, "step": 7061 }, { "epoch": 1.14, "grad_norm": 10.222334861755371, "learning_rate": 1.4266220102908617e-05, "loss": 1.3539, "step": 7062 }, { "epoch": 1.14, "grad_norm": 10.600372314453125, "learning_rate": 1.4264648147188317e-05, "loss": 1.1272, "step": 7063 }, { "epoch": 1.14, "grad_norm": 10.643264770507812, "learning_rate": 1.4263076062649503e-05, "loss": 1.2098, "step": 7064 }, { "epoch": 1.14, "grad_norm": 10.6288423538208, "learning_rate": 1.4261503849339666e-05, "loss": 1.0264, "step": 7065 }, { "epoch": 1.14, "grad_norm": 6.950641632080078, "learning_rate": 1.4259931507306288e-05, "loss": 0.6766, "step": 7066 }, { "epoch": 1.14, "grad_norm": 9.848997116088867, "learning_rate": 1.4258359036596867e-05, "loss": 1.0196, "step": 7067 }, { "epoch": 1.14, "grad_norm": 9.59841537475586, "learning_rate": 1.4256786437258903e-05, "loss": 1.1364, "step": 7068 }, { "epoch": 1.14, "grad_norm": 9.483355522155762, "learning_rate": 1.4255213709339899e-05, "loss": 1.1547, "step": 7069 }, { "epoch": 1.14, "grad_norm": 8.89937973022461, "learning_rate": 1.4253640852887354e-05, "loss": 0.8895, "step": 7070 }, { "epoch": 1.14, "grad_norm": 10.068015098571777, "learning_rate": 1.4252067867948784e-05, "loss": 1.0568, "step": 7071 }, { "epoch": 1.14, "grad_norm": 9.383438110351562, "learning_rate": 1.4250494754571704e-05, "loss": 1.0782, "step": 7072 }, { "epoch": 1.14, "grad_norm": 12.432585716247559, "learning_rate": 1.424892151280363e-05, "loss": 1.2941, "step": 7073 }, { "epoch": 1.14, "grad_norm": 9.01949405670166, "learning_rate": 1.4247348142692081e-05, "loss": 1.2183, "step": 7074 }, { "epoch": 1.14, "grad_norm": 10.347647666931152, "learning_rate": 1.4245774644284585e-05, "loss": 1.3012, "step": 7075 }, { "epoch": 1.14, "grad_norm": 8.887221336364746, "learning_rate": 1.424420101762867e-05, "loss": 1.0556, "step": 7076 }, { "epoch": 1.14, "grad_norm": 8.760503768920898, "learning_rate": 1.424262726277187e-05, "loss": 1.0199, "step": 7077 }, { "epoch": 1.14, "grad_norm": 10.726546287536621, "learning_rate": 1.4241053379761723e-05, "loss": 0.8904, "step": 7078 }, { "epoch": 1.14, "grad_norm": 9.928226470947266, "learning_rate": 1.4239479368645768e-05, "loss": 1.0619, "step": 7079 }, { "epoch": 1.14, "grad_norm": 8.92959213256836, "learning_rate": 1.4237905229471549e-05, "loss": 1.1343, "step": 7080 }, { "epoch": 1.14, "grad_norm": 9.04601764678955, "learning_rate": 1.4236330962286618e-05, "loss": 0.9076, "step": 7081 }, { "epoch": 1.14, "grad_norm": 9.061758995056152, "learning_rate": 1.4234756567138524e-05, "loss": 1.1051, "step": 7082 }, { "epoch": 1.14, "grad_norm": 7.710172653198242, "learning_rate": 1.4233182044074824e-05, "loss": 1.1098, "step": 7083 }, { "epoch": 1.14, "grad_norm": 8.502965927124023, "learning_rate": 1.423160739314308e-05, "loss": 0.9714, "step": 7084 }, { "epoch": 1.14, "grad_norm": 11.370712280273438, "learning_rate": 1.423003261439086e-05, "loss": 1.2477, "step": 7085 }, { "epoch": 1.14, "grad_norm": 10.962138175964355, "learning_rate": 1.4228457707865722e-05, "loss": 1.2423, "step": 7086 }, { "epoch": 1.14, "grad_norm": 9.973814964294434, "learning_rate": 1.4226882673615245e-05, "loss": 1.0666, "step": 7087 }, { "epoch": 1.14, "grad_norm": 8.637410163879395, "learning_rate": 1.4225307511687003e-05, "loss": 1.2036, "step": 7088 }, { "epoch": 1.14, "grad_norm": 11.658403396606445, "learning_rate": 1.4223732222128575e-05, "loss": 0.9684, "step": 7089 }, { "epoch": 1.14, "grad_norm": 8.347193717956543, "learning_rate": 1.4222156804987545e-05, "loss": 0.9624, "step": 7090 }, { "epoch": 1.14, "grad_norm": 10.05363655090332, "learning_rate": 1.4220581260311499e-05, "loss": 1.1846, "step": 7091 }, { "epoch": 1.14, "grad_norm": 7.669898509979248, "learning_rate": 1.4219005588148031e-05, "loss": 1.056, "step": 7092 }, { "epoch": 1.14, "grad_norm": 8.53329086303711, "learning_rate": 1.4217429788544732e-05, "loss": 1.1174, "step": 7093 }, { "epoch": 1.14, "grad_norm": 7.460669040679932, "learning_rate": 1.4215853861549205e-05, "loss": 1.2057, "step": 7094 }, { "epoch": 1.14, "grad_norm": 7.580000877380371, "learning_rate": 1.4214277807209049e-05, "loss": 1.1693, "step": 7095 }, { "epoch": 1.14, "grad_norm": 7.709446907043457, "learning_rate": 1.4212701625571877e-05, "loss": 0.786, "step": 7096 }, { "epoch": 1.14, "grad_norm": 8.492490768432617, "learning_rate": 1.4211125316685289e-05, "loss": 1.1008, "step": 7097 }, { "epoch": 1.14, "grad_norm": 9.98327922821045, "learning_rate": 1.4209548880596907e-05, "loss": 1.38, "step": 7098 }, { "epoch": 1.14, "grad_norm": 9.483236312866211, "learning_rate": 1.4207972317354343e-05, "loss": 1.0962, "step": 7099 }, { "epoch": 1.14, "grad_norm": 9.843877792358398, "learning_rate": 1.4206395627005226e-05, "loss": 0.9546, "step": 7100 }, { "epoch": 1.14, "grad_norm": 10.009406089782715, "learning_rate": 1.4204818809597177e-05, "loss": 1.075, "step": 7101 }, { "epoch": 1.14, "grad_norm": 9.240193367004395, "learning_rate": 1.4203241865177826e-05, "loss": 1.1363, "step": 7102 }, { "epoch": 1.14, "grad_norm": 8.17448902130127, "learning_rate": 1.4201664793794809e-05, "loss": 1.0153, "step": 7103 }, { "epoch": 1.14, "grad_norm": 8.686166763305664, "learning_rate": 1.4200087595495764e-05, "loss": 0.9784, "step": 7104 }, { "epoch": 1.14, "grad_norm": 8.554121971130371, "learning_rate": 1.4198510270328324e-05, "loss": 1.105, "step": 7105 }, { "epoch": 1.14, "grad_norm": 8.486175537109375, "learning_rate": 1.4196932818340142e-05, "loss": 1.003, "step": 7106 }, { "epoch": 1.14, "grad_norm": 9.914017677307129, "learning_rate": 1.4195355239578863e-05, "loss": 1.2769, "step": 7107 }, { "epoch": 1.14, "grad_norm": 8.44678020477295, "learning_rate": 1.4193777534092144e-05, "loss": 1.1738, "step": 7108 }, { "epoch": 1.14, "grad_norm": 8.988994598388672, "learning_rate": 1.4192199701927637e-05, "loss": 1.1011, "step": 7109 }, { "epoch": 1.14, "grad_norm": 10.21877384185791, "learning_rate": 1.4190621743133e-05, "loss": 1.1004, "step": 7110 }, { "epoch": 1.14, "grad_norm": 8.787614822387695, "learning_rate": 1.4189043657755898e-05, "loss": 0.952, "step": 7111 }, { "epoch": 1.14, "grad_norm": 8.56800365447998, "learning_rate": 1.4187465445844008e-05, "loss": 1.0584, "step": 7112 }, { "epoch": 1.15, "grad_norm": 9.484332084655762, "learning_rate": 1.4185887107444991e-05, "loss": 1.0503, "step": 7113 }, { "epoch": 1.15, "grad_norm": 8.679442405700684, "learning_rate": 1.4184308642606529e-05, "loss": 1.2682, "step": 7114 }, { "epoch": 1.15, "grad_norm": 8.72862434387207, "learning_rate": 1.4182730051376295e-05, "loss": 0.9025, "step": 7115 }, { "epoch": 1.15, "grad_norm": 7.627735137939453, "learning_rate": 1.418115133380198e-05, "loss": 1.0078, "step": 7116 }, { "epoch": 1.15, "grad_norm": 10.70093059539795, "learning_rate": 1.4179572489931262e-05, "loss": 1.146, "step": 7117 }, { "epoch": 1.15, "grad_norm": 8.509746551513672, "learning_rate": 1.4177993519811839e-05, "loss": 1.1834, "step": 7118 }, { "epoch": 1.15, "grad_norm": 7.523044109344482, "learning_rate": 1.41764144234914e-05, "loss": 0.983, "step": 7119 }, { "epoch": 1.15, "grad_norm": 8.584774017333984, "learning_rate": 1.4174835201017652e-05, "loss": 1.1642, "step": 7120 }, { "epoch": 1.15, "grad_norm": 7.401889324188232, "learning_rate": 1.4173255852438285e-05, "loss": 0.9313, "step": 7121 }, { "epoch": 1.15, "grad_norm": 7.3206281661987305, "learning_rate": 1.4171676377801014e-05, "loss": 1.2272, "step": 7122 }, { "epoch": 1.15, "grad_norm": 9.795525550842285, "learning_rate": 1.4170096777153548e-05, "loss": 1.0834, "step": 7123 }, { "epoch": 1.15, "grad_norm": 9.609731674194336, "learning_rate": 1.41685170505436e-05, "loss": 1.1483, "step": 7124 }, { "epoch": 1.15, "grad_norm": 8.680994033813477, "learning_rate": 1.4166937198018884e-05, "loss": 1.2743, "step": 7125 }, { "epoch": 1.15, "grad_norm": 7.691725730895996, "learning_rate": 1.4165357219627124e-05, "loss": 1.1257, "step": 7126 }, { "epoch": 1.15, "grad_norm": 9.12700366973877, "learning_rate": 1.4163777115416044e-05, "loss": 1.129, "step": 7127 }, { "epoch": 1.15, "grad_norm": 8.393863677978516, "learning_rate": 1.4162196885433376e-05, "loss": 1.1235, "step": 7128 }, { "epoch": 1.15, "grad_norm": 9.9513578414917, "learning_rate": 1.416061652972685e-05, "loss": 1.1893, "step": 7129 }, { "epoch": 1.15, "grad_norm": 7.4598917961120605, "learning_rate": 1.4159036048344202e-05, "loss": 0.9824, "step": 7130 }, { "epoch": 1.15, "grad_norm": 12.32394027709961, "learning_rate": 1.415745544133317e-05, "loss": 1.1046, "step": 7131 }, { "epoch": 1.15, "grad_norm": 9.535552024841309, "learning_rate": 1.4155874708741503e-05, "loss": 1.1885, "step": 7132 }, { "epoch": 1.15, "grad_norm": 8.429561614990234, "learning_rate": 1.4154293850616947e-05, "loss": 1.1542, "step": 7133 }, { "epoch": 1.15, "grad_norm": 7.791080951690674, "learning_rate": 1.4152712867007252e-05, "loss": 1.0761, "step": 7134 }, { "epoch": 1.15, "grad_norm": 8.051886558532715, "learning_rate": 1.4151131757960178e-05, "loss": 1.007, "step": 7135 }, { "epoch": 1.15, "grad_norm": 7.467377185821533, "learning_rate": 1.4149550523523476e-05, "loss": 0.7151, "step": 7136 }, { "epoch": 1.15, "grad_norm": 9.203689575195312, "learning_rate": 1.4147969163744917e-05, "loss": 1.1622, "step": 7137 }, { "epoch": 1.15, "grad_norm": 11.49134635925293, "learning_rate": 1.4146387678672264e-05, "loss": 1.2812, "step": 7138 }, { "epoch": 1.15, "grad_norm": 9.037611961364746, "learning_rate": 1.4144806068353287e-05, "loss": 1.1709, "step": 7139 }, { "epoch": 1.15, "grad_norm": 8.859214782714844, "learning_rate": 1.4143224332835763e-05, "loss": 1.3004, "step": 7140 }, { "epoch": 1.15, "grad_norm": 8.344008445739746, "learning_rate": 1.4141642472167468e-05, "loss": 1.1006, "step": 7141 }, { "epoch": 1.15, "grad_norm": 10.36047649383545, "learning_rate": 1.4140060486396184e-05, "loss": 1.3993, "step": 7142 }, { "epoch": 1.15, "grad_norm": 8.619219779968262, "learning_rate": 1.4138478375569698e-05, "loss": 1.015, "step": 7143 }, { "epoch": 1.15, "grad_norm": 9.171050071716309, "learning_rate": 1.4136896139735797e-05, "loss": 1.0274, "step": 7144 }, { "epoch": 1.15, "grad_norm": 8.113643646240234, "learning_rate": 1.4135313778942275e-05, "loss": 1.0825, "step": 7145 }, { "epoch": 1.15, "grad_norm": 10.682025909423828, "learning_rate": 1.4133731293236933e-05, "loss": 1.1476, "step": 7146 }, { "epoch": 1.15, "grad_norm": 9.26974868774414, "learning_rate": 1.4132148682667569e-05, "loss": 0.982, "step": 7147 }, { "epoch": 1.15, "grad_norm": 8.887011528015137, "learning_rate": 1.4130565947281984e-05, "loss": 1.465, "step": 7148 }, { "epoch": 1.15, "grad_norm": 8.681208610534668, "learning_rate": 1.4128983087127989e-05, "loss": 1.0672, "step": 7149 }, { "epoch": 1.15, "grad_norm": 8.173463821411133, "learning_rate": 1.4127400102253397e-05, "loss": 0.8881, "step": 7150 }, { "epoch": 1.15, "grad_norm": 8.64274787902832, "learning_rate": 1.4125816992706026e-05, "loss": 1.0483, "step": 7151 }, { "epoch": 1.15, "grad_norm": 9.206334114074707, "learning_rate": 1.412423375853369e-05, "loss": 1.1922, "step": 7152 }, { "epoch": 1.15, "grad_norm": 7.4503607749938965, "learning_rate": 1.4122650399784214e-05, "loss": 1.0782, "step": 7153 }, { "epoch": 1.15, "grad_norm": 7.928387641906738, "learning_rate": 1.412106691650543e-05, "loss": 0.8195, "step": 7154 }, { "epoch": 1.15, "grad_norm": 9.273557662963867, "learning_rate": 1.4119483308745165e-05, "loss": 1.13, "step": 7155 }, { "epoch": 1.15, "grad_norm": 7.979517936706543, "learning_rate": 1.4117899576551253e-05, "loss": 0.8173, "step": 7156 }, { "epoch": 1.15, "grad_norm": 10.201181411743164, "learning_rate": 1.4116315719971534e-05, "loss": 1.3349, "step": 7157 }, { "epoch": 1.15, "grad_norm": 9.248017311096191, "learning_rate": 1.4114731739053848e-05, "loss": 1.3639, "step": 7158 }, { "epoch": 1.15, "grad_norm": 9.92277717590332, "learning_rate": 1.4113147633846044e-05, "loss": 1.1078, "step": 7159 }, { "epoch": 1.15, "grad_norm": 7.942895412445068, "learning_rate": 1.4111563404395971e-05, "loss": 1.0228, "step": 7160 }, { "epoch": 1.15, "grad_norm": 8.621039390563965, "learning_rate": 1.410997905075148e-05, "loss": 1.2583, "step": 7161 }, { "epoch": 1.15, "grad_norm": 8.518583297729492, "learning_rate": 1.4108394572960427e-05, "loss": 1.2062, "step": 7162 }, { "epoch": 1.15, "grad_norm": 10.563159942626953, "learning_rate": 1.4106809971070683e-05, "loss": 0.9533, "step": 7163 }, { "epoch": 1.15, "grad_norm": 8.137927055358887, "learning_rate": 1.4105225245130098e-05, "loss": 0.993, "step": 7164 }, { "epoch": 1.15, "grad_norm": 9.641478538513184, "learning_rate": 1.4103640395186553e-05, "loss": 1.1285, "step": 7165 }, { "epoch": 1.15, "grad_norm": 7.683690547943115, "learning_rate": 1.4102055421287912e-05, "loss": 1.0638, "step": 7166 }, { "epoch": 1.15, "grad_norm": 7.815800666809082, "learning_rate": 1.4100470323482058e-05, "loss": 1.2442, "step": 7167 }, { "epoch": 1.15, "grad_norm": 9.194905281066895, "learning_rate": 1.4098885101816864e-05, "loss": 1.2376, "step": 7168 }, { "epoch": 1.15, "grad_norm": 7.359624862670898, "learning_rate": 1.4097299756340217e-05, "loss": 0.9363, "step": 7169 }, { "epoch": 1.15, "grad_norm": 10.441441535949707, "learning_rate": 1.4095714287100003e-05, "loss": 1.2862, "step": 7170 }, { "epoch": 1.15, "grad_norm": 8.716414451599121, "learning_rate": 1.4094128694144113e-05, "loss": 1.1325, "step": 7171 }, { "epoch": 1.15, "grad_norm": 9.505396842956543, "learning_rate": 1.4092542977520439e-05, "loss": 1.1253, "step": 7172 }, { "epoch": 1.15, "grad_norm": 7.679351806640625, "learning_rate": 1.4090957137276883e-05, "loss": 1.1201, "step": 7173 }, { "epoch": 1.15, "grad_norm": 10.585515975952148, "learning_rate": 1.408937117346135e-05, "loss": 1.1599, "step": 7174 }, { "epoch": 1.16, "grad_norm": 9.543107032775879, "learning_rate": 1.408778508612174e-05, "loss": 1.1977, "step": 7175 }, { "epoch": 1.16, "grad_norm": 8.919164657592773, "learning_rate": 1.4086198875305965e-05, "loss": 1.1092, "step": 7176 }, { "epoch": 1.16, "grad_norm": 9.579047203063965, "learning_rate": 1.4084612541061934e-05, "loss": 1.178, "step": 7177 }, { "epoch": 1.16, "grad_norm": 8.122328758239746, "learning_rate": 1.4083026083437572e-05, "loss": 0.9383, "step": 7178 }, { "epoch": 1.16, "grad_norm": 8.382328987121582, "learning_rate": 1.4081439502480799e-05, "loss": 1.3815, "step": 7179 }, { "epoch": 1.16, "grad_norm": 9.569046020507812, "learning_rate": 1.407985279823953e-05, "loss": 0.93, "step": 7180 }, { "epoch": 1.16, "grad_norm": 9.80721378326416, "learning_rate": 1.4078265970761699e-05, "loss": 1.0632, "step": 7181 }, { "epoch": 1.16, "grad_norm": 7.825000762939453, "learning_rate": 1.4076679020095243e-05, "loss": 0.9636, "step": 7182 }, { "epoch": 1.16, "grad_norm": 8.689888000488281, "learning_rate": 1.4075091946288089e-05, "loss": 1.0709, "step": 7183 }, { "epoch": 1.16, "grad_norm": 7.551395893096924, "learning_rate": 1.4073504749388183e-05, "loss": 0.9124, "step": 7184 }, { "epoch": 1.16, "grad_norm": 9.176518440246582, "learning_rate": 1.4071917429443463e-05, "loss": 1.1436, "step": 7185 }, { "epoch": 1.16, "grad_norm": 10.090978622436523, "learning_rate": 1.4070329986501881e-05, "loss": 1.1651, "step": 7186 }, { "epoch": 1.16, "grad_norm": 7.684496879577637, "learning_rate": 1.406874242061138e-05, "loss": 1.1143, "step": 7187 }, { "epoch": 1.16, "grad_norm": 8.85959243774414, "learning_rate": 1.4067154731819923e-05, "loss": 0.8743, "step": 7188 }, { "epoch": 1.16, "grad_norm": 7.329366207122803, "learning_rate": 1.4065566920175462e-05, "loss": 1.0524, "step": 7189 }, { "epoch": 1.16, "grad_norm": 9.301898956298828, "learning_rate": 1.4063978985725959e-05, "loss": 0.9121, "step": 7190 }, { "epoch": 1.16, "grad_norm": 9.26626205444336, "learning_rate": 1.4062390928519386e-05, "loss": 1.1938, "step": 7191 }, { "epoch": 1.16, "grad_norm": 9.475969314575195, "learning_rate": 1.4060802748603699e-05, "loss": 1.1691, "step": 7192 }, { "epoch": 1.16, "grad_norm": 8.927179336547852, "learning_rate": 1.4059214446026883e-05, "loss": 1.0899, "step": 7193 }, { "epoch": 1.16, "grad_norm": 11.611841201782227, "learning_rate": 1.4057626020836912e-05, "loss": 1.0633, "step": 7194 }, { "epoch": 1.16, "grad_norm": 9.403193473815918, "learning_rate": 1.405603747308176e-05, "loss": 1.2708, "step": 7195 }, { "epoch": 1.16, "grad_norm": 7.839602947235107, "learning_rate": 1.4054448802809416e-05, "loss": 0.7809, "step": 7196 }, { "epoch": 1.16, "grad_norm": 8.705498695373535, "learning_rate": 1.405286001006787e-05, "loss": 1.0335, "step": 7197 }, { "epoch": 1.16, "grad_norm": 9.805883407592773, "learning_rate": 1.4051271094905108e-05, "loss": 1.1522, "step": 7198 }, { "epoch": 1.16, "grad_norm": 9.165678977966309, "learning_rate": 1.4049682057369128e-05, "loss": 0.8683, "step": 7199 }, { "epoch": 1.16, "grad_norm": 7.660426139831543, "learning_rate": 1.4048092897507925e-05, "loss": 1.1557, "step": 7200 }, { "epoch": 1.16, "grad_norm": 8.317886352539062, "learning_rate": 1.4046503615369507e-05, "loss": 0.9982, "step": 7201 }, { "epoch": 1.16, "grad_norm": 8.772041320800781, "learning_rate": 1.4044914211001874e-05, "loss": 1.0223, "step": 7202 }, { "epoch": 1.16, "grad_norm": 10.005915641784668, "learning_rate": 1.404332468445304e-05, "loss": 1.2541, "step": 7203 }, { "epoch": 1.16, "grad_norm": 11.00283432006836, "learning_rate": 1.4041735035771015e-05, "loss": 1.1356, "step": 7204 }, { "epoch": 1.16, "grad_norm": 9.495680809020996, "learning_rate": 1.4040145265003822e-05, "loss": 1.0878, "step": 7205 }, { "epoch": 1.16, "grad_norm": 8.40898609161377, "learning_rate": 1.4038555372199479e-05, "loss": 1.3461, "step": 7206 }, { "epoch": 1.16, "grad_norm": 10.545427322387695, "learning_rate": 1.403696535740601e-05, "loss": 1.0976, "step": 7207 }, { "epoch": 1.16, "grad_norm": 10.654854774475098, "learning_rate": 1.4035375220671439e-05, "loss": 1.0626, "step": 7208 }, { "epoch": 1.16, "grad_norm": 7.856832027435303, "learning_rate": 1.4033784962043803e-05, "loss": 0.9715, "step": 7209 }, { "epoch": 1.16, "grad_norm": 8.569067001342773, "learning_rate": 1.4032194581571137e-05, "loss": 1.0137, "step": 7210 }, { "epoch": 1.16, "grad_norm": 10.56462287902832, "learning_rate": 1.403060407930148e-05, "loss": 1.2323, "step": 7211 }, { "epoch": 1.16, "grad_norm": 9.304344177246094, "learning_rate": 1.4029013455282872e-05, "loss": 0.694, "step": 7212 }, { "epoch": 1.16, "grad_norm": 8.462626457214355, "learning_rate": 1.4027422709563365e-05, "loss": 0.9898, "step": 7213 }, { "epoch": 1.16, "grad_norm": 9.86455249786377, "learning_rate": 1.4025831842191004e-05, "loss": 1.114, "step": 7214 }, { "epoch": 1.16, "grad_norm": 8.934321403503418, "learning_rate": 1.4024240853213848e-05, "loss": 0.8841, "step": 7215 }, { "epoch": 1.16, "grad_norm": 10.409475326538086, "learning_rate": 1.402264974267995e-05, "loss": 1.1618, "step": 7216 }, { "epoch": 1.16, "grad_norm": 8.82647705078125, "learning_rate": 1.4021058510637374e-05, "loss": 0.7292, "step": 7217 }, { "epoch": 1.16, "grad_norm": 10.533852577209473, "learning_rate": 1.4019467157134185e-05, "loss": 1.0906, "step": 7218 }, { "epoch": 1.16, "grad_norm": 7.996520519256592, "learning_rate": 1.4017875682218449e-05, "loss": 0.9508, "step": 7219 }, { "epoch": 1.16, "grad_norm": 10.320810317993164, "learning_rate": 1.4016284085938241e-05, "loss": 1.2304, "step": 7220 }, { "epoch": 1.16, "grad_norm": 9.23729419708252, "learning_rate": 1.4014692368341636e-05, "loss": 1.0878, "step": 7221 }, { "epoch": 1.16, "grad_norm": 10.162352561950684, "learning_rate": 1.4013100529476714e-05, "loss": 1.2138, "step": 7222 }, { "epoch": 1.16, "grad_norm": 9.098790168762207, "learning_rate": 1.4011508569391556e-05, "loss": 0.8763, "step": 7223 }, { "epoch": 1.16, "grad_norm": 10.631902694702148, "learning_rate": 1.4009916488134254e-05, "loss": 1.1316, "step": 7224 }, { "epoch": 1.16, "grad_norm": 7.441483020782471, "learning_rate": 1.4008324285752894e-05, "loss": 0.911, "step": 7225 }, { "epoch": 1.16, "grad_norm": 11.130902290344238, "learning_rate": 1.4006731962295572e-05, "loss": 1.1006, "step": 7226 }, { "epoch": 1.16, "grad_norm": 9.443947792053223, "learning_rate": 1.4005139517810386e-05, "loss": 1.0436, "step": 7227 }, { "epoch": 1.16, "grad_norm": 7.835217475891113, "learning_rate": 1.4003546952345438e-05, "loss": 0.8734, "step": 7228 }, { "epoch": 1.16, "grad_norm": 8.7227144241333, "learning_rate": 1.4001954265948831e-05, "loss": 1.2196, "step": 7229 }, { "epoch": 1.16, "grad_norm": 8.80202579498291, "learning_rate": 1.4000361458668677e-05, "loss": 1.208, "step": 7230 }, { "epoch": 1.16, "grad_norm": 9.423768043518066, "learning_rate": 1.3998768530553086e-05, "loss": 1.0962, "step": 7231 }, { "epoch": 1.16, "grad_norm": 9.643233299255371, "learning_rate": 1.3997175481650176e-05, "loss": 1.0153, "step": 7232 }, { "epoch": 1.16, "grad_norm": 10.164977073669434, "learning_rate": 1.3995582312008065e-05, "loss": 1.0192, "step": 7233 }, { "epoch": 1.16, "grad_norm": 9.067978858947754, "learning_rate": 1.3993989021674877e-05, "loss": 1.122, "step": 7234 }, { "epoch": 1.16, "grad_norm": 9.646366119384766, "learning_rate": 1.3992395610698739e-05, "loss": 0.9926, "step": 7235 }, { "epoch": 1.16, "grad_norm": 7.2099480628967285, "learning_rate": 1.3990802079127785e-05, "loss": 0.8562, "step": 7236 }, { "epoch": 1.17, "grad_norm": 9.904199600219727, "learning_rate": 1.3989208427010147e-05, "loss": 1.2732, "step": 7237 }, { "epoch": 1.17, "grad_norm": 9.018601417541504, "learning_rate": 1.3987614654393962e-05, "loss": 0.9921, "step": 7238 }, { "epoch": 1.17, "grad_norm": 10.617436408996582, "learning_rate": 1.3986020761327373e-05, "loss": 0.9118, "step": 7239 }, { "epoch": 1.17, "grad_norm": 10.710189819335938, "learning_rate": 1.3984426747858524e-05, "loss": 0.9624, "step": 7240 }, { "epoch": 1.17, "grad_norm": 8.250940322875977, "learning_rate": 1.3982832614035564e-05, "loss": 1.2398, "step": 7241 }, { "epoch": 1.17, "grad_norm": 9.011235237121582, "learning_rate": 1.398123835990665e-05, "loss": 0.8313, "step": 7242 }, { "epoch": 1.17, "grad_norm": 9.337346076965332, "learning_rate": 1.397964398551993e-05, "loss": 0.9953, "step": 7243 }, { "epoch": 1.17, "grad_norm": 9.764412879943848, "learning_rate": 1.397804949092357e-05, "loss": 1.0148, "step": 7244 }, { "epoch": 1.17, "grad_norm": 8.958796501159668, "learning_rate": 1.3976454876165737e-05, "loss": 1.1306, "step": 7245 }, { "epoch": 1.17, "grad_norm": 10.948305130004883, "learning_rate": 1.3974860141294589e-05, "loss": 1.1741, "step": 7246 }, { "epoch": 1.17, "grad_norm": 10.108753204345703, "learning_rate": 1.3973265286358302e-05, "loss": 0.9107, "step": 7247 }, { "epoch": 1.17, "grad_norm": 9.269431114196777, "learning_rate": 1.397167031140505e-05, "loss": 1.1672, "step": 7248 }, { "epoch": 1.17, "grad_norm": 9.60107421875, "learning_rate": 1.3970075216483012e-05, "loss": 0.8789, "step": 7249 }, { "epoch": 1.17, "grad_norm": 9.130064010620117, "learning_rate": 1.3968480001640366e-05, "loss": 0.925, "step": 7250 }, { "epoch": 1.17, "grad_norm": 8.440929412841797, "learning_rate": 1.3966884666925297e-05, "loss": 1.1254, "step": 7251 }, { "epoch": 1.17, "grad_norm": 10.307888984680176, "learning_rate": 1.3965289212386e-05, "loss": 1.0392, "step": 7252 }, { "epoch": 1.17, "grad_norm": 9.717070579528809, "learning_rate": 1.3963693638070664e-05, "loss": 1.0159, "step": 7253 }, { "epoch": 1.17, "grad_norm": 10.015424728393555, "learning_rate": 1.3962097944027481e-05, "loss": 1.1101, "step": 7254 }, { "epoch": 1.17, "grad_norm": 9.700809478759766, "learning_rate": 1.3960502130304659e-05, "loss": 1.2313, "step": 7255 }, { "epoch": 1.17, "grad_norm": 8.474374771118164, "learning_rate": 1.3958906196950394e-05, "loss": 1.0783, "step": 7256 }, { "epoch": 1.17, "grad_norm": 7.8078107833862305, "learning_rate": 1.3957310144012901e-05, "loss": 1.0885, "step": 7257 }, { "epoch": 1.17, "grad_norm": 7.842254161834717, "learning_rate": 1.3955713971540382e-05, "loss": 0.9898, "step": 7258 }, { "epoch": 1.17, "grad_norm": 8.944291114807129, "learning_rate": 1.3954117679581055e-05, "loss": 1.2161, "step": 7259 }, { "epoch": 1.17, "grad_norm": 9.361133575439453, "learning_rate": 1.3952521268183135e-05, "loss": 1.1258, "step": 7260 }, { "epoch": 1.17, "grad_norm": 10.048871994018555, "learning_rate": 1.395092473739485e-05, "loss": 1.2551, "step": 7261 }, { "epoch": 1.17, "grad_norm": 8.58267879486084, "learning_rate": 1.394932808726442e-05, "loss": 0.8383, "step": 7262 }, { "epoch": 1.17, "grad_norm": 9.963112831115723, "learning_rate": 1.3947731317840072e-05, "loss": 1.078, "step": 7263 }, { "epoch": 1.17, "grad_norm": 11.048529624938965, "learning_rate": 1.3946134429170044e-05, "loss": 1.3068, "step": 7264 }, { "epoch": 1.17, "grad_norm": 11.49722671508789, "learning_rate": 1.3944537421302568e-05, "loss": 1.157, "step": 7265 }, { "epoch": 1.17, "grad_norm": 7.80727481842041, "learning_rate": 1.3942940294285883e-05, "loss": 0.9242, "step": 7266 }, { "epoch": 1.17, "grad_norm": 9.076489448547363, "learning_rate": 1.3941343048168232e-05, "loss": 0.9573, "step": 7267 }, { "epoch": 1.17, "grad_norm": 9.685714721679688, "learning_rate": 1.3939745682997866e-05, "loss": 1.2178, "step": 7268 }, { "epoch": 1.17, "grad_norm": 9.81823444366455, "learning_rate": 1.3938148198823032e-05, "loss": 0.8912, "step": 7269 }, { "epoch": 1.17, "grad_norm": 9.179210662841797, "learning_rate": 1.3936550595691983e-05, "loss": 1.2893, "step": 7270 }, { "epoch": 1.17, "grad_norm": 9.258890151977539, "learning_rate": 1.3934952873652975e-05, "loss": 0.88, "step": 7271 }, { "epoch": 1.17, "grad_norm": 8.44430923461914, "learning_rate": 1.3933355032754272e-05, "loss": 1.1646, "step": 7272 }, { "epoch": 1.17, "grad_norm": 8.630928039550781, "learning_rate": 1.3931757073044139e-05, "loss": 1.1368, "step": 7273 }, { "epoch": 1.17, "grad_norm": 9.547074317932129, "learning_rate": 1.3930158994570841e-05, "loss": 1.0705, "step": 7274 }, { "epoch": 1.17, "grad_norm": 9.642616271972656, "learning_rate": 1.3928560797382656e-05, "loss": 0.9948, "step": 7275 }, { "epoch": 1.17, "grad_norm": 11.322842597961426, "learning_rate": 1.3926962481527853e-05, "loss": 1.026, "step": 7276 }, { "epoch": 1.17, "grad_norm": 11.77092456817627, "learning_rate": 1.3925364047054712e-05, "loss": 1.282, "step": 7277 }, { "epoch": 1.17, "grad_norm": 8.230789184570312, "learning_rate": 1.3923765494011517e-05, "loss": 1.3458, "step": 7278 }, { "epoch": 1.17, "grad_norm": 9.33173656463623, "learning_rate": 1.3922166822446553e-05, "loss": 1.1358, "step": 7279 }, { "epoch": 1.17, "grad_norm": 8.612099647521973, "learning_rate": 1.392056803240811e-05, "loss": 1.0542, "step": 7280 }, { "epoch": 1.17, "grad_norm": 10.509111404418945, "learning_rate": 1.3918969123944483e-05, "loss": 1.2937, "step": 7281 }, { "epoch": 1.17, "grad_norm": 9.452993392944336, "learning_rate": 1.3917370097103967e-05, "loss": 1.1961, "step": 7282 }, { "epoch": 1.17, "grad_norm": 9.034357070922852, "learning_rate": 1.391577095193486e-05, "loss": 1.102, "step": 7283 }, { "epoch": 1.17, "grad_norm": 8.108059883117676, "learning_rate": 1.3914171688485473e-05, "loss": 1.2206, "step": 7284 }, { "epoch": 1.17, "grad_norm": 9.029460906982422, "learning_rate": 1.3912572306804108e-05, "loss": 1.1029, "step": 7285 }, { "epoch": 1.17, "grad_norm": 8.878877639770508, "learning_rate": 1.3910972806939076e-05, "loss": 1.0068, "step": 7286 }, { "epoch": 1.17, "grad_norm": 7.118765354156494, "learning_rate": 1.3909373188938695e-05, "loss": 0.8312, "step": 7287 }, { "epoch": 1.17, "grad_norm": 9.851134300231934, "learning_rate": 1.3907773452851283e-05, "loss": 1.0726, "step": 7288 }, { "epoch": 1.17, "grad_norm": 7.697023868560791, "learning_rate": 1.3906173598725156e-05, "loss": 1.03, "step": 7289 }, { "epoch": 1.17, "grad_norm": 11.162067413330078, "learning_rate": 1.3904573626608645e-05, "loss": 1.2503, "step": 7290 }, { "epoch": 1.17, "grad_norm": 7.353150844573975, "learning_rate": 1.390297353655008e-05, "loss": 0.6889, "step": 7291 }, { "epoch": 1.17, "grad_norm": 7.588124752044678, "learning_rate": 1.3901373328597792e-05, "loss": 1.1405, "step": 7292 }, { "epoch": 1.17, "grad_norm": 8.119577407836914, "learning_rate": 1.3899773002800113e-05, "loss": 0.9596, "step": 7293 }, { "epoch": 1.17, "grad_norm": 9.2183198928833, "learning_rate": 1.3898172559205385e-05, "loss": 1.1235, "step": 7294 }, { "epoch": 1.17, "grad_norm": 8.780250549316406, "learning_rate": 1.3896571997861953e-05, "loss": 0.8836, "step": 7295 }, { "epoch": 1.17, "grad_norm": 7.312885284423828, "learning_rate": 1.3894971318818167e-05, "loss": 1.0418, "step": 7296 }, { "epoch": 1.17, "grad_norm": 7.998175144195557, "learning_rate": 1.3893370522122372e-05, "loss": 1.0664, "step": 7297 }, { "epoch": 1.17, "grad_norm": 7.569463729858398, "learning_rate": 1.389176960782292e-05, "loss": 0.8908, "step": 7298 }, { "epoch": 1.17, "grad_norm": 9.332140922546387, "learning_rate": 1.3890168575968173e-05, "loss": 1.0748, "step": 7299 }, { "epoch": 1.18, "grad_norm": 10.051441192626953, "learning_rate": 1.3888567426606491e-05, "loss": 1.3043, "step": 7300 }, { "epoch": 1.18, "grad_norm": 8.308073997497559, "learning_rate": 1.3886966159786236e-05, "loss": 1.1197, "step": 7301 }, { "epoch": 1.18, "grad_norm": 8.70195198059082, "learning_rate": 1.3885364775555781e-05, "loss": 1.153, "step": 7302 }, { "epoch": 1.18, "grad_norm": 8.91375732421875, "learning_rate": 1.3883763273963492e-05, "loss": 0.9904, "step": 7303 }, { "epoch": 1.18, "grad_norm": 9.263591766357422, "learning_rate": 1.3882161655057748e-05, "loss": 1.0006, "step": 7304 }, { "epoch": 1.18, "grad_norm": 9.171334266662598, "learning_rate": 1.3880559918886927e-05, "loss": 1.0019, "step": 7305 }, { "epoch": 1.18, "grad_norm": 8.799835205078125, "learning_rate": 1.3878958065499408e-05, "loss": 1.0353, "step": 7306 }, { "epoch": 1.18, "grad_norm": 8.577808380126953, "learning_rate": 1.3877356094943584e-05, "loss": 1.1287, "step": 7307 }, { "epoch": 1.18, "grad_norm": 8.30741024017334, "learning_rate": 1.3875754007267839e-05, "loss": 0.9434, "step": 7308 }, { "epoch": 1.18, "grad_norm": 9.58609390258789, "learning_rate": 1.3874151802520564e-05, "loss": 1.3607, "step": 7309 }, { "epoch": 1.18, "grad_norm": 9.278654098510742, "learning_rate": 1.387254948075016e-05, "loss": 1.532, "step": 7310 }, { "epoch": 1.18, "grad_norm": 8.844694137573242, "learning_rate": 1.3870947042005024e-05, "loss": 1.1935, "step": 7311 }, { "epoch": 1.18, "grad_norm": 9.585789680480957, "learning_rate": 1.3869344486333565e-05, "loss": 0.8991, "step": 7312 }, { "epoch": 1.18, "grad_norm": 8.593530654907227, "learning_rate": 1.3867741813784182e-05, "loss": 1.0312, "step": 7313 }, { "epoch": 1.18, "grad_norm": 7.755086898803711, "learning_rate": 1.3866139024405287e-05, "loss": 0.8259, "step": 7314 }, { "epoch": 1.18, "grad_norm": 10.018855094909668, "learning_rate": 1.3864536118245302e-05, "loss": 1.2076, "step": 7315 }, { "epoch": 1.18, "grad_norm": 8.901195526123047, "learning_rate": 1.3862933095352636e-05, "loss": 1.123, "step": 7316 }, { "epoch": 1.18, "grad_norm": 8.46088981628418, "learning_rate": 1.3861329955775711e-05, "loss": 0.9295, "step": 7317 }, { "epoch": 1.18, "grad_norm": 9.47457218170166, "learning_rate": 1.3859726699562954e-05, "loss": 1.0302, "step": 7318 }, { "epoch": 1.18, "grad_norm": 7.556437015533447, "learning_rate": 1.3858123326762793e-05, "loss": 1.0186, "step": 7319 }, { "epoch": 1.18, "grad_norm": 9.01715087890625, "learning_rate": 1.3856519837423664e-05, "loss": 0.9588, "step": 7320 }, { "epoch": 1.18, "grad_norm": 11.234413146972656, "learning_rate": 1.3854916231593994e-05, "loss": 0.9484, "step": 7321 }, { "epoch": 1.18, "grad_norm": 9.058263778686523, "learning_rate": 1.3853312509322224e-05, "loss": 1.1039, "step": 7322 }, { "epoch": 1.18, "grad_norm": 10.42995548248291, "learning_rate": 1.3851708670656798e-05, "loss": 1.0197, "step": 7323 }, { "epoch": 1.18, "grad_norm": 10.149832725524902, "learning_rate": 1.3850104715646162e-05, "loss": 1.273, "step": 7324 }, { "epoch": 1.18, "grad_norm": 9.670454978942871, "learning_rate": 1.3848500644338766e-05, "loss": 1.0848, "step": 7325 }, { "epoch": 1.18, "grad_norm": 9.080032348632812, "learning_rate": 1.384689645678306e-05, "loss": 0.9967, "step": 7326 }, { "epoch": 1.18, "grad_norm": 9.157983779907227, "learning_rate": 1.3845292153027502e-05, "loss": 0.9657, "step": 7327 }, { "epoch": 1.18, "grad_norm": 9.446929931640625, "learning_rate": 1.3843687733120552e-05, "loss": 1.0961, "step": 7328 }, { "epoch": 1.18, "grad_norm": 7.157984733581543, "learning_rate": 1.3842083197110672e-05, "loss": 0.9206, "step": 7329 }, { "epoch": 1.18, "grad_norm": 8.211211204528809, "learning_rate": 1.384047854504633e-05, "loss": 1.0256, "step": 7330 }, { "epoch": 1.18, "grad_norm": 8.474953651428223, "learning_rate": 1.3838873776975999e-05, "loss": 1.1184, "step": 7331 }, { "epoch": 1.18, "grad_norm": 8.039143562316895, "learning_rate": 1.3837268892948147e-05, "loss": 1.0275, "step": 7332 }, { "epoch": 1.18, "grad_norm": 11.240325927734375, "learning_rate": 1.3835663893011253e-05, "loss": 1.4622, "step": 7333 }, { "epoch": 1.18, "grad_norm": 8.885868072509766, "learning_rate": 1.3834058777213797e-05, "loss": 1.1292, "step": 7334 }, { "epoch": 1.18, "grad_norm": 11.194472312927246, "learning_rate": 1.383245354560427e-05, "loss": 1.4544, "step": 7335 }, { "epoch": 1.18, "grad_norm": 11.854950904846191, "learning_rate": 1.3830848198231156e-05, "loss": 1.2309, "step": 7336 }, { "epoch": 1.18, "grad_norm": 8.133222579956055, "learning_rate": 1.382924273514294e-05, "loss": 0.9263, "step": 7337 }, { "epoch": 1.18, "grad_norm": 9.761937141418457, "learning_rate": 1.3827637156388126e-05, "loss": 1.1253, "step": 7338 }, { "epoch": 1.18, "grad_norm": 10.0444974899292, "learning_rate": 1.382603146201521e-05, "loss": 1.1891, "step": 7339 }, { "epoch": 1.18, "grad_norm": 9.581192016601562, "learning_rate": 1.3824425652072691e-05, "loss": 1.2929, "step": 7340 }, { "epoch": 1.18, "grad_norm": 7.730834484100342, "learning_rate": 1.3822819726609073e-05, "loss": 0.9938, "step": 7341 }, { "epoch": 1.18, "grad_norm": 8.329505920410156, "learning_rate": 1.3821213685672869e-05, "loss": 1.3234, "step": 7342 }, { "epoch": 1.18, "grad_norm": 6.323145389556885, "learning_rate": 1.3819607529312593e-05, "loss": 0.7261, "step": 7343 }, { "epoch": 1.18, "grad_norm": 10.024317741394043, "learning_rate": 1.3818001257576754e-05, "loss": 1.1522, "step": 7344 }, { "epoch": 1.18, "grad_norm": 8.82708740234375, "learning_rate": 1.3816394870513877e-05, "loss": 1.0415, "step": 7345 }, { "epoch": 1.18, "grad_norm": 8.887053489685059, "learning_rate": 1.381478836817248e-05, "loss": 0.9538, "step": 7346 }, { "epoch": 1.18, "grad_norm": 9.555830001831055, "learning_rate": 1.3813181750601094e-05, "loss": 0.9818, "step": 7347 }, { "epoch": 1.18, "grad_norm": 8.998812675476074, "learning_rate": 1.3811575017848246e-05, "loss": 0.7748, "step": 7348 }, { "epoch": 1.18, "grad_norm": 7.800058841705322, "learning_rate": 1.3809968169962469e-05, "loss": 0.8082, "step": 7349 }, { "epoch": 1.18, "grad_norm": 7.305813789367676, "learning_rate": 1.3808361206992302e-05, "loss": 0.9501, "step": 7350 }, { "epoch": 1.18, "grad_norm": 9.696552276611328, "learning_rate": 1.3806754128986284e-05, "loss": 1.1304, "step": 7351 }, { "epoch": 1.18, "grad_norm": 7.9839677810668945, "learning_rate": 1.3805146935992957e-05, "loss": 0.974, "step": 7352 }, { "epoch": 1.18, "grad_norm": 10.142980575561523, "learning_rate": 1.3803539628060867e-05, "loss": 1.145, "step": 7353 }, { "epoch": 1.18, "grad_norm": 8.234893798828125, "learning_rate": 1.3801932205238568e-05, "loss": 1.0927, "step": 7354 }, { "epoch": 1.18, "grad_norm": 12.747953414916992, "learning_rate": 1.3800324667574614e-05, "loss": 1.1701, "step": 7355 }, { "epoch": 1.18, "grad_norm": 9.621549606323242, "learning_rate": 1.379871701511756e-05, "loss": 1.0509, "step": 7356 }, { "epoch": 1.18, "grad_norm": 10.801701545715332, "learning_rate": 1.3797109247915968e-05, "loss": 0.9361, "step": 7357 }, { "epoch": 1.18, "grad_norm": 10.681868553161621, "learning_rate": 1.3795501366018405e-05, "loss": 1.1291, "step": 7358 }, { "epoch": 1.18, "grad_norm": 7.411014080047607, "learning_rate": 1.3793893369473431e-05, "loss": 0.8737, "step": 7359 }, { "epoch": 1.18, "grad_norm": 8.824344635009766, "learning_rate": 1.379228525832963e-05, "loss": 1.0216, "step": 7360 }, { "epoch": 1.18, "grad_norm": 9.287163734436035, "learning_rate": 1.3790677032635565e-05, "loss": 1.1927, "step": 7361 }, { "epoch": 1.19, "grad_norm": 9.8884859085083, "learning_rate": 1.378906869243982e-05, "loss": 1.0644, "step": 7362 }, { "epoch": 1.19, "grad_norm": 8.322142601013184, "learning_rate": 1.3787460237790975e-05, "loss": 1.0367, "step": 7363 }, { "epoch": 1.19, "grad_norm": 9.932706832885742, "learning_rate": 1.3785851668737615e-05, "loss": 0.9793, "step": 7364 }, { "epoch": 1.19, "grad_norm": 9.283082962036133, "learning_rate": 1.378424298532833e-05, "loss": 1.0759, "step": 7365 }, { "epoch": 1.19, "grad_norm": 8.57979679107666, "learning_rate": 1.3782634187611717e-05, "loss": 0.97, "step": 7366 }, { "epoch": 1.19, "grad_norm": 10.535256385803223, "learning_rate": 1.378102527563636e-05, "loss": 1.1443, "step": 7367 }, { "epoch": 1.19, "grad_norm": 8.82798957824707, "learning_rate": 1.3779416249450865e-05, "loss": 0.9986, "step": 7368 }, { "epoch": 1.19, "grad_norm": 8.65873908996582, "learning_rate": 1.3777807109103833e-05, "loss": 0.88, "step": 7369 }, { "epoch": 1.19, "grad_norm": 8.760062217712402, "learning_rate": 1.3776197854643876e-05, "loss": 1.1624, "step": 7370 }, { "epoch": 1.19, "grad_norm": 10.012904167175293, "learning_rate": 1.3774588486119594e-05, "loss": 1.1702, "step": 7371 }, { "epoch": 1.19, "grad_norm": 8.395774841308594, "learning_rate": 1.3772979003579601e-05, "loss": 1.0978, "step": 7372 }, { "epoch": 1.19, "grad_norm": 7.358222484588623, "learning_rate": 1.3771369407072518e-05, "loss": 0.9953, "step": 7373 }, { "epoch": 1.19, "grad_norm": 11.640914916992188, "learning_rate": 1.3769759696646963e-05, "loss": 0.9562, "step": 7374 }, { "epoch": 1.19, "grad_norm": 9.644978523254395, "learning_rate": 1.376814987235156e-05, "loss": 1.0693, "step": 7375 }, { "epoch": 1.19, "grad_norm": 10.387388229370117, "learning_rate": 1.3766539934234931e-05, "loss": 1.3538, "step": 7376 }, { "epoch": 1.19, "grad_norm": 8.471427917480469, "learning_rate": 1.376492988234571e-05, "loss": 1.076, "step": 7377 }, { "epoch": 1.19, "grad_norm": 9.869526863098145, "learning_rate": 1.3763319716732534e-05, "loss": 1.0815, "step": 7378 }, { "epoch": 1.19, "grad_norm": 8.16489028930664, "learning_rate": 1.376170943744403e-05, "loss": 1.1641, "step": 7379 }, { "epoch": 1.19, "grad_norm": 7.234323978424072, "learning_rate": 1.3760099044528845e-05, "loss": 0.9265, "step": 7380 }, { "epoch": 1.19, "grad_norm": 9.99488639831543, "learning_rate": 1.3758488538035624e-05, "loss": 1.298, "step": 7381 }, { "epoch": 1.19, "grad_norm": 8.937723159790039, "learning_rate": 1.3756877918013011e-05, "loss": 0.9929, "step": 7382 }, { "epoch": 1.19, "grad_norm": 8.123658180236816, "learning_rate": 1.3755267184509658e-05, "loss": 0.9721, "step": 7383 }, { "epoch": 1.19, "grad_norm": 8.655593872070312, "learning_rate": 1.3753656337574217e-05, "loss": 0.9669, "step": 7384 }, { "epoch": 1.19, "grad_norm": 9.60050106048584, "learning_rate": 1.3752045377255344e-05, "loss": 0.9846, "step": 7385 }, { "epoch": 1.19, "grad_norm": 8.548792839050293, "learning_rate": 1.3750434303601707e-05, "loss": 1.0291, "step": 7386 }, { "epoch": 1.19, "grad_norm": 9.465025901794434, "learning_rate": 1.3748823116661964e-05, "loss": 0.9979, "step": 7387 }, { "epoch": 1.19, "grad_norm": 7.8705854415893555, "learning_rate": 1.3747211816484787e-05, "loss": 1.174, "step": 7388 }, { "epoch": 1.19, "grad_norm": 11.172404289245605, "learning_rate": 1.3745600403118842e-05, "loss": 1.3937, "step": 7389 }, { "epoch": 1.19, "grad_norm": 10.93533992767334, "learning_rate": 1.3743988876612808e-05, "loss": 1.2192, "step": 7390 }, { "epoch": 1.19, "grad_norm": 9.88508129119873, "learning_rate": 1.3742377237015361e-05, "loss": 0.932, "step": 7391 }, { "epoch": 1.19, "grad_norm": 7.691385269165039, "learning_rate": 1.3740765484375183e-05, "loss": 0.8396, "step": 7392 }, { "epoch": 1.19, "grad_norm": 8.570921897888184, "learning_rate": 1.373915361874096e-05, "loss": 1.138, "step": 7393 }, { "epoch": 1.19, "grad_norm": 8.499258041381836, "learning_rate": 1.3737541640161378e-05, "loss": 1.0378, "step": 7394 }, { "epoch": 1.19, "grad_norm": 9.02650260925293, "learning_rate": 1.3735929548685127e-05, "loss": 1.2355, "step": 7395 }, { "epoch": 1.19, "grad_norm": 9.060402870178223, "learning_rate": 1.3734317344360906e-05, "loss": 0.9297, "step": 7396 }, { "epoch": 1.19, "grad_norm": 9.118252754211426, "learning_rate": 1.3732705027237414e-05, "loss": 1.0096, "step": 7397 }, { "epoch": 1.19, "grad_norm": 9.452189445495605, "learning_rate": 1.3731092597363346e-05, "loss": 1.2893, "step": 7398 }, { "epoch": 1.19, "grad_norm": 7.982141494750977, "learning_rate": 1.3729480054787416e-05, "loss": 1.0807, "step": 7399 }, { "epoch": 1.19, "grad_norm": 8.049386024475098, "learning_rate": 1.3727867399558327e-05, "loss": 1.1783, "step": 7400 }, { "epoch": 1.19, "grad_norm": 9.33231258392334, "learning_rate": 1.3726254631724796e-05, "loss": 1.2149, "step": 7401 }, { "epoch": 1.19, "grad_norm": 8.692231178283691, "learning_rate": 1.3724641751335535e-05, "loss": 0.9884, "step": 7402 }, { "epoch": 1.19, "grad_norm": 11.254082679748535, "learning_rate": 1.3723028758439261e-05, "loss": 0.8596, "step": 7403 }, { "epoch": 1.19, "grad_norm": 9.02309799194336, "learning_rate": 1.3721415653084698e-05, "loss": 1.2098, "step": 7404 }, { "epoch": 1.19, "grad_norm": 9.606834411621094, "learning_rate": 1.3719802435320573e-05, "loss": 1.2483, "step": 7405 }, { "epoch": 1.19, "grad_norm": 9.548405647277832, "learning_rate": 1.3718189105195614e-05, "loss": 1.0712, "step": 7406 }, { "epoch": 1.19, "grad_norm": 8.467147827148438, "learning_rate": 1.3716575662758551e-05, "loss": 1.0141, "step": 7407 }, { "epoch": 1.19, "grad_norm": 8.899988174438477, "learning_rate": 1.3714962108058125e-05, "loss": 1.1059, "step": 7408 }, { "epoch": 1.19, "grad_norm": 7.860326290130615, "learning_rate": 1.3713348441143074e-05, "loss": 1.2757, "step": 7409 }, { "epoch": 1.19, "grad_norm": 8.482535362243652, "learning_rate": 1.371173466206214e-05, "loss": 1.1045, "step": 7410 }, { "epoch": 1.19, "grad_norm": 8.002320289611816, "learning_rate": 1.3710120770864066e-05, "loss": 1.1099, "step": 7411 }, { "epoch": 1.19, "grad_norm": 9.90564250946045, "learning_rate": 1.3708506767597605e-05, "loss": 1.0365, "step": 7412 }, { "epoch": 1.19, "grad_norm": 10.48697280883789, "learning_rate": 1.3706892652311505e-05, "loss": 1.1267, "step": 7413 }, { "epoch": 1.19, "grad_norm": 8.485709190368652, "learning_rate": 1.3705278425054532e-05, "loss": 1.0561, "step": 7414 }, { "epoch": 1.19, "grad_norm": 10.673157691955566, "learning_rate": 1.3703664085875435e-05, "loss": 0.9529, "step": 7415 }, { "epoch": 1.19, "grad_norm": 10.785857200622559, "learning_rate": 1.370204963482298e-05, "loss": 1.0939, "step": 7416 }, { "epoch": 1.19, "grad_norm": 7.548821449279785, "learning_rate": 1.370043507194594e-05, "loss": 0.7862, "step": 7417 }, { "epoch": 1.19, "grad_norm": 10.091911315917969, "learning_rate": 1.3698820397293075e-05, "loss": 1.0987, "step": 7418 }, { "epoch": 1.19, "grad_norm": 8.793737411499023, "learning_rate": 1.3697205610913162e-05, "loss": 1.0516, "step": 7419 }, { "epoch": 1.19, "grad_norm": 8.545439720153809, "learning_rate": 1.369559071285498e-05, "loss": 0.8447, "step": 7420 }, { "epoch": 1.19, "grad_norm": 9.905478477478027, "learning_rate": 1.3693975703167308e-05, "loss": 1.0064, "step": 7421 }, { "epoch": 1.19, "grad_norm": 8.68583869934082, "learning_rate": 1.3692360581898924e-05, "loss": 0.9072, "step": 7422 }, { "epoch": 1.19, "grad_norm": 9.945733070373535, "learning_rate": 1.3690745349098623e-05, "loss": 1.293, "step": 7423 }, { "epoch": 1.2, "grad_norm": 10.101339340209961, "learning_rate": 1.3689130004815185e-05, "loss": 1.1034, "step": 7424 }, { "epoch": 1.2, "grad_norm": 8.43459701538086, "learning_rate": 1.3687514549097411e-05, "loss": 1.2127, "step": 7425 }, { "epoch": 1.2, "grad_norm": 8.816518783569336, "learning_rate": 1.3685898981994095e-05, "loss": 1.0917, "step": 7426 }, { "epoch": 1.2, "grad_norm": 9.800806999206543, "learning_rate": 1.3684283303554037e-05, "loss": 0.9188, "step": 7427 }, { "epoch": 1.2, "grad_norm": 9.778701782226562, "learning_rate": 1.368266751382604e-05, "loss": 1.2984, "step": 7428 }, { "epoch": 1.2, "grad_norm": 9.661686897277832, "learning_rate": 1.3681051612858915e-05, "loss": 0.9862, "step": 7429 }, { "epoch": 1.2, "grad_norm": 8.710205078125, "learning_rate": 1.3679435600701467e-05, "loss": 1.1596, "step": 7430 }, { "epoch": 1.2, "grad_norm": 9.439435958862305, "learning_rate": 1.367781947740251e-05, "loss": 1.0015, "step": 7431 }, { "epoch": 1.2, "grad_norm": 8.685832977294922, "learning_rate": 1.3676203243010862e-05, "loss": 1.1008, "step": 7432 }, { "epoch": 1.2, "grad_norm": 7.999553680419922, "learning_rate": 1.3674586897575343e-05, "loss": 0.9379, "step": 7433 }, { "epoch": 1.2, "grad_norm": 9.863821983337402, "learning_rate": 1.3672970441144777e-05, "loss": 1.2738, "step": 7434 }, { "epoch": 1.2, "grad_norm": 7.691089153289795, "learning_rate": 1.3671353873767988e-05, "loss": 1.1316, "step": 7435 }, { "epoch": 1.2, "grad_norm": 9.81249713897705, "learning_rate": 1.3669737195493814e-05, "loss": 1.1315, "step": 7436 }, { "epoch": 1.2, "grad_norm": 9.599532127380371, "learning_rate": 1.3668120406371077e-05, "loss": 1.1257, "step": 7437 }, { "epoch": 1.2, "grad_norm": 12.734354019165039, "learning_rate": 1.3666503506448623e-05, "loss": 1.1007, "step": 7438 }, { "epoch": 1.2, "grad_norm": 8.186909675598145, "learning_rate": 1.366488649577529e-05, "loss": 1.0741, "step": 7439 }, { "epoch": 1.2, "grad_norm": 8.176855087280273, "learning_rate": 1.3663269374399922e-05, "loss": 1.2577, "step": 7440 }, { "epoch": 1.2, "grad_norm": 12.264549255371094, "learning_rate": 1.3661652142371365e-05, "loss": 1.3982, "step": 7441 }, { "epoch": 1.2, "grad_norm": 9.186836242675781, "learning_rate": 1.3660034799738466e-05, "loss": 1.0068, "step": 7442 }, { "epoch": 1.2, "grad_norm": 8.092926025390625, "learning_rate": 1.3658417346550086e-05, "loss": 0.9786, "step": 7443 }, { "epoch": 1.2, "grad_norm": 11.217039108276367, "learning_rate": 1.3656799782855077e-05, "loss": 1.3386, "step": 7444 }, { "epoch": 1.2, "grad_norm": 10.483612060546875, "learning_rate": 1.3655182108702301e-05, "loss": 1.2563, "step": 7445 }, { "epoch": 1.2, "grad_norm": 8.145255088806152, "learning_rate": 1.3653564324140617e-05, "loss": 1.174, "step": 7446 }, { "epoch": 1.2, "grad_norm": 7.54550838470459, "learning_rate": 1.36519464292189e-05, "loss": 0.9847, "step": 7447 }, { "epoch": 1.2, "grad_norm": 10.0186767578125, "learning_rate": 1.3650328423986018e-05, "loss": 1.1263, "step": 7448 }, { "epoch": 1.2, "grad_norm": 8.412940979003906, "learning_rate": 1.3648710308490842e-05, "loss": 0.9884, "step": 7449 }, { "epoch": 1.2, "grad_norm": 9.130136489868164, "learning_rate": 1.364709208278225e-05, "loss": 1.1208, "step": 7450 }, { "epoch": 1.2, "grad_norm": 7.8915696144104, "learning_rate": 1.3645473746909119e-05, "loss": 1.1591, "step": 7451 }, { "epoch": 1.2, "grad_norm": 7.084768772125244, "learning_rate": 1.3643855300920339e-05, "loss": 0.8021, "step": 7452 }, { "epoch": 1.2, "grad_norm": 6.761204242706299, "learning_rate": 1.3642236744864797e-05, "loss": 0.9495, "step": 7453 }, { "epoch": 1.2, "grad_norm": 10.200043678283691, "learning_rate": 1.3640618078791376e-05, "loss": 0.9294, "step": 7454 }, { "epoch": 1.2, "grad_norm": 7.6911468505859375, "learning_rate": 1.3638999302748976e-05, "loss": 1.177, "step": 7455 }, { "epoch": 1.2, "grad_norm": 7.842493534088135, "learning_rate": 1.3637380416786491e-05, "loss": 0.9855, "step": 7456 }, { "epoch": 1.2, "grad_norm": 10.209400177001953, "learning_rate": 1.3635761420952824e-05, "loss": 1.3811, "step": 7457 }, { "epoch": 1.2, "grad_norm": 9.611639976501465, "learning_rate": 1.3634142315296876e-05, "loss": 1.3063, "step": 7458 }, { "epoch": 1.2, "grad_norm": 8.236730575561523, "learning_rate": 1.3632523099867554e-05, "loss": 1.1947, "step": 7459 }, { "epoch": 1.2, "grad_norm": 8.387004852294922, "learning_rate": 1.363090377471377e-05, "loss": 0.9669, "step": 7460 }, { "epoch": 1.2, "grad_norm": 10.600809097290039, "learning_rate": 1.3629284339884438e-05, "loss": 1.2345, "step": 7461 }, { "epoch": 1.2, "grad_norm": 9.03909683227539, "learning_rate": 1.3627664795428473e-05, "loss": 1.2345, "step": 7462 }, { "epoch": 1.2, "grad_norm": 10.159571647644043, "learning_rate": 1.3626045141394793e-05, "loss": 0.8098, "step": 7463 }, { "epoch": 1.2, "grad_norm": 9.359383583068848, "learning_rate": 1.3624425377832327e-05, "loss": 0.8902, "step": 7464 }, { "epoch": 1.2, "grad_norm": 8.048516273498535, "learning_rate": 1.362280550479e-05, "loss": 1.1089, "step": 7465 }, { "epoch": 1.2, "grad_norm": 9.96839427947998, "learning_rate": 1.3621185522316736e-05, "loss": 0.8953, "step": 7466 }, { "epoch": 1.2, "grad_norm": 12.966384887695312, "learning_rate": 1.3619565430461475e-05, "loss": 1.1276, "step": 7467 }, { "epoch": 1.2, "grad_norm": 9.599138259887695, "learning_rate": 1.3617945229273156e-05, "loss": 1.1324, "step": 7468 }, { "epoch": 1.2, "grad_norm": 8.644073486328125, "learning_rate": 1.3616324918800712e-05, "loss": 0.8962, "step": 7469 }, { "epoch": 1.2, "grad_norm": 9.069342613220215, "learning_rate": 1.361470449909309e-05, "loss": 1.2237, "step": 7470 }, { "epoch": 1.2, "grad_norm": 11.261436462402344, "learning_rate": 1.3613083970199234e-05, "loss": 1.3076, "step": 7471 }, { "epoch": 1.2, "grad_norm": 8.40231990814209, "learning_rate": 1.36114633321681e-05, "loss": 1.2, "step": 7472 }, { "epoch": 1.2, "grad_norm": 7.896827697753906, "learning_rate": 1.3609842585048636e-05, "loss": 1.0904, "step": 7473 }, { "epoch": 1.2, "grad_norm": 9.17746353149414, "learning_rate": 1.3608221728889798e-05, "loss": 1.113, "step": 7474 }, { "epoch": 1.2, "grad_norm": 10.879009246826172, "learning_rate": 1.3606600763740548e-05, "loss": 1.1578, "step": 7475 }, { "epoch": 1.2, "grad_norm": 8.70734977722168, "learning_rate": 1.3604979689649848e-05, "loss": 1.0816, "step": 7476 }, { "epoch": 1.2, "grad_norm": 8.12201976776123, "learning_rate": 1.3603358506666667e-05, "loss": 0.8515, "step": 7477 }, { "epoch": 1.2, "grad_norm": 7.754678249359131, "learning_rate": 1.360173721483997e-05, "loss": 0.7729, "step": 7478 }, { "epoch": 1.2, "grad_norm": 7.941917419433594, "learning_rate": 1.3600115814218733e-05, "loss": 1.0741, "step": 7479 }, { "epoch": 1.2, "grad_norm": 10.043691635131836, "learning_rate": 1.3598494304851934e-05, "loss": 1.2425, "step": 7480 }, { "epoch": 1.2, "grad_norm": 8.836338996887207, "learning_rate": 1.359687268678855e-05, "loss": 1.1729, "step": 7481 }, { "epoch": 1.2, "grad_norm": 8.926053047180176, "learning_rate": 1.3595250960077561e-05, "loss": 1.1348, "step": 7482 }, { "epoch": 1.2, "grad_norm": 7.309967517852783, "learning_rate": 1.3593629124767957e-05, "loss": 1.2991, "step": 7483 }, { "epoch": 1.2, "grad_norm": 8.956528663635254, "learning_rate": 1.359200718090873e-05, "loss": 0.9763, "step": 7484 }, { "epoch": 1.2, "grad_norm": 9.599937438964844, "learning_rate": 1.3590385128548866e-05, "loss": 1.0656, "step": 7485 }, { "epoch": 1.21, "grad_norm": 8.363785743713379, "learning_rate": 1.3588762967737364e-05, "loss": 0.9008, "step": 7486 }, { "epoch": 1.21, "grad_norm": 10.195618629455566, "learning_rate": 1.3587140698523225e-05, "loss": 1.1094, "step": 7487 }, { "epoch": 1.21, "grad_norm": 8.970442771911621, "learning_rate": 1.3585518320955449e-05, "loss": 1.0674, "step": 7488 }, { "epoch": 1.21, "grad_norm": 8.400004386901855, "learning_rate": 1.3583895835083042e-05, "loss": 1.1799, "step": 7489 }, { "epoch": 1.21, "grad_norm": 8.154132843017578, "learning_rate": 1.3582273240955013e-05, "loss": 1.0235, "step": 7490 }, { "epoch": 1.21, "grad_norm": 10.886543273925781, "learning_rate": 1.3580650538620374e-05, "loss": 1.1221, "step": 7491 }, { "epoch": 1.21, "grad_norm": 7.746565341949463, "learning_rate": 1.3579027728128146e-05, "loss": 0.8777, "step": 7492 }, { "epoch": 1.21, "grad_norm": 7.189146995544434, "learning_rate": 1.3577404809527342e-05, "loss": 0.8855, "step": 7493 }, { "epoch": 1.21, "grad_norm": 7.448007106781006, "learning_rate": 1.3575781782866982e-05, "loss": 0.8666, "step": 7494 }, { "epoch": 1.21, "grad_norm": 9.519905090332031, "learning_rate": 1.3574158648196096e-05, "loss": 0.8531, "step": 7495 }, { "epoch": 1.21, "grad_norm": 8.347193717956543, "learning_rate": 1.3572535405563712e-05, "loss": 0.9965, "step": 7496 }, { "epoch": 1.21, "grad_norm": 7.973623275756836, "learning_rate": 1.357091205501886e-05, "loss": 0.8498, "step": 7497 }, { "epoch": 1.21, "grad_norm": 11.956195831298828, "learning_rate": 1.3569288596610576e-05, "loss": 1.1033, "step": 7498 }, { "epoch": 1.21, "grad_norm": 9.900035858154297, "learning_rate": 1.3567665030387901e-05, "loss": 0.9558, "step": 7499 }, { "epoch": 1.21, "grad_norm": 8.748536109924316, "learning_rate": 1.3566041356399872e-05, "loss": 0.6432, "step": 7500 }, { "epoch": 1.21, "grad_norm": 9.056909561157227, "learning_rate": 1.3564417574695539e-05, "loss": 0.8693, "step": 7501 }, { "epoch": 1.21, "grad_norm": 11.058225631713867, "learning_rate": 1.3562793685323945e-05, "loss": 1.0712, "step": 7502 }, { "epoch": 1.21, "grad_norm": 9.18092155456543, "learning_rate": 1.3561169688334144e-05, "loss": 1.2037, "step": 7503 }, { "epoch": 1.21, "grad_norm": 9.691411018371582, "learning_rate": 1.3559545583775193e-05, "loss": 1.1413, "step": 7504 }, { "epoch": 1.21, "grad_norm": 9.821565628051758, "learning_rate": 1.3557921371696145e-05, "loss": 1.1267, "step": 7505 }, { "epoch": 1.21, "grad_norm": 9.252023696899414, "learning_rate": 1.3556297052146062e-05, "loss": 0.9182, "step": 7506 }, { "epoch": 1.21, "grad_norm": 10.157856941223145, "learning_rate": 1.3554672625174013e-05, "loss": 1.0011, "step": 7507 }, { "epoch": 1.21, "grad_norm": 10.127416610717773, "learning_rate": 1.355304809082906e-05, "loss": 1.1908, "step": 7508 }, { "epoch": 1.21, "grad_norm": 9.379813194274902, "learning_rate": 1.355142344916028e-05, "loss": 0.9862, "step": 7509 }, { "epoch": 1.21, "grad_norm": 10.407733917236328, "learning_rate": 1.3549798700216742e-05, "loss": 0.8777, "step": 7510 }, { "epoch": 1.21, "grad_norm": 11.599815368652344, "learning_rate": 1.3548173844047523e-05, "loss": 1.5017, "step": 7511 }, { "epoch": 1.21, "grad_norm": 8.612593650817871, "learning_rate": 1.3546548880701708e-05, "loss": 1.0846, "step": 7512 }, { "epoch": 1.21, "grad_norm": 10.846104621887207, "learning_rate": 1.3544923810228376e-05, "loss": 1.1611, "step": 7513 }, { "epoch": 1.21, "grad_norm": 10.924612045288086, "learning_rate": 1.3543298632676617e-05, "loss": 1.1274, "step": 7514 }, { "epoch": 1.21, "grad_norm": 7.2229485511779785, "learning_rate": 1.3541673348095526e-05, "loss": 1.0138, "step": 7515 }, { "epoch": 1.21, "grad_norm": 9.1146879196167, "learning_rate": 1.3540047956534186e-05, "loss": 1.0596, "step": 7516 }, { "epoch": 1.21, "grad_norm": 9.028702735900879, "learning_rate": 1.35384224580417e-05, "loss": 1.1555, "step": 7517 }, { "epoch": 1.21, "grad_norm": 10.901616096496582, "learning_rate": 1.3536796852667166e-05, "loss": 1.4124, "step": 7518 }, { "epoch": 1.21, "grad_norm": 7.688583850860596, "learning_rate": 1.3535171140459694e-05, "loss": 0.9195, "step": 7519 }, { "epoch": 1.21, "grad_norm": 9.230728149414062, "learning_rate": 1.353354532146838e-05, "loss": 0.8441, "step": 7520 }, { "epoch": 1.21, "grad_norm": 8.238652229309082, "learning_rate": 1.3531919395742342e-05, "loss": 1.2857, "step": 7521 }, { "epoch": 1.21, "grad_norm": 9.893803596496582, "learning_rate": 1.3530293363330687e-05, "loss": 1.1364, "step": 7522 }, { "epoch": 1.21, "grad_norm": 8.611265182495117, "learning_rate": 1.3528667224282535e-05, "loss": 0.9835, "step": 7523 }, { "epoch": 1.21, "grad_norm": 7.271573066711426, "learning_rate": 1.3527040978647004e-05, "loss": 0.9086, "step": 7524 }, { "epoch": 1.21, "grad_norm": 10.615224838256836, "learning_rate": 1.3525414626473214e-05, "loss": 1.2246, "step": 7525 }, { "epoch": 1.21, "grad_norm": 10.034602165222168, "learning_rate": 1.3523788167810296e-05, "loss": 1.0341, "step": 7526 }, { "epoch": 1.21, "grad_norm": 10.18492603302002, "learning_rate": 1.3522161602707377e-05, "loss": 1.1794, "step": 7527 }, { "epoch": 1.21, "grad_norm": 8.977193832397461, "learning_rate": 1.3520534931213587e-05, "loss": 1.214, "step": 7528 }, { "epoch": 1.21, "grad_norm": 8.92109203338623, "learning_rate": 1.3518908153378065e-05, "loss": 1.1438, "step": 7529 }, { "epoch": 1.21, "grad_norm": 7.723555564880371, "learning_rate": 1.3517281269249947e-05, "loss": 1.1053, "step": 7530 }, { "epoch": 1.21, "grad_norm": 10.001875877380371, "learning_rate": 1.3515654278878377e-05, "loss": 1.182, "step": 7531 }, { "epoch": 1.21, "grad_norm": 11.36411190032959, "learning_rate": 1.3514027182312497e-05, "loss": 1.3515, "step": 7532 }, { "epoch": 1.21, "grad_norm": 9.358919143676758, "learning_rate": 1.3512399979601457e-05, "loss": 0.9568, "step": 7533 }, { "epoch": 1.21, "grad_norm": 8.454988479614258, "learning_rate": 1.351077267079441e-05, "loss": 0.9701, "step": 7534 }, { "epoch": 1.21, "grad_norm": 10.7344388961792, "learning_rate": 1.350914525594051e-05, "loss": 1.1751, "step": 7535 }, { "epoch": 1.21, "grad_norm": 8.754159927368164, "learning_rate": 1.3507517735088913e-05, "loss": 1.0621, "step": 7536 }, { "epoch": 1.21, "grad_norm": 9.13187026977539, "learning_rate": 1.350589010828878e-05, "loss": 1.1177, "step": 7537 }, { "epoch": 1.21, "grad_norm": 9.488889694213867, "learning_rate": 1.3504262375589279e-05, "loss": 1.2702, "step": 7538 }, { "epoch": 1.21, "grad_norm": 7.402856349945068, "learning_rate": 1.3502634537039575e-05, "loss": 1.143, "step": 7539 }, { "epoch": 1.21, "grad_norm": 8.317760467529297, "learning_rate": 1.3501006592688834e-05, "loss": 1.1003, "step": 7540 }, { "epoch": 1.21, "grad_norm": 8.658185958862305, "learning_rate": 1.3499378542586241e-05, "loss": 1.2886, "step": 7541 }, { "epoch": 1.21, "grad_norm": 9.62724781036377, "learning_rate": 1.3497750386780963e-05, "loss": 1.1803, "step": 7542 }, { "epoch": 1.21, "grad_norm": 9.384613990783691, "learning_rate": 1.3496122125322187e-05, "loss": 0.9734, "step": 7543 }, { "epoch": 1.21, "grad_norm": 9.56485366821289, "learning_rate": 1.349449375825909e-05, "loss": 0.9839, "step": 7544 }, { "epoch": 1.21, "grad_norm": 8.986946105957031, "learning_rate": 1.3492865285640865e-05, "loss": 1.0866, "step": 7545 }, { "epoch": 1.21, "grad_norm": 8.0443754196167, "learning_rate": 1.34912367075167e-05, "loss": 1.057, "step": 7546 }, { "epoch": 1.21, "grad_norm": 10.677006721496582, "learning_rate": 1.3489608023935783e-05, "loss": 1.2958, "step": 7547 }, { "epoch": 1.22, "grad_norm": 9.630624771118164, "learning_rate": 1.3487979234947317e-05, "loss": 1.0438, "step": 7548 }, { "epoch": 1.22, "grad_norm": 8.137940406799316, "learning_rate": 1.3486350340600498e-05, "loss": 0.7692, "step": 7549 }, { "epoch": 1.22, "grad_norm": 8.7324857711792, "learning_rate": 1.3484721340944533e-05, "loss": 1.0784, "step": 7550 }, { "epoch": 1.22, "grad_norm": 10.448552131652832, "learning_rate": 1.348309223602862e-05, "loss": 1.148, "step": 7551 }, { "epoch": 1.22, "grad_norm": 7.394539833068848, "learning_rate": 1.3481463025901973e-05, "loss": 0.8652, "step": 7552 }, { "epoch": 1.22, "grad_norm": 8.545572280883789, "learning_rate": 1.3479833710613803e-05, "loss": 1.0385, "step": 7553 }, { "epoch": 1.22, "grad_norm": 8.995245933532715, "learning_rate": 1.3478204290213329e-05, "loss": 0.9989, "step": 7554 }, { "epoch": 1.22, "grad_norm": 7.8021464347839355, "learning_rate": 1.347657476474976e-05, "loss": 0.8178, "step": 7555 }, { "epoch": 1.22, "grad_norm": 7.975088596343994, "learning_rate": 1.3474945134272329e-05, "loss": 0.8649, "step": 7556 }, { "epoch": 1.22, "grad_norm": 9.004959106445312, "learning_rate": 1.3473315398830252e-05, "loss": 1.0063, "step": 7557 }, { "epoch": 1.22, "grad_norm": 9.665156364440918, "learning_rate": 1.3471685558472762e-05, "loss": 1.1671, "step": 7558 }, { "epoch": 1.22, "grad_norm": 8.04263687133789, "learning_rate": 1.3470055613249086e-05, "loss": 0.8839, "step": 7559 }, { "epoch": 1.22, "grad_norm": 7.892040729522705, "learning_rate": 1.3468425563208464e-05, "loss": 0.9349, "step": 7560 }, { "epoch": 1.22, "grad_norm": 7.78003454208374, "learning_rate": 1.3466795408400131e-05, "loss": 0.9965, "step": 7561 }, { "epoch": 1.22, "grad_norm": 9.093727111816406, "learning_rate": 1.3465165148873327e-05, "loss": 1.0807, "step": 7562 }, { "epoch": 1.22, "grad_norm": 9.357039451599121, "learning_rate": 1.3463534784677298e-05, "loss": 1.1342, "step": 7563 }, { "epoch": 1.22, "grad_norm": 8.819849967956543, "learning_rate": 1.3461904315861283e-05, "loss": 0.9494, "step": 7564 }, { "epoch": 1.22, "grad_norm": 7.853682994842529, "learning_rate": 1.3460273742474541e-05, "loss": 0.9419, "step": 7565 }, { "epoch": 1.22, "grad_norm": 11.327653884887695, "learning_rate": 1.3458643064566326e-05, "loss": 1.2201, "step": 7566 }, { "epoch": 1.22, "grad_norm": 10.194704055786133, "learning_rate": 1.3457012282185883e-05, "loss": 1.1472, "step": 7567 }, { "epoch": 1.22, "grad_norm": 8.724865913391113, "learning_rate": 1.3455381395382486e-05, "loss": 0.984, "step": 7568 }, { "epoch": 1.22, "grad_norm": 10.614148139953613, "learning_rate": 1.3453750404205388e-05, "loss": 1.1584, "step": 7569 }, { "epoch": 1.22, "grad_norm": 10.80233383178711, "learning_rate": 1.3452119308703861e-05, "loss": 0.8728, "step": 7570 }, { "epoch": 1.22, "grad_norm": 9.441656112670898, "learning_rate": 1.3450488108927171e-05, "loss": 1.1678, "step": 7571 }, { "epoch": 1.22, "grad_norm": 11.784070014953613, "learning_rate": 1.344885680492459e-05, "loss": 1.2403, "step": 7572 }, { "epoch": 1.22, "grad_norm": 10.740382194519043, "learning_rate": 1.3447225396745393e-05, "loss": 0.8809, "step": 7573 }, { "epoch": 1.22, "grad_norm": 8.72681999206543, "learning_rate": 1.3445593884438864e-05, "loss": 0.9771, "step": 7574 }, { "epoch": 1.22, "grad_norm": 10.73802375793457, "learning_rate": 1.3443962268054274e-05, "loss": 1.2083, "step": 7575 }, { "epoch": 1.22, "grad_norm": 11.691032409667969, "learning_rate": 1.3442330547640916e-05, "loss": 1.3337, "step": 7576 }, { "epoch": 1.22, "grad_norm": 9.023886680603027, "learning_rate": 1.3440698723248079e-05, "loss": 0.9791, "step": 7577 }, { "epoch": 1.22, "grad_norm": 7.8253374099731445, "learning_rate": 1.3439066794925046e-05, "loss": 0.9282, "step": 7578 }, { "epoch": 1.22, "grad_norm": 8.572802543640137, "learning_rate": 1.3437434762721118e-05, "loss": 0.8652, "step": 7579 }, { "epoch": 1.22, "grad_norm": 9.681546211242676, "learning_rate": 1.3435802626685593e-05, "loss": 1.0904, "step": 7580 }, { "epoch": 1.22, "grad_norm": 9.735576629638672, "learning_rate": 1.3434170386867766e-05, "loss": 1.1632, "step": 7581 }, { "epoch": 1.22, "grad_norm": 7.79676628112793, "learning_rate": 1.3432538043316944e-05, "loss": 1.0959, "step": 7582 }, { "epoch": 1.22, "grad_norm": 9.858942031860352, "learning_rate": 1.3430905596082437e-05, "loss": 1.0541, "step": 7583 }, { "epoch": 1.22, "grad_norm": 12.093055725097656, "learning_rate": 1.342927304521355e-05, "loss": 1.1372, "step": 7584 }, { "epoch": 1.22, "grad_norm": 7.4894633293151855, "learning_rate": 1.3427640390759595e-05, "loss": 0.9513, "step": 7585 }, { "epoch": 1.22, "grad_norm": 10.87883472442627, "learning_rate": 1.3426007632769896e-05, "loss": 1.2028, "step": 7586 }, { "epoch": 1.22, "grad_norm": 9.950940132141113, "learning_rate": 1.342437477129376e-05, "loss": 0.9758, "step": 7587 }, { "epoch": 1.22, "grad_norm": 9.9876070022583, "learning_rate": 1.3422741806380519e-05, "loss": 1.0205, "step": 7588 }, { "epoch": 1.22, "grad_norm": 9.465653419494629, "learning_rate": 1.34211087380795e-05, "loss": 1.0635, "step": 7589 }, { "epoch": 1.22, "grad_norm": 12.839250564575195, "learning_rate": 1.3419475566440023e-05, "loss": 0.8858, "step": 7590 }, { "epoch": 1.22, "grad_norm": 9.568900108337402, "learning_rate": 1.3417842291511427e-05, "loss": 1.3174, "step": 7591 }, { "epoch": 1.22, "grad_norm": 9.905402183532715, "learning_rate": 1.3416208913343042e-05, "loss": 1.3185, "step": 7592 }, { "epoch": 1.22, "grad_norm": 9.578526496887207, "learning_rate": 1.3414575431984213e-05, "loss": 1.2864, "step": 7593 }, { "epoch": 1.22, "grad_norm": 8.684876441955566, "learning_rate": 1.3412941847484274e-05, "loss": 1.0911, "step": 7594 }, { "epoch": 1.22, "grad_norm": 11.302541732788086, "learning_rate": 1.3411308159892571e-05, "loss": 0.9602, "step": 7595 }, { "epoch": 1.22, "grad_norm": 7.895712375640869, "learning_rate": 1.3409674369258452e-05, "loss": 1.1322, "step": 7596 }, { "epoch": 1.22, "grad_norm": 7.172527313232422, "learning_rate": 1.3408040475631267e-05, "loss": 0.7666, "step": 7597 }, { "epoch": 1.22, "grad_norm": 8.314620971679688, "learning_rate": 1.3406406479060372e-05, "loss": 1.2145, "step": 7598 }, { "epoch": 1.22, "grad_norm": 9.470501899719238, "learning_rate": 1.3404772379595121e-05, "loss": 1.1655, "step": 7599 }, { "epoch": 1.22, "grad_norm": 10.777369499206543, "learning_rate": 1.3403138177284873e-05, "loss": 1.1151, "step": 7600 }, { "epoch": 1.22, "grad_norm": 8.963749885559082, "learning_rate": 1.3401503872178998e-05, "loss": 0.8474, "step": 7601 }, { "epoch": 1.22, "grad_norm": 9.259711265563965, "learning_rate": 1.3399869464326853e-05, "loss": 0.9711, "step": 7602 }, { "epoch": 1.22, "grad_norm": 7.837844371795654, "learning_rate": 1.339823495377781e-05, "loss": 1.0006, "step": 7603 }, { "epoch": 1.22, "grad_norm": 9.428647994995117, "learning_rate": 1.3396600340581246e-05, "loss": 0.9969, "step": 7604 }, { "epoch": 1.22, "grad_norm": 10.371742248535156, "learning_rate": 1.339496562478653e-05, "loss": 1.078, "step": 7605 }, { "epoch": 1.22, "grad_norm": 8.59001636505127, "learning_rate": 1.3393330806443045e-05, "loss": 0.9372, "step": 7606 }, { "epoch": 1.22, "grad_norm": 9.4668550491333, "learning_rate": 1.3391695885600169e-05, "loss": 1.0607, "step": 7607 }, { "epoch": 1.22, "grad_norm": 7.889372825622559, "learning_rate": 1.3390060862307285e-05, "loss": 0.9973, "step": 7608 }, { "epoch": 1.22, "grad_norm": 9.652132987976074, "learning_rate": 1.3388425736613789e-05, "loss": 1.1524, "step": 7609 }, { "epoch": 1.23, "grad_norm": 9.084736824035645, "learning_rate": 1.3386790508569062e-05, "loss": 1.097, "step": 7610 }, { "epoch": 1.23, "grad_norm": 9.752715110778809, "learning_rate": 1.3385155178222507e-05, "loss": 1.1954, "step": 7611 }, { "epoch": 1.23, "grad_norm": 9.527169227600098, "learning_rate": 1.3383519745623512e-05, "loss": 1.0402, "step": 7612 }, { "epoch": 1.23, "grad_norm": 10.691752433776855, "learning_rate": 1.3381884210821487e-05, "loss": 0.914, "step": 7613 }, { "epoch": 1.23, "grad_norm": 8.281649589538574, "learning_rate": 1.3380248573865825e-05, "loss": 0.8863, "step": 7614 }, { "epoch": 1.23, "grad_norm": 11.444551467895508, "learning_rate": 1.3378612834805939e-05, "loss": 1.267, "step": 7615 }, { "epoch": 1.23, "grad_norm": 11.241826057434082, "learning_rate": 1.3376976993691237e-05, "loss": 1.3286, "step": 7616 }, { "epoch": 1.23, "grad_norm": 8.885002136230469, "learning_rate": 1.337534105057113e-05, "loss": 0.9482, "step": 7617 }, { "epoch": 1.23, "grad_norm": 9.173299789428711, "learning_rate": 1.3373705005495034e-05, "loss": 0.8152, "step": 7618 }, { "epoch": 1.23, "grad_norm": 7.093904495239258, "learning_rate": 1.3372068858512367e-05, "loss": 0.8208, "step": 7619 }, { "epoch": 1.23, "grad_norm": 9.368677139282227, "learning_rate": 1.3370432609672554e-05, "loss": 1.1176, "step": 7620 }, { "epoch": 1.23, "grad_norm": 7.810057163238525, "learning_rate": 1.3368796259025019e-05, "loss": 0.9452, "step": 7621 }, { "epoch": 1.23, "grad_norm": 9.606534957885742, "learning_rate": 1.3367159806619185e-05, "loss": 1.1702, "step": 7622 }, { "epoch": 1.23, "grad_norm": 9.32617473602295, "learning_rate": 1.3365523252504486e-05, "loss": 1.078, "step": 7623 }, { "epoch": 1.23, "grad_norm": 10.086827278137207, "learning_rate": 1.3363886596730359e-05, "loss": 0.9287, "step": 7624 }, { "epoch": 1.23, "grad_norm": 7.76769495010376, "learning_rate": 1.3362249839346235e-05, "loss": 0.9808, "step": 7625 }, { "epoch": 1.23, "grad_norm": 8.815394401550293, "learning_rate": 1.3360612980401558e-05, "loss": 1.2243, "step": 7626 }, { "epoch": 1.23, "grad_norm": 7.85776948928833, "learning_rate": 1.335897601994577e-05, "loss": 0.8588, "step": 7627 }, { "epoch": 1.23, "grad_norm": 8.392475128173828, "learning_rate": 1.3357338958028318e-05, "loss": 1.0948, "step": 7628 }, { "epoch": 1.23, "grad_norm": 9.585761070251465, "learning_rate": 1.335570179469865e-05, "loss": 0.9362, "step": 7629 }, { "epoch": 1.23, "grad_norm": 8.189934730529785, "learning_rate": 1.335406453000622e-05, "loss": 1.0109, "step": 7630 }, { "epoch": 1.23, "grad_norm": 9.245251655578613, "learning_rate": 1.3352427164000481e-05, "loss": 1.3478, "step": 7631 }, { "epoch": 1.23, "grad_norm": 10.894214630126953, "learning_rate": 1.3350789696730897e-05, "loss": 1.0435, "step": 7632 }, { "epoch": 1.23, "grad_norm": 10.261283874511719, "learning_rate": 1.3349152128246922e-05, "loss": 1.1687, "step": 7633 }, { "epoch": 1.23, "grad_norm": 7.886005401611328, "learning_rate": 1.3347514458598028e-05, "loss": 0.9607, "step": 7634 }, { "epoch": 1.23, "grad_norm": 8.600652694702148, "learning_rate": 1.3345876687833677e-05, "loss": 0.9994, "step": 7635 }, { "epoch": 1.23, "grad_norm": 8.057025909423828, "learning_rate": 1.3344238816003341e-05, "loss": 0.7193, "step": 7636 }, { "epoch": 1.23, "grad_norm": 9.191790580749512, "learning_rate": 1.3342600843156498e-05, "loss": 1.1648, "step": 7637 }, { "epoch": 1.23, "grad_norm": 8.588001251220703, "learning_rate": 1.3340962769342615e-05, "loss": 1.0951, "step": 7638 }, { "epoch": 1.23, "grad_norm": 8.334392547607422, "learning_rate": 1.3339324594611181e-05, "loss": 0.8619, "step": 7639 }, { "epoch": 1.23, "grad_norm": 8.253408432006836, "learning_rate": 1.3337686319011677e-05, "loss": 1.0924, "step": 7640 }, { "epoch": 1.23, "grad_norm": 10.789095878601074, "learning_rate": 1.333604794259359e-05, "loss": 1.1096, "step": 7641 }, { "epoch": 1.23, "grad_norm": 10.093347549438477, "learning_rate": 1.3334409465406404e-05, "loss": 1.0552, "step": 7642 }, { "epoch": 1.23, "grad_norm": 10.381689071655273, "learning_rate": 1.3332770887499613e-05, "loss": 1.0303, "step": 7643 }, { "epoch": 1.23, "grad_norm": 10.271485328674316, "learning_rate": 1.3331132208922717e-05, "loss": 1.3438, "step": 7644 }, { "epoch": 1.23, "grad_norm": 8.819138526916504, "learning_rate": 1.3329493429725209e-05, "loss": 1.1532, "step": 7645 }, { "epoch": 1.23, "grad_norm": 10.691271781921387, "learning_rate": 1.3327854549956589e-05, "loss": 0.9747, "step": 7646 }, { "epoch": 1.23, "grad_norm": 10.626565933227539, "learning_rate": 1.3326215569666368e-05, "loss": 1.2718, "step": 7647 }, { "epoch": 1.23, "grad_norm": 8.159613609313965, "learning_rate": 1.3324576488904045e-05, "loss": 0.9914, "step": 7648 }, { "epoch": 1.23, "grad_norm": 9.606034278869629, "learning_rate": 1.3322937307719137e-05, "loss": 1.1852, "step": 7649 }, { "epoch": 1.23, "grad_norm": 9.027243614196777, "learning_rate": 1.3321298026161151e-05, "loss": 1.2188, "step": 7650 }, { "epoch": 1.23, "grad_norm": 8.909415245056152, "learning_rate": 1.3319658644279612e-05, "loss": 0.8256, "step": 7651 }, { "epoch": 1.23, "grad_norm": 11.147261619567871, "learning_rate": 1.3318019162124031e-05, "loss": 1.1573, "step": 7652 }, { "epoch": 1.23, "grad_norm": 9.557108879089355, "learning_rate": 1.3316379579743934e-05, "loss": 1.0787, "step": 7653 }, { "epoch": 1.23, "grad_norm": 8.529247283935547, "learning_rate": 1.3314739897188846e-05, "loss": 1.108, "step": 7654 }, { "epoch": 1.23, "grad_norm": 10.28093433380127, "learning_rate": 1.3313100114508297e-05, "loss": 1.0944, "step": 7655 }, { "epoch": 1.23, "grad_norm": 11.870556831359863, "learning_rate": 1.3311460231751818e-05, "loss": 1.4892, "step": 7656 }, { "epoch": 1.23, "grad_norm": 7.3594231605529785, "learning_rate": 1.330982024896894e-05, "loss": 1.1251, "step": 7657 }, { "epoch": 1.23, "grad_norm": 10.118523597717285, "learning_rate": 1.3308180166209204e-05, "loss": 1.0442, "step": 7658 }, { "epoch": 1.23, "grad_norm": 8.412755966186523, "learning_rate": 1.3306539983522151e-05, "loss": 1.0175, "step": 7659 }, { "epoch": 1.23, "grad_norm": 7.183824062347412, "learning_rate": 1.3304899700957324e-05, "loss": 1.0019, "step": 7660 }, { "epoch": 1.23, "grad_norm": 10.494199752807617, "learning_rate": 1.3303259318564265e-05, "loss": 1.036, "step": 7661 }, { "epoch": 1.23, "grad_norm": 9.397789001464844, "learning_rate": 1.3301618836392529e-05, "loss": 1.1854, "step": 7662 }, { "epoch": 1.23, "grad_norm": 8.36605453491211, "learning_rate": 1.3299978254491669e-05, "loss": 1.0407, "step": 7663 }, { "epoch": 1.23, "grad_norm": 7.804925918579102, "learning_rate": 1.329833757291124e-05, "loss": 0.9757, "step": 7664 }, { "epoch": 1.23, "grad_norm": 9.163960456848145, "learning_rate": 1.3296696791700796e-05, "loss": 1.2073, "step": 7665 }, { "epoch": 1.23, "grad_norm": 7.672532558441162, "learning_rate": 1.3295055910909902e-05, "loss": 1.1099, "step": 7666 }, { "epoch": 1.23, "grad_norm": 8.75365924835205, "learning_rate": 1.3293414930588125e-05, "loss": 0.928, "step": 7667 }, { "epoch": 1.23, "grad_norm": 10.978270530700684, "learning_rate": 1.3291773850785025e-05, "loss": 1.0025, "step": 7668 }, { "epoch": 1.23, "grad_norm": 9.07325267791748, "learning_rate": 1.3290132671550185e-05, "loss": 0.9961, "step": 7669 }, { "epoch": 1.23, "grad_norm": 9.12264347076416, "learning_rate": 1.3288491392933167e-05, "loss": 1.1714, "step": 7670 }, { "epoch": 1.23, "grad_norm": 7.429653644561768, "learning_rate": 1.3286850014983555e-05, "loss": 0.8145, "step": 7671 }, { "epoch": 1.24, "grad_norm": 8.749836921691895, "learning_rate": 1.3285208537750925e-05, "loss": 0.9168, "step": 7672 }, { "epoch": 1.24, "grad_norm": 8.188065528869629, "learning_rate": 1.3283566961284859e-05, "loss": 0.9306, "step": 7673 }, { "epoch": 1.24, "grad_norm": 9.738105773925781, "learning_rate": 1.3281925285634947e-05, "loss": 1.2153, "step": 7674 }, { "epoch": 1.24, "grad_norm": 9.3132963180542, "learning_rate": 1.3280283510850774e-05, "loss": 1.1331, "step": 7675 }, { "epoch": 1.24, "grad_norm": 9.861857414245605, "learning_rate": 1.3278641636981935e-05, "loss": 1.3703, "step": 7676 }, { "epoch": 1.24, "grad_norm": 10.982272148132324, "learning_rate": 1.3276999664078019e-05, "loss": 1.1263, "step": 7677 }, { "epoch": 1.24, "grad_norm": 11.961945533752441, "learning_rate": 1.3275357592188627e-05, "loss": 1.0566, "step": 7678 }, { "epoch": 1.24, "grad_norm": 8.634065628051758, "learning_rate": 1.3273715421363362e-05, "loss": 0.9066, "step": 7679 }, { "epoch": 1.24, "grad_norm": 9.783744812011719, "learning_rate": 1.3272073151651822e-05, "loss": 1.2884, "step": 7680 }, { "epoch": 1.24, "grad_norm": 9.189825057983398, "learning_rate": 1.3270430783103621e-05, "loss": 1.1735, "step": 7681 }, { "epoch": 1.24, "grad_norm": 10.051164627075195, "learning_rate": 1.3268788315768361e-05, "loss": 1.2217, "step": 7682 }, { "epoch": 1.24, "grad_norm": 7.7734599113464355, "learning_rate": 1.3267145749695662e-05, "loss": 0.994, "step": 7683 }, { "epoch": 1.24, "grad_norm": 9.217801094055176, "learning_rate": 1.3265503084935131e-05, "loss": 1.0857, "step": 7684 }, { "epoch": 1.24, "grad_norm": 7.571924686431885, "learning_rate": 1.3263860321536394e-05, "loss": 1.1978, "step": 7685 }, { "epoch": 1.24, "grad_norm": 8.368816375732422, "learning_rate": 1.3262217459549069e-05, "loss": 0.9677, "step": 7686 }, { "epoch": 1.24, "grad_norm": 8.732857704162598, "learning_rate": 1.326057449902278e-05, "loss": 1.2157, "step": 7687 }, { "epoch": 1.24, "grad_norm": 9.183658599853516, "learning_rate": 1.3258931440007159e-05, "loss": 1.2665, "step": 7688 }, { "epoch": 1.24, "grad_norm": 7.805944442749023, "learning_rate": 1.3257288282551827e-05, "loss": 1.1685, "step": 7689 }, { "epoch": 1.24, "grad_norm": 10.011189460754395, "learning_rate": 1.3255645026706429e-05, "loss": 1.096, "step": 7690 }, { "epoch": 1.24, "grad_norm": 8.579048156738281, "learning_rate": 1.3254001672520596e-05, "loss": 0.917, "step": 7691 }, { "epoch": 1.24, "grad_norm": 8.462843894958496, "learning_rate": 1.3252358220043966e-05, "loss": 1.0296, "step": 7692 }, { "epoch": 1.24, "grad_norm": 8.095311164855957, "learning_rate": 1.3250714669326182e-05, "loss": 0.887, "step": 7693 }, { "epoch": 1.24, "grad_norm": 9.422250747680664, "learning_rate": 1.3249071020416891e-05, "loss": 0.9897, "step": 7694 }, { "epoch": 1.24, "grad_norm": 8.142679214477539, "learning_rate": 1.3247427273365738e-05, "loss": 1.3051, "step": 7695 }, { "epoch": 1.24, "grad_norm": 9.125531196594238, "learning_rate": 1.3245783428222378e-05, "loss": 1.0214, "step": 7696 }, { "epoch": 1.24, "grad_norm": 8.793488502502441, "learning_rate": 1.3244139485036464e-05, "loss": 1.0516, "step": 7697 }, { "epoch": 1.24, "grad_norm": 7.004580020904541, "learning_rate": 1.3242495443857652e-05, "loss": 0.7843, "step": 7698 }, { "epoch": 1.24, "grad_norm": 8.30892562866211, "learning_rate": 1.3240851304735605e-05, "loss": 0.9673, "step": 7699 }, { "epoch": 1.24, "grad_norm": 7.476660251617432, "learning_rate": 1.3239207067719982e-05, "loss": 1.1174, "step": 7700 }, { "epoch": 1.24, "grad_norm": 9.955266952514648, "learning_rate": 1.3237562732860449e-05, "loss": 1.0557, "step": 7701 }, { "epoch": 1.24, "grad_norm": 8.947866439819336, "learning_rate": 1.323591830020668e-05, "loss": 0.8381, "step": 7702 }, { "epoch": 1.24, "grad_norm": 9.121808052062988, "learning_rate": 1.3234273769808346e-05, "loss": 0.7985, "step": 7703 }, { "epoch": 1.24, "grad_norm": 8.69579792022705, "learning_rate": 1.3232629141715116e-05, "loss": 1.1129, "step": 7704 }, { "epoch": 1.24, "grad_norm": 8.46009635925293, "learning_rate": 1.3230984415976673e-05, "loss": 1.2128, "step": 7705 }, { "epoch": 1.24, "grad_norm": 8.552483558654785, "learning_rate": 1.3229339592642697e-05, "loss": 1.1238, "step": 7706 }, { "epoch": 1.24, "grad_norm": 8.15689468383789, "learning_rate": 1.3227694671762871e-05, "loss": 0.9634, "step": 7707 }, { "epoch": 1.24, "grad_norm": 9.848986625671387, "learning_rate": 1.3226049653386883e-05, "loss": 0.8844, "step": 7708 }, { "epoch": 1.24, "grad_norm": 14.175806999206543, "learning_rate": 1.322440453756442e-05, "loss": 0.9845, "step": 7709 }, { "epoch": 1.24, "grad_norm": 9.381083488464355, "learning_rate": 1.3222759324345178e-05, "loss": 0.9372, "step": 7710 }, { "epoch": 1.24, "grad_norm": 10.276142120361328, "learning_rate": 1.3221114013778849e-05, "loss": 1.0899, "step": 7711 }, { "epoch": 1.24, "grad_norm": 10.655620574951172, "learning_rate": 1.3219468605915134e-05, "loss": 1.0129, "step": 7712 }, { "epoch": 1.24, "grad_norm": 9.73383617401123, "learning_rate": 1.3217823100803733e-05, "loss": 1.1723, "step": 7713 }, { "epoch": 1.24, "grad_norm": 8.789922714233398, "learning_rate": 1.3216177498494349e-05, "loss": 0.9688, "step": 7714 }, { "epoch": 1.24, "grad_norm": 8.989805221557617, "learning_rate": 1.3214531799036696e-05, "loss": 0.9518, "step": 7715 }, { "epoch": 1.24, "grad_norm": 9.526802062988281, "learning_rate": 1.3212886002480479e-05, "loss": 1.3986, "step": 7716 }, { "epoch": 1.24, "grad_norm": 8.767379760742188, "learning_rate": 1.3211240108875407e-05, "loss": 1.2109, "step": 7717 }, { "epoch": 1.24, "grad_norm": 9.387832641601562, "learning_rate": 1.3209594118271204e-05, "loss": 1.0907, "step": 7718 }, { "epoch": 1.24, "grad_norm": 9.228643417358398, "learning_rate": 1.3207948030717583e-05, "loss": 1.2674, "step": 7719 }, { "epoch": 1.24, "grad_norm": 8.306742668151855, "learning_rate": 1.3206301846264272e-05, "loss": 0.999, "step": 7720 }, { "epoch": 1.24, "grad_norm": 8.352768898010254, "learning_rate": 1.3204655564960989e-05, "loss": 1.0641, "step": 7721 }, { "epoch": 1.24, "grad_norm": 8.252992630004883, "learning_rate": 1.3203009186857467e-05, "loss": 0.8514, "step": 7722 }, { "epoch": 1.24, "grad_norm": 9.35330581665039, "learning_rate": 1.3201362712003435e-05, "loss": 0.877, "step": 7723 }, { "epoch": 1.24, "grad_norm": 11.961780548095703, "learning_rate": 1.3199716140448629e-05, "loss": 1.1779, "step": 7724 }, { "epoch": 1.24, "grad_norm": 9.97429370880127, "learning_rate": 1.3198069472242778e-05, "loss": 1.0823, "step": 7725 }, { "epoch": 1.24, "grad_norm": 7.402737140655518, "learning_rate": 1.319642270743563e-05, "loss": 0.8624, "step": 7726 }, { "epoch": 1.24, "grad_norm": 8.772634506225586, "learning_rate": 1.3194775846076924e-05, "loss": 1.096, "step": 7727 }, { "epoch": 1.24, "grad_norm": 8.491979598999023, "learning_rate": 1.3193128888216405e-05, "loss": 1.0728, "step": 7728 }, { "epoch": 1.24, "grad_norm": 8.345480918884277, "learning_rate": 1.3191481833903818e-05, "loss": 0.8746, "step": 7729 }, { "epoch": 1.24, "grad_norm": 9.109600067138672, "learning_rate": 1.3189834683188923e-05, "loss": 1.0641, "step": 7730 }, { "epoch": 1.24, "grad_norm": 8.09581184387207, "learning_rate": 1.3188187436121465e-05, "loss": 0.8997, "step": 7731 }, { "epoch": 1.24, "grad_norm": 10.654870986938477, "learning_rate": 1.3186540092751208e-05, "loss": 1.2665, "step": 7732 }, { "epoch": 1.24, "grad_norm": 8.64168643951416, "learning_rate": 1.3184892653127905e-05, "loss": 0.8926, "step": 7733 }, { "epoch": 1.25, "grad_norm": 9.231263160705566, "learning_rate": 1.3183245117301326e-05, "loss": 1.0257, "step": 7734 }, { "epoch": 1.25, "grad_norm": 9.143143653869629, "learning_rate": 1.318159748532123e-05, "loss": 1.0498, "step": 7735 }, { "epoch": 1.25, "grad_norm": 9.97415828704834, "learning_rate": 1.3179949757237388e-05, "loss": 1.1112, "step": 7736 }, { "epoch": 1.25, "grad_norm": 10.211091041564941, "learning_rate": 1.3178301933099574e-05, "loss": 1.212, "step": 7737 }, { "epoch": 1.25, "grad_norm": 8.241179466247559, "learning_rate": 1.3176654012957562e-05, "loss": 0.7953, "step": 7738 }, { "epoch": 1.25, "grad_norm": 9.216915130615234, "learning_rate": 1.3175005996861123e-05, "loss": 1.301, "step": 7739 }, { "epoch": 1.25, "grad_norm": 8.896624565124512, "learning_rate": 1.3173357884860043e-05, "loss": 1.2134, "step": 7740 }, { "epoch": 1.25, "grad_norm": 9.75898551940918, "learning_rate": 1.3171709677004105e-05, "loss": 1.2257, "step": 7741 }, { "epoch": 1.25, "grad_norm": 9.072392463684082, "learning_rate": 1.3170061373343096e-05, "loss": 1.0387, "step": 7742 }, { "epoch": 1.25, "grad_norm": 10.910096168518066, "learning_rate": 1.3168412973926801e-05, "loss": 1.0093, "step": 7743 }, { "epoch": 1.25, "grad_norm": 10.387418746948242, "learning_rate": 1.3166764478805015e-05, "loss": 0.9366, "step": 7744 }, { "epoch": 1.25, "grad_norm": 8.883485794067383, "learning_rate": 1.316511588802753e-05, "loss": 0.9083, "step": 7745 }, { "epoch": 1.25, "grad_norm": 10.388602256774902, "learning_rate": 1.3163467201644147e-05, "loss": 1.3482, "step": 7746 }, { "epoch": 1.25, "grad_norm": 9.189458847045898, "learning_rate": 1.3161818419704662e-05, "loss": 1.083, "step": 7747 }, { "epoch": 1.25, "grad_norm": 11.271482467651367, "learning_rate": 1.316016954225888e-05, "loss": 1.0357, "step": 7748 }, { "epoch": 1.25, "grad_norm": 8.506208419799805, "learning_rate": 1.3158520569356608e-05, "loss": 1.2515, "step": 7749 }, { "epoch": 1.25, "grad_norm": 8.824917793273926, "learning_rate": 1.3156871501047656e-05, "loss": 1.1608, "step": 7750 }, { "epoch": 1.25, "grad_norm": 8.967997550964355, "learning_rate": 1.3155222337381833e-05, "loss": 1.149, "step": 7751 }, { "epoch": 1.25, "grad_norm": 8.787381172180176, "learning_rate": 1.3153573078408957e-05, "loss": 1.0474, "step": 7752 }, { "epoch": 1.25, "grad_norm": 9.679085731506348, "learning_rate": 1.3151923724178843e-05, "loss": 1.1446, "step": 7753 }, { "epoch": 1.25, "grad_norm": 9.390310287475586, "learning_rate": 1.3150274274741317e-05, "loss": 1.0487, "step": 7754 }, { "epoch": 1.25, "grad_norm": 9.38336181640625, "learning_rate": 1.3148624730146197e-05, "loss": 1.0687, "step": 7755 }, { "epoch": 1.25, "grad_norm": 8.95144271850586, "learning_rate": 1.314697509044331e-05, "loss": 1.0658, "step": 7756 }, { "epoch": 1.25, "grad_norm": 9.108800888061523, "learning_rate": 1.3145325355682485e-05, "loss": 0.8558, "step": 7757 }, { "epoch": 1.25, "grad_norm": 7.777554512023926, "learning_rate": 1.3143675525913557e-05, "loss": 0.9996, "step": 7758 }, { "epoch": 1.25, "grad_norm": 7.4341511726379395, "learning_rate": 1.3142025601186356e-05, "loss": 1.0352, "step": 7759 }, { "epoch": 1.25, "grad_norm": 8.319255828857422, "learning_rate": 1.3140375581550726e-05, "loss": 0.9523, "step": 7760 }, { "epoch": 1.25, "grad_norm": 9.580455780029297, "learning_rate": 1.3138725467056505e-05, "loss": 0.8859, "step": 7761 }, { "epoch": 1.25, "grad_norm": 9.649746894836426, "learning_rate": 1.3137075257753535e-05, "loss": 1.0444, "step": 7762 }, { "epoch": 1.25, "grad_norm": 11.128397941589355, "learning_rate": 1.3135424953691665e-05, "loss": 0.9235, "step": 7763 }, { "epoch": 1.25, "grad_norm": 9.663445472717285, "learning_rate": 1.3133774554920742e-05, "loss": 1.4342, "step": 7764 }, { "epoch": 1.25, "grad_norm": 9.276820182800293, "learning_rate": 1.313212406149062e-05, "loss": 1.1843, "step": 7765 }, { "epoch": 1.25, "grad_norm": 9.097221374511719, "learning_rate": 1.3130473473451153e-05, "loss": 0.9624, "step": 7766 }, { "epoch": 1.25, "grad_norm": 9.41855525970459, "learning_rate": 1.3128822790852198e-05, "loss": 1.1898, "step": 7767 }, { "epoch": 1.25, "grad_norm": 9.362732887268066, "learning_rate": 1.3127172013743618e-05, "loss": 0.9959, "step": 7768 }, { "epoch": 1.25, "grad_norm": 8.567946434020996, "learning_rate": 1.3125521142175275e-05, "loss": 1.1899, "step": 7769 }, { "epoch": 1.25, "grad_norm": 7.915991306304932, "learning_rate": 1.3123870176197033e-05, "loss": 0.9677, "step": 7770 }, { "epoch": 1.25, "grad_norm": 8.113676071166992, "learning_rate": 1.3122219115858767e-05, "loss": 0.9831, "step": 7771 }, { "epoch": 1.25, "grad_norm": 9.982929229736328, "learning_rate": 1.3120567961210347e-05, "loss": 1.0297, "step": 7772 }, { "epoch": 1.25, "grad_norm": 9.968209266662598, "learning_rate": 1.3118916712301648e-05, "loss": 0.9767, "step": 7773 }, { "epoch": 1.25, "grad_norm": 9.550176620483398, "learning_rate": 1.3117265369182546e-05, "loss": 1.1076, "step": 7774 }, { "epoch": 1.25, "grad_norm": 8.112566947937012, "learning_rate": 1.3115613931902922e-05, "loss": 0.941, "step": 7775 }, { "epoch": 1.25, "grad_norm": 10.306586265563965, "learning_rate": 1.311396240051266e-05, "loss": 1.1486, "step": 7776 }, { "epoch": 1.25, "grad_norm": 9.246103286743164, "learning_rate": 1.311231077506165e-05, "loss": 0.9793, "step": 7777 }, { "epoch": 1.25, "grad_norm": 8.483206748962402, "learning_rate": 1.3110659055599775e-05, "loss": 1.0028, "step": 7778 }, { "epoch": 1.25, "grad_norm": 11.186309814453125, "learning_rate": 1.3109007242176928e-05, "loss": 1.0879, "step": 7779 }, { "epoch": 1.25, "grad_norm": 10.883111000061035, "learning_rate": 1.3107355334843006e-05, "loss": 1.3579, "step": 7780 }, { "epoch": 1.25, "grad_norm": 9.808588027954102, "learning_rate": 1.310570333364791e-05, "loss": 1.0885, "step": 7781 }, { "epoch": 1.25, "grad_norm": 9.13330364227295, "learning_rate": 1.3104051238641536e-05, "loss": 1.2166, "step": 7782 }, { "epoch": 1.25, "grad_norm": 9.7435941696167, "learning_rate": 1.3102399049873787e-05, "loss": 1.1402, "step": 7783 }, { "epoch": 1.25, "grad_norm": 8.628811836242676, "learning_rate": 1.3100746767394571e-05, "loss": 0.9858, "step": 7784 }, { "epoch": 1.25, "grad_norm": 10.551746368408203, "learning_rate": 1.30990943912538e-05, "loss": 1.0158, "step": 7785 }, { "epoch": 1.25, "grad_norm": 10.781769752502441, "learning_rate": 1.3097441921501382e-05, "loss": 0.9346, "step": 7786 }, { "epoch": 1.25, "grad_norm": 7.757033348083496, "learning_rate": 1.309578935818723e-05, "loss": 1.0281, "step": 7787 }, { "epoch": 1.25, "grad_norm": 10.577881813049316, "learning_rate": 1.3094136701361264e-05, "loss": 0.9105, "step": 7788 }, { "epoch": 1.25, "grad_norm": 9.134134292602539, "learning_rate": 1.3092483951073407e-05, "loss": 1.2341, "step": 7789 }, { "epoch": 1.25, "grad_norm": 9.09199047088623, "learning_rate": 1.3090831107373574e-05, "loss": 1.0705, "step": 7790 }, { "epoch": 1.25, "grad_norm": 8.95346736907959, "learning_rate": 1.30891781703117e-05, "loss": 0.7666, "step": 7791 }, { "epoch": 1.25, "grad_norm": 7.118836402893066, "learning_rate": 1.3087525139937712e-05, "loss": 0.9794, "step": 7792 }, { "epoch": 1.25, "grad_norm": 9.187361717224121, "learning_rate": 1.308587201630154e-05, "loss": 1.3485, "step": 7793 }, { "epoch": 1.25, "grad_norm": 8.942667961120605, "learning_rate": 1.3084218799453116e-05, "loss": 1.3415, "step": 7794 }, { "epoch": 1.25, "grad_norm": 10.30611801147461, "learning_rate": 1.3082565489442381e-05, "loss": 1.396, "step": 7795 }, { "epoch": 1.25, "grad_norm": 7.732736110687256, "learning_rate": 1.3080912086319273e-05, "loss": 0.9519, "step": 7796 }, { "epoch": 1.26, "grad_norm": 8.829622268676758, "learning_rate": 1.3079258590133738e-05, "loss": 1.2849, "step": 7797 }, { "epoch": 1.26, "grad_norm": 8.976762771606445, "learning_rate": 1.3077605000935718e-05, "loss": 1.333, "step": 7798 }, { "epoch": 1.26, "grad_norm": 8.390103340148926, "learning_rate": 1.3075951318775162e-05, "loss": 1.2223, "step": 7799 }, { "epoch": 1.26, "grad_norm": 12.668810844421387, "learning_rate": 1.3074297543702024e-05, "loss": 1.2384, "step": 7800 }, { "epoch": 1.26, "grad_norm": 9.751971244812012, "learning_rate": 1.3072643675766255e-05, "loss": 0.9251, "step": 7801 }, { "epoch": 1.26, "grad_norm": 10.568097114562988, "learning_rate": 1.3070989715017813e-05, "loss": 1.0692, "step": 7802 }, { "epoch": 1.26, "grad_norm": 9.566096305847168, "learning_rate": 1.306933566150666e-05, "loss": 1.0997, "step": 7803 }, { "epoch": 1.26, "grad_norm": 6.5835862159729, "learning_rate": 1.3067681515282756e-05, "loss": 0.8037, "step": 7804 }, { "epoch": 1.26, "grad_norm": 8.01578140258789, "learning_rate": 1.3066027276396067e-05, "loss": 1.22, "step": 7805 }, { "epoch": 1.26, "grad_norm": 11.178568840026855, "learning_rate": 1.3064372944896562e-05, "loss": 1.2038, "step": 7806 }, { "epoch": 1.26, "grad_norm": 8.905705451965332, "learning_rate": 1.306271852083421e-05, "loss": 0.9811, "step": 7807 }, { "epoch": 1.26, "grad_norm": 8.609950065612793, "learning_rate": 1.3061064004258988e-05, "loss": 0.6944, "step": 7808 }, { "epoch": 1.26, "grad_norm": 9.16146183013916, "learning_rate": 1.305940939522087e-05, "loss": 1.1576, "step": 7809 }, { "epoch": 1.26, "grad_norm": 10.283378601074219, "learning_rate": 1.3057754693769833e-05, "loss": 1.3302, "step": 7810 }, { "epoch": 1.26, "grad_norm": 8.061758995056152, "learning_rate": 1.3056099899955863e-05, "loss": 1.0784, "step": 7811 }, { "epoch": 1.26, "grad_norm": 8.011393547058105, "learning_rate": 1.3054445013828948e-05, "loss": 1.1524, "step": 7812 }, { "epoch": 1.26, "grad_norm": 7.155457973480225, "learning_rate": 1.3052790035439066e-05, "loss": 0.9494, "step": 7813 }, { "epoch": 1.26, "grad_norm": 10.854462623596191, "learning_rate": 1.3051134964836213e-05, "loss": 1.3309, "step": 7814 }, { "epoch": 1.26, "grad_norm": 10.468097686767578, "learning_rate": 1.3049479802070385e-05, "loss": 0.9656, "step": 7815 }, { "epoch": 1.26, "grad_norm": 8.186324119567871, "learning_rate": 1.304782454719158e-05, "loss": 0.9156, "step": 7816 }, { "epoch": 1.26, "grad_norm": 7.604979515075684, "learning_rate": 1.3046169200249786e-05, "loss": 1.0972, "step": 7817 }, { "epoch": 1.26, "grad_norm": 8.909279823303223, "learning_rate": 1.3044513761295013e-05, "loss": 1.2301, "step": 7818 }, { "epoch": 1.26, "grad_norm": 9.840529441833496, "learning_rate": 1.304285823037726e-05, "loss": 1.1857, "step": 7819 }, { "epoch": 1.26, "grad_norm": 9.605782508850098, "learning_rate": 1.3041202607546542e-05, "loss": 1.005, "step": 7820 }, { "epoch": 1.26, "grad_norm": 9.264012336730957, "learning_rate": 1.303954689285286e-05, "loss": 0.8333, "step": 7821 }, { "epoch": 1.26, "grad_norm": 10.546844482421875, "learning_rate": 1.3037891086346233e-05, "loss": 1.1741, "step": 7822 }, { "epoch": 1.26, "grad_norm": 8.20643138885498, "learning_rate": 1.3036235188076676e-05, "loss": 1.1537, "step": 7823 }, { "epoch": 1.26, "grad_norm": 9.1904296875, "learning_rate": 1.3034579198094206e-05, "loss": 0.9387, "step": 7824 }, { "epoch": 1.26, "grad_norm": 9.51103687286377, "learning_rate": 1.3032923116448844e-05, "loss": 1.3003, "step": 7825 }, { "epoch": 1.26, "grad_norm": 10.510720252990723, "learning_rate": 1.3031266943190613e-05, "loss": 1.175, "step": 7826 }, { "epoch": 1.26, "grad_norm": 11.021276473999023, "learning_rate": 1.3029610678369538e-05, "loss": 1.5207, "step": 7827 }, { "epoch": 1.26, "grad_norm": 9.730228424072266, "learning_rate": 1.3027954322035656e-05, "loss": 0.8897, "step": 7828 }, { "epoch": 1.26, "grad_norm": 9.61731243133545, "learning_rate": 1.302629787423899e-05, "loss": 1.256, "step": 7829 }, { "epoch": 1.26, "grad_norm": 8.822906494140625, "learning_rate": 1.3024641335029577e-05, "loss": 1.0911, "step": 7830 }, { "epoch": 1.26, "grad_norm": 9.770011901855469, "learning_rate": 1.3022984704457458e-05, "loss": 1.1096, "step": 7831 }, { "epoch": 1.26, "grad_norm": 7.647899627685547, "learning_rate": 1.3021327982572673e-05, "loss": 0.9709, "step": 7832 }, { "epoch": 1.26, "grad_norm": 12.690098762512207, "learning_rate": 1.3019671169425266e-05, "loss": 1.2984, "step": 7833 }, { "epoch": 1.26, "grad_norm": 9.603443145751953, "learning_rate": 1.3018014265065278e-05, "loss": 1.0185, "step": 7834 }, { "epoch": 1.26, "grad_norm": 7.764188766479492, "learning_rate": 1.3016357269542759e-05, "loss": 1.02, "step": 7835 }, { "epoch": 1.26, "grad_norm": 8.850163459777832, "learning_rate": 1.3014700182907767e-05, "loss": 1.1984, "step": 7836 }, { "epoch": 1.26, "grad_norm": 9.269359588623047, "learning_rate": 1.3013043005210347e-05, "loss": 0.918, "step": 7837 }, { "epoch": 1.26, "grad_norm": 9.232842445373535, "learning_rate": 1.301138573650056e-05, "loss": 1.125, "step": 7838 }, { "epoch": 1.26, "grad_norm": 8.876250267028809, "learning_rate": 1.3009728376828465e-05, "loss": 1.0741, "step": 7839 }, { "epoch": 1.26, "grad_norm": 8.821188926696777, "learning_rate": 1.3008070926244129e-05, "loss": 0.9972, "step": 7840 }, { "epoch": 1.26, "grad_norm": 8.716287612915039, "learning_rate": 1.3006413384797609e-05, "loss": 1.1709, "step": 7841 }, { "epoch": 1.26, "grad_norm": 9.486701965332031, "learning_rate": 1.3004755752538978e-05, "loss": 1.0123, "step": 7842 }, { "epoch": 1.26, "grad_norm": 8.12418270111084, "learning_rate": 1.3003098029518306e-05, "loss": 0.904, "step": 7843 }, { "epoch": 1.26, "grad_norm": 9.14276123046875, "learning_rate": 1.300144021578567e-05, "loss": 1.1756, "step": 7844 }, { "epoch": 1.26, "grad_norm": 8.514847755432129, "learning_rate": 1.2999782311391137e-05, "loss": 0.9253, "step": 7845 }, { "epoch": 1.26, "grad_norm": 8.387863159179688, "learning_rate": 1.2998124316384794e-05, "loss": 0.9952, "step": 7846 }, { "epoch": 1.26, "grad_norm": 7.889824867248535, "learning_rate": 1.299646623081672e-05, "loss": 1.2023, "step": 7847 }, { "epoch": 1.26, "grad_norm": 9.552443504333496, "learning_rate": 1.2994808054737e-05, "loss": 1.533, "step": 7848 }, { "epoch": 1.26, "grad_norm": 8.535758018493652, "learning_rate": 1.299314978819572e-05, "loss": 0.9805, "step": 7849 }, { "epoch": 1.26, "grad_norm": 9.977616310119629, "learning_rate": 1.2991491431242969e-05, "loss": 1.5278, "step": 7850 }, { "epoch": 1.26, "grad_norm": 8.976029396057129, "learning_rate": 1.298983298392884e-05, "loss": 1.1834, "step": 7851 }, { "epoch": 1.26, "grad_norm": 12.985530853271484, "learning_rate": 1.298817444630343e-05, "loss": 1.266, "step": 7852 }, { "epoch": 1.26, "grad_norm": 9.961029052734375, "learning_rate": 1.2986515818416833e-05, "loss": 1.1352, "step": 7853 }, { "epoch": 1.26, "grad_norm": 10.071492195129395, "learning_rate": 1.2984857100319154e-05, "loss": 1.1928, "step": 7854 }, { "epoch": 1.26, "grad_norm": 11.2651948928833, "learning_rate": 1.29831982920605e-05, "loss": 1.002, "step": 7855 }, { "epoch": 1.26, "grad_norm": 8.509544372558594, "learning_rate": 1.2981539393690969e-05, "loss": 1.1061, "step": 7856 }, { "epoch": 1.26, "grad_norm": 9.113401412963867, "learning_rate": 1.2979880405260672e-05, "loss": 1.2263, "step": 7857 }, { "epoch": 1.26, "grad_norm": 7.718954563140869, "learning_rate": 1.2978221326819722e-05, "loss": 0.9374, "step": 7858 }, { "epoch": 1.27, "grad_norm": 11.285362243652344, "learning_rate": 1.2976562158418234e-05, "loss": 0.9566, "step": 7859 }, { "epoch": 1.27, "grad_norm": 9.397941589355469, "learning_rate": 1.2974902900106325e-05, "loss": 0.9668, "step": 7860 }, { "epoch": 1.27, "grad_norm": 9.868742942810059, "learning_rate": 1.297324355193411e-05, "loss": 0.9888, "step": 7861 }, { "epoch": 1.27, "grad_norm": 8.689286231994629, "learning_rate": 1.2971584113951718e-05, "loss": 0.9023, "step": 7862 }, { "epoch": 1.27, "grad_norm": 8.139344215393066, "learning_rate": 1.2969924586209275e-05, "loss": 1.2497, "step": 7863 }, { "epoch": 1.27, "grad_norm": 10.002039909362793, "learning_rate": 1.29682649687569e-05, "loss": 1.1246, "step": 7864 }, { "epoch": 1.27, "grad_norm": 7.950613021850586, "learning_rate": 1.2966605261644731e-05, "loss": 0.9841, "step": 7865 }, { "epoch": 1.27, "grad_norm": 9.143797874450684, "learning_rate": 1.2964945464922902e-05, "loss": 1.1516, "step": 7866 }, { "epoch": 1.27, "grad_norm": 8.323271751403809, "learning_rate": 1.2963285578641544e-05, "loss": 0.9013, "step": 7867 }, { "epoch": 1.27, "grad_norm": 9.139511108398438, "learning_rate": 1.29616256028508e-05, "loss": 1.1861, "step": 7868 }, { "epoch": 1.27, "grad_norm": 8.843741416931152, "learning_rate": 1.295996553760081e-05, "loss": 1.0304, "step": 7869 }, { "epoch": 1.27, "grad_norm": 9.315539360046387, "learning_rate": 1.2958305382941716e-05, "loss": 1.0195, "step": 7870 }, { "epoch": 1.27, "grad_norm": 7.8546037673950195, "learning_rate": 1.2956645138923665e-05, "loss": 0.965, "step": 7871 }, { "epoch": 1.27, "grad_norm": 10.63003158569336, "learning_rate": 1.295498480559681e-05, "loss": 1.2369, "step": 7872 }, { "epoch": 1.27, "grad_norm": 9.651518821716309, "learning_rate": 1.2953324383011303e-05, "loss": 0.9241, "step": 7873 }, { "epoch": 1.27, "grad_norm": 10.22028636932373, "learning_rate": 1.2951663871217295e-05, "loss": 1.1114, "step": 7874 }, { "epoch": 1.27, "grad_norm": 8.980947494506836, "learning_rate": 1.295000327026495e-05, "loss": 1.0961, "step": 7875 }, { "epoch": 1.27, "grad_norm": 7.226788520812988, "learning_rate": 1.294834258020442e-05, "loss": 1.0739, "step": 7876 }, { "epoch": 1.27, "grad_norm": 8.828466415405273, "learning_rate": 1.2946681801085874e-05, "loss": 0.8973, "step": 7877 }, { "epoch": 1.27, "grad_norm": 9.050293922424316, "learning_rate": 1.2945020932959478e-05, "loss": 0.9801, "step": 7878 }, { "epoch": 1.27, "grad_norm": 8.768471717834473, "learning_rate": 1.2943359975875398e-05, "loss": 0.8543, "step": 7879 }, { "epoch": 1.27, "grad_norm": 7.912567138671875, "learning_rate": 1.2941698929883804e-05, "loss": 1.2572, "step": 7880 }, { "epoch": 1.27, "grad_norm": 11.720332145690918, "learning_rate": 1.294003779503487e-05, "loss": 1.0821, "step": 7881 }, { "epoch": 1.27, "grad_norm": 9.744093894958496, "learning_rate": 1.2938376571378774e-05, "loss": 1.0306, "step": 7882 }, { "epoch": 1.27, "grad_norm": 7.97709321975708, "learning_rate": 1.2936715258965699e-05, "loss": 0.7947, "step": 7883 }, { "epoch": 1.27, "grad_norm": 8.930002212524414, "learning_rate": 1.293505385784582e-05, "loss": 1.1354, "step": 7884 }, { "epoch": 1.27, "grad_norm": 9.48984432220459, "learning_rate": 1.2933392368069325e-05, "loss": 1.094, "step": 7885 }, { "epoch": 1.27, "grad_norm": 10.359574317932129, "learning_rate": 1.2931730789686402e-05, "loss": 1.1207, "step": 7886 }, { "epoch": 1.27, "grad_norm": 8.145645141601562, "learning_rate": 1.2930069122747238e-05, "loss": 0.8929, "step": 7887 }, { "epoch": 1.27, "grad_norm": 8.454989433288574, "learning_rate": 1.2928407367302025e-05, "loss": 1.0257, "step": 7888 }, { "epoch": 1.27, "grad_norm": 8.688287734985352, "learning_rate": 1.2926745523400961e-05, "loss": 1.2983, "step": 7889 }, { "epoch": 1.27, "grad_norm": 8.161575317382812, "learning_rate": 1.2925083591094242e-05, "loss": 1.0393, "step": 7890 }, { "epoch": 1.27, "grad_norm": 8.72389030456543, "learning_rate": 1.2923421570432071e-05, "loss": 1.0816, "step": 7891 }, { "epoch": 1.27, "grad_norm": 9.436544418334961, "learning_rate": 1.2921759461464649e-05, "loss": 0.8275, "step": 7892 }, { "epoch": 1.27, "grad_norm": 9.429834365844727, "learning_rate": 1.2920097264242184e-05, "loss": 1.0138, "step": 7893 }, { "epoch": 1.27, "grad_norm": 9.038750648498535, "learning_rate": 1.2918434978814882e-05, "loss": 0.96, "step": 7894 }, { "epoch": 1.27, "grad_norm": 10.410148620605469, "learning_rate": 1.2916772605232957e-05, "loss": 1.0827, "step": 7895 }, { "epoch": 1.27, "grad_norm": 9.919502258300781, "learning_rate": 1.291511014354662e-05, "loss": 1.0706, "step": 7896 }, { "epoch": 1.27, "grad_norm": 10.403770446777344, "learning_rate": 1.2913447593806087e-05, "loss": 1.4374, "step": 7897 }, { "epoch": 1.27, "grad_norm": 7.4134721755981445, "learning_rate": 1.291178495606158e-05, "loss": 1.2123, "step": 7898 }, { "epoch": 1.27, "grad_norm": 10.172492027282715, "learning_rate": 1.291012223036332e-05, "loss": 1.1318, "step": 7899 }, { "epoch": 1.27, "grad_norm": 8.953547477722168, "learning_rate": 1.2908459416761531e-05, "loss": 0.9172, "step": 7900 }, { "epoch": 1.27, "grad_norm": 9.55982780456543, "learning_rate": 1.290679651530644e-05, "loss": 1.237, "step": 7901 }, { "epoch": 1.27, "grad_norm": 11.589181900024414, "learning_rate": 1.2905133526048279e-05, "loss": 1.0913, "step": 7902 }, { "epoch": 1.27, "grad_norm": 9.397300720214844, "learning_rate": 1.2903470449037277e-05, "loss": 1.2924, "step": 7903 }, { "epoch": 1.27, "grad_norm": 10.819665908813477, "learning_rate": 1.2901807284323671e-05, "loss": 1.3143, "step": 7904 }, { "epoch": 1.27, "grad_norm": 9.668045997619629, "learning_rate": 1.2900144031957698e-05, "loss": 1.1699, "step": 7905 }, { "epoch": 1.27, "grad_norm": 11.38367748260498, "learning_rate": 1.2898480691989601e-05, "loss": 1.3324, "step": 7906 }, { "epoch": 1.27, "grad_norm": 10.839625358581543, "learning_rate": 1.289681726446962e-05, "loss": 1.3203, "step": 7907 }, { "epoch": 1.27, "grad_norm": 9.330257415771484, "learning_rate": 1.2895153749448002e-05, "loss": 1.0467, "step": 7908 }, { "epoch": 1.27, "grad_norm": 8.020482063293457, "learning_rate": 1.2893490146974994e-05, "loss": 1.1284, "step": 7909 }, { "epoch": 1.27, "grad_norm": 7.7205939292907715, "learning_rate": 1.2891826457100849e-05, "loss": 1.2308, "step": 7910 }, { "epoch": 1.27, "grad_norm": 9.344206809997559, "learning_rate": 1.289016267987582e-05, "loss": 1.2319, "step": 7911 }, { "epoch": 1.27, "grad_norm": 7.636039733886719, "learning_rate": 1.2888498815350161e-05, "loss": 0.9399, "step": 7912 }, { "epoch": 1.27, "grad_norm": 10.754063606262207, "learning_rate": 1.2886834863574133e-05, "loss": 1.4408, "step": 7913 }, { "epoch": 1.27, "grad_norm": 9.805378913879395, "learning_rate": 1.2885170824598004e-05, "loss": 0.9773, "step": 7914 }, { "epoch": 1.27, "grad_norm": 8.298972129821777, "learning_rate": 1.2883506698472026e-05, "loss": 0.9492, "step": 7915 }, { "epoch": 1.27, "grad_norm": 10.172210693359375, "learning_rate": 1.288184248524647e-05, "loss": 0.8404, "step": 7916 }, { "epoch": 1.27, "grad_norm": 9.9281587600708, "learning_rate": 1.288017818497161e-05, "loss": 1.356, "step": 7917 }, { "epoch": 1.27, "grad_norm": 8.722026824951172, "learning_rate": 1.2878513797697716e-05, "loss": 1.1307, "step": 7918 }, { "epoch": 1.27, "grad_norm": 8.248839378356934, "learning_rate": 1.287684932347506e-05, "loss": 1.0563, "step": 7919 }, { "epoch": 1.27, "grad_norm": 9.361556053161621, "learning_rate": 1.2875184762353921e-05, "loss": 1.0123, "step": 7920 }, { "epoch": 1.28, "grad_norm": 8.06340217590332, "learning_rate": 1.2873520114384577e-05, "loss": 0.6956, "step": 7921 }, { "epoch": 1.28, "grad_norm": 8.06775188446045, "learning_rate": 1.2871855379617312e-05, "loss": 1.1016, "step": 7922 }, { "epoch": 1.28, "grad_norm": 8.593332290649414, "learning_rate": 1.2870190558102416e-05, "loss": 1.0422, "step": 7923 }, { "epoch": 1.28, "grad_norm": 8.179484367370605, "learning_rate": 1.286852564989017e-05, "loss": 1.0457, "step": 7924 }, { "epoch": 1.28, "grad_norm": 8.981158256530762, "learning_rate": 1.2866860655030866e-05, "loss": 1.1228, "step": 7925 }, { "epoch": 1.28, "grad_norm": 7.365686416625977, "learning_rate": 1.28651955735748e-05, "loss": 1.172, "step": 7926 }, { "epoch": 1.28, "grad_norm": 8.542713165283203, "learning_rate": 1.2863530405572267e-05, "loss": 1.0034, "step": 7927 }, { "epoch": 1.28, "grad_norm": 8.59614372253418, "learning_rate": 1.286186515107356e-05, "loss": 1.2715, "step": 7928 }, { "epoch": 1.28, "grad_norm": 8.5868501663208, "learning_rate": 1.2860199810128985e-05, "loss": 0.9429, "step": 7929 }, { "epoch": 1.28, "grad_norm": 10.169672012329102, "learning_rate": 1.2858534382788848e-05, "loss": 1.1856, "step": 7930 }, { "epoch": 1.28, "grad_norm": 9.852327346801758, "learning_rate": 1.2856868869103447e-05, "loss": 0.9336, "step": 7931 }, { "epoch": 1.28, "grad_norm": 11.302541732788086, "learning_rate": 1.2855203269123093e-05, "loss": 1.1328, "step": 7932 }, { "epoch": 1.28, "grad_norm": 9.232741355895996, "learning_rate": 1.2853537582898107e-05, "loss": 1.0596, "step": 7933 }, { "epoch": 1.28, "grad_norm": 9.46962833404541, "learning_rate": 1.285187181047879e-05, "loss": 1.143, "step": 7934 }, { "epoch": 1.28, "grad_norm": 9.006767272949219, "learning_rate": 1.2850205951915467e-05, "loss": 1.0548, "step": 7935 }, { "epoch": 1.28, "grad_norm": 8.988259315490723, "learning_rate": 1.2848540007258452e-05, "loss": 0.9208, "step": 7936 }, { "epoch": 1.28, "grad_norm": 9.991284370422363, "learning_rate": 1.284687397655807e-05, "loss": 1.1571, "step": 7937 }, { "epoch": 1.28, "grad_norm": 6.72796106338501, "learning_rate": 1.2845207859864644e-05, "loss": 1.1437, "step": 7938 }, { "epoch": 1.28, "grad_norm": 10.916630744934082, "learning_rate": 1.2843541657228502e-05, "loss": 0.9623, "step": 7939 }, { "epoch": 1.28, "grad_norm": 8.925271034240723, "learning_rate": 1.284187536869997e-05, "loss": 1.1253, "step": 7940 }, { "epoch": 1.28, "grad_norm": 10.036187171936035, "learning_rate": 1.2840208994329385e-05, "loss": 1.1042, "step": 7941 }, { "epoch": 1.28, "grad_norm": 9.729796409606934, "learning_rate": 1.2838542534167078e-05, "loss": 1.1852, "step": 7942 }, { "epoch": 1.28, "grad_norm": 8.287057876586914, "learning_rate": 1.2836875988263388e-05, "loss": 0.8501, "step": 7943 }, { "epoch": 1.28, "grad_norm": 8.759517669677734, "learning_rate": 1.2835209356668654e-05, "loss": 0.905, "step": 7944 }, { "epoch": 1.28, "grad_norm": 10.404149055480957, "learning_rate": 1.2833542639433224e-05, "loss": 1.0568, "step": 7945 }, { "epoch": 1.28, "grad_norm": 9.529040336608887, "learning_rate": 1.2831875836607432e-05, "loss": 0.8937, "step": 7946 }, { "epoch": 1.28, "grad_norm": 7.862434387207031, "learning_rate": 1.2830208948241636e-05, "loss": 1.1142, "step": 7947 }, { "epoch": 1.28, "grad_norm": 9.478126525878906, "learning_rate": 1.282854197438618e-05, "loss": 1.1804, "step": 7948 }, { "epoch": 1.28, "grad_norm": 8.844690322875977, "learning_rate": 1.282687491509142e-05, "loss": 1.083, "step": 7949 }, { "epoch": 1.28, "grad_norm": 7.4618611335754395, "learning_rate": 1.2825207770407712e-05, "loss": 0.9434, "step": 7950 }, { "epoch": 1.28, "grad_norm": 9.423995018005371, "learning_rate": 1.282354054038541e-05, "loss": 1.1537, "step": 7951 }, { "epoch": 1.28, "grad_norm": 8.644889831542969, "learning_rate": 1.2821873225074877e-05, "loss": 1.2931, "step": 7952 }, { "epoch": 1.28, "grad_norm": 9.834925651550293, "learning_rate": 1.2820205824526477e-05, "loss": 1.1793, "step": 7953 }, { "epoch": 1.28, "grad_norm": 7.954033851623535, "learning_rate": 1.2818538338790577e-05, "loss": 1.0497, "step": 7954 }, { "epoch": 1.28, "grad_norm": 9.6549711227417, "learning_rate": 1.281687076791754e-05, "loss": 1.0832, "step": 7955 }, { "epoch": 1.28, "grad_norm": 7.593142986297607, "learning_rate": 1.281520311195774e-05, "loss": 1.0415, "step": 7956 }, { "epoch": 1.28, "grad_norm": 11.009641647338867, "learning_rate": 1.2813535370961555e-05, "loss": 0.9085, "step": 7957 }, { "epoch": 1.28, "grad_norm": 7.989189147949219, "learning_rate": 1.281186754497935e-05, "loss": 1.186, "step": 7958 }, { "epoch": 1.28, "grad_norm": 8.36125659942627, "learning_rate": 1.2810199634061512e-05, "loss": 1.171, "step": 7959 }, { "epoch": 1.28, "grad_norm": 10.472156524658203, "learning_rate": 1.280853163825842e-05, "loss": 0.89, "step": 7960 }, { "epoch": 1.28, "grad_norm": 9.26711368560791, "learning_rate": 1.2806863557620464e-05, "loss": 1.1528, "step": 7961 }, { "epoch": 1.28, "grad_norm": 7.6454362869262695, "learning_rate": 1.2805195392198015e-05, "loss": 0.9483, "step": 7962 }, { "epoch": 1.28, "grad_norm": 8.262041091918945, "learning_rate": 1.2803527142041475e-05, "loss": 0.9687, "step": 7963 }, { "epoch": 1.28, "grad_norm": 8.516702651977539, "learning_rate": 1.280185880720123e-05, "loss": 0.8353, "step": 7964 }, { "epoch": 1.28, "grad_norm": 10.542271614074707, "learning_rate": 1.2800190387727677e-05, "loss": 0.863, "step": 7965 }, { "epoch": 1.28, "grad_norm": 7.777996063232422, "learning_rate": 1.279852188367121e-05, "loss": 0.9421, "step": 7966 }, { "epoch": 1.28, "grad_norm": 10.952712059020996, "learning_rate": 1.2796853295082227e-05, "loss": 1.2492, "step": 7967 }, { "epoch": 1.28, "grad_norm": 11.031183242797852, "learning_rate": 1.2795184622011134e-05, "loss": 0.9582, "step": 7968 }, { "epoch": 1.28, "grad_norm": 9.005718231201172, "learning_rate": 1.2793515864508334e-05, "loss": 0.9295, "step": 7969 }, { "epoch": 1.28, "grad_norm": 10.311837196350098, "learning_rate": 1.2791847022624228e-05, "loss": 1.1662, "step": 7970 }, { "epoch": 1.28, "grad_norm": 8.123924255371094, "learning_rate": 1.2790178096409234e-05, "loss": 1.0469, "step": 7971 }, { "epoch": 1.28, "grad_norm": 8.811382293701172, "learning_rate": 1.2788509085913753e-05, "loss": 1.0768, "step": 7972 }, { "epoch": 1.28, "grad_norm": 8.50505542755127, "learning_rate": 1.2786839991188212e-05, "loss": 1.1435, "step": 7973 }, { "epoch": 1.28, "grad_norm": 11.059039115905762, "learning_rate": 1.2785170812283016e-05, "loss": 0.8621, "step": 7974 }, { "epoch": 1.28, "grad_norm": 8.861427307128906, "learning_rate": 1.2783501549248594e-05, "loss": 0.937, "step": 7975 }, { "epoch": 1.28, "grad_norm": 10.144104957580566, "learning_rate": 1.2781832202135363e-05, "loss": 0.8598, "step": 7976 }, { "epoch": 1.28, "grad_norm": 9.711907386779785, "learning_rate": 1.2780162770993749e-05, "loss": 1.0859, "step": 7977 }, { "epoch": 1.28, "grad_norm": 8.590290069580078, "learning_rate": 1.2778493255874178e-05, "loss": 1.0611, "step": 7978 }, { "epoch": 1.28, "grad_norm": 9.824150085449219, "learning_rate": 1.2776823656827078e-05, "loss": 1.0732, "step": 7979 }, { "epoch": 1.28, "grad_norm": 8.465592384338379, "learning_rate": 1.2775153973902885e-05, "loss": 1.2203, "step": 7980 }, { "epoch": 1.28, "grad_norm": 9.445971488952637, "learning_rate": 1.2773484207152035e-05, "loss": 0.6914, "step": 7981 }, { "epoch": 1.28, "grad_norm": 9.275524139404297, "learning_rate": 1.2771814356624956e-05, "loss": 0.925, "step": 7982 }, { "epoch": 1.29, "grad_norm": 8.718316078186035, "learning_rate": 1.2770144422372095e-05, "loss": 1.1239, "step": 7983 }, { "epoch": 1.29, "grad_norm": 10.29118537902832, "learning_rate": 1.2768474404443896e-05, "loss": 1.135, "step": 7984 }, { "epoch": 1.29, "grad_norm": 9.722614288330078, "learning_rate": 1.27668043028908e-05, "loss": 1.0691, "step": 7985 }, { "epoch": 1.29, "grad_norm": 8.98334789276123, "learning_rate": 1.2765134117763254e-05, "loss": 1.0835, "step": 7986 }, { "epoch": 1.29, "grad_norm": 9.50262451171875, "learning_rate": 1.276346384911171e-05, "loss": 0.9858, "step": 7987 }, { "epoch": 1.29, "grad_norm": 9.742130279541016, "learning_rate": 1.2761793496986619e-05, "loss": 1.183, "step": 7988 }, { "epoch": 1.29, "grad_norm": 11.851419448852539, "learning_rate": 1.2760123061438437e-05, "loss": 1.0902, "step": 7989 }, { "epoch": 1.29, "grad_norm": 9.378663063049316, "learning_rate": 1.2758452542517618e-05, "loss": 1.0898, "step": 7990 }, { "epoch": 1.29, "grad_norm": 10.490520477294922, "learning_rate": 1.2756781940274628e-05, "loss": 1.2356, "step": 7991 }, { "epoch": 1.29, "grad_norm": 8.501007080078125, "learning_rate": 1.2755111254759922e-05, "loss": 0.7896, "step": 7992 }, { "epoch": 1.29, "grad_norm": 9.352506637573242, "learning_rate": 1.2753440486023971e-05, "loss": 1.3469, "step": 7993 }, { "epoch": 1.29, "grad_norm": 8.825334548950195, "learning_rate": 1.2751769634117239e-05, "loss": 1.2698, "step": 7994 }, { "epoch": 1.29, "grad_norm": 8.739362716674805, "learning_rate": 1.27500986990902e-05, "loss": 1.1517, "step": 7995 }, { "epoch": 1.29, "grad_norm": 8.182854652404785, "learning_rate": 1.2748427680993323e-05, "loss": 1.0182, "step": 7996 }, { "epoch": 1.29, "grad_norm": 8.370899200439453, "learning_rate": 1.2746756579877081e-05, "loss": 0.9756, "step": 7997 }, { "epoch": 1.29, "grad_norm": 9.179727554321289, "learning_rate": 1.2745085395791957e-05, "loss": 1.1446, "step": 7998 }, { "epoch": 1.29, "grad_norm": 11.713898658752441, "learning_rate": 1.2743414128788428e-05, "loss": 1.2017, "step": 7999 }, { "epoch": 1.29, "grad_norm": 8.100281715393066, "learning_rate": 1.2741742778916977e-05, "loss": 1.0369, "step": 8000 }, { "epoch": 1.29, "grad_norm": 8.701146125793457, "learning_rate": 1.2740071346228086e-05, "loss": 1.3493, "step": 8001 }, { "epoch": 1.29, "grad_norm": 11.405668258666992, "learning_rate": 1.2738399830772248e-05, "loss": 1.1071, "step": 8002 }, { "epoch": 1.29, "grad_norm": 9.386673927307129, "learning_rate": 1.2736728232599948e-05, "loss": 1.0807, "step": 8003 }, { "epoch": 1.29, "grad_norm": 11.02056884765625, "learning_rate": 1.2735056551761684e-05, "loss": 1.1513, "step": 8004 }, { "epoch": 1.29, "grad_norm": 9.99721908569336, "learning_rate": 1.2733384788307945e-05, "loss": 1.2514, "step": 8005 }, { "epoch": 1.29, "grad_norm": 8.85513973236084, "learning_rate": 1.2731712942289232e-05, "loss": 1.1998, "step": 8006 }, { "epoch": 1.29, "grad_norm": 10.377619743347168, "learning_rate": 1.2730041013756044e-05, "loss": 1.0598, "step": 8007 }, { "epoch": 1.29, "grad_norm": 12.071699142456055, "learning_rate": 1.2728369002758886e-05, "loss": 1.1698, "step": 8008 }, { "epoch": 1.29, "grad_norm": 8.518943786621094, "learning_rate": 1.272669690934826e-05, "loss": 1.0997, "step": 8009 }, { "epoch": 1.29, "grad_norm": 9.615177154541016, "learning_rate": 1.2725024733574673e-05, "loss": 1.066, "step": 8010 }, { "epoch": 1.29, "grad_norm": 8.86892318725586, "learning_rate": 1.2723352475488634e-05, "loss": 0.9118, "step": 8011 }, { "epoch": 1.29, "grad_norm": 8.203946113586426, "learning_rate": 1.2721680135140664e-05, "loss": 1.2552, "step": 8012 }, { "epoch": 1.29, "grad_norm": 10.19176197052002, "learning_rate": 1.2720007712581266e-05, "loss": 1.1666, "step": 8013 }, { "epoch": 1.29, "grad_norm": 9.63184642791748, "learning_rate": 1.2718335207860963e-05, "loss": 0.873, "step": 8014 }, { "epoch": 1.29, "grad_norm": 8.578431129455566, "learning_rate": 1.2716662621030275e-05, "loss": 0.9801, "step": 8015 }, { "epoch": 1.29, "grad_norm": 9.94552993774414, "learning_rate": 1.2714989952139728e-05, "loss": 1.0776, "step": 8016 }, { "epoch": 1.29, "grad_norm": 9.76415729522705, "learning_rate": 1.2713317201239842e-05, "loss": 1.0848, "step": 8017 }, { "epoch": 1.29, "grad_norm": 8.79328441619873, "learning_rate": 1.2711644368381145e-05, "loss": 1.1066, "step": 8018 }, { "epoch": 1.29, "grad_norm": 13.772893905639648, "learning_rate": 1.2709971453614166e-05, "loss": 1.0674, "step": 8019 }, { "epoch": 1.29, "grad_norm": 10.180010795593262, "learning_rate": 1.2708298456989441e-05, "loss": 1.1023, "step": 8020 }, { "epoch": 1.29, "grad_norm": 8.429521560668945, "learning_rate": 1.27066253785575e-05, "loss": 1.1094, "step": 8021 }, { "epoch": 1.29, "grad_norm": 9.753701210021973, "learning_rate": 1.2704952218368882e-05, "loss": 0.9711, "step": 8022 }, { "epoch": 1.29, "grad_norm": 10.188759803771973, "learning_rate": 1.2703278976474127e-05, "loss": 1.153, "step": 8023 }, { "epoch": 1.29, "grad_norm": 9.051807403564453, "learning_rate": 1.2701605652923778e-05, "loss": 1.3114, "step": 8024 }, { "epoch": 1.29, "grad_norm": 9.033620834350586, "learning_rate": 1.269993224776838e-05, "loss": 1.1445, "step": 8025 }, { "epoch": 1.29, "grad_norm": 8.52669620513916, "learning_rate": 1.2698258761058478e-05, "loss": 1.1547, "step": 8026 }, { "epoch": 1.29, "grad_norm": 7.678494453430176, "learning_rate": 1.2696585192844623e-05, "loss": 0.8879, "step": 8027 }, { "epoch": 1.29, "grad_norm": 10.77424144744873, "learning_rate": 1.2694911543177368e-05, "loss": 1.264, "step": 8028 }, { "epoch": 1.29, "grad_norm": 8.239550590515137, "learning_rate": 1.2693237812107263e-05, "loss": 0.9793, "step": 8029 }, { "epoch": 1.29, "grad_norm": 10.828645706176758, "learning_rate": 1.269156399968487e-05, "loss": 0.9731, "step": 8030 }, { "epoch": 1.29, "grad_norm": 9.7666015625, "learning_rate": 1.2689890105960747e-05, "loss": 1.0595, "step": 8031 }, { "epoch": 1.29, "grad_norm": 9.089059829711914, "learning_rate": 1.2688216130985458e-05, "loss": 1.0022, "step": 8032 }, { "epoch": 1.29, "grad_norm": 8.397418022155762, "learning_rate": 1.2686542074809558e-05, "loss": 0.865, "step": 8033 }, { "epoch": 1.29, "grad_norm": 12.178043365478516, "learning_rate": 1.2684867937483624e-05, "loss": 1.1969, "step": 8034 }, { "epoch": 1.29, "grad_norm": 9.440690040588379, "learning_rate": 1.2683193719058222e-05, "loss": 0.8766, "step": 8035 }, { "epoch": 1.29, "grad_norm": 9.388773918151855, "learning_rate": 1.2681519419583922e-05, "loss": 0.9296, "step": 8036 }, { "epoch": 1.29, "grad_norm": 8.99452018737793, "learning_rate": 1.26798450391113e-05, "loss": 1.1079, "step": 8037 }, { "epoch": 1.29, "grad_norm": 8.079367637634277, "learning_rate": 1.2678170577690932e-05, "loss": 1.149, "step": 8038 }, { "epoch": 1.29, "grad_norm": 9.441272735595703, "learning_rate": 1.2676496035373397e-05, "loss": 1.1817, "step": 8039 }, { "epoch": 1.29, "grad_norm": 9.188779830932617, "learning_rate": 1.2674821412209277e-05, "loss": 1.0355, "step": 8040 }, { "epoch": 1.29, "grad_norm": 9.547160148620605, "learning_rate": 1.2673146708249153e-05, "loss": 0.9817, "step": 8041 }, { "epoch": 1.29, "grad_norm": 8.798491477966309, "learning_rate": 1.2671471923543615e-05, "loss": 1.1818, "step": 8042 }, { "epoch": 1.29, "grad_norm": 8.549185752868652, "learning_rate": 1.2669797058143249e-05, "loss": 1.0803, "step": 8043 }, { "epoch": 1.29, "grad_norm": 9.943790435791016, "learning_rate": 1.266812211209865e-05, "loss": 1.0426, "step": 8044 }, { "epoch": 1.3, "grad_norm": 9.310147285461426, "learning_rate": 1.2666447085460406e-05, "loss": 1.0672, "step": 8045 }, { "epoch": 1.3, "grad_norm": 11.235713958740234, "learning_rate": 1.2664771978279119e-05, "loss": 1.3569, "step": 8046 }, { "epoch": 1.3, "grad_norm": 8.70229434967041, "learning_rate": 1.2663096790605384e-05, "loss": 1.1229, "step": 8047 }, { "epoch": 1.3, "grad_norm": 11.059810638427734, "learning_rate": 1.2661421522489803e-05, "loss": 1.1587, "step": 8048 }, { "epoch": 1.3, "grad_norm": 10.944975852966309, "learning_rate": 1.2659746173982979e-05, "loss": 1.3341, "step": 8049 }, { "epoch": 1.3, "grad_norm": 7.862634658813477, "learning_rate": 1.2658070745135515e-05, "loss": 1.1462, "step": 8050 }, { "epoch": 1.3, "grad_norm": 9.803573608398438, "learning_rate": 1.2656395235998027e-05, "loss": 0.9536, "step": 8051 }, { "epoch": 1.3, "grad_norm": 10.446722984313965, "learning_rate": 1.2654719646621116e-05, "loss": 1.3006, "step": 8052 }, { "epoch": 1.3, "grad_norm": 9.833248138427734, "learning_rate": 1.26530439770554e-05, "loss": 0.9264, "step": 8053 }, { "epoch": 1.3, "grad_norm": 9.803627014160156, "learning_rate": 1.2651368227351495e-05, "loss": 0.9887, "step": 8054 }, { "epoch": 1.3, "grad_norm": 8.273345947265625, "learning_rate": 1.2649692397560021e-05, "loss": 1.1203, "step": 8055 }, { "epoch": 1.3, "grad_norm": 9.367097854614258, "learning_rate": 1.2648016487731593e-05, "loss": 1.006, "step": 8056 }, { "epoch": 1.3, "grad_norm": 8.296889305114746, "learning_rate": 1.2646340497916838e-05, "loss": 0.7625, "step": 8057 }, { "epoch": 1.3, "grad_norm": 8.881486892700195, "learning_rate": 1.2644664428166377e-05, "loss": 0.9371, "step": 8058 }, { "epoch": 1.3, "grad_norm": 8.47104263305664, "learning_rate": 1.2642988278530843e-05, "loss": 0.9082, "step": 8059 }, { "epoch": 1.3, "grad_norm": 11.453200340270996, "learning_rate": 1.2641312049060861e-05, "loss": 1.0964, "step": 8060 }, { "epoch": 1.3, "grad_norm": 8.006354331970215, "learning_rate": 1.2639635739807066e-05, "loss": 0.8558, "step": 8061 }, { "epoch": 1.3, "grad_norm": 12.688447952270508, "learning_rate": 1.263795935082009e-05, "loss": 1.1308, "step": 8062 }, { "epoch": 1.3, "grad_norm": 12.568965911865234, "learning_rate": 1.2636282882150574e-05, "loss": 1.389, "step": 8063 }, { "epoch": 1.3, "grad_norm": 12.188516616821289, "learning_rate": 1.2634606333849155e-05, "loss": 1.2249, "step": 8064 }, { "epoch": 1.3, "grad_norm": 10.68343734741211, "learning_rate": 1.2632929705966478e-05, "loss": 1.0402, "step": 8065 }, { "epoch": 1.3, "grad_norm": 11.074851989746094, "learning_rate": 1.2631252998553183e-05, "loss": 1.0753, "step": 8066 }, { "epoch": 1.3, "grad_norm": 8.43974494934082, "learning_rate": 1.2629576211659925e-05, "loss": 1.0471, "step": 8067 }, { "epoch": 1.3, "grad_norm": 7.802322864532471, "learning_rate": 1.2627899345337342e-05, "loss": 1.0688, "step": 8068 }, { "epoch": 1.3, "grad_norm": 9.143792152404785, "learning_rate": 1.2626222399636092e-05, "loss": 1.1254, "step": 8069 }, { "epoch": 1.3, "grad_norm": 9.133472442626953, "learning_rate": 1.2624545374606828e-05, "loss": 0.9127, "step": 8070 }, { "epoch": 1.3, "grad_norm": 9.327655792236328, "learning_rate": 1.262286827030021e-05, "loss": 1.0732, "step": 8071 }, { "epoch": 1.3, "grad_norm": 9.07388687133789, "learning_rate": 1.2621191086766893e-05, "loss": 1.317, "step": 8072 }, { "epoch": 1.3, "grad_norm": 9.954452514648438, "learning_rate": 1.2619513824057537e-05, "loss": 1.1817, "step": 8073 }, { "epoch": 1.3, "grad_norm": 8.903608322143555, "learning_rate": 1.2617836482222807e-05, "loss": 1.2887, "step": 8074 }, { "epoch": 1.3, "grad_norm": 9.025039672851562, "learning_rate": 1.2616159061313368e-05, "loss": 1.0252, "step": 8075 }, { "epoch": 1.3, "grad_norm": 9.405394554138184, "learning_rate": 1.2614481561379892e-05, "loss": 1.0146, "step": 8076 }, { "epoch": 1.3, "grad_norm": 10.017191886901855, "learning_rate": 1.2612803982473047e-05, "loss": 1.224, "step": 8077 }, { "epoch": 1.3, "grad_norm": 8.201797485351562, "learning_rate": 1.261112632464351e-05, "loss": 1.1154, "step": 8078 }, { "epoch": 1.3, "grad_norm": 7.451838970184326, "learning_rate": 1.260944858794195e-05, "loss": 0.7547, "step": 8079 }, { "epoch": 1.3, "grad_norm": 9.021035194396973, "learning_rate": 1.2607770772419048e-05, "loss": 1.1494, "step": 8080 }, { "epoch": 1.3, "grad_norm": 8.762139320373535, "learning_rate": 1.2606092878125485e-05, "loss": 1.2565, "step": 8081 }, { "epoch": 1.3, "grad_norm": 8.074348449707031, "learning_rate": 1.2604414905111944e-05, "loss": 0.9391, "step": 8082 }, { "epoch": 1.3, "grad_norm": 8.36871337890625, "learning_rate": 1.2602736853429109e-05, "loss": 1.2798, "step": 8083 }, { "epoch": 1.3, "grad_norm": 9.270718574523926, "learning_rate": 1.2601058723127665e-05, "loss": 1.0603, "step": 8084 }, { "epoch": 1.3, "grad_norm": 9.2818021774292, "learning_rate": 1.2599380514258309e-05, "loss": 1.1645, "step": 8085 }, { "epoch": 1.3, "grad_norm": 9.965156555175781, "learning_rate": 1.2597702226871726e-05, "loss": 1.0302, "step": 8086 }, { "epoch": 1.3, "grad_norm": 7.6104350090026855, "learning_rate": 1.2596023861018614e-05, "loss": 1.0164, "step": 8087 }, { "epoch": 1.3, "grad_norm": 10.655776977539062, "learning_rate": 1.2594345416749668e-05, "loss": 1.0421, "step": 8088 }, { "epoch": 1.3, "grad_norm": 8.590712547302246, "learning_rate": 1.2592666894115591e-05, "loss": 0.9727, "step": 8089 }, { "epoch": 1.3, "grad_norm": 10.146337509155273, "learning_rate": 1.259098829316708e-05, "loss": 1.0974, "step": 8090 }, { "epoch": 1.3, "grad_norm": 10.701457023620605, "learning_rate": 1.2589309613954843e-05, "loss": 1.3647, "step": 8091 }, { "epoch": 1.3, "grad_norm": 8.93839168548584, "learning_rate": 1.2587630856529584e-05, "loss": 1.0306, "step": 8092 }, { "epoch": 1.3, "grad_norm": 9.161248207092285, "learning_rate": 1.2585952020942012e-05, "loss": 1.2864, "step": 8093 }, { "epoch": 1.3, "grad_norm": 8.038894653320312, "learning_rate": 1.258427310724284e-05, "loss": 0.8954, "step": 8094 }, { "epoch": 1.3, "grad_norm": 8.960345268249512, "learning_rate": 1.2582594115482779e-05, "loss": 1.0378, "step": 8095 }, { "epoch": 1.3, "grad_norm": 9.659955024719238, "learning_rate": 1.2580915045712545e-05, "loss": 1.156, "step": 8096 }, { "epoch": 1.3, "grad_norm": 9.099059104919434, "learning_rate": 1.2579235897982857e-05, "loss": 0.9737, "step": 8097 }, { "epoch": 1.3, "grad_norm": 8.37035083770752, "learning_rate": 1.2577556672344438e-05, "loss": 1.0851, "step": 8098 }, { "epoch": 1.3, "grad_norm": 9.539139747619629, "learning_rate": 1.2575877368848006e-05, "loss": 1.0661, "step": 8099 }, { "epoch": 1.3, "grad_norm": 9.198266983032227, "learning_rate": 1.257419798754429e-05, "loss": 0.9828, "step": 8100 }, { "epoch": 1.3, "grad_norm": 8.953300476074219, "learning_rate": 1.2572518528484016e-05, "loss": 1.2431, "step": 8101 }, { "epoch": 1.3, "grad_norm": 9.199150085449219, "learning_rate": 1.2570838991717917e-05, "loss": 1.1122, "step": 8102 }, { "epoch": 1.3, "grad_norm": 9.122822761535645, "learning_rate": 1.2569159377296721e-05, "loss": 1.2201, "step": 8103 }, { "epoch": 1.3, "grad_norm": 9.548478126525879, "learning_rate": 1.256747968527116e-05, "loss": 0.9692, "step": 8104 }, { "epoch": 1.3, "grad_norm": 8.181448936462402, "learning_rate": 1.256579991569198e-05, "loss": 1.0745, "step": 8105 }, { "epoch": 1.3, "grad_norm": 9.317072868347168, "learning_rate": 1.2564120068609915e-05, "loss": 1.0154, "step": 8106 }, { "epoch": 1.31, "grad_norm": 8.338805198669434, "learning_rate": 1.2562440144075708e-05, "loss": 1.1661, "step": 8107 }, { "epoch": 1.31, "grad_norm": 9.990743637084961, "learning_rate": 1.2560760142140098e-05, "loss": 1.3673, "step": 8108 }, { "epoch": 1.31, "grad_norm": 9.258550643920898, "learning_rate": 1.2559080062853841e-05, "loss": 1.2207, "step": 8109 }, { "epoch": 1.31, "grad_norm": 7.945727825164795, "learning_rate": 1.255739990626768e-05, "loss": 1.0355, "step": 8110 }, { "epoch": 1.31, "grad_norm": 9.888557434082031, "learning_rate": 1.2555719672432365e-05, "loss": 1.1295, "step": 8111 }, { "epoch": 1.31, "grad_norm": 9.814072608947754, "learning_rate": 1.2554039361398649e-05, "loss": 1.109, "step": 8112 }, { "epoch": 1.31, "grad_norm": 8.015119552612305, "learning_rate": 1.255235897321729e-05, "loss": 0.9077, "step": 8113 }, { "epoch": 1.31, "grad_norm": 10.388680458068848, "learning_rate": 1.2550678507939046e-05, "loss": 1.1494, "step": 8114 }, { "epoch": 1.31, "grad_norm": 9.396378517150879, "learning_rate": 1.2548997965614678e-05, "loss": 1.0801, "step": 8115 }, { "epoch": 1.31, "grad_norm": 10.047404289245605, "learning_rate": 1.2547317346294947e-05, "loss": 1.0633, "step": 8116 }, { "epoch": 1.31, "grad_norm": 8.608885765075684, "learning_rate": 1.2545636650030621e-05, "loss": 1.0279, "step": 8117 }, { "epoch": 1.31, "grad_norm": 8.662505149841309, "learning_rate": 1.2543955876872462e-05, "loss": 1.0189, "step": 8118 }, { "epoch": 1.31, "grad_norm": 9.64085865020752, "learning_rate": 1.2542275026871243e-05, "loss": 0.9323, "step": 8119 }, { "epoch": 1.31, "grad_norm": 7.758790969848633, "learning_rate": 1.2540594100077734e-05, "loss": 1.1444, "step": 8120 }, { "epoch": 1.31, "grad_norm": 10.133110046386719, "learning_rate": 1.2538913096542711e-05, "loss": 1.065, "step": 8121 }, { "epoch": 1.31, "grad_norm": 8.84595775604248, "learning_rate": 1.2537232016316953e-05, "loss": 1.1538, "step": 8122 }, { "epoch": 1.31, "grad_norm": 8.46859073638916, "learning_rate": 1.2535550859451235e-05, "loss": 0.9017, "step": 8123 }, { "epoch": 1.31, "grad_norm": 9.080431938171387, "learning_rate": 1.2533869625996338e-05, "loss": 1.0885, "step": 8124 }, { "epoch": 1.31, "grad_norm": 9.03188419342041, "learning_rate": 1.2532188316003051e-05, "loss": 0.9085, "step": 8125 }, { "epoch": 1.31, "grad_norm": 7.513949871063232, "learning_rate": 1.2530506929522153e-05, "loss": 0.8933, "step": 8126 }, { "epoch": 1.31, "grad_norm": 10.749948501586914, "learning_rate": 1.2528825466604434e-05, "loss": 1.0378, "step": 8127 }, { "epoch": 1.31, "grad_norm": 9.508856773376465, "learning_rate": 1.2527143927300686e-05, "loss": 0.8894, "step": 8128 }, { "epoch": 1.31, "grad_norm": 9.650359153747559, "learning_rate": 1.2525462311661707e-05, "loss": 1.0701, "step": 8129 }, { "epoch": 1.31, "grad_norm": 10.187325477600098, "learning_rate": 1.2523780619738282e-05, "loss": 0.9405, "step": 8130 }, { "epoch": 1.31, "grad_norm": 9.166903495788574, "learning_rate": 1.2522098851581212e-05, "loss": 0.959, "step": 8131 }, { "epoch": 1.31, "grad_norm": 9.805614471435547, "learning_rate": 1.25204170072413e-05, "loss": 1.1252, "step": 8132 }, { "epoch": 1.31, "grad_norm": 9.620482444763184, "learning_rate": 1.2518735086769342e-05, "loss": 1.0699, "step": 8133 }, { "epoch": 1.31, "grad_norm": 8.861191749572754, "learning_rate": 1.251705309021615e-05, "loss": 1.1277, "step": 8134 }, { "epoch": 1.31, "grad_norm": 8.453865051269531, "learning_rate": 1.2515371017632525e-05, "loss": 1.1201, "step": 8135 }, { "epoch": 1.31, "grad_norm": 8.523538589477539, "learning_rate": 1.2513688869069276e-05, "loss": 0.8636, "step": 8136 }, { "epoch": 1.31, "grad_norm": 9.769068717956543, "learning_rate": 1.2512006644577219e-05, "loss": 1.2994, "step": 8137 }, { "epoch": 1.31, "grad_norm": 8.339034080505371, "learning_rate": 1.2510324344207163e-05, "loss": 0.9859, "step": 8138 }, { "epoch": 1.31, "grad_norm": 8.95606517791748, "learning_rate": 1.2508641968009924e-05, "loss": 1.0455, "step": 8139 }, { "epoch": 1.31, "grad_norm": 9.192084312438965, "learning_rate": 1.250695951603632e-05, "loss": 1.0305, "step": 8140 }, { "epoch": 1.31, "grad_norm": 10.085875511169434, "learning_rate": 1.2505276988337175e-05, "loss": 1.2042, "step": 8141 }, { "epoch": 1.31, "grad_norm": 8.225942611694336, "learning_rate": 1.2503594384963309e-05, "loss": 1.0508, "step": 8142 }, { "epoch": 1.31, "grad_norm": 8.31572151184082, "learning_rate": 1.2501911705965546e-05, "loss": 0.9901, "step": 8143 }, { "epoch": 1.31, "grad_norm": 9.622300148010254, "learning_rate": 1.2500228951394712e-05, "loss": 1.0253, "step": 8144 }, { "epoch": 1.31, "grad_norm": 10.149693489074707, "learning_rate": 1.249854612130164e-05, "loss": 0.9569, "step": 8145 }, { "epoch": 1.31, "grad_norm": 8.043124198913574, "learning_rate": 1.2496863215737163e-05, "loss": 1.242, "step": 8146 }, { "epoch": 1.31, "grad_norm": 9.984318733215332, "learning_rate": 1.2495180234752109e-05, "loss": 1.2694, "step": 8147 }, { "epoch": 1.31, "grad_norm": 8.575921058654785, "learning_rate": 1.249349717839732e-05, "loss": 1.1913, "step": 8148 }, { "epoch": 1.31, "grad_norm": 9.864653587341309, "learning_rate": 1.2491814046723635e-05, "loss": 1.2154, "step": 8149 }, { "epoch": 1.31, "grad_norm": 9.295510292053223, "learning_rate": 1.2490130839781886e-05, "loss": 1.0067, "step": 8150 }, { "epoch": 1.31, "grad_norm": 9.027060508728027, "learning_rate": 1.2488447557622927e-05, "loss": 0.9932, "step": 8151 }, { "epoch": 1.31, "grad_norm": 7.713151454925537, "learning_rate": 1.2486764200297598e-05, "loss": 0.9305, "step": 8152 }, { "epoch": 1.31, "grad_norm": 8.930241584777832, "learning_rate": 1.248508076785675e-05, "loss": 0.8921, "step": 8153 }, { "epoch": 1.31, "grad_norm": 9.640175819396973, "learning_rate": 1.2483397260351226e-05, "loss": 1.1537, "step": 8154 }, { "epoch": 1.31, "grad_norm": 10.377091407775879, "learning_rate": 1.2481713677831884e-05, "loss": 1.074, "step": 8155 }, { "epoch": 1.31, "grad_norm": 8.54776668548584, "learning_rate": 1.248003002034958e-05, "loss": 1.1877, "step": 8156 }, { "epoch": 1.31, "grad_norm": 9.851296424865723, "learning_rate": 1.2478346287955167e-05, "loss": 0.9149, "step": 8157 }, { "epoch": 1.31, "grad_norm": 9.22622013092041, "learning_rate": 1.2476662480699507e-05, "loss": 1.0684, "step": 8158 }, { "epoch": 1.31, "grad_norm": 8.441078186035156, "learning_rate": 1.2474978598633456e-05, "loss": 1.2492, "step": 8159 }, { "epoch": 1.31, "grad_norm": 9.885077476501465, "learning_rate": 1.2473294641807882e-05, "loss": 0.998, "step": 8160 }, { "epoch": 1.31, "grad_norm": 9.542981147766113, "learning_rate": 1.2471610610273652e-05, "loss": 1.0236, "step": 8161 }, { "epoch": 1.31, "grad_norm": 10.240263938903809, "learning_rate": 1.2469926504081631e-05, "loss": 1.061, "step": 8162 }, { "epoch": 1.31, "grad_norm": 8.55312442779541, "learning_rate": 1.2468242323282691e-05, "loss": 0.8908, "step": 8163 }, { "epoch": 1.31, "grad_norm": 8.839367866516113, "learning_rate": 1.24665580679277e-05, "loss": 1.0273, "step": 8164 }, { "epoch": 1.31, "grad_norm": 9.588611602783203, "learning_rate": 1.2464873738067542e-05, "loss": 1.1668, "step": 8165 }, { "epoch": 1.31, "grad_norm": 9.02283000946045, "learning_rate": 1.2463189333753085e-05, "loss": 1.2864, "step": 8166 }, { "epoch": 1.31, "grad_norm": 7.971770763397217, "learning_rate": 1.2461504855035213e-05, "loss": 1.0577, "step": 8167 }, { "epoch": 1.31, "grad_norm": 10.725852966308594, "learning_rate": 1.2459820301964812e-05, "loss": 1.1522, "step": 8168 }, { "epoch": 1.32, "grad_norm": 11.62744426727295, "learning_rate": 1.2458135674592754e-05, "loss": 1.1061, "step": 8169 }, { "epoch": 1.32, "grad_norm": 9.725098609924316, "learning_rate": 1.2456450972969933e-05, "loss": 0.9744, "step": 8170 }, { "epoch": 1.32, "grad_norm": 9.714059829711914, "learning_rate": 1.2454766197147237e-05, "loss": 1.2811, "step": 8171 }, { "epoch": 1.32, "grad_norm": 9.34706974029541, "learning_rate": 1.2453081347175555e-05, "loss": 1.0119, "step": 8172 }, { "epoch": 1.32, "grad_norm": 10.226219177246094, "learning_rate": 1.2451396423105781e-05, "loss": 1.2125, "step": 8173 }, { "epoch": 1.32, "grad_norm": 10.679482460021973, "learning_rate": 1.244971142498881e-05, "loss": 0.929, "step": 8174 }, { "epoch": 1.32, "grad_norm": 9.568779945373535, "learning_rate": 1.2448026352875535e-05, "loss": 1.2934, "step": 8175 }, { "epoch": 1.32, "grad_norm": 11.984358787536621, "learning_rate": 1.2446341206816864e-05, "loss": 1.2585, "step": 8176 }, { "epoch": 1.32, "grad_norm": 9.470468521118164, "learning_rate": 1.2444655986863692e-05, "loss": 1.1788, "step": 8177 }, { "epoch": 1.32, "grad_norm": 10.254358291625977, "learning_rate": 1.2442970693066922e-05, "loss": 1.0149, "step": 8178 }, { "epoch": 1.32, "grad_norm": 9.549386024475098, "learning_rate": 1.2441285325477466e-05, "loss": 0.9516, "step": 8179 }, { "epoch": 1.32, "grad_norm": 11.38807201385498, "learning_rate": 1.2439599884146232e-05, "loss": 1.1011, "step": 8180 }, { "epoch": 1.32, "grad_norm": 7.7853217124938965, "learning_rate": 1.2437914369124124e-05, "loss": 0.9668, "step": 8181 }, { "epoch": 1.32, "grad_norm": 10.281902313232422, "learning_rate": 1.2436228780462061e-05, "loss": 1.2655, "step": 8182 }, { "epoch": 1.32, "grad_norm": 10.683178901672363, "learning_rate": 1.2434543118210953e-05, "loss": 0.9449, "step": 8183 }, { "epoch": 1.32, "grad_norm": 9.797994613647461, "learning_rate": 1.2432857382421726e-05, "loss": 1.0239, "step": 8184 }, { "epoch": 1.32, "grad_norm": 7.475193023681641, "learning_rate": 1.2431171573145289e-05, "loss": 0.8387, "step": 8185 }, { "epoch": 1.32, "grad_norm": 9.119708061218262, "learning_rate": 1.2429485690432568e-05, "loss": 1.1336, "step": 8186 }, { "epoch": 1.32, "grad_norm": 9.08276081085205, "learning_rate": 1.242779973433449e-05, "loss": 1.0881, "step": 8187 }, { "epoch": 1.32, "grad_norm": 10.170533180236816, "learning_rate": 1.2426113704901983e-05, "loss": 0.912, "step": 8188 }, { "epoch": 1.32, "grad_norm": 8.141352653503418, "learning_rate": 1.2424427602185966e-05, "loss": 0.9535, "step": 8189 }, { "epoch": 1.32, "grad_norm": 9.330641746520996, "learning_rate": 1.2422741426237377e-05, "loss": 1.3284, "step": 8190 }, { "epoch": 1.32, "grad_norm": 7.808426856994629, "learning_rate": 1.2421055177107146e-05, "loss": 0.8573, "step": 8191 }, { "epoch": 1.32, "grad_norm": 9.178750991821289, "learning_rate": 1.2419368854846212e-05, "loss": 0.9697, "step": 8192 }, { "epoch": 1.32, "grad_norm": 8.510722160339355, "learning_rate": 1.2417682459505506e-05, "loss": 0.9564, "step": 8193 }, { "epoch": 1.32, "grad_norm": 13.654335975646973, "learning_rate": 1.2415995991135971e-05, "loss": 1.1505, "step": 8194 }, { "epoch": 1.32, "grad_norm": 7.99569845199585, "learning_rate": 1.241430944978855e-05, "loss": 0.9989, "step": 8195 }, { "epoch": 1.32, "grad_norm": 10.066258430480957, "learning_rate": 1.2412622835514182e-05, "loss": 1.1296, "step": 8196 }, { "epoch": 1.32, "grad_norm": 11.561749458312988, "learning_rate": 1.241093614836382e-05, "loss": 1.4296, "step": 8197 }, { "epoch": 1.32, "grad_norm": 10.390600204467773, "learning_rate": 1.2409249388388407e-05, "loss": 1.1298, "step": 8198 }, { "epoch": 1.32, "grad_norm": 9.801735877990723, "learning_rate": 1.2407562555638894e-05, "loss": 1.182, "step": 8199 }, { "epoch": 1.32, "grad_norm": 9.338227272033691, "learning_rate": 1.240587565016624e-05, "loss": 1.0417, "step": 8200 }, { "epoch": 1.32, "grad_norm": 9.365557670593262, "learning_rate": 1.240418867202139e-05, "loss": 1.1122, "step": 8201 }, { "epoch": 1.32, "grad_norm": 10.546323776245117, "learning_rate": 1.2402501621255305e-05, "loss": 1.1646, "step": 8202 }, { "epoch": 1.32, "grad_norm": 10.572735786437988, "learning_rate": 1.2400814497918946e-05, "loss": 1.1144, "step": 8203 }, { "epoch": 1.32, "grad_norm": 10.596478462219238, "learning_rate": 1.2399127302063279e-05, "loss": 0.9703, "step": 8204 }, { "epoch": 1.32, "grad_norm": 9.593536376953125, "learning_rate": 1.2397440033739254e-05, "loss": 0.9574, "step": 8205 }, { "epoch": 1.32, "grad_norm": 9.589612007141113, "learning_rate": 1.2395752692997849e-05, "loss": 1.308, "step": 8206 }, { "epoch": 1.32, "grad_norm": 10.333534240722656, "learning_rate": 1.2394065279890028e-05, "loss": 0.9761, "step": 8207 }, { "epoch": 1.32, "grad_norm": 7.905910491943359, "learning_rate": 1.239237779446676e-05, "loss": 0.8925, "step": 8208 }, { "epoch": 1.32, "grad_norm": 7.711134433746338, "learning_rate": 1.2390690236779022e-05, "loss": 1.1001, "step": 8209 }, { "epoch": 1.32, "grad_norm": 10.107604026794434, "learning_rate": 1.2389002606877782e-05, "loss": 1.0401, "step": 8210 }, { "epoch": 1.32, "grad_norm": 8.777593612670898, "learning_rate": 1.238731490481402e-05, "loss": 0.9784, "step": 8211 }, { "epoch": 1.32, "grad_norm": 8.61187744140625, "learning_rate": 1.2385627130638717e-05, "loss": 0.9084, "step": 8212 }, { "epoch": 1.32, "grad_norm": 8.897802352905273, "learning_rate": 1.238393928440285e-05, "loss": 0.6776, "step": 8213 }, { "epoch": 1.32, "grad_norm": 9.2466459274292, "learning_rate": 1.2382251366157408e-05, "loss": 1.0901, "step": 8214 }, { "epoch": 1.32, "grad_norm": 8.944961547851562, "learning_rate": 1.238056337595337e-05, "loss": 0.931, "step": 8215 }, { "epoch": 1.32, "grad_norm": 11.172381401062012, "learning_rate": 1.2378875313841725e-05, "loss": 1.2686, "step": 8216 }, { "epoch": 1.32, "grad_norm": 11.125609397888184, "learning_rate": 1.2377187179873467e-05, "loss": 1.0118, "step": 8217 }, { "epoch": 1.32, "grad_norm": 8.80868911743164, "learning_rate": 1.2375498974099585e-05, "loss": 1.1977, "step": 8218 }, { "epoch": 1.32, "grad_norm": 10.478521347045898, "learning_rate": 1.2373810696571074e-05, "loss": 1.0503, "step": 8219 }, { "epoch": 1.32, "grad_norm": 8.613948822021484, "learning_rate": 1.2372122347338928e-05, "loss": 1.2959, "step": 8220 }, { "epoch": 1.32, "grad_norm": 10.842209815979004, "learning_rate": 1.237043392645415e-05, "loss": 1.0397, "step": 8221 }, { "epoch": 1.32, "grad_norm": 9.798501968383789, "learning_rate": 1.2368745433967734e-05, "loss": 1.1787, "step": 8222 }, { "epoch": 1.32, "grad_norm": 10.190831184387207, "learning_rate": 1.2367056869930691e-05, "loss": 1.1353, "step": 8223 }, { "epoch": 1.32, "grad_norm": 9.615723609924316, "learning_rate": 1.2365368234394021e-05, "loss": 1.0686, "step": 8224 }, { "epoch": 1.32, "grad_norm": 9.07605266571045, "learning_rate": 1.236367952740873e-05, "loss": 0.9922, "step": 8225 }, { "epoch": 1.32, "grad_norm": 8.172075271606445, "learning_rate": 1.236199074902583e-05, "loss": 0.8694, "step": 8226 }, { "epoch": 1.32, "grad_norm": 8.794626235961914, "learning_rate": 1.2360301899296332e-05, "loss": 1.0026, "step": 8227 }, { "epoch": 1.32, "grad_norm": 7.9502668380737305, "learning_rate": 1.235861297827125e-05, "loss": 0.9663, "step": 8228 }, { "epoch": 1.32, "grad_norm": 8.654829025268555, "learning_rate": 1.2356923986001599e-05, "loss": 0.8705, "step": 8229 }, { "epoch": 1.32, "grad_norm": 8.088213920593262, "learning_rate": 1.2355234922538395e-05, "loss": 0.9026, "step": 8230 }, { "epoch": 1.33, "grad_norm": 10.726116180419922, "learning_rate": 1.2353545787932664e-05, "loss": 0.8591, "step": 8231 }, { "epoch": 1.33, "grad_norm": 11.140254020690918, "learning_rate": 1.2351856582235419e-05, "loss": 0.949, "step": 8232 }, { "epoch": 1.33, "grad_norm": 11.580648422241211, "learning_rate": 1.2350167305497693e-05, "loss": 1.3812, "step": 8233 }, { "epoch": 1.33, "grad_norm": 8.701211929321289, "learning_rate": 1.2348477957770509e-05, "loss": 0.9672, "step": 8234 }, { "epoch": 1.33, "grad_norm": 9.579631805419922, "learning_rate": 1.2346788539104898e-05, "loss": 1.1231, "step": 8235 }, { "epoch": 1.33, "grad_norm": 10.467826843261719, "learning_rate": 1.2345099049551883e-05, "loss": 0.9292, "step": 8236 }, { "epoch": 1.33, "grad_norm": 11.613069534301758, "learning_rate": 1.2343409489162505e-05, "loss": 1.0622, "step": 8237 }, { "epoch": 1.33, "grad_norm": 10.440292358398438, "learning_rate": 1.23417198579878e-05, "loss": 1.3915, "step": 8238 }, { "epoch": 1.33, "grad_norm": 7.871937274932861, "learning_rate": 1.23400301560788e-05, "loss": 0.9865, "step": 8239 }, { "epoch": 1.33, "grad_norm": 8.115575790405273, "learning_rate": 1.2338340383486545e-05, "loss": 1.1058, "step": 8240 }, { "epoch": 1.33, "grad_norm": 9.630644798278809, "learning_rate": 1.2336650540262078e-05, "loss": 1.0384, "step": 8241 }, { "epoch": 1.33, "grad_norm": 8.905858993530273, "learning_rate": 1.2334960626456442e-05, "loss": 0.8851, "step": 8242 }, { "epoch": 1.33, "grad_norm": 10.61520004272461, "learning_rate": 1.2333270642120687e-05, "loss": 1.1918, "step": 8243 }, { "epoch": 1.33, "grad_norm": 10.3370943069458, "learning_rate": 1.2331580587305852e-05, "loss": 1.2194, "step": 8244 }, { "epoch": 1.33, "grad_norm": 8.991053581237793, "learning_rate": 1.2329890462062994e-05, "loss": 1.0095, "step": 8245 }, { "epoch": 1.33, "grad_norm": 9.877159118652344, "learning_rate": 1.2328200266443162e-05, "loss": 1.0344, "step": 8246 }, { "epoch": 1.33, "grad_norm": 10.179010391235352, "learning_rate": 1.2326510000497412e-05, "loss": 1.2807, "step": 8247 }, { "epoch": 1.33, "grad_norm": 10.855178833007812, "learning_rate": 1.2324819664276801e-05, "loss": 1.1518, "step": 8248 }, { "epoch": 1.33, "grad_norm": 9.63491439819336, "learning_rate": 1.2323129257832384e-05, "loss": 0.995, "step": 8249 }, { "epoch": 1.33, "grad_norm": 7.849437713623047, "learning_rate": 1.2321438781215225e-05, "loss": 0.9266, "step": 8250 }, { "epoch": 1.33, "grad_norm": 9.147154808044434, "learning_rate": 1.2319748234476387e-05, "loss": 1.0688, "step": 8251 }, { "epoch": 1.33, "grad_norm": 10.761174201965332, "learning_rate": 1.2318057617666934e-05, "loss": 1.017, "step": 8252 }, { "epoch": 1.33, "grad_norm": 9.15611457824707, "learning_rate": 1.2316366930837931e-05, "loss": 0.9693, "step": 8253 }, { "epoch": 1.33, "grad_norm": 8.530378341674805, "learning_rate": 1.2314676174040448e-05, "loss": 0.9492, "step": 8254 }, { "epoch": 1.33, "grad_norm": 8.897086143493652, "learning_rate": 1.231298534732556e-05, "loss": 0.7869, "step": 8255 }, { "epoch": 1.33, "grad_norm": 8.299312591552734, "learning_rate": 1.2311294450744333e-05, "loss": 0.9757, "step": 8256 }, { "epoch": 1.33, "grad_norm": 8.905709266662598, "learning_rate": 1.2309603484347847e-05, "loss": 1.2221, "step": 8257 }, { "epoch": 1.33, "grad_norm": 8.678620338439941, "learning_rate": 1.2307912448187183e-05, "loss": 1.1175, "step": 8258 }, { "epoch": 1.33, "grad_norm": 10.357931137084961, "learning_rate": 1.2306221342313415e-05, "loss": 1.0163, "step": 8259 }, { "epoch": 1.33, "grad_norm": 9.046357154846191, "learning_rate": 1.2304530166777625e-05, "loss": 1.0831, "step": 8260 }, { "epoch": 1.33, "grad_norm": 12.953594207763672, "learning_rate": 1.2302838921630903e-05, "loss": 1.1213, "step": 8261 }, { "epoch": 1.33, "grad_norm": 8.748960494995117, "learning_rate": 1.2301147606924326e-05, "loss": 1.0304, "step": 8262 }, { "epoch": 1.33, "grad_norm": 7.961829662322998, "learning_rate": 1.229945622270899e-05, "loss": 0.9078, "step": 8263 }, { "epoch": 1.33, "grad_norm": 10.178199768066406, "learning_rate": 1.2297764769035978e-05, "loss": 0.9297, "step": 8264 }, { "epoch": 1.33, "grad_norm": 9.38026237487793, "learning_rate": 1.2296073245956388e-05, "loss": 0.965, "step": 8265 }, { "epoch": 1.33, "grad_norm": 9.2213716506958, "learning_rate": 1.2294381653521313e-05, "loss": 1.0133, "step": 8266 }, { "epoch": 1.33, "grad_norm": 9.713032722473145, "learning_rate": 1.2292689991781845e-05, "loss": 1.0728, "step": 8267 }, { "epoch": 1.33, "grad_norm": 8.543131828308105, "learning_rate": 1.229099826078909e-05, "loss": 1.0801, "step": 8268 }, { "epoch": 1.33, "grad_norm": 8.908429145812988, "learning_rate": 1.2289306460594144e-05, "loss": 1.0763, "step": 8269 }, { "epoch": 1.33, "grad_norm": 8.427881240844727, "learning_rate": 1.2287614591248112e-05, "loss": 0.9779, "step": 8270 }, { "epoch": 1.33, "grad_norm": 10.06396484375, "learning_rate": 1.2285922652802098e-05, "loss": 1.1074, "step": 8271 }, { "epoch": 1.33, "grad_norm": 11.202256202697754, "learning_rate": 1.2284230645307205e-05, "loss": 1.1833, "step": 8272 }, { "epoch": 1.33, "grad_norm": 9.723554611206055, "learning_rate": 1.2282538568814547e-05, "loss": 1.333, "step": 8273 }, { "epoch": 1.33, "grad_norm": 10.689554214477539, "learning_rate": 1.2280846423375236e-05, "loss": 1.1204, "step": 8274 }, { "epoch": 1.33, "grad_norm": 11.055412292480469, "learning_rate": 1.2279154209040381e-05, "loss": 0.9769, "step": 8275 }, { "epoch": 1.33, "grad_norm": 9.376474380493164, "learning_rate": 1.2277461925861097e-05, "loss": 1.2446, "step": 8276 }, { "epoch": 1.33, "grad_norm": 9.678864479064941, "learning_rate": 1.2275769573888505e-05, "loss": 1.2407, "step": 8277 }, { "epoch": 1.33, "grad_norm": 8.828664779663086, "learning_rate": 1.2274077153173725e-05, "loss": 0.9943, "step": 8278 }, { "epoch": 1.33, "grad_norm": 8.98000717163086, "learning_rate": 1.2272384663767873e-05, "loss": 1.0967, "step": 8279 }, { "epoch": 1.33, "grad_norm": 8.653829574584961, "learning_rate": 1.2270692105722076e-05, "loss": 0.8888, "step": 8280 }, { "epoch": 1.33, "grad_norm": 9.374685287475586, "learning_rate": 1.226899947908746e-05, "loss": 1.099, "step": 8281 }, { "epoch": 1.33, "grad_norm": 10.736892700195312, "learning_rate": 1.2267306783915154e-05, "loss": 1.0107, "step": 8282 }, { "epoch": 1.33, "grad_norm": 9.36003589630127, "learning_rate": 1.2265614020256283e-05, "loss": 1.2381, "step": 8283 }, { "epoch": 1.33, "grad_norm": 9.615129470825195, "learning_rate": 1.2263921188161983e-05, "loss": 1.1812, "step": 8284 }, { "epoch": 1.33, "grad_norm": 9.797057151794434, "learning_rate": 1.2262228287683384e-05, "loss": 0.9323, "step": 8285 }, { "epoch": 1.33, "grad_norm": 7.49188232421875, "learning_rate": 1.2260535318871628e-05, "loss": 1.0442, "step": 8286 }, { "epoch": 1.33, "grad_norm": 8.80662727355957, "learning_rate": 1.2258842281777848e-05, "loss": 1.1015, "step": 8287 }, { "epoch": 1.33, "grad_norm": 10.206768989562988, "learning_rate": 1.2257149176453184e-05, "loss": 1.0327, "step": 8288 }, { "epoch": 1.33, "grad_norm": 10.311288833618164, "learning_rate": 1.2255456002948782e-05, "loss": 1.1876, "step": 8289 }, { "epoch": 1.33, "grad_norm": 8.526837348937988, "learning_rate": 1.2253762761315787e-05, "loss": 1.0905, "step": 8290 }, { "epoch": 1.33, "grad_norm": 9.064933776855469, "learning_rate": 1.2252069451605337e-05, "loss": 1.2169, "step": 8291 }, { "epoch": 1.33, "grad_norm": 8.02882194519043, "learning_rate": 1.2250376073868588e-05, "loss": 1.0717, "step": 8292 }, { "epoch": 1.33, "grad_norm": 13.4649658203125, "learning_rate": 1.2248682628156687e-05, "loss": 0.9028, "step": 8293 }, { "epoch": 1.34, "grad_norm": 9.830317497253418, "learning_rate": 1.224698911452079e-05, "loss": 1.2025, "step": 8294 }, { "epoch": 1.34, "grad_norm": 8.899795532226562, "learning_rate": 1.2245295533012047e-05, "loss": 0.9998, "step": 8295 }, { "epoch": 1.34, "grad_norm": 8.710738182067871, "learning_rate": 1.2243601883681617e-05, "loss": 0.9084, "step": 8296 }, { "epoch": 1.34, "grad_norm": 10.385910987854004, "learning_rate": 1.2241908166580658e-05, "loss": 1.1097, "step": 8297 }, { "epoch": 1.34, "grad_norm": 8.499393463134766, "learning_rate": 1.224021438176033e-05, "loss": 0.9081, "step": 8298 }, { "epoch": 1.34, "grad_norm": 8.311919212341309, "learning_rate": 1.2238520529271798e-05, "loss": 0.9284, "step": 8299 }, { "epoch": 1.34, "grad_norm": 10.536434173583984, "learning_rate": 1.2236826609166223e-05, "loss": 1.1454, "step": 8300 }, { "epoch": 1.34, "grad_norm": 10.388214111328125, "learning_rate": 1.2235132621494777e-05, "loss": 1.3271, "step": 8301 }, { "epoch": 1.34, "grad_norm": 13.27013111114502, "learning_rate": 1.2233438566308625e-05, "loss": 1.2221, "step": 8302 }, { "epoch": 1.34, "grad_norm": 9.359106063842773, "learning_rate": 1.2231744443658935e-05, "loss": 1.0459, "step": 8303 }, { "epoch": 1.34, "grad_norm": 7.6850481033325195, "learning_rate": 1.2230050253596885e-05, "loss": 0.8681, "step": 8304 }, { "epoch": 1.34, "grad_norm": 8.282557487487793, "learning_rate": 1.2228355996173652e-05, "loss": 0.9481, "step": 8305 }, { "epoch": 1.34, "grad_norm": 10.14206600189209, "learning_rate": 1.2226661671440408e-05, "loss": 1.0596, "step": 8306 }, { "epoch": 1.34, "grad_norm": 9.657909393310547, "learning_rate": 1.222496727944833e-05, "loss": 0.8746, "step": 8307 }, { "epoch": 1.34, "grad_norm": 9.886724472045898, "learning_rate": 1.2223272820248604e-05, "loss": 1.1289, "step": 8308 }, { "epoch": 1.34, "grad_norm": 10.350300788879395, "learning_rate": 1.2221578293892413e-05, "loss": 1.197, "step": 8309 }, { "epoch": 1.34, "grad_norm": 8.055806159973145, "learning_rate": 1.221988370043094e-05, "loss": 1.0198, "step": 8310 }, { "epoch": 1.34, "grad_norm": 9.282902717590332, "learning_rate": 1.2218189039915371e-05, "loss": 0.837, "step": 8311 }, { "epoch": 1.34, "grad_norm": 10.429673194885254, "learning_rate": 1.2216494312396895e-05, "loss": 1.1082, "step": 8312 }, { "epoch": 1.34, "grad_norm": 10.568690299987793, "learning_rate": 1.2214799517926713e-05, "loss": 1.0165, "step": 8313 }, { "epoch": 1.34, "grad_norm": 11.51242446899414, "learning_rate": 1.2213104656556003e-05, "loss": 1.2542, "step": 8314 }, { "epoch": 1.34, "grad_norm": 13.246382713317871, "learning_rate": 1.221140972833597e-05, "loss": 1.0721, "step": 8315 }, { "epoch": 1.34, "grad_norm": 9.603836059570312, "learning_rate": 1.2209714733317806e-05, "loss": 1.0479, "step": 8316 }, { "epoch": 1.34, "grad_norm": 10.107606887817383, "learning_rate": 1.2208019671552712e-05, "loss": 1.0336, "step": 8317 }, { "epoch": 1.34, "grad_norm": 8.343608856201172, "learning_rate": 1.2206324543091894e-05, "loss": 1.2263, "step": 8318 }, { "epoch": 1.34, "grad_norm": 8.755287170410156, "learning_rate": 1.220462934798655e-05, "loss": 1.1306, "step": 8319 }, { "epoch": 1.34, "grad_norm": 9.640155792236328, "learning_rate": 1.2202934086287886e-05, "loss": 1.2465, "step": 8320 }, { "epoch": 1.34, "grad_norm": 10.192360877990723, "learning_rate": 1.2201238758047111e-05, "loss": 1.162, "step": 8321 }, { "epoch": 1.34, "grad_norm": 9.86327838897705, "learning_rate": 1.2199543363315433e-05, "loss": 1.0844, "step": 8322 }, { "epoch": 1.34, "grad_norm": 9.32214641571045, "learning_rate": 1.2197847902144062e-05, "loss": 1.1995, "step": 8323 }, { "epoch": 1.34, "grad_norm": 8.66114330291748, "learning_rate": 1.2196152374584214e-05, "loss": 1.0707, "step": 8324 }, { "epoch": 1.34, "grad_norm": 8.351189613342285, "learning_rate": 1.2194456780687106e-05, "loss": 1.2162, "step": 8325 }, { "epoch": 1.34, "grad_norm": 8.163651466369629, "learning_rate": 1.2192761120503946e-05, "loss": 1.2377, "step": 8326 }, { "epoch": 1.34, "grad_norm": 8.4091157913208, "learning_rate": 1.2191065394085962e-05, "loss": 0.9382, "step": 8327 }, { "epoch": 1.34, "grad_norm": 11.360486030578613, "learning_rate": 1.2189369601484373e-05, "loss": 1.017, "step": 8328 }, { "epoch": 1.34, "grad_norm": 8.566356658935547, "learning_rate": 1.2187673742750406e-05, "loss": 0.9124, "step": 8329 }, { "epoch": 1.34, "grad_norm": 8.58873462677002, "learning_rate": 1.2185977817935279e-05, "loss": 0.9947, "step": 8330 }, { "epoch": 1.34, "grad_norm": 9.538103103637695, "learning_rate": 1.2184281827090225e-05, "loss": 1.3022, "step": 8331 }, { "epoch": 1.34, "grad_norm": 8.023083686828613, "learning_rate": 1.2182585770266471e-05, "loss": 0.9004, "step": 8332 }, { "epoch": 1.34, "grad_norm": 7.901673316955566, "learning_rate": 1.218088964751525e-05, "loss": 0.9567, "step": 8333 }, { "epoch": 1.34, "grad_norm": 9.30582046508789, "learning_rate": 1.2179193458887793e-05, "loss": 1.0026, "step": 8334 }, { "epoch": 1.34, "grad_norm": 8.26823902130127, "learning_rate": 1.2177497204435334e-05, "loss": 1.0393, "step": 8335 }, { "epoch": 1.34, "grad_norm": 8.83266830444336, "learning_rate": 1.2175800884209114e-05, "loss": 0.9348, "step": 8336 }, { "epoch": 1.34, "grad_norm": 10.471627235412598, "learning_rate": 1.2174104498260372e-05, "loss": 1.2853, "step": 8337 }, { "epoch": 1.34, "grad_norm": 9.339386940002441, "learning_rate": 1.2172408046640347e-05, "loss": 0.9543, "step": 8338 }, { "epoch": 1.34, "grad_norm": 9.090646743774414, "learning_rate": 1.2170711529400283e-05, "loss": 0.9931, "step": 8339 }, { "epoch": 1.34, "grad_norm": 11.257972717285156, "learning_rate": 1.216901494659143e-05, "loss": 1.1783, "step": 8340 }, { "epoch": 1.34, "grad_norm": 8.801546096801758, "learning_rate": 1.2167318298265024e-05, "loss": 1.0832, "step": 8341 }, { "epoch": 1.34, "grad_norm": 8.408296585083008, "learning_rate": 1.2165621584472325e-05, "loss": 0.9679, "step": 8342 }, { "epoch": 1.34, "grad_norm": 9.693060874938965, "learning_rate": 1.216392480526458e-05, "loss": 0.9689, "step": 8343 }, { "epoch": 1.34, "grad_norm": 9.319230079650879, "learning_rate": 1.216222796069304e-05, "loss": 1.2329, "step": 8344 }, { "epoch": 1.34, "grad_norm": 10.388529777526855, "learning_rate": 1.2160531050808966e-05, "loss": 1.0692, "step": 8345 }, { "epoch": 1.34, "grad_norm": 8.775931358337402, "learning_rate": 1.2158834075663611e-05, "loss": 1.0039, "step": 8346 }, { "epoch": 1.34, "grad_norm": 7.845221996307373, "learning_rate": 1.2157137035308228e-05, "loss": 1.1615, "step": 8347 }, { "epoch": 1.34, "grad_norm": 8.094454765319824, "learning_rate": 1.2155439929794093e-05, "loss": 0.9953, "step": 8348 }, { "epoch": 1.34, "grad_norm": 8.920541763305664, "learning_rate": 1.2153742759172456e-05, "loss": 0.8857, "step": 8349 }, { "epoch": 1.34, "grad_norm": 8.587845802307129, "learning_rate": 1.2152045523494586e-05, "loss": 1.1768, "step": 8350 }, { "epoch": 1.34, "grad_norm": 7.560948848724365, "learning_rate": 1.2150348222811752e-05, "loss": 0.9505, "step": 8351 }, { "epoch": 1.34, "grad_norm": 9.547173500061035, "learning_rate": 1.2148650857175224e-05, "loss": 1.2436, "step": 8352 }, { "epoch": 1.34, "grad_norm": 8.977837562561035, "learning_rate": 1.2146953426636268e-05, "loss": 1.1534, "step": 8353 }, { "epoch": 1.34, "grad_norm": 10.929222106933594, "learning_rate": 1.2145255931246156e-05, "loss": 1.0828, "step": 8354 }, { "epoch": 1.34, "grad_norm": 9.642088890075684, "learning_rate": 1.2143558371056168e-05, "loss": 0.981, "step": 8355 }, { "epoch": 1.35, "grad_norm": 9.793185234069824, "learning_rate": 1.2141860746117578e-05, "loss": 1.1651, "step": 8356 }, { "epoch": 1.35, "grad_norm": 10.922478675842285, "learning_rate": 1.2140163056481665e-05, "loss": 1.234, "step": 8357 }, { "epoch": 1.35, "grad_norm": 7.891573905944824, "learning_rate": 1.2138465302199709e-05, "loss": 1.139, "step": 8358 }, { "epoch": 1.35, "grad_norm": 8.74614429473877, "learning_rate": 1.2136767483322993e-05, "loss": 0.9859, "step": 8359 }, { "epoch": 1.35, "grad_norm": 8.137582778930664, "learning_rate": 1.2135069599902805e-05, "loss": 0.9757, "step": 8360 }, { "epoch": 1.35, "grad_norm": 11.96048641204834, "learning_rate": 1.2133371651990427e-05, "loss": 1.1344, "step": 8361 }, { "epoch": 1.35, "grad_norm": 8.594531059265137, "learning_rate": 1.2131673639637148e-05, "loss": 1.1934, "step": 8362 }, { "epoch": 1.35, "grad_norm": 11.34569263458252, "learning_rate": 1.2129975562894258e-05, "loss": 0.9756, "step": 8363 }, { "epoch": 1.35, "grad_norm": 9.229321479797363, "learning_rate": 1.2128277421813052e-05, "loss": 1.0224, "step": 8364 }, { "epoch": 1.35, "grad_norm": 12.145389556884766, "learning_rate": 1.2126579216444825e-05, "loss": 1.2005, "step": 8365 }, { "epoch": 1.35, "grad_norm": 9.472363471984863, "learning_rate": 1.2124880946840867e-05, "loss": 1.3197, "step": 8366 }, { "epoch": 1.35, "grad_norm": 9.37363052368164, "learning_rate": 1.212318261305248e-05, "loss": 1.1973, "step": 8367 }, { "epoch": 1.35, "grad_norm": 9.631837844848633, "learning_rate": 1.2121484215130965e-05, "loss": 0.9778, "step": 8368 }, { "epoch": 1.35, "grad_norm": 7.324979305267334, "learning_rate": 1.2119785753127625e-05, "loss": 0.9844, "step": 8369 }, { "epoch": 1.35, "grad_norm": 8.275768280029297, "learning_rate": 1.211808722709376e-05, "loss": 0.9355, "step": 8370 }, { "epoch": 1.35, "grad_norm": 8.580102920532227, "learning_rate": 1.211638863708068e-05, "loss": 1.0194, "step": 8371 }, { "epoch": 1.35, "grad_norm": 7.799231052398682, "learning_rate": 1.2114689983139691e-05, "loss": 0.7928, "step": 8372 }, { "epoch": 1.35, "grad_norm": 10.290327072143555, "learning_rate": 1.2112991265322103e-05, "loss": 0.991, "step": 8373 }, { "epoch": 1.35, "grad_norm": 8.983916282653809, "learning_rate": 1.2111292483679225e-05, "loss": 1.0875, "step": 8374 }, { "epoch": 1.35, "grad_norm": 9.547517776489258, "learning_rate": 1.2109593638262373e-05, "loss": 1.1308, "step": 8375 }, { "epoch": 1.35, "grad_norm": 8.623374938964844, "learning_rate": 1.2107894729122866e-05, "loss": 1.286, "step": 8376 }, { "epoch": 1.35, "grad_norm": 8.08962345123291, "learning_rate": 1.2106195756312016e-05, "loss": 0.9233, "step": 8377 }, { "epoch": 1.35, "grad_norm": 10.145258903503418, "learning_rate": 1.210449671988114e-05, "loss": 1.3581, "step": 8378 }, { "epoch": 1.35, "grad_norm": 9.863043785095215, "learning_rate": 1.210279761988157e-05, "loss": 1.2397, "step": 8379 }, { "epoch": 1.35, "grad_norm": 8.329998016357422, "learning_rate": 1.210109845636462e-05, "loss": 0.953, "step": 8380 }, { "epoch": 1.35, "grad_norm": 8.339771270751953, "learning_rate": 1.2099399229381618e-05, "loss": 1.009, "step": 8381 }, { "epoch": 1.35, "grad_norm": 10.481443405151367, "learning_rate": 1.2097699938983891e-05, "loss": 1.0804, "step": 8382 }, { "epoch": 1.35, "grad_norm": 9.087063789367676, "learning_rate": 1.2096000585222769e-05, "loss": 1.2159, "step": 8383 }, { "epoch": 1.35, "grad_norm": 9.86121940612793, "learning_rate": 1.2094301168149586e-05, "loss": 0.991, "step": 8384 }, { "epoch": 1.35, "grad_norm": 9.190698623657227, "learning_rate": 1.2092601687815664e-05, "loss": 1.0887, "step": 8385 }, { "epoch": 1.35, "grad_norm": 9.127113342285156, "learning_rate": 1.2090902144272347e-05, "loss": 1.0615, "step": 8386 }, { "epoch": 1.35, "grad_norm": 8.947484970092773, "learning_rate": 1.208920253757097e-05, "loss": 0.9322, "step": 8387 }, { "epoch": 1.35, "grad_norm": 9.06398868560791, "learning_rate": 1.208750286776287e-05, "loss": 1.1028, "step": 8388 }, { "epoch": 1.35, "grad_norm": 8.303559303283691, "learning_rate": 1.2085803134899387e-05, "loss": 1.1026, "step": 8389 }, { "epoch": 1.35, "grad_norm": 9.697983741760254, "learning_rate": 1.2084103339031867e-05, "loss": 0.8075, "step": 8390 }, { "epoch": 1.35, "grad_norm": 8.383864402770996, "learning_rate": 1.2082403480211653e-05, "loss": 1.0573, "step": 8391 }, { "epoch": 1.35, "grad_norm": 9.07331371307373, "learning_rate": 1.2080703558490087e-05, "loss": 1.2396, "step": 8392 }, { "epoch": 1.35, "grad_norm": 10.099796295166016, "learning_rate": 1.207900357391852e-05, "loss": 1.0558, "step": 8393 }, { "epoch": 1.35, "grad_norm": 9.434783935546875, "learning_rate": 1.2077303526548303e-05, "loss": 1.0951, "step": 8394 }, { "epoch": 1.35, "grad_norm": 10.077035903930664, "learning_rate": 1.2075603416430785e-05, "loss": 1.127, "step": 8395 }, { "epoch": 1.35, "grad_norm": 7.8953471183776855, "learning_rate": 1.2073903243617326e-05, "loss": 1.0099, "step": 8396 }, { "epoch": 1.35, "grad_norm": 10.359907150268555, "learning_rate": 1.2072203008159273e-05, "loss": 1.1767, "step": 8397 }, { "epoch": 1.35, "grad_norm": 9.154914855957031, "learning_rate": 1.2070502710107987e-05, "loss": 1.112, "step": 8398 }, { "epoch": 1.35, "grad_norm": 8.307169914245605, "learning_rate": 1.2068802349514833e-05, "loss": 0.8783, "step": 8399 }, { "epoch": 1.35, "grad_norm": 9.670125961303711, "learning_rate": 1.2067101926431164e-05, "loss": 1.2307, "step": 8400 }, { "epoch": 1.35, "grad_norm": 9.51944351196289, "learning_rate": 1.2065401440908346e-05, "loss": 0.9753, "step": 8401 }, { "epoch": 1.35, "grad_norm": 8.879389762878418, "learning_rate": 1.2063700892997745e-05, "loss": 1.2039, "step": 8402 }, { "epoch": 1.35, "grad_norm": 8.101125717163086, "learning_rate": 1.2062000282750732e-05, "loss": 1.0883, "step": 8403 }, { "epoch": 1.35, "grad_norm": 8.5396146774292, "learning_rate": 1.2060299610218669e-05, "loss": 0.8983, "step": 8404 }, { "epoch": 1.35, "grad_norm": 7.618843078613281, "learning_rate": 1.2058598875452927e-05, "loss": 0.8609, "step": 8405 }, { "epoch": 1.35, "grad_norm": 9.886930465698242, "learning_rate": 1.2056898078504881e-05, "loss": 1.2473, "step": 8406 }, { "epoch": 1.35, "grad_norm": 12.006465911865234, "learning_rate": 1.2055197219425909e-05, "loss": 1.08, "step": 8407 }, { "epoch": 1.35, "grad_norm": 9.414081573486328, "learning_rate": 1.205349629826738e-05, "loss": 1.0241, "step": 8408 }, { "epoch": 1.35, "grad_norm": 8.669534683227539, "learning_rate": 1.205179531508068e-05, "loss": 1.0339, "step": 8409 }, { "epoch": 1.35, "grad_norm": 10.092880249023438, "learning_rate": 1.2050094269917181e-05, "loss": 1.1866, "step": 8410 }, { "epoch": 1.35, "grad_norm": 11.131454467773438, "learning_rate": 1.2048393162828273e-05, "loss": 1.4035, "step": 8411 }, { "epoch": 1.35, "grad_norm": 8.785684585571289, "learning_rate": 1.2046691993865334e-05, "loss": 0.9235, "step": 8412 }, { "epoch": 1.35, "grad_norm": 9.656632423400879, "learning_rate": 1.2044990763079751e-05, "loss": 1.1541, "step": 8413 }, { "epoch": 1.35, "grad_norm": 9.113061904907227, "learning_rate": 1.2043289470522911e-05, "loss": 1.0358, "step": 8414 }, { "epoch": 1.35, "grad_norm": 10.786919593811035, "learning_rate": 1.204158811624621e-05, "loss": 1.2591, "step": 8415 }, { "epoch": 1.35, "grad_norm": 10.690130233764648, "learning_rate": 1.2039886700301031e-05, "loss": 0.9022, "step": 8416 }, { "epoch": 1.35, "grad_norm": 8.64431381225586, "learning_rate": 1.203818522273877e-05, "loss": 0.9424, "step": 8417 }, { "epoch": 1.36, "grad_norm": 10.757328987121582, "learning_rate": 1.2036483683610823e-05, "loss": 1.0999, "step": 8418 }, { "epoch": 1.36, "grad_norm": 7.787920951843262, "learning_rate": 1.2034782082968586e-05, "loss": 1.0447, "step": 8419 }, { "epoch": 1.36, "grad_norm": 9.516997337341309, "learning_rate": 1.2033080420863457e-05, "loss": 1.1414, "step": 8420 }, { "epoch": 1.36, "grad_norm": 9.350044250488281, "learning_rate": 1.2031378697346838e-05, "loss": 0.9048, "step": 8421 }, { "epoch": 1.36, "grad_norm": 8.62447452545166, "learning_rate": 1.2029676912470132e-05, "loss": 1.1547, "step": 8422 }, { "epoch": 1.36, "grad_norm": 8.916031837463379, "learning_rate": 1.2027975066284745e-05, "loss": 1.1743, "step": 8423 }, { "epoch": 1.36, "grad_norm": 9.509088516235352, "learning_rate": 1.2026273158842076e-05, "loss": 0.9899, "step": 8424 }, { "epoch": 1.36, "grad_norm": 8.527909278869629, "learning_rate": 1.202457119019354e-05, "loss": 0.9647, "step": 8425 }, { "epoch": 1.36, "grad_norm": 10.716343879699707, "learning_rate": 1.2022869160390545e-05, "loss": 0.8542, "step": 8426 }, { "epoch": 1.36, "grad_norm": 8.95804500579834, "learning_rate": 1.2021167069484502e-05, "loss": 1.1302, "step": 8427 }, { "epoch": 1.36, "grad_norm": 8.101163864135742, "learning_rate": 1.2019464917526823e-05, "loss": 0.9126, "step": 8428 }, { "epoch": 1.36, "grad_norm": 10.783391952514648, "learning_rate": 1.2017762704568927e-05, "loss": 1.2844, "step": 8429 }, { "epoch": 1.36, "grad_norm": 10.366037368774414, "learning_rate": 1.201606043066223e-05, "loss": 1.2182, "step": 8430 }, { "epoch": 1.36, "grad_norm": 9.973821640014648, "learning_rate": 1.201435809585815e-05, "loss": 1.0484, "step": 8431 }, { "epoch": 1.36, "grad_norm": 9.066903114318848, "learning_rate": 1.2012655700208108e-05, "loss": 1.0598, "step": 8432 }, { "epoch": 1.36, "grad_norm": 9.764816284179688, "learning_rate": 1.2010953243763528e-05, "loss": 1.054, "step": 8433 }, { "epoch": 1.36, "grad_norm": 7.178859233856201, "learning_rate": 1.2009250726575835e-05, "loss": 0.9754, "step": 8434 }, { "epoch": 1.36, "grad_norm": 9.626058578491211, "learning_rate": 1.2007548148696455e-05, "loss": 1.1027, "step": 8435 }, { "epoch": 1.36, "grad_norm": 7.941903591156006, "learning_rate": 1.2005845510176815e-05, "loss": 1.1754, "step": 8436 }, { "epoch": 1.36, "grad_norm": 8.995183944702148, "learning_rate": 1.2004142811068344e-05, "loss": 1.0409, "step": 8437 }, { "epoch": 1.36, "grad_norm": 9.196853637695312, "learning_rate": 1.2002440051422477e-05, "loss": 0.8892, "step": 8438 }, { "epoch": 1.36, "grad_norm": 9.959275245666504, "learning_rate": 1.2000737231290647e-05, "loss": 1.2541, "step": 8439 }, { "epoch": 1.36, "grad_norm": 9.886734008789062, "learning_rate": 1.1999034350724288e-05, "loss": 0.9582, "step": 8440 }, { "epoch": 1.36, "grad_norm": 13.416372299194336, "learning_rate": 1.1997331409774839e-05, "loss": 1.2043, "step": 8441 }, { "epoch": 1.36, "grad_norm": 9.798691749572754, "learning_rate": 1.1995628408493742e-05, "loss": 1.049, "step": 8442 }, { "epoch": 1.36, "grad_norm": 8.060497283935547, "learning_rate": 1.199392534693243e-05, "loss": 1.1763, "step": 8443 }, { "epoch": 1.36, "grad_norm": 8.563474655151367, "learning_rate": 1.1992222225142356e-05, "loss": 0.9332, "step": 8444 }, { "epoch": 1.36, "grad_norm": 10.304430961608887, "learning_rate": 1.1990519043174955e-05, "loss": 1.0274, "step": 8445 }, { "epoch": 1.36, "grad_norm": 10.245302200317383, "learning_rate": 1.1988815801081681e-05, "loss": 1.064, "step": 8446 }, { "epoch": 1.36, "grad_norm": 11.003527641296387, "learning_rate": 1.1987112498913982e-05, "loss": 0.9024, "step": 8447 }, { "epoch": 1.36, "grad_norm": 11.418623924255371, "learning_rate": 1.1985409136723298e-05, "loss": 1.144, "step": 8448 }, { "epoch": 1.36, "grad_norm": 8.802877426147461, "learning_rate": 1.1983705714561093e-05, "loss": 0.9332, "step": 8449 }, { "epoch": 1.36, "grad_norm": 10.765286445617676, "learning_rate": 1.198200223247882e-05, "loss": 1.2365, "step": 8450 }, { "epoch": 1.36, "grad_norm": 9.069960594177246, "learning_rate": 1.1980298690527927e-05, "loss": 1.0447, "step": 8451 }, { "epoch": 1.36, "grad_norm": 9.831049919128418, "learning_rate": 1.1978595088759875e-05, "loss": 0.8917, "step": 8452 }, { "epoch": 1.36, "grad_norm": 8.27538013458252, "learning_rate": 1.1976891427226125e-05, "loss": 0.8199, "step": 8453 }, { "epoch": 1.36, "grad_norm": 9.337748527526855, "learning_rate": 1.1975187705978137e-05, "loss": 1.085, "step": 8454 }, { "epoch": 1.36, "grad_norm": 10.768242835998535, "learning_rate": 1.1973483925067374e-05, "loss": 1.1912, "step": 8455 }, { "epoch": 1.36, "grad_norm": 8.995013236999512, "learning_rate": 1.1971780084545297e-05, "loss": 1.0821, "step": 8456 }, { "epoch": 1.36, "grad_norm": 10.17776870727539, "learning_rate": 1.197007618446338e-05, "loss": 1.2711, "step": 8457 }, { "epoch": 1.36, "grad_norm": 8.292927742004395, "learning_rate": 1.1968372224873084e-05, "loss": 0.874, "step": 8458 }, { "epoch": 1.36, "grad_norm": 9.907370567321777, "learning_rate": 1.196666820582588e-05, "loss": 1.0733, "step": 8459 }, { "epoch": 1.36, "grad_norm": 9.053165435791016, "learning_rate": 1.1964964127373245e-05, "loss": 1.192, "step": 8460 }, { "epoch": 1.36, "grad_norm": 8.409262657165527, "learning_rate": 1.1963259989566646e-05, "loss": 1.1373, "step": 8461 }, { "epoch": 1.36, "grad_norm": 8.671050071716309, "learning_rate": 1.1961555792457567e-05, "loss": 1.1296, "step": 8462 }, { "epoch": 1.36, "grad_norm": 8.265107154846191, "learning_rate": 1.1959851536097476e-05, "loss": 0.9755, "step": 8463 }, { "epoch": 1.36, "grad_norm": 7.077902793884277, "learning_rate": 1.1958147220537855e-05, "loss": 0.8986, "step": 8464 }, { "epoch": 1.36, "grad_norm": 9.191844940185547, "learning_rate": 1.1956442845830187e-05, "loss": 0.9882, "step": 8465 }, { "epoch": 1.36, "grad_norm": 9.156405448913574, "learning_rate": 1.1954738412025955e-05, "loss": 1.2212, "step": 8466 }, { "epoch": 1.36, "grad_norm": 9.103591918945312, "learning_rate": 1.195303391917664e-05, "loss": 1.058, "step": 8467 }, { "epoch": 1.36, "grad_norm": 10.2634859085083, "learning_rate": 1.1951329367333728e-05, "loss": 1.1835, "step": 8468 }, { "epoch": 1.36, "grad_norm": 8.467863082885742, "learning_rate": 1.194962475654871e-05, "loss": 1.2182, "step": 8469 }, { "epoch": 1.36, "grad_norm": 9.8978910446167, "learning_rate": 1.1947920086873072e-05, "loss": 1.0727, "step": 8470 }, { "epoch": 1.36, "grad_norm": 9.267866134643555, "learning_rate": 1.194621535835831e-05, "loss": 0.971, "step": 8471 }, { "epoch": 1.36, "grad_norm": 11.36441421508789, "learning_rate": 1.1944510571055916e-05, "loss": 1.2788, "step": 8472 }, { "epoch": 1.36, "grad_norm": 8.629197120666504, "learning_rate": 1.194280572501738e-05, "loss": 0.9378, "step": 8473 }, { "epoch": 1.36, "grad_norm": 8.615674018859863, "learning_rate": 1.194110082029421e-05, "loss": 1.0269, "step": 8474 }, { "epoch": 1.36, "grad_norm": 8.832582473754883, "learning_rate": 1.1939395856937892e-05, "loss": 1.1331, "step": 8475 }, { "epoch": 1.36, "grad_norm": 8.769596099853516, "learning_rate": 1.1937690834999933e-05, "loss": 0.9134, "step": 8476 }, { "epoch": 1.36, "grad_norm": 9.563786506652832, "learning_rate": 1.1935985754531833e-05, "loss": 1.2276, "step": 8477 }, { "epoch": 1.36, "grad_norm": 10.995072364807129, "learning_rate": 1.1934280615585098e-05, "loss": 1.1677, "step": 8478 }, { "epoch": 1.36, "grad_norm": 8.805563926696777, "learning_rate": 1.1932575418211232e-05, "loss": 0.971, "step": 8479 }, { "epoch": 1.37, "grad_norm": 9.5982666015625, "learning_rate": 1.193087016246174e-05, "loss": 1.187, "step": 8480 }, { "epoch": 1.37, "grad_norm": 8.491183280944824, "learning_rate": 1.192916484838814e-05, "loss": 1.0176, "step": 8481 }, { "epoch": 1.37, "grad_norm": 9.52129077911377, "learning_rate": 1.1927459476041932e-05, "loss": 1.1853, "step": 8482 }, { "epoch": 1.37, "grad_norm": 8.859310150146484, "learning_rate": 1.1925754045474635e-05, "loss": 0.8611, "step": 8483 }, { "epoch": 1.37, "grad_norm": 10.648866653442383, "learning_rate": 1.1924048556737762e-05, "loss": 1.146, "step": 8484 }, { "epoch": 1.37, "grad_norm": 8.817496299743652, "learning_rate": 1.1922343009882828e-05, "loss": 1.066, "step": 8485 }, { "epoch": 1.37, "grad_norm": 8.86705207824707, "learning_rate": 1.1920637404961357e-05, "loss": 1.1233, "step": 8486 }, { "epoch": 1.37, "grad_norm": 9.445427894592285, "learning_rate": 1.191893174202486e-05, "loss": 1.2233, "step": 8487 }, { "epoch": 1.37, "grad_norm": 11.967300415039062, "learning_rate": 1.191722602112486e-05, "loss": 0.9803, "step": 8488 }, { "epoch": 1.37, "grad_norm": 8.924722671508789, "learning_rate": 1.1915520242312888e-05, "loss": 1.0685, "step": 8489 }, { "epoch": 1.37, "grad_norm": 9.762182235717773, "learning_rate": 1.1913814405640461e-05, "loss": 1.2165, "step": 8490 }, { "epoch": 1.37, "grad_norm": 11.709348678588867, "learning_rate": 1.1912108511159106e-05, "loss": 1.1559, "step": 8491 }, { "epoch": 1.37, "grad_norm": 10.96057415008545, "learning_rate": 1.1910402558920358e-05, "loss": 1.1445, "step": 8492 }, { "epoch": 1.37, "grad_norm": 8.083996772766113, "learning_rate": 1.1908696548975742e-05, "loss": 0.969, "step": 8493 }, { "epoch": 1.37, "grad_norm": 9.59740924835205, "learning_rate": 1.190699048137679e-05, "loss": 1.1554, "step": 8494 }, { "epoch": 1.37, "grad_norm": 10.898894309997559, "learning_rate": 1.1905284356175037e-05, "loss": 1.0361, "step": 8495 }, { "epoch": 1.37, "grad_norm": 9.242585182189941, "learning_rate": 1.1903578173422019e-05, "loss": 1.125, "step": 8496 }, { "epoch": 1.37, "grad_norm": 8.57413101196289, "learning_rate": 1.1901871933169274e-05, "loss": 1.3159, "step": 8497 }, { "epoch": 1.37, "grad_norm": 10.522805213928223, "learning_rate": 1.1900165635468335e-05, "loss": 1.192, "step": 8498 }, { "epoch": 1.37, "grad_norm": 19.281551361083984, "learning_rate": 1.1898459280370747e-05, "loss": 0.9167, "step": 8499 }, { "epoch": 1.37, "grad_norm": 8.746087074279785, "learning_rate": 1.1896752867928054e-05, "loss": 0.9815, "step": 8500 }, { "epoch": 1.37, "grad_norm": 8.875988960266113, "learning_rate": 1.18950463981918e-05, "loss": 1.0383, "step": 8501 }, { "epoch": 1.37, "grad_norm": 10.46386432647705, "learning_rate": 1.1893339871213527e-05, "loss": 1.1296, "step": 8502 }, { "epoch": 1.37, "grad_norm": 11.49790096282959, "learning_rate": 1.1891633287044783e-05, "loss": 1.1387, "step": 8503 }, { "epoch": 1.37, "grad_norm": 9.123188972473145, "learning_rate": 1.1889926645737121e-05, "loss": 1.1967, "step": 8504 }, { "epoch": 1.37, "grad_norm": 9.869046211242676, "learning_rate": 1.1888219947342095e-05, "loss": 1.0022, "step": 8505 }, { "epoch": 1.37, "grad_norm": 9.75482177734375, "learning_rate": 1.1886513191911246e-05, "loss": 1.141, "step": 8506 }, { "epoch": 1.37, "grad_norm": 10.703893661499023, "learning_rate": 1.1884806379496139e-05, "loss": 0.9483, "step": 8507 }, { "epoch": 1.37, "grad_norm": 9.76930046081543, "learning_rate": 1.1883099510148324e-05, "loss": 1.2208, "step": 8508 }, { "epoch": 1.37, "grad_norm": 12.807409286499023, "learning_rate": 1.1881392583919363e-05, "loss": 1.1571, "step": 8509 }, { "epoch": 1.37, "grad_norm": 10.655352592468262, "learning_rate": 1.1879685600860814e-05, "loss": 1.0031, "step": 8510 }, { "epoch": 1.37, "grad_norm": 12.262521743774414, "learning_rate": 1.1877978561024236e-05, "loss": 1.1072, "step": 8511 }, { "epoch": 1.37, "grad_norm": 9.818887710571289, "learning_rate": 1.1876271464461198e-05, "loss": 1.032, "step": 8512 }, { "epoch": 1.37, "grad_norm": 9.314018249511719, "learning_rate": 1.1874564311223262e-05, "loss": 0.8226, "step": 8513 }, { "epoch": 1.37, "grad_norm": 10.823141098022461, "learning_rate": 1.1872857101361992e-05, "loss": 1.2098, "step": 8514 }, { "epoch": 1.37, "grad_norm": 11.385955810546875, "learning_rate": 1.1871149834928956e-05, "loss": 0.9682, "step": 8515 }, { "epoch": 1.37, "grad_norm": 8.468506813049316, "learning_rate": 1.1869442511975728e-05, "loss": 1.2307, "step": 8516 }, { "epoch": 1.37, "grad_norm": 10.126565933227539, "learning_rate": 1.1867735132553879e-05, "loss": 0.9158, "step": 