{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9998924384209961, "eval_steps": 500, "global_step": 13944, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "grad_norm": 21.06913729083478, "learning_rate": 2.3866348448687353e-08, "loss": 3.5549, "step": 1 }, { "epoch": 0.0, "grad_norm": 25.549658787343287, "learning_rate": 4.7732696897374706e-08, "loss": 2.2746, "step": 2 }, { "epoch": 0.0, "grad_norm": 29.165813149290287, "learning_rate": 7.159904534606206e-08, "loss": 2.2025, "step": 3 }, { "epoch": 0.0, "grad_norm": 23.995384705330235, "learning_rate": 9.546539379474941e-08, "loss": 2.334, "step": 4 }, { "epoch": 0.0, "grad_norm": 44.43121603288465, "learning_rate": 1.1933174224343676e-07, "loss": 5.2402, "step": 5 }, { "epoch": 0.0, "grad_norm": 27.8754590869235, "learning_rate": 1.431980906921241e-07, "loss": 2.301, "step": 6 }, { "epoch": 0.0, "grad_norm": 23.61943672154416, "learning_rate": 1.6706443914081147e-07, "loss": 2.1829, "step": 7 }, { "epoch": 0.0, "grad_norm": 52.6457290470021, "learning_rate": 1.9093078758949883e-07, "loss": 5.5723, "step": 8 }, { "epoch": 0.0, "grad_norm": 39.63303593010135, "learning_rate": 2.1479713603818616e-07, "loss": 5.4004, "step": 9 }, { "epoch": 0.0, "grad_norm": 28.20764589310639, "learning_rate": 2.386634844868735e-07, "loss": 2.3303, "step": 10 }, { "epoch": 0.0, "grad_norm": 24.239616327298545, "learning_rate": 2.625298329355609e-07, "loss": 2.3906, "step": 11 }, { "epoch": 0.0, "grad_norm": 29.178055838588794, "learning_rate": 2.863961813842482e-07, "loss": 2.2638, "step": 12 }, { "epoch": 0.0, "grad_norm": 37.16423615268669, "learning_rate": 3.1026252983293556e-07, "loss": 5.4111, "step": 13 }, { "epoch": 0.0, "grad_norm": 24.865195289608277, "learning_rate": 3.3412887828162294e-07, "loss": 2.3675, "step": 14 }, { "epoch": 0.0, "grad_norm": 24.222673910943623, "learning_rate": 3.579952267303103e-07, "loss": 2.483, "step": 15 }, { "epoch": 0.0, "grad_norm": 18.99256532702526, "learning_rate": 3.8186157517899765e-07, "loss": 2.2357, "step": 16 }, { "epoch": 0.0, "grad_norm": 19.102317126809563, "learning_rate": 4.0572792362768503e-07, "loss": 3.5833, "step": 17 }, { "epoch": 0.0, "grad_norm": 21.823383440751204, "learning_rate": 4.295942720763723e-07, "loss": 2.3158, "step": 18 }, { "epoch": 0.0, "grad_norm": 17.233239516255836, "learning_rate": 4.534606205250597e-07, "loss": 2.2257, "step": 19 }, { "epoch": 0.0, "grad_norm": 14.647916113362964, "learning_rate": 4.77326968973747e-07, "loss": 2.1609, "step": 20 }, { "epoch": 0.0, "grad_norm": 13.392112466439517, "learning_rate": 5.011933174224344e-07, "loss": 2.2729, "step": 21 }, { "epoch": 0.0, "grad_norm": 9.789944402849779, "learning_rate": 5.250596658711218e-07, "loss": 2.1906, "step": 22 }, { "epoch": 0.0, "grad_norm": 29.36332965073352, "learning_rate": 5.489260143198091e-07, "loss": 5.3398, "step": 23 }, { "epoch": 0.0, "grad_norm": 20.44867721270815, "learning_rate": 5.727923627684965e-07, "loss": 3.5847, "step": 24 }, { "epoch": 0.0, "grad_norm": 18.897308744747754, "learning_rate": 5.966587112171838e-07, "loss": 3.5535, "step": 25 }, { "epoch": 0.0, "grad_norm": 8.336670973010802, "learning_rate": 6.205250596658711e-07, "loss": 2.1624, "step": 26 }, { "epoch": 0.0, "grad_norm": 46.79112450548547, "learning_rate": 6.443914081145585e-07, "loss": 5.4492, "step": 27 }, { "epoch": 0.0, "grad_norm": 12.342888268147274, "learning_rate": 6.682577565632459e-07, "loss": 2.1405, "step": 28 }, { "epoch": 0.0, "grad_norm": 24.310286103412338, "learning_rate": 6.921241050119332e-07, "loss": 5.4336, "step": 29 }, { "epoch": 0.0, "grad_norm": 57.48771468233883, "learning_rate": 7.159904534606206e-07, "loss": 5.5312, "step": 30 }, { "epoch": 0.0, "grad_norm": 8.02082317188957, "learning_rate": 7.398568019093079e-07, "loss": 2.1275, "step": 31 }, { "epoch": 0.0, "grad_norm": 7.974249583930672, "learning_rate": 7.637231503579953e-07, "loss": 2.1906, "step": 32 }, { "epoch": 0.0, "grad_norm": 34.59661753237589, "learning_rate": 7.875894988066826e-07, "loss": 4.9521, "step": 33 }, { "epoch": 0.0, "grad_norm": 8.075413109926286, "learning_rate": 8.114558472553701e-07, "loss": 2.0775, "step": 34 }, { "epoch": 0.0, "grad_norm": 30.331859581188382, "learning_rate": 8.353221957040573e-07, "loss": 5.375, "step": 35 }, { "epoch": 0.0, "grad_norm": 41.339257138043585, "learning_rate": 8.591885441527446e-07, "loss": 5.2559, "step": 36 }, { "epoch": 0.0, "grad_norm": 8.795925689578722, "learning_rate": 8.830548926014321e-07, "loss": 2.0637, "step": 37 }, { "epoch": 0.0, "grad_norm": 39.85753668889101, "learning_rate": 9.069212410501194e-07, "loss": 5.1768, "step": 38 }, { "epoch": 0.0, "grad_norm": 45.724723297000146, "learning_rate": 9.307875894988068e-07, "loss": 5.127, "step": 39 }, { "epoch": 0.0, "grad_norm": 9.443936529889722, "learning_rate": 9.54653937947494e-07, "loss": 2.1573, "step": 40 }, { "epoch": 0.0, "grad_norm": 16.741297077868722, "learning_rate": 9.785202863961816e-07, "loss": 3.4359, "step": 41 }, { "epoch": 0.0, "grad_norm": 43.58239967182763, "learning_rate": 1.0023866348448687e-06, "loss": 5.3232, "step": 42 }, { "epoch": 0.0, "grad_norm": 15.685233279050415, "learning_rate": 1.026252983293556e-06, "loss": 3.4489, "step": 43 }, { "epoch": 0.0, "grad_norm": 34.05157687713875, "learning_rate": 1.0501193317422436e-06, "loss": 5.2295, "step": 44 }, { "epoch": 0.0, "grad_norm": 7.817230873743135, "learning_rate": 1.073985680190931e-06, "loss": 2.0275, "step": 45 }, { "epoch": 0.0, "grad_norm": 7.688315698183592, "learning_rate": 1.0978520286396182e-06, "loss": 2.1314, "step": 46 }, { "epoch": 0.0, "grad_norm": 8.505179702919218, "learning_rate": 1.1217183770883056e-06, "loss": 2.0533, "step": 47 }, { "epoch": 0.0, "grad_norm": 8.461252530294823, "learning_rate": 1.145584725536993e-06, "loss": 2.1772, "step": 48 }, { "epoch": 0.0, "grad_norm": 12.094869913723766, "learning_rate": 1.1694510739856802e-06, "loss": 2.0387, "step": 49 }, { "epoch": 0.0, "grad_norm": 37.82726171065179, "learning_rate": 1.1933174224343676e-06, "loss": 5.1719, "step": 50 }, { "epoch": 0.0, "grad_norm": 6.912203428329138, "learning_rate": 1.217183770883055e-06, "loss": 1.991, "step": 51 }, { "epoch": 0.0, "grad_norm": 10.065091827777783, "learning_rate": 1.2410501193317422e-06, "loss": 2.0314, "step": 52 }, { "epoch": 0.0, "grad_norm": 32.20071754361512, "learning_rate": 1.2649164677804296e-06, "loss": 4.9766, "step": 53 }, { "epoch": 0.0, "grad_norm": 6.213774197714604, "learning_rate": 1.288782816229117e-06, "loss": 1.9138, "step": 54 }, { "epoch": 0.0, "grad_norm": 6.8771485884543, "learning_rate": 1.3126491646778044e-06, "loss": 1.9771, "step": 55 }, { "epoch": 0.0, "grad_norm": 7.225403664134467, "learning_rate": 1.3365155131264918e-06, "loss": 1.9403, "step": 56 }, { "epoch": 0.0, "grad_norm": 8.7158417953512, "learning_rate": 1.360381861575179e-06, "loss": 1.8917, "step": 57 }, { "epoch": 0.0, "grad_norm": 8.861645345937653, "learning_rate": 1.3842482100238664e-06, "loss": 1.9767, "step": 58 }, { "epoch": 0.0, "grad_norm": 30.116413552398612, "learning_rate": 1.408114558472554e-06, "loss": 5.3672, "step": 59 }, { "epoch": 0.0, "grad_norm": 25.725948332943553, "learning_rate": 1.4319809069212413e-06, "loss": 4.8711, "step": 60 }, { "epoch": 0.0, "grad_norm": 5.983044963706182, "learning_rate": 1.4558472553699284e-06, "loss": 1.7758, "step": 61 }, { "epoch": 0.0, "grad_norm": 30.726165506157102, "learning_rate": 1.4797136038186157e-06, "loss": 4.8047, "step": 62 }, { "epoch": 0.0, "grad_norm": 18.784473728649264, "learning_rate": 1.5035799522673033e-06, "loss": 3.0519, "step": 63 }, { "epoch": 0.0, "grad_norm": 28.3969398489548, "learning_rate": 1.5274463007159906e-06, "loss": 4.6807, "step": 64 }, { "epoch": 0.0, "grad_norm": 7.5393990805526725, "learning_rate": 1.551312649164678e-06, "loss": 1.7661, "step": 65 }, { "epoch": 0.0, "grad_norm": 23.863552330540234, "learning_rate": 1.5751789976133653e-06, "loss": 4.7383, "step": 66 }, { "epoch": 0.0, "grad_norm": 11.239895900554616, "learning_rate": 1.5990453460620526e-06, "loss": 2.8233, "step": 67 }, { "epoch": 0.0, "grad_norm": 23.994892004636423, "learning_rate": 1.6229116945107401e-06, "loss": 4.708, "step": 68 }, { "epoch": 0.0, "grad_norm": 7.067858518719917, "learning_rate": 1.6467780429594275e-06, "loss": 1.7285, "step": 69 }, { "epoch": 0.01, "grad_norm": 18.959969254823356, "learning_rate": 1.6706443914081146e-06, "loss": 4.2939, "step": 70 }, { "epoch": 0.01, "grad_norm": 31.193841942073668, "learning_rate": 1.694510739856802e-06, "loss": 4.5771, "step": 71 }, { "epoch": 0.01, "grad_norm": 12.128467974016134, "learning_rate": 1.7183770883054892e-06, "loss": 2.5941, "step": 72 }, { "epoch": 0.01, "grad_norm": 35.64001458818317, "learning_rate": 1.7422434367541768e-06, "loss": 4.6592, "step": 73 }, { "epoch": 0.01, "grad_norm": 7.857222243013525, "learning_rate": 1.7661097852028641e-06, "loss": 1.5848, "step": 74 }, { "epoch": 0.01, "grad_norm": 7.297026383983365, "learning_rate": 1.7899761336515514e-06, "loss": 1.4273, "step": 75 }, { "epoch": 0.01, "grad_norm": 21.207235744200766, "learning_rate": 1.8138424821002388e-06, "loss": 4.7324, "step": 76 }, { "epoch": 0.01, "grad_norm": 20.748419022304567, "learning_rate": 1.8377088305489263e-06, "loss": 4.8906, "step": 77 }, { "epoch": 0.01, "grad_norm": 8.136077064231058, "learning_rate": 1.8615751789976136e-06, "loss": 1.3786, "step": 78 }, { "epoch": 0.01, "grad_norm": 6.388083322770488, "learning_rate": 1.885441527446301e-06, "loss": 1.2973, "step": 79 }, { "epoch": 0.01, "grad_norm": 6.50811836267904, "learning_rate": 1.909307875894988e-06, "loss": 1.2668, "step": 80 }, { "epoch": 0.01, "grad_norm": 17.600039704985633, "learning_rate": 1.9331742243436756e-06, "loss": 4.2891, "step": 81 }, { "epoch": 0.01, "grad_norm": 6.609138465501021, "learning_rate": 1.957040572792363e-06, "loss": 1.2467, "step": 82 }, { "epoch": 0.01, "grad_norm": 6.163915450475028, "learning_rate": 1.9809069212410503e-06, "loss": 1.1986, "step": 83 }, { "epoch": 0.01, "grad_norm": 7.479721812838956, "learning_rate": 2.0047732696897374e-06, "loss": 1.242, "step": 84 }, { "epoch": 0.01, "grad_norm": 9.817408922014158, "learning_rate": 2.028639618138425e-06, "loss": 2.2839, "step": 85 }, { "epoch": 0.01, "grad_norm": 16.729280976925313, "learning_rate": 2.052505966587112e-06, "loss": 4.0283, "step": 86 }, { "epoch": 0.01, "grad_norm": 6.3305745619980955, "learning_rate": 2.0763723150357996e-06, "loss": 0.9697, "step": 87 }, { "epoch": 0.01, "grad_norm": 8.196376433212697, "learning_rate": 2.100238663484487e-06, "loss": 1.1518, "step": 88 }, { "epoch": 0.01, "grad_norm": 8.012672969282763, "learning_rate": 2.1241050119331743e-06, "loss": 0.9725, "step": 89 }, { "epoch": 0.01, "grad_norm": 6.5369013066101145, "learning_rate": 2.147971360381862e-06, "loss": 0.9569, "step": 90 }, { "epoch": 0.01, "grad_norm": 5.933866089546631, "learning_rate": 2.1718377088305494e-06, "loss": 0.8956, "step": 91 }, { "epoch": 0.01, "grad_norm": 7.1219441686886835, "learning_rate": 2.1957040572792365e-06, "loss": 0.9532, "step": 92 }, { "epoch": 0.01, "grad_norm": 14.108742423233466, "learning_rate": 2.219570405727924e-06, "loss": 4.0566, "step": 93 }, { "epoch": 0.01, "grad_norm": 8.959782432253737, "learning_rate": 2.243436754176611e-06, "loss": 1.9001, "step": 94 }, { "epoch": 0.01, "grad_norm": 9.106748384302287, "learning_rate": 2.2673031026252983e-06, "loss": 1.8901, "step": 95 }, { "epoch": 0.01, "grad_norm": 15.988754911411538, "learning_rate": 2.291169451073986e-06, "loss": 4.0728, "step": 96 }, { "epoch": 0.01, "grad_norm": 6.85684362094771, "learning_rate": 2.3150357995226733e-06, "loss": 0.8349, "step": 97 }, { "epoch": 0.01, "grad_norm": 15.052057178765267, "learning_rate": 2.3389021479713605e-06, "loss": 3.8545, "step": 98 }, { "epoch": 0.01, "grad_norm": 7.161050435851496, "learning_rate": 2.362768496420048e-06, "loss": 0.8861, "step": 99 }, { "epoch": 0.01, "grad_norm": 6.372054667123127, "learning_rate": 2.386634844868735e-06, "loss": 0.6957, "step": 100 }, { "epoch": 0.01, "grad_norm": 8.230091750899888, "learning_rate": 2.4105011933174227e-06, "loss": 0.9634, "step": 101 }, { "epoch": 0.01, "grad_norm": 6.993268073495224, "learning_rate": 2.43436754176611e-06, "loss": 0.7829, "step": 102 }, { "epoch": 0.01, "grad_norm": 14.339672152941587, "learning_rate": 2.4582338902147973e-06, "loss": 3.7251, "step": 103 }, { "epoch": 0.01, "grad_norm": 8.534854375554868, "learning_rate": 2.4821002386634844e-06, "loss": 1.7221, "step": 104 }, { "epoch": 0.01, "grad_norm": 17.551644742774698, "learning_rate": 2.505966587112172e-06, "loss": 3.623, "step": 105 }, { "epoch": 0.01, "grad_norm": 20.905599777198415, "learning_rate": 2.529832935560859e-06, "loss": 3.7168, "step": 106 }, { "epoch": 0.01, "grad_norm": 12.346669447043382, "learning_rate": 2.5536992840095466e-06, "loss": 3.9224, "step": 107 }, { "epoch": 0.01, "grad_norm": 12.400618349635598, "learning_rate": 2.577565632458234e-06, "loss": 3.7358, "step": 108 }, { "epoch": 0.01, "grad_norm": 6.3405382404312265, "learning_rate": 2.6014319809069217e-06, "loss": 0.6995, "step": 109 }, { "epoch": 0.01, "grad_norm": 8.337232985187772, "learning_rate": 2.625298329355609e-06, "loss": 0.6838, "step": 110 }, { "epoch": 0.01, "grad_norm": 23.570896040973015, "learning_rate": 2.6491646778042964e-06, "loss": 3.3179, "step": 111 }, { "epoch": 0.01, "grad_norm": 7.668610887948283, "learning_rate": 2.6730310262529835e-06, "loss": 0.7199, "step": 112 }, { "epoch": 0.01, "grad_norm": 8.332279431163434, "learning_rate": 2.6968973747016706e-06, "loss": 0.7372, "step": 113 }, { "epoch": 0.01, "grad_norm": 9.471522255019442, "learning_rate": 2.720763723150358e-06, "loss": 1.6035, "step": 114 }, { "epoch": 0.01, "grad_norm": 11.142327627488601, "learning_rate": 2.7446300715990453e-06, "loss": 1.6751, "step": 115 }, { "epoch": 0.01, "grad_norm": 14.342823153294171, "learning_rate": 2.768496420047733e-06, "loss": 3.6992, "step": 116 }, { "epoch": 0.01, "grad_norm": 6.692280831157525, "learning_rate": 2.7923627684964204e-06, "loss": 0.7968, "step": 117 }, { "epoch": 0.01, "grad_norm": 6.079785036380918, "learning_rate": 2.816229116945108e-06, "loss": 0.6583, "step": 118 }, { "epoch": 0.01, "grad_norm": 10.75060930581779, "learning_rate": 2.840095465393795e-06, "loss": 3.459, "step": 119 }, { "epoch": 0.01, "grad_norm": 14.063851884704066, "learning_rate": 2.8639618138424826e-06, "loss": 3.4121, "step": 120 }, { "epoch": 0.01, "grad_norm": 23.268065460105625, "learning_rate": 2.8878281622911697e-06, "loss": 4.0439, "step": 121 }, { "epoch": 0.01, "grad_norm": 10.34766044598067, "learning_rate": 2.911694510739857e-06, "loss": 1.4622, "step": 122 }, { "epoch": 0.01, "grad_norm": 27.467192810414996, "learning_rate": 2.9355608591885443e-06, "loss": 3.0957, "step": 123 }, { "epoch": 0.01, "grad_norm": 9.51929236074937, "learning_rate": 2.9594272076372315e-06, "loss": 1.5903, "step": 124 }, { "epoch": 0.01, "grad_norm": 8.315531056502394, "learning_rate": 2.983293556085919e-06, "loss": 1.6946, "step": 125 }, { "epoch": 0.01, "grad_norm": 39.20846351886089, "learning_rate": 3.0071599045346066e-06, "loss": 3.9521, "step": 126 }, { "epoch": 0.01, "grad_norm": 19.67517037162141, "learning_rate": 3.031026252983294e-06, "loss": 3.7104, "step": 127 }, { "epoch": 0.01, "grad_norm": 10.038142446279359, "learning_rate": 3.054892601431981e-06, "loss": 0.6101, "step": 128 }, { "epoch": 0.01, "grad_norm": 45.99769605929335, "learning_rate": 3.0787589498806688e-06, "loss": 3.1899, "step": 129 }, { "epoch": 0.01, "grad_norm": 8.364881314920572, "learning_rate": 3.102625298329356e-06, "loss": 1.604, "step": 130 }, { "epoch": 0.01, "grad_norm": 11.104729502095145, "learning_rate": 3.126491646778043e-06, "loss": 3.9141, "step": 131 }, { "epoch": 0.01, "grad_norm": 7.00668978652339, "learning_rate": 3.1503579952267305e-06, "loss": 0.6393, "step": 132 }, { "epoch": 0.01, "grad_norm": 5.838740067311361, "learning_rate": 3.1742243436754176e-06, "loss": 0.6697, "step": 133 }, { "epoch": 0.01, "grad_norm": 6.693773883283984, "learning_rate": 3.198090692124105e-06, "loss": 0.645, "step": 134 }, { "epoch": 0.01, "grad_norm": 5.603772974131767, "learning_rate": 3.2219570405727923e-06, "loss": 0.4947, "step": 135 }, { "epoch": 0.01, "grad_norm": 11.20935104827344, "learning_rate": 3.2458233890214803e-06, "loss": 3.3855, "step": 136 }, { "epoch": 0.01, "grad_norm": 23.674698407459164, "learning_rate": 3.2696897374701674e-06, "loss": 3.1797, "step": 137 }, { "epoch": 0.01, "grad_norm": 6.788270137141241, "learning_rate": 3.293556085918855e-06, "loss": 0.6735, "step": 138 }, { "epoch": 0.01, "grad_norm": 7.000023249490818, "learning_rate": 3.317422434367542e-06, "loss": 0.6491, "step": 139 }, { "epoch": 0.01, "grad_norm": 19.88096323812759, "learning_rate": 3.341288782816229e-06, "loss": 3.1335, "step": 140 }, { "epoch": 0.01, "grad_norm": 7.637710656524167, "learning_rate": 3.3651551312649167e-06, "loss": 0.7012, "step": 141 }, { "epoch": 0.01, "grad_norm": 27.031503177588675, "learning_rate": 3.389021479713604e-06, "loss": 3.9536, "step": 142 }, { "epoch": 0.01, "grad_norm": 7.1904317884127265, "learning_rate": 3.4128878281622914e-06, "loss": 0.6312, "step": 143 }, { "epoch": 0.01, "grad_norm": 53.1289279559027, "learning_rate": 3.4367541766109785e-06, "loss": 3.751, "step": 144 }, { "epoch": 0.01, "grad_norm": 9.895740745243366, "learning_rate": 3.4606205250596665e-06, "loss": 0.628, "step": 145 }, { "epoch": 0.01, "grad_norm": 9.686437712071445, "learning_rate": 3.4844868735083536e-06, "loss": 0.6114, "step": 146 }, { "epoch": 0.01, "grad_norm": 24.320094908749102, "learning_rate": 3.508353221957041e-06, "loss": 3.5254, "step": 147 }, { "epoch": 0.01, "grad_norm": 13.513922171772306, "learning_rate": 3.5322195704057282e-06, "loss": 3.2949, "step": 148 }, { "epoch": 0.01, "grad_norm": 8.216338429899288, "learning_rate": 3.5560859188544154e-06, "loss": 1.4706, "step": 149 }, { "epoch": 0.01, "grad_norm": 24.779673101548784, "learning_rate": 3.579952267303103e-06, "loss": 3.3164, "step": 150 }, { "epoch": 0.01, "grad_norm": 11.801564318449037, "learning_rate": 3.60381861575179e-06, "loss": 3.0073, "step": 151 }, { "epoch": 0.01, "grad_norm": 10.52650984290866, "learning_rate": 3.6276849642004776e-06, "loss": 3.1406, "step": 152 }, { "epoch": 0.01, "grad_norm": 11.292114417140233, "learning_rate": 3.6515513126491647e-06, "loss": 3.2847, "step": 153 }, { "epoch": 0.01, "grad_norm": 14.674162231319773, "learning_rate": 3.6754176610978526e-06, "loss": 3.3779, "step": 154 }, { "epoch": 0.01, "grad_norm": 7.967915420411316, "learning_rate": 3.6992840095465398e-06, "loss": 1.5841, "step": 155 }, { "epoch": 0.01, "grad_norm": 10.856785400053196, "learning_rate": 3.7231503579952273e-06, "loss": 2.8315, "step": 156 }, { "epoch": 0.01, "grad_norm": 12.978697413211018, "learning_rate": 3.7470167064439144e-06, "loss": 3.2979, "step": 157 }, { "epoch": 0.01, "grad_norm": 14.146274445056331, "learning_rate": 3.770883054892602e-06, "loss": 3.4351, "step": 158 }, { "epoch": 0.01, "grad_norm": 7.407390662827984, "learning_rate": 3.794749403341289e-06, "loss": 0.7598, "step": 159 }, { "epoch": 0.01, "grad_norm": 6.045692175661569, "learning_rate": 3.818615751789976e-06, "loss": 0.517, "step": 160 }, { "epoch": 0.01, "grad_norm": 11.767308685677122, "learning_rate": 3.842482100238663e-06, "loss": 3.1045, "step": 161 }, { "epoch": 0.01, "grad_norm": 7.067704346293094, "learning_rate": 3.866348448687351e-06, "loss": 0.6455, "step": 162 }, { "epoch": 0.01, "grad_norm": 5.350362179988843, "learning_rate": 3.890214797136038e-06, "loss": 0.5251, "step": 163 }, { "epoch": 0.01, "grad_norm": 6.06443864014607, "learning_rate": 3.914081145584726e-06, "loss": 0.6487, "step": 164 }, { "epoch": 0.01, "grad_norm": 13.849055781318475, "learning_rate": 3.9379474940334135e-06, "loss": 2.8286, "step": 165 }, { "epoch": 0.01, "grad_norm": 8.42154064824658, "learning_rate": 3.961813842482101e-06, "loss": 0.6066, "step": 166 }, { "epoch": 0.01, "grad_norm": 9.1251020711232, "learning_rate": 3.985680190930788e-06, "loss": 1.7285, "step": 167 }, { "epoch": 0.01, "grad_norm": 29.106749578061926, "learning_rate": 4.009546539379475e-06, "loss": 3.3359, "step": 168 }, { "epoch": 0.01, "grad_norm": 6.618012610742909, "learning_rate": 4.033412887828163e-06, "loss": 0.5871, "step": 169 }, { "epoch": 0.01, "grad_norm": 13.436295457755978, "learning_rate": 4.05727923627685e-06, "loss": 3.3174, "step": 170 }, { "epoch": 0.01, "grad_norm": 6.041072545182817, "learning_rate": 4.081145584725537e-06, "loss": 0.5395, "step": 171 }, { "epoch": 0.01, "grad_norm": 5.887691273185817, "learning_rate": 4.105011933174224e-06, "loss": 0.541, "step": 172 }, { "epoch": 0.01, "grad_norm": 14.939642800360572, "learning_rate": 4.128878281622912e-06, "loss": 2.9009, "step": 173 }, { "epoch": 0.01, "grad_norm": 21.630954402885628, "learning_rate": 4.152744630071599e-06, "loss": 3.1162, "step": 174 }, { "epoch": 0.01, "grad_norm": 6.562219992959233, "learning_rate": 4.176610978520287e-06, "loss": 0.5912, "step": 175 }, { "epoch": 0.01, "grad_norm": 8.87761197130283, "learning_rate": 4.200477326968974e-06, "loss": 1.5514, "step": 176 }, { "epoch": 0.01, "grad_norm": 6.8508210665656915, "learning_rate": 4.2243436754176614e-06, "loss": 0.5939, "step": 177 }, { "epoch": 0.01, "grad_norm": 8.359830479189194, "learning_rate": 4.2482100238663486e-06, "loss": 1.371, "step": 178 }, { "epoch": 0.01, "grad_norm": 7.681500675275438, "learning_rate": 4.272076372315036e-06, "loss": 0.7275, "step": 179 }, { "epoch": 0.01, "grad_norm": 6.584081880582891, "learning_rate": 4.295942720763724e-06, "loss": 0.541, "step": 180 }, { "epoch": 0.01, "grad_norm": 6.712230225470642, "learning_rate": 4.319809069212411e-06, "loss": 0.7163, "step": 181 }, { "epoch": 0.01, "grad_norm": 7.019598318387011, "learning_rate": 4.343675417661099e-06, "loss": 0.6074, "step": 182 }, { "epoch": 0.01, "grad_norm": 6.075711418660591, "learning_rate": 4.367541766109786e-06, "loss": 0.5562, "step": 183 }, { "epoch": 0.01, "grad_norm": 16.445767406993284, "learning_rate": 4.391408114558473e-06, "loss": 3.1299, "step": 184 }, { "epoch": 0.01, "grad_norm": 6.063400904252105, "learning_rate": 4.41527446300716e-06, "loss": 0.5538, "step": 185 }, { "epoch": 0.01, "grad_norm": 7.326465843550126, "learning_rate": 4.439140811455848e-06, "loss": 0.6189, "step": 186 }, { "epoch": 0.01, "grad_norm": 7.114859356263161, "learning_rate": 4.463007159904535e-06, "loss": 0.6058, "step": 187 }, { "epoch": 0.01, "grad_norm": 16.70203282896011, "learning_rate": 4.486873508353222e-06, "loss": 2.3582, "step": 188 }, { "epoch": 0.01, "grad_norm": 8.166255281735108, "learning_rate": 4.510739856801909e-06, "loss": 1.5235, "step": 189 }, { "epoch": 0.01, "grad_norm": 23.748976480235132, "learning_rate": 4.5346062052505965e-06, "loss": 2.8718, "step": 190 }, { "epoch": 0.01, "grad_norm": 8.017401611293606, "learning_rate": 4.5584725536992845e-06, "loss": 0.4686, "step": 191 }, { "epoch": 0.01, "grad_norm": 62.57903412305517, "learning_rate": 4.582338902147972e-06, "loss": 3.3533, "step": 192 }, { "epoch": 0.01, "grad_norm": 12.98026404261257, "learning_rate": 4.6062052505966596e-06, "loss": 2.521, "step": 193 }, { "epoch": 0.01, "grad_norm": 7.705986484193463, "learning_rate": 4.630071599045347e-06, "loss": 0.6022, "step": 194 }, { "epoch": 0.01, "grad_norm": 7.373265599847962, "learning_rate": 4.653937947494034e-06, "loss": 0.603, "step": 195 }, { "epoch": 0.01, "grad_norm": 6.7660549700827195, "learning_rate": 4.677804295942721e-06, "loss": 0.4762, "step": 196 }, { "epoch": 0.01, "grad_norm": 9.509832914619578, "learning_rate": 4.701670644391408e-06, "loss": 0.582, "step": 197 }, { "epoch": 0.01, "grad_norm": 7.80660289622208, "learning_rate": 4.725536992840096e-06, "loss": 0.6957, "step": 198 }, { "epoch": 0.01, "grad_norm": 8.1604041650588, "learning_rate": 4.749403341288783e-06, "loss": 1.4616, "step": 199 }, { "epoch": 0.01, "grad_norm": 19.734263447160302, "learning_rate": 4.77326968973747e-06, "loss": 2.8047, "step": 200 }, { "epoch": 0.01, "grad_norm": 6.833201671839634, "learning_rate": 4.797136038186158e-06, "loss": 0.5702, "step": 201 }, { "epoch": 0.01, "grad_norm": 7.532754259531044, "learning_rate": 4.821002386634845e-06, "loss": 0.5888, "step": 202 }, { "epoch": 0.01, "grad_norm": 8.718833455505697, "learning_rate": 4.8448687350835324e-06, "loss": 1.4651, "step": 203 }, { "epoch": 0.01, "grad_norm": 7.287785519553843, "learning_rate": 4.86873508353222e-06, "loss": 0.6145, "step": 204 }, { "epoch": 0.01, "grad_norm": 8.935448727752648, "learning_rate": 4.8926014319809075e-06, "loss": 1.6023, "step": 205 }, { "epoch": 0.01, "grad_norm": 8.355981015661111, "learning_rate": 4.916467780429595e-06, "loss": 1.5177, "step": 206 }, { "epoch": 0.01, "grad_norm": 14.92200338190543, "learning_rate": 4.940334128878282e-06, "loss": 2.619, "step": 207 }, { "epoch": 0.01, "grad_norm": 7.199278885581448, "learning_rate": 4.964200477326969e-06, "loss": 0.6258, "step": 208 }, { "epoch": 0.01, "grad_norm": 8.754137762999193, "learning_rate": 4.988066825775657e-06, "loss": 1.446, "step": 209 }, { "epoch": 0.02, "grad_norm": 6.232454790417008, "learning_rate": 5.011933174224344e-06, "loss": 0.5506, "step": 210 }, { "epoch": 0.02, "grad_norm": 70.71727161685175, "learning_rate": 5.035799522673032e-06, "loss": 2.7161, "step": 211 }, { "epoch": 0.02, "grad_norm": 8.866538067790406, "learning_rate": 5.059665871121718e-06, "loss": 0.605, "step": 212 }, { "epoch": 0.02, "grad_norm": 13.81837088036001, "learning_rate": 5.083532219570406e-06, "loss": 2.4746, "step": 213 }, { "epoch": 0.02, "grad_norm": 7.342826745232244, "learning_rate": 5.107398568019093e-06, "loss": 0.6954, "step": 214 }, { "epoch": 0.02, "grad_norm": 20.310748856731866, "learning_rate": 5.13126491646778e-06, "loss": 2.4471, "step": 215 }, { "epoch": 0.02, "grad_norm": 14.72326457925547, "learning_rate": 5.155131264916468e-06, "loss": 2.0164, "step": 216 }, { "epoch": 0.02, "grad_norm": 6.821039717843807, "learning_rate": 5.1789976133651555e-06, "loss": 0.652, "step": 217 }, { "epoch": 0.02, "grad_norm": 5.667959517471572, "learning_rate": 5.2028639618138435e-06, "loss": 0.5512, "step": 218 }, { "epoch": 0.02, "grad_norm": 14.132432863314015, "learning_rate": 5.22673031026253e-06, "loss": 2.625, "step": 219 }, { "epoch": 0.02, "grad_norm": 6.713320976461797, "learning_rate": 5.250596658711218e-06, "loss": 0.5268, "step": 220 }, { "epoch": 0.02, "grad_norm": 10.197985766746019, "learning_rate": 5.274463007159905e-06, "loss": 1.3446, "step": 221 }, { "epoch": 0.02, "grad_norm": 7.581266020294474, "learning_rate": 5.298329355608593e-06, "loss": 0.5693, "step": 222 }, { "epoch": 0.02, "grad_norm": 9.829916505284933, "learning_rate": 5.322195704057279e-06, "loss": 0.7419, "step": 223 }, { "epoch": 0.02, "grad_norm": 16.647849735845256, "learning_rate": 5.346062052505967e-06, "loss": 2.2246, "step": 224 }, { "epoch": 0.02, "grad_norm": 6.85932498670985, "learning_rate": 5.369928400954655e-06, "loss": 0.5861, "step": 225 }, { "epoch": 0.02, "grad_norm": 7.2906162830846135, "learning_rate": 5.393794749403341e-06, "loss": 0.5917, "step": 226 }, { "epoch": 0.02, "grad_norm": 6.647792344200744, "learning_rate": 5.417661097852029e-06, "loss": 0.5657, "step": 227 }, { "epoch": 0.02, "grad_norm": 58.55233515090872, "learning_rate": 5.441527446300716e-06, "loss": 2.4912, "step": 228 }, { "epoch": 0.02, "grad_norm": 7.927736981778081, "learning_rate": 5.465393794749404e-06, "loss": 0.5803, "step": 229 }, { "epoch": 0.02, "grad_norm": 9.874050620911866, "learning_rate": 5.4892601431980906e-06, "loss": 1.5193, "step": 230 }, { "epoch": 0.02, "grad_norm": 9.833575475737405, "learning_rate": 5.5131264916467785e-06, "loss": 1.4932, "step": 231 }, { "epoch": 0.02, "grad_norm": 7.217992406058742, "learning_rate": 5.536992840095466e-06, "loss": 0.4929, "step": 232 }, { "epoch": 0.02, "grad_norm": 70.4232028571082, "learning_rate": 5.560859188544153e-06, "loss": 3.1343, "step": 233 }, { "epoch": 0.02, "grad_norm": 10.992643679178872, "learning_rate": 5.584725536992841e-06, "loss": 0.6417, "step": 234 }, { "epoch": 0.02, "grad_norm": 21.044753095996757, "learning_rate": 5.608591885441528e-06, "loss": 2.2366, "step": 235 }, { "epoch": 0.02, "grad_norm": 8.482846129042201, "learning_rate": 5.632458233890216e-06, "loss": 0.5832, "step": 236 }, { "epoch": 0.02, "grad_norm": 6.887489484819354, "learning_rate": 5.656324582338902e-06, "loss": 0.5691, "step": 237 }, { "epoch": 0.02, "grad_norm": 6.243101577105483, "learning_rate": 5.68019093078759e-06, "loss": 0.4656, "step": 238 }, { "epoch": 0.02, "grad_norm": 7.653526569322158, "learning_rate": 5.704057279236277e-06, "loss": 0.6885, "step": 239 }, { "epoch": 0.02, "grad_norm": 7.5113063184157935, "learning_rate": 5.727923627684965e-06, "loss": 0.4817, "step": 240 }, { "epoch": 0.02, "grad_norm": 8.920491509907414, "learning_rate": 5.751789976133651e-06, "loss": 1.4045, "step": 241 }, { "epoch": 0.02, "grad_norm": 7.937192887634819, "learning_rate": 5.775656324582339e-06, "loss": 0.5435, "step": 242 }, { "epoch": 0.02, "grad_norm": 14.738189478891146, "learning_rate": 5.799522673031027e-06, "loss": 2.4761, "step": 243 }, { "epoch": 0.02, "grad_norm": 8.252244493395645, "learning_rate": 5.823389021479714e-06, "loss": 1.5884, "step": 244 }, { "epoch": 0.02, "grad_norm": 7.833741071187496, "learning_rate": 5.847255369928402e-06, "loss": 0.6439, "step": 245 }, { "epoch": 0.02, "grad_norm": 7.798380041772368, "learning_rate": 5.871121718377089e-06, "loss": 1.5044, "step": 246 }, { "epoch": 0.02, "grad_norm": 6.9935704496423, "learning_rate": 5.894988066825777e-06, "loss": 0.6106, "step": 247 }, { "epoch": 0.02, "grad_norm": 13.154632010659002, "learning_rate": 5.918854415274463e-06, "loss": 2.0779, "step": 248 }, { "epoch": 0.02, "grad_norm": 6.658057623331864, "learning_rate": 5.942720763723151e-06, "loss": 0.5571, "step": 249 }, { "epoch": 0.02, "grad_norm": 5.878372182951577, "learning_rate": 5.966587112171838e-06, "loss": 0.5398, "step": 250 }, { "epoch": 0.02, "grad_norm": 7.192364759194606, "learning_rate": 5.990453460620526e-06, "loss": 0.5492, "step": 251 }, { "epoch": 0.02, "grad_norm": 6.636112825414706, "learning_rate": 6.014319809069213e-06, "loss": 0.6031, "step": 252 }, { "epoch": 0.02, "grad_norm": 6.865840408166868, "learning_rate": 6.0381861575179e-06, "loss": 0.5871, "step": 253 }, { "epoch": 0.02, "grad_norm": 5.59279916468544, "learning_rate": 6.062052505966588e-06, "loss": 0.5326, "step": 254 }, { "epoch": 0.02, "grad_norm": 21.12749984735492, "learning_rate": 6.0859188544152745e-06, "loss": 2.5752, "step": 255 }, { "epoch": 0.02, "grad_norm": 7.159116503912414, "learning_rate": 6.109785202863962e-06, "loss": 0.6395, "step": 256 }, { "epoch": 0.02, "grad_norm": 7.9403026314425995, "learning_rate": 6.1336515513126495e-06, "loss": 0.5586, "step": 257 }, { "epoch": 0.02, "grad_norm": 8.131794871156831, "learning_rate": 6.1575178997613375e-06, "loss": 1.5734, "step": 258 }, { "epoch": 0.02, "grad_norm": 6.872488177091165, "learning_rate": 6.181384248210024e-06, "loss": 0.6016, "step": 259 }, { "epoch": 0.02, "grad_norm": 6.846990836217444, "learning_rate": 6.205250596658712e-06, "loss": 0.5617, "step": 260 }, { "epoch": 0.02, "grad_norm": 8.182273355687148, "learning_rate": 6.2291169451074e-06, "loss": 1.2774, "step": 261 }, { "epoch": 0.02, "grad_norm": 17.15033819440034, "learning_rate": 6.252983293556086e-06, "loss": 2.2043, "step": 262 }, { "epoch": 0.02, "grad_norm": 8.46693697683212, "learning_rate": 6.276849642004774e-06, "loss": 0.5981, "step": 263 }, { "epoch": 0.02, "grad_norm": 6.909791162360944, "learning_rate": 6.300715990453461e-06, "loss": 0.5529, "step": 264 }, { "epoch": 0.02, "grad_norm": 6.51646136689271, "learning_rate": 6.324582338902149e-06, "loss": 0.5615, "step": 265 }, { "epoch": 0.02, "grad_norm": 7.49798391665265, "learning_rate": 6.348448687350835e-06, "loss": 1.3299, "step": 266 }, { "epoch": 0.02, "grad_norm": 7.092056734506342, "learning_rate": 6.372315035799523e-06, "loss": 0.6266, "step": 267 }, { "epoch": 0.02, "grad_norm": 8.283754975966033, "learning_rate": 6.39618138424821e-06, "loss": 1.532, "step": 268 }, { "epoch": 0.02, "grad_norm": 19.190216283997025, "learning_rate": 6.420047732696898e-06, "loss": 3.1016, "step": 269 }, { "epoch": 0.02, "grad_norm": 15.499803001311696, "learning_rate": 6.443914081145585e-06, "loss": 2.0278, "step": 270 }, { "epoch": 0.02, "grad_norm": 20.733448974113042, "learning_rate": 6.467780429594273e-06, "loss": 2.5032, "step": 271 }, { "epoch": 0.02, "grad_norm": 44.79019899709578, "learning_rate": 6.4916467780429605e-06, "loss": 2.1743, "step": 272 }, { "epoch": 0.02, "grad_norm": 30.98278341850108, "learning_rate": 6.515513126491647e-06, "loss": 2.6951, "step": 273 }, { "epoch": 0.02, "grad_norm": 7.94567856617445, "learning_rate": 6.539379474940335e-06, "loss": 0.4874, "step": 274 }, { "epoch": 0.02, "grad_norm": 7.618266767508625, "learning_rate": 6.563245823389022e-06, "loss": 0.5268, "step": 275 }, { "epoch": 0.02, "grad_norm": 7.752892237581922, "learning_rate": 6.58711217183771e-06, "loss": 0.575, "step": 276 }, { "epoch": 0.02, "grad_norm": 7.949531722214186, "learning_rate": 6.610978520286396e-06, "loss": 0.5819, "step": 277 }, { "epoch": 0.02, "grad_norm": 6.560430858530162, "learning_rate": 6.634844868735084e-06, "loss": 0.4767, "step": 278 }, { "epoch": 0.02, "grad_norm": 6.707223854181556, "learning_rate": 6.658711217183771e-06, "loss": 0.5592, "step": 279 }, { "epoch": 0.02, "grad_norm": 7.61063057619553, "learning_rate": 6.682577565632458e-06, "loss": 0.6857, "step": 280 }, { "epoch": 0.02, "grad_norm": 25.033931653597094, "learning_rate": 6.706443914081146e-06, "loss": 2.8706, "step": 281 }, { "epoch": 0.02, "grad_norm": 16.110089710864898, "learning_rate": 6.7303102625298334e-06, "loss": 2.3521, "step": 282 }, { "epoch": 0.02, "grad_norm": 13.465908609878221, "learning_rate": 6.754176610978521e-06, "loss": 2.165, "step": 283 }, { "epoch": 0.02, "grad_norm": 5.955565591437533, "learning_rate": 6.778042959427208e-06, "loss": 0.5685, "step": 284 }, { "epoch": 0.02, "grad_norm": 5.452319465742983, "learning_rate": 6.801909307875896e-06, "loss": 0.525, "step": 285 }, { "epoch": 0.02, "grad_norm": 6.94706533026497, "learning_rate": 6.825775656324583e-06, "loss": 0.6671, "step": 286 }, { "epoch": 0.02, "grad_norm": 19.665573038555447, "learning_rate": 6.849642004773271e-06, "loss": 2.9375, "step": 287 }, { "epoch": 0.02, "grad_norm": 7.846480741213654, "learning_rate": 6.873508353221957e-06, "loss": 0.4554, "step": 288 }, { "epoch": 0.02, "grad_norm": 16.200875399605557, "learning_rate": 6.897374701670645e-06, "loss": 2.1436, "step": 289 }, { "epoch": 0.02, "grad_norm": 5.644538383347044, "learning_rate": 6.921241050119333e-06, "loss": 0.3895, "step": 290 }, { "epoch": 0.02, "grad_norm": 16.035348126147078, "learning_rate": 6.945107398568019e-06, "loss": 2.4424, "step": 291 }, { "epoch": 0.02, "grad_norm": 6.696599033949783, "learning_rate": 6.968973747016707e-06, "loss": 0.4531, "step": 292 }, { "epoch": 0.02, "grad_norm": 21.92607040951585, "learning_rate": 6.992840095465394e-06, "loss": 2.6357, "step": 293 }, { "epoch": 0.02, "grad_norm": 6.289807276169132, "learning_rate": 7.016706443914082e-06, "loss": 0.5123, "step": 294 }, { "epoch": 0.02, "grad_norm": 16.045029981945177, "learning_rate": 7.0405727923627685e-06, "loss": 2.1221, "step": 295 }, { "epoch": 0.02, "grad_norm": 7.6954330962576405, "learning_rate": 7.0644391408114565e-06, "loss": 0.6445, "step": 296 }, { "epoch": 0.02, "grad_norm": 68.29605921478806, "learning_rate": 7.088305489260144e-06, "loss": 2.5156, "step": 297 }, { "epoch": 0.02, "grad_norm": 6.713255492306333, "learning_rate": 7.112171837708831e-06, "loss": 0.541, "step": 298 }, { "epoch": 0.02, "grad_norm": 8.51454345210288, "learning_rate": 7.136038186157519e-06, "loss": 1.5266, "step": 299 }, { "epoch": 0.02, "grad_norm": 7.692309202471743, "learning_rate": 7.159904534606206e-06, "loss": 0.4852, "step": 300 }, { "epoch": 0.02, "grad_norm": 23.724883157849238, "learning_rate": 7.183770883054894e-06, "loss": 2.5049, "step": 301 }, { "epoch": 0.02, "grad_norm": 7.528851184814045, "learning_rate": 7.20763723150358e-06, "loss": 0.624, "step": 302 }, { "epoch": 0.02, "grad_norm": 7.235983491559889, "learning_rate": 7.231503579952268e-06, "loss": 0.5937, "step": 303 }, { "epoch": 0.02, "grad_norm": 18.535943756449935, "learning_rate": 7.255369928400955e-06, "loss": 2.3817, "step": 304 }, { "epoch": 0.02, "grad_norm": 6.730144273929128, "learning_rate": 7.279236276849643e-06, "loss": 0.5209, "step": 305 }, { "epoch": 0.02, "grad_norm": 7.868903667207591, "learning_rate": 7.303102625298329e-06, "loss": 1.4545, "step": 306 }, { "epoch": 0.02, "grad_norm": 54.14905726438483, "learning_rate": 7.326968973747017e-06, "loss": 2.1632, "step": 307 }, { "epoch": 0.02, "grad_norm": 6.187734287795712, "learning_rate": 7.350835322195705e-06, "loss": 0.5592, "step": 308 }, { "epoch": 0.02, "grad_norm": 6.688629906398678, "learning_rate": 7.3747016706443915e-06, "loss": 0.4714, "step": 309 }, { "epoch": 0.02, "grad_norm": 8.383084080275946, "learning_rate": 7.3985680190930795e-06, "loss": 0.537, "step": 310 }, { "epoch": 0.02, "grad_norm": 25.560969054926026, "learning_rate": 7.422434367541767e-06, "loss": 2.3887, "step": 311 }, { "epoch": 0.02, "grad_norm": 28.0804704334018, "learning_rate": 7.446300715990455e-06, "loss": 2.646, "step": 312 }, { "epoch": 0.02, "grad_norm": 18.686953473132363, "learning_rate": 7.470167064439141e-06, "loss": 2.1851, "step": 313 }, { "epoch": 0.02, "grad_norm": 6.666068908577361, "learning_rate": 7.494033412887829e-06, "loss": 0.4546, "step": 314 }, { "epoch": 0.02, "grad_norm": 21.202356493039137, "learning_rate": 7.517899761336516e-06, "loss": 2.0972, "step": 315 }, { "epoch": 0.02, "grad_norm": 20.687542811899778, "learning_rate": 7.541766109785204e-06, "loss": 1.7666, "step": 316 }, { "epoch": 0.02, "grad_norm": 16.264815454432572, "learning_rate": 7.565632458233891e-06, "loss": 2.1479, "step": 317 }, { "epoch": 0.02, "grad_norm": 8.340460323982494, "learning_rate": 7.589498806682578e-06, "loss": 1.3883, "step": 318 }, { "epoch": 0.02, "grad_norm": 7.044561656895319, "learning_rate": 7.613365155131266e-06, "loss": 0.5866, "step": 319 }, { "epoch": 0.02, "grad_norm": 6.305274360995857, "learning_rate": 7.637231503579952e-06, "loss": 0.5284, "step": 320 }, { "epoch": 0.02, "grad_norm": 6.2123023400476916, "learning_rate": 7.661097852028641e-06, "loss": 0.5171, "step": 321 }, { "epoch": 0.02, "grad_norm": 7.231427817562743, "learning_rate": 7.684964200477327e-06, "loss": 0.5799, "step": 322 }, { "epoch": 0.02, "grad_norm": 6.537617205799375, "learning_rate": 7.708830548926015e-06, "loss": 0.5241, "step": 323 }, { "epoch": 0.02, "grad_norm": 17.099080278847307, "learning_rate": 7.732696897374703e-06, "loss": 1.8303, "step": 324 }, { "epoch": 0.02, "grad_norm": 6.549512741128894, "learning_rate": 7.75656324582339e-06, "loss": 0.4787, "step": 325 }, { "epoch": 0.02, "grad_norm": 7.394139657182929, "learning_rate": 7.780429594272077e-06, "loss": 0.627, "step": 326 }, { "epoch": 0.02, "grad_norm": 7.043442898036025, "learning_rate": 7.804295942720764e-06, "loss": 0.5122, "step": 327 }, { "epoch": 0.02, "grad_norm": 5.410757814788419, "learning_rate": 7.828162291169453e-06, "loss": 0.4941, "step": 328 }, { "epoch": 0.02, "grad_norm": 5.961593639280332, "learning_rate": 7.852028639618138e-06, "loss": 0.5169, "step": 329 }, { "epoch": 0.02, "grad_norm": 5.760143745983816, "learning_rate": 7.875894988066827e-06, "loss": 0.4691, "step": 330 }, { "epoch": 0.02, "grad_norm": 17.72860361231765, "learning_rate": 7.899761336515514e-06, "loss": 2.5703, "step": 331 }, { "epoch": 0.02, "grad_norm": 6.700013263489254, "learning_rate": 7.923627684964201e-06, "loss": 0.5943, "step": 332 }, { "epoch": 0.02, "grad_norm": 22.551329196992132, "learning_rate": 7.947494033412888e-06, "loss": 2.5942, "step": 333 }, { "epoch": 0.02, "grad_norm": 8.101767758555406, "learning_rate": 7.971360381861575e-06, "loss": 0.5296, "step": 334 }, { "epoch": 0.02, "grad_norm": 9.017809413617684, "learning_rate": 7.995226730310263e-06, "loss": 1.4977, "step": 335 }, { "epoch": 0.02, "grad_norm": 8.316191553319227, "learning_rate": 8.01909307875895e-06, "loss": 1.2584, "step": 336 }, { "epoch": 0.02, "grad_norm": 15.74216235270556, "learning_rate": 8.042959427207638e-06, "loss": 1.7373, "step": 337 }, { "epoch": 0.02, "grad_norm": 6.426339838040901, "learning_rate": 8.066825775656326e-06, "loss": 0.5611, "step": 338 }, { "epoch": 0.02, "grad_norm": 6.997478697386499, "learning_rate": 8.090692124105013e-06, "loss": 1.3327, "step": 339 }, { "epoch": 0.02, "grad_norm": 7.0895625488763825, "learning_rate": 8.1145584725537e-06, "loss": 0.4858, "step": 340 }, { "epoch": 0.02, "grad_norm": 7.098391310510913, "learning_rate": 8.138424821002387e-06, "loss": 1.4742, "step": 341 }, { "epoch": 0.02, "grad_norm": 5.7433615624994845, "learning_rate": 8.162291169451074e-06, "loss": 0.5558, "step": 342 }, { "epoch": 0.02, "grad_norm": 19.004349550471098, "learning_rate": 8.186157517899763e-06, "loss": 2.4226, "step": 343 }, { "epoch": 0.02, "grad_norm": 8.718869313048527, "learning_rate": 8.210023866348448e-06, "loss": 0.4079, "step": 344 }, { "epoch": 0.02, "grad_norm": 9.38868207269892, "learning_rate": 8.233890214797137e-06, "loss": 0.5084, "step": 345 }, { "epoch": 0.02, "grad_norm": 9.326104483690232, "learning_rate": 8.257756563245824e-06, "loss": 0.4856, "step": 346 }, { "epoch": 0.02, "grad_norm": 8.98764497862847, "learning_rate": 8.281622911694511e-06, "loss": 0.5485, "step": 347 }, { "epoch": 0.02, "grad_norm": 10.81041914144272, "learning_rate": 8.305489260143198e-06, "loss": 1.5223, "step": 348 }, { "epoch": 0.03, "grad_norm": 9.403970258217486, "learning_rate": 8.329355608591886e-06, "loss": 0.5785, "step": 349 }, { "epoch": 0.03, "grad_norm": 8.894599877202195, "learning_rate": 8.353221957040574e-06, "loss": 0.4974, "step": 350 }, { "epoch": 0.03, "grad_norm": 8.716930416084566, "learning_rate": 8.37708830548926e-06, "loss": 0.4614, "step": 351 }, { "epoch": 0.03, "grad_norm": 10.201299023442774, "learning_rate": 8.400954653937949e-06, "loss": 0.5885, "step": 352 }, { "epoch": 0.03, "grad_norm": 9.261116317890176, "learning_rate": 8.424821002386636e-06, "loss": 0.5513, "step": 353 }, { "epoch": 0.03, "grad_norm": 20.832083239178026, "learning_rate": 8.448687350835323e-06, "loss": 2.5447, "step": 354 }, { "epoch": 0.03, "grad_norm": 7.015445941008297, "learning_rate": 8.47255369928401e-06, "loss": 0.5315, "step": 355 }, { "epoch": 0.03, "grad_norm": 6.786300924466145, "learning_rate": 8.496420047732697e-06, "loss": 0.5829, "step": 356 }, { "epoch": 0.03, "grad_norm": 6.738296465176958, "learning_rate": 8.520286396181386e-06, "loss": 0.4844, "step": 357 }, { "epoch": 0.03, "grad_norm": 7.014244503430905, "learning_rate": 8.544152744630071e-06, "loss": 0.5556, "step": 358 }, { "epoch": 0.03, "grad_norm": 22.462022789122415, "learning_rate": 8.56801909307876e-06, "loss": 2.1438, "step": 359 }, { "epoch": 0.03, "grad_norm": 6.617632650985344, "learning_rate": 8.591885441527447e-06, "loss": 0.4773, "step": 360 }, { "epoch": 0.03, "grad_norm": 7.186399243468252, "learning_rate": 8.615751789976134e-06, "loss": 0.59, "step": 361 }, { "epoch": 0.03, "grad_norm": 5.877336011260149, "learning_rate": 8.639618138424822e-06, "loss": 0.4748, "step": 362 }, { "epoch": 0.03, "grad_norm": 6.092991170436388, "learning_rate": 8.663484486873509e-06, "loss": 0.4761, "step": 363 }, { "epoch": 0.03, "grad_norm": 14.284977109321037, "learning_rate": 8.687350835322197e-06, "loss": 1.8435, "step": 364 }, { "epoch": 0.03, "grad_norm": 5.663179139310932, "learning_rate": 8.711217183770883e-06, "loss": 0.4062, "step": 365 }, { "epoch": 0.03, "grad_norm": 14.620274698637035, "learning_rate": 8.735083532219572e-06, "loss": 1.8573, "step": 366 }, { "epoch": 0.03, "grad_norm": 5.2201358907227435, "learning_rate": 8.758949880668259e-06, "loss": 0.3996, "step": 367 }, { "epoch": 0.03, "grad_norm": 7.623632669384718, "learning_rate": 8.782816229116946e-06, "loss": 1.2442, "step": 368 }, { "epoch": 0.03, "grad_norm": 5.655405946214731, "learning_rate": 8.806682577565633e-06, "loss": 0.4256, "step": 369 }, { "epoch": 0.03, "grad_norm": 6.420645171004693, "learning_rate": 8.83054892601432e-06, "loss": 0.5504, "step": 370 }, { "epoch": 0.03, "grad_norm": 15.601090056181983, "learning_rate": 8.854415274463007e-06, "loss": 2.5271, "step": 371 }, { "epoch": 0.03, "grad_norm": 6.735698502439842, "learning_rate": 8.878281622911696e-06, "loss": 0.5077, "step": 372 }, { "epoch": 0.03, "grad_norm": 6.767367540760681, "learning_rate": 8.902147971360383e-06, "loss": 0.495, "step": 373 }, { "epoch": 0.03, "grad_norm": 20.043412177833893, "learning_rate": 8.92601431980907e-06, "loss": 2.3521, "step": 374 }, { "epoch": 0.03, "grad_norm": 6.619292776004069, "learning_rate": 8.949880668257757e-06, "loss": 0.5362, "step": 375 }, { "epoch": 0.03, "grad_norm": 26.66724807003266, "learning_rate": 8.973747016706445e-06, "loss": 2.5602, "step": 376 }, { "epoch": 0.03, "grad_norm": 7.435695784835214, "learning_rate": 8.997613365155132e-06, "loss": 0.4946, "step": 377 }, { "epoch": 0.03, "grad_norm": 17.340544061345845, "learning_rate": 9.021479713603819e-06, "loss": 2.0405, "step": 378 }, { "epoch": 0.03, "grad_norm": 17.61151414047877, "learning_rate": 9.045346062052508e-06, "loss": 2.0137, "step": 379 }, { "epoch": 0.03, "grad_norm": 5.765441038975913, "learning_rate": 9.069212410501193e-06, "loss": 0.4239, "step": 380 }, { "epoch": 0.03, "grad_norm": 7.314460100454243, "learning_rate": 9.093078758949882e-06, "loss": 0.5404, "step": 381 }, { "epoch": 0.03, "grad_norm": 5.74499455157619, "learning_rate": 9.116945107398569e-06, "loss": 0.5211, "step": 382 }, { "epoch": 0.03, "grad_norm": 7.26582103994141, "learning_rate": 9.140811455847256e-06, "loss": 0.5681, "step": 383 }, { "epoch": 0.03, "grad_norm": 15.954770386204627, "learning_rate": 9.164677804295943e-06, "loss": 2.4993, "step": 384 }, { "epoch": 0.03, "grad_norm": 58.24094435290336, "learning_rate": 9.18854415274463e-06, "loss": 2.3823, "step": 385 }, { "epoch": 0.03, "grad_norm": 23.10177305818425, "learning_rate": 9.212410501193319e-06, "loss": 2.0168, "step": 386 }, { "epoch": 0.03, "grad_norm": 12.514046359620904, "learning_rate": 9.236276849642005e-06, "loss": 1.5558, "step": 387 }, { "epoch": 0.03, "grad_norm": 6.366165018224452, "learning_rate": 9.260143198090693e-06, "loss": 0.5755, "step": 388 }, { "epoch": 0.03, "grad_norm": 6.144688245658441, "learning_rate": 9.28400954653938e-06, "loss": 0.6527, "step": 389 }, { "epoch": 0.03, "grad_norm": 5.439984879606275, "learning_rate": 9.307875894988068e-06, "loss": 0.4877, "step": 390 }, { "epoch": 0.03, "grad_norm": 6.165039379431597, "learning_rate": 9.331742243436755e-06, "loss": 0.518, "step": 391 }, { "epoch": 0.03, "grad_norm": 5.824787889574625, "learning_rate": 9.355608591885442e-06, "loss": 0.4696, "step": 392 }, { "epoch": 0.03, "grad_norm": 111.00325671327036, "learning_rate": 9.37947494033413e-06, "loss": 3.8964, "step": 393 }, { "epoch": 0.03, "grad_norm": 13.85492569090323, "learning_rate": 9.403341288782816e-06, "loss": 1.3045, "step": 394 }, { "epoch": 0.03, "grad_norm": 13.961083064339558, "learning_rate": 9.427207637231505e-06, "loss": 0.5952, "step": 395 }, { "epoch": 0.03, "grad_norm": 26.609723663165127, "learning_rate": 9.451073985680192e-06, "loss": 2.4878, "step": 396 }, { "epoch": 0.03, "grad_norm": 21.976308598733386, "learning_rate": 9.474940334128879e-06, "loss": 2.689, "step": 397 }, { "epoch": 0.03, "grad_norm": 7.182548052040225, "learning_rate": 9.498806682577566e-06, "loss": 0.5391, "step": 398 }, { "epoch": 0.03, "grad_norm": 23.642476481421916, "learning_rate": 9.522673031026253e-06, "loss": 2.3073, "step": 399 }, { "epoch": 0.03, "grad_norm": 7.247616173282442, "learning_rate": 9.54653937947494e-06, "loss": 1.482, "step": 400 }, { "epoch": 0.03, "grad_norm": 6.841677804030877, "learning_rate": 9.570405727923628e-06, "loss": 0.4485, "step": 401 }, { "epoch": 0.03, "grad_norm": 7.136883901921768, "learning_rate": 9.594272076372316e-06, "loss": 0.5777, "step": 402 }, { "epoch": 0.03, "grad_norm": 21.325519309418425, "learning_rate": 9.618138424821004e-06, "loss": 2.3315, "step": 403 }, { "epoch": 0.03, "grad_norm": 9.211450959946715, "learning_rate": 9.64200477326969e-06, "loss": 1.3166, "step": 404 }, { "epoch": 0.03, "grad_norm": 8.649402631486625, "learning_rate": 9.665871121718378e-06, "loss": 0.637, "step": 405 }, { "epoch": 0.03, "grad_norm": 31.86286778850959, "learning_rate": 9.689737470167065e-06, "loss": 1.9042, "step": 406 }, { "epoch": 0.03, "grad_norm": 19.458810080395633, "learning_rate": 9.713603818615752e-06, "loss": 2.2773, "step": 407 }, { "epoch": 0.03, "grad_norm": 7.8471294098038396, "learning_rate": 9.73747016706444e-06, "loss": 0.5752, "step": 408 }, { "epoch": 0.03, "grad_norm": 16.25205289021579, "learning_rate": 9.761336515513126e-06, "loss": 2.3625, "step": 409 }, { "epoch": 0.03, "grad_norm": 7.886619303605399, "learning_rate": 9.785202863961815e-06, "loss": 0.5184, "step": 410 }, { "epoch": 0.03, "grad_norm": 64.19928352523118, "learning_rate": 9.809069212410502e-06, "loss": 2.3348, "step": 411 }, { "epoch": 0.03, "grad_norm": 29.307285741892837, "learning_rate": 9.83293556085919e-06, "loss": 2.8472, "step": 412 }, { "epoch": 0.03, "grad_norm": 7.958193368366796, "learning_rate": 9.856801909307876e-06, "loss": 0.5062, "step": 413 }, { "epoch": 0.03, "grad_norm": 37.165229901619696, "learning_rate": 9.880668257756564e-06, "loss": 2.4521, "step": 414 }, { "epoch": 0.03, "grad_norm": 21.80830337609787, "learning_rate": 9.904534606205252e-06, "loss": 2.1421, "step": 415 }, { "epoch": 0.03, "grad_norm": 16.172367009775144, "learning_rate": 9.928400954653938e-06, "loss": 2.1333, "step": 416 }, { "epoch": 0.03, "grad_norm": 8.33637684634632, "learning_rate": 9.952267303102627e-06, "loss": 1.4923, "step": 417 }, { "epoch": 0.03, "grad_norm": 7.1695485924718305, "learning_rate": 9.976133651551314e-06, "loss": 1.2803, "step": 418 }, { "epoch": 0.03, "grad_norm": 7.201903719458476, "learning_rate": 1e-05, "loss": 0.6431, "step": 419 }, { "epoch": 0.03, "grad_norm": 6.6312694660628475, "learning_rate": 9.999999865134465e-06, "loss": 0.4465, "step": 420 }, { "epoch": 0.03, "grad_norm": 7.422711402023177, "learning_rate": 9.999999460537861e-06, "loss": 1.4324, "step": 421 }, { "epoch": 0.03, "grad_norm": 6.307099821385748, "learning_rate": 9.999998786210212e-06, "loss": 0.4439, "step": 422 }, { "epoch": 0.03, "grad_norm": 16.382009183203323, "learning_rate": 9.999997842151555e-06, "loss": 1.5945, "step": 423 }, { "epoch": 0.03, "grad_norm": 8.288376818158763, "learning_rate": 9.99999662836194e-06, "loss": 1.5003, "step": 424 }, { "epoch": 0.03, "grad_norm": 13.71468699332148, "learning_rate": 9.999995144841433e-06, "loss": 1.8373, "step": 425 }, { "epoch": 0.03, "grad_norm": 15.799110641057544, "learning_rate": 9.999993391590115e-06, "loss": 2.0803, "step": 426 }, { "epoch": 0.03, "grad_norm": 5.855412983862229, "learning_rate": 9.999991368608079e-06, "loss": 0.4181, "step": 427 }, { "epoch": 0.03, "grad_norm": 6.531482587514036, "learning_rate": 9.999989075895434e-06, "loss": 0.4362, "step": 428 }, { "epoch": 0.03, "grad_norm": 6.092018523369561, "learning_rate": 9.999986513452305e-06, "loss": 0.4115, "step": 429 }, { "epoch": 0.03, "grad_norm": 5.815447994288978, "learning_rate": 9.99998368127883e-06, "loss": 0.3334, "step": 430 }, { "epoch": 0.03, "grad_norm": 22.662607234962593, "learning_rate": 9.999980579375158e-06, "loss": 2.0103, "step": 431 }, { "epoch": 0.03, "grad_norm": 6.462573339745543, "learning_rate": 9.999977207741466e-06, "loss": 0.5103, "step": 432 }, { "epoch": 0.03, "grad_norm": 6.972333699664965, "learning_rate": 9.999973566377923e-06, "loss": 0.5554, "step": 433 }, { "epoch": 0.03, "grad_norm": 27.335105758632995, "learning_rate": 9.999969655284737e-06, "loss": 1.9734, "step": 434 }, { "epoch": 0.03, "grad_norm": 7.393147336269817, "learning_rate": 9.99996547446211e-06, "loss": 0.4582, "step": 435 }, { "epoch": 0.03, "grad_norm": 11.92975699468618, "learning_rate": 9.999961023910274e-06, "loss": 1.4122, "step": 436 }, { "epoch": 0.03, "grad_norm": 19.622712503233558, "learning_rate": 9.999956303629467e-06, "loss": 2.3147, "step": 437 }, { "epoch": 0.03, "grad_norm": 7.1993664724663065, "learning_rate": 9.999951313619942e-06, "loss": 0.4809, "step": 438 }, { "epoch": 0.03, "grad_norm": 10.487899984901096, "learning_rate": 9.999946053881971e-06, "loss": 1.3677, "step": 439 }, { "epoch": 0.03, "grad_norm": 8.346535275027493, "learning_rate": 9.999940524415835e-06, "loss": 1.245, "step": 440 }, { "epoch": 0.03, "grad_norm": 5.884720219123637, "learning_rate": 9.999934725221834e-06, "loss": 0.4143, "step": 441 }, { "epoch": 0.03, "grad_norm": 6.507004453542836, "learning_rate": 9.99992865630028e-06, "loss": 0.5878, "step": 442 }, { "epoch": 0.03, "grad_norm": 28.586286013344537, "learning_rate": 9.9999223176515e-06, "loss": 1.5312, "step": 443 }, { "epoch": 0.03, "grad_norm": 8.673738905161304, "learning_rate": 9.999915709275839e-06, "loss": 1.2873, "step": 444 }, { "epoch": 0.03, "grad_norm": 14.636611215028559, "learning_rate": 9.999908831173649e-06, "loss": 1.649, "step": 445 }, { "epoch": 0.03, "grad_norm": 4.978036689096491, "learning_rate": 9.999901683345304e-06, "loss": 0.3665, "step": 446 }, { "epoch": 0.03, "grad_norm": 8.063360968528857, "learning_rate": 9.999894265791188e-06, "loss": 0.4417, "step": 447 }, { "epoch": 0.03, "grad_norm": 7.944924353624397, "learning_rate": 9.999886578511703e-06, "loss": 0.5303, "step": 448 }, { "epoch": 0.03, "grad_norm": 37.42346614399741, "learning_rate": 9.999878621507263e-06, "loss": 1.266, "step": 449 }, { "epoch": 0.03, "grad_norm": 7.919630064600641, "learning_rate": 9.999870394778294e-06, "loss": 1.4381, "step": 450 }, { "epoch": 0.03, "grad_norm": 7.590563349951277, "learning_rate": 9.999861898325246e-06, "loss": 1.3605, "step": 451 }, { "epoch": 0.03, "grad_norm": 123.982025606754, "learning_rate": 9.999853132148573e-06, "loss": 2.4904, "step": 452 }, { "epoch": 0.03, "grad_norm": 24.498650391998897, "learning_rate": 9.99984409624875e-06, "loss": 2.9536, "step": 453 }, { "epoch": 0.03, "grad_norm": 9.308664842631174, "learning_rate": 9.999834790626262e-06, "loss": 1.3443, "step": 454 }, { "epoch": 0.03, "grad_norm": 105.54675278087764, "learning_rate": 9.999825215281613e-06, "loss": 2.6169, "step": 455 }, { "epoch": 0.03, "grad_norm": 8.326831340001187, "learning_rate": 9.99981537021532e-06, "loss": 1.3431, "step": 456 }, { "epoch": 0.03, "grad_norm": 7.968027515366357, "learning_rate": 9.999805255427912e-06, "loss": 0.4609, "step": 457 }, { "epoch": 0.03, "grad_norm": 31.879793887746757, "learning_rate": 9.999794870919936e-06, "loss": 1.7493, "step": 458 }, { "epoch": 0.03, "grad_norm": 56.74478341830899, "learning_rate": 9.999784216691953e-06, "loss": 1.7128, "step": 459 }, { "epoch": 0.03, "grad_norm": 6.945240481536872, "learning_rate": 9.999773292744534e-06, "loss": 1.2526, "step": 460 }, { "epoch": 0.03, "grad_norm": 6.9519734543394085, "learning_rate": 9.999762099078274e-06, "loss": 0.5967, "step": 461 }, { "epoch": 0.03, "grad_norm": 6.971391190718617, "learning_rate": 9.999750635693772e-06, "loss": 1.2751, "step": 462 }, { "epoch": 0.03, "grad_norm": 6.399827628193818, "learning_rate": 9.999738902591648e-06, "loss": 0.4855, "step": 463 }, { "epoch": 0.03, "grad_norm": 32.69503992118082, "learning_rate": 9.999726899772537e-06, "loss": 2.4089, "step": 464 }, { "epoch": 0.03, "grad_norm": 25.213028636124058, "learning_rate": 9.999714627237084e-06, "loss": 1.6688, "step": 465 }, { "epoch": 0.03, "grad_norm": 5.894807368226712, "learning_rate": 9.999702084985952e-06, "loss": 0.5124, "step": 466 }, { "epoch": 0.03, "grad_norm": 8.277914705930494, "learning_rate": 9.999689273019817e-06, "loss": 1.3963, "step": 467 }, { "epoch": 0.03, "grad_norm": 6.98488955312875, "learning_rate": 9.999676191339369e-06, "loss": 0.382, "step": 468 }, { "epoch": 0.03, "grad_norm": 6.1357642810185276, "learning_rate": 9.999662839945317e-06, "loss": 0.3074, "step": 469 }, { "epoch": 0.03, "grad_norm": 6.066717790822141, "learning_rate": 9.999649218838381e-06, "loss": 0.4195, "step": 470 }, { "epoch": 0.03, "grad_norm": 32.48355900998288, "learning_rate": 9.999635328019293e-06, "loss": 1.9042, "step": 471 }, { "epoch": 0.03, "grad_norm": 13.37381239057721, "learning_rate": 9.999621167488803e-06, "loss": 1.6222, "step": 472 }, { "epoch": 0.03, "grad_norm": 25.98090921410072, "learning_rate": 9.999606737247675e-06, "loss": 1.972, "step": 473 }, { "epoch": 0.03, "grad_norm": 16.643290758293798, "learning_rate": 9.99959203729669e-06, "loss": 1.8824, "step": 474 }, { "epoch": 0.03, "grad_norm": 6.361403265440034, "learning_rate": 9.999577067636638e-06, "loss": 0.445, "step": 475 }, { "epoch": 0.03, "grad_norm": 7.605969665092573, "learning_rate": 9.999561828268328e-06, "loss": 0.487, "step": 476 }, { "epoch": 0.03, "grad_norm": 8.23905842657315, "learning_rate": 9.999546319192581e-06, "loss": 1.3209, "step": 477 }, { "epoch": 0.03, "grad_norm": 6.8153471667269345, "learning_rate": 9.999530540410236e-06, "loss": 0.4154, "step": 478 }, { "epoch": 0.03, "grad_norm": 16.45079514542793, "learning_rate": 9.999514491922141e-06, "loss": 1.8352, "step": 479 }, { "epoch": 0.03, "grad_norm": 14.864776075501698, "learning_rate": 9.999498173729164e-06, "loss": 1.6729, "step": 480 }, { "epoch": 0.03, "grad_norm": 6.219076372496219, "learning_rate": 9.999481585832185e-06, "loss": 0.4468, "step": 481 }, { "epoch": 0.03, "grad_norm": 6.3843807828415455, "learning_rate": 9.999464728232099e-06, "loss": 0.433, "step": 482 }, { "epoch": 0.03, "grad_norm": 17.94054960173829, "learning_rate": 9.999447600929813e-06, "loss": 1.8193, "step": 483 }, { "epoch": 0.03, "grad_norm": 16.066716440707452, "learning_rate": 9.999430203926253e-06, "loss": 1.5386, "step": 484 }, { "epoch": 0.03, "grad_norm": 6.628826944470622, "learning_rate": 9.99941253722236e-06, "loss": 0.4078, "step": 485 }, { "epoch": 0.03, "grad_norm": 7.441152186253721, "learning_rate": 9.999394600819082e-06, "loss": 0.5394, "step": 486 }, { "epoch": 0.03, "grad_norm": 6.6363219739859, "learning_rate": 9.99937639471739e-06, "loss": 0.364, "step": 487 }, { "epoch": 0.03, "grad_norm": 15.341154558040248, "learning_rate": 9.999357918918264e-06, "loss": 1.478, "step": 488 }, { "epoch": 0.04, "grad_norm": 13.43478414883832, "learning_rate": 9.999339173422702e-06, "loss": 1.6501, "step": 489 }, { "epoch": 0.04, "grad_norm": 6.503184037117259, "learning_rate": 9.999320158231717e-06, "loss": 0.4487, "step": 490 }, { "epoch": 0.04, "grad_norm": 9.726314735416649, "learning_rate": 9.999300873346332e-06, "loss": 1.2324, "step": 491 }, { "epoch": 0.04, "grad_norm": 9.75363490848292, "learning_rate": 9.999281318767588e-06, "loss": 1.3616, "step": 492 }, { "epoch": 0.04, "grad_norm": 20.414536832132868, "learning_rate": 9.99926149449654e-06, "loss": 1.8557, "step": 493 }, { "epoch": 0.04, "grad_norm": 22.24082806429749, "learning_rate": 9.999241400534257e-06, "loss": 2.1653, "step": 494 }, { "epoch": 0.04, "grad_norm": 32.426882973918985, "learning_rate": 9.999221036881825e-06, "loss": 2.2471, "step": 495 }, { "epoch": 0.04, "grad_norm": 8.463379036570567, "learning_rate": 9.99920040354034e-06, "loss": 1.2341, "step": 496 }, { "epoch": 0.04, "grad_norm": 6.217912586805053, "learning_rate": 9.999179500510916e-06, "loss": 0.3707, "step": 497 }, { "epoch": 0.04, "grad_norm": 6.985550610160262, "learning_rate": 9.999158327794682e-06, "loss": 0.5201, "step": 498 }, { "epoch": 0.04, "grad_norm": 6.121174282302712, "learning_rate": 9.99913688539278e-06, "loss": 0.3244, "step": 499 }, { "epoch": 0.04, "grad_norm": 14.626394902613917, "learning_rate": 9.999115173306365e-06, "loss": 1.8064, "step": 500 }, { "epoch": 0.04, "grad_norm": 16.523235661826877, "learning_rate": 9.999093191536607e-06, "loss": 1.5515, "step": 501 }, { "epoch": 0.04, "grad_norm": 20.36778837006366, "learning_rate": 9.999070940084697e-06, "loss": 2.0349, "step": 502 }, { "epoch": 0.04, "grad_norm": 5.530845272219675, "learning_rate": 9.99904841895183e-06, "loss": 0.3801, "step": 503 }, { "epoch": 0.04, "grad_norm": 5.932637460987958, "learning_rate": 9.999025628139224e-06, "loss": 0.378, "step": 504 }, { "epoch": 0.04, "grad_norm": 5.700810913613519, "learning_rate": 9.999002567648109e-06, "loss": 0.3151, "step": 505 }, { "epoch": 0.04, "grad_norm": 6.923677770139593, "learning_rate": 9.998979237479725e-06, "loss": 1.2841, "step": 506 }, { "epoch": 0.04, "grad_norm": 5.618472961535218, "learning_rate": 9.998955637635334e-06, "loss": 0.3595, "step": 507 }, { "epoch": 0.04, "grad_norm": 23.80997706339047, "learning_rate": 9.99893176811621e-06, "loss": 1.6941, "step": 508 }, { "epoch": 0.04, "grad_norm": 47.09224678104177, "learning_rate": 9.998907628923638e-06, "loss": 1.2878, "step": 509 }, { "epoch": 0.04, "grad_norm": 7.105236881770997, "learning_rate": 9.998883220058921e-06, "loss": 0.484, "step": 510 }, { "epoch": 0.04, "grad_norm": 6.44470024623023, "learning_rate": 9.998858541523378e-06, "loss": 0.4612, "step": 511 }, { "epoch": 0.04, "grad_norm": 18.798623886900973, "learning_rate": 9.998833593318336e-06, "loss": 1.8616, "step": 512 }, { "epoch": 0.04, "grad_norm": 12.39121589429627, "learning_rate": 9.998808375445145e-06, "loss": 0.3426, "step": 513 }, { "epoch": 0.04, "grad_norm": 27.488071917382168, "learning_rate": 9.998782887905162e-06, "loss": 1.8767, "step": 514 }, { "epoch": 0.04, "grad_norm": 9.261028390106619, "learning_rate": 9.998757130699766e-06, "loss": 1.3715, "step": 515 }, { "epoch": 0.04, "grad_norm": 6.685563431264771, "learning_rate": 9.998731103830342e-06, "loss": 0.4801, "step": 516 }, { "epoch": 0.04, "grad_norm": 13.73773445499547, "learning_rate": 9.998704807298297e-06, "loss": 1.6589, "step": 517 }, { "epoch": 0.04, "grad_norm": 7.206967157647374, "learning_rate": 9.998678241105048e-06, "loss": 0.6539, "step": 518 }, { "epoch": 0.04, "grad_norm": 5.2331199359895875, "learning_rate": 9.99865140525203e-06, "loss": 0.341, "step": 519 }, { "epoch": 0.04, "grad_norm": 5.868388876839941, "learning_rate": 9.99862429974069e-06, "loss": 0.3563, "step": 520 }, { "epoch": 0.04, "grad_norm": 5.496857774297562, "learning_rate": 9.998596924572488e-06, "loss": 0.345, "step": 521 }, { "epoch": 0.04, "grad_norm": 7.045478807554108, "learning_rate": 9.998569279748904e-06, "loss": 1.3604, "step": 522 }, { "epoch": 0.04, "grad_norm": 6.645978538628741, "learning_rate": 9.998541365271428e-06, "loss": 0.4065, "step": 523 }, { "epoch": 0.04, "grad_norm": 25.92649874997683, "learning_rate": 9.998513181141565e-06, "loss": 1.6674, "step": 524 }, { "epoch": 0.04, "grad_norm": 18.05363183086338, "learning_rate": 9.998484727360836e-06, "loss": 2.3994, "step": 525 }, { "epoch": 0.04, "grad_norm": 6.225804796440901, "learning_rate": 9.998456003930777e-06, "loss": 0.4407, "step": 526 }, { "epoch": 0.04, "grad_norm": 6.335178482297757, "learning_rate": 9.998427010852938e-06, "loss": 0.4152, "step": 527 }, { "epoch": 0.04, "grad_norm": 6.622612828947076, "learning_rate": 9.99839774812888e-06, "loss": 0.5512, "step": 528 }, { "epoch": 0.04, "grad_norm": 6.2422907521594615, "learning_rate": 9.998368215760184e-06, "loss": 0.4489, "step": 529 }, { "epoch": 0.04, "grad_norm": 6.167761479688236, "learning_rate": 9.998338413748443e-06, "loss": 0.4149, "step": 530 }, { "epoch": 0.04, "grad_norm": 19.77028115833888, "learning_rate": 9.998308342095265e-06, "loss": 2.2864, "step": 531 }, { "epoch": 0.04, "grad_norm": 39.807581623804985, "learning_rate": 9.99827800080227e-06, "loss": 1.7183, "step": 532 }, { "epoch": 0.04, "grad_norm": 7.716108245671921, "learning_rate": 9.998247389871096e-06, "loss": 0.5095, "step": 533 }, { "epoch": 0.04, "grad_norm": 6.736486789971277, "learning_rate": 9.998216509303395e-06, "loss": 0.4055, "step": 534 }, { "epoch": 0.04, "grad_norm": 12.700368110057985, "learning_rate": 9.998185359100834e-06, "loss": 1.4054, "step": 535 }, { "epoch": 0.04, "grad_norm": 5.502116242365439, "learning_rate": 9.998153939265091e-06, "loss": 0.3723, "step": 536 }, { "epoch": 0.04, "grad_norm": 44.55996309381452, "learning_rate": 9.998122249797863e-06, "loss": 1.6631, "step": 537 }, { "epoch": 0.04, "grad_norm": 28.059550210424568, "learning_rate": 9.998090290700856e-06, "loss": 2.0108, "step": 538 }, { "epoch": 0.04, "grad_norm": 92.27813114341173, "learning_rate": 9.9980580619758e-06, "loss": 1.4612, "step": 539 }, { "epoch": 0.04, "grad_norm": 5.930120062992984, "learning_rate": 9.998025563624429e-06, "loss": 0.4539, "step": 540 }, { "epoch": 0.04, "grad_norm": 6.554863476975863, "learning_rate": 9.997992795648496e-06, "loss": 1.3211, "step": 541 }, { "epoch": 0.04, "grad_norm": 5.2790495468744085, "learning_rate": 9.99795975804977e-06, "loss": 0.2945, "step": 542 }, { "epoch": 0.04, "grad_norm": 8.169375339063253, "learning_rate": 9.997926450830036e-06, "loss": 1.5464, "step": 543 }, { "epoch": 0.04, "grad_norm": 5.648224494284476, "learning_rate": 9.997892873991086e-06, "loss": 0.4821, "step": 544 }, { "epoch": 0.04, "grad_norm": 5.508527599810656, "learning_rate": 9.997859027534735e-06, "loss": 0.4085, "step": 545 }, { "epoch": 0.04, "grad_norm": 17.292020694754175, "learning_rate": 9.997824911462805e-06, "loss": 1.9139, "step": 546 }, { "epoch": 0.04, "grad_norm": 20.381600354720902, "learning_rate": 9.997790525777142e-06, "loss": 1.8778, "step": 547 }, { "epoch": 0.04, "grad_norm": 7.729765438207888, "learning_rate": 9.997755870479595e-06, "loss": 0.4055, "step": 548 }, { "epoch": 0.04, "grad_norm": 16.82690118049962, "learning_rate": 9.997720945572038e-06, "loss": 1.1285, "step": 549 }, { "epoch": 0.04, "grad_norm": 10.37558927431236, "learning_rate": 9.997685751056354e-06, "loss": 1.588, "step": 550 }, { "epoch": 0.04, "grad_norm": 8.136910041204832, "learning_rate": 9.99765028693444e-06, "loss": 0.5216, "step": 551 }, { "epoch": 0.04, "grad_norm": 17.271898168894012, "learning_rate": 9.99761455320821e-06, "loss": 1.7515, "step": 552 }, { "epoch": 0.04, "grad_norm": 7.291579163178733, "learning_rate": 9.997578549879592e-06, "loss": 0.4363, "step": 553 }, { "epoch": 0.04, "grad_norm": 6.163716389821705, "learning_rate": 9.997542276950528e-06, "loss": 0.4286, "step": 554 }, { "epoch": 0.04, "grad_norm": 17.83786731421156, "learning_rate": 9.997505734422974e-06, "loss": 1.8761, "step": 555 }, { "epoch": 0.04, "grad_norm": 15.5257700818671, "learning_rate": 9.997468922298904e-06, "loss": 1.4453, "step": 556 }, { "epoch": 0.04, "grad_norm": 12.333379192567259, "learning_rate": 9.997431840580301e-06, "loss": 1.9024, "step": 557 }, { "epoch": 0.04, "grad_norm": 37.045720464439725, "learning_rate": 9.997394489269165e-06, "loss": 2.1091, "step": 558 }, { "epoch": 0.04, "grad_norm": 29.639221478933536, "learning_rate": 9.997356868367517e-06, "loss": 2.1572, "step": 559 }, { "epoch": 0.04, "grad_norm": 16.14489064080532, "learning_rate": 9.997318977877377e-06, "loss": 2.0302, "step": 560 }, { "epoch": 0.04, "grad_norm": 7.331559376244857, "learning_rate": 9.997280817800797e-06, "loss": 1.2744, "step": 561 }, { "epoch": 0.04, "grad_norm": 6.317732890394334, "learning_rate": 9.997242388139832e-06, "loss": 0.4135, "step": 562 }, { "epoch": 0.04, "grad_norm": 5.198250458394617, "learning_rate": 9.997203688896554e-06, "loss": 0.3188, "step": 563 }, { "epoch": 0.04, "grad_norm": 24.98479280913613, "learning_rate": 9.997164720073055e-06, "loss": 2.4004, "step": 564 }, { "epoch": 0.04, "grad_norm": 7.8089745323809785, "learning_rate": 9.997125481671433e-06, "loss": 0.477, "step": 565 }, { "epoch": 0.04, "grad_norm": 6.44579198128922, "learning_rate": 9.997085973693807e-06, "loss": 