8517 }, { "epoch": 1.37, "grad_norm": 9.790985107421875, "learning_rate": 1.1866027696714979e-05, "loss": 0.989, "step": 8518 }, { "epoch": 1.37, "grad_norm": 8.520512580871582, "learning_rate": 1.1864320204510605e-05, "loss": 0.9194, "step": 8519 }, { "epoch": 1.37, "grad_norm": 9.724624633789062, "learning_rate": 1.1862612655992333e-05, "loss": 0.952, "step": 8520 }, { "epoch": 1.37, "grad_norm": 8.949705123901367, "learning_rate": 1.1860905051211745e-05, "loss": 1.3239, "step": 8521 }, { "epoch": 1.37, "grad_norm": 10.473997116088867, "learning_rate": 1.1859197390220413e-05, "loss": 1.0139, "step": 8522 }, { "epoch": 1.37, "grad_norm": 6.944491386413574, "learning_rate": 1.1857489673069926e-05, "loss": 0.8478, "step": 8523 }, { "epoch": 1.37, "grad_norm": 9.544146537780762, "learning_rate": 1.1855781899811867e-05, "loss": 1.081, "step": 8524 }, { "epoch": 1.37, "grad_norm": 9.256446838378906, "learning_rate": 1.1854074070497821e-05, "loss": 1.0064, "step": 8525 }, { "epoch": 1.37, "grad_norm": 9.745307922363281, "learning_rate": 1.1852366185179372e-05, "loss": 1.2009, "step": 8526 }, { "epoch": 1.37, "grad_norm": 7.748474597930908, "learning_rate": 1.1850658243908111e-05, "loss": 0.8706, "step": 8527 }, { "epoch": 1.37, "grad_norm": 8.712176322937012, "learning_rate": 1.1848950246735625e-05, "loss": 0.9059, "step": 8528 }, { "epoch": 1.37, "grad_norm": 10.023297309875488, "learning_rate": 1.1847242193713513e-05, "loss": 1.0886, "step": 8529 }, { "epoch": 1.37, "grad_norm": 9.109620094299316, "learning_rate": 1.1845534084893357e-05, "loss": 1.0653, "step": 8530 }, { "epoch": 1.37, "grad_norm": 8.717031478881836, "learning_rate": 1.1843825920326764e-05, "loss": 1.1307, "step": 8531 }, { "epoch": 1.37, "grad_norm": 9.991861343383789, "learning_rate": 1.1842117700065327e-05, "loss": 1.0316, "step": 8532 }, { "epoch": 1.37, "grad_norm": 8.300185203552246, "learning_rate": 1.1840409424160646e-05, "loss": 0.9439, "step": 8533 }, { "epoch": 1.37, "grad_norm": 10.64600944519043, "learning_rate": 1.1838701092664317e-05, "loss": 1.3072, "step": 8534 }, { "epoch": 1.37, "grad_norm": 9.662247657775879, "learning_rate": 1.1836992705627946e-05, "loss": 1.2673, "step": 8535 }, { "epoch": 1.37, "grad_norm": 7.833940505981445, "learning_rate": 1.1835284263103136e-05, "loss": 0.7971, "step": 8536 }, { "epoch": 1.37, "grad_norm": 9.084775924682617, "learning_rate": 1.183357576514149e-05, "loss": 1.2962, "step": 8537 }, { "epoch": 1.37, "grad_norm": 9.17919921875, "learning_rate": 1.1831867211794616e-05, "loss": 1.0201, "step": 8538 }, { "epoch": 1.37, "grad_norm": 10.551295280456543, "learning_rate": 1.1830158603114127e-05, "loss": 1.0506, "step": 8539 }, { "epoch": 1.37, "grad_norm": 11.931495666503906, "learning_rate": 1.182844993915163e-05, "loss": 1.1734, "step": 8540 }, { "epoch": 1.37, "grad_norm": 11.024288177490234, "learning_rate": 1.1826741219958736e-05, "loss": 1.2197, "step": 8541 }, { "epoch": 1.38, "grad_norm": 9.908084869384766, "learning_rate": 1.1825032445587061e-05, "loss": 1.0639, "step": 8542 }, { "epoch": 1.38, "grad_norm": 8.381449699401855, "learning_rate": 1.1823323616088219e-05, "loss": 1.0618, "step": 8543 }, { "epoch": 1.38, "grad_norm": 8.957762718200684, "learning_rate": 1.182161473151383e-05, "loss": 1.2059, "step": 8544 }, { "epoch": 1.38, "grad_norm": 7.492696762084961, "learning_rate": 1.181990579191551e-05, "loss": 1.1346, "step": 8545 }, { "epoch": 1.38, "grad_norm": 8.688375473022461, "learning_rate": 1.1818196797344879e-05, "loss": 0.8535, "step": 8546 }, { "epoch": 1.38, "grad_norm": 8.271204948425293, "learning_rate": 1.181648774785356e-05, "loss": 1.0646, "step": 8547 }, { "epoch": 1.38, "grad_norm": 8.632378578186035, "learning_rate": 1.1814778643493179e-05, "loss": 1.096, "step": 8548 }, { "epoch": 1.38, "grad_norm": 8.095931053161621, "learning_rate": 1.1813069484315357e-05, "loss": 1.0105, "step": 8549 }, { "epoch": 1.38, "grad_norm": 8.90494441986084, "learning_rate": 1.1811360270371723e-05, "loss": 0.9989, "step": 8550 }, { "epoch": 1.38, "grad_norm": 7.761898040771484, "learning_rate": 1.1809651001713906e-05, "loss": 0.8328, "step": 8551 }, { "epoch": 1.38, "grad_norm": 9.687747955322266, "learning_rate": 1.1807941678393542e-05, "loss": 1.0477, "step": 8552 }, { "epoch": 1.38, "grad_norm": 8.61856460571289, "learning_rate": 1.1806232300462253e-05, "loss": 0.8192, "step": 8553 }, { "epoch": 1.38, "grad_norm": 8.995780944824219, "learning_rate": 1.1804522867971677e-05, "loss": 0.7894, "step": 8554 }, { "epoch": 1.38, "grad_norm": 9.781829833984375, "learning_rate": 1.1802813380973453e-05, "loss": 1.1264, "step": 8555 }, { "epoch": 1.38, "grad_norm": 10.56016731262207, "learning_rate": 1.1801103839519214e-05, "loss": 1.2559, "step": 8556 }, { "epoch": 1.38, "grad_norm": 9.627685546875, "learning_rate": 1.1799394243660597e-05, "loss": 1.0806, "step": 8557 }, { "epoch": 1.38, "grad_norm": 11.907939910888672, "learning_rate": 1.1797684593449245e-05, "loss": 1.3029, "step": 8558 }, { "epoch": 1.38, "grad_norm": 10.342926979064941, "learning_rate": 1.1795974888936801e-05, "loss": 1.096, "step": 8559 }, { "epoch": 1.38, "grad_norm": 7.715500831604004, "learning_rate": 1.1794265130174904e-05, "loss": 1.0568, "step": 8560 }, { "epoch": 1.38, "grad_norm": 10.119330406188965, "learning_rate": 1.1792555317215206e-05, "loss": 1.2635, "step": 8561 }, { "epoch": 1.38, "grad_norm": 8.368582725524902, "learning_rate": 1.1790845450109348e-05, "loss": 1.0975, "step": 8562 }, { "epoch": 1.38, "grad_norm": 10.560912132263184, "learning_rate": 1.178913552890898e-05, "loss": 1.1338, "step": 8563 }, { "epoch": 1.38, "grad_norm": 10.776888847351074, "learning_rate": 1.1787425553665756e-05, "loss": 1.1476, "step": 8564 }, { "epoch": 1.38, "grad_norm": 10.095072746276855, "learning_rate": 1.178571552443132e-05, "loss": 1.0821, "step": 8565 }, { "epoch": 1.38, "grad_norm": 11.03558349609375, "learning_rate": 1.1784005441257333e-05, "loss": 1.1937, "step": 8566 }, { "epoch": 1.38, "grad_norm": 9.424412727355957, "learning_rate": 1.1782295304195446e-05, "loss": 0.9945, "step": 8567 }, { "epoch": 1.38, "grad_norm": 8.72793197631836, "learning_rate": 1.1780585113297315e-05, "loss": 1.1804, "step": 8568 }, { "epoch": 1.38, "grad_norm": 9.724885940551758, "learning_rate": 1.1778874868614602e-05, "loss": 1.1948, "step": 8569 }, { "epoch": 1.38, "grad_norm": 10.540080070495605, "learning_rate": 1.177716457019896e-05, "loss": 1.4454, "step": 8570 }, { "epoch": 1.38, "grad_norm": 9.54478645324707, "learning_rate": 1.177545421810206e-05, "loss": 1.0898, "step": 8571 }, { "epoch": 1.38, "grad_norm": 10.68580150604248, "learning_rate": 1.1773743812375557e-05, "loss": 0.9297, "step": 8572 }, { "epoch": 1.38, "grad_norm": 8.298565864562988, "learning_rate": 1.1772033353071119e-05, "loss": 0.9488, "step": 8573 }, { "epoch": 1.38, "grad_norm": 9.11838150024414, "learning_rate": 1.1770322840240414e-05, "loss": 0.939, "step": 8574 }, { "epoch": 1.38, "grad_norm": 8.5731782913208, "learning_rate": 1.176861227393511e-05, "loss": 0.9339, "step": 8575 }, { "epoch": 1.38, "grad_norm": 9.789067268371582, "learning_rate": 1.1766901654206871e-05, "loss": 1.1071, "step": 8576 }, { "epoch": 1.38, "grad_norm": 9.164030075073242, "learning_rate": 1.1765190981107373e-05, "loss": 0.9135, "step": 8577 }, { "epoch": 1.38, "grad_norm": 9.271501541137695, "learning_rate": 1.1763480254688286e-05, "loss": 0.8973, "step": 8578 }, { "epoch": 1.38, "grad_norm": 8.717572212219238, "learning_rate": 1.1761769475001288e-05, "loss": 1.1849, "step": 8579 }, { "epoch": 1.38, "grad_norm": 9.196983337402344, "learning_rate": 1.1760058642098053e-05, "loss": 1.0812, "step": 8580 }, { "epoch": 1.38, "grad_norm": 8.695676803588867, "learning_rate": 1.1758347756030257e-05, "loss": 1.0569, "step": 8581 }, { "epoch": 1.38, "grad_norm": 10.887636184692383, "learning_rate": 1.1756636816849584e-05, "loss": 1.1062, "step": 8582 }, { "epoch": 1.38, "grad_norm": 8.645452499389648, "learning_rate": 1.1754925824607713e-05, "loss": 0.7922, "step": 8583 }, { "epoch": 1.38, "grad_norm": 10.522032737731934, "learning_rate": 1.1753214779356324e-05, "loss": 0.9485, "step": 8584 }, { "epoch": 1.38, "grad_norm": 7.650611877441406, "learning_rate": 1.17515036811471e-05, "loss": 0.9686, "step": 8585 }, { "epoch": 1.38, "grad_norm": 12.640496253967285, "learning_rate": 1.174979253003173e-05, "loss": 0.9616, "step": 8586 }, { "epoch": 1.38, "grad_norm": 11.669798851013184, "learning_rate": 1.1748081326061904e-05, "loss": 1.2028, "step": 8587 }, { "epoch": 1.38, "grad_norm": 10.230435371398926, "learning_rate": 1.1746370069289304e-05, "loss": 1.1436, "step": 8588 }, { "epoch": 1.38, "grad_norm": 8.650699615478516, "learning_rate": 1.1744658759765624e-05, "loss": 1.2321, "step": 8589 }, { "epoch": 1.38, "grad_norm": 10.11892318725586, "learning_rate": 1.1742947397542553e-05, "loss": 1.0772, "step": 8590 }, { "epoch": 1.38, "grad_norm": 8.5770263671875, "learning_rate": 1.1741235982671792e-05, "loss": 1.0273, "step": 8591 }, { "epoch": 1.38, "grad_norm": 9.570523262023926, "learning_rate": 1.173952451520503e-05, "loss": 0.869, "step": 8592 }, { "epoch": 1.38, "grad_norm": 9.692338943481445, "learning_rate": 1.1737812995193964e-05, "loss": 0.9872, "step": 8593 }, { "epoch": 1.38, "grad_norm": 9.158181190490723, "learning_rate": 1.1736101422690293e-05, "loss": 0.9093, "step": 8594 }, { "epoch": 1.38, "grad_norm": 10.704499244689941, "learning_rate": 1.1734389797745722e-05, "loss": 1.1057, "step": 8595 }, { "epoch": 1.38, "grad_norm": 9.043095588684082, "learning_rate": 1.1732678120411944e-05, "loss": 1.1221, "step": 8596 }, { "epoch": 1.38, "grad_norm": 8.701105117797852, "learning_rate": 1.173096639074067e-05, "loss": 0.9046, "step": 8597 }, { "epoch": 1.38, "grad_norm": 9.95256233215332, "learning_rate": 1.1729254608783598e-05, "loss": 1.0602, "step": 8598 }, { "epoch": 1.38, "grad_norm": 10.3738374710083, "learning_rate": 1.1727542774592442e-05, "loss": 0.9303, "step": 8599 }, { "epoch": 1.38, "grad_norm": 9.184557914733887, "learning_rate": 1.1725830888218899e-05, "loss": 1.0195, "step": 8600 }, { "epoch": 1.38, "grad_norm": 9.96288013458252, "learning_rate": 1.1724118949714687e-05, "loss": 0.9613, "step": 8601 }, { "epoch": 1.38, "grad_norm": 9.262516975402832, "learning_rate": 1.1722406959131516e-05, "loss": 0.831, "step": 8602 }, { "epoch": 1.38, "grad_norm": 10.745186805725098, "learning_rate": 1.1720694916521099e-05, "loss": 1.1478, "step": 8603 }, { "epoch": 1.39, "grad_norm": 7.834399700164795, "learning_rate": 1.1718982821935146e-05, "loss": 1.0558, "step": 8604 }, { "epoch": 1.39, "grad_norm": 7.477068901062012, "learning_rate": 1.1717270675425377e-05, "loss": 1.0619, "step": 8605 }, { "epoch": 1.39, "grad_norm": 10.531725883483887, "learning_rate": 1.1715558477043506e-05, "loss": 1.3529, "step": 8606 }, { "epoch": 1.39, "grad_norm": 10.497946739196777, "learning_rate": 1.1713846226841256e-05, "loss": 0.851, "step": 8607 }, { "epoch": 1.39, "grad_norm": 10.00692081451416, "learning_rate": 1.1712133924870343e-05, "loss": 0.968, "step": 8608 }, { "epoch": 1.39, "grad_norm": 11.06290340423584, "learning_rate": 1.1710421571182493e-05, "loss": 1.1523, "step": 8609 }, { "epoch": 1.39, "grad_norm": 7.591170787811279, "learning_rate": 1.1708709165829427e-05, "loss": 1.0208, "step": 8610 }, { "epoch": 1.39, "grad_norm": 9.842040061950684, "learning_rate": 1.1706996708862868e-05, "loss": 1.1674, "step": 8611 }, { "epoch": 1.39, "grad_norm": 8.777356147766113, "learning_rate": 1.1705284200334548e-05, "loss": 1.0998, "step": 8612 }, { "epoch": 1.39, "grad_norm": 9.63936710357666, "learning_rate": 1.1703571640296193e-05, "loss": 0.9143, "step": 8613 }, { "epoch": 1.39, "grad_norm": 8.380809783935547, "learning_rate": 1.1701859028799532e-05, "loss": 0.9657, "step": 8614 }, { "epoch": 1.39, "grad_norm": 9.813278198242188, "learning_rate": 1.1700146365896296e-05, "loss": 1.0385, "step": 8615 }, { "epoch": 1.39, "grad_norm": 9.289072036743164, "learning_rate": 1.1698433651638221e-05, "loss": 1.031, "step": 8616 }, { "epoch": 1.39, "grad_norm": 10.7036771774292, "learning_rate": 1.1696720886077038e-05, "loss": 1.0303, "step": 8617 }, { "epoch": 1.39, "grad_norm": 8.78547191619873, "learning_rate": 1.1695008069264482e-05, "loss": 1.043, "step": 8618 }, { "epoch": 1.39, "grad_norm": 11.460537910461426, "learning_rate": 1.1693295201252297e-05, "loss": 0.9696, "step": 8619 }, { "epoch": 1.39, "grad_norm": 10.128870964050293, "learning_rate": 1.1691582282092215e-05, "loss": 1.0135, "step": 8620 }, { "epoch": 1.39, "grad_norm": 9.97949504852295, "learning_rate": 1.1689869311835979e-05, "loss": 1.2654, "step": 8621 }, { "epoch": 1.39, "grad_norm": 8.822039604187012, "learning_rate": 1.1688156290535336e-05, "loss": 1.3538, "step": 8622 }, { "epoch": 1.39, "grad_norm": 10.38455581665039, "learning_rate": 1.1686443218242023e-05, "loss": 1.2897, "step": 8623 }, { "epoch": 1.39, "grad_norm": 10.053686141967773, "learning_rate": 1.1684730095007788e-05, "loss": 1.2852, "step": 8624 }, { "epoch": 1.39, "grad_norm": 9.029911994934082, "learning_rate": 1.1683016920884376e-05, "loss": 1.1536, "step": 8625 }, { "epoch": 1.39, "grad_norm": 7.250888824462891, "learning_rate": 1.1681303695923541e-05, "loss": 0.8829, "step": 8626 }, { "epoch": 1.39, "grad_norm": 10.138742446899414, "learning_rate": 1.1679590420177028e-05, "loss": 1.0729, "step": 8627 }, { "epoch": 1.39, "grad_norm": 10.144735336303711, "learning_rate": 1.1677877093696585e-05, "loss": 1.0801, "step": 8628 }, { "epoch": 1.39, "grad_norm": 9.205601692199707, "learning_rate": 1.1676163716533973e-05, "loss": 1.0645, "step": 8629 }, { "epoch": 1.39, "grad_norm": 9.191351890563965, "learning_rate": 1.1674450288740941e-05, "loss": 1.136, "step": 8630 }, { "epoch": 1.39, "grad_norm": 10.361217498779297, "learning_rate": 1.167273681036925e-05, "loss": 0.9646, "step": 8631 }, { "epoch": 1.39, "grad_norm": 8.2035493850708, "learning_rate": 1.1671023281470651e-05, "loss": 0.9471, "step": 8632 }, { "epoch": 1.39, "grad_norm": 10.133186340332031, "learning_rate": 1.1669309702096907e-05, "loss": 1.1233, "step": 8633 }, { "epoch": 1.39, "grad_norm": 11.86074161529541, "learning_rate": 1.1667596072299782e-05, "loss": 1.2362, "step": 8634 }, { "epoch": 1.39, "grad_norm": 8.74050521850586, "learning_rate": 1.1665882392131032e-05, "loss": 1.0013, "step": 8635 }, { "epoch": 1.39, "grad_norm": 9.062507629394531, "learning_rate": 1.1664168661642421e-05, "loss": 1.1455, "step": 8636 }, { "epoch": 1.39, "grad_norm": 8.888609886169434, "learning_rate": 1.1662454880885716e-05, "loss": 0.9026, "step": 8637 }, { "epoch": 1.39, "grad_norm": 7.955594539642334, "learning_rate": 1.1660741049912687e-05, "loss": 0.9091, "step": 8638 }, { "epoch": 1.39, "grad_norm": 9.075098991394043, "learning_rate": 1.1659027168775096e-05, "loss": 0.8689, "step": 8639 }, { "epoch": 1.39, "grad_norm": 10.265660285949707, "learning_rate": 1.1657313237524716e-05, "loss": 0.8742, "step": 8640 }, { "epoch": 1.39, "grad_norm": 8.31779956817627, "learning_rate": 1.1655599256213317e-05, "loss": 1.0347, "step": 8641 }, { "epoch": 1.39, "grad_norm": 9.10093879699707, "learning_rate": 1.1653885224892676e-05, "loss": 1.1087, "step": 8642 }, { "epoch": 1.39, "grad_norm": 10.914402961730957, "learning_rate": 1.165217114361456e-05, "loss": 1.119, "step": 8643 }, { "epoch": 1.39, "grad_norm": 7.7645039558410645, "learning_rate": 1.165045701243075e-05, "loss": 0.9624, "step": 8644 }, { "epoch": 1.39, "grad_norm": 9.693778038024902, "learning_rate": 1.1648742831393023e-05, "loss": 1.1812, "step": 8645 }, { "epoch": 1.39, "grad_norm": 10.25252628326416, "learning_rate": 1.1647028600553157e-05, "loss": 1.2839, "step": 8646 }, { "epoch": 1.39, "grad_norm": 8.979985237121582, "learning_rate": 1.164531431996293e-05, "loss": 0.8553, "step": 8647 }, { "epoch": 1.39, "grad_norm": 8.875469207763672, "learning_rate": 1.1643599989674126e-05, "loss": 1.204, "step": 8648 }, { "epoch": 1.39, "grad_norm": 9.603376388549805, "learning_rate": 1.1641885609738527e-05, "loss": 0.8637, "step": 8649 }, { "epoch": 1.39, "grad_norm": 9.255916595458984, "learning_rate": 1.1640171180207921e-05, "loss": 0.9632, "step": 8650 }, { "epoch": 1.39, "grad_norm": 10.970808982849121, "learning_rate": 1.163845670113409e-05, "loss": 1.066, "step": 8651 }, { "epoch": 1.39, "grad_norm": 10.758720397949219, "learning_rate": 1.1636742172568824e-05, "loss": 1.0577, "step": 8652 }, { "epoch": 1.39, "grad_norm": 8.917692184448242, "learning_rate": 1.1635027594563915e-05, "loss": 1.1401, "step": 8653 }, { "epoch": 1.39, "grad_norm": 9.860697746276855, "learning_rate": 1.1633312967171149e-05, "loss": 1.0921, "step": 8654 }, { "epoch": 1.39, "grad_norm": 9.50571346282959, "learning_rate": 1.1631598290442319e-05, "loss": 1.0758, "step": 8655 }, { "epoch": 1.39, "grad_norm": 9.447583198547363, "learning_rate": 1.162988356442922e-05, "loss": 1.1659, "step": 8656 }, { "epoch": 1.39, "grad_norm": 10.496480941772461, "learning_rate": 1.1628168789183649e-05, "loss": 1.1143, "step": 8657 }, { "epoch": 1.39, "grad_norm": 8.112518310546875, "learning_rate": 1.1626453964757402e-05, "loss": 0.7368, "step": 8658 }, { "epoch": 1.39, "grad_norm": 11.410563468933105, "learning_rate": 1.1624739091202273e-05, "loss": 1.2466, "step": 8659 }, { "epoch": 1.39, "grad_norm": 11.064168930053711, "learning_rate": 1.1623024168570066e-05, "loss": 1.2689, "step": 8660 }, { "epoch": 1.39, "grad_norm": 9.564220428466797, "learning_rate": 1.1621309196912578e-05, "loss": 0.8093, "step": 8661 }, { "epoch": 1.39, "grad_norm": 10.719756126403809, "learning_rate": 1.1619594176281619e-05, "loss": 1.3273, "step": 8662 }, { "epoch": 1.39, "grad_norm": 9.022262573242188, "learning_rate": 1.1617879106728984e-05, "loss": 1.1406, "step": 8663 }, { "epoch": 1.39, "grad_norm": 10.239145278930664, "learning_rate": 1.1616163988306485e-05, "loss": 1.126, "step": 8664 }, { "epoch": 1.39, "grad_norm": 10.533602714538574, "learning_rate": 1.161444882106593e-05, "loss": 1.0784, "step": 8665 }, { "epoch": 1.4, "grad_norm": 9.060189247131348, "learning_rate": 1.1612733605059126e-05, "loss": 0.9576, "step": 8666 }, { "epoch": 1.4, "grad_norm": 9.488410949707031, "learning_rate": 1.161101834033788e-05, "loss": 1.2397, "step": 8667 }, { "epoch": 1.4, "grad_norm": 10.854926109313965, "learning_rate": 1.1609303026954002e-05, "loss": 1.4662, "step": 8668 }, { "epoch": 1.4, "grad_norm": 8.438189506530762, "learning_rate": 1.1607587664959312e-05, "loss": 1.1574, "step": 8669 }, { "epoch": 1.4, "grad_norm": 10.731496810913086, "learning_rate": 1.1605872254405621e-05, "loss": 1.3255, "step": 8670 }, { "epoch": 1.4, "grad_norm": 8.31380558013916, "learning_rate": 1.1604156795344741e-05, "loss": 1.0028, "step": 8671 }, { "epoch": 1.4, "grad_norm": 9.480840682983398, "learning_rate": 1.1602441287828497e-05, "loss": 1.1578, "step": 8672 }, { "epoch": 1.4, "grad_norm": 8.498090744018555, "learning_rate": 1.1600725731908704e-05, "loss": 1.1395, "step": 8673 }, { "epoch": 1.4, "grad_norm": 9.585493087768555, "learning_rate": 1.1599010127637183e-05, "loss": 1.1818, "step": 8674 }, { "epoch": 1.4, "grad_norm": 8.01059627532959, "learning_rate": 1.1597294475065752e-05, "loss": 0.9444, "step": 8675 }, { "epoch": 1.4, "grad_norm": 9.626176834106445, "learning_rate": 1.1595578774246238e-05, "loss": 0.9949, "step": 8676 }, { "epoch": 1.4, "grad_norm": 7.418781757354736, "learning_rate": 1.159386302523047e-05, "loss": 1.0362, "step": 8677 }, { "epoch": 1.4, "grad_norm": 8.28995418548584, "learning_rate": 1.1592147228070262e-05, "loss": 1.0468, "step": 8678 }, { "epoch": 1.4, "grad_norm": 9.783069610595703, "learning_rate": 1.1590431382817453e-05, "loss": 1.3424, "step": 8679 }, { "epoch": 1.4, "grad_norm": 9.521727561950684, "learning_rate": 1.1588715489523868e-05, "loss": 1.2017, "step": 8680 }, { "epoch": 1.4, "grad_norm": 7.3139753341674805, "learning_rate": 1.1586999548241338e-05, "loss": 0.9125, "step": 8681 }, { "epoch": 1.4, "grad_norm": 8.225567817687988, "learning_rate": 1.158528355902169e-05, "loss": 0.978, "step": 8682 }, { "epoch": 1.4, "grad_norm": 9.615612983703613, "learning_rate": 1.1583567521916764e-05, "loss": 1.0365, "step": 8683 }, { "epoch": 1.4, "grad_norm": 9.748412132263184, "learning_rate": 1.1581851436978391e-05, "loss": 1.2983, "step": 8684 }, { "epoch": 1.4, "grad_norm": 8.186086654663086, "learning_rate": 1.158013530425841e-05, "loss": 1.0107, "step": 8685 }, { "epoch": 1.4, "grad_norm": 9.479317665100098, "learning_rate": 1.1578419123808659e-05, "loss": 1.2147, "step": 8686 }, { "epoch": 1.4, "grad_norm": 9.752513885498047, "learning_rate": 1.1576702895680974e-05, "loss": 1.0226, "step": 8687 }, { "epoch": 1.4, "grad_norm": 10.557892799377441, "learning_rate": 1.1574986619927197e-05, "loss": 0.982, "step": 8688 }, { "epoch": 1.4, "grad_norm": 8.557389259338379, "learning_rate": 1.1573270296599172e-05, "loss": 1.3673, "step": 8689 }, { "epoch": 1.4, "grad_norm": 9.592784881591797, "learning_rate": 1.157155392574874e-05, "loss": 0.9693, "step": 8690 }, { "epoch": 1.4, "grad_norm": 9.595076560974121, "learning_rate": 1.1569837507427745e-05, "loss": 0.9884, "step": 8691 }, { "epoch": 1.4, "grad_norm": 11.185836791992188, "learning_rate": 1.1568121041688039e-05, "loss": 0.981, "step": 8692 }, { "epoch": 1.4, "grad_norm": 9.908195495605469, "learning_rate": 1.156640452858146e-05, "loss": 1.0743, "step": 8693 }, { "epoch": 1.4, "grad_norm": 9.914475440979004, "learning_rate": 1.1564687968159869e-05, "loss": 0.8982, "step": 8694 }, { "epoch": 1.4, "grad_norm": 9.835756301879883, "learning_rate": 1.1562971360475106e-05, "loss": 0.8303, "step": 8695 }, { "epoch": 1.4, "grad_norm": 9.871074676513672, "learning_rate": 1.1561254705579031e-05, "loss": 0.8759, "step": 8696 }, { "epoch": 1.4, "grad_norm": 8.129189491271973, "learning_rate": 1.1559538003523493e-05, "loss": 0.9578, "step": 8697 }, { "epoch": 1.4, "grad_norm": 7.568936824798584, "learning_rate": 1.1557821254360349e-05, "loss": 1.0695, "step": 8698 }, { "epoch": 1.4, "grad_norm": 9.71351146697998, "learning_rate": 1.1556104458141455e-05, "loss": 1.0883, "step": 8699 }, { "epoch": 1.4, "grad_norm": 12.071471214294434, "learning_rate": 1.1554387614918666e-05, "loss": 1.2799, "step": 8700 }, { "epoch": 1.4, "grad_norm": 8.58426570892334, "learning_rate": 1.1552670724743846e-05, "loss": 1.0374, "step": 8701 }, { "epoch": 1.4, "grad_norm": 9.163373947143555, "learning_rate": 1.155095378766885e-05, "loss": 0.9625, "step": 8702 }, { "epoch": 1.4, "grad_norm": 9.581388473510742, "learning_rate": 1.1549236803745545e-05, "loss": 1.1959, "step": 8703 }, { "epoch": 1.4, "grad_norm": 8.307679176330566, "learning_rate": 1.1547519773025794e-05, "loss": 0.9403, "step": 8704 }, { "epoch": 1.4, "grad_norm": 9.992786407470703, "learning_rate": 1.1545802695561458e-05, "loss": 1.146, "step": 8705 }, { "epoch": 1.4, "grad_norm": 10.231366157531738, "learning_rate": 1.1544085571404409e-05, "loss": 1.1689, "step": 8706 }, { "epoch": 1.4, "grad_norm": 8.473706245422363, "learning_rate": 1.1542368400606508e-05, "loss": 1.025, "step": 8707 }, { "epoch": 1.4, "grad_norm": 10.069857597351074, "learning_rate": 1.1540651183219629e-05, "loss": 1.1219, "step": 8708 }, { "epoch": 1.4, "grad_norm": 9.8873291015625, "learning_rate": 1.1538933919295641e-05, "loss": 0.9127, "step": 8709 }, { "epoch": 1.4, "grad_norm": 12.049819946289062, "learning_rate": 1.1537216608886414e-05, "loss": 0.9541, "step": 8710 }, { "epoch": 1.4, "grad_norm": 8.184112548828125, "learning_rate": 1.1535499252043824e-05, "loss": 1.1306, "step": 8711 }, { "epoch": 1.4, "grad_norm": 10.42968463897705, "learning_rate": 1.1533781848819747e-05, "loss": 1.1897, "step": 8712 }, { "epoch": 1.4, "grad_norm": 9.078106880187988, "learning_rate": 1.1532064399266055e-05, "loss": 0.9284, "step": 8713 }, { "epoch": 1.4, "grad_norm": 9.320977210998535, "learning_rate": 1.1530346903434628e-05, "loss": 0.9097, "step": 8714 }, { "epoch": 1.4, "grad_norm": 9.060495376586914, "learning_rate": 1.1528629361377343e-05, "loss": 1.0671, "step": 8715 }, { "epoch": 1.4, "grad_norm": 7.251411437988281, "learning_rate": 1.1526911773146085e-05, "loss": 0.7991, "step": 8716 }, { "epoch": 1.4, "grad_norm": 10.21705150604248, "learning_rate": 1.152519413879273e-05, "loss": 0.8218, "step": 8717 }, { "epoch": 1.4, "grad_norm": 10.939642906188965, "learning_rate": 1.1523476458369162e-05, "loss": 1.0811, "step": 8718 }, { "epoch": 1.4, "grad_norm": 10.85505199432373, "learning_rate": 1.152175873192727e-05, "loss": 1.1849, "step": 8719 }, { "epoch": 1.4, "grad_norm": 7.589674949645996, "learning_rate": 1.1520040959518937e-05, "loss": 0.8113, "step": 8720 }, { "epoch": 1.4, "grad_norm": 8.76654052734375, "learning_rate": 1.1518323141196048e-05, "loss": 1.1139, "step": 8721 }, { "epoch": 1.4, "grad_norm": 7.235544204711914, "learning_rate": 1.1516605277010493e-05, "loss": 1.0282, "step": 8722 }, { "epoch": 1.4, "grad_norm": 8.517206192016602, "learning_rate": 1.1514887367014163e-05, "loss": 1.102, "step": 8723 }, { "epoch": 1.4, "grad_norm": 8.193649291992188, "learning_rate": 1.151316941125895e-05, "loss": 0.9453, "step": 8724 }, { "epoch": 1.4, "grad_norm": 8.42213249206543, "learning_rate": 1.1511451409796748e-05, "loss": 1.064, "step": 8725 }, { "epoch": 1.4, "grad_norm": 9.525140762329102, "learning_rate": 1.1509733362679445e-05, "loss": 0.9694, "step": 8726 }, { "epoch": 1.4, "grad_norm": 9.55888557434082, "learning_rate": 1.1508015269958943e-05, "loss": 1.1241, "step": 8727 }, { "epoch": 1.41, "grad_norm": 9.908879280090332, "learning_rate": 1.1506297131687139e-05, "loss": 1.1529, "step": 8728 }, { "epoch": 1.41, "grad_norm": 8.641965866088867, "learning_rate": 1.1504578947915923e-05, "loss": 1.0255, "step": 8729 }, { "epoch": 1.41, "grad_norm": 9.925009727478027, "learning_rate": 1.1502860718697203e-05, "loss": 1.0478, "step": 8730 }, { "epoch": 1.41, "grad_norm": 8.384849548339844, "learning_rate": 1.150114244408288e-05, "loss": 1.0807, "step": 8731 }, { "epoch": 1.41, "grad_norm": 8.633588790893555, "learning_rate": 1.1499424124124851e-05, "loss": 1.0682, "step": 8732 }, { "epoch": 1.41, "grad_norm": 10.143594741821289, "learning_rate": 1.1497705758875024e-05, "loss": 1.1447, "step": 8733 }, { "epoch": 1.41, "grad_norm": 8.826456069946289, "learning_rate": 1.1495987348385303e-05, "loss": 1.0996, "step": 8734 }, { "epoch": 1.41, "grad_norm": 9.409987449645996, "learning_rate": 1.1494268892707595e-05, "loss": 1.1585, "step": 8735 }, { "epoch": 1.41, "grad_norm": 11.06940746307373, "learning_rate": 1.1492550391893809e-05, "loss": 1.0213, "step": 8736 }, { "epoch": 1.41, "grad_norm": 8.965081214904785, "learning_rate": 1.1490831845995849e-05, "loss": 1.1457, "step": 8737 }, { "epoch": 1.41, "grad_norm": 7.512459754943848, "learning_rate": 1.148911325506563e-05, "loss": 0.8253, "step": 8738 }, { "epoch": 1.41, "grad_norm": 11.120688438415527, "learning_rate": 1.1487394619155065e-05, "loss": 0.9422, "step": 8739 }, { "epoch": 1.41, "grad_norm": 9.178497314453125, "learning_rate": 1.1485675938316066e-05, "loss": 0.9431, "step": 8740 }, { "epoch": 1.41, "grad_norm": 9.303777694702148, "learning_rate": 1.1483957212600548e-05, "loss": 1.0257, "step": 8741 }, { "epoch": 1.41, "grad_norm": 10.362102508544922, "learning_rate": 1.148223844206042e-05, "loss": 1.3868, "step": 8742 }, { "epoch": 1.41, "grad_norm": 11.556050300598145, "learning_rate": 1.1480519626747613e-05, "loss": 1.0591, "step": 8743 }, { "epoch": 1.41, "grad_norm": 10.963555335998535, "learning_rate": 1.1478800766714035e-05, "loss": 1.1283, "step": 8744 }, { "epoch": 1.41, "grad_norm": 8.784693717956543, "learning_rate": 1.147708186201161e-05, "loss": 1.0602, "step": 8745 }, { "epoch": 1.41, "grad_norm": 9.868545532226562, "learning_rate": 1.1475362912692261e-05, "loss": 1.2619, "step": 8746 }, { "epoch": 1.41, "grad_norm": 9.062516212463379, "learning_rate": 1.1473643918807906e-05, "loss": 0.8362, "step": 8747 }, { "epoch": 1.41, "grad_norm": 9.03751277923584, "learning_rate": 1.1471924880410477e-05, "loss": 1.0442, "step": 8748 }, { "epoch": 1.41, "grad_norm": 8.337180137634277, "learning_rate": 1.147020579755189e-05, "loss": 0.9765, "step": 8749 }, { "epoch": 1.41, "grad_norm": 7.847895622253418, "learning_rate": 1.1468486670284075e-05, "loss": 1.1189, "step": 8750 }, { "epoch": 1.41, "grad_norm": 10.651226043701172, "learning_rate": 1.1466767498658967e-05, "loss": 1.3261, "step": 8751 }, { "epoch": 1.41, "grad_norm": 8.354716300964355, "learning_rate": 1.1465048282728486e-05, "loss": 0.9821, "step": 8752 }, { "epoch": 1.41, "grad_norm": 10.220036506652832, "learning_rate": 1.1463329022544566e-05, "loss": 1.0231, "step": 8753 }, { "epoch": 1.41, "grad_norm": 10.14195442199707, "learning_rate": 1.1461609718159143e-05, "loss": 1.1539, "step": 8754 }, { "epoch": 1.41, "grad_norm": 10.42103099822998, "learning_rate": 1.145989036962415e-05, "loss": 1.203, "step": 8755 }, { "epoch": 1.41, "grad_norm": 7.903524398803711, "learning_rate": 1.1458170976991514e-05, "loss": 0.9977, "step": 8756 }, { "epoch": 1.41, "grad_norm": 10.314189910888672, "learning_rate": 1.145645154031318e-05, "loss": 0.9846, "step": 8757 }, { "epoch": 1.41, "grad_norm": 9.349506378173828, "learning_rate": 1.145473205964108e-05, "loss": 1.034, "step": 8758 }, { "epoch": 1.41, "grad_norm": 8.462251663208008, "learning_rate": 1.145301253502716e-05, "loss": 1.224, "step": 8759 }, { "epoch": 1.41, "grad_norm": 10.422578811645508, "learning_rate": 1.1451292966523351e-05, "loss": 1.3863, "step": 8760 }, { "epoch": 1.41, "grad_norm": 9.531153678894043, "learning_rate": 1.14495733541816e-05, "loss": 0.9316, "step": 8761 }, { "epoch": 1.41, "grad_norm": 10.96926498413086, "learning_rate": 1.1447853698053847e-05, "loss": 1.0312, "step": 8762 }, { "epoch": 1.41, "grad_norm": 10.975422859191895, "learning_rate": 1.1446133998192039e-05, "loss": 1.2435, "step": 8763 }, { "epoch": 1.41, "grad_norm": 9.743330001831055, "learning_rate": 1.144441425464812e-05, "loss": 0.8511, "step": 8764 }, { "epoch": 1.41, "grad_norm": 11.057815551757812, "learning_rate": 1.1442694467474038e-05, "loss": 1.0655, "step": 8765 }, { "epoch": 1.41, "grad_norm": 11.986302375793457, "learning_rate": 1.144097463672174e-05, "loss": 1.0837, "step": 8766 }, { "epoch": 1.41, "grad_norm": 9.745168685913086, "learning_rate": 1.1439254762443177e-05, "loss": 1.058, "step": 8767 }, { "epoch": 1.41, "grad_norm": 10.413104057312012, "learning_rate": 1.1437534844690296e-05, "loss": 1.0036, "step": 8768 }, { "epoch": 1.41, "grad_norm": 8.172361373901367, "learning_rate": 1.143581488351505e-05, "loss": 0.88, "step": 8769 }, { "epoch": 1.41, "grad_norm": 9.32629680633545, "learning_rate": 1.1434094878969398e-05, "loss": 1.0343, "step": 8770 }, { "epoch": 1.41, "grad_norm": 6.965032577514648, "learning_rate": 1.1432374831105289e-05, "loss": 0.9118, "step": 8771 }, { "epoch": 1.41, "grad_norm": 9.315482139587402, "learning_rate": 1.143065473997468e-05, "loss": 1.0182, "step": 8772 }, { "epoch": 1.41, "grad_norm": 16.083208084106445, "learning_rate": 1.1428934605629524e-05, "loss": 0.9515, "step": 8773 }, { "epoch": 1.41, "grad_norm": 7.704286098480225, "learning_rate": 1.1427214428121791e-05, "loss": 0.9006, "step": 8774 }, { "epoch": 1.41, "grad_norm": 8.630053520202637, "learning_rate": 1.1425494207503434e-05, "loss": 0.9134, "step": 8775 }, { "epoch": 1.41, "grad_norm": 8.399646759033203, "learning_rate": 1.1423773943826411e-05, "loss": 1.1556, "step": 8776 }, { "epoch": 1.41, "grad_norm": 9.782368659973145, "learning_rate": 1.142205363714269e-05, "loss": 1.2837, "step": 8777 }, { "epoch": 1.41, "grad_norm": 7.422635078430176, "learning_rate": 1.1420333287504233e-05, "loss": 1.1533, "step": 8778 }, { "epoch": 1.41, "grad_norm": 10.061473846435547, "learning_rate": 1.1418612894963007e-05, "loss": 0.9487, "step": 8779 }, { "epoch": 1.41, "grad_norm": 11.720726013183594, "learning_rate": 1.1416892459570972e-05, "loss": 1.1216, "step": 8780 }, { "epoch": 1.41, "grad_norm": 9.710474014282227, "learning_rate": 1.14151719813801e-05, "loss": 1.0119, "step": 8781 }, { "epoch": 1.41, "grad_norm": 9.63257122039795, "learning_rate": 1.1413451460442361e-05, "loss": 1.1977, "step": 8782 }, { "epoch": 1.41, "grad_norm": 9.228403091430664, "learning_rate": 1.1411730896809724e-05, "loss": 0.9277, "step": 8783 }, { "epoch": 1.41, "grad_norm": 11.078217506408691, "learning_rate": 1.1410010290534164e-05, "loss": 1.158, "step": 8784 }, { "epoch": 1.41, "grad_norm": 9.600780487060547, "learning_rate": 1.1408289641667646e-05, "loss": 1.1014, "step": 8785 }, { "epoch": 1.41, "grad_norm": 12.104105949401855, "learning_rate": 1.1406568950262151e-05, "loss": 1.2514, "step": 8786 }, { "epoch": 1.41, "grad_norm": 8.837778091430664, "learning_rate": 1.1404848216369655e-05, "loss": 1.0391, "step": 8787 }, { "epoch": 1.41, "grad_norm": 9.68649959564209, "learning_rate": 1.1403127440042127e-05, "loss": 0.9411, "step": 8788 }, { "epoch": 1.41, "grad_norm": 10.779777526855469, "learning_rate": 1.1401406621331552e-05, "loss": 1.0619, "step": 8789 }, { "epoch": 1.42, "grad_norm": 8.727643966674805, "learning_rate": 1.1399685760289908e-05, "loss": 1.0545, "step": 8790 }, { "epoch": 1.42, "grad_norm": 10.248259544372559, "learning_rate": 1.1397964856969177e-05, "loss": 1.1892, "step": 8791 }, { "epoch": 1.42, "grad_norm": 11.009552001953125, "learning_rate": 1.1396243911421335e-05, "loss": 0.9863, "step": 8792 }, { "epoch": 1.42, "grad_norm": 9.240480422973633, "learning_rate": 1.1394522923698366e-05, "loss": 0.8666, "step": 8793 }, { "epoch": 1.42, "grad_norm": 10.169309616088867, "learning_rate": 1.1392801893852263e-05, "loss": 0.9541, "step": 8794 }, { "epoch": 1.42, "grad_norm": 9.21965217590332, "learning_rate": 1.1391080821935004e-05, "loss": 0.9259, "step": 8795 }, { "epoch": 1.42, "grad_norm": 9.057586669921875, "learning_rate": 1.1389359707998575e-05, "loss": 1.4242, "step": 8796 }, { "epoch": 1.42, "grad_norm": 11.490063667297363, "learning_rate": 1.1387638552094969e-05, "loss": 1.2319, "step": 8797 }, { "epoch": 1.42, "grad_norm": 7.70916223526001, "learning_rate": 1.1385917354276174e-05, "loss": 0.8547, "step": 8798 }, { "epoch": 1.42, "grad_norm": 8.868145942687988, "learning_rate": 1.1384196114594179e-05, "loss": 0.9232, "step": 8799 }, { "epoch": 1.42, "grad_norm": 10.6431303024292, "learning_rate": 1.1382474833100975e-05, "loss": 0.8632, "step": 8800 }, { "epoch": 1.42, "grad_norm": 9.756356239318848, "learning_rate": 1.1380753509848561e-05, "loss": 0.953, "step": 8801 }, { "epoch": 1.42, "grad_norm": 7.986217498779297, "learning_rate": 1.1379032144888926e-05, "loss": 1.1917, "step": 8802 }, { "epoch": 1.42, "grad_norm": 8.354192733764648, "learning_rate": 1.1377310738274065e-05, "loss": 0.8944, "step": 8803 }, { "epoch": 1.42, "grad_norm": 8.227241516113281, "learning_rate": 1.1375589290055982e-05, "loss": 0.9053, "step": 8804 }, { "epoch": 1.42, "grad_norm": 11.127243995666504, "learning_rate": 1.1373867800286669e-05, "loss": 1.3625, "step": 8805 }, { "epoch": 1.42, "grad_norm": 8.283116340637207, "learning_rate": 1.137214626901813e-05, "loss": 1.0217, "step": 8806 }, { "epoch": 1.42, "grad_norm": 9.315832138061523, "learning_rate": 1.1370424696302358e-05, "loss": 1.2931, "step": 8807 }, { "epoch": 1.42, "grad_norm": 11.32941722869873, "learning_rate": 1.1368703082191364e-05, "loss": 1.1179, "step": 8808 }, { "epoch": 1.42, "grad_norm": 9.838630676269531, "learning_rate": 1.1366981426737145e-05, "loss": 1.0701, "step": 8809 }, { "epoch": 1.42, "grad_norm": 10.30350399017334, "learning_rate": 1.136525972999171e-05, "loss": 1.4192, "step": 8810 }, { "epoch": 1.42, "grad_norm": 10.707569122314453, "learning_rate": 1.1363537992007063e-05, "loss": 1.3179, "step": 8811 }, { "epoch": 1.42, "grad_norm": 9.548698425292969, "learning_rate": 1.1361816212835211e-05, "loss": 0.9698, "step": 8812 }, { "epoch": 1.42, "grad_norm": 9.420453071594238, "learning_rate": 1.1360094392528158e-05, "loss": 0.8765, "step": 8813 }, { "epoch": 1.42, "grad_norm": 9.126103401184082, "learning_rate": 1.1358372531137925e-05, "loss": 1.0383, "step": 8814 }, { "epoch": 1.42, "grad_norm": 9.839584350585938, "learning_rate": 1.135665062871651e-05, "loss": 1.2522, "step": 8815 }, { "epoch": 1.42, "grad_norm": 10.28671646118164, "learning_rate": 1.1354928685315932e-05, "loss": 1.0255, "step": 8816 }, { "epoch": 1.42, "grad_norm": 8.405959129333496, "learning_rate": 1.1353206700988202e-05, "loss": 0.9495, "step": 8817 }, { "epoch": 1.42, "grad_norm": 9.451350212097168, "learning_rate": 1.1351484675785338e-05, "loss": 1.1831, "step": 8818 }, { "epoch": 1.42, "grad_norm": 8.278254508972168, "learning_rate": 1.1349762609759351e-05, "loss": 1.0223, "step": 8819 }, { "epoch": 1.42, "grad_norm": 8.737197875976562, "learning_rate": 1.134804050296226e-05, "loss": 1.2116, "step": 8820 }, { "epoch": 1.42, "grad_norm": 9.797645568847656, "learning_rate": 1.1346318355446085e-05, "loss": 1.2083, "step": 8821 }, { "epoch": 1.42, "grad_norm": 9.339595794677734, "learning_rate": 1.1344596167262843e-05, "loss": 1.4377, "step": 8822 }, { "epoch": 1.42, "grad_norm": 9.36262321472168, "learning_rate": 1.1342873938464552e-05, "loss": 1.1209, "step": 8823 }, { "epoch": 1.42, "grad_norm": 9.264435768127441, "learning_rate": 1.1341151669103238e-05, "loss": 1.1526, "step": 8824 }, { "epoch": 1.42, "grad_norm": 8.621060371398926, "learning_rate": 1.1339429359230923e-05, "loss": 1.2045, "step": 8825 }, { "epoch": 1.42, "grad_norm": 8.26057243347168, "learning_rate": 1.1337707008899634e-05, "loss": 0.882, "step": 8826 }, { "epoch": 1.42, "grad_norm": 9.339046478271484, "learning_rate": 1.1335984618161393e-05, "loss": 0.8206, "step": 8827 }, { "epoch": 1.42, "grad_norm": 9.43063735961914, "learning_rate": 1.1334262187068226e-05, "loss": 1.1027, "step": 8828 }, { "epoch": 1.42, "grad_norm": 11.409778594970703, "learning_rate": 1.1332539715672164e-05, "loss": 1.0871, "step": 8829 }, { "epoch": 1.42, "grad_norm": 7.67844820022583, "learning_rate": 1.1330817204025237e-05, "loss": 0.9476, "step": 8830 }, { "epoch": 1.42, "grad_norm": 8.278314590454102, "learning_rate": 1.132909465217947e-05, "loss": 1.3038, "step": 8831 }, { "epoch": 1.42, "grad_norm": 9.042930603027344, "learning_rate": 1.1327372060186898e-05, "loss": 1.027, "step": 8832 }, { "epoch": 1.42, "grad_norm": 8.483733177185059, "learning_rate": 1.1325649428099555e-05, "loss": 1.1008, "step": 8833 }, { "epoch": 1.42, "grad_norm": 10.420451164245605, "learning_rate": 1.1323926755969471e-05, "loss": 1.0086, "step": 8834 }, { "epoch": 1.42, "grad_norm": 7.888525009155273, "learning_rate": 1.1322204043848687e-05, "loss": 1.196, "step": 8835 }, { "epoch": 1.42, "grad_norm": 11.7529935836792, "learning_rate": 1.1320481291789236e-05, "loss": 1.1002, "step": 8836 }, { "epoch": 1.42, "grad_norm": 8.27626895904541, "learning_rate": 1.1318758499843159e-05, "loss": 1.1488, "step": 8837 }, { "epoch": 1.42, "grad_norm": 9.225081443786621, "learning_rate": 1.1317035668062488e-05, "loss": 0.9485, "step": 8838 }, { "epoch": 1.42, "grad_norm": 10.283782958984375, "learning_rate": 1.131531279649927e-05, "loss": 0.9151, "step": 8839 }, { "epoch": 1.42, "grad_norm": 9.084816932678223, "learning_rate": 1.131358988520554e-05, "loss": 1.0269, "step": 8840 }, { "epoch": 1.42, "grad_norm": 8.224564552307129, "learning_rate": 1.1311866934233344e-05, "loss": 0.9731, "step": 8841 }, { "epoch": 1.42, "grad_norm": 10.590027809143066, "learning_rate": 1.131014394363473e-05, "loss": 0.8905, "step": 8842 }, { "epoch": 1.42, "grad_norm": 10.302529335021973, "learning_rate": 1.1308420913461735e-05, "loss": 1.1828, "step": 8843 }, { "epoch": 1.42, "grad_norm": 7.849194526672363, "learning_rate": 1.1306697843766407e-05, "loss": 0.8068, "step": 8844 }, { "epoch": 1.42, "grad_norm": 8.869213104248047, "learning_rate": 1.13049747346008e-05, "loss": 0.8992, "step": 8845 }, { "epoch": 1.42, "grad_norm": 9.257824897766113, "learning_rate": 1.1303251586016954e-05, "loss": 1.192, "step": 8846 }, { "epoch": 1.42, "grad_norm": 8.657248497009277, "learning_rate": 1.1301528398066922e-05, "loss": 1.2471, "step": 8847 }, { "epoch": 1.42, "grad_norm": 10.474995613098145, "learning_rate": 1.1299805170802754e-05, "loss": 1.287, "step": 8848 }, { "epoch": 1.42, "grad_norm": 10.867180824279785, "learning_rate": 1.1298081904276505e-05, "loss": 0.9167, "step": 8849 }, { "epoch": 1.42, "grad_norm": 12.45731258392334, "learning_rate": 1.1296358598540224e-05, "loss": 1.178, "step": 8850 }, { "epoch": 1.42, "grad_norm": 8.64318561553955, "learning_rate": 1.1294635253645969e-05, "loss": 0.9238, "step": 8851 }, { "epoch": 1.42, "grad_norm": 9.057111740112305, "learning_rate": 1.1292911869645793e-05, "loss": 1.0858, "step": 8852 }, { "epoch": 1.43, "grad_norm": 9.138837814331055, "learning_rate": 1.1291188446591754e-05, "loss": 0.9944, "step": 8853 }, { "epoch": 1.43, "grad_norm": 15.187965393066406, "learning_rate": 1.128946498453591e-05, "loss": 1.7713, "step": 8854 }, { "epoch": 1.43, "grad_norm": 9.35268497467041, "learning_rate": 1.1287741483530317e-05, "loss": 0.9595, "step": 8855 }, { "epoch": 1.43, "grad_norm": 8.439664840698242, "learning_rate": 1.1286017943627041e-05, "loss": 1.0148, "step": 8856 }, { "epoch": 1.43, "grad_norm": 9.855683326721191, "learning_rate": 1.1284294364878141e-05, "loss": 1.1019, "step": 8857 }, { "epoch": 1.43, "grad_norm": 8.067959785461426, "learning_rate": 1.128257074733568e-05, "loss": 0.9992, "step": 8858 }, { "epoch": 1.43, "grad_norm": 9.802413940429688, "learning_rate": 1.1280847091051716e-05, "loss": 1.1479, "step": 8859 }, { "epoch": 1.43, "grad_norm": 10.445685386657715, "learning_rate": 1.1279123396078322e-05, "loss": 1.0965, "step": 8860 }, { "epoch": 1.43, "grad_norm": 8.901530265808105, "learning_rate": 1.1277399662467564e-05, "loss": 0.9987, "step": 8861 }, { "epoch": 1.43, "grad_norm": 8.828705787658691, "learning_rate": 1.1275675890271502e-05, "loss": 1.0163, "step": 8862 }, { "epoch": 1.43, "grad_norm": 7.846931457519531, "learning_rate": 1.1273952079542212e-05, "loss": 0.9868, "step": 8863 }, { "epoch": 1.43, "grad_norm": 9.444869041442871, "learning_rate": 1.1272228230331757e-05, "loss": 0.9296, "step": 8864 }, { "epoch": 1.43, "grad_norm": 9.503297805786133, "learning_rate": 1.1270504342692215e-05, "loss": 1.0543, "step": 8865 }, { "epoch": 1.43, "grad_norm": 11.042890548706055, "learning_rate": 1.1268780416675652e-05, "loss": 1.1709, "step": 8866 }, { "epoch": 1.43, "grad_norm": 10.105138778686523, "learning_rate": 1.1267056452334144e-05, "loss": 1.2597, "step": 8867 }, { "epoch": 1.43, "grad_norm": 10.331710815429688, "learning_rate": 1.1265332449719766e-05, "loss": 0.8603, "step": 8868 }, { "epoch": 1.43, "grad_norm": 9.24232292175293, "learning_rate": 1.1263608408884595e-05, "loss": 1.1245, "step": 8869 }, { "epoch": 1.43, "grad_norm": 10.31771183013916, "learning_rate": 1.1261884329880699e-05, "loss": 0.9847, "step": 8870 }, { "epoch": 1.43, "grad_norm": 8.372215270996094, "learning_rate": 1.1260160212760165e-05, "loss": 0.9374, "step": 8871 }, { "epoch": 1.43, "grad_norm": 8.979998588562012, "learning_rate": 1.1258436057575066e-05, "loss": 1.2217, "step": 8872 }, { "epoch": 1.43, "grad_norm": 7.64735221862793, "learning_rate": 1.1256711864377487e-05, "loss": 0.8448, "step": 8873 }, { "epoch": 1.43, "grad_norm": 9.689702987670898, "learning_rate": 1.1254987633219505e-05, "loss": 0.9471, "step": 8874 }, { "epoch": 1.43, "grad_norm": 8.582514762878418, "learning_rate": 1.1253263364153205e-05, "loss": 1.1884, "step": 8875 }, { "epoch": 1.43, "grad_norm": 10.40808391571045, "learning_rate": 1.1251539057230671e-05, "loss": 1.1784, "step": 8876 }, { "epoch": 1.43, "grad_norm": 9.937272071838379, "learning_rate": 1.1249814712503983e-05, "loss": 1.1724, "step": 8877 }, { "epoch": 1.43, "grad_norm": 8.30956745147705, "learning_rate": 1.1248090330025232e-05, "loss": 1.011, "step": 8878 }, { "epoch": 1.43, "grad_norm": 8.994194030761719, "learning_rate": 1.1246365909846504e-05, "loss": 1.0701, "step": 8879 }, { "epoch": 1.43, "grad_norm": 9.767606735229492, "learning_rate": 1.1244641452019884e-05, "loss": 1.1523, "step": 8880 }, { "epoch": 1.43, "grad_norm": 9.42658519744873, "learning_rate": 1.1242916956597464e-05, "loss": 0.9571, "step": 8881 }, { "epoch": 1.43, "grad_norm": 9.058157920837402, "learning_rate": 1.1241192423631335e-05, "loss": 1.4116, "step": 8882 }, { "epoch": 1.43, "grad_norm": 9.949177742004395, "learning_rate": 1.1239467853173585e-05, "loss": 1.0397, "step": 8883 }, { "epoch": 1.43, "grad_norm": 8.881141662597656, "learning_rate": 1.123774324527631e-05, "loss": 1.0307, "step": 8884 }, { "epoch": 1.43, "grad_norm": 9.178098678588867, "learning_rate": 1.12360185999916e-05, "loss": 0.8139, "step": 8885 }, { "epoch": 1.43, "grad_norm": 10.61792278289795, "learning_rate": 1.1234293917371555e-05, "loss": 1.1147, "step": 8886 }, { "epoch": 1.43, "grad_norm": 7.960679054260254, "learning_rate": 1.1232569197468266e-05, "loss": 0.9617, "step": 8887 }, { "epoch": 1.43, "grad_norm": 10.619744300842285, "learning_rate": 1.1230844440333835e-05, "loss": 1.2158, "step": 8888 }, { "epoch": 1.43, "grad_norm": 8.048303604125977, "learning_rate": 1.1229119646020354e-05, "loss": 1.234, "step": 8889 }, { "epoch": 1.43, "grad_norm": 9.58914852142334, "learning_rate": 1.1227394814579929e-05, "loss": 1.0901, "step": 8890 }, { "epoch": 1.43, "grad_norm": 9.569875717163086, "learning_rate": 1.1225669946064655e-05, "loss": 1.1669, "step": 8891 }, { "epoch": 1.43, "grad_norm": 10.734358787536621, "learning_rate": 1.1223945040526636e-05, "loss": 1.2155, "step": 8892 }, { "epoch": 1.43, "grad_norm": 13.128368377685547, "learning_rate": 1.1222220098017977e-05, "loss": 1.3136, "step": 8893 }, { "epoch": 1.43, "grad_norm": 9.013988494873047, "learning_rate": 1.1220495118590777e-05, "loss": 1.2023, "step": 8894 }, { "epoch": 1.43, "grad_norm": 10.81729793548584, "learning_rate": 1.1218770102297144e-05, "loss": 1.2177, "step": 8895 }, { "epoch": 1.43, "grad_norm": 9.12613582611084, "learning_rate": 1.1217045049189186e-05, "loss": 0.873, "step": 8896 }, { "epoch": 1.43, "grad_norm": 7.343288898468018, "learning_rate": 1.1215319959319007e-05, "loss": 1.1454, "step": 8897 }, { "epoch": 1.43, "grad_norm": 8.760014533996582, "learning_rate": 1.1213594832738713e-05, "loss": 0.8514, "step": 8898 }, { "epoch": 1.43, "grad_norm": 9.235699653625488, "learning_rate": 1.1211869669500419e-05, "loss": 1.1246, "step": 8899 }, { "epoch": 1.43, "grad_norm": 8.977665901184082, "learning_rate": 1.1210144469656235e-05, "loss": 1.1485, "step": 8900 }, { "epoch": 1.43, "grad_norm": 8.098644256591797, "learning_rate": 1.1208419233258265e-05, "loss": 0.8981, "step": 8901 }, { "epoch": 1.43, "grad_norm": 8.97492504119873, "learning_rate": 1.120669396035863e-05, "loss": 0.942, "step": 8902 }, { "epoch": 1.43, "grad_norm": 7.2018327713012695, "learning_rate": 1.120496865100944e-05, "loss": 0.7966, "step": 8903 }, { "epoch": 1.43, "grad_norm": 8.86258602142334, "learning_rate": 1.1203243305262813e-05, "loss": 0.7824, "step": 8904 }, { "epoch": 1.43, "grad_norm": 9.866602897644043, "learning_rate": 1.1201517923170861e-05, "loss": 1.1242, "step": 8905 }, { "epoch": 1.43, "grad_norm": 8.999842643737793, "learning_rate": 1.1199792504785704e-05, "loss": 1.1294, "step": 8906 }, { "epoch": 1.43, "grad_norm": 8.982446670532227, "learning_rate": 1.119806705015946e-05, "loss": 1.2253, "step": 8907 }, { "epoch": 1.43, "grad_norm": 8.333460807800293, "learning_rate": 1.1196341559344248e-05, "loss": 1.0498, "step": 8908 }, { "epoch": 1.43, "grad_norm": 8.533005714416504, "learning_rate": 1.1194616032392186e-05, "loss": 0.9486, "step": 8909 }, { "epoch": 1.43, "grad_norm": 7.534408092498779, "learning_rate": 1.11928904693554e-05, "loss": 1.0425, "step": 8910 }, { "epoch": 1.43, "grad_norm": 9.723685264587402, "learning_rate": 1.1191164870286006e-05, "loss": 1.077, "step": 8911 }, { "epoch": 1.43, "grad_norm": 9.68529224395752, "learning_rate": 1.1189439235236137e-05, "loss": 1.1534, "step": 8912 }, { "epoch": 1.43, "grad_norm": 8.468893051147461, "learning_rate": 1.118771356425791e-05, "loss": 1.0123, "step": 8913 }, { "epoch": 1.43, "grad_norm": 12.053483009338379, "learning_rate": 1.1185987857403453e-05, "loss": 0.9228, "step": 8914 }, { "epoch": 1.44, "grad_norm": 11.815377235412598, "learning_rate": 1.1184262114724894e-05, "loss": 1.0142, "step": 8915 }, { "epoch": 1.44, "grad_norm": 9.574235916137695, "learning_rate": 1.1182536336274359e-05, "loss": 0.9698, "step": 8916 }, { "epoch": 1.44, "grad_norm": 9.764286994934082, "learning_rate": 1.118081052210398e-05, "loss": 1.167, "step": 8917 }, { "epoch": 1.44, "grad_norm": 10.193795204162598, "learning_rate": 1.1179084672265883e-05, "loss": 1.3203, "step": 8918 }, { "epoch": 1.44, "grad_norm": 10.510171890258789, "learning_rate": 1.1177358786812202e-05, "loss": 1.2951, "step": 8919 }, { "epoch": 1.44, "grad_norm": 9.343962669372559, "learning_rate": 1.1175632865795074e-05, "loss": 1.0605, "step": 8920 }, { "epoch": 1.44, "grad_norm": 11.701326370239258, "learning_rate": 1.1173906909266623e-05, "loss": 0.965, "step": 8921 }, { "epoch": 1.44, "grad_norm": 11.351218223571777, "learning_rate": 1.117218091727899e-05, "loss": 1.3805, "step": 8922 }, { "epoch": 1.44, "grad_norm": 9.594338417053223, "learning_rate": 1.1170454889884305e-05, "loss": 0.9867, "step": 8923 }, { "epoch": 1.44, "grad_norm": 9.712342262268066, "learning_rate": 1.1168728827134714e-05, "loss": 1.0573, "step": 8924 }, { "epoch": 1.44, "grad_norm": 8.897856712341309, "learning_rate": 1.1167002729082344e-05, "loss": 1.1214, "step": 8925 }, { "epoch": 1.44, "grad_norm": 12.760669708251953, "learning_rate": 1.1165276595779337e-05, "loss": 1.0179, "step": 8926 }, { "epoch": 1.44, "grad_norm": 7.753700256347656, "learning_rate": 1.1163550427277841e-05, "loss": 1.1123, "step": 8927 }, { "epoch": 1.44, "grad_norm": 9.530531883239746, "learning_rate": 1.1161824223629987e-05, "loss": 1.262, "step": 8928 }, { "epoch": 1.44, "grad_norm": 9.379775047302246, "learning_rate": 1.1160097984887919e-05, "loss": 0.944, "step": 8929 }, { "epoch": 1.44, "grad_norm": 10.698143005371094, "learning_rate": 1.1158371711103779e-05, "loss": 1.2198, "step": 8930 }, { "epoch": 1.44, "grad_norm": 8.534671783447266, "learning_rate": 1.1156645402329717e-05, "loss": 1.0079, "step": 8931 }, { "epoch": 1.44, "grad_norm": 8.38311767578125, "learning_rate": 1.1154919058617874e-05, "loss": 1.0815, "step": 8932 }, { "epoch": 1.44, "grad_norm": 8.541891098022461, "learning_rate": 1.1153192680020396e-05, "loss": 0.972, "step": 8933 }, { "epoch": 1.44, "grad_norm": 10.840177536010742, "learning_rate": 1.1151466266589431e-05, "loss": 1.1486, "step": 8934 }, { "epoch": 1.44, "grad_norm": 8.350440979003906, "learning_rate": 1.1149739818377126e-05, "loss": 0.9872, "step": 8935 }, { "epoch": 1.44, "grad_norm": 10.377314567565918, "learning_rate": 1.114801333543563e-05, "loss": 1.1087, "step": 8936 }, { "epoch": 1.44, "grad_norm": 8.52670955657959, "learning_rate": 1.1146286817817096e-05, "loss": 0.8993, "step": 8937 }, { "epoch": 1.44, "grad_norm": 8.295090675354004, "learning_rate": 1.1144560265573674e-05, "loss": 0.8771, "step": 8938 }, { "epoch": 1.44, "grad_norm": 9.274189949035645, "learning_rate": 1.1142833678757516e-05, "loss": 1.1762, "step": 8939 }, { "epoch": 1.44, "grad_norm": 7.979900360107422, "learning_rate": 1.1141107057420777e-05, "loss": 1.0108, "step": 8940 }, { "epoch": 1.44, "grad_norm": 11.341294288635254, "learning_rate": 1.1139380401615609e-05, "loss": 0.9777, "step": 8941 }, { "epoch": 1.44, "grad_norm": 8.860527992248535, "learning_rate": 1.1137653711394171e-05, "loss": 1.0494, "step": 8942 }, { "epoch": 1.44, "grad_norm": 9.093894004821777, "learning_rate": 1.1135926986808619e-05, "loss": 0.9695, "step": 8943 }, { "epoch": 1.44, "grad_norm": 8.559446334838867, "learning_rate": 1.1134200227911107e-05, "loss": 1.1078, "step": 8944 }, { "epoch": 1.44, "grad_norm": 9.122729301452637, "learning_rate": 1.1132473434753794e-05, "loss": 1.3559, "step": 8945 }, { "epoch": 1.44, "grad_norm": 9.59448528289795, "learning_rate": 1.1130746607388843e-05, "loss": 1.0583, "step": 8946 }, { "epoch": 1.44, "grad_norm": 9.07730770111084, "learning_rate": 1.1129019745868418e-05, "loss": 1.1988, "step": 8947 }, { "epoch": 1.44, "grad_norm": 8.402775764465332, "learning_rate": 1.1127292850244672e-05, "loss": 0.9784, "step": 8948 }, { "epoch": 1.44, "grad_norm": 10.096881866455078, "learning_rate": 1.1125565920569774e-05, "loss": 1.0039, "step": 8949 }, { "epoch": 1.44, "grad_norm": 8.768854141235352, "learning_rate": 1.1123838956895884e-05, "loss": 0.8156, "step": 8950 }, { "epoch": 1.44, "grad_norm": 8.924034118652344, "learning_rate": 1.1122111959275172e-05, "loss": 1.1365, "step": 8951 }, { "epoch": 1.44, "grad_norm": 9.12421989440918, "learning_rate": 1.1120384927759799e-05, "loss": 1.1169, "step": 8952 }, { "epoch": 1.44, "grad_norm": 8.604540824890137, "learning_rate": 1.1118657862401936e-05, "loss": 1.0954, "step": 8953 }, { "epoch": 1.44, "grad_norm": 8.61596393585205, "learning_rate": 1.1116930763253746e-05, "loss": 0.9407, "step": 8954 }, { "epoch": 1.44, "grad_norm": 8.14713191986084, "learning_rate": 1.1115203630367401e-05, "loss": 0.8299, "step": 8955 }, { "epoch": 1.44, "grad_norm": 9.701546669006348, "learning_rate": 1.1113476463795073e-05, "loss": 0.8765, "step": 8956 }, { "epoch": 1.44, "grad_norm": 10.77208137512207, "learning_rate": 1.111174926358893e-05, "loss": 1.2047, "step": 8957 }, { "epoch": 1.44, "grad_norm": 9.634111404418945, "learning_rate": 1.1110022029801144e-05, "loss": 0.9995, "step": 8958 }, { "epoch": 1.44, "grad_norm": 9.005359649658203, "learning_rate": 1.110829476248389e-05, "loss": 0.951, "step": 8959 }, { "epoch": 1.44, "grad_norm": 9.230931282043457, "learning_rate": 1.1106567461689341e-05, "loss": 0.888, "step": 8960 }, { "epoch": 1.44, "grad_norm": 9.928302764892578, "learning_rate": 1.1104840127469672e-05, "loss": 0.9623, "step": 8961 }, { "epoch": 1.44, "grad_norm": 8.357536315917969, "learning_rate": 1.1103112759877057e-05, "loss": 0.9847, "step": 8962 }, { "epoch": 1.44, "grad_norm": 8.741085052490234, "learning_rate": 1.1101385358963678e-05, "loss": 0.9422, "step": 8963 }, { "epoch": 1.44, "grad_norm": 10.139527320861816, "learning_rate": 1.1099657924781709e-05, "loss": 0.9523, "step": 8964 }, { "epoch": 1.44, "grad_norm": 8.153939247131348, "learning_rate": 1.1097930457383327e-05, "loss": 0.8809, "step": 8965 }, { "epoch": 1.44, "grad_norm": 8.378819465637207, "learning_rate": 1.109620295682072e-05, "loss": 1.011, "step": 8966 }, { "epoch": 1.44, "grad_norm": 9.26948070526123, "learning_rate": 1.1094475423146065e-05, "loss": 1.3256, "step": 8967 }, { "epoch": 1.44, "grad_norm": 9.522857666015625, "learning_rate": 1.1092747856411542e-05, "loss": 1.097, "step": 8968 }, { "epoch": 1.44, "grad_norm": 9.352413177490234, "learning_rate": 1.1091020256669336e-05, "loss": 0.9024, "step": 8969 }, { "epoch": 1.44, "grad_norm": 9.087797164916992, "learning_rate": 1.108929262397163e-05, "loss": 1.0497, "step": 8970 }, { "epoch": 1.44, "grad_norm": 9.154051780700684, "learning_rate": 1.1087564958370614e-05, "loss": 1.0116, "step": 8971 }, { "epoch": 1.44, "grad_norm": 9.83405876159668, "learning_rate": 1.1085837259918466e-05, "loss": 1.1096, "step": 8972 }, { "epoch": 1.44, "grad_norm": 8.263423919677734, "learning_rate": 1.1084109528667377e-05, "loss": 0.9243, "step": 8973 }, { "epoch": 1.44, "grad_norm": 12.673748970031738, "learning_rate": 1.1082381764669536e-05, "loss": 1.5497, "step": 8974 }, { "epoch": 1.44, "grad_norm": 10.173299789428711, "learning_rate": 1.1080653967977133e-05, "loss": 0.9491, "step": 8975 }, { "epoch": 1.44, "grad_norm": 9.156599044799805, "learning_rate": 1.1078926138642354e-05, "loss": 1.2149, "step": 8976 }, { "epoch": 1.45, "grad_norm": 9.954391479492188, "learning_rate": 1.1077198276717394e-05, "loss": 1.0682, "step": 8977 }, { "epoch": 1.45, "grad_norm": 10.059001922607422, "learning_rate": 1.1075470382254445e-05, "loss": 1.0989, "step": 8978 }, { "epoch": 1.45, "grad_norm": 10.80988883972168, "learning_rate": 1.1073742455305697e-05, "loss": 0.8596, "step": 8979 }, { "epoch": 1.45, "grad_norm": 9.968308448791504, "learning_rate": 1.1072014495923342e-05, "loss": 1.0498, "step": 8980 }, { "epoch": 1.45, "grad_norm": 8.872786521911621, "learning_rate": 1.1070286504159582e-05, "loss": 1.0742, "step": 8981 }, { "epoch": 1.45, "grad_norm": 8.150594711303711, "learning_rate": 1.1068558480066612e-05, "loss": 0.8288, "step": 8982 }, { "epoch": 1.45, "grad_norm": 13.850050926208496, "learning_rate": 1.1066830423696621e-05, "loss": 1.283, "step": 8983 }, { "epoch": 1.45, "grad_norm": 10.91064167022705, "learning_rate": 1.1065102335101813e-05, "loss": 1.0204, "step": 8984 }, { "epoch": 1.45, "grad_norm": 9.55488395690918, "learning_rate": 1.1063374214334386e-05, "loss": 0.7903, "step": 8985 }, { "epoch": 1.45, "grad_norm": 10.150982856750488, "learning_rate": 1.1061646061446543e-05, "loss": 1.1924, "step": 8986 }, { "epoch": 1.45, "grad_norm": 10.107674598693848, "learning_rate": 1.1059917876490478e-05, "loss": 1.0863, "step": 8987 }, { "epoch": 1.45, "grad_norm": 10.283514022827148, "learning_rate": 1.1058189659518396e-05, "loss": 1.0691, "step": 8988 }, { "epoch": 1.45, "grad_norm": 8.979832649230957, "learning_rate": 1.1056461410582504e-05, "loss": 1.165, "step": 8989 }, { "epoch": 1.45, "grad_norm": 10.965469360351562, "learning_rate": 1.1054733129735003e-05, "loss": 1.1998, "step": 8990 }, { "epoch": 1.45, "grad_norm": 7.96166467666626, "learning_rate": 1.1053004817028093e-05, "loss": 1.159, "step": 8991 }, { "epoch": 1.45, "grad_norm": 11.563400268554688, "learning_rate": 1.1051276472513984e-05, "loss": 1.1588, "step": 8992 }, { "epoch": 1.45, "grad_norm": 8.665205001831055, "learning_rate": 1.104954809624488e-05, "loss": 1.1034, "step": 8993 }, { "epoch": 1.45, "grad_norm": 10.999883651733398, "learning_rate": 1.1047819688272994e-05, "loss": 1.1091, "step": 8994 }, { "epoch": 1.45, "grad_norm": 12.520602226257324, "learning_rate": 1.1046091248650528e-05, "loss": 1.2895, "step": 8995 }, { "epoch": 1.45, "grad_norm": 11.912552833557129, "learning_rate": 1.1044362777429694e-05, "loss": 1.0501, "step": 8996 }, { "epoch": 1.45, "grad_norm": 9.22199535369873, "learning_rate": 1.1042634274662703e-05, "loss": 1.1214, "step": 8997 }, { "epoch": 1.45, "grad_norm": 8.157876014709473, "learning_rate": 1.1040905740401769e-05, "loss": 1.0698, "step": 8998 }, { "epoch": 1.45, "grad_norm": 8.072982788085938, "learning_rate": 1.1039177174699098e-05, "loss": 0.8534, "step": 8999 }, { "epoch": 1.45, "grad_norm": 10.504476547241211, "learning_rate": 1.1037448577606908e-05, "loss": 1.0388, "step": 9000 }, { "epoch": 1.45, "grad_norm": 9.204122543334961, "learning_rate": 1.1035719949177413e-05, "loss": 1.0508, "step": 9001 }, { "epoch": 1.45, "grad_norm": 7.680531024932861, "learning_rate": 1.1033991289462829e-05, "loss": 0.7327, "step": 9002 }, { "epoch": 1.45, "grad_norm": 9.139699935913086, "learning_rate": 1.1032262598515366e-05, "loss": 1.0089, "step": 9003 }, { "epoch": 1.45, "grad_norm": 9.594439506530762, "learning_rate": 1.1030533876387248e-05, "loss": 1.1026, "step": 9004 }, { "epoch": 1.45, "grad_norm": 11.024833679199219, "learning_rate": 1.1028805123130688e-05, "loss": 1.051, "step": 9005 }, { "epoch": 1.45, "grad_norm": 7.841409206390381, "learning_rate": 1.1027076338797908e-05, "loss": 1.0184, "step": 9006 }, { "epoch": 1.45, "grad_norm": 8.463164329528809, "learning_rate": 1.1025347523441128e-05, "loss": 0.9397, "step": 9007 }, { "epoch": 1.45, "grad_norm": 9.064257621765137, "learning_rate": 1.1023618677112568e-05, "loss": 1.0689, "step": 9008 }, { "epoch": 1.45, "grad_norm": 7.51349401473999, "learning_rate": 1.1021889799864449e-05, "loss": 0.8454, "step": 9009 }, { "epoch": 1.45, "grad_norm": 10.865575790405273, "learning_rate": 1.1020160891748996e-05, "loss": 1.4793, "step": 9010 }, { "epoch": 1.45, "grad_norm": 9.767374038696289, "learning_rate": 1.1018431952818432e-05, "loss": 0.9855, "step": 9011 }, { "epoch": 1.45, "grad_norm": 10.979771614074707, "learning_rate": 1.1016702983124978e-05, "loss": 1.2105, "step": 9012 }, { "epoch": 1.45, "grad_norm": 8.312263488769531, "learning_rate": 1.1014973982720863e-05, "loss": 1.0853, "step": 9013 }, { "epoch": 1.45, "grad_norm": 9.37486743927002, "learning_rate": 1.1013244951658313e-05, "loss": 0.9792, "step": 9014 }, { "epoch": 1.45, "grad_norm": 9.96323013305664, "learning_rate": 1.1011515889989555e-05, "loss": 1.173, "step": 9015 }, { "epoch": 1.45, "grad_norm": 9.904023170471191, "learning_rate": 1.1009786797766813e-05, "loss": 1.0178, "step": 9016 }, { "epoch": 1.45, "grad_norm": 9.569625854492188, "learning_rate": 1.1008057675042326e-05, "loss": 1.1642, "step": 9017 }, { "epoch": 1.45, "grad_norm": 8.709975242614746, "learning_rate": 1.1006328521868315e-05, "loss": 1.0084, "step": 9018 }, { "epoch": 1.45, "grad_norm": 9.670247077941895, "learning_rate": 1.1004599338297016e-05, "loss": 0.9603, "step": 9019 }, { "epoch": 1.45, "grad_norm": 9.736413955688477, "learning_rate": 1.1002870124380658e-05, "loss": 1.0862, "step": 9020 }, { "epoch": 1.45, "grad_norm": 8.053393363952637, "learning_rate": 1.100114088017148e-05, "loss": 1.0218, "step": 9021 }, { "epoch": 1.45, "grad_norm": 9.138590812683105, "learning_rate": 1.0999411605721706e-05, "loss": 1.1453, "step": 9022 }, { "epoch": 1.45, "grad_norm": 9.591506958007812, "learning_rate": 1.0997682301083574e-05, "loss": 1.327, "step": 9023 }, { "epoch": 1.45, "grad_norm": 9.204362869262695, "learning_rate": 1.0995952966309325e-05, "loss": 0.9996, "step": 9024 }, { "epoch": 1.45, "grad_norm": 8.861879348754883, "learning_rate": 1.099422360145119e-05, "loss": 0.8859, "step": 9025 }, { "epoch": 1.45, "grad_norm": 9.878985404968262, "learning_rate": 1.099249420656141e-05, "loss": 0.9188, "step": 9026 }, { "epoch": 1.45, "grad_norm": 9.616911888122559, "learning_rate": 1.099076478169222e-05, "loss": 1.1156, "step": 9027 }, { "epoch": 1.45, "grad_norm": 9.351366996765137, "learning_rate": 1.098903532689586e-05, "loss": 1.0395, "step": 9028 }, { "epoch": 1.45, "grad_norm": 8.339628219604492, "learning_rate": 1.0987305842224574e-05, "loss": 0.9432, "step": 9029 }, { "epoch": 1.45, "grad_norm": 9.680224418640137, "learning_rate": 1.0985576327730598e-05, "loss": 1.2049, "step": 9030 }, { "epoch": 1.45, "grad_norm": 7.835319995880127, "learning_rate": 1.0983846783466176e-05, "loss": 1.0255, "step": 9031 }, { "epoch": 1.45, "grad_norm": 10.060213088989258, "learning_rate": 1.0982117209483551e-05, "loss": 1.0608, "step": 9032 }, { "epoch": 1.45, "grad_norm": 10.85662841796875, "learning_rate": 1.0980387605834968e-05, "loss": 1.1342, "step": 9033 }, { "epoch": 1.45, "grad_norm": 8.739278793334961, "learning_rate": 1.0978657972572669e-05, "loss": 0.9209, "step": 9034 }, { "epoch": 1.45, "grad_norm": 8.643269538879395, "learning_rate": 1.0976928309748898e-05, "loss": 0.9071, "step": 9035 }, { "epoch": 1.45, "grad_norm": 8.472731590270996, "learning_rate": 1.0975198617415902e-05, "loss": 0.9569, "step": 9036 }, { "epoch": 1.45, "grad_norm": 7.438924789428711, "learning_rate": 1.0973468895625938e-05, "loss": 0.9358, "step": 9037 }, { "epoch": 1.45, "grad_norm": 11.070772171020508, "learning_rate": 1.0971739144431242e-05, "loss": 1.1046, "step": 9038 }, { "epoch": 1.46, "grad_norm": 8.30114459991455, "learning_rate": 1.0970009363884068e-05, "loss": 0.9142, "step": 9039 }, { "epoch": 1.46, "grad_norm": 8.423574447631836, "learning_rate": 1.0968279554036667e-05, "loss": 1.1349, "step": 9040 }, { "epoch": 1.46, "grad_norm": 10.231897354125977, "learning_rate": 1.0966549714941288e-05, "loss": 1.0594, "step": 9041 }, { "epoch": 1.46, "grad_norm": 10.970378875732422, "learning_rate": 1.0964819846650184e-05, "loss": 1.0517, "step": 9042 }, { "epoch": 1.46, "grad_norm": 9.936517715454102, "learning_rate": 1.0963089949215604e-05, "loss": 1.0413, "step": 9043 }, { "epoch": 1.46, "grad_norm": 9.108235359191895, "learning_rate": 1.0961360022689806e-05, "loss": 1.1313, "step": 9044 }, { "epoch": 1.46, "grad_norm": 8.896692276000977, "learning_rate": 1.0959630067125045e-05, "loss": 1.0452, "step": 9045 }, { "epoch": 1.46, "grad_norm": 8.600378036499023, "learning_rate": 1.095790008257357e-05, "loss": 0.8217, "step": 9046 }, { "epoch": 1.46, "grad_norm": 8.186086654663086, "learning_rate": 1.095617006908764e-05, "loss": 0.7927, "step": 9047 }, { "epoch": 1.46, "grad_norm": 8.244657516479492, "learning_rate": 1.0954440026719518e-05, "loss": 1.1202, "step": 9048 }, { "epoch": 1.46, "grad_norm": 8.37477970123291, "learning_rate": 1.0952709955521455e-05, "loss": 0.9845, "step": 9049 }, { "epoch": 1.46, "grad_norm": 9.127017974853516, "learning_rate": 1.0950979855545711e-05, "loss": 1.058, "step": 9050 }, { "epoch": 1.46, "grad_norm": 10.672998428344727, "learning_rate": 1.0949249726844548e-05, "loss": 1.1514, "step": 9051 }, { "epoch": 1.46, "grad_norm": 9.644412994384766, "learning_rate": 1.0947519569470224e-05, "loss": 0.8391, "step": 9052 }, { "epoch": 1.46, "grad_norm": 9.577810287475586, "learning_rate": 1.0945789383475003e-05, "loss": 1.2901, "step": 9053 }, { "epoch": 1.46, "grad_norm": 8.024423599243164, "learning_rate": 1.0944059168911143e-05, "loss": 0.8521, "step": 9054 }, { "epoch": 1.46, "grad_norm": 11.784697532653809, "learning_rate": 1.094232892583091e-05, "loss": 1.2374, "step": 9055 }, { "epoch": 1.46, "grad_norm": 11.444992065429688, "learning_rate": 1.0940598654286568e-05, "loss": 1.1629, "step": 9056 }, { "epoch": 1.46, "grad_norm": 9.841588020324707, "learning_rate": 1.093886835433038e-05, "loss": 1.0322, "step": 9057 }, { "epoch": 1.46, "grad_norm": 7.485451698303223, "learning_rate": 1.0937138026014613e-05, "loss": 0.8228, "step": 9058 }, { "epoch": 1.46, "grad_norm": 11.712140083312988, "learning_rate": 1.0935407669391535e-05, "loss": 1.1726, "step": 9059 }, { "epoch": 1.46, "grad_norm": 11.925313949584961, "learning_rate": 1.0933677284513414e-05, "loss": 1.0221, "step": 9060 }, { "epoch": 1.46, "grad_norm": 9.926311492919922, "learning_rate": 1.0931946871432513e-05, "loss": 0.9821, "step": 9061 }, { "epoch": 1.46, "grad_norm": 13.214910507202148, "learning_rate": 1.0930216430201103e-05, "loss": 1.0631, "step": 9062 }, { "epoch": 1.46, "grad_norm": 10.900362968444824, "learning_rate": 1.0928485960871458e-05, "loss": 1.0786, "step": 9063 }, { "epoch": 1.46, "grad_norm": 9.050116539001465, "learning_rate": 1.0926755463495845e-05, "loss": 1.114, "step": 9064 }, { "epoch": 1.46, "grad_norm": 9.43106460571289, "learning_rate": 1.0925024938126537e-05, "loss": 1.1236, "step": 9065 }, { "epoch": 1.46, "grad_norm": 9.200946807861328, "learning_rate": 1.0923294384815804e-05, "loss": 1.207, "step": 9066 }, { "epoch": 1.46, "grad_norm": 8.875152587890625, "learning_rate": 1.092156380361592e-05, "loss": 0.8857, "step": 9067 }, { "epoch": 1.46, "grad_norm": 12.018338203430176, "learning_rate": 1.0919833194579166e-05, "loss": 1.2159, "step": 9068 }, { "epoch": 1.46, "grad_norm": 8.555865287780762, "learning_rate": 1.0918102557757808e-05, "loss": 1.0361, "step": 9069 }, { "epoch": 1.46, "grad_norm": 11.899869918823242, "learning_rate": 1.0916371893204125e-05, "loss": 0.9811, "step": 9070 }, { "epoch": 1.46, "grad_norm": 11.432775497436523, "learning_rate": 1.0914641200970392e-05, "loss": 1.1172, "step": 9071 }, { "epoch": 1.46, "grad_norm": 11.945446968078613, "learning_rate": 1.0912910481108896e-05, "loss": 1.2197, "step": 9072 }, { "epoch": 1.46, "grad_norm": 12.100567817687988, "learning_rate": 1.0911179733671901e-05, "loss": 1.3609, "step": 9073 }, { "epoch": 1.46, "grad_norm": 9.076876640319824, "learning_rate": 1.0909448958711694e-05, "loss": 1.0617, "step": 9074 }, { "epoch": 1.46, "grad_norm": 8.955251693725586, "learning_rate": 1.0907718156280555e-05, "loss": 0.7987, "step": 9075 }, { "epoch": 1.46, "grad_norm": 9.09023666381836, "learning_rate": 1.0905987326430765e-05, "loss": 0.9542, "step": 9076 }, { "epoch": 1.46, "grad_norm": 10.73046875, "learning_rate": 1.0904256469214602e-05, "loss": 1.0703, "step": 9077 }, { "epoch": 1.46, "grad_norm": 10.286418914794922, "learning_rate": 1.0902525584684354e-05, "loss": 1.0692, "step": 9078 }, { "epoch": 1.46, "grad_norm": 8.740153312683105, "learning_rate": 1.0900794672892298e-05, "loss": 1.2109, "step": 9079 }, { "epoch": 1.46, "grad_norm": 8.484541893005371, "learning_rate": 1.0899063733890727e-05, "loss": 1.1981, "step": 9080 }, { "epoch": 1.46, "grad_norm": 9.235450744628906, "learning_rate": 1.0897332767731916e-05, "loss": 1.1629, "step": 9081 }, { "epoch": 1.46, "grad_norm": 9.608075141906738, "learning_rate": 1.0895601774468157e-05, "loss": 1.0519, "step": 9082 }, { "epoch": 1.46, "grad_norm": 9.807875633239746, "learning_rate": 1.0893870754151735e-05, "loss": 0.785, "step": 9083 }, { "epoch": 1.46, "grad_norm": 9.345196723937988, "learning_rate": 1.0892139706834937e-05, "loss": 0.8026, "step": 9084 }, { "epoch": 1.46, "grad_norm": 8.94063949584961, "learning_rate": 1.0890408632570051e-05, "loss": 0.9912, "step": 9085 }, { "epoch": 1.46, "grad_norm": 8.126753807067871, "learning_rate": 1.0888677531409365e-05, "loss": 0.8602, "step": 9086 }, { "epoch": 1.46, "grad_norm": 9.891738891601562, "learning_rate": 1.088694640340517e-05, "loss": 1.2028, "step": 9087 }, { "epoch": 1.46, "grad_norm": 9.650118827819824, "learning_rate": 1.0885215248609762e-05, "loss": 1.1745, "step": 9088 }, { "epoch": 1.46, "grad_norm": 9.672348022460938, "learning_rate": 1.0883484067075422e-05, "loss": 0.9967, "step": 9089 }, { "epoch": 1.46, "grad_norm": 9.204292297363281, "learning_rate": 1.0881752858854453e-05, "loss": 1.0207, "step": 9090 }, { "epoch": 1.46, "grad_norm": 10.449312210083008, "learning_rate": 1.088002162399914e-05, "loss": 1.2557, "step": 9091 }, { "epoch": 1.46, "grad_norm": 8.578234672546387, "learning_rate": 1.0878290362561784e-05, "loss": 1.2524, "step": 9092 }, { "epoch": 1.46, "grad_norm": 10.376391410827637, "learning_rate": 1.0876559074594671e-05, "loss": 1.0627, "step": 9093 }, { "epoch": 1.46, "grad_norm": 9.579155921936035, "learning_rate": 1.0874827760150101e-05, "loss": 1.0273, "step": 9094 }, { "epoch": 1.46, "grad_norm": 9.983850479125977, "learning_rate": 1.087309641928037e-05, "loss": 0.9529, "step": 9095 }, { "epoch": 1.46, "grad_norm": 10.197100639343262, "learning_rate": 1.087136505203778e-05, "loss": 0.9738, "step": 9096 }, { "epoch": 1.46, "grad_norm": 8.70720386505127, "learning_rate": 1.0869633658474621e-05, "loss": 1.1113, "step": 9097 }, { "epoch": 1.46, "grad_norm": 9.476698875427246, "learning_rate": 1.0867902238643195e-05, "loss": 1.1539, "step": 9098 }, { "epoch": 1.46, "grad_norm": 8.236330032348633, "learning_rate": 1.0866170792595806e-05, "loss": 0.991, "step": 9099 }, { "epoch": 1.46, "grad_norm": 9.490012168884277, "learning_rate": 1.0864439320384746e-05, "loss": 1.1196, "step": 9100 }, { "epoch": 1.47, "grad_norm": 7.778077125549316, "learning_rate": 1.0862707822062319e-05, "loss": 0.7079, "step": 9101 }, { "epoch": 1.47, "grad_norm": 8.53055191040039, "learning_rate": 1.086097629768083e-05, "loss": 1.0704, "step": 9102 }, { "epoch": 1.47, "grad_norm": 7.953604221343994, "learning_rate": 1.085924474729258e-05, "loss": 1.1367, "step": 9103 }, { "epoch": 1.47, "grad_norm": 8.000472068786621, "learning_rate": 1.0857513170949871e-05, "loss": 1.0856, "step": 9104 }, { "epoch": 1.47, "grad_norm": 8.3329439163208, "learning_rate": 1.0855781568705009e-05, "loss": 1.1827, "step": 9105 }, { "epoch": 1.47, "grad_norm": 8.946693420410156, "learning_rate": 1.0854049940610297e-05, "loss": 0.7645, "step": 9106 }, { "epoch": 1.47, "grad_norm": 9.993228912353516, "learning_rate": 1.0852318286718042e-05, "loss": 1.0027, "step": 9107 }, { "epoch": 1.47, "grad_norm": 11.661247253417969, "learning_rate": 1.085058660708055e-05, "loss": 1.0346, "step": 9108 }, { "epoch": 1.47, "grad_norm": 8.593469619750977, "learning_rate": 1.084885490175013e-05, "loss": 1.2521, "step": 9109 }, { "epoch": 1.47, "grad_norm": 8.842730522155762, "learning_rate": 1.0847123170779087e-05, "loss": 1.0575, "step": 9110 }, { "epoch": 1.47, "grad_norm": 9.36446762084961, "learning_rate": 1.0845391414219736e-05, "loss": 0.8378, "step": 9111 }, { "epoch": 1.47, "grad_norm": 8.509005546569824, "learning_rate": 1.0843659632124378e-05, "loss": 0.989, "step": 9112 }, { "epoch": 1.47, "grad_norm": 9.316180229187012, "learning_rate": 1.0841927824545331e-05, "loss": 1.2235, "step": 9113 }, { "epoch": 1.47, "grad_norm": 9.904093742370605, "learning_rate": 1.0840195991534903e-05, "loss": 1.1378, "step": 9114 }, { "epoch": 1.47, "grad_norm": 8.989542007446289, "learning_rate": 1.0838464133145405e-05, "loss": 0.9456, "step": 9115 }, { "epoch": 1.47, "grad_norm": 9.518021583557129, "learning_rate": 1.0836732249429151e-05, "loss": 0.9959, "step": 9116 }, { "epoch": 1.47, "grad_norm": 10.272656440734863, "learning_rate": 1.0835000340438453e-05, "loss": 1.1091, "step": 9117 }, { "epoch": 1.47, "grad_norm": 10.954307556152344, "learning_rate": 1.0833268406225628e-05, "loss": 1.276, "step": 9118 }, { "epoch": 1.47, "grad_norm": 12.138862609863281, "learning_rate": 1.0831536446842992e-05, "loss": 1.1271, "step": 9119 }, { "epoch": 1.47, "grad_norm": 8.351027488708496, "learning_rate": 1.0829804462342856e-05, "loss": 1.216, "step": 9120 }, { "epoch": 1.47, "grad_norm": 8.14403247833252, "learning_rate": 1.0828072452777537e-05, "loss": 1.031, "step": 9121 }, { "epoch": 1.47, "grad_norm": 9.932721138000488, "learning_rate": 1.082634041819936e-05, "loss": 0.9522, "step": 9122 }, { "epoch": 1.47, "grad_norm": 12.413127899169922, "learning_rate": 1.0824608358660633e-05, "loss": 1.2208, "step": 9123 }, { "epoch": 1.47, "grad_norm": 10.266237258911133, "learning_rate": 1.0822876274213681e-05, "loss": 1.0096, "step": 9124 }, { "epoch": 1.47, "grad_norm": 7.925777912139893, "learning_rate": 1.0821144164910822e-05, "loss": 1.1563, "step": 9125 }, { "epoch": 1.47, "grad_norm": 11.534676551818848, "learning_rate": 1.0819412030804372e-05, "loss": 1.1162, "step": 9126 }, { "epoch": 1.47, "grad_norm": 9.833671569824219, "learning_rate": 1.081767987194666e-05, "loss": 1.0906, "step": 9127 }, { "epoch": 1.47, "grad_norm": 9.089000701904297, "learning_rate": 1.0815947688390003e-05, "loss": 0.8353, "step": 9128 }, { "epoch": 1.47, "grad_norm": 9.321063041687012, "learning_rate": 1.0814215480186725e-05, "loss": 1.0453, "step": 9129 }, { "epoch": 1.47, "grad_norm": 7.866451740264893, "learning_rate": 1.0812483247389147e-05, "loss": 1.1397, "step": 9130 }, { "epoch": 1.47, "grad_norm": 11.73098087310791, "learning_rate": 1.0810750990049598e-05, "loss": 1.0217, "step": 9131 }, { "epoch": 1.47, "grad_norm": 12.605908393859863, "learning_rate": 1.0809018708220399e-05, "loss": 1.4302, "step": 9132 }, { "epoch": 1.47, "grad_norm": 11.978370666503906, "learning_rate": 1.0807286401953873e-05, "loss": 0.8996, "step": 9133 }, { "epoch": 1.47, "grad_norm": 9.959700584411621, "learning_rate": 1.0805554071302354e-05, "loss": 1.2648, "step": 9134 }, { "epoch": 1.47, "grad_norm": 8.643314361572266, "learning_rate": 1.0803821716318161e-05, "loss": 1.0088, "step": 9135 }, { "epoch": 1.47, "grad_norm": 9.88506031036377, "learning_rate": 1.0802089337053627e-05, "loss": 1.0257, "step": 9136 }, { "epoch": 1.47, "grad_norm": 8.589159965515137, "learning_rate": 1.0800356933561077e-05, "loss": 0.9842, "step": 9137 }, { "epoch": 1.47, "grad_norm": 8.337607383728027, "learning_rate": 1.0798624505892843e-05, "loss": 1.0736, "step": 9138 }, { "epoch": 1.47, "grad_norm": 10.46334171295166, "learning_rate": 1.0796892054101253e-05, "loss": 1.1412, "step": 9139 }, { "epoch": 1.47, "grad_norm": 9.319353103637695, "learning_rate": 1.0795159578238638e-05, "loss": 0.8233, "step": 9140 }, { "epoch": 1.47, "grad_norm": 12.031571388244629, "learning_rate": 1.0793427078357332e-05, "loss": 0.8022, "step": 9141 }, { "epoch": 1.47, "grad_norm": 9.429220199584961, "learning_rate": 1.0791694554509664e-05, "loss": 1.1182, "step": 9142 }, { "epoch": 1.47, "grad_norm": 8.030879020690918, "learning_rate": 1.0789962006747969e-05, "loss": 1.3109, "step": 9143 }, { "epoch": 1.47, "grad_norm": 10.633646965026855, "learning_rate": 1.0788229435124578e-05, "loss": 1.4163, "step": 9144 }, { "epoch": 1.47, "grad_norm": 9.72641372680664, "learning_rate": 1.0786496839691826e-05, "loss": 1.1051, "step": 9145 }, { "epoch": 1.47, "grad_norm": 10.429793357849121, "learning_rate": 1.0784764220502052e-05, "loss": 1.004, "step": 9146 }, { "epoch": 1.47, "grad_norm": 9.63525390625, "learning_rate": 1.0783031577607586e-05, "loss": 1.1404, "step": 9147 }, { "epoch": 1.47, "grad_norm": 10.304875373840332, "learning_rate": 1.0781298911060765e-05, "loss": 1.1312, "step": 9148 }, { "epoch": 1.47, "grad_norm": 8.424322128295898, "learning_rate": 1.0779566220913929e-05, "loss": 1.167, "step": 9149 }, { "epoch": 1.47, "grad_norm": 11.513104438781738, "learning_rate": 1.0777833507219418e-05, "loss": 1.1163, "step": 9150 }, { "epoch": 1.47, "grad_norm": 10.556303977966309, "learning_rate": 1.0776100770029563e-05, "loss": 1.0858, "step": 9151 }, { "epoch": 1.47, "grad_norm": 10.617037773132324, "learning_rate": 1.077436800939671e-05, "loss": 1.1517, "step": 9152 }, { "epoch": 1.47, "grad_norm": 9.611712455749512, "learning_rate": 1.0772635225373195e-05, "loss": 0.9395, "step": 9153 }, { "epoch": 1.47, "grad_norm": 10.298248291015625, "learning_rate": 1.0770902418011362e-05, "loss": 1.1386, "step": 9154 }, { "epoch": 1.47, "grad_norm": 9.358138084411621, "learning_rate": 1.076916958736355e-05, "loss": 1.01, "step": 9155 }, { "epoch": 1.47, "grad_norm": 8.704716682434082, "learning_rate": 1.0767436733482104e-05, "loss": 0.8862, "step": 9156 }, { "epoch": 1.47, "grad_norm": 8.223893165588379, "learning_rate": 1.0765703856419362e-05, "loss": 1.002, "step": 9157 }, { "epoch": 1.47, "grad_norm": 9.294219017028809, "learning_rate": 1.0763970956227671e-05, "loss": 1.0624, "step": 9158 }, { "epoch": 1.47, "grad_norm": 8.352044105529785, "learning_rate": 1.0762238032959374e-05, "loss": 1.0388, "step": 9159 }, { "epoch": 1.47, "grad_norm": 8.576701164245605, "learning_rate": 1.0760505086666817e-05, "loss": 0.9407, "step": 9160 }, { "epoch": 1.47, "grad_norm": 10.300621032714844, "learning_rate": 1.0758772117402346e-05, "loss": 1.0012, "step": 9161 }, { "epoch": 1.47, "grad_norm": 8.82954216003418, "learning_rate": 1.0757039125218304e-05, "loss": 1.1316, "step": 9162 }, { "epoch": 1.48, "grad_norm": 10.382974624633789, "learning_rate": 1.0755306110167043e-05, "loss": 1.2185, "step": 9163 }, { "epoch": 1.48, "grad_norm": 9.16775894165039, "learning_rate": 1.0753573072300905e-05, "loss": 1.0321, "step": 9164 }, { "epoch": 1.48, "grad_norm": 11.678339958190918, "learning_rate": 1.0751840011672243e-05, "loss": 1.0136, "step": 9165 }, { "epoch": 1.48, "grad_norm": 11.018516540527344, "learning_rate": 1.0750106928333406e-05, "loss": 1.0922, "step": 9166 }, { "epoch": 1.48, "grad_norm": 10.775525093078613, "learning_rate": 1.0748373822336738e-05, "loss": 1.144, "step": 9167 }, { "epoch": 1.48, "grad_norm": 9.906719207763672, "learning_rate": 1.0746640693734594e-05, "loss": 1.0448, "step": 9168 }, { "epoch": 1.48, "grad_norm": 7.983192443847656, "learning_rate": 1.0744907542579323e-05, "loss": 0.8525, "step": 9169 }, { "epoch": 1.48, "grad_norm": 10.057882308959961, "learning_rate": 1.0743174368923284e-05, "loss": 1.4177, "step": 9170 }, { "epoch": 1.48, "grad_norm": 9.8334321975708, "learning_rate": 1.074144117281882e-05, "loss": 1.1894, "step": 9171 }, { "epoch": 1.48, "grad_norm": 10.081474304199219, "learning_rate": 1.073970795431829e-05, "loss": 1.0599, "step": 9172 }, { "epoch": 1.48, "grad_norm": 8.045815467834473, "learning_rate": 1.0737974713474044e-05, "loss": 0.8962, "step": 9173 }, { "epoch": 1.48, "grad_norm": 8.503161430358887, "learning_rate": 1.073624145033844e-05, "loss": 0.9762, "step": 9174 }, { "epoch": 1.48, "grad_norm": 9.679357528686523, "learning_rate": 1.0734508164963831e-05, "loss": 0.9305, "step": 9175 }, { "epoch": 1.48, "grad_norm": 11.491013526916504, "learning_rate": 1.0732774857402572e-05, "loss": 0.9093, "step": 9176 }, { "epoch": 1.48, "grad_norm": 13.002850532531738, "learning_rate": 1.0731041527707022e-05, "loss": 1.1227, "step": 9177 }, { "epoch": 1.48, "grad_norm": 10.979693412780762, "learning_rate": 1.0729308175929535e-05, "loss": 0.9212, "step": 9178 }, { "epoch": 1.48, "grad_norm": 10.753897666931152, "learning_rate": 1.0727574802122474e-05, "loss": 1.1519, "step": 9179 }, { "epoch": 1.48, "grad_norm": 9.220229148864746, "learning_rate": 1.0725841406338192e-05, "loss": 0.9884, "step": 9180 }, { "epoch": 1.48, "grad_norm": 7.462979793548584, "learning_rate": 1.0724107988629048e-05, "loss": 1.0488, "step": 9181 }, { "epoch": 1.48, "grad_norm": 11.070197105407715, "learning_rate": 1.0722374549047412e-05, "loss": 1.0164, "step": 9182 }, { "epoch": 1.48, "grad_norm": 9.983266830444336, "learning_rate": 1.072064108764563e-05, "loss": 1.0084, "step": 9183 }, { "epoch": 1.48, "grad_norm": 10.964164733886719, "learning_rate": 1.0718907604476075e-05, "loss": 1.1891, "step": 9184 }, { "epoch": 1.48, "grad_norm": 10.318084716796875, "learning_rate": 1.0717174099591098e-05, "loss": 1.0569, "step": 9185 }, { "epoch": 1.48, "grad_norm": 8.195928573608398, "learning_rate": 1.0715440573043075e-05, "loss": 0.8928, "step": 9186 }, { "epoch": 1.48, "grad_norm": 9.908004760742188, "learning_rate": 1.0713707024884357e-05, "loss": 1.1502, "step": 9187 }, { "epoch": 1.48, "grad_norm": 9.971860885620117, "learning_rate": 1.0711973455167311e-05, "loss": 1.1011, "step": 9188 }, { "epoch": 1.48, "grad_norm": 7.973855018615723, "learning_rate": 1.0710239863944305e-05, "loss": 1.1559, "step": 9189 }, { "epoch": 1.48, "grad_norm": 10.592543601989746, "learning_rate": 1.0708506251267701e-05, "loss": 0.9887, "step": 9190 }, { "epoch": 1.48, "grad_norm": 9.887815475463867, "learning_rate": 1.0706772617189865e-05, "loss": 1.2557, "step": 9191 }, { "epoch": 1.48, "grad_norm": 7.778311729431152, "learning_rate": 1.0705038961763164e-05, "loss": 0.9786, "step": 9192 }, { "epoch": 1.48, "grad_norm": 11.08390998840332, "learning_rate": 1.0703305285039965e-05, "loss": 0.8545, "step": 9193 }, { "epoch": 1.48, "grad_norm": 9.915020942687988, "learning_rate": 1.0701571587072637e-05, "loss": 1.4121, "step": 9194 }, { "epoch": 1.48, "grad_norm": 8.0883207321167, "learning_rate": 1.0699837867913543e-05, "loss": 1.1263, "step": 9195 }, { "epoch": 1.48, "grad_norm": 9.673298835754395, "learning_rate": 1.0698104127615059e-05, "loss": 1.1689, "step": 9196 }, { "epoch": 1.48, "grad_norm": 8.499080657958984, "learning_rate": 1.069637036622955e-05, "loss": 0.9025, "step": 9197 }, { "epoch": 1.48, "grad_norm": 11.227961540222168, "learning_rate": 1.0694636583809388e-05, "loss": 1.1518, "step": 9198 }, { "epoch": 1.48, "grad_norm": 8.378768920898438, "learning_rate": 1.0692902780406943e-05, "loss": 0.9021, "step": 9199 }, { "epoch": 1.48, "grad_norm": 9.484610557556152, "learning_rate": 1.0691168956074586e-05, "loss": 1.0187, "step": 9200 }, { "epoch": 1.48, "grad_norm": 9.938895225524902, "learning_rate": 1.0689435110864694e-05, "loss": 1.1688, "step": 9201 }, { "epoch": 1.48, "grad_norm": 9.03451156616211, "learning_rate": 1.0687701244829634e-05, "loss": 0.9667, "step": 9202 }, { "epoch": 1.48, "grad_norm": 8.899949073791504, "learning_rate": 1.068596735802178e-05, "loss": 1.0588, "step": 9203 }, { "epoch": 1.48, "grad_norm": 12.072940826416016, "learning_rate": 1.0684233450493506e-05, "loss": 0.9494, "step": 9204 }, { "epoch": 1.48, "grad_norm": 10.860241889953613, "learning_rate": 1.0682499522297192e-05, "loss": 1.1048, "step": 9205 }, { "epoch": 1.48, "grad_norm": 7.700897216796875, "learning_rate": 1.0680765573485206e-05, "loss": 1.0514, "step": 9206 }, { "epoch": 1.48, "grad_norm": 10.027112007141113, "learning_rate": 1.067903160410993e-05, "loss": 1.236, "step": 9207 }, { "epoch": 1.48, "grad_norm": 8.612650871276855, "learning_rate": 1.067729761422373e-05, "loss": 1.2053, "step": 9208 }, { "epoch": 1.48, "grad_norm": 11.686474800109863, "learning_rate": 1.0675563603878998e-05, "loss": 1.2642, "step": 9209 }, { "epoch": 1.48, "grad_norm": 8.177781105041504, "learning_rate": 1.0673829573128103e-05, "loss": 0.8208, "step": 9210 }, { "epoch": 1.48, "grad_norm": 9.94293212890625, "learning_rate": 1.0672095522023422e-05, "loss": 1.2367, "step": 9211 }, { "epoch": 1.48, "grad_norm": 7.753229141235352, "learning_rate": 1.0670361450617339e-05, "loss": 0.9822, "step": 9212 }, { "epoch": 1.48, "grad_norm": 10.5778226852417, "learning_rate": 1.0668627358962231e-05, "loss": 1.0435, "step": 9213 }, { "epoch": 1.48, "grad_norm": 8.559431076049805, "learning_rate": 1.0666893247110479e-05, "loss": 0.9199, "step": 9214 }, { "epoch": 1.48, "grad_norm": 9.134481430053711, "learning_rate": 1.0665159115114461e-05, "loss": 0.8901, "step": 9215 }, { "epoch": 1.48, "grad_norm": 8.775625228881836, "learning_rate": 1.0663424963026563e-05, "loss": 0.884, "step": 9216 }, { "epoch": 1.48, "grad_norm": 7.569181442260742, "learning_rate": 1.0661690790899163e-05, "loss": 1.0966, "step": 9217 }, { "epoch": 1.48, "grad_norm": 9.971135139465332, "learning_rate": 1.0659956598784649e-05, "loss": 1.2468, "step": 9218 }, { "epoch": 1.48, "grad_norm": 8.432012557983398, "learning_rate": 1.0658222386735395e-05, "loss": 0.859, "step": 9219 }, { "epoch": 1.48, "grad_norm": 9.764556884765625, "learning_rate": 1.0656488154803793e-05, "loss": 1.0948, "step": 9220 }, { "epoch": 1.48, "grad_norm": 9.241677284240723, "learning_rate": 1.0654753903042225e-05, "loss": 1.1996, "step": 9221 }, { "epoch": 1.48, "grad_norm": 8.866830825805664, "learning_rate": 1.0653019631503076e-05, "loss": 0.9961, "step": 9222 }, { "epoch": 1.48, "grad_norm": 9.649445533752441, "learning_rate": 1.065128534023873e-05, "loss": 1.0084, "step": 9223 }, { "epoch": 1.48, "grad_norm": 8.738807678222656, "learning_rate": 1.0649551029301577e-05, "loss": 0.8581, "step": 9224 }, { "epoch": 1.49, "grad_norm": 8.58802318572998, "learning_rate": 1.0647816698744003e-05, "loss": 1.1113, "step": 9225 }, { "epoch": 1.49, "grad_norm": 8.218884468078613, "learning_rate": 1.064608234861839e-05, "loss": 0.8868, "step": 9226 }, { "epoch": 1.49, "grad_norm": 9.058073043823242, "learning_rate": 1.0644347978977132e-05, "loss": 0.8068, "step": 9227 }, { "epoch": 1.49, "grad_norm": 9.087472915649414, "learning_rate": 1.0642613589872614e-05, "loss": 1.2103, "step": 9228 }, { "epoch": 1.49, "grad_norm": 9.947583198547363, "learning_rate": 1.0640879181357226e-05, "loss": 0.9015, "step": 9229 }, { "epoch": 1.49, "grad_norm": 10.008271217346191, "learning_rate": 1.063914475348336e-05, "loss": 1.0399, "step": 9230 }, { "epoch": 1.49, "grad_norm": 7.469876289367676, "learning_rate": 1.0637410306303405e-05, "loss": 0.9609, "step": 9231 }, { "epoch": 1.49, "grad_norm": 10.029854774475098, "learning_rate": 1.0635675839869753e-05, "loss": 1.4934, "step": 9232 }, { "epoch": 1.49, "grad_norm": 8.070829391479492, "learning_rate": 1.0633941354234793e-05, "loss": 0.9473, "step": 9233 }, { "epoch": 1.49, "grad_norm": 10.33123779296875, "learning_rate": 1.0632206849450919e-05, "loss": 1.1742, "step": 9234 }, { "epoch": 1.49, "grad_norm": 9.726564407348633, "learning_rate": 1.0630472325570523e-05, "loss": 1.1159, "step": 9235 }, { "epoch": 1.49, "grad_norm": 7.810680866241455, "learning_rate": 1.0628737782645997e-05, "loss": 0.8842, "step": 9236 }, { "epoch": 1.49, "grad_norm": 10.594240188598633, "learning_rate": 1.0627003220729739e-05, "loss": 1.2011, "step": 9237 }, { "epoch": 1.49, "grad_norm": 9.724960327148438, "learning_rate": 1.0625268639874137e-05, "loss": 1.1688, "step": 9238 }, { "epoch": 1.49, "grad_norm": 10.829680442810059, "learning_rate": 1.062353404013159e-05, "loss": 1.0124, "step": 9239 }, { "epoch": 1.49, "grad_norm": 8.447861671447754, "learning_rate": 1.0621799421554496e-05, "loss": 0.8756, "step": 9240 }, { "epoch": 1.49, "grad_norm": 9.427242279052734, "learning_rate": 1.0620064784195246e-05, "loss": 1.1464, "step": 9241 }, { "epoch": 1.49, "grad_norm": 9.63206672668457, "learning_rate": 1.0618330128106237e-05, "loss": 0.8411, "step": 9242 }, { "epoch": 1.49, "grad_norm": 8.098657608032227, "learning_rate": 1.0616595453339872e-05, "loss": 0.9212, "step": 9243 }, { "epoch": 1.49, "grad_norm": 12.074040412902832, "learning_rate": 1.0614860759948544e-05, "loss": 1.0986, "step": 9244 }, { "epoch": 1.49, "grad_norm": 9.295522689819336, "learning_rate": 1.0613126047984651e-05, "loss": 1.149, "step": 9245 }, { "epoch": 1.49, "grad_norm": 9.029932022094727, "learning_rate": 1.0611391317500595e-05, "loss": 1.1399, "step": 9246 }, { "epoch": 1.49, "grad_norm": 9.531654357910156, "learning_rate": 1.0609656568548773e-05, "loss": 1.2473, "step": 9247 }, { "epoch": 1.49, "grad_norm": 9.528849601745605, "learning_rate": 1.0607921801181585e-05, "loss": 0.9491, "step": 9248 }, { "epoch": 1.49, "grad_norm": 8.743585586547852, "learning_rate": 1.0606187015451432e-05, "loss": 0.9568, "step": 9249 }, { "epoch": 1.49, "grad_norm": 9.67586898803711, "learning_rate": 1.0604452211410716e-05, "loss": 0.9426, "step": 9250 }, { "epoch": 1.49, "grad_norm": 8.657380104064941, "learning_rate": 1.060271738911184e-05, "loss": 1.2524, "step": 9251 }, { "epoch": 1.49, "grad_norm": 8.205078125, "learning_rate": 1.0600982548607204e-05, "loss": 1.057, "step": 9252 }, { "epoch": 1.49, "grad_norm": 9.18691349029541, "learning_rate": 1.059924768994921e-05, "loss": 1.1383, "step": 9253 }, { "epoch": 1.49, "grad_norm": 11.360904693603516, "learning_rate": 1.0597512813190263e-05, "loss": 0.994, "step": 9254 }, { "epoch": 1.49, "grad_norm": 9.006814002990723, "learning_rate": 1.0595777918382766e-05, "loss": 1.2222, "step": 9255 }, { "epoch": 1.49, "grad_norm": 8.57945728302002, "learning_rate": 1.0594043005579126e-05, "loss": 1.1387, "step": 9256 }, { "epoch": 1.49, "grad_norm": 9.569820404052734, "learning_rate": 1.0592308074831744e-05, "loss": 1.3454, "step": 9257 }, { "epoch": 1.49, "grad_norm": 9.047025680541992, "learning_rate": 1.0590573126193031e-05, "loss": 1.1124, "step": 9258 }, { "epoch": 1.49, "grad_norm": 9.26516056060791, "learning_rate": 1.0588838159715384e-05, "loss": 1.1455, "step": 9259 }, { "epoch": 1.49, "grad_norm": 7.819277286529541, "learning_rate": 1.058710317545122e-05, "loss": 0.8953, "step": 9260 }, { "epoch": 1.49, "grad_norm": 7.727189064025879, "learning_rate": 1.0585368173452942e-05, "loss": 1.1996, "step": 9261 }, { "epoch": 1.49, "grad_norm": 9.821616172790527, "learning_rate": 1.0583633153772955e-05, "loss": 0.8047, "step": 9262 }, { "epoch": 1.49, "grad_norm": 9.750168800354004, "learning_rate": 1.0581898116463666e-05, "loss": 1.0037, "step": 9263 }, { "epoch": 1.49, "grad_norm": 9.825235366821289, "learning_rate": 1.0580163061577496e-05, "loss": 1.0152, "step": 9264 }, { "epoch": 1.49, "grad_norm": 9.241201400756836, "learning_rate": 1.057842798916684e-05, "loss": 1.0825, "step": 9265 }, { "epoch": 1.49, "grad_norm": 7.768532752990723, "learning_rate": 1.0576692899284114e-05, "loss": 1.2904, "step": 9266 }, { "epoch": 1.49, "grad_norm": 11.100035667419434, "learning_rate": 1.0574957791981727e-05, "loss": 1.3535, "step": 9267 }, { "epoch": 1.49, "grad_norm": 10.124248504638672, "learning_rate": 1.0573222667312091e-05, "loss": 1.3098, "step": 9268 }, { "epoch": 1.49, "grad_norm": 8.950118064880371, "learning_rate": 1.0571487525327614e-05, "loss": 1.2562, "step": 9269 }, { "epoch": 1.49, "grad_norm": 10.003716468811035, "learning_rate": 1.0569752366080715e-05, "loss": 1.2105, "step": 9270 }, { "epoch": 1.49, "grad_norm": 9.90822982788086, "learning_rate": 1.05680171896238e-05, "loss": 1.0925, "step": 9271 }, { "epoch": 1.49, "grad_norm": 9.403013229370117, "learning_rate": 1.0566281996009287e-05, "loss": 1.2453, "step": 9272 }, { "epoch": 1.49, "grad_norm": 8.23205280303955, "learning_rate": 1.0564546785289584e-05, "loss": 1.0338, "step": 9273 }, { "epoch": 1.49, "grad_norm": 8.773736953735352, "learning_rate": 1.0562811557517108e-05, "loss": 1.1581, "step": 9274 }, { "epoch": 1.49, "grad_norm": 8.921112060546875, "learning_rate": 1.0561076312744273e-05, "loss": 0.9409, "step": 9275 }, { "epoch": 1.49, "grad_norm": 9.166671752929688, "learning_rate": 1.0559341051023497e-05, "loss": 1.0404, "step": 9276 }, { "epoch": 1.49, "grad_norm": 9.612711906433105, "learning_rate": 1.055760577240719e-05, "loss": 0.876, "step": 9277 }, { "epoch": 1.49, "grad_norm": 7.871773719787598, "learning_rate": 1.0555870476947774e-05, "loss": 1.0466, "step": 9278 }, { "epoch": 1.49, "grad_norm": 8.353067398071289, "learning_rate": 1.0554135164697659e-05, "loss": 1.2077, "step": 9279 }, { "epoch": 1.49, "grad_norm": 7.820023059844971, "learning_rate": 1.0552399835709267e-05, "loss": 1.1457, "step": 9280 }, { "epoch": 1.49, "grad_norm": 12.916105270385742, "learning_rate": 1.0550664490035015e-05, "loss": 1.231, "step": 9281 }, { "epoch": 1.49, "grad_norm": 8.956562995910645, "learning_rate": 1.0548929127727316e-05, "loss": 1.1112, "step": 9282 }, { "epoch": 1.49, "grad_norm": 8.33600902557373, "learning_rate": 1.0547193748838598e-05, "loss": 0.9391, "step": 9283 }, { "epoch": 1.49, "grad_norm": 9.174134254455566, "learning_rate": 1.0545458353421271e-05, "loss": 1.0527, "step": 9284 }, { "epoch": 1.49, "grad_norm": 9.35098934173584, "learning_rate": 1.054372294152776e-05, "loss": 1.1669, "step": 9285 }, { "epoch": 1.49, "grad_norm": 10.135420799255371, "learning_rate": 1.0541987513210486e-05, "loss": 1.0004, "step": 9286 }, { "epoch": 1.5, "grad_norm": 9.259611129760742, "learning_rate": 1.0540252068521864e-05, "loss": 0.8903, "step": 9287 }, { "epoch": 1.5, "grad_norm": 8.900341033935547, "learning_rate": 1.053851660751432e-05, "loss": 0.901, "step": 9288 }, { "epoch": 1.5, "grad_norm": 10.793084144592285, "learning_rate": 1.0536781130240267e-05, "loss": 1.0311, "step": 9289 }, { "epoch": 1.5, "grad_norm": 9.453390121459961, "learning_rate": 1.0535045636752139e-05, "loss": 1.2224, "step": 9290 }, { "epoch": 1.5, "grad_norm": 9.308440208435059, "learning_rate": 1.0533310127102356e-05, "loss": 1.4447, "step": 9291 }, { "epoch": 1.5, "grad_norm": 8.786527633666992, "learning_rate": 1.0531574601343334e-05, "loss": 0.8077, "step": 9292 }, { "epoch": 1.5, "grad_norm": 9.370739936828613, "learning_rate": 1.0529839059527502e-05, "loss": 1.1688, "step": 9293 }, { "epoch": 1.5, "grad_norm": 12.260124206542969, "learning_rate": 1.0528103501707283e-05, "loss": 1.3672, "step": 9294 }, { "epoch": 1.5, "grad_norm": 8.632808685302734, "learning_rate": 1.0526367927935103e-05, "loss": 1.0605, "step": 9295 }, { "epoch": 1.5, "grad_norm": 8.294246673583984, "learning_rate": 1.0524632338263383e-05, "loss": 1.148, "step": 9296 }, { "epoch": 1.5, "grad_norm": 9.332630157470703, "learning_rate": 1.0522896732744551e-05, "loss": 1.155, "step": 9297 }, { "epoch": 1.5, "grad_norm": 7.72158145904541, "learning_rate": 1.0521161111431033e-05, "loss": 0.9944, "step": 9298 }, { "epoch": 1.5, "grad_norm": 8.133183479309082, "learning_rate": 1.0519425474375254e-05, "loss": 1.1214, "step": 9299 }, { "epoch": 1.5, "grad_norm": 9.321603775024414, "learning_rate": 1.0517689821629642e-05, "loss": 1.0414, "step": 9300 }, { "epoch": 1.5, "grad_norm": 10.859755516052246, "learning_rate": 1.0515954153246625e-05, "loss": 1.0319, "step": 9301 }, { "epoch": 1.5, "grad_norm": 9.629032135009766, "learning_rate": 1.0514218469278629e-05, "loss": 1.1031, "step": 9302 }, { "epoch": 1.5, "grad_norm": 9.142333030700684, "learning_rate": 1.0512482769778088e-05, "loss": 1.0058, "step": 9303 }, { "epoch": 1.5, "grad_norm": 8.694067001342773, "learning_rate": 1.051074705479742e-05, "loss": 0.9347, "step": 9304 }, { "epoch": 1.5, "grad_norm": 9.44105339050293, "learning_rate": 1.0509011324389064e-05, "loss": 0.8985, "step": 9305 }, { "epoch": 1.5, "grad_norm": 9.570385932922363, "learning_rate": 1.0507275578605445e-05, "loss": 0.7706, "step": 9306 }, { "epoch": 1.5, "grad_norm": 10.981972694396973, "learning_rate": 1.0505539817498995e-05, "loss": 1.1543, "step": 9307 }, { "epoch": 1.5, "grad_norm": 8.863058090209961, "learning_rate": 1.0503804041122144e-05, "loss": 0.7727, "step": 9308 }, { "epoch": 1.5, "grad_norm": 9.568235397338867, "learning_rate": 1.0502068249527319e-05, "loss": 1.1469, "step": 9309 }, { "epoch": 1.5, "grad_norm": 10.084925651550293, "learning_rate": 1.0500332442766957e-05, "loss": 0.9858, "step": 9310 }, { "epoch": 1.5, "grad_norm": 9.288152694702148, "learning_rate": 1.049859662089349e-05, "loss": 0.9684, "step": 9311 }, { "epoch": 1.5, "grad_norm": 10.047151565551758, "learning_rate": 1.049686078395935e-05, "loss": 1.21, "step": 9312 }, { "epoch": 1.5, "grad_norm": 10.13843822479248, "learning_rate": 1.0495124932016967e-05, "loss": 1.137, "step": 9313 }, { "epoch": 1.5, "grad_norm": 10.171976089477539, "learning_rate": 1.0493389065118777e-05, "loss": 1.0958, "step": 9314 }, { "epoch": 1.5, "grad_norm": 10.117730140686035, "learning_rate": 1.0491653183317215e-05, "loss": 1.2375, "step": 9315 }, { "epoch": 1.5, "grad_norm": 10.088691711425781, "learning_rate": 1.048991728666471e-05, "loss": 0.9593, "step": 9316 }, { "epoch": 1.5, "grad_norm": 9.084636688232422, "learning_rate": 1.0488181375213702e-05, "loss": 1.041, "step": 9317 }, { "epoch": 1.5, "grad_norm": 9.356351852416992, "learning_rate": 1.0486445449016622e-05, "loss": 1.3137, "step": 9318 }, { "epoch": 1.5, "grad_norm": 9.751383781433105, "learning_rate": 1.0484709508125911e-05, "loss": 1.3405, "step": 9319 }, { "epoch": 1.5, "grad_norm": 9.437201499938965, "learning_rate": 1.0482973552593997e-05, "loss": 0.9968, "step": 9320 }, { "epoch": 1.5, "grad_norm": 9.647934913635254, "learning_rate": 1.0481237582473324e-05, "loss": 1.1192, "step": 9321 }, { "epoch": 1.5, "grad_norm": 9.67324447631836, "learning_rate": 1.0479501597816331e-05, "loss": 1.1376, "step": 9322 }, { "epoch": 1.5, "grad_norm": 11.773865699768066, "learning_rate": 1.0477765598675444e-05, "loss": 1.1399, "step": 9323 }, { "epoch": 1.5, "grad_norm": 7.975586414337158, "learning_rate": 1.047602958510311e-05, "loss": 0.9491, "step": 9324 }, { "epoch": 1.5, "grad_norm": 9.152156829833984, "learning_rate": 1.0474293557151764e-05, "loss": 0.9791, "step": 9325 }, { "epoch": 1.5, "grad_norm": 10.046436309814453, "learning_rate": 1.0472557514873847e-05, "loss": 1.1359, "step": 9326 }, { "epoch": 1.5, "grad_norm": 10.872398376464844, "learning_rate": 1.0470821458321797e-05, "loss": 1.0234, "step": 9327 }, { "epoch": 1.5, "grad_norm": 9.181329727172852, "learning_rate": 1.046908538754805e-05, "loss": 1.1987, "step": 9328 }, { "epoch": 1.5, "grad_norm": 10.192960739135742, "learning_rate": 1.0467349302605053e-05, "loss": 1.0535, "step": 9329 }, { "epoch": 1.5, "grad_norm": 9.141129493713379, "learning_rate": 1.0465613203545237e-05, "loss": 1.1352, "step": 9330 }, { "epoch": 1.5, "grad_norm": 10.33355712890625, "learning_rate": 1.0463877090421054e-05, "loss": 1.1977, "step": 9331 }, { "epoch": 1.5, "grad_norm": 10.261902809143066, "learning_rate": 1.0462140963284935e-05, "loss": 1.4668, "step": 9332 }, { "epoch": 1.5, "grad_norm": 8.963512420654297, "learning_rate": 1.0460404822189327e-05, "loss": 0.8925, "step": 9333 }, { "epoch": 1.5, "grad_norm": 11.237618446350098, "learning_rate": 1.0458668667186676e-05, "loss": 1.293, "step": 9334 }, { "epoch": 1.5, "grad_norm": 12.188799858093262, "learning_rate": 1.0456932498329416e-05, "loss": 1.0292, "step": 9335 }, { "epoch": 1.5, "grad_norm": 8.952702522277832, "learning_rate": 1.0455196315669993e-05, "loss": 1.0932, "step": 9336 }, { "epoch": 1.5, "grad_norm": 10.616881370544434, "learning_rate": 1.0453460119260851e-05, "loss": 1.0951, "step": 9337 }, { "epoch": 1.5, "grad_norm": 7.947702884674072, "learning_rate": 1.0451723909154437e-05, "loss": 0.8994, "step": 9338 }, { "epoch": 1.5, "grad_norm": 9.188767433166504, "learning_rate": 1.0449987685403194e-05, "loss": 1.2322, "step": 9339 }, { "epoch": 1.5, "grad_norm": 10.083270072937012, "learning_rate": 1.0448251448059557e-05, "loss": 1.1696, "step": 9340 }, { "epoch": 1.5, "grad_norm": 10.159248352050781, "learning_rate": 1.0446515197175984e-05, "loss": 1.0422, "step": 9341 }, { "epoch": 1.5, "grad_norm": 9.065364837646484, "learning_rate": 1.0444778932804915e-05, "loss": 1.0961, "step": 9342 }, { "epoch": 1.5, "grad_norm": 9.518685340881348, "learning_rate": 1.0443042654998796e-05, "loss": 0.8152, "step": 9343 }, { "epoch": 1.5, "grad_norm": 9.522418022155762, "learning_rate": 1.044130636381007e-05, "loss": 1.1928, "step": 9344 }, { "epoch": 1.5, "grad_norm": 9.284378051757812, "learning_rate": 1.043957005929119e-05, "loss": 1.1372, "step": 9345 }, { "epoch": 1.5, "grad_norm": 10.34170150756836, "learning_rate": 1.0437833741494601e-05, "loss": 1.0945, "step": 9346 }, { "epoch": 1.5, "grad_norm": 10.662858009338379, "learning_rate": 1.0436097410472747e-05, "loss": 1.2392, "step": 9347 }, { "epoch": 1.5, "grad_norm": 10.129921913146973, "learning_rate": 1.0434361066278078e-05, "loss": 1.1878, "step": 9348 }, { "epoch": 1.5, "grad_norm": 8.490201950073242, "learning_rate": 1.0432624708963044e-05, "loss": 1.0675, "step": 9349 }, { "epoch": 1.51, "grad_norm": 8.019257545471191, "learning_rate": 1.043088833858009e-05, "loss": 0.8115, "step": 9350 }, { "epoch": 1.51, "grad_norm": 10.071760177612305, "learning_rate": 1.042915195518167e-05, "loss": 0.9855, "step": 9351 }, { "epoch": 1.51, "grad_norm": 9.114986419677734, "learning_rate": 1.0427415558820226e-05, "loss": 1.1882, "step": 9352 }, { "epoch": 1.51, "grad_norm": 9.287166595458984, "learning_rate": 1.0425679149548215e-05, "loss": 0.9857, "step": 9353 }, { "epoch": 1.51, "grad_norm": 9.955830574035645, "learning_rate": 1.0423942727418089e-05, "loss": 1.1123, "step": 9354 }, { "epoch": 1.51, "grad_norm": 11.993186950683594, "learning_rate": 1.0422206292482286e-05, "loss": 1.0883, "step": 9355 }, { "epoch": 1.51, "grad_norm": 8.869826316833496, "learning_rate": 1.042046984479327e-05, "loss": 0.911, "step": 9356 }, { "epoch": 1.51, "grad_norm": 8.327558517456055, "learning_rate": 1.0418733384403485e-05, "loss": 1.1808, "step": 9357 }, { "epoch": 1.51, "grad_norm": 9.636103630065918, "learning_rate": 1.0416996911365389e-05, "loss": 1.1205, "step": 9358 }, { "epoch": 1.51, "grad_norm": 8.526098251342773, "learning_rate": 1.0415260425731428e-05, "loss": 1.1569, "step": 9359 }, { "epoch": 1.51, "grad_norm": 10.73121452331543, "learning_rate": 1.0413523927554055e-05, "loss": 1.0663, "step": 9360 }, { "epoch": 1.51, "grad_norm": 10.112167358398438, "learning_rate": 1.0411787416885725e-05, "loss": 1.0899, "step": 9361 }, { "epoch": 1.51, "grad_norm": 9.255352020263672, "learning_rate": 1.0410050893778894e-05, "loss": 0.944, "step": 9362 }, { "epoch": 1.51, "grad_norm": 8.65122127532959, "learning_rate": 1.040831435828601e-05, "loss": 1.1911, "step": 9363 }, { "epoch": 1.51, "grad_norm": 7.619572639465332, "learning_rate": 1.040657781045953e-05, "loss": 1.1279, "step": 9364 }, { "epoch": 1.51, "grad_norm": 8.089902877807617, "learning_rate": 1.0404841250351909e-05, "loss": 1.0322, "step": 9365 }, { "epoch": 1.51, "grad_norm": 9.138777732849121, "learning_rate": 1.0403104678015601e-05, "loss": 1.0127, "step": 9366 }, { "epoch": 1.51, "grad_norm": 8.694751739501953, "learning_rate": 1.0401368093503062e-05, "loss": 0.9764, "step": 9367 }, { "epoch": 1.51, "grad_norm": 11.540404319763184, "learning_rate": 1.0399631496866743e-05, "loss": 1.3552, "step": 9368 }, { "epoch": 1.51, "grad_norm": 8.764853477478027, "learning_rate": 1.0397894888159105e-05, "loss": 1.2177, "step": 9369 }, { "epoch": 1.51, "grad_norm": 9.643059730529785, "learning_rate": 1.0396158267432603e-05, "loss": 0.889, "step": 9370 }, { "epoch": 1.51, "grad_norm": 8.94125747680664, "learning_rate": 1.0394421634739691e-05, "loss": 0.8402, "step": 9371 }, { "epoch": 1.51, "grad_norm": 9.811589241027832, "learning_rate": 1.0392684990132829e-05, "loss": 0.8791, "step": 9372 }, { "epoch": 1.51, "grad_norm": 10.530486106872559, "learning_rate": 1.0390948333664476e-05, "loss": 0.8354, "step": 9373 }, { "epoch": 1.51, "grad_norm": 8.804139137268066, "learning_rate": 1.0389211665387085e-05, "loss": 0.9986, "step": 9374 }, { "epoch": 1.51, "grad_norm": 10.295344352722168, "learning_rate": 1.0387474985353116e-05, "loss": 1.2294, "step": 9375 }, { "epoch": 1.51, "grad_norm": 8.420621871948242, "learning_rate": 1.0385738293615027e-05, "loss": 1.1131, "step": 9376 }, { "epoch": 1.51, "grad_norm": 8.827919960021973, "learning_rate": 1.0384001590225277e-05, "loss": 1.0968, "step": 9377 }, { "epoch": 1.51, "grad_norm": 10.33518123626709, "learning_rate": 1.0382264875236328e-05, "loss": 1.1547, "step": 9378 }, { "epoch": 1.51, "grad_norm": 10.598359107971191, "learning_rate": 1.0380528148700635e-05, "loss": 1.0638, "step": 9379 }, { "epoch": 1.51, "grad_norm": 10.035212516784668, "learning_rate": 1.0378791410670658e-05, "loss": 1.0292, "step": 9380 }, { "epoch": 1.51, "grad_norm": 9.595609664916992, "learning_rate": 1.037705466119886e-05, "loss": 1.2057, "step": 9381 }, { "epoch": 1.51, "grad_norm": 11.633925437927246, "learning_rate": 1.03753179003377e-05, "loss": 0.9275, "step": 9382 }, { "epoch": 1.51, "grad_norm": 11.69899845123291, "learning_rate": 1.0373581128139639e-05, "loss": 0.9727, "step": 9383 }, { "epoch": 1.51, "grad_norm": 10.575268745422363, "learning_rate": 1.0371844344657135e-05, "loss": 1.021, "step": 9384 }, { "epoch": 1.51, "grad_norm": 11.177937507629395, "learning_rate": 1.0370107549942658e-05, "loss": 1.1144, "step": 9385 }, { "epoch": 1.51, "grad_norm": 10.342194557189941, "learning_rate": 1.0368370744048664e-05, "loss": 1.0428, "step": 9386 }, { "epoch": 1.51, "grad_norm": 10.098932266235352, "learning_rate": 1.0366633927027611e-05, "loss": 1.2548, "step": 9387 }, { "epoch": 1.51, "grad_norm": 9.52348518371582, "learning_rate": 1.0364897098931969e-05, "loss": 0.9885, "step": 9388 }, { "epoch": 1.51, "grad_norm": 10.348405838012695, "learning_rate": 1.03631602598142e-05, "loss": 1.2995, "step": 9389 }, { "epoch": 1.51, "grad_norm": 10.946537017822266, "learning_rate": 1.0361423409726764e-05, "loss": 1.1562, "step": 9390 }, { "epoch": 1.51, "grad_norm": 9.886399269104004, "learning_rate": 1.0359686548722121e-05, "loss": 0.9422, "step": 9391 }, { "epoch": 1.51, "grad_norm": 10.450546264648438, "learning_rate": 1.0357949676852743e-05, "loss": 1.3705, "step": 9392 }, { "epoch": 1.51, "grad_norm": 8.489690780639648, "learning_rate": 1.0356212794171092e-05, "loss": 1.1284, "step": 9393 }, { "epoch": 1.51, "grad_norm": 9.413919448852539, "learning_rate": 1.035447590072963e-05, "loss": 1.2138, "step": 9394 }, { "epoch": 1.51, "grad_norm": 8.27306842803955, "learning_rate": 1.0352738996580823e-05, "loss": 1.0096, "step": 9395 }, { "epoch": 1.51, "grad_norm": 9.813817977905273, "learning_rate": 1.0351002081777137e-05, "loss": 1.0659, "step": 9396 }, { "epoch": 1.51, "grad_norm": 10.267542839050293, "learning_rate": 1.0349265156371037e-05, "loss": 1.0134, "step": 9397 }, { "epoch": 1.51, "grad_norm": 7.333067417144775, "learning_rate": 1.0347528220414987e-05, "loss": 0.9402, "step": 9398 }, { "epoch": 1.51, "grad_norm": 9.138651847839355, "learning_rate": 1.0345791273961457e-05, "loss": 0.9704, "step": 9399 }, { "epoch": 1.51, "grad_norm": 9.67267894744873, "learning_rate": 1.0344054317062908e-05, "loss": 0.9119, "step": 9400 }, { "epoch": 1.51, "grad_norm": 7.826971530914307, "learning_rate": 1.0342317349771809e-05, "loss": 0.8395, "step": 9401 }, { "epoch": 1.51, "grad_norm": 10.395764350891113, "learning_rate": 1.034058037214063e-05, "loss": 1.2748, "step": 9402 }, { "epoch": 1.51, "grad_norm": 9.248342514038086, "learning_rate": 1.0338843384221835e-05, "loss": 1.2067, "step": 9403 }, { "epoch": 1.51, "grad_norm": 9.317780494689941, "learning_rate": 1.0337106386067895e-05, "loss": 0.9426, "step": 9404 }, { "epoch": 1.51, "grad_norm": 11.825797080993652, "learning_rate": 1.0335369377731276e-05, "loss": 1.0687, "step": 9405 }, { "epoch": 1.51, "grad_norm": 8.541487693786621, "learning_rate": 1.0333632359264445e-05, "loss": 0.9541, "step": 9406 }, { "epoch": 1.51, "grad_norm": 8.586686134338379, "learning_rate": 1.033189533071987e-05, "loss": 0.8969, "step": 9407 }, { "epoch": 1.51, "grad_norm": 10.79940414428711, "learning_rate": 1.0330158292150024e-05, "loss": 1.3905, "step": 9408 }, { "epoch": 1.51, "grad_norm": 7.815842628479004, "learning_rate": 1.0328421243607375e-05, "loss": 0.8692, "step": 9409 }, { "epoch": 1.51, "grad_norm": 10.656755447387695, "learning_rate": 1.0326684185144389e-05, "loss": 1.3146, "step": 9410 }, { "epoch": 1.51, "grad_norm": 10.568086624145508, "learning_rate": 1.0324947116813538e-05, "loss": 1.1973, "step": 9411 }, { "epoch": 1.52, "grad_norm": 8.062192916870117, "learning_rate": 1.0323210038667293e-05, "loss": 0.8588, "step": 9412 }, { "epoch": 1.52, "grad_norm": 10.877677917480469, "learning_rate": 1.0321472950758125e-05, "loss": 1.3425, "step": 9413 }, { "epoch": 1.52, "grad_norm": 8.438748359680176, "learning_rate": 1.0319735853138501e-05, "loss": 1.1227, "step": 9414 }, { "epoch": 1.52, "grad_norm": 10.043001174926758, "learning_rate": 1.0317998745860896e-05, "loss": 1.275, "step": 9415 }, { "epoch": 1.52, "grad_norm": 8.352615356445312, "learning_rate": 1.031626162897778e-05, "loss": 1.1196, "step": 9416 }, { "epoch": 1.52, "grad_norm": 8.750296592712402, "learning_rate": 1.0314524502541625e-05, "loss": 1.2128, "step": 9417 }, { "epoch": 1.52, "grad_norm": 9.304763793945312, "learning_rate": 1.03127873666049e-05, "loss": 0.9696, "step": 9418 }, { "epoch": 1.52, "grad_norm": 8.311053276062012, "learning_rate": 1.031105022122008e-05, "loss": 0.9239, "step": 9419 }, { "epoch": 1.52, "grad_norm": 9.709759712219238, "learning_rate": 1.0309313066439637e-05, "loss": 0.9493, "step": 9420 }, { "epoch": 1.52, "grad_norm": 10.752945899963379, "learning_rate": 1.0307575902316044e-05, "loss": 1.0377, "step": 9421 }, { "epoch": 1.52, "grad_norm": 8.198071479797363, "learning_rate": 1.0305838728901773e-05, "loss": 1.2228, "step": 9422 }, { "epoch": 1.52, "grad_norm": 11.013042449951172, "learning_rate": 1.0304101546249296e-05, "loss": 1.186, "step": 9423 }, { "epoch": 1.52, "grad_norm": 8.492866516113281, "learning_rate": 1.0302364354411094e-05, "loss": 1.1301, "step": 9424 }, { "epoch": 1.52, "grad_norm": 9.185976028442383, "learning_rate": 1.030062715343963e-05, "loss": 0.8357, "step": 9425 }, { "epoch": 1.52, "grad_norm": 10.057279586791992, "learning_rate": 1.0298889943387384e-05, "loss": 1.223, "step": 9426 }, { "epoch": 1.52, "grad_norm": 10.156328201293945, "learning_rate": 1.0297152724306832e-05, "loss": 1.0369, "step": 9427 }, { "epoch": 1.52, "grad_norm": 8.15373420715332, "learning_rate": 1.0295415496250445e-05, "loss": 0.9678, "step": 9428 }, { "epoch": 1.52, "grad_norm": 9.462176322937012, "learning_rate": 1.02936782592707e-05, "loss": 0.9916, "step": 9429 }, { "epoch": 1.52, "grad_norm": 9.950915336608887, "learning_rate": 1.0291941013420067e-05, "loss": 0.897, "step": 9430 }, { "epoch": 1.52, "grad_norm": 7.214657306671143, "learning_rate": 1.029020375875103e-05, "loss": 1.0481, "step": 9431 }, { "epoch": 1.52, "grad_norm": 9.54089641571045, "learning_rate": 1.028846649531606e-05, "loss": 0.9239, "step": 9432 }, { "epoch": 1.52, "grad_norm": 11.485329627990723, "learning_rate": 1.0286729223167635e-05, "loss": 1.1449, "step": 9433 }, { "epoch": 1.52, "grad_norm": 11.4404296875, "learning_rate": 1.0284991942358226e-05, "loss": 1.3243, "step": 9434 }, { "epoch": 1.52, "grad_norm": 8.983026504516602, "learning_rate": 1.0283254652940317e-05, "loss": 0.8439, "step": 9435 }, { "epoch": 1.52, "grad_norm": 10.98844051361084, "learning_rate": 1.0281517354966383e-05, "loss": 1.0272, "step": 9436 }, { "epoch": 1.52, "grad_norm": 9.154129981994629, "learning_rate": 1.0279780048488897e-05, "loss": 0.9893, "step": 9437 }, { "epoch": 1.52, "grad_norm": 12.317439079284668, "learning_rate": 1.0278042733560338e-05, "loss": 1.2428, "step": 9438 }, { "epoch": 1.52, "grad_norm": 12.032211303710938, "learning_rate": 1.0276305410233185e-05, "loss": 1.0339, "step": 9439 }, { "epoch": 1.52, "grad_norm": 11.576085090637207, "learning_rate": 1.0274568078559916e-05, "loss": 1.2812, "step": 9440 }, { "epoch": 1.52, "grad_norm": 8.4525785446167, "learning_rate": 1.0272830738593004e-05, "loss": 1.0649, "step": 9441 }, { "epoch": 1.52, "grad_norm": 12.5263032913208, "learning_rate": 1.0271093390384934e-05, "loss": 1.2669, "step": 9442 }, { "epoch": 1.52, "grad_norm": 11.574996948242188, "learning_rate": 1.0269356033988183e-05, "loss": 0.9499, "step": 9443 }, { "epoch": 1.52, "grad_norm": 10.367234230041504, "learning_rate": 1.026761866945523e-05, "loss": 1.2699, "step": 9444 }, { "epoch": 1.52, "grad_norm": 8.429825782775879, "learning_rate": 1.0265881296838548e-05, "loss": 1.1221, "step": 9445 }, { "epoch": 1.52, "grad_norm": 9.281919479370117, "learning_rate": 1.0264143916190626e-05, "loss": 0.9247, "step": 9446 }, { "epoch": 1.52, "grad_norm": 10.389471054077148, "learning_rate": 1.0262406527563936e-05, "loss": 1.1723, "step": 9447 }, { "epoch": 1.52, "grad_norm": 9.550578117370605, "learning_rate": 1.0260669131010964e-05, "loss": 1.2414, "step": 9448 }, { "epoch": 1.52, "grad_norm": 8.735834121704102, "learning_rate": 1.0258931726584182e-05, "loss": 1.0876, "step": 9449 }, { "epoch": 1.52, "grad_norm": 9.544821739196777, "learning_rate": 1.0257194314336076e-05, "loss": 0.9558, "step": 9450 }, { "epoch": 1.52, "grad_norm": 9.699499130249023, "learning_rate": 1.0255456894319124e-05, "loss": 1.0858, "step": 9451 }, { "epoch": 1.52, "grad_norm": 7.828131675720215, "learning_rate": 1.025371946658581e-05, "loss": 0.9062, "step": 9452 }, { "epoch": 1.52, "grad_norm": 11.204119682312012, "learning_rate": 1.0251982031188612e-05, "loss": 1.2122, "step": 9453 }, { "epoch": 1.52, "grad_norm": 9.060705184936523, "learning_rate": 1.0250244588180013e-05, "loss": 1.1182, "step": 9454 }, { "epoch": 1.52, "grad_norm": 9.11817455291748, "learning_rate": 1.0248507137612494e-05, "loss": 0.9813, "step": 9455 }, { "epoch": 1.52, "grad_norm": 10.626602172851562, "learning_rate": 1.0246769679538538e-05, "loss": 0.9523, "step": 9456 }, { "epoch": 1.52, "grad_norm": 9.840293884277344, "learning_rate": 1.0245032214010621e-05, "loss": 1.16, "step": 9457 }, { "epoch": 1.52, "grad_norm": 7.158971309661865, "learning_rate": 1.024329474108123e-05, "loss": 0.8542, "step": 9458 }, { "epoch": 1.52, "grad_norm": 9.601316452026367, "learning_rate": 1.0241557260802848e-05, "loss": 1.1881, "step": 9459 }, { "epoch": 1.52, "grad_norm": 11.25939655303955, "learning_rate": 1.0239819773227956e-05, "loss": 0.9122, "step": 9460 }, { "epoch": 1.52, "grad_norm": 9.721025466918945, "learning_rate": 1.0238082278409039e-05, "loss": 1.3136, "step": 9461 }, { "epoch": 1.52, "grad_norm": 7.7837934494018555, "learning_rate": 1.0236344776398571e-05, "loss": 1.1001, "step": 9462 }, { "epoch": 1.52, "grad_norm": 9.928828239440918, "learning_rate": 1.0234607267249048e-05, "loss": 1.0135, "step": 9463 }, { "epoch": 1.52, "grad_norm": 9.875723838806152, "learning_rate": 1.0232869751012945e-05, "loss": 1.1661, "step": 9464 }, { "epoch": 1.52, "grad_norm": 8.997498512268066, "learning_rate": 1.023113222774275e-05, "loss": 0.9684, "step": 9465 }, { "epoch": 1.52, "grad_norm": 8.838376998901367, "learning_rate": 1.0229394697490945e-05, "loss": 0.8005, "step": 9466 }, { "epoch": 1.52, "grad_norm": 9.617403030395508, "learning_rate": 1.0227657160310014e-05, "loss": 1.2114, "step": 9467 }, { "epoch": 1.52, "grad_norm": 8.969348907470703, "learning_rate": 1.022591961625244e-05, "loss": 1.0273, "step": 9468 }, { "epoch": 1.52, "grad_norm": 12.23903751373291, "learning_rate": 1.022418206537071e-05, "loss": 1.1803, "step": 9469 }, { "epoch": 1.52, "grad_norm": 8.281360626220703, "learning_rate": 1.0222444507717306e-05, "loss": 0.965, "step": 9470 }, { "epoch": 1.52, "grad_norm": 8.998023986816406, "learning_rate": 1.0220706943344713e-05, "loss": 1.2147, "step": 9471 }, { "epoch": 1.52, "grad_norm": 7.803172588348389, "learning_rate": 1.021896937230542e-05, "loss": 1.0173, "step": 9472 }, { "epoch": 1.52, "grad_norm": 10.924809455871582, "learning_rate": 1.0217231794651909e-05, "loss": 1.2138, "step": 9473 }, { "epoch": 1.53, "grad_norm": 7.529858112335205, "learning_rate": 1.0215494210436666e-05, "loss": 1.0343, "step": 9474 }, { "epoch": 1.53, "grad_norm": 9.217971801757812, "learning_rate": 1.0213756619712181e-05, "loss": 1.0004, "step": 9475 }, { "epoch": 1.53, "grad_norm": 8.084440231323242, "learning_rate": 1.0212019022530933e-05, "loss": 1.2314, "step": 9476 }, { "epoch": 1.53, "grad_norm": 9.590014457702637, "learning_rate": 1.0210281418945409e-05, "loss": 1.2521, "step": 9477 }, { "epoch": 1.53, "grad_norm": 8.55125904083252, "learning_rate": 1.02085438090081e-05, "loss": 0.7737, "step": 9478 }, { "epoch": 1.53, "grad_norm": 9.030948638916016, "learning_rate": 1.020680619277149e-05, "loss": 1.1877, "step": 9479 }, { "epoch": 1.53, "grad_norm": 8.032390594482422, "learning_rate": 1.0205068570288063e-05, "loss": 0.8626, "step": 9480 }, { "epoch": 1.53, "grad_norm": 10.981328964233398, "learning_rate": 1.020333094161031e-05, "loss": 0.9882, "step": 9481 }, { "epoch": 1.53, "grad_norm": 9.517863273620605, "learning_rate": 1.0201593306790714e-05, "loss": 1.061, "step": 9482 }, { "epoch": 1.53, "grad_norm": 10.063821792602539, "learning_rate": 1.0199855665881765e-05, "loss": 0.9257, "step": 9483 }, { "epoch": 1.53, "grad_norm": 10.508227348327637, "learning_rate": 1.0198118018935953e-05, "loss": 1.4561, "step": 9484 }, { "epoch": 1.53, "grad_norm": 10.13498592376709, "learning_rate": 1.019638036600576e-05, "loss": 1.1178, "step": 9485 }, { "epoch": 1.53, "grad_norm": 11.183286666870117, "learning_rate": 1.0194642707143676e-05, "loss": 1.3217, "step": 9486 }, { "epoch": 1.53, "grad_norm": 9.484804153442383, "learning_rate": 1.0192905042402192e-05, "loss": 1.1303, "step": 9487 }, { "epoch": 1.53, "grad_norm": 10.817388534545898, "learning_rate": 1.019116737183379e-05, "loss": 1.1555, "step": 9488 }, { "epoch": 1.53, "grad_norm": 9.685019493103027, "learning_rate": 1.0189429695490962e-05, "loss": 1.2706, "step": 9489 }, { "epoch": 1.53, "grad_norm": 11.7936429977417, "learning_rate": 1.0187692013426198e-05, "loss": 1.148, "step": 9490 }, { "epoch": 1.53, "grad_norm": 9.895279884338379, "learning_rate": 1.0185954325691984e-05, "loss": 1.0373, "step": 9491 }, { "epoch": 1.53, "grad_norm": 8.543269157409668, "learning_rate": 1.0184216632340807e-05, "loss": 1.0444, "step": 9492 }, { "epoch": 1.53, "grad_norm": 9.420666694641113, "learning_rate": 1.0182478933425162e-05, "loss": 1.1202, "step": 9493 }, { "epoch": 1.53, "grad_norm": 8.195995330810547, "learning_rate": 1.0180741228997532e-05, "loss": 1.1998, "step": 9494 }, { "epoch": 1.53, "grad_norm": 10.569952011108398, "learning_rate": 1.0179003519110413e-05, "loss": 1.1789, "step": 9495 }, { "epoch": 1.53, "grad_norm": 9.896920204162598, "learning_rate": 1.0177265803816287e-05, "loss": 1.1891, "step": 9496 }, { "epoch": 1.53, "grad_norm": 8.69053840637207, "learning_rate": 1.0175528083167647e-05, "loss": 0.8865, "step": 9497 }, { "epoch": 1.53, "grad_norm": 9.520544052124023, "learning_rate": 1.0173790357216986e-05, "loss": 0.8445, "step": 9498 }, { "epoch": 1.53, "grad_norm": 9.806665420532227, "learning_rate": 1.017205262601679e-05, "loss": 0.9471, "step": 9499 }, { "epoch": 1.53, "grad_norm": 10.78055477142334, "learning_rate": 1.0170314889619548e-05, "loss": 1.0927, "step": 9500 }, { "epoch": 1.53, "grad_norm": 10.5518798828125, "learning_rate": 1.0168577148077754e-05, "loss": 1.2087, "step": 9501 }, { "epoch": 1.53, "grad_norm": 10.32021427154541, "learning_rate": 1.0166839401443897e-05, "loss": 0.951, "step": 9502 }, { "epoch": 1.53, "grad_norm": 7.478814601898193, "learning_rate": 1.0165101649770466e-05, "loss": 0.8109, "step": 9503 }, { "epoch": 1.53, "grad_norm": 9.614328384399414, "learning_rate": 1.0163363893109954e-05, "loss": 0.857, "step": 9504 }, { "epoch": 1.53, "grad_norm": 11.123396873474121, "learning_rate": 1.0161626131514852e-05, "loss": 1.061, "step": 9505 }, { "epoch": 1.53, "grad_norm": 10.755674362182617, "learning_rate": 1.015988836503765e-05, "loss": 0.9797, "step": 9506 }, { "epoch": 1.53, "grad_norm": 11.678714752197266, "learning_rate": 1.015815059373084e-05, "loss": 1.1878, "step": 9507 }, { "epoch": 1.53, "grad_norm": 11.25098705291748, "learning_rate": 1.015641281764691e-05, "loss": 0.9688, "step": 9508 }, { "epoch": 1.53, "grad_norm": 10.732644081115723, "learning_rate": 1.0154675036838355e-05, "loss": 1.2338, "step": 9509 }, { "epoch": 1.53, "grad_norm": 9.724671363830566, "learning_rate": 1.0152937251357666e-05, "loss": 1.2207, "step": 9510 }, { "epoch": 1.53, "grad_norm": 9.784506797790527, "learning_rate": 1.0151199461257337e-05, "loss": 1.1598, "step": 9511 }, { "epoch": 1.53, "grad_norm": 8.64903736114502, "learning_rate": 1.0149461666589856e-05, "loss": 1.1296, "step": 9512 }, { "epoch": 1.53, "grad_norm": 8.37851333618164, "learning_rate": 1.0147723867407712e-05, "loss": 0.8521, "step": 9513 }, { "epoch": 1.53, "grad_norm": 10.823356628417969, "learning_rate": 1.014598606376341e-05, "loss": 1.3661, "step": 9514 }, { "epoch": 1.53, "grad_norm": 10.236513137817383, "learning_rate": 1.014424825570943e-05, "loss": 1.0632, "step": 9515 }, { "epoch": 1.53, "grad_norm": 7.3473286628723145, "learning_rate": 1.0142510443298268e-05, "loss": 0.8774, "step": 9516 }, { "epoch": 1.53, "grad_norm": 11.058585166931152, "learning_rate": 1.0140772626582416e-05, "loss": 1.2107, "step": 9517 }, { "epoch": 1.53, "grad_norm": 8.886554718017578, "learning_rate": 1.0139034805614371e-05, "loss": 0.89, "step": 9518 }, { "epoch": 1.53, "grad_norm": 8.540613174438477, "learning_rate": 1.0137296980446623e-05, "loss": 1.1599, "step": 9519 }, { "epoch": 1.53, "grad_norm": 8.394298553466797, "learning_rate": 1.0135559151131662e-05, "loss": 1.1333, "step": 9520 }, { "epoch": 1.53, "grad_norm": 9.306245803833008, "learning_rate": 1.0133821317721984e-05, "loss": 0.8773, "step": 9521 }, { "epoch": 1.53, "grad_norm": 10.136090278625488, "learning_rate": 1.013208348027008e-05, "loss": 0.956, "step": 9522 }, { "epoch": 1.53, "grad_norm": 9.670939445495605, "learning_rate": 1.0130345638828449e-05, "loss": 1.0874, "step": 9523 }, { "epoch": 1.53, "grad_norm": 10.1639986038208, "learning_rate": 1.012860779344958e-05, "loss": 1.0067, "step": 9524 }, { "epoch": 1.53, "grad_norm": 10.395684242248535, "learning_rate": 1.0126869944185964e-05, "loss": 1.1583, "step": 9525 }, { "epoch": 1.53, "grad_norm": 10.106088638305664, "learning_rate": 1.0125132091090104e-05, "loss": 1.214, "step": 9526 }, { "epoch": 1.53, "grad_norm": 11.132328987121582, "learning_rate": 1.0123394234214485e-05, "loss": 1.0018, "step": 9527 }, { "epoch": 1.53, "grad_norm": 8.989890098571777, "learning_rate": 1.0121656373611604e-05, "loss": 0.9702, "step": 9528 }, { "epoch": 1.53, "grad_norm": 8.164880752563477, "learning_rate": 1.0119918509333952e-05, "loss": 0.9162, "step": 9529 }, { "epoch": 1.53, "grad_norm": 12.028085708618164, "learning_rate": 1.011818064143403e-05, "loss": 1.0103, "step": 9530 }, { "epoch": 1.53, "grad_norm": 9.171199798583984, "learning_rate": 1.0116442769964328e-05, "loss": 0.9578, "step": 9531 }, { "epoch": 1.53, "grad_norm": 10.53958511352539, "learning_rate": 1.0114704894977339e-05, "loss": 1.1989, "step": 9532 }, { "epoch": 1.53, "grad_norm": 10.233777046203613, "learning_rate": 1.0112967016525554e-05, "loss": 1.0314, "step": 9533 }, { "epoch": 1.53, "grad_norm": 8.489853858947754, "learning_rate": 1.0111229134661483e-05, "loss": 0.8343, "step": 9534 }, { "epoch": 1.53, "grad_norm": 9.649458885192871, "learning_rate": 1.0109491249437604e-05, "loss": 1.1523, "step": 9535 }, { "epoch": 1.54, "grad_norm": 10.43073558807373, "learning_rate": 1.0107753360906419e-05, "loss": 1.0889, "step": 9536 }, { "epoch": 1.54, "grad_norm": 7.965679168701172, "learning_rate": 1.0106015469120423e-05, "loss": 1.0808, "step": 9537 }, { "epoch": 1.54, "grad_norm": 12.172493934631348, "learning_rate": 1.0104277574132111e-05, "loss": 1.3882, "step": 9538 }, { "epoch": 1.54, "grad_norm": 12.297978401184082, "learning_rate": 1.0102539675993975e-05, "loss": 1.066, "step": 9539 }, { "epoch": 1.54, "grad_norm": 8.322209358215332, "learning_rate": 1.0100801774758514e-05, "loss": 1.1438, "step": 9540 }, { "epoch": 1.54, "grad_norm": 16.434600830078125, "learning_rate": 1.0099063870478218e-05, "loss": 0.8644, "step": 9541 }, { "epoch": 1.54, "grad_norm": 9.134875297546387, "learning_rate": 1.009732596320559e-05, "loss": 1.1351, "step": 9542 }, { "epoch": 1.54, "grad_norm": 11.016353607177734, "learning_rate": 1.0095588052993118e-05, "loss": 1.0153, "step": 9543 }, { "epoch": 1.54, "grad_norm": 8.47156047821045, "learning_rate": 1.0093850139893302e-05, "loss": 0.8825, "step": 9544 }, { "epoch": 1.54, "grad_norm": 10.868711471557617, "learning_rate": 1.009211222395864e-05, "loss": 1.0282, "step": 9545 }, { "epoch": 1.54, "grad_norm": 8.787107467651367, "learning_rate": 1.0090374305241619e-05, "loss": 0.8494, "step": 9546 }, { "epoch": 1.54, "grad_norm": 11.722097396850586, "learning_rate": 1.0088636383794744e-05, "loss": 1.1885, "step": 9547 }, { "epoch": 1.54, "grad_norm": 12.252402305603027, "learning_rate": 1.0086898459670504e-05, "loss": 1.2819, "step": 9548 }, { "epoch": 1.54, "grad_norm": 8.78125, "learning_rate": 1.0085160532921399e-05, "loss": 1.2391, "step": 9549 }, { "epoch": 1.54, "grad_norm": 9.656005859375, "learning_rate": 1.0083422603599926e-05, "loss": 1.1772, "step": 9550 }, { "epoch": 1.54, "grad_norm": 10.282559394836426, "learning_rate": 1.0081684671758576e-05, "loss": 1.158, "step": 9551 }, { "epoch": 1.54, "grad_norm": 8.845426559448242, "learning_rate": 1.0079946737449851e-05, "loss": 1.1801, "step": 9552 }, { "epoch": 1.54, "grad_norm": 8.98217487335205, "learning_rate": 1.0078208800726242e-05, "loss": 0.9965, "step": 9553 }, { "epoch": 1.54, "grad_norm": 10.09439754486084, "learning_rate": 1.0076470861640248e-05, "loss": 1.2649, "step": 9554 }, { "epoch": 1.54, "grad_norm": 8.634113311767578, "learning_rate": 1.0074732920244366e-05, "loss": 1.0057, "step": 9555 }, { "epoch": 1.54, "grad_norm": 7.95496940612793, "learning_rate": 1.007299497659109e-05, "loss": 1.0654, "step": 9556 }, { "epoch": 1.54, "grad_norm": 8.815203666687012, "learning_rate": 1.0071257030732924e-05, "loss": 1.1049, "step": 9557 }, { "epoch": 1.54, "grad_norm": 9.09001350402832, "learning_rate": 1.0069519082722355e-05, "loss": 0.9668, "step": 9558 }, { "epoch": 1.54, "grad_norm": 10.362190246582031, "learning_rate": 1.0067781132611883e-05, "loss": 1.0386, "step": 9559 }, { "epoch": 1.54, "grad_norm": 6.776011943817139, "learning_rate": 1.0066043180454004e-05, "loss": 0.7813, "step": 9560 }, { "epoch": 1.54, "grad_norm": 8.60266399383545, "learning_rate": 1.0064305226301218e-05, "loss": 0.9603, "step": 9561 }, { "epoch": 1.54, "grad_norm": 8.020768165588379, "learning_rate": 1.0062567270206023e-05, "loss": 0.8533, "step": 9562 }, { "epoch": 1.54, "grad_norm": 8.809276580810547, "learning_rate": 1.0060829312220906e-05, "loss": 1.3217, "step": 9563 }, { "epoch": 1.54, "grad_norm": 9.531461715698242, "learning_rate": 1.0059091352398376e-05, "loss": 1.1586, "step": 9564 }, { "epoch": 1.54, "grad_norm": 7.495376110076904, "learning_rate": 1.0057353390790925e-05, "loss": 0.8943, "step": 9565 }, { "epoch": 1.54, "grad_norm": 9.518547058105469, "learning_rate": 1.0055615427451049e-05, "loss": 1.0063, "step": 9566 }, { "epoch": 1.54, "grad_norm": 9.139299392700195, "learning_rate": 1.0053877462431245e-05, "loss": 0.9791, "step": 9567 }, { "epoch": 1.54, "grad_norm": 9.922714233398438, "learning_rate": 1.0052139495784012e-05, "loss": 1.1279, "step": 9568 }, { "epoch": 1.54, "grad_norm": 8.72696590423584, "learning_rate": 1.0050401527561848e-05, "loss": 0.9376, "step": 9569 }, { "epoch": 1.54, "grad_norm": 9.367863655090332, "learning_rate": 1.0048663557817249e-05, "loss": 1.2547, "step": 9570 }, { "epoch": 1.54, "grad_norm": 9.508394241333008, "learning_rate": 1.0046925586602712e-05, "loss": 0.922, "step": 9571 }, { "epoch": 1.54, "grad_norm": 9.654035568237305, "learning_rate": 1.0045187613970733e-05, "loss": 1.0081, "step": 9572 }, { "epoch": 1.54, "grad_norm": 9.001102447509766, "learning_rate": 1.004344963997381e-05, "loss": 1.1696, "step": 9573 }, { "epoch": 1.54, "grad_norm": 9.135662078857422, "learning_rate": 1.0041711664664444e-05, "loss": 0.891, "step": 9574 }, { "epoch": 1.54, "grad_norm": 9.238807678222656, "learning_rate": 1.0039973688095131e-05, "loss": 1.0164, "step": 9575 }, { "epoch": 1.54, "grad_norm": 9.88713550567627, "learning_rate": 1.0038235710318364e-05, "loss": 0.9172, "step": 9576 }, { "epoch": 1.54, "grad_norm": 10.681159973144531, "learning_rate": 1.003649773138665e-05, "loss": 0.9763, "step": 9577 }, { "epoch": 1.54, "grad_norm": 9.171643257141113, "learning_rate": 1.0034759751352477e-05, "loss": 1.1364, "step": 9578 }, { "epoch": 1.54, "grad_norm": 11.820283889770508, "learning_rate": 1.0033021770268345e-05, "loss": 1.1474, "step": 9579 }, { "epoch": 1.54, "grad_norm": 10.008898735046387, "learning_rate": 1.0031283788186755e-05, "loss": 0.8651, "step": 9580 }, { "epoch": 1.54, "grad_norm": 9.752025604248047, "learning_rate": 1.0029545805160206e-05, "loss": 0.9341, "step": 9581 }, { "epoch": 1.54, "grad_norm": 9.135138511657715, "learning_rate": 1.0027807821241187e-05, "loss": 0.9397, "step": 9582 }, { "epoch": 1.54, "grad_norm": 9.63477897644043, "learning_rate": 1.0026069836482202e-05, "loss": 1.023, "step": 9583 }, { "epoch": 1.54, "grad_norm": 10.693130493164062, "learning_rate": 1.0024331850935751e-05, "loss": 0.8397, "step": 9584 }, { "epoch": 1.54, "grad_norm": 11.215875625610352, "learning_rate": 1.0022593864654328e-05, "loss": 1.1107, "step": 9585 }, { "epoch": 1.54, "grad_norm": 11.422174453735352, "learning_rate": 1.0020855877690432e-05, "loss": 1.1147, "step": 9586 }, { "epoch": 1.54, "grad_norm": 9.836671829223633, "learning_rate": 1.001911789009656e-05, "loss": 1.2792, "step": 9587 }, { "epoch": 1.54, "grad_norm": 9.578567504882812, "learning_rate": 1.001737990192521e-05, "loss": 1.2052, "step": 9588 }, { "epoch": 1.54, "grad_norm": 9.940349578857422, "learning_rate": 1.0015641913228882e-05, "loss": 1.2753, "step": 9589 }, { "epoch": 1.54, "grad_norm": 8.145795822143555, "learning_rate": 1.0013903924060071e-05, "loss": 0.8533, "step": 9590 }, { "epoch": 1.54, "grad_norm": 7.7693190574646, "learning_rate": 1.0012165934471276e-05, "loss": 0.8553, "step": 9591 }, { "epoch": 1.54, "grad_norm": 9.384831428527832, "learning_rate": 1.0010427944514998e-05, "loss": 1.0326, "step": 9592 }, { "epoch": 1.54, "grad_norm": 9.892685890197754, "learning_rate": 1.0008689954243728e-05, "loss": 1.0163, "step": 9593 }, { "epoch": 1.54, "grad_norm": 11.09688663482666, "learning_rate": 1.000695196370997e-05, "loss": 1.1583, "step": 9594 }, { "epoch": 1.54, "grad_norm": 10.023750305175781, "learning_rate": 1.000521397296622e-05, "loss": 0.9911, "step": 9595 }, { "epoch": 1.54, "grad_norm": 8.232357025146484, "learning_rate": 1.0003475982064978e-05, "loss": 0.8934, "step": 9596 }, { "epoch": 1.54, "grad_norm": 8.820530891418457, "learning_rate": 1.0001737991058739e-05, "loss": 1.1402, "step": 9597 }, { "epoch": 1.55, "grad_norm": 8.4398193359375, "learning_rate": 1e-05, "loss": 0.7575, "step": 9598 }, { "epoch": 1.55, "grad_norm": 9.923101425170898, "learning_rate": 9.998262008941264e-06, "loss": 0.8284, "step": 9599 }, { "epoch": 1.55, "grad_norm": 11.87346363067627, "learning_rate": 9.996524017935024e-06, "loss": 1.3197, "step": 9600 }, { "epoch": 1.55, "grad_norm": 9.417652130126953, "learning_rate": 9.994786027033781e-06, "loss": 0.9572, "step": 9601 }, { "epoch": 1.55, "grad_norm": 10.420198440551758, "learning_rate": 9.993048036290031e-06, "loss": 1.0351, "step": 9602 }, { "epoch": 1.55, "grad_norm": 10.007031440734863, "learning_rate": 9.991310045756274e-06, "loss": 0.9642, "step": 9603 }, { "epoch": 1.55, "grad_norm": 11.772028923034668, "learning_rate": 9.989572055485007e-06, "loss": 1.1597, "step": 9604 }, { "epoch": 1.55, "grad_norm": 9.755829811096191, "learning_rate": 9.987834065528728e-06, "loss": 1.2431, "step": 9605 }, { "epoch": 1.55, "grad_norm": 9.660898208618164, "learning_rate": 9.986096075939932e-06, "loss": 1.3753, "step": 9606 }, { "epoch": 1.55, "grad_norm": 9.857905387878418, "learning_rate": 9.984358086771121e-06, "loss": 1.1689, "step": 9607 }, { "epoch": 1.55, "grad_norm": 8.858871459960938, "learning_rate": 9.982620098074794e-06, "loss": 0.9542, "step": 9608 }, { "epoch": 1.55, "grad_norm": 9.991793632507324, "learning_rate": 9.980882109903442e-06, "loss": 1.1655, "step": 9609 }, { "epoch": 1.55, "grad_norm": 10.475544929504395, "learning_rate": 9.97914412230957e-06, "loss": 1.0407, "step": 9610 }, { "epoch": 1.55, "grad_norm": 8.68616008758545, "learning_rate": 9.977406135345674e-06, "loss": 0.9447, "step": 9611 }, { "epoch": 1.55, "grad_norm": 9.645654678344727, "learning_rate": 9.975668149064252e-06, "loss": 1.049, "step": 9612 }, { "epoch": 1.55, "grad_norm": 11.883746147155762, "learning_rate": 9.9739301635178e-06, "loss": 1.1359, "step": 9613 }, { "epoch": 1.55, "grad_norm": 9.708772659301758, "learning_rate": 9.972192178758815e-06, "loss": 1.1068, "step": 9614 }, { "epoch": 1.55, "grad_norm": 9.25830364227295, "learning_rate": 9.970454194839799e-06, "loss": 0.9586, "step": 9615 }, { "epoch": 1.55, "grad_norm": 10.174774169921875, "learning_rate": 9.968716211813247e-06, "loss": 1.1382, "step": 9616 }, { "epoch": 1.55, "grad_norm": 10.510536193847656, "learning_rate": 9.966978229731657e-06, "loss": 1.0653, "step": 9617 }, { "epoch": 1.55, "grad_norm": 9.647294044494629, "learning_rate": 9.965240248647526e-06, "loss": 1.1673, "step": 9618 }, { "epoch": 1.55, "grad_norm": 9.30296802520752, "learning_rate": 9.963502268613352e-06, "loss": 0.9126, "step": 9619 }, { "epoch": 1.55, "grad_norm": 11.731636047363281, "learning_rate": 9.961764289681637e-06, "loss": 1.0583, "step": 9620 }, { "epoch": 1.55, "grad_norm": 10.622647285461426, "learning_rate": 9.960026311904874e-06, "loss": 0.9568, "step": 9621 }, { "epoch": 1.55, "grad_norm": 10.082159042358398, "learning_rate": 9.958288335335558e-06, "loss": 1.1925, "step": 9622 }, { "epoch": 1.55, "grad_norm": 12.043702125549316, "learning_rate": 9.956550360026191e-06, "loss": 1.4072, "step": 9623 }, { "epoch": 1.55, "grad_norm": 11.052908897399902, "learning_rate": 9.954812386029272e-06, "loss": 1.1494, "step": 9624 }, { "epoch": 1.55, "grad_norm": 8.810569763183594, "learning_rate": 9.953074413397295e-06, "loss": 0.9059, "step": 9625 }, { "epoch": 1.55, "grad_norm": 11.573979377746582, "learning_rate": 9.951336442182756e-06, "loss": 1.1899, "step": 9626 }, { "epoch": 1.55, "grad_norm": 9.257967948913574, "learning_rate": 9.949598472438156e-06, "loss": 1.1666, "step": 9627 }, { "epoch": 1.55, "grad_norm": 9.605501174926758, "learning_rate": 9.94786050421599e-06, "loss": 1.1733, "step": 9628 }, { "epoch": 1.55, "grad_norm": 10.103167533874512, "learning_rate": 9.946122537568755e-06, "loss": 1.2437, "step": 9629 }, { "epoch": 1.55, "grad_norm": 10.384405136108398, "learning_rate": 9.944384572548953e-06, "loss": 1.0133, "step": 9630 }, { "epoch": 1.55, "grad_norm": 9.811921119689941, "learning_rate": 9.942646609209079e-06, "loss": 1.1061, "step": 9631 }, { "epoch": 1.55, "grad_norm": 11.865966796875, "learning_rate": 9.940908647601625e-06, "loss": 1.105, "step": 9632 }, { "epoch": 1.55, "grad_norm": 9.481188774108887, "learning_rate": 9.939170687779094e-06, "loss": 1.0471, "step": 9633 }, { "epoch": 1.55, "grad_norm": 9.18903923034668, "learning_rate": 9.93743272979398e-06, "loss": 1.087, "step": 9634 }, { "epoch": 1.55, "grad_norm": 10.093328475952148, "learning_rate": 9.935694773698785e-06, "loss": 1.0644, "step": 9635 }, { "epoch": 1.55, "grad_norm": 10.032806396484375, "learning_rate": 9.933956819546e-06, "loss": 1.1687, "step": 9636 }, { "epoch": 1.55, "grad_norm": 7.735827445983887, "learning_rate": 9.932218867388122e-06, "loss": 0.9367, "step": 9637 }, { "epoch": 1.55, "grad_norm": 10.058444023132324, "learning_rate": 9.93048091727765e-06, "loss": 0.9077, "step": 9638 }, { "epoch": 1.55, "grad_norm": 10.151171684265137, "learning_rate": 9.92874296926708e-06, "loss": 1.1212, "step": 9639 }, { "epoch": 1.55, "grad_norm": 8.96534538269043, "learning_rate": 9.927005023408913e-06, "loss": 1.0462, "step": 9640 }, { "epoch": 1.55, "grad_norm": 8.970867156982422, "learning_rate": 9.925267079755637e-06, "loss": 1.0003, "step": 9641 }, { "epoch": 1.55, "grad_norm": 10.346421241760254, "learning_rate": 9.923529138359756e-06, "loss": 1.098, "step": 9642 }, { "epoch": 1.55, "grad_norm": 14.651737213134766, "learning_rate": 9.921791199273761e-06, "loss": 1.2241, "step": 9643 }, { "epoch": 1.55, "grad_norm": 9.753533363342285, "learning_rate": 9.920053262550156e-06, "loss": 1.0212, "step": 9644 }, { "epoch": 1.55, "grad_norm": 9.194049835205078, "learning_rate": 9.918315328241427e-06, "loss": 0.8846, "step": 9645 }, { "epoch": 1.55, "grad_norm": 10.331952095031738, "learning_rate": 9.916577396400079e-06, "loss": 1.012, "step": 9646 }, { "epoch": 1.55, "grad_norm": 9.59816837310791, "learning_rate": 9.914839467078604e-06, "loss": 0.9062, "step": 9647 }, { "epoch": 1.55, "grad_norm": 13.440356254577637, "learning_rate": 9.913101540329501e-06, "loss": 1.324, "step": 9648 }, { "epoch": 1.55, "grad_norm": 11.577521324157715, "learning_rate": 9.911363616205258e-06, "loss": 1.1236, "step": 9649 }, { "epoch": 1.55, "grad_norm": 10.23145580291748, "learning_rate": 9.909625694758381e-06, "loss": 0.9984, "step": 9650 }, { "epoch": 1.55, "grad_norm": 8.973830223083496, "learning_rate": 9.907887776041364e-06, "loss": 0.7059, "step": 9651 }, { "epoch": 1.55, "grad_norm": 7.863427639007568, "learning_rate": 9.906149860106698e-06, "loss": 1.1222, "step": 9652 }, { "epoch": 1.55, "grad_norm": 6.9923272132873535, "learning_rate": 9.904411947006884e-06, "loss": 0.8432, "step": 9653 }, { "epoch": 1.55, "grad_norm": 8.953509330749512, "learning_rate": 9.902674036794413e-06, "loss": 1.1765, "step": 9654 }, { "epoch": 1.55, "grad_norm": 7.748394012451172, "learning_rate": 9.900936129521784e-06, "loss": 0.9391, "step": 9655 }, { "epoch": 1.55, "grad_norm": 7.621736526489258, "learning_rate": 9.899198225241491e-06, "loss": 0.8375, "step": 9656 }, { "epoch": 1.55, "grad_norm": 8.079564094543457, "learning_rate": 9.897460324006029e-06, "loss": 1.1435, "step": 9657 }, { "epoch": 1.55, "grad_norm": 7.866931915283203, "learning_rate": 9.895722425867892e-06, "loss": 1.1049, "step": 9658 }, { "epoch": 1.55, "grad_norm": 9.326428413391113, "learning_rate": 9.893984530879582e-06, "loss": 0.9724, "step": 9659 }, { "epoch": 1.56, "grad_norm": 9.106100082397461, "learning_rate": 9.892246639093585e-06, "loss": 0.8633, "step": 9660 }, { "epoch": 1.56, "grad_norm": 9.090494155883789, "learning_rate": 9.8905087505624e-06, "loss": 1.1225, "step": 9661 }, { "epoch": 1.56, "grad_norm": 11.386738777160645, "learning_rate": 9.888770865338522e-06, "loss": 1.0, "step": 9662 }, { "epoch": 1.56, "grad_norm": 13.918513298034668, "learning_rate": 9.887032983474447e-06, "loss": 1.1196, "step": 9663 }, { "epoch": 1.56, "grad_norm": 10.13464641571045, "learning_rate": 9.885295105022668e-06, "loss": 0.8683, "step": 9664 }, { "epoch": 1.56, "grad_norm": 9.382020950317383, "learning_rate": 9.883557230035678e-06, "loss": 0.8436, "step": 9665 }, { "epoch": 1.56, "grad_norm": 13.008647918701172, "learning_rate": 9.881819358565973e-06, "loss": 1.1466, "step": 9666 }, { "epoch": 1.56, "grad_norm": 11.122110366821289, "learning_rate": 9.880081490666051e-06, "loss": 1.1383, "step": 9667 }, { "epoch": 1.56, "grad_norm": 9.2191162109375, "learning_rate": 9.878343626388402e-06, "loss": 0.924, "step": 9668 }, { "epoch": 1.56, "grad_norm": 10.115800857543945, "learning_rate": 9.876605765785516e-06, "loss": 1.28, "step": 9669 }, { "epoch": 1.56, "grad_norm": 11.570159912109375, "learning_rate": 9.874867908909899e-06, "loss": 0.941, "step": 9670 }, { "epoch": 1.56, "grad_norm": 8.861311912536621, "learning_rate": 9.873130055814036e-06, "loss": 0.9172, "step": 9671 }, { "epoch": 1.56, "grad_norm": 8.835729598999023, "learning_rate": 9.871392206550422e-06, "loss": 1.139, "step": 9672 }, { "epoch": 1.56, "grad_norm": 8.558974266052246, "learning_rate": 9.869654361171553e-06, "loss": 1.1508, "step": 9673 }, { "epoch": 1.56, "grad_norm": 9.155954360961914, "learning_rate": 9.867916519729922e-06, "loss": 1.2494, "step": 9674 }, { "epoch": 1.56, "grad_norm": 9.572027206420898, "learning_rate": 9.866178682278019e-06, "loss": 1.1942, "step": 9675 }, { "epoch": 1.56, "grad_norm": 7.6390299797058105, "learning_rate": 9.864440848868342e-06, "loss": 0.9679, "step": 9676 }, { "epoch": 1.56, "grad_norm": 8.664241790771484, "learning_rate": 9.86270301955338e-06, "loss": 0.9594, "step": 9677 }, { "epoch": 1.56, "grad_norm": 9.677865028381348, "learning_rate": 9.86096519438563e-06, "loss": 1.1059, "step": 9678 }, { "epoch": 1.56, "grad_norm": 7.166117191314697, "learning_rate": 9.859227373417586e-06, "loss": 0.838, "step": 9679 }, { "epoch": 1.56, "grad_norm": 12.326985359191895, "learning_rate": 9.857489556701735e-06, "loss": 1.2783, "step": 9680 }, { "epoch": 1.56, "grad_norm": 8.966672897338867, "learning_rate": 9.855751744290574e-06, "loss": 1.0121, "step": 9681 }, { "epoch": 1.56, "grad_norm": 7.9620842933654785, "learning_rate": 9.854013936236594e-06, "loss": 1.107, "step": 9682 }, { "epoch": 1.56, "grad_norm": 9.531159400939941, "learning_rate": 9.85227613259229e-06, "loss": 0.9077, "step": 9683 }, { "epoch": 1.56, "grad_norm": 10.390778541564941, "learning_rate": 9.85053833341015e-06, "loss": 1.2314, "step": 9684 }, { "epoch": 1.56, "grad_norm": 11.175250053405762, "learning_rate": 9.848800538742668e-06, "loss": 0.7829, "step": 9685 }, { "epoch": 1.56, "grad_norm": 10.086162567138672, "learning_rate": 9.847062748642336e-06, "loss": 0.9919, "step": 9686 }, { "epoch": 1.56, "grad_norm": 8.528877258300781, "learning_rate": 9.84532496316165e-06, "loss": 1.287, "step": 9687 }, { "epoch": 1.56, "grad_norm": 10.413688659667969, "learning_rate": 9.843587182353095e-06, "loss": 0.966, "step": 9688 }, { "epoch": 1.56, "grad_norm": 9.899103164672852, "learning_rate": 9.841849406269165e-06, "loss": 0.9275, "step": 9689 }, { "epoch": 1.56, "grad_norm": 9.294397354125977, "learning_rate": 9.840111634962353e-06, "loss": 1.259, "step": 9690 }, { "epoch": 1.56, "grad_norm": 8.636621475219727, "learning_rate": 9.83837386848515e-06, "loss": 1.1356, "step": 9691 }, { "epoch": 1.56, "grad_norm": 10.79664421081543, "learning_rate": 9.836636106890046e-06, "loss": 1.152, "step": 9692 }, { "epoch": 1.56, "grad_norm": 9.720829010009766, "learning_rate": 9.834898350229534e-06, "loss": 1.1755, "step": 9693 }, { "epoch": 1.56, "grad_norm": 10.395153999328613, "learning_rate": 9.833160598556107e-06, "loss": 0.7369, "step": 9694 }, { "epoch": 1.56, "grad_norm": 10.364947319030762, "learning_rate": 9.831422851922249e-06, "loss": 0.9894, "step": 9695 }, { "epoch": 1.56, "grad_norm": 9.730558395385742, "learning_rate": 9.829685110380454e-06, "loss": 1.0715, "step": 9696 }, { "epoch": 1.56, "grad_norm": 9.535979270935059, "learning_rate": 9.827947373983213e-06, "loss": 0.9062, "step": 9697 }, { "epoch": 1.56, "grad_norm": 8.951814651489258, "learning_rate": 9.826209642783018e-06, "loss": 1.0508, "step": 9698 }, { "epoch": 1.56, "grad_norm": 11.401058197021484, "learning_rate": 9.824471916832355e-06, "loss": 1.1381, "step": 9699 }, { "epoch": 1.56, "grad_norm": 10.198217391967773, "learning_rate": 9.822734196183715e-06, "loss": 0.917, "step": 9700 }, { "epoch": 1.56, "grad_norm": 10.204384803771973, "learning_rate": 9.82099648088959e-06, "loss": 1.0015, "step": 9701 }, { "epoch": 1.56, "grad_norm": 10.528212547302246, "learning_rate": 9.819258771002471e-06, "loss": 1.3673, "step": 9702 }, { "epoch": 1.56, "grad_norm": 8.222389221191406, "learning_rate": 9.817521066574843e-06, "loss": 0.8597, "step": 9703 }, { "epoch": 1.56, "grad_norm": 8.1177396774292, "learning_rate": 9.815783367659195e-06, "loss": 1.1388, "step": 9704 }, { "epoch": 1.56, "grad_norm": 10.908649444580078, "learning_rate": 9.81404567430802e-06, "loss": 1.0981, "step": 9705 }, { "epoch": 1.56, "grad_norm": 10.846389770507812, "learning_rate": 9.812307986573806e-06, "loss": 1.1378, "step": 9706 }, { "epoch": 1.56, "grad_norm": 8.66542911529541, "learning_rate": 9.810570304509041e-06, "loss": 0.9528, "step": 9707 }, { "epoch": 1.56, "grad_norm": 9.860577583312988, "learning_rate": 9.808832628166214e-06, "loss": 0.9296, "step": 9708 }, { "epoch": 1.56, "grad_norm": 8.68062973022461, "learning_rate": 9.807094957597813e-06, "loss": 0.9705, "step": 9709 }, { "epoch": 1.56, "grad_norm": 10.168205261230469, "learning_rate": 9.805357292856324e-06, "loss": 1.0502, "step": 9710 }, { "epoch": 1.56, "grad_norm": 10.179237365722656, "learning_rate": 9.803619633994241e-06, "loss": 1.2379, "step": 9711 }, { "epoch": 1.56, "grad_norm": 11.461735725402832, "learning_rate": 9.801881981064049e-06, "loss": 1.091, "step": 9712 }, { "epoch": 1.56, "grad_norm": 9.81933307647705, "learning_rate": 9.800144334118236e-06, "loss": 1.086, "step": 9713 }, { "epoch": 1.56, "grad_norm": 9.334803581237793, "learning_rate": 9.798406693209288e-06, "loss": 1.1308, "step": 9714 }, { "epoch": 1.56, "grad_norm": 10.797350883483887, "learning_rate": 9.796669058389694e-06, "loss": 0.934, "step": 9715 }, { "epoch": 1.56, "grad_norm": 9.285636901855469, "learning_rate": 9.794931429711938e-06, "loss": 0.8525, "step": 9716 }, { "epoch": 1.56, "grad_norm": 10.235344886779785, "learning_rate": 9.793193807228513e-06, "loss": 1.0713, "step": 9717 }, { "epoch": 1.56, "grad_norm": 10.35103702545166, "learning_rate": 9.791456190991903e-06, "loss": 1.0921, "step": 9718 }, { "epoch": 1.56, "grad_norm": 8.197799682617188, "learning_rate": 9.789718581054593e-06, "loss": 0.8217, "step": 9719 }, { "epoch": 1.56, "grad_norm": 8.810091972351074, "learning_rate": 9.78798097746907e-06, "loss": 0.8895, "step": 9720 }, { "epoch": 1.56, "grad_norm": 9.267091751098633, "learning_rate": 9.78624338028782e-06, "loss": 0.924, "step": 9721 }, { "epoch": 1.57, "grad_norm": 8.970949172973633, "learning_rate": 9.784505789563335e-06, "loss": 1.2895, "step": 9722 }, { "epoch": 1.57, "grad_norm": 8.834470748901367, "learning_rate": 9.782768205348093e-06, "loss": 0.9125, "step": 9723 }, { "epoch": 1.57, "grad_norm": 9.845524787902832, "learning_rate": 9.781030627694583e-06, "loss": 1.1059, "step": 9724 }, { "epoch": 1.57, "grad_norm": 9.64852523803711, "learning_rate": 9.779293056655288e-06, "loss": 0.8448, "step": 9725 }, { "epoch": 1.57, "grad_norm": 7.957540035247803, "learning_rate": 9.777555492282699e-06, "loss": 1.0028, "step": 9726 }, { "epoch": 1.57, "grad_norm": 8.443038940429688, "learning_rate": 9.775817934629295e-06, "loss": 0.9234, "step": 9727 }, { "epoch": 1.57, "grad_norm": 8.011575698852539, "learning_rate": 9.774080383747566e-06, "loss": 0.9568, "step": 9728 }, { "epoch": 1.57, "grad_norm": 10.4710054397583, "learning_rate": 9.772342839689991e-06, "loss": 1.0065, "step": 9729 }, { "epoch": 1.57, "grad_norm": 10.051148414611816, "learning_rate": 9.770605302509057e-06, "loss": 1.3032, "step": 9730 }, { "epoch": 1.57, "grad_norm": 7.5730085372924805, "learning_rate": 9.768867772257252e-06, "loss": 0.9725, "step": 9731 }, { "epoch": 1.57, "grad_norm": 9.078097343444824, "learning_rate": 9.767130248987055e-06, "loss": 1.0376, "step": 9732 }, { "epoch": 1.57, "grad_norm": 9.22132682800293, "learning_rate": 9.765392732750955e-06, "loss": 1.1638, "step": 9733 }, { "epoch": 1.57, "grad_norm": 10.013542175292969, "learning_rate": 9.763655223601429e-06, "loss": 1.041, "step": 9734 }, { "epoch": 1.57, "grad_norm": 10.310665130615234, "learning_rate": 9.761917721590966e-06, "loss": 0.9303, "step": 9735 }, { "epoch": 1.57, "grad_norm": 10.270554542541504, "learning_rate": 9.760180226772045e-06, "loss": 1.216, "step": 9736 }, { "epoch": 1.57, "grad_norm": 9.84939956665039, "learning_rate": 9.758442739197155e-06, "loss": 0.8358, "step": 9737 }, { "epoch": 1.57, "grad_norm": 13.339763641357422, "learning_rate": 9.756705258918773e-06, "loss": 1.0224, "step": 9738 }, { "epoch": 1.57, "grad_norm": 10.154288291931152, "learning_rate": 9.754967785989382e-06, "loss": 1.1481, "step": 9739 }, { "epoch": 1.57, "grad_norm": 7.683777332305908, "learning_rate": 9.753230320461467e-06, "loss": 0.8107, "step": 9740 }, { "epoch": 1.57, "grad_norm": 10.68794059753418, "learning_rate": 9.75149286238751e-06, "loss": 1.2164, "step": 9741 }, { "epoch": 1.57, "grad_norm": 10.61011791229248, "learning_rate": 9.74975541181999e-06, "loss": 1.0965, "step": 9742 }, { "epoch": 1.57, "grad_norm": 9.895999908447266, "learning_rate": 9.74801796881139e-06, "loss": 0.9608, "step": 9743 }, { "epoch": 1.57, "grad_norm": 8.86890983581543, "learning_rate": 9.746280533414191e-06, "loss": 1.0761, "step": 9744 }, { "epoch": 1.57, "grad_norm": 8.6688871383667, "learning_rate": 9.744543105680878e-06, "loss": 1.0804, "step": 9745 }, { "epoch": 1.57, "grad_norm": 9.447872161865234, "learning_rate": 9.74280568566393e-06, "loss": 1.0233, "step": 9746 }, { "epoch": 1.57, "grad_norm": 10.113616943359375, "learning_rate": 9.741068273415823e-06, "loss": 1.0759, "step": 9747 }, { "epoch": 1.57, "grad_norm": 7.703508377075195, "learning_rate": 9.739330868989041e-06, "loss": 0.9078, "step": 9748 }, { "epoch": 1.57, "grad_norm": 9.820138931274414, "learning_rate": 9.737593472436066e-06, "loss": 0.9506, "step": 9749 }, { "epoch": 1.57, "grad_norm": 9.055255889892578, "learning_rate": 9.735856083809376e-06, "loss": 0.9115, "step": 9750 }, { "epoch": 1.57, "grad_norm": 13.002237319946289, "learning_rate": 9.73411870316145e-06, "loss": 1.1822, "step": 9751 }, { "epoch": 1.57, "grad_norm": 9.81750202178955, "learning_rate": 9.732381330544773e-06, "loss": 0.9274, "step": 9752 }, { "epoch": 1.57, "grad_norm": 11.242961883544922, "learning_rate": 9.730643966011819e-06, "loss": 1.054, "step": 9753 }, { "epoch": 1.57, "grad_norm": 12.060163497924805, "learning_rate": 9.728906609615066e-06, "loss": 1.1111, "step": 9754 }, { "epoch": 1.57, "grad_norm": 10.151363372802734, "learning_rate": 9.727169261406996e-06, "loss": 1.2137, "step": 9755 }, { "epoch": 1.57, "grad_norm": 9.555657386779785, "learning_rate": 9.725431921440085e-06, "loss": 0.9294, "step": 9756 }, { "epoch": 1.57, "grad_norm": 9.517926216125488, "learning_rate": 9.723694589766818e-06, "loss": 1.0358, "step": 9757 }, { "epoch": 1.57, "grad_norm": 9.644425392150879, "learning_rate": 9.721957266439665e-06, "loss": 1.1062, "step": 9758 }, { "epoch": 1.57, "grad_norm": 9.267372131347656, "learning_rate": 9.720219951511106e-06, "loss": 1.0593, "step": 9759 }, { "epoch": 1.57, "grad_norm": 7.2005462646484375, "learning_rate": 9.71848264503362e-06, "loss": 0.8739, "step": 9760 }, { "epoch": 1.57, "grad_norm": 9.436237335205078, "learning_rate": 9.716745347059684e-06, "loss": 1.196, "step": 9761 }, { "epoch": 1.57, "grad_norm": 10.502276420593262, "learning_rate": 9.715008057641776e-06, "loss": 1.0, "step": 9762 }, { "epoch": 1.57, "grad_norm": 11.011892318725586, "learning_rate": 9.713270776832368e-06, "loss": 1.0345, "step": 9763 }, { "epoch": 1.57, "grad_norm": 9.473453521728516, "learning_rate": 9.711533504683942e-06, "loss": 0.9219, "step": 9764 }, { "epoch": 1.57, "grad_norm": 10.322002410888672, "learning_rate": 9.709796241248975e-06, "loss": 1.1454, "step": 9765 }, { "epoch": 1.57, "grad_norm": 13.047889709472656, "learning_rate": 9.708058986579936e-06, "loss": 0.9081, "step": 9766 }, { "epoch": 1.57, "grad_norm": 10.391130447387695, "learning_rate": 9.706321740729306e-06, "loss": 1.2765, "step": 9767 }, { "epoch": 1.57, "grad_norm": 9.88739013671875, "learning_rate": 9.70458450374956e-06, "loss": 1.2984, "step": 9768 }, { "epoch": 1.57, "grad_norm": 13.844283103942871, "learning_rate": 9.702847275693175e-06, "loss": 1.3737, "step": 9769 }, { "epoch": 1.57, "grad_norm": 9.408422470092773, "learning_rate": 9.701110056612621e-06, "loss": 0.9663, "step": 9770 }, { "epoch": 1.57, "grad_norm": 10.476956367492676, "learning_rate": 9.69937284656037e-06, "loss": 1.2538, "step": 9771 }, { "epoch": 1.57, "grad_norm": 9.129587173461914, "learning_rate": 9.69763564558891e-06, "loss": 1.1251, "step": 9772 }, { "epoch": 1.57, "grad_norm": 9.106635093688965, "learning_rate": 9.695898453750704e-06, "loss": 1.1179, "step": 9773 }, { "epoch": 1.57, "grad_norm": 10.36557674407959, "learning_rate": 9.694161271098229e-06, "loss": 1.2712, "step": 9774 }, { "epoch": 1.57, "grad_norm": 8.362799644470215, "learning_rate": 9.692424097683957e-06, "loss": 0.8528, "step": 9775 }, { "epoch": 1.57, "grad_norm": 11.49002742767334, "learning_rate": 9.690686933560365e-06, "loss": 1.0615, "step": 9776 }, { "epoch": 1.57, "grad_norm": 8.932138442993164, "learning_rate": 9.688949778779922e-06, "loss": 1.0362, "step": 9777 }, { "epoch": 1.57, "grad_norm": 8.05931568145752, "learning_rate": 9.687212633395102e-06, "loss": 0.8998, "step": 9778 }, { "epoch": 1.57, "grad_norm": 9.296249389648438, "learning_rate": 9.685475497458378e-06, "loss": 0.8894, "step": 9779 }, { "epoch": 1.57, "grad_norm": 9.469305038452148, "learning_rate": 9.683738371022222e-06, "loss": 1.1505, "step": 9780 }, { "epoch": 1.57, "grad_norm": 9.137923240661621, "learning_rate": 9.682001254139106e-06, "loss": 1.1954, "step": 9781 }, { "epoch": 1.57, "grad_norm": 9.051243782043457, "learning_rate": 9.6802641468615e-06, "loss": 1.1489, "step": 9782 }, { "epoch": 1.57, "grad_norm": 9.120643615722656, "learning_rate": 9.678527049241879e-06, "loss": 1.0977, "step": 9783 }, { "epoch": 1.58, "grad_norm": 7.349173069000244, "learning_rate": 9.676789961332708e-06, "loss": 0.7687, "step": 9784 }, { "epoch": 1.58, "grad_norm": 10.713516235351562, "learning_rate": 9.675052883186467e-06, "loss": 0.965, "step": 9785 }, { "epoch": 1.58, "grad_norm": 9.9493989944458, "learning_rate": 9.673315814855614e-06, "loss": 0.9506, "step": 9786 }, { "epoch": 1.58, "grad_norm": 10.01050090789795, "learning_rate": 9.67157875639263e-06, "loss": 0.9902, "step": 9787 }, { "epoch": 1.58, "grad_norm": 10.235652923583984, "learning_rate": 9.669841707849978e-06, "loss": 1.0108, "step": 9788 }, { "epoch": 1.58, "grad_norm": 10.652853012084961, "learning_rate": 9.668104669280135e-06, "loss": 1.1389, "step": 9789 }, { "epoch": 1.58, "grad_norm": 10.981058120727539, "learning_rate": 9.66636764073556e-06, "loss": 1.1586, "step": 9790 }, { "epoch": 1.58, "grad_norm": 10.152029037475586, "learning_rate": 9.664630622268727e-06, "loss": 1.1234, "step": 9791 }, { "epoch": 1.58, "grad_norm": 8.492720603942871, "learning_rate": 9.662893613932105e-06, "loss": 0.8099, "step": 9792 }, { "epoch": 1.58, "grad_norm": 11.605338096618652, "learning_rate": 9.661156615778165e-06, "loss": 1.06, "step": 9793 }, { "epoch": 1.58, "grad_norm": 9.880850791931152, "learning_rate": 9.65941962785937e-06, "loss": 1.148, "step": 9794 }, { "epoch": 1.58, "grad_norm": 8.297750473022461, "learning_rate": 9.657682650228191e-06, "loss": 0.9322, "step": 9795 }, { "epoch": 1.58, "grad_norm": 11.483136177062988, "learning_rate": 9.655945682937095e-06, "loss": 1.0874, "step": 9796 }, { "epoch": 1.58, "grad_norm": 9.940581321716309, "learning_rate": 9.654208726038548e-06, "loss": 1.3244, "step": 9797 }, { "epoch": 1.58, "grad_norm": 11.827006340026855, "learning_rate": 9.652471779585015e-06, "loss": 1.2233, "step": 9798 }, { "epoch": 1.58, "grad_norm": 8.999048233032227, "learning_rate": 9.650734843628966e-06, "loss": 0.7952, "step": 9799 }, { "epoch": 1.58, "grad_norm": 10.198712348937988, "learning_rate": 9.648997918222867e-06, "loss": 1.1989, "step": 9800 }, { "epoch": 1.58, "grad_norm": 9.82645034790039, "learning_rate": 9.647261003419179e-06, "loss": 1.0864, "step": 9801 }, { "epoch": 1.58, "grad_norm": 10.014256477355957, "learning_rate": 9.645524099270373e-06, "loss": 0.9298, "step": 9802 }, { "epoch": 1.58, "grad_norm": 8.628936767578125, "learning_rate": 9.64378720582891e-06, "loss": 0.9733, "step": 9803 }, { "epoch": 1.58, "grad_norm": 9.444310188293457, "learning_rate": 9.64205032314726e-06, "loss": 1.1036, "step": 9804 }, { "epoch": 1.58, "grad_norm": 11.678709983825684, "learning_rate": 9.640313451277882e-06, "loss": 1.0392, "step": 9805 }, { "epoch": 1.58, "grad_norm": 9.96900463104248, "learning_rate": 9.638576590273241e-06, "loss": 1.1745, "step": 9806 }, { "epoch": 1.58, "grad_norm": 8.650875091552734, "learning_rate": 9.636839740185805e-06, "loss": 1.1863, "step": 9807 }, { "epoch": 1.58, "grad_norm": 12.191363334655762, "learning_rate": 9.635102901068036e-06, "loss": 0.9077, "step": 9808 }, { "epoch": 1.58, "grad_norm": 7.059864044189453, "learning_rate": 9.633366072972394e-06, "loss": 0.9673, "step": 9809 }, { "epoch": 1.58, "grad_norm": 8.839533805847168, "learning_rate": 9.631629255951343e-06, "loss": 1.029, "step": 9810 }, { "epoch": 1.58, "grad_norm": 8.98092269897461, "learning_rate": 9.629892450057344e-06, "loss": 1.0292, "step": 9811 }, { "epoch": 1.58, "grad_norm": 7.501487731933594, "learning_rate": 9.628155655342863e-06, "loss": 0.9056, "step": 9812 }, { "epoch": 1.58, "grad_norm": 9.056244850158691, "learning_rate": 9.626418871860363e-06, "loss": 1.3314, "step": 9813 }, { "epoch": 1.58, "grad_norm": 10.092915534973145, "learning_rate": 9.624682099662302e-06, "loss": 0.9808, "step": 9814 }, { "epoch": 1.58, "grad_norm": 10.236181259155273, "learning_rate": 9.622945338801143e-06, "loss": 1.2369, "step": 9815 }, { "epoch": 1.58, "grad_norm": 9.048460006713867, "learning_rate": 9.621208589329343e-06, "loss": 1.138, "step": 9816 }, { "epoch": 1.58, "grad_norm": 9.311137199401855, "learning_rate": 9.619471851299368e-06, "loss": 0.9393, "step": 9817 }, { "epoch": 1.58, "grad_norm": 8.431495666503906, "learning_rate": 9.617735124763674e-06, "loss": 1.1599, "step": 9818 }, { "epoch": 1.58, "grad_norm": 9.497734069824219, "learning_rate": 9.615998409774725e-06, "loss": 0.961, "step": 9819 }, { "epoch": 1.58, "grad_norm": 10.32052230834961, "learning_rate": 9.614261706384975e-06, "loss": 1.1996, "step": 9820 }, { "epoch": 1.58, "grad_norm": 9.108368873596191, "learning_rate": 9.612525014646887e-06, "loss": 1.0832, "step": 9821 }, { "epoch": 1.58, "grad_norm": 7.988372325897217, "learning_rate": 9.610788334612917e-06, "loss": 1.0615, "step": 9822 }, { "epoch": 1.58, "grad_norm": 11.170912742614746, "learning_rate": 9.609051666335526e-06, "loss": 1.211, "step": 9823 }, { "epoch": 1.58, "grad_norm": 8.911822319030762, "learning_rate": 9.607315009867173e-06, "loss": 0.9461, "step": 9824 }, { "epoch": 1.58, "grad_norm": 8.20160961151123, "learning_rate": 9.605578365260312e-06, "loss": 1.1015, "step": 9825 }, { "epoch": 1.58, "grad_norm": 9.734532356262207, "learning_rate": 9.6038417325674e-06, "loss": 0.9391, "step": 9826 }, { "epoch": 1.58, "grad_norm": 9.81745433807373, "learning_rate": 9.602105111840898e-06, "loss": 0.8158, "step": 9827 }, { "epoch": 1.58, "grad_norm": 8.69641399383545, "learning_rate": 9.600368503133262e-06, "loss": 1.1543, "step": 9828 }, { "epoch": 1.58, "grad_norm": 7.838602542877197, "learning_rate": 9.598631906496945e-06, "loss": 1.0118, "step": 9829 }, { "epoch": 1.58, "grad_norm": 9.0143461227417, "learning_rate": 9.596895321984404e-06, "loss": 0.9062, "step": 9830 }, { "epoch": 1.58, "grad_norm": 10.564569473266602, "learning_rate": 9.595158749648095e-06, "loss": 1.1144, "step": 9831 }, { "epoch": 1.58, "grad_norm": 9.785386085510254, "learning_rate": 9.59342218954047e-06, "loss": 0.9907, "step": 9832 }, { "epoch": 1.58, "grad_norm": 8.617402076721191, "learning_rate": 9.591685641713991e-06, "loss": 0.959, "step": 9833 }, { "epoch": 1.58, "grad_norm": 12.395439147949219, "learning_rate": 9.589949106221106e-06, "loss": 0.9279, "step": 9834 }, { "epoch": 1.58, "grad_norm": 8.765829086303711, "learning_rate": 9.588212583114276e-06, "loss": 1.0633, "step": 9835 }, { "epoch": 1.58, "grad_norm": 10.83083724975586, "learning_rate": 9.586476072445947e-06, "loss": 1.182, "step": 9836 }, { "epoch": 1.58, "grad_norm": 9.889278411865234, "learning_rate": 9.584739574268575e-06, "loss": 0.94, "step": 9837 }, { "epoch": 1.58, "grad_norm": 12.311555862426758, "learning_rate": 9.583003088634613e-06, "loss": 1.0416, "step": 9838 }, { "epoch": 1.58, "grad_norm": 9.718318939208984, "learning_rate": 9.581266615596517e-06, "loss": 1.0419, "step": 9839 }, { "epoch": 1.58, "grad_norm": 9.30341625213623, "learning_rate": 9.579530155206733e-06, "loss": 1.1264, "step": 9840 }, { "epoch": 1.58, "grad_norm": 10.324909210205078, "learning_rate": 9.577793707517716e-06, "loss": 1.4214, "step": 9841 }, { "epoch": 1.58, "grad_norm": 9.81649398803711, "learning_rate": 9.576057272581916e-06, "loss": 1.062, "step": 9842 }, { "epoch": 1.58, "grad_norm": 9.60545539855957, "learning_rate": 9.574320850451788e-06, "loss": 0.9892, "step": 9843 }, { "epoch": 1.58, "grad_norm": 9.571098327636719, "learning_rate": 9.572584441179775e-06, "loss": 0.9439, "step": 9844 }, { "epoch": 1.58, "grad_norm": 10.484070777893066, "learning_rate": 9.570848044818334e-06, "loss": 1.2737, "step": 9845 }, { "epoch": 1.58, "grad_norm": 11.582733154296875, "learning_rate": 9.569111661419912e-06, "loss": 1.1233, "step": 9846 }, { "epoch": 1.59, "grad_norm": 7.283797264099121, "learning_rate": 9.56737529103696e-06, "loss": 0.9868, "step": 9847 }, { "epoch": 1.59, "grad_norm": 10.4609375, "learning_rate": 9.565638933721925e-06, "loss": 1.3209, "step": 9848 }, { "epoch": 1.59, "grad_norm": 8.29853343963623, "learning_rate": 9.563902589527258e-06, "loss": 1.0143, "step": 9849 }, { "epoch": 1.59, "grad_norm": 8.956411361694336, "learning_rate": 9.562166258505402e-06, "loss": 0.998, "step": 9850 }, { "epoch": 1.59, "grad_norm": 11.78260612487793, "learning_rate": 9.560429940708813e-06, "loss": 1.3662, "step": 9851 }, { "epoch": 1.59, "grad_norm": 10.7981538772583, "learning_rate": 9.55869363618993e-06, "loss": 1.0639, "step": 9852 }, { "epoch": 1.59, "grad_norm": 9.572795867919922, "learning_rate": 9.556957345001206e-06, "loss": 0.9482, "step": 9853 }, { "epoch": 1.59, "grad_norm": 10.457071304321289, "learning_rate": 9.555221067195087e-06, "loss": 1.1372, "step": 9854 }, { "epoch": 1.59, "grad_norm": 8.611754417419434, "learning_rate": 9.553484802824018e-06, "loss": 0.9623, "step": 9855 }, { "epoch": 1.59, "grad_norm": 8.600587844848633, "learning_rate": 9.551748551940443e-06, "loss": 1.1667, "step": 9856 }, { "epoch": 1.59, "grad_norm": 9.63795280456543, "learning_rate": 9.55001231459681e-06, "loss": 1.2501, "step": 9857 }, { "epoch": 1.59, "grad_norm": 12.661371231079102, "learning_rate": 9.548276090845565e-06, "loss": 0.9696, "step": 9858 }, { "epoch": 1.59, "grad_norm": 10.647643089294434, "learning_rate": 9.54653988073915e-06, "loss": 1.0033, "step": 9859 }, { "epoch": 1.59, "grad_norm": 9.042380332946777, "learning_rate": 9.544803684330009e-06, "loss": 0.8653, "step": 9860 }, { "epoch": 1.59, "grad_norm": 9.543395042419434, "learning_rate": 9.543067501670587e-06, "loss": 1.546, "step": 9861 }, { "epoch": 1.59, "grad_norm": 10.632137298583984, "learning_rate": 9.541331332813327e-06, "loss": 1.2244, "step": 9862 }, { "epoch": 1.59, "grad_norm": 9.228414535522461, "learning_rate": 9.539595177810675e-06, "loss": 1.0278, "step": 9863 }, { "epoch": 1.59, "grad_norm": 9.591316223144531, "learning_rate": 9.537859036715068e-06, "loss": 0.9595, "step": 9864 }, { "epoch": 1.59, "grad_norm": 9.566337585449219, "learning_rate": 9.536122909578951e-06, "loss": 1.1825, "step": 9865 }, { "epoch": 1.59, "grad_norm": 9.714930534362793, "learning_rate": 9.534386796454765e-06, "loss": 1.0589, "step": 9866 }, { "epoch": 1.59, "grad_norm": 10.857295036315918, "learning_rate": 9.532650697394954e-06, "loss": 1.0934, "step": 9867 }, { "epoch": 1.59, "grad_norm": 9.350980758666992, "learning_rate": 9.530914612451955e-06, "loss": 1.007, "step": 9868 }, { "epoch": 1.59, "grad_norm": 8.950355529785156, "learning_rate": 9.529178541678208e-06, "loss": 1.0857, "step": 9869 }, { "epoch": 1.59, "grad_norm": 9.296462059020996, "learning_rate": 9.527442485126157e-06, "loss": 0.9957, "step": 9870 }, { "epoch": 1.59, "grad_norm": 9.878778457641602, "learning_rate": 9.525706442848241e-06, "loss": 1.2863, "step": 9871 }, { "epoch": 1.59, "grad_norm": 10.421249389648438, "learning_rate": 9.52397041489689e-06, "loss": 1.3178, "step": 9872 }, { "epoch": 1.59, "grad_norm": 9.77691650390625, "learning_rate": 9.522234401324556e-06, "loss": 1.0358, "step": 9873 }, { "epoch": 1.59, "grad_norm": 9.568880081176758, "learning_rate": 9.520498402183674e-06, "loss": 1.0689, "step": 9874 }, { "epoch": 1.59, "grad_norm": 7.574872970581055, "learning_rate": 9.518762417526676e-06, "loss": 1.1585, "step": 9875 }, { "epoch": 1.59, "grad_norm": 9.93519115447998, "learning_rate": 9.517026447406003e-06, "loss": 0.8308, "step": 9876 }, { "epoch": 1.59, "grad_norm": 10.435273170471191, "learning_rate": 9.515290491874092e-06, "loss": 1.1419, "step": 9877 }, { "epoch": 1.59, "grad_norm": 9.916200637817383, "learning_rate": 9.513554550983381e-06, "loss": 0.8475, "step": 9878 }, { "epoch": 1.59, "grad_norm": 8.910568237304688, "learning_rate": 9.511818624786301e-06, "loss": 1.1195, "step": 9879 }, { "epoch": 1.59, "grad_norm": 8.227954864501953, "learning_rate": 9.510082713335293e-06, "loss": 1.119, "step": 9880 }, { "epoch": 1.59, "grad_norm": 9.40676498413086, "learning_rate": 9.508346816682789e-06, "loss": 1.024, "step": 9881 }, { "epoch": 1.59, "grad_norm": 8.766545295715332, "learning_rate": 9.506610934881227e-06, "loss": 0.7522, "step": 9882 }, { "epoch": 1.59, "grad_norm": 9.196215629577637, "learning_rate": 9.504875067983036e-06, "loss": 1.0612, "step": 9883 }, { "epoch": 1.59, "grad_norm": 9.974863052368164, "learning_rate": 9.503139216040653e-06, "loss": 0.9611, "step": 9884 }, { "epoch": 1.59, "grad_norm": 8.222906112670898, "learning_rate": 9.501403379106511e-06, "loss": 1.1948, "step": 9885 }, { "epoch": 1.59, "grad_norm": 9.54184627532959, "learning_rate": 9.499667557233045e-06, "loss": 1.3281, "step": 9886 }, { "epoch": 1.59, "grad_norm": 8.8922700881958, "learning_rate": 9.497931750472685e-06, "loss": 1.1141, "step": 9887 }, { "epoch": 1.59, "grad_norm": 9.101654052734375, "learning_rate": 9.496195958877861e-06, "loss": 0.947, "step": 9888 }, { "epoch": 1.59, "grad_norm": 8.708882331848145, "learning_rate": 9.49446018250101e-06, "loss": 1.4242, "step": 9889 }, { "epoch": 1.59, "grad_norm": 10.421089172363281, "learning_rate": 9.492724421394558e-06, "loss": 1.2111, "step": 9890 }, { "epoch": 1.59, "grad_norm": 10.410638809204102, "learning_rate": 9.490988675610941e-06, "loss": 0.9422, "step": 9891 }, { "epoch": 1.59, "grad_norm": 10.542848587036133, "learning_rate": 9.48925294520258e-06, "loss": 1.0268, "step": 9892 }, { "epoch": 1.59, "grad_norm": 11.034849166870117, "learning_rate": 9.487517230221915e-06, "loss": 1.2204, "step": 9893 }, { "epoch": 1.59, "grad_norm": 9.267524719238281, "learning_rate": 9.485781530721371e-06, "loss": 0.9958, "step": 9894 }, { "epoch": 1.59, "grad_norm": 9.384053230285645, "learning_rate": 9.484045846753376e-06, "loss": 1.2336, "step": 9895 }, { "epoch": 1.59, "grad_norm": 8.024710655212402, "learning_rate": 9.482310178370358e-06, "loss": 0.8338, "step": 9896 }, { "epoch": 1.59, "grad_norm": 10.345980644226074, "learning_rate": 9.480574525624748e-06, "loss": 1.149, "step": 9897 }, { "epoch": 1.59, "grad_norm": 10.01057243347168, "learning_rate": 9.47883888856897e-06, "loss": 0.9223, "step": 9898 }, { "epoch": 1.59, "grad_norm": 10.07332706451416, "learning_rate": 9.477103267255452e-06, "loss": 0.9788, "step": 9899 }, { "epoch": 1.59, "grad_norm": 8.457908630371094, "learning_rate": 9.47536766173662e-06, "loss": 0.7787, "step": 9900 }, { "epoch": 1.59, "grad_norm": 8.798223495483398, "learning_rate": 9.473632072064899e-06, "loss": 1.0357, "step": 9901 }, { "epoch": 1.59, "grad_norm": 9.154986381530762, "learning_rate": 9.471896498292719e-06, "loss": 0.9972, "step": 9902 }, { "epoch": 1.59, "grad_norm": 11.8653564453125, "learning_rate": 9.470160940472501e-06, "loss": 1.0896, "step": 9903 }, { "epoch": 1.59, "grad_norm": 11.572003364562988, "learning_rate": 9.468425398656667e-06, "loss": 1.0411, "step": 9904 }, { "epoch": 1.59, "grad_norm": 10.626205444335938, "learning_rate": 9.466689872897646e-06, "loss": 1.2242, "step": 9905 }, { "epoch": 1.59, "grad_norm": 10.269410133361816, "learning_rate": 9.464954363247863e-06, "loss": 0.9794, "step": 9906 }, { "epoch": 1.59, "grad_norm": 9.383099555969238, "learning_rate": 9.463218869759734e-06, "loss": 1.3536, "step": 9907 }, { "epoch": 1.59, "grad_norm": 9.825037956237793, "learning_rate": 9.461483392485687e-06, "loss": 0.8772, "step": 9908 }, { "epoch": 1.6, "grad_norm": 9.829222679138184, "learning_rate": 9.45974793147814e-06, "loss": 1.2935, "step": 9909 }, { "epoch": 1.6, "grad_norm": 8.510048866271973, "learning_rate": 9.45801248678952e-06, "loss": 1.0497, "step": 9910 }, { "epoch": 1.6, "grad_norm": 8.524959564208984, "learning_rate": 9.456277058472245e-06, "loss": 0.9883, "step": 9911 }, { "epoch": 1.6, "grad_norm": 8.841841697692871, "learning_rate": 9.454541646578732e-06, "loss": 0.9273, "step": 9912 }, { "epoch": 1.6, "grad_norm": 10.012062072753906, "learning_rate": 9.452806251161403e-06, "loss": 1.2708, "step": 9913 }, { "epoch": 1.6, "grad_norm": 10.17232894897461, "learning_rate": 9.451070872272684e-06, "loss": 1.3184, "step": 9914 }, { "epoch": 1.6, "grad_norm": 9.236360549926758, "learning_rate": 9.449335509964987e-06, "loss": 0.8901, "step": 9915 }, { "epoch": 1.6, "grad_norm": 7.809539794921875, "learning_rate": 9.447600164290733e-06, "loss": 1.0503, "step": 9916 }, { "epoch": 1.6, "grad_norm": 10.49884033203125, "learning_rate": 9.445864835302343e-06, "loss": 1.0308, "step": 9917 }, { "epoch": 1.6, "grad_norm": 9.098092079162598, "learning_rate": 9.44412952305223e-06, "loss": 0.9163, "step": 9918 }, { "epoch": 1.6, "grad_norm": 11.136625289916992, "learning_rate": 9.442394227592811e-06, "loss": 0.9552, "step": 9919 }, { "epoch": 1.6, "grad_norm": 10.48814582824707, "learning_rate": 9.440658948976504e-06, "loss": 1.2455, "step": 9920 }, { "epoch": 1.6, "grad_norm": 8.227246284484863, "learning_rate": 9.438923687255728e-06, "loss": 1.043, "step": 9921 }, { "epoch": 1.6, "grad_norm": 9.233478546142578, "learning_rate": 9.437188442482894e-06, "loss": 1.1379, "step": 9922 }, { "epoch": 1.6, "grad_norm": 7.789116382598877, "learning_rate": 9.435453214710419e-06, "loss": 1.1722, "step": 9923 }, { "epoch": 1.6, "grad_norm": 8.837724685668945, "learning_rate": 9.433718003990716e-06, "loss": 1.0981, "step": 9924 }, { "epoch": 1.6, "grad_norm": 9.690499305725098, "learning_rate": 9.431982810376204e-06, "loss": 1.0632, "step": 9925 }, { "epoch": 1.6, "grad_norm": 10.61090087890625, "learning_rate": 9.43024763391929e-06, "loss": 0.8608, "step": 9926 }, { "epoch": 1.6, "grad_norm": 9.553791999816895, "learning_rate": 9.428512474672389e-06, "loss": 1.0581, "step": 9927 }, { "epoch": 1.6, "grad_norm": 9.796736717224121, "learning_rate": 9.426777332687914e-06, "loss": 1.0739, "step": 9928 }, { "epoch": 1.6, "grad_norm": 11.081151008605957, "learning_rate": 9.425042208018278e-06, "loss": 1.1614, "step": 9929 }, { "epoch": 1.6, "grad_norm": 9.135129928588867, "learning_rate": 9.423307100715893e-06, "loss": 1.0053, "step": 9930 }, { "epoch": 1.6, "grad_norm": 10.753915786743164, "learning_rate": 9.421572010833167e-06, "loss": 1.158, "step": 9931 }, { "epoch": 1.6, "grad_norm": 9.323060035705566, "learning_rate": 9.419836938422511e-06, "loss": 0.9935, "step": 9932 }, { "epoch": 1.6, "grad_norm": 8.937492370605469, "learning_rate": 9.418101883536332e-06, "loss": 1.0857, "step": 9933 }, { "epoch": 1.6, "grad_norm": 12.272089958190918, "learning_rate": 9.416366846227047e-06, "loss": 1.0681, "step": 9934 }, { "epoch": 1.6, "grad_norm": 11.14603042602539, "learning_rate": 9.41463182654706e-06, "loss": 1.1081, "step": 9935 }, { "epoch": 1.6, "grad_norm": 8.384875297546387, "learning_rate": 9.412896824548783e-06, "loss": 0.913, "step": 9936 }, { "epoch": 1.6, "grad_norm": 9.105727195739746, "learning_rate": 9.411161840284618e-06, "loss": 0.8604, "step": 9937 }, { "epoch": 1.6, "grad_norm": 9.3529691696167, "learning_rate": 9.409426873806972e-06, "loss": 1.1004, "step": 9938 }, { "epoch": 1.6, "grad_norm": 8.435514450073242, "learning_rate": 9.407691925168257e-06, "loss": 0.9191, "step": 9939 }, { "epoch": 1.6, "grad_norm": 9.899925231933594, "learning_rate": 9.405956994420877e-06, "loss": 1.1238, "step": 9940 }, { "epoch": 1.6, "grad_norm": 9.629050254821777, "learning_rate": 9.404222081617238e-06, "loss": 1.1189, "step": 9941 }, { "epoch": 1.6, "grad_norm": 8.754836082458496, "learning_rate": 9.40248718680974e-06, "loss": 1.0238, "step": 9942 }, { "epoch": 1.6, "grad_norm": 12.045461654663086, "learning_rate": 9.400752310050793e-06, "loss": 1.151, "step": 9943 }, { "epoch": 1.6, "grad_norm": 9.681299209594727, "learning_rate": 9.3990174513928e-06, "loss": 1.2509, "step": 9944 }, { "epoch": 1.6, "grad_norm": 10.480367660522461, "learning_rate": 9.397282610888164e-06, "loss": 1.0759, "step": 9945 }, { "epoch": 1.6, "grad_norm": 11.1443510055542, "learning_rate": 9.395547788589287e-06, "loss": 1.2455, "step": 9946 }, { "epoch": 1.6, "grad_norm": 10.322112083435059, "learning_rate": 9.393812984548571e-06, "loss": 1.1206, "step": 9947 }, { "epoch": 1.6, "grad_norm": 10.021554946899414, "learning_rate": 9.392078198818419e-06, "loss": 1.1278, "step": 9948 }, { "epoch": 1.6, "grad_norm": 12.146849632263184, "learning_rate": 9.390343431451232e-06, "loss": 1.105, "step": 9949 }, { "epoch": 1.6, "grad_norm": 8.725370407104492, "learning_rate": 9.38860868249941e-06, "loss": 1.0049, "step": 9950 }, { "epoch": 1.6, "grad_norm": 9.882455825805664, "learning_rate": 9.386873952015354e-06, "loss": 1.2057, "step": 9951 }, { "epoch": 1.6, "grad_norm": 7.472190856933594, "learning_rate": 9.385139240051459e-06, "loss": 0.8042, "step": 9952 }, { "epoch": 1.6, "grad_norm": 11.713898658752441, "learning_rate": 9.38340454666013e-06, "loss": 1.0245, "step": 9953 }, { "epoch": 1.6, "grad_norm": 12.255290031433105, "learning_rate": 9.381669871893761e-06, "loss": 1.1259, "step": 9954 }, { "epoch": 1.6, "grad_norm": 10.62094783782959, "learning_rate": 9.379935215804756e-06, "loss": 1.2225, "step": 9955 }, { "epoch": 1.6, "grad_norm": 9.851110458374023, "learning_rate": 9.378200578445506e-06, "loss": 0.9922, "step": 9956 }, { "epoch": 1.6, "grad_norm": 7.658028602600098, "learning_rate": 9.37646595986841e-06, "loss": 0.8821, "step": 9957 }, { "epoch": 1.6, "grad_norm": 8.886914253234863, "learning_rate": 9.374731360125864e-06, "loss": 1.0773, "step": 9958 }, { "epoch": 1.6, "grad_norm": 8.26418685913086, "learning_rate": 9.372996779270263e-06, "loss": 1.1891, "step": 9959 }, { "epoch": 1.6, "grad_norm": 9.757431030273438, "learning_rate": 9.371262217354006e-06, "loss": 0.9667, "step": 9960 }, { "epoch": 1.6, "grad_norm": 7.690340995788574, "learning_rate": 9.36952767442948e-06, "loss": 1.0299, "step": 9961 }, { "epoch": 1.6, "grad_norm": 7.824804306030273, "learning_rate": 9.367793150549084e-06, "loss": 0.9696, "step": 9962 }, { "epoch": 1.6, "grad_norm": 9.437699317932129, "learning_rate": 9.366058645765209e-06, "loss": 1.0483, "step": 9963 }, { "epoch": 1.6, "grad_norm": 10.412949562072754, "learning_rate": 9.36432416013025e-06, "loss": 1.0193, "step": 9964 }, { "epoch": 1.6, "grad_norm": 11.14052963256836, "learning_rate": 9.362589693696598e-06, "loss": 1.1466, "step": 9965 }, { "epoch": 1.6, "grad_norm": 8.571157455444336, "learning_rate": 9.360855246516643e-06, "loss": 0.973, "step": 9966 }, { "epoch": 1.6, "grad_norm": 9.866808891296387, "learning_rate": 9.359120818642776e-06, "loss": 1.0159, "step": 9967 }, { "epoch": 1.6, "grad_norm": 8.466739654541016, "learning_rate": 9.35738641012739e-06, "loss": 0.9868, "step": 9968 }, { "epoch": 1.6, "grad_norm": 9.629084587097168, "learning_rate": 9.355652021022875e-06, "loss": 0.8385, "step": 9969 }, { "epoch": 1.6, "grad_norm": 9.117419242858887, "learning_rate": 9.353917651381615e-06, "loss": 1.2035, "step": 9970 }, { "epoch": 1.61, "grad_norm": 9.580408096313477, "learning_rate": 9.352183301256004e-06, "loss": 0.8736, "step": 9971 }, { "epoch": 1.61, "grad_norm": 9.504549980163574, "learning_rate": 9.350448970698427e-06, "loss": 1.0833, "step": 9972 }, { "epoch": 1.61, "grad_norm": 10.133064270019531, "learning_rate": 9.34871465976127e-06, "loss": 0.847, "step": 9973 }, { "epoch": 1.61, "grad_norm": 8.07766342163086, "learning_rate": 9.346980368496926e-06, "loss": 1.0529, "step": 9974 }, { "epoch": 1.61, "grad_norm": 7.4618449211120605, "learning_rate": 9.345246096957777e-06, "loss": 1.0099, "step": 9975 }, { "epoch": 1.61, "grad_norm": 8.620978355407715, "learning_rate": 9.343511845196208e-06, "loss": 