0.3884, "step": 566 }, { "epoch": 0.04, "grad_norm": 11.063290778000852, "learning_rate": 9.997046196142309e-06, "loss": 1.0975, "step": 567 }, { "epoch": 0.04, "grad_norm": 5.381849260806226, "learning_rate": 9.997006149019081e-06, "loss": 0.3841, "step": 568 }, { "epoch": 0.04, "grad_norm": 4.9815855479582885, "learning_rate": 9.996965832326287e-06, "loss": 0.4106, "step": 569 }, { "epoch": 0.04, "grad_norm": 15.802798358157482, "learning_rate": 9.996925246066101e-06, "loss": 1.5474, "step": 570 }, { "epoch": 0.04, "grad_norm": 5.994611039347963, "learning_rate": 9.996884390240713e-06, "loss": 0.3045, "step": 571 }, { "epoch": 0.04, "grad_norm": 6.8040897301277505, "learning_rate": 9.996843264852327e-06, "loss": 1.3475, "step": 572 }, { "epoch": 0.04, "grad_norm": 6.222179320051345, "learning_rate": 9.996801869903159e-06, "loss": 0.4893, "step": 573 }, { "epoch": 0.04, "grad_norm": 12.748782256803391, "learning_rate": 9.996760205395445e-06, "loss": 1.5183, "step": 574 }, { "epoch": 0.04, "grad_norm": 7.823063784250277, "learning_rate": 9.99671827133143e-06, "loss": 0.4204, "step": 575 }, { "epoch": 0.04, "grad_norm": 6.525677545904485, "learning_rate": 9.99667606771338e-06, "loss": 0.3685, "step": 576 }, { "epoch": 0.04, "grad_norm": 9.143472797341614, "learning_rate": 9.99663359454357e-06, "loss": 0.4029, "step": 577 }, { "epoch": 0.04, "grad_norm": 5.825200134389126, "learning_rate": 9.996590851824289e-06, "loss": 0.3794, "step": 578 }, { "epoch": 0.04, "grad_norm": 6.762291263062169, "learning_rate": 9.996547839557846e-06, "loss": 0.3947, "step": 579 }, { "epoch": 0.04, "grad_norm": 9.404620291939024, "learning_rate": 9.99650455774656e-06, "loss": 1.0084, "step": 580 }, { "epoch": 0.04, "grad_norm": 10.086737055919395, "learning_rate": 9.996461006392764e-06, "loss": 0.4893, "step": 581 }, { "epoch": 0.04, "grad_norm": 5.978857531192779, "learning_rate": 9.996417185498812e-06, "loss": 0.4656, "step": 582 }, { "epoch": 0.04, "grad_norm": 17.795489525480072, "learning_rate": 9.996373095067064e-06, "loss": 2.0092, "step": 583 }, { "epoch": 0.04, "grad_norm": 6.170711965190347, "learning_rate": 9.9963287350999e-06, "loss": 0.5013, "step": 584 }, { "epoch": 0.04, "grad_norm": 5.3616518882737445, "learning_rate": 9.996284105599712e-06, "loss": 0.3326, "step": 585 }, { "epoch": 0.04, "grad_norm": 8.093846169506016, "learning_rate": 9.996239206568909e-06, "loss": 1.1289, "step": 586 }, { "epoch": 0.04, "grad_norm": 21.227624812815062, "learning_rate": 9.996194038009914e-06, "loss": 2.1037, "step": 587 }, { "epoch": 0.04, "grad_norm": 5.759507908449696, "learning_rate": 9.996148599925161e-06, "loss": 0.3211, "step": 588 }, { "epoch": 0.04, "grad_norm": 50.26645198643791, "learning_rate": 9.996102892317102e-06, "loss": 1.6641, "step": 589 }, { "epoch": 0.04, "grad_norm": 6.02130847920457, "learning_rate": 9.996056915188204e-06, "loss": 0.3365, "step": 590 }, { "epoch": 0.04, "grad_norm": 6.661985511313164, "learning_rate": 9.996010668540946e-06, "loss": 0.4449, "step": 591 }, { "epoch": 0.04, "grad_norm": 7.5310887027002416, "learning_rate": 9.995964152377824e-06, "loss": 1.2483, "step": 592 }, { "epoch": 0.04, "grad_norm": 17.491717752804597, "learning_rate": 9.995917366701346e-06, "loss": 1.8191, "step": 593 }, { "epoch": 0.04, "grad_norm": 17.289208230216303, "learning_rate": 9.995870311514037e-06, "loss": 2.7817, "step": 594 }, { "epoch": 0.04, "grad_norm": 19.149208900568006, "learning_rate": 9.995822986818436e-06, "loss": 1.2349, "step": 595 }, { "epoch": 0.04, "grad_norm": 6.905949179993832, "learning_rate": 9.995775392617095e-06, "loss": 1.2618, "step": 596 }, { "epoch": 0.04, "grad_norm": 5.938346076126658, "learning_rate": 9.99572752891258e-06, "loss": 0.4068, "step": 597 }, { "epoch": 0.04, "grad_norm": 5.383401383786106, "learning_rate": 9.995679395707476e-06, "loss": 0.401, "step": 598 }, { "epoch": 0.04, "grad_norm": 18.58269921361879, "learning_rate": 9.995630993004376e-06, "loss": 1.2841, "step": 599 }, { "epoch": 0.04, "grad_norm": 32.955368657004946, "learning_rate": 9.995582320805896e-06, "loss": 1.3485, "step": 600 }, { "epoch": 0.04, "grad_norm": 17.770792634354464, "learning_rate": 9.995533379114659e-06, "loss": 1.9818, "step": 601 }, { "epoch": 0.04, "grad_norm": 7.426561794143236, "learning_rate": 9.995484167933305e-06, "loss": 1.3788, "step": 602 }, { "epoch": 0.04, "grad_norm": 17.026472894835926, "learning_rate": 9.995434687264489e-06, "loss": 1.4047, "step": 603 }, { "epoch": 0.04, "grad_norm": 5.615192453698795, "learning_rate": 9.99538493711088e-06, "loss": 0.4171, "step": 604 }, { "epoch": 0.04, "grad_norm": 7.123853482483525, "learning_rate": 9.995334917475162e-06, "loss": 1.1241, "step": 605 }, { "epoch": 0.04, "grad_norm": 37.067507887053196, "learning_rate": 9.995284628360033e-06, "loss": 2.2207, "step": 606 }, { "epoch": 0.04, "grad_norm": 8.43137775558408, "learning_rate": 9.995234069768208e-06, "loss": 1.3611, "step": 607 }, { "epoch": 0.04, "grad_norm": 43.05807903643525, "learning_rate": 9.995183241702412e-06, "loss": 1.7412, "step": 608 }, { "epoch": 0.04, "grad_norm": 6.648302923837131, "learning_rate": 9.99513214416539e-06, "loss": 0.398, "step": 609 }, { "epoch": 0.04, "grad_norm": 7.236487614186066, "learning_rate": 9.995080777159894e-06, "loss": 0.4914, "step": 610 }, { "epoch": 0.04, "grad_norm": 22.751609612314653, "learning_rate": 9.995029140688697e-06, "loss": 2.036, "step": 611 }, { "epoch": 0.04, "grad_norm": 17.306641766872204, "learning_rate": 9.994977234754585e-06, "loss": 1.7559, "step": 612 }, { "epoch": 0.04, "grad_norm": 7.522572141188867, "learning_rate": 9.99492505936036e-06, "loss": 1.0862, "step": 613 }, { "epoch": 0.04, "grad_norm": 6.510948528262795, "learning_rate": 9.994872614508835e-06, "loss": 0.4381, "step": 614 }, { "epoch": 0.04, "grad_norm": 5.466282461831215, "learning_rate": 9.994819900202838e-06, "loss": 0.39, "step": 615 }, { "epoch": 0.04, "grad_norm": 6.054720433849098, "learning_rate": 9.994766916445215e-06, "loss": 0.4464, "step": 616 }, { "epoch": 0.04, "grad_norm": 11.383241548149549, "learning_rate": 9.994713663238822e-06, "loss": 1.3896, "step": 617 }, { "epoch": 0.04, "grad_norm": 13.174649717970862, "learning_rate": 9.994660140586532e-06, "loss": 1.4447, "step": 618 }, { "epoch": 0.04, "grad_norm": 19.84470705364458, "learning_rate": 9.994606348491236e-06, "loss": 1.6749, "step": 619 }, { "epoch": 0.04, "grad_norm": 5.876476311717249, "learning_rate": 9.99455228695583e-06, "loss": 0.3961, "step": 620 }, { "epoch": 0.04, "grad_norm": 5.781304178633378, "learning_rate": 9.994497955983235e-06, "loss": 0.2728, "step": 621 }, { "epoch": 0.04, "grad_norm": 17.70231024608206, "learning_rate": 9.994443355576379e-06, "loss": 2.3326, "step": 622 }, { "epoch": 0.04, "grad_norm": 7.046157921996084, "learning_rate": 9.99438848573821e-06, "loss": 1.1473, "step": 623 }, { "epoch": 0.04, "grad_norm": 6.019624315795207, "learning_rate": 9.99433334647169e-06, "loss": 0.3864, "step": 624 }, { "epoch": 0.04, "grad_norm": 6.235280359976735, "learning_rate": 9.994277937779787e-06, "loss": 0.5066, "step": 625 }, { "epoch": 0.04, "grad_norm": 7.3342447776467585, "learning_rate": 9.994222259665494e-06, "loss": 1.2656, "step": 626 }, { "epoch": 0.04, "grad_norm": 5.939091623534494, "learning_rate": 9.994166312131814e-06, "loss": 0.35, "step": 627 }, { "epoch": 0.05, "grad_norm": 15.252148308114348, "learning_rate": 9.994110095181767e-06, "loss": 1.4609, "step": 628 }, { "epoch": 0.05, "grad_norm": 11.664430480188717, "learning_rate": 9.994053608818383e-06, "loss": 1.5376, "step": 629 }, { "epoch": 0.05, "grad_norm": 5.213109153420738, "learning_rate": 9.993996853044711e-06, "loss": 0.3496, "step": 630 }, { "epoch": 0.05, "grad_norm": 5.156974653727794, "learning_rate": 9.993939827863812e-06, "loss": 0.4376, "step": 631 }, { "epoch": 0.05, "grad_norm": 4.954289144296864, "learning_rate": 9.993882533278762e-06, "loss": 0.2661, "step": 632 }, { "epoch": 0.05, "grad_norm": 25.312436276555875, "learning_rate": 9.993824969292651e-06, "loss": 1.9376, "step": 633 }, { "epoch": 0.05, "grad_norm": 15.174606111972908, "learning_rate": 9.993767135908587e-06, "loss": 2.0, "step": 634 }, { "epoch": 0.05, "grad_norm": 6.137964754151268, "learning_rate": 9.99370903312969e-06, "loss": 0.3754, "step": 635 }, { "epoch": 0.05, "grad_norm": 15.333037151270432, "learning_rate": 9.99365066095909e-06, "loss": 1.4809, "step": 636 }, { "epoch": 0.05, "grad_norm": 13.869742223764913, "learning_rate": 9.99359201939994e-06, "loss": 1.2272, "step": 637 }, { "epoch": 0.05, "grad_norm": 5.804536910811676, "learning_rate": 9.993533108455402e-06, "loss": 0.3846, "step": 638 }, { "epoch": 0.05, "grad_norm": 5.192508304759837, "learning_rate": 9.993473928128655e-06, "loss": 0.3389, "step": 639 }, { "epoch": 0.05, "grad_norm": 7.519077750823338, "learning_rate": 9.99341447842289e-06, "loss": 1.2475, "step": 640 }, { "epoch": 0.05, "grad_norm": 235.96810463205387, "learning_rate": 9.993354759341316e-06, "loss": 1.7788, "step": 641 }, { "epoch": 0.05, "grad_norm": 109.64207612677767, "learning_rate": 9.993294770887153e-06, "loss": 1.312, "step": 642 }, { "epoch": 0.05, "grad_norm": 5.445770442585691, "learning_rate": 9.993234513063639e-06, "loss": 0.3756, "step": 643 }, { "epoch": 0.05, "grad_norm": 14.638668491604406, "learning_rate": 9.993173985874021e-06, "loss": 1.6448, "step": 644 }, { "epoch": 0.05, "grad_norm": 37.335145909034686, "learning_rate": 9.993113189321569e-06, "loss": 1.7988, "step": 645 }, { "epoch": 0.05, "grad_norm": 10.441746963647262, "learning_rate": 9.99305212340956e-06, "loss": 1.1802, "step": 646 }, { "epoch": 0.05, "grad_norm": 8.738262872245624, "learning_rate": 9.992990788141288e-06, "loss": 1.1514, "step": 647 }, { "epoch": 0.05, "grad_norm": 20.209303708271317, "learning_rate": 9.992929183520062e-06, "loss": 2.1846, "step": 648 }, { "epoch": 0.05, "grad_norm": 24.56885736432164, "learning_rate": 9.992867309549206e-06, "loss": 1.4152, "step": 649 }, { "epoch": 0.05, "grad_norm": 5.7450801356422065, "learning_rate": 9.992805166232058e-06, "loss": 0.4138, "step": 650 }, { "epoch": 0.05, "grad_norm": 34.16507846959877, "learning_rate": 9.992742753571969e-06, "loss": 1.5226, "step": 651 }, { "epoch": 0.05, "grad_norm": 5.914194247217231, "learning_rate": 9.992680071572309e-06, "loss": 0.4843, "step": 652 }, { "epoch": 0.05, "grad_norm": 8.739983379895683, "learning_rate": 9.992617120236457e-06, "loss": 1.1726, "step": 653 }, { "epoch": 0.05, "grad_norm": 33.89108307171769, "learning_rate": 9.992553899567809e-06, "loss": 1.5948, "step": 654 }, { "epoch": 0.05, "grad_norm": 20.728688180241914, "learning_rate": 9.992490409569775e-06, "loss": 1.7136, "step": 655 }, { "epoch": 0.05, "grad_norm": 7.734934501447232, "learning_rate": 9.992426650245783e-06, "loss": 1.4091, "step": 656 }, { "epoch": 0.05, "grad_norm": 6.090752630763269, "learning_rate": 9.992362621599269e-06, "loss": 0.3191, "step": 657 }, { "epoch": 0.05, "grad_norm": 13.86875718673051, "learning_rate": 9.992298323633688e-06, "loss": 1.317, "step": 658 }, { "epoch": 0.05, "grad_norm": 14.193877599383733, "learning_rate": 9.992233756352511e-06, "loss": 1.4613, "step": 659 }, { "epoch": 0.05, "grad_norm": 4.88344653498568, "learning_rate": 9.99216891975922e-06, "loss": 0.362, "step": 660 }, { "epoch": 0.05, "grad_norm": 25.790998044509237, "learning_rate": 9.99210381385731e-06, "loss": 1.4337, "step": 661 }, { "epoch": 0.05, "grad_norm": 5.102766934772444, "learning_rate": 9.992038438650297e-06, "loss": 0.3416, "step": 662 }, { "epoch": 0.05, "grad_norm": 5.051294813447945, "learning_rate": 9.991972794141706e-06, "loss": 0.3638, "step": 663 }, { "epoch": 0.05, "grad_norm": 5.099975109506273, "learning_rate": 9.991906880335079e-06, "loss": 0.3986, "step": 664 }, { "epoch": 0.05, "grad_norm": 5.17088813911077, "learning_rate": 9.99184069723397e-06, "loss": 0.2989, "step": 665 }, { "epoch": 0.05, "grad_norm": 7.3132352703298915, "learning_rate": 9.99177424484195e-06, "loss": 1.1399, "step": 666 }, { "epoch": 0.05, "grad_norm": 7.434671073133902, "learning_rate": 9.991707523162606e-06, "loss": 1.0018, "step": 667 }, { "epoch": 0.05, "grad_norm": 17.735946034326137, "learning_rate": 9.991640532199534e-06, "loss": 1.6912, "step": 668 }, { "epoch": 0.05, "grad_norm": 5.770527771204879, "learning_rate": 9.99157327195635e-06, "loss": 0.3531, "step": 669 }, { "epoch": 0.05, "grad_norm": 5.680454386057738, "learning_rate": 9.991505742436683e-06, "loss": 0.3559, "step": 670 }, { "epoch": 0.05, "grad_norm": 19.686954773254712, "learning_rate": 9.991437943644175e-06, "loss": 1.4747, "step": 671 }, { "epoch": 0.05, "grad_norm": 26.206435725908094, "learning_rate": 9.991369875582483e-06, "loss": 1.3204, "step": 672 }, { "epoch": 0.05, "grad_norm": 5.293662071299467, "learning_rate": 9.99130153825528e-06, "loss": 0.3433, "step": 673 }, { "epoch": 0.05, "grad_norm": 18.287274032518724, "learning_rate": 9.991232931666253e-06, "loss": 1.8752, "step": 674 }, { "epoch": 0.05, "grad_norm": 5.880992742792822, "learning_rate": 9.9911640558191e-06, "loss": 0.325, "step": 675 }, { "epoch": 0.05, "grad_norm": 7.929288144653811, "learning_rate": 9.99109491071754e-06, "loss": 1.2986, "step": 676 }, { "epoch": 0.05, "grad_norm": 12.562210682834909, "learning_rate": 9.991025496365303e-06, "loss": 0.9443, "step": 677 }, { "epoch": 0.05, "grad_norm": 8.30736220446281, "learning_rate": 9.990955812766133e-06, "loss": 1.3017, "step": 678 }, { "epoch": 0.05, "grad_norm": 6.136882661121462, "learning_rate": 9.990885859923786e-06, "loss": 0.5033, "step": 679 }, { "epoch": 0.05, "grad_norm": 8.715734197472553, "learning_rate": 9.990815637842039e-06, "loss": 1.2645, "step": 680 }, { "epoch": 0.05, "grad_norm": 3.965314998697503, "learning_rate": 9.990745146524681e-06, "loss": 0.2487, "step": 681 }, { "epoch": 0.05, "grad_norm": 17.01651061365127, "learning_rate": 9.990674385975512e-06, "loss": 1.5356, "step": 682 }, { "epoch": 0.05, "grad_norm": 6.128195134230789, "learning_rate": 9.990603356198351e-06, "loss": 1.2871, "step": 683 }, { "epoch": 0.05, "grad_norm": 4.776382081593076, "learning_rate": 9.99053205719703e-06, "loss": 0.3173, "step": 684 }, { "epoch": 0.05, "grad_norm": 19.9924798202342, "learning_rate": 9.990460488975392e-06, "loss": 2.2217, "step": 685 }, { "epoch": 0.05, "grad_norm": 16.75673076904733, "learning_rate": 9.990388651537303e-06, "loss": 2.2024, "step": 686 }, { "epoch": 0.05, "grad_norm": 6.203961942069068, "learning_rate": 9.990316544886636e-06, "loss": 0.5582, "step": 687 }, { "epoch": 0.05, "grad_norm": 18.87738399697242, "learning_rate": 9.990244169027281e-06, "loss": 1.2745, "step": 688 }, { "epoch": 0.05, "grad_norm": 16.934334936690767, "learning_rate": 9.990171523963141e-06, "loss": 1.7115, "step": 689 }, { "epoch": 0.05, "grad_norm": 27.00607477189788, "learning_rate": 9.990098609698136e-06, "loss": 2.4062, "step": 690 }, { "epoch": 0.05, "grad_norm": 7.531075668168414, "learning_rate": 9.9900254262362e-06, "loss": 0.4164, "step": 691 }, { "epoch": 0.05, "grad_norm": 16.903489468294914, "learning_rate": 9.989951973581283e-06, "loss": 2.4335, "step": 692 }, { "epoch": 0.05, "grad_norm": 7.269809527653588, "learning_rate": 9.989878251737342e-06, "loss": 0.9159, "step": 693 }, { "epoch": 0.05, "grad_norm": 4.407487751858617, "learning_rate": 9.989804260708359e-06, "loss": 0.4153, "step": 694 }, { "epoch": 0.05, "grad_norm": 12.609495937054897, "learning_rate": 9.98973000049832e-06, "loss": 1.3797, "step": 695 }, { "epoch": 0.05, "grad_norm": 5.081032907924076, "learning_rate": 9.989655471111237e-06, "loss": 0.3592, "step": 696 }, { "epoch": 0.05, "grad_norm": 5.8816881392091975, "learning_rate": 9.98958067255113e-06, "loss": 0.3676, "step": 697 }, { "epoch": 0.05, "grad_norm": 23.490315615573675, "learning_rate": 9.98950560482203e-06, "loss": 1.9027, "step": 698 }, { "epoch": 0.05, "grad_norm": 24.925656494964198, "learning_rate": 9.989430267927988e-06, "loss": 1.6682, "step": 699 }, { "epoch": 0.05, "grad_norm": 32.3033716842048, "learning_rate": 9.989354661873072e-06, "loss": 1.9612, "step": 700 }, { "epoch": 0.05, "grad_norm": 27.390257687856366, "learning_rate": 9.989278786661356e-06, "loss": 2.2134, "step": 701 }, { "epoch": 0.05, "grad_norm": 5.01855553613523, "learning_rate": 9.989202642296935e-06, "loss": 0.3147, "step": 702 }, { "epoch": 0.05, "grad_norm": 5.660067336670666, "learning_rate": 9.989126228783916e-06, "loss": 0.4209, "step": 703 }, { "epoch": 0.05, "grad_norm": 5.120688017480608, "learning_rate": 9.989049546126422e-06, "loss": 0.3355, "step": 704 }, { "epoch": 0.05, "grad_norm": 8.643504623310877, "learning_rate": 9.988972594328591e-06, "loss": 1.1614, "step": 705 }, { "epoch": 0.05, "grad_norm": 5.693931582900803, "learning_rate": 9.988895373394571e-06, "loss": 0.3913, "step": 706 }, { "epoch": 0.05, "grad_norm": 22.75056595161672, "learning_rate": 9.988817883328532e-06, "loss": 1.8802, "step": 707 }, { "epoch": 0.05, "grad_norm": 7.279412245516647, "learning_rate": 9.988740124134651e-06, "loss": 0.4299, "step": 708 }, { "epoch": 0.05, "grad_norm": 39.029075006610746, "learning_rate": 9.988662095817122e-06, "loss": 1.5709, "step": 709 }, { "epoch": 0.05, "grad_norm": 6.3800006809041925, "learning_rate": 9.988583798380157e-06, "loss": 0.4508, "step": 710 }, { "epoch": 0.05, "grad_norm": 5.687287665315964, "learning_rate": 9.988505231827979e-06, "loss": 0.3402, "step": 711 }, { "epoch": 0.05, "grad_norm": 26.662210625435918, "learning_rate": 9.988426396164826e-06, "loss": 1.949, "step": 712 }, { "epoch": 0.05, "grad_norm": 5.671322090476272, "learning_rate": 9.988347291394952e-06, "loss": 0.3751, "step": 713 }, { "epoch": 0.05, "grad_norm": 5.736327697021255, "learning_rate": 9.988267917522623e-06, "loss": 0.3151, "step": 714 }, { "epoch": 0.05, "grad_norm": 12.694638083306522, "learning_rate": 9.988188274552122e-06, "loss": 1.9137, "step": 715 }, { "epoch": 0.05, "grad_norm": 6.5731476496455254, "learning_rate": 9.988108362487743e-06, "loss": 0.4494, "step": 716 }, { "epoch": 0.05, "grad_norm": 4.572634623442389, "learning_rate": 9.9880281813338e-06, "loss": 0.3038, "step": 717 }, { "epoch": 0.05, "grad_norm": 17.555059693659235, "learning_rate": 9.987947731094618e-06, "loss": 1.7709, "step": 718 }, { "epoch": 0.05, "grad_norm": 11.870597777814927, "learning_rate": 9.987867011774537e-06, "loss": 1.6958, "step": 719 }, { "epoch": 0.05, "grad_norm": 9.845349976356928, "learning_rate": 9.987786023377909e-06, "loss": 1.2455, "step": 720 }, { "epoch": 0.05, "grad_norm": 5.815714593487619, "learning_rate": 9.987704765909106e-06, "loss": 0.3197, "step": 721 }, { "epoch": 0.05, "grad_norm": 29.58862824393832, "learning_rate": 9.98762323937251e-06, "loss": 1.385, "step": 722 }, { "epoch": 0.05, "grad_norm": 13.492823834779802, "learning_rate": 9.987541443772519e-06, "loss": 1.5471, "step": 723 }, { "epoch": 0.05, "grad_norm": 5.678217159809218, "learning_rate": 9.987459379113546e-06, "loss": 0.2999, "step": 724 }, { "epoch": 0.05, "grad_norm": 6.960585918393814, "learning_rate": 9.987377045400018e-06, "loss": 0.3114, "step": 725 }, { "epoch": 0.05, "grad_norm": 17.53339602385623, "learning_rate": 9.987294442636378e-06, "loss": 2.1826, "step": 726 }, { "epoch": 0.05, "grad_norm": 16.780477673209052, "learning_rate": 9.98721157082708e-06, "loss": 2.6162, "step": 727 }, { "epoch": 0.05, "grad_norm": 7.38306219100748, "learning_rate": 9.987128429976595e-06, "loss": 0.4617, "step": 728 }, { "epoch": 0.05, "grad_norm": 62.5502690907789, "learning_rate": 9.987045020089408e-06, "loss": 1.5225, "step": 729 }, { "epoch": 0.05, "grad_norm": 9.708140150912904, "learning_rate": 9.98696134117002e-06, "loss": 1.6987, "step": 730 }, { "epoch": 0.05, "grad_norm": 5.810306869407679, "learning_rate": 9.986877393222945e-06, "loss": 0.4935, "step": 731 }, { "epoch": 0.05, "grad_norm": 22.13583643311101, "learning_rate": 9.98679317625271e-06, "loss": 2.3969, "step": 732 }, { "epoch": 0.05, "grad_norm": 14.643674023235029, "learning_rate": 9.986708690263858e-06, "loss": 1.793, "step": 733 }, { "epoch": 0.05, "grad_norm": 18.889672962030186, "learning_rate": 9.98662393526095e-06, "loss": 1.8608, "step": 734 }, { "epoch": 0.05, "grad_norm": 6.274324131589975, "learning_rate": 9.986538911248555e-06, "loss": 0.4715, "step": 735 }, { "epoch": 0.05, "grad_norm": 5.926227425473273, "learning_rate": 9.98645361823126e-06, "loss": 0.3791, "step": 736 }, { "epoch": 0.05, "grad_norm": 6.166851901249635, "learning_rate": 9.986368056213669e-06, "loss": 0.4283, "step": 737 }, { "epoch": 0.05, "grad_norm": 5.335655743912343, "learning_rate": 9.986282225200394e-06, "loss": 0.3445, "step": 738 }, { "epoch": 0.05, "grad_norm": 34.80461472130051, "learning_rate": 9.986196125196067e-06, "loss": 1.8123, "step": 739 }, { "epoch": 0.05, "grad_norm": 8.77620283450862, "learning_rate": 9.986109756205335e-06, "loss": 0.9373, "step": 740 }, { "epoch": 0.05, "grad_norm": 6.76750085016676, "learning_rate": 9.986023118232855e-06, "loss": 0.3898, "step": 741 }, { "epoch": 0.05, "grad_norm": 4.445999742919224, "learning_rate": 9.9859362112833e-06, "loss": 0.3971, "step": 742 }, { "epoch": 0.05, "grad_norm": 5.298562083842315, "learning_rate": 9.98584903536136e-06, "loss": 0.5216, "step": 743 }, { "epoch": 0.05, "grad_norm": 18.353761603252337, "learning_rate": 9.985761590471735e-06, "loss": 1.515, "step": 744 }, { "epoch": 0.05, "grad_norm": 7.939506782817247, "learning_rate": 9.985673876619143e-06, "loss": 1.0848, "step": 745 }, { "epoch": 0.05, "grad_norm": 7.966292854991644, "learning_rate": 9.985585893808321e-06, "loss": 1.1222, "step": 746 }, { "epoch": 0.05, "grad_norm": 6.983006284780603, "learning_rate": 9.98549764204401e-06, "loss": 1.2186, "step": 747 }, { "epoch": 0.05, "grad_norm": 7.631503404909291, "learning_rate": 9.985409121330972e-06, "loss": 1.3331, "step": 748 }, { "epoch": 0.05, "grad_norm": 16.377567037326912, "learning_rate": 9.985320331673983e-06, "loss": 1.7046, "step": 749 }, { "epoch": 0.05, "grad_norm": 16.901265489222457, "learning_rate": 9.985231273077834e-06, "loss": 1.647, "step": 750 }, { "epoch": 0.05, "grad_norm": 6.080467476491907, "learning_rate": 9.985141945547325e-06, "loss": 0.3114, "step": 751 }, { "epoch": 0.05, "grad_norm": 7.601558619023492, "learning_rate": 9.98505234908728e-06, "loss": 1.025, "step": 752 }, { "epoch": 0.05, "grad_norm": 5.700331096710796, "learning_rate": 9.98496248370253e-06, "loss": 0.3667, "step": 753 }, { "epoch": 0.05, "grad_norm": 5.623434059590175, "learning_rate": 9.984872349397923e-06, "loss": 0.3666, "step": 754 }, { "epoch": 0.05, "grad_norm": 5.41868488903656, "learning_rate": 9.98478194617832e-06, "loss": 0.3669, "step": 755 }, { "epoch": 0.05, "grad_norm": 6.300208595968777, "learning_rate": 9.9846912740486e-06, "loss": 0.394, "step": 756 }, { "epoch": 0.05, "grad_norm": 5.862715697431143, "learning_rate": 9.984600333013655e-06, "loss": 0.4699, "step": 757 }, { "epoch": 0.05, "grad_norm": 17.459590321402207, "learning_rate": 9.98450912307839e-06, "loss": 2.0923, "step": 758 }, { "epoch": 0.05, "grad_norm": 15.785599447373983, "learning_rate": 9.984417644247724e-06, "loss": 1.2459, "step": 759 }, { "epoch": 0.05, "grad_norm": 14.791671140280906, "learning_rate": 9.984325896526594e-06, "loss": 1.5767, "step": 760 }, { "epoch": 0.05, "grad_norm": 6.025631938432606, "learning_rate": 9.984233879919947e-06, "loss": 0.3058, "step": 761 }, { "epoch": 0.05, "grad_norm": 15.16166496276692, "learning_rate": 9.98414159443275e-06, "loss": 1.671, "step": 762 }, { "epoch": 0.05, "grad_norm": 5.691280658772873, "learning_rate": 9.984049040069981e-06, "loss": 0.3394, "step": 763 }, { "epoch": 0.05, "grad_norm": 18.371954783539437, "learning_rate": 9.98395621683663e-06, "loss": 1.3425, "step": 764 }, { "epoch": 0.05, "grad_norm": 5.9259873623617425, "learning_rate": 9.983863124737707e-06, "loss": 0.3418, "step": 765 }, { "epoch": 0.05, "grad_norm": 7.095976847463342, "learning_rate": 9.983769763778232e-06, "loss": 0.3994, "step": 766 }, { "epoch": 0.05, "grad_norm": 5.893261830370111, "learning_rate": 9.983676133963245e-06, "loss": 0.3363, "step": 767 }, { "epoch": 0.06, "grad_norm": 6.278992300826001, "learning_rate": 9.983582235297794e-06, "loss": 0.4693, "step": 768 }, { "epoch": 0.06, "grad_norm": 17.24294701020808, "learning_rate": 9.983488067786946e-06, "loss": 2.0157, "step": 769 }, { "epoch": 0.06, "grad_norm": 5.75295390084294, "learning_rate": 9.983393631435778e-06, "loss": 0.3185, "step": 770 }, { "epoch": 0.06, "grad_norm": 19.39412075895508, "learning_rate": 9.983298926249389e-06, "loss": 2.0945, "step": 771 }, { "epoch": 0.06, "grad_norm": 13.280196715391478, "learning_rate": 9.983203952232885e-06, "loss": 1.6199, "step": 772 }, { "epoch": 0.06, "grad_norm": 8.79737560869856, "learning_rate": 9.983108709391392e-06, "loss": 1.2601, "step": 773 }, { "epoch": 0.06, "grad_norm": 20.53231274531508, "learning_rate": 9.983013197730043e-06, "loss": 1.9176, "step": 774 }, { "epoch": 0.06, "grad_norm": 9.831178573734556, "learning_rate": 9.982917417253995e-06, "loss": 1.3363, "step": 775 }, { "epoch": 0.06, "grad_norm": 15.835031836954624, "learning_rate": 9.982821367968413e-06, "loss": 1.3892, "step": 776 }, { "epoch": 0.06, "grad_norm": 28.37370259009455, "learning_rate": 9.98272504987848e-06, "loss": 1.991, "step": 777 }, { "epoch": 0.06, "grad_norm": 7.026242633059297, "learning_rate": 9.982628462989393e-06, "loss": 0.4553, "step": 778 }, { "epoch": 0.06, "grad_norm": 14.083218566934503, "learning_rate": 9.982531607306358e-06, "loss": 1.2833, "step": 779 }, { "epoch": 0.06, "grad_norm": 5.256741111637278, "learning_rate": 9.982434482834603e-06, "loss": 0.4264, "step": 780 }, { "epoch": 0.06, "grad_norm": 5.072423413693832, "learning_rate": 9.982337089579369e-06, "loss": 0.3712, "step": 781 }, { "epoch": 0.06, "grad_norm": 4.90993157369915, "learning_rate": 9.982239427545908e-06, "loss": 0.4642, "step": 782 }, { "epoch": 0.06, "grad_norm": 6.787160893496931, "learning_rate": 9.982141496739486e-06, "loss": 1.3785, "step": 783 }, { "epoch": 0.06, "grad_norm": 7.614410117029195, "learning_rate": 9.982043297165392e-06, "loss": 1.1912, "step": 784 }, { "epoch": 0.06, "grad_norm": 5.567111007556725, "learning_rate": 9.981944828828917e-06, "loss": 0.3589, "step": 785 }, { "epoch": 0.06, "grad_norm": 4.802227006845754, "learning_rate": 9.981846091735378e-06, "loss": 0.3117, "step": 786 }, { "epoch": 0.06, "grad_norm": 5.496183048037362, "learning_rate": 9.981747085890101e-06, "loss": 0.3595, "step": 787 }, { "epoch": 0.06, "grad_norm": 17.150391452975843, "learning_rate": 9.981647811298424e-06, "loss": 2.2228, "step": 788 }, { "epoch": 0.06, "grad_norm": 5.67254974596712, "learning_rate": 9.981548267965704e-06, "loss": 0.3518, "step": 789 }, { "epoch": 0.06, "grad_norm": 5.950727609959436, "learning_rate": 9.981448455897312e-06, "loss": 0.3217, "step": 790 }, { "epoch": 0.06, "grad_norm": 5.717014782610819, "learning_rate": 9.981348375098629e-06, "loss": 0.3331, "step": 791 }, { "epoch": 0.06, "grad_norm": 5.303249389146543, "learning_rate": 9.98124802557506e-06, "loss": 0.2781, "step": 792 }, { "epoch": 0.06, "grad_norm": 19.75310004674832, "learning_rate": 9.981147407332014e-06, "loss": 1.8468, "step": 793 }, { "epoch": 0.06, "grad_norm": 6.240907112616244, "learning_rate": 9.981046520374919e-06, "loss": 0.3678, "step": 794 }, { "epoch": 0.06, "grad_norm": 6.163612779324094, "learning_rate": 9.980945364709218e-06, "loss": 0.3974, "step": 795 }, { "epoch": 0.06, "grad_norm": 6.100546699973333, "learning_rate": 9.98084394034037e-06, "loss": 0.2776, "step": 796 }, { "epoch": 0.06, "grad_norm": 12.578474787977598, "learning_rate": 9.980742247273846e-06, "loss": 1.1661, "step": 797 }, { "epoch": 0.06, "grad_norm": 6.8367873792503815, "learning_rate": 9.980640285515128e-06, "loss": 1.2031, "step": 798 }, { "epoch": 0.06, "grad_norm": 5.056223390321942, "learning_rate": 9.980538055069721e-06, "loss": 0.2996, "step": 799 }, { "epoch": 0.06, "grad_norm": 4.82781616597233, "learning_rate": 9.98043555594314e-06, "loss": 0.3086, "step": 800 }, { "epoch": 0.06, "grad_norm": 7.178372190233899, "learning_rate": 9.980332788140909e-06, "loss": 1.2207, "step": 801 }, { "epoch": 0.06, "grad_norm": 6.461369169374659, "learning_rate": 9.980229751668578e-06, "loss": 1.2111, "step": 802 }, { "epoch": 0.06, "grad_norm": 4.809861562753653, "learning_rate": 9.980126446531705e-06, "loss": 0.3355, "step": 803 }, { "epoch": 0.06, "grad_norm": 15.749484923850744, "learning_rate": 9.980022872735857e-06, "loss": 1.1191, "step": 804 }, { "epoch": 0.06, "grad_norm": 6.384917325848477, "learning_rate": 9.979919030286626e-06, "loss": 1.072, "step": 805 }, { "epoch": 0.06, "grad_norm": 5.799450789212387, "learning_rate": 9.979814919189616e-06, "loss": 0.4823, "step": 806 }, { "epoch": 0.06, "grad_norm": 28.705215539534628, "learning_rate": 9.979710539450438e-06, "loss": 0.7896, "step": 807 }, { "epoch": 0.06, "grad_norm": 4.5562898891743915, "learning_rate": 9.979605891074727e-06, "loss": 0.4015, "step": 808 }, { "epoch": 0.06, "grad_norm": 22.7386533665346, "learning_rate": 9.979500974068126e-06, "loss": 2.1357, "step": 809 }, { "epoch": 0.06, "grad_norm": 6.257188530142576, "learning_rate": 9.979395788436297e-06, "loss": 0.3941, "step": 810 }, { "epoch": 0.06, "grad_norm": 5.739537388738633, "learning_rate": 9.979290334184913e-06, "loss": 0.3919, "step": 811 }, { "epoch": 0.06, "grad_norm": 5.449095324167142, "learning_rate": 9.979184611319664e-06, "loss": 0.3105, "step": 812 }, { "epoch": 0.06, "grad_norm": 7.294430447049135, "learning_rate": 9.979078619846251e-06, "loss": 1.0968, "step": 813 }, { "epoch": 0.06, "grad_norm": 5.627320272062385, "learning_rate": 9.978972359770394e-06, "loss": 0.3419, "step": 814 }, { "epoch": 0.06, "grad_norm": 6.577753683374242, "learning_rate": 9.978865831097824e-06, "loss": 0.271, "step": 815 }, { "epoch": 0.06, "grad_norm": 5.476958273268101, "learning_rate": 9.978759033834289e-06, "loss": 0.3117, "step": 816 }, { "epoch": 0.06, "grad_norm": 5.519669586848756, "learning_rate": 9.978651967985548e-06, "loss": 0.2614, "step": 817 }, { "epoch": 0.06, "grad_norm": 6.057683352512534, "learning_rate": 9.978544633557381e-06, "loss": 0.4043, "step": 818 }, { "epoch": 0.06, "grad_norm": 25.790003512686205, "learning_rate": 9.978437030555576e-06, "loss": 2.2561, "step": 819 }, { "epoch": 0.06, "grad_norm": 7.599720432282954, "learning_rate": 9.978329158985936e-06, "loss": 0.3807, "step": 820 }, { "epoch": 0.06, "grad_norm": 6.352841605969592, "learning_rate": 9.978221018854282e-06, "loss": 0.3264, "step": 821 }, { "epoch": 0.06, "grad_norm": 40.973374478463775, "learning_rate": 9.978112610166448e-06, "loss": 1.7838, "step": 822 }, { "epoch": 0.06, "grad_norm": 6.557598033221641, "learning_rate": 9.978003932928281e-06, "loss": 0.4332, "step": 823 }, { "epoch": 0.06, "grad_norm": 4.960222394779172, "learning_rate": 9.977894987145646e-06, "loss": 0.2747, "step": 824 }, { "epoch": 0.06, "grad_norm": 6.388961602919242, "learning_rate": 9.97778577282442e-06, "loss": 0.3767, "step": 825 }, { "epoch": 0.06, "grad_norm": 21.071876991128377, "learning_rate": 9.97767628997049e-06, "loss": 1.4724, "step": 826 }, { "epoch": 0.06, "grad_norm": 6.879910072265154, "learning_rate": 9.97756653858977e-06, "loss": 0.4383, "step": 827 }, { "epoch": 0.06, "grad_norm": 7.2339079062905896, "learning_rate": 9.977456518688173e-06, "loss": 1.1849, "step": 828 }, { "epoch": 0.06, "grad_norm": 6.544048328689085, "learning_rate": 9.977346230271639e-06, "loss": 0.3682, "step": 829 }, { "epoch": 0.06, "grad_norm": 6.792861094184693, "learning_rate": 9.977235673346115e-06, "loss": 1.1679, "step": 830 }, { "epoch": 0.06, "grad_norm": 12.444522857628685, "learning_rate": 9.977124847917565e-06, "loss": 1.4763, "step": 831 }, { "epoch": 0.06, "grad_norm": 19.926014576287002, "learning_rate": 9.97701375399197e-06, "loss": 1.1312, "step": 832 }, { "epoch": 0.06, "grad_norm": 5.758445555104973, "learning_rate": 9.976902391575325e-06, "loss": 0.4318, "step": 833 }, { "epoch": 0.06, "grad_norm": 26.450332509316276, "learning_rate": 9.976790760673632e-06, "loss": 1.8424, "step": 834 }, { "epoch": 0.06, "grad_norm": 6.434731381872068, "learning_rate": 9.976678861292915e-06, "loss": 1.0542, "step": 835 }, { "epoch": 0.06, "grad_norm": 7.20936964885956, "learning_rate": 9.976566693439212e-06, "loss": 0.4311, "step": 836 }, { "epoch": 0.06, "grad_norm": 6.166706361884313, "learning_rate": 9.976454257118573e-06, "loss": 0.3804, "step": 837 }, { "epoch": 0.06, "grad_norm": 6.120966510286055, "learning_rate": 9.976341552337066e-06, "loss": 0.3668, "step": 838 }, { "epoch": 0.06, "grad_norm": 12.15795230932173, "learning_rate": 9.976228579100765e-06, "loss": 1.7278, "step": 839 }, { "epoch": 0.06, "grad_norm": 5.617353970210674, "learning_rate": 9.97611533741577e-06, "loss": 0.3858, "step": 840 }, { "epoch": 0.06, "grad_norm": 55.23069508268783, "learning_rate": 9.97600182728819e-06, "loss": 2.1087, "step": 841 }, { "epoch": 0.06, "grad_norm": 5.907793839311431, "learning_rate": 9.975888048724143e-06, "loss": 0.4466, "step": 842 }, { "epoch": 0.06, "grad_norm": 9.942932494903141, "learning_rate": 9.975774001729773e-06, "loss": 1.169, "step": 843 }, { "epoch": 0.06, "grad_norm": 5.8371451102578105, "learning_rate": 9.975659686311228e-06, "loss": 0.3549, "step": 844 }, { "epoch": 0.06, "grad_norm": 19.478083499560064, "learning_rate": 9.975545102474679e-06, "loss": 1.5589, "step": 845 }, { "epoch": 0.06, "grad_norm": 5.624617406164848, "learning_rate": 9.975430250226304e-06, "loss": 0.2977, "step": 846 }, { "epoch": 0.06, "grad_norm": 12.631878626847246, "learning_rate": 9.9753151295723e-06, "loss": 1.4055, "step": 847 }, { "epoch": 0.06, "grad_norm": 4.985199516505867, "learning_rate": 9.975199740518877e-06, "loss": 0.3323, "step": 848 }, { "epoch": 0.06, "grad_norm": 11.44996401680669, "learning_rate": 9.975084083072262e-06, "loss": 1.7471, "step": 849 }, { "epoch": 0.06, "grad_norm": 6.700852204636575, "learning_rate": 9.974968157238691e-06, "loss": 1.265, "step": 850 }, { "epoch": 0.06, "grad_norm": 15.509341702625987, "learning_rate": 9.974851963024419e-06, "loss": 1.8355, "step": 851 }, { "epoch": 0.06, "grad_norm": 6.355712957500305, "learning_rate": 9.974735500435715e-06, "loss": 0.4392, "step": 852 }, { "epoch": 0.06, "grad_norm": 17.703167647391655, "learning_rate": 9.974618769478862e-06, "loss": 1.2286, "step": 853 }, { "epoch": 0.06, "grad_norm": 4.493910716473474, "learning_rate": 9.974501770160154e-06, "loss": 0.2297, "step": 854 }, { "epoch": 0.06, "grad_norm": 6.301535701423258, "learning_rate": 9.974384502485906e-06, "loss": 0.3666, "step": 855 }, { "epoch": 0.06, "grad_norm": 5.9072008242049785, "learning_rate": 9.974266966462444e-06, "loss": 0.3282, "step": 856 }, { "epoch": 0.06, "grad_norm": 30.138332097702648, "learning_rate": 9.974149162096108e-06, "loss": 1.2407, "step": 857 }, { "epoch": 0.06, "grad_norm": 7.13538702325037, "learning_rate": 9.974031089393251e-06, "loss": 1.1132, "step": 858 }, { "epoch": 0.06, "grad_norm": 13.39480230057385, "learning_rate": 9.973912748360245e-06, "loss": 1.2384, "step": 859 }, { "epoch": 0.06, "grad_norm": 6.418631552218406, "learning_rate": 9.973794139003475e-06, "loss": 0.3869, "step": 860 }, { "epoch": 0.06, "grad_norm": 7.288715525943443, "learning_rate": 9.973675261329336e-06, "loss": 1.4376, "step": 861 }, { "epoch": 0.06, "grad_norm": 6.02477583479047, "learning_rate": 9.973556115344246e-06, "loss": 0.3436, "step": 862 }, { "epoch": 0.06, "grad_norm": 9.833443891767491, "learning_rate": 9.973436701054627e-06, "loss": 1.7609, "step": 863 }, { "epoch": 0.06, "grad_norm": 5.60357778777062, "learning_rate": 9.973317018466924e-06, "loss": 0.3881, "step": 864 }, { "epoch": 0.06, "grad_norm": 29.930397131936218, "learning_rate": 9.973197067587593e-06, "loss": 2.4631, "step": 865 }, { "epoch": 0.06, "grad_norm": 7.20683770914926, "learning_rate": 9.973076848423105e-06, "loss": 1.1058, "step": 866 }, { "epoch": 0.06, "grad_norm": 33.38546941793342, "learning_rate": 9.972956360979944e-06, "loss": 2.0342, "step": 867 }, { "epoch": 0.06, "grad_norm": 25.706813034788812, "learning_rate": 9.972835605264612e-06, "loss": 1.6913, "step": 868 }, { "epoch": 0.06, "grad_norm": 5.764026949312878, "learning_rate": 9.972714581283623e-06, "loss": 0.3295, "step": 869 }, { "epoch": 0.06, "grad_norm": 5.650118757982493, "learning_rate": 9.972593289043504e-06, "loss": 0.4256, "step": 870 }, { "epoch": 0.06, "grad_norm": 5.9176048498236495, "learning_rate": 9.972471728550799e-06, "loss": 0.3888, "step": 871 }, { "epoch": 0.06, "grad_norm": 26.951711896185838, "learning_rate": 9.972349899812067e-06, "loss": 1.6469, "step": 872 }, { "epoch": 0.06, "grad_norm": 11.855227291190483, "learning_rate": 9.97222780283388e-06, "loss": 1.5638, "step": 873 }, { "epoch": 0.06, "grad_norm": 4.970473961956003, "learning_rate": 9.972105437622822e-06, "loss": 0.3276, "step": 874 }, { "epoch": 0.06, "grad_norm": 4.72113021699222, "learning_rate": 9.971982804185497e-06, "loss": 0.3458, "step": 875 }, { "epoch": 0.06, "grad_norm": 4.902484479696756, "learning_rate": 9.97185990252852e-06, "loss": 0.3725, "step": 876 }, { "epoch": 0.06, "grad_norm": 8.29920636949284, "learning_rate": 9.97173673265852e-06, "loss": 1.3157, "step": 877 }, { "epoch": 0.06, "grad_norm": 6.261444949953518, "learning_rate": 9.971613294582143e-06, "loss": 1.1674, "step": 878 }, { "epoch": 0.06, "grad_norm": 4.801842982918444, "learning_rate": 9.971489588306047e-06, "loss": 0.3359, "step": 879 }, { "epoch": 0.06, "grad_norm": 13.360789566978315, "learning_rate": 9.971365613836905e-06, "loss": 1.4698, "step": 880 }, { "epoch": 0.06, "grad_norm": 20.5121170351854, "learning_rate": 9.971241371181409e-06, "loss": 1.692, "step": 881 }, { "epoch": 0.06, "grad_norm": 6.823476554237536, "learning_rate": 9.971116860346254e-06, "loss": 1.294, "step": 882 }, { "epoch": 0.06, "grad_norm": 5.855121018138487, "learning_rate": 9.970992081338163e-06, "loss": 0.3272, "step": 883 }, { "epoch": 0.06, "grad_norm": 34.58197456916178, "learning_rate": 9.970867034163865e-06, "loss": 1.8668, "step": 884 }, { "epoch": 0.06, "grad_norm": 6.29342863765811, "learning_rate": 9.970741718830108e-06, "loss": 0.4667, "step": 885 }, { "epoch": 0.06, "grad_norm": 34.3987992866788, "learning_rate": 9.970616135343649e-06, "loss": 1.4978, "step": 886 }, { "epoch": 0.06, "grad_norm": 5.64189817050778, "learning_rate": 9.970490283711265e-06, "loss": 0.4138, "step": 887 }, { "epoch": 0.06, "grad_norm": 4.332712595443798, "learning_rate": 9.970364163939742e-06, "loss": 0.3293, "step": 888 }, { "epoch": 0.06, "grad_norm": 7.487097753395645, "learning_rate": 9.970237776035888e-06, "loss": 1.3783, "step": 889 }, { "epoch": 0.06, "grad_norm": 7.378199176115007, "learning_rate": 9.97011112000652e-06, "loss": 1.1545, "step": 890 }, { "epoch": 0.06, "grad_norm": 5.491905297123394, "learning_rate": 9.96998419585847e-06, "loss": 0.5568, "step": 891 }, { "epoch": 0.06, "grad_norm": 5.513461171738478, "learning_rate": 9.969857003598586e-06, "loss": 0.2594, "step": 892 }, { "epoch": 0.06, "grad_norm": 6.363909675416937, "learning_rate": 9.969729543233727e-06, "loss": 0.3068, "step": 893 }, { "epoch": 0.06, "grad_norm": 13.115869437698292, "learning_rate": 9.969601814770771e-06, "loss": 1.1539, "step": 894 }, { "epoch": 0.06, "grad_norm": 6.962411225447852, "learning_rate": 9.969473818216608e-06, "loss": 1.2432, "step": 895 }, { "epoch": 0.06, "grad_norm": 5.05029870335266, "learning_rate": 9.969345553578143e-06, "loss": 0.2867, "step": 896 }, { "epoch": 0.06, "grad_norm": 5.535025038669533, "learning_rate": 9.969217020862294e-06, "loss": 0.4313, "step": 897 }, { "epoch": 0.06, "grad_norm": 5.27786123986962, "learning_rate": 9.969088220075998e-06, "loss": 0.4229, "step": 898 }, { "epoch": 0.06, "grad_norm": 20.19853511243369, "learning_rate": 9.968959151226201e-06, "loss": 1.5716, "step": 899 }, { "epoch": 0.06, "grad_norm": 5.199446677933602, "learning_rate": 9.968829814319865e-06, "loss": 0.2875, "step": 900 }, { "epoch": 0.06, "grad_norm": 7.187228606265145, "learning_rate": 9.968700209363971e-06, "loss": 0.5012, "step": 901 }, { "epoch": 0.06, "grad_norm": 19.276932183887588, "learning_rate": 9.968570336365507e-06, "loss": 1.6692, "step": 902 }, { "epoch": 0.06, "grad_norm": 5.908030685835896, "learning_rate": 9.968440195331482e-06, "loss": 0.4427, "step": 903 }, { "epoch": 0.06, "grad_norm": 25.750629557320035, "learning_rate": 9.968309786268913e-06, "loss": 1.5736, "step": 904 }, { "epoch": 0.06, "grad_norm": 5.726750698708755, "learning_rate": 9.968179109184837e-06, "loss": 0.3636, "step": 905 }, { "epoch": 0.06, "grad_norm": 5.054806818471459, "learning_rate": 9.968048164086304e-06, "loss": 0.2865, "step": 906 }, { "epoch": 0.07, "grad_norm": 8.249624307530075, "learning_rate": 9.967916950980379e-06, "loss": 1.0693, "step": 907 }, { "epoch": 0.07, "grad_norm": 5.273146849550737, "learning_rate": 9.967785469874137e-06, "loss": 0.4127, "step": 908 }, { "epoch": 0.07, "grad_norm": 13.05947684551704, "learning_rate": 9.967653720774674e-06, "loss": 1.2374, "step": 909 }, { "epoch": 0.07, "grad_norm": 5.028416893263092, "learning_rate": 9.967521703689096e-06, "loss": 0.2787, "step": 910 }, { "epoch": 0.07, "grad_norm": 10.860667734201327, "learning_rate": 9.967389418624527e-06, "loss": 1.2968, "step": 911 }, { "epoch": 0.07, "grad_norm": 6.321519219018819, "learning_rate": 9.9672568655881e-06, "loss": 1.238, "step": 912 }, { "epoch": 0.07, "grad_norm": 8.705972303661627, "learning_rate": 9.967124044586966e-06, "loss": 1.2052, "step": 913 }, { "epoch": 0.07, "grad_norm": 68.5188688483464, "learning_rate": 9.966990955628293e-06, "loss": 1.6506, "step": 914 }, { "epoch": 0.07, "grad_norm": 7.0437967842740195, "learning_rate": 9.966857598719259e-06, "loss": 1.5203, "step": 915 }, { "epoch": 0.07, "grad_norm": 18.964284893985745, "learning_rate": 9.966723973867058e-06, "loss": 0.993, "step": 916 }, { "epoch": 0.07, "grad_norm": 5.221152063603391, "learning_rate": 9.966590081078898e-06, "loss": 0.309, "step": 917 }, { "epoch": 0.07, "grad_norm": 12.947690490653901, "learning_rate": 9.966455920362003e-06, "loss": 1.3987, "step": 918 }, { "epoch": 0.07, "grad_norm": 5.638790936128257, "learning_rate": 9.966321491723609e-06, "loss": 0.3465, "step": 919 }, { "epoch": 0.07, "grad_norm": 4.921238740038057, "learning_rate": 9.966186795170971e-06, "loss": 0.3635, "step": 920 }, { "epoch": 0.07, "grad_norm": 12.056286334720255, "learning_rate": 9.966051830711353e-06, "loss": 0.9218, "step": 921 }, { "epoch": 0.07, "grad_norm": 4.540509014089636, "learning_rate": 9.965916598352037e-06, "loss": 0.3094, "step": 922 }, { "epoch": 0.07, "grad_norm": 5.172574748253067, "learning_rate": 9.965781098100315e-06, "loss": 0.4505, "step": 923 }, { "epoch": 0.07, "grad_norm": 40.42357117131758, "learning_rate": 9.965645329963502e-06, "loss": 1.2526, "step": 924 }, { "epoch": 0.07, "grad_norm": 5.31549831347472, "learning_rate": 9.965509293948918e-06, "loss": 0.3589, "step": 925 }, { "epoch": 0.07, "grad_norm": 16.77939184401447, "learning_rate": 9.965372990063903e-06, "loss": 1.4749, "step": 926 }, { "epoch": 0.07, "grad_norm": 13.565753303724183, "learning_rate": 9.96523641831581e-06, "loss": 1.4499, "step": 927 }, { "epoch": 0.07, "grad_norm": 14.137153955092034, "learning_rate": 9.965099578712008e-06, "loss": 0.9267, "step": 928 }, { "epoch": 0.07, "grad_norm": 5.756812166594939, "learning_rate": 9.964962471259875e-06, "loss": 0.4062, "step": 929 }, { "epoch": 0.07, "grad_norm": 5.497810584322844, "learning_rate": 9.964825095966814e-06, "loss": 0.4523, "step": 930 }, { "epoch": 0.07, "grad_norm": 8.119038352322315, "learning_rate": 9.96468745284023e-06, "loss": 1.2851, "step": 931 }, { "epoch": 0.07, "grad_norm": 19.635559874002425, "learning_rate": 9.96454954188755e-06, "loss": 1.7642, "step": 932 }, { "epoch": 0.07, "grad_norm": 8.503843916688236, "learning_rate": 9.964411363116213e-06, "loss": 1.1896, "step": 933 }, { "epoch": 0.07, "grad_norm": 5.475199368245788, "learning_rate": 9.964272916533677e-06, "loss": 0.4363, "step": 934 }, { "epoch": 0.07, "grad_norm": 7.885747406364, "learning_rate": 9.964134202147406e-06, "loss": 1.1669, "step": 935 }, { "epoch": 0.07, "grad_norm": 19.680293267742602, "learning_rate": 9.963995219964884e-06, "loss": 1.6956, "step": 936 }, { "epoch": 0.07, "grad_norm": 6.185469271588979, "learning_rate": 9.963855969993611e-06, "loss": 0.323, "step": 937 }, { "epoch": 0.07, "grad_norm": 17.545311813027002, "learning_rate": 9.963716452241099e-06, "loss": 1.1386, "step": 938 }, { "epoch": 0.07, "grad_norm": 5.316631385727141, "learning_rate": 9.963576666714871e-06, "loss": 0.4429, "step": 939 }, { "epoch": 0.07, "grad_norm": 6.8498513260139715, "learning_rate": 9.96343661342247e-06, "loss": 1.1093, "step": 940 }, { "epoch": 0.07, "grad_norm": 5.637894169952689, "learning_rate": 9.963296292371454e-06, "loss": 0.4928, "step": 941 }, { "epoch": 0.07, "grad_norm": 4.436449780132322, "learning_rate": 9.963155703569388e-06, "loss": 0.364, "step": 942 }, { "epoch": 0.07, "grad_norm": 7.201030147438301, "learning_rate": 9.963014847023858e-06, "loss": 0.3647, "step": 943 }, { "epoch": 0.07, "grad_norm": 23.365208041452593, "learning_rate": 9.962873722742463e-06, "loss": 1.5272, "step": 944 }, { "epoch": 0.07, "grad_norm": 7.489850994600788, "learning_rate": 9.962732330732818e-06, "loss": 0.4904, "step": 945 }, { "epoch": 0.07, "grad_norm": 5.581441447316493, "learning_rate": 9.962590671002546e-06, "loss": 0.3287, "step": 946 }, { "epoch": 0.07, "grad_norm": 5.934664194727276, "learning_rate": 9.962448743559295e-06, "loss": 0.4212, "step": 947 }, { "epoch": 0.07, "grad_norm": 16.605366815050633, "learning_rate": 9.962306548410717e-06, "loss": 1.6262, "step": 948 }, { "epoch": 0.07, "grad_norm": 19.68771984160559, "learning_rate": 9.962164085564482e-06, "loss": 1.6384, "step": 949 }, { "epoch": 0.07, "grad_norm": 5.536886180046085, "learning_rate": 9.96202135502828e-06, "loss": 0.3619, "step": 950 }, { "epoch": 0.07, "grad_norm": 5.318501343844009, "learning_rate": 9.961878356809807e-06, "loss": 0.3416, "step": 951 }, { "epoch": 0.07, "grad_norm": 27.310396008779065, "learning_rate": 9.961735090916778e-06, "loss": 1.9451, "step": 952 }, { "epoch": 0.07, "grad_norm": 6.824467378447831, "learning_rate": 9.961591557356922e-06, "loss": 1.2615, "step": 953 }, { "epoch": 0.07, "grad_norm": 4.568540234859305, "learning_rate": 9.961447756137982e-06, "loss": 0.3132, "step": 954 }, { "epoch": 0.07, "grad_norm": 5.688857399962662, "learning_rate": 9.961303687267716e-06, "loss": 0.4383, "step": 955 }, { "epoch": 0.07, "grad_norm": 5.345592108067759, "learning_rate": 9.961159350753895e-06, "loss": 0.383, "step": 956 }, { "epoch": 0.07, "grad_norm": 6.101109969425101, "learning_rate": 9.961014746604308e-06, "loss": 0.342, "step": 957 }, { "epoch": 0.07, "grad_norm": 5.034635507682679, "learning_rate": 9.960869874826753e-06, "loss": 0.3425, "step": 958 }, { "epoch": 0.07, "grad_norm": 5.4233303026801485, "learning_rate": 9.960724735429047e-06, "loss": 0.453, "step": 959 }, { "epoch": 0.07, "grad_norm": 6.085356118665362, "learning_rate": 9.960579328419017e-06, "loss": 0.5502, "step": 960 }, { "epoch": 0.07, "grad_norm": 19.019591513999057, "learning_rate": 9.960433653804509e-06, "loss": 1.7599, "step": 961 }, { "epoch": 0.07, "grad_norm": 21.845001149000442, "learning_rate": 9.960287711593383e-06, "loss": 2.542, "step": 962 }, { "epoch": 0.07, "grad_norm": 20.50984896281616, "learning_rate": 9.96014150179351e-06, "loss": 1.8325, "step": 963 }, { "epoch": 0.07, "grad_norm": 6.265904609136085, "learning_rate": 9.959995024412778e-06, "loss": 0.4834, "step": 964 }, { "epoch": 0.07, "grad_norm": 5.544882485363742, "learning_rate": 9.95984827945909e-06, "loss": 0.3986, "step": 965 }, { "epoch": 0.07, "grad_norm": 5.734402517137861, "learning_rate": 9.95970126694036e-06, "loss": 0.4136, "step": 966 }, { "epoch": 0.07, "grad_norm": 6.698785302988662, "learning_rate": 9.959553986864521e-06, "loss": 1.3106, "step": 967 }, { "epoch": 0.07, "grad_norm": 5.693272662168439, "learning_rate": 9.959406439239517e-06, "loss": 0.4032, "step": 968 }, { "epoch": 0.07, "grad_norm": 7.067641116407061, "learning_rate": 9.95925862407331e-06, "loss": 1.1136, "step": 969 }, { "epoch": 0.07, "grad_norm": 5.1368905021307985, "learning_rate": 9.95911054137387e-06, "loss": 0.3062, "step": 970 }, { "epoch": 0.07, "grad_norm": 24.270960925546333, "learning_rate": 9.958962191149187e-06, "loss": 1.6283, "step": 971 }, { "epoch": 0.07, "grad_norm": 19.775968335441426, "learning_rate": 9.958813573407264e-06, "loss": 1.8884, "step": 972 }, { "epoch": 0.07, "grad_norm": 6.961706249427559, "learning_rate": 9.95866468815612e-06, "loss": 1.3003, "step": 973 }, { "epoch": 0.07, "grad_norm": 34.35828028717145, "learning_rate": 9.958515535403787e-06, "loss": 1.8409, "step": 974 }, { "epoch": 0.07, "grad_norm": 5.9689500869515415, "learning_rate": 9.958366115158308e-06, "loss": 0.3799, "step": 975 }, { "epoch": 0.07, "grad_norm": 7.107440978678895, "learning_rate": 9.958216427427746e-06, "loss": 1.1245, "step": 976 }, { "epoch": 0.07, "grad_norm": 4.944155562618358, "learning_rate": 9.958066472220176e-06, "loss": 0.2876, "step": 977 }, { "epoch": 0.07, "grad_norm": 12.707895684938817, "learning_rate": 9.957916249543686e-06, "loss": 0.8633, "step": 978 }, { "epoch": 0.07, "grad_norm": 17.10923705980783, "learning_rate": 9.957765759406383e-06, "loss": 1.5197, "step": 979 }, { "epoch": 0.07, "grad_norm": 16.27085814102429, "learning_rate": 9.957615001816381e-06, "loss": 1.5703, "step": 980 }, { "epoch": 0.07, "grad_norm": 5.65849784723309, "learning_rate": 9.957463976781818e-06, "loss": 0.2959, "step": 981 }, { "epoch": 0.07, "grad_norm": 7.2960338256841775, "learning_rate": 9.957312684310835e-06, "loss": 0.9801, "step": 982 }, { "epoch": 0.07, "grad_norm": 5.6564182727691135, "learning_rate": 9.957161124411601e-06, "loss": 0.4231, "step": 983 }, { "epoch": 0.07, "grad_norm": 5.880173097038521, "learning_rate": 9.957009297092286e-06, "loss": 0.4023, "step": 984 }, { "epoch": 0.07, "grad_norm": 21.08604915509229, "learning_rate": 9.956857202361082e-06, "loss": 1.5123, "step": 985 }, { "epoch": 0.07, "grad_norm": 23.192260541665203, "learning_rate": 9.956704840226197e-06, "loss": 1.8645, "step": 986 }, { "epoch": 0.07, "grad_norm": 5.435463435502005, "learning_rate": 9.956552210695847e-06, "loss": 0.3, "step": 987 }, { "epoch": 0.07, "grad_norm": 5.400527925494013, "learning_rate": 9.956399313778267e-06, "loss": 0.3257, "step": 988 }, { "epoch": 0.07, "grad_norm": 5.420886520815356, "learning_rate": 9.956246149481703e-06, "loss": 0.3624, "step": 989 }, { "epoch": 0.07, "grad_norm": 5.0141731065171165, "learning_rate": 9.956092717814421e-06, "loss": 0.4364, "step": 990 }, { "epoch": 0.07, "grad_norm": 6.148511754345486, "learning_rate": 9.955939018784698e-06, "loss": 0.4651, "step": 991 }, { "epoch": 0.07, "grad_norm": 5.615607661713724, "learning_rate": 9.955785052400824e-06, "loss": 0.3708, "step": 992 }, { "epoch": 0.07, "grad_norm": 5.539233987228891, "learning_rate": 9.955630818671103e-06, "loss": 0.3474, "step": 993 }, { "epoch": 0.07, "grad_norm": 5.363178636888007, "learning_rate": 9.955476317603858e-06, "loss": 0.3529, "step": 994 }, { "epoch": 0.07, "grad_norm": 5.107227899540807, "learning_rate": 9.955321549207423e-06, "loss": 0.3373, "step": 995 }, { "epoch": 0.07, "grad_norm": 11.593312002178083, "learning_rate": 9.955166513490149e-06, "loss": 0.8126, "step": 996 }, { "epoch": 0.07, "grad_norm": 4.731562163039969, "learning_rate": 9.955011210460398e-06, "loss": 0.2857, "step": 997 }, { "epoch": 0.07, "grad_norm": 5.967403916660322, "learning_rate": 9.954855640126546e-06, "loss": 0.4081, "step": 998 }, { "epoch": 0.07, "grad_norm": 3.7979571970567494, "learning_rate": 9.95469980249699e-06, "loss": 0.2705, "step": 999 }, { "epoch": 0.07, "grad_norm": 3.706720541514571, "learning_rate": 9.954543697580132e-06, "loss": 0.2934, "step": 1000 }, { "epoch": 0.07, "grad_norm": 12.085960245869398, "learning_rate": 9.954387325384397e-06, "loss": 1.4348, "step": 1001 }, { "epoch": 0.07, "grad_norm": 4.597784901757409, "learning_rate": 9.954230685918216e-06, "loss": 0.2528, "step": 1002 }, { "epoch": 0.07, "grad_norm": 28.290737324980824, "learning_rate": 9.954073779190046e-06, "loss": 1.4832, "step": 1003 }, { "epoch": 0.07, "grad_norm": 7.09847218724948, "learning_rate": 9.953916605208346e-06, "loss": 1.1272, "step": 1004 }, { "epoch": 0.07, "grad_norm": 27.997403507772862, "learning_rate": 9.953759163981597e-06, "loss": 1.347, "step": 1005 }, { "epoch": 0.07, "grad_norm": 5.289316943910545, "learning_rate": 9.953601455518292e-06, "loss": 0.3832, "step": 1006 }, { "epoch": 0.07, "grad_norm": 13.994811690774062, "learning_rate": 9.953443479826939e-06, "loss": 2.051, "step": 1007 }, { "epoch": 0.07, "grad_norm": 18.15377022072487, "learning_rate": 9.95328523691606e-06, "loss": 1.614, "step": 1008 }, { "epoch": 0.07, "grad_norm": 6.364919687643424, "learning_rate": 9.953126726794192e-06, "loss": 0.4582, "step": 1009 }, { "epoch": 0.07, "grad_norm": 5.230884767372851, "learning_rate": 9.952967949469886e-06, "loss": 0.3094, "step": 1010 }, { "epoch": 0.07, "grad_norm": 5.534639455503117, "learning_rate": 9.952808904951706e-06, "loss": 0.3437, "step": 1011 }, { "epoch": 0.07, "grad_norm": 5.467071134000427, "learning_rate": 9.952649593248233e-06, "loss": 0.2978, "step": 1012 }, { "epoch": 0.07, "grad_norm": 31.991239667227628, "learning_rate": 9.952490014368062e-06, "loss": 1.5194, "step": 1013 }, { "epoch": 0.07, "grad_norm": 5.639346194158878, "learning_rate": 9.9523301683198e-06, "loss": 0.497, "step": 1014 }, { "epoch": 0.07, "grad_norm": 4.889495375852533, "learning_rate": 9.952170055112072e-06, "loss": 0.3087, "step": 1015 }, { "epoch": 0.07, "grad_norm": 22.83445367992443, "learning_rate": 9.952009674753514e-06, "loss": 1.8757, "step": 1016 }, { "epoch": 0.07, "grad_norm": 5.248395288479579, "learning_rate": 9.95184902725278e-06, "loss": 0.3024, "step": 1017 }, { "epoch": 0.07, "grad_norm": 5.402962744381052, "learning_rate": 9.951688112618531e-06, "loss": 0.3347, "step": 1018 }, { "epoch": 0.07, "grad_norm": 5.854509889560893, "learning_rate": 9.951526930859455e-06, "loss": 0.3991, "step": 1019 }, { "epoch": 0.07, "grad_norm": 15.729710374665407, "learning_rate": 9.951365481984243e-06, "loss": 1.437, "step": 1020 }, { "epoch": 0.07, "grad_norm": 6.7489135891101055, "learning_rate": 9.951203766001605e-06, "loss": 1.0157, "step": 1021 }, { "epoch": 0.07, "grad_norm": 14.632384184928805, "learning_rate": 9.951041782920263e-06, "loss": 1.6548, "step": 1022 }, { "epoch": 0.07, "grad_norm": 5.603951083577964, "learning_rate": 9.95087953274896e-06, "loss": 0.3645, "step": 1023 }, { "epoch": 0.07, "grad_norm": 16.737753575898076, "learning_rate": 9.950717015496447e-06, "loss": 1.73, "step": 1024 }, { "epoch": 0.07, "grad_norm": 6.193476981369923, "learning_rate": 9.950554231171488e-06, "loss": 1.2339, "step": 1025 }, { "epoch": 0.07, "grad_norm": 7.131817106834927, "learning_rate": 9.950391179782868e-06, "loss": 1.2499, "step": 1026 }, { "epoch": 0.07, "grad_norm": 12.89538465451892, "learning_rate": 9.950227861339383e-06, "loss": 1.3586, "step": 1027 }, { "epoch": 0.07, "grad_norm": 4.810675238182825, "learning_rate": 9.950064275849842e-06, "loss": 0.3705, "step": 1028 }, { "epoch": 0.07, "grad_norm": 23.57769534543603, "learning_rate": 9.949900423323068e-06, "loss": 2.3605, "step": 1029 }, { "epoch": 0.07, "grad_norm": 5.665244522038758, "learning_rate": 9.949736303767906e-06, "loss": 0.3523, "step": 1030 }, { "epoch": 0.07, "grad_norm": 7.908834260493989, "learning_rate": 9.949571917193206e-06, "loss": 1.0841, "step": 1031 }, { "epoch": 0.07, "grad_norm": 6.9777189687166, "learning_rate": 9.949407263607835e-06, "loss": 0.539, "step": 1032 }, { "epoch": 0.07, "grad_norm": 6.353303182192173, "learning_rate": 9.949242343020676e-06, "loss": 0.3545, "step": 1033 }, { "epoch": 0.07, "grad_norm": 6.2091902628422675, "learning_rate": 9.949077155440627e-06, "loss": 1.0884, "step": 1034 }, { "epoch": 0.07, "grad_norm": 5.543321347266183, "learning_rate": 9.9489117008766e-06, "loss": 0.4023, "step": 1035 }, { "epoch": 0.07, "grad_norm": 5.75811954156553, "learning_rate": 9.948745979337518e-06, "loss": 0.3013, "step": 1036 }, { "epoch": 0.07, "grad_norm": 14.749982138127548, "learning_rate": 9.948579990832322e-06, "loss": 1.1332, "step": 1037 }, { "epoch": 0.07, "grad_norm": 4.322267110913301, "learning_rate": 9.948413735369968e-06, "loss": 0.2799, "step": 1038 }, { "epoch": 0.07, "grad_norm": 16.93852327438852, "learning_rate": 9.948247212959424e-06, "loss": 1.1368, "step": 1039 }, { "epoch": 0.07, "grad_norm": 16.398640285176025, "learning_rate": 9.948080423609672e-06, "loss": 1.896, "step": 1040 }, { "epoch": 0.07, "grad_norm": 5.90465021139947, "learning_rate": 9.947913367329711e-06, "loss": 0.441, "step": 1041 }, { "epoch": 0.07, "grad_norm": 7.631804916786856, "learning_rate": 9.947746044128552e-06, "loss": 1.3376, "step": 1042 }, { "epoch": 0.07, "grad_norm": 5.378910594378567, "learning_rate": 9.947578454015224e-06, "loss": 0.3068, "step": 1043 }, { "epoch": 0.07, "grad_norm": 5.4056635602377625, "learning_rate": 9.947410596998765e-06, "loss": 0.4328, "step": 1044 }, { "epoch": 0.07, "grad_norm": 4.931706299191818, "learning_rate": 9.947242473088232e-06, "loss": 0.3638, "step": 1045 }, { "epoch": 0.08, "grad_norm": 9.566069963078, "learning_rate": 9.947074082292692e-06, "loss": 1.1144, "step": 1046 }, { "epoch": 0.08, "grad_norm": 19.543633689053063, "learning_rate": 9.946905424621232e-06, "loss": 1.3741, "step": 1047 }, { "epoch": 0.08, "grad_norm": 5.239230839580761, "learning_rate": 9.946736500082949e-06, "loss": 0.4573, "step": 1048 }, { "epoch": 0.08, "grad_norm": 17.683425190561326, "learning_rate": 9.946567308686957e-06, "loss": 1.5906, "step": 1049 }, { "epoch": 0.08, "grad_norm": 6.4088814673137575, "learning_rate": 9.946397850442382e-06, "loss": 1.218, "step": 1050 }, { "epoch": 0.08, "grad_norm": 19.329542397952913, "learning_rate": 9.946228125358365e-06, "loss": 1.5673, "step": 1051 }, { "epoch": 0.08, "grad_norm": 18.369177878706314, "learning_rate": 9.946058133444066e-06, "loss": 1.6964, "step": 1052 }, { "epoch": 0.08, "grad_norm": 5.120700788540297, "learning_rate": 9.945887874708651e-06, "loss": 0.2682, "step": 1053 }, { "epoch": 0.08, "grad_norm": 27.331211388815802, "learning_rate": 9.945717349161306e-06, "loss": 1.5013, "step": 1054 }, { "epoch": 0.08, "grad_norm": 4.2029339747112004, "learning_rate": 9.94554655681123e-06, "loss": 0.2744, "step": 1055 }, { "epoch": 0.08, "grad_norm": 4.906296225824991, "learning_rate": 9.94537549766764e-06, "loss": 0.3679, "step": 1056 }, { "epoch": 0.08, "grad_norm": 13.483784820571286, "learning_rate": 9.945204171739759e-06, "loss": 1.8083, "step": 1057 }, { "epoch": 0.08, "grad_norm": 4.173440761249849, "learning_rate": 9.94503257903683e-06, "loss": 0.2623, "step": 1058 }, { "epoch": 0.08, "grad_norm": 4.869852062467082, "learning_rate": 9.944860719568115e-06, "loss": 0.4118, "step": 1059 }, { "epoch": 0.08, "grad_norm": 3.936096491886314, "learning_rate": 9.94468859334288e-06, "loss": 0.2732, "step": 1060 }, { "epoch": 0.08, "grad_norm": 12.83054942134101, "learning_rate": 9.94451620037041e-06, "loss": 1.6245, "step": 1061 }, { "epoch": 0.08, "grad_norm": 5.56855344885911, "learning_rate": 9.944343540660009e-06, "loss": 0.4662, "step": 1062 }, { "epoch": 0.08, "grad_norm": 18.26516369783222, "learning_rate": 9.94417061422099e-06, "loss": 1.8569, "step": 1063 }, { "epoch": 0.08, "grad_norm": 25.68537608629959, "learning_rate": 9.943997421062679e-06, "loss": 1.994, "step": 1064 }, { "epoch": 0.08, "grad_norm": 5.147561404053151, "learning_rate": 9.943823961194422e-06, "loss": 0.2685, "step": 1065 }, { "epoch": 0.08, "grad_norm": 7.074082567133689, "learning_rate": 9.943650234625576e-06, "loss": 1.1113, "step": 1066 }, { "epoch": 0.08, "grad_norm": 5.689468180921053, "learning_rate": 9.943476241365513e-06, "loss": 0.3502, "step": 1067 }, { "epoch": 0.08, "grad_norm": 5.468950719950903, "learning_rate": 9.943301981423618e-06, "loss": 0.3943, "step": 1068 }, { "epoch": 0.08, "grad_norm": 5.268829213913781, "learning_rate": 9.943127454809293e-06, "loss": 0.3001, "step": 1069 }, { "epoch": 0.08, "grad_norm": 6.311940250663888, "learning_rate": 9.942952661531952e-06, "loss": 0.4237, "step": 1070 }, { "epoch": 0.08, "grad_norm": 12.544382625903383, "learning_rate": 9.942777601601025e-06, "loss": 1.0909, "step": 1071 }, { "epoch": 0.08, "grad_norm": 4.528665502871194, "learning_rate": 9.942602275025956e-06, "loss": 0.3691, "step": 1072 }, { "epoch": 0.08, "grad_norm": 18.296288006049455, "learning_rate": 9.942426681816204e-06, "loss": 1.3732, "step": 1073 }, { "epoch": 0.08, "grad_norm": 19.458138257340256, "learning_rate": 9.94225082198124e-06, "loss": 1.545, "step": 1074 }, { "epoch": 0.08, "grad_norm": 6.8080106187965495, "learning_rate": 9.94207469553055e-06, "loss": 1.1458, "step": 1075 }, { "epoch": 0.08, "grad_norm": 4.952627945605125, "learning_rate": 9.94189830247364e-06, "loss": 0.4497, "step": 1076 }, { "epoch": 0.08, "grad_norm": 14.363210144969107, "learning_rate": 9.94172164282002e-06, "loss": 1.3193, "step": 1077 }, { "epoch": 0.08, "grad_norm": 13.792267118164284, "learning_rate": 9.941544716579224e-06, "loss": 1.3253, "step": 1078 }, { "epoch": 0.08, "grad_norm": 6.21081993457736, "learning_rate": 9.941367523760796e-06, "loss": 0.5212, "step": 1079 }, { "epoch": 0.08, "grad_norm": 4.944630531261782, "learning_rate": 9.941190064374294e-06, "loss": 0.3335, "step": 1080 }, { "epoch": 0.08, "grad_norm": 8.817041742493007, "learning_rate": 9.94101233842929e-06, "loss": 0.9346, "step": 1081 }, { "epoch": 0.08, "grad_norm": 4.8077440391916655, "learning_rate": 9.940834345935374e-06, "loss": 0.4845, "step": 1082 }, { "epoch": 0.08, "grad_norm": 4.298431786020821, "learning_rate": 9.940656086902147e-06, "loss": 0.4098, "step": 1083 }, { "epoch": 0.08, "grad_norm": 4.442624139382252, "learning_rate": 9.940477561339225e-06, "loss": 0.3548, "step": 1084 }, { "epoch": 0.08, "grad_norm": 3.7024898169378404, "learning_rate": 9.94029876925624e-06, "loss": 0.2541, "step": 1085 }, { "epoch": 0.08, "grad_norm": 4.725580127432489, "learning_rate": 9.940119710662837e-06, "loss": 0.4102, "step": 1086 }, { "epoch": 0.08, "grad_norm": 4.416017805799512, "learning_rate": 9.939940385568673e-06, "loss": 0.3751, "step": 1087 }, { "epoch": 0.08, "grad_norm": 5.416591158307206, "learning_rate": 9.939760793983426e-06, "loss": 1.3413, "step": 1088 }, { "epoch": 0.08, "grad_norm": 4.381594715212119, "learning_rate": 9.939580935916782e-06, "loss": 0.4114, "step": 1089 }, { "epoch": 0.08, "grad_norm": 5.344216621949239, "learning_rate": 9.93940081137844e-06, "loss": 1.0416, "step": 1090 }, { "epoch": 0.08, "grad_norm": 3.705184850456659, "learning_rate": 9.939220420378126e-06, "loss": 0.2969, "step": 1091 }, { "epoch": 0.08, "grad_norm": 37.52896015451271, "learning_rate": 9.939039762925565e-06, "loss": 1.2524, "step": 1092 }, { "epoch": 0.08, "grad_norm": 5.856914723581127, "learning_rate": 9.938858839030503e-06, "loss": 1.2074, "step": 1093 }, { "epoch": 0.08, "grad_norm": 4.318587687812586, "learning_rate": 9.9386776487027e-06, "loss": 0.3047, "step": 1094 }, { "epoch": 0.08, "grad_norm": 5.3500887138880895, "learning_rate": 9.938496191951934e-06, "loss": 0.4185, "step": 1095 }, { "epoch": 0.08, "grad_norm": 25.39778279219018, "learning_rate": 9.938314468787992e-06, "loss": 1.7622, "step": 1096 }, { "epoch": 0.08, "grad_norm": 7.915661860363944, "learning_rate": 9.938132479220675e-06, "loss": 0.4615, "step": 1097 }, { "epoch": 0.08, "grad_norm": 18.81773359338087, "learning_rate": 9.937950223259804e-06, "loss": 0.8173, "step": 1098 }, { "epoch": 0.08, "grad_norm": 5.033826286261849, "learning_rate": 9.93776770091521e-06, "loss": 0.3894, "step": 1099 }, { "epoch": 0.08, "grad_norm": 6.450395183888688, "learning_rate": 9.937584912196738e-06, "loss": 0.9072, "step": 1100 }, { "epoch": 0.08, "grad_norm": 22.083181147233276, "learning_rate": 9.937401857114249e-06, "loss": 1.7759, "step": 1101 }, { "epoch": 0.08, "grad_norm": 4.713742935416431, "learning_rate": 9.93721853567762e-06, "loss": 0.2838, "step": 1102 }, { "epoch": 0.08, "grad_norm": 7.434156322083973, "learning_rate": 9.937034947896739e-06, "loss": 1.204, "step": 1103 }, { "epoch": 0.08, "grad_norm": 8.528561967912896, "learning_rate": 9.936851093781512e-06, "loss": 0.3468, "step": 1104 }, { "epoch": 0.08, "grad_norm": 5.468158765236823, "learning_rate": 9.936666973341853e-06, "loss": 0.44, "step": 1105 }, { "epoch": 0.08, "grad_norm": 7.380932824940822, "learning_rate": 9.936482586587698e-06, "loss": 1.1382, "step": 1106 }, { "epoch": 0.08, "grad_norm": 5.05999832778143, "learning_rate": 9.936297933528992e-06, "loss": 0.3325, "step": 1107 }, { "epoch": 0.08, "grad_norm": 6.004351799921172, "learning_rate": 9.9361130141757e-06, "loss": 0.3072, "step": 1108 }, { "epoch": 0.08, "grad_norm": 7.104545421210339, "learning_rate": 9.935927828537793e-06, "loss": 1.0785, "step": 1109 }, { "epoch": 0.08, "grad_norm": 24.285907817704224, "learning_rate": 9.935742376625264e-06, "loss": 0.9317, "step": 1110 }, { "epoch": 0.08, "grad_norm": 12.376630075119275, "learning_rate": 9.935556658448115e-06, "loss": 1.6474, "step": 1111 }, { "epoch": 0.08, "grad_norm": 5.212213427571981, "learning_rate": 9.935370674016369e-06, "loss": 0.3369, "step": 1112 }, { "epoch": 0.08, "grad_norm": 3.8341694858671795, "learning_rate": 9.935184423340054e-06, "loss": 0.3136, "step": 1113 }, { "epoch": 0.08, "grad_norm": 4.353295132514207, "learning_rate": 9.934997906429222e-06, "loss": 0.3176, "step": 1114 }, { "epoch": 0.08, "grad_norm": 5.602180183960513, "learning_rate": 9.934811123293933e-06, "loss": 0.3624, "step": 1115 }, { "epoch": 0.08, "grad_norm": 6.726793720621401, "learning_rate": 9.93462407394426e-06, "loss": 1.1558, "step": 1116 }, { "epoch": 0.08, "grad_norm": 14.247437444204623, "learning_rate": 9.9344367583903e-06, "loss": 1.32, "step": 1117 }, { "epoch": 0.08, "grad_norm": 21.854293743458218, "learning_rate": 9.934249176642155e-06, "loss": 2.2574, "step": 1118 }, { "epoch": 0.08, "grad_norm": 7.219911681316433, "learning_rate": 9.934061328709943e-06, "loss": 0.497, "step": 1119 }, { "epoch": 0.08, "grad_norm": 20.645673635098998, "learning_rate": 9.933873214603798e-06, "loss": 1.7166, "step": 1120 }, { "epoch": 0.08, "grad_norm": 18.712593773696987, "learning_rate": 9.93368483433387e-06, "loss": 1.3808, "step": 1121 }, { "epoch": 0.08, "grad_norm": 4.791226475087332, "learning_rate": 9.933496187910319e-06, "loss": 0.3515, "step": 1122 }, { "epoch": 0.08, "grad_norm": 5.065983280980811, "learning_rate": 9.933307275343324e-06, "loss": 0.3303, "step": 1123 }, { "epoch": 0.08, "grad_norm": 5.088660449815601, "learning_rate": 9.933118096643075e-06, "loss": 0.2788, "step": 1124 }, { "epoch": 0.08, "grad_norm": 5.075078021971279, "learning_rate": 9.932928651819777e-06, "loss": 0.3771, "step": 1125 }, { "epoch": 0.08, "grad_norm": 17.264024737401037, "learning_rate": 9.93273894088365e-06, "loss": 1.0819, "step": 1126 }, { "epoch": 0.08, "grad_norm": 14.633356517802534, "learning_rate": 9.93254896384493e-06, "loss": 1.4891, "step": 1127 }, { "epoch": 0.08, "grad_norm": 6.19287524812615, "learning_rate": 9.932358720713863e-06, "loss": 0.3406, "step": 1128 }, { "epoch": 0.08, "grad_norm": 16.311806477368993, "learning_rate": 9.932168211500714e-06, "loss": 1.546, "step": 1129 }, { "epoch": 0.08, "grad_norm": 8.105308658040636, "learning_rate": 9.931977436215759e-06, "loss": 1.1361, "step": 1130 }, { "epoch": 0.08, "grad_norm": 31.08025094919286, "learning_rate": 9.93178639486929e-06, "loss": 1.3839, "step": 1131 }, { "epoch": 0.08, "grad_norm": 23.50142115529522, "learning_rate": 9.93159508747161e-06, "loss": 1.181, "step": 1132 }, { "epoch": 0.08, "grad_norm": 7.222085964395209, "learning_rate": 9.931403514033046e-06, "loss": 0.9178, "step": 1133 }, { "epoch": 0.08, "grad_norm": 22.04299557331215, "learning_rate": 9.931211674563926e-06, "loss": 1.1702, "step": 1134 }, { "epoch": 0.08, "grad_norm": 7.08775632263463, "learning_rate": 9.931019569074604e-06, "loss": 1.0792, "step": 1135 }, { "epoch": 0.08, "grad_norm": 6.01629511803259, "learning_rate": 9.93082719757544e-06, "loss": 1.0139, "step": 1136 }, { "epoch": 0.08, "grad_norm": 7.382177863666987, "learning_rate": 9.930634560076813e-06, "loss": 0.9481, "step": 1137 }, { "epoch": 0.08, "grad_norm": 6.5326407693380455, "learning_rate": 9.930441656589115e-06, "loss": 1.2244, "step": 1138 }, { "epoch": 0.08, "grad_norm": 4.381004849797672, "learning_rate": 9.930248487122752e-06, "loss": 0.2905, "step": 1139 }, { "epoch": 0.08, "grad_norm": 19.1175493805288, "learning_rate": 9.930055051688146e-06, "loss": 1.4895, "step": 1140 }, { "epoch": 0.08, "grad_norm": 30.976611426800304, "learning_rate": 9.929861350295732e-06, "loss": 1.775, "step": 1141 }, { "epoch": 0.08, "grad_norm": 11.592504775286267, "learning_rate": 9.929667382955956e-06, "loss": 1.2247, "step": 1142 }, { "epoch": 0.08, "grad_norm": 5.407250179488634, "learning_rate": 9.929473149679285e-06, "loss": 0.3693, "step": 1143 }, { "epoch": 0.08, "grad_norm": 15.567716540715288, "learning_rate": 9.929278650476199e-06, "loss": 1.7574, "step": 1144 }, { "epoch": 0.08, "grad_norm": 6.198809729838973, "learning_rate": 9.929083885357186e-06, "loss": 0.4056, "step": 1145 }, { "epoch": 0.08, "grad_norm": 5.1679334504216445, "learning_rate": 9.928888854332756e-06, "loss": 0.3297, "step": 1146 }, { "epoch": 0.08, "grad_norm": 6.107192116874614, "learning_rate": 9.928693557413429e-06, "loss": 0.477, "step": 1147 }, { "epoch": 0.08, "grad_norm": 5.144230948571216, "learning_rate": 9.928497994609741e-06, "loss": 0.3067, "step": 1148 }, { "epoch": 0.08, "grad_norm": 157.78541130573043, "learning_rate": 9.92830216593224e-06, "loss": 1.1939, "step": 1149 }, { "epoch": 0.08, "grad_norm": 28.04859282471981, "learning_rate": 9.928106071391493e-06, "loss": 1.5722, "step": 1150 }, { "epoch": 0.08, "grad_norm": 11.549141040217606, "learning_rate": 9.927909710998079e-06, "loss": 1.5009, "step": 1151 }, { "epoch": 0.08, "grad_norm": 17.43210957753329, "learning_rate": 9.927713084762587e-06, "loss": 1.5925, "step": 1152 }, { "epoch": 0.08, "grad_norm": 5.341714506660686, "learning_rate": 9.927516192695625e-06, "loss": 0.3023, "step": 1153 }, { "epoch": 0.08, "grad_norm": 29.76643864947851, "learning_rate": 9.927319034807819e-06, "loss": 2.4743, "step": 1154 }, { "epoch": 0.08, "grad_norm": 8.141131764821056, "learning_rate": 9.927121611109802e-06, "loss": 1.1532, "step": 1155 }, { "epoch": 0.08, "grad_norm": 6.5915363700495435, "learning_rate": 9.926923921612222e-06, "loss": 0.3201, "step": 1156 }, { "epoch": 0.08, "grad_norm": 14.047076314080094, "learning_rate": 9.926725966325748e-06, "loss": 1.5111, "step": 1157 }, { "epoch": 0.08, "grad_norm": 5.024769847908029, "learning_rate": 9.926527745261055e-06, "loss": 0.3488, "step": 1158 }, { "epoch": 0.08, "grad_norm": 5.099904776186859, "learning_rate": 9.92632925842884e-06, "loss": 0.394, "step": 1159 }, { "epoch": 0.08, "grad_norm": 5.066861829703255, "learning_rate": 9.926130505839807e-06, "loss": 0.3636, "step": 1160 }, { "epoch": 0.08, "grad_norm": 11.763266091933415, "learning_rate": 9.92593148750468e-06, "loss": 1.4117, "step": 1161 }, { "epoch": 0.08, "grad_norm": 4.970164276442249, "learning_rate": 9.925732203434195e-06, "loss": 0.3798, "step": 1162 }, { "epoch": 0.08, "grad_norm": 4.213292105766587, "learning_rate": 9.925532653639102e-06, "loss": 0.2701, "step": 1163 }, { "epoch": 0.08, "grad_norm": 17.55421519533059, "learning_rate": 9.925332838130169e-06, "loss": 1.5495, "step": 1164 }, { "epoch": 0.08, "grad_norm": 10.289551776685038, "learning_rate": 9.92513275691817e-06, "loss": 1.382, "step": 1165 }, { "epoch": 0.08, "grad_norm": 6.719776523943642, "learning_rate": 9.924932410013902e-06, "loss": 1.1311, "step": 1166 }, { "epoch": 0.08, "grad_norm": 14.610662795345672, "learning_rate": 9.924731797428174e-06, "loss": 0.9874, "step": 1167 }, { "epoch": 0.08, "grad_norm": 20.755465625546716, "learning_rate": 9.924530919171806e-06, "loss": 1.5499, "step": 1168 }, { "epoch": 0.08, "grad_norm": 8.376327759143498, "learning_rate": 9.924329775255635e-06, "loss": 0.9628, "step": 1169 }, { "epoch": 0.08, "grad_norm": 37.92296774953209, "learning_rate": 9.92412836569051e-06, "loss": 1.524, "step": 1170 }, { "epoch": 0.08, "grad_norm": 5.248220566429882, "learning_rate": 9.9239266904873e-06, "loss": 0.3, "step": 1171 }, { "epoch": 0.08, "grad_norm": 12.171527186090684, "learning_rate": 9.923724749656884e-06, "loss": 1.8038, "step": 1172 }, { "epoch": 0.08, "grad_norm": 4.542996612341364, "learning_rate": 9.923522543210155e-06, "loss": 0.3213, "step": 1173 }, { "epoch": 0.08, "grad_norm": 16.132583867495494, "learning_rate": 9.923320071158022e-06, "loss": 1.6418, "step": 1174 }, { "epoch": 0.08, "grad_norm": 5.633682955391493, "learning_rate": 9.923117333511405e-06, "loss": 0.3331, "step": 1175 }, { "epoch": 0.08, "grad_norm": 5.4602941527538515, "learning_rate": 9.922914330281244e-06, "loss": 0.4163, "step": 1176 }, { "epoch": 0.08, "grad_norm": 20.279298951935978, "learning_rate": 9.922711061478489e-06, "loss": 1.2136, "step": 1177 }, { "epoch": 0.08, "grad_norm": 27.37246396574736, "learning_rate": 9.922507527114105e-06, "loss": 1.3649, "step": 1178 }, { "epoch": 0.08, "grad_norm": 7.446713021202938, "learning_rate": 9.922303727199074e-06, "loss": 1.2471, "step": 1179 }, { "epoch": 0.08, "grad_norm": 15.578583664717113, "learning_rate": 9.922099661744388e-06, "loss": 1.5964, "step": 1180 }, { "epoch": 0.08, "grad_norm": 22.183804925187086, "learning_rate": 9.921895330761055e-06, "loss": 1.894, "step": 1181 }, { "epoch": 0.08, "grad_norm": 12.749067332029538, "learning_rate": 9.921690734260102e-06, "loss": 1.1586, "step": 1182 }, { "epoch": 0.08, "grad_norm": 4.971068322238837, "learning_rate": 9.921485872252563e-06, "loss": 0.3214, "step": 1183 }, { "epoch": 0.08, "grad_norm": 5.305113241248176, "learning_rate": 9.921280744749488e-06, "loss": 0.4237, "step": 1184 }, { "epoch": 0.08, "grad_norm": 49.41343021323377, "learning_rate": 9.921075351761947e-06, "loss": 1.425, "step": 1185 }, { "epoch": 0.09, "grad_norm": 5.743917907226199, "learning_rate": 9.920869693301016e-06, "loss": 0.3012, "step": 1186 }, { "epoch": 0.09, "grad_norm": 11.375249437126435, "learning_rate": 9.920663769377794e-06, "loss": 0.993, "step": 1187 }, { "epoch": 0.09, "grad_norm": 16.954126560142665, "learning_rate": 9.920457580003385e-06, "loss": 1.0991, "step": 1188 }, { "epoch": 0.09, "grad_norm": 4.065047662513692, "learning_rate": 9.920251125188915e-06, "loss": 0.3507, "step": 1189 }, { "epoch": 0.09, "grad_norm": 3.773478707379328, "learning_rate": 9.920044404945522e-06, "loss": 0.2056, "step": 1190 }, { "epoch": 0.09, "grad_norm": 4.258208494886032, "learning_rate": 9.919837419284357e-06, "loss": 0.3328, "step": 1191 }, { "epoch": 0.09, "grad_norm": 8.42585039975603, "learning_rate": 9.919630168216586e-06, "loss": 1.3871, "step": 1192 }, { "epoch": 0.09, "grad_norm": 8.29256512403175, "learning_rate": 9.919422651753388e-06, "loss": 1.2209, "step": 1193 }, { "epoch": 0.09, "grad_norm": 8.125990430753061, "learning_rate": 9.919214869905959e-06, "loss": 1.2191, "step": 1194 }, { "epoch": 0.09, "grad_norm": 4.313315459503424, "learning_rate": 9.919006822685508e-06, "loss": 0.2754, "step": 1195 }, { "epoch": 0.09, "grad_norm": 18.35277464449867, "learning_rate": 9.91879851010326e-06, "loss": 1.4576, "step": 1196 }, { "epoch": 0.09, "grad_norm": 5.747557232947898, "learning_rate": 9.918589932170448e-06, "loss": 0.4487, "step": 1197 }, { "epoch": 0.09, "grad_norm": 5.161043940040854, "learning_rate": 9.91838108889833e-06, "loss": 0.3825, "step": 1198 }, { "epoch": 0.09, "grad_norm": 24.851469825251666, "learning_rate": 9.918171980298167e-06, "loss": 1.9792, "step": 1199 }, { "epoch": 0.09, "grad_norm": 15.363669861303384, "learning_rate": 9.917962606381244e-06, "loss": 1.1014, "step": 1200 }, { "epoch": 0.09, "grad_norm": 10.624535928590173, "learning_rate": 9.917752967158852e-06, "loss": 0.8628, "step": 1201 }, { "epoch": 0.09, "grad_norm": 14.280665638473083, "learning_rate": 9.917543062642303e-06, "loss": 1.3525, "step": 1202 }, { "epoch": 0.09, "grad_norm": 4.705723612738509, "learning_rate": 9.917332892842922e-06, "loss": 0.3352, "step": 1203 }, { "epoch": 0.09, "grad_norm": 12.840995854415139, "learning_rate": 9.917122457772042e-06, "loss": 1.3151, "step": 1204 }, { "epoch": 0.09, "grad_norm": 15.0162277808523, "learning_rate": 9.916911757441017e-06, "loss": 1.1447, "step": 1205 }, { "epoch": 0.09, "grad_norm": 4.7158582991153875, "learning_rate": 9.916700791861215e-06, "loss": 0.3091, "step": 1206 }, { "epoch": 0.09, "grad_norm": 4.366505652041513, "learning_rate": 9.916489561044016e-06, "loss": 0.2852, "step": 1207 }, { "epoch": 0.09, "grad_norm": 4.654151163509973, "learning_rate": 9.916278065000815e-06, "loss": 0.3032, "step": 1208 }, { "epoch": 0.09, "grad_norm": 4.073522836361969, "learning_rate": 9.916066303743023e-06, "loss": 0.233, "step": 1209 }, { "epoch": 0.09, "grad_norm": 4.475662547519268, "learning_rate": 9.915854277282061e-06, "loss": 0.2917, "step": 1210 }, { "epoch": 0.09, "grad_norm": 7.649294838389349, "learning_rate": 9.915641985629368e-06, "loss": 1.1066, "step": 1211 }, { "epoch": 0.09, "grad_norm": 28.844202771412654, "learning_rate": 9.915429428796397e-06, "loss": 1.5577, "step": 1212 }, { "epoch": 0.09, "grad_norm": 5.679702425459298, "learning_rate": 9.915216606794615e-06, "loss": 0.4029, "step": 1213 }, { "epoch": 0.09, "grad_norm": 6.786145675265349, "learning_rate": 9.9150035196355e-06, "loss": 1.1171, "step": 1214 }, { "epoch": 0.09, "grad_norm": 5.188169151698947, "learning_rate": 9.914790167330552e-06, "loss": 0.3584, "step": 1215 }, { "epoch": 0.09, "grad_norm": 7.860687116166146, "learning_rate": 9.914576549891276e-06, "loss": 1.3512, "step": 1216 }, { "epoch": 0.09, "grad_norm": 5.598513658890038, "learning_rate": 9.914362667329197e-06, "loss": 0.3912, "step": 1217 }, { "epoch": 0.09, "grad_norm": 18.024241247272045, "learning_rate": 9.914148519655857e-06, "loss": 1.4113, "step": 1218 }, { "epoch": 0.09, "grad_norm": 7.228976044728462, "learning_rate": 9.913934106882805e-06, "loss": 0.4196, "step": 1219 }, { "epoch": 0.09, "grad_norm": 4.435238602482972, "learning_rate": 9.913719429021607e-06, "loss": 0.2215, "step": 1220 }, { "epoch": 0.09, "grad_norm": 29.804118052424897, "learning_rate": 9.913504486083845e-06, "loss": 1.1697, "step": 1221 }, { "epoch": 0.09, "grad_norm": 25.393238965936575, "learning_rate": 9.913289278081115e-06, "loss": 1.3743, "step": 1222 }, { "epoch": 0.09, "grad_norm": 4.892946407956693, "learning_rate": 9.913073805025027e-06, "loss": 0.4065, "step": 1223 }, { "epoch": 0.09, "grad_norm": 24.521292714117212, "learning_rate": 9.912858066927202e-06, "loss": 1.5423, "step": 1224 }, { "epoch": 0.09, "grad_norm": 4.398168338062987, "learning_rate": 9.912642063799283e-06, "loss": 0.3186, "step": 1225 }, { "epoch": 0.09, "grad_norm": 4.350462071666284, "learning_rate": 9.912425795652918e-06, "loss": 0.2911, "step": 1226 }, { "epoch": 0.09, "grad_norm": 5.542112129858853, "learning_rate": 9.912209262499777e-06, "loss": 1.158, "step": 1227 }, { "epoch": 0.09, "grad_norm": 8.779621546978603, "learning_rate": 9.91199246435154e-06, "loss": 1.1249, "step": 1228 }, { "epoch": 0.09, "grad_norm": 4.280481555617682, "learning_rate": 9.911775401219902e-06, "loss": 0.2727, "step": 1229 }, { "epoch": 0.09, "grad_norm": 5.945467206874026, "learning_rate": 9.911558073116574e-06, "loss": 0.286, "step": 1230 }, { "epoch": 0.09, "grad_norm": 4.239505087915251, "learning_rate": 9.911340480053277e-06, "loss": 0.333, "step": 1231 }, { "epoch": 0.09, "grad_norm": 4.580893016767791, "learning_rate": 9.911122622041754e-06, "loss": 0.2597, "step": 1232 }, { "epoch": 0.09, "grad_norm": 7.120841973051503, "learning_rate": 9.910904499093754e-06, "loss": 1.1175, "step": 1233 }, { "epoch": 0.09, "grad_norm": 4.690895910028979, "learning_rate": 9.910686111221047e-06, "loss": 0.3714, "step": 1234 }, { "epoch": 0.09, "grad_norm": 18.701028749368874, "learning_rate": 9.91046745843541e-06, "loss": 1.8589, "step": 1235 }, { "epoch": 0.09, "grad_norm": 15.225908787396168, "learning_rate": 9.910248540748642e-06, "loss": 1.4354, "step": 1236 }, { "epoch": 0.09, "grad_norm": 16.580877209753062, "learning_rate": 9.91002935817255e-06, "loss": 1.4449, "step": 1237 }, { "epoch": 0.09, "grad_norm": 5.023451683857193, "learning_rate": 9.90980991071896e-06, "loss": 0.2827, "step": 1238 }, { "epoch": 0.09, "grad_norm": 4.006140774579467, "learning_rate": 9.90959019839971e-06, "loss": 0.2765, "step": 1239 }, { "epoch": 0.09, "grad_norm": 4.945510365415166, "learning_rate": 9.909370221226654e-06, "loss": 0.39, "step": 1240 }, { "epoch": 0.09, "grad_norm": 4.518758267998571, "learning_rate": 9.909149979211656e-06, "loss": 0.3171, "step": 1241 }, { "epoch": 0.09, "grad_norm": 55.994392599333345, "learning_rate": 9.9089294723666e-06, "loss": 2.3019, "step": 1242 }, { "epoch": 0.09, "grad_norm": 7.432094373874257, "learning_rate": 9.90870870070338e-06, "loss": 1.082, "step": 1243 }, { "epoch": 0.09, "grad_norm": 5.497388317017798, "learning_rate": 9.908487664233906e-06, "loss": 0.3132, "step": 1244 }, { "epoch": 0.09, "grad_norm": 37.268582521947486, "learning_rate": 9.908266362970101e-06, "loss": 1.3628, "step": 1245 }, { "epoch": 0.09, "grad_norm": 31.876236727970532, "learning_rate": 9.908044796923906e-06, "loss": 1.4446, "step": 1246 }, { "epoch": 0.09, "grad_norm": 21.68515281119724, "learning_rate": 9.90782296610727e-06, "loss": 1.2736, "step": 1247 }, { "epoch": 0.09, "grad_norm": 5.316617318523804, "learning_rate": 9.907600870532165e-06, "loss": 0.3883, "step": 1248 }, { "epoch": 0.09, "grad_norm": 5.329344640700329, "learning_rate": 9.907378510210568e-06, "loss": 0.506, "step": 1249 }, { "epoch": 0.09, "grad_norm": 21.55816843272218, "learning_rate": 9.907155885154476e-06, "loss": 1.6653, "step": 1250 }, { "epoch": 0.09, "grad_norm": 4.924155049239697, "learning_rate": 9.906932995375898e-06, "loss": 0.4057, "step": 1251 }, { "epoch": 0.09, "grad_norm": 20.322846242245273, "learning_rate": 9.906709840886858e-06, "loss": 1.5156, "step": 1252 }, { "epoch": 0.09, "grad_norm": 4.294226843366683, "learning_rate": 9.906486421699397e-06, "loss": 0.2694, "step": 1253 }, { "epoch": 0.09, "grad_norm": 23.898311546609726, "learning_rate": 9.906262737825565e-06, "loss": 1.6885, "step": 1254 }, { "epoch": 0.09, "grad_norm": 4.985205322199002, "learning_rate": 9.906038789277428e-06, "loss": 0.3274, "step": 1255 }, { "epoch": 0.09, "grad_norm": 5.263434571394311, "learning_rate": 9.90581457606707e-06, "loss": 0.3907, "step": 1256 }, { "epoch": 0.09, "grad_norm": 6.020018097671901, "learning_rate": 9.905590098206585e-06, "loss": 0.3903, "step": 1257 }, { "epoch": 0.09, "grad_norm": 17.817662363942986, "learning_rate": 9.905365355708084e-06, "loss": 1.6544, "step": 1258 }, { "epoch": 0.09, "grad_norm": 24.14432311054375, "learning_rate": 9.905140348583688e-06, "loss": 1.7175, "step": 1259 }, { "epoch": 0.09, "grad_norm": 5.201150725494538, "learning_rate": 9.904915076845538e-06, "loss": 0.2825, "step": 1260 }, { "epoch": 0.09, "grad_norm": 5.819896841266351, "learning_rate": 9.904689540505785e-06, "loss": 1.0498, "step": 1261 }, { "epoch": 0.09, "grad_norm": 24.20183397240531, "learning_rate": 9.904463739576598e-06, "loss": 1.553, "step": 1262 }, { "epoch": 0.09, "grad_norm": 5.242140879999844, "learning_rate": 9.904237674070155e-06, "loss": 0.3399, "step": 1263 }, { "epoch": 0.09, "grad_norm": 5.010725040635945, "learning_rate": 9.904011343998654e-06, "loss": 0.3077, "step": 1264 }, { "epoch": 0.09, "grad_norm": 13.591049370208275, "learning_rate": 9.903784749374303e-06, "loss": 1.4313, "step": 1265 }, { "epoch": 0.09, "grad_norm": 35.37124869106234, "learning_rate": 9.903557890209327e-06, "loss": 1.6835, "step": 1266 }, { "epoch": 0.09, "grad_norm": 18.17257942435633, "learning_rate": 9.903330766515965e-06, "loss": 2.055, "step": 1267 }, { "epoch": 0.09, "grad_norm": 24.201525488474633, "learning_rate": 9.903103378306466e-06, "loss": 2.1203, "step": 1268 }, { "epoch": 0.09, "grad_norm": 6.811445086982506, "learning_rate": 9.9028757255931e-06, "loss": 1.2969, "step": 1269 }, { "epoch": 0.09, "grad_norm": 7.259580020642507, "learning_rate": 9.902647808388149e-06, "loss": 1.0453, "step": 1270 }, { "epoch": 0.09, "grad_norm": 14.351334043758037, "learning_rate": 9.902419626703903e-06, "loss": 1.119, "step": 1271 }, { "epoch": 0.09, "grad_norm": 3.804908652398972, "learning_rate": 9.902191180552676e-06, "loss": 0.2465, "step": 1272 }, { "epoch": 0.09, "grad_norm": 6.760530514693415, "learning_rate": 9.90196246994679e-06, "loss": 1.0584, "step": 1273 }, { "epoch": 0.09, "grad_norm": 50.898487431190006, "learning_rate": 9.901733494898585e-06, "loss": 1.1598, "step": 1274 }, { "epoch": 0.09, "grad_norm": 7.643096989274367, "learning_rate": 9.901504255420411e-06, "loss": 0.9434, "step": 1275 }, { "epoch": 0.09, "grad_norm": 3.924485180599043, "learning_rate": 9.901274751524636e-06, "loss": 0.271, "step": 1276 }, { "epoch": 0.09, "grad_norm": 5.24033833131271, "learning_rate": 9.901044983223642e-06, "loss": 0.4015, "step": 1277 }, { "epoch": 0.09, "grad_norm": 4.230000439167124, "learning_rate": 9.90081495052982e-06, "loss": 0.2793, "step": 1278 }, { "epoch": 0.09, "grad_norm": 5.781826810634509, "learning_rate": 9.900584653455584e-06, "loss": 0.4217, "step": 1279 }, { "epoch": 0.09, "grad_norm": 16.116117184038842, "learning_rate": 9.900354092013355e-06, "loss": 1.2585, "step": 1280 }, { "epoch": 0.09, "grad_norm": 6.954456853022862, "learning_rate": 9.90012326621557e-06, "loss": 1.1552, "step": 1281 }, { "epoch": 0.09, "grad_norm": 5.568854672496837, "learning_rate": 9.899892176074687e-06, "loss": 0.9508, "step": 1282 }, { "epoch": 0.09, "grad_norm": 4.9399696754580935, "learning_rate": 9.899660821603165e-06, "loss": 0.3042, "step": 1283 }, { "epoch": 0.09, "grad_norm": 4.522219148579822, "learning_rate": 9.899429202813487e-06, "loss": 0.3654, "step": 1284 }, { "epoch": 0.09, "grad_norm": 7.112608231483578, "learning_rate": 9.89919731971815e-06, "loss": 1.244, "step": 1285 }, { "epoch": 0.09, "grad_norm": 14.733330562870412, "learning_rate": 9.898965172329663e-06, "loss": 1.1815, "step": 1286 }, { "epoch": 0.09, "grad_norm": 14.11278625999045, "learning_rate": 9.898732760660549e-06, "loss": 1.2629, "step": 1287 }, { "epoch": 0.09, "grad_norm": 26.19470012879902, "learning_rate": 9.898500084723344e-06, "loss": 1.6996, "step": 1288 }, { "epoch": 0.09, "grad_norm": 6.084476872534638, "learning_rate": 9.8982671445306e-06, "loss": 1.0742, "step": 1289 }, { "epoch": 0.09, "grad_norm": 11.867059452505172, "learning_rate": 9.898033940094886e-06, "loss": 1.3206, "step": 1290 }, { "epoch": 0.09, "grad_norm": 4.848601832202022, "learning_rate": 9.89780047142878e-06, "loss": 0.2845, "step": 1291 }, { "epoch": 0.09, "grad_norm": 4.946486423625804, "learning_rate": 9.89756673854488e-06, "loss": 0.3171, "step": 1292 }, { "epoch": 0.09, "grad_norm": 6.200312379842299, "learning_rate": 9.897332741455791e-06, "loss": 1.1281, "step": 1293 }, { "epoch": 0.09, "grad_norm": 16.90388627295913, "learning_rate": 9.897098480174138e-06, "loss": 1.5188, "step": 1294 }, { "epoch": 0.09, "grad_norm": 16.09567285893331, "learning_rate": 9.896863954712558e-06, "loss": 1.3705, "step": 1295 }, { "epoch": 0.09, "grad_norm": 18.026428764666807, "learning_rate": 9.896629165083705e-06, "loss": 1.1288, "step": 1296 }, { "epoch": 0.09, "grad_norm": 22.677048596510026, "learning_rate": 9.896394111300242e-06, "loss": 1.201, "step": 1297 }, { "epoch": 0.09, "grad_norm": 6.706941268920996, "learning_rate": 9.896158793374852e-06, "loss": 0.3358, "step": 1298 }, { "epoch": 0.09, "grad_norm": 7.12468976306126, "learning_rate": 9.895923211320227e-06, "loss": 1.1204, "step": 1299 }, { "epoch": 0.09, "grad_norm": 5.798170473391381, "learning_rate": 9.895687365149077e-06, "loss": 0.3666, "step": 1300 }, { "epoch": 0.09, "grad_norm": 5.001667377747705, "learning_rate": 9.895451254874124e-06, "loss": 0.2938, "step": 1301 }, { "epoch": 0.09, "grad_norm": 5.199983610022982, "learning_rate": 9.895214880508107e-06, "loss": 0.2697, "step": 1302 }, { "epoch": 0.09, "grad_norm": 20.207407653466795, "learning_rate": 9.894978242063777e-06, "loss": 1.5186, "step": 1303 }, { "epoch": 0.09, "grad_norm": 5.147844971812504, "learning_rate": 9.894741339553898e-06, "loss": 0.4068, "step": 1304 }, { "epoch": 0.09, "grad_norm": 4.371043665444003, "learning_rate": 9.894504172991252e-06, "loss": 0.2841, "step": 1305 }, { "epoch": 0.09, "grad_norm": 9.35831096073354, "learning_rate": 9.894266742388633e-06, "loss": 0.8248, "step": 1306 }, { "epoch": 0.09, "grad_norm": 12.243268022855707, "learning_rate": 9.894029047758848e-06, "loss": 0.9716, "step": 1307 }, { "epoch": 0.09, "grad_norm": 15.794495262285114, "learning_rate": 9.893791089114722e-06, "loss": 1.3405, "step": 1308 }, { "epoch": 0.09, "grad_norm": 21.614676668686165, "learning_rate": 9.89355286646909e-06, "loss": 1.7143, "step": 1309 }, { "epoch": 0.09, "grad_norm": 5.508179254462062, "learning_rate": 9.893314379834804e-06, "loss": 0.3135, "step": 1310 }, { "epoch": 0.09, "grad_norm": 5.315310423311075, "learning_rate": 9.89307562922473e-06, "loss": 0.3068, "step": 1311 }, { "epoch": 0.09, "grad_norm": 11.968495042707564, "learning_rate": 9.892836614651747e-06, "loss": 1.2032, "step": 1312 }, { "epoch": 0.09, "grad_norm": 4.7986181166595605, "learning_rate": 9.89259733612875e-06, "loss": 0.3783, "step": 1313 }, { "epoch": 0.09, "grad_norm": 43.475833191279555, "learning_rate": 9.892357793668643e-06, "loss": 1.6647, "step": 1314 }, { "epoch": 0.09, "grad_norm": 6.687739312085653, "learning_rate": 9.892117987284354e-06, "loss": 1.1149, "step": 1315 }, { "epoch": 0.09, "grad_norm": 4.641594503382766, "learning_rate": 9.891877916988817e-06, "loss": 0.2625, "step": 1316 }, { "epoch": 0.09, "grad_norm": 6.878703363898584, "learning_rate": 9.891637582794983e-06, "loss": 1.1062, "step": 1317 }, { "epoch": 0.09, "grad_norm": 5.505525224808377, "learning_rate": 9.891396984715817e-06, "loss": 0.3389, "step": 1318 }, { "epoch": 0.09, "grad_norm": 25.83678496119158, "learning_rate": 9.891156122764297e-06, "loss": 1.1458, "step": 1319 }, { "epoch": 0.09, "grad_norm": 4.97262714267416, "learning_rate": 9.890914996953422e-06, "loss": 0.2587, "step": 1320 }, { "epoch": 0.09, "grad_norm": 4.831352324389743, "learning_rate": 9.890673607296194e-06, "loss": 0.3261, "step": 1321 }, { "epoch": 0.09, "grad_norm": 8.672819541079743, "learning_rate": 9.890431953805637e-06, "loss": 0.853, "step": 1322 }, { "epoch": 0.09, "grad_norm": 11.368585121523786, "learning_rate": 9.890190036494787e-06, "loss": 1.2384, "step": 1323 }, { "epoch": 0.09, "grad_norm": 3.8191135565020184, "learning_rate": 9.889947855376696e-06, "loss": 0.2938, "step": 1324 }, { "epoch": 0.1, "grad_norm": 4.550898580453021, "learning_rate": 9.889705410464428e-06, "loss": 0.3296, "step": 1325 }, { "epoch": 0.1, "grad_norm": 13.601766443925673, "learning_rate": 9.88946270177106e-06, "loss": 1.1481, "step": 1326 }, { "epoch": 0.1, "grad_norm": 6.366163809626671, "learning_rate": 9.889219729309686e-06, "loss": 1.1183, "step": 1327 }, { "epoch": 0.1, "grad_norm": 5.9617296855980335, "learning_rate": 9.888976493093417e-06, "loss": 0.4315, "step": 1328 }, { "epoch": 0.1, "grad_norm": 19.225460792792326, "learning_rate": 9.88873299313537e-06, "loss": 1.143, "step": 1329 }, { "epoch": 0.1, "grad_norm": 4.8182726669345755, "learning_rate": 9.888489229448686e-06, "loss": 0.4115, "step": 1330 }, { "epoch": 0.1, "grad_norm": 24.756089917078956, "learning_rate": 9.888245202046509e-06, "loss": 1.6824, "step": 1331 }, { "epoch": 0.1, "grad_norm": 4.910112858362978, "learning_rate": 9.888000910942006e-06, "loss": 0.3385, "step": 1332 }, { "epoch": 0.1, "grad_norm": 20.066431704311352, "learning_rate": 9.887756356148359e-06, "loss": 1.0628, "step": 1333 }, { "epoch": 0.1, "grad_norm": 6.85523491891757, "learning_rate": 9.887511537678754e-06, "loss": 1.0606, "step": 1334 }, { "epoch": 0.1, "grad_norm": 12.428313522664938, "learning_rate": 9.887266455546404e-06, "loss": 1.4244, "step": 1335 }, { "epoch": 0.1, "grad_norm": 5.790346798564289, "learning_rate": 9.887021109764527e-06, "loss": 0.3336, "step": 1336 }, { "epoch": 0.1, "grad_norm": 5.950599238012521, "learning_rate": 9.886775500346362e-06, "loss": 0.3542, "step": 1337 }, { "epoch": 0.1, "grad_norm": 5.043588114477745, "learning_rate": 9.886529627305152e-06, "loss": 0.4066, "step": 1338 }, { "epoch": 0.1, "grad_norm": 5.592918685593219, "learning_rate": 9.88628349065417e-06, "loss": 0.3573, "step": 1339 }, { "epoch": 0.1, "grad_norm": 4.727108416655391, "learning_rate": 9.886037090406685e-06, "loss": 0.2942, "step": 1340 }, { "epoch": 0.1, "grad_norm": 6.676571072029495, "learning_rate": 9.885790426575996e-06, "loss": 1.3514, "step": 1341 }, { "epoch": 0.1, "grad_norm": 6.898387875818178, "learning_rate": 9.885543499175406e-06, "loss": 1.1127, "step": 1342 }, { "epoch": 0.1, "grad_norm": 26.073646052590473, "learning_rate": 9.885296308218238e-06, "loss": 1.756, "step": 1343 }, { "epoch": 0.1, "grad_norm": 5.561358621397341, "learning_rate": 9.885048853717825e-06, "loss": 0.3222, "step": 1344 }, { "epoch": 0.1, "grad_norm": 5.659254178641458, "learning_rate": 9.884801135687519e-06, "loss": 0.3753, "step": 1345 }, { "epoch": 0.1, "grad_norm": 46.4242721519916, "learning_rate": 9.88455315414068e-06, "loss": 1.7321, "step": 1346 }, { "epoch": 0.1, "grad_norm": 5.684214898851853, "learning_rate": 9.884304909090688e-06, "loss": 0.3525, "step": 1347 }, { "epoch": 0.1, "grad_norm": 6.396836766697932, "learning_rate": 9.884056400550934e-06, "loss": 1.2595, "step": 1348 }, { "epoch": 0.1, "grad_norm": 4.68320085206405, "learning_rate": 9.883807628534825e-06, "loss": 0.2474, "step": 1349 }, { "epoch": 0.1, "grad_norm": 16.980079174607532, "learning_rate": 9.88355859305578e-06, "loss": 1.9181, "step": 1350 }, { "epoch": 0.1, "grad_norm": 58.04351344432298, "learning_rate": 9.883309294127234e-06, "loss": 1.7086, "step": 1351 }, { "epoch": 0.1, "grad_norm": 5.464367049346169, "learning_rate": 9.883059731762637e-06, "loss": 0.3326, "step": 1352 }, { "epoch": 0.1, "grad_norm": 15.853757682894628, "learning_rate": 9.88280990597545e-06, "loss": 1.8081, "step": 1353 }, { "epoch": 0.1, "grad_norm": 5.325067937598506, "learning_rate": 9.882559816779152e-06, "loss": 0.3302, "step": 1354 }, { "epoch": 0.1, "grad_norm": 11.948476510401436, "learning_rate": 9.882309464187231e-06, "loss": 1.2746, "step": 1355 }, { "epoch": 0.1, "grad_norm": 5.1060136685275195, "learning_rate": 9.882058848213196e-06, "loss": 0.3652, "step": 1356 }, { "epoch": 0.1, "grad_norm": 4.5338220628758785, "learning_rate": 9.881807968870568e-06, "loss": 0.2868, "step": 1357 }, { "epoch": 0.1, "grad_norm": 6.275412158529945, "learning_rate": 9.881556826172875e-06, "loss": 1.1602, "step": 1358 }, { "epoch": 0.1, "grad_norm": 4.584273375856434, "learning_rate": 9.881305420133671e-06, "loss": 0.3502, "step": 1359 }, { "epoch": 0.1, "grad_norm": 23.037403901430746, "learning_rate": 9.881053750766517e-06, "loss": 1.9366, "step": 1360 }, { "epoch": 0.1, "grad_norm": 29.20259830983615, "learning_rate": 9.880801818084988e-06, "loss": 1.039, "step": 1361 }, { "epoch": 0.1, "grad_norm": 26.010550919283276, "learning_rate": 9.880549622102676e-06, "loss": 1.1591, "step": 1362 }, { "epoch": 0.1, "grad_norm": 4.6158962904133585, "learning_rate": 9.880297162833186e-06, "loss": 0.3047, "step": 1363 }, { "epoch": 0.1, "grad_norm": 27.585259372947437, "learning_rate": 9.880044440290136e-06, "loss": 2.1043, "step": 1364 }, { "epoch": 0.1, "grad_norm": 4.957247115129802, "learning_rate": 9.879791454487161e-06, "loss": 0.2522, "step": 1365 }, { "epoch": 0.1, "grad_norm": 5.619304533845806, "learning_rate": 9.879538205437909e-06, "loss": 0.3055, "step": 1366 }, { "epoch": 0.1, "grad_norm": 5.148346189607, "learning_rate": 9.87928469315604e-06, "loss": 0.2807, "step": 1367 }, { "epoch": 0.1, "grad_norm": 13.052044139859898, "learning_rate": 9.87903091765523e-06, "loss": 1.6371, "step": 1368 }, { "epoch": 0.1, "grad_norm": 21.993748308240807, "learning_rate": 9.878776878949171e-06, "loss": 2.1045, "step": 1369 }, { "epoch": 0.1, "grad_norm": 6.282276290444303, "learning_rate": 9.878522577051567e-06, "loss": 0.281, "step": 1370 }, { "epoch": 0.1, "grad_norm": 20.43024433803282, "learning_rate": 9.878268011976135e-06, "loss": 1.5659, "step": 1371 }, { "epoch": 0.1, "grad_norm": 7.35843985669414, "learning_rate": 9.87801318373661e-06, "loss": 1.2471, "step": 1372 }, { "epoch": 0.1, "grad_norm": 17.174632153392146, "learning_rate": 9.87775809234674e-06, "loss": 1.5793, "step": 1373 }, { "epoch": 0.1, "grad_norm": 7.306313001719329, "learning_rate": 9.877502737820279e-06, "loss": 1.0768, "step": 1374 }, { "epoch": 0.1, "grad_norm": 5.850947738384945, "learning_rate": 9.877247120171012e-06, "loss": 1.1336, "step": 1375 }, { "epoch": 0.1, "grad_norm": 4.554009128289451, "learning_rate": 9.876991239412723e-06, "loss": 0.3276, "step": 1376 }, { "epoch": 0.1, "grad_norm": 4.763471464187428, "learning_rate": 9.876735095559217e-06, "loss": 0.3609, "step": 1377 }, { "epoch": 0.1, "grad_norm": 12.019536749616996, "learning_rate": 9.87647868862431e-06, "loss": 0.9612, "step": 1378 }, { "epoch": 0.1, "grad_norm": 4.45747471939822, "learning_rate": 9.876222018621839e-06, "loss": 0.3057, "step": 1379 }, { "epoch": 0.1, "grad_norm": 4.219325490055066, "learning_rate": 9.875965085565648e-06, "loss": 0.2931, "step": 1380 }, { "epoch": 0.1, "grad_norm": 4.6016696626765015, "learning_rate": 9.875707889469595e-06, "loss": 0.3734, "step": 1381 }, { "epoch": 0.1, "grad_norm": 5.384818774413637, "learning_rate": 9.875450430347558e-06, "loss": 0.4056, "step": 1382 }, { "epoch": 0.1, "grad_norm": 6.857956341029196, "learning_rate": 9.875192708213425e-06, "loss": 1.081, "step": 1383 }, { "epoch": 0.1, "grad_norm": 5.6209250362636505, "learning_rate": 9.874934723081099e-06, "loss": 0.3656, "step": 1384 }, { "epoch": 0.1, "grad_norm": 4.054143942782579, "learning_rate": 9.874676474964497e-06, "loss": 0.3515, "step": 1385 }, { "epoch": 0.1, "grad_norm": 6.685589590670473, "learning_rate": 9.87441796387755e-06, "loss": 1.1409, "step": 1386 }, { "epoch": 0.1, "grad_norm": 4.737923752279675, "learning_rate": 9.874159189834206e-06, "loss": 0.3699, "step": 1387 }, { "epoch": 0.1, "grad_norm": 18.3550220956042, "learning_rate": 9.873900152848424e-06, "loss": 1.0443, "step": 1388 }, { "epoch": 0.1, "grad_norm": 4.872795167869886, "learning_rate": 9.873640852934176e-06, "loss": 0.297, "step": 1389 }, { "epoch": 0.1, "grad_norm": 5.056199930779924, "learning_rate": 9.873381290105451e-06, "loss": 0.2897, "step": 1390 }, { "epoch": 0.1, "grad_norm": 6.421893751745245, "learning_rate": 9.873121464376255e-06, "loss": 0.8992, "step": 1391 }, { "epoch": 0.1, "grad_norm": 4.638568698029766, "learning_rate": 9.872861375760598e-06, "loss": 0.2792, "step": 1392 }, { "epoch": 0.1, "grad_norm": 42.13755170095363, "learning_rate": 9.872601024272517e-06, "loss": 1.2021, "step": 1393 }, { "epoch": 0.1, "grad_norm": 6.488601995410183, "learning_rate": 9.872340409926053e-06, "loss": 0.4034, "step": 1394 }, { "epoch": 0.1, "grad_norm": 3.63745395650679, "learning_rate": 9.872079532735268e-06, "loss": 0.1997, "step": 1395 }, { "epoch": 0.1, "grad_norm": 5.84268263873479, "learning_rate": 9.871818392714234e-06, "loss": 0.3585, "step": 1396 }, { "epoch": 0.1, "grad_norm": 13.59203324967616, "learning_rate": 9.871556989877038e-06, "loss": 1.5026, "step": 1397 }, { "epoch": 0.1, "grad_norm": 22.34363087868927, "learning_rate": 9.871295324237783e-06, "loss": 1.179, "step": 1398 }, { "epoch": 0.1, "grad_norm": 4.1350219286699295, "learning_rate": 9.871033395810583e-06, "loss": 0.2581, "step": 1399 }, { "epoch": 0.1, "grad_norm": 5.68043936350287, "learning_rate": 9.870771204609571e-06, "loss": 0.3879, "step": 1400 }, { "epoch": 0.1, "grad_norm": 6.3387264331396524, "learning_rate": 9.870508750648887e-06, "loss": 1.2051, "step": 1401 }, { "epoch": 0.1, "grad_norm": 4.516809135644811, "learning_rate": 9.870246033942694e-06, "loss": 0.3125, "step": 1402 }, { "epoch": 0.1, "grad_norm": 4.36183669281403, "learning_rate": 9.869983054505161e-06, "loss": 0.3192, "step": 1403 }, { "epoch": 0.1, "grad_norm": 14.044488859275376, "learning_rate": 9.869719812350475e-06, "loss": 1.0833, "step": 1404 }, { "epoch": 0.1, "grad_norm": 11.441801757229044, "learning_rate": 9.86945630749284e-06, "loss": 0.985, "step": 1405 }, { "epoch": 0.1, "grad_norm": 5.882904808233509, "learning_rate": 9.869192539946468e-06, "loss": 1.0125, "step": 1406 }, { "epoch": 0.1, "grad_norm": 5.029420465912323, "learning_rate": 9.868928509725591e-06, "loss": 0.4163, "step": 1407 }, { "epoch": 0.1, "grad_norm": 30.636311405318533, "learning_rate": 9.86866421684445e-06, "loss": 1.4581, "step": 1408 }, { "epoch": 0.1, "grad_norm": 19.216213932496327, "learning_rate": 9.868399661317303e-06, "loss": 1.4564, "step": 1409 }, { "epoch": 0.1, "grad_norm": 4.4064629515362705, "learning_rate": 9.868134843158421e-06, "loss": 0.3178, "step": 1410 }, { "epoch": 0.1, "grad_norm": 5.999185370996733, "learning_rate": 9.867869762382093e-06, "loss": 1.3086, "step": 1411 }, { "epoch": 0.1, "grad_norm": 4.459184976672309, "learning_rate": 9.867604419002616e-06, "loss": 0.2299, "step": 1412 }, { "epoch": 0.1, "grad_norm": 41.22627156736703, "learning_rate": 9.867338813034307e-06, "loss": 1.073, "step": 1413 }, { "epoch": 0.1, "grad_norm": 24.679293615004948, "learning_rate": 9.867072944491492e-06, "loss": 1.1923, "step": 1414 }, { "epoch": 0.1, "grad_norm": 62.26540115917003, "learning_rate": 9.866806813388514e-06, "loss": 1.4362, "step": 1415 }, { "epoch": 0.1, "grad_norm": 5.078503646331537, "learning_rate": 9.86654041973973e-06, "loss": 0.2778, "step": 1416 }, { "epoch": 0.1, "grad_norm": 43.147692979758794, "learning_rate": 9.866273763559512e-06, "loss": 1.1343, "step": 1417 }, { "epoch": 0.1, "grad_norm": 132.4568427231583, "learning_rate": 9.866006844862246e-06, "loss": 1.3682, "step": 1418 }, { "epoch": 0.1, "grad_norm": 37.67420435242833, "learning_rate": 9.865739663662326e-06, "loss": 1.2453, "step": 1419 }, { "epoch": 0.1, "grad_norm": 5.2259211539443635, "learning_rate": 9.86547221997417e-06, "loss": 0.3631, "step": 1420 }, { "epoch": 0.1, "grad_norm": 30.061495619690408, "learning_rate": 9.865204513812208e-06, "loss": 1.9703, "step": 1421 }, { "epoch": 0.1, "grad_norm": 6.423781905553708, "learning_rate": 9.864936545190876e-06, "loss": 1.0404, "step": 1422 }, { "epoch": 0.1, "grad_norm": 46.71335659050631, "learning_rate": 9.86466831412463e-06, "loss": 2.0088, "step": 1423 }, { "epoch": 0.1, "grad_norm": 20.013186761857337, "learning_rate": 9.864399820627944e-06, "loss": 1.5016, "step": 1424 }, { "epoch": 0.1, "grad_norm": 6.209219186562817, "learning_rate": 9.864131064715302e-06, "loss": 1.2177, "step": 1425 }, { "epoch": 0.1, "grad_norm": 4.529105743075609, "learning_rate": 9.863862046401198e-06, "loss": 0.289, "step": 1426 }, { "epoch": 0.1, "grad_norm": 35.97369799964603, "learning_rate": 9.863592765700149e-06, "loss": 1.1767, "step": 1427 }, { "epoch": 0.1, "grad_norm": 5.932141790366115, "learning_rate": 9.86332322262668e-06, "loss": 1.0334, "step": 1428 }, { "epoch": 0.1, "grad_norm": 4.163359964839368, "learning_rate": 9.86305341719533e-06, "loss": 0.2239, "step": 1429 }, { "epoch": 0.1, "grad_norm": 5.329239809398957, "learning_rate": 9.862783349420657e-06, "loss": 0.3301, "step": 1430 }, { "epoch": 0.1, "grad_norm": 4.421857333362876, "learning_rate": 9.86251301931723e-06, "loss": 0.2568, "step": 1431 }, { "epoch": 0.1, "grad_norm": 4.574716655784341, "learning_rate": 9.86224242689963e-06, "loss": 0.2252, "step": 1432 }, { "epoch": 0.1, "grad_norm": 5.363801588386903, "learning_rate": 9.861971572182456e-06, "loss": 0.3499, "step": 1433 }, { "epoch": 0.1, "grad_norm": 20.94547495152175, "learning_rate": 9.86170045518032e-06, "loss": 1.4086, "step": 1434 }, { "epoch": 0.1, "grad_norm": 11.83501347352298, "learning_rate": 9.861429075907845e-06, "loss": 1.3248, "step": 1435 }, { "epoch": 0.1, "grad_norm": 4.344101825817167, "learning_rate": 9.861157434379673e-06, "loss": 0.2533, "step": 1436 }, { "epoch": 0.1, "grad_norm": 8.366757263912225, "learning_rate": 9.86088553061046e-06, "loss": 0.4044, "step": 1437 }, { "epoch": 0.1, "grad_norm": 33.59606782619112, "learning_rate": 9.86061336461487e-06, "loss": 1.4213, "step": 1438 }, { "epoch": 0.1, "grad_norm": 5.302834116987351, "learning_rate": 9.860340936407589e-06, "loss": 0.3034, "step": 1439 }, { "epoch": 0.1, "grad_norm": 17.63278689273178, "learning_rate": 9.860068246003311e-06, "loss": 0.9845, "step": 1440 }, { "epoch": 0.1, "grad_norm": 23.838813853620003, "learning_rate": 9.859795293416748e-06, "loss": 1.7426, "step": 1441 }, { "epoch": 0.1, "grad_norm": 10.031071443261364, "learning_rate": 9.859522078662625e-06, "loss": 0.7877, "step": 1442 }, { "epoch": 0.1, "grad_norm": 12.95981137611716, "learning_rate": 9.859248601755679e-06, "loss": 2.1613, "step": 1443 }, { "epoch": 0.1, "grad_norm": 4.812761535411927, "learning_rate": 9.858974862710664e-06, "loss": 0.2825, "step": 1444 }, { "epoch": 0.1, "grad_norm": 20.491398793599497, "learning_rate": 9.858700861542348e-06, "loss": 1.7268, "step": 1445 }, { "epoch": 0.1, "grad_norm": 12.223218149090828, "learning_rate": 9.858426598265511e-06, "loss": 1.2715, "step": 1446 }, { "epoch": 0.1, "grad_norm": 4.742308984936539, "learning_rate": 9.85815207289495e-06, "loss": 0.4204, "step": 1447 }, { "epoch": 0.1, "grad_norm": 8.702807231600783, "learning_rate": 9.857877285445473e-06, "loss": 0.8975, "step": 1448 }, { "epoch": 0.1, "grad_norm": 3.508502071409034, "learning_rate": 9.857602235931904e-06, "loss": 0.2629, "step": 1449 }, { "epoch": 0.1, "grad_norm": 5.016806360437113, "learning_rate": 9.857326924369082e-06, "loss": 0.3244, "step": 1450 }, { "epoch": 0.1, "grad_norm": 4.099873145723555, "learning_rate": 9.85705135077186e-06, "loss": 0.354, "step": 1451 }, { "epoch": 0.1, "grad_norm": 20.68859621573412, "learning_rate": 9.856775515155101e-06, "loss": 1.2943, "step": 1452 }, { "epoch": 0.1, "grad_norm": 71.8438487923445, "learning_rate": 9.856499417533687e-06, "loss": 1.2604, "step": 1453 }, { "epoch": 0.1, "grad_norm": 5.841615786612452, "learning_rate": 9.856223057922512e-06, "loss": 1.0565, "step": 1454 }, { "epoch": 0.1, "grad_norm": 4.436352514979919, "learning_rate": 9.855946436336486e-06, "loss": 0.2744, "step": 1455 }, { "epoch": 0.1, "grad_norm": 4.196633112142781, "learning_rate": 9.85566955279053e-06, "loss": 0.2877, "step": 1456 }, { "epoch": 0.1, "grad_norm": 13.600560286933629, "learning_rate": 9.85539240729958e-06, "loss": 1.16, "step": 1457 }, { "epoch": 0.1, "grad_norm": 19.1653296568493, "learning_rate": 9.85511499987859e-06, "loss": 1.3801, "step": 1458 }, { "epoch": 0.1, "grad_norm": 4.885736282890637, "learning_rate": 9.854837330542521e-06, "loss": 0.3162, "step": 1459 }, { "epoch": 0.1, "grad_norm": 13.306617206904662, "learning_rate": 9.854559399306358e-06, "loss": 0.8373, "step": 1460 }, { "epoch": 0.1, "grad_norm": 23.80051992231294, "learning_rate": 9.854281206185088e-06, "loss": 1.7506, "step": 1461 }, { "epoch": 0.1, "grad_norm": 23.128383773873942, "learning_rate": 9.854002751193724e-06, "loss": 1.5354, "step": 1462 }, { "epoch": 0.1, "grad_norm": 4.264844399753477, "learning_rate": 9.853724034347283e-06, "loss": 0.2569, "step": 1463 }, { "epoch": 0.1, "grad_norm": 5.069033882711562, "learning_rate": 9.853445055660803e-06, "loss": 0.4192, "step": 1464 }, { "epoch": 0.11, "grad_norm": 5.19359118208667, "learning_rate": 9.853165815149333e-06, "loss": 0.4709, "step": 1465 }, { "epoch": 0.11, "grad_norm": 16.92734794388201, "learning_rate": 9.852886312827938e-06, "loss": 2.2453, "step": 1466 }, { "epoch": 0.11, "grad_norm": 6.100021256683218, "learning_rate": 9.852606548711697e-06, "loss": 0.9647, "step": 1467 }, { "epoch": 0.11, "grad_norm": 5.3806783403153275, "learning_rate": 9.8523265228157e-06, "loss": 0.3187, "step": 1468 }, { "epoch": 0.11, "grad_norm": 5.273146628064586, "learning_rate": 9.852046235155053e-06, "loss": 0.3432, "step": 1469 }, { "epoch": 0.11, "grad_norm": 4.8817055324808525, "learning_rate": 9.85176568574488e-06, "loss": 0.2197, "step": 1470 }, { "epoch": 0.11, "grad_norm": 18.662363777493137, "learning_rate": 9.85148487460031e-06, "loss": 1.2295, "step": 1471 }, { "epoch": 0.11, "grad_norm": 28.93167793330691, "learning_rate": 9.851203801736496e-06, "loss": 2.0028, "step": 1472 }, { "epoch": 0.11, "grad_norm": 4.505832912078425, "learning_rate": 9.8509224671686e-06, "loss": 0.1828, "step": 1473 }, { "epoch": 0.11, "grad_norm": 27.27503517993237, "learning_rate": 9.8506408709118e-06, "loss": 1.9235, "step": 1474 }, { "epoch": 0.11, "grad_norm": 5.560514706627873, "learning_rate": 9.850359012981284e-06, "loss": 0.4062, "step": 1475 }, { "epoch": 0.11, "grad_norm": 6.803645897992183, "learning_rate": 9.85007689339226e-06, "loss": 1.2437, "step": 1476 }, { "epoch": 0.11, "grad_norm": 5.347878371964562, "learning_rate": 9.849794512159944e-06, "loss": 0.3022, "step": 1477 }, { "epoch": 0.11, "grad_norm": 4.9740229597277965, "learning_rate": 9.849511869299574e-06, "loss": 0.3783, "step": 1478 }, { "epoch": 0.11, "grad_norm": 5.4114252315944364, "learning_rate": 9.849228964826394e-06, "loss": 0.3966, "step": 1479 }, { "epoch": 0.11, "grad_norm": 16.23008842669428, "learning_rate": 9.848945798755667e-06, "loss": 2.1039, "step": 1480 }, { "epoch": 0.11, "grad_norm": 4.9088839444376875, "learning_rate": 9.848662371102668e-06, "loss": 0.399, "step": 1481 }, { "epoch": 0.11, "grad_norm": 4.452486351231861, "learning_rate": 9.848378681882687e-06, "loss": 0.2983, "step": 1482 }, { "epoch": 0.11, "grad_norm": 4.913187477849116, "learning_rate": 9.848094731111028e-06, "loss": 0.3487, "step": 1483 }, { "epoch": 0.11, "grad_norm": 4.998405054633763, "learning_rate": 9.847810518803011e-06, "loss": 0.3672, "step": 1484 }, { "epoch": 0.11, "grad_norm": 11.17664466471127, "learning_rate": 9.847526044973964e-06, "loss": 1.0794, "step": 1485 }, { "epoch": 0.11, "grad_norm": 6.675598079210604, "learning_rate": 9.847241309639238e-06, "loss": 1.1942, "step": 1486 }, { "epoch": 0.11, "grad_norm": 4.23373600131044, "learning_rate": 9.84695631281419e-06, "loss": 0.3374, "step": 1487 }, { "epoch": 0.11, "grad_norm": 4.933958690993196, "learning_rate": 9.846671054514196e-06, "loss": 0.3581, "step": 1488 }, { "epoch": 0.11, "grad_norm": 4.11600524387837, "learning_rate": 9.846385534754646e-06, "loss": 0.3061, "step": 1489 }, { "epoch": 0.11, "grad_norm": 5.723165147192148, "learning_rate": 9.846099753550938e-06, "loss": 1.1092, "step": 1490 }, { "epoch": 0.11, "grad_norm": 3.9830723867786917, "learning_rate": 9.845813710918492e-06, "loss": 0.2921, "step": 1491 }, { "epoch": 0.11, "grad_norm": 22.030064497855893, "learning_rate": 9.845527406872742e-06, "loss": 1.2318, "step": 1492 }, { "epoch": 0.11, "grad_norm": 17.709662115154803, "learning_rate": 9.845240841429126e-06, "loss": 1.6123, "step": 1493 }, { "epoch": 0.11, "grad_norm": 23.10591655306058, "learning_rate": 9.844954014603108e-06, "loss": 1.649, "step": 1494 }, { "epoch": 0.11, "grad_norm": 16.078875855297788, "learning_rate": 9.844666926410161e-06, "loss": 1.1141, "step": 1495 }, { "epoch": 0.11, "grad_norm": 4.242917251661823, "learning_rate": 9.84437957686577e-06, "loss": 0.2749, "step": 1496 }, { "epoch": 0.11, "grad_norm": 3.7806055039366395, "learning_rate": 9.844091965985438e-06, "loss": 0.3143, "step": 1497 }, { "epoch": 0.11, "grad_norm": 16.94659989892063, "learning_rate": 9.843804093784682e-06, "loss": 1.2417, "step": 1498 }, { "epoch": 0.11, "grad_norm": 22.090153932918483, "learning_rate": 9.843515960279029e-06, "loss": 1.8486, "step": 1499 }, { "epoch": 0.11, "grad_norm": 22.18293053739052, "learning_rate": 9.843227565484022e-06, "loss": 0.3447, "step": 1500 }, { "epoch": 0.11, "grad_norm": 13.879885656635574, "learning_rate": 9.842938909415222e-06, "loss": 1.7577, "step": 1501 }, { "epoch": 0.11, "grad_norm": 150.653103203567, "learning_rate": 9.842649992088199e-06, "loss": 0.5913, "step": 1502 }, { "epoch": 0.11, "grad_norm": 6.750097050529787, "learning_rate": 9.842360813518538e-06, "loss": 1.2376, "step": 1503 }, { "epoch": 0.11, "grad_norm": 9.713064425678432, "learning_rate": 9.842071373721844e-06, "loss": 0.3856, "step": 1504 }, { "epoch": 0.11, "grad_norm": 4.783031999022843, "learning_rate": 9.841781672713725e-06, "loss": 0.268, "step": 1505 }, { "epoch": 0.11, "grad_norm": 23.511295040802146, "learning_rate": 9.841491710509813e-06, "loss": 1.4474, "step": 1506 }, { "epoch": 0.11, "grad_norm": 46.69761410213684, "learning_rate": 9.841201487125749e-06, "loss": 1.8481, "step": 1507 }, { "epoch": 0.11, "grad_norm": 84.60529131039323, "learning_rate": 9.840911002577189e-06, "loss": 3.3148, "step": 1508 }, { "epoch": 0.11, "grad_norm": 14.202858256558043, "learning_rate": 9.840620256879806e-06, "loss": 0.5852, "step": 1509 }, { "epoch": 0.11, "grad_norm": 11.457733835449007, "learning_rate": 9.840329250049282e-06, "loss": 0.7554, "step": 1510 }, { "epoch": 0.11, "grad_norm": 12.7626153456412, "learning_rate": 9.840037982101314e-06, "loss": 1.0149, "step": 1511 }, { "epoch": 0.11, "grad_norm": 17.441192174596047, "learning_rate": 9.839746453051619e-06, "loss": 1.2469, "step": 1512 }, { "epoch": 0.11, "grad_norm": 5.590140273683809, "learning_rate": 9.839454662915923e-06, "loss": 1.328, "step": 1513 }, { "epoch": 0.11, "grad_norm": 4.704952244452666, "learning_rate": 9.839162611709968e-06, "loss": 0.2662, "step": 1514 }, { "epoch": 0.11, "grad_norm": 25.924819418792634, "learning_rate": 9.838870299449504e-06, "loss": 1.5872, "step": 1515 }, { "epoch": 0.11, "grad_norm": 5.498179859135677, "learning_rate": 9.838577726150305e-06, "loss": 0.3311, "step": 1516 }, { "epoch": 0.11, "grad_norm": 4.842836270894532, "learning_rate": 9.838284891828154e-06, "loss": 0.3692, "step": 1517 }, { "epoch": 0.11, "grad_norm": 18.739482528368853, "learning_rate": 9.837991796498845e-06, "loss": 1.1884, "step": 1518 }, { "epoch": 0.11, "grad_norm": 15.99550765445871, "learning_rate": 9.837698440178191e-06, "loss": 1.844, "step": 1519 }, { "epoch": 0.11, "grad_norm": 22.211055748304226, "learning_rate": 9.83740482288202e-06, "loss": 1.5663, "step": 1520 }, { "epoch": 0.11, "grad_norm": 22.291371407930647, "learning_rate": 9.837110944626168e-06, "loss": 1.2775, "step": 1521 }, { "epoch": 0.11, "grad_norm": 11.434626868305536, "learning_rate": 9.83681680542649e-06, "loss": 1.1583, "step": 1522 }, { "epoch": 0.11, "grad_norm": 4.0063171499273444, "learning_rate": 9.836522405298853e-06, "loss": 0.283, "step": 1523 }, { "epoch": 0.11, "grad_norm": 6.045492063524387, "learning_rate": 9.836227744259141e-06, "loss": 1.0052, "step": 1524 }, { "epoch": 0.11, "grad_norm": 4.820692424601162, "learning_rate": 9.835932822323247e-06, "loss": 0.2873, "step": 1525 }, { "epoch": 0.11, "grad_norm": 4.88952560871193, "learning_rate": 9.835637639507082e-06, "loss": 0.2839, "step": 1526 }, { "epoch": 0.11, "grad_norm": 5.989247916425711, "learning_rate": 9.835342195826572e-06, "loss": 1.132, "step": 1527 }, { "epoch": 0.11, "grad_norm": 4.7847622962418415, "learning_rate": 9.835046491297653e-06, "loss": 0.3579, "step": 1528 }, { "epoch": 0.11, "grad_norm": 6.36738838989337, "learning_rate": 9.834750525936277e-06, "loss": 1.234, "step": 1529 }, { "epoch": 0.11, "grad_norm": 6.005236066837438, "learning_rate": 9.83445429975841e-06, "loss": 0.2991, "step": 1530 }, { "epoch": 0.11, "grad_norm": 26.55214373438119, "learning_rate": 9.834157812780035e-06, "loss": 1.8473, "step": 1531 }, { "epoch": 0.11, "grad_norm": 5.670390997634389, "learning_rate": 9.833861065017142e-06, "loss": 0.3601, "step": 1532 }, { "epoch": 0.11, "grad_norm": 20.07877998341654, "learning_rate": 9.833564056485742e-06, "loss": 1.7294, "step": 1533 }, { "epoch": 0.11, "grad_norm": 6.992298809724057, "learning_rate": 9.833266787201858e-06, "loss": 1.0212, "step": 1534 }, { "epoch": 0.11, "grad_norm": 19.299719082800426, "learning_rate": 9.832969257181525e-06, "loss": 1.0593, "step": 1535 }, { "epoch": 0.11, "grad_norm": 31.15028309697567, "learning_rate": 9.832671466440795e-06, "loss": 1.4423, "step": 1536 }, { "epoch": 0.11, "grad_norm": 7.012396565820055, "learning_rate": 9.83237341499573e-06, "loss": 1.0369, "step": 1537 }, { "epoch": 0.11, "grad_norm": 5.327756459883463, "learning_rate": 9.832075102862414e-06, "loss": 0.9199, "step": 1538 }, { "epoch": 0.11, "grad_norm": 5.3959572512890315, "learning_rate": 9.831776530056936e-06, "loss": 0.3778, "step": 1539 }, { "epoch": 0.11, "grad_norm": 12.8141570073646, "learning_rate": 9.831477696595403e-06, "loss": 1.2423, "step": 1540 }, { "epoch": 0.11, "grad_norm": 4.173228378415182, "learning_rate": 9.831178602493935e-06, "loss": 0.2588, "step": 1541 }, { "epoch": 0.11, "grad_norm": 5.755634514485555, "learning_rate": 9.830879247768669e-06, "loss": 0.3659, "step": 1542 }, { "epoch": 0.11, "grad_norm": 4.021154331860064, "learning_rate": 9.830579632435754e-06, "loss": 0.2552, "step": 1543 }, { "epoch": 0.11, "grad_norm": 5.4076428414296505, "learning_rate": 9.830279756511352e-06, "loss": 1.2255, "step": 1544 }, { "epoch": 0.11, "grad_norm": 31.815116356789225, "learning_rate": 9.829979620011641e-06, "loss": 2.0372, "step": 1545 }, { "epoch": 0.11, "grad_norm": 5.624485435736779, "learning_rate": 9.829679222952813e-06, "loss": 0.2866, "step": 1546 }, { "epoch": 0.11, "grad_norm": 31.6771344229947, "learning_rate": 9.829378565351071e-06, "loss": 1.6704, "step": 1547 }, { "epoch": 0.11, "grad_norm": 6.545343402100654, "learning_rate": 9.829077647222636e-06, "loss": 1.0901, "step": 1548 }, { "epoch": 0.11, "grad_norm": 6.276884690760708, "learning_rate": 9.82877646858374e-06, "loss": 0.9723, "step": 1549 }, { "epoch": 0.11, "grad_norm": 5.07606277169552, "learning_rate": 9.828475029450633e-06, "loss": 0.2483, "step": 1550 }, { "epoch": 0.11, "grad_norm": 5.443378494591427, "learning_rate": 9.828173329839574e-06, "loss": 0.3329, "step": 1551 }, { "epoch": 0.11, "grad_norm": 10.940032604545681, "learning_rate": 9.827871369766841e-06, "loss": 1.103, "step": 1552 }, { "epoch": 0.11, "grad_norm": 4.179283887848109, "learning_rate": 9.82756914924872e-06, "loss": 0.3037, "step": 1553 }, { "epoch": 0.11, "grad_norm": 25.840059663685448, "learning_rate": 9.827266668301518e-06, "loss": 1.7104, "step": 1554 }, { "epoch": 0.11, "grad_norm": 6.5196662587163114, "learning_rate": 9.82696392694155e-06, "loss": 1.1843, "step": 1555 }, { "epoch": 0.11, "grad_norm": 13.725853134451993, "learning_rate": 9.826660925185151e-06, "loss": 1.1158, "step": 1556 }, { "epoch": 0.11, "grad_norm": 11.107126898448891, "learning_rate": 9.826357663048665e-06, "loss": 1.2987, "step": 1557 }, { "epoch": 0.11, "grad_norm": 5.28472454365378, "learning_rate": 9.826054140548451e-06, "loss": 0.9569, "step": 1558 }, { "epoch": 0.11, "grad_norm": 12.382953972356116, "learning_rate": 9.825750357700884e-06, "loss": 1.2803, "step": 1559 }, { "epoch": 0.11, "grad_norm": 9.380207107892963, "learning_rate": 9.825446314522351e-06, "loss": 1.2825, "step": 1560 }, { "epoch": 0.11, "grad_norm": 5.270037144579977, "learning_rate": 9.825142011029255e-06, "loss": 0.2464, "step": 1561 }, { "epoch": 0.11, "grad_norm": 13.07583673297471, "learning_rate": 9.824837447238012e-06, "loss": 0.4906, "step": 1562 }, { "epoch": 0.11, "grad_norm": 6.018398599654464, "learning_rate": 9.824532623165051e-06, "loss": 1.0292, "step": 1563 }, { "epoch": 0.11, "grad_norm": 21.196598569492515, "learning_rate": 9.824227538826818e-06, "loss": 1.5482, "step": 1564 }, { "epoch": 0.11, "grad_norm": 6.7365057175812035, "learning_rate": 9.823922194239769e-06, "loss": 0.9293, "step": 1565 }, { "epoch": 0.11, "grad_norm": 11.896201869086635, "learning_rate": 9.823616589420378e-06, "loss": 0.9124, "step": 1566 }, { "epoch": 0.11, "grad_norm": 18.480193360433482, "learning_rate": 9.823310724385128e-06, "loss": 1.0347, "step": 1567 }, { "epoch": 0.11, "grad_norm": 4.958502695685892, "learning_rate": 9.823004599150525e-06, "loss": 0.2645, "step": 1568 }, { "epoch": 0.11, "grad_norm": 4.370956441008158, "learning_rate": 9.822698213733077e-06, "loss": 0.293, "step": 1569 }, { "epoch": 0.11, "grad_norm": 24.39833936700523, "learning_rate": 9.822391568149318e-06, "loss": 1.0333, "step": 1570 }, { "epoch": 0.11, "grad_norm": 6.428175619000391, "learning_rate": 9.822084662415786e-06, "loss": 0.9403, "step": 1571 }, { "epoch": 0.11, "grad_norm": 5.5559901770512035, "learning_rate": 9.821777496549039e-06, "loss": 0.3613, "step": 1572 }, { "epoch": 0.11, "grad_norm": 4.821486527020876, "learning_rate": 9.821470070565648e-06, "loss": 0.3507, "step": 1573 }, { "epoch": 0.11, "grad_norm": 5.038189956701842, "learning_rate": 9.821162384482196e-06, "loss": 0.3824, "step": 1574 }, { "epoch": 0.11, "grad_norm": 24.139310479210764, "learning_rate": 9.820854438315283e-06, "loss": 1.4192, "step": 1575 }, { "epoch": 0.11, "grad_norm": 3.912142866149468, "learning_rate": 9.820546232081523e-06, "loss": 0.3036, "step": 1576 }, { "epoch": 0.11, "grad_norm": 13.809840183843997, "learning_rate": 9.820237765797538e-06, "loss": 1.3464, "step": 1577 }, { "epoch": 0.11, "grad_norm": 4.143345328155439, "learning_rate": 9.819929039479972e-06, "loss": 0.2726, "step": 1578 }, { "epoch": 0.11, "grad_norm": 4.836772723092644, "learning_rate": 9.819620053145478e-06, "loss": 0.3164, "step": 1579 }, { "epoch": 0.11, "grad_norm": 4.542443081001037, "learning_rate": 9.819310806810727e-06, "loss": 0.33, "step": 1580 }, { "epoch": 0.11, "grad_norm": 4.914537296041724, "learning_rate": 9.819001300492398e-06, "loss": 0.23, "step": 1581 }, { "epoch": 0.11, "grad_norm": 31.573658621095145, "learning_rate": 9.818691534207192e-06, "loss": 1.4051, "step": 1582 }, { "epoch": 0.11, "grad_norm": 20.829546775440114, "learning_rate": 9.818381507971815e-06, "loss": 1.1823, "step": 1583 }, { "epoch": 0.11, "grad_norm": 21.935615785976026, "learning_rate": 9.818071221802996e-06, "loss": 2.3178, "step": 1584 }, { "epoch": 0.11, "grad_norm": 5.49847246133438, "learning_rate": 9.81776067571747e-06, "loss": 0.4049, "step": 1585 }, { "epoch": 0.11, "grad_norm": 5.51701690084627, "learning_rate": 9.817449869731995e-06, "loss": 0.373, "step": 1586 }, { "epoch": 0.11, "grad_norm": 5.158742090211531, "learning_rate": 9.817138803863332e-06, "loss": 0.4371, "step": 1587 }, { "epoch": 0.11, "grad_norm": 6.074356634337482, "learning_rate": 9.816827478128265e-06, "loss": 1.0197, "step": 1588 }, { "epoch": 0.11, "grad_norm": 17.09046211372591, "learning_rate": 9.81651589254359e-06, "loss": 0.9033, "step": 1589 }, { "epoch": 0.11, "grad_norm": 7.952593853191333, "learning_rate": 9.816204047126111e-06, "loss": 1.2145, "step": 1590 }, { "epoch": 0.11, "grad_norm": 17.96610651158857, "learning_rate": 9.815891941892657e-06, "loss": 1.2969, "step": 1591 }, { "epoch": 0.11, "grad_norm": 16.725698294270774, "learning_rate": 9.81557957686006e-06, "loss": 0.9907, "step": 1592 }, { "epoch": 0.11, "grad_norm": 4.181037355902699, "learning_rate": 9.815266952045173e-06, "loss": 0.2462, "step": 1593 }, { "epoch": 0.11, "grad_norm": 6.101769406894543, "learning_rate": 9.81495406746486e-06, "loss": 1.0752, "step": 1594 }, { "epoch": 0.11, "grad_norm": 4.227895830486115, "learning_rate": 9.814640923136003e-06, "loss": 0.3383, "step": 1595 }, { "epoch": 0.11, "grad_norm": 68.63729338523854, "learning_rate": 9.81432751907549e-06, "loss": 1.1707, "step": 1596 }, { "epoch": 0.11, "grad_norm": 4.782363750994288, "learning_rate": 9.814013855300231e-06, "loss": 0.2671, "step": 1597 }, { "epoch": 0.11, "grad_norm": 5.102851235701723, "learning_rate": 9.813699931827147e-06, "loss": 0.2883, "step": 1598 }, { "epoch": 0.11, "grad_norm": 6.453811899740123, "learning_rate": 9.813385748673175e-06, "loss": 0.9477, "step": 1599 }, { "epoch": 0.11, "grad_norm": 20.461559371281655, "learning_rate": 9.813071305855259e-06, "loss": 1.0259, "step": 1600 }, { "epoch": 0.11, "grad_norm": 5.198737948784659, "learning_rate": 9.812756603390362e-06, "loss": 0.2361, "step": 1601 }, { "epoch": 0.11, "grad_norm": 3.6837415758341043, "learning_rate": 9.812441641295468e-06, "loss": 0.2733, "step": 1602 }, { "epoch": 0.11, "grad_norm": 4.514740885856748, "learning_rate": 9.812126419587562e-06, "loss": 0.3905, "step": 1603 }, { "epoch": 0.12, "grad_norm": 8.423160465896782, "learning_rate": 9.81181093828365e-06, "loss": 1.0962, "step": 1604 }, { "epoch": 0.12, "grad_norm": 5.185946900213324, "learning_rate": 9.811495197400753e-06, "loss": 0.3093, "step": 1605 }, { "epoch": 0.12, "grad_norm": 9.933524519318546, "learning_rate": 9.811179196955902e-06, "loss": 0.7541, "step": 1606 }, { "epoch": 0.12, "grad_norm": 13.069429988855928, "learning_rate": 9.810862936966144e-06, "loss": 1.4568, "step": 1607 }, { "epoch": 0.12, "grad_norm": 4.522214996857623, "learning_rate": 9.81054641744854e-06, "loss": 0.3023, "step": 1608 }, { "epoch": 0.12, "grad_norm": 3.849248257456824, "learning_rate": 9.810229638420166e-06, "loss": 0.2731, "step": 1609 }, { "epoch": 0.12, "grad_norm": 3.9135476321869698, "learning_rate": 9.809912599898112e-06, "loss": 0.2522, "step": 1610 }, { "epoch": 0.12, "grad_norm": 6.34285558021622, "learning_rate": 9.80959530189948e-06, "loss": 0.3228, "step": 1611 }, { "epoch": 0.12, "grad_norm": 19.148041064861303, "learning_rate": 9.809277744441384e-06, "loss": 1.752, "step": 1612 }, { "epoch": 0.12, "grad_norm": 4.6721884918454055, "learning_rate": 9.808959927540959e-06, "loss": 0.2882, "step": 1613 }, { "epoch": 0.12, "grad_norm": 18.69819155124576, "learning_rate": 9.80864185121535e-06, "loss": 1.1317, "step": 1614 }, { "epoch": 0.12, "grad_norm": 16.31302450685389, "learning_rate": 9.808323515481714e-06, "loss": 1.5158, "step": 1615 }, { "epoch": 0.12, "grad_norm": 4.209649989593727, "learning_rate": 9.808004920357226e-06, "loss": 0.2652, "step": 1616 }, { "epoch": 0.12, "grad_norm": 5.216616760305834, "learning_rate": 9.807686065859071e-06, "loss": 0.3171, "step": 1617 }, { "epoch": 0.12, "grad_norm": 4.702350145051927, "learning_rate": 9.80736695200445e-06, "loss": 0.3662, "step": 1618 }, { "epoch": 0.12, "grad_norm": 9.31870520148133, "learning_rate": 9.80704757881058e-06, "loss": 1.2838, "step": 1619 }, { "epoch": 0.12, "grad_norm": 3.8469541181382376, "learning_rate": 9.806727946294688e-06, "loss": 0.2492, "step": 1620 }, { "epoch": 0.12, "grad_norm": 5.721183142063519, "learning_rate": 9.806408054474017e-06, "loss": 1.2157, "step": 1621 }, { "epoch": 0.12, "grad_norm": 4.157439008923661, "learning_rate": 9.806087903365826e-06, "loss": 0.2788, "step": 1622 }, { "epoch": 0.12, "grad_norm": 4.270082018964796, "learning_rate": 9.805767492987385e-06, "loss": 0.3185, "step": 1623 }, { "epoch": 0.12, "grad_norm": 9.053316424712301, "learning_rate": 9.805446823355978e-06, "loss": 1.1001, "step": 1624 }, { "epoch": 0.12, "grad_norm": 4.186039589012566, "learning_rate": 9.805125894488905e-06, "loss": 0.3634, "step": 1625 }, { "epoch": 0.12, "grad_norm": 3.4666559896886637, "learning_rate": 9.804804706403476e-06, "loss": 0.242, "step": 1626 }, { "epoch": 0.12, "grad_norm": 5.679546140614748, "learning_rate": 9.804483259117024e-06, "loss": 1.2985, "step": 1627 }, { "epoch": 0.12, "grad_norm": 14.305005329202714, "learning_rate": 9.804161552646882e-06, "loss": 0.7452, "step": 1628 }, { "epoch": 0.12, "grad_norm": 13.149769742842656, "learning_rate": 9.803839587010413e-06, "loss": 1.487, "step": 1629 }, { "epoch": 0.12, "grad_norm": 4.418770746962146, "learning_rate": 9.80351736222498e-06, "loss": 0.2783, "step": 1630 }, { "epoch": 0.12, "grad_norm": 4.6675046989958995, "learning_rate": 9.803194878307967e-06, "loss": 0.2652, "step": 1631 }, { "epoch": 0.12, "grad_norm": 19.302821572537702, "learning_rate": 9.802872135276771e-06, "loss": 1.4421, "step": 1632 }, { "epoch": 0.12, "grad_norm": 4.2349495552830385, "learning_rate": 9.802549133148804e-06, "loss": 0.271, "step": 1633 }, { "epoch": 0.12, "grad_norm": 23.573765419282285, "learning_rate": 9.80222587194149e-06, "loss": 1.2778, "step": 1634 }, { "epoch": 0.12, "grad_norm": 13.102067014066703, "learning_rate": 9.80190235167227e-06, "loss": 1.3876, "step": 1635 }, { "epoch": 0.12, "grad_norm": 16.27613298152478, "learning_rate": 9.80157857235859e-06, "loss": 1.3221, "step": 1636 }, { "epoch": 0.12, "grad_norm": 5.3843329892257605, "learning_rate": 9.801254534017923e-06, "loss": 0.4153, "step": 1637 }, { "epoch": 0.12, "grad_norm": 16.733742668322567, "learning_rate": 9.800930236667748e-06, "loss": 1.1098, "step": 1638 }, { "epoch": 0.12, "grad_norm": 20.922478343169622, "learning_rate": 9.80060568032556e-06, "loss": 1.417, "step": 1639 }, { "epoch": 0.12, "grad_norm": 22.861316032401575, "learning_rate": 9.800280865008864e-06, "loss": 1.0773, "step": 1640 }, { "epoch": 0.12, "grad_norm": 4.497197635642287, "learning_rate": 9.799955790735187e-06, "loss": 0.3288, "step": 1641 }, { "epoch": 0.12, "grad_norm": 14.71452049858688, "learning_rate": 9.799630457522065e-06, "loss": 2.0367, "step": 1642 }, { "epoch": 0.12, "grad_norm": 4.9826909785244515, "learning_rate": 9.799304865387045e-06, "loss": 0.2908, "step": 1643 }, { "epoch": 0.12, "grad_norm": 43.46236665826924, "learning_rate": 9.798979014347696e-06, "loss": 1.5461, "step": 1644 }, { "epoch": 0.12, "grad_norm": 5.919654047451032, "learning_rate": 9.798652904421593e-06, "loss": 1.0345, "step": 1645 }, { "epoch": 0.12, "grad_norm": 24.164475412900767, "learning_rate": 9.798326535626331e-06, "loss": 1.4445, "step": 1646 }, { "epoch": 0.12, "grad_norm": 5.314097946246953, "learning_rate": 9.797999907979515e-06, "loss": 0.3691, "step": 1647 }, { "epoch": 0.12, "grad_norm": 4.516172222867311, "learning_rate": 9.797673021498764e-06, "loss": 0.2851, "step": 1648 }, { "epoch": 0.12, "grad_norm": 4.405550556953849, "learning_rate": 9.797345876201714e-06, "loss": 0.2172, "step": 1649 }, { "epoch": 0.12, "grad_norm": 7.280934582937967, "learning_rate": 9.797018472106013e-06, "loss": 1.0812, "step": 1650 }, { "epoch": 0.12, "grad_norm": 4.814044384034644, "learning_rate": 9.796690809229323e-06, "loss": 0.3048, "step": 1651 }, { "epoch": 0.12, "grad_norm": 6.842986607743116, "learning_rate": 9.796362887589321e-06, "loss": 1.1234, "step": 1652 }, { "epoch": 0.12, "grad_norm": 4.973258921790116, "learning_rate": 9.796034707203697e-06, "loss": 0.3545, "step": 1653 }, { "epoch": 0.12, "grad_norm": 4.896131009128511, "learning_rate": 9.795706268090153e-06, "loss": 0.2303, "step": 1654 }, { "epoch": 0.12, "grad_norm": 4.801339576616092, "learning_rate": 9.795377570266409e-06, "loss": 0.2953, "step": 1655 }, { "epoch": 0.12, "grad_norm": 5.576739567614082, "learning_rate": 9.795048613750197e-06, "loss": 0.2982, "step": 1656 }, { "epoch": 0.12, "grad_norm": 16.659840059846733, "learning_rate": 9.794719398559263e-06, "loss": 1.7663, "step": 1657 }, { "epoch": 0.12, "grad_norm": 5.22317503870554, "learning_rate": 9.794389924711365e-06, "loss": 0.3321, "step": 1658 }, { "epoch": 0.12, "grad_norm": 10.211930953333182, "learning_rate": 9.79406019222428e-06, "loss": 0.8676, "step": 1659 }, { "epoch": 0.12, "grad_norm": 16.142922760140166, "learning_rate": 9.793730201115792e-06, "loss": 1.1873, "step": 1660 }, { "epoch": 0.12, "grad_norm": 4.116040979292729, "learning_rate": 9.793399951403706e-06, "loss": 0.3188, "step": 1661 }, { "epoch": 0.12, "grad_norm": 29.397411780450565, "learning_rate": 9.793069443105836e-06, "loss": 1.6628, "step": 1662 }, { "epoch": 0.12, "grad_norm": 5.49478651624847, "learning_rate": 9.792738676240013e-06, "loss": 0.2939, "step": 1663 }, { "epoch": 0.12, "grad_norm": 5.512212964796319, "learning_rate": 9.792407650824081e-06, "loss": 0.3641, "step": 1664 }, { "epoch": 0.12, "grad_norm": 5.925788300244167, "learning_rate": 9.792076366875895e-06, "loss": 0.3806, "step": 1665 }, { "epoch": 0.12, "grad_norm": 13.16858306249579, "learning_rate": 9.791744824413329e-06, "loss": 1.3723, "step": 1666 }, { "epoch": 0.12, "grad_norm": 14.659566777494458, "learning_rate": 9.791413023454266e-06, "loss": 1.2411, "step": 1667 }, { "epoch": 0.12, "grad_norm": 4.645063611479164, "learning_rate": 9.791080964016609e-06, "loss": 0.3392, "step": 1668 }, { "epoch": 0.12, "grad_norm": 5.655255295501133, "learning_rate": 9.790748646118266e-06, "loss": 0.3577, "step": 1669 }, { "epoch": 0.12, "grad_norm": 6.725334657389284, "learning_rate": 9.79041606977717e-06, "loss": 0.4524, "step": 1670 }, { "epoch": 0.12, "grad_norm": 6.950720144613855, "learning_rate": 9.79008323501126e-06, "loss": 1.0739, "step": 1671 }, { "epoch": 0.12, "grad_norm": 12.374826596716991, "learning_rate": 9.78975014183849e-06, "loss": 1.4496, "step": 1672 }, { "epoch": 0.12, "grad_norm": 13.176955274702555, "learning_rate": 9.78941679027683e-06, "loss": 1.0418, "step": 1673 }, { "epoch": 0.12, "grad_norm": 5.551475348459286, "learning_rate": 9.789083180344263e-06, "loss": 0.3492, "step": 1674 }, { "epoch": 0.12, "grad_norm": 12.572094366201078, "learning_rate": 9.788749312058787e-06, "loss": 1.5618, "step": 1675 }, { "epoch": 0.12, "grad_norm": 23.61678636345783, "learning_rate": 9.788415185438411e-06, "loss": 0.9572, "step": 1676 }, { "epoch": 0.12, "grad_norm": 4.69243053379465, "learning_rate": 9.78808080050116e-06, "loss": 0.3596, "step": 1677 }, { "epoch": 0.12, "grad_norm": 4.663387800646603, "learning_rate": 9.787746157265077e-06, "loss": 0.3144, "step": 1678 }, { "epoch": 0.12, "grad_norm": 13.758390541091654, "learning_rate": 9.787411255748207e-06, "loss": 1.5502, "step": 1679 }, { "epoch": 0.12, "grad_norm": 5.129382018343231, "learning_rate": 9.787076095968624e-06, "loss": 0.3175, "step": 1680 }, { "epoch": 0.12, "grad_norm": 4.938194942456906, "learning_rate": 9.786740677944407e-06, "loss": 0.3613, "step": 1681 }, { "epoch": 0.12, "grad_norm": 11.72500834894702, "learning_rate": 9.786405001693646e-06, "loss": 1.6884, "step": 1682 }, { "epoch": 0.12, "grad_norm": 4.885325285144449, "learning_rate": 9.786069067234454e-06, "loss": 0.3864, "step": 1683 }, { "epoch": 0.12, "grad_norm": 4.957255386552923, "learning_rate": 9.785732874584951e-06, "loss": 0.3439, "step": 1684 }, { "epoch": 0.12, "grad_norm": 26.1068914093648, "learning_rate": 9.785396423763276e-06, "loss": 1.2392, "step": 1685 }, { "epoch": 0.12, "grad_norm": 15.36649404061814, "learning_rate": 9.785059714787579e-06, "loss": 1.4016, "step": 1686 }, { "epoch": 0.12, "grad_norm": 5.7811471303553565, "learning_rate": 9.78472274767602e-06, "loss": 1.0286, "step": 1687 }, { "epoch": 0.12, "grad_norm": 11.395694542922111, "learning_rate": 9.78438552244678e-06, "loss": 1.1421, "step": 1688 }, { "epoch": 0.12, "grad_norm": 6.14538357939311, "learning_rate": 9.784048039118053e-06, "loss": 1.0453, "step": 1689 }, { "epoch": 0.12, "grad_norm": 5.067680998953025, "learning_rate": 9.78371029770804e-06, "loss": 0.3719, "step": 1690 }, { "epoch": 0.12, "grad_norm": 13.272872758559776, "learning_rate": 9.783372298234966e-06, "loss": 0.8581, "step": 1691 }, { "epoch": 0.12, "grad_norm": 7.971570767366137, "learning_rate": 9.783034040717063e-06, "loss": 1.0162, "step": 1692 }, { "epoch": 0.12, "grad_norm": 28.002432898400734, "learning_rate": 9.782695525172577e-06, "loss": 1.1312, "step": 1693 }, { "epoch": 0.12, "grad_norm": 4.817659029971965, "learning_rate": 9.78235675161977e-06, "loss": 0.3343, "step": 1694 }, { "epoch": 0.12, "grad_norm": 3.8724486864529624, "learning_rate": 9.782017720076922e-06, "loss": 