0.9329, "step": 9976 }, { "epoch": 1.61, "grad_norm": 9.287510871887207, "learning_rate": 9.341777613264606e-06, "loss": 1.0302, "step": 9977 }, { "epoch": 1.61, "grad_norm": 9.824735641479492, "learning_rate": 9.340043401215354e-06, "loss": 0.8337, "step": 9978 }, { "epoch": 1.61, "grad_norm": 12.252457618713379, "learning_rate": 9.338309209100838e-06, "loss": 1.2231, "step": 9979 }, { "epoch": 1.61, "grad_norm": 10.437310218811035, "learning_rate": 9.33657503697344e-06, "loss": 0.9055, "step": 9980 }, { "epoch": 1.61, "grad_norm": 9.165313720703125, "learning_rate": 9.33484088488554e-06, "loss": 1.2403, "step": 9981 }, { "epoch": 1.61, "grad_norm": 10.00022029876709, "learning_rate": 9.333106752889525e-06, "loss": 1.0856, "step": 9982 }, { "epoch": 1.61, "grad_norm": 11.03398323059082, "learning_rate": 9.331372641037772e-06, "loss": 1.4351, "step": 9983 }, { "epoch": 1.61, "grad_norm": 9.097747802734375, "learning_rate": 9.329638549382664e-06, "loss": 1.1007, "step": 9984 }, { "epoch": 1.61, "grad_norm": 9.240408897399902, "learning_rate": 9.32790447797658e-06, "loss": 0.9639, "step": 9985 }, { "epoch": 1.61, "grad_norm": 9.394697189331055, "learning_rate": 9.326170426871902e-06, "loss": 0.8051, "step": 9986 }, { "epoch": 1.61, "grad_norm": 9.20727252960205, "learning_rate": 9.324436396121004e-06, "loss": 1.0069, "step": 9987 }, { "epoch": 1.61, "grad_norm": 12.941003799438477, "learning_rate": 9.322702385776272e-06, "loss": 1.347, "step": 9988 }, { "epoch": 1.61, "grad_norm": 10.177952766418457, "learning_rate": 9.320968395890077e-06, "loss": 1.2603, "step": 9989 }, { "epoch": 1.61, "grad_norm": 11.844254493713379, "learning_rate": 9.319234426514798e-06, "loss": 1.2866, "step": 9990 }, { "epoch": 1.61, "grad_norm": 12.919477462768555, "learning_rate": 9.317500477702813e-06, "loss": 1.2405, "step": 9991 }, { "epoch": 1.61, "grad_norm": 12.622065544128418, "learning_rate": 9.315766549506499e-06, "loss": 1.0172, "step": 9992 }, { "epoch": 1.61, "grad_norm": 9.771249771118164, "learning_rate": 9.314032641978226e-06, "loss": 1.11, "step": 9993 }, { "epoch": 1.61, "grad_norm": 11.77660083770752, "learning_rate": 9.312298755170368e-06, "loss": 1.0204, "step": 9994 }, { "epoch": 1.61, "grad_norm": 9.74564266204834, "learning_rate": 9.310564889135308e-06, "loss": 1.1617, "step": 9995 }, { "epoch": 1.61, "grad_norm": 12.680848121643066, "learning_rate": 9.308831043925414e-06, "loss": 0.9855, "step": 9996 }, { "epoch": 1.61, "grad_norm": 8.375446319580078, "learning_rate": 9.307097219593057e-06, "loss": 0.8315, "step": 9997 }, { "epoch": 1.61, "grad_norm": 8.158291816711426, "learning_rate": 9.305363416190614e-06, "loss": 0.7664, "step": 9998 }, { "epoch": 1.61, "grad_norm": 7.807108402252197, "learning_rate": 9.303629633770454e-06, "loss": 0.9984, "step": 9999 }, { "epoch": 1.61, "grad_norm": 9.793712615966797, "learning_rate": 9.301895872384944e-06, "loss": 1.099, "step": 10000 }, { "epoch": 1.61, "grad_norm": 10.052451133728027, "learning_rate": 9.300162132086459e-06, "loss": 1.1387, "step": 10001 }, { "epoch": 1.61, "grad_norm": 10.965188980102539, "learning_rate": 9.298428412927368e-06, "loss": 0.9947, "step": 10002 }, { "epoch": 1.61, "grad_norm": 10.391338348388672, "learning_rate": 9.29669471496004e-06, "loss": 0.9898, "step": 10003 }, { "epoch": 1.61, "grad_norm": 12.38438892364502, "learning_rate": 9.294961038236841e-06, "loss": 0.9949, "step": 10004 }, { "epoch": 1.61, "grad_norm": 9.928091049194336, "learning_rate": 9.293227382810139e-06, "loss": 1.1187, "step": 10005 }, { "epoch": 1.61, "grad_norm": 9.479415893554688, "learning_rate": 9.291493748732304e-06, "loss": 0.8558, "step": 10006 }, { "epoch": 1.61, "grad_norm": 8.390427589416504, "learning_rate": 9.289760136055698e-06, "loss": 1.0144, "step": 10007 }, { "epoch": 1.61, "grad_norm": 8.318469047546387, "learning_rate": 9.288026544832694e-06, "loss": 1.193, "step": 10008 }, { "epoch": 1.61, "grad_norm": 9.335315704345703, "learning_rate": 9.286292975115648e-06, "loss": 1.0202, "step": 10009 }, { "epoch": 1.61, "grad_norm": 8.020299911499023, "learning_rate": 9.28455942695693e-06, "loss": 1.0237, "step": 10010 }, { "epoch": 1.61, "grad_norm": 9.944053649902344, "learning_rate": 9.282825900408903e-06, "loss": 1.0712, "step": 10011 }, { "epoch": 1.61, "grad_norm": 9.598895072937012, "learning_rate": 9.28109239552393e-06, "loss": 1.0176, "step": 10012 }, { "epoch": 1.61, "grad_norm": 8.7806978225708, "learning_rate": 9.279358912354375e-06, "loss": 0.9011, "step": 10013 }, { "epoch": 1.61, "grad_norm": 7.9702653884887695, "learning_rate": 9.277625450952592e-06, "loss": 1.0068, "step": 10014 }, { "epoch": 1.61, "grad_norm": 11.272817611694336, "learning_rate": 9.27589201137095e-06, "loss": 1.055, "step": 10015 }, { "epoch": 1.61, "grad_norm": 9.683531761169434, "learning_rate": 9.27415859366181e-06, "loss": 1.0541, "step": 10016 }, { "epoch": 1.61, "grad_norm": 10.628477096557617, "learning_rate": 9.272425197877528e-06, "loss": 0.9162, "step": 10017 }, { "epoch": 1.61, "grad_norm": 8.761810302734375, "learning_rate": 9.270691824070465e-06, "loss": 0.9228, "step": 10018 }, { "epoch": 1.61, "grad_norm": 9.710772514343262, "learning_rate": 9.268958472292982e-06, "loss": 1.1429, "step": 10019 }, { "epoch": 1.61, "grad_norm": 8.08480453491211, "learning_rate": 9.267225142597431e-06, "loss": 0.8424, "step": 10020 }, { "epoch": 1.61, "grad_norm": 9.52198600769043, "learning_rate": 9.265491835036172e-06, "loss": 0.9493, "step": 10021 }, { "epoch": 1.61, "grad_norm": 9.989058494567871, "learning_rate": 9.263758549661563e-06, "loss": 1.2609, "step": 10022 }, { "epoch": 1.61, "grad_norm": 12.455161094665527, "learning_rate": 9.26202528652596e-06, "loss": 1.1719, "step": 10023 }, { "epoch": 1.61, "grad_norm": 9.430258750915527, "learning_rate": 9.260292045681713e-06, "loss": 1.2137, "step": 10024 }, { "epoch": 1.61, "grad_norm": 9.154953956604004, "learning_rate": 9.258558827181182e-06, "loss": 1.0247, "step": 10025 }, { "epoch": 1.61, "grad_norm": 10.118905067443848, "learning_rate": 9.25682563107672e-06, "loss": 1.1484, "step": 10026 }, { "epoch": 1.61, "grad_norm": 8.951108932495117, "learning_rate": 9.255092457420679e-06, "loss": 0.9581, "step": 10027 }, { "epoch": 1.61, "grad_norm": 11.820976257324219, "learning_rate": 9.25335930626541e-06, "loss": 1.232, "step": 10028 }, { "epoch": 1.61, "grad_norm": 11.148765563964844, "learning_rate": 9.251626177663266e-06, "loss": 0.999, "step": 10029 }, { "epoch": 1.61, "grad_norm": 10.885052680969238, "learning_rate": 9.2498930716666e-06, "loss": 1.2874, "step": 10030 }, { "epoch": 1.61, "grad_norm": 8.016573905944824, "learning_rate": 9.248159988327764e-06, "loss": 0.8289, "step": 10031 }, { "epoch": 1.61, "grad_norm": 10.714200973510742, "learning_rate": 9.2464269276991e-06, "loss": 1.2722, "step": 10032 }, { "epoch": 1.62, "grad_norm": 8.502298355102539, "learning_rate": 9.244693889832964e-06, "loss": 0.9794, "step": 10033 }, { "epoch": 1.62, "grad_norm": 8.892507553100586, "learning_rate": 9.242960874781697e-06, "loss": 0.9109, "step": 10034 }, { "epoch": 1.62, "grad_norm": 9.542866706848145, "learning_rate": 9.241227882597656e-06, "loss": 0.9549, "step": 10035 }, { "epoch": 1.62, "grad_norm": 8.605606079101562, "learning_rate": 9.239494913333183e-06, "loss": 1.0989, "step": 10036 }, { "epoch": 1.62, "grad_norm": 11.906689643859863, "learning_rate": 9.237761967040626e-06, "loss": 1.1497, "step": 10037 }, { "epoch": 1.62, "grad_norm": 10.498234748840332, "learning_rate": 9.236029043772332e-06, "loss": 0.8307, "step": 10038 }, { "epoch": 1.62, "grad_norm": 8.748475074768066, "learning_rate": 9.234296143580641e-06, "loss": 0.9879, "step": 10039 }, { "epoch": 1.62, "grad_norm": 9.702882766723633, "learning_rate": 9.2325632665179e-06, "loss": 0.89, "step": 10040 }, { "epoch": 1.62, "grad_norm": 10.737500190734863, "learning_rate": 9.23083041263645e-06, "loss": 1.1304, "step": 10041 }, { "epoch": 1.62, "grad_norm": 8.538617134094238, "learning_rate": 9.229097581988641e-06, "loss": 0.7868, "step": 10042 }, { "epoch": 1.62, "grad_norm": 9.109926223754883, "learning_rate": 9.227364774626808e-06, "loss": 0.9249, "step": 10043 }, { "epoch": 1.62, "grad_norm": 9.329023361206055, "learning_rate": 9.225631990603294e-06, "loss": 1.0356, "step": 10044 }, { "epoch": 1.62, "grad_norm": 8.747638702392578, "learning_rate": 9.22389922997044e-06, "loss": 0.9103, "step": 10045 }, { "epoch": 1.62, "grad_norm": 9.034165382385254, "learning_rate": 9.222166492780587e-06, "loss": 1.2671, "step": 10046 }, { "epoch": 1.62, "grad_norm": 9.731645584106445, "learning_rate": 9.220433779086076e-06, "loss": 1.2149, "step": 10047 }, { "epoch": 1.62, "grad_norm": 11.904221534729004, "learning_rate": 9.218701088939238e-06, "loss": 1.1811, "step": 10048 }, { "epoch": 1.62, "grad_norm": 8.541128158569336, "learning_rate": 9.21696842239242e-06, "loss": 1.1488, "step": 10049 }, { "epoch": 1.62, "grad_norm": 9.531354904174805, "learning_rate": 9.215235779497953e-06, "loss": 0.8729, "step": 10050 }, { "epoch": 1.62, "grad_norm": 10.061274528503418, "learning_rate": 9.213503160308179e-06, "loss": 1.3296, "step": 10051 }, { "epoch": 1.62, "grad_norm": 9.409268379211426, "learning_rate": 9.211770564875425e-06, "loss": 0.9116, "step": 10052 }, { "epoch": 1.62, "grad_norm": 11.946135520935059, "learning_rate": 9.210037993252036e-06, "loss": 1.0498, "step": 10053 }, { "epoch": 1.62, "grad_norm": 8.102170944213867, "learning_rate": 9.208305445490336e-06, "loss": 0.9417, "step": 10054 }, { "epoch": 1.62, "grad_norm": 12.25761890411377, "learning_rate": 9.206572921642668e-06, "loss": 1.2195, "step": 10055 }, { "epoch": 1.62, "grad_norm": 8.624197959899902, "learning_rate": 9.20484042176136e-06, "loss": 1.0449, "step": 10056 }, { "epoch": 1.62, "grad_norm": 12.40884780883789, "learning_rate": 9.203107945898747e-06, "loss": 1.1089, "step": 10057 }, { "epoch": 1.62, "grad_norm": 9.332245826721191, "learning_rate": 9.20137549410716e-06, "loss": 0.8842, "step": 10058 }, { "epoch": 1.62, "grad_norm": 11.279446601867676, "learning_rate": 9.199643066438925e-06, "loss": 1.078, "step": 10059 }, { "epoch": 1.62, "grad_norm": 10.142560958862305, "learning_rate": 9.197910662946375e-06, "loss": 1.1218, "step": 10060 }, { "epoch": 1.62, "grad_norm": 10.694259643554688, "learning_rate": 9.19617828368184e-06, "loss": 1.1284, "step": 10061 }, { "epoch": 1.62, "grad_norm": 8.280013084411621, "learning_rate": 9.194445928697651e-06, "loss": 0.8125, "step": 10062 }, { "epoch": 1.62, "grad_norm": 10.168797492980957, "learning_rate": 9.192713598046128e-06, "loss": 1.0738, "step": 10063 }, { "epoch": 1.62, "grad_norm": 9.16048812866211, "learning_rate": 9.190981291779605e-06, "loss": 0.9379, "step": 10064 }, { "epoch": 1.62, "grad_norm": 11.502350807189941, "learning_rate": 9.189249009950403e-06, "loss": 1.1947, "step": 10065 }, { "epoch": 1.62, "grad_norm": 10.32861328125, "learning_rate": 9.187516752610856e-06, "loss": 1.1656, "step": 10066 }, { "epoch": 1.62, "grad_norm": 10.785831451416016, "learning_rate": 9.185784519813278e-06, "loss": 1.1244, "step": 10067 }, { "epoch": 1.62, "grad_norm": 8.635169982910156, "learning_rate": 9.18405231161e-06, "loss": 1.2553, "step": 10068 }, { "epoch": 1.62, "grad_norm": 10.130090713500977, "learning_rate": 9.182320128053343e-06, "loss": 0.8882, "step": 10069 }, { "epoch": 1.62, "grad_norm": 10.618176460266113, "learning_rate": 9.180587969195631e-06, "loss": 0.9449, "step": 10070 }, { "epoch": 1.62, "grad_norm": 10.63758659362793, "learning_rate": 9.178855835089184e-06, "loss": 1.0512, "step": 10071 }, { "epoch": 1.62, "grad_norm": 9.140970230102539, "learning_rate": 9.177123725786324e-06, "loss": 1.0281, "step": 10072 }, { "epoch": 1.62, "grad_norm": 9.755851745605469, "learning_rate": 9.17539164133937e-06, "loss": 1.188, "step": 10073 }, { "epoch": 1.62, "grad_norm": 8.038928985595703, "learning_rate": 9.173659581800644e-06, "loss": 0.8968, "step": 10074 }, { "epoch": 1.62, "grad_norm": 7.9647979736328125, "learning_rate": 9.17192754722246e-06, "loss": 1.0361, "step": 10075 }, { "epoch": 1.62, "grad_norm": 9.801630973815918, "learning_rate": 9.170195537657145e-06, "loss": 1.1391, "step": 10076 }, { "epoch": 1.62, "grad_norm": 8.772465705871582, "learning_rate": 9.168463553157012e-06, "loss": 0.9223, "step": 10077 }, { "epoch": 1.62, "grad_norm": 8.582460403442383, "learning_rate": 9.166731593774371e-06, "loss": 1.1882, "step": 10078 }, { "epoch": 1.62, "grad_norm": 11.081696510314941, "learning_rate": 9.164999659561547e-06, "loss": 1.2155, "step": 10079 }, { "epoch": 1.62, "grad_norm": 8.823257446289062, "learning_rate": 9.16326775057085e-06, "loss": 0.8521, "step": 10080 }, { "epoch": 1.62, "grad_norm": 10.15770435333252, "learning_rate": 9.161535866854598e-06, "loss": 1.0512, "step": 10081 }, { "epoch": 1.62, "grad_norm": 10.439329147338867, "learning_rate": 9.1598040084651e-06, "loss": 1.0132, "step": 10082 }, { "epoch": 1.62, "grad_norm": 10.607532501220703, "learning_rate": 9.15807217545467e-06, "loss": 1.1928, "step": 10083 }, { "epoch": 1.62, "grad_norm": 7.8877458572387695, "learning_rate": 9.156340367875623e-06, "loss": 0.9867, "step": 10084 }, { "epoch": 1.62, "grad_norm": 8.355027198791504, "learning_rate": 9.154608585780267e-06, "loss": 0.8256, "step": 10085 }, { "epoch": 1.62, "grad_norm": 11.117722511291504, "learning_rate": 9.152876829220914e-06, "loss": 1.0969, "step": 10086 }, { "epoch": 1.62, "grad_norm": 11.39747142791748, "learning_rate": 9.151145098249873e-06, "loss": 1.0268, "step": 10087 }, { "epoch": 1.62, "grad_norm": 8.636739730834961, "learning_rate": 9.149413392919453e-06, "loss": 1.0582, "step": 10088 }, { "epoch": 1.62, "grad_norm": 8.975168228149414, "learning_rate": 9.147681713281961e-06, "loss": 1.0213, "step": 10089 }, { "epoch": 1.62, "grad_norm": 8.970503807067871, "learning_rate": 9.145950059389708e-06, "loss": 1.1208, "step": 10090 }, { "epoch": 1.62, "grad_norm": 11.83212947845459, "learning_rate": 9.144218431294996e-06, "loss": 1.127, "step": 10091 }, { "epoch": 1.62, "grad_norm": 9.909232139587402, "learning_rate": 9.142486829050134e-06, "loss": 1.2244, "step": 10092 }, { "epoch": 1.62, "grad_norm": 11.396310806274414, "learning_rate": 9.140755252707424e-06, "loss": 0.9108, "step": 10093 }, { "epoch": 1.62, "grad_norm": 8.517279624938965, "learning_rate": 9.139023702319175e-06, "loss": 1.1074, "step": 10094 }, { "epoch": 1.63, "grad_norm": 9.196682929992676, "learning_rate": 9.137292177937681e-06, "loss": 1.0069, "step": 10095 }, { "epoch": 1.63, "grad_norm": 10.774308204650879, "learning_rate": 9.135560679615256e-06, "loss": 1.0787, "step": 10096 }, { "epoch": 1.63, "grad_norm": 10.379582405090332, "learning_rate": 9.133829207404198e-06, "loss": 1.0275, "step": 10097 }, { "epoch": 1.63, "grad_norm": 8.924412727355957, "learning_rate": 9.132097761356805e-06, "loss": 1.1082, "step": 10098 }, { "epoch": 1.63, "grad_norm": 9.529256820678711, "learning_rate": 9.13036634152538e-06, "loss": 0.905, "step": 10099 }, { "epoch": 1.63, "grad_norm": 10.202817916870117, "learning_rate": 9.128634947962221e-06, "loss": 1.4121, "step": 10100 }, { "epoch": 1.63, "grad_norm": 10.37219524383545, "learning_rate": 9.126903580719631e-06, "loss": 1.2357, "step": 10101 }, { "epoch": 1.63, "grad_norm": 11.753195762634277, "learning_rate": 9.1251722398499e-06, "loss": 1.2611, "step": 10102 }, { "epoch": 1.63, "grad_norm": 9.767892837524414, "learning_rate": 9.123440925405332e-06, "loss": 1.0943, "step": 10103 }, { "epoch": 1.63, "grad_norm": 10.726329803466797, "learning_rate": 9.121709637438221e-06, "loss": 1.2605, "step": 10104 }, { "epoch": 1.63, "grad_norm": 8.273752212524414, "learning_rate": 9.119978376000863e-06, "loss": 0.9116, "step": 10105 }, { "epoch": 1.63, "grad_norm": 8.239296913146973, "learning_rate": 9.11824714114555e-06, "loss": 0.8365, "step": 10106 }, { "epoch": 1.63, "grad_norm": 9.673738479614258, "learning_rate": 9.11651593292458e-06, "loss": 1.3789, "step": 10107 }, { "epoch": 1.63, "grad_norm": 9.06603717803955, "learning_rate": 9.114784751390241e-06, "loss": 1.2607, "step": 10108 }, { "epoch": 1.63, "grad_norm": 7.945883274078369, "learning_rate": 9.113053596594833e-06, "loss": 0.8652, "step": 10109 }, { "epoch": 1.63, "grad_norm": 8.584768295288086, "learning_rate": 9.11132246859064e-06, "loss": 1.2035, "step": 10110 }, { "epoch": 1.63, "grad_norm": 9.600319862365723, "learning_rate": 9.109591367429955e-06, "loss": 1.0649, "step": 10111 }, { "epoch": 1.63, "grad_norm": 9.345719337463379, "learning_rate": 9.107860293165068e-06, "loss": 0.8174, "step": 10112 }, { "epoch": 1.63, "grad_norm": 8.782646179199219, "learning_rate": 9.10612924584827e-06, "loss": 0.8443, "step": 10113 }, { "epoch": 1.63, "grad_norm": 9.26189136505127, "learning_rate": 9.104398225531848e-06, "loss": 0.9186, "step": 10114 }, { "epoch": 1.63, "grad_norm": 10.70612621307373, "learning_rate": 9.102667232268084e-06, "loss": 1.2676, "step": 10115 }, { "epoch": 1.63, "grad_norm": 8.839248657226562, "learning_rate": 9.100936266109276e-06, "loss": 0.9221, "step": 10116 }, { "epoch": 1.63, "grad_norm": 11.530211448669434, "learning_rate": 9.099205327107702e-06, "loss": 1.08, "step": 10117 }, { "epoch": 1.63, "grad_norm": 9.652966499328613, "learning_rate": 9.097474415315647e-06, "loss": 1.2825, "step": 10118 }, { "epoch": 1.63, "grad_norm": 8.603626251220703, "learning_rate": 9.095743530785396e-06, "loss": 1.2469, "step": 10119 }, { "epoch": 1.63, "grad_norm": 9.24801254272461, "learning_rate": 9.094012673569239e-06, "loss": 1.0347, "step": 10120 }, { "epoch": 1.63, "grad_norm": 8.385614395141602, "learning_rate": 9.092281843719447e-06, "loss": 1.1371, "step": 10121 }, { "epoch": 1.63, "grad_norm": 11.089197158813477, "learning_rate": 9.090551041288308e-06, "loss": 1.2373, "step": 10122 }, { "epoch": 1.63, "grad_norm": 10.480076789855957, "learning_rate": 9.0888202663281e-06, "loss": 1.1003, "step": 10123 }, { "epoch": 1.63, "grad_norm": 9.543375968933105, "learning_rate": 9.087089518891108e-06, "loss": 1.1417, "step": 10124 }, { "epoch": 1.63, "grad_norm": 11.002969741821289, "learning_rate": 9.08535879902961e-06, "loss": 1.0907, "step": 10125 }, { "epoch": 1.63, "grad_norm": 8.008588790893555, "learning_rate": 9.083628106795878e-06, "loss": 0.9396, "step": 10126 }, { "epoch": 1.63, "grad_norm": 10.671624183654785, "learning_rate": 9.081897442242196e-06, "loss": 1.2105, "step": 10127 }, { "epoch": 1.63, "grad_norm": 9.542630195617676, "learning_rate": 9.080166805420838e-06, "loss": 1.1133, "step": 10128 }, { "epoch": 1.63, "grad_norm": 9.278451919555664, "learning_rate": 9.078436196384083e-06, "loss": 1.27, "step": 10129 }, { "epoch": 1.63, "grad_norm": 8.732817649841309, "learning_rate": 9.0767056151842e-06, "loss": 0.9071, "step": 10130 }, { "epoch": 1.63, "grad_norm": 10.08709716796875, "learning_rate": 9.074975061873468e-06, "loss": 0.8851, "step": 10131 }, { "epoch": 1.63, "grad_norm": 11.2030611038208, "learning_rate": 9.07324453650416e-06, "loss": 1.199, "step": 10132 }, { "epoch": 1.63, "grad_norm": 8.940796852111816, "learning_rate": 9.071514039128548e-06, "loss": 1.058, "step": 10133 }, { "epoch": 1.63, "grad_norm": 10.021181106567383, "learning_rate": 9.069783569798902e-06, "loss": 1.0965, "step": 10134 }, { "epoch": 1.63, "grad_norm": 7.222158432006836, "learning_rate": 9.068053128567494e-06, "loss": 0.8653, "step": 10135 }, { "epoch": 1.63, "grad_norm": 10.674227714538574, "learning_rate": 9.06632271548659e-06, "loss": 1.356, "step": 10136 }, { "epoch": 1.63, "grad_norm": 7.466738700866699, "learning_rate": 9.064592330608466e-06, "loss": 1.145, "step": 10137 }, { "epoch": 1.63, "grad_norm": 10.461626052856445, "learning_rate": 9.062861973985387e-06, "loss": 1.0157, "step": 10138 }, { "epoch": 1.63, "grad_norm": 10.022461891174316, "learning_rate": 9.061131645669621e-06, "loss": 1.2067, "step": 10139 }, { "epoch": 1.63, "grad_norm": 11.490577697753906, "learning_rate": 9.059401345713436e-06, "loss": 1.193, "step": 10140 }, { "epoch": 1.63, "grad_norm": 9.559431076049805, "learning_rate": 9.057671074169094e-06, "loss": 1.0048, "step": 10141 }, { "epoch": 1.63, "grad_norm": 8.229459762573242, "learning_rate": 9.05594083108886e-06, "loss": 1.0524, "step": 10142 }, { "epoch": 1.63, "grad_norm": 9.830191612243652, "learning_rate": 9.054210616525e-06, "loss": 0.9192, "step": 10143 }, { "epoch": 1.63, "grad_norm": 9.327859878540039, "learning_rate": 9.05248043052978e-06, "loss": 1.1597, "step": 10144 }, { "epoch": 1.63, "grad_norm": 8.029162406921387, "learning_rate": 9.050750273155455e-06, "loss": 0.9738, "step": 10145 }, { "epoch": 1.63, "grad_norm": 8.96815299987793, "learning_rate": 9.049020144454292e-06, "loss": 1.1968, "step": 10146 }, { "epoch": 1.63, "grad_norm": 10.110795974731445, "learning_rate": 9.047290044478547e-06, "loss": 1.0216, "step": 10147 }, { "epoch": 1.63, "grad_norm": 11.673413276672363, "learning_rate": 9.045559973280487e-06, "loss": 1.1582, "step": 10148 }, { "epoch": 1.63, "grad_norm": 9.95622730255127, "learning_rate": 9.043829930912363e-06, "loss": 1.1988, "step": 10149 }, { "epoch": 1.63, "grad_norm": 9.272645950317383, "learning_rate": 9.042099917426435e-06, "loss": 1.0055, "step": 10150 }, { "epoch": 1.63, "grad_norm": 8.193156242370605, "learning_rate": 9.04036993287496e-06, "loss": 1.0014, "step": 10151 }, { "epoch": 1.63, "grad_norm": 9.499282836914062, "learning_rate": 9.038639977310198e-06, "loss": 1.1116, "step": 10152 }, { "epoch": 1.63, "grad_norm": 9.76231575012207, "learning_rate": 9.036910050784401e-06, "loss": 1.0341, "step": 10153 }, { "epoch": 1.63, "grad_norm": 8.806164741516113, "learning_rate": 9.035180153349821e-06, "loss": 0.9135, "step": 10154 }, { "epoch": 1.63, "grad_norm": 9.649251937866211, "learning_rate": 9.033450285058715e-06, "loss": 1.2236, "step": 10155 }, { "epoch": 1.63, "grad_norm": 10.391223907470703, "learning_rate": 9.031720445963335e-06, "loss": 1.1154, "step": 10156 }, { "epoch": 1.64, "grad_norm": 8.341584205627441, "learning_rate": 9.029990636115932e-06, "loss": 0.8103, "step": 10157 }, { "epoch": 1.64, "grad_norm": 9.139148712158203, "learning_rate": 9.028260855568758e-06, "loss": 0.8251, "step": 10158 }, { "epoch": 1.64, "grad_norm": 10.17272663116455, "learning_rate": 9.026531104374065e-06, "loss": 1.1538, "step": 10159 }, { "epoch": 1.64, "grad_norm": 8.885531425476074, "learning_rate": 9.024801382584096e-06, "loss": 1.236, "step": 10160 }, { "epoch": 1.64, "grad_norm": 9.204554557800293, "learning_rate": 9.023071690251103e-06, "loss": 1.0387, "step": 10161 }, { "epoch": 1.64, "grad_norm": 9.390750885009766, "learning_rate": 9.021342027427336e-06, "loss": 0.9854, "step": 10162 }, { "epoch": 1.64, "grad_norm": 8.806624412536621, "learning_rate": 9.019612394165034e-06, "loss": 1.1399, "step": 10163 }, { "epoch": 1.64, "grad_norm": 9.376649856567383, "learning_rate": 9.017882790516452e-06, "loss": 1.1067, "step": 10164 }, { "epoch": 1.64, "grad_norm": 8.94560432434082, "learning_rate": 9.016153216533827e-06, "loss": 0.9944, "step": 10165 }, { "epoch": 1.64, "grad_norm": 9.30579662322998, "learning_rate": 9.014423672269404e-06, "loss": 0.9865, "step": 10166 }, { "epoch": 1.64, "grad_norm": 11.408844947814941, "learning_rate": 9.012694157775427e-06, "loss": 1.3944, "step": 10167 }, { "epoch": 1.64, "grad_norm": 9.102020263671875, "learning_rate": 9.010964673104143e-06, "loss": 1.247, "step": 10168 }, { "epoch": 1.64, "grad_norm": 9.423690795898438, "learning_rate": 9.009235218307784e-06, "loss": 0.9124, "step": 10169 }, { "epoch": 1.64, "grad_norm": 10.15422534942627, "learning_rate": 9.007505793438593e-06, "loss": 1.1224, "step": 10170 }, { "epoch": 1.64, "grad_norm": 12.058938026428223, "learning_rate": 9.005776398548813e-06, "loss": 1.1927, "step": 10171 }, { "epoch": 1.64, "grad_norm": 11.24283218383789, "learning_rate": 9.004047033690679e-06, "loss": 0.9473, "step": 10172 }, { "epoch": 1.64, "grad_norm": 9.373087882995605, "learning_rate": 9.00231769891643e-06, "loss": 0.9547, "step": 10173 }, { "epoch": 1.64, "grad_norm": 10.6878662109375, "learning_rate": 9.0005883942783e-06, "loss": 0.9546, "step": 10174 }, { "epoch": 1.64, "grad_norm": 8.245781898498535, "learning_rate": 8.998859119828527e-06, "loss": 1.0338, "step": 10175 }, { "epoch": 1.64, "grad_norm": 9.563484191894531, "learning_rate": 8.997129875619342e-06, "loss": 1.0551, "step": 10176 }, { "epoch": 1.64, "grad_norm": 8.842090606689453, "learning_rate": 8.995400661702984e-06, "loss": 1.0251, "step": 10177 }, { "epoch": 1.64, "grad_norm": 11.943108558654785, "learning_rate": 8.993671478131685e-06, "loss": 1.1454, "step": 10178 }, { "epoch": 1.64, "grad_norm": 11.223715782165527, "learning_rate": 8.991942324957676e-06, "loss": 1.1755, "step": 10179 }, { "epoch": 1.64, "grad_norm": 8.505759239196777, "learning_rate": 8.990213202233187e-06, "loss": 0.8646, "step": 10180 }, { "epoch": 1.64, "grad_norm": 9.672707557678223, "learning_rate": 8.988484110010449e-06, "loss": 1.2105, "step": 10181 }, { "epoch": 1.64, "grad_norm": 9.55074691772461, "learning_rate": 8.986755048341689e-06, "loss": 0.9281, "step": 10182 }, { "epoch": 1.64, "grad_norm": 8.209156036376953, "learning_rate": 8.98502601727914e-06, "loss": 0.9155, "step": 10183 }, { "epoch": 1.64, "grad_norm": 10.77332592010498, "learning_rate": 8.983297016875025e-06, "loss": 0.9319, "step": 10184 }, { "epoch": 1.64, "grad_norm": 9.7913236618042, "learning_rate": 8.981568047181571e-06, "loss": 1.03, "step": 10185 }, { "epoch": 1.64, "grad_norm": 10.909546852111816, "learning_rate": 8.979839108251005e-06, "loss": 1.1264, "step": 10186 }, { "epoch": 1.64, "grad_norm": 8.36247444152832, "learning_rate": 8.978110200135554e-06, "loss": 1.1181, "step": 10187 }, { "epoch": 1.64, "grad_norm": 10.685992240905762, "learning_rate": 8.976381322887436e-06, "loss": 1.3833, "step": 10188 }, { "epoch": 1.64, "grad_norm": 11.918974876403809, "learning_rate": 8.974652476558874e-06, "loss": 1.1008, "step": 10189 }, { "epoch": 1.64, "grad_norm": 11.523036003112793, "learning_rate": 8.972923661202094e-06, "loss": 1.0853, "step": 10190 }, { "epoch": 1.64, "grad_norm": 12.939376831054688, "learning_rate": 8.971194876869315e-06, "loss": 0.8779, "step": 10191 }, { "epoch": 1.64, "grad_norm": 9.452629089355469, "learning_rate": 8.969466123612757e-06, "loss": 0.9792, "step": 10192 }, { "epoch": 1.64, "grad_norm": 7.712236404418945, "learning_rate": 8.967737401484639e-06, "loss": 0.9665, "step": 10193 }, { "epoch": 1.64, "grad_norm": 8.878003120422363, "learning_rate": 8.966008710537178e-06, "loss": 1.012, "step": 10194 }, { "epoch": 1.64, "grad_norm": 8.645519256591797, "learning_rate": 8.96428005082259e-06, "loss": 0.9802, "step": 10195 }, { "epoch": 1.64, "grad_norm": 9.156025886535645, "learning_rate": 8.962551422393092e-06, "loss": 1.0396, "step": 10196 }, { "epoch": 1.64, "grad_norm": 10.744400024414062, "learning_rate": 8.960822825300902e-06, "loss": 1.0999, "step": 10197 }, { "epoch": 1.64, "grad_norm": 9.25108814239502, "learning_rate": 8.959094259598235e-06, "loss": 1.1127, "step": 10198 }, { "epoch": 1.64, "grad_norm": 9.652527809143066, "learning_rate": 8.957365725337299e-06, "loss": 0.9742, "step": 10199 }, { "epoch": 1.64, "grad_norm": 8.707611083984375, "learning_rate": 8.955637222570308e-06, "loss": 1.0253, "step": 10200 }, { "epoch": 1.64, "grad_norm": 11.294142723083496, "learning_rate": 8.953908751349476e-06, "loss": 1.0585, "step": 10201 }, { "epoch": 1.64, "grad_norm": 10.304471015930176, "learning_rate": 8.952180311727009e-06, "loss": 1.1247, "step": 10202 }, { "epoch": 1.64, "grad_norm": 10.009775161743164, "learning_rate": 8.950451903755123e-06, "loss": 0.8784, "step": 10203 }, { "epoch": 1.64, "grad_norm": 11.869070053100586, "learning_rate": 8.948723527486021e-06, "loss": 1.1837, "step": 10204 }, { "epoch": 1.64, "grad_norm": 9.278985023498535, "learning_rate": 8.94699518297191e-06, "loss": 1.0006, "step": 10205 }, { "epoch": 1.64, "grad_norm": 10.580105781555176, "learning_rate": 8.945266870265002e-06, "loss": 1.0947, "step": 10206 }, { "epoch": 1.64, "grad_norm": 9.457925796508789, "learning_rate": 8.9435385894175e-06, "loss": 1.2105, "step": 10207 }, { "epoch": 1.64, "grad_norm": 9.181243896484375, "learning_rate": 8.941810340481605e-06, "loss": 1.0729, "step": 10208 }, { "epoch": 1.64, "grad_norm": 9.295180320739746, "learning_rate": 8.940082123509525e-06, "loss": 1.3422, "step": 10209 }, { "epoch": 1.64, "grad_norm": 9.853972434997559, "learning_rate": 8.938353938553462e-06, "loss": 0.9153, "step": 10210 }, { "epoch": 1.64, "grad_norm": 8.85544204711914, "learning_rate": 8.936625785665619e-06, "loss": 1.1295, "step": 10211 }, { "epoch": 1.64, "grad_norm": 8.69329833984375, "learning_rate": 8.934897664898192e-06, "loss": 0.8247, "step": 10212 }, { "epoch": 1.64, "grad_norm": 8.495741844177246, "learning_rate": 8.933169576303384e-06, "loss": 1.0276, "step": 10213 }, { "epoch": 1.64, "grad_norm": 9.617684364318848, "learning_rate": 8.931441519933395e-06, "loss": 1.0966, "step": 10214 }, { "epoch": 1.64, "grad_norm": 8.690799713134766, "learning_rate": 8.929713495840423e-06, "loss": 0.9186, "step": 10215 }, { "epoch": 1.64, "grad_norm": 9.32871150970459, "learning_rate": 8.927985504076661e-06, "loss": 0.9534, "step": 10216 }, { "epoch": 1.64, "grad_norm": 11.641688346862793, "learning_rate": 8.926257544694305e-06, "loss": 1.3304, "step": 10217 }, { "epoch": 1.64, "grad_norm": 12.700575828552246, "learning_rate": 8.924529617745558e-06, "loss": 1.1575, "step": 10218 }, { "epoch": 1.65, "grad_norm": 8.999786376953125, "learning_rate": 8.922801723282607e-06, "loss": 0.9718, "step": 10219 }, { "epoch": 1.65, "grad_norm": 8.703035354614258, "learning_rate": 8.921073861357646e-06, "loss": 0.9463, "step": 10220 }, { "epoch": 1.65, "grad_norm": 9.769275665283203, "learning_rate": 8.919346032022869e-06, "loss": 1.0922, "step": 10221 }, { "epoch": 1.65, "grad_norm": 10.41085433959961, "learning_rate": 8.917618235330466e-06, "loss": 1.0648, "step": 10222 }, { "epoch": 1.65, "grad_norm": 10.27573013305664, "learning_rate": 8.915890471332625e-06, "loss": 1.0722, "step": 10223 }, { "epoch": 1.65, "grad_norm": 9.793176651000977, "learning_rate": 8.914162740081537e-06, "loss": 0.8992, "step": 10224 }, { "epoch": 1.65, "grad_norm": 8.037250518798828, "learning_rate": 8.912435041629391e-06, "loss": 0.9703, "step": 10225 }, { "epoch": 1.65, "grad_norm": 9.939002990722656, "learning_rate": 8.910707376028373e-06, "loss": 1.0435, "step": 10226 }, { "epoch": 1.65, "grad_norm": 9.090798377990723, "learning_rate": 8.908979743330668e-06, "loss": 0.9641, "step": 10227 }, { "epoch": 1.65, "grad_norm": 9.592355728149414, "learning_rate": 8.90725214358846e-06, "loss": 1.0498, "step": 10228 }, { "epoch": 1.65, "grad_norm": 10.182291984558105, "learning_rate": 8.905524576853937e-06, "loss": 1.1011, "step": 10229 }, { "epoch": 1.65, "grad_norm": 10.06892204284668, "learning_rate": 8.903797043179281e-06, "loss": 0.9562, "step": 10230 }, { "epoch": 1.65, "grad_norm": 9.455360412597656, "learning_rate": 8.902069542616674e-06, "loss": 0.9681, "step": 10231 }, { "epoch": 1.65, "grad_norm": 11.242548942565918, "learning_rate": 8.900342075218296e-06, "loss": 0.9074, "step": 10232 }, { "epoch": 1.65, "grad_norm": 10.257858276367188, "learning_rate": 8.898614641036327e-06, "loss": 0.9316, "step": 10233 }, { "epoch": 1.65, "grad_norm": 8.79616928100586, "learning_rate": 8.896887240122947e-06, "loss": 1.1155, "step": 10234 }, { "epoch": 1.65, "grad_norm": 11.907526016235352, "learning_rate": 8.895159872530335e-06, "loss": 1.3145, "step": 10235 }, { "epoch": 1.65, "grad_norm": 12.61099910736084, "learning_rate": 8.893432538310664e-06, "loss": 1.1196, "step": 10236 }, { "epoch": 1.65, "grad_norm": 10.194406509399414, "learning_rate": 8.891705237516113e-06, "loss": 0.9124, "step": 10237 }, { "epoch": 1.65, "grad_norm": 8.388185501098633, "learning_rate": 8.889977970198858e-06, "loss": 0.8672, "step": 10238 }, { "epoch": 1.65, "grad_norm": 10.268661499023438, "learning_rate": 8.888250736411072e-06, "loss": 1.22, "step": 10239 }, { "epoch": 1.65, "grad_norm": 10.04408073425293, "learning_rate": 8.886523536204929e-06, "loss": 0.8154, "step": 10240 }, { "epoch": 1.65, "grad_norm": 9.919377326965332, "learning_rate": 8.884796369632599e-06, "loss": 1.0873, "step": 10241 }, { "epoch": 1.65, "grad_norm": 10.732098579406738, "learning_rate": 8.883069236746258e-06, "loss": 1.1467, "step": 10242 }, { "epoch": 1.65, "grad_norm": 7.756825923919678, "learning_rate": 8.881342137598069e-06, "loss": 0.8423, "step": 10243 }, { "epoch": 1.65, "grad_norm": 10.796624183654785, "learning_rate": 8.879615072240203e-06, "loss": 1.251, "step": 10244 }, { "epoch": 1.65, "grad_norm": 9.685190200805664, "learning_rate": 8.87788804072483e-06, "loss": 1.0626, "step": 10245 }, { "epoch": 1.65, "grad_norm": 10.015571594238281, "learning_rate": 8.876161043104119e-06, "loss": 1.0097, "step": 10246 }, { "epoch": 1.65, "grad_norm": 8.999908447265625, "learning_rate": 8.874434079430229e-06, "loss": 1.1274, "step": 10247 }, { "epoch": 1.65, "grad_norm": 9.428043365478516, "learning_rate": 8.87270714975533e-06, "loss": 0.8868, "step": 10248 }, { "epoch": 1.65, "grad_norm": 11.314033508300781, "learning_rate": 8.870980254131586e-06, "loss": 0.9634, "step": 10249 }, { "epoch": 1.65, "grad_norm": 8.547321319580078, "learning_rate": 8.869253392611158e-06, "loss": 0.9204, "step": 10250 }, { "epoch": 1.65, "grad_norm": 11.39773941040039, "learning_rate": 8.86752656524621e-06, "loss": 0.8628, "step": 10251 }, { "epoch": 1.65, "grad_norm": 9.076767921447754, "learning_rate": 8.865799772088898e-06, "loss": 0.8466, "step": 10252 }, { "epoch": 1.65, "grad_norm": 8.20698070526123, "learning_rate": 8.864073013191386e-06, "loss": 0.9348, "step": 10253 }, { "epoch": 1.65, "grad_norm": 10.623591423034668, "learning_rate": 8.862346288605834e-06, "loss": 1.1015, "step": 10254 }, { "epoch": 1.65, "grad_norm": 9.645718574523926, "learning_rate": 8.860619598384394e-06, "loss": 1.0075, "step": 10255 }, { "epoch": 1.65, "grad_norm": 9.144474983215332, "learning_rate": 8.858892942579227e-06, "loss": 1.0646, "step": 10256 }, { "epoch": 1.65, "grad_norm": 9.845263481140137, "learning_rate": 8.857166321242486e-06, "loss": 1.2447, "step": 10257 }, { "epoch": 1.65, "grad_norm": 10.601579666137695, "learning_rate": 8.855439734426327e-06, "loss": 0.9567, "step": 10258 }, { "epoch": 1.65, "grad_norm": 12.347790718078613, "learning_rate": 8.853713182182904e-06, "loss": 1.2509, "step": 10259 }, { "epoch": 1.65, "grad_norm": 8.654329299926758, "learning_rate": 8.85198666456437e-06, "loss": 0.9339, "step": 10260 }, { "epoch": 1.65, "grad_norm": 9.784215927124023, "learning_rate": 8.850260181622878e-06, "loss": 1.0337, "step": 10261 }, { "epoch": 1.65, "grad_norm": 8.969554901123047, "learning_rate": 8.848533733410572e-06, "loss": 0.8327, "step": 10262 }, { "epoch": 1.65, "grad_norm": 12.121798515319824, "learning_rate": 8.846807319979607e-06, "loss": 1.3272, "step": 10263 }, { "epoch": 1.65, "grad_norm": 8.995776176452637, "learning_rate": 8.845080941382128e-06, "loss": 1.0134, "step": 10264 }, { "epoch": 1.65, "grad_norm": 10.903916358947754, "learning_rate": 8.843354597670286e-06, "loss": 1.1359, "step": 10265 }, { "epoch": 1.65, "grad_norm": 10.284523010253906, "learning_rate": 8.841628288896223e-06, "loss": 1.0881, "step": 10266 }, { "epoch": 1.65, "grad_norm": 8.265789031982422, "learning_rate": 8.839902015112085e-06, "loss": 0.903, "step": 10267 }, { "epoch": 1.65, "grad_norm": 8.719467163085938, "learning_rate": 8.838175776370018e-06, "loss": 0.9479, "step": 10268 }, { "epoch": 1.65, "grad_norm": 10.133682250976562, "learning_rate": 8.836449572722162e-06, "loss": 1.1316, "step": 10269 }, { "epoch": 1.65, "grad_norm": 10.706965446472168, "learning_rate": 8.834723404220665e-06, "loss": 1.2445, "step": 10270 }, { "epoch": 1.65, "grad_norm": 9.54693603515625, "learning_rate": 8.83299727091766e-06, "loss": 0.9815, "step": 10271 }, { "epoch": 1.65, "grad_norm": 10.935381889343262, "learning_rate": 8.831271172865291e-06, "loss": 0.8453, "step": 10272 }, { "epoch": 1.65, "grad_norm": 12.41945743560791, "learning_rate": 8.829545110115697e-06, "loss": 1.0546, "step": 10273 }, { "epoch": 1.65, "grad_norm": 11.062373161315918, "learning_rate": 8.827819082721016e-06, "loss": 1.1659, "step": 10274 }, { "epoch": 1.65, "grad_norm": 8.983428955078125, "learning_rate": 8.826093090733383e-06, "loss": 0.8982, "step": 10275 }, { "epoch": 1.65, "grad_norm": 12.004337310791016, "learning_rate": 8.824367134204931e-06, "loss": 1.059, "step": 10276 }, { "epoch": 1.65, "grad_norm": 10.08498764038086, "learning_rate": 8.822641213187798e-06, "loss": 1.1093, "step": 10277 }, { "epoch": 1.65, "grad_norm": 9.15675163269043, "learning_rate": 8.820915327734118e-06, "loss": 0.8138, "step": 10278 }, { "epoch": 1.65, "grad_norm": 8.849617958068848, "learning_rate": 8.819189477896021e-06, "loss": 0.777, "step": 10279 }, { "epoch": 1.65, "grad_norm": 10.800093650817871, "learning_rate": 8.817463663725641e-06, "loss": 1.1406, "step": 10280 }, { "epoch": 1.66, "grad_norm": 10.289297103881836, "learning_rate": 8.81573788527511e-06, "loss": 1.0784, "step": 10281 }, { "epoch": 1.66, "grad_norm": 10.39457893371582, "learning_rate": 8.81401214259655e-06, "loss": 1.2214, "step": 10282 }, { "epoch": 1.66, "grad_norm": 8.314409255981445, "learning_rate": 8.812286435742092e-06, "loss": 1.0901, "step": 10283 }, { "epoch": 1.66, "grad_norm": 9.273276329040527, "learning_rate": 8.810560764763865e-06, "loss": 0.8426, "step": 10284 }, { "epoch": 1.66, "grad_norm": 9.527162551879883, "learning_rate": 8.808835129713996e-06, "loss": 1.0177, "step": 10285 }, { "epoch": 1.66, "grad_norm": 9.535789489746094, "learning_rate": 8.807109530644603e-06, "loss": 0.8453, "step": 10286 }, { "epoch": 1.66, "grad_norm": 11.382050514221191, "learning_rate": 8.805383967607815e-06, "loss": 1.1738, "step": 10287 }, { "epoch": 1.66, "grad_norm": 9.007426261901855, "learning_rate": 8.803658440655755e-06, "loss": 0.7676, "step": 10288 }, { "epoch": 1.66, "grad_norm": 9.556915283203125, "learning_rate": 8.801932949840543e-06, "loss": 0.6743, "step": 10289 }, { "epoch": 1.66, "grad_norm": 9.490485191345215, "learning_rate": 8.800207495214299e-06, "loss": 1.012, "step": 10290 }, { "epoch": 1.66, "grad_norm": 9.785247802734375, "learning_rate": 8.79848207682914e-06, "loss": 0.9401, "step": 10291 }, { "epoch": 1.66, "grad_norm": 10.23543930053711, "learning_rate": 8.796756694737189e-06, "loss": 0.9606, "step": 10292 }, { "epoch": 1.66, "grad_norm": 10.587368965148926, "learning_rate": 8.795031348990564e-06, "loss": 0.9537, "step": 10293 }, { "epoch": 1.66, "grad_norm": 9.319991111755371, "learning_rate": 8.793306039641375e-06, "loss": 0.8239, "step": 10294 }, { "epoch": 1.66, "grad_norm": 11.3405179977417, "learning_rate": 8.79158076674174e-06, "loss": 1.105, "step": 10295 }, { "epoch": 1.66, "grad_norm": 12.201011657714844, "learning_rate": 8.789855530343771e-06, "loss": 1.4612, "step": 10296 }, { "epoch": 1.66, "grad_norm": 8.320329666137695, "learning_rate": 8.788130330499585e-06, "loss": 0.95, "step": 10297 }, { "epoch": 1.66, "grad_norm": 7.796040058135986, "learning_rate": 8.786405167261287e-06, "loss": 1.1437, "step": 10298 }, { "epoch": 1.66, "grad_norm": 9.834344863891602, "learning_rate": 8.784680040680995e-06, "loss": 0.9929, "step": 10299 }, { "epoch": 1.66, "grad_norm": 10.540983200073242, "learning_rate": 8.782954950810817e-06, "loss": 0.911, "step": 10300 }, { "epoch": 1.66, "grad_norm": 8.878936767578125, "learning_rate": 8.781229897702857e-06, "loss": 0.7995, "step": 10301 }, { "epoch": 1.66, "grad_norm": 9.497072219848633, "learning_rate": 8.779504881409224e-06, "loss": 1.0313, "step": 10302 }, { "epoch": 1.66, "grad_norm": 11.087697982788086, "learning_rate": 8.777779901982024e-06, "loss": 1.1956, "step": 10303 }, { "epoch": 1.66, "grad_norm": 9.537330627441406, "learning_rate": 8.776054959473366e-06, "loss": 0.933, "step": 10304 }, { "epoch": 1.66, "grad_norm": 9.65231704711914, "learning_rate": 8.774330053935348e-06, "loss": 0.8508, "step": 10305 }, { "epoch": 1.66, "grad_norm": 10.17624282836914, "learning_rate": 8.772605185420074e-06, "loss": 1.0732, "step": 10306 }, { "epoch": 1.66, "grad_norm": 10.256624221801758, "learning_rate": 8.770880353979647e-06, "loss": 0.9242, "step": 10307 }, { "epoch": 1.66, "grad_norm": 8.009420394897461, "learning_rate": 8.769155559666169e-06, "loss": 0.7646, "step": 10308 }, { "epoch": 1.66, "grad_norm": 9.326203346252441, "learning_rate": 8.767430802531737e-06, "loss": 1.0167, "step": 10309 }, { "epoch": 1.66, "grad_norm": 10.009002685546875, "learning_rate": 8.76570608262845e-06, "loss": 1.1174, "step": 10310 }, { "epoch": 1.66, "grad_norm": 11.283535957336426, "learning_rate": 8.763981400008401e-06, "loss": 1.1613, "step": 10311 }, { "epoch": 1.66, "grad_norm": 9.679832458496094, "learning_rate": 8.762256754723693e-06, "loss": 1.1588, "step": 10312 }, { "epoch": 1.66, "grad_norm": 12.446978569030762, "learning_rate": 8.760532146826418e-06, "loss": 1.201, "step": 10313 }, { "epoch": 1.66, "grad_norm": 10.10582160949707, "learning_rate": 8.75880757636867e-06, "loss": 1.1859, "step": 10314 }, { "epoch": 1.66, "grad_norm": 9.773992538452148, "learning_rate": 8.757083043402539e-06, "loss": 1.0051, "step": 10315 }, { "epoch": 1.66, "grad_norm": 9.340230941772461, "learning_rate": 8.755358547980118e-06, "loss": 1.0167, "step": 10316 }, { "epoch": 1.66, "grad_norm": 9.55312442779541, "learning_rate": 8.753634090153503e-06, "loss": 1.0319, "step": 10317 }, { "epoch": 1.66, "grad_norm": 10.398131370544434, "learning_rate": 8.751909669974767e-06, "loss": 1.1309, "step": 10318 }, { "epoch": 1.66, "grad_norm": 11.080013275146484, "learning_rate": 8.750185287496015e-06, "loss": 1.025, "step": 10319 }, { "epoch": 1.66, "grad_norm": 10.041033744812012, "learning_rate": 8.748460942769332e-06, "loss": 1.0128, "step": 10320 }, { "epoch": 1.66, "grad_norm": 8.57843017578125, "learning_rate": 8.746736635846797e-06, "loss": 1.15, "step": 10321 }, { "epoch": 1.66, "grad_norm": 9.240894317626953, "learning_rate": 8.745012366780497e-06, "loss": 0.7104, "step": 10322 }, { "epoch": 1.66, "grad_norm": 7.961857795715332, "learning_rate": 8.743288135622515e-06, "loss": 0.7829, "step": 10323 }, { "epoch": 1.66, "grad_norm": 11.014869689941406, "learning_rate": 8.741563942424937e-06, "loss": 1.2462, "step": 10324 }, { "epoch": 1.66, "grad_norm": 10.231104850769043, "learning_rate": 8.739839787239838e-06, "loss": 0.981, "step": 10325 }, { "epoch": 1.66, "grad_norm": 11.539667129516602, "learning_rate": 8.738115670119304e-06, "loss": 1.1046, "step": 10326 }, { "epoch": 1.66, "grad_norm": 11.648531913757324, "learning_rate": 8.73639159111541e-06, "loss": 1.1073, "step": 10327 }, { "epoch": 1.66, "grad_norm": 10.271300315856934, "learning_rate": 8.734667550280237e-06, "loss": 1.2081, "step": 10328 }, { "epoch": 1.66, "grad_norm": 13.362391471862793, "learning_rate": 8.732943547665857e-06, "loss": 0.8006, "step": 10329 }, { "epoch": 1.66, "grad_norm": 8.91307258605957, "learning_rate": 8.731219583324351e-06, "loss": 1.0082, "step": 10330 }, { "epoch": 1.66, "grad_norm": 10.364097595214844, "learning_rate": 8.729495657307789e-06, "loss": 1.063, "step": 10331 }, { "epoch": 1.66, "grad_norm": 9.800495147705078, "learning_rate": 8.727771769668246e-06, "loss": 1.1591, "step": 10332 }, { "epoch": 1.66, "grad_norm": 10.752896308898926, "learning_rate": 8.726047920457794e-06, "loss": 1.0619, "step": 10333 }, { "epoch": 1.66, "grad_norm": 9.844344139099121, "learning_rate": 8.724324109728501e-06, "loss": 0.9263, "step": 10334 }, { "epoch": 1.66, "grad_norm": 10.902743339538574, "learning_rate": 8.722600337532442e-06, "loss": 1.0233, "step": 10335 }, { "epoch": 1.66, "grad_norm": 9.10323429107666, "learning_rate": 8.72087660392168e-06, "loss": 1.1837, "step": 10336 }, { "epoch": 1.66, "grad_norm": 11.474164962768555, "learning_rate": 8.719152908948287e-06, "loss": 1.0915, "step": 10337 }, { "epoch": 1.66, "grad_norm": 10.273921966552734, "learning_rate": 8.717429252664322e-06, "loss": 1.046, "step": 10338 }, { "epoch": 1.66, "grad_norm": 7.551628112792969, "learning_rate": 8.715705635121862e-06, "loss": 0.9462, "step": 10339 }, { "epoch": 1.66, "grad_norm": 7.894081115722656, "learning_rate": 8.71398205637296e-06, "loss": 0.988, "step": 10340 }, { "epoch": 1.66, "grad_norm": 10.251689910888672, "learning_rate": 8.712258516469683e-06, "loss": 1.2461, "step": 10341 }, { "epoch": 1.66, "grad_norm": 9.178407669067383, "learning_rate": 8.710535015464092e-06, "loss": 0.9733, "step": 10342 }, { "epoch": 1.67, "grad_norm": 8.990031242370605, "learning_rate": 8.70881155340825e-06, "loss": 1.1772, "step": 10343 }, { "epoch": 1.67, "grad_norm": 11.40604019165039, "learning_rate": 8.70708813035421e-06, "loss": 1.3399, "step": 10344 }, { "epoch": 1.67, "grad_norm": 9.162848472595215, "learning_rate": 8.705364746354033e-06, "loss": 0.9811, "step": 10345 }, { "epoch": 1.67, "grad_norm": 11.166258811950684, "learning_rate": 8.703641401459778e-06, "loss": 1.045, "step": 10346 }, { "epoch": 1.67, "grad_norm": 11.433478355407715, "learning_rate": 8.701918095723498e-06, "loss": 1.2879, "step": 10347 }, { "epoch": 1.67, "grad_norm": 10.424176216125488, "learning_rate": 8.700194829197248e-06, "loss": 0.9853, "step": 10348 }, { "epoch": 1.67, "grad_norm": 9.376261711120605, "learning_rate": 8.69847160193308e-06, "loss": 1.1423, "step": 10349 }, { "epoch": 1.67, "grad_norm": 10.411149978637695, "learning_rate": 8.696748413983049e-06, "loss": 0.9752, "step": 10350 }, { "epoch": 1.67, "grad_norm": 8.769329071044922, "learning_rate": 8.695025265399203e-06, "loss": 0.8555, "step": 10351 }, { "epoch": 1.67, "grad_norm": 8.527161598205566, "learning_rate": 8.693302156233595e-06, "loss": 1.0101, "step": 10352 }, { "epoch": 1.67, "grad_norm": 9.3849458694458, "learning_rate": 8.691579086538269e-06, "loss": 1.1319, "step": 10353 }, { "epoch": 1.67, "grad_norm": 8.877303123474121, "learning_rate": 8.689856056365275e-06, "loss": 1.2686, "step": 10354 }, { "epoch": 1.67, "grad_norm": 9.735931396484375, "learning_rate": 8.688133065766658e-06, "loss": 0.7588, "step": 10355 }, { "epoch": 1.67, "grad_norm": 9.041625022888184, "learning_rate": 8.686410114794467e-06, "loss": 0.9951, "step": 10356 }, { "epoch": 1.67, "grad_norm": 9.791471481323242, "learning_rate": 8.684687203500737e-06, "loss": 0.9698, "step": 10357 }, { "epoch": 1.67, "grad_norm": 9.830397605895996, "learning_rate": 8.682964331937512e-06, "loss": 0.8152, "step": 10358 }, { "epoch": 1.67, "grad_norm": 10.895719528198242, "learning_rate": 8.681241500156844e-06, "loss": 0.8087, "step": 10359 }, { "epoch": 1.67, "grad_norm": 10.218473434448242, "learning_rate": 8.679518708210764e-06, "loss": 1.047, "step": 10360 }, { "epoch": 1.67, "grad_norm": 8.433815956115723, "learning_rate": 8.677795956151313e-06, "loss": 0.8215, "step": 10361 }, { "epoch": 1.67, "grad_norm": 9.811558723449707, "learning_rate": 8.676073244030527e-06, "loss": 1.0187, "step": 10362 }, { "epoch": 1.67, "grad_norm": 9.040799140930176, "learning_rate": 8.674350571900448e-06, "loss": 0.9633, "step": 10363 }, { "epoch": 1.67, "grad_norm": 9.390914916992188, "learning_rate": 8.672627939813104e-06, "loss": 1.1081, "step": 10364 }, { "epoch": 1.67, "grad_norm": 12.742118835449219, "learning_rate": 8.670905347820533e-06, "loss": 1.307, "step": 10365 }, { "epoch": 1.67, "grad_norm": 9.690698623657227, "learning_rate": 8.669182795974766e-06, "loss": 1.034, "step": 10366 }, { "epoch": 1.67, "grad_norm": 9.25924301147461, "learning_rate": 8.667460284327839e-06, "loss": 1.233, "step": 10367 }, { "epoch": 1.67, "grad_norm": 7.3264384269714355, "learning_rate": 8.665737812931775e-06, "loss": 1.0182, "step": 10368 }, { "epoch": 1.67, "grad_norm": 11.227302551269531, "learning_rate": 8.66401538183861e-06, "loss": 1.0683, "step": 10369 }, { "epoch": 1.67, "grad_norm": 9.468266487121582, "learning_rate": 8.662292991100368e-06, "loss": 0.9437, "step": 10370 }, { "epoch": 1.67, "grad_norm": 11.791919708251953, "learning_rate": 8.660570640769079e-06, "loss": 0.9523, "step": 10371 }, { "epoch": 1.67, "grad_norm": 7.96798038482666, "learning_rate": 8.658848330896765e-06, "loss": 0.9153, "step": 10372 }, { "epoch": 1.67, "grad_norm": 7.432399272918701, "learning_rate": 8.657126061535453e-06, "loss": 1.2752, "step": 10373 }, { "epoch": 1.67, "grad_norm": 13.68910026550293, "learning_rate": 8.655403832737162e-06, "loss": 0.8058, "step": 10374 }, { "epoch": 1.67, "grad_norm": 9.024897575378418, "learning_rate": 8.653681644553919e-06, "loss": 0.9826, "step": 10375 }, { "epoch": 1.67, "grad_norm": 10.156255722045898, "learning_rate": 8.651959497037745e-06, "loss": 1.0058, "step": 10376 }, { "epoch": 1.67, "grad_norm": 11.037110328674316, "learning_rate": 8.650237390240652e-06, "loss": 0.897, "step": 10377 }, { "epoch": 1.67, "grad_norm": 9.76752758026123, "learning_rate": 8.648515324214666e-06, "loss": 0.9699, "step": 10378 }, { "epoch": 1.67, "grad_norm": 8.207178115844727, "learning_rate": 8.646793299011798e-06, "loss": 1.002, "step": 10379 }, { "epoch": 1.67, "grad_norm": 9.594473838806152, "learning_rate": 8.645071314684068e-06, "loss": 0.9948, "step": 10380 }, { "epoch": 1.67, "grad_norm": 10.455323219299316, "learning_rate": 8.64334937128349e-06, "loss": 1.0857, "step": 10381 }, { "epoch": 1.67, "grad_norm": 10.716087341308594, "learning_rate": 8.64162746886208e-06, "loss": 0.9836, "step": 10382 }, { "epoch": 1.67, "grad_norm": 8.831436157226562, "learning_rate": 8.639905607471842e-06, "loss": 0.8815, "step": 10383 }, { "epoch": 1.67, "grad_norm": 11.08683967590332, "learning_rate": 8.638183787164792e-06, "loss": 0.9561, "step": 10384 }, { "epoch": 1.67, "grad_norm": 9.856661796569824, "learning_rate": 8.636462007992939e-06, "loss": 1.0286, "step": 10385 }, { "epoch": 1.67, "grad_norm": 8.208895683288574, "learning_rate": 8.634740270008291e-06, "loss": 0.9049, "step": 10386 }, { "epoch": 1.67, "grad_norm": 10.496788024902344, "learning_rate": 8.633018573262859e-06, "loss": 1.0174, "step": 10387 }, { "epoch": 1.67, "grad_norm": 9.728913307189941, "learning_rate": 8.63129691780864e-06, "loss": 0.9086, "step": 10388 }, { "epoch": 1.67, "grad_norm": 10.818419456481934, "learning_rate": 8.629575303697644e-06, "loss": 1.0458, "step": 10389 }, { "epoch": 1.67, "grad_norm": 8.99259090423584, "learning_rate": 8.627853730981875e-06, "loss": 1.0069, "step": 10390 }, { "epoch": 1.67, "grad_norm": 9.543231010437012, "learning_rate": 8.626132199713335e-06, "loss": 1.0296, "step": 10391 }, { "epoch": 1.67, "grad_norm": 9.652273178100586, "learning_rate": 8.624410709944021e-06, "loss": 0.8857, "step": 10392 }, { "epoch": 1.67, "grad_norm": 11.095564842224121, "learning_rate": 8.622689261725936e-06, "loss": 1.0762, "step": 10393 }, { "epoch": 1.67, "grad_norm": 10.168563842773438, "learning_rate": 8.620967855111077e-06, "loss": 0.9143, "step": 10394 }, { "epoch": 1.67, "grad_norm": 11.158447265625, "learning_rate": 8.619246490151446e-06, "loss": 1.0476, "step": 10395 }, { "epoch": 1.67, "grad_norm": 10.11742877960205, "learning_rate": 8.617525166899028e-06, "loss": 1.0802, "step": 10396 }, { "epoch": 1.67, "grad_norm": 9.742365837097168, "learning_rate": 8.615803885405826e-06, "loss": 1.1678, "step": 10397 }, { "epoch": 1.67, "grad_norm": 10.125767707824707, "learning_rate": 8.614082645723831e-06, "loss": 1.2739, "step": 10398 }, { "epoch": 1.67, "grad_norm": 12.329707145690918, "learning_rate": 8.612361447905031e-06, "loss": 1.1378, "step": 10399 }, { "epoch": 1.67, "grad_norm": 8.999894142150879, "learning_rate": 8.610640292001425e-06, "loss": 1.1239, "step": 10400 }, { "epoch": 1.67, "grad_norm": 9.089468955993652, "learning_rate": 8.608919178064998e-06, "loss": 1.1087, "step": 10401 }, { "epoch": 1.67, "grad_norm": 7.993200302124023, "learning_rate": 8.60719810614774e-06, "loss": 1.0115, "step": 10402 }, { "epoch": 1.67, "grad_norm": 8.196378707885742, "learning_rate": 8.605477076301634e-06, "loss": 1.0329, "step": 10403 }, { "epoch": 1.67, "grad_norm": 7.634374618530273, "learning_rate": 8.603756088578669e-06, "loss": 1.2004, "step": 10404 }, { "epoch": 1.67, "grad_norm": 9.194205284118652, "learning_rate": 8.602035143030826e-06, "loss": 0.8649, "step": 10405 }, { "epoch": 1.68, "grad_norm": 10.620634078979492, "learning_rate": 8.600314239710095e-06, "loss": 1.1784, "step": 10406 }, { "epoch": 1.68, "grad_norm": 8.960501670837402, "learning_rate": 8.59859337866845e-06, "loss": 0.8681, "step": 10407 }, { "epoch": 1.68, "grad_norm": 8.51034164428711, "learning_rate": 8.596872559957875e-06, "loss": 1.0319, "step": 10408 }, { "epoch": 1.68, "grad_norm": 8.750327110290527, "learning_rate": 8.595151783630349e-06, "loss": 0.8825, "step": 10409 }, { "epoch": 1.68, "grad_norm": 9.252950668334961, "learning_rate": 8.593431049737852e-06, "loss": 1.0076, "step": 10410 }, { "epoch": 1.68, "grad_norm": 9.17077922821045, "learning_rate": 8.591710358332357e-06, "loss": 1.0122, "step": 10411 }, { "epoch": 1.68, "grad_norm": 9.244945526123047, "learning_rate": 8.58998970946584e-06, "loss": 0.8519, "step": 10412 }, { "epoch": 1.68, "grad_norm": 11.454215049743652, "learning_rate": 8.588269103190277e-06, "loss": 1.0306, "step": 10413 }, { "epoch": 1.68, "grad_norm": 12.252029418945312, "learning_rate": 8.58654853955764e-06, "loss": 1.2646, "step": 10414 }, { "epoch": 1.68, "grad_norm": 9.502103805541992, "learning_rate": 8.584828018619905e-06, "loss": 1.0832, "step": 10415 }, { "epoch": 1.68, "grad_norm": 10.882272720336914, "learning_rate": 8.583107540429035e-06, "loss": 1.0002, "step": 10416 }, { "epoch": 1.68, "grad_norm": 12.15086841583252, "learning_rate": 8.581387105037e-06, "loss": 1.3101, "step": 10417 }, { "epoch": 1.68, "grad_norm": 10.016934394836426, "learning_rate": 8.57966671249577e-06, "loss": 0.8697, "step": 10418 }, { "epoch": 1.68, "grad_norm": 10.018710136413574, "learning_rate": 8.57794636285731e-06, "loss": 1.0194, "step": 10419 }, { "epoch": 1.68, "grad_norm": 11.049835205078125, "learning_rate": 8.576226056173589e-06, "loss": 1.0504, "step": 10420 }, { "epoch": 1.68, "grad_norm": 9.238582611083984, "learning_rate": 8.574505792496571e-06, "loss": 1.1474, "step": 10421 }, { "epoch": 1.68, "grad_norm": 11.15945816040039, "learning_rate": 8.57278557187821e-06, "loss": 1.0775, "step": 10422 }, { "epoch": 1.68, "grad_norm": 7.8610615730285645, "learning_rate": 8.571065394370474e-06, "loss": 0.7498, "step": 10423 }, { "epoch": 1.68, "grad_norm": 7.641897678375244, "learning_rate": 8.569345260025324e-06, "loss": 1.0136, "step": 10424 }, { "epoch": 1.68, "grad_norm": 10.815028190612793, "learning_rate": 8.567625168894713e-06, "loss": 1.0918, "step": 10425 }, { "epoch": 1.68, "grad_norm": 9.294710159301758, "learning_rate": 8.565905121030607e-06, "loss": 1.0511, "step": 10426 }, { "epoch": 1.68, "grad_norm": 10.770072937011719, "learning_rate": 8.56418511648495e-06, "loss": 0.9671, "step": 10427 }, { "epoch": 1.68, "grad_norm": 9.314627647399902, "learning_rate": 8.562465155309707e-06, "loss": 1.0002, "step": 10428 }, { "epoch": 1.68, "grad_norm": 9.455596923828125, "learning_rate": 8.560745237556826e-06, "loss": 1.0337, "step": 10429 }, { "epoch": 1.68, "grad_norm": 9.537055969238281, "learning_rate": 8.559025363278264e-06, "loss": 0.9474, "step": 10430 }, { "epoch": 1.68, "grad_norm": 11.695694923400879, "learning_rate": 8.557305532525965e-06, "loss": 0.8774, "step": 10431 }, { "epoch": 1.68, "grad_norm": 11.54189682006836, "learning_rate": 8.555585745351882e-06, "loss": 0.9385, "step": 10432 }, { "epoch": 1.68, "grad_norm": 10.957010269165039, "learning_rate": 8.553866001807963e-06, "loss": 1.3443, "step": 10433 }, { "epoch": 1.68, "grad_norm": 10.3035306930542, "learning_rate": 8.552146301946158e-06, "loss": 0.9931, "step": 10434 }, { "epoch": 1.68, "grad_norm": 10.130879402160645, "learning_rate": 8.550426645818405e-06, "loss": 1.0609, "step": 10435 }, { "epoch": 1.68, "grad_norm": 9.359132766723633, "learning_rate": 8.548707033476654e-06, "loss": 1.2497, "step": 10436 }, { "epoch": 1.68, "grad_norm": 8.975214004516602, "learning_rate": 8.546987464972845e-06, "loss": 0.8529, "step": 10437 }, { "epoch": 1.68, "grad_norm": 8.645697593688965, "learning_rate": 8.545267940358923e-06, "loss": 0.764, "step": 10438 }, { "epoch": 1.68, "grad_norm": 8.850238800048828, "learning_rate": 8.543548459686822e-06, "loss": 0.9394, "step": 10439 }, { "epoch": 1.68, "grad_norm": 9.647502899169922, "learning_rate": 8.541829023008486e-06, "loss": 0.952, "step": 10440 }, { "epoch": 1.68, "grad_norm": 8.70608139038086, "learning_rate": 8.540109630375854e-06, "loss": 0.9593, "step": 10441 }, { "epoch": 1.68, "grad_norm": 10.905399322509766, "learning_rate": 8.538390281840859e-06, "loss": 1.1005, "step": 10442 }, { "epoch": 1.68, "grad_norm": 11.054603576660156, "learning_rate": 8.536670977455434e-06, "loss": 1.179, "step": 10443 }, { "epoch": 1.68, "grad_norm": 11.751795768737793, "learning_rate": 8.534951717271516e-06, "loss": 0.892, "step": 10444 }, { "epoch": 1.68, "grad_norm": 10.566350936889648, "learning_rate": 8.533232501341038e-06, "loss": 1.0781, "step": 10445 }, { "epoch": 1.68, "grad_norm": 9.302145957946777, "learning_rate": 8.531513329715927e-06, "loss": 1.0106, "step": 10446 }, { "epoch": 1.68, "grad_norm": 9.66838264465332, "learning_rate": 8.529794202448114e-06, "loss": 0.8813, "step": 10447 }, { "epoch": 1.68, "grad_norm": 8.52883529663086, "learning_rate": 8.528075119589528e-06, "loss": 1.0306, "step": 10448 }, { "epoch": 1.68, "grad_norm": 11.697381019592285, "learning_rate": 8.526356081192096e-06, "loss": 1.2232, "step": 10449 }, { "epoch": 1.68, "grad_norm": 12.653457641601562, "learning_rate": 8.524637087307744e-06, "loss": 1.1639, "step": 10450 }, { "epoch": 1.68, "grad_norm": 8.848342895507812, "learning_rate": 8.522918137988391e-06, "loss": 1.0725, "step": 10451 }, { "epoch": 1.68, "grad_norm": 11.016559600830078, "learning_rate": 8.521199233285967e-06, "loss": 1.256, "step": 10452 }, { "epoch": 1.68, "grad_norm": 9.256407737731934, "learning_rate": 8.51948037325239e-06, "loss": 0.9026, "step": 10453 }, { "epoch": 1.68, "grad_norm": 10.655709266662598, "learning_rate": 8.517761557939583e-06, "loss": 0.9384, "step": 10454 }, { "epoch": 1.68, "grad_norm": 9.466388702392578, "learning_rate": 8.516042787399459e-06, "loss": 1.0983, "step": 10455 }, { "epoch": 1.68, "grad_norm": 10.483145713806152, "learning_rate": 8.51432406168394e-06, "loss": 1.2493, "step": 10456 }, { "epoch": 1.68, "grad_norm": 9.578505516052246, "learning_rate": 8.512605380844938e-06, "loss": 1.0872, "step": 10457 }, { "epoch": 1.68, "grad_norm": 11.674986839294434, "learning_rate": 8.510886744934374e-06, "loss": 1.0777, "step": 10458 }, { "epoch": 1.68, "grad_norm": 9.27660083770752, "learning_rate": 8.509168154004156e-06, "loss": 1.0444, "step": 10459 }, { "epoch": 1.68, "grad_norm": 9.533966064453125, "learning_rate": 8.507449608106195e-06, "loss": 1.0192, "step": 10460 }, { "epoch": 1.68, "grad_norm": 8.874855041503906, "learning_rate": 8.505731107292407e-06, "loss": 1.0333, "step": 10461 }, { "epoch": 1.68, "grad_norm": 10.206679344177246, "learning_rate": 8.504012651614697e-06, "loss": 0.9347, "step": 10462 }, { "epoch": 1.68, "grad_norm": 9.51660442352295, "learning_rate": 8.502294241124976e-06, "loss": 1.146, "step": 10463 }, { "epoch": 1.68, "grad_norm": 12.591144561767578, "learning_rate": 8.50057587587515e-06, "loss": 1.0206, "step": 10464 }, { "epoch": 1.68, "grad_norm": 8.01523208618164, "learning_rate": 8.498857555917124e-06, "loss": 0.8683, "step": 10465 }, { "epoch": 1.68, "grad_norm": 10.822035789489746, "learning_rate": 8.497139281302798e-06, "loss": 0.9691, "step": 10466 }, { "epoch": 1.68, "grad_norm": 8.161030769348145, "learning_rate": 8.495421052084079e-06, "loss": 0.77, "step": 10467 }, { "epoch": 1.69, "grad_norm": 10.7968111038208, "learning_rate": 8.493702868312865e-06, "loss": 1.0684, "step": 10468 }, { "epoch": 1.69, "grad_norm": 9.675104141235352, "learning_rate": 8.49198473004106e-06, "loss": 1.1044, "step": 10469 }, { "epoch": 1.69, "grad_norm": 9.849163055419922, "learning_rate": 8.490266637320559e-06, "loss": 0.8532, "step": 10470 }, { "epoch": 1.69, "grad_norm": 10.365654945373535, "learning_rate": 8.488548590203255e-06, "loss": 1.2909, "step": 10471 }, { "epoch": 1.69, "grad_norm": 11.252812385559082, "learning_rate": 8.486830588741052e-06, "loss": 1.1017, "step": 10472 }, { "epoch": 1.69, "grad_norm": 10.016301155090332, "learning_rate": 8.48511263298584e-06, "loss": 1.1254, "step": 10473 }, { "epoch": 1.69, "grad_norm": 10.00963306427002, "learning_rate": 8.483394722989512e-06, "loss": 1.1966, "step": 10474 }, { "epoch": 1.69, "grad_norm": 11.61076831817627, "learning_rate": 8.481676858803957e-06, "loss": 1.1767, "step": 10475 }, { "epoch": 1.69, "grad_norm": 8.824531555175781, "learning_rate": 8.479959040481068e-06, "loss": 1.1084, "step": 10476 }, { "epoch": 1.69, "grad_norm": 9.916624069213867, "learning_rate": 8.478241268072733e-06, "loss": 0.9347, "step": 10477 }, { "epoch": 1.69, "grad_norm": 8.6437406539917, "learning_rate": 8.476523541630841e-06, "loss": 1.0579, "step": 10478 }, { "epoch": 1.69, "grad_norm": 9.21183967590332, "learning_rate": 8.474805861207275e-06, "loss": 1.1623, "step": 10479 }, { "epoch": 1.69, "grad_norm": 12.257196426391602, "learning_rate": 8.473088226853916e-06, "loss": 1.2968, "step": 10480 }, { "epoch": 1.69, "grad_norm": 11.710298538208008, "learning_rate": 8.471370638622657e-06, "loss": 1.0542, "step": 10481 }, { "epoch": 1.69, "grad_norm": 9.056360244750977, "learning_rate": 8.469653096565374e-06, "loss": 1.0151, "step": 10482 }, { "epoch": 1.69, "grad_norm": 11.173225402832031, "learning_rate": 8.467935600733945e-06, "loss": 0.9106, "step": 10483 }, { "epoch": 1.69, "grad_norm": 10.019292831420898, "learning_rate": 8.466218151180256e-06, "loss": 0.8351, "step": 10484 }, { "epoch": 1.69, "grad_norm": 11.352899551391602, "learning_rate": 8.464500747956178e-06, "loss": 1.1013, "step": 10485 }, { "epoch": 1.69, "grad_norm": 11.325071334838867, "learning_rate": 8.462783391113587e-06, "loss": 1.0568, "step": 10486 }, { "epoch": 1.69, "grad_norm": 11.323219299316406, "learning_rate": 8.46106608070436e-06, "loss": 0.9928, "step": 10487 }, { "epoch": 1.69, "grad_norm": 8.504738807678223, "learning_rate": 8.459348816780375e-06, "loss": 1.1041, "step": 10488 }, { "epoch": 1.69, "grad_norm": 7.890020370483398, "learning_rate": 8.457631599393495e-06, "loss": 0.9224, "step": 10489 }, { "epoch": 1.69, "grad_norm": 11.353263854980469, "learning_rate": 8.455914428595595e-06, "loss": 0.8374, "step": 10490 }, { "epoch": 1.69, "grad_norm": 11.485442161560059, "learning_rate": 8.454197304438544e-06, "loss": 0.9956, "step": 10491 }, { "epoch": 1.69, "grad_norm": 10.656789779663086, "learning_rate": 8.452480226974209e-06, "loss": 1.3034, "step": 10492 }, { "epoch": 1.69, "grad_norm": 8.19329833984375, "learning_rate": 8.450763196254458e-06, "loss": 0.8326, "step": 10493 }, { "epoch": 1.69, "grad_norm": 10.18601131439209, "learning_rate": 8.449046212331154e-06, "loss": 0.926, "step": 10494 }, { "epoch": 1.69, "grad_norm": 9.243255615234375, "learning_rate": 8.447329275256159e-06, "loss": 1.3367, "step": 10495 }, { "epoch": 1.69, "grad_norm": 10.601457595825195, "learning_rate": 8.445612385081336e-06, "loss": 1.1654, "step": 10496 }, { "epoch": 1.69, "grad_norm": 9.762310028076172, "learning_rate": 8.44389554185855e-06, "loss": 1.2257, "step": 10497 }, { "epoch": 1.69, "grad_norm": 7.915304660797119, "learning_rate": 8.442178745639656e-06, "loss": 0.9551, "step": 10498 }, { "epoch": 1.69, "grad_norm": 8.456809997558594, "learning_rate": 8.440461996476512e-06, "loss": 0.7439, "step": 10499 }, { "epoch": 1.69, "grad_norm": 12.526815414428711, "learning_rate": 8.43874529442097e-06, "loss": 1.2137, "step": 10500 }, { "epoch": 1.69, "grad_norm": 12.818961143493652, "learning_rate": 8.437028639524894e-06, "loss": 0.9477, "step": 10501 }, { "epoch": 1.69, "grad_norm": 8.834997177124023, "learning_rate": 8.435312031840133e-06, "loss": 1.3177, "step": 10502 }, { "epoch": 1.69, "grad_norm": 10.315476417541504, "learning_rate": 8.433595471418538e-06, "loss": 1.0529, "step": 10503 }, { "epoch": 1.69, "grad_norm": 9.1786470413208, "learning_rate": 8.431878958311965e-06, "loss": 1.2762, "step": 10504 }, { "epoch": 1.69, "grad_norm": 10.55362319946289, "learning_rate": 8.430162492572256e-06, "loss": 0.8099, "step": 10505 }, { "epoch": 1.69, "grad_norm": 9.893152236938477, "learning_rate": 8.428446074251262e-06, "loss": 0.9527, "step": 10506 }, { "epoch": 1.69, "grad_norm": 10.685412406921387, "learning_rate": 8.42672970340083e-06, "loss": 0.9826, "step": 10507 }, { "epoch": 1.69, "grad_norm": 9.660523414611816, "learning_rate": 8.425013380072805e-06, "loss": 0.9978, "step": 10508 }, { "epoch": 1.69, "grad_norm": 10.15749740600586, "learning_rate": 8.423297104319027e-06, "loss": 0.8378, "step": 10509 }, { "epoch": 1.69, "grad_norm": 9.910400390625, "learning_rate": 8.421580876191344e-06, "loss": 0.9949, "step": 10510 }, { "epoch": 1.69, "grad_norm": 9.938251495361328, "learning_rate": 8.419864695741591e-06, "loss": 0.9471, "step": 10511 }, { "epoch": 1.69, "grad_norm": 11.69528579711914, "learning_rate": 8.418148563021612e-06, "loss": 0.9513, "step": 10512 }, { "epoch": 1.69, "grad_norm": 9.954427719116211, "learning_rate": 8.416432478083241e-06, "loss": 1.0654, "step": 10513 }, { "epoch": 1.69, "grad_norm": 9.480574607849121, "learning_rate": 8.414716440978315e-06, "loss": 1.169, "step": 10514 }, { "epoch": 1.69, "grad_norm": 9.512276649475098, "learning_rate": 8.413000451758669e-06, "loss": 0.8685, "step": 10515 }, { "epoch": 1.69, "grad_norm": 8.990449905395508, "learning_rate": 8.411284510476139e-06, "loss": 0.875, "step": 10516 }, { "epoch": 1.69, "grad_norm": 11.851139068603516, "learning_rate": 8.409568617182552e-06, "loss": 1.2745, "step": 10517 }, { "epoch": 1.69, "grad_norm": 9.306056022644043, "learning_rate": 8.407852771929741e-06, "loss": 0.7978, "step": 10518 }, { "epoch": 1.69, "grad_norm": 9.934894561767578, "learning_rate": 8.406136974769537e-06, "loss": 1.0034, "step": 10519 }, { "epoch": 1.69, "grad_norm": 11.946309089660645, "learning_rate": 8.404421225753763e-06, "loss": 1.0563, "step": 10520 }, { "epoch": 1.69, "grad_norm": 12.034780502319336, "learning_rate": 8.402705524934248e-06, "loss": 1.0295, "step": 10521 }, { "epoch": 1.69, "grad_norm": 9.863040924072266, "learning_rate": 8.400989872362818e-06, "loss": 1.1804, "step": 10522 }, { "epoch": 1.69, "grad_norm": 9.83173656463623, "learning_rate": 8.399274268091298e-06, "loss": 0.9502, "step": 10523 }, { "epoch": 1.69, "grad_norm": 9.62880802154541, "learning_rate": 8.397558712171505e-06, "loss": 1.2945, "step": 10524 }, { "epoch": 1.69, "grad_norm": 9.844205856323242, "learning_rate": 8.395843204655259e-06, "loss": 1.1556, "step": 10525 }, { "epoch": 1.69, "grad_norm": 8.66625690460205, "learning_rate": 8.39412774559438e-06, "loss": 0.8803, "step": 10526 }, { "epoch": 1.69, "grad_norm": 10.591991424560547, "learning_rate": 8.392412335040691e-06, "loss": 1.1212, "step": 10527 }, { "epoch": 1.69, "grad_norm": 8.214179039001465, "learning_rate": 8.390696973046001e-06, "loss": 1.1441, "step": 10528 }, { "epoch": 1.69, "grad_norm": 9.644364356994629, "learning_rate": 8.388981659662126e-06, "loss": 0.9918, "step": 10529 }, { "epoch": 1.7, "grad_norm": 10.27985954284668, "learning_rate": 8.387266394940879e-06, "loss": 1.0251, "step": 10530 }, { "epoch": 1.7, "grad_norm": 9.579004287719727, "learning_rate": 8.385551178934071e-06, "loss": 0.8608, "step": 10531 }, { "epoch": 1.7, "grad_norm": 9.639556884765625, "learning_rate": 8.383836011693516e-06, "loss": 1.2276, "step": 10532 }, { "epoch": 1.7, "grad_norm": 8.956311225891113, "learning_rate": 8.382120893271018e-06, "loss": 1.071, "step": 10533 }, { "epoch": 1.7, "grad_norm": 12.311996459960938, "learning_rate": 8.380405823718385e-06, "loss": 1.1825, "step": 10534 }, { "epoch": 1.7, "grad_norm": 11.358734130859375, "learning_rate": 8.378690803087424e-06, "loss": 1.1416, "step": 10535 }, { "epoch": 1.7, "grad_norm": 11.30906867980957, "learning_rate": 8.37697583142994e-06, "loss": 1.1226, "step": 10536 }, { "epoch": 1.7, "grad_norm": 10.663459777832031, "learning_rate": 8.375260908797732e-06, "loss": 1.0908, "step": 10537 }, { "epoch": 1.7, "grad_norm": 9.695487022399902, "learning_rate": 8.373546035242603e-06, "loss": 1.248, "step": 10538 }, { "epoch": 1.7, "grad_norm": 8.050065994262695, "learning_rate": 8.371831210816353e-06, "loss": 1.0751, "step": 10539 }, { "epoch": 1.7, "grad_norm": 9.608640670776367, "learning_rate": 8.370116435570783e-06, "loss": 0.921, "step": 10540 }, { "epoch": 1.7, "grad_norm": 9.954824447631836, "learning_rate": 8.368401709557681e-06, "loss": 1.2484, "step": 10541 }, { "epoch": 1.7, "grad_norm": 10.133045196533203, "learning_rate": 8.366687032828852e-06, "loss": 1.0363, "step": 10542 }, { "epoch": 1.7, "grad_norm": 8.014018058776855, "learning_rate": 8.364972405436087e-06, "loss": 1.0373, "step": 10543 }, { "epoch": 1.7, "grad_norm": 8.8778076171875, "learning_rate": 8.363257827431176e-06, "loss": 1.0739, "step": 10544 }, { "epoch": 1.7, "grad_norm": 8.822988510131836, "learning_rate": 8.361543298865911e-06, "loss": 0.9553, "step": 10545 }, { "epoch": 1.7, "grad_norm": 9.407669067382812, "learning_rate": 8.35982881979208e-06, "loss": 1.0288, "step": 10546 }, { "epoch": 1.7, "grad_norm": 11.169160842895508, "learning_rate": 8.358114390261474e-06, "loss": 1.3612, "step": 10547 }, { "epoch": 1.7, "grad_norm": 10.261102676391602, "learning_rate": 8.356400010325877e-06, "loss": 1.0674, "step": 10548 }, { "epoch": 1.7, "grad_norm": 9.085053443908691, "learning_rate": 8.354685680037073e-06, "loss": 1.0414, "step": 10549 }, { "epoch": 1.7, "grad_norm": 9.632563591003418, "learning_rate": 8.352971399446847e-06, "loss": 0.9941, "step": 10550 }, { "epoch": 1.7, "grad_norm": 7.958868026733398, "learning_rate": 8.351257168606982e-06, "loss": 0.8754, "step": 10551 }, { "epoch": 1.7, "grad_norm": 9.49696159362793, "learning_rate": 8.349542987569252e-06, "loss": 0.8346, "step": 10552 }, { "epoch": 1.7, "grad_norm": 10.214794158935547, "learning_rate": 8.347828856385443e-06, "loss": 1.3554, "step": 10553 }, { "epoch": 1.7, "grad_norm": 8.131903648376465, "learning_rate": 8.346114775107329e-06, "loss": 0.847, "step": 10554 }, { "epoch": 1.7, "grad_norm": 8.666454315185547, "learning_rate": 8.344400743786688e-06, "loss": 0.9012, "step": 10555 }, { "epoch": 1.7, "grad_norm": 8.890863418579102, "learning_rate": 8.342686762475289e-06, "loss": 1.0792, "step": 10556 }, { "epoch": 1.7, "grad_norm": 8.970747947692871, "learning_rate": 8.340972831224909e-06, "loss": 1.3496, "step": 10557 }, { "epoch": 1.7, "grad_norm": 8.949664115905762, "learning_rate": 8.339258950087318e-06, "loss": 0.8414, "step": 10558 }, { "epoch": 1.7, "grad_norm": 10.798364639282227, "learning_rate": 8.337545119114287e-06, "loss": 1.2122, "step": 10559 }, { "epoch": 1.7, "grad_norm": 9.120484352111816, "learning_rate": 8.335831338357584e-06, "loss": 0.9708, "step": 10560 }, { "epoch": 1.7, "grad_norm": 9.684435844421387, "learning_rate": 8.33411760786897e-06, "loss": 1.0928, "step": 10561 }, { "epoch": 1.7, "grad_norm": 9.349882125854492, "learning_rate": 8.332403927700221e-06, "loss": 1.0624, "step": 10562 }, { "epoch": 1.7, "grad_norm": 9.632173538208008, "learning_rate": 8.330690297903093e-06, "loss": 0.9248, "step": 10563 }, { "epoch": 1.7, "grad_norm": 9.820125579833984, "learning_rate": 8.328976718529349e-06, "loss": 1.0019, "step": 10564 }, { "epoch": 1.7, "grad_norm": 8.080605506896973, "learning_rate": 8.32726318963075e-06, "loss": 0.9415, "step": 10565 }, { "epoch": 1.7, "grad_norm": 9.961236000061035, "learning_rate": 8.32554971125906e-06, "loss": 1.0777, "step": 10566 }, { "epoch": 1.7, "grad_norm": 8.911270141601562, "learning_rate": 8.323836283466028e-06, "loss": 1.2191, "step": 10567 }, { "epoch": 1.7, "grad_norm": 7.439207077026367, "learning_rate": 8.322122906303417e-06, "loss": 0.8221, "step": 10568 }, { "epoch": 1.7, "grad_norm": 8.83049201965332, "learning_rate": 8.320409579822977e-06, "loss": 1.0789, "step": 10569 }, { "epoch": 1.7, "grad_norm": 10.537064552307129, "learning_rate": 8.31869630407646e-06, "loss": 1.0695, "step": 10570 }, { "epoch": 1.7, "grad_norm": 8.445265769958496, "learning_rate": 8.316983079115625e-06, "loss": 1.3025, "step": 10571 }, { "epoch": 1.7, "grad_norm": 8.772432327270508, "learning_rate": 8.315269904992215e-06, "loss": 1.1565, "step": 10572 }, { "epoch": 1.7, "grad_norm": 9.00655460357666, "learning_rate": 8.31355678175798e-06, "loss": 1.1178, "step": 10573 }, { "epoch": 1.7, "grad_norm": 9.876689910888672, "learning_rate": 8.311843709464666e-06, "loss": 1.0408, "step": 10574 }, { "epoch": 1.7, "grad_norm": 9.170642852783203, "learning_rate": 8.310130688164023e-06, "loss": 1.1865, "step": 10575 }, { "epoch": 1.7, "grad_norm": 8.599453926086426, "learning_rate": 8.308417717907789e-06, "loss": 0.938, "step": 10576 }, { "epoch": 1.7, "grad_norm": 9.908254623413086, "learning_rate": 8.306704798747708e-06, "loss": 1.1067, "step": 10577 }, { "epoch": 1.7, "grad_norm": 11.775700569152832, "learning_rate": 8.30499193073552e-06, "loss": 0.9711, "step": 10578 }, { "epoch": 1.7, "grad_norm": 10.566617965698242, "learning_rate": 8.303279113922968e-06, "loss": 1.0818, "step": 10579 }, { "epoch": 1.7, "grad_norm": 10.760965347290039, "learning_rate": 8.301566348361785e-06, "loss": 1.0102, "step": 10580 }, { "epoch": 1.7, "grad_norm": 9.607488632202148, "learning_rate": 8.299853634103704e-06, "loss": 0.9523, "step": 10581 }, { "epoch": 1.7, "grad_norm": 9.72463321685791, "learning_rate": 8.29814097120047e-06, "loss": 1.0259, "step": 10582 }, { "epoch": 1.7, "grad_norm": 11.486945152282715, "learning_rate": 8.296428359703809e-06, "loss": 1.0507, "step": 10583 }, { "epoch": 1.7, "grad_norm": 9.463421821594238, "learning_rate": 8.294715799665453e-06, "loss": 0.8226, "step": 10584 }, { "epoch": 1.7, "grad_norm": 10.102078437805176, "learning_rate": 8.293003291137133e-06, "loss": 1.0382, "step": 10585 }, { "epoch": 1.7, "grad_norm": 7.959240436553955, "learning_rate": 8.291290834170578e-06, "loss": 0.8562, "step": 10586 }, { "epoch": 1.7, "grad_norm": 12.639799118041992, "learning_rate": 8.28957842881751e-06, "loss": 1.1182, "step": 10587 }, { "epoch": 1.7, "grad_norm": 12.6864013671875, "learning_rate": 8.287866075129658e-06, "loss": 0.9695, "step": 10588 }, { "epoch": 1.7, "grad_norm": 10.418194770812988, "learning_rate": 8.286153773158745e-06, "loss": 0.8645, "step": 10589 }, { "epoch": 1.7, "grad_norm": 9.062028884887695, "learning_rate": 8.284441522956497e-06, "loss": 0.9352, "step": 10590 }, { "epoch": 1.7, "grad_norm": 9.205877304077148, "learning_rate": 8.282729324574627e-06, "loss": 1.039, "step": 10591 }, { "epoch": 1.71, "grad_norm": 9.568358421325684, "learning_rate": 8.281017178064856e-06, "loss": 1.18, "step": 10592 }, { "epoch": 1.71, "grad_norm": 9.313754081726074, "learning_rate": 8.279305083478903e-06, "loss": 0.9404, "step": 10593 }, { "epoch": 1.71, "grad_norm": 10.420875549316406, "learning_rate": 8.277593040868487e-06, "loss": 1.1063, "step": 10594 }, { "epoch": 1.71, "grad_norm": 10.934425354003906, "learning_rate": 8.275881050285315e-06, "loss": 1.245, "step": 10595 }, { "epoch": 1.71, "grad_norm": 9.623065948486328, "learning_rate": 8.274169111781105e-06, "loss": 1.0949, "step": 10596 }, { "epoch": 1.71, "grad_norm": 9.348462104797363, "learning_rate": 8.272457225407564e-06, "loss": 1.1195, "step": 10597 }, { "epoch": 1.71, "grad_norm": 10.884109497070312, "learning_rate": 8.270745391216405e-06, "loss": 1.1638, "step": 10598 }, { "epoch": 1.71, "grad_norm": 8.921932220458984, "learning_rate": 8.269033609259335e-06, "loss": 1.0066, "step": 10599 }, { "epoch": 1.71, "grad_norm": 8.875936508178711, "learning_rate": 8.26732187958806e-06, "loss": 0.8823, "step": 10600 }, { "epoch": 1.71, "grad_norm": 9.222378730773926, "learning_rate": 8.265610202254283e-06, "loss": 0.9075, "step": 10601 }, { "epoch": 1.71, "grad_norm": 9.498490333557129, "learning_rate": 8.263898577309707e-06, "loss": 1.0636, "step": 10602 }, { "epoch": 1.71, "grad_norm": 9.308379173278809, "learning_rate": 8.262187004806038e-06, "loss": 0.8136, "step": 10603 }, { "epoch": 1.71, "grad_norm": 11.930837631225586, "learning_rate": 8.260475484794972e-06, "loss": 1.2121, "step": 10604 }, { "epoch": 1.71, "grad_norm": 10.079641342163086, "learning_rate": 8.258764017328212e-06, "loss": 0.9899, "step": 10605 }, { "epoch": 1.71, "grad_norm": 8.543861389160156, "learning_rate": 8.257052602457449e-06, "loss": 0.7546, "step": 10606 }, { "epoch": 1.71, "grad_norm": 10.838953018188477, "learning_rate": 8.255341240234381e-06, "loss": 1.2039, "step": 10607 }, { "epoch": 1.71, "grad_norm": 9.245062828063965, "learning_rate": 8.2536299307107e-06, "loss": 1.215, "step": 10608 }, { "epoch": 1.71, "grad_norm": 8.231213569641113, "learning_rate": 8.2519186739381e-06, "loss": 0.9878, "step": 10609 }, { "epoch": 1.71, "grad_norm": 8.123528480529785, "learning_rate": 8.250207469968272e-06, "loss": 1.0641, "step": 10610 }, { "epoch": 1.71, "grad_norm": 8.682766914367676, "learning_rate": 8.248496318852903e-06, "loss": 1.013, "step": 10611 }, { "epoch": 1.71, "grad_norm": 9.120838165283203, "learning_rate": 8.246785220643681e-06, "loss": 1.1713, "step": 10612 }, { "epoch": 1.71, "grad_norm": 9.080404281616211, "learning_rate": 8.24507417539229e-06, "loss": 1.1497, "step": 10613 }, { "epoch": 1.71, "grad_norm": 9.638849258422852, "learning_rate": 8.24336318315042e-06, "loss": 1.0368, "step": 10614 }, { "epoch": 1.71, "grad_norm": 9.300198554992676, "learning_rate": 8.241652243969744e-06, "loss": 1.1565, "step": 10615 }, { "epoch": 1.71, "grad_norm": 9.980443954467773, "learning_rate": 8.23994135790195e-06, "loss": 1.1106, "step": 10616 }, { "epoch": 1.71, "grad_norm": 9.836926460266113, "learning_rate": 8.238230524998716e-06, "loss": 0.9911, "step": 10617 }, { "epoch": 1.71, "grad_norm": 9.731222152709961, "learning_rate": 8.236519745311719e-06, "loss": 0.9781, "step": 10618 }, { "epoch": 1.71, "grad_norm": 10.311442375183105, "learning_rate": 8.234809018892634e-06, "loss": 0.9402, "step": 10619 }, { "epoch": 1.71, "grad_norm": 8.896905899047852, "learning_rate": 8.233098345793134e-06, "loss": 0.9715, "step": 10620 }, { "epoch": 1.71, "grad_norm": 10.004738807678223, "learning_rate": 8.231387726064897e-06, "loss": 1.1961, "step": 10621 }, { "epoch": 1.71, "grad_norm": 10.345427513122559, "learning_rate": 8.229677159759588e-06, "loss": 1.0435, "step": 10622 }, { "epoch": 1.71, "grad_norm": 9.391587257385254, "learning_rate": 8.22796664692888e-06, "loss": 1.1922, "step": 10623 }, { "epoch": 1.71, "grad_norm": 13.196383476257324, "learning_rate": 8.226256187624443e-06, "loss": 1.0952, "step": 10624 }, { "epoch": 1.71, "grad_norm": 10.762418746948242, "learning_rate": 8.224545781897944e-06, "loss": 1.1845, "step": 10625 }, { "epoch": 1.71, "grad_norm": 9.697091102600098, "learning_rate": 8.22283542980104e-06, "loss": 1.1556, "step": 10626 }, { "epoch": 1.71, "grad_norm": 10.60952377319336, "learning_rate": 8.221125131385401e-06, "loss": 0.8952, "step": 10627 }, { "epoch": 1.71, "grad_norm": 9.252754211425781, "learning_rate": 8.219414886702686e-06, "loss": 1.0332, "step": 10628 }, { "epoch": 1.71, "grad_norm": 12.352110862731934, "learning_rate": 8.217704695804558e-06, "loss": 0.8385, "step": 10629 }, { "epoch": 1.71, "grad_norm": 11.718947410583496, "learning_rate": 8.21599455874267e-06, "loss": 1.1107, "step": 10630 }, { "epoch": 1.71, "grad_norm": 9.368708610534668, "learning_rate": 8.214284475568681e-06, "loss": 0.9955, "step": 10631 }, { "epoch": 1.71, "grad_norm": 9.948530197143555, "learning_rate": 8.212574446334247e-06, "loss": 1.0254, "step": 10632 }, { "epoch": 1.71, "grad_norm": 12.781576156616211, "learning_rate": 8.210864471091022e-06, "loss": 0.9493, "step": 10633 }, { "epoch": 1.71, "grad_norm": 10.804616928100586, "learning_rate": 8.209154549890655e-06, "loss": 1.1908, "step": 10634 }, { "epoch": 1.71, "grad_norm": 8.808266639709473, "learning_rate": 8.207444682784798e-06, "loss": 1.0953, "step": 10635 }, { "epoch": 1.71, "grad_norm": 13.190524101257324, "learning_rate": 8.205734869825098e-06, "loss": 0.9661, "step": 10636 }, { "epoch": 1.71, "grad_norm": 9.687877655029297, "learning_rate": 8.204025111063204e-06, "loss": 1.1645, "step": 10637 }, { "epoch": 1.71, "grad_norm": 10.13005256652832, "learning_rate": 8.20231540655076e-06, "loss": 1.2698, "step": 10638 }, { "epoch": 1.71, "grad_norm": 11.664241790771484, "learning_rate": 8.200605756339408e-06, "loss": 1.1228, "step": 10639 }, { "epoch": 1.71, "grad_norm": 12.704874038696289, "learning_rate": 8.198896160480791e-06, "loss": 1.1278, "step": 10640 }, { "epoch": 1.71, "grad_norm": 12.357856750488281, "learning_rate": 8.197186619026552e-06, "loss": 0.9321, "step": 10641 }, { "epoch": 1.71, "grad_norm": 10.627269744873047, "learning_rate": 8.195477132028322e-06, "loss": 1.3356, "step": 10642 }, { "epoch": 1.71, "grad_norm": 10.552791595458984, "learning_rate": 8.193767699537747e-06, "loss": 1.1125, "step": 10643 }, { "epoch": 1.71, "grad_norm": 8.345928192138672, "learning_rate": 8.192058321606463e-06, "loss": 0.886, "step": 10644 }, { "epoch": 1.71, "grad_norm": 9.39791488647461, "learning_rate": 8.190348998286092e-06, "loss": 1.1497, "step": 10645 }, { "epoch": 1.71, "grad_norm": 9.361605644226074, "learning_rate": 8.188639729628278e-06, "loss": 0.925, "step": 10646 }, { "epoch": 1.71, "grad_norm": 10.980850219726562, "learning_rate": 8.186930515684644e-06, "loss": 1.0247, "step": 10647 }, { "epoch": 1.71, "grad_norm": 13.982170104980469, "learning_rate": 8.185221356506825e-06, "loss": 1.3141, "step": 10648 }, { "epoch": 1.71, "grad_norm": 9.501710891723633, "learning_rate": 8.183512252146443e-06, "loss": 1.0732, "step": 10649 }, { "epoch": 1.71, "grad_norm": 12.163394927978516, "learning_rate": 8.181803202655125e-06, "loss": 1.005, "step": 10650 }, { "epoch": 1.71, "grad_norm": 9.367094993591309, "learning_rate": 8.180094208084494e-06, "loss": 1.1852, "step": 10651 }, { "epoch": 1.71, "grad_norm": 11.168050765991211, "learning_rate": 8.178385268486171e-06, "loss": 0.966, "step": 10652 }, { "epoch": 1.71, "grad_norm": 11.081774711608887, "learning_rate": 8.176676383911783e-06, "loss": 0.974, "step": 10653 }, { "epoch": 1.72, "grad_norm": 10.85493278503418, "learning_rate": 8.174967554412942e-06, "loss": 1.0034, "step": 10654 }, { "epoch": 1.72, "grad_norm": 9.858939170837402, "learning_rate": 8.173258780041266e-06, "loss": 1.1424, "step": 10655 }, { "epoch": 1.72, "grad_norm": 9.970380783081055, "learning_rate": 8.171550060848373e-06, "loss": 0.9785, "step": 10656 }, { "epoch": 1.72, "grad_norm": 8.787631034851074, "learning_rate": 8.169841396885878e-06, "loss": 1.1471, "step": 10657 }, { "epoch": 1.72, "grad_norm": 9.933363914489746, "learning_rate": 8.168132788205387e-06, "loss": 1.0286, "step": 10658 }, { "epoch": 1.72, "grad_norm": 10.117094993591309, "learning_rate": 8.166424234858515e-06, "loss": 1.175, "step": 10659 }, { "epoch": 1.72, "grad_norm": 10.038126945495605, "learning_rate": 8.16471573689687e-06, "loss": 0.9239, "step": 10660 }, { "epoch": 1.72, "grad_norm": 8.577765464782715, "learning_rate": 8.16300729437206e-06, "loss": 0.9122, "step": 10661 }, { "epoch": 1.72, "grad_norm": 10.476363182067871, "learning_rate": 8.161298907335685e-06, "loss": 1.191, "step": 10662 }, { "epoch": 1.72, "grad_norm": 8.554032325744629, "learning_rate": 8.159590575839355e-06, "loss": 0.9856, "step": 10663 }, { "epoch": 1.72, "grad_norm": 12.369261741638184, "learning_rate": 8.157882299934674e-06, "loss": 1.1273, "step": 10664 }, { "epoch": 1.72, "grad_norm": 9.744478225708008, "learning_rate": 8.156174079673236e-06, "loss": 1.0368, "step": 10665 }, { "epoch": 1.72, "grad_norm": 11.759880065917969, "learning_rate": 8.154465915106643e-06, "loss": 1.2649, "step": 10666 }, { "epoch": 1.72, "grad_norm": 10.258906364440918, "learning_rate": 8.152757806286492e-06, "loss": 1.0346, "step": 10667 }, { "epoch": 1.72, "grad_norm": 9.265522956848145, "learning_rate": 8.151049753264377e-06, "loss": 1.1034, "step": 10668 }, { "epoch": 1.72, "grad_norm": 10.684480667114258, "learning_rate": 8.149341756091892e-06, "loss": 1.2393, "step": 10669 }, { "epoch": 1.72, "grad_norm": 9.410978317260742, "learning_rate": 8.14763381482063e-06, "loss": 1.1186, "step": 10670 }, { "epoch": 1.72, "grad_norm": 8.898233413696289, "learning_rate": 8.14592592950218e-06, "loss": 1.037, "step": 10671 }, { "epoch": 1.72, "grad_norm": 10.951035499572754, "learning_rate": 8.144218100188134e-06, "loss": 1.0809, "step": 10672 }, { "epoch": 1.72, "grad_norm": 10.57317066192627, "learning_rate": 8.142510326930075e-06, "loss": 1.0541, "step": 10673 }, { "epoch": 1.72, "grad_norm": 8.311800003051758, "learning_rate": 8.140802609779588e-06, "loss": 0.9615, "step": 10674 }, { "epoch": 1.72, "grad_norm": 10.55412483215332, "learning_rate": 8.13909494878826e-06, "loss": 0.9672, "step": 10675 }, { "epoch": 1.72, "grad_norm": 9.551758766174316, "learning_rate": 8.137387344007668e-06, "loss": 0.7706, "step": 10676 }, { "epoch": 1.72, "grad_norm": 9.764894485473633, "learning_rate": 8.1356797954894e-06, "loss": 1.0804, "step": 10677 }, { "epoch": 1.72, "grad_norm": 9.056065559387207, "learning_rate": 8.133972303285024e-06, "loss": 1.1967, "step": 10678 }, { "epoch": 1.72, "grad_norm": 8.708596229553223, "learning_rate": 8.132264867446124e-06, "loss": 1.035, "step": 10679 }, { "epoch": 1.72, "grad_norm": 10.040414810180664, "learning_rate": 8.130557488024274e-06, "loss": 0.9623, "step": 10680 }, { "epoch": 1.72, "grad_norm": 10.101503372192383, "learning_rate": 8.128850165071047e-06, "loss": 0.9088, "step": 10681 }, { "epoch": 1.72, "grad_norm": 9.725116729736328, "learning_rate": 8.127142898638013e-06, "loss": 0.9146, "step": 10682 }, { "epoch": 1.72, "grad_norm": 9.057064056396484, "learning_rate": 8.125435688776741e-06, "loss": 0.9629, "step": 10683 }, { "epoch": 1.72, "grad_norm": 9.971576690673828, "learning_rate": 8.123728535538803e-06, "loss": 1.2466, "step": 10684 }, { "epoch": 1.72, "grad_norm": 9.331036567687988, "learning_rate": 8.122021438975764e-06, "loss": 1.0473, "step": 10685 }, { "epoch": 1.72, "grad_norm": 10.959858894348145, "learning_rate": 8.120314399139188e-06, "loss": 1.3079, "step": 10686 }, { "epoch": 1.72, "grad_norm": 10.729276657104492, "learning_rate": 8.118607416080637e-06, "loss": 1.1351, "step": 10687 }, { "epoch": 1.72, "grad_norm": 10.591802597045898, "learning_rate": 8.116900489851678e-06, "loss": 1.0927, "step": 10688 }, { "epoch": 1.72, "grad_norm": 9.684961318969727, "learning_rate": 8.115193620503864e-06, "loss": 1.2768, "step": 10689 }, { "epoch": 1.72, "grad_norm": 9.473621368408203, "learning_rate": 8.113486808088755e-06, "loss": 1.2041, "step": 10690 }, { "epoch": 1.72, "grad_norm": 7.4697794914245605, "learning_rate": 8.11178005265791e-06, "loss": 0.8309, "step": 10691 }, { "epoch": 1.72, "grad_norm": 13.127312660217285, "learning_rate": 8.11007335426288e-06, "loss": 1.1475, "step": 10692 }, { "epoch": 1.72, "grad_norm": 11.159492492675781, "learning_rate": 8.108366712955218e-06, "loss": 1.2049, "step": 10693 }, { "epoch": 1.72, "grad_norm": 9.408565521240234, "learning_rate": 8.106660128786476e-06, "loss": 1.1747, "step": 10694 }, { "epoch": 1.72, "grad_norm": 9.128327369689941, "learning_rate": 8.104953601808202e-06, "loss": 1.0447, "step": 10695 }, { "epoch": 1.72, "grad_norm": 10.083331108093262, "learning_rate": 8.10324713207195e-06, "loss": 1.0055, "step": 10696 }, { "epoch": 1.72, "grad_norm": 10.217788696289062, "learning_rate": 8.101540719629256e-06, "loss": 1.1719, "step": 10697 }, { "epoch": 1.72, "grad_norm": 11.64471435546875, "learning_rate": 8.099834364531669e-06, "loss": 1.2416, "step": 10698 }, { "epoch": 1.72, "grad_norm": 10.458252906799316, "learning_rate": 8.098128066830733e-06, "loss": 1.1483, "step": 10699 }, { "epoch": 1.72, "grad_norm": 11.574922561645508, "learning_rate": 8.096421826577984e-06, "loss": 1.1441, "step": 10700 }, { "epoch": 1.72, "grad_norm": 10.38573932647705, "learning_rate": 8.094715643824968e-06, "loss": 1.0225, "step": 10701 }, { "epoch": 1.72, "grad_norm": 10.606298446655273, "learning_rate": 8.093009518623213e-06, "loss": 0.9994, "step": 10702 }, { "epoch": 1.72, "grad_norm": 9.434958457946777, "learning_rate": 8.09130345102426e-06, "loss": 1.2731, "step": 10703 }, { "epoch": 1.72, "grad_norm": 9.334310531616211, "learning_rate": 8.089597441079644e-06, "loss": 0.9746, "step": 10704 }, { "epoch": 1.72, "grad_norm": 12.434534072875977, "learning_rate": 8.087891488840893e-06, "loss": 1.1675, "step": 10705 }, { "epoch": 1.72, "grad_norm": 10.706040382385254, "learning_rate": 8.08618559435954e-06, "loss": 1.2885, "step": 10706 }, { "epoch": 1.72, "grad_norm": 9.229020118713379, "learning_rate": 8.084479757687115e-06, "loss": 0.8118, "step": 10707 }, { "epoch": 1.72, "grad_norm": 10.269172668457031, "learning_rate": 8.082773978875141e-06, "loss": 1.3305, "step": 10708 }, { "epoch": 1.72, "grad_norm": 8.97283935546875, "learning_rate": 8.081068257975145e-06, "loss": 1.1327, "step": 10709 }, { "epoch": 1.72, "grad_norm": 10.342401504516602, "learning_rate": 8.079362595038647e-06, "loss": 1.046, "step": 10710 }, { "epoch": 1.72, "grad_norm": 10.53890609741211, "learning_rate": 8.077656990117175e-06, "loss": 1.2196, "step": 10711 }, { "epoch": 1.72, "grad_norm": 9.88355827331543, "learning_rate": 8.075951443262241e-06, "loss": 1.2991, "step": 10712 }, { "epoch": 1.72, "grad_norm": 10.603734970092773, "learning_rate": 8.074245954525368e-06, "loss": 1.0175, "step": 10713 }, { "epoch": 1.72, "grad_norm": 9.760869026184082, "learning_rate": 8.07254052395807e-06, "loss": 0.8741, "step": 10714 }, { "epoch": 1.72, "grad_norm": 7.799349308013916, "learning_rate": 8.070835151611864e-06, "loss": 0.7316, "step": 10715 }, { "epoch": 1.73, "grad_norm": 9.30738353729248, "learning_rate": 8.069129837538262e-06, "loss": 0.9024, "step": 10716 }, { "epoch": 1.73, "grad_norm": 9.594524383544922, "learning_rate": 8.067424581788772e-06, "loss": 0.9434, "step": 10717 }, { "epoch": 1.73, "grad_norm": 8.383943557739258, "learning_rate": 8.065719384414905e-06, "loss": 0.8479, "step": 10718 }, { "epoch": 1.73, "grad_norm": 10.038871765136719, "learning_rate": 8.06401424546817e-06, "loss": 1.1475, "step": 10719 }, { "epoch": 1.73, "grad_norm": 9.325434684753418, "learning_rate": 8.062309165000072e-06, "loss": 1.2278, "step": 10720 }, { "epoch": 1.73, "grad_norm": 11.689745903015137, "learning_rate": 8.060604143062113e-06, "loss": 0.847, "step": 10721 }, { "epoch": 1.73, "grad_norm": 9.866172790527344, "learning_rate": 8.058899179705794e-06, "loss": 1.0124, "step": 10722 }, { "epoch": 1.73, "grad_norm": 8.487788200378418, "learning_rate": 8.057194274982617e-06, "loss": 1.066, "step": 10723 }, { "epoch": 1.73, "grad_norm": 10.182668685913086, "learning_rate": 8.055489428944086e-06, "loss": 1.2407, "step": 10724 }, { "epoch": 1.73, "grad_norm": 8.229372024536133, "learning_rate": 8.05378464164169e-06, "loss": 1.1436, "step": 10725 }, { "epoch": 1.73, "grad_norm": 12.513999938964844, "learning_rate": 8.052079913126926e-06, "loss": 0.9842, "step": 10726 }, { "epoch": 1.73, "grad_norm": 11.047411918640137, "learning_rate": 8.050375243451294e-06, "loss": 0.8762, "step": 10727 }, { "epoch": 1.73, "grad_norm": 9.298521041870117, "learning_rate": 8.048670632666275e-06, "loss": 1.0223, "step": 10728 }, { "epoch": 1.73, "grad_norm": 8.96346378326416, "learning_rate": 8.046966080823363e-06, "loss": 0.8838, "step": 10729 }, { "epoch": 1.73, "grad_norm": 9.837844848632812, "learning_rate": 8.045261587974048e-06, "loss": 1.0624, "step": 10730 }, { "epoch": 1.73, "grad_norm": 10.142890930175781, "learning_rate": 8.043557154169816e-06, "loss": 1.1894, "step": 10731 }, { "epoch": 1.73, "grad_norm": 8.147934913635254, "learning_rate": 8.041852779462148e-06, "loss": 1.048, "step": 10732 }, { "epoch": 1.73, "grad_norm": 9.85078239440918, "learning_rate": 8.040148463902528e-06, "loss": 0.8407, "step": 10733 }, { "epoch": 1.73, "grad_norm": 10.518953323364258, "learning_rate": 8.038444207542438e-06, "loss": 1.2673, "step": 10734 }, { "epoch": 1.73, "grad_norm": 8.902942657470703, "learning_rate": 8.036740010433357e-06, "loss": 1.0448, "step": 10735 }, { "epoch": 1.73, "grad_norm": 9.430381774902344, "learning_rate": 8.03503587262676e-06, "loss": 0.8808, "step": 10736 }, { "epoch": 1.73, "grad_norm": 9.968650817871094, "learning_rate": 8.033331794174123e-06, "loss": 1.2183, "step": 10737 }, { "epoch": 1.73, "grad_norm": 14.129423141479492, "learning_rate": 8.03162777512692e-06, "loss": 1.1277, "step": 10738 }, { "epoch": 1.73, "grad_norm": 7.680530071258545, "learning_rate": 8.029923815536624e-06, "loss": 1.1841, "step": 10739 }, { "epoch": 1.73, "grad_norm": 11.005273818969727, "learning_rate": 8.028219915454708e-06, "loss": 1.3092, "step": 10740 }, { "epoch": 1.73, "grad_norm": 11.084786415100098, "learning_rate": 8.026516074932631e-06, "loss": 1.1, "step": 10741 }, { "epoch": 1.73, "grad_norm": 10.528536796569824, "learning_rate": 8.024812294021868e-06, "loss": 1.1112, "step": 10742 }, { "epoch": 1.73, "grad_norm": 10.001192092895508, "learning_rate": 8.023108572773877e-06, "loss": 0.939, "step": 10743 }, { "epoch": 1.73, "grad_norm": 11.194819450378418, "learning_rate": 8.021404911240125e-06, "loss": 0.9014, "step": 10744 }, { "epoch": 1.73, "grad_norm": 10.288969039916992, "learning_rate": 8.019701309472074e-06, "loss": 1.2622, "step": 10745 }, { "epoch": 1.73, "grad_norm": 9.3636474609375, "learning_rate": 8.017997767521184e-06, "loss": 1.0511, "step": 10746 }, { "epoch": 1.73, "grad_norm": 9.086624145507812, "learning_rate": 8.016294285438907e-06, "loss": 0.9546, "step": 10747 }, { "epoch": 1.73, "grad_norm": 8.291153907775879, "learning_rate": 8.014590863276702e-06, "loss": 0.8528, "step": 10748 }, { "epoch": 1.73, "grad_norm": 9.365188598632812, "learning_rate": 8.012887501086022e-06, "loss": 0.8925, "step": 10749 }, { "epoch": 1.73, "grad_norm": 9.878602981567383, "learning_rate": 8.011184198918322e-06, "loss": 0.8458, "step": 10750 }, { "epoch": 1.73, "grad_norm": 9.592986106872559, "learning_rate": 8.009480956825047e-06, "loss": 1.0177, "step": 10751 }, { "epoch": 1.73, "grad_norm": 8.453824043273926, "learning_rate": 8.007777774857648e-06, "loss": 1.1234, "step": 10752 }, { "epoch": 1.73, "grad_norm": 9.931802749633789, "learning_rate": 8.006074653067571e-06, "loss": 1.0206, "step": 10753 }, { "epoch": 1.73, "grad_norm": 10.548913955688477, "learning_rate": 8.004371591506261e-06, "loss": 0.9932, "step": 10754 }, { "epoch": 1.73, "grad_norm": 9.743250846862793, "learning_rate": 8.002668590225163e-06, "loss": 1.2033, "step": 10755 }, { "epoch": 1.73, "grad_norm": 9.767607688903809, "learning_rate": 8.000965649275716e-06, "loss": 0.8538, "step": 10756 }, { "epoch": 1.73, "grad_norm": 9.980690956115723, "learning_rate": 7.999262768709356e-06, "loss": 1.1674, "step": 10757 }, { "epoch": 1.73, "grad_norm": 7.944430351257324, "learning_rate": 7.997559948577526e-06, "loss": 0.9002, "step": 10758 }, { "epoch": 1.73, "grad_norm": 10.003509521484375, "learning_rate": 7.995857188931661e-06, "loss": 0.947, "step": 10759 }, { "epoch": 1.73, "grad_norm": 11.391385078430176, "learning_rate": 7.994154489823192e-06, "loss": 0.9268, "step": 10760 }, { "epoch": 1.73, "grad_norm": 9.884571075439453, "learning_rate": 7.992451851303551e-06, "loss": 0.8797, "step": 10761 }, { "epoch": 1.73, "grad_norm": 9.471447944641113, "learning_rate": 7.990749273424169e-06, "loss": 0.9565, "step": 10762 }, { "epoch": 1.73, "grad_norm": 10.009471893310547, "learning_rate": 7.989046756236477e-06, "loss": 1.0776, "step": 10763 }, { "epoch": 1.73, "grad_norm": 10.073739051818848, "learning_rate": 7.987344299791893e-06, "loss": 0.9466, "step": 10764 }, { "epoch": 1.73, "grad_norm": 10.289148330688477, "learning_rate": 7.98564190414185e-06, "loss": 0.9014, "step": 10765 }, { "epoch": 1.73, "grad_norm": 10.23338508605957, "learning_rate": 7.983939569337773e-06, "loss": 0.9945, "step": 10766 }, { "epoch": 1.73, "grad_norm": 8.848124504089355, "learning_rate": 7.982237295431074e-06, "loss": 1.0098, "step": 10767 }, { "epoch": 1.73, "grad_norm": 11.812799453735352, "learning_rate": 7.980535082473177e-06, "loss": 0.925, "step": 10768 }, { "epoch": 1.73, "grad_norm": 12.001595497131348, "learning_rate": 7.9788329305155e-06, "loss": 1.1538, "step": 10769 }, { "epoch": 1.73, "grad_norm": 11.109489440917969, "learning_rate": 7.97713083960946e-06, "loss": 1.0905, "step": 10770 }, { "epoch": 1.73, "grad_norm": 11.421743392944336, "learning_rate": 7.975428809806462e-06, "loss": 0.974, "step": 10771 }, { "epoch": 1.73, "grad_norm": 9.13493824005127, "learning_rate": 7.973726841157926e-06, "loss": 0.9498, "step": 10772 }, { "epoch": 1.73, "grad_norm": 8.318462371826172, "learning_rate": 7.972024933715258e-06, "loss": 1.0729, "step": 10773 }, { "epoch": 1.73, "grad_norm": 10.18762493133545, "learning_rate": 7.970323087529871e-06, "loss": 1.0982, "step": 10774 }, { "epoch": 1.73, "grad_norm": 9.615864753723145, "learning_rate": 7.968621302653164e-06, "loss": 0.8788, "step": 10775 }, { "epoch": 1.73, "grad_norm": 9.939043998718262, "learning_rate": 7.966919579136547e-06, "loss": 0.9273, "step": 10776 }, { "epoch": 1.73, "grad_norm": 9.399320602416992, "learning_rate": 7.965217917031417e-06, "loss": 1.0723, "step": 10777 }, { "epoch": 1.74, "grad_norm": 9.868369102478027, "learning_rate": 7.96351631638918e-06, "loss": 0.7702, "step": 10778 }, { "epoch": 1.74, "grad_norm": 13.159271240234375, "learning_rate": 7.961814777261235e-06, "loss": 1.0824, "step": 10779 }, { "epoch": 1.74, "grad_norm": 9.858407974243164, "learning_rate": 7.960113299698974e-06, "loss": 0.9045, "step": 10780 }, { "epoch": 1.74, "grad_norm": 9.754734992980957, "learning_rate": 7.958411883753794e-06, "loss": 1.0559, "step": 10781 }, { "epoch": 1.74, "grad_norm": 12.312691688537598, "learning_rate": 7.95671052947709e-06, "loss": 1.2145, "step": 10782 }, { "epoch": 1.74, "grad_norm": 9.10234546661377, "learning_rate": 7.955009236920254e-06, "loss": 0.929, "step": 10783 }, { "epoch": 1.74, "grad_norm": 10.080877304077148, "learning_rate": 7.953308006134667e-06, "loss": 1.0519, "step": 10784 }, { "epoch": 1.74, "grad_norm": 9.623851776123047, "learning_rate": 7.95160683717173e-06, "loss": 0.965, "step": 10785 }, { "epoch": 1.74, "grad_norm": 10.131790161132812, "learning_rate": 7.94990573008282e-06, "loss": 1.0108, "step": 10786 }, { "epoch": 1.74, "grad_norm": 9.160386085510254, "learning_rate": 7.948204684919322e-06, "loss": 0.9184, "step": 10787 }, { "epoch": 1.74, "grad_norm": 10.716294288635254, "learning_rate": 7.94650370173262e-06, "loss": 1.0265, "step": 10788 }, { "epoch": 1.74, "grad_norm": 10.819839477539062, "learning_rate": 7.944802780574094e-06, "loss": 1.1346, "step": 10789 }, { "epoch": 1.74, "grad_norm": 9.530900955200195, "learning_rate": 7.94310192149512e-06, "loss": 0.9999, "step": 10790 }, { "epoch": 1.74, "grad_norm": 15.961904525756836, "learning_rate": 7.941401124547075e-06, "loss": 1.2506, "step": 10791 }, { "epoch": 1.74, "grad_norm": 9.54695987701416, "learning_rate": 7.939700389781335e-06, "loss": 0.9057, "step": 10792 }, { "epoch": 1.74, "grad_norm": 9.907915115356445, "learning_rate": 7.937999717249271e-06, "loss": 1.0638, "step": 10793 }, { "epoch": 1.74, "grad_norm": 12.316096305847168, "learning_rate": 7.936299107002256e-06, "loss": 0.941, "step": 10794 }, { "epoch": 1.74, "grad_norm": 9.398335456848145, "learning_rate": 7.934598559091656e-06, "loss": 0.8438, "step": 10795 }, { "epoch": 1.74, "grad_norm": 12.431930541992188, "learning_rate": 7.93289807356884e-06, "loss": 1.2248, "step": 10796 }, { "epoch": 1.74, "grad_norm": 9.01996898651123, "learning_rate": 7.93119765048517e-06, "loss": 0.8313, "step": 10797 }, { "epoch": 1.74, "grad_norm": 9.491063117980957, "learning_rate": 7.929497289892014e-06, "loss": 1.1259, "step": 10798 }, { "epoch": 1.74, "grad_norm": 10.803705215454102, "learning_rate": 7.927796991840732e-06, "loss": 1.1045, "step": 10799 }, { "epoch": 1.74, "grad_norm": 8.1168851852417, "learning_rate": 7.92609675638268e-06, "loss": 0.932, "step": 10800 }, { "epoch": 1.74, "grad_norm": 10.62687873840332, "learning_rate": 7.924396583569218e-06, "loss": 1.3036, "step": 10801 }, { "epoch": 1.74, "grad_norm": 9.912290573120117, "learning_rate": 7.922696473451704e-06, "loss": 0.9421, "step": 10802 }, { "epoch": 1.74, "grad_norm": 10.32163143157959, "learning_rate": 7.920996426081486e-06, "loss": 1.1164, "step": 10803 }, { "epoch": 1.74, "grad_norm": 10.781240463256836, "learning_rate": 7.919296441509915e-06, "loss": 1.1352, "step": 10804 }, { "epoch": 1.74, "grad_norm": 10.180005073547363, "learning_rate": 7.91759651978835e-06, "loss": 0.6842, "step": 10805 }, { "epoch": 1.74, "grad_norm": 9.53583812713623, "learning_rate": 7.915896660968134e-06, "loss": 0.8628, "step": 10806 }, { "epoch": 1.74, "grad_norm": 9.29875659942627, "learning_rate": 7.914196865100613e-06, "loss": 1.1797, "step": 10807 }, { "epoch": 1.74, "grad_norm": 11.132226943969727, "learning_rate": 7.912497132237131e-06, "loss": 1.3109, "step": 10808 }, { "epoch": 1.74, "grad_norm": 10.610543251037598, "learning_rate": 7.910797462429032e-06, "loss": 1.0471, "step": 10809 }, { "epoch": 1.74, "grad_norm": 9.102270126342773, "learning_rate": 7.909097855727656e-06, "loss": 0.9689, "step": 10810 }, { "epoch": 1.74, "grad_norm": 13.386378288269043, "learning_rate": 7.907398312184338e-06, "loss": 1.1477, "step": 10811 }, { "epoch": 1.74, "grad_norm": 9.100871086120605, "learning_rate": 7.90569883185042e-06, "loss": 1.0578, "step": 10812 }, { "epoch": 1.74, "grad_norm": 11.051337242126465, "learning_rate": 7.903999414777233e-06, "loss": 0.9786, "step": 10813 }, { "epoch": 1.74, "grad_norm": 9.15422534942627, "learning_rate": 7.902300061016112e-06, "loss": 1.1363, "step": 10814 }, { "epoch": 1.74, "grad_norm": 9.654504776000977, "learning_rate": 7.900600770618385e-06, "loss": 1.0652, "step": 10815 }, { "epoch": 1.74, "grad_norm": 10.474180221557617, "learning_rate": 7.898901543635384e-06, "loss": 0.969, "step": 10816 }, { "epoch": 1.74, "grad_norm": 10.021347045898438, "learning_rate": 7.897202380118432e-06, "loss": 1.0384, "step": 10817 }, { "epoch": 1.74, "grad_norm": 10.286195755004883, "learning_rate": 7.895503280118862e-06, "loss": 1.0304, "step": 10818 }, { "epoch": 1.74, "grad_norm": 9.198371887207031, "learning_rate": 7.89380424368799e-06, "loss": 0.9037, "step": 10819 }, { "epoch": 1.74, "grad_norm": 8.659295082092285, "learning_rate": 7.892105270877139e-06, "loss": 0.8929, "step": 10820 }, { "epoch": 1.74, "grad_norm": 9.78312873840332, "learning_rate": 7.890406361737629e-06, "loss": 0.9076, "step": 10821 }, { "epoch": 1.74, "grad_norm": 11.874560356140137, "learning_rate": 7.88870751632078e-06, "loss": 0.9994, "step": 10822 }, { "epoch": 1.74, "grad_norm": 11.729401588439941, "learning_rate": 7.887008734677904e-06, "loss": 1.0973, "step": 10823 }, { "epoch": 1.74, "grad_norm": 8.90114688873291, "learning_rate": 7.885310016860314e-06, "loss": 1.0527, "step": 10824 }, { "epoch": 1.74, "grad_norm": 8.241116523742676, "learning_rate": 7.883611362919322e-06, "loss": 1.0521, "step": 10825 }, { "epoch": 1.74, "grad_norm": 9.990718841552734, "learning_rate": 7.881912772906239e-06, "loss": 0.9721, "step": 10826 }, { "epoch": 1.74, "grad_norm": 10.2610502243042, "learning_rate": 7.880214246872375e-06, "loss": 0.968, "step": 10827 }, { "epoch": 1.74, "grad_norm": 8.714742660522461, "learning_rate": 7.878515784869037e-06, "loss": 0.9854, "step": 10828 }, { "epoch": 1.74, "grad_norm": 9.09510326385498, "learning_rate": 7.876817386947522e-06, "loss": 0.836, "step": 10829 }, { "epoch": 1.74, "grad_norm": 8.868566513061523, "learning_rate": 7.875119053159136e-06, "loss": 0.6934, "step": 10830 }, { "epoch": 1.74, "grad_norm": 10.155335426330566, "learning_rate": 7.87342078355518e-06, "loss": 1.017, "step": 10831 }, { "epoch": 1.74, "grad_norm": 10.58687973022461, "learning_rate": 7.87172257818695e-06, "loss": 0.9968, "step": 10832 }, { "epoch": 1.74, "grad_norm": 7.876615524291992, "learning_rate": 7.870024437105745e-06, "loss": 0.8392, "step": 10833 }, { "epoch": 1.74, "grad_norm": 12.472697257995605, "learning_rate": 7.868326360362856e-06, "loss": 1.0785, "step": 10834 }, { "epoch": 1.74, "grad_norm": 8.455198287963867, "learning_rate": 7.866628348009577e-06, "loss": 0.7342, "step": 10835 }, { "epoch": 1.74, "grad_norm": 11.059806823730469, "learning_rate": 7.864930400097197e-06, "loss": 1.2117, "step": 10836 }, { "epoch": 1.74, "grad_norm": 9.801218032836914, "learning_rate": 7.86323251667701e-06, "loss": 1.0962, "step": 10837 }, { "epoch": 1.74, "grad_norm": 10.429189682006836, "learning_rate": 7.861534697800295e-06, "loss": 1.3382, "step": 10838 }, { "epoch": 1.74, "grad_norm": 10.64383316040039, "learning_rate": 7.859836943518339e-06, "loss": 1.1878, "step": 10839 }, { "epoch": 1.75, "grad_norm": 9.912885665893555, "learning_rate": 7.858139253882426e-06, "loss": 1.0104, "step": 10840 }, { "epoch": 1.75, "grad_norm": 8.735389709472656, "learning_rate": 7.856441628943837e-06, "loss": 0.9287, "step": 10841 }, { "epoch": 1.75, "grad_norm": 13.14442253112793, "learning_rate": 7.854744068753849e-06, "loss": 1.3094, "step": 10842 }, { "epoch": 1.75, "grad_norm": 9.202324867248535, "learning_rate": 7.853046573363739e-06, "loss": 0.8767, "step": 10843 }, { "epoch": 1.75, "grad_norm": 11.817461013793945, "learning_rate": 7.85134914282478e-06, "loss": 1.0975, "step": 10844 }, { "epoch": 1.75, "grad_norm": 11.112704277038574, "learning_rate": 7.849651777188248e-06, "loss": 1.1112, "step": 10845 }, { "epoch": 1.75, "grad_norm": 8.763949394226074, "learning_rate": 7.847954476505412e-06, "loss": 1.0937, "step": 10846 }, { "epoch": 1.75, "grad_norm": 9.451986312866211, "learning_rate": 7.846257240827544e-06, "loss": 1.0779, "step": 10847 }, { "epoch": 1.75, "grad_norm": 9.392555236816406, "learning_rate": 7.844560070205909e-06, "loss": 1.4623, "step": 10848 }, { "epoch": 1.75, "grad_norm": 15.795612335205078, "learning_rate": 7.842862964691772e-06, "loss": 1.0306, "step": 10849 }, { "epoch": 1.75, "grad_norm": 10.088780403137207, "learning_rate": 7.841165924336392e-06, "loss": 1.2161, "step": 10850 }, { "epoch": 1.75, "grad_norm": 9.206900596618652, "learning_rate": 7.839468949191036e-06, "loss": 0.8884, "step": 10851 }, { "epoch": 1.75, "grad_norm": 13.445713996887207, "learning_rate": 7.837772039306962e-06, "loss": 1.2018, "step": 10852 }, { "epoch": 1.75, "grad_norm": 11.187013626098633, "learning_rate": 7.836075194735423e-06, "loss": 0.9234, "step": 10853 }, { "epoch": 1.75, "grad_norm": 9.25241756439209, "learning_rate": 7.834378415527676e-06, "loss": 0.9164, "step": 10854 }, { "epoch": 1.75, "grad_norm": 10.927952766418457, "learning_rate": 7.832681701734977e-06, "loss": 1.1158, "step": 10855 }, { "epoch": 1.75, "grad_norm": 9.291306495666504, "learning_rate": 7.830985053408574e-06, "loss": 0.7935, "step": 10856 }, { "epoch": 1.75, "grad_norm": 9.717034339904785, "learning_rate": 7.829288470599718e-06, "loss": 1.2227, "step": 10857 }, { "epoch": 1.75, "grad_norm": 9.067829132080078, "learning_rate": 7.827591953359656e-06, "loss": 1.2447, "step": 10858 }, { "epoch": 1.75, "grad_norm": 10.510499954223633, "learning_rate": 7.825895501739631e-06, "loss": 1.1979, "step": 10859 }, { "epoch": 1.75, "grad_norm": 10.648818016052246, "learning_rate": 7.824199115790889e-06, "loss": 0.9633, "step": 10860 }, { "epoch": 1.75, "grad_norm": 10.994062423706055, "learning_rate": 7.82250279556467e-06, "loss": 0.9299, "step": 10861 }, { "epoch": 1.75, "grad_norm": 9.199800491333008, "learning_rate": 7.820806541112214e-06, "loss": 1.035, "step": 10862 }, { "epoch": 1.75, "grad_norm": 9.467835426330566, "learning_rate": 7.819110352484755e-06, "loss": 0.9633, "step": 10863 }, { "epoch": 1.75, "grad_norm": 10.26461124420166, "learning_rate": 7.817414229733532e-06, "loss": 1.0393, "step": 10864 }, { "epoch": 1.75, "grad_norm": 8.69938850402832, "learning_rate": 7.815718172909777e-06, "loss": 1.038, "step": 10865 }, { "epoch": 1.75, "grad_norm": 9.562411308288574, "learning_rate": 7.814022182064721e-06, "loss": 1.1327, "step": 10866 }, { "epoch": 1.75, "grad_norm": 8.65554428100586, "learning_rate": 7.812326257249597e-06, "loss": 0.916, "step": 10867 }, { "epoch": 1.75, "grad_norm": 12.584653854370117, "learning_rate": 7.810630398515627e-06, "loss": 1.088, "step": 10868 }, { "epoch": 1.75, "grad_norm": 9.541769027709961, "learning_rate": 7.80893460591404e-06, "loss": 1.185, "step": 10869 }, { "epoch": 1.75, "grad_norm": 9.823993682861328, "learning_rate": 7.807238879496055e-06, "loss": 1.0634, "step": 10870 }, { "epoch": 1.75, "grad_norm": 9.729758262634277, "learning_rate": 7.8055432193129e-06, "loss": 1.0173, "step": 10871 }, { "epoch": 1.75, "grad_norm": 9.64382553100586, "learning_rate": 7.803847625415789e-06, "loss": 0.9849, "step": 10872 }, { "epoch": 1.75, "grad_norm": 9.357784271240234, "learning_rate": 7.802152097855941e-06, "loss": 0.8978, "step": 10873 }, { "epoch": 1.75, "grad_norm": 9.948729515075684, "learning_rate": 7.80045663668457e-06, "loss": 1.0256, "step": 10874 }, { "epoch": 1.75, "grad_norm": 11.006592750549316, "learning_rate": 7.79876124195289e-06, "loss": 1.2328, "step": 10875 }, { "epoch": 1.75, "grad_norm": 11.019736289978027, "learning_rate": 7.797065913712116e-06, "loss": 1.1575, "step": 10876 }, { "epoch": 1.75, "grad_norm": 9.09212589263916, "learning_rate": 7.795370652013453e-06, "loss": 0.8433, "step": 10877 }, { "epoch": 1.75, "grad_norm": 8.301081657409668, "learning_rate": 7.793675456908108e-06, "loss": 0.8, "step": 10878 }, { "epoch": 1.75, "grad_norm": 8.509711265563965, "learning_rate": 7.79198032844729e-06, "loss": 0.8396, "step": 10879 }, { "epoch": 1.75, "grad_norm": 9.047666549682617, "learning_rate": 7.790285266682199e-06, "loss": 1.0283, "step": 10880 }, { "epoch": 1.75, "grad_norm": 9.184345245361328, "learning_rate": 7.788590271664037e-06, "loss": 1.0792, "step": 10881 }, { "epoch": 1.75, "grad_norm": 10.395829200744629, "learning_rate": 7.786895343444002e-06, "loss": 0.926, "step": 10882 }, { "epoch": 1.75, "grad_norm": 9.496036529541016, "learning_rate": 7.785200482073292e-06, "loss": 1.004, "step": 10883 }, { "epoch": 1.75, "grad_norm": 14.524200439453125, "learning_rate": 7.783505687603106e-06, "loss": 1.1355, "step": 10884 }, { "epoch": 1.75, "grad_norm": 10.478168487548828, "learning_rate": 7.78181096008463e-06, "loss": 1.0313, "step": 10885 }, { "epoch": 1.75, "grad_norm": 9.458756446838379, "learning_rate": 7.780116299569062e-06, "loss": 0.8627, "step": 10886 }, { "epoch": 1.75, "grad_norm": 10.663823127746582, "learning_rate": 7.77842170610759e-06, "loss": 1.3908, "step": 10887 }, { "epoch": 1.75, "grad_norm": 10.60153865814209, "learning_rate": 7.776727179751398e-06, "loss": 1.1511, "step": 10888 }, { "epoch": 1.75, "grad_norm": 9.34366226196289, "learning_rate": 7.775032720551672e-06, "loss": 1.0025, "step": 10889 }, { "epoch": 1.75, "grad_norm": 9.739483833312988, "learning_rate": 7.773338328559596e-06, "loss": 0.8586, "step": 10890 }, { "epoch": 1.75, "grad_norm": 8.812017440795898, "learning_rate": 7.771644003826353e-06, "loss": 0.9182, "step": 10891 }, { "epoch": 1.75, "grad_norm": 10.266109466552734, "learning_rate": 7.769949746403116e-06, "loss": 1.0046, "step": 10892 }, { "epoch": 1.75, "grad_norm": 11.705665588378906, "learning_rate": 7.768255556341067e-06, "loss": 1.1223, "step": 10893 }, { "epoch": 1.75, "grad_norm": 11.985193252563477, "learning_rate": 7.76656143369138e-06, "loss": 1.0716, "step": 10894 }, { "epoch": 1.75, "grad_norm": 8.712373733520508, "learning_rate": 7.764867378505225e-06, "loss": 1.0483, "step": 10895 }, { "epoch": 1.75, "grad_norm": 9.084407806396484, "learning_rate": 7.76317339083378e-06, "loss": 0.8724, "step": 10896 }, { "epoch": 1.75, "grad_norm": 11.529627799987793, "learning_rate": 7.761479470728206e-06, "loss": 1.1116, "step": 10897 }, { "epoch": 1.75, "grad_norm": 8.829850196838379, "learning_rate": 7.759785618239672e-06, "loss": 0.8644, "step": 10898 }, { "epoch": 1.75, "grad_norm": 8.587772369384766, "learning_rate": 7.758091833419345e-06, "loss": 0.8386, "step": 10899 }, { "epoch": 1.75, "grad_norm": 8.32234001159668, "learning_rate": 7.756398116318388e-06, "loss": 0.934, "step": 10900 }, { "epoch": 1.75, "grad_norm": 11.495996475219727, "learning_rate": 7.754704466987958e-06, "loss": 1.2419, "step": 10901 }, { "epoch": 1.75, "grad_norm": 12.412040710449219, "learning_rate": 7.753010885479215e-06, "loss": 1.105, "step": 10902 }, { "epoch": 1.76, "grad_norm": 12.529654502868652, "learning_rate": 7.751317371843315e-06, "loss": 1.2078, "step": 10903 }, { "epoch": 1.76, "grad_norm": 8.390175819396973, "learning_rate": 7.749623926131417e-06, "loss": 0.7804, "step": 10904 }, { "epoch": 1.76, "grad_norm": 9.647787094116211, "learning_rate": 7.747930548394668e-06, "loss": 0.8968, "step": 10905 }, { "epoch": 1.76, "grad_norm": 11.440832138061523, "learning_rate": 7.746237238684217e-06, "loss": 1.3016, "step": 10906 }, { "epoch": 1.76, "grad_norm": 8.217344284057617, "learning_rate": 7.744543997051218e-06, "loss": 1.133, "step": 10907 }, { "epoch": 1.76, "grad_norm": 10.118692398071289, "learning_rate": 7.742850823546816e-06, "loss": 1.0896, "step": 10908 }, { "epoch": 1.76, "grad_norm": 8.005043983459473, "learning_rate": 7.741157718222154e-06, "loss": 0.8606, "step": 10909 }, { "epoch": 1.76, "grad_norm": 11.26623821258545, "learning_rate": 7.739464681128374e-06, "loss": 1.0196, "step": 10910 }, { "epoch": 1.76, "grad_norm": 12.111655235290527, "learning_rate": 7.737771712316617e-06, "loss": 0.9535, "step": 10911 }, { "epoch": 1.76, "grad_norm": 9.994841575622559, "learning_rate": 7.73607881183802e-06, "loss": 1.2411, "step": 10912 }, { "epoch": 1.76, "grad_norm": 9.372267723083496, "learning_rate": 7.734385979743719e-06, "loss": 1.3266, "step": 10913 }, { "epoch": 1.76, "grad_norm": 9.530750274658203, "learning_rate": 7.73269321608485e-06, "loss": 0.9617, "step": 10914 }, { "epoch": 1.76, "grad_norm": 10.174567222595215, "learning_rate": 7.731000520912545e-06, "loss": 0.9852, "step": 10915 }, { "epoch": 1.76, "grad_norm": 9.056234359741211, "learning_rate": 7.729307894277926e-06, "loss": 1.2179, "step": 10916 }, { "epoch": 1.76, "grad_norm": 9.484989166259766, "learning_rate": 7.72761533623213e-06, "loss": 1.1015, "step": 10917 }, { "epoch": 1.76, "grad_norm": 11.278918266296387, "learning_rate": 7.72592284682628e-06, "loss": 0.9396, "step": 10918 }, { "epoch": 1.76, "grad_norm": 10.513127326965332, "learning_rate": 7.724230426111498e-06, "loss": 1.116, "step": 10919 }, { "epoch": 1.76, "grad_norm": 7.616156101226807, "learning_rate": 7.722538074138907e-06, "loss": 0.8095, "step": 10920 }, { "epoch": 1.76, "grad_norm": 8.342070579528809, "learning_rate": 7.720845790959624e-06, "loss": 1.0122, "step": 10921 }, { "epoch": 1.76, "grad_norm": 9.463356018066406, "learning_rate": 7.719153576624769e-06, "loss": 1.0262, "step": 10922 }, { "epoch": 1.76, "grad_norm": 8.195104598999023, "learning_rate": 7.717461431185456e-06, "loss": 0.8291, "step": 10923 }, { "epoch": 1.76, "grad_norm": 9.247546195983887, "learning_rate": 7.715769354692799e-06, "loss": 1.2286, "step": 10924 }, { "epoch": 1.76, "grad_norm": 9.031793594360352, "learning_rate": 7.714077347197907e-06, "loss": 1.1257, "step": 10925 }, { "epoch": 1.76, "grad_norm": 8.752458572387695, "learning_rate": 7.71238540875189e-06, "loss": 1.2589, "step": 10926 }, { "epoch": 1.76, "grad_norm": 8.578310012817383, "learning_rate": 7.710693539405858e-06, "loss": 1.0721, "step": 10927 }, { "epoch": 1.76, "grad_norm": 10.0105562210083, "learning_rate": 7.709001739210911e-06, "loss": 1.1424, "step": 10928 }, { "epoch": 1.76, "grad_norm": 11.33489990234375, "learning_rate": 7.707310008218155e-06, "loss": 1.0492, "step": 10929 }, { "epoch": 1.76, "grad_norm": 10.01025390625, "learning_rate": 7.705618346478692e-06, "loss": 0.836, "step": 10930 }, { "epoch": 1.76, "grad_norm": 7.318192958831787, "learning_rate": 7.703926754043614e-06, "loss": 0.9614, "step": 10931 }, { "epoch": 1.76, "grad_norm": 8.9723482131958, "learning_rate": 7.702235230964024e-06, "loss": 1.0816, "step": 10932 }, { "epoch": 1.76, "grad_norm": 9.134045600891113, "learning_rate": 7.700543777291014e-06, "loss": 1.0176, "step": 10933 }, { "epoch": 1.76, "grad_norm": 9.195708274841309, "learning_rate": 7.698852393075675e-06, "loss": 1.1575, "step": 10934 }, { "epoch": 1.76, "grad_norm": 8.832854270935059, "learning_rate": 7.697161078369102e-06, "loss": 0.9647, "step": 10935 }, { "epoch": 1.76, "grad_norm": 11.03170108795166, "learning_rate": 7.695469833222376e-06, "loss": 1.0536, "step": 10936 }, { "epoch": 1.76, "grad_norm": 9.69261646270752, "learning_rate": 7.693778657686588e-06, "loss": 0.9479, "step": 10937 }, { "epoch": 1.76, "grad_norm": 10.539185523986816, "learning_rate": 7.692087551812819e-06, "loss": 1.0007, "step": 10938 }, { "epoch": 1.76, "grad_norm": 11.04742431640625, "learning_rate": 7.690396515652155e-06, "loss": 1.1445, "step": 10939 }, { "epoch": 1.76, "grad_norm": 9.968602180480957, "learning_rate": 7.68870554925567e-06, "loss": 1.0494, "step": 10940 }, { "epoch": 1.76, "grad_norm": 11.608884811401367, "learning_rate": 7.687014652674445e-06, "loss": 1.3164, "step": 10941 }, { "epoch": 1.76, "grad_norm": 9.841617584228516, "learning_rate": 7.685323825959554e-06, "loss": 1.0615, "step": 10942 }, { "epoch": 1.76, "grad_norm": 11.807960510253906, "learning_rate": 7.683633069162074e-06, "loss": 1.0772, "step": 10943 }, { "epoch": 1.76, "grad_norm": 10.374061584472656, "learning_rate": 7.681942382333073e-06, "loss": 1.0348, "step": 10944 }, { "epoch": 1.76, "grad_norm": 9.524828910827637, "learning_rate": 7.680251765523617e-06, "loss": 1.2655, "step": 10945 }, { "epoch": 1.76, "grad_norm": 8.860960006713867, "learning_rate": 7.678561218784775e-06, "loss": 0.8663, "step": 10946 }, { "epoch": 1.76, "grad_norm": 7.5890421867370605, "learning_rate": 7.676870742167618e-06, "loss": 0.8816, "step": 10947 }, { "epoch": 1.76, "grad_norm": 13.397626876831055, "learning_rate": 7.6751803357232e-06, "loss": 1.009, "step": 10948 }, { "epoch": 1.76, "grad_norm": 8.587095260620117, "learning_rate": 7.673489999502588e-06, "loss": 0.7655, "step": 10949 }, { "epoch": 1.76, "grad_norm": 9.92002010345459, "learning_rate": 7.67179973355684e-06, "loss": 1.0974, "step": 10950 }, { "epoch": 1.76, "grad_norm": 8.613136291503906, "learning_rate": 7.670109537937009e-06, "loss": 0.931, "step": 10951 }, { "epoch": 1.76, "grad_norm": 9.61013126373291, "learning_rate": 7.66841941269415e-06, "loss": 1.1693, "step": 10952 }, { "epoch": 1.76, "grad_norm": 10.129918098449707, "learning_rate": 7.666729357879316e-06, "loss": 1.2564, "step": 10953 }, { "epoch": 1.76, "grad_norm": 10.864644050598145, "learning_rate": 7.66503937354356e-06, "loss": 1.0617, "step": 10954 }, { "epoch": 1.76, "grad_norm": 8.562555313110352, "learning_rate": 7.663349459737926e-06, "loss": 0.8405, "step": 10955 }, { "epoch": 1.76, "grad_norm": 10.529007911682129, "learning_rate": 7.661659616513459e-06, "loss": 0.8625, "step": 10956 }, { "epoch": 1.76, "grad_norm": 8.927618026733398, "learning_rate": 7.659969843921204e-06, "loss": 1.1386, "step": 10957 }, { "epoch": 1.76, "grad_norm": 9.488431930541992, "learning_rate": 7.658280142012205e-06, "loss": 0.8797, "step": 10958 }, { "epoch": 1.76, "grad_norm": 9.707995414733887, "learning_rate": 7.656590510837496e-06, "loss": 0.7546, "step": 10959 }, { "epoch": 1.76, "grad_norm": 11.819957733154297, "learning_rate": 7.654900950448118e-06, "loss": 0.9698, "step": 10960 }, { "epoch": 1.76, "grad_norm": 8.485337257385254, "learning_rate": 7.653211460895107e-06, "loss": 0.9451, "step": 10961 }, { "epoch": 1.76, "grad_norm": 9.879924774169922, "learning_rate": 7.651522042229493e-06, "loss": 1.1206, "step": 10962 }, { "epoch": 1.76, "grad_norm": 10.95543384552002, "learning_rate": 7.64983269450231e-06, "loss": 1.4121, "step": 10963 }, { "epoch": 1.76, "grad_norm": 11.548290252685547, "learning_rate": 7.648143417764584e-06, "loss": 1.0643, "step": 10964 }, { "epoch": 1.77, "grad_norm": 9.207322120666504, "learning_rate": 7.646454212067342e-06, "loss": 0.9566, "step": 10965 }, { "epoch": 1.77, "grad_norm": 13.841961860656738, "learning_rate": 7.644765077461606e-06, "loss": 1.2186, "step": 10966 }, { "epoch": 1.77, "grad_norm": 12.202719688415527, "learning_rate": 7.643076013998403e-06, "loss": 1.0745, "step": 10967 }, { "epoch": 1.77, "grad_norm": 9.413201332092285, "learning_rate": 7.64138702172875e-06, "loss": 0.9893, "step": 10968 }, { "epoch": 1.77, "grad_norm": 10.993517875671387, "learning_rate": 7.639698100703671e-06, "loss": 0.8696, "step": 10969 }, { "epoch": 1.77, "grad_norm": 9.448067665100098, "learning_rate": 7.638009250974172e-06, "loss": 1.2283, "step": 10970 }, { "epoch": 1.77, "grad_norm": 9.536093711853027, "learning_rate": 7.636320472591272e-06, "loss": 1.1157, "step": 10971 }, { "epoch": 1.77, "grad_norm": 9.8342866897583, "learning_rate": 7.634631765605982e-06, "loss": 1.2228, "step": 10972 }, { "epoch": 1.77, "grad_norm": 7.615771770477295, "learning_rate": 7.63294313006931e-06, "loss": 0.6574, "step": 10973 }, { "epoch": 1.77, "grad_norm": 9.903133392333984, "learning_rate": 7.631254566032268e-06, "loss": 0.9012, "step": 10974 }, { "epoch": 1.77, "grad_norm": 11.397275924682617, "learning_rate": 7.629566073545854e-06, "loss": 1.0226, "step": 10975 }, { "epoch": 1.77, "grad_norm": 11.037900924682617, "learning_rate": 7.627877652661075e-06, "loss": 1.1131, "step": 10976 }, { "epoch": 1.77, "grad_norm": 8.553816795349121, "learning_rate": 7.626189303428929e-06, "loss": 0.8921, "step": 10977 }, { "epoch": 1.77, "grad_norm": 9.658330917358398, "learning_rate": 7.624501025900419e-06, "loss": 0.9686, "step": 10978 }, { "epoch": 1.77, "grad_norm": 9.239699363708496, "learning_rate": 7.622812820126536e-06, "loss": 1.0246, "step": 10979 }, { "epoch": 1.77, "grad_norm": 7.936397075653076, "learning_rate": 7.621124686158278e-06, "loss": 0.928, "step": 10980 }, { "epoch": 1.77, "grad_norm": 8.83549690246582, "learning_rate": 7.619436624046634e-06, "loss": 1.1013, "step": 10981 }, { "epoch": 1.77, "grad_norm": 11.64378547668457, "learning_rate": 7.617748633842598e-06, "loss": 0.9961, "step": 10982 }, { "epoch": 1.77, "grad_norm": 9.084144592285156, "learning_rate": 7.616060715597154e-06, "loss": 0.8396, "step": 10983 }, { "epoch": 1.77, "grad_norm": 7.997262001037598, "learning_rate": 7.6143728693612876e-06, "loss": 0.7606, "step": 10984 }, { "epoch": 1.77, "grad_norm": 10.661142349243164, "learning_rate": 7.612685095185983e-06, "loss": 1.2409, "step": 10985 }, { "epoch": 1.77, "grad_norm": 10.141115188598633, "learning_rate": 7.610997393122223e-06, "loss": 1.0099, "step": 10986 }, { "epoch": 1.77, "grad_norm": 9.636256217956543, "learning_rate": 7.60930976322098e-06, "loss": 1.0381, "step": 10987 }, { "epoch": 1.77, "grad_norm": 10.883467674255371, "learning_rate": 7.607622205533239e-06, "loss": 1.036, "step": 10988 }, { "epoch": 1.77, "grad_norm": 11.190778732299805, "learning_rate": 7.6059347201099746e-06, "loss": 1.2778, "step": 10989 }, { "epoch": 1.77, "grad_norm": 9.79898738861084, "learning_rate": 7.604247307002153e-06, "loss": 1.1625, "step": 10990 }, { "epoch": 1.77, "grad_norm": 10.88470458984375, "learning_rate": 7.602559966260748e-06, "loss": 1.2536, "step": 10991 }, { "epoch": 1.77, "grad_norm": 8.853248596191406, "learning_rate": 7.600872697936725e-06, "loss": 1.0249, "step": 10992 }, { "epoch": 1.77, "grad_norm": 9.943136215209961, "learning_rate": 7.599185502081055e-06, "loss": 1.1589, "step": 10993 }, { "epoch": 1.77, "grad_norm": 10.091877937316895, "learning_rate": 7.5974983787446965e-06, "loss": 1.1156, "step": 10994 }, { "epoch": 1.77, "grad_norm": 9.118120193481445, "learning_rate": 7.595811327978613e-06, "loss": 1.1521, "step": 10995 }, { "epoch": 1.77, "grad_norm": 9.924443244934082, "learning_rate": 7.594124349833765e-06, "loss": 1.108, "step": 10996 }, { "epoch": 1.77, "grad_norm": 10.081510543823242, "learning_rate": 7.592437444361108e-06, "loss": 1.193, "step": 10997 }, { "epoch": 1.77, "grad_norm": 9.551275253295898, "learning_rate": 7.590750611611596e-06, "loss": 0.9823, "step": 10998 }, { "epoch": 1.77, "grad_norm": 8.873190879821777, "learning_rate": 7.589063851636183e-06, "loss": 1.1874, "step": 10999 }, { "epoch": 1.77, "grad_norm": 12.178361892700195, "learning_rate": 7.587377164485819e-06, "loss": 1.0255, "step": 11000 }, { "epoch": 1.77, "grad_norm": 11.497492790222168, "learning_rate": 7.585690550211453e-06, "loss": 1.1455, "step": 11001 }, { "epoch": 1.77, "grad_norm": 9.835134506225586, "learning_rate": 7.584004008864034e-06, "loss": 0.9819, "step": 11002 }, { "epoch": 1.77, "grad_norm": 10.244441032409668, "learning_rate": 7.582317540494499e-06, "loss": 0.9432, "step": 11003 }, { "epoch": 1.77, "grad_norm": 10.165095329284668, "learning_rate": 7.580631145153793e-06, "loss": 1.233, "step": 11004 }, { "epoch": 1.77, "grad_norm": 9.803842544555664, "learning_rate": 7.578944822892856e-06, "loss": 1.1312, "step": 11005 }, { "epoch": 1.77, "grad_norm": 10.11072826385498, "learning_rate": 7.5772585737626285e-06, "loss": 1.1911, "step": 11006 }, { "epoch": 1.77, "grad_norm": 9.0833101272583, "learning_rate": 7.575572397814034e-06, "loss": 0.9962, "step": 11007 }, { "epoch": 1.77, "grad_norm": 8.777718544006348, "learning_rate": 7.573886295098021e-06, "loss": 1.1221, "step": 11008 }, { "epoch": 1.77, "grad_norm": 8.476156234741211, "learning_rate": 7.57220026566551e-06, "loss": 1.1111, "step": 11009 }, { "epoch": 1.77, "grad_norm": 10.050824165344238, "learning_rate": 7.5705143095674315e-06, "loss": 0.9958, "step": 11010 }, { "epoch": 1.77, "grad_norm": 7.90841817855835, "learning_rate": 7.568828426854713e-06, "loss": 1.1454, "step": 11011 }, { "epoch": 1.77, "grad_norm": 10.229833602905273, "learning_rate": 7.567142617578278e-06, "loss": 1.0084, "step": 11012 }, { "epoch": 1.77, "grad_norm": 9.782305717468262, "learning_rate": 7.56545688178905e-06, "loss": 1.0501, "step": 11013 }, { "epoch": 1.77, "grad_norm": 11.348135948181152, "learning_rate": 7.563771219537943e-06, "loss": 1.1486, "step": 11014 }, { "epoch": 1.77, "grad_norm": 9.517054557800293, "learning_rate": 7.562085630875879e-06, "loss": 1.1463, "step": 11015 }, { "epoch": 1.77, "grad_norm": 9.611099243164062, "learning_rate": 7.560400115853772e-06, "loss": 1.1374, "step": 11016 }, { "epoch": 1.77, "grad_norm": 8.670116424560547, "learning_rate": 7.558714674522536e-06, "loss": 1.0055, "step": 11017 }, { "epoch": 1.77, "grad_norm": 10.833765029907227, "learning_rate": 7.557029306933079e-06, "loss": 0.8694, "step": 11018 }, { "epoch": 1.77, "grad_norm": 10.459366798400879, "learning_rate": 7.555344013136312e-06, "loss": 1.0179, "step": 11019 }, { "epoch": 1.77, "grad_norm": 9.143763542175293, "learning_rate": 7.553658793183139e-06, "loss": 0.8262, "step": 11020 }, { "epoch": 1.77, "grad_norm": 11.532112121582031, "learning_rate": 7.551973647124468e-06, "loss": 1.1811, "step": 11021 }, { "epoch": 1.77, "grad_norm": 9.031648635864258, "learning_rate": 7.550288575011195e-06, "loss": 1.1234, "step": 11022 }, { "epoch": 1.77, "grad_norm": 9.773283004760742, "learning_rate": 7.548603576894222e-06, "loss": 0.7473, "step": 11023 }, { "epoch": 1.77, "grad_norm": 11.012144088745117, "learning_rate": 7.546918652824448e-06, "loss": 1.0028, "step": 11024 }, { "epoch": 1.77, "grad_norm": 10.246960639953613, "learning_rate": 7.545233802852768e-06, "loss": 0.9639, "step": 11025 }, { "epoch": 1.77, "grad_norm": 9.928784370422363, "learning_rate": 7.543549027030071e-06, "loss": 1.1306, "step": 11026 }, { "epoch": 1.78, "grad_norm": 10.137900352478027, "learning_rate": 7.5418643254072465e-06, "loss": 0.8881, "step": 11027 }, { "epoch": 1.78, "grad_norm": 9.7407808303833, "learning_rate": 7.540179698035193e-06, "loss": 0.8362, "step": 11028 }, { "epoch": 1.78, "grad_norm": 9.428109169006348, "learning_rate": 7.538495144964787e-06, "loss": 0.7824, "step": 11029 }, { "epoch": 1.78, "grad_norm": 10.002202987670898, "learning_rate": 7.536810666246915e-06, "loss": 1.0151, "step": 11030 }, { "epoch": 1.78, "grad_norm": 12.466228485107422, "learning_rate": 7.535126261932459e-06, "loss": 1.1274, "step": 11031 }, { "epoch": 1.78, "grad_norm": 10.14093017578125, "learning_rate": 7.533441932072301e-06, "loss": 1.0659, "step": 11032 }, { "epoch": 1.78, "grad_norm": 8.728034973144531, "learning_rate": 7.531757676717313e-06, "loss": 1.0285, "step": 11033 }, { "epoch": 1.78, "grad_norm": 8.959463119506836, "learning_rate": 7.530073495918372e-06, "loss": 0.8762, "step": 11034 }, { "epoch": 1.78, "grad_norm": 11.756721496582031, "learning_rate": 7.5283893897263495e-06, "loss": 1.0316, "step": 11035 }, { "epoch": 1.78, "grad_norm": 13.033513069152832, "learning_rate": 7.526705358192121e-06, "loss": 1.1905, "step": 11036 }, { "epoch": 1.78, "grad_norm": 11.66870403289795, "learning_rate": 7.525021401366547e-06, "loss": 1.0966, "step": 11037 }, { "epoch": 1.78, "grad_norm": 8.633148193359375, "learning_rate": 7.5233375193004975e-06, "loss": 1.0477, "step": 11038 }, { "epoch": 1.78, "grad_norm": 8.108680725097656, "learning_rate": 7.521653712044835e-06, "loss": 0.987, "step": 11039 }, { "epoch": 1.78, "grad_norm": 9.798735618591309, "learning_rate": 7.519969979650422e-06, "loss": 0.862, "step": 11040 }, { "epoch": 1.78, "grad_norm": 8.54336929321289, "learning_rate": 7.518286322168118e-06, "loss": 1.148, "step": 11041 }, { "epoch": 1.78, "grad_norm": 11.55567455291748, "learning_rate": 7.516602739648777e-06, "loss": 1.1487, "step": 11042 }, { "epoch": 1.78, "grad_norm": 9.953097343444824, "learning_rate": 7.5149192321432554e-06, "loss": 1.1845, "step": 11043 }, { "epoch": 1.78, "grad_norm": 8.707038879394531, "learning_rate": 7.513235799702405e-06, "loss": 0.9482, "step": 11044 }, { "epoch": 1.78, "grad_norm": 9.719653129577637, "learning_rate": 7.511552442377078e-06, "loss": 1.107, "step": 11045 }, { "epoch": 1.78, "grad_norm": 8.624241828918457, "learning_rate": 7.5098691602181176e-06, "loss": 0.918, "step": 11046 }, { "epoch": 1.78, "grad_norm": 11.013026237487793, "learning_rate": 7.508185953276369e-06, "loss": 1.3333, "step": 11047 }, { "epoch": 1.78, "grad_norm": 12.898642539978027, "learning_rate": 7.506502821602682e-06, "loss": 0.6551, "step": 11048 }, { "epoch": 1.78, "grad_norm": 10.350264549255371, "learning_rate": 7.504819765247892e-06, "loss": 0.9521, "step": 11049 }, { "epoch": 1.78, "grad_norm": 10.714600563049316, "learning_rate": 7.503136784262839e-06, "loss": 0.9744, "step": 11050 }, { "epoch": 1.78, "grad_norm": 8.840672492980957, "learning_rate": 7.501453878698359e-06, "loss": 1.1983, "step": 11051 }, { "epoch": 1.78, "grad_norm": 10.324922561645508, "learning_rate": 7.4997710486052895e-06, "loss": 1.1934, "step": 11052 }, { "epoch": 1.78, "grad_norm": 10.255182266235352, "learning_rate": 7.498088294034458e-06, "loss": 0.9853, "step": 11053 }, { "epoch": 1.78, "grad_norm": 14.279082298278809, "learning_rate": 7.496405615036694e-06, "loss": 1.3028, "step": 11054 }, { "epoch": 1.78, "grad_norm": 8.854156494140625, "learning_rate": 7.4947230116628255e-06, "loss": 1.1489, "step": 11055 }, { "epoch": 1.78, "grad_norm": 8.329690933227539, "learning_rate": 7.493040483963683e-06, "loss": 0.9776, "step": 11056 }, { "epoch": 1.78, "grad_norm": 10.752214431762695, "learning_rate": 7.491358031990079e-06, "loss": 1.2127, "step": 11057 }, { "epoch": 1.78, "grad_norm": 11.225701332092285, "learning_rate": 7.4896756557928406e-06, "loss": 1.2871, "step": 11058 }, { "epoch": 1.78, "grad_norm": 12.653210639953613, "learning_rate": 7.487993355422784e-06, "loss": 1.4502, "step": 11059 }, { "epoch": 1.78, "grad_norm": 9.302322387695312, "learning_rate": 7.486311130930726e-06, "loss": 0.997, "step": 11060 }, { "epoch": 1.78, "grad_norm": 12.900890350341797, "learning_rate": 7.484628982367479e-06, "loss": 1.2066, "step": 11061 }, { "epoch": 1.78, "grad_norm": 10.283018112182617, "learning_rate": 7.482946909783853e-06, "loss": 1.3111, "step": 11062 }, { "epoch": 1.78, "grad_norm": 10.594795227050781, "learning_rate": 7.481264913230659e-06, "loss": 1.0215, "step": 11063 }, { "epoch": 1.78, "grad_norm": 8.169666290283203, "learning_rate": 7.479582992758706e-06, "loss": 0.9639, "step": 11064 }, { "epoch": 1.78, "grad_norm": 8.25244140625, "learning_rate": 7.477901148418792e-06, "loss": 0.8376, "step": 11065 }, { "epoch": 1.78, "grad_norm": 9.318009376525879, "learning_rate": 7.476219380261725e-06, "loss": 1.0856, "step": 11066 }, { "epoch": 1.78, "grad_norm": 12.608102798461914, "learning_rate": 7.4745376883382996e-06, "loss": 1.1157, "step": 11067 }, { "epoch": 1.78, "grad_norm": 9.312918663024902, "learning_rate": 7.472856072699313e-06, "loss": 1.204, "step": 11068 }, { "epoch": 1.78, "grad_norm": 10.342690467834473, "learning_rate": 7.4711745333955665e-06, "loss": 0.9806, "step": 11069 }, { "epoch": 1.78, "grad_norm": 11.089240074157715, "learning_rate": 7.469493070477849e-06, "loss": 1.1386, "step": 11070 }, { "epoch": 1.78, "grad_norm": 9.884279251098633, "learning_rate": 7.467811683996953e-06, "loss": 0.9092, "step": 11071 }, { "epoch": 1.78, "grad_norm": 8.715991973876953, "learning_rate": 7.466130374003662e-06, "loss": 1.0696, "step": 11072 }, { "epoch": 1.78, "grad_norm": 12.584758758544922, "learning_rate": 7.464449140548766e-06, "loss": 1.1938, "step": 11073 }, { "epoch": 1.78, "grad_norm": 10.326139450073242, "learning_rate": 7.462767983683048e-06, "loss": 1.113, "step": 11074 }, { "epoch": 1.78, "grad_norm": 9.574725151062012, "learning_rate": 7.461086903457291e-06, "loss": 0.8993, "step": 11075 }, { "epoch": 1.78, "grad_norm": 7.786775588989258, "learning_rate": 7.459405899922269e-06, "loss": 0.8149, "step": 11076 }, { "epoch": 1.78, "grad_norm": 7.9986796379089355, "learning_rate": 7.457724973128761e-06, "loss": 0.8331, "step": 11077 }, { "epoch": 1.78, "grad_norm": 8.609919548034668, "learning_rate": 7.4560441231275415e-06, "loss": 0.9892, "step": 11078 }, { "epoch": 1.78, "grad_norm": 9.427331924438477, "learning_rate": 7.454363349969383e-06, "loss": 1.2222, "step": 11079 }, { "epoch": 1.78, "grad_norm": 8.697035789489746, "learning_rate": 7.452682653705055e-06, "loss": 1.0565, "step": 11080 }, { "epoch": 1.78, "grad_norm": 11.245888710021973, "learning_rate": 7.451002034385324e-06, "loss": 0.8905, "step": 11081 }, { "epoch": 1.78, "grad_norm": 8.856905937194824, "learning_rate": 7.4493214920609555e-06, "loss": 1.1392, "step": 11082 }, { "epoch": 1.78, "grad_norm": 8.585033416748047, "learning_rate": 7.447641026782713e-06, "loss": 0.9135, "step": 11083 }, { "epoch": 1.78, "grad_norm": 10.460742950439453, "learning_rate": 7.445960638601355e-06, "loss": 1.1167, "step": 11084 }, { "epoch": 1.78, "grad_norm": 9.192001342773438, "learning_rate": 7.444280327567641e-06, "loss": 1.1846, "step": 11085 }, { "epoch": 1.78, "grad_norm": 8.782791137695312, "learning_rate": 7.442600093732326e-06, "loss": 1.0122, "step": 11086 }, { "epoch": 1.78, "grad_norm": 9.915477752685547, "learning_rate": 7.440919937146162e-06, "loss": 1.38, "step": 11087 }, { "epoch": 1.78, "grad_norm": 8.8304443359375, "learning_rate": 7.439239857859901e-06, "loss": 0.8592, "step": 11088 }, { "epoch": 1.79, "grad_norm": 10.41665267944336, "learning_rate": 7.4375598559242945e-06, "loss": 1.1098, "step": 11089 }, { "epoch": 1.79, "grad_norm": 8.799095153808594, "learning_rate": 7.435879931390086e-06, "loss": 0.8806, "step": 11090 }, { "epoch": 1.79, "grad_norm": 11.775016784667969, "learning_rate": 7.434200084308022e-06, "loss": 1.1057, "step": 11091 }, { "epoch": 1.79, "grad_norm": 9.20869255065918, "learning_rate": 7.432520314728841e-06, "loss": 0.9491, "step": 11092 }, { "epoch": 1.79, "grad_norm": 9.827202796936035, "learning_rate": 7.430840622703283e-06, "loss": 1.2527, "step": 11093 }, { "epoch": 1.79, "grad_norm": 8.58143424987793, "learning_rate": 7.4291610082820865e-06, "loss": 0.7919, "step": 11094 }, { "epoch": 1.79, "grad_norm": 9.998540878295898, "learning_rate": 7.427481471515987e-06, "loss": 1.3441, "step": 11095 }, { "epoch": 1.79, "grad_norm": 10.8495512008667, "learning_rate": 7.4258020124557116e-06, "loss": 0.9595, "step": 11096 }, { "epoch": 1.79, "grad_norm": 10.454127311706543, "learning_rate": 7.424122631151995e-06, "loss": 1.035, "step": 11097 }, { "epoch": 1.79, "grad_norm": 8.940568923950195, "learning_rate": 7.422443327655564e-06, "loss": 1.1033, "step": 11098 }, { "epoch": 1.79, "grad_norm": 10.357941627502441, "learning_rate": 7.420764102017146e-06, "loss": 0.8393, "step": 11099 }, { "epoch": 1.79, "grad_norm": 9.967439651489258, "learning_rate": 7.4190849542874585e-06, "loss": 1.004, "step": 11100 }, { "epoch": 1.79, "grad_norm": 10.309737205505371, "learning_rate": 7.4174058845172255e-06, "loss": 1.0664, "step": 11101 }, { "epoch": 1.79, "grad_norm": 10.38182544708252, "learning_rate": 7.415726892757163e-06, "loss": 0.9015, "step": 11102 }, { "epoch": 1.79, "grad_norm": 10.143177032470703, "learning_rate": 7.414047979057992e-06, "loss": 1.0656, "step": 11103 }, { "epoch": 1.79, "grad_norm": 9.422165870666504, "learning_rate": 7.41236914347042e-06, "loss": 1.0304, "step": 11104 }, { "epoch": 1.79, "grad_norm": 9.359963417053223, "learning_rate": 7.410690386045161e-06, "loss": 0.7541, "step": 11105 }, { "epoch": 1.79, "grad_norm": 10.08726978302002, "learning_rate": 7.409011706832923e-06, "loss": 1.1375, "step": 11106 }, { "epoch": 1.79, "grad_norm": 9.137347221374512, "learning_rate": 7.407333105884414e-06, "loss": 1.0377, "step": 11107 }, { "epoch": 1.79, "grad_norm": 7.555431842803955, "learning_rate": 7.405654583250332e-06, "loss": 0.6849, "step": 11108 }, { "epoch": 1.79, "grad_norm": 6.815184116363525, "learning_rate": 7.403976138981387e-06, "loss": 0.8799, "step": 11109 }, { "epoch": 1.79, "grad_norm": 9.410575866699219, "learning_rate": 7.402297773128276e-06, "loss": 0.9551, "step": 11110 }, { "epoch": 1.79, "grad_norm": 7.662389755249023, "learning_rate": 7.400619485741694e-06, "loss": 0.9343, "step": 11111 }, { "epoch": 1.79, "grad_norm": 10.141061782836914, "learning_rate": 7.398941276872335e-06, "loss": 1.1006, "step": 11112 }, { "epoch": 1.79, "grad_norm": 14.144503593444824, "learning_rate": 7.397263146570894e-06, "loss": 1.4694, "step": 11113 }, { "epoch": 1.79, "grad_norm": 11.111653327941895, "learning_rate": 7.39558509488806e-06, "loss": 1.1946, "step": 11114 }, { "epoch": 1.79, "grad_norm": 9.463920593261719, "learning_rate": 7.393907121874517e-06, "loss": 1.1461, "step": 11115 }, { "epoch": 1.79, "grad_norm": 10.519299507141113, "learning_rate": 7.392229227580955e-06, "loss": 1.0767, "step": 11116 }, { "epoch": 1.79, "grad_norm": 10.373526573181152, "learning_rate": 7.3905514120580535e-06, "loss": 1.0187, "step": 11117 }, { "epoch": 1.79, "grad_norm": 8.888680458068848, "learning_rate": 7.388873675356494e-06, "loss": 0.9025, "step": 11118 }, { "epoch": 1.79, "grad_norm": 9.64496898651123, "learning_rate": 7.387196017526955e-06, "loss": 1.0819, "step": 11119 }, { "epoch": 1.79, "grad_norm": 9.705327033996582, "learning_rate": 7.3855184386201095e-06, "loss": 0.9897, "step": 11120 }, { "epoch": 1.79, "grad_norm": 7.964367389678955, "learning_rate": 7.3838409386866324e-06, "loss": 0.9136, "step": 11121 }, { "epoch": 1.79, "grad_norm": 12.869272232055664, "learning_rate": 7.382163517777197e-06, "loss": 1.1772, "step": 11122 }, { "epoch": 1.79, "grad_norm": 9.244118690490723, "learning_rate": 7.380486175942468e-06, "loss": 0.7902, "step": 11123 }, { "epoch": 1.79, "grad_norm": 9.22569465637207, "learning_rate": 7.378808913233113e-06, "loss": 0.9321, "step": 11124 }, { "epoch": 1.79, "grad_norm": 12.461862564086914, "learning_rate": 7.3771317296997935e-06, "loss": 0.874, "step": 11125 }, { "epoch": 1.79, "grad_norm": 13.390774726867676, "learning_rate": 7.375454625393173e-06, "loss": 1.0155, "step": 11126 }, { "epoch": 1.79, "grad_norm": 11.506125450134277, "learning_rate": 7.373777600363911e-06, "loss": 1.1123, "step": 11127 }, { "epoch": 1.79, "grad_norm": 10.974295616149902, "learning_rate": 7.3721006546626594e-06, "loss": 1.0911, "step": 11128 }, { "epoch": 1.79, "grad_norm": 11.452479362487793, "learning_rate": 7.370423788340078e-06, "loss": 0.9566, "step": 11129 }, { "epoch": 1.79, "grad_norm": 11.421765327453613, "learning_rate": 7.368747001446817e-06, "loss": 1.2137, "step": 11130 }, { "epoch": 1.79, "grad_norm": 9.858844757080078, "learning_rate": 7.367070294033522e-06, "loss": 0.9588, "step": 11131 }, { "epoch": 1.79, "grad_norm": 8.263654708862305, "learning_rate": 7.365393666150845e-06, "loss": 0.8698, "step": 11132 }, { "epoch": 1.79, "grad_norm": 12.32444953918457, "learning_rate": 7.363717117849427e-06, "loss": 0.9157, "step": 11133 }, { "epoch": 1.79, "grad_norm": 10.89411449432373, "learning_rate": 7.3620406491799136e-06, "loss": 0.8694, "step": 11134 }, { "epoch": 1.79, "grad_norm": 9.994794845581055, "learning_rate": 7.360364260192938e-06, "loss": 1.0715, "step": 11135 }, { "epoch": 1.79, "grad_norm": 9.221176147460938, "learning_rate": 7.358687950939143e-06, "loss": 1.2584, "step": 11136 }, { "epoch": 1.79, "grad_norm": 11.873196601867676, "learning_rate": 7.35701172146916e-06, "loss": 0.9611, "step": 11137 }, { "epoch": 1.79, "grad_norm": 12.159513473510742, "learning_rate": 7.355335571833626e-06, "loss": 1.0472, "step": 11138 }, { "epoch": 1.79, "grad_norm": 10.234678268432617, "learning_rate": 7.353659502083166e-06, "loss": 0.9448, "step": 11139 }, { "epoch": 1.79, "grad_norm": 8.242931365966797, "learning_rate": 7.351983512268409e-06, "loss": 0.8738, "step": 11140 }, { "epoch": 1.79, "grad_norm": 10.459123611450195, "learning_rate": 7.350307602439981e-06, "loss": 1.0564, "step": 11141 }, { "epoch": 1.79, "grad_norm": 10.68353271484375, "learning_rate": 7.348631772648507e-06, "loss": 1.0715, "step": 11142 }, { "epoch": 1.79, "grad_norm": 10.645827293395996, "learning_rate": 7.346956022944603e-06, "loss": 0.8411, "step": 11143 }, { "epoch": 1.79, "grad_norm": 11.639969825744629, "learning_rate": 7.345280353378887e-06, "loss": 1.5345, "step": 11144 }, { "epoch": 1.79, "grad_norm": 10.471051216125488, "learning_rate": 7.343604764001979e-06, "loss": 1.3183, "step": 11145 }, { "epoch": 1.79, "grad_norm": 11.320557594299316, "learning_rate": 7.341929254864488e-06, "loss": 1.1496, "step": 11146 }, { "epoch": 1.79, "grad_norm": 8.416150093078613, "learning_rate": 7.340253826017028e-06, "loss": 1.0575, "step": 11147 }, { "epoch": 1.79, "grad_norm": 10.228596687316895, "learning_rate": 7.338578477510203e-06, "loss": 1.1087, "step": 11148 }, { "epoch": 1.79, "grad_norm": 9.883306503295898, "learning_rate": 7.336903209394619e-06, "loss": 1.2556, "step": 11149 }, { "epoch": 1.79, "grad_norm": 9.793478012084961, "learning_rate": 7.335228021720883e-06, "loss": 0.948, "step": 11150 }, { "epoch": 1.8, "grad_norm": 9.025318145751953, "learning_rate": 7.333552914539594e-06, "loss": 1.0903, "step": 11151 }, { "epoch": 1.8, "grad_norm": 10.990206718444824, "learning_rate": 7.331877887901351e-06, "loss": 1.0488, "step": 11152 }, { "epoch": 1.8, "grad_norm": 9.513139724731445, "learning_rate": 7.330202941856752e-06, "loss": 1.0595, "step": 11153 }, { "epoch": 1.8, "grad_norm": 9.575713157653809, "learning_rate": 7.328528076456387e-06, "loss": 1.1717, "step": 11154 }, { "epoch": 1.8, "grad_norm": 9.6473970413208, "learning_rate": 7.326853291750848e-06, "loss": 1.055, "step": 11155 }, { "epoch": 1.8, "grad_norm": 9.782004356384277, "learning_rate": 7.3251785877907266e-06, "loss": 0.8453, "step": 11156 }, { "epoch": 1.8, "grad_norm": 7.5721917152404785, "learning_rate": 7.323503964626605e-06, "loss": 1.0368, "step": 11157 }, { "epoch": 1.8, "grad_norm": 10.625822067260742, "learning_rate": 7.321829422309071e-06, "loss": 0.8869, "step": 11158 }, { "epoch": 1.8, "grad_norm": 9.129467010498047, "learning_rate": 7.320154960888703e-06, "loss": 1.15, "step": 11159 }, { "epoch": 1.8, "grad_norm": 9.977665901184082, "learning_rate": 7.318480580416081e-06, "loss": 0.9717, "step": 11160 }, { "epoch": 1.8, "grad_norm": 10.444822311401367, "learning_rate": 7.316806280941779e-06, "loss": 1.0759, "step": 11161 }, { "epoch": 1.8, "grad_norm": 8.937582969665527, "learning_rate": 7.315132062516379e-06, "loss": 1.0899, "step": 11162 }, { "epoch": 1.8, "grad_norm": 10.360182762145996, "learning_rate": 7.3134579251904444e-06, "loss": 1.1236, "step": 11163 }, { "epoch": 1.8, "grad_norm": 9.158037185668945, "learning_rate": 7.3117838690145485e-06, "loss": 1.0637, "step": 11164 }, { "epoch": 1.8, "grad_norm": 9.923482894897461, "learning_rate": 7.310109894039254e-06, "loss": 1.0664, "step": 11165 }, { "epoch": 1.8, "grad_norm": 8.248041152954102, "learning_rate": 7.308436000315133e-06, "loss": 0.9143, "step": 11166 }, { "epoch": 1.8, "grad_norm": 12.429424285888672, "learning_rate": 7.30676218789274e-06, "loss": 0.9883, "step": 11167 }, { "epoch": 1.8, "grad_norm": 9.604990005493164, "learning_rate": 7.305088456822637e-06, "loss": 1.2396, "step": 11168 }, { "epoch": 1.8, "grad_norm": 9.151468276977539, "learning_rate": 7.3034148071553775e-06, "loss": 1.0711, "step": 11169 }, { "epoch": 1.8, "grad_norm": 10.057896614074707, "learning_rate": 7.301741238941522e-06, "loss": 1.0732, "step": 11170 }, { "epoch": 1.8, "grad_norm": 9.614232063293457, "learning_rate": 7.300067752231621e-06, "loss": 1.006, "step": 11171 }, { "epoch": 1.8, "grad_norm": 8.847525596618652, "learning_rate": 7.298394347076222e-06, "loss": 1.0151, "step": 11172 }, { "epoch": 1.8, "grad_norm": 14.057356834411621, "learning_rate": 7.296721023525874e-06, "loss": 1.0045, "step": 11173 }, { "epoch": 1.8, "grad_norm": 12.005979537963867, "learning_rate": 7.295047781631122e-06, "loss": 1.3796, "step": 11174 }, { "epoch": 1.8, "grad_norm": 9.302236557006836, "learning_rate": 7.293374621442504e-06, "loss": 0.9625, "step": 11175 }, { "epoch": 1.8, "grad_norm": 9.706341743469238, "learning_rate": 7.291701543010563e-06, "loss": 1.1012, "step": 11176 }, { "epoch": 1.8, "grad_norm": 10.045055389404297, "learning_rate": 7.2900285463858376e-06, "loss": 1.1724, "step": 11177 }, { "epoch": 1.8, "grad_norm": 8.795933723449707, "learning_rate": 7.28835563161886e-06, "loss": 0.88, "step": 11178 }, { "epoch": 1.8, "grad_norm": 8.812190055847168, "learning_rate": 7.286682798760161e-06, "loss": 1.0848, "step": 11179 }, { "epoch": 1.8, "grad_norm": 8.649932861328125, "learning_rate": 7.285010047860274e-06, "loss": 0.9814, "step": 11180 }, { "epoch": 1.8, "grad_norm": 11.004129409790039, "learning_rate": 7.283337378969726e-06, "loss": 1.1202, "step": 11181 }, { "epoch": 1.8, "grad_norm": 8.472674369812012, "learning_rate": 7.281664792139041e-06, "loss": 1.0894, "step": 11182 }, { "epoch": 1.8, "grad_norm": 9.344942092895508, "learning_rate": 7.279992287418738e-06, "loss": 1.0426, "step": 11183 }, { "epoch": 1.8, "grad_norm": 13.72140884399414, "learning_rate": 7.278319864859342e-06, "loss": 0.9371, "step": 11184 }, { "epoch": 1.8, "grad_norm": 9.020346641540527, "learning_rate": 7.276647524511367e-06, "loss": 1.0303, "step": 11185 }, { "epoch": 1.8, "grad_norm": 7.714916229248047, "learning_rate": 7.274975266425332e-06, "loss": 0.948, "step": 11186 }, { "epoch": 1.8, "grad_norm": 9.49906063079834, "learning_rate": 7.273303090651746e-06, "loss": 1.0469, "step": 11187 }, { "epoch": 1.8, "grad_norm": 9.768245697021484, "learning_rate": 7.2716309972411194e-06, "loss": 1.1614, "step": 11188 }, { "epoch": 1.8, "grad_norm": 8.688207626342773, "learning_rate": 7.2699589862439564e-06, "loss": 1.1406, "step": 11189 }, { "epoch": 1.8, "grad_norm": 11.114315032958984, "learning_rate": 7.2682870577107685e-06, "loss": 1.0646, "step": 11190 }, { "epoch": 1.8, "grad_norm": 8.183496475219727, "learning_rate": 7.266615211692057e-06, "loss": 0.8978, "step": 11191 }, { "epoch": 1.8, "grad_norm": 10.50224781036377, "learning_rate": 7.264943448238319e-06, "loss": 1.157, "step": 11192 }, { "epoch": 1.8, "grad_norm": 13.869485855102539, "learning_rate": 7.263271767400053e-06, "loss": 1.2908, "step": 11193 }, { "epoch": 1.8, "grad_norm": 10.289290428161621, "learning_rate": 7.261600169227754e-06, "loss": 1.321, "step": 11194 }, { "epoch": 1.8, "grad_norm": 9.661539077758789, "learning_rate": 7.259928653771915e-06, "loss": 1.2231, "step": 11195 }, { "epoch": 1.8, "grad_norm": 10.501514434814453, "learning_rate": 7.258257221083025e-06, "loss": 1.1749, "step": 11196 }, { "epoch": 1.8, "grad_norm": 9.22153091430664, "learning_rate": 7.2565858712115755e-06, "loss": 1.1302, "step": 11197 }, { "epoch": 1.8, "grad_norm": 7.6269636154174805, "learning_rate": 7.2549146042080455e-06, "loss": 0.7897, "step": 11198 }, { "epoch": 1.8, "grad_norm": 9.671659469604492, "learning_rate": 7.2532434201229195e-06, "loss": 0.8846, "step": 11199 }, { "epoch": 1.8, "grad_norm": 10.312692642211914, "learning_rate": 7.25157231900668e-06, "loss": 1.0814, "step": 11200 }, { "epoch": 1.8, "grad_norm": 10.067712783813477, "learning_rate": 7.249901300909803e-06, "loss": 0.852, "step": 11201 }, { "epoch": 1.8, "grad_norm": 9.706750869750977, "learning_rate": 7.248230365882763e-06, "loss": 0.8267, "step": 11202 }, { "epoch": 1.8, "grad_norm": 8.62846565246582, "learning_rate": 7.246559513976031e-06, "loss": 1.122, "step": 11203 }, { "epoch": 1.8, "grad_norm": 11.116385459899902, "learning_rate": 7.24488874524008e-06, "loss": 1.1719, "step": 11204 }, { "epoch": 1.8, "grad_norm": 10.523747444152832, "learning_rate": 7.243218059725378e-06, "loss": 0.9973, "step": 11205 }, { "epoch": 1.8, "grad_norm": 8.631180763244629, "learning_rate": 7.241547457482385e-06, "loss": 0.901, "step": 11206 }, { "epoch": 1.8, "grad_norm": 9.674816131591797, "learning_rate": 7.2398769385615675e-06, "loss": 0.996, "step": 11207 }, { "epoch": 1.8, "grad_norm": 12.706466674804688, "learning_rate": 7.238206503013386e-06, "loss": 1.1486, "step": 11208 }, { "epoch": 1.8, "grad_norm": 8.652405738830566, "learning_rate": 7.2365361508882955e-06, "loss": 0.8434, "step": 11209 }, { "epoch": 1.8, "grad_norm": 12.169316291809082, "learning_rate": 7.234865882236745e-06, "loss": 1.0779, "step": 11210 }, { "epoch": 1.8, "grad_norm": 10.787803649902344, "learning_rate": 7.233195697109201e-06, "loss": 1.0316, "step": 11211 }, { "epoch": 1.8, "grad_norm": 9.44864559173584, "learning_rate": 7.231525595556106e-06, "loss": 0.8315, "step": 11212 }, { "epoch": 1.81, "grad_norm": 9.220970153808594, "learning_rate": 7.229855577627905e-06, "loss": 0.8309, "step": 11213 }, { "epoch": 1.81, "grad_norm": 12.527715682983398, "learning_rate": 7.228185643375045e-06, "loss": 0.9746, "step": 11214 }, { "epoch": 1.81, "grad_norm": 12.094051361083984, "learning_rate": 7.226515792847969e-06, "loss": 1.2208, "step": 11215 }, { "epoch": 1.81, "grad_norm": 13.707056045532227, "learning_rate": 7.224846026097118e-06, "loss": 0.8796, "step": 11216 }, { "epoch": 1.81, "grad_norm": 8.670120239257812, "learning_rate": 7.223176343172924e-06, "loss": 0.7388, "step": 11217 }, { "epoch": 1.81, "grad_norm": 8.908087730407715, "learning_rate": 7.221506744125826e-06, "loss": 0.9683, "step": 11218 }, { "epoch": 1.81, "grad_norm": 8.217893600463867, "learning_rate": 7.219837229006255e-06, "loss": 1.1656, "step": 11219 }, { "epoch": 1.81, "grad_norm": 8.420437812805176, "learning_rate": 7.218167797864642e-06, "loss": 0.9197, "step": 11220 }, { "epoch": 1.81, "grad_norm": 11.271729469299316, "learning_rate": 7.216498450751409e-06, "loss": 1.1593, "step": 11221 }, { "epoch": 1.81, "grad_norm": 13.282042503356934, "learning_rate": 7.2148291877169855e-06, "loss": 0.9603, "step": 11222 }, { "epoch": 1.81, "grad_norm": 9.347780227661133, "learning_rate": 7.213160008811793e-06, "loss": 0.9052, "step": 11223 }, { "epoch": 1.81, "grad_norm": 9.360759735107422, "learning_rate": 7.211490914086248e-06, "loss": 0.779, "step": 11224 }, { "epoch": 1.81, "grad_norm": 12.449838638305664, "learning_rate": 7.209821903590773e-06, "loss": 1.0429, "step": 11225 }, { "epoch": 1.81, "grad_norm": 8.89786148071289, "learning_rate": 7.208152977375775e-06, "loss": 0.9788, "step": 11226 }, { "epoch": 1.81, "grad_norm": 10.15019702911377, "learning_rate": 7.206484135491671e-06, "loss": 1.082, "step": 11227 }, { "epoch": 1.81, "grad_norm": 10.741724967956543, "learning_rate": 7.204815377988867e-06, "loss": 1.0301, "step": 11228 }, { "epoch": 1.81, "grad_norm": 9.820343017578125, "learning_rate": 7.203146704917776e-06, "loss": 1.1109, "step": 11229 }, { "epoch": 1.81, "grad_norm": 11.046553611755371, "learning_rate": 7.2014781163287905e-06, "loss": 1.09, "step": 11230 }, { "epoch": 1.81, "grad_norm": 9.361418724060059, "learning_rate": 7.199809612272325e-06, "loss": 1.0609, "step": 11231 }, { "epoch": 1.81, "grad_norm": 9.596717834472656, "learning_rate": 7.198141192798771e-06, "loss": 1.0615, "step": 11232 }, { "epoch": 1.81, "grad_norm": 8.800626754760742, "learning_rate": 7.196472857958526e-06, "loss": 1.0766, "step": 11233 }, { "epoch": 1.81, "grad_norm": 11.48789119720459, "learning_rate": 7.1948046078019865e-06, "loss": 0.8975, "step": 11234 }, { "epoch": 1.81, "grad_norm": 12.622903823852539, "learning_rate": 7.19313644237954e-06, "loss": 1.0063, "step": 11235 }, { "epoch": 1.81, "grad_norm": 10.879334449768066, "learning_rate": 7.19146836174158e-06, "loss": 1.2132, "step": 11236 }, { "epoch": 1.81, "grad_norm": 11.145344734191895, "learning_rate": 7.189800365938489e-06, "loss": 0.8917, "step": 11237 }, { "epoch": 1.81, "grad_norm": 9.756256103515625, "learning_rate": 7.188132455020652e-06, "loss": 1.0858, "step": 11238 }, { "epoch": 1.81, "grad_norm": 9.14054012298584, "learning_rate": 7.18646462903845e-06, "loss": 0.8433, "step": 11239 }, { "epoch": 1.81, "grad_norm": 9.747541427612305, "learning_rate": 7.184796888042263e-06, "loss": 0.9241, "step": 11240 }, { "epoch": 1.81, "grad_norm": 11.926947593688965, "learning_rate": 7.183129232082464e-06, "loss": 1.2857, "step": 11241 }, { "epoch": 1.81, "grad_norm": 10.572906494140625, "learning_rate": 7.181461661209427e-06, "loss": 1.1078, "step": 11242 }, { "epoch": 1.81, "grad_norm": 9.523693084716797, "learning_rate": 7.1797941754735245e-06, "loss": 1.0598, "step": 11243 }, { "epoch": 1.81, "grad_norm": 8.327829360961914, "learning_rate": 7.178126774925126e-06, "loss": 0.9217, "step": 11244 }, { "epoch": 1.81, "grad_norm": 7.973998546600342, "learning_rate": 7.176459459614593e-06, "loss": 0.8532, "step": 11245 }, { "epoch": 1.81, "grad_norm": 11.3278169631958, "learning_rate": 7.174792229592292e-06, "loss": 1.4194, "step": 11246 }, { "epoch": 1.81, "grad_norm": 9.801393508911133, "learning_rate": 7.173125084908583e-06, "loss": 1.1227, "step": 11247 }, { "epoch": 1.81, "grad_norm": 9.903380393981934, "learning_rate": 7.171458025613825e-06, "loss": 1.0319, "step": 11248 }, { "epoch": 1.81, "grad_norm": 9.528738021850586, "learning_rate": 7.169791051758369e-06, "loss": 1.0112, "step": 11249 }, { "epoch": 1.81, "grad_norm": 9.59813404083252, "learning_rate": 7.168124163392567e-06, "loss": 1.0438, "step": 11250 }, { "epoch": 1.81, "grad_norm": 9.860272407531738, "learning_rate": 7.16645736056678e-06, "loss": 1.2369, "step": 11251 }, { "epoch": 1.81, "grad_norm": 10.131597518920898, "learning_rate": 7.164790643331345e-06, "loss": 1.0511, "step": 11252 }, { "epoch": 1.81, "grad_norm": 18.408838272094727, "learning_rate": 7.163124011736613e-06, "loss": 1.0161, "step": 11253 }, { "epoch": 1.81, "grad_norm": 9.1908597946167, "learning_rate": 7.161457465832924e-06, "loss": 0.891, "step": 11254 }, { "epoch": 1.81, "grad_norm": 12.424495697021484, "learning_rate": 7.159791005670619e-06, "loss": 1.0508, "step": 11255 }, { "epoch": 1.81, "grad_norm": 11.77525806427002, "learning_rate": 7.158124631300033e-06, "loss": 1.2314, "step": 11256 }, { "epoch": 1.81, "grad_norm": 9.636482238769531, "learning_rate": 7.156458342771502e-06, "loss": 0.9736, "step": 11257 }, { "epoch": 1.81, "grad_norm": 10.65584945678711, "learning_rate": 7.154792140135358e-06, "loss": 1.2308, "step": 11258 }, { "epoch": 1.81, "grad_norm": 11.350427627563477, "learning_rate": 7.153126023441935e-06, "loss": 1.1875, "step": 11259 }, { "epoch": 1.81, "grad_norm": 10.986883163452148, "learning_rate": 7.151459992741552e-06, "loss": 1.1974, "step": 11260 }, { "epoch": 1.81, "grad_norm": 10.649548530578613, "learning_rate": 7.149794048084537e-06, "loss": 1.1913, "step": 11261 }, { "epoch": 1.81, "grad_norm": 10.077411651611328, "learning_rate": 7.1481281895212114e-06, "loss": 1.1932, "step": 11262 }, { "epoch": 1.81, "grad_norm": 8.025239944458008, "learning_rate": 7.146462417101896e-06, "loss": 0.9517, "step": 11263 }, { "epoch": 1.81, "grad_norm": 10.530876159667969, "learning_rate": 7.144796730876908e-06, "loss": 1.093, "step": 11264 }, { "epoch": 1.81, "grad_norm": 10.57749080657959, "learning_rate": 7.143131130896557e-06, "loss": 0.8693, "step": 11265 }, { "epoch": 1.81, "grad_norm": 10.780783653259277, "learning_rate": 7.141465617211159e-06, "loss": 1.1282, "step": 11266 }, { "epoch": 1.81, "grad_norm": 9.067227363586426, "learning_rate": 7.139800189871018e-06, "loss": 1.2184, "step": 11267 }, { "epoch": 1.81, "grad_norm": 9.796907424926758, "learning_rate": 7.138134848926445e-06, "loss": 1.1271, "step": 11268 }, { "epoch": 1.81, "grad_norm": 10.17958927154541, "learning_rate": 7.13646959442774e-06, "loss": 1.0237, "step": 11269 }, { "epoch": 1.81, "grad_norm": 9.15368938446045, "learning_rate": 7.134804426425202e-06, "loss": 1.15, "step": 11270 }, { "epoch": 1.81, "grad_norm": 11.013123512268066, "learning_rate": 7.133139344969135e-06, "loss": 1.1092, "step": 11271 }, { "epoch": 1.81, "grad_norm": 10.634209632873535, "learning_rate": 7.131474350109832e-06, "loss": 1.1577, "step": 11272 }, { "epoch": 1.81, "grad_norm": 9.535362243652344, "learning_rate": 7.129809441897586e-06, "loss": 1.0296, "step": 11273 }, { "epoch": 1.81, "grad_norm": 8.960949897766113, "learning_rate": 7.128144620382686e-06, "loss": 1.1742, "step": 11274 }, { "epoch": 1.82, "grad_norm": 10.463630676269531, "learning_rate": 7.126479885615426e-06, "loss": 1.064, "step": 11275 }, { "epoch": 1.82, "grad_norm": 10.23473834991455, "learning_rate": 7.124815237646083e-06, "loss": 1.2277, "step": 11276 }, { "epoch": 1.82, "grad_norm": 9.251821517944336, "learning_rate": 7.123150676524944e-06, "loss": 0.7001, "step": 11277 }, { "epoch": 1.82, "grad_norm": 9.894457817077637, "learning_rate": 7.121486202302288e-06, "loss": 0.7628, "step": 11278 }, { "epoch": 1.82, "grad_norm": 10.066452026367188, "learning_rate": 7.1198218150283934e-06, "loss": 1.3614, "step": 11279 }, { "epoch": 1.82, "grad_norm": 10.016611099243164, "learning_rate": 7.118157514753532e-06, "loss": 1.1414, "step": 11280 }, { "epoch": 1.82, "grad_norm": 11.12070369720459, "learning_rate": 7.116493301527978e-06, "loss": 1.0431, "step": 11281 }, { "epoch": 1.82, "grad_norm": 9.149873733520508, "learning_rate": 7.114829175402001e-06, "loss": 0.838, "step": 11282 }, { "epoch": 1.82, "grad_norm": 10.508620262145996, "learning_rate": 7.113165136425868e-06, "loss": 1.421, "step": 11283 }, { "epoch": 1.82, "grad_norm": 12.383711814880371, "learning_rate": 7.111501184649841e-06, "loss": 1.3287, "step": 11284 }, { "epoch": 1.82, "grad_norm": 8.974215507507324, "learning_rate": 7.109837320124184e-06, "loss": 1.3178, "step": 11285 }, { "epoch": 1.82, "grad_norm": 11.047325134277344, "learning_rate": 7.108173542899154e-06, "loss": 1.0792, "step": 11286 }, { "epoch": 1.82, "grad_norm": 10.951199531555176, "learning_rate": 7.106509853025011e-06, "loss": 1.1302, "step": 11287 }, { "epoch": 1.82, "grad_norm": 10.467276573181152, "learning_rate": 7.104846250552003e-06, "loss": 1.2347, "step": 11288 }, { "epoch": 1.82, "grad_norm": 8.879556655883789, "learning_rate": 7.103182735530385e-06, "loss": 1.0365, "step": 11289 }, { "epoch": 1.82, "grad_norm": 10.03795337677002, "learning_rate": 7.101519308010403e-06, "loss": 1.0293, "step": 11290 }, { "epoch": 1.82, "grad_norm": 8.222227096557617, "learning_rate": 7.099855968042303e-06, "loss": 0.8783, "step": 11291 }, { "epoch": 1.82, "grad_norm": 10.323409080505371, "learning_rate": 7.098192715676331e-06, "loss": 1.1489, "step": 11292 }, { "epoch": 1.82, "grad_norm": 9.966104507446289, "learning_rate": 7.0965295509627245e-06, "loss": 0.9245, "step": 11293 }, { "epoch": 1.82, "grad_norm": 9.33207893371582, "learning_rate": 7.094866473951725e-06, "loss": 1.0408, "step": 11294 }, { "epoch": 1.82, "grad_norm": 12.25447940826416, "learning_rate": 7.093203484693561e-06, "loss": 1.1111, "step": 11295 }, { "epoch": 1.82, "grad_norm": 8.191222190856934, "learning_rate": 7.091540583238471e-06, "loss": 0.7761, "step": 11296 }, { "epoch": 1.82, "grad_norm": 12.427823066711426, "learning_rate": 7.0898777696366814e-06, "loss": 1.2454, "step": 11297 }, { "epoch": 1.82, "grad_norm": 10.664142608642578, "learning_rate": 7.088215043938423e-06, "loss": 1.158, "step": 11298 }, { "epoch": 1.82, "grad_norm": 10.3229341506958, "learning_rate": 7.086552406193916e-06, "loss": 1.0141, "step": 11299 }, { "epoch": 1.82, "grad_norm": 9.138306617736816, "learning_rate": 7.084889856453384e-06, "loss": 1.1501, "step": 11300 }, { "epoch": 1.82, "grad_norm": 10.832728385925293, "learning_rate": 7.083227394767047e-06, "loss": 1.1595, "step": 11301 }, { "epoch": 1.82, "grad_norm": 9.341615676879883, "learning_rate": 7.081565021185119e-06, "loss": 0.9829, "step": 11302 }, { "epoch": 1.82, "grad_norm": 10.848153114318848, "learning_rate": 7.07990273575782e-06, "loss": 0.9428, "step": 11303 }, { "epoch": 1.82, "grad_norm": 10.327873229980469, "learning_rate": 7.0782405385353525e-06, "loss": 1.242, "step": 11304 }, { "epoch": 1.82, "grad_norm": 8.561949729919434, "learning_rate": 7.076578429567931e-06, "loss": 0.8233, "step": 11305 }, { "epoch": 1.82, "grad_norm": 9.646471977233887, "learning_rate": 7.074916408905759e-06, "loss": 1.0544, "step": 11306 }, { "epoch": 1.82, "grad_norm": 9.284799575805664, "learning_rate": 7.073254476599043e-06, "loss": 0.9239, "step": 11307 }, { "epoch": 1.82, "grad_norm": 11.799560546875, "learning_rate": 7.071592632697979e-06, "loss": 1.0835, "step": 11308 }, { "epoch": 1.82, "grad_norm": 9.998270988464355, "learning_rate": 7.069930877252767e-06, "loss": 1.1045, "step": 11309 }, { "epoch": 1.82, "grad_norm": 9.183110237121582, "learning_rate": 7.068269210313603e-06, "loss": 0.968, "step": 11310 }, { "epoch": 1.82, "grad_norm": 11.479042053222656, "learning_rate": 7.066607631930675e-06, "loss": 1.1372, "step": 11311 }, { "epoch": 1.82, "grad_norm": 10.384650230407715, "learning_rate": 7.064946142154181e-06, "loss": 1.1016, "step": 11312 }, { "epoch": 1.82, "grad_norm": 9.65196704864502, "learning_rate": 7.063284741034302e-06, "loss": 0.6811, "step": 11313 }, { "epoch": 1.82, "grad_norm": 9.178101539611816, "learning_rate": 7.061623428621226e-06, "loss": 0.9636, "step": 11314 }, { "epoch": 1.82, "grad_norm": 9.662672996520996, "learning_rate": 7.059962204965132e-06, "loss": 0.908, "step": 11315 }, { "epoch": 1.82, "grad_norm": 11.983469009399414, "learning_rate": 7.058301070116198e-06, "loss": 1.2611, "step": 11316 }, { "epoch": 1.82, "grad_norm": 10.20732593536377, "learning_rate": 7.056640024124606e-06, "loss": 1.1556, "step": 11317 }, { "epoch": 1.82, "grad_norm": 8.933221817016602, "learning_rate": 7.054979067040526e-06, "loss": 1.1053, "step": 11318 }, { "epoch": 1.82, "grad_norm": 9.77493667602539, "learning_rate": 7.053318198914127e-06, "loss": 1.0121, "step": 11319 }, { "epoch": 1.82, "grad_norm": 10.542869567871094, "learning_rate": 7.051657419795581e-06, "loss": 1.1641, "step": 11320 }, { "epoch": 1.82, "grad_norm": 11.21940803527832, "learning_rate": 7.049996729735053e-06, "loss": 0.7975, "step": 11321 }, { "epoch": 1.82, "grad_norm": 10.101104736328125, "learning_rate": 7.048336128782706e-06, "loss": 0.938, "step": 11322 }, { "epoch": 1.82, "grad_norm": 9.185027122497559, "learning_rate": 7.0466756169887e-06, "loss": 0.7159, "step": 11323 }, { "epoch": 1.82, "grad_norm": 10.268172264099121, "learning_rate": 7.045015194403192e-06, "loss": 0.9235, "step": 11324 }, { "epoch": 1.82, "grad_norm": 10.070494651794434, "learning_rate": 7.043354861076337e-06, "loss": 1.0145, "step": 11325 }, { "epoch": 1.82, "grad_norm": 8.86202335357666, "learning_rate": 7.04169461705829e-06, "loss": 1.0246, "step": 11326 }, { "epoch": 1.82, "grad_norm": 12.778836250305176, "learning_rate": 7.040034462399197e-06, "loss": 1.1687, "step": 11327 }, { "epoch": 1.82, "grad_norm": 10.165241241455078, "learning_rate": 7.0383743971492044e-06, "loss": 0.8067, "step": 11328 }, { "epoch": 1.82, "grad_norm": 10.670361518859863, "learning_rate": 7.03671442135846e-06, "loss": 1.278, "step": 11329 }, { "epoch": 1.82, "grad_norm": 11.306258201599121, "learning_rate": 7.035054535077102e-06, "loss": 1.3256, "step": 11330 }, { "epoch": 1.82, "grad_norm": 13.278694152832031, "learning_rate": 7.033394738355268e-06, "loss": 1.2938, "step": 11331 }, { "epoch": 1.82, "grad_norm": 10.309737205505371, "learning_rate": 7.031735031243099e-06, "loss": 1.0656, "step": 11332 }, { "epoch": 1.82, "grad_norm": 10.34362506866455, "learning_rate": 7.030075413790729e-06, "loss": 1.0686, "step": 11333 }, { "epoch": 1.82, "grad_norm": 8.836679458618164, "learning_rate": 7.028415886048282e-06, "loss": 1.0186, "step": 11334 }, { "epoch": 1.82, "grad_norm": 10.53238296508789, "learning_rate": 7.0267564480658905e-06, "loss": 1.0931, "step": 11335 }, { "epoch": 1.82, "grad_norm": 10.196008682250977, "learning_rate": 7.025097099893679e-06, "loss": 1.143, "step": 11336 }, { "epoch": 1.83, "grad_norm": 11.63076114654541, "learning_rate": 7.023437841581769e-06, "loss": 1.1489, "step": 11337 }, { "epoch": 1.83, "grad_norm": 9.58734130859375, "learning_rate": 7.02177867318028e-06, "loss": 1.3171, "step": 11338 }, { "epoch": 1.83, "grad_norm": 9.525572776794434, "learning_rate": 7.020119594739331e-06, "loss": 1.1753, "step": 11339 }, { "epoch": 1.83, "grad_norm": 9.893715858459473, "learning_rate": 7.018460606309034e-06, "loss": 0.9804, "step": 11340 }, { "epoch": 1.83, "grad_norm": 9.75057315826416, "learning_rate": 7.0168017079395025e-06, "loss": 1.1271, "step": 11341 }, { "epoch": 1.83, "grad_norm": 10.701029777526855, "learning_rate": 7.015142899680847e-06, "loss": 1.1679, "step": 11342 }, { "epoch": 1.83, "grad_norm": 11.435273170471191, "learning_rate": 7.013484181583168e-06, "loss": 1.1101, "step": 11343 }, { "epoch": 1.83, "grad_norm": 9.35757827758789, "learning_rate": 7.011825553696574e-06, "loss": 1.038, "step": 11344 }, { "epoch": 1.83, "grad_norm": 13.558486938476562, "learning_rate": 7.0101670160711635e-06, "loss": 1.207, "step": 11345 }, { "epoch": 1.83, "grad_norm": 11.229312896728516, "learning_rate": 7.008508568757036e-06, "loss": 1.0673, "step": 11346 }, { "epoch": 1.83, "grad_norm": 8.895739555358887, "learning_rate": 7.0068502118042854e-06, "loss": 1.1673, "step": 11347 }, { "epoch": 1.83, "grad_norm": 10.978230476379395, "learning_rate": 7.005191945263004e-06, "loss": 0.966, "step": 11348 }, { "epoch": 1.83, "grad_norm": 8.31130599975586, "learning_rate": 7.003533769183283e-06, "loss": 1.0883, "step": 11349 }, { "epoch": 1.83, "grad_norm": 8.854103088378906, "learning_rate": 7.001875683615209e-06, "loss": 0.9908, "step": 11350 }, { "epoch": 1.83, "grad_norm": 9.506172180175781, "learning_rate": 7.000217688608861e-06, "loss": 1.0449, "step": 11351 }, { "epoch": 1.83, "grad_norm": 9.104424476623535, "learning_rate": 6.9985597842143315e-06, "loss": 1.3101, "step": 11352 }, { "epoch": 1.83, "grad_norm": 12.57719612121582, "learning_rate": 6.996901970481693e-06, "loss": 1.0316, "step": 11353 }, { "epoch": 1.83, "grad_norm": 12.94775676727295, "learning_rate": 6.995244247461023e-06, "loss": 1.2317, "step": 11354 }, { "epoch": 1.83, "grad_norm": 10.130752563476562, "learning_rate": 6.993586615202392e-06, "loss": 1.0297, "step": 11355 }, { "epoch": 1.83, "grad_norm": 11.762679100036621, "learning_rate": 6.991929073755874e-06, "loss": 0.8633, "step": 11356 }, { "epoch": 1.83, "grad_norm": 10.124074935913086, "learning_rate": 6.990271623171538e-06, "loss": 1.1103, "step": 11357 }, { "epoch": 1.83, "grad_norm": 9.592601776123047, "learning_rate": 6.988614263499443e-06, "loss": 1.0741, "step": 11358 }, { "epoch": 1.83, "grad_norm": 8.911783218383789, "learning_rate": 6.9869569947896565e-06, "loss": 1.2914, "step": 11359 }, { "epoch": 1.83, "grad_norm": 8.573812484741211, "learning_rate": 6.985299817092237e-06, "loss": 0.9508, "step": 11360 }, { "epoch": 1.83, "grad_norm": 10.666746139526367, "learning_rate": 6.983642730457243e-06, "loss": 0.9552, "step": 11361 }, { "epoch": 1.83, "grad_norm": 9.793705940246582, "learning_rate": 6.981985734934727e-06, "loss": 0.9055, "step": 11362 }, { "epoch": 1.83, "grad_norm": 9.875439643859863, "learning_rate": 6.980328830574739e-06, "loss": 1.1053, "step": 11363 }, { "epoch": 1.83, "grad_norm": 9.516029357910156, "learning_rate": 6.978672017427328e-06, "loss": 1.1427, "step": 11364 }, { "epoch": 1.83, "grad_norm": 9.037063598632812, "learning_rate": 6.9770152955425445e-06, "loss": 0.9049, "step": 11365 }, { "epoch": 1.83, "grad_norm": 10.038900375366211, "learning_rate": 6.975358664970426e-06, "loss": 0.928, "step": 11366 }, { "epoch": 1.83, "grad_norm": 10.316590309143066, "learning_rate": 6.973702125761015e-06, "loss": 0.9707, "step": 11367 }, { "epoch": 1.83, "grad_norm": 12.401111602783203, "learning_rate": 6.9720456779643495e-06, "loss": 1.0616, "step": 11368 }, { "epoch": 1.83, "grad_norm": 10.326516151428223, "learning_rate": 6.970389321630464e-06, "loss": 1.0388, "step": 11369 }, { "epoch": 1.83, "grad_norm": 11.078513145446777, "learning_rate": 6.968733056809394e-06, "loss": 1.1527, "step": 11370 }, { "epoch": 1.83, "grad_norm": 9.004120826721191, "learning_rate": 6.9670768835511625e-06, "loss": 1.1533, "step": 11371 }, { "epoch": 1.83, "grad_norm": 8.758746147155762, "learning_rate": 6.9654208019057965e-06, "loss": 1.072, "step": 11372 }, { "epoch": 1.83, "grad_norm": 9.148944854736328, "learning_rate": 6.963764811923326e-06, "loss": 0.8369, "step": 11373 }, { "epoch": 1.83, "grad_norm": 10.637622833251953, "learning_rate": 6.9621089136537665e-06, "loss": 1.0788, "step": 11374 }, { "epoch": 1.83, "grad_norm": 10.93022632598877, "learning_rate": 6.9604531071471406e-06, "loss": 1.1928, "step": 11375 }, { "epoch": 1.83, "grad_norm": 8.470088005065918, "learning_rate": 6.9587973924534625e-06, "loss": 0.765, "step": 11376 }, { "epoch": 1.83, "grad_norm": 8.186100959777832, "learning_rate": 6.957141769622742e-06, "loss": 1.155, "step": 11377 }, { "epoch": 1.83, "grad_norm": 8.397470474243164, "learning_rate": 6.9554862387049914e-06, "loss": 0.897, "step": 11378 }, { "epoch": 1.83, "grad_norm": 8.961688041687012, "learning_rate": 6.953830799750218e-06, "loss": 0.9646, "step": 11379 }, { "epoch": 1.83, "grad_norm": 8.767284393310547, "learning_rate": 6.952175452808424e-06, "loss": 1.1726, "step": 11380 }, { "epoch": 1.83, "grad_norm": 10.97884750366211, "learning_rate": 6.950520197929615e-06, "loss": 1.1725, "step": 11381 }, { "epoch": 1.83, "grad_norm": 8.62563419342041, "learning_rate": 6.948865035163787e-06, "loss": 0.8558, "step": 11382 }, { "epoch": 1.83, "grad_norm": 9.43350887298584, "learning_rate": 6.947209964560937e-06, "loss": 0.9614, "step": 11383 }, { "epoch": 1.83, "grad_norm": 9.089910507202148, "learning_rate": 6.945554986171057e-06, "loss": 0.8926, "step": 11384 }, { "epoch": 1.83, "grad_norm": 8.699004173278809, "learning_rate": 6.943900100044141e-06, "loss": 0.7444, "step": 11385 }, { "epoch": 1.83, "grad_norm": 7.672641754150391, "learning_rate": 6.942245306230171e-06, "loss": 0.9503, "step": 11386 }, { "epoch": 1.83, "grad_norm": 12.193541526794434, "learning_rate": 6.940590604779135e-06, "loss": 1.1615, "step": 11387 }, { "epoch": 1.83, "grad_norm": 9.683999061584473, "learning_rate": 6.938935995741017e-06, "loss": 1.0162, "step": 11388 }, { "epoch": 1.83, "grad_norm": 10.678272247314453, "learning_rate": 6.937281479165794e-06, "loss": 1.0392, "step": 11389 }, { "epoch": 1.83, "grad_norm": 10.273432731628418, "learning_rate": 6.935627055103442e-06, "loss": 1.2021, "step": 11390 }, { "epoch": 1.83, "grad_norm": 14.236480712890625, "learning_rate": 6.9339727236039365e-06, "loss": 0.9509, "step": 11391 }, { "epoch": 1.83, "grad_norm": 9.268193244934082, "learning_rate": 6.932318484717246e-06, "loss": 1.2854, "step": 11392 }, { "epoch": 1.83, "grad_norm": 10.051055908203125, "learning_rate": 6.93066433849334e-06, "loss": 1.0007, "step": 11393 }, { "epoch": 1.83, "grad_norm": 9.364421844482422, "learning_rate": 6.929010284982187e-06, "loss": 0.8429, "step": 11394 }, { "epoch": 1.83, "grad_norm": 10.070293426513672, "learning_rate": 6.927356324233746e-06, "loss": 1.1659, "step": 11395 }, { "epoch": 1.83, "grad_norm": 10.909967422485352, "learning_rate": 6.92570245629798e-06, "loss": 1.0317, "step": 11396 }, { "epoch": 1.83, "grad_norm": 9.250237464904785, "learning_rate": 6.92404868122484e-06, "loss": 0.9469, "step": 11397 }, { "epoch": 1.83, "grad_norm": 10.344779014587402, "learning_rate": 6.922394999064285e-06, "loss": 0.9945, "step": 11398 }, { "epoch": 1.83, "grad_norm": 9.589859008789062, "learning_rate": 6.920741409866265e-06, "loss": 1.1201, "step": 11399 }, { "epoch": 1.84, "grad_norm": 9.50057601928711, "learning_rate": 6.91908791368073e-06, "loss": 1.3759, "step": 11400 }, { "epoch": 1.84, "grad_norm": 9.850645065307617, "learning_rate": 6.917434510557621e-06, "loss": 1.2834, "step": 11401 }, { "epoch": 1.84, "grad_norm": 9.75755500793457, "learning_rate": 6.915781200546887e-06, "loss": 1.0578, "step": 11402 }, { "epoch": 1.84, "grad_norm": 9.397345542907715, "learning_rate": 6.914127983698463e-06, "loss": 0.7967, "step": 11403 }, { "epoch": 1.84, "grad_norm": 9.904341697692871, "learning_rate": 6.912474860062292e-06, "loss": 1.24, "step": 11404 }, { "epoch": 1.84, "grad_norm": 9.908153533935547, "learning_rate": 6.910821829688301e-06, "loss": 1.3009, "step": 11405 }, { "epoch": 1.84, "grad_norm": 10.981375694274902, "learning_rate": 6.9091688926264275e-06, "loss": 1.0357, "step": 11406 }, { "epoch": 1.84, "grad_norm": 9.71947956085205, "learning_rate": 6.907516048926598e-06, "loss": 0.9522, "step": 11407 }, { "epoch": 1.84, "grad_norm": 9.408036231994629, "learning_rate": 6.905863298638739e-06, "loss": 0.9197, "step": 11408 }, { "epoch": 1.84, "grad_norm": 9.00528335571289, "learning_rate": 6.904210641812775e-06, "loss": 1.046, "step": 11409 }, { "epoch": 1.84, "grad_norm": 8.851059913635254, "learning_rate": 6.902558078498625e-06, "loss": 0.9747, "step": 11410 }, { "epoch": 1.84, "grad_norm": 8.200298309326172, "learning_rate": 6.900905608746204e-06, "loss": 1.2164, "step": 11411 }, { "epoch": 1.84, "grad_norm": 9.076205253601074, "learning_rate": 6.899253232605429e-06, "loss": 0.9849, "step": 11412 }, { "epoch": 1.84, "grad_norm": 9.336552619934082, "learning_rate": 6.897600950126213e-06, "loss": 1.4578, "step": 11413 }, { "epoch": 1.84, "grad_norm": 6.801361560821533, "learning_rate": 6.895948761358465e-06, "loss": 0.9256, "step": 11414 }, { "epoch": 1.84, "grad_norm": 9.741875648498535, "learning_rate": 6.894296666352093e-06, "loss": 0.7661, "step": 11415 }, { "epoch": 1.84, "grad_norm": 9.461530685424805, "learning_rate": 6.8926446651569954e-06, "loss": 1.3503, "step": 11416 }, { "epoch": 1.84, "grad_norm": 11.21316909790039, "learning_rate": 6.890992757823074e-06, "loss": 1.0117, "step": 11417 }, { "epoch": 1.84, "grad_norm": 11.195963859558105, "learning_rate": 6.889340944400229e-06, "loss": 1.0536, "step": 11418 }, { "epoch": 1.84, "grad_norm": 10.869650840759277, "learning_rate": 6.887689224938354e-06, "loss": 1.1713, "step": 11419 }, { "epoch": 1.84, "grad_norm": 10.384806632995605, "learning_rate": 6.886037599487343e-06, "loss": 1.0449, "step": 11420 }, { "epoch": 1.84, "grad_norm": 8.821720123291016, "learning_rate": 6.884386068097082e-06, "loss": 1.2171, "step": 11421 }, { "epoch": 1.84, "grad_norm": 10.089770317077637, "learning_rate": 6.882734630817458e-06, "loss": 0.9474, "step": 11422 }, { "epoch": 1.84, "grad_norm": 7.707110404968262, "learning_rate": 6.881083287698355e-06, "loss": 0.8218, "step": 11423 }, { "epoch": 1.84, "grad_norm": 11.108263969421387, "learning_rate": 6.879432038789655e-06, "loss": 1.0137, "step": 11424 }, { "epoch": 1.84, "grad_norm": 9.041072845458984, "learning_rate": 6.877780884141234e-06, "loss": 1.3385, "step": 11425 }, { "epoch": 1.84, "grad_norm": 9.39964771270752, "learning_rate": 6.876129823802968e-06, "loss": 0.9077, "step": 11426 }, { "epoch": 1.84, "grad_norm": 8.89806842803955, "learning_rate": 6.874478857824729e-06, "loss": 0.9381, "step": 11427 }, { "epoch": 1.84, "grad_norm": 11.749567985534668, "learning_rate": 6.872827986256388e-06, "loss": 0.9105, "step": 11428 }, { "epoch": 1.84, "grad_norm": 10.394810676574707, "learning_rate": 