0.3091, "step": 1695 }, { "epoch": 0.12, "grad_norm": 15.81902789577758, "learning_rate": 9.781678430562315e-06, "loss": 1.2923, "step": 1696 }, { "epoch": 0.12, "grad_norm": 4.534192070332152, "learning_rate": 9.781338883094258e-06, "loss": 0.309, "step": 1697 }, { "epoch": 0.12, "grad_norm": 4.052459864089427, "learning_rate": 9.780999077691067e-06, "loss": 0.2798, "step": 1698 }, { "epoch": 0.12, "grad_norm": 21.86644592143455, "learning_rate": 9.780659014371072e-06, "loss": 1.4228, "step": 1699 }, { "epoch": 0.12, "grad_norm": 5.3653115447259125, "learning_rate": 9.78031869315262e-06, "loss": 0.5179, "step": 1700 }, { "epoch": 0.12, "grad_norm": 6.314007187379518, "learning_rate": 9.779978114054067e-06, "loss": 1.0977, "step": 1701 }, { "epoch": 0.12, "grad_norm": 4.447866267466741, "learning_rate": 9.77963727709379e-06, "loss": 0.3203, "step": 1702 }, { "epoch": 0.12, "grad_norm": 5.308718654910777, "learning_rate": 9.779296182290173e-06, "loss": 0.2757, "step": 1703 }, { "epoch": 0.12, "grad_norm": 14.166775450317697, "learning_rate": 9.778954829661618e-06, "loss": 1.6265, "step": 1704 }, { "epoch": 0.12, "grad_norm": 4.897563074411955, "learning_rate": 9.778613219226539e-06, "loss": 0.2692, "step": 1705 }, { "epoch": 0.12, "grad_norm": 4.589764762325046, "learning_rate": 9.778271351003365e-06, "loss": 0.2949, "step": 1706 }, { "epoch": 0.12, "grad_norm": 33.0101532820508, "learning_rate": 9.777929225010538e-06, "loss": 1.4852, "step": 1707 }, { "epoch": 0.12, "grad_norm": 5.401973017752019, "learning_rate": 9.777586841266514e-06, "loss": 0.4315, "step": 1708 }, { "epoch": 0.12, "grad_norm": 4.912875194346227, "learning_rate": 9.777244199789767e-06, "loss": 0.2738, "step": 1709 }, { "epoch": 0.12, "grad_norm": 6.396243895200103, "learning_rate": 9.776901300598775e-06, "loss": 1.1721, "step": 1710 }, { "epoch": 0.12, "grad_norm": 14.758624328829065, "learning_rate": 9.77655814371204e-06, "loss": 0.9071, "step": 1711 }, { "epoch": 0.12, "grad_norm": 4.724655717932264, "learning_rate": 9.776214729148075e-06, "loss": 0.3401, "step": 1712 }, { "epoch": 0.12, "grad_norm": 3.6216176875214967, "learning_rate": 9.775871056925403e-06, "loss": 0.3085, "step": 1713 }, { "epoch": 0.12, "grad_norm": 23.05280361147133, "learning_rate": 9.775527127062566e-06, "loss": 1.3738, "step": 1714 }, { "epoch": 0.12, "grad_norm": 14.219736173972596, "learning_rate": 9.775182939578115e-06, "loss": 1.2758, "step": 1715 }, { "epoch": 0.12, "grad_norm": 4.1498145997233395, "learning_rate": 9.77483849449062e-06, "loss": 0.287, "step": 1716 }, { "epoch": 0.12, "grad_norm": 3.9355913857956697, "learning_rate": 9.774493791818663e-06, "loss": 0.302, "step": 1717 }, { "epoch": 0.12, "grad_norm": 5.680304418145706, "learning_rate": 9.774148831580838e-06, "loss": 1.0742, "step": 1718 }, { "epoch": 0.12, "grad_norm": 25.39678949251965, "learning_rate": 9.773803613795757e-06, "loss": 1.509, "step": 1719 }, { "epoch": 0.12, "grad_norm": 4.559516361970822, "learning_rate": 9.773458138482037e-06, "loss": 0.3679, "step": 1720 }, { "epoch": 0.12, "grad_norm": 4.596542877021742, "learning_rate": 9.77311240565832e-06, "loss": 0.2775, "step": 1721 }, { "epoch": 0.12, "grad_norm": 5.345739164153695, "learning_rate": 9.772766415343257e-06, "loss": 1.0143, "step": 1722 }, { "epoch": 0.12, "grad_norm": 5.193793270339112, "learning_rate": 9.772420167555512e-06, "loss": 0.2714, "step": 1723 }, { "epoch": 0.12, "grad_norm": 4.044941486965281, "learning_rate": 9.772073662313761e-06, "loss": 0.2895, "step": 1724 }, { "epoch": 0.12, "grad_norm": 3.609466752600603, "learning_rate": 9.771726899636701e-06, "loss": 0.2025, "step": 1725 }, { "epoch": 0.12, "grad_norm": 4.39644588262929, "learning_rate": 9.771379879543038e-06, "loss": 0.2378, "step": 1726 }, { "epoch": 0.12, "grad_norm": 4.844858204547516, "learning_rate": 9.771032602051489e-06, "loss": 0.3839, "step": 1727 }, { "epoch": 0.12, "grad_norm": 5.453306462405535, "learning_rate": 9.770685067180792e-06, "loss": 0.9622, "step": 1728 }, { "epoch": 0.12, "grad_norm": 45.03982261797562, "learning_rate": 9.770337274949693e-06, "loss": 1.5667, "step": 1729 }, { "epoch": 0.12, "grad_norm": 4.93995602525646, "learning_rate": 9.769989225376954e-06, "loss": 0.3102, "step": 1730 }, { "epoch": 0.12, "grad_norm": 16.302667245425468, "learning_rate": 9.769640918481354e-06, "loss": 0.9859, "step": 1731 }, { "epoch": 0.12, "grad_norm": 3.7271120876564523, "learning_rate": 9.769292354281679e-06, "loss": 0.2182, "step": 1732 }, { "epoch": 0.12, "grad_norm": 5.207575788898936, "learning_rate": 9.768943532796732e-06, "loss": 0.3936, "step": 1733 }, { "epoch": 0.12, "grad_norm": 12.922927064785128, "learning_rate": 9.768594454045337e-06, "loss": 1.0762, "step": 1734 }, { "epoch": 0.12, "grad_norm": 4.3787568920002915, "learning_rate": 9.768245118046319e-06, "loss": 0.3175, "step": 1735 }, { "epoch": 0.12, "grad_norm": 18.91208536402075, "learning_rate": 9.767895524818526e-06, "loss": 0.951, "step": 1736 }, { "epoch": 0.12, "grad_norm": 4.107411578717551, "learning_rate": 9.767545674380816e-06, "loss": 0.2771, "step": 1737 }, { "epoch": 0.12, "grad_norm": 19.96317415877599, "learning_rate": 9.767195566752064e-06, "loss": 1.2397, "step": 1738 }, { "epoch": 0.12, "grad_norm": 21.576510689468975, "learning_rate": 9.766845201951156e-06, "loss": 1.686, "step": 1739 }, { "epoch": 0.12, "grad_norm": 35.19212110709528, "learning_rate": 9.766494579996992e-06, "loss": 1.6499, "step": 1740 }, { "epoch": 0.12, "grad_norm": 4.720987943786792, "learning_rate": 9.766143700908489e-06, "loss": 0.3161, "step": 1741 }, { "epoch": 0.12, "grad_norm": 5.867511314459896, "learning_rate": 9.765792564704574e-06, "loss": 0.4169, "step": 1742 }, { "epoch": 0.12, "grad_norm": 5.031450408096444, "learning_rate": 9.765441171404188e-06, "loss": 0.2623, "step": 1743 }, { "epoch": 0.13, "grad_norm": 4.932264749726195, "learning_rate": 9.76508952102629e-06, "loss": 0.3133, "step": 1744 }, { "epoch": 0.13, "grad_norm": 4.984146709902565, "learning_rate": 9.76473761358985e-06, "loss": 0.3227, "step": 1745 }, { "epoch": 0.13, "grad_norm": 6.196775565530507, "learning_rate": 9.764385449113852e-06, "loss": 0.9848, "step": 1746 }, { "epoch": 0.13, "grad_norm": 6.304901083090228, "learning_rate": 9.76403302761729e-06, "loss": 1.2184, "step": 1747 }, { "epoch": 0.13, "grad_norm": 4.808801224445701, "learning_rate": 9.763680349119181e-06, "loss": 0.2867, "step": 1748 }, { "epoch": 0.13, "grad_norm": 4.889860410675962, "learning_rate": 9.763327413638549e-06, "loss": 0.3491, "step": 1749 }, { "epoch": 0.13, "grad_norm": 4.54227378551419, "learning_rate": 9.762974221194433e-06, "loss": 0.27, "step": 1750 }, { "epoch": 0.13, "grad_norm": 52.69755857608532, "learning_rate": 9.762620771805884e-06, "loss": 1.0049, "step": 1751 }, { "epoch": 0.13, "grad_norm": 4.150898581703361, "learning_rate": 9.762267065491975e-06, "loss": 0.2386, "step": 1752 }, { "epoch": 0.13, "grad_norm": 14.2092460956909, "learning_rate": 9.761913102271782e-06, "loss": 1.4099, "step": 1753 }, { "epoch": 0.13, "grad_norm": 5.833991611678696, "learning_rate": 9.761558882164404e-06, "loss": 1.1285, "step": 1754 }, { "epoch": 0.13, "grad_norm": 15.66405211105054, "learning_rate": 9.761204405188947e-06, "loss": 1.3516, "step": 1755 }, { "epoch": 0.13, "grad_norm": 4.03318387371403, "learning_rate": 9.760849671364533e-06, "loss": 0.2465, "step": 1756 }, { "epoch": 0.13, "grad_norm": 6.713838621945019, "learning_rate": 9.760494680710303e-06, "loss": 1.2093, "step": 1757 }, { "epoch": 0.13, "grad_norm": 73.39585854865697, "learning_rate": 9.760139433245403e-06, "loss": 1.949, "step": 1758 }, { "epoch": 0.13, "grad_norm": 24.055523803928196, "learning_rate": 9.759783928988997e-06, "loss": 1.8036, "step": 1759 }, { "epoch": 0.13, "grad_norm": 10.948337816159555, "learning_rate": 9.759428167960267e-06, "loss": 1.5068, "step": 1760 }, { "epoch": 0.13, "grad_norm": 18.407064490161368, "learning_rate": 9.7590721501784e-06, "loss": 1.3687, "step": 1761 }, { "epoch": 0.13, "grad_norm": 32.017464849975234, "learning_rate": 9.758715875662608e-06, "loss": 1.7231, "step": 1762 }, { "epoch": 0.13, "grad_norm": 7.462976980789892, "learning_rate": 9.758359344432107e-06, "loss": 1.043, "step": 1763 }, { "epoch": 0.13, "grad_norm": 19.881857790486297, "learning_rate": 9.75800255650613e-06, "loss": 1.7092, "step": 1764 }, { "epoch": 0.13, "grad_norm": 4.769467277372865, "learning_rate": 9.757645511903924e-06, "loss": 0.3859, "step": 1765 }, { "epoch": 0.13, "grad_norm": 11.14343545592022, "learning_rate": 9.75728821064475e-06, "loss": 0.9705, "step": 1766 }, { "epoch": 0.13, "grad_norm": 4.9595668452137955, "learning_rate": 9.756930652747887e-06, "loss": 0.2885, "step": 1767 }, { "epoch": 0.13, "grad_norm": 15.61940375941115, "learning_rate": 9.756572838232621e-06, "loss": 1.3147, "step": 1768 }, { "epoch": 0.13, "grad_norm": 5.354153198288985, "learning_rate": 9.756214767118255e-06, "loss": 0.507, "step": 1769 }, { "epoch": 0.13, "grad_norm": 21.37600150540276, "learning_rate": 9.755856439424105e-06, "loss": 1.7124, "step": 1770 }, { "epoch": 0.13, "grad_norm": 6.297350410456241, "learning_rate": 9.755497855169502e-06, "loss": 1.0691, "step": 1771 }, { "epoch": 0.13, "grad_norm": 22.706572612969957, "learning_rate": 9.755139014373789e-06, "loss": 1.6248, "step": 1772 }, { "epoch": 0.13, "grad_norm": 5.089595363366967, "learning_rate": 9.754779917056327e-06, "loss": 0.3735, "step": 1773 }, { "epoch": 0.13, "grad_norm": 5.310847825128942, "learning_rate": 9.754420563236487e-06, "loss": 0.3348, "step": 1774 }, { "epoch": 0.13, "grad_norm": 4.551162664000545, "learning_rate": 9.754060952933652e-06, "loss": 0.3312, "step": 1775 }, { "epoch": 0.13, "grad_norm": 9.35974563520835, "learning_rate": 9.753701086167224e-06, "loss": 0.7712, "step": 1776 }, { "epoch": 0.13, "grad_norm": 6.006217310842812, "learning_rate": 9.753340962956617e-06, "loss": 1.1503, "step": 1777 }, { "epoch": 0.13, "grad_norm": 12.873761003918581, "learning_rate": 9.752980583321258e-06, "loss": 0.9711, "step": 1778 }, { "epoch": 0.13, "grad_norm": 3.288887211746414, "learning_rate": 9.752619947280587e-06, "loss": 0.3013, "step": 1779 }, { "epoch": 0.13, "grad_norm": 31.394367151997688, "learning_rate": 9.752259054854058e-06, "loss": 1.2314, "step": 1780 }, { "epoch": 0.13, "grad_norm": 8.950050921432567, "learning_rate": 9.751897906061143e-06, "loss": 0.8209, "step": 1781 }, { "epoch": 0.13, "grad_norm": 4.020010130909499, "learning_rate": 9.751536500921322e-06, "loss": 0.3075, "step": 1782 }, { "epoch": 0.13, "grad_norm": 3.686646051480571, "learning_rate": 9.751174839454094e-06, "loss": 0.3165, "step": 1783 }, { "epoch": 0.13, "grad_norm": 3.85199235102652, "learning_rate": 9.750812921678965e-06, "loss": 0.2926, "step": 1784 }, { "epoch": 0.13, "grad_norm": 34.28915624999759, "learning_rate": 9.750450747615463e-06, "loss": 1.1453, "step": 1785 }, { "epoch": 0.13, "grad_norm": 3.7486603064655335, "learning_rate": 9.750088317283125e-06, "loss": 0.2121, "step": 1786 }, { "epoch": 0.13, "grad_norm": 3.5501520358343974, "learning_rate": 9.749725630701503e-06, "loss": 0.2397, "step": 1787 }, { "epoch": 0.13, "grad_norm": 18.913037264147192, "learning_rate": 9.74936268789016e-06, "loss": 1.4517, "step": 1788 }, { "epoch": 0.13, "grad_norm": 21.68902033621489, "learning_rate": 9.748999488868677e-06, "loss": 1.6537, "step": 1789 }, { "epoch": 0.13, "grad_norm": 6.728258210522582, "learning_rate": 9.74863603365665e-06, "loss": 1.2571, "step": 1790 }, { "epoch": 0.13, "grad_norm": 4.075801095172629, "learning_rate": 9.748272322273681e-06, "loss": 0.2054, "step": 1791 }, { "epoch": 0.13, "grad_norm": 4.798903174284147, "learning_rate": 9.747908354739394e-06, "loss": 0.254, "step": 1792 }, { "epoch": 0.13, "grad_norm": 12.918456556004495, "learning_rate": 9.747544131073425e-06, "loss": 1.1838, "step": 1793 }, { "epoch": 0.13, "grad_norm": 5.100726715789696, "learning_rate": 9.74717965129542e-06, "loss": 0.3441, "step": 1794 }, { "epoch": 0.13, "grad_norm": 6.215950769767821, "learning_rate": 9.74681491542504e-06, "loss": 1.1518, "step": 1795 }, { "epoch": 0.13, "grad_norm": 5.416940531639362, "learning_rate": 9.746449923481965e-06, "loss": 1.0457, "step": 1796 }, { "epoch": 0.13, "grad_norm": 19.41324162097771, "learning_rate": 9.746084675485881e-06, "loss": 1.3266, "step": 1797 }, { "epoch": 0.13, "grad_norm": 14.45956510405657, "learning_rate": 9.745719171456495e-06, "loss": 1.4575, "step": 1798 }, { "epoch": 0.13, "grad_norm": 4.497229403196966, "learning_rate": 9.745353411413523e-06, "loss": 0.3391, "step": 1799 }, { "epoch": 0.13, "grad_norm": 4.657173789796294, "learning_rate": 9.744987395376696e-06, "loss": 0.3518, "step": 1800 }, { "epoch": 0.13, "grad_norm": 20.474495807926722, "learning_rate": 9.74462112336576e-06, "loss": 1.8693, "step": 1801 }, { "epoch": 0.13, "grad_norm": 25.930018617507827, "learning_rate": 9.744254595400474e-06, "loss": 2.0084, "step": 1802 }, { "epoch": 0.13, "grad_norm": 5.297859159649251, "learning_rate": 9.74388781150061e-06, "loss": 0.3496, "step": 1803 }, { "epoch": 0.13, "grad_norm": 5.464225815275087, "learning_rate": 9.743520771685957e-06, "loss": 0.4307, "step": 1804 }, { "epoch": 0.13, "grad_norm": 5.079522383904176, "learning_rate": 9.743153475976312e-06, "loss": 0.2106, "step": 1805 }, { "epoch": 0.13, "grad_norm": 5.247139150153015, "learning_rate": 9.74278592439149e-06, "loss": 0.3377, "step": 1806 }, { "epoch": 0.13, "grad_norm": 49.669293533157436, "learning_rate": 9.742418116951322e-06, "loss": 1.5724, "step": 1807 }, { "epoch": 0.13, "grad_norm": 6.449372265289701, "learning_rate": 9.742050053675645e-06, "loss": 0.3255, "step": 1808 }, { "epoch": 0.13, "grad_norm": 6.668511022104206, "learning_rate": 9.74168173458432e-06, "loss": 1.1609, "step": 1809 }, { "epoch": 0.13, "grad_norm": 4.7693748569514565, "learning_rate": 9.741313159697212e-06, "loss": 0.3024, "step": 1810 }, { "epoch": 0.13, "grad_norm": 6.062103313067339, "learning_rate": 9.740944329034206e-06, "loss": 0.9025, "step": 1811 }, { "epoch": 0.13, "grad_norm": 4.733405639760099, "learning_rate": 9.7405752426152e-06, "loss": 0.3187, "step": 1812 }, { "epoch": 0.13, "grad_norm": 6.2166327863019335, "learning_rate": 9.740205900460102e-06, "loss": 1.0285, "step": 1813 }, { "epoch": 0.13, "grad_norm": 5.120608812046843, "learning_rate": 9.739836302588839e-06, "loss": 0.2791, "step": 1814 }, { "epoch": 0.13, "grad_norm": 16.322302251486487, "learning_rate": 9.739466449021348e-06, "loss": 1.0189, "step": 1815 }, { "epoch": 0.13, "grad_norm": 5.139705480046142, "learning_rate": 9.739096339777584e-06, "loss": 0.3765, "step": 1816 }, { "epoch": 0.13, "grad_norm": 33.53582847754766, "learning_rate": 9.73872597487751e-06, "loss": 2.0067, "step": 1817 }, { "epoch": 0.13, "grad_norm": 7.301760859297956, "learning_rate": 9.738355354341107e-06, "loss": 1.1393, "step": 1818 }, { "epoch": 0.13, "grad_norm": 6.802185736757348, "learning_rate": 9.737984478188368e-06, "loss": 0.993, "step": 1819 }, { "epoch": 0.13, "grad_norm": 5.519177869369563, "learning_rate": 9.7376133464393e-06, "loss": 0.3884, "step": 1820 }, { "epoch": 0.13, "grad_norm": 7.247054318792038, "learning_rate": 9.737241959113925e-06, "loss": 0.3352, "step": 1821 }, { "epoch": 0.13, "grad_norm": 5.632182032548367, "learning_rate": 9.736870316232278e-06, "loss": 0.3324, "step": 1822 }, { "epoch": 0.13, "grad_norm": 5.351423783553331, "learning_rate": 9.736498417814406e-06, "loss": 0.2232, "step": 1823 }, { "epoch": 0.13, "grad_norm": 7.696426653492521, "learning_rate": 9.736126263880376e-06, "loss": 1.0165, "step": 1824 }, { "epoch": 0.13, "grad_norm": 5.685822074792867, "learning_rate": 9.735753854450258e-06, "loss": 0.2742, "step": 1825 }, { "epoch": 0.13, "grad_norm": 70.47479865744431, "learning_rate": 9.735381189544149e-06, "loss": 1.0779, "step": 1826 }, { "epoch": 0.13, "grad_norm": 4.570916904175144, "learning_rate": 9.735008269182147e-06, "loss": 0.3065, "step": 1827 }, { "epoch": 0.13, "grad_norm": 7.136899130407513, "learning_rate": 9.734635093384372e-06, "loss": 0.3202, "step": 1828 }, { "epoch": 0.13, "grad_norm": 21.401027204020114, "learning_rate": 9.734261662170953e-06, "loss": 0.742, "step": 1829 }, { "epoch": 0.13, "grad_norm": 13.152973493996177, "learning_rate": 9.733887975562041e-06, "loss": 1.5035, "step": 1830 }, { "epoch": 0.13, "grad_norm": 4.6995890627652805, "learning_rate": 9.733514033577791e-06, "loss": 0.3212, "step": 1831 }, { "epoch": 0.13, "grad_norm": 4.1417243394302625, "learning_rate": 9.733139836238376e-06, "loss": 0.3272, "step": 1832 }, { "epoch": 0.13, "grad_norm": 19.33884643008215, "learning_rate": 9.732765383563983e-06, "loss": 0.8278, "step": 1833 }, { "epoch": 0.13, "grad_norm": 4.040789759905044, "learning_rate": 9.73239067557481e-06, "loss": 0.29, "step": 1834 }, { "epoch": 0.13, "grad_norm": 4.322829661748336, "learning_rate": 9.732015712291075e-06, "loss": 0.3709, "step": 1835 }, { "epoch": 0.13, "grad_norm": 13.74076214022332, "learning_rate": 9.731640493733004e-06, "loss": 1.1644, "step": 1836 }, { "epoch": 0.13, "grad_norm": 19.584792072046525, "learning_rate": 9.731265019920836e-06, "loss": 1.6632, "step": 1837 }, { "epoch": 0.13, "grad_norm": 4.991368786296432, "learning_rate": 9.730889290874833e-06, "loss": 0.3244, "step": 1838 }, { "epoch": 0.13, "grad_norm": 6.311992454537559, "learning_rate": 9.730513306615258e-06, "loss": 1.2583, "step": 1839 }, { "epoch": 0.13, "grad_norm": 10.88180158735713, "learning_rate": 9.730137067162396e-06, "loss": 1.3311, "step": 1840 }, { "epoch": 0.13, "grad_norm": 10.375404530043603, "learning_rate": 9.729760572536542e-06, "loss": 0.7283, "step": 1841 }, { "epoch": 0.13, "grad_norm": 4.1883310195876895, "learning_rate": 9.729383822758012e-06, "loss": 0.3695, "step": 1842 }, { "epoch": 0.13, "grad_norm": 5.07347635138009, "learning_rate": 9.729006817847124e-06, "loss": 1.0737, "step": 1843 }, { "epoch": 0.13, "grad_norm": 5.792061244319384, "learning_rate": 9.728629557824218e-06, "loss": 1.2445, "step": 1844 }, { "epoch": 0.13, "grad_norm": 15.646203908552451, "learning_rate": 9.728252042709648e-06, "loss": 0.961, "step": 1845 }, { "epoch": 0.13, "grad_norm": 22.92258488083376, "learning_rate": 9.727874272523777e-06, "loss": 1.1222, "step": 1846 }, { "epoch": 0.13, "grad_norm": 8.470246679773757, "learning_rate": 9.727496247286985e-06, "loss": 0.9659, "step": 1847 }, { "epoch": 0.13, "grad_norm": 3.384046929921896, "learning_rate": 9.727117967019664e-06, "loss": 0.3091, "step": 1848 }, { "epoch": 0.13, "grad_norm": 4.065098243299034, "learning_rate": 9.726739431742224e-06, "loss": 0.2893, "step": 1849 }, { "epoch": 0.13, "grad_norm": 17.56765267238573, "learning_rate": 9.726360641475081e-06, "loss": 1.3726, "step": 1850 }, { "epoch": 0.13, "grad_norm": 5.885316779952114, "learning_rate": 9.725981596238675e-06, "loss": 1.1804, "step": 1851 }, { "epoch": 0.13, "grad_norm": 24.69299037317376, "learning_rate": 9.725602296053447e-06, "loss": 1.4223, "step": 1852 }, { "epoch": 0.13, "grad_norm": 5.430072488811135, "learning_rate": 9.725222740939863e-06, "loss": 0.3337, "step": 1853 }, { "epoch": 0.13, "grad_norm": 4.393994779068372, "learning_rate": 9.724842930918398e-06, "loss": 0.2485, "step": 1854 }, { "epoch": 0.13, "grad_norm": 21.807928319010877, "learning_rate": 9.724462866009544e-06, "loss": 1.7913, "step": 1855 }, { "epoch": 0.13, "grad_norm": 3.6712344693173344, "learning_rate": 9.7240825462338e-06, "loss": 0.2769, "step": 1856 }, { "epoch": 0.13, "grad_norm": 4.703195003894077, "learning_rate": 9.723701971611683e-06, "loss": 0.36, "step": 1857 }, { "epoch": 0.13, "grad_norm": 6.418563884313357, "learning_rate": 9.723321142163726e-06, "loss": 1.0781, "step": 1858 }, { "epoch": 0.13, "grad_norm": 12.095682884077354, "learning_rate": 9.722940057910473e-06, "loss": 1.4251, "step": 1859 }, { "epoch": 0.13, "grad_norm": 14.38178151956185, "learning_rate": 9.72255871887248e-06, "loss": 1.2208, "step": 1860 }, { "epoch": 0.13, "grad_norm": 4.314489970204228, "learning_rate": 9.72217712507032e-06, "loss": 0.2448, "step": 1861 }, { "epoch": 0.13, "grad_norm": 4.899022862153132, "learning_rate": 9.721795276524579e-06, "loss": 0.4179, "step": 1862 }, { "epoch": 0.13, "grad_norm": 26.74108310206918, "learning_rate": 9.721413173255857e-06, "loss": 1.6118, "step": 1863 }, { "epoch": 0.13, "grad_norm": 5.3381303130517646, "learning_rate": 9.721030815284763e-06, "loss": 0.3644, "step": 1864 }, { "epoch": 0.13, "grad_norm": 14.616493878386002, "learning_rate": 9.720648202631927e-06, "loss": 1.2907, "step": 1865 }, { "epoch": 0.13, "grad_norm": 5.616429187232846, "learning_rate": 9.72026533531799e-06, "loss": 1.2762, "step": 1866 }, { "epoch": 0.13, "grad_norm": 20.666631222763336, "learning_rate": 9.719882213363605e-06, "loss": 1.31, "step": 1867 }, { "epoch": 0.13, "grad_norm": 21.24661622691467, "learning_rate": 9.719498836789441e-06, "loss": 1.7563, "step": 1868 }, { "epoch": 0.13, "grad_norm": 5.592128270080609, "learning_rate": 9.719115205616178e-06, "loss": 0.3192, "step": 1869 }, { "epoch": 0.13, "grad_norm": 19.023445494292737, "learning_rate": 9.718731319864514e-06, "loss": 1.1271, "step": 1870 }, { "epoch": 0.13, "grad_norm": 9.33032184492464, "learning_rate": 9.718347179555155e-06, "loss": 0.848, "step": 1871 }, { "epoch": 0.13, "grad_norm": 22.011410773759213, "learning_rate": 9.717962784708827e-06, "loss": 1.5894, "step": 1872 }, { "epoch": 0.13, "grad_norm": 12.781926484837133, "learning_rate": 9.717578135346264e-06, "loss": 0.8345, "step": 1873 }, { "epoch": 0.13, "grad_norm": 6.653053655928593, "learning_rate": 9.717193231488218e-06, "loss": 1.1755, "step": 1874 }, { "epoch": 0.13, "grad_norm": 4.713005549613308, "learning_rate": 9.716808073155454e-06, "loss": 0.3789, "step": 1875 }, { "epoch": 0.13, "grad_norm": 15.612258762697875, "learning_rate": 9.716422660368747e-06, "loss": 1.0048, "step": 1876 }, { "epoch": 0.13, "grad_norm": 4.799159059448839, "learning_rate": 9.71603699314889e-06, "loss": 0.3593, "step": 1877 }, { "epoch": 0.13, "grad_norm": 19.430898084043587, "learning_rate": 9.71565107151669e-06, "loss": 1.4432, "step": 1878 }, { "epoch": 0.13, "grad_norm": 4.062150841349138, "learning_rate": 9.715264895492962e-06, "loss": 0.2822, "step": 1879 }, { "epoch": 0.13, "grad_norm": 15.835753525765702, "learning_rate": 9.714878465098543e-06, "loss": 1.6582, "step": 1880 }, { "epoch": 0.13, "grad_norm": 14.684708609018505, "learning_rate": 9.714491780354277e-06, "loss": 1.0606, "step": 1881 }, { "epoch": 0.13, "grad_norm": 4.773977718066168, "learning_rate": 9.714104841281024e-06, "loss": 0.3605, "step": 1882 }, { "epoch": 0.14, "grad_norm": 3.7983128275310514, "learning_rate": 9.713717647899662e-06, "loss": 0.2604, "step": 1883 }, { "epoch": 0.14, "grad_norm": 5.739524288519371, "learning_rate": 9.71333020023107e-06, "loss": 0.2924, "step": 1884 }, { "epoch": 0.14, "grad_norm": 5.707246434242864, "learning_rate": 9.712942498296159e-06, "loss": 1.0666, "step": 1885 }, { "epoch": 0.14, "grad_norm": 4.842753235294485, "learning_rate": 9.712554542115838e-06, "loss": 0.9689, "step": 1886 }, { "epoch": 0.14, "grad_norm": 3.9666930893191275, "learning_rate": 9.712166331711037e-06, "loss": 0.3089, "step": 1887 }, { "epoch": 0.14, "grad_norm": 24.553126132852338, "learning_rate": 9.7117778671027e-06, "loss": 1.9651, "step": 1888 }, { "epoch": 0.14, "grad_norm": 6.850846247363415, "learning_rate": 9.711389148311781e-06, "loss": 0.4166, "step": 1889 }, { "epoch": 0.14, "grad_norm": 5.18088494516111, "learning_rate": 9.711000175359251e-06, "loss": 0.3118, "step": 1890 }, { "epoch": 0.14, "grad_norm": 5.772388388851158, "learning_rate": 9.710610948266096e-06, "loss": 0.2482, "step": 1891 }, { "epoch": 0.14, "grad_norm": 16.53313545240447, "learning_rate": 9.710221467053308e-06, "loss": 1.2521, "step": 1892 }, { "epoch": 0.14, "grad_norm": 3.776493397886543, "learning_rate": 9.709831731741902e-06, "loss": 0.2715, "step": 1893 }, { "epoch": 0.14, "grad_norm": 12.097399527371394, "learning_rate": 9.709441742352902e-06, "loss": 1.2332, "step": 1894 }, { "epoch": 0.14, "grad_norm": 17.115806766830378, "learning_rate": 9.709051498907345e-06, "loss": 1.2739, "step": 1895 }, { "epoch": 0.14, "grad_norm": 21.73966155624622, "learning_rate": 9.708661001426287e-06, "loss": 2.1197, "step": 1896 }, { "epoch": 0.14, "grad_norm": 6.009176914800183, "learning_rate": 9.708270249930789e-06, "loss": 0.7938, "step": 1897 }, { "epoch": 0.14, "grad_norm": 5.227663077663519, "learning_rate": 9.707879244441933e-06, "loss": 0.2961, "step": 1898 }, { "epoch": 0.14, "grad_norm": 5.796815470316867, "learning_rate": 9.707487984980814e-06, "loss": 0.3347, "step": 1899 }, { "epoch": 0.14, "grad_norm": 5.895895275056373, "learning_rate": 9.707096471568536e-06, "loss": 0.1959, "step": 1900 }, { "epoch": 0.14, "grad_norm": 5.0383923165998254, "learning_rate": 9.70670470422622e-06, "loss": 0.2137, "step": 1901 }, { "epoch": 0.14, "grad_norm": 13.115524230172381, "learning_rate": 9.706312682975002e-06, "loss": 1.162, "step": 1902 }, { "epoch": 0.14, "grad_norm": 47.898333666111995, "learning_rate": 9.705920407836026e-06, "loss": 2.1325, "step": 1903 }, { "epoch": 0.14, "grad_norm": 7.2306264198618315, "learning_rate": 9.70552787883046e-06, "loss": 1.2892, "step": 1904 }, { "epoch": 0.14, "grad_norm": 4.919028496298654, "learning_rate": 9.705135095979477e-06, "loss": 0.2467, "step": 1905 }, { "epoch": 0.14, "grad_norm": 4.62315611819921, "learning_rate": 9.704742059304263e-06, "loss": 0.1881, "step": 1906 }, { "epoch": 0.14, "grad_norm": 25.111462051142055, "learning_rate": 9.704348768826023e-06, "loss": 2.0912, "step": 1907 }, { "epoch": 0.14, "grad_norm": 4.711707518951179, "learning_rate": 9.703955224565977e-06, "loss": 0.2576, "step": 1908 }, { "epoch": 0.14, "grad_norm": 4.994138038714543, "learning_rate": 9.70356142654535e-06, "loss": 0.2149, "step": 1909 }, { "epoch": 0.14, "grad_norm": 33.58224692178425, "learning_rate": 9.703167374785388e-06, "loss": 1.1316, "step": 1910 }, { "epoch": 0.14, "grad_norm": 16.887901856710016, "learning_rate": 9.70277306930735e-06, "loss": 1.137, "step": 1911 }, { "epoch": 0.14, "grad_norm": 4.603810640583134, "learning_rate": 9.702378510132505e-06, "loss": 0.3568, "step": 1912 }, { "epoch": 0.14, "grad_norm": 19.004214278466215, "learning_rate": 9.701983697282138e-06, "loss": 1.3985, "step": 1913 }, { "epoch": 0.14, "grad_norm": 10.516629225994956, "learning_rate": 9.70158863077755e-06, "loss": 0.6293, "step": 1914 }, { "epoch": 0.14, "grad_norm": 4.543675413352612, "learning_rate": 9.701193310640049e-06, "loss": 0.3555, "step": 1915 }, { "epoch": 0.14, "grad_norm": 24.71727387958877, "learning_rate": 9.700797736890967e-06, "loss": 1.5417, "step": 1916 }, { "epoch": 0.14, "grad_norm": 17.252924134947758, "learning_rate": 9.700401909551638e-06, "loss": 0.8242, "step": 1917 }, { "epoch": 0.14, "grad_norm": 4.04677162716801, "learning_rate": 9.70000582864342e-06, "loss": 0.2962, "step": 1918 }, { "epoch": 0.14, "grad_norm": 5.592504727276112, "learning_rate": 9.699609494187679e-06, "loss": 1.108, "step": 1919 }, { "epoch": 0.14, "grad_norm": 5.096932509079619, "learning_rate": 9.699212906205793e-06, "loss": 0.9326, "step": 1920 }, { "epoch": 0.14, "grad_norm": 3.516977284702795, "learning_rate": 9.698816064719158e-06, "loss": 0.2436, "step": 1921 }, { "epoch": 0.14, "grad_norm": 4.498445605047461, "learning_rate": 9.698418969749181e-06, "loss": 0.4073, "step": 1922 }, { "epoch": 0.14, "grad_norm": 4.417119684988516, "learning_rate": 9.698021621317288e-06, "loss": 0.3013, "step": 1923 }, { "epoch": 0.14, "grad_norm": 16.72987775077967, "learning_rate": 9.697624019444911e-06, "loss": 1.3848, "step": 1924 }, { "epoch": 0.14, "grad_norm": 19.168830211444906, "learning_rate": 9.697226164153498e-06, "loss": 1.2487, "step": 1925 }, { "epoch": 0.14, "grad_norm": 5.895460398078848, "learning_rate": 9.696828055464515e-06, "loss": 0.947, "step": 1926 }, { "epoch": 0.14, "grad_norm": 5.986607442570045, "learning_rate": 9.696429693399436e-06, "loss": 1.1489, "step": 1927 }, { "epoch": 0.14, "grad_norm": 5.374691138689757, "learning_rate": 9.696031077979752e-06, "loss": 0.2553, "step": 1928 }, { "epoch": 0.14, "grad_norm": 5.783264279856037, "learning_rate": 9.695632209226967e-06, "loss": 0.4005, "step": 1929 }, { "epoch": 0.14, "grad_norm": 19.748919640505083, "learning_rate": 9.6952330871626e-06, "loss": 1.2433, "step": 1930 }, { "epoch": 0.14, "grad_norm": 4.574090105264361, "learning_rate": 9.694833711808175e-06, "loss": 0.2742, "step": 1931 }, { "epoch": 0.14, "grad_norm": 3.7166575569181655, "learning_rate": 9.694434083185247e-06, "loss": 0.1507, "step": 1932 }, { "epoch": 0.14, "grad_norm": 26.940374855045356, "learning_rate": 9.69403420131537e-06, "loss": 1.4651, "step": 1933 }, { "epoch": 0.14, "grad_norm": 4.627544691227969, "learning_rate": 9.693634066220112e-06, "loss": 0.2753, "step": 1934 }, { "epoch": 0.14, "grad_norm": 4.521473282374634, "learning_rate": 9.693233677921065e-06, "loss": 0.2724, "step": 1935 }, { "epoch": 0.14, "grad_norm": 4.6463310353892675, "learning_rate": 9.692833036439827e-06, "loss": 0.2648, "step": 1936 }, { "epoch": 0.14, "grad_norm": 6.8969021847939915, "learning_rate": 9.692432141798008e-06, "loss": 0.9053, "step": 1937 }, { "epoch": 0.14, "grad_norm": 5.687749772684217, "learning_rate": 9.692030994017236e-06, "loss": 0.3388, "step": 1938 }, { "epoch": 0.14, "grad_norm": 16.784665303878175, "learning_rate": 9.691629593119154e-06, "loss": 1.0241, "step": 1939 }, { "epoch": 0.14, "grad_norm": 4.1508957411573695, "learning_rate": 9.691227939125414e-06, "loss": 0.3022, "step": 1940 }, { "epoch": 0.14, "grad_norm": 29.71881962116551, "learning_rate": 9.690826032057683e-06, "loss": 1.2905, "step": 1941 }, { "epoch": 0.14, "grad_norm": 4.805254796663094, "learning_rate": 9.690423871937645e-06, "loss": 0.3569, "step": 1942 }, { "epoch": 0.14, "grad_norm": 4.03182136153043, "learning_rate": 9.690021458786991e-06, "loss": 0.2145, "step": 1943 }, { "epoch": 0.14, "grad_norm": 12.564863274852978, "learning_rate": 9.689618792627434e-06, "loss": 1.3168, "step": 1944 }, { "epoch": 0.14, "grad_norm": 6.295381814068741, "learning_rate": 9.689215873480694e-06, "loss": 1.249, "step": 1945 }, { "epoch": 0.14, "grad_norm": 17.17291407591979, "learning_rate": 9.688812701368506e-06, "loss": 1.5504, "step": 1946 }, { "epoch": 0.14, "grad_norm": 4.069416917350583, "learning_rate": 9.688409276312622e-06, "loss": 0.317, "step": 1947 }, { "epoch": 0.14, "grad_norm": 5.755823029594329, "learning_rate": 9.688005598334804e-06, "loss": 1.0722, "step": 1948 }, { "epoch": 0.14, "grad_norm": 8.933425716714234, "learning_rate": 9.687601667456829e-06, "loss": 0.6097, "step": 1949 }, { "epoch": 0.14, "grad_norm": 3.951681699828475, "learning_rate": 9.687197483700488e-06, "loss": 0.3707, "step": 1950 }, { "epoch": 0.14, "grad_norm": 4.746962021676798, "learning_rate": 9.686793047087584e-06, "loss": 0.9561, "step": 1951 }, { "epoch": 0.14, "grad_norm": 6.720968508541444, "learning_rate": 9.686388357639936e-06, "loss": 1.2805, "step": 1952 }, { "epoch": 0.14, "grad_norm": 6.836254172834971, "learning_rate": 9.685983415379374e-06, "loss": 0.5916, "step": 1953 }, { "epoch": 0.14, "grad_norm": 4.224570771376216, "learning_rate": 9.685578220327744e-06, "loss": 0.3333, "step": 1954 }, { "epoch": 0.14, "grad_norm": 12.841334747752814, "learning_rate": 9.685172772506908e-06, "loss": 1.0514, "step": 1955 }, { "epoch": 0.14, "grad_norm": 19.109950284497373, "learning_rate": 9.684767071938732e-06, "loss": 0.7262, "step": 1956 }, { "epoch": 0.14, "grad_norm": 18.65737690127425, "learning_rate": 9.684361118645108e-06, "loss": 1.2959, "step": 1957 }, { "epoch": 0.14, "grad_norm": 3.8453133668547665, "learning_rate": 9.683954912647931e-06, "loss": 0.2943, "step": 1958 }, { "epoch": 0.14, "grad_norm": 17.86352561858731, "learning_rate": 9.683548453969117e-06, "loss": 1.7066, "step": 1959 }, { "epoch": 0.14, "grad_norm": 5.393108288030331, "learning_rate": 9.683141742630591e-06, "loss": 0.2798, "step": 1960 }, { "epoch": 0.14, "grad_norm": 7.181431207831538, "learning_rate": 9.682734778654294e-06, "loss": 1.1845, "step": 1961 }, { "epoch": 0.14, "grad_norm": 19.224908655446303, "learning_rate": 9.682327562062183e-06, "loss": 1.6221, "step": 1962 }, { "epoch": 0.14, "grad_norm": 4.465422539473171, "learning_rate": 9.681920092876222e-06, "loss": 0.4109, "step": 1963 }, { "epoch": 0.14, "grad_norm": 3.354397577962642, "learning_rate": 9.681512371118395e-06, "loss": 0.2017, "step": 1964 }, { "epoch": 0.14, "grad_norm": 4.5098668069343635, "learning_rate": 9.681104396810696e-06, "loss": 0.2585, "step": 1965 }, { "epoch": 0.14, "grad_norm": 8.642357458772334, "learning_rate": 9.680696169975133e-06, "loss": 0.4807, "step": 1966 }, { "epoch": 0.14, "grad_norm": 3.681225814029964, "learning_rate": 9.68028769063373e-06, "loss": 0.341, "step": 1967 }, { "epoch": 0.14, "grad_norm": 20.185750406074497, "learning_rate": 9.679878958808521e-06, "loss": 1.1166, "step": 1968 }, { "epoch": 0.14, "grad_norm": 5.382258677844892, "learning_rate": 9.679469974521558e-06, "loss": 1.0166, "step": 1969 }, { "epoch": 0.14, "grad_norm": 27.76574381502368, "learning_rate": 9.6790607377949e-06, "loss": 1.2734, "step": 1970 }, { "epoch": 0.14, "grad_norm": 5.55072188066238, "learning_rate": 9.678651248650628e-06, "loss": 0.4432, "step": 1971 }, { "epoch": 0.14, "grad_norm": 4.723559644302558, "learning_rate": 9.678241507110833e-06, "loss": 0.3431, "step": 1972 }, { "epoch": 0.14, "grad_norm": 4.648116468352233, "learning_rate": 9.677831513197614e-06, "loss": 0.2803, "step": 1973 }, { "epoch": 0.14, "grad_norm": 5.056210234157562, "learning_rate": 9.677421266933091e-06, "loss": 0.3809, "step": 1974 }, { "epoch": 0.14, "grad_norm": 3.727812154804654, "learning_rate": 9.677010768339397e-06, "loss": 0.2426, "step": 1975 }, { "epoch": 0.14, "grad_norm": 5.022450750361038, "learning_rate": 9.676600017438675e-06, "loss": 0.4186, "step": 1976 }, { "epoch": 0.14, "grad_norm": 3.762601622428056, "learning_rate": 9.676189014253085e-06, "loss": 0.2559, "step": 1977 }, { "epoch": 0.14, "grad_norm": 17.43002862747351, "learning_rate": 9.675777758804796e-06, "loss": 1.0893, "step": 1978 }, { "epoch": 0.14, "grad_norm": 21.084223633801432, "learning_rate": 9.675366251115998e-06, "loss": 1.933, "step": 1979 }, { "epoch": 0.14, "grad_norm": 7.553559186915883, "learning_rate": 9.674954491208886e-06, "loss": 0.6355, "step": 1980 }, { "epoch": 0.14, "grad_norm": 13.459395336618664, "learning_rate": 9.674542479105676e-06, "loss": 1.353, "step": 1981 }, { "epoch": 0.14, "grad_norm": 6.998278907503723, "learning_rate": 9.674130214828592e-06, "loss": 1.1458, "step": 1982 }, { "epoch": 0.14, "grad_norm": 4.357834585207393, "learning_rate": 9.673717698399878e-06, "loss": 0.2722, "step": 1983 }, { "epoch": 0.14, "grad_norm": 5.037401086380017, "learning_rate": 9.673304929841782e-06, "loss": 0.4409, "step": 1984 }, { "epoch": 0.14, "grad_norm": 4.227294554799323, "learning_rate": 9.672891909176577e-06, "loss": 0.2452, "step": 1985 }, { "epoch": 0.14, "grad_norm": 15.719441928832124, "learning_rate": 9.67247863642654e-06, "loss": 1.033, "step": 1986 }, { "epoch": 0.14, "grad_norm": 3.722476151793504, "learning_rate": 9.672065111613968e-06, "loss": 0.2496, "step": 1987 }, { "epoch": 0.14, "grad_norm": 4.894175144865544, "learning_rate": 9.671651334761167e-06, "loss": 0.3158, "step": 1988 }, { "epoch": 0.14, "grad_norm": 5.423555715421603, "learning_rate": 9.67123730589046e-06, "loss": 1.0404, "step": 1989 }, { "epoch": 0.14, "grad_norm": 3.4670903428609656, "learning_rate": 9.670823025024183e-06, "loss": 0.264, "step": 1990 }, { "epoch": 0.14, "grad_norm": 5.113020696651939, "learning_rate": 9.670408492184682e-06, "loss": 0.9961, "step": 1991 }, { "epoch": 0.14, "grad_norm": 3.793700723109595, "learning_rate": 9.66999370739432e-06, "loss": 0.28, "step": 1992 }, { "epoch": 0.14, "grad_norm": 18.298782986698065, "learning_rate": 9.669578670675476e-06, "loss": 1.0967, "step": 1993 }, { "epoch": 0.14, "grad_norm": 12.728473592419576, "learning_rate": 9.669163382050539e-06, "loss": 1.3356, "step": 1994 }, { "epoch": 0.14, "grad_norm": 21.016373721229225, "learning_rate": 9.668747841541912e-06, "loss": 1.2379, "step": 1995 }, { "epoch": 0.14, "grad_norm": 19.316357282255, "learning_rate": 9.668332049172009e-06, "loss": 1.6906, "step": 1996 }, { "epoch": 0.14, "grad_norm": 4.587276604558562, "learning_rate": 9.667916004963264e-06, "loss": 0.2749, "step": 1997 }, { "epoch": 0.14, "grad_norm": 37.52759065628752, "learning_rate": 9.667499708938119e-06, "loss": 1.4549, "step": 1998 }, { "epoch": 0.14, "grad_norm": 18.222888767241386, "learning_rate": 9.66708316111903e-06, "loss": 1.5248, "step": 1999 }, { "epoch": 0.14, "grad_norm": 4.114027983702515, "learning_rate": 9.666666361528474e-06, "loss": 0.246, "step": 2000 }, { "epoch": 0.14, "grad_norm": 11.669687287094135, "learning_rate": 9.666249310188932e-06, "loss": 1.1417, "step": 2001 }, { "epoch": 0.14, "grad_norm": 3.8164213634695017, "learning_rate": 9.665832007122899e-06, "loss": 0.2351, "step": 2002 }, { "epoch": 0.14, "grad_norm": 3.578866759880426, "learning_rate": 9.665414452352893e-06, "loss": 0.2555, "step": 2003 }, { "epoch": 0.14, "grad_norm": 32.25217812054379, "learning_rate": 9.664996645901436e-06, "loss": 1.8286, "step": 2004 }, { "epoch": 0.14, "grad_norm": 4.144784805389486, "learning_rate": 9.664578587791067e-06, "loss": 0.2539, "step": 2005 }, { "epoch": 0.14, "grad_norm": 20.579277302208915, "learning_rate": 9.66416027804434e-06, "loss": 1.6017, "step": 2006 }, { "epoch": 0.14, "grad_norm": 4.4144087651364625, "learning_rate": 9.663741716683822e-06, "loss": 0.2802, "step": 2007 }, { "epoch": 0.14, "grad_norm": 4.599302575573245, "learning_rate": 9.66332290373209e-06, "loss": 0.3624, "step": 2008 }, { "epoch": 0.14, "grad_norm": 4.974603242175658, "learning_rate": 9.662903839211742e-06, "loss": 0.3273, "step": 2009 }, { "epoch": 0.14, "grad_norm": 4.316348698630639, "learning_rate": 9.662484523145379e-06, "loss": 0.2954, "step": 2010 }, { "epoch": 0.14, "grad_norm": 4.7306470082132055, "learning_rate": 9.662064955555625e-06, "loss": 0.3795, "step": 2011 }, { "epoch": 0.14, "grad_norm": 4.571045297395084, "learning_rate": 9.661645136465114e-06, "loss": 0.2639, "step": 2012 }, { "epoch": 0.14, "grad_norm": 5.518552762328232, "learning_rate": 9.661225065896493e-06, "loss": 1.0746, "step": 2013 }, { "epoch": 0.14, "grad_norm": 3.6431341039185647, "learning_rate": 9.660804743872423e-06, "loss": 0.2033, "step": 2014 }, { "epoch": 0.14, "grad_norm": 5.038808854581303, "learning_rate": 9.66038417041558e-06, "loss": 0.5142, "step": 2015 }, { "epoch": 0.14, "grad_norm": 4.285060202917356, "learning_rate": 9.65996334554865e-06, "loss": 0.2693, "step": 2016 }, { "epoch": 0.14, "grad_norm": 11.142069165404635, "learning_rate": 9.659542269294338e-06, "loss": 1.001, "step": 2017 }, { "epoch": 0.14, "grad_norm": 4.070511644483254, "learning_rate": 9.659120941675357e-06, "loss": 0.3599, "step": 2018 }, { "epoch": 0.14, "grad_norm": 4.245181645865383, "learning_rate": 9.658699362714438e-06, "loss": 0.2816, "step": 2019 }, { "epoch": 0.14, "grad_norm": 15.010056938528939, "learning_rate": 9.658277532434322e-06, "loss": 1.5425, "step": 2020 }, { "epoch": 0.14, "grad_norm": 3.513341499315179, "learning_rate": 9.657855450857765e-06, "loss": 0.2394, "step": 2021 }, { "epoch": 0.14, "grad_norm": 4.0357115111012165, "learning_rate": 9.657433118007538e-06, "loss": 0.2337, "step": 2022 }, { "epoch": 0.15, "grad_norm": 23.277171950488537, "learning_rate": 9.657010533906424e-06, "loss": 1.8257, "step": 2023 }, { "epoch": 0.15, "grad_norm": 5.859859260710464, "learning_rate": 9.656587698577219e-06, "loss": 1.1511, "step": 2024 }, { "epoch": 0.15, "grad_norm": 23.22110923384935, "learning_rate": 9.656164612042734e-06, "loss": 1.5645, "step": 2025 }, { "epoch": 0.15, "grad_norm": 4.942802331405245, "learning_rate": 9.655741274325791e-06, "loss": 0.2712, "step": 2026 }, { "epoch": 0.15, "grad_norm": 6.07962031162716, "learning_rate": 9.655317685449233e-06, "loss": 1.029, "step": 2027 }, { "epoch": 0.15, "grad_norm": 6.046121958901695, "learning_rate": 9.654893845435904e-06, "loss": 1.1736, "step": 2028 }, { "epoch": 0.15, "grad_norm": 15.920302864611502, "learning_rate": 9.654469754308673e-06, "loss": 1.2078, "step": 2029 }, { "epoch": 0.15, "grad_norm": 4.253289363090582, "learning_rate": 9.654045412090415e-06, "loss": 0.2307, "step": 2030 }, { "epoch": 0.15, "grad_norm": 16.01262102059351, "learning_rate": 9.653620818804026e-06, "loss": 1.4296, "step": 2031 }, { "epoch": 0.15, "grad_norm": 5.564301307599986, "learning_rate": 9.653195974472407e-06, "loss": 0.3255, "step": 2032 }, { "epoch": 0.15, "grad_norm": 12.18492377598219, "learning_rate": 9.65277087911848e-06, "loss": 1.2206, "step": 2033 }, { "epoch": 0.15, "grad_norm": 27.95030738522187, "learning_rate": 9.652345532765174e-06, "loss": 0.933, "step": 2034 }, { "epoch": 0.15, "grad_norm": 3.865675867515654, "learning_rate": 9.651919935435438e-06, "loss": 0.258, "step": 2035 }, { "epoch": 0.15, "grad_norm": 19.635409680603253, "learning_rate": 9.65149408715223e-06, "loss": 1.3345, "step": 2036 }, { "epoch": 0.15, "grad_norm": 9.868252175248829, "learning_rate": 9.651067987938522e-06, "loss": 1.0637, "step": 2037 }, { "epoch": 0.15, "grad_norm": 12.887302360159206, "learning_rate": 9.650641637817301e-06, "loss": 0.9296, "step": 2038 }, { "epoch": 0.15, "grad_norm": 21.03686906778295, "learning_rate": 9.650215036811568e-06, "loss": 1.6777, "step": 2039 }, { "epoch": 0.15, "grad_norm": 4.7163312493990235, "learning_rate": 9.649788184944337e-06, "loss": 0.3928, "step": 2040 }, { "epoch": 0.15, "grad_norm": 5.608881073935905, "learning_rate": 9.649361082238633e-06, "loss": 1.1226, "step": 2041 }, { "epoch": 0.15, "grad_norm": 4.969690556816054, "learning_rate": 9.648933728717497e-06, "loss": 0.304, "step": 2042 }, { "epoch": 0.15, "grad_norm": 17.959783830497454, "learning_rate": 9.648506124403982e-06, "loss": 1.383, "step": 2043 }, { "epoch": 0.15, "grad_norm": 7.763308117177654, "learning_rate": 9.64807826932116e-06, "loss": 1.066, "step": 2044 }, { "epoch": 0.15, "grad_norm": 4.115975946646112, "learning_rate": 9.647650163492108e-06, "loss": 0.3039, "step": 2045 }, { "epoch": 0.15, "grad_norm": 5.391840380172674, "learning_rate": 9.647221806939922e-06, "loss": 0.876, "step": 2046 }, { "epoch": 0.15, "grad_norm": 31.35617400039428, "learning_rate": 9.646793199687709e-06, "loss": 0.9623, "step": 2047 }, { "epoch": 0.15, "grad_norm": 5.914439096394987, "learning_rate": 9.646364341758593e-06, "loss": 1.2152, "step": 2048 }, { "epoch": 0.15, "grad_norm": 4.3479879347198365, "learning_rate": 9.64593523317571e-06, "loss": 0.2298, "step": 2049 }, { "epoch": 0.15, "grad_norm": 21.026145083275445, "learning_rate": 9.645505873962205e-06, "loss": 1.2514, "step": 2050 }, { "epoch": 0.15, "grad_norm": 5.2172902611341545, "learning_rate": 9.645076264141242e-06, "loss": 1.0679, "step": 2051 }, { "epoch": 0.15, "grad_norm": 3.317408152457814, "learning_rate": 9.644646403735998e-06, "loss": 0.2264, "step": 2052 }, { "epoch": 0.15, "grad_norm": 5.315414477964954, "learning_rate": 9.644216292769661e-06, "loss": 1.0942, "step": 2053 }, { "epoch": 0.15, "grad_norm": 4.970904329894231, "learning_rate": 9.643785931265435e-06, "loss": 1.056, "step": 2054 }, { "epoch": 0.15, "grad_norm": 5.06074789357795, "learning_rate": 9.643355319246535e-06, "loss": 0.2942, "step": 2055 }, { "epoch": 0.15, "grad_norm": 4.775906608290938, "learning_rate": 9.64292445673619e-06, "loss": 0.3269, "step": 2056 }, { "epoch": 0.15, "grad_norm": 23.655167086046045, "learning_rate": 9.642493343757648e-06, "loss": 1.0322, "step": 2057 }, { "epoch": 0.15, "grad_norm": 146.58243041970462, "learning_rate": 9.642061980334161e-06, "loss": 1.671, "step": 2058 }, { "epoch": 0.15, "grad_norm": 5.166645058243217, "learning_rate": 9.641630366489003e-06, "loss": 0.4292, "step": 2059 }, { "epoch": 0.15, "grad_norm": 6.330649493305766, "learning_rate": 9.641198502245455e-06, "loss": 0.9241, "step": 2060 }, { "epoch": 0.15, "grad_norm": 5.121570879688442, "learning_rate": 9.640766387626815e-06, "loss": 0.3462, "step": 2061 }, { "epoch": 0.15, "grad_norm": 14.719661116106291, "learning_rate": 9.640334022656396e-06, "loss": 1.2949, "step": 2062 }, { "epoch": 0.15, "grad_norm": 15.09685821540329, "learning_rate": 9.639901407357521e-06, "loss": 0.9373, "step": 2063 }, { "epoch": 0.15, "grad_norm": 5.491960801208608, "learning_rate": 9.639468541753527e-06, "loss": 1.1525, "step": 2064 }, { "epoch": 0.15, "grad_norm": 20.047866682545063, "learning_rate": 9.639035425867767e-06, "loss": 1.3472, "step": 2065 }, { "epoch": 0.15, "grad_norm": 27.93108753213777, "learning_rate": 9.638602059723607e-06, "loss": 1.0867, "step": 2066 }, { "epoch": 0.15, "grad_norm": 4.256719052515827, "learning_rate": 9.638168443344422e-06, "loss": 0.3088, "step": 2067 }, { "epoch": 0.15, "grad_norm": 19.763446869704595, "learning_rate": 9.637734576753606e-06, "loss": 1.4745, "step": 2068 }, { "epoch": 0.15, "grad_norm": 19.667386695408695, "learning_rate": 9.637300459974566e-06, "loss": 1.1977, "step": 2069 }, { "epoch": 0.15, "grad_norm": 4.259933926657804, "learning_rate": 9.636866093030717e-06, "loss": 0.1991, "step": 2070 }, { "epoch": 0.15, "grad_norm": 10.75597082294718, "learning_rate": 9.636431475945495e-06, "loss": 0.8964, "step": 2071 }, { "epoch": 0.15, "grad_norm": 4.138952957402652, "learning_rate": 9.635996608742345e-06, "loss": 0.2655, "step": 2072 }, { "epoch": 0.15, "grad_norm": 63.977176614484534, "learning_rate": 9.635561491444726e-06, "loss": 1.3559, "step": 2073 }, { "epoch": 0.15, "grad_norm": 14.34380283204377, "learning_rate": 9.63512612407611e-06, "loss": 1.6576, "step": 2074 }, { "epoch": 0.15, "grad_norm": 13.410061876657954, "learning_rate": 9.634690506659986e-06, "loss": 1.1382, "step": 2075 }, { "epoch": 0.15, "grad_norm": 4.355924268274909, "learning_rate": 9.634254639219852e-06, "loss": 0.2769, "step": 2076 }, { "epoch": 0.15, "grad_norm": 4.366776792032409, "learning_rate": 9.633818521779221e-06, "loss": 0.3479, "step": 2077 }, { "epoch": 0.15, "grad_norm": 23.181696671942845, "learning_rate": 9.633382154361623e-06, "loss": 1.7554, "step": 2078 }, { "epoch": 0.15, "grad_norm": 4.423685753304198, "learning_rate": 9.632945536990595e-06, "loss": 0.2788, "step": 2079 }, { "epoch": 0.15, "grad_norm": 6.23100372150776, "learning_rate": 9.632508669689691e-06, "loss": 1.0036, "step": 2080 }, { "epoch": 0.15, "grad_norm": 4.860976512492658, "learning_rate": 9.632071552482479e-06, "loss": 0.2948, "step": 2081 }, { "epoch": 0.15, "grad_norm": 10.644346544104458, "learning_rate": 9.631634185392541e-06, "loss": 0.7393, "step": 2082 }, { "epoch": 0.15, "grad_norm": 4.052690119405541, "learning_rate": 9.631196568443471e-06, "loss": 0.2654, "step": 2083 }, { "epoch": 0.15, "grad_norm": 3.4009264277267603, "learning_rate": 9.630758701658874e-06, "loss": 0.2794, "step": 2084 }, { "epoch": 0.15, "grad_norm": 24.36335197105553, "learning_rate": 9.630320585062375e-06, "loss": 1.7592, "step": 2085 }, { "epoch": 0.15, "grad_norm": 4.858905877109441, "learning_rate": 9.629882218677606e-06, "loss": 0.3083, "step": 2086 }, { "epoch": 0.15, "grad_norm": 4.449628617190748, "learning_rate": 9.629443602528217e-06, "loss": 0.2138, "step": 2087 }, { "epoch": 0.15, "grad_norm": 7.605712236404369, "learning_rate": 9.62900473663787e-06, "loss": 1.0227, "step": 2088 }, { "epoch": 0.15, "grad_norm": 5.07881440166514, "learning_rate": 9.62856562103024e-06, "loss": 0.3566, "step": 2089 }, { "epoch": 0.15, "grad_norm": 4.796400963928402, "learning_rate": 9.628126255729013e-06, "loss": 0.2849, "step": 2090 }, { "epoch": 0.15, "grad_norm": 9.43073401027991, "learning_rate": 9.627686640757891e-06, "loss": 0.7816, "step": 2091 }, { "epoch": 0.15, "grad_norm": 4.048441024288282, "learning_rate": 9.627246776140595e-06, "loss": 0.2317, "step": 2092 }, { "epoch": 0.15, "grad_norm": 48.361460974010406, "learning_rate": 9.62680666190085e-06, "loss": 1.5941, "step": 2093 }, { "epoch": 0.15, "grad_norm": 4.993976273639079, "learning_rate": 9.626366298062397e-06, "loss": 0.3182, "step": 2094 }, { "epoch": 0.15, "grad_norm": 6.408548390935406, "learning_rate": 9.625925684648995e-06, "loss": 1.0641, "step": 2095 }, { "epoch": 0.15, "grad_norm": 4.7271050857045305, "learning_rate": 9.625484821684413e-06, "loss": 0.2512, "step": 2096 }, { "epoch": 0.15, "grad_norm": 4.974381899869801, "learning_rate": 9.625043709192433e-06, "loss": 0.2582, "step": 2097 }, { "epoch": 0.15, "grad_norm": 5.1928183468623095, "learning_rate": 9.624602347196852e-06, "loss": 0.2868, "step": 2098 }, { "epoch": 0.15, "grad_norm": 5.1712374821890155, "learning_rate": 9.624160735721479e-06, "loss": 0.2851, "step": 2099 }, { "epoch": 0.15, "grad_norm": 28.46567167796357, "learning_rate": 9.623718874790137e-06, "loss": 1.5244, "step": 2100 }, { "epoch": 0.15, "grad_norm": 12.862733554892149, "learning_rate": 9.623276764426663e-06, "loss": 1.3344, "step": 2101 }, { "epoch": 0.15, "grad_norm": 18.230549868958143, "learning_rate": 9.62283440465491e-06, "loss": 1.6537, "step": 2102 }, { "epoch": 0.15, "grad_norm": 17.51565282649639, "learning_rate": 9.622391795498738e-06, "loss": 1.3221, "step": 2103 }, { "epoch": 0.15, "grad_norm": 3.599525965399324, "learning_rate": 9.621948936982024e-06, "loss": 0.1855, "step": 2104 }, { "epoch": 0.15, "grad_norm": 3.870174304242464, "learning_rate": 9.621505829128661e-06, "loss": 0.2221, "step": 2105 }, { "epoch": 0.15, "grad_norm": 4.412923864748455, "learning_rate": 9.621062471962553e-06, "loss": 0.2622, "step": 2106 }, { "epoch": 0.15, "grad_norm": 4.534097865892773, "learning_rate": 9.620618865507615e-06, "loss": 0.3702, "step": 2107 }, { "epoch": 0.15, "grad_norm": 10.367962678212463, "learning_rate": 9.62017500978778e-06, "loss": 1.4266, "step": 2108 }, { "epoch": 0.15, "grad_norm": 19.078524189917253, "learning_rate": 9.61973090482699e-06, "loss": 1.5033, "step": 2109 }, { "epoch": 0.15, "grad_norm": 3.700877506026893, "learning_rate": 9.619286550649205e-06, "loss": 0.3138, "step": 2110 }, { "epoch": 0.15, "grad_norm": 5.923972148792611, "learning_rate": 9.618841947278396e-06, "loss": 1.0844, "step": 2111 }, { "epoch": 0.15, "grad_norm": 3.9831881212785163, "learning_rate": 9.618397094738547e-06, "loss": 0.2486, "step": 2112 }, { "epoch": 0.15, "grad_norm": 4.053073478314815, "learning_rate": 9.617951993053658e-06, "loss": 0.2364, "step": 2113 }, { "epoch": 0.15, "grad_norm": 26.076744624541117, "learning_rate": 9.617506642247735e-06, "loss": 1.6619, "step": 2114 }, { "epoch": 0.15, "grad_norm": 22.43104111016753, "learning_rate": 9.617061042344812e-06, "loss": 1.234, "step": 2115 }, { "epoch": 0.15, "grad_norm": 29.518258213245158, "learning_rate": 9.616615193368917e-06, "loss": 1.9504, "step": 2116 }, { "epoch": 0.15, "grad_norm": 5.628151395555626, "learning_rate": 9.61616909534411e-06, "loss": 0.3011, "step": 2117 }, { "epoch": 0.15, "grad_norm": 6.146304164089872, "learning_rate": 9.615722748294454e-06, "loss": 1.0785, "step": 2118 }, { "epoch": 0.15, "grad_norm": 5.797899136454844, "learning_rate": 9.615276152244027e-06, "loss": 1.1425, "step": 2119 }, { "epoch": 0.15, "grad_norm": 4.912014577761085, "learning_rate": 9.61482930721692e-06, "loss": 0.3448, "step": 2120 }, { "epoch": 0.15, "grad_norm": 6.460791716677355, "learning_rate": 9.61438221323724e-06, "loss": 0.5963, "step": 2121 }, { "epoch": 0.15, "grad_norm": 4.785264321916628, "learning_rate": 9.613934870329107e-06, "loss": 1.1906, "step": 2122 }, { "epoch": 0.15, "grad_norm": 16.590273009893856, "learning_rate": 9.61348727851665e-06, "loss": 1.3946, "step": 2123 }, { "epoch": 0.15, "grad_norm": 13.419782537642263, "learning_rate": 9.613039437824022e-06, "loss": 1.3941, "step": 2124 }, { "epoch": 0.15, "grad_norm": 5.281648304517405, "learning_rate": 9.612591348275373e-06, "loss": 0.9349, "step": 2125 }, { "epoch": 0.15, "grad_norm": 29.076444256625415, "learning_rate": 9.612143009894883e-06, "loss": 1.6662, "step": 2126 }, { "epoch": 0.15, "grad_norm": 3.7486762901569826, "learning_rate": 9.611694422706734e-06, "loss": 0.208, "step": 2127 }, { "epoch": 0.15, "grad_norm": 4.760267873580344, "learning_rate": 9.611245586735128e-06, "loss": 0.3663, "step": 2128 }, { "epoch": 0.15, "grad_norm": 7.214638259525724, "learning_rate": 9.610796502004278e-06, "loss": 1.0091, "step": 2129 }, { "epoch": 0.15, "grad_norm": 4.596106819161184, "learning_rate": 9.610347168538408e-06, "loss": 0.3869, "step": 2130 }, { "epoch": 0.15, "grad_norm": 4.7995131424249315, "learning_rate": 9.60989758636176e-06, "loss": 0.4001, "step": 2131 }, { "epoch": 0.15, "grad_norm": 3.8012807181749335, "learning_rate": 9.609447755498586e-06, "loss": 0.2229, "step": 2132 }, { "epoch": 0.15, "grad_norm": 6.620762066982747, "learning_rate": 9.608997675973155e-06, "loss": 1.0433, "step": 2133 }, { "epoch": 0.15, "grad_norm": 4.042181355221697, "learning_rate": 9.608547347809744e-06, "loss": 0.2567, "step": 2134 }, { "epoch": 0.15, "grad_norm": 3.8826941602803178, "learning_rate": 9.608096771032648e-06, "loss": 0.2119, "step": 2135 }, { "epoch": 0.15, "grad_norm": 4.178514462949838, "learning_rate": 9.607645945666174e-06, "loss": 0.2619, "step": 2136 }, { "epoch": 0.15, "grad_norm": 4.491965425170396, "learning_rate": 9.607194871734643e-06, "loss": 0.2293, "step": 2137 }, { "epoch": 0.15, "grad_norm": 4.357731628415745, "learning_rate": 9.606743549262386e-06, "loss": 0.2284, "step": 2138 }, { "epoch": 0.15, "grad_norm": 5.015473594877808, "learning_rate": 9.606291978273752e-06, "loss": 0.4379, "step": 2139 }, { "epoch": 0.15, "grad_norm": 5.147123062534966, "learning_rate": 9.605840158793102e-06, "loss": 0.4441, "step": 2140 }, { "epoch": 0.15, "grad_norm": 13.004911840142565, "learning_rate": 9.60538809084481e-06, "loss": 0.7459, "step": 2141 }, { "epoch": 0.15, "grad_norm": 4.198480042660495, "learning_rate": 9.604935774453263e-06, "loss": 0.2684, "step": 2142 }, { "epoch": 0.15, "grad_norm": 6.787895734996168, "learning_rate": 9.604483209642861e-06, "loss": 1.2286, "step": 2143 }, { "epoch": 0.15, "grad_norm": 5.654459691718224, "learning_rate": 9.60403039643802e-06, "loss": 1.2122, "step": 2144 }, { "epoch": 0.15, "grad_norm": 4.180237672042812, "learning_rate": 9.603577334863166e-06, "loss": 0.259, "step": 2145 }, { "epoch": 0.15, "grad_norm": 4.054323986531529, "learning_rate": 9.603124024942739e-06, "loss": 0.3097, "step": 2146 }, { "epoch": 0.15, "grad_norm": 15.964730159917568, "learning_rate": 9.602670466701195e-06, "loss": 0.8417, "step": 2147 }, { "epoch": 0.15, "grad_norm": 6.034481494264666, "learning_rate": 9.602216660163001e-06, "loss": 0.9585, "step": 2148 }, { "epoch": 0.15, "grad_norm": 3.6964118949927323, "learning_rate": 9.601762605352641e-06, "loss": 0.2608, "step": 2149 }, { "epoch": 0.15, "grad_norm": 9.181304399596776, "learning_rate": 9.601308302294605e-06, "loss": 1.0255, "step": 2150 }, { "epoch": 0.15, "grad_norm": 9.722278158803338, "learning_rate": 9.600853751013402e-06, "loss": 1.0793, "step": 2151 }, { "epoch": 0.15, "grad_norm": 5.6086691080977475, "learning_rate": 9.600398951533557e-06, "loss": 1.0962, "step": 2152 }, { "epoch": 0.15, "grad_norm": 4.3696131448635835, "learning_rate": 9.599943903879599e-06, "loss": 0.3097, "step": 2153 }, { "epoch": 0.15, "grad_norm": 15.49184395017155, "learning_rate": 9.59948860807608e-06, "loss": 1.3488, "step": 2154 }, { "epoch": 0.15, "grad_norm": 5.569257587713978, "learning_rate": 9.599033064147563e-06, "loss": 0.3662, "step": 2155 }, { "epoch": 0.15, "grad_norm": 4.274761228074361, "learning_rate": 9.598577272118618e-06, "loss": 0.3318, "step": 2156 }, { "epoch": 0.15, "grad_norm": 4.9418568072682785, "learning_rate": 9.598121232013837e-06, "loss": 0.4725, "step": 2157 }, { "epoch": 0.15, "grad_norm": 4.840078883419235, "learning_rate": 9.59766494385782e-06, "loss": 0.2671, "step": 2158 }, { "epoch": 0.15, "grad_norm": 13.46319992444348, "learning_rate": 9.597208407675181e-06, "loss": 1.2802, "step": 2159 }, { "epoch": 0.15, "grad_norm": 3.8551320575643007, "learning_rate": 9.596751623490553e-06, "loss": 0.2231, "step": 2160 }, { "epoch": 0.15, "grad_norm": 3.9536984942698137, "learning_rate": 9.596294591328571e-06, "loss": 0.1973, "step": 2161 }, { "epoch": 0.16, "grad_norm": 12.64085105621602, "learning_rate": 9.595837311213896e-06, "loss": 0.9517, "step": 2162 }, { "epoch": 0.16, "grad_norm": 5.4477384726087035, "learning_rate": 9.595379783171192e-06, "loss": 0.9952, "step": 2163 }, { "epoch": 0.16, "grad_norm": 8.975709093125042, "learning_rate": 9.594922007225144e-06, "loss": 0.7792, "step": 2164 }, { "epoch": 0.16, "grad_norm": 3.757763033320141, "learning_rate": 9.594463983400449e-06, "loss": 0.3026, "step": 2165 }, { "epoch": 0.16, "grad_norm": 4.520504908215212, "learning_rate": 9.594005711721811e-06, "loss": 0.2315, "step": 2166 }, { "epoch": 0.16, "grad_norm": 12.840897490878271, "learning_rate": 9.593547192213955e-06, "loss": 1.3753, "step": 2167 }, { "epoch": 0.16, "grad_norm": 12.363268968821615, "learning_rate": 9.593088424901615e-06, "loss": 1.0912, "step": 2168 }, { "epoch": 0.16, "grad_norm": 9.087406079696676, "learning_rate": 9.59262940980954e-06, "loss": 1.1975, "step": 2169 }, { "epoch": 0.16, "grad_norm": 4.50003891867464, "learning_rate": 9.592170146962492e-06, "loss": 0.3088, "step": 2170 }, { "epoch": 0.16, "grad_norm": 6.117574581060358, "learning_rate": 9.591710636385248e-06, "loss": 0.3265, "step": 2171 }, { "epoch": 0.16, "grad_norm": 12.178741316934062, "learning_rate": 9.591250878102594e-06, "loss": 0.7754, "step": 2172 }, { "epoch": 0.16, "grad_norm": 6.237307002695604, "learning_rate": 9.590790872139335e-06, "loss": 0.9163, "step": 2173 }, { "epoch": 0.16, "grad_norm": 13.706625993650823, "learning_rate": 9.590330618520284e-06, "loss": 1.0684, "step": 2174 }, { "epoch": 0.16, "grad_norm": 4.572286324068097, "learning_rate": 9.589870117270274e-06, "loss": 0.3455, "step": 2175 }, { "epoch": 0.16, "grad_norm": 6.213533094660169, "learning_rate": 9.589409368414144e-06, "loss": 1.2535, "step": 2176 }, { "epoch": 0.16, "grad_norm": 4.9363622181163995, "learning_rate": 9.58894837197675e-06, "loss": 0.3324, "step": 2177 }, { "epoch": 0.16, "grad_norm": 29.6424454253752, "learning_rate": 9.588487127982962e-06, "loss": 1.4149, "step": 2178 }, { "epoch": 0.16, "grad_norm": 4.9306968764912735, "learning_rate": 9.588025636457662e-06, "loss": 0.2881, "step": 2179 }, { "epoch": 0.16, "grad_norm": 7.445712143541607, "learning_rate": 9.587563897425744e-06, "loss": 0.96, "step": 2180 }, { "epoch": 0.16, "grad_norm": 5.046911015319469, "learning_rate": 9.587101910912121e-06, "loss": 0.2316, "step": 2181 }, { "epoch": 0.16, "grad_norm": 20.628357970763535, "learning_rate": 9.586639676941712e-06, "loss": 1.1197, "step": 2182 }, { "epoch": 0.16, "grad_norm": 26.142908657883694, "learning_rate": 9.586177195539452e-06, "loss": 1.4661, "step": 2183 }, { "epoch": 0.16, "grad_norm": 5.623606969156054, "learning_rate": 9.585714466730294e-06, "loss": 0.2823, "step": 2184 }, { "epoch": 0.16, "grad_norm": 5.459381132159393, "learning_rate": 9.585251490539198e-06, "loss": 0.3318, "step": 2185 }, { "epoch": 0.16, "grad_norm": 22.175830198985953, "learning_rate": 9.58478826699114e-06, "loss": 1.1869, "step": 2186 }, { "epoch": 0.16, "grad_norm": 4.964060074239939, "learning_rate": 9.58432479611111e-06, "loss": 0.4087, "step": 2187 }, { "epoch": 0.16, "grad_norm": 4.190470087059749, "learning_rate": 9.583861077924109e-06, "loss": 0.2318, "step": 2188 }, { "epoch": 0.16, "grad_norm": 33.449303509947605, "learning_rate": 9.583397112455152e-06, "loss": 0.9284, "step": 2189 }, { "epoch": 0.16, "grad_norm": 4.152355366786511, "learning_rate": 9.582932899729274e-06, "loss": 0.3046, "step": 2190 }, { "epoch": 0.16, "grad_norm": 4.9494334622074785, "learning_rate": 9.582468439771511e-06, "loss": 1.1826, "step": 2191 }, { "epoch": 0.16, "grad_norm": 4.238502715866378, "learning_rate": 9.582003732606921e-06, "loss": 0.3161, "step": 2192 }, { "epoch": 0.16, "grad_norm": 5.501134761109598, "learning_rate": 9.581538778260574e-06, "loss": 1.1154, "step": 2193 }, { "epoch": 0.16, "grad_norm": 4.073311320399983, "learning_rate": 9.581073576757551e-06, "loss": 0.3497, "step": 2194 }, { "epoch": 0.16, "grad_norm": 3.860138189874805, "learning_rate": 9.58060812812295e-06, "loss": 0.2775, "step": 2195 }, { "epoch": 0.16, "grad_norm": 5.684499311615184, "learning_rate": 9.580142432381881e-06, "loss": 1.0086, "step": 2196 }, { "epoch": 0.16, "grad_norm": 3.71629868597286, "learning_rate": 9.579676489559462e-06, "loss": 0.2361, "step": 2197 }, { "epoch": 0.16, "grad_norm": 4.297174547927817, "learning_rate": 9.57921029968083e-06, "loss": 0.33, "step": 2198 }, { "epoch": 0.16, "grad_norm": 3.8316958944330173, "learning_rate": 9.578743862771137e-06, "loss": 0.3321, "step": 2199 }, { "epoch": 0.16, "grad_norm": 5.579780912091275, "learning_rate": 9.578277178855545e-06, "loss": 1.0257, "step": 2200 }, { "epoch": 0.16, "grad_norm": 5.036273898997859, "learning_rate": 9.57781024795923e-06, "loss": 0.2648, "step": 2201 }, { "epoch": 0.16, "grad_norm": 30.152611079742094, "learning_rate": 9.577343070107377e-06, "loss": 1.7156, "step": 2202 }, { "epoch": 0.16, "grad_norm": 5.0633457546521665, "learning_rate": 9.576875645325193e-06, "loss": 0.2556, "step": 2203 }, { "epoch": 0.16, "grad_norm": 5.143537469469841, "learning_rate": 9.576407973637892e-06, "loss": 0.3533, "step": 2204 }, { "epoch": 0.16, "grad_norm": 4.780456644302459, "learning_rate": 9.575940055070705e-06, "loss": 0.2716, "step": 2205 }, { "epoch": 0.16, "grad_norm": 42.81959272925734, "learning_rate": 9.575471889648871e-06, "loss": 1.7869, "step": 2206 }, { "epoch": 0.16, "grad_norm": 4.753147171369584, "learning_rate": 9.57500347739765e-06, "loss": 0.2036, "step": 2207 }, { "epoch": 0.16, "grad_norm": 26.462101198307565, "learning_rate": 9.574534818342304e-06, "loss": 1.6812, "step": 2208 }, { "epoch": 0.16, "grad_norm": 6.024593273638091, "learning_rate": 9.574065912508124e-06, "loss": 0.2988, "step": 2209 }, { "epoch": 0.16, "grad_norm": 4.786248048387668, "learning_rate": 9.573596759920401e-06, "loss": 0.2468, "step": 2210 }, { "epoch": 0.16, "grad_norm": 6.310851544106439, "learning_rate": 9.573127360604445e-06, "loss": 1.1513, "step": 2211 }, { "epoch": 0.16, "grad_norm": 6.289312034059576, "learning_rate": 9.572657714585578e-06, "loss": 0.4178, "step": 2212 }, { "epoch": 0.16, "grad_norm": 14.582800485916243, "learning_rate": 9.572187821889136e-06, "loss": 0.9432, "step": 2213 }, { "epoch": 0.16, "grad_norm": 25.007247713812724, "learning_rate": 9.571717682540467e-06, "loss": 1.9315, "step": 2214 }, { "epoch": 0.16, "grad_norm": 34.520879221511656, "learning_rate": 9.571247296564933e-06, "loss": 1.3885, "step": 2215 }, { "epoch": 0.16, "grad_norm": 11.426485239912026, "learning_rate": 9.570776663987914e-06, "loss": 1.1363, "step": 2216 }, { "epoch": 0.16, "grad_norm": 4.858178819217395, "learning_rate": 9.570305784834792e-06, "loss": 0.4128, "step": 2217 }, { "epoch": 0.16, "grad_norm": 4.1099784586681665, "learning_rate": 9.569834659130974e-06, "loss": 0.2517, "step": 2218 }, { "epoch": 0.16, "grad_norm": 4.090809354145801, "learning_rate": 9.569363286901873e-06, "loss": 0.2963, "step": 2219 }, { "epoch": 0.16, "grad_norm": 5.6069534054759, "learning_rate": 9.568891668172919e-06, "loss": 0.2267, "step": 2220 }, { "epoch": 0.16, "grad_norm": 4.296794631090411, "learning_rate": 9.568419802969553e-06, "loss": 0.3593, "step": 2221 }, { "epoch": 0.16, "grad_norm": 16.305342653092577, "learning_rate": 9.567947691317231e-06, "loss": 1.6352, "step": 2222 }, { "epoch": 0.16, "grad_norm": 7.846746257331388, "learning_rate": 9.567475333241423e-06, "loss": 0.7932, "step": 2223 }, { "epoch": 0.16, "grad_norm": 11.285005775798815, "learning_rate": 9.567002728767609e-06, "loss": 1.0349, "step": 2224 }, { "epoch": 0.16, "grad_norm": 28.140452828236786, "learning_rate": 9.566529877921284e-06, "loss": 1.7522, "step": 2225 }, { "epoch": 0.16, "grad_norm": 6.9550994733255, "learning_rate": 9.56605678072796e-06, "loss": 1.0105, "step": 2226 }, { "epoch": 0.16, "grad_norm": 18.345056103312317, "learning_rate": 9.565583437213154e-06, "loss": 1.1075, "step": 2227 }, { "epoch": 0.16, "grad_norm": 4.4129875968680885, "learning_rate": 9.565109847402403e-06, "loss": 0.2607, "step": 2228 }, { "epoch": 0.16, "grad_norm": 4.807916485103817, "learning_rate": 9.564636011321256e-06, "loss": 0.3228, "step": 2229 }, { "epoch": 0.16, "grad_norm": 4.1846345315545905, "learning_rate": 9.564161928995275e-06, "loss": 0.2722, "step": 2230 }, { "epoch": 0.16, "grad_norm": 6.40835437208897, "learning_rate": 9.563687600450034e-06, "loss": 1.2226, "step": 2231 }, { "epoch": 0.16, "grad_norm": 18.880045221851926, "learning_rate": 9.563213025711122e-06, "loss": 1.6371, "step": 2232 }, { "epoch": 0.16, "grad_norm": 4.865283345162097, "learning_rate": 9.562738204804138e-06, "loss": 0.3302, "step": 2233 }, { "epoch": 0.16, "grad_norm": 5.7176730982031785, "learning_rate": 9.5622631377547e-06, "loss": 0.2832, "step": 2234 }, { "epoch": 0.16, "grad_norm": 31.148465456115368, "learning_rate": 9.561787824588434e-06, "loss": 1.4567, "step": 2235 }, { "epoch": 0.16, "grad_norm": 4.569280801830411, "learning_rate": 9.561312265330983e-06, "loss": 0.3077, "step": 2236 }, { "epoch": 0.16, "grad_norm": 5.803927115176648, "learning_rate": 9.560836460008e-06, "loss": 0.3529, "step": 2237 }, { "epoch": 0.16, "grad_norm": 4.495075353415189, "learning_rate": 9.560360408645155e-06, "loss": 0.2463, "step": 2238 }, { "epoch": 0.16, "grad_norm": 14.515174272935697, "learning_rate": 9.559884111268128e-06, "loss": 1.0119, "step": 2239 }, { "epoch": 0.16, "grad_norm": 4.209476751288751, "learning_rate": 9.559407567902612e-06, "loss": 0.3858, "step": 2240 }, { "epoch": 0.16, "grad_norm": 3.376001672198274, "learning_rate": 9.558930778574316e-06, "loss": 0.2304, "step": 2241 }, { "epoch": 0.16, "grad_norm": 21.428979481208387, "learning_rate": 9.558453743308962e-06, "loss": 1.078, "step": 2242 }, { "epoch": 0.16, "grad_norm": 4.241988558432934, "learning_rate": 9.557976462132282e-06, "loss": 0.2297, "step": 2243 }, { "epoch": 0.16, "grad_norm": 6.17794675879827, "learning_rate": 9.557498935070027e-06, "loss": 1.2098, "step": 2244 }, { "epoch": 0.16, "grad_norm": 4.9004927261105875, "learning_rate": 9.557021162147952e-06, "loss": 0.9925, "step": 2245 }, { "epoch": 0.16, "grad_norm": 5.666511893851699, "learning_rate": 9.556543143391836e-06, "loss": 1.1719, "step": 2246 }, { "epoch": 0.16, "grad_norm": 4.256898242662345, "learning_rate": 9.556064878827467e-06, "loss": 0.2696, "step": 2247 }, { "epoch": 0.16, "grad_norm": 4.5665977127479636, "learning_rate": 9.555586368480642e-06, "loss": 0.3339, "step": 2248 }, { "epoch": 0.16, "grad_norm": 5.746025976596114, "learning_rate": 9.555107612377174e-06, "loss": 1.0468, "step": 2249 }, { "epoch": 0.16, "grad_norm": 23.26544341854004, "learning_rate": 9.554628610542895e-06, "loss": 1.1873, "step": 2250 }, { "epoch": 0.16, "grad_norm": 4.736913247542832, "learning_rate": 9.554149363003643e-06, "loss": 0.4579, "step": 2251 }, { "epoch": 0.16, "grad_norm": 15.768746366748653, "learning_rate": 9.55366986978527e-06, "loss": 1.4366, "step": 2252 }, { "epoch": 0.16, "grad_norm": 15.80121907956133, "learning_rate": 9.553190130913646e-06, "loss": 1.2532, "step": 2253 }, { "epoch": 0.16, "grad_norm": 13.328616589900546, "learning_rate": 9.552710146414646e-06, "loss": 1.1618, "step": 2254 }, { "epoch": 0.16, "grad_norm": 4.570310918622819, "learning_rate": 9.552229916314168e-06, "loss": 0.3959, "step": 2255 }, { "epoch": 0.16, "grad_norm": 26.66590769221558, "learning_rate": 9.55174944063812e-06, "loss": 1.516, "step": 2256 }, { "epoch": 0.16, "grad_norm": 5.486302260850816, "learning_rate": 