6.871177209147806e-06, "loss": 0.8749, "step": 11429 }, { "epoch": 1.84, "grad_norm": 10.234658241271973, "learning_rate": 6.869526526548852e-06, "loss": 1.0662, "step": 11430 }, { "epoch": 1.84, "grad_norm": 9.324426651000977, "learning_rate": 6.867875938509384e-06, "loss": 1.0822, "step": 11431 }, { "epoch": 1.84, "grad_norm": 8.539039611816406, "learning_rate": 6.86622544507926e-06, "loss": 1.0659, "step": 11432 }, { "epoch": 1.84, "grad_norm": 10.1553373336792, "learning_rate": 6.864575046308336e-06, "loss": 1.1099, "step": 11433 }, { "epoch": 1.84, "grad_norm": 9.66071605682373, "learning_rate": 6.862924742246464e-06, "loss": 1.1811, "step": 11434 }, { "epoch": 1.84, "grad_norm": 9.735432624816895, "learning_rate": 6.8612745329434975e-06, "loss": 1.0904, "step": 11435 }, { "epoch": 1.84, "grad_norm": 9.645402908325195, "learning_rate": 6.859624418449275e-06, "loss": 1.1847, "step": 11436 }, { "epoch": 1.84, "grad_norm": 9.668392181396484, "learning_rate": 6.857974398813644e-06, "loss": 0.9638, "step": 11437 }, { "epoch": 1.84, "grad_norm": 9.24555492401123, "learning_rate": 6.856324474086444e-06, "loss": 1.1102, "step": 11438 }, { "epoch": 1.84, "grad_norm": 10.340551376342773, "learning_rate": 6.8546746443175185e-06, "loss": 1.0438, "step": 11439 }, { "epoch": 1.84, "grad_norm": 11.97131061553955, "learning_rate": 6.853024909556694e-06, "loss": 1.1931, "step": 11440 }, { "epoch": 1.84, "grad_norm": 12.286450386047363, "learning_rate": 6.851375269853806e-06, "loss": 0.9861, "step": 11441 }, { "epoch": 1.84, "grad_norm": 9.353731155395508, "learning_rate": 6.8497257252586846e-06, "loss": 1.0186, "step": 11442 }, { "epoch": 1.84, "grad_norm": 11.002727508544922, "learning_rate": 6.848076275821158e-06, "loss": 1.3357, "step": 11443 }, { "epoch": 1.84, "grad_norm": 9.439421653747559, "learning_rate": 6.846426921591045e-06, "loss": 1.4018, "step": 11444 }, { "epoch": 1.84, "grad_norm": 9.66151237487793, "learning_rate": 6.8447776626181695e-06, "loss": 0.8018, "step": 11445 }, { "epoch": 1.84, "grad_norm": 9.346192359924316, "learning_rate": 6.8431284989523475e-06, "loss": 0.9346, "step": 11446 }, { "epoch": 1.84, "grad_norm": 9.571820259094238, "learning_rate": 6.841479430643394e-06, "loss": 0.9105, "step": 11447 }, { "epoch": 1.84, "grad_norm": 8.999404907226562, "learning_rate": 6.839830457741126e-06, "loss": 0.794, "step": 11448 }, { "epoch": 1.84, "grad_norm": 11.581459045410156, "learning_rate": 6.838181580295343e-06, "loss": 1.1689, "step": 11449 }, { "epoch": 1.84, "grad_norm": 9.754386901855469, "learning_rate": 6.836532798355858e-06, "loss": 0.8839, "step": 11450 }, { "epoch": 1.84, "grad_norm": 9.593833923339844, "learning_rate": 6.834884111972473e-06, "loss": 0.9507, "step": 11451 }, { "epoch": 1.84, "grad_norm": 12.00739860534668, "learning_rate": 6.83323552119499e-06, "loss": 0.9939, "step": 11452 }, { "epoch": 1.84, "grad_norm": 10.0620756149292, "learning_rate": 6.831587026073199e-06, "loss": 1.1146, "step": 11453 }, { "epoch": 1.84, "grad_norm": 9.189847946166992, "learning_rate": 6.829938626656906e-06, "loss": 1.0818, "step": 11454 }, { "epoch": 1.84, "grad_norm": 9.975480079650879, "learning_rate": 6.8282903229958945e-06, "loss": 0.9502, "step": 11455 }, { "epoch": 1.84, "grad_norm": 11.8085355758667, "learning_rate": 6.8266421151399565e-06, "loss": 1.0226, "step": 11456 }, { "epoch": 1.84, "grad_norm": 7.954242706298828, "learning_rate": 6.824994003138878e-06, "loss": 0.8346, "step": 11457 }, { "epoch": 1.84, "grad_norm": 8.110919952392578, "learning_rate": 6.823345987042441e-06, "loss": 0.9457, "step": 11458 }, { "epoch": 1.84, "grad_norm": 10.642152786254883, "learning_rate": 6.821698066900429e-06, "loss": 1.074, "step": 11459 }, { "epoch": 1.84, "grad_norm": 9.292789459228516, "learning_rate": 6.820050242762615e-06, "loss": 1.0209, "step": 11460 }, { "epoch": 1.84, "grad_norm": 9.302181243896484, "learning_rate": 6.818402514678773e-06, "loss": 0.9221, "step": 11461 }, { "epoch": 1.85, "grad_norm": 8.506098747253418, "learning_rate": 6.816754882698677e-06, "loss": 1.1441, "step": 11462 }, { "epoch": 1.85, "grad_norm": 10.713751792907715, "learning_rate": 6.815107346872099e-06, "loss": 0.8878, "step": 11463 }, { "epoch": 1.85, "grad_norm": 9.830744743347168, "learning_rate": 6.813459907248796e-06, "loss": 0.8478, "step": 11464 }, { "epoch": 1.85, "grad_norm": 10.330422401428223, "learning_rate": 6.811812563878537e-06, "loss": 1.0611, "step": 11465 }, { "epoch": 1.85, "grad_norm": 8.75916862487793, "learning_rate": 6.81016531681108e-06, "loss": 0.8958, "step": 11466 }, { "epoch": 1.85, "grad_norm": 11.24899673461914, "learning_rate": 6.808518166096185e-06, "loss": 0.9749, "step": 11467 }, { "epoch": 1.85, "grad_norm": 10.345475196838379, "learning_rate": 6.806871111783599e-06, "loss": 1.0012, "step": 11468 }, { "epoch": 1.85, "grad_norm": 10.935301780700684, "learning_rate": 6.80522415392308e-06, "loss": 1.0325, "step": 11469 }, { "epoch": 1.85, "grad_norm": 8.879318237304688, "learning_rate": 6.8035772925643736e-06, "loss": 0.9945, "step": 11470 }, { "epoch": 1.85, "grad_norm": 9.631021499633789, "learning_rate": 6.801930527757226e-06, "loss": 1.0709, "step": 11471 }, { "epoch": 1.85, "grad_norm": 12.278826713562012, "learning_rate": 6.800283859551378e-06, "loss": 1.2234, "step": 11472 }, { "epoch": 1.85, "grad_norm": 9.604084014892578, "learning_rate": 6.798637287996565e-06, "loss": 1.054, "step": 11473 }, { "epoch": 1.85, "grad_norm": 9.329843521118164, "learning_rate": 6.796990813142534e-06, "loss": 0.9256, "step": 11474 }, { "epoch": 1.85, "grad_norm": 11.575267791748047, "learning_rate": 6.795344435039012e-06, "loss": 1.174, "step": 11475 }, { "epoch": 1.85, "grad_norm": 10.862556457519531, "learning_rate": 6.7936981537357305e-06, "loss": 0.9845, "step": 11476 }, { "epoch": 1.85, "grad_norm": 10.827798843383789, "learning_rate": 6.792051969282417e-06, "loss": 1.0302, "step": 11477 }, { "epoch": 1.85, "grad_norm": 8.742730140686035, "learning_rate": 6.7904058817288e-06, "loss": 0.8346, "step": 11478 }, { "epoch": 1.85, "grad_norm": 10.402436256408691, "learning_rate": 6.788759891124596e-06, "loss": 1.0024, "step": 11479 }, { "epoch": 1.85, "grad_norm": 11.85515308380127, "learning_rate": 6.787113997519526e-06, "loss": 1.1719, "step": 11480 }, { "epoch": 1.85, "grad_norm": 8.621066093444824, "learning_rate": 6.785468200963306e-06, "loss": 0.9733, "step": 11481 }, { "epoch": 1.85, "grad_norm": 11.304368019104004, "learning_rate": 6.783822501505653e-06, "loss": 0.8498, "step": 11482 }, { "epoch": 1.85, "grad_norm": 13.768343925476074, "learning_rate": 6.782176899196271e-06, "loss": 0.8544, "step": 11483 }, { "epoch": 1.85, "grad_norm": 11.144747734069824, "learning_rate": 6.780531394084869e-06, "loss": 1.0069, "step": 11484 }, { "epoch": 1.85, "grad_norm": 10.059405326843262, "learning_rate": 6.778885986221155e-06, "loss": 1.1433, "step": 11485 }, { "epoch": 1.85, "grad_norm": 13.662365913391113, "learning_rate": 6.777240675654826e-06, "loss": 1.1832, "step": 11486 }, { "epoch": 1.85, "grad_norm": 8.86040210723877, "learning_rate": 6.775595462435585e-06, "loss": 1.141, "step": 11487 }, { "epoch": 1.85, "grad_norm": 8.0265531539917, "learning_rate": 6.7739503466131205e-06, "loss": 0.6878, "step": 11488 }, { "epoch": 1.85, "grad_norm": 10.202220916748047, "learning_rate": 6.772305328237132e-06, "loss": 0.9918, "step": 11489 }, { "epoch": 1.85, "grad_norm": 9.725358009338379, "learning_rate": 6.770660407357306e-06, "loss": 0.7737, "step": 11490 }, { "epoch": 1.85, "grad_norm": 10.074305534362793, "learning_rate": 6.76901558402333e-06, "loss": 1.1929, "step": 11491 }, { "epoch": 1.85, "grad_norm": 9.929702758789062, "learning_rate": 6.767370858284887e-06, "loss": 1.2237, "step": 11492 }, { "epoch": 1.85, "grad_norm": 9.647497177124023, "learning_rate": 6.765726230191657e-06, "loss": 0.8665, "step": 11493 }, { "epoch": 1.85, "grad_norm": 11.952396392822266, "learning_rate": 6.76408169979332e-06, "loss": 1.0805, "step": 11494 }, { "epoch": 1.85, "grad_norm": 9.178717613220215, "learning_rate": 6.76243726713955e-06, "loss": 1.0185, "step": 11495 }, { "epoch": 1.85, "grad_norm": 9.930334091186523, "learning_rate": 6.7607929322800205e-06, "loss": 1.3488, "step": 11496 }, { "epoch": 1.85, "grad_norm": 10.514318466186523, "learning_rate": 6.759148695264397e-06, "loss": 1.1174, "step": 11497 }, { "epoch": 1.85, "grad_norm": 9.691190719604492, "learning_rate": 6.75750455614235e-06, "loss": 1.3714, "step": 11498 }, { "epoch": 1.85, "grad_norm": 10.810307502746582, "learning_rate": 6.755860514963538e-06, "loss": 1.1585, "step": 11499 }, { "epoch": 1.85, "grad_norm": 10.639803886413574, "learning_rate": 6.754216571777623e-06, "loss": 0.9509, "step": 11500 }, { "epoch": 1.85, "grad_norm": 10.709664344787598, "learning_rate": 6.752572726634263e-06, "loss": 1.0864, "step": 11501 }, { "epoch": 1.85, "grad_norm": 9.846979141235352, "learning_rate": 6.750928979583114e-06, "loss": 1.1988, "step": 11502 }, { "epoch": 1.85, "grad_norm": 10.300000190734863, "learning_rate": 6.749285330673821e-06, "loss": 0.9787, "step": 11503 }, { "epoch": 1.85, "grad_norm": 8.998629570007324, "learning_rate": 6.747641779956038e-06, "loss": 1.1693, "step": 11504 }, { "epoch": 1.85, "grad_norm": 10.398783683776855, "learning_rate": 6.745998327479406e-06, "loss": 1.0278, "step": 11505 }, { "epoch": 1.85, "grad_norm": 11.437451362609863, "learning_rate": 6.744354973293574e-06, "loss": 1.205, "step": 11506 }, { "epoch": 1.85, "grad_norm": 11.035109519958496, "learning_rate": 6.7427117174481735e-06, "loss": 1.1442, "step": 11507 }, { "epoch": 1.85, "grad_norm": 9.421690940856934, "learning_rate": 6.741068559992845e-06, "loss": 0.8559, "step": 11508 }, { "epoch": 1.85, "grad_norm": 10.55331802368164, "learning_rate": 6.739425500977221e-06, "loss": 1.0286, "step": 11509 }, { "epoch": 1.85, "grad_norm": 11.703124046325684, "learning_rate": 6.7377825404509365e-06, "loss": 1.1769, "step": 11510 }, { "epoch": 1.85, "grad_norm": 12.70662784576416, "learning_rate": 6.7361396784636115e-06, "loss": 1.003, "step": 11511 }, { "epoch": 1.85, "grad_norm": 8.971592903137207, "learning_rate": 6.734496915064872e-06, "loss": 1.0844, "step": 11512 }, { "epoch": 1.85, "grad_norm": 9.80948257446289, "learning_rate": 6.732854250304343e-06, "loss": 0.7955, "step": 11513 }, { "epoch": 1.85, "grad_norm": 7.468273639678955, "learning_rate": 6.73121168423164e-06, "loss": 1.0695, "step": 11514 }, { "epoch": 1.85, "grad_norm": 9.961029052734375, "learning_rate": 6.729569216896381e-06, "loss": 1.0027, "step": 11515 }, { "epoch": 1.85, "grad_norm": 9.401206016540527, "learning_rate": 6.727926848348177e-06, "loss": 1.1905, "step": 11516 }, { "epoch": 1.85, "grad_norm": 11.767326354980469, "learning_rate": 6.726284578636642e-06, "loss": 1.3087, "step": 11517 }, { "epoch": 1.85, "grad_norm": 8.943862915039062, "learning_rate": 6.724642407811375e-06, "loss": 0.915, "step": 11518 }, { "epoch": 1.85, "grad_norm": 10.04813289642334, "learning_rate": 6.723000335921983e-06, "loss": 0.9796, "step": 11519 }, { "epoch": 1.85, "grad_norm": 9.28287124633789, "learning_rate": 6.721358363018068e-06, "loss": 1.087, "step": 11520 }, { "epoch": 1.85, "grad_norm": 9.561849594116211, "learning_rate": 6.719716489149228e-06, "loss": 1.1479, "step": 11521 }, { "epoch": 1.85, "grad_norm": 11.161852836608887, "learning_rate": 6.718074714365055e-06, "loss": 0.9867, "step": 11522 }, { "epoch": 1.85, "grad_norm": 8.707901000976562, "learning_rate": 6.716433038715142e-06, "loss": 0.9777, "step": 11523 }, { "epoch": 1.86, "grad_norm": 8.452688217163086, "learning_rate": 6.714791462249077e-06, "loss": 0.9859, "step": 11524 }, { "epoch": 1.86, "grad_norm": 9.503576278686523, "learning_rate": 6.713149985016447e-06, "loss": 1.1524, "step": 11525 }, { "epoch": 1.86, "grad_norm": 11.097372055053711, "learning_rate": 6.7115086070668364e-06, "loss": 1.2126, "step": 11526 }, { "epoch": 1.86, "grad_norm": 11.989733695983887, "learning_rate": 6.709867328449819e-06, "loss": 0.9784, "step": 11527 }, { "epoch": 1.86, "grad_norm": 11.086761474609375, "learning_rate": 6.7082261492149756e-06, "loss": 0.9353, "step": 11528 }, { "epoch": 1.86, "grad_norm": 9.029619216918945, "learning_rate": 6.706585069411879e-06, "loss": 1.1774, "step": 11529 }, { "epoch": 1.86, "grad_norm": 10.909443855285645, "learning_rate": 6.704944089090103e-06, "loss": 1.0414, "step": 11530 }, { "epoch": 1.86, "grad_norm": 8.749832153320312, "learning_rate": 6.70330320829921e-06, "loss": 0.7865, "step": 11531 }, { "epoch": 1.86, "grad_norm": 9.389444351196289, "learning_rate": 6.701662427088766e-06, "loss": 1.072, "step": 11532 }, { "epoch": 1.86, "grad_norm": 11.859413146972656, "learning_rate": 6.700021745508334e-06, "loss": 1.2066, "step": 11533 }, { "epoch": 1.86, "grad_norm": 8.091217994689941, "learning_rate": 6.6983811636074705e-06, "loss": 0.9144, "step": 11534 }, { "epoch": 1.86, "grad_norm": 10.698904991149902, "learning_rate": 6.696740681435735e-06, "loss": 0.9152, "step": 11535 }, { "epoch": 1.86, "grad_norm": 9.829184532165527, "learning_rate": 6.695100299042679e-06, "loss": 1.0087, "step": 11536 }, { "epoch": 1.86, "grad_norm": 10.213753700256348, "learning_rate": 6.69346001647785e-06, "loss": 0.9732, "step": 11537 }, { "epoch": 1.86, "grad_norm": 11.62052059173584, "learning_rate": 6.691819833790797e-06, "loss": 0.9978, "step": 11538 }, { "epoch": 1.86, "grad_norm": 10.562382698059082, "learning_rate": 6.6901797510310605e-06, "loss": 1.0057, "step": 11539 }, { "epoch": 1.86, "grad_norm": 18.719318389892578, "learning_rate": 6.688539768248183e-06, "loss": 0.8303, "step": 11540 }, { "epoch": 1.86, "grad_norm": 9.263192176818848, "learning_rate": 6.686899885491704e-06, "loss": 1.1281, "step": 11541 }, { "epoch": 1.86, "grad_norm": 12.680733680725098, "learning_rate": 6.685260102811155e-06, "loss": 1.2481, "step": 11542 }, { "epoch": 1.86, "grad_norm": 7.9231085777282715, "learning_rate": 6.683620420256067e-06, "loss": 0.9853, "step": 11543 }, { "epoch": 1.86, "grad_norm": 9.774957656860352, "learning_rate": 6.6819808378759705e-06, "loss": 0.9386, "step": 11544 }, { "epoch": 1.86, "grad_norm": 11.472914695739746, "learning_rate": 6.680341355720393e-06, "loss": 1.1436, "step": 11545 }, { "epoch": 1.86, "grad_norm": 8.45081615447998, "learning_rate": 6.67870197383885e-06, "loss": 1.0174, "step": 11546 }, { "epoch": 1.86, "grad_norm": 11.070265769958496, "learning_rate": 6.6770626922808666e-06, "loss": 1.0395, "step": 11547 }, { "epoch": 1.86, "grad_norm": 8.24082088470459, "learning_rate": 6.675423511095958e-06, "loss": 1.1723, "step": 11548 }, { "epoch": 1.86, "grad_norm": 11.311243057250977, "learning_rate": 6.673784430333639e-06, "loss": 1.0764, "step": 11549 }, { "epoch": 1.86, "grad_norm": 9.199922561645508, "learning_rate": 6.672145450043415e-06, "loss": 1.1823, "step": 11550 }, { "epoch": 1.86, "grad_norm": 8.472206115722656, "learning_rate": 6.670506570274796e-06, "loss": 1.0058, "step": 11551 }, { "epoch": 1.86, "grad_norm": 10.826889991760254, "learning_rate": 6.668867791077287e-06, "loss": 1.0621, "step": 11552 }, { "epoch": 1.86, "grad_norm": 9.47222900390625, "learning_rate": 6.667229112500388e-06, "loss": 0.8248, "step": 11553 }, { "epoch": 1.86, "grad_norm": 11.80243968963623, "learning_rate": 6.665590534593598e-06, "loss": 0.994, "step": 11554 }, { "epoch": 1.86, "grad_norm": 9.14521598815918, "learning_rate": 6.6639520574064124e-06, "loss": 1.1225, "step": 11555 }, { "epoch": 1.86, "grad_norm": 9.352897644042969, "learning_rate": 6.662313680988325e-06, "loss": 1.0918, "step": 11556 }, { "epoch": 1.86, "grad_norm": 9.077078819274902, "learning_rate": 6.66067540538882e-06, "loss": 1.2032, "step": 11557 }, { "epoch": 1.86, "grad_norm": 8.687614440917969, "learning_rate": 6.6590372306573855e-06, "loss": 0.9355, "step": 11558 }, { "epoch": 1.86, "grad_norm": 9.813076972961426, "learning_rate": 6.657399156843507e-06, "loss": 0.8762, "step": 11559 }, { "epoch": 1.86, "grad_norm": 8.407483100891113, "learning_rate": 6.655761183996663e-06, "loss": 1.0424, "step": 11560 }, { "epoch": 1.86, "grad_norm": 10.205848693847656, "learning_rate": 6.654123312166327e-06, "loss": 0.8058, "step": 11561 }, { "epoch": 1.86, "grad_norm": 10.135934829711914, "learning_rate": 6.652485541401975e-06, "loss": 1.2432, "step": 11562 }, { "epoch": 1.86, "grad_norm": 10.07236385345459, "learning_rate": 6.650847871753079e-06, "loss": 0.8291, "step": 11563 }, { "epoch": 1.86, "grad_norm": 10.578246116638184, "learning_rate": 6.6492103032691045e-06, "loss": 0.9812, "step": 11564 }, { "epoch": 1.86, "grad_norm": 8.492730140686035, "learning_rate": 6.64757283599952e-06, "loss": 1.0955, "step": 11565 }, { "epoch": 1.86, "grad_norm": 9.98934555053711, "learning_rate": 6.645935469993783e-06, "loss": 0.9876, "step": 11566 }, { "epoch": 1.86, "grad_norm": 11.81386661529541, "learning_rate": 6.644298205301354e-06, "loss": 1.0756, "step": 11567 }, { "epoch": 1.86, "grad_norm": 10.448673248291016, "learning_rate": 6.642661041971685e-06, "loss": 1.1368, "step": 11568 }, { "epoch": 1.86, "grad_norm": 10.662616729736328, "learning_rate": 6.641023980054235e-06, "loss": 1.2064, "step": 11569 }, { "epoch": 1.86, "grad_norm": 10.540395736694336, "learning_rate": 6.6393870195984465e-06, "loss": 1.1788, "step": 11570 }, { "epoch": 1.86, "grad_norm": 10.914529800415039, "learning_rate": 6.6377501606537695e-06, "loss": 1.1035, "step": 11571 }, { "epoch": 1.86, "grad_norm": 8.593530654907227, "learning_rate": 6.6361134032696465e-06, "loss": 1.1048, "step": 11572 }, { "epoch": 1.86, "grad_norm": 10.637581825256348, "learning_rate": 6.6344767474955196e-06, "loss": 1.1555, "step": 11573 }, { "epoch": 1.86, "grad_norm": 11.106059074401855, "learning_rate": 6.6328401933808165e-06, "loss": 1.1708, "step": 11574 }, { "epoch": 1.86, "grad_norm": 10.693071365356445, "learning_rate": 6.631203740974983e-06, "loss": 1.306, "step": 11575 }, { "epoch": 1.86, "grad_norm": 8.41303825378418, "learning_rate": 6.629567390327447e-06, "loss": 1.0285, "step": 11576 }, { "epoch": 1.86, "grad_norm": 8.034507751464844, "learning_rate": 6.6279311414876325e-06, "loss": 0.7979, "step": 11577 }, { "epoch": 1.86, "grad_norm": 10.818307876586914, "learning_rate": 6.626294994504967e-06, "loss": 0.8662, "step": 11578 }, { "epoch": 1.86, "grad_norm": 8.287376403808594, "learning_rate": 6.62465894942887e-06, "loss": 0.856, "step": 11579 }, { "epoch": 1.86, "grad_norm": 10.193559646606445, "learning_rate": 6.623023006308766e-06, "loss": 0.9036, "step": 11580 }, { "epoch": 1.86, "grad_norm": 9.919461250305176, "learning_rate": 6.621387165194062e-06, "loss": 0.9515, "step": 11581 }, { "epoch": 1.86, "grad_norm": 10.00851821899414, "learning_rate": 6.6197514261341754e-06, "loss": 0.8894, "step": 11582 }, { "epoch": 1.86, "grad_norm": 11.42906379699707, "learning_rate": 6.618115789178516e-06, "loss": 1.0052, "step": 11583 }, { "epoch": 1.86, "grad_norm": 9.801844596862793, "learning_rate": 6.616480254376489e-06, "loss": 0.9937, "step": 11584 }, { "epoch": 1.86, "grad_norm": 10.075911521911621, "learning_rate": 6.614844821777497e-06, "loss": 0.8549, "step": 11585 }, { "epoch": 1.87, "grad_norm": 24.00823402404785, "learning_rate": 6.6132094914309385e-06, "loss": 0.885, "step": 11586 }, { "epoch": 1.87, "grad_norm": 9.736702919006348, "learning_rate": 6.611574263386214e-06, "loss": 1.3601, "step": 11587 }, { "epoch": 1.87, "grad_norm": 10.395511627197266, "learning_rate": 6.609939137692718e-06, "loss": 0.9408, "step": 11588 }, { "epoch": 1.87, "grad_norm": 12.274099349975586, "learning_rate": 6.608304114399836e-06, "loss": 1.1946, "step": 11589 }, { "epoch": 1.87, "grad_norm": 11.579317092895508, "learning_rate": 6.606669193556961e-06, "loss": 1.152, "step": 11590 }, { "epoch": 1.87, "grad_norm": 8.332711219787598, "learning_rate": 6.605034375213473e-06, "loss": 1.082, "step": 11591 }, { "epoch": 1.87, "grad_norm": 11.794041633605957, "learning_rate": 6.603399659418757e-06, "loss": 1.1816, "step": 11592 }, { "epoch": 1.87, "grad_norm": 8.94080924987793, "learning_rate": 6.6017650462221925e-06, "loss": 1.0098, "step": 11593 }, { "epoch": 1.87, "grad_norm": 9.825125694274902, "learning_rate": 6.600130535673151e-06, "loss": 1.1677, "step": 11594 }, { "epoch": 1.87, "grad_norm": 9.567375183105469, "learning_rate": 6.598496127821003e-06, "loss": 1.0183, "step": 11595 }, { "epoch": 1.87, "grad_norm": 10.292393684387207, "learning_rate": 6.596861822715126e-06, "loss": 0.9423, "step": 11596 }, { "epoch": 1.87, "grad_norm": 9.78488540649414, "learning_rate": 6.595227620404881e-06, "loss": 1.0224, "step": 11597 }, { "epoch": 1.87, "grad_norm": 11.061544418334961, "learning_rate": 6.59359352093963e-06, "loss": 1.0331, "step": 11598 }, { "epoch": 1.87, "grad_norm": 8.601569175720215, "learning_rate": 6.591959524368736e-06, "loss": 1.1532, "step": 11599 }, { "epoch": 1.87, "grad_norm": 11.492502212524414, "learning_rate": 6.590325630741552e-06, "loss": 1.1352, "step": 11600 }, { "epoch": 1.87, "grad_norm": 9.531599044799805, "learning_rate": 6.588691840107433e-06, "loss": 1.0359, "step": 11601 }, { "epoch": 1.87, "grad_norm": 8.501736640930176, "learning_rate": 6.587058152515729e-06, "loss": 0.9209, "step": 11602 }, { "epoch": 1.87, "grad_norm": 10.626047134399414, "learning_rate": 6.58542456801579e-06, "loss": 1.0909, "step": 11603 }, { "epoch": 1.87, "grad_norm": 10.30967903137207, "learning_rate": 6.583791086656958e-06, "loss": 1.0338, "step": 11604 }, { "epoch": 1.87, "grad_norm": 9.569877624511719, "learning_rate": 6.582157708488576e-06, "loss": 1.0129, "step": 11605 }, { "epoch": 1.87, "grad_norm": 11.5526762008667, "learning_rate": 6.5805244335599785e-06, "loss": 1.0821, "step": 11606 }, { "epoch": 1.87, "grad_norm": 8.999485969543457, "learning_rate": 6.578891261920504e-06, "loss": 1.1209, "step": 11607 }, { "epoch": 1.87, "grad_norm": 9.460272789001465, "learning_rate": 6.577258193619483e-06, "loss": 1.2526, "step": 11608 }, { "epoch": 1.87, "grad_norm": 10.145529747009277, "learning_rate": 6.575625228706243e-06, "loss": 0.7261, "step": 11609 }, { "epoch": 1.87, "grad_norm": 9.338624954223633, "learning_rate": 6.57399236723011e-06, "loss": 1.0188, "step": 11610 }, { "epoch": 1.87, "grad_norm": 10.361489295959473, "learning_rate": 6.572359609240408e-06, "loss": 1.2175, "step": 11611 }, { "epoch": 1.87, "grad_norm": 10.590182304382324, "learning_rate": 6.570726954786456e-06, "loss": 1.0515, "step": 11612 }, { "epoch": 1.87, "grad_norm": 9.833967208862305, "learning_rate": 6.569094403917568e-06, "loss": 1.0416, "step": 11613 }, { "epoch": 1.87, "grad_norm": 10.609030723571777, "learning_rate": 6.567461956683059e-06, "loss": 0.9599, "step": 11614 }, { "epoch": 1.87, "grad_norm": 9.028974533081055, "learning_rate": 6.565829613132235e-06, "loss": 1.1068, "step": 11615 }, { "epoch": 1.87, "grad_norm": 9.790750503540039, "learning_rate": 6.56419737331441e-06, "loss": 0.8691, "step": 11616 }, { "epoch": 1.87, "grad_norm": 11.400256156921387, "learning_rate": 6.562565237278883e-06, "loss": 1.2104, "step": 11617 }, { "epoch": 1.87, "grad_norm": 9.073598861694336, "learning_rate": 6.560933205074955e-06, "loss": 1.2493, "step": 11618 }, { "epoch": 1.87, "grad_norm": 10.74148941040039, "learning_rate": 6.5593012767519255e-06, "loss": 0.9627, "step": 11619 }, { "epoch": 1.87, "grad_norm": 10.392804145812988, "learning_rate": 6.557669452359086e-06, "loss": 0.7806, "step": 11620 }, { "epoch": 1.87, "grad_norm": 10.053848266601562, "learning_rate": 6.556037731945728e-06, "loss": 0.9076, "step": 11621 }, { "epoch": 1.87, "grad_norm": 8.87427806854248, "learning_rate": 6.55440611556114e-06, "loss": 1.1212, "step": 11622 }, { "epoch": 1.87, "grad_norm": 9.42735481262207, "learning_rate": 6.55277460325461e-06, "loss": 0.9433, "step": 11623 }, { "epoch": 1.87, "grad_norm": 10.422046661376953, "learning_rate": 6.551143195075414e-06, "loss": 1.0858, "step": 11624 }, { "epoch": 1.87, "grad_norm": 8.325125694274902, "learning_rate": 6.549511891072832e-06, "loss": 0.8429, "step": 11625 }, { "epoch": 1.87, "grad_norm": 8.176721572875977, "learning_rate": 6.5478806912961405e-06, "loss": 0.9385, "step": 11626 }, { "epoch": 1.87, "grad_norm": 10.177098274230957, "learning_rate": 6.546249595794614e-06, "loss": 1.0157, "step": 11627 }, { "epoch": 1.87, "grad_norm": 11.420289039611816, "learning_rate": 6.544618604617518e-06, "loss": 1.1292, "step": 11628 }, { "epoch": 1.87, "grad_norm": 10.665033340454102, "learning_rate": 6.542987717814117e-06, "loss": 1.0093, "step": 11629 }, { "epoch": 1.87, "grad_norm": 9.724946975708008, "learning_rate": 6.54135693543368e-06, "loss": 0.9651, "step": 11630 }, { "epoch": 1.87, "grad_norm": 9.018881797790527, "learning_rate": 6.539726257525461e-06, "loss": 0.9539, "step": 11631 }, { "epoch": 1.87, "grad_norm": 9.549958229064941, "learning_rate": 6.538095684138721e-06, "loss": 0.9609, "step": 11632 }, { "epoch": 1.87, "grad_norm": 10.726776123046875, "learning_rate": 6.536465215322709e-06, "loss": 0.9654, "step": 11633 }, { "epoch": 1.87, "grad_norm": 9.61429500579834, "learning_rate": 6.534834851126677e-06, "loss": 1.1421, "step": 11634 }, { "epoch": 1.87, "grad_norm": 11.513001441955566, "learning_rate": 6.53320459159987e-06, "loss": 1.3355, "step": 11635 }, { "epoch": 1.87, "grad_norm": 9.653207778930664, "learning_rate": 6.531574436791536e-06, "loss": 1.2346, "step": 11636 }, { "epoch": 1.87, "grad_norm": 9.99555778503418, "learning_rate": 6.529944386750912e-06, "loss": 1.265, "step": 11637 }, { "epoch": 1.87, "grad_norm": 10.4620943069458, "learning_rate": 6.528314441527242e-06, "loss": 0.8863, "step": 11638 }, { "epoch": 1.87, "grad_norm": 13.862539291381836, "learning_rate": 6.5266846011697505e-06, "loss": 1.0642, "step": 11639 }, { "epoch": 1.87, "grad_norm": 11.719244956970215, "learning_rate": 6.525054865727674e-06, "loss": 0.9786, "step": 11640 }, { "epoch": 1.87, "grad_norm": 10.214909553527832, "learning_rate": 6.52342523525024e-06, "loss": 1.1586, "step": 11641 }, { "epoch": 1.87, "grad_norm": 8.823651313781738, "learning_rate": 6.521795709786675e-06, "loss": 0.9392, "step": 11642 }, { "epoch": 1.87, "grad_norm": 11.214540481567383, "learning_rate": 6.5201662893861994e-06, "loss": 1.0399, "step": 11643 }, { "epoch": 1.87, "grad_norm": 9.930496215820312, "learning_rate": 6.518536974098029e-06, "loss": 0.9397, "step": 11644 }, { "epoch": 1.87, "grad_norm": 8.180281639099121, "learning_rate": 6.516907763971383e-06, "loss": 0.8932, "step": 11645 }, { "epoch": 1.87, "grad_norm": 10.529170989990234, "learning_rate": 6.51527865905547e-06, "loss": 1.2478, "step": 11646 }, { "epoch": 1.87, "grad_norm": 10.683130264282227, "learning_rate": 6.5136496593995035e-06, "loss": 1.1507, "step": 11647 }, { "epoch": 1.88, "grad_norm": 9.102161407470703, "learning_rate": 6.512020765052685e-06, "loss": 0.8312, "step": 11648 }, { "epoch": 1.88, "grad_norm": 9.922212600708008, "learning_rate": 6.510391976064219e-06, "loss": 0.7502, "step": 11649 }, { "epoch": 1.88, "grad_norm": 9.899606704711914, "learning_rate": 6.508763292483305e-06, "loss": 1.0656, "step": 11650 }, { "epoch": 1.88, "grad_norm": 10.381114959716797, "learning_rate": 6.507134714359139e-06, "loss": 1.0058, "step": 11651 }, { "epoch": 1.88, "grad_norm": 9.997389793395996, "learning_rate": 6.505506241740913e-06, "loss": 1.0654, "step": 11652 }, { "epoch": 1.88, "grad_norm": 10.723604202270508, "learning_rate": 6.5038778746778175e-06, "loss": 1.0313, "step": 11653 }, { "epoch": 1.88, "grad_norm": 11.134504318237305, "learning_rate": 6.502249613219041e-06, "loss": 1.0663, "step": 11654 }, { "epoch": 1.88, "grad_norm": 12.5777587890625, "learning_rate": 6.500621457413761e-06, "loss": 1.1926, "step": 11655 }, { "epoch": 1.88, "grad_norm": 14.342912673950195, "learning_rate": 6.498993407311165e-06, "loss": 1.2805, "step": 11656 }, { "epoch": 1.88, "grad_norm": 9.367772102355957, "learning_rate": 6.497365462960428e-06, "loss": 1.1179, "step": 11657 }, { "epoch": 1.88, "grad_norm": 9.423087120056152, "learning_rate": 6.495737624410724e-06, "loss": 1.0905, "step": 11658 }, { "epoch": 1.88, "grad_norm": 7.947801113128662, "learning_rate": 6.494109891711222e-06, "loss": 0.6215, "step": 11659 }, { "epoch": 1.88, "grad_norm": 8.80301284790039, "learning_rate": 6.49248226491109e-06, "loss": 1.2113, "step": 11660 }, { "epoch": 1.88, "grad_norm": 10.000467300415039, "learning_rate": 6.4908547440594915e-06, "loss": 1.0891, "step": 11661 }, { "epoch": 1.88, "grad_norm": 9.739564895629883, "learning_rate": 6.489227329205593e-06, "loss": 1.3249, "step": 11662 }, { "epoch": 1.88, "grad_norm": 8.130902290344238, "learning_rate": 6.487600020398546e-06, "loss": 0.9399, "step": 11663 }, { "epoch": 1.88, "grad_norm": 9.938782691955566, "learning_rate": 6.485972817687505e-06, "loss": 1.0653, "step": 11664 }, { "epoch": 1.88, "grad_norm": 10.817096710205078, "learning_rate": 6.4843457211216255e-06, "loss": 1.2514, "step": 11665 }, { "epoch": 1.88, "grad_norm": 10.740968704223633, "learning_rate": 6.482718730750056e-06, "loss": 1.0271, "step": 11666 }, { "epoch": 1.88, "grad_norm": 13.460774421691895, "learning_rate": 6.481091846621939e-06, "loss": 1.1356, "step": 11667 }, { "epoch": 1.88, "grad_norm": 9.4525146484375, "learning_rate": 6.4794650687864145e-06, "loss": 0.9962, "step": 11668 }, { "epoch": 1.88, "grad_norm": 9.758150100708008, "learning_rate": 6.4778383972926265e-06, "loss": 1.2362, "step": 11669 }, { "epoch": 1.88, "grad_norm": 12.662677764892578, "learning_rate": 6.476211832189706e-06, "loss": 1.1069, "step": 11670 }, { "epoch": 1.88, "grad_norm": 9.271357536315918, "learning_rate": 6.474585373526789e-06, "loss": 0.9143, "step": 11671 }, { "epoch": 1.88, "grad_norm": 9.744542121887207, "learning_rate": 6.472959021353001e-06, "loss": 1.0207, "step": 11672 }, { "epoch": 1.88, "grad_norm": 9.261029243469238, "learning_rate": 6.471332775717469e-06, "loss": 0.9033, "step": 11673 }, { "epoch": 1.88, "grad_norm": 9.880751609802246, "learning_rate": 6.469706636669317e-06, "loss": 1.1298, "step": 11674 }, { "epoch": 1.88, "grad_norm": 10.461175918579102, "learning_rate": 6.468080604257665e-06, "loss": 0.8975, "step": 11675 }, { "epoch": 1.88, "grad_norm": 12.674612045288086, "learning_rate": 6.46645467853162e-06, "loss": 1.2141, "step": 11676 }, { "epoch": 1.88, "grad_norm": 11.169845581054688, "learning_rate": 6.464828859540309e-06, "loss": 1.1359, "step": 11677 }, { "epoch": 1.88, "grad_norm": 9.137018203735352, "learning_rate": 6.463203147332833e-06, "loss": 0.7847, "step": 11678 }, { "epoch": 1.88, "grad_norm": 9.199676513671875, "learning_rate": 6.461577541958302e-06, "loss": 0.9932, "step": 11679 }, { "epoch": 1.88, "grad_norm": 10.030517578125, "learning_rate": 6.459952043465816e-06, "loss": 1.0045, "step": 11680 }, { "epoch": 1.88, "grad_norm": 10.372407913208008, "learning_rate": 6.458326651904476e-06, "loss": 1.0605, "step": 11681 }, { "epoch": 1.88, "grad_norm": 9.372344970703125, "learning_rate": 6.4567013673233836e-06, "loss": 0.9614, "step": 11682 }, { "epoch": 1.88, "grad_norm": 10.5633544921875, "learning_rate": 6.455076189771625e-06, "loss": 1.1326, "step": 11683 }, { "epoch": 1.88, "grad_norm": 8.683487892150879, "learning_rate": 6.453451119298295e-06, "loss": 1.0275, "step": 11684 }, { "epoch": 1.88, "grad_norm": 9.129803657531738, "learning_rate": 6.451826155952478e-06, "loss": 1.1131, "step": 11685 }, { "epoch": 1.88, "grad_norm": 9.89596176147461, "learning_rate": 6.450201299783263e-06, "loss": 0.9096, "step": 11686 }, { "epoch": 1.88, "grad_norm": 9.8145112991333, "learning_rate": 6.448576550839725e-06, "loss": 1.0094, "step": 11687 }, { "epoch": 1.88, "grad_norm": 11.02927017211914, "learning_rate": 6.4469519091709425e-06, "loss": 1.0187, "step": 11688 }, { "epoch": 1.88, "grad_norm": 10.565868377685547, "learning_rate": 6.44532737482599e-06, "loss": 1.0393, "step": 11689 }, { "epoch": 1.88, "grad_norm": 10.240077018737793, "learning_rate": 6.443702947853942e-06, "loss": 1.0454, "step": 11690 }, { "epoch": 1.88, "grad_norm": 9.617318153381348, "learning_rate": 6.442078628303861e-06, "loss": 1.0664, "step": 11691 }, { "epoch": 1.88, "grad_norm": 11.068036079406738, "learning_rate": 6.440454416224813e-06, "loss": 1.1607, "step": 11692 }, { "epoch": 1.88, "grad_norm": 10.842475891113281, "learning_rate": 6.438830311665858e-06, "loss": 1.0282, "step": 11693 }, { "epoch": 1.88, "grad_norm": 10.273509979248047, "learning_rate": 6.43720631467606e-06, "loss": 0.9861, "step": 11694 }, { "epoch": 1.88, "grad_norm": 11.109564781188965, "learning_rate": 6.435582425304466e-06, "loss": 1.1049, "step": 11695 }, { "epoch": 1.88, "grad_norm": 10.417994499206543, "learning_rate": 6.433958643600127e-06, "loss": 1.0602, "step": 11696 }, { "epoch": 1.88, "grad_norm": 10.090373039245605, "learning_rate": 6.432334969612101e-06, "loss": 0.906, "step": 11697 }, { "epoch": 1.88, "grad_norm": 10.231889724731445, "learning_rate": 6.430711403389424e-06, "loss": 1.1937, "step": 11698 }, { "epoch": 1.88, "grad_norm": 8.605525016784668, "learning_rate": 6.429087944981141e-06, "loss": 1.0179, "step": 11699 }, { "epoch": 1.88, "grad_norm": 9.689519882202148, "learning_rate": 6.42746459443629e-06, "loss": 0.9854, "step": 11700 }, { "epoch": 1.88, "grad_norm": 9.44299602508545, "learning_rate": 6.425841351803906e-06, "loss": 1.2443, "step": 11701 }, { "epoch": 1.88, "grad_norm": 10.408792495727539, "learning_rate": 6.42421821713302e-06, "loss": 1.1402, "step": 11702 }, { "epoch": 1.88, "grad_norm": 10.990373611450195, "learning_rate": 6.422595190472663e-06, "loss": 0.9504, "step": 11703 }, { "epoch": 1.88, "grad_norm": 10.323821067810059, "learning_rate": 6.420972271871856e-06, "loss": 1.0251, "step": 11704 }, { "epoch": 1.88, "grad_norm": 9.654081344604492, "learning_rate": 6.419349461379627e-06, "loss": 1.1229, "step": 11705 }, { "epoch": 1.88, "grad_norm": 10.573192596435547, "learning_rate": 6.417726759044989e-06, "loss": 1.0085, "step": 11706 }, { "epoch": 1.88, "grad_norm": 11.73401165008545, "learning_rate": 6.416104164916961e-06, "loss": 1.0462, "step": 11707 }, { "epoch": 1.88, "grad_norm": 9.738654136657715, "learning_rate": 6.414481679044554e-06, "loss": 0.9563, "step": 11708 }, { "epoch": 1.88, "grad_norm": 10.007696151733398, "learning_rate": 6.412859301476777e-06, "loss": 0.8691, "step": 11709 }, { "epoch": 1.89, "grad_norm": 8.813308715820312, "learning_rate": 6.4112370322626406e-06, "loss": 1.0249, "step": 11710 }, { "epoch": 1.89, "grad_norm": 11.075464248657227, "learning_rate": 6.409614871451137e-06, "loss": 1.1265, "step": 11711 }, { "epoch": 1.89, "grad_norm": 8.146275520324707, "learning_rate": 6.407992819091274e-06, "loss": 1.005, "step": 11712 }, { "epoch": 1.89, "grad_norm": 9.995144844055176, "learning_rate": 6.406370875232044e-06, "loss": 1.011, "step": 11713 }, { "epoch": 1.89, "grad_norm": 9.82216739654541, "learning_rate": 6.404749039922445e-06, "loss": 1.0084, "step": 11714 }, { "epoch": 1.89, "grad_norm": 9.603577613830566, "learning_rate": 6.403127313211457e-06, "loss": 1.0353, "step": 11715 }, { "epoch": 1.89, "grad_norm": 9.904534339904785, "learning_rate": 6.4015056951480695e-06, "loss": 0.994, "step": 11716 }, { "epoch": 1.89, "grad_norm": 11.233139038085938, "learning_rate": 6.399884185781268e-06, "loss": 1.1425, "step": 11717 }, { "epoch": 1.89, "grad_norm": 10.344070434570312, "learning_rate": 6.398262785160031e-06, "loss": 1.3314, "step": 11718 }, { "epoch": 1.89, "grad_norm": 11.366830825805664, "learning_rate": 6.396641493333336e-06, "loss": 1.0998, "step": 11719 }, { "epoch": 1.89, "grad_norm": 8.459970474243164, "learning_rate": 6.395020310350153e-06, "loss": 1.0193, "step": 11720 }, { "epoch": 1.89, "grad_norm": 12.229266166687012, "learning_rate": 6.3933992362594545e-06, "loss": 1.0952, "step": 11721 }, { "epoch": 1.89, "grad_norm": 10.84189510345459, "learning_rate": 6.391778271110205e-06, "loss": 0.9641, "step": 11722 }, { "epoch": 1.89, "grad_norm": 9.85059642791748, "learning_rate": 6.390157414951368e-06, "loss": 1.1838, "step": 11723 }, { "epoch": 1.89, "grad_norm": 10.149850845336914, "learning_rate": 6.388536667831902e-06, "loss": 1.0744, "step": 11724 }, { "epoch": 1.89, "grad_norm": 11.223365783691406, "learning_rate": 6.386916029800768e-06, "loss": 1.195, "step": 11725 }, { "epoch": 1.89, "grad_norm": 14.246970176696777, "learning_rate": 6.385295500906914e-06, "loss": 1.0056, "step": 11726 }, { "epoch": 1.89, "grad_norm": 9.689533233642578, "learning_rate": 6.383675081199292e-06, "loss": 0.8967, "step": 11727 }, { "epoch": 1.89, "grad_norm": 11.20120906829834, "learning_rate": 6.382054770726848e-06, "loss": 1.0933, "step": 11728 }, { "epoch": 1.89, "grad_norm": 9.52241039276123, "learning_rate": 6.380434569538527e-06, "loss": 1.0375, "step": 11729 }, { "epoch": 1.89, "grad_norm": 10.42801570892334, "learning_rate": 6.378814477683268e-06, "loss": 1.3576, "step": 11730 }, { "epoch": 1.89, "grad_norm": 8.844100952148438, "learning_rate": 6.377194495210006e-06, "loss": 1.2654, "step": 11731 }, { "epoch": 1.89, "grad_norm": 9.004854202270508, "learning_rate": 6.375574622167677e-06, "loss": 0.9272, "step": 11732 }, { "epoch": 1.89, "grad_norm": 10.518181800842285, "learning_rate": 6.373954858605212e-06, "loss": 1.0568, "step": 11733 }, { "epoch": 1.89, "grad_norm": 10.985841751098633, "learning_rate": 6.372335204571534e-06, "loss": 1.156, "step": 11734 }, { "epoch": 1.89, "grad_norm": 9.13522720336914, "learning_rate": 6.370715660115567e-06, "loss": 0.9747, "step": 11735 }, { "epoch": 1.89, "grad_norm": 9.107662200927734, "learning_rate": 6.369096225286231e-06, "loss": 0.9261, "step": 11736 }, { "epoch": 1.89, "grad_norm": 9.316096305847168, "learning_rate": 6.367476900132446e-06, "loss": 1.1052, "step": 11737 }, { "epoch": 1.89, "grad_norm": 8.79886245727539, "learning_rate": 6.365857684703125e-06, "loss": 0.9213, "step": 11738 }, { "epoch": 1.89, "grad_norm": 9.890955924987793, "learning_rate": 6.3642385790471775e-06, "loss": 1.191, "step": 11739 }, { "epoch": 1.89, "grad_norm": 9.798803329467773, "learning_rate": 6.362619583213512e-06, "loss": 1.0703, "step": 11740 }, { "epoch": 1.89, "grad_norm": 10.708648681640625, "learning_rate": 6.361000697251027e-06, "loss": 1.3282, "step": 11741 }, { "epoch": 1.89, "grad_norm": 10.65541934967041, "learning_rate": 6.359381921208626e-06, "loss": 0.9805, "step": 11742 }, { "epoch": 1.89, "grad_norm": 8.997314453125, "learning_rate": 6.3577632551352065e-06, "loss": 0.9954, "step": 11743 }, { "epoch": 1.89, "grad_norm": 9.298646926879883, "learning_rate": 6.356144699079663e-06, "loss": 0.8996, "step": 11744 }, { "epoch": 1.89, "grad_norm": 11.39840030670166, "learning_rate": 6.3545262530908826e-06, "loss": 1.0275, "step": 11745 }, { "epoch": 1.89, "grad_norm": 11.468342781066895, "learning_rate": 6.352907917217754e-06, "loss": 0.8678, "step": 11746 }, { "epoch": 1.89, "grad_norm": 9.069664001464844, "learning_rate": 6.351289691509161e-06, "loss": 1.0094, "step": 11747 }, { "epoch": 1.89, "grad_norm": 11.89263916015625, "learning_rate": 6.349671576013985e-06, "loss": 0.9898, "step": 11748 }, { "epoch": 1.89, "grad_norm": 9.738363265991211, "learning_rate": 6.348053570781102e-06, "loss": 1.0146, "step": 11749 }, { "epoch": 1.89, "grad_norm": 10.298697471618652, "learning_rate": 6.346435675859383e-06, "loss": 0.9811, "step": 11750 }, { "epoch": 1.89, "grad_norm": 10.338604927062988, "learning_rate": 6.344817891297703e-06, "loss": 1.13, "step": 11751 }, { "epoch": 1.89, "grad_norm": 10.19503116607666, "learning_rate": 6.343200217144927e-06, "loss": 0.9676, "step": 11752 }, { "epoch": 1.89, "grad_norm": 9.512686729431152, "learning_rate": 6.341582653449918e-06, "loss": 0.9376, "step": 11753 }, { "epoch": 1.89, "grad_norm": 12.608262062072754, "learning_rate": 6.339965200261538e-06, "loss": 1.0786, "step": 11754 }, { "epoch": 1.89, "grad_norm": 10.486418724060059, "learning_rate": 6.338347857628641e-06, "loss": 1.1163, "step": 11755 }, { "epoch": 1.89, "grad_norm": 10.05710506439209, "learning_rate": 6.3367306256000825e-06, "loss": 1.0206, "step": 11756 }, { "epoch": 1.89, "grad_norm": 9.374459266662598, "learning_rate": 6.335113504224711e-06, "loss": 1.0525, "step": 11757 }, { "epoch": 1.89, "grad_norm": 12.79289436340332, "learning_rate": 6.333496493551376e-06, "loss": 1.1027, "step": 11758 }, { "epoch": 1.89, "grad_norm": 8.771172523498535, "learning_rate": 6.331879593628923e-06, "loss": 0.9097, "step": 11759 }, { "epoch": 1.89, "grad_norm": 8.268567085266113, "learning_rate": 6.330262804506191e-06, "loss": 0.7968, "step": 11760 }, { "epoch": 1.89, "grad_norm": 11.407547950744629, "learning_rate": 6.328646126232013e-06, "loss": 1.171, "step": 11761 }, { "epoch": 1.89, "grad_norm": 9.859395980834961, "learning_rate": 6.327029558855227e-06, "loss": 1.0675, "step": 11762 }, { "epoch": 1.89, "grad_norm": 9.6920804977417, "learning_rate": 6.32541310242466e-06, "loss": 0.9186, "step": 11763 }, { "epoch": 1.89, "grad_norm": 10.87502670288086, "learning_rate": 6.323796756989142e-06, "loss": 1.1161, "step": 11764 }, { "epoch": 1.89, "grad_norm": 9.446253776550293, "learning_rate": 6.322180522597494e-06, "loss": 0.9515, "step": 11765 }, { "epoch": 1.89, "grad_norm": 10.68687915802002, "learning_rate": 6.320564399298537e-06, "loss": 1.4203, "step": 11766 }, { "epoch": 1.89, "grad_norm": 9.555695533752441, "learning_rate": 6.318948387141087e-06, "loss": 1.233, "step": 11767 }, { "epoch": 1.89, "grad_norm": 8.840615272521973, "learning_rate": 6.317332486173962e-06, "loss": 0.9447, "step": 11768 }, { "epoch": 1.89, "grad_norm": 8.242074966430664, "learning_rate": 6.315716696445966e-06, "loss": 0.8985, "step": 11769 }, { "epoch": 1.89, "grad_norm": 11.064844131469727, "learning_rate": 6.314101018005907e-06, "loss": 0.9094, "step": 11770 }, { "epoch": 1.89, "grad_norm": 12.0986328125, "learning_rate": 6.312485450902592e-06, "loss": 0.8428, "step": 11771 }, { "epoch": 1.9, "grad_norm": 10.73291301727295, "learning_rate": 6.31086999518482e-06, "loss": 1.1488, "step": 11772 }, { "epoch": 1.9, "grad_norm": 8.764458656311035, "learning_rate": 6.309254650901384e-06, "loss": 0.9079, "step": 11773 }, { "epoch": 1.9, "grad_norm": 8.847305297851562, "learning_rate": 6.307639418101079e-06, "loss": 1.1174, "step": 11774 }, { "epoch": 1.9, "grad_norm": 10.541531562805176, "learning_rate": 6.306024296832696e-06, "loss": 0.9766, "step": 11775 }, { "epoch": 1.9, "grad_norm": 9.242361068725586, "learning_rate": 6.304409287145023e-06, "loss": 1.1027, "step": 11776 }, { "epoch": 1.9, "grad_norm": 11.78476333618164, "learning_rate": 6.302794389086837e-06, "loss": 1.1072, "step": 11777 }, { "epoch": 1.9, "grad_norm": 11.866009712219238, "learning_rate": 6.301179602706926e-06, "loss": 1.1701, "step": 11778 }, { "epoch": 1.9, "grad_norm": 10.9415922164917, "learning_rate": 6.299564928054062e-06, "loss": 0.8718, "step": 11779 }, { "epoch": 1.9, "grad_norm": 11.894000053405762, "learning_rate": 6.297950365177019e-06, "loss": 1.2068, "step": 11780 }, { "epoch": 1.9, "grad_norm": 11.707715034484863, "learning_rate": 6.296335914124567e-06, "loss": 0.9745, "step": 11781 }, { "epoch": 1.9, "grad_norm": 8.68844223022461, "learning_rate": 6.294721574945471e-06, "loss": 0.6277, "step": 11782 }, { "epoch": 1.9, "grad_norm": 12.777017593383789, "learning_rate": 6.293107347688496e-06, "loss": 1.3762, "step": 11783 }, { "epoch": 1.9, "grad_norm": 10.109893798828125, "learning_rate": 6.291493232402399e-06, "loss": 1.2995, "step": 11784 }, { "epoch": 1.9, "grad_norm": 10.616735458374023, "learning_rate": 6.2898792291359366e-06, "loss": 1.3183, "step": 11785 }, { "epoch": 1.9, "grad_norm": 8.838390350341797, "learning_rate": 6.288265337937863e-06, "loss": 0.8885, "step": 11786 }, { "epoch": 1.9, "grad_norm": 10.441041946411133, "learning_rate": 6.286651558856926e-06, "loss": 0.9756, "step": 11787 }, { "epoch": 1.9, "grad_norm": 9.017684936523438, "learning_rate": 6.285037891941876e-06, "loss": 1.0799, "step": 11788 }, { "epoch": 1.9, "grad_norm": 9.390618324279785, "learning_rate": 6.2834243372414505e-06, "loss": 1.0733, "step": 11789 }, { "epoch": 1.9, "grad_norm": 10.334569931030273, "learning_rate": 6.2818108948043896e-06, "loss": 0.9205, "step": 11790 }, { "epoch": 1.9, "grad_norm": 8.716232299804688, "learning_rate": 6.280197564679431e-06, "loss": 0.9748, "step": 11791 }, { "epoch": 1.9, "grad_norm": 8.819136619567871, "learning_rate": 6.278584346915307e-06, "loss": 0.9734, "step": 11792 }, { "epoch": 1.9, "grad_norm": 7.975505352020264, "learning_rate": 6.2769712415607455e-06, "loss": 0.8253, "step": 11793 }, { "epoch": 1.9, "grad_norm": 9.458056449890137, "learning_rate": 6.275358248664471e-06, "loss": 1.1417, "step": 11794 }, { "epoch": 1.9, "grad_norm": 10.159311294555664, "learning_rate": 6.273745368275208e-06, "loss": 1.2415, "step": 11795 }, { "epoch": 1.9, "grad_norm": 11.202865600585938, "learning_rate": 6.272132600441676e-06, "loss": 1.0315, "step": 11796 }, { "epoch": 1.9, "grad_norm": 8.646817207336426, "learning_rate": 6.2705199452125834e-06, "loss": 0.8601, "step": 11797 }, { "epoch": 1.9, "grad_norm": 9.31694221496582, "learning_rate": 6.268907402636653e-06, "loss": 1.0463, "step": 11798 }, { "epoch": 1.9, "grad_norm": 9.221246719360352, "learning_rate": 6.26729497276259e-06, "loss": 1.0427, "step": 11799 }, { "epoch": 1.9, "grad_norm": 9.914287567138672, "learning_rate": 6.265682655639096e-06, "loss": 1.0789, "step": 11800 }, { "epoch": 1.9, "grad_norm": 8.10064697265625, "learning_rate": 6.264070451314874e-06, "loss": 0.9634, "step": 11801 }, { "epoch": 1.9, "grad_norm": 9.398557662963867, "learning_rate": 6.262458359838625e-06, "loss": 1.1609, "step": 11802 }, { "epoch": 1.9, "grad_norm": 9.230433464050293, "learning_rate": 6.2608463812590445e-06, "loss": 1.264, "step": 11803 }, { "epoch": 1.9, "grad_norm": 8.705421447753906, "learning_rate": 6.259234515624819e-06, "loss": 1.0546, "step": 11804 }, { "epoch": 1.9, "grad_norm": 10.16740894317627, "learning_rate": 6.257622762984641e-06, "loss": 1.3232, "step": 11805 }, { "epoch": 1.9, "grad_norm": 9.395401000976562, "learning_rate": 6.256011123387194e-06, "loss": 1.1558, "step": 11806 }, { "epoch": 1.9, "grad_norm": 12.187703132629395, "learning_rate": 6.254399596881161e-06, "loss": 0.9206, "step": 11807 }, { "epoch": 1.9, "grad_norm": 8.3053617477417, "learning_rate": 6.252788183515217e-06, "loss": 1.2291, "step": 11808 }, { "epoch": 1.9, "grad_norm": 9.256299018859863, "learning_rate": 6.251176883338038e-06, "loss": 0.9562, "step": 11809 }, { "epoch": 1.9, "grad_norm": 8.713618278503418, "learning_rate": 6.2495656963982956e-06, "loss": 1.0743, "step": 11810 }, { "epoch": 1.9, "grad_norm": 10.155661582946777, "learning_rate": 6.247954622744659e-06, "loss": 0.9258, "step": 11811 }, { "epoch": 1.9, "grad_norm": 9.446645736694336, "learning_rate": 6.246343662425789e-06, "loss": 0.8783, "step": 11812 }, { "epoch": 1.9, "grad_norm": 9.084805488586426, "learning_rate": 6.244732815490347e-06, "loss": 1.0596, "step": 11813 }, { "epoch": 1.9, "grad_norm": 10.90672779083252, "learning_rate": 6.243122081986993e-06, "loss": 1.1917, "step": 11814 }, { "epoch": 1.9, "grad_norm": 9.064903259277344, "learning_rate": 6.2415114619643795e-06, "loss": 1.0969, "step": 11815 }, { "epoch": 1.9, "grad_norm": 10.25096607208252, "learning_rate": 6.239900955471159e-06, "loss": 1.1644, "step": 11816 }, { "epoch": 1.9, "grad_norm": 10.041339874267578, "learning_rate": 6.238290562555971e-06, "loss": 1.1776, "step": 11817 }, { "epoch": 1.9, "grad_norm": 10.817834854125977, "learning_rate": 6.23668028326747e-06, "loss": 1.0805, "step": 11818 }, { "epoch": 1.9, "grad_norm": 10.702778816223145, "learning_rate": 6.23507011765429e-06, "loss": 1.0288, "step": 11819 }, { "epoch": 1.9, "grad_norm": 11.8143310546875, "learning_rate": 6.23346006576507e-06, "loss": 1.1927, "step": 11820 }, { "epoch": 1.9, "grad_norm": 10.380281448364258, "learning_rate": 6.231850127648442e-06, "loss": 1.127, "step": 11821 }, { "epoch": 1.9, "grad_norm": 9.1336088180542, "learning_rate": 6.23024030335304e-06, "loss": 0.8861, "step": 11822 }, { "epoch": 1.9, "grad_norm": 8.605545043945312, "learning_rate": 6.228630592927485e-06, "loss": 0.8375, "step": 11823 }, { "epoch": 1.9, "grad_norm": 9.295137405395508, "learning_rate": 6.227020996420401e-06, "loss": 1.3132, "step": 11824 }, { "epoch": 1.9, "grad_norm": 10.085424423217773, "learning_rate": 6.225411513880411e-06, "loss": 1.0099, "step": 11825 }, { "epoch": 1.9, "grad_norm": 11.051328659057617, "learning_rate": 6.223802145356128e-06, "loss": 1.1921, "step": 11826 }, { "epoch": 1.9, "grad_norm": 8.476014137268066, "learning_rate": 6.222192890896168e-06, "loss": 0.9114, "step": 11827 }, { "epoch": 1.9, "grad_norm": 8.933418273925781, "learning_rate": 6.220583750549138e-06, "loss": 1.2199, "step": 11828 }, { "epoch": 1.9, "grad_norm": 9.811952590942383, "learning_rate": 6.218974724363643e-06, "loss": 0.8272, "step": 11829 }, { "epoch": 1.9, "grad_norm": 10.809555053710938, "learning_rate": 6.2173658123882876e-06, "loss": 1.2004, "step": 11830 }, { "epoch": 1.9, "grad_norm": 11.450773239135742, "learning_rate": 6.215757014671672e-06, "loss": 1.3023, "step": 11831 }, { "epoch": 1.9, "grad_norm": 10.54633903503418, "learning_rate": 6.2141483312623875e-06, "loss": 0.9281, "step": 11832 }, { "epoch": 1.9, "grad_norm": 9.912932395935059, "learning_rate": 6.2125397622090285e-06, "loss": 1.1625, "step": 11833 }, { "epoch": 1.91, "grad_norm": 8.555800437927246, "learning_rate": 6.210931307560184e-06, "loss": 1.126, "step": 11834 }, { "epoch": 1.91, "grad_norm": 10.298884391784668, "learning_rate": 6.209322967364441e-06, "loss": 1.1206, "step": 11835 }, { "epoch": 1.91, "grad_norm": 11.904735565185547, "learning_rate": 6.2077147416703765e-06, "loss": 0.9616, "step": 11836 }, { "epoch": 1.91, "grad_norm": 9.302872657775879, "learning_rate": 6.206106630526572e-06, "loss": 1.1327, "step": 11837 }, { "epoch": 1.91, "grad_norm": 9.369917869567871, "learning_rate": 6.2044986339815995e-06, "loss": 1.158, "step": 11838 }, { "epoch": 1.91, "grad_norm": 13.682343482971191, "learning_rate": 6.202890752084034e-06, "loss": 1.2093, "step": 11839 }, { "epoch": 1.91, "grad_norm": 10.480544090270996, "learning_rate": 6.201282984882442e-06, "loss": 1.074, "step": 11840 }, { "epoch": 1.91, "grad_norm": 8.511099815368652, "learning_rate": 6.1996753324253876e-06, "loss": 0.975, "step": 11841 }, { "epoch": 1.91, "grad_norm": 9.261804580688477, "learning_rate": 6.198067794761434e-06, "loss": 0.9869, "step": 11842 }, { "epoch": 1.91, "grad_norm": 11.171143531799316, "learning_rate": 6.1964603719391345e-06, "loss": 1.0884, "step": 11843 }, { "epoch": 1.91, "grad_norm": 8.591567993164062, "learning_rate": 6.194853064007047e-06, "loss": 0.9352, "step": 11844 }, { "epoch": 1.91, "grad_norm": 12.655078887939453, "learning_rate": 6.193245871013719e-06, "loss": 0.8781, "step": 11845 }, { "epoch": 1.91, "grad_norm": 8.61307430267334, "learning_rate": 6.191638793007702e-06, "loss": 0.9558, "step": 11846 }, { "epoch": 1.91, "grad_norm": 10.258783340454102, "learning_rate": 6.1900318300375325e-06, "loss": 1.0189, "step": 11847 }, { "epoch": 1.91, "grad_norm": 9.953834533691406, "learning_rate": 6.188424982151756e-06, "loss": 1.1495, "step": 11848 }, { "epoch": 1.91, "grad_norm": 12.247810363769531, "learning_rate": 6.186818249398907e-06, "loss": 1.2648, "step": 11849 }, { "epoch": 1.91, "grad_norm": 10.871578216552734, "learning_rate": 6.185211631827523e-06, "loss": 0.9863, "step": 11850 }, { "epoch": 1.91, "grad_norm": 9.494473457336426, "learning_rate": 6.183605129486127e-06, "loss": 0.9483, "step": 11851 }, { "epoch": 1.91, "grad_norm": 11.50161075592041, "learning_rate": 6.18199874242325e-06, "loss": 0.8192, "step": 11852 }, { "epoch": 1.91, "grad_norm": 12.45760726928711, "learning_rate": 6.180392470687411e-06, "loss": 1.0504, "step": 11853 }, { "epoch": 1.91, "grad_norm": 11.12961483001709, "learning_rate": 6.178786314327132e-06, "loss": 0.9921, "step": 11854 }, { "epoch": 1.91, "grad_norm": 11.841431617736816, "learning_rate": 6.17718027339093e-06, "loss": 0.853, "step": 11855 }, { "epoch": 1.91, "grad_norm": 9.88151741027832, "learning_rate": 6.175574347927316e-06, "loss": 1.0734, "step": 11856 }, { "epoch": 1.91, "grad_norm": 11.759256362915039, "learning_rate": 6.1739685379847954e-06, "loss": 1.3391, "step": 11857 }, { "epoch": 1.91, "grad_norm": 9.595921516418457, "learning_rate": 6.172362843611874e-06, "loss": 0.9339, "step": 11858 }, { "epoch": 1.91, "grad_norm": 9.354071617126465, "learning_rate": 6.17075726485706e-06, "loss": 0.8787, "step": 11859 }, { "epoch": 1.91, "grad_norm": 7.521214008331299, "learning_rate": 6.169151801768847e-06, "loss": 0.9096, "step": 11860 }, { "epoch": 1.91, "grad_norm": 10.342849731445312, "learning_rate": 6.167546454395732e-06, "loss": 0.994, "step": 11861 }, { "epoch": 1.91, "grad_norm": 9.917847633361816, "learning_rate": 6.165941222786203e-06, "loss": 1.098, "step": 11862 }, { "epoch": 1.91, "grad_norm": 10.99695873260498, "learning_rate": 6.164336106988751e-06, "loss": 1.1395, "step": 11863 }, { "epoch": 1.91, "grad_norm": 10.080394744873047, "learning_rate": 6.162731107051857e-06, "loss": 1.0481, "step": 11864 }, { "epoch": 1.91, "grad_norm": 10.914823532104492, "learning_rate": 6.161126223024005e-06, "loss": 1.0941, "step": 11865 }, { "epoch": 1.91, "grad_norm": 11.003535270690918, "learning_rate": 6.159521454953672e-06, "loss": 1.071, "step": 11866 }, { "epoch": 1.91, "grad_norm": 9.4467191696167, "learning_rate": 6.15791680288933e-06, "loss": 0.9434, "step": 11867 }, { "epoch": 1.91, "grad_norm": 9.929179191589355, "learning_rate": 6.156312266879451e-06, "loss": 0.9045, "step": 11868 }, { "epoch": 1.91, "grad_norm": 9.925546646118164, "learning_rate": 6.154707846972499e-06, "loss": 1.0051, "step": 11869 }, { "epoch": 1.91, "grad_norm": 11.525944709777832, "learning_rate": 6.153103543216944e-06, "loss": 1.1655, "step": 11870 }, { "epoch": 1.91, "grad_norm": 11.828234672546387, "learning_rate": 6.151499355661238e-06, "loss": 1.2911, "step": 11871 }, { "epoch": 1.91, "grad_norm": 14.167251586914062, "learning_rate": 6.149895284353841e-06, "loss": 1.0776, "step": 11872 }, { "epoch": 1.91, "grad_norm": 9.640710830688477, "learning_rate": 6.1482913293432045e-06, "loss": 1.0608, "step": 11873 }, { "epoch": 1.91, "grad_norm": 12.738809585571289, "learning_rate": 6.146687490677781e-06, "loss": 1.323, "step": 11874 }, { "epoch": 1.91, "grad_norm": 10.477381706237793, "learning_rate": 6.145083768406013e-06, "loss": 1.0277, "step": 11875 }, { "epoch": 1.91, "grad_norm": 10.444498062133789, "learning_rate": 6.143480162576342e-06, "loss": 1.0539, "step": 11876 }, { "epoch": 1.91, "grad_norm": 11.555755615234375, "learning_rate": 6.141876673237208e-06, "loss": 1.233, "step": 11877 }, { "epoch": 1.91, "grad_norm": 11.78912353515625, "learning_rate": 6.140273300437047e-06, "loss": 1.1464, "step": 11878 }, { "epoch": 1.91, "grad_norm": 10.107641220092773, "learning_rate": 6.1386700442242905e-06, "loss": 1.0573, "step": 11879 }, { "epoch": 1.91, "grad_norm": 9.943780899047852, "learning_rate": 6.137066904647366e-06, "loss": 0.9958, "step": 11880 }, { "epoch": 1.91, "grad_norm": 12.06394100189209, "learning_rate": 6.135463881754702e-06, "loss": 0.9893, "step": 11881 }, { "epoch": 1.91, "grad_norm": 13.40414047241211, "learning_rate": 6.133860975594714e-06, "loss": 1.0103, "step": 11882 }, { "epoch": 1.91, "grad_norm": 9.151612281799316, "learning_rate": 6.13225818621582e-06, "loss": 0.895, "step": 11883 }, { "epoch": 1.91, "grad_norm": 9.64433765411377, "learning_rate": 6.130655513666438e-06, "loss": 1.1953, "step": 11884 }, { "epoch": 1.91, "grad_norm": 15.079039573669434, "learning_rate": 6.129052957994977e-06, "loss": 1.0027, "step": 11885 }, { "epoch": 1.91, "grad_norm": 8.99399185180664, "learning_rate": 6.1274505192498425e-06, "loss": 1.2092, "step": 11886 }, { "epoch": 1.91, "grad_norm": 12.477707862854004, "learning_rate": 6.1258481974794385e-06, "loss": 1.0342, "step": 11887 }, { "epoch": 1.91, "grad_norm": 9.799588203430176, "learning_rate": 6.124245992732165e-06, "loss": 0.9535, "step": 11888 }, { "epoch": 1.91, "grad_norm": 9.170333862304688, "learning_rate": 6.12264390505642e-06, "loss": 1.3014, "step": 11889 }, { "epoch": 1.91, "grad_norm": 8.647687911987305, "learning_rate": 6.121041934500593e-06, "loss": 1.164, "step": 11890 }, { "epoch": 1.91, "grad_norm": 8.941753387451172, "learning_rate": 6.1194400811130775e-06, "loss": 0.953, "step": 11891 }, { "epoch": 1.91, "grad_norm": 10.486196517944336, "learning_rate": 6.117838344942255e-06, "loss": 1.1697, "step": 11892 }, { "epoch": 1.91, "grad_norm": 10.002883911132812, "learning_rate": 6.116236726036511e-06, "loss": 1.0859, "step": 11893 }, { "epoch": 1.91, "grad_norm": 9.948970794677734, "learning_rate": 6.114635224444225e-06, "loss": 1.0447, "step": 11894 }, { "epoch": 1.91, "grad_norm": 10.195362091064453, "learning_rate": 6.113033840213768e-06, "loss": 0.9019, "step": 11895 }, { "epoch": 1.92, "grad_norm": 9.751442909240723, "learning_rate": 6.1114325733935145e-06, "loss": 1.0492, "step": 11896 }, { "epoch": 1.92, "grad_norm": 8.082964897155762, "learning_rate": 6.1098314240318315e-06, "loss": 0.9526, "step": 11897 }, { "epoch": 1.92, "grad_norm": 9.016351699829102, "learning_rate": 6.108230392177086e-06, "loss": 0.8907, "step": 11898 }, { "epoch": 1.92, "grad_norm": 10.126168251037598, "learning_rate": 6.1066294778776305e-06, "loss": 0.9439, "step": 11899 }, { "epoch": 1.92, "grad_norm": 10.2437105178833, "learning_rate": 6.105028681181835e-06, "loss": 1.1583, "step": 11900 }, { "epoch": 1.92, "grad_norm": 14.290885925292969, "learning_rate": 6.103428002138046e-06, "loss": 1.1763, "step": 11901 }, { "epoch": 1.92, "grad_norm": 11.48946762084961, "learning_rate": 6.1018274407946145e-06, "loss": 0.7992, "step": 11902 }, { "epoch": 1.92, "grad_norm": 9.903789520263672, "learning_rate": 6.100226997199889e-06, "loss": 0.9887, "step": 11903 }, { "epoch": 1.92, "grad_norm": 9.075465202331543, "learning_rate": 6.0986266714022104e-06, "loss": 0.9185, "step": 11904 }, { "epoch": 1.92, "grad_norm": 13.282084465026855, "learning_rate": 6.097026463449922e-06, "loss": 0.9438, "step": 11905 }, { "epoch": 1.92, "grad_norm": 10.019116401672363, "learning_rate": 6.0954263733913554e-06, "loss": 1.1466, "step": 11906 }, { "epoch": 1.92, "grad_norm": 10.824414253234863, "learning_rate": 6.093826401274845e-06, "loss": 1.0257, "step": 11907 }, { "epoch": 1.92, "grad_norm": 11.92030143737793, "learning_rate": 6.092226547148719e-06, "loss": 1.0825, "step": 11908 }, { "epoch": 1.92, "grad_norm": 12.283059120178223, "learning_rate": 6.090626811061308e-06, "loss": 1.0801, "step": 11909 }, { "epoch": 1.92, "grad_norm": 10.1920747756958, "learning_rate": 6.089027193060926e-06, "loss": 0.9587, "step": 11910 }, { "epoch": 1.92, "grad_norm": 9.536388397216797, "learning_rate": 6.087427693195895e-06, "loss": 1.0048, "step": 11911 }, { "epoch": 1.92, "grad_norm": 10.378960609436035, "learning_rate": 6.085828311514529e-06, "loss": 1.0478, "step": 11912 }, { "epoch": 1.92, "grad_norm": 13.194337844848633, "learning_rate": 6.084229048065142e-06, "loss": 1.1632, "step": 11913 }, { "epoch": 1.92, "grad_norm": 10.593866348266602, "learning_rate": 6.0826299028960375e-06, "loss": 1.1719, "step": 11914 }, { "epoch": 1.92, "grad_norm": 10.726511001586914, "learning_rate": 6.08103087605552e-06, "loss": 0.9298, "step": 11915 }, { "epoch": 1.92, "grad_norm": 9.366445541381836, "learning_rate": 6.079431967591893e-06, "loss": 0.8258, "step": 11916 }, { "epoch": 1.92, "grad_norm": 10.195136070251465, "learning_rate": 6.077833177553453e-06, "loss": 1.178, "step": 11917 }, { "epoch": 1.92, "grad_norm": 10.228541374206543, "learning_rate": 6.07623450598849e-06, "loss": 1.0458, "step": 11918 }, { "epoch": 1.92, "grad_norm": 9.978096961975098, "learning_rate": 6.074635952945289e-06, "loss": 1.2107, "step": 11919 }, { "epoch": 1.92, "grad_norm": 10.617253303527832, "learning_rate": 6.0730375184721515e-06, "loss": 0.8496, "step": 11920 }, { "epoch": 1.92, "grad_norm": 10.375738143920898, "learning_rate": 6.071439202617346e-06, "loss": 1.3179, "step": 11921 }, { "epoch": 1.92, "grad_norm": 9.436944007873535, "learning_rate": 6.069841005429158e-06, "loss": 1.0008, "step": 11922 }, { "epoch": 1.92, "grad_norm": 9.384358406066895, "learning_rate": 6.068242926955862e-06, "loss": 1.136, "step": 11923 }, { "epoch": 1.92, "grad_norm": 9.680198669433594, "learning_rate": 6.066644967245729e-06, "loss": 1.1634, "step": 11924 }, { "epoch": 1.92, "grad_norm": 9.730475425720215, "learning_rate": 6.065047126347028e-06, "loss": 1.0787, "step": 11925 }, { "epoch": 1.92, "grad_norm": 10.43274974822998, "learning_rate": 6.0634494043080205e-06, "loss": 1.0399, "step": 11926 }, { "epoch": 1.92, "grad_norm": 10.851452827453613, "learning_rate": 6.06185180117697e-06, "loss": 0.9572, "step": 11927 }, { "epoch": 1.92, "grad_norm": 9.4778470993042, "learning_rate": 6.060254317002137e-06, "loss": 1.1418, "step": 11928 }, { "epoch": 1.92, "grad_norm": 10.447619438171387, "learning_rate": 6.05865695183177e-06, "loss": 1.0658, "step": 11929 }, { "epoch": 1.92, "grad_norm": 11.431365013122559, "learning_rate": 6.05705970571412e-06, "loss": 0.8963, "step": 11930 }, { "epoch": 1.92, "grad_norm": 7.988107204437256, "learning_rate": 6.055462578697435e-06, "loss": 0.8051, "step": 11931 }, { "epoch": 1.92, "grad_norm": 10.648680686950684, "learning_rate": 6.0538655708299585e-06, "loss": 1.0417, "step": 11932 }, { "epoch": 1.92, "grad_norm": 9.813803672790527, "learning_rate": 6.052268682159932e-06, "loss": 0.7828, "step": 11933 }, { "epoch": 1.92, "grad_norm": 8.915632247924805, "learning_rate": 6.050671912735585e-06, "loss": 0.8577, "step": 11934 }, { "epoch": 1.92, "grad_norm": 8.271072387695312, "learning_rate": 6.049075262605154e-06, "loss": 0.9867, "step": 11935 }, { "epoch": 1.92, "grad_norm": 10.10841178894043, "learning_rate": 6.047478731816867e-06, "loss": 1.2285, "step": 11936 }, { "epoch": 1.92, "grad_norm": 11.080459594726562, "learning_rate": 6.045882320418952e-06, "loss": 0.8854, "step": 11937 }, { "epoch": 1.92, "grad_norm": 9.273027420043945, "learning_rate": 6.044286028459625e-06, "loss": 1.0238, "step": 11938 }, { "epoch": 1.92, "grad_norm": 9.162826538085938, "learning_rate": 6.0426898559871025e-06, "loss": 1.0665, "step": 11939 }, { "epoch": 1.92, "grad_norm": 11.563640594482422, "learning_rate": 6.041093803049607e-06, "loss": 1.3844, "step": 11940 }, { "epoch": 1.92, "grad_norm": 9.747645378112793, "learning_rate": 6.039497869695342e-06, "loss": 1.1341, "step": 11941 }, { "epoch": 1.92, "grad_norm": 8.869515419006348, "learning_rate": 6.037902055972517e-06, "loss": 1.0506, "step": 11942 }, { "epoch": 1.92, "grad_norm": 10.9703369140625, "learning_rate": 6.036306361929338e-06, "loss": 1.0477, "step": 11943 }, { "epoch": 1.92, "grad_norm": 9.699127197265625, "learning_rate": 6.034710787614002e-06, "loss": 1.1177, "step": 11944 }, { "epoch": 1.92, "grad_norm": 8.983991622924805, "learning_rate": 6.0331153330747036e-06, "loss": 1.055, "step": 11945 }, { "epoch": 1.92, "grad_norm": 10.33842945098877, "learning_rate": 6.031519998359637e-06, "loss": 1.152, "step": 11946 }, { "epoch": 1.92, "grad_norm": 11.095094680786133, "learning_rate": 6.029924783516992e-06, "loss": 1.1994, "step": 11947 }, { "epoch": 1.92, "grad_norm": 10.195596694946289, "learning_rate": 6.028329688594954e-06, "loss": 1.2522, "step": 11948 }, { "epoch": 1.92, "grad_norm": 10.299652099609375, "learning_rate": 6.0267347136417e-06, "loss": 1.1077, "step": 11949 }, { "epoch": 1.92, "grad_norm": 11.933868408203125, "learning_rate": 6.025139858705413e-06, "loss": 1.0983, "step": 11950 }, { "epoch": 1.92, "grad_norm": 11.269064903259277, "learning_rate": 6.023545123834266e-06, "loss": 0.9395, "step": 11951 }, { "epoch": 1.92, "grad_norm": 9.618195533752441, "learning_rate": 6.02195050907643e-06, "loss": 0.9709, "step": 11952 }, { "epoch": 1.92, "grad_norm": 11.440296173095703, "learning_rate": 6.020356014480072e-06, "loss": 0.8509, "step": 11953 }, { "epoch": 1.92, "grad_norm": 9.978790283203125, "learning_rate": 6.0187616400933545e-06, "loss": 1.099, "step": 11954 }, { "epoch": 1.92, "grad_norm": 10.215601921081543, "learning_rate": 6.017167385964437e-06, "loss": 1.2442, "step": 11955 }, { "epoch": 1.92, "grad_norm": 10.043633460998535, "learning_rate": 6.015573252141481e-06, "loss": 1.0062, "step": 11956 }, { "epoch": 1.92, "grad_norm": 10.466383934020996, "learning_rate": 6.013979238672633e-06, "loss": 0.9482, "step": 11957 }, { "epoch": 1.92, "grad_norm": 8.747199058532715, "learning_rate": 6.012385345606042e-06, "loss": 0.7174, "step": 11958 }, { "epoch": 1.93, "grad_norm": 9.404154777526855, "learning_rate": 6.010791572989856e-06, "loss": 1.1155, "step": 11959 }, { "epoch": 1.93, "grad_norm": 10.422321319580078, "learning_rate": 6.009197920872216e-06, "loss": 1.2015, "step": 11960 }, { "epoch": 1.93, "grad_norm": 9.558212280273438, "learning_rate": 6.00760438930126e-06, "loss": 0.8428, "step": 11961 }, { "epoch": 1.93, "grad_norm": 13.382176399230957, "learning_rate": 6.006010978325123e-06, "loss": 1.0291, "step": 11962 }, { "epoch": 1.93, "grad_norm": 11.344481468200684, "learning_rate": 6.004417687991939e-06, "loss": 1.0936, "step": 11963 }, { "epoch": 1.93, "grad_norm": 12.09130859375, "learning_rate": 6.002824518349828e-06, "loss": 0.6725, "step": 11964 }, { "epoch": 1.93, "grad_norm": 13.258770942687988, "learning_rate": 6.0012314694469165e-06, "loss": 0.9202, "step": 11965 }, { "epoch": 1.93, "grad_norm": 12.191964149475098, "learning_rate": 5.9996385413313255e-06, "loss": 1.0928, "step": 11966 }, { "epoch": 1.93, "grad_norm": 13.90890884399414, "learning_rate": 5.998045734051172e-06, "loss": 1.3527, "step": 11967 }, { "epoch": 1.93, "grad_norm": 10.332609176635742, "learning_rate": 5.996453047654566e-06, "loss": 0.8511, "step": 11968 }, { "epoch": 1.93, "grad_norm": 9.024627685546875, "learning_rate": 5.994860482189616e-06, "loss": 1.1144, "step": 11969 }, { "epoch": 1.93, "grad_norm": 11.346692085266113, "learning_rate": 5.99326803770443e-06, "loss": 1.0109, "step": 11970 }, { "epoch": 1.93, "grad_norm": 9.490910530090332, "learning_rate": 5.991675714247107e-06, "loss": 1.3669, "step": 11971 }, { "epoch": 1.93, "grad_norm": 11.327676773071289, "learning_rate": 5.9900835118657495e-06, "loss": 1.1552, "step": 11972 }, { "epoch": 1.93, "grad_norm": 11.920929908752441, "learning_rate": 5.988491430608445e-06, "loss": 1.1073, "step": 11973 }, { "epoch": 1.93, "grad_norm": 11.018111228942871, "learning_rate": 5.986899470523289e-06, "loss": 0.8601, "step": 11974 }, { "epoch": 1.93, "grad_norm": 12.430706977844238, "learning_rate": 5.985307631658366e-06, "loss": 0.9465, "step": 11975 }, { "epoch": 1.93, "grad_norm": 10.69361400604248, "learning_rate": 5.983715914061763e-06, "loss": 1.082, "step": 11976 }, { "epoch": 1.93, "grad_norm": 8.69422721862793, "learning_rate": 5.982124317781555e-06, "loss": 1.0361, "step": 11977 }, { "epoch": 1.93, "grad_norm": 9.882251739501953, "learning_rate": 5.980532842865819e-06, "loss": 0.9138, "step": 11978 }, { "epoch": 1.93, "grad_norm": 9.26475715637207, "learning_rate": 5.978941489362629e-06, "loss": 0.892, "step": 11979 }, { "epoch": 1.93, "grad_norm": 10.600890159606934, "learning_rate": 5.9773502573200506e-06, "loss": 0.9615, "step": 11980 }, { "epoch": 1.93, "grad_norm": 9.110000610351562, "learning_rate": 5.975759146786153e-06, "loss": 1.0367, "step": 11981 }, { "epoch": 1.93, "grad_norm": 11.374601364135742, "learning_rate": 5.974168157808995e-06, "loss": 1.1058, "step": 11982 }, { "epoch": 1.93, "grad_norm": 10.261268615722656, "learning_rate": 5.972577290436636e-06, "loss": 0.9833, "step": 11983 }, { "epoch": 1.93, "grad_norm": 9.235880851745605, "learning_rate": 5.970986544717129e-06, "loss": 0.976, "step": 11984 }, { "epoch": 1.93, "grad_norm": 8.374212265014648, "learning_rate": 5.969395920698522e-06, "loss": 1.0815, "step": 11985 }, { "epoch": 1.93, "grad_norm": 10.950727462768555, "learning_rate": 5.9678054184288645e-06, "loss": 1.1751, "step": 11986 }, { "epoch": 1.93, "grad_norm": 10.240667343139648, "learning_rate": 5.9662150379562e-06, "loss": 0.7891, "step": 11987 }, { "epoch": 1.93, "grad_norm": 9.845800399780273, "learning_rate": 5.964624779328565e-06, "loss": 1.1241, "step": 11988 }, { "epoch": 1.93, "grad_norm": 11.72935676574707, "learning_rate": 5.963034642593994e-06, "loss": 1.0636, "step": 11989 }, { "epoch": 1.93, "grad_norm": 11.269789695739746, "learning_rate": 5.961444627800524e-06, "loss": 1.2486, "step": 11990 }, { "epoch": 1.93, "grad_norm": 10.88929271697998, "learning_rate": 5.95985473499618e-06, "loss": 1.008, "step": 11991 }, { "epoch": 1.93, "grad_norm": 12.262020111083984, "learning_rate": 5.958264964228986e-06, "loss": 1.0671, "step": 11992 }, { "epoch": 1.93, "grad_norm": 10.145861625671387, "learning_rate": 5.956675315546963e-06, "loss": 1.2432, "step": 11993 }, { "epoch": 1.93, "grad_norm": 11.468953132629395, "learning_rate": 5.95508578899813e-06, "loss": 1.3042, "step": 11994 }, { "epoch": 1.93, "grad_norm": 8.77471923828125, "learning_rate": 5.953496384630499e-06, "loss": 0.9034, "step": 11995 }, { "epoch": 1.93, "grad_norm": 9.56347370147705, "learning_rate": 5.951907102492078e-06, "loss": 0.8869, "step": 11996 }, { "epoch": 1.93, "grad_norm": 8.875147819519043, "learning_rate": 5.950317942630877e-06, "loss": 0.8939, "step": 11997 }, { "epoch": 1.93, "grad_norm": 9.819106101989746, "learning_rate": 5.948728905094896e-06, "loss": 1.3153, "step": 11998 }, { "epoch": 1.93, "grad_norm": 10.143754005432129, "learning_rate": 5.947139989932134e-06, "loss": 1.0343, "step": 11999 }, { "epoch": 1.93, "grad_norm": 10.275972366333008, "learning_rate": 5.945551197190582e-06, "loss": 1.0978, "step": 12000 }, { "epoch": 1.93, "grad_norm": 11.26082992553711, "learning_rate": 5.943962526918239e-06, "loss": 0.8558, "step": 12001 }, { "epoch": 1.93, "grad_norm": 10.279254913330078, "learning_rate": 5.942373979163092e-06, "loss": 0.9093, "step": 12002 }, { "epoch": 1.93, "grad_norm": 10.953083992004395, "learning_rate": 5.940785553973118e-06, "loss": 1.0536, "step": 12003 }, { "epoch": 1.93, "grad_norm": 10.586421966552734, "learning_rate": 5.939197251396301e-06, "loss": 1.1106, "step": 12004 }, { "epoch": 1.93, "grad_norm": 10.63644790649414, "learning_rate": 5.937609071480618e-06, "loss": 1.1372, "step": 12005 }, { "epoch": 1.93, "grad_norm": 9.880873680114746, "learning_rate": 5.936021014274043e-06, "loss": 0.8393, "step": 12006 }, { "epoch": 1.93, "grad_norm": 11.098893165588379, "learning_rate": 5.934433079824542e-06, "loss": 1.1234, "step": 12007 }, { "epoch": 1.93, "grad_norm": 11.74860954284668, "learning_rate": 5.9328452681800805e-06, "loss": 1.0794, "step": 12008 }, { "epoch": 1.93, "grad_norm": 9.967442512512207, "learning_rate": 5.931257579388621e-06, "loss": 1.1301, "step": 12009 }, { "epoch": 1.93, "grad_norm": 10.767728805541992, "learning_rate": 5.929670013498124e-06, "loss": 1.2436, "step": 12010 }, { "epoch": 1.93, "grad_norm": 12.001029014587402, "learning_rate": 5.928082570556541e-06, "loss": 1.229, "step": 12011 }, { "epoch": 1.93, "grad_norm": 9.105242729187012, "learning_rate": 5.9264952506118215e-06, "loss": 1.1278, "step": 12012 }, { "epoch": 1.93, "grad_norm": 10.356303215026855, "learning_rate": 5.924908053711913e-06, "loss": 1.0475, "step": 12013 }, { "epoch": 1.93, "grad_norm": 13.991397857666016, "learning_rate": 5.9233209799047604e-06, "loss": 1.3347, "step": 12014 }, { "epoch": 1.93, "grad_norm": 11.669990539550781, "learning_rate": 5.921734029238304e-06, "loss": 1.1097, "step": 12015 }, { "epoch": 1.93, "grad_norm": 9.782371520996094, "learning_rate": 5.920147201760475e-06, "loss": 1.0764, "step": 12016 }, { "epoch": 1.93, "grad_norm": 9.397774696350098, "learning_rate": 5.918560497519208e-06, "loss": 1.0574, "step": 12017 }, { "epoch": 1.93, "grad_norm": 10.148541450500488, "learning_rate": 5.91697391656243e-06, "loss": 1.1954, "step": 12018 }, { "epoch": 1.93, "grad_norm": 9.775381088256836, "learning_rate": 5.915387458938069e-06, "loss": 0.8671, "step": 12019 }, { "epoch": 1.93, "grad_norm": 10.576160430908203, "learning_rate": 5.913801124694036e-06, "loss": 1.2884, "step": 12020 }, { "epoch": 1.94, "grad_norm": 10.317849159240723, "learning_rate": 5.912214913878261e-06, "loss": 1.1254, "step": 12021 }, { "epoch": 1.94, "grad_norm": 10.0360689163208, "learning_rate": 5.910628826538653e-06, "loss": 1.2693, "step": 12022 }, { "epoch": 1.94, "grad_norm": 10.579145431518555, "learning_rate": 5.909042862723117e-06, "loss": 0.9516, "step": 12023 }, { "epoch": 1.94, "grad_norm": 9.324698448181152, "learning_rate": 5.907457022479561e-06, "loss": 1.086, "step": 12024 }, { "epoch": 1.94, "grad_norm": 11.021392822265625, "learning_rate": 5.9058713058558904e-06, "loss": 1.0209, "step": 12025 }, { "epoch": 1.94, "grad_norm": 8.193814277648926, "learning_rate": 5.904285712900002e-06, "loss": 0.8415, "step": 12026 }, { "epoch": 1.94, "grad_norm": 8.736254692077637, "learning_rate": 5.9027002436597865e-06, "loss": 0.9336, "step": 12027 }, { "epoch": 1.94, "grad_norm": 9.80414867401123, "learning_rate": 5.901114898183138e-06, "loss": 0.9955, "step": 12028 }, { "epoch": 1.94, "grad_norm": 12.750441551208496, "learning_rate": 5.899529676517944e-06, "loss": 1.0554, "step": 12029 }, { "epoch": 1.94, "grad_norm": 12.791108131408691, "learning_rate": 5.897944578712089e-06, "loss": 1.1435, "step": 12030 }, { "epoch": 1.94, "grad_norm": 8.474806785583496, "learning_rate": 5.896359604813449e-06, "loss": 0.9304, "step": 12031 }, { "epoch": 1.94, "grad_norm": 10.798619270324707, "learning_rate": 5.894774754869903e-06, "loss": 1.2044, "step": 12032 }, { "epoch": 1.94, "grad_norm": 9.447649002075195, "learning_rate": 5.893190028929321e-06, "loss": 1.2555, "step": 12033 }, { "epoch": 1.94, "grad_norm": 10.213974952697754, "learning_rate": 5.891605427039575e-06, "loss": 1.1395, "step": 12034 }, { "epoch": 1.94, "grad_norm": 10.146360397338867, "learning_rate": 5.8900209492485246e-06, "loss": 1.1512, "step": 12035 }, { "epoch": 1.94, "grad_norm": 10.173511505126953, "learning_rate": 5.8884365956040345e-06, "loss": 0.9123, "step": 12036 }, { "epoch": 1.94, "grad_norm": 11.583128929138184, "learning_rate": 5.886852366153959e-06, "loss": 1.1899, "step": 12037 }, { "epoch": 1.94, "grad_norm": 11.96764850616455, "learning_rate": 5.885268260946155e-06, "loss": 0.9644, "step": 12038 }, { "epoch": 1.94, "grad_norm": 10.031564712524414, "learning_rate": 5.883684280028472e-06, "loss": 1.0643, "step": 12039 }, { "epoch": 1.94, "grad_norm": 11.181452751159668, "learning_rate": 5.882100423448747e-06, "loss": 0.8418, "step": 12040 }, { "epoch": 1.94, "grad_norm": 13.649035453796387, "learning_rate": 5.880516691254837e-06, "loss": 1.1245, "step": 12041 }, { "epoch": 1.94, "grad_norm": 12.417610168457031, "learning_rate": 5.878933083494571e-06, "loss": 1.0607, "step": 12042 }, { "epoch": 1.94, "grad_norm": 9.94146728515625, "learning_rate": 5.877349600215785e-06, "loss": 1.1421, "step": 12043 }, { "epoch": 1.94, "grad_norm": 9.761393547058105, "learning_rate": 5.8757662414663115e-06, "loss": 1.2265, "step": 12044 }, { "epoch": 1.94, "grad_norm": 12.732902526855469, "learning_rate": 5.874183007293978e-06, "loss": 0.9974, "step": 12045 }, { "epoch": 1.94, "grad_norm": 11.270859718322754, "learning_rate": 5.872599897746605e-06, "loss": 1.0878, "step": 12046 }, { "epoch": 1.94, "grad_norm": 11.404195785522461, "learning_rate": 5.871016912872014e-06, "loss": 1.1387, "step": 12047 }, { "epoch": 1.94, "grad_norm": 10.837663650512695, "learning_rate": 5.869434052718019e-06, "loss": 1.3578, "step": 12048 }, { "epoch": 1.94, "grad_norm": 9.722037315368652, "learning_rate": 5.8678513173324355e-06, "loss": 1.0108, "step": 12049 }, { "epoch": 1.94, "grad_norm": 11.48898696899414, "learning_rate": 5.866268706763071e-06, "loss": 1.0704, "step": 12050 }, { "epoch": 1.94, "grad_norm": 8.86991024017334, "learning_rate": 5.864686221057727e-06, "loss": 1.2889, "step": 12051 }, { "epoch": 1.94, "grad_norm": 8.619532585144043, "learning_rate": 5.8631038602642055e-06, "loss": 1.2655, "step": 12052 }, { "epoch": 1.94, "grad_norm": 10.084687232971191, "learning_rate": 5.8615216244303065e-06, "loss": 1.1731, "step": 12053 }, { "epoch": 1.94, "grad_norm": 10.52996826171875, "learning_rate": 5.85993951360382e-06, "loss": 0.9, "step": 12054 }, { "epoch": 1.94, "grad_norm": 7.888055324554443, "learning_rate": 5.858357527832536e-06, "loss": 0.8072, "step": 12055 }, { "epoch": 1.94, "grad_norm": 10.293028831481934, "learning_rate": 5.856775667164241e-06, "loss": 1.2718, "step": 12056 }, { "epoch": 1.94, "grad_norm": 8.455496788024902, "learning_rate": 5.855193931646717e-06, "loss": 1.0292, "step": 12057 }, { "epoch": 1.94, "grad_norm": 10.22907829284668, "learning_rate": 5.853612321327742e-06, "loss": 1.0545, "step": 12058 }, { "epoch": 1.94, "grad_norm": 11.008984565734863, "learning_rate": 5.852030836255088e-06, "loss": 1.0867, "step": 12059 }, { "epoch": 1.94, "grad_norm": 8.796072959899902, "learning_rate": 5.8504494764765274e-06, "loss": 0.8473, "step": 12060 }, { "epoch": 1.94, "grad_norm": 9.175500869750977, "learning_rate": 5.848868242039826e-06, "loss": 1.0843, "step": 12061 }, { "epoch": 1.94, "grad_norm": 8.948625564575195, "learning_rate": 5.847287132992749e-06, "loss": 1.2656, "step": 12062 }, { "epoch": 1.94, "grad_norm": 10.480110168457031, "learning_rate": 5.8457061493830545e-06, "loss": 1.0792, "step": 12063 }, { "epoch": 1.94, "grad_norm": 8.69139289855957, "learning_rate": 5.8441252912584975e-06, "loss": 1.0159, "step": 12064 }, { "epoch": 1.94, "grad_norm": 9.511798858642578, "learning_rate": 5.842544558666833e-06, "loss": 1.0605, "step": 12065 }, { "epoch": 1.94, "grad_norm": 10.809141159057617, "learning_rate": 5.840963951655802e-06, "loss": 1.369, "step": 12066 }, { "epoch": 1.94, "grad_norm": 12.403824806213379, "learning_rate": 5.839383470273154e-06, "loss": 0.9834, "step": 12067 }, { "epoch": 1.94, "grad_norm": 9.860513687133789, "learning_rate": 5.837803114566628e-06, "loss": 0.8218, "step": 12068 }, { "epoch": 1.94, "grad_norm": 11.489761352539062, "learning_rate": 5.836222884583955e-06, "loss": 1.079, "step": 12069 }, { "epoch": 1.94, "grad_norm": 13.026054382324219, "learning_rate": 5.834642780372879e-06, "loss": 1.0159, "step": 12070 }, { "epoch": 1.94, "grad_norm": 9.24618148803711, "learning_rate": 5.833062801981121e-06, "loss": 0.9641, "step": 12071 }, { "epoch": 1.94, "grad_norm": 9.355790138244629, "learning_rate": 5.831482949456403e-06, "loss": 0.9248, "step": 12072 }, { "epoch": 1.94, "grad_norm": 9.258820533752441, "learning_rate": 5.829903222846456e-06, "loss": 1.1703, "step": 12073 }, { "epoch": 1.94, "grad_norm": 10.016402244567871, "learning_rate": 5.828323622198986e-06, "loss": 0.9995, "step": 12074 }, { "epoch": 1.94, "grad_norm": 11.896210670471191, "learning_rate": 5.826744147561718e-06, "loss": 0.8266, "step": 12075 }, { "epoch": 1.94, "grad_norm": 12.435508728027344, "learning_rate": 5.825164798982354e-06, "loss": 1.2092, "step": 12076 }, { "epoch": 1.94, "grad_norm": 9.012574195861816, "learning_rate": 5.823585576508602e-06, "loss": 1.0542, "step": 12077 }, { "epoch": 1.94, "grad_norm": 9.738543510437012, "learning_rate": 5.822006480188167e-06, "loss": 0.9454, "step": 12078 }, { "epoch": 1.94, "grad_norm": 10.683672904968262, "learning_rate": 5.82042751006874e-06, "loss": 1.0372, "step": 12079 }, { "epoch": 1.94, "grad_norm": 11.361994743347168, "learning_rate": 5.818848666198027e-06, "loss": 1.0244, "step": 12080 }, { "epoch": 1.94, "grad_norm": 8.480433464050293, "learning_rate": 5.817269948623706e-06, "loss": 1.15, "step": 12081 }, { "epoch": 1.94, "grad_norm": 8.88582706451416, "learning_rate": 5.8156913573934756e-06, "loss": 1.1501, "step": 12082 }, { "epoch": 1.95, "grad_norm": 11.064122200012207, "learning_rate": 5.8141128925550096e-06, "loss": 1.1251, "step": 12083 }, { "epoch": 1.95, "grad_norm": 10.340367317199707, "learning_rate": 5.812534554155995e-06, "loss": 1.1472, "step": 12084 }, { "epoch": 1.95, "grad_norm": 10.851198196411133, "learning_rate": 5.8109563422441e-06, "loss": 1.044, "step": 12085 }, { "epoch": 1.95, "grad_norm": 9.8154296875, "learning_rate": 5.809378256867003e-06, "loss": 1.0984, "step": 12086 }, { "epoch": 1.95, "grad_norm": 12.983672142028809, "learning_rate": 5.807800298072369e-06, "loss": 1.0911, "step": 12087 }, { "epoch": 1.95, "grad_norm": 9.202999114990234, "learning_rate": 5.8062224659078584e-06, "loss": 1.1091, "step": 12088 }, { "epoch": 1.95, "grad_norm": 9.333219528198242, "learning_rate": 5.804644760421139e-06, "loss": 0.6862, "step": 12089 }, { "epoch": 1.95, "grad_norm": 9.432520866394043, "learning_rate": 5.803067181659858e-06, "loss": 1.0615, "step": 12090 }, { "epoch": 1.95, "grad_norm": 8.93860912322998, "learning_rate": 5.801489729671677e-06, "loss": 1.1906, "step": 12091 }, { "epoch": 1.95, "grad_norm": 9.767716407775879, "learning_rate": 5.799912404504242e-06, "loss": 0.9807, "step": 12092 }, { "epoch": 1.95, "grad_norm": 11.87053108215332, "learning_rate": 5.798335206205191e-06, "loss": 0.9654, "step": 12093 }, { "epoch": 1.95, "grad_norm": 8.466815948486328, "learning_rate": 5.796758134822176e-06, "loss": 1.0107, "step": 12094 }, { "epoch": 1.95, "grad_norm": 11.238714218139648, "learning_rate": 5.7951811904028274e-06, "loss": 0.7603, "step": 12095 }, { "epoch": 1.95, "grad_norm": 12.264284133911133, "learning_rate": 5.793604372994777e-06, "loss": 1.4023, "step": 12096 }, { "epoch": 1.95, "grad_norm": 11.530834197998047, "learning_rate": 5.792027682645661e-06, "loss": 1.2234, "step": 12097 }, { "epoch": 1.95, "grad_norm": 10.274724006652832, "learning_rate": 5.790451119403098e-06, "loss": 0.9239, "step": 12098 }, { "epoch": 1.95, "grad_norm": 9.339176177978516, "learning_rate": 5.7888746833147154e-06, "loss": 0.8912, "step": 12099 }, { "epoch": 1.95, "grad_norm": 10.346163749694824, "learning_rate": 5.787298374428132e-06, "loss": 1.0226, "step": 12100 }, { "epoch": 1.95, "grad_norm": 9.531190872192383, "learning_rate": 5.785722192790949e-06, "loss": 1.1708, "step": 12101 }, { "epoch": 1.95, "grad_norm": 9.086275100708008, "learning_rate": 5.784146138450796e-06, "loss": 0.8395, "step": 12102 }, { "epoch": 1.95, "grad_norm": 8.930980682373047, "learning_rate": 5.782570211455269e-06, "loss": 1.0831, "step": 12103 }, { "epoch": 1.95, "grad_norm": 9.358715057373047, "learning_rate": 5.78099441185197e-06, "loss": 1.1207, "step": 12104 }, { "epoch": 1.95, "grad_norm": 8.81629753112793, "learning_rate": 5.779418739688503e-06, "loss": 0.8333, "step": 12105 }, { "epoch": 1.95, "grad_norm": 9.615286827087402, "learning_rate": 5.777843195012459e-06, "loss": 1.0623, "step": 12106 }, { "epoch": 1.95, "grad_norm": 9.545907974243164, "learning_rate": 5.776267777871427e-06, "loss": 1.0056, "step": 12107 }, { "epoch": 1.95, "grad_norm": 9.547313690185547, "learning_rate": 5.7746924883130005e-06, "loss": 1.2542, "step": 12108 }, { "epoch": 1.95, "grad_norm": 10.635091781616211, "learning_rate": 5.7731173263847565e-06, "loss": 1.1363, "step": 12109 }, { "epoch": 1.95, "grad_norm": 9.840900421142578, "learning_rate": 5.77154229213428e-06, "loss": 0.8569, "step": 12110 }, { "epoch": 1.95, "grad_norm": 9.677077293395996, "learning_rate": 5.769967385609145e-06, "loss": 0.9008, "step": 12111 }, { "epoch": 1.95, "grad_norm": 8.914107322692871, "learning_rate": 5.7683926068569206e-06, "loss": 1.1963, "step": 12112 }, { "epoch": 1.95, "grad_norm": 8.787310600280762, "learning_rate": 5.766817955925179e-06, "loss": 1.0005, "step": 12113 }, { "epoch": 1.95, "grad_norm": 10.008349418640137, "learning_rate": 5.7652434328614785e-06, "loss": 1.0173, "step": 12114 }, { "epoch": 1.95, "grad_norm": 12.729647636413574, "learning_rate": 5.763669037713386e-06, "loss": 1.2837, "step": 12115 }, { "epoch": 1.95, "grad_norm": 8.634952545166016, "learning_rate": 5.762094770528457e-06, "loss": 0.9215, "step": 12116 }, { "epoch": 1.95, "grad_norm": 10.561948776245117, "learning_rate": 5.760520631354236e-06, "loss": 1.0313, "step": 12117 }, { "epoch": 1.95, "grad_norm": 11.153362274169922, "learning_rate": 5.7589466202382815e-06, "loss": 1.1989, "step": 12118 }, { "epoch": 1.95, "grad_norm": 9.237065315246582, "learning_rate": 5.7573727372281354e-06, "loss": 0.9948, "step": 12119 }, { "epoch": 1.95, "grad_norm": 9.525321006774902, "learning_rate": 5.755798982371332e-06, "loss": 1.123, "step": 12120 }, { "epoch": 1.95, "grad_norm": 10.13761043548584, "learning_rate": 5.754225355715415e-06, "loss": 1.0436, "step": 12121 }, { "epoch": 1.95, "grad_norm": 9.217230796813965, "learning_rate": 5.75265185730792e-06, "loss": 1.2844, "step": 12122 }, { "epoch": 1.95, "grad_norm": 10.35692024230957, "learning_rate": 5.75107848719637e-06, "loss": 0.8417, "step": 12123 }, { "epoch": 1.95, "grad_norm": 10.503165245056152, "learning_rate": 5.749505245428296e-06, "loss": 1.1315, "step": 12124 }, { "epoch": 1.95, "grad_norm": 9.405068397521973, "learning_rate": 5.747932132051213e-06, "loss": 1.093, "step": 12125 }, { "epoch": 1.95, "grad_norm": 11.016826629638672, "learning_rate": 5.746359147112647e-06, "loss": 0.9955, "step": 12126 }, { "epoch": 1.95, "grad_norm": 10.349523544311523, "learning_rate": 5.744786290660106e-06, "loss": 0.9265, "step": 12127 }, { "epoch": 1.95, "grad_norm": 9.773428916931152, "learning_rate": 5.7432135627410964e-06, "loss": 1.1098, "step": 12128 }, { "epoch": 1.95, "grad_norm": 9.699347496032715, "learning_rate": 5.741640963403134e-06, "loss": 1.2092, "step": 12129 }, { "epoch": 1.95, "grad_norm": 8.388760566711426, "learning_rate": 5.7400684926937165e-06, "loss": 0.899, "step": 12130 }, { "epoch": 1.95, "grad_norm": 11.140007019042969, "learning_rate": 5.7384961506603376e-06, "loss": 1.0522, "step": 12131 }, { "epoch": 1.95, "grad_norm": 8.552356719970703, "learning_rate": 5.736923937350498e-06, "loss": 0.8735, "step": 12132 }, { "epoch": 1.95, "grad_norm": 10.090824127197266, "learning_rate": 5.7353518528116825e-06, "loss": 1.099, "step": 12133 }, { "epoch": 1.95, "grad_norm": 9.096418380737305, "learning_rate": 5.733779897091385e-06, "loss": 0.9, "step": 12134 }, { "epoch": 1.95, "grad_norm": 10.185140609741211, "learning_rate": 5.732208070237084e-06, "loss": 0.8037, "step": 12135 }, { "epoch": 1.95, "grad_norm": 9.427190780639648, "learning_rate": 5.730636372296254e-06, "loss": 1.1352, "step": 12136 }, { "epoch": 1.95, "grad_norm": 9.867833137512207, "learning_rate": 5.72906480331638e-06, "loss": 0.6924, "step": 12137 }, { "epoch": 1.95, "grad_norm": 9.982563972473145, "learning_rate": 5.727493363344925e-06, "loss": 1.2586, "step": 12138 }, { "epoch": 1.95, "grad_norm": 10.984248161315918, "learning_rate": 5.7259220524293555e-06, "loss": 1.1076, "step": 12139 }, { "epoch": 1.95, "grad_norm": 10.597521781921387, "learning_rate": 5.724350870617141e-06, "loss": 1.0094, "step": 12140 }, { "epoch": 1.95, "grad_norm": 11.847128868103027, "learning_rate": 5.722779817955735e-06, "loss": 1.0031, "step": 12141 }, { "epoch": 1.95, "grad_norm": 10.970166206359863, "learning_rate": 5.721208894492594e-06, "loss": 0.9616, "step": 12142 }, { "epoch": 1.95, "grad_norm": 11.753026962280273, "learning_rate": 5.7196381002751756e-06, "loss": 1.2685, "step": 12143 }, { "epoch": 1.95, "grad_norm": 9.878722190856934, "learning_rate": 5.718067435350919e-06, "loss": 0.8401, "step": 12144 }, { "epoch": 1.96, "grad_norm": 9.9688138961792, "learning_rate": 5.7164968997672764e-06, "loss": 1.2465, "step": 12145 }, { "epoch": 1.96, "grad_norm": 11.634140014648438, "learning_rate": 5.7149264935716816e-06, "loss": 0.9268, "step": 12146 }, { "epoch": 1.96, "grad_norm": 9.45816421508789, "learning_rate": 5.7133562168115675e-06, "loss": 0.9298, "step": 12147 }, { "epoch": 1.96, "grad_norm": 9.882491111755371, "learning_rate": 5.711786069534376e-06, "loss": 0.9117, "step": 12148 }, { "epoch": 1.96, "grad_norm": 9.215002059936523, "learning_rate": 5.710216051787527e-06, "loss": 0.8685, "step": 12149 }, { "epoch": 1.96, "grad_norm": 12.310786247253418, "learning_rate": 5.7086461636184456e-06, "loss": 1.1491, "step": 12150 }, { "epoch": 1.96, "grad_norm": 13.828518867492676, "learning_rate": 5.707076405074556e-06, "loss": 1.0951, "step": 12151 }, { "epoch": 1.96, "grad_norm": 9.177794456481934, "learning_rate": 5.705506776203269e-06, "loss": 1.0841, "step": 12152 }, { "epoch": 1.96, "grad_norm": 10.696274757385254, "learning_rate": 5.703937277052003e-06, "loss": 1.1133, "step": 12153 }, { "epoch": 1.96, "grad_norm": 10.574313163757324, "learning_rate": 5.7023679076681645e-06, "loss": 1.0622, "step": 12154 }, { "epoch": 1.96, "grad_norm": 9.583380699157715, "learning_rate": 5.700798668099153e-06, "loss": 0.9174, "step": 12155 }, { "epoch": 1.96, "grad_norm": 9.97715950012207, "learning_rate": 5.699229558392378e-06, "loss": 1.0794, "step": 12156 }, { "epoch": 1.96, "grad_norm": 9.87026596069336, "learning_rate": 5.6976605785952274e-06, "loss": 0.9876, "step": 12157 }, { "epoch": 1.96, "grad_norm": 10.507868766784668, "learning_rate": 5.696091728755102e-06, "loss": 1.2209, "step": 12158 }, { "epoch": 1.96, "grad_norm": 11.925355911254883, "learning_rate": 5.6945230089193855e-06, "loss": 1.0832, "step": 12159 }, { "epoch": 1.96, "grad_norm": 9.655942916870117, "learning_rate": 5.692954419135462e-06, "loss": 0.877, "step": 12160 }, { "epoch": 1.96, "grad_norm": 10.343550682067871, "learning_rate": 5.69138595945072e-06, "loss": 1.1913, "step": 12161 }, { "epoch": 1.96, "grad_norm": 11.401619911193848, "learning_rate": 5.689817629912524e-06, "loss": 1.0038, "step": 12162 }, { "epoch": 1.96, "grad_norm": 10.64479923248291, "learning_rate": 5.6882494305682574e-06, "loss": 0.9901, "step": 12163 }, { "epoch": 1.96, "grad_norm": 9.09349536895752, "learning_rate": 5.686681361465289e-06, "loss": 1.0196, "step": 12164 }, { "epoch": 1.96, "grad_norm": 10.2520170211792, "learning_rate": 5.685113422650983e-06, "loss": 1.1131, "step": 12165 }, { "epoch": 1.96, "grad_norm": 10.65221881866455, "learning_rate": 5.683545614172694e-06, "loss": 0.9182, "step": 12166 }, { "epoch": 1.96, "grad_norm": 9.97018814086914, "learning_rate": 5.681977936077789e-06, "loss": 1.1193, "step": 12167 }, { "epoch": 1.96, "grad_norm": 9.636335372924805, "learning_rate": 5.6804103884136145e-06, "loss": 1.0628, "step": 12168 }, { "epoch": 1.96, "grad_norm": 10.16123104095459, "learning_rate": 5.6788429712275265e-06, "loss": 1.3082, "step": 12169 }, { "epoch": 1.96, "grad_norm": 8.247173309326172, "learning_rate": 5.677275684566867e-06, "loss": 0.9136, "step": 12170 }, { "epoch": 1.96, "grad_norm": 11.789912223815918, "learning_rate": 5.675708528478973e-06, "loss": 1.2388, "step": 12171 }, { "epoch": 1.96, "grad_norm": 10.392256736755371, "learning_rate": 5.674141503011191e-06, "loss": 1.0307, "step": 12172 }, { "epoch": 1.96, "grad_norm": 8.889545440673828, "learning_rate": 5.67257460821085e-06, "loss": 1.0194, "step": 12173 }, { "epoch": 1.96, "grad_norm": 12.324381828308105, "learning_rate": 5.671007844125278e-06, "loss": 1.1616, "step": 12174 }, { "epoch": 1.96, "grad_norm": 11.580276489257812, "learning_rate": 5.669441210801807e-06, "loss": 0.8404, "step": 12175 }, { "epoch": 1.96, "grad_norm": 9.905903816223145, "learning_rate": 5.66787470828775e-06, "loss": 1.1149, "step": 12176 }, { "epoch": 1.96, "grad_norm": 8.350510597229004, "learning_rate": 5.666308336630437e-06, "loss": 0.8547, "step": 12177 }, { "epoch": 1.96, "grad_norm": 11.003134727478027, "learning_rate": 5.664742095877174e-06, "loss": 0.9449, "step": 12178 }, { "epoch": 1.96, "grad_norm": 9.549059867858887, "learning_rate": 5.663175986075267e-06, "loss": 0.887, "step": 12179 }, { "epoch": 1.96, "grad_norm": 12.301119804382324, "learning_rate": 5.661610007272033e-06, "loss": 1.0487, "step": 12180 }, { "epoch": 1.96, "grad_norm": 10.402434349060059, "learning_rate": 5.660044159514764e-06, "loss": 0.9351, "step": 12181 }, { "epoch": 1.96, "grad_norm": 9.450153350830078, "learning_rate": 5.658478442850763e-06, "loss": 0.9519, "step": 12182 }, { "epoch": 1.96, "grad_norm": 10.79446029663086, "learning_rate": 5.656912857327328e-06, "loss": 1.2488, "step": 12183 }, { "epoch": 1.96, "grad_norm": 9.7369966506958, "learning_rate": 5.655347402991746e-06, "loss": 0.8453, "step": 12184 }, { "epoch": 1.96, "grad_norm": 11.051411628723145, "learning_rate": 5.653782079891297e-06, "loss": 1.011, "step": 12185 }, { "epoch": 1.96, "grad_norm": 11.62017822265625, "learning_rate": 5.652216888073274e-06, "loss": 1.1714, "step": 12186 }, { "epoch": 1.96, "grad_norm": 9.959413528442383, "learning_rate": 5.650651827584947e-06, "loss": 1.1561, "step": 12187 }, { "epoch": 1.96, "grad_norm": 9.357078552246094, "learning_rate": 5.649086898473597e-06, "loss": 0.8753, "step": 12188 }, { "epoch": 1.96, "grad_norm": 10.598587989807129, "learning_rate": 5.647522100786492e-06, "loss": 1.0551, "step": 12189 }, { "epoch": 1.96, "grad_norm": 10.170001983642578, "learning_rate": 5.645957434570894e-06, "loss": 1.2487, "step": 12190 }, { "epoch": 1.96, "grad_norm": 10.168906211853027, "learning_rate": 5.644392899874071e-06, "loss": 1.0696, "step": 12191 }, { "epoch": 1.96, "grad_norm": 9.824997901916504, "learning_rate": 5.642828496743279e-06, "loss": 1.1317, "step": 12192 }, { "epoch": 1.96, "grad_norm": 9.203592300415039, "learning_rate": 5.641264225225776e-06, "loss": 1.015, "step": 12193 }, { "epoch": 1.96, "grad_norm": 11.103717803955078, "learning_rate": 5.639700085368809e-06, "loss": 1.4523, "step": 12194 }, { "epoch": 1.96, "grad_norm": 8.4263277053833, "learning_rate": 5.638136077219623e-06, "loss": 0.9137, "step": 12195 }, { "epoch": 1.96, "grad_norm": 10.3767728805542, "learning_rate": 5.636572200825466e-06, "loss": 1.1056, "step": 12196 }, { "epoch": 1.96, "grad_norm": 10.441678047180176, "learning_rate": 5.635008456233575e-06, "loss": 0.9964, "step": 12197 }, { "epoch": 1.96, "grad_norm": 10.044212341308594, "learning_rate": 5.6334448434911794e-06, "loss": 1.4164, "step": 12198 }, { "epoch": 1.96, "grad_norm": 9.440491676330566, "learning_rate": 5.631881362645517e-06, "loss": 1.0877, "step": 12199 }, { "epoch": 1.96, "grad_norm": 11.419833183288574, "learning_rate": 5.6303180137438086e-06, "loss": 0.9769, "step": 12200 }, { "epoch": 1.96, "grad_norm": 9.034775733947754, "learning_rate": 5.628754796833283e-06, "loss": 0.9564, "step": 12201 }, { "epoch": 1.96, "grad_norm": 9.460938453674316, "learning_rate": 5.627191711961153e-06, "loss": 1.1066, "step": 12202 }, { "epoch": 1.96, "grad_norm": 10.980731964111328, "learning_rate": 5.625628759174635e-06, "loss": 1.0619, "step": 12203 }, { "epoch": 1.96, "grad_norm": 11.387149810791016, "learning_rate": 5.624065938520945e-06, "loss": 0.9271, "step": 12204 }, { "epoch": 1.96, "grad_norm": 12.4774808883667, "learning_rate": 5.622503250047286e-06, "loss": 0.8695, "step": 12205 }, { "epoch": 1.96, "grad_norm": 10.032876014709473, "learning_rate": 5.620940693800856e-06, "loss": 0.9571, "step": 12206 }, { "epoch": 1.97, "grad_norm": 9.553010940551758, "learning_rate": 5.619378269828864e-06, "loss": 1.1477, "step": 12207 }, { "epoch": 1.97, "grad_norm": 8.627188682556152, "learning_rate": 5.617815978178496e-06, "loss": 1.2422, "step": 12208 }, { "epoch": 1.97, "grad_norm": 9.285420417785645, "learning_rate": 5.616253818896944e-06, "loss": 0.9752, "step": 12209 }, { "epoch": 1.97, "grad_norm": 9.7142915725708, "learning_rate": 5.614691792031399e-06, "loss": 1.178, "step": 12210 }, { "epoch": 1.97, "grad_norm": 10.034797668457031, "learning_rate": 5.613129897629038e-06, "loss": 1.1809, "step": 12211 }, { "epoch": 1.97, "grad_norm": 7.537208080291748, "learning_rate": 5.611568135737047e-06, "loss": 0.7366, "step": 12212 }, { "epoch": 1.97, "grad_norm": 10.467606544494629, "learning_rate": 5.6100065064025975e-06, "loss": 1.158, "step": 12213 }, { "epoch": 1.97, "grad_norm": 10.330316543579102, "learning_rate": 5.608445009672856e-06, "loss": 1.087, "step": 12214 }, { "epoch": 1.97, "grad_norm": 10.61528491973877, "learning_rate": 5.606883645594996e-06, "loss": 1.0919, "step": 12215 }, { "epoch": 1.97, "grad_norm": 7.905320644378662, "learning_rate": 5.605322414216177e-06, "loss": 0.9685, "step": 12216 }, { "epoch": 1.97, "grad_norm": 10.57320499420166, "learning_rate": 5.603761315583553e-06, "loss": 1.09, "step": 12217 }, { "epoch": 1.97, "grad_norm": 9.870580673217773, "learning_rate": 5.60220034974429e-06, "loss": 0.9068, "step": 12218 }, { "epoch": 1.97, "grad_norm": 11.653286933898926, "learning_rate": 5.600639516745527e-06, "loss": 1.1589, "step": 12219 }, { "epoch": 1.97, "grad_norm": 12.378500938415527, "learning_rate": 5.5990788166344215e-06, "loss": 1.1687, "step": 12220 }, { "epoch": 1.97, "grad_norm": 9.719097137451172, "learning_rate": 5.597518249458109e-06, "loss": 1.1109, "step": 12221 }, { "epoch": 1.97, "grad_norm": 8.895511627197266, "learning_rate": 5.595957815263728e-06, "loss": 1.1165, "step": 12222 }, { "epoch": 1.97, "grad_norm": 10.72024154663086, "learning_rate": 5.5943975140984145e-06, "loss": 1.0892, "step": 12223 }, { "epoch": 1.97, "grad_norm": 10.022798538208008, "learning_rate": 5.592837346009304e-06, "loss": 0.9612, "step": 12224 }, { "epoch": 1.97, "grad_norm": 8.449677467346191, "learning_rate": 5.591277311043515e-06, "loss": 0.9919, "step": 12225 }, { "epoch": 1.97, "grad_norm": 9.192092895507812, "learning_rate": 5.589717409248176e-06, "loss": 0.9572, "step": 12226 }, { "epoch": 1.97, "grad_norm": 10.311389923095703, "learning_rate": 5.588157640670406e-06, "loss": 1.1238, "step": 12227 }, { "epoch": 1.97, "grad_norm": 9.764378547668457, "learning_rate": 5.5865980053573135e-06, "loss": 1.0368, "step": 12228 }, { "epoch": 1.97, "grad_norm": 9.864463806152344, "learning_rate": 5.585038503356018e-06, "loss": 1.2048, "step": 12229 }, { "epoch": 1.97, "grad_norm": 9.429475784301758, "learning_rate": 5.583479134713614e-06, "loss": 0.9615, "step": 12230 }, { "epoch": 1.97, "grad_norm": 8.851400375366211, "learning_rate": 5.581919899477217e-06, "loss": 1.0382, "step": 12231 }, { "epoch": 1.97, "grad_norm": 11.235156059265137, "learning_rate": 5.580360797693918e-06, "loss": 1.381, "step": 12232 }, { "epoch": 1.97, "grad_norm": 11.5936279296875, "learning_rate": 5.578801829410809e-06, "loss": 0.8155, "step": 12233 }, { "epoch": 1.97, "grad_norm": 8.175942420959473, "learning_rate": 5.577242994674989e-06, "loss": 0.8789, "step": 12234 }, { "epoch": 1.97, "grad_norm": 11.359100341796875, "learning_rate": 5.575684293533535e-06, "loss": 1.2065, "step": 12235 }, { "epoch": 1.97, "grad_norm": 8.211591720581055, "learning_rate": 5.574125726033538e-06, "loss": 1.1065, "step": 12236 }, { "epoch": 1.97, "grad_norm": 10.33486270904541, "learning_rate": 5.572567292222072e-06, "loss": 1.048, "step": 12237 }, { "epoch": 1.97, "grad_norm": 10.935480117797852, "learning_rate": 5.571008992146207e-06, "loss": 1.0454, "step": 12238 }, { "epoch": 1.97, "grad_norm": 10.834856986999512, "learning_rate": 5.569450825853022e-06, "loss": 1.2618, "step": 12239 }, { "epoch": 1.97, "grad_norm": 8.528762817382812, "learning_rate": 5.56789279338958e-06, "loss": 0.8051, "step": 12240 }, { "epoch": 1.97, "grad_norm": 10.123093605041504, "learning_rate": 5.566334894802936e-06, "loss": 1.0047, "step": 12241 }, { "epoch": 1.97, "grad_norm": 10.094438552856445, "learning_rate": 5.56477713014016e-06, "loss": 0.9089, "step": 12242 }, { "epoch": 1.97, "grad_norm": 10.33353328704834, "learning_rate": 5.563219499448295e-06, "loss": 0.9807, "step": 12243 }, { "epoch": 1.97, "grad_norm": 9.940031051635742, "learning_rate": 5.561662002774395e-06, "loss": 1.0827, "step": 12244 }, { "epoch": 1.97, "grad_norm": 11.146240234375, "learning_rate": 5.560104640165513e-06, "loss": 1.0095, "step": 12245 }, { "epoch": 1.97, "grad_norm": 11.092449188232422, "learning_rate": 5.55854741166868e-06, "loss": 0.8097, "step": 12246 }, { "epoch": 1.97, "grad_norm": 10.721333503723145, "learning_rate": 5.556990317330942e-06, "loss": 1.3457, "step": 12247 }, { "epoch": 1.97, "grad_norm": 8.757911682128906, "learning_rate": 5.555433357199331e-06, "loss": 1.0487, "step": 12248 }, { "epoch": 1.97, "grad_norm": 11.338973999023438, "learning_rate": 5.55387653132087e-06, "loss": 1.0946, "step": 12249 }, { "epoch": 1.97, "grad_norm": 8.663117408752441, "learning_rate": 5.5523198397425945e-06, "loss": 0.9612, "step": 12250 }, { "epoch": 1.97, "grad_norm": 8.029363632202148, "learning_rate": 5.550763282511522e-06, "loss": 1.1794, "step": 12251 }, { "epoch": 1.97, "grad_norm": 10.799120903015137, "learning_rate": 5.549206859674665e-06, "loss": 0.6695, "step": 12252 }, { "epoch": 1.97, "grad_norm": 10.254781723022461, "learning_rate": 5.547650571279046e-06, "loss": 1.014, "step": 12253 }, { "epoch": 1.97, "grad_norm": 10.423540115356445, "learning_rate": 5.546094417371666e-06, "loss": 1.0046, "step": 12254 }, { "epoch": 1.97, "grad_norm": 9.615004539489746, "learning_rate": 5.544538397999538e-06, "loss": 0.8054, "step": 12255 }, { "epoch": 1.97, "grad_norm": 9.686446189880371, "learning_rate": 5.542982513209661e-06, "loss": 1.0297, "step": 12256 }, { "epoch": 1.97, "grad_norm": 11.81408977508545, "learning_rate": 5.541426763049026e-06, "loss": 1.1595, "step": 12257 }, { "epoch": 1.97, "grad_norm": 11.314461708068848, "learning_rate": 5.539871147564635e-06, "loss": 0.9986, "step": 12258 }, { "epoch": 1.97, "grad_norm": 10.40050983428955, "learning_rate": 5.538315666803471e-06, "loss": 0.9293, "step": 12259 }, { "epoch": 1.97, "grad_norm": 11.775063514709473, "learning_rate": 5.536760320812524e-06, "loss": 1.3345, "step": 12260 }, { "epoch": 1.97, "grad_norm": 16.724475860595703, "learning_rate": 5.535205109638773e-06, "loss": 1.1105, "step": 12261 }, { "epoch": 1.97, "grad_norm": 10.45237922668457, "learning_rate": 5.53365003332919e-06, "loss": 0.9243, "step": 12262 }, { "epoch": 1.97, "grad_norm": 9.839364051818848, "learning_rate": 5.532095091930753e-06, "loss": 1.1036, "step": 12263 }, { "epoch": 1.97, "grad_norm": 8.007378578186035, "learning_rate": 5.530540285490433e-06, "loss": 0.8849, "step": 12264 }, { "epoch": 1.97, "grad_norm": 10.385551452636719, "learning_rate": 5.528985614055188e-06, "loss": 0.9993, "step": 12265 }, { "epoch": 1.97, "grad_norm": 10.997855186462402, "learning_rate": 5.527431077671987e-06, "loss": 1.2796, "step": 12266 }, { "epoch": 1.97, "grad_norm": 10.934616088867188, "learning_rate": 5.5258766763877806e-06, "loss": 0.9741, "step": 12267 }, { "epoch": 1.97, "grad_norm": 9.828282356262207, "learning_rate": 5.5243224102495185e-06, "loss": 1.0306, "step": 12268 }, { "epoch": 1.98, "grad_norm": 11.906984329223633, "learning_rate": 5.522768279304157e-06, "loss": 1.1916, "step": 12269 }, { "epoch": 1.98, "grad_norm": 8.604134559631348, "learning_rate": 5.521214283598633e-06, "loss": 0.8851, "step": 12270 }, { "epoch": 1.98, "grad_norm": 10.510156631469727, "learning_rate": 5.519660423179894e-06, "loss": 1.2477, "step": 12271 }, { "epoch": 1.98, "grad_norm": 7.6445417404174805, "learning_rate": 5.518106698094872e-06, "loss": 1.0003, "step": 12272 }, { "epoch": 1.98, "grad_norm": 12.38883113861084, "learning_rate": 5.5165531083904945e-06, "loss": 0.962, "step": 12273 }, { "epoch": 1.98, "grad_norm": 10.3975248336792, "learning_rate": 5.514999654113698e-06, "loss": 1.0445, "step": 12274 }, { "epoch": 1.98, "grad_norm": 10.461248397827148, "learning_rate": 5.513446335311403e-06, "loss": 0.9478, "step": 12275 }, { "epoch": 1.98, "grad_norm": 9.074675559997559, "learning_rate": 5.511893152030526e-06, "loss": 1.2021, "step": 12276 }, { "epoch": 1.98, "grad_norm": 11.716673851013184, "learning_rate": 5.510340104317986e-06, "loss": 1.1071, "step": 12277 }, { "epoch": 1.98, "grad_norm": 12.06120777130127, "learning_rate": 5.5087871922206925e-06, "loss": 1.4288, "step": 12278 }, { "epoch": 1.98, "grad_norm": 12.611865997314453, "learning_rate": 5.507234415785558e-06, "loss": 1.0274, "step": 12279 }, { "epoch": 1.98, "grad_norm": 9.591243743896484, "learning_rate": 5.505681775059481e-06, "loss": 1.1718, "step": 12280 }, { "epoch": 1.98, "grad_norm": 10.260027885437012, "learning_rate": 5.50412927008936e-06, "loss": 1.0424, "step": 12281 }, { "epoch": 1.98, "grad_norm": 11.589995384216309, "learning_rate": 5.5025769009220954e-06, "loss": 1.243, "step": 12282 }, { "epoch": 1.98, "grad_norm": 11.238142013549805, "learning_rate": 5.501024667604575e-06, "loss": 1.0648, "step": 12283 }, { "epoch": 1.98, "grad_norm": 11.196178436279297, "learning_rate": 5.499472570183677e-06, "loss": 1.0829, "step": 12284 }, { "epoch": 1.98, "grad_norm": 13.325652122497559, "learning_rate": 5.497920608706302e-06, "loss": 0.9624, "step": 12285 }, { "epoch": 1.98, "grad_norm": 8.665038108825684, "learning_rate": 5.496368783219319e-06, "loss": 1.0806, "step": 12286 }, { "epoch": 1.98, "grad_norm": 8.349264144897461, "learning_rate": 5.494817093769601e-06, "loss": 0.881, "step": 12287 }, { "epoch": 1.98, "grad_norm": 10.0282564163208, "learning_rate": 5.493265540404023e-06, "loss": 0.9318, "step": 12288 }, { "epoch": 1.98, "grad_norm": 9.352577209472656, "learning_rate": 5.491714123169447e-06, "loss": 0.9784, "step": 12289 }, { "epoch": 1.98, "grad_norm": 9.368733406066895, "learning_rate": 5.490162842112741e-06, "loss": 1.0636, "step": 12290 }, { "epoch": 1.98, "grad_norm": 10.739365577697754, "learning_rate": 5.48861169728076e-06, "loss": 1.3011, "step": 12291 }, { "epoch": 1.98, "grad_norm": 9.6536865234375, "learning_rate": 5.487060688720353e-06, "loss": 1.0262, "step": 12292 }, { "epoch": 1.98, "grad_norm": 9.760369300842285, "learning_rate": 5.485509816478381e-06, "loss": 0.9962, "step": 12293 }, { "epoch": 1.98, "grad_norm": 9.371121406555176, "learning_rate": 5.4839590806016815e-06, "loss": 1.0533, "step": 12294 }, { "epoch": 1.98, "grad_norm": 9.422480583190918, "learning_rate": 5.482408481137097e-06, "loss": 0.9022, "step": 12295 }, { "epoch": 1.98, "grad_norm": 11.21680736541748, "learning_rate": 5.48085801813147e-06, "loss": 1.1764, "step": 12296 }, { "epoch": 1.98, "grad_norm": 9.09838581085205, "learning_rate": 5.479307691631626e-06, "loss": 1.16, "step": 12297 }, { "epoch": 1.98, "grad_norm": 10.435876846313477, "learning_rate": 5.477757501684402e-06, "loss": 1.2782, "step": 12298 }, { "epoch": 1.98, "grad_norm": 10.423819541931152, "learning_rate": 5.476207448336621e-06, "loss": 1.1987, "step": 12299 }, { "epoch": 1.98, "grad_norm": 8.804409980773926, "learning_rate": 5.474657531635099e-06, "loss": 0.8785, "step": 12300 }, { "epoch": 1.98, "grad_norm": 9.874476432800293, "learning_rate": 5.473107751626663e-06, "loss": 1.1836, "step": 12301 }, { "epoch": 1.98, "grad_norm": 11.09592056274414, "learning_rate": 5.471558108358115e-06, "loss": 1.3562, "step": 12302 }, { "epoch": 1.98, "grad_norm": 10.359063148498535, "learning_rate": 5.470008601876272e-06, "loss": 0.8685, "step": 12303 }, { "epoch": 1.98, "grad_norm": 10.889838218688965, "learning_rate": 5.468459232227933e-06, "loss": 1.1094, "step": 12304 }, { "epoch": 1.98, "grad_norm": 7.224134922027588, "learning_rate": 5.466909999459904e-06, "loss": 0.8912, "step": 12305 }, { "epoch": 1.98, "grad_norm": 10.055533409118652, "learning_rate": 5.465360903618975e-06, "loss": 1.0345, "step": 12306 }, { "epoch": 1.98, "grad_norm": 8.569899559020996, "learning_rate": 5.463811944751944e-06, "loss": 1.0138, "step": 12307 }, { "epoch": 1.98, "grad_norm": 8.408988952636719, "learning_rate": 5.4622631229055935e-06, "loss": 0.9075, "step": 12308 }, { "epoch": 1.98, "grad_norm": 9.530165672302246, "learning_rate": 5.4607144381267154e-06, "loss": 0.8618, "step": 12309 }, { "epoch": 1.98, "grad_norm": 9.273575782775879, "learning_rate": 5.4591658904620835e-06, "loss": 1.0691, "step": 12310 }, { "epoch": 1.98, "grad_norm": 8.751155853271484, "learning_rate": 5.45761747995847e-06, "loss": 0.796, "step": 12311 }, { "epoch": 1.98, "grad_norm": 10.815272331237793, "learning_rate": 5.4560692066626555e-06, "loss": 1.3154, "step": 12312 }, { "epoch": 1.98, "grad_norm": 9.721183776855469, "learning_rate": 5.454521070621399e-06, "loss": 1.083, "step": 12313 }, { "epoch": 1.98, "grad_norm": 10.309368133544922, "learning_rate": 5.452973071881471e-06, "loss": 1.0346, "step": 12314 }, { "epoch": 1.98, "grad_norm": 9.676621437072754, "learning_rate": 5.451425210489627e-06, "loss": 0.7478, "step": 12315 }, { "epoch": 1.98, "grad_norm": 9.56501293182373, "learning_rate": 5.449877486492617e-06, "loss": 0.8449, "step": 12316 }, { "epoch": 1.98, "grad_norm": 9.385271072387695, "learning_rate": 5.4483298999372014e-06, "loss": 1.3205, "step": 12317 }, { "epoch": 1.98, "grad_norm": 9.546551704406738, "learning_rate": 5.446782450870122e-06, "loss": 1.0205, "step": 12318 }, { "epoch": 1.98, "grad_norm": 8.808656692504883, "learning_rate": 5.4452351393381166e-06, "loss": 0.9649, "step": 12319 }, { "epoch": 1.98, "grad_norm": 11.9871187210083, "learning_rate": 5.443687965387933e-06, "loss": 1.0044, "step": 12320 }, { "epoch": 1.98, "grad_norm": 11.318099021911621, "learning_rate": 5.442140929066295e-06, "loss": 0.8346, "step": 12321 }, { "epoch": 1.98, "grad_norm": 9.536417961120605, "learning_rate": 5.4405940304199435e-06, "loss": 0.8274, "step": 12322 }, { "epoch": 1.98, "grad_norm": 7.984951496124268, "learning_rate": 5.439047269495598e-06, "loss": 0.7487, "step": 12323 }, { "epoch": 1.98, "grad_norm": 9.772154808044434, "learning_rate": 5.437500646339974e-06, "loss": 1.2372, "step": 12324 }, { "epoch": 1.98, "grad_norm": 11.601224899291992, "learning_rate": 5.435954160999804e-06, "loss": 0.9374, "step": 12325 }, { "epoch": 1.98, "grad_norm": 12.52043342590332, "learning_rate": 5.434407813521791e-06, "loss": 1.2568, "step": 12326 }, { "epoch": 1.98, "grad_norm": 11.050657272338867, "learning_rate": 5.432861603952645e-06, "loss": 0.9665, "step": 12327 }, { "epoch": 1.98, "grad_norm": 12.873161315917969, "learning_rate": 5.431315532339076e-06, "loss": 1.2437, "step": 12328 }, { "epoch": 1.98, "grad_norm": 10.906844139099121, "learning_rate": 5.429769598727781e-06, "loss": 1.074, "step": 12329 }, { "epoch": 1.98, "grad_norm": 10.092970848083496, "learning_rate": 5.4282238031654535e-06, "loss": 1.0875, "step": 12330 }, { "epoch": 1.99, "grad_norm": 9.401339530944824, "learning_rate": 5.426678145698792e-06, "loss": 1.1077, "step": 12331 }, { "epoch": 1.99, "grad_norm": 9.599982261657715, "learning_rate": 5.4251326263744795e-06, "loss": 1.0572, "step": 12332 }, { "epoch": 1.99, "grad_norm": 10.615650177001953, "learning_rate": 5.423587245239207e-06, "loss": 1.0268, "step": 12333 }, { "epoch": 1.99, "grad_norm": 12.025517463684082, "learning_rate": 5.422042002339651e-06, "loss": 1.1291, "step": 12334 }, { "epoch": 1.99, "grad_norm": 8.361981391906738, "learning_rate": 5.420496897722483e-06, "loss": 1.0929, "step": 12335 }, { "epoch": 1.99, "grad_norm": 9.736005783081055, "learning_rate": 5.418951931434381e-06, "loss": 0.9732, "step": 12336 }, { "epoch": 1.99, "grad_norm": 8.226288795471191, "learning_rate": 5.417407103522005e-06, "loss": 1.0406, "step": 12337 }, { "epoch": 1.99, "grad_norm": 11.453705787658691, "learning_rate": 5.415862414032029e-06, "loss": 0.8214, "step": 12338 }, { "epoch": 1.99, "grad_norm": 8.58832836151123, "learning_rate": 5.414317863011107e-06, "loss": 0.9554, "step": 12339 }, { "epoch": 1.99, "grad_norm": 10.556119918823242, "learning_rate": 5.412773450505888e-06, "loss": 1.3459, "step": 12340 }, { "epoch": 1.99, "grad_norm": 9.16950511932373, "learning_rate": 5.411229176563032e-06, "loss": 0.8794, "step": 12341 }, { "epoch": 1.99, "grad_norm": 10.371865272521973, "learning_rate": 5.409685041229182e-06, "loss": 1.1314, "step": 12342 }, { "epoch": 1.99, "grad_norm": 11.422155380249023, "learning_rate": 5.408141044550974e-06, "loss": 0.9884, "step": 12343 }, { "epoch": 1.99, "grad_norm": 13.447515487670898, "learning_rate": 5.406597186575053e-06, "loss": 1.0962, "step": 12344 }, { "epoch": 1.99, "grad_norm": 10.505743980407715, "learning_rate": 5.405053467348057e-06, "loss": 1.1072, "step": 12345 }, { "epoch": 1.99, "grad_norm": 11.547542572021484, "learning_rate": 5.403509886916606e-06, "loss": 1.0587, "step": 12346 }, { "epoch": 1.99, "grad_norm": 12.145130157470703, "learning_rate": 5.401966445327334e-06, "loss": 1.1592, "step": 12347 }, { "epoch": 1.99, "grad_norm": 8.581976890563965, "learning_rate": 5.400423142626855e-06, "loss": 1.2166, "step": 12348 }, { "epoch": 1.99, "grad_norm": 9.95896053314209, "learning_rate": 5.398879978861793e-06, "loss": 0.9545, "step": 12349 }, { "epoch": 1.99, "grad_norm": 14.314779281616211, "learning_rate": 5.397336954078759e-06, "loss": 0.9065, "step": 12350 }, { "epoch": 1.99, "grad_norm": 9.517128944396973, "learning_rate": 5.395794068324356e-06, "loss": 1.1718, "step": 12351 }, { "epoch": 1.99, "grad_norm": 8.79114055633545, "learning_rate": 5.394251321645198e-06, "loss": 1.1235, "step": 12352 }, { "epoch": 1.99, "grad_norm": 11.976866722106934, "learning_rate": 5.39270871408788e-06, "loss": 1.024, "step": 12353 }, { "epoch": 1.99, "grad_norm": 9.30492877960205, "learning_rate": 5.391166245698996e-06, "loss": 0.9849, "step": 12354 }, { "epoch": 1.99, "grad_norm": 10.758397102355957, "learning_rate": 5.389623916525143e-06, "loss": 1.2033, "step": 12355 }, { "epoch": 1.99, "grad_norm": 10.968560218811035, "learning_rate": 5.388081726612904e-06, "loss": 1.0995, "step": 12356 }, { "epoch": 1.99, "grad_norm": 9.576171875, "learning_rate": 5.386539676008869e-06, "loss": 0.909, "step": 12357 }, { "epoch": 1.99, "grad_norm": 10.306168556213379, "learning_rate": 5.384997764759613e-06, "loss": 1.0959, "step": 12358 }, { "epoch": 1.99, "grad_norm": 10.005227088928223, "learning_rate": 5.383455992911708e-06, "loss": 0.8784, "step": 12359 }, { "epoch": 1.99, "grad_norm": 10.674871444702148, "learning_rate": 5.381914360511734e-06, "loss": 1.235, "step": 12360 }, { "epoch": 1.99, "grad_norm": 10.326966285705566, "learning_rate": 5.380372867606251e-06, "loss": 0.9578, "step": 12361 }, { "epoch": 1.99, "grad_norm": 10.005859375, "learning_rate": 5.37883151424182e-06, "loss": 1.0106, "step": 12362 }, { "epoch": 1.99, "grad_norm": 10.006570816040039, "learning_rate": 5.377290300465008e-06, "loss": 1.1107, "step": 12363 }, { "epoch": 1.99, "grad_norm": 9.353981971740723, "learning_rate": 5.3757492263223575e-06, "loss": 0.7395, "step": 12364 }, { "epoch": 1.99, "grad_norm": 9.033102035522461, "learning_rate": 5.374208291860424e-06, "loss": 1.2967, "step": 12365 }, { "epoch": 1.99, "grad_norm": 8.818419456481934, "learning_rate": 5.3726674971257585e-06, "loss": 1.0004, "step": 12366 }, { "epoch": 1.99, "grad_norm": 11.25857925415039, "learning_rate": 5.3711268421648935e-06, "loss": 0.8627, "step": 12367 }, { "epoch": 1.99, "grad_norm": 8.884232521057129, "learning_rate": 5.3695863270243745e-06, "loss": 0.7325, "step": 12368 }, { "epoch": 1.99, "grad_norm": 11.069210052490234, "learning_rate": 5.368045951750729e-06, "loss": 1.3762, "step": 12369 }, { "epoch": 1.99, "grad_norm": 10.068622589111328, "learning_rate": 5.366505716390484e-06, "loss": 1.087, "step": 12370 }, { "epoch": 1.99, "grad_norm": 9.434056282043457, "learning_rate": 5.36496562099017e-06, "loss": 1.1613, "step": 12371 }, { "epoch": 1.99, "grad_norm": 10.869549751281738, "learning_rate": 5.363425665596306e-06, "loss": 1.0361, "step": 12372 }, { "epoch": 1.99, "grad_norm": 9.287919044494629, "learning_rate": 5.361885850255402e-06, "loss": 1.1518, "step": 12373 }, { "epoch": 1.99, "grad_norm": 9.800503730773926, "learning_rate": 5.360346175013977e-06, "loss": 1.2663, "step": 12374 }, { "epoch": 1.99, "grad_norm": 11.905893325805664, "learning_rate": 5.358806639918534e-06, "loss": 1.0228, "step": 12375 }, { "epoch": 1.99, "grad_norm": 10.65897274017334, "learning_rate": 5.35726724501558e-06, "loss": 1.0488, "step": 12376 }, { "epoch": 1.99, "grad_norm": 8.370326042175293, "learning_rate": 5.355727990351614e-06, "loss": 0.9148, "step": 12377 }, { "epoch": 1.99, "grad_norm": 9.30842399597168, "learning_rate": 5.3541888759731256e-06, "loss": 0.8519, "step": 12378 }, { "epoch": 1.99, "grad_norm": 8.248221397399902, "learning_rate": 5.352649901926612e-06, "loss": 0.9509, "step": 12379 }, { "epoch": 1.99, "grad_norm": 11.812748908996582, "learning_rate": 5.3511110682585545e-06, "loss": 0.8929, "step": 12380 }, { "epoch": 1.99, "grad_norm": 9.487847328186035, "learning_rate": 5.349572375015441e-06, "loss": 1.1746, "step": 12381 }, { "epoch": 1.99, "grad_norm": 9.287680625915527, "learning_rate": 5.348033822243747e-06, "loss": 1.0303, "step": 12382 }, { "epoch": 1.99, "grad_norm": 11.401528358459473, "learning_rate": 5.34649540998994e-06, "loss": 1.0051, "step": 12383 }, { "epoch": 1.99, "grad_norm": 9.499245643615723, "learning_rate": 5.344957138300499e-06, "loss": 1.1723, "step": 12384 }, { "epoch": 1.99, "grad_norm": 11.352566719055176, "learning_rate": 5.343419007221884e-06, "loss": 0.9964, "step": 12385 }, { "epoch": 1.99, "grad_norm": 9.737066268920898, "learning_rate": 5.341881016800554e-06, "loss": 1.0253, "step": 12386 }, { "epoch": 1.99, "grad_norm": 11.181805610656738, "learning_rate": 5.340343167082974e-06, "loss": 0.9515, "step": 12387 }, { "epoch": 1.99, "grad_norm": 10.256671905517578, "learning_rate": 5.338805458115593e-06, "loss": 1.077, "step": 12388 }, { "epoch": 1.99, "grad_norm": 7.960834503173828, "learning_rate": 5.3372678899448504e-06, "loss": 0.9158, "step": 12389 }, { "epoch": 1.99, "grad_norm": 9.368620872497559, "learning_rate": 5.3357304626172034e-06, "loss": 0.8705, "step": 12390 }, { "epoch": 1.99, "grad_norm": 7.867173671722412, "learning_rate": 5.334193176179081e-06, "loss": 0.8604, "step": 12391 }, { "epoch": 1.99, "grad_norm": 10.881461143493652, "learning_rate": 5.332656030676928e-06, "loss": 0.683, "step": 12392 }, { "epoch": 2.0, "grad_norm": 10.571563720703125, "learning_rate": 5.331119026157172e-06, "loss": 0.9608, "step": 12393 }, { "epoch": 2.0, "grad_norm": 9.325986862182617, "learning_rate": 5.329582162666234e-06, "loss": 0.9394, "step": 12394 }, { "epoch": 2.0, "grad_norm": 10.907469749450684, "learning_rate": 5.328045440250544e-06, "loss": 0.7658, "step": 12395 }, { "epoch": 2.0, "grad_norm": 9.741559982299805, "learning_rate": 5.326508858956521e-06, "loss": 1.2027, "step": 12396 }, { "epoch": 2.0, "grad_norm": 12.235082626342773, "learning_rate": 5.324972418830572e-06, "loss": 1.0388, "step": 12397 }, { "epoch": 2.0, "grad_norm": 9.264439582824707, "learning_rate": 5.323436119919113e-06, "loss": 0.8813, "step": 12398 }, { "epoch": 2.0, "grad_norm": 9.273944854736328, "learning_rate": 5.321899962268545e-06, "loss": 0.7782, "step": 12399 }, { "epoch": 2.0, "grad_norm": 10.69243049621582, "learning_rate": 5.320363945925277e-06, "loss": 1.0425, "step": 12400 }, { "epoch": 2.0, "grad_norm": 10.706639289855957, "learning_rate": 5.3188280709357e-06, "loss": 1.0779, "step": 12401 }, { "epoch": 2.0, "grad_norm": 9.254609107971191, "learning_rate": 5.317292337346205e-06, "loss": 1.1757, "step": 12402 }, { "epoch": 2.0, "grad_norm": 11.630356788635254, "learning_rate": 5.315756745203188e-06, "loss": 1.0751, "step": 12403 }, { "epoch": 2.0, "grad_norm": 11.707691192626953, "learning_rate": 5.314221294553024e-06, "loss": 1.321, "step": 12404 }, { "epoch": 2.0, "grad_norm": 9.11882495880127, "learning_rate": 5.312685985442099e-06, "loss": 0.9872, "step": 12405 }, { "epoch": 2.0, "grad_norm": 10.152944564819336, "learning_rate": 5.311150817916791e-06, "loss": 1.0651, "step": 12406 }, { "epoch": 2.0, "grad_norm": 9.753883361816406, "learning_rate": 5.309615792023469e-06, "loss": 1.217, "step": 12407 }, { "epoch": 2.0, "grad_norm": 8.417499542236328, "learning_rate": 5.308080907808495e-06, "loss": 1.0099, "step": 12408 }, { "epoch": 2.0, "grad_norm": 10.702997207641602, "learning_rate": 5.30654616531824e-06, "loss": 0.9748, "step": 12409 }, { "epoch": 2.0, "grad_norm": 12.083967208862305, "learning_rate": 5.3050115645990565e-06, "loss": 1.0166, "step": 12410 }, { "epoch": 2.0, "grad_norm": 8.444317817687988, "learning_rate": 5.303477105697304e-06, "loss": 0.8759, "step": 12411 }, { "epoch": 2.0, "grad_norm": 11.931290626525879, "learning_rate": 5.301942788659331e-06, "loss": 1.1906, "step": 12412 }, { "epoch": 2.0, "grad_norm": 10.82311725616455, "learning_rate": 5.300408613531478e-06, "loss": 1.2027, "step": 12413 }, { "epoch": 2.0, "grad_norm": 11.414722442626953, "learning_rate": 5.298874580360094e-06, "loss": 1.177, "step": 12414 }, { "epoch": 2.0, "grad_norm": 9.331140518188477, "learning_rate": 5.297340689191509e-06, "loss": 0.9147, "step": 12415 }, { "epoch": 2.0, "grad_norm": 12.120293617248535, "learning_rate": 5.295806940072063e-06, "loss": 1.1257, "step": 12416 }, { "epoch": 2.0, "grad_norm": 8.97998046875, "learning_rate": 5.2942733330480835e-06, "loss": 1.0844, "step": 12417 }, { "epoch": 2.0, "grad_norm": 9.979399681091309, "learning_rate": 5.292739868165888e-06, "loss": 1.0334, "step": 12418 }, { "epoch": 2.0, "grad_norm": 11.77750015258789, "learning_rate": 5.291206545471806e-06, "loss": 1.2152, "step": 12419 }, { "epoch": 2.0, "grad_norm": 11.483024597167969, "learning_rate": 5.289673365012147e-06, "loss": 1.1446, "step": 12420 }, { "epoch": 2.0, "grad_norm": 13.250231742858887, "learning_rate": 5.28814032683322e-06, "loss": 1.4279, "step": 12421 }, { "epoch": 2.0, "grad_norm": 10.589608192443848, "learning_rate": 5.286607430981341e-06, "loss": 1.008, "step": 12422 }, { "epoch": 2.0, "grad_norm": 10.556191444396973, "learning_rate": 5.285074677502804e-06, "loss": 1.2612, "step": 12423 }, { "epoch": 2.0, "grad_norm": 8.62913990020752, "learning_rate": 5.283542066443914e-06, "loss": 0.9518, "step": 12424 }, { "epoch": 2.0, "grad_norm": 9.676910400390625, "learning_rate": 5.282009597850961e-06, "loss": 0.8404, "step": 12425 }, { "epoch": 2.0, "grad_norm": 7.646076679229736, "learning_rate": 5.280477271770235e-06, "loss": 0.6053, "step": 12426 }, { "epoch": 2.0, "grad_norm": 9.830353736877441, "learning_rate": 5.278945088248029e-06, "loss": 0.8121, "step": 12427 }, { "epoch": 2.0, "grad_norm": 7.278806209564209, "learning_rate": 5.277413047330617e-06, "loss": 0.6112, "step": 12428 }, { "epoch": 2.0, "grad_norm": 8.089902877807617, "learning_rate": 5.2758811490642744e-06, "loss": 0.7622, "step": 12429 }, { "epoch": 2.0, "grad_norm": 9.254305839538574, "learning_rate": 5.274349393495282e-06, "loss": 0.8747, "step": 12430 }, { "epoch": 2.0, "grad_norm": 9.129323959350586, "learning_rate": 5.272817780669903e-06, "loss": 0.9159, "step": 12431 }, { "epoch": 2.0, "grad_norm": 7.575631141662598, "learning_rate": 5.271286310634398e-06, "loss": 0.8166, "step": 12432 }, { "epoch": 2.0, "grad_norm": 9.061738967895508, "learning_rate": 5.269754983435033e-06, "loss": 0.8772, "step": 12433 }, { "epoch": 2.0, "grad_norm": 9.337082862854004, "learning_rate": 5.26822379911806e-06, "loss": 0.7962, "step": 12434 }, { "epoch": 2.0, "grad_norm": 9.124088287353516, "learning_rate": 5.266692757729735e-06, "loss": 0.8685, "step": 12435 }, { "epoch": 2.0, "grad_norm": 12.41002082824707, "learning_rate": 5.265161859316299e-06, "loss": 0.8542, "step": 12436 }, { "epoch": 2.0, "grad_norm": 9.106858253479004, "learning_rate": 5.263631103923994e-06, "loss": 0.8352, "step": 12437 }, { "epoch": 2.0, "grad_norm": 7.846491813659668, "learning_rate": 5.262100491599066e-06, "loss": 0.6308, "step": 12438 }, { "epoch": 2.0, "grad_norm": 8.877237319946289, "learning_rate": 5.260570022387743e-06, "loss": 0.761, "step": 12439 }, { "epoch": 2.0, "grad_norm": 8.187100410461426, "learning_rate": 5.25903969633625e-06, "loss": 0.5429, "step": 12440 }, { "epoch": 2.0, "grad_norm": 8.372653007507324, "learning_rate": 5.2575095134908236e-06, "loss": 0.636, "step": 12441 }, { "epoch": 2.0, "grad_norm": 10.333099365234375, "learning_rate": 5.255979473897673e-06, "loss": 0.5983, "step": 12442 }, { "epoch": 2.0, "grad_norm": 9.776063919067383, "learning_rate": 5.2544495776030246e-06, "loss": 0.7862, "step": 12443 }, { "epoch": 2.0, "grad_norm": 9.8468017578125, "learning_rate": 5.252919824653088e-06, "loss": 0.7895, "step": 12444 }, { "epoch": 2.0, "grad_norm": 9.609167098999023, "learning_rate": 5.251390215094065e-06, "loss": 0.7334, "step": 12445 }, { "epoch": 2.0, "grad_norm": 10.345320701599121, "learning_rate": 5.249860748972163e-06, "loss": 0.7081, "step": 12446 }, { "epoch": 2.0, "grad_norm": 12.419885635375977, "learning_rate": 5.248331426333586e-06, "loss": 0.9403, "step": 12447 }, { "epoch": 2.0, "grad_norm": 10.973722457885742, "learning_rate": 5.246802247224522e-06, "loss": 0.7488, "step": 12448 }, { "epoch": 2.0, "grad_norm": 13.575888633728027, "learning_rate": 5.245273211691168e-06, "loss": 0.7136, "step": 12449 }, { "epoch": 2.0, "grad_norm": 13.162348747253418, "learning_rate": 5.243744319779706e-06, "loss": 0.6248, "step": 12450 }, { "epoch": 2.0, "grad_norm": 12.614338874816895, "learning_rate": 5.242215571536315e-06, "loss": 0.7648, "step": 12451 }, { "epoch": 2.0, "grad_norm": 10.554455757141113, "learning_rate": 5.2406869670071815e-06, "loss": 0.5373, "step": 12452 }, { "epoch": 2.0, "grad_norm": 11.358837127685547, "learning_rate": 5.239158506238468e-06, "loss": 0.6955, "step": 12453 }, { "epoch": 2.0, "grad_norm": 10.061408042907715, "learning_rate": 5.237630189276354e-06, "loss": 0.7012, "step": 12454 }, { "epoch": 2.0, "grad_norm": 11.772706985473633, "learning_rate": 5.236102016166999e-06, "loss": 0.6578, "step": 12455 }, { "epoch": 2.01, "grad_norm": 10.063806533813477, "learning_rate": 5.234573986956557e-06, "loss": 0.5388, "step": 12456 }, { "epoch": 2.01, "grad_norm": 13.990145683288574, "learning_rate": 5.233046101691196e-06, "loss": 0.6798, "step": 12457 }, { "epoch": 2.01, "grad_norm": 11.734992980957031, "learning_rate": 5.231518360417057e-06, "loss": 0.7206, "step": 12458 }, { "epoch": 2.01, "grad_norm": 9.863633155822754, "learning_rate": 5.229990763180295e-06, "loss": 0.7495, "step": 12459 }, { "epoch": 2.01, "grad_norm": 9.373311042785645, "learning_rate": 5.22846331002705e-06, "loss": 0.5589, "step": 12460 }, { "epoch": 2.01, "grad_norm": 10.808032989501953, "learning_rate": 5.226936001003455e-06, "loss": 0.7267, "step": 12461 }, { "epoch": 2.01, "grad_norm": 8.626222610473633, "learning_rate": 5.2254088361556544e-06, "loss": 0.6232, "step": 12462 }, { "epoch": 2.01, "grad_norm": 9.708314895629883, "learning_rate": 5.223881815529772e-06, "loss": 0.7872, "step": 12463 }, { "epoch": 2.01, "grad_norm": 9.465377807617188, "learning_rate": 5.222354939171929e-06, "loss": 0.5136, "step": 12464 }, { "epoch": 2.01, "grad_norm": 10.838030815124512, "learning_rate": 5.220828207128256e-06, "loss": 0.5118, "step": 12465 }, { "epoch": 2.01, "grad_norm": 11.821063995361328, "learning_rate": 5.21930161944486e-06, "loss": 0.6858, "step": 12466 }, { "epoch": 2.01, "grad_norm": 12.674602508544922, "learning_rate": 5.217775176167858e-06, "loss": 0.6291, "step": 12467 }, { "epoch": 2.01, "grad_norm": 10.83066463470459, "learning_rate": 5.216248877343362e-06, "loss": 0.6585, "step": 12468 }, { "epoch": 2.01, "grad_norm": 12.566767692565918, "learning_rate": 5.214722723017468e-06, "loss": 0.6587, "step": 12469 }, { "epoch": 2.01, "grad_norm": 10.860517501831055, "learning_rate": 5.213196713236283e-06, "loss": 0.8187, "step": 12470 }, { "epoch": 2.01, "grad_norm": 9.484622955322266, "learning_rate": 5.211670848045897e-06, "loss": 0.949, "step": 12471 }, { "epoch": 2.01, "grad_norm": 10.545905113220215, "learning_rate": 5.2101451274923966e-06, "loss": 0.496, "step": 12472 }, { "epoch": 2.01, "grad_norm": 8.568082809448242, "learning_rate": 5.208619551621877e-06, "loss": 0.5996, "step": 12473 }, { "epoch": 2.01, "grad_norm": 11.490488052368164, "learning_rate": 5.207094120480416e-06, "loss": 0.7943, "step": 12474 }, { "epoch": 2.01, "grad_norm": 11.064520835876465, "learning_rate": 5.205568834114085e-06, "loss": 0.6111, "step": 12475 }, { "epoch": 2.01, "grad_norm": 11.588642120361328, "learning_rate": 5.2040436925689674e-06, "loss": 0.5486, "step": 12476 }, { "epoch": 2.01, "grad_norm": 12.491960525512695, "learning_rate": 5.202518695891123e-06, "loss": 0.8047, "step": 12477 }, { "epoch": 2.01, "grad_norm": 10.824260711669922, "learning_rate": 5.200993844126624e-06, "loss": 0.9011, "step": 12478 }, { "epoch": 2.01, "grad_norm": 11.654483795166016, "learning_rate": 5.1994691373215245e-06, "loss": 0.6434, "step": 12479 }, { "epoch": 2.01, "grad_norm": 11.841235160827637, "learning_rate": 5.197944575521879e-06, "loss": 0.7345, "step": 12480 }, { "epoch": 2.01, "grad_norm": 10.0422945022583, "learning_rate": 5.196420158773745e-06, "loss": 0.5563, "step": 12481 }, { "epoch": 2.01, "grad_norm": 11.320544242858887, "learning_rate": 5.194895887123161e-06, "loss": 0.5202, "step": 12482 }, { "epoch": 2.01, "grad_norm": 10.326213836669922, "learning_rate": 5.1933717606161785e-06, "loss": 0.5149, "step": 12483 }, { "epoch": 2.01, "grad_norm": 9.938033103942871, "learning_rate": 5.19184777929883e-06, "loss": 0.5639, "step": 12484 }, { "epoch": 2.01, "grad_norm": 9.591961860656738, "learning_rate": 5.190323943217146e-06, "loss": 0.5467, "step": 12485 }, { "epoch": 2.01, "grad_norm": 11.095175743103027, "learning_rate": 5.1888002524171585e-06, "loss": 0.8321, "step": 12486 }, { "epoch": 2.01, "grad_norm": 10.983893394470215, "learning_rate": 5.187276706944899e-06, "loss": 0.6916, "step": 12487 }, { "epoch": 2.01, "grad_norm": 12.497862815856934, "learning_rate": 5.1857533068463775e-06, "loss": 0.6715, "step": 12488 }, { "epoch": 2.01, "grad_norm": 11.767560005187988, "learning_rate": 5.184230052167618e-06, "loss": 0.8209, "step": 12489 }, { "epoch": 2.01, "grad_norm": 11.654065132141113, "learning_rate": 5.182706942954629e-06, "loss": 0.6622, "step": 12490 }, { "epoch": 2.01, "grad_norm": 12.558140754699707, "learning_rate": 5.1811839792534145e-06, "loss": 0.7523, "step": 12491 }, { "epoch": 2.01, "grad_norm": 10.946495056152344, "learning_rate": 5.1796611611099844e-06, "loss": 0.6426, "step": 12492 }, { "epoch": 2.01, "grad_norm": 12.20384693145752, "learning_rate": 5.178138488570329e-06, "loss": 0.8539, "step": 12493 }, { "epoch": 2.01, "grad_norm": 12.475811958312988, "learning_rate": 5.17661596168045e-06, "loss": 0.7342, "step": 12494 }, { "epoch": 2.01, "grad_norm": 10.434700965881348, "learning_rate": 5.175093580486335e-06, "loss": 0.7711, "step": 12495 }, { "epoch": 2.01, "grad_norm": 12.729820251464844, "learning_rate": 5.1735713450339605e-06, "loss": 0.6139, "step": 12496 }, { "epoch": 2.01, "grad_norm": 9.76039981842041, "learning_rate": 5.172049255369323e-06, "loss": 0.7623, "step": 12497 }, { "epoch": 2.01, "grad_norm": 9.698270797729492, "learning_rate": 5.170527311538387e-06, "loss": 0.9422, "step": 12498 }, { "epoch": 2.01, "grad_norm": 12.957048416137695, "learning_rate": 5.169005513587127e-06, "loss": 0.8307, "step": 12499 }, { "epoch": 2.01, "grad_norm": 10.588516235351562, "learning_rate": 5.167483861561513e-06, "loss": 0.845, "step": 12500 }, { "epoch": 2.01, "grad_norm": 9.901951789855957, "learning_rate": 5.165962355507503e-06, "loss": 0.6385, "step": 12501 }, { "epoch": 2.01, "grad_norm": 9.781957626342773, "learning_rate": 5.164440995471066e-06, "loss": 0.722, "step": 12502 }, { "epoch": 2.01, "grad_norm": 13.01928997039795, "learning_rate": 5.162919781498149e-06, "loss": 0.6665, "step": 12503 }, { "epoch": 2.01, "grad_norm": 11.124961853027344, "learning_rate": 5.1613987136347e-06, "loss": 0.6822, "step": 12504 }, { "epoch": 2.01, "grad_norm": 10.106086730957031, "learning_rate": 5.159877791926671e-06, "loss": 0.6442, "step": 12505 }, { "epoch": 2.01, "grad_norm": 12.685047149658203, "learning_rate": 5.158357016419995e-06, "loss": 0.8161, "step": 12506 }, { "epoch": 2.01, "grad_norm": 13.166938781738281, "learning_rate": 5.1568363871606155e-06, "loss": 0.6894, "step": 12507 }, { "epoch": 2.01, "grad_norm": 9.853486061096191, "learning_rate": 5.155315904194465e-06, "loss": 0.7654, "step": 12508 }, { "epoch": 2.01, "grad_norm": 9.901727676391602, "learning_rate": 5.15379556756747e-06, "loss": 0.7504, "step": 12509 }, { "epoch": 2.01, "grad_norm": 11.87558364868164, "learning_rate": 5.152275377325549e-06, "loss": 0.7585, "step": 12510 }, { "epoch": 2.01, "grad_norm": 10.65035629272461, "learning_rate": 5.15075533351463e-06, "loss": 0.8828, "step": 12511 }, { "epoch": 2.01, "grad_norm": 10.028762817382812, "learning_rate": 5.14923543618062e-06, "loss": 0.6598, "step": 12512 }, { "epoch": 2.01, "grad_norm": 11.068658828735352, "learning_rate": 5.147715685369434e-06, "loss": 0.6573, "step": 12513 }, { "epoch": 2.01, "grad_norm": 12.559087753295898, "learning_rate": 5.1461960811269775e-06, "loss": 0.7427, "step": 12514 }, { "epoch": 2.01, "grad_norm": 12.45153522491455, "learning_rate": 5.144676623499148e-06, "loss": 0.6952, "step": 12515 }, { "epoch": 2.01, "grad_norm": 10.885061264038086, "learning_rate": 5.1431573125318455e-06, "loss": 0.6513, "step": 12516 }, { "epoch": 2.01, "grad_norm": 12.152203559875488, "learning_rate": 5.141638148270964e-06, "loss": 0.6577, "step": 12517 }, { "epoch": 2.02, "grad_norm": 12.6124849319458, "learning_rate": 5.1401191307623846e-06, "loss": 0.8135, "step": 12518 }, { "epoch": 2.02, "grad_norm": 12.171961784362793, "learning_rate": 5.138600260052e-06, "loss": 0.7969, "step": 12519 }, { "epoch": 2.02, "grad_norm": 11.557629585266113, "learning_rate": 5.137081536185681e-06, "loss": 0.6152, "step": 12520 }, { "epoch": 2.02, "grad_norm": 13.792393684387207, "learning_rate": 5.135562959209311e-06, "loss": 0.9034, "step": 12521 }, { "epoch": 2.02, "grad_norm": 11.098159790039062, "learning_rate": 5.134044529168756e-06, "loss": 0.947, "step": 12522 }, { "epoch": 2.02, "grad_norm": 10.566754341125488, "learning_rate": 5.132526246109877e-06, "loss": 0.7202, "step": 12523 }, { "epoch": 2.02, "grad_norm": 11.185636520385742, "learning_rate": 5.131008110078543e-06, "loss": 0.6843, "step": 12524 }, { "epoch": 2.02, "grad_norm": 10.467916488647461, "learning_rate": 5.129490121120606e-06, "loss": 0.6356, "step": 12525 }, { "epoch": 2.02, "grad_norm": 12.764823913574219, "learning_rate": 5.127972279281925e-06, "loss": 0.4701, "step": 12526 }, { "epoch": 2.02, "grad_norm": 11.91551685333252, "learning_rate": 5.1264545846083394e-06, "loss": 0.8146, "step": 12527 }, { "epoch": 2.02, "grad_norm": 12.826473236083984, "learning_rate": 5.124937037145701e-06, "loss": 0.6842, "step": 12528 }, { "epoch": 2.02, "grad_norm": 10.912027359008789, "learning_rate": 5.123419636939841e-06, "loss": 0.7483, "step": 12529 }, { "epoch": 2.02, "grad_norm": 11.223811149597168, "learning_rate": 5.121902384036605e-06, "loss": 0.5413, "step": 12530 }, { "epoch": 2.02, "grad_norm": 11.73653507232666, "learning_rate": 5.120385278481811e-06, "loss": 0.7947, "step": 12531 }, { "epoch": 2.02, "grad_norm": 11.532663345336914, "learning_rate": 5.118868320321295e-06, "loss": 0.55, "step": 12532 }, { "epoch": 2.02, "grad_norm": 12.468786239624023, "learning_rate": 5.117351509600874e-06, "loss": 0.5833, "step": 12533 }, { "epoch": 2.02, "grad_norm": 10.7660493850708, "learning_rate": 5.1158348463663635e-06, "loss": 0.6521, "step": 12534 }, { "epoch": 2.02, "grad_norm": 10.995595932006836, "learning_rate": 5.114318330663579e-06, "loss": 0.6631, "step": 12535 }, { "epoch": 2.02, "grad_norm": 11.062714576721191, "learning_rate": 5.1128019625383256e-06, "loss": 0.6186, "step": 12536 }, { "epoch": 2.02, "grad_norm": 10.740823745727539, "learning_rate": 5.111285742036411e-06, "loss": 0.7472, "step": 12537 }, { "epoch": 2.02, "grad_norm": 12.139381408691406, "learning_rate": 5.109769669203632e-06, "loss": 0.6875, "step": 12538 }, { "epoch": 2.02, "grad_norm": 11.101632118225098, "learning_rate": 5.1082537440857805e-06, "loss": 0.7816, "step": 12539 }, { "epoch": 2.02, "grad_norm": 9.178001403808594, "learning_rate": 5.106737966728652e-06, "loss": 0.6889, "step": 12540 }, { "epoch": 2.02, "grad_norm": 13.190622329711914, "learning_rate": 5.105222337178031e-06, "loss": 0.6906, "step": 12541 }, { "epoch": 2.02, "grad_norm": 10.731115341186523, "learning_rate": 5.103706855479693e-06, "loss": 0.6838, "step": 12542 }, { "epoch": 2.02, "grad_norm": 13.04898452758789, "learning_rate": 5.102191521679425e-06, "loss": 0.6565, "step": 12543 }, { "epoch": 2.02, "grad_norm": 9.838912010192871, "learning_rate": 5.100676335822989e-06, "loss": 0.6222, "step": 12544 }, { "epoch": 2.02, "grad_norm": 10.886514663696289, "learning_rate": 5.099161297956161e-06, "loss": 0.6453, "step": 12545 }, { "epoch": 2.02, "grad_norm": 10.616896629333496, "learning_rate": 5.097646408124701e-06, "loss": 0.9468, "step": 12546 }, { "epoch": 2.02, "grad_norm": 11.316773414611816, "learning_rate": 5.096131666374361e-06, "loss": 0.9156, "step": 12547 }, { "epoch": 2.02, "grad_norm": 10.873462677001953, "learning_rate": 5.09461707275091e-06, "loss": 0.6966, "step": 12548 }, { "epoch": 2.02, "grad_norm": 9.666352272033691, "learning_rate": 5.09310262730009e-06, "loss": 0.5894, "step": 12549 }, { "epoch": 2.02, "grad_norm": 10.891236305236816, "learning_rate": 5.091588330067645e-06, "loss": 0.7714, "step": 12550 }, { "epoch": 2.02, "grad_norm": 12.439064979553223, "learning_rate": 5.0900741810993206e-06, "loss": 0.8896, "step": 12551 }, { "epoch": 2.02, "grad_norm": 10.89145278930664, "learning_rate": 5.088560180440852e-06, "loss": 0.7734, "step": 12552 }, { "epoch": 2.02, "grad_norm": 10.814911842346191, "learning_rate": 5.087046328137966e-06, "loss": 0.7526, "step": 12553 }, { "epoch": 2.02, "grad_norm": 11.078155517578125, "learning_rate": 5.085532624236398e-06, "loss": 0.6945, "step": 12554 }, { "epoch": 2.02, "grad_norm": 10.715252876281738, "learning_rate": 5.084019068781865e-06, "loss": 0.5829, "step": 12555 }, { "epoch": 2.02, "grad_norm": 9.752778053283691, "learning_rate": 5.082505661820091e-06, "loss": 0.7789, "step": 12556 }, { "epoch": 2.02, "grad_norm": 11.577156066894531, "learning_rate": 5.080992403396787e-06, "loss": 0.7742, "step": 12557 }, { "epoch": 2.02, "grad_norm": 12.469882011413574, "learning_rate": 5.079479293557661e-06, "loss": 0.7723, "step": 12558 }, { "epoch": 2.02, "grad_norm": 11.234755516052246, "learning_rate": 5.077966332348423e-06, "loss": 0.8901, "step": 12559 }, { "epoch": 2.02, "grad_norm": 10.811241149902344, "learning_rate": 5.076453519814766e-06, "loss": 0.7894, "step": 12560 }, { "epoch": 2.02, "grad_norm": 11.416762351989746, "learning_rate": 5.074940856002397e-06, "loss": 0.6422, "step": 12561 }, { "epoch": 2.02, "grad_norm": 13.804884910583496, "learning_rate": 5.073428340957001e-06, "loss": 0.8585, "step": 12562 }, { "epoch": 2.02, "grad_norm": 10.179993629455566, "learning_rate": 5.071915974724263e-06, "loss": 0.6348, "step": 12563 }, { "epoch": 2.02, "grad_norm": 12.181130409240723, "learning_rate": 5.070403757349872e-06, "loss": 0.7503, "step": 12564 }, { "epoch": 2.02, "grad_norm": 11.087581634521484, "learning_rate": 5.068891688879503e-06, "loss": 0.831, "step": 12565 }, { "epoch": 2.02, "grad_norm": 11.190536499023438, "learning_rate": 5.067379769358827e-06, "loss": 0.5358, "step": 12566 }, { "epoch": 2.02, "grad_norm": 10.796066284179688, "learning_rate": 5.065867998833515e-06, "loss": 0.685, "step": 12567 }, { "epoch": 2.02, "grad_norm": 10.238362312316895, "learning_rate": 5.064356377349239e-06, "loss": 0.6918, "step": 12568 }, { "epoch": 2.02, "grad_norm": 11.237403869628906, "learning_rate": 5.062844904951647e-06, "loss": 0.7792, "step": 12569 }, { "epoch": 2.02, "grad_norm": 10.229557991027832, "learning_rate": 5.061333581686405e-06, "loss": 0.8969, "step": 12570 }, { "epoch": 2.02, "grad_norm": 10.281889915466309, "learning_rate": 5.059822407599158e-06, "loss": 0.6148, "step": 12571 }, { "epoch": 2.02, "grad_norm": 12.64513874053955, "learning_rate": 5.058311382735557e-06, "loss": 0.6097, "step": 12572 }, { "epoch": 2.02, "grad_norm": 12.884180068969727, "learning_rate": 5.056800507141242e-06, "loss": 0.7271, "step": 12573 }, { "epoch": 2.02, "grad_norm": 11.670076370239258, "learning_rate": 5.055289780861846e-06, "loss": 0.8235, "step": 12574 }, { "epoch": 2.02, "grad_norm": 12.616667747497559, "learning_rate": 5.053779203943013e-06, "loss": 0.6577, "step": 12575 }, { "epoch": 2.02, "grad_norm": 11.482131958007812, "learning_rate": 5.052268776430363e-06, "loss": 0.7443, "step": 12576 }, { "epoch": 2.02, "grad_norm": 10.142080307006836, "learning_rate": 5.0507584983695215e-06, "loss": 0.6514, "step": 12577 }, { "epoch": 2.02, "grad_norm": 8.886216163635254, "learning_rate": 5.04924836980611e-06, "loss": 0.6312, "step": 12578 }, { "epoch": 2.02, "grad_norm": 12.101327896118164, "learning_rate": 5.047738390785741e-06, "loss": 0.7175, "step": 12579 }, { "epoch": 2.03, "grad_norm": 11.936960220336914, "learning_rate": 5.046228561354032e-06, "loss": 0.6652, "step": 12580 }, { "epoch": 2.03, "grad_norm": 11.457783699035645, "learning_rate": 5.044718881556583e-06, "loss": 0.5967, "step": 12581 }, { "epoch": 2.03, "grad_norm": 10.84190845489502, "learning_rate": 5.0432093514389934e-06, "loss": 0.7011, "step": 12582 }, { "epoch": 2.03, "grad_norm": 10.084906578063965, "learning_rate": 5.0416999710468675e-06, "loss": 0.6357, "step": 12583 }, { "epoch": 2.03, "grad_norm": 11.994935989379883, "learning_rate": 5.0401907404257945e-06, "loss": 0.8139, "step": 12584 }, { "epoch": 2.03, "grad_norm": 12.368054389953613, "learning_rate": 5.0386816596213565e-06, "loss": 0.8057, "step": 12585 }, { "epoch": 2.03, "grad_norm": 11.889386177062988, "learning_rate": 5.037172728679147e-06, "loss": 0.6094, "step": 12586 }, { "epoch": 2.03, "grad_norm": 11.624444961547852, "learning_rate": 5.0356639476447374e-06, "loss": 0.7867, "step": 12587 }, { "epoch": 2.03, "grad_norm": 9.319465637207031, "learning_rate": 5.034155316563705e-06, "loss": 0.7774, "step": 12588 }, { "epoch": 2.03, "grad_norm": 11.731148719787598, "learning_rate": 5.0326468354816226e-06, "loss": 0.6768, "step": 12589 }, { "epoch": 2.03, "grad_norm": 13.473905563354492, "learning_rate": 5.031138504444051e-06, "loss": 0.7826, "step": 12590 }, { "epoch": 2.03, "grad_norm": 11.301259994506836, "learning_rate": 5.0296303234965545e-06, "loss": 0.7348, "step": 12591 }, { "epoch": 2.03, "grad_norm": 9.912999153137207, "learning_rate": 5.028122292684688e-06, "loss": 0.6701, "step": 12592 }, { "epoch": 2.03, "grad_norm": 10.328194618225098, "learning_rate": 5.0266144120540005e-06, "loss": 0.5763, "step": 12593 }, { "epoch": 2.03, "grad_norm": 10.605493545532227, "learning_rate": 5.025106681650045e-06, "loss": 0.6637, "step": 12594 }, { "epoch": 2.03, "grad_norm": 10.029848098754883, "learning_rate": 5.02359910151836e-06, "loss": 0.5552, "step": 12595 }, { "epoch": 2.03, "grad_norm": 11.340587615966797, "learning_rate": 5.022091671704482e-06, "loss": 0.7269, "step": 12596 }, { "epoch": 2.03, "grad_norm": 12.704208374023438, "learning_rate": 5.020584392253951e-06, "loss": 0.6544, "step": 12597 }, { "epoch": 2.03, "grad_norm": 10.270509719848633, "learning_rate": 5.019077263212289e-06, "loss": 0.6093, "step": 12598 }, { "epoch": 2.03, "grad_norm": 9.185380935668945, "learning_rate": 5.017570284625026e-06, "loss": 0.6893, "step": 12599 }, { "epoch": 2.03, "grad_norm": 10.511547088623047, "learning_rate": 5.016063456537681e-06, "loss": 0.8868, "step": 12600 }, { "epoch": 2.03, "grad_norm": 12.866063117980957, "learning_rate": 5.014556778995765e-06, "loss": 0.8476, "step": 12601 }, { "epoch": 2.03, "grad_norm": 10.711024284362793, "learning_rate": 5.013050252044795e-06, "loss": 0.6673, "step": 12602 }, { "epoch": 2.03, "grad_norm": 12.156463623046875, "learning_rate": 5.011543875730271e-06, "loss": 0.6954, "step": 12603 }, { "epoch": 2.03, "grad_norm": 9.866658210754395, "learning_rate": 5.010037650097702e-06, "loss": 0.4965, "step": 12604 }, { "epoch": 2.03, "grad_norm": 9.886350631713867, "learning_rate": 5.0085315751925825e-06, "loss": 0.6026, "step": 12605 }, { "epoch": 2.03, "grad_norm": 13.234502792358398, "learning_rate": 5.0070256510604e-06, "loss": 0.7636, "step": 12606 }, { "epoch": 2.03, "grad_norm": 11.425684928894043, "learning_rate": 5.00551987774665e-06, "loss": 0.4707, "step": 12607 }, { "epoch": 2.03, "grad_norm": 12.998605728149414, "learning_rate": 5.004014255296809e-06, "loss": 0.6921, "step": 12608 }, { "epoch": 2.03, "grad_norm": 11.739645004272461, "learning_rate": 5.002508783756362e-06, "loss": 0.722, "step": 12609 }, { "epoch": 2.03, "grad_norm": 11.504841804504395, "learning_rate": 5.0010034631707835e-06, "loss": 0.6249, "step": 12610 }, { "epoch": 2.03, "grad_norm": 12.032495498657227, "learning_rate": 4.999498293585542e-06, "loss": 0.8628, "step": 12611 }, { "epoch": 2.03, "grad_norm": 10.289895057678223, "learning_rate": 4.9979932750460995e-06, "loss": 0.4854, "step": 12612 }, { "epoch": 2.03, "grad_norm": 10.588204383850098, "learning_rate": 4.996488407597922e-06, "loss": 0.6038, "step": 12613 }, { "epoch": 2.03, "grad_norm": 13.19759750366211, "learning_rate": 4.994983691286459e-06, "loss": 0.7553, "step": 12614 }, { "epoch": 2.03, "grad_norm": 15.159613609313965, "learning_rate": 4.993479126157172e-06, "loss": 0.7084, "step": 12615 }, { "epoch": 2.03, "grad_norm": 9.758037567138672, "learning_rate": 4.9919747122555e-06, "loss": 0.6097, "step": 12616 }, { "epoch": 2.03, "grad_norm": 12.066262245178223, "learning_rate": 4.990470449626887e-06, "loss": 0.6146, "step": 12617 }, { "epoch": 2.03, "grad_norm": 10.861419677734375, "learning_rate": 4.988966338316774e-06, "loss": 0.7084, "step": 12618 }, { "epoch": 2.03, "grad_norm": 11.576335906982422, "learning_rate": 4.987462378370594e-06, "loss": 0.7087, "step": 12619 }, { "epoch": 2.03, "grad_norm": 13.439332008361816, "learning_rate": 4.9859585698337686e-06, "loss": 0.6414, "step": 12620 }, { "epoch": 2.03, "grad_norm": 11.794988632202148, "learning_rate": 4.984454912751733e-06, "loss": 0.8471, "step": 12621 }, { "epoch": 2.03, "grad_norm": 9.486930847167969, "learning_rate": 4.982951407169896e-06, "loss": 0.8689, "step": 12622 }, { "epoch": 2.03, "grad_norm": 12.550749778747559, "learning_rate": 4.981448053133683e-06, "loss": 0.6965, "step": 12623 }, { "epoch": 2.03, "grad_norm": 9.539897918701172, "learning_rate": 4.9799448506885e-06, "loss": 0.6427, "step": 12624 }, { "epoch": 2.03, "grad_norm": 10.487832069396973, "learning_rate": 4.978441799879747e-06, "loss": 0.7334, "step": 12625 }, { "epoch": 2.03, "grad_norm": 10.32807445526123, "learning_rate": 4.976938900752837e-06, "loss": 0.8152, "step": 12626 }, { "epoch": 2.03, "grad_norm": 12.637872695922852, "learning_rate": 4.975436153353155e-06, "loss": 0.6861, "step": 12627 }, { "epoch": 2.03, "grad_norm": 9.324996948242188, "learning_rate": 4.973933557726099e-06, "loss": 0.5477, "step": 12628 }, { "epoch": 2.03, "grad_norm": 8.253968238830566, "learning_rate": 4.972431113917062e-06, "loss": 0.6534, "step": 12629 }, { "epoch": 2.03, "grad_norm": 11.093362808227539, "learning_rate": 4.97092882197142e-06, "loss": 0.8665, "step": 12630 }, { "epoch": 2.03, "grad_norm": 10.571358680725098, "learning_rate": 4.969426681934551e-06, "loss": 0.6301, "step": 12631 }, { "epoch": 2.03, "grad_norm": 10.907541275024414, "learning_rate": 4.967924693851832e-06, "loss": 0.6631, "step": 12632 }, { "epoch": 2.03, "grad_norm": 33.73777770996094, "learning_rate": 4.966422857768629e-06, "loss": 0.6064, "step": 12633 }, { "epoch": 2.03, "grad_norm": 10.677095413208008, "learning_rate": 4.964921173730312e-06, "loss": 0.8218, "step": 12634 }, { "epoch": 2.03, "grad_norm": 11.77441120147705, "learning_rate": 4.963419641782237e-06, "loss": 0.5532, "step": 12635 }, { "epoch": 2.03, "grad_norm": 9.33170223236084, "learning_rate": 4.961918261969757e-06, "loss": 0.6472, "step": 12636 }, { "epoch": 2.03, "grad_norm": 11.7896146774292, "learning_rate": 4.960417034338229e-06, "loss": 0.686, "step": 12637 }, { "epoch": 2.03, "grad_norm": 13.184670448303223, "learning_rate": 4.958915958932994e-06, "loss": 0.7304, "step": 12638 }, { "epoch": 2.03, "grad_norm": 14.163545608520508, "learning_rate": 4.957415035799399e-06, "loss": 0.5342, "step": 12639 }, { "epoch": 2.03, "grad_norm": 12.082200050354004, "learning_rate": 4.955914264982779e-06, "loss": 0.6526, "step": 12640 }, { "epoch": 2.03, "grad_norm": 10.969477653503418, "learning_rate": 4.954413646528461e-06, "loss": 0.6659, "step": 12641 }, { "epoch": 2.04, "grad_norm": 13.951376914978027, "learning_rate": 4.952913180481781e-06, "loss": 0.6405, "step": 12642 }, { "epoch": 2.04, "grad_norm": 13.270712852478027, "learning_rate": 4.951412866888059e-06, "loss": 0.5669, "step": 12643 }, { "epoch": 2.04, "grad_norm": 11.353180885314941, "learning_rate": 4.949912705792609e-06, "loss": 0.6826, "step": 12644 }, { "epoch": 2.04, "grad_norm": 11.784310340881348, "learning_rate": 4.9484126972407544e-06, "loss": 0.8889, "step": 12645 }, { "epoch": 2.04, "grad_norm": 12.137612342834473, "learning_rate": 4.946912841277794e-06, "loss": 0.771, "step": 12646 }, { "epoch": 2.04, "grad_norm": 10.49862289428711, "learning_rate": 4.945413137949044e-06, "loss": 0.5767, "step": 12647 }, { "epoch": 2.04, "grad_norm": 12.98116683959961, "learning_rate": 4.943913587299794e-06, "loss": 0.6065, "step": 12648 }, { "epoch": 2.04, "grad_norm": 8.813631057739258, "learning_rate": 4.9424141893753466e-06, "loss": 0.7404, "step": 12649 }, { "epoch": 2.04, "grad_norm": 13.435746192932129, "learning_rate": 4.940914944220992e-06, "loss": 0.6765, "step": 12650 }, { "epoch": 2.04, "grad_norm": 10.22187614440918, "learning_rate": 4.939415851882018e-06, "loss": 0.4759, "step": 12651 }, { "epoch": 2.04, "grad_norm": 13.830843925476074, "learning_rate": 4.9379169124036986e-06, "loss": 0.6604, "step": 12652 }, { "epoch": 2.04, "grad_norm": 11.921392440795898, "learning_rate": 4.936418125831319e-06, "loss": 0.6246, "step": 12653 }, { "epoch": 2.04, "grad_norm": 12.543511390686035, "learning_rate": 4.9349194922101505e-06, "loss": 0.8868, "step": 12654 }, { "epoch": 2.04, "grad_norm": 13.204132080078125, "learning_rate": 4.933421011585454e-06, "loss": 0.6591, "step": 12655 }, { "epoch": 2.04, "grad_norm": 11.007098197937012, "learning_rate": 4.931922684002504e-06, "loss": 0.7084, "step": 12656 }, { "epoch": 2.04, "grad_norm": 8.692081451416016, "learning_rate": 4.930424509506549e-06, "loss": 0.7136, "step": 12657 }, { "epoch": 2.04, "grad_norm": 13.301933288574219, "learning_rate": 4.92892648814285e-06, "loss": 0.7123, "step": 12658 }, { "epoch": 2.04, "grad_norm": 10.39992904663086, "learning_rate": 4.9274286199566545e-06, "loss": 0.7602, "step": 12659 }, { "epoch": 2.04, "grad_norm": 9.913996696472168, "learning_rate": 4.925930904993203e-06, "loss": 0.8744, "step": 12660 }, { "epoch": 2.04, "grad_norm": 12.787332534790039, "learning_rate": 4.924433343297742e-06, "loss": 0.618, "step": 12661 }, { "epoch": 2.04, "grad_norm": 13.005840301513672, "learning_rate": 4.922935934915506e-06, "loss": 0.6674, "step": 12662 }, { "epoch": 2.04, "grad_norm": 13.14383602142334, "learning_rate": 4.92143867989172e-06, "loss": 0.7467, "step": 12663 }, { "epoch": 2.04, "grad_norm": 11.693413734436035, "learning_rate": 4.919941578271617e-06, "loss": 0.8512, "step": 12664 }, { "epoch": 2.04, "grad_norm": 10.956408500671387, "learning_rate": 4.918444630100414e-06, "loss": 0.63, "step": 12665 }, { "epoch": 2.04, "grad_norm": 10.769841194152832, "learning_rate": 4.916947835423335e-06, "loss": 0.5604, "step": 12666 }, { "epoch": 2.04, "grad_norm": 9.783354759216309, "learning_rate": 4.915451194285585e-06, "loss": 0.6761, "step": 12667 }, { "epoch": 2.04, "grad_norm": 11.513266563415527, "learning_rate": 4.913954706732372e-06, "loss": 0.7705, "step": 12668 }, { "epoch": 2.04, "grad_norm": 11.573944091796875, "learning_rate": 4.912458372808901e-06, "loss": 0.7203, "step": 12669 }, { "epoch": 2.04, "grad_norm": 16.3270263671875, "learning_rate": 4.910962192560374e-06, "loss": 0.5643, "step": 12670 }, { "epoch": 2.04, "grad_norm": 10.238847732543945, "learning_rate": 4.90946616603198e-06, "loss": 0.661, "step": 12671 }, { "epoch": 2.04, "grad_norm": 11.04511547088623, "learning_rate": 4.907970293268912e-06, "loss": 0.6321, "step": 12672 }, { "epoch": 2.04, "grad_norm": 12.65092658996582, "learning_rate": 4.906474574316353e-06, "loss": 0.7016, "step": 12673 }, { "epoch": 2.04, "grad_norm": 13.021450996398926, "learning_rate": 4.904979009219478e-06, "loss": 0.5338, "step": 12674 }, { "epoch": 2.04, "grad_norm": 10.529449462890625, "learning_rate": 4.903483598023471e-06, "loss": 0.6004, "step": 12675 }, { "epoch": 2.04, "grad_norm": 13.351075172424316, "learning_rate": 4.901988340773494e-06, "loss": 0.687, "step": 12676 }, { "epoch": 2.04, "grad_norm": 11.543601989746094, "learning_rate": 4.90049323751472e-06, "loss": 0.6897, "step": 12677 }, { "epoch": 2.04, "grad_norm": 10.60282039642334, "learning_rate": 4.898998288292308e-06, "loss": 0.6073, "step": 12678 }, { "epoch": 2.04, "grad_norm": 11.68294906616211, "learning_rate": 4.89750349315141e-06, "loss": 0.7936, "step": 12679 }, { "epoch": 2.04, "grad_norm": 9.589031219482422, "learning_rate": 4.896008852137186e-06, "loss": 0.6205, "step": 12680 }, { "epoch": 2.04, "grad_norm": 11.854345321655273, "learning_rate": 4.894514365294775e-06, "loss": 0.6574, "step": 12681 }, { "epoch": 2.04, "grad_norm": 11.271988868713379, "learning_rate": 4.893020032669328e-06, "loss": 0.5577, "step": 12682 }, { "epoch": 2.04, "grad_norm": 11.569131851196289, "learning_rate": 4.8915258543059784e-06, "loss": 0.9323, "step": 12683 }, { "epoch": 2.04, "grad_norm": 12.310521125793457, "learning_rate": 4.890031830249857e-06, "loss": 0.6239, "step": 12684 }, { "epoch": 2.04, "grad_norm": 10.664739608764648, "learning_rate": 4.8885379605460994e-06, "loss": 0.7014, "step": 12685 }, { "epoch": 2.04, "grad_norm": 12.09815788269043, "learning_rate": 4.887044245239825e-06, "loss": 0.7048, "step": 12686 }, { "epoch": 2.04, "grad_norm": 11.549942970275879, "learning_rate": 4.88555068437615e-06, "loss": 0.7734, "step": 12687 }, { "epoch": 2.04, "grad_norm": 11.557669639587402, "learning_rate": 4.884057278000197e-06, "loss": 0.5882, "step": 12688 }, { "epoch": 2.04, "grad_norm": 13.171676635742188, "learning_rate": 4.882564026157068e-06, "loss": 0.6563, "step": 12689 }, { "epoch": 2.04, "grad_norm": 12.814623832702637, "learning_rate": 4.881070928891873e-06, "loss": 0.6291, "step": 12690 }, { "epoch": 2.04, "grad_norm": 13.779662132263184, "learning_rate": 4.879577986249716e-06, "loss": 0.6795, "step": 12691 }, { "epoch": 2.04, "grad_norm": 10.43092155456543, "learning_rate": 4.878085198275685e-06, "loss": 0.4965, "step": 12692 }, { "epoch": 2.04, "grad_norm": 11.109199523925781, "learning_rate": 4.876592565014879e-06, "loss": 0.6204, "step": 12693 }, { "epoch": 2.04, "grad_norm": 12.986095428466797, "learning_rate": 4.875100086512381e-06, "loss": 0.4975, "step": 12694 }, { "epoch": 2.04, "grad_norm": 12.125391006469727, "learning_rate": 4.87360776281327e-06, "loss": 0.9407, "step": 12695 }, { "epoch": 2.04, "grad_norm": 9.429791450500488, "learning_rate": 4.87211559396263e-06, "loss": 0.6906, "step": 12696 }, { "epoch": 2.04, "grad_norm": 11.790526390075684, "learning_rate": 4.870623580005529e-06, "loss": 0.6343, "step": 12697 }, { "epoch": 2.04, "grad_norm": 10.038444519042969, "learning_rate": 4.8691317209870335e-06, "loss": 0.7477, "step": 12698 }, { "epoch": 2.04, "grad_norm": 10.593018531799316, "learning_rate": 4.867640016952213e-06, "loss": 0.6795, "step": 12699 }, { "epoch": 2.04, "grad_norm": 11.709702491760254, "learning_rate": 4.866148467946118e-06, "loss": 0.6933, "step": 12700 }, { "epoch": 2.04, "grad_norm": 10.081239700317383, "learning_rate": 4.864657074013811e-06, "loss": 0.7326, "step": 12701 }, { "epoch": 2.04, "grad_norm": 10.503482818603516, "learning_rate": 4.863165835200338e-06, "loss": 0.759, "step": 12702 }, { "epoch": 2.04, "grad_norm": 11.051582336425781, "learning_rate": 4.8616747515507366e-06, "loss": 0.8131, "step": 12703 }, { "epoch": 2.05, "grad_norm": 13.042076110839844, "learning_rate": 4.860183823110058e-06, "loss": 0.8351, "step": 12704 }, { "epoch": 2.05, "grad_norm": 10.636490821838379, "learning_rate": 4.858693049923329e-06, "loss": 0.6167, "step": 12705 }, { "epoch": 2.05, "grad_norm": 11.051466941833496, "learning_rate": 4.857202432035586e-06, "loss": 0.7974, "step": 12706 }, { "epoch": 2.05, "grad_norm": 10.786310195922852, "learning_rate": 4.855711969491852e-06, "loss": 0.5456, "step": 12707 }, { "epoch": 2.05, "grad_norm": 13.89157485961914, "learning_rate": 4.854221662337144e-06, "loss": 0.7762, "step": 12708 }, { "epoch": 2.05, "grad_norm": 12.900909423828125, "learning_rate": 4.852731510616483e-06, "loss": 0.6969, "step": 12709 }, { "epoch": 2.05, "grad_norm": 12.873169898986816, "learning_rate": 4.851241514374884e-06, "loss": 0.7525, "step": 12710 }, { "epoch": 2.05, "grad_norm": 9.769445419311523, "learning_rate": 4.849751673657346e-06, "loss": 0.7468, "step": 12711 }, { "epoch": 2.05, "grad_norm": 10.347565650939941, "learning_rate": 4.84826198850888e-06, "loss": 0.7302, "step": 12712 }, { "epoch": 2.05, "grad_norm": 11.208226203918457, "learning_rate": 4.846772458974479e-06, "loss": 0.8485, "step": 12713 }, { "epoch": 2.05, "grad_norm": 11.967630386352539, "learning_rate": 4.845283085099132e-06, "loss": 0.6607, "step": 12714 }, { "epoch": 2.05, "grad_norm": 14.210322380065918, "learning_rate": 4.843793866927835e-06, "loss": 0.7069, "step": 12715 }, { "epoch": 2.05, "grad_norm": 13.679717063903809, "learning_rate": 4.842304804505565e-06, "loss": 0.8098, "step": 12716 }, { "epoch": 2.05, "grad_norm": 11.412985801696777, "learning_rate": 4.840815897877306e-06, "loss": 0.6624, "step": 12717 }, { "epoch": 2.05, "grad_norm": 10.574965476989746, "learning_rate": 4.8393271470880295e-06, "loss": 0.5821, "step": 12718 }, { "epoch": 2.05, "grad_norm": 10.465807914733887, "learning_rate": 4.837838552182703e-06, "loss": 0.7469, "step": 12719 }, { "epoch": 2.05, "grad_norm": 12.080510139465332, "learning_rate": 4.836350113206296e-06, "loss": 0.5398, "step": 12720 }, { "epoch": 2.05, "grad_norm": 11.456509590148926, "learning_rate": 4.834861830203766e-06, "loss": 0.7954, "step": 12721 }, { "epoch": 2.05, "grad_norm": 12.257381439208984, "learning_rate": 4.833373703220065e-06, "loss": 0.7547, "step": 12722 }, { "epoch": 2.05, "grad_norm": 11.963224411010742, "learning_rate": 4.83188573230015e-06, "loss": 0.5413, "step": 12723 }, { "epoch": 2.05, "grad_norm": 12.593356132507324, "learning_rate": 4.83039791748896e-06, "loss": 0.8589, "step": 12724 }, { "epoch": 2.05, "grad_norm": 12.902804374694824, "learning_rate": 4.828910258831443e-06, "loss": 0.6161, "step": 12725 }, { "epoch": 2.05, "grad_norm": 11.186161994934082, "learning_rate": 4.8274227563725316e-06, "loss": 0.6459, "step": 12726 }, { "epoch": 2.05, "grad_norm": 12.680699348449707, "learning_rate": 4.825935410157155e-06, "loss": 0.824, "step": 12727 }, { "epoch": 2.05, "grad_norm": 10.5338773727417, "learning_rate": 4.824448220230246e-06, "loss": 0.7102, "step": 12728 }, { "epoch": 2.05, "grad_norm": 10.921072006225586, "learning_rate": 4.822961186636721e-06, "loss": 0.8228, "step": 12729 }, { "epoch": 2.05, "grad_norm": 10.076069831848145, "learning_rate": 4.8214743094215e-06, "loss": 0.7686, "step": 12730 }, { "epoch": 2.05, "grad_norm": 13.092434883117676, "learning_rate": 4.8199875886295e-06, "loss": 0.5702, "step": 12731 }, { "epoch": 2.05, "grad_norm": 10.591981887817383, "learning_rate": 4.818501024305625e-06, "loss": 0.5131, "step": 12732 }, { "epoch": 2.05, "grad_norm": 10.778672218322754, "learning_rate": 4.817014616494776e-06, "loss": 0.7203, "step": 12733 }, { "epoch": 2.05, "grad_norm": 13.214312553405762, "learning_rate": 4.815528365241857e-06, "loss": 0.8393, "step": 12734 }, { "epoch": 2.05, "grad_norm": 11.979135513305664, "learning_rate": 4.814042270591757e-06, "loss": 0.7852, "step": 12735 }, { "epoch": 2.05, "grad_norm": 11.163859367370605, "learning_rate": 4.81255633258937e-06, "loss": 0.7339, "step": 12736 }, { "epoch": 2.05, "grad_norm": 11.642580032348633, "learning_rate": 4.811070551279578e-06, "loss": 0.588, "step": 12737 }, { "epoch": 2.05, "grad_norm": 11.193102836608887, "learning_rate": 4.809584926707256e-06, "loss": 0.5661, "step": 12738 }, { "epoch": 2.05, "grad_norm": 12.477843284606934, "learning_rate": 4.808099458917288e-06, "loss": 0.5715, "step": 12739 }, { "epoch": 2.05, "grad_norm": 19.23539924621582, "learning_rate": 4.8066141479545405e-06, "loss": 0.7282, "step": 12740 }, { "epoch": 2.05, "grad_norm": 13.495647430419922, "learning_rate": 4.805128993863873e-06, "loss": 0.6977, "step": 12741 }, { "epoch": 2.05, "grad_norm": 11.423006057739258, "learning_rate": 4.803643996690156e-06, "loss": 0.9219, "step": 12742 }, { "epoch": 2.05, "grad_norm": 12.368017196655273, "learning_rate": 4.8021591564782365e-06, "loss": 0.8347, "step": 12743 }, { "epoch": 2.05, "grad_norm": 12.279464721679688, "learning_rate": 4.800674473272975e-06, "loss": 0.6256, "step": 12744 }, { "epoch": 2.05, "grad_norm": 13.381113052368164, "learning_rate": 4.799189947119214e-06, "loss": 0.773, "step": 12745 }, { "epoch": 2.05, "grad_norm": 13.616872787475586, "learning_rate": 4.797705578061789e-06, "loss": 0.5921, "step": 12746 }, { "epoch": 2.05, "grad_norm": 11.043035507202148, "learning_rate": 4.796221366145547e-06, "loss": 0.7266, "step": 12747 }, { "epoch": 2.05, "grad_norm": 21.416118621826172, "learning_rate": 4.7947373114153125e-06, "loss": 0.8127, "step": 12748 }, { "epoch": 2.05, "grad_norm": 11.465535163879395, "learning_rate": 4.79325341391592e-06, "loss": 0.7601, "step": 12749 }, { "epoch": 2.05, "grad_norm": 12.669341087341309, "learning_rate": 4.791769673692186e-06, "loss": 0.776, "step": 12750 }, { "epoch": 2.05, "grad_norm": 12.172280311584473, "learning_rate": 4.790286090788934e-06, "loss": 0.8316, "step": 12751 }, { "epoch": 2.05, "grad_norm": 12.609841346740723, "learning_rate": 4.788802665250971e-06, "loss": 0.5378, "step": 12752 }, { "epoch": 2.05, "grad_norm": 10.67297077178955, "learning_rate": 4.7873193971231135e-06, "loss": 0.6989, "step": 12753 }, { "epoch": 2.05, "grad_norm": 11.117085456848145, "learning_rate": 4.785836286450157e-06, "loss": 0.5861, "step": 12754 }, { "epoch": 2.05, "grad_norm": 13.053116798400879, "learning_rate": 4.784353333276907e-06, "loss": 0.5762, "step": 12755 }, { "epoch": 2.05, "grad_norm": 10.990001678466797, "learning_rate": 4.782870537648158e-06, "loss": 0.6961, "step": 12756 }, { "epoch": 2.05, "grad_norm": 17.536563873291016, "learning_rate": 4.7813878996086905e-06, "loss": 0.6914, "step": 12757 }, { "epoch": 2.05, "grad_norm": 11.376727104187012, "learning_rate": 4.7799054192033e-06, "loss": 0.6475, "step": 12758 }, { "epoch": 2.05, "grad_norm": 10.431741714477539, "learning_rate": 4.778423096476758e-06, "loss": 0.7065, "step": 12759 }, { "epoch": 2.05, "grad_norm": 10.305206298828125, "learning_rate": 4.776940931473849e-06, "loss": 0.9606, "step": 12760 }, { "epoch": 2.05, "grad_norm": 12.736348152160645, "learning_rate": 4.7754589242393376e-06, "loss": 0.6859, "step": 12761 }, { "epoch": 2.05, "grad_norm": 12.453690528869629, "learning_rate": 4.773977074817986e-06, "loss": 0.6282, "step": 12762 }, { "epoch": 2.05, "grad_norm": 10.67434024810791, "learning_rate": 4.7724953832545625e-06, "loss": 0.9002, "step": 12763 }, { "epoch": 2.05, "grad_norm": 11.856401443481445, "learning_rate": 4.771013849593822e-06, "loss": 0.8205, "step": 12764 }, { "epoch": 2.05, "grad_norm": 13.490784645080566, "learning_rate": 4.769532473880508e-06, "loss": 0.666, "step": 12765 }, { "epoch": 2.06, "grad_norm": 12.017183303833008, "learning_rate": 4.768051256159378e-06, "loss": 0.7177, "step": 12766 }, { "epoch": 2.06, "grad_norm": 10.014714241027832, "learning_rate": 4.766570196475165e-06, "loss": 0.5918, "step": 12767 }, { "epoch": 2.06, "grad_norm": 10.107979774475098, "learning_rate": 4.765089294872613e-06, "loss": 0.9476, "step": 12768 }, { "epoch": 2.06, "grad_norm": 11.732234001159668, "learning_rate": 4.763608551396453e-06, "loss": 0.7798, "step": 12769 }, { "epoch": 2.06, "grad_norm": 12.540228843688965, "learning_rate": 4.762127966091402e-06, "loss": 0.7355, "step": 12770 }, { "epoch": 2.06, "grad_norm": 13.34303092956543, "learning_rate": 4.7606475390022e-06, "loss": 0.7228, "step": 12771 }, { "epoch": 2.06, "grad_norm": 10.598856925964355, "learning_rate": 4.7591672701735555e-06, "loss": 0.626, "step": 12772 }, { "epoch": 2.06, "grad_norm": 10.90831470489502, "learning_rate": 4.75768715965018e-06, "loss": 0.6819, "step": 12773 }, { "epoch": 2.06, "grad_norm": 9.436334609985352, "learning_rate": 4.756207207476787e-06, "loss": 0.5556, "step": 12774 }, { "epoch": 2.06, "grad_norm": 13.654433250427246, "learning_rate": 4.7547274136980794e-06, "loss": 0.7091, "step": 12775 }, { "epoch": 2.06, "grad_norm": 9.966647148132324, "learning_rate": 4.75324777835875e-06, "loss": 0.5741, "step": 12776 }, { "epoch": 2.06, "grad_norm": 11.623149871826172, "learning_rate": 4.751768301503501e-06, "loss": 0.7683, "step": 12777 }, { "epoch": 2.06, "grad_norm": 10.788333892822266, "learning_rate": 4.750288983177013e-06, "loss": 0.8326, "step": 12778 }, { "epoch": 2.06, "grad_norm": 12.276298522949219, "learning_rate": 4.748809823423982e-06, "loss": 0.7944, "step": 12779 }, { "epoch": 2.06, "grad_norm": 12.579583168029785, "learning_rate": 4.747330822289079e-06, "loss": 0.8921, "step": 12780 }, { "epoch": 2.06, "grad_norm": 11.806264877319336, "learning_rate": 4.745851979816979e-06, "loss": 0.8076, "step": 12781 }, { "epoch": 2.06, "grad_norm": 13.03809642791748, "learning_rate": 4.7443732960523555e-06, "loss": 0.7713, "step": 12782 }, { "epoch": 2.06, "grad_norm": 11.856563568115234, "learning_rate": 4.742894771039871e-06, "loss": 0.869, "step": 12783 }, { "epoch": 2.06, "grad_norm": 10.798720359802246, "learning_rate": 4.741416404824189e-06, "loss": 0.5937, "step": 12784 }, { "epoch": 2.06, "grad_norm": 10.657962799072266, "learning_rate": 4.739938197449966e-06, "loss": 0.6223, "step": 12785 }, { "epoch": 2.06, "grad_norm": 12.814309120178223, "learning_rate": 4.738460148961845e-06, "loss": 0.7658, "step": 12786 }, { "epoch": 2.06, "grad_norm": 10.429557800292969, "learning_rate": 4.7369822594044835e-06, "loss": 0.7511, "step": 12787 }, { "epoch": 2.06, "grad_norm": 12.408181190490723, "learning_rate": 4.735504528822516e-06, "loss": 0.8832, "step": 12788 }, { "epoch": 2.06, "grad_norm": 10.615962982177734, "learning_rate": 4.7340269572605755e-06, "loss": 0.8289, "step": 12789 }, { "epoch": 2.06, "grad_norm": 10.013772964477539, "learning_rate": 4.7325495447633e-06, "loss": 0.5107, "step": 12790 }, { "epoch": 2.06, "grad_norm": 13.335679054260254, "learning_rate": 4.731072291375317e-06, "loss": 0.8331, "step": 12791 }, { "epoch": 2.06, "grad_norm": 8.906449317932129, "learning_rate": 4.729595197141243e-06, "loss": 0.7405, "step": 12792 }, { "epoch": 2.06, "grad_norm": 9.755775451660156, "learning_rate": 4.728118262105702e-06, "loss": 0.6568, "step": 12793 }, { "epoch": 2.06, "grad_norm": 12.917089462280273, "learning_rate": 4.7266414863133e-06, "loss": 0.5761, "step": 12794 }, { "epoch": 2.06, "grad_norm": 10.522432327270508, "learning_rate": 4.725164869808652e-06, "loss": 0.6878, "step": 12795 }, { "epoch": 2.06, "grad_norm": 12.847689628601074, "learning_rate": 4.723688412636356e-06, "loss": 0.7487, "step": 12796 }, { "epoch": 2.06, "grad_norm": 8.83850383758545, "learning_rate": 4.722212114841006e-06, "loss": 0.4963, "step": 12797 }, { "epoch": 2.06, "grad_norm": 9.81480598449707, "learning_rate": 4.720735976467205e-06, "loss": 0.6476, "step": 12798 }, { "epoch": 2.06, "grad_norm": 11.264976501464844, "learning_rate": 4.719259997559537e-06, "loss": 0.7183, "step": 12799 }, { "epoch": 2.06, "grad_norm": 14.432208061218262, "learning_rate": 4.71778417816258e-06, "loss": 0.6256, "step": 12800 }, { "epoch": 2.06, "grad_norm": 9.091652870178223, "learning_rate": 4.716308518320921e-06, "loss": 0.7319, "step": 12801 }, { "epoch": 2.06, "grad_norm": 11.040436744689941, "learning_rate": 4.714833018079129e-06, "loss": 0.7302, "step": 12802 }, { "epoch": 2.06, "grad_norm": 11.006303787231445, "learning_rate": 4.713357677481777e-06, "loss": 0.5661, "step": 12803 }, { "epoch": 2.06, "grad_norm": 11.965200424194336, "learning_rate": 4.7118824965734275e-06, "loss": 0.6117, "step": 12804 }, { "epoch": 2.06, "grad_norm": 13.311362266540527, "learning_rate": 4.710407475398637e-06, "loss": 0.6916, "step": 12805 }, { "epoch": 2.06, "grad_norm": 10.322787284851074, "learning_rate": 4.708932614001967e-06, "loss": 0.7662, "step": 12806 }, { "epoch": 2.06, "grad_norm": 12.575461387634277, "learning_rate": 4.707457912427962e-06, "loss": 0.6877, "step": 12807 }, { "epoch": 2.06, "grad_norm": 13.104768753051758, "learning_rate": 4.705983370721165e-06, "loss": 0.6739, "step": 12808 }, { "epoch": 2.06, "grad_norm": 10.928337097167969, "learning_rate": 4.704508988926122e-06, "loss": 0.4892, "step": 12809 }, { "epoch": 2.06, "grad_norm": 12.850425720214844, "learning_rate": 4.703034767087363e-06, "loss": 0.7382, "step": 12810 }, { "epoch": 2.06, "grad_norm": 10.626716613769531, "learning_rate": 4.7015607052494225e-06, "loss": 0.598, "step": 12811 }, { "epoch": 2.06, "grad_norm": 16.22136878967285, "learning_rate": 4.700086803456828e-06, "loss": 0.6757, "step": 12812 }, { "epoch": 2.06, "grad_norm": 11.37704849243164, "learning_rate": 4.698613061754094e-06, "loss": 0.6548, "step": 12813 }, { "epoch": 2.06, "grad_norm": 10.882416725158691, "learning_rate": 4.697139480185743e-06, "loss": 0.5064, "step": 12814 }, { "epoch": 2.06, "grad_norm": 11.510451316833496, "learning_rate": 4.695666058796284e-06, "loss": 0.8396, "step": 12815 }, { "epoch": 2.06, "grad_norm": 11.519611358642578, "learning_rate": 4.694192797630218e-06, "loss": 0.5136, "step": 12816 }, { "epoch": 2.06, "grad_norm": 10.714570999145508, "learning_rate": 4.692719696732056e-06, "loss": 0.55, "step": 12817 }, { "epoch": 2.06, "grad_norm": 11.20350456237793, "learning_rate": 4.691246756146289e-06, "loss": 0.6488, "step": 12818 }, { "epoch": 2.06, "grad_norm": 9.789166450500488, "learning_rate": 4.689773975917407e-06, "loss": 0.5565, "step": 12819 }, { "epoch": 2.06, "grad_norm": 11.201627731323242, "learning_rate": 4.6883013560899014e-06, "loss": 0.6924, "step": 12820 }, { "epoch": 2.06, "grad_norm": 12.963919639587402, "learning_rate": 4.6868288967082495e-06, "loss": 0.6477, "step": 12821 }, { "epoch": 2.06, "grad_norm": 10.06932544708252, "learning_rate": 4.685356597816937e-06, "loss": 0.6063, "step": 12822 }, { "epoch": 2.06, "grad_norm": 12.438987731933594, "learning_rate": 4.683884459460429e-06, "loss": 0.6243, "step": 12823 }, { "epoch": 2.06, "grad_norm": 12.144668579101562, "learning_rate": 4.6824124816831915e-06, "loss": 0.403, "step": 12824 }, { "epoch": 2.06, "grad_norm": 13.659475326538086, "learning_rate": 4.680940664529695e-06, "loss": 0.7354, "step": 12825 }, { "epoch": 2.06, "grad_norm": 11.974043846130371, "learning_rate": 4.6794690080443905e-06, "loss": 0.5982, "step": 12826 }, { "epoch": 2.06, "grad_norm": 13.867949485778809, "learning_rate": 4.677997512271736e-06, "loss": 0.9097, "step": 12827 }, { "epoch": 2.07, "grad_norm": 13.669520378112793, "learning_rate": 4.676526177256179e-06, "loss": 0.7924, "step": 12828 }, { "epoch": 2.07, "grad_norm": 14.381863594055176, "learning_rate": 4.675055003042156e-06, "loss": 0.8605, "step": 12829 }, { "epoch": 2.07, "grad_norm": 12.76263427734375, "learning_rate": 4.673583989674115e-06, "loss": 0.698, "step": 12830 }, { "epoch": 2.07, "grad_norm": 9.998507499694824, "learning_rate": 4.672113137196482e-06, "loss": 0.6455, "step": 12831 }, { "epoch": 2.07, "grad_norm": 11.694927215576172, "learning_rate": 4.6706424456536896e-06, "loss": 0.5997, "step": 12832 }, { "epoch": 2.07, "grad_norm": 10.207916259765625, "learning_rate": 4.669171915090164e-06, "loss": 0.6953, "step": 12833 }, { "epoch": 2.07, "grad_norm": 13.43675708770752, "learning_rate": 4.667701545550322e-06, "loss": 0.8271, "step": 12834 }, { "epoch": 2.07, "grad_norm": 12.309857368469238, "learning_rate": 4.6662313370785724e-06, "loss": 0.5584, "step": 12835 }, { "epoch": 2.07, "grad_norm": 10.684440612792969, "learning_rate": 4.664761289719335e-06, "loss": 0.5927, "step": 12836 }, { "epoch": 2.07, "grad_norm": 11.000016212463379, "learning_rate": 4.663291403517004e-06, "loss": 0.5229, "step": 12837 }, { "epoch": 2.07, "grad_norm": 16.475364685058594, "learning_rate": 4.661821678515988e-06, "loss": 0.6394, "step": 12838 }, { "epoch": 2.07, "grad_norm": 13.897146224975586, "learning_rate": 4.6603521147606754e-06, "loss": 0.687, "step": 12839 }, { "epoch": 2.07, "grad_norm": 12.312759399414062, "learning_rate": 4.6588827122954565e-06, "loss": 0.7854, "step": 12840 }, { "epoch": 2.07, "grad_norm": 11.401483535766602, "learning_rate": 4.65741347116472e-06, "loss": 0.6008, "step": 12841 }, { "epoch": 2.07, "grad_norm": 11.651519775390625, "learning_rate": 4.655944391412844e-06, "loss": 0.6152, "step": 12842 }, { "epoch": 2.07, "grad_norm": 18.72403335571289, "learning_rate": 4.6544754730842005e-06, "loss": 0.7017, "step": 12843 }, { "epoch": 2.07, "grad_norm": 11.70914363861084, "learning_rate": 4.653006716223165e-06, "loss": 0.7841, "step": 12844 }, { "epoch": 2.07, "grad_norm": 12.582624435424805, "learning_rate": 4.651538120874099e-06, "loss": 0.7084, "step": 12845 }, { "epoch": 2.07, "grad_norm": 12.422016143798828, "learning_rate": 4.650069687081366e-06, "loss": 0.6116, "step": 12846 }, { "epoch": 2.07, "grad_norm": 12.237881660461426, "learning_rate": 4.648601414889322e-06, "loss": 0.6006, "step": 12847 }, { "epoch": 2.07, "grad_norm": 13.879839897155762, "learning_rate": 4.647133304342312e-06, "loss": 0.8798, "step": 12848 }, { "epoch": 2.07, "grad_norm": 13.648749351501465, "learning_rate": 4.64566535548469e-06, "loss": 0.7499, "step": 12849 }, { "epoch": 2.07, "grad_norm": 11.491506576538086, "learning_rate": 4.6441975683607895e-06, "loss": 0.8219, "step": 12850 }, { "epoch": 2.07, "grad_norm": 10.402905464172363, "learning_rate": 4.6427299430149505e-06, "loss": 0.7077, "step": 12851 }, { "epoch": 2.07, "grad_norm": 13.915287017822266, "learning_rate": 4.641262479491509e-06, "loss": 0.7953, "step": 12852 }, { "epoch": 2.07, "grad_norm": 12.464155197143555, "learning_rate": 4.639795177834786e-06, "loss": 0.6252, "step": 12853 }, { "epoch": 2.07, "grad_norm": 12.440953254699707, "learning_rate": 4.638328038089099e-06, "loss": 0.7833, "step": 12854 }, { "epoch": 2.07, "grad_norm": 12.58263111114502, "learning_rate": 4.636861060298774e-06, "loss": 0.7604, "step": 12855 }, { "epoch": 2.07, "grad_norm": 11.657628059387207, "learning_rate": 4.635394244508114e-06, "loss": 0.6968, "step": 12856 }, { "epoch": 2.07, "grad_norm": 14.637052536010742, "learning_rate": 4.633927590761434e-06, "loss": 0.7655, "step": 12857 }, { "epoch": 2.07, "grad_norm": 11.08665657043457, "learning_rate": 4.63246109910303e-06, "loss": 0.5872, "step": 12858 }, { "epoch": 2.07, "grad_norm": 11.238606452941895, "learning_rate": 4.6309947695772e-06, "loss": 0.7201, "step": 12859 }, { "epoch": 2.07, "grad_norm": 9.801209449768066, "learning_rate": 4.629528602228238e-06, "loss": 0.6509, "step": 12860 }, { "epoch": 2.07, "grad_norm": 12.467313766479492, "learning_rate": 4.628062597100428e-06, "loss": 0.6627, "step": 12861 }, { "epoch": 2.07, "grad_norm": 11.476099014282227, "learning_rate": 4.626596754238057e-06, "loss": 0.5586, "step": 12862 }, { "epoch": 2.07, "grad_norm": 8.868212699890137, "learning_rate": 4.625131073685401e-06, "loss": 0.6596, "step": 12863 }, { "epoch": 2.07, "grad_norm": 12.698614120483398, "learning_rate": 4.623665555486726e-06, "loss": 0.7647, "step": 12864 }, { "epoch": 2.07, "grad_norm": 11.44686508178711, "learning_rate": 4.622200199686311e-06, "loss": 0.8313, "step": 12865 }, { "epoch": 2.07, "grad_norm": 13.138679504394531, "learning_rate": 4.6207350063284106e-06, "loss": 0.5535, "step": 12866 }, { "epoch": 2.07, "grad_norm": 11.861519813537598, "learning_rate": 4.6192699754572834e-06, "loss": 0.7161, "step": 12867 }, { "epoch": 2.07, "grad_norm": 11.904197692871094, "learning_rate": 4.617805107117186e-06, "loss": 0.7057, "step": 12868 }, { "epoch": 2.07, "grad_norm": 15.595900535583496, "learning_rate": 4.61634040135236e-06, "loss": 0.6915, "step": 12869 }, { "epoch": 2.07, "grad_norm": 10.447653770446777, "learning_rate": 4.614875858207057e-06, "loss": 0.6061, "step": 12870 }, { "epoch": 2.07, "grad_norm": 12.93587589263916, "learning_rate": 4.613411477725508e-06, "loss": 0.7539, "step": 12871 }, { "epoch": 2.07, "grad_norm": 14.020979881286621, "learning_rate": 4.611947259951948e-06, "loss": 0.7369, "step": 12872 }, { "epoch": 2.07, "grad_norm": 10.082112312316895, "learning_rate": 4.610483204930611e-06, "loss": 0.5431, "step": 12873 }, { "epoch": 2.07, "grad_norm": 14.464605331420898, "learning_rate": 4.609019312705716e-06, "loss": 0.6958, "step": 12874 }, { "epoch": 2.07, "grad_norm": 11.368803024291992, "learning_rate": 4.607555583321477e-06, "loss": 0.6133, "step": 12875 }, { "epoch": 2.07, "grad_norm": 10.108997344970703, "learning_rate": 4.6060920168221154e-06, "loss": 0.6653, "step": 12876 }, { "epoch": 2.07, "grad_norm": 12.698660850524902, "learning_rate": 4.604628613251837e-06, "loss": 0.6136, "step": 12877 }, { "epoch": 2.07, "grad_norm": 11.949407577514648, "learning_rate": 4.603165372654841e-06, "loss": 0.6782, "step": 12878 }, { "epoch": 2.07, "grad_norm": 11.65782642364502, "learning_rate": 4.601702295075334e-06, "loss": 0.6321, "step": 12879 }, { "epoch": 2.07, "grad_norm": 13.10770320892334, "learning_rate": 4.600239380557503e-06, "loss": 0.727, "step": 12880 }, { "epoch": 2.07, "grad_norm": 12.997252464294434, "learning_rate": 4.598776629145544e-06, "loss": 0.7809, "step": 12881 }, { "epoch": 2.07, "grad_norm": 12.103215217590332, "learning_rate": 4.597314040883637e-06, "loss": 0.5727, "step": 12882 }, { "epoch": 2.07, "grad_norm": 11.674296379089355, "learning_rate": 4.595851615815957e-06, "loss": 0.7709, "step": 12883 }, { "epoch": 2.07, "grad_norm": 12.239984512329102, "learning_rate": 4.594389353986687e-06, "loss": 0.7368, "step": 12884 }, { "epoch": 2.07, "grad_norm": 10.361051559448242, "learning_rate": 4.59292725543999e-06, "loss": 0.6649, "step": 12885 }, { "epoch": 2.07, "grad_norm": 11.689462661743164, "learning_rate": 4.591465320220031e-06, "loss": 0.8981, "step": 12886 }, { "epoch": 2.07, "grad_norm": 15.40504264831543, "learning_rate": 4.590003548370974e-06, "loss": 0.7829, "step": 12887 }, { "epoch": 2.07, "grad_norm": 11.820805549621582, "learning_rate": 4.588541939936965e-06, "loss": 0.7117, "step": 12888 }, { "epoch": 2.07, "grad_norm": 11.6724853515625, "learning_rate": 4.587080494962163e-06, "loss": 0.6932, "step": 12889 }, { "epoch": 2.08, "grad_norm": 11.526752471923828, "learning_rate": 4.585619213490707e-06, "loss": 0.6512, "step": 12890 }, { "epoch": 2.08, "grad_norm": 10.273880004882812, "learning_rate": 4.584158095566732e-06, "loss": 0.8384, "step": 12891 }, { "epoch": 2.08, "grad_norm": 13.999848365783691, "learning_rate": 4.582697141234386e-06, "loss": 0.7278, "step": 12892 }, { "epoch": 2.08, "grad_norm": 10.566195487976074, "learning_rate": 4.58123635053779e-06, "loss": 0.748, "step": 12893 }, { "epoch": 2.08, "grad_norm": 10.413511276245117, "learning_rate": 4.5797757235210675e-06, "loss": 0.7692, "step": 12894 }, { "epoch": 2.08, "grad_norm": 12.15650463104248, "learning_rate": 4.5783152602283446e-06, "loss": 0.9374, "step": 12895 }, { "epoch": 2.08, "grad_norm": 10.220779418945312, "learning_rate": 4.576854960703731e-06, "loss": 0.6087, "step": 12896 }, { "epoch": 2.08, "grad_norm": 9.33862018585205, "learning_rate": 4.575394824991337e-06, "loss": 0.5297, "step": 12897 }, { "epoch": 2.08, "grad_norm": 11.432191848754883, "learning_rate": 4.5739348531352725e-06, "loss": 0.7765, "step": 12898 }, { "epoch": 2.08, "grad_norm": 12.510437965393066, "learning_rate": 4.572475045179629e-06, "loss": 0.6616, "step": 12899 }, { "epoch": 2.08, "grad_norm": 13.29851245880127, "learning_rate": 4.5710154011685115e-06, "loss": 0.6985, "step": 12900 }, { "epoch": 2.08, "grad_norm": 11.788554191589355, "learning_rate": 4.569555921146005e-06, "loss": 0.6636, "step": 12901 }, { "epoch": 2.08, "grad_norm": 11.105588912963867, "learning_rate": 4.568096605156193e-06, "loss": 0.7575, "step": 12902 }, { "epoch": 2.08, "grad_norm": 11.81989574432373, "learning_rate": 4.566637453243159e-06, "loss": 0.6113, "step": 12903 }, { "epoch": 2.08, "grad_norm": 11.70374584197998, "learning_rate": 4.565178465450976e-06, "loss": 0.7873, "step": 12904 }, { "epoch": 2.08, "grad_norm": 11.212965965270996, "learning_rate": 4.563719641823719e-06, "loss": 0.7815, "step": 12905 }, { "epoch": 2.08, "grad_norm": 14.238959312438965, "learning_rate": 4.562260982405449e-06, "loss": 0.6984, "step": 12906 }, { "epoch": 2.08, "grad_norm": 11.164651870727539, "learning_rate": 4.560802487240224e-06, "loss": 0.6383, "step": 12907 }, { "epoch": 2.08, "grad_norm": 11.527359008789062, "learning_rate": 4.559344156372108e-06, "loss": 0.7523, "step": 12908 }, { "epoch": 2.08, "grad_norm": 13.092981338500977, "learning_rate": 4.557885989845146e-06, "loss": 0.8319, "step": 12909 }, { "epoch": 2.08, "grad_norm": 11.647616386413574, "learning_rate": 4.556427987703379e-06, "loss": 0.7062, "step": 12910 }, { "epoch": 2.08, "grad_norm": 13.558351516723633, "learning_rate": 4.554970149990858e-06, "loss": 0.727, "step": 12911 }, { "epoch": 2.08, "grad_norm": 12.144123077392578, "learning_rate": 4.553512476751609e-06, "loss": 0.6798, "step": 12912 }, { "epoch": 2.08, "grad_norm": 11.180291175842285, "learning_rate": 4.552054968029669e-06, "loss": 0.7126, "step": 12913 }, { "epoch": 2.08, "grad_norm": 11.162674903869629, "learning_rate": 4.550597623869063e-06, "loss": 0.6823, "step": 12914 }, { "epoch": 2.08, "grad_norm": 11.109075546264648, "learning_rate": 4.549140444313808e-06, "loss": 0.612, "step": 12915 }, { "epoch": 2.08, "grad_norm": 12.00794506072998, "learning_rate": 4.5476834294079264e-06, "loss": 0.6512, "step": 12916 }, { "epoch": 2.08, "grad_norm": 11.354310035705566, "learning_rate": 4.546226579195424e-06, "loss": 0.5597, "step": 12917 }, { "epoch": 2.08, "grad_norm": 13.011600494384766, "learning_rate": 4.544769893720303e-06, "loss": 0.6672, "step": 12918 }, { "epoch": 2.08, "grad_norm": 11.431343078613281, "learning_rate": 4.543313373026574e-06, "loss": 0.7658, "step": 12919 }, { "epoch": 2.08, "grad_norm": 13.383502006530762, "learning_rate": 4.5418570171582265e-06, "loss": 0.55, "step": 12920 }, { "epoch": 2.08, "grad_norm": 11.768524169921875, "learning_rate": 4.540400826159249e-06, "loss": 0.7317, "step": 12921 }, { "epoch": 2.08, "grad_norm": 12.76537036895752, "learning_rate": 4.538944800073635e-06, "loss": 0.6157, "step": 12922 }, { "epoch": 2.08, "grad_norm": 13.355819702148438, "learning_rate": 4.537488938945357e-06, "loss": 0.8356, "step": 12923 }, { "epoch": 2.08, "grad_norm": 10.131932258605957, "learning_rate": 4.5360332428183985e-06, "loss": 0.7797, "step": 12924 }, { "epoch": 2.08, "grad_norm": 10.916218757629395, "learning_rate": 4.534577711736727e-06, "loss": 0.5683, "step": 12925 }, { "epoch": 2.08, "grad_norm": 11.752087593078613, "learning_rate": 4.5331223457443055e-06, "loss": 0.6223, "step": 12926 }, { "epoch": 2.08, "grad_norm": 9.671523094177246, "learning_rate": 4.5316671448851016e-06, "loss": 0.4964, "step": 12927 }, { "epoch": 2.08, "grad_norm": 9.658611297607422, "learning_rate": 4.5302121092030635e-06, "loss": 0.698, "step": 12928 }, { "epoch": 2.08, "grad_norm": 12.954034805297852, "learning_rate": 4.528757238742151e-06, "loss": 0.7669, "step": 12929 }, { "epoch": 2.08, "grad_norm": 9.3908109664917, "learning_rate": 4.527302533546304e-06, "loss": 0.6288, "step": 12930 }, { "epoch": 2.08, "grad_norm": 10.779650688171387, "learning_rate": 4.5258479936594624e-06, "loss": 0.4419, "step": 12931 }, { "epoch": 2.08, "grad_norm": 12.44583511352539, "learning_rate": 4.524393619125564e-06, "loss": 0.6109, "step": 12932 }, { "epoch": 2.08, "grad_norm": 12.48363971710205, "learning_rate": 4.522939409988546e-06, "loss": 0.7526, "step": 12933 }, { "epoch": 2.08, "grad_norm": 11.21918773651123, "learning_rate": 4.521485366292324e-06, "loss": 0.7221, "step": 12934 }, { "epoch": 2.08, "grad_norm": 13.34183406829834, "learning_rate": 4.520031488080828e-06, "loss": 0.6713, "step": 12935 }, { "epoch": 2.08, "grad_norm": 10.55435848236084, "learning_rate": 4.518577775397971e-06, "loss": 0.6941, "step": 12936 }, { "epoch": 2.08, "grad_norm": 14.48320198059082, "learning_rate": 4.517124228287658e-06, "loss": 0.7366, "step": 12937 }, { "epoch": 2.08, "grad_norm": 10.414401054382324, "learning_rate": 4.515670846793805e-06, "loss": 0.7672, "step": 12938 }, { "epoch": 2.08, "grad_norm": 10.372855186462402, "learning_rate": 4.514217630960305e-06, "loss": 0.6673, "step": 12939 }, { "epoch": 2.08, "grad_norm": 11.063325881958008, "learning_rate": 4.51276458083106e-06, "loss": 0.6893, "step": 12940 }, { "epoch": 2.08, "grad_norm": 11.350659370422363, "learning_rate": 4.511311696449958e-06, "loss": 0.6998, "step": 12941 }, { "epoch": 2.08, "grad_norm": 12.521023750305176, "learning_rate": 4.509858977860882e-06, "loss": 0.7506, "step": 12942 }, { "epoch": 2.08, "grad_norm": 10.677099227905273, "learning_rate": 4.5084064251077195e-06, "loss": 0.8117, "step": 12943 }, { "epoch": 2.08, "grad_norm": 12.109318733215332, "learning_rate": 4.506954038234345e-06, "loss": 0.6403, "step": 12944 }, { "epoch": 2.08, "grad_norm": 10.334181785583496, "learning_rate": 4.505501817284622e-06, "loss": 0.6864, "step": 12945 }, { "epoch": 2.08, "grad_norm": 12.919219970703125, "learning_rate": 4.5040497623024275e-06, "loss": 0.7576, "step": 12946 }, { "epoch": 2.08, "grad_norm": 12.322911262512207, "learning_rate": 4.5025978733316135e-06, "loss": 0.7766, "step": 12947 }, { "epoch": 2.08, "grad_norm": 13.117436408996582, "learning_rate": 4.501146150416042e-06, "loss": 0.7256, "step": 12948 }, { "epoch": 2.08, "grad_norm": 9.877079010009766, "learning_rate": 4.499694593599563e-06, "loss": 0.7872, "step": 12949 }, { "epoch": 2.08, "grad_norm": 13.057059288024902, "learning_rate": 4.498243202926017e-06, "loss": 0.653, "step": 12950 }, { "epoch": 2.08, "grad_norm": 11.53433895111084, "learning_rate": 4.496791978439252e-06, "loss": 0.6789, "step": 12951 }, { "epoch": 2.08, "grad_norm": 12.68272876739502, "learning_rate": 4.495340920183097e-06, "loss": 0.6693, "step": 12952 }, { "epoch": 2.09, "grad_norm": 21.795028686523438, "learning_rate": 4.493890028201387e-06, "loss": 0.8724, "step": 12953 }, { "epoch": 2.09, "grad_norm": 12.028498649597168, "learning_rate": 4.492439302537952e-06, "loss": 0.5454, "step": 12954 }, { "epoch": 2.09, "grad_norm": 11.962446212768555, "learning_rate": 4.490988743236607e-06, "loss": 0.7932, "step": 12955 }, { "epoch": 2.09, "grad_norm": 12.123146057128906, "learning_rate": 4.489538350341166e-06, "loss": 0.6648, "step": 12956 }, { "epoch": 2.09, "grad_norm": 11.921433448791504, "learning_rate": 4.4880881238954465e-06, "loss": 0.6372, "step": 12957 }, { "epoch": 2.09, "grad_norm": 13.034584999084473, "learning_rate": 4.4866380639432475e-06, "loss": 0.7937, "step": 12958 }, { "epoch": 2.09, "grad_norm": 11.434906005859375, "learning_rate": 4.485188170528375e-06, "loss": 0.9509, "step": 12959 }, { "epoch": 2.09, "grad_norm": 12.3272123336792, "learning_rate": 4.483738443694623e-06, "loss": 0.5724, "step": 12960 }, { "epoch": 2.09, "grad_norm": 12.511157989501953, "learning_rate": 4.482288883485778e-06, "loss": 0.9319, "step": 12961 }, { "epoch": 2.09, "grad_norm": 11.791299819946289, "learning_rate": 4.4808394899456334e-06, "loss": 0.8156, "step": 12962 }, { "epoch": 2.09, "grad_norm": 10.741766929626465, "learning_rate": 4.479390263117964e-06, "loss": 0.6466, "step": 12963 }, { "epoch": 2.09, "grad_norm": 11.643645286560059, "learning_rate": 4.477941203046544e-06, "loss": 0.7249, "step": 12964 }, { "epoch": 2.09, "grad_norm": 13.88283634185791, "learning_rate": 4.476492309775151e-06, "loss": 0.643, "step": 12965 }, { "epoch": 2.09, "grad_norm": 11.624343872070312, "learning_rate": 4.475043583347542e-06, "loss": 0.6231, "step": 12966 }, { "epoch": 2.09, "grad_norm": 11.294124603271484, "learning_rate": 4.473595023807485e-06, "loss": 0.9077, "step": 12967 }, { "epoch": 2.09, "grad_norm": 10.66901969909668, "learning_rate": 4.472146631198733e-06, "loss": 0.7683, "step": 12968 }, { "epoch": 2.09, "grad_norm": 10.401240348815918, "learning_rate": 4.4706984055650305e-06, "loss": 0.5099, "step": 12969 }, { "epoch": 2.09, "grad_norm": 11.179754257202148, "learning_rate": 4.469250346950131e-06, "loss": 0.6526, "step": 12970 }, { "epoch": 2.09, "grad_norm": 12.523865699768066, "learning_rate": 4.467802455397769e-06, "loss": 0.7287, "step": 12971 }, { "epoch": 2.09, "grad_norm": 10.350992202758789, "learning_rate": 4.466354730951686e-06, "loss": 0.7766, "step": 12972 }, { "epoch": 2.09, "grad_norm": 12.834651947021484, "learning_rate": 4.4649071736556025e-06, "loss": 0.6305, "step": 12973 }, { "epoch": 2.09, "grad_norm": 13.109524726867676, "learning_rate": 4.463459783553255e-06, "loss": 0.7178, "step": 12974 }, { "epoch": 2.09, "grad_norm": 11.40176773071289, "learning_rate": 4.462012560688354e-06, "loss": 0.7101, "step": 12975 }, { "epoch": 2.09, "grad_norm": 11.241202354431152, "learning_rate": 4.46056550510462e-06, "loss": 0.8084, "step": 12976 }, { "epoch": 2.09, "grad_norm": 10.676040649414062, "learning_rate": 4.45911861684576e-06, "loss": 0.8842, "step": 12977 }, { "epoch": 2.09, "grad_norm": 13.231348991394043, "learning_rate": 4.4576718959554836e-06, "loss": 0.5302, "step": 12978 }, { "epoch": 2.09, "grad_norm": 13.218182563781738, "learning_rate": 4.456225342477486e-06, "loss": 0.7408, "step": 12979 }, { "epoch": 2.09, "grad_norm": 12.290169715881348, "learning_rate": 4.45477895645546e-06, "loss": 0.6891, "step": 12980 }, { "epoch": 2.09, "grad_norm": 11.040861129760742, "learning_rate": 4.453332737933102e-06, "loss": 0.4959, "step": 12981 }, { "epoch": 2.09, "grad_norm": 9.960318565368652, "learning_rate": 4.451886686954091e-06, "loss": 0.6917, "step": 12982 }, { "epoch": 2.09, "grad_norm": 13.866846084594727, "learning_rate": 4.450440803562112e-06, "loss": 0.9413, "step": 12983 }, { "epoch": 2.09, "grad_norm": 12.000985145568848, "learning_rate": 4.448995087800835e-06, "loss": 0.5938, "step": 12984 }, { "epoch": 2.09, "grad_norm": 12.058096885681152, "learning_rate": 4.447549539713928e-06, "loss": 0.6138, "step": 12985 }, { "epoch": 2.09, "grad_norm": 11.2161865234375, "learning_rate": 4.446104159345063e-06, "loss": 0.6779, "step": 12986 }, { "epoch": 2.09, "grad_norm": 11.0939359664917, "learning_rate": 4.4446589467378935e-06, "loss": 0.6564, "step": 12987 }, { "epoch": 2.09, "grad_norm": 12.080598831176758, "learning_rate": 4.443213901936071e-06, "loss": 0.7017, "step": 12988 }, { "epoch": 2.09, "grad_norm": 14.061561584472656, "learning_rate": 4.441769024983253e-06, "loss": 0.5711, "step": 12989 }, { "epoch": 2.09, "grad_norm": 13.073915481567383, "learning_rate": 4.440324315923074e-06, "loss": 0.6274, "step": 12990 }, { "epoch": 2.09, "grad_norm": 10.502005577087402, "learning_rate": 4.438879774799184e-06, "loss": 0.4987, "step": 12991 }, { "epoch": 2.09, "grad_norm": 13.268001556396484, "learning_rate": 4.4374354016552105e-06, "loss": 0.6472, "step": 12992 }, { "epoch": 2.09, "grad_norm": 12.355119705200195, "learning_rate": 4.435991196534775e-06, "loss": 0.8311, "step": 12993 }, { "epoch": 2.09, "grad_norm": 12.415160179138184, "learning_rate": 4.434547159481518e-06, "loss": 0.5839, "step": 12994 }, { "epoch": 2.09, "grad_norm": 13.079767227172852, "learning_rate": 4.43310329053905e-06, "loss": 0.8118, "step": 12995 }, { "epoch": 2.09, "grad_norm": 11.517544746398926, "learning_rate": 4.43165958975098e-06, "loss": 0.6613, "step": 12996 }, { "epoch": 2.09, "grad_norm": 11.84786319732666, "learning_rate": 4.430216057160926e-06, "loss": 0.6304, "step": 12997 }, { "epoch": 2.09, "grad_norm": 12.832987785339355, "learning_rate": 4.428772692812485e-06, "loss": 0.8324, "step": 12998 }, { "epoch": 2.09, "grad_norm": 15.623682022094727, "learning_rate": 4.427329496749255e-06, "loss": 0.7797, "step": 12999 }, { "epoch": 2.09, "grad_norm": 12.922292709350586, "learning_rate": 4.425886469014834e-06, "loss": 0.6551, "step": 13000 }, { "epoch": 2.09, "grad_norm": 12.3284330368042, "learning_rate": 4.424443609652805e-06, "loss": 0.7159, "step": 13001 }, { "epoch": 2.09, "grad_norm": 13.03540325164795, "learning_rate": 4.423000918706758e-06, "loss": 0.6805, "step": 13002 }, { "epoch": 2.09, "grad_norm": 13.336910247802734, "learning_rate": 4.421558396220265e-06, "loss": 0.68, "step": 13003 }, { "epoch": 2.09, "grad_norm": 13.84523868560791, "learning_rate": 4.4201160422369e-06, "loss": 0.8575, "step": 13004 }, { "epoch": 2.09, "grad_norm": 10.362661361694336, "learning_rate": 4.418673856800234e-06, "loss": 0.863, "step": 13005 }, { "epoch": 2.09, "grad_norm": 11.50277328491211, "learning_rate": 4.417231839953825e-06, "loss": 0.7589, "step": 13006 }, { "epoch": 2.09, "grad_norm": 11.886651039123535, "learning_rate": 4.415789991741238e-06, "loss": 0.8476, "step": 13007 }, { "epoch": 2.09, "grad_norm": 12.451122283935547, "learning_rate": 4.41434831220602e-06, "loss": 0.7862, "step": 13008 }, { "epoch": 2.09, "grad_norm": 9.882279396057129, "learning_rate": 4.4129068013917165e-06, "loss": 0.6924, "step": 13009 }, { "epoch": 2.09, "grad_norm": 11.367234230041504, "learning_rate": 4.411465459341876e-06, "loss": 0.6683, "step": 13010 }, { "epoch": 2.09, "grad_norm": 39.0401496887207, "learning_rate": 4.410024286100034e-06, "loss": 0.5803, "step": 13011 }, { "epoch": 2.09, "grad_norm": 36.47720718383789, "learning_rate": 4.408583281709718e-06, "loss": 0.6982, "step": 13012 }, { "epoch": 2.09, "grad_norm": 10.175093650817871, "learning_rate": 4.407142446214461e-06, "loss": 0.684, "step": 13013 }, { "epoch": 2.09, "grad_norm": 11.83129596710205, "learning_rate": 4.405701779657785e-06, "loss": 0.6479, "step": 13014 }, { "epoch": 2.1, "grad_norm": 12.242757797241211, "learning_rate": 4.404261282083201e-06, "loss": 0.7304, "step": 13015 }, { "epoch": 2.1, "grad_norm": 10.64401912689209, "learning_rate": 4.402820953534229e-06, "loss": 0.644, "step": 13016 }, { "epoch": 2.1, "grad_norm": 13.839162826538086, "learning_rate": 4.40138079405437e-06, "loss": 0.7975, "step": 13017 }, { "epoch": 2.1, "grad_norm": 9.268471717834473, "learning_rate": 4.399940803687128e-06, "loss": 0.6635, "step": 13018 }, { "epoch": 2.1, "grad_norm": 11.187289237976074, "learning_rate": 4.398500982476e-06, "loss": 0.9019, "step": 13019 }, { "epoch": 2.1, "grad_norm": 12.39281940460205, "learning_rate": 4.397061330464474e-06, "loss": 0.8652, "step": 13020 }, { "epoch": 2.1, "grad_norm": 10.701412200927734, "learning_rate": 4.395621847696041e-06, "loss": 0.8734, "step": 13021 }, { "epoch": 2.1, "grad_norm": 11.712069511413574, "learning_rate": 4.3941825342141805e-06, "loss": 0.7596, "step": 13022 }, { "epoch": 2.1, "grad_norm": 10.454383850097656, "learning_rate": 4.392743390062364e-06, "loss": 0.5845, "step": 13023 }, { "epoch": 2.1, "grad_norm": 9.039703369140625, "learning_rate": 4.39130441528407e-06, "loss": 0.611, "step": 13024 }, { "epoch": 2.1, "grad_norm": 12.906179428100586, "learning_rate": 4.389865609922758e-06, "loss": 0.71, "step": 13025 }, { "epoch": 2.1, "grad_norm": 11.842355728149414, "learning_rate": 4.388426974021893e-06, "loss": 0.6955, "step": 13026 }, { "epoch": 2.1, "grad_norm": 11.494688034057617, "learning_rate": 4.386988507624931e-06, "loss": 0.6132, "step": 13027 }, { "epoch": 2.1, "grad_norm": 10.956018447875977, "learning_rate": 4.385550210775317e-06, "loss": 0.6741, "step": 13028 }, { "epoch": 2.1, "grad_norm": 10.533178329467773, "learning_rate": 4.384112083516503e-06, "loss": 0.6536, "step": 13029 }, { "epoch": 2.1, "grad_norm": 11.647600173950195, "learning_rate": 4.382674125891926e-06, "loss": 0.884, "step": 13030 }, { "epoch": 2.1, "grad_norm": 13.422847747802734, "learning_rate": 4.381236337945018e-06, "loss": 0.6377, "step": 13031 }, { "epoch": 2.1, "grad_norm": 12.088994979858398, "learning_rate": 4.379798719719216e-06, "loss": 0.8109, "step": 13032 }, { "epoch": 2.1, "grad_norm": 10.156963348388672, "learning_rate": 4.3783612712579386e-06, "loss": 0.4777, "step": 13033 }, { "epoch": 2.1, "grad_norm": 13.863247871398926, "learning_rate": 4.376923992604607e-06, "loss": 0.8797, "step": 13034 }, { "epoch": 2.1, "grad_norm": 12.365462303161621, "learning_rate": 4.375486883802641e-06, "loss": 0.8033, "step": 13035 }, { "epoch": 2.1, "grad_norm": 13.667513847351074, "learning_rate": 4.374049944895443e-06, "loss": 0.8082, "step": 13036 }, { "epoch": 2.1, "grad_norm": 11.809648513793945, "learning_rate": 4.372613175926423e-06, "loss": 0.5721, "step": 13037 }, { "epoch": 2.1, "grad_norm": 11.142278671264648, "learning_rate": 4.371176576938979e-06, "loss": 0.5769, "step": 13038 }, { "epoch": 2.1, "grad_norm": 14.131026268005371, "learning_rate": 4.3697401479765e-06, "loss": 0.6107, "step": 13039 }, { "epoch": 2.1, "grad_norm": 9.729517936706543, "learning_rate": 4.368303889082383e-06, "loss": 0.4587, "step": 13040 }, { "epoch": 2.1, "grad_norm": 15.108165740966797, "learning_rate": 4.366867800300006e-06, "loss": 0.5648, "step": 13041 }, { "epoch": 2.1, "grad_norm": 12.706459999084473, "learning_rate": 4.365431881672748e-06, "loss": 0.6876, "step": 13042 }, { "epoch": 2.1, "grad_norm": 13.244943618774414, "learning_rate": 4.3639961332439865e-06, "loss": 0.6744, "step": 13043 }, { "epoch": 2.1, "grad_norm": 10.80400562286377, "learning_rate": 4.362560555057083e-06, "loss": 0.4846, "step": 13044 }, { "epoch": 2.1, "grad_norm": 14.040507316589355, "learning_rate": 4.361125147155408e-06, "loss": 0.6334, "step": 13045 }, { "epoch": 2.1, "grad_norm": 14.033916473388672, "learning_rate": 4.359689909582319e-06, "loss": 0.702, "step": 13046 }, { "epoch": 2.1, "grad_norm": 13.636590003967285, "learning_rate": 4.358254842381161e-06, "loss": 0.7483, "step": 13047 }, { "epoch": 2.1, "grad_norm": 15.893241882324219, "learning_rate": 4.3568199455952916e-06, "loss": 0.8349, "step": 13048 }, { "epoch": 2.1, "grad_norm": 11.984163284301758, "learning_rate": 4.355385219268046e-06, "loss": 0.8711, "step": 13049 }, { "epoch": 2.1, "grad_norm": 12.987523078918457, "learning_rate": 4.353950663442768e-06, "loss": 0.6289, "step": 13050 }, { "epoch": 2.1, "grad_norm": 11.2054443359375, "learning_rate": 4.352516278162788e-06, "loss": 0.5321, "step": 13051 }, { "epoch": 2.1, "grad_norm": 10.330416679382324, "learning_rate": 4.351082063471428e-06, "loss": 0.4428, "step": 13052 }, { "epoch": 2.1, "grad_norm": 11.346882820129395, "learning_rate": 4.3496480194120185e-06, "loss": 0.8193, "step": 13053 }, { "epoch": 2.1, "grad_norm": 11.857279777526855, "learning_rate": 4.348214146027868e-06, "loss": 0.9469, "step": 13054 }, { "epoch": 2.1, "grad_norm": 10.894936561584473, "learning_rate": 4.346780443362293e-06, "loss": 0.5797, "step": 13055 }, { "epoch": 2.1, "grad_norm": 12.87219524383545, "learning_rate": 4.3453469114586024e-06, "loss": 0.63, "step": 13056 }, { "epoch": 2.1, "grad_norm": 12.49949836730957, "learning_rate": 4.343913550360096e-06, "loss": 0.5857, "step": 13057 }, { "epoch": 2.1, "grad_norm": 11.685762405395508, "learning_rate": 4.342480360110064e-06, "loss": 0.5943, "step": 13058 }, { "epoch": 2.1, "grad_norm": 10.591136932373047, "learning_rate": 4.341047340751808e-06, "loss": 0.708, "step": 13059 }, { "epoch": 2.1, "grad_norm": 11.589292526245117, "learning_rate": 4.3396144923286034e-06, "loss": 0.6881, "step": 13060 }, { "epoch": 2.1, "grad_norm": 10.813325881958008, "learning_rate": 4.33818181488374e-06, "loss": 0.73, "step": 13061 }, { "epoch": 2.1, "grad_norm": 10.51683521270752, "learning_rate": 4.3367493084604895e-06, "loss": 0.6383, "step": 13062 }, { "epoch": 2.1, "grad_norm": 11.566009521484375, "learning_rate": 4.3353169731021194e-06, "loss": 0.8162, "step": 13063 }, { "epoch": 2.1, "grad_norm": 11.754449844360352, "learning_rate": 4.3338848088519e-06, "loss": 0.8075, "step": 13064 }, { "epoch": 2.1, "grad_norm": 11.000585556030273, "learning_rate": 4.332452815753091e-06, "loss": 0.5992, "step": 13065 }, { "epoch": 2.1, "grad_norm": 13.650142669677734, "learning_rate": 4.331020993848942e-06, "loss": 0.5932, "step": 13066 }, { "epoch": 2.1, "grad_norm": 14.137187004089355, "learning_rate": 4.329589343182709e-06, "loss": 0.598, "step": 13067 }, { "epoch": 2.1, "grad_norm": 10.553098678588867, "learning_rate": 4.328157863797632e-06, "loss": 0.7891, "step": 13068 }, { "epoch": 2.1, "grad_norm": 11.36505126953125, "learning_rate": 4.326726555736956e-06, "loss": 0.9457, "step": 13069 }, { "epoch": 2.1, "grad_norm": 13.51855182647705, "learning_rate": 4.325295419043912e-06, "loss": 0.8586, "step": 13070 }, { "epoch": 2.1, "grad_norm": 12.098846435546875, "learning_rate": 4.323864453761725e-06, "loss": 0.8194, "step": 13071 }, { "epoch": 2.1, "grad_norm": 11.705934524536133, "learning_rate": 4.322433659933627e-06, "loss": 0.4483, "step": 13072 }, { "epoch": 2.1, "grad_norm": 12.582748413085938, "learning_rate": 4.321003037602829e-06, "loss": 0.6223, "step": 13073 }, { "epoch": 2.1, "grad_norm": 12.45284652709961, "learning_rate": 4.319572586812548e-06, "loss": 0.6347, "step": 13074 }, { "epoch": 2.1, "grad_norm": 11.816912651062012, "learning_rate": 4.318142307605997e-06, "loss": 0.6528, "step": 13075 }, { "epoch": 2.1, "grad_norm": 10.901350975036621, "learning_rate": 4.316712200026374e-06, "loss": 0.6986, "step": 13076 }, { "epoch": 2.11, "grad_norm": 13.37185001373291, "learning_rate": 4.3152822641168745e-06, "loss": 0.9577, "step": 13077 }, { "epoch": 2.11, "grad_norm": 15.133543968200684, "learning_rate": 4.313852499920699e-06, "loss": 0.7196, "step": 13078 }, { "epoch": 2.11, "grad_norm": 13.959724426269531, "learning_rate": 4.312422907481026e-06, "loss": 0.7316, "step": 13079 }, { "epoch": 2.11, "grad_norm": 17.937604904174805, "learning_rate": 4.310993486841047e-06, "loss": 0.5665, "step": 13080 }, { "epoch": 2.11, "grad_norm": 10.644942283630371, "learning_rate": 4.309564238043935e-06, "loss": 0.4858, "step": 13081 }, { "epoch": 2.11, "grad_norm": 12.608840942382812, "learning_rate": 4.308135161132859e-06, "loss": 0.7358, "step": 13082 }, { "epoch": 2.11, "grad_norm": 11.104070663452148, "learning_rate": 4.306706256150992e-06, "loss": 0.6037, "step": 13083 }, { "epoch": 2.11, "grad_norm": 12.898590087890625, "learning_rate": 4.305277523141489e-06, "loss": 0.617, "step": 13084 }, { "epoch": 2.11, "grad_norm": 13.705656051635742, "learning_rate": 4.303848962147515e-06, "loss": 0.7079, "step": 13085 }, { "epoch": 2.11, "grad_norm": 13.817610740661621, "learning_rate": 4.302420573212216e-06, "loss": 0.7644, "step": 13086 }, { "epoch": 2.11, "grad_norm": 14.270401954650879, "learning_rate": 4.300992356378734e-06, "loss": 0.8538, "step": 13087 }, { "epoch": 2.11, "grad_norm": 9.770310401916504, "learning_rate": 4.2995643116902185e-06, "loss": 0.7887, "step": 13088 }, { "epoch": 2.11, "grad_norm": 11.400402069091797, "learning_rate": 4.298136439189802e-06, "loss": 0.7681, "step": 13089 }, { "epoch": 2.11, "grad_norm": 10.487993240356445, "learning_rate": 4.29670873892061e-06, "loss": 0.6328, "step": 13090 }, { "epoch": 2.11, "grad_norm": 14.29349422454834, "learning_rate": 4.295281210925776e-06, "loss": 0.541, "step": 13091 }, { "epoch": 2.11, "grad_norm": 13.703295707702637, "learning_rate": 4.293853855248412e-06, "loss": 0.5961, "step": 13092 }, { "epoch": 2.11, "grad_norm": 16.88484764099121, "learning_rate": 4.292426671931641e-06, "loss": 0.7612, "step": 13093 }, { "epoch": 2.11, "grad_norm": 10.576945304870605, "learning_rate": 4.290999661018566e-06, "loss": 0.5184, "step": 13094 }, { "epoch": 2.11, "grad_norm": 15.430489540100098, "learning_rate": 4.289572822552293e-06, "loss": 0.6973, "step": 13095 }, { "epoch": 2.11, "grad_norm": 12.53591251373291, "learning_rate": 4.2881461565759254e-06, "loss": 0.6908, "step": 13096 }, { "epoch": 2.11, "grad_norm": 12.739206314086914, "learning_rate": 4.2867196631325546e-06, "loss": 0.7297, "step": 13097 }, { "epoch": 2.11, "grad_norm": 13.955353736877441, "learning_rate": 4.285293342265266e-06, "loss": 0.7594, "step": 13098 }, { "epoch": 2.11, "grad_norm": 14.570098876953125, "learning_rate": 4.283867194017151e-06, "loss": 0.6775, "step": 13099 }, { "epoch": 2.11, "grad_norm": 12.167202949523926, "learning_rate": 4.282441218431281e-06, "loss": 0.6439, "step": 13100 }, { "epoch": 2.11, "grad_norm": 13.481572151184082, "learning_rate": 4.28101541555073e-06, "loss": 0.582, "step": 13101 }, { "epoch": 2.11, "grad_norm": 10.707139015197754, "learning_rate": 4.27958978541857e-06, "loss": 0.5984, "step": 13102 }, { "epoch": 2.11, "grad_norm": 13.203180313110352, "learning_rate": 4.278164328077859e-06, "loss": 0.5782, "step": 13103 }, { "epoch": 2.11, "grad_norm": 12.561996459960938, "learning_rate": 4.2767390435716595e-06, "loss": 0.6415, "step": 13104 }, { "epoch": 2.11, "grad_norm": 11.048269271850586, "learning_rate": 4.2753139319430215e-06, "loss": 0.7353, "step": 13105 }, { "epoch": 2.11, "grad_norm": 12.999939918518066, "learning_rate": 4.273888993234988e-06, "loss": 0.7823, "step": 13106 }, { "epoch": 2.11, "grad_norm": 11.294305801391602, "learning_rate": 4.272464227490608e-06, "loss": 0.6373, "step": 13107 }, { "epoch": 2.11, "grad_norm": 11.00481128692627, "learning_rate": 4.271039634752915e-06, "loss": 0.6473, "step": 13108 }, { "epoch": 2.11, "grad_norm": 13.542494773864746, "learning_rate": 4.269615215064937e-06, "loss": 0.8527, "step": 13109 }, { "epoch": 2.11, "grad_norm": 12.015798568725586, "learning_rate": 4.268190968469707e-06, "loss": 0.609, "step": 13110 }, { "epoch": 2.11, "grad_norm": 16.169462203979492, "learning_rate": 4.266766895010239e-06, "loss": 0.592, "step": 13111 }, { "epoch": 2.11, "grad_norm": 9.636604309082031, "learning_rate": 4.265342994729557e-06, "loss": 0.6052, "step": 13112 }, { "epoch": 2.11, "grad_norm": 12.736017227172852, "learning_rate": 4.263919267670664e-06, "loss": 0.7931, "step": 13113 }, { "epoch": 2.11, "grad_norm": 12.958231925964355, "learning_rate": 4.262495713876563e-06, "loss": 0.8403, "step": 13114 }, { "epoch": 2.11, "grad_norm": 14.059408187866211, "learning_rate": 4.261072333390266e-06, "loss": 0.774, "step": 13115 }, { "epoch": 2.11, "grad_norm": 11.55235481262207, "learning_rate": 4.25964912625476e-06, "loss": 0.8995, "step": 13116 }, { "epoch": 2.11, "grad_norm": 10.475696563720703, "learning_rate": 4.258226092513032e-06, "loss": 0.7436, "step": 13117 }, { "epoch": 2.11, "grad_norm": 14.980792999267578, "learning_rate": 4.256803232208074e-06, "loss": 0.5381, "step": 13118 }, { "epoch": 2.11, "grad_norm": 13.913346290588379, "learning_rate": 4.2553805453828614e-06, "loss": 0.7334, "step": 13119 }, { "epoch": 2.11, "grad_norm": 11.782927513122559, "learning_rate": 4.253958032080363e-06, "loss": 0.9859, "step": 13120 }, { "epoch": 2.11, "grad_norm": 10.402548789978027, "learning_rate": 4.2525356923435555e-06, "loss": 0.5321, "step": 13121 }, { "epoch": 2.11, "grad_norm": 10.396164894104004, "learning_rate": 4.251113526215395e-06, "loss": 0.4196, "step": 13122 }, { "epoch": 2.11, "grad_norm": 12.88965892791748, "learning_rate": 4.249691533738849e-06, "loss": 0.6633, "step": 13123 }, { "epoch": 2.11, "grad_norm": 12.519240379333496, "learning_rate": 4.248269714956863e-06, "loss": 0.6049, "step": 13124 }, { "epoch": 2.11, "grad_norm": 12.947442054748535, "learning_rate": 4.246848069912383e-06, "loss": 0.6314, "step": 13125 }, { "epoch": 2.11, "grad_norm": 13.154919624328613, "learning_rate": 4.245426598648359e-06, "loss": 0.6025, "step": 13126 }, { "epoch": 2.11, "grad_norm": 12.943358421325684, "learning_rate": 4.24400530120772e-06, "loss": 0.8796, "step": 13127 }, { "epoch": 2.11, "grad_norm": 9.272388458251953, "learning_rate": 4.242584177633406e-06, "loss": 0.5456, "step": 13128 }, { "epoch": 2.11, "grad_norm": 12.806514739990234, "learning_rate": 4.241163227968338e-06, "loss": 0.7104, "step": 13129 }, { "epoch": 2.11, "grad_norm": 11.831007957458496, "learning_rate": 4.239742452255437e-06, "loss": 0.6555, "step": 13130 }, { "epoch": 2.11, "grad_norm": 11.686688423156738, "learning_rate": 4.238321850537624e-06, "loss": 0.7427, "step": 13131 }, { "epoch": 2.11, "grad_norm": 12.283079147338867, "learning_rate": 4.236901422857807e-06, "loss": 0.6803, "step": 13132 }, { "epoch": 2.11, "grad_norm": 10.685929298400879, "learning_rate": 4.235481169258887e-06, "loss": 0.5808, "step": 13133 }, { "epoch": 2.11, "grad_norm": 11.868038177490234, "learning_rate": 4.234061089783773e-06, "loss": 0.6488, "step": 13134 }, { "epoch": 2.11, "grad_norm": 13.037371635437012, "learning_rate": 4.2326411844753534e-06, "loss": 0.6655, "step": 13135 }, { "epoch": 2.11, "grad_norm": 12.052841186523438, "learning_rate": 4.23122145337652e-06, "loss": 0.7291, "step": 13136 }, { "epoch": 2.11, "grad_norm": 13.117271423339844, "learning_rate": 4.229801896530161e-06, "loss": 0.6639, "step": 13137 }, { "epoch": 2.11, "grad_norm": 10.597001075744629, "learning_rate": 4.22838251397915e-06, "loss": 0.7862, "step": 13138 }, { "epoch": 2.12, "grad_norm": 12.100082397460938, "learning_rate": 4.226963305766367e-06, "loss": 0.7852, "step": 13139 }, { "epoch": 2.12, "grad_norm": 15.457925796508789, "learning_rate": 4.225544271934679e-06, "loss": 0.6087, "step": 13140 }, { "epoch": 2.12, "grad_norm": 14.164060592651367, "learning_rate": 4.224125412526943e-06, "loss": 0.6203, "step": 13141 }, { "epoch": 2.12, "grad_norm": 10.454606056213379, "learning_rate": 4.2227067275860254e-06, "loss": 0.5313, "step": 13142 }, { "epoch": 2.12, "grad_norm": 11.783950805664062, "learning_rate": 4.221288217154777e-06, "loss": 0.6963, "step": 13143 }, { "epoch": 2.12, "grad_norm": 10.765212059020996, "learning_rate": 4.219869881276041e-06, "loss": 0.6353, "step": 13144 }, { "epoch": 2.12, "grad_norm": 11.911127090454102, "learning_rate": 4.218451719992668e-06, "loss": 0.7481, "step": 13145 }, { "epoch": 2.12, "grad_norm": 10.579022407531738, "learning_rate": 4.217033733347485e-06, "loss": 0.6501, "step": 13146 }, { "epoch": 2.12, "grad_norm": 12.16681957244873, "learning_rate": 4.2156159213833345e-06, "loss": 0.8718, "step": 13147 }, { "epoch": 2.12, "grad_norm": 11.049644470214844, "learning_rate": 4.214198284143039e-06, "loss": 0.6259, "step": 13148 }, { "epoch": 2.12, "grad_norm": 11.613667488098145, "learning_rate": 4.212780821669413e-06, "loss": 0.7024, "step": 13149 }, { "epoch": 2.12, "grad_norm": 14.423735618591309, "learning_rate": 4.211363534005284e-06, "loss": 0.7401, "step": 13150 }, { "epoch": 2.12, "grad_norm": 13.019482612609863, "learning_rate": 4.209946421193454e-06, "loss": 0.6305, "step": 13151 }, { "epoch": 2.12, "grad_norm": 12.230917930603027, "learning_rate": 4.208529483276734e-06, "loss": 0.6459, "step": 13152 }, { "epoch": 2.12, "grad_norm": 13.21733283996582, "learning_rate": 4.207112720297923e-06, "loss": 0.7882, "step": 13153 }, { "epoch": 2.12, "grad_norm": 10.950257301330566, "learning_rate": 4.205696132299812e-06, "loss": 0.5224, "step": 13154 }, { "epoch": 2.12, "grad_norm": 10.857575416564941, "learning_rate": 4.2042797193251916e-06, "loss": 0.7235, "step": 13155 }, { "epoch": 2.12, "grad_norm": 13.434481620788574, "learning_rate": 4.202863481416852e-06, "loss": 0.7551, "step": 13156 }, { "epoch": 2.12, "grad_norm": 11.181676864624023, "learning_rate": 4.201447418617566e-06, "loss": 0.725, "step": 13157 }, { "epoch": 2.12, "grad_norm": 13.382033348083496, "learning_rate": 4.200031530970113e-06, "loss": 0.8748, "step": 13158 }, { "epoch": 2.12, "grad_norm": 9.380223274230957, "learning_rate": 4.198615818517259e-06, "loss": 0.8645, "step": 13159 }, { "epoch": 2.12, "grad_norm": 11.772706031799316, "learning_rate": 4.197200281301762e-06, "loss": 0.7606, "step": 13160 }, { "epoch": 2.12, "grad_norm": 14.033838272094727, "learning_rate": 4.195784919366388e-06, "loss": 0.8275, "step": 13161 }, { "epoch": 2.12, "grad_norm": 14.447166442871094, "learning_rate": 4.194369732753882e-06, "loss": 0.7442, "step": 13162 }, { "epoch": 2.12, "grad_norm": 12.081453323364258, "learning_rate": 4.192954721506999e-06, "loss": 0.6628, "step": 13163 }, { "epoch": 2.12, "grad_norm": 11.047130584716797, "learning_rate": 4.191539885668478e-06, "loss": 0.7641, "step": 13164 }, { "epoch": 2.12, "grad_norm": 17.465051651000977, "learning_rate": 4.190125225281051e-06, "loss": 0.8094, "step": 13165 }, { "epoch": 2.12, "grad_norm": 12.793261528015137, "learning_rate": 4.188710740387457e-06, "loss": 0.7774, "step": 13166 }, { "epoch": 2.12, "grad_norm": 12.599306106567383, "learning_rate": 4.187296431030419e-06, "loss": 0.8063, "step": 13167 }, { "epoch": 2.12, "grad_norm": 11.700716972351074, "learning_rate": 4.185882297252653e-06, "loss": 0.7782, "step": 13168 }, { "epoch": 2.12, "grad_norm": 11.834148406982422, "learning_rate": 4.184468339096882e-06, "loss": 0.7789, "step": 13169 }, { "epoch": 2.12, "grad_norm": 12.084786415100098, "learning_rate": 4.183054556605811e-06, "loss": 0.7312, "step": 13170 }, { "epoch": 2.12, "grad_norm": 11.460000991821289, "learning_rate": 4.18164094982215e-06, "loss": 0.714, "step": 13171 }, { "epoch": 2.12, "grad_norm": 13.80907154083252, "learning_rate": 4.180227518788595e-06, "loss": 0.7007, "step": 13172 }, { "epoch": 2.12, "grad_norm": 13.9202880859375, "learning_rate": 4.178814263547837e-06, "loss": 0.6702, "step": 13173 }, { "epoch": 2.12, "grad_norm": 11.119004249572754, "learning_rate": 4.1774011841425735e-06, "loss": 0.7888, "step": 13174 }, { "epoch": 2.12, "grad_norm": 13.57375717163086, "learning_rate": 4.17598828061548e-06, "loss": 0.8872, "step": 13175 }, { "epoch": 2.12, "grad_norm": 12.179381370544434, "learning_rate": 4.174575553009237e-06, "loss": 0.7273, "step": 13176 }, { "epoch": 2.12, "grad_norm": 11.205327033996582, "learning_rate": 4.173163001366522e-06, "loss": 0.6287, "step": 13177 }, { "epoch": 2.12, "grad_norm": 11.715020179748535, "learning_rate": 4.171750625730001e-06, "loss": 0.6833, "step": 13178 }, { "epoch": 2.12, "grad_norm": 12.331339836120605, "learning_rate": 4.170338426142331e-06, "loss": 0.8265, "step": 13179 }, { "epoch": 2.12, "grad_norm": 13.655326843261719, "learning_rate": 4.168926402646176e-06, "loss": 0.7161, "step": 13180 }, { "epoch": 2.12, "grad_norm": 11.980544090270996, "learning_rate": 4.167514555284182e-06, "loss": 0.6761, "step": 13181 }, { "epoch": 2.12, "grad_norm": 10.492053985595703, "learning_rate": 4.166102884099002e-06, "loss": 0.8954, "step": 13182 }, { "epoch": 2.12, "grad_norm": 11.54273796081543, "learning_rate": 4.164691389133273e-06, "loss": 0.6264, "step": 13183 }, { "epoch": 2.12, "grad_norm": 11.592178344726562, "learning_rate": 4.163280070429627e-06, "loss": 0.6939, "step": 13184 }, { "epoch": 2.12, "grad_norm": 11.706202507019043, "learning_rate": 4.161868928030703e-06, "loss": 0.7202, "step": 13185 }, { "epoch": 2.12, "grad_norm": 11.293428421020508, "learning_rate": 4.1604579619791215e-06, "loss": 0.8584, "step": 13186 }, { "epoch": 2.12, "grad_norm": 12.139104843139648, "learning_rate": 4.1590471723175e-06, "loss": 0.7173, "step": 13187 }, { "epoch": 2.12, "grad_norm": 13.548622131347656, "learning_rate": 4.157636559088458e-06, "loss": 0.7458, "step": 13188 }, { "epoch": 2.12, "grad_norm": 12.912568092346191, "learning_rate": 4.156226122334599e-06, "loss": 0.5689, "step": 13189 }, { "epoch": 2.12, "grad_norm": 13.735430717468262, "learning_rate": 4.154815862098533e-06, "loss": 0.5491, "step": 13190 }, { "epoch": 2.12, "grad_norm": 14.004851341247559, "learning_rate": 4.153405778422856e-06, "loss": 0.8133, "step": 13191 }, { "epoch": 2.12, "grad_norm": 12.227031707763672, "learning_rate": 4.1519958713501585e-06, "loss": 0.8691, "step": 13192 }, { "epoch": 2.12, "grad_norm": 12.598095893859863, "learning_rate": 4.150586140923033e-06, "loss": 0.6939, "step": 13193 }, { "epoch": 2.12, "grad_norm": 11.206369400024414, "learning_rate": 4.149176587184056e-06, "loss": 0.5978, "step": 13194 }, { "epoch": 2.12, "grad_norm": 12.691729545593262, "learning_rate": 4.147767210175808e-06, "loss": 0.7737, "step": 13195 }, { "epoch": 2.12, "grad_norm": 13.552940368652344, "learning_rate": 4.146358009940866e-06, "loss": 0.7507, "step": 13196 }, { "epoch": 2.12, "grad_norm": 11.257843971252441, "learning_rate": 4.1449489865217904e-06, "loss": 0.7185, "step": 13197 }, { "epoch": 2.12, "grad_norm": 13.098954200744629, "learning_rate": 4.14354013996114e-06, "loss": 0.6974, "step": 13198 }, { "epoch": 2.12, "grad_norm": 10.148655891418457, "learning_rate": 4.142131470301478e-06, "loss": 0.6736, "step": 13199 }, { "epoch": 2.12, "grad_norm": 11.630424499511719, "learning_rate": 4.140722977585349e-06, "loss": 0.6323, "step": 13200 }, { "epoch": 2.13, "grad_norm": 12.44581127166748, "learning_rate": 4.139314661855302e-06, "loss": 0.6688, "step": 13201 }, { "epoch": 2.13, "grad_norm": 13.369316101074219, "learning_rate": 4.137906523153874e-06, "loss": 0.7037, "step": 13202 }, { "epoch": 2.13, "grad_norm": 10.841238021850586, "learning_rate": 4.136498561523599e-06, "loss": 0.7468, "step": 13203 }, { "epoch": 2.13, "grad_norm": 15.599444389343262, "learning_rate": 4.1350907770070095e-06, "loss": 0.7954, "step": 13204 }, { "epoch": 2.13, "grad_norm": 12.400595664978027, "learning_rate": 4.1336831696466226e-06, "loss": 0.8291, "step": 13205 }, { "epoch": 2.13, "grad_norm": 13.0269775390625, "learning_rate": 4.132275739484967e-06, "loss": 0.8653, "step": 13206 }, { "epoch": 2.13, "grad_norm": 16.584213256835938, "learning_rate": 4.130868486564547e-06, "loss": 0.7441, "step": 13207 }, { "epoch": 2.13, "grad_norm": 12.726947784423828, "learning_rate": 4.129461410927872e-06, "loss": 0.676, "step": 13208 }, { "epoch": 2.13, "grad_norm": 12.412823677062988, "learning_rate": 4.1280545126174486e-06, "loss": 0.8104, "step": 13209 }, { "epoch": 2.13, "grad_norm": 13.491554260253906, "learning_rate": 4.126647791675771e-06, "loss": 0.5513, "step": 13210 }, { "epoch": 2.13, "grad_norm": 11.880386352539062, "learning_rate": 4.125241248145325e-06, "loss": 0.7554, "step": 13211 }, { "epoch": 2.13, "grad_norm": 12.146072387695312, "learning_rate": 4.123834882068608e-06, "loss": 0.6796, "step": 13212 }, { "epoch": 2.13, "grad_norm": 16.546449661254883, "learning_rate": 4.1224286934880895e-06, "loss": 0.6829, "step": 13213 }, { "epoch": 2.13, "grad_norm": 12.06916332244873, "learning_rate": 4.121022682446256e-06, "loss": 0.6896, "step": 13214 }, { "epoch": 2.13, "grad_norm": 11.68299674987793, "learning_rate": 4.119616848985572e-06, "loss": 0.7145, "step": 13215 }, { "epoch": 2.13, "grad_norm": 12.147544860839844, "learning_rate": 4.118211193148493e-06, "loss": 0.7676, "step": 13216 }, { "epoch": 2.13, "grad_norm": 13.159710884094238, "learning_rate": 4.1168057149774975e-06, "loss": 0.8334, "step": 13217 }, { "epoch": 2.13, "grad_norm": 11.51146125793457, "learning_rate": 4.11540041451503e-06, "loss": 0.6385, "step": 13218 }, { "epoch": 2.13, "grad_norm": 16.433351516723633, "learning_rate": 4.113995291803535e-06, "loss": 0.7283, "step": 13219 }, { "epoch": 2.13, "grad_norm": 14.753560066223145, "learning_rate": 4.112590346885464e-06, "loss": 0.5746, "step": 13220 }, { "epoch": 2.13, "grad_norm": 11.522761344909668, "learning_rate": 4.111185579803251e-06, "loss": 0.8219, "step": 13221 }, { "epoch": 2.13, "grad_norm": 11.579889297485352, "learning_rate": 4.109780990599325e-06, "loss": 0.6918, "step": 13222 }, { "epoch": 2.13, "grad_norm": 13.249029159545898, "learning_rate": 4.1083765793161205e-06, "loss": 0.7031, "step": 13223 }, { "epoch": 2.13, "grad_norm": 13.948681831359863, "learning_rate": 4.106972345996051e-06, "loss": 0.5591, "step": 13224 }, { "epoch": 2.13, "grad_norm": 10.579185485839844, "learning_rate": 4.105568290681542e-06, "loss": 0.5886, "step": 13225 }, { "epoch": 2.13, "grad_norm": 10.456690788269043, "learning_rate": 4.1041644134149996e-06, "loss": 0.6628, "step": 13226 }, { "epoch": 2.13, "grad_norm": 10.134514808654785, "learning_rate": 4.102760714238827e-06, "loss": 0.7487, "step": 13227 }, { "epoch": 2.13, "grad_norm": 10.184786796569824, "learning_rate": 4.101357193195431e-06, "loss": 0.6986, "step": 13228 }, { "epoch": 2.13, "grad_norm": 13.13692569732666, "learning_rate": 4.099953850327199e-06, "loss": 0.719, "step": 13229 }, { "epoch": 2.13, "grad_norm": 10.688353538513184, "learning_rate": 4.098550685676529e-06, "loss": 0.6793, "step": 13230 }, { "epoch": 2.13, "grad_norm": 11.390592575073242, "learning_rate": 4.0971476992858e-06, "loss": 0.71, "step": 13231 }, { "epoch": 2.13, "grad_norm": 11.441871643066406, "learning_rate": 4.09574489119739e-06, "loss": 0.6982, "step": 13232 }, { "epoch": 2.13, "grad_norm": 13.425838470458984, "learning_rate": 4.094342261453676e-06, "loss": 0.5754, "step": 13233 }, { "epoch": 2.13, "grad_norm": 11.215255737304688, "learning_rate": 4.092939810097023e-06, "loss": 0.7827, "step": 13234 }, { "epoch": 2.13, "grad_norm": 11.189996719360352, "learning_rate": 4.091537537169793e-06, "loss": 0.6475, "step": 13235 }, { "epoch": 2.13, "grad_norm": 13.138500213623047, "learning_rate": 4.090135442714344e-06, "loss": 0.5841, "step": 13236 }, { "epoch": 2.13, "grad_norm": 11.87710189819336, "learning_rate": 4.088733526773032e-06, "loss": 0.7867, "step": 13237 }, { "epoch": 2.13, "grad_norm": 11.6138334274292, "learning_rate": 4.087331789388197e-06, "loss": 0.6545, "step": 13238 }, { "epoch": 2.13, "grad_norm": 9.46182918548584, "learning_rate": 4.085930230602186e-06, "loss": 0.6054, "step": 13239 }, { "epoch": 2.13, "grad_norm": 12.285294532775879, "learning_rate": 4.08452885045733e-06, "loss": 0.8118, "step": 13240 }, { "epoch": 2.13, "grad_norm": 11.651135444641113, "learning_rate": 4.083127648995964e-06, "loss": 0.563, "step": 13241 }, { "epoch": 2.13, "grad_norm": 10.7769775390625, "learning_rate": 4.081726626260409e-06, "loss": 0.6491, "step": 13242 }, { "epoch": 2.13, "grad_norm": 10.090285301208496, "learning_rate": 4.080325782292983e-06, "loss": 0.8863, "step": 13243 }, { "epoch": 2.13, "grad_norm": 11.348220825195312, "learning_rate": 4.078925117136007e-06, "loss": 0.7358, "step": 13244 }, { "epoch": 2.13, "grad_norm": 9.986002922058105, "learning_rate": 4.077524630831784e-06, "loss": 0.7834, "step": 13245 }, { "epoch": 2.13, "grad_norm": 14.07390022277832, "learning_rate": 4.076124323422615e-06, "loss": 0.6909, "step": 13246 }, { "epoch": 2.13, "grad_norm": 14.307321548461914, "learning_rate": 4.074724194950807e-06, "loss": 0.6232, "step": 13247 }, { "epoch": 2.13, "grad_norm": 13.49746322631836, "learning_rate": 4.073324245458641e-06, "loss": 0.7573, "step": 13248 }, { "epoch": 2.13, "grad_norm": 14.050689697265625, "learning_rate": 4.071924474988413e-06, "loss": 0.8848, "step": 13249 }, { "epoch": 2.13, "grad_norm": 11.78173542022705, "learning_rate": 4.070524883582404e-06, "loss": 0.6102, "step": 13250 }, { "epoch": 2.13, "grad_norm": 11.635157585144043, "learning_rate": 4.069125471282882e-06, "loss": 0.6906, "step": 13251 }, { "epoch": 2.13, "grad_norm": 12.744173049926758, "learning_rate": 4.067726238132129e-06, "loss": 0.7513, "step": 13252 }, { "epoch": 2.13, "grad_norm": 11.766043663024902, "learning_rate": 4.066327184172404e-06, "loss": 0.6093, "step": 13253 }, { "epoch": 2.13, "grad_norm": 14.077844619750977, "learning_rate": 4.0649283094459645e-06, "loss": 0.7193, "step": 13254 }, { "epoch": 2.13, "grad_norm": 11.151285171508789, "learning_rate": 4.063529613995073e-06, "loss": 0.6148, "step": 13255 }, { "epoch": 2.13, "grad_norm": 13.165556907653809, "learning_rate": 4.06213109786197e-06, "loss": 0.4899, "step": 13256 }, { "epoch": 2.13, "grad_norm": 11.827590942382812, "learning_rate": 4.060732761088905e-06, "loss": 0.5434, "step": 13257 }, { "epoch": 2.13, "grad_norm": 12.252252578735352, "learning_rate": 4.059334603718117e-06, "loss": 0.6694, "step": 13258 }, { "epoch": 2.13, "grad_norm": 10.945722579956055, "learning_rate": 4.0579366257918326e-06, "loss": 0.6653, "step": 13259 }, { "epoch": 2.13, "grad_norm": 12.433176040649414, "learning_rate": 4.056538827352289e-06, "loss": 0.566, "step": 13260 }, { "epoch": 2.13, "grad_norm": 13.133793830871582, "learning_rate": 4.055141208441702e-06, "loss": 0.9021, "step": 13261 }, { "epoch": 2.13, "grad_norm": 13.236906051635742, "learning_rate": 4.0537437691022865e-06, "loss": 0.6603, "step": 13262 }, { "epoch": 2.14, "grad_norm": 12.752548217773438, "learning_rate": 4.052346509376258e-06, "loss": 0.5123, "step": 13263 }, { "epoch": 2.14, "grad_norm": 13.92476749420166, "learning_rate": 4.050949429305821e-06, "loss": 0.5761, "step": 13264 }, { "epoch": 2.14, "grad_norm": 12.686821937561035, "learning_rate": 4.0495525289331736e-06, "loss": 0.6279, "step": 13265 }, { "epoch": 2.14, "grad_norm": 12.508296966552734, "learning_rate": 4.048155808300516e-06, "loss": 0.752, "step": 13266 }, { "epoch": 2.14, "grad_norm": 11.711949348449707, "learning_rate": 4.04675926745003e-06, "loss": 0.659, "step": 13267 }, { "epoch": 2.14, "grad_norm": 14.080224990844727, "learning_rate": 4.045362906423907e-06, "loss": 0.8235, "step": 13268 }, { "epoch": 2.14, "grad_norm": 11.88773250579834, "learning_rate": 4.0439667252643235e-06, "loss": 0.6919, "step": 13269 }, { "epoch": 2.14, "grad_norm": 12.176828384399414, "learning_rate": 4.042570724013449e-06, "loss": 0.6028, "step": 13270 }, { "epoch": 2.14, "grad_norm": 12.273731231689453, "learning_rate": 4.041174902713457e-06, "loss": 0.6836, "step": 13271 }, { "epoch": 2.14, "grad_norm": 11.418708801269531, "learning_rate": 4.0397792614065046e-06, "loss": 0.6756, "step": 13272 }, { "epoch": 2.14, "grad_norm": 14.566581726074219, "learning_rate": 4.038383800134754e-06, "loss": 0.5928, "step": 13273 }, { "epoch": 2.14, "grad_norm": 11.951894760131836, "learning_rate": 4.036988518940354e-06, "loss": 0.7249, "step": 13274 }, { "epoch": 2.14, "grad_norm": 15.050311088562012, "learning_rate": 4.035593417865447e-06, "loss": 0.8206, "step": 13275 }, { "epoch": 2.14, "grad_norm": 13.52076244354248, "learning_rate": 4.034198496952181e-06, "loss": 0.7132, "step": 13276 }, { "epoch": 2.14, "grad_norm": 13.66706371307373, "learning_rate": 4.032803756242684e-06, "loss": 0.7248, "step": 13277 }, { "epoch": 2.14, "grad_norm": 11.632607460021973, "learning_rate": 4.031409195779088e-06, "loss": 0.91, "step": 13278 }, { "epoch": 2.14, "grad_norm": 13.070016860961914, "learning_rate": 4.0300148156035224e-06, "loss": 0.9235, "step": 13279 }, { "epoch": 2.14, "grad_norm": 11.268048286437988, "learning_rate": 4.028620615758103e-06, "loss": 0.7633, "step": 13280 }, { "epoch": 2.14, "grad_norm": 10.79310131072998, "learning_rate": 4.027226596284937e-06, "loss": 0.5269, "step": 13281 }, { "epoch": 2.14, "grad_norm": 10.92171859741211, "learning_rate": 4.025832757226142e-06, "loss": 0.8595, "step": 13282 }, { "epoch": 2.14, "grad_norm": 10.664582252502441, "learning_rate": 4.024439098623813e-06, "loss": 0.7284, "step": 13283 }, { "epoch": 2.14, "grad_norm": 13.29946231842041, "learning_rate": 4.023045620520053e-06, "loss": 0.8467, "step": 13284 }, { "epoch": 2.14, "grad_norm": 11.52742862701416, "learning_rate": 4.02165232295695e-06, "loss": 0.6134, "step": 13285 }, { "epoch": 2.14, "grad_norm": 11.233243942260742, "learning_rate": 4.020259205976588e-06, "loss": 0.5565, "step": 13286 }, { "epoch": 2.14, "grad_norm": 12.900266647338867, "learning_rate": 4.018866269621053e-06, "loss": 0.8605, "step": 13287 }, { "epoch": 2.14, "grad_norm": 10.89041805267334, "learning_rate": 4.017473513932418e-06, "loss": 0.68, "step": 13288 }, { "epoch": 2.14, "grad_norm": 12.989538192749023, "learning_rate": 4.01608093895275e-06, "loss": 0.8433, "step": 13289 }, { "epoch": 2.14, "grad_norm": 13.280867576599121, "learning_rate": 4.014688544724118e-06, "loss": 0.6974, "step": 13290 }, { "epoch": 2.14, "grad_norm": 11.449285507202148, "learning_rate": 4.013296331288575e-06, "loss": 0.6976, "step": 13291 }, { "epoch": 2.14, "grad_norm": 10.8220853805542, "learning_rate": 4.011904298688182e-06, "loss": 0.6776, "step": 13292 }, { "epoch": 2.14, "grad_norm": 9.994172096252441, "learning_rate": 4.010512446964981e-06, "loss": 0.6495, "step": 13293 }, { "epoch": 2.14, "grad_norm": 13.745748519897461, "learning_rate": 4.009120776161015e-06, "loss": 1.0451, "step": 13294 }, { "epoch": 2.14, "grad_norm": 11.22635269165039, "learning_rate": 4.007729286318325e-06, "loss": 0.6855, "step": 13295 }, { "epoch": 2.14, "grad_norm": 11.286404609680176, "learning_rate": 4.006337977478935e-06, "loss": 0.67, "step": 13296 }, { "epoch": 2.14, "grad_norm": 12.975160598754883, "learning_rate": 4.004946849684877e-06, "loss": 0.8027, "step": 13297 }, { "epoch": 2.14, "grad_norm": 11.513765335083008, "learning_rate": 4.003555902978174e-06, "loss": 0.7213, "step": 13298 }, { "epoch": 2.14, "grad_norm": 11.223782539367676, "learning_rate": 4.002165137400838e-06, "loss": 0.5998, "step": 13299 }, { "epoch": 2.14, "grad_norm": 11.382551193237305, "learning_rate": 4.000774552994874e-06, "loss": 0.6418, "step": 13300 }, { "epoch": 2.14, "grad_norm": 10.648417472839355, "learning_rate": 3.999384149802294e-06, "loss": 0.679, "step": 13301 }, { "epoch": 2.14, "grad_norm": 12.406426429748535, "learning_rate": 3.997993927865089e-06, "loss": 0.8531, "step": 13302 }, { "epoch": 2.14, "grad_norm": 10.28695297241211, "learning_rate": 3.99660388722526e-06, "loss": 0.5297, "step": 13303 }, { "epoch": 2.14, "grad_norm": 11.197439193725586, "learning_rate": 3.995214027924791e-06, "loss": 0.6664, "step": 13304 }, { "epoch": 2.14, "grad_norm": 11.859546661376953, "learning_rate": 3.993824350005661e-06, "loss": 0.6166, "step": 13305 }, { "epoch": 2.14, "grad_norm": 11.780414581298828, "learning_rate": 3.992434853509853e-06, "loss": 0.6137, "step": 13306 }, { "epoch": 2.14, "grad_norm": 13.20660400390625, "learning_rate": 3.991045538479334e-06, "loss": 0.5964, "step": 13307 }, { "epoch": 2.14, "grad_norm": 13.280477523803711, "learning_rate": 3.9896564049560714e-06, "loss": 0.8078, "step": 13308 }, { "epoch": 2.14, "grad_norm": 13.968748092651367, "learning_rate": 3.988267452982028e-06, "loss": 0.7901, "step": 13309 }, { "epoch": 2.14, "grad_norm": 14.320993423461914, "learning_rate": 3.9868786825991516e-06, "loss": 0.64, "step": 13310 }, { "epoch": 2.14, "grad_norm": 11.962262153625488, "learning_rate": 3.985490093849399e-06, "loss": 0.7527, "step": 13311 }, { "epoch": 2.14, "grad_norm": 11.765613555908203, "learning_rate": 3.984101686774712e-06, "loss": 0.7853, "step": 13312 }, { "epoch": 2.14, "grad_norm": 11.06541633605957, "learning_rate": 3.9827134614170236e-06, "loss": 0.5546, "step": 13313 }, { "epoch": 2.14, "grad_norm": 10.93126392364502, "learning_rate": 3.981325417818275e-06, "loss": 0.6529, "step": 13314 }, { "epoch": 2.14, "grad_norm": 11.26536750793457, "learning_rate": 3.9799375560203865e-06, "loss": 0.7389, "step": 13315 }, { "epoch": 2.14, "grad_norm": 12.24007797241211, "learning_rate": 3.978549876065288e-06, "loss": 0.5928, "step": 13316 }, { "epoch": 2.14, "grad_norm": 11.566500663757324, "learning_rate": 3.977162377994885e-06, "loss": 0.5607, "step": 13317 }, { "epoch": 2.14, "grad_norm": 9.074713706970215, "learning_rate": 3.975775061851097e-06, "loss": 0.6045, "step": 13318 }, { "epoch": 2.14, "grad_norm": 11.557842254638672, "learning_rate": 3.97438792767583e-06, "loss": 0.6071, "step": 13319 }, { "epoch": 2.14, "grad_norm": 10.562651634216309, "learning_rate": 3.9730009755109825e-06, "loss": 0.6747, "step": 13320 }, { "epoch": 2.14, "grad_norm": 13.532082557678223, "learning_rate": 3.971614205398442e-06, "loss": 0.6383, "step": 13321 }, { "epoch": 2.14, "grad_norm": 10.79951286315918, "learning_rate": 3.970227617380108e-06, "loss": 0.7049, "step": 13322 }, { "epoch": 2.14, "grad_norm": 13.053462028503418, "learning_rate": 3.968841211497858e-06, "loss": 0.7393, "step": 13323 }, { "epoch": 2.14, "grad_norm": 14.001468658447266, "learning_rate": 3.967454987793568e-06, "loss": 0.8, "step": 13324 }, { "epoch": 2.15, "grad_norm": 12.960244178771973, "learning_rate": 3.966068946309118e-06, "loss": 0.6668, "step": 13325 }, { "epoch": 2.15, "grad_norm": 12.955012321472168, "learning_rate": 3.964683087086365e-06, "loss": 0.592, "step": 13326 }, { "epoch": 2.15, "grad_norm": 11.24893856048584, "learning_rate": 3.963297410167179e-06, "loss": 0.816, "step": 13327 }, { "epoch": 2.15, "grad_norm": 11.80557918548584, "learning_rate": 3.961911915593415e-06, "loss": 0.6748, "step": 13328 }, { "epoch": 2.15, "grad_norm": 11.87230396270752, "learning_rate": 3.960526603406917e-06, "loss": 0.5219, "step": 13329 }, { "epoch": 2.15, "grad_norm": 10.80111312866211, "learning_rate": 3.959141473649537e-06, "loss": 0.8648, "step": 13330 }, { "epoch": 2.15, "grad_norm": 13.09548568725586, "learning_rate": 3.957756526363111e-06, "loss": 0.6708, "step": 13331 }, { "epoch": 2.15, "grad_norm": 14.916831970214844, "learning_rate": 3.956371761589472e-06, "loss": 0.7037, "step": 13332 }, { "epoch": 2.15, "grad_norm": 18.147071838378906, "learning_rate": 3.954987179370451e-06, "loss": 0.8044, "step": 13333 }, { "epoch": 2.15, "grad_norm": 13.041703224182129, "learning_rate": 3.953602779747867e-06, "loss": 0.5698, "step": 13334 }, { "epoch": 2.15, "grad_norm": 12.58448314666748, "learning_rate": 3.952218562763544e-06, "loss": 0.6143, "step": 13335 }, { "epoch": 2.15, "grad_norm": 9.638748168945312, "learning_rate": 3.9508345284592895e-06, "loss": 0.4522, "step": 13336 }, { "epoch": 2.15, "grad_norm": 12.652425765991211, "learning_rate": 3.949450676876902e-06, "loss": 0.7156, "step": 13337 }, { "epoch": 2.15, "grad_norm": 13.273500442504883, "learning_rate": 3.9480670080582e-06, "loss": 0.6715, "step": 13338 }, { "epoch": 2.15, "grad_norm": 12.587462425231934, "learning_rate": 3.946683522044967e-06, "loss": 0.9437, "step": 13339 }, { "epoch": 2.15, "grad_norm": 13.819836616516113, "learning_rate": 3.945300218878992e-06, "loss": 0.7257, "step": 13340 }, { "epoch": 2.15, "grad_norm": 10.886545181274414, "learning_rate": 3.943917098602067e-06, "loss": 0.7918, "step": 13341 }, { "epoch": 2.15, "grad_norm": 13.48812484741211, "learning_rate": 3.942534161255965e-06, "loss": 0.4737, "step": 13342 }, { "epoch": 2.15, "grad_norm": 12.679656982421875, "learning_rate": 3.941151406882457e-06, "loss": 0.6081, "step": 13343 }, { "epoch": 2.15, "grad_norm": 12.535574913024902, "learning_rate": 3.939768835523318e-06, "loss": 0.8055, "step": 13344 }, { "epoch": 2.15, "grad_norm": 12.398079872131348, "learning_rate": 3.9383864472203005e-06, "loss": 0.6098, "step": 13345 }, { "epoch": 2.15, "grad_norm": 11.805970191955566, "learning_rate": 3.937004242015171e-06, "loss": 0.7472, "step": 13346 }, { "epoch": 2.15, "grad_norm": 11.543157577514648, "learning_rate": 3.935622219949677e-06, "loss": 0.8892, "step": 13347 }, { "epoch": 2.15, "grad_norm": 13.395393371582031, "learning_rate": 3.934240381065558e-06, "loss": 0.6314, "step": 13348 }, { "epoch": 2.15, "grad_norm": 13.09306812286377, "learning_rate": 3.9328587254045634e-06, "loss": 0.7244, "step": 13349 }, { "epoch": 2.15, "grad_norm": 13.002276420593262, "learning_rate": 3.93147725300842e-06, "loss": 0.7205, "step": 13350 }, { "epoch": 2.15, "grad_norm": 10.843378067016602, "learning_rate": 3.9300959639188645e-06, "loss": 0.8182, "step": 13351 }, { "epoch": 2.15, "grad_norm": 12.229039192199707, "learning_rate": 3.928714858177615e-06, "loss": 0.7435, "step": 13352 }, { "epoch": 2.15, "grad_norm": 12.39107608795166, "learning_rate": 3.927333935826386e-06, "loss": 0.7646, "step": 13353 }, { "epoch": 2.15, "grad_norm": 10.443879127502441, "learning_rate": 3.925953196906899e-06, "loss": 0.6359, "step": 13354 }, { "epoch": 2.15, "grad_norm": 9.773480415344238, "learning_rate": 3.924572641460855e-06, "loss": 0.5468, "step": 13355 }, { "epoch": 2.15, "grad_norm": 10.135126113891602, "learning_rate": 3.923192269529953e-06, "loss": 0.4383, "step": 13356 }, { "epoch": 2.15, "grad_norm": 11.530794143676758, "learning_rate": 3.921812081155896e-06, "loss": 0.7041, "step": 13357 }, { "epoch": 2.15, "grad_norm": 11.798891067504883, "learning_rate": 3.920432076380365e-06, "loss": 0.6859, "step": 13358 }, { "epoch": 2.15, "grad_norm": 13.232733726501465, "learning_rate": 3.9190522552450505e-06, "loss": 0.6028, "step": 13359 }, { "epoch": 2.15, "grad_norm": 10.696468353271484, "learning_rate": 3.917672617791634e-06, "loss": 0.6242, "step": 13360 }, { "epoch": 2.15, "grad_norm": 13.735764503479004, "learning_rate": 3.916293164061783e-06, "loss": 0.7967, "step": 13361 }, { "epoch": 2.15, "grad_norm": 13.23879337310791, "learning_rate": 3.914913894097171e-06, "loss": 0.8434, "step": 13362 }, { "epoch": 2.15, "grad_norm": 10.519664764404297, "learning_rate": 3.9135348079394566e-06, "loss": 0.675, "step": 13363 }, { "epoch": 2.15, "grad_norm": 11.92054557800293, "learning_rate": 3.912155905630295e-06, "loss": 0.848, "step": 13364 }, { "epoch": 2.15, "grad_norm": 12.133615493774414, "learning_rate": 3.9107771872113434e-06, "loss": 0.6631, "step": 13365 }, { "epoch": 2.15, "grad_norm": 12.614675521850586, "learning_rate": 3.9093986527242445e-06, "loss": 0.5299, "step": 13366 }, { "epoch": 2.15, "grad_norm": 11.021882057189941, "learning_rate": 3.908020302210634e-06, "loss": 0.9243, "step": 13367 }, { "epoch": 2.15, "grad_norm": 12.080282211303711, "learning_rate": 3.906642135712154e-06, "loss": 0.7643, "step": 13368 }, { "epoch": 2.15, "grad_norm": 14.690120697021484, "learning_rate": 3.905264153270428e-06, "loss": 0.6524, "step": 13369 }, { "epoch": 2.15, "grad_norm": 12.842116355895996, "learning_rate": 3.903886354927085e-06, "loss": 0.8242, "step": 13370 }, { "epoch": 2.15, "grad_norm": 10.6424560546875, "learning_rate": 3.902508740723739e-06, "loss": 0.728, "step": 13371 }, { "epoch": 2.15, "grad_norm": 12.443855285644531, "learning_rate": 3.901131310702001e-06, "loss": 0.715, "step": 13372 }, { "epoch": 2.15, "grad_norm": 11.899459838867188, "learning_rate": 3.899754064903482e-06, "loss": 0.5461, "step": 13373 }, { "epoch": 2.15, "grad_norm": 12.324376106262207, "learning_rate": 3.8983770033697785e-06, "loss": 0.7798, "step": 13374 }, { "epoch": 2.15, "grad_norm": 10.715084075927734, "learning_rate": 3.897000126142493e-06, "loss": 0.571, "step": 13375 }, { "epoch": 2.15, "grad_norm": 12.412308692932129, "learning_rate": 3.8956234332632115e-06, "loss": 0.9522, "step": 13376 }, { "epoch": 2.15, "grad_norm": 12.558634757995605, "learning_rate": 3.894246924773513e-06, "loss": 0.5567, "step": 13377 }, { "epoch": 2.15, "grad_norm": 10.902666091918945, "learning_rate": 3.892870600714985e-06, "loss": 0.6404, "step": 13378 }, { "epoch": 2.15, "grad_norm": 11.200098991394043, "learning_rate": 3.8914944611292005e-06, "loss": 0.6603, "step": 13379 }, { "epoch": 2.15, "grad_norm": 13.66608715057373, "learning_rate": 3.890118506057722e-06, "loss": 0.7783, "step": 13380 }, { "epoch": 2.15, "grad_norm": 11.773191452026367, "learning_rate": 3.888742735542117e-06, "loss": 0.678, "step": 13381 }, { "epoch": 2.15, "grad_norm": 12.428950309753418, "learning_rate": 3.887367149623941e-06, "loss": 0.9699, "step": 13382 }, { "epoch": 2.15, "grad_norm": 10.795469284057617, "learning_rate": 3.88599174834474e-06, "loss": 0.8857, "step": 13383 }, { "epoch": 2.15, "grad_norm": 11.118612289428711, "learning_rate": 3.8846165317460674e-06, "loss": 0.7987, "step": 13384 }, { "epoch": 2.15, "grad_norm": 11.14153003692627, "learning_rate": 3.883241499869456e-06, "loss": 0.5473, "step": 13385 }, { "epoch": 2.15, "grad_norm": 8.671971321105957, "learning_rate": 3.881866652756446e-06, "loss": 0.501, "step": 13386 }, { "epoch": 2.16, "grad_norm": 13.66712760925293, "learning_rate": 3.880491990448565e-06, "loss": 0.7733, "step": 13387 }, { "epoch": 2.16, "grad_norm": 11.527284622192383, "learning_rate": 3.87911751298733e-06, "loss": 0.6122, "step": 13388 }, { "epoch": 2.16, "grad_norm": 12.367544174194336, "learning_rate": 3.877743220414268e-06, "loss": 0.8691, "step": 13389 }, { "epoch": 2.16, "grad_norm": 12.400146484375, "learning_rate": 3.876369112770887e-06, "loss": 0.7085, "step": 13390 }, { "epoch": 2.16, "grad_norm": 12.987017631530762, "learning_rate": 3.874995190098689e-06, "loss": 0.6792, "step": 13391 }, { "epoch": 2.16, "grad_norm": 12.883241653442383, "learning_rate": 3.873621452439183e-06, "loss": 0.7054, "step": 13392 }, { "epoch": 2.16, "grad_norm": 11.66698932647705, "learning_rate": 3.8722478998338575e-06, "loss": 0.691, "step": 13393 }, { "epoch": 2.16, "grad_norm": 15.568439483642578, "learning_rate": 3.870874532324208e-06, "loss": 0.7913, "step": 13394 }, { "epoch": 2.16, "grad_norm": 14.301761627197266, "learning_rate": 3.869501349951716e-06, "loss": 0.6998, "step": 13395 }, { "epoch": 2.16, "grad_norm": 15.760286331176758, "learning_rate": 3.868128352757855e-06, "loss": 0.6365, "step": 13396 }, { "epoch": 2.16, "grad_norm": 11.961130142211914, "learning_rate": 3.866755540784107e-06, "loss": 0.5745, "step": 13397 }, { "epoch": 2.16, "grad_norm": 11.858901023864746, "learning_rate": 3.865382914071931e-06, "loss": 0.4647, "step": 13398 }, { "epoch": 2.16, "grad_norm": 13.360028266906738, "learning_rate": 3.864010472662794e-06, "loss": 0.5645, "step": 13399 }, { "epoch": 2.16, "grad_norm": 11.42043685913086, "learning_rate": 3.862638216598153e-06, "loss": 0.5859, "step": 13400 }, { "epoch": 2.16, "grad_norm": 14.376325607299805, "learning_rate": 3.861266145919457e-06, "loss": 0.6932, "step": 13401 }, { "epoch": 2.16, "grad_norm": 10.897605895996094, "learning_rate": 3.859894260668146e-06, "loss": 0.8217, "step": 13402 }, { "epoch": 2.16, "grad_norm": 11.207610130310059, "learning_rate": 3.858522560885668e-06, "loss": 0.6403, "step": 13403 }, { "epoch": 2.16, "grad_norm": 11.198938369750977, "learning_rate": 3.857151046613449e-06, "loss": 0.5646, "step": 13404 }, { "epoch": 2.16, "grad_norm": 13.200602531433105, "learning_rate": 3.855779717892925e-06, "loss": 1.0179, "step": 13405 }, { "epoch": 2.16, "grad_norm": 15.687408447265625, "learning_rate": 3.854408574765513e-06, "loss": 0.8667, "step": 13406 }, { "epoch": 2.16, "grad_norm": 9.730778694152832, "learning_rate": 3.853037617272627e-06, "loss": 0.4206, "step": 13407 }, { "epoch": 2.16, "grad_norm": 10.048262596130371, "learning_rate": 3.851666845455686e-06, "loss": 0.5645, "step": 13408 }, { "epoch": 2.16, "grad_norm": 13.168700218200684, "learning_rate": 3.850296259356092e-06, "loss": 0.7148, "step": 13409 }, { "epoch": 2.16, "grad_norm": 12.241252899169922, "learning_rate": 3.848925859015242e-06, "loss": 0.6253, "step": 13410 }, { "epoch": 2.16, "grad_norm": 13.779437065124512, "learning_rate": 3.8475556444745365e-06, "loss": 0.5541, "step": 13411 }, { "epoch": 2.16, "grad_norm": 13.816573143005371, "learning_rate": 3.846185615775359e-06, "loss": 0.5615, "step": 13412 }, { "epoch": 2.16, "grad_norm": 10.430996894836426, "learning_rate": 3.844815772959099e-06, "loss": 0.7237, "step": 13413 }, { "epoch": 2.16, "grad_norm": 12.712170600891113, "learning_rate": 3.8434461160671285e-06, "loss": 0.686, "step": 13414 }, { "epoch": 2.16, "grad_norm": 10.008559226989746, "learning_rate": 3.84207664514082e-06, "loss": 0.6421, "step": 13415 }, { "epoch": 2.16, "grad_norm": 12.694354057312012, "learning_rate": 3.840707360221544e-06, "loss": 0.5287, "step": 13416 }, { "epoch": 2.16, "grad_norm": 10.58513069152832, "learning_rate": 3.839338261350656e-06, "loss": 0.6713, "step": 13417 }, { "epoch": 2.16, "grad_norm": 11.164470672607422, "learning_rate": 3.8379693485695125e-06, "loss": 0.7509, "step": 13418 }, { "epoch": 2.16, "grad_norm": 11.803023338317871, "learning_rate": 3.836600621919468e-06, "loss": 0.79, "step": 13419 }, { "epoch": 2.16, "grad_norm": 10.872671127319336, "learning_rate": 3.835232081441864e-06, "loss": 0.6085, "step": 13420 }, { "epoch": 2.16, "grad_norm": 11.23892879486084, "learning_rate": 3.8338637271780345e-06, "loss": 0.769, "step": 13421 }, { "epoch": 2.16, "grad_norm": 9.171369552612305, "learning_rate": 3.832495559169318e-06, "loss": 0.527, "step": 13422 }, { "epoch": 2.16, "grad_norm": 23.644168853759766, "learning_rate": 3.8311275774570344e-06, "loss": 0.5375, "step": 13423 }, { "epoch": 2.16, "grad_norm": 15.186527252197266, "learning_rate": 3.829759782082515e-06, "loss": 0.6857, "step": 13424 }, { "epoch": 2.16, "grad_norm": 11.763575553894043, "learning_rate": 3.828392173087069e-06, "loss": 0.7026, "step": 13425 }, { "epoch": 2.16, "grad_norm": 14.352941513061523, "learning_rate": 3.827024750512005e-06, "loss": 0.7086, "step": 13426 }, { "epoch": 2.16, "grad_norm": 9.636502265930176, "learning_rate": 3.825657514398633e-06, "loss": 0.5662, "step": 13427 }, { "epoch": 2.16, "grad_norm": 10.58492374420166, "learning_rate": 3.824290464788246e-06, "loss": 0.7504, "step": 13428 }, { "epoch": 2.16, "grad_norm": 10.018174171447754, "learning_rate": 3.8229236017221445e-06, "loss": 0.7848, "step": 13429 }, { "epoch": 2.16, "grad_norm": 13.787984848022461, "learning_rate": 3.821556925241612e-06, "loss": 0.885, "step": 13430 }, { "epoch": 2.16, "grad_norm": 9.512341499328613, "learning_rate": 3.820190435387927e-06, "loss": 0.6836, "step": 13431 }, { "epoch": 2.16, "grad_norm": 14.812521934509277, "learning_rate": 3.818824132202374e-06, "loss": 0.7442, "step": 13432 }, { "epoch": 2.16, "grad_norm": 11.723814964294434, "learning_rate": 3.817458015726218e-06, "loss": 0.8042, "step": 13433 }, { "epoch": 2.16, "grad_norm": 13.233024597167969, "learning_rate": 3.8160920860007234e-06, "loss": 0.6385, "step": 13434 }, { "epoch": 2.16, "grad_norm": 14.936574935913086, "learning_rate": 3.814726343067153e-06, "loss": 0.7887, "step": 13435 }, { "epoch": 2.16, "grad_norm": 11.593880653381348, "learning_rate": 3.8133607869667577e-06, "loss": 0.6907, "step": 13436 }, { "epoch": 2.16, "grad_norm": 13.862101554870605, "learning_rate": 3.8119954177407903e-06, "loss": 0.6296, "step": 13437 }, { "epoch": 2.16, "grad_norm": 11.825685501098633, "learning_rate": 3.8106302354304893e-06, "loss": 0.6611, "step": 13438 }, { "epoch": 2.16, "grad_norm": 11.463763236999512, "learning_rate": 3.8092652400770857e-06, "loss": 0.5428, "step": 13439 }, { "epoch": 2.16, "grad_norm": 14.039365768432617, "learning_rate": 3.807900431721826e-06, "loss": 0.7642, "step": 13440 }, { "epoch": 2.16, "grad_norm": 13.290915489196777, "learning_rate": 3.8065358104059257e-06, "loss": 0.7796, "step": 13441 }, { "epoch": 2.16, "grad_norm": 11.967253684997559, "learning_rate": 3.8051713761706034e-06, "loss": 0.7945, "step": 13442 }, { "epoch": 2.16, "grad_norm": 11.10606575012207, "learning_rate": 3.80380712905708e-06, "loss": 0.7872, "step": 13443 }, { "epoch": 2.16, "grad_norm": 12.25906753540039, "learning_rate": 3.8024430691065594e-06, "loss": 0.7, "step": 13444 }, { "epoch": 2.16, "grad_norm": 10.697172164916992, "learning_rate": 3.8010791963602434e-06, "loss": 0.4324, "step": 13445 }, { "epoch": 2.16, "grad_norm": 13.123783111572266, "learning_rate": 3.7997155108593344e-06, "loss": 0.8588, "step": 13446 }, { "epoch": 2.16, "grad_norm": 12.445487976074219, "learning_rate": 3.798352012645018e-06, "loss": 0.7947, "step": 13447 }, { "epoch": 2.16, "grad_norm": 16.193706512451172, "learning_rate": 3.7969887017584873e-06, "loss": 0.6434, "step": 13448 }, { "epoch": 2.17, "grad_norm": 12.318675994873047, "learning_rate": 3.7956255782409177e-06, "loss": 0.854, "step": 13449 }, { "epoch": 2.17, "grad_norm": 10.63500690460205, "learning_rate": 3.7942626421334814e-06, "loss": 0.5886, "step": 13450 }, { "epoch": 2.17, "grad_norm": 12.074268341064453, "learning_rate": 3.7928998934773552e-06, "loss": 0.7934, "step": 13451 }, { "epoch": 2.17, "grad_norm": 13.754836082458496, "learning_rate": 3.791537332313695e-06, "loss": 0.746, "step": 13452 }, { "epoch": 2.17, "grad_norm": 11.564836502075195, "learning_rate": 3.7901749586836646e-06, "loss": 0.8961, "step": 13453 }, { "epoch": 2.17, "grad_norm": 10.646203994750977, "learning_rate": 3.788812772628413e-06, "loss": 0.6347, "step": 13454 }, { "epoch": 2.17, "grad_norm": 11.40857982635498, "learning_rate": 3.7874507741890833e-06, "loss": 0.7155, "step": 13455 }, { "epoch": 2.17, "grad_norm": 13.120046615600586, "learning_rate": 3.786088963406823e-06, "loss": 0.489, "step": 13456 }, { "epoch": 2.17, "grad_norm": 10.743443489074707, "learning_rate": 3.7847273403227636e-06, "loss": 0.4337, "step": 13457 }, { "epoch": 2.17, "grad_norm": 10.984326362609863, "learning_rate": 3.7833659049780313e-06, "loss": 0.6443, "step": 13458 }, { "epoch": 2.17, "grad_norm": 12.275099754333496, "learning_rate": 3.782004657413754e-06, "loss": 0.7616, "step": 13459 }, { "epoch": 2.17, "grad_norm": 12.852831840515137, "learning_rate": 3.7806435976710508e-06, "loss": 0.7066, "step": 13460 }, { "epoch": 2.17, "grad_norm": 10.908408164978027, "learning_rate": 3.7792827257910304e-06, "loss": 0.7863, "step": 13461 }, { "epoch": 2.17, "grad_norm": 11.244071006774902, "learning_rate": 3.7779220418148034e-06, "loss": 0.7858, "step": 13462 }, { "epoch": 2.17, "grad_norm": 11.812234878540039, "learning_rate": 3.776561545783465e-06, "loss": 0.6668, "step": 13463 }, { "epoch": 2.17, "grad_norm": 10.931988716125488, "learning_rate": 3.7752012377381177e-06, "loss": 0.5739, "step": 13464 }, { "epoch": 2.17, "grad_norm": 10.914281845092773, "learning_rate": 3.7738411177198477e-06, "loss": 0.801, "step": 13465 }, { "epoch": 2.17, "grad_norm": 15.357460975646973, "learning_rate": 3.772481185769734e-06, "loss": 0.6363, "step": 13466 }, { "epoch": 2.17, "grad_norm": 12.615407943725586, "learning_rate": 3.7711214419288643e-06, "loss": 0.9492, "step": 13467 }, { "epoch": 2.17, "grad_norm": 12.112326622009277, "learning_rate": 3.7697618862383067e-06, "loss": 0.8402, "step": 13468 }, { "epoch": 2.17, "grad_norm": 11.582280158996582, "learning_rate": 3.768402518739125e-06, "loss": 0.7713, "step": 13469 }, { "epoch": 2.17, "grad_norm": 13.816643714904785, "learning_rate": 3.767043339472386e-06, "loss": 0.7176, "step": 13470 }, { "epoch": 2.17, "grad_norm": 14.430333137512207, "learning_rate": 3.7656843484791384e-06, "loss": 0.8906, "step": 13471 }, { "epoch": 2.17, "grad_norm": 12.731280326843262, "learning_rate": 3.7643255458004414e-06, "loss": 0.6272, "step": 13472 }, { "epoch": 2.17, "grad_norm": 9.445330619812012, "learning_rate": 3.7629669314773334e-06, "loss": 0.5502, "step": 13473 }, { "epoch": 2.17, "grad_norm": 9.665116310119629, "learning_rate": 3.7616085055508488e-06, "loss": 0.6719, "step": 13474 }, { "epoch": 2.17, "grad_norm": 11.70251178741455, "learning_rate": 3.76025026806203e-06, "loss": 0.7079, "step": 13475 }, { "epoch": 2.17, "grad_norm": 14.317314147949219, "learning_rate": 3.7588922190518985e-06, "loss": 0.7762, "step": 13476 }, { "epoch": 2.17, "grad_norm": 12.305415153503418, "learning_rate": 3.757534358561473e-06, "loss": 0.9053, "step": 13477 }, { "epoch": 2.17, "grad_norm": 11.528664588928223, "learning_rate": 3.7561766866317774e-06, "loss": 0.7636, "step": 13478 }, { "epoch": 2.17, "grad_norm": 14.947358131408691, "learning_rate": 3.7548192033038124e-06, "loss": 0.9352, "step": 13479 }, { "epoch": 2.17, "grad_norm": 12.569247245788574, "learning_rate": 3.7534619086185865e-06, "loss": 0.8371, "step": 13480 }, { "epoch": 2.17, "grad_norm": 12.681726455688477, "learning_rate": 3.7521048026171023e-06, "loss": 0.6897, "step": 13481 }, { "epoch": 2.17, "grad_norm": 13.090911865234375, "learning_rate": 3.7507478853403466e-06, "loss": 0.8426, "step": 13482 }, { "epoch": 2.17, "grad_norm": 14.337562561035156, "learning_rate": 3.749391156829312e-06, "loss": 0.6421, "step": 13483 }, { "epoch": 2.17, "grad_norm": 9.964288711547852, "learning_rate": 3.7480346171249772e-06, "loss": 0.6751, "step": 13484 }, { "epoch": 2.17, "grad_norm": 12.738601684570312, "learning_rate": 3.7466782662683144e-06, "loss": 0.6455, "step": 13485 }, { "epoch": 2.17, "grad_norm": 9.775585174560547, "learning_rate": 3.745322104300301e-06, "loss": 0.7111, "step": 13486 }, { "epoch": 2.17, "grad_norm": 11.407108306884766, "learning_rate": 3.743966131261897e-06, "loss": 0.656, "step": 13487 }, { "epoch": 2.17, "grad_norm": 12.874272346496582, "learning_rate": 3.742610347194059e-06, "loss": 0.7121, "step": 13488 }, { "epoch": 2.17, "grad_norm": 13.573436737060547, "learning_rate": 3.741254752137745e-06, "loss": 0.7847, "step": 13489 }, { "epoch": 2.17, "grad_norm": 10.780649185180664, "learning_rate": 3.739899346133897e-06, "loss": 0.727, "step": 13490 }, { "epoch": 2.17, "grad_norm": 11.755598068237305, "learning_rate": 3.738544129223465e-06, "loss": 0.8152, "step": 13491 }, { "epoch": 2.17, "grad_norm": 11.001387596130371, "learning_rate": 3.737189101447377e-06, "loss": 0.5871, "step": 13492 }, { "epoch": 2.17, "grad_norm": 13.358091354370117, "learning_rate": 3.735834262846564e-06, "loss": 0.6554, "step": 13493 }, { "epoch": 2.17, "grad_norm": 12.06052017211914, "learning_rate": 3.7344796134619545e-06, "loss": 0.6838, "step": 13494 }, { "epoch": 2.17, "grad_norm": 11.661130905151367, "learning_rate": 3.7331251533344602e-06, "loss": 0.7611, "step": 13495 }, { "epoch": 2.17, "grad_norm": 13.113045692443848, "learning_rate": 3.7317708825050036e-06, "loss": 0.7696, "step": 13496 }, { "epoch": 2.17, "grad_norm": 10.757185935974121, "learning_rate": 3.730416801014487e-06, "loss": 0.5742, "step": 13497 }, { "epoch": 2.17, "grad_norm": 11.921380996704102, "learning_rate": 3.7290629089038076e-06, "loss": 0.6933, "step": 13498 }, { "epoch": 2.17, "grad_norm": 13.140458106994629, "learning_rate": 3.7277092062138666e-06, "loss": 0.6453, "step": 13499 }, { "epoch": 2.17, "grad_norm": 11.898303985595703, "learning_rate": 3.726355692985556e-06, "loss": 0.7152, "step": 13500 }, { "epoch": 2.17, "grad_norm": 11.084818840026855, "learning_rate": 3.725002369259755e-06, "loss": 0.7337, "step": 13501 }, { "epoch": 2.17, "grad_norm": 13.718039512634277, "learning_rate": 3.7236492350773466e-06, "loss": 0.702, "step": 13502 }, { "epoch": 2.17, "grad_norm": 12.90601921081543, "learning_rate": 3.7222962904792028e-06, "loss": 0.5577, "step": 13503 }, { "epoch": 2.17, "grad_norm": 11.497772216796875, "learning_rate": 3.720943535506185e-06, "loss": 0.6089, "step": 13504 }, { "epoch": 2.17, "grad_norm": 13.163113594055176, "learning_rate": 3.719590970199164e-06, "loss": 0.7254, "step": 13505 }, { "epoch": 2.17, "grad_norm": 11.835063934326172, "learning_rate": 3.7182385945989872e-06, "loss": 0.5421, "step": 13506 }, { "epoch": 2.17, "grad_norm": 14.528085708618164, "learning_rate": 3.7168864087465128e-06, "loss": 0.8146, "step": 13507 }, { "epoch": 2.17, "grad_norm": 10.158440589904785, "learning_rate": 3.71553441268258e-06, "loss": 0.6318, "step": 13508 }, { "epoch": 2.17, "grad_norm": 10.264688491821289, "learning_rate": 3.7141826064480247e-06, "loss": 0.6195, "step": 13509 }, { "epoch": 2.17, "grad_norm": 11.037772178649902, "learning_rate": 3.7128309900836867e-06, "loss": 0.8852, "step": 13510 }, { "epoch": 2.17, "grad_norm": 9.612547874450684, "learning_rate": 3.711479563630389e-06, "loss": 0.7227, "step": 13511 }, { "epoch": 2.18, "grad_norm": 14.566020011901855, "learning_rate": 3.7101283271289512e-06, "loss": 0.708, "step": 13512 }, { "epoch": 2.18, "grad_norm": 13.245173454284668, "learning_rate": 3.708777280620194e-06, "loss": 0.7601, "step": 13513 }, { "epoch": 2.18, "grad_norm": 12.117178916931152, "learning_rate": 3.707426424144921e-06, "loss": 0.6821, "step": 13514 }, { "epoch": 2.18, "grad_norm": 11.570508003234863, "learning_rate": 3.706075757743943e-06, "loss": 0.6993, "step": 13515 }, { "epoch": 2.18, "grad_norm": 11.98021411895752, "learning_rate": 3.704725281458057e-06, "loss": 0.6955, "step": 13516 }, { "epoch": 2.18, "grad_norm": 14.737309455871582, "learning_rate": 3.7033749953280483e-06, "loss": 0.7638, "step": 13517 }, { "epoch": 2.18, "grad_norm": 11.05715560913086, "learning_rate": 3.702024899394715e-06, "loss": 0.7836, "step": 13518 }, { "epoch": 2.18, "grad_norm": 12.765270233154297, "learning_rate": 3.7006749936988274e-06, "loss": 0.8365, "step": 13519 }, { "epoch": 2.18, "grad_norm": 10.029741287231445, "learning_rate": 3.6993252782811674e-06, "loss": 0.5962, "step": 13520 }, { "epoch": 2.18, "grad_norm": 15.256351470947266, "learning_rate": 3.697975753182507e-06, "loss": 0.7658, "step": 13521 }, { "epoch": 2.18, "grad_norm": 10.576062202453613, "learning_rate": 3.696626418443606e-06, "loss": 0.8698, "step": 13522 }, { "epoch": 2.18, "grad_norm": 11.487269401550293, "learning_rate": 3.69527727410522e-06, "loss": 0.6341, "step": 13523 }, { "epoch": 2.18, "grad_norm": 14.178804397583008, "learning_rate": 3.693928320208109e-06, "loss": 0.5982, "step": 13524 }, { "epoch": 2.18, "grad_norm": 9.948418617248535, "learning_rate": 3.6925795567930113e-06, "loss": 0.5779, "step": 13525 }, { "epoch": 2.18, "grad_norm": 12.460953712463379, "learning_rate": 3.6912309839006745e-06, "loss": 0.7525, "step": 13526 }, { "epoch": 2.18, "grad_norm": 11.846333503723145, "learning_rate": 3.6898826015718327e-06, "loss": 0.9261, "step": 13527 }, { "epoch": 2.18, "grad_norm": 13.207422256469727, "learning_rate": 3.688534409847209e-06, "loss": 0.7557, "step": 13528 }, { "epoch": 2.18, "grad_norm": 13.088552474975586, "learning_rate": 3.687186408767536e-06, "loss": 0.7294, "step": 13529 }, { "epoch": 2.18, "grad_norm": 14.981836318969727, "learning_rate": 3.6858385983735233e-06, "loss": 0.6446, "step": 13530 }, { "epoch": 2.18, "grad_norm": 9.695425987243652, "learning_rate": 3.684490978705891e-06, "loss": 0.6561, "step": 13531 }, { "epoch": 2.18, "grad_norm": 15.615863800048828, "learning_rate": 3.6831435498053414e-06, "loss": 0.6369, "step": 13532 }, { "epoch": 2.18, "grad_norm": 11.971968650817871, "learning_rate": 3.6817963117125723e-06, "loss": 0.6223, "step": 13533 }, { "epoch": 2.18, "grad_norm": 12.084741592407227, "learning_rate": 3.680449264468284e-06, "loss": 0.8605, "step": 13534 }, { "epoch": 2.18, "grad_norm": 9.624689102172852, "learning_rate": 3.6791024081131644e-06, "loss": 0.5918, "step": 13535 }, { "epoch": 2.18, "grad_norm": 11.548216819763184, "learning_rate": 3.677755742687892e-06, "loss": 0.5548, "step": 13536 }, { "epoch": 2.18, "grad_norm": 11.834342002868652, "learning_rate": 3.6764092682331507e-06, "loss": 0.7074, "step": 13537 }, { "epoch": 2.18, "grad_norm": 12.379305839538574, "learning_rate": 3.675062984789607e-06, "loss": 0.6249, "step": 13538 }, { "epoch": 2.18, "grad_norm": 13.299857139587402, "learning_rate": 3.6737168923979326e-06, "loss": 0.7157, "step": 13539 }, { "epoch": 2.18, "grad_norm": 12.380189895629883, "learning_rate": 3.6723709910987802e-06, "loss": 0.6825, "step": 13540 }, { "epoch": 2.18, "grad_norm": 12.858535766601562, "learning_rate": 3.6710252809328093e-06, "loss": 0.6038, "step": 13541 }, { "epoch": 2.18, "grad_norm": 12.077049255371094, "learning_rate": 3.669679761940672e-06, "loss": 0.6891, "step": 13542 }, { "epoch": 2.18, "grad_norm": 10.134716033935547, "learning_rate": 3.6683344341630067e-06, "loss": 0.4991, "step": 13543 }, { "epoch": 2.18, "grad_norm": 11.766980171203613, "learning_rate": 3.6669892976404465e-06, "loss": 0.5537, "step": 13544 }, { "epoch": 2.18, "grad_norm": 13.55622673034668, "learning_rate": 3.665644352413632e-06, "loss": 0.5428, "step": 13545 }, { "epoch": 2.18, "grad_norm": 13.638168334960938, "learning_rate": 3.6642995985231833e-06, "loss": 0.6936, "step": 13546 }, { "epoch": 2.18, "grad_norm": 12.296395301818848, "learning_rate": 3.662955036009718e-06, "loss": 0.7279, "step": 13547 }, { "epoch": 2.18, "grad_norm": 14.087075233459473, "learning_rate": 3.661610664913856e-06, "loss": 0.6477, "step": 13548 }, { "epoch": 2.18, "grad_norm": 10.763713836669922, "learning_rate": 3.6602664852762003e-06, "loss": 0.6146, "step": 13549 }, { "epoch": 2.18, "grad_norm": 12.820921897888184, "learning_rate": 3.6589224971373583e-06, "loss": 0.6839, "step": 13550 }, { "epoch": 2.18, "grad_norm": 13.293704986572266, "learning_rate": 3.6575787005379237e-06, "loss": 0.6789, "step": 13551 }, { "epoch": 2.18, "grad_norm": 14.434130668640137, "learning_rate": 3.6562350955184843e-06, "loss": 0.5932, "step": 13552 }, { "epoch": 2.18, "grad_norm": 12.007386207580566, "learning_rate": 3.6548916821196333e-06, "loss": 0.8367, "step": 13553 }, { "epoch": 2.18, "grad_norm": 11.451279640197754, "learning_rate": 3.653548460381945e-06, "loss": 0.803, "step": 13554 }, { "epoch": 2.18, "grad_norm": 12.131444931030273, "learning_rate": 3.6522054303459896e-06, "loss": 0.7121, "step": 13555 }, { "epoch": 2.18, "grad_norm": 11.306639671325684, "learning_rate": 3.6508625920523423e-06, "loss": 0.6591, "step": 13556 }, { "epoch": 2.18, "grad_norm": 11.375528335571289, "learning_rate": 3.6495199455415575e-06, "loss": 0.7726, "step": 13557 }, { "epoch": 2.18, "grad_norm": 13.610944747924805, "learning_rate": 3.648177490854198e-06, "loss": 0.5626, "step": 13558 }, { "epoch": 2.18, "grad_norm": 13.355375289916992, "learning_rate": 3.646835228030813e-06, "loss": 0.7028, "step": 13559 }, { "epoch": 2.18, "grad_norm": 11.527449607849121, "learning_rate": 3.645493157111937e-06, "loss": 0.6605, "step": 13560 }, { "epoch": 2.18, "grad_norm": 11.392024993896484, "learning_rate": 3.644151278138125e-06, "loss": 0.6034, "step": 13561 }, { "epoch": 2.18, "grad_norm": 10.80158519744873, "learning_rate": 3.642809591149902e-06, "loss": 0.6307, "step": 13562 }, { "epoch": 2.18, "grad_norm": 15.330710411071777, "learning_rate": 3.641468096187791e-06, "loss": 0.8999, "step": 13563 }, { "epoch": 2.18, "grad_norm": 12.78217601776123, "learning_rate": 3.640126793292321e-06, "loss": 0.6543, "step": 13564 }, { "epoch": 2.18, "grad_norm": 11.356146812438965, "learning_rate": 3.638785682504006e-06, "loss": 0.644, "step": 13565 }, { "epoch": 2.18, "grad_norm": 10.903900146484375, "learning_rate": 3.6374447638633493e-06, "loss": 0.4978, "step": 13566 }, { "epoch": 2.18, "grad_norm": 11.046875, "learning_rate": 3.6361040374108626e-06, "loss": 0.5551, "step": 13567 }, { "epoch": 2.18, "grad_norm": 12.256712913513184, "learning_rate": 3.6347635031870386e-06, "loss": 0.6532, "step": 13568 }, { "epoch": 2.18, "grad_norm": 9.782703399658203, "learning_rate": 3.6334231612323757e-06, "loss": 0.7367, "step": 13569 }, { "epoch": 2.18, "grad_norm": 14.319284439086914, "learning_rate": 3.632083011587356e-06, "loss": 0.7408, "step": 13570 }, { "epoch": 2.18, "grad_norm": 13.423768997192383, "learning_rate": 3.6307430542924584e-06, "loss": 0.6908, "step": 13571 }, { "epoch": 2.18, "grad_norm": 13.42258071899414, "learning_rate": 3.6294032893881625e-06, "loss": 0.6398, "step": 13572 }, { "epoch": 2.18, "grad_norm": 14.379169464111328, "learning_rate": 3.6280637169149334e-06, "loss": 0.8569, "step": 13573 }, { "epoch": 2.19, "grad_norm": 11.887478828430176, "learning_rate": 3.626724336913239e-06, "loss": 0.7042, "step": 13574 }, { "epoch": 2.19, "grad_norm": 11.837560653686523, "learning_rate": 3.625385149423535e-06, "loss": 0.6708, "step": 13575 }, { "epoch": 2.19, "grad_norm": 10.863818168640137, "learning_rate": 3.6240461544862684e-06, "loss": 0.7298, "step": 13576 }, { "epoch": 2.19, "grad_norm": 14.318840026855469, "learning_rate": 3.622707352141892e-06, "loss": 0.9961, "step": 13577 }, { "epoch": 2.19, "grad_norm": 14.250882148742676, "learning_rate": 3.621368742430842e-06, "loss": 0.7266, "step": 13578 }, { "epoch": 2.19, "grad_norm": 10.671754837036133, "learning_rate": 3.620030325393551e-06, "loss": 0.7074, "step": 13579 }, { "epoch": 2.19, "grad_norm": 68.7439193725586, "learning_rate": 3.61869210107045e-06, "loss": 0.7849, "step": 13580 }, { "epoch": 2.19, "grad_norm": 10.992599487304688, "learning_rate": 3.617354069501965e-06, "loss": 0.5876, "step": 13581 }, { "epoch": 2.19, "grad_norm": 13.391499519348145, "learning_rate": 3.6160162307285042e-06, "loss": 0.7514, "step": 13582 }, { "epoch": 2.19, "grad_norm": 12.504860877990723, "learning_rate": 3.6146785847904877e-06, "loss": 0.8184, "step": 13583 }, { "epoch": 2.19, "grad_norm": 13.031143188476562, "learning_rate": 3.613341131728313e-06, "loss": 0.623, "step": 13584 }, { "epoch": 2.19, "grad_norm": 12.187909126281738, "learning_rate": 3.6120038715823856e-06, "loss": 0.6124, "step": 13585 }, { "epoch": 2.19, "grad_norm": 15.255424499511719, "learning_rate": 3.610666804393097e-06, "loss": 0.8358, "step": 13586 }, { "epoch": 2.19, "grad_norm": 14.605413436889648, "learning_rate": 3.60932993020083e-06, "loss": 0.7355, "step": 13587 }, { "epoch": 2.19, "grad_norm": 12.5642671585083, "learning_rate": 3.6079932490459735e-06, "loss": 0.7215, "step": 13588 }, { "epoch": 2.19, "grad_norm": 16.229490280151367, "learning_rate": 3.6066567609689006e-06, "loss": 0.7438, "step": 13589 }, { "epoch": 2.19, "grad_norm": 10.514364242553711, "learning_rate": 3.605320466009977e-06, "loss": 0.6423, "step": 13590 }, { "epoch": 2.19, "grad_norm": 16.515832901000977, "learning_rate": 3.603984364209575e-06, "loss": 0.7191, "step": 13591 }, { "epoch": 2.19, "grad_norm": 12.3890380859375, "learning_rate": 3.6026484556080453e-06, "loss": 0.6797, "step": 13592 }, { "epoch": 2.19, "grad_norm": 12.451580047607422, "learning_rate": 3.601312740245748e-06, "loss": 0.6841, "step": 13593 }, { "epoch": 2.19, "grad_norm": 10.906017303466797, "learning_rate": 3.5999772181630267e-06, "loss": 0.7124, "step": 13594 }, { "epoch": 2.19, "grad_norm": 13.67624568939209, "learning_rate": 3.5986418894002184e-06, "loss": 0.7622, "step": 13595 }, { "epoch": 2.19, "grad_norm": 11.699976921081543, "learning_rate": 3.5973067539976647e-06, "loss": 0.9061, "step": 13596 }, { "epoch": 2.19, "grad_norm": 12.638691902160645, "learning_rate": 3.595971811995689e-06, "loss": 0.6755, "step": 13597 }, { "epoch": 2.19, "grad_norm": 13.418326377868652, "learning_rate": 3.594637063434621e-06, "loss": 0.6002, "step": 13598 }, { "epoch": 2.19, "grad_norm": 11.393160820007324, "learning_rate": 3.5933025083547744e-06, "loss": 0.5873, "step": 13599 }, { "epoch": 2.19, "grad_norm": 10.849506378173828, "learning_rate": 3.5919681467964582e-06, "loss": 0.5845, "step": 13600 }, { "epoch": 2.19, "grad_norm": 13.643810272216797, "learning_rate": 3.590633978799982e-06, "loss": 0.787, "step": 13601 }, { "epoch": 2.19, "grad_norm": 13.524782180786133, "learning_rate": 3.589300004405648e-06, "loss": 0.6781, "step": 13602 }, { "epoch": 2.19, "grad_norm": 11.628657341003418, "learning_rate": 3.587966223653745e-06, "loss": 0.59, "step": 13603 }, { "epoch": 2.19, "grad_norm": 13.475594520568848, "learning_rate": 3.586632636584567e-06, "loss": 0.8934, "step": 13604 }, { "epoch": 2.19, "grad_norm": 10.332905769348145, "learning_rate": 3.585299243238395e-06, "loss": 0.7613, "step": 13605 }, { "epoch": 2.19, "grad_norm": 21.767606735229492, "learning_rate": 3.583966043655499e-06, "loss": 0.5124, "step": 13606 }, { "epoch": 2.19, "grad_norm": 11.164484024047852, "learning_rate": 3.58263303787616e-06, "loss": 0.7307, "step": 13607 }, { "epoch": 2.19, "grad_norm": 11.2407865524292, "learning_rate": 3.5813002259406337e-06, "loss": 0.8431, "step": 13608 }, { "epoch": 2.19, "grad_norm": 11.47693157196045, "learning_rate": 3.5799676078891876e-06, "loss": 0.677, "step": 13609 }, { "epoch": 2.19, "grad_norm": 13.670344352722168, "learning_rate": 3.5786351837620715e-06, "loss": 0.6603, "step": 13610 }, { "epoch": 2.19, "grad_norm": 10.375473976135254, "learning_rate": 3.5773029535995274e-06, "loss": 0.7666, "step": 13611 }, { "epoch": 2.19, "grad_norm": 10.285274505615234, "learning_rate": 3.5759709174418056e-06, "loss": 0.7093, "step": 13612 }, { "epoch": 2.19, "grad_norm": 10.993366241455078, "learning_rate": 3.574639075329137e-06, "loss": 0.7466, "step": 13613 }, { "epoch": 2.19, "grad_norm": 12.563950538635254, "learning_rate": 3.5733074273017497e-06, "loss": 0.7796, "step": 13614 }, { "epoch": 2.19, "grad_norm": 10.683525085449219, "learning_rate": 3.571975973399874e-06, "loss": 0.6804, "step": 13615 }, { "epoch": 2.19, "grad_norm": 13.41348648071289, "learning_rate": 3.570644713663719e-06, "loss": 0.8287, "step": 13616 }, { "epoch": 2.19, "grad_norm": 13.807990074157715, "learning_rate": 3.5693136481335066e-06, "loss": 0.5227, "step": 13617 }, { "epoch": 2.19, "grad_norm": 12.111851692199707, "learning_rate": 3.567982776849438e-06, "loss": 0.735, "step": 13618 }, { "epoch": 2.19, "grad_norm": 11.28735065460205, "learning_rate": 3.5666520998517108e-06, "loss": 0.7372, "step": 13619 }, { "epoch": 2.19, "grad_norm": 13.58043384552002, "learning_rate": 3.565321617180526e-06, "loss": 0.6746, "step": 13620 }, { "epoch": 2.19, "grad_norm": 11.129376411437988, "learning_rate": 3.5639913288760663e-06, "loss": 0.6256, "step": 13621 }, { "epoch": 2.19, "grad_norm": 12.543471336364746, "learning_rate": 3.5626612349785174e-06, "loss": 0.6955, "step": 13622 }, { "epoch": 2.19, "grad_norm": 11.6318359375, "learning_rate": 3.5613313355280598e-06, "loss": 0.6419, "step": 13623 }, { "epoch": 2.19, "grad_norm": 15.372294425964355, "learning_rate": 3.560001630564861e-06, "loss": 0.5107, "step": 13624 }, { "epoch": 2.19, "grad_norm": 13.659976959228516, "learning_rate": 3.558672120129083e-06, "loss": 1.1114, "step": 13625 }, { "epoch": 2.19, "grad_norm": 13.324705123901367, "learning_rate": 3.5573428042608926e-06, "loss": 0.7762, "step": 13626 }, { "epoch": 2.19, "grad_norm": 11.663328170776367, "learning_rate": 3.5560136830004368e-06, "loss": 0.7142, "step": 13627 }, { "epoch": 2.19, "grad_norm": 11.644559860229492, "learning_rate": 3.554684756387867e-06, "loss": 0.6805, "step": 13628 }, { "epoch": 2.19, "grad_norm": 13.676506996154785, "learning_rate": 3.5533560244633248e-06, "loss": 0.6155, "step": 13629 }, { "epoch": 2.19, "grad_norm": 12.43027400970459, "learning_rate": 3.5520274872669415e-06, "loss": 0.7412, "step": 13630 }, { "epoch": 2.19, "grad_norm": 12.230316162109375, "learning_rate": 3.550699144838854e-06, "loss": 0.9213, "step": 13631 }, { "epoch": 2.19, "grad_norm": 13.142516136169434, "learning_rate": 3.5493709972191825e-06, "loss": 0.7702, "step": 13632 }, { "epoch": 2.19, "grad_norm": 11.223775863647461, "learning_rate": 3.5480430444480428e-06, "loss": 0.6756, "step": 13633 }, { "epoch": 2.19, "grad_norm": 11.879623413085938, "learning_rate": 3.546715286565553e-06, "loss": 0.7945, "step": 13634 }, { "epoch": 2.19, "grad_norm": 20.911151885986328, "learning_rate": 3.5453877236118128e-06, "loss": 0.7421, "step": 13635 }, { "epoch": 2.2, "grad_norm": 10.8605375289917, "learning_rate": 3.5440603556269305e-06, "loss": 0.7252, "step": 13636 }, { "epoch": 2.2, "grad_norm": 11.536102294921875, "learning_rate": 3.542733182650996e-06, "loss": 0.6515, "step": 13637 }, { "epoch": 2.2, "grad_norm": 13.812068939208984, "learning_rate": 3.541406204724096e-06, "loss": 0.7275, "step": 13638 }, { "epoch": 2.2, "grad_norm": 11.540541648864746, "learning_rate": 3.5400794218863187e-06, "loss": 0.6924, "step": 13639 }, { "epoch": 2.2, "grad_norm": 11.888813018798828, "learning_rate": 3.5387528341777356e-06, "loss": 0.6024, "step": 13640 }, { "epoch": 2.2, "grad_norm": 10.240141868591309, "learning_rate": 3.537426441638421e-06, "loss": 0.5537, "step": 13641 }, { "epoch": 2.2, "grad_norm": 10.919690132141113, "learning_rate": 3.5361002443084423e-06, "loss": 0.7986, "step": 13642 }, { "epoch": 2.2, "grad_norm": 19.728252410888672, "learning_rate": 3.5347742422278565e-06, "loss": 0.6578, "step": 13643 }, { "epoch": 2.2, "grad_norm": 12.385407447814941, "learning_rate": 3.5334484354367136e-06, "loss": 0.673, "step": 13644 }, { "epoch": 2.2, "grad_norm": 10.71657943725586, "learning_rate": 3.5321228239750682e-06, "loss": 0.7556, "step": 13645 }, { "epoch": 2.2, "grad_norm": 12.192108154296875, "learning_rate": 3.5307974078829532e-06, "loss": 0.762, "step": 13646 }, { "epoch": 2.2, "grad_norm": 10.729167938232422, "learning_rate": 3.5294721872004133e-06, "loss": 0.8145, "step": 13647 }, { "epoch": 2.2, "grad_norm": 10.97718334197998, "learning_rate": 3.5281471619674747e-06, "loss": 0.6224, "step": 13648 }, { "epoch": 2.2, "grad_norm": 12.791241645812988, "learning_rate": 3.526822332224156e-06, "loss": 0.5391, "step": 13649 }, { "epoch": 2.2, "grad_norm": 12.827373504638672, "learning_rate": 3.525497698010484e-06, "loss": 0.8034, "step": 13650 }, { "epoch": 2.2, "grad_norm": 9.669990539550781, "learning_rate": 3.524173259366462e-06, "loss": 0.4787, "step": 13651 }, { "epoch": 2.2, "grad_norm": 16.000131607055664, "learning_rate": 3.522849016332105e-06, "loss": 0.718, "step": 13652 }, { "epoch": 2.2, "grad_norm": 13.23630428314209, "learning_rate": 3.5215249689474085e-06, "loss": 0.6783, "step": 13653 }, { "epoch": 2.2, "grad_norm": 11.726489067077637, "learning_rate": 3.520201117252364e-06, "loss": 0.6337, "step": 13654 }, { "epoch": 2.2, "grad_norm": 11.621535301208496, "learning_rate": 3.518877461286967e-06, "loss": 0.6556, "step": 13655 }, { "epoch": 2.2, "grad_norm": 10.964327812194824, "learning_rate": 3.5175540010911957e-06, "loss": 0.6412, "step": 13656 }, { "epoch": 2.2, "grad_norm": 10.6217622756958, "learning_rate": 3.516230736705024e-06, "loss": 0.5851, "step": 13657 }, { "epoch": 2.2, "grad_norm": 12.668769836425781, "learning_rate": 3.514907668168429e-06, "loss": 0.7485, "step": 13658 }, { "epoch": 2.2, "grad_norm": 12.597722053527832, "learning_rate": 3.5135847955213697e-06, "loss": 0.7435, "step": 13659 }, { "epoch": 2.2, "grad_norm": 14.595346450805664, "learning_rate": 3.512262118803811e-06, "loss": 0.6302, "step": 13660 }, { "epoch": 2.2, "grad_norm": 10.882545471191406, "learning_rate": 3.5109396380557014e-06, "loss": 0.5484, "step": 13661 }, { "epoch": 2.2, "grad_norm": 12.811838150024414, "learning_rate": 3.509617353316983e-06, "loss": 0.6857, "step": 13662 }, { "epoch": 2.2, "grad_norm": 12.137211799621582, "learning_rate": 3.5082952646276105e-06, "loss": 0.7436, "step": 13663 }, { "epoch": 2.2, "grad_norm": 9.937260627746582, "learning_rate": 3.50697337202751e-06, "loss": 0.5953, "step": 13664 }, { "epoch": 2.2, "grad_norm": 11.976336479187012, "learning_rate": 3.5056516755566094e-06, "loss": 0.7838, "step": 13665 }, { "epoch": 2.2, "grad_norm": 11.11626148223877, "learning_rate": 3.5043301752548374e-06, "loss": 0.682, "step": 13666 }, { "epoch": 2.2, "grad_norm": 20.455368041992188, "learning_rate": 3.50300887116211e-06, "loss": 0.7492, "step": 13667 }, { "epoch": 2.2, "grad_norm": 11.300485610961914, "learning_rate": 3.5016877633183345e-06, "loss": 0.5321, "step": 13668 }, { "epoch": 2.2, "grad_norm": 13.923616409301758, "learning_rate": 3.5003668517634216e-06, "loss": 0.6024, "step": 13669 }, { "epoch": 2.2, "grad_norm": 13.092029571533203, "learning_rate": 3.499046136537266e-06, "loss": 0.6148, "step": 13670 }, { "epoch": 2.2, "grad_norm": 12.150940895080566, "learning_rate": 3.497725617679768e-06, "loss": 0.8556, "step": 13671 }, { "epoch": 2.2, "grad_norm": 13.807439804077148, "learning_rate": 3.4964052952308113e-06, "loss": 0.5702, "step": 13672 }, { "epoch": 2.2, "grad_norm": 12.411285400390625, "learning_rate": 3.4950851692302745e-06, "loss": 0.6787, "step": 13673 }, { "epoch": 2.2, "grad_norm": 14.293190002441406, "learning_rate": 3.49376523971804e-06, "loss": 0.6316, "step": 13674 }, { "epoch": 2.2, "grad_norm": 10.577542304992676, "learning_rate": 3.4924455067339726e-06, "loss": 0.6385, "step": 13675 }, { "epoch": 2.2, "grad_norm": 12.048298835754395, "learning_rate": 3.4911259703179413e-06, "loss": 0.6191, "step": 13676 }, { "epoch": 2.2, "grad_norm": 14.32175350189209, "learning_rate": 3.489806630509801e-06, "loss": 0.613, "step": 13677 }, { "epoch": 2.2, "grad_norm": 11.532848358154297, "learning_rate": 3.4884874873494013e-06, "loss": 0.6811, "step": 13678 }, { "epoch": 2.2, "grad_norm": 10.94014835357666, "learning_rate": 3.4871685408765945e-06, "loss": 0.5427, "step": 13679 }, { "epoch": 2.2, "grad_norm": 11.865645408630371, "learning_rate": 3.4858497911312173e-06, "loss": 0.6636, "step": 13680 }, { "epoch": 2.2, "grad_norm": 12.759921073913574, "learning_rate": 3.4845312381531015e-06, "loss": 0.7673, "step": 13681 }, { "epoch": 2.2, "grad_norm": 14.452494621276855, "learning_rate": 3.4832128819820788e-06, "loss": 0.5632, "step": 13682 }, { "epoch": 2.2, "grad_norm": 12.354889869689941, "learning_rate": 3.4818947226579737e-06, "loss": 0.7436, "step": 13683 }, { "epoch": 2.2, "grad_norm": 13.530842781066895, "learning_rate": 3.4805767602205976e-06, "loss": 0.6514, "step": 13684 }, { "epoch": 2.2, "grad_norm": 12.391562461853027, "learning_rate": 3.4792589947097666e-06, "loss": 0.647, "step": 13685 }, { "epoch": 2.2, "grad_norm": 13.015018463134766, "learning_rate": 3.47794142616528e-06, "loss": 0.7194, "step": 13686 }, { "epoch": 2.2, "grad_norm": 10.460793495178223, "learning_rate": 3.4766240546269413e-06, "loss": 0.7767, "step": 13687 }, { "epoch": 2.2, "grad_norm": 12.153456687927246, "learning_rate": 3.47530688013454e-06, "loss": 0.6949, "step": 13688 }, { "epoch": 2.2, "grad_norm": 14.57514762878418, "learning_rate": 3.4739899027278623e-06, "loss": 0.7366, "step": 13689 }, { "epoch": 2.2, "grad_norm": 14.271513938903809, "learning_rate": 3.4726731224466912e-06, "loss": 0.6684, "step": 13690 }, { "epoch": 2.2, "grad_norm": 11.300211906433105, "learning_rate": 3.4713565393308023e-06, "loss": 0.6467, "step": 13691 }, { "epoch": 2.2, "grad_norm": 15.359403610229492, "learning_rate": 3.470040153419959e-06, "loss": 0.6816, "step": 13692 }, { "epoch": 2.2, "grad_norm": 12.30099105834961, "learning_rate": 3.4687239647539318e-06, "loss": 0.6766, "step": 13693 }, { "epoch": 2.2, "grad_norm": 12.819659233093262, "learning_rate": 3.4674079733724697e-06, "loss": 0.7184, "step": 13694 }, { "epoch": 2.2, "grad_norm": 13.94396686553955, "learning_rate": 3.4660921793153324e-06, "loss": 0.8276, "step": 13695 }, { "epoch": 2.2, "grad_norm": 14.429658889770508, "learning_rate": 3.464776582622259e-06, "loss": 0.8242, "step": 13696 }, { "epoch": 2.2, "grad_norm": 14.041991233825684, "learning_rate": 3.463461183332988e-06, "loss": 0.7774, "step": 13697 }, { "epoch": 2.21, "grad_norm": 12.394572257995605, "learning_rate": 3.462145981487256e-06, "loss": 0.8248, "step": 13698 }, { "epoch": 2.21, "grad_norm": 13.321371078491211, "learning_rate": 3.4608309771247904e-06, "loss": 0.7703, "step": 13699 }, { "epoch": 2.21, "grad_norm": 11.214197158813477, "learning_rate": 3.459516170285306e-06, "loss": 0.7499, "step": 13700 }, { "epoch": 2.21, "grad_norm": 10.339448928833008, "learning_rate": 3.4582015610085273e-06, "loss": 0.5722, "step": 13701 }, { "epoch": 2.21, "grad_norm": 12.788793563842773, "learning_rate": 3.456887149334155e-06, "loss": 0.6627, "step": 13702 }, { "epoch": 2.21, "grad_norm": 11.17427921295166, "learning_rate": 3.4555729353018963e-06, "loss": 0.7105, "step": 13703 }, { "epoch": 2.21, "grad_norm": 11.704593658447266, "learning_rate": 3.4542589189514517e-06, "loss": 0.8086, "step": 13704 }, { "epoch": 2.21, "grad_norm": 12.644407272338867, "learning_rate": 3.4529451003225067e-06, "loss": 0.8335, "step": 13705 }, { "epoch": 2.21, "grad_norm": 11.28946304321289, "learning_rate": 3.4516314794547534e-06, "loss": 0.6311, "step": 13706 }, { "epoch": 2.21, "grad_norm": 10.51651382446289, "learning_rate": 3.450318056387866e-06, "loss": 0.7084, "step": 13707 }, { "epoch": 2.21, "grad_norm": 12.661309242248535, "learning_rate": 3.4490048311615154e-06, "loss": 0.6711, "step": 13708 }, { "epoch": 2.21, "grad_norm": 11.26526927947998, "learning_rate": 3.4476918038153773e-06, "loss": 0.7047, "step": 13709 }, { "epoch": 2.21, "grad_norm": 12.832565307617188, "learning_rate": 3.4463789743891086e-06, "loss": 0.6321, "step": 13710 }, { "epoch": 2.21, "grad_norm": 12.475379943847656, "learning_rate": 3.445066342922361e-06, "loss": 0.6646, "step": 13711 }, { "epoch": 2.21, "grad_norm": 11.872003555297852, "learning_rate": 3.44375390945479e-06, "loss": 0.6572, "step": 13712 }, { "epoch": 2.21, "grad_norm": 11.531255722045898, "learning_rate": 3.4424416740260357e-06, "loss": 0.7121, "step": 13713 }, { "epoch": 2.21, "grad_norm": 15.943255424499512, "learning_rate": 3.4411296366757386e-06, "loss": 0.565, "step": 13714 }, { "epoch": 2.21, "grad_norm": 12.215544700622559, "learning_rate": 3.43981779744353e-06, "loss": 0.7098, "step": 13715 }, { "epoch": 2.21, "grad_norm": 12.16107177734375, "learning_rate": 3.4385061563690303e-06, "loss": 0.7708, "step": 13716 }, { "epoch": 2.21, "grad_norm": 11.134607315063477, "learning_rate": 3.4371947134918658e-06, "loss": 0.8021, "step": 13717 }, { "epoch": 2.21, "grad_norm": 13.729071617126465, "learning_rate": 3.435883468851644e-06, "loss": 0.6232, "step": 13718 }, { "epoch": 2.21, "grad_norm": 11.081302642822266, "learning_rate": 3.4345724224879793e-06, "loss": 0.8251, "step": 13719 }, { "epoch": 2.21, "grad_norm": 10.947044372558594, "learning_rate": 3.4332615744404694e-06, "loss": 0.5849, "step": 13720 }, { "epoch": 2.21, "grad_norm": 13.666549682617188, "learning_rate": 3.4319509247487083e-06, "loss": 0.6473, "step": 13721 }, { "epoch": 2.21, "grad_norm": 12.578267097473145, "learning_rate": 3.430640473452286e-06, "loss": 0.6254, "step": 13722 }, { "epoch": 2.21, "grad_norm": 11.486944198608398, "learning_rate": 3.429330220590792e-06, "loss": 0.4342, "step": 13723 }, { "epoch": 2.21, "grad_norm": 11.591387748718262, "learning_rate": 3.428020166203797e-06, "loss": 0.8349, "step": 13724 }, { "epoch": 2.21, "grad_norm": 11.235136985778809, "learning_rate": 3.4267103103308774e-06, "loss": 0.5444, "step": 13725 }, { "epoch": 2.21, "grad_norm": 13.182741165161133, "learning_rate": 3.425400653011598e-06, "loss": 0.6387, "step": 13726 }, { "epoch": 2.21, "grad_norm": 13.661701202392578, "learning_rate": 3.4240911942855136e-06, "loss": 0.8261, "step": 13727 }, { "epoch": 2.21, "grad_norm": 11.59208869934082, "learning_rate": 3.4227819341921854e-06, "loss": 0.6991, "step": 13728 }, { "epoch": 2.21, "grad_norm": 9.956355094909668, "learning_rate": 3.4214728727711545e-06, "loss": 0.7191, "step": 13729 }, { "epoch": 2.21, "grad_norm": 14.13731575012207, "learning_rate": 3.4201640100619672e-06, "loss": 0.8278, "step": 13730 }, { "epoch": 2.21, "grad_norm": 12.440681457519531, "learning_rate": 3.4188553461041584e-06, "loss": 0.7109, "step": 13731 }, { "epoch": 2.21, "grad_norm": 12.856305122375488, "learning_rate": 3.4175468809372535e-06, "loss": 0.6937, "step": 13732 }, { "epoch": 2.21, "grad_norm": 11.894834518432617, "learning_rate": 3.416238614600783e-06, "loss": 0.7712, "step": 13733 }, { "epoch": 2.21, "grad_norm": 13.082294464111328, "learning_rate": 3.4149305471342607e-06, "loss": 0.6949, "step": 13734 }, { "epoch": 2.21, "grad_norm": 12.070653915405273, "learning_rate": 3.4136226785771963e-06, "loss": 0.6809, "step": 13735 }, { "epoch": 2.21, "grad_norm": 10.722041130065918, "learning_rate": 3.4123150089691003e-06, "loss": 0.5566, "step": 13736 }, { "epoch": 2.21, "grad_norm": 12.7687406539917, "learning_rate": 3.4110075383494655e-06, "loss": 0.59, "step": 13737 }, { "epoch": 2.21, "grad_norm": 9.708044052124023, "learning_rate": 3.409700266757795e-06, "loss": 0.6953, "step": 13738 }, { "epoch": 2.21, "grad_norm": 14.22176456451416, "learning_rate": 3.40839319423357e-06, "loss": 0.5955, "step": 13739 }, { "epoch": 2.21, "grad_norm": 13.915772438049316, "learning_rate": 3.40708632081627e-06, "loss": 0.5626, "step": 13740 }, { "epoch": 2.21, "grad_norm": 11.559593200683594, "learning_rate": 3.4057796465453775e-06, "loss": 0.7857, "step": 13741 }, { "epoch": 2.21, "grad_norm": 12.699880599975586, "learning_rate": 3.404473171460355e-06, "loss": 0.744, "step": 13742 }, { "epoch": 2.21, "grad_norm": 13.30777645111084, "learning_rate": 3.40316689560067e-06, "loss": 0.7805, "step": 13743 }, { "epoch": 2.21, "grad_norm": 13.449222564697266, "learning_rate": 3.401860819005782e-06, "loss": 0.5819, "step": 13744 }, { "epoch": 2.21, "grad_norm": 12.598854064941406, "learning_rate": 3.400554941715142e-06, "loss": 0.6464, "step": 13745 }, { "epoch": 2.21, "grad_norm": 12.80246353149414, "learning_rate": 3.3992492637681882e-06, "loss": 0.7074, "step": 13746 }, { "epoch": 2.21, "grad_norm": 11.562127113342285, "learning_rate": 3.3979437852043696e-06, "loss": 0.8134, "step": 13747 }, { "epoch": 2.21, "grad_norm": 13.081464767456055, "learning_rate": 3.3966385060631124e-06, "loss": 0.8079, "step": 13748 }, { "epoch": 2.21, "grad_norm": 12.653249740600586, "learning_rate": 3.39533342638385e-06, "loss": 0.5887, "step": 13749 }, { "epoch": 2.21, "grad_norm": 13.622257232666016, "learning_rate": 3.394028546206002e-06, "loss": 0.6758, "step": 13750 }, { "epoch": 2.21, "grad_norm": 12.538056373596191, "learning_rate": 3.3927238655689785e-06, "loss": 0.6869, "step": 13751 }, { "epoch": 2.21, "grad_norm": 13.779552459716797, "learning_rate": 3.391419384512197e-06, "loss": 0.8369, "step": 13752 }, { "epoch": 2.21, "grad_norm": 11.497603416442871, "learning_rate": 3.3901151030750524e-06, "loss": 0.5591, "step": 13753 }, { "epoch": 2.21, "grad_norm": 13.070073127746582, "learning_rate": 3.3888110212969504e-06, "loss": 0.8384, "step": 13754 }, { "epoch": 2.21, "grad_norm": 14.8796968460083, "learning_rate": 3.38750713921728e-06, "loss": 0.6394, "step": 13755 }, { "epoch": 2.21, "grad_norm": 13.744235038757324, "learning_rate": 3.3862034568754197e-06, "loss": 0.6285, "step": 13756 }, { "epoch": 2.21, "grad_norm": 12.375431060791016, "learning_rate": 3.384899974310757e-06, "loss": 0.6802, "step": 13757 }, { "epoch": 2.21, "grad_norm": 12.570847511291504, "learning_rate": 3.383596691562663e-06, "loss": 0.7185, "step": 13758 }, { "epoch": 2.21, "grad_norm": 11.743610382080078, "learning_rate": 3.3822936086704993e-06, "loss": 0.9262, "step": 13759 }, { "epoch": 2.22, "grad_norm": 15.148598670959473, "learning_rate": 3.380990725673635e-06, "loss": 0.7348, "step": 13760 }, { "epoch": 2.22, "grad_norm": 11.204832077026367, "learning_rate": 3.3796880426114185e-06, "loss": 0.5715, "step": 13761 }, { "epoch": 2.22, "grad_norm": 13.976383209228516, "learning_rate": 3.378385559523205e-06, "loss": 0.6152, "step": 13762 }, { "epoch": 2.22, "grad_norm": 9.96169376373291, "learning_rate": 3.3770832764483317e-06, "loss": 0.616, "step": 13763 }, { "epoch": 2.22, "grad_norm": 11.392471313476562, "learning_rate": 3.3757811934261374e-06, "loss": 0.8171, "step": 13764 }, { "epoch": 2.22, "grad_norm": 11.082906723022461, "learning_rate": 3.374479310495957e-06, "loss": 0.5746, "step": 13765 }, { "epoch": 2.22, "grad_norm": 12.500931739807129, "learning_rate": 3.3731776276971128e-06, "loss": 0.9236, "step": 13766 }, { "epoch": 2.22, "grad_norm": 11.312355041503906, "learning_rate": 3.3718761450689197e-06, "loss": 0.711, "step": 13767 }, { "epoch": 2.22, "grad_norm": 12.22961139678955, "learning_rate": 3.3705748626506963e-06, "loss": 0.682, "step": 13768 }, { "epoch": 2.22, "grad_norm": 12.711225509643555, "learning_rate": 3.3692737804817464e-06, "loss": 0.7529, "step": 13769 }, { "epoch": 2.22, "grad_norm": 11.18406867980957, "learning_rate": 3.3679728986013683e-06, "loss": 0.7059, "step": 13770 }, { "epoch": 2.22, "grad_norm": 12.894728660583496, "learning_rate": 3.3666722170488617e-06, "loss": 0.803, "step": 13771 }, { "epoch": 2.22, "grad_norm": 16.143436431884766, "learning_rate": 3.3653717358635095e-06, "loss": 0.638, "step": 13772 }, { "epoch": 2.22, "grad_norm": 12.214151382446289, "learning_rate": 3.364071455084601e-06, "loss": 0.791, "step": 13773 }, { "epoch": 2.22, "grad_norm": 11.084589958190918, "learning_rate": 3.3627713747514092e-06, "loss": 0.827, "step": 13774 }, { "epoch": 2.22, "grad_norm": 13.084232330322266, "learning_rate": 3.361471494903201e-06, "loss": 0.652, "step": 13775 }, { "epoch": 2.22, "grad_norm": 10.726874351501465, "learning_rate": 3.3601718155792473e-06, "loss": 0.7225, "step": 13776 }, { "epoch": 2.22, "grad_norm": 12.169782638549805, "learning_rate": 3.3588723368188024e-06, "loss": 0.8035, "step": 13777 }, { "epoch": 2.22, "grad_norm": 13.029605865478516, "learning_rate": 3.357573058661118e-06, "loss": 0.667, "step": 13778 }, { "epoch": 2.22, "grad_norm": 11.53877067565918, "learning_rate": 3.3562739811454427e-06, "loss": 0.546, "step": 13779 }, { "epoch": 2.22, "grad_norm": 10.464457511901855, "learning_rate": 3.3549751043110123e-06, "loss": 0.5589, "step": 13780 }, { "epoch": 2.22, "grad_norm": 11.851531028747559, "learning_rate": 3.353676428197068e-06, "loss": 0.5062, "step": 13781 }, { "epoch": 2.22, "grad_norm": 10.865971565246582, "learning_rate": 3.3523779528428337e-06, "loss": 0.6821, "step": 13782 }, { "epoch": 2.22, "grad_norm": 11.405611991882324, "learning_rate": 3.3510796782875242e-06, "loss": 0.4409, "step": 13783 }, { "epoch": 2.22, "grad_norm": 13.280759811401367, "learning_rate": 3.3497816045703702e-06, "loss": 0.69, "step": 13784 }, { "epoch": 2.22, "grad_norm": 13.368683815002441, "learning_rate": 3.348483731730573e-06, "loss": 0.7627, "step": 13785 }, { "epoch": 2.22, "grad_norm": 16.223058700561523, "learning_rate": 3.3471860598073336e-06, "loss": 0.6469, "step": 13786 }, { "epoch": 2.22, "grad_norm": 12.466206550598145, "learning_rate": 3.3458885888398574e-06, "loss": 0.6128, "step": 13787 }, { "epoch": 2.22, "grad_norm": 10.077193260192871, "learning_rate": 3.3445913188673306e-06, "loss": 0.5633, "step": 13788 }, { "epoch": 2.22, "grad_norm": 12.880247116088867, "learning_rate": 3.343294249928937e-06, "loss": 0.6416, "step": 13789 }, { "epoch": 2.22, "grad_norm": 11.02946662902832, "learning_rate": 3.341997382063863e-06, "loss": 0.902, "step": 13790 }, { "epoch": 2.22, "grad_norm": 12.377063751220703, "learning_rate": 3.340700715311274e-06, "loss": 0.6029, "step": 13791 }, { "epoch": 2.22, "grad_norm": 13.484081268310547, "learning_rate": 3.339404249710343e-06, "loss": 0.8298, "step": 13792 }, { "epoch": 2.22, "grad_norm": 11.558128356933594, "learning_rate": 3.3381079853002307e-06, "loss": 0.5299, "step": 13793 }, { "epoch": 2.22, "grad_norm": 12.812358856201172, "learning_rate": 3.336811922120088e-06, "loss": 0.782, "step": 13794 }, { "epoch": 2.22, "grad_norm": 11.788616180419922, "learning_rate": 3.335516060209069e-06, "loss": 0.7672, "step": 13795 }, { "epoch": 2.22, "grad_norm": 13.208732604980469, "learning_rate": 3.3342203996063115e-06, "loss": 0.6666, "step": 13796 }, { "epoch": 2.22, "grad_norm": 12.509199142456055, "learning_rate": 3.332924940350959e-06, "loss": 0.8486, "step": 13797 }, { "epoch": 2.22, "grad_norm": 12.306282997131348, "learning_rate": 3.3316296824821382e-06, "loss": 0.9014, "step": 13798 }, { "epoch": 2.22, "grad_norm": 12.450238227844238, "learning_rate": 3.3303346260389713e-06, "loss": 0.7237, "step": 13799 }, { "epoch": 2.22, "grad_norm": 13.697738647460938, "learning_rate": 3.3290397710605825e-06, "loss": 0.7862, "step": 13800 }, { "epoch": 2.22, "grad_norm": 12.172200202941895, "learning_rate": 3.3277451175860833e-06, "loss": 0.64, "step": 13801 }, { "epoch": 2.22, "grad_norm": 10.795671463012695, "learning_rate": 3.3264506656545747e-06, "loss": 0.7747, "step": 13802 }, { "epoch": 2.22, "grad_norm": 11.871382713317871, "learning_rate": 3.325156415305161e-06, "loss": 0.6494, "step": 13803 }, { "epoch": 2.22, "grad_norm": 13.074551582336426, "learning_rate": 3.323862366576939e-06, "loss": 0.5614, "step": 13804 }, { "epoch": 2.22, "grad_norm": 11.635284423828125, "learning_rate": 3.322568519508992e-06, "loss": 0.5754, "step": 13805 }, { "epoch": 2.22, "grad_norm": 12.243230819702148, "learning_rate": 3.3212748741404076e-06, "loss": 0.6155, "step": 13806 }, { "epoch": 2.22, "grad_norm": 12.042880058288574, "learning_rate": 3.319981430510256e-06, "loss": 0.8038, "step": 13807 }, { "epoch": 2.22, "grad_norm": 9.439020156860352, "learning_rate": 3.3186881886576127e-06, "loss": 0.6521, "step": 13808 }, { "epoch": 2.22, "grad_norm": 12.483598709106445, "learning_rate": 3.3173951486215396e-06, "loss": 0.7576, "step": 13809 }, { "epoch": 2.22, "grad_norm": 11.266645431518555, "learning_rate": 3.3161023104410895e-06, "loss": 0.8451, "step": 13810 }, { "epoch": 2.22, "grad_norm": 14.075155258178711, "learning_rate": 3.3148096741553214e-06, "loss": 0.6293, "step": 13811 }, { "epoch": 2.22, "grad_norm": 10.939785957336426, "learning_rate": 3.313517239803278e-06, "loss": 0.6749, "step": 13812 }, { "epoch": 2.22, "grad_norm": 11.96951961517334, "learning_rate": 3.3122250074239938e-06, "loss": 0.6223, "step": 13813 }, { "epoch": 2.22, "grad_norm": 11.511959075927734, "learning_rate": 3.310932977056511e-06, "loss": 0.6993, "step": 13814 }, { "epoch": 2.22, "grad_norm": 57.7967643737793, "learning_rate": 3.3096411487398484e-06, "loss": 0.8666, "step": 13815 }, { "epoch": 2.22, "grad_norm": 13.79617691040039, "learning_rate": 3.3083495225130356e-06, "loss": 0.6778, "step": 13816 }, { "epoch": 2.22, "grad_norm": 15.75503921508789, "learning_rate": 3.3070580984150825e-06, "loss": 0.7275, "step": 13817 }, { "epoch": 2.22, "grad_norm": 13.29902172088623, "learning_rate": 3.305766876484996e-06, "loss": 0.6597, "step": 13818 }, { "epoch": 2.22, "grad_norm": 13.925023078918457, "learning_rate": 3.304475856761784e-06, "loss": 0.734, "step": 13819 }, { "epoch": 2.22, "grad_norm": 12.892030715942383, "learning_rate": 3.303185039284438e-06, "loss": 0.907, "step": 13820 }, { "epoch": 2.22, "grad_norm": 12.042755126953125, "learning_rate": 3.301894424091955e-06, "loss": 0.7889, "step": 13821 }, { "epoch": 2.23, "grad_norm": 14.549577713012695, "learning_rate": 3.300604011223316e-06, "loss": 0.7797, "step": 13822 }, { "epoch": 2.23, "grad_norm": 9.972399711608887, "learning_rate": 3.299313800717496e-06, "loss": 0.6366, "step": 13823 }, { "epoch": 2.23, "grad_norm": 9.694838523864746, "learning_rate": 3.2980237926134707e-06, "loss": 0.6241, "step": 13824 }, { "epoch": 2.23, "grad_norm": 15.017084121704102, "learning_rate": 3.296733986950209e-06, "loss": 0.6971, "step": 13825 }, { "epoch": 2.23, "grad_norm": 11.675095558166504, "learning_rate": 3.295444383766665e-06, "loss": 0.5545, "step": 13826 }, { "epoch": 2.23, "grad_norm": 10.65314769744873, "learning_rate": 3.2941549831017994e-06, "loss": 0.7581, "step": 13827 }, { "epoch": 2.23, "grad_norm": 12.608674049377441, "learning_rate": 3.2928657849945554e-06, "loss": 0.5657, "step": 13828 }, { "epoch": 2.23, "grad_norm": 11.317018508911133, "learning_rate": 3.2915767894838734e-06, "loss": 0.6131, "step": 13829 }, { "epoch": 2.23, "grad_norm": 11.881402969360352, "learning_rate": 3.290287996608694e-06, "loss": 0.5867, "step": 13830 }, { "epoch": 2.23, "grad_norm": 13.817447662353516, "learning_rate": 3.288999406407941e-06, "loss": 0.7745, "step": 13831 }, { "epoch": 2.23, "grad_norm": 14.196622848510742, "learning_rate": 3.287711018920543e-06, "loss": 0.7566, "step": 13832 }, { "epoch": 2.23, "grad_norm": 10.46370792388916, "learning_rate": 3.2864228341854155e-06, "loss": 0.7473, "step": 13833 }, { "epoch": 2.23, "grad_norm": 10.409895896911621, "learning_rate": 3.285134852241465e-06, "loss": 0.7684, "step": 13834 }, { "epoch": 2.23, "grad_norm": 14.125809669494629, "learning_rate": 3.283847073127604e-06, "loss": 0.721, "step": 13835 }, { "epoch": 2.23, "grad_norm": 11.201644897460938, "learning_rate": 3.2825594968827267e-06, "loss": 0.8099, "step": 13836 }, { "epoch": 2.23, "grad_norm": 14.878535270690918, "learning_rate": 3.2812721235457247e-06, "loss": 0.7744, "step": 13837 }, { "epoch": 2.23, "grad_norm": 12.422591209411621, "learning_rate": 3.2799849531554884e-06, "loss": 0.5783, "step": 13838 }, { "epoch": 2.23, "grad_norm": 11.385499954223633, "learning_rate": 3.2786979857508937e-06, "loss": 0.6599, "step": 13839 }, { "epoch": 2.23, "grad_norm": 12.281511306762695, "learning_rate": 3.277411221370821e-06, "loss": 0.5333, "step": 13840 }, { "epoch": 2.23, "grad_norm": 11.83083724975586, "learning_rate": 3.276124660054135e-06, "loss": 0.6504, "step": 13841 }, { "epoch": 2.23, "grad_norm": 11.984477996826172, "learning_rate": 3.2748383018396946e-06, "loss": 0.51, "step": 13842 }, { "epoch": 2.23, "grad_norm": 11.437212944030762, "learning_rate": 3.273552146766362e-06, "loss": 0.7429, "step": 13843 }, { "epoch": 2.23, "grad_norm": 14.443803787231445, "learning_rate": 3.2722661948729805e-06, "loss": 0.7194, "step": 13844 }, { "epoch": 2.23, "grad_norm": 10.644340515136719, "learning_rate": 3.2709804461983973e-06, "loss": 0.5685, "step": 13845 }, { "epoch": 2.23, "grad_norm": 12.729004859924316, "learning_rate": 3.269694900781454e-06, "loss": 0.7364, "step": 13846 }, { "epoch": 2.23, "grad_norm": 11.713141441345215, "learning_rate": 3.2684095586609767e-06, "loss": 0.7187, "step": 13847 }, { "epoch": 2.23, "grad_norm": 11.266249656677246, "learning_rate": 3.267124419875788e-06, "loss": 0.458, "step": 13848 }, { "epoch": 2.23, "grad_norm": 9.694705963134766, "learning_rate": 3.2658394844647134e-06, "loss": 0.5485, "step": 13849 }, { "epoch": 2.23, "grad_norm": 11.628888130187988, "learning_rate": 3.26455475246656e-06, "loss": 0.6129, "step": 13850 }, { "epoch": 2.23, "grad_norm": 12.422158241271973, "learning_rate": 3.2632702239201406e-06, "loss": 0.7151, "step": 13851 }, { "epoch": 2.23, "grad_norm": 15.590764999389648, "learning_rate": 3.2619858988642537e-06, "loss": 0.6808, "step": 13852 }, { "epoch": 2.23, "grad_norm": 11.6080904006958, "learning_rate": 3.260701777337689e-06, "loss": 0.7103, "step": 13853 }, { "epoch": 2.23, "grad_norm": 10.758478164672852, "learning_rate": 3.2594178593792427e-06, "loss": 0.7014, "step": 13854 }, { "epoch": 2.23, "grad_norm": 11.631827354431152, "learning_rate": 3.2581341450276915e-06, "loss": 0.7609, "step": 13855 }, { "epoch": 2.23, "grad_norm": 12.597065925598145, "learning_rate": 3.256850634321811e-06, "loss": 0.7308, "step": 13856 }, { "epoch": 2.23, "grad_norm": 12.047779083251953, "learning_rate": 3.2555673273003752e-06, "loss": 0.5978, "step": 13857 }, { "epoch": 2.23, "grad_norm": 12.333248138427734, "learning_rate": 3.2542842240021433e-06, "loss": 0.8109, "step": 13858 }, { "epoch": 2.23, "grad_norm": 13.308510780334473, "learning_rate": 3.2530013244658776e-06, "loss": 0.7155, "step": 13859 }, { "epoch": 2.23, "grad_norm": 11.006643295288086, "learning_rate": 3.2517186287303283e-06, "loss": 0.822, "step": 13860 }, { "epoch": 2.23, "grad_norm": 13.306588172912598, "learning_rate": 3.250436136834235e-06, "loss": 0.7262, "step": 13861 }, { "epoch": 2.23, "grad_norm": 12.52178955078125, "learning_rate": 3.2491538488163454e-06, "loss": 0.6372, "step": 13862 }, { "epoch": 2.23, "grad_norm": 12.053284645080566, "learning_rate": 3.2478717647153845e-06, "loss": 0.6237, "step": 13863 }, { "epoch": 2.23, "grad_norm": 12.097578048706055, "learning_rate": 3.2465898845700838e-06, "loss": 0.8447, "step": 13864 }, { "epoch": 2.23, "grad_norm": 9.932350158691406, "learning_rate": 3.2453082084191655e-06, "loss": 0.9054, "step": 13865 }, { "epoch": 2.23, "grad_norm": 11.868432998657227, "learning_rate": 3.2440267363013423e-06, "loss": 0.6814, "step": 13866 }, { "epoch": 2.23, "grad_norm": 9.70926284790039, "learning_rate": 3.2427454682553195e-06, "loss": 0.6247, "step": 13867 }, { "epoch": 2.23, "grad_norm": 11.24247932434082, "learning_rate": 3.241464404319804e-06, "loss": 0.6347, "step": 13868 }, { "epoch": 2.23, "grad_norm": 14.569125175476074, "learning_rate": 3.2401835445334863e-06, "loss": 0.7579, "step": 13869 }, { "epoch": 2.23, "grad_norm": 12.60348892211914, "learning_rate": 3.2389028889350625e-06, "loss": 0.4658, "step": 13870 }, { "epoch": 2.23, "grad_norm": 14.858159065246582, "learning_rate": 3.2376224375632126e-06, "loss": 0.5296, "step": 13871 }, { "epoch": 2.23, "grad_norm": 12.768889427185059, "learning_rate": 3.2363421904566128e-06, "loss": 0.7731, "step": 13872 }, { "epoch": 2.23, "grad_norm": 12.27547836303711, "learning_rate": 3.235062147653938e-06, "loss": 0.6156, "step": 13873 }, { "epoch": 2.23, "grad_norm": 12.680144309997559, "learning_rate": 3.2337823091938494e-06, "loss": 0.9455, "step": 13874 }, { "epoch": 2.23, "grad_norm": 11.171124458312988, "learning_rate": 3.2325026751150114e-06, "loss": 0.8471, "step": 13875 }, { "epoch": 2.23, "grad_norm": 11.463830947875977, "learning_rate": 3.231223245456073e-06, "loss": 0.7243, "step": 13876 }, { "epoch": 2.23, "grad_norm": 12.054468154907227, "learning_rate": 3.2299440202556775e-06, "loss": 0.6952, "step": 13877 }, { "epoch": 2.23, "grad_norm": 14.404387474060059, "learning_rate": 3.2286649995524733e-06, "loss": 0.6927, "step": 13878 }, { "epoch": 2.23, "grad_norm": 14.083311080932617, "learning_rate": 3.2273861833850906e-06, "loss": 0.7676, "step": 13879 }, { "epoch": 2.23, "grad_norm": 10.925580024719238, "learning_rate": 3.226107571792154e-06, "loss": 0.6843, "step": 13880 }, { "epoch": 2.23, "grad_norm": 10.731063842773438, "learning_rate": 3.224829164812292e-06, "loss": 0.5825, "step": 13881 }, { "epoch": 2.23, "grad_norm": 13.519906044006348, "learning_rate": 3.223550962484114e-06, "loss": 0.6815, "step": 13882 }, { "epoch": 2.23, "grad_norm": 15.952306747436523, "learning_rate": 3.2222729648462357e-06, "loss": 0.7528, "step": 13883 }, { "epoch": 2.24, "grad_norm": 12.073259353637695, "learning_rate": 3.2209951719372544e-06, "loss": 0.5422, "step": 13884 }, { "epoch": 2.24, "grad_norm": 13.543663024902344, "learning_rate": 3.2197175837957707e-06, "loss": 0.8251, "step": 13885 }, { "epoch": 2.24, "grad_norm": 10.650972366333008, "learning_rate": 3.2184402004603778e-06, "loss": 0.5339, "step": 13886 }, { "epoch": 2.24, "grad_norm": 12.565300941467285, "learning_rate": 3.2171630219696582e-06, "loss": 0.7336, "step": 13887 }, { "epoch": 2.24, "grad_norm": 10.762319564819336, "learning_rate": 3.215886048362187e-06, "loss": 0.6555, "step": 13888 }, { "epoch": 2.24, "grad_norm": 12.544095039367676, "learning_rate": 3.214609279676544e-06, "loss": 0.6596, "step": 13889 }, { "epoch": 2.24, "grad_norm": 12.047447204589844, "learning_rate": 3.2133327159512904e-06, "loss": 0.8566, "step": 13890 }, { "epoch": 2.24, "grad_norm": 12.0676851272583, "learning_rate": 3.212056357224984e-06, "loss": 0.7321, "step": 13891 }, { "epoch": 2.24, "grad_norm": 11.878808975219727, "learning_rate": 3.2107802035361847e-06, "loss": 0.8602, "step": 13892 }, { "epoch": 2.24, "grad_norm": 13.344858169555664, "learning_rate": 3.2095042549234344e-06, "loss": 0.5518, "step": 13893 }, { "epoch": 2.24, "grad_norm": 11.017452239990234, "learning_rate": 3.208228511425281e-06, "loss": 0.5594, "step": 13894 }, { "epoch": 2.24, "grad_norm": 12.217209815979004, "learning_rate": 3.2069529730802563e-06, "loss": 0.7159, "step": 13895 }, { "epoch": 2.24, "grad_norm": 11.113956451416016, "learning_rate": 3.205677639926886e-06, "loss": 0.5757, "step": 13896 }, { "epoch": 2.24, "grad_norm": 13.190755844116211, "learning_rate": 3.2044025120036994e-06, "loss": 0.6858, "step": 13897 }, { "epoch": 2.24, "grad_norm": 12.365775108337402, "learning_rate": 3.2031275893492074e-06, "loss": 0.8455, "step": 13898 }, { "epoch": 2.24, "grad_norm": 10.79280948638916, "learning_rate": 3.201852872001926e-06, "loss": 0.5481, "step": 13899 }, { "epoch": 2.24, "grad_norm": 11.59760856628418, "learning_rate": 3.200578360000356e-06, "loss": 0.5186, "step": 13900 }, { "epoch": 2.24, "grad_norm": 10.135321617126465, "learning_rate": 3.1993040533829934e-06, "loss": 0.7047, "step": 13901 }, { "epoch": 2.24, "grad_norm": 10.740212440490723, "learning_rate": 3.198029952188336e-06, "loss": 0.6195, "step": 13902 }, { "epoch": 2.24, "grad_norm": 12.934005737304688, "learning_rate": 3.196756056454866e-06, "loss": 0.5085, "step": 13903 }, { "epoch": 2.24, "grad_norm": 9.714393615722656, "learning_rate": 3.195482366221061e-06, "loss": 0.5798, "step": 13904 }, { "epoch": 2.24, "grad_norm": 16.496009826660156, "learning_rate": 3.194208881525395e-06, "loss": 0.7499, "step": 13905 }, { "epoch": 2.24, "grad_norm": 12.76415729522705, "learning_rate": 3.192935602406341e-06, "loss": 0.6139, "step": 13906 }, { "epoch": 2.24, "grad_norm": 13.151311874389648, "learning_rate": 3.1916625289023504e-06, "loss": 0.5518, "step": 13907 }, { "epoch": 2.24, "grad_norm": 13.440533638000488, "learning_rate": 3.1903896610518868e-06, "loss": 0.7786, "step": 13908 }, { "epoch": 2.24, "grad_norm": 12.084943771362305, "learning_rate": 3.189116998893391e-06, "loss": 0.6953, "step": 13909 }, { "epoch": 2.24, "grad_norm": 14.84096622467041, "learning_rate": 3.1878445424653114e-06, "loss": 0.6953, "step": 13910 }, { "epoch": 2.24, "grad_norm": 13.650896072387695, "learning_rate": 3.186572291806083e-06, "loss": 0.7885, "step": 13911 }, { "epoch": 2.24, "grad_norm": 12.373856544494629, "learning_rate": 3.1853002469541285e-06, "loss": 0.6996, "step": 13912 }, { "epoch": 2.24, "grad_norm": 15.340411186218262, "learning_rate": 3.1840284079478813e-06, "loss": 0.792, "step": 13913 }, { "epoch": 2.24, "grad_norm": 15.109197616577148, "learning_rate": 3.1827567748257537e-06, "loss": 0.662, "step": 13914 }, { "epoch": 2.24, "grad_norm": 14.845292091369629, "learning_rate": 3.181485347626154e-06, "loss": 0.9825, "step": 13915 }, { "epoch": 2.24, "grad_norm": 10.14783000946045, "learning_rate": 3.180214126387494e-06, "loss": 0.7938, "step": 13916 }, { "epoch": 2.24, "grad_norm": 12.340073585510254, "learning_rate": 3.178943111148165e-06, "loss": 0.6544, "step": 13917 }, { "epoch": 2.24, "grad_norm": 10.256020545959473, "learning_rate": 3.1776723019465672e-06, "loss": 0.6815, "step": 13918 }, { "epoch": 2.24, "grad_norm": 11.50663948059082, "learning_rate": 3.176401698821082e-06, "loss": 0.6237, "step": 13919 }, { "epoch": 2.24, "grad_norm": 11.441509246826172, "learning_rate": 3.175131301810088e-06, "loss": 0.8128, "step": 13920 }, { "epoch": 2.24, "grad_norm": 16.586149215698242, "learning_rate": 3.1738611109519646e-06, "loss": 0.6887, "step": 13921 }, { "epoch": 2.24, "grad_norm": 14.64870548248291, "learning_rate": 3.1725911262850716e-06, "loss": 0.6341, "step": 13922 }, { "epoch": 2.24, "grad_norm": 12.438934326171875, "learning_rate": 3.171321347847779e-06, "loss": 0.856, "step": 13923 }, { "epoch": 2.24, "grad_norm": 13.431340217590332, "learning_rate": 3.1700517756784366e-06, "loss": 0.6732, "step": 13924 }, { "epoch": 2.24, "grad_norm": 10.854561805725098, "learning_rate": 3.1687824098153917e-06, "loss": 0.5944, "step": 13925 }, { "epoch": 2.24, "grad_norm": 12.281331062316895, "learning_rate": 3.16751325029699e-06, "loss": 0.7691, "step": 13926 }, { "epoch": 2.24, "grad_norm": 15.214825630187988, "learning_rate": 3.16624429716157e-06, "loss": 0.5834, "step": 13927 }, { "epoch": 2.24, "grad_norm": 14.107604026794434, "learning_rate": 3.1649755504474555e-06, "loss": 0.8027, "step": 13928 }, { "epoch": 2.24, "grad_norm": 13.34264087677002, "learning_rate": 3.163707010192978e-06, "loss": 0.7835, "step": 13929 }, { "epoch": 2.24, "grad_norm": 12.258047103881836, "learning_rate": 3.1624386764364525e-06, "loss": 0.6312, "step": 13930 }, { "epoch": 2.24, "grad_norm": 12.671016693115234, "learning_rate": 3.1611705492161837e-06, "loss": 0.5837, "step": 13931 }, { "epoch": 2.24, "grad_norm": 11.30209732055664, "learning_rate": 3.159902628570487e-06, "loss": 0.5991, "step": 13932 }, { "epoch": 2.24, "grad_norm": 16.29055404663086, "learning_rate": 3.158634914537657e-06, "loss": 0.8556, "step": 13933 }, { "epoch": 2.24, "grad_norm": 12.375520706176758, "learning_rate": 3.1573674071559833e-06, "loss": 0.8517, "step": 13934 }, { "epoch": 2.24, "grad_norm": 10.486623764038086, "learning_rate": 3.1561001064637576e-06, "loss": 0.4191, "step": 13935 }, { "epoch": 2.24, "grad_norm": 11.843594551086426, "learning_rate": 3.1548330124992554e-06, "loss": 0.7315, "step": 13936 }, { "epoch": 2.24, "grad_norm": 14.305244445800781, "learning_rate": 3.1535661253007556e-06, "loss": 0.7177, "step": 13937 }, { "epoch": 2.24, "grad_norm": 12.420496940612793, "learning_rate": 3.1522994449065247e-06, "loss": 0.8195, "step": 13938 }, { "epoch": 2.24, "grad_norm": 15.007628440856934, "learning_rate": 3.1510329713548194e-06, "loss": 0.6681, "step": 13939 }, { "epoch": 2.24, "grad_norm": 16.391738891601562, "learning_rate": 3.149766704683901e-06, "loss": 0.5451, "step": 13940 }, { "epoch": 2.24, "grad_norm": 36.81021499633789, "learning_rate": 3.1485006449320143e-06, "loss": 0.5686, "step": 13941 }, { "epoch": 2.24, "grad_norm": 14.694609642028809, "learning_rate": 3.147234792137406e-06, "loss": 0.8791, "step": 13942 }, { "epoch": 2.24, "grad_norm": 12.921276092529297, "learning_rate": 3.1459691463383114e-06, "loss": 0.8455, "step": 13943 }, { "epoch": 2.24, "grad_norm": 12.733038902282715, "learning_rate": 3.1447037075729557e-06, "loss": 0.8007, "step": 13944 }, { "epoch": 2.24, "grad_norm": 10.902371406555176, "learning_rate": 3.1434384758795677e-06, "loss": 0.6681, "step": 13945 }, { "epoch": 2.25, "grad_norm": 13.241373062133789, "learning_rate": 3.1421734512963676e-06, "loss": 0.6768, "step": 13946 }, { "epoch": 2.25, "grad_norm": 11.359862327575684, "learning_rate": 3.1409086338615614e-06, "loss": 0.5979, "step": 13947 }, { "epoch": 2.25, "grad_norm": 11.297750473022461, "learning_rate": 3.1396440236133587e-06, "loss": 0.8231, "step": 13948 }, { "epoch": 2.25, "grad_norm": 13.604874610900879, "learning_rate": 3.1383796205899577e-06, "loss": 0.7974, "step": 13949 }, { "epoch": 2.25, "grad_norm": 14.73217487335205, "learning_rate": 3.1371154248295456e-06, "loss": 0.649, "step": 13950 }, { "epoch": 2.25, "grad_norm": 12.88476848602295, "learning_rate": 3.1358514363703172e-06, "loss": 0.7405, "step": 13951 }, { "epoch": 2.25, "grad_norm": 10.793289184570312, "learning_rate": 3.1345876552504463e-06, "loss": 0.7385, "step": 13952 }, { "epoch": 2.25, "grad_norm": 11.806509017944336, "learning_rate": 3.133324081508111e-06, "loss": 0.6869, "step": 13953 }, { "epoch": 2.25, "grad_norm": 11.097764015197754, "learning_rate": 3.1320607151814776e-06, "loss": 0.5287, "step": 13954 }, { "epoch": 2.25, "grad_norm": 13.591097831726074, "learning_rate": 3.1307975563087043e-06, "loss": 0.7216, "step": 13955 }, { "epoch": 2.25, "grad_norm": 10.718832015991211, "learning_rate": 3.129534604927953e-06, "loss": 0.6042, "step": 13956 }, { "epoch": 2.25, "grad_norm": 13.915060043334961, "learning_rate": 3.1282718610773676e-06, "loss": 0.7202, "step": 13957 }, { "epoch": 2.25, "grad_norm": 12.454606056213379, "learning_rate": 3.1270093247950885e-06, "loss": 0.6987, "step": 13958 }, { "epoch": 2.25, "grad_norm": 13.149246215820312, "learning_rate": 3.1257469961192586e-06, "loss": 0.7913, "step": 13959 }, { "epoch": 2.25, "grad_norm": 18.754169464111328, "learning_rate": 3.124484875088002e-06, "loss": 0.7127, "step": 13960 }, { "epoch": 2.25, "grad_norm": 12.356348037719727, "learning_rate": 3.123222961739448e-06, "loss": 0.705, "step": 13961 }, { "epoch": 2.25, "grad_norm": 13.081499099731445, "learning_rate": 3.1219612561117118e-06, "loss": 0.6083, "step": 13962 }, { "epoch": 2.25, "grad_norm": 14.674216270446777, "learning_rate": 3.120699758242901e-06, "loss": 0.5556, "step": 13963 }, { "epoch": 2.25, "grad_norm": 11.372075080871582, "learning_rate": 3.1194384681711264e-06, "loss": 0.6715, "step": 13964 }, { "epoch": 2.25, "grad_norm": 11.781878471374512, "learning_rate": 3.118177385934481e-06, "loss": 0.6435, "step": 13965 }, { "epoch": 2.25, "grad_norm": 12.958809852600098, "learning_rate": 3.11691651157106e-06, "loss": 0.8306, "step": 13966 }, { "epoch": 2.25, "grad_norm": 11.600931167602539, "learning_rate": 3.115655845118952e-06, "loss": 0.7939, "step": 13967 }, { "epoch": 2.25, "grad_norm": 12.188460350036621, "learning_rate": 3.1143953866162357e-06, "loss": 0.6844, "step": 13968 }, { "epoch": 2.25, "grad_norm": 11.956095695495605, "learning_rate": 3.113135136100981e-06, "loss": 0.6307, "step": 13969 }, { "epoch": 2.25, "grad_norm": 12.3541841506958, "learning_rate": 3.1118750936112596e-06, "loss": 0.8384, "step": 13970 }, { "epoch": 2.25, "grad_norm": 11.613493919372559, "learning_rate": 3.110615259185128e-06, "loss": 0.6814, "step": 13971 }, { "epoch": 2.25, "grad_norm": 11.754972457885742, "learning_rate": 3.1093556328606467e-06, "loss": 0.7902, "step": 13972 }, { "epoch": 2.25, "grad_norm": 13.27954387664795, "learning_rate": 3.108096214675862e-06, "loss": 0.6943, "step": 13973 }, { "epoch": 2.25, "grad_norm": 11.299904823303223, "learning_rate": 3.10683700466881e-06, "loss": 0.6849, "step": 13974 }, { "epoch": 2.25, "grad_norm": 14.302513122558594, "learning_rate": 3.1055780028775363e-06, "loss": 0.7336, "step": 13975 }, { "epoch": 2.25, "grad_norm": 11.607367515563965, "learning_rate": 3.104319209340062e-06, "loss": 0.5718, "step": 13976 }, { "epoch": 2.25, "grad_norm": 10.758736610412598, "learning_rate": 3.1030606240944174e-06, "loss": 0.8263, "step": 13977 }, { "epoch": 2.25, "grad_norm": 10.787579536437988, "learning_rate": 3.1018022471786167e-06, "loss": 0.6519, "step": 13978 }, { "epoch": 2.25, "grad_norm": 11.760787963867188, "learning_rate": 3.1005440786306662e-06, "loss": 0.7895, "step": 13979 }, { "epoch": 2.25, "grad_norm": 14.05795669555664, "learning_rate": 3.099286118488578e-06, "loss": 0.7263, "step": 13980 }, { "epoch": 2.25, "grad_norm": 13.795382499694824, "learning_rate": 3.0980283667903465e-06, "loss": 0.86, "step": 13981 }, { "epoch": 2.25, "grad_norm": 13.804909706115723, "learning_rate": 3.09677082357396e-06, "loss": 0.657, "step": 13982 }, { "epoch": 2.25, "grad_norm": 14.54361629486084, "learning_rate": 3.0955134888774106e-06, "loss": 0.7742, "step": 13983 }, { "epoch": 2.25, "grad_norm": 13.243157386779785, "learning_rate": 3.0942563627386712e-06, "loss": 0.754, "step": 13984 }, { "epoch": 2.25, "grad_norm": 9.699474334716797, "learning_rate": 3.0929994451957225e-06, "loss": 0.6421, "step": 13985 }, { "epoch": 2.25, "grad_norm": 11.646024703979492, "learning_rate": 3.0917427362865216e-06, "loss": 0.5009, "step": 13986 }, { "epoch": 2.25, "grad_norm": 10.433953285217285, "learning_rate": 3.090486236049034e-06, "loss": 0.5892, "step": 13987 }, { "epoch": 2.25, "grad_norm": 14.39709758758545, "learning_rate": 3.089229944521217e-06, "loss": 0.7539, "step": 13988 }, { "epoch": 2.25, "grad_norm": 13.927103042602539, "learning_rate": 3.0879738617410148e-06, "loss": 0.4977, "step": 13989 }, { "epoch": 2.25, "grad_norm": 12.759675025939941, "learning_rate": 3.086717987746366e-06, "loss": 0.7348, "step": 13990 }, { "epoch": 2.25, "grad_norm": 12.307424545288086, "learning_rate": 3.085462322575211e-06, "loss": 0.5464, "step": 13991 }, { "epoch": 2.25, "grad_norm": 13.631464958190918, "learning_rate": 3.0842068662654756e-06, "loss": 0.9409, "step": 13992 }, { "epoch": 2.25, "grad_norm": 13.63018798828125, "learning_rate": 3.08295161885508e-06, "loss": 0.7118, "step": 13993 }, { "epoch": 2.25, "grad_norm": 11.732832908630371, "learning_rate": 3.0816965803819466e-06, "loss": 0.6515, "step": 13994 }, { "epoch": 2.25, "grad_norm": 17.71990203857422, "learning_rate": 3.080441750883979e-06, "loss": 0.7048, "step": 13995 }, { "epoch": 2.25, "grad_norm": 12.679186820983887, "learning_rate": 3.079187130399085e-06, "loss": 0.7047, "step": 13996 }, { "epoch": 2.25, "grad_norm": 12.56080436706543, "learning_rate": 3.077932718965162e-06, "loss": 0.8046, "step": 13997 }, { "epoch": 2.25, "grad_norm": 10.158127784729004, "learning_rate": 3.076678516620095e-06, "loss": 0.8813, "step": 13998 }, { "epoch": 2.25, "grad_norm": 16.529617309570312, "learning_rate": 3.0754245234017765e-06, "loss": 0.4833, "step": 13999 }, { "epoch": 2.25, "grad_norm": 15.316454887390137, "learning_rate": 3.074170739348078e-06, "loss": 0.7274, "step": 14000 }, { "epoch": 2.25, "grad_norm": 11.37846851348877, "learning_rate": 3.072917164496877e-06, "loss": 0.5082, "step": 14001 }, { "epoch": 2.25, "grad_norm": 12.958210945129395, "learning_rate": 3.0716637988860375e-06, "loss": 0.5423, "step": 14002 }, { "epoch": 2.25, "grad_norm": 12.993581771850586, "learning_rate": 3.070410642553414e-06, "loss": 0.6847, "step": 14003 }, { "epoch": 2.25, "grad_norm": 11.633346557617188, "learning_rate": 3.0691576955368674e-06, "loss": 0.609, "step": 14004 }, { "epoch": 2.25, "grad_norm": 12.465727806091309, "learning_rate": 3.0679049578742405e-06, "loss": 0.8932, "step": 14005 }, { "epoch": 2.25, "grad_norm": 12.253046035766602, "learning_rate": 3.0666524296033673e-06, "loss": 0.715, "step": 14006 }, { "epoch": 2.25, "grad_norm": 11.478759765625, "learning_rate": 3.0654001107620947e-06, "loss": 0.7563, "step": 14007 }, { "epoch": 2.25, "grad_norm": 11.534449577331543, "learning_rate": 3.064148001388243e-06, "loss": 0.6587, "step": 14008 }, { "epoch": 2.26, "grad_norm": 13.787835121154785, "learning_rate": 3.0628961015196323e-06, "loss": 0.7666, "step": 14009 }, { "epoch": 2.26, "grad_norm": 11.621779441833496, "learning_rate": 3.0616444111940822e-06, "loss": 0.6248, "step": 14010 }, { "epoch": 2.26, "grad_norm": 12.948789596557617, "learning_rate": 3.0603929304493996e-06, "loss": 0.764, "step": 14011 }, { "epoch": 2.26, "grad_norm": 10.859140396118164, "learning_rate": 3.0591416593233837e-06, "loss": 0.7647, "step": 14012 }, { "epoch": 2.26, "grad_norm": 13.39992904663086, "learning_rate": 3.0578905978538355e-06, "loss": 0.5618, "step": 14013 }, { "epoch": 2.26, "grad_norm": 11.400094985961914, "learning_rate": 3.056639746078539e-06, "loss": 0.8609, "step": 14014 }, { "epoch": 2.26, "grad_norm": 12.283214569091797, "learning_rate": 3.0553891040352847e-06, "loss": 0.6869, "step": 14015 }, { "epoch": 2.26, "grad_norm": 17.539793014526367, "learning_rate": 3.054138671761846e-06, "loss": 0.8099, "step": 14016 }, { "epoch": 2.26, "grad_norm": 11.753694534301758, "learning_rate": 3.052888449295991e-06, "loss": 0.5389, "step": 14017 }, { "epoch": 2.26, "grad_norm": 11.748899459838867, "learning_rate": 3.0516384366754883e-06, "loss": 0.5725, "step": 14018 }, { "epoch": 2.26, "grad_norm": 14.485027313232422, "learning_rate": 3.050388633938093e-06, "loss": 0.8543, "step": 14019 }, { "epoch": 2.26, "grad_norm": 11.769094467163086, "learning_rate": 3.049139041121559e-06, "loss": 0.7835, "step": 14020 }, { "epoch": 2.26, "grad_norm": 12.988518714904785, "learning_rate": 3.047889658263633e-06, "loss": 0.8444, "step": 14021 }, { "epoch": 2.26, "grad_norm": 12.82972526550293, "learning_rate": 3.046640485402047e-06, "loss": 0.8793, "step": 14022 }, { "epoch": 2.26, "grad_norm": 11.597343444824219, "learning_rate": 3.0453915225745425e-06, "loss": 0.6374, "step": 14023 }, { "epoch": 2.26, "grad_norm": 16.12457275390625, "learning_rate": 3.044142769818842e-06, "loss": 0.6753, "step": 14024 }, { "epoch": 2.26, "grad_norm": 14.319079399108887, "learning_rate": 3.0428942271726614e-06, "loss": 0.5519, "step": 14025 }, { "epoch": 2.26, "grad_norm": 14.172259330749512, "learning_rate": 3.0416458946737195e-06, "loss": 0.7785, "step": 14026 }, { "epoch": 2.26, "grad_norm": 12.063751220703125, "learning_rate": 3.0403977723597244e-06, "loss": 0.6579, "step": 14027 }, { "epoch": 2.26, "grad_norm": 10.435256004333496, "learning_rate": 3.039149860268372e-06, "loss": 0.5245, "step": 14028 }, { "epoch": 2.26, "grad_norm": 12.082159042358398, "learning_rate": 3.0379021584373627e-06, "loss": 0.6158, "step": 14029 }, { "epoch": 2.26, "grad_norm": 15.249298095703125, "learning_rate": 3.0366546669043794e-06, "loss": 0.7247, "step": 14030 }, { "epoch": 2.26, "grad_norm": 12.353586196899414, "learning_rate": 3.0354073857071097e-06, "loss": 0.7146, "step": 14031 }, { "epoch": 2.26, "grad_norm": 13.636646270751953, "learning_rate": 3.0341603148832253e-06, "loss": 0.5994, "step": 14032 }, { "epoch": 2.26, "grad_norm": 13.61181640625, "learning_rate": 3.032913454470393e-06, "loss": 0.6944, "step": 14033 }, { "epoch": 2.26, "grad_norm": 11.504694938659668, "learning_rate": 3.0316668045062826e-06, "loss": 0.4906, "step": 14034 }, { "epoch": 2.26, "grad_norm": 13.950690269470215, "learning_rate": 3.0304203650285458e-06, "loss": 0.5975, "step": 14035 }, { "epoch": 2.26, "grad_norm": 11.450212478637695, "learning_rate": 3.0291741360748306e-06, "loss": 0.6245, "step": 14036 }, { "epoch": 2.26, "grad_norm": 13.868668556213379, "learning_rate": 3.027928117682787e-06, "loss": 0.7758, "step": 14037 }, { "epoch": 2.26, "grad_norm": 13.40746021270752, "learning_rate": 3.0266823098900456e-06, "loss": 0.8031, "step": 14038 }, { "epoch": 2.26, "grad_norm": 12.583304405212402, "learning_rate": 3.0254367127342442e-06, "loss": 0.755, "step": 14039 }, { "epoch": 2.26, "grad_norm": 10.592495918273926, "learning_rate": 3.0241913262530044e-06, "loss": 0.752, "step": 14040 }, { "epoch": 2.26, "grad_norm": 11.93242359161377, "learning_rate": 3.0229461504839406e-06, "loss": 0.7404, "step": 14041 }, { "epoch": 2.26, "grad_norm": 15.682474136352539, "learning_rate": 3.0217011854646714e-06, "loss": 0.7284, "step": 14042 }, { "epoch": 2.26, "grad_norm": 13.455151557922363, "learning_rate": 3.0204564312327966e-06, "loss": 0.7142, "step": 14043 }, { "epoch": 2.26, "grad_norm": 12.18856143951416, "learning_rate": 3.0192118878259213e-06, "loss": 0.7306, "step": 14044 }, { "epoch": 2.26, "grad_norm": 13.95029354095459, "learning_rate": 3.017967555281636e-06, "loss": 0.9264, "step": 14045 }, { "epoch": 2.26, "grad_norm": 14.050076484680176, "learning_rate": 3.0167234336375228e-06, "loss": 0.7775, "step": 14046 }, { "epoch": 2.26, "grad_norm": 11.451024055480957, "learning_rate": 3.015479522931165e-06, "loss": 0.7267, "step": 14047 }, { "epoch": 2.26, "grad_norm": 12.524242401123047, "learning_rate": 3.0142358232001402e-06, "loss": 0.8913, "step": 14048 }, { "epoch": 2.26, "grad_norm": 13.163844108581543, "learning_rate": 3.012992334482009e-06, "loss": 0.6962, "step": 14049 }, { "epoch": 2.26, "grad_norm": 13.1419095993042, "learning_rate": 3.011749056814339e-06, "loss": 0.7969, "step": 14050 }, { "epoch": 2.26, "grad_norm": 11.204214096069336, "learning_rate": 3.010505990234682e-06, "loss": 0.5692, "step": 14051 }, { "epoch": 2.26, "grad_norm": 13.97629451751709, "learning_rate": 3.0092631347805824e-06, "loss": 0.8048, "step": 14052 }, { "epoch": 2.26, "grad_norm": 15.083244323730469, "learning_rate": 3.008020490489587e-06, "loss": 0.6847, "step": 14053 }, { "epoch": 2.26, "grad_norm": 14.003612518310547, "learning_rate": 3.006778057399228e-06, "loss": 0.9245, "step": 14054 }, { "epoch": 2.26, "grad_norm": 11.322100639343262, "learning_rate": 3.005535835547039e-06, "loss": 0.7769, "step": 14055 }, { "epoch": 2.26, "grad_norm": 12.724455833435059, "learning_rate": 3.00429382497054e-06, "loss": 0.6628, "step": 14056 }, { "epoch": 2.26, "grad_norm": 13.190323829650879, "learning_rate": 3.0030520257072437e-06, "loss": 0.6696, "step": 14057 }, { "epoch": 2.26, "grad_norm": 11.35865592956543, "learning_rate": 3.0018104377946667e-06, "loss": 0.5373, "step": 14058 }, { "epoch": 2.26, "grad_norm": 12.081011772155762, "learning_rate": 3.00056906127031e-06, "loss": 0.7487, "step": 14059 }, { "epoch": 2.26, "grad_norm": 11.534547805786133, "learning_rate": 2.9993278961716667e-06, "loss": 0.8331, "step": 14060 }, { "epoch": 2.26, "grad_norm": 12.791847229003906, "learning_rate": 2.9980869425362337e-06, "loss": 0.8527, "step": 14061 }, { "epoch": 2.26, "grad_norm": 12.803717613220215, "learning_rate": 2.99684620040149e-06, "loss": 0.6511, "step": 14062 }, { "epoch": 2.26, "grad_norm": 10.393051147460938, "learning_rate": 2.9956056698049183e-06, "loss": 0.6552, "step": 14063 }, { "epoch": 2.26, "grad_norm": 10.296919822692871, "learning_rate": 2.99436535078399e-06, "loss": 0.8184, "step": 14064 }, { "epoch": 2.26, "grad_norm": 12.851297378540039, "learning_rate": 2.9931252433761648e-06, "loss": 0.8285, "step": 14065 }, { "epoch": 2.26, "grad_norm": 13.095675468444824, "learning_rate": 2.991885347618909e-06, "loss": 0.7211, "step": 14066 }, { "epoch": 2.26, "grad_norm": 12.772725105285645, "learning_rate": 2.9906456635496684e-06, "loss": 0.6155, "step": 14067 }, { "epoch": 2.26, "grad_norm": 12.58869743347168, "learning_rate": 2.9894061912058915e-06, "loss": 0.5751, "step": 14068 }, { "epoch": 2.26, "grad_norm": 11.015572547912598, "learning_rate": 2.988166930625023e-06, "loss": 0.6084, "step": 14069 }, { "epoch": 2.26, "grad_norm": 10.405526161193848, "learning_rate": 2.98692788184449e-06, "loss": 0.8862, "step": 14070 }, { "epoch": 2.27, "grad_norm": 10.668440818786621, "learning_rate": 2.9856890449017184e-06, "loss": 0.7187, "step": 14071 }, { "epoch": 2.27, "grad_norm": 12.780158042907715, "learning_rate": 2.984450419834135e-06, "loss": 0.6677, "step": 14072 }, { "epoch": 2.27, "grad_norm": 11.940274238586426, "learning_rate": 2.983212006679147e-06, "loss": 0.7301, "step": 14073 }, { "epoch": 2.27, "grad_norm": 14.301942825317383, "learning_rate": 2.9819738054741688e-06, "loss": 0.7733, "step": 14074 }, { "epoch": 2.27, "grad_norm": 13.900796890258789, "learning_rate": 2.9807358162565968e-06, "loss": 0.6807, "step": 14075 }, { "epoch": 2.27, "grad_norm": 13.895931243896484, "learning_rate": 2.9794980390638263e-06, "loss": 0.8091, "step": 14076 }, { "epoch": 2.27, "grad_norm": 11.55366325378418, "learning_rate": 2.9782604739332478e-06, "loss": 0.7781, "step": 14077 }, { "epoch": 2.27, "grad_norm": 13.50900650024414, "learning_rate": 2.9770231209022405e-06, "loss": 0.5801, "step": 14078 }, { "epoch": 2.27, "grad_norm": 13.300097465515137, "learning_rate": 2.9757859800081835e-06, "loss": 0.7149, "step": 14079 }, { "epoch": 2.27, "grad_norm": 12.339354515075684, "learning_rate": 2.9745490512884454e-06, "loss": 0.5398, "step": 14080 }, { "epoch": 2.27, "grad_norm": 12.0985107421875, "learning_rate": 2.973312334780385e-06, "loss": 0.6801, "step": 14081 }, { "epoch": 2.27, "grad_norm": 13.365615844726562, "learning_rate": 2.972075830521365e-06, "loss": 0.5545, "step": 14082 }, { "epoch": 2.27, "grad_norm": 12.266376495361328, "learning_rate": 2.970839538548731e-06, "loss": 0.7037, "step": 14083 }, { "epoch": 2.27, "grad_norm": 11.981812477111816, "learning_rate": 2.9696034588998247e-06, "loss": 0.6159, "step": 14084 }, { "epoch": 2.27, "grad_norm": 12.299262046813965, "learning_rate": 2.9683675916119903e-06, "loss": 0.569, "step": 14085 }, { "epoch": 2.27, "grad_norm": 12.232950210571289, "learning_rate": 2.9671319367225506e-06, "loss": 0.6924, "step": 14086 }, { "epoch": 2.27, "grad_norm": 12.805572509765625, "learning_rate": 2.965896494268834e-06, "loss": 0.7382, "step": 14087 }, { "epoch": 2.27, "grad_norm": 12.258649826049805, "learning_rate": 2.964661264288162e-06, "loss": 0.7219, "step": 14088 }, { "epoch": 2.27, "grad_norm": 11.969663619995117, "learning_rate": 2.963426246817842e-06, "loss": 0.8435, "step": 14089 }, { "epoch": 2.27, "grad_norm": 10.5441312789917, "learning_rate": 2.962191441895177e-06, "loss": 0.7895, "step": 14090 }, { "epoch": 2.27, "grad_norm": 11.194040298461914, "learning_rate": 2.9609568495574703e-06, "loss": 0.7149, "step": 14091 }, { "epoch": 2.27, "grad_norm": 16.482450485229492, "learning_rate": 2.959722469842009e-06, "loss": 0.8567, "step": 14092 }, { "epoch": 2.27, "grad_norm": 17.441381454467773, "learning_rate": 2.9584883027860854e-06, "loss": 0.7811, "step": 14093 }, { "epoch": 2.27, "grad_norm": 10.892592430114746, "learning_rate": 2.9572543484269766e-06, "loss": 0.7945, "step": 14094 }, { "epoch": 2.27, "grad_norm": 10.428545951843262, "learning_rate": 2.95602060680195e-06, "loss": 0.7355, "step": 14095 }, { "epoch": 2.27, "grad_norm": 13.40952205657959, "learning_rate": 2.95478707794828e-06, "loss": 0.8476, "step": 14096 }, { "epoch": 2.27, "grad_norm": 11.072880744934082, "learning_rate": 2.9535537619032207e-06, "loss": 0.5553, "step": 14097 }, { "epoch": 2.27, "grad_norm": 18.457250595092773, "learning_rate": 2.952320658704032e-06, "loss": 0.8395, "step": 14098 }, { "epoch": 2.27, "grad_norm": 13.11003589630127, "learning_rate": 2.951087768387957e-06, "loss": 0.9493, "step": 14099 }, { "epoch": 2.27, "grad_norm": 11.50267505645752, "learning_rate": 2.949855090992234e-06, "loss": 0.6764, "step": 14100 }, { "epoch": 2.27, "grad_norm": 11.790090560913086, "learning_rate": 2.948622626554104e-06, "loss": 0.8721, "step": 14101 }, { "epoch": 2.27, "grad_norm": 15.509903907775879, "learning_rate": 2.9473903751107913e-06, "loss": 0.6108, "step": 14102 }, { "epoch": 2.27, "grad_norm": 13.300724029541016, "learning_rate": 2.946158336699515e-06, "loss": 0.7867, "step": 14103 }, { "epoch": 2.27, "grad_norm": 12.283202171325684, "learning_rate": 2.9449265113574967e-06, "loss": 0.7402, "step": 14104 }, { "epoch": 2.27, "grad_norm": 11.451959609985352, "learning_rate": 2.9436948991219373e-06, "loss": 0.5756, "step": 14105 }, { "epoch": 2.27, "grad_norm": 12.184850692749023, "learning_rate": 2.9424635000300463e-06, "loss": 0.6794, "step": 14106 }, { "epoch": 2.27, "grad_norm": 11.460447311401367, "learning_rate": 2.9412323141190124e-06, "loss": 0.5122, "step": 14107 }, { "epoch": 2.27, "grad_norm": 11.244790077209473, "learning_rate": 2.94000134142603e-06, "loss": 0.6013, "step": 14108 }, { "epoch": 2.27, "grad_norm": 11.590060234069824, "learning_rate": 2.9387705819882838e-06, "loss": 0.7089, "step": 14109 }, { "epoch": 2.27, "grad_norm": 11.88374137878418, "learning_rate": 2.937540035842947e-06, "loss": 0.5871, "step": 14110 }, { "epoch": 2.27, "grad_norm": 14.62315559387207, "learning_rate": 2.936309703027187e-06, "loss": 0.7872, "step": 14111 }, { "epoch": 2.27, "grad_norm": 12.769976615905762, "learning_rate": 2.9350795835781743e-06, "loss": 0.6467, "step": 14112 }, { "epoch": 2.27, "grad_norm": 13.602313041687012, "learning_rate": 2.9338496775330614e-06, "loss": 0.7457, "step": 14113 }, { "epoch": 2.27, "grad_norm": 11.965944290161133, "learning_rate": 2.932619984928996e-06, "loss": 0.7528, "step": 14114 }, { "epoch": 2.27, "grad_norm": 11.375523567199707, "learning_rate": 2.93139050580313e-06, "loss": 0.6804, "step": 14115 }, { "epoch": 2.27, "grad_norm": 9.910517692565918, "learning_rate": 2.930161240192594e-06, "loss": 0.5906, "step": 14116 }, { "epoch": 2.27, "grad_norm": 13.527094841003418, "learning_rate": 2.9289321881345257e-06, "loss": 0.5979, "step": 14117 }, { "epoch": 2.27, "grad_norm": 13.192401885986328, "learning_rate": 2.9277033496660467e-06, "loss": 0.7527, "step": 14118 }, { "epoch": 2.27, "grad_norm": 12.02705192565918, "learning_rate": 2.9264747248242732e-06, "loss": 0.4892, "step": 14119 }, { "epoch": 2.27, "grad_norm": 12.277974128723145, "learning_rate": 2.9252463136463214e-06, "loss": 0.7051, "step": 14120 }, { "epoch": 2.27, "grad_norm": 11.21389389038086, "learning_rate": 2.924018116169294e-06, "loss": 0.8041, "step": 14121 }, { "epoch": 2.27, "grad_norm": 12.674562454223633, "learning_rate": 2.922790132430293e-06, "loss": 0.7189, "step": 14122 }, { "epoch": 2.27, "grad_norm": 11.023805618286133, "learning_rate": 2.9215623624664104e-06, "loss": 0.8462, "step": 14123 }, { "epoch": 2.27, "grad_norm": 11.150649070739746, "learning_rate": 2.920334806314726e-06, "loss": 0.6739, "step": 14124 }, { "epoch": 2.27, "grad_norm": 13.277143478393555, "learning_rate": 2.9191074640123295e-06, "loss": 0.9028, "step": 14125 }, { "epoch": 2.27, "grad_norm": 13.42502212524414, "learning_rate": 2.917880335596289e-06, "loss": 0.4916, "step": 14126 }, { "epoch": 2.27, "grad_norm": 12.760501861572266, "learning_rate": 2.916653421103669e-06, "loss": 0.9111, "step": 14127 }, { "epoch": 2.27, "grad_norm": 12.493404388427734, "learning_rate": 2.9154267205715325e-06, "loss": 0.6658, "step": 14128 }, { "epoch": 2.27, "grad_norm": 12.007773399353027, "learning_rate": 2.9142002340369368e-06, "loss": 0.681, "step": 14129 }, { "epoch": 2.27, "grad_norm": 12.69605827331543, "learning_rate": 2.9129739615369225e-06, "loss": 0.8095, "step": 14130 }, { "epoch": 2.27, "grad_norm": 12.502954483032227, "learning_rate": 2.9117479031085373e-06, "loss": 0.7478, "step": 14131 }, { "epoch": 2.27, "grad_norm": 11.520373344421387, "learning_rate": 2.9105220587888084e-06, "loss": 0.7334, "step": 14132 }, { "epoch": 2.28, "grad_norm": 12.106308937072754, "learning_rate": 2.909296428614772e-06, "loss": 0.8325, "step": 14133 }, { "epoch": 2.28, "grad_norm": 11.256649017333984, "learning_rate": 2.9080710126234447e-06, "loss": 0.4853, "step": 14134 }, { "epoch": 2.28, "grad_norm": 12.421395301818848, "learning_rate": 2.9068458108518392e-06, "loss": 0.6153, "step": 14135 }, { "epoch": 2.28, "grad_norm": 13.015289306640625, "learning_rate": 2.905620823336971e-06, "loss": 0.6752, "step": 14136 }, { "epoch": 2.28, "grad_norm": 15.551491737365723, "learning_rate": 2.9043960501158376e-06, "loss": 0.8232, "step": 14137 }, { "epoch": 2.28, "grad_norm": 15.538625717163086, "learning_rate": 2.9031714912254328e-06, "loss": 0.7306, "step": 14138 }, { "epoch": 2.28, "grad_norm": 16.030004501342773, "learning_rate": 2.9019471467027517e-06, "loss": 0.7615, "step": 14139 }, { "epoch": 2.28, "grad_norm": 13.018270492553711, "learning_rate": 2.90072301658477e-06, "loss": 0.6761, "step": 14140 }, { "epoch": 2.28, "grad_norm": 13.649727821350098, "learning_rate": 2.8994991009084717e-06, "loss": 0.9178, "step": 14141 }, { "epoch": 2.28, "grad_norm": 12.839516639709473, "learning_rate": 2.8982753997108225e-06, "loss": 0.7504, "step": 14142 }, { "epoch": 2.28, "grad_norm": 11.733331680297852, "learning_rate": 2.8970519130287823e-06, "loss": 0.755, "step": 14143 }, { "epoch": 2.28, "grad_norm": 12.905942916870117, "learning_rate": 2.895828640899314e-06, "loss": 0.6746, "step": 14144 }, { "epoch": 2.28, "grad_norm": 14.998985290527344, "learning_rate": 2.8946055833593633e-06, "loss": 0.5497, "step": 14145 }, { "epoch": 2.28, "grad_norm": 12.225640296936035, "learning_rate": 2.893382740445879e-06, "loss": 0.5682, "step": 14146 }, { "epoch": 2.28, "grad_norm": 17.37342643737793, "learning_rate": 2.8921601121957944e-06, "loss": 0.7098, "step": 14147 }, { "epoch": 2.28, "grad_norm": 13.722911834716797, "learning_rate": 2.890937698646039e-06, "loss": 0.7289, "step": 14148 }, { "epoch": 2.28, "grad_norm": 12.68720531463623, "learning_rate": 2.8897154998335398e-06, "loss": 0.6074, "step": 14149 }, { "epoch": 2.28, "grad_norm": 13.826258659362793, "learning_rate": 2.8884935157952177e-06, "loss": 0.5643, "step": 14150 }, { "epoch": 2.28, "grad_norm": 13.52357292175293, "learning_rate": 2.8872717465679768e-06, "loss": 0.549, "step": 14151 }, { "epoch": 2.28, "grad_norm": 12.355257987976074, "learning_rate": 2.8860501921887295e-06, "loss": 0.7882, "step": 14152 }, { "epoch": 2.28, "grad_norm": 13.133645057678223, "learning_rate": 2.8848288526943713e-06, "loss": 0.7408, "step": 14153 }, { "epoch": 2.28, "grad_norm": 10.915849685668945, "learning_rate": 2.883607728121791e-06, "loss": 0.8331, "step": 14154 }, { "epoch": 2.28, "grad_norm": 11.552910804748535, "learning_rate": 2.8823868185078796e-06, "loss": 0.7641, "step": 14155 }, { "epoch": 2.28, "grad_norm": 12.836934089660645, "learning_rate": 2.8811661238895126e-06, "loss": 0.8701, "step": 14156 }, { "epoch": 2.28, "grad_norm": 12.45274829864502, "learning_rate": 2.8799456443035602e-06, "loss": 0.8326, "step": 14157 }, { "epoch": 2.28, "grad_norm": 10.387954711914062, "learning_rate": 2.8787253797868952e-06, "loss": 0.5757, "step": 14158 }, { "epoch": 2.28, "grad_norm": 13.749917030334473, "learning_rate": 2.877505330376369e-06, "loss": 0.89, "step": 14159 }, { "epoch": 2.28, "grad_norm": 10.59924030303955, "learning_rate": 2.8762854961088414e-06, "loss": 0.7664, "step": 14160 }, { "epoch": 2.28, "grad_norm": 10.941755294799805, "learning_rate": 2.875065877021157e-06, "loss": 0.538, "step": 14161 }, { "epoch": 2.28, "grad_norm": 9.737736701965332, "learning_rate": 2.8738464731501512e-06, "loss": 0.4755, "step": 14162 }, { "epoch": 2.28, "grad_norm": 13.26190185546875, "learning_rate": 2.8726272845326653e-06, "loss": 0.8205, "step": 14163 }, { "epoch": 2.28, "grad_norm": 11.951502799987793, "learning_rate": 2.871408311205518e-06, "loss": 0.5384, "step": 14164 }, { "epoch": 2.28, "grad_norm": 12.163859367370605, "learning_rate": 2.870189553205538e-06, "loss": 0.6503, "step": 14165 }, { "epoch": 2.28, "grad_norm": 11.383238792419434, "learning_rate": 2.8689710105695344e-06, "loss": 0.8285, "step": 14166 }, { "epoch": 2.28, "grad_norm": 11.582521438598633, "learning_rate": 2.8677526833343115e-06, "loss": 0.6561, "step": 14167 }, { "epoch": 2.28, "grad_norm": 13.92456340789795, "learning_rate": 2.8665345715366756e-06, "loss": 0.8083, "step": 14168 }, { "epoch": 2.28, "grad_norm": 12.54181957244873, "learning_rate": 2.865316675213422e-06, "loss": 0.665, "step": 14169 }, { "epoch": 2.28, "grad_norm": 11.260804176330566, "learning_rate": 2.864098994401334e-06, "loss": 0.7071, "step": 14170 }, { "epoch": 2.28, "grad_norm": 16.422832489013672, "learning_rate": 2.862881529137198e-06, "loss": 0.7526, "step": 14171 }, { "epoch": 2.28, "grad_norm": 11.123437881469727, "learning_rate": 2.8616642794577863e-06, "loss": 0.7625, "step": 14172 }, { "epoch": 2.28, "grad_norm": 12.715960502624512, "learning_rate": 2.8604472453998643e-06, "loss": 0.8683, "step": 14173 }, { "epoch": 2.28, "grad_norm": 10.641883850097656, "learning_rate": 2.8592304270002004e-06, "loss": 0.7111, "step": 14174 }, { "epoch": 2.28, "grad_norm": 13.112040519714355, "learning_rate": 2.8580138242955422e-06, "loss": 0.8409, "step": 14175 }, { "epoch": 2.28, "grad_norm": 11.227264404296875, "learning_rate": 2.856797437322647e-06, "loss": 0.7472, "step": 14176 }, { "epoch": 2.28, "grad_norm": 9.972023963928223, "learning_rate": 2.8555812661182524e-06, "loss": 0.599, "step": 14177 }, { "epoch": 2.28, "grad_norm": 10.69520378112793, "learning_rate": 2.8543653107190927e-06, "loss": 0.6354, "step": 14178 }, { "epoch": 2.28, "grad_norm": 14.679167747497559, "learning_rate": 2.853149571161901e-06, "loss": 0.8336, "step": 14179 }, { "epoch": 2.28, "grad_norm": 15.009227752685547, "learning_rate": 2.851934047483399e-06, "loss": 0.6783, "step": 14180 }, { "epoch": 2.28, "grad_norm": 14.24785041809082, "learning_rate": 2.8507187397202994e-06, "loss": 0.6315, "step": 14181 }, { "epoch": 2.28, "grad_norm": 12.895377159118652, "learning_rate": 2.8495036479093176e-06, "loss": 0.9785, "step": 14182 }, { "epoch": 2.28, "grad_norm": 13.836038589477539, "learning_rate": 2.848288772087151e-06, "loss": 0.5246, "step": 14183 }, { "epoch": 2.28, "grad_norm": 12.698700904846191, "learning_rate": 2.8470741122905034e-06, "loss": 0.6361, "step": 14184 }, { "epoch": 2.28, "grad_norm": 12.00793743133545, "learning_rate": 2.8458596685560593e-06, "loss": 0.6851, "step": 14185 }, { "epoch": 2.28, "grad_norm": 13.58007526397705, "learning_rate": 2.844645440920503e-06, "loss": 0.5981, "step": 14186 }, { "epoch": 2.28, "grad_norm": 12.451151847839355, "learning_rate": 2.8434314294205135e-06, "loss": 0.8207, "step": 14187 }, { "epoch": 2.28, "grad_norm": 12.200835227966309, "learning_rate": 2.842217634092759e-06, "loss": 0.6084, "step": 14188 }, { "epoch": 2.28, "grad_norm": 12.97877025604248, "learning_rate": 2.841004054973905e-06, "loss": 0.7003, "step": 14189 }, { "epoch": 2.28, "grad_norm": 11.85130786895752, "learning_rate": 2.8397906921006114e-06, "loss": 0.7303, "step": 14190 }, { "epoch": 2.28, "grad_norm": 9.948705673217773, "learning_rate": 2.8385775455095267e-06, "loss": 0.4219, "step": 14191 }, { "epoch": 2.28, "grad_norm": 12.032998085021973, "learning_rate": 2.8373646152372935e-06, "loss": 0.6147, "step": 14192 }, { "epoch": 2.28, "grad_norm": 11.377031326293945, "learning_rate": 2.836151901320554e-06, "loss": 0.693, "step": 14193 }, { "epoch": 2.28, "grad_norm": 12.855668067932129, "learning_rate": 2.8349394037959353e-06, "loss": 0.4722, "step": 14194 }, { "epoch": 2.29, "grad_norm": 11.922121047973633, "learning_rate": 2.8337271227000664e-06, "loss": 0.6305, "step": 14195 }, { "epoch": 2.29, "grad_norm": 12.131421089172363, "learning_rate": 2.832515058069565e-06, "loss": 0.5087, "step": 14196 }, { "epoch": 2.29, "grad_norm": 15.0011568069458, "learning_rate": 2.8313032099410367e-06, "loss": 0.6622, "step": 14197 }, { "epoch": 2.29, "grad_norm": 12.13260269165039, "learning_rate": 2.830091578351095e-06, "loss": 0.7554, "step": 14198 }, { "epoch": 2.29, "grad_norm": 15.66455364227295, "learning_rate": 2.828880163336333e-06, "loss": 0.7951, "step": 14199 }, { "epoch": 2.29, "grad_norm": 11.25623607635498, "learning_rate": 2.8276689649333466e-06, "loss": 0.8859, "step": 14200 }, { "epoch": 2.29, "grad_norm": 12.633910179138184, "learning_rate": 2.8264579831787207e-06, "loss": 0.6428, "step": 14201 }, { "epoch": 2.29, "grad_norm": 12.254424095153809, "learning_rate": 2.8252472181090305e-06, "loss": 0.6104, "step": 14202 }, { "epoch": 2.29, "grad_norm": 13.450672149658203, "learning_rate": 2.8240366697608545e-06, "loss": 0.698, "step": 14203 }, { "epoch": 2.29, "grad_norm": 16.6152400970459, "learning_rate": 2.8228263381707554e-06, "loss": 0.8162, "step": 14204 }, { "epoch": 2.29, "grad_norm": 11.530540466308594, "learning_rate": 2.8216162233752897e-06, "loss": 0.6495, "step": 14205 }, { "epoch": 2.29, "grad_norm": 12.607344627380371, "learning_rate": 2.820406325411016e-06, "loss": 0.8318, "step": 14206 }, { "epoch": 2.29, "grad_norm": 11.036916732788086, "learning_rate": 2.819196644314476e-06, "loss": 0.8321, "step": 14207 }, { "epoch": 2.29, "grad_norm": 12.976076126098633, "learning_rate": 2.8179871801222138e-06, "loss": 0.7979, "step": 14208 }, { "epoch": 2.29, "grad_norm": 11.184213638305664, "learning_rate": 2.8167779328707578e-06, "loss": 0.7014, "step": 14209 }, { "epoch": 2.29, "grad_norm": 15.153360366821289, "learning_rate": 2.815568902596637e-06, "loss": 0.6944, "step": 14210 }, { "epoch": 2.29, "grad_norm": 11.448412895202637, "learning_rate": 2.8143600893363755e-06, "loss": 0.8996, "step": 14211 }, { "epoch": 2.29, "grad_norm": 14.29887580871582, "learning_rate": 2.8131514931264825e-06, "loss": 0.7615, "step": 14212 }, { "epoch": 2.29, "grad_norm": 13.333821296691895, "learning_rate": 2.8119431140034636e-06, "loss": 0.6806, "step": 14213 }, { "epoch": 2.29, "grad_norm": 13.058589935302734, "learning_rate": 2.810734952003824e-06, "loss": 0.7955, "step": 14214 }, { "epoch": 2.29, "grad_norm": 10.67871379852295, "learning_rate": 2.8095270071640544e-06, "loss": 0.6987, "step": 14215 }, { "epoch": 2.29, "grad_norm": 12.99176025390625, "learning_rate": 2.8083192795206403e-06, "loss": 0.7486, "step": 14216 }, { "epoch": 2.29, "grad_norm": 13.068930625915527, "learning_rate": 2.807111769110067e-06, "loss": 0.718, "step": 14217 }, { "epoch": 2.29, "grad_norm": 10.516531944274902, "learning_rate": 2.805904475968805e-06, "loss": 0.4701, "step": 14218 }, { "epoch": 2.29, "grad_norm": 10.903457641601562, "learning_rate": 2.804697400133325e-06, "loss": 0.6192, "step": 14219 }, { "epoch": 2.29, "grad_norm": 14.752325057983398, "learning_rate": 2.803490541640087e-06, "loss": 0.6836, "step": 14220 }, { "epoch": 2.29, "grad_norm": 12.195388793945312, "learning_rate": 2.802283900525543e-06, "loss": 0.7193, "step": 14221 }, { "epoch": 2.29, "grad_norm": 15.638937950134277, "learning_rate": 2.801077476826145e-06, "loss": 0.5846, "step": 14222 }, { "epoch": 2.29, "grad_norm": 10.83126163482666, "learning_rate": 2.7998712705783302e-06, "loss": 0.6604, "step": 14223 }, { "epoch": 2.29, "grad_norm": 13.402342796325684, "learning_rate": 2.798665281818539e-06, "loss": 0.7411, "step": 14224 }, { "epoch": 2.29, "grad_norm": 8.928559303283691, "learning_rate": 2.797459510583196e-06, "loss": 0.6032, "step": 14225 }, { "epoch": 2.29, "grad_norm": 12.279151916503906, "learning_rate": 2.79625395690872e-06, "loss": 0.6025, "step": 14226 }, { "epoch": 2.29, "grad_norm": 11.482906341552734, "learning_rate": 2.795048620831533e-06, "loss": 0.7405, "step": 14227 }, { "epoch": 2.29, "grad_norm": 12.700695991516113, "learning_rate": 2.7938435023880396e-06, "loss": 0.758, "step": 14228 }, { "epoch": 2.29, "grad_norm": 14.343259811401367, "learning_rate": 2.792638601614636e-06, "loss": 0.8733, "step": 14229 }, { "epoch": 2.29, "grad_norm": 9.109003067016602, "learning_rate": 2.791433918547729e-06, "loss": 0.5611, "step": 14230 }, { "epoch": 2.29, "grad_norm": 13.590248107910156, "learning_rate": 2.7902294532237027e-06, "loss": 0.6438, "step": 14231 }, { "epoch": 2.29, "grad_norm": 11.82324504852295, "learning_rate": 2.7890252056789356e-06, "loss": 0.5572, "step": 14232 }, { "epoch": 2.29, "grad_norm": 11.548332214355469, "learning_rate": 2.78782117594981e-06, "loss": 0.6474, "step": 14233 }, { "epoch": 2.29, "grad_norm": 12.258543014526367, "learning_rate": 2.7866173640726903e-06, "loss": 0.826, "step": 14234 }, { "epoch": 2.29, "grad_norm": 10.64721965789795, "learning_rate": 2.785413770083938e-06, "loss": 0.5819, "step": 14235 }, { "epoch": 2.29, "grad_norm": 13.277681350708008, "learning_rate": 2.784210394019914e-06, "loss": 0.8258, "step": 14236 }, { "epoch": 2.29, "grad_norm": 14.267335891723633, "learning_rate": 2.7830072359169614e-06, "loss": 0.6899, "step": 14237 }, { "epoch": 2.29, "grad_norm": 11.164027214050293, "learning_rate": 2.78180429581143e-06, "loss": 0.7545, "step": 14238 }, { "epoch": 2.29, "grad_norm": 10.115851402282715, "learning_rate": 2.780601573739652e-06, "loss": 0.5243, "step": 14239 }, { "epoch": 2.29, "grad_norm": 14.690295219421387, "learning_rate": 2.7793990697379537e-06, "loss": 0.6956, "step": 14240 }, { "epoch": 2.29, "grad_norm": 12.025527954101562, "learning_rate": 2.778196783842665e-06, "loss": 0.7366, "step": 14241 }, { "epoch": 2.29, "grad_norm": 14.779480934143066, "learning_rate": 2.776994716090097e-06, "loss": 0.8979, "step": 14242 }, { "epoch": 2.29, "grad_norm": 10.436677932739258, "learning_rate": 2.7757928665165634e-06, "loss": 0.6602, "step": 14243 }, { "epoch": 2.29, "grad_norm": 10.534110069274902, "learning_rate": 2.774591235158366e-06, "loss": 0.6744, "step": 14244 }, { "epoch": 2.29, "grad_norm": 11.129396438598633, "learning_rate": 2.7733898220517975e-06, "loss": 0.5965, "step": 14245 }, { "epoch": 2.29, "grad_norm": 14.317723274230957, "learning_rate": 2.7721886272331545e-06, "loss": 0.5853, "step": 14246 }, { "epoch": 2.29, "grad_norm": 13.725811958312988, "learning_rate": 2.7709876507387166e-06, "loss": 0.6922, "step": 14247 }, { "epoch": 2.29, "grad_norm": 13.510064125061035, "learning_rate": 2.76978689260476e-06, "loss": 0.59, "step": 14248 }, { "epoch": 2.29, "grad_norm": 10.944544792175293, "learning_rate": 2.768586352867555e-06, "loss": 0.643, "step": 14249 }, { "epoch": 2.29, "grad_norm": 11.065998077392578, "learning_rate": 2.7673860315633703e-06, "loss": 0.6576, "step": 14250 }, { "epoch": 2.29, "grad_norm": 11.114583969116211, "learning_rate": 2.766185928728455e-06, "loss": 0.7941, "step": 14251 }, { "epoch": 2.29, "grad_norm": 14.940369606018066, "learning_rate": 2.7649860443990685e-06, "loss": 0.7282, "step": 14252 }, { "epoch": 2.29, "grad_norm": 11.244258880615234, "learning_rate": 2.763786378611446e-06, "loss": 0.7336, "step": 14253 }, { "epoch": 2.29, "grad_norm": 12.45068645477295, "learning_rate": 2.7625869314018326e-06, "loss": 0.8245, "step": 14254 }, { "epoch": 2.29, "grad_norm": 14.178512573242188, "learning_rate": 2.761387702806455e-06, "loss": 0.6898, "step": 14255 }, { "epoch": 2.29, "grad_norm": 12.905345916748047, "learning_rate": 2.760188692861534e-06, "loss": 0.6585, "step": 14256 }, { "epoch": 2.3, "grad_norm": 12.750982284545898, "learning_rate": 2.7589899016032927e-06, "loss": 0.9096, "step": 14257 }, { "epoch": 2.3, "grad_norm": 13.444672584533691, "learning_rate": 2.7577913290679404e-06, "loss": 0.7909, "step": 14258 }, { "epoch": 2.3, "grad_norm": 12.306925773620605, "learning_rate": 2.7565929752916775e-06, "loss": 0.7458, "step": 14259 }, { "epoch": 2.3, "grad_norm": 10.588184356689453, "learning_rate": 2.7553948403107076e-06, "loss": 0.645, "step": 14260 }, { "epoch": 2.3, "grad_norm": 34.783607482910156, "learning_rate": 2.7541969241612155e-06, "loss": 0.6045, "step": 14261 }, { "epoch": 2.3, "grad_norm": 13.196828842163086, "learning_rate": 2.7529992268793927e-06, "loss": 0.5569, "step": 14262 }, { "epoch": 2.3, "grad_norm": 14.504570007324219, "learning_rate": 2.7518017485014126e-06, "loss": 1.2028, "step": 14263 }, { "epoch": 2.3, "grad_norm": 11.440496444702148, "learning_rate": 2.750604489063444e-06, "loss": 0.619, "step": 14264 }, { "epoch": 2.3, "grad_norm": 12.569992065429688, "learning_rate": 2.7494074486016575e-06, "loss": 0.7463, "step": 14265 }, { "epoch": 2.3, "grad_norm": 13.148749351501465, "learning_rate": 2.748210627152206e-06, "loss": 0.5443, "step": 14266 }, { "epoch": 2.3, "grad_norm": 15.817272186279297, "learning_rate": 2.7470140247512456e-06, "loss": 0.6317, "step": 14267 }, { "epoch": 2.3, "grad_norm": 12.335779190063477, "learning_rate": 2.745817641434917e-06, "loss": 0.8154, "step": 14268 }, { "epoch": 2.3, "grad_norm": 11.586612701416016, "learning_rate": 2.7446214772393578e-06, "loss": 0.7864, "step": 14269 }, { "epoch": 2.3, "grad_norm": 11.388457298278809, "learning_rate": 2.743425532200701e-06, "loss": 0.6716, "step": 14270 }, { "epoch": 2.3, "grad_norm": 13.653364181518555, "learning_rate": 2.7422298063550746e-06, "loss": 0.709, "step": 14271 }, { "epoch": 2.3, "grad_norm": 13.471330642700195, "learning_rate": 2.741034299738592e-06, "loss": 0.7775, "step": 14272 }, { "epoch": 2.3, "grad_norm": 11.415299415588379, "learning_rate": 2.7398390123873686e-06, "loss": 0.6942, "step": 14273 }, { "epoch": 2.3, "grad_norm": 11.266831398010254, "learning_rate": 2.738643944337509e-06, "loss": 0.6926, "step": 14274 }, { "epoch": 2.3, "grad_norm": 15.42720890045166, "learning_rate": 2.7374490956251067e-06, "loss": 0.6585, "step": 14275 }, { "epoch": 2.3, "grad_norm": 11.673988342285156, "learning_rate": 2.7362544662862588e-06, "loss": 0.7312, "step": 14276 }, { "epoch": 2.3, "grad_norm": 11.224820137023926, "learning_rate": 2.7350600563570475e-06, "loss": 0.5545, "step": 14277 }, { "epoch": 2.3, "grad_norm": 11.792537689208984, "learning_rate": 2.733865865873554e-06, "loss": 0.7257, "step": 14278 }, { "epoch": 2.3, "grad_norm": 10.265223503112793, "learning_rate": 2.73267189487185e-06, "loss": 0.6084, "step": 14279 }, { "epoch": 2.3, "grad_norm": 12.159116744995117, "learning_rate": 2.7314781433879958e-06, "loss": 0.6674, "step": 14280 }, { "epoch": 2.3, "grad_norm": 12.118756294250488, "learning_rate": 2.730284611458056e-06, "loss": 0.5377, "step": 14281 }, { "epoch": 2.3, "grad_norm": 11.873248100280762, "learning_rate": 2.7290912991180808e-06, "loss": 0.6907, "step": 14282 }, { "epoch": 2.3, "grad_norm": 11.972882270812988, "learning_rate": 2.7278982064041116e-06, "loss": 0.7797, "step": 14283 }, { "epoch": 2.3, "grad_norm": 11.54602336883545, "learning_rate": 2.726705333352193e-06, "loss": 0.7065, "step": 14284 }, { "epoch": 2.3, "grad_norm": 13.579597473144531, "learning_rate": 2.7255126799983523e-06, "loss": 0.6021, "step": 14285 }, { "epoch": 2.3, "grad_norm": 11.281060218811035, "learning_rate": 2.72432024637862e-06, "loss": 0.6578, "step": 14286 }, { "epoch": 2.3, "grad_norm": 12.168644905090332, "learning_rate": 2.723128032529011e-06, "loss": 0.6392, "step": 14287 }, { "epoch": 2.3, "grad_norm": 9.229133605957031, "learning_rate": 2.7219360384855364e-06, "loss": 0.6815, "step": 14288 }, { "epoch": 2.3, "grad_norm": 11.311813354492188, "learning_rate": 2.7207442642842076e-06, "loss": 0.6797, "step": 14289 }, { "epoch": 2.3, "grad_norm": 10.716745376586914, "learning_rate": 2.7195527099610153e-06, "loss": 0.5541, "step": 14290 }, { "epoch": 2.3, "grad_norm": 14.900607109069824, "learning_rate": 2.7183613755519566e-06, "loss": 0.7664, "step": 14291 }, { "epoch": 2.3, "grad_norm": 12.80508804321289, "learning_rate": 2.7171702610930195e-06, "loss": 0.6177, "step": 14292 }, { "epoch": 2.3, "grad_norm": 11.109173774719238, "learning_rate": 2.7159793666201804e-06, "loss": 0.7958, "step": 14293 }, { "epoch": 2.3, "grad_norm": 10.885211944580078, "learning_rate": 2.714788692169409e-06, "loss": 0.5901, "step": 14294 }, { "epoch": 2.3, "grad_norm": 13.558517456054688, "learning_rate": 2.7135982377766746e-06, "loss": 0.4832, "step": 14295 }, { "epoch": 2.3, "grad_norm": 11.412652969360352, "learning_rate": 2.7124080034779332e-06, "loss": 0.6941, "step": 14296 }, { "epoch": 2.3, "grad_norm": 13.310372352600098, "learning_rate": 2.7112179893091415e-06, "loss": 0.729, "step": 14297 }, { "epoch": 2.3, "grad_norm": 13.81208610534668, "learning_rate": 2.7100281953062424e-06, "loss": 0.5794, "step": 14298 }, { "epoch": 2.3, "grad_norm": 10.54415225982666, "learning_rate": 2.708838621505173e-06, "loss": 0.7023, "step": 14299 }, { "epoch": 2.3, "grad_norm": 13.888278007507324, "learning_rate": 2.707649267941871e-06, "loss": 0.7585, "step": 14300 }, { "epoch": 2.3, "grad_norm": 12.762472152709961, "learning_rate": 2.7064601346522546e-06, "loss": 0.7787, "step": 14301 }, { "epoch": 2.3, "grad_norm": 11.741864204406738, "learning_rate": 2.705271221672251e-06, "loss": 0.7502, "step": 14302 }, { "epoch": 2.3, "grad_norm": 13.952034950256348, "learning_rate": 2.7040825290377693e-06, "loss": 0.6391, "step": 14303 }, { "epoch": 2.3, "grad_norm": 12.583242416381836, "learning_rate": 2.702894056784713e-06, "loss": 0.9749, "step": 14304 }, { "epoch": 2.3, "grad_norm": 12.966887474060059, "learning_rate": 2.701705804948984e-06, "loss": 0.6514, "step": 14305 }, { "epoch": 2.3, "grad_norm": 16.47513198852539, "learning_rate": 2.7005177735664755e-06, "loss": 0.8597, "step": 14306 }, { "epoch": 2.3, "grad_norm": 12.322226524353027, "learning_rate": 2.6993299626730695e-06, "loss": 0.6406, "step": 14307 }, { "epoch": 2.3, "grad_norm": 12.418919563293457, "learning_rate": 2.6981423723046494e-06, "loss": 0.7263, "step": 14308 }, { "epoch": 2.3, "grad_norm": 12.209009170532227, "learning_rate": 2.6969550024970836e-06, "loss": 0.7031, "step": 14309 }, { "epoch": 2.3, "grad_norm": 12.956315040588379, "learning_rate": 2.69576785328624e-06, "loss": 0.6696, "step": 14310 }, { "epoch": 2.3, "grad_norm": 10.842154502868652, "learning_rate": 2.6945809247079813e-06, "loss": 0.7245, "step": 14311 }, { "epoch": 2.3, "grad_norm": 11.52360725402832, "learning_rate": 2.6933942167981563e-06, "loss": 0.6116, "step": 14312 }, { "epoch": 2.3, "grad_norm": 49.401161193847656, "learning_rate": 2.6922077295926086e-06, "loss": 0.8962, "step": 14313 }, { "epoch": 2.3, "grad_norm": 14.137031555175781, "learning_rate": 2.6910214631271826e-06, "loss": 0.6485, "step": 14314 }, { "epoch": 2.3, "grad_norm": 12.87243938446045, "learning_rate": 2.6898354174377052e-06, "loss": 0.6462, "step": 14315 }, { "epoch": 2.3, "grad_norm": 11.360735893249512, "learning_rate": 2.6886495925600085e-06, "loss": 0.7511, "step": 14316 }, { "epoch": 2.3, "grad_norm": 13.460502624511719, "learning_rate": 2.6874639885299092e-06, "loss": 0.7147, "step": 14317 }, { "epoch": 2.3, "grad_norm": 11.449450492858887, "learning_rate": 2.6862786053832157e-06, "loss": 0.5555, "step": 14318 }, { "epoch": 2.31, "grad_norm": 11.912785530090332, "learning_rate": 