9.551268719412414e-06, "loss": 0.356, "step": 2257 }, { "epoch": 0.16, "grad_norm": 4.805208327524164, "learning_rate": 9.550787752662992e-06, "loss": 0.3207, "step": 2258 }, { "epoch": 0.16, "grad_norm": 12.455925741730866, "learning_rate": 9.550306540415793e-06, "loss": 0.8817, "step": 2259 }, { "epoch": 0.16, "grad_norm": 14.941562800878623, "learning_rate": 9.549825082696782e-06, "loss": 1.2664, "step": 2260 }, { "epoch": 0.16, "grad_norm": 4.703267667446763, "learning_rate": 9.549343379531931e-06, "loss": 0.9352, "step": 2261 }, { "epoch": 0.16, "grad_norm": 20.526338729881125, "learning_rate": 9.548861430947222e-06, "loss": 1.8405, "step": 2262 }, { "epoch": 0.16, "grad_norm": 21.82875006896252, "learning_rate": 9.548379236968659e-06, "loss": 0.9321, "step": 2263 }, { "epoch": 0.16, "grad_norm": 5.949578251844159, "learning_rate": 9.547896797622252e-06, "loss": 1.1707, "step": 2264 }, { "epoch": 0.16, "grad_norm": 4.025471952620486, "learning_rate": 9.547414112934028e-06, "loss": 0.2758, "step": 2265 }, { "epoch": 0.16, "grad_norm": 4.291611517823496, "learning_rate": 9.546931182930024e-06, "loss": 0.2429, "step": 2266 }, { "epoch": 0.16, "grad_norm": 5.853501056468595, "learning_rate": 9.546448007636295e-06, "loss": 1.0059, "step": 2267 }, { "epoch": 0.16, "grad_norm": 80.7407410094363, "learning_rate": 9.545964587078906e-06, "loss": 1.3418, "step": 2268 }, { "epoch": 0.16, "grad_norm": 99.5225708776974, "learning_rate": 9.545480921283934e-06, "loss": 1.0338, "step": 2269 }, { "epoch": 0.16, "grad_norm": 16.50885062917954, "learning_rate": 9.544997010277472e-06, "loss": 1.4423, "step": 2270 }, { "epoch": 0.16, "grad_norm": 4.040947614781685, "learning_rate": 9.544512854085626e-06, "loss": 0.262, "step": 2271 }, { "epoch": 0.16, "grad_norm": 5.107492777005974, "learning_rate": 9.544028452734513e-06, "loss": 0.9681, "step": 2272 }, { "epoch": 0.16, "grad_norm": 5.791864179377169, "learning_rate": 9.543543806250264e-06, "loss": 1.1866, "step": 2273 }, { "epoch": 0.16, "grad_norm": 12.118896095608445, "learning_rate": 9.543058914659026e-06, "loss": 0.7247, "step": 2274 }, { "epoch": 0.16, "grad_norm": 54.532892011686734, "learning_rate": 9.542573777986955e-06, "loss": 1.2825, "step": 2275 }, { "epoch": 0.16, "grad_norm": 5.546351746683935, "learning_rate": 9.542088396260224e-06, "loss": 1.048, "step": 2276 }, { "epoch": 0.16, "grad_norm": 20.43081149205147, "learning_rate": 9.541602769505017e-06, "loss": 1.3003, "step": 2277 }, { "epoch": 0.16, "grad_norm": 7.3569255231839294, "learning_rate": 9.541116897747532e-06, "loss": 0.4398, "step": 2278 }, { "epoch": 0.16, "grad_norm": 4.278335385763088, "learning_rate": 9.540630781013978e-06, "loss": 0.2741, "step": 2279 }, { "epoch": 0.16, "grad_norm": 6.258723725265775, "learning_rate": 9.54014441933058e-06, "loss": 1.0504, "step": 2280 }, { "epoch": 0.16, "grad_norm": 4.480861546748723, "learning_rate": 9.539657812723578e-06, "loss": 0.3674, "step": 2281 }, { "epoch": 0.16, "grad_norm": 3.6882745464377167, "learning_rate": 9.539170961219219e-06, "loss": 0.2315, "step": 2282 }, { "epoch": 0.16, "grad_norm": 15.056873469188686, "learning_rate": 9.538683864843769e-06, "loss": 1.1658, "step": 2283 }, { "epoch": 0.16, "grad_norm": 13.16046520164018, "learning_rate": 9.538196523623505e-06, "loss": 1.366, "step": 2284 }, { "epoch": 0.16, "grad_norm": 28.09278305571031, "learning_rate": 9.537708937584714e-06, "loss": 0.7763, "step": 2285 }, { "epoch": 0.16, "grad_norm": 9.430231726228708, "learning_rate": 9.537221106753705e-06, "loss": 1.0045, "step": 2286 }, { "epoch": 0.16, "grad_norm": 4.0817904887389505, "learning_rate": 9.536733031156789e-06, "loss": 0.2351, "step": 2287 }, { "epoch": 0.16, "grad_norm": 3.1923488047885673, "learning_rate": 9.5362447108203e-06, "loss": 0.2493, "step": 2288 }, { "epoch": 0.16, "grad_norm": 21.65713632839914, "learning_rate": 9.53575614577058e-06, "loss": 0.7515, "step": 2289 }, { "epoch": 0.16, "grad_norm": 15.75071580974788, "learning_rate": 9.535267336033984e-06, "loss": 1.0232, "step": 2290 }, { "epoch": 0.16, "grad_norm": 5.198540439665363, "learning_rate": 9.534778281636882e-06, "loss": 0.3978, "step": 2291 }, { "epoch": 0.16, "grad_norm": 6.288718029524899, "learning_rate": 9.534288982605658e-06, "loss": 1.1605, "step": 2292 }, { "epoch": 0.16, "grad_norm": 4.511738970764019, "learning_rate": 9.533799438966704e-06, "loss": 0.2544, "step": 2293 }, { "epoch": 0.16, "grad_norm": 4.1111458537255325, "learning_rate": 9.533309650746433e-06, "loss": 0.2568, "step": 2294 }, { "epoch": 0.16, "grad_norm": 9.023945239512608, "learning_rate": 9.532819617971266e-06, "loss": 0.6984, "step": 2295 }, { "epoch": 0.16, "grad_norm": 3.454212685834147, "learning_rate": 9.532329340667638e-06, "loss": 0.3413, "step": 2296 }, { "epoch": 0.16, "grad_norm": 4.158799784245985, "learning_rate": 9.531838818861998e-06, "loss": 0.3035, "step": 2297 }, { "epoch": 0.16, "grad_norm": 3.9123758684160004, "learning_rate": 9.53134805258081e-06, "loss": 0.3866, "step": 2298 }, { "epoch": 0.16, "grad_norm": 4.1538854251060044, "learning_rate": 9.530857041850542e-06, "loss": 0.3222, "step": 2299 }, { "epoch": 0.16, "grad_norm": 22.11449124879467, "learning_rate": 9.530365786697691e-06, "loss": 1.1103, "step": 2300 }, { "epoch": 0.16, "grad_norm": 5.502111984026648, "learning_rate": 9.52987428714875e-06, "loss": 0.3223, "step": 2301 }, { "epoch": 0.17, "grad_norm": 5.431250789070153, "learning_rate": 9.529382543230242e-06, "loss": 0.3028, "step": 2302 }, { "epoch": 0.17, "grad_norm": 5.732451306850395, "learning_rate": 9.528890554968688e-06, "loss": 1.0477, "step": 2303 }, { "epoch": 0.17, "grad_norm": 4.686810720552399, "learning_rate": 9.528398322390631e-06, "loss": 0.2598, "step": 2304 }, { "epoch": 0.17, "grad_norm": 4.4754590835846955, "learning_rate": 9.527905845522628e-06, "loss": 0.2517, "step": 2305 }, { "epoch": 0.17, "grad_norm": 12.203085681051137, "learning_rate": 9.52741312439124e-06, "loss": 1.8408, "step": 2306 }, { "epoch": 0.17, "grad_norm": 15.471493387419807, "learning_rate": 9.526920159023055e-06, "loss": 1.5786, "step": 2307 }, { "epoch": 0.17, "grad_norm": 4.015468189114433, "learning_rate": 9.52642694944466e-06, "loss": 0.2819, "step": 2308 }, { "epoch": 0.17, "grad_norm": 4.507394145501732, "learning_rate": 9.525933495682666e-06, "loss": 0.312, "step": 2309 }, { "epoch": 0.17, "grad_norm": 3.8006148221729315, "learning_rate": 9.52543979776369e-06, "loss": 0.2589, "step": 2310 }, { "epoch": 0.17, "grad_norm": 21.082028179341197, "learning_rate": 9.524945855714368e-06, "loss": 1.4771, "step": 2311 }, { "epoch": 0.17, "grad_norm": 19.855227957985498, "learning_rate": 9.524451669561345e-06, "loss": 1.6426, "step": 2312 }, { "epoch": 0.17, "grad_norm": 4.426261546186628, "learning_rate": 9.523957239331278e-06, "loss": 0.3039, "step": 2313 }, { "epoch": 0.17, "grad_norm": 24.185459769290258, "learning_rate": 9.523462565050845e-06, "loss": 1.3997, "step": 2314 }, { "epoch": 0.17, "grad_norm": 14.277503665583131, "learning_rate": 9.522967646746728e-06, "loss": 0.9796, "step": 2315 }, { "epoch": 0.17, "grad_norm": 29.954724200989002, "learning_rate": 9.522472484445626e-06, "loss": 1.7014, "step": 2316 }, { "epoch": 0.17, "grad_norm": 20.963925277807252, "learning_rate": 9.521977078174253e-06, "loss": 1.3378, "step": 2317 }, { "epoch": 0.17, "grad_norm": 6.7006777080004465, "learning_rate": 9.521481427959332e-06, "loss": 0.3143, "step": 2318 }, { "epoch": 0.17, "grad_norm": 18.833617288455095, "learning_rate": 9.520985533827605e-06, "loss": 1.466, "step": 2319 }, { "epoch": 0.17, "grad_norm": 5.334395622982927, "learning_rate": 9.52048939580582e-06, "loss": 0.4401, "step": 2320 }, { "epoch": 0.17, "grad_norm": 5.549917749182734, "learning_rate": 9.519993013920742e-06, "loss": 0.971, "step": 2321 }, { "epoch": 0.17, "grad_norm": 4.719629563502181, "learning_rate": 9.51949638819915e-06, "loss": 0.2219, "step": 2322 }, { "epoch": 0.17, "grad_norm": 33.38412892497667, "learning_rate": 9.518999518667837e-06, "loss": 1.2048, "step": 2323 }, { "epoch": 0.17, "grad_norm": 4.625790800698172, "learning_rate": 9.518502405353605e-06, "loss": 0.2398, "step": 2324 }, { "epoch": 0.17, "grad_norm": 4.521136059467783, "learning_rate": 9.51800504828327e-06, "loss": 0.3015, "step": 2325 }, { "epoch": 0.17, "grad_norm": 46.814485754016474, "learning_rate": 9.517507447483665e-06, "loss": 2.0557, "step": 2326 }, { "epoch": 0.17, "grad_norm": 4.651792548729165, "learning_rate": 9.517009602981632e-06, "loss": 0.2667, "step": 2327 }, { "epoch": 0.17, "grad_norm": 5.043820923855837, "learning_rate": 9.516511514804031e-06, "loss": 0.3739, "step": 2328 }, { "epoch": 0.17, "grad_norm": 16.98798210033119, "learning_rate": 9.516013182977728e-06, "loss": 1.0924, "step": 2329 }, { "epoch": 0.17, "grad_norm": 4.6793625337664, "learning_rate": 9.51551460752961e-06, "loss": 0.3626, "step": 2330 }, { "epoch": 0.17, "grad_norm": 22.597939976105582, "learning_rate": 9.515015788486568e-06, "loss": 1.4341, "step": 2331 }, { "epoch": 0.17, "grad_norm": 4.255490050145015, "learning_rate": 9.514516725875517e-06, "loss": 0.256, "step": 2332 }, { "epoch": 0.17, "grad_norm": 4.651853562355636, "learning_rate": 9.514017419723376e-06, "loss": 0.326, "step": 2333 }, { "epoch": 0.17, "grad_norm": 6.208524462447209, "learning_rate": 9.513517870057081e-06, "loss": 0.3509, "step": 2334 }, { "epoch": 0.17, "grad_norm": 10.527684303806346, "learning_rate": 9.513018076903581e-06, "loss": 0.7808, "step": 2335 }, { "epoch": 0.17, "grad_norm": 5.182713960357863, "learning_rate": 9.51251804028984e-06, "loss": 0.8607, "step": 2336 }, { "epoch": 0.17, "grad_norm": 3.9264731207537626, "learning_rate": 9.51201776024283e-06, "loss": 0.2359, "step": 2337 }, { "epoch": 0.17, "grad_norm": 3.8641071475436806, "learning_rate": 9.511517236789543e-06, "loss": 0.342, "step": 2338 }, { "epoch": 0.17, "grad_norm": 5.015467337850116, "learning_rate": 9.511016469956977e-06, "loss": 0.2852, "step": 2339 }, { "epoch": 0.17, "grad_norm": 4.021143080732048, "learning_rate": 9.510515459772148e-06, "loss": 0.2599, "step": 2340 }, { "epoch": 0.17, "grad_norm": 3.6551019052367586, "learning_rate": 9.510014206262083e-06, "loss": 0.2006, "step": 2341 }, { "epoch": 0.17, "grad_norm": 10.375903741125079, "learning_rate": 9.509512709453821e-06, "loss": 1.2694, "step": 2342 }, { "epoch": 0.17, "grad_norm": 4.100873072636362, "learning_rate": 9.509010969374421e-06, "loss": 0.3506, "step": 2343 }, { "epoch": 0.17, "grad_norm": 5.734159696724969, "learning_rate": 9.508508986050945e-06, "loss": 0.9728, "step": 2344 }, { "epoch": 0.17, "grad_norm": 21.0687825138783, "learning_rate": 9.508006759510474e-06, "loss": 1.0785, "step": 2345 }, { "epoch": 0.17, "grad_norm": 5.827227768518582, "learning_rate": 9.507504289780103e-06, "loss": 1.1454, "step": 2346 }, { "epoch": 0.17, "grad_norm": 4.587419042657272, "learning_rate": 9.507001576886939e-06, "loss": 0.3321, "step": 2347 }, { "epoch": 0.17, "grad_norm": 4.75944806801238, "learning_rate": 9.506498620858097e-06, "loss": 0.9205, "step": 2348 }, { "epoch": 0.17, "grad_norm": 3.9258188050032783, "learning_rate": 9.505995421720713e-06, "loss": 0.3413, "step": 2349 }, { "epoch": 0.17, "grad_norm": 14.960451850399574, "learning_rate": 9.505491979501934e-06, "loss": 0.9979, "step": 2350 }, { "epoch": 0.17, "grad_norm": 3.828591269031335, "learning_rate": 9.504988294228914e-06, "loss": 0.2781, "step": 2351 }, { "epoch": 0.17, "grad_norm": 4.611877134672539, "learning_rate": 9.50448436592883e-06, "loss": 0.3286, "step": 2352 }, { "epoch": 0.17, "grad_norm": 5.360946030127727, "learning_rate": 9.503980194628865e-06, "loss": 1.082, "step": 2353 }, { "epoch": 0.17, "grad_norm": 16.69055895257568, "learning_rate": 9.503475780356216e-06, "loss": 1.4202, "step": 2354 }, { "epoch": 0.17, "grad_norm": 4.22179697845835, "learning_rate": 9.502971123138097e-06, "loss": 0.265, "step": 2355 }, { "epoch": 0.17, "grad_norm": 4.778589001001336, "learning_rate": 9.502466223001728e-06, "loss": 0.3369, "step": 2356 }, { "epoch": 0.17, "grad_norm": 13.641616065212194, "learning_rate": 9.50196107997435e-06, "loss": 0.849, "step": 2357 }, { "epoch": 0.17, "grad_norm": 5.426788592186985, "learning_rate": 9.501455694083213e-06, "loss": 1.0031, "step": 2358 }, { "epoch": 0.17, "grad_norm": 3.5904677424160933, "learning_rate": 9.500950065355579e-06, "loss": 0.2221, "step": 2359 }, { "epoch": 0.17, "grad_norm": 6.5642972438794756, "learning_rate": 9.500444193818728e-06, "loss": 0.9621, "step": 2360 }, { "epoch": 0.17, "grad_norm": 9.398012390659936, "learning_rate": 9.499938079499947e-06, "loss": 0.973, "step": 2361 }, { "epoch": 0.17, "grad_norm": 23.998473012768194, "learning_rate": 9.49943172242654e-06, "loss": 1.4404, "step": 2362 }, { "epoch": 0.17, "grad_norm": 8.8929621724448, "learning_rate": 9.498925122625822e-06, "loss": 0.9965, "step": 2363 }, { "epoch": 0.17, "grad_norm": 6.768102924637826, "learning_rate": 9.498418280125124e-06, "loss": 1.0784, "step": 2364 }, { "epoch": 0.17, "grad_norm": 18.87742729178073, "learning_rate": 9.497911194951786e-06, "loss": 1.7284, "step": 2365 }, { "epoch": 0.17, "grad_norm": 10.1982042437282, "learning_rate": 9.497403867133166e-06, "loss": 0.7715, "step": 2366 }, { "epoch": 0.17, "grad_norm": 3.0734108005834644, "learning_rate": 9.49689629669663e-06, "loss": 0.2358, "step": 2367 }, { "epoch": 0.17, "grad_norm": 5.842617253737317, "learning_rate": 9.49638848366956e-06, "loss": 1.0094, "step": 2368 }, { "epoch": 0.17, "grad_norm": 3.8500975200674294, "learning_rate": 9.495880428079352e-06, "loss": 0.2748, "step": 2369 }, { "epoch": 0.17, "grad_norm": 2.985411273933532, "learning_rate": 9.495372129953412e-06, "loss": 0.1765, "step": 2370 }, { "epoch": 0.17, "grad_norm": 4.238321928610432, "learning_rate": 9.494863589319161e-06, "loss": 0.2482, "step": 2371 }, { "epoch": 0.17, "grad_norm": 3.9296320838499144, "learning_rate": 9.494354806204036e-06, "loss": 0.3162, "step": 2372 }, { "epoch": 0.17, "grad_norm": 52.21292570194964, "learning_rate": 9.493845780635478e-06, "loss": 1.2793, "step": 2373 }, { "epoch": 0.17, "grad_norm": 4.41479127043637, "learning_rate": 9.493336512640955e-06, "loss": 0.3135, "step": 2374 }, { "epoch": 0.17, "grad_norm": 15.253276602734081, "learning_rate": 9.492827002247931e-06, "loss": 1.0645, "step": 2375 }, { "epoch": 0.17, "grad_norm": 3.8584526261719, "learning_rate": 9.492317249483897e-06, "loss": 0.2372, "step": 2376 }, { "epoch": 0.17, "grad_norm": 4.939060804980733, "learning_rate": 9.491807254376353e-06, "loss": 0.9378, "step": 2377 }, { "epoch": 0.17, "grad_norm": 3.7908362608910853, "learning_rate": 9.491297016952809e-06, "loss": 0.2794, "step": 2378 }, { "epoch": 0.17, "grad_norm": 3.9469526003651136, "learning_rate": 9.490786537240792e-06, "loss": 0.2455, "step": 2379 }, { "epoch": 0.17, "grad_norm": 19.510118753247525, "learning_rate": 9.490275815267839e-06, "loss": 1.3065, "step": 2380 }, { "epoch": 0.17, "grad_norm": 10.028862012112718, "learning_rate": 9.489764851061504e-06, "loss": 1.0298, "step": 2381 }, { "epoch": 0.17, "grad_norm": 19.084617302042265, "learning_rate": 9.489253644649348e-06, "loss": 0.8255, "step": 2382 }, { "epoch": 0.17, "grad_norm": 12.361152784671656, "learning_rate": 9.488742196058953e-06, "loss": 1.1129, "step": 2383 }, { "epoch": 0.17, "grad_norm": 3.776202941794618, "learning_rate": 9.488230505317905e-06, "loss": 0.3191, "step": 2384 }, { "epoch": 0.17, "grad_norm": 20.938943174030065, "learning_rate": 9.487718572453812e-06, "loss": 1.3387, "step": 2385 }, { "epoch": 0.17, "grad_norm": 14.301372084091383, "learning_rate": 9.48720639749429e-06, "loss": 1.6382, "step": 2386 }, { "epoch": 0.17, "grad_norm": 3.9161146614959255, "learning_rate": 9.486693980466963e-06, "loss": 0.2802, "step": 2387 }, { "epoch": 0.17, "grad_norm": 4.973852228293753, "learning_rate": 9.486181321399483e-06, "loss": 0.3344, "step": 2388 }, { "epoch": 0.17, "grad_norm": 6.516914959609167, "learning_rate": 9.4856684203195e-06, "loss": 1.1669, "step": 2389 }, { "epoch": 0.17, "grad_norm": 16.360728600632232, "learning_rate": 9.485155277254686e-06, "loss": 1.2332, "step": 2390 }, { "epoch": 0.17, "grad_norm": 25.838681405770796, "learning_rate": 9.48464189223272e-06, "loss": 1.7098, "step": 2391 }, { "epoch": 0.17, "grad_norm": 5.188645672197203, "learning_rate": 9.484128265281303e-06, "loss": 0.3517, "step": 2392 }, { "epoch": 0.17, "grad_norm": 5.440795467774527, "learning_rate": 9.483614396428136e-06, "loss": 1.0473, "step": 2393 }, { "epoch": 0.17, "grad_norm": 8.761714304897641, "learning_rate": 9.483100285700946e-06, "loss": 0.6505, "step": 2394 }, { "epoch": 0.17, "grad_norm": 5.533815393058471, "learning_rate": 9.482585933127463e-06, "loss": 1.1868, "step": 2395 }, { "epoch": 0.17, "grad_norm": 3.570885995791422, "learning_rate": 9.482071338735437e-06, "loss": 0.2879, "step": 2396 }, { "epoch": 0.17, "grad_norm": 5.943953417945761, "learning_rate": 9.481556502552629e-06, "loss": 1.0855, "step": 2397 }, { "epoch": 0.17, "grad_norm": 3.885060709658576, "learning_rate": 9.48104142460681e-06, "loss": 0.2919, "step": 2398 }, { "epoch": 0.17, "grad_norm": 3.308054690455102, "learning_rate": 9.480526104925769e-06, "loss": 0.3379, "step": 2399 }, { "epoch": 0.17, "grad_norm": 3.7587338676349056, "learning_rate": 9.480010543537303e-06, "loss": 0.2715, "step": 2400 }, { "epoch": 0.17, "grad_norm": 13.144715709450681, "learning_rate": 9.479494740469226e-06, "loss": 0.8036, "step": 2401 }, { "epoch": 0.17, "grad_norm": 3.7255331548556403, "learning_rate": 9.478978695749366e-06, "loss": 0.3312, "step": 2402 }, { "epoch": 0.17, "grad_norm": 6.8150560406267555, "learning_rate": 9.478462409405555e-06, "loss": 0.7653, "step": 2403 }, { "epoch": 0.17, "grad_norm": 5.964154362625553, "learning_rate": 9.477945881465654e-06, "loss": 1.097, "step": 2404 }, { "epoch": 0.17, "grad_norm": 3.6083105004112155, "learning_rate": 9.47742911195752e-06, "loss": 0.3227, "step": 2405 }, { "epoch": 0.17, "grad_norm": 3.666374692115333, "learning_rate": 9.476912100909032e-06, "loss": 0.3458, "step": 2406 }, { "epoch": 0.17, "grad_norm": 3.5604622825942065, "learning_rate": 9.476394848348083e-06, "loss": 0.2522, "step": 2407 }, { "epoch": 0.17, "grad_norm": 18.338330096199808, "learning_rate": 9.475877354302578e-06, "loss": 1.7365, "step": 2408 }, { "epoch": 0.17, "grad_norm": 4.294908837325487, "learning_rate": 9.47535961880043e-06, "loss": 0.348, "step": 2409 }, { "epoch": 0.17, "grad_norm": 4.677718793125498, "learning_rate": 9.47484164186957e-06, "loss": 0.3336, "step": 2410 }, { "epoch": 0.17, "grad_norm": 4.2806297458567775, "learning_rate": 9.474323423537944e-06, "loss": 0.3059, "step": 2411 }, { "epoch": 0.17, "grad_norm": 4.704372757605186, "learning_rate": 9.473804963833503e-06, "loss": 0.3778, "step": 2412 }, { "epoch": 0.17, "grad_norm": 11.966282155950747, "learning_rate": 9.473286262784221e-06, "loss": 1.145, "step": 2413 }, { "epoch": 0.17, "grad_norm": 4.011908935056265, "learning_rate": 9.472767320418074e-06, "loss": 0.3184, "step": 2414 }, { "epoch": 0.17, "grad_norm": 3.4080600042725475, "learning_rate": 9.472248136763063e-06, "loss": 0.2234, "step": 2415 }, { "epoch": 0.17, "grad_norm": 8.113409780705096, "learning_rate": 9.471728711847195e-06, "loss": 0.9624, "step": 2416 }, { "epoch": 0.17, "grad_norm": 19.84133753556202, "learning_rate": 9.471209045698486e-06, "loss": 1.1741, "step": 2417 }, { "epoch": 0.17, "grad_norm": 5.920496751386947, "learning_rate": 9.470689138344975e-06, "loss": 1.0077, "step": 2418 }, { "epoch": 0.17, "grad_norm": 16.605927125275088, "learning_rate": 9.470168989814708e-06, "loss": 1.5417, "step": 2419 }, { "epoch": 0.17, "grad_norm": 5.708789103257953, "learning_rate": 9.469648600135743e-06, "loss": 0.4101, "step": 2420 }, { "epoch": 0.17, "grad_norm": 4.9821019126120625, "learning_rate": 9.469127969336156e-06, "loss": 0.4474, "step": 2421 }, { "epoch": 0.17, "grad_norm": 8.746153338900756, "learning_rate": 9.468607097444031e-06, "loss": 0.8478, "step": 2422 }, { "epoch": 0.17, "grad_norm": 4.093819023977977, "learning_rate": 9.468085984487467e-06, "loss": 0.2516, "step": 2423 }, { "epoch": 0.17, "grad_norm": 23.281379331354508, "learning_rate": 9.467564630494577e-06, "loss": 0.7874, "step": 2424 }, { "epoch": 0.17, "grad_norm": 11.50730064730521, "learning_rate": 9.467043035493486e-06, "loss": 0.8589, "step": 2425 }, { "epoch": 0.17, "grad_norm": 20.128191541478262, "learning_rate": 9.466521199512333e-06, "loss": 1.5395, "step": 2426 }, { "epoch": 0.17, "grad_norm": 19.510650532527343, "learning_rate": 9.465999122579267e-06, "loss": 1.3684, "step": 2427 }, { "epoch": 0.17, "grad_norm": 26.12169526509801, "learning_rate": 9.465476804722453e-06, "loss": 2.062, "step": 2428 }, { "epoch": 0.17, "grad_norm": 4.3169662737208006, "learning_rate": 9.464954245970068e-06, "loss": 0.2588, "step": 2429 }, { "epoch": 0.17, "grad_norm": 16.103656916036, "learning_rate": 9.464431446350304e-06, "loss": 0.9089, "step": 2430 }, { "epoch": 0.17, "grad_norm": 15.008427385698274, "learning_rate": 9.46390840589136e-06, "loss": 1.0465, "step": 2431 }, { "epoch": 0.17, "grad_norm": 4.593140558557531, "learning_rate": 9.463385124621454e-06, "loss": 0.2242, "step": 2432 }, { "epoch": 0.17, "grad_norm": 4.309317327828996, "learning_rate": 9.462861602568817e-06, "loss": 0.256, "step": 2433 }, { "epoch": 0.17, "grad_norm": 17.187271309133937, "learning_rate": 9.46233783976169e-06, "loss": 0.9724, "step": 2434 }, { "epoch": 0.17, "grad_norm": 4.335647777971435, "learning_rate": 9.461813836228326e-06, "loss": 0.313, "step": 2435 }, { "epoch": 0.17, "grad_norm": 6.382001789098776, "learning_rate": 9.461289591996994e-06, "loss": 1.1307, "step": 2436 }, { "epoch": 0.17, "grad_norm": 15.152740557944613, "learning_rate": 9.460765107095976e-06, "loss": 1.291, "step": 2437 }, { "epoch": 0.17, "grad_norm": 4.9684303502620155, "learning_rate": 9.460240381553564e-06, "loss": 0.3089, "step": 2438 }, { "epoch": 0.17, "grad_norm": 6.552946757559765, "learning_rate": 9.459715415398069e-06, "loss": 0.5805, "step": 2439 }, { "epoch": 0.17, "grad_norm": 4.754656505117285, "learning_rate": 9.459190208657807e-06, "loss": 0.2534, "step": 2440 }, { "epoch": 0.18, "grad_norm": 3.4209822125963316, "learning_rate": 9.458664761361112e-06, "loss": 0.2273, "step": 2441 }, { "epoch": 0.18, "grad_norm": 6.926693305026101, "learning_rate": 9.458139073536331e-06, "loss": 1.1226, "step": 2442 }, { "epoch": 0.18, "grad_norm": 5.283672602738862, "learning_rate": 9.457613145211821e-06, "loss": 0.9495, "step": 2443 }, { "epoch": 0.18, "grad_norm": 9.663834959425555, "learning_rate": 9.457086976415956e-06, "loss": 0.7104, "step": 2444 }, { "epoch": 0.18, "grad_norm": 14.24846804092817, "learning_rate": 9.456560567177118e-06, "loss": 1.4363, "step": 2445 }, { "epoch": 0.18, "grad_norm": 3.82306888478567, "learning_rate": 9.456033917523708e-06, "loss": 0.3135, "step": 2446 }, { "epoch": 0.18, "grad_norm": 4.250309425863985, "learning_rate": 9.455507027484134e-06, "loss": 0.2457, "step": 2447 }, { "epoch": 0.18, "grad_norm": 3.873209993838987, "learning_rate": 9.45497989708682e-06, "loss": 0.2738, "step": 2448 }, { "epoch": 0.18, "grad_norm": 4.134143266170794, "learning_rate": 9.454452526360205e-06, "loss": 0.2909, "step": 2449 }, { "epoch": 0.18, "grad_norm": 15.78222872526678, "learning_rate": 9.453924915332737e-06, "loss": 1.0771, "step": 2450 }, { "epoch": 0.18, "grad_norm": 3.4606216313951372, "learning_rate": 9.453397064032879e-06, "loss": 0.1953, "step": 2451 }, { "epoch": 0.18, "grad_norm": 20.09573540427869, "learning_rate": 9.452868972489106e-06, "loss": 2.3815, "step": 2452 }, { "epoch": 0.18, "grad_norm": 82.79766529211383, "learning_rate": 9.452340640729908e-06, "loss": 1.419, "step": 2453 }, { "epoch": 0.18, "grad_norm": 3.4228640387214764, "learning_rate": 9.451812068783784e-06, "loss": 0.1914, "step": 2454 }, { "epoch": 0.18, "grad_norm": 8.67391679726409, "learning_rate": 9.45128325667925e-06, "loss": 0.3375, "step": 2455 }, { "epoch": 0.18, "grad_norm": 4.312000969032345, "learning_rate": 9.450754204444834e-06, "loss": 0.2709, "step": 2456 }, { "epoch": 0.18, "grad_norm": 6.177871601488237, "learning_rate": 9.450224912109076e-06, "loss": 1.2171, "step": 2457 }, { "epoch": 0.18, "grad_norm": 23.2857354663469, "learning_rate": 9.44969537970053e-06, "loss": 1.6907, "step": 2458 }, { "epoch": 0.18, "grad_norm": 4.129750442438945, "learning_rate": 9.44916560724776e-06, "loss": 0.1937, "step": 2459 }, { "epoch": 0.18, "grad_norm": 5.82596992726019, "learning_rate": 9.448635594779345e-06, "loss": 0.2928, "step": 2460 }, { "epoch": 0.18, "grad_norm": 4.900697170937419, "learning_rate": 9.44810534232388e-06, "loss": 0.3056, "step": 2461 }, { "epoch": 0.18, "grad_norm": 10.982473264145344, "learning_rate": 9.447574849909969e-06, "loss": 0.9476, "step": 2462 }, { "epoch": 0.18, "grad_norm": 4.560498234774412, "learning_rate": 9.447044117566228e-06, "loss": 0.2663, "step": 2463 }, { "epoch": 0.18, "grad_norm": 3.2349245379176175, "learning_rate": 9.446513145321293e-06, "loss": 0.2614, "step": 2464 }, { "epoch": 0.18, "grad_norm": 21.791890774137794, "learning_rate": 9.445981933203803e-06, "loss": 0.9462, "step": 2465 }, { "epoch": 0.18, "grad_norm": 4.208083570189662, "learning_rate": 9.445450481242417e-06, "loss": 0.2998, "step": 2466 }, { "epoch": 0.18, "grad_norm": 4.500792177318873, "learning_rate": 9.444918789465804e-06, "loss": 0.2913, "step": 2467 }, { "epoch": 0.18, "grad_norm": 26.965872626902534, "learning_rate": 9.444386857902647e-06, "loss": 1.5692, "step": 2468 }, { "epoch": 0.18, "grad_norm": 30.143837268944715, "learning_rate": 9.443854686581644e-06, "loss": 1.1185, "step": 2469 }, { "epoch": 0.18, "grad_norm": 6.318852649606908, "learning_rate": 9.443322275531498e-06, "loss": 1.1188, "step": 2470 }, { "epoch": 0.18, "grad_norm": 5.162129643021403, "learning_rate": 9.442789624780937e-06, "loss": 0.2832, "step": 2471 }, { "epoch": 0.18, "grad_norm": 19.724897285669662, "learning_rate": 9.442256734358692e-06, "loss": 1.719, "step": 2472 }, { "epoch": 0.18, "grad_norm": 4.835523379272493, "learning_rate": 9.44172360429351e-06, "loss": 0.2204, "step": 2473 }, { "epoch": 0.18, "grad_norm": 4.5337937326788955, "learning_rate": 9.441190234614155e-06, "loss": 0.2997, "step": 2474 }, { "epoch": 0.18, "grad_norm": 14.039195799577112, "learning_rate": 9.440656625349396e-06, "loss": 1.0563, "step": 2475 }, { "epoch": 0.18, "grad_norm": 3.9115682836068615, "learning_rate": 9.44012277652802e-06, "loss": 0.2347, "step": 2476 }, { "epoch": 0.18, "grad_norm": 4.301861866622937, "learning_rate": 9.439588688178829e-06, "loss": 0.3161, "step": 2477 }, { "epoch": 0.18, "grad_norm": 18.49277053049604, "learning_rate": 9.439054360330631e-06, "loss": 1.1937, "step": 2478 }, { "epoch": 0.18, "grad_norm": 14.854058329821525, "learning_rate": 9.438519793012254e-06, "loss": 0.8752, "step": 2479 }, { "epoch": 0.18, "grad_norm": 3.461986042630739, "learning_rate": 9.437984986252536e-06, "loss": 0.2365, "step": 2480 }, { "epoch": 0.18, "grad_norm": 26.102795569448833, "learning_rate": 9.437449940080322e-06, "loss": 1.323, "step": 2481 }, { "epoch": 0.18, "grad_norm": 68.59473558049311, "learning_rate": 9.436914654524486e-06, "loss": 1.4974, "step": 2482 }, { "epoch": 0.18, "grad_norm": 12.847648325100641, "learning_rate": 9.436379129613896e-06, "loss": 0.7327, "step": 2483 }, { "epoch": 0.18, "grad_norm": 3.275778042209461, "learning_rate": 9.435843365377446e-06, "loss": 0.2306, "step": 2484 }, { "epoch": 0.18, "grad_norm": 17.19933189919114, "learning_rate": 9.435307361844035e-06, "loss": 1.4885, "step": 2485 }, { "epoch": 0.18, "grad_norm": 10.406647621046533, "learning_rate": 9.434771119042584e-06, "loss": 0.7883, "step": 2486 }, { "epoch": 0.18, "grad_norm": 6.095939836263039, "learning_rate": 9.434234637002014e-06, "loss": 0.9717, "step": 2487 }, { "epoch": 0.18, "grad_norm": 3.825397451304484, "learning_rate": 9.433697915751271e-06, "loss": 0.3199, "step": 2488 }, { "epoch": 0.18, "grad_norm": 4.798807398919432, "learning_rate": 9.433160955319308e-06, "loss": 0.2988, "step": 2489 }, { "epoch": 0.18, "grad_norm": 4.219195016571574, "learning_rate": 9.432623755735092e-06, "loss": 0.3381, "step": 2490 }, { "epoch": 0.18, "grad_norm": 3.131220683896367, "learning_rate": 9.432086317027602e-06, "loss": 0.1904, "step": 2491 }, { "epoch": 0.18, "grad_norm": 3.638009783992325, "learning_rate": 9.431548639225834e-06, "loss": 0.2322, "step": 2492 }, { "epoch": 0.18, "grad_norm": 60.51389824579636, "learning_rate": 9.431010722358788e-06, "loss": 1.3523, "step": 2493 }, { "epoch": 0.18, "grad_norm": 4.810982593979537, "learning_rate": 9.430472566455488e-06, "loss": 0.8653, "step": 2494 }, { "epoch": 0.18, "grad_norm": 8.178608607977823, "learning_rate": 9.429934171544964e-06, "loss": 1.139, "step": 2495 }, { "epoch": 0.18, "grad_norm": 5.484265402000523, "learning_rate": 9.429395537656257e-06, "loss": 0.989, "step": 2496 }, { "epoch": 0.18, "grad_norm": 4.877925940608203, "learning_rate": 9.428856664818427e-06, "loss": 0.2364, "step": 2497 }, { "epoch": 0.18, "grad_norm": 14.53594040814951, "learning_rate": 9.428317553060546e-06, "loss": 1.2095, "step": 2498 }, { "epoch": 0.18, "grad_norm": 3.802534250069596, "learning_rate": 9.427778202411695e-06, "loss": 0.1852, "step": 2499 }, { "epoch": 0.18, "grad_norm": 7.001365942632924, "learning_rate": 9.427238612900968e-06, "loss": 1.0255, "step": 2500 }, { "epoch": 0.18, "grad_norm": 6.465113431784772, "learning_rate": 9.426698784557478e-06, "loss": 1.1331, "step": 2501 }, { "epoch": 0.18, "grad_norm": 3.708066359606912, "learning_rate": 9.426158717410345e-06, "loss": 0.2938, "step": 2502 }, { "epoch": 0.18, "grad_norm": 3.725603287021532, "learning_rate": 9.425618411488702e-06, "loss": 0.2602, "step": 2503 }, { "epoch": 0.18, "grad_norm": 5.207336934192528, "learning_rate": 9.425077866821696e-06, "loss": 0.8741, "step": 2504 }, { "epoch": 0.18, "grad_norm": 16.87616555726878, "learning_rate": 9.42453708343849e-06, "loss": 1.6335, "step": 2505 }, { "epoch": 0.18, "grad_norm": 4.907822441174371, "learning_rate": 9.423996061368258e-06, "loss": 0.322, "step": 2506 }, { "epoch": 0.18, "grad_norm": 3.978470383669298, "learning_rate": 9.423454800640182e-06, "loss": 0.1823, "step": 2507 }, { "epoch": 0.18, "grad_norm": 4.976826962319542, "learning_rate": 9.422913301283465e-06, "loss": 0.3501, "step": 2508 }, { "epoch": 0.18, "grad_norm": 23.677508169766703, "learning_rate": 9.422371563327316e-06, "loss": 2.0829, "step": 2509 }, { "epoch": 0.18, "grad_norm": 4.895141541941725, "learning_rate": 9.421829586800959e-06, "loss": 0.3147, "step": 2510 }, { "epoch": 0.18, "grad_norm": 5.354684182753213, "learning_rate": 9.421287371733636e-06, "loss": 0.3041, "step": 2511 }, { "epoch": 0.18, "grad_norm": 4.616811182507276, "learning_rate": 9.420744918154594e-06, "loss": 0.2331, "step": 2512 }, { "epoch": 0.18, "grad_norm": 7.622165522699272, "learning_rate": 9.420202226093096e-06, "loss": 0.4713, "step": 2513 }, { "epoch": 0.18, "grad_norm": 11.055723501836301, "learning_rate": 9.41965929557842e-06, "loss": 0.926, "step": 2514 }, { "epoch": 0.18, "grad_norm": 4.590350968785904, "learning_rate": 9.419116126639855e-06, "loss": 0.3281, "step": 2515 }, { "epoch": 0.18, "grad_norm": 13.481506764917746, "learning_rate": 9.418572719306703e-06, "loss": 1.0982, "step": 2516 }, { "epoch": 0.18, "grad_norm": 40.13744694577149, "learning_rate": 9.418029073608274e-06, "loss": 1.4553, "step": 2517 }, { "epoch": 0.18, "grad_norm": 13.336230980146489, "learning_rate": 9.417485189573903e-06, "loss": 1.1265, "step": 2518 }, { "epoch": 0.18, "grad_norm": 22.40801370356513, "learning_rate": 9.416941067232927e-06, "loss": 1.44, "step": 2519 }, { "epoch": 0.18, "grad_norm": 5.785629679365978, "learning_rate": 9.4163967066147e-06, "loss": 0.2826, "step": 2520 }, { "epoch": 0.18, "grad_norm": 17.625355160450123, "learning_rate": 9.415852107748587e-06, "loss": 1.0598, "step": 2521 }, { "epoch": 0.18, "grad_norm": 4.8882662479655545, "learning_rate": 9.415307270663966e-06, "loss": 0.3272, "step": 2522 }, { "epoch": 0.18, "grad_norm": 4.071108473001199, "learning_rate": 9.414762195390233e-06, "loss": 0.229, "step": 2523 }, { "epoch": 0.18, "grad_norm": 5.534621605368517, "learning_rate": 9.414216881956787e-06, "loss": 1.191, "step": 2524 }, { "epoch": 0.18, "grad_norm": 22.241191281377397, "learning_rate": 9.413671330393053e-06, "loss": 1.4251, "step": 2525 }, { "epoch": 0.18, "grad_norm": 20.527516913790365, "learning_rate": 9.413125540728456e-06, "loss": 1.4613, "step": 2526 }, { "epoch": 0.18, "grad_norm": 6.124806103533203, "learning_rate": 9.41257951299244e-06, "loss": 0.3504, "step": 2527 }, { "epoch": 0.18, "grad_norm": 18.40032113345306, "learning_rate": 9.412033247214461e-06, "loss": 1.4985, "step": 2528 }, { "epoch": 0.18, "grad_norm": 17.621172998257116, "learning_rate": 9.41148674342399e-06, "loss": 1.167, "step": 2529 }, { "epoch": 0.18, "grad_norm": 5.427771856420341, "learning_rate": 9.410940001650507e-06, "loss": 1.0725, "step": 2530 }, { "epoch": 0.18, "grad_norm": 4.120403830757458, "learning_rate": 9.410393021923508e-06, "loss": 0.365, "step": 2531 }, { "epoch": 0.18, "grad_norm": 3.874258656672931, "learning_rate": 9.409845804272499e-06, "loss": 0.2988, "step": 2532 }, { "epoch": 0.18, "grad_norm": 4.336707650097172, "learning_rate": 9.409298348727001e-06, "loss": 0.2544, "step": 2533 }, { "epoch": 0.18, "grad_norm": 6.271546843323089, "learning_rate": 9.408750655316547e-06, "loss": 0.3218, "step": 2534 }, { "epoch": 0.18, "grad_norm": 3.559104971850839, "learning_rate": 9.408202724070683e-06, "loss": 0.2413, "step": 2535 }, { "epoch": 0.18, "grad_norm": 14.469180440379183, "learning_rate": 9.40765455501897e-06, "loss": 1.1387, "step": 2536 }, { "epoch": 0.18, "grad_norm": 5.60904693920918, "learning_rate": 9.407106148190977e-06, "loss": 1.0144, "step": 2537 }, { "epoch": 0.18, "grad_norm": 9.09464369554944, "learning_rate": 9.406557503616286e-06, "loss": 0.7815, "step": 2538 }, { "epoch": 0.18, "grad_norm": 17.353246336202474, "learning_rate": 9.406008621324501e-06, "loss": 1.3445, "step": 2539 }, { "epoch": 0.18, "grad_norm": 5.115648362609031, "learning_rate": 9.405459501345228e-06, "loss": 1.1608, "step": 2540 }, { "epoch": 0.18, "grad_norm": 18.401283920982117, "learning_rate": 9.404910143708089e-06, "loss": 1.4691, "step": 2541 }, { "epoch": 0.18, "grad_norm": 29.872509562727846, "learning_rate": 9.404360548442722e-06, "loss": 0.8454, "step": 2542 }, { "epoch": 0.18, "grad_norm": 14.737507565887329, "learning_rate": 9.403810715578776e-06, "loss": 1.5845, "step": 2543 }, { "epoch": 0.18, "grad_norm": 12.28847242757309, "learning_rate": 9.403260645145908e-06, "loss": 1.1224, "step": 2544 }, { "epoch": 0.18, "grad_norm": 4.998488527311884, "learning_rate": 9.402710337173799e-06, "loss": 1.0132, "step": 2545 }, { "epoch": 0.18, "grad_norm": 6.2120205468105105, "learning_rate": 9.402159791692132e-06, "loss": 1.1975, "step": 2546 }, { "epoch": 0.18, "grad_norm": 19.160572113778755, "learning_rate": 9.401609008730605e-06, "loss": 1.154, "step": 2547 }, { "epoch": 0.18, "grad_norm": 4.55503885661299, "learning_rate": 9.401057988318933e-06, "loss": 1.0154, "step": 2548 }, { "epoch": 0.18, "grad_norm": 3.528744900809983, "learning_rate": 9.400506730486845e-06, "loss": 0.2484, "step": 2549 }, { "epoch": 0.18, "grad_norm": 28.264039695418152, "learning_rate": 9.399955235264072e-06, "loss": 1.5256, "step": 2550 }, { "epoch": 0.18, "grad_norm": 17.171968036289584, "learning_rate": 9.399403502680371e-06, "loss": 1.908, "step": 2551 }, { "epoch": 0.18, "grad_norm": 5.60563491310798, "learning_rate": 9.398851532765502e-06, "loss": 0.9859, "step": 2552 }, { "epoch": 0.18, "grad_norm": 8.541143962454145, "learning_rate": 9.398299325549244e-06, "loss": 0.9737, "step": 2553 }, { "epoch": 0.18, "grad_norm": 17.605693150382496, "learning_rate": 9.397746881061387e-06, "loss": 1.569, "step": 2554 }, { "epoch": 0.18, "grad_norm": 6.190542649819119, "learning_rate": 9.397194199331732e-06, "loss": 1.1312, "step": 2555 }, { "epoch": 0.18, "grad_norm": 47.42675355770323, "learning_rate": 9.396641280390092e-06, "loss": 0.9553, "step": 2556 }, { "epoch": 0.18, "grad_norm": 4.58208723674808, "learning_rate": 9.396088124266298e-06, "loss": 0.3481, "step": 2557 }, { "epoch": 0.18, "grad_norm": 26.462148799604282, "learning_rate": 9.395534730990192e-06, "loss": 1.4472, "step": 2558 }, { "epoch": 0.18, "grad_norm": 4.717409626196407, "learning_rate": 9.394981100591622e-06, "loss": 0.2652, "step": 2559 }, { "epoch": 0.18, "grad_norm": 5.199969643633739, "learning_rate": 9.39442723310046e-06, "loss": 0.3851, "step": 2560 }, { "epoch": 0.18, "grad_norm": 4.160432393225838, "learning_rate": 9.393873128546582e-06, "loss": 0.2232, "step": 2561 }, { "epoch": 0.18, "grad_norm": 4.033633893594269, "learning_rate": 9.39331878695988e-06, "loss": 0.1797, "step": 2562 }, { "epoch": 0.18, "grad_norm": 5.024905847862226, "learning_rate": 9.39276420837026e-06, "loss": 0.3717, "step": 2563 }, { "epoch": 0.18, "grad_norm": 18.588860541387998, "learning_rate": 9.392209392807637e-06, "loss": 1.1566, "step": 2564 }, { "epoch": 0.18, "grad_norm": 5.782743990556194, "learning_rate": 9.391654340301944e-06, "loss": 0.8599, "step": 2565 }, { "epoch": 0.18, "grad_norm": 9.43584493075981, "learning_rate": 9.391099050883122e-06, "loss": 0.7262, "step": 2566 }, { "epoch": 0.18, "grad_norm": 9.812867225550027, "learning_rate": 9.390543524581126e-06, "loss": 0.9147, "step": 2567 }, { "epoch": 0.18, "grad_norm": 3.028890784665136, "learning_rate": 9.389987761425929e-06, "loss": 0.166, "step": 2568 }, { "epoch": 0.18, "grad_norm": 5.510084171406703, "learning_rate": 9.389431761447507e-06, "loss": 1.1303, "step": 2569 }, { "epoch": 0.18, "grad_norm": 10.9615694867442, "learning_rate": 9.388875524675857e-06, "loss": 0.9115, "step": 2570 }, { "epoch": 0.18, "grad_norm": 18.569344439819638, "learning_rate": 9.388319051140986e-06, "loss": 1.0371, "step": 2571 }, { "epoch": 0.18, "grad_norm": 5.318375009664656, "learning_rate": 9.387762340872912e-06, "loss": 0.946, "step": 2572 }, { "epoch": 0.18, "grad_norm": 3.708301082794265, "learning_rate": 9.387205393901668e-06, "loss": 0.2143, "step": 2573 }, { "epoch": 0.18, "grad_norm": 4.131546783545086, "learning_rate": 9.3866482102573e-06, "loss": 0.2185, "step": 2574 }, { "epoch": 0.18, "grad_norm": 12.466063800941473, "learning_rate": 9.386090789969866e-06, "loss": 1.027, "step": 2575 }, { "epoch": 0.18, "grad_norm": 19.064434804370382, "learning_rate": 9.385533133069436e-06, "loss": 0.8564, "step": 2576 }, { "epoch": 0.18, "grad_norm": 14.605212731250575, "learning_rate": 9.384975239586092e-06, "loss": 0.9259, "step": 2577 }, { "epoch": 0.18, "grad_norm": 5.068789790301012, "learning_rate": 9.384417109549933e-06, "loss": 1.0429, "step": 2578 }, { "epoch": 0.18, "grad_norm": 9.509595068324183, "learning_rate": 9.383858742991067e-06, "loss": 0.7004, "step": 2579 }, { "epoch": 0.19, "grad_norm": 5.605733059664155, "learning_rate": 9.383300139939615e-06, "loss": 0.8523, "step": 2580 }, { "epoch": 0.19, "grad_norm": 5.35448315295613, "learning_rate": 9.382741300425713e-06, "loss": 1.1083, "step": 2581 }, { "epoch": 0.19, "grad_norm": 5.0516128135205625, "learning_rate": 9.382182224479508e-06, "loss": 1.2206, "step": 2582 }, { "epoch": 0.19, "grad_norm": 3.939090134899647, "learning_rate": 9.381622912131158e-06, "loss": 0.275, "step": 2583 }, { "epoch": 0.19, "grad_norm": 5.915507864958991, "learning_rate": 9.381063363410837e-06, "loss": 0.9363, "step": 2584 }, { "epoch": 0.19, "grad_norm": 3.184182210575551, "learning_rate": 9.380503578348732e-06, "loss": 0.2269, "step": 2585 }, { "epoch": 0.19, "grad_norm": 3.6213224681114453, "learning_rate": 9.37994355697504e-06, "loss": 0.2749, "step": 2586 }, { "epoch": 0.19, "grad_norm": 8.922031370872531, "learning_rate": 9.37938329931997e-06, "loss": 0.87, "step": 2587 }, { "epoch": 0.19, "grad_norm": 10.89675144041056, "learning_rate": 9.37882280541375e-06, "loss": 0.8293, "step": 2588 }, { "epoch": 0.19, "grad_norm": 14.951040439314593, "learning_rate": 9.378262075286615e-06, "loss": 0.8318, "step": 2589 }, { "epoch": 0.19, "grad_norm": 3.974761912927434, "learning_rate": 9.377701108968811e-06, "loss": 0.3079, "step": 2590 }, { "epoch": 0.19, "grad_norm": 3.7512723631700475, "learning_rate": 9.377139906490605e-06, "loss": 0.2025, "step": 2591 }, { "epoch": 0.19, "grad_norm": 5.51734426579955, "learning_rate": 9.37657846788227e-06, "loss": 1.0065, "step": 2592 }, { "epoch": 0.19, "grad_norm": 4.280219820230315, "learning_rate": 9.37601679317409e-06, "loss": 0.3053, "step": 2593 }, { "epoch": 0.19, "grad_norm": 31.72314485268923, "learning_rate": 9.37545488239637e-06, "loss": 1.7108, "step": 2594 }, { "epoch": 0.19, "grad_norm": 25.598461937830507, "learning_rate": 9.374892735579422e-06, "loss": 1.6702, "step": 2595 }, { "epoch": 0.19, "grad_norm": 4.544192228298524, "learning_rate": 9.374330352753569e-06, "loss": 0.2749, "step": 2596 }, { "epoch": 0.19, "grad_norm": 4.453599386408801, "learning_rate": 9.373767733949151e-06, "loss": 0.3466, "step": 2597 }, { "epoch": 0.19, "grad_norm": 8.224342184669595, "learning_rate": 9.373204879196521e-06, "loss": 0.9273, "step": 2598 }, { "epoch": 0.19, "grad_norm": 4.117920168622737, "learning_rate": 9.37264178852604e-06, "loss": 0.1932, "step": 2599 }, { "epoch": 0.19, "grad_norm": 5.513663544232114, "learning_rate": 9.372078461968086e-06, "loss": 1.0964, "step": 2600 }, { "epoch": 0.19, "grad_norm": 45.77233671029602, "learning_rate": 9.371514899553047e-06, "loss": 1.1885, "step": 2601 }, { "epoch": 0.19, "grad_norm": 3.8279321733939677, "learning_rate": 9.37095110131133e-06, "loss": 0.2056, "step": 2602 }, { "epoch": 0.19, "grad_norm": 6.758699766457204, "learning_rate": 9.370387067273342e-06, "loss": 1.2042, "step": 2603 }, { "epoch": 0.19, "grad_norm": 4.953614151881974, "learning_rate": 9.369822797469517e-06, "loss": 0.2518, "step": 2604 }, { "epoch": 0.19, "grad_norm": 4.9829315701110595, "learning_rate": 9.369258291930292e-06, "loss": 0.9322, "step": 2605 }, { "epoch": 0.19, "grad_norm": 4.150651132765117, "learning_rate": 9.36869355068612e-06, "loss": 0.2012, "step": 2606 }, { "epoch": 0.19, "grad_norm": 4.191541892538787, "learning_rate": 9.368128573767468e-06, "loss": 0.3092, "step": 2607 }, { "epoch": 0.19, "grad_norm": 3.7330198052412262, "learning_rate": 9.367563361204814e-06, "loss": 0.2186, "step": 2608 }, { "epoch": 0.19, "grad_norm": 4.043583369781259, "learning_rate": 9.366997913028649e-06, "loss": 0.3288, "step": 2609 }, { "epoch": 0.19, "grad_norm": 28.87555895325285, "learning_rate": 9.366432229269476e-06, "loss": 1.3613, "step": 2610 }, { "epoch": 0.19, "grad_norm": 5.888672362655057, "learning_rate": 9.365866309957812e-06, "loss": 1.1574, "step": 2611 }, { "epoch": 0.19, "grad_norm": 5.235648790137745, "learning_rate": 9.365300155124186e-06, "loss": 1.0702, "step": 2612 }, { "epoch": 0.19, "grad_norm": 5.296395373066473, "learning_rate": 9.36473376479914e-06, "loss": 0.4297, "step": 2613 }, { "epoch": 0.19, "grad_norm": 4.716600542520914, "learning_rate": 9.36416713901323e-06, "loss": 0.2168, "step": 2614 }, { "epoch": 0.19, "grad_norm": 48.901292617285414, "learning_rate": 9.363600277797023e-06, "loss": 1.425, "step": 2615 }, { "epoch": 0.19, "grad_norm": 15.290509729208582, "learning_rate": 9.363033181181096e-06, "loss": 1.2533, "step": 2616 }, { "epoch": 0.19, "grad_norm": 5.612506203188058, "learning_rate": 9.362465849196044e-06, "loss": 1.0516, "step": 2617 }, { "epoch": 0.19, "grad_norm": 4.310201201871275, "learning_rate": 9.361898281872474e-06, "loss": 0.2112, "step": 2618 }, { "epoch": 0.19, "grad_norm": 26.30210890852297, "learning_rate": 9.361330479241e-06, "loss": 1.6096, "step": 2619 }, { "epoch": 0.19, "grad_norm": 33.847807783558984, "learning_rate": 9.360762441332256e-06, "loss": 2.1028, "step": 2620 }, { "epoch": 0.19, "grad_norm": 5.218576043653057, "learning_rate": 9.360194168176887e-06, "loss": 0.2466, "step": 2621 }, { "epoch": 0.19, "grad_norm": 4.550440430131345, "learning_rate": 9.359625659805545e-06, "loss": 0.1949, "step": 2622 }, { "epoch": 0.19, "grad_norm": 15.359353899449085, "learning_rate": 9.359056916248898e-06, "loss": 1.0746, "step": 2623 }, { "epoch": 0.19, "grad_norm": 5.635907297385802, "learning_rate": 9.358487937537635e-06, "loss": 1.1309, "step": 2624 }, { "epoch": 0.19, "grad_norm": 18.09044720022509, "learning_rate": 9.357918723702443e-06, "loss": 1.065, "step": 2625 }, { "epoch": 0.19, "grad_norm": 10.054538369075098, "learning_rate": 9.357349274774032e-06, "loss": 0.8771, "step": 2626 }, { "epoch": 0.19, "grad_norm": 4.752084179929212, "learning_rate": 9.35677959078312e-06, "loss": 0.9093, "step": 2627 }, { "epoch": 0.19, "grad_norm": 4.305078515131577, "learning_rate": 9.356209671760443e-06, "loss": 0.2574, "step": 2628 }, { "epoch": 0.19, "grad_norm": 4.370568277125726, "learning_rate": 9.355639517736742e-06, "loss": 0.2419, "step": 2629 }, { "epoch": 0.19, "grad_norm": 19.961566189942367, "learning_rate": 9.355069128742775e-06, "loss": 1.9363, "step": 2630 }, { "epoch": 0.19, "grad_norm": 5.187566203730704, "learning_rate": 9.354498504809313e-06, "loss": 0.1655, "step": 2631 }, { "epoch": 0.19, "grad_norm": 5.650416915890465, "learning_rate": 9.353927645967142e-06, "loss": 0.2766, "step": 2632 }, { "epoch": 0.19, "grad_norm": 4.962840239157679, "learning_rate": 9.353356552247052e-06, "loss": 0.2, "step": 2633 }, { "epoch": 0.19, "grad_norm": 12.114856502781338, "learning_rate": 9.352785223679857e-06, "loss": 1.1423, "step": 2634 }, { "epoch": 0.19, "grad_norm": 3.7244920279079907, "learning_rate": 9.352213660296374e-06, "loss": 0.2329, "step": 2635 }, { "epoch": 0.19, "grad_norm": 3.7229950052026566, "learning_rate": 9.351641862127438e-06, "loss": 0.2729, "step": 2636 }, { "epoch": 0.19, "grad_norm": 13.565083230696155, "learning_rate": 9.351069829203896e-06, "loss": 0.9966, "step": 2637 }, { "epoch": 0.19, "grad_norm": 5.088358147992582, "learning_rate": 9.350497561556606e-06, "loss": 0.3811, "step": 2638 }, { "epoch": 0.19, "grad_norm": 5.773553323818067, "learning_rate": 9.349925059216441e-06, "loss": 0.7636, "step": 2639 }, { "epoch": 0.19, "grad_norm": 4.47672387053809, "learning_rate": 9.349352322214282e-06, "loss": 0.2881, "step": 2640 }, { "epoch": 0.19, "grad_norm": 14.155499733985225, "learning_rate": 9.34877935058103e-06, "loss": 0.9218, "step": 2641 }, { "epoch": 0.19, "grad_norm": 39.34068554997418, "learning_rate": 9.348206144347593e-06, "loss": 0.9454, "step": 2642 }, { "epoch": 0.19, "grad_norm": 19.01815980875869, "learning_rate": 9.347632703544893e-06, "loss": 1.2929, "step": 2643 }, { "epoch": 0.19, "grad_norm": 19.145436028094284, "learning_rate": 9.347059028203866e-06, "loss": 1.4423, "step": 2644 }, { "epoch": 0.19, "grad_norm": 27.20203022857791, "learning_rate": 9.34648511835546e-06, "loss": 1.0694, "step": 2645 }, { "epoch": 0.19, "grad_norm": 14.877630581143936, "learning_rate": 9.345910974030632e-06, "loss": 1.2237, "step": 2646 }, { "epoch": 0.19, "grad_norm": 5.375467791775614, "learning_rate": 9.345336595260358e-06, "loss": 1.1515, "step": 2647 }, { "epoch": 0.19, "grad_norm": 62.72022801803814, "learning_rate": 9.344761982075622e-06, "loss": 1.1255, "step": 2648 }, { "epoch": 0.19, "grad_norm": 5.034778886150017, "learning_rate": 9.344187134507425e-06, "loss": 0.3126, "step": 2649 }, { "epoch": 0.19, "grad_norm": 4.150752865448122, "learning_rate": 9.343612052586774e-06, "loss": 0.2388, "step": 2650 }, { "epoch": 0.19, "grad_norm": 50.748698247117964, "learning_rate": 9.343036736344695e-06, "loss": 1.7213, "step": 2651 }, { "epoch": 0.19, "grad_norm": 16.38216549955065, "learning_rate": 9.34246118581222e-06, "loss": 0.7722, "step": 2652 }, { "epoch": 0.19, "grad_norm": 4.702676798507811, "learning_rate": 9.341885401020405e-06, "loss": 0.3358, "step": 2653 }, { "epoch": 0.19, "grad_norm": 5.578573965155407, "learning_rate": 9.341309382000307e-06, "loss": 0.9551, "step": 2654 }, { "epoch": 0.19, "grad_norm": 4.456898806341073, "learning_rate": 9.340733128783e-06, "loss": 0.3133, "step": 2655 }, { "epoch": 0.19, "grad_norm": 4.3277468805622945, "learning_rate": 9.340156641399572e-06, "loss": 0.2765, "step": 2656 }, { "epoch": 0.19, "grad_norm": 3.794853336575069, "learning_rate": 9.339579919881122e-06, "loss": 0.2491, "step": 2657 }, { "epoch": 0.19, "grad_norm": 13.732386328770435, "learning_rate": 9.339002964258758e-06, "loss": 1.0039, "step": 2658 }, { "epoch": 0.19, "grad_norm": 21.25081894412962, "learning_rate": 9.33842577456361e-06, "loss": 0.912, "step": 2659 }, { "epoch": 0.19, "grad_norm": 5.316769683107164, "learning_rate": 9.337848350826815e-06, "loss": 1.0614, "step": 2660 }, { "epoch": 0.19, "grad_norm": 3.9997721504105925, "learning_rate": 9.337270693079521e-06, "loss": 0.2096, "step": 2661 }, { "epoch": 0.19, "grad_norm": 5.228753239582995, "learning_rate": 9.336692801352889e-06, "loss": 0.3357, "step": 2662 }, { "epoch": 0.19, "grad_norm": 3.546603977672413, "learning_rate": 9.336114675678096e-06, "loss": 0.2076, "step": 2663 }, { "epoch": 0.19, "grad_norm": 13.107387801107883, "learning_rate": 9.33553631608633e-06, "loss": 1.2107, "step": 2664 }, { "epoch": 0.19, "grad_norm": 5.656522216916619, "learning_rate": 9.33495772260879e-06, "loss": 1.0975, "step": 2665 }, { "epoch": 0.19, "grad_norm": 4.918614655829753, "learning_rate": 9.33437889527669e-06, "loss": 0.2142, "step": 2666 }, { "epoch": 0.19, "grad_norm": 3.873534747516526, "learning_rate": 9.333799834121256e-06, "loss": 0.2546, "step": 2667 }, { "epoch": 0.19, "grad_norm": 30.402562885535215, "learning_rate": 9.333220539173722e-06, "loss": 1.5331, "step": 2668 }, { "epoch": 0.19, "grad_norm": 15.165476289085376, "learning_rate": 9.332641010465346e-06, "loss": 1.5466, "step": 2669 }, { "epoch": 0.19, "grad_norm": 4.825217409963464, "learning_rate": 9.332061248027385e-06, "loss": 0.2669, "step": 2670 }, { "epoch": 0.19, "grad_norm": 4.205816816344974, "learning_rate": 9.331481251891117e-06, "loss": 0.2323, "step": 2671 }, { "epoch": 0.19, "grad_norm": 20.909515797829982, "learning_rate": 9.330901022087833e-06, "loss": 1.5818, "step": 2672 }, { "epoch": 0.19, "grad_norm": 14.98311975766101, "learning_rate": 9.330320558648831e-06, "loss": 1.1445, "step": 2673 }, { "epoch": 0.19, "grad_norm": 4.397611332010402, "learning_rate": 9.329739861605427e-06, "loss": 0.2704, "step": 2674 }, { "epoch": 0.19, "grad_norm": 3.9018256114604184, "learning_rate": 9.329158930988944e-06, "loss": 0.2801, "step": 2675 }, { "epoch": 0.19, "grad_norm": 4.028541589083036, "learning_rate": 9.328577766830726e-06, "loss": 0.3044, "step": 2676 }, { "epoch": 0.19, "grad_norm": 4.207639198719313, "learning_rate": 9.32799636916212e-06, "loss": 0.3376, "step": 2677 }, { "epoch": 0.19, "grad_norm": 4.290812025811822, "learning_rate": 9.327414738014496e-06, "loss": 0.3919, "step": 2678 }, { "epoch": 0.19, "grad_norm": 3.9863332453302327, "learning_rate": 9.326832873419222e-06, "loss": 0.3025, "step": 2679 }, { "epoch": 0.19, "grad_norm": 4.729619349476116, "learning_rate": 9.326250775407696e-06, "loss": 1.004, "step": 2680 }, { "epoch": 0.19, "grad_norm": 19.092791632583467, "learning_rate": 9.325668444011315e-06, "loss": 1.2729, "step": 2681 }, { "epoch": 0.19, "grad_norm": 15.972446609147955, "learning_rate": 9.325085879261494e-06, "loss": 0.8452, "step": 2682 }, { "epoch": 0.19, "grad_norm": 3.8224176199035402, "learning_rate": 9.324503081189665e-06, "loss": 0.3129, "step": 2683 }, { "epoch": 0.19, "grad_norm": 14.731487500310603, "learning_rate": 9.323920049827259e-06, "loss": 0.9328, "step": 2684 }, { "epoch": 0.19, "grad_norm": 10.817561373821752, "learning_rate": 9.323336785205737e-06, "loss": 0.8857, "step": 2685 }, { "epoch": 0.19, "grad_norm": 4.597030632364986, "learning_rate": 9.322753287356559e-06, "loss": 1.0692, "step": 2686 }, { "epoch": 0.19, "grad_norm": 3.7114181859875437, "learning_rate": 9.322169556311203e-06, "loss": 0.4019, "step": 2687 }, { "epoch": 0.19, "grad_norm": 16.33584707615083, "learning_rate": 9.32158559210116e-06, "loss": 1.1266, "step": 2688 }, { "epoch": 0.19, "grad_norm": 4.181133587983451, "learning_rate": 9.321001394757932e-06, "loss": 0.2708, "step": 2689 }, { "epoch": 0.19, "grad_norm": 5.943331288181301, "learning_rate": 9.320416964313034e-06, "loss": 0.8972, "step": 2690 }, { "epoch": 0.19, "grad_norm": 17.537645561694106, "learning_rate": 9.319832300797997e-06, "loss": 1.1452, "step": 2691 }, { "epoch": 0.19, "grad_norm": 4.186710702317397, "learning_rate": 9.319247404244357e-06, "loss": 0.2285, "step": 2692 }, { "epoch": 0.19, "grad_norm": 3.5779657061429955, "learning_rate": 9.31866227468367e-06, "loss": 0.2392, "step": 2693 }, { "epoch": 0.19, "grad_norm": 3.850781400001379, "learning_rate": 9.3180769121475e-06, "loss": 0.2406, "step": 2694 }, { "epoch": 0.19, "grad_norm": 21.78435479769487, "learning_rate": 9.317491316667423e-06, "loss": 1.6115, "step": 2695 }, { "epoch": 0.19, "grad_norm": 4.37568050848537, "learning_rate": 9.316905488275035e-06, "loss": 0.2674, "step": 2696 }, { "epoch": 0.19, "grad_norm": 53.3413772819399, "learning_rate": 9.316319427001936e-06, "loss": 1.4604, "step": 2697 }, { "epoch": 0.19, "grad_norm": 4.302124907804513, "learning_rate": 9.31573313287974e-06, "loss": 0.2525, "step": 2698 }, { "epoch": 0.19, "grad_norm": 4.076214130642502, "learning_rate": 9.31514660594008e-06, "loss": 0.2324, "step": 2699 }, { "epoch": 0.19, "grad_norm": 4.682004923272012, "learning_rate": 9.314559846214593e-06, "loss": 0.3338, "step": 2700 }, { "epoch": 0.19, "grad_norm": 4.886346511163569, "learning_rate": 9.313972853734932e-06, "loss": 0.4524, "step": 2701 }, { "epoch": 0.19, "grad_norm": 5.620601490340341, "learning_rate": 9.313385628532767e-06, "loss": 0.8898, "step": 2702 }, { "epoch": 0.19, "grad_norm": 6.828000766324256, "learning_rate": 9.312798170639774e-06, "loss": 0.2832, "step": 2703 }, { "epoch": 0.19, "grad_norm": 6.022865096956735, "learning_rate": 9.312210480087645e-06, "loss": 1.0913, "step": 2704 }, { "epoch": 0.19, "grad_norm": 4.24287599096266, "learning_rate": 9.311622556908082e-06, "loss": 0.225, "step": 2705 }, { "epoch": 0.19, "grad_norm": 5.802566789378067, "learning_rate": 9.311034401132804e-06, "loss": 0.8747, "step": 2706 }, { "epoch": 0.19, "grad_norm": 4.508281837046009, "learning_rate": 9.310446012793537e-06, "loss": 0.2726, "step": 2707 }, { "epoch": 0.19, "grad_norm": 3.7456085812518367, "learning_rate": 9.309857391922024e-06, "loss": 0.1889, "step": 2708 }, { "epoch": 0.19, "grad_norm": 26.49520518231017, "learning_rate": 9.309268538550017e-06, "loss": 1.8501, "step": 2709 }, { "epoch": 0.19, "grad_norm": 9.258543449799134, "learning_rate": 9.308679452709287e-06, "loss": 0.9864, "step": 2710 }, { "epoch": 0.19, "grad_norm": 18.22041087606338, "learning_rate": 9.308090134431607e-06, "loss": 1.1243, "step": 2711 }, { "epoch": 0.19, "grad_norm": 11.423617769081206, "learning_rate": 9.307500583748772e-06, "loss": 1.1375, "step": 2712 }, { "epoch": 0.19, "grad_norm": 16.984419906636653, "learning_rate": 9.306910800692586e-06, "loss": 1.1002, "step": 2713 }, { "epoch": 0.19, "grad_norm": 5.714476641236882, "learning_rate": 9.306320785294865e-06, "loss": 0.3425, "step": 2714 }, { "epoch": 0.19, "grad_norm": 19.283695653151387, "learning_rate": 9.305730537587436e-06, "loss": 1.1352, "step": 2715 }, { "epoch": 0.19, "grad_norm": 4.199543690766468, "learning_rate": 9.305140057602145e-06, "loss": 0.1897, "step": 2716 }, { "epoch": 0.19, "grad_norm": 3.9751781035885014, "learning_rate": 9.304549345370841e-06, "loss": 0.2719, "step": 2717 }, { "epoch": 0.19, "grad_norm": 7.95996383722991, "learning_rate": 9.303958400925396e-06, "loss": 0.8446, "step": 2718 }, { "epoch": 0.19, "grad_norm": 9.820970798600381, "learning_rate": 9.303367224297687e-06, "loss": 0.9608, "step": 2719 }, { "epoch": 0.2, "grad_norm": 14.412173269805901, "learning_rate": 9.302775815519603e-06, "loss": 1.0596, "step": 2720 }, { "epoch": 0.2, "grad_norm": 30.122006751372425, "learning_rate": 9.302184174623052e-06, "loss": 0.9512, "step": 2721 }, { "epoch": 0.2, "grad_norm": 17.214635008085377, "learning_rate": 9.301592301639948e-06, "loss": 1.302, "step": 2722 }, { "epoch": 0.2, "grad_norm": 7.692272686348032, "learning_rate": 9.301000196602223e-06, "loss": 0.7072, "step": 2723 }, { "epoch": 0.2, "grad_norm": 2.838633403654442, "learning_rate": 9.300407859541817e-06, "loss": 0.1955, "step": 2724 }, { "epoch": 0.2, "grad_norm": 7.617672704720693, "learning_rate": 9.299815290490686e-06, "loss": 0.8237, "step": 2725 }, { "epoch": 0.2, "grad_norm": 7.9847480480951605, "learning_rate": 9.299222489480795e-06, "loss": 1.055, "step": 2726 }, { "epoch": 0.2, "grad_norm": 5.016043876451588, "learning_rate": 9.298629456544123e-06, "loss": 1.1962, "step": 2727 }, { "epoch": 0.2, "grad_norm": 13.043150545311418, "learning_rate": 9.298036191712665e-06, "loss": 0.7367, "step": 2728 }, { "epoch": 0.2, "grad_norm": 7.87640882839761, "learning_rate": 9.297442695018421e-06, "loss": 0.6807, "step": 2729 }, { "epoch": 0.2, "grad_norm": 5.057084682705021, "learning_rate": 9.296848966493412e-06, "loss": 0.9118, "step": 2730 }, { "epoch": 0.2, "grad_norm": 15.49524762018196, "learning_rate": 9.296255006169667e-06, "loss": 1.3123, "step": 2731 }, { "epoch": 0.2, "grad_norm": 5.871023197806695, "learning_rate": 9.295660814079224e-06, "loss": 0.9529, "step": 2732 }, { "epoch": 0.2, "grad_norm": 18.441657477501902, "learning_rate": 9.29506639025414e-06, "loss": 1.0666, "step": 2733 }, { "epoch": 0.2, "grad_norm": 3.7933806309585707, "learning_rate": 9.294471734726486e-06, "loss": 0.2512, "step": 2734 }, { "epoch": 0.2, "grad_norm": 15.78287430580173, "learning_rate": 9.293876847528333e-06, "loss": 0.5545, "step": 2735 }, { "epoch": 0.2, "grad_norm": 49.55496824592135, "learning_rate": 9.29328172869178e-06, "loss": 1.2743, "step": 2736 }, { "epoch": 0.2, "grad_norm": 9.017166339863355, "learning_rate": 9.292686378248927e-06, "loss": 0.7016, "step": 2737 }, { "epoch": 0.2, "grad_norm": 19.30421170912958, "learning_rate": 9.292090796231893e-06, "loss": 1.4845, "step": 2738 }, { "epoch": 0.2, "grad_norm": 4.120375210922695, "learning_rate": 9.291494982672808e-06, "loss": 0.3057, "step": 2739 }, { "epoch": 0.2, "grad_norm": 16.030355328251698, "learning_rate": 9.290898937603811e-06, "loss": 1.2458, "step": 2740 }, { "epoch": 0.2, "grad_norm": 21.870904373937844, "learning_rate": 9.29030266105706e-06, "loss": 1.9131, "step": 2741 }, { "epoch": 0.2, "grad_norm": 32.68623753018599, "learning_rate": 9.28970615306472e-06, "loss": 1.0962, "step": 2742 }, { "epoch": 0.2, "grad_norm": 4.211754535317358, "learning_rate": 9.28910941365897e-06, "loss": 0.3021, "step": 2743 }, { "epoch": 0.2, "grad_norm": 16.28599147415597, "learning_rate": 9.288512442872002e-06, "loss": 0.969, "step": 2744 }, { "epoch": 0.2, "grad_norm": 17.065804889002376, "learning_rate": 9.28791524073602e-06, "loss": 1.1367, "step": 2745 }, { "epoch": 0.2, "grad_norm": 4.405988103075916, "learning_rate": 9.287317807283243e-06, "loss": 0.2393, "step": 2746 }, { "epoch": 0.2, "grad_norm": 5.969840677691524, "learning_rate": 9.286720142545897e-06, "loss": 1.0744, "step": 2747 }, { "epoch": 0.2, "grad_norm": 4.703892875501522, "learning_rate": 9.286122246556227e-06, "loss": 0.305, "step": 2748 }, { "epoch": 0.2, "grad_norm": 10.429256236065939, "learning_rate": 9.285524119346483e-06, "loss": 1.3711, "step": 2749 }, { "epoch": 0.2, "grad_norm": 3.364378399737358, "learning_rate": 9.284925760948938e-06, "loss": 0.2083, "step": 2750 }, { "epoch": 0.2, "grad_norm": 5.416510983316586, "learning_rate": 9.284327171395866e-06, "loss": 0.5306, "step": 2751 }, { "epoch": 0.2, "grad_norm": 11.547256222558211, "learning_rate": 9.283728350719559e-06, "loss": 0.9671, "step": 2752 }, { "epoch": 0.2, "grad_norm": 3.723525097538658, "learning_rate": 9.283129298952322e-06, "loss": 0.3678, "step": 2753 }, { "epoch": 0.2, "grad_norm": 2.9311802007721615, "learning_rate": 9.282530016126474e-06, "loss": 0.1645, "step": 2754 }, { "epoch": 0.2, "grad_norm": 3.8413389460234333, "learning_rate": 9.281930502274341e-06, "loss": 0.2833, "step": 2755 }, { "epoch": 0.2, "grad_norm": 3.9948265485479566, "learning_rate": 9.281330757428265e-06, "loss": 0.2571, "step": 2756 }, { "epoch": 0.2, "grad_norm": 36.25088508624206, "learning_rate": 9.280730781620602e-06, "loss": 2.0341, "step": 2757 }, { "epoch": 0.2, "grad_norm": 20.352522212667992, "learning_rate": 9.280130574883714e-06, "loss": 1.2502, "step": 2758 }, { "epoch": 0.2, "grad_norm": 6.078198047557033, "learning_rate": 9.279530137249985e-06, "loss": 0.9452, "step": 2759 }, { "epoch": 0.2, "grad_norm": 4.687917272436424, "learning_rate": 9.278929468751804e-06, "loss": 0.2524, "step": 2760 }, { "epoch": 0.2, "grad_norm": 10.647087018203047, "learning_rate": 9.278328569421574e-06, "loss": 0.6725, "step": 2761 }, { "epoch": 0.2, "grad_norm": 14.020815844267561, "learning_rate": 9.27772743929171e-06, "loss": 0.5613, "step": 2762 }, { "epoch": 0.2, "grad_norm": 4.319222989577761, "learning_rate": 9.277126078394645e-06, "loss": 0.2658, "step": 2763 }, { "epoch": 0.2, "grad_norm": 11.70082901252625, "learning_rate": 9.276524486762819e-06, "loss": 0.6479, "step": 2764 }, { "epoch": 0.2, "grad_norm": 13.563499936975463, "learning_rate": 9.275922664428682e-06, "loss": 0.8429, "step": 2765 }, { "epoch": 0.2, "grad_norm": 4.172871260662168, "learning_rate": 9.275320611424703e-06, "loss": 0.2823, "step": 2766 }, { "epoch": 0.2, "grad_norm": 5.27962422480369, "learning_rate": 9.27471832778336e-06, "loss": 0.9894, "step": 2767 }, { "epoch": 0.2, "grad_norm": 4.784361557009436, "learning_rate": 9.274115813537145e-06, "loss": 0.4024, "step": 2768 }, { "epoch": 0.2, "grad_norm": 4.992884036916852, "learning_rate": 9.27351306871856e-06, "loss": 1.0178, "step": 2769 }, { "epoch": 0.2, "grad_norm": 4.646270418290151, "learning_rate": 9.272910093360122e-06, "loss": 0.3132, "step": 2770 }, { "epoch": 0.2, "grad_norm": 4.344269598332943, "learning_rate": 9.272306887494355e-06, "loss": 0.2093, "step": 2771 }, { "epoch": 0.2, "grad_norm": 3.377425246445786, "learning_rate": 9.271703451153806e-06, "loss": 0.2551, "step": 2772 }, { "epoch": 0.2, "grad_norm": 3.9204089867106164, "learning_rate": 9.271099784371024e-06, "loss": 0.1992, "step": 2773 }, { "epoch": 0.2, "grad_norm": 4.209068129164345, "learning_rate": 9.270495887178575e-06, "loss": 0.2791, "step": 2774 }, { "epoch": 0.2, "grad_norm": 4.021597462232536, "learning_rate": 9.269891759609038e-06, "loss": 0.3406, "step": 2775 }, { "epoch": 0.2, "grad_norm": 33.27703225320053, "learning_rate": 9.269287401695005e-06, "loss": 1.1883, "step": 2776 }, { "epoch": 0.2, "grad_norm": 5.571804136403926, "learning_rate": 9.268682813469074e-06, "loss": 0.3293, "step": 2777 }, { "epoch": 0.2, "grad_norm": 22.944914529148285, "learning_rate": 9.268077994963864e-06, "loss": 1.4397, "step": 2778 }, { "epoch": 0.2, "grad_norm": 14.155581009613924, "learning_rate": 9.267472946212002e-06, "loss": 1.2901, "step": 2779 }, { "epoch": 0.2, "grad_norm": 4.376430637645879, "learning_rate": 9.266867667246127e-06, "loss": 0.3369, "step": 2780 }, { "epoch": 0.2, "grad_norm": 31.493758364053242, "learning_rate": 9.266262158098893e-06, "loss": 1.6866, "step": 2781 }, { "epoch": 0.2, "grad_norm": 5.32536402582835, "learning_rate": 9.265656418802964e-06, "loss": 0.2022, "step": 2782 }, { "epoch": 0.2, "grad_norm": 6.31155175680463, "learning_rate": 9.265050449391018e-06, "loss": 0.2407, "step": 2783 }, { "epoch": 0.2, "grad_norm": 5.625454459577603, "learning_rate": 9.264444249895745e-06, "loss": 0.3738, "step": 2784 }, { "epoch": 0.2, "grad_norm": 19.328594909445762, "learning_rate": 9.263837820349846e-06, "loss": 1.2652, "step": 2785 }, { "epoch": 0.2, "grad_norm": 3.620434354673233, "learning_rate": 9.263231160786035e-06, "loss": 0.2092, "step": 2786 }, { "epoch": 0.2, "grad_norm": 16.775737735386517, "learning_rate": 9.262624271237042e-06, "loss": 1.0757, "step": 2787 }, { "epoch": 0.2, "grad_norm": 17.961108106032242, "learning_rate": 9.262017151735602e-06, "loss": 1.2476, "step": 2788 }, { "epoch": 0.2, "grad_norm": 14.9551950095577, "learning_rate": 9.261409802314472e-06, "loss": 1.1033, "step": 2789 }, { "epoch": 0.2, "grad_norm": 11.753873330313821, "learning_rate": 9.26080222300641e-06, "loss": 1.0457, "step": 2790 }, { "epoch": 0.2, "grad_norm": 28.62130096880451, "learning_rate": 9.2601944138442e-06, "loss": 1.4329, "step": 2791 }, { "epoch": 0.2, "grad_norm": 5.533362093336271, "learning_rate": 9.259586374860625e-06, "loss": 1.1133, "step": 2792 }, { "epoch": 0.2, "grad_norm": 4.590869443119343, "learning_rate": 9.258978106088488e-06, "loss": 0.2165, "step": 2793 }, { "epoch": 0.2, "grad_norm": 17.358940511636646, "learning_rate": 9.258369607560605e-06, "loss": 1.3204, "step": 2794 }, { "epoch": 0.2, "grad_norm": 18.56226885525285, "learning_rate": 9.257760879309799e-06, "loss": 1.5928, "step": 2795 }, { "epoch": 0.2, "grad_norm": 4.425166651255054, "learning_rate": 9.25715192136891e-06, "loss": 0.2462, "step": 2796 }, { "epoch": 0.2, "grad_norm": 4.685727705915668, "learning_rate": 9.25654273377079e-06, "loss": 0.2777, "step": 2797 }, { "epoch": 0.2, "grad_norm": 35.362754989762465, "learning_rate": 9.255933316548302e-06, "loss": 1.3705, "step": 2798 }, { "epoch": 0.2, "grad_norm": 20.965799762663977, "learning_rate": 9.25532366973432e-06, "loss": 1.2749, "step": 2799 }, { "epoch": 0.2, "grad_norm": 17.64105215527218, "learning_rate": 9.254713793361733e-06, "loss": 1.0433, "step": 2800 }, { "epoch": 0.2, "grad_norm": 4.217001183930291, "learning_rate": 9.254103687463443e-06, "loss": 0.2846, "step": 2801 }, { "epoch": 0.2, "grad_norm": 4.290856290087183, "learning_rate": 9.25349335207236e-06, "loss": 0.2045, "step": 2802 }, { "epoch": 0.2, "grad_norm": 3.983955220735674, "learning_rate": 9.252882787221413e-06, "loss": 0.1693, "step": 2803 }, { "epoch": 0.2, "grad_norm": 23.298450494343342, "learning_rate": 9.252271992943536e-06, "loss": 1.1349, "step": 2804 }, { "epoch": 0.2, "grad_norm": 14.714952262336014, "learning_rate": 9.251660969271682e-06, "loss": 1.2158, "step": 2805 }, { "epoch": 0.2, "grad_norm": 4.995315610922858, "learning_rate": 9.251049716238811e-06, "loss": 0.2503, "step": 2806 }, { "epoch": 0.2, "grad_norm": 4.2835393687707715, "learning_rate": 9.2504382338779e-06, "loss": 0.304, "step": 2807 }, { "epoch": 0.2, "grad_norm": 6.3152191219289175, "learning_rate": 9.249826522221935e-06, "loss": 0.3355, "step": 2808 }, { "epoch": 0.2, "grad_norm": 5.118680382696014, "learning_rate": 9.249214581303913e-06, "loss": 0.2291, "step": 2809 }, { "epoch": 0.2, "grad_norm": 4.465382932851689, "learning_rate": 9.248602411156852e-06, "loss": 0.3146, "step": 2810 }, { "epoch": 0.2, "grad_norm": 31.583091244966592, "learning_rate": 9.247990011813771e-06, "loss": 0.5507, "step": 2811 }, { "epoch": 0.2, "grad_norm": 3.7041967268009266, "learning_rate": 9.247377383307709e-06, "loss": 0.3247, "step": 2812 }, { "epoch": 0.2, "grad_norm": 4.711921779165656, "learning_rate": 9.246764525671714e-06, "loss": 0.3583, "step": 2813 }, { "epoch": 0.2, "grad_norm": 4.765745836983188, "learning_rate": 9.246151438938847e-06, "loss": 0.9989, "step": 2814 }, { "epoch": 0.2, "grad_norm": 3.263027180389333, "learning_rate": 9.245538123142184e-06, "loss": 0.2449, "step": 2815 }, { "epoch": 0.2, "grad_norm": 4.702492497004213, "learning_rate": 9.244924578314808e-06, "loss": 0.939, "step": 2816 }, { "epoch": 0.2, "grad_norm": 3.9034228932106756, "learning_rate": 9.24431080448982e-06, "loss": 0.3314, "step": 2817 }, { "epoch": 0.2, "grad_norm": 4.035067653929308, "learning_rate": 9.24369680170033e-06, "loss": 0.2226, "step": 2818 }, { "epoch": 0.2, "grad_norm": 5.4779343171317345, "learning_rate": 9.243082569979458e-06, "loss": 1.0762, "step": 2819 }, { "epoch": 