2.6850934431557397e-06, "loss": 0.5851, "step": 14319 }, { "epoch": 2.31, "grad_norm": 12.518824577331543, "learning_rate": 2.6839085018832754e-06, "loss": 0.5835, "step": 14320 }, { "epoch": 2.31, "grad_norm": 13.219623565673828, "learning_rate": 2.6827237816016216e-06, "loss": 0.795, "step": 14321 }, { "epoch": 2.31, "grad_norm": 13.366808891296387, "learning_rate": 2.681539282346559e-06, "loss": 0.6521, "step": 14322 }, { "epoch": 2.31, "grad_norm": 11.614775657653809, "learning_rate": 2.680355004153866e-06, "loss": 0.5549, "step": 14323 }, { "epoch": 2.31, "grad_norm": 13.886558532714844, "learning_rate": 2.6791709470593197e-06, "loss": 0.7442, "step": 14324 }, { "epoch": 2.31, "grad_norm": 12.554370880126953, "learning_rate": 2.677987111098683e-06, "loss": 0.7102, "step": 14325 }, { "epoch": 2.31, "grad_norm": 12.5820894241333, "learning_rate": 2.6768034963077126e-06, "loss": 0.6753, "step": 14326 }, { "epoch": 2.31, "grad_norm": 9.111964225769043, "learning_rate": 2.6756201027221673e-06, "loss": 0.675, "step": 14327 }, { "epoch": 2.31, "grad_norm": 13.628631591796875, "learning_rate": 2.6744369303777853e-06, "loss": 0.5622, "step": 14328 }, { "epoch": 2.31, "grad_norm": 11.091557502746582, "learning_rate": 2.673253979310313e-06, "loss": 0.5463, "step": 14329 }, { "epoch": 2.31, "grad_norm": 11.012641906738281, "learning_rate": 2.672071249555477e-06, "loss": 0.6686, "step": 14330 }, { "epoch": 2.31, "grad_norm": 11.61453628540039, "learning_rate": 2.6708887411490035e-06, "loss": 0.6302, "step": 14331 }, { "epoch": 2.31, "grad_norm": 11.270559310913086, "learning_rate": 2.669706454126617e-06, "loss": 0.72, "step": 14332 }, { "epoch": 2.31, "grad_norm": 14.230050086975098, "learning_rate": 2.6685243885240266e-06, "loss": 0.8145, "step": 14333 }, { "epoch": 2.31, "grad_norm": 12.068037986755371, "learning_rate": 2.667342544376934e-06, "loss": 0.5643, "step": 14334 }, { "epoch": 2.31, "grad_norm": 10.407644271850586, "learning_rate": 2.666160921721045e-06, "loss": 0.7191, "step": 14335 }, { "epoch": 2.31, "grad_norm": 11.078216552734375, "learning_rate": 2.6649795205920482e-06, "loss": 0.8697, "step": 14336 }, { "epoch": 2.31, "grad_norm": 11.629767417907715, "learning_rate": 2.663798341025626e-06, "loss": 1.0055, "step": 14337 }, { "epoch": 2.31, "grad_norm": 11.184457778930664, "learning_rate": 2.662617383057464e-06, "loss": 0.7557, "step": 14338 }, { "epoch": 2.31, "grad_norm": 10.387423515319824, "learning_rate": 2.6614366467232287e-06, "loss": 0.7402, "step": 14339 }, { "epoch": 2.31, "grad_norm": 14.930948257446289, "learning_rate": 2.660256132058591e-06, "loss": 0.7943, "step": 14340 }, { "epoch": 2.31, "grad_norm": 11.190339088439941, "learning_rate": 2.6590758390992056e-06, "loss": 0.8514, "step": 14341 }, { "epoch": 2.31, "grad_norm": 10.753263473510742, "learning_rate": 2.657895767880723e-06, "loss": 0.5164, "step": 14342 }, { "epoch": 2.31, "grad_norm": 10.268621444702148, "learning_rate": 2.656715918438794e-06, "loss": 0.7663, "step": 14343 }, { "epoch": 2.31, "grad_norm": 11.878010749816895, "learning_rate": 2.65553629080905e-06, "loss": 0.7718, "step": 14344 }, { "epoch": 2.31, "grad_norm": 10.968085289001465, "learning_rate": 2.654356885027132e-06, "loss": 0.6566, "step": 14345 }, { "epoch": 2.31, "grad_norm": 12.056906700134277, "learning_rate": 2.6531777011286607e-06, "loss": 0.5461, "step": 14346 }, { "epoch": 2.31, "grad_norm": 14.88599681854248, "learning_rate": 2.651998739149251e-06, "loss": 0.658, "step": 14347 }, { "epoch": 2.31, "grad_norm": 13.216568946838379, "learning_rate": 2.6508199991245222e-06, "loss": 0.6478, "step": 14348 }, { "epoch": 2.31, "grad_norm": 11.348012924194336, "learning_rate": 2.649641481090075e-06, "loss": 0.4629, "step": 14349 }, { "epoch": 2.31, "grad_norm": 11.709750175476074, "learning_rate": 2.6484631850815056e-06, "loss": 0.8666, "step": 14350 }, { "epoch": 2.31, "grad_norm": 10.667969703674316, "learning_rate": 2.647285111134409e-06, "loss": 0.5821, "step": 14351 }, { "epoch": 2.31, "grad_norm": 11.69218921661377, "learning_rate": 2.646107259284373e-06, "loss": 0.8803, "step": 14352 }, { "epoch": 2.31, "grad_norm": 13.679262161254883, "learning_rate": 2.64492962956697e-06, "loss": 0.8067, "step": 14353 }, { "epoch": 2.31, "grad_norm": 14.20540714263916, "learning_rate": 2.643752222017778e-06, "loss": 0.7495, "step": 14354 }, { "epoch": 2.31, "grad_norm": 13.214000701904297, "learning_rate": 2.6425750366723544e-06, "loss": 0.7791, "step": 14355 }, { "epoch": 2.31, "grad_norm": 11.087979316711426, "learning_rate": 2.6413980735662647e-06, "loss": 0.6902, "step": 14356 }, { "epoch": 2.31, "grad_norm": 11.651284217834473, "learning_rate": 2.6402213327350578e-06, "loss": 0.6995, "step": 14357 }, { "epoch": 2.31, "grad_norm": 15.537287712097168, "learning_rate": 2.6390448142142754e-06, "loss": 0.7889, "step": 14358 }, { "epoch": 2.31, "grad_norm": 13.233222961425781, "learning_rate": 2.6378685180394604e-06, "loss": 0.6087, "step": 14359 }, { "epoch": 2.31, "grad_norm": 13.606812477111816, "learning_rate": 2.6366924442461416e-06, "loss": 0.7187, "step": 14360 }, { "epoch": 2.31, "grad_norm": 12.871872901916504, "learning_rate": 2.635516592869841e-06, "loss": 0.6925, "step": 14361 }, { "epoch": 2.31, "grad_norm": 14.49001693725586, "learning_rate": 2.634340963946083e-06, "loss": 0.6682, "step": 14362 }, { "epoch": 2.31, "grad_norm": 13.894009590148926, "learning_rate": 2.633165557510372e-06, "loss": 0.7805, "step": 14363 }, { "epoch": 2.31, "grad_norm": 11.504862785339355, "learning_rate": 2.631990373598218e-06, "loss": 0.7642, "step": 14364 }, { "epoch": 2.31, "grad_norm": 13.751893997192383, "learning_rate": 2.630815412245117e-06, "loss": 0.77, "step": 14365 }, { "epoch": 2.31, "grad_norm": 10.778913497924805, "learning_rate": 2.6296406734865563e-06, "loss": 0.6056, "step": 14366 }, { "epoch": 2.31, "grad_norm": 13.244331359863281, "learning_rate": 2.6284661573580263e-06, "loss": 0.6421, "step": 14367 }, { "epoch": 2.31, "grad_norm": 10.768790245056152, "learning_rate": 2.6272918638949984e-06, "loss": 0.5061, "step": 14368 }, { "epoch": 2.31, "grad_norm": 11.326924324035645, "learning_rate": 2.6261177931329497e-06, "loss": 0.8369, "step": 14369 }, { "epoch": 2.31, "grad_norm": 13.000722885131836, "learning_rate": 2.624943945107342e-06, "loss": 0.8824, "step": 14370 }, { "epoch": 2.31, "grad_norm": 11.066144943237305, "learning_rate": 2.623770319853629e-06, "loss": 0.7984, "step": 14371 }, { "epoch": 2.31, "grad_norm": 11.339174270629883, "learning_rate": 2.6225969174072643e-06, "loss": 0.6978, "step": 14372 }, { "epoch": 2.31, "grad_norm": 12.557601928710938, "learning_rate": 2.6214237378036945e-06, "loss": 0.8203, "step": 14373 }, { "epoch": 2.31, "grad_norm": 13.106759071350098, "learning_rate": 2.6202507810783516e-06, "loss": 0.6821, "step": 14374 }, { "epoch": 2.31, "grad_norm": 13.682652473449707, "learning_rate": 2.619078047266671e-06, "loss": 0.672, "step": 14375 }, { "epoch": 2.31, "grad_norm": 12.76230239868164, "learning_rate": 2.617905536404075e-06, "loss": 0.6188, "step": 14376 }, { "epoch": 2.31, "grad_norm": 12.366868019104004, "learning_rate": 2.6167332485259767e-06, "loss": 0.715, "step": 14377 }, { "epoch": 2.31, "grad_norm": 11.981959342956543, "learning_rate": 2.615561183667793e-06, "loss": 0.6648, "step": 14378 }, { "epoch": 2.31, "grad_norm": 12.486673355102539, "learning_rate": 2.6143893418649203e-06, "loss": 0.653, "step": 14379 }, { "epoch": 2.31, "grad_norm": 12.556445121765137, "learning_rate": 2.6132177231527622e-06, "loss": 0.7637, "step": 14380 }, { "epoch": 2.32, "grad_norm": 10.052746772766113, "learning_rate": 2.6120463275667063e-06, "loss": 0.5798, "step": 14381 }, { "epoch": 2.32, "grad_norm": 14.5535306930542, "learning_rate": 2.6108751551421317e-06, "loss": 0.5818, "step": 14382 }, { "epoch": 2.32, "grad_norm": 13.981943130493164, "learning_rate": 2.6097042059144217e-06, "loss": 0.6324, "step": 14383 }, { "epoch": 2.32, "grad_norm": 11.04856014251709, "learning_rate": 2.608533479918943e-06, "loss": 0.6344, "step": 14384 }, { "epoch": 2.32, "grad_norm": 13.928092956542969, "learning_rate": 2.607362977191056e-06, "loss": 0.8786, "step": 14385 }, { "epoch": 2.32, "grad_norm": 12.983397483825684, "learning_rate": 2.606192697766121e-06, "loss": 0.7605, "step": 14386 }, { "epoch": 2.32, "grad_norm": 13.13330078125, "learning_rate": 2.6050226416794855e-06, "loss": 0.6492, "step": 14387 }, { "epoch": 2.32, "grad_norm": 14.136921882629395, "learning_rate": 2.6038528089664947e-06, "loss": 0.9152, "step": 14388 }, { "epoch": 2.32, "grad_norm": 14.485403060913086, "learning_rate": 2.6026831996624846e-06, "loss": 0.6224, "step": 14389 }, { "epoch": 2.32, "grad_norm": 13.890463829040527, "learning_rate": 2.6015138138027797e-06, "loss": 0.5175, "step": 14390 }, { "epoch": 2.32, "grad_norm": 12.272083282470703, "learning_rate": 2.600344651422706e-06, "loss": 0.8963, "step": 14391 }, { "epoch": 2.32, "grad_norm": 11.270157814025879, "learning_rate": 2.5991757125575823e-06, "loss": 0.6713, "step": 14392 }, { "epoch": 2.32, "grad_norm": 15.565813064575195, "learning_rate": 2.598006997242711e-06, "loss": 0.845, "step": 14393 }, { "epoch": 2.32, "grad_norm": 14.40086841583252, "learning_rate": 2.596838505513403e-06, "loss": 0.6612, "step": 14394 }, { "epoch": 2.32, "grad_norm": 11.921039581298828, "learning_rate": 2.595670237404949e-06, "loss": 0.7169, "step": 14395 }, { "epoch": 2.32, "grad_norm": 12.958331108093262, "learning_rate": 2.594502192952635e-06, "loss": 0.6123, "step": 14396 }, { "epoch": 2.32, "grad_norm": 15.166528701782227, "learning_rate": 2.593334372191749e-06, "loss": 0.6405, "step": 14397 }, { "epoch": 2.32, "grad_norm": 10.616456985473633, "learning_rate": 2.5921667751575606e-06, "loss": 0.6655, "step": 14398 }, { "epoch": 2.32, "grad_norm": 12.178664207458496, "learning_rate": 2.5909994018853448e-06, "loss": 0.8825, "step": 14399 }, { "epoch": 2.32, "grad_norm": 11.455501556396484, "learning_rate": 2.5898322524103602e-06, "loss": 0.8646, "step": 14400 }, { "epoch": 2.32, "grad_norm": 14.970016479492188, "learning_rate": 2.5886653267678575e-06, "loss": 0.7771, "step": 14401 }, { "epoch": 2.32, "grad_norm": 12.159623146057129, "learning_rate": 2.5874986249930935e-06, "loss": 0.5491, "step": 14402 }, { "epoch": 2.32, "grad_norm": 10.46264362335205, "learning_rate": 2.5863321471213044e-06, "loss": 0.7702, "step": 14403 }, { "epoch": 2.32, "grad_norm": 13.160093307495117, "learning_rate": 2.5851658931877233e-06, "loss": 0.863, "step": 14404 }, { "epoch": 2.32, "grad_norm": 12.60438346862793, "learning_rate": 2.5839998632275844e-06, "loss": 0.6328, "step": 14405 }, { "epoch": 2.32, "grad_norm": 13.253693580627441, "learning_rate": 2.5828340572761025e-06, "loss": 0.7533, "step": 14406 }, { "epoch": 2.32, "grad_norm": 12.498937606811523, "learning_rate": 2.5816684753684973e-06, "loss": 0.8168, "step": 14407 }, { "epoch": 2.32, "grad_norm": 11.222399711608887, "learning_rate": 2.5805031175399753e-06, "loss": 0.5405, "step": 14408 }, { "epoch": 2.32, "grad_norm": 16.622058868408203, "learning_rate": 2.5793379838257325e-06, "loss": 0.8214, "step": 14409 }, { "epoch": 2.32, "grad_norm": 11.266271591186523, "learning_rate": 2.5781730742609714e-06, "loss": 0.5572, "step": 14410 }, { "epoch": 2.32, "grad_norm": 10.35236930847168, "learning_rate": 2.5770083888808706e-06, "loss": 0.6676, "step": 14411 }, { "epoch": 2.32, "grad_norm": 11.982670783996582, "learning_rate": 2.575843927720616e-06, "loss": 0.851, "step": 14412 }, { "epoch": 2.32, "grad_norm": 10.109899520874023, "learning_rate": 2.5746796908153826e-06, "loss": 0.5243, "step": 14413 }, { "epoch": 2.32, "grad_norm": 11.038264274597168, "learning_rate": 2.5735156782003368e-06, "loss": 0.7047, "step": 14414 }, { "epoch": 2.32, "grad_norm": 12.94550609588623, "learning_rate": 2.5723518899106337e-06, "loss": 0.7495, "step": 14415 }, { "epoch": 2.32, "grad_norm": 12.42089557647705, "learning_rate": 2.571188325981434e-06, "loss": 0.7813, "step": 14416 }, { "epoch": 2.32, "grad_norm": 13.470848083496094, "learning_rate": 2.5700249864478786e-06, "loss": 0.8788, "step": 14417 }, { "epoch": 2.32, "grad_norm": 12.87117862701416, "learning_rate": 2.568861871345113e-06, "loss": 0.5594, "step": 14418 }, { "epoch": 2.32, "grad_norm": 15.94211483001709, "learning_rate": 2.5676989807082675e-06, "loss": 0.7803, "step": 14419 }, { "epoch": 2.32, "grad_norm": 11.011310577392578, "learning_rate": 2.5665363145724652e-06, "loss": 0.8044, "step": 14420 }, { "epoch": 2.32, "grad_norm": 13.069588661193848, "learning_rate": 2.5653738729728327e-06, "loss": 0.8641, "step": 14421 }, { "epoch": 2.32, "grad_norm": 11.446248054504395, "learning_rate": 2.5642116559444764e-06, "loss": 0.4821, "step": 14422 }, { "epoch": 2.32, "grad_norm": 11.293981552124023, "learning_rate": 2.5630496635225076e-06, "loss": 0.5792, "step": 14423 }, { "epoch": 2.32, "grad_norm": 11.962512016296387, "learning_rate": 2.5618878957420235e-06, "loss": 0.7633, "step": 14424 }, { "epoch": 2.32, "grad_norm": 11.752633094787598, "learning_rate": 2.5607263526381133e-06, "loss": 0.9123, "step": 14425 }, { "epoch": 2.32, "grad_norm": 16.651269912719727, "learning_rate": 2.5595650342458687e-06, "loss": 0.7334, "step": 14426 }, { "epoch": 2.32, "grad_norm": 15.344560623168945, "learning_rate": 2.5584039406003658e-06, "loss": 0.7141, "step": 14427 }, { "epoch": 2.32, "grad_norm": 12.615743637084961, "learning_rate": 2.5572430717366738e-06, "loss": 0.6717, "step": 14428 }, { "epoch": 2.32, "grad_norm": 12.937851905822754, "learning_rate": 2.5560824276898633e-06, "loss": 0.7264, "step": 14429 }, { "epoch": 2.32, "grad_norm": 10.639363288879395, "learning_rate": 2.554922008494988e-06, "loss": 0.5216, "step": 14430 }, { "epoch": 2.32, "grad_norm": 14.540587425231934, "learning_rate": 2.5537618141871047e-06, "loss": 0.8172, "step": 14431 }, { "epoch": 2.32, "grad_norm": 10.257243156433105, "learning_rate": 2.5526018448012536e-06, "loss": 0.6385, "step": 14432 }, { "epoch": 2.32, "grad_norm": 13.223289489746094, "learning_rate": 2.5514421003724744e-06, "loss": 0.7603, "step": 14433 }, { "epoch": 2.32, "grad_norm": 11.010809898376465, "learning_rate": 2.5502825809358023e-06, "loss": 0.4779, "step": 14434 }, { "epoch": 2.32, "grad_norm": 14.055280685424805, "learning_rate": 2.5491232865262585e-06, "loss": 0.8689, "step": 14435 }, { "epoch": 2.32, "grad_norm": 22.166444778442383, "learning_rate": 2.5479642171788587e-06, "loss": 0.5926, "step": 14436 }, { "epoch": 2.32, "grad_norm": 14.92623233795166, "learning_rate": 2.546805372928619e-06, "loss": 0.5786, "step": 14437 }, { "epoch": 2.32, "grad_norm": 10.911227226257324, "learning_rate": 2.54564675381054e-06, "loss": 0.7372, "step": 14438 }, { "epoch": 2.32, "grad_norm": 11.938567161560059, "learning_rate": 2.5444883598596183e-06, "loss": 0.6196, "step": 14439 }, { "epoch": 2.32, "grad_norm": 12.540617942810059, "learning_rate": 2.543330191110849e-06, "loss": 0.8284, "step": 14440 }, { "epoch": 2.32, "grad_norm": 10.611568450927734, "learning_rate": 2.5421722475992095e-06, "loss": 0.7589, "step": 14441 }, { "epoch": 2.32, "grad_norm": 13.305675506591797, "learning_rate": 2.541014529359683e-06, "loss": 0.624, "step": 14442 }, { "epoch": 2.33, "grad_norm": 10.77589225769043, "learning_rate": 2.539857036427238e-06, "loss": 0.5263, "step": 14443 }, { "epoch": 2.33, "grad_norm": 14.615789413452148, "learning_rate": 2.538699768836834e-06, "loss": 0.7589, "step": 14444 }, { "epoch": 2.33, "grad_norm": 14.352485656738281, "learning_rate": 2.5375427266234332e-06, "loss": 0.8613, "step": 14445 }, { "epoch": 2.33, "grad_norm": 11.18929672241211, "learning_rate": 2.5363859098219803e-06, "loss": 0.5518, "step": 14446 }, { "epoch": 2.33, "grad_norm": 13.75903034210205, "learning_rate": 2.5352293184674226e-06, "loss": 0.7248, "step": 14447 }, { "epoch": 2.33, "grad_norm": 12.234545707702637, "learning_rate": 2.534072952594694e-06, "loss": 0.8578, "step": 14448 }, { "epoch": 2.33, "grad_norm": 11.583377838134766, "learning_rate": 2.5329168122387213e-06, "loss": 0.6449, "step": 14449 }, { "epoch": 2.33, "grad_norm": 12.193074226379395, "learning_rate": 2.5317608974344343e-06, "loss": 0.7406, "step": 14450 }, { "epoch": 2.33, "grad_norm": 12.655213356018066, "learning_rate": 2.5306052082167432e-06, "loss": 0.665, "step": 14451 }, { "epoch": 2.33, "grad_norm": 13.362578392028809, "learning_rate": 2.529449744620551e-06, "loss": 0.7228, "step": 14452 }, { "epoch": 2.33, "grad_norm": 11.016059875488281, "learning_rate": 2.528294506680774e-06, "loss": 0.5125, "step": 14453 }, { "epoch": 2.33, "grad_norm": 12.034854888916016, "learning_rate": 2.5271394944323e-06, "loss": 0.7731, "step": 14454 }, { "epoch": 2.33, "grad_norm": 14.480422019958496, "learning_rate": 2.525984707910015e-06, "loss": 0.5997, "step": 14455 }, { "epoch": 2.33, "grad_norm": 11.729199409484863, "learning_rate": 2.5248301471488056e-06, "loss": 0.6803, "step": 14456 }, { "epoch": 2.33, "grad_norm": 14.052785873413086, "learning_rate": 2.5236758121835415e-06, "loss": 0.6999, "step": 14457 }, { "epoch": 2.33, "grad_norm": 10.838650703430176, "learning_rate": 2.5225217030490966e-06, "loss": 0.5853, "step": 14458 }, { "epoch": 2.33, "grad_norm": 13.220365524291992, "learning_rate": 2.521367819780328e-06, "loss": 0.7818, "step": 14459 }, { "epoch": 2.33, "grad_norm": 11.470269203186035, "learning_rate": 2.520214162412089e-06, "loss": 0.6546, "step": 14460 }, { "epoch": 2.33, "grad_norm": 11.581206321716309, "learning_rate": 2.5190607309792326e-06, "loss": 0.6032, "step": 14461 }, { "epoch": 2.33, "grad_norm": 15.143156051635742, "learning_rate": 2.5179075255165954e-06, "loss": 0.8634, "step": 14462 }, { "epoch": 2.33, "grad_norm": 12.835502624511719, "learning_rate": 2.5167545460590092e-06, "loss": 0.8621, "step": 14463 }, { "epoch": 2.33, "grad_norm": 12.706756591796875, "learning_rate": 2.515601792641306e-06, "loss": 0.6475, "step": 14464 }, { "epoch": 2.33, "grad_norm": 10.393942832946777, "learning_rate": 2.514449265298301e-06, "loss": 0.7205, "step": 14465 }, { "epoch": 2.33, "grad_norm": 13.647705078125, "learning_rate": 2.5132969640648143e-06, "loss": 0.7481, "step": 14466 }, { "epoch": 2.33, "grad_norm": 13.520646095275879, "learning_rate": 2.512144888975647e-06, "loss": 0.6364, "step": 14467 }, { "epoch": 2.33, "grad_norm": 13.314960479736328, "learning_rate": 2.510993040065598e-06, "loss": 0.7342, "step": 14468 }, { "epoch": 2.33, "grad_norm": 12.12971305847168, "learning_rate": 2.5098414173694663e-06, "loss": 0.669, "step": 14469 }, { "epoch": 2.33, "grad_norm": 13.604351043701172, "learning_rate": 2.508690020922033e-06, "loss": 0.7007, "step": 14470 }, { "epoch": 2.33, "grad_norm": 11.945422172546387, "learning_rate": 2.5075388507580757e-06, "loss": 0.5677, "step": 14471 }, { "epoch": 2.33, "grad_norm": 12.710224151611328, "learning_rate": 2.5063879069123708e-06, "loss": 0.6587, "step": 14472 }, { "epoch": 2.33, "grad_norm": 11.4595947265625, "learning_rate": 2.5052371894196848e-06, "loss": 0.6981, "step": 14473 }, { "epoch": 2.33, "grad_norm": 12.121622085571289, "learning_rate": 2.5040866983147715e-06, "loss": 0.5619, "step": 14474 }, { "epoch": 2.33, "grad_norm": 12.76108169555664, "learning_rate": 2.502936433632388e-06, "loss": 0.6052, "step": 14475 }, { "epoch": 2.33, "grad_norm": 11.2529935836792, "learning_rate": 2.501786395407274e-06, "loss": 0.6903, "step": 14476 }, { "epoch": 2.33, "grad_norm": 12.393826484680176, "learning_rate": 2.5006365836741744e-06, "loss": 0.7391, "step": 14477 }, { "epoch": 2.33, "grad_norm": 12.974416732788086, "learning_rate": 2.499486998467816e-06, "loss": 0.808, "step": 14478 }, { "epoch": 2.33, "grad_norm": 13.724331855773926, "learning_rate": 2.498337639822922e-06, "loss": 0.6497, "step": 14479 }, { "epoch": 2.33, "grad_norm": 12.892427444458008, "learning_rate": 2.497188507774215e-06, "loss": 0.8082, "step": 14480 }, { "epoch": 2.33, "grad_norm": 11.353060722351074, "learning_rate": 2.4960396023564024e-06, "loss": 0.5816, "step": 14481 }, { "epoch": 2.33, "grad_norm": 12.093057632446289, "learning_rate": 2.4948909236041863e-06, "loss": 0.8023, "step": 14482 }, { "epoch": 2.33, "grad_norm": 12.208967208862305, "learning_rate": 2.4937424715522697e-06, "loss": 0.4834, "step": 14483 }, { "epoch": 2.33, "grad_norm": 11.43121337890625, "learning_rate": 2.492594246235336e-06, "loss": 0.7858, "step": 14484 }, { "epoch": 2.33, "grad_norm": 11.362277030944824, "learning_rate": 2.491446247688075e-06, "loss": 0.6337, "step": 14485 }, { "epoch": 2.33, "grad_norm": 12.49179744720459, "learning_rate": 2.490298475945161e-06, "loss": 0.8372, "step": 14486 }, { "epoch": 2.33, "grad_norm": 11.95728588104248, "learning_rate": 2.4891509310412596e-06, "loss": 0.7078, "step": 14487 }, { "epoch": 2.33, "grad_norm": 12.843602180480957, "learning_rate": 2.4880036130110407e-06, "loss": 0.7552, "step": 14488 }, { "epoch": 2.33, "grad_norm": 14.747889518737793, "learning_rate": 2.4868565218891538e-06, "loss": 0.6839, "step": 14489 }, { "epoch": 2.33, "grad_norm": 14.571763038635254, "learning_rate": 2.4857096577102545e-06, "loss": 0.8697, "step": 14490 }, { "epoch": 2.33, "grad_norm": 13.05185317993164, "learning_rate": 2.484563020508982e-06, "loss": 0.6844, "step": 14491 }, { "epoch": 2.33, "grad_norm": 12.069109916687012, "learning_rate": 2.4834166103199685e-06, "loss": 0.8499, "step": 14492 }, { "epoch": 2.33, "grad_norm": 12.343888282775879, "learning_rate": 2.482270427177845e-06, "loss": 0.4955, "step": 14493 }, { "epoch": 2.33, "grad_norm": 14.738757133483887, "learning_rate": 2.4811244711172376e-06, "loss": 0.8902, "step": 14494 }, { "epoch": 2.33, "grad_norm": 12.493922233581543, "learning_rate": 2.479978742172755e-06, "loss": 0.6778, "step": 14495 }, { "epoch": 2.33, "grad_norm": 12.88740062713623, "learning_rate": 2.4788332403790106e-06, "loss": 0.7467, "step": 14496 }, { "epoch": 2.33, "grad_norm": 11.405842781066895, "learning_rate": 2.4776879657706033e-06, "loss": 0.8355, "step": 14497 }, { "epoch": 2.33, "grad_norm": 9.80137825012207, "learning_rate": 2.476542918382123e-06, "loss": 0.7546, "step": 14498 }, { "epoch": 2.33, "grad_norm": 12.133323669433594, "learning_rate": 2.475398098248164e-06, "loss": 0.6011, "step": 14499 }, { "epoch": 2.33, "grad_norm": 12.211130142211914, "learning_rate": 2.474253505403301e-06, "loss": 0.7553, "step": 14500 }, { "epoch": 2.33, "grad_norm": 11.330057144165039, "learning_rate": 2.473109139882115e-06, "loss": 0.8363, "step": 14501 }, { "epoch": 2.33, "grad_norm": 13.933272361755371, "learning_rate": 2.4719650017191663e-06, "loss": 0.6665, "step": 14502 }, { "epoch": 2.33, "grad_norm": 12.133094787597656, "learning_rate": 2.470821090949016e-06, "loss": 0.8214, "step": 14503 }, { "epoch": 2.33, "grad_norm": 14.027365684509277, "learning_rate": 2.4696774076062204e-06, "loss": 0.6614, "step": 14504 }, { "epoch": 2.33, "grad_norm": 12.22065258026123, "learning_rate": 2.468533951725324e-06, "loss": 0.8004, "step": 14505 }, { "epoch": 2.34, "grad_norm": 14.56304931640625, "learning_rate": 2.467390723340863e-06, "loss": 0.7399, "step": 14506 }, { "epoch": 2.34, "grad_norm": 13.427105903625488, "learning_rate": 2.466247722487376e-06, "loss": 0.6147, "step": 14507 }, { "epoch": 2.34, "grad_norm": 12.258171081542969, "learning_rate": 2.4651049491993818e-06, "loss": 0.6423, "step": 14508 }, { "epoch": 2.34, "grad_norm": 12.828506469726562, "learning_rate": 2.4639624035114064e-06, "loss": 0.6533, "step": 14509 }, { "epoch": 2.34, "grad_norm": 12.418567657470703, "learning_rate": 2.4628200854579563e-06, "loss": 0.7399, "step": 14510 }, { "epoch": 2.34, "grad_norm": 12.269503593444824, "learning_rate": 2.461677995073537e-06, "loss": 0.9676, "step": 14511 }, { "epoch": 2.34, "grad_norm": 11.764290809631348, "learning_rate": 2.4605361323926504e-06, "loss": 0.7997, "step": 14512 }, { "epoch": 2.34, "grad_norm": 13.881061553955078, "learning_rate": 2.459394497449782e-06, "loss": 0.667, "step": 14513 }, { "epoch": 2.34, "grad_norm": 13.673123359680176, "learning_rate": 2.4582530902794198e-06, "loss": 0.7229, "step": 14514 }, { "epoch": 2.34, "grad_norm": 11.228458404541016, "learning_rate": 2.4571119109160436e-06, "loss": 0.622, "step": 14515 }, { "epoch": 2.34, "grad_norm": 10.274445533752441, "learning_rate": 2.4559709593941215e-06, "loss": 0.7819, "step": 14516 }, { "epoch": 2.34, "grad_norm": 12.778470993041992, "learning_rate": 2.4548302357481133e-06, "loss": 0.7502, "step": 14517 }, { "epoch": 2.34, "grad_norm": 12.883094787597656, "learning_rate": 2.4536897400124835e-06, "loss": 0.7034, "step": 14518 }, { "epoch": 2.34, "grad_norm": 11.619505882263184, "learning_rate": 2.4525494722216746e-06, "loss": 0.6091, "step": 14519 }, { "epoch": 2.34, "grad_norm": 11.568253517150879, "learning_rate": 2.4514094324101366e-06, "loss": 0.8603, "step": 14520 }, { "epoch": 2.34, "grad_norm": 12.495233535766602, "learning_rate": 2.4502696206123023e-06, "loss": 0.6132, "step": 14521 }, { "epoch": 2.34, "grad_norm": 12.968381881713867, "learning_rate": 2.449130036862598e-06, "loss": 0.629, "step": 14522 }, { "epoch": 2.34, "grad_norm": 10.532212257385254, "learning_rate": 2.4479906811954525e-06, "loss": 0.5553, "step": 14523 }, { "epoch": 2.34, "grad_norm": 13.260364532470703, "learning_rate": 2.446851553645274e-06, "loss": 0.6993, "step": 14524 }, { "epoch": 2.34, "grad_norm": 11.769173622131348, "learning_rate": 2.4457126542464782e-06, "loss": 0.6852, "step": 14525 }, { "epoch": 2.34, "grad_norm": 9.560994148254395, "learning_rate": 2.4445739830334638e-06, "loss": 0.6608, "step": 14526 }, { "epoch": 2.34, "grad_norm": 12.017579078674316, "learning_rate": 2.443435540040623e-06, "loss": 0.6525, "step": 14527 }, { "epoch": 2.34, "grad_norm": 11.327738761901855, "learning_rate": 2.4422973253023487e-06, "loss": 0.7039, "step": 14528 }, { "epoch": 2.34, "grad_norm": 11.61999797821045, "learning_rate": 2.4411593388530207e-06, "loss": 0.8368, "step": 14529 }, { "epoch": 2.34, "grad_norm": 11.564184188842773, "learning_rate": 2.440021580727008e-06, "loss": 0.6966, "step": 14530 }, { "epoch": 2.34, "grad_norm": 13.242446899414062, "learning_rate": 2.4388840509586855e-06, "loss": 0.72, "step": 14531 }, { "epoch": 2.34, "grad_norm": 12.58862018585205, "learning_rate": 2.437746749582407e-06, "loss": 0.7813, "step": 14532 }, { "epoch": 2.34, "grad_norm": 12.80109977722168, "learning_rate": 2.4366096766325287e-06, "loss": 0.7729, "step": 14533 }, { "epoch": 2.34, "grad_norm": 11.831705093383789, "learning_rate": 2.4354728321433996e-06, "loss": 0.5051, "step": 14534 }, { "epoch": 2.34, "grad_norm": 13.570789337158203, "learning_rate": 2.4343362161493554e-06, "loss": 0.82, "step": 14535 }, { "epoch": 2.34, "grad_norm": 11.578669548034668, "learning_rate": 2.4331998286847335e-06, "loss": 0.7414, "step": 14536 }, { "epoch": 2.34, "grad_norm": 12.783172607421875, "learning_rate": 2.432063669783857e-06, "loss": 0.6901, "step": 14537 }, { "epoch": 2.34, "grad_norm": 15.248430252075195, "learning_rate": 2.4309277394810417e-06, "loss": 0.634, "step": 14538 }, { "epoch": 2.34, "grad_norm": 12.558835983276367, "learning_rate": 2.4297920378106053e-06, "loss": 0.9401, "step": 14539 }, { "epoch": 2.34, "grad_norm": 14.277409553527832, "learning_rate": 2.4286565648068507e-06, "loss": 0.8708, "step": 14540 }, { "epoch": 2.34, "grad_norm": 9.050511360168457, "learning_rate": 2.427521320504074e-06, "loss": 0.5319, "step": 14541 }, { "epoch": 2.34, "grad_norm": 11.446208000183105, "learning_rate": 2.42638630493657e-06, "loss": 0.45, "step": 14542 }, { "epoch": 2.34, "grad_norm": 14.302610397338867, "learning_rate": 2.425251518138618e-06, "loss": 0.7789, "step": 14543 }, { "epoch": 2.34, "grad_norm": 11.558633804321289, "learning_rate": 2.4241169601445037e-06, "loss": 0.6978, "step": 14544 }, { "epoch": 2.34, "grad_norm": 12.354599952697754, "learning_rate": 2.422982630988492e-06, "loss": 0.5094, "step": 14545 }, { "epoch": 2.34, "grad_norm": 12.566329956054688, "learning_rate": 2.4218485307048457e-06, "loss": 0.8364, "step": 14546 }, { "epoch": 2.34, "grad_norm": 11.582292556762695, "learning_rate": 2.4207146593278254e-06, "loss": 0.6291, "step": 14547 }, { "epoch": 2.34, "grad_norm": 12.082343101501465, "learning_rate": 2.419581016891679e-06, "loss": 0.6261, "step": 14548 }, { "epoch": 2.34, "grad_norm": 12.161102294921875, "learning_rate": 2.418447603430647e-06, "loss": 0.6795, "step": 14549 }, { "epoch": 2.34, "grad_norm": 12.740278244018555, "learning_rate": 2.417314418978971e-06, "loss": 0.6023, "step": 14550 }, { "epoch": 2.34, "grad_norm": 11.188189506530762, "learning_rate": 2.416181463570875e-06, "loss": 0.5822, "step": 14551 }, { "epoch": 2.34, "grad_norm": 10.959596633911133, "learning_rate": 2.415048737240585e-06, "loss": 0.7755, "step": 14552 }, { "epoch": 2.34, "grad_norm": 10.498271942138672, "learning_rate": 2.4139162400223116e-06, "loss": 0.8734, "step": 14553 }, { "epoch": 2.34, "grad_norm": 10.56540584564209, "learning_rate": 2.4127839719502668e-06, "loss": 0.572, "step": 14554 }, { "epoch": 2.34, "grad_norm": 11.316020011901855, "learning_rate": 2.4116519330586542e-06, "loss": 0.5963, "step": 14555 }, { "epoch": 2.34, "grad_norm": 11.604574203491211, "learning_rate": 2.4105201233816657e-06, "loss": 0.8368, "step": 14556 }, { "epoch": 2.34, "grad_norm": 12.745898246765137, "learning_rate": 2.4093885429534846e-06, "loss": 0.6776, "step": 14557 }, { "epoch": 2.34, "grad_norm": 10.881851196289062, "learning_rate": 2.4082571918083e-06, "loss": 0.5775, "step": 14558 }, { "epoch": 2.34, "grad_norm": 12.346575736999512, "learning_rate": 2.40712606998028e-06, "loss": 0.7655, "step": 14559 }, { "epoch": 2.34, "grad_norm": 16.712257385253906, "learning_rate": 2.4059951775035904e-06, "loss": 0.7929, "step": 14560 }, { "epoch": 2.34, "grad_norm": 14.035721778869629, "learning_rate": 2.404864514412396e-06, "loss": 0.8154, "step": 14561 }, { "epoch": 2.34, "grad_norm": 14.842108726501465, "learning_rate": 2.403734080740845e-06, "loss": 0.8732, "step": 14562 }, { "epoch": 2.34, "grad_norm": 13.932010650634766, "learning_rate": 2.4026038765230875e-06, "loss": 0.7406, "step": 14563 }, { "epoch": 2.34, "grad_norm": 11.889596939086914, "learning_rate": 2.401473901793261e-06, "loss": 0.6509, "step": 14564 }, { "epoch": 2.34, "grad_norm": 13.175738334655762, "learning_rate": 2.4003441565854944e-06, "loss": 0.6134, "step": 14565 }, { "epoch": 2.34, "grad_norm": 12.118597030639648, "learning_rate": 2.399214640933919e-06, "loss": 0.7607, "step": 14566 }, { "epoch": 2.34, "grad_norm": 13.83961009979248, "learning_rate": 2.3980853548726468e-06, "loss": 0.6644, "step": 14567 }, { "epoch": 2.35, "grad_norm": 11.252650260925293, "learning_rate": 2.3969562984357952e-06, "loss": 0.6644, "step": 14568 }, { "epoch": 2.35, "grad_norm": 14.920836448669434, "learning_rate": 2.3958274716574647e-06, "loss": 0.6381, "step": 14569 }, { "epoch": 2.35, "grad_norm": 11.809601783752441, "learning_rate": 2.3946988745717524e-06, "loss": 0.5725, "step": 14570 }, { "epoch": 2.35, "grad_norm": 13.648818016052246, "learning_rate": 2.3935705072127523e-06, "loss": 0.667, "step": 14571 }, { "epoch": 2.35, "grad_norm": 11.92619514465332, "learning_rate": 2.392442369614547e-06, "loss": 0.8278, "step": 14572 }, { "epoch": 2.35, "grad_norm": 10.389835357666016, "learning_rate": 2.391314461811205e-06, "loss": 0.6807, "step": 14573 }, { "epoch": 2.35, "grad_norm": 13.882111549377441, "learning_rate": 2.3901867838368097e-06, "loss": 0.6647, "step": 14574 }, { "epoch": 2.35, "grad_norm": 11.934013366699219, "learning_rate": 2.3890593357254178e-06, "loss": 0.5912, "step": 14575 }, { "epoch": 2.35, "grad_norm": 12.152617454528809, "learning_rate": 2.38793211751108e-06, "loss": 0.6243, "step": 14576 }, { "epoch": 2.35, "grad_norm": 13.942753791809082, "learning_rate": 2.386805129227855e-06, "loss": 0.9683, "step": 14577 }, { "epoch": 2.35, "grad_norm": 10.798162460327148, "learning_rate": 2.385678370909775e-06, "loss": 0.5396, "step": 14578 }, { "epoch": 2.35, "grad_norm": 12.879944801330566, "learning_rate": 2.3845518425908843e-06, "loss": 0.7155, "step": 14579 }, { "epoch": 2.35, "grad_norm": 20.246660232543945, "learning_rate": 2.3834255443052044e-06, "loss": 0.6645, "step": 14580 }, { "epoch": 2.35, "grad_norm": 11.699856758117676, "learning_rate": 2.3822994760867568e-06, "loss": 0.7183, "step": 14581 }, { "epoch": 2.35, "grad_norm": 12.099349975585938, "learning_rate": 2.3811736379695594e-06, "loss": 0.6721, "step": 14582 }, { "epoch": 2.35, "grad_norm": 12.106697082519531, "learning_rate": 2.380048029987618e-06, "loss": 0.741, "step": 14583 }, { "epoch": 2.35, "grad_norm": 12.591113090515137, "learning_rate": 2.3789226521749286e-06, "loss": 0.7268, "step": 14584 }, { "epoch": 2.35, "grad_norm": 15.636309623718262, "learning_rate": 2.3777975045654912e-06, "loss": 0.6851, "step": 14585 }, { "epoch": 2.35, "grad_norm": 10.7044677734375, "learning_rate": 2.376672587193285e-06, "loss": 0.5764, "step": 14586 }, { "epoch": 2.35, "grad_norm": 11.081801414489746, "learning_rate": 2.3755479000922974e-06, "loss": 0.6063, "step": 14587 }, { "epoch": 2.35, "grad_norm": 12.59935474395752, "learning_rate": 2.3744234432964964e-06, "loss": 0.7816, "step": 14588 }, { "epoch": 2.35, "grad_norm": 12.810367584228516, "learning_rate": 2.373299216839845e-06, "loss": 0.5484, "step": 14589 }, { "epoch": 2.35, "grad_norm": 12.760719299316406, "learning_rate": 2.3721752207563075e-06, "loss": 0.6669, "step": 14590 }, { "epoch": 2.35, "grad_norm": 14.134513854980469, "learning_rate": 2.3710514550798293e-06, "loss": 0.73, "step": 14591 }, { "epoch": 2.35, "grad_norm": 11.81628704071045, "learning_rate": 2.3699279198443615e-06, "loss": 0.8282, "step": 14592 }, { "epoch": 2.35, "grad_norm": 13.015434265136719, "learning_rate": 2.3688046150838385e-06, "loss": 0.6457, "step": 14593 }, { "epoch": 2.35, "grad_norm": 12.386065483093262, "learning_rate": 2.367681540832187e-06, "loss": 0.7702, "step": 14594 }, { "epoch": 2.35, "grad_norm": 11.287806510925293, "learning_rate": 2.366558697123337e-06, "loss": 0.6928, "step": 14595 }, { "epoch": 2.35, "grad_norm": 12.811656951904297, "learning_rate": 2.365436083991204e-06, "loss": 0.7256, "step": 14596 }, { "epoch": 2.35, "grad_norm": 12.41495418548584, "learning_rate": 2.364313701469695e-06, "loss": 0.8525, "step": 14597 }, { "epoch": 2.35, "grad_norm": 10.625036239624023, "learning_rate": 2.363191549592716e-06, "loss": 0.5895, "step": 14598 }, { "epoch": 2.35, "grad_norm": 15.290607452392578, "learning_rate": 2.3620696283941623e-06, "loss": 0.727, "step": 14599 }, { "epoch": 2.35, "grad_norm": 15.304091453552246, "learning_rate": 2.3609479379079205e-06, "loss": 0.7285, "step": 14600 }, { "epoch": 2.35, "grad_norm": 13.965896606445312, "learning_rate": 2.3598264781678747e-06, "loss": 0.8667, "step": 14601 }, { "epoch": 2.35, "grad_norm": 13.633964538574219, "learning_rate": 2.3587052492078977e-06, "loss": 0.8716, "step": 14602 }, { "epoch": 2.35, "grad_norm": 13.550782203674316, "learning_rate": 2.357584251061862e-06, "loss": 0.9106, "step": 14603 }, { "epoch": 2.35, "grad_norm": 9.682012557983398, "learning_rate": 2.3564634837636248e-06, "loss": 0.5266, "step": 14604 }, { "epoch": 2.35, "grad_norm": 13.05029582977295, "learning_rate": 2.355342947347038e-06, "loss": 0.5709, "step": 14605 }, { "epoch": 2.35, "grad_norm": 10.310932159423828, "learning_rate": 2.354222641845955e-06, "loss": 0.675, "step": 14606 }, { "epoch": 2.35, "grad_norm": 13.073463439941406, "learning_rate": 2.353102567294212e-06, "loss": 0.58, "step": 14607 }, { "epoch": 2.35, "grad_norm": 12.144227981567383, "learning_rate": 2.3519827237256398e-06, "loss": 0.7352, "step": 14608 }, { "epoch": 2.35, "grad_norm": 10.931670188903809, "learning_rate": 2.350863111174071e-06, "loss": 0.8121, "step": 14609 }, { "epoch": 2.35, "grad_norm": 12.358681678771973, "learning_rate": 2.3497437296733173e-06, "loss": 0.7429, "step": 14610 }, { "epoch": 2.35, "grad_norm": 12.930477142333984, "learning_rate": 2.348624579257197e-06, "loss": 0.7848, "step": 14611 }, { "epoch": 2.35, "grad_norm": 12.713467597961426, "learning_rate": 2.3475056599595145e-06, "loss": 0.7951, "step": 14612 }, { "epoch": 2.35, "grad_norm": 12.591358184814453, "learning_rate": 2.346386971814063e-06, "loss": 0.6866, "step": 14613 }, { "epoch": 2.35, "grad_norm": 14.27762508392334, "learning_rate": 2.3452685148546373e-06, "loss": 0.6073, "step": 14614 }, { "epoch": 2.35, "grad_norm": 13.545770645141602, "learning_rate": 2.3441502891150236e-06, "loss": 0.6037, "step": 14615 }, { "epoch": 2.35, "grad_norm": 13.346944808959961, "learning_rate": 2.343032294628995e-06, "loss": 0.6095, "step": 14616 }, { "epoch": 2.35, "grad_norm": 11.305973052978516, "learning_rate": 2.341914531430327e-06, "loss": 0.6244, "step": 14617 }, { "epoch": 2.35, "grad_norm": 13.149638175964355, "learning_rate": 2.34079699955278e-06, "loss": 0.7552, "step": 14618 }, { "epoch": 2.35, "grad_norm": 11.431471824645996, "learning_rate": 2.3396796990301073e-06, "loss": 0.7259, "step": 14619 }, { "epoch": 2.35, "grad_norm": 14.634617805480957, "learning_rate": 2.338562629896064e-06, "loss": 0.6974, "step": 14620 }, { "epoch": 2.35, "grad_norm": 13.937216758728027, "learning_rate": 2.3374457921843863e-06, "loss": 0.5307, "step": 14621 }, { "epoch": 2.35, "grad_norm": 12.56019401550293, "learning_rate": 2.336329185928815e-06, "loss": 0.565, "step": 14622 }, { "epoch": 2.35, "grad_norm": 10.71655559539795, "learning_rate": 2.335212811163077e-06, "loss": 0.6769, "step": 14623 }, { "epoch": 2.35, "grad_norm": 11.749988555908203, "learning_rate": 2.3340966679208897e-06, "loss": 0.6108, "step": 14624 }, { "epoch": 2.35, "grad_norm": 13.46679401397705, "learning_rate": 2.332980756235974e-06, "loss": 0.6934, "step": 14625 }, { "epoch": 2.35, "grad_norm": 14.11075496673584, "learning_rate": 2.3318650761420335e-06, "loss": 0.8078, "step": 14626 }, { "epoch": 2.35, "grad_norm": 13.635009765625, "learning_rate": 2.3307496276727658e-06, "loss": 0.8199, "step": 14627 }, { "epoch": 2.35, "grad_norm": 10.842926025390625, "learning_rate": 2.329634410861872e-06, "loss": 0.7147, "step": 14628 }, { "epoch": 2.35, "grad_norm": 11.398038864135742, "learning_rate": 2.3285194257430286e-06, "loss": 0.8477, "step": 14629 }, { "epoch": 2.36, "grad_norm": 13.899744033813477, "learning_rate": 2.3274046723499254e-06, "loss": 0.8864, "step": 14630 }, { "epoch": 2.36, "grad_norm": 16.680158615112305, "learning_rate": 2.3262901507162283e-06, "loss": 0.5877, "step": 14631 }, { "epoch": 2.36, "grad_norm": 12.7188138961792, "learning_rate": 2.3251758608756024e-06, "loss": 0.8272, "step": 14632 }, { "epoch": 2.36, "grad_norm": 12.247641563415527, "learning_rate": 2.32406180286171e-06, "loss": 0.9816, "step": 14633 }, { "epoch": 2.36, "grad_norm": 11.127222061157227, "learning_rate": 2.3229479767081977e-06, "loss": 0.5955, "step": 14634 }, { "epoch": 2.36, "grad_norm": 12.190834999084473, "learning_rate": 2.3218343824487123e-06, "loss": 0.8008, "step": 14635 }, { "epoch": 2.36, "grad_norm": 15.192176818847656, "learning_rate": 2.320721020116894e-06, "loss": 0.8212, "step": 14636 }, { "epoch": 2.36, "grad_norm": 14.432473182678223, "learning_rate": 2.31960788974637e-06, "loss": 0.7396, "step": 14637 }, { "epoch": 2.36, "grad_norm": 12.362863540649414, "learning_rate": 2.318494991370762e-06, "loss": 0.7257, "step": 14638 }, { "epoch": 2.36, "grad_norm": 13.430099487304688, "learning_rate": 2.3173823250236917e-06, "loss": 0.6354, "step": 14639 }, { "epoch": 2.36, "grad_norm": 11.296304702758789, "learning_rate": 2.3162698907387627e-06, "loss": 0.7231, "step": 14640 }, { "epoch": 2.36, "grad_norm": 14.315195083618164, "learning_rate": 2.3151576885495818e-06, "loss": 0.8055, "step": 14641 }, { "epoch": 2.36, "grad_norm": 10.383102416992188, "learning_rate": 2.3140457184897424e-06, "loss": 0.5736, "step": 14642 }, { "epoch": 2.36, "grad_norm": 11.327825546264648, "learning_rate": 2.3129339805928307e-06, "loss": 0.718, "step": 14643 }, { "epoch": 2.36, "grad_norm": 11.862800598144531, "learning_rate": 2.3118224748924323e-06, "loss": 0.7605, "step": 14644 }, { "epoch": 2.36, "grad_norm": 15.92684555053711, "learning_rate": 2.310711201422117e-06, "loss": 0.8113, "step": 14645 }, { "epoch": 2.36, "grad_norm": 10.843137741088867, "learning_rate": 2.3096001602154573e-06, "loss": 0.675, "step": 14646 }, { "epoch": 2.36, "grad_norm": 12.566396713256836, "learning_rate": 2.30848935130601e-06, "loss": 0.723, "step": 14647 }, { "epoch": 2.36, "grad_norm": 9.569572448730469, "learning_rate": 2.3073787747273256e-06, "loss": 0.5907, "step": 14648 }, { "epoch": 2.36, "grad_norm": 10.904139518737793, "learning_rate": 2.306268430512958e-06, "loss": 0.6349, "step": 14649 }, { "epoch": 2.36, "grad_norm": 11.01569938659668, "learning_rate": 2.30515831869644e-06, "loss": 0.7813, "step": 14650 }, { "epoch": 2.36, "grad_norm": 12.935622215270996, "learning_rate": 2.304048439311305e-06, "loss": 0.6126, "step": 14651 }, { "epoch": 2.36, "grad_norm": 13.744715690612793, "learning_rate": 2.3029387923910807e-06, "loss": 0.6995, "step": 14652 }, { "epoch": 2.36, "grad_norm": 11.966907501220703, "learning_rate": 2.30182937796928e-06, "loss": 0.6966, "step": 14653 }, { "epoch": 2.36, "grad_norm": 12.715737342834473, "learning_rate": 2.300720196079421e-06, "loss": 0.5539, "step": 14654 }, { "epoch": 2.36, "grad_norm": 11.431225776672363, "learning_rate": 2.2996112467550014e-06, "loss": 0.7758, "step": 14655 }, { "epoch": 2.36, "grad_norm": 13.014755249023438, "learning_rate": 2.298502530029522e-06, "loss": 0.6565, "step": 14656 }, { "epoch": 2.36, "grad_norm": 13.138182640075684, "learning_rate": 2.297394045936473e-06, "loss": 0.6919, "step": 14657 }, { "epoch": 2.36, "grad_norm": 11.183084487915039, "learning_rate": 2.2962857945093377e-06, "loss": 0.6619, "step": 14658 }, { "epoch": 2.36, "grad_norm": 12.666234970092773, "learning_rate": 2.2951777757815874e-06, "loss": 0.7287, "step": 14659 }, { "epoch": 2.36, "grad_norm": 15.737833023071289, "learning_rate": 2.294069989786698e-06, "loss": 0.8518, "step": 14660 }, { "epoch": 2.36, "grad_norm": 11.035269737243652, "learning_rate": 2.292962436558127e-06, "loss": 0.567, "step": 14661 }, { "epoch": 2.36, "grad_norm": 11.998120307922363, "learning_rate": 2.291855116129328e-06, "loss": 0.6181, "step": 14662 }, { "epoch": 2.36, "grad_norm": 12.564933776855469, "learning_rate": 2.2907480285337537e-06, "loss": 0.4722, "step": 14663 }, { "epoch": 2.36, "grad_norm": 13.486348152160645, "learning_rate": 2.2896411738048395e-06, "loss": 0.7293, "step": 14664 }, { "epoch": 2.36, "grad_norm": 14.73842716217041, "learning_rate": 2.288534551976025e-06, "loss": 0.6794, "step": 14665 }, { "epoch": 2.36, "grad_norm": 12.539525985717773, "learning_rate": 2.2874281630807325e-06, "loss": 0.7665, "step": 14666 }, { "epoch": 2.36, "grad_norm": 11.212377548217773, "learning_rate": 2.2863220071523816e-06, "loss": 0.4678, "step": 14667 }, { "epoch": 2.36, "grad_norm": 9.66005802154541, "learning_rate": 2.2852160842243897e-06, "loss": 0.6962, "step": 14668 }, { "epoch": 2.36, "grad_norm": 13.516281127929688, "learning_rate": 2.2841103943301557e-06, "loss": 0.8108, "step": 14669 }, { "epoch": 2.36, "grad_norm": 13.1104736328125, "learning_rate": 2.283004937503085e-06, "loss": 0.6031, "step": 14670 }, { "epoch": 2.36, "grad_norm": 10.521419525146484, "learning_rate": 2.281899713776565e-06, "loss": 0.8041, "step": 14671 }, { "epoch": 2.36, "grad_norm": 11.426438331604004, "learning_rate": 2.28079472318398e-06, "loss": 0.7048, "step": 14672 }, { "epoch": 2.36, "grad_norm": 12.057008743286133, "learning_rate": 2.27968996575871e-06, "loss": 0.6175, "step": 14673 }, { "epoch": 2.36, "grad_norm": 11.52430534362793, "learning_rate": 2.278585441534126e-06, "loss": 0.6127, "step": 14674 }, { "epoch": 2.36, "grad_norm": 13.216154098510742, "learning_rate": 2.2774811505435824e-06, "loss": 0.6795, "step": 14675 }, { "epoch": 2.36, "grad_norm": 10.532110214233398, "learning_rate": 2.2763770928204487e-06, "loss": 0.6468, "step": 14676 }, { "epoch": 2.36, "grad_norm": 13.425366401672363, "learning_rate": 2.2752732683980683e-06, "loss": 0.6568, "step": 14677 }, { "epoch": 2.36, "grad_norm": 11.615131378173828, "learning_rate": 2.27416967730978e-06, "loss": 0.5179, "step": 14678 }, { "epoch": 2.36, "grad_norm": 12.540204048156738, "learning_rate": 2.2730663195889257e-06, "loss": 0.7181, "step": 14679 }, { "epoch": 2.36, "grad_norm": 9.880234718322754, "learning_rate": 2.271963195268827e-06, "loss": 0.5485, "step": 14680 }, { "epoch": 2.36, "grad_norm": 11.354337692260742, "learning_rate": 2.2708603043828116e-06, "loss": 0.7524, "step": 14681 }, { "epoch": 2.36, "grad_norm": 11.530858039855957, "learning_rate": 2.26975764696419e-06, "loss": 0.5299, "step": 14682 }, { "epoch": 2.36, "grad_norm": 12.238303184509277, "learning_rate": 2.2686552230462678e-06, "loss": 0.8547, "step": 14683 }, { "epoch": 2.36, "grad_norm": 13.640875816345215, "learning_rate": 2.2675530326623476e-06, "loss": 0.7468, "step": 14684 }, { "epoch": 2.36, "grad_norm": 11.421207427978516, "learning_rate": 2.2664510758457225e-06, "loss": 0.5367, "step": 14685 }, { "epoch": 2.36, "grad_norm": 12.321694374084473, "learning_rate": 2.2653493526296733e-06, "loss": 0.5193, "step": 14686 }, { "epoch": 2.36, "grad_norm": 11.9963960647583, "learning_rate": 2.2642478630474863e-06, "loss": 0.8058, "step": 14687 }, { "epoch": 2.36, "grad_norm": 11.652522087097168, "learning_rate": 2.263146607132427e-06, "loss": 0.7212, "step": 14688 }, { "epoch": 2.36, "grad_norm": 13.623882293701172, "learning_rate": 2.2620455849177647e-06, "loss": 0.6068, "step": 14689 }, { "epoch": 2.36, "grad_norm": 11.680227279663086, "learning_rate": 2.260944796436755e-06, "loss": 0.7161, "step": 14690 }, { "epoch": 2.36, "grad_norm": 12.487935066223145, "learning_rate": 2.2598442417226464e-06, "loss": 0.8138, "step": 14691 }, { "epoch": 2.37, "grad_norm": 10.490584373474121, "learning_rate": 2.258743920808687e-06, "loss": 0.5476, "step": 14692 }, { "epoch": 2.37, "grad_norm": 14.705185890197754, "learning_rate": 2.2576438337281102e-06, "loss": 0.7244, "step": 14693 }, { "epoch": 2.37, "grad_norm": 14.486978530883789, "learning_rate": 2.2565439805141433e-06, "loss": 0.6131, "step": 14694 }, { "epoch": 2.37, "grad_norm": 10.975566864013672, "learning_rate": 2.255444361200011e-06, "loss": 0.7998, "step": 14695 }, { "epoch": 2.37, "grad_norm": 11.811187744140625, "learning_rate": 2.2543449758189317e-06, "loss": 0.7675, "step": 14696 }, { "epoch": 2.37, "grad_norm": 11.387557983398438, "learning_rate": 2.2532458244041076e-06, "loss": 0.5794, "step": 14697 }, { "epoch": 2.37, "grad_norm": 13.514288902282715, "learning_rate": 2.2521469069887468e-06, "loss": 0.6317, "step": 14698 }, { "epoch": 2.37, "grad_norm": 12.515429496765137, "learning_rate": 2.251048223606035e-06, "loss": 0.7313, "step": 14699 }, { "epoch": 2.37, "grad_norm": 13.149578094482422, "learning_rate": 2.2499497742891675e-06, "loss": 0.6403, "step": 14700 }, { "epoch": 2.37, "grad_norm": 12.348041534423828, "learning_rate": 2.248851559071319e-06, "loss": 0.8669, "step": 14701 }, { "epoch": 2.37, "grad_norm": 14.254931449890137, "learning_rate": 2.2477535779856617e-06, "loss": 0.8071, "step": 14702 }, { "epoch": 2.37, "grad_norm": 14.128544807434082, "learning_rate": 2.246655831065365e-06, "loss": 0.6736, "step": 14703 }, { "epoch": 2.37, "grad_norm": 12.52774429321289, "learning_rate": 2.245558318343587e-06, "loss": 0.8478, "step": 14704 }, { "epoch": 2.37, "grad_norm": 14.16899299621582, "learning_rate": 2.244461039853475e-06, "loss": 0.6818, "step": 14705 }, { "epoch": 2.37, "grad_norm": 12.796497344970703, "learning_rate": 2.243363995628178e-06, "loss": 0.6638, "step": 14706 }, { "epoch": 2.37, "grad_norm": 14.72799015045166, "learning_rate": 2.2422671857008304e-06, "loss": 0.7646, "step": 14707 }, { "epoch": 2.37, "grad_norm": 14.426480293273926, "learning_rate": 2.241170610104568e-06, "loss": 0.6378, "step": 14708 }, { "epoch": 2.37, "grad_norm": 12.453461647033691, "learning_rate": 2.2400742688725098e-06, "loss": 0.6805, "step": 14709 }, { "epoch": 2.37, "grad_norm": 11.489238739013672, "learning_rate": 2.2389781620377695e-06, "loss": 0.7911, "step": 14710 }, { "epoch": 2.37, "grad_norm": 12.858954429626465, "learning_rate": 2.237882289633462e-06, "loss": 0.7316, "step": 14711 }, { "epoch": 2.37, "grad_norm": 13.531366348266602, "learning_rate": 2.236786651692684e-06, "loss": 0.6855, "step": 14712 }, { "epoch": 2.37, "grad_norm": 11.821423530578613, "learning_rate": 2.235691248248536e-06, "loss": 0.5548, "step": 14713 }, { "epoch": 2.37, "grad_norm": 9.948687553405762, "learning_rate": 2.2345960793341024e-06, "loss": 0.7151, "step": 14714 }, { "epoch": 2.37, "grad_norm": 13.949053764343262, "learning_rate": 2.2335011449824627e-06, "loss": 0.7934, "step": 14715 }, { "epoch": 2.37, "grad_norm": 13.488882064819336, "learning_rate": 2.2324064452266927e-06, "loss": 0.8753, "step": 14716 }, { "epoch": 2.37, "grad_norm": 12.080770492553711, "learning_rate": 2.2313119800998618e-06, "loss": 0.5669, "step": 14717 }, { "epoch": 2.37, "grad_norm": 12.673301696777344, "learning_rate": 2.230217749635024e-06, "loss": 0.6968, "step": 14718 }, { "epoch": 2.37, "grad_norm": 13.623655319213867, "learning_rate": 2.2291237538652366e-06, "loss": 0.7923, "step": 14719 }, { "epoch": 2.37, "grad_norm": 10.142783164978027, "learning_rate": 2.228029992823544e-06, "loss": 0.5688, "step": 14720 }, { "epoch": 2.37, "grad_norm": 13.281983375549316, "learning_rate": 2.2269364665429805e-06, "loss": 0.6865, "step": 14721 }, { "epoch": 2.37, "grad_norm": 12.845333099365234, "learning_rate": 2.225843175056582e-06, "loss": 0.7655, "step": 14722 }, { "epoch": 2.37, "grad_norm": 15.2527437210083, "learning_rate": 2.2247501183973685e-06, "loss": 0.7243, "step": 14723 }, { "epoch": 2.37, "grad_norm": 11.96027946472168, "learning_rate": 2.2236572965983615e-06, "loss": 0.6518, "step": 14724 }, { "epoch": 2.37, "grad_norm": 11.259908676147461, "learning_rate": 2.2225647096925696e-06, "loss": 0.5602, "step": 14725 }, { "epoch": 2.37, "grad_norm": 15.23027515411377, "learning_rate": 2.2214723577129915e-06, "loss": 0.9106, "step": 14726 }, { "epoch": 2.37, "grad_norm": 11.04971981048584, "learning_rate": 2.220380240692629e-06, "loss": 0.6504, "step": 14727 }, { "epoch": 2.37, "grad_norm": 11.680681228637695, "learning_rate": 2.2192883586644677e-06, "loss": 0.5589, "step": 14728 }, { "epoch": 2.37, "grad_norm": 12.123357772827148, "learning_rate": 2.218196711661487e-06, "loss": 0.8208, "step": 14729 }, { "epoch": 2.37, "grad_norm": 12.879476547241211, "learning_rate": 2.2171052997166654e-06, "loss": 0.8137, "step": 14730 }, { "epoch": 2.37, "grad_norm": 17.643125534057617, "learning_rate": 2.2160141228629663e-06, "loss": 0.5924, "step": 14731 }, { "epoch": 2.37, "grad_norm": 14.094585418701172, "learning_rate": 2.214923181133355e-06, "loss": 0.6985, "step": 14732 }, { "epoch": 2.37, "grad_norm": 19.096263885498047, "learning_rate": 2.213832474560782e-06, "loss": 0.8162, "step": 14733 }, { "epoch": 2.37, "grad_norm": 15.8463773727417, "learning_rate": 2.21274200317819e-06, "loss": 0.8486, "step": 14734 }, { "epoch": 2.37, "grad_norm": 13.456482887268066, "learning_rate": 2.2116517670185233e-06, "loss": 0.7239, "step": 14735 }, { "epoch": 2.37, "grad_norm": 14.863273620605469, "learning_rate": 2.2105617661147082e-06, "loss": 0.8896, "step": 14736 }, { "epoch": 2.37, "grad_norm": 12.60537338256836, "learning_rate": 2.209472000499674e-06, "loss": 0.7328, "step": 14737 }, { "epoch": 2.37, "grad_norm": 13.296514511108398, "learning_rate": 2.2083824702063385e-06, "loss": 0.4821, "step": 14738 }, { "epoch": 2.37, "grad_norm": 12.739935874938965, "learning_rate": 2.2072931752676107e-06, "loss": 0.6203, "step": 14739 }, { "epoch": 2.37, "grad_norm": 13.12365436553955, "learning_rate": 2.2062041157163916e-06, "loss": 0.6413, "step": 14740 }, { "epoch": 2.37, "grad_norm": 14.786291122436523, "learning_rate": 2.205115291585582e-06, "loss": 0.6704, "step": 14741 }, { "epoch": 2.37, "grad_norm": 10.741128921508789, "learning_rate": 2.2040267029080664e-06, "loss": 0.7209, "step": 14742 }, { "epoch": 2.37, "grad_norm": 12.569857597351074, "learning_rate": 2.2029383497167324e-06, "loss": 0.6848, "step": 14743 }, { "epoch": 2.37, "grad_norm": 14.630379676818848, "learning_rate": 2.2018502320444514e-06, "loss": 0.717, "step": 14744 }, { "epoch": 2.37, "grad_norm": 11.270121574401855, "learning_rate": 2.2007623499240895e-06, "loss": 0.7146, "step": 14745 }, { "epoch": 2.37, "grad_norm": 10.38469409942627, "learning_rate": 2.1996747033885113e-06, "loss": 0.5018, "step": 14746 }, { "epoch": 2.37, "grad_norm": 15.417118072509766, "learning_rate": 2.1985872924705654e-06, "loss": 0.7381, "step": 14747 }, { "epoch": 2.37, "grad_norm": 11.997312545776367, "learning_rate": 2.197500117203105e-06, "loss": 0.5976, "step": 14748 }, { "epoch": 2.37, "grad_norm": 12.808249473571777, "learning_rate": 2.196413177618967e-06, "loss": 0.558, "step": 14749 }, { "epoch": 2.37, "grad_norm": 10.622509956359863, "learning_rate": 2.1953264737509783e-06, "loss": 0.4222, "step": 14750 }, { "epoch": 2.37, "grad_norm": 10.723481178283691, "learning_rate": 2.194240005631971e-06, "loss": 0.6248, "step": 14751 }, { "epoch": 2.37, "grad_norm": 12.928491592407227, "learning_rate": 2.1931537732947615e-06, "loss": 0.7349, "step": 14752 }, { "epoch": 2.37, "grad_norm": 12.108513832092285, "learning_rate": 2.1920677767721554e-06, "loss": 1.0182, "step": 14753 }, { "epoch": 2.38, "grad_norm": 13.561178207397461, "learning_rate": 2.190982016096964e-06, "loss": 0.6575, "step": 14754 }, { "epoch": 2.38, "grad_norm": 12.326555252075195, "learning_rate": 2.1898964913019783e-06, "loss": 0.8053, "step": 14755 }, { "epoch": 2.38, "grad_norm": 14.517414093017578, "learning_rate": 2.18881120241999e-06, "loss": 0.7076, "step": 14756 }, { "epoch": 2.38, "grad_norm": 17.057884216308594, "learning_rate": 2.1877261494837843e-06, "loss": 0.7045, "step": 14757 }, { "epoch": 2.38, "grad_norm": 14.072837829589844, "learning_rate": 2.1866413325261302e-06, "loss": 0.6143, "step": 14758 }, { "epoch": 2.38, "grad_norm": 11.935178756713867, "learning_rate": 2.1855567515798027e-06, "loss": 0.6241, "step": 14759 }, { "epoch": 2.38, "grad_norm": 12.43177604675293, "learning_rate": 2.184472406677559e-06, "loss": 0.9616, "step": 14760 }, { "epoch": 2.38, "grad_norm": 12.485292434692383, "learning_rate": 2.1833882978521502e-06, "loss": 0.6581, "step": 14761 }, { "epoch": 2.38, "grad_norm": 12.487054824829102, "learning_rate": 2.18230442513633e-06, "loss": 0.814, "step": 14762 }, { "epoch": 2.38, "grad_norm": 14.018542289733887, "learning_rate": 2.1812207885628335e-06, "loss": 0.6678, "step": 14763 }, { "epoch": 2.38, "grad_norm": 14.307856559753418, "learning_rate": 2.1801373881643915e-06, "loss": 0.5651, "step": 14764 }, { "epoch": 2.38, "grad_norm": 11.176826477050781, "learning_rate": 2.1790542239737343e-06, "loss": 0.7061, "step": 14765 }, { "epoch": 2.38, "grad_norm": 11.761308670043945, "learning_rate": 2.1779712960235744e-06, "loss": 0.5387, "step": 14766 }, { "epoch": 2.38, "grad_norm": 11.245362281799316, "learning_rate": 2.1768886043466296e-06, "loss": 0.5492, "step": 14767 }, { "epoch": 2.38, "grad_norm": 10.980292320251465, "learning_rate": 2.175806148975599e-06, "loss": 0.6281, "step": 14768 }, { "epoch": 2.38, "grad_norm": 12.898702621459961, "learning_rate": 2.174723929943179e-06, "loss": 0.8237, "step": 14769 }, { "epoch": 2.38, "grad_norm": 14.567761421203613, "learning_rate": 2.1736419472820624e-06, "loss": 0.7633, "step": 14770 }, { "epoch": 2.38, "grad_norm": 13.171661376953125, "learning_rate": 2.1725602010249304e-06, "loss": 0.7367, "step": 14771 }, { "epoch": 2.38, "grad_norm": 10.827173233032227, "learning_rate": 2.171478691204456e-06, "loss": 0.7704, "step": 14772 }, { "epoch": 2.38, "grad_norm": 10.683528900146484, "learning_rate": 2.1703974178533114e-06, "loss": 0.5633, "step": 14773 }, { "epoch": 2.38, "grad_norm": 11.679052352905273, "learning_rate": 2.1693163810041527e-06, "loss": 0.5704, "step": 14774 }, { "epoch": 2.38, "grad_norm": 16.11127281188965, "learning_rate": 2.1682355806896404e-06, "loss": 0.6578, "step": 14775 }, { "epoch": 2.38, "grad_norm": 14.025016784667969, "learning_rate": 2.167155016942416e-06, "loss": 0.7429, "step": 14776 }, { "epoch": 2.38, "grad_norm": 13.50238037109375, "learning_rate": 2.1660746897951203e-06, "loss": 0.7706, "step": 14777 }, { "epoch": 2.38, "grad_norm": 13.3353271484375, "learning_rate": 2.164994599280388e-06, "loss": 0.601, "step": 14778 }, { "epoch": 2.38, "grad_norm": 12.1929349899292, "learning_rate": 2.163914745430844e-06, "loss": 0.6354, "step": 14779 }, { "epoch": 2.38, "grad_norm": 14.76216983795166, "learning_rate": 2.1628351282791027e-06, "loss": 0.635, "step": 14780 }, { "epoch": 2.38, "grad_norm": 13.400191307067871, "learning_rate": 2.16175574785778e-06, "loss": 0.8187, "step": 14781 }, { "epoch": 2.38, "grad_norm": 12.694388389587402, "learning_rate": 2.1606766041994785e-06, "loss": 0.7678, "step": 14782 }, { "epoch": 2.38, "grad_norm": 12.235447883605957, "learning_rate": 2.1595976973367918e-06, "loss": 0.6457, "step": 14783 }, { "epoch": 2.38, "grad_norm": 14.1260986328125, "learning_rate": 2.1585190273023127e-06, "loss": 0.5144, "step": 14784 }, { "epoch": 2.38, "grad_norm": 15.56763744354248, "learning_rate": 2.1574405941286215e-06, "loss": 0.6826, "step": 14785 }, { "epoch": 2.38, "grad_norm": 11.528478622436523, "learning_rate": 2.1563623978482972e-06, "loss": 0.5729, "step": 14786 }, { "epoch": 2.38, "grad_norm": 13.10938835144043, "learning_rate": 2.1552844384939054e-06, "loss": 0.813, "step": 14787 }, { "epoch": 2.38, "grad_norm": 11.810047149658203, "learning_rate": 2.154206716098005e-06, "loss": 0.6896, "step": 14788 }, { "epoch": 2.38, "grad_norm": 12.028420448303223, "learning_rate": 2.153129230693154e-06, "loss": 0.7662, "step": 14789 }, { "epoch": 2.38, "grad_norm": 12.58189868927002, "learning_rate": 2.1520519823118947e-06, "loss": 0.8611, "step": 14790 }, { "epoch": 2.38, "grad_norm": 14.203292846679688, "learning_rate": 2.150974970986772e-06, "loss": 0.5717, "step": 14791 }, { "epoch": 2.38, "grad_norm": 12.755772590637207, "learning_rate": 2.149898196750315e-06, "loss": 0.7298, "step": 14792 }, { "epoch": 2.38, "grad_norm": 12.826224327087402, "learning_rate": 2.1488216596350464e-06, "loss": 0.8783, "step": 14793 }, { "epoch": 2.38, "grad_norm": 11.602682113647461, "learning_rate": 2.1477453596734887e-06, "loss": 0.8084, "step": 14794 }, { "epoch": 2.38, "grad_norm": 13.26528263092041, "learning_rate": 2.146669296898153e-06, "loss": 0.659, "step": 14795 }, { "epoch": 2.38, "grad_norm": 11.737380981445312, "learning_rate": 2.145593471341534e-06, "loss": 0.6534, "step": 14796 }, { "epoch": 2.38, "grad_norm": 12.465666770935059, "learning_rate": 2.1445178830361413e-06, "loss": 0.6035, "step": 14797 }, { "epoch": 2.38, "grad_norm": 13.177637100219727, "learning_rate": 2.143442532014457e-06, "loss": 0.7324, "step": 14798 }, { "epoch": 2.38, "grad_norm": 12.586277961730957, "learning_rate": 2.1423674183089627e-06, "loss": 0.5922, "step": 14799 }, { "epoch": 2.38, "grad_norm": 12.071619987487793, "learning_rate": 2.141292541952138e-06, "loss": 0.8349, "step": 14800 }, { "epoch": 2.38, "grad_norm": 11.980542182922363, "learning_rate": 2.140217902976446e-06, "loss": 0.7373, "step": 14801 }, { "epoch": 2.38, "grad_norm": 12.026190757751465, "learning_rate": 2.1391435014143504e-06, "loss": 0.6273, "step": 14802 }, { "epoch": 2.38, "grad_norm": 11.967222213745117, "learning_rate": 2.138069337298304e-06, "loss": 0.6105, "step": 14803 }, { "epoch": 2.38, "grad_norm": 13.39240550994873, "learning_rate": 2.1369954106607506e-06, "loss": 0.5401, "step": 14804 }, { "epoch": 2.38, "grad_norm": 12.980944633483887, "learning_rate": 2.1359217215341344e-06, "loss": 0.5871, "step": 14805 }, { "epoch": 2.38, "grad_norm": 11.469355583190918, "learning_rate": 2.1348482699508842e-06, "loss": 0.6429, "step": 14806 }, { "epoch": 2.38, "grad_norm": 13.350205421447754, "learning_rate": 2.1337750559434224e-06, "loss": 0.7523, "step": 14807 }, { "epoch": 2.38, "grad_norm": 13.225598335266113, "learning_rate": 2.132702079544172e-06, "loss": 0.5225, "step": 14808 }, { "epoch": 2.38, "grad_norm": 13.89864730834961, "learning_rate": 2.131629340785538e-06, "loss": 0.7611, "step": 14809 }, { "epoch": 2.38, "grad_norm": 13.761798858642578, "learning_rate": 2.13055683969993e-06, "loss": 0.6803, "step": 14810 }, { "epoch": 2.38, "grad_norm": 11.914122581481934, "learning_rate": 2.12948457631974e-06, "loss": 0.7944, "step": 14811 }, { "epoch": 2.38, "grad_norm": 11.279928207397461, "learning_rate": 2.1284125506773547e-06, "loss": 0.7003, "step": 14812 }, { "epoch": 2.38, "grad_norm": 14.47415542602539, "learning_rate": 2.127340762805162e-06, "loss": 0.7217, "step": 14813 }, { "epoch": 2.38, "grad_norm": 12.898543357849121, "learning_rate": 2.12626921273553e-06, "loss": 0.7416, "step": 14814 }, { "epoch": 2.38, "grad_norm": 23.5933780670166, "learning_rate": 2.125197900500833e-06, "loss": 0.7606, "step": 14815 }, { "epoch": 2.39, "grad_norm": 12.35644245147705, "learning_rate": 2.1241268261334257e-06, "loss": 0.7616, "step": 14816 }, { "epoch": 2.39, "grad_norm": 12.1074800491333, "learning_rate": 2.123055989665661e-06, "loss": 0.63, "step": 14817 }, { "epoch": 2.39, "grad_norm": 14.104744911193848, "learning_rate": 2.1219853911298872e-06, "loss": 0.537, "step": 14818 }, { "epoch": 2.39, "grad_norm": 15.133925437927246, "learning_rate": 2.1209150305584446e-06, "loss": 0.8541, "step": 14819 }, { "epoch": 2.39, "grad_norm": 11.909462928771973, "learning_rate": 2.1198449079836593e-06, "loss": 0.6913, "step": 14820 }, { "epoch": 2.39, "grad_norm": 11.465744018554688, "learning_rate": 2.1187750234378623e-06, "loss": 0.9161, "step": 14821 }, { "epoch": 2.39, "grad_norm": 12.452245712280273, "learning_rate": 2.117705376953366e-06, "loss": 0.6311, "step": 14822 }, { "epoch": 2.39, "grad_norm": 12.343655586242676, "learning_rate": 2.1166359685624792e-06, "loss": 0.4694, "step": 14823 }, { "epoch": 2.39, "grad_norm": 11.353242874145508, "learning_rate": 2.115566798297509e-06, "loss": 0.7197, "step": 14824 }, { "epoch": 2.39, "grad_norm": 13.167218208312988, "learning_rate": 2.1144978661907454e-06, "loss": 0.7289, "step": 14825 }, { "epoch": 2.39, "grad_norm": 11.877285957336426, "learning_rate": 2.113429172274484e-06, "loss": 0.7899, "step": 14826 }, { "epoch": 2.39, "grad_norm": 12.324966430664062, "learning_rate": 2.112360716581001e-06, "loss": 0.6567, "step": 14827 }, { "epoch": 2.39, "grad_norm": 10.623032569885254, "learning_rate": 2.1112924991425675e-06, "loss": 0.8881, "step": 14828 }, { "epoch": 2.39, "grad_norm": 13.552242279052734, "learning_rate": 2.1102245199914582e-06, "loss": 0.6898, "step": 14829 }, { "epoch": 2.39, "grad_norm": 12.595361709594727, "learning_rate": 2.1091567791599275e-06, "loss": 0.5819, "step": 14830 }, { "epoch": 2.39, "grad_norm": 11.870534896850586, "learning_rate": 2.1080892766802253e-06, "loss": 0.5756, "step": 14831 }, { "epoch": 2.39, "grad_norm": 20.408897399902344, "learning_rate": 2.107022012584603e-06, "loss": 0.7979, "step": 14832 }, { "epoch": 2.39, "grad_norm": 13.568726539611816, "learning_rate": 2.105954986905293e-06, "loss": 0.7317, "step": 14833 }, { "epoch": 2.39, "grad_norm": 10.617545127868652, "learning_rate": 2.104888199674531e-06, "loss": 0.6034, "step": 14834 }, { "epoch": 2.39, "grad_norm": 13.152807235717773, "learning_rate": 2.103821650924537e-06, "loss": 0.6192, "step": 14835 }, { "epoch": 2.39, "grad_norm": 9.379603385925293, "learning_rate": 2.1027553406875254e-06, "loss": 0.4565, "step": 14836 }, { "epoch": 2.39, "grad_norm": 13.90480899810791, "learning_rate": 2.101689268995709e-06, "loss": 0.6039, "step": 14837 }, { "epoch": 2.39, "grad_norm": 14.907737731933594, "learning_rate": 2.1006234358812915e-06, "loss": 0.5543, "step": 14838 }, { "epoch": 2.39, "grad_norm": 13.337625503540039, "learning_rate": 2.099557841376463e-06, "loss": 0.5365, "step": 14839 }, { "epoch": 2.39, "grad_norm": 12.554641723632812, "learning_rate": 2.0984924855134147e-06, "loss": 0.7329, "step": 14840 }, { "epoch": 2.39, "grad_norm": 10.908984184265137, "learning_rate": 2.0974273683243253e-06, "loss": 0.5917, "step": 14841 }, { "epoch": 2.39, "grad_norm": 10.493473052978516, "learning_rate": 2.096362489841364e-06, "loss": 0.8983, "step": 14842 }, { "epoch": 2.39, "grad_norm": 16.296733856201172, "learning_rate": 2.0952978500967038e-06, "loss": 0.5351, "step": 14843 }, { "epoch": 2.39, "grad_norm": 14.951189994812012, "learning_rate": 2.094233449122497e-06, "loss": 0.7562, "step": 14844 }, { "epoch": 2.39, "grad_norm": 13.729289054870605, "learning_rate": 2.093169286950901e-06, "loss": 0.8082, "step": 14845 }, { "epoch": 2.39, "grad_norm": 10.401632308959961, "learning_rate": 2.0921053636140565e-06, "loss": 0.6646, "step": 14846 }, { "epoch": 2.39, "grad_norm": 14.59920883178711, "learning_rate": 2.0910416791440977e-06, "loss": 0.7977, "step": 14847 }, { "epoch": 2.39, "grad_norm": 12.133580207824707, "learning_rate": 2.089978233573161e-06, "loss": 0.8068, "step": 14848 }, { "epoch": 2.39, "grad_norm": 10.742266654968262, "learning_rate": 2.0889150269333656e-06, "loss": 0.529, "step": 14849 }, { "epoch": 2.39, "grad_norm": 11.85191535949707, "learning_rate": 2.087852059256824e-06, "loss": 0.7592, "step": 14850 }, { "epoch": 2.39, "grad_norm": 13.012702941894531, "learning_rate": 2.0867893305756505e-06, "loss": 0.7258, "step": 14851 }, { "epoch": 2.39, "grad_norm": 11.585708618164062, "learning_rate": 2.0857268409219388e-06, "loss": 0.6503, "step": 14852 }, { "epoch": 2.39, "grad_norm": 14.32262134552002, "learning_rate": 2.0846645903277897e-06, "loss": 0.5921, "step": 14853 }, { "epoch": 2.39, "grad_norm": 12.995149612426758, "learning_rate": 2.0836025788252867e-06, "loss": 0.8138, "step": 14854 }, { "epoch": 2.39, "grad_norm": 13.007683753967285, "learning_rate": 2.0825408064465058e-06, "loss": 0.8208, "step": 14855 }, { "epoch": 2.39, "grad_norm": 13.023737907409668, "learning_rate": 2.0814792732235256e-06, "loss": 0.7269, "step": 14856 }, { "epoch": 2.39, "grad_norm": 11.246078491210938, "learning_rate": 2.080417979188404e-06, "loss": 0.7088, "step": 14857 }, { "epoch": 2.39, "grad_norm": 13.186361312866211, "learning_rate": 2.0793569243732014e-06, "loss": 0.5867, "step": 14858 }, { "epoch": 2.39, "grad_norm": 11.554381370544434, "learning_rate": 2.078296108809972e-06, "loss": 0.7573, "step": 14859 }, { "epoch": 2.39, "grad_norm": 12.96829605102539, "learning_rate": 2.077235532530756e-06, "loss": 0.6823, "step": 14860 }, { "epoch": 2.39, "grad_norm": 12.417482376098633, "learning_rate": 2.076175195567586e-06, "loss": 0.5652, "step": 14861 }, { "epoch": 2.39, "grad_norm": 10.084071159362793, "learning_rate": 2.075115097952496e-06, "loss": 0.6256, "step": 14862 }, { "epoch": 2.39, "grad_norm": 12.987436294555664, "learning_rate": 2.074055239717503e-06, "loss": 0.7856, "step": 14863 }, { "epoch": 2.39, "grad_norm": 11.261073112487793, "learning_rate": 2.0729956208946257e-06, "loss": 0.7227, "step": 14864 }, { "epoch": 2.39, "grad_norm": 11.331233978271484, "learning_rate": 2.0719362415158686e-06, "loss": 0.7074, "step": 14865 }, { "epoch": 2.39, "grad_norm": 12.74025821685791, "learning_rate": 2.070877101613228e-06, "loss": 0.8443, "step": 14866 }, { "epoch": 2.39, "grad_norm": 13.352799415588379, "learning_rate": 2.069818201218703e-06, "loss": 0.7415, "step": 14867 }, { "epoch": 2.39, "grad_norm": 11.217409133911133, "learning_rate": 2.0687595403642735e-06, "loss": 0.8024, "step": 14868 }, { "epoch": 2.39, "grad_norm": 11.279478073120117, "learning_rate": 2.067701119081923e-06, "loss": 0.7291, "step": 14869 }, { "epoch": 2.39, "grad_norm": 13.510546684265137, "learning_rate": 2.066642937403619e-06, "loss": 0.621, "step": 14870 }, { "epoch": 2.39, "grad_norm": 13.833948135375977, "learning_rate": 2.0655849953613215e-06, "loss": 0.6285, "step": 14871 }, { "epoch": 2.39, "grad_norm": 11.435037612915039, "learning_rate": 2.064527292986995e-06, "loss": 0.7173, "step": 14872 }, { "epoch": 2.39, "grad_norm": 12.669169425964355, "learning_rate": 2.0634698303125834e-06, "loss": 0.7013, "step": 14873 }, { "epoch": 2.39, "grad_norm": 12.086406707763672, "learning_rate": 2.0624126073700255e-06, "loss": 0.8061, "step": 14874 }, { "epoch": 2.39, "grad_norm": 13.057968139648438, "learning_rate": 2.0613556241912645e-06, "loss": 0.7939, "step": 14875 }, { "epoch": 2.39, "grad_norm": 14.690372467041016, "learning_rate": 2.060298880808219e-06, "loss": 0.9127, "step": 14876 }, { "epoch": 2.39, "grad_norm": 12.683839797973633, "learning_rate": 2.0592423772528135e-06, "loss": 0.762, "step": 14877 }, { "epoch": 2.4, "grad_norm": 17.681547164916992, "learning_rate": 2.0581861135569627e-06, "loss": 0.8177, "step": 14878 }, { "epoch": 2.4, "grad_norm": 13.844377517700195, "learning_rate": 2.057130089752568e-06, "loss": 0.8741, "step": 14879 }, { "epoch": 2.4, "grad_norm": 12.162461280822754, "learning_rate": 2.0560743058715325e-06, "loss": 0.6202, "step": 14880 }, { "epoch": 2.4, "grad_norm": 12.356973648071289, "learning_rate": 2.0550187619457453e-06, "loss": 0.7203, "step": 14881 }, { "epoch": 2.4, "grad_norm": 13.185091018676758, "learning_rate": 2.0539634580070854e-06, "loss": 0.7375, "step": 14882 }, { "epoch": 2.4, "grad_norm": 11.6066312789917, "learning_rate": 2.052908394087437e-06, "loss": 0.8815, "step": 14883 }, { "epoch": 2.4, "grad_norm": 11.738616943359375, "learning_rate": 2.051853570218667e-06, "loss": 0.7523, "step": 14884 }, { "epoch": 2.4, "grad_norm": 12.036212921142578, "learning_rate": 2.0507989864326338e-06, "loss": 0.6719, "step": 14885 }, { "epoch": 2.4, "grad_norm": 10.57332706451416, "learning_rate": 2.0497446427611976e-06, "loss": 0.565, "step": 14886 }, { "epoch": 2.4, "grad_norm": 15.406952857971191, "learning_rate": 2.0486905392362024e-06, "loss": 0.697, "step": 14887 }, { "epoch": 2.4, "grad_norm": 11.747030258178711, "learning_rate": 2.0476366758894907e-06, "loss": 0.818, "step": 14888 }, { "epoch": 2.4, "grad_norm": 13.835562705993652, "learning_rate": 2.046583052752896e-06, "loss": 0.6381, "step": 14889 }, { "epoch": 2.4, "grad_norm": 13.394745826721191, "learning_rate": 2.0455296698582393e-06, "loss": 0.6166, "step": 14890 }, { "epoch": 2.4, "grad_norm": 10.84762191772461, "learning_rate": 2.044476527237347e-06, "loss": 0.7242, "step": 14891 }, { "epoch": 2.4, "grad_norm": 11.642838478088379, "learning_rate": 2.043423624922023e-06, "loss": 0.6851, "step": 14892 }, { "epoch": 2.4, "grad_norm": 11.77966022491455, "learning_rate": 2.0423709629440767e-06, "loss": 0.7613, "step": 14893 }, { "epoch": 2.4, "grad_norm": 11.735633850097656, "learning_rate": 2.0413185413353044e-06, "loss": 0.702, "step": 14894 }, { "epoch": 2.4, "grad_norm": 16.3760929107666, "learning_rate": 2.040266360127491e-06, "loss": 0.7387, "step": 14895 }, { "epoch": 2.4, "grad_norm": 12.819955825805664, "learning_rate": 2.0392144193524256e-06, "loss": 0.509, "step": 14896 }, { "epoch": 2.4, "grad_norm": 10.986716270446777, "learning_rate": 2.0381627190418784e-06, "loss": 0.6557, "step": 14897 }, { "epoch": 2.4, "grad_norm": 12.341724395751953, "learning_rate": 2.0371112592276145e-06, "loss": 0.6941, "step": 14898 }, { "epoch": 2.4, "grad_norm": 13.536761283874512, "learning_rate": 2.0360600399414033e-06, "loss": 0.6352, "step": 14899 }, { "epoch": 2.4, "grad_norm": 12.307097434997559, "learning_rate": 2.035009061214994e-06, "loss": 0.9005, "step": 14900 }, { "epoch": 2.4, "grad_norm": 12.638150215148926, "learning_rate": 2.0339583230801285e-06, "loss": 0.6614, "step": 14901 }, { "epoch": 2.4, "grad_norm": 11.822005271911621, "learning_rate": 2.0329078255685518e-06, "loss": 0.7847, "step": 14902 }, { "epoch": 2.4, "grad_norm": 10.757953643798828, "learning_rate": 2.0318575687119903e-06, "loss": 0.8837, "step": 14903 }, { "epoch": 2.4, "grad_norm": 16.480037689208984, "learning_rate": 2.030807552542172e-06, "loss": 0.7637, "step": 14904 }, { "epoch": 2.4, "grad_norm": 13.070754051208496, "learning_rate": 2.0297577770908126e-06, "loss": 0.707, "step": 14905 }, { "epoch": 2.4, "grad_norm": 14.604192733764648, "learning_rate": 2.0287082423896197e-06, "loss": 0.8943, "step": 14906 }, { "epoch": 2.4, "grad_norm": 10.934795379638672, "learning_rate": 2.027658948470299e-06, "loss": 0.6875, "step": 14907 }, { "epoch": 2.4, "grad_norm": 10.868366241455078, "learning_rate": 2.026609895364544e-06, "loss": 0.7814, "step": 14908 }, { "epoch": 2.4, "grad_norm": 15.575431823730469, "learning_rate": 2.0255610831040407e-06, "loss": 0.8918, "step": 14909 }, { "epoch": 2.4, "grad_norm": 11.136120796203613, "learning_rate": 2.0245125117204735e-06, "loss": 0.6852, "step": 14910 }, { "epoch": 2.4, "grad_norm": 11.020831108093262, "learning_rate": 2.02346418124551e-06, "loss": 0.7398, "step": 14911 }, { "epoch": 2.4, "grad_norm": 13.302665710449219, "learning_rate": 2.0224160917108236e-06, "loss": 0.8218, "step": 14912 }, { "epoch": 2.4, "grad_norm": 14.496776580810547, "learning_rate": 2.0213682431480698e-06, "loss": 0.7805, "step": 14913 }, { "epoch": 2.4, "grad_norm": 13.568679809570312, "learning_rate": 2.0203206355888962e-06, "loss": 0.8621, "step": 14914 }, { "epoch": 2.4, "grad_norm": 11.339086532592773, "learning_rate": 2.019273269064953e-06, "loss": 0.7438, "step": 14915 }, { "epoch": 2.4, "grad_norm": 11.64747428894043, "learning_rate": 2.018226143607874e-06, "loss": 0.4892, "step": 14916 }, { "epoch": 2.4, "grad_norm": 13.12724781036377, "learning_rate": 2.017179259249288e-06, "loss": 0.9894, "step": 14917 }, { "epoch": 2.4, "grad_norm": 14.42973518371582, "learning_rate": 2.0161326160208173e-06, "loss": 0.7432, "step": 14918 }, { "epoch": 2.4, "grad_norm": 10.319452285766602, "learning_rate": 2.0150862139540804e-06, "loss": 0.4692, "step": 14919 }, { "epoch": 2.4, "grad_norm": 13.978754997253418, "learning_rate": 2.0140400530806803e-06, "loss": 0.7072, "step": 14920 }, { "epoch": 2.4, "grad_norm": 11.791096687316895, "learning_rate": 2.0129941334322234e-06, "loss": 0.6484, "step": 14921 }, { "epoch": 2.4, "grad_norm": 13.596470832824707, "learning_rate": 2.0119484550402966e-06, "loss": 0.6063, "step": 14922 }, { "epoch": 2.4, "grad_norm": 11.992594718933105, "learning_rate": 2.010903017936491e-06, "loss": 0.7607, "step": 14923 }, { "epoch": 2.4, "grad_norm": 14.561432838439941, "learning_rate": 2.0098578221523835e-06, "loss": 0.8035, "step": 14924 }, { "epoch": 2.4, "grad_norm": 13.689969062805176, "learning_rate": 2.0088128677195417e-06, "loss": 0.5079, "step": 14925 }, { "epoch": 2.4, "grad_norm": 13.56838607788086, "learning_rate": 2.007768154669535e-06, "loss": 0.6868, "step": 14926 }, { "epoch": 2.4, "grad_norm": 12.413878440856934, "learning_rate": 2.006723683033919e-06, "loss": 0.6325, "step": 14927 }, { "epoch": 2.4, "grad_norm": 10.468189239501953, "learning_rate": 2.005679452844238e-06, "loss": 0.4984, "step": 14928 }, { "epoch": 2.4, "grad_norm": 12.089171409606934, "learning_rate": 2.0046354641320407e-06, "loss": 0.7319, "step": 14929 }, { "epoch": 2.4, "grad_norm": 11.976189613342285, "learning_rate": 2.003591716928858e-06, "loss": 0.6175, "step": 14930 }, { "epoch": 2.4, "grad_norm": 13.188603401184082, "learning_rate": 2.0025482112662207e-06, "loss": 0.7378, "step": 14931 }, { "epoch": 2.4, "grad_norm": 10.919572830200195, "learning_rate": 2.0015049471756476e-06, "loss": 0.6758, "step": 14932 }, { "epoch": 2.4, "grad_norm": 14.255935668945312, "learning_rate": 2.0004619246886493e-06, "loss": 0.8614, "step": 14933 }, { "epoch": 2.4, "grad_norm": 15.821033477783203, "learning_rate": 1.999419143836735e-06, "loss": 0.7319, "step": 14934 }, { "epoch": 2.4, "grad_norm": 16.37641716003418, "learning_rate": 1.9983766046513997e-06, "loss": 0.7557, "step": 14935 }, { "epoch": 2.4, "grad_norm": 11.913209915161133, "learning_rate": 1.9973343071641384e-06, "loss": 0.7376, "step": 14936 }, { "epoch": 2.4, "grad_norm": 12.963312149047852, "learning_rate": 1.9962922514064333e-06, "loss": 0.909, "step": 14937 }, { "epoch": 2.4, "grad_norm": 15.599122047424316, "learning_rate": 1.995250437409757e-06, "loss": 0.782, "step": 14938 }, { "epoch": 2.4, "grad_norm": 11.992032051086426, "learning_rate": 1.9942088652055823e-06, "loss": 0.6554, "step": 14939 }, { "epoch": 2.41, "grad_norm": 13.912259101867676, "learning_rate": 1.9931675348253743e-06, "loss": 0.7484, "step": 14940 }, { "epoch": 2.41, "grad_norm": 9.12031364440918, "learning_rate": 1.9921264463005806e-06, "loss": 0.5428, "step": 14941 }, { "epoch": 2.41, "grad_norm": 10.7279634475708, "learning_rate": 1.9910855996626553e-06, "loss": 0.596, "step": 14942 }, { "epoch": 2.41, "grad_norm": 15.055706024169922, "learning_rate": 1.9900449949430343e-06, "loss": 0.6987, "step": 14943 }, { "epoch": 2.41, "grad_norm": 12.791927337646484, "learning_rate": 1.989004632173148e-06, "loss": 0.6926, "step": 14944 }, { "epoch": 2.41, "grad_norm": 13.402039527893066, "learning_rate": 1.987964511384428e-06, "loss": 0.7664, "step": 14945 }, { "epoch": 2.41, "grad_norm": 12.861078262329102, "learning_rate": 1.9869246326082857e-06, "loss": 0.7323, "step": 14946 }, { "epoch": 2.41, "grad_norm": 11.89201831817627, "learning_rate": 1.9858849958761385e-06, "loss": 0.7462, "step": 14947 }, { "epoch": 2.41, "grad_norm": 10.908440589904785, "learning_rate": 1.9848456012193864e-06, "loss": 0.7689, "step": 14948 }, { "epoch": 2.41, "grad_norm": 10.906953811645508, "learning_rate": 1.983806448669422e-06, "loss": 0.7614, "step": 14949 }, { "epoch": 2.41, "grad_norm": 12.332466125488281, "learning_rate": 1.9827675382576404e-06, "loss": 0.6531, "step": 14950 }, { "epoch": 2.41, "grad_norm": 13.185309410095215, "learning_rate": 1.98172887001542e-06, "loss": 1.0106, "step": 14951 }, { "epoch": 2.41, "grad_norm": 11.73438835144043, "learning_rate": 1.9806904439741337e-06, "loss": 0.7104, "step": 14952 }, { "epoch": 2.41, "grad_norm": 13.901150703430176, "learning_rate": 1.9796522601651514e-06, "loss": 0.6361, "step": 14953 }, { "epoch": 2.41, "grad_norm": 10.15471076965332, "learning_rate": 1.978614318619829e-06, "loss": 0.59, "step": 14954 }, { "epoch": 2.41, "grad_norm": 12.820252418518066, "learning_rate": 1.9775766193695223e-06, "loss": 0.7958, "step": 14955 }, { "epoch": 2.41, "grad_norm": 12.59489631652832, "learning_rate": 1.9765391624455753e-06, "loss": 0.6228, "step": 14956 }, { "epoch": 2.41, "grad_norm": 12.097057342529297, "learning_rate": 1.9755019478793226e-06, "loss": 0.9605, "step": 14957 }, { "epoch": 2.41, "grad_norm": 13.34349536895752, "learning_rate": 1.9744649757020963e-06, "loss": 0.6322, "step": 14958 }, { "epoch": 2.41, "grad_norm": 11.124251365661621, "learning_rate": 1.9734282459452226e-06, "loss": 0.8044, "step": 14959 }, { "epoch": 2.41, "grad_norm": 13.8936767578125, "learning_rate": 1.972391758640011e-06, "loss": 0.6432, "step": 14960 }, { "epoch": 2.41, "grad_norm": 12.130380630493164, "learning_rate": 1.9713555138177753e-06, "loss": 0.7261, "step": 14961 }, { "epoch": 2.41, "grad_norm": 12.440560340881348, "learning_rate": 1.970319511509815e-06, "loss": 0.6543, "step": 14962 }, { "epoch": 2.41, "grad_norm": 12.778541564941406, "learning_rate": 1.9692837517474193e-06, "loss": 0.9117, "step": 14963 }, { "epoch": 2.41, "grad_norm": 12.201854705810547, "learning_rate": 1.9682482345618813e-06, "loss": 0.5354, "step": 14964 }, { "epoch": 2.41, "grad_norm": 13.606093406677246, "learning_rate": 1.9672129599844735e-06, "loss": 0.6066, "step": 14965 }, { "epoch": 2.41, "grad_norm": 14.747942924499512, "learning_rate": 1.9661779280464745e-06, "loss": 0.6996, "step": 14966 }, { "epoch": 2.41, "grad_norm": 13.770517349243164, "learning_rate": 1.965143138779144e-06, "loss": 0.7305, "step": 14967 }, { "epoch": 2.41, "grad_norm": 13.454383850097656, "learning_rate": 1.964108592213737e-06, "loss": 0.4808, "step": 14968 }, { "epoch": 2.41, "grad_norm": 14.124985694885254, "learning_rate": 1.963074288381509e-06, "loss": 0.7873, "step": 14969 }, { "epoch": 2.41, "grad_norm": 13.42049789428711, "learning_rate": 1.962040227313696e-06, "loss": 0.508, "step": 14970 }, { "epoch": 2.41, "grad_norm": 12.174226760864258, "learning_rate": 1.9610064090415404e-06, "loss": 0.6271, "step": 14971 }, { "epoch": 2.41, "grad_norm": 11.493759155273438, "learning_rate": 1.959972833596264e-06, "loss": 0.7235, "step": 14972 }, { "epoch": 2.41, "grad_norm": 12.446700096130371, "learning_rate": 1.9589395010090874e-06, "loss": 0.4868, "step": 14973 }, { "epoch": 2.41, "grad_norm": 12.904891014099121, "learning_rate": 1.9579064113112266e-06, "loss": 0.6514, "step": 14974 }, { "epoch": 2.41, "grad_norm": 13.084122657775879, "learning_rate": 1.9568735645338867e-06, "loss": 0.7074, "step": 14975 }, { "epoch": 2.41, "grad_norm": 14.08216381072998, "learning_rate": 1.9558409607082605e-06, "loss": 0.705, "step": 14976 }, { "epoch": 2.41, "grad_norm": 11.884218215942383, "learning_rate": 1.9548085998655477e-06, "loss": 0.6716, "step": 14977 }, { "epoch": 2.41, "grad_norm": 14.35486125946045, "learning_rate": 1.953776482036924e-06, "loss": 0.8671, "step": 14978 }, { "epoch": 2.41, "grad_norm": 132.53599548339844, "learning_rate": 1.9527446072535704e-06, "loss": 0.5773, "step": 14979 }, { "epoch": 2.41, "grad_norm": 12.446440696716309, "learning_rate": 1.951712975546657e-06, "loss": 0.6155, "step": 14980 }, { "epoch": 2.41, "grad_norm": 12.843986511230469, "learning_rate": 1.9506815869473404e-06, "loss": 0.9653, "step": 14981 }, { "epoch": 2.41, "grad_norm": 11.256377220153809, "learning_rate": 1.94965044148678e-06, "loss": 0.6787, "step": 14982 }, { "epoch": 2.41, "grad_norm": 15.447009086608887, "learning_rate": 1.9486195391961204e-06, "loss": 0.6922, "step": 14983 }, { "epoch": 2.41, "grad_norm": 13.17794132232666, "learning_rate": 1.9475888801064993e-06, "loss": 0.7587, "step": 14984 }, { "epoch": 2.41, "grad_norm": 11.885536193847656, "learning_rate": 1.946558464249052e-06, "loss": 0.857, "step": 14985 }, { "epoch": 2.41, "grad_norm": 13.703692436218262, "learning_rate": 1.9455282916549035e-06, "loss": 0.6835, "step": 14986 }, { "epoch": 2.41, "grad_norm": 15.749753952026367, "learning_rate": 1.944498362355166e-06, "loss": 0.6476, "step": 14987 }, { "epoch": 2.41, "grad_norm": 11.192694664001465, "learning_rate": 1.9434686763809564e-06, "loss": 0.4882, "step": 14988 }, { "epoch": 2.41, "grad_norm": 14.16451358795166, "learning_rate": 1.9424392337633725e-06, "loss": 0.9541, "step": 14989 }, { "epoch": 2.41, "grad_norm": 15.400262832641602, "learning_rate": 1.941410034533515e-06, "loss": 0.7427, "step": 14990 }, { "epoch": 2.41, "grad_norm": 14.004274368286133, "learning_rate": 1.9403810787224686e-06, "loss": 0.7667, "step": 14991 }, { "epoch": 2.41, "grad_norm": 15.698310852050781, "learning_rate": 1.9393523663613123e-06, "loss": 0.4551, "step": 14992 }, { "epoch": 2.41, "grad_norm": 14.037957191467285, "learning_rate": 1.9383238974811235e-06, "loss": 0.8, "step": 14993 }, { "epoch": 2.41, "grad_norm": 14.451607704162598, "learning_rate": 1.9372956721129667e-06, "loss": 0.4696, "step": 14994 }, { "epoch": 2.41, "grad_norm": 12.036993980407715, "learning_rate": 1.9362676902878983e-06, "loss": 0.8909, "step": 14995 }, { "epoch": 2.41, "grad_norm": 15.008509635925293, "learning_rate": 1.9352399520369736e-06, "loss": 0.6304, "step": 14996 }, { "epoch": 2.41, "grad_norm": 12.954778671264648, "learning_rate": 1.9342124573912325e-06, "loss": 0.603, "step": 14997 }, { "epoch": 2.41, "grad_norm": 12.161654472351074, "learning_rate": 1.933185206381717e-06, "loss": 0.6689, "step": 14998 }, { "epoch": 2.41, "grad_norm": 12.759628295898438, "learning_rate": 1.93215819903945e-06, "loss": 0.7298, "step": 14999 }, { "epoch": 2.41, "grad_norm": 15.608778953552246, "learning_rate": 1.931131435395457e-06, "loss": 0.774, "step": 15000 }, { "epoch": 2.41, "grad_norm": 16.02450942993164, "learning_rate": 1.930104915480754e-06, "loss": 0.7127, "step": 15001 }, { "epoch": 2.42, "grad_norm": 14.871063232421875, "learning_rate": 1.9290786393263474e-06, "loss": 0.8604, "step": 15002 }, { "epoch": 2.42, "grad_norm": 13.056259155273438, "learning_rate": 1.928052606963233e-06, "loss": 0.5778, "step": 15003 }, { "epoch": 2.42, "grad_norm": 13.02949333190918, "learning_rate": 1.9270268184224094e-06, "loss": 0.7782, "step": 15004 }, { "epoch": 2.42, "grad_norm": 11.992437362670898, "learning_rate": 1.9260012737348575e-06, "loss": 0.4916, "step": 15005 }, { "epoch": 2.42, "grad_norm": 16.360149383544922, "learning_rate": 1.9249759729315544e-06, "loss": 0.7495, "step": 15006 }, { "epoch": 2.42, "grad_norm": 12.777135848999023, "learning_rate": 1.9239509160434756e-06, "loss": 0.5321, "step": 15007 }, { "epoch": 2.42, "grad_norm": 12.892610549926758, "learning_rate": 1.922926103101577e-06, "loss": 0.6353, "step": 15008 }, { "epoch": 2.42, "grad_norm": 11.524744033813477, "learning_rate": 1.921901534136821e-06, "loss": 0.806, "step": 15009 }, { "epoch": 2.42, "grad_norm": 14.360599517822266, "learning_rate": 1.920877209180152e-06, "loss": 0.7403, "step": 15010 }, { "epoch": 2.42, "grad_norm": 12.964513778686523, "learning_rate": 1.9198531282625097e-06, "loss": 0.7284, "step": 15011 }, { "epoch": 2.42, "grad_norm": 13.555685043334961, "learning_rate": 1.918829291414832e-06, "loss": 0.7118, "step": 15012 }, { "epoch": 2.42, "grad_norm": 9.299534797668457, "learning_rate": 1.9178056986680403e-06, "loss": 0.5472, "step": 15013 }, { "epoch": 2.42, "grad_norm": 15.940386772155762, "learning_rate": 1.916782350053058e-06, "loss": 0.7621, "step": 15014 }, { "epoch": 2.42, "grad_norm": 12.306598663330078, "learning_rate": 1.9157592456007943e-06, "loss": 0.6743, "step": 15015 }, { "epoch": 2.42, "grad_norm": 17.19573402404785, "learning_rate": 1.914736385342152e-06, "loss": 0.6718, "step": 15016 }, { "epoch": 2.42, "grad_norm": 10.308869361877441, "learning_rate": 1.9137137693080297e-06, "loss": 0.7396, "step": 15017 }, { "epoch": 2.42, "grad_norm": 10.752898216247559, "learning_rate": 1.912691397529317e-06, "loss": 0.8828, "step": 15018 }, { "epoch": 2.42, "grad_norm": 11.518190383911133, "learning_rate": 1.9116692700368887e-06, "loss": 0.5668, "step": 15019 }, { "epoch": 2.42, "grad_norm": 13.001172065734863, "learning_rate": 1.9106473868616317e-06, "loss": 0.7754, "step": 15020 }, { "epoch": 2.42, "grad_norm": 16.139400482177734, "learning_rate": 1.909625748034405e-06, "loss": 0.6817, "step": 15021 }, { "epoch": 2.42, "grad_norm": 13.606070518493652, "learning_rate": 1.9086043535860687e-06, "loss": 0.8899, "step": 15022 }, { "epoch": 2.42, "grad_norm": 12.18032169342041, "learning_rate": 1.9075832035474784e-06, "loss": 0.6574, "step": 15023 }, { "epoch": 2.42, "grad_norm": 12.083642959594727, "learning_rate": 1.9065622979494759e-06, "loss": 0.5755, "step": 15024 }, { "epoch": 2.42, "grad_norm": 12.380026817321777, "learning_rate": 1.9055416368229019e-06, "loss": 0.7098, "step": 15025 }, { "epoch": 2.42, "grad_norm": 16.19944190979004, "learning_rate": 1.9045212201985852e-06, "loss": 0.7029, "step": 15026 }, { "epoch": 2.42, "grad_norm": 14.745762825012207, "learning_rate": 1.9035010481073456e-06, "loss": 0.6407, "step": 15027 }, { "epoch": 2.42, "grad_norm": 13.660149574279785, "learning_rate": 1.9024811205800032e-06, "loss": 0.6709, "step": 15028 }, { "epoch": 2.42, "grad_norm": 13.120356559753418, "learning_rate": 1.901461437647365e-06, "loss": 0.6464, "step": 15029 }, { "epoch": 2.42, "grad_norm": 10.849971771240234, "learning_rate": 1.900441999340228e-06, "loss": 0.5712, "step": 15030 }, { "epoch": 2.42, "grad_norm": 10.308330535888672, "learning_rate": 1.8994228056893904e-06, "loss": 0.5892, "step": 15031 }, { "epoch": 2.42, "grad_norm": 13.34398365020752, "learning_rate": 1.8984038567256346e-06, "loss": 0.8632, "step": 15032 }, { "epoch": 2.42, "grad_norm": 11.748462677001953, "learning_rate": 1.8973851524797416e-06, "loss": 0.6544, "step": 15033 }, { "epoch": 2.42, "grad_norm": 10.292579650878906, "learning_rate": 1.896366692982482e-06, "loss": 0.7824, "step": 15034 }, { "epoch": 2.42, "grad_norm": 12.788456916809082, "learning_rate": 1.895348478264617e-06, "loss": 0.6192, "step": 15035 }, { "epoch": 2.42, "grad_norm": 15.149060249328613, "learning_rate": 1.8943305083569074e-06, "loss": 0.8458, "step": 15036 }, { "epoch": 2.42, "grad_norm": 12.386805534362793, "learning_rate": 1.8933127832900955e-06, "loss": 0.8116, "step": 15037 }, { "epoch": 2.42, "grad_norm": 12.678102493286133, "learning_rate": 1.8922953030949309e-06, "loss": 0.6475, "step": 15038 }, { "epoch": 2.42, "grad_norm": 13.996480941772461, "learning_rate": 1.8912780678021435e-06, "loss": 0.5247, "step": 15039 }, { "epoch": 2.42, "grad_norm": 11.3610258102417, "learning_rate": 1.890261077442458e-06, "loss": 0.6291, "step": 15040 }, { "epoch": 2.42, "grad_norm": 14.109458923339844, "learning_rate": 1.8892443320465959e-06, "loss": 0.6209, "step": 15041 }, { "epoch": 2.42, "grad_norm": 13.377487182617188, "learning_rate": 1.8882278316452706e-06, "loss": 0.792, "step": 15042 }, { "epoch": 2.42, "grad_norm": 10.504385948181152, "learning_rate": 1.8872115762691845e-06, "loss": 0.6438, "step": 15043 }, { "epoch": 2.42, "grad_norm": 10.775482177734375, "learning_rate": 1.8861955659490373e-06, "loss": 0.7218, "step": 15044 }, { "epoch": 2.42, "grad_norm": 10.866290092468262, "learning_rate": 1.8851798007155175e-06, "loss": 0.7118, "step": 15045 }, { "epoch": 2.42, "grad_norm": 13.33940601348877, "learning_rate": 1.8841642805993033e-06, "loss": 0.7734, "step": 15046 }, { "epoch": 2.42, "grad_norm": 12.17989730834961, "learning_rate": 1.8831490056310764e-06, "loss": 0.5964, "step": 15047 }, { "epoch": 2.42, "grad_norm": 14.679061889648438, "learning_rate": 1.8821339758414991e-06, "loss": 0.8267, "step": 15048 }, { "epoch": 2.42, "grad_norm": 9.980814933776855, "learning_rate": 1.881119191261236e-06, "loss": 0.5508, "step": 15049 }, { "epoch": 2.42, "grad_norm": 11.051575660705566, "learning_rate": 1.8801046519209377e-06, "loss": 0.5661, "step": 15050 }, { "epoch": 2.42, "grad_norm": 13.502153396606445, "learning_rate": 1.8790903578512465e-06, "loss": 0.6211, "step": 15051 }, { "epoch": 2.42, "grad_norm": 11.603987693786621, "learning_rate": 1.8780763090828047e-06, "loss": 0.5702, "step": 15052 }, { "epoch": 2.42, "grad_norm": 14.045448303222656, "learning_rate": 1.8770625056462421e-06, "loss": 0.7009, "step": 15053 }, { "epoch": 2.42, "grad_norm": 11.444086074829102, "learning_rate": 1.8760489475721777e-06, "loss": 0.6822, "step": 15054 }, { "epoch": 2.42, "grad_norm": 18.80925941467285, "learning_rate": 1.8750356348912336e-06, "loss": 0.747, "step": 15055 }, { "epoch": 2.42, "grad_norm": 11.561334609985352, "learning_rate": 1.8740225676340118e-06, "loss": 0.763, "step": 15056 }, { "epoch": 2.42, "grad_norm": 12.182785987854004, "learning_rate": 1.873009745831118e-06, "loss": 0.6433, "step": 15057 }, { "epoch": 2.42, "grad_norm": 13.506847381591797, "learning_rate": 1.8719971695131445e-06, "loss": 0.797, "step": 15058 }, { "epoch": 2.42, "grad_norm": 13.412565231323242, "learning_rate": 1.8709848387106734e-06, "loss": 0.6032, "step": 15059 }, { "epoch": 2.42, "grad_norm": 10.886734962463379, "learning_rate": 1.8699727534542867e-06, "loss": 0.6714, "step": 15060 }, { "epoch": 2.42, "grad_norm": 13.967975616455078, "learning_rate": 1.8689609137745578e-06, "loss": 0.6872, "step": 15061 }, { "epoch": 2.42, "grad_norm": 12.528322219848633, "learning_rate": 1.867949319702046e-06, "loss": 0.7596, "step": 15062 }, { "epoch": 2.42, "grad_norm": 12.208088874816895, "learning_rate": 1.8669379712673107e-06, "loss": 0.8045, "step": 15063 }, { "epoch": 2.42, "grad_norm": 18.594633102416992, "learning_rate": 1.8659268685009012e-06, "loss": 0.7328, "step": 15064 }, { "epoch": 2.43, "grad_norm": 13.367499351501465, "learning_rate": 1.8649160114333542e-06, "loss": 0.7196, "step": 15065 }, { "epoch": 2.43, "grad_norm": 10.301528930664062, "learning_rate": 1.8639054000952094e-06, "loss": 0.6953, "step": 15066 }, { "epoch": 2.43, "grad_norm": 11.45982837677002, "learning_rate": 1.8628950345169893e-06, "loss": 0.7846, "step": 15067 }, { "epoch": 2.43, "grad_norm": 12.411449432373047, "learning_rate": 1.8618849147292173e-06, "loss": 0.6503, "step": 15068 }, { "epoch": 2.43, "grad_norm": 11.712636947631836, "learning_rate": 1.8608750407624033e-06, "loss": 0.8231, "step": 15069 }, { "epoch": 2.43, "grad_norm": 23.918893814086914, "learning_rate": 1.8598654126470483e-06, "loss": 0.7899, "step": 15070 }, { "epoch": 2.43, "grad_norm": 13.217419624328613, "learning_rate": 1.858856030413654e-06, "loss": 0.8469, "step": 15071 }, { "epoch": 2.43, "grad_norm": 11.965479850769043, "learning_rate": 1.8578468940927085e-06, "loss": 0.7358, "step": 15072 }, { "epoch": 2.43, "grad_norm": 12.541685104370117, "learning_rate": 1.8568380037146915e-06, "loss": 0.4999, "step": 15073 }, { "epoch": 2.43, "grad_norm": 9.554365158081055, "learning_rate": 1.8558293593100828e-06, "loss": 0.6946, "step": 15074 }, { "epoch": 2.43, "grad_norm": 10.382884979248047, "learning_rate": 1.8548209609093426e-06, "loss": 0.7474, "step": 15075 }, { "epoch": 2.43, "grad_norm": 11.74459171295166, "learning_rate": 1.8538128085429386e-06, "loss": 0.6115, "step": 15076 }, { "epoch": 2.43, "grad_norm": 12.533788681030273, "learning_rate": 1.8528049022413187e-06, "loss": 0.7364, "step": 15077 }, { "epoch": 2.43, "grad_norm": 12.351668357849121, "learning_rate": 1.851797242034926e-06, "loss": 0.693, "step": 15078 }, { "epoch": 2.43, "grad_norm": 13.581830978393555, "learning_rate": 1.8507898279542025e-06, "loss": 0.9254, "step": 15079 }, { "epoch": 2.43, "grad_norm": 13.432891845703125, "learning_rate": 1.8497826600295742e-06, "loss": 0.6296, "step": 15080 }, { "epoch": 2.43, "grad_norm": 10.350330352783203, "learning_rate": 1.8487757382914652e-06, "loss": 0.5798, "step": 15081 }, { "epoch": 2.43, "grad_norm": 9.97535514831543, "learning_rate": 1.8477690627702938e-06, "loss": 0.7188, "step": 15082 }, { "epoch": 2.43, "grad_norm": 19.23162269592285, "learning_rate": 1.846762633496466e-06, "loss": 0.6249, "step": 15083 }, { "epoch": 2.43, "grad_norm": 13.722614288330078, "learning_rate": 1.845756450500379e-06, "loss": 0.7941, "step": 15084 }, { "epoch": 2.43, "grad_norm": 11.352540016174316, "learning_rate": 1.84475051381243e-06, "loss": 0.8534, "step": 15085 }, { "epoch": 2.43, "grad_norm": 14.644373893737793, "learning_rate": 1.8437448234629995e-06, "loss": 0.7936, "step": 15086 }, { "epoch": 2.43, "grad_norm": 12.280058860778809, "learning_rate": 1.8427393794824723e-06, "loss": 0.8157, "step": 15087 }, { "epoch": 2.43, "grad_norm": 12.659215927124023, "learning_rate": 1.8417341819012135e-06, "loss": 0.6994, "step": 15088 }, { "epoch": 2.43, "grad_norm": 13.065800666809082, "learning_rate": 1.8407292307495873e-06, "loss": 0.7271, "step": 15089 }, { "epoch": 2.43, "grad_norm": 10.73980712890625, "learning_rate": 1.839724526057951e-06, "loss": 0.5771, "step": 15090 }, { "epoch": 2.43, "grad_norm": 11.648645401000977, "learning_rate": 1.8387200678566497e-06, "loss": 0.8052, "step": 15091 }, { "epoch": 2.43, "grad_norm": 13.845401763916016, "learning_rate": 1.83771585617603e-06, "loss": 0.7255, "step": 15092 }, { "epoch": 2.43, "grad_norm": 17.747737884521484, "learning_rate": 1.83671189104642e-06, "loss": 0.7664, "step": 15093 }, { "epoch": 2.43, "grad_norm": 13.049107551574707, "learning_rate": 1.8357081724981452e-06, "loss": 0.8584, "step": 15094 }, { "epoch": 2.43, "grad_norm": 13.104120254516602, "learning_rate": 1.8347047005615294e-06, "loss": 0.7824, "step": 15095 }, { "epoch": 2.43, "grad_norm": 10.921590805053711, "learning_rate": 1.8337014752668792e-06, "loss": 0.5916, "step": 15096 }, { "epoch": 2.43, "grad_norm": 11.672029495239258, "learning_rate": 1.832698496644497e-06, "loss": 0.6171, "step": 15097 }, { "epoch": 2.43, "grad_norm": 12.098347663879395, "learning_rate": 1.8316957647246846e-06, "loss": 0.5689, "step": 15098 }, { "epoch": 2.43, "grad_norm": 9.80162239074707, "learning_rate": 1.830693279537723e-06, "loss": 0.4905, "step": 15099 }, { "epoch": 2.43, "grad_norm": 13.223467826843262, "learning_rate": 1.829691041113899e-06, "loss": 0.6657, "step": 15100 }, { "epoch": 2.43, "grad_norm": 12.880925178527832, "learning_rate": 1.8286890494834863e-06, "loss": 0.774, "step": 15101 }, { "epoch": 2.43, "grad_norm": 11.592848777770996, "learning_rate": 1.8276873046767473e-06, "loss": 0.4999, "step": 15102 }, { "epoch": 2.43, "grad_norm": 13.988537788391113, "learning_rate": 1.8266858067239468e-06, "loss": 0.7214, "step": 15103 }, { "epoch": 2.43, "grad_norm": 15.267597198486328, "learning_rate": 1.8256845556553326e-06, "loss": 0.567, "step": 15104 }, { "epoch": 2.43, "grad_norm": 13.166998863220215, "learning_rate": 1.8246835515011452e-06, "loss": 0.7397, "step": 15105 }, { "epoch": 2.43, "grad_norm": 11.953243255615234, "learning_rate": 1.823682794291628e-06, "loss": 0.7182, "step": 15106 }, { "epoch": 2.43, "grad_norm": 11.515719413757324, "learning_rate": 1.8226822840570058e-06, "loss": 0.7573, "step": 15107 }, { "epoch": 2.43, "grad_norm": 13.31420612335205, "learning_rate": 1.8216820208274987e-06, "loss": 0.8598, "step": 15108 }, { "epoch": 2.43, "grad_norm": 12.117562294006348, "learning_rate": 1.8206820046333263e-06, "loss": 0.6936, "step": 15109 }, { "epoch": 2.43, "grad_norm": 10.7721586227417, "learning_rate": 1.8196822355046884e-06, "loss": 0.7255, "step": 15110 }, { "epoch": 2.43, "grad_norm": 12.872395515441895, "learning_rate": 1.8186827134717911e-06, "loss": 0.8583, "step": 15111 }, { "epoch": 2.43, "grad_norm": 15.502740859985352, "learning_rate": 1.8176834385648224e-06, "loss": 0.7579, "step": 15112 }, { "epoch": 2.43, "grad_norm": 11.416234970092773, "learning_rate": 1.8166844108139636e-06, "loss": 0.5006, "step": 15113 }, { "epoch": 2.43, "grad_norm": 14.411821365356445, "learning_rate": 1.8156856302493974e-06, "loss": 0.9099, "step": 15114 }, { "epoch": 2.43, "grad_norm": 14.738378524780273, "learning_rate": 1.8146870969012885e-06, "loss": 0.7469, "step": 15115 }, { "epoch": 2.43, "grad_norm": 12.48839282989502, "learning_rate": 1.813688810799802e-06, "loss": 0.5823, "step": 15116 }, { "epoch": 2.43, "grad_norm": 12.245722770690918, "learning_rate": 1.812690771975092e-06, "loss": 0.6465, "step": 15117 }, { "epoch": 2.43, "grad_norm": 12.904131889343262, "learning_rate": 1.8116929804573013e-06, "loss": 0.7167, "step": 15118 }, { "epoch": 2.43, "grad_norm": 11.617898941040039, "learning_rate": 1.810695436276575e-06, "loss": 0.5878, "step": 15119 }, { "epoch": 2.43, "grad_norm": 10.997345924377441, "learning_rate": 1.809698139463042e-06, "loss": 0.6139, "step": 15120 }, { "epoch": 2.43, "grad_norm": 13.868865966796875, "learning_rate": 1.8087010900468215e-06, "loss": 0.6716, "step": 15121 }, { "epoch": 2.43, "grad_norm": 15.552732467651367, "learning_rate": 1.807704288058042e-06, "loss": 0.6591, "step": 15122 }, { "epoch": 2.43, "grad_norm": 11.251041412353516, "learning_rate": 1.8067077335268058e-06, "loss": 0.6513, "step": 15123 }, { "epoch": 2.43, "grad_norm": 13.077792167663574, "learning_rate": 1.8057114264832154e-06, "loss": 0.603, "step": 15124 }, { "epoch": 2.43, "grad_norm": 13.005455017089844, "learning_rate": 1.804715366957368e-06, "loss": 0.7543, "step": 15125 }, { "epoch": 2.43, "grad_norm": 13.619866371154785, "learning_rate": 1.8037195549793462e-06, "loss": 0.7254, "step": 15126 }, { "epoch": 2.44, "grad_norm": 12.383164405822754, "learning_rate": 1.802723990579235e-06, "loss": 0.7151, "step": 15127 }, { "epoch": 2.44, "grad_norm": 12.379307746887207, "learning_rate": 1.8017286737871043e-06, "loss": 0.468, "step": 15128 }, { "epoch": 2.44, "grad_norm": 15.421106338500977, "learning_rate": 1.800733604633016e-06, "loss": 0.7051, "step": 15129 }, { "epoch": 2.44, "grad_norm": 14.771071434020996, "learning_rate": 1.7997387831470314e-06, "loss": 0.7675, "step": 15130 }, { "epoch": 2.44, "grad_norm": 14.652132987976074, "learning_rate": 1.798744209359199e-06, "loss": 0.6322, "step": 15131 }, { "epoch": 2.44, "grad_norm": 13.048860549926758, "learning_rate": 1.7977498832995576e-06, "loss": 0.6999, "step": 15132 }, { "epoch": 2.44, "grad_norm": 15.22170352935791, "learning_rate": 1.7967558049981482e-06, "loss": 0.7882, "step": 15133 }, { "epoch": 2.44, "grad_norm": 14.821084022521973, "learning_rate": 1.7957619744849909e-06, "loss": 0.5905, "step": 15134 }, { "epoch": 2.44, "grad_norm": 13.350455284118652, "learning_rate": 1.7947683917901127e-06, "loss": 0.8758, "step": 15135 }, { "epoch": 2.44, "grad_norm": 13.359879493713379, "learning_rate": 1.7937750569435232e-06, "loss": 0.6913, "step": 15136 }, { "epoch": 2.44, "grad_norm": 12.827271461486816, "learning_rate": 1.7927819699752224e-06, "loss": 0.6186, "step": 15137 }, { "epoch": 2.44, "grad_norm": 9.953083038330078, "learning_rate": 1.7917891309152148e-06, "loss": 0.6131, "step": 15138 }, { "epoch": 2.44, "grad_norm": 12.729016304016113, "learning_rate": 1.7907965397934868e-06, "loss": 0.638, "step": 15139 }, { "epoch": 2.44, "grad_norm": 13.171788215637207, "learning_rate": 1.7898041966400193e-06, "loss": 0.7377, "step": 15140 }, { "epoch": 2.44, "grad_norm": 13.369694709777832, "learning_rate": 1.7888121014847882e-06, "loss": 0.5456, "step": 15141 }, { "epoch": 2.44, "grad_norm": 11.745532989501953, "learning_rate": 1.787820254357765e-06, "loss": 0.5854, "step": 15142 }, { "epoch": 2.44, "grad_norm": 13.949707984924316, "learning_rate": 1.7868286552889025e-06, "loss": 0.9009, "step": 15143 }, { "epoch": 2.44, "grad_norm": 12.71138858795166, "learning_rate": 1.78583730430816e-06, "loss": 0.7801, "step": 15144 }, { "epoch": 2.44, "grad_norm": 10.437506675720215, "learning_rate": 1.7848462014454758e-06, "loss": 0.6407, "step": 15145 }, { "epoch": 2.44, "grad_norm": 12.6980619430542, "learning_rate": 1.783855346730794e-06, "loss": 0.8028, "step": 15146 }, { "epoch": 2.44, "grad_norm": 14.13335132598877, "learning_rate": 1.7828647401940401e-06, "loss": 0.8041, "step": 15147 }, { "epoch": 2.44, "grad_norm": 13.36018180847168, "learning_rate": 1.7818743818651353e-06, "loss": 0.6171, "step": 15148 }, { "epoch": 2.44, "grad_norm": 12.473706245422363, "learning_rate": 1.7808842717739984e-06, "loss": 0.7623, "step": 15149 }, { "epoch": 2.44, "grad_norm": 11.17952823638916, "learning_rate": 1.7798944099505355e-06, "loss": 0.783, "step": 15150 }, { "epoch": 2.44, "grad_norm": 13.485700607299805, "learning_rate": 1.7789047964246441e-06, "loss": 0.5838, "step": 15151 }, { "epoch": 2.44, "grad_norm": 12.827330589294434, "learning_rate": 1.7779154312262204e-06, "loss": 0.644, "step": 15152 }, { "epoch": 2.44, "grad_norm": 14.211213111877441, "learning_rate": 1.7769263143851444e-06, "loss": 0.6494, "step": 15153 }, { "epoch": 2.44, "grad_norm": 12.612112998962402, "learning_rate": 1.7759374459313007e-06, "loss": 0.6245, "step": 15154 }, { "epoch": 2.44, "grad_norm": 13.091064453125, "learning_rate": 1.7749488258945535e-06, "loss": 0.7393, "step": 15155 }, { "epoch": 2.44, "grad_norm": 14.912373542785645, "learning_rate": 1.773960454304764e-06, "loss": 0.5843, "step": 15156 }, { "epoch": 2.44, "grad_norm": 15.836407661437988, "learning_rate": 1.7729723311917923e-06, "loss": 0.6602, "step": 15157 }, { "epoch": 2.44, "grad_norm": 13.313238143920898, "learning_rate": 1.7719844565854815e-06, "loss": 0.6602, "step": 15158 }, { "epoch": 2.44, "grad_norm": 12.898643493652344, "learning_rate": 1.7709968305156756e-06, "loss": 0.8547, "step": 15159 }, { "epoch": 2.44, "grad_norm": 14.19674015045166, "learning_rate": 1.7700094530122037e-06, "loss": 0.7695, "step": 15160 }, { "epoch": 2.44, "grad_norm": 13.448600769042969, "learning_rate": 1.7690223241048887e-06, "loss": 0.6629, "step": 15161 }, { "epoch": 2.44, "grad_norm": 11.20067024230957, "learning_rate": 1.7680354438235515e-06, "loss": 0.7476, "step": 15162 }, { "epoch": 2.44, "grad_norm": 13.939240455627441, "learning_rate": 1.7670488121980046e-06, "loss": 0.8639, "step": 15163 }, { "epoch": 2.44, "grad_norm": 11.292817115783691, "learning_rate": 1.766062429258043e-06, "loss": 0.7703, "step": 15164 }, { "epoch": 2.44, "grad_norm": 14.192742347717285, "learning_rate": 1.7650762950334677e-06, "loss": 0.6017, "step": 15165 }, { "epoch": 2.44, "grad_norm": 15.72856330871582, "learning_rate": 1.7640904095540646e-06, "loss": 0.7127, "step": 15166 }, { "epoch": 2.44, "grad_norm": 11.809040069580078, "learning_rate": 1.7631047728496099e-06, "loss": 0.7481, "step": 15167 }, { "epoch": 2.44, "grad_norm": 14.067428588867188, "learning_rate": 1.7621193849498808e-06, "loss": 0.8002, "step": 15168 }, { "epoch": 2.44, "grad_norm": 13.830978393554688, "learning_rate": 1.761134245884637e-06, "loss": 0.7418, "step": 15169 }, { "epoch": 2.44, "grad_norm": 11.965044975280762, "learning_rate": 1.7601493556836412e-06, "loss": 0.7161, "step": 15170 }, { "epoch": 2.44, "grad_norm": 11.833348274230957, "learning_rate": 1.759164714376641e-06, "loss": 0.7237, "step": 15171 }, { "epoch": 2.44, "grad_norm": 13.656649589538574, "learning_rate": 1.7581803219933747e-06, "loss": 0.838, "step": 15172 }, { "epoch": 2.44, "grad_norm": 11.7652587890625, "learning_rate": 1.7571961785635827e-06, "loss": 0.9219, "step": 15173 }, { "epoch": 2.44, "grad_norm": 11.691814422607422, "learning_rate": 1.7562122841169904e-06, "loss": 0.5578, "step": 15174 }, { "epoch": 2.44, "grad_norm": 14.47452163696289, "learning_rate": 1.755228638683315e-06, "loss": 0.8305, "step": 15175 }, { "epoch": 2.44, "grad_norm": 12.030233383178711, "learning_rate": 1.7542452422922718e-06, "loss": 0.6151, "step": 15176 }, { "epoch": 2.44, "grad_norm": 18.350753784179688, "learning_rate": 1.7532620949735625e-06, "loss": 0.9022, "step": 15177 }, { "epoch": 2.44, "grad_norm": 10.392240524291992, "learning_rate": 1.7522791967568875e-06, "loss": 0.6676, "step": 15178 }, { "epoch": 2.44, "grad_norm": 10.760268211364746, "learning_rate": 1.7512965476719347e-06, "loss": 0.7418, "step": 15179 }, { "epoch": 2.44, "grad_norm": 13.321133613586426, "learning_rate": 1.7503141477483843e-06, "loss": 0.8366, "step": 15180 }, { "epoch": 2.44, "grad_norm": 11.144084930419922, "learning_rate": 1.7493319970159118e-06, "loss": 0.7293, "step": 15181 }, { "epoch": 2.44, "grad_norm": 12.425421714782715, "learning_rate": 1.7483500955041877e-06, "loss": 0.6429, "step": 15182 }, { "epoch": 2.44, "grad_norm": 13.171883583068848, "learning_rate": 1.747368443242865e-06, "loss": 0.9346, "step": 15183 }, { "epoch": 2.44, "grad_norm": 12.210801124572754, "learning_rate": 1.7463870402616035e-06, "loss": 0.6191, "step": 15184 }, { "epoch": 2.44, "grad_norm": 13.00148868560791, "learning_rate": 1.7454058865900425e-06, "loss": 0.6705, "step": 15185 }, { "epoch": 2.44, "grad_norm": 14.294443130493164, "learning_rate": 1.7444249822578163e-06, "loss": 0.7957, "step": 15186 }, { "epoch": 2.44, "grad_norm": 18.229816436767578, "learning_rate": 1.743444327294561e-06, "loss": 0.7113, "step": 15187 }, { "epoch": 2.44, "grad_norm": 16.531814575195312, "learning_rate": 1.7424639217298933e-06, "loss": 0.5741, "step": 15188 }, { "epoch": 2.45, "grad_norm": 14.426569938659668, "learning_rate": 1.7414837655934303e-06, "loss": 0.6953, "step": 15189 }, { "epoch": 2.45, "grad_norm": 12.459735870361328, "learning_rate": 1.7405038589147783e-06, "loss": 0.7416, "step": 15190 }, { "epoch": 2.45, "grad_norm": 11.31510066986084, "learning_rate": 1.7395242017235337e-06, "loss": 0.8059, "step": 15191 }, { "epoch": 2.45, "grad_norm": 15.132434844970703, "learning_rate": 1.7385447940492916e-06, "loss": 0.8419, "step": 15192 }, { "epoch": 2.45, "grad_norm": 13.667401313781738, "learning_rate": 1.7375656359216331e-06, "loss": 0.7318, "step": 15193 }, { "epoch": 2.45, "grad_norm": 13.602690696716309, "learning_rate": 1.7365867273701386e-06, "loss": 0.724, "step": 15194 }, { "epoch": 2.45, "grad_norm": 12.617474555969238, "learning_rate": 1.7356080684243747e-06, "loss": 0.6013, "step": 15195 }, { "epoch": 2.45, "grad_norm": 15.964821815490723, "learning_rate": 1.7346296591139012e-06, "loss": 0.6269, "step": 15196 }, { "epoch": 2.45, "grad_norm": 11.750268936157227, "learning_rate": 1.7336514994682751e-06, "loss": 0.5588, "step": 15197 }, { "epoch": 2.45, "grad_norm": 13.128682136535645, "learning_rate": 1.7326735895170433e-06, "loss": 0.5988, "step": 15198 }, { "epoch": 2.45, "grad_norm": 11.999489784240723, "learning_rate": 1.7316959292897406e-06, "loss": 0.5485, "step": 15199 }, { "epoch": 2.45, "grad_norm": 13.216214179992676, "learning_rate": 1.7307185188159026e-06, "loss": 0.6002, "step": 15200 }, { "epoch": 2.45, "grad_norm": 11.672734260559082, "learning_rate": 1.7297413581250488e-06, "loss": 0.7405, "step": 15201 }, { "epoch": 2.45, "grad_norm": 13.365192413330078, "learning_rate": 1.728764447246698e-06, "loss": 0.5886, "step": 15202 }, { "epoch": 2.45, "grad_norm": 12.190292358398438, "learning_rate": 1.727787786210362e-06, "loss": 0.7184, "step": 15203 }, { "epoch": 2.45, "grad_norm": 13.956670761108398, "learning_rate": 1.7268113750455363e-06, "loss": 0.5707, "step": 15204 }, { "epoch": 2.45, "grad_norm": 13.484271049499512, "learning_rate": 1.725835213781718e-06, "loss": 0.7453, "step": 15205 }, { "epoch": 2.45, "grad_norm": 11.968306541442871, "learning_rate": 1.724859302448394e-06, "loss": 0.8301, "step": 15206 }, { "epoch": 2.45, "grad_norm": 14.938620567321777, "learning_rate": 1.723883641075037e-06, "loss": 0.7723, "step": 15207 }, { "epoch": 2.45, "grad_norm": 12.202920913696289, "learning_rate": 1.7229082296911259e-06, "loss": 0.7629, "step": 15208 }, { "epoch": 2.45, "grad_norm": 14.942760467529297, "learning_rate": 1.7219330683261192e-06, "loss": 0.711, "step": 15209 }, { "epoch": 2.45, "grad_norm": 14.694249153137207, "learning_rate": 1.7209581570094714e-06, "loss": 0.5774, "step": 15210 }, { "epoch": 2.45, "grad_norm": 12.094614028930664, "learning_rate": 1.7199834957706352e-06, "loss": 0.8741, "step": 15211 }, { "epoch": 2.45, "grad_norm": 14.108295440673828, "learning_rate": 1.7190090846390473e-06, "loss": 0.6413, "step": 15212 }, { "epoch": 2.45, "grad_norm": 14.26606559753418, "learning_rate": 1.7180349236441451e-06, "loss": 0.6803, "step": 15213 }, { "epoch": 2.45, "grad_norm": 11.608448028564453, "learning_rate": 1.717061012815352e-06, "loss": 0.6304, "step": 15214 }, { "epoch": 2.45, "grad_norm": 12.38991928100586, "learning_rate": 1.7160873521820831e-06, "loss": 0.575, "step": 15215 }, { "epoch": 2.45, "grad_norm": 13.394113540649414, "learning_rate": 1.715113941773755e-06, "loss": 0.6299, "step": 15216 }, { "epoch": 2.45, "grad_norm": 12.336610794067383, "learning_rate": 1.7141407816197664e-06, "loss": 0.8931, "step": 15217 }, { "epoch": 2.45, "grad_norm": 14.77391529083252, "learning_rate": 1.7131678717495125e-06, "loss": 0.8636, "step": 15218 }, { "epoch": 2.45, "grad_norm": 11.118376731872559, "learning_rate": 1.7121952121923846e-06, "loss": 0.6325, "step": 15219 }, { "epoch": 2.45, "grad_norm": 12.869673728942871, "learning_rate": 1.7112228029777578e-06, "loss": 0.627, "step": 15220 }, { "epoch": 2.45, "grad_norm": 11.224828720092773, "learning_rate": 1.7102506441350108e-06, "loss": 0.6207, "step": 15221 }, { "epoch": 2.45, "grad_norm": 12.883739471435547, "learning_rate": 1.7092787356935026e-06, "loss": 0.6181, "step": 15222 }, { "epoch": 2.45, "grad_norm": 12.77195930480957, "learning_rate": 1.708307077682595e-06, "loss": 0.673, "step": 15223 }, { "epoch": 2.45, "grad_norm": 13.694755554199219, "learning_rate": 1.707335670131639e-06, "loss": 0.6077, "step": 15224 }, { "epoch": 2.45, "grad_norm": 12.83258056640625, "learning_rate": 1.7063645130699757e-06, "loss": 0.5943, "step": 15225 }, { "epoch": 2.45, "grad_norm": 15.712864875793457, "learning_rate": 1.7053936065269361e-06, "loss": 0.6447, "step": 15226 }, { "epoch": 2.45, "grad_norm": 12.731231689453125, "learning_rate": 1.7044229505318532e-06, "loss": 0.8056, "step": 15227 }, { "epoch": 2.45, "grad_norm": 13.077162742614746, "learning_rate": 1.703452545114045e-06, "loss": 0.6226, "step": 15228 }, { "epoch": 2.45, "grad_norm": 10.434369087219238, "learning_rate": 1.7024823903028199e-06, "loss": 0.5691, "step": 15229 }, { "epoch": 2.45, "grad_norm": 12.086158752441406, "learning_rate": 1.701512486127489e-06, "loss": 0.8303, "step": 15230 }, { "epoch": 2.45, "grad_norm": 9.717229843139648, "learning_rate": 1.7005428326173424e-06, "loss": 0.6365, "step": 15231 }, { "epoch": 2.45, "grad_norm": 12.509182929992676, "learning_rate": 1.6995734298016763e-06, "loss": 0.6393, "step": 15232 }, { "epoch": 2.45, "grad_norm": 12.546295166015625, "learning_rate": 1.69860427770977e-06, "loss": 0.538, "step": 15233 }, { "epoch": 2.45, "grad_norm": 11.588700294494629, "learning_rate": 1.6976353763708942e-06, "loss": 0.4861, "step": 15234 }, { "epoch": 2.45, "grad_norm": 20.506717681884766, "learning_rate": 1.696666725814321e-06, "loss": 0.619, "step": 15235 }, { "epoch": 2.45, "grad_norm": 12.513443946838379, "learning_rate": 1.6956983260693061e-06, "loss": 0.6385, "step": 15236 }, { "epoch": 2.45, "grad_norm": 14.303053855895996, "learning_rate": 1.6947301771651036e-06, "loss": 0.7255, "step": 15237 }, { "epoch": 2.45, "grad_norm": 11.78548812866211, "learning_rate": 1.693762279130957e-06, "loss": 0.7773, "step": 15238 }, { "epoch": 2.45, "grad_norm": 12.92032241821289, "learning_rate": 1.6927946319960998e-06, "loss": 0.7015, "step": 15239 }, { "epoch": 2.45, "grad_norm": 13.485921859741211, "learning_rate": 1.6918272357897647e-06, "loss": 0.673, "step": 15240 }, { "epoch": 2.45, "grad_norm": 12.961800575256348, "learning_rate": 1.6908600905411722e-06, "loss": 0.7622, "step": 15241 }, { "epoch": 2.45, "grad_norm": 11.51069450378418, "learning_rate": 1.6898931962795306e-06, "loss": 0.7714, "step": 15242 }, { "epoch": 2.45, "grad_norm": 14.363199234008789, "learning_rate": 1.6889265530340549e-06, "loss": 0.6648, "step": 15243 }, { "epoch": 2.45, "grad_norm": 13.613775253295898, "learning_rate": 1.6879601608339392e-06, "loss": 0.6105, "step": 15244 }, { "epoch": 2.45, "grad_norm": 14.091550827026367, "learning_rate": 1.6869940197083734e-06, "loss": 0.6112, "step": 15245 }, { "epoch": 2.45, "grad_norm": 15.301308631896973, "learning_rate": 1.6860281296865432e-06, "loss": 0.7012, "step": 15246 }, { "epoch": 2.45, "grad_norm": 11.843894958496094, "learning_rate": 1.685062490797622e-06, "loss": 0.5961, "step": 15247 }, { "epoch": 2.45, "grad_norm": 13.173165321350098, "learning_rate": 1.684097103070782e-06, "loss": 0.6876, "step": 15248 }, { "epoch": 2.45, "grad_norm": 12.991406440734863, "learning_rate": 1.6831319665351808e-06, "loss": 0.7679, "step": 15249 }, { "epoch": 2.45, "grad_norm": 13.653010368347168, "learning_rate": 1.6821670812199697e-06, "loss": 0.6835, "step": 15250 }, { "epoch": 2.46, "grad_norm": 12.073171615600586, "learning_rate": 1.6812024471542987e-06, "loss": 0.8132, "step": 15251 }, { "epoch": 2.46, "grad_norm": 13.313260078430176, "learning_rate": 1.6802380643673034e-06, "loss": 0.5833, "step": 15252 }, { "epoch": 2.46, "grad_norm": 13.044588088989258, "learning_rate": 1.6792739328881113e-06, "loss": 0.5179, "step": 15253 }, { "epoch": 2.46, "grad_norm": 13.36376953125, "learning_rate": 1.6783100527458495e-06, "loss": 0.7863, "step": 15254 }, { "epoch": 2.46, "grad_norm": 14.845663070678711, "learning_rate": 1.6773464239696301e-06, "loss": 0.6394, "step": 15255 }, { "epoch": 2.46, "grad_norm": 11.035890579223633, "learning_rate": 1.6763830465885632e-06, "loss": 0.6653, "step": 15256 }, { "epoch": 2.46, "grad_norm": 11.439372062683105, "learning_rate": 1.6754199206317478e-06, "loss": 0.6644, "step": 15257 }, { "epoch": 2.46, "grad_norm": 12.03481674194336, "learning_rate": 1.6744570461282739e-06, "loss": 0.5548, "step": 15258 }, { "epoch": 2.46, "grad_norm": 14.31899642944336, "learning_rate": 1.6734944231072292e-06, "loss": 0.7258, "step": 15259 }, { "epoch": 2.46, "grad_norm": 13.916017532348633, "learning_rate": 1.6725320515976885e-06, "loss": 0.4846, "step": 15260 }, { "epoch": 2.46, "grad_norm": 12.692648887634277, "learning_rate": 1.6715699316287236e-06, "loss": 0.7758, "step": 15261 }, { "epoch": 2.46, "grad_norm": 13.697608947753906, "learning_rate": 1.6706080632293942e-06, "loss": 0.6112, "step": 15262 }, { "epoch": 2.46, "grad_norm": 11.453349113464355, "learning_rate": 1.6696464464287577e-06, "loss": 0.6065, "step": 15263 }, { "epoch": 2.46, "grad_norm": 12.81518840789795, "learning_rate": 1.6686850812558565e-06, "loss": 0.6554, "step": 15264 }, { "epoch": 2.46, "grad_norm": 14.155313491821289, "learning_rate": 1.667723967739734e-06, "loss": 0.8494, "step": 15265 }, { "epoch": 2.46, "grad_norm": 13.873418807983398, "learning_rate": 1.666763105909418e-06, "loss": 0.8777, "step": 15266 }, { "epoch": 2.46, "grad_norm": 13.588421821594238, "learning_rate": 1.6658024957939357e-06, "loss": 0.7008, "step": 15267 }, { "epoch": 2.46, "grad_norm": 14.688725471496582, "learning_rate": 1.6648421374223022e-06, "loss": 0.7539, "step": 15268 }, { "epoch": 2.46, "grad_norm": 13.407025337219238, "learning_rate": 1.6638820308235237e-06, "loss": 0.5332, "step": 15269 }, { "epoch": 2.46, "grad_norm": 13.529122352600098, "learning_rate": 1.6629221760266068e-06, "loss": 0.7197, "step": 15270 }, { "epoch": 2.46, "grad_norm": 13.6989107131958, "learning_rate": 1.6619625730605383e-06, "loss": 0.6879, "step": 15271 }, { "epoch": 2.46, "grad_norm": 12.319098472595215, "learning_rate": 1.6610032219543105e-06, "loss": 0.5878, "step": 15272 }, { "epoch": 2.46, "grad_norm": 13.574219703674316, "learning_rate": 1.6600441227368979e-06, "loss": 0.7319, "step": 15273 }, { "epoch": 2.46, "grad_norm": 17.24573516845703, "learning_rate": 1.65908527543727e-06, "loss": 0.6087, "step": 15274 }, { "epoch": 2.46, "grad_norm": 10.45390796661377, "learning_rate": 1.6581266800843942e-06, "loss": 0.4641, "step": 15275 }, { "epoch": 2.46, "grad_norm": 13.287314414978027, "learning_rate": 1.6571683367072243e-06, "loss": 0.7503, "step": 15276 }, { "epoch": 2.46, "grad_norm": 17.510683059692383, "learning_rate": 1.656210245334704e-06, "loss": 0.6894, "step": 15277 }, { "epoch": 2.46, "grad_norm": 12.685973167419434, "learning_rate": 1.6552524059957797e-06, "loss": 0.8889, "step": 15278 }, { "epoch": 2.46, "grad_norm": 11.624869346618652, "learning_rate": 1.6542948187193786e-06, "loss": 0.6088, "step": 15279 }, { "epoch": 2.46, "grad_norm": 11.790342330932617, "learning_rate": 1.6533374835344319e-06, "loss": 0.6579, "step": 15280 }, { "epoch": 2.46, "grad_norm": 13.200867652893066, "learning_rate": 1.6523804004698529e-06, "loss": 0.6455, "step": 15281 }, { "epoch": 2.46, "grad_norm": 15.8153657913208, "learning_rate": 1.651423569554549e-06, "loss": 0.5997, "step": 15282 }, { "epoch": 2.46, "grad_norm": 12.354574203491211, "learning_rate": 1.6504669908174254e-06, "loss": 0.7077, "step": 15283 }, { "epoch": 2.46, "grad_norm": 15.90696907043457, "learning_rate": 1.6495106642873792e-06, "loss": 0.6803, "step": 15284 }, { "epoch": 2.46, "grad_norm": 13.988961219787598, "learning_rate": 1.6485545899932932e-06, "loss": 0.6514, "step": 15285 }, { "epoch": 2.46, "grad_norm": 13.017922401428223, "learning_rate": 1.6475987679640493e-06, "loss": 0.6033, "step": 15286 }, { "epoch": 2.46, "grad_norm": 12.58869743347168, "learning_rate": 1.646643198228518e-06, "loss": 0.6389, "step": 15287 }, { "epoch": 2.46, "grad_norm": 13.149908065795898, "learning_rate": 1.645687880815562e-06, "loss": 0.8106, "step": 15288 }, { "epoch": 2.46, "grad_norm": 30.092411041259766, "learning_rate": 1.644732815754041e-06, "loss": 0.8256, "step": 15289 }, { "epoch": 2.46, "grad_norm": 12.214750289916992, "learning_rate": 1.6437780030728001e-06, "loss": 0.601, "step": 15290 }, { "epoch": 2.46, "grad_norm": 12.862939834594727, "learning_rate": 1.642823442800684e-06, "loss": 0.7257, "step": 15291 }, { "epoch": 2.46, "grad_norm": 13.319385528564453, "learning_rate": 1.641869134966525e-06, "loss": 0.634, "step": 15292 }, { "epoch": 2.46, "grad_norm": 13.740805625915527, "learning_rate": 1.6409150795991468e-06, "loss": 0.8686, "step": 15293 }, { "epoch": 2.46, "grad_norm": 15.10390853881836, "learning_rate": 1.639961276727372e-06, "loss": 0.7628, "step": 15294 }, { "epoch": 2.46, "grad_norm": 11.221640586853027, "learning_rate": 1.6390077263800086e-06, "loss": 0.699, "step": 15295 }, { "epoch": 2.46, "grad_norm": 14.025724411010742, "learning_rate": 1.6380544285858568e-06, "loss": 0.838, "step": 15296 }, { "epoch": 2.46, "grad_norm": 15.277618408203125, "learning_rate": 1.6371013833737181e-06, "loss": 0.9634, "step": 15297 }, { "epoch": 2.46, "grad_norm": 13.229567527770996, "learning_rate": 1.6361485907723763e-06, "loss": 0.7006, "step": 15298 }, { "epoch": 2.46, "grad_norm": 12.500696182250977, "learning_rate": 1.6351960508106136e-06, "loss": 0.6101, "step": 15299 }, { "epoch": 2.46, "grad_norm": 12.878235816955566, "learning_rate": 1.6342437635172025e-06, "loss": 0.6585, "step": 15300 }, { "epoch": 2.46, "grad_norm": 10.536783218383789, "learning_rate": 1.633291728920904e-06, "loss": 0.6706, "step": 15301 }, { "epoch": 2.46, "grad_norm": 12.068364143371582, "learning_rate": 1.6323399470504808e-06, "loss": 0.8632, "step": 15302 }, { "epoch": 2.46, "grad_norm": 11.867012977600098, "learning_rate": 1.6313884179346784e-06, "loss": 0.839, "step": 15303 }, { "epoch": 2.46, "grad_norm": 11.45445728302002, "learning_rate": 1.6304371416022403e-06, "loss": 0.7565, "step": 15304 }, { "epoch": 2.46, "grad_norm": 13.35621166229248, "learning_rate": 1.6294861180819033e-06, "loss": 0.6892, "step": 15305 }, { "epoch": 2.46, "grad_norm": 12.567089080810547, "learning_rate": 1.6285353474023925e-06, "loss": 0.612, "step": 15306 }, { "epoch": 2.46, "grad_norm": 15.09801197052002, "learning_rate": 1.6275848295924246e-06, "loss": 0.8666, "step": 15307 }, { "epoch": 2.46, "grad_norm": 16.01844596862793, "learning_rate": 1.6266345646807147e-06, "loss": 0.7733, "step": 15308 }, { "epoch": 2.46, "grad_norm": 15.843323707580566, "learning_rate": 1.625684552695962e-06, "loss": 0.6114, "step": 15309 }, { "epoch": 2.46, "grad_norm": 13.01530647277832, "learning_rate": 1.624734793666869e-06, "loss": 0.8751, "step": 15310 }, { "epoch": 2.46, "grad_norm": 13.161575317382812, "learning_rate": 1.6237852876221204e-06, "loss": 0.7404, "step": 15311 }, { "epoch": 2.46, "grad_norm": 12.328112602233887, "learning_rate": 1.6228360345903949e-06, "loss": 0.7965, "step": 15312 }, { "epoch": 2.47, "grad_norm": 11.805947303771973, "learning_rate": 1.6218870346003713e-06, "loss": 0.5479, "step": 15313 }, { "epoch": 2.47, "grad_norm": 14.286974906921387, "learning_rate": 1.620938287680709e-06, "loss": 0.606, "step": 15314 }, { "epoch": 2.47, "grad_norm": 12.150994300842285, "learning_rate": 1.6199897938600729e-06, "loss": 0.4889, "step": 15315 }, { "epoch": 2.47, "grad_norm": 11.058347702026367, "learning_rate": 1.619041553167109e-06, "loss": 0.7608, "step": 15316 }, { "epoch": 2.47, "grad_norm": 15.060937881469727, "learning_rate": 1.6180935656304586e-06, "loss": 0.8612, "step": 15317 }, { "epoch": 2.47, "grad_norm": 13.242809295654297, "learning_rate": 1.6171458312787603e-06, "loss": 0.8039, "step": 15318 }, { "epoch": 2.47, "grad_norm": 12.9075927734375, "learning_rate": 1.616198350140641e-06, "loss": 0.6674, "step": 15319 }, { "epoch": 2.47, "grad_norm": 12.339947700500488, "learning_rate": 1.6152511222447165e-06, "loss": 0.6015, "step": 15320 }, { "epoch": 2.47, "grad_norm": 13.616165161132812, "learning_rate": 1.6143041476196053e-06, "loss": 0.7169, "step": 15321 }, { "epoch": 2.47, "grad_norm": 12.329244613647461, "learning_rate": 1.6133574262939044e-06, "loss": 0.6328, "step": 15322 }, { "epoch": 2.47, "grad_norm": 12.096589088439941, "learning_rate": 1.6124109582962154e-06, "loss": 0.6901, "step": 15323 }, { "epoch": 2.47, "grad_norm": 12.126250267028809, "learning_rate": 1.61146474365513e-06, "loss": 0.5151, "step": 15324 }, { "epoch": 2.47, "grad_norm": 11.114460945129395, "learning_rate": 1.610518782399223e-06, "loss": 0.7873, "step": 15325 }, { "epoch": 2.47, "grad_norm": 12.905040740966797, "learning_rate": 1.6095730745570736e-06, "loss": 0.9603, "step": 15326 }, { "epoch": 2.47, "grad_norm": 12.647748947143555, "learning_rate": 1.608627620157246e-06, "loss": 0.6875, "step": 15327 }, { "epoch": 2.47, "grad_norm": 14.104867935180664, "learning_rate": 1.6076824192282959e-06, "loss": 0.8053, "step": 15328 }, { "epoch": 2.47, "grad_norm": 14.300207138061523, "learning_rate": 1.6067374717987805e-06, "loss": 0.7338, "step": 15329 }, { "epoch": 2.47, "grad_norm": 14.4021635055542, "learning_rate": 1.6057927778972382e-06, "loss": 0.5968, "step": 15330 }, { "epoch": 2.47, "grad_norm": 11.186334609985352, "learning_rate": 1.6048483375522038e-06, "loss": 0.7484, "step": 15331 }, { "epoch": 2.47, "grad_norm": 13.332114219665527, "learning_rate": 1.6039041507922092e-06, "loss": 0.8387, "step": 15332 }, { "epoch": 2.47, "grad_norm": 11.269887924194336, "learning_rate": 1.6029602176457704e-06, "loss": 0.6224, "step": 15333 }, { "epoch": 2.47, "grad_norm": 16.638578414916992, "learning_rate": 1.6020165381414042e-06, "loss": 0.7589, "step": 15334 }, { "epoch": 2.47, "grad_norm": 13.19171142578125, "learning_rate": 1.6010731123076139e-06, "loss": 0.7283, "step": 15335 }, { "epoch": 2.47, "grad_norm": 10.609964370727539, "learning_rate": 1.600129940172893e-06, "loss": 0.6124, "step": 15336 }, { "epoch": 2.47, "grad_norm": 14.30286979675293, "learning_rate": 1.599187021765738e-06, "loss": 0.7162, "step": 15337 }, { "epoch": 2.47, "grad_norm": 13.949379920959473, "learning_rate": 1.5982443571146244e-06, "loss": 0.7774, "step": 15338 }, { "epoch": 2.47, "grad_norm": 16.904232025146484, "learning_rate": 1.597301946248031e-06, "loss": 0.6533, "step": 15339 }, { "epoch": 2.47, "grad_norm": 10.702392578125, "learning_rate": 1.5963597891944227e-06, "loss": 0.5895, "step": 15340 }, { "epoch": 2.47, "grad_norm": 11.062052726745605, "learning_rate": 1.5954178859822556e-06, "loss": 0.656, "step": 15341 }, { "epoch": 2.47, "grad_norm": 10.966934204101562, "learning_rate": 1.5944762366399869e-06, "loss": 0.755, "step": 15342 }, { "epoch": 2.47, "grad_norm": 10.851221084594727, "learning_rate": 1.593534841196056e-06, "loss": 0.7268, "step": 15343 }, { "epoch": 2.47, "grad_norm": 12.307884216308594, "learning_rate": 1.592593699678896e-06, "loss": 0.6697, "step": 15344 }, { "epoch": 2.47, "grad_norm": 12.774857521057129, "learning_rate": 1.5916528121169427e-06, "loss": 0.6906, "step": 15345 }, { "epoch": 2.47, "grad_norm": 12.684959411621094, "learning_rate": 1.5907121785386136e-06, "loss": 0.6927, "step": 15346 }, { "epoch": 2.47, "grad_norm": 14.684557914733887, "learning_rate": 1.5897717989723182e-06, "loss": 0.5396, "step": 15347 }, { "epoch": 2.47, "grad_norm": 13.585108757019043, "learning_rate": 1.588831673446467e-06, "loss": 0.604, "step": 15348 }, { "epoch": 2.47, "grad_norm": 10.639694213867188, "learning_rate": 1.5878918019894519e-06, "loss": 0.6323, "step": 15349 }, { "epoch": 2.47, "grad_norm": 13.552522659301758, "learning_rate": 1.5869521846296686e-06, "loss": 0.5326, "step": 15350 }, { "epoch": 2.47, "grad_norm": 14.487656593322754, "learning_rate": 1.5860128213954962e-06, "loss": 0.7261, "step": 15351 }, { "epoch": 2.47, "grad_norm": 15.091046333312988, "learning_rate": 1.5850737123153082e-06, "loss": 0.7767, "step": 15352 }, { "epoch": 2.47, "grad_norm": 13.977635383605957, "learning_rate": 1.5841348574174743e-06, "loss": 0.8195, "step": 15353 }, { "epoch": 2.47, "grad_norm": 12.410151481628418, "learning_rate": 1.5831962567303515e-06, "loss": 0.6577, "step": 15354 }, { "epoch": 2.47, "grad_norm": 12.396675109863281, "learning_rate": 1.582257910282291e-06, "loss": 0.9021, "step": 15355 }, { "epoch": 2.47, "grad_norm": 11.95895767211914, "learning_rate": 1.5813198181016388e-06, "loss": 0.678, "step": 15356 }, { "epoch": 2.47, "grad_norm": 13.499462127685547, "learning_rate": 1.5803819802167286e-06, "loss": 0.7051, "step": 15357 }, { "epoch": 2.47, "grad_norm": 10.66985034942627, "learning_rate": 1.5794443966558915e-06, "loss": 0.4936, "step": 15358 }, { "epoch": 2.47, "grad_norm": 11.775331497192383, "learning_rate": 1.5785070674474468e-06, "loss": 0.7657, "step": 15359 }, { "epoch": 2.47, "grad_norm": 11.192644119262695, "learning_rate": 1.5775699926197064e-06, "loss": 0.6801, "step": 15360 }, { "epoch": 2.47, "grad_norm": 12.812450408935547, "learning_rate": 1.5766331722009775e-06, "loss": 0.5598, "step": 15361 }, { "epoch": 2.47, "grad_norm": 12.330737113952637, "learning_rate": 1.575696606219559e-06, "loss": 0.6342, "step": 15362 }, { "epoch": 2.47, "grad_norm": 13.496638298034668, "learning_rate": 1.5747602947037354e-06, "loss": 0.8274, "step": 15363 }, { "epoch": 2.47, "grad_norm": 13.667977333068848, "learning_rate": 1.5738242376817936e-06, "loss": 0.9523, "step": 15364 }, { "epoch": 2.47, "grad_norm": 13.633906364440918, "learning_rate": 1.5728884351820095e-06, "loss": 0.7176, "step": 15365 }, { "epoch": 2.47, "grad_norm": 12.244625091552734, "learning_rate": 1.571952887232646e-06, "loss": 0.7791, "step": 15366 }, { "epoch": 2.47, "grad_norm": 13.146330833435059, "learning_rate": 1.5710175938619664e-06, "loss": 0.7967, "step": 15367 }, { "epoch": 2.47, "grad_norm": 17.20656394958496, "learning_rate": 1.5700825550982191e-06, "loss": 0.7704, "step": 15368 }, { "epoch": 2.47, "grad_norm": 13.853477478027344, "learning_rate": 1.56914777096965e-06, "loss": 0.6688, "step": 15369 }, { "epoch": 2.47, "grad_norm": 13.772408485412598, "learning_rate": 1.568213241504496e-06, "loss": 0.9472, "step": 15370 }, { "epoch": 2.47, "grad_norm": 14.311575889587402, "learning_rate": 1.5672789667309807e-06, "loss": 0.7747, "step": 15371 }, { "epoch": 2.47, "grad_norm": 14.849573135375977, "learning_rate": 1.5663449466773307e-06, "loss": 0.7898, "step": 15372 }, { "epoch": 2.47, "grad_norm": 15.68067741394043, "learning_rate": 1.5654111813717576e-06, "loss": 0.6065, "step": 15373 }, { "epoch": 2.47, "grad_norm": 13.271224975585938, "learning_rate": 1.5644776708424635e-06, "loss": 0.6925, "step": 15374 }, { "epoch": 2.48, "grad_norm": 18.629663467407227, "learning_rate": 1.5635444151176503e-06, "loss": 0.7692, "step": 15375 }, { "epoch": 2.48, "grad_norm": 12.199653625488281, "learning_rate": 1.5626114142255045e-06, "loss": 0.5466, "step": 15376 }, { "epoch": 2.48, "grad_norm": 14.44385814666748, "learning_rate": 1.5616786681942131e-06, "loss": 0.6844, "step": 15377 }, { "epoch": 2.48, "grad_norm": 15.25374698638916, "learning_rate": 1.5607461770519484e-06, "loss": 0.6772, "step": 15378 }, { "epoch": 2.48, "grad_norm": 11.394113540649414, "learning_rate": 1.5598139408268732e-06, "loss": 0.7251, "step": 15379 }, { "epoch": 2.48, "grad_norm": 12.91492748260498, "learning_rate": 1.5588819595471537e-06, "loss": 0.7619, "step": 15380 }, { "epoch": 2.48, "grad_norm": 13.795732498168945, "learning_rate": 1.5579502332409357e-06, "loss": 0.7332, "step": 15381 }, { "epoch": 2.48, "grad_norm": 12.343878746032715, "learning_rate": 1.5570187619363685e-06, "loss": 0.8011, "step": 15382 }, { "epoch": 2.48, "grad_norm": 12.626893997192383, "learning_rate": 1.5560875456615853e-06, "loss": 0.698, "step": 15383 }, { "epoch": 2.48, "grad_norm": 15.08489990234375, "learning_rate": 1.5551565844447124e-06, "loss": 0.7553, "step": 15384 }, { "epoch": 2.48, "grad_norm": 13.57047176361084, "learning_rate": 1.554225878313872e-06, "loss": 0.7124, "step": 15385 }, { "epoch": 2.48, "grad_norm": 12.91247844696045, "learning_rate": 1.55329542729718e-06, "loss": 0.5911, "step": 15386 }, { "epoch": 2.48, "grad_norm": 11.279742240905762, "learning_rate": 1.5523652314227378e-06, "loss": 0.6254, "step": 15387 }, { "epoch": 2.48, "grad_norm": 13.203861236572266, "learning_rate": 1.5514352907186469e-06, "loss": 0.6465, "step": 15388 }, { "epoch": 2.48, "grad_norm": 13.384325981140137, "learning_rate": 1.5505056052129963e-06, "loss": 0.6377, "step": 15389 }, { "epoch": 2.48, "grad_norm": 12.303154945373535, "learning_rate": 1.5495761749338633e-06, "loss": 0.6201, "step": 15390 }, { "epoch": 2.48, "grad_norm": 12.126197814941406, "learning_rate": 1.548646999909329e-06, "loss": 0.832, "step": 15391 }, { "epoch": 2.48, "grad_norm": 14.849882125854492, "learning_rate": 1.5477180801674541e-06, "loss": 0.7833, "step": 15392 }, { "epoch": 2.48, "grad_norm": 12.075172424316406, "learning_rate": 1.546789415736304e-06, "loss": 0.6758, "step": 15393 }, { "epoch": 2.48, "grad_norm": 13.464334487915039, "learning_rate": 1.5458610066439273e-06, "loss": 0.6934, "step": 15394 }, { "epoch": 2.48, "grad_norm": 11.380163192749023, "learning_rate": 1.544932852918364e-06, "loss": 0.6383, "step": 15395 }, { "epoch": 2.48, "grad_norm": 15.62685489654541, "learning_rate": 1.5440049545876568e-06, "loss": 0.7025, "step": 15396 }, { "epoch": 2.48, "grad_norm": 11.472492218017578, "learning_rate": 1.5430773116798291e-06, "loss": 0.6279, "step": 15397 }, { "epoch": 2.48, "grad_norm": 13.306644439697266, "learning_rate": 1.542149924222901e-06, "loss": 0.8299, "step": 15398 }, { "epoch": 2.48, "grad_norm": 11.920374870300293, "learning_rate": 1.5412227922448885e-06, "loss": 0.6246, "step": 15399 }, { "epoch": 2.48, "grad_norm": 13.03657054901123, "learning_rate": 1.540295915773794e-06, "loss": 0.6125, "step": 15400 }, { "epoch": 2.48, "grad_norm": 11.273000717163086, "learning_rate": 1.5393692948376182e-06, "loss": 0.6009, "step": 15401 }, { "epoch": 2.48, "grad_norm": 11.26830768585205, "learning_rate": 1.5384429294643476e-06, "loss": 0.715, "step": 15402 }, { "epoch": 2.48, "grad_norm": 12.011242866516113, "learning_rate": 1.5375168196819634e-06, "loss": 0.6846, "step": 15403 }, { "epoch": 2.48, "grad_norm": 11.888870239257812, "learning_rate": 1.536590965518442e-06, "loss": 0.8938, "step": 15404 }, { "epoch": 2.48, "grad_norm": 13.307415962219238, "learning_rate": 1.5356653670017508e-06, "loss": 0.8269, "step": 15405 }, { "epoch": 2.48, "grad_norm": 11.13288688659668, "learning_rate": 1.5347400241598453e-06, "loss": 0.6732, "step": 15406 }, { "epoch": 2.48, "grad_norm": 14.93846607208252, "learning_rate": 1.53381493702068e-06, "loss": 0.7425, "step": 15407 }, { "epoch": 2.48, "grad_norm": 10.49014663696289, "learning_rate": 1.532890105612197e-06, "loss": 0.7634, "step": 15408 }, { "epoch": 2.48, "grad_norm": 10.51843547821045, "learning_rate": 1.5319655299623293e-06, "loss": 0.6498, "step": 15409 }, { "epoch": 2.48, "grad_norm": 10.722650527954102, "learning_rate": 1.531041210099008e-06, "loss": 0.7803, "step": 15410 }, { "epoch": 2.48, "grad_norm": 12.906949996948242, "learning_rate": 1.5301171460501506e-06, "loss": 0.7024, "step": 15411 }, { "epoch": 2.48, "grad_norm": 10.488061904907227, "learning_rate": 1.529193337843673e-06, "loss": 0.5638, "step": 15412 }, { "epoch": 2.48, "grad_norm": 11.503173828125, "learning_rate": 1.528269785507478e-06, "loss": 0.4967, "step": 15413 }, { "epoch": 2.48, "grad_norm": 13.181910514831543, "learning_rate": 1.52734648906946e-06, "loss": 0.8081, "step": 15414 }, { "epoch": 2.48, "grad_norm": 16.167993545532227, "learning_rate": 1.5264234485575135e-06, "loss": 0.665, "step": 15415 }, { "epoch": 2.48, "grad_norm": 13.363153457641602, "learning_rate": 1.525500663999514e-06, "loss": 0.8485, "step": 15416 }, { "epoch": 2.48, "grad_norm": 10.23305606842041, "learning_rate": 1.5245781354233403e-06, "loss": 0.7749, "step": 15417 }, { "epoch": 2.48, "grad_norm": 12.5826416015625, "learning_rate": 1.5236558628568576e-06, "loss": 0.6899, "step": 15418 }, { "epoch": 2.48, "grad_norm": 10.988680839538574, "learning_rate": 1.5227338463279195e-06, "loss": 0.5911, "step": 15419 }, { "epoch": 2.48, "grad_norm": 13.423551559448242, "learning_rate": 1.5218120858643826e-06, "loss": 0.8622, "step": 15420 }, { "epoch": 2.48, "grad_norm": 11.53038501739502, "learning_rate": 1.5208905814940878e-06, "loss": 0.6562, "step": 15421 }, { "epoch": 2.48, "grad_norm": 11.830982208251953, "learning_rate": 1.519969333244866e-06, "loss": 0.6422, "step": 15422 }, { "epoch": 2.48, "grad_norm": 13.775753021240234, "learning_rate": 1.5190483411445522e-06, "loss": 0.5694, "step": 15423 }, { "epoch": 2.48, "grad_norm": 14.786882400512695, "learning_rate": 1.518127605220958e-06, "loss": 0.6651, "step": 15424 }, { "epoch": 2.48, "grad_norm": 11.039108276367188, "learning_rate": 1.5172071255019004e-06, "loss": 0.5837, "step": 15425 }, { "epoch": 2.48, "grad_norm": 14.191548347473145, "learning_rate": 1.5162869020151838e-06, "loss": 0.8659, "step": 15426 }, { "epoch": 2.48, "grad_norm": 15.154044151306152, "learning_rate": 1.5153669347886013e-06, "loss": 0.6575, "step": 15427 }, { "epoch": 2.48, "grad_norm": 11.765307426452637, "learning_rate": 1.5144472238499452e-06, "loss": 0.7166, "step": 15428 }, { "epoch": 2.48, "grad_norm": 12.018013000488281, "learning_rate": 1.5135277692269945e-06, "loss": 0.6912, "step": 15429 }, { "epoch": 2.48, "grad_norm": 10.258482933044434, "learning_rate": 1.5126085709475202e-06, "loss": 0.6117, "step": 15430 }, { "epoch": 2.48, "grad_norm": 14.071603775024414, "learning_rate": 1.5116896290392924e-06, "loss": 0.665, "step": 15431 }, { "epoch": 2.48, "grad_norm": 10.559242248535156, "learning_rate": 1.5107709435300666e-06, "loss": 0.5634, "step": 15432 }, { "epoch": 2.48, "grad_norm": 13.557914733886719, "learning_rate": 1.5098525144475894e-06, "loss": 0.669, "step": 15433 }, { "epoch": 2.48, "grad_norm": 14.340085983276367, "learning_rate": 1.508934341819609e-06, "loss": 0.6114, "step": 15434 }, { "epoch": 2.48, "grad_norm": 16.47535514831543, "learning_rate": 1.5080164256738539e-06, "loss": 0.783, "step": 15435 }, { "epoch": 2.48, "grad_norm": 13.468257904052734, "learning_rate": 1.5070987660380565e-06, "loss": 0.8803, "step": 15436 }, { "epoch": 2.49, "grad_norm": 10.05400276184082, "learning_rate": 1.5061813629399324e-06, "loss": 0.5682, "step": 15437 }, { "epoch": 2.49, "grad_norm": 14.278826713562012, "learning_rate": 1.5052642164071918e-06, "loss": 0.6211, "step": 15438 }, { "epoch": 2.49, "grad_norm": 11.833308219909668, "learning_rate": 1.5043473264675424e-06, "loss": 0.461, "step": 15439 }, { "epoch": 2.49, "grad_norm": 12.416600227355957, "learning_rate": 1.5034306931486764e-06, "loss": 0.6447, "step": 15440 }, { "epoch": 2.49, "grad_norm": 14.393973350524902, "learning_rate": 1.5025143164782808e-06, "loss": 0.7883, "step": 15441 }, { "epoch": 2.49, "grad_norm": 12.950082778930664, "learning_rate": 1.50159819648404e-06, "loss": 0.7598, "step": 15442 }, { "epoch": 2.49, "grad_norm": 12.297454833984375, "learning_rate": 1.5006823331936216e-06, "loss": 0.7937, "step": 15443 }, { "epoch": 2.49, "grad_norm": 12.035539627075195, "learning_rate": 1.499766726634695e-06, "loss": 0.5987, "step": 15444 }, { "epoch": 2.49, "grad_norm": 10.682028770446777, "learning_rate": 1.4988513768349123e-06, "loss": 0.5905, "step": 15445 }, { "epoch": 2.49, "grad_norm": 11.936257362365723, "learning_rate": 1.4979362838219257e-06, "loss": 0.6824, "step": 15446 }, { "epoch": 2.49, "grad_norm": 11.101282119750977, "learning_rate": 1.497021447623378e-06, "loss": 0.6154, "step": 15447 }, { "epoch": 2.49, "grad_norm": 11.820021629333496, "learning_rate": 1.496106868266901e-06, "loss": 0.6616, "step": 15448 }, { "epoch": 2.49, "grad_norm": 13.936561584472656, "learning_rate": 1.4951925457801187e-06, "loss": 0.8629, "step": 15449 }, { "epoch": 2.49, "grad_norm": 11.860840797424316, "learning_rate": 1.4942784801906528e-06, "loss": 0.6974, "step": 15450 }, { "epoch": 2.49, "grad_norm": 13.529972076416016, "learning_rate": 1.493364671526113e-06, "loss": 0.7027, "step": 15451 }, { "epoch": 2.49, "grad_norm": 15.486799240112305, "learning_rate": 1.492451119814098e-06, "loss": 0.7016, "step": 15452 }, { "epoch": 2.49, "grad_norm": 13.007368087768555, "learning_rate": 1.4915378250822076e-06, "loss": 0.8037, "step": 15453 }, { "epoch": 2.49, "grad_norm": 12.974559783935547, "learning_rate": 1.490624787358025e-06, "loss": 0.7377, "step": 15454 }, { "epoch": 2.49, "grad_norm": 11.788244247436523, "learning_rate": 1.489712006669134e-06, "loss": 0.6533, "step": 15455 }, { "epoch": 2.49, "grad_norm": 13.444769859313965, "learning_rate": 1.488799483043103e-06, "loss": 0.659, "step": 15456 }, { "epoch": 2.49, "grad_norm": 15.264571189880371, "learning_rate": 1.4878872165074943e-06, "loss": 0.7591, "step": 15457 }, { "epoch": 2.49, "grad_norm": 13.523055076599121, "learning_rate": 1.486975207089869e-06, "loss": 0.6906, "step": 15458 }, { "epoch": 2.49, "grad_norm": 11.633119583129883, "learning_rate": 1.4860634548177688e-06, "loss": 0.5773, "step": 15459 }, { "epoch": 2.49, "grad_norm": 10.856258392333984, "learning_rate": 1.4851519597187414e-06, "loss": 0.5848, "step": 15460 }, { "epoch": 2.49, "grad_norm": 15.70493221282959, "learning_rate": 1.484240721820316e-06, "loss": 0.727, "step": 15461 }, { "epoch": 2.49, "grad_norm": 16.152881622314453, "learning_rate": 1.4833297411500147e-06, "loss": 0.6149, "step": 15462 }, { "epoch": 2.49, "grad_norm": 13.759637832641602, "learning_rate": 1.4824190177353604e-06, "loss": 0.7658, "step": 15463 }, { "epoch": 2.49, "grad_norm": 13.41067123413086, "learning_rate": 1.4815085516038586e-06, "loss": 0.7376, "step": 15464 }, { "epoch": 2.49, "grad_norm": 16.158597946166992, "learning_rate": 1.4805983427830085e-06, "loss": 0.6502, "step": 15465 }, { "epoch": 2.49, "grad_norm": 14.349373817443848, "learning_rate": 1.4796883913003112e-06, "loss": 0.8496, "step": 15466 }, { "epoch": 2.49, "grad_norm": 11.803304672241211, "learning_rate": 1.4787786971832508e-06, "loss": 0.5215, "step": 15467 }, { "epoch": 2.49, "grad_norm": 12.386442184448242, "learning_rate": 1.4778692604593004e-06, "loss": 0.58, "step": 15468 }, { "epoch": 2.49, "grad_norm": 12.892292976379395, "learning_rate": 1.476960081155936e-06, "loss": 0.8234, "step": 15469 }, { "epoch": 2.49, "grad_norm": 15.174295425415039, "learning_rate": 1.4760511593006176e-06, "loss": 0.7401, "step": 15470 }, { "epoch": 2.49, "grad_norm": 10.64924430847168, "learning_rate": 1.4751424949208026e-06, "loss": 0.5669, "step": 15471 }, { "epoch": 2.49, "grad_norm": 14.573526382446289, "learning_rate": 1.4742340880439364e-06, "loss": 0.7651, "step": 15472 }, { "epoch": 2.49, "grad_norm": 14.912727355957031, "learning_rate": 1.4733259386974575e-06, "loss": 0.5893, "step": 15473 }, { "epoch": 2.49, "grad_norm": 11.420672416687012, "learning_rate": 1.4724180469088002e-06, "loss": 0.7985, "step": 15474 }, { "epoch": 2.49, "grad_norm": 11.225051879882812, "learning_rate": 1.4715104127053882e-06, "loss": 0.6513, "step": 15475 }, { "epoch": 2.49, "grad_norm": 11.371099472045898, "learning_rate": 1.470603036114634e-06, "loss": 0.7313, "step": 15476 }, { "epoch": 2.49, "grad_norm": 13.707052230834961, "learning_rate": 1.4696959171639502e-06, "loss": 0.9598, "step": 15477 }, { "epoch": 2.49, "grad_norm": 10.77384090423584, "learning_rate": 1.4687890558807328e-06, "loss": 0.7042, "step": 15478 }, { "epoch": 2.49, "grad_norm": 12.18677043914795, "learning_rate": 1.4678824522923808e-06, "loss": 0.7116, "step": 15479 }, { "epoch": 2.49, "grad_norm": 9.954672813415527, "learning_rate": 1.4669761064262755e-06, "loss": 0.6509, "step": 15480 }, { "epoch": 2.49, "grad_norm": 10.773672103881836, "learning_rate": 1.466070018309792e-06, "loss": 0.6849, "step": 15481 }, { "epoch": 2.49, "grad_norm": 13.898436546325684, "learning_rate": 1.4651641879703037e-06, "loss": 0.7794, "step": 15482 }, { "epoch": 2.49, "grad_norm": 13.168085098266602, "learning_rate": 1.464258615435169e-06, "loss": 0.7156, "step": 15483 }, { "epoch": 2.49, "grad_norm": 12.883172035217285, "learning_rate": 1.4633533007317457e-06, "loss": 0.7124, "step": 15484 }, { "epoch": 2.49, "grad_norm": 14.353068351745605, "learning_rate": 1.4624482438873756e-06, "loss": 0.8549, "step": 15485 }, { "epoch": 2.49, "grad_norm": 13.554404258728027, "learning_rate": 1.461543444929401e-06, "loss": 0.7898, "step": 15486 }, { "epoch": 2.49, "grad_norm": 16.246522903442383, "learning_rate": 1.4606389038851465e-06, "loss": 0.939, "step": 15487 }, { "epoch": 2.49, "grad_norm": 13.855128288269043, "learning_rate": 1.4597346207819418e-06, "loss": 0.5727, "step": 15488 }, { "epoch": 2.49, "grad_norm": 14.128890991210938, "learning_rate": 1.4588305956470972e-06, "loss": 0.5974, "step": 15489 }, { "epoch": 2.49, "grad_norm": 12.47793197631836, "learning_rate": 1.4579268285079218e-06, "loss": 0.6554, "step": 15490 }, { "epoch": 2.49, "grad_norm": 12.342560768127441, "learning_rate": 1.457023319391715e-06, "loss": 0.6215, "step": 15491 }, { "epoch": 2.49, "grad_norm": 17.462881088256836, "learning_rate": 1.456120068325766e-06, "loss": 0.8014, "step": 15492 }, { "epoch": 2.49, "grad_norm": 11.95397663116455, "learning_rate": 1.4552170753373606e-06, "loss": 0.7006, "step": 15493 }, { "epoch": 2.49, "grad_norm": 13.208619117736816, "learning_rate": 1.454314340453773e-06, "loss": 0.6886, "step": 15494 }, { "epoch": 2.49, "grad_norm": 14.848821640014648, "learning_rate": 1.4534118637022753e-06, "loss": 0.7751, "step": 15495 }, { "epoch": 2.49, "grad_norm": 13.948941230773926, "learning_rate": 1.452509645110124e-06, "loss": 0.8408, "step": 15496 }, { "epoch": 2.49, "grad_norm": 12.80854606628418, "learning_rate": 1.4516076847045712e-06, "loss": 0.7382, "step": 15497 }, { "epoch": 2.49, "grad_norm": 13.054611206054688, "learning_rate": 1.4507059825128645e-06, "loss": 0.5492, "step": 15498 }, { "epoch": 2.5, "grad_norm": 11.760187149047852, "learning_rate": 1.4498045385622395e-06, "loss": 0.6649, "step": 15499 }, { "epoch": 2.5, "grad_norm": 11.701194763183594, "learning_rate": 1.4489033528799235e-06, "loss": 0.6434, "step": 15500 }, { "epoch": 2.5, "grad_norm": 12.087006568908691, "learning_rate": 1.4480024254931423e-06, "loss": 0.6809, "step": 15501 }, { "epoch": 2.5, "grad_norm": 11.87508773803711, "learning_rate": 1.4471017564291034e-06, "loss": 0.6226, "step": 15502 }, { "epoch": 2.5, "grad_norm": 11.696725845336914, "learning_rate": 1.4462013457150192e-06, "loss": 0.6989, "step": 15503 }, { "epoch": 2.5, "grad_norm": 11.786161422729492, "learning_rate": 1.4453011933780837e-06, "loss": 0.5721, "step": 15504 }, { "epoch": 2.5, "grad_norm": 13.488842010498047, "learning_rate": 1.4444012994454847e-06, "loss": 0.8392, "step": 15505 }, { "epoch": 2.5, "grad_norm": 12.571853637695312, "learning_rate": 1.4435016639444077e-06, "loss": 0.5342, "step": 15506 }, { "epoch": 2.5, "grad_norm": 12.15267562866211, "learning_rate": 1.4426022869020295e-06, "loss": 0.7994, "step": 15507 }, { "epoch": 2.5, "grad_norm": 13.06521224975586, "learning_rate": 1.4417031683455119e-06, "loss": 0.7561, "step": 15508 }, { "epoch": 2.5, "grad_norm": 13.467472076416016, "learning_rate": 1.4408043083020185e-06, "loss": 0.6476, "step": 15509 }, { "epoch": 2.5, "grad_norm": 13.805400848388672, "learning_rate": 1.4399057067986978e-06, "loss": 0.7039, "step": 15510 }, { "epoch": 2.5, "grad_norm": 14.22705364227295, "learning_rate": 1.4390073638626912e-06, "loss": 0.813, "step": 15511 }, { "epoch": 2.5, "grad_norm": 15.462966918945312, "learning_rate": 1.4381092795211382e-06, "loss": 0.7292, "step": 15512 }, { "epoch": 2.5, "grad_norm": 23.32217025756836, "learning_rate": 1.4372114538011629e-06, "loss": 0.7379, "step": 15513 }, { "epoch": 2.5, "grad_norm": 13.263551712036133, "learning_rate": 1.4363138867298886e-06, "loss": 0.658, "step": 15514 }, { "epoch": 2.5, "grad_norm": 12.331191062927246, "learning_rate": 1.4354165783344254e-06, "loss": 0.4566, "step": 15515 }, { "epoch": 2.5, "grad_norm": 12.601456642150879, "learning_rate": 1.4345195286418745e-06, "loss": 0.6751, "step": 15516 }, { "epoch": 2.5, "grad_norm": 12.044414520263672, "learning_rate": 1.4336227376793388e-06, "loss": 0.6305, "step": 15517 }, { "epoch": 2.5, "grad_norm": 12.983116149902344, "learning_rate": 1.4327262054739032e-06, "loss": 0.5599, "step": 15518 }, { "epoch": 2.5, "grad_norm": 14.092938423156738, "learning_rate": 1.4318299320526462e-06, "loss": 0.5634, "step": 15519 }, { "epoch": 2.5, "grad_norm": 14.8533935546875, "learning_rate": 1.430933917442645e-06, "loss": 0.573, "step": 15520 }, { "epoch": 2.5, "grad_norm": 11.638264656066895, "learning_rate": 1.4300381616709614e-06, "loss": 0.5847, "step": 15521 }, { "epoch": 2.5, "grad_norm": 12.24913215637207, "learning_rate": 1.4291426647646566e-06, "loss": 0.715, "step": 15522 }, { "epoch": 2.5, "grad_norm": 13.974331855773926, "learning_rate": 1.4282474267507774e-06, "loss": 0.8127, "step": 15523 }, { "epoch": 2.5, "grad_norm": 13.279916763305664, "learning_rate": 1.4273524476563638e-06, "loss": 0.6746, "step": 15524 }, { "epoch": 2.5, "grad_norm": 10.966400146484375, "learning_rate": 1.4264577275084534e-06, "loss": 0.7341, "step": 15525 }, { "epoch": 2.5, "grad_norm": 15.574411392211914, "learning_rate": 1.425563266334069e-06, "loss": 0.8027, "step": 15526 }, { "epoch": 2.5, "grad_norm": 13.549518585205078, "learning_rate": 1.4246690641602302e-06, "loss": 0.6695, "step": 15527 }, { "epoch": 2.5, "grad_norm": 14.290847778320312, "learning_rate": 1.4237751210139505e-06, "loss": 0.7319, "step": 15528 }, { "epoch": 2.5, "grad_norm": 12.165393829345703, "learning_rate": 1.4228814369222289e-06, "loss": 0.666, "step": 15529 }, { "epoch": 2.5, "grad_norm": 13.352300643920898, "learning_rate": 1.4219880119120587e-06, "loss": 0.7542, "step": 15530 }, { "epoch": 2.5, "grad_norm": 12.227351188659668, "learning_rate": 1.4210948460104301e-06, "loss": 0.677, "step": 15531 }, { "epoch": 2.5, "grad_norm": 12.51025676727295, "learning_rate": 1.42020193924432e-06, "loss": 0.7494, "step": 15532 }, { "epoch": 2.5, "grad_norm": 13.075247764587402, "learning_rate": 1.4193092916407025e-06, "loss": 0.7715, "step": 15533 }, { "epoch": 2.5, "grad_norm": 14.206429481506348, "learning_rate": 1.4184169032265394e-06, "loss": 0.7307, "step": 15534 }, { "epoch": 2.5, "grad_norm": 12.411334991455078, "learning_rate": 1.4175247740287835e-06, "loss": 0.5445, "step": 15535 }, { "epoch": 2.5, "grad_norm": 14.382774353027344, "learning_rate": 1.4166329040743865e-06, "loss": 0.8108, "step": 15536 }, { "epoch": 2.5, "grad_norm": 11.97026538848877, "learning_rate": 1.415741293390286e-06, "loss": 0.6498, "step": 15537 }, { "epoch": 2.5, "grad_norm": 11.565046310424805, "learning_rate": 1.4148499420034168e-06, "loss": 0.7335, "step": 15538 }, { "epoch": 2.5, "grad_norm": 13.01684284210205, "learning_rate": 1.4139588499407009e-06, "loss": 0.7551, "step": 15539 }, { "epoch": 2.5, "grad_norm": 12.581379890441895, "learning_rate": 1.4130680172290534e-06, "loss": 0.753, "step": 15540 }, { "epoch": 2.5, "grad_norm": 13.266470909118652, "learning_rate": 1.4121774438953873e-06, "loss": 0.6926, "step": 15541 }, { "epoch": 2.5, "grad_norm": 15.08002758026123, "learning_rate": 1.411287129966601e-06, "loss": 0.7616, "step": 15542 }, { "epoch": 2.5, "grad_norm": 11.378644943237305, "learning_rate": 1.410397075469585e-06, "loss": 0.6412, "step": 15543 }, { "epoch": 2.5, "grad_norm": 19.000259399414062, "learning_rate": 1.4095072804312282e-06, "loss": 0.6896, "step": 15544 }, { "epoch": 2.5, "grad_norm": 13.340860366821289, "learning_rate": 1.4086177448784044e-06, "loss": 0.7283, "step": 15545 }, { "epoch": 2.5, "grad_norm": 14.69062328338623, "learning_rate": 1.4077284688379856e-06, "loss": 0.7172, "step": 15546 }, { "epoch": 2.5, "grad_norm": 11.916916847229004, "learning_rate": 1.4068394523368345e-06, "loss": 0.4761, "step": 15547 }, { "epoch": 2.5, "grad_norm": 14.649787902832031, "learning_rate": 1.405950695401801e-06, "loss": 0.6047, "step": 15548 }, { "epoch": 2.5, "grad_norm": 15.868139266967773, "learning_rate": 1.4050621980597357e-06, "loss": 0.8285, "step": 15549 }, { "epoch": 2.5, "grad_norm": 10.899357795715332, "learning_rate": 1.4041739603374738e-06, "loss": 0.4878, "step": 15550 }, { "epoch": 2.5, "grad_norm": 11.173470497131348, "learning_rate": 1.4032859822618439e-06, "loss": 0.6951, "step": 15551 }, { "epoch": 2.5, "grad_norm": 13.833328247070312, "learning_rate": 1.4023982638596722e-06, "loss": 0.5297, "step": 15552 }, { "epoch": 2.5, "grad_norm": 15.286774635314941, "learning_rate": 1.4015108051577709e-06, "loss": 0.5701, "step": 15553 }, { "epoch": 2.5, "grad_norm": 11.65628433227539, "learning_rate": 1.4006236061829458e-06, "loss": 0.659, "step": 15554 }, { "epoch": 2.5, "grad_norm": 15.188671112060547, "learning_rate": 1.3997366669619983e-06, "loss": 0.9779, "step": 15555 }, { "epoch": 2.5, "grad_norm": 13.134767532348633, "learning_rate": 1.3988499875217166e-06, "loss": 0.7299, "step": 15556 }, { "epoch": 2.5, "grad_norm": 10.909499168395996, "learning_rate": 1.397963567888888e-06, "loss": 0.5503, "step": 15557 }, { "epoch": 2.5, "grad_norm": 13.318017959594727, "learning_rate": 1.397077408090285e-06, "loss": 0.7246, "step": 15558 }, { "epoch": 2.5, "grad_norm": 10.765690803527832, "learning_rate": 1.3961915081526722e-06, "loss": 0.6185, "step": 15559 }, { "epoch": 2.5, "grad_norm": 11.644402503967285, "learning_rate": 1.3953058681028153e-06, "loss": 0.846, "step": 15560 }, { "epoch": 2.5, "grad_norm": 11.015924453735352, "learning_rate": 1.3944204879674605e-06, "loss": 0.7771, "step": 15561 }, { "epoch": 2.51, "grad_norm": 11.391026496887207, "learning_rate": 1.3935353677733565e-06, "loss": 0.6548, "step": 15562 }, { "epoch": 2.51, "grad_norm": 11.340167999267578, "learning_rate": 1.3926505075472374e-06, "loss": 0.718, "step": 15563 }, { "epoch": 2.51, "grad_norm": 12.318453788757324, "learning_rate": 1.3917659073158286e-06, "loss": 0.8395, "step": 15564 }, { "epoch": 2.51, "grad_norm": 11.648287773132324, "learning_rate": 1.3908815671058563e-06, "loss": 0.5252, "step": 15565 }, { "epoch": 2.51, "grad_norm": 12.013345718383789, "learning_rate": 1.3899974869440259e-06, "loss": 0.6821, "step": 15566 }, { "epoch": 2.51, "grad_norm": 14.095012664794922, "learning_rate": 1.389113666857047e-06, "loss": 0.7308, "step": 15567 }, { "epoch": 2.51, "grad_norm": 12.146424293518066, "learning_rate": 1.3882301068716164e-06, "loss": 0.6194, "step": 15568 }, { "epoch": 2.51, "grad_norm": 14.931600570678711, "learning_rate": 1.3873468070144225e-06, "loss": 0.9509, "step": 15569 }, { "epoch": 2.51, "grad_norm": 10.273578643798828, "learning_rate": 1.3864637673121428e-06, "loss": 0.6964, "step": 15570 }, { "epoch": 2.51, "grad_norm": 13.012807846069336, "learning_rate": 1.385580987791456e-06, "loss": 0.6115, "step": 15571 }, { "epoch": 2.51, "grad_norm": 12.14644718170166, "learning_rate": 1.3846984684790232e-06, "loss": 0.786, "step": 15572 }, { "epoch": 2.51, "grad_norm": 14.445185661315918, "learning_rate": 1.383816209401505e-06, "loss": 0.6551, "step": 15573 }, { "epoch": 2.51, "grad_norm": 10.863605499267578, "learning_rate": 1.3829342105855504e-06, "loss": 0.8258, "step": 15574 }, { "epoch": 2.51, "grad_norm": 16.706748962402344, "learning_rate": 1.3820524720577978e-06, "loss": 0.5674, "step": 15575 }, { "epoch": 2.51, "grad_norm": 11.605435371398926, "learning_rate": 1.3811709938448848e-06, "loss": 0.6995, "step": 15576 }, { "epoch": 2.51, "grad_norm": 11.187952041625977, "learning_rate": 1.380289775973438e-06, "loss": 0.7241, "step": 15577 }, { "epoch": 2.51, "grad_norm": 15.835265159606934, "learning_rate": 1.3794088184700706e-06, "loss": 0.7243, "step": 15578 }, { "epoch": 2.51, "grad_norm": 18.072935104370117, "learning_rate": 1.3785281213613987e-06, "loss": 0.7665, "step": 15579 }, { "epoch": 2.51, "grad_norm": 11.652743339538574, "learning_rate": 1.3776476846740194e-06, "loss": 0.5802, "step": 15580 }, { "epoch": 2.51, "grad_norm": 15.575814247131348, "learning_rate": 1.3767675084345334e-06, "loss": 0.7004, "step": 15581 }, { "epoch": 2.51, "grad_norm": 14.255316734313965, "learning_rate": 1.3758875926695236e-06, "loss": 0.6817, "step": 15582 }, { "epoch": 2.51, "grad_norm": 16.398408889770508, "learning_rate": 1.3750079374055668e-06, "loss": 0.6383, "step": 15583 }, { "epoch": 2.51, "grad_norm": 12.385923385620117, "learning_rate": 1.3741285426692398e-06, "loss": 0.896, "step": 15584 }, { "epoch": 2.51, "grad_norm": 13.506317138671875, "learning_rate": 1.3732494084871017e-06, "loss": 0.6518, "step": 15585 }, { "epoch": 2.51, "grad_norm": 13.928990364074707, "learning_rate": 1.3723705348857064e-06, "loss": 0.5651, "step": 15586 }, { "epoch": 2.51, "grad_norm": 12.340824127197266, "learning_rate": 1.3714919218916033e-06, "loss": 0.68, "step": 15587 }, { "epoch": 2.51, "grad_norm": 11.2272310256958, "learning_rate": 1.370613569531335e-06, "loss": 0.5862, "step": 15588 }, { "epoch": 2.51, "grad_norm": 13.209577560424805, "learning_rate": 1.3697354778314277e-06, "loss": 0.7046, "step": 15589 }, { "epoch": 2.51, "grad_norm": 15.194750785827637, "learning_rate": 1.368857646818409e-06, "loss": 0.7529, "step": 15590 }, { "epoch": 2.51, "grad_norm": 13.9146146774292, "learning_rate": 1.3679800765187912e-06, "loss": 0.687, "step": 15591 }, { "epoch": 2.51, "grad_norm": 10.335442543029785, "learning_rate": 1.3671027669590874e-06, "loss": 0.6462, "step": 15592 }, { "epoch": 2.51, "grad_norm": 13.389842987060547, "learning_rate": 1.3662257181657945e-06, "loss": 0.5849, "step": 15593 }, { "epoch": 2.51, "grad_norm": 12.775282859802246, "learning_rate": 1.3653489301654032e-06, "loss": 0.635, "step": 15594 }, { "epoch": 2.51, "grad_norm": 11.427776336669922, "learning_rate": 1.3644724029844014e-06, "loss": 0.6685, "step": 15595 }, { "epoch": 2.51, "grad_norm": 10.887638092041016, "learning_rate": 1.3635961366492634e-06, "loss": 0.7469, "step": 15596 }, { "epoch": 2.51, "grad_norm": 11.587614059448242, "learning_rate": 1.362720131186457e-06, "loss": 0.7579, "step": 15597 }, { "epoch": 2.51, "grad_norm": 13.630132675170898, "learning_rate": 1.3618443866224463e-06, "loss": 0.8488, "step": 15598 }, { "epoch": 2.51, "grad_norm": 12.103933334350586, "learning_rate": 1.3609689029836793e-06, "loss": 0.64, "step": 15599 }, { "epoch": 2.51, "grad_norm": 10.226766586303711, "learning_rate": 1.3600936802966069e-06, "loss": 0.6678, "step": 15600 }, { "epoch": 2.51, "grad_norm": 11.357922554016113, "learning_rate": 1.3592187185876625e-06, "loss": 0.5861, "step": 15601 }, { "epoch": 2.51, "grad_norm": 15.193888664245605, "learning_rate": 1.358344017883274e-06, "loss": 0.6285, "step": 15602 }, { "epoch": 2.51, "grad_norm": 13.56960678100586, "learning_rate": 1.3574695782098667e-06, "loss": 0.8821, "step": 15603 }, { "epoch": 2.51, "grad_norm": 12.451345443725586, "learning_rate": 1.3565953995938487e-06, "loss": 0.7125, "step": 15604 }, { "epoch": 2.51, "grad_norm": 13.326070785522461, "learning_rate": 1.3557214820616327e-06, "loss": 0.815, "step": 15605 }, { "epoch": 2.51, "grad_norm": 11.524558067321777, "learning_rate": 1.3548478256396114e-06, "loss": 0.6075, "step": 15606 }, { "epoch": 2.51, "grad_norm": 12.131808280944824, "learning_rate": 1.3539744303541736e-06, "loss": 0.7394, "step": 15607 }, { "epoch": 2.51, "grad_norm": 12.177013397216797, "learning_rate": 1.3531012962317025e-06, "loss": 0.6231, "step": 15608 }, { "epoch": 2.51, "grad_norm": 11.395357131958008, "learning_rate": 1.352228423298576e-06, "loss": 0.6121, "step": 15609 }, { "epoch": 2.51, "grad_norm": 13.128849983215332, "learning_rate": 1.3513558115811554e-06, "loss": 0.6419, "step": 15610 }, { "epoch": 2.51, "grad_norm": 14.285999298095703, "learning_rate": 1.3504834611058015e-06, "loss": 0.831, "step": 15611 }, { "epoch": 2.51, "grad_norm": 12.56495475769043, "learning_rate": 1.3496113718988646e-06, "loss": 0.7372, "step": 15612 }, { "epoch": 2.51, "grad_norm": 12.560389518737793, "learning_rate": 1.3487395439866836e-06, "loss": 0.6978, "step": 15613 }, { "epoch": 2.51, "grad_norm": 13.19845199584961, "learning_rate": 1.3478679773955972e-06, "loss": 0.7245, "step": 15614 }, { "epoch": 2.51, "grad_norm": 10.462955474853516, "learning_rate": 1.3469966721519289e-06, "loss": 0.6265, "step": 15615 }, { "epoch": 2.51, "grad_norm": 15.475346565246582, "learning_rate": 1.3461256282820013e-06, "loss": 0.7416, "step": 15616 }, { "epoch": 2.51, "grad_norm": 12.599262237548828, "learning_rate": 1.3452548458121228e-06, "loss": 0.7203, "step": 15617 }, { "epoch": 2.51, "grad_norm": 12.74963092803955, "learning_rate": 1.344384324768595e-06, "loss": 0.6888, "step": 15618 }, { "epoch": 2.51, "grad_norm": 13.548279762268066, "learning_rate": 1.3435140651777168e-06, "loss": 0.6665, "step": 15619 }, { "epoch": 2.51, "grad_norm": 13.439437866210938, "learning_rate": 1.3426440670657725e-06, "loss": 0.6481, "step": 15620 }, { "epoch": 2.51, "grad_norm": 9.447029113769531, "learning_rate": 1.3417743304590403e-06, "loss": 0.5418, "step": 15621 }, { "epoch": 2.51, "grad_norm": 10.05942153930664, "learning_rate": 1.3409048553837956e-06, "loss": 0.6099, "step": 15622 }, { "epoch": 2.51, "grad_norm": 12.83955192565918, "learning_rate": 1.3400356418662973e-06, "loss": 0.6972, "step": 15623 }, { "epoch": 2.52, "grad_norm": 12.17416763305664, "learning_rate": 1.3391666899328048e-06, "loss": 0.5433, "step": 15624 }, { "epoch": 2.52, "grad_norm": 14.751606941223145, "learning_rate": 1.3382979996095658e-06, "loss": 0.7344, "step": 15625 }, { "epoch": 2.52, "grad_norm": 12.187142372131348, "learning_rate": 1.3374295709228147e-06, "loss": 0.5765, "step": 15626 }, { "epoch": 2.52, "grad_norm": 12.41790771484375, "learning_rate": 1.3365614038987885e-06, "loss": 0.5044, "step": 15627 }, { "epoch": 2.52, "grad_norm": 10.94605541229248, "learning_rate": 1.3356934985637105e-06, "loss": 0.4422, "step": 15628 }, { "epoch": 2.52, "grad_norm": 13.528230667114258, "learning_rate": 1.3348258549437953e-06, "loss": 0.8778, "step": 15629 }, { "epoch": 2.52, "grad_norm": 12.209335327148438, "learning_rate": 1.3339584730652532e-06, "loss": 0.5832, "step": 15630 }, { "epoch": 2.52, "grad_norm": 12.6979341506958, "learning_rate": 1.3330913529542834e-06, "loss": 0.6262, "step": 15631 }, { "epoch": 2.52, "grad_norm": 12.7472562789917, "learning_rate": 1.3322244946370756e-06, "loss": 0.5686, "step": 15632 }, { "epoch": 2.52, "grad_norm": 11.777902603149414, "learning_rate": 1.331357898139819e-06, "loss": 0.6063, "step": 15633 }, { "epoch": 2.52, "grad_norm": 11.917747497558594, "learning_rate": 1.3304915634886861e-06, "loss": 0.6999, "step": 15634 }, { "epoch": 2.52, "grad_norm": 12.96570873260498, "learning_rate": 1.3296254907098482e-06, "loss": 0.6563, "step": 15635 }, { "epoch": 2.52, "grad_norm": 14.720052719116211, "learning_rate": 1.3287596798294655e-06, "loss": 0.6404, "step": 15636 }, { "epoch": 2.52, "grad_norm": 13.636098861694336, "learning_rate": 1.3278941308736881e-06, "loss": 0.7843, "step": 15637 }, { "epoch": 2.52, "grad_norm": 12.338536262512207, "learning_rate": 1.3270288438686652e-06, "loss": 0.6426, "step": 15638 }, { "epoch": 2.52, "grad_norm": 13.645265579223633, "learning_rate": 1.326163818840529e-06, "loss": 0.878, "step": 15639 }, { "epoch": 2.52, "grad_norm": 11.983019828796387, "learning_rate": 1.3252990558154144e-06, "loss": 0.9014, "step": 15640 }, { "epoch": 2.52, "grad_norm": 13.836356163024902, "learning_rate": 1.3244345548194383e-06, "loss": 0.6296, "step": 15641 }, { "epoch": 2.52, "grad_norm": 13.111029624938965, "learning_rate": 1.3235703158787127e-06, "loss": 0.5692, "step": 15642 }, { "epoch": 2.52, "grad_norm": 17.920583724975586, "learning_rate": 1.3227063390193474e-06, "loss": 0.8163, "step": 15643 }, { "epoch": 2.52, "grad_norm": 11.217041969299316, "learning_rate": 1.3218426242674375e-06, "loss": 0.4416, "step": 15644 }, { "epoch": 2.52, "grad_norm": 13.477275848388672, "learning_rate": 1.3209791716490717e-06, "loss": 0.4779, "step": 15645 }, { "epoch": 2.52, "grad_norm": 11.82948112487793, "learning_rate": 1.320115981190333e-06, "loss": 0.6799, "step": 15646 }, { "epoch": 2.52, "grad_norm": 13.78136157989502, "learning_rate": 1.319253052917292e-06, "loss": 0.8756, "step": 15647 }, { "epoch": 2.52, "grad_norm": 11.9583158493042, "learning_rate": 1.3183903868560176e-06, "loss": 0.539, "step": 15648 }, { "epoch": 2.52, "grad_norm": 13.700122833251953, "learning_rate": 1.3175279830325693e-06, "loss": 0.7233, "step": 15649 }, { "epoch": 2.52, "grad_norm": 13.193828582763672, "learning_rate": 1.3166658414729916e-06, "loss": 0.7037, "step": 15650 }, { "epoch": 2.52, "grad_norm": 13.54025936126709, "learning_rate": 1.3158039622033325e-06, "loss": 0.7307, "step": 15651 }, { "epoch": 2.52, "grad_norm": 11.164494514465332, "learning_rate": 1.3149423452496224e-06, "loss": 0.466, "step": 15652 }, { "epoch": 2.52, "grad_norm": 12.527068138122559, "learning_rate": 1.3140809906378848e-06, "loss": 0.5643, "step": 15653 }, { "epoch": 2.52, "grad_norm": 10.577214241027832, "learning_rate": 1.3132198983941447e-06, "loss": 0.4392, "step": 15654 }, { "epoch": 2.52, "grad_norm": 13.209145545959473, "learning_rate": 1.3123590685444077e-06, "loss": 0.8253, "step": 15655 }, { "epoch": 2.52, "grad_norm": 14.26163387298584, "learning_rate": 1.3114985011146752e-06, "loss": 0.7134, "step": 15656 }, { "epoch": 2.52, "grad_norm": 12.817800521850586, "learning_rate": 1.3106381961309467e-06, "loss": 0.6337, "step": 15657 }, { "epoch": 2.52, "grad_norm": 12.114507675170898, "learning_rate": 1.3097781536192023e-06, "loss": 0.7417, "step": 15658 }, { "epoch": 2.52, "grad_norm": 10.95812702178955, "learning_rate": 1.3089183736054256e-06, "loss": 0.7336, "step": 15659 }, { "epoch": 2.52, "grad_norm": 14.547685623168945, "learning_rate": 1.308058856115586e-06, "loss": 0.6445, "step": 15660 }, { "epoch": 2.52, "grad_norm": 18.95323944091797, "learning_rate": 1.3071996011756438e-06, "loss": 0.8792, "step": 15661 }, { "epoch": 2.52, "grad_norm": 15.244637489318848, "learning_rate": 1.3063406088115571e-06, "loss": 0.6715, "step": 15662 }, { "epoch": 2.52, "grad_norm": 12.446930885314941, "learning_rate": 1.3054818790492717e-06, "loss": 0.6335, "step": 15663 }, { "epoch": 2.52, "grad_norm": 12.281404495239258, "learning_rate": 1.3046234119147228e-06, "loss": 0.6313, "step": 15664 }, { "epoch": 2.52, "grad_norm": 15.090219497680664, "learning_rate": 1.303765207433848e-06, "loss": 0.842, "step": 15665 }, { "epoch": 2.52, "grad_norm": 13.533982276916504, "learning_rate": 1.3029072656325636e-06, "loss": 0.7001, "step": 15666 }, { "epoch": 2.52, "grad_norm": 11.78221321105957, "learning_rate": 1.3020495865367911e-06, "loss": 0.6796, "step": 15667 }, { "epoch": 2.52, "grad_norm": 11.25285816192627, "learning_rate": 1.3011921701724317e-06, "loss": 0.5638, "step": 15668 }, { "epoch": 2.52, "grad_norm": 15.02225399017334, "learning_rate": 1.3003350165653884e-06, "loss": 0.7587, "step": 15669 }, { "epoch": 2.52, "grad_norm": 12.072039604187012, "learning_rate": 1.2994781257415534e-06, "loss": 0.5014, "step": 15670 }, { "epoch": 2.52, "grad_norm": 13.30292797088623, "learning_rate": 1.2986214977268075e-06, "loss": 0.7039, "step": 15671 }, { "epoch": 2.52, "grad_norm": 12.481364250183105, "learning_rate": 1.2977651325470253e-06, "loss": 0.5694, "step": 15672 }, { "epoch": 2.52, "grad_norm": 11.06974983215332, "learning_rate": 1.2969090302280784e-06, "loss": 0.6617, "step": 15673 }, { "epoch": 2.52, "grad_norm": 13.036054611206055, "learning_rate": 1.296053190795823e-06, "loss": 0.7916, "step": 15674 }, { "epoch": 2.52, "grad_norm": 13.304882049560547, "learning_rate": 1.2951976142761092e-06, "loss": 0.5785, "step": 15675 }, { "epoch": 2.52, "grad_norm": 18.368873596191406, "learning_rate": 1.2943423006947864e-06, "loss": 0.707, "step": 15676 }, { "epoch": 2.52, "grad_norm": 12.739824295043945, "learning_rate": 1.293487250077683e-06, "loss": 0.6452, "step": 15677 }, { "epoch": 2.52, "grad_norm": 12.815579414367676, "learning_rate": 1.2926324624506337e-06, "loss": 0.5727, "step": 15678 }, { "epoch": 2.52, "grad_norm": 14.135299682617188, "learning_rate": 1.2917779378394558e-06, "loss": 0.6906, "step": 15679 }, { "epoch": 2.52, "grad_norm": 13.33063793182373, "learning_rate": 1.2909236762699574e-06, "loss": 0.8111, "step": 15680 }, { "epoch": 2.52, "grad_norm": 9.079649925231934, "learning_rate": 1.2900696777679478e-06, "loss": 0.4801, "step": 15681 }, { "epoch": 2.52, "grad_norm": 14.891900062561035, "learning_rate": 1.2892159423592198e-06, "loss": 0.7355, "step": 15682 }, { "epoch": 2.52, "grad_norm": 11.334016799926758, "learning_rate": 1.2883624700695641e-06, "loss": 0.6348, "step": 15683 }, { "epoch": 2.52, "grad_norm": 11.437433242797852, "learning_rate": 1.2875092609247587e-06, "loss": 0.4582, "step": 15684 }, { "epoch": 2.52, "grad_norm": 13.190686225891113, "learning_rate": 1.286656314950574e-06, "loss": 0.6849, "step": 15685 }, { "epoch": 2.53, "grad_norm": 12.484782218933105, "learning_rate": 1.2858036321727796e-06, "loss": 0.6109, "step": 15686 }, { "epoch": 2.53, "grad_norm": 14.210933685302734, "learning_rate": 1.284951212617127e-06, "loss": 0.7318, "step": 15687 }, { "epoch": 2.53, "grad_norm": 12.847311973571777, "learning_rate": 1.2840990563093636e-06, "loss": 0.9311, "step": 15688 }, { "epoch": 2.53, "grad_norm": 14.499556541442871, "learning_rate": 1.2832471632752352e-06, "loss": 0.8205, "step": 15689 }, { "epoch": 2.53, "grad_norm": 11.720589637756348, "learning_rate": 1.2823955335404714e-06, "loss": 0.8468, "step": 15690 }, { "epoch": 2.53, "grad_norm": 21.45446014404297, "learning_rate": 1.281544167130795e-06, "loss": 0.8962, "step": 15691 }, { "epoch": 2.53, "grad_norm": 13.213733673095703, "learning_rate": 1.280693064071925e-06, "loss": 0.5366, "step": 15692 }, { "epoch": 2.53, "grad_norm": 13.416913986206055, "learning_rate": 1.2798422243895681e-06, "loss": 0.7219, "step": 15693 }, { "epoch": 2.53, "grad_norm": 11.45380687713623, "learning_rate": 1.2789916481094266e-06, "loss": 0.7099, "step": 15694 }, { "epoch": 2.53, "grad_norm": 12.795600891113281, "learning_rate": 1.2781413352571925e-06, "loss": 0.6851, "step": 15695 }, { "epoch": 2.53, "grad_norm": 11.216293334960938, "learning_rate": 1.2772912858585474e-06, "loss": 0.8317, "step": 15696 }, { "epoch": 2.53, "grad_norm": 11.894095420837402, "learning_rate": 1.2764414999391739e-06, "loss": 0.6888, "step": 15697 }, { "epoch": 2.53, "grad_norm": 11.748401641845703, "learning_rate": 1.2755919775247373e-06, "loss": 0.7174, "step": 15698 }, { "epoch": 2.53, "grad_norm": 11.620028495788574, "learning_rate": 1.2747427186408956e-06, "loss": 0.8011, "step": 15699 }, { "epoch": 2.53, "grad_norm": 12.387465476989746, "learning_rate": 1.2738937233133075e-06, "loss": 0.7952, "step": 15700 }, { "epoch": 2.53, "grad_norm": 13.401028633117676, "learning_rate": 1.2730449915676135e-06, "loss": 0.6759, "step": 15701 }, { "epoch": 2.53, "grad_norm": 13.25818920135498, "learning_rate": 1.2721965234294532e-06, "loss": 0.9636, "step": 15702 }, { "epoch": 2.53, "grad_norm": 11.396533012390137, "learning_rate": 1.2713483189244547e-06, "loss": 0.7114, "step": 15703 }, { "epoch": 2.53, "grad_norm": 12.900742530822754, "learning_rate": 1.2705003780782365e-06, "loss": 0.8602, "step": 15704 }, { "epoch": 2.53, "grad_norm": 14.851959228515625, "learning_rate": 1.269652700916415e-06, "loss": 0.6197, "step": 15705 }, { "epoch": 2.53, "grad_norm": 13.252408027648926, "learning_rate": 1.268805287464593e-06, "loss": 0.7724, "step": 15706 }, { "epoch": 2.53, "grad_norm": 11.23181438446045, "learning_rate": 1.2679581377483708e-06, "loss": 0.5401, "step": 15707 }, { "epoch": 2.53, "grad_norm": 11.423641204833984, "learning_rate": 1.2671112517933327e-06, "loss": 0.5273, "step": 15708 }, { "epoch": 2.53, "grad_norm": 14.410189628601074, "learning_rate": 1.2662646296250648e-06, "loss": 0.749, "step": 15709 }, { "epoch": 2.53, "grad_norm": 12.014962196350098, "learning_rate": 1.2654182712691355e-06, "loss": 0.7014, "step": 15710 }, { "epoch": 2.53, "grad_norm": 15.869093894958496, "learning_rate": 1.2645721767511143e-06, "loss": 0.6667, "step": 15711 }, { "epoch": 2.53, "grad_norm": 13.899969100952148, "learning_rate": 1.2637263460965543e-06, "loss": 0.5796, "step": 15712 }, { "epoch": 2.53, "grad_norm": 10.43159294128418, "learning_rate": 1.2628807793310106e-06, "loss": 0.7251, "step": 15713 }, { "epoch": 2.53, "grad_norm": 14.626909255981445, "learning_rate": 1.2620354764800192e-06, "loss": 0.9553, "step": 15714 }, { "epoch": 2.53, "grad_norm": 12.809988021850586, "learning_rate": 1.2611904375691143e-06, "loss": 0.7197, "step": 15715 }, { "epoch": 2.53, "grad_norm": 12.845405578613281, "learning_rate": 1.2603456626238231e-06, "loss": 0.8445, "step": 15716 }, { "epoch": 2.53, "grad_norm": 13.511175155639648, "learning_rate": 1.2595011516696609e-06, "loss": 0.5719, "step": 15717 }, { "epoch": 2.53, "grad_norm": 11.193355560302734, "learning_rate": 1.2586569047321396e-06, "loss": 0.6051, "step": 15718 }, { "epoch": 2.53, "grad_norm": 13.273648262023926, "learning_rate": 1.2578129218367597e-06, "loss": 0.7512, "step": 15719 }, { "epoch": 2.53, "grad_norm": 16.002269744873047, "learning_rate": 1.256969203009012e-06, "loss": 0.7093, "step": 15720 }, { "epoch": 2.53, "grad_norm": 11.340113639831543, "learning_rate": 1.256125748274386e-06, "loss": 0.4203, "step": 15721 }, { "epoch": 2.53, "grad_norm": 12.773256301879883, "learning_rate": 1.2552825576583572e-06, "loss": 0.6597, "step": 15722 }, { "epoch": 2.53, "grad_norm": 12.787429809570312, "learning_rate": 1.2544396311863937e-06, "loss": 0.7802, "step": 15723 }, { "epoch": 2.53, "grad_norm": 12.795746803283691, "learning_rate": 1.253596968883961e-06, "loss": 0.6577, "step": 15724 }, { "epoch": 2.53, "grad_norm": 12.572124481201172, "learning_rate": 1.2527545707765076e-06, "loss": 0.6838, "step": 15725 }, { "epoch": 2.53, "grad_norm": 12.545751571655273, "learning_rate": 1.2519124368894842e-06, "loss": 0.5094, "step": 15726 }, { "epoch": 2.53, "grad_norm": 13.234102249145508, "learning_rate": 1.251070567248326e-06, "loss": 0.741, "step": 15727 }, { "epoch": 2.53, "grad_norm": 13.46155834197998, "learning_rate": 1.2502289618784602e-06, "loss": 0.6602, "step": 15728 }, { "epoch": 2.53, "grad_norm": 14.804207801818848, "learning_rate": 1.2493876208053113e-06, "loss": 0.6838, "step": 15729 }, { "epoch": 2.53, "grad_norm": 11.48079776763916, "learning_rate": 1.2485465440542942e-06, "loss": 0.6397, "step": 15730 }, { "epoch": 2.53, "grad_norm": 13.191171646118164, "learning_rate": 1.2477057316508113e-06, "loss": 0.6676, "step": 15731 }, { "epoch": 2.53, "grad_norm": 11.489663124084473, "learning_rate": 1.246865183620264e-06, "loss": 0.8729, "step": 15732 }, { "epoch": 2.53, "grad_norm": 13.103424072265625, "learning_rate": 1.2460248999880388e-06, "loss": 0.673, "step": 15733 }, { "epoch": 2.53, "grad_norm": 9.942505836486816, "learning_rate": 1.2451848807795176e-06, "loss": 0.5856, "step": 15734 }, { "epoch": 2.53, "grad_norm": 10.95872688293457, "learning_rate": 1.2443451260200767e-06, "loss": 0.533, "step": 15735 }, { "epoch": 2.53, "grad_norm": 11.050756454467773, "learning_rate": 1.2435056357350783e-06, "loss": 0.5505, "step": 15736 }, { "epoch": 2.53, "grad_norm": 12.564580917358398, "learning_rate": 1.2426664099498841e-06, "loss": 0.6196, "step": 15737 }, { "epoch": 2.53, "grad_norm": 12.717220306396484, "learning_rate": 1.2418274486898407e-06, "loss": 0.7323, "step": 15738 }, { "epoch": 2.53, "grad_norm": 11.894512176513672, "learning_rate": 1.2409887519802899e-06, "loss": 0.4305, "step": 15739 }, { "epoch": 2.53, "grad_norm": 13.164680480957031, "learning_rate": 1.2401503198465693e-06, "loss": 0.6433, "step": 15740 }, { "epoch": 2.53, "grad_norm": 13.138057708740234, "learning_rate": 1.2393121523140006e-06, "loss": 0.69, "step": 15741 }, { "epoch": 2.53, "grad_norm": 13.060948371887207, "learning_rate": 1.2384742494079016e-06, "loss": 0.6009, "step": 15742 }, { "epoch": 2.53, "grad_norm": 15.363201141357422, "learning_rate": 1.2376366111535854e-06, "loss": 0.8307, "step": 15743 }, { "epoch": 2.53, "grad_norm": 12.206727027893066, "learning_rate": 1.2367992375763494e-06, "loss": 0.537, "step": 15744 }, { "epoch": 2.53, "grad_norm": 13.444838523864746, "learning_rate": 1.2359621287014923e-06, "loss": 0.9626, "step": 15745 }, { "epoch": 2.53, "grad_norm": 11.208117485046387, "learning_rate": 1.2351252845542972e-06, "loss": 0.5257, "step": 15746 }, { "epoch": 2.53, "grad_norm": 12.220450401306152, "learning_rate": 1.2342887051600405e-06, "loss": 0.9427, "step": 15747 }, { "epoch": 2.54, "grad_norm": 13.540336608886719, "learning_rate": 1.2334523905439955e-06, "loss": 0.7355, "step": 15748 }, { "epoch": 2.54, "grad_norm": 12.163780212402344, "learning_rate": 1.2326163407314206e-06, "loss": 0.679, "step": 15749 }, { "epoch": 2.54, "grad_norm": 11.78265380859375, "learning_rate": 1.2317805557475704e-06, "loss": 0.5948, "step": 15750 }, { "epoch": 2.54, "grad_norm": 9.801835060119629, "learning_rate": 1.2309450356176945e-06, "loss": 0.6058, "step": 15751 }, { "epoch": 2.54, "grad_norm": 13.714118003845215, "learning_rate": 1.230109780367028e-06, "loss": 0.7448, "step": 15752 }, { "epoch": 2.54, "grad_norm": 11.491365432739258, "learning_rate": 1.229274790020798e-06, "loss": 0.7589, "step": 15753 }, { "epoch": 2.54, "grad_norm": 11.942326545715332, "learning_rate": 1.2284400646042306e-06, "loss": 0.7103, "step": 15754 }, { "epoch": 2.54, "grad_norm": 11.584724426269531, "learning_rate": 1.2276056041425367e-06, "loss": 0.8932, "step": 15755 }, { "epoch": 2.54, "grad_norm": 13.92947769165039, "learning_rate": 1.226771408660925e-06, "loss": 0.6596, "step": 15756 }, { "epoch": 2.54, "grad_norm": 13.288429260253906, "learning_rate": 1.2259374781845911e-06, "loss": 0.631, "step": 15757 }, { "epoch": 2.54, "grad_norm": 11.767611503601074, "learning_rate": 1.2251038127387239e-06, "loss": 0.599, "step": 15758 }, { "epoch": 2.54, "grad_norm": 14.035139083862305, "learning_rate": 1.2242704123485082e-06, "loss": 0.6202, "step": 15759 }, { "epoch": 2.54, "grad_norm": 14.619248390197754, "learning_rate": 1.2234372770391156e-06, "loss": 0.6636, "step": 15760 }, { "epoch": 2.54, "grad_norm": 13.463908195495605, "learning_rate": 1.2226044068357135e-06, "loss": 0.7822, "step": 15761 }, { "epoch": 2.54, "grad_norm": 16.25530242919922, "learning_rate": 1.2217718017634583e-06, "loss": 0.6825, "step": 15762 }, { "epoch": 2.54, "grad_norm": 15.162927627563477, "learning_rate": 1.220939461847499e-06, "loss": 0.7025, "step": 15763 }, { "epoch": 2.54, "grad_norm": 11.39052963256836, "learning_rate": 1.2201073871129808e-06, "loss": 0.6094, "step": 15764 }, { "epoch": 2.54, "grad_norm": 11.902430534362793, "learning_rate": 1.2192755775850352e-06, "loss": 0.7423, "step": 15765 }, { "epoch": 2.54, "grad_norm": 13.408760070800781, "learning_rate": 1.218444033288786e-06, "loss": 0.7969, "step": 15766 }, { "epoch": 2.54, "grad_norm": 11.545096397399902, "learning_rate": 1.2176127542493543e-06, "loss": 0.9161, "step": 15767 }, { "epoch": 2.54, "grad_norm": 14.720930099487305, "learning_rate": 1.216781740491847e-06, "loss": 0.5462, "step": 15768 }, { "epoch": 2.54, "grad_norm": 14.496755599975586, "learning_rate": 1.215950992041367e-06, "loss": 0.5816, "step": 15769 }, { "epoch": 2.54, "grad_norm": 10.024435043334961, "learning_rate": 1.21512050892301e-06, "loss": 0.8017, "step": 15770 }, { "epoch": 2.54, "grad_norm": 16.169124603271484, "learning_rate": 1.214290291161858e-06, "loss": 0.7431, "step": 15771 }, { "epoch": 2.54, "grad_norm": 14.136407852172852, "learning_rate": 1.2134603387829935e-06, "loss": 0.5349, "step": 15772 }, { "epoch": 2.54, "grad_norm": 11.629849433898926, "learning_rate": 1.2126306518114817e-06, "loss": 0.6365, "step": 15773 }, { "epoch": 2.54, "grad_norm": 13.499943733215332, "learning_rate": 1.2118012302723846e-06, "loss": 0.6568, "step": 15774 }, { "epoch": 2.54, "grad_norm": 12.606051445007324, "learning_rate": 1.2109720741907593e-06, "loss": 0.5424, "step": 15775 }, { "epoch": 2.54, "grad_norm": 13.158441543579102, "learning_rate": 1.2101431835916488e-06, "loss": 0.5895, "step": 15776 }, { "epoch": 2.54, "grad_norm": 14.671664237976074, "learning_rate": 1.2093145585000898e-06, "loss": 0.8116, "step": 15777 }, { "epoch": 2.54, "grad_norm": 12.57923412322998, "learning_rate": 1.2084861989411144e-06, "loss": 0.6775, "step": 15778 }, { "epoch": 2.54, "grad_norm": 15.209615707397461, "learning_rate": 1.2076581049397407e-06, "loss": 0.7897, "step": 15779 }, { "epoch": 2.54, "grad_norm": 13.503595352172852, "learning_rate": 1.206830276520986e-06, "loss": 0.5659, "step": 15780 }, { "epoch": 2.54, "grad_norm": 11.493941307067871, "learning_rate": 1.206002713709854e-06, "loss": 0.688, "step": 15781 }, { "epoch": 2.54, "grad_norm": 15.29110336303711, "learning_rate": 1.2051754165313411e-06, "loss": 0.7916, "step": 15782 }, { "epoch": 2.54, "grad_norm": 10.082084655761719, "learning_rate": 1.2043483850104388e-06, "loss": 0.5149, "step": 15783 }, { "epoch": 2.54, "grad_norm": 14.13663101196289, "learning_rate": 1.2035216191721266e-06, "loss": 0.534, "step": 15784 }, { "epoch": 2.54, "grad_norm": 14.21479606628418, "learning_rate": 1.2026951190413804e-06, "loss": 0.6947, "step": 15785 }, { "epoch": 2.54, "grad_norm": 11.602694511413574, "learning_rate": 1.2018688846431636e-06, "loss": 0.8231, "step": 15786 }, { "epoch": 2.54, "grad_norm": 14.94628620147705, "learning_rate": 1.2010429160024327e-06, "loss": 0.7665, "step": 15787 }, { "epoch": 2.54, "grad_norm": 14.512489318847656, "learning_rate": 1.2002172131441403e-06, "loss": 0.7705, "step": 15788 }, { "epoch": 2.54, "grad_norm": 14.01821517944336, "learning_rate": 1.199391776093224e-06, "loss": 0.7136, "step": 15789 }, { "epoch": 2.54, "grad_norm": 13.50716495513916, "learning_rate": 1.198566604874618e-06, "loss": 0.8265, "step": 15790 }, { "epoch": 2.54, "grad_norm": 11.124046325683594, "learning_rate": 1.1977416995132507e-06, "loss": 0.8243, "step": 15791 }, { "epoch": 2.54, "grad_norm": 12.51465892791748, "learning_rate": 1.1969170600340374e-06, "loss": 0.7273, "step": 15792 }, { "epoch": 2.54, "grad_norm": 10.27945613861084, "learning_rate": 1.196092686461884e-06, "loss": 0.7459, "step": 15793 }, { "epoch": 2.54, "grad_norm": 14.695173263549805, "learning_rate": 1.1952685788216967e-06, "loss": 0.6348, "step": 15794 }, { "epoch": 2.54, "grad_norm": 12.06604290008545, "learning_rate": 1.1944447371383638e-06, "loss": 0.6169, "step": 15795 }, { "epoch": 2.54, "grad_norm": 13.112309455871582, "learning_rate": 1.1936211614367754e-06, "loss": 0.7686, "step": 15796 }, { "epoch": 2.54, "grad_norm": 12.424444198608398, "learning_rate": 1.192797851741806e-06, "loss": 0.9022, "step": 15797 }, { "epoch": 2.54, "grad_norm": 15.049335479736328, "learning_rate": 1.1919748080783233e-06, "loss": 0.6983, "step": 15798 }, { "epoch": 2.54, "grad_norm": 10.532785415649414, "learning_rate": 1.191152030471191e-06, "loss": 0.7204, "step": 15799 }, { "epoch": 2.54, "grad_norm": 14.851444244384766, "learning_rate": 1.1903295189452602e-06, "loss": 0.8025, "step": 15800 }, { "epoch": 2.54, "grad_norm": 11.463001251220703, "learning_rate": 1.1895072735253743e-06, "loss": 0.6732, "step": 15801 }, { "epoch": 2.54, "grad_norm": 13.50349235534668, "learning_rate": 1.1886852942363747e-06, "loss": 0.6395, "step": 15802 }, { "epoch": 2.54, "grad_norm": 13.536004066467285, "learning_rate": 1.1878635811030848e-06, "loss": 0.5961, "step": 15803 }, { "epoch": 2.54, "grad_norm": 12.730245590209961, "learning_rate": 1.1870421341503301e-06, "loss": 0.7409, "step": 15804 }, { "epoch": 2.54, "grad_norm": 12.819905281066895, "learning_rate": 1.1862209534029211e-06, "loss": 0.6757, "step": 15805 }, { "epoch": 2.54, "grad_norm": 13.592710494995117, "learning_rate": 1.1854000388856624e-06, "loss": 0.655, "step": 15806 }, { "epoch": 2.54, "grad_norm": 12.73769474029541, "learning_rate": 1.1845793906233515e-06, "loss": 0.7988, "step": 15807 }, { "epoch": 2.54, "grad_norm": 14.352254867553711, "learning_rate": 1.1837590086407768e-06, "loss": 0.712, "step": 15808 }, { "epoch": 2.54, "grad_norm": 17.602893829345703, "learning_rate": 1.1829388929627173e-06, "loss": 0.6775, "step": 15809 }, { "epoch": 2.55, "grad_norm": 12.753792762756348, "learning_rate": 1.1821190436139452e-06, "loss": 0.6671, "step": 15810 }, { "epoch": 2.55, "grad_norm": 13.515700340270996, "learning_rate": 1.1812994606192295e-06, "loss": 0.7979, "step": 15811 }, { "epoch": 2.55, "grad_norm": 12.13917350769043, "learning_rate": 1.180480144003322e-06, "loss": 0.7062, "step": 15812 }, { "epoch": 2.55, "grad_norm": 13.317110061645508, "learning_rate": 1.1796610937909747e-06, "loss": 0.923, "step": 15813 }, { "epoch": 2.55, "grad_norm": 11.812921524047852, "learning_rate": 1.1788423100069234e-06, "loss": 0.7429, "step": 15814 }, { "epoch": 2.55, "grad_norm": 14.199690818786621, "learning_rate": 1.1780237926759064e-06, "loss": 0.7995, "step": 15815 }, { "epoch": 2.55, "grad_norm": 12.685035705566406, "learning_rate": 1.1772055418226435e-06, "loss": 0.6146, "step": 15816 }, { "epoch": 2.55, "grad_norm": 13.743947982788086, "learning_rate": 1.1763875574718498e-06, "loss": 0.6282, "step": 15817 }, { "epoch": 2.55, "grad_norm": 10.422783851623535, "learning_rate": 1.1755698396482385e-06, "loss": 0.7206, "step": 15818 }, { "epoch": 2.55, "grad_norm": 13.934513092041016, "learning_rate": 1.1747523883765056e-06, "loss": 0.7382, "step": 15819 }, { "epoch": 2.55, "grad_norm": 10.107342720031738, "learning_rate": 1.1739352036813434e-06, "loss": 0.5854, "step": 15820 }, { "epoch": 2.55, "grad_norm": 14.562911033630371, "learning_rate": 1.1731182855874379e-06, "loss": 0.8427, "step": 15821 }, { "epoch": 2.55, "grad_norm": 13.650704383850098, "learning_rate": 1.1723016341194625e-06, "loss": 0.6979, "step": 15822 }, { "epoch": 2.55, "grad_norm": 12.988292694091797, "learning_rate": 1.1714852493020878e-06, "loss": 0.8134, "step": 15823 }, { "epoch": 2.55, "grad_norm": 12.580354690551758, "learning_rate": 1.1706691311599728e-06, "loss": 0.8486, "step": 15824 }, { "epoch": 2.55, "grad_norm": 13.173389434814453, "learning_rate": 1.1698532797177675e-06, "loss": 0.7786, "step": 15825 }, { "epoch": 2.55, "grad_norm": 12.781723022460938, "learning_rate": 1.1690376950001182e-06, "loss": 0.6817, "step": 15826 }, { "epoch": 2.55, "grad_norm": 11.47353744506836, "learning_rate": 1.168222377031657e-06, "loss": 0.6646, "step": 15827 }, { "epoch": 2.55, "grad_norm": 14.41264820098877, "learning_rate": 1.1674073258370166e-06, "loss": 0.7226, "step": 15828 }, { "epoch": 2.55, "grad_norm": 11.933300018310547, "learning_rate": 1.166592541440813e-06, "loss": 0.655, "step": 15829 }, { "epoch": 2.55, "grad_norm": 13.208888053894043, "learning_rate": 1.1657780238676564e-06, "loss": 0.6349, "step": 15830 }, { "epoch": 2.55, "grad_norm": 11.119831085205078, "learning_rate": 1.1649637731421525e-06, "loss": 0.757, "step": 15831 }, { "epoch": 2.55, "grad_norm": 11.716748237609863, "learning_rate": 1.1641497892888987e-06, "loss": 0.7273, "step": 15832 }, { "epoch": 2.55, "grad_norm": 12.311466217041016, "learning_rate": 1.1633360723324781e-06, "loss": 0.8193, "step": 15833 }, { "epoch": 2.55, "grad_norm": 10.879053115844727, "learning_rate": 1.1625226222974729e-06, "loss": 0.548, "step": 15834 }, { "epoch": 2.55, "grad_norm": 12.72623062133789, "learning_rate": 1.1617094392084538e-06, "loss": 0.6245, "step": 15835 }, { "epoch": 2.55, "grad_norm": 10.529508590698242, "learning_rate": 1.1608965230899815e-06, "loss": 0.488, "step": 15836 }, { "epoch": 2.55, "grad_norm": 12.124876976013184, "learning_rate": 1.1600838739666143e-06, "loss": 0.6802, "step": 15837 }, { "epoch": 2.55, "grad_norm": 12.114812850952148, "learning_rate": 1.1592714918628967e-06, "loss": 0.8767, "step": 15838 }, { "epoch": 2.55, "grad_norm": 16.014583587646484, "learning_rate": 1.1584593768033703e-06, "loss": 0.7273, "step": 15839 }, { "epoch": 2.55, "grad_norm": 10.988627433776855, "learning_rate": 1.1576475288125633e-06, "loss": 0.6243, "step": 15840 }, { "epoch": 2.55, "grad_norm": 13.075977325439453, "learning_rate": 1.1568359479149982e-06, "loss": 0.7583, "step": 15841 }, { "epoch": 2.55, "grad_norm": 10.072051048278809, "learning_rate": 1.1560246341351933e-06, "loss": 0.5573, "step": 15842 }, { "epoch": 2.55, "grad_norm": 11.368294715881348, "learning_rate": 1.1552135874976533e-06, "loss": 0.9321, "step": 15843 }, { "epoch": 2.55, "grad_norm": 11.611559867858887, "learning_rate": 1.1544028080268744e-06, "loss": 0.4983, "step": 15844 }, { "epoch": 2.55, "grad_norm": 10.572578430175781, "learning_rate": 1.15359229574735e-06, "loss": 0.7692, "step": 15845 }, { "epoch": 2.55, "grad_norm": 12.91631031036377, "learning_rate": 1.1527820506835608e-06, "loss": 0.7387, "step": 15846 }, { "epoch": 2.55, "grad_norm": 14.280963897705078, "learning_rate": 1.1519720728599837e-06, "loss": 0.6956, "step": 15847 }, { "epoch": 2.55, "grad_norm": 14.321401596069336, "learning_rate": 1.1511623623010837e-06, "loss": 0.7634, "step": 15848 }, { "epoch": 2.55, "grad_norm": 12.046777725219727, "learning_rate": 1.1503529190313157e-06, "loss": 0.6293, "step": 15849 }, { "epoch": 2.55, "grad_norm": 13.805158615112305, "learning_rate": 1.1495437430751343e-06, "loss": 0.8505, "step": 15850 }, { "epoch": 2.55, "grad_norm": 10.997642517089844, "learning_rate": 1.1487348344569815e-06, "loss": 0.517, "step": 15851 }, { "epoch": 2.55, "grad_norm": 13.930452346801758, "learning_rate": 1.1479261932012875e-06, "loss": 0.829, "step": 15852 }, { "epoch": 2.55, "grad_norm": 12.699389457702637, "learning_rate": 1.1471178193324828e-06, "loss": 0.5588, "step": 15853 }, { "epoch": 2.55, "grad_norm": 16.755218505859375, "learning_rate": 1.1463097128749834e-06, "loss": 0.9146, "step": 15854 }, { "epoch": 2.55, "grad_norm": 11.344087600708008, "learning_rate": 1.1455018738531964e-06, "loss": 0.5841, "step": 15855 }, { "epoch": 2.55, "grad_norm": 14.091534614562988, "learning_rate": 1.144694302291528e-06, "loss": 0.6749, "step": 15856 }, { "epoch": 2.55, "grad_norm": 11.310439109802246, "learning_rate": 1.1438869982143673e-06, "loss": 0.7197, "step": 15857 }, { "epoch": 2.55, "grad_norm": 13.051438331604004, "learning_rate": 1.1430799616461053e-06, "loss": 0.8243, "step": 15858 }, { "epoch": 2.55, "grad_norm": 14.775137901306152, "learning_rate": 1.1422731926111152e-06, "loss": 0.7804, "step": 15859 }, { "epoch": 2.55, "grad_norm": 13.496670722961426, "learning_rate": 1.1414666911337657e-06, "loss": 0.6866, "step": 15860 }, { "epoch": 2.55, "grad_norm": 15.662786483764648, "learning_rate": 1.1406604572384228e-06, "loss": 0.9694, "step": 15861 }, { "epoch": 2.55, "grad_norm": 14.13022518157959, "learning_rate": 1.1398544909494335e-06, "loss": 0.9885, "step": 15862 }, { "epoch": 2.55, "grad_norm": 12.788618087768555, "learning_rate": 1.1390487922911487e-06, "loss": 0.7697, "step": 15863 }, { "epoch": 2.55, "grad_norm": 15.664199829101562, "learning_rate": 1.138243361287903e-06, "loss": 0.6752, "step": 15864 }, { "epoch": 2.55, "grad_norm": 11.587272644042969, "learning_rate": 1.1374381979640237e-06, "loss": 0.7144, "step": 15865 }, { "epoch": 2.55, "grad_norm": 11.766079902648926, "learning_rate": 1.1366333023438348e-06, "loss": 0.7772, "step": 15866 }, { "epoch": 2.55, "grad_norm": 10.896615028381348, "learning_rate": 1.1358286744516477e-06, "loss": 0.7877, "step": 15867 }, { "epoch": 2.55, "grad_norm": 10.864091873168945, "learning_rate": 1.1350243143117656e-06, "loss": 0.7072, "step": 15868 }, { "epoch": 2.55, "grad_norm": 12.234174728393555, "learning_rate": 1.1342202219484877e-06, "loss": 0.5581, "step": 15869 }, { "epoch": 2.55, "grad_norm": 12.407252311706543, "learning_rate": 1.133416397386099e-06, "loss": 0.7569, "step": 15870 }, { "epoch": 2.55, "grad_norm": 11.21428394317627, "learning_rate": 1.1326128406488822e-06, "loss": 0.6215, "step": 15871 }, { "epoch": 2.56, "grad_norm": 11.4256010055542, "learning_rate": 1.1318095517611128e-06, "loss": 0.5853, "step": 15872 }, { "epoch": 2.56, "grad_norm": 15.075692176818848, "learning_rate": 1.1310065307470486e-06, "loss": 0.6405, "step": 15873 }, { "epoch": 2.56, "grad_norm": 13.829323768615723, "learning_rate": 1.1302037776309515e-06, "loss": 0.6574, "step": 15874 }, { "epoch": 2.56, "grad_norm": 11.037328720092773, "learning_rate": 1.129401292437068e-06, "loss": 0.6753, "step": 15875 }, { "epoch": 2.56, "grad_norm": 12.317118644714355, "learning_rate": 1.1285990751896348e-06, "loss": 0.7795, "step": 15876 }, { "epoch": 2.56, "grad_norm": 14.269745826721191, "learning_rate": 1.1277971259128873e-06, "loss": 0.5952, "step": 15877 }, { "epoch": 2.56, "grad_norm": 12.450862884521484, "learning_rate": 1.1269954446310482e-06, "loss": 0.6772, "step": 15878 }, { "epoch": 2.56, "grad_norm": 14.861771583557129, "learning_rate": 1.1261940313683318e-06, "loss": 0.5737, "step": 15879 }, { "epoch": 2.56, "grad_norm": 13.615477561950684, "learning_rate": 1.1253928861489493e-06, "loss": 0.7506, "step": 15880 }, { "epoch": 2.56, "grad_norm": 14.437846183776855, "learning_rate": 1.1245920089970952e-06, "loss": 0.724, "step": 15881 }, { "epoch": 2.56, "grad_norm": 14.644294738769531, "learning_rate": 1.1237913999369654e-06, "loss": 0.7693, "step": 15882 }, { "epoch": 2.56, "grad_norm": 14.8317289352417, "learning_rate": 1.1229910589927417e-06, "loss": 0.7087, "step": 15883 }, { "epoch": 2.56, "grad_norm": 16.162242889404297, "learning_rate": 1.1221909861885982e-06, "loss": 0.6465, "step": 15884 }, { "epoch": 2.56, "grad_norm": 12.492536544799805, "learning_rate": 1.1213911815487033e-06, "loss": 0.699, "step": 15885 }, { "epoch": 2.56, "grad_norm": 15.024027824401855, "learning_rate": 1.1205916450972165e-06, "loss": 0.6372, "step": 15886 }, { "epoch": 2.56, "grad_norm": 11.739967346191406, "learning_rate": 1.1197923768582853e-06, "loss": 0.8063, "step": 15887 }, { "epoch": 2.56, "grad_norm": 10.437329292297363, "learning_rate": 1.1189933768560567e-06, "loss": 0.4114, "step": 15888 }, { "epoch": 2.56, "grad_norm": 11.664177894592285, "learning_rate": 1.118194645114662e-06, "loss": 0.6026, "step": 15889 }, { "epoch": 2.56, "grad_norm": 12.718663215637207, "learning_rate": 1.1173961816582313e-06, "loss": 0.6293, "step": 15890 }, { "epoch": 2.56, "grad_norm": 11.620457649230957, "learning_rate": 1.116597986510879e-06, "loss": 0.8004, "step": 15891 }, { "epoch": 2.56, "grad_norm": 13.290009498596191, "learning_rate": 1.1158000596967177e-06, "loss": 0.6107, "step": 15892 }, { "epoch": 2.56, "grad_norm": 14.143988609313965, "learning_rate": 1.1150024012398518e-06, "loss": 0.9046, "step": 15893 }, { "epoch": 2.56, "grad_norm": 10.36371898651123, "learning_rate": 1.1142050111643743e-06, "loss": 0.5277, "step": 15894 }, { "epoch": 2.56, "grad_norm": 12.914802551269531, "learning_rate": 1.1134078894943667e-06, "loss": 0.6579, "step": 15895 }, { "epoch": 2.56, "grad_norm": 13.574467658996582, "learning_rate": 1.1126110362539133e-06, "loss": 0.7476, "step": 15896 }, { "epoch": 2.56, "grad_norm": 12.086150169372559, "learning_rate": 1.1118144514670814e-06, "loss": 0.835, "step": 15897 }, { "epoch": 2.56, "grad_norm": 14.751201629638672, "learning_rate": 1.1110181351579308e-06, "loss": 0.5714, "step": 15898 }, { "epoch": 2.56, "grad_norm": 15.923868179321289, "learning_rate": 1.1102220873505187e-06, "loss": 0.935, "step": 15899 }, { "epoch": 2.56, "grad_norm": 13.220343589782715, "learning_rate": 1.1094263080688871e-06, "loss": 0.7615, "step": 15900 }, { "epoch": 2.56, "grad_norm": 14.136944770812988, "learning_rate": 1.1086307973370768e-06, "loss": 0.7084, "step": 15901 }, { "epoch": 2.56, "grad_norm": 11.476109504699707, "learning_rate": 1.1078355551791154e-06, "loss": 0.6804, "step": 15902 }, { "epoch": 2.56, "grad_norm": 19.361371994018555, "learning_rate": 1.1070405816190221e-06, "loss": 0.8473, "step": 15903 }, { "epoch": 2.56, "grad_norm": 12.909680366516113, "learning_rate": 1.1062458766808138e-06, "loss": 0.583, "step": 15904 }, { "epoch": 2.56, "grad_norm": 13.025752067565918, "learning_rate": 1.105451440388492e-06, "loss": 0.8334, "step": 15905 }, { "epoch": 2.56, "grad_norm": 13.072250366210938, "learning_rate": 1.1046572727660577e-06, "loss": 0.5514, "step": 15906 }, { "epoch": 2.56, "grad_norm": 13.31177806854248, "learning_rate": 1.103863373837496e-06, "loss": 0.6273, "step": 15907 }, { "epoch": 2.56, "grad_norm": 13.359855651855469, "learning_rate": 1.1030697436267878e-06, "loss": 0.7404, "step": 15908 }, { "epoch": 2.56, "grad_norm": 13.225665092468262, "learning_rate": 1.1022763821579074e-06, "loss": 0.6529, "step": 15909 }, { "epoch": 2.56, "grad_norm": 15.230620384216309, "learning_rate": 1.1014832894548188e-06, "loss": 0.6729, "step": 15910 }, { "epoch": 2.56, "grad_norm": 12.69225788116455, "learning_rate": 1.100690465541474e-06, "loss": 0.6452, "step": 15911 }, { "epoch": 2.56, "grad_norm": 13.907411575317383, "learning_rate": 1.0998979104418284e-06, "loss": 0.7724, "step": 15912 }, { "epoch": 2.56, "grad_norm": 15.26888656616211, "learning_rate": 1.0991056241798192e-06, "loss": 0.6856, "step": 15913 }, { "epoch": 2.56, "grad_norm": 13.830888748168945, "learning_rate": 1.0983136067793753e-06, "loss": 0.7742, "step": 15914 }, { "epoch": 2.56, "grad_norm": 11.09990406036377, "learning_rate": 1.0975218582644242e-06, "loss": 0.6995, "step": 15915 }, { "epoch": 2.56, "grad_norm": 11.391911506652832, "learning_rate": 1.0967303786588778e-06, "loss": 0.7024, "step": 15916 }, { "epoch": 2.56, "grad_norm": 12.914360046386719, "learning_rate": 1.0959391679866482e-06, "loss": 0.6412, "step": 15917 }, { "epoch": 2.56, "grad_norm": 13.245083808898926, "learning_rate": 1.0951482262716328e-06, "loss": 0.5971, "step": 15918 }, { "epoch": 2.56, "grad_norm": 12.176937103271484, "learning_rate": 1.0943575535377216e-06, "loss": 0.7503, "step": 15919 }, { "epoch": 2.56, "grad_norm": 11.520828247070312, "learning_rate": 1.0935671498087997e-06, "loss": 0.8352, "step": 15920 }, { "epoch": 2.56, "grad_norm": 12.006301879882812, "learning_rate": 1.0927770151087413e-06, "loss": 0.5824, "step": 15921 }, { "epoch": 2.56, "grad_norm": 13.975054740905762, "learning_rate": 1.091987149461411e-06, "loss": 0.7851, "step": 15922 }, { "epoch": 2.56, "grad_norm": 13.26622200012207, "learning_rate": 1.0911975528906726e-06, "loss": 0.6556, "step": 15923 }, { "epoch": 2.56, "grad_norm": 15.8460693359375, "learning_rate": 1.090408225420372e-06, "loss": 0.6752, "step": 15924 }, { "epoch": 2.56, "grad_norm": 12.32133960723877, "learning_rate": 1.089619167074355e-06, "loss": 0.7359, "step": 15925 }, { "epoch": 2.56, "grad_norm": 13.65911865234375, "learning_rate": 1.0888303778764542e-06, "loss": 0.7, "step": 15926 }, { "epoch": 2.56, "grad_norm": 15.697420120239258, "learning_rate": 1.088041857850496e-06, "loss": 0.692, "step": 15927 }, { "epoch": 2.56, "grad_norm": 12.68509578704834, "learning_rate": 1.0872536070202999e-06, "loss": 0.662, "step": 15928 }, { "epoch": 2.56, "grad_norm": 11.606513977050781, "learning_rate": 1.0864656254096728e-06, "loss": 0.6154, "step": 15929 }, { "epoch": 2.56, "grad_norm": 10.990392684936523, "learning_rate": 1.0856779130424212e-06, "loss": 0.5296, "step": 15930 }, { "epoch": 2.56, "grad_norm": 13.388726234436035, "learning_rate": 1.0848904699423346e-06, "loss": 0.6706, "step": 15931 }, { "epoch": 2.56, "grad_norm": 11.886165618896484, "learning_rate": 1.084103296133202e-06, "loss": 0.8329, "step": 15932 }, { "epoch": 2.56, "grad_norm": 13.738221168518066, "learning_rate": 1.0833163916387968e-06, "loss": 0.821, "step": 15933 }, { "epoch": 2.57, "grad_norm": 32.165557861328125, "learning_rate": 1.082529756482893e-06, "loss": 0.9117, "step": 15934 }, { "epoch": 2.57, "grad_norm": 15.182150840759277, "learning_rate": 1.0817433906892483e-06, "loss": 0.7556, "step": 15935 }, { "epoch": 2.57, "grad_norm": 15.174762725830078, "learning_rate": 1.0809572942816181e-06, "loss": 0.6143, "step": 15936 }, { "epoch": 2.57, "grad_norm": 13.855459213256836, "learning_rate": 1.080171467283746e-06, "loss": 0.6734, "step": 15937 }, { "epoch": 2.57, "grad_norm": 12.173686027526855, "learning_rate": 1.0793859097193682e-06, "loss": 0.6226, "step": 15938 }, { "epoch": 2.57, "grad_norm": 14.942700386047363, "learning_rate": 1.0786006216122159e-06, "loss": 0.5957, "step": 15939 }, { "epoch": 2.57, "grad_norm": 13.486266136169434, "learning_rate": 1.0778156029860054e-06, "loss": 0.789, "step": 15940 }, { "epoch": 2.57, "grad_norm": 17.917259216308594, "learning_rate": 1.0770308538644537e-06, "loss": 0.822, "step": 15941 }, { "epoch": 2.57, "grad_norm": 11.42549991607666, "learning_rate": 1.0762463742712636e-06, "loss": 0.842, "step": 15942 }, { "epoch": 2.57, "grad_norm": 13.650060653686523, "learning_rate": 1.0754621642301277e-06, "loss": 0.7854, "step": 15943 }, { "epoch": 2.57, "grad_norm": 12.29038143157959, "learning_rate": 1.074678223764739e-06, "loss": 0.5905, "step": 15944 }, { "epoch": 2.57, "grad_norm": 11.08845043182373, "learning_rate": 1.0738945528987754e-06, "loss": 0.7774, "step": 15945 }, { "epoch": 2.57, "grad_norm": 10.722929954528809, "learning_rate": 1.0731111516559055e-06, "loss": 0.6442, "step": 15946 }, { "epoch": 2.57, "grad_norm": 11.28419017791748, "learning_rate": 1.0723280200597985e-06, "loss": 0.6232, "step": 15947 }, { "epoch": 2.57, "grad_norm": 14.297760963439941, "learning_rate": 1.0715451581341052e-06, "loss": 0.4945, "step": 15948 }, { "epoch": 2.57, "grad_norm": 13.019721031188965, "learning_rate": 1.0707625659024757e-06, "loss": 0.8313, "step": 15949 }, { "epoch": 2.57, "grad_norm": 12.943368911743164, "learning_rate": 1.0699802433885475e-06, "loss": 0.694, "step": 15950 }, { "epoch": 2.57, "grad_norm": 13.778288841247559, "learning_rate": 1.0691981906159476e-06, "loss": 0.6555, "step": 15951 }, { "epoch": 2.57, "grad_norm": 12.153862953186035, "learning_rate": 1.068416407608308e-06, "loss": 0.6359, "step": 15952 }, { "epoch": 2.57, "grad_norm": 11.738192558288574, "learning_rate": 1.0676348943892389e-06, "loss": 0.925, "step": 15953 }, { "epoch": 2.57, "grad_norm": 13.911554336547852, "learning_rate": 1.0668536509823436e-06, "loss": 0.7002, "step": 15954 }, { "epoch": 2.57, "grad_norm": 13.593722343444824, "learning_rate": 1.0660726774112251e-06, "loss": 0.8931, "step": 15955 }, { "epoch": 2.57, "grad_norm": 10.568656921386719, "learning_rate": 1.0652919736994715e-06, "loss": 0.7952, "step": 15956 }, { "epoch": 2.57, "grad_norm": 12.149627685546875, "learning_rate": 1.064511539870663e-06, "loss": 0.7614, "step": 15957 }, { "epoch": 2.57, "grad_norm": 11.035429954528809, "learning_rate": 1.063731375948377e-06, "loss": 0.5135, "step": 15958 }, { "epoch": 2.57, "grad_norm": 10.572637557983398, "learning_rate": 1.062951481956177e-06, "loss": 0.6696, "step": 15959 }, { "epoch": 2.57, "grad_norm": 12.589635848999023, "learning_rate": 1.0621718579176222e-06, "loss": 0.6751, "step": 15960 }, { "epoch": 2.57, "grad_norm": 13.436567306518555, "learning_rate": 1.061392503856261e-06, "loss": 0.746, "step": 15961 }, { "epoch": 2.57, "grad_norm": 14.863202095031738, "learning_rate": 1.060613419795633e-06, "loss": 0.6239, "step": 15962 }, { "epoch": 2.57, "grad_norm": 12.51802921295166, "learning_rate": 1.0598346057592757e-06, "loss": 0.6999, "step": 15963 }, { "epoch": 2.57, "grad_norm": 16.209228515625, "learning_rate": 1.0590560617707114e-06, "loss": 0.7304, "step": 15964 }, { "epoch": 2.57, "grad_norm": 12.017683029174805, "learning_rate": 1.0582777878534544e-06, "loss": 0.7316, "step": 15965 }, { "epoch": 2.57, "grad_norm": 14.264137268066406, "learning_rate": 1.0574997840310174e-06, "loss": 0.7446, "step": 15966 }, { "epoch": 2.57, "grad_norm": 11.543377876281738, "learning_rate": 1.0567220503268983e-06, "loss": 0.8526, "step": 15967 }, { "epoch": 2.57, "grad_norm": 12.380215644836426, "learning_rate": 1.0559445867645924e-06, "loss": 0.7305, "step": 15968 }, { "epoch": 2.57, "grad_norm": 14.368566513061523, "learning_rate": 1.055167393367581e-06, "loss": 0.6963, "step": 15969 }, { "epoch": 2.57, "grad_norm": 13.234843254089355, "learning_rate": 1.05439047015934e-06, "loss": 0.5609, "step": 15970 }, { "epoch": 2.57, "grad_norm": 11.089264869689941, "learning_rate": 1.0536138171633393e-06, "loss": 0.6024, "step": 15971 }, { "epoch": 2.57, "grad_norm": 13.773154258728027, "learning_rate": 1.0528374344030369e-06, "loss": 0.5239, "step": 15972 }, { "epoch": 2.57, "grad_norm": 12.180124282836914, "learning_rate": 1.0520613219018839e-06, "loss": 0.7013, "step": 15973 }, { "epoch": 2.57, "grad_norm": 12.571145057678223, "learning_rate": 1.0512854796833273e-06, "loss": 0.5756, "step": 15974 }, { "epoch": 2.57, "grad_norm": 10.58207893371582, "learning_rate": 1.0505099077707991e-06, "loss": 0.6454, "step": 15975 }, { "epoch": 2.57, "grad_norm": 13.512482643127441, "learning_rate": 1.0497346061877256e-06, "loss": 0.7129, "step": 15976 }, { "epoch": 2.57, "grad_norm": 10.731863021850586, "learning_rate": 1.0489595749575298e-06, "loss": 0.7112, "step": 15977 }, { "epoch": 2.57, "grad_norm": 12.979812622070312, "learning_rate": 1.0481848141036166e-06, "loss": 0.7023, "step": 15978 }, { "epoch": 2.57, "grad_norm": 12.7056884765625, "learning_rate": 1.0474103236493937e-06, "loss": 0.6384, "step": 15979 }, { "epoch": 2.57, "grad_norm": 15.774789810180664, "learning_rate": 1.0466361036182526e-06, "loss": 0.6565, "step": 15980 }, { "epoch": 2.57, "grad_norm": 14.30910873413086, "learning_rate": 1.0458621540335789e-06, "loss": 0.7921, "step": 15981 }, { "epoch": 2.57, "grad_norm": 13.654842376708984, "learning_rate": 1.045088474918754e-06, "loss": 0.8659, "step": 15982 }, { "epoch": 2.57, "grad_norm": 11.238126754760742, "learning_rate": 1.0443150662971435e-06, "loss": 0.6188, "step": 15983 }, { "epoch": 2.57, "grad_norm": 12.543938636779785, "learning_rate": 1.0435419281921133e-06, "loss": 0.6443, "step": 15984 }, { "epoch": 2.57, "grad_norm": 12.369755744934082, "learning_rate": 1.0427690606270157e-06, "loss": 0.7428, "step": 15985 }, { "epoch": 2.57, "grad_norm": 9.645598411560059, "learning_rate": 1.0419964636251922e-06, "loss": 0.6884, "step": 15986 }, { "epoch": 2.57, "grad_norm": 11.621573448181152, "learning_rate": 1.0412241372099853e-06, "loss": 0.7296, "step": 15987 }, { "epoch": 2.57, "grad_norm": 13.453417778015137, "learning_rate": 1.0404520814047214e-06, "loss": 0.6465, "step": 15988 }, { "epoch": 2.57, "grad_norm": 13.936653137207031, "learning_rate": 1.039680296232719e-06, "loss": 0.6909, "step": 15989 }, { "epoch": 2.57, "grad_norm": 11.396711349487305, "learning_rate": 1.038908781717296e-06, "loss": 0.6373, "step": 15990 }, { "epoch": 2.57, "grad_norm": 11.868265151977539, "learning_rate": 1.0381375378817526e-06, "loss": 0.6298, "step": 15991 }, { "epoch": 2.57, "grad_norm": 12.545140266418457, "learning_rate": 1.0373665647493858e-06, "loss": 0.8213, "step": 15992 }, { "epoch": 2.57, "grad_norm": 11.859458923339844, "learning_rate": 1.0365958623434869e-06, "loss": 0.6663, "step": 15993 }, { "epoch": 2.57, "grad_norm": 14.515107154846191, "learning_rate": 1.0358254306873317e-06, "loss": 0.8039, "step": 15994 }, { "epoch": 2.57, "grad_norm": 11.37221908569336, "learning_rate": 1.035055269804196e-06, "loss": 0.5808, "step": 15995 }, { "epoch": 2.58, "grad_norm": 12.188213348388672, "learning_rate": 1.0342853797173401e-06, "loss": 0.8078, "step": 15996 }, { "epoch": 2.58, "grad_norm": 12.953604698181152, "learning_rate": 1.03351576045002e-06, "loss": 0.6237, "step": 15997 }, { "epoch": 2.58, "grad_norm": 11.85701847076416, "learning_rate": 1.0327464120254848e-06, "loss": 0.5333, "step": 15998 }, { "epoch": 2.58, "grad_norm": 12.49289321899414, "learning_rate": 1.0319773344669716e-06, "loss": 0.7415, "step": 15999 }, { "epoch": 2.58, "grad_norm": 12.196063041687012, "learning_rate": 1.03120852779771e-06, "loss": 0.6638, "step": 16000 }, { "epoch": 2.58, "grad_norm": 15.23846435546875, "learning_rate": 1.0304399920409269e-06, "loss": 0.6668, "step": 16001 }, { "epoch": 2.58, "grad_norm": 12.251228332519531, "learning_rate": 1.029671727219832e-06, "loss": 0.7084, "step": 16002 }, { "epoch": 2.58, "grad_norm": 11.925174713134766, "learning_rate": 1.0289037333576356e-06, "loss": 0.6402, "step": 16003 }, { "epoch": 2.58, "grad_norm": 18.448671340942383, "learning_rate": 1.028136010477534e-06, "loss": 0.6429, "step": 16004 }, { "epoch": 2.58, "grad_norm": 13.685737609863281, "learning_rate": 1.027368558602716e-06, "loss": 0.8037, "step": 16005 }, { "epoch": 2.58, "grad_norm": 13.692362785339355, "learning_rate": 1.0266013777563666e-06, "loss": 0.6976, "step": 16006 }, { "epoch": 2.58, "grad_norm": 13.379262924194336, "learning_rate": 1.025834467961655e-06, "loss": 0.6834, "step": 16007 }, { "epoch": 2.58, "grad_norm": 11.424851417541504, "learning_rate": 1.0250678292417515e-06, "loss": 0.6615, "step": 16008 }, { "epoch": 2.58, "grad_norm": 14.320791244506836, "learning_rate": 1.02430146161981e-06, "loss": 0.9264, "step": 16009 }, { "epoch": 2.58, "grad_norm": 14.005616188049316, "learning_rate": 1.0235353651189794e-06, "loss": 0.7387, "step": 16010 }, { "epoch": 2.58, "grad_norm": 11.331933975219727, "learning_rate": 1.0227695397624026e-06, "loss": 0.6736, "step": 16011 }, { "epoch": 2.58, "grad_norm": 16.53226089477539, "learning_rate": 1.0220039855732089e-06, "loss": 0.734, "step": 16012 }, { "epoch": 2.58, "grad_norm": 14.271464347839355, "learning_rate": 1.0212387025745251e-06, "loss": 0.7088, "step": 16013 }, { "epoch": 2.58, "grad_norm": 13.6920747756958, "learning_rate": 1.0204736907894697e-06, "loss": 0.7084, "step": 16014 }, { "epoch": 2.58, "grad_norm": 11.126187324523926, "learning_rate": 1.0197089502411473e-06, "loss": 0.6761, "step": 16015 }, { "epoch": 2.58, "grad_norm": 11.246447563171387, "learning_rate": 1.0189444809526572e-06, "loss": 0.6458, "step": 16016 }, { "epoch": 2.58, "grad_norm": 11.75427532196045, "learning_rate": 1.0181802829470943e-06, "loss": 0.6953, "step": 16017 }, { "epoch": 2.58, "grad_norm": 12.799896240234375, "learning_rate": 1.017416356247538e-06, "loss": 0.6631, "step": 16018 }, { "epoch": 2.58, "grad_norm": 11.095145225524902, "learning_rate": 1.0166527008770688e-06, "loss": 0.8792, "step": 16019 }, { "epoch": 2.58, "grad_norm": 10.48936939239502, "learning_rate": 1.0158893168587503e-06, "loss": 0.7365, "step": 16020 }, { "epoch": 2.58, "grad_norm": 12.400814056396484, "learning_rate": 1.0151262042156396e-06, "loss": 0.753, "step": 16021 }, { "epoch": 2.58, "grad_norm": 11.458908081054688, "learning_rate": 1.0143633629707928e-06, "loss": 0.6088, "step": 16022 }, { "epoch": 2.58, "grad_norm": 12.550536155700684, "learning_rate": 1.013600793147248e-06, "loss": 0.5505, "step": 16023 }, { "epoch": 2.58, "grad_norm": 12.6378755569458, "learning_rate": 1.0128384947680392e-06, "loss": 0.6051, "step": 16024 }, { "epoch": 2.58, "grad_norm": 13.734397888183594, "learning_rate": 1.0120764678561967e-06, "loss": 0.7095, "step": 16025 }, { "epoch": 2.58, "grad_norm": 10.730868339538574, "learning_rate": 1.0113147124347322e-06, "loss": 0.5231, "step": 16026 }, { "epoch": 2.58, "grad_norm": 13.606572151184082, "learning_rate": 1.010553228526663e-06, "loss": 0.6425, "step": 16027 }, { "epoch": 2.58, "grad_norm": 11.802181243896484, "learning_rate": 1.0097920161549845e-06, "loss": 0.5239, "step": 16028 }, { "epoch": 2.58, "grad_norm": 12.590686798095703, "learning_rate": 1.0090310753426913e-06, "loss": 0.9303, "step": 16029 }, { "epoch": 2.58, "grad_norm": 12.545289993286133, "learning_rate": 1.0082704061127701e-06, "loss": 0.6549, "step": 16030 }, { "epoch": 2.58, "grad_norm": 12.14645004272461, "learning_rate": 1.0075100084881972e-06, "loss": 0.6487, "step": 16031 }, { "epoch": 2.58, "grad_norm": 11.945558547973633, "learning_rate": 1.0067498824919397e-06, "loss": 0.6633, "step": 16032 }, { "epoch": 2.58, "grad_norm": 13.531434059143066, "learning_rate": 1.0059900281469581e-06, "loss": 0.7279, "step": 16033 }, { "epoch": 2.58, "grad_norm": 10.99870491027832, "learning_rate": 1.0052304454762084e-06, "loss": 0.6244, "step": 16034 }, { "epoch": 2.58, "grad_norm": 13.40913200378418, "learning_rate": 1.0044711345026304e-06, "loss": 0.6651, "step": 16035 }, { "epoch": 2.58, "grad_norm": 13.617815017700195, "learning_rate": 1.003712095249163e-06, "loss": 0.7775, "step": 16036 }, { "epoch": 2.58, "grad_norm": 12.567173957824707, "learning_rate": 1.0029533277387304e-06, "loss": 0.7891, "step": 16037 }, { "epoch": 2.58, "grad_norm": 13.944074630737305, "learning_rate": 1.0021948319942565e-06, "loss": 0.6466, "step": 16038 }, { "epoch": 2.58, "grad_norm": 13.778718948364258, "learning_rate": 1.0014366080386496e-06, "loss": 0.5753, "step": 16039 }, { "epoch": 2.58, "grad_norm": 11.4246244430542, "learning_rate": 1.000678655894811e-06, "loss": 0.6635, "step": 16040 }, { "epoch": 2.58, "grad_norm": 17.563390731811523, "learning_rate": 9.99920975585641e-07, "loss": 0.6045, "step": 16041 }, { "epoch": 2.58, "grad_norm": 13.034673690795898, "learning_rate": 9.991635671340216e-07, "loss": 0.7211, "step": 16042 }, { "epoch": 2.58, "grad_norm": 12.205900192260742, "learning_rate": 9.984064305628317e-07, "loss": 0.4776, "step": 16043 }, { "epoch": 2.58, "grad_norm": 16.676509857177734, "learning_rate": 9.976495658949436e-07, "loss": 0.6215, "step": 16044 }, { "epoch": 2.58, "grad_norm": 12.195619583129883, "learning_rate": 9.96892973153215e-07, "loss": 0.7336, "step": 16045 }, { "epoch": 2.58, "grad_norm": 14.293347358703613, "learning_rate": 9.961366523605053e-07, "loss": 0.7717, "step": 16046 }, { "epoch": 2.58, "grad_norm": 14.727777481079102, "learning_rate": 9.953806035396575e-07, "loss": 0.7546, "step": 16047 }, { "epoch": 2.58, "grad_norm": 14.099296569824219, "learning_rate": 9.946248267135073e-07, "loss": 0.9393, "step": 16048 }, { "epoch": 2.58, "grad_norm": 15.47183609008789, "learning_rate": 9.938693219048856e-07, "loss": 0.7128, "step": 16049 }, { "epoch": 2.58, "grad_norm": 13.857392311096191, "learning_rate": 9.93114089136612e-07, "loss": 0.6133, "step": 16050 }, { "epoch": 2.58, "grad_norm": 13.993717193603516, "learning_rate": 9.923591284315014e-07, "loss": 0.5028, "step": 16051 }, { "epoch": 2.58, "grad_norm": 12.230819702148438, "learning_rate": 9.91604439812357e-07, "loss": 0.744, "step": 16052 }, { "epoch": 2.58, "grad_norm": 13.446261405944824, "learning_rate": 9.908500233019725e-07, "loss": 0.7481, "step": 16053 }, { "epoch": 2.58, "grad_norm": 12.875041007995605, "learning_rate": 9.900958789231386e-07, "loss": 0.7309, "step": 16054 }, { "epoch": 2.58, "grad_norm": 13.802298545837402, "learning_rate": 9.89342006698637e-07, "loss": 0.7015, "step": 16055 }, { "epoch": 2.58, "grad_norm": 14.659494400024414, "learning_rate": 9.885884066512341e-07, "loss": 0.6977, "step": 16056 }, { "epoch": 2.58, "grad_norm": 12.536287307739258, "learning_rate": 9.878350788036995e-07, "loss": 0.5992, "step": 16057 }, { "epoch": 2.58, "grad_norm": 12.122113227844238, "learning_rate": 9.870820231787836e-07, "loss": 0.6838, "step": 16058 }, { "epoch": 2.59, "grad_norm": 15.190640449523926, "learning_rate": 9.86329239799234e-07, "loss": 0.7466, "step": 16059 }, { "epoch": 2.59, "grad_norm": 12.527674674987793, "learning_rate": 9.855767286877903e-07, "loss": 0.6527, "step": 16060 }, { "epoch": 2.59, "grad_norm": 13.193482398986816, "learning_rate": 9.848244898671823e-07, "loss": 0.6864, "step": 16061 }, { "epoch": 2.59, "grad_norm": 15.095897674560547, "learning_rate": 9.840725233601345e-07, "loss": 0.6467, "step": 16062 }, { "epoch": 2.59, "grad_norm": 14.499490737915039, "learning_rate": 9.83320829189358e-07, "loss": 0.6292, "step": 16063 }, { "epoch": 2.59, "grad_norm": 13.476340293884277, "learning_rate": 9.825694073775582e-07, "loss": 0.8597, "step": 16064 }, { "epoch": 2.59, "grad_norm": 12.524626731872559, "learning_rate": 9.818182579474355e-07, "loss": 0.6272, "step": 16065 }, { "epoch": 2.59, "grad_norm": 16.70398712158203, "learning_rate": 9.810673809216786e-07, "loss": 0.8309, "step": 16066 }, { "epoch": 2.59, "grad_norm": 15.59360408782959, "learning_rate": 9.803167763229648e-07, "loss": 0.7484, "step": 16067 }, { "epoch": 2.59, "grad_norm": 14.512162208557129, "learning_rate": 9.795664441739727e-07, "loss": 0.5994, "step": 16068 }, { "epoch": 2.59, "grad_norm": 11.928245544433594, "learning_rate": 9.788163844973608e-07, "loss": 0.7049, "step": 16069 }, { "epoch": 2.59, "grad_norm": 12.3699951171875, "learning_rate": 9.780665973157922e-07, "loss": 0.5915, "step": 16070 }, { "epoch": 2.59, "grad_norm": 12.43802261352539, "learning_rate": 9.773170826519107e-07, "loss": 0.9344, "step": 16071 }, { "epoch": 2.59, "grad_norm": 12.31638240814209, "learning_rate": 9.765678405283562e-07, "loss": 1.0259, "step": 16072 }, { "epoch": 2.59, "grad_norm": 13.610198974609375, "learning_rate": 9.758188709677606e-07, "loss": 0.9736, "step": 16073 }, { "epoch": 2.59, "grad_norm": 12.10821533203125, "learning_rate": 9.750701739927505e-07, "loss": 0.6195, "step": 16074 }, { "epoch": 2.59, "grad_norm": 12.171731948852539, "learning_rate": 9.743217496259372e-07, "loss": 0.6921, "step": 16075 }, { "epoch": 2.59, "grad_norm": 11.664434432983398, "learning_rate": 9.735735978899309e-07, "loss": 0.6594, "step": 16076 }, { "epoch": 2.59, "grad_norm": 14.963361740112305, "learning_rate": 9.728257188073286e-07, "loss": 0.7373, "step": 16077 }, { "epoch": 2.59, "grad_norm": 12.518905639648438, "learning_rate": 9.720781124007194e-07, "loss": 0.9768, "step": 16078 }, { "epoch": 2.59, "grad_norm": 15.644893646240234, "learning_rate": 9.713307786926895e-07, "loss": 0.7113, "step": 16079 }, { "epoch": 2.59, "grad_norm": 12.64805793762207, "learning_rate": 9.705837177058076e-07, "loss": 0.6922, "step": 16080 }, { "epoch": 2.59, "grad_norm": 12.544464111328125, "learning_rate": 9.698369294626452e-07, "loss": 0.8026, "step": 16081 }, { "epoch": 2.59, "grad_norm": 11.047045707702637, "learning_rate": 9.690904139857581e-07, "loss": 0.7089, "step": 16082 }, { "epoch": 2.59, "grad_norm": 15.626395225524902, "learning_rate": 9.683441712976916e-07, "loss": 0.5094, "step": 16083 }, { "epoch": 2.59, "grad_norm": 14.810670852661133, "learning_rate": 9.675982014209927e-07, "loss": 0.8443, "step": 16084 }, { "epoch": 2.59, "grad_norm": 10.644698143005371, "learning_rate": 9.668525043781895e-07, "loss": 0.7048, "step": 16085 }, { "epoch": 2.59, "grad_norm": 11.024961471557617, "learning_rate": 9.661070801918104e-07, "loss": 0.7642, "step": 16086 }, { "epoch": 2.59, "grad_norm": 11.027898788452148, "learning_rate": 9.653619288843707e-07, "loss": 0.6772, "step": 16087 }, { "epoch": 2.59, "grad_norm": 13.149645805358887, "learning_rate": 9.64617050478376e-07, "loss": 0.7331, "step": 16088 }, { "epoch": 2.59, "grad_norm": 15.325765609741211, "learning_rate": 9.638724449963288e-07, "loss": 0.9428, "step": 16089 }, { "epoch": 2.59, "grad_norm": 13.109357833862305, "learning_rate": 9.631281124607216e-07, "loss": 0.821, "step": 16090 }, { "epoch": 2.59, "grad_norm": 11.565656661987305, "learning_rate": 9.623840528940332e-07, "loss": 0.8094, "step": 16091 }, { "epoch": 2.59, "grad_norm": 11.453165054321289, "learning_rate": 9.61640266318744e-07, "loss": 0.5229, "step": 16092 }, { "epoch": 2.59, "grad_norm": 13.203222274780273, "learning_rate": 9.608967527573176e-07, "loss": 0.8205, "step": 16093 }, { "epoch": 2.59, "grad_norm": 13.268071174621582, "learning_rate": 9.601535122322137e-07, "loss": 0.7223, "step": 16094 }, { "epoch": 2.59, "grad_norm": 12.432148933410645, "learning_rate": 9.594105447658842e-07, "loss": 0.6464, "step": 16095 }, { "epoch": 2.59, "grad_norm": 12.652562141418457, "learning_rate": 9.586678503807679e-07, "loss": 0.8563, "step": 16096 }, { "epoch": 2.59, "grad_norm": 11.889005661010742, "learning_rate": 9.579254290993034e-07, "loss": 0.7151, "step": 16097 }, { "epoch": 2.59, "grad_norm": 13.57694149017334, "learning_rate": 9.57183280943913e-07, "loss": 0.73, "step": 16098 }, { "epoch": 2.59, "grad_norm": 11.089669227600098, "learning_rate": 9.56441405937013e-07, "loss": 0.5882, "step": 16099 }, { "epoch": 2.59, "grad_norm": 14.091691017150879, "learning_rate": 9.556998041010169e-07, "loss": 0.804, "step": 16100 }, { "epoch": 2.59, "grad_norm": 11.236974716186523, "learning_rate": 9.54958475458323e-07, "loss": 0.5894, "step": 16101 }, { "epoch": 2.59, "grad_norm": 13.527978897094727, "learning_rate": 9.542174200313214e-07, "loss": 0.6328, "step": 16102 }, { "epoch": 2.59, "grad_norm": 11.620747566223145, "learning_rate": 9.534766378424021e-07, "loss": 0.6652, "step": 16103 }, { "epoch": 2.59, "grad_norm": 13.878585815429688, "learning_rate": 9.527361289139358e-07, "loss": 0.7934, "step": 16104 }, { "epoch": 2.59, "grad_norm": 13.279525756835938, "learning_rate": 9.519958932682949e-07, "loss": 0.7781, "step": 16105 }, { "epoch": 2.59, "grad_norm": 11.575139999389648, "learning_rate": 9.512559309278368e-07, "loss": 0.8634, "step": 16106 }, { "epoch": 2.59, "grad_norm": 10.415773391723633, "learning_rate": 9.505162419149128e-07, "loss": 0.7494, "step": 16107 }, { "epoch": 2.59, "grad_norm": 12.990849494934082, "learning_rate": 9.497768262518669e-07, "loss": 0.6227, "step": 16108 }, { "epoch": 2.59, "grad_norm": 12.149023056030273, "learning_rate": 9.490376839610338e-07, "loss": 0.7246, "step": 16109 }, { "epoch": 2.59, "grad_norm": 14.640853881835938, "learning_rate": 9.482988150647388e-07, "loss": 0.6945, "step": 16110 }, { "epoch": 2.59, "grad_norm": 12.363799095153809, "learning_rate": 9.475602195853029e-07, "loss": 0.6899, "step": 16111 }, { "epoch": 2.59, "grad_norm": 12.150640487670898, "learning_rate": 9.468218975450328e-07, "loss": 0.6157, "step": 16112 }, { "epoch": 2.59, "grad_norm": 12.233428955078125, "learning_rate": 9.460838489662339e-07, "loss": 0.6348, "step": 16113 }, { "epoch": 2.59, "grad_norm": 13.09646224975586, "learning_rate": 9.453460738711962e-07, "loss": 0.6858, "step": 16114 }, { "epoch": 2.59, "grad_norm": 13.62552547454834, "learning_rate": 9.446085722822063e-07, "loss": 0.6056, "step": 16115 }, { "epoch": 2.59, "grad_norm": 11.43606948852539, "learning_rate": 9.438713442215453e-07, "loss": 0.7994, "step": 16116 }, { "epoch": 2.59, "grad_norm": 12.178174018859863, "learning_rate": 9.431343897114775e-07, "loss": 0.6387, "step": 16117 }, { "epoch": 2.59, "grad_norm": 18.243165969848633, "learning_rate": 9.423977087742631e-07, "loss": 0.8567, "step": 16118 }, { "epoch": 2.59, "grad_norm": 10.546464920043945, "learning_rate": 9.416613014321574e-07, "loss": 0.6163, "step": 16119 }, { "epoch": 2.59, "grad_norm": 11.306106567382812, "learning_rate": 9.409251677074027e-07, "loss": 0.6873, "step": 16120 }, { "epoch": 2.6, "grad_norm": 13.058008193969727, "learning_rate": 9.401893076222335e-07, "loss": 0.7559, "step": 16121 }, { "epoch": 2.6, "grad_norm": 13.398466110229492, "learning_rate": 9.394537211988808e-07, "loss": 0.8854, "step": 16122 }, { "epoch": 2.6, "grad_norm": 13.946686744689941, "learning_rate": 9.387184084595591e-07, "loss": 0.748, "step": 16123 }, { "epoch": 2.6, "grad_norm": 13.913105010986328, "learning_rate": 9.379833694264851e-07, "loss": 0.7106, "step": 16124 }, { "epoch": 2.6, "grad_norm": 11.730652809143066, "learning_rate": 9.372486041218576e-07, "loss": 0.769, "step": 16125 }, { "epoch": 2.6, "grad_norm": 16.905872344970703, "learning_rate": 9.36514112567869e-07, "loss": 0.7648, "step": 16126 }, { "epoch": 2.6, "grad_norm": 13.55738353729248, "learning_rate": 9.357798947867114e-07, "loss": 0.7539, "step": 16127 }, { "epoch": 2.6, "grad_norm": 11.45435619354248, "learning_rate": 9.350459508005572e-07, "loss": 0.6853, "step": 16128 }, { "epoch": 2.6, "grad_norm": 18.712177276611328, "learning_rate": 9.343122806315796e-07, "loss": 0.8409, "step": 16129 }, { "epoch": 2.6, "grad_norm": 13.191758155822754, "learning_rate": 9.335788843019389e-07, "loss": 1.0507, "step": 16130 }, { "epoch": 2.6, "grad_norm": 11.27585506439209, "learning_rate": 9.328457618337861e-07, "loss": 0.6335, "step": 16131 }, { "epoch": 2.6, "grad_norm": 11.71250057220459, "learning_rate": 9.321129132492701e-07, "loss": 0.8836, "step": 16132 }, { "epoch": 2.6, "grad_norm": 13.082473754882812, "learning_rate": 9.313803385705245e-07, "loss": 0.5632, "step": 16133 }, { "epoch": 2.6, "grad_norm": 9.898172378540039, "learning_rate": 9.306480378196747e-07, "loss": 0.5979, "step": 16134 }, { "epoch": 2.6, "grad_norm": 12.724959373474121, "learning_rate": 9.299160110188477e-07, "loss": 0.5714, "step": 16135 }, { "epoch": 2.6, "grad_norm": 14.277619361877441, "learning_rate": 9.291842581901522e-07, "loss": 0.6921, "step": 16136 }, { "epoch": 2.6, "grad_norm": 13.585400581359863, "learning_rate": 9.284527793556885e-07, "loss": 0.6124, "step": 16137 }, { "epoch": 2.6, "grad_norm": 12.958040237426758, "learning_rate": 9.277215745375567e-07, "loss": 0.619, "step": 16138 }, { "epoch": 2.6, "grad_norm": 12.21574592590332, "learning_rate": 9.269906437578391e-07, "loss": 0.7905, "step": 16139 }, { "epoch": 2.6, "grad_norm": 14.141522407531738, "learning_rate": 9.26259987038618e-07, "loss": 0.6607, "step": 16140 }, { "epoch": 2.6, "grad_norm": 11.885636329650879, "learning_rate": 9.255296044019623e-07, "loss": 0.4714, "step": 16141 }, { "epoch": 2.6, "grad_norm": 10.09424877166748, "learning_rate": 9.247994958699325e-07, "loss": 0.4897, "step": 16142 }, { "epoch": 2.6, "grad_norm": 11.113844871520996, "learning_rate": 9.24069661464585e-07, "loss": 0.5636, "step": 16143 }, { "epoch": 2.6, "grad_norm": 10.75014877319336, "learning_rate": 9.233401012079634e-07, "loss": 0.6302, "step": 16144 }, { "epoch": 2.6, "grad_norm": 12.96597957611084, "learning_rate": 9.226108151221047e-07, "loss": 0.6132, "step": 16145 }, { "epoch": 2.6, "grad_norm": 11.764209747314453, "learning_rate": 9.2188180322904e-07, "loss": 0.4813, "step": 16146 }, { "epoch": 2.6, "grad_norm": 11.510229110717773, "learning_rate": 9.211530655507872e-07, "loss": 0.747, "step": 16147 }, { "epoch": 2.6, "grad_norm": 12.953246116638184, "learning_rate": 9.204246021093621e-07, "loss": 0.6762, "step": 16148 }, { "epoch": 2.6, "grad_norm": 10.641793251037598, "learning_rate": 9.19696412926766e-07, "loss": 0.6984, "step": 16149 }, { "epoch": 2.6, "grad_norm": 11.165714263916016, "learning_rate": 9.189684980249936e-07, "loss": 0.7197, "step": 16150 }, { "epoch": 2.6, "grad_norm": 10.977747917175293, "learning_rate": 9.18240857426036e-07, "loss": 0.6793, "step": 16151 }, { "epoch": 2.6, "grad_norm": 13.405228614807129, "learning_rate": 9.17513491151868e-07, "loss": 0.7018, "step": 16152 }, { "epoch": 2.6, "grad_norm": 17.091562271118164, "learning_rate": 9.167863992244652e-07, "loss": 0.7179, "step": 16153 }, { "epoch": 2.6, "grad_norm": 11.890341758728027, "learning_rate": 9.160595816657869e-07, "loss": 0.6086, "step": 16154 }, { "epoch": 2.6, "grad_norm": 14.619281768798828, "learning_rate": 9.153330384977899e-07, "loss": 0.8502, "step": 16155 }, { "epoch": 2.6, "grad_norm": 11.332098960876465, "learning_rate": 9.146067697424177e-07, "loss": 0.5592, "step": 16156 }, { "epoch": 2.6, "grad_norm": 11.143513679504395, "learning_rate": 9.138807754216106e-07, "loss": 0.6251, "step": 16157 }, { "epoch": 2.6, "grad_norm": 11.981992721557617, "learning_rate": 9.131550555572955e-07, "loss": 0.7508, "step": 16158 }, { "epoch": 2.6, "grad_norm": 14.803958892822266, "learning_rate": 9.124296101713958e-07, "loss": 0.8983, "step": 16159 }, { "epoch": 2.6, "grad_norm": 15.442015647888184, "learning_rate": 9.117044392858243e-07, "loss": 0.566, "step": 16160 }, { "epoch": 2.6, "grad_norm": 9.938088417053223, "learning_rate": 9.109795429224832e-07, "loss": 0.7519, "step": 16161 }, { "epoch": 2.6, "grad_norm": 14.003653526306152, "learning_rate": 9.102549211032719e-07, "loss": 0.9462, "step": 16162 }, { "epoch": 2.6, "grad_norm": 14.22864055633545, "learning_rate": 9.095305738500748e-07, "loss": 0.6111, "step": 16163 }, { "epoch": 2.6, "grad_norm": 13.613162994384766, "learning_rate": 9.088065011847757e-07, "loss": 0.7564, "step": 16164 }, { "epoch": 2.6, "grad_norm": 12.921209335327148, "learning_rate": 9.080827031292439e-07, "loss": 0.6316, "step": 16165 }, { "epoch": 2.6, "grad_norm": 12.233020782470703, "learning_rate": 9.073591797053416e-07, "loss": 0.7313, "step": 16166 }, { "epoch": 2.6, "grad_norm": 12.121339797973633, "learning_rate": 9.06635930934926e-07, "loss": 0.6224, "step": 16167 }, { "epoch": 2.6, "grad_norm": 15.75848388671875, "learning_rate": 9.059129568398428e-07, "loss": 0.6185, "step": 16168 }, { "epoch": 2.6, "grad_norm": 16.099533081054688, "learning_rate": 9.051902574419281e-07, "loss": 0.581, "step": 16169 }, { "epoch": 2.6, "grad_norm": 11.449186325073242, "learning_rate": 9.044678327630152e-07, "loss": 0.5765, "step": 16170 }, { "epoch": 2.6, "grad_norm": 12.136520385742188, "learning_rate": 9.037456828249225e-07, "loss": 0.6479, "step": 16171 }, { "epoch": 2.6, "grad_norm": 15.01107406616211, "learning_rate": 9.030238076494669e-07, "loss": 0.5724, "step": 16172 }, { "epoch": 2.6, "grad_norm": 14.678437232971191, "learning_rate": 9.02302207258452e-07, "loss": 0.7907, "step": 16173 }, { "epoch": 2.6, "grad_norm": 13.602142333984375, "learning_rate": 9.015808816736693e-07, "loss": 0.6038, "step": 16174 }, { "epoch": 2.6, "grad_norm": 10.217289924621582, "learning_rate": 9.00859830916917e-07, "loss": 0.7399, "step": 16175 }, { "epoch": 2.6, "grad_norm": 14.209371566772461, "learning_rate": 9.001390550099698e-07, "loss": 0.859, "step": 16176 }, { "epoch": 2.6, "grad_norm": 13.196614265441895, "learning_rate": 8.994185539745981e-07, "loss": 0.6066, "step": 16177 }, { "epoch": 2.6, "grad_norm": 12.430866241455078, "learning_rate": 8.9869832783257e-07, "loss": 0.6744, "step": 16178 }, { "epoch": 2.6, "grad_norm": 13.53692626953125, "learning_rate": 8.979783766056372e-07, "loss": 0.8281, "step": 16179 }, { "epoch": 2.6, "grad_norm": 14.310097694396973, "learning_rate": 8.972587003155464e-07, "loss": 0.6802, "step": 16180 }, { "epoch": 2.6, "grad_norm": 11.718964576721191, "learning_rate": 8.965392989840393e-07, "loss": 0.544, "step": 16181 }, { "epoch": 2.6, "grad_norm": 9.48316478729248, "learning_rate": 8.958201726328431e-07, "loss": 0.5441, "step": 16182 }, { "epoch": 2.61, "grad_norm": 10.824013710021973, "learning_rate": 8.951013212836823e-07, "loss": 0.6083, "step": 16183 }, { "epoch": 2.61, "grad_norm": 21.46253204345703, "learning_rate": 8.943827449582687e-07, "loss": 0.9963, "step": 16184 }, { "epoch": 2.61, "grad_norm": 12.479293823242188, "learning_rate": 8.936644436783082e-07, "loss": 0.6817, "step": 16185 }, { "epoch": 2.61, "grad_norm": 13.052973747253418, "learning_rate": 8.929464174654989e-07, "loss": 0.7811, "step": 16186 }, { "epoch": 2.61, "grad_norm": 10.88259506225586, "learning_rate": 8.922286663415292e-07, "loss": 0.5728, "step": 16187 }, { "epoch": 2.61, "grad_norm": 17.173912048339844, "learning_rate": 8.915111903280771e-07, "loss": 0.7411, "step": 16188 }, { "epoch": 2.61, "grad_norm": 13.692073822021484, "learning_rate": 8.907939894468187e-07, "loss": 0.8428, "step": 16189 }, { "epoch": 2.61, "grad_norm": 13.6526460647583, "learning_rate": 8.900770637194145e-07, "loss": 0.6642, "step": 16190 }, { "epoch": 2.61, "grad_norm": 12.53333568572998, "learning_rate": 8.893604131675237e-07, "loss": 0.915, "step": 16191 }, { "epoch": 2.61, "grad_norm": 21.143667221069336, "learning_rate": 8.886440378127915e-07, "loss": 0.6567, "step": 16192 }, { "epoch": 2.61, "grad_norm": 14.410518646240234, "learning_rate": 8.879279376768546e-07, "loss": 0.6039, "step": 16193 }, { "epoch": 2.61, "grad_norm": 13.597268104553223, "learning_rate": 8.872121127813471e-07, "loss": 0.8242, "step": 16194 }, { "epoch": 2.61, "grad_norm": 12.777588844299316, "learning_rate": 8.864965631478894e-07, "loss": 0.8843, "step": 16195 }, { "epoch": 2.61, "grad_norm": 10.78562068939209, "learning_rate": 8.857812887980955e-07, "loss": 1.0502, "step": 16196 }, { "epoch": 2.61, "grad_norm": 13.161672592163086, "learning_rate": 8.850662897535734e-07, "loss": 0.4523, "step": 16197 }, { "epoch": 2.61, "grad_norm": 12.268275260925293, "learning_rate": 8.843515660359181e-07, "loss": 0.7112, "step": 16198 }, { "epoch": 2.61, "grad_norm": 12.151436805725098, "learning_rate": 8.83637117666718e-07, "loss": 0.677, "step": 16199 }, { "epoch": 2.61, "grad_norm": 13.925649642944336, "learning_rate": 8.829229446675569e-07, "loss": 0.6417, "step": 16200 }, { "epoch": 2.61, "grad_norm": 12.213343620300293, "learning_rate": 8.822090470600031e-07, "loss": 0.711, "step": 16201 }, { "epoch": 2.61, "grad_norm": 11.05718994140625, "learning_rate": 8.81495424865626e-07, "loss": 0.706, "step": 16202 }, { "epoch": 2.61, "grad_norm": 12.735037803649902, "learning_rate": 8.807820781059773e-07, "loss": 0.5993, "step": 16203 }, { "epoch": 2.61, "grad_norm": 10.96480941772461, "learning_rate": 8.800690068026041e-07, "loss": 0.7548, "step": 16204 }, { "epoch": 2.61, "grad_norm": 14.079178810119629, "learning_rate": 8.793562109770481e-07, "loss": 0.6447, "step": 16205 }, { "epoch": 2.61, "grad_norm": 11.988511085510254, "learning_rate": 8.786436906508378e-07, "loss": 0.8479, "step": 16206 }, { "epoch": 2.61, "grad_norm": 13.617137908935547, "learning_rate": 8.779314458454979e-07, "loss": 0.6934, "step": 16207 }, { "epoch": 2.61, "grad_norm": 11.554999351501465, "learning_rate": 8.772194765825415e-07, "loss": 0.7126, "step": 16208 }, { "epoch": 2.61, "grad_norm": 15.02997875213623, "learning_rate": 8.765077828834734e-07, "loss": 0.7463, "step": 16209 }, { "epoch": 2.61, "grad_norm": 12.347091674804688, "learning_rate": 8.757963647697932e-07, "loss": 0.7881, "step": 16210 }, { "epoch": 2.61, "grad_norm": 13.532048225402832, "learning_rate": 8.750852222629891e-07, "loss": 0.7908, "step": 16211 }, { "epoch": 2.61, "grad_norm": 11.683059692382812, "learning_rate": 8.743743553845396e-07, "loss": 0.6545, "step": 16212 }, { "epoch": 2.61, "grad_norm": 13.06139087677002, "learning_rate": 8.736637641559209e-07, "loss": 0.7066, "step": 16213 }, { "epoch": 2.61, "grad_norm": 13.261357307434082, "learning_rate": 8.729534485985946e-07, "loss": 0.6705, "step": 16214 }, { "epoch": 2.61, "grad_norm": 12.381916999816895, "learning_rate": 8.722434087340181e-07, "loss": 0.5326, "step": 16215 }, { "epoch": 2.61, "grad_norm": 11.906087875366211, "learning_rate": 8.715336445836398e-07, "loss": 0.6189, "step": 16216 }, { "epoch": 2.61, "grad_norm": 13.218316078186035, "learning_rate": 8.708241561688968e-07, "loss": 0.6406, "step": 16217 }, { "epoch": 2.61, "grad_norm": 11.585087776184082, "learning_rate": 8.701149435112221e-07, "loss": 0.6466, "step": 16218 }, { "epoch": 2.61, "grad_norm": 14.262598991394043, "learning_rate": 8.694060066320375e-07, "loss": 0.6508, "step": 16219 }, { "epoch": 2.61, "grad_norm": 13.9454927444458, "learning_rate": 8.686973455527548e-07, "loss": 0.9977, "step": 16220 }, { "epoch": 2.61, "grad_norm": 14.109379768371582, "learning_rate": 8.679889602947833e-07, "loss": 0.8297, "step": 16221 }, { "epoch": 2.61, "grad_norm": 15.297662734985352, "learning_rate": 8.672808508795194e-07, "loss": 0.6669, "step": 16222 }, { "epoch": 2.61, "grad_norm": 11.751460075378418, "learning_rate": 8.665730173283505e-07, "loss": 0.722, "step": 16223 }, { "epoch": 2.61, "grad_norm": 12.17702865600586, "learning_rate": 8.658654596626614e-07, "loss": 0.7949, "step": 16224 }, { "epoch": 2.61, "grad_norm": 13.935408592224121, "learning_rate": 8.651581779038198e-07, "loss": 0.7379, "step": 16225 }, { "epoch": 2.61, "grad_norm": 12.580584526062012, "learning_rate": 8.644511720731951e-07, "loss": 0.7751, "step": 16226 }, { "epoch": 2.61, "grad_norm": 14.204924583435059, "learning_rate": 8.637444421921404e-07, "loss": 0.6014, "step": 16227 }, { "epoch": 2.61, "grad_norm": 9.530113220214844, "learning_rate": 8.630379882820017e-07, "loss": 0.6203, "step": 16228 }, { "epoch": 2.61, "grad_norm": 12.044952392578125, "learning_rate": 8.623318103641221e-07, "loss": 0.8173, "step": 16229 }, { "epoch": 2.61, "grad_norm": 10.06562328338623, "learning_rate": 8.616259084598289e-07, "loss": 0.6256, "step": 16230 }, { "epoch": 2.61, "grad_norm": 14.232207298278809, "learning_rate": 8.609202825904483e-07, "loss": 0.8028, "step": 16231 }, { "epoch": 2.61, "grad_norm": 12.746594429016113, "learning_rate": 8.602149327772924e-07, "loss": 0.7276, "step": 16232 }, { "epoch": 2.61, "grad_norm": 13.538406372070312, "learning_rate": 8.595098590416651e-07, "loss": 0.7858, "step": 16233 }, { "epoch": 2.61, "grad_norm": 13.497604370117188, "learning_rate": 8.588050614048682e-07, "loss": 0.6163, "step": 16234 }, { "epoch": 2.61, "grad_norm": 12.566205024719238, "learning_rate": 8.581005398881859e-07, "loss": 0.5354, "step": 16235 }, { "epoch": 2.61, "grad_norm": 38.51247024536133, "learning_rate": 8.573962945129033e-07, "loss": 0.735, "step": 16236 }, { "epoch": 2.61, "grad_norm": 14.106757164001465, "learning_rate": 8.566923253002923e-07, "loss": 0.7544, "step": 16237 }, { "epoch": 2.61, "grad_norm": 13.832324981689453, "learning_rate": 8.559886322716171e-07, "loss": 0.6277, "step": 16238 }, { "epoch": 2.61, "grad_norm": 14.862414360046387, "learning_rate": 8.552852154481305e-07, "loss": 0.6919, "step": 16239 }, { "epoch": 2.61, "grad_norm": 14.185715675354004, "learning_rate": 8.545820748510846e-07, "loss": 0.6233, "step": 16240 }, { "epoch": 2.61, "grad_norm": 14.199687004089355, "learning_rate": 8.538792105017146e-07, "loss": 0.742, "step": 16241 }, { "epoch": 2.61, "grad_norm": 11.767413139343262, "learning_rate": 8.531766224212545e-07, "loss": 0.7339, "step": 16242 }, { "epoch": 2.61, "grad_norm": 12.635167121887207, "learning_rate": 8.524743106309241e-07, "loss": 0.9056, "step": 16243 }, { "epoch": 2.61, "grad_norm": 10.422102928161621, "learning_rate": 8.517722751519386e-07, "loss": 0.7048, "step": 16244 }, { "epoch": 2.62, "grad_norm": 13.448286056518555, "learning_rate": 8.510705160055044e-07, "loss": 0.6573, "step": 16245 }, { "epoch": 2.62, "grad_norm": 13.247983932495117, "learning_rate": 8.50369033212819e-07, "loss": 0.6351, "step": 16246 }, { "epoch": 2.62, "grad_norm": 14.2216796875, "learning_rate": 8.496678267950686e-07, "loss": 0.8019, "step": 16247 }, { "epoch": 2.62, "grad_norm": 13.544185638427734, "learning_rate": 8.489668967734377e-07, "loss": 0.7899, "step": 16248 }, { "epoch": 2.62, "grad_norm": 12.781389236450195, "learning_rate": 8.482662431690958e-07, "loss": 0.8838, "step": 16249 }, { "epoch": 2.62, "grad_norm": 16.40169334411621, "learning_rate": 8.475658660032105e-07, "loss": 0.7837, "step": 16250 }, { "epoch": 2.62, "grad_norm": 13.4356689453125, "learning_rate": 8.468657652969348e-07, "loss": 0.6823, "step": 16251 }, { "epoch": 2.62, "grad_norm": 11.255145072937012, "learning_rate": 8.461659410714151e-07, "loss": 0.707, "step": 16252 }, { "epoch": 2.62, "grad_norm": 13.337703704833984, "learning_rate": 8.454663933477924e-07, "loss": 0.7251, "step": 16253 }, { "epoch": 2.62, "grad_norm": 11.24280071258545, "learning_rate": 8.447671221471964e-07, "loss": 0.8695, "step": 16254 }, { "epoch": 2.62, "grad_norm": 12.569541931152344, "learning_rate": 8.440681274907492e-07, "loss": 0.7878, "step": 16255 }, { "epoch": 2.62, "grad_norm": 12.88735580444336, "learning_rate": 8.433694093995659e-07, "loss": 0.8375, "step": 16256 }, { "epoch": 2.62, "grad_norm": 11.977472305297852, "learning_rate": 8.426709678947531e-07, "loss": 0.7872, "step": 16257 }, { "epoch": 2.62, "grad_norm": 13.530481338500977, "learning_rate": 8.419728029974029e-07, "loss": 0.9442, "step": 16258 }, { "epoch": 2.62, "grad_norm": 9.532068252563477, "learning_rate": 8.412749147286104e-07, "loss": 0.6532, "step": 16259 }, { "epoch": 2.62, "grad_norm": 12.221599578857422, "learning_rate": 8.40577303109451e-07, "loss": 0.4991, "step": 16260 }, { "epoch": 2.62, "grad_norm": 15.574475288391113, "learning_rate": 8.398799681610004e-07, "loss": 0.6752, "step": 16261 }, { "epoch": 2.62, "grad_norm": 14.866681098937988, "learning_rate": 8.391829099043225e-07, "loss": 0.6053, "step": 16262 }, { "epoch": 2.62, "grad_norm": 11.91663646697998, "learning_rate": 8.384861283604684e-07, "loss": 0.6058, "step": 16263 }, { "epoch": 2.62, "grad_norm": 9.802345275878906, "learning_rate": 8.377896235504901e-07, "loss": 0.5915, "step": 16264 }, { "epoch": 2.62, "grad_norm": 10.980978012084961, "learning_rate": 8.370933954954241e-07, "loss": 0.8141, "step": 16265 }, { "epoch": 2.62, "grad_norm": 13.9406156539917, "learning_rate": 8.363974442162992e-07, "loss": 0.5773, "step": 16266 }, { "epoch": 2.62, "grad_norm": 11.826465606689453, "learning_rate": 8.357017697341418e-07, "loss": 0.8641, "step": 16267 }, { "epoch": 2.62, "grad_norm": 12.261756896972656, "learning_rate": 8.350063720699609e-07, "loss": 0.7276, "step": 16268 }, { "epoch": 2.62, "grad_norm": 11.240229606628418, "learning_rate": 8.343112512447648e-07, "loss": 0.6584, "step": 16269 }, { "epoch": 2.62, "grad_norm": 16.11573028564453, "learning_rate": 8.336164072795494e-07, "loss": 0.8805, "step": 16270 }, { "epoch": 2.62, "grad_norm": 14.3691987991333, "learning_rate": 8.329218401953021e-07, "loss": 0.6937, "step": 16271 }, { "epoch": 2.62, "grad_norm": 12.62606143951416, "learning_rate": 8.322275500130051e-07, "loss": 0.6522, "step": 16272 }, { "epoch": 2.62, "grad_norm": 12.026778221130371, "learning_rate": 8.315335367536282e-07, "loss": 0.7514, "step": 16273 }, { "epoch": 2.62, "grad_norm": 11.920098304748535, "learning_rate": 8.30839800438138e-07, "loss": 0.6301, "step": 16274 }, { "epoch": 2.62, "grad_norm": 12.37228012084961, "learning_rate": 8.301463410874866e-07, "loss": 0.7354, "step": 16275 }, { "epoch": 2.62, "grad_norm": 9.640738487243652, "learning_rate": 8.294531587226206e-07, "loss": 0.6097, "step": 16276 }, { "epoch": 2.62, "grad_norm": 11.726897239685059, "learning_rate": 8.28760253364479e-07, "loss": 1.0677, "step": 16277 }, { "epoch": 2.62, "grad_norm": 12.153362274169922, "learning_rate": 8.280676250339948e-07, "loss": 0.6008, "step": 16278 }, { "epoch": 2.62, "grad_norm": 12.848245620727539, "learning_rate": 8.273752737520846e-07, "loss": 0.6986, "step": 16279 }, { "epoch": 2.62, "grad_norm": 11.339568138122559, "learning_rate": 8.266831995396651e-07, "loss": 0.765, "step": 16280 }, { "epoch": 2.62, "grad_norm": 14.798484802246094, "learning_rate": 8.259914024176418e-07, "loss": 0.6336, "step": 16281 }, { "epoch": 2.62, "grad_norm": 15.675130844116211, "learning_rate": 8.25299882406907e-07, "loss": 0.5516, "step": 16282 }, { "epoch": 2.62, "grad_norm": 11.502623558044434, "learning_rate": 8.246086395283536e-07, "loss": 0.6682, "step": 16283 }, { "epoch": 2.62, "grad_norm": 11.976016998291016, "learning_rate": 8.239176738028565e-07, "loss": 0.8231, "step": 16284 }, { "epoch": 2.62, "grad_norm": 15.05435562133789, "learning_rate": 8.23226985251293e-07, "loss": 0.6879, "step": 16285 }, { "epoch": 2.62, "grad_norm": 11.543936729431152, "learning_rate": 8.225365738945213e-07, "loss": 0.7747, "step": 16286 }, { "epoch": 2.62, "grad_norm": 12.390830993652344, "learning_rate": 8.218464397533976e-07, "loss": 0.6657, "step": 16287 }, { "epoch": 2.62, "grad_norm": 14.505204200744629, "learning_rate": 8.211565828487689e-07, "loss": 0.6321, "step": 16288 }, { "epoch": 2.62, "grad_norm": 12.991232872009277, "learning_rate": 8.204670032014728e-07, "loss": 0.8468, "step": 16289 }, { "epoch": 2.62, "grad_norm": 12.57693862915039, "learning_rate": 8.197777008323371e-07, "loss": 0.878, "step": 16290 }, { "epoch": 2.62, "grad_norm": 11.51851749420166, "learning_rate": 8.190886757621853e-07, "loss": 0.5814, "step": 16291 }, { "epoch": 2.62, "grad_norm": 14.151938438415527, "learning_rate": 8.183999280118282e-07, "loss": 0.5688, "step": 16292 }, { "epoch": 2.62, "grad_norm": 13.291248321533203, "learning_rate": 8.177114576020728e-07, "loss": 0.6273, "step": 16293 }, { "epoch": 2.62, "grad_norm": 12.86715030670166, "learning_rate": 8.170232645537135e-07, "loss": 0.8265, "step": 16294 }, { "epoch": 2.62, "grad_norm": 11.334006309509277, "learning_rate": 8.163353488875369e-07, "loss": 0.7164, "step": 16295 }, { "epoch": 2.62, "grad_norm": 12.306537628173828, "learning_rate": 8.156477106243222e-07, "loss": 0.8994, "step": 16296 }, { "epoch": 2.62, "grad_norm": 14.691401481628418, "learning_rate": 8.149603497848446e-07, "loss": 0.5087, "step": 16297 }, { "epoch": 2.62, "grad_norm": 13.346244812011719, "learning_rate": 8.142732663898612e-07, "loss": 0.8062, "step": 16298 }, { "epoch": 2.62, "grad_norm": 22.49064064025879, "learning_rate": 8.135864604601295e-07, "loss": 0.9306, "step": 16299 }, { "epoch": 2.62, "grad_norm": 13.255011558532715, "learning_rate": 8.128999320163944e-07, "loss": 0.5227, "step": 16300 }, { "epoch": 2.62, "grad_norm": 12.339977264404297, "learning_rate": 8.122136810793924e-07, "loss": 0.6605, "step": 16301 }, { "epoch": 2.62, "grad_norm": 14.054682731628418, "learning_rate": 8.115277076698536e-07, "loss": 0.8597, "step": 16302 }, { "epoch": 2.62, "grad_norm": 11.199727058410645, "learning_rate": 8.108420118084969e-07, "loss": 0.8142, "step": 16303 }, { "epoch": 2.62, "grad_norm": 15.857794761657715, "learning_rate": 8.101565935160371e-07, "loss": 0.6513, "step": 16304 }, { "epoch": 2.62, "grad_norm": 10.641287803649902, "learning_rate": 8.094714528131775e-07, "loss": 0.4758, "step": 16305 }, { "epoch": 2.62, "grad_norm": 11.656545639038086, "learning_rate": 8.087865897206105e-07, "loss": 0.6489, "step": 16306 }, { "epoch": 2.63, "grad_norm": 12.842636108398438, "learning_rate": 8.08102004259027e-07, "loss": 0.7513, "step": 16307 }, { "epoch": 2.63, "grad_norm": 16.98959732055664, "learning_rate": 8.074176964491031e-07, "loss": 0.7259, "step": 16308 }, { "epoch": 2.63, "grad_norm": 13.797452926635742, "learning_rate": 8.067336663115111e-07, "loss": 0.6854, "step": 16309 }, { "epoch": 2.63, "grad_norm": 13.205221176147461, "learning_rate": 8.06049913866912e-07, "loss": 0.7133, "step": 16310 }, { "epoch": 2.63, "grad_norm": 12.107046127319336, "learning_rate": 8.053664391359572e-07, "loss": 0.9202, "step": 16311 }, { "epoch": 2.63, "grad_norm": 12.103900909423828, "learning_rate": 8.046832421392958e-07, "loss": 0.7618, "step": 16312 }, { "epoch": 2.63, "grad_norm": 11.835478782653809, "learning_rate": 8.040003228975623e-07, "loss": 0.739, "step": 16313 }, { "epoch": 2.63, "grad_norm": 14.422621726989746, "learning_rate": 8.033176814313837e-07, "loss": 0.7078, "step": 16314 }, { "epoch": 2.63, "grad_norm": 14.182406425476074, "learning_rate": 8.026353177613833e-07, "loss": 0.7553, "step": 16315 }, { "epoch": 2.63, "grad_norm": 12.508708953857422, "learning_rate": 8.019532319081691e-07, "loss": 0.6336, "step": 16316 }, { "epoch": 2.63, "grad_norm": 12.957587242126465, "learning_rate": 8.012714238923469e-07, "loss": 0.8285, "step": 16317 }, { "epoch": 2.63, "grad_norm": 12.920079231262207, "learning_rate": 8.005898937345114e-07, "loss": 0.6986, "step": 16318 }, { "epoch": 2.63, "grad_norm": 14.818649291992188, "learning_rate": 7.99908641455247e-07, "loss": 0.6331, "step": 16319 }, { "epoch": 2.63, "grad_norm": 14.536676406860352, "learning_rate": 7.992276670751343e-07, "loss": 0.6172, "step": 16320 }, { "epoch": 2.63, "grad_norm": 12.175451278686523, "learning_rate": 7.98546970614742e-07, "loss": 0.6587, "step": 16321 }, { "epoch": 2.63, "grad_norm": 13.244772911071777, "learning_rate": 7.978665520946282e-07, "loss": 0.6626, "step": 16322 }, { "epoch": 2.63, "grad_norm": 15.101068496704102, "learning_rate": 7.971864115353512e-07, "loss": 0.8998, "step": 16323 }, { "epoch": 2.63, "grad_norm": 12.596935272216797, "learning_rate": 7.965065489574519e-07, "loss": 0.7339, "step": 16324 }, { "epoch": 2.63, "grad_norm": 13.069745063781738, "learning_rate": 7.958269643814664e-07, "loss": 0.6894, "step": 16325 }, { "epoch": 2.63, "grad_norm": 11.815857887268066, "learning_rate": 7.951476578279238e-07, "loss": 0.8704, "step": 16326 }, { "epoch": 2.63, "grad_norm": 13.046610832214355, "learning_rate": 7.944686293173398e-07, "loss": 0.6629, "step": 16327 }, { "epoch": 2.63, "grad_norm": 11.61388874053955, "learning_rate": 7.937898788702314e-07, "loss": 0.7331, "step": 16328 }, { "epoch": 2.63, "grad_norm": 14.682429313659668, "learning_rate": 7.931114065070967e-07, "loss": 0.6676, "step": 16329 }, { "epoch": 2.63, "grad_norm": 12.284936904907227, "learning_rate": 7.92433212248428e-07, "loss": 0.6433, "step": 16330 }, { "epoch": 2.63, "grad_norm": 11.085169792175293, "learning_rate": 7.917552961147157e-07, "loss": 0.6003, "step": 16331 }, { "epoch": 2.63, "grad_norm": 12.524728775024414, "learning_rate": 7.910776581264334e-07, "loss": 0.7461, "step": 16332 }, { "epoch": 2.63, "grad_norm": 13.030937194824219, "learning_rate": 7.904002983040526e-07, "loss": 0.7452, "step": 16333 }, { "epoch": 2.63, "grad_norm": 15.154402732849121, "learning_rate": 7.897232166680324e-07, "loss": 0.7335, "step": 16334 }, { "epoch": 2.63, "grad_norm": 16.707008361816406, "learning_rate": 7.890464132388232e-07, "loss": 0.7563, "step": 16335 }, { "epoch": 2.63, "grad_norm": 12.5404691696167, "learning_rate": 7.883698880368707e-07, "loss": 0.6509, "step": 16336 }, { "epoch": 2.63, "grad_norm": 14.519432067871094, "learning_rate": 7.876936410826108e-07, "loss": 0.8433, "step": 16337 }, { "epoch": 2.63, "grad_norm": 13.142688751220703, "learning_rate": 7.870176723964673e-07, "loss": 0.477, "step": 16338 }, { "epoch": 2.63, "grad_norm": 12.038910865783691, "learning_rate": 7.863419819988627e-07, "loss": 0.5916, "step": 16339 }, { "epoch": 2.63, "grad_norm": 15.245637893676758, "learning_rate": 7.856665699102039e-07, "loss": 0.8101, "step": 16340 }, { "epoch": 2.63, "grad_norm": 15.694668769836426, "learning_rate": 7.849914361508925e-07, "loss": 0.6392, "step": 16341 }, { "epoch": 2.63, "grad_norm": 13.111193656921387, "learning_rate": 7.843165807413233e-07, "loss": 0.4562, "step": 16342 }, { "epoch": 2.63, "grad_norm": 12.989191055297852, "learning_rate": 7.83642003701881e-07, "loss": 0.7666, "step": 16343 }, { "epoch": 2.63, "grad_norm": 11.739825248718262, "learning_rate": 7.829677050529394e-07, "loss": 0.7087, "step": 16344 }, { "epoch": 2.63, "grad_norm": 12.566017150878906, "learning_rate": 7.822936848148699e-07, "loss": 0.548, "step": 16345 }, { "epoch": 2.63, "grad_norm": 13.981128692626953, "learning_rate": 7.816199430080285e-07, "loss": 0.728, "step": 16346 }, { "epoch": 2.63, "grad_norm": 13.67607593536377, "learning_rate": 7.809464796527699e-07, "loss": 0.5503, "step": 16347 }, { "epoch": 2.63, "grad_norm": 10.916146278381348, "learning_rate": 7.802732947694347e-07, "loss": 0.6865, "step": 16348 }, { "epoch": 2.63, "grad_norm": 12.771625518798828, "learning_rate": 7.796003883783565e-07, "loss": 0.6694, "step": 16349 }, { "epoch": 2.63, "grad_norm": 12.711926460266113, "learning_rate": 7.789277604998646e-07, "loss": 0.8577, "step": 16350 }, { "epoch": 2.63, "grad_norm": 12.73311996459961, "learning_rate": 7.782554111542718e-07, "loss": 0.5406, "step": 16351 }, { "epoch": 2.63, "grad_norm": 10.665966033935547, "learning_rate": 7.775833403618915e-07, "loss": 0.5866, "step": 16352 }, { "epoch": 2.63, "grad_norm": 12.45982551574707, "learning_rate": 7.769115481430234e-07, "loss": 0.8315, "step": 16353 }, { "epoch": 2.63, "grad_norm": 11.77833366394043, "learning_rate": 7.762400345179555e-07, "loss": 0.8751, "step": 16354 }, { "epoch": 2.63, "grad_norm": 13.010929107666016, "learning_rate": 7.755687995069783e-07, "loss": 0.6713, "step": 16355 }, { "epoch": 2.63, "grad_norm": 10.807249069213867, "learning_rate": 7.748978431303622e-07, "loss": 0.615, "step": 16356 }, { "epoch": 2.63, "grad_norm": 11.947830200195312, "learning_rate": 7.742271654083733e-07, "loss": 0.8006, "step": 16357 }, { "epoch": 2.63, "grad_norm": 13.439453125, "learning_rate": 7.735567663612765e-07, "loss": 0.5767, "step": 16358 }, { "epoch": 2.63, "grad_norm": 13.15829849243164, "learning_rate": 7.728866460093177e-07, "loss": 0.7501, "step": 16359 }, { "epoch": 2.63, "grad_norm": 13.322922706604004, "learning_rate": 7.722168043727373e-07, "loss": 0.6296, "step": 16360 }, { "epoch": 2.63, "grad_norm": 12.11036491394043, "learning_rate": 7.715472414717729e-07, "loss": 1.0054, "step": 16361 }, { "epoch": 2.63, "grad_norm": 12.510289192199707, "learning_rate": 7.708779573266445e-07, "loss": 0.7029, "step": 16362 }, { "epoch": 2.63, "grad_norm": 10.065898895263672, "learning_rate": 7.702089519575739e-07, "loss": 0.5306, "step": 16363 }, { "epoch": 2.63, "grad_norm": 11.907344818115234, "learning_rate": 7.69540225384765e-07, "loss": 0.8941, "step": 16364 }, { "epoch": 2.63, "grad_norm": 12.818624496459961, "learning_rate": 7.688717776284171e-07, "loss": 0.7301, "step": 16365 }, { "epoch": 2.63, "grad_norm": 11.732428550720215, "learning_rate": 7.682036087087252e-07, "loss": 0.713, "step": 16366 }, { "epoch": 2.63, "grad_norm": 13.251052856445312, "learning_rate": 7.675357186458709e-07, "loss": 0.6568, "step": 16367 }, { "epoch": 2.63, "grad_norm": 10.498558044433594, "learning_rate": 7.668681074600248e-07, "loss": 0.7437, "step": 16368 }, { "epoch": 2.64, "grad_norm": 12.60705852508545, "learning_rate": 7.662007751713573e-07, "loss": 0.7423, "step": 16369 }, { "epoch": 2.64, "grad_norm": 10.801237106323242, "learning_rate": 7.655337218000236e-07, "loss": 0.7208, "step": 16370 }, { "epoch": 2.64, "grad_norm": 14.203338623046875, "learning_rate": 7.648669473661741e-07, "loss": 0.8765, "step": 16371 }, { "epoch": 2.64, "grad_norm": 10.547239303588867, "learning_rate": 7.642004518899493e-07, "loss": 0.6874, "step": 16372 }, { "epoch": 2.64, "grad_norm": 13.775918960571289, "learning_rate": 7.635342353914799e-07, "loss": 0.775, "step": 16373 }, { "epoch": 2.64, "grad_norm": 11.21902084350586, "learning_rate": 7.628682978908919e-07, "loss": 0.8233, "step": 16374 }, { "epoch": 2.64, "grad_norm": 11.024986267089844, "learning_rate": 7.622026394082982e-07, "loss": 0.7638, "step": 16375 }, { "epoch": 2.64, "grad_norm": 13.115191459655762, "learning_rate": 7.615372599638082e-07, "loss": 0.509, "step": 16376 }, { "epoch": 2.64, "grad_norm": 12.20356273651123, "learning_rate": 7.608721595775192e-07, "loss": 0.7432, "step": 16377 }, { "epoch": 2.64, "grad_norm": 12.413921356201172, "learning_rate": 7.602073382695218e-07, "loss": 0.7274, "step": 16378 }, { "epoch": 2.64, "grad_norm": 13.083385467529297, "learning_rate": 7.595427960598967e-07, "loss": 0.6266, "step": 16379 }, { "epoch": 2.64, "grad_norm": 11.305678367614746, "learning_rate": 7.588785329687198e-07, "loss": 0.5192, "step": 16380 }, { "epoch": 2.64, "grad_norm": 12.89303207397461, "learning_rate": 7.58214549016052e-07, "loss": 0.7315, "step": 16381 }, { "epoch": 2.64, "grad_norm": 13.154669761657715, "learning_rate": 7.575508442219525e-07, "loss": 0.5325, "step": 16382 }, { "epoch": 2.64, "grad_norm": 10.934035301208496, "learning_rate": 7.568874186064701e-07, "loss": 0.7189, "step": 16383 }, { "epoch": 2.64, "grad_norm": 15.350048065185547, "learning_rate": 7.562242721896396e-07, "loss": 0.7175, "step": 16384 }, { "epoch": 2.64, "grad_norm": 15.796473503112793, "learning_rate": 7.555614049914973e-07, "loss": 0.9103, "step": 16385 }, { "epoch": 2.64, "grad_norm": 11.53281021118164, "learning_rate": 7.548988170320625e-07, "loss": 0.7481, "step": 16386 }, { "epoch": 2.64, "grad_norm": 11.202617645263672, "learning_rate": 7.542365083313519e-07, "loss": 0.6242, "step": 16387 }, { "epoch": 2.64, "grad_norm": 11.90116024017334, "learning_rate": 7.535744789093702e-07, "loss": 0.568, "step": 16388 }, { "epoch": 2.64, "grad_norm": 13.567298889160156, "learning_rate": 7.529127287861127e-07, "loss": 0.7911, "step": 16389 }, { "epoch": 2.64, "grad_norm": 15.114910125732422, "learning_rate": 7.522512579815721e-07, "loss": 0.7967, "step": 16390 }, { "epoch": 2.64, "grad_norm": 12.722458839416504, "learning_rate": 7.515900665157261e-07, "loss": 0.8096, "step": 16391 }, { "epoch": 2.64, "grad_norm": 12.115718841552734, "learning_rate": 7.509291544085474e-07, "loss": 0.8718, "step": 16392 }, { "epoch": 2.64, "grad_norm": 12.334766387939453, "learning_rate": 7.502685216799998e-07, "loss": 0.6851, "step": 16393 }, { "epoch": 2.64, "grad_norm": 11.355199813842773, "learning_rate": 7.496081683500378e-07, "loss": 0.624, "step": 16394 }, { "epoch": 2.64, "grad_norm": 12.880934715270996, "learning_rate": 7.489480944386096e-07, "loss": 0.6202, "step": 16395 }, { "epoch": 2.64, "grad_norm": 10.652776718139648, "learning_rate": 7.482882999656538e-07, "loss": 0.6434, "step": 16396 }, { "epoch": 2.64, "grad_norm": 11.903657913208008, "learning_rate": 7.476287849510944e-07, "loss": 0.652, "step": 16397 }, { "epoch": 2.64, "grad_norm": 10.568659782409668, "learning_rate": 7.469695494148621e-07, "loss": 0.7006, "step": 16398 }, { "epoch": 2.64, "grad_norm": 13.121891021728516, "learning_rate": 7.463105933768644e-07, "loss": 0.7333, "step": 16399 }, { "epoch": 2.64, "grad_norm": 10.877726554870605, "learning_rate": 7.456519168570054e-07, "loss": 0.5411, "step": 16400 }, { "epoch": 2.64, "grad_norm": 9.388611793518066, "learning_rate": 7.449935198751834e-07, "loss": 0.572, "step": 16401 }, { "epoch": 2.64, "grad_norm": 11.801629066467285, "learning_rate": 7.443354024512861e-07, "loss": 0.5107, "step": 16402 }, { "epoch": 2.64, "grad_norm": 13.441144943237305, "learning_rate": 7.436775646051886e-07, "loss": 0.6048, "step": 16403 }, { "epoch": 2.64, "grad_norm": 10.483705520629883, "learning_rate": 7.430200063567672e-07, "loss": 0.5957, "step": 16404 }, { "epoch": 2.64, "grad_norm": 15.706741333007812, "learning_rate": 7.423627277258804e-07, "loss": 0.7605, "step": 16405 }, { "epoch": 2.64, "grad_norm": 12.311861991882324, "learning_rate": 7.417057287323837e-07, "loss": 0.5113, "step": 16406 }, { "epoch": 2.64, "grad_norm": 13.06833267211914, "learning_rate": 7.410490093961231e-07, "loss": 0.9804, "step": 16407 }, { "epoch": 2.64, "grad_norm": 14.056097030639648, "learning_rate": 7.403925697369319e-07, "loss": 0.698, "step": 16408 }, { "epoch": 2.64, "grad_norm": 12.809821128845215, "learning_rate": 7.39736409774644e-07, "loss": 0.5181, "step": 16409 }, { "epoch": 2.64, "grad_norm": 15.258140563964844, "learning_rate": 7.390805295290738e-07, "loss": 0.6645, "step": 16410 }, { "epoch": 2.64, "grad_norm": 14.373905181884766, "learning_rate": 7.384249290200385e-07, "loss": 0.6671, "step": 16411 }, { "epoch": 2.64, "grad_norm": 18.526691436767578, "learning_rate": 7.37769608267338e-07, "loss": 0.8034, "step": 16412 }, { "epoch": 2.64, "grad_norm": 12.707304954528809, "learning_rate": 7.371145672907654e-07, "loss": 0.7178, "step": 16413 }, { "epoch": 2.64, "grad_norm": 13.072712898254395, "learning_rate": 7.364598061101114e-07, "loss": 0.8419, "step": 16414 }, { "epoch": 2.64, "grad_norm": 11.747241020202637, "learning_rate": 7.358053247451513e-07, "loss": 0.61, "step": 16415 }, { "epoch": 2.64, "grad_norm": 12.185014724731445, "learning_rate": 7.351511232156527e-07, "loss": 0.6959, "step": 16416 }, { "epoch": 2.64, "grad_norm": 14.054444313049316, "learning_rate": 7.344972015413809e-07, "loss": 0.6446, "step": 16417 }, { "epoch": 2.64, "grad_norm": 12.403172492980957, "learning_rate": 7.338435597420835e-07, "loss": 0.5892, "step": 16418 }, { "epoch": 2.64, "grad_norm": 14.477336883544922, "learning_rate": 7.331901978375066e-07, "loss": 0.7219, "step": 16419 }, { "epoch": 2.64, "grad_norm": 11.233479499816895, "learning_rate": 7.325371158473882e-07, "loss": 0.5834, "step": 16420 }, { "epoch": 2.64, "grad_norm": 14.035325050354004, "learning_rate": 7.318843137914533e-07, "loss": 0.7288, "step": 16421 }, { "epoch": 2.64, "grad_norm": 13.05270767211914, "learning_rate": 7.312317916894185e-07, "loss": 0.7521, "step": 16422 }, { "epoch": 2.64, "grad_norm": 11.893444061279297, "learning_rate": 7.305795495609979e-07, "loss": 0.6144, "step": 16423 }, { "epoch": 2.64, "grad_norm": 14.207250595092773, "learning_rate": 7.299275874258893e-07, "loss": 0.6004, "step": 16424 }, { "epoch": 2.64, "grad_norm": 14.281034469604492, "learning_rate": 7.29275905303789e-07, "loss": 0.8413, "step": 16425 }, { "epoch": 2.64, "grad_norm": 14.501853942871094, "learning_rate": 7.286245032143812e-07, "loss": 0.6497, "step": 16426 }, { "epoch": 2.64, "grad_norm": 11.80456829071045, "learning_rate": 7.279733811773393e-07, "loss": 0.6775, "step": 16427 }, { "epoch": 2.64, "grad_norm": 11.564451217651367, "learning_rate": 7.273225392123362e-07, "loss": 0.7463, "step": 16428 }, { "epoch": 2.64, "grad_norm": 12.099447250366211, "learning_rate": 7.266719773390263e-07, "loss": 0.6498, "step": 16429 }, { "epoch": 2.64, "grad_norm": 12.355546951293945, "learning_rate": 7.26021695577065e-07, "loss": 0.8124, "step": 16430 }, { "epoch": 2.65, "grad_norm": 12.239715576171875, "learning_rate": 7.25371693946092e-07, "loss": 0.8607, "step": 16431 }, { "epoch": 2.65, "grad_norm": 12.86723804473877, "learning_rate": 7.247219724657406e-07, "loss": 0.6592, "step": 16432 }, { "epoch": 2.65, "grad_norm": 11.505767822265625, "learning_rate": 7.240725311556396e-07, "loss": 0.7223, "step": 16433 }, { "epoch": 2.65, "grad_norm": 10.55742073059082, "learning_rate": 7.234233700354043e-07, "loss": 0.5631, "step": 16434 }, { "epoch": 2.65, "grad_norm": 12.536115646362305, "learning_rate": 7.227744891246425e-07, "loss": 0.6939, "step": 16435 }, { "epoch": 2.65, "grad_norm": 13.650063514709473, "learning_rate": 7.22125888442956e-07, "loss": 0.8853, "step": 16436 }, { "epoch": 2.65, "grad_norm": 15.399978637695312, "learning_rate": 7.214775680099339e-07, "loss": 0.5925, "step": 16437 }, { "epoch": 2.65, "grad_norm": 11.186888694763184, "learning_rate": 7.208295278451615e-07, "loss": 0.6715, "step": 16438 }, { "epoch": 2.65, "grad_norm": 13.795565605163574, "learning_rate": 7.201817679682154e-07, "loss": 0.6393, "step": 16439 }, { "epoch": 2.65, "grad_norm": 13.930750846862793, "learning_rate": 7.195342883986578e-07, "loss": 0.7308, "step": 16440 }, { "epoch": 2.65, "grad_norm": 16.018848419189453, "learning_rate": 7.188870891560507e-07, "loss": 0.6031, "step": 16441 }, { "epoch": 2.65, "grad_norm": 14.81799602508545, "learning_rate": 7.182401702599418e-07, "loss": 0.6872, "step": 16442 }, { "epoch": 2.65, "grad_norm": 11.436456680297852, "learning_rate": 7.175935317298688e-07, "loss": 0.6164, "step": 16443 }, { "epoch": 2.65, "grad_norm": 12.70578670501709, "learning_rate": 7.169471735853694e-07, "loss": 0.5144, "step": 16444 }, { "epoch": 2.65, "grad_norm": 16.107656478881836, "learning_rate": 7.16301095845966e-07, "loss": 0.7621, "step": 16445 }, { "epoch": 2.65, "grad_norm": 14.07430648803711, "learning_rate": 7.156552985311705e-07, "loss": 0.5842, "step": 16446 }, { "epoch": 2.65, "grad_norm": 13.66617488861084, "learning_rate": 7.150097816604951e-07, "loss": 0.6873, "step": 16447 }, { "epoch": 2.65, "grad_norm": 10.419227600097656, "learning_rate": 7.143645452534331e-07, "loss": 0.5818, "step": 16448 }, { "epoch": 2.65, "grad_norm": 11.159344673156738, "learning_rate": 7.137195893294802e-07, "loss": 0.5436, "step": 16449 }, { "epoch": 2.65, "grad_norm": 13.124348640441895, "learning_rate": 7.13074913908115e-07, "loss": 0.6669, "step": 16450 }, { "epoch": 2.65, "grad_norm": 11.279080390930176, "learning_rate": 7.124305190088099e-07, "loss": 0.652, "step": 16451 }, { "epoch": 2.65, "grad_norm": 11.78475570678711, "learning_rate": 7.117864046510315e-07, "loss": 0.4875, "step": 16452 }, { "epoch": 2.65, "grad_norm": 12.487457275390625, "learning_rate": 7.111425708542341e-07, "loss": 0.7195, "step": 16453 }, { "epoch": 2.65, "grad_norm": 14.295140266418457, "learning_rate": 7.104990176378678e-07, "loss": 0.6002, "step": 16454 }, { "epoch": 2.65, "grad_norm": 12.390015602111816, "learning_rate": 7.098557450213706e-07, "loss": 0.6451, "step": 16455 }, { "epoch": 2.65, "grad_norm": 12.611427307128906, "learning_rate": 7.092127530241711e-07, "loss": 0.8518, "step": 16456 }, { "epoch": 2.65, "grad_norm": 13.021202087402344, "learning_rate": 7.085700416656959e-07, "loss": 0.7843, "step": 16457 }, { "epoch": 2.65, "grad_norm": 14.764209747314453, "learning_rate": 7.079276109653544e-07, "loss": 0.9031, "step": 16458 }, { "epoch": 2.65, "grad_norm": 12.372721672058105, "learning_rate": 7.07285460942555e-07, "loss": 0.8352, "step": 16459 }, { "epoch": 2.65, "grad_norm": 16.282976150512695, "learning_rate": 7.066435916166947e-07, "loss": 0.6523, "step": 16460 }, { "epoch": 2.65, "grad_norm": 12.266618728637695, "learning_rate": 7.060020030071613e-07, "loss": 0.7214, "step": 16461 }, { "epoch": 2.65, "grad_norm": 14.583456039428711, "learning_rate": 7.053606951333325e-07, "loss": 0.5501, "step": 16462 }, { "epoch": 2.65, "grad_norm": 16.059444427490234, "learning_rate": 7.047196680145829e-07, "loss": 0.9873, "step": 16463 }, { "epoch": 2.65, "grad_norm": 10.8161039352417, "learning_rate": 7.040789216702726e-07, "loss": 0.4622, "step": 16464 }, { "epoch": 2.65, "grad_norm": 11.67310619354248, "learning_rate": 7.034384561197594e-07, "loss": 0.6555, "step": 16465 }, { "epoch": 2.65, "grad_norm": 14.239501953125, "learning_rate": 7.027982713823878e-07, "loss": 0.9136, "step": 16466 }, { "epoch": 2.65, "grad_norm": 13.63229751586914, "learning_rate": 7.021583674774934e-07, "loss": 0.5965, "step": 16467 }, { "epoch": 2.65, "grad_norm": 11.762157440185547, "learning_rate": 7.015187444244075e-07, "loss": 0.6352, "step": 16468 }, { "epoch": 2.65, "grad_norm": 12.94653034210205, "learning_rate": 7.008794022424503e-07, "loss": 0.6367, "step": 16469 }, { "epoch": 2.65, "grad_norm": 12.841147422790527, "learning_rate": 7.002403409509317e-07, "loss": 0.7337, "step": 16470 }, { "epoch": 2.65, "grad_norm": 11.426185607910156, "learning_rate": 6.996015605691598e-07, "loss": 0.6217, "step": 16471 }, { "epoch": 2.65, "grad_norm": 10.972729682922363, "learning_rate": 6.989630611164244e-07, "loss": 0.8023, "step": 16472 }, { "epoch": 2.65, "grad_norm": 12.906617164611816, "learning_rate": 6.98324842612016e-07, "loss": 0.6261, "step": 16473 }, { "epoch": 2.65, "grad_norm": 16.170513153076172, "learning_rate": 6.976869050752111e-07, "loss": 0.5979, "step": 16474 }, { "epoch": 2.65, "grad_norm": 10.707565307617188, "learning_rate": 6.970492485252789e-07, "loss": 0.6026, "step": 16475 }, { "epoch": 2.65, "grad_norm": 12.692472457885742, "learning_rate": 6.964118729814817e-07, "loss": 0.5845, "step": 16476 }, { "epoch": 2.65, "grad_norm": 12.848132133483887, "learning_rate": 6.957747784630709e-07, "loss": 0.6719, "step": 16477 }, { "epoch": 2.65, "grad_norm": 13.614164352416992, "learning_rate": 6.951379649892908e-07, "loss": 0.4949, "step": 16478 }, { "epoch": 2.65, "grad_norm": 11.876726150512695, "learning_rate": 6.945014325793797e-07, "loss": 0.5922, "step": 16479 }, { "epoch": 2.65, "grad_norm": 12.786704063415527, "learning_rate": 6.938651812525631e-07, "loss": 0.8457, "step": 16480 }, { "epoch": 2.65, "grad_norm": 13.424882888793945, "learning_rate": 6.93229211028057e-07, "loss": 0.6062, "step": 16481 }, { "epoch": 2.65, "grad_norm": 12.073450088500977, "learning_rate": 6.925935219250756e-07, "loss": 0.7611, "step": 16482 }, { "epoch": 2.65, "grad_norm": 13.191503524780273, "learning_rate": 6.919581139628173e-07, "loss": 0.6117, "step": 16483 }, { "epoch": 2.65, "grad_norm": 14.507493019104004, "learning_rate": 6.913229871604788e-07, "loss": 0.8446, "step": 16484 }, { "epoch": 2.65, "grad_norm": 13.203849792480469, "learning_rate": 6.906881415372435e-07, "loss": 0.8655, "step": 16485 }, { "epoch": 2.65, "grad_norm": 14.96715259552002, "learning_rate": 6.900535771122841e-07, "loss": 0.9089, "step": 16486 }, { "epoch": 2.65, "grad_norm": 14.674302101135254, "learning_rate": 6.894192939047739e-07, "loss": 0.758, "step": 16487 }, { "epoch": 2.65, "grad_norm": 11.661604881286621, "learning_rate": 6.887852919338678e-07, "loss": 0.7952, "step": 16488 }, { "epoch": 2.65, "grad_norm": 13.140823364257812, "learning_rate": 6.881515712187203e-07, "loss": 0.8501, "step": 16489 }, { "epoch": 2.65, "grad_norm": 17.707101821899414, "learning_rate": 6.875181317784718e-07, "loss": 0.7611, "step": 16490 }, { "epoch": 2.65, "grad_norm": 13.67084789276123, "learning_rate": 6.868849736322536e-07, "loss": 0.9349, "step": 16491 }, { "epoch": 2.65, "grad_norm": 13.33730411529541, "learning_rate": 6.862520967991948e-07, "loss": 0.7206, "step": 16492 }, { "epoch": 2.66, "grad_norm": 13.876432418823242, "learning_rate": 6.856195012984113e-07, "loss": 0.7243, "step": 16493 }, { "epoch": 2.66, "grad_norm": 13.544193267822266, "learning_rate": 6.849871871490088e-07, "loss": 0.6897, "step": 16494 }, { "epoch": 2.66, "grad_norm": 83.51890563964844, "learning_rate": 6.843551543700911e-07, "loss": 0.7601, "step": 16495 }, { "epoch": 2.66, "grad_norm": 14.508230209350586, "learning_rate": 6.837234029807449e-07, "loss": 0.7807, "step": 16496 }, { "epoch": 2.66, "grad_norm": 14.42561149597168, "learning_rate": 6.830919330000574e-07, "loss": 0.7347, "step": 16497 }, { "epoch": 2.66, "grad_norm": 13.459421157836914, "learning_rate": 6.82460744447101e-07, "loss": 0.8085, "step": 16498 }, { "epoch": 2.66, "grad_norm": 13.162909507751465, "learning_rate": 6.818298373409393e-07, "loss": 0.8325, "step": 16499 }, { "epoch": 2.66, "grad_norm": 11.96353816986084, "learning_rate": 6.811992117006316e-07, "loss": 0.5225, "step": 16500 }, { "epoch": 2.66, "grad_norm": 15.168801307678223, "learning_rate": 6.805688675452293e-07, "loss": 0.6533, "step": 16501 }, { "epoch": 2.66, "grad_norm": 14.916773796081543, "learning_rate": 6.799388048937671e-07, "loss": 0.6659, "step": 16502 }, { "epoch": 2.66, "grad_norm": 11.692219734191895, "learning_rate": 6.793090237652822e-07, "loss": 0.4402, "step": 16503 }, { "epoch": 2.66, "grad_norm": 13.131360054016113, "learning_rate": 6.786795241787947e-07, "loss": 0.7042, "step": 16504 }, { "epoch": 2.66, "grad_norm": 13.025115966796875, "learning_rate": 6.780503061533183e-07, "loss": 0.7277, "step": 16505 }, { "epoch": 2.66, "grad_norm": 12.265660285949707, "learning_rate": 6.774213697078624e-07, "loss": 0.5067, "step": 16506 }, { "epoch": 2.66, "grad_norm": 13.559223175048828, "learning_rate": 6.767927148614228e-07, "loss": 0.5667, "step": 16507 }, { "epoch": 2.66, "grad_norm": 11.8176908493042, "learning_rate": 6.7616434163299e-07, "loss": 0.7241, "step": 16508 }, { "epoch": 2.66, "grad_norm": 12.37086296081543, "learning_rate": 6.75536250041543e-07, "loss": 0.7004, "step": 16509 }, { "epoch": 2.66, "grad_norm": 16.00615882873535, "learning_rate": 6.749084401060546e-07, "loss": 0.9652, "step": 16510 }, { "epoch": 2.66, "grad_norm": 14.75199031829834, "learning_rate": 6.742809118454896e-07, "loss": 0.5158, "step": 16511 }, { "epoch": 2.66, "grad_norm": 15.733480453491211, "learning_rate": 6.736536652788029e-07, "loss": 0.7029, "step": 16512 }, { "epoch": 2.66, "grad_norm": 12.814377784729004, "learning_rate": 6.730267004249391e-07, "loss": 0.4187, "step": 16513 }, { "epoch": 2.66, "grad_norm": 13.518099784851074, "learning_rate": 6.724000173028389e-07, "loss": 0.6148, "step": 16514 }, { "epoch": 2.66, "grad_norm": 13.422697067260742, "learning_rate": 6.717736159314303e-07, "loss": 0.8826, "step": 16515 }, { "epoch": 2.66, "grad_norm": 15.175776481628418, "learning_rate": 6.711474963296372e-07, "loss": 0.6241, "step": 16516 }, { "epoch": 2.66, "grad_norm": 13.335365295410156, "learning_rate": 6.705216585163687e-07, "loss": 0.7268, "step": 16517 }, { "epoch": 2.66, "grad_norm": 15.010405540466309, "learning_rate": 6.698961025105299e-07, "loss": 0.8792, "step": 16518 }, { "epoch": 2.66, "grad_norm": 13.34945297241211, "learning_rate": 6.692708283310167e-07, "loss": 0.7125, "step": 16519 }, { "epoch": 2.66, "grad_norm": 14.85758113861084, "learning_rate": 6.686458359967185e-07, "loss": 0.7818, "step": 16520 }, { "epoch": 2.66, "grad_norm": 10.591180801391602, "learning_rate": 6.680211255265101e-07, "loss": 0.4357, "step": 16521 }, { "epoch": 2.66, "grad_norm": 13.13106918334961, "learning_rate": 6.673966969392653e-07, "loss": 0.7096, "step": 16522 }, { "epoch": 2.66, "grad_norm": 13.197123527526855, "learning_rate": 6.667725502538436e-07, "loss": 1.0365, "step": 16523 }, { "epoch": 2.66, "grad_norm": 11.872964859008789, "learning_rate": 6.661486854890964e-07, "loss": 0.6049, "step": 16524 }, { "epoch": 2.66, "grad_norm": 15.00702953338623, "learning_rate": 6.65525102663872e-07, "loss": 0.852, "step": 16525 }, { "epoch": 2.66, "grad_norm": 11.214427947998047, "learning_rate": 6.649018017970032e-07, "loss": 0.582, "step": 16526 }, { "epoch": 2.66, "grad_norm": 10.603194236755371, "learning_rate": 6.642787829073205e-07, "loss": 0.5541, "step": 16527 }, { "epoch": 2.66, "grad_norm": 11.373429298400879, "learning_rate": 6.636560460136399e-07, "loss": 0.6974, "step": 16528 }, { "epoch": 2.66, "grad_norm": 11.113858222961426, "learning_rate": 6.630335911347729e-07, "loss": 0.5996, "step": 16529 }, { "epoch": 2.66, "grad_norm": 14.349442481994629, "learning_rate": 6.624114182895224e-07, "loss": 0.7252, "step": 16530 }, { "epoch": 2.66, "grad_norm": 13.072288513183594, "learning_rate": 6.617895274966801e-07, "loss": 0.665, "step": 16531 }, { "epoch": 2.66, "grad_norm": 12.382986068725586, "learning_rate": 6.611679187750342e-07, "loss": 0.5869, "step": 16532 }, { "epoch": 2.66, "grad_norm": 12.296388626098633, "learning_rate": 6.605465921433585e-07, "loss": 0.7143, "step": 16533 }, { "epoch": 2.66, "grad_norm": 11.395082473754883, "learning_rate": 6.599255476204192e-07, "loss": 0.5498, "step": 16534 }, { "epoch": 2.66, "grad_norm": 14.083349227905273, "learning_rate": 6.59304785224979e-07, "loss": 0.8631, "step": 16535 }, { "epoch": 2.66, "grad_norm": 16.541969299316406, "learning_rate": 6.586843049757885e-07, "loss": 0.6261, "step": 16536 }, { "epoch": 2.66, "grad_norm": 13.891805648803711, "learning_rate": 6.580641068915861e-07, "loss": 0.6563, "step": 16537 }, { "epoch": 2.66, "grad_norm": 12.814629554748535, "learning_rate": 6.574441909911111e-07, "loss": 0.757, "step": 16538 }, { "epoch": 2.66, "grad_norm": 13.922740936279297, "learning_rate": 6.568245572930842e-07, "loss": 0.53, "step": 16539 }, { "epoch": 2.66, "grad_norm": 11.294578552246094, "learning_rate": 6.562052058162238e-07, "loss": 0.9366, "step": 16540 }, { "epoch": 2.66, "grad_norm": 11.238543510437012, "learning_rate": 6.555861365792393e-07, "loss": 0.7064, "step": 16541 }, { "epoch": 2.66, "grad_norm": 11.170866012573242, "learning_rate": 6.549673496008291e-07, "loss": 0.7842, "step": 16542 }, { "epoch": 2.66, "grad_norm": 13.034015655517578, "learning_rate": 6.54348844899686e-07, "loss": 0.7014, "step": 16543 }, { "epoch": 2.66, "grad_norm": 11.348367691040039, "learning_rate": 6.537306224944906e-07, "loss": 0.9095, "step": 16544 }, { "epoch": 2.66, "grad_norm": 14.785415649414062, "learning_rate": 6.531126824039169e-07, "loss": 0.765, "step": 16545 }, { "epoch": 2.66, "grad_norm": 13.210151672363281, "learning_rate": 6.524950246466322e-07, "loss": 0.6254, "step": 16546 }, { "epoch": 2.66, "grad_norm": 14.488924026489258, "learning_rate": 6.518776492412926e-07, "loss": 0.8955, "step": 16547 }, { "epoch": 2.66, "grad_norm": 11.87473201751709, "learning_rate": 6.512605562065455e-07, "loss": 0.6793, "step": 16548 }, { "epoch": 2.66, "grad_norm": 10.975974082946777, "learning_rate": 6.506437455610337e-07, "loss": 0.6163, "step": 16549 }, { "epoch": 2.66, "grad_norm": 17.636320114135742, "learning_rate": 6.500272173233845e-07, "loss": 0.8629, "step": 16550 }, { "epoch": 2.66, "grad_norm": 11.618457794189453, "learning_rate": 6.494109715122265e-07, "loss": 0.6049, "step": 16551 }, { "epoch": 2.66, "grad_norm": 11.867557525634766, "learning_rate": 6.487950081461703e-07, "loss": 0.5359, "step": 16552 }, { "epoch": 2.66, "grad_norm": 15.636114120483398, "learning_rate": 6.481793272438208e-07, "loss": 0.6254, "step": 16553 }, { "epoch": 2.66, "grad_norm": 12.367016792297363, "learning_rate": 6.475639288237778e-07, "loss": 0.6195, "step": 16554 }, { "epoch": 2.67, "grad_norm": 13.326300621032715, "learning_rate": 6.469488129046287e-07, "loss": 0.8435, "step": 16555 }, { "epoch": 2.67, "grad_norm": 17.59032440185547, "learning_rate": 6.463339795049551e-07, "loss": 0.7335, "step": 16556 }, { "epoch": 2.67, "grad_norm": 11.405693054199219, "learning_rate": 6.457194286433288e-07, "loss": 0.632, "step": 16557 }, { "epoch": 2.67, "grad_norm": 22.963150024414062, "learning_rate": 6.451051603383096e-07, "loss": 0.7242, "step": 16558 }, { "epoch": 2.67, "grad_norm": 12.237308502197266, "learning_rate": 6.444911746084559e-07, "loss": 0.7692, "step": 16559 }, { "epoch": 2.67, "grad_norm": 12.425680160522461, "learning_rate": 6.438774714723129e-07, "loss": 0.6574, "step": 16560 }, { "epoch": 2.67, "grad_norm": 15.554630279541016, "learning_rate": 6.432640509484178e-07, "loss": 0.6845, "step": 16561 }, { "epoch": 2.67, "grad_norm": 14.213847160339355, "learning_rate": 6.426509130553005e-07, "loss": 0.6045, "step": 16562 }, { "epoch": 2.67, "grad_norm": 11.959364891052246, "learning_rate": 6.420380578114805e-07, "loss": 0.631, "step": 16563 }, { "epoch": 2.67, "grad_norm": 14.663695335388184, "learning_rate": 6.414254852354696e-07, "loss": 0.6129, "step": 16564 }, { "epoch": 2.67, "grad_norm": 11.067789077758789, "learning_rate": 6.40813195345773e-07, "loss": 0.5059, "step": 16565 }, { "epoch": 2.67, "grad_norm": 14.948331832885742, "learning_rate": 6.402011881608827e-07, "loss": 0.5974, "step": 16566 }, { "epoch": 2.67, "grad_norm": 15.01725959777832, "learning_rate": 6.395894636992884e-07, "loss": 0.6326, "step": 16567 }, { "epoch": 2.67, "grad_norm": 12.902585983276367, "learning_rate": 6.389780219794662e-07, "loss": 0.7353, "step": 16568 }, { "epoch": 2.67, "grad_norm": 14.41609001159668, "learning_rate": 6.383668630198836e-07, "loss": 0.7885, "step": 16569 }, { "epoch": 2.67, "grad_norm": 13.08315372467041, "learning_rate": 6.377559868390048e-07, "loss": 0.6208, "step": 16570 }, { "epoch": 2.67, "grad_norm": 17.78893280029297, "learning_rate": 6.371453934552807e-07, "loss": 0.6862, "step": 16571 }, { "epoch": 2.67, "grad_norm": 12.694583892822266, "learning_rate": 6.36535082887153e-07, "loss": 0.9152, "step": 16572 }, { "epoch": 2.67, "grad_norm": 14.784668922424316, "learning_rate": 6.359250551530605e-07, "loss": 0.8167, "step": 16573 }, { "epoch": 2.67, "grad_norm": 13.722972869873047, "learning_rate": 6.35315310271426e-07, "loss": 0.7568, "step": 16574 }, { "epoch": 2.67, "grad_norm": 11.697863578796387, "learning_rate": 6.347058482606705e-07, "loss": 0.5911, "step": 16575 }, { "epoch": 2.67, "grad_norm": 14.345376014709473, "learning_rate": 6.340966691392025e-07, "loss": 0.8468, "step": 16576 }, { "epoch": 2.67, "grad_norm": 13.447046279907227, "learning_rate": 6.334877729254219e-07, "loss": 0.7434, "step": 16577 }, { "epoch": 2.67, "grad_norm": 12.272762298583984, "learning_rate": 6.328791596377226e-07, "loss": 0.6136, "step": 16578 }, { "epoch": 2.67, "grad_norm": 14.268927574157715, "learning_rate": 6.322708292944879e-07, "loss": 0.6942, "step": 16579 }, { "epoch": 2.67, "grad_norm": 11.567676544189453, "learning_rate": 6.316627819140908e-07, "loss": 0.7395, "step": 16580 }, { "epoch": 2.67, "grad_norm": 11.311169624328613, "learning_rate": 6.310550175149022e-07, "loss": 0.4621, "step": 16581 }, { "epoch": 2.67, "grad_norm": 11.998205184936523, "learning_rate": 6.304475361152784e-07, "loss": 0.6961, "step": 16582 }, { "epoch": 2.67, "grad_norm": 11.518183708190918, "learning_rate": 6.298403377335682e-07, "loss": 0.7908, "step": 16583 }, { "epoch": 2.67, "grad_norm": 12.649112701416016, "learning_rate": 6.292334223881147e-07, "loss": 0.6296, "step": 16584 }, { "epoch": 2.67, "grad_norm": 15.501100540161133, "learning_rate": 6.286267900972476e-07, "loss": 0.87, "step": 16585 }, { "epoch": 2.67, "grad_norm": 13.019389152526855, "learning_rate": 6.280204408792945e-07, "loss": 0.6423, "step": 16586 }, { "epoch": 2.67, "grad_norm": 13.66777515411377, "learning_rate": 6.274143747525696e-07, "loss": 0.5703, "step": 16587 }, { "epoch": 2.67, "grad_norm": 11.964396476745605, "learning_rate": 6.268085917353761e-07, "loss": 1.0705, "step": 16588 }, { "epoch": 2.67, "grad_norm": 11.701813697814941, "learning_rate": 6.262030918460182e-07, "loss": 0.6241, "step": 16589 }, { "epoch": 2.67, "grad_norm": 12.014178276062012, "learning_rate": 6.255978751027825e-07, "loss": 0.775, "step": 16590 }, { "epoch": 2.67, "grad_norm": 13.772254943847656, "learning_rate": 6.249929415239498e-07, "loss": 0.6021, "step": 16591 }, { "epoch": 2.67, "grad_norm": 12.104971885681152, "learning_rate": 6.243882911277943e-07, "loss": 0.7308, "step": 16592 }, { "epoch": 2.67, "grad_norm": 12.753291130065918, "learning_rate": 6.237839239325783e-07, "loss": 0.5091, "step": 16593 }, { "epoch": 2.67, "grad_norm": 13.377991676330566, "learning_rate": 6.231798399565603e-07, "loss": 0.4503, "step": 16594 }, { "epoch": 2.67, "grad_norm": 16.193777084350586, "learning_rate": 6.225760392179858e-07, "loss": 0.6959, "step": 16595 }, { "epoch": 2.67, "grad_norm": 12.240992546081543, "learning_rate": 6.219725217350925e-07, "loss": 0.7011, "step": 16596 }, { "epoch": 2.67, "grad_norm": 17.03488540649414, "learning_rate": 6.213692875261113e-07, "loss": 0.7388, "step": 16597 }, { "epoch": 2.67, "grad_norm": 13.667428970336914, "learning_rate": 6.20766336609262e-07, "loss": 0.6584, "step": 16598 }, { "epoch": 2.67, "grad_norm": 15.092317581176758, "learning_rate": 6.201636690027602e-07, "loss": 0.8003, "step": 16599 }, { "epoch": 2.67, "grad_norm": 13.14689826965332, "learning_rate": 6.195612847248078e-07, "loss": 0.7654, "step": 16600 }, { "epoch": 2.67, "grad_norm": 15.563765525817871, "learning_rate": 6.189591837936016e-07, "loss": 0.6386, "step": 16601 }, { "epoch": 2.67, "grad_norm": 11.985808372497559, "learning_rate": 6.183573662273279e-07, "loss": 0.7194, "step": 16602 }, { "epoch": 2.67, "grad_norm": 12.64080810546875, "learning_rate": 6.177558320441679e-07, "loss": 0.8756, "step": 16603 }, { "epoch": 2.67, "grad_norm": 13.278112411499023, "learning_rate": 6.17154581262287e-07, "loss": 0.7538, "step": 16604 }, { "epoch": 2.67, "grad_norm": 12.629366874694824, "learning_rate": 6.165536138998507e-07, "loss": 0.7715, "step": 16605 }, { "epoch": 2.67, "grad_norm": 14.384873390197754, "learning_rate": 6.159529299750111e-07, "loss": 0.8803, "step": 16606 }, { "epoch": 2.67, "grad_norm": 12.812180519104004, "learning_rate": 6.153525295059104e-07, "loss": 0.6547, "step": 16607 }, { "epoch": 2.67, "grad_norm": 11.346089363098145, "learning_rate": 6.147524125106874e-07, "loss": 0.8313, "step": 16608 }, { "epoch": 2.67, "grad_norm": 12.169885635375977, "learning_rate": 6.141525790074653e-07, "loss": 0.7547, "step": 16609 }, { "epoch": 2.67, "grad_norm": 12.676332473754883, "learning_rate": 6.135530290143676e-07, "loss": 0.7512, "step": 16610 }, { "epoch": 2.67, "grad_norm": 17.073171615600586, "learning_rate": 6.129537625495019e-07, "loss": 0.7673, "step": 16611 }, { "epoch": 2.67, "grad_norm": 12.928325653076172, "learning_rate": 6.123547796309681e-07, "loss": 0.734, "step": 16612 }, { "epoch": 2.67, "grad_norm": 11.689065933227539, "learning_rate": 6.117560802768629e-07, "loss": 0.6152, "step": 16613 }, { "epoch": 2.67, "grad_norm": 11.942442893981934, "learning_rate": 6.111576645052675e-07, "loss": 0.5962, "step": 16614 }, { "epoch": 2.67, "grad_norm": 11.584870338439941, "learning_rate": 6.105595323342583e-07, "loss": 0.7066, "step": 16615 }, { "epoch": 2.67, "grad_norm": 12.066985130310059, "learning_rate": 6.099616837819045e-07, "loss": 0.6723, "step": 16616 }, { "epoch": 2.67, "grad_norm": 14.501131057739258, "learning_rate": 6.093641188662624e-07, "loss": 0.7903, "step": 16617 }, { "epoch": 2.68, "grad_norm": 12.227418899536133, "learning_rate": 6.087668376053834e-07, "loss": 0.7932, "step": 16618 }, { "epoch": 2.68, "grad_norm": 14.95718002319336, "learning_rate": 6.081698400173097e-07, "loss": 0.8095, "step": 16619 }, { "epoch": 2.68, "grad_norm": 13.612425804138184, "learning_rate": 6.075731261200701e-07, "loss": 0.9551, "step": 16620 }, { "epoch": 2.68, "grad_norm": 13.382899284362793, "learning_rate": 6.069766959316958e-07, "loss": 0.685, "step": 16621 }, { "epoch": 2.68, "grad_norm": 13.662158012390137, "learning_rate": 6.063805494701991e-07, "loss": 0.6621, "step": 16622 }, { "epoch": 2.68, "grad_norm": 11.681133270263672, "learning_rate": 6.057846867535854e-07, "loss": 0.7701, "step": 16623 }, { "epoch": 2.68, "grad_norm": 13.221078872680664, "learning_rate": 6.051891077998573e-07, "loss": 0.7382, "step": 16624 }, { "epoch": 2.68, "grad_norm": 13.695981979370117, "learning_rate": 6.045938126270034e-07, "loss": 0.8471, "step": 16625 }, { "epoch": 2.68, "grad_norm": 12.089920997619629, "learning_rate": 6.039988012530018e-07, "loss": 0.6724, "step": 16626 }, { "epoch": 2.68, "grad_norm": 13.992083549499512, "learning_rate": 6.034040736958313e-07, "loss": 0.983, "step": 16627 }, { "epoch": 2.68, "grad_norm": 16.53521728515625, "learning_rate": 6.02809629973452e-07, "loss": 0.6984, "step": 16628 }, { "epoch": 2.68, "grad_norm": 12.864591598510742, "learning_rate": 6.022154701038218e-07, "loss": 0.622, "step": 16629 }, { "epoch": 2.68, "grad_norm": 10.959638595581055, "learning_rate": 6.016215941048886e-07, "loss": 0.6422, "step": 16630 }, { "epoch": 2.68, "grad_norm": 13.315329551696777, "learning_rate": 6.01028001994588e-07, "loss": 0.6874, "step": 16631 }, { "epoch": 2.68, "grad_norm": 9.618961334228516, "learning_rate": 6.004346937908534e-07, "loss": 0.6355, "step": 16632 }, { "epoch": 2.68, "grad_norm": 10.908180236816406, "learning_rate": 5.998416695116038e-07, "loss": 0.9228, "step": 16633 }, { "epoch": 2.68, "grad_norm": 11.738523483276367, "learning_rate": 5.992489291747549e-07, "loss": 0.6107, "step": 16634 }, { "epoch": 2.68, "grad_norm": 12.620335578918457, "learning_rate": 5.986564727982102e-07, "loss": 0.7542, "step": 16635 }, { "epoch": 2.68, "grad_norm": 12.4873685836792, "learning_rate": 5.98064300399862e-07, "loss": 0.6085, "step": 16636 }, { "epoch": 2.68, "grad_norm": 12.696812629699707, "learning_rate": 5.974724119976028e-07, "loss": 0.7233, "step": 16637 }, { "epoch": 2.68, "grad_norm": 12.730569839477539, "learning_rate": 5.968808076093091e-07, "loss": 0.5182, "step": 16638 }, { "epoch": 2.68, "grad_norm": 13.15621280670166, "learning_rate": 5.962894872528491e-07, "loss": 0.5652, "step": 16639 }, { "epoch": 2.68, "grad_norm": 13.684877395629883, "learning_rate": 5.956984509460861e-07, "loss": 0.6998, "step": 16640 }, { "epoch": 2.68, "grad_norm": 12.83592414855957, "learning_rate": 5.951076987068738e-07, "loss": 0.581, "step": 16641 }, { "epoch": 2.68, "grad_norm": 16.694080352783203, "learning_rate": 5.945172305530544e-07, "loss": 0.5665, "step": 16642 }, { "epoch": 2.68, "grad_norm": 13.394996643066406, "learning_rate": 5.939270465024671e-07, "loss": 0.5195, "step": 16643 }, { "epoch": 2.68, "grad_norm": 13.417571067810059, "learning_rate": 5.933371465729343e-07, "loss": 0.6391, "step": 16644 }, { "epoch": 2.68, "grad_norm": 13.717377662658691, "learning_rate": 5.927475307822794e-07, "loss": 0.8125, "step": 16645 }, { "epoch": 2.68, "grad_norm": 12.608503341674805, "learning_rate": 5.921581991483094e-07, "loss": 0.4919, "step": 16646 }, { "epoch": 2.68, "grad_norm": 24.48321533203125, "learning_rate": 5.915691516888256e-07, "loss": 0.6071, "step": 16647 }, { "epoch": 2.68, "grad_norm": 13.235633850097656, "learning_rate": 5.909803884216236e-07, "loss": 0.6988, "step": 16648 }, { "epoch": 2.68, "grad_norm": 16.46625328063965, "learning_rate": 5.90391909364485e-07, "loss": 0.6727, "step": 16649 }, { "epoch": 2.68, "grad_norm": 11.664665222167969, "learning_rate": 5.898037145351854e-07, "loss": 0.6617, "step": 16650 }, { "epoch": 2.68, "grad_norm": 12.542356491088867, "learning_rate": 5.892158039514939e-07, "loss": 0.6227, "step": 16651 }, { "epoch": 2.68, "grad_norm": 15.336048126220703, "learning_rate": 5.886281776311664e-07, "loss": 0.8291, "step": 16652 }, { "epoch": 2.68, "grad_norm": 12.913111686706543, "learning_rate": 5.880408355919554e-07, "loss": 0.8218, "step": 16653 }, { "epoch": 2.68, "grad_norm": 12.941222190856934, "learning_rate": 5.874537778516021e-07, "loss": 0.6633, "step": 16654 }, { "epoch": 2.68, "grad_norm": 19.505516052246094, "learning_rate": 5.868670044278357e-07, "loss": 0.9135, "step": 16655 }, { "epoch": 2.68, "grad_norm": 12.3898286819458, "learning_rate": 5.862805153383855e-07, "loss": 0.5102, "step": 16656 }, { "epoch": 2.68, "grad_norm": 10.724065780639648, "learning_rate": 5.856943106009638e-07, "loss": 0.6176, "step": 16657 }, { "epoch": 2.68, "grad_norm": 11.406146049499512, "learning_rate": 5.851083902332765e-07, "loss": 0.6243, "step": 16658 }, { "epoch": 2.68, "grad_norm": 15.245439529418945, "learning_rate": 5.845227542530252e-07, "loss": 0.7183, "step": 16659 }, { "epoch": 2.68, "grad_norm": 12.850418090820312, "learning_rate": 5.839374026778977e-07, "loss": 1.0863, "step": 16660 }, { "epoch": 2.68, "grad_norm": 16.13894271850586, "learning_rate": 5.833523355255755e-07, "loss": 0.7158, "step": 16661 }, { "epoch": 2.68, "grad_norm": 15.224364280700684, "learning_rate": 5.827675528137323e-07, "loss": 0.7758, "step": 16662 }, { "epoch": 2.68, "grad_norm": 13.104381561279297, "learning_rate": 5.821830545600304e-07, "loss": 0.657, "step": 16663 }, { "epoch": 2.68, "grad_norm": 12.388555526733398, "learning_rate": 5.815988407821271e-07, "loss": 0.8798, "step": 16664 }, { "epoch": 2.68, "grad_norm": 13.36024284362793, "learning_rate": 5.810149114976682e-07, "loss": 0.8879, "step": 16665 }, { "epoch": 2.68, "grad_norm": 14.367112159729004, "learning_rate": 5.804312667242906e-07, "loss": 0.7378, "step": 16666 }, { "epoch": 2.68, "grad_norm": 12.70202922821045, "learning_rate": 5.79847906479627e-07, "loss": 0.6787, "step": 16667 }, { "epoch": 2.68, "grad_norm": 11.700053215026855, "learning_rate": 5.792648307812964e-07, "loss": 0.5603, "step": 16668 }, { "epoch": 2.68, "grad_norm": 14.490561485290527, "learning_rate": 5.786820396469106e-07, "loss": 0.7877, "step": 16669 }, { "epoch": 2.68, "grad_norm": 14.461421966552734, "learning_rate": 5.780995330940764e-07, "loss": 0.7085, "step": 16670 }, { "epoch": 2.68, "grad_norm": 11.144890785217285, "learning_rate": 5.775173111403853e-07, "loss": 0.6434, "step": 16671 }, { "epoch": 2.68, "grad_norm": 14.575858116149902, "learning_rate": 5.769353738034267e-07, "loss": 0.8632, "step": 16672 }, { "epoch": 2.68, "grad_norm": 12.95238208770752, "learning_rate": 5.763537211007785e-07, "loss": 0.6736, "step": 16673 }, { "epoch": 2.68, "grad_norm": 12.25656795501709, "learning_rate": 5.757723530500081e-07, "loss": 0.6884, "step": 16674 }, { "epoch": 2.68, "grad_norm": 14.924490928649902, "learning_rate": 5.75191269668679e-07, "loss": 0.6576, "step": 16675 }, { "epoch": 2.68, "grad_norm": 11.67580509185791, "learning_rate": 5.746104709743416e-07, "loss": 0.6051, "step": 16676 }, { "epoch": 2.68, "grad_norm": 14.649404525756836, "learning_rate": 5.740299569845409e-07, "loss": 0.7833, "step": 16677 }, { "epoch": 2.68, "grad_norm": 15.086240768432617, "learning_rate": 5.734497277168116e-07, "loss": 0.7662, "step": 16678 }, { "epoch": 2.68, "grad_norm": 14.27561092376709, "learning_rate": 5.728697831886776e-07, "loss": 0.6559, "step": 16679 }, { "epoch": 2.69, "grad_norm": 13.300237655639648, "learning_rate": 5.722901234176614e-07, "loss": 0.6376, "step": 16680 }, { "epoch": 2.69, "grad_norm": 13.760710716247559, "learning_rate": 5.71710748421268e-07, "loss": 0.8968, "step": 16681 }, { "epoch": 2.69, "grad_norm": 13.216819763183594, "learning_rate": 5.711316582170013e-07, "loss": 0.7861, "step": 16682 }, { "epoch": 2.69, "grad_norm": 14.569887161254883, "learning_rate": 5.705528528223525e-07, "loss": 0.6443, "step": 16683 }, { "epoch": 2.69, "grad_norm": 16.287891387939453, "learning_rate": 5.699743322548046e-07, "loss": 0.618, "step": 16684 }, { "epoch": 2.69, "grad_norm": 13.431755065917969, "learning_rate": 5.693960965318313e-07, "loss": 0.9012, "step": 16685 }, { "epoch": 2.69, "grad_norm": 12.833324432373047, "learning_rate": 5.688181456709019e-07, "loss": 0.9539, "step": 16686 }, { "epoch": 2.69, "grad_norm": 11.926124572753906, "learning_rate": 5.682404796894703e-07, "loss": 0.5696, "step": 16687 }, { "epoch": 2.69, "grad_norm": 11.511601448059082, "learning_rate": 5.676630986049891e-07, "loss": 0.6918, "step": 16688 }, { "epoch": 2.69, "grad_norm": 17.232852935791016, "learning_rate": 5.670860024348968e-07, "loss": 0.6195, "step": 16689 }, { "epoch": 2.69, "grad_norm": 14.827371597290039, "learning_rate": 5.665091911966247e-07, "loss": 0.5414, "step": 16690 }, { "epoch": 2.69, "grad_norm": 14.739311218261719, "learning_rate": 5.659326649075992e-07, "loss": 0.9738, "step": 16691 }, { "epoch": 2.69, "grad_norm": 10.93570327758789, "learning_rate": 5.653564235852316e-07, "loss": 0.6533, "step": 16692 }, { "epoch": 2.69, "grad_norm": 13.278923988342285, "learning_rate": 5.64780467246927e-07, "loss": 0.5915, "step": 16693 }, { "epoch": 2.69, "grad_norm": 12.546268463134766, "learning_rate": 5.642047959100871e-07, "loss": 0.5813, "step": 16694 }, { "epoch": 2.69, "grad_norm": 15.337167739868164, "learning_rate": 5.63629409592098e-07, "loss": 0.9105, "step": 16695 }, { "epoch": 2.69, "grad_norm": 15.513779640197754, "learning_rate": 5.630543083103402e-07, "loss": 0.8399, "step": 16696 }, { "epoch": 2.69, "grad_norm": 10.88771915435791, "learning_rate": 5.624794920821863e-07, "loss": 0.6826, "step": 16697 }, { "epoch": 2.69, "grad_norm": 12.31085205078125, "learning_rate": 5.61904960924996e-07, "loss": 0.6963, "step": 16698 }, { "epoch": 2.69, "grad_norm": 12.005014419555664, "learning_rate": 5.613307148561276e-07, "loss": 0.9348, "step": 16699 }, { "epoch": 2.69, "grad_norm": 12.122157096862793, "learning_rate": 5.607567538929248e-07, "loss": 0.7034, "step": 16700 }, { "epoch": 2.69, "grad_norm": 12.695786476135254, "learning_rate": 5.60183078052724e-07, "loss": 0.6874, "step": 16701 }, { "epoch": 2.69, "grad_norm": 11.171300888061523, "learning_rate": 5.596096873528578e-07, "loss": 0.6684, "step": 16702 }, { "epoch": 2.69, "grad_norm": 12.710497856140137, "learning_rate": 5.590365818106425e-07, "loss": 0.7366, "step": 16703 }, { "epoch": 2.69, "grad_norm": 11.465296745300293, "learning_rate": 5.584637614433886e-07, "loss": 0.671, "step": 16704 }, { "epoch": 2.69, "grad_norm": 12.291322708129883, "learning_rate": 5.57891226268401e-07, "loss": 0.6854, "step": 16705 }, { "epoch": 2.69, "grad_norm": 12.72737979888916, "learning_rate": 5.573189763029729e-07, "loss": 0.8367, "step": 16706 }, { "epoch": 2.69, "grad_norm": 12.048196792602539, "learning_rate": 5.567470115643914e-07, "loss": 0.732, "step": 16707 }, { "epoch": 2.69, "grad_norm": 13.54349136352539, "learning_rate": 5.561753320699314e-07, "loss": 0.6691, "step": 16708 }, { "epoch": 2.69, "grad_norm": 11.727082252502441, "learning_rate": 5.556039378368605e-07, "loss": 0.4733, "step": 16709 }, { "epoch": 2.69, "grad_norm": 14.19430160522461, "learning_rate": 5.550328288824403e-07, "loss": 0.4785, "step": 16710 }, { "epoch": 2.69, "grad_norm": 15.999554634094238, "learning_rate": 5.544620052239202e-07, "loss": 0.6652, "step": 16711 }, { "epoch": 2.69, "grad_norm": 11.987587928771973, "learning_rate": 5.538914668785444e-07, "loss": 0.6801, "step": 16712 }, { "epoch": 2.69, "grad_norm": 15.233930587768555, "learning_rate": 5.533212138635458e-07, "loss": 0.6542, "step": 16713 }, { "epoch": 2.69, "grad_norm": 12.4055814743042, "learning_rate": 5.527512461961482e-07, "loss": 0.8278, "step": 16714 }, { "epoch": 2.69, "grad_norm": 13.365851402282715, "learning_rate": 5.521815638935701e-07, "loss": 0.744, "step": 16715 }, { "epoch": 2.69, "grad_norm": 12.1037015914917, "learning_rate": 5.51612166973019e-07, "loss": 0.553, "step": 16716 }, { "epoch": 2.69, "grad_norm": 16.073055267333984, "learning_rate": 5.51043055451691e-07, "loss": 0.7105, "step": 16717 }, { "epoch": 2.69, "grad_norm": 15.573775291442871, "learning_rate": 5.504742293467825e-07, "loss": 0.7182, "step": 16718 }, { "epoch": 2.69, "grad_norm": 13.126832962036133, "learning_rate": 5.499056886754695e-07, "loss": 0.5789, "step": 16719 }, { "epoch": 2.69, "grad_norm": 13.602715492248535, "learning_rate": 5.493374334549306e-07, "loss": 0.7677, "step": 16720 }, { "epoch": 2.69, "grad_norm": 12.102395057678223, "learning_rate": 5.487694637023278e-07, "loss": 0.6701, "step": 16721 }, { "epoch": 2.69, "grad_norm": 12.958053588867188, "learning_rate": 5.482017794348172e-07, "loss": 0.5991, "step": 16722 }, { "epoch": 2.69, "grad_norm": 11.310650825500488, "learning_rate": 5.476343806695484e-07, "loss": 0.6787, "step": 16723 }, { "epoch": 2.69, "grad_norm": 14.314990043640137, "learning_rate": 5.470672674236587e-07, "loss": 0.7317, "step": 16724 }, { "epoch": 2.69, "grad_norm": 13.669279098510742, "learning_rate": 5.46500439714277e-07, "loss": 0.6296, "step": 16725 }, { "epoch": 2.69, "grad_norm": 15.576201438903809, "learning_rate": 5.459338975585282e-07, "loss": 0.595, "step": 16726 }, { "epoch": 2.69, "grad_norm": 12.824385643005371, "learning_rate": 5.453676409735242e-07, "loss": 0.7998, "step": 16727 }, { "epoch": 2.69, "grad_norm": 12.05334758758545, "learning_rate": 5.44801669976367e-07, "loss": 0.5518, "step": 16728 }, { "epoch": 2.69, "grad_norm": 13.538006782531738, "learning_rate": 5.442359845841561e-07, "loss": 0.7143, "step": 16729 }, { "epoch": 2.69, "grad_norm": 12.871546745300293, "learning_rate": 5.436705848139756e-07, "loss": 0.5177, "step": 16730 }, { "epoch": 2.69, "grad_norm": 13.73995590209961, "learning_rate": 5.431054706829064e-07, "loss": 0.72, "step": 16731 }, { "epoch": 2.69, "grad_norm": 12.08593463897705, "learning_rate": 5.42540642208017e-07, "loss": 0.6397, "step": 16732 }, { "epoch": 2.69, "grad_norm": 11.473496437072754, "learning_rate": 5.419760994063683e-07, "loss": 0.648, "step": 16733 }, { "epoch": 2.69, "grad_norm": 12.445462226867676, "learning_rate": 5.414118422950154e-07, "loss": 0.6605, "step": 16734 }, { "epoch": 2.69, "grad_norm": 15.713927268981934, "learning_rate": 5.408478708910003e-07, "loss": 0.7854, "step": 16735 }, { "epoch": 2.69, "grad_norm": 11.604799270629883, "learning_rate": 5.402841852113571e-07, "loss": 0.5996, "step": 16736 }, { "epoch": 2.69, "grad_norm": 13.073006629943848, "learning_rate": 5.397207852731156e-07, "loss": 0.8814, "step": 16737 }, { "epoch": 2.69, "grad_norm": 13.690985679626465, "learning_rate": 5.391576710932911e-07, "loss": 0.613, "step": 16738 }, { "epoch": 2.69, "grad_norm": 14.662281036376953, "learning_rate": 5.385948426888954e-07, "loss": 0.5666, "step": 16739 }, { "epoch": 2.69, "grad_norm": 12.758642196655273, "learning_rate": 5.380323000769294e-07, "loss": 0.7105, "step": 16740 }, { "epoch": 2.69, "grad_norm": 11.015885353088379, "learning_rate": 5.374700432743829e-07, "loss": 0.692, "step": 16741 }, { "epoch": 2.7, "grad_norm": 12.349203109741211, "learning_rate": 5.3690807229824e-07, "loss": 0.5853, "step": 16742 }, { "epoch": 2.7, "grad_norm": 14.819334983825684, "learning_rate": 5.363463871654795e-07, "loss": 0.6663, "step": 16743 }, { "epoch": 2.7, "grad_norm": 12.130471229553223, "learning_rate": 5.357849878930632e-07, "loss": 0.7499, "step": 16744 }, { "epoch": 2.7, "grad_norm": 13.215261459350586, "learning_rate": 5.352238744979509e-07, "loss": 0.5992, "step": 16745 }, { "epoch": 2.7, "grad_norm": 11.911916732788086, "learning_rate": 5.346630469970926e-07, "loss": 0.5902, "step": 16746 }, { "epoch": 2.7, "grad_norm": 14.65562915802002, "learning_rate": 5.341025054074245e-07, "loss": 0.7267, "step": 16747 }, { "epoch": 2.7, "grad_norm": 12.407029151916504, "learning_rate": 5.335422497458842e-07, "loss": 0.7179, "step": 16748 }, { "epoch": 2.7, "grad_norm": 10.922433853149414, "learning_rate": 5.329822800293893e-07, "loss": 0.6477, "step": 16749 }, { "epoch": 2.7, "grad_norm": 12.072164535522461, "learning_rate": 5.324225962748597e-07, "loss": 0.8729, "step": 16750 }, { "epoch": 2.7, "grad_norm": 13.607532501220703, "learning_rate": 5.318631984991973e-07, "loss": 0.5063, "step": 16751 }, { "epoch": 2.7, "grad_norm": 14.968043327331543, "learning_rate": 5.313040867192997e-07, "loss": 0.7063, "step": 16752 }, { "epoch": 2.7, "grad_norm": 12.53005599975586, "learning_rate": 5.307452609520581e-07, "loss": 0.7005, "step": 16753 }, { "epoch": 2.7, "grad_norm": 11.73477840423584, "learning_rate": 5.301867212143497e-07, "loss": 0.8047, "step": 16754 }, { "epoch": 2.7, "grad_norm": 14.094094276428223, "learning_rate": 5.296284675230478e-07, "loss": 0.6216, "step": 16755 }, { "epoch": 2.7, "grad_norm": 15.562019348144531, "learning_rate": 5.290704998950136e-07, "loss": 0.8338, "step": 16756 }, { "epoch": 2.7, "grad_norm": 12.105653762817383, "learning_rate": 5.285128183471011e-07, "loss": 0.8375, "step": 16757 }, { "epoch": 2.7, "grad_norm": 10.670614242553711, "learning_rate": 5.27955422896157e-07, "loss": 0.4618, "step": 16758 }, { "epoch": 2.7, "grad_norm": 12.382811546325684, "learning_rate": 5.273983135590178e-07, "loss": 0.8188, "step": 16759 }, { "epoch": 2.7, "grad_norm": 11.703875541687012, "learning_rate": 5.268414903525098e-07, "loss": 0.6023, "step": 16760 }, { "epoch": 2.7, "grad_norm": 11.992313385009766, "learning_rate": 5.262849532934555e-07, "loss": 0.6373, "step": 16761 }, { "epoch": 2.7, "grad_norm": 11.90916919708252, "learning_rate": 5.257287023986623e-07, "loss": 0.718, "step": 16762 }, { "epoch": 2.7, "grad_norm": 11.915592193603516, "learning_rate": 5.251727376849336e-07, "loss": 0.6921, "step": 16763 }, { "epoch": 2.7, "grad_norm": 15.242721557617188, "learning_rate": 5.246170591690658e-07, "loss": 0.6479, "step": 16764 }, { "epoch": 2.7, "grad_norm": 14.331765174865723, "learning_rate": 5.240616668678399e-07, "loss": 0.8407, "step": 16765 }, { "epoch": 2.7, "grad_norm": 12.290966033935547, "learning_rate": 5.23506560798035e-07, "loss": 0.6649, "step": 16766 }, { "epoch": 2.7, "grad_norm": 11.710064888000488, "learning_rate": 5.229517409764173e-07, "loss": 0.7075, "step": 16767 }, { "epoch": 2.7, "grad_norm": 15.375615119934082, "learning_rate": 5.223972074197459e-07, "loss": 0.6739, "step": 16768 }, { "epoch": 2.7, "grad_norm": 13.645621299743652, "learning_rate": 5.218429601447705e-07, "loss": 0.775, "step": 16769 }, { "epoch": 2.7, "grad_norm": 12.614700317382812, "learning_rate": 5.212889991682346e-07, "loss": 0.9486, "step": 16770 }, { "epoch": 2.7, "grad_norm": 11.764952659606934, "learning_rate": 5.207353245068691e-07, "loss": 0.6562, "step": 16771 }, { "epoch": 2.7, "grad_norm": 12.632116317749023, "learning_rate": 5.201819361773996e-07, "loss": 0.6123, "step": 16772 }, { "epoch": 2.7, "grad_norm": 15.715967178344727, "learning_rate": 5.196288341965405e-07, "loss": 0.6942, "step": 16773 }, { "epoch": 2.7, "grad_norm": 16.865293502807617, "learning_rate": 5.190760185810018e-07, "loss": 0.688, "step": 16774 }, { "epoch": 2.7, "grad_norm": 16.159366607666016, "learning_rate": 5.185234893474788e-07, "loss": 0.6183, "step": 16775 }, { "epoch": 2.7, "grad_norm": 11.387162208557129, "learning_rate": 5.179712465126619e-07, "loss": 0.5633, "step": 16776 }, { "epoch": 2.7, "grad_norm": 13.436744689941406, "learning_rate": 5.17419290093234e-07, "loss": 0.6601, "step": 16777 }, { "epoch": 2.7, "grad_norm": 9.828197479248047, "learning_rate": 5.168676201058653e-07, "loss": 0.5694, "step": 16778 }, { "epoch": 2.7, "grad_norm": 10.201630592346191, "learning_rate": 5.163162365672226e-07, "loss": 0.7366, "step": 16779 }, { "epoch": 2.7, "grad_norm": 9.427828788757324, "learning_rate": 5.157651394939578e-07, "loss": 0.535, "step": 16780 }, { "epoch": 2.7, "grad_norm": 13.091231346130371, "learning_rate": 5.152143289027178e-07, "loss": 1.0307, "step": 16781 }, { "epoch": 2.7, "grad_norm": 10.217658996582031, "learning_rate": 5.146638048101416e-07, "loss": 0.5752, "step": 16782 }, { "epoch": 2.7, "grad_norm": 12.976922988891602, "learning_rate": 5.14113567232859e-07, "loss": 0.6884, "step": 16783 }, { "epoch": 2.7, "grad_norm": 15.711454391479492, "learning_rate": 5.13563616187489e-07, "loss": 0.7176, "step": 16784 }, { "epoch": 2.7, "grad_norm": 11.446221351623535, "learning_rate": 5.130139516906463e-07, "loss": 0.7806, "step": 16785 }, { "epoch": 2.7, "grad_norm": 14.184653282165527, "learning_rate": 5.124645737589307e-07, "loss": 0.592, "step": 16786 }, { "epoch": 2.7, "grad_norm": 12.721135139465332, "learning_rate": 5.119154824089378e-07, "loss": 0.5888, "step": 16787 }, { "epoch": 2.7, "grad_norm": 11.194150924682617, "learning_rate": 5.113666776572546e-07, "loss": 0.5578, "step": 16788 }, { "epoch": 2.7, "grad_norm": 11.622297286987305, "learning_rate": 5.108181595204564e-07, "loss": 0.7043, "step": 16789 }, { "epoch": 2.7, "grad_norm": 14.563100814819336, "learning_rate": 5.102699280151146e-07, "loss": 0.6822, "step": 16790 }, { "epoch": 2.7, "grad_norm": 14.946285247802734, "learning_rate": 5.097219831577882e-07, "loss": 0.7327, "step": 16791 }, { "epoch": 2.7, "grad_norm": 13.620695114135742, "learning_rate": 5.091743249650249e-07, "loss": 0.5744, "step": 16792 }, { "epoch": 2.7, "grad_norm": 11.239108085632324, "learning_rate": 5.086269534533727e-07, "loss": 0.918, "step": 16793 }, { "epoch": 2.7, "grad_norm": 11.931276321411133, "learning_rate": 5.08079868639364e-07, "loss": 0.7118, "step": 16794 }, { "epoch": 2.7, "grad_norm": 12.80199146270752, "learning_rate": 5.075330705395209e-07, "loss": 0.87, "step": 16795 }, { "epoch": 2.7, "grad_norm": 11.257622718811035, "learning_rate": 5.069865591703649e-07, "loss": 0.8051, "step": 16796 }, { "epoch": 2.7, "grad_norm": 14.615680694580078, "learning_rate": 5.064403345483992e-07, "loss": 0.7963, "step": 16797 }, { "epoch": 2.7, "grad_norm": 12.70396614074707, "learning_rate": 5.058943966901264e-07, "loss": 0.6644, "step": 16798 }, { "epoch": 2.7, "grad_norm": 14.811797142028809, "learning_rate": 5.053487456120376e-07, "loss": 0.6905, "step": 16799 }, { "epoch": 2.7, "grad_norm": 13.64098834991455, "learning_rate": 5.048033813306119e-07, "loss": 0.7853, "step": 16800 }, { "epoch": 2.7, "grad_norm": 13.553001403808594, "learning_rate": 5.04258303862325e-07, "loss": 0.4437, "step": 16801 }, { "epoch": 2.7, "grad_norm": 14.504003524780273, "learning_rate": 5.037135132236415e-07, "loss": 0.8549, "step": 16802 }, { "epoch": 2.7, "grad_norm": 12.975951194763184, "learning_rate": 5.031690094310126e-07, "loss": 0.6004, "step": 16803 }, { "epoch": 2.71, "grad_norm": 12.635008811950684, "learning_rate": 5.026247925008942e-07, "loss": 0.5427, "step": 16804 }, { "epoch": 2.71, "grad_norm": 10.881264686584473, "learning_rate": 5.020808624497197e-07, "loss": 0.62, "step": 16805 }, { "epoch": 2.71, "grad_norm": 11.977352142333984, "learning_rate": 5.015372192939183e-07, "loss": 0.527, "step": 16806 }, { "epoch": 2.71, "grad_norm": 14.219353675842285, "learning_rate": 5.009938630499156e-07, "loss": 0.7263, "step": 16807 }, { "epoch": 2.71, "grad_norm": 14.728202819824219, "learning_rate": 5.004507937341197e-07, "loss": 0.8223, "step": 16808 }, { "epoch": 2.71, "grad_norm": 14.257620811462402, "learning_rate": 4.999080113629373e-07, "loss": 0.751, "step": 16809 }, { "epoch": 2.71, "grad_norm": 11.902642250061035, "learning_rate": 4.993655159527644e-07, "loss": 0.7277, "step": 16810 }, { "epoch": 2.71, "grad_norm": 10.195540428161621, "learning_rate": 4.988233075199845e-07, "loss": 0.5841, "step": 16811 }, { "epoch": 2.71, "grad_norm": 12.517765998840332, "learning_rate": 4.982813860809788e-07, "loss": 0.891, "step": 16812 }, { "epoch": 2.71, "grad_norm": 10.587021827697754, "learning_rate": 4.977397516521154e-07, "loss": 0.6669, "step": 16813 }, { "epoch": 2.71, "grad_norm": 14.377518653869629, "learning_rate": 4.971984042497535e-07, "loss": 0.7081, "step": 16814 }, { "epoch": 2.71, "grad_norm": 11.42037296295166, "learning_rate": 4.966573438902478e-07, "loss": 0.7217, "step": 16815 }, { "epoch": 2.71, "grad_norm": 12.293549537658691, "learning_rate": 4.961165705899396e-07, "loss": 0.5215, "step": 16816 }, { "epoch": 2.71, "grad_norm": 17.806488037109375, "learning_rate": 4.955760843651647e-07, "loss": 0.5557, "step": 16817 }, { "epoch": 2.71, "grad_norm": 17.916629791259766, "learning_rate": 4.95035885232249e-07, "loss": 0.6261, "step": 16818 }, { "epoch": 2.71, "grad_norm": 14.01123046875, "learning_rate": 4.944959732075094e-07, "loss": 0.7585, "step": 16819 }, { "epoch": 2.71, "grad_norm": 11.83133316040039, "learning_rate": 4.939563483072552e-07, "loss": 0.7543, "step": 16820 }, { "epoch": 2.71, "grad_norm": 14.004530906677246, "learning_rate": 4.934170105477842e-07, "loss": 0.8093, "step": 16821 }, { "epoch": 2.71, "grad_norm": 12.389403343200684, "learning_rate": 4.928779599453914e-07, "loss": 0.4991, "step": 16822 }, { "epoch": 2.71, "grad_norm": 14.461899757385254, "learning_rate": 4.92339196516356e-07, "loss": 0.9458, "step": 16823 }, { "epoch": 2.71, "grad_norm": 14.130778312683105, "learning_rate": 4.918007202769549e-07, "loss": 0.6198, "step": 16824 }, { "epoch": 2.71, "grad_norm": 11.23616886138916, "learning_rate": 4.912625312434505e-07, "loss": 0.7271, "step": 16825 }, { "epoch": 2.71, "grad_norm": 17.171276092529297, "learning_rate": 4.907246294321022e-07, "loss": 0.7533, "step": 16826 }, { "epoch": 2.71, "grad_norm": 10.714205741882324, "learning_rate": 4.901870148591547e-07, "loss": 0.6511, "step": 16827 }, { "epoch": 2.71, "grad_norm": 13.032515525817871, "learning_rate": 4.896496875408507e-07, "loss": 0.6033, "step": 16828 }, { "epoch": 2.71, "grad_norm": 12.780158996582031, "learning_rate": 4.891126474934194e-07, "loss": 0.8303, "step": 16829 }, { "epoch": 2.71, "grad_norm": 14.392252922058105, "learning_rate": 4.88575894733081e-07, "loss": 0.7048, "step": 16830 }, { "epoch": 2.71, "grad_norm": 11.38508129119873, "learning_rate": 4.880394292760515e-07, "loss": 0.7327, "step": 16831 }, { "epoch": 2.71, "grad_norm": 16.014564514160156, "learning_rate": 4.875032511385325e-07, "loss": 0.7674, "step": 16832 }, { "epoch": 2.71, "grad_norm": 13.036696434020996, "learning_rate": 4.86967360336723e-07, "loss": 0.7927, "step": 16833 }, { "epoch": 2.71, "grad_norm": 14.330127716064453, "learning_rate": 4.864317568868083e-07, "loss": 0.9322, "step": 16834 }, { "epoch": 2.71, "grad_norm": 11.459784507751465, "learning_rate": 4.858964408049672e-07, "loss": 0.7669, "step": 16835 }, { "epoch": 2.71, "grad_norm": 13.977653503417969, "learning_rate": 4.853614121073701e-07, "loss": 0.5959, "step": 16836 }, { "epoch": 2.71, "grad_norm": 11.388348579406738, "learning_rate": 4.848266708101779e-07, "loss": 0.8477, "step": 16837 }, { "epoch": 2.71, "grad_norm": 14.269319534301758, "learning_rate": 4.842922169295405e-07, "loss": 0.6626, "step": 16838 }, { "epoch": 2.71, "grad_norm": 15.014448165893555, "learning_rate": 4.837580504816064e-07, "loss": 0.6837, "step": 16839 }, { "epoch": 2.71, "grad_norm": 12.961579322814941, "learning_rate": 4.83224171482507e-07, "loss": 0.8151, "step": 16840 }, { "epoch": 2.71, "grad_norm": 11.845412254333496, "learning_rate": 4.826905799483716e-07, "loss": 0.6154, "step": 16841 }, { "epoch": 2.71, "grad_norm": 16.71058464050293, "learning_rate": 4.821572758953164e-07, "loss": 0.7324, "step": 16842 }, { "epoch": 2.71, "grad_norm": 11.887480735778809, "learning_rate": 4.816242593394472e-07, "loss": 0.7053, "step": 16843 }, { "epoch": 2.71, "grad_norm": 14.230380058288574, "learning_rate": 4.810915302968699e-07, "loss": 0.7531, "step": 16844 }, { "epoch": 2.71, "grad_norm": 16.610422134399414, "learning_rate": 4.805590887836742e-07, "loss": 0.7914, "step": 16845 }, { "epoch": 2.71, "grad_norm": 16.54098129272461, "learning_rate": 4.800269348159415e-07, "loss": 0.6502, "step": 16846 }, { "epoch": 2.71, "grad_norm": 13.97498893737793, "learning_rate": 4.794950684097488e-07, "loss": 0.7129, "step": 16847 }, { "epoch": 2.71, "grad_norm": 12.327433586120605, "learning_rate": 4.789634895811613e-07, "loss": 0.7503, "step": 16848 }, { "epoch": 2.71, "grad_norm": 15.4524564743042, "learning_rate": 4.784321983462325e-07, "loss": 0.5878, "step": 16849 }, { "epoch": 2.71, "grad_norm": 11.696182250976562, "learning_rate": 4.779011947210144e-07, "loss": 0.5976, "step": 16850 }, { "epoch": 2.71, "grad_norm": 15.501358985900879, "learning_rate": 4.773704787215439e-07, "loss": 0.7761, "step": 16851 }, { "epoch": 2.71, "grad_norm": 13.288084030151367, "learning_rate": 4.768400503638548e-07, "loss": 0.6052, "step": 16852 }, { "epoch": 2.71, "grad_norm": 19.081937789916992, "learning_rate": 4.7630990966396784e-07, "loss": 0.9789, "step": 16853 }, { "epoch": 2.71, "grad_norm": 14.01154899597168, "learning_rate": 4.757800566378945e-07, "loss": 0.6205, "step": 16854 }, { "epoch": 2.71, "grad_norm": 12.259657859802246, "learning_rate": 4.7525049130164315e-07, "loss": 0.5498, "step": 16855 }, { "epoch": 2.71, "grad_norm": 12.804495811462402, "learning_rate": 4.747212136712076e-07, "loss": 0.7816, "step": 16856 }, { "epoch": 2.71, "grad_norm": 13.0283203125, "learning_rate": 4.7419222376257625e-07, "loss": 0.8942, "step": 16857 }, { "epoch": 2.71, "grad_norm": 11.648752212524414, "learning_rate": 4.7366352159172846e-07, "loss": 0.6309, "step": 16858 }, { "epoch": 2.71, "grad_norm": 10.973705291748047, "learning_rate": 4.731351071746315e-07, "loss": 0.6161, "step": 16859 }, { "epoch": 2.71, "grad_norm": 11.540661811828613, "learning_rate": 4.726069805272493e-07, "loss": 0.7584, "step": 16860 }, { "epoch": 2.71, "grad_norm": 13.192880630493164, "learning_rate": 4.720791416655346e-07, "loss": 0.6123, "step": 16861 }, { "epoch": 2.71, "grad_norm": 15.47452449798584, "learning_rate": 4.715515906054291e-07, "loss": 0.7386, "step": 16862 }, { "epoch": 2.71, "grad_norm": 12.476178169250488, "learning_rate": 4.71024327362869e-07, "loss": 0.649, "step": 16863 }, { "epoch": 2.71, "grad_norm": 13.18753433227539, "learning_rate": 4.7049735195378367e-07, "loss": 0.807, "step": 16864 }, { "epoch": 2.71, "grad_norm": 10.708168983459473, "learning_rate": 4.699706643940871e-07, "loss": 0.5962, "step": 16865 }, { "epoch": 2.72, "grad_norm": 13.716076850891113, "learning_rate": 4.6944426469969105e-07, "loss": 0.7046, "step": 16866 }, { "epoch": 2.72, "grad_norm": 11.92705249786377, "learning_rate": 4.689181528864939e-07, "loss": 0.8312, "step": 16867 }, { "epoch": 2.72, "grad_norm": 12.672679901123047, "learning_rate": 4.6839232897039065e-07, "loss": 0.5195, "step": 16868 }, { "epoch": 2.72, "grad_norm": 11.639534950256348, "learning_rate": 4.678667929672609e-07, "loss": 0.6496, "step": 16869 }, { "epoch": 2.72, "grad_norm": 12.71862506866455, "learning_rate": 4.6734154489298077e-07, "loss": 0.5898, "step": 16870 }, { "epoch": 2.72, "grad_norm": 11.739806175231934, "learning_rate": 4.668165847634165e-07, "loss": 0.736, "step": 16871 }, { "epoch": 2.72, "grad_norm": 13.134547233581543, "learning_rate": 4.662919125944243e-07, "loss": 0.5985, "step": 16872 }, { "epoch": 2.72, "grad_norm": 11.79456615447998, "learning_rate": 4.657675284018504e-07, "loss": 0.7409, "step": 16873 }, { "epoch": 2.72, "grad_norm": 11.131449699401855, "learning_rate": 4.652434322015387e-07, "loss": 0.5563, "step": 16874 }, { "epoch": 2.72, "grad_norm": 14.801873207092285, "learning_rate": 4.647196240093166e-07, "loss": 0.9353, "step": 16875 }, { "epoch": 2.72, "grad_norm": 13.647656440734863, "learning_rate": 4.641961038410092e-07, "loss": 0.7451, "step": 16876 }, { "epoch": 2.72, "grad_norm": 14.988424301147461, "learning_rate": 4.6367287171242835e-07, "loss": 0.662, "step": 16877 }, { "epoch": 2.72, "grad_norm": 13.578213691711426, "learning_rate": 4.6314992763937807e-07, "loss": 0.7817, "step": 16878 }, { "epoch": 2.72, "grad_norm": 12.331387519836426, "learning_rate": 4.6262727163765673e-07, "loss": 0.6144, "step": 16879 }, { "epoch": 2.72, "grad_norm": 14.257325172424316, "learning_rate": 4.621049037230496e-07, "loss": 0.6827, "step": 16880 }, { "epoch": 2.72, "grad_norm": 12.998882293701172, "learning_rate": 4.6158282391133514e-07, "loss": 0.4802, "step": 16881 }, { "epoch": 2.72, "grad_norm": 16.163480758666992, "learning_rate": 4.610610322182862e-07, "loss": 0.6757, "step": 16882 }, { "epoch": 2.72, "grad_norm": 12.821555137634277, "learning_rate": 4.605395286596614e-07, "loss": 0.6494, "step": 16883 }, { "epoch": 2.72, "grad_norm": 12.962103843688965, "learning_rate": 4.600183132512137e-07, "loss": 0.8002, "step": 16884 }, { "epoch": 2.72, "grad_norm": 13.601217269897461, "learning_rate": 4.5949738600868934e-07, "loss": 0.6186, "step": 16885 }, { "epoch": 2.72, "grad_norm": 13.240089416503906, "learning_rate": 4.5897674694782013e-07, "loss": 0.5112, "step": 16886 }, { "epoch": 2.72, "grad_norm": 11.933688163757324, "learning_rate": 4.5845639608433467e-07, "loss": 0.722, "step": 16887 }, { "epoch": 2.72, "grad_norm": 13.491591453552246, "learning_rate": 4.579363334339515e-07, "loss": 0.7484, "step": 16888 }, { "epoch": 2.72, "grad_norm": 11.893794059753418, "learning_rate": 4.5741655901237694e-07, "loss": 0.6966, "step": 16889 }, { "epoch": 2.72, "grad_norm": 13.879430770874023, "learning_rate": 4.568970728353139e-07, "loss": 0.709, "step": 16890 }, { "epoch": 2.72, "grad_norm": 12.338702201843262, "learning_rate": 4.563778749184533e-07, "loss": 0.5125, "step": 16891 }, { "epoch": 2.72, "grad_norm": 12.143988609313965, "learning_rate": 4.55858965277477e-07, "loss": 0.5747, "step": 16892 }, { "epoch": 2.72, "grad_norm": 12.9732084274292, "learning_rate": 4.5534034392806015e-07, "loss": 0.7253, "step": 16893 }, { "epoch": 2.72, "grad_norm": 12.242629051208496, "learning_rate": 4.5482201088586806e-07, "loss": 0.732, "step": 16894 }, { "epoch": 2.72, "grad_norm": 10.721735954284668, "learning_rate": 4.5430396616655825e-07, "loss": 0.6223, "step": 16895 }, { "epoch": 2.72, "grad_norm": 12.419071197509766, "learning_rate": 4.5378620978577923e-07, "loss": 0.8209, "step": 16896 }, { "epoch": 2.72, "grad_norm": 14.7969331741333, "learning_rate": 4.532687417591675e-07, "loss": 0.7266, "step": 16897 }, { "epoch": 2.72, "grad_norm": 12.965803146362305, "learning_rate": 4.52751562102357e-07, "loss": 0.6308, "step": 16898 }, { "epoch": 2.72, "grad_norm": 17.275083541870117, "learning_rate": 4.522346708309677e-07, "loss": 0.84, "step": 16899 }, { "epoch": 2.72, "grad_norm": 12.579776763916016, "learning_rate": 4.5171806796061477e-07, "loss": 0.618, "step": 16900 }, { "epoch": 2.72, "grad_norm": 12.864587783813477, "learning_rate": 4.5120175350690244e-07, "loss": 0.7645, "step": 16901 }, { "epoch": 2.72, "grad_norm": 14.218485832214355, "learning_rate": 4.506857274854237e-07, "loss": 0.8389, "step": 16902 }, { "epoch": 2.72, "grad_norm": 12.666221618652344, "learning_rate": 4.5016998991176953e-07, "loss": 0.9103, "step": 16903 }, { "epoch": 2.72, "grad_norm": 10.615608215332031, "learning_rate": 4.4965454080151515e-07, "loss": 0.5029, "step": 16904 }, { "epoch": 2.72, "grad_norm": 15.478677749633789, "learning_rate": 4.4913938017023153e-07, "loss": 0.7488, "step": 16905 }, { "epoch": 2.72, "grad_norm": 11.659276008605957, "learning_rate": 4.486245080334817e-07, "loss": 0.5923, "step": 16906 }, { "epoch": 2.72, "grad_norm": 11.53505802154541, "learning_rate": 4.481099244068166e-07, "loss": 0.6643, "step": 16907 }, { "epoch": 2.72, "grad_norm": 13.448037147521973, "learning_rate": 4.47595629305777e-07, "loss": 0.761, "step": 16908 }, { "epoch": 2.72, "grad_norm": 12.696718215942383, "learning_rate": 4.4708162274590294e-07, "loss": 0.888, "step": 16909 }, { "epoch": 2.72, "grad_norm": 10.925153732299805, "learning_rate": 4.4656790474271515e-07, "loss": 0.6736, "step": 16910 }, { "epoch": 2.72, "grad_norm": 13.393399238586426, "learning_rate": 4.4605447531173574e-07, "loss": 0.6503, "step": 16911 }, { "epoch": 2.72, "grad_norm": 11.6984224319458, "learning_rate": 4.4554133446847113e-07, "loss": 0.6244, "step": 16912 }, { "epoch": 2.72, "grad_norm": 13.435236930847168, "learning_rate": 4.4502848222842124e-07, "loss": 0.5988, "step": 16913 }, { "epoch": 2.72, "grad_norm": 11.691847801208496, "learning_rate": 4.4451591860707796e-07, "loss": 0.6841, "step": 16914 }, { "epoch": 2.72, "grad_norm": 13.659193992614746, "learning_rate": 4.4400364361992355e-07, "loss": 0.5549, "step": 16915 }, { "epoch": 2.72, "grad_norm": 11.843528747558594, "learning_rate": 4.4349165728243105e-07, "loss": 0.4751, "step": 16916 }, { "epoch": 2.72, "grad_norm": 12.97943115234375, "learning_rate": 4.4297995961006815e-07, "loss": 0.6352, "step": 16917 }, { "epoch": 2.72, "grad_norm": 12.629823684692383, "learning_rate": 4.42468550618288e-07, "loss": 0.7919, "step": 16918 }, { "epoch": 2.72, "grad_norm": 13.809969902038574, "learning_rate": 4.419574303225405e-07, "loss": 0.8122, "step": 16919 }, { "epoch": 2.72, "grad_norm": 16.311126708984375, "learning_rate": 4.4144659873826435e-07, "loss": 0.6816, "step": 16920 }, { "epoch": 2.72, "grad_norm": 12.839885711669922, "learning_rate": 4.409360558808884e-07, "loss": 0.6873, "step": 16921 }, { "epoch": 2.72, "grad_norm": 11.910829544067383, "learning_rate": 4.4042580176583583e-07, "loss": 0.7872, "step": 16922 }, { "epoch": 2.72, "grad_norm": 13.980581283569336, "learning_rate": 4.3991583640851877e-07, "loss": 0.6413, "step": 16923 }, { "epoch": 2.72, "grad_norm": 12.61154556274414, "learning_rate": 4.3940615982434044e-07, "loss": 0.8242, "step": 16924 }, { "epoch": 2.72, "grad_norm": 11.941628456115723, "learning_rate": 4.388967720286985e-07, "loss": 0.8474, "step": 16925 }, { "epoch": 2.72, "grad_norm": 11.490296363830566, "learning_rate": 4.3838767303697736e-07, "loss": 0.6842, "step": 16926 }, { "epoch": 2.72, "grad_norm": 12.201648712158203, "learning_rate": 4.3787886286455584e-07, "loss": 0.576, "step": 16927 }, { "epoch": 2.73, "grad_norm": 12.956887245178223, "learning_rate": 4.373703415268027e-07, "loss": 0.6832, "step": 16928 }, { "epoch": 2.73, "grad_norm": 13.033747673034668, "learning_rate": 4.36862109039079e-07, "loss": 0.609, "step": 16929 }, { "epoch": 2.73, "grad_norm": 12.937912940979004, "learning_rate": 4.363541654167369e-07, "loss": 0.8102, "step": 16930 }, { "epoch": 2.73, "grad_norm": 10.693394660949707, "learning_rate": 4.3584651067511863e-07, "loss": 0.7392, "step": 16931 }, { "epoch": 2.73, "grad_norm": 12.681771278381348, "learning_rate": 4.3533914482955744e-07, "loss": 0.6006, "step": 16932 }, { "epoch": 2.73, "grad_norm": 13.481291770935059, "learning_rate": 4.34832067895381e-07, "loss": 0.6804, "step": 16933 }, { "epoch": 2.73, "grad_norm": 12.898534774780273, "learning_rate": 4.3432527988790385e-07, "loss": 0.6533, "step": 16934 }, { "epoch": 2.73, "grad_norm": 14.088793754577637, "learning_rate": 4.33818780822437e-07, "loss": 0.6972, "step": 16935 }, { "epoch": 2.73, "grad_norm": 15.209991455078125, "learning_rate": 4.3331257071427823e-07, "loss": 0.7479, "step": 16936 }, { "epoch": 2.73, "grad_norm": 13.707514762878418, "learning_rate": 4.3280664957871646e-07, "loss": 0.636, "step": 16937 }, { "epoch": 2.73, "grad_norm": 12.548197746276855, "learning_rate": 4.3230101743103713e-07, "loss": 0.6953, "step": 16938 }, { "epoch": 2.73, "grad_norm": 12.331273078918457, "learning_rate": 4.3179567428651145e-07, "loss": 0.6872, "step": 16939 }, { "epoch": 2.73, "grad_norm": 14.047616004943848, "learning_rate": 4.312906201604028e-07, "loss": 0.628, "step": 16940 }, { "epoch": 2.73, "grad_norm": 15.079174995422363, "learning_rate": 4.307858550679689e-07, "loss": 0.8649, "step": 16941 }, { "epoch": 2.73, "grad_norm": 10.634965896606445, "learning_rate": 4.302813790244553e-07, "loss": 0.7279, "step": 16942 }, { "epoch": 2.73, "grad_norm": 13.596089363098145, "learning_rate": 4.2977719204510213e-07, "loss": 0.5773, "step": 16943 }, { "epoch": 2.73, "grad_norm": 13.66063117980957, "learning_rate": 4.292732941451361e-07, "loss": 0.5329, "step": 16944 }, { "epoch": 2.73, "grad_norm": 10.372331619262695, "learning_rate": 4.2876968533978045e-07, "loss": 0.5719, "step": 16945 }, { "epoch": 2.73, "grad_norm": 11.92078685760498, "learning_rate": 4.282663656442465e-07, "loss": 0.8029, "step": 16946 }, { "epoch": 2.73, "grad_norm": 10.416719436645508, "learning_rate": 4.277633350737387e-07, "loss": 0.7185, "step": 16947 }, { "epoch": 2.73, "grad_norm": 14.323592185974121, "learning_rate": 4.272605936434493e-07, "loss": 0.7062, "step": 16948 }, { "epoch": 2.73, "grad_norm": 13.75416374206543, "learning_rate": 4.267581413685662e-07, "loss": 0.6433, "step": 16949 }, { "epoch": 2.73, "grad_norm": 11.095015525817871, "learning_rate": 4.262559782642661e-07, "loss": 0.785, "step": 16950 }, { "epoch": 2.73, "grad_norm": 11.806336402893066, "learning_rate": 4.2575410434571583e-07, "loss": 0.4906, "step": 16951 }, { "epoch": 2.73, "grad_norm": 12.654584884643555, "learning_rate": 4.252525196280777e-07, "loss": 0.8257, "step": 16952 }, { "epoch": 2.73, "grad_norm": 11.055566787719727, "learning_rate": 4.247512241264995e-07, "loss": 0.6881, "step": 16953 }, { "epoch": 2.73, "grad_norm": 12.314323425292969, "learning_rate": 4.2425021785612584e-07, "loss": 0.6993, "step": 16954 }, { "epoch": 2.73, "grad_norm": 12.929557800292969, "learning_rate": 4.2374950083209023e-07, "loss": 0.8443, "step": 16955 }, { "epoch": 2.73, "grad_norm": 12.551236152648926, "learning_rate": 4.23249073069516e-07, "loss": 0.8041, "step": 16956 }, { "epoch": 2.73, "grad_norm": 14.659672737121582, "learning_rate": 4.2274893458352004e-07, "loss": 0.6015, "step": 16957 }, { "epoch": 2.73, "grad_norm": 13.687840461730957, "learning_rate": 4.222490853892103e-07, "loss": 0.6986, "step": 16958 }, { "epoch": 2.73, "grad_norm": 13.67963981628418, "learning_rate": 4.2174952550168237e-07, "loss": 0.6428, "step": 16959 }, { "epoch": 2.73, "grad_norm": 12.910179138183594, "learning_rate": 4.2125025493602865e-07, "loss": 0.9292, "step": 16960 }, { "epoch": 2.73, "grad_norm": 15.352704048156738, "learning_rate": 4.207512737073294e-07, "loss": 0.771, "step": 16961 }, { "epoch": 2.73, "grad_norm": 11.914045333862305, "learning_rate": 4.2025258183065796e-07, "loss": 0.6707, "step": 16962 }, { "epoch": 2.73, "grad_norm": 13.925125122070312, "learning_rate": 4.197541793210769e-07, "loss": 0.7304, "step": 16963 }, { "epoch": 2.73, "grad_norm": 13.878433227539062, "learning_rate": 4.1925606619363955e-07, "loss": 0.8659, "step": 16964 }, { "epoch": 2.73, "grad_norm": 12.5641508102417, "learning_rate": 4.1875824246339404e-07, "loss": 0.7009, "step": 16965 }, { "epoch": 2.73, "grad_norm": 12.609345436096191, "learning_rate": 4.1826070814537824e-07, "loss": 0.7925, "step": 16966 }, { "epoch": 2.73, "grad_norm": 13.707077026367188, "learning_rate": 4.177634632546179e-07, "loss": 0.7908, "step": 16967 }, { "epoch": 2.73, "grad_norm": 13.96368408203125, "learning_rate": 4.172665078061366e-07, "loss": 0.9169, "step": 16968 }, { "epoch": 2.73, "grad_norm": 11.120444297790527, "learning_rate": 4.167698418149435e-07, "loss": 0.7851, "step": 16969 }, { "epoch": 2.73, "grad_norm": 12.65653133392334, "learning_rate": 4.1627346529603873e-07, "loss": 0.8823, "step": 16970 }, { "epoch": 2.73, "grad_norm": 13.230376243591309, "learning_rate": 4.1577737826442034e-07, "loss": 0.852, "step": 16971 }, { "epoch": 2.73, "grad_norm": 11.471854209899902, "learning_rate": 4.1528158073506854e-07, "loss": 0.6142, "step": 16972 }, { "epoch": 2.73, "grad_norm": 15.401537895202637, "learning_rate": 4.1478607272296466e-07, "loss": 0.7916, "step": 16973 }, { "epoch": 2.73, "grad_norm": 12.46120834350586, "learning_rate": 4.1429085424307237e-07, "loss": 0.4967, "step": 16974 }, { "epoch": 2.73, "grad_norm": 12.524118423461914, "learning_rate": 4.1379592531034964e-07, "loss": 0.6617, "step": 16975 }, { "epoch": 2.73, "grad_norm": 16.195476531982422, "learning_rate": 4.1330128593975005e-07, "loss": 0.626, "step": 16976 }, { "epoch": 2.73, "grad_norm": 12.016060829162598, "learning_rate": 4.128069361462106e-07, "loss": 0.6646, "step": 16977 }, { "epoch": 2.73, "grad_norm": 14.44582748413086, "learning_rate": 4.1231287594466706e-07, "loss": 0.5969, "step": 16978 }, { "epoch": 2.73, "grad_norm": 13.803455352783203, "learning_rate": 4.11819105350042e-07, "loss": 0.5502, "step": 16979 }, { "epoch": 2.73, "grad_norm": 13.153036117553711, "learning_rate": 4.1132562437724897e-07, "loss": 0.7785, "step": 16980 }, { "epoch": 2.73, "grad_norm": 13.80507755279541, "learning_rate": 4.1083243304119615e-07, "loss": 0.5891, "step": 16981 }, { "epoch": 2.73, "grad_norm": 12.674708366394043, "learning_rate": 4.103395313567804e-07, "loss": 0.804, "step": 16982 }, { "epoch": 2.73, "grad_norm": 13.375829696655273, "learning_rate": 4.0984691933888875e-07, "loss": 0.8522, "step": 16983 }, { "epoch": 2.73, "grad_norm": 13.833945274353027, "learning_rate": 4.0935459700240266e-07, "loss": 0.6143, "step": 16984 }, { "epoch": 2.73, "grad_norm": 10.140266418457031, "learning_rate": 4.088625643621924e-07, "loss": 0.9015, "step": 16985 }, { "epoch": 2.73, "grad_norm": 13.344747543334961, "learning_rate": 4.0837082143312057e-07, "loss": 0.6503, "step": 16986 }, { "epoch": 2.73, "grad_norm": 16.98493194580078, "learning_rate": 4.0787936823004304e-07, "loss": 0.7011, "step": 16987 }, { "epoch": 2.73, "grad_norm": 15.182029724121094, "learning_rate": 4.0738820476780127e-07, "loss": 0.5382, "step": 16988 }, { "epoch": 2.73, "grad_norm": 12.662104606628418, "learning_rate": 4.0689733106123453e-07, "loss": 0.8327, "step": 16989 }, { "epoch": 2.74, "grad_norm": 12.711507797241211, "learning_rate": 4.064067471251676e-07, "loss": 0.8124, "step": 16990 }, { "epoch": 2.74, "grad_norm": 16.570398330688477, "learning_rate": 4.0591645297441973e-07, "loss": 0.6851, "step": 16991 }, { "epoch": 2.74, "grad_norm": 13.960704803466797, "learning_rate": 4.054264486238024e-07, "loss": 0.5083, "step": 16992 }, { "epoch": 2.74, "grad_norm": 12.983905792236328, "learning_rate": 4.0493673408811497e-07, "loss": 0.4601, "step": 16993 }, { "epoch": 2.74, "grad_norm": 14.264571189880371, "learning_rate": 4.0444730938214996e-07, "loss": 0.766, "step": 16994 }, { "epoch": 2.74, "grad_norm": 12.475005149841309, "learning_rate": 4.039581745206922e-07, "loss": 0.7917, "step": 16995 }, { "epoch": 2.74, "grad_norm": 12.603415489196777, "learning_rate": 4.0346932951851545e-07, "loss": 0.6244, "step": 16996 }, { "epoch": 2.74, "grad_norm": 11.980860710144043, "learning_rate": 4.0298077439038687e-07, "loss": 0.8068, "step": 16997 }, { "epoch": 2.74, "grad_norm": 15.498637199401855, "learning_rate": 4.024925091510634e-07, "loss": 0.847, "step": 16998 }, { "epoch": 2.74, "grad_norm": 14.086267471313477, "learning_rate": 4.0200453381529225e-07, "loss": 0.7923, "step": 16999 }, { "epoch": 2.74, "grad_norm": 13.966263771057129, "learning_rate": 4.015168483978149e-07, "loss": 0.8024, "step": 17000 }, { "epoch": 2.74, "grad_norm": 15.718989372253418, "learning_rate": 4.010294529133607e-07, "loss": 0.7902, "step": 17001 }, { "epoch": 2.74, "grad_norm": 15.804038047790527, "learning_rate": 4.0054234737665565e-07, "loss": 0.8502, "step": 17002 }, { "epoch": 2.74, "grad_norm": 11.817569732666016, "learning_rate": 4.000555318024102e-07, "loss": 0.856, "step": 17003 }, { "epoch": 2.74, "grad_norm": 16.396888732910156, "learning_rate": 3.995690062053281e-07, "loss": 0.7872, "step": 17004 }, { "epoch": 2.74, "grad_norm": 15.532628059387207, "learning_rate": 3.990827706001088e-07, "loss": 0.6978, "step": 17005 }, { "epoch": 2.74, "grad_norm": 10.984856605529785, "learning_rate": 3.9859682500143827e-07, "loss": 0.7294, "step": 17006 }, { "epoch": 2.74, "grad_norm": 13.359439849853516, "learning_rate": 3.981111694239925e-07, "loss": 0.7671, "step": 17007 }, { "epoch": 2.74, "grad_norm": 13.991003036499023, "learning_rate": 3.976258038824465e-07, "loss": 0.6459, "step": 17008 }, { "epoch": 2.74, "grad_norm": 14.630630493164062, "learning_rate": 3.9714072839145746e-07, "loss": 0.5615, "step": 17009 }, { "epoch": 2.74, "grad_norm": 13.870077133178711, "learning_rate": 3.966559429656769e-07, "loss": 0.6625, "step": 17010 }, { "epoch": 2.74, "grad_norm": 15.464744567871094, "learning_rate": 3.96171447619752e-07, "loss": 0.6799, "step": 17011 }, { "epoch": 2.74, "grad_norm": 14.45838737487793, "learning_rate": 3.9568724236831334e-07, "loss": 0.7296, "step": 17012 }, { "epoch": 2.74, "grad_norm": 13.630173683166504, "learning_rate": 3.952033272259903e-07, "loss": 0.8166, "step": 17013 }, { "epoch": 2.74, "grad_norm": 15.807062149047852, "learning_rate": 3.94719702207399e-07, "loss": 0.8714, "step": 17014 }, { "epoch": 2.74, "grad_norm": 14.008188247680664, "learning_rate": 3.9423636732714654e-07, "loss": 0.7026, "step": 17015 }, { "epoch": 2.74, "grad_norm": 16.019804000854492, "learning_rate": 3.9375332259983466e-07, "loss": 0.6303, "step": 17016 }, { "epoch": 2.74, "grad_norm": 12.06755256652832, "learning_rate": 3.932705680400528e-07, "loss": 0.6703, "step": 17017 }, { "epoch": 2.74, "grad_norm": 10.922445297241211, "learning_rate": 3.9278810366238375e-07, "loss": 0.558, "step": 17018 }, { "epoch": 2.74, "grad_norm": 13.747437477111816, "learning_rate": 3.923059294814002e-07, "loss": 0.624, "step": 17019 }, { "epoch": 2.74, "grad_norm": 15.377154350280762, "learning_rate": 3.9182404551166733e-07, "loss": 0.8076, "step": 17020 }, { "epoch": 2.74, "grad_norm": 15.025723457336426, "learning_rate": 3.913424517677411e-07, "loss": 0.8322, "step": 17021 }, { "epoch": 2.74, "grad_norm": 10.779449462890625, "learning_rate": 3.9086114826416885e-07, "loss": 0.7475, "step": 17022 }, { "epoch": 2.74, "grad_norm": 13.423276901245117, "learning_rate": 3.903801350154879e-07, "loss": 0.7404, "step": 17023 }, { "epoch": 2.74, "grad_norm": 16.414384841918945, "learning_rate": 3.898994120362287e-07, "loss": 0.7845, "step": 17024 }, { "epoch": 2.74, "grad_norm": 13.419139862060547, "learning_rate": 3.8941897934091086e-07, "loss": 0.7062, "step": 17025 }, { "epoch": 2.74, "grad_norm": 11.772848129272461, "learning_rate": 3.8893883694404723e-07, "loss": 0.7554, "step": 17026 }, { "epoch": 2.74, "grad_norm": 14.876153945922852, "learning_rate": 3.884589848601428e-07, "loss": 0.795, "step": 17027 }, { "epoch": 2.74, "grad_norm": 11.855659484863281, "learning_rate": 3.879794231036893e-07, "loss": 0.7542, "step": 17028 }, { "epoch": 2.74, "grad_norm": 12.536873817443848, "learning_rate": 3.8750015168917297e-07, "loss": 0.6578, "step": 17029 }, { "epoch": 2.74, "grad_norm": 12.049474716186523, "learning_rate": 3.8702117063107225e-07, "loss": 0.5357, "step": 17030 }, { "epoch": 2.74, "grad_norm": 12.819375038146973, "learning_rate": 3.865424799438522e-07, "loss": 0.5809, "step": 17031 }, { "epoch": 2.74, "grad_norm": 16.116504669189453, "learning_rate": 3.860640796419768e-07, "loss": 0.701, "step": 17032 }, { "epoch": 2.74, "grad_norm": 13.861618041992188, "learning_rate": 3.855859697398934e-07, "loss": 0.8774, "step": 17033 }, { "epoch": 2.74, "grad_norm": 14.451087951660156, "learning_rate": 3.8510815025204376e-07, "loss": 0.7316, "step": 17034 }, { "epoch": 2.74, "grad_norm": 13.574958801269531, "learning_rate": 3.84630621192863e-07, "loss": 0.7861, "step": 17035 }, { "epoch": 2.74, "grad_norm": 11.920523643493652, "learning_rate": 3.8415338257677406e-07, "loss": 0.9038, "step": 17036 }, { "epoch": 2.74, "grad_norm": 14.010454177856445, "learning_rate": 3.836764344181909e-07, "loss": 0.5602, "step": 17037 }, { "epoch": 2.74, "grad_norm": 12.772309303283691, "learning_rate": 3.831997767315243e-07, "loss": 0.6996, "step": 17038 }, { "epoch": 2.74, "grad_norm": 12.420598983764648, "learning_rate": 3.8272340953116717e-07, "loss": 0.5168, "step": 17039 }, { "epoch": 2.74, "grad_norm": 14.554437637329102, "learning_rate": 3.822473328315146e-07, "loss": 0.5684, "step": 17040 }, { "epoch": 2.74, "grad_norm": 12.246811866760254, "learning_rate": 3.817715466469429e-07, "loss": 0.6733, "step": 17041 }, { "epoch": 2.74, "grad_norm": 12.76605224609375, "learning_rate": 3.812960509918229e-07, "loss": 0.6145, "step": 17042 }, { "epoch": 2.74, "grad_norm": 14.18738079071045, "learning_rate": 3.808208458805218e-07, "loss": 0.7712, "step": 17043 }, { "epoch": 2.74, "grad_norm": 12.905562400817871, "learning_rate": 3.803459313273894e-07, "loss": 0.5901, "step": 17044 }, { "epoch": 2.74, "grad_norm": 14.903559684753418, "learning_rate": 3.7987130734677413e-07, "loss": 0.6964, "step": 17045 }, { "epoch": 2.74, "grad_norm": 14.433860778808594, "learning_rate": 3.793969739530101e-07, "loss": 0.5786, "step": 17046 }, { "epoch": 2.74, "grad_norm": 12.99157428741455, "learning_rate": 3.789229311604281e-07, "loss": 0.6324, "step": 17047 }, { "epoch": 2.74, "grad_norm": 13.212919235229492, "learning_rate": 3.784491789833433e-07, "loss": 0.8679, "step": 17048 }, { "epoch": 2.74, "grad_norm": 14.403200149536133, "learning_rate": 3.7797571743606985e-07, "loss": 0.686, "step": 17049 }, { "epoch": 2.74, "grad_norm": 15.59200382232666, "learning_rate": 3.775025465329052e-07, "loss": 0.6606, "step": 17050 }, { "epoch": 2.74, "grad_norm": 12.067008018493652, "learning_rate": 3.770296662881456e-07, "loss": 0.7031, "step": 17051 }, { "epoch": 2.75, "grad_norm": 14.974419593811035, "learning_rate": 3.765570767160731e-07, "loss": 0.6972, "step": 17052 }, { "epoch": 2.75, "grad_norm": 10.039846420288086, "learning_rate": 3.7608477783096286e-07, "loss": 0.6316, "step": 17053 }, { "epoch": 2.75, "grad_norm": 11.464544296264648, "learning_rate": 3.756127696470824e-07, "loss": 0.7514, "step": 17054 }, { "epoch": 2.75, "grad_norm": 14.698112487792969, "learning_rate": 3.751410521786858e-07, "loss": 0.6823, "step": 17055 }, { "epoch": 2.75, "grad_norm": 14.703133583068848, "learning_rate": 3.746696254400273e-07, "loss": 0.6192, "step": 17056 }, { "epoch": 2.75, "grad_norm": 12.277192115783691, "learning_rate": 3.741984894453432e-07, "loss": 0.9362, "step": 17057 }, { "epoch": 2.75, "grad_norm": 12.549749374389648, "learning_rate": 3.737276442088633e-07, "loss": 0.6052, "step": 17058 }, { "epoch": 2.75, "grad_norm": 13.786620140075684, "learning_rate": 3.732570897448151e-07, "loss": 0.85, "step": 17059 }, { "epoch": 2.75, "grad_norm": 14.7428560256958, "learning_rate": 3.7278682606740833e-07, "loss": 0.7413, "step": 17060 }, { "epoch": 2.75, "grad_norm": 12.326373100280762, "learning_rate": 3.723168531908472e-07, "loss": 0.7642, "step": 17061 }, { "epoch": 2.75, "grad_norm": 13.514104843139648, "learning_rate": 3.7184717112933033e-07, "loss": 0.5883, "step": 17062 }, { "epoch": 2.75, "grad_norm": 11.458490371704102, "learning_rate": 3.7137777989704307e-07, "loss": 0.7458, "step": 17063 }, { "epoch": 2.75, "grad_norm": 15.375515937805176, "learning_rate": 3.7090867950816625e-07, "loss": 0.7751, "step": 17064 }, { "epoch": 2.75, "grad_norm": 11.010238647460938, "learning_rate": 3.704398699768685e-07, "loss": 0.7807, "step": 17065 }, { "epoch": 2.75, "grad_norm": 13.571187019348145, "learning_rate": 3.6997135131730754e-07, "loss": 0.6814, "step": 17066 }, { "epoch": 2.75, "grad_norm": 13.990699768066406, "learning_rate": 3.695031235436408e-07, "loss": 0.6887, "step": 17067 }, { "epoch": 2.75, "grad_norm": 12.923371315002441, "learning_rate": 3.6903518667000924e-07, "loss": 0.6691, "step": 17068 }, { "epoch": 2.75, "grad_norm": 15.737504959106445, "learning_rate": 3.68567540710546e-07, "loss": 0.6873, "step": 17069 }, { "epoch": 2.75, "grad_norm": 10.563807487487793, "learning_rate": 3.6810018567937977e-07, "loss": 0.5738, "step": 17070 }, { "epoch": 2.75, "grad_norm": 14.498382568359375, "learning_rate": 3.676331215906259e-07, "loss": 0.6611, "step": 17071 }, { "epoch": 2.75, "grad_norm": 12.726734161376953, "learning_rate": 3.6716634845839206e-07, "loss": 0.7805, "step": 17072 }, { "epoch": 2.75, "grad_norm": 19.977575302124023, "learning_rate": 3.66699866296778e-07, "loss": 0.7757, "step": 17073 }, { "epoch": 2.75, "grad_norm": 11.546829223632812, "learning_rate": 3.6623367511987476e-07, "loss": 0.7041, "step": 17074 }, { "epoch": 2.75, "grad_norm": 14.2133150100708, "learning_rate": 3.657677749417643e-07, "loss": 0.9961, "step": 17075 }, { "epoch": 2.75, "grad_norm": 15.007633209228516, "learning_rate": 3.653021657765199e-07, "loss": 0.9841, "step": 17076 }, { "epoch": 2.75, "grad_norm": 12.080599784851074, "learning_rate": 3.6483684763820473e-07, "loss": 0.7299, "step": 17077 }, { "epoch": 2.75, "grad_norm": 13.16733169555664, "learning_rate": 3.643718205408753e-07, "loss": 0.7642, "step": 17078 }, { "epoch": 2.75, "grad_norm": 11.992762565612793, "learning_rate": 3.639070844985759e-07, "loss": 0.7027, "step": 17079 }, { "epoch": 2.75, "grad_norm": 10.776074409484863, "learning_rate": 3.6344263952534875e-07, "loss": 0.6119, "step": 17080 }, { "epoch": 2.75, "grad_norm": 13.046452522277832, "learning_rate": 3.6297848563522034e-07, "loss": 0.8655, "step": 17081 }, { "epoch": 2.75, "grad_norm": 11.44250202178955, "learning_rate": 3.625146228422094e-07, "loss": 0.5661, "step": 17082 }, { "epoch": 2.75, "grad_norm": 11.285481452941895, "learning_rate": 3.6205105116033036e-07, "loss": 0.6826, "step": 17083 }, { "epoch": 2.75, "grad_norm": 12.023452758789062, "learning_rate": 3.615877706035853e-07, "loss": 0.9311, "step": 17084 }, { "epoch": 2.75, "grad_norm": 14.610660552978516, "learning_rate": 3.6112478118596526e-07, "loss": 0.7253, "step": 17085 }, { "epoch": 2.75, "grad_norm": 10.097095489501953, "learning_rate": 3.60662082921458e-07, "loss": 0.685, "step": 17086 }, { "epoch": 2.75, "grad_norm": 16.115802764892578, "learning_rate": 3.6019967582404113e-07, "loss": 0.7592, "step": 17087 }, { "epoch": 2.75, "grad_norm": 10.734920501708984, "learning_rate": 3.5973755990767913e-07, "loss": 0.7934, "step": 17088 }, { "epoch": 2.75, "grad_norm": 11.56509780883789, "learning_rate": 3.592757351863341e-07, "loss": 0.7799, "step": 17089 }, { "epoch": 2.75, "grad_norm": 10.74535083770752, "learning_rate": 3.588142016739515e-07, "loss": 0.7351, "step": 17090 }, { "epoch": 2.75, "grad_norm": 15.57562255859375, "learning_rate": 3.58352959384477e-07, "loss": 0.9495, "step": 17091 }, { "epoch": 2.75, "grad_norm": 11.601235389709473, "learning_rate": 3.5789200833184047e-07, "loss": 0.5318, "step": 17092 }, { "epoch": 2.75, "grad_norm": 12.801468849182129, "learning_rate": 3.574313485299652e-07, "loss": 0.5758, "step": 17093 }, { "epoch": 2.75, "grad_norm": 11.770383834838867, "learning_rate": 3.569709799927679e-07, "loss": 0.6238, "step": 17094 }, { "epoch": 2.75, "grad_norm": 10.952903747558594, "learning_rate": 3.565109027341529e-07, "loss": 0.4641, "step": 17095 }, { "epoch": 2.75, "grad_norm": 12.305373191833496, "learning_rate": 3.5605111676801694e-07, "loss": 0.5756, "step": 17096 }, { "epoch": 2.75, "grad_norm": 12.569740295410156, "learning_rate": 3.5559162210825005e-07, "loss": 0.6386, "step": 17097 }, { "epoch": 2.75, "grad_norm": 13.96555233001709, "learning_rate": 3.5513241876872995e-07, "loss": 0.5776, "step": 17098 }, { "epoch": 2.75, "grad_norm": 13.433380126953125, "learning_rate": 3.546735067633289e-07, "loss": 0.8841, "step": 17099 }, { "epoch": 2.75, "grad_norm": 12.744640350341797, "learning_rate": 3.5421488610590917e-07, "loss": 0.8334, "step": 17100 }, { "epoch": 2.75, "grad_norm": 13.380901336669922, "learning_rate": 3.5375655681032186e-07, "loss": 0.8254, "step": 17101 }, { "epoch": 2.75, "grad_norm": 9.6464204788208, "learning_rate": 3.5329851889041257e-07, "loss": 0.5729, "step": 17102 }, { "epoch": 2.75, "grad_norm": 11.19174861907959, "learning_rate": 3.5284077236001804e-07, "loss": 0.6278, "step": 17103 }, { "epoch": 2.75, "grad_norm": 12.070789337158203, "learning_rate": 3.523833172329616e-07, "loss": 0.6737, "step": 17104 }, { "epoch": 2.75, "grad_norm": 11.92337417602539, "learning_rate": 3.519261535230656e-07, "loss": 0.8077, "step": 17105 }, { "epoch": 2.75, "grad_norm": 13.680571556091309, "learning_rate": 3.514692812441345e-07, "loss": 0.6948, "step": 17106 }, { "epoch": 2.75, "grad_norm": 13.653011322021484, "learning_rate": 3.510127004099717e-07, "loss": 0.7406, "step": 17107 }, { "epoch": 2.75, "grad_norm": 12.825762748718262, "learning_rate": 3.5055641103436957e-07, "loss": 0.551, "step": 17108 }, { "epoch": 2.75, "grad_norm": 10.779454231262207, "learning_rate": 3.501004131311081e-07, "loss": 0.6312, "step": 17109 }, { "epoch": 2.75, "grad_norm": 15.186586380004883, "learning_rate": 3.496447067139641e-07, "loss": 0.7051, "step": 17110 }, { "epoch": 2.75, "grad_norm": 13.199462890625, "learning_rate": 3.49189291796701e-07, "loss": 0.5799, "step": 17111 }, { "epoch": 2.75, "grad_norm": 14.196423530578613, "learning_rate": 3.4873416839307337e-07, "loss": 0.8311, "step": 17112 }, { "epoch": 2.75, "grad_norm": 14.40078067779541, "learning_rate": 3.482793365168324e-07, "loss": 0.6774, "step": 17113 }, { "epoch": 2.75, "grad_norm": 14.645258903503418, "learning_rate": 3.4782479618171496e-07, "loss": 0.5677, "step": 17114 }, { "epoch": 2.76, "grad_norm": 14.203293800354004, "learning_rate": 3.4737054740144993e-07, "loss": 0.717, "step": 17115 }, { "epoch": 2.76, "grad_norm": 10.693045616149902, "learning_rate": 3.4691659018976085e-07, "loss": 0.6112, "step": 17116 }, { "epoch": 2.76, "grad_norm": 11.55189323425293, "learning_rate": 3.464629245603568e-07, "loss": 0.4784, "step": 17117 }, { "epoch": 2.76, "grad_norm": 14.115577697753906, "learning_rate": 3.460095505269445e-07, "loss": 0.7669, "step": 17118 }, { "epoch": 2.76, "grad_norm": 11.510232925415039, "learning_rate": 3.455564681032175e-07, "loss": 0.833, "step": 17119 }, { "epoch": 2.76, "grad_norm": 11.94815444946289, "learning_rate": 3.451036773028604e-07, "loss": 0.6393, "step": 17120 }, { "epoch": 2.76, "grad_norm": 10.774641036987305, "learning_rate": 3.4465117813955227e-07, "loss": 0.4909, "step": 17121 }, { "epoch": 2.76, "grad_norm": 11.707094192504883, "learning_rate": 3.4419897062695885e-07, "loss": 0.6652, "step": 17122 }, { "epoch": 2.76, "grad_norm": 14.033576965332031, "learning_rate": 3.437470547787425e-07, "loss": 0.7742, "step": 17123 }, { "epoch": 2.76, "grad_norm": 12.856758117675781, "learning_rate": 3.432954306085523e-07, "loss": 0.7267, "step": 17124 }, { "epoch": 2.76, "grad_norm": 13.19748592376709, "learning_rate": 3.428440981300296e-07, "loss": 0.5256, "step": 17125 }, { "epoch": 2.76, "grad_norm": 13.366097450256348, "learning_rate": 3.423930573568079e-07, "loss": 0.6663, "step": 17126 }, { "epoch": 2.76, "grad_norm": 15.36151123046875, "learning_rate": 3.4194230830251195e-07, "loss": 0.8337, "step": 17127 }, { "epoch": 2.76, "grad_norm": 11.983166694641113, "learning_rate": 3.414918509807552e-07, "loss": 0.8038, "step": 17128 }, { "epoch": 2.76, "grad_norm": 12.623821258544922, "learning_rate": 3.4104168540514794e-07, "loss": 0.7905, "step": 17129 }, { "epoch": 2.76, "grad_norm": 13.7134428024292, "learning_rate": 3.405918115892848e-07, "loss": 0.7772, "step": 17130 }, { "epoch": 2.76, "grad_norm": 14.22107219696045, "learning_rate": 3.401422295467549e-07, "loss": 0.8117, "step": 17131 }, { "epoch": 2.76, "grad_norm": 11.030845642089844, "learning_rate": 3.396929392911408e-07, "loss": 0.4852, "step": 17132 }, { "epoch": 2.76, "grad_norm": 20.680055618286133, "learning_rate": 3.392439408360104e-07, "loss": 0.8886, "step": 17133 }, { "epoch": 2.76, "grad_norm": 12.356870651245117, "learning_rate": 3.3879523419492855e-07, "loss": 0.749, "step": 17134 }, { "epoch": 2.76, "grad_norm": 17.143096923828125, "learning_rate": 3.3834681938144875e-07, "loss": 0.6758, "step": 17135 }, { "epoch": 2.76, "grad_norm": 12.916812896728516, "learning_rate": 3.3789869640911465e-07, "loss": 0.6436, "step": 17136 }, { "epoch": 2.76, "grad_norm": 12.849555969238281, "learning_rate": 3.374508652914643e-07, "loss": 0.5139, "step": 17137 }, { "epoch": 2.76, "grad_norm": 12.335688591003418, "learning_rate": 3.370033260420236e-07, "loss": 0.8394, "step": 17138 }, { "epoch": 2.76, "grad_norm": 13.129561424255371, "learning_rate": 3.365560786743094e-07, "loss": 0.7538, "step": 17139 }, { "epoch": 2.76, "grad_norm": 13.955615043640137, "learning_rate": 3.361091232018354e-07, "loss": 0.7177, "step": 17140 }, { "epoch": 2.76, "grad_norm": 14.050724983215332, "learning_rate": 3.356624596380975e-07, "loss": 0.5735, "step": 17141 }, { "epoch": 2.76, "grad_norm": 12.148797035217285, "learning_rate": 3.3521608799659265e-07, "loss": 0.8253, "step": 17142 }, { "epoch": 2.76, "grad_norm": 13.338228225708008, "learning_rate": 3.3477000829080007e-07, "loss": 0.8169, "step": 17143 }, { "epoch": 2.76, "grad_norm": 11.214555740356445, "learning_rate": 3.343242205341957e-07, "loss": 0.5522, "step": 17144 }, { "epoch": 2.76, "grad_norm": 14.402498245239258, "learning_rate": 3.338787247402453e-07, "loss": 0.7881, "step": 17145 }, { "epoch": 2.76, "grad_norm": 12.880171775817871, "learning_rate": 3.334335209224049e-07, "loss": 0.7808, "step": 17146 }, { "epoch": 2.76, "grad_norm": 13.756706237792969, "learning_rate": 3.3298860909412257e-07, "loss": 0.7884, "step": 17147 }, { "epoch": 2.76, "grad_norm": 15.580781936645508, "learning_rate": 3.3254398926883757e-07, "loss": 0.6476, "step": 17148 }, { "epoch": 2.76, "grad_norm": 13.096831321716309, "learning_rate": 3.320996614599814e-07, "loss": 0.7919, "step": 17149 }, { "epoch": 2.76, "grad_norm": 12.872457504272461, "learning_rate": 3.316556256809722e-07, "loss": 0.7143, "step": 17150 }, { "epoch": 2.76, "grad_norm": 14.736122131347656, "learning_rate": 3.3121188194522594e-07, "loss": 0.8889, "step": 17151 }, { "epoch": 2.76, "grad_norm": 13.169218063354492, "learning_rate": 3.30768430266144e-07, "loss": 0.7015, "step": 17152 }, { "epoch": 2.76, "grad_norm": 11.535347938537598, "learning_rate": 3.303252706571236e-07, "loss": 0.5947, "step": 17153 }, { "epoch": 2.76, "grad_norm": 11.123355865478516, "learning_rate": 3.2988240313154953e-07, "loss": 0.5616, "step": 17154 }, { "epoch": 2.76, "grad_norm": 26.335241317749023, "learning_rate": 3.294398277027977e-07, "loss": 0.7578, "step": 17155 }, { "epoch": 2.76, "grad_norm": 11.34741497039795, "learning_rate": 3.289975443842397e-07, "loss": 0.6584, "step": 17156 }, { "epoch": 2.76, "grad_norm": 21.81805419921875, "learning_rate": 3.285555531892326e-07, "loss": 0.5673, "step": 17157 }, { "epoch": 2.76, "grad_norm": 11.972759246826172, "learning_rate": 3.2811385413112906e-07, "loss": 0.8819, "step": 17158 }, { "epoch": 2.76, "grad_norm": 16.97593879699707, "learning_rate": 3.276724472232695e-07, "loss": 0.758, "step": 17159 }, { "epoch": 2.76, "grad_norm": 12.033792495727539, "learning_rate": 3.272313324789878e-07, "loss": 0.4342, "step": 17160 }, { "epoch": 2.76, "grad_norm": 12.289908409118652, "learning_rate": 3.2679050991160976e-07, "loss": 0.7293, "step": 17161 }, { "epoch": 2.76, "grad_norm": 14.87761402130127, "learning_rate": 3.2634997953444823e-07, "loss": 0.6521, "step": 17162 }, { "epoch": 2.76, "grad_norm": 12.273324012756348, "learning_rate": 3.259097413608114e-07, "loss": 0.5926, "step": 17163 }, { "epoch": 2.76, "grad_norm": 13.269035339355469, "learning_rate": 3.2546979540399757e-07, "loss": 0.8868, "step": 17164 }, { "epoch": 2.76, "grad_norm": 15.103445053100586, "learning_rate": 3.25030141677295e-07, "loss": 0.7112, "step": 17165 }, { "epoch": 2.76, "grad_norm": 14.939096450805664, "learning_rate": 3.2459078019398515e-07, "loss": 0.7542, "step": 17166 }, { "epoch": 2.76, "grad_norm": 15.280985832214355, "learning_rate": 3.2415171096733646e-07, "loss": 0.7986, "step": 17167 }, { "epoch": 2.76, "grad_norm": 15.874504089355469, "learning_rate": 3.237129340106138e-07, "loss": 0.8358, "step": 17168 }, { "epoch": 2.76, "grad_norm": 12.51274585723877, "learning_rate": 3.232744493370721e-07, "loss": 0.6515, "step": 17169 }, { "epoch": 2.76, "grad_norm": 11.693306922912598, "learning_rate": 3.228362569599552e-07, "loss": 0.8111, "step": 17170 }, { "epoch": 2.76, "grad_norm": 14.286714553833008, "learning_rate": 3.223983568924971e-07, "loss": 0.7508, "step": 17171 }, { "epoch": 2.76, "grad_norm": 12.392738342285156, "learning_rate": 3.219607491479282e-07, "loss": 0.7149, "step": 17172 }, { "epoch": 2.76, "grad_norm": 13.509178161621094, "learning_rate": 3.215234337394657e-07, "loss": 0.7227, "step": 17173 }, { "epoch": 2.76, "grad_norm": 13.718589782714844, "learning_rate": 3.210864106803169e-07, "loss": 0.7185, "step": 17174 }, { "epoch": 2.76, "grad_norm": 15.928108215332031, "learning_rate": 3.206496799836867e-07, "loss": 0.6861, "step": 17175 }, { "epoch": 2.76, "grad_norm": 12.994976997375488, "learning_rate": 3.2021324166276457e-07, "loss": 0.6599, "step": 17176 }, { "epoch": 2.77, "grad_norm": 13.058454513549805, "learning_rate": 3.1977709573073444e-07, "loss": 0.645, "step": 17177 }, { "epoch": 2.77, "grad_norm": 10.050790786743164, "learning_rate": 3.193412422007702e-07, "loss": 0.642, "step": 17178 }, { "epoch": 2.77, "grad_norm": 12.048048973083496, "learning_rate": 3.1890568108603694e-07, "loss": 0.6273, "step": 17179 }, { "epoch": 2.77, "grad_norm": 11.225380897521973, "learning_rate": 3.184704123996918e-07, "loss": 0.6161, "step": 17180 }, { "epoch": 2.77, "grad_norm": 13.619810104370117, "learning_rate": 3.1803543615488317e-07, "loss": 0.8094, "step": 17181 }, { "epoch": 2.77, "grad_norm": 13.828503608703613, "learning_rate": 3.176007523647484e-07, "loss": 0.7151, "step": 17182 }, { "epoch": 2.77, "grad_norm": 11.283801078796387, "learning_rate": 3.1716636104241915e-07, "loss": 0.7227, "step": 17183 }, { "epoch": 2.77, "grad_norm": 15.730470657348633, "learning_rate": 3.1673226220101495e-07, "loss": 0.902, "step": 17184 }, { "epoch": 2.77, "grad_norm": 11.254029273986816, "learning_rate": 3.162984558536497e-07, "loss": 0.6582, "step": 17185 }, { "epoch": 2.77, "grad_norm": 15.863310813903809, "learning_rate": 3.1586494201342855e-07, "loss": 0.8282, "step": 17186 }, { "epoch": 2.77, "grad_norm": 14.241203308105469, "learning_rate": 3.1543172069344206e-07, "loss": 0.6561, "step": 17187 }, { "epoch": 2.77, "grad_norm": 12.618341445922852, "learning_rate": 3.1499879190677874e-07, "loss": 0.6587, "step": 17188 }, { "epoch": 2.77, "grad_norm": 14.929046630859375, "learning_rate": 3.1456615566651585e-07, "loss": 0.6269, "step": 17189 }, { "epoch": 2.77, "grad_norm": 11.950883865356445, "learning_rate": 3.141338119857207e-07, "loss": 0.5574, "step": 17190 }, { "epoch": 2.77, "grad_norm": 14.546874046325684, "learning_rate": 3.137017608774551e-07, "loss": 0.5458, "step": 17191 }, { "epoch": 2.77, "grad_norm": 13.271037101745605, "learning_rate": 3.1327000235476636e-07, "loss": 0.8516, "step": 17192 }, { "epoch": 2.77, "grad_norm": 12.19403076171875, "learning_rate": 3.128385364306974e-07, "loss": 0.7029, "step": 17193 }, { "epoch": 2.77, "grad_norm": 12.14681625366211, "learning_rate": 3.124073631182833e-07, "loss": 0.9201, "step": 17194 }, { "epoch": 2.77, "grad_norm": 15.217073440551758, "learning_rate": 3.119764824305438e-07, "loss": 0.7918, "step": 17195 }, { "epoch": 2.77, "grad_norm": 12.627062797546387, "learning_rate": 3.115458943804983e-07, "loss": 0.8454, "step": 17196 }, { "epoch": 2.77, "grad_norm": 12.301673889160156, "learning_rate": 3.1111559898115205e-07, "loss": 0.8963, "step": 17197 }, { "epoch": 2.77, "grad_norm": 12.225071907043457, "learning_rate": 3.106855962455002e-07, "loss": 0.6696, "step": 17198 }, { "epoch": 2.77, "grad_norm": 15.659175872802734, "learning_rate": 3.102558861865346e-07, "loss": 0.8413, "step": 17199 }, { "epoch": 2.77, "grad_norm": 12.061664581298828, "learning_rate": 3.0982646881723254e-07, "loss": 0.5329, "step": 17200 }, { "epoch": 2.77, "grad_norm": 13.833744049072266, "learning_rate": 3.093973441505682e-07, "loss": 0.8488, "step": 17201 }, { "epoch": 2.77, "grad_norm": 14.197656631469727, "learning_rate": 3.089685121995012e-07, "loss": 0.7869, "step": 17202 }, { "epoch": 2.77, "grad_norm": 15.442768096923828, "learning_rate": 3.0853997297698446e-07, "loss": 0.7534, "step": 17203 }, { "epoch": 2.77, "grad_norm": 12.302254676818848, "learning_rate": 3.081117264959654e-07, "loss": 0.901, "step": 17204 }, { "epoch": 2.77, "grad_norm": 11.962790489196777, "learning_rate": 3.076837727693771e-07, "loss": 0.7095, "step": 17205 }, { "epoch": 2.77, "grad_norm": 10.625670433044434, "learning_rate": 3.072561118101469e-07, "loss": 0.4012, "step": 17206 }, { "epoch": 2.77, "grad_norm": 14.637264251708984, "learning_rate": 3.0682874363119455e-07, "loss": 0.6369, "step": 17207 }, { "epoch": 2.77, "grad_norm": 15.610468864440918, "learning_rate": 3.0640166824542516e-07, "loss": 0.8781, "step": 17208 }, { "epoch": 2.77, "grad_norm": 14.22201919555664, "learning_rate": 3.05974885665743e-07, "loss": 0.9192, "step": 17209 }, { "epoch": 2.77, "grad_norm": 12.451343536376953, "learning_rate": 3.055483959050387e-07, "loss": 0.7837, "step": 17210 }, { "epoch": 2.77, "grad_norm": 17.174705505371094, "learning_rate": 3.0512219897619324e-07, "loss": 0.568, "step": 17211 }, { "epoch": 2.77, "grad_norm": 18.75341033935547, "learning_rate": 3.0469629489208284e-07, "loss": 0.6066, "step": 17212 }, { "epoch": 2.77, "grad_norm": 12.383856773376465, "learning_rate": 3.0427068366557064e-07, "loss": 0.7645, "step": 17213 }, { "epoch": 2.77, "grad_norm": 12.29114055633545, "learning_rate": 3.038453653095119e-07, "loss": 0.7096, "step": 17214 }, { "epoch": 2.77, "grad_norm": 14.904086112976074, "learning_rate": 3.0342033983675744e-07, "loss": 0.9233, "step": 17215 }, { "epoch": 2.77, "grad_norm": 13.53648567199707, "learning_rate": 3.0299560726014145e-07, "loss": 0.7016, "step": 17216 }, { "epoch": 2.77, "grad_norm": 10.170770645141602, "learning_rate": 3.025711675924947e-07, "loss": 0.6549, "step": 17217 }, { "epoch": 2.77, "grad_norm": 12.193245887756348, "learning_rate": 3.021470208466404e-07, "loss": 0.5689, "step": 17218 }, { "epoch": 2.77, "grad_norm": 13.046211242675781, "learning_rate": 3.0172316703538594e-07, "loss": 0.6495, "step": 17219 }, { "epoch": 2.77, "grad_norm": 13.249829292297363, "learning_rate": 3.0129960617153897e-07, "loss": 0.684, "step": 17220 }, { "epoch": 2.77, "grad_norm": 12.550537109375, "learning_rate": 3.008763382678903e-07, "loss": 0.9091, "step": 17221 }, { "epoch": 2.77, "grad_norm": 15.53890609741211, "learning_rate": 3.004533633372264e-07, "loss": 0.9223, "step": 17222 }, { "epoch": 2.77, "grad_norm": 13.82072639465332, "learning_rate": 3.000306813923237e-07, "loss": 0.5907, "step": 17223 }, { "epoch": 2.77, "grad_norm": 11.409341812133789, "learning_rate": 2.996082924459498e-07, "loss": 0.5669, "step": 17224 }, { "epoch": 2.77, "grad_norm": 10.077442169189453, "learning_rate": 2.9918619651086334e-07, "loss": 0.5151, "step": 17225 }, { "epoch": 2.77, "grad_norm": 13.190171241760254, "learning_rate": 2.987643935998152e-07, "loss": 0.8063, "step": 17226 }, { "epoch": 2.77, "grad_norm": 10.802550315856934, "learning_rate": 2.983428837255431e-07, "loss": 0.4744, "step": 17227 }, { "epoch": 2.77, "grad_norm": 13.7037353515625, "learning_rate": 2.979216669007823e-07, "loss": 0.7832, "step": 17228 }, { "epoch": 2.77, "grad_norm": 12.510295867919922, "learning_rate": 2.9750074313825595e-07, "loss": 0.7784, "step": 17229 }, { "epoch": 2.77, "grad_norm": 13.68331527709961, "learning_rate": 2.9708011245067727e-07, "loss": 0.6701, "step": 17230 }, { "epoch": 2.77, "grad_norm": 12.868864059448242, "learning_rate": 2.966597748507538e-07, "loss": 0.8856, "step": 17231 }, { "epoch": 2.77, "grad_norm": 14.887064933776855, "learning_rate": 2.96239730351181e-07, "loss": 0.8733, "step": 17232 }, { "epoch": 2.77, "grad_norm": 13.299318313598633, "learning_rate": 2.9581997896464653e-07, "loss": 0.5949, "step": 17233 }, { "epoch": 2.77, "grad_norm": 14.110349655151367, "learning_rate": 2.9540052070382905e-07, "loss": 0.6151, "step": 17234 }, { "epoch": 2.77, "grad_norm": 14.1228609085083, "learning_rate": 2.949813555813996e-07, "loss": 0.6333, "step": 17235 }, { "epoch": 2.77, "grad_norm": 14.210408210754395, "learning_rate": 2.945624836100203e-07, "loss": 0.5314, "step": 17236 }, { "epoch": 2.77, "grad_norm": 11.531035423278809, "learning_rate": 2.941439048023431e-07, "loss": 0.7377, "step": 17237 }, { "epoch": 2.77, "grad_norm": 13.251955032348633, "learning_rate": 2.937256191710103e-07, "loss": 0.6, "step": 17238 }, { "epoch": 2.78, "grad_norm": 13.133447647094727, "learning_rate": 2.9330762672865833e-07, "loss": 0.8189, "step": 17239 }, { "epoch": 2.78, "grad_norm": 12.21458625793457, "learning_rate": 2.9288992748791265e-07, "loss": 0.6017, "step": 17240 }, { "epoch": 2.78, "grad_norm": 11.784235954284668, "learning_rate": 2.924725214613899e-07, "loss": 0.5924, "step": 17241 }, { "epoch": 2.78, "grad_norm": 12.753366470336914, "learning_rate": 2.920554086616989e-07, "loss": 0.9561, "step": 17242 }, { "epoch": 2.78, "grad_norm": 11.210955619812012, "learning_rate": 2.916385891014373e-07, "loss": 0.647, "step": 17243 }, { "epoch": 2.78, "grad_norm": 12.569671630859375, "learning_rate": 2.9122206279319834e-07, "loss": 0.7799, "step": 17244 }, { "epoch": 2.78, "grad_norm": 12.153632164001465, "learning_rate": 2.9080582974956307e-07, "loss": 0.6931, "step": 17245 }, { "epoch": 2.78, "grad_norm": 10.9486083984375, "learning_rate": 2.9038988998310256e-07, "loss": 0.6462, "step": 17246 }, { "epoch": 2.78, "grad_norm": 10.252972602844238, "learning_rate": 2.899742435063824e-07, "loss": 0.5714, "step": 17247 }, { "epoch": 2.78, "grad_norm": 11.805915832519531, "learning_rate": 2.895588903319557e-07, "loss": 0.5455, "step": 17248 }, { "epoch": 2.78, "grad_norm": 12.055535316467285, "learning_rate": 2.891438304723693e-07, "loss": 0.6073, "step": 17249 }, { "epoch": 2.78, "grad_norm": 11.966803550720215, "learning_rate": 2.8872906394016407e-07, "loss": 0.7249, "step": 17250 }, { "epoch": 2.78, "grad_norm": 11.851898193359375, "learning_rate": 2.883145907478646e-07, "loss": 0.5921, "step": 17251 }, { "epoch": 2.78, "grad_norm": 15.89577865600586, "learning_rate": 2.879004109079908e-07, "loss": 0.6415, "step": 17252 }, { "epoch": 2.78, "grad_norm": 12.81173324584961, "learning_rate": 2.874865244330549e-07, "loss": 0.6464, "step": 17253 }, { "epoch": 2.78, "grad_norm": 12.596261024475098, "learning_rate": 2.870729313355569e-07, "loss": 0.6177, "step": 17254 }, { "epoch": 2.78, "grad_norm": 14.002067565917969, "learning_rate": 2.866596316279935e-07, "loss": 0.9782, "step": 17255 }, { "epoch": 2.78, "grad_norm": 14.528013229370117, "learning_rate": 2.8624662532284465e-07, "loss": 0.7102, "step": 17256 }, { "epoch": 2.78, "grad_norm": 9.762451171875, "learning_rate": 2.8583391243258816e-07, "loss": 0.5519, "step": 17257 }, { "epoch": 2.78, "grad_norm": 14.096756935119629, "learning_rate": 2.854214929696897e-07, "loss": 0.8734, "step": 17258 }, { "epoch": 2.78, "grad_norm": 14.948452949523926, "learning_rate": 2.85009366946607e-07, "loss": 0.6968, "step": 17259 }, { "epoch": 2.78, "grad_norm": 11.705714225769043, "learning_rate": 2.8459753437578897e-07, "loss": 0.6668, "step": 17260 }, { "epoch": 2.78, "grad_norm": 12.232789039611816, "learning_rate": 2.8418599526967463e-07, "loss": 0.5774, "step": 17261 }, { "epoch": 2.78, "grad_norm": 13.54786491394043, "learning_rate": 2.8377474964069617e-07, "loss": 0.7943, "step": 17262 }, { "epoch": 2.78, "grad_norm": 14.249347686767578, "learning_rate": 2.8336379750127487e-07, "loss": 0.7144, "step": 17263 }, { "epoch": 2.78, "grad_norm": 13.947790145874023, "learning_rate": 2.8295313886382514e-07, "loss": 0.8231, "step": 17264 }, { "epoch": 2.78, "grad_norm": 14.234322547912598, "learning_rate": 2.8254277374074936e-07, "loss": 0.8876, "step": 17265 }, { "epoch": 2.78, "grad_norm": 16.439409255981445, "learning_rate": 2.8213270214444533e-07, "loss": 0.7907, "step": 17266 }, { "epoch": 2.78, "grad_norm": 11.137853622436523, "learning_rate": 2.8172292408729763e-07, "loss": 0.7573, "step": 17267 }, { "epoch": 2.78, "grad_norm": 11.08419418334961, "learning_rate": 2.8131343958168633e-07, "loss": 0.6749, "step": 17268 }, { "epoch": 2.78, "grad_norm": 12.22900390625, "learning_rate": 2.809042486399782e-07, "loss": 0.5114, "step": 17269 }, { "epoch": 2.78, "grad_norm": 11.349751472473145, "learning_rate": 2.804953512745345e-07, "loss": 0.5039, "step": 17270 }, { "epoch": 2.78, "grad_norm": 10.937249183654785, "learning_rate": 2.8008674749770647e-07, "loss": 0.6488, "step": 17271 }, { "epoch": 2.78, "grad_norm": 13.046422004699707, "learning_rate": 2.7967843732183644e-07, "loss": 0.6378, "step": 17272 }, { "epoch": 2.78, "grad_norm": 11.688849449157715, "learning_rate": 2.792704207592567e-07, "loss": 0.5331, "step": 17273 }, { "epoch": 2.78, "grad_norm": 14.047077178955078, "learning_rate": 2.788626978222941e-07, "loss": 0.6283, "step": 17274 }, { "epoch": 2.78, "grad_norm": 9.424240112304688, "learning_rate": 2.7845526852326223e-07, "loss": 0.4162, "step": 17275 }, { "epoch": 2.78, "grad_norm": 14.501726150512695, "learning_rate": 2.7804813287446885e-07, "loss": 0.5705, "step": 17276 }, { "epoch": 2.78, "grad_norm": 11.088190078735352, "learning_rate": 2.77641290888212e-07, "loss": 0.5757, "step": 17277 }, { "epoch": 2.78, "grad_norm": 13.328866004943848, "learning_rate": 2.7723474257678073e-07, "loss": 0.5513, "step": 17278 }, { "epoch": 2.78, "grad_norm": 13.196175575256348, "learning_rate": 2.768284879524552e-07, "loss": 0.8197, "step": 17279 }, { "epoch": 2.78, "grad_norm": 11.38140869140625, "learning_rate": 2.764225270275067e-07, "loss": 0.825, "step": 17280 }, { "epoch": 2.78, "grad_norm": 13.97841739654541, "learning_rate": 2.7601685981419766e-07, "loss": 0.7252, "step": 17281 }, { "epoch": 2.78, "grad_norm": 14.20519733428955, "learning_rate": 2.756114863247827e-07, "loss": 0.8949, "step": 17282 }, { "epoch": 2.78, "grad_norm": 12.04537296295166, "learning_rate": 2.7520640657150655e-07, "loss": 0.7917, "step": 17283 }, { "epoch": 2.78, "grad_norm": 14.033782958984375, "learning_rate": 2.748016205666015e-07, "loss": 0.765, "step": 17284 }, { "epoch": 2.78, "grad_norm": 13.268153190612793, "learning_rate": 2.7439712832230015e-07, "loss": 0.7169, "step": 17285 }, { "epoch": 2.78, "grad_norm": 10.963571548461914, "learning_rate": 2.739929298508159e-07, "loss": 0.8945, "step": 17286 }, { "epoch": 2.78, "grad_norm": 14.181879043579102, "learning_rate": 2.7358902516436247e-07, "loss": 0.6384, "step": 17287 }, { "epoch": 2.78, "grad_norm": 14.289237976074219, "learning_rate": 2.731854142751367e-07, "loss": 0.7115, "step": 17288 }, { "epoch": 2.78, "grad_norm": 12.547645568847656, "learning_rate": 2.727820971953288e-07, "loss": 0.6889, "step": 17289 }, { "epoch": 2.78, "grad_norm": 12.428003311157227, "learning_rate": 2.723790739371268e-07, "loss": 0.662, "step": 17290 }, { "epoch": 2.78, "grad_norm": 11.801429748535156, "learning_rate": 2.7197634451270105e-07, "loss": 0.9012, "step": 17291 }, { "epoch": 2.78, "grad_norm": 13.01628589630127, "learning_rate": 2.715739089342162e-07, "loss": 0.667, "step": 17292 }, { "epoch": 2.78, "grad_norm": 10.440999031066895, "learning_rate": 2.711717672138303e-07, "loss": 0.5826, "step": 17293 }, { "epoch": 2.78, "grad_norm": 14.965437889099121, "learning_rate": 2.707699193636881e-07, "loss": 0.5863, "step": 17294 }, { "epoch": 2.78, "grad_norm": 12.33018970489502, "learning_rate": 2.703683653959288e-07, "loss": 0.6122, "step": 17295 }, { "epoch": 2.78, "grad_norm": 17.11710548400879, "learning_rate": 2.699671053226838e-07, "loss": 0.8756, "step": 17296 }, { "epoch": 2.78, "grad_norm": 12.672492027282715, "learning_rate": 2.6956613915606886e-07, "loss": 0.5442, "step": 17297 }, { "epoch": 2.78, "grad_norm": 14.767308235168457, "learning_rate": 2.691654669082011e-07, "loss": 0.6897, "step": 17298 }, { "epoch": 2.78, "grad_norm": 11.896382331848145, "learning_rate": 2.687650885911808e-07, "loss": 0.6965, "step": 17299 }, { "epoch": 2.78, "grad_norm": 21.849559783935547, "learning_rate": 2.683650042170993e-07, "loss": 0.7816, "step": 17300 }, { "epoch": 2.79, "grad_norm": 11.050015449523926, "learning_rate": 2.6796521379804594e-07, "loss": 0.8035, "step": 17301 }, { "epoch": 2.79, "grad_norm": 12.971807479858398, "learning_rate": 2.6756571734609436e-07, "loss": 0.7085, "step": 17302 }, { "epoch": 2.79, "grad_norm": 14.342254638671875, "learning_rate": 2.671665148733127e-07, "loss": 0.6413, "step": 17303 }, { "epoch": 2.79, "grad_norm": 13.450302124023438, "learning_rate": 2.6676760639175904e-07, "loss": 0.5174, "step": 17304 }, { "epoch": 2.79, "grad_norm": 14.958980560302734, "learning_rate": 2.6636899191348264e-07, "loss": 0.806, "step": 17305 }, { "epoch": 2.79, "grad_norm": 10.391402244567871, "learning_rate": 2.659706714505239e-07, "loss": 0.693, "step": 17306 }, { "epoch": 2.79, "grad_norm": 12.355745315551758, "learning_rate": 2.6557264501491654e-07, "loss": 0.6205, "step": 17307 }, { "epoch": 2.79, "grad_norm": 14.063451766967773, "learning_rate": 2.6517491261867976e-07, "loss": 0.6199, "step": 17308 }, { "epoch": 2.79, "grad_norm": 14.932655334472656, "learning_rate": 2.647774742738296e-07, "loss": 0.6205, "step": 17309 }, { "epoch": 2.79, "grad_norm": 15.974666595458984, "learning_rate": 2.6438032999237196e-07, "loss": 0.6905, "step": 17310 }, { "epoch": 2.79, "grad_norm": 12.907462120056152, "learning_rate": 2.6398347978630167e-07, "loss": 0.7919, "step": 17311 }, { "epoch": 2.79, "grad_norm": 11.77615737915039, "learning_rate": 2.635869236676081e-07, "loss": 0.6817, "step": 17312 }, { "epoch": 2.79, "grad_norm": 14.372336387634277, "learning_rate": 2.63190661648266e-07, "loss": 0.6274, "step": 17313 }, { "epoch": 2.79, "grad_norm": 13.900601387023926, "learning_rate": 2.6279469374024816e-07, "loss": 0.8601, "step": 17314 }, { "epoch": 2.79, "grad_norm": 12.47009563446045, "learning_rate": 2.6239901995551485e-07, "loss": 0.8332, "step": 17315 }, { "epoch": 2.79, "grad_norm": 13.99741268157959, "learning_rate": 2.6200364030601555e-07, "loss": 0.6396, "step": 17316 }, { "epoch": 2.79, "grad_norm": 15.10606861114502, "learning_rate": 2.6160855480369506e-07, "loss": 0.5411, "step": 17317 }, { "epoch": 2.79, "grad_norm": 13.360368728637695, "learning_rate": 2.612137634604872e-07, "loss": 0.6997, "step": 17318 }, { "epoch": 2.79, "grad_norm": 12.67432975769043, "learning_rate": 2.608192662883169e-07, "loss": 0.7727, "step": 17319 }, { "epoch": 2.79, "grad_norm": 14.335585594177246, "learning_rate": 2.604250632991012e-07, "loss": 0.777, "step": 17320 }, { "epoch": 2.79, "grad_norm": 10.454397201538086, "learning_rate": 2.600311545047451e-07, "loss": 0.6466, "step": 17321 }, { "epoch": 2.79, "grad_norm": 13.166046142578125, "learning_rate": 2.5963753991714915e-07, "loss": 0.6696, "step": 17322 }, { "epoch": 2.79, "grad_norm": 13.275713920593262, "learning_rate": 2.592442195482037e-07, "loss": 0.6864, "step": 17323 }, { "epoch": 2.79, "grad_norm": 12.114606857299805, "learning_rate": 2.588511934097859e-07, "loss": 0.6069, "step": 17324 }, { "epoch": 2.79, "grad_norm": 13.323229789733887, "learning_rate": 2.5845846151377083e-07, "loss": 0.6147, "step": 17325 }, { "epoch": 2.79, "grad_norm": 14.322208404541016, "learning_rate": 2.5806602387202117e-07, "loss": 0.8808, "step": 17326 }, { "epoch": 2.79, "grad_norm": 14.36049747467041, "learning_rate": 2.576738804963885e-07, "loss": 0.7052, "step": 17327 }, { "epoch": 2.79, "grad_norm": 15.290732383728027, "learning_rate": 2.572820313987212e-07, "loss": 0.894, "step": 17328 }, { "epoch": 2.79, "grad_norm": 13.181634902954102, "learning_rate": 2.5689047659085196e-07, "loss": 0.7649, "step": 17329 }, { "epoch": 2.79, "grad_norm": 11.141104698181152, "learning_rate": 2.5649921608461024e-07, "loss": 0.6655, "step": 17330 }, { "epoch": 2.79, "grad_norm": 10.931347846984863, "learning_rate": 2.561082498918155e-07, "loss": 0.5531, "step": 17331 }, { "epoch": 2.79, "grad_norm": 13.557074546813965, "learning_rate": 2.5571757802427487e-07, "loss": 0.703, "step": 17332 }, { "epoch": 2.79, "grad_norm": 13.22552490234375, "learning_rate": 2.553272004937923e-07, "loss": 0.6907, "step": 17333 }, { "epoch": 2.79, "grad_norm": 35.3502082824707, "learning_rate": 2.549371173121562e-07, "loss": 0.7636, "step": 17334 }, { "epoch": 2.79, "grad_norm": 12.446189880371094, "learning_rate": 2.5454732849115037e-07, "loss": 0.7635, "step": 17335 }, { "epoch": 2.79, "grad_norm": 13.651082992553711, "learning_rate": 2.54157834042551e-07, "loss": 0.7714, "step": 17336 }, { "epoch": 2.79, "grad_norm": 13.350526809692383, "learning_rate": 2.537686339781209e-07, "loss": 0.7333, "step": 17337 }, { "epoch": 2.79, "grad_norm": 13.281366348266602, "learning_rate": 2.533797283096162e-07, "loss": 0.6586, "step": 17338 }, { "epoch": 2.79, "grad_norm": 10.708218574523926, "learning_rate": 2.529911170487853e-07, "loss": 0.6854, "step": 17339 }, { "epoch": 2.79, "grad_norm": 12.640164375305176, "learning_rate": 2.526028002073655e-07, "loss": 0.6791, "step": 17340 }, { "epoch": 2.79, "grad_norm": 12.288103103637695, "learning_rate": 2.5221477779708847e-07, "loss": 0.6248, "step": 17341 }, { "epoch": 2.79, "grad_norm": 13.235777854919434, "learning_rate": 2.518270498296727e-07, "loss": 0.6554, "step": 17342 }, { "epoch": 2.79, "grad_norm": 10.717010498046875, "learning_rate": 2.514396163168309e-07, "loss": 0.6139, "step": 17343 }, { "epoch": 2.79, "grad_norm": 12.930956840515137, "learning_rate": 2.5105247727026603e-07, "loss": 0.7476, "step": 17344 }, { "epoch": 2.79, "grad_norm": 15.283554077148438, "learning_rate": 2.5066563270167097e-07, "loss": 0.6876, "step": 17345 }, { "epoch": 2.79, "grad_norm": 13.86506175994873, "learning_rate": 2.5027908262273306e-07, "loss": 0.5653, "step": 17346 }, { "epoch": 2.79, "grad_norm": 12.362582206726074, "learning_rate": 2.498928270451273e-07, "loss": 0.7208, "step": 17347 }, { "epoch": 2.79, "grad_norm": 17.68942642211914, "learning_rate": 2.4950686598051885e-07, "loss": 0.8204, "step": 17348 }, { "epoch": 2.79, "grad_norm": 13.552144050598145, "learning_rate": 2.491211994405696e-07, "loss": 0.7456, "step": 17349 }, { "epoch": 2.79, "grad_norm": 12.034585952758789, "learning_rate": 2.487358274369267e-07, "loss": 0.692, "step": 17350 }, { "epoch": 2.79, "grad_norm": 12.946642875671387, "learning_rate": 2.48350749981231e-07, "loss": 0.7059, "step": 17351 }, { "epoch": 2.79, "grad_norm": 14.961898803710938, "learning_rate": 2.479659670851153e-07, "loss": 1.0384, "step": 17352 }, { "epoch": 2.79, "grad_norm": 13.15845012664795, "learning_rate": 2.4758147876020266e-07, "loss": 0.6227, "step": 17353 }, { "epoch": 2.79, "grad_norm": 12.787467002868652, "learning_rate": 2.471972850181048e-07, "loss": 0.5941, "step": 17354 }, { "epoch": 2.79, "grad_norm": 13.47130012512207, "learning_rate": 2.4681338587042913e-07, "loss": 0.6516, "step": 17355 }, { "epoch": 2.79, "grad_norm": 13.987688064575195, "learning_rate": 2.464297813287697e-07, "loss": 0.7372, "step": 17356 }, { "epoch": 2.79, "grad_norm": 14.867541313171387, "learning_rate": 2.4604647140471506e-07, "loss": 0.9259, "step": 17357 }, { "epoch": 2.79, "grad_norm": 13.49594497680664, "learning_rate": 2.4566345610984475e-07, "loss": 0.7233, "step": 17358 }, { "epoch": 2.79, "grad_norm": 13.991541862487793, "learning_rate": 2.4528073545572406e-07, "loss": 0.5963, "step": 17359 }, { "epoch": 2.79, "grad_norm": 12.994969367980957, "learning_rate": 2.448983094539181e-07, "loss": 0.7861, "step": 17360 }, { "epoch": 2.79, "grad_norm": 14.626374244689941, "learning_rate": 2.445161781159755e-07, "loss": 0.7026, "step": 17361 }, { "epoch": 2.79, "grad_norm": 14.182730674743652, "learning_rate": 2.4413434145343916e-07, "loss": 0.5818, "step": 17362 }, { "epoch": 2.8, "grad_norm": 11.203014373779297, "learning_rate": 2.4375279947784436e-07, "loss": 0.5978, "step": 17363 }, { "epoch": 2.8, "grad_norm": 13.193748474121094, "learning_rate": 2.4337155220071516e-07, "loss": 0.8529, "step": 17364 }, { "epoch": 2.8, "grad_norm": 21.203937530517578, "learning_rate": 2.42990599633568e-07, "loss": 0.7926, "step": 17365 }, { "epoch": 2.8, "grad_norm": 13.094276428222656, "learning_rate": 2.426099417879102e-07, "loss": 0.8643, "step": 17366 }, { "epoch": 2.8, "grad_norm": 13.03530502319336, "learning_rate": 2.422295786752371e-07, "loss": 0.8683, "step": 17367 }, { "epoch": 2.8, "grad_norm": 14.008289337158203, "learning_rate": 2.4184951030704286e-07, "loss": 0.7397, "step": 17368 }, { "epoch": 2.8, "grad_norm": 13.035429954528809, "learning_rate": 2.4146973669480376e-07, "loss": 0.8453, "step": 17369 }, { "epoch": 2.8, "grad_norm": 10.601617813110352, "learning_rate": 2.410902578499919e-07, "loss": 0.6086, "step": 17370 }, { "epoch": 2.8, "grad_norm": 11.979691505432129, "learning_rate": 2.4071107378407343e-07, "loss": 0.6924, "step": 17371 }, { "epoch": 2.8, "grad_norm": 12.822272300720215, "learning_rate": 2.4033218450849825e-07, "loss": 0.8067, "step": 17372 }, { "epoch": 2.8, "grad_norm": 14.788488388061523, "learning_rate": 2.3995359003471165e-07, "loss": 0.6506, "step": 17373 }, { "epoch": 2.8, "grad_norm": 12.876717567443848, "learning_rate": 2.3957529037415107e-07, "loss": 1.0512, "step": 17374 }, { "epoch": 2.8, "grad_norm": 10.975702285766602, "learning_rate": 2.391972855382418e-07, "loss": 0.6017, "step": 17375 }, { "epoch": 2.8, "grad_norm": 11.88399600982666, "learning_rate": 2.3881957553840375e-07, "loss": 0.6894, "step": 17376 }, { "epoch": 2.8, "grad_norm": 13.822021484375, "learning_rate": 2.384421603860443e-07, "loss": 0.8233, "step": 17377 }, { "epoch": 2.8, "grad_norm": 12.75015640258789, "learning_rate": 2.3806504009256436e-07, "loss": 0.6346, "step": 17378 }, { "epoch": 2.8, "grad_norm": 13.136469841003418, "learning_rate": 2.37688214669356e-07, "loss": 0.6812, "step": 17379 }, { "epoch": 2.8, "grad_norm": 11.344566345214844, "learning_rate": 2.3731168412779892e-07, "loss": 0.7058, "step": 17380 }, { "epoch": 2.8, "grad_norm": 15.274054527282715, "learning_rate": 2.3693544847927076e-07, "loss": 0.8158, "step": 17381 }, { "epoch": 2.8, "grad_norm": 11.830816268920898, "learning_rate": 2.3655950773513458e-07, "loss": 0.6995, "step": 17382 }, { "epoch": 2.8, "grad_norm": 11.545465469360352, "learning_rate": 2.3618386190674357e-07, "loss": 0.7217, "step": 17383 }, { "epoch": 2.8, "grad_norm": 12.1463041305542, "learning_rate": 2.3580851100544755e-07, "loss": 0.8027, "step": 17384 }, { "epoch": 2.8, "grad_norm": 11.951529502868652, "learning_rate": 2.3543345504258408e-07, "loss": 0.8684, "step": 17385 }, { "epoch": 2.8, "grad_norm": 11.413716316223145, "learning_rate": 2.3505869402948079e-07, "loss": 0.6695, "step": 17386 }, { "epoch": 2.8, "grad_norm": 13.815349578857422, "learning_rate": 2.346842279774586e-07, "loss": 0.6487, "step": 17387 }, { "epoch": 2.8, "grad_norm": 13.963998794555664, "learning_rate": 2.3431005689782737e-07, "loss": 0.914, "step": 17388 }, { "epoch": 2.8, "grad_norm": 14.446622848510742, "learning_rate": 2.339361808018925e-07, "loss": 0.7714, "step": 17389 }, { "epoch": 2.8, "grad_norm": 12.201531410217285, "learning_rate": 2.3356259970094386e-07, "loss": 0.8075, "step": 17390 }, { "epoch": 2.8, "grad_norm": 14.552314758300781, "learning_rate": 2.3318931360626794e-07, "loss": 0.7746, "step": 17391 }, { "epoch": 2.8, "grad_norm": 14.716663360595703, "learning_rate": 2.3281632252914023e-07, "loss": 0.7013, "step": 17392 }, { "epoch": 2.8, "grad_norm": 11.053814888000488, "learning_rate": 2.3244362648082607e-07, "loss": 0.658, "step": 17393 }, { "epoch": 2.8, "grad_norm": 16.412216186523438, "learning_rate": 2.320712254725843e-07, "loss": 0.7929, "step": 17394 }, { "epoch": 2.8, "grad_norm": 13.274374961853027, "learning_rate": 2.316991195156637e-07, "loss": 0.6545, "step": 17395 }, { "epoch": 2.8, "grad_norm": 12.089221954345703, "learning_rate": 2.3132730862130414e-07, "loss": 0.4912, "step": 17396 }, { "epoch": 2.8, "grad_norm": 12.310432434082031, "learning_rate": 2.3095579280073442e-07, "loss": 0.7819, "step": 17397 }, { "epoch": 2.8, "grad_norm": 12.690649032592773, "learning_rate": 2.3058457206518004e-07, "loss": 0.7343, "step": 17398 }, { "epoch": 2.8, "grad_norm": 11.06368350982666, "learning_rate": 2.3021364642585197e-07, "loss": 0.7255, "step": 17399 }, { "epoch": 2.8, "grad_norm": 14.459490776062012, "learning_rate": 2.2984301589395574e-07, "loss": 0.7059, "step": 17400 }, { "epoch": 2.8, "grad_norm": 12.103425979614258, "learning_rate": 2.2947268048068573e-07, "loss": 0.8568, "step": 17401 }, { "epoch": 2.8, "grad_norm": 11.416049003601074, "learning_rate": 2.291026401972274e-07, "loss": 0.7388, "step": 17402 }, { "epoch": 2.8, "grad_norm": 10.959258079528809, "learning_rate": 2.287328950547607e-07, "loss": 0.6767, "step": 17403 }, { "epoch": 2.8, "grad_norm": 12.261178016662598, "learning_rate": 2.283634450644523e-07, "loss": 0.6956, "step": 17404 }, { "epoch": 2.8, "grad_norm": 15.3592529296875, "learning_rate": 2.279942902374621e-07, "loss": 0.5903, "step": 17405 }, { "epoch": 2.8, "grad_norm": 11.816207885742188, "learning_rate": 2.2762543058494234e-07, "loss": 0.5948, "step": 17406 }, { "epoch": 2.8, "grad_norm": 15.678462028503418, "learning_rate": 2.272568661180319e-07, "loss": 0.753, "step": 17407 }, { "epoch": 2.8, "grad_norm": 13.733558654785156, "learning_rate": 2.2688859684786736e-07, "loss": 0.734, "step": 17408 }, { "epoch": 2.8, "grad_norm": 23.946313858032227, "learning_rate": 2.2652062278556985e-07, "loss": 0.5818, "step": 17409 }, { "epoch": 2.8, "grad_norm": 13.924897193908691, "learning_rate": 2.2615294394225386e-07, "loss": 0.434, "step": 17410 }, { "epoch": 2.8, "grad_norm": 13.510436058044434, "learning_rate": 2.257855603290293e-07, "loss": 0.7014, "step": 17411 }, { "epoch": 2.8, "grad_norm": 16.518234252929688, "learning_rate": 2.2541847195699074e-07, "loss": 0.6889, "step": 17412 }, { "epoch": 2.8, "grad_norm": 13.086050987243652, "learning_rate": 2.2505167883722589e-07, "loss": 0.5511, "step": 17413 }, { "epoch": 2.8, "grad_norm": 14.600798606872559, "learning_rate": 2.2468518098081703e-07, "loss": 0.6552, "step": 17414 }, { "epoch": 2.8, "grad_norm": 12.52357292175293, "learning_rate": 2.2431897839883199e-07, "loss": 0.7334, "step": 17415 }, { "epoch": 2.8, "grad_norm": 13.711094856262207, "learning_rate": 2.239530711023319e-07, "loss": 0.5844, "step": 17416 }, { "epoch": 2.8, "grad_norm": 15.708949089050293, "learning_rate": 2.2358745910237235e-07, "loss": 0.7268, "step": 17417 }, { "epoch": 2.8, "grad_norm": 11.175263404846191, "learning_rate": 2.2322214240999452e-07, "loss": 0.7637, "step": 17418 }, { "epoch": 2.8, "grad_norm": 16.165159225463867, "learning_rate": 2.2285712103623402e-07, "loss": 0.9887, "step": 17419 }, { "epoch": 2.8, "grad_norm": 12.560385704040527, "learning_rate": 2.224923949921176e-07, "loss": 0.7281, "step": 17420 }, { "epoch": 2.8, "grad_norm": 11.572635650634766, "learning_rate": 2.221279642886609e-07, "loss": 0.7081, "step": 17421 }, { "epoch": 2.8, "grad_norm": 10.77751636505127, "learning_rate": 2.2176382893687287e-07, "loss": 0.5052, "step": 17422 }, { "epoch": 2.8, "grad_norm": 12.597639083862305, "learning_rate": 2.2139998894775138e-07, "loss": 0.5046, "step": 17423 }, { "epoch": 2.8, "grad_norm": 15.013311386108398, "learning_rate": 2.2103644433228765e-07, "loss": 0.6915, "step": 17424 }, { "epoch": 2.81, "grad_norm": 11.31260871887207, "learning_rate": 2.20673195101464e-07, "loss": 0.5869, "step": 17425 }, { "epoch": 2.81, "grad_norm": 15.781749725341797, "learning_rate": 2.2031024126624945e-07, "loss": 0.718, "step": 17426 }, { "epoch": 2.81, "grad_norm": 13.474342346191406, "learning_rate": 2.1994758283761187e-07, "loss": 0.6508, "step": 17427 }, { "epoch": 2.81, "grad_norm": 12.795770645141602, "learning_rate": 2.195852198265025e-07, "loss": 0.6095, "step": 17428 }, { "epoch": 2.81, "grad_norm": 13.548770904541016, "learning_rate": 2.1922315224386703e-07, "loss": 0.7527, "step": 17429 }, { "epoch": 2.81, "grad_norm": 12.466227531433105, "learning_rate": 2.188613801006434e-07, "loss": 0.7138, "step": 17430 }, { "epoch": 2.81, "grad_norm": 9.65539264678955, "learning_rate": 2.184999034077584e-07, "loss": 0.5361, "step": 17431 }, { "epoch": 2.81, "grad_norm": 14.216897964477539, "learning_rate": 2.1813872217613218e-07, "loss": 0.8598, "step": 17432 }, { "epoch": 2.81, "grad_norm": 13.241607666015625, "learning_rate": 2.1777783641667382e-07, "loss": 0.617, "step": 17433 }, { "epoch": 2.81, "grad_norm": 11.742177963256836, "learning_rate": 2.1741724614028347e-07, "loss": 0.8402, "step": 17434 }, { "epoch": 2.81, "grad_norm": 11.278752326965332, "learning_rate": 2.1705695135785466e-07, "loss": 0.4665, "step": 17435 }, { "epoch": 2.81, "grad_norm": 12.51927375793457, "learning_rate": 2.1669695208027086e-07, "loss": 0.6566, "step": 17436 }, { "epoch": 2.81, "grad_norm": 14.500905990600586, "learning_rate": 2.163372483184034e-07, "loss": 0.6845, "step": 17437 }, { "epoch": 2.81, "grad_norm": 17.56340789794922, "learning_rate": 2.1597784008312027e-07, "loss": 0.837, "step": 17438 }, { "epoch": 2.81, "grad_norm": 11.608621597290039, "learning_rate": 2.1561872738527612e-07, "loss": 0.5749, "step": 17439 }, { "epoch": 2.81, "grad_norm": 13.650694847106934, "learning_rate": 2.152599102357189e-07, "loss": 0.662, "step": 17440 }, { "epoch": 2.81, "grad_norm": 14.007797241210938, "learning_rate": 2.1490138864528887e-07, "loss": 0.8448, "step": 17441 }, { "epoch": 2.81, "grad_norm": 11.584637641906738, "learning_rate": 2.1454316262481179e-07, "loss": 0.7705, "step": 17442 }, { "epoch": 2.81, "grad_norm": 12.441326141357422, "learning_rate": 2.1418523218511233e-07, "loss": 0.5604, "step": 17443 }, { "epoch": 2.81, "grad_norm": 11.100163459777832, "learning_rate": 2.1382759733699965e-07, "loss": 0.746, "step": 17444 }, { "epoch": 2.81, "grad_norm": 11.169844627380371, "learning_rate": 2.134702580912773e-07, "loss": 0.6561, "step": 17445 }, { "epoch": 2.81, "grad_norm": 14.037781715393066, "learning_rate": 2.131132144587389e-07, "loss": 0.7442, "step": 17446 }, { "epoch": 2.81, "grad_norm": 12.43049144744873, "learning_rate": 2.1275646645016913e-07, "loss": 0.7393, "step": 17447 }, { "epoch": 2.81, "grad_norm": 12.381129264831543, "learning_rate": 2.1240001407634493e-07, "loss": 0.8027, "step": 17448 }, { "epoch": 2.81, "grad_norm": 15.942267417907715, "learning_rate": 2.1204385734803324e-07, "loss": 0.7784, "step": 17449 }, { "epoch": 2.81, "grad_norm": 11.31628131866455, "learning_rate": 2.116879962759899e-07, "loss": 0.4382, "step": 17450 }, { "epoch": 2.81, "grad_norm": 13.980582237243652, "learning_rate": 2.1133243087096634e-07, "loss": 0.8161, "step": 17451 }, { "epoch": 2.81, "grad_norm": 12.552541732788086, "learning_rate": 2.1097716114370282e-07, "loss": 0.6871, "step": 17452 }, { "epoch": 2.81, "grad_norm": 10.546614646911621, "learning_rate": 2.1062218710492966e-07, "loss": 0.6282, "step": 17453 }, { "epoch": 2.81, "grad_norm": 12.673643112182617, "learning_rate": 2.102675087653705e-07, "loss": 0.6467, "step": 17454 }, { "epoch": 2.81, "grad_norm": 11.300934791564941, "learning_rate": 2.099131261357379e-07, "loss": 0.5678, "step": 17455 }, { "epoch": 2.81, "grad_norm": 11.697251319885254, "learning_rate": 2.095590392267366e-07, "loss": 0.5701, "step": 17456 }, { "epoch": 2.81, "grad_norm": 10.63065242767334, "learning_rate": 2.092052480490625e-07, "loss": 0.6513, "step": 17457 }, { "epoch": 2.81, "grad_norm": 12.293802261352539, "learning_rate": 2.088517526134004e-07, "loss": 0.5061, "step": 17458 }, { "epoch": 2.81, "grad_norm": 13.09984302520752, "learning_rate": 2.0849855293043063e-07, "loss": 0.7487, "step": 17459 }, { "epoch": 2.81, "grad_norm": 13.84980297088623, "learning_rate": 2.0814564901082134e-07, "loss": 0.8107, "step": 17460 }, { "epoch": 2.81, "grad_norm": 11.772089958190918, "learning_rate": 2.0779304086523066e-07, "loss": 0.6601, "step": 17461 }, { "epoch": 2.81, "grad_norm": 13.010693550109863, "learning_rate": 2.074407285043123e-07, "loss": 0.8928, "step": 17462 }, { "epoch": 2.81, "grad_norm": 16.715740203857422, "learning_rate": 2.070887119387066e-07, "loss": 0.7999, "step": 17463 }, { "epoch": 2.81, "grad_norm": 14.855914115905762, "learning_rate": 2.067369911790451e-07, "loss": 0.8745, "step": 17464 }, { "epoch": 2.81, "grad_norm": 13.311135292053223, "learning_rate": 2.0638556623595484e-07, "loss": 0.7372, "step": 17465 }, { "epoch": 2.81, "grad_norm": 11.760697364807129, "learning_rate": 2.0603443712004956e-07, "loss": 0.8894, "step": 17466 }, { "epoch": 2.81, "grad_norm": 13.466821670532227, "learning_rate": 2.0568360384193632e-07, "loss": 0.8522, "step": 17467 }, { "epoch": 2.81, "grad_norm": 14.580968856811523, "learning_rate": 2.0533306641221108e-07, "loss": 0.8849, "step": 17468 }, { "epoch": 2.81, "grad_norm": 14.355229377746582, "learning_rate": 2.0498282484146315e-07, "loss": 0.6658, "step": 17469 }, { "epoch": 2.81, "grad_norm": 13.364330291748047, "learning_rate": 2.0463287914027185e-07, "loss": 0.5404, "step": 17470 }, { "epoch": 2.81, "grad_norm": 12.14380931854248, "learning_rate": 2.0428322931920652e-07, "loss": 0.5341, "step": 17471 }, { "epoch": 2.81, "grad_norm": 11.929542541503906, "learning_rate": 2.0393387538883093e-07, "loss": 0.8604, "step": 17472 }, { "epoch": 2.81, "grad_norm": 11.861749649047852, "learning_rate": 2.035848173596966e-07, "loss": 0.6996, "step": 17473 }, { "epoch": 2.81, "grad_norm": 16.28923797607422, "learning_rate": 2.0323605524234734e-07, "loss": 0.8976, "step": 17474 }, { "epoch": 2.81, "grad_norm": 17.244274139404297, "learning_rate": 2.0288758904731698e-07, "loss": 0.7197, "step": 17475 }, { "epoch": 2.81, "grad_norm": 11.564323425292969, "learning_rate": 2.0253941878513373e-07, "loss": 0.8171, "step": 17476 }, { "epoch": 2.81, "grad_norm": 11.676496505737305, "learning_rate": 2.0219154446631139e-07, "loss": 0.7695, "step": 17477 }, { "epoch": 2.81, "grad_norm": 11.77883243560791, "learning_rate": 2.018439661013605e-07, "loss": 0.7636, "step": 17478 }, { "epoch": 2.81, "grad_norm": 13.312015533447266, "learning_rate": 2.0149668370077813e-07, "loss": 0.8488, "step": 17479 }, { "epoch": 2.81, "grad_norm": 13.703601837158203, "learning_rate": 2.0114969727505485e-07, "loss": 0.736, "step": 17480 }, { "epoch": 2.81, "grad_norm": 13.218164443969727, "learning_rate": 2.0080300683467336e-07, "loss": 0.7872, "step": 17481 }, { "epoch": 2.81, "grad_norm": 13.52225399017334, "learning_rate": 2.0045661239010418e-07, "loss": 0.9245, "step": 17482 }, { "epoch": 2.81, "grad_norm": 11.424899101257324, "learning_rate": 2.0011051395181001e-07, "loss": 0.8542, "step": 17483 }, { "epoch": 2.81, "grad_norm": 15.905137062072754, "learning_rate": 1.9976471153024702e-07, "loss": 0.6872, "step": 17484 }, { "epoch": 2.81, "grad_norm": 11.92124080657959, "learning_rate": 1.99419205135859e-07, "loss": 0.4767, "step": 17485 }, { "epoch": 2.81, "grad_norm": 12.525111198425293, "learning_rate": 1.9907399477908317e-07, "loss": 0.7446, "step": 17486 }, { "epoch": 2.82, "grad_norm": 15.924829483032227, "learning_rate": 1.9872908047034679e-07, "loss": 0.6537, "step": 17487 }, { "epoch": 2.82, "grad_norm": 10.778554916381836, "learning_rate": 1.9838446222006813e-07, "loss": 0.7631, "step": 17488 }, { "epoch": 2.82, "grad_norm": 11.344083786010742, "learning_rate": 1.9804014003865779e-07, "loss": 0.8182, "step": 17489 }, { "epoch": 2.82, "grad_norm": 14.251593589782715, "learning_rate": 1.9769611393651521e-07, "loss": 0.7707, "step": 17490 }, { "epoch": 2.82, "grad_norm": 14.289315223693848, "learning_rate": 1.973523839240321e-07, "loss": 0.7102, "step": 17491 }, { "epoch": 2.82, "grad_norm": 13.06806755065918, "learning_rate": 1.9700895001159238e-07, "loss": 0.7168, "step": 17492 }, { "epoch": 2.82, "grad_norm": 12.999747276306152, "learning_rate": 1.9666581220956882e-07, "loss": 0.4814, "step": 17493 }, { "epoch": 2.82, "grad_norm": 14.246321678161621, "learning_rate": 1.963229705283265e-07, "loss": 0.7311, "step": 17494 }, { "epoch": 2.82, "grad_norm": 14.283530235290527, "learning_rate": 1.9598042497822156e-07, "loss": 0.59, "step": 17495 }, { "epoch": 2.82, "grad_norm": 12.029518127441406, "learning_rate": 1.956381755696002e-07, "loss": 0.7329, "step": 17496 }, { "epoch": 2.82, "grad_norm": 11.319716453552246, "learning_rate": 1.9529622231280187e-07, "loss": 0.6614, "step": 17497 }, { "epoch": 2.82, "grad_norm": 15.759984016418457, "learning_rate": 1.9495456521815503e-07, "loss": 0.7161, "step": 17498 }, { "epoch": 2.82, "grad_norm": 15.35332202911377, "learning_rate": 1.946132042959792e-07, "loss": 0.5545, "step": 17499 }, { "epoch": 2.82, "grad_norm": 12.084650993347168, "learning_rate": 1.9427213955658607e-07, "loss": 0.7027, "step": 17500 }, { "epoch": 2.82, "grad_norm": 14.781041145324707, "learning_rate": 1.939313710102775e-07, "loss": 0.6102, "step": 17501 }, { "epoch": 2.82, "grad_norm": 12.421965599060059, "learning_rate": 1.9359089866734738e-07, "loss": 0.6443, "step": 17502 }, { "epoch": 2.82, "grad_norm": 13.13315486907959, "learning_rate": 1.9325072253808087e-07, "loss": 0.7711, "step": 17503 }, { "epoch": 2.82, "grad_norm": 14.71030044555664, "learning_rate": 1.9291084263275083e-07, "loss": 0.7031, "step": 17504 }, { "epoch": 2.82, "grad_norm": 14.404196739196777, "learning_rate": 1.9257125896162686e-07, "loss": 0.6578, "step": 17505 }, { "epoch": 2.82, "grad_norm": 14.925909996032715, "learning_rate": 1.922319715349641e-07, "loss": 0.7546, "step": 17506 }, { "epoch": 2.82, "grad_norm": 13.057920455932617, "learning_rate": 1.9189298036301097e-07, "loss": 0.5744, "step": 17507 }, { "epoch": 2.82, "grad_norm": 13.48172378540039, "learning_rate": 1.9155428545600818e-07, "loss": 0.5206, "step": 17508 }, { "epoch": 2.82, "grad_norm": 10.52629566192627, "learning_rate": 1.9121588682418645e-07, "loss": 0.5366, "step": 17509 }, { "epoch": 2.82, "grad_norm": 13.839326858520508, "learning_rate": 1.9087778447776762e-07, "loss": 0.6273, "step": 17510 }, { "epoch": 2.82, "grad_norm": 15.605077743530273, "learning_rate": 1.9053997842696348e-07, "loss": 0.6336, "step": 17511 }, { "epoch": 2.82, "grad_norm": 15.717698097229004, "learning_rate": 1.9020246868197812e-07, "loss": 0.6859, "step": 17512 }, { "epoch": 2.82, "grad_norm": 15.17597770690918, "learning_rate": 1.898652552530067e-07, "loss": 0.6802, "step": 17513 }, { "epoch": 2.82, "grad_norm": 14.953091621398926, "learning_rate": 1.8952833815023663e-07, "loss": 0.8206, "step": 17514 }, { "epoch": 2.82, "grad_norm": 18.964597702026367, "learning_rate": 1.89191717383842e-07, "loss": 0.6392, "step": 17515 }, { "epoch": 2.82, "grad_norm": 13.496438026428223, "learning_rate": 1.8885539296399247e-07, "loss": 0.8674, "step": 17516 }, { "epoch": 2.82, "grad_norm": 10.894965171813965, "learning_rate": 1.8851936490084656e-07, "loss": 0.7336, "step": 17517 }, { "epoch": 2.82, "grad_norm": 11.571937561035156, "learning_rate": 1.8818363320455502e-07, "loss": 0.6865, "step": 17518 }, { "epoch": 2.82, "grad_norm": 12.624906539916992, "learning_rate": 1.8784819788525865e-07, "loss": 0.6818, "step": 17519 }, { "epoch": 2.82, "grad_norm": 10.141417503356934, "learning_rate": 1.8751305895308824e-07, "loss": 0.562, "step": 17520 }, { "epoch": 2.82, "grad_norm": 18.577709197998047, "learning_rate": 1.8717821641817013e-07, "loss": 0.6025, "step": 17521 }, { "epoch": 2.82, "grad_norm": 12.601417541503906, "learning_rate": 1.8684367029061734e-07, "loss": 0.5709, "step": 17522 }, { "epoch": 2.82, "grad_norm": 11.380352020263672, "learning_rate": 1.8650942058053288e-07, "loss": 0.7173, "step": 17523 }, { "epoch": 2.82, "grad_norm": 14.163986206054688, "learning_rate": 1.861754672980165e-07, "loss": 0.6361, "step": 17524 }, { "epoch": 2.82, "grad_norm": 13.250154495239258, "learning_rate": 1.858418104531523e-07, "loss": 0.7234, "step": 17525 }, { "epoch": 2.82, "grad_norm": 12.803732872009277, "learning_rate": 1.8550845005602224e-07, "loss": 0.7248, "step": 17526 }, { "epoch": 2.82, "grad_norm": 19.793428421020508, "learning_rate": 1.8517538611669495e-07, "loss": 0.7228, "step": 17527 }, { "epoch": 2.82, "grad_norm": 14.53492259979248, "learning_rate": 1.8484261864522902e-07, "loss": 0.6653, "step": 17528 }, { "epoch": 2.82, "grad_norm": 12.362724304199219, "learning_rate": 1.845101476516775e-07, "loss": 0.7532, "step": 17529 }, { "epoch": 2.82, "grad_norm": 14.14944076538086, "learning_rate": 1.8417797314608354e-07, "loss": 0.5726, "step": 17530 }, { "epoch": 2.82, "grad_norm": 16.29692840576172, "learning_rate": 1.8384609513847905e-07, "loss": 0.7418, "step": 17531 }, { "epoch": 2.82, "grad_norm": 12.714656829833984, "learning_rate": 1.835145136388894e-07, "loss": 0.5433, "step": 17532 }, { "epoch": 2.82, "grad_norm": 12.322111129760742, "learning_rate": 1.8318322865733207e-07, "loss": 0.723, "step": 17533 }, { "epoch": 2.82, "grad_norm": 15.154440879821777, "learning_rate": 1.8285224020381243e-07, "loss": 0.7007, "step": 17534 }, { "epoch": 2.82, "grad_norm": 12.341790199279785, "learning_rate": 1.8252154828832914e-07, "loss": 0.6909, "step": 17535 }, { "epoch": 2.82, "grad_norm": 16.981687545776367, "learning_rate": 1.8219115292086975e-07, "loss": 0.6769, "step": 17536 }, { "epoch": 2.82, "grad_norm": 12.26753044128418, "learning_rate": 1.818610541114163e-07, "loss": 0.758, "step": 17537 }, { "epoch": 2.82, "grad_norm": 13.750948905944824, "learning_rate": 1.815312518699386e-07, "loss": 0.7609, "step": 17538 }, { "epoch": 2.82, "grad_norm": 15.08204460144043, "learning_rate": 1.812017462063975e-07, "loss": 0.5046, "step": 17539 }, { "epoch": 2.82, "grad_norm": 13.291589736938477, "learning_rate": 1.8087253713074848e-07, "loss": 0.8109, "step": 17540 }, { "epoch": 2.82, "grad_norm": 11.817320823669434, "learning_rate": 1.8054362465293462e-07, "loss": 0.8359, "step": 17541 }, { "epoch": 2.82, "grad_norm": 13.967153549194336, "learning_rate": 1.802150087828891e-07, "loss": 0.6496, "step": 17542 }, { "epoch": 2.82, "grad_norm": 17.133377075195312, "learning_rate": 1.7988668953054177e-07, "loss": 0.7476, "step": 17543 }, { "epoch": 2.82, "grad_norm": 14.580260276794434, "learning_rate": 1.7955866690580693e-07, "loss": 0.7605, "step": 17544 }, { "epoch": 2.82, "grad_norm": 15.400751113891602, "learning_rate": 1.792309409185955e-07, "loss": 0.7762, "step": 17545 }, { "epoch": 2.82, "grad_norm": 13.54776668548584, "learning_rate": 1.7890351157880514e-07, "loss": 0.7767, "step": 17546 }, { "epoch": 2.82, "grad_norm": 12.25486946105957, "learning_rate": 1.7857637889632462e-07, "loss": 0.747, "step": 17547 }, { "epoch": 2.82, "grad_norm": 13.594804763793945, "learning_rate": 1.7824954288103934e-07, "loss": 0.6226, "step": 17548 }, { "epoch": 2.83, "grad_norm": 15.193601608276367, "learning_rate": 1.7792300354281922e-07, "loss": 0.7585, "step": 17549 }, { "epoch": 2.83, "grad_norm": 12.629776954650879, "learning_rate": 1.7759676089152743e-07, "loss": 0.8993, "step": 17550 }, { "epoch": 2.83, "grad_norm": 10.80228328704834, "learning_rate": 1.7727081493701946e-07, "loss": 0.71, "step": 17551 }, { "epoch": 2.83, "grad_norm": 14.358522415161133, "learning_rate": 1.7694516568913967e-07, "loss": 0.7246, "step": 17552 }, { "epoch": 2.83, "grad_norm": 11.738210678100586, "learning_rate": 1.766198131577268e-07, "loss": 0.6436, "step": 17553 }, { "epoch": 2.83, "grad_norm": 11.330829620361328, "learning_rate": 1.7629475735260637e-07, "loss": 0.4767, "step": 17554 }, { "epoch": 2.83, "grad_norm": 14.248316764831543, "learning_rate": 1.7596999828359828e-07, "loss": 0.8537, "step": 17555 }, { "epoch": 2.83, "grad_norm": 12.277316093444824, "learning_rate": 1.7564553596051248e-07, "loss": 0.8104, "step": 17556 }, { "epoch": 2.83, "grad_norm": 13.876946449279785, "learning_rate": 1.7532137039315e-07, "loss": 0.5294, "step": 17557 }, { "epoch": 2.83, "grad_norm": 13.478476524353027, "learning_rate": 1.7499750159129968e-07, "loss": 0.8368, "step": 17558 }, { "epoch": 2.83, "grad_norm": 13.961743354797363, "learning_rate": 1.7467392956474815e-07, "loss": 1.012, "step": 17559 }, { "epoch": 2.83, "grad_norm": 12.709190368652344, "learning_rate": 1.743506543232676e-07, "loss": 0.7155, "step": 17560 }, { "epoch": 2.83, "grad_norm": 11.460631370544434, "learning_rate": 1.740276758766224e-07, "loss": 0.599, "step": 17561 }, { "epoch": 2.83, "grad_norm": 12.561744689941406, "learning_rate": 1.7370499423456922e-07, "loss": 0.5696, "step": 17562 }, { "epoch": 2.83, "grad_norm": 14.097164154052734, "learning_rate": 1.7338260940685358e-07, "loss": 0.6874, "step": 17563 }, { "epoch": 2.83, "grad_norm": 12.497209548950195, "learning_rate": 1.7306052140321661e-07, "loss": 0.7902, "step": 17564 }, { "epoch": 2.83, "grad_norm": 11.881738662719727, "learning_rate": 1.7273873023338494e-07, "loss": 0.6987, "step": 17565 }, { "epoch": 2.83, "grad_norm": 13.453486442565918, "learning_rate": 1.724172359070786e-07, "loss": 0.7338, "step": 17566 }, { "epoch": 2.83, "grad_norm": 12.613449096679688, "learning_rate": 1.720960384340098e-07, "loss": 0.7695, "step": 17567 }, { "epoch": 2.83, "grad_norm": 13.523924827575684, "learning_rate": 1.7177513782387857e-07, "loss": 0.7763, "step": 17568 }, { "epoch": 2.83, "grad_norm": 12.7627592086792, "learning_rate": 1.7145453408638158e-07, "loss": 0.6536, "step": 17569 }, { "epoch": 2.83, "grad_norm": 11.780336380004883, "learning_rate": 1.7113422723119998e-07, "loss": 0.7441, "step": 17570 }, { "epoch": 2.83, "grad_norm": 12.159027099609375, "learning_rate": 1.7081421726801052e-07, "loss": 0.8949, "step": 17571 }, { "epoch": 2.83, "grad_norm": 12.881075859069824, "learning_rate": 1.7049450420647873e-07, "loss": 0.7644, "step": 17572 }, { "epoch": 2.83, "grad_norm": 15.710997581481934, "learning_rate": 1.7017508805626139e-07, "loss": 0.8969, "step": 17573 }, { "epoch": 2.83, "grad_norm": 10.997550964355469, "learning_rate": 1.698559688270085e-07, "loss": 0.6872, "step": 17574 }, { "epoch": 2.83, "grad_norm": 10.185035705566406, "learning_rate": 1.6953714652835794e-07, "loss": 0.6883, "step": 17575 }, { "epoch": 2.83, "grad_norm": 12.734915733337402, "learning_rate": 1.6921862116994202e-07, "loss": 0.5954, "step": 17576 }, { "epoch": 2.83, "grad_norm": 12.267407417297363, "learning_rate": 1.689003927613797e-07, "loss": 0.7657, "step": 17577 }, { "epoch": 2.83, "grad_norm": 14.837028503417969, "learning_rate": 1.685824613122855e-07, "loss": 0.6753, "step": 17578 }, { "epoch": 2.83, "grad_norm": 12.529995918273926, "learning_rate": 1.6826482683226065e-07, "loss": 0.5862, "step": 17579 }, { "epoch": 2.83, "grad_norm": 13.202025413513184, "learning_rate": 1.6794748933090188e-07, "loss": 0.602, "step": 17580 }, { "epoch": 2.83, "grad_norm": 14.668063163757324, "learning_rate": 1.6763044881779378e-07, "loss": 0.8073, "step": 17581 }, { "epoch": 2.83, "grad_norm": 10.93523120880127, "learning_rate": 1.673137053025131e-07, "loss": 0.6622, "step": 17582 }, { "epoch": 2.83, "grad_norm": 12.644364356994629, "learning_rate": 1.6699725879462668e-07, "loss": 0.8002, "step": 17583 }, { "epoch": 2.83, "grad_norm": 12.253982543945312, "learning_rate": 1.6668110930369464e-07, "loss": 0.7981, "step": 17584 }, { "epoch": 2.83, "grad_norm": 12.019499778747559, "learning_rate": 1.6636525683926486e-07, "loss": 0.7138, "step": 17585 }, { "epoch": 2.83, "grad_norm": 11.653581619262695, "learning_rate": 1.6604970141087973e-07, "loss": 0.6444, "step": 17586 }, { "epoch": 2.83, "grad_norm": 13.864800453186035, "learning_rate": 1.657344430280683e-07, "loss": 0.7754, "step": 17587 }, { "epoch": 2.83, "grad_norm": 11.329835891723633, "learning_rate": 1.6541948170035737e-07, "loss": 0.5204, "step": 17588 }, { "epoch": 2.83, "grad_norm": 15.080527305603027, "learning_rate": 1.6510481743725714e-07, "loss": 0.8226, "step": 17589 }, { "epoch": 2.83, "grad_norm": 12.951338768005371, "learning_rate": 1.6479045024827444e-07, "loss": 0.5672, "step": 17590 }, { "epoch": 2.83, "grad_norm": 12.150445938110352, "learning_rate": 1.644763801429039e-07, "loss": 0.7147, "step": 17591 }, { "epoch": 2.83, "grad_norm": 14.236128807067871, "learning_rate": 1.6416260713063238e-07, "loss": 0.567, "step": 17592 }, { "epoch": 2.83, "grad_norm": 10.712925910949707, "learning_rate": 1.6384913122093893e-07, "loss": 0.5814, "step": 17593 }, { "epoch": 2.83, "grad_norm": 14.301984786987305, "learning_rate": 1.6353595242329156e-07, "loss": 0.7445, "step": 17594 }, { "epoch": 2.83, "grad_norm": 12.528240203857422, "learning_rate": 1.632230707471516e-07, "loss": 0.775, "step": 17595 }, { "epoch": 2.83, "grad_norm": 10.709465980529785, "learning_rate": 1.6291048620196702e-07, "loss": 0.5598, "step": 17596 }, { "epoch": 2.83, "grad_norm": 17.0640811920166, "learning_rate": 1.6259819879718142e-07, "loss": 0.7014, "step": 17597 }, { "epoch": 2.83, "grad_norm": 13.942523956298828, "learning_rate": 1.6228620854222831e-07, "loss": 0.6762, "step": 17598 }, { "epoch": 2.83, "grad_norm": 12.50727653503418, "learning_rate": 1.6197451544653243e-07, "loss": 0.572, "step": 17599 }, { "epoch": 2.83, "grad_norm": 12.921422958374023, "learning_rate": 1.616631195195073e-07, "loss": 0.7865, "step": 17600 }, { "epoch": 2.83, "grad_norm": 13.19313907623291, "learning_rate": 1.6135202077055767e-07, "loss": 0.4978, "step": 17601 }, { "epoch": 2.83, "grad_norm": 13.698925018310547, "learning_rate": 1.6104121920908378e-07, "loss": 0.7391, "step": 17602 }, { "epoch": 2.83, "grad_norm": 15.00839614868164, "learning_rate": 1.6073071484447145e-07, "loss": 0.763, "step": 17603 }, { "epoch": 2.83, "grad_norm": 14.31610107421875, "learning_rate": 1.6042050768610208e-07, "loss": 0.7056, "step": 17604 }, { "epoch": 2.83, "grad_norm": 13.431987762451172, "learning_rate": 1.601105977433437e-07, "loss": 0.8303, "step": 17605 }, { "epoch": 2.83, "grad_norm": 15.436760902404785, "learning_rate": 1.5980098502555775e-07, "loss": 0.7641, "step": 17606 }, { "epoch": 2.83, "grad_norm": 13.906438827514648, "learning_rate": 1.5949166954209672e-07, "loss": 0.6943, "step": 17607 }, { "epoch": 2.83, "grad_norm": 13.531106948852539, "learning_rate": 1.5918265130230538e-07, "loss": 0.7269, "step": 17608 }, { "epoch": 2.83, "grad_norm": 16.1612491607666, "learning_rate": 1.5887393031551512e-07, "loss": 0.8722, "step": 17609 }, { "epoch": 2.83, "grad_norm": 13.392038345336914, "learning_rate": 1.5856550659105297e-07, "loss": 0.9153, "step": 17610 }, { "epoch": 2.83, "grad_norm": 14.710379600524902, "learning_rate": 1.5825738013823479e-07, "loss": 0.9085, "step": 17611 }, { "epoch": 2.84, "grad_norm": 12.42878532409668, "learning_rate": 1.5794955096636866e-07, "loss": 0.7634, "step": 17612 }, { "epoch": 2.84, "grad_norm": 14.06384563446045, "learning_rate": 1.5764201908475162e-07, "loss": 0.6352, "step": 17613 }, { "epoch": 2.84, "grad_norm": 11.923492431640625, "learning_rate": 1.5733478450267293e-07, "loss": 0.7921, "step": 17614 }, { "epoch": 2.84, "grad_norm": 11.684235572814941, "learning_rate": 1.570278472294151e-07, "loss": 0.4722, "step": 17615 }, { "epoch": 2.84, "grad_norm": 10.950822830200195, "learning_rate": 1.5672120727424743e-07, "loss": 0.6687, "step": 17616 }, { "epoch": 2.84, "grad_norm": 17.943511962890625, "learning_rate": 1.5641486464643363e-07, "loss": 0.82, "step": 17617 }, { "epoch": 2.84, "grad_norm": 11.48311710357666, "learning_rate": 1.5610881935522626e-07, "loss": 0.6549, "step": 17618 }, { "epoch": 2.84, "grad_norm": 13.092803955078125, "learning_rate": 1.5580307140987017e-07, "loss": 0.7158, "step": 17619 }, { "epoch": 2.84, "grad_norm": 12.020448684692383, "learning_rate": 1.5549762081959906e-07, "loss": 0.7642, "step": 17620 }, { "epoch": 2.84, "grad_norm": 13.554622650146484, "learning_rate": 1.5519246759364336e-07, "loss": 0.7815, "step": 17621 }, { "epoch": 2.84, "grad_norm": 12.889756202697754, "learning_rate": 1.548876117412168e-07, "loss": 0.7076, "step": 17622 }, { "epoch": 2.84, "grad_norm": 12.830543518066406, "learning_rate": 1.545830532715298e-07, "loss": 0.8381, "step": 17623 }, { "epoch": 2.84, "grad_norm": 13.173347473144531, "learning_rate": 1.5427879219378162e-07, "loss": 0.924, "step": 17624 }, { "epoch": 2.84, "grad_norm": 14.670403480529785, "learning_rate": 1.5397482851716162e-07, "loss": 0.4613, "step": 17625 }, { "epoch": 2.84, "grad_norm": 14.91518497467041, "learning_rate": 1.5367116225085244e-07, "loss": 0.8067, "step": 17626 }, { "epoch": 2.84, "grad_norm": 16.233442306518555, "learning_rate": 1.5336779340402785e-07, "loss": 0.7352, "step": 17627 }, { "epoch": 2.84, "grad_norm": 18.46342658996582, "learning_rate": 1.5306472198584833e-07, "loss": 0.6865, "step": 17628 }, { "epoch": 2.84, "grad_norm": 12.406187057495117, "learning_rate": 1.5276194800547207e-07, "loss": 0.6558, "step": 17629 }, { "epoch": 2.84, "grad_norm": 14.097366333007812, "learning_rate": 1.5245947147204065e-07, "loss": 0.6343, "step": 17630 }, { "epoch": 2.84, "grad_norm": 16.852725982666016, "learning_rate": 1.5215729239469568e-07, "loss": 0.6598, "step": 17631 }, { "epoch": 2.84, "grad_norm": 15.814703941345215, "learning_rate": 1.5185541078256094e-07, "loss": 0.7037, "step": 17632 }, { "epoch": 2.84, "grad_norm": 12.545933723449707, "learning_rate": 1.5155382664475472e-07, "loss": 0.8907, "step": 17633 }, { "epoch": 2.84, "grad_norm": 12.070053100585938, "learning_rate": 1.512525399903908e-07, "loss": 0.5723, "step": 17634 }, { "epoch": 2.84, "grad_norm": 15.236196517944336, "learning_rate": 1.5095155082856638e-07, "loss": 0.6884, "step": 17635 }, { "epoch": 2.84, "grad_norm": 15.22330379486084, "learning_rate": 1.506508591683742e-07, "loss": 0.7164, "step": 17636 }, { "epoch": 2.84, "grad_norm": 10.747969627380371, "learning_rate": 1.5035046501889695e-07, "loss": 0.5505, "step": 17637 }, { "epoch": 2.84, "grad_norm": 12.365898132324219, "learning_rate": 1.500503683892085e-07, "loss": 0.5327, "step": 17638 }, { "epoch": 2.84, "grad_norm": 18.54802703857422, "learning_rate": 1.4975056928837385e-07, "loss": 0.9236, "step": 17639 }, { "epoch": 2.84, "grad_norm": 12.917338371276855, "learning_rate": 1.4945106772544792e-07, "loss": 0.6288, "step": 17640 }, { "epoch": 2.84, "grad_norm": 11.232304573059082, "learning_rate": 1.49151863709478e-07, "loss": 0.7715, "step": 17641 }, { "epoch": 2.84, "grad_norm": 15.095602035522461, "learning_rate": 1.4885295724950234e-07, "loss": 0.7721, "step": 17642 }, { "epoch": 2.84, "grad_norm": 13.709527969360352, "learning_rate": 1.485543483545493e-07, "loss": 0.8703, "step": 17643 }, { "epoch": 2.84, "grad_norm": 13.861236572265625, "learning_rate": 1.4825603703363722e-07, "loss": 0.6594, "step": 17644 }, { "epoch": 2.84, "grad_norm": 13.972373962402344, "learning_rate": 1.4795802329578003e-07, "loss": 0.9103, "step": 17645 }, { "epoch": 2.84, "grad_norm": 14.660730361938477, "learning_rate": 1.4766030714997715e-07, "loss": 0.6405, "step": 17646 }, { "epoch": 2.84, "grad_norm": 16.40887451171875, "learning_rate": 1.4736288860522252e-07, "loss": 0.7848, "step": 17647 }, { "epoch": 2.84, "grad_norm": 21.300575256347656, "learning_rate": 1.4706576767050006e-07, "loss": 0.6412, "step": 17648 }, { "epoch": 2.84, "grad_norm": 14.358390808105469, "learning_rate": 1.4676894435478374e-07, "loss": 0.6529, "step": 17649 }, { "epoch": 2.84, "grad_norm": 14.707022666931152, "learning_rate": 1.4647241866704077e-07, "loss": 0.6658, "step": 17650 }, { "epoch": 2.84, "grad_norm": 15.067825317382812, "learning_rate": 1.4617619061622735e-07, "loss": 0.821, "step": 17651 }, { "epoch": 2.84, "grad_norm": 14.547242164611816, "learning_rate": 1.4588026021128964e-07, "loss": 0.4561, "step": 17652 }, { "epoch": 2.84, "grad_norm": 17.460355758666992, "learning_rate": 1.455846274611694e-07, "loss": 0.7012, "step": 17653 }, { "epoch": 2.84, "grad_norm": 12.996641159057617, "learning_rate": 1.4528929237479506e-07, "loss": 0.7463, "step": 17654 }, { "epoch": 2.84, "grad_norm": 14.818375587463379, "learning_rate": 1.4499425496108832e-07, "loss": 0.8587, "step": 17655 }, { "epoch": 2.84, "grad_norm": 13.0093994140625, "learning_rate": 1.4469951522895987e-07, "loss": 0.9161, "step": 17656 }, { "epoch": 2.84, "grad_norm": 15.06141185760498, "learning_rate": 1.4440507318731366e-07, "loss": 0.6439, "step": 17657 }, { "epoch": 2.84, "grad_norm": 13.70244312286377, "learning_rate": 1.4411092884504373e-07, "loss": 0.739, "step": 17658 }, { "epoch": 2.84, "grad_norm": 10.544483184814453, "learning_rate": 1.438170822110352e-07, "loss": 0.498, "step": 17659 }, { "epoch": 2.84, "grad_norm": 12.216438293457031, "learning_rate": 1.4352353329416314e-07, "loss": 0.7436, "step": 17660 }, { "epoch": 2.84, "grad_norm": 12.489131927490234, "learning_rate": 1.4323028210329494e-07, "loss": 0.7943, "step": 17661 }, { "epoch": 2.84, "grad_norm": 13.389548301696777, "learning_rate": 1.4293732864728903e-07, "loss": 0.5466, "step": 17662 }, { "epoch": 2.84, "grad_norm": 13.27060604095459, "learning_rate": 1.4264467293499284e-07, "loss": 0.6124, "step": 17663 }, { "epoch": 2.84, "grad_norm": 13.381196022033691, "learning_rate": 1.4235231497524814e-07, "loss": 0.617, "step": 17664 }, { "epoch": 2.84, "grad_norm": 13.705243110656738, "learning_rate": 1.4206025477688569e-07, "loss": 0.808, "step": 17665 }, { "epoch": 2.84, "grad_norm": 16.939760208129883, "learning_rate": 1.4176849234872615e-07, "loss": 0.6274, "step": 17666 }, { "epoch": 2.84, "grad_norm": 14.952703475952148, "learning_rate": 1.4147702769958472e-07, "loss": 0.8294, "step": 17667 }, { "epoch": 2.84, "grad_norm": 13.248278617858887, "learning_rate": 1.4118586083826324e-07, "loss": 0.8092, "step": 17668 }, { "epoch": 2.84, "grad_norm": 14.176932334899902, "learning_rate": 1.4089499177355804e-07, "loss": 0.606, "step": 17669 }, { "epoch": 2.84, "grad_norm": 13.036931037902832, "learning_rate": 1.4060442051425426e-07, "loss": 0.5315, "step": 17670 }, { "epoch": 2.84, "grad_norm": 14.529269218444824, "learning_rate": 1.403141470691305e-07, "loss": 0.4875, "step": 17671 }, { "epoch": 2.84, "grad_norm": 16.068862915039062, "learning_rate": 1.4002417144695414e-07, "loss": 0.4703, "step": 17672 }, { "epoch": 2.84, "grad_norm": 13.12965202331543, "learning_rate": 1.3973449365648152e-07, "loss": 0.7011, "step": 17673 }, { "epoch": 2.85, "grad_norm": 13.470033645629883, "learning_rate": 1.3944511370646675e-07, "loss": 0.7126, "step": 17674 }, { "epoch": 2.85, "grad_norm": 12.017206192016602, "learning_rate": 1.391560316056484e-07, "loss": 0.7564, "step": 17675 }, { "epoch": 2.85, "grad_norm": 15.706214904785156, "learning_rate": 1.3886724736275948e-07, "loss": 0.7494, "step": 17676 }, { "epoch": 2.85, "grad_norm": 16.720643997192383, "learning_rate": 1.38578760986523e-07, "loss": 0.6004, "step": 17677 }, { "epoch": 2.85, "grad_norm": 13.356062889099121, "learning_rate": 1.3829057248565313e-07, "loss": 0.6701, "step": 17678 }, { "epoch": 2.85, "grad_norm": 13.124581336975098, "learning_rate": 1.3800268186885403e-07, "loss": 0.6575, "step": 17679 }, { "epoch": 2.85, "grad_norm": 11.913304328918457, "learning_rate": 1.3771508914482312e-07, "loss": 0.8025, "step": 17680 }, { "epoch": 2.85, "grad_norm": 16.211698532104492, "learning_rate": 1.3742779432224574e-07, "loss": 0.6224, "step": 17681 }, { "epoch": 2.85, "grad_norm": 12.482089042663574, "learning_rate": 1.3714079740980158e-07, "loss": 0.6347, "step": 17682 }, { "epoch": 2.85, "grad_norm": 13.07577133178711, "learning_rate": 1.3685409841615925e-07, "loss": 0.6137, "step": 17683 }, { "epoch": 2.85, "grad_norm": 14.721270561218262, "learning_rate": 1.3656769734997744e-07, "loss": 0.6736, "step": 17684 }, { "epoch": 2.85, "grad_norm": 11.56810474395752, "learning_rate": 1.3628159421991028e-07, "loss": 0.8523, "step": 17685 }, { "epoch": 2.85, "grad_norm": 13.805439949035645, "learning_rate": 1.3599578903459642e-07, "loss": 0.6684, "step": 17686 }, { "epoch": 2.85, "grad_norm": 12.45947265625, "learning_rate": 1.3571028180267121e-07, "loss": 0.5678, "step": 17687 }, { "epoch": 2.85, "grad_norm": 13.421834945678711, "learning_rate": 1.354250725327577e-07, "loss": 0.7861, "step": 17688 }, { "epoch": 2.85, "grad_norm": 18.732810974121094, "learning_rate": 1.3514016123347019e-07, "loss": 0.8926, "step": 17689 }, { "epoch": 2.85, "grad_norm": 12.380578994750977, "learning_rate": 1.3485554791341727e-07, "loss": 0.566, "step": 17690 }, { "epoch": 2.85, "grad_norm": 11.322660446166992, "learning_rate": 1.3457123258119432e-07, "loss": 0.6487, "step": 17691 }, { "epoch": 2.85, "grad_norm": 12.930215835571289, "learning_rate": 1.3428721524538891e-07, "loss": 0.6779, "step": 17692 }, { "epoch": 2.85, "grad_norm": 15.148664474487305, "learning_rate": 1.3400349591458195e-07, "loss": 0.7812, "step": 17693 }, { "epoch": 2.85, "grad_norm": 14.487943649291992, "learning_rate": 1.3372007459734104e-07, "loss": 0.7722, "step": 17694 }, { "epoch": 2.85, "grad_norm": 13.075395584106445, "learning_rate": 1.3343695130222822e-07, "loss": 0.5616, "step": 17695 }, { "epoch": 2.85, "grad_norm": 12.521550178527832, "learning_rate": 1.3315412603779777e-07, "loss": 0.5476, "step": 17696 }, { "epoch": 2.85, "grad_norm": 14.762956619262695, "learning_rate": 1.328715988125906e-07, "loss": 0.8566, "step": 17697 }, { "epoch": 2.85, "grad_norm": 12.509098052978516, "learning_rate": 1.325893696351388e-07, "loss": 0.4023, "step": 17698 }, { "epoch": 2.85, "grad_norm": 13.900825500488281, "learning_rate": 1.323074385139722e-07, "loss": 0.6988, "step": 17699 }, { "epoch": 2.85, "grad_norm": 15.321306228637695, "learning_rate": 1.320258054576029e-07, "loss": 0.887, "step": 17700 }, { "epoch": 2.85, "grad_norm": 14.16132926940918, "learning_rate": 1.3174447047453965e-07, "loss": 0.7544, "step": 17701 }, { "epoch": 2.85, "grad_norm": 13.72001838684082, "learning_rate": 1.3146343357328005e-07, "loss": 0.6428, "step": 17702 }, { "epoch": 2.85, "grad_norm": 10.441218376159668, "learning_rate": 1.3118269476231405e-07, "loss": 0.5341, "step": 17703 }, { "epoch": 2.85, "grad_norm": 17.268165588378906, "learning_rate": 1.3090225405012035e-07, "loss": 0.7673, "step": 17704 }, { "epoch": 2.85, "grad_norm": 10.512121200561523, "learning_rate": 1.3062211144517002e-07, "loss": 0.8118, "step": 17705 }, { "epoch": 2.85, "grad_norm": 10.608842849731445, "learning_rate": 1.3034226695592623e-07, "loss": 0.6934, "step": 17706 }, { "epoch": 2.85, "grad_norm": 13.863417625427246, "learning_rate": 1.3006272059084112e-07, "loss": 0.8454, "step": 17707 }, { "epoch": 2.85, "grad_norm": 11.926459312438965, "learning_rate": 1.2978347235835908e-07, "loss": 0.6819, "step": 17708 }, { "epoch": 2.85, "grad_norm": 12.778199195861816, "learning_rate": 1.2950452226691446e-07, "loss": 0.7668, "step": 17709 }, { "epoch": 2.85, "grad_norm": 16.45713233947754, "learning_rate": 1.2922587032493494e-07, "loss": 0.6634, "step": 17710 }, { "epoch": 2.85, "grad_norm": 13.024417877197266, "learning_rate": 1.2894751654083492e-07, "loss": 0.7884, "step": 17711 }, { "epoch": 2.85, "grad_norm": 15.107769966125488, "learning_rate": 1.2866946092302546e-07, "loss": 0.7653, "step": 17712 }, { "epoch": 2.85, "grad_norm": 13.271153450012207, "learning_rate": 1.2839170347990204e-07, "loss": 0.8604, "step": 17713 }, { "epoch": 2.85, "grad_norm": 11.482175827026367, "learning_rate": 1.2811424421985684e-07, "loss": 0.7105, "step": 17714 }, { "epoch": 2.85, "grad_norm": 14.207906723022461, "learning_rate": 1.2783708315127208e-07, "loss": 0.7736, "step": 17715 }, { "epoch": 2.85, "grad_norm": 12.345966339111328, "learning_rate": 1.275602202825177e-07, "loss": 0.6975, "step": 17716 }, { "epoch": 2.85, "grad_norm": 19.248374938964844, "learning_rate": 1.272836556219559e-07, "loss": 0.6927, "step": 17717 }, { "epoch": 2.85, "grad_norm": 13.24837589263916, "learning_rate": 1.270073891779433e-07, "loss": 0.8784, "step": 17718 }, { "epoch": 2.85, "grad_norm": 14.738431930541992, "learning_rate": 1.2673142095882217e-07, "loss": 0.5981, "step": 17719 }, { "epoch": 2.85, "grad_norm": 14.555586814880371, "learning_rate": 1.2645575097293028e-07, "loss": 0.5034, "step": 17720 }, { "epoch": 2.85, "grad_norm": 11.566095352172852, "learning_rate": 1.2618037922859316e-07, "loss": 0.6497, "step": 17721 }, { "epoch": 2.85, "grad_norm": 14.28179931640625, "learning_rate": 1.2590530573412974e-07, "loss": 0.8409, "step": 17722 }, { "epoch": 2.85, "grad_norm": 12.607216835021973, "learning_rate": 1.2563053049785e-07, "loss": 0.5586, "step": 17723 }, { "epoch": 2.85, "grad_norm": 13.849665641784668, "learning_rate": 1.253560535280507e-07, "loss": 0.819, "step": 17724 }, { "epoch": 2.85, "grad_norm": 12.49821949005127, "learning_rate": 1.2508187483302514e-07, "loss": 0.6082, "step": 17725 }, { "epoch": 2.85, "grad_norm": 11.410171508789062, "learning_rate": 1.2480799442105562e-07, "loss": 0.5972, "step": 17726 }, { "epoch": 2.85, "grad_norm": 9.852184295654297, "learning_rate": 1.2453441230041218e-07, "loss": 0.6949, "step": 17727 }, { "epoch": 2.85, "grad_norm": 15.927093505859375, "learning_rate": 1.2426112847936156e-07, "loss": 0.704, "step": 17728 }, { "epoch": 2.85, "grad_norm": 15.45547103881836, "learning_rate": 1.2398814296615826e-07, "loss": 0.6785, "step": 17729 }, { "epoch": 2.85, "grad_norm": 11.478398323059082, "learning_rate": 1.2371545576904565e-07, "loss": 0.5407, "step": 17730 }, { "epoch": 2.85, "grad_norm": 14.969673156738281, "learning_rate": 1.2344306689626383e-07, "loss": 0.7266, "step": 17731 }, { "epoch": 2.85, "grad_norm": 11.988761901855469, "learning_rate": 1.2317097635603736e-07, "loss": 0.6145, "step": 17732 }, { "epoch": 2.85, "grad_norm": 14.5076265335083, "learning_rate": 1.2289918415658852e-07, "loss": 0.5736, "step": 17733 }, { "epoch": 2.85, "grad_norm": 13.533211708068848, "learning_rate": 1.226276903061241e-07, "loss": 0.6338, "step": 17734 }, { "epoch": 2.85, "grad_norm": 13.82583999633789, "learning_rate": 1.2235649481284639e-07, "loss": 0.8544, "step": 17735 }, { "epoch": 2.86, "grad_norm": 14.3781099319458, "learning_rate": 1.2208559768494665e-07, "loss": 0.8014, "step": 17736 }, { "epoch": 2.86, "grad_norm": 12.948044776916504, "learning_rate": 1.2181499893060834e-07, "loss": 0.8132, "step": 17737 }, { "epoch": 2.86, "grad_norm": 12.853020668029785, "learning_rate": 1.2154469855800488e-07, "loss": 0.7694, "step": 17738 }, { "epoch": 2.86, "grad_norm": 14.050294876098633, "learning_rate": 1.212746965752998e-07, "loss": 0.8633, "step": 17739 }, { "epoch": 2.86, "grad_norm": 13.159438133239746, "learning_rate": 1.2100499299065094e-07, "loss": 0.8628, "step": 17740 }, { "epoch": 2.86, "grad_norm": 14.176741600036621, "learning_rate": 1.2073558781220297e-07, "loss": 0.6055, "step": 17741 }, { "epoch": 2.86, "grad_norm": 11.71650218963623, "learning_rate": 1.20466481048096e-07, "loss": 0.7072, "step": 17742 }, { "epoch": 2.86, "grad_norm": 13.118359565734863, "learning_rate": 1.2019767270645576e-07, "loss": 0.7033, "step": 17743 }, { "epoch": 2.86, "grad_norm": 14.79788589477539, "learning_rate": 1.1992916279540356e-07, "loss": 0.6663, "step": 17744 }, { "epoch": 2.86, "grad_norm": 13.55276870727539, "learning_rate": 1.1966095132305067e-07, "loss": 0.6677, "step": 17745 }, { "epoch": 2.86, "grad_norm": 13.360150337219238, "learning_rate": 1.1939303829749726e-07, "loss": 0.6761, "step": 17746 }, { "epoch": 2.86, "grad_norm": 13.256351470947266, "learning_rate": 1.1912542372683688e-07, "loss": 0.6633, "step": 17747 }, { "epoch": 2.86, "grad_norm": 10.78696060180664, "learning_rate": 1.1885810761915195e-07, "loss": 0.5031, "step": 17748 }, { "epoch": 2.86, "grad_norm": 12.565303802490234, "learning_rate": 1.1859108998251933e-07, "loss": 0.757, "step": 17749 }, { "epoch": 2.86, "grad_norm": 12.87475299835205, "learning_rate": 1.1832437082500259e-07, "loss": 0.7887, "step": 17750 }, { "epoch": 2.86, "grad_norm": 11.11595344543457, "learning_rate": 1.180579501546586e-07, "loss": 0.7792, "step": 17751 }, { "epoch": 2.86, "grad_norm": 10.65345287322998, "learning_rate": 1.1779182797953648e-07, "loss": 0.5429, "step": 17752 }, { "epoch": 2.86, "grad_norm": 12.047231674194336, "learning_rate": 1.1752600430767203e-07, "loss": 0.7379, "step": 17753 }, { "epoch": 2.86, "grad_norm": 18.051090240478516, "learning_rate": 1.172604791470966e-07, "loss": 0.8133, "step": 17754 }, { "epoch": 2.86, "grad_norm": 12.18431282043457, "learning_rate": 1.1699525250582933e-07, "loss": 0.6923, "step": 17755 }, { "epoch": 2.86, "grad_norm": 13.125835418701172, "learning_rate": 1.1673032439188381e-07, "loss": 0.8602, "step": 17756 }, { "epoch": 2.86, "grad_norm": 12.369242668151855, "learning_rate": 1.1646569481326142e-07, "loss": 0.4602, "step": 17757 }, { "epoch": 2.86, "grad_norm": 13.458449363708496, "learning_rate": 1.1620136377795466e-07, "loss": 0.6619, "step": 17758 }, { "epoch": 2.86, "grad_norm": 10.553388595581055, "learning_rate": 1.1593733129394935e-07, "loss": 0.6262, "step": 17759 }, { "epoch": 2.86, "grad_norm": 11.697053909301758, "learning_rate": 1.1567359736922023e-07, "loss": 0.8378, "step": 17760 }, { "epoch": 2.86, "grad_norm": 14.13017749786377, "learning_rate": 1.1541016201173427e-07, "loss": 0.8216, "step": 17761 }, { "epoch": 2.86, "grad_norm": 15.187655448913574, "learning_rate": 1.1514702522944732e-07, "loss": 0.7371, "step": 17762 }, { "epoch": 2.86, "grad_norm": 12.960156440734863, "learning_rate": 1.1488418703030969e-07, "loss": 0.6256, "step": 17763 }, { "epoch": 2.86, "grad_norm": 14.176733016967773, "learning_rate": 1.1462164742225945e-07, "loss": 0.863, "step": 17764 }, { "epoch": 2.86, "grad_norm": 15.737659454345703, "learning_rate": 1.1435940641322585e-07, "loss": 0.6072, "step": 17765 }, { "epoch": 2.86, "grad_norm": 13.780329704284668, "learning_rate": 1.1409746401113363e-07, "loss": 0.686, "step": 17766 }, { "epoch": 2.86, "grad_norm": 14.500290870666504, "learning_rate": 1.1383582022389095e-07, "loss": 0.7157, "step": 17767 }, { "epoch": 2.86, "grad_norm": 12.427379608154297, "learning_rate": 1.1357447505940477e-07, "loss": 0.5901, "step": 17768 }, { "epoch": 2.86, "grad_norm": 16.904443740844727, "learning_rate": 1.133134285255677e-07, "loss": 0.8435, "step": 17769 }, { "epoch": 2.86, "grad_norm": 13.754667282104492, "learning_rate": 1.1305268063026343e-07, "loss": 0.5862, "step": 17770 }, { "epoch": 2.86, "grad_norm": 12.472590446472168, "learning_rate": 1.1279223138137007e-07, "loss": 0.7084, "step": 17771 }, { "epoch": 2.86, "grad_norm": 11.091780662536621, "learning_rate": 1.125320807867547e-07, "loss": 0.7531, "step": 17772 }, { "epoch": 2.86, "grad_norm": 11.187952041625977, "learning_rate": 1.1227222885427436e-07, "loss": 0.6409, "step": 17773 }, { "epoch": 2.86, "grad_norm": 15.010985374450684, "learning_rate": 1.1201267559177942e-07, "loss": 0.7152, "step": 17774 }, { "epoch": 2.86, "grad_norm": 11.129051208496094, "learning_rate": 1.1175342100710917e-07, "loss": 0.5276, "step": 17775 }, { "epoch": 2.86, "grad_norm": 15.574932098388672, "learning_rate": 1.1149446510809403e-07, "loss": 0.7425, "step": 17776 }, { "epoch": 2.86, "grad_norm": 14.253874778747559, "learning_rate": 1.1123580790255883e-07, "loss": 0.7928, "step": 17777 }, { "epoch": 2.86, "grad_norm": 11.374269485473633, "learning_rate": 1.1097744939831289e-07, "loss": 0.6104, "step": 17778 }, { "epoch": 2.86, "grad_norm": 17.250041961669922, "learning_rate": 1.107193896031633e-07, "loss": 0.6782, "step": 17779 }, { "epoch": 2.86, "grad_norm": 11.699338912963867, "learning_rate": 1.1046162852490383e-07, "loss": 0.4563, "step": 17780 }, { "epoch": 2.86, "grad_norm": 13.15056037902832, "learning_rate": 1.1020416617132047e-07, "loss": 0.7271, "step": 17781 }, { "epoch": 2.86, "grad_norm": 16.08403205871582, "learning_rate": 1.099470025501892e-07, "loss": 0.8531, "step": 17782 }, { "epoch": 2.86, "grad_norm": 11.602266311645508, "learning_rate": 1.0969013766927938e-07, "loss": 0.7788, "step": 17783 }, { "epoch": 2.86, "grad_norm": 13.366692543029785, "learning_rate": 1.0943357153634926e-07, "loss": 0.5376, "step": 17784 }, { "epoch": 2.86, "grad_norm": 16.335567474365234, "learning_rate": 1.0917730415914929e-07, "loss": 0.511, "step": 17785 }, { "epoch": 2.86, "grad_norm": 14.523283004760742, "learning_rate": 1.0892133554541883e-07, "loss": 0.6413, "step": 17786 }, { "epoch": 2.86, "grad_norm": 12.148004531860352, "learning_rate": 1.0866566570289173e-07, "loss": 0.8327, "step": 17787 }, { "epoch": 2.86, "grad_norm": 13.68998908996582, "learning_rate": 1.0841029463928954e-07, "loss": 0.7304, "step": 17788 }, { "epoch": 2.86, "grad_norm": 21.634803771972656, "learning_rate": 1.0815522236232611e-07, "loss": 0.8207, "step": 17789 }, { "epoch": 2.86, "grad_norm": 21.057409286499023, "learning_rate": 1.0790044887970641e-07, "loss": 0.7195, "step": 17790 }, { "epoch": 2.86, "grad_norm": 16.459436416625977, "learning_rate": 1.0764597419912537e-07, "loss": 0.7364, "step": 17791 }, { "epoch": 2.86, "grad_norm": 13.531325340270996, "learning_rate": 1.0739179832827129e-07, "loss": 0.6667, "step": 17792 }, { "epoch": 2.86, "grad_norm": 29.707380294799805, "learning_rate": 1.0713792127482137e-07, "loss": 0.7296, "step": 17793 }, { "epoch": 2.86, "grad_norm": 15.628758430480957, "learning_rate": 1.068843430464428e-07, "loss": 0.8322, "step": 17794 }, { "epoch": 2.86, "grad_norm": 15.995365142822266, "learning_rate": 1.0663106365079723e-07, "loss": 0.7741, "step": 17795 }, { "epoch": 2.86, "grad_norm": 12.604220390319824, "learning_rate": 1.0637808309553299e-07, "loss": 0.9986, "step": 17796 }, { "epoch": 2.86, "grad_norm": 13.969849586486816, "learning_rate": 1.0612540138829286e-07, "loss": 0.6336, "step": 17797 }, { "epoch": 2.87, "grad_norm": 11.588327407836914, "learning_rate": 1.0587301853671072e-07, "loss": 0.5617, "step": 17798 }, { "epoch": 2.87, "grad_norm": 10.961610794067383, "learning_rate": 1.0562093454840827e-07, "loss": 0.6122, "step": 17799 }, { "epoch": 2.87, "grad_norm": 11.347624778747559, "learning_rate": 1.0536914943100052e-07, "loss": 0.8091, "step": 17800 }, { "epoch": 2.87, "grad_norm": 14.903216361999512, "learning_rate": 1.0511766319209248e-07, "loss": 0.8325, "step": 17801 }, { "epoch": 2.87, "grad_norm": 13.08217716217041, "learning_rate": 1.0486647583928144e-07, "loss": 0.834, "step": 17802 }, { "epoch": 2.87, "grad_norm": 12.263548851013184, "learning_rate": 1.0461558738015465e-07, "loss": 0.4833, "step": 17803 }, { "epoch": 2.87, "grad_norm": 13.303665161132812, "learning_rate": 1.0436499782229048e-07, "loss": 0.842, "step": 17804 }, { "epoch": 2.87, "grad_norm": 13.5858793258667, "learning_rate": 1.0411470717325622e-07, "loss": 0.8301, "step": 17805 }, { "epoch": 2.87, "grad_norm": 11.223636627197266, "learning_rate": 1.0386471544061583e-07, "loss": 0.7239, "step": 17806 }, { "epoch": 2.87, "grad_norm": 12.43907356262207, "learning_rate": 1.036150226319177e-07, "loss": 0.6106, "step": 17807 }, { "epoch": 2.87, "grad_norm": 13.906054496765137, "learning_rate": 1.0336562875470469e-07, "loss": 0.603, "step": 17808 }, { "epoch": 2.87, "grad_norm": 11.184511184692383, "learning_rate": 1.0311653381651187e-07, "loss": 0.6347, "step": 17809 }, { "epoch": 2.87, "grad_norm": 13.12658405303955, "learning_rate": 1.0286773782486103e-07, "loss": 0.8533, "step": 17810 }, { "epoch": 2.87, "grad_norm": 14.009721755981445, "learning_rate": 1.0261924078726836e-07, "loss": 0.8481, "step": 17811 }, { "epoch": 2.87, "grad_norm": 12.987624168395996, "learning_rate": 1.023710427112401e-07, "loss": 0.7593, "step": 17812 }, { "epoch": 2.87, "grad_norm": 15.089010238647461, "learning_rate": 1.0212314360427244e-07, "loss": 0.6212, "step": 17813 }, { "epoch": 2.87, "grad_norm": 12.704780578613281, "learning_rate": 1.0187554347385608e-07, "loss": 0.7213, "step": 17814 }, { "epoch": 2.87, "grad_norm": 13.158267974853516, "learning_rate": 1.0162824232746616e-07, "loss": 0.6522, "step": 17815 }, { "epoch": 2.87, "grad_norm": 13.140162467956543, "learning_rate": 1.0138124017257556e-07, "loss": 0.7971, "step": 17816 }, { "epoch": 2.87, "grad_norm": 26.024717330932617, "learning_rate": 1.0113453701664388e-07, "loss": 0.6658, "step": 17817 }, { "epoch": 2.87, "grad_norm": 14.490948677062988, "learning_rate": 1.0088813286712406e-07, "loss": 0.8141, "step": 17818 }, { "epoch": 2.87, "grad_norm": 11.866509437561035, "learning_rate": 1.0064202773145903e-07, "loss": 0.6523, "step": 17819 }, { "epoch": 2.87, "grad_norm": 11.314922332763672, "learning_rate": 1.0039622161708173e-07, "loss": 0.7343, "step": 17820 }, { "epoch": 2.87, "grad_norm": 12.788570404052734, "learning_rate": 1.0015071453141734e-07, "loss": 0.738, "step": 17821 }, { "epoch": 2.87, "grad_norm": 11.435528755187988, "learning_rate": 9.990550648188213e-08, "loss": 0.699, "step": 17822 }, { "epoch": 2.87, "grad_norm": 12.701212882995605, "learning_rate": 9.966059747588241e-08, "loss": 0.6608, "step": 17823 }, { "epoch": 2.87, "grad_norm": 12.374526977539062, "learning_rate": 9.94159875208156e-08, "loss": 0.6233, "step": 17824 }, { "epoch": 2.87, "grad_norm": 11.47159481048584, "learning_rate": 9.917167662407135e-08, "loss": 0.8238, "step": 17825 }, { "epoch": 2.87, "grad_norm": 12.802717208862305, "learning_rate": 9.892766479302817e-08, "loss": 0.906, "step": 17826 }, { "epoch": 2.87, "grad_norm": 11.931520462036133, "learning_rate": 9.868395203505909e-08, "loss": 0.4796, "step": 17827 }, { "epoch": 2.87, "grad_norm": 12.706780433654785, "learning_rate": 9.844053835752265e-08, "loss": 0.5769, "step": 17828 }, { "epoch": 2.87, "grad_norm": 13.696517944335938, "learning_rate": 9.819742376777297e-08, "loss": 0.9676, "step": 17829 }, { "epoch": 2.87, "grad_norm": 14.06647777557373, "learning_rate": 9.795460827315418e-08, "loss": 0.8255, "step": 17830 }, { "epoch": 2.87, "grad_norm": 20.75092124938965, "learning_rate": 9.771209188100039e-08, "loss": 0.8687, "step": 17831 }, { "epoch": 2.87, "grad_norm": 14.941205024719238, "learning_rate": 9.746987459863578e-08, "loss": 0.8757, "step": 17832 }, { "epoch": 2.87, "grad_norm": 12.203629493713379, "learning_rate": 9.72279564333789e-08, "loss": 0.6482, "step": 17833 }, { "epoch": 2.87, "grad_norm": 13.465462684631348, "learning_rate": 9.698633739253505e-08, "loss": 0.7044, "step": 17834 }, { "epoch": 2.87, "grad_norm": 12.533132553100586, "learning_rate": 9.674501748340503e-08, "loss": 0.8605, "step": 17835 }, { "epoch": 2.87, "grad_norm": 10.239126205444336, "learning_rate": 9.650399671327526e-08, "loss": 0.7765, "step": 17836 }, { "epoch": 2.87, "grad_norm": 13.889060974121094, "learning_rate": 9.626327508942767e-08, "loss": 0.7091, "step": 17837 }, { "epoch": 2.87, "grad_norm": 11.31440544128418, "learning_rate": 9.602285261913424e-08, "loss": 0.7889, "step": 17838 }, { "epoch": 2.87, "grad_norm": 13.145280838012695, "learning_rate": 9.578272930965582e-08, "loss": 0.6831, "step": 17839 }, { "epoch": 2.87, "grad_norm": 13.768684387207031, "learning_rate": 9.55429051682466e-08, "loss": 0.8106, "step": 17840 }, { "epoch": 2.87, "grad_norm": 10.31796932220459, "learning_rate": 9.53033802021508e-08, "loss": 0.5506, "step": 17841 }, { "epoch": 2.87, "grad_norm": 14.499152183532715, "learning_rate": 9.506415441860261e-08, "loss": 0.6382, "step": 17842 }, { "epoch": 2.87, "grad_norm": 12.307422637939453, "learning_rate": 9.482522782482739e-08, "loss": 0.7364, "step": 17843 }, { "epoch": 2.87, "grad_norm": 13.037363052368164, "learning_rate": 9.45866004280449e-08, "loss": 0.6, "step": 17844 }, { "epoch": 2.87, "grad_norm": 13.566401481628418, "learning_rate": 9.434827223546162e-08, "loss": 0.8532, "step": 17845 }, { "epoch": 2.87, "grad_norm": 12.383255958557129, "learning_rate": 9.41102432542762e-08, "loss": 0.6872, "step": 17846 }, { "epoch": 2.87, "grad_norm": 12.657033920288086, "learning_rate": 9.387251349167848e-08, "loss": 0.6605, "step": 17847 }, { "epoch": 2.87, "grad_norm": 13.83809757232666, "learning_rate": 9.363508295484936e-08, "loss": 0.7153, "step": 17848 }, { "epoch": 2.87, "grad_norm": 15.192131042480469, "learning_rate": 9.3397951650962e-08, "loss": 0.674, "step": 17849 }, { "epoch": 2.87, "grad_norm": 16.755281448364258, "learning_rate": 9.316111958717844e-08, "loss": 0.5803, "step": 17850 }, { "epoch": 2.87, "grad_norm": 15.257122039794922, "learning_rate": 9.292458677065186e-08, "loss": 0.7649, "step": 17851 }, { "epoch": 2.87, "grad_norm": 13.109073638916016, "learning_rate": 9.268835320852876e-08, "loss": 0.6705, "step": 17852 }, { "epoch": 2.87, "grad_norm": 15.602392196655273, "learning_rate": 9.245241890794121e-08, "loss": 0.804, "step": 17853 }, { "epoch": 2.87, "grad_norm": 15.967536926269531, "learning_rate": 9.221678387602018e-08, "loss": 0.8328, "step": 17854 }, { "epoch": 2.87, "grad_norm": 13.94946002960205, "learning_rate": 9.198144811988108e-08, "loss": 0.8742, "step": 17855 }, { "epoch": 2.87, "grad_norm": 13.732185363769531, "learning_rate": 9.174641164663156e-08, "loss": 0.7241, "step": 17856 }, { "epoch": 2.87, "grad_norm": 11.220407485961914, "learning_rate": 9.151167446337262e-08, "loss": 0.4989, "step": 17857 }, { "epoch": 2.87, "grad_norm": 13.892367362976074, "learning_rate": 9.127723657719522e-08, "loss": 0.8198, "step": 17858 }, { "epoch": 2.87, "grad_norm": 13.606117248535156, "learning_rate": 9.104309799518041e-08, "loss": 1.0352, "step": 17859 }, { "epoch": 2.88, "grad_norm": 11.052453994750977, "learning_rate": 9.080925872440027e-08, "loss": 0.7486, "step": 17860 }, { "epoch": 2.88, "grad_norm": 14.576070785522461, "learning_rate": 9.057571877191695e-08, "loss": 0.7198, "step": 17861 }, { "epoch": 2.88, "grad_norm": 15.169914245605469, "learning_rate": 9.034247814478702e-08, "loss": 0.7096, "step": 17862 }, { "epoch": 2.88, "grad_norm": 14.718538284301758, "learning_rate": 9.010953685005486e-08, "loss": 0.6387, "step": 17863 }, { "epoch": 2.88, "grad_norm": 15.56336498260498, "learning_rate": 8.987689489475593e-08, "loss": 0.7247, "step": 17864 }, { "epoch": 2.88, "grad_norm": 13.078034400939941, "learning_rate": 8.964455228591906e-08, "loss": 0.8374, "step": 17865 }, { "epoch": 2.88, "grad_norm": 11.563272476196289, "learning_rate": 8.941250903056198e-08, "loss": 0.6755, "step": 17866 }, { "epoch": 2.88, "grad_norm": 13.564401626586914, "learning_rate": 8.91807651356924e-08, "loss": 0.602, "step": 17867 }, { "epoch": 2.88, "grad_norm": 14.843628883361816, "learning_rate": 8.894932060831141e-08, "loss": 0.7538, "step": 17868 }, { "epoch": 2.88, "grad_norm": 14.743680000305176, "learning_rate": 8.871817545541006e-08, "loss": 0.8282, "step": 17869 }, { "epoch": 2.88, "grad_norm": 13.380881309509277, "learning_rate": 8.848732968397056e-08, "loss": 0.7257, "step": 17870 }, { "epoch": 2.88, "grad_norm": 12.088550567626953, "learning_rate": 8.825678330096621e-08, "loss": 0.6688, "step": 17871 }, { "epoch": 2.88, "grad_norm": 17.083629608154297, "learning_rate": 8.802653631335923e-08, "loss": 0.7705, "step": 17872 }, { "epoch": 2.88, "grad_norm": 14.936816215515137, "learning_rate": 8.779658872810626e-08, "loss": 0.6173, "step": 17873 }, { "epoch": 2.88, "grad_norm": 15.429998397827148, "learning_rate": 8.756694055215175e-08, "loss": 0.6637, "step": 17874 }, { "epoch": 2.88, "grad_norm": 12.284310340881348, "learning_rate": 8.73375917924335e-08, "loss": 0.6809, "step": 17875 }, { "epoch": 2.88, "grad_norm": 16.107494354248047, "learning_rate": 8.710854245588041e-08, "loss": 0.6741, "step": 17876 }, { "epoch": 2.88, "grad_norm": 15.523759841918945, "learning_rate": 8.687979254940804e-08, "loss": 0.5789, "step": 17877 }, { "epoch": 2.88, "grad_norm": 13.528928756713867, "learning_rate": 8.665134207992753e-08, "loss": 0.7019, "step": 17878 }, { "epoch": 2.88, "grad_norm": 13.655326843261719, "learning_rate": 8.642319105434005e-08, "loss": 0.7616, "step": 17879 }, { "epoch": 2.88, "grad_norm": 13.551168441772461, "learning_rate": 8.619533947953673e-08, "loss": 0.931, "step": 17880 }, { "epoch": 2.88, "grad_norm": 12.567424774169922, "learning_rate": 8.596778736240097e-08, "loss": 0.6163, "step": 17881 }, { "epoch": 2.88, "grad_norm": 12.491299629211426, "learning_rate": 8.574053470980504e-08, "loss": 0.5794, "step": 17882 }, { "epoch": 2.88, "grad_norm": 12.161306381225586, "learning_rate": 8.551358152861345e-08, "loss": 0.9273, "step": 17883 }, { "epoch": 2.88, "grad_norm": 11.514862060546875, "learning_rate": 8.528692782568293e-08, "loss": 0.7897, "step": 17884 }, { "epoch": 2.88, "grad_norm": 13.244270324707031, "learning_rate": 8.506057360785802e-08, "loss": 0.7245, "step": 17885 }, { "epoch": 2.88, "grad_norm": 11.174567222595215, "learning_rate": 8.483451888197547e-08, "loss": 0.8047, "step": 17886 }, { "epoch": 2.88, "grad_norm": 13.890679359436035, "learning_rate": 8.460876365486648e-08, "loss": 0.8613, "step": 17887 }, { "epoch": 2.88, "grad_norm": 19.009233474731445, "learning_rate": 8.43833079333467e-08, "loss": 0.6404, "step": 17888 }, { "epoch": 2.88, "grad_norm": 10.272289276123047, "learning_rate": 8.415815172422959e-08, "loss": 0.6638, "step": 17889 }, { "epoch": 2.88, "grad_norm": 13.837200164794922, "learning_rate": 8.393329503431525e-08, "loss": 0.6967, "step": 17890 }, { "epoch": 2.88, "grad_norm": 13.933040618896484, "learning_rate": 8.37087378703938e-08, "loss": 0.7141, "step": 17891 }, { "epoch": 2.88, "grad_norm": 13.105712890625, "learning_rate": 8.348448023925093e-08, "loss": 0.7221, "step": 17892 }, { "epoch": 2.88, "grad_norm": 14.080294609069824, "learning_rate": 8.32605221476579e-08, "loss": 0.585, "step": 17893 }, { "epoch": 2.88, "grad_norm": 13.925236701965332, "learning_rate": 8.30368636023826e-08, "loss": 0.5862, "step": 17894 }, { "epoch": 2.88, "grad_norm": 10.826465606689453, "learning_rate": 8.281350461017856e-08, "loss": 0.5834, "step": 17895 }, { "epoch": 2.88, "grad_norm": 12.085087776184082, "learning_rate": 8.259044517779368e-08, "loss": 0.7048, "step": 17896 }, { "epoch": 2.88, "grad_norm": 13.762364387512207, "learning_rate": 8.236768531196482e-08, "loss": 0.9328, "step": 17897 }, { "epoch": 2.88, "grad_norm": 11.871880531311035, "learning_rate": 8.214522501942213e-08, "loss": 0.7129, "step": 17898 }, { "epoch": 2.88, "grad_norm": 14.837783813476562, "learning_rate": 8.192306430688358e-08, "loss": 0.7926, "step": 17899 }, { "epoch": 2.88, "grad_norm": 13.924664497375488, "learning_rate": 8.170120318106045e-08, "loss": 0.8943, "step": 17900 }, { "epoch": 2.88, "grad_norm": 14.657526969909668, "learning_rate": 8.147964164865629e-08, "loss": 0.7203, "step": 17901 }, { "epoch": 2.88, "grad_norm": 11.4554443359375, "learning_rate": 8.125837971636019e-08, "loss": 0.7672, "step": 17902 }, { "epoch": 2.88, "grad_norm": 12.727299690246582, "learning_rate": 8.10374173908579e-08, "loss": 0.82, "step": 17903 }, { "epoch": 2.88, "grad_norm": 11.844175338745117, "learning_rate": 8.0816754678823e-08, "loss": 0.7197, "step": 17904 }, { "epoch": 2.88, "grad_norm": 14.957755088806152, "learning_rate": 8.059639158692123e-08, "loss": 0.6837, "step": 17905 }, { "epoch": 2.88, "grad_norm": 13.200798034667969, "learning_rate": 8.03763281218084e-08, "loss": 0.8232, "step": 17906 }, { "epoch": 2.88, "grad_norm": 13.917341232299805, "learning_rate": 8.015656429013252e-08, "loss": 0.6894, "step": 17907 }, { "epoch": 2.88, "grad_norm": 16.069313049316406, "learning_rate": 7.993710009853161e-08, "loss": 0.7082, "step": 17908 }, { "epoch": 2.88, "grad_norm": 12.195878028869629, "learning_rate": 7.971793555363372e-08, "loss": 0.7494, "step": 17909 }, { "epoch": 2.88, "grad_norm": 13.865134239196777, "learning_rate": 7.94990706620602e-08, "loss": 0.8678, "step": 17910 }, { "epoch": 2.88, "grad_norm": 12.972784042358398, "learning_rate": 7.928050543042242e-08, "loss": 0.7872, "step": 17911 }, { "epoch": 2.88, "grad_norm": 15.038873672485352, "learning_rate": 7.906223986532068e-08, "loss": 1.0223, "step": 17912 }, { "epoch": 2.88, "grad_norm": 16.337554931640625, "learning_rate": 7.88442739733497e-08, "loss": 0.7851, "step": 17913 }, { "epoch": 2.88, "grad_norm": 10.746503829956055, "learning_rate": 7.86266077610931e-08, "loss": 0.632, "step": 17914 }, { "epoch": 2.88, "grad_norm": 16.014934539794922, "learning_rate": 7.84092412351245e-08, "loss": 0.8748, "step": 17915 }, { "epoch": 2.88, "grad_norm": 10.376233100891113, "learning_rate": 7.819217440201088e-08, "loss": 0.5768, "step": 17916 }, { "epoch": 2.88, "grad_norm": 11.930068969726562, "learning_rate": 7.79754072683081e-08, "loss": 0.5647, "step": 17917 }, { "epoch": 2.88, "grad_norm": 11.553930282592773, "learning_rate": 7.775893984056427e-08, "loss": 0.7896, "step": 17918 }, { "epoch": 2.88, "grad_norm": 13.407122611999512, "learning_rate": 7.754277212531858e-08, "loss": 0.7794, "step": 17919 }, { "epoch": 2.88, "grad_norm": 15.262395858764648, "learning_rate": 7.732690412910137e-08, "loss": 0.669, "step": 17920 }, { "epoch": 2.88, "grad_norm": 11.417848587036133, "learning_rate": 7.711133585842968e-08, "loss": 0.8983, "step": 17921 }, { "epoch": 2.89, "grad_norm": 15.135976791381836, "learning_rate": 7.689606731981936e-08, "loss": 0.6734, "step": 17922 }, { "epoch": 2.89, "grad_norm": 14.691567420959473, "learning_rate": 7.668109851976968e-08, "loss": 0.8613, "step": 17923 }, { "epoch": 2.89, "grad_norm": 12.063918113708496, "learning_rate": 7.646642946477545e-08, "loss": 0.8761, "step": 17924 }, { "epoch": 2.89, "grad_norm": 13.324129104614258, "learning_rate": 7.625206016132037e-08, "loss": 0.7607, "step": 17925 }, { "epoch": 2.89, "grad_norm": 10.879660606384277, "learning_rate": 7.603799061588036e-08, "loss": 0.5507, "step": 17926 }, { "epoch": 2.89, "grad_norm": 16.082273483276367, "learning_rate": 7.582422083492136e-08, "loss": 0.7061, "step": 17927 }, { "epoch": 2.89, "grad_norm": 13.717996597290039, "learning_rate": 7.561075082490043e-08, "loss": 0.8367, "step": 17928 }, { "epoch": 2.89, "grad_norm": 13.38643741607666, "learning_rate": 7.539758059226465e-08, "loss": 0.7301, "step": 17929 }, { "epoch": 2.89, "grad_norm": 12.53702449798584, "learning_rate": 7.518471014345551e-08, "loss": 0.7305, "step": 17930 }, { "epoch": 2.89, "grad_norm": 12.429471969604492, "learning_rate": 7.497213948490012e-08, "loss": 0.5737, "step": 17931 }, { "epoch": 2.89, "grad_norm": 13.891256332397461, "learning_rate": 7.475986862302221e-08, "loss": 0.6482, "step": 17932 }, { "epoch": 2.89, "grad_norm": 13.739278793334961, "learning_rate": 7.45478975642322e-08, "loss": 0.5979, "step": 17933 }, { "epoch": 2.89, "grad_norm": 14.4802827835083, "learning_rate": 7.433622631493276e-08, "loss": 0.6837, "step": 17934 }, { "epoch": 2.89, "grad_norm": 13.040003776550293, "learning_rate": 7.412485488151767e-08, "loss": 0.9378, "step": 17935 }, { "epoch": 2.89, "grad_norm": 14.924588203430176, "learning_rate": 7.39137832703718e-08, "loss": 0.7758, "step": 17936 }, { "epoch": 2.89, "grad_norm": 15.000533103942871, "learning_rate": 7.370301148787117e-08, "loss": 0.778, "step": 17937 }, { "epoch": 2.89, "grad_norm": 13.293333053588867, "learning_rate": 7.349253954038294e-08, "loss": 0.9047, "step": 17938 }, { "epoch": 2.89, "grad_norm": 10.313798904418945, "learning_rate": 7.328236743426309e-08, "loss": 0.6041, "step": 17939 }, { "epoch": 2.89, "grad_norm": 11.450403213500977, "learning_rate": 7.307249517586101e-08, "loss": 0.7038, "step": 17940 }, { "epoch": 2.89, "grad_norm": 12.132770538330078, "learning_rate": 7.286292277151607e-08, "loss": 0.7219, "step": 17941 }, { "epoch": 2.89, "grad_norm": 13.861038208007812, "learning_rate": 7.265365022755766e-08, "loss": 0.7038, "step": 17942 }, { "epoch": 2.89, "grad_norm": 12.663773536682129, "learning_rate": 7.24446775503096e-08, "loss": 0.5917, "step": 17943 }, { "epoch": 2.89, "grad_norm": 15.120028495788574, "learning_rate": 7.223600474608128e-08, "loss": 0.7312, "step": 17944 }, { "epoch": 2.89, "grad_norm": 14.091525077819824, "learning_rate": 7.202763182117766e-08, "loss": 0.7293, "step": 17945 }, { "epoch": 2.89, "grad_norm": 12.531930923461914, "learning_rate": 7.181955878189262e-08, "loss": 0.5908, "step": 17946 }, { "epoch": 2.89, "grad_norm": 13.500904083251953, "learning_rate": 7.161178563450999e-08, "loss": 0.7896, "step": 17947 }, { "epoch": 2.89, "grad_norm": 11.056924819946289, "learning_rate": 7.140431238530809e-08, "loss": 0.53, "step": 17948 }, { "epoch": 2.89, "grad_norm": 9.807199478149414, "learning_rate": 7.119713904055193e-08, "loss": 0.7609, "step": 17949 }, { "epoch": 2.89, "grad_norm": 15.948718070983887, "learning_rate": 7.09902656064998e-08, "loss": 0.6751, "step": 17950 }, { "epoch": 2.89, "grad_norm": 12.026347160339355, "learning_rate": 7.078369208940228e-08, "loss": 0.6533, "step": 17951 }, { "epoch": 2.89, "grad_norm": 14.513198852539062, "learning_rate": 7.05774184954966e-08, "loss": 0.8522, "step": 17952 }, { "epoch": 2.89, "grad_norm": 14.61759090423584, "learning_rate": 7.037144483101444e-08, "loss": 0.7859, "step": 17953 }, { "epoch": 2.89, "grad_norm": 13.899045944213867, "learning_rate": 7.016577110217748e-08, "loss": 0.7335, "step": 17954 }, { "epoch": 2.89, "grad_norm": 13.727455139160156, "learning_rate": 6.996039731519855e-08, "loss": 0.9578, "step": 17955 }, { "epoch": 2.89, "grad_norm": 11.178793907165527, "learning_rate": 6.975532347628156e-08, "loss": 0.6289, "step": 17956 }, { "epoch": 2.89, "grad_norm": 12.880179405212402, "learning_rate": 6.955054959162044e-08, "loss": 0.695, "step": 17957 }, { "epoch": 2.89, "grad_norm": 18.35594367980957, "learning_rate": 6.934607566740026e-08, "loss": 0.7465, "step": 17958 }, { "epoch": 2.89, "grad_norm": 13.728365898132324, "learning_rate": 6.914190170979829e-08, "loss": 0.6765, "step": 17959 }, { "epoch": 2.89, "grad_norm": 11.575758934020996, "learning_rate": 6.893802772498071e-08, "loss": 0.5723, "step": 17960 }, { "epoch": 2.89, "grad_norm": 14.48218822479248, "learning_rate": 6.873445371910703e-08, "loss": 0.8252, "step": 17961 }, { "epoch": 2.89, "grad_norm": 11.216116905212402, "learning_rate": 6.853117969832568e-08, "loss": 0.6784, "step": 17962 }, { "epoch": 2.89, "grad_norm": 17.498830795288086, "learning_rate": 6.832820566877729e-08, "loss": 0.83, "step": 17963 }, { "epoch": 2.89, "grad_norm": 12.041218757629395, "learning_rate": 6.812553163659253e-08, "loss": 0.7204, "step": 17964 }, { "epoch": 2.89, "grad_norm": 11.595818519592285, "learning_rate": 6.792315760789315e-08, "loss": 0.7805, "step": 17965 }, { "epoch": 2.89, "grad_norm": 12.414258003234863, "learning_rate": 6.772108358879314e-08, "loss": 0.6634, "step": 17966 }, { "epoch": 2.89, "grad_norm": 13.438610076904297, "learning_rate": 6.751930958539432e-08, "loss": 0.5705, "step": 17967 }, { "epoch": 2.89, "grad_norm": 10.770030975341797, "learning_rate": 6.731783560379401e-08, "loss": 0.7183, "step": 17968 }, { "epoch": 2.89, "grad_norm": 15.983882904052734, "learning_rate": 6.711666165007625e-08, "loss": 0.9148, "step": 17969 }, { "epoch": 2.89, "grad_norm": 12.74716854095459, "learning_rate": 6.691578773031837e-08, "loss": 0.8843, "step": 17970 }, { "epoch": 2.89, "grad_norm": 11.228052139282227, "learning_rate": 6.671521385058776e-08, "loss": 0.8031, "step": 17971 }, { "epoch": 2.89, "grad_norm": 12.776596069335938, "learning_rate": 6.651494001694291e-08, "loss": 0.8669, "step": 17972 }, { "epoch": 2.89, "grad_norm": 13.939168930053711, "learning_rate": 6.631496623543343e-08, "loss": 0.5612, "step": 17973 }, { "epoch": 2.89, "grad_norm": 14.41109848022461, "learning_rate": 6.611529251210002e-08, "loss": 0.4636, "step": 17974 }, { "epoch": 2.89, "grad_norm": 10.920304298400879, "learning_rate": 6.591591885297344e-08, "loss": 0.5809, "step": 17975 }, { "epoch": 2.89, "grad_norm": 10.672182083129883, "learning_rate": 6.571684526407773e-08, "loss": 0.7924, "step": 17976 }, { "epoch": 2.89, "grad_norm": 15.788121223449707, "learning_rate": 6.551807175142366e-08, "loss": 0.9421, "step": 17977 }, { "epoch": 2.89, "grad_norm": 13.916746139526367, "learning_rate": 6.531959832101642e-08, "loss": 0.7218, "step": 17978 }, { "epoch": 2.89, "grad_norm": 15.519804000854492, "learning_rate": 6.512142497885121e-08, "loss": 1.0006, "step": 17979 }, { "epoch": 2.89, "grad_norm": 13.641891479492188, "learning_rate": 6.492355173091436e-08, "loss": 0.6466, "step": 17980 }, { "epoch": 2.89, "grad_norm": 14.271591186523438, "learning_rate": 6.472597858318219e-08, "loss": 0.6388, "step": 17981 }, { "epoch": 2.89, "grad_norm": 12.331515312194824, "learning_rate": 6.452870554162328e-08, "loss": 0.7026, "step": 17982 }, { "epoch": 2.89, "grad_norm": 13.915984153747559, "learning_rate": 6.433173261219728e-08, "loss": 0.5954, "step": 17983 }, { "epoch": 2.9, "grad_norm": 11.825956344604492, "learning_rate": 6.413505980085167e-08, "loss": 0.8559, "step": 17984 }, { "epoch": 2.9, "grad_norm": 11.737167358398438, "learning_rate": 6.393868711352946e-08, "loss": 0.7873, "step": 17985 }, { "epoch": 2.9, "grad_norm": 13.49916934967041, "learning_rate": 6.374261455616037e-08, "loss": 0.5881, "step": 17986 }, { "epoch": 2.9, "grad_norm": 12.035187721252441, "learning_rate": 6.354684213466855e-08, "loss": 0.7391, "step": 17987 }, { "epoch": 2.9, "grad_norm": 12.258048057556152, "learning_rate": 6.335136985496704e-08, "loss": 0.7689, "step": 17988 }, { "epoch": 2.9, "grad_norm": 16.19281578063965, "learning_rate": 6.315619772295889e-08, "loss": 0.7304, "step": 17989 }, { "epoch": 2.9, "grad_norm": 12.598958015441895, "learning_rate": 6.296132574454161e-08, "loss": 0.6096, "step": 17990 }, { "epoch": 2.9, "grad_norm": 11.797625541687012, "learning_rate": 6.27667539256016e-08, "loss": 0.5768, "step": 17991 }, { "epoch": 2.9, "grad_norm": 12.5440673828125, "learning_rate": 6.257248227201418e-08, "loss": 0.6394, "step": 17992 }, { "epoch": 2.9, "grad_norm": 16.863924026489258, "learning_rate": 6.237851078964796e-08, "loss": 0.529, "step": 17993 }, { "epoch": 2.9, "grad_norm": 12.854960441589355, "learning_rate": 6.218483948436382e-08, "loss": 0.5264, "step": 17994 }, { "epoch": 2.9, "grad_norm": 14.788203239440918, "learning_rate": 6.199146836201043e-08, "loss": 0.8065, "step": 17995 }, { "epoch": 2.9, "grad_norm": 13.116286277770996, "learning_rate": 6.179839742842864e-08, "loss": 0.4753, "step": 17996 }, { "epoch": 2.9, "grad_norm": 14.08203125, "learning_rate": 6.160562668945158e-08, "loss": 0.5877, "step": 17997 }, { "epoch": 2.9, "grad_norm": 14.604032516479492, "learning_rate": 6.141315615090016e-08, "loss": 0.7433, "step": 17998 }, { "epoch": 2.9, "grad_norm": 15.240473747253418, "learning_rate": 6.12209858185897e-08, "loss": 0.7241, "step": 17999 }, { "epoch": 2.9, "grad_norm": 15.872359275817871, "learning_rate": 6.102911569832449e-08, "loss": 0.5532, "step": 18000 }, { "epoch": 2.9, "grad_norm": 11.288747787475586, "learning_rate": 6.083754579590095e-08, "loss": 0.8253, "step": 18001 }, { "epoch": 2.9, "grad_norm": 13.906719207763672, "learning_rate": 6.06462761171045e-08, "loss": 0.5747, "step": 18002 }, { "epoch": 2.9, "grad_norm": 13.402262687683105, "learning_rate": 6.04553066677127e-08, "loss": 0.6945, "step": 18003 }, { "epoch": 2.9, "grad_norm": 13.982550621032715, "learning_rate": 6.026463745349432e-08, "loss": 0.6537, "step": 18004 }, { "epoch": 2.9, "grad_norm": 16.972002029418945, "learning_rate": 6.007426848020914e-08, "loss": 0.6673, "step": 18005 }, { "epoch": 2.9, "grad_norm": 14.725257873535156, "learning_rate": 5.988419975360815e-08, "loss": 0.7715, "step": 18006 }, { "epoch": 2.9, "grad_norm": 11.677030563354492, "learning_rate": 5.969443127943009e-08, "loss": 0.5361, "step": 18007 }, { "epoch": 2.9, "grad_norm": 11.094731330871582, "learning_rate": 5.950496306340925e-08, "loss": 0.7246, "step": 18008 }, { "epoch": 2.9, "grad_norm": 15.789285659790039, "learning_rate": 5.931579511126773e-08, "loss": 0.6872, "step": 18009 }, { "epoch": 2.9, "grad_norm": 14.988811492919922, "learning_rate": 5.912692742871984e-08, "loss": 0.7007, "step": 18010 }, { "epoch": 2.9, "grad_norm": 12.24384880065918, "learning_rate": 5.893836002147102e-08, "loss": 0.8105, "step": 18011 }, { "epoch": 2.9, "grad_norm": 13.360861778259277, "learning_rate": 5.875009289521672e-08, "loss": 0.6093, "step": 18012 }, { "epoch": 2.9, "grad_norm": 13.306947708129883, "learning_rate": 5.85621260556446e-08, "loss": 0.6755, "step": 18013 }, { "epoch": 2.9, "grad_norm": 13.584049224853516, "learning_rate": 5.8374459508430125e-08, "loss": 0.6866, "step": 18014 }, { "epoch": 2.9, "grad_norm": 12.176275253295898, "learning_rate": 5.8187093259244323e-08, "loss": 0.8713, "step": 18015 }, { "epoch": 2.9, "grad_norm": 14.414400100708008, "learning_rate": 5.800002731374599e-08, "loss": 0.8296, "step": 18016 }, { "epoch": 2.9, "grad_norm": 14.483294486999512, "learning_rate": 5.7813261677585055e-08, "loss": 0.7525, "step": 18017 }, { "epoch": 2.9, "grad_norm": 13.682679176330566, "learning_rate": 5.762679635640367e-08, "loss": 0.8667, "step": 18018 }, { "epoch": 2.9, "grad_norm": 16.355310440063477, "learning_rate": 5.744063135583511e-08, "loss": 0.7252, "step": 18019 }, { "epoch": 2.9, "grad_norm": 13.328461647033691, "learning_rate": 5.725476668150043e-08, "loss": 0.5796, "step": 18020 }, { "epoch": 2.9, "grad_norm": 14.078534126281738, "learning_rate": 5.706920233901625e-08, "loss": 0.7698, "step": 18021 }, { "epoch": 2.9, "grad_norm": 12.89498233795166, "learning_rate": 5.688393833398587e-08, "loss": 0.6698, "step": 18022 }, { "epoch": 2.9, "grad_norm": 14.31320858001709, "learning_rate": 5.669897467200591e-08, "loss": 0.5982, "step": 18023 }, { "epoch": 2.9, "grad_norm": 13.045031547546387, "learning_rate": 5.651431135866525e-08, "loss": 0.7346, "step": 18024 }, { "epoch": 2.9, "grad_norm": 12.698686599731445, "learning_rate": 5.632994839953831e-08, "loss": 0.8006, "step": 18025 }, { "epoch": 2.9, "grad_norm": 10.768258094787598, "learning_rate": 5.614588580019731e-08, "loss": 0.6756, "step": 18026 }, { "epoch": 2.9, "grad_norm": 13.687732696533203, "learning_rate": 5.5962123566201124e-08, "loss": 0.6377, "step": 18027 }, { "epoch": 2.9, "grad_norm": 11.991080284118652, "learning_rate": 5.5778661703098644e-08, "loss": 0.5282, "step": 18028 }, { "epoch": 2.9, "grad_norm": 12.414840698242188, "learning_rate": 5.559550021643434e-08, "loss": 0.5255, "step": 18029 }, { "epoch": 2.9, "grad_norm": 16.576873779296875, "learning_rate": 5.541263911173822e-08, "loss": 0.6946, "step": 18030 }, { "epoch": 2.9, "grad_norm": 12.324524879455566, "learning_rate": 5.523007839453587e-08, "loss": 0.8129, "step": 18031 }, { "epoch": 2.9, "grad_norm": 16.0091609954834, "learning_rate": 5.504781807034065e-08, "loss": 0.7227, "step": 18032 }, { "epoch": 2.9, "grad_norm": 23.970727920532227, "learning_rate": 5.486585814465817e-08, "loss": 0.647, "step": 18033 }, { "epoch": 2.9, "grad_norm": 14.34186840057373, "learning_rate": 5.468419862298513e-08, "loss": 0.7418, "step": 18034 }, { "epoch": 2.9, "grad_norm": 12.722845077514648, "learning_rate": 5.450283951080826e-08, "loss": 0.6594, "step": 18035 }, { "epoch": 2.9, "grad_norm": 16.833709716796875, "learning_rate": 5.43217808136054e-08, "loss": 0.6219, "step": 18036 }, { "epoch": 2.9, "grad_norm": 15.57608413696289, "learning_rate": 5.414102253684661e-08, "loss": 0.7887, "step": 18037 }, { "epoch": 2.9, "grad_norm": 10.363922119140625, "learning_rate": 5.396056468599087e-08, "loss": 0.7665, "step": 18038 }, { "epoch": 2.9, "grad_norm": 14.558969497680664, "learning_rate": 5.378040726649048e-08, "loss": 0.61, "step": 18039 }, { "epoch": 2.9, "grad_norm": 12.40320110321045, "learning_rate": 5.3600550283785525e-08, "loss": 0.7351, "step": 18040 }, { "epoch": 2.9, "grad_norm": 13.158323287963867, "learning_rate": 5.3420993743310554e-08, "loss": 0.7417, "step": 18041 }, { "epoch": 2.9, "grad_norm": 14.428232192993164, "learning_rate": 5.32417376504879e-08, "loss": 0.714, "step": 18042 }, { "epoch": 2.9, "grad_norm": 12.140799522399902, "learning_rate": 5.3062782010733224e-08, "loss": 0.7766, "step": 18043 }, { "epoch": 2.9, "grad_norm": 13.072244644165039, "learning_rate": 5.28841268294511e-08, "loss": 0.6144, "step": 18044 }, { "epoch": 2.9, "grad_norm": 11.707916259765625, "learning_rate": 5.270577211203942e-08, "loss": 0.7344, "step": 18045 }, { "epoch": 2.91, "grad_norm": 15.93117618560791, "learning_rate": 5.2527717863883885e-08, "loss": 0.5654, "step": 18046 }, { "epoch": 2.91, "grad_norm": 15.883872985839844, "learning_rate": 5.2349964090363524e-08, "loss": 0.8806, "step": 18047 }, { "epoch": 2.91, "grad_norm": 11.860645294189453, "learning_rate": 5.217251079684737e-08, "loss": 0.7694, "step": 18048 }, { "epoch": 2.91, "grad_norm": 14.355692863464355, "learning_rate": 5.19953579886967e-08, "loss": 0.6826, "step": 18049 }, { "epoch": 2.91, "grad_norm": 18.994707107543945, "learning_rate": 5.181850567126168e-08, "loss": 0.8455, "step": 18050 }, { "epoch": 2.91, "grad_norm": 12.564094543457031, "learning_rate": 5.1641953849884684e-08, "loss": 0.7928, "step": 18051 }, { "epoch": 2.91, "grad_norm": 12.146014213562012, "learning_rate": 5.146570252989702e-08, "loss": 0.8864, "step": 18052 }, { "epoch": 2.91, "grad_norm": 10.505155563354492, "learning_rate": 5.1289751716625536e-08, "loss": 0.6102, "step": 18053 }, { "epoch": 2.91, "grad_norm": 11.063529014587402, "learning_rate": 5.111410141538375e-08, "loss": 0.8689, "step": 18054 }, { "epoch": 2.91, "grad_norm": 11.368171691894531, "learning_rate": 5.093875163147521e-08, "loss": 0.7841, "step": 18055 }, { "epoch": 2.91, "grad_norm": 11.401674270629883, "learning_rate": 5.076370237020012e-08, "loss": 0.791, "step": 18056 }, { "epoch": 2.91, "grad_norm": 10.933671951293945, "learning_rate": 5.058895363684313e-08, "loss": 0.7116, "step": 18057 }, { "epoch": 2.91, "grad_norm": 10.488622665405273, "learning_rate": 5.041450543668558e-08, "loss": 0.6188, "step": 18058 }, { "epoch": 2.91, "grad_norm": 15.579252243041992, "learning_rate": 5.0240357774994367e-08, "loss": 0.7562, "step": 18059 }, { "epoch": 2.91, "grad_norm": 12.40957260131836, "learning_rate": 5.006651065702972e-08, "loss": 0.6726, "step": 18060 }, { "epoch": 2.91, "grad_norm": 12.622690200805664, "learning_rate": 4.989296408804523e-08, "loss": 0.7001, "step": 18061 }, { "epoch": 2.91, "grad_norm": 19.53687286376953, "learning_rate": 4.971971807328113e-08, "loss": 0.7721, "step": 18062 }, { "epoch": 2.91, "grad_norm": 11.63586139678955, "learning_rate": 4.9546772617971025e-08, "loss": 0.597, "step": 18063 }, { "epoch": 2.91, "grad_norm": 11.842578887939453, "learning_rate": 4.937412772733963e-08, "loss": 0.5597, "step": 18064 }, { "epoch": 2.91, "grad_norm": 14.054990768432617, "learning_rate": 4.920178340660164e-08, "loss": 0.9986, "step": 18065 }, { "epoch": 2.91, "grad_norm": 14.11400032043457, "learning_rate": 4.9029739660960696e-08, "loss": 0.7233, "step": 18066 }, { "epoch": 2.91, "grad_norm": 13.732521057128906, "learning_rate": 4.885799649561707e-08, "loss": 0.7289, "step": 18067 }, { "epoch": 2.91, "grad_norm": 16.84702491760254, "learning_rate": 4.8686553915755494e-08, "loss": 0.7538, "step": 18068 }, { "epoch": 2.91, "grad_norm": 11.738222122192383, "learning_rate": 4.85154119265574e-08, "loss": 0.6314, "step": 18069 }, { "epoch": 2.91, "grad_norm": 12.02350902557373, "learning_rate": 4.834457053318975e-08, "loss": 0.8237, "step": 18070 }, { "epoch": 2.91, "grad_norm": 12.922024726867676, "learning_rate": 4.8174029740813975e-08, "loss": 0.8823, "step": 18071 }, { "epoch": 2.91, "grad_norm": 14.449167251586914, "learning_rate": 4.8003789554582626e-08, "loss": 0.7008, "step": 18072 }, { "epoch": 2.91, "grad_norm": 14.519124984741211, "learning_rate": 4.7833849979636026e-08, "loss": 0.6674, "step": 18073 }, { "epoch": 2.91, "grad_norm": 12.753299713134766, "learning_rate": 4.7664211021107856e-08, "loss": 0.6487, "step": 18074 }, { "epoch": 2.91, "grad_norm": 14.027508735656738, "learning_rate": 4.74948726841229e-08, "loss": 0.7222, "step": 18075 }, { "epoch": 2.91, "grad_norm": 12.443041801452637, "learning_rate": 4.732583497379595e-08, "loss": 0.5969, "step": 18076 }, { "epoch": 2.91, "grad_norm": 13.453248023986816, "learning_rate": 4.7157097895234037e-08, "loss": 0.8799, "step": 18077 }, { "epoch": 2.91, "grad_norm": 12.328461647033691, "learning_rate": 4.698866145353198e-08, "loss": 0.7063, "step": 18078 }, { "epoch": 2.91, "grad_norm": 12.971776008605957, "learning_rate": 4.6820525653779036e-08, "loss": 0.6786, "step": 18079 }, { "epoch": 2.91, "grad_norm": 11.74740219116211, "learning_rate": 4.665269050105337e-08, "loss": 0.7885, "step": 18080 }, { "epoch": 2.91, "grad_norm": 15.030835151672363, "learning_rate": 4.6485156000425355e-08, "loss": 0.737, "step": 18081 }, { "epoch": 2.91, "grad_norm": 15.50411319732666, "learning_rate": 4.63179221569543e-08, "loss": 0.8461, "step": 18082 }, { "epoch": 2.91, "grad_norm": 12.513697624206543, "learning_rate": 4.6150988975692815e-08, "loss": 0.7276, "step": 18083 }, { "epoch": 2.91, "grad_norm": 14.867579460144043, "learning_rate": 4.598435646168242e-08, "loss": 0.8944, "step": 18084 }, { "epoch": 2.91, "grad_norm": 10.344060897827148, "learning_rate": 4.581802461995688e-08, "loss": 0.5843, "step": 18085 }, { "epoch": 2.91, "grad_norm": 14.094297409057617, "learning_rate": 4.5651993455541054e-08, "loss": 0.5504, "step": 18086 }, { "epoch": 2.91, "grad_norm": 11.019305229187012, "learning_rate": 4.5486262973448716e-08, "loss": 0.6536, "step": 18087 }, { "epoch": 2.91, "grad_norm": 10.228381156921387, "learning_rate": 4.5320833178688074e-08, "loss": 0.616, "step": 18088 }, { "epoch": 2.91, "grad_norm": 14.544503211975098, "learning_rate": 4.5155704076254024e-08, "loss": 0.8753, "step": 18089 }, { "epoch": 2.91, "grad_norm": 13.150290489196777, "learning_rate": 4.49908756711348e-08, "loss": 0.6502, "step": 18090 }, { "epoch": 2.91, "grad_norm": 15.243496894836426, "learning_rate": 4.482634796830976e-08, "loss": 0.7158, "step": 18091 }, { "epoch": 2.91, "grad_norm": 15.231287002563477, "learning_rate": 4.466212097274936e-08, "loss": 0.8305, "step": 18092 }, { "epoch": 2.91, "grad_norm": 11.93482780456543, "learning_rate": 4.449819468941297e-08, "loss": 0.8493, "step": 18093 }, { "epoch": 2.91, "grad_norm": 13.153339385986328, "learning_rate": 4.433456912325218e-08, "loss": 0.5703, "step": 18094 }, { "epoch": 2.91, "grad_norm": 11.160998344421387, "learning_rate": 4.417124427921082e-08, "loss": 0.5879, "step": 18095 }, { "epoch": 2.91, "grad_norm": 11.576652526855469, "learning_rate": 4.4008220162220504e-08, "loss": 0.8051, "step": 18096 }, { "epoch": 2.91, "grad_norm": 15.053762435913086, "learning_rate": 4.384549677720729e-08, "loss": 0.677, "step": 18097 }, { "epoch": 2.91, "grad_norm": 15.05565071105957, "learning_rate": 4.3683074129085014e-08, "loss": 0.64, "step": 18098 }, { "epoch": 2.91, "grad_norm": 14.354764938354492, "learning_rate": 4.352095222276087e-08, "loss": 0.8278, "step": 18099 }, { "epoch": 2.91, "grad_norm": 13.084097862243652, "learning_rate": 4.3359131063132056e-08, "loss": 0.6016, "step": 18100 }, { "epoch": 2.91, "grad_norm": 13.341669082641602, "learning_rate": 4.3197610655084656e-08, "loss": 0.5556, "step": 18101 }, { "epoch": 2.91, "grad_norm": 14.183496475219727, "learning_rate": 4.303639100350032e-08, "loss": 0.8237, "step": 18102 }, { "epoch": 2.91, "grad_norm": 13.298700332641602, "learning_rate": 4.287547211324628e-08, "loss": 0.9384, "step": 18103 }, { "epoch": 2.91, "grad_norm": 13.87012767791748, "learning_rate": 4.271485398918529e-08, "loss": 0.6867, "step": 18104 }, { "epoch": 2.91, "grad_norm": 12.888526916503906, "learning_rate": 4.255453663616793e-08, "loss": 0.7404, "step": 18105 }, { "epoch": 2.91, "grad_norm": 13.021793365478516, "learning_rate": 4.2394520059036995e-08, "loss": 0.6296, "step": 18106 }, { "epoch": 2.91, "grad_norm": 15.182023048400879, "learning_rate": 4.223480426262638e-08, "loss": 0.7967, "step": 18107 }, { "epoch": 2.92, "grad_norm": 14.358070373535156, "learning_rate": 4.207538925176002e-08, "loss": 0.7996, "step": 18108 }, { "epoch": 2.92, "grad_norm": 13.55644416809082, "learning_rate": 4.191627503125295e-08, "loss": 0.4839, "step": 18109 }, { "epoch": 2.92, "grad_norm": 14.370988845825195, "learning_rate": 4.175746160591243e-08, "loss": 0.9661, "step": 18110 }, { "epoch": 2.92, "grad_norm": 15.684582710266113, "learning_rate": 4.1598948980533516e-08, "loss": 0.7017, "step": 18111 }, { "epoch": 2.92, "grad_norm": 11.524748802185059, "learning_rate": 4.1440737159906816e-08, "loss": 0.728, "step": 18112 }, { "epoch": 2.92, "grad_norm": 15.39791488647461, "learning_rate": 4.128282614881074e-08, "loss": 0.7211, "step": 18113 }, { "epoch": 2.92, "grad_norm": 13.25031852722168, "learning_rate": 4.1125215952013686e-08, "loss": 0.7615, "step": 18114 }, { "epoch": 2.92, "grad_norm": 15.660287857055664, "learning_rate": 4.09679065742774e-08, "loss": 0.8386, "step": 18115 }, { "epoch": 2.92, "grad_norm": 12.535170555114746, "learning_rate": 4.081089802035254e-08, "loss": 0.8418, "step": 18116 }, { "epoch": 2.92, "grad_norm": 11.217687606811523, "learning_rate": 4.065419029498419e-08, "loss": 0.7001, "step": 18117 }, { "epoch": 2.92, "grad_norm": 12.622515678405762, "learning_rate": 4.049778340290411e-08, "loss": 0.6853, "step": 18118 }, { "epoch": 2.92, "grad_norm": 13.568390846252441, "learning_rate": 4.034167734883743e-08, "loss": 0.7692, "step": 18119 }, { "epoch": 2.92, "grad_norm": 14.80380916595459, "learning_rate": 4.0185872137498136e-08, "loss": 0.8557, "step": 18120 }, { "epoch": 2.92, "grad_norm": 13.001415252685547, "learning_rate": 4.003036777359359e-08, "loss": 0.8754, "step": 18121 }, { "epoch": 2.92, "grad_norm": 12.567269325256348, "learning_rate": 3.9875164261821144e-08, "loss": 0.5917, "step": 18122 }, { "epoch": 2.92, "grad_norm": 11.64382266998291, "learning_rate": 3.972026160686926e-08, "loss": 0.705, "step": 18123 }, { "epoch": 2.92, "grad_norm": 12.76932144165039, "learning_rate": 3.9565659813415314e-08, "loss": 0.871, "step": 18124 }, { "epoch": 2.92, "grad_norm": 13.534687995910645, "learning_rate": 3.9411358886131124e-08, "loss": 0.5597, "step": 18125 }, { "epoch": 2.92, "grad_norm": 17.8497371673584, "learning_rate": 3.9257358829676296e-08, "loss": 0.6855, "step": 18126 }, { "epoch": 2.92, "grad_norm": 12.15739917755127, "learning_rate": 3.910365964870266e-08, "loss": 0.686, "step": 18127 }, { "epoch": 2.92, "grad_norm": 13.022040367126465, "learning_rate": 3.895026134785318e-08, "loss": 0.629, "step": 18128 }, { "epoch": 2.92, "grad_norm": 29.163881301879883, "learning_rate": 3.879716393176191e-08, "loss": 0.6681, "step": 18129 }, { "epoch": 2.92, "grad_norm": 13.663366317749023, "learning_rate": 3.864436740505184e-08, "loss": 0.6262, "step": 18130 }, { "epoch": 2.92, "grad_norm": 12.599732398986816, "learning_rate": 3.849187177234037e-08, "loss": 0.9167, "step": 18131 }, { "epoch": 2.92, "grad_norm": 14.084474563598633, "learning_rate": 3.8339677038231604e-08, "loss": 0.8138, "step": 18132 }, { "epoch": 2.92, "grad_norm": 12.42926025390625, "learning_rate": 3.818778320732519e-08, "loss": 0.7455, "step": 18133 }, { "epoch": 2.92, "grad_norm": 12.56714916229248, "learning_rate": 3.8036190284207464e-08, "loss": 0.6436, "step": 18134 }, { "epoch": 2.92, "grad_norm": 13.205321311950684, "learning_rate": 3.788489827345698e-08, "loss": 0.6223, "step": 18135 }, { "epoch": 2.92, "grad_norm": 13.848678588867188, "learning_rate": 3.773390717964565e-08, "loss": 0.666, "step": 18136 }, { "epoch": 2.92, "grad_norm": 13.369233131408691, "learning_rate": 3.758321700733314e-08, "loss": 0.6854, "step": 18137 }, { "epoch": 2.92, "grad_norm": 9.766326904296875, "learning_rate": 3.7432827761071375e-08, "loss": 0.7522, "step": 18138 }, { "epoch": 2.92, "grad_norm": 11.7957124710083, "learning_rate": 3.728273944540339e-08, "loss": 0.428, "step": 18139 }, { "epoch": 2.92, "grad_norm": 16.754533767700195, "learning_rate": 3.7132952064862227e-08, "loss": 0.7668, "step": 18140 }, { "epoch": 2.92, "grad_norm": 13.547317504882812, "learning_rate": 3.6983465623972035e-08, "loss": 0.6968, "step": 18141 }, { "epoch": 2.92, "grad_norm": 12.761724472045898, "learning_rate": 3.683428012724921e-08, "loss": 0.8279, "step": 18142 }, { "epoch": 2.92, "grad_norm": 12.247265815734863, "learning_rate": 3.668539557920015e-08, "loss": 0.6405, "step": 18143 }, { "epoch": 2.92, "grad_norm": 11.79145622253418, "learning_rate": 3.653681198432124e-08, "loss": 0.5685, "step": 18144 }, { "epoch": 2.92, "grad_norm": 13.643294334411621, "learning_rate": 3.638852934710113e-08, "loss": 0.6891, "step": 18145 }, { "epoch": 2.92, "grad_norm": 12.775535583496094, "learning_rate": 3.6240547672018455e-08, "loss": 0.8461, "step": 18146 }, { "epoch": 2.92, "grad_norm": 15.642449378967285, "learning_rate": 3.609286696354297e-08, "loss": 0.8433, "step": 18147 }, { "epoch": 2.92, "grad_norm": 12.538681983947754, "learning_rate": 3.594548722613667e-08, "loss": 0.8442, "step": 18148 }, { "epoch": 2.92, "grad_norm": 11.496614456176758, "learning_rate": 3.5798408464250425e-08, "loss": 0.6468, "step": 18149 }, { "epoch": 2.92, "grad_norm": 12.189624786376953, "learning_rate": 3.565163068232735e-08, "loss": 0.6757, "step": 18150 }, { "epoch": 2.92, "grad_norm": 12.072832107543945, "learning_rate": 3.550515388480058e-08, "loss": 0.8229, "step": 18151 }, { "epoch": 2.92, "grad_norm": 12.39178466796875, "learning_rate": 3.5358978076095454e-08, "loss": 0.7357, "step": 18152 }, { "epoch": 2.92, "grad_norm": 13.435826301574707, "learning_rate": 3.5213103260626217e-08, "loss": 0.8268, "step": 18153 }, { "epoch": 2.92, "grad_norm": 12.598527908325195, "learning_rate": 3.506752944279934e-08, "loss": 0.524, "step": 18154 }, { "epoch": 2.92, "grad_norm": 12.485937118530273, "learning_rate": 3.492225662701243e-08, "loss": 0.4962, "step": 18155 }, { "epoch": 2.92, "grad_norm": 12.839794158935547, "learning_rate": 3.477728481765419e-08, "loss": 0.5128, "step": 18156 }, { "epoch": 2.92, "grad_norm": 12.168061256408691, "learning_rate": 3.463261401910334e-08, "loss": 1.0217, "step": 18157 }, { "epoch": 2.92, "grad_norm": 12.343234062194824, "learning_rate": 3.448824423572861e-08, "loss": 0.6987, "step": 18158 }, { "epoch": 2.92, "grad_norm": 12.453084945678711, "learning_rate": 3.434417547189317e-08, "loss": 0.6043, "step": 18159 }, { "epoch": 2.92, "grad_norm": 17.335920333862305, "learning_rate": 3.4200407731945776e-08, "loss": 0.6925, "step": 18160 }, { "epoch": 2.92, "grad_norm": 11.944136619567871, "learning_rate": 3.405694102023183e-08, "loss": 0.6115, "step": 18161 }, { "epoch": 2.92, "grad_norm": 12.689346313476562, "learning_rate": 3.391377534108342e-08, "loss": 0.7173, "step": 18162 }, { "epoch": 2.92, "grad_norm": 14.105360984802246, "learning_rate": 3.377091069882599e-08, "loss": 0.6293, "step": 18163 }, { "epoch": 2.92, "grad_norm": 13.468061447143555, "learning_rate": 3.362834709777385e-08, "loss": 0.7599, "step": 18164 }, { "epoch": 2.92, "grad_norm": 13.999707221984863, "learning_rate": 3.348608454223357e-08, "loss": 0.7138, "step": 18165 }, { "epoch": 2.92, "grad_norm": 13.284599304199219, "learning_rate": 3.3344123036502806e-08, "loss": 0.686, "step": 18166 }, { "epoch": 2.92, "grad_norm": 12.46347713470459, "learning_rate": 3.3202462584870364e-08, "loss": 0.7247, "step": 18167 }, { "epoch": 2.92, "grad_norm": 10.072062492370605, "learning_rate": 3.306110319161282e-08, "loss": 0.6763, "step": 18168 }, { "epoch": 2.92, "grad_norm": 14.49172592163086, "learning_rate": 3.292004486100231e-08, "loss": 0.7154, "step": 18169 }, { "epoch": 2.92, "grad_norm": 11.746784210205078, "learning_rate": 3.2779287597298757e-08, "loss": 0.6805, "step": 18170 }, { "epoch": 2.93, "grad_norm": 13.755681037902832, "learning_rate": 3.2638831404755436e-08, "loss": 0.5911, "step": 18171 }, { "epoch": 2.93, "grad_norm": 11.647961616516113, "learning_rate": 3.2498676287612275e-08, "loss": 0.7548, "step": 18172 }, { "epoch": 2.93, "grad_norm": 13.37229061126709, "learning_rate": 3.235882225010478e-08, "loss": 0.6197, "step": 18173 }, { "epoch": 2.93, "grad_norm": 12.186705589294434, "learning_rate": 3.2219269296456246e-08, "loss": 0.6182, "step": 18174 }, { "epoch": 2.93, "grad_norm": 15.003739356994629, "learning_rate": 3.2080017430883293e-08, "loss": 0.6483, "step": 18175 }, { "epoch": 2.93, "grad_norm": 12.832655906677246, "learning_rate": 3.194106665759034e-08, "loss": 0.9072, "step": 18176 }, { "epoch": 2.93, "grad_norm": 15.904157638549805, "learning_rate": 3.180241698077735e-08, "loss": 0.8369, "step": 18177 }, { "epoch": 2.93, "grad_norm": 11.736681938171387, "learning_rate": 3.1664068404628764e-08, "loss": 0.8164, "step": 18178 }, { "epoch": 2.93, "grad_norm": 12.462164878845215, "learning_rate": 3.15260209333268e-08, "loss": 0.6275, "step": 18179 }, { "epoch": 2.93, "grad_norm": 45.61677932739258, "learning_rate": 3.138827457104032e-08, "loss": 0.8672, "step": 18180 }, { "epoch": 2.93, "grad_norm": 14.32400131225586, "learning_rate": 3.125082932192935e-08, "loss": 0.7934, "step": 18181 }, { "epoch": 2.93, "grad_norm": 15.427705764770508, "learning_rate": 3.1113685190146124e-08, "loss": 0.7617, "step": 18182 }, { "epoch": 2.93, "grad_norm": 15.97910213470459, "learning_rate": 3.097684217983288e-08, "loss": 0.9261, "step": 18183 }, { "epoch": 2.93, "grad_norm": 16.95047950744629, "learning_rate": 3.084030029512408e-08, "loss": 0.7612, "step": 18184 }, { "epoch": 2.93, "grad_norm": 15.043904304504395, "learning_rate": 3.070405954014311e-08, "loss": 0.9082, "step": 18185 }, { "epoch": 2.93, "grad_norm": 14.143671035766602, "learning_rate": 3.056811991900554e-08, "loss": 0.7064, "step": 18186 }, { "epoch": 2.93, "grad_norm": 12.681839942932129, "learning_rate": 3.043248143581812e-08, "loss": 0.7385, "step": 18187 }, { "epoch": 2.93, "grad_norm": 12.53005313873291, "learning_rate": 3.029714409467755e-08, "loss": 0.651, "step": 18188 }, { "epoch": 2.93, "grad_norm": 11.905235290527344, "learning_rate": 3.0162107899671664e-08, "loss": 0.6995, "step": 18189 }, { "epoch": 2.93, "grad_norm": 12.357404708862305, "learning_rate": 3.0027372854879445e-08, "loss": 0.7526, "step": 18190 }, { "epoch": 2.93, "grad_norm": 14.930429458618164, "learning_rate": 2.9892938964370954e-08, "loss": 0.6449, "step": 18191 }, { "epoch": 2.93, "grad_norm": 13.9242582321167, "learning_rate": 2.9758806232206284e-08, "loss": 0.781, "step": 18192 }, { "epoch": 2.93, "grad_norm": 12.00407600402832, "learning_rate": 2.9624974662438856e-08, "loss": 0.7969, "step": 18193 }, { "epoch": 2.93, "grad_norm": 13.008734703063965, "learning_rate": 2.949144425910877e-08, "loss": 0.6525, "step": 18194 }, { "epoch": 2.93, "grad_norm": 12.543116569519043, "learning_rate": 2.935821502625058e-08, "loss": 0.5945, "step": 18195 }, { "epoch": 2.93, "grad_norm": 13.772336959838867, "learning_rate": 2.9225286967888843e-08, "loss": 0.8216, "step": 18196 }, { "epoch": 2.93, "grad_norm": 11.13066577911377, "learning_rate": 2.9092660088039237e-08, "loss": 0.6192, "step": 18197 }, { "epoch": 2.93, "grad_norm": 13.050521850585938, "learning_rate": 2.8960334390706335e-08, "loss": 0.5286, "step": 18198 }, { "epoch": 2.93, "grad_norm": 13.079680442810059, "learning_rate": 2.882830987988916e-08, "loss": 0.7004, "step": 18199 }, { "epoch": 2.93, "grad_norm": 15.520329475402832, "learning_rate": 2.8696586559573413e-08, "loss": 0.6604, "step": 18200 }, { "epoch": 2.93, "grad_norm": 13.471854209899902, "learning_rate": 2.8565164433739245e-08, "loss": 0.6846, "step": 18201 }, { "epoch": 2.93, "grad_norm": 14.775557518005371, "learning_rate": 2.843404350635681e-08, "loss": 0.7811, "step": 18202 }, { "epoch": 2.93, "grad_norm": 14.602192878723145, "learning_rate": 2.830322378138628e-08, "loss": 0.841, "step": 18203 }, { "epoch": 2.93, "grad_norm": 12.138294219970703, "learning_rate": 2.8172705262780043e-08, "loss": 0.6432, "step": 18204 }, { "epoch": 2.93, "grad_norm": 12.730666160583496, "learning_rate": 2.8042487954478283e-08, "loss": 0.6499, "step": 18205 }, { "epoch": 2.93, "grad_norm": 12.537435531616211, "learning_rate": 2.7912571860416738e-08, "loss": 0.6607, "step": 18206 }, { "epoch": 2.93, "grad_norm": 14.160653114318848, "learning_rate": 2.778295698451894e-08, "loss": 0.8121, "step": 18207 }, { "epoch": 2.93, "grad_norm": 17.45957374572754, "learning_rate": 2.7653643330699532e-08, "loss": 0.806, "step": 18208 }, { "epoch": 2.93, "grad_norm": 12.816316604614258, "learning_rate": 2.752463090286428e-08, "loss": 0.7953, "step": 18209 }, { "epoch": 2.93, "grad_norm": 13.023723602294922, "learning_rate": 2.7395919704911178e-08, "loss": 0.8493, "step": 18210 }, { "epoch": 2.93, "grad_norm": 11.682615280151367, "learning_rate": 2.7267509740728227e-08, "loss": 0.541, "step": 18211 }, { "epoch": 2.93, "grad_norm": 14.957657814025879, "learning_rate": 2.7139401014193434e-08, "loss": 0.5787, "step": 18212 }, { "epoch": 2.93, "grad_norm": 11.01461410522461, "learning_rate": 2.701159352917593e-08, "loss": 0.5785, "step": 18213 }, { "epoch": 2.93, "grad_norm": 13.603705406188965, "learning_rate": 2.6884087289537063e-08, "loss": 0.9098, "step": 18214 }, { "epoch": 2.93, "grad_norm": 13.251869201660156, "learning_rate": 2.6756882299129316e-08, "loss": 0.7714, "step": 18215 }, { "epoch": 2.93, "grad_norm": 14.455004692077637, "learning_rate": 2.6629978561792945e-08, "loss": 0.8235, "step": 18216 }, { "epoch": 2.93, "grad_norm": 14.005915641784668, "learning_rate": 2.6503376081363775e-08, "loss": 0.5819, "step": 18217 }, { "epoch": 2.93, "grad_norm": 12.44593620300293, "learning_rate": 2.6377074861663187e-08, "loss": 0.8401, "step": 18218 }, { "epoch": 2.93, "grad_norm": 12.156037330627441, "learning_rate": 2.6251074906508135e-08, "loss": 0.5343, "step": 18219 }, { "epoch": 2.93, "grad_norm": 13.647317886352539, "learning_rate": 2.6125376219703348e-08, "loss": 0.6115, "step": 18220 }, { "epoch": 2.93, "grad_norm": 16.671239852905273, "learning_rate": 2.5999978805046898e-08, "loss": 0.8358, "step": 18221 }, { "epoch": 2.93, "grad_norm": 15.763985633850098, "learning_rate": 2.587488266632576e-08, "loss": 0.6007, "step": 18222 }, { "epoch": 2.93, "grad_norm": 12.641700744628906, "learning_rate": 2.5750087807319136e-08, "loss": 0.7399, "step": 18223 }, { "epoch": 2.93, "grad_norm": 16.304649353027344, "learning_rate": 2.5625594231796224e-08, "loss": 0.8074, "step": 18224 }, { "epoch": 2.93, "grad_norm": 11.581478118896484, "learning_rate": 2.5501401943518467e-08, "loss": 0.7885, "step": 18225 }, { "epoch": 2.93, "grad_norm": 13.96335506439209, "learning_rate": 2.5377510946235084e-08, "loss": 0.7319, "step": 18226 }, { "epoch": 2.93, "grad_norm": 19.323076248168945, "learning_rate": 2.5253921243689748e-08, "loss": 0.7606, "step": 18227 }, { "epoch": 2.93, "grad_norm": 12.92330265045166, "learning_rate": 2.5130632839616142e-08, "loss": 0.6087, "step": 18228 }, { "epoch": 2.93, "grad_norm": 10.896894454956055, "learning_rate": 2.500764573773795e-08, "loss": 0.6036, "step": 18229 }, { "epoch": 2.93, "grad_norm": 16.46769142150879, "learning_rate": 2.4884959941768872e-08, "loss": 0.8205, "step": 18230 }, { "epoch": 2.93, "grad_norm": 14.257057189941406, "learning_rate": 2.476257545541705e-08, "loss": 0.7059, "step": 18231 }, { "epoch": 2.93, "grad_norm": 12.938505172729492, "learning_rate": 2.4640492282376193e-08, "loss": 0.6732, "step": 18232 }, { "epoch": 2.94, "grad_norm": 13.925249099731445, "learning_rate": 2.451871042633669e-08, "loss": 0.8277, "step": 18233 }, { "epoch": 2.94, "grad_norm": 11.021002769470215, "learning_rate": 2.4397229890976702e-08, "loss": 0.8165, "step": 18234 }, { "epoch": 2.94, "grad_norm": 12.153266906738281, "learning_rate": 2.4276050679964415e-08, "loss": 0.8135, "step": 18235 }, { "epoch": 2.94, "grad_norm": 12.492788314819336, "learning_rate": 2.4155172796961335e-08, "loss": 0.6954, "step": 18236 }, { "epoch": 2.94, "grad_norm": 12.230813026428223, "learning_rate": 2.403459624561788e-08, "loss": 0.7026, "step": 18237 }, { "epoch": 2.94, "grad_norm": 15.743910789489746, "learning_rate": 2.391432102957669e-08, "loss": 0.7574, "step": 18238 }, { "epoch": 2.94, "grad_norm": 13.930253982543945, "learning_rate": 2.379434715247153e-08, "loss": 0.7158, "step": 18239 }, { "epoch": 2.94, "grad_norm": 13.79272174835205, "learning_rate": 2.3674674617925054e-08, "loss": 0.8047, "step": 18240 }, { "epoch": 2.94, "grad_norm": 12.574254989624023, "learning_rate": 2.3555303429552145e-08, "loss": 0.6976, "step": 18241 }, { "epoch": 2.94, "grad_norm": 12.22000789642334, "learning_rate": 2.3436233590958814e-08, "loss": 0.6682, "step": 18242 }, { "epoch": 2.94, "grad_norm": 14.436078071594238, "learning_rate": 2.3317465105742176e-08, "loss": 0.8239, "step": 18243 }, { "epoch": 2.94, "grad_norm": 14.012027740478516, "learning_rate": 2.319899797748937e-08, "loss": 0.918, "step": 18244 }, { "epoch": 2.94, "grad_norm": 13.489728927612305, "learning_rate": 2.3080832209779746e-08, "loss": 0.7246, "step": 18245 }, { "epoch": 2.94, "grad_norm": 12.442814826965332, "learning_rate": 2.296296780618157e-08, "loss": 0.7138, "step": 18246 }, { "epoch": 2.94, "grad_norm": 12.379711151123047, "learning_rate": 2.284540477025532e-08, "loss": 0.5788, "step": 18247 }, { "epoch": 2.94, "grad_norm": 14.494603157043457, "learning_rate": 2.272814310555149e-08, "loss": 0.5023, "step": 18248 }, { "epoch": 2.94, "grad_norm": 13.541939735412598, "learning_rate": 2.2611182815612808e-08, "loss": 0.6423, "step": 18249 }, { "epoch": 2.94, "grad_norm": 14.268632888793945, "learning_rate": 2.249452390397311e-08, "loss": 0.6296, "step": 18250 }, { "epoch": 2.94, "grad_norm": 13.227568626403809, "learning_rate": 2.237816637415402e-08, "loss": 0.7179, "step": 18251 }, { "epoch": 2.94, "grad_norm": 14.82807731628418, "learning_rate": 2.2262110229671617e-08, "loss": 0.6364, "step": 18252 }, { "epoch": 2.94, "grad_norm": 13.194549560546875, "learning_rate": 2.214635547403199e-08, "loss": 0.6287, "step": 18253 }, { "epoch": 2.94, "grad_norm": 13.113615989685059, "learning_rate": 2.203090211073011e-08, "loss": 0.7429, "step": 18254 }, { "epoch": 2.94, "grad_norm": 13.235201835632324, "learning_rate": 2.1915750143255422e-08, "loss": 0.5541, "step": 18255 }, { "epoch": 2.94, "grad_norm": 15.046636581420898, "learning_rate": 2.1800899575084022e-08, "loss": 0.7089, "step": 18256 }, { "epoch": 2.94, "grad_norm": 14.12403392791748, "learning_rate": 2.1686350409686473e-08, "loss": 0.7534, "step": 18257 }, { "epoch": 2.94, "grad_norm": 12.70588207244873, "learning_rate": 2.157210265052334e-08, "loss": 0.7748, "step": 18258 }, { "epoch": 2.94, "grad_norm": 11.944669723510742, "learning_rate": 2.1458156301042977e-08, "loss": 0.6728, "step": 18259 }, { "epoch": 2.94, "grad_norm": 12.5424165725708, "learning_rate": 2.1344511364690403e-08, "loss": 0.8781, "step": 18260 }, { "epoch": 2.94, "grad_norm": 13.25770378112793, "learning_rate": 2.1231167844897316e-08, "loss": 0.5733, "step": 18261 }, { "epoch": 2.94, "grad_norm": 15.998299598693848, "learning_rate": 2.1118125745085426e-08, "loss": 0.7686, "step": 18262 }, { "epoch": 2.94, "grad_norm": 11.222919464111328, "learning_rate": 2.1005385068673112e-08, "loss": 0.6435, "step": 18263 }, { "epoch": 2.94, "grad_norm": 12.44929027557373, "learning_rate": 2.0892945819062092e-08, "loss": 0.6194, "step": 18264 }, { "epoch": 2.94, "grad_norm": 17.611312866210938, "learning_rate": 2.0780807999651874e-08, "loss": 0.8538, "step": 18265 }, { "epoch": 2.94, "grad_norm": 12.385518074035645, "learning_rate": 2.0668971613827527e-08, "loss": 0.5025, "step": 18266 }, { "epoch": 2.94, "grad_norm": 11.817940711975098, "learning_rate": 2.0557436664967457e-08, "loss": 0.8193, "step": 18267 }, { "epoch": 2.94, "grad_norm": 12.54820442199707, "learning_rate": 2.044620315644119e-08, "loss": 0.7259, "step": 18268 }, { "epoch": 2.94, "grad_norm": 15.38715648651123, "learning_rate": 2.0335271091608268e-08, "loss": 0.6187, "step": 18269 }, { "epoch": 2.94, "grad_norm": 13.613404273986816, "learning_rate": 2.0224640473820445e-08, "loss": 0.6246, "step": 18270 }, { "epoch": 2.94, "grad_norm": 13.956619262695312, "learning_rate": 2.0114311306418388e-08, "loss": 0.8349, "step": 18271 }, { "epoch": 2.94, "grad_norm": 12.392587661743164, "learning_rate": 2.0004283592736096e-08, "loss": 0.6181, "step": 18272 }, { "epoch": 2.94, "grad_norm": 12.607587814331055, "learning_rate": 1.9894557336095356e-08, "loss": 0.7776, "step": 18273 }, { "epoch": 2.94, "grad_norm": 12.640157699584961, "learning_rate": 1.9785132539811292e-08, "loss": 0.8686, "step": 18274 }, { "epoch": 2.94, "grad_norm": 10.74914264678955, "learning_rate": 1.967600920718904e-08, "loss": 0.5943, "step": 18275 }, { "epoch": 2.94, "grad_norm": 14.274636268615723, "learning_rate": 1.9567187341525963e-08, "loss": 0.7717, "step": 18276 }, { "epoch": 2.94, "grad_norm": 11.150214195251465, "learning_rate": 1.9458666946107207e-08, "loss": 0.7227, "step": 18277 }, { "epoch": 2.94, "grad_norm": 11.719749450683594, "learning_rate": 1.9350448024212376e-08, "loss": 0.677, "step": 18278 }, { "epoch": 2.94, "grad_norm": 13.9136962890625, "learning_rate": 1.9242530579108854e-08, "loss": 0.5382, "step": 18279 }, { "epoch": 2.94, "grad_norm": 11.13231372833252, "learning_rate": 1.9134914614058475e-08, "loss": 0.6779, "step": 18280 }, { "epoch": 2.94, "grad_norm": 14.818075180053711, "learning_rate": 1.9027600132309754e-08, "loss": 0.8485, "step": 18281 }, { "epoch": 2.94, "grad_norm": 11.8096342086792, "learning_rate": 1.892058713710565e-08, "loss": 0.6375, "step": 18282 }, { "epoch": 2.94, "grad_norm": 11.984164237976074, "learning_rate": 1.8813875631678025e-08, "loss": 0.5589, "step": 18283 }, { "epoch": 2.94, "grad_norm": 13.478617668151855, "learning_rate": 1.8707465619249852e-08, "loss": 0.6903, "step": 18284 }, { "epoch": 2.94, "grad_norm": 13.803831100463867, "learning_rate": 1.8601357103037455e-08, "loss": 0.748, "step": 18285 }, { "epoch": 2.94, "grad_norm": 13.501051902770996, "learning_rate": 1.849555008624271e-08, "loss": 0.7549, "step": 18286 }, { "epoch": 2.94, "grad_norm": 11.492620468139648, "learning_rate": 1.839004457206417e-08, "loss": 0.7596, "step": 18287 }, { "epoch": 2.94, "grad_norm": 12.327062606811523, "learning_rate": 1.828484056368707e-08, "loss": 0.7498, "step": 18288 }, { "epoch": 2.94, "grad_norm": 13.21873664855957, "learning_rate": 1.817993806429108e-08, "loss": 0.6229, "step": 18289 }, { "epoch": 2.94, "grad_norm": 15.514348030090332, "learning_rate": 1.8075337077042566e-08, "loss": 0.7754, "step": 18290 }, { "epoch": 2.94, "grad_norm": 12.085771560668945, "learning_rate": 1.7971037605103438e-08, "loss": 0.8277, "step": 18291 }, { "epoch": 2.94, "grad_norm": 14.633098602294922, "learning_rate": 1.7867039651622287e-08, "loss": 0.8165, "step": 18292 }, { "epoch": 2.94, "grad_norm": 13.229398727416992, "learning_rate": 1.7763343219742156e-08, "loss": 0.6133, "step": 18293 }, { "epoch": 2.94, "grad_norm": 11.65726375579834, "learning_rate": 1.7659948312593877e-08, "loss": 0.5711, "step": 18294 }, { "epoch": 2.95, "grad_norm": 10.126896858215332, "learning_rate": 1.7556854933301615e-08, "loss": 0.6247, "step": 18295 }, { "epoch": 2.95, "grad_norm": 13.736069679260254, "learning_rate": 1.7454063084979545e-08, "loss": 0.7882, "step": 18296 }, { "epoch": 2.95, "grad_norm": 12.787213325500488, "learning_rate": 1.7351572770730742e-08, "loss": 0.6295, "step": 18297 }, { "epoch": 2.95, "grad_norm": 11.732090950012207, "learning_rate": 1.7249383993653835e-08, "loss": 0.8011, "step": 18298 }, { "epoch": 2.95, "grad_norm": 13.710756301879883, "learning_rate": 1.7147496756833028e-08, "loss": 0.6474, "step": 18299 }, { "epoch": 2.95, "grad_norm": 13.697274208068848, "learning_rate": 1.7045911063346964e-08, "loss": 0.7848, "step": 18300 }, { "epoch": 2.95, "grad_norm": 14.069718360900879, "learning_rate": 1.6944626916265417e-08, "loss": 0.7703, "step": 18301 }, { "epoch": 2.95, "grad_norm": 14.399087905883789, "learning_rate": 1.6843644318644826e-08, "loss": 0.7377, "step": 18302 }, { "epoch": 2.95, "grad_norm": 15.588358879089355, "learning_rate": 1.6742963273538303e-08, "loss": 0.5976, "step": 18303 }, { "epoch": 2.95, "grad_norm": 12.020490646362305, "learning_rate": 1.6642583783985643e-08, "loss": 0.6557, "step": 18304 }, { "epoch": 2.95, "grad_norm": 15.000590324401855, "learning_rate": 1.6542505853019974e-08, "loss": 0.6492, "step": 18305 }, { "epoch": 2.95, "grad_norm": 14.181147575378418, "learning_rate": 1.6442729483663324e-08, "loss": 0.7865, "step": 18306 }, { "epoch": 2.95, "grad_norm": 17.552688598632812, "learning_rate": 1.6343254678928834e-08, "loss": 0.7866, "step": 18307 }, { "epoch": 2.95, "grad_norm": 12.391105651855469, "learning_rate": 1.624408144182299e-08, "loss": 0.7097, "step": 18308 }, { "epoch": 2.95, "grad_norm": 11.875736236572266, "learning_rate": 1.6145209775340066e-08, "loss": 0.7841, "step": 18309 }, { "epoch": 2.95, "grad_norm": 12.798725128173828, "learning_rate": 1.604663968246767e-08, "loss": 0.7308, "step": 18310 }, { "epoch": 2.95, "grad_norm": 12.088085174560547, "learning_rate": 1.5948371166183418e-08, "loss": 0.6411, "step": 18311 }, { "epoch": 2.95, "grad_norm": 13.773581504821777, "learning_rate": 1.585040422945383e-08, "loss": 0.7611, "step": 18312 }, { "epoch": 2.95, "grad_norm": 10.733262062072754, "learning_rate": 1.575273887523876e-08, "loss": 0.5404, "step": 18313 }, { "epoch": 2.95, "grad_norm": 14.892440795898438, "learning_rate": 1.565537510649029e-08, "loss": 0.7944, "step": 18314 }, { "epoch": 2.95, "grad_norm": 14.07401180267334, "learning_rate": 1.555831292614607e-08, "loss": 0.5447, "step": 18315 }, { "epoch": 2.95, "grad_norm": 11.82101058959961, "learning_rate": 1.5461552337140417e-08, "loss": 0.688, "step": 18316 }, { "epoch": 2.95, "grad_norm": 11.658995628356934, "learning_rate": 1.5365093342396554e-08, "loss": 0.6344, "step": 18317 }, { "epoch": 2.95, "grad_norm": 15.186817169189453, "learning_rate": 1.526893594482548e-08, "loss": 0.6211, "step": 18318 }, { "epoch": 2.95, "grad_norm": 12.390427589416504, "learning_rate": 1.5173080147333763e-08, "loss": 0.7517, "step": 18319 }, { "epoch": 2.95, "grad_norm": 13.049918174743652, "learning_rate": 1.5077525952816862e-08, "loss": 0.5895, "step": 18320 }, { "epoch": 2.95, "grad_norm": 14.726509094238281, "learning_rate": 1.498227336416025e-08, "loss": 0.8607, "step": 18321 }, { "epoch": 2.95, "grad_norm": 11.037551879882812, "learning_rate": 1.488732238424051e-08, "loss": 0.6892, "step": 18322 }, { "epoch": 2.95, "grad_norm": 10.27558422088623, "learning_rate": 1.4792673015927572e-08, "loss": 0.6142, "step": 18323 }, { "epoch": 2.95, "grad_norm": 14.369157791137695, "learning_rate": 1.4698325262079149e-08, "loss": 0.7756, "step": 18324 }, { "epoch": 2.95, "grad_norm": 16.167612075805664, "learning_rate": 1.4604279125546295e-08, "loss": 0.7779, "step": 18325 }, { "epoch": 2.95, "grad_norm": 11.726631164550781, "learning_rate": 1.4510534609168959e-08, "loss": 0.7781, "step": 18326 }, { "epoch": 2.95, "grad_norm": 12.109603881835938, "learning_rate": 1.441709171577821e-08, "loss": 0.5635, "step": 18327 }, { "epoch": 2.95, "grad_norm": 13.380163192749023, "learning_rate": 1.4323950448197343e-08, "loss": 0.5733, "step": 18328 }, { "epoch": 2.95, "grad_norm": 10.939910888671875, "learning_rate": 1.4231110809239668e-08, "loss": 0.7655, "step": 18329 }, { "epoch": 2.95, "grad_norm": 12.96142578125, "learning_rate": 1.4138572801709605e-08, "loss": 0.674, "step": 18330 }, { "epoch": 2.95, "grad_norm": 10.73221206665039, "learning_rate": 1.4046336428402696e-08, "loss": 0.611, "step": 18331 }, { "epoch": 2.95, "grad_norm": 16.36337661743164, "learning_rate": 1.3954401692104491e-08, "loss": 0.809, "step": 18332 }, { "epoch": 2.95, "grad_norm": 13.857033729553223, "learning_rate": 1.3862768595591657e-08, "loss": 0.7026, "step": 18333 }, { "epoch": 2.95, "grad_norm": 15.042264938354492, "learning_rate": 1.3771437141631982e-08, "loss": 0.6153, "step": 18334 }, { "epoch": 2.95, "grad_norm": 13.62148666381836, "learning_rate": 1.368040733298659e-08, "loss": 0.6181, "step": 18335 }, { "epoch": 2.95, "grad_norm": 12.979022979736328, "learning_rate": 1.3589679172402171e-08, "loss": 0.7425, "step": 18336 }, { "epoch": 2.95, "grad_norm": 14.924558639526367, "learning_rate": 1.3499252662619865e-08, "loss": 0.8019, "step": 18337 }, { "epoch": 2.95, "grad_norm": 11.695053100585938, "learning_rate": 1.3409127806373045e-08, "loss": 0.8561, "step": 18338 }, { "epoch": 2.95, "grad_norm": 9.64231014251709, "learning_rate": 1.3319304606381756e-08, "loss": 0.6352, "step": 18339 }, { "epoch": 2.95, "grad_norm": 14.142059326171875, "learning_rate": 1.322978306536049e-08, "loss": 0.6791, "step": 18340 }, { "epoch": 2.95, "grad_norm": 12.538155555725098, "learning_rate": 1.3140563186013755e-08, "loss": 0.4138, "step": 18341 }, { "epoch": 2.95, "grad_norm": 12.334827423095703, "learning_rate": 1.3051644971034948e-08, "loss": 0.8294, "step": 18342 }, { "epoch": 2.95, "grad_norm": 13.991744041442871, "learning_rate": 1.2963028423109703e-08, "loss": 0.8287, "step": 18343 }, { "epoch": 2.95, "grad_norm": 16.682748794555664, "learning_rate": 1.2874713544918093e-08, "loss": 0.7052, "step": 18344 }, { "epoch": 2.95, "grad_norm": 12.926849365234375, "learning_rate": 1.2786700339123548e-08, "loss": 0.7882, "step": 18345 }, { "epoch": 2.95, "grad_norm": 12.946344375610352, "learning_rate": 1.2698988808387269e-08, "loss": 0.7393, "step": 18346 }, { "epoch": 2.95, "grad_norm": 13.257113456726074, "learning_rate": 1.261157895535825e-08, "loss": 0.6245, "step": 18347 }, { "epoch": 2.95, "grad_norm": 12.714485168457031, "learning_rate": 1.2524470782676601e-08, "loss": 0.6318, "step": 18348 }, { "epoch": 2.95, "grad_norm": 12.777900695800781, "learning_rate": 1.243766429297355e-08, "loss": 0.7973, "step": 18349 }, { "epoch": 2.95, "grad_norm": 13.298683166503906, "learning_rate": 1.2351159488870335e-08, "loss": 0.7809, "step": 18350 }, { "epoch": 2.95, "grad_norm": 14.463227272033691, "learning_rate": 1.2264956372981529e-08, "loss": 0.6445, "step": 18351 }, { "epoch": 2.95, "grad_norm": 11.618583679199219, "learning_rate": 1.2179054947910607e-08, "loss": 0.7107, "step": 18352 }, { "epoch": 2.95, "grad_norm": 13.886042594909668, "learning_rate": 1.2093455216251049e-08, "loss": 0.8367, "step": 18353 }, { "epoch": 2.95, "grad_norm": 14.074278831481934, "learning_rate": 1.2008157180589675e-08, "loss": 0.611, "step": 18354 }, { "epoch": 2.95, "grad_norm": 12.156647682189941, "learning_rate": 1.1923160843502201e-08, "loss": 0.7467, "step": 18355 }, { "epoch": 2.95, "grad_norm": 16.363786697387695, "learning_rate": 1.1838466207556576e-08, "loss": 0.5014, "step": 18356 }, { "epoch": 2.96, "grad_norm": 13.700263977050781, "learning_rate": 1.175407327531075e-08, "loss": 0.6665, "step": 18357 }, { "epoch": 2.96, "grad_norm": 12.151684761047363, "learning_rate": 1.1669982049314909e-08, "loss": 0.8499, "step": 18358 }, { "epoch": 2.96, "grad_norm": 12.387845039367676, "learning_rate": 1.1586192532108131e-08, "loss": 0.5778, "step": 18359 }, { "epoch": 2.96, "grad_norm": 11.998550415039062, "learning_rate": 1.1502704726221725e-08, "loss": 0.6548, "step": 18360 }, { "epoch": 2.96, "grad_norm": 12.434925079345703, "learning_rate": 1.1419518634177006e-08, "loss": 0.8049, "step": 18361 }, { "epoch": 2.96, "grad_norm": 13.328468322753906, "learning_rate": 1.1336634258487523e-08, "loss": 0.8569, "step": 18362 }, { "epoch": 2.96, "grad_norm": 11.574623107910156, "learning_rate": 1.1254051601656824e-08, "loss": 0.726, "step": 18363 }, { "epoch": 2.96, "grad_norm": 16.194726943969727, "learning_rate": 1.1171770666178471e-08, "loss": 0.6812, "step": 18364 }, { "epoch": 2.96, "grad_norm": 12.103528022766113, "learning_rate": 1.1089791454538256e-08, "loss": 0.7055, "step": 18365 }, { "epoch": 2.96, "grad_norm": 11.98067855834961, "learning_rate": 1.1008113969213085e-08, "loss": 0.7449, "step": 18366 }, { "epoch": 2.96, "grad_norm": 16.26300621032715, "learning_rate": 1.0926738212669874e-08, "loss": 0.8704, "step": 18367 }, { "epoch": 2.96, "grad_norm": 15.411333084106445, "learning_rate": 1.0845664187366656e-08, "loss": 0.7503, "step": 18368 }, { "epoch": 2.96, "grad_norm": 14.834178924560547, "learning_rate": 1.0764891895751472e-08, "loss": 0.7629, "step": 18369 }, { "epoch": 2.96, "grad_norm": 12.159164428710938, "learning_rate": 1.0684421340265705e-08, "loss": 0.8011, "step": 18370 }, { "epoch": 2.96, "grad_norm": 12.855365753173828, "learning_rate": 1.060425252333852e-08, "loss": 0.8754, "step": 18371 }, { "epoch": 2.96, "grad_norm": 11.375733375549316, "learning_rate": 1.0524385447392427e-08, "loss": 0.5762, "step": 18372 }, { "epoch": 2.96, "grad_norm": 13.357880592346191, "learning_rate": 1.0444820114839937e-08, "loss": 0.6819, "step": 18373 }, { "epoch": 2.96, "grad_norm": 12.714884757995605, "learning_rate": 1.0365556528083575e-08, "loss": 0.8012, "step": 18374 }, { "epoch": 2.96, "grad_norm": 15.291219711303711, "learning_rate": 1.0286594689519203e-08, "loss": 0.7807, "step": 18375 }, { "epoch": 2.96, "grad_norm": 11.762540817260742, "learning_rate": 1.0207934601529356e-08, "loss": 0.6303, "step": 18376 }, { "epoch": 2.96, "grad_norm": 11.972405433654785, "learning_rate": 1.0129576266493247e-08, "loss": 0.7596, "step": 18377 }, { "epoch": 2.96, "grad_norm": 12.33154010772705, "learning_rate": 1.0051519686775646e-08, "loss": 0.4874, "step": 18378 }, { "epoch": 2.96, "grad_norm": 14.343194007873535, "learning_rate": 9.97376486473467e-09, "loss": 0.7078, "step": 18379 }, { "epoch": 2.96, "grad_norm": 15.019058227539062, "learning_rate": 9.89631180271844e-09, "loss": 0.7291, "step": 18380 }, { "epoch": 2.96, "grad_norm": 12.4888916015625, "learning_rate": 9.819160503068415e-09, "loss": 0.5774, "step": 18381 }, { "epoch": 2.96, "grad_norm": 15.961119651794434, "learning_rate": 9.742310968113844e-09, "loss": 0.7171, "step": 18382 }, { "epoch": 2.96, "grad_norm": 12.108296394348145, "learning_rate": 9.665763200175094e-09, "loss": 0.6308, "step": 18383 }, { "epoch": 2.96, "grad_norm": 12.745380401611328, "learning_rate": 9.589517201565868e-09, "loss": 0.9084, "step": 18384 }, { "epoch": 2.96, "grad_norm": 16.213449478149414, "learning_rate": 9.51357297458877e-09, "loss": 0.5491, "step": 18385 }, { "epoch": 2.96, "grad_norm": 15.855937957763672, "learning_rate": 9.43793052153752e-09, "loss": 0.7154, "step": 18386 }, { "epoch": 2.96, "grad_norm": 702.8896484375, "learning_rate": 9.362589844698067e-09, "loss": 0.9503, "step": 18387 }, { "epoch": 2.96, "grad_norm": 13.612625122070312, "learning_rate": 9.287550946344148e-09, "loss": 0.8017, "step": 18388 }, { "epoch": 2.96, "grad_norm": 14.201367378234863, "learning_rate": 9.21281382874506e-09, "loss": 0.6184, "step": 18389 }, { "epoch": 2.96, "grad_norm": 13.726297378540039, "learning_rate": 9.138378494155665e-09, "loss": 0.7399, "step": 18390 }, { "epoch": 2.96, "grad_norm": 11.22170352935791, "learning_rate": 9.064244944825273e-09, "loss": 0.6655, "step": 18391 }, { "epoch": 2.96, "grad_norm": 12.420404434204102, "learning_rate": 8.990413182994317e-09, "loss": 0.7305, "step": 18392 }, { "epoch": 2.96, "grad_norm": 10.260968208312988, "learning_rate": 8.916883210892125e-09, "loss": 0.6643, "step": 18393 }, { "epoch": 2.96, "grad_norm": 14.591136932373047, "learning_rate": 8.84365503073914e-09, "loss": 0.9431, "step": 18394 }, { "epoch": 2.96, "grad_norm": 14.526165962219238, "learning_rate": 8.770728644748038e-09, "loss": 0.8413, "step": 18395 }, { "epoch": 2.96, "grad_norm": 13.114657402038574, "learning_rate": 8.698104055121504e-09, "loss": 0.7302, "step": 18396 }, { "epoch": 2.96, "grad_norm": 14.15397834777832, "learning_rate": 8.625781264053335e-09, "loss": 0.8401, "step": 18397 }, { "epoch": 2.96, "grad_norm": 15.28298568725586, "learning_rate": 8.55376027372734e-09, "loss": 0.7566, "step": 18398 }, { "epoch": 2.96, "grad_norm": 13.068824768066406, "learning_rate": 8.482041086320669e-09, "loss": 0.7005, "step": 18399 }, { "epoch": 2.96, "grad_norm": 11.648422241210938, "learning_rate": 8.410623703998254e-09, "loss": 0.6257, "step": 18400 }, { "epoch": 2.96, "grad_norm": 18.390520095825195, "learning_rate": 8.339508128917262e-09, "loss": 0.7722, "step": 18401 }, { "epoch": 2.96, "grad_norm": 13.578031539916992, "learning_rate": 8.268694363227082e-09, "loss": 0.7953, "step": 18402 }, { "epoch": 2.96, "grad_norm": 11.994275093078613, "learning_rate": 8.198182409064892e-09, "loss": 0.6762, "step": 18403 }, { "epoch": 2.96, "grad_norm": 13.209615707397461, "learning_rate": 8.127972268562324e-09, "loss": 0.7701, "step": 18404 }, { "epoch": 2.96, "grad_norm": 13.062788009643555, "learning_rate": 8.0580639438399e-09, "loss": 0.6433, "step": 18405 }, { "epoch": 2.96, "grad_norm": 12.525254249572754, "learning_rate": 7.988457437009267e-09, "loss": 0.5775, "step": 18406 }, { "epoch": 2.96, "grad_norm": 13.98147964477539, "learning_rate": 7.919152750170966e-09, "loss": 0.6394, "step": 18407 }, { "epoch": 2.96, "grad_norm": 13.607685089111328, "learning_rate": 7.8501498854211e-09, "loss": 0.8193, "step": 18408 }, { "epoch": 2.96, "grad_norm": 13.560277938842773, "learning_rate": 7.781448844842443e-09, "loss": 0.8703, "step": 18409 }, { "epoch": 2.96, "grad_norm": 13.268526077270508, "learning_rate": 7.713049630511116e-09, "loss": 0.7984, "step": 18410 }, { "epoch": 2.96, "grad_norm": 14.281205177307129, "learning_rate": 7.644952244492132e-09, "loss": 0.9267, "step": 18411 }, { "epoch": 2.96, "grad_norm": 18.115934371948242, "learning_rate": 7.577156688843846e-09, "loss": 0.833, "step": 18412 }, { "epoch": 2.96, "grad_norm": 11.839315414428711, "learning_rate": 7.509662965612397e-09, "loss": 0.7404, "step": 18413 }, { "epoch": 2.96, "grad_norm": 15.177070617675781, "learning_rate": 7.442471076838376e-09, "loss": 0.7846, "step": 18414 }, { "epoch": 2.96, "grad_norm": 14.355082511901855, "learning_rate": 7.37558102454905e-09, "loss": 0.7544, "step": 18415 }, { "epoch": 2.96, "grad_norm": 15.254695892333984, "learning_rate": 7.308992810767246e-09, "loss": 0.6269, "step": 18416 }, { "epoch": 2.96, "grad_norm": 19.730466842651367, "learning_rate": 7.242706437503577e-09, "loss": 0.8087, "step": 18417 }, { "epoch": 2.96, "grad_norm": 14.659628868103027, "learning_rate": 7.176721906759776e-09, "loss": 0.7016, "step": 18418 }, { "epoch": 2.97, "grad_norm": 11.451011657714844, "learning_rate": 7.111039220528693e-09, "loss": 0.9671, "step": 18419 }, { "epoch": 2.97, "grad_norm": 12.917047500610352, "learning_rate": 7.045658380795406e-09, "loss": 0.8952, "step": 18420 }, { "epoch": 2.97, "grad_norm": 18.73456382751465, "learning_rate": 6.980579389535003e-09, "loss": 0.6771, "step": 18421 }, { "epoch": 2.97, "grad_norm": 14.980415344238281, "learning_rate": 6.915802248712577e-09, "loss": 0.622, "step": 18422 }, { "epoch": 2.97, "grad_norm": 12.663207054138184, "learning_rate": 6.8513269602832335e-09, "loss": 0.7346, "step": 18423 }, { "epoch": 2.97, "grad_norm": 13.280390739440918, "learning_rate": 6.787153526197632e-09, "loss": 0.495, "step": 18424 }, { "epoch": 2.97, "grad_norm": 17.20097541809082, "learning_rate": 6.723281948392002e-09, "loss": 0.5711, "step": 18425 }, { "epoch": 2.97, "grad_norm": 15.67855453491211, "learning_rate": 6.6597122287970215e-09, "loss": 0.718, "step": 18426 }, { "epoch": 2.97, "grad_norm": 16.40863800048828, "learning_rate": 6.5964443693311565e-09, "loss": 0.8023, "step": 18427 }, { "epoch": 2.97, "grad_norm": 12.537879943847656, "learning_rate": 6.53347837190732e-09, "loss": 0.7479, "step": 18428 }, { "epoch": 2.97, "grad_norm": 11.368473052978516, "learning_rate": 6.4708142384273255e-09, "loss": 0.6884, "step": 18429 }, { "epoch": 2.97, "grad_norm": 11.500768661499023, "learning_rate": 6.408451970781882e-09, "loss": 0.6834, "step": 18430 }, { "epoch": 2.97, "grad_norm": 12.842957496643066, "learning_rate": 6.3463915708572575e-09, "loss": 0.6341, "step": 18431 }, { "epoch": 2.97, "grad_norm": 14.406207084655762, "learning_rate": 6.2846330405264e-09, "loss": 0.5224, "step": 18432 }, { "epoch": 2.97, "grad_norm": 13.058472633361816, "learning_rate": 6.223176381656704e-09, "loss": 0.756, "step": 18433 }, { "epoch": 2.97, "grad_norm": 15.398465156555176, "learning_rate": 6.162021596102241e-09, "loss": 0.7689, "step": 18434 }, { "epoch": 2.97, "grad_norm": 15.306796073913574, "learning_rate": 6.101168685710424e-09, "loss": 0.836, "step": 18435 }, { "epoch": 2.97, "grad_norm": 11.934399604797363, "learning_rate": 6.04061765232089e-09, "loss": 0.6503, "step": 18436 }, { "epoch": 2.97, "grad_norm": 13.350362777709961, "learning_rate": 5.980368497763289e-09, "loss": 0.6997, "step": 18437 }, { "epoch": 2.97, "grad_norm": 18.402524948120117, "learning_rate": 5.920421223855055e-09, "loss": 0.612, "step": 18438 }, { "epoch": 2.97, "grad_norm": 11.455743789672852, "learning_rate": 5.860775832408072e-09, "loss": 0.637, "step": 18439 }, { "epoch": 2.97, "grad_norm": 13.749480247497559, "learning_rate": 5.801432325224232e-09, "loss": 0.7504, "step": 18440 }, { "epoch": 2.97, "grad_norm": 17.737125396728516, "learning_rate": 5.742390704095436e-09, "loss": 0.5877, "step": 18441 }, { "epoch": 2.97, "grad_norm": 15.871542930603027, "learning_rate": 5.6836509708058095e-09, "loss": 0.6057, "step": 18442 }, { "epoch": 2.97, "grad_norm": 13.527583122253418, "learning_rate": 5.625213127129492e-09, "loss": 0.7095, "step": 18443 }, { "epoch": 2.97, "grad_norm": 13.139853477478027, "learning_rate": 5.567077174831736e-09, "loss": 0.6243, "step": 18444 }, { "epoch": 2.97, "grad_norm": 11.445660591125488, "learning_rate": 5.509243115667806e-09, "loss": 0.6346, "step": 18445 }, { "epoch": 2.97, "grad_norm": 13.961519241333008, "learning_rate": 5.451710951386302e-09, "loss": 0.5062, "step": 18446 }, { "epoch": 2.97, "grad_norm": 13.527542114257812, "learning_rate": 5.394480683722503e-09, "loss": 0.6874, "step": 18447 }, { "epoch": 2.97, "grad_norm": 11.622748374938965, "learning_rate": 5.337552314408356e-09, "loss": 0.7216, "step": 18448 }, { "epoch": 2.97, "grad_norm": 13.168573379516602, "learning_rate": 5.280925845160268e-09, "loss": 0.7201, "step": 18449 }, { "epoch": 2.97, "grad_norm": 15.712119102478027, "learning_rate": 5.22460127769131e-09, "loss": 0.7985, "step": 18450 }, { "epoch": 2.97, "grad_norm": 13.70095443725586, "learning_rate": 5.168578613701236e-09, "loss": 0.7491, "step": 18451 }, { "epoch": 2.97, "grad_norm": 15.520651817321777, "learning_rate": 5.112857854882025e-09, "loss": 1.0453, "step": 18452 }, { "epoch": 2.97, "grad_norm": 13.934590339660645, "learning_rate": 5.057439002918996e-09, "loss": 0.8983, "step": 18453 }, { "epoch": 2.97, "grad_norm": 10.241159439086914, "learning_rate": 5.0023220594830334e-09, "loss": 0.5105, "step": 18454 }, { "epoch": 2.97, "grad_norm": 9.154168128967285, "learning_rate": 4.9475070262416935e-09, "loss": 0.6945, "step": 18455 }, { "epoch": 2.97, "grad_norm": 13.487972259521484, "learning_rate": 4.892993904849208e-09, "loss": 0.6215, "step": 18456 }, { "epoch": 2.97, "grad_norm": 14.377559661865234, "learning_rate": 4.838782696952038e-09, "loss": 0.8419, "step": 18457 }, { "epoch": 2.97, "grad_norm": 15.297707557678223, "learning_rate": 4.784873404188872e-09, "loss": 0.7347, "step": 18458 }, { "epoch": 2.97, "grad_norm": 14.262493133544922, "learning_rate": 4.7312660281884085e-09, "loss": 0.8522, "step": 18459 }, { "epoch": 2.97, "grad_norm": 13.945051193237305, "learning_rate": 4.677960570567131e-09, "loss": 0.7642, "step": 18460 }, { "epoch": 2.97, "grad_norm": 13.600565910339355, "learning_rate": 4.6249570329381934e-09, "loss": 0.8969, "step": 18461 }, { "epoch": 2.97, "grad_norm": 11.634024620056152, "learning_rate": 4.572255416901428e-09, "loss": 0.7348, "step": 18462 }, { "epoch": 2.97, "grad_norm": 15.275955200195312, "learning_rate": 4.519855724048894e-09, "loss": 0.8185, "step": 18463 }, { "epoch": 2.97, "grad_norm": 13.390677452087402, "learning_rate": 4.467757955963769e-09, "loss": 0.684, "step": 18464 }, { "epoch": 2.97, "grad_norm": 12.006450653076172, "learning_rate": 4.41596211421813e-09, "loss": 0.8053, "step": 18465 }, { "epoch": 2.97, "grad_norm": 17.162137985229492, "learning_rate": 4.364468200378502e-09, "loss": 0.7098, "step": 18466 }, { "epoch": 2.97, "grad_norm": 10.353546142578125, "learning_rate": 4.313276215999196e-09, "loss": 0.8234, "step": 18467 }, { "epoch": 2.97, "grad_norm": 13.209822654724121, "learning_rate": 4.262386162626753e-09, "loss": 0.7054, "step": 18468 }, { "epoch": 2.97, "grad_norm": 9.628573417663574, "learning_rate": 4.211798041798831e-09, "loss": 0.5636, "step": 18469 }, { "epoch": 2.97, "grad_norm": 12.07054328918457, "learning_rate": 4.161511855041989e-09, "loss": 0.7038, "step": 18470 }, { "epoch": 2.97, "grad_norm": 12.442705154418945, "learning_rate": 4.11152760387723e-09, "loss": 0.5265, "step": 18471 }, { "epoch": 2.97, "grad_norm": 13.847627639770508, "learning_rate": 4.061845289813349e-09, "loss": 0.5944, "step": 18472 }, { "epoch": 2.97, "grad_norm": 12.166309356689453, "learning_rate": 4.012464914351366e-09, "loss": 0.6629, "step": 18473 }, { "epoch": 2.97, "grad_norm": 11.79554271697998, "learning_rate": 3.963386478982312e-09, "loss": 0.5955, "step": 18474 }, { "epoch": 2.97, "grad_norm": 16.237850189208984, "learning_rate": 3.914609985189444e-09, "loss": 0.7281, "step": 18475 }, { "epoch": 2.97, "grad_norm": 12.781585693359375, "learning_rate": 3.866135434444917e-09, "loss": 0.7933, "step": 18476 }, { "epoch": 2.97, "grad_norm": 15.522683143615723, "learning_rate": 3.817962828214228e-09, "loss": 0.6413, "step": 18477 }, { "epoch": 2.97, "grad_norm": 14.496162414550781, "learning_rate": 3.770092167951767e-09, "loss": 0.6798, "step": 18478 }, { "epoch": 2.97, "grad_norm": 13.867478370666504, "learning_rate": 3.7225234551041543e-09, "loss": 0.6597, "step": 18479 }, { "epoch": 2.97, "grad_norm": 13.438447952270508, "learning_rate": 3.67525669110691e-09, "loss": 0.6603, "step": 18480 }, { "epoch": 2.98, "grad_norm": 13.452140808105469, "learning_rate": 3.62829187739e-09, "loss": 0.7271, "step": 18481 }, { "epoch": 2.98, "grad_norm": 12.082940101623535, "learning_rate": 3.5816290153700695e-09, "loss": 0.6492, "step": 18482 }, { "epoch": 2.98, "grad_norm": 13.252360343933105, "learning_rate": 3.5352681064571015e-09, "loss": 0.7683, "step": 18483 }, { "epoch": 2.98, "grad_norm": 13.800535202026367, "learning_rate": 3.4892091520510873e-09, "loss": 0.8193, "step": 18484 }, { "epoch": 2.98, "grad_norm": 12.640094757080078, "learning_rate": 3.443452153544247e-09, "loss": 0.9084, "step": 18485 }, { "epoch": 2.98, "grad_norm": 12.818406105041504, "learning_rate": 3.397997112318807e-09, "loss": 0.8168, "step": 18486 }, { "epoch": 2.98, "grad_norm": 12.239572525024414, "learning_rate": 3.3528440297481145e-09, "loss": 0.6012, "step": 18487 }, { "epoch": 2.98, "grad_norm": 12.288581848144531, "learning_rate": 3.3079929071933028e-09, "loss": 0.7111, "step": 18488 }, { "epoch": 2.98, "grad_norm": 20.114013671875, "learning_rate": 3.263443746013284e-09, "loss": 0.6091, "step": 18489 }, { "epoch": 2.98, "grad_norm": 15.194079399108887, "learning_rate": 3.2191965475503182e-09, "loss": 0.7755, "step": 18490 }, { "epoch": 2.98, "grad_norm": 12.154823303222656, "learning_rate": 3.1752513131422247e-09, "loss": 0.5244, "step": 18491 }, { "epoch": 2.98, "grad_norm": 21.40390396118164, "learning_rate": 3.1316080441168296e-09, "loss": 0.6926, "step": 18492 }, { "epoch": 2.98, "grad_norm": 15.5429048538208, "learning_rate": 3.0882667417908573e-09, "loss": 0.7013, "step": 18493 }, { "epoch": 2.98, "grad_norm": 15.01576042175293, "learning_rate": 3.0452274074754817e-09, "loss": 0.8633, "step": 18494 }, { "epoch": 2.98, "grad_norm": 14.485649108886719, "learning_rate": 3.0024900424696633e-09, "loss": 0.8479, "step": 18495 }, { "epoch": 2.98, "grad_norm": 16.142818450927734, "learning_rate": 2.960054648064592e-09, "loss": 0.6394, "step": 18496 }, { "epoch": 2.98, "grad_norm": 18.2862548828125, "learning_rate": 2.917921225542575e-09, "loss": 0.684, "step": 18497 }, { "epoch": 2.98, "grad_norm": 11.376521110534668, "learning_rate": 2.8760897761759275e-09, "loss": 0.7985, "step": 18498 }, { "epoch": 2.98, "grad_norm": 15.234249114990234, "learning_rate": 2.834560301226974e-09, "loss": 0.7548, "step": 18499 }, { "epoch": 2.98, "grad_norm": 15.792909622192383, "learning_rate": 2.793332801951376e-09, "loss": 0.6428, "step": 18500 }, { "epoch": 2.98, "grad_norm": 16.27790641784668, "learning_rate": 2.7524072795948042e-09, "loss": 0.8188, "step": 18501 }, { "epoch": 2.98, "grad_norm": 11.043492317199707, "learning_rate": 2.7117837353929365e-09, "loss": 0.7328, "step": 18502 }, { "epoch": 2.98, "grad_norm": 13.619319915771484, "learning_rate": 2.6714621705725696e-09, "loss": 0.6297, "step": 18503 }, { "epoch": 2.98, "grad_norm": 13.631651878356934, "learning_rate": 2.631442586352728e-09, "loss": 0.646, "step": 18504 }, { "epoch": 2.98, "grad_norm": 18.487424850463867, "learning_rate": 2.591724983940225e-09, "loss": 0.6893, "step": 18505 }, { "epoch": 2.98, "grad_norm": 13.745434761047363, "learning_rate": 2.552309364537431e-09, "loss": 0.8059, "step": 18506 }, { "epoch": 2.98, "grad_norm": 13.324219703674316, "learning_rate": 2.513195729332285e-09, "loss": 0.7217, "step": 18507 }, { "epoch": 2.98, "grad_norm": 13.970855712890625, "learning_rate": 2.4743840795082852e-09, "loss": 0.9192, "step": 18508 }, { "epoch": 2.98, "grad_norm": 13.152749061584473, "learning_rate": 2.4358744162367166e-09, "loss": 0.7887, "step": 18509 }, { "epoch": 2.98, "grad_norm": 15.245701789855957, "learning_rate": 2.3976667406810927e-09, "loss": 0.6719, "step": 18510 }, { "epoch": 2.98, "grad_norm": 14.687318801879883, "learning_rate": 2.359761053996046e-09, "loss": 0.96, "step": 18511 }, { "epoch": 2.98, "grad_norm": 13.642560958862305, "learning_rate": 2.3221573573262156e-09, "loss": 0.5896, "step": 18512 }, { "epoch": 2.98, "grad_norm": 12.382451057434082, "learning_rate": 2.2848556518062504e-09, "loss": 0.8114, "step": 18513 }, { "epoch": 2.98, "grad_norm": 13.826456069946289, "learning_rate": 2.2478559385641363e-09, "loss": 0.852, "step": 18514 }, { "epoch": 2.98, "grad_norm": 20.77797508239746, "learning_rate": 2.2111582187178684e-09, "loss": 0.7787, "step": 18515 }, { "epoch": 2.98, "grad_norm": 14.03465461730957, "learning_rate": 2.1747624933754484e-09, "loss": 0.7649, "step": 18516 }, { "epoch": 2.98, "grad_norm": 13.528194427490234, "learning_rate": 2.138668763635998e-09, "loss": 0.4939, "step": 18517 }, { "epoch": 2.98, "grad_norm": 14.27251148223877, "learning_rate": 2.102877030589756e-09, "loss": 0.7352, "step": 18518 }, { "epoch": 2.98, "grad_norm": 13.777583122253418, "learning_rate": 2.0673872953180798e-09, "loss": 0.767, "step": 18519 }, { "epoch": 2.98, "grad_norm": 11.738167762756348, "learning_rate": 2.032199558893444e-09, "loss": 0.6837, "step": 18520 }, { "epoch": 2.98, "grad_norm": 12.00861930847168, "learning_rate": 1.9973138223772224e-09, "loss": 0.8048, "step": 18521 }, { "epoch": 2.98, "grad_norm": 12.144767761230469, "learning_rate": 1.9627300868252375e-09, "loss": 0.6436, "step": 18522 }, { "epoch": 2.98, "grad_norm": 10.79563045501709, "learning_rate": 1.928448353279988e-09, "loss": 0.5526, "step": 18523 }, { "epoch": 2.98, "grad_norm": 13.276565551757812, "learning_rate": 1.894468622779533e-09, "loss": 0.6321, "step": 18524 }, { "epoch": 2.98, "grad_norm": 12.169682502746582, "learning_rate": 1.8607908963474974e-09, "loss": 0.7331, "step": 18525 }, { "epoch": 2.98, "grad_norm": 18.09868621826172, "learning_rate": 1.8274151750030666e-09, "loss": 0.7172, "step": 18526 }, { "epoch": 2.98, "grad_norm": 12.175918579101562, "learning_rate": 1.7943414597532126e-09, "loss": 0.6569, "step": 18527 }, { "epoch": 2.98, "grad_norm": 13.752628326416016, "learning_rate": 1.7615697515982466e-09, "loss": 0.7038, "step": 18528 }, { "epoch": 2.98, "grad_norm": 12.139412879943848, "learning_rate": 1.729100051527377e-09, "loss": 0.6045, "step": 18529 }, { "epoch": 2.98, "grad_norm": 11.158076286315918, "learning_rate": 1.6969323605209309e-09, "loss": 0.5223, "step": 18530 }, { "epoch": 2.98, "grad_norm": 15.965680122375488, "learning_rate": 1.6650666795514636e-09, "loss": 0.6762, "step": 18531 }, { "epoch": 2.98, "grad_norm": 11.169571876525879, "learning_rate": 1.6335030095815386e-09, "loss": 0.6377, "step": 18532 }, { "epoch": 2.98, "grad_norm": 13.394927024841309, "learning_rate": 1.6022413515626168e-09, "loss": 0.7479, "step": 18533 }, { "epoch": 2.98, "grad_norm": 14.559053421020508, "learning_rate": 1.5712817064417185e-09, "loss": 0.6804, "step": 18534 }, { "epoch": 2.98, "grad_norm": 10.881452560424805, "learning_rate": 1.5406240751525415e-09, "loss": 0.4671, "step": 18535 }, { "epoch": 2.98, "grad_norm": 14.048636436462402, "learning_rate": 1.5102684586210115e-09, "loss": 0.7566, "step": 18536 }, { "epoch": 2.98, "grad_norm": 13.739083290100098, "learning_rate": 1.480214857765283e-09, "loss": 0.6751, "step": 18537 }, { "epoch": 2.98, "grad_norm": 13.913891792297363, "learning_rate": 1.4504632734912982e-09, "loss": 0.7984, "step": 18538 }, { "epoch": 2.98, "grad_norm": 11.324922561645508, "learning_rate": 1.4210137066994478e-09, "loss": 0.6924, "step": 18539 }, { "epoch": 2.98, "grad_norm": 12.700151443481445, "learning_rate": 1.3918661582779102e-09, "loss": 0.764, "step": 18540 }, { "epoch": 2.98, "grad_norm": 11.918251037597656, "learning_rate": 1.3630206291082028e-09, "loss": 0.6401, "step": 18541 }, { "epoch": 2.98, "grad_norm": 15.749191284179688, "learning_rate": 1.3344771200607398e-09, "loss": 0.482, "step": 18542 }, { "epoch": 2.99, "grad_norm": 12.611735343933105, "learning_rate": 1.3062356319992753e-09, "loss": 0.5214, "step": 18543 }, { "epoch": 2.99, "grad_norm": 15.904733657836914, "learning_rate": 1.27829616577535e-09, "loss": 0.6492, "step": 18544 }, { "epoch": 2.99, "grad_norm": 10.82391357421875, "learning_rate": 1.2506587222327338e-09, "loss": 0.6965, "step": 18545 }, { "epoch": 2.99, "grad_norm": 11.647525787353516, "learning_rate": 1.2233233022074242e-09, "loss": 0.7055, "step": 18546 }, { "epoch": 2.99, "grad_norm": 13.883509635925293, "learning_rate": 1.1962899065243172e-09, "loss": 0.7838, "step": 18547 }, { "epoch": 2.99, "grad_norm": 14.535102844238281, "learning_rate": 1.1695585360005368e-09, "loss": 0.7216, "step": 18548 }, { "epoch": 2.99, "grad_norm": 12.613668441772461, "learning_rate": 1.1431291914432152e-09, "loss": 0.814, "step": 18549 }, { "epoch": 2.99, "grad_norm": 15.031326293945312, "learning_rate": 1.1170018736506028e-09, "loss": 0.8873, "step": 18550 }, { "epoch": 2.99, "grad_norm": 11.657931327819824, "learning_rate": 1.0911765834109579e-09, "loss": 0.619, "step": 18551 }, { "epoch": 2.99, "grad_norm": 12.993629455566406, "learning_rate": 1.0656533215069875e-09, "loss": 0.8422, "step": 18552 }, { "epoch": 2.99, "grad_norm": 14.904200553894043, "learning_rate": 1.0404320887069663e-09, "loss": 0.6552, "step": 18553 }, { "epoch": 2.99, "grad_norm": 11.91603946685791, "learning_rate": 1.0155128857747277e-09, "loss": 0.4888, "step": 18554 }, { "epoch": 2.99, "grad_norm": 10.408578872680664, "learning_rate": 9.908957134618924e-10, "loss": 0.666, "step": 18555 }, { "epoch": 2.99, "grad_norm": 13.378522872924805, "learning_rate": 9.665805725111998e-10, "loss": 0.7, "step": 18556 }, { "epoch": 2.99, "grad_norm": 11.857918739318848, "learning_rate": 9.425674636587278e-10, "loss": 0.662, "step": 18557 }, { "epoch": 2.99, "grad_norm": 17.416414260864258, "learning_rate": 9.188563876294521e-10, "loss": 0.8725, "step": 18558 }, { "epoch": 2.99, "grad_norm": 12.975193977355957, "learning_rate": 8.954473451394663e-10, "loss": 0.5804, "step": 18559 }, { "epoch": 2.99, "grad_norm": 11.415436744689941, "learning_rate": 8.723403368959826e-10, "loss": 0.658, "step": 18560 }, { "epoch": 2.99, "grad_norm": 13.965126991271973, "learning_rate": 8.495353635962211e-10, "loss": 0.6773, "step": 18561 }, { "epoch": 2.99, "grad_norm": 17.11789321899414, "learning_rate": 8.270324259296303e-10, "loss": 0.7958, "step": 18562 }, { "epoch": 2.99, "grad_norm": 13.210973739624023, "learning_rate": 8.048315245767768e-10, "loss": 0.6024, "step": 18563 }, { "epoch": 2.99, "grad_norm": 13.055146217346191, "learning_rate": 7.829326602071253e-10, "loss": 0.6536, "step": 18564 }, { "epoch": 2.99, "grad_norm": 14.112260818481445, "learning_rate": 7.613358334812582e-10, "loss": 0.7408, "step": 18565 }, { "epoch": 2.99, "grad_norm": 13.577360153198242, "learning_rate": 7.400410450542073e-10, "loss": 0.672, "step": 18566 }, { "epoch": 2.99, "grad_norm": 13.273663520812988, "learning_rate": 7.190482955665712e-10, "loss": 0.7666, "step": 18567 }, { "epoch": 2.99, "grad_norm": 13.878864288330078, "learning_rate": 6.983575856533975e-10, "loss": 0.7253, "step": 18568 }, { "epoch": 2.99, "grad_norm": 13.405611038208008, "learning_rate": 6.77968915940852e-10, "loss": 0.7286, "step": 18569 }, { "epoch": 2.99, "grad_norm": 11.557073593139648, "learning_rate": 6.57882287042888e-10, "loss": 0.594, "step": 18570 }, { "epoch": 2.99, "grad_norm": 56.045894622802734, "learning_rate": 6.380976995679078e-10, "loss": 0.8684, "step": 18571 }, { "epoch": 2.99, "grad_norm": 13.182633399963379, "learning_rate": 6.18615154112101e-10, "loss": 0.5535, "step": 18572 }, { "epoch": 2.99, "grad_norm": 11.608549118041992, "learning_rate": 5.994346512649962e-10, "loss": 0.6211, "step": 18573 }, { "epoch": 2.99, "grad_norm": 11.839028358459473, "learning_rate": 5.805561916061298e-10, "loss": 0.628, "step": 18574 }, { "epoch": 2.99, "grad_norm": 24.046281814575195, "learning_rate": 5.61979775705046e-10, "loss": 0.5879, "step": 18575 }, { "epoch": 2.99, "grad_norm": 12.342273712158203, "learning_rate": 5.437054041224077e-10, "loss": 0.7929, "step": 18576 }, { "epoch": 2.99, "grad_norm": 11.715555191040039, "learning_rate": 5.257330774111058e-10, "loss": 0.7858, "step": 18577 }, { "epoch": 2.99, "grad_norm": 11.073930740356445, "learning_rate": 5.080627961140394e-10, "loss": 0.5194, "step": 18578 }, { "epoch": 2.99, "grad_norm": 12.691548347473145, "learning_rate": 4.906945607641156e-10, "loss": 0.7323, "step": 18579 }, { "epoch": 2.99, "grad_norm": 13.428982734680176, "learning_rate": 4.736283718864698e-10, "loss": 0.6182, "step": 18580 }, { "epoch": 2.99, "grad_norm": 11.988503456115723, "learning_rate": 4.568642299973558e-10, "loss": 0.9591, "step": 18581 }, { "epoch": 2.99, "grad_norm": 22.554468154907227, "learning_rate": 4.40402135601925e-10, "loss": 0.576, "step": 18582 }, { "epoch": 2.99, "grad_norm": 14.834395408630371, "learning_rate": 4.242420891975574e-10, "loss": 0.6473, "step": 18583 }, { "epoch": 2.99, "grad_norm": 16.425737380981445, "learning_rate": 4.0838409127386126e-10, "loss": 0.9056, "step": 18584 }, { "epoch": 2.99, "grad_norm": 13.201130867004395, "learning_rate": 3.928281423082325e-10, "loss": 0.8555, "step": 18585 }, { "epoch": 2.99, "grad_norm": 10.741613388061523, "learning_rate": 3.7757424277029553e-10, "loss": 0.6678, "step": 18586 }, { "epoch": 2.99, "grad_norm": 14.082887649536133, "learning_rate": 3.626223931230133e-10, "loss": 0.6956, "step": 18587 }, { "epoch": 2.99, "grad_norm": 11.289261817932129, "learning_rate": 3.479725938160261e-10, "loss": 0.72, "step": 18588 }, { "epoch": 2.99, "grad_norm": 13.37094783782959, "learning_rate": 3.3362484529231297e-10, "loss": 0.6357, "step": 18589 }, { "epoch": 2.99, "grad_norm": 11.894315719604492, "learning_rate": 3.195791479848609e-10, "loss": 0.703, "step": 18590 }, { "epoch": 2.99, "grad_norm": 11.589043617248535, "learning_rate": 3.058355023199955e-10, "loss": 0.6614, "step": 18591 }, { "epoch": 2.99, "grad_norm": 16.761716842651367, "learning_rate": 2.9239390870960947e-10, "loss": 0.6564, "step": 18592 }, { "epoch": 2.99, "grad_norm": 12.132682800292969, "learning_rate": 2.79254367562265e-10, "loss": 0.582, "step": 18593 }, { "epoch": 2.99, "grad_norm": 12.999547004699707, "learning_rate": 2.664168792743116e-10, "loss": 0.9165, "step": 18594 }, { "epoch": 2.99, "grad_norm": 13.004779815673828, "learning_rate": 2.5388144423321715e-10, "loss": 0.5611, "step": 18595 }, { "epoch": 2.99, "grad_norm": 13.968632698059082, "learning_rate": 2.416480628175677e-10, "loss": 0.6124, "step": 18596 }, { "epoch": 2.99, "grad_norm": 15.871668815612793, "learning_rate": 2.297167353959573e-10, "loss": 0.7831, "step": 18597 }, { "epoch": 2.99, "grad_norm": 16.616249084472656, "learning_rate": 2.1808746233142885e-10, "loss": 0.6898, "step": 18598 }, { "epoch": 2.99, "grad_norm": 15.476476669311523, "learning_rate": 2.0676024397259243e-10, "loss": 0.9122, "step": 18599 }, { "epoch": 2.99, "grad_norm": 14.11747932434082, "learning_rate": 1.957350806625069e-10, "loss": 0.871, "step": 18600 }, { "epoch": 2.99, "grad_norm": 14.731621742248535, "learning_rate": 1.8501197273534944e-10, "loss": 0.7598, "step": 18601 }, { "epoch": 2.99, "grad_norm": 13.632750511169434, "learning_rate": 1.7459092051308468e-10, "loss": 0.8226, "step": 18602 }, { "epoch": 2.99, "grad_norm": 14.998197555541992, "learning_rate": 1.64471924311016e-10, "loss": 0.6575, "step": 18603 }, { "epoch": 2.99, "grad_norm": 13.59372329711914, "learning_rate": 1.5465498443667514e-10, "loss": 0.6639, "step": 18604 }, { "epoch": 3.0, "grad_norm": 17.3103084564209, "learning_rate": 1.4514010118427123e-10, "loss": 0.6397, "step": 18605 }, { "epoch": 3.0, "grad_norm": 13.496805191040039, "learning_rate": 1.3592727484135203e-10, "loss": 0.7001, "step": 18606 }, { "epoch": 3.0, "grad_norm": 17.105398178100586, "learning_rate": 1.2701650568769375e-10, "loss": 0.6808, "step": 18607 }, { "epoch": 3.0, "grad_norm": 13.594084739685059, "learning_rate": 1.1840779399197034e-10, "loss": 0.4935, "step": 18608 }, { "epoch": 3.0, "grad_norm": 11.413239479064941, "learning_rate": 1.1010114001397399e-10, "loss": 0.6767, "step": 18609 }, { "epoch": 3.0, "grad_norm": 15.338213920593262, "learning_rate": 1.0209654400350488e-10, "loss": 0.6242, "step": 18610 }, { "epoch": 3.0, "grad_norm": 15.738765716552734, "learning_rate": 9.439400620370187e-11, "loss": 0.7323, "step": 18611 }, { "epoch": 3.0, "grad_norm": 13.347275733947754, "learning_rate": 8.699352684771179e-11, "loss": 0.7543, "step": 18612 }, { "epoch": 3.0, "grad_norm": 12.076729774475098, "learning_rate": 7.989510615757923e-11, "loss": 0.6829, "step": 18613 }, { "epoch": 3.0, "grad_norm": 19.989437103271484, "learning_rate": 7.309874434868746e-11, "loss": 0.9317, "step": 18614 }, { "epoch": 3.0, "grad_norm": 13.78690242767334, "learning_rate": 6.660444162642776e-11, "loss": 0.9343, "step": 18615 }, { "epoch": 3.0, "grad_norm": 12.39465618133545, "learning_rate": 6.041219818619937e-11, "loss": 0.5482, "step": 18616 }, { "epoch": 3.0, "grad_norm": 14.61097526550293, "learning_rate": 5.452201421562997e-11, "loss": 0.6093, "step": 18617 }, { "epoch": 3.0, "grad_norm": 14.167916297912598, "learning_rate": 4.893388989235526e-11, "loss": 0.7314, "step": 18618 }, { "epoch": 3.0, "grad_norm": 14.955461502075195, "learning_rate": 4.3647825385129125e-11, "loss": 0.9538, "step": 18619 }, { "epoch": 3.0, "grad_norm": 15.828787803649902, "learning_rate": 3.86638208538237e-11, "loss": 0.9402, "step": 18620 }, { "epoch": 3.0, "grad_norm": 11.837417602539062, "learning_rate": 3.3981876448319076e-11, "loss": 0.7558, "step": 18621 }, { "epoch": 3.0, "grad_norm": 13.793660163879395, "learning_rate": 2.960199231183403e-11, "loss": 0.7023, "step": 18622 }, { "epoch": 3.0, "grad_norm": 11.758991241455078, "learning_rate": 2.5524168575374874e-11, "loss": 0.7195, "step": 18623 }, { "epoch": 3.0, "grad_norm": 16.327640533447266, "learning_rate": 2.174840536106615e-11, "loss": 0.5718, "step": 18624 }, { "epoch": 3.0, "grad_norm": 12.622299194335938, "learning_rate": 1.827470278548127e-11, "loss": 0.5658, "step": 18625 }, { "epoch": 3.0, "grad_norm": 13.504253387451172, "learning_rate": 1.5103060951870972e-11, "loss": 0.6793, "step": 18626 }, { "epoch": 3.0, "grad_norm": 15.826436996459961, "learning_rate": 1.2233479955714445e-11, "loss": 0.5835, "step": 18627 }, { "epoch": 3.0, "grad_norm": 11.063828468322754, "learning_rate": 9.6659598847193e-12, "loss": 0.6417, "step": 18628 }, { "epoch": 3.0, "grad_norm": 12.916427612304688, "learning_rate": 7.400500816601153e-12, "loss": 0.8455, "step": 18629 }, { "epoch": 3.0, "grad_norm": 12.593755722045898, "learning_rate": 5.437102817973383e-12, "loss": 0.6668, "step": 18630 }, { "epoch": 3.0, "grad_norm": 14.284482955932617, "learning_rate": 3.775765951008481e-12, "loss": 0.6385, "step": 18631 }, { "epoch": 3.0, "grad_norm": 13.897321701049805, "learning_rate": 2.4164902634460362e-12, "loss": 0.6602, "step": 18632 }, { "epoch": 3.0, "grad_norm": 21.2341251373291, "learning_rate": 1.3592757974745241e-12, "loss": 0.5908, "step": 18633 }, { "epoch": 3.0, "grad_norm": 12.301837921142578, "learning_rate": 6.04122584180189e-13, "loss": 0.8789, "step": 18634 }, { "epoch": 3.0, "grad_norm": 13.960826873779297, "learning_rate": 1.5103064687771452e-13, "loss": 0.7577, "step": 18635 }, { "epoch": 3.0, "grad_norm": 13.134809494018555, "learning_rate": 0.0, "loss": 0.5701, "step": 18636 }, { "epoch": 3.0, "step": 18636, "total_flos": 6.957976257990492e+17, "train_loss": 0.674467591889863, "train_runtime": 47470.9717, "train_samples_per_second": 6.281, "train_steps_per_second": 0.393 } ], "logging_steps": 1.0, "max_steps": 18636, "num_input_tokens_seen": 0, "num_train_epochs": 3, "save_steps": 5000, "total_flos": 6.957976257990492e+17, "train_batch_size": 4, "trial_name": null, "trial_params": null }