0.2, "grad_norm": 4.175447495712283, "learning_rate": 9.242468109360345e-06, "loss": 0.2603, "step": 2820 }, { "epoch": 0.2, "grad_norm": 14.416338471582941, "learning_rate": 9.241853419876136e-06, "loss": 1.5403, "step": 2821 }, { "epoch": 0.2, "grad_norm": 4.686788073842483, "learning_rate": 9.24123850155999e-06, "loss": 0.3023, "step": 2822 }, { "epoch": 0.2, "grad_norm": 11.965954439773926, "learning_rate": 9.240623354445083e-06, "loss": 0.7216, "step": 2823 }, { "epoch": 0.2, "grad_norm": 9.480257494219407, "learning_rate": 9.240007978564596e-06, "loss": 0.6805, "step": 2824 }, { "epoch": 0.2, "grad_norm": 3.9623235926535303, "learning_rate": 9.239392373951727e-06, "loss": 0.2776, "step": 2825 }, { "epoch": 0.2, "grad_norm": 4.879394573805734, "learning_rate": 9.238776540639688e-06, "loss": 1.0134, "step": 2826 }, { "epoch": 0.2, "grad_norm": 4.385703760282075, "learning_rate": 9.2381604786617e-06, "loss": 0.2543, "step": 2827 }, { "epoch": 0.2, "grad_norm": 3.9601306190737717, "learning_rate": 9.237544188050994e-06, "loss": 0.2366, "step": 2828 }, { "epoch": 0.2, "grad_norm": 18.552246612525785, "learning_rate": 9.23692766884082e-06, "loss": 1.3338, "step": 2829 }, { "epoch": 0.2, "grad_norm": 4.213582894437286, "learning_rate": 9.236310921064437e-06, "loss": 0.2457, "step": 2830 }, { "epoch": 0.2, "grad_norm": 5.479740795956514, "learning_rate": 9.235693944755114e-06, "loss": 1.0599, "step": 2831 }, { "epoch": 0.2, "grad_norm": 25.152260653742232, "learning_rate": 9.235076739946136e-06, "loss": 1.7029, "step": 2832 }, { "epoch": 0.2, "grad_norm": 28.05386144395111, "learning_rate": 9.234459306670798e-06, "loss": 0.8652, "step": 2833 }, { "epoch": 0.2, "grad_norm": 15.499614845630694, "learning_rate": 9.23384164496241e-06, "loss": 1.4732, "step": 2834 }, { "epoch": 0.2, "grad_norm": 6.088551905781854, "learning_rate": 9.23322375485429e-06, "loss": 0.985, "step": 2835 }, { "epoch": 0.2, "grad_norm": 3.7920005061329327, "learning_rate": 9.232605636379772e-06, "loss": 0.2179, "step": 2836 }, { "epoch": 0.2, "grad_norm": 4.191701883600488, "learning_rate": 9.231987289572201e-06, "loss": 0.2577, "step": 2837 }, { "epoch": 0.2, "grad_norm": 12.836061855186742, "learning_rate": 9.231368714464937e-06, "loss": 0.8471, "step": 2838 }, { "epoch": 0.2, "grad_norm": 3.702723047115035, "learning_rate": 9.230749911091346e-06, "loss": 0.2787, "step": 2839 }, { "epoch": 0.2, "grad_norm": 3.984074168777282, "learning_rate": 9.230130879484811e-06, "loss": 0.3259, "step": 2840 }, { "epoch": 0.2, "grad_norm": 31.2012185759383, "learning_rate": 9.229511619678727e-06, "loss": 1.3915, "step": 2841 }, { "epoch": 0.2, "grad_norm": 14.7036376088298, "learning_rate": 9.228892131706502e-06, "loss": 0.8939, "step": 2842 }, { "epoch": 0.2, "grad_norm": 3.572116035824458, "learning_rate": 9.228272415601552e-06, "loss": 0.2101, "step": 2843 }, { "epoch": 0.2, "grad_norm": 7.14408853281272, "learning_rate": 9.22765247139731e-06, "loss": 0.9249, "step": 2844 }, { "epoch": 0.2, "grad_norm": 6.058269574223726, "learning_rate": 9.227032299127221e-06, "loss": 1.0679, "step": 2845 }, { "epoch": 0.2, "grad_norm": 6.200225245542236, "learning_rate": 9.226411898824738e-06, "loss": 1.0264, "step": 2846 }, { "epoch": 0.2, "grad_norm": 3.9960667351876986, "learning_rate": 9.225791270523332e-06, "loss": 0.2551, "step": 2847 }, { "epoch": 0.2, "grad_norm": 8.628831714530966, "learning_rate": 9.225170414256483e-06, "loss": 0.5565, "step": 2848 }, { "epoch": 0.2, "grad_norm": 3.539032853374798, "learning_rate": 9.224549330057684e-06, "loss": 0.3188, "step": 2849 }, { "epoch": 0.2, "grad_norm": 17.900955732544315, "learning_rate": 9.223928017960437e-06, "loss": 1.4008, "step": 2850 }, { "epoch": 0.2, "grad_norm": 6.147714684084873, "learning_rate": 9.223306477998265e-06, "loss": 0.3503, "step": 2851 }, { "epoch": 0.2, "grad_norm": 13.765732072832181, "learning_rate": 9.22268471020469e-06, "loss": 1.0067, "step": 2852 }, { "epoch": 0.2, "grad_norm": 5.764160624244239, "learning_rate": 9.222062714613263e-06, "loss": 1.0016, "step": 2853 }, { "epoch": 0.2, "grad_norm": 2.565716544587312, "learning_rate": 9.221440491257531e-06, "loss": 0.1979, "step": 2854 }, { "epoch": 0.2, "grad_norm": 4.309617968305675, "learning_rate": 9.220818040171065e-06, "loss": 0.4708, "step": 2855 }, { "epoch": 0.2, "grad_norm": 7.153126624025964, "learning_rate": 9.220195361387442e-06, "loss": 1.0418, "step": 2856 }, { "epoch": 0.2, "grad_norm": 8.82733995227428, "learning_rate": 9.219572454940253e-06, "loss": 0.8544, "step": 2857 }, { "epoch": 0.2, "grad_norm": 3.456571179579801, "learning_rate": 9.218949320863104e-06, "loss": 0.2653, "step": 2858 }, { "epoch": 0.21, "grad_norm": 3.748814694353519, "learning_rate": 9.218325959189608e-06, "loss": 0.2106, "step": 2859 }, { "epoch": 0.21, "grad_norm": 3.9965071842202833, "learning_rate": 9.217702369953392e-06, "loss": 0.3634, "step": 2860 }, { "epoch": 0.21, "grad_norm": 2.9965133387287888, "learning_rate": 9.217078553188099e-06, "loss": 0.1937, "step": 2861 }, { "epoch": 0.21, "grad_norm": 3.2392950958011313, "learning_rate": 9.216454508927381e-06, "loss": 0.2502, "step": 2862 }, { "epoch": 0.21, "grad_norm": 5.0638071192677305, "learning_rate": 9.215830237204902e-06, "loss": 1.0381, "step": 2863 }, { "epoch": 0.21, "grad_norm": 3.90029378344728, "learning_rate": 9.215205738054339e-06, "loss": 0.2064, "step": 2864 }, { "epoch": 0.21, "grad_norm": 4.742145673923127, "learning_rate": 9.214581011509383e-06, "loss": 1.0942, "step": 2865 }, { "epoch": 0.21, "grad_norm": 3.3834428060135844, "learning_rate": 9.213956057603734e-06, "loss": 0.2687, "step": 2866 }, { "epoch": 0.21, "grad_norm": 3.408031588885121, "learning_rate": 9.213330876371105e-06, "loss": 0.2531, "step": 2867 }, { "epoch": 0.21, "grad_norm": 3.138902594886283, "learning_rate": 9.212705467845226e-06, "loss": 0.2072, "step": 2868 }, { "epoch": 0.21, "grad_norm": 3.5878437789368336, "learning_rate": 9.212079832059831e-06, "loss": 0.3282, "step": 2869 }, { "epoch": 0.21, "grad_norm": 25.067389056178335, "learning_rate": 9.211453969048673e-06, "loss": 1.3829, "step": 2870 }, { "epoch": 0.21, "grad_norm": 4.8216025750719265, "learning_rate": 9.210827878845516e-06, "loss": 0.2925, "step": 2871 }, { "epoch": 0.21, "grad_norm": 17.59800737520669, "learning_rate": 9.210201561484132e-06, "loss": 1.4005, "step": 2872 }, { "epoch": 0.21, "grad_norm": 26.07998589959625, "learning_rate": 9.20957501699831e-06, "loss": 2.0891, "step": 2873 }, { "epoch": 0.21, "grad_norm": 5.396479353545754, "learning_rate": 9.208948245421853e-06, "loss": 0.3352, "step": 2874 }, { "epoch": 0.21, "grad_norm": 5.593782377944323, "learning_rate": 9.208321246788567e-06, "loss": 1.0724, "step": 2875 }, { "epoch": 0.21, "grad_norm": 22.92307090240068, "learning_rate": 9.207694021132281e-06, "loss": 0.6486, "step": 2876 }, { "epoch": 0.21, "grad_norm": 4.655829906852639, "learning_rate": 9.207066568486827e-06, "loss": 0.8796, "step": 2877 }, { "epoch": 0.21, "grad_norm": 3.0283165341062137, "learning_rate": 9.206438888886058e-06, "loss": 0.1468, "step": 2878 }, { "epoch": 0.21, "grad_norm": 3.9076637460355665, "learning_rate": 9.205810982363835e-06, "loss": 0.2617, "step": 2879 }, { "epoch": 0.21, "grad_norm": 28.033422536334495, "learning_rate": 9.205182848954027e-06, "loss": 1.9197, "step": 2880 }, { "epoch": 0.21, "grad_norm": 21.381719564345527, "learning_rate": 9.20455448869052e-06, "loss": 1.4031, "step": 2881 }, { "epoch": 0.21, "grad_norm": 14.89734639636495, "learning_rate": 9.203925901607216e-06, "loss": 1.2527, "step": 2882 }, { "epoch": 0.21, "grad_norm": 5.085331563056285, "learning_rate": 9.203297087738023e-06, "loss": 1.1421, "step": 2883 }, { "epoch": 0.21, "grad_norm": 3.9349152090793287, "learning_rate": 9.20266804711686e-06, "loss": 0.2909, "step": 2884 }, { "epoch": 0.21, "grad_norm": 3.6520546907357208, "learning_rate": 9.202038779777666e-06, "loss": 0.3114, "step": 2885 }, { "epoch": 0.21, "grad_norm": 5.347355476425321, "learning_rate": 9.201409285754384e-06, "loss": 0.2998, "step": 2886 }, { "epoch": 0.21, "grad_norm": 5.108944819667609, "learning_rate": 9.200779565080976e-06, "loss": 1.1157, "step": 2887 }, { "epoch": 0.21, "grad_norm": 3.8154791190907242, "learning_rate": 9.20014961779141e-06, "loss": 0.3725, "step": 2888 }, { "epoch": 0.21, "grad_norm": 4.230376053809538, "learning_rate": 9.19951944391967e-06, "loss": 0.2451, "step": 2889 }, { "epoch": 0.21, "grad_norm": 16.63046203436031, "learning_rate": 9.198889043499752e-06, "loss": 0.9703, "step": 2890 }, { "epoch": 0.21, "grad_norm": 4.008632941120039, "learning_rate": 9.198258416565663e-06, "loss": 0.2227, "step": 2891 }, { "epoch": 0.21, "grad_norm": 3.8706021227219294, "learning_rate": 9.197627563151427e-06, "loss": 0.2118, "step": 2892 }, { "epoch": 0.21, "grad_norm": 4.5919975032374385, "learning_rate": 9.196996483291071e-06, "loss": 0.2936, "step": 2893 }, { "epoch": 0.21, "grad_norm": 17.047000815062148, "learning_rate": 9.19636517701864e-06, "loss": 1.1216, "step": 2894 }, { "epoch": 0.21, "grad_norm": 4.0137150009789275, "learning_rate": 9.195733644368195e-06, "loss": 0.3035, "step": 2895 }, { "epoch": 0.21, "grad_norm": 5.001936387273719, "learning_rate": 9.195101885373798e-06, "loss": 1.112, "step": 2896 }, { "epoch": 0.21, "grad_norm": 3.1587089129805754, "learning_rate": 9.194469900069536e-06, "loss": 0.1847, "step": 2897 }, { "epoch": 0.21, "grad_norm": 23.435165811213633, "learning_rate": 9.193837688489499e-06, "loss": 1.8416, "step": 2898 }, { "epoch": 0.21, "grad_norm": 3.943746724115875, "learning_rate": 9.193205250667792e-06, "loss": 0.2597, "step": 2899 }, { "epoch": 0.21, "grad_norm": 4.942842050621309, "learning_rate": 9.192572586638537e-06, "loss": 0.2462, "step": 2900 }, { "epoch": 0.21, "grad_norm": 21.599862772886592, "learning_rate": 9.191939696435857e-06, "loss": 1.4984, "step": 2901 }, { "epoch": 0.21, "grad_norm": 18.225313976865017, "learning_rate": 9.191306580093901e-06, "loss": 0.8834, "step": 2902 }, { "epoch": 0.21, "grad_norm": 28.369165939633376, "learning_rate": 9.190673237646817e-06, "loss": 1.2046, "step": 2903 }, { "epoch": 0.21, "grad_norm": 5.042259594517012, "learning_rate": 9.190039669128776e-06, "loss": 0.3075, "step": 2904 }, { "epoch": 0.21, "grad_norm": 3.9314532292123068, "learning_rate": 9.189405874573955e-06, "loss": 0.1545, "step": 2905 }, { "epoch": 0.21, "grad_norm": 23.836250782402765, "learning_rate": 9.188771854016546e-06, "loss": 1.1721, "step": 2906 }, { "epoch": 0.21, "grad_norm": 4.413036825823862, "learning_rate": 9.18813760749075e-06, "loss": 0.2839, "step": 2907 }, { "epoch": 0.21, "grad_norm": 5.318744509514477, "learning_rate": 9.187503135030782e-06, "loss": 1.1474, "step": 2908 }, { "epoch": 0.21, "grad_norm": 5.390355710565617, "learning_rate": 9.18686843667087e-06, "loss": 1.1225, "step": 2909 }, { "epoch": 0.21, "grad_norm": 3.838994882528301, "learning_rate": 9.186233512445255e-06, "loss": 0.2455, "step": 2910 }, { "epoch": 0.21, "grad_norm": 5.371340321095615, "learning_rate": 9.185598362388188e-06, "loss": 1.0914, "step": 2911 }, { "epoch": 0.21, "grad_norm": 17.016145774602247, "learning_rate": 9.184962986533933e-06, "loss": 1.6154, "step": 2912 }, { "epoch": 0.21, "grad_norm": 4.775729566857688, "learning_rate": 9.184327384916765e-06, "loss": 0.3213, "step": 2913 }, { "epoch": 0.21, "grad_norm": 4.262209421122261, "learning_rate": 9.183691557570973e-06, "loss": 0.2775, "step": 2914 }, { "epoch": 0.21, "grad_norm": 4.226581387061766, "learning_rate": 9.183055504530857e-06, "loss": 0.1996, "step": 2915 }, { "epoch": 0.21, "grad_norm": 3.934979447000386, "learning_rate": 9.182419225830733e-06, "loss": 0.2437, "step": 2916 }, { "epoch": 0.21, "grad_norm": 4.932828944771242, "learning_rate": 9.18178272150492e-06, "loss": 0.3063, "step": 2917 }, { "epoch": 0.21, "grad_norm": 4.6524999649663545, "learning_rate": 9.181145991587762e-06, "loss": 0.2271, "step": 2918 }, { "epoch": 0.21, "grad_norm": 5.4030022685175645, "learning_rate": 9.180509036113601e-06, "loss": 0.9673, "step": 2919 }, { "epoch": 0.21, "grad_norm": 4.851016833154705, "learning_rate": 9.179871855116802e-06, "loss": 0.3336, "step": 2920 }, { "epoch": 0.21, "grad_norm": 12.041130767693144, "learning_rate": 9.179234448631737e-06, "loss": 1.2238, "step": 2921 }, { "epoch": 0.21, "grad_norm": 3.539408169964969, "learning_rate": 9.178596816692796e-06, "loss": 0.2559, "step": 2922 }, { "epoch": 0.21, "grad_norm": 12.493663142943097, "learning_rate": 9.17795895933437e-06, "loss": 1.4622, "step": 2923 }, { "epoch": 0.21, "grad_norm": 4.920441136152477, "learning_rate": 9.177320876590874e-06, "loss": 0.9925, "step": 2924 }, { "epoch": 0.21, "grad_norm": 10.57439474292571, "learning_rate": 9.17668256849673e-06, "loss": 0.8799, "step": 2925 }, { "epoch": 0.21, "grad_norm": 3.5926176624635144, "learning_rate": 9.176044035086372e-06, "loss": 0.2166, "step": 2926 }, { "epoch": 0.21, "grad_norm": 4.8637432974316805, "learning_rate": 9.175405276394243e-06, "loss": 0.3377, "step": 2927 }, { "epoch": 0.21, "grad_norm": 14.36676989291613, "learning_rate": 9.174766292454804e-06, "loss": 1.5897, "step": 2928 }, { "epoch": 0.21, "grad_norm": 8.938036159675432, "learning_rate": 9.174127083302527e-06, "loss": 1.3826, "step": 2929 }, { "epoch": 0.21, "grad_norm": 3.4262217109814785, "learning_rate": 9.173487648971893e-06, "loss": 0.2359, "step": 2930 }, { "epoch": 0.21, "grad_norm": 19.603825116241413, "learning_rate": 9.172847989497397e-06, "loss": 1.5044, "step": 2931 }, { "epoch": 0.21, "grad_norm": 5.677777404036348, "learning_rate": 9.172208104913549e-06, "loss": 0.2992, "step": 2932 }, { "epoch": 0.21, "grad_norm": 4.379070539263362, "learning_rate": 9.171567995254867e-06, "loss": 0.2367, "step": 2933 }, { "epoch": 0.21, "grad_norm": 5.422013626963518, "learning_rate": 9.17092766055588e-06, "loss": 0.8806, "step": 2934 }, { "epoch": 0.21, "grad_norm": 4.99136035890354, "learning_rate": 9.170287100851133e-06, "loss": 0.3123, "step": 2935 }, { "epoch": 0.21, "grad_norm": 4.606353559624557, "learning_rate": 9.169646316175183e-06, "loss": 0.3039, "step": 2936 }, { "epoch": 0.21, "grad_norm": 7.305311573057084, "learning_rate": 9.169005306562598e-06, "loss": 0.9835, "step": 2937 }, { "epoch": 0.21, "grad_norm": 4.254737849736089, "learning_rate": 9.168364072047957e-06, "loss": 0.2872, "step": 2938 }, { "epoch": 0.21, "grad_norm": 13.396774404287425, "learning_rate": 9.167722612665853e-06, "loss": 0.7748, "step": 2939 }, { "epoch": 0.21, "grad_norm": 4.808856632922249, "learning_rate": 9.167080928450888e-06, "loss": 0.9185, "step": 2940 }, { "epoch": 0.21, "grad_norm": 3.382331524494441, "learning_rate": 9.16643901943768e-06, "loss": 0.2929, "step": 2941 }, { "epoch": 0.21, "grad_norm": 4.029659850495301, "learning_rate": 9.16579688566086e-06, "loss": 0.2415, "step": 2942 }, { "epoch": 0.21, "grad_norm": 3.8844487154442646, "learning_rate": 9.165154527155065e-06, "loss": 0.2773, "step": 2943 }, { "epoch": 0.21, "grad_norm": 20.22868915613621, "learning_rate": 9.16451194395495e-06, "loss": 0.8585, "step": 2944 }, { "epoch": 0.21, "grad_norm": 3.876093890622652, "learning_rate": 9.163869136095178e-06, "loss": 0.2341, "step": 2945 }, { "epoch": 0.21, "grad_norm": 12.95404452363587, "learning_rate": 9.163226103610429e-06, "loss": 1.3266, "step": 2946 }, { "epoch": 0.21, "grad_norm": 4.880261890005765, "learning_rate": 9.162582846535388e-06, "loss": 0.2546, "step": 2947 }, { "epoch": 0.21, "grad_norm": 3.9978277121194994, "learning_rate": 9.161939364904761e-06, "loss": 0.2854, "step": 2948 }, { "epoch": 0.21, "grad_norm": 7.473379120748451, "learning_rate": 9.161295658753257e-06, "loss": 1.0916, "step": 2949 }, { "epoch": 0.21, "grad_norm": 4.3316596645467955, "learning_rate": 9.160651728115606e-06, "loss": 0.2518, "step": 2950 }, { "epoch": 0.21, "grad_norm": 3.957368194807853, "learning_rate": 9.160007573026542e-06, "loss": 0.2225, "step": 2951 }, { "epoch": 0.21, "grad_norm": 4.000786276215673, "learning_rate": 9.159363193520814e-06, "loss": 0.3228, "step": 2952 }, { "epoch": 0.21, "grad_norm": 14.327287493163258, "learning_rate": 9.158718589633189e-06, "loss": 1.4397, "step": 2953 }, { "epoch": 0.21, "grad_norm": 5.170700617957154, "learning_rate": 9.158073761398435e-06, "loss": 0.9848, "step": 2954 }, { "epoch": 0.21, "grad_norm": 11.902417199217684, "learning_rate": 9.157428708851341e-06, "loss": 0.7632, "step": 2955 }, { "epoch": 0.21, "grad_norm": 5.509802023540688, "learning_rate": 9.156783432026707e-06, "loss": 0.9451, "step": 2956 }, { "epoch": 0.21, "grad_norm": 3.6670381561548777, "learning_rate": 9.15613793095934e-06, "loss": 0.2904, "step": 2957 }, { "epoch": 0.21, "grad_norm": 4.061009163644482, "learning_rate": 9.155492205684062e-06, "loss": 0.3269, "step": 2958 }, { "epoch": 0.21, "grad_norm": 16.769727677640375, "learning_rate": 9.154846256235712e-06, "loss": 1.7513, "step": 2959 }, { "epoch": 0.21, "grad_norm": 4.547107785921117, "learning_rate": 9.154200082649131e-06, "loss": 0.3524, "step": 2960 }, { "epoch": 0.21, "grad_norm": 15.45304702200982, "learning_rate": 9.15355368495918e-06, "loss": 1.2063, "step": 2961 }, { "epoch": 0.21, "grad_norm": 14.534864915858433, "learning_rate": 9.15290706320073e-06, "loss": 0.739, "step": 2962 }, { "epoch": 0.21, "grad_norm": 16.004556412588066, "learning_rate": 9.152260217408664e-06, "loss": 1.3458, "step": 2963 }, { "epoch": 0.21, "grad_norm": 3.835940699053152, "learning_rate": 9.151613147617878e-06, "loss": 0.1826, "step": 2964 }, { "epoch": 0.21, "grad_norm": 5.012151942394052, "learning_rate": 9.150965853863275e-06, "loss": 0.9662, "step": 2965 }, { "epoch": 0.21, "grad_norm": 3.82173137010373, "learning_rate": 9.150318336179778e-06, "loss": 0.1776, "step": 2966 }, { "epoch": 0.21, "grad_norm": 12.627596142899328, "learning_rate": 9.149670594602315e-06, "loss": 1.4246, "step": 2967 }, { "epoch": 0.21, "grad_norm": 9.147041249914807, "learning_rate": 9.149022629165831e-06, "loss": 0.8576, "step": 2968 }, { "epoch": 0.21, "grad_norm": 4.815217174968651, "learning_rate": 9.148374439905281e-06, "loss": 1.0106, "step": 2969 }, { "epoch": 0.21, "grad_norm": 12.666875123457684, "learning_rate": 9.147726026855635e-06, "loss": 0.9872, "step": 2970 }, { "epoch": 0.21, "grad_norm": 4.066761142999372, "learning_rate": 9.147077390051867e-06, "loss": 0.3099, "step": 2971 }, { "epoch": 0.21, "grad_norm": 17.52407140914102, "learning_rate": 9.146428529528972e-06, "loss": 1.1046, "step": 2972 }, { "epoch": 0.21, "grad_norm": 3.7666080708779686, "learning_rate": 9.145779445321953e-06, "loss": 0.2365, "step": 2973 }, { "epoch": 0.21, "grad_norm": 4.831745659087322, "learning_rate": 9.145130137465826e-06, "loss": 0.2433, "step": 2974 }, { "epoch": 0.21, "grad_norm": 3.8985294918902764, "learning_rate": 9.144480605995618e-06, "loss": 0.2762, "step": 2975 }, { "epoch": 0.21, "grad_norm": 3.761222244867176, "learning_rate": 9.143830850946369e-06, "loss": 0.3295, "step": 2976 }, { "epoch": 0.21, "grad_norm": 12.617109345339271, "learning_rate": 9.143180872353131e-06, "loss": 1.0032, "step": 2977 }, { "epoch": 0.21, "grad_norm": 3.69214621506076, "learning_rate": 9.142530670250967e-06, "loss": 0.1748, "step": 2978 }, { "epoch": 0.21, "grad_norm": 43.377737037424225, "learning_rate": 9.141880244674954e-06, "loss": 1.4075, "step": 2979 }, { "epoch": 0.21, "grad_norm": 4.32516818247225, "learning_rate": 9.141229595660181e-06, "loss": 0.2484, "step": 2980 }, { "epoch": 0.21, "grad_norm": 4.624865745530652, "learning_rate": 9.140578723241745e-06, "loss": 0.3606, "step": 2981 }, { "epoch": 0.21, "grad_norm": 4.030107869364527, "learning_rate": 9.139927627454761e-06, "loss": 0.2201, "step": 2982 }, { "epoch": 0.21, "grad_norm": 5.012580164796776, "learning_rate": 9.139276308334352e-06, "loss": 0.2741, "step": 2983 }, { "epoch": 0.21, "grad_norm": 37.03264482473502, "learning_rate": 9.138624765915653e-06, "loss": 1.1115, "step": 2984 }, { "epoch": 0.21, "grad_norm": 15.041575470961238, "learning_rate": 9.137973000233814e-06, "loss": 1.4379, "step": 2985 }, { "epoch": 0.21, "grad_norm": 10.499808583771935, "learning_rate": 9.137321011323994e-06, "loss": 1.0248, "step": 2986 }, { "epoch": 0.21, "grad_norm": 4.4835075950486205, "learning_rate": 9.136668799221367e-06, "loss": 0.3403, "step": 2987 }, { "epoch": 0.21, "grad_norm": 20.246092330176943, "learning_rate": 9.136016363961117e-06, "loss": 1.0709, "step": 2988 }, { "epoch": 0.21, "grad_norm": 4.456954559820949, "learning_rate": 9.13536370557844e-06, "loss": 0.2976, "step": 2989 }, { "epoch": 0.21, "grad_norm": 5.223066451368159, "learning_rate": 9.13471082410854e-06, "loss": 1.1041, "step": 2990 }, { "epoch": 0.21, "grad_norm": 39.94908019782297, "learning_rate": 9.134057719586647e-06, "loss": 1.0406, "step": 2991 }, { "epoch": 0.21, "grad_norm": 3.6016144200563565, "learning_rate": 9.133404392047986e-06, "loss": 0.2491, "step": 2992 }, { "epoch": 0.21, "grad_norm": 3.1866669555457094, "learning_rate": 9.132750841527804e-06, "loss": 0.2405, "step": 2993 }, { "epoch": 0.21, "grad_norm": 3.3097343063679827, "learning_rate": 9.13209706806136e-06, "loss": 0.1951, "step": 2994 }, { "epoch": 0.21, "grad_norm": 3.2313710574564842, "learning_rate": 9.131443071683917e-06, "loss": 0.2613, "step": 2995 }, { "epoch": 0.21, "grad_norm": 4.6421787233273815, "learning_rate": 9.130788852430759e-06, "loss": 1.0173, "step": 2996 }, { "epoch": 0.21, "grad_norm": 20.138316207444213, "learning_rate": 9.130134410337179e-06, "loss": 1.0243, "step": 2997 }, { "epoch": 0.21, "grad_norm": 5.224835210420804, "learning_rate": 9.12947974543848e-06, "loss": 0.889, "step": 2998 }, { "epoch": 0.22, "grad_norm": 3.848019379852227, "learning_rate": 9.12882485776998e-06, "loss": 0.2698, "step": 2999 }, { "epoch": 0.22, "grad_norm": 29.653744930298682, "learning_rate": 9.128169747367007e-06, "loss": 1.429, "step": 3000 }, { "epoch": 0.22, "grad_norm": 4.425110579350178, "learning_rate": 9.127514414264902e-06, "loss": 0.3498, "step": 3001 }, { "epoch": 0.22, "grad_norm": 5.0647835219109645, "learning_rate": 9.12685885849902e-06, "loss": 1.0074, "step": 3002 }, { "epoch": 0.22, "grad_norm": 3.7990330381146746, "learning_rate": 9.126203080104721e-06, "loss": 0.2891, "step": 3003 }, { "epoch": 0.22, "grad_norm": 19.431703288603227, "learning_rate": 9.125547079117386e-06, "loss": 0.9659, "step": 3004 }, { "epoch": 0.22, "grad_norm": 19.01398996862045, "learning_rate": 9.124890855572402e-06, "loss": 1.0138, "step": 3005 }, { "epoch": 0.22, "grad_norm": 20.590234361051767, "learning_rate": 9.12423440950517e-06, "loss": 0.8698, "step": 3006 }, { "epoch": 0.22, "grad_norm": 3.8457487060621545, "learning_rate": 9.123577740951102e-06, "loss": 0.2709, "step": 3007 }, { "epoch": 0.22, "grad_norm": 9.575490655891313, "learning_rate": 9.122920849945623e-06, "loss": 0.5798, "step": 3008 }, { "epoch": 0.22, "grad_norm": 28.757370700857667, "learning_rate": 9.12226373652417e-06, "loss": 1.5868, "step": 3009 }, { "epoch": 0.22, "grad_norm": 4.256161808471672, "learning_rate": 9.121606400722194e-06, "loss": 0.1311, "step": 3010 }, { "epoch": 0.22, "grad_norm": 23.36404067604309, "learning_rate": 9.120948842575152e-06, "loss": 0.7396, "step": 3011 }, { "epoch": 0.22, "grad_norm": 4.490569736541902, "learning_rate": 9.12029106211852e-06, "loss": 0.3351, "step": 3012 }, { "epoch": 0.22, "grad_norm": 12.214088588404346, "learning_rate": 9.11963305938778e-06, "loss": 0.9057, "step": 3013 }, { "epoch": 0.22, "grad_norm": 4.714453915599862, "learning_rate": 9.118974834418432e-06, "loss": 0.3522, "step": 3014 }, { "epoch": 0.22, "grad_norm": 4.296444638993057, "learning_rate": 9.11831638724598e-06, "loss": 0.278, "step": 3015 }, { "epoch": 0.22, "grad_norm": 3.1321991055010696, "learning_rate": 9.11765771790595e-06, "loss": 0.215, "step": 3016 }, { "epoch": 0.22, "grad_norm": 5.826666758938889, "learning_rate": 9.116998826433872e-06, "loss": 0.7907, "step": 3017 }, { "epoch": 0.22, "grad_norm": 15.21443572730196, "learning_rate": 9.11633971286529e-06, "loss": 0.991, "step": 3018 }, { "epoch": 0.22, "grad_norm": 4.461536093145366, "learning_rate": 9.115680377235762e-06, "loss": 0.3409, "step": 3019 }, { "epoch": 0.22, "grad_norm": 4.718922399126962, "learning_rate": 9.115020819580858e-06, "loss": 0.2893, "step": 3020 }, { "epoch": 0.22, "grad_norm": 11.896643283932837, "learning_rate": 9.114361039936155e-06, "loss": 1.4468, "step": 3021 }, { "epoch": 0.22, "grad_norm": 13.118662400741558, "learning_rate": 9.113701038337248e-06, "loss": 1.6696, "step": 3022 }, { "epoch": 0.22, "grad_norm": 12.985564197357302, "learning_rate": 9.11304081481974e-06, "loss": 1.0117, "step": 3023 }, { "epoch": 0.22, "grad_norm": 28.353152703483033, "learning_rate": 9.112380369419252e-06, "loss": 1.0707, "step": 3024 }, { "epoch": 0.22, "grad_norm": 16.550740240660662, "learning_rate": 9.111719702171407e-06, "loss": 0.7664, "step": 3025 }, { "epoch": 0.22, "grad_norm": 4.012673891711476, "learning_rate": 9.111058813111848e-06, "loss": 0.3028, "step": 3026 }, { "epoch": 0.22, "grad_norm": 3.9097664271997243, "learning_rate": 9.110397702276226e-06, "loss": 0.3521, "step": 3027 }, { "epoch": 0.22, "grad_norm": 4.032451573618989, "learning_rate": 9.109736369700209e-06, "loss": 0.2531, "step": 3028 }, { "epoch": 0.22, "grad_norm": 4.280831794310595, "learning_rate": 9.10907481541947e-06, "loss": 0.2879, "step": 3029 }, { "epoch": 0.22, "grad_norm": 13.475478358318531, "learning_rate": 9.108413039469699e-06, "loss": 1.0749, "step": 3030 }, { "epoch": 0.22, "grad_norm": 4.119478724836763, "learning_rate": 9.107751041886594e-06, "loss": 0.352, "step": 3031 }, { "epoch": 0.22, "grad_norm": 26.159167509098946, "learning_rate": 9.107088822705871e-06, "loss": 1.2778, "step": 3032 }, { "epoch": 0.22, "grad_norm": 5.558048189335842, "learning_rate": 9.106426381963252e-06, "loss": 0.3716, "step": 3033 }, { "epoch": 0.22, "grad_norm": 14.253086200186997, "learning_rate": 9.105763719694472e-06, "loss": 0.8884, "step": 3034 }, { "epoch": 0.22, "grad_norm": 6.474544088875947, "learning_rate": 9.105100835935281e-06, "loss": 1.1663, "step": 3035 }, { "epoch": 0.22, "grad_norm": 34.298664352562135, "learning_rate": 9.104437730721439e-06, "loss": 1.6986, "step": 3036 }, { "epoch": 0.22, "grad_norm": 18.298261332550837, "learning_rate": 9.103774404088717e-06, "loss": 1.491, "step": 3037 }, { "epoch": 0.22, "grad_norm": 10.742823316676247, "learning_rate": 9.1031108560729e-06, "loss": 1.6279, "step": 3038 }, { "epoch": 0.22, "grad_norm": 8.167337016898724, "learning_rate": 9.102447086709783e-06, "loss": 0.6464, "step": 3039 }, { "epoch": 0.22, "grad_norm": 3.211276830678782, "learning_rate": 9.101783096035174e-06, "loss": 0.2177, "step": 3040 }, { "epoch": 0.22, "grad_norm": 23.359798230146446, "learning_rate": 9.101118884084893e-06, "loss": 1.4114, "step": 3041 }, { "epoch": 0.22, "grad_norm": 6.097414906881208, "learning_rate": 9.100454450894774e-06, "loss": 0.3107, "step": 3042 }, { "epoch": 0.22, "grad_norm": 4.565051450318181, "learning_rate": 9.099789796500656e-06, "loss": 0.3183, "step": 3043 }, { "epoch": 0.22, "grad_norm": 3.8594348742633793, "learning_rate": 9.099124920938398e-06, "loss": 0.2511, "step": 3044 }, { "epoch": 0.22, "grad_norm": 3.4363399721550354, "learning_rate": 9.098459824243865e-06, "loss": 0.1809, "step": 3045 }, { "epoch": 0.22, "grad_norm": 13.181276557642667, "learning_rate": 9.09779450645294e-06, "loss": 1.0088, "step": 3046 }, { "epoch": 0.22, "grad_norm": 3.834315091892576, "learning_rate": 9.097128967601512e-06, "loss": 0.2554, "step": 3047 }, { "epoch": 0.22, "grad_norm": 3.4915738932993063, "learning_rate": 9.096463207725485e-06, "loss": 0.2561, "step": 3048 }, { "epoch": 0.22, "grad_norm": 3.814637741320984, "learning_rate": 9.095797226860774e-06, "loss": 0.216, "step": 3049 }, { "epoch": 0.22, "grad_norm": 5.664132913527118, "learning_rate": 9.095131025043305e-06, "loss": 1.0062, "step": 3050 }, { "epoch": 0.22, "grad_norm": 4.501842791894912, "learning_rate": 9.094464602309017e-06, "loss": 0.3428, "step": 3051 }, { "epoch": 0.22, "grad_norm": 33.71595611272709, "learning_rate": 9.093797958693864e-06, "loss": 1.4817, "step": 3052 }, { "epoch": 0.22, "grad_norm": 3.856483545224815, "learning_rate": 9.093131094233807e-06, "loss": 0.2275, "step": 3053 }, { "epoch": 0.22, "grad_norm": 22.963134100212137, "learning_rate": 9.09246400896482e-06, "loss": 1.3246, "step": 3054 }, { "epoch": 0.22, "grad_norm": 4.1212897560152815, "learning_rate": 9.09179670292289e-06, "loss": 0.274, "step": 3055 }, { "epoch": 0.22, "grad_norm": 3.696008218233656, "learning_rate": 9.091129176144018e-06, "loss": 0.2525, "step": 3056 }, { "epoch": 0.22, "grad_norm": 6.50021889744355, "learning_rate": 9.090461428664212e-06, "loss": 1.1587, "step": 3057 }, { "epoch": 0.22, "grad_norm": 8.444153022698698, "learning_rate": 9.089793460519496e-06, "loss": 0.3, "step": 3058 }, { "epoch": 0.22, "grad_norm": 30.57174670079192, "learning_rate": 9.0891252717459e-06, "loss": 1.6473, "step": 3059 }, { "epoch": 0.22, "grad_norm": 9.72553475062995, "learning_rate": 9.088456862379477e-06, "loss": 0.6927, "step": 3060 }, { "epoch": 0.22, "grad_norm": 9.10210901934939, "learning_rate": 9.087788232456282e-06, "loss": 1.0045, "step": 3061 }, { "epoch": 0.22, "grad_norm": 5.459692735294457, "learning_rate": 9.087119382012384e-06, "loss": 0.9006, "step": 3062 }, { "epoch": 0.22, "grad_norm": 4.19765859321397, "learning_rate": 9.086450311083865e-06, "loss": 0.3108, "step": 3063 }, { "epoch": 0.22, "grad_norm": 12.484398235868989, "learning_rate": 9.085781019706821e-06, "loss": 1.2344, "step": 3064 }, { "epoch": 0.22, "grad_norm": 13.684279720718674, "learning_rate": 9.085111507917356e-06, "loss": 1.2535, "step": 3065 }, { "epoch": 0.22, "grad_norm": 4.439319892574831, "learning_rate": 9.084441775751589e-06, "loss": 0.3879, "step": 3066 }, { "epoch": 0.22, "grad_norm": 21.84675286434982, "learning_rate": 9.083771823245648e-06, "loss": 1.1721, "step": 3067 }, { "epoch": 0.22, "grad_norm": 3.7259369384234984, "learning_rate": 9.083101650435676e-06, "loss": 0.1981, "step": 3068 }, { "epoch": 0.22, "grad_norm": 4.093834709435731, "learning_rate": 9.082431257357825e-06, "loss": 0.2528, "step": 3069 }, { "epoch": 0.22, "grad_norm": 5.6197741681424915, "learning_rate": 9.081760644048258e-06, "loss": 0.8186, "step": 3070 }, { "epoch": 0.22, "grad_norm": 3.9563960636307316, "learning_rate": 9.081089810543158e-06, "loss": 0.1927, "step": 3071 }, { "epoch": 0.22, "grad_norm": 11.761776455395054, "learning_rate": 9.080418756878709e-06, "loss": 1.0194, "step": 3072 }, { "epoch": 0.22, "grad_norm": 17.210776969571857, "learning_rate": 9.079747483091112e-06, "loss": 1.3138, "step": 3073 }, { "epoch": 0.22, "grad_norm": 19.302956247441426, "learning_rate": 9.079075989216583e-06, "loss": 1.2023, "step": 3074 }, { "epoch": 0.22, "grad_norm": 4.137604605831166, "learning_rate": 9.078404275291344e-06, "loss": 0.2327, "step": 3075 }, { "epoch": 0.22, "grad_norm": 4.054098564847477, "learning_rate": 9.077732341351632e-06, "loss": 0.237, "step": 3076 }, { "epoch": 0.22, "grad_norm": 3.743179567563054, "learning_rate": 9.077060187433696e-06, "loss": 0.243, "step": 3077 }, { "epoch": 0.22, "grad_norm": 4.968872240213748, "learning_rate": 9.076387813573793e-06, "loss": 0.3082, "step": 3078 }, { "epoch": 0.22, "grad_norm": 5.5041155654147635, "learning_rate": 9.075715219808199e-06, "loss": 1.0025, "step": 3079 }, { "epoch": 0.22, "grad_norm": 3.9712586879371323, "learning_rate": 9.075042406173196e-06, "loss": 0.2235, "step": 3080 }, { "epoch": 0.22, "grad_norm": 4.731897521630615, "learning_rate": 9.07436937270508e-06, "loss": 0.4035, "step": 3081 }, { "epoch": 0.22, "grad_norm": 10.189875058600094, "learning_rate": 9.07369611944016e-06, "loss": 0.2776, "step": 3082 }, { "epoch": 0.22, "grad_norm": 3.9946399362965486, "learning_rate": 9.073022646414753e-06, "loss": 0.2668, "step": 3083 }, { "epoch": 0.22, "grad_norm": 3.549443156475615, "learning_rate": 9.07234895366519e-06, "loss": 0.2008, "step": 3084 }, { "epoch": 0.22, "grad_norm": 19.39446233273275, "learning_rate": 9.071675041227819e-06, "loss": 1.442, "step": 3085 }, { "epoch": 0.22, "grad_norm": 4.220355738617231, "learning_rate": 9.071000909138989e-06, "loss": 0.2659, "step": 3086 }, { "epoch": 0.22, "grad_norm": 12.752489494984523, "learning_rate": 9.070326557435069e-06, "loss": 0.8649, "step": 3087 }, { "epoch": 0.22, "grad_norm": 13.815222229745697, "learning_rate": 9.069651986152442e-06, "loss": 1.2939, "step": 3088 }, { "epoch": 0.22, "grad_norm": 5.3265447359072144, "learning_rate": 9.068977195327491e-06, "loss": 1.0042, "step": 3089 }, { "epoch": 0.22, "grad_norm": 4.040505536035547, "learning_rate": 9.068302184996624e-06, "loss": 0.2388, "step": 3090 }, { "epoch": 0.22, "grad_norm": 4.010194455804641, "learning_rate": 9.067626955196253e-06, "loss": 0.2393, "step": 3091 }, { "epoch": 0.22, "grad_norm": 32.785266831414475, "learning_rate": 9.066951505962804e-06, "loss": 1.566, "step": 3092 }, { "epoch": 0.22, "grad_norm": 4.446171203109338, "learning_rate": 9.066275837332716e-06, "loss": 0.3215, "step": 3093 }, { "epoch": 0.22, "grad_norm": 4.102925369729014, "learning_rate": 9.065599949342439e-06, "loss": 0.1781, "step": 3094 }, { "epoch": 0.22, "grad_norm": 5.120539711577363, "learning_rate": 9.064923842028432e-06, "loss": 0.7928, "step": 3095 }, { "epoch": 0.22, "grad_norm": 16.792130722305068, "learning_rate": 9.064247515427172e-06, "loss": 1.0548, "step": 3096 }, { "epoch": 0.22, "grad_norm": 13.84970109826724, "learning_rate": 9.063570969575142e-06, "loss": 1.1138, "step": 3097 }, { "epoch": 0.22, "grad_norm": 4.698920379712482, "learning_rate": 9.06289420450884e-06, "loss": 0.3354, "step": 3098 }, { "epoch": 0.22, "grad_norm": 20.59437191971976, "learning_rate": 9.062217220264774e-06, "loss": 1.2197, "step": 3099 }, { "epoch": 0.22, "grad_norm": 11.407309864555211, "learning_rate": 9.061540016879467e-06, "loss": 0.9476, "step": 3100 }, { "epoch": 0.22, "grad_norm": 5.767306254027288, "learning_rate": 9.060862594389448e-06, "loss": 1.1811, "step": 3101 }, { "epoch": 0.22, "grad_norm": 32.19838072447533, "learning_rate": 9.060184952831263e-06, "loss": 1.3624, "step": 3102 }, { "epoch": 0.22, "grad_norm": 10.799626893066838, "learning_rate": 9.059507092241469e-06, "loss": 0.8217, "step": 3103 }, { "epoch": 0.22, "grad_norm": 17.391831182620678, "learning_rate": 9.058829012656634e-06, "loss": 1.0407, "step": 3104 }, { "epoch": 0.22, "grad_norm": 12.260096329832566, "learning_rate": 9.058150714113336e-06, "loss": 0.8614, "step": 3105 }, { "epoch": 0.22, "grad_norm": 15.121362803650692, "learning_rate": 9.05747219664817e-06, "loss": 1.1689, "step": 3106 }, { "epoch": 0.22, "grad_norm": 3.8516311864561863, "learning_rate": 9.056793460297734e-06, "loss": 0.284, "step": 3107 }, { "epoch": 0.22, "grad_norm": 9.568370533352361, "learning_rate": 9.05611450509865e-06, "loss": 0.8792, "step": 3108 }, { "epoch": 0.22, "grad_norm": 7.865772146359768, "learning_rate": 9.055435331087541e-06, "loss": 0.9715, "step": 3109 }, { "epoch": 0.22, "grad_norm": 5.474663111363375, "learning_rate": 9.054755938301046e-06, "loss": 0.9838, "step": 3110 }, { "epoch": 0.22, "grad_norm": 3.592969319951981, "learning_rate": 9.054076326775815e-06, "loss": 0.3098, "step": 3111 }, { "epoch": 0.22, "grad_norm": 4.869558824753809, "learning_rate": 9.053396496548514e-06, "loss": 1.1408, "step": 3112 }, { "epoch": 0.22, "grad_norm": 3.093376685333904, "learning_rate": 9.052716447655814e-06, "loss": 0.2312, "step": 3113 }, { "epoch": 0.22, "grad_norm": 14.331614628975265, "learning_rate": 9.0520361801344e-06, "loss": 1.1707, "step": 3114 }, { "epoch": 0.22, "grad_norm": 11.961054939265223, "learning_rate": 9.051355694020975e-06, "loss": 0.7031, "step": 3115 }, { "epoch": 0.22, "grad_norm": 12.675939459183414, "learning_rate": 9.050674989352243e-06, "loss": 0.8222, "step": 3116 }, { "epoch": 0.22, "grad_norm": 6.962778745435146, "learning_rate": 9.04999406616493e-06, "loss": 0.9555, "step": 3117 }, { "epoch": 0.22, "grad_norm": 15.697542382373138, "learning_rate": 9.049312924495765e-06, "loss": 1.4266, "step": 3118 }, { "epoch": 0.22, "grad_norm": 58.57967179437246, "learning_rate": 9.048631564381498e-06, "loss": 0.9795, "step": 3119 }, { "epoch": 0.22, "grad_norm": 4.52166431610517, "learning_rate": 9.047949985858882e-06, "loss": 0.3784, "step": 3120 }, { "epoch": 0.22, "grad_norm": 4.0197692440787725, "learning_rate": 9.047268188964687e-06, "loss": 0.2065, "step": 3121 }, { "epoch": 0.22, "grad_norm": 3.7874915657690504, "learning_rate": 9.046586173735691e-06, "loss": 0.2173, "step": 3122 }, { "epoch": 0.22, "grad_norm": 12.231070927284591, "learning_rate": 9.04590394020869e-06, "loss": 1.1384, "step": 3123 }, { "epoch": 0.22, "grad_norm": 25.655142005206365, "learning_rate": 9.045221488420484e-06, "loss": 0.9696, "step": 3124 }, { "epoch": 0.22, "grad_norm": 4.2238389083629535, "learning_rate": 9.04453881840789e-06, "loss": 0.2975, "step": 3125 }, { "epoch": 0.22, "grad_norm": 20.99081564694988, "learning_rate": 9.043855930207739e-06, "loss": 1.2881, "step": 3126 }, { "epoch": 0.22, "grad_norm": 8.178757531743987, "learning_rate": 9.043172823856867e-06, "loss": 0.9618, "step": 3127 }, { "epoch": 0.22, "grad_norm": 10.696047159631123, "learning_rate": 9.042489499392125e-06, "loss": 0.7559, "step": 3128 }, { "epoch": 0.22, "grad_norm": 3.2700858036360656, "learning_rate": 9.041805956850377e-06, "loss": 0.1879, "step": 3129 }, { "epoch": 0.22, "grad_norm": 3.6279469697839732, "learning_rate": 9.041122196268495e-06, "loss": 0.2117, "step": 3130 }, { "epoch": 0.22, "grad_norm": 3.1615121483098174, "learning_rate": 9.040438217683367e-06, "loss": 0.2034, "step": 3131 }, { "epoch": 0.22, "grad_norm": 4.065363351455208, "learning_rate": 9.039754021131892e-06, "loss": 0.2987, "step": 3132 }, { "epoch": 0.22, "grad_norm": 18.110635097930036, "learning_rate": 9.039069606650978e-06, "loss": 0.9578, "step": 3133 }, { "epoch": 0.22, "grad_norm": 5.7099772942447, "learning_rate": 9.038384974277548e-06, "loss": 1.0443, "step": 3134 }, { "epoch": 0.22, "grad_norm": 5.876761938657452, "learning_rate": 9.037700124048534e-06, "loss": 1.1231, "step": 3135 }, { "epoch": 0.22, "grad_norm": 3.803346115626495, "learning_rate": 9.037015056000885e-06, "loss": 0.2619, "step": 3136 }, { "epoch": 0.22, "grad_norm": 10.561596666590795, "learning_rate": 9.03632977017155e-06, "loss": 0.6873, "step": 3137 }, { "epoch": 0.23, "grad_norm": 3.3851582030509264, "learning_rate": 9.035644266597505e-06, "loss": 0.2281, "step": 3138 }, { "epoch": 0.23, "grad_norm": 17.027063798660667, "learning_rate": 9.034958545315728e-06, "loss": 0.8871, "step": 3139 }, { "epoch": 0.23, "grad_norm": 15.54772923127436, "learning_rate": 9.034272606363209e-06, "loss": 1.1155, "step": 3140 }, { "epoch": 0.23, "grad_norm": 23.35817772630401, "learning_rate": 9.033586449776954e-06, "loss": 1.1525, "step": 3141 }, { "epoch": 0.23, "grad_norm": 7.028276457376609, "learning_rate": 9.032900075593978e-06, "loss": 0.8794, "step": 3142 }, { "epoch": 0.23, "grad_norm": 3.826431563285959, "learning_rate": 9.032213483851308e-06, "loss": 0.2372, "step": 3143 }, { "epoch": 0.23, "grad_norm": 3.5884119846856586, "learning_rate": 9.031526674585985e-06, "loss": 0.1357, "step": 3144 }, { "epoch": 0.23, "grad_norm": 4.987409522007411, "learning_rate": 9.030839647835055e-06, "loss": 0.798, "step": 3145 }, { "epoch": 0.23, "grad_norm": 4.420160763720122, "learning_rate": 9.030152403635586e-06, "loss": 0.1972, "step": 3146 }, { "epoch": 0.23, "grad_norm": 3.883683005711612, "learning_rate": 9.029464942024648e-06, "loss": 0.2564, "step": 3147 }, { "epoch": 0.23, "grad_norm": 3.783066930281787, "learning_rate": 9.02877726303933e-06, "loss": 0.1512, "step": 3148 }, { "epoch": 0.23, "grad_norm": 5.875059037823014, "learning_rate": 9.028089366716729e-06, "loss": 1.1293, "step": 3149 }, { "epoch": 0.23, "grad_norm": 3.7116448040500876, "learning_rate": 9.027401253093953e-06, "loss": 0.171, "step": 3150 }, { "epoch": 0.23, "grad_norm": 5.809950903078356, "learning_rate": 9.026712922208123e-06, "loss": 0.9327, "step": 3151 }, { "epoch": 0.23, "grad_norm": 5.557295335310203, "learning_rate": 9.026024374096374e-06, "loss": 0.9996, "step": 3152 }, { "epoch": 0.23, "grad_norm": 3.208843044214837, "learning_rate": 9.02533560879585e-06, "loss": 0.1303, "step": 3153 }, { "epoch": 0.23, "grad_norm": 4.327370277508283, "learning_rate": 9.024646626343704e-06, "loss": 0.2681, "step": 3154 }, { "epoch": 0.23, "grad_norm": 4.243197504705438, "learning_rate": 9.02395742677711e-06, "loss": 0.2495, "step": 3155 }, { "epoch": 0.23, "grad_norm": 5.378334940327313, "learning_rate": 9.023268010133241e-06, "loss": 0.9804, "step": 3156 }, { "epoch": 0.23, "grad_norm": 5.975571813095269, "learning_rate": 9.022578376449294e-06, "loss": 1.1003, "step": 3157 }, { "epoch": 0.23, "grad_norm": 3.5669777267954843, "learning_rate": 9.02188852576247e-06, "loss": 0.1501, "step": 3158 }, { "epoch": 0.23, "grad_norm": 5.305539965466266, "learning_rate": 9.021198458109982e-06, "loss": 0.9518, "step": 3159 }, { "epoch": 0.23, "grad_norm": 43.529186438520746, "learning_rate": 9.02050817352906e-06, "loss": 1.7188, "step": 3160 }, { "epoch": 0.23, "grad_norm": 4.216888206243557, "learning_rate": 9.01981767205694e-06, "loss": 0.2045, "step": 3161 }, { "epoch": 0.23, "grad_norm": 5.3029952600879815, "learning_rate": 9.019126953730873e-06, "loss": 0.3395, "step": 3162 }, { "epoch": 0.23, "grad_norm": 24.64430003879729, "learning_rate": 9.018436018588118e-06, "loss": 2.2782, "step": 3163 }, { "epoch": 0.23, "grad_norm": 16.696290508905733, "learning_rate": 9.017744866665952e-06, "loss": 1.4988, "step": 3164 }, { "epoch": 0.23, "grad_norm": 5.39418191512009, "learning_rate": 9.017053498001658e-06, "loss": 1.0094, "step": 3165 }, { "epoch": 0.23, "grad_norm": 22.68113836322811, "learning_rate": 9.016361912632533e-06, "loss": 0.9503, "step": 3166 }, { "epoch": 0.23, "grad_norm": 4.922467126180567, "learning_rate": 9.015670110595886e-06, "loss": 0.2836, "step": 3167 }, { "epoch": 0.23, "grad_norm": 5.20051232065624, "learning_rate": 9.014978091929037e-06, "loss": 0.8919, "step": 3168 }, { "epoch": 0.23, "grad_norm": 5.674578695924063, "learning_rate": 9.014285856669316e-06, "loss": 0.8456, "step": 3169 }, { "epoch": 0.23, "grad_norm": 4.727191893504204, "learning_rate": 9.013593404854068e-06, "loss": 0.7891, "step": 3170 }, { "epoch": 0.23, "grad_norm": 4.507623422146453, "learning_rate": 9.012900736520648e-06, "loss": 0.2247, "step": 3171 }, { "epoch": 0.23, "grad_norm": 3.6849291404194955, "learning_rate": 9.012207851706425e-06, "loss": 0.2751, "step": 3172 }, { "epoch": 0.23, "grad_norm": 9.35432848340249, "learning_rate": 9.011514750448775e-06, "loss": 0.8899, "step": 3173 }, { "epoch": 0.23, "grad_norm": 3.5989702781729993, "learning_rate": 9.010821432785087e-06, "loss": 0.3255, "step": 3174 }, { "epoch": 0.23, "grad_norm": 4.695454601175248, "learning_rate": 9.010127898752765e-06, "loss": 0.3807, "step": 3175 }, { "epoch": 0.23, "grad_norm": 3.668853581445928, "learning_rate": 9.009434148389224e-06, "loss": 0.3215, "step": 3176 }, { "epoch": 0.23, "grad_norm": 7.111874825491742, "learning_rate": 9.008740181731885e-06, "loss": 0.7288, "step": 3177 }, { "epoch": 0.23, "grad_norm": 3.43676260714438, "learning_rate": 9.008045998818187e-06, "loss": 0.2727, "step": 3178 }, { "epoch": 0.23, "grad_norm": 3.8158256430684823, "learning_rate": 9.007351599685581e-06, "loss": 0.2691, "step": 3179 }, { "epoch": 0.23, "grad_norm": 4.24767115985275, "learning_rate": 9.006656984371523e-06, "loss": 1.0173, "step": 3180 }, { "epoch": 0.23, "grad_norm": 6.406250715723423, "learning_rate": 9.005962152913485e-06, "loss": 0.3561, "step": 3181 }, { "epoch": 0.23, "grad_norm": 4.352256978486719, "learning_rate": 9.005267105348957e-06, "loss": 0.7716, "step": 3182 }, { "epoch": 0.23, "grad_norm": 2.6985004355047413, "learning_rate": 9.004571841715424e-06, "loss": 0.2464, "step": 3183 }, { "epoch": 0.23, "grad_norm": 3.3433273418015648, "learning_rate": 9.003876362050401e-06, "loss": 0.2515, "step": 3184 }, { "epoch": 0.23, "grad_norm": 19.57064202575789, "learning_rate": 9.003180666391405e-06, "loss": 1.1664, "step": 3185 }, { "epoch": 0.23, "grad_norm": 54.14814153096701, "learning_rate": 9.002484754775964e-06, "loss": 1.0507, "step": 3186 }, { "epoch": 0.23, "grad_norm": 18.304385057260394, "learning_rate": 9.001788627241619e-06, "loss": 2.0468, "step": 3187 }, { "epoch": 0.23, "grad_norm": 4.043368658410733, "learning_rate": 9.001092283825927e-06, "loss": 0.2165, "step": 3188 }, { "epoch": 0.23, "grad_norm": 4.694544877153527, "learning_rate": 9.000395724566451e-06, "loss": 0.308, "step": 3189 }, { "epoch": 0.23, "grad_norm": 5.033621270747784, "learning_rate": 8.999698949500767e-06, "loss": 0.8249, "step": 3190 }, { "epoch": 0.23, "grad_norm": 11.247314131162245, "learning_rate": 8.999001958666464e-06, "loss": 0.7815, "step": 3191 }, { "epoch": 0.23, "grad_norm": 5.600890392658712, "learning_rate": 8.998304752101143e-06, "loss": 1.1326, "step": 3192 }, { "epoch": 0.23, "grad_norm": 3.293199885167078, "learning_rate": 8.997607329842416e-06, "loss": 0.2573, "step": 3193 }, { "epoch": 0.23, "grad_norm": 3.3867039382351796, "learning_rate": 8.996909691927906e-06, "loss": 0.2038, "step": 3194 }, { "epoch": 0.23, "grad_norm": 4.139280144963931, "learning_rate": 8.996211838395246e-06, "loss": 0.2606, "step": 3195 }, { "epoch": 0.23, "grad_norm": 11.31129410482767, "learning_rate": 8.995513769282083e-06, "loss": 1.011, "step": 3196 }, { "epoch": 0.23, "grad_norm": 13.086581084243903, "learning_rate": 8.994815484626076e-06, "loss": 0.9068, "step": 3197 }, { "epoch": 0.23, "grad_norm": 9.859842865839338, "learning_rate": 8.994116984464896e-06, "loss": 0.6927, "step": 3198 }, { "epoch": 0.23, "grad_norm": 14.995608585686869, "learning_rate": 8.993418268836223e-06, "loss": 1.0286, "step": 3199 }, { "epoch": 0.23, "grad_norm": 3.5768286807088536, "learning_rate": 8.99271933777775e-06, "loss": 0.1983, "step": 3200 }, { "epoch": 0.23, "grad_norm": 6.402672104053578, "learning_rate": 8.992020191327181e-06, "loss": 0.9493, "step": 3201 }, { "epoch": 0.23, "grad_norm": 25.36798695020706, "learning_rate": 8.991320829522235e-06, "loss": 1.4844, "step": 3202 }, { "epoch": 0.23, "grad_norm": 13.264134019448553, "learning_rate": 8.990621252400637e-06, "loss": 0.7594, "step": 3203 }, { "epoch": 0.23, "grad_norm": 17.121244204388702, "learning_rate": 8.989921460000128e-06, "loss": 1.8798, "step": 3204 }, { "epoch": 0.23, "grad_norm": 4.4398747768394164, "learning_rate": 8.989221452358462e-06, "loss": 0.2925, "step": 3205 }, { "epoch": 0.23, "grad_norm": 16.85612445866481, "learning_rate": 8.988521229513394e-06, "loss": 0.8841, "step": 3206 }, { "epoch": 0.23, "grad_norm": 5.1827110049379925, "learning_rate": 8.987820791502704e-06, "loss": 1.0237, "step": 3207 }, { "epoch": 0.23, "grad_norm": 4.545782472603784, "learning_rate": 8.98712013836418e-06, "loss": 0.4307, "step": 3208 }, { "epoch": 0.23, "grad_norm": 3.828029303595765, "learning_rate": 8.986419270135613e-06, "loss": 0.1833, "step": 3209 }, { "epoch": 0.23, "grad_norm": 4.768873879222793, "learning_rate": 8.985718186854818e-06, "loss": 1.0566, "step": 3210 }, { "epoch": 0.23, "grad_norm": 4.245858831854081, "learning_rate": 8.985016888559612e-06, "loss": 0.2397, "step": 3211 }, { "epoch": 0.23, "grad_norm": 3.48187832645726, "learning_rate": 8.984315375287831e-06, "loss": 0.2878, "step": 3212 }, { "epoch": 0.23, "grad_norm": 13.827700006437912, "learning_rate": 8.983613647077315e-06, "loss": 0.9494, "step": 3213 }, { "epoch": 0.23, "grad_norm": 3.96464398241598, "learning_rate": 8.982911703965923e-06, "loss": 0.3857, "step": 3214 }, { "epoch": 0.23, "grad_norm": 4.306325026331212, "learning_rate": 8.98220954599152e-06, "loss": 0.3646, "step": 3215 }, { "epoch": 0.23, "grad_norm": 14.715218268830547, "learning_rate": 8.981507173191986e-06, "loss": 1.0095, "step": 3216 }, { "epoch": 0.23, "grad_norm": 4.584236283075338, "learning_rate": 8.980804585605211e-06, "loss": 0.3872, "step": 3217 }, { "epoch": 0.23, "grad_norm": 22.13854221510169, "learning_rate": 8.980101783269095e-06, "loss": 1.5774, "step": 3218 }, { "epoch": 0.23, "grad_norm": 3.594880517075353, "learning_rate": 8.979398766221556e-06, "loss": 0.2481, "step": 3219 }, { "epoch": 0.23, "grad_norm": 3.39196108316887, "learning_rate": 8.978695534500515e-06, "loss": 0.21, "step": 3220 }, { "epoch": 0.23, "grad_norm": 3.741358152504647, "learning_rate": 8.97799208814391e-06, "loss": 0.2392, "step": 3221 }, { "epoch": 0.23, "grad_norm": 4.859690190009779, "learning_rate": 8.977288427189688e-06, "loss": 0.8581, "step": 3222 }, { "epoch": 0.23, "grad_norm": 3.871398351365207, "learning_rate": 8.976584551675812e-06, "loss": 0.2332, "step": 3223 }, { "epoch": 0.23, "grad_norm": 22.001169674163847, "learning_rate": 8.975880461640251e-06, "loss": 1.2067, "step": 3224 }, { "epoch": 0.23, "grad_norm": 14.79444196345308, "learning_rate": 8.975176157120988e-06, "loss": 1.516, "step": 3225 }, { "epoch": 0.23, "grad_norm": 4.075331383278395, "learning_rate": 8.97447163815602e-06, "loss": 0.2203, "step": 3226 }, { "epoch": 0.23, "grad_norm": 15.752745441006404, "learning_rate": 8.97376690478335e-06, "loss": 0.9652, "step": 3227 }, { "epoch": 0.23, "grad_norm": 4.6645878761591435, "learning_rate": 8.973061957040998e-06, "loss": 0.9344, "step": 3228 }, { "epoch": 0.23, "grad_norm": 13.497711430973776, "learning_rate": 8.972356794966991e-06, "loss": 0.8275, "step": 3229 }, { "epoch": 0.23, "grad_norm": 2.97757416790762, "learning_rate": 8.971651418599371e-06, "loss": 0.153, "step": 3230 }, { "epoch": 0.23, "grad_norm": 16.409836657648064, "learning_rate": 8.970945827976192e-06, "loss": 0.8913, "step": 3231 }, { "epoch": 0.23, "grad_norm": 4.228064666774703, "learning_rate": 8.970240023135514e-06, "loss": 0.3289, "step": 3232 }, { "epoch": 0.23, "grad_norm": 10.094375166595631, "learning_rate": 8.969534004115419e-06, "loss": 0.701, "step": 3233 }, { "epoch": 0.23, "grad_norm": 3.7280270028489206, "learning_rate": 8.968827770953986e-06, "loss": 0.2207, "step": 3234 }, { "epoch": 0.23, "grad_norm": 4.9777957202395315, "learning_rate": 8.96812132368932e-06, "loss": 0.9646, "step": 3235 }, { "epoch": 0.23, "grad_norm": 10.012558373622058, "learning_rate": 8.967414662359529e-06, "loss": 1.2025, "step": 3236 }, { "epoch": 0.23, "grad_norm": 3.694448021867293, "learning_rate": 8.966707787002732e-06, "loss": 0.2597, "step": 3237 }, { "epoch": 0.23, "grad_norm": 3.489653293790695, "learning_rate": 8.966000697657064e-06, "loss": 0.2327, "step": 3238 }, { "epoch": 0.23, "grad_norm": 3.546111338781134, "learning_rate": 8.965293394360674e-06, "loss": 0.309, "step": 3239 }, { "epoch": 0.23, "grad_norm": 3.57327411183886, "learning_rate": 8.964585877151714e-06, "loss": 0.2, "step": 3240 }, { "epoch": 0.23, "grad_norm": 16.518619347220298, "learning_rate": 8.96387814606835e-06, "loss": 1.5625, "step": 3241 }, { "epoch": 0.23, "grad_norm": 13.815037549863552, "learning_rate": 8.963170201148767e-06, "loss": 1.26, "step": 3242 }, { "epoch": 0.23, "grad_norm": 12.067388491824602, "learning_rate": 8.962462042431152e-06, "loss": 0.9526, "step": 3243 }, { "epoch": 0.23, "grad_norm": 3.6865645368088087, "learning_rate": 8.96175366995371e-06, "loss": 0.2924, "step": 3244 }, { "epoch": 0.23, "grad_norm": 3.2304684362621034, "learning_rate": 8.961045083754651e-06, "loss": 0.2222, "step": 3245 }, { "epoch": 0.23, "grad_norm": 3.8983520406764063, "learning_rate": 8.960336283872204e-06, "loss": 0.2371, "step": 3246 }, { "epoch": 0.23, "grad_norm": 9.481121724657722, "learning_rate": 8.959627270344606e-06, "loss": 0.6339, "step": 3247 }, { "epoch": 0.23, "grad_norm": 19.034638750901994, "learning_rate": 8.958918043210106e-06, "loss": 1.2102, "step": 3248 }, { "epoch": 0.23, "grad_norm": 5.995096588757508, "learning_rate": 8.95820860250696e-06, "loss": 0.9068, "step": 3249 }, { "epoch": 0.23, "grad_norm": 14.982922545895974, "learning_rate": 8.957498948273444e-06, "loss": 0.8803, "step": 3250 }, { "epoch": 0.23, "grad_norm": 20.46865739727977, "learning_rate": 8.956789080547838e-06, "loss": 1.0562, "step": 3251 }, { "epoch": 0.23, "grad_norm": 4.204918749216148, "learning_rate": 8.95607899936844e-06, "loss": 0.2661, "step": 3252 }, { "epoch": 0.23, "grad_norm": 42.09817072066939, "learning_rate": 8.955368704773555e-06, "loss": 0.7878, "step": 3253 }, { "epoch": 0.23, "grad_norm": 5.503970825489459, "learning_rate": 8.9546581968015e-06, "loss": 0.3611, "step": 3254 }, { "epoch": 0.23, "grad_norm": 19.333234702164734, "learning_rate": 8.953947475490603e-06, "loss": 1.0428, "step": 3255 }, { "epoch": 0.23, "grad_norm": 4.246580971641582, "learning_rate": 8.953236540879208e-06, "loss": 0.3026, "step": 3256 }, { "epoch": 0.23, "grad_norm": 13.804238669965516, "learning_rate": 8.952525393005664e-06, "loss": 0.9583, "step": 3257 }, { "epoch": 0.23, "grad_norm": 3.3802613694023935, "learning_rate": 8.951814031908337e-06, "loss": 0.2122, "step": 3258 }, { "epoch": 0.23, "grad_norm": 40.60443055597151, "learning_rate": 8.951102457625603e-06, "loss": 1.621, "step": 3259 }, { "epoch": 0.23, "grad_norm": 3.4855189699978495, "learning_rate": 8.950390670195843e-06, "loss": 0.2108, "step": 3260 }, { "epoch": 0.23, "grad_norm": 11.236477056481496, "learning_rate": 8.949678669657462e-06, "loss": 1.2656, "step": 3261 }, { "epoch": 0.23, "grad_norm": 4.314676649315473, "learning_rate": 8.948966456048868e-06, "loss": 0.2263, "step": 3262 }, { "epoch": 0.23, "grad_norm": 16.503429753305475, "learning_rate": 8.94825402940848e-06, "loss": 1.1781, "step": 3263 }, { "epoch": 0.23, "grad_norm": 5.582379208232339, "learning_rate": 8.947541389774732e-06, "loss": 0.8398, "step": 3264 }, { "epoch": 0.23, "grad_norm": 6.205494406210702, "learning_rate": 8.946828537186068e-06, "loss": 0.9572, "step": 3265 }, { "epoch": 0.23, "grad_norm": 5.055097566932504, "learning_rate": 8.946115471680946e-06, "loss": 0.3267, "step": 3266 }, { "epoch": 0.23, "grad_norm": 18.53912094633665, "learning_rate": 8.94540219329783e-06, "loss": 1.4947, "step": 3267 }, { "epoch": 0.23, "grad_norm": 4.814424669256443, "learning_rate": 8.9446887020752e-06, "loss": 0.349, "step": 3268 }, { "epoch": 0.23, "grad_norm": 24.69275704052415, "learning_rate": 8.943974998051544e-06, "loss": 1.7226, "step": 3269 }, { "epoch": 0.23, "grad_norm": 4.842840069851701, "learning_rate": 8.943261081265367e-06, "loss": 0.2827, "step": 3270 }, { "epoch": 0.23, "grad_norm": 9.66352134653314, "learning_rate": 8.942546951755182e-06, "loss": 0.638, "step": 3271 }, { "epoch": 0.23, "grad_norm": 10.803541148703541, "learning_rate": 8.94183260955951e-06, "loss": 0.9006, "step": 3272 }, { "epoch": 0.23, "grad_norm": 4.73453161641732, "learning_rate": 8.941118054716891e-06, "loss": 1.0728, "step": 3273 }, { "epoch": 0.23, "grad_norm": 22.600702517524812, "learning_rate": 8.94040328726587e-06, "loss": 1.2363, "step": 3274 }, { "epoch": 0.23, "grad_norm": 7.560888857312164, "learning_rate": 8.939688307245009e-06, "loss": 0.2001, "step": 3275 }, { "epoch": 0.23, "grad_norm": 13.345847913899707, "learning_rate": 8.938973114692874e-06, "loss": 1.3959, "step": 3276 }, { "epoch": 0.23, "grad_norm": 3.8874941957563283, "learning_rate": 8.93825770964805e-06, "loss": 0.3069, "step": 3277 }, { "epoch": 0.24, "grad_norm": 5.173470946548737, "learning_rate": 8.93754209214913e-06, "loss": 1.0575, "step": 3278 }, { "epoch": 0.24, "grad_norm": 3.840058325221302, "learning_rate": 8.936826262234717e-06, "loss": 0.1471, "step": 3279 }, { "epoch": 0.24, "grad_norm": 17.165192172873052, "learning_rate": 8.936110219943431e-06, "loss": 1.8939, "step": 3280 }, { "epoch": 0.24, "grad_norm": 4.402413814723578, "learning_rate": 8.935393965313897e-06, "loss": 0.2682, "step": 3281 }, { "epoch": 0.24, "grad_norm": 6.29324253468793, "learning_rate": 8.934677498384753e-06, "loss": 0.3241, "step": 3282 }, { "epoch": 0.24, "grad_norm": 4.50827297239662, "learning_rate": 8.933960819194654e-06, "loss": 0.2768, "step": 3283 }, { "epoch": 0.24, "grad_norm": 4.002170685611326, "learning_rate": 8.933243927782259e-06, "loss": 0.2377, "step": 3284 }, { "epoch": 0.24, "grad_norm": 4.410298398042183, "learning_rate": 8.93252682418624e-06, "loss": 0.2577, "step": 3285 }, { "epoch": 0.24, "grad_norm": 4.227478692932564, "learning_rate": 8.931809508445287e-06, "loss": 0.2534, "step": 3286 }, { "epoch": 0.24, "grad_norm": 5.694318809144289, "learning_rate": 8.931091980598093e-06, "loss": 1.1391, "step": 3287 }, { "epoch": 0.24, "grad_norm": 4.555679129540708, "learning_rate": 8.930374240683366e-06, "loss": 0.2674, "step": 3288 }, { "epoch": 0.24, "grad_norm": 5.8460469528303936, "learning_rate": 8.929656288739828e-06, "loss": 0.3369, "step": 3289 }, { "epoch": 0.24, "grad_norm": 12.83072622972392, "learning_rate": 8.928938124806206e-06, "loss": 1.1787, "step": 3290 }, { "epoch": 0.24, "grad_norm": 5.8217681370585375, "learning_rate": 8.928219748921244e-06, "loss": 0.3025, "step": 3291 }, { "epoch": 0.24, "grad_norm": 3.9973984250735772, "learning_rate": 8.927501161123695e-06, "loss": 0.4121, "step": 3292 }, { "epoch": 0.24, "grad_norm": 13.465943301286321, "learning_rate": 8.926782361452326e-06, "loss": 0.9546, "step": 3293 }, { "epoch": 0.24, "grad_norm": 5.0409249461113195, "learning_rate": 8.926063349945912e-06, "loss": 1.0463, "step": 3294 }, { "epoch": 0.24, "grad_norm": 16.081162871972552, "learning_rate": 8.925344126643241e-06, "loss": 1.43, "step": 3295 }, { "epoch": 0.24, "grad_norm": 4.116485970366455, "learning_rate": 8.924624691583114e-06, "loss": 0.2286, "step": 3296 }, { "epoch": 0.24, "grad_norm": 4.261192807978535, "learning_rate": 8.923905044804338e-06, "loss": 0.3023, "step": 3297 }, { "epoch": 0.24, "grad_norm": 24.165513380020727, "learning_rate": 8.92318518634574e-06, "loss": 1.4301, "step": 3298 }, { "epoch": 0.24, "grad_norm": 12.593005066410992, "learning_rate": 8.92246511624615e-06, "loss": 1.0466, "step": 3299 }, { "epoch": 0.24, "grad_norm": 7.647323214617221, "learning_rate": 8.921744834544415e-06, "loss": 0.7019, "step": 3300 }, { "epoch": 0.24, "grad_norm": 3.455943012494048, "learning_rate": 8.92102434127939e-06, "loss": 0.2886, "step": 3301 }, { "epoch": 0.24, "grad_norm": 4.189737968803568, "learning_rate": 8.920303636489943e-06, "loss": 0.2773, "step": 3302 }, { "epoch": 0.24, "grad_norm": 2.4598497435571933, "learning_rate": 8.919582720214955e-06, "loss": 0.1815, "step": 3303 }, { "epoch": 0.24, "grad_norm": 3.016252184968179, "learning_rate": 8.918861592493316e-06, "loss": 0.2075, "step": 3304 }, { "epoch": 0.24, "grad_norm": 3.4172361825505226, "learning_rate": 8.918140253363927e-06, "loss": 0.1961, "step": 3305 }, { "epoch": 0.24, "grad_norm": 15.011921648159811, "learning_rate": 8.917418702865701e-06, "loss": 1.8914, "step": 3306 }, { "epoch": 0.24, "grad_norm": 4.588787703711708, "learning_rate": 8.916696941037566e-06, "loss": 0.2979, "step": 3307 }, { "epoch": 0.24, "grad_norm": 4.542853606507522, "learning_rate": 8.915974967918457e-06, "loss": 0.2184, "step": 3308 }, { "epoch": 0.24, "grad_norm": 4.707154054758463, "learning_rate": 8.915252783547321e-06, "loss": 0.2814, "step": 3309 }, { "epoch": 0.24, "grad_norm": 3.9020202870265255, "learning_rate": 8.914530387963116e-06, "loss": 0.1664, "step": 3310 }, { "epoch": 0.24, "grad_norm": 6.924808544675785, "learning_rate": 8.913807781204816e-06, "loss": 0.936, "step": 3311 }, { "epoch": 0.24, "grad_norm": 5.3699249126237065, "learning_rate": 8.9130849633114e-06, "loss": 1.0825, "step": 3312 }, { "epoch": 0.24, "grad_norm": 4.75221026718126, "learning_rate": 8.91236193432186e-06, "loss": 0.2597, "step": 3313 }, { "epoch": 0.24, "grad_norm": 4.706702513079442, "learning_rate": 8.911638694275207e-06, "loss": 0.2458, "step": 3314 }, { "epoch": 0.24, "grad_norm": 4.430527941666228, "learning_rate": 8.910915243210448e-06, "loss": 0.2252, "step": 3315 }, { "epoch": 0.24, "grad_norm": 16.659350802284695, "learning_rate": 8.910191581166619e-06, "loss": 1.2278, "step": 3316 }, { "epoch": 0.24, "grad_norm": 16.71933944457013, "learning_rate": 8.909467708182753e-06, "loss": 0.9846, "step": 3317 }, { "epoch": 0.24, "grad_norm": 18.31592245929859, "learning_rate": 8.908743624297901e-06, "loss": 1.3496, "step": 3318 }, { "epoch": 0.24, "grad_norm": 4.6023731779291595, "learning_rate": 8.90801932955113e-06, "loss": 0.2194, "step": 3319 }, { "epoch": 0.24, "grad_norm": 4.058519871502417, "learning_rate": 8.907294823981507e-06, "loss": 0.2877, "step": 3320 }, { "epoch": 0.24, "grad_norm": 11.656013911420812, "learning_rate": 8.906570107628116e-06, "loss": 0.6621, "step": 3321 }, { "epoch": 0.24, "grad_norm": 12.89410559939675, "learning_rate": 8.905845180530057e-06, "loss": 0.9176, "step": 3322 }, { "epoch": 0.24, "grad_norm": 9.265604881012768, "learning_rate": 8.905120042726434e-06, "loss": 1.1479, "step": 3323 }, { "epoch": 0.24, "grad_norm": 3.1897199480161023, "learning_rate": 8.904394694256367e-06, "loss": 0.1772, "step": 3324 }, { "epoch": 0.24, "grad_norm": 4.242914915533952, "learning_rate": 8.903669135158985e-06, "loss": 0.2827, "step": 3325 }, { "epoch": 0.24, "grad_norm": 3.865718708682746, "learning_rate": 8.90294336547343e-06, "loss": 0.2326, "step": 3326 }, { "epoch": 0.24, "grad_norm": 4.09267723959932, "learning_rate": 8.902217385238852e-06, "loss": 0.3445, "step": 3327 }, { "epoch": 0.24, "grad_norm": 13.656643657030019, "learning_rate": 8.901491194494418e-06, "loss": 1.4921, "step": 3328 }, { "epoch": 0.24, "grad_norm": 3.937546462632868, "learning_rate": 8.900764793279302e-06, "loss": 0.2876, "step": 3329 }, { "epoch": 0.24, "grad_norm": 4.646455555109227, "learning_rate": 8.90003818163269e-06, "loss": 0.291, "step": 3330 }, { "epoch": 0.24, "grad_norm": 4.873853699737183, "learning_rate": 8.899311359593784e-06, "loss": 0.8268, "step": 3331 }, { "epoch": 0.24, "grad_norm": 11.935566210829982, "learning_rate": 8.898584327201785e-06, "loss": 0.9784, "step": 3332 }, { "epoch": 0.24, "grad_norm": 8.2642597140022, "learning_rate": 8.897857084495921e-06, "loss": 0.8022, "step": 3333 }, { "epoch": 0.24, "grad_norm": 4.92672848018341, "learning_rate": 8.897129631515422e-06, "loss": 0.9392, "step": 3334 }, { "epoch": 0.24, "grad_norm": 4.150338926647549, "learning_rate": 8.896401968299528e-06, "loss": 1.0659, "step": 3335 }, { "epoch": 0.24, "grad_norm": 7.085489007725124, "learning_rate": 8.895674094887498e-06, "loss": 0.562, "step": 3336 }, { "epoch": 0.24, "grad_norm": 4.598071902332324, "learning_rate": 8.894946011318597e-06, "loss": 0.3824, "step": 3337 }, { "epoch": 0.24, "grad_norm": 16.88269653540521, "learning_rate": 8.894217717632104e-06, "loss": 0.9771, "step": 3338 }, { "epoch": 0.24, "grad_norm": 3.5304448947657967, "learning_rate": 8.893489213867302e-06, "loss": 0.1986, "step": 3339 }, { "epoch": 0.24, "grad_norm": 4.406368303444756, "learning_rate": 8.892760500063495e-06, "loss": 0.311, "step": 3340 }, { "epoch": 0.24, "grad_norm": 4.805567996780148, "learning_rate": 8.892031576259996e-06, "loss": 0.2606, "step": 3341 }, { "epoch": 0.24, "grad_norm": 11.70720338783415, "learning_rate": 8.891302442496123e-06, "loss": 0.5899, "step": 3342 }, { "epoch": 0.24, "grad_norm": 32.416345071126045, "learning_rate": 8.890573098811213e-06, "loss": 1.7609, "step": 3343 }, { "epoch": 0.24, "grad_norm": 4.831361557068844, "learning_rate": 8.889843545244614e-06, "loss": 0.3104, "step": 3344 }, { "epoch": 0.24, "grad_norm": 4.889735395263048, "learning_rate": 8.889113781835676e-06, "loss": 0.1874, "step": 3345 }, { "epoch": 0.24, "grad_norm": 5.330557662714733, "learning_rate": 8.888383808623772e-06, "loss": 0.2891, "step": 3346 }, { "epoch": 0.24, "grad_norm": 4.463374451114171, "learning_rate": 8.88765362564828e-06, "loss": 0.3189, "step": 3347 }, { "epoch": 0.24, "grad_norm": 30.827072106581344, "learning_rate": 8.88692323294859e-06, "loss": 1.0162, "step": 3348 }, { "epoch": 0.24, "grad_norm": 14.499223919261158, "learning_rate": 8.886192630564106e-06, "loss": 1.2155, "step": 3349 }, { "epoch": 0.24, "grad_norm": 3.682654699813397, "learning_rate": 8.885461818534237e-06, "loss": 0.2269, "step": 3350 }, { "epoch": 0.24, "grad_norm": 14.993381964943161, "learning_rate": 8.884730796898412e-06, "loss": 0.8332, "step": 3351 }, { "epoch": 0.24, "grad_norm": 4.8556172730122835, "learning_rate": 8.883999565696064e-06, "loss": 0.8771, "step": 3352 }, { "epoch": 0.24, "grad_norm": 15.169403332468026, "learning_rate": 8.883268124966643e-06, "loss": 1.0991, "step": 3353 }, { "epoch": 0.24, "grad_norm": 4.584608648383221, "learning_rate": 8.882536474749604e-06, "loss": 0.2111, "step": 3354 }, { "epoch": 0.24, "grad_norm": 11.01403483288694, "learning_rate": 8.881804615084418e-06, "loss": 0.9642, "step": 3355 }, { "epoch": 0.24, "grad_norm": 3.3593309081581646, "learning_rate": 8.881072546010567e-06, "loss": 0.2558, "step": 3356 }, { "epoch": 0.24, "grad_norm": 80.1944191692998, "learning_rate": 8.880340267567543e-06, "loss": 1.4671, "step": 3357 }, { "epoch": 0.24, "grad_norm": 5.769392211167972, "learning_rate": 8.87960777979485e-06, "loss": 1.1207, "step": 3358 }, { "epoch": 0.24, "grad_norm": 9.931734056935342, "learning_rate": 8.878875082732e-06, "loss": 0.2458, "step": 3359 }, { "epoch": 0.24, "grad_norm": 10.558703000074152, "learning_rate": 8.878142176418525e-06, "loss": 0.8969, "step": 3360 }, { "epoch": 0.24, "grad_norm": 12.954619958218164, "learning_rate": 8.877409060893956e-06, "loss": 0.8949, "step": 3361 }, { "epoch": 0.24, "grad_norm": 26.71338533198731, "learning_rate": 8.876675736197847e-06, "loss": 0.8531, "step": 3362 }, { "epoch": 0.24, "grad_norm": 5.071718808072766, "learning_rate": 8.875942202369757e-06, "loss": 0.3154, "step": 3363 }, { "epoch": 0.24, "grad_norm": 4.064793329818831, "learning_rate": 8.875208459449254e-06, "loss": 0.1912, "step": 3364 }, { "epoch": 0.24, "grad_norm": 14.947984185663072, "learning_rate": 8.874474507475923e-06, "loss": 1.4466, "step": 3365 }, { "epoch": 0.24, "grad_norm": 5.825040837012071, "learning_rate": 8.87374034648936e-06, "loss": 1.0288, "step": 3366 }, { "epoch": 0.24, "grad_norm": 5.557599508995844, "learning_rate": 8.873005976529167e-06, "loss": 1.0789, "step": 3367 }, { "epoch": 0.24, "grad_norm": 5.354210557688084, "learning_rate": 8.872271397634962e-06, "loss": 0.2729, "step": 3368 }, { "epoch": 0.24, "grad_norm": 29.43220939169348, "learning_rate": 8.871536609846371e-06, "loss": 1.3246, "step": 3369 }, { "epoch": 0.24, "grad_norm": 4.824709203415624, "learning_rate": 8.870801613203037e-06, "loss": 0.4326, "step": 3370 }, { "epoch": 0.24, "grad_norm": 3.8649065836806624, "learning_rate": 8.870066407744607e-06, "loss": 0.2467, "step": 3371 }, { "epoch": 0.24, "grad_norm": 3.6660969640082115, "learning_rate": 8.869330993510743e-06, "loss": 0.2259, "step": 3372 }, { "epoch": 0.24, "grad_norm": 3.8122128698485356, "learning_rate": 8.868595370541117e-06, "loss": 0.1928, "step": 3373 }, { "epoch": 0.24, "grad_norm": 3.7955019213428463, "learning_rate": 8.867859538875417e-06, "loss": 0.2178, "step": 3374 }, { "epoch": 0.24, "grad_norm": 19.204408829328823, "learning_rate": 8.867123498553333e-06, "loss": 1.0919, "step": 3375 }, { "epoch": 0.24, "grad_norm": 3.9908628230373555, "learning_rate": 8.866387249614573e-06, "loss": 0.2226, "step": 3376 }, { "epoch": 0.24, "grad_norm": 7.892311060351732, "learning_rate": 8.86565079209886e-06, "loss": 0.5873, "step": 3377 }, { "epoch": 0.24, "grad_norm": 4.347883995595635, "learning_rate": 8.864914126045916e-06, "loss": 0.3854, "step": 3378 }, { "epoch": 0.24, "grad_norm": 14.645928865340059, "learning_rate": 8.864177251495484e-06, "loss": 1.4958, "step": 3379 }, { "epoch": 0.24, "grad_norm": 17.54482347228709, "learning_rate": 8.863440168487318e-06, "loss": 1.6096, "step": 3380 }, { "epoch": 0.24, "grad_norm": 4.413500859577843, "learning_rate": 8.862702877061178e-06, "loss": 0.3402, "step": 3381 }, { "epoch": 0.24, "grad_norm": 13.621708774954312, "learning_rate": 8.86196537725684e-06, "loss": 0.8138, "step": 3382 }, { "epoch": 0.24, "grad_norm": 4.68047066603562, "learning_rate": 8.861227669114085e-06, "loss": 0.9381, "step": 3383 }, { "epoch": 0.24, "grad_norm": 4.918862247029942, "learning_rate": 8.860489752672715e-06, "loss": 0.9498, "step": 3384 }, { "epoch": 0.24, "grad_norm": 13.494451556001726, "learning_rate": 8.859751627972536e-06, "loss": 0.8214, "step": 3385 }, { "epoch": 0.24, "grad_norm": 3.740434009224841, "learning_rate": 8.859013295053366e-06, "loss": 0.2295, "step": 3386 }, { "epoch": 0.24, "grad_norm": 9.062197369882597, "learning_rate": 8.858274753955036e-06, "loss": 0.8192, "step": 3387 }, { "epoch": 0.24, "grad_norm": 28.060530136808936, "learning_rate": 8.857536004717389e-06, "loss": 1.4789, "step": 3388 }, { "epoch": 0.24, "grad_norm": 5.565783675386464, "learning_rate": 8.856797047380275e-06, "loss": 1.1023, "step": 3389 }, { "epoch": 0.24, "grad_norm": 27.916631136673463, "learning_rate": 8.856057881983557e-06, "loss": 1.462, "step": 3390 }, { "epoch": 0.24, "grad_norm": 22.567198792685474, "learning_rate": 8.855318508567115e-06, "loss": 1.0962, "step": 3391 }, { "epoch": 0.24, "grad_norm": 4.018987422295366, "learning_rate": 8.854578927170832e-06, "loss": 0.2074, "step": 3392 }, { "epoch": 0.24, "grad_norm": 5.258243640675407, "learning_rate": 8.853839137834605e-06, "loss": 1.013, "step": 3393 }, { "epoch": 0.24, "grad_norm": 19.35868701206039, "learning_rate": 8.853099140598344e-06, "loss": 1.377, "step": 3394 }, { "epoch": 0.24, "grad_norm": 4.954873389907151, "learning_rate": 8.85235893550197e-06, "loss": 0.9638, "step": 3395 }, { "epoch": 0.24, "grad_norm": 4.32532550217047, "learning_rate": 8.851618522585414e-06, "loss": 0.2816, "step": 3396 }, { "epoch": 0.24, "grad_norm": 4.439120512043488, "learning_rate": 8.850877901888618e-06, "loss": 0.2256, "step": 3397 }, { "epoch": 0.24, "grad_norm": 3.5598788815660987, "learning_rate": 8.850137073451534e-06, "loss": 0.2586, "step": 3398 }, { "epoch": 0.24, "grad_norm": 4.915392005321474, "learning_rate": 8.849396037314128e-06, "loss": 0.7781, "step": 3399 }, { "epoch": 0.24, "grad_norm": 3.9981453726102645, "learning_rate": 8.848654793516379e-06, "loss": 0.2339, "step": 3400 }, { "epoch": 0.24, "grad_norm": 3.463939852247123, "learning_rate": 8.84791334209827e-06, "loss": 0.2068, "step": 3401 }, { "epoch": 0.24, "grad_norm": 5.533700657207025, "learning_rate": 8.847171683099803e-06, "loss": 1.0074, "step": 3402 }, { "epoch": 0.24, "grad_norm": 5.716876126972258, "learning_rate": 8.846429816560984e-06, "loss": 0.9792, "step": 3403 }, { "epoch": 0.24, "grad_norm": 10.58128244181294, "learning_rate": 8.845687742521836e-06, "loss": 0.7943, "step": 3404 }, { "epoch": 0.24, "grad_norm": 10.183095953210575, "learning_rate": 8.844945461022393e-06, "loss": 0.7155, "step": 3405 }, { "epoch": 0.24, "grad_norm": 3.9830814303853566, "learning_rate": 8.844202972102696e-06, "loss": 0.2859, "step": 3406 }, { "epoch": 0.24, "grad_norm": 5.36773487636452, "learning_rate": 8.843460275802798e-06, "loss": 1.0362, "step": 3407 }, { "epoch": 0.24, "grad_norm": 9.887044153421785, "learning_rate": 8.842717372162768e-06, "loss": 0.7925, "step": 3408 }, { "epoch": 0.24, "grad_norm": 3.690376236485158, "learning_rate": 8.84197426122268e-06, "loss": 0.2752, "step": 3409 }, { "epoch": 0.24, "grad_norm": 3.7501335961472626, "learning_rate": 8.841230943022624e-06, "loss": 0.2178, "step": 3410 }, { "epoch": 0.24, "grad_norm": 3.535725710243205, "learning_rate": 8.840487417602699e-06, "loss": 0.1809, "step": 3411 }, { "epoch": 0.24, "grad_norm": 20.807817619552132, "learning_rate": 8.839743685003014e-06, "loss": 1.1337, "step": 3412 }, { "epoch": 0.24, "grad_norm": 4.97996057286586, "learning_rate": 8.83899974526369e-06, "loss": 0.9914, "step": 3413 }, { "epoch": 0.24, "grad_norm": 6.205351281892952, "learning_rate": 8.838255598424864e-06, "loss": 1.0934, "step": 3414 }, { "epoch": 0.24, "grad_norm": 11.923387528397786, "learning_rate": 8.837511244526675e-06, "loss": 1.3191, "step": 3415 }, { "epoch": 0.24, "grad_norm": 3.588083731203287, "learning_rate": 8.836766683609282e-06, "loss": 0.241, "step": 3416 }, { "epoch": 0.25, "grad_norm": 4.20334761592457, "learning_rate": 8.836021915712847e-06, "loss": 0.3066, "step": 3417 }, { "epoch": 0.25, "grad_norm": 5.3386360106900455, "learning_rate": 8.835276940877551e-06, "loss": 0.9279, "step": 3418 }, { "epoch": 0.25, "grad_norm": 3.7911579067179124, "learning_rate": 8.834531759143583e-06, "loss": 0.304, "step": 3419 }, { "epoch": 0.25, "grad_norm": 3.618566500015867, "learning_rate": 8.833786370551138e-06, "loss": 0.2979, "step": 3420 }, { "epoch": 0.25, "grad_norm": 13.11697188171173, "learning_rate": 8.833040775140432e-06, "loss": 1.0677, "step": 3421 }, { "epoch": 0.25, "grad_norm": 4.797000619136129, "learning_rate": 8.832294972951685e-06, "loss": 0.3636, "step": 3422 }, { "epoch": 0.25, "grad_norm": 3.7634363532213366, "learning_rate": 8.831548964025129e-06, "loss": 0.224, "step": 3423 }, { "epoch": 0.25, "grad_norm": 17.42895726057002, "learning_rate": 8.830802748401011e-06, "loss": 0.9808, "step": 3424 }, { "epoch": 0.25, "grad_norm": 4.788580242048133, "learning_rate": 8.830056326119586e-06, "loss": 0.8891, "step": 3425 }, { "epoch": 0.25, "grad_norm": 14.532401283081938, "learning_rate": 8.829309697221116e-06, "loss": 1.1534, "step": 3426 }, { "epoch": 0.25, "grad_norm": 4.553019594110495, "learning_rate": 8.828562861745886e-06, "loss": 0.2797, "step": 3427 }, { "epoch": 0.25, "grad_norm": 13.709655999834764, "learning_rate": 8.82781581973418e-06, "loss": 0.9856, "step": 3428 }, { "epoch": 0.25, "grad_norm": 4.673836372947105, "learning_rate": 8.8270685712263e-06, "loss": 1.0852, "step": 3429 }, { "epoch": 0.25, "grad_norm": 4.3769364484671325, "learning_rate": 8.826321116262556e-06, "loss": 0.9234, "step": 3430 }, { "epoch": 0.25, "grad_norm": 4.596343636852009, "learning_rate": 8.825573454883271e-06, "loss": 0.8887, "step": 3431 }, { "epoch": 0.25, "grad_norm": 14.030331610843733, "learning_rate": 8.82482558712878e-06, "loss": 1.0634, "step": 3432 }, { "epoch": 0.25, "grad_norm": 12.4996225280366, "learning_rate": 8.824077513039426e-06, "loss": 0.8746, "step": 3433 }, { "epoch": 0.25, "grad_norm": 2.8545984581730637, "learning_rate": 8.823329232655566e-06, "loss": 0.1766, "step": 3434 }, { "epoch": 0.25, "grad_norm": 4.360465931703093, "learning_rate": 8.822580746017564e-06, "loss": 1.1442, "step": 3435 }, { "epoch": 0.25, "grad_norm": 23.60237278325861, "learning_rate": 8.821832053165802e-06, "loss": 1.1897, "step": 3436 }, { "epoch": 0.25, "grad_norm": 30.05570815113164, "learning_rate": 8.821083154140667e-06, "loss": 1.1196, "step": 3437 }, { "epoch": 0.25, "grad_norm": 5.594652872948515, "learning_rate": 8.820334048982557e-06, "loss": 0.9935, "step": 3438 }, { "epoch": 0.25, "grad_norm": 28.02627961233972, "learning_rate": 8.819584737731888e-06, "loss": 1.6519, "step": 3439 }, { "epoch": 0.25, "grad_norm": 5.090453445780805, "learning_rate": 8.81883522042908e-06, "loss": 0.255, "step": 3440 }, { "epoch": 0.25, "grad_norm": 4.766351131279261, "learning_rate": 8.818085497114565e-06, "loss": 0.2389, "step": 3441 }, { "epoch": 0.25, "grad_norm": 22.556931367253764, "learning_rate": 8.817335567828792e-06, "loss": 1.1198, "step": 3442 }, { "epoch": 0.25, "grad_norm": 5.595251984454754, "learning_rate": 8.816585432612215e-06, "loss": 1.0368, "step": 3443 }, { "epoch": 0.25, "grad_norm": 4.992248733731112, "learning_rate": 8.815835091505298e-06, "loss": 0.9374, "step": 3444 }, { "epoch": 0.25, "grad_norm": 4.302001008576596, "learning_rate": 8.815084544548523e-06, "loss": 0.2839, "step": 3445 }, { "epoch": 0.25, "grad_norm": 4.387876130364773, "learning_rate": 8.814333791782377e-06, "loss": 0.3193, "step": 3446 }, { "epoch": 0.25, "grad_norm": 4.002496448777856, "learning_rate": 8.813582833247363e-06, "loss": 0.1859, "step": 3447 }, { "epoch": 0.25, "grad_norm": 4.019996899650367, "learning_rate": 8.812831668983987e-06, "loss": 0.2886, "step": 3448 }, { "epoch": 0.25, "grad_norm": 3.6566849010112175, "learning_rate": 8.812080299032776e-06, "loss": 0.1931, "step": 3449 }, { "epoch": 0.25, "grad_norm": 16.432349910373706, "learning_rate": 8.811328723434265e-06, "loss": 0.88, "step": 3450 }, { "epoch": 0.25, "grad_norm": 12.204031138796658, "learning_rate": 8.810576942228994e-06, "loss": 0.8384, "step": 3451 }, { "epoch": 0.25, "grad_norm": 36.32684212023799, "learning_rate": 8.80982495545752e-06, "loss": 1.4442, "step": 3452 }, { "epoch": 0.25, "grad_norm": 4.250323723071887, "learning_rate": 8.80907276316041e-06, "loss": 0.3236, "step": 3453 }, { "epoch": 0.25, "grad_norm": 4.02752547964483, "learning_rate": 8.808320365378246e-06, "loss": 0.2244, "step": 3454 }, { "epoch": 0.25, "grad_norm": 3.6735423481552982, "learning_rate": 8.807567762151611e-06, "loss": 0.1722, "step": 3455 }, { "epoch": 0.25, "grad_norm": 3.8266634575101586, "learning_rate": 8.80681495352111e-06, "loss": 0.1616, "step": 3456 }, { "epoch": 0.25, "grad_norm": 21.991001765729102, "learning_rate": 8.806061939527348e-06, "loss": 1.0022, "step": 3457 }, { "epoch": 0.25, "grad_norm": 4.681422115570555, "learning_rate": 8.805308720210955e-06, "loss": 0.4446, "step": 3458 }, { "epoch": 0.25, "grad_norm": 10.805700950470722, "learning_rate": 8.804555295612558e-06, "loss": 0.7005, "step": 3459 }, { "epoch": 0.25, "grad_norm": 2.912432226653979, "learning_rate": 8.803801665772804e-06, "loss": 0.1802, "step": 3460 }, { "epoch": 0.25, "grad_norm": 4.931350882128485, "learning_rate": 8.803047830732348e-06, "loss": 0.3103, "step": 3461 }, { "epoch": 0.25, "grad_norm": 3.367597602063936, "learning_rate": 8.802293790531858e-06, "loss": 0.2595, "step": 3462 }, { "epoch": 0.25, "grad_norm": 4.864510915732313, "learning_rate": 8.80153954521201e-06, "loss": 0.3826, "step": 3463 }, { "epoch": 0.25, "grad_norm": 3.2846984930628467, "learning_rate": 8.800785094813493e-06, "loss": 0.2239, "step": 3464 }, { "epoch": 0.25, "grad_norm": 5.447969355441047, "learning_rate": 8.800030439377007e-06, "loss": 0.9639, "step": 3465 }, { "epoch": 0.25, "grad_norm": 20.33932588471536, "learning_rate": 8.799275578943263e-06, "loss": 1.6828, "step": 3466 }, { "epoch": 0.25, "grad_norm": 5.079855341340237, "learning_rate": 8.798520513552981e-06, "loss": 0.2682, "step": 3467 }, { "epoch": 0.25, "grad_norm": 4.323653746575499, "learning_rate": 8.797765243246898e-06, "loss": 0.2261, "step": 3468 }, { "epoch": 0.25, "grad_norm": 3.8101355461715567, "learning_rate": 8.797009768065755e-06, "loss": 0.168, "step": 3469 }, { "epoch": 0.25, "grad_norm": 5.633308567189916, "learning_rate": 8.796254088050305e-06, "loss": 0.8192, "step": 3470 }, { "epoch": 0.25, "grad_norm": 13.494070990675546, "learning_rate": 8.795498203241319e-06, "loss": 1.1055, "step": 3471 }, { "epoch": 0.25, "grad_norm": 5.079463174727847, "learning_rate": 8.79474211367957e-06, "loss": 0.9587, "step": 3472 }, { "epoch": 0.25, "grad_norm": 4.3182613399866545, "learning_rate": 8.793985819405851e-06, "loss": 0.2143, "step": 3473 }, { "epoch": 0.25, "grad_norm": 3.3294603235738625, "learning_rate": 8.793229320460955e-06, "loss": 0.1861, "step": 3474 }, { "epoch": 0.25, "grad_norm": 18.77956464535233, "learning_rate": 8.792472616885696e-06, "loss": 0.9867, "step": 3475 }, { "epoch": 0.25, "grad_norm": 4.364746510423789, "learning_rate": 8.791715708720893e-06, "loss": 0.2821, "step": 3476 }, { "epoch": 0.25, "grad_norm": 20.774046183358454, "learning_rate": 8.790958596007382e-06, "loss": 0.8359, "step": 3477 }, { "epoch": 0.25, "grad_norm": 15.227216586420589, "learning_rate": 8.790201278786004e-06, "loss": 1.7802, "step": 3478 }, { "epoch": 0.25, "grad_norm": 24.561010377941862, "learning_rate": 8.789443757097615e-06, "loss": 1.5504, "step": 3479 }, { "epoch": 0.25, "grad_norm": 5.096299615805466, "learning_rate": 8.788686030983077e-06, "loss": 0.2991, "step": 3480 }, { "epoch": 0.25, "grad_norm": 4.279908396864872, "learning_rate": 8.787928100483269e-06, "loss": 0.3868, "step": 3481 }, { "epoch": 0.25, "grad_norm": 4.506536699554388, "learning_rate": 8.78716996563908e-06, "loss": 0.3411, "step": 3482 }, { "epoch": 0.25, "grad_norm": 16.637412860765643, "learning_rate": 8.786411626491405e-06, "loss": 0.9889, "step": 3483 }, { "epoch": 0.25, "grad_norm": 13.784617909044002, "learning_rate": 8.785653083081153e-06, "loss": 0.7728, "step": 3484 }, { "epoch": 0.25, "grad_norm": 3.1975253704955664, "learning_rate": 8.78489433544925e-06, "loss": 0.2256, "step": 3485 }, { "epoch": 0.25, "grad_norm": 3.0899125477341385, "learning_rate": 8.784135383636623e-06, "loss": 0.214, "step": 3486 }, { "epoch": 0.25, "grad_norm": 5.067997652331653, "learning_rate": 8.783376227684215e-06, "loss": 1.3452, "step": 3487 }, { "epoch": 0.25, "grad_norm": 3.1119038990069376, "learning_rate": 8.782616867632982e-06, "loss": 0.1997, "step": 3488 }, { "epoch": 0.25, "grad_norm": 16.342248247795006, "learning_rate": 8.781857303523886e-06, "loss": 1.0208, "step": 3489 }, { "epoch": 0.25, "grad_norm": 5.691832929522086, "learning_rate": 8.781097535397904e-06, "loss": 0.9971, "step": 3490 }, { "epoch": 0.25, "grad_norm": 15.316284133124205, "learning_rate": 8.780337563296022e-06, "loss": 1.1416, "step": 3491 }, { "epoch": 0.25, "grad_norm": 12.330041457673314, "learning_rate": 8.779577387259237e-06, "loss": 0.9236, "step": 3492 }, { "epoch": 0.25, "grad_norm": 3.75700371512181, "learning_rate": 8.778817007328559e-06, "loss": 0.3362, "step": 3493 }, { "epoch": 0.25, "grad_norm": 18.95419053553714, "learning_rate": 8.778056423545008e-06, "loss": 1.4273, "step": 3494 }, { "epoch": 0.25, "grad_norm": 5.347693375002226, "learning_rate": 8.777295635949612e-06, "loss": 0.9034, "step": 3495 }, { "epoch": 0.25, "grad_norm": 4.188203361519384, "learning_rate": 8.776534644583416e-06, "loss": 0.2027, "step": 3496 }, { "epoch": 0.25, "grad_norm": 17.597021003694373, "learning_rate": 8.775773449487469e-06, "loss": 1.2495, "step": 3497 }, { "epoch": 0.25, "grad_norm": 3.932482166406905, "learning_rate": 8.775012050702838e-06, "loss": 0.2561, "step": 3498 }, { "epoch": 0.25, "grad_norm": 4.175831254271052, "learning_rate": 8.774250448270596e-06, "loss": 0.2375, "step": 3499 }, { "epoch": 0.25, "grad_norm": 4.6713406356824905, "learning_rate": 8.773488642231828e-06, "loss": 0.266, "step": 3500 }, { "epoch": 0.25, "grad_norm": 4.4633422385579165, "learning_rate": 8.77272663262763e-06, "loss": 0.3146, "step": 3501 }, { "epoch": 0.25, "grad_norm": 3.809916119736843, "learning_rate": 8.771964419499113e-06, "loss": 0.2366, "step": 3502 }, { "epoch": 0.25, "grad_norm": 4.135787043812379, "learning_rate": 8.771202002887392e-06, "loss": 0.2303, "step": 3503 }, { "epoch": 0.25, "grad_norm": 3.446258598202745, "learning_rate": 8.770439382833598e-06, "loss": 0.2108, "step": 3504 }, { "epoch": 0.25, "grad_norm": 16.224155198697193, "learning_rate": 8.76967655937887e-06, "loss": 1.367, "step": 3505 }, { "epoch": 0.25, "grad_norm": 3.6870023583241074, "learning_rate": 8.768913532564363e-06, "loss": 0.1772, "step": 3506 }, { "epoch": 0.25, "grad_norm": 20.214449041979076, "learning_rate": 8.768150302431235e-06, "loss": 0.9483, "step": 3507 }, { "epoch": 0.25, "grad_norm": 14.19997116075005, "learning_rate": 8.767386869020662e-06, "loss": 1.1929, "step": 3508 }, { "epoch": 0.25, "grad_norm": 7.924915678743067, "learning_rate": 8.766623232373829e-06, "loss": 0.6846, "step": 3509 }, { "epoch": 0.25, "grad_norm": 13.001623340928983, "learning_rate": 8.76585939253193e-06, "loss": 0.8639, "step": 3510 }, { "epoch": 0.25, "grad_norm": 3.869256077302689, "learning_rate": 8.76509534953617e-06, "loss": 0.2557, "step": 3511 }, { "epoch": 0.25, "grad_norm": 3.669703518033605, "learning_rate": 8.764331103427768e-06, "loss": 0.2294, "step": 3512 }, { "epoch": 0.25, "grad_norm": 15.49628512329716, "learning_rate": 8.763566654247951e-06, "loss": 1.4454, "step": 3513 }, { "epoch": 0.25, "grad_norm": 3.863427594555248, "learning_rate": 8.76280200203796e-06, "loss": 0.2668, "step": 3514 }, { "epoch": 0.25, "grad_norm": 13.654409370239241, "learning_rate": 8.762037146839043e-06, "loss": 1.3337, "step": 3515 }, { "epoch": 0.25, "grad_norm": 10.441211700304356, "learning_rate": 8.761272088692463e-06, "loss": 0.6724, "step": 3516 }, { "epoch": 0.25, "grad_norm": 3.5470324085723326, "learning_rate": 8.76050682763949e-06, "loss": 0.2577, "step": 3517 }, { "epoch": 0.25, "grad_norm": 3.5530671397782547, "learning_rate": 8.759741363721408e-06, "loss": 0.1945, "step": 3518 }, { "epoch": 0.25, "grad_norm": 2.9529072570379573, "learning_rate": 8.758975696979513e-06, "loss": 0.203, "step": 3519 }, { "epoch": 0.25, "grad_norm": 20.587606689404012, "learning_rate": 8.758209827455105e-06, "loss": 1.0008, "step": 3520 }, { "epoch": 0.25, "grad_norm": 3.8707211203848266, "learning_rate": 8.757443755189503e-06, "loss": 0.2976, "step": 3521 }, { "epoch": 0.25, "grad_norm": 4.932341224259321, "learning_rate": 8.756677480224034e-06, "loss": 1.2423, "step": 3522 }, { "epoch": 0.25, "grad_norm": 5.448603634936699, "learning_rate": 8.755911002600034e-06, "loss": 1.0875, "step": 3523 }, { "epoch": 0.25, "grad_norm": 18.78513022826373, "learning_rate": 8.755144322358853e-06, "loss": 1.4726, "step": 3524 }, { "epoch": 0.25, "grad_norm": 4.147731793281164, "learning_rate": 8.754377439541848e-06, "loss": 0.3527, "step": 3525 }, { "epoch": 0.25, "grad_norm": 11.958090307553633, "learning_rate": 8.753610354190394e-06, "loss": 0.9437, "step": 3526 }, { "epoch": 0.25, "grad_norm": 12.909663508815738, "learning_rate": 8.752843066345868e-06, "loss": 1.3425, "step": 3527 }, { "epoch": 0.25, "grad_norm": 9.517352123137712, "learning_rate": 8.752075576049664e-06, "loss": 0.8854, "step": 3528 }, { "epoch": 0.25, "grad_norm": 3.9753020758834237, "learning_rate": 8.751307883343186e-06, "loss": 0.2095, "step": 3529 }, { "epoch": 0.25, "grad_norm": 3.513558080154856, "learning_rate": 8.750539988267845e-06, "loss": 0.25, "step": 3530 }, { "epoch": 0.25, "grad_norm": 4.031658854398174, "learning_rate": 8.749771890865069e-06, "loss": 0.2896, "step": 3531 }, { "epoch": 0.25, "grad_norm": 4.91442571958915, "learning_rate": 8.749003591176293e-06, "loss": 0.4936, "step": 3532 }, { "epoch": 0.25, "grad_norm": 3.333581490625913, "learning_rate": 8.748235089242964e-06, "loss": 0.2465, "step": 3533 }, { "epoch": 0.25, "grad_norm": 3.589413427752514, "learning_rate": 8.747466385106542e-06, "loss": 0.2268, "step": 3534 }, { "epoch": 0.25, "grad_norm": 4.283675318720279, "learning_rate": 8.746697478808491e-06, "loss": 0.375, "step": 3535 }, { "epoch": 0.25, "grad_norm": 3.4339812966585783, "learning_rate": 8.745928370390295e-06, "loss": 0.275, "step": 3536 }, { "epoch": 0.25, "grad_norm": 4.26549364132014, "learning_rate": 8.74515905989344e-06, "loss": 0.4679, "step": 3537 }, { "epoch": 0.25, "grad_norm": 11.058871411314021, "learning_rate": 8.744389547359432e-06, "loss": 1.0513, "step": 3538 }, { "epoch": 0.25, "grad_norm": 3.761983699492477, "learning_rate": 8.74361983282978e-06, "loss": 0.1789, "step": 3539 }, { "epoch": 0.25, "grad_norm": 3.5183465642272673, "learning_rate": 8.742849916346008e-06, "loss": 0.2718, "step": 3540 }, { "epoch": 0.25, "grad_norm": 20.511439714317287, "learning_rate": 8.742079797949653e-06, "loss": 1.4716, "step": 3541 }, { "epoch": 0.25, "grad_norm": 14.21684950632311, "learning_rate": 8.741309477682255e-06, "loss": 1.3184, "step": 3542 }, { "epoch": 0.25, "grad_norm": 3.9181572285675648, "learning_rate": 8.740538955585374e-06, "loss": 0.2552, "step": 3543 }, { "epoch": 0.25, "grad_norm": 4.21906782512944, "learning_rate": 8.739768231700574e-06, "loss": 0.1715, "step": 3544 }, { "epoch": 0.25, "grad_norm": 3.6671575581892104, "learning_rate": 8.738997306069432e-06, "loss": 0.2233, "step": 3545 }, { "epoch": 0.25, "grad_norm": 3.5096040002131574, "learning_rate": 8.73822617873354e-06, "loss": 0.1949, "step": 3546 }, { "epoch": 0.25, "grad_norm": 25.581599856097114, "learning_rate": 8.737454849734498e-06, "loss": 1.0376, "step": 3547 }, { "epoch": 0.25, "grad_norm": 30.474576345614047, "learning_rate": 8.73668331911391e-06, "loss": 1.4457, "step": 3548 }, { "epoch": 0.25, "grad_norm": 4.018773464622966, "learning_rate": 8.735911586913402e-06, "loss": 0.2245, "step": 3549 }, { "epoch": 0.25, "grad_norm": 3.8982425908450242, "learning_rate": 8.735139653174608e-06, "loss": 0.217, "step": 3550 }, { "epoch": 0.25, "grad_norm": 4.9680589066388725, "learning_rate": 8.734367517939163e-06, "loss": 0.9424, "step": 3551 }, { "epoch": 0.25, "grad_norm": 15.564284298678075, "learning_rate": 8.733595181248729e-06, "loss": 1.1448, "step": 3552 }, { "epoch": 0.25, "grad_norm": 20.448695286726114, "learning_rate": 8.732822643144967e-06, "loss": 1.3744, "step": 3553 }, { "epoch": 0.25, "grad_norm": 29.296662420369177, "learning_rate": 8.732049903669552e-06, "loss": 1.1522, "step": 3554 }, { "epoch": 0.25, "grad_norm": 13.614971483159856, "learning_rate": 8.731276962864172e-06, "loss": 0.7317, "step": 3555 }, { "epoch": 0.25, "grad_norm": 4.3025581448274925, "learning_rate": 8.730503820770523e-06, "loss": 0.2499, "step": 3556 }, { "epoch": 0.26, "grad_norm": 2.849701542525207, "learning_rate": 8.729730477430315e-06, "loss": 0.1864, "step": 3557 }, { "epoch": 0.26, "grad_norm": 3.2709741936148125, "learning_rate": 8.728956932885264e-06, "loss": 0.2527, "step": 3558 }, { "epoch": 0.26, "grad_norm": 5.606525082047341, "learning_rate": 8.728183187177101e-06, "loss": 0.2724, "step": 3559 }, { "epoch": 0.26, "grad_norm": 3.6724727554133367, "learning_rate": 8.727409240347569e-06, "loss": 0.2011, "step": 3560 }, { "epoch": 0.26, "grad_norm": 2.874015656254199, "learning_rate": 8.726635092438416e-06, "loss": 0.2239, "step": 3561 }, { "epoch": 0.26, "grad_norm": 15.458527188326878, "learning_rate": 8.725860743491407e-06, "loss": 0.6044, "step": 3562 }, { "epoch": 0.26, "grad_norm": 18.809532987159955, "learning_rate": 8.725086193548313e-06, "loss": 1.1343, "step": 3563 }, { "epoch": 0.26, "grad_norm": 3.8309354437306156, "learning_rate": 8.72431144265092e-06, "loss": 0.2128, "step": 3564 }, { "epoch": 0.26, "grad_norm": 4.6281711254411775, "learning_rate": 8.72353649084102e-06, "loss": 0.2845, "step": 3565 }, { "epoch": 0.26, "grad_norm": 20.17650129696855, "learning_rate": 8.722761338160422e-06, "loss": 1.5878, "step": 3566 }, { "epoch": 0.26, "grad_norm": 17.729995411319546, "learning_rate": 8.721985984650942e-06, "loss": 0.9973, "step": 3567 }, { "epoch": 0.26, "grad_norm": 3.4210945294954254, "learning_rate": 8.721210430354406e-06, "loss": 0.268, "step": 3568 }, { "epoch": 0.26, "grad_norm": 3.0807044656122593, "learning_rate": 8.720434675312652e-06, "loss": 0.185, "step": 3569 }, { "epoch": 0.26, "grad_norm": 12.135516287016376, "learning_rate": 8.719658719567532e-06, "loss": 1.0762, "step": 3570 }, { "epoch": 0.26, "grad_norm": 3.9962324390155217, "learning_rate": 8.718882563160902e-06, "loss": 0.2043, "step": 3571 }, { "epoch": 0.26, "grad_norm": 108.84708597601701, "learning_rate": 8.718106206134633e-06, "loss": 1.3436, "step": 3572 }, { "epoch": 0.26, "grad_norm": 13.92469080227643, "learning_rate": 8.71732964853061e-06, "loss": 0.6485, "step": 3573 }, { "epoch": 0.26, "grad_norm": 70.78801666020067, "learning_rate": 8.716552890390723e-06, "loss": 1.0614, "step": 3574 }, { "epoch": 0.26, "grad_norm": 9.53773931966149, "learning_rate": 8.715775931756876e-06, "loss": 0.798, "step": 3575 }, { "epoch": 0.26, "grad_norm": 3.0549943603157597, "learning_rate": 8.714998772670981e-06, "loss": 0.1712, "step": 3576 }, { "epoch": 0.26, "grad_norm": 21.352413086620505, "learning_rate": 8.714221413174967e-06, "loss": 1.2098, "step": 3577 }, { "epoch": 0.26, "grad_norm": 6.398937620985986, "learning_rate": 8.713443853310765e-06, "loss": 0.9663, "step": 3578 }, { "epoch": 0.26, "grad_norm": 10.27215881138572, "learning_rate": 8.712666093120323e-06, "loss": 0.8678, "step": 3579 }, { "epoch": 0.26, "grad_norm": 3.159814068564374, "learning_rate": 8.711888132645598e-06, "loss": 0.206, "step": 3580 }, { "epoch": 0.26, "grad_norm": 4.57298603933516, "learning_rate": 8.71110997192856e-06, "loss": 0.9097, "step": 3581 }, { "epoch": 0.26, "grad_norm": 5.311389839239068, "learning_rate": 8.710331611011184e-06, "loss": 1.0395, "step": 3582 }, { "epoch": 0.26, "grad_norm": 3.274073478031992, "learning_rate": 8.709553049935464e-06, "loss": 0.1827, "step": 3583 }, { "epoch": 0.26, "grad_norm": 4.222778794384306, "learning_rate": 8.708774288743399e-06, "loss": 0.334, "step": 3584 }, { "epoch": 0.26, "grad_norm": 5.201828977964347, "learning_rate": 8.707995327476998e-06, "loss": 0.9988, "step": 3585 }, { "epoch": 0.26, "grad_norm": 3.1390715226223866, "learning_rate": 8.707216166178285e-06, "loss": 0.2061, "step": 3586 }, { "epoch": 0.26, "grad_norm": 17.00602234678057, "learning_rate": 8.706436804889293e-06, "loss": 1.5399, "step": 3587 }, { "epoch": 0.26, "grad_norm": 12.391067346941341, "learning_rate": 8.705657243652064e-06, "loss": 0.8286, "step": 3588 }, { "epoch": 0.26, "grad_norm": 4.1990328012922955, "learning_rate": 8.704877482508655e-06, "loss": 0.1464, "step": 3589 }, { "epoch": 0.26, "grad_norm": 3.451453108890848, "learning_rate": 8.704097521501128e-06, "loss": 0.2521, "step": 3590 }, { "epoch": 0.26, "grad_norm": 23.992044012446943, "learning_rate": 8.703317360671562e-06, "loss": 1.3899, "step": 3591 }, { "epoch": 0.26, "grad_norm": 21.752626059810023, "learning_rate": 8.70253700006204e-06, "loss": 1.6241, "step": 3592 }, { "epoch": 0.26, "grad_norm": 3.8486872216131247, "learning_rate": 8.701756439714664e-06, "loss": 0.1796, "step": 3593 }, { "epoch": 0.26, "grad_norm": 4.873041659262702, "learning_rate": 8.700975679671538e-06, "loss": 0.3176, "step": 3594 }, { "epoch": 0.26, "grad_norm": 4.334582468779019, "learning_rate": 8.700194719974783e-06, "loss": 0.2433, "step": 3595 }, { "epoch": 0.26, "grad_norm": 17.40685472089483, "learning_rate": 8.69941356066653e-06, "loss": 1.2773, "step": 3596 }, { "epoch": 0.26, "grad_norm": 4.2026470894016645, "learning_rate": 8.698632201788919e-06, "loss": 0.3718, "step": 3597 }, { "epoch": 0.26, "grad_norm": 4.990091045700442, "learning_rate": 8.6978506433841e-06, "loss": 0.8569, "step": 3598 }, { "epoch": 0.26, "grad_norm": 5.363895327843206, "learning_rate": 8.697068885494236e-06, "loss": 0.8521, "step": 3599 }, { "epoch": 0.26, "grad_norm": 15.259394088592165, "learning_rate": 8.696286928161499e-06, "loss": 0.7722, "step": 3600 }, { "epoch": 0.26, "grad_norm": 3.6695315572807172, "learning_rate": 8.695504771428075e-06, "loss": 0.2118, "step": 3601 }, { "epoch": 0.26, "grad_norm": 5.175651246784312, "learning_rate": 8.694722415336155e-06, "loss": 1.0179, "step": 3602 }, { "epoch": 0.26, "grad_norm": 5.206016284319846, "learning_rate": 8.693939859927946e-06, "loss": 1.0849, "step": 3603 }, { "epoch": 0.26, "grad_norm": 3.2349610386446854, "learning_rate": 8.693157105245667e-06, "loss": 0.1954, "step": 3604 }, { "epoch": 0.26, "grad_norm": 17.055003929974447, "learning_rate": 8.692374151331539e-06, "loss": 1.0036, "step": 3605 }, { "epoch": 0.26, "grad_norm": 3.519767391003272, "learning_rate": 8.6915909982278e-06, "loss": 0.2202, "step": 3606 }, { "epoch": 0.26, "grad_norm": 3.955610259906687, "learning_rate": 8.690807645976705e-06, "loss": 0.2796, "step": 3607 }, { "epoch": 0.26, "grad_norm": 5.156723322199491, "learning_rate": 8.690024094620504e-06, "loss": 0.7622, "step": 3608 }, { "epoch": 0.26, "grad_norm": 10.375752172444642, "learning_rate": 8.689240344201473e-06, "loss": 0.7041, "step": 3609 }, { "epoch": 0.26, "grad_norm": 3.2878197174215664, "learning_rate": 8.688456394761888e-06, "loss": 0.2492, "step": 3610 }, { "epoch": 0.26, "grad_norm": 2.922194502959371, "learning_rate": 8.687672246344044e-06, "loss": 0.2237, "step": 3611 }, { "epoch": 0.26, "grad_norm": 14.053354732297194, "learning_rate": 8.686887898990239e-06, "loss": 1.5059, "step": 3612 }, { "epoch": 0.26, "grad_norm": 11.768127045166288, "learning_rate": 8.686103352742788e-06, "loss": 1.0806, "step": 3613 }, { "epoch": 0.26, "grad_norm": 18.98366430288276, "learning_rate": 8.685318607644015e-06, "loss": 0.645, "step": 3614 }, { "epoch": 0.26, "grad_norm": 18.744031706727615, "learning_rate": 8.684533663736252e-06, "loss": 0.7505, "step": 3615 }, { "epoch": 0.26, "grad_norm": 19.326846199827788, "learning_rate": 8.683748521061844e-06, "loss": 1.4157, "step": 3616 }, { "epoch": 0.26, "grad_norm": 24.495734946501013, "learning_rate": 8.682963179663147e-06, "loss": 1.0582, "step": 3617 }, { "epoch": 0.26, "grad_norm": 4.462232287506697, "learning_rate": 8.682177639582529e-06, "loss": 0.2812, "step": 3618 }, { "epoch": 0.26, "grad_norm": 5.2896355001675746, "learning_rate": 8.681391900862364e-06, "loss": 0.85, "step": 3619 }, { "epoch": 0.26, "grad_norm": 5.768145586739297, "learning_rate": 8.68060596354504e-06, "loss": 0.8437, "step": 3620 }, { "epoch": 0.26, "grad_norm": 13.0319362796861, "learning_rate": 8.679819827672957e-06, "loss": 0.9075, "step": 3621 }, { "epoch": 0.26, "grad_norm": 10.374166752691895, "learning_rate": 8.679033493288522e-06, "loss": 0.8825, "step": 3622 }, { "epoch": 0.26, "grad_norm": 2.9472888497392464, "learning_rate": 8.678246960434157e-06, "loss": 0.1852, "step": 3623 }, { "epoch": 0.26, "grad_norm": 3.150123751737153, "learning_rate": 8.677460229152292e-06, "loss": 0.2378, "step": 3624 }, { "epoch": 0.26, "grad_norm": 4.680964552691582, "learning_rate": 8.676673299485367e-06, "loss": 1.047, "step": 3625 }, { "epoch": 0.26, "grad_norm": 7.470057434681238, "learning_rate": 8.675886171475833e-06, "loss": 0.7387, "step": 3626 }, { "epoch": 0.26, "grad_norm": 3.1681828411166, "learning_rate": 8.675098845166157e-06, "loss": 0.2793, "step": 3627 }, { "epoch": 0.26, "grad_norm": 3.7918875902029407, "learning_rate": 8.674311320598807e-06, "loss": 0.2524, "step": 3628 }, { "epoch": 0.26, "grad_norm": 3.3655894611400847, "learning_rate": 8.67352359781627e-06, "loss": 0.1804, "step": 3629 }, { "epoch": 0.26, "grad_norm": 3.6655890058193283, "learning_rate": 8.67273567686104e-06, "loss": 0.2686, "step": 3630 }, { "epoch": 0.26, "grad_norm": 3.061432314487415, "learning_rate": 8.671947557775622e-06, "loss": 0.2081, "step": 3631 }, { "epoch": 0.26, "grad_norm": 3.033083368924105, "learning_rate": 8.671159240602533e-06, "loss": 0.26, "step": 3632 }, { "epoch": 0.26, "grad_norm": 33.81260112568463, "learning_rate": 8.670370725384298e-06, "loss": 1.3626, "step": 3633 }, { "epoch": 0.26, "grad_norm": 23.385913126561917, "learning_rate": 8.669582012163457e-06, "loss": 1.1701, "step": 3634 }, { "epoch": 0.26, "grad_norm": 4.843967563907713, "learning_rate": 8.668793100982555e-06, "loss": 0.282, "step": 3635 }, { "epoch": 0.26, "grad_norm": 13.940621771079071, "learning_rate": 8.668003991884154e-06, "loss": 0.7439, "step": 3636 }, { "epoch": 0.26, "grad_norm": 3.745941014291613, "learning_rate": 8.66721468491082e-06, "loss": 0.2238, "step": 3637 }, { "epoch": 0.26, "grad_norm": 23.632955542645007, "learning_rate": 8.666425180105136e-06, "loss": 1.1981, "step": 3638 }, { "epoch": 0.26, "grad_norm": 19.29417215852253, "learning_rate": 8.66563547750969e-06, "loss": 1.5867, "step": 3639 }, { "epoch": 0.26, "grad_norm": 5.354513110509464, "learning_rate": 8.664845577167087e-06, "loss": 0.8307, "step": 3640 }, { "epoch": 0.26, "grad_norm": 13.903088738722166, "learning_rate": 8.664055479119937e-06, "loss": 0.7766, "step": 3641 }, { "epoch": 0.26, "grad_norm": 4.048908785625099, "learning_rate": 8.663265183410863e-06, "loss": 0.2137, "step": 3642 }, { "epoch": 0.26, "grad_norm": 5.338696249667881, "learning_rate": 8.662474690082498e-06, "loss": 0.8939, "step": 3643 }, { "epoch": 0.26, "grad_norm": 4.444574849817186, "learning_rate": 8.661683999177489e-06, "loss": 0.8969, "step": 3644 }, { "epoch": 0.26, "grad_norm": 6.1122718167120285, "learning_rate": 8.660893110738485e-06, "loss": 1.0112, "step": 3645 }, { "epoch": 0.26, "grad_norm": 19.798499495996758, "learning_rate": 8.660102024808158e-06, "loss": 1.6006, "step": 3646 }, { "epoch": 0.26, "grad_norm": 5.1147797097898495, "learning_rate": 8.659310741429178e-06, "loss": 0.4147, "step": 3647 }, { "epoch": 0.26, "grad_norm": 37.81698178815399, "learning_rate": 8.658519260644236e-06, "loss": 1.2098, "step": 3648 }, { "epoch": 0.26, "grad_norm": 19.773062577612695, "learning_rate": 8.657727582496029e-06, "loss": 1.3399, "step": 3649 }, { "epoch": 0.26, "grad_norm": 22.423111146349626, "learning_rate": 8.656935707027264e-06, "loss": 1.2183, "step": 3650 }, { "epoch": 0.26, "grad_norm": 4.281342819913707, "learning_rate": 8.656143634280659e-06, "loss": 0.1731, "step": 3651 }, { "epoch": 0.26, "grad_norm": 5.073854133673658, "learning_rate": 8.655351364298945e-06, "loss": 0.4124, "step": 3652 }, { "epoch": 0.26, "grad_norm": 13.200947781946482, "learning_rate": 8.65455889712486e-06, "loss": 1.0016, "step": 3653 }, { "epoch": 0.26, "grad_norm": 4.152267406629177, "learning_rate": 8.653766232801158e-06, "loss": 0.3319, "step": 3654 }, { "epoch": 0.26, "grad_norm": 18.794596964310497, "learning_rate": 8.652973371370597e-06, "loss": 1.1134, "step": 3655 }, { "epoch": 0.26, "grad_norm": 17.379148383133813, "learning_rate": 8.652180312875948e-06, "loss": 0.9313, "step": 3656 }, { "epoch": 0.26, "grad_norm": 4.1441576628747585, "learning_rate": 8.651387057359996e-06, "loss": 0.9566, "step": 3657 }, { "epoch": 0.26, "grad_norm": 3.4609002189731664, "learning_rate": 8.650593604865535e-06, "loss": 0.2716, "step": 3658 }, { "epoch": 0.26, "grad_norm": 3.6332299379230646, "learning_rate": 8.649799955435367e-06, "loss": 0.2701, "step": 3659 }, { "epoch": 0.26, "grad_norm": 4.089579898242848, "learning_rate": 8.649006109112305e-06, "loss": 0.1881, "step": 3660 }, { "epoch": 0.26, "grad_norm": 2.834465448208203, "learning_rate": 8.64821206593918e-06, "loss": 0.1542, "step": 3661 }, { "epoch": 0.26, "grad_norm": 9.181285723990403, "learning_rate": 8.64741782595882e-06, "loss": 0.5342, "step": 3662 }, { "epoch": 0.26, "grad_norm": 3.555067369428993, "learning_rate": 8.646623389214073e-06, "loss": 0.2716, "step": 3663 }, { "epoch": 0.26, "grad_norm": 13.213401721387795, "learning_rate": 8.6458287557478e-06, "loss": 1.1366, "step": 3664 }, { "epoch": 0.26, "grad_norm": 13.354065930550671, "learning_rate": 8.645033925602864e-06, "loss": 1.0143, "step": 3665 }, { "epoch": 0.26, "grad_norm": 6.216733240083664, "learning_rate": 8.644238898822146e-06, "loss": 1.1489, "step": 3666 }, { "epoch": 0.26, "grad_norm": 3.4933572290484753, "learning_rate": 8.643443675448533e-06, "loss": 0.1675, "step": 3667 }, { "epoch": 0.26, "grad_norm": 3.9563892232974394, "learning_rate": 8.642648255524926e-06, "loss": 0.2816, "step": 3668 }, { "epoch": 0.26, "grad_norm": 23.11352833216446, "learning_rate": 8.641852639094232e-06, "loss": 1.4323, "step": 3669 }, { "epoch": 0.26, "grad_norm": 31.478032891059023, "learning_rate": 8.641056826199375e-06, "loss": 1.5564, "step": 3670 }, { "epoch": 0.26, "grad_norm": 5.972025633988419, "learning_rate": 8.640260816883284e-06, "loss": 1.0006, "step": 3671 }, { "epoch": 0.26, "grad_norm": 17.45930575686076, "learning_rate": 8.6394646111889e-06, "loss": 1.0012, "step": 3672 }, { "epoch": 0.26, "grad_norm": 3.8559963441388927, "learning_rate": 8.638668209159177e-06, "loss": 0.2331, "step": 3673 }, { "epoch": 0.26, "grad_norm": 3.8879897372968624, "learning_rate": 8.637871610837078e-06, "loss": 0.2668, "step": 3674 }, { "epoch": 0.26, "grad_norm": 4.1816754464938395, "learning_rate": 8.637074816265574e-06, "loss": 0.271, "step": 3675 }, { "epoch": 0.26, "grad_norm": 2.950621341212956, "learning_rate": 8.636277825487654e-06, "loss": 0.1775, "step": 3676 }, { "epoch": 0.26, "grad_norm": 4.613700188019994, "learning_rate": 8.635480638546305e-06, "loss": 0.3059, "step": 3677 }, { "epoch": 0.26, "grad_norm": 3.6297176115707175, "learning_rate": 8.63468325548454e-06, "loss": 0.2644, "step": 3678 }, { "epoch": 0.26, "grad_norm": 5.4999291406974855, "learning_rate": 8.63388567634537e-06, "loss": 1.1992, "step": 3679 }, { "epoch": 0.26, "grad_norm": 3.6248611017153665, "learning_rate": 8.633087901171822e-06, "loss": 0.23, "step": 3680 }, { "epoch": 0.26, "grad_norm": 3.544626054659625, "learning_rate": 8.632289930006935e-06, "loss": 0.2841, "step": 3681 }, { "epoch": 0.26, "grad_norm": 4.017506157647747, "learning_rate": 8.631491762893755e-06, "loss": 0.3198, "step": 3682 }, { "epoch": 0.26, "grad_norm": 16.789704369468712, "learning_rate": 8.630693399875341e-06, "loss": 1.0162, "step": 3683 }, { "epoch": 0.26, "grad_norm": 5.472027703196137, "learning_rate": 8.62989484099476e-06, "loss": 1.0252, "step": 3684 }, { "epoch": 0.26, "grad_norm": 4.018325859514503, "learning_rate": 8.629096086295092e-06, "loss": 0.2826, "step": 3685 }, { "epoch": 0.26, "grad_norm": 11.090868884839301, "learning_rate": 8.628297135819428e-06, "loss": 0.7535, "step": 3686 }, { "epoch": 0.26, "grad_norm": 3.694846795664302, "learning_rate": 8.627497989610868e-06, "loss": 0.2654, "step": 3687 }, { "epoch": 0.26, "grad_norm": 15.018421357752922, "learning_rate": 8.62669864771252e-06, "loss": 0.9681, "step": 3688 }, { "epoch": 0.26, "grad_norm": 4.391072015990822, "learning_rate": 8.625899110167508e-06, "loss": 0.2071, "step": 3689 }, { "epoch": 0.26, "grad_norm": 5.082309185463799, "learning_rate": 8.625099377018965e-06, "loss": 1.0044, "step": 3690 }, { "epoch": 0.26, "grad_norm": 4.4550280515168295, "learning_rate": 8.624299448310031e-06, "loss": 0.2452, "step": 3691 }, { "epoch": 0.26, "grad_norm": 5.506666112007027, "learning_rate": 8.623499324083862e-06, "loss": 1.2264, "step": 3692 }, { "epoch": 0.26, "grad_norm": 17.858630410919165, "learning_rate": 8.62269900438362e-06, "loss": 1.0143, "step": 3693 }, { "epoch": 0.26, "grad_norm": 4.145910650237702, "learning_rate": 8.621898489252479e-06, "loss": 0.2543, "step": 3694 }, { "epoch": 0.26, "grad_norm": 17.53913722563536, "learning_rate": 8.621097778733624e-06, "loss": 1.3509, "step": 3695 }, { "epoch": 0.27, "grad_norm": 40.63188294927123, "learning_rate": 8.620296872870251e-06, "loss": 1.1461, "step": 3696 }, { "epoch": 0.27, "grad_norm": 3.987139481135104, "learning_rate": 8.619495771705564e-06, "loss": 0.2048, "step": 3697 }, { "epoch": 0.27, "grad_norm": 5.407469457754169, "learning_rate": 8.618694475282783e-06, "loss": 1.0336, "step": 3698 }, { "epoch": 0.27, "grad_norm": 3.538792352783614, "learning_rate": 8.61789298364513e-06, "loss": 0.1721, "step": 3699 }, { "epoch": 0.27, "grad_norm": 3.8386509620068647, "learning_rate": 8.617091296835847e-06, "loss": 0.2048, "step": 3700 }, { "epoch": 0.27, "grad_norm": 5.3632037848281495, "learning_rate": 8.616289414898178e-06, "loss": 0.8375, "step": 3701 }, { "epoch": 0.27, "grad_norm": 4.13718288281943, "learning_rate": 8.615487337875385e-06, "loss": 0.2077, "step": 3702 }, { "epoch": 0.27, "grad_norm": 10.888107441270357, "learning_rate": 8.614685065810734e-06, "loss": 1.0885, "step": 3703 }, { "epoch": 0.27, "grad_norm": 3.161086402425564, "learning_rate": 8.613882598747509e-06, "loss": 0.1658, "step": 3704 }, { "epoch": 0.27, "grad_norm": 13.075354219306654, "learning_rate": 8.613079936728994e-06, "loss": 1.3306, "step": 3705 }, { "epoch": 0.27, "grad_norm": 4.065144983113022, "learning_rate": 8.612277079798494e-06, "loss": 0.3253, "step": 3706 }, { "epoch": 0.27, "grad_norm": 16.042894888184186, "learning_rate": 8.611474027999319e-06, "loss": 0.965, "step": 3707 }, { "epoch": 0.27, "grad_norm": 3.814024309478831, "learning_rate": 8.610670781374789e-06, "loss": 0.246, "step": 3708 }, { "epoch": 0.27, "grad_norm": 4.744158957031691, "learning_rate": 8.609867339968239e-06, "loss": 0.2241, "step": 3709 }, { "epoch": 0.27, "grad_norm": 3.1385083811395984, "learning_rate": 8.609063703823009e-06, "loss": 0.1462, "step": 3710 }, { "epoch": 0.27, "grad_norm": 4.927049556008404, "learning_rate": 8.608259872982454e-06, "loss": 0.9888, "step": 3711 }, { "epoch": 0.27, "grad_norm": 32.47003910176378, "learning_rate": 8.607455847489935e-06, "loss": 1.3294, "step": 3712 }, { "epoch": 0.27, "grad_norm": 4.906722227232288, "learning_rate": 8.606651627388831e-06, "loss": 0.4129, "step": 3713 }, { "epoch": 0.27, "grad_norm": 5.394078470836299, "learning_rate": 8.605847212722523e-06, "loss": 0.982, "step": 3714 }, { "epoch": 0.27, "grad_norm": 4.074483750590205, "learning_rate": 8.605042603534405e-06, "loss": 0.2563, "step": 3715 }, { "epoch": 0.27, "grad_norm": 5.036156761206086, "learning_rate": 8.604237799867885e-06, "loss": 0.3624, "step": 3716 }, { "epoch": 0.27, "grad_norm": 5.8491517266726705, "learning_rate": 8.603432801766379e-06, "loss": 0.2564, "step": 3717 }, { "epoch": 0.27, "grad_norm": 4.42344084307025, "learning_rate": 8.602627609273313e-06, "loss": 0.3827, "step": 3718 }, { "epoch": 0.27, "grad_norm": 17.12826522065824, "learning_rate": 8.601822222432124e-06, "loss": 1.4523, "step": 3719 }, { "epoch": 0.27, "grad_norm": 4.129151455129166, "learning_rate": 8.601016641286259e-06, "loss": 0.2429, "step": 3720 }, { "epoch": 0.27, "grad_norm": 6.139437057730459, "learning_rate": 8.600210865879178e-06, "loss": 1.3165, "step": 3721 }, { "epoch": 0.27, "grad_norm": 4.266962830697113, "learning_rate": 8.599404896254349e-06, "loss": 0.2325, "step": 3722 }, { "epoch": 0.27, "grad_norm": 4.350181910680772, "learning_rate": 8.598598732455248e-06, "loss": 0.3462, "step": 3723 }, { "epoch": 0.27, "grad_norm": 3.7849029619182355, "learning_rate": 8.59779237452537e-06, "loss": 0.2656, "step": 3724 }, { "epoch": 0.27, "grad_norm": 4.070305120977494, "learning_rate": 8.596985822508208e-06, "loss": 0.2615, "step": 3725 }, { "epoch": 0.27, "grad_norm": 23.54773677802838, "learning_rate": 8.596179076447278e-06, "loss": 1.2628, "step": 3726 }, { "epoch": 0.27, "grad_norm": 3.5111155260598665, "learning_rate": 8.5953721363861e-06, "loss": 0.2144, "step": 3727 }, { "epoch": 0.27, "grad_norm": 3.1015967280584986, "learning_rate": 8.594565002368202e-06, "loss": 0.1347, "step": 3728 }, { "epoch": 0.27, "grad_norm": 3.1630324969170553, "learning_rate": 8.593757674437131e-06, "loss": 0.1741, "step": 3729 }, { "epoch": 0.27, "grad_norm": 3.5482059552898364, "learning_rate": 8.592950152636435e-06, "loss": 0.2553, "step": 3730 }, { "epoch": 0.27, "grad_norm": 4.22572671018048, "learning_rate": 8.592142437009678e-06, "loss": 0.8934, "step": 3731 }, { "epoch": 0.27, "grad_norm": 3.809927478493268, "learning_rate": 8.591334527600434e-06, "loss": 0.1585, "step": 3732 }, { "epoch": 0.27, "grad_norm": 8.989084924548722, "learning_rate": 8.590526424452287e-06, "loss": 0.6927, "step": 3733 }, { "epoch": 0.27, "grad_norm": 3.522140318790771, "learning_rate": 8.58971812760883e-06, "loss": 0.2398, "step": 3734 }, { "epoch": 0.27, "grad_norm": 3.41346504691419, "learning_rate": 8.588909637113668e-06, "loss": 0.1737, "step": 3735 }, { "epoch": 0.27, "grad_norm": 3.4472158494338565, "learning_rate": 8.588100953010415e-06, "loss": 0.2504, "step": 3736 }, { "epoch": 0.27, "grad_norm": 3.416144651337662, "learning_rate": 8.587292075342698e-06, "loss": 0.1907, "step": 3737 }, { "epoch": 0.27, "grad_norm": 5.664966505103805, "learning_rate": 8.586483004154151e-06, "loss": 1.1593, "step": 3738 }, { "epoch": 0.27, "grad_norm": 2.8767589936703697, "learning_rate": 8.585673739488422e-06, "loss": 0.1984, "step": 3739 }, { "epoch": 0.27, "grad_norm": 20.33498142576689, "learning_rate": 8.584864281389168e-06, "loss": 1.3825, "step": 3740 }, { "epoch": 0.27, "grad_norm": 15.488242787251785, "learning_rate": 8.584054629900057e-06, "loss": 1.0994, "step": 3741 }, { "epoch": 0.27, "grad_norm": 5.3786022441352035, "learning_rate": 8.583244785064762e-06, "loss": 1.0985, "step": 3742 }, { "epoch": 0.27, "grad_norm": 9.742156742536459, "learning_rate": 8.582434746926976e-06, "loss": 0.7347, "step": 3743 }, { "epoch": 0.27, "grad_norm": 12.475362457578319, "learning_rate": 8.581624515530393e-06, "loss": 1.2705, "step": 3744 }, { "epoch": 0.27, "grad_norm": 17.05835996492873, "learning_rate": 8.580814090918727e-06, "loss": 1.0455, "step": 3745 }, { "epoch": 0.27, "grad_norm": 14.791972182473677, "learning_rate": 8.580003473135695e-06, "loss": 0.6239, "step": 3746 }, { "epoch": 0.27, "grad_norm": 18.164996153579406, "learning_rate": 8.579192662225026e-06, "loss": 1.2016, "step": 3747 }, { "epoch": 0.27, "grad_norm": 4.014930000210543, "learning_rate": 8.57838165823046e-06, "loss": 0.2488, "step": 3748 }, { "epoch": 0.27, "grad_norm": 15.63093771570149, "learning_rate": 8.57757046119575e-06, "loss": 1.13, "step": 3749 }, { "epoch": 0.27, "grad_norm": 5.816608205307317, "learning_rate": 8.576759071164656e-06, "loss": 0.3689, "step": 3750 }, { "epoch": 0.27, "grad_norm": 4.99621132504866, "learning_rate": 8.575947488180946e-06, "loss": 0.9539, "step": 3751 }, { "epoch": 0.27, "grad_norm": 3.6005575631041586, "learning_rate": 8.575135712288409e-06, "loss": 0.1893, "step": 3752 }, { "epoch": 0.27, "grad_norm": 4.197154087996343, "learning_rate": 8.57432374353083e-06, "loss": 0.2687, "step": 3753 }, { "epoch": 0.27, "grad_norm": 23.213053622552035, "learning_rate": 8.573511581952015e-06, "loss": 1.4792, "step": 3754 }, { "epoch": 0.27, "grad_norm": 4.6964706880648, "learning_rate": 8.572699227595775e-06, "loss": 0.1432, "step": 3755 }, { "epoch": 0.27, "grad_norm": 12.261134447147962, "learning_rate": 8.571886680505938e-06, "loss": 0.8238, "step": 3756 }, { "epoch": 0.27, "grad_norm": 3.7535754065490115, "learning_rate": 8.571073940726334e-06, "loss": 0.2479, "step": 3757 }, { "epoch": 0.27, "grad_norm": 10.467212490672248, "learning_rate": 8.570261008300808e-06, "loss": 0.8168, "step": 3758 }, { "epoch": 0.27, "grad_norm": 4.212238153945301, "learning_rate": 8.569447883273215e-06, "loss": 0.2334, "step": 3759 }, { "epoch": 0.27, "grad_norm": 6.4392639696148795, "learning_rate": 8.568634565687418e-06, "loss": 0.7862, "step": 3760 }, { "epoch": 0.27, "grad_norm": 27.108905859779103, "learning_rate": 8.567821055587295e-06, "loss": 1.8871, "step": 3761 }, { "epoch": 0.27, "grad_norm": 6.677978133595711, "learning_rate": 8.567007353016732e-06, "loss": 0.9276, "step": 3762 }, { "epoch": 0.27, "grad_norm": 5.200572897721333, "learning_rate": 8.566193458019624e-06, "loss": 0.533, "step": 3763 }, { "epoch": 0.27, "grad_norm": 4.333601495574828, "learning_rate": 8.565379370639876e-06, "loss": 1.0075, "step": 3764 }, { "epoch": 0.27, "grad_norm": 14.436049574450934, "learning_rate": 8.56456509092141e-06, "loss": 1.2567, "step": 3765 }, { "epoch": 0.27, "grad_norm": 3.39893180996515, "learning_rate": 8.563750618908148e-06, "loss": 0.1711, "step": 3766 }, { "epoch": 0.27, "grad_norm": 4.148868697217046, "learning_rate": 8.56293595464403e-06, "loss": 0.2445, "step": 3767 }, { "epoch": 0.27, "grad_norm": 10.778944675806496, "learning_rate": 8.562121098173002e-06, "loss": 0.7538, "step": 3768 }, { "epoch": 0.27, "grad_norm": 4.076335141937408, "learning_rate": 8.561306049539027e-06, "loss": 0.1905, "step": 3769 }, { "epoch": 0.27, "grad_norm": 4.53827526339687, "learning_rate": 8.560490808786069e-06, "loss": 1.0388, "step": 3770 }, { "epoch": 0.27, "grad_norm": 4.235931081299815, "learning_rate": 8.559675375958108e-06, "loss": 0.3964, "step": 3771 }, { "epoch": 0.27, "grad_norm": 13.03989548447869, "learning_rate": 8.558859751099137e-06, "loss": 0.7772, "step": 3772 }, { "epoch": 0.27, "grad_norm": 4.243368605553692, "learning_rate": 8.558043934253152e-06, "loss": 0.2439, "step": 3773 }, { "epoch": 0.27, "grad_norm": 4.196411204739771, "learning_rate": 8.557227925464165e-06, "loss": 0.3563, "step": 3774 }, { "epoch": 0.27, "grad_norm": 4.945112456560713, "learning_rate": 8.556411724776196e-06, "loss": 1.1519, "step": 3775 }, { "epoch": 0.27, "grad_norm": 12.818021800055178, "learning_rate": 8.555595332233277e-06, "loss": 0.7438, "step": 3776 }, { "epoch": 0.27, "grad_norm": 3.705678427768827, "learning_rate": 8.554778747879448e-06, "loss": 0.2928, "step": 3777 }, { "epoch": 0.27, "grad_norm": 3.8392592313947134, "learning_rate": 8.553961971758762e-06, "loss": 0.2886, "step": 3778 }, { "epoch": 0.27, "grad_norm": 14.462576193205734, "learning_rate": 8.55314500391528e-06, "loss": 1.1738, "step": 3779 }, { "epoch": 0.27, "grad_norm": 2.546803709425161, "learning_rate": 8.552327844393072e-06, "loss": 0.1442, "step": 3780 }, { "epoch": 0.27, "grad_norm": 3.8507567046453506, "learning_rate": 8.551510493236227e-06, "loss": 0.2472, "step": 3781 }, { "epoch": 0.27, "grad_norm": 3.566025316051921, "learning_rate": 8.55069295048883e-06, "loss": 0.2211, "step": 3782 }, { "epoch": 0.27, "grad_norm": 4.812638980891375, "learning_rate": 8.549875216194991e-06, "loss": 0.9588, "step": 3783 }, { "epoch": 0.27, "grad_norm": 20.937754635304092, "learning_rate": 8.54905729039882e-06, "loss": 1.2122, "step": 3784 }, { "epoch": 0.27, "grad_norm": 8.294448325561634, "learning_rate": 8.548239173144443e-06, "loss": 0.6856, "step": 3785 }, { "epoch": 0.27, "grad_norm": 13.511629256821506, "learning_rate": 8.547420864475992e-06, "loss": 0.6623, "step": 3786 }, { "epoch": 0.27, "grad_norm": 15.006542048287548, "learning_rate": 8.546602364437615e-06, "loss": 1.5427, "step": 3787 }, { "epoch": 0.27, "grad_norm": 6.1602783369857885, "learning_rate": 8.545783673073463e-06, "loss": 0.3091, "step": 3788 }, { "epoch": 0.27, "grad_norm": 19.47940816032135, "learning_rate": 8.544964790427704e-06, "loss": 1.5092, "step": 3789 }, { "epoch": 0.27, "grad_norm": 5.81451136221218, "learning_rate": 8.544145716544515e-06, "loss": 0.3444, "step": 3790 }, { "epoch": 0.27, "grad_norm": 4.212661544285519, "learning_rate": 8.543326451468078e-06, "loss": 0.2525, "step": 3791 }, { "epoch": 0.27, "grad_norm": 46.596024169970896, "learning_rate": 8.542506995242591e-06, "loss": 2.4728, "step": 3792 }, { "epoch": 0.27, "grad_norm": 11.837417544701681, "learning_rate": 8.541687347912262e-06, "loss": 0.7958, "step": 3793 }, { "epoch": 0.27, "grad_norm": 4.144972288912679, "learning_rate": 8.540867509521305e-06, "loss": 0.387, "step": 3794 }, { "epoch": 0.27, "grad_norm": 4.792865865582306, "learning_rate": 8.54004748011395e-06, "loss": 1.0058, "step": 3795 }, { "epoch": 0.27, "grad_norm": 18.755697809049153, "learning_rate": 8.539227259734432e-06, "loss": 0.8732, "step": 3796 }, { "epoch": 0.27, "grad_norm": 13.217087676073366, "learning_rate": 8.538406848427002e-06, "loss": 1.2165, "step": 3797 }, { "epoch": 0.27, "grad_norm": 3.694744817562581, "learning_rate": 8.537586246235914e-06, "loss": 0.2348, "step": 3798 }, { "epoch": 0.27, "grad_norm": 15.724527428394063, "learning_rate": 8.53676545320544e-06, "loss": 0.8856, "step": 3799 }, { "epoch": 0.27, "grad_norm": 3.8679772837171034, "learning_rate": 8.535944469379855e-06, "loss": 0.2544, "step": 3800 }, { "epoch": 0.27, "grad_norm": 4.8894568665648235, "learning_rate": 8.535123294803453e-06, "loss": 0.9394, "step": 3801 }, { "epoch": 0.27, "grad_norm": 4.7957888755170615, "learning_rate": 8.534301929520529e-06, "loss": 0.9276, "step": 3802 }, { "epoch": 0.27, "grad_norm": 11.476516069366419, "learning_rate": 8.533480373575393e-06, "loss": 0.8427, "step": 3803 }, { "epoch": 0.27, "grad_norm": 14.773027241128847, "learning_rate": 8.532658627012369e-06, "loss": 1.0759, "step": 3804 }, { "epoch": 0.27, "grad_norm": 19.242050820025998, "learning_rate": 8.531836689875781e-06, "loss": 0.9285, "step": 3805 }, { "epoch": 0.27, "grad_norm": 15.944888207938803, "learning_rate": 8.531014562209974e-06, "loss": 1.2705, "step": 3806 }, { "epoch": 0.27, "grad_norm": 3.3017926263161876, "learning_rate": 8.530192244059296e-06, "loss": 0.2204, "step": 3807 }, { "epoch": 0.27, "grad_norm": 14.080493170655409, "learning_rate": 8.529369735468109e-06, "loss": 0.6125, "step": 3808 }, { "epoch": 0.27, "grad_norm": 21.43400562807047, "learning_rate": 8.528547036480782e-06, "loss": 1.5715, "step": 3809 }, { "epoch": 0.27, "grad_norm": 4.377873784788969, "learning_rate": 8.527724147141702e-06, "loss": 0.4225, "step": 3810 }, { "epoch": 0.27, "grad_norm": 32.584967383884084, "learning_rate": 8.526901067495257e-06, "loss": 1.185, "step": 3811 }, { "epoch": 0.27, "grad_norm": 5.867084757253306, "learning_rate": 8.526077797585849e-06, "loss": 0.9472, "step": 3812 }, { "epoch": 0.27, "grad_norm": 8.11009943973136, "learning_rate": 8.525254337457891e-06, "loss": 0.7732, "step": 3813 }, { "epoch": 0.27, "grad_norm": 4.827113447039869, "learning_rate": 8.524430687155805e-06, "loss": 0.8602, "step": 3814 }, { "epoch": 0.27, "grad_norm": 4.431324225172715, "learning_rate": 8.523606846724023e-06, "loss": 0.8608, "step": 3815 }, { "epoch": 0.27, "grad_norm": 2.690418767199702, "learning_rate": 8.522782816206992e-06, "loss": 0.2021, "step": 3816 }, { "epoch": 0.27, "grad_norm": 3.2163189885920946, "learning_rate": 8.52195859564916e-06, "loss": 0.178, "step": 3817 }, { "epoch": 0.27, "grad_norm": 2.7478001183519236, "learning_rate": 8.521134185094992e-06, "loss": 0.1799, "step": 3818 }, { "epoch": 0.27, "grad_norm": 12.999900535408901, "learning_rate": 8.520309584588965e-06, "loss": 1.0471, "step": 3819 }, { "epoch": 0.27, "grad_norm": 4.473862286809004, "learning_rate": 8.51948479417556e-06, "loss": 1.0193, "step": 3820 }, { "epoch": 0.27, "grad_norm": 3.96889237403533, "learning_rate": 8.518659813899275e-06, "loss": 0.2897, "step": 3821 }, { "epoch": 0.27, "grad_norm": 3.465528934426226, "learning_rate": 8.51783464380461e-06, "loss": 0.2396, "step": 3822 }, { "epoch": 0.27, "grad_norm": 3.4115315117556766, "learning_rate": 8.517009283936082e-06, "loss": 0.244, "step": 3823 }, { "epoch": 0.27, "grad_norm": 3.9103662036373192, "learning_rate": 8.516183734338216e-06, "loss": 0.3085, "step": 3824 }, { "epoch": 0.27, "grad_norm": 21.94163153560491, "learning_rate": 8.515357995055545e-06, "loss": 1.366, "step": 3825 }, { "epoch": 0.27, "grad_norm": 4.570794310558312, "learning_rate": 8.514532066132618e-06, "loss": 0.3375, "step": 3826 }, { "epoch": 0.27, "grad_norm": 5.213189066365377, "learning_rate": 8.513705947613988e-06, "loss": 0.297, "step": 3827 }, { "epoch": 0.27, "grad_norm": 14.825639086524095, "learning_rate": 8.512879639544224e-06, "loss": 0.7363, "step": 3828 }, { "epoch": 0.27, "grad_norm": 4.759366442317506, "learning_rate": 8.512053141967902e-06, "loss": 0.7803, "step": 3829 }, { "epoch": 0.27, "grad_norm": 4.928334742517857, "learning_rate": 8.511226454929604e-06, "loss": 0.8867, "step": 3830 }, { "epoch": 0.27, "grad_norm": 23.085071898133467, "learning_rate": 8.510399578473929e-06, "loss": 1.2622, "step": 3831 }, { "epoch": 0.27, "grad_norm": 5.840679221637579, "learning_rate": 8.509572512645486e-06, "loss": 0.9769, "step": 3832 }, { "epoch": 0.27, "grad_norm": 5.061397196274653, "learning_rate": 8.508745257488888e-06, "loss": 0.8271, "step": 3833 }, { "epoch": 0.27, "grad_norm": 5.369637383365995, "learning_rate": 8.507917813048768e-06, "loss": 0.8047, "step": 3834 }, { "epoch": 0.27, "grad_norm": 22.515799464970815, "learning_rate": 8.50709017936976e-06, "loss": 1.8573, "step": 3835 }, { "epoch": 0.28, "grad_norm": 17.505942863370894, "learning_rate": 8.506262356496509e-06, "loss": 0.908, "step": 3836 }, { "epoch": 0.28, "grad_norm": 17.80856493901718, "learning_rate": 8.505434344473678e-06, "loss": 0.8733, "step": 3837 }, { "epoch": 0.28, "grad_norm": 3.336683205803644, "learning_rate": 8.504606143345931e-06, "loss": 0.2216, "step": 3838 }, { "epoch": 0.28, "grad_norm": 3.6738058610339586, "learning_rate": 8.50377775315795e-06, "loss": 0.2713, "step": 3839 }, { "epoch": 0.28, "grad_norm": 3.6626674659797587, "learning_rate": 8.502949173954423e-06, "loss": 0.1933, "step": 3840 }, { "epoch": 0.28, "grad_norm": 5.208912433594024, "learning_rate": 8.502120405780045e-06, "loss": 1.06, "step": 3841 }, { "epoch": 0.28, "grad_norm": 4.872502474413096, "learning_rate": 8.501291448679527e-06, "loss": 0.7746, "step": 3842 }, { "epoch": 0.28, "grad_norm": 27.92634609794288, "learning_rate": 8.50046230269759e-06, "loss": 0.9678, "step": 3843 }, { "epoch": 0.28, "grad_norm": 49.092314729013204, "learning_rate": 8.499632967878962e-06, "loss": 0.8746, "step": 3844 }, { "epoch": 0.28, "grad_norm": 32.67324352758971, "learning_rate": 8.49880344426838e-06, "loss": 1.7191, "step": 3845 }, { "epoch": 0.28, "grad_norm": 4.275402671843022, "learning_rate": 8.497973731910599e-06, "loss": 0.3432, "step": 3846 }, { "epoch": 0.28, "grad_norm": 4.810196325862556, "learning_rate": 8.497143830850374e-06, "loss": 0.4273, "step": 3847 }, { "epoch": 0.28, "grad_norm": 16.389715872965382, "learning_rate": 8.496313741132476e-06, "loss": 0.9827, "step": 3848 }, { "epoch": 0.28, "grad_norm": 3.7996417543420686, "learning_rate": 8.495483462801686e-06, "loss": 0.2432, "step": 3849 }, { "epoch": 0.28, "grad_norm": 3.4257347359889594, "learning_rate": 8.494652995902795e-06, "loss": 0.1936, "step": 3850 }, { "epoch": 0.28, "grad_norm": 3.596809502206199, "learning_rate": 8.493822340480602e-06, "loss": 0.1739, "step": 3851 }, { "epoch": 0.28, "grad_norm": 3.995259079879438, "learning_rate": 8.492991496579921e-06, "loss": 0.2476, "step": 3852 }, { "epoch": 0.28, "grad_norm": 4.689995431439387, "learning_rate": 8.492160464245569e-06, "loss": 0.4026, "step": 3853 }, { "epoch": 0.28, "grad_norm": 18.032999468516177, "learning_rate": 8.491329243522378e-06, "loss": 1.4032, "step": 3854 }, { "epoch": 0.28, "grad_norm": 25.52515496056896, "learning_rate": 8.490497834455189e-06, "loss": 1.3044, "step": 3855 }, { "epoch": 0.28, "grad_norm": 21.730955572347597, "learning_rate": 8.489666237088857e-06, "loss": 1.1448, "step": 3856 }, { "epoch": 0.28, "grad_norm": 3.99188600357338, "learning_rate": 8.488834451468238e-06, "loss": 0.3204, "step": 3857 }, { "epoch": 0.28, "grad_norm": 11.864570418162948, "learning_rate": 8.488002477638207e-06, "loss": 1.1159, "step": 3858 }, { "epoch": 0.28, "grad_norm": 11.195781545151707, "learning_rate": 8.487170315643645e-06, "loss": 0.9209, "step": 3859 }, { "epoch": 0.28, "grad_norm": 4.419034029018861, "learning_rate": 8.486337965529445e-06, "loss": 0.3637, "step": 3860 }, { "epoch": 0.28, "grad_norm": 4.9327355750378015, "learning_rate": 8.485505427340508e-06, "loss": 1.2338, "step": 3861 }, { "epoch": 0.28, "grad_norm": 3.6616723200834667, "learning_rate": 8.484672701121746e-06, "loss": 0.2438, "step": 3862 }, { "epoch": 0.28, "grad_norm": 4.9626018727597865, "learning_rate": 8.48383978691808e-06, "loss": 0.9222, "step": 3863 }, { "epoch": 0.28, "grad_norm": 4.964522790047121, "learning_rate": 8.483006684774448e-06, "loss": 0.9526, "step": 3864 }, { "epoch": 0.28, "grad_norm": 4.5721168124900196, "learning_rate": 8.48217339473579e-06, "loss": 0.8065, "step": 3865 }, { "epoch": 0.28, "grad_norm": 3.9106544460376167, "learning_rate": 8.481339916847054e-06, "loss": 0.218, "step": 3866 }, { "epoch": 0.28, "grad_norm": 4.561635518449303, "learning_rate": 8.480506251153208e-06, "loss": 0.1485, "step": 3867 }, { "epoch": 0.28, "grad_norm": 4.954615690541229, "learning_rate": 8.479672397699227e-06, "loss": 1.0755, "step": 3868 }, { "epoch": 0.28, "grad_norm": 2.8207032190121017, "learning_rate": 8.478838356530089e-06, "loss": 0.2077, "step": 3869 }, { "epoch": 0.28, "grad_norm": 27.166161414140824, "learning_rate": 8.478004127690791e-06, "loss": 1.8641, "step": 3870 }, { "epoch": 0.28, "grad_norm": 4.898044264915411, "learning_rate": 8.477169711226335e-06, "loss": 0.1979, "step": 3871 }, { "epoch": 0.28, "grad_norm": 4.3668855134214875, "learning_rate": 8.476335107181734e-06, "loss": 0.1427, "step": 3872 }, { "epoch": 0.28, "grad_norm": 6.455278017761061, "learning_rate": 8.475500315602013e-06, "loss": 1.0109, "step": 3873 }, { "epoch": 0.28, "grad_norm": 5.152741065425498, "learning_rate": 8.474665336532207e-06, "loss": 0.3032, "step": 3874 }, { "epoch": 0.28, "grad_norm": 4.921826325441647, "learning_rate": 8.473830170017357e-06, "loss": 0.2254, "step": 3875 }, { "epoch": 0.28, "grad_norm": 22.68031094345012, "learning_rate": 8.472994816102518e-06, "loss": 1.8139, "step": 3876 }, { "epoch": 0.28, "grad_norm": 26.34410680223311, "learning_rate": 8.472159274832755e-06, "loss": 1.4996, "step": 3877 }, { "epoch": 0.28, "grad_norm": 6.266382137577809, "learning_rate": 8.471323546253143e-06, "loss": 0.989, "step": 3878 }, { "epoch": 0.28, "grad_norm": 4.243199529099097, "learning_rate": 8.470487630408763e-06, "loss": 0.2006, "step": 3879 }, { "epoch": 0.28, "grad_norm": 4.675598142835519, "learning_rate": 8.469651527344714e-06, "loss": 0.2146, "step": 3880 }, { "epoch": 0.28, "grad_norm": 5.545215565367382, "learning_rate": 8.468815237106098e-06, "loss": 0.905, "step": 3881 }, { "epoch": 0.28, "grad_norm": 5.684693815504427, "learning_rate": 8.46797875973803e-06, "loss": 1.0062, "step": 3882 }, { "epoch": 0.28, "grad_norm": 11.667647001339889, "learning_rate": 8.467142095285635e-06, "loss": 0.708, "step": 3883 }, { "epoch": 0.28, "grad_norm": 3.1998431411206463, "learning_rate": 8.466305243794049e-06, "loss": 0.1924, "step": 3884 }, { "epoch": 0.28, "grad_norm": 3.3508711629431196, "learning_rate": 8.465468205308414e-06, "loss": 0.2674, "step": 3885 }, { "epoch": 0.28, "grad_norm": 23.177878596027018, "learning_rate": 8.464630979873887e-06, "loss": 0.9259, "step": 3886 }, { "epoch": 0.28, "grad_norm": 3.8770772243031244, "learning_rate": 8.463793567535633e-06, "loss": 0.2136, "step": 3887 }, { "epoch": 0.28, "grad_norm": 3.799605498264615, "learning_rate": 8.462955968338828e-06, "loss": 0.2479, "step": 3888 }, { "epoch": 0.28, "grad_norm": 3.9158351157873543, "learning_rate": 8.462118182328655e-06, "loss": 0.3342, "step": 3889 }, { "epoch": 0.28, "grad_norm": 4.384147724329073, "learning_rate": 8.461280209550313e-06, "loss": 0.32, "step": 3890 }, { "epoch": 0.28, "grad_norm": 3.6022876067363065, "learning_rate": 8.460442050049004e-06, "loss": 0.2772, "step": 3891 }, { "epoch": 0.28, "grad_norm": 4.373394788749662, "learning_rate": 8.459603703869945e-06, "loss": 0.3704, "step": 3892 }, { "epoch": 0.28, "grad_norm": 15.193853789748001, "learning_rate": 8.45876517105836e-06, "loss": 0.7592, "step": 3893 }, { "epoch": 0.28, "grad_norm": 2.9143264177435078, "learning_rate": 8.45792645165949e-06, "loss": 0.1757, "step": 3894 }, { "epoch": 0.28, "grad_norm": 5.506963540446746, "learning_rate": 8.457087545718574e-06, "loss": 0.3396, "step": 3895 }, { "epoch": 0.28, "grad_norm": 3.146399398666843, "learning_rate": 8.456248453280873e-06, "loss": 0.2855, "step": 3896 }, { "epoch": 0.28, "grad_norm": 3.814083436658194, "learning_rate": 8.455409174391648e-06, "loss": 0.3079, "step": 3897 }, { "epoch": 0.28, "grad_norm": 3.6276820460111394, "learning_rate": 8.454569709096179e-06, "loss": 0.2599, "step": 3898 }, { "epoch": 0.28, "grad_norm": 7.706884022122633, "learning_rate": 8.453730057439749e-06, "loss": 0.3601, "step": 3899 }, { "epoch": 0.28, "grad_norm": 3.0993184358356327, "learning_rate": 8.452890219467657e-06, "loss": 0.1913, "step": 3900 }, { "epoch": 0.28, "grad_norm": 10.429434049449492, "learning_rate": 8.452050195225207e-06, "loss": 0.6627, "step": 3901 }, { "epoch": 0.28, "grad_norm": 11.006070813645554, "learning_rate": 8.451209984757715e-06, "loss": 0.5422, "step": 3902 }, { "epoch": 0.28, "grad_norm": 26.37677339927823, "learning_rate": 8.450369588110508e-06, "loss": 0.8902, "step": 3903 }, { "epoch": 0.28, "grad_norm": 26.592717950817224, "learning_rate": 8.449529005328922e-06, "loss": 1.2908, "step": 3904 }, { "epoch": 0.28, "grad_norm": 5.036890731129747, "learning_rate": 8.448688236458305e-06, "loss": 0.3773, "step": 3905 }, { "epoch": 0.28, "grad_norm": 42.32059067900691, "learning_rate": 8.44784728154401e-06, "loss": 1.4075, "step": 3906 }, { "epoch": 0.28, "grad_norm": 4.653614681356294, "learning_rate": 8.447006140631407e-06, "loss": 0.2332, "step": 3907 }, { "epoch": 0.28, "grad_norm": 9.961548973128902, "learning_rate": 8.446164813765868e-06, "loss": 0.5554, "step": 3908 }, { "epoch": 0.28, "grad_norm": 185.08289246331702, "learning_rate": 8.445323300992783e-06, "loss": 1.4116, "step": 3909 }, { "epoch": 0.28, "grad_norm": 22.44345756039768, "learning_rate": 8.444481602357546e-06, "loss": 1.5681, "step": 3910 }, { "epoch": 0.28, "grad_norm": 4.2777956593256405, "learning_rate": 8.443639717905565e-06, "loss": 0.3002, "step": 3911 }, { "epoch": 0.28, "grad_norm": 17.42750474797714, "learning_rate": 8.442797647682257e-06, "loss": 1.1043, "step": 3912 }, { "epoch": 0.28, "grad_norm": 4.138211137935797, "learning_rate": 8.441955391733048e-06, "loss": 0.2838, "step": 3913 }, { "epoch": 0.28, "grad_norm": 15.221493284675057, "learning_rate": 8.441112950103371e-06, "loss": 1.1591, "step": 3914 }, { "epoch": 0.28, "grad_norm": 5.054482049489833, "learning_rate": 8.440270322838679e-06, "loss": 0.911, "step": 3915 }, { "epoch": 0.28, "grad_norm": 4.13903997721406, "learning_rate": 8.439427509984424e-06, "loss": 0.3267, "step": 3916 }, { "epoch": 0.28, "grad_norm": 4.698576944334966, "learning_rate": 8.438584511586074e-06, "loss": 0.2874, "step": 3917 }, { "epoch": 0.28, "grad_norm": 21.91533783070259, "learning_rate": 8.437741327689105e-06, "loss": 1.0053, "step": 3918 }, { "epoch": 0.28, "grad_norm": 2.7680158731856657, "learning_rate": 8.436897958339005e-06, "loss": 0.1995, "step": 3919 }, { "epoch": 0.28, "grad_norm": 3.6289586008489816, "learning_rate": 8.436054403581268e-06, "loss": 0.2379, "step": 3920 }, { "epoch": 0.28, "grad_norm": 15.813016853414387, "learning_rate": 8.435210663461403e-06, "loss": 1.2444, "step": 3921 }, { "epoch": 0.28, "grad_norm": 3.0975492585906044, "learning_rate": 8.434366738024925e-06, "loss": 0.1763, "step": 3922 }, { "epoch": 0.28, "grad_norm": 5.632467966022395, "learning_rate": 8.433522627317361e-06, "loss": 1.0383, "step": 3923 }, { "epoch": 0.28, "grad_norm": 5.7371301379299515, "learning_rate": 8.432678331384249e-06, "loss": 0.2148, "step": 3924 }, { "epoch": 0.28, "grad_norm": 3.5337052017609305, "learning_rate": 8.431833850271134e-06, "loss": 0.2963, "step": 3925 }, { "epoch": 0.28, "grad_norm": 3.527870371118004, "learning_rate": 8.430989184023572e-06, "loss": 0.164, "step": 3926 }, { "epoch": 0.28, "grad_norm": 3.4564029327804815, "learning_rate": 8.430144332687133e-06, "loss": 0.2037, "step": 3927 }, { "epoch": 0.28, "grad_norm": 4.568186922847124, "learning_rate": 8.42929929630739e-06, "loss": 0.2972, "step": 3928 }, { "epoch": 0.28, "grad_norm": 29.724853216989768, "learning_rate": 8.428454074929929e-06, "loss": 0.9778, "step": 3929 }, { "epoch": 0.28, "grad_norm": 3.8580297854495034, "learning_rate": 8.427608668600348e-06, "loss": 0.2518, "step": 3930 }, { "epoch": 0.28, "grad_norm": 5.79712546812697, "learning_rate": 8.426763077364257e-06, "loss": 0.9891, "step": 3931 }, { "epoch": 0.28, "grad_norm": 19.96716711217093, "learning_rate": 8.425917301267266e-06, "loss": 1.3817, "step": 3932 }, { "epoch": 0.28, "grad_norm": 5.568025184931114, "learning_rate": 8.425071340355005e-06, "loss": 0.974, "step": 3933 }, { "epoch": 0.28, "grad_norm": 14.374875235093995, "learning_rate": 8.42422519467311e-06, "loss": 0.9532, "step": 3934 }, { "epoch": 0.28, "grad_norm": 3.5931674160185487, "learning_rate": 8.423378864267226e-06, "loss": 0.2251, "step": 3935 }, { "epoch": 0.28, "grad_norm": 54.83486710250049, "learning_rate": 8.422532349183012e-06, "loss": 1.7503, "step": 3936 }, { "epoch": 0.28, "grad_norm": 5.824929224817277, "learning_rate": 8.421685649466133e-06, "loss": 0.8687, "step": 3937 }, { "epoch": 0.28, "grad_norm": 3.9958780847582296, "learning_rate": 8.420838765162264e-06, "loss": 0.1867, "step": 3938 }, { "epoch": 0.28, "grad_norm": 24.545758618426607, "learning_rate": 8.419991696317091e-06, "loss": 1.0758, "step": 3939 }, { "epoch": 0.28, "grad_norm": 24.063575578912587, "learning_rate": 8.419144442976313e-06, "loss": 0.831, "step": 3940 }, { "epoch": 0.28, "grad_norm": 4.351169789775591, "learning_rate": 8.418297005185634e-06, "loss": 0.1864, "step": 3941 }, { "epoch": 0.28, "grad_norm": 3.7847748584291505, "learning_rate": 8.417449382990771e-06, "loss": 0.2546, "step": 3942 }, { "epoch": 0.28, "grad_norm": 3.43020818493393, "learning_rate": 8.41660157643745e-06, "loss": 0.2004, "step": 3943 }, { "epoch": 0.28, "grad_norm": 22.06780616920303, "learning_rate": 8.415753585571406e-06, "loss": 1.1, "step": 3944 }, { "epoch": 0.28, "grad_norm": 4.153346042274948, "learning_rate": 8.414905410438385e-06, "loss": 0.194, "step": 3945 }, { "epoch": 0.28, "grad_norm": 18.65996142092308, "learning_rate": 8.414057051084144e-06, "loss": 1.1536, "step": 3946 }, { "epoch": 0.28, "grad_norm": 4.292967037992706, "learning_rate": 8.413208507554447e-06, "loss": 0.3372, "step": 3947 }, { "epoch": 0.28, "grad_norm": 17.698082850522564, "learning_rate": 8.41235977989507e-06, "loss": 1.0697, "step": 3948 }, { "epoch": 0.28, "grad_norm": 4.128266835414966, "learning_rate": 8.411510868151801e-06, "loss": 0.3195, "step": 3949 }, { "epoch": 0.28, "grad_norm": 23.645983705878123, "learning_rate": 8.410661772370433e-06, "loss": 1.5507, "step": 3950 }, { "epoch": 0.28, "grad_norm": 22.703975678075263, "learning_rate": 8.409812492596774e-06, "loss": 1.2635, "step": 3951 }, { "epoch": 0.28, "grad_norm": 22.16325986229377, "learning_rate": 8.408963028876636e-06, "loss": 1.1219, "step": 3952 }, { "epoch": 0.28, "grad_norm": 9.658337350806827, "learning_rate": 8.408113381255847e-06, "loss": 0.6047, "step": 3953 }, { "epoch": 0.28, "grad_norm": 2.8886472724374688, "learning_rate": 8.40726354978024e-06, "loss": 0.1844, "step": 3954 }, { "epoch": 0.28, "grad_norm": 3.9312886103601663, "learning_rate": 8.406413534495664e-06, "loss": 0.2472, "step": 3955 }, { "epoch": 0.28, "grad_norm": 3.440530316180248, "learning_rate": 8.40556333544797e-06, "loss": 0.3106, "step": 3956 }, { "epoch": 0.28, "grad_norm": 2.7246319919138795, "learning_rate": 8.404712952683027e-06, "loss": 0.1872, "step": 3957 }, { "epoch": 0.28, "grad_norm": 3.2575935369716613, "learning_rate": 8.403862386246706e-06, "loss": 0.1777, "step": 3958 }, { "epoch": 0.28, "grad_norm": 4.7846474359475915, "learning_rate": 8.403011636184894e-06, "loss": 0.3806, "step": 3959 }, { "epoch": 0.28, "grad_norm": 4.470439676139121, "learning_rate": 8.402160702543485e-06, "loss": 1.0498, "step": 3960 }, { "epoch": 0.28, "grad_norm": 19.14463807581333, "learning_rate": 8.401309585368385e-06, "loss": 1.0571, "step": 3961 }, { "epoch": 0.28, "grad_norm": 14.066226931005671, "learning_rate": 8.400458284705506e-06, "loss": 0.9251, "step": 3962 }, { "epoch": 0.28, "grad_norm": 41.409755463402256, "learning_rate": 8.399606800600774e-06, "loss": 1.4047, "step": 3963 }, { "epoch": 0.28, "grad_norm": 4.119358819323716, "learning_rate": 8.398755133100125e-06, "loss": 0.1736, "step": 3964 }, { "epoch": 0.28, "grad_norm": 4.269036091887605, "learning_rate": 8.3979032822495e-06, "loss": 0.2238, "step": 3965 }, { "epoch": 0.28, "grad_norm": 11.398391968904392, "learning_rate": 8.397051248094856e-06, "loss": 0.6853, "step": 3966 }, { "epoch": 0.28, "grad_norm": 3.7209167468350204, "learning_rate": 8.396199030682156e-06, "loss": 0.2559, "step": 3967 }, { "epoch": 0.28, "grad_norm": 14.784053400606446, "learning_rate": 8.395346630057372e-06, "loss": 1.2254, "step": 3968 }, { "epoch": 0.28, "grad_norm": 4.390758247976648, "learning_rate": 8.39449404626649e-06, "loss": 0.3401, "step": 3969 }, { "epoch": 0.28, "grad_norm": 3.6310318909368218, "learning_rate": 8.393641279355502e-06, "loss": 0.1841, "step": 3970 }, { "epoch": 0.28, "grad_norm": 4.754282035820674, "learning_rate": 8.392788329370414e-06, "loss": 0.9007, "step": 3971 }, { "epoch": 0.28, "grad_norm": 3.5081123044747673, "learning_rate": 8.391935196357239e-06, "loss": 0.16, "step": 3972 }, { "epoch": 0.28, "grad_norm": 3.337566417568712, "learning_rate": 8.391081880361998e-06, "loss": 0.1722, "step": 3973 }, { "epoch": 0.28, "grad_norm": 3.6817713501218794, "learning_rate": 8.390228381430725e-06, "loss": 0.1477, "step": 3974 }, { "epoch": 0.29, "grad_norm": 12.503268935203886, "learning_rate": 8.389374699609465e-06, "loss": 1.1296, "step": 3975 }, { "epoch": 0.29, "grad_norm": 4.1904073880885235, "learning_rate": 8.388520834944268e-06, "loss": 0.316, "step": 3976 }, { "epoch": 0.29, "grad_norm": 3.0850018704252733, "learning_rate": 8.3876667874812e-06, "loss": 0.1784, "step": 3977 }, { "epoch": 0.29, "grad_norm": 10.754579656573641, "learning_rate": 8.386812557266329e-06, "loss": 0.8584, "step": 3978 }, { "epoch": 0.29, "grad_norm": 3.7510343889028377, "learning_rate": 8.385958144345743e-06, "loss": 0.391, "step": 3979 }, { "epoch": 0.29, "grad_norm": 3.5969130403284226, "learning_rate": 8.38510354876553e-06, "loss": 0.2287, "step": 3980 }, { "epoch": 0.29, "grad_norm": 9.510478759098657, "learning_rate": 8.384248770571794e-06, "loss": 0.9312, "step": 3981 }, { "epoch": 0.29, "grad_norm": 17.535813936755723, "learning_rate": 8.383393809810648e-06, "loss": 1.5834, "step": 3982 }, { "epoch": 0.29, "grad_norm": 4.175583889138343, "learning_rate": 8.382538666528212e-06, "loss": 0.2805, "step": 3983 }, { "epoch": 0.29, "grad_norm": 14.420008826811287, "learning_rate": 8.381683340770618e-06, "loss": 2.068, "step": 3984 }, { "epoch": 0.29, "grad_norm": 4.799319075099816, "learning_rate": 8.380827832584008e-06, "loss": 0.4037, "step": 3985 }, { "epoch": 0.29, "grad_norm": 4.782334372258805, "learning_rate": 8.379972142014534e-06, "loss": 0.8556, "step": 3986 }, { "epoch": 0.29, "grad_norm": 5.548830925931665, "learning_rate": 8.379116269108359e-06, "loss": 0.9698, "step": 3987 }, { "epoch": 0.29, "grad_norm": 4.358117572566381, "learning_rate": 8.378260213911648e-06, "loss": 0.1892, "step": 3988 }, { "epoch": 0.29, "grad_norm": 4.510480482978773, "learning_rate": 8.37740397647059e-06, "loss": 0.2778, "step": 3989 }, { "epoch": 0.29, "grad_norm": 15.943032716167217, "learning_rate": 8.376547556831369e-06, "loss": 1.3908, "step": 3990 }, { "epoch": 0.29, "grad_norm": 3.7547444376527217, "learning_rate": 8.37569095504019e-06, "loss": 0.1654, "step": 3991 }, { "epoch": 0.29, "grad_norm": 4.000080866003662, "learning_rate": 8.37483417114326e-06, "loss": 0.1739, "step": 3992 }, { "epoch": 0.29, "grad_norm": 4.922499647499341, "learning_rate": 8.373977205186801e-06, "loss": 1.0659, "step": 3993 }, { "epoch": 0.29, "grad_norm": 13.297110796355625, "learning_rate": 8.373120057217044e-06, "loss": 1.2723, "step": 3994 }, { "epoch": 0.29, "grad_norm": 4.07878824998737, "learning_rate": 8.372262727280228e-06, "loss": 0.2829, "step": 3995 }, { "epoch": 0.29, "grad_norm": 5.60091434141493, "learning_rate": 8.3714052154226e-06, "loss": 0.265, "step": 3996 }, { "epoch": 0.29, "grad_norm": 5.125882688857994, "learning_rate": 8.370547521690424e-06, "loss": 0.8174, "step": 3997 }, { "epoch": 0.29, "grad_norm": 4.7413982357283135, "learning_rate": 8.369689646129968e-06, "loss": 0.9578, "step": 3998 }, { "epoch": 0.29, "grad_norm": 35.125187856107814, "learning_rate": 8.36883158878751e-06, "loss": 2.003, "step": 3999 }, { "epoch": 0.29, "grad_norm": 5.077719423536201, "learning_rate": 8.36797334970934e-06, "loss": 0.3341, "step": 4000 }, { "epoch": 0.29, "grad_norm": 5.563968180827309, "learning_rate": 8.367114928941755e-06, "loss": 1.0398, "step": 4001 }, { "epoch": 0.29, "grad_norm": 13.101465561073873, "learning_rate": 8.366256326531065e-06, "loss": 1.2757, "step": 4002 }, { "epoch": 0.29, "grad_norm": 20.93452446922676, "learning_rate": 8.36539754252359e-06, "loss": 1.3644, "step": 4003 }, { "epoch": 0.29, "grad_norm": 4.921346561973997, "learning_rate": 8.364538576965653e-06, "loss": 0.3037, "step": 4004 }, { "epoch": 0.29, "grad_norm": 9.007382723960834, "learning_rate": 8.363679429903597e-06, "loss": 0.7474, "step": 4005 }, { "epoch": 0.29, "grad_norm": 16.587608115524596, "learning_rate": 8.36282010138377e-06, "loss": 1.331, "step": 4006 }, { "epoch": 0.29, "grad_norm": 4.230219763242356, "learning_rate": 8.361960591452525e-06, "loss": 0.1971, "step": 4007 }, { "epoch": 0.29, "grad_norm": 4.474585105400044, "learning_rate": 8.361100900156232e-06, "loss": 0.2761, "step": 4008 }, { "epoch": 0.29, "grad_norm": 4.229516407630052, "learning_rate": 8.36024102754127e-06, "loss": 0.2992, "step": 4009 }, { "epoch": 0.29, "grad_norm": 6.555140302610431, "learning_rate": 8.35938097365402e-06, "loss": 0.3307, "step": 4010 }, { "epoch": 0.29, "grad_norm": 4.956457297760118, "learning_rate": 8.358520738540886e-06, "loss": 0.9394, "step": 4011 }, { "epoch": 0.29, "grad_norm": 4.453088203715537, "learning_rate": 8.357660322248269e-06, "loss": 0.1972, "step": 4012 }, { "epoch": 0.29, "grad_norm": 7.9164047011231515, "learning_rate": 8.356799724822586e-06, "loss": 0.8126, "step": 4013 }, { "epoch": 0.29, "grad_norm": 4.895580792103942, "learning_rate": 8.355938946310265e-06, "loss": 0.8622, "step": 4014 }, { "epoch": 0.29, "grad_norm": 18.202151571200098, "learning_rate": 8.355077986757743e-06, "loss": 1.2151, "step": 4015 }, { "epoch": 0.29, "grad_norm": 8.471348986700896, "learning_rate": 8.354216846211462e-06, "loss": 0.7858, "step": 4016 }, { "epoch": 0.29, "grad_norm": 7.585653664533867, "learning_rate": 8.353355524717877e-06, "loss": 0.8195, "step": 4017 }, { "epoch": 0.29, "grad_norm": 10.496323260214673, "learning_rate": 8.352494022323457e-06, "loss": 0.7505, "step": 4018 }, { "epoch": 0.29, "grad_norm": 3.814156256587538, "learning_rate": 8.35163233907467e-06, "loss": 0.2759, "step": 4019 }, { "epoch": 0.29, "grad_norm": 4.37820734755874, "learning_rate": 8.350770475018009e-06, "loss": 0.922, "step": 4020 }, { "epoch": 0.29, "grad_norm": 3.62466253281944, "learning_rate": 8.349908430199965e-06, "loss": 0.1946, "step": 4021 }, { "epoch": 0.29, "grad_norm": 10.29322141194682, "learning_rate": 8.349046204667038e-06, "loss": 1.0392, "step": 4022 }, { "epoch": 0.29, "grad_norm": 3.258010019220926, "learning_rate": 8.348183798465746e-06, "loss": 0.2114, "step": 4023 }, { "epoch": 0.29, "grad_norm": 13.55987437620624, "learning_rate": 8.347321211642611e-06, "loss": 0.9283, "step": 4024 }, { "epoch": 0.29, "grad_norm": 3.8994989938351075, "learning_rate": 8.34645844424417e-06, "loss": 0.3457, "step": 4025 }, { "epoch": 0.29, "grad_norm": 3.6225484381950994, "learning_rate": 8.345595496316961e-06, "loss": 0.2677, "step": 4026 }, { "epoch": 0.29, "grad_norm": 8.386250117962742, "learning_rate": 8.344732367907538e-06, "loss": 0.6844, "step": 4027 }, { "epoch": 0.29, "grad_norm": 13.144157056698496, "learning_rate": 8.343869059062466e-06, "loss": 1.0597, "step": 4028 }, { "epoch": 0.29, "grad_norm": 4.446347921282385, "learning_rate": 8.343005569828313e-06, "loss": 0.2538, "step": 4029 }, { "epoch": 0.29, "grad_norm": 3.798363798110741, "learning_rate": 8.342141900251666e-06, "loss": 0.2206, "step": 4030 }, { "epoch": 0.29, "grad_norm": 3.7552248923747418, "learning_rate": 8.341278050379113e-06, "loss": 0.2268, "step": 4031 }, { "epoch": 0.29, "grad_norm": 30.40025094460224, "learning_rate": 8.340414020257258e-06, "loss": 1.6974, "step": 4032 }, { "epoch": 0.29, "grad_norm": 14.790871512031991, "learning_rate": 8.339549809932709e-06, "loss": 0.8645, "step": 4033 }, { "epoch": 0.29, "grad_norm": 4.167655977633774, "learning_rate": 8.338685419452089e-06, "loss": 0.8456, "step": 4034 }, { "epoch": 0.29, "grad_norm": 3.394190702275888, "learning_rate": 8.337820848862028e-06, "loss": 0.2765, "step": 4035 }, { "epoch": 0.29, "grad_norm": 4.532679475303648, "learning_rate": 8.336956098209169e-06, "loss": 0.2115, "step": 4036 }, { "epoch": 0.29, "grad_norm": 15.366728019066198, "learning_rate": 8.336091167540154e-06, "loss": 1.5901, "step": 4037 }, { "epoch": 0.29, "grad_norm": 3.903651337673075, "learning_rate": 8.335226056901653e-06, "loss": 0.2387, "step": 4038 }, { "epoch": 0.29, "grad_norm": 4.536392083242983, "learning_rate": 8.33436076634033e-06, "loss": 0.43, "step": 4039 }, { "epoch": 0.29, "grad_norm": 4.249019471652361, "learning_rate": 8.333495295902865e-06, "loss": 0.2538, "step": 4040 }, { "epoch": 0.29, "grad_norm": 4.461652340859804, "learning_rate": 8.332629645635946e-06, "loss": 0.2656, "step": 4041 }, { "epoch": 0.29, "grad_norm": 25.752277970851026, "learning_rate": 8.331763815586272e-06, "loss": 1.2026, "step": 4042 }, { "epoch": 0.29, "grad_norm": 4.052816220528876, "learning_rate": 8.330897805800553e-06, "loss": 0.2279, "step": 4043 }, { "epoch": 0.29, "grad_norm": 12.728307466736545, "learning_rate": 8.330031616325504e-06, "loss": 1.0168, "step": 4044 }, { "epoch": 0.29, "grad_norm": 17.376490136450037, "learning_rate": 8.329165247207856e-06, "loss": 1.0016, "step": 4045 }, { "epoch": 0.29, "grad_norm": 11.53737219066984, "learning_rate": 8.328298698494342e-06, "loss": 0.9912, "step": 4046 }, { "epoch": 0.29, "grad_norm": 5.198705338591846, "learning_rate": 8.327431970231712e-06, "loss": 1.0173, "step": 4047 }, { "epoch": 0.29, "grad_norm": 4.774959199008686, "learning_rate": 8.326565062466723e-06, "loss": 0.7894, "step": 4048 }, { "epoch": 0.29, "grad_norm": 11.234444451404647, "learning_rate": 8.325697975246142e-06, "loss": 1.0363, "step": 4049 }, { "epoch": 0.29, "grad_norm": 4.211012496572706, "learning_rate": 8.324830708616741e-06, "loss": 0.2771, "step": 4050 }, { "epoch": 0.29, "grad_norm": 5.830067489890288, "learning_rate": 8.32396326262531e-06, "loss": 0.9182, "step": 4051 }, { "epoch": 0.29, "grad_norm": 3.625397162319384, "learning_rate": 8.323095637318643e-06, "loss": 0.2622, "step": 4052 }, { "epoch": 0.29, "grad_norm": 5.181879788892342, "learning_rate": 8.322227832743544e-06, "loss": 1.0509, "step": 4053 }, { "epoch": 0.29, "grad_norm": 3.446914506125653, "learning_rate": 8.321359848946828e-06, "loss": 0.2674, "step": 4054 }, { "epoch": 0.29, "grad_norm": 3.1505379619963665, "learning_rate": 8.320491685975322e-06, "loss": 0.1673, "step": 4055 }, { "epoch": 0.29, "grad_norm": 31.789087686169125, "learning_rate": 8.319623343875858e-06, "loss": 1.013, "step": 4056 }, { "epoch": 0.29, "grad_norm": 5.196253849370992, "learning_rate": 8.318754822695282e-06, "loss": 0.3186, "step": 4057 }, { "epoch": 0.29, "grad_norm": 14.222742446763128, "learning_rate": 8.317886122480441e-06, "loss": 0.8626, "step": 4058 }, { "epoch": 0.29, "grad_norm": 3.020027024668826, "learning_rate": 8.317017243278205e-06, "loss": 0.2277, "step": 4059 }, { "epoch": 0.29, "grad_norm": 2.981449523406235, "learning_rate": 8.316148185135444e-06, "loss": 0.21, "step": 4060 }, { "epoch": 0.29, "grad_norm": 4.789948999100519, "learning_rate": 8.315278948099041e-06, "loss": 1.0996, "step": 4061 }, { "epoch": 0.29, "grad_norm": 4.328884785824113, "learning_rate": 8.314409532215887e-06, "loss": 0.7487, "step": 4062 }, { "epoch": 0.29, "grad_norm": 3.274424675445406, "learning_rate": 8.313539937532886e-06, "loss": 0.1934, "step": 4063 }, { "epoch": 0.29, "grad_norm": 3.6779563752021107, "learning_rate": 8.312670164096947e-06, "loss": 0.2575, "step": 4064 }, { "epoch": 0.29, "grad_norm": 23.712059107611807, "learning_rate": 8.311800211954993e-06, "loss": 1.1616, "step": 4065 }, { "epoch": 0.29, "grad_norm": 4.829928245284333, "learning_rate": 8.310930081153951e-06, "loss": 0.3683, "step": 4066 }, { "epoch": 0.29, "grad_norm": 3.655244093057447, "learning_rate": 8.310059771740764e-06, "loss": 0.2829, "step": 4067 }, { "epoch": 0.29, "grad_norm": 16.478631003864578, "learning_rate": 8.309189283762382e-06, "loss": 1.2811, "step": 4068 }, { "epoch": 0.29, "grad_norm": 4.492922449646144, "learning_rate": 8.308318617265764e-06, "loss": 0.2291, "step": 4069 }, { "epoch": 0.29, "grad_norm": 19.306632613974774, "learning_rate": 8.307447772297882e-06, "loss": 1.2791, "step": 4070 }, { "epoch": 0.29, "grad_norm": 19.55853871688624, "learning_rate": 8.306576748905709e-06, "loss": 0.7459, "step": 4071 }, { "epoch": 0.29, "grad_norm": 3.4550483077783025, "learning_rate": 8.305705547136237e-06, "loss": 0.1677, "step": 4072 }, { "epoch": 0.29, "grad_norm": 3.6287829862521224, "learning_rate": 8.304834167036464e-06, "loss": 0.2748, "step": 4073 }, { "epoch": 0.29, "grad_norm": 2.7998685404021173, "learning_rate": 8.303962608653396e-06, "loss": 0.1744, "step": 4074 }, { "epoch": 0.29, "grad_norm": 5.8516960344446085, "learning_rate": 8.303090872034053e-06, "loss": 0.9963, "step": 4075 }, { "epoch": 0.29, "grad_norm": 20.412088441831163, "learning_rate": 8.30221895722546e-06, "loss": 1.4638, "step": 4076 }, { "epoch": 0.29, "grad_norm": 4.115312422172461, "learning_rate": 8.301346864274652e-06, "loss": 0.3029, "step": 4077 }, { "epoch": 0.29, "grad_norm": 4.515115177142103, "learning_rate": 8.30047459322868e-06, "loss": 0.2607, "step": 4078 }, { "epoch": 0.29, "grad_norm": 32.11354242649532, "learning_rate": 8.299602144134594e-06, "loss": 1.3782, "step": 4079 }, { "epoch": 0.29, "grad_norm": 16.783261863796152, "learning_rate": 8.298729517039463e-06, "loss": 1.3586, "step": 4080 }, { "epoch": 0.29, "grad_norm": 17.600338700808628, "learning_rate": 8.297856711990361e-06, "loss": 0.976, "step": 4081 }, { "epoch": 0.29, "grad_norm": 4.402275080221205, "learning_rate": 8.296983729034374e-06, "loss": 0.9069, "step": 4082 }, { "epoch": 0.29, "grad_norm": 24.288844574992023, "learning_rate": 8.296110568218592e-06, "loss": 1.4437, "step": 4083 }, { "epoch": 0.29, "grad_norm": 3.9492898700849963, "learning_rate": 8.295237229590123e-06, "loss": 0.2342, "step": 4084 }, { "epoch": 0.29, "grad_norm": 14.003948031362668, "learning_rate": 8.294363713196077e-06, "loss": 0.9724, "step": 4085 }, { "epoch": 0.29, "grad_norm": 3.4998014447477286, "learning_rate": 8.29349001908358e-06, "loss": 0.2364, "step": 4086 }, { "epoch": 0.29, "grad_norm": 4.301031342393336, "learning_rate": 8.292616147299763e-06, "loss": 0.2916, "step": 4087 }, { "epoch": 0.29, "grad_norm": 4.853184569382207, "learning_rate": 8.291742097891768e-06, "loss": 1.0164, "step": 4088 }, { "epoch": 0.29, "grad_norm": 5.247704472074128, "learning_rate": 8.290867870906746e-06, "loss": 1.1131, "step": 4089 }, { "epoch": 0.29, "grad_norm": 23.813185171597674, "learning_rate": 8.28999346639186e-06, "loss": 1.0399, "step": 4090 }, { "epoch": 0.29, "grad_norm": 17.55238092950328, "learning_rate": 8.289118884394279e-06, "loss": 1.0692, "step": 4091 }, { "epoch": 0.29, "grad_norm": 4.753645870137814, "learning_rate": 8.288244124961182e-06, "loss": 0.9519, "step": 4092 }, { "epoch": 0.29, "grad_norm": 3.448501234370902, "learning_rate": 8.287369188139765e-06, "loss": 0.2528, "step": 4093 }, { "epoch": 0.29, "grad_norm": 4.780902730147667, "learning_rate": 8.286494073977221e-06, "loss": 1.0012, "step": 4094 }, { "epoch": 0.29, "grad_norm": 3.574069133686736, "learning_rate": 8.285618782520763e-06, "loss": 0.2894, "step": 4095 }, { "epoch": 0.29, "grad_norm": 3.5690193814887823, "learning_rate": 8.284743313817609e-06, "loss": 0.3067, "step": 4096 }, { "epoch": 0.29, "grad_norm": 13.187709574239861, "learning_rate": 8.283867667914986e-06, "loss": 0.7886, "step": 4097 }, { "epoch": 0.29, "grad_norm": 27.080689787869606, "learning_rate": 8.28299184486013e-06, "loss": 1.1796, "step": 4098 }, { "epoch": 0.29, "grad_norm": 8.531241212255074, "learning_rate": 8.282115844700296e-06, "loss": 0.9876, "step": 4099 }, { "epoch": 0.29, "grad_norm": 2.7784730776052027, "learning_rate": 8.281239667482733e-06, "loss": 0.1995, "step": 4100 }, { "epoch": 0.29, "grad_norm": 3.0816287278929515, "learning_rate": 8.280363313254708e-06, "loss": 0.2022, "step": 4101 }, { "epoch": 0.29, "grad_norm": 27.161879322009078, "learning_rate": 8.279486782063502e-06, "loss": 1.1134, "step": 4102 }, { "epoch": 0.29, "grad_norm": 3.9182670288036303, "learning_rate": 8.278610073956396e-06, "loss": 0.3005, "step": 4103 }, { "epoch": 0.29, "grad_norm": 20.768887695717144, "learning_rate": 8.277733188980688e-06, "loss": 1.0137, "step": 4104 }, { "epoch": 0.29, "grad_norm": 3.5837502829249526, "learning_rate": 8.276856127183681e-06, "loss": 0.2699, "step": 4105 }, { "epoch": 0.29, "grad_norm": 3.029056294170075, "learning_rate": 8.275978888612688e-06, "loss": 0.2395, "step": 4106 }, { "epoch": 0.29, "grad_norm": 4.8941635333782205, "learning_rate": 8.275101473315037e-06, "loss": 0.3438, "step": 4107 }, { "epoch": 0.29, "grad_norm": 4.925863879862476, "learning_rate": 8.274223881338057e-06, "loss": 1.0227, "step": 4108 }, { "epoch": 0.29, "grad_norm": 3.407587846545786, "learning_rate": 8.27334611272909e-06, "loss": 0.2112, "step": 4109 }, { "epoch": 0.29, "grad_norm": 4.518311639000504, "learning_rate": 8.272468167535494e-06, "loss": 0.9965, "step": 4110 }, { "epoch": 0.29, "grad_norm": 13.583069223716334, "learning_rate": 8.271590045804624e-06, "loss": 0.8817, "step": 4111 }, { "epoch": 0.29, "grad_norm": 2.900386100509699, "learning_rate": 8.270711747583858e-06, "loss": 0.1655, "step": 4112 }, { "epoch": 0.29, "grad_norm": 13.381653766686119, "learning_rate": 8.269833272920572e-06, "loss": 0.9179, "step": 4113 }, { "epoch": 0.3, "grad_norm": 3.834339841237416, "learning_rate": 8.268954621862158e-06, "loss": 0.3567, "step": 4114 }, { "epoch": 0.3, "grad_norm": 3.2015925178823803, "learning_rate": 8.268075794456016e-06, "loss": 0.1883, "step": 4115 }, { "epoch": 0.3, "grad_norm": 3.510491209194888, "learning_rate": 8.267196790749555e-06, "loss": 0.1853, "step": 4116 }, { "epoch": 0.3, "grad_norm": 25.299889886101905, "learning_rate": 8.266317610790194e-06, "loss": 1.2143, "step": 4117 }, { "epoch": 0.3, "grad_norm": 4.364501738822559, "learning_rate": 8.265438254625362e-06, "loss": 0.211, "step": 4118 }, { "epoch": 0.3, "grad_norm": 12.29231094740549, "learning_rate": 8.264558722302496e-06, "loss": 0.8839, "step": 4119 }, { "epoch": 0.3, "grad_norm": 4.726081592653966, "learning_rate": 8.263679013869044e-06, "loss": 0.8738, "step": 4120 }, { "epoch": 0.3, "grad_norm": 4.86426442978126, "learning_rate": 8.262799129372464e-06, "loss": 0.9298, "step": 4121 }, { "epoch": 0.3, "grad_norm": 3.1351544978218358, "learning_rate": 8.261919068860221e-06, "loss": 0.2167, "step": 4122 }, { "epoch": 0.3, "grad_norm": 16.100594800418428, "learning_rate": 8.26103883237979e-06, "loss": 0.9885, "step": 4123 }, { "epoch": 0.3, "grad_norm": 3.6016111948307215, "learning_rate": 8.260158419978659e-06, "loss": 0.2185, "step": 4124 }, { "epoch": 0.3, "grad_norm": 4.511943525618138, "learning_rate": 8.25927783170432e-06, "loss": 0.3613, "step": 4125 }, { "epoch": 0.3, "grad_norm": 2.657030202961488, "learning_rate": 8.25839706760428e-06, "loss": 0.1893, "step": 4126 }, { "epoch": 0.3, "grad_norm": 3.8972820510951673, "learning_rate": 8.257516127726054e-06, "loss": 0.2631, "step": 4127 }, { "epoch": 0.3, "grad_norm": 40.129023223667915, "learning_rate": 8.25663501211716e-06, "loss": 1.1503, "step": 4128 }, { "epoch": 0.3, "grad_norm": 3.6886494048239853, "learning_rate": 8.255753720825134e-06, "loss": 0.2936, "step": 4129 }, { "epoch": 0.3, "grad_norm": 11.783098971484957, "learning_rate": 8.25487225389752e-06, "loss": 0.8338, "step": 4130 }, { "epoch": 0.3, "grad_norm": 11.450294423622214, "learning_rate": 8.25399061138187e-06, "loss": 0.6596, "step": 4131 }, { "epoch": 0.3, "grad_norm": 3.7537747948076166, "learning_rate": 8.253108793325742e-06, "loss": 0.2653, "step": 4132 }, { "epoch": 0.3, "grad_norm": 4.5003490460641915, "learning_rate": 8.252226799776706e-06, "loss": 1.0352, "step": 4133 }, { "epoch": 0.3, "grad_norm": 24.6729516115949, "learning_rate": 8.251344630782348e-06, "loss": 1.7179, "step": 4134 }, { "epoch": 0.3, "grad_norm": 3.9555603573124714, "learning_rate": 8.250462286390253e-06, "loss": 0.1911, "step": 4135 }, { "epoch": 0.3, "grad_norm": 14.270272518930414, "learning_rate": 8.24957976664802e-06, "loss": 0.8954, "step": 4136 }, { "epoch": 0.3, "grad_norm": 14.46757973614137, "learning_rate": 8.24869707160326e-06, "loss": 0.7497, "step": 4137 }, { "epoch": 0.3, "grad_norm": 5.575583033878565, "learning_rate": 8.24781420130359e-06, "loss": 0.9989, "step": 4138 }, { "epoch": 0.3, "grad_norm": 11.151495329377754, "learning_rate": 8.246931155796637e-06, "loss": 1.1518, "step": 4139 }, { "epoch": 0.3, "grad_norm": 5.6947291435657235, "learning_rate": 8.24604793513004e-06, "loss": 0.8836, "step": 4140 }, { "epoch": 0.3, "grad_norm": 3.9333020223977164, "learning_rate": 8.245164539351443e-06, "loss": 0.1869, "step": 4141 }, { "epoch": 0.3, "grad_norm": 5.039808601461706, "learning_rate": 8.2442809685085e-06, "loss": 1.1025, "step": 4142 }, { "epoch": 0.3, "grad_norm": 7.960119460645415, "learning_rate": 8.243397222648883e-06, "loss": 0.4993, "step": 4143 }, { "epoch": 0.3, "grad_norm": 3.147525895065272, "learning_rate": 8.24251330182026e-06, "loss": 0.1783, "step": 4144 }, { "epoch": 0.3, "grad_norm": 4.1885699084587875, "learning_rate": 8.24162920607032e-06, "loss": 0.9259, "step": 4145 }, { "epoch": 0.3, "grad_norm": 3.4967605782426268, "learning_rate": 8.240744935446752e-06, "loss": 0.1752, "step": 4146 }, { "epoch": 0.3, "grad_norm": 5.195846040535868, "learning_rate": 8.239860489997265e-06, "loss": 0.9119, "step": 4147 }, { "epoch": 0.3, "grad_norm": 74.19351208085337, "learning_rate": 8.238975869769565e-06, "loss": 1.1478, "step": 4148 }, { "epoch": 0.3, "grad_norm": 3.9697710893352283, "learning_rate": 8.23809107481138e-06, "loss": 0.2643, "step": 4149 }, { "epoch": 0.3, "grad_norm": 15.04876180695703, "learning_rate": 8.237206105170436e-06, "loss": 1.1481, "step": 4150 }, { "epoch": 0.3, "grad_norm": 4.982927039666577, "learning_rate": 8.236320960894478e-06, "loss": 0.9029, "step": 4151 }, { "epoch": 0.3, "grad_norm": 4.464844166559851, "learning_rate": 8.235435642031252e-06, "loss": 0.3363, "step": 4152 }, { "epoch": 0.3, "grad_norm": 3.7796789261326604, "learning_rate": 8.234550148628521e-06, "loss": 0.211, "step": 4153 }, { "epoch": 0.3, "grad_norm": 4.7040041796740475, "learning_rate": 8.233664480734054e-06, "loss": 1.0395, "step": 4154 }, { "epoch": 0.3, "grad_norm": 3.5647283296553653, "learning_rate": 8.232778638395627e-06, "loss": 0.2065, "step": 4155 }, { "epoch": 0.3, "grad_norm": 30.655046878224812, "learning_rate": 8.231892621661029e-06, "loss": 1.1338, "step": 4156 }, { "epoch": 0.3, "grad_norm": 30.344501578323147, "learning_rate": 8.231006430578058e-06, "loss": 2.056, "step": 4157 }, { "epoch": 0.3, "grad_norm": 4.678131087158774, "learning_rate": 8.23012006519452e-06, "loss": 0.2535, "step": 4158 }, { "epoch": 0.3, "grad_norm": 11.590389950222145, "learning_rate": 8.22923352555823e-06, "loss": 0.6774, "step": 4159 }, { "epoch": 0.3, "grad_norm": 3.927235198657844, "learning_rate": 8.228346811717018e-06, "loss": 0.2112, "step": 4160 }, { "epoch": 0.3, "grad_norm": 15.714470768357847, "learning_rate": 8.227459923718713e-06, "loss": 0.8534, "step": 4161 }, { "epoch": 0.3, "grad_norm": 2.946686850936666, "learning_rate": 8.226572861611162e-06, "loss": 0.1572, "step": 4162 }, { "epoch": 0.3, "grad_norm": 35.723772768097795, "learning_rate": 8.225685625442217e-06, "loss": 1.0595, "step": 4163 }, { "epoch": 0.3, "grad_norm": 17.986043591816696, "learning_rate": 8.224798215259743e-06, "loss": 1.0784, "step": 4164 }, { "epoch": 0.3, "grad_norm": 4.035520344319598, "learning_rate": 8.223910631111613e-06, "loss": 0.2892, "step": 4165 }, { "epoch": 0.3, "grad_norm": 12.17493791246933, "learning_rate": 8.223022873045708e-06, "loss": 0.9269, "step": 4166 }, { "epoch": 0.3, "grad_norm": 4.644786981831307, "learning_rate": 8.222134941109919e-06, "loss": 1.0706, "step": 4167 }, { "epoch": 0.3, "grad_norm": 3.3372037138885178, "learning_rate": 8.221246835352145e-06, "loss": 0.1975, "step": 4168 }, { "epoch": 0.3, "grad_norm": 14.257054202887357, "learning_rate": 8.220358555820296e-06, "loss": 0.6141, "step": 4169 }, { "epoch": 0.3, "grad_norm": 4.129853065540987, "learning_rate": 8.219470102562294e-06, "loss": 0.3117, "step": 4170 }, { "epoch": 0.3, "grad_norm": 3.230633334594431, "learning_rate": 8.218581475626067e-06, "loss": 0.1673, "step": 4171 }, { "epoch": 0.3, "grad_norm": 6.955599585951888, "learning_rate": 8.217692675059552e-06, "loss": 0.6231, "step": 4172 }, { "epoch": 0.3, "grad_norm": 2.561151484402475, "learning_rate": 8.216803700910698e-06, "loss": 0.1442, "step": 4173 }, { "epoch": 0.3, "grad_norm": 3.1102440806253364, "learning_rate": 8.215914553227459e-06, "loss": 0.1756, "step": 4174 }, { "epoch": 0.3, "grad_norm": 20.10095717155767, "learning_rate": 8.215025232057803e-06, "loss": 1.0975, "step": 4175 }, { "epoch": 0.3, "grad_norm": 4.235931836998976, "learning_rate": 8.214135737449704e-06, "loss": 0.2447, "step": 4176 }, { "epoch": 0.3, "grad_norm": 3.9885478641849437, "learning_rate": 8.21324606945115e-06, "loss": 0.272, "step": 4177 }, { "epoch": 0.3, "grad_norm": 4.6936747762269535, "learning_rate": 8.212356228110134e-06, "loss": 0.3454, "step": 4178 }, { "epoch": 0.3, "grad_norm": 3.41874110854345, "learning_rate": 8.211466213474657e-06, "loss": 0.1644, "step": 4179 }, { "epoch": 0.3, "grad_norm": 4.601991073318406, "learning_rate": 8.210576025592736e-06, "loss": 0.3133, "step": 4180 }, { "epoch": 0.3, "grad_norm": 13.44343757185444, "learning_rate": 8.20968566451239e-06, "loss": 0.8198, "step": 4181 }, { "epoch": 0.3, "grad_norm": 10.71325596340526, "learning_rate": 8.208795130281651e-06, "loss": 1.3701, "step": 4182 }, { "epoch": 0.3, "grad_norm": 47.87094327751903, "learning_rate": 8.20790442294856e-06, "loss": 1.6592, "step": 4183 }, { "epoch": 0.3, "grad_norm": 16.15307762526497, "learning_rate": 8.207013542561171e-06, "loss": 0.7129, "step": 4184 }, { "epoch": 0.3, "grad_norm": 2.9271251793348223, "learning_rate": 8.206122489167537e-06, "loss": 0.1649, "step": 4185 }, { "epoch": 0.3, "grad_norm": 4.422393928570067, "learning_rate": 8.205231262815733e-06, "loss": 0.265, "step": 4186 }, { "epoch": 0.3, "grad_norm": 4.898323969296549, "learning_rate": 8.204339863553834e-06, "loss": 0.3158, "step": 4187 }, { "epoch": 0.3, "grad_norm": 13.24347444319629, "learning_rate": 8.203448291429927e-06, "loss": 1.3117, "step": 4188 }, { "epoch": 0.3, "grad_norm": 3.272093483728161, "learning_rate": 8.202556546492113e-06, "loss": 0.202, "step": 4189 }, { "epoch": 0.3, "grad_norm": 20.780590051438548, "learning_rate": 8.201664628788494e-06, "loss": 1.2457, "step": 4190 }, { "epoch": 0.3, "grad_norm": 14.66554842612509, "learning_rate": 8.200772538367188e-06, "loss": 0.9938, "step": 4191 }, { "epoch": 0.3, "grad_norm": 8.650897257055435, "learning_rate": 8.199880275276319e-06, "loss": 0.5692, "step": 4192 }, { "epoch": 0.3, "grad_norm": 12.175477310804375, "learning_rate": 8.19898783956402e-06, "loss": 1.303, "step": 4193 }, { "epoch": 0.3, "grad_norm": 9.16670802432847, "learning_rate": 8.198095231278438e-06, "loss": 0.6343, "step": 4194 }, { "epoch": 0.3, "grad_norm": 5.4247595538043445, "learning_rate": 8.197202450467723e-06, "loss": 0.9753, "step": 4195 }, { "epoch": 0.3, "grad_norm": 5.700904815599983, "learning_rate": 8.196309497180038e-06, "loss": 0.8724, "step": 4196 }, { "epoch": 0.3, "grad_norm": 13.566103271230459, "learning_rate": 8.195416371463555e-06, "loss": 1.1136, "step": 4197 }, { "epoch": 0.3, "grad_norm": 3.7684591636713622, "learning_rate": 8.194523073366452e-06, "loss": 0.2768, "step": 4198 }, { "epoch": 0.3, "grad_norm": 4.647236746561421, "learning_rate": 8.193629602936922e-06, "loss": 0.995, "step": 4199 }, { "epoch": 0.3, "grad_norm": 4.116961698584988, "learning_rate": 8.192735960223164e-06, "loss": 0.2813, "step": 4200 }, { "epoch": 0.3, "grad_norm": 11.844613448393183, "learning_rate": 8.191842145273386e-06, "loss": 1.4223, "step": 4201 }, { "epoch": 0.3, "grad_norm": 24.75502961308866, "learning_rate": 8.190948158135806e-06, "loss": 1.1467, "step": 4202 }, { "epoch": 0.3, "grad_norm": 3.8512759704087833, "learning_rate": 8.19005399885865e-06, "loss": 0.1907, "step": 4203 }, { "epoch": 0.3, "grad_norm": 5.0713290201018575, "learning_rate": 8.189159667490159e-06, "loss": 1.1042, "step": 4204 }, { "epoch": 0.3, "grad_norm": 36.7366188958066, "learning_rate": 8.188265164078573e-06, "loss": 1.7901, "step": 4205 }, { "epoch": 0.3, "grad_norm": 13.042786021754438, "learning_rate": 8.187370488672149e-06, "loss": 0.9891, "step": 4206 }, { "epoch": 0.3, "grad_norm": 3.7160304499867407, "learning_rate": 8.186475641319153e-06, "loss": 0.3217, "step": 4207 }, { "epoch": 0.3, "grad_norm": 5.388446369883554, "learning_rate": 8.18558062206786e-06, "loss": 0.8531, "step": 4208 }, { "epoch": 0.3, "grad_norm": 3.0254788440489038, "learning_rate": 8.184685430966547e-06, "loss": 0.19, "step": 4209 }, { "epoch": 0.3, "grad_norm": 26.511212479703115, "learning_rate": 8.18379006806351e-06, "loss": 1.2014, "step": 4210 }, { "epoch": 0.3, "grad_norm": 3.621057048209442, "learning_rate": 8.182894533407051e-06, "loss": 0.2573, "step": 4211 }, { "epoch": 0.3, "grad_norm": 4.302849841443182, "learning_rate": 8.18199882704548e-06, "loss": 0.2769, "step": 4212 }, { "epoch": 0.3, "grad_norm": 10.736946005260423, "learning_rate": 8.181102949027116e-06, "loss": 1.1954, "step": 4213 }, { "epoch": 0.3, "grad_norm": 13.145355140943506, "learning_rate": 8.180206899400291e-06, "loss": 0.6461, "step": 4214 }, { "epoch": 0.3, "grad_norm": 3.576501545579592, "learning_rate": 8.17931067821334e-06, "loss": 0.247, "step": 4215 }, { "epoch": 0.3, "grad_norm": 20.2309924448999, "learning_rate": 8.178414285514613e-06, "loss": 1.3725, "step": 4216 }, { "epoch": 0.3, "grad_norm": 4.120001035121884, "learning_rate": 8.177517721352464e-06, "loss": 0.2427, "step": 4217 }, { "epoch": 0.3, "grad_norm": 3.880683605427927, "learning_rate": 8.176620985775264e-06, "loss": 0.1869, "step": 4218 }, { "epoch": 0.3, "grad_norm": 4.349868401369177, "learning_rate": 8.175724078831386e-06, "loss": 0.233, "step": 4219 }, { "epoch": 0.3, "grad_norm": 3.6229334444910632, "learning_rate": 8.174827000569214e-06, "loss": 0.1305, "step": 4220 }, { "epoch": 0.3, "grad_norm": 22.06896653311238, "learning_rate": 8.173929751037142e-06, "loss": 0.8515, "step": 4221 }, { "epoch": 0.3, "grad_norm": 4.745140045059064, "learning_rate": 8.173032330283574e-06, "loss": 0.3205, "step": 4222 }, { "epoch": 0.3, "grad_norm": 5.7795989439392645, "learning_rate": 8.172134738356923e-06, "loss": 0.957, "step": 4223 }, { "epoch": 0.3, "grad_norm": 4.147166618606368, "learning_rate": 8.17123697530561e-06, "loss": 0.3412, "step": 4224 }, { "epoch": 0.3, "grad_norm": 3.9486710035815227, "learning_rate": 8.170339041178067e-06, "loss": 0.2409, "step": 4225 }, { "epoch": 0.3, "grad_norm": 3.8673221817001555, "learning_rate": 8.169440936022731e-06, "loss": 0.2079, "step": 4226 }, { "epoch": 0.3, "grad_norm": 39.929700934270095, "learning_rate": 8.168542659888054e-06, "loss": 0.8986, "step": 4227 }, { "epoch": 0.3, "grad_norm": 3.5381111399251646, "learning_rate": 8.167644212822495e-06, "loss": 0.2044, "step": 4228 }, { "epoch": 0.3, "grad_norm": 3.4608943456327634, "learning_rate": 8.166745594874521e-06, "loss": 0.2021, "step": 4229 }, { "epoch": 0.3, "grad_norm": 5.295398261111294, "learning_rate": 8.165846806092607e-06, "loss": 1.1363, "step": 4230 }, { "epoch": 0.3, "grad_norm": 5.121104574559692, "learning_rate": 8.164947846525243e-06, "loss": 0.9239, "step": 4231 }, { "epoch": 0.3, "grad_norm": 3.753820193314077, "learning_rate": 8.164048716220922e-06, "loss": 0.224, "step": 4232 }, { "epoch": 0.3, "grad_norm": 9.969408994957858, "learning_rate": 8.163149415228152e-06, "loss": 0.6595, "step": 4233 }, { "epoch": 0.3, "grad_norm": 4.543620391099807, "learning_rate": 8.162249943595439e-06, "loss": 0.7831, "step": 4234 }, { "epoch": 0.3, "grad_norm": 4.971966489465689, "learning_rate": 8.161350301371315e-06, "loss": 1.0823, "step": 4235 }, { "epoch": 0.3, "grad_norm": 5.972013461588552, "learning_rate": 8.160450488604307e-06, "loss": 0.9567, "step": 4236 }, { "epoch": 0.3, "grad_norm": 13.771338854290578, "learning_rate": 8.159550505342959e-06, "loss": 1.4442, "step": 4237 }, { "epoch": 0.3, "grad_norm": 4.3209265783205355, "learning_rate": 8.158650351635821e-06, "loss": 0.2812, "step": 4238 }, { "epoch": 0.3, "grad_norm": 3.9685145168198037, "learning_rate": 8.157750027531452e-06, "loss": 0.31, "step": 4239 }, { "epoch": 0.3, "grad_norm": 3.290516921068645, "learning_rate": 8.156849533078422e-06, "loss": 0.1766, "step": 4240 }, { "epoch": 0.3, "grad_norm": 18.407984184755897, "learning_rate": 8.15594886832531e-06, "loss": 0.7839, "step": 4241 }, { "epoch": 0.3, "grad_norm": 15.142700271709469, "learning_rate": 8.155048033320701e-06, "loss": 0.7492, "step": 4242 }, { "epoch": 0.3, "grad_norm": 3.4406140668370537, "learning_rate": 8.154147028113194e-06, "loss": 0.2497, "step": 4243 }, { "epoch": 0.3, "grad_norm": 75.05961369447668, "learning_rate": 8.153245852751394e-06, "loss": 1.4403, "step": 4244 }, { "epoch": 0.3, "grad_norm": 5.236494708810347, "learning_rate": 8.152344507283914e-06, "loss": 0.8938, "step": 4245 }, { "epoch": 0.3, "grad_norm": 23.347130402669084, "learning_rate": 8.151442991759381e-06, "loss": 1.0936, "step": 4246 }, { "epoch": 0.3, "grad_norm": 3.936199938663366, "learning_rate": 8.150541306226429e-06, "loss": 0.2049, "step": 4247 }, { "epoch": 0.3, "grad_norm": 14.02493019017644, "learning_rate": 8.149639450733698e-06, "loss": 1.0327, "step": 4248 }, { "epoch": 0.3, "grad_norm": 3.829544334739744, "learning_rate": 8.14873742532984e-06, "loss": 0.2043, "step": 4249 }, { "epoch": 0.3, "grad_norm": 14.04112503354219, "learning_rate": 8.14783523006352e-06, "loss": 0.9351, "step": 4250 }, { "epoch": 0.3, "grad_norm": 4.575158072217435, "learning_rate": 8.1469328649834e-06, "loss": 0.8591, "step": 4251 }, { "epoch": 0.3, "grad_norm": 16.600999905345912, "learning_rate": 8.146030330138166e-06, "loss": 1.7322, "step": 4252 }, { "epoch": 0.3, "grad_norm": 15.633795887671646, "learning_rate": 8.145127625576502e-06, "loss": 0.4411, "step": 4253 }, { "epoch": 0.31, "grad_norm": 6.293755765035601, "learning_rate": 8.14422475134711e-06, "loss": 1.1219, "step": 4254 }, { "epoch": 0.31, "grad_norm": 4.5218638334219925, "learning_rate": 8.143321707498692e-06, "loss": 1.0351, "step": 4255 }, { "epoch": 0.31, "grad_norm": 12.931510941490824, "learning_rate": 8.142418494079969e-06, "loss": 0.937, "step": 4256 }, { "epoch": 0.31, "grad_norm": 3.671738816506961, "learning_rate": 8.14151511113966e-06, "loss": 0.2506, "step": 4257 }, { "epoch": 0.31, "grad_norm": 3.675608712101329, "learning_rate": 8.140611558726503e-06, "loss": 0.1934, "step": 4258 }, { "epoch": 0.31, "grad_norm": 4.410428169520217, "learning_rate": 8.139707836889242e-06, "loss": 0.8519, "step": 4259 }, { "epoch": 0.31, "grad_norm": 32.63772975267225, "learning_rate": 8.138803945676626e-06, "loss": 0.8548, "step": 4260 }, { "epoch": 0.31, "grad_norm": 19.458121344993387, "learning_rate": 8.137899885137417e-06, "loss": 0.988, "step": 4261 }, { "epoch": 0.31, "grad_norm": 4.828918201770102, "learning_rate": 8.136995655320389e-06, "loss": 0.9735, "step": 4262 }, { "epoch": 0.31, "grad_norm": 3.778987579218402, "learning_rate": 8.13609125627432e-06, "loss": 0.2461, "step": 4263 }, { "epoch": 0.31, "grad_norm": 4.730034324222002, "learning_rate": 8.135186688047995e-06, "loss": 0.8521, "step": 4264 }, { "epoch": 0.31, "grad_norm": 3.301545996418199, "learning_rate": 8.134281950690218e-06, "loss": 0.1691, "step": 4265 }, { "epoch": 0.31, "grad_norm": 3.5996983329388974, "learning_rate": 8.133377044249793e-06, "loss": 0.1696, "step": 4266 }, { "epoch": 0.31, "grad_norm": 6.998500947750497, "learning_rate": 8.132471968775537e-06, "loss": 1.1503, "step": 4267 }, { "epoch": 0.31, "grad_norm": 3.3648582873189294, "learning_rate": 8.131566724316276e-06, "loss": 0.1811, "step": 4268 }, { "epoch": 0.31, "grad_norm": 3.8786130541777095, "learning_rate": 8.130661310920843e-06, "loss": 0.2559, "step": 4269 }, { "epoch": 0.31, "grad_norm": 5.29769680851489, "learning_rate": 8.129755728638082e-06, "loss": 1.0014, "step": 4270 }, { "epoch": 0.31, "grad_norm": 4.544503922257582, "learning_rate": 8.128849977516848e-06, "loss": 0.3763, "step": 4271 }, { "epoch": 0.31, "grad_norm": 3.7626235806602266, "learning_rate": 8.127944057606e-06, "loss": 0.2199, "step": 4272 }, { "epoch": 0.31, "grad_norm": 17.92124707118956, "learning_rate": 8.12703796895441e-06, "loss": 1.2137, "step": 4273 }, { "epoch": 0.31, "grad_norm": 4.977482050994651, "learning_rate": 8.126131711610958e-06, "loss": 0.8938, "step": 4274 }, { "epoch": 0.31, "grad_norm": 6.825315447081463, "learning_rate": 8.125225285624534e-06, "loss": 0.2557, "step": 4275 }, { "epoch": 0.31, "grad_norm": 3.9074608264777497, "learning_rate": 8.124318691044036e-06, "loss": 0.1871, "step": 4276 }, { "epoch": 0.31, "grad_norm": 4.187471274194671, "learning_rate": 8.123411927918369e-06, "loss": 0.187, "step": 4277 }, { "epoch": 0.31, "grad_norm": 6.237632693033877, "learning_rate": 8.122504996296453e-06, "loss": 0.9035, "step": 4278 }, { "epoch": 0.31, "grad_norm": 19.412093219951515, "learning_rate": 8.121597896227208e-06, "loss": 1.1137, "step": 4279 }, { "epoch": 0.31, "grad_norm": 14.11980319875913, "learning_rate": 8.120690627759576e-06, "loss": 1.0215, "step": 4280 }, { "epoch": 0.31, "grad_norm": 4.193043700224426, "learning_rate": 8.119783190942497e-06, "loss": 0.2007, "step": 4281 }, { "epoch": 0.31, "grad_norm": 11.57563170261041, "learning_rate": 8.118875585824923e-06, "loss": 0.9332, "step": 4282 }, { "epoch": 0.31, "grad_norm": 3.240932265048769, "learning_rate": 8.117967812455816e-06, "loss": 0.2451, "step": 4283 }, { "epoch": 0.31, "grad_norm": 3.510249841366919, "learning_rate": 8.117059870884148e-06, "loss": 0.2291, "step": 4284 }, { "epoch": 0.31, "grad_norm": 3.332621941699944, "learning_rate": 8.116151761158898e-06, "loss": 0.2233, "step": 4285 }, { "epoch": 0.31, "grad_norm": 4.6394657356364615, "learning_rate": 8.115243483329059e-06, "loss": 0.426, "step": 4286 }, { "epoch": 0.31, "grad_norm": 4.087773164604455, "learning_rate": 8.114335037443624e-06, "loss": 0.3078, "step": 4287 }, { "epoch": 0.31, "grad_norm": 3.3454585495621583, "learning_rate": 8.113426423551602e-06, "loss": 0.2066, "step": 4288 }, { "epoch": 0.31, "grad_norm": 3.456096994017377, "learning_rate": 8.112517641702009e-06, "loss": 0.2104, "step": 4289 }, { "epoch": 0.31, "grad_norm": 3.0139648173176217, "learning_rate": 8.111608691943872e-06, "loss": 0.2145, "step": 4290 }, { "epoch": 0.31, "grad_norm": 3.231390993602078, "learning_rate": 8.110699574326222e-06, "loss": 0.1804, "step": 4291 }, { "epoch": 0.31, "grad_norm": 3.519036935882012, "learning_rate": 8.109790288898108e-06, "loss": 0.2714, "step": 4292 }, { "epoch": 0.31, "grad_norm": 6.175887019297188, "learning_rate": 8.108880835708574e-06, "loss": 1.0268, "step": 4293 }, { "epoch": 0.31, "grad_norm": 2.88764429804173, "learning_rate": 8.107971214806691e-06, "loss": 0.1272, "step": 4294 }, { "epoch": 0.31, "grad_norm": 5.289294141473225, "learning_rate": 8.107061426241523e-06, "loss": 0.8491, "step": 4295 }, { "epoch": 0.31, "grad_norm": 3.8432568172791917, "learning_rate": 8.106151470062154e-06, "loss": 0.301, "step": 4296 }, { "epoch": 0.31, "grad_norm": 2.983830249874769, "learning_rate": 8.105241346317668e-06, "loss": 0.2236, "step": 4297 }, { "epoch": 0.31, "grad_norm": 19.837628380720584, "learning_rate": 8.104331055057168e-06, "loss": 1.5014, "step": 4298 }, { "epoch": 0.31, "grad_norm": 15.236685471440355, "learning_rate": 8.103420596329756e-06, "loss": 1.0117, "step": 4299 }, { "epoch": 0.31, "grad_norm": 3.5074261316881548, "learning_rate": 8.102509970184552e-06, "loss": 0.2373, "step": 4300 }, { "epoch": 0.31, "grad_norm": 3.9653210905582266, "learning_rate": 8.101599176670676e-06, "loss": 0.1991, "step": 4301 }, { "epoch": 0.31, "grad_norm": 5.50540712848279, "learning_rate": 8.100688215837267e-06, "loss": 0.9107, "step": 4302 }, { "epoch": 0.31, "grad_norm": 3.305887480050254, "learning_rate": 8.099777087733464e-06, "loss": 0.154, "step": 4303 }, { "epoch": 0.31, "grad_norm": 10.988217719537989, "learning_rate": 8.098865792408422e-06, "loss": 0.7905, "step": 4304 }, { "epoch": 0.31, "grad_norm": 4.851605630628994, "learning_rate": 8.0979543299113e-06, "loss": 0.4389, "step": 4305 }, { "epoch": 0.31, "grad_norm": 4.101744281772184, "learning_rate": 8.097042700291266e-06, "loss": 0.236, "step": 4306 }, { "epoch": 0.31, "grad_norm": 4.588948771246663, "learning_rate": 8.096130903597503e-06, "loss": 0.8946, "step": 4307 }, { "epoch": 0.31, "grad_norm": 3.5747883835370096, "learning_rate": 8.095218939879198e-06, "loss": 0.1894, "step": 4308 }, { "epoch": 0.31, "grad_norm": 3.6251152623306457, "learning_rate": 8.094306809185546e-06, "loss": 0.2983, "step": 4309 }, { "epoch": 0.31, "grad_norm": 2.963436802044704, "learning_rate": 8.093394511565755e-06, "loss": 0.184, "step": 4310 }, { "epoch": 0.31, "grad_norm": 4.01940280728474, "learning_rate": 8.09248204706904e-06, "loss": 0.3197, "step": 4311 }, { "epoch": 0.31, "grad_norm": 20.6835874686248, "learning_rate": 8.091569415744622e-06, "loss": 1.5793, "step": 4312 }, { "epoch": 0.31, "grad_norm": 3.4583743904762634, "learning_rate": 8.090656617641738e-06, "loss": 0.1939, "step": 4313 }, { "epoch": 0.31, "grad_norm": 3.6204251144042803, "learning_rate": 8.089743652809626e-06, "loss": 0.2653, "step": 4314 }, { "epoch": 0.31, "grad_norm": 4.136007213481752, "learning_rate": 8.088830521297542e-06, "loss": 0.2507, "step": 4315 }, { "epoch": 0.31, "grad_norm": 3.8520997871149945, "learning_rate": 8.087917223154742e-06, "loss": 0.2102, "step": 4316 }, { "epoch": 0.31, "grad_norm": 5.845759604512922, "learning_rate": 8.087003758430496e-06, "loss": 0.9035, "step": 4317 }, { "epoch": 0.31, "grad_norm": 11.903226333247853, "learning_rate": 8.086090127174082e-06, "loss": 1.0571, "step": 4318 }, { "epoch": 0.31, "grad_norm": 3.59232320825685, "learning_rate": 8.085176329434787e-06, "loss": 0.2666, "step": 4319 }, { "epoch": 0.31, "grad_norm": 5.731722388249887, "learning_rate": 8.084262365261907e-06, "loss": 0.8922, "step": 4320 }, { "epoch": 0.31, "grad_norm": 4.2155827317296986, "learning_rate": 8.083348234704748e-06, "loss": 0.283, "step": 4321 }, { "epoch": 0.31, "grad_norm": 20.168058809629887, "learning_rate": 8.082433937812624e-06, "loss": 1.3431, "step": 4322 }, { "epoch": 0.31, "grad_norm": 9.529654073771017, "learning_rate": 8.081519474634855e-06, "loss": 0.538, "step": 4323 }, { "epoch": 0.31, "grad_norm": 3.498073046350503, "learning_rate": 8.080604845220774e-06, "loss": 0.3482, "step": 4324 }, { "epoch": 0.31, "grad_norm": 3.5794394357134935, "learning_rate": 8.079690049619725e-06, "loss": 0.2605, "step": 4325 }, { "epoch": 0.31, "grad_norm": 3.137466660960086, "learning_rate": 8.078775087881052e-06, "loss": 0.1503, "step": 4326 }, { "epoch": 0.31, "grad_norm": 14.673549444005932, "learning_rate": 8.077859960054118e-06, "loss": 0.8663, "step": 4327 }, { "epoch": 0.31, "grad_norm": 5.543028480286919, "learning_rate": 8.076944666188288e-06, "loss": 1.0326, "step": 4328 }, { "epoch": 0.31, "grad_norm": 2.8388391795123273, "learning_rate": 8.076029206332943e-06, "loss": 0.1781, "step": 4329 }, { "epoch": 0.31, "grad_norm": 4.985943306108271, "learning_rate": 8.075113580537463e-06, "loss": 0.9802, "step": 4330 }, { "epoch": 0.31, "grad_norm": 3.484141317095105, "learning_rate": 8.074197788851247e-06, "loss": 0.3017, "step": 4331 }, { "epoch": 0.31, "grad_norm": 3.816569138388383, "learning_rate": 8.073281831323695e-06, "loss": 0.199, "step": 4332 }, { "epoch": 0.31, "grad_norm": 3.6203130644409223, "learning_rate": 8.072365708004224e-06, "loss": 0.2251, "step": 4333 }, { "epoch": 0.31, "grad_norm": 4.7564454653770385, "learning_rate": 8.071449418942251e-06, "loss": 1.1995, "step": 4334 }, { "epoch": 0.31, "grad_norm": 19.785239666137848, "learning_rate": 8.070532964187208e-06, "loss": 1.0494, "step": 4335 }, { "epoch": 0.31, "grad_norm": 5.340754234488707, "learning_rate": 8.069616343788534e-06, "loss": 1.0515, "step": 4336 }, { "epoch": 0.31, "grad_norm": 3.5660627903645197, "learning_rate": 8.068699557795677e-06, "loss": 0.1727, "step": 4337 }, { "epoch": 0.31, "grad_norm": 3.8683848370469214, "learning_rate": 8.067782606258094e-06, "loss": 0.1871, "step": 4338 }, { "epoch": 0.31, "grad_norm": 35.16758961722644, "learning_rate": 8.066865489225254e-06, "loss": 1.0133, "step": 4339 }, { "epoch": 0.31, "grad_norm": 12.811540342396237, "learning_rate": 8.065948206746627e-06, "loss": 0.8283, "step": 4340 }, { "epoch": 0.31, "grad_norm": 16.215598237974397, "learning_rate": 8.065030758871702e-06, "loss": 1.0432, "step": 4341 }, { "epoch": 0.31, "grad_norm": 3.8429839300774296, "learning_rate": 8.064113145649968e-06, "loss": 0.3057, "step": 4342 }, { "epoch": 0.31, "grad_norm": 3.446762351270462, "learning_rate": 8.063195367130928e-06, "loss": 0.2041, "step": 4343 }, { "epoch": 0.31, "grad_norm": 10.163176852248611, "learning_rate": 8.062277423364094e-06, "loss": 0.834, "step": 4344 }, { "epoch": 0.31, "grad_norm": 5.538968383417999, "learning_rate": 8.061359314398982e-06, "loss": 0.8673, "step": 4345 }, { "epoch": 0.31, "grad_norm": 3.114955405688384, "learning_rate": 8.060441040285125e-06, "loss": 0.1779, "step": 4346 }, { "epoch": 0.31, "grad_norm": 4.183412935214845, "learning_rate": 8.059522601072057e-06, "loss": 1.1319, "step": 4347 }, { "epoch": 0.31, "grad_norm": 27.47019702664458, "learning_rate": 8.058603996809325e-06, "loss": 1.0424, "step": 4348 }, { "epoch": 0.31, "grad_norm": 15.24947636995908, "learning_rate": 8.057685227546487e-06, "loss": 1.1528, "step": 4349 }, { "epoch": 0.31, "grad_norm": 13.659058497689244, "learning_rate": 8.056766293333103e-06, "loss": 1.179, "step": 4350 }, { "epoch": 0.31, "grad_norm": 6.104855096519298, "learning_rate": 8.05584719421875e-06, "loss": 1.0398, "step": 4351 }, { "epoch": 0.31, "grad_norm": 4.320157124736112, "learning_rate": 8.054927930253005e-06, "loss": 1.1214, "step": 4352 }, { "epoch": 0.31, "grad_norm": 3.4942645532078878, "learning_rate": 8.054008501485465e-06, "loss": 0.1931, "step": 4353 }, { "epoch": 0.31, "grad_norm": 4.469019632902201, "learning_rate": 8.053088907965722e-06, "loss": 0.2734, "step": 4354 }, { "epoch": 0.31, "grad_norm": 7.196559440603233, "learning_rate": 8.052169149743393e-06, "loss": 0.7412, "step": 4355 }, { "epoch": 0.31, "grad_norm": 2.528753264619427, "learning_rate": 8.051249226868089e-06, "loss": 0.152, "step": 4356 }, { "epoch": 0.31, "grad_norm": 3.9908201183379406, "learning_rate": 8.050329139389442e-06, "loss": 0.4264, "step": 4357 }, { "epoch": 0.31, "grad_norm": 3.3999235917588573, "learning_rate": 8.04940888735708e-06, "loss": 0.2849, "step": 4358 }, { "epoch": 0.31, "grad_norm": 3.2210706782889647, "learning_rate": 8.048488470820655e-06, "loss": 0.1864, "step": 4359 }, { "epoch": 0.31, "grad_norm": 3.7340502307827763, "learning_rate": 8.047567889829814e-06, "loss": 0.2019, "step": 4360 }, { "epoch": 0.31, "grad_norm": 4.095705606276281, "learning_rate": 8.04664714443422e-06, "loss": 0.3885, "step": 4361 }, { "epoch": 0.31, "grad_norm": 4.624146175280227, "learning_rate": 8.045726234683547e-06, "loss": 0.3021, "step": 4362 }, { "epoch": 0.31, "grad_norm": 2.633418547929613, "learning_rate": 8.04480516062747e-06, "loss": 0.1531, "step": 4363 }, { "epoch": 0.31, "grad_norm": 34.22105301981577, "learning_rate": 8.043883922315683e-06, "loss": 1.1139, "step": 4364 }, { "epoch": 0.31, "grad_norm": 4.674946166869121, "learning_rate": 8.042962519797877e-06, "loss": 1.0156, "step": 4365 }, { "epoch": 0.31, "grad_norm": 9.159731142716735, "learning_rate": 8.042040953123764e-06, "loss": 0.5151, "step": 4366 }, { "epoch": 0.31, "grad_norm": 3.668525819007993, "learning_rate": 8.041119222343054e-06, "loss": 0.2934, "step": 4367 }, { "epoch": 0.31, "grad_norm": 2.5225071208740135, "learning_rate": 8.040197327505475e-06, "loss": 0.2129, "step": 4368 }, { "epoch": 0.31, "grad_norm": 4.3409981948864775, "learning_rate": 8.039275268660756e-06, "loss": 0.8132, "step": 4369 }, { "epoch": 0.31, "grad_norm": 41.33122837865549, "learning_rate": 8.038353045858642e-06, "loss": 1.3328, "step": 4370 }, { "epoch": 0.31, "grad_norm": 12.949960928206476, "learning_rate": 8.037430659148882e-06, "loss": 0.9923, "step": 4371 }, { "epoch": 0.31, "grad_norm": 4.836630545347667, "learning_rate": 8.036508108581235e-06, "loss": 0.9568, "step": 4372 }, { "epoch": 0.31, "grad_norm": 3.323587912503248, "learning_rate": 8.03558539420547e-06, "loss": 0.1912, "step": 4373 }, { "epoch": 0.31, "grad_norm": 4.548941787787227, "learning_rate": 8.034662516071361e-06, "loss": 0.1905, "step": 4374 }, { "epoch": 0.31, "grad_norm": 3.2131302893144604, "learning_rate": 8.033739474228696e-06, "loss": 0.1768, "step": 4375 }, { "epoch": 0.31, "grad_norm": 3.6930527417878523, "learning_rate": 8.032816268727273e-06, "loss": 0.2014, "step": 4376 }, { "epoch": 0.31, "grad_norm": 3.2087071560775424, "learning_rate": 8.031892899616891e-06, "loss": 0.2016, "step": 4377 }, { "epoch": 0.31, "grad_norm": 3.755410348982927, "learning_rate": 8.030969366947361e-06, "loss": 0.2566, "step": 4378 }, { "epoch": 0.31, "grad_norm": 27.985368034859512, "learning_rate": 8.030045670768509e-06, "loss": 1.0126, "step": 4379 }, { "epoch": 0.31, "grad_norm": 8.47395067025554, "learning_rate": 8.029121811130163e-06, "loss": 0.5644, "step": 4380 }, { "epoch": 0.31, "grad_norm": 3.4581906734602055, "learning_rate": 8.02819778808216e-06, "loss": 0.1933, "step": 4381 }, { "epoch": 0.31, "grad_norm": 4.140762511595161, "learning_rate": 8.027273601674346e-06, "loss": 0.3382, "step": 4382 }, { "epoch": 0.31, "grad_norm": 2.324872196849748, "learning_rate": 8.026349251956585e-06, "loss": 0.1277, "step": 4383 }, { "epoch": 0.31, "grad_norm": 2.9285673407799604, "learning_rate": 8.025424738978733e-06, "loss": 0.1481, "step": 4384 }, { "epoch": 0.31, "grad_norm": 5.555077443599472, "learning_rate": 8.02450006279067e-06, "loss": 0.2323, "step": 4385 }, { "epoch": 0.31, "grad_norm": 35.70173312831585, "learning_rate": 8.023575223442278e-06, "loss": 0.7296, "step": 4386 }, { "epoch": 0.31, "grad_norm": 3.3868693746623295, "learning_rate": 8.022650220983446e-06, "loss": 0.1544, "step": 4387 }, { "epoch": 0.31, "grad_norm": 3.6360781686528085, "learning_rate": 8.021725055464076e-06, "loss": 0.3309, "step": 4388 }, { "epoch": 0.31, "grad_norm": 5.499298746708726, "learning_rate": 8.020799726934079e-06, "loss": 0.8328, "step": 4389 }, { "epoch": 0.31, "grad_norm": 5.445408614604851, "learning_rate": 8.019874235443368e-06, "loss": 1.0928, "step": 4390 }, { "epoch": 0.31, "grad_norm": 3.27599287429893, "learning_rate": 8.018948581041874e-06, "loss": 0.1617, "step": 4391 }, { "epoch": 0.31, "grad_norm": 3.6463573753630967, "learning_rate": 8.018022763779533e-06, "loss": 0.2619, "step": 4392 }, { "epoch": 0.32, "grad_norm": 21.520894247324296, "learning_rate": 8.017096783706286e-06, "loss": 1.1798, "step": 4393 }, { "epoch": 0.32, "grad_norm": 3.9820769778498506, "learning_rate": 8.01617064087209e-06, "loss": 0.1961, "step": 4394 }, { "epoch": 0.32, "grad_norm": 4.863015276514836, "learning_rate": 8.015244335326901e-06, "loss": 0.8522, "step": 4395 }, { "epoch": 0.32, "grad_norm": 5.326802462234032, "learning_rate": 8.014317867120696e-06, "loss": 0.9444, "step": 4396 }, { "epoch": 0.32, "grad_norm": 14.169751612338496, "learning_rate": 8.013391236303451e-06, "loss": 0.544, "step": 4397 }, { "epoch": 0.32, "grad_norm": 20.332269042842903, "learning_rate": 8.012464442925156e-06, "loss": 0.8389, "step": 4398 }, { "epoch": 0.32, "grad_norm": 4.177851107916144, "learning_rate": 8.011537487035807e-06, "loss": 0.2797, "step": 4399 }, { "epoch": 0.32, "grad_norm": 13.08896141953075, "learning_rate": 8.01061036868541e-06, "loss": 1.372, "step": 4400 }, { "epoch": 0.32, "grad_norm": 16.85501037032431, "learning_rate": 8.009683087923979e-06, "loss": 1.5898, "step": 4401 }, { "epoch": 0.32, "grad_norm": 11.462489563307429, "learning_rate": 8.008755644801538e-06, "loss": 0.8053, "step": 4402 }, { "epoch": 0.32, "grad_norm": 11.085668930961623, "learning_rate": 8.007828039368118e-06, "loss": 1.1021, "step": 4403 }, { "epoch": 0.32, "grad_norm": 3.8656389719972895, "learning_rate": 8.00690027167376e-06, "loss": 0.2265, "step": 4404 }, { "epoch": 0.32, "grad_norm": 2.8744930626194463, "learning_rate": 8.005972341768515e-06, "loss": 0.1832, "step": 4405 }, { "epoch": 0.32, "grad_norm": 3.2538694577743223, "learning_rate": 8.00504424970244e-06, "loss": 0.2554, "step": 4406 }, { "epoch": 0.32, "grad_norm": 24.822691756285746, "learning_rate": 8.004115995525605e-06, "loss": 1.167, "step": 4407 }, { "epoch": 0.32, "grad_norm": 4.092886242568198, "learning_rate": 8.00318757928808e-06, "loss": 0.2265, "step": 4408 }, { "epoch": 0.32, "grad_norm": 4.600577582862276, "learning_rate": 8.002259001039954e-06, "loss": 0.2894, "step": 4409 }, { "epoch": 0.32, "grad_norm": 23.047682321994444, "learning_rate": 8.001330260831319e-06, "loss": 1.9879, "step": 4410 }, { "epoch": 0.32, "grad_norm": 3.885685747033397, "learning_rate": 8.000401358712277e-06, "loss": 0.1885, "step": 4411 }, { "epoch": 0.32, "grad_norm": 4.050213977707369, "learning_rate": 7.999472294732938e-06, "loss": 0.2391, "step": 4412 }, { "epoch": 0.32, "grad_norm": 19.48798500179324, "learning_rate": 7.998543068943422e-06, "loss": 1.4088, "step": 4413 }, { "epoch": 0.32, "grad_norm": 4.440696528518378, "learning_rate": 7.99761368139386e-06, "loss": 0.3089, "step": 4414 }, { "epoch": 0.32, "grad_norm": 4.461933860427275, "learning_rate": 7.996684132134384e-06, "loss": 0.8297, "step": 4415 }, { "epoch": 0.32, "grad_norm": 12.659925478124697, "learning_rate": 7.995754421215144e-06, "loss": 0.8611, "step": 4416 }, { "epoch": 0.32, "grad_norm": 5.804119350893872, "learning_rate": 7.99482454868629e-06, "loss": 1.0208, "step": 4417 }, { "epoch": 0.32, "grad_norm": 5.127457046321896, "learning_rate": 7.993894514597991e-06, "loss": 0.9395, "step": 4418 }, { "epoch": 0.32, "grad_norm": 3.7778179481110863, "learning_rate": 7.992964319000414e-06, "loss": 0.2743, "step": 4419 }, { "epoch": 0.32, "grad_norm": 5.498868395421986, "learning_rate": 7.992033961943742e-06, "loss": 1.0373, "step": 4420 }, { "epoch": 0.32, "grad_norm": 11.815266423521843, "learning_rate": 7.99110344347816e-06, "loss": 0.8841, "step": 4421 }, { "epoch": 0.32, "grad_norm": 20.690730020892797, "learning_rate": 7.990172763653872e-06, "loss": 1.3561, "step": 4422 }, { "epoch": 0.32, "grad_norm": 3.711603514459684, "learning_rate": 7.989241922521081e-06, "loss": 0.1446, "step": 4423 }, { "epoch": 0.32, "grad_norm": 3.9922446594870187, "learning_rate": 7.988310920130002e-06, "loss": 0.2273, "step": 4424 }, { "epoch": 0.32, "grad_norm": 3.853140851521321, "learning_rate": 7.987379756530862e-06, "loss": 0.2356, "step": 4425 }, { "epoch": 0.32, "grad_norm": 15.398325813290448, "learning_rate": 7.986448431773893e-06, "loss": 1.1623, "step": 4426 }, { "epoch": 0.32, "grad_norm": 4.5045759336278275, "learning_rate": 7.985516945909333e-06, "loss": 0.3512, "step": 4427 }, { "epoch": 0.32, "grad_norm": 4.418833211818708, "learning_rate": 7.984585298987435e-06, "loss": 0.8013, "step": 4428 }, { "epoch": 0.32, "grad_norm": 7.590002496194331, "learning_rate": 7.983653491058457e-06, "loss": 0.267, "step": 4429 }, { "epoch": 0.32, "grad_norm": 23.02535611230473, "learning_rate": 7.982721522172667e-06, "loss": 0.9938, "step": 4430 }, { "epoch": 0.32, "grad_norm": 13.190792843975993, "learning_rate": 7.981789392380344e-06, "loss": 0.9892, "step": 4431 }, { "epoch": 0.32, "grad_norm": 4.617068379634271, "learning_rate": 7.980857101731766e-06, "loss": 0.9099, "step": 4432 }, { "epoch": 0.32, "grad_norm": 4.593950910224185, "learning_rate": 7.979924650277233e-06, "loss": 0.2588, "step": 4433 }, { "epoch": 0.32, "grad_norm": 3.1933055306340625, "learning_rate": 7.978992038067044e-06, "loss": 0.1608, "step": 4434 }, { "epoch": 0.32, "grad_norm": 3.6106594122650804, "learning_rate": 7.97805926515151e-06, "loss": 0.1684, "step": 4435 }, { "epoch": 0.32, "grad_norm": 3.436685443796199, "learning_rate": 7.97712633158095e-06, "loss": 0.22, "step": 4436 }, { "epoch": 0.32, "grad_norm": 3.8182201592051563, "learning_rate": 7.976193237405699e-06, "loss": 0.2127, "step": 4437 }, { "epoch": 0.32, "grad_norm": 3.6551833375758074, "learning_rate": 7.975259982676083e-06, "loss": 0.2065, "step": 4438 }, { "epoch": 0.32, "grad_norm": 18.329878707135272, "learning_rate": 7.974326567442454e-06, "loss": 1.1479, "step": 4439 }, { "epoch": 0.32, "grad_norm": 4.050499925381428, "learning_rate": 7.973392991755167e-06, "loss": 0.2617, "step": 4440 }, { "epoch": 0.32, "grad_norm": 4.284774378742361, "learning_rate": 7.97245925566458e-06, "loss": 0.2967, "step": 4441 }, { "epoch": 0.32, "grad_norm": 4.201570115803305, "learning_rate": 7.97152535922107e-06, "loss": 0.2213, "step": 4442 }, { "epoch": 0.32, "grad_norm": 3.8929585503064494, "learning_rate": 7.970591302475015e-06, "loss": 0.234, "step": 4443 }, { "epoch": 0.32, "grad_norm": 15.798083869102294, "learning_rate": 7.969657085476802e-06, "loss": 1.3439, "step": 4444 }, { "epoch": 0.32, "grad_norm": 4.371678734551295, "learning_rate": 7.96872270827683e-06, "loss": 0.2279, "step": 4445 }, { "epoch": 0.32, "grad_norm": 4.361745899977507, "learning_rate": 7.967788170925505e-06, "loss": 0.2098, "step": 4446 }, { "epoch": 0.32, "grad_norm": 14.89771076153433, "learning_rate": 7.966853473473242e-06, "loss": 0.8988, "step": 4447 }, { "epoch": 0.32, "grad_norm": 3.916264625720098, "learning_rate": 7.965918615970466e-06, "loss": 0.2632, "step": 4448 }, { "epoch": 0.32, "grad_norm": 3.5223901706676704, "learning_rate": 7.964983598467606e-06, "loss": 0.1869, "step": 4449 }, { "epoch": 0.32, "grad_norm": 4.840022357475904, "learning_rate": 7.964048421015102e-06, "loss": 0.9141, "step": 4450 }, { "epoch": 0.32, "grad_norm": 4.453418725377098, "learning_rate": 7.963113083663408e-06, "loss": 0.3204, "step": 4451 }, { "epoch": 0.32, "grad_norm": 3.0735515023655084, "learning_rate": 7.962177586462978e-06, "loss": 0.1336, "step": 4452 }, { "epoch": 0.32, "grad_norm": 4.29237431180832, "learning_rate": 7.961241929464279e-06, "loss": 0.3485, "step": 4453 }, { "epoch": 0.32, "grad_norm": 20.123332131610965, "learning_rate": 7.960306112717788e-06, "loss": 1.2759, "step": 4454 }, { "epoch": 0.32, "grad_norm": 16.617119209114378, "learning_rate": 7.959370136273988e-06, "loss": 1.4273, "step": 4455 }, { "epoch": 0.32, "grad_norm": 13.823074465969565, "learning_rate": 7.95843400018337e-06, "loss": 1.0291, "step": 4456 }, { "epoch": 0.32, "grad_norm": 3.449247271750496, "learning_rate": 7.957497704496436e-06, "loss": 0.1883, "step": 4457 }, { "epoch": 0.32, "grad_norm": 3.2740743012305042, "learning_rate": 7.956561249263696e-06, "loss": 0.1937, "step": 4458 }, { "epoch": 0.32, "grad_norm": 4.615054882920035, "learning_rate": 7.955624634535669e-06, "loss": 0.2569, "step": 4459 }, { "epoch": 0.32, "grad_norm": 19.122493987029596, "learning_rate": 7.954687860362877e-06, "loss": 1.2191, "step": 4460 }, { "epoch": 0.32, "grad_norm": 3.5993027163843228, "learning_rate": 7.953750926795863e-06, "loss": 0.1596, "step": 4461 }, { "epoch": 0.32, "grad_norm": 4.861622406086245, "learning_rate": 7.952813833885165e-06, "loss": 0.3268, "step": 4462 }, { "epoch": 0.32, "grad_norm": 5.063072966254527, "learning_rate": 7.951876581681337e-06, "loss": 0.8929, "step": 4463 }, { "epoch": 0.32, "grad_norm": 3.8128664611531966, "learning_rate": 7.95093917023494e-06, "loss": 0.1709, "step": 4464 }, { "epoch": 0.32, "grad_norm": 3.7829435084167984, "learning_rate": 7.950001599596549e-06, "loss": 0.2465, "step": 4465 }, { "epoch": 0.32, "grad_norm": 4.399891026354249, "learning_rate": 7.949063869816736e-06, "loss": 0.2093, "step": 4466 }, { "epoch": 0.32, "grad_norm": 16.60752994527519, "learning_rate": 7.94812598094609e-06, "loss": 1.0127, "step": 4467 }, { "epoch": 0.32, "grad_norm": 3.9490470590363578, "learning_rate": 7.947187933035206e-06, "loss": 0.3175, "step": 4468 }, { "epoch": 0.32, "grad_norm": 3.3967750694982883, "learning_rate": 7.946249726134687e-06, "loss": 0.2122, "step": 4469 }, { "epoch": 0.32, "grad_norm": 13.874450261016058, "learning_rate": 7.945311360295151e-06, "loss": 0.957, "step": 4470 }, { "epoch": 0.32, "grad_norm": 3.9794528691852524, "learning_rate": 7.944372835567213e-06, "loss": 0.2816, "step": 4471 }, { "epoch": 0.32, "grad_norm": 4.502390902318126, "learning_rate": 7.943434152001506e-06, "loss": 0.3621, "step": 4472 }, { "epoch": 0.32, "grad_norm": 3.8608483971110834, "learning_rate": 7.94249530964867e-06, "loss": 0.1903, "step": 4473 }, { "epoch": 0.32, "grad_norm": 4.499040052348915, "learning_rate": 7.941556308559349e-06, "loss": 0.2092, "step": 4474 }, { "epoch": 0.32, "grad_norm": 3.6133422873303735, "learning_rate": 7.940617148784198e-06, "loss": 0.2465, "step": 4475 }, { "epoch": 0.32, "grad_norm": 3.5363926630810525, "learning_rate": 7.939677830373884e-06, "loss": 0.2532, "step": 4476 }, { "epoch": 0.32, "grad_norm": 3.836992127498262, "learning_rate": 7.938738353379079e-06, "loss": 0.2415, "step": 4477 }, { "epoch": 0.32, "grad_norm": 3.883617836443506, "learning_rate": 7.937798717850462e-06, "loss": 0.2098, "step": 4478 }, { "epoch": 0.32, "grad_norm": 24.61173406088024, "learning_rate": 7.936858923838726e-06, "loss": 1.2198, "step": 4479 }, { "epoch": 0.32, "grad_norm": 16.16234067975995, "learning_rate": 7.935918971394566e-06, "loss": 1.4421, "step": 4480 }, { "epoch": 0.32, "grad_norm": 17.131837461122984, "learning_rate": 7.934978860568692e-06, "loss": 1.1647, "step": 4481 }, { "epoch": 0.32, "grad_norm": 4.896601777866347, "learning_rate": 7.934038591411817e-06, "loss": 0.8365, "step": 4482 }, { "epoch": 0.32, "grad_norm": 3.8557623891989925, "learning_rate": 7.933098163974667e-06, "loss": 0.2528, "step": 4483 }, { "epoch": 0.32, "grad_norm": 4.085543433933361, "learning_rate": 7.932157578307971e-06, "loss": 0.2259, "step": 4484 }, { "epoch": 0.32, "grad_norm": 5.172375689873203, "learning_rate": 7.931216834462475e-06, "loss": 1.0351, "step": 4485 }, { "epoch": 0.32, "grad_norm": 12.027098289416566, "learning_rate": 7.930275932488925e-06, "loss": 0.7188, "step": 4486 }, { "epoch": 0.32, "grad_norm": 18.38005100157309, "learning_rate": 7.929334872438081e-06, "loss": 1.5973, "step": 4487 }, { "epoch": 0.32, "grad_norm": 4.56916310619876, "learning_rate": 7.928393654360708e-06, "loss": 0.82, "step": 4488 }, { "epoch": 0.32, "grad_norm": 15.430493333246398, "learning_rate": 7.927452278307581e-06, "loss": 0.6548, "step": 4489 }, { "epoch": 0.32, "grad_norm": 4.057945044844267, "learning_rate": 7.926510744329486e-06, "loss": 0.7215, "step": 4490 }, { "epoch": 0.32, "grad_norm": 19.0863113442018, "learning_rate": 7.925569052477214e-06, "loss": 0.737, "step": 4491 }, { "epoch": 0.32, "grad_norm": 17.47999166503429, "learning_rate": 7.924627202801564e-06, "loss": 1.2766, "step": 4492 }, { "epoch": 0.32, "grad_norm": 8.058640980641247, "learning_rate": 7.923685195353347e-06, "loss": 0.4125, "step": 4493 }, { "epoch": 0.32, "grad_norm": 24.540405363810894, "learning_rate": 7.922743030183383e-06, "loss": 0.8944, "step": 4494 }, { "epoch": 0.32, "grad_norm": 3.367960630690193, "learning_rate": 7.921800707342494e-06, "loss": 0.173, "step": 4495 }, { "epoch": 0.32, "grad_norm": 19.20444045092211, "learning_rate": 7.920858226881515e-06, "loss": 1.4598, "step": 4496 }, { "epoch": 0.32, "grad_norm": 3.844413485523745, "learning_rate": 7.919915588851292e-06, "loss": 0.2099, "step": 4497 }, { "epoch": 0.32, "grad_norm": 5.19091017016505, "learning_rate": 7.918972793302674e-06, "loss": 0.2517, "step": 4498 }, { "epoch": 0.32, "grad_norm": 6.826876343015901, "learning_rate": 7.918029840286525e-06, "loss": 0.663, "step": 4499 }, { "epoch": 0.32, "grad_norm": 3.017792402497853, "learning_rate": 7.91708672985371e-06, "loss": 0.1863, "step": 4500 }, { "epoch": 0.32, "grad_norm": 3.2001624933738952, "learning_rate": 7.916143462055109e-06, "loss": 0.2986, "step": 4501 }, { "epoch": 0.32, "grad_norm": 2.2307402269162226, "learning_rate": 7.915200036941606e-06, "loss": 0.1675, "step": 4502 }, { "epoch": 0.32, "grad_norm": 9.780429127546704, "learning_rate": 7.914256454564095e-06, "loss": 1.0369, "step": 4503 }, { "epoch": 0.32, "grad_norm": 17.02939870418041, "learning_rate": 7.913312714973481e-06, "loss": 0.7253, "step": 4504 }, { "epoch": 0.32, "grad_norm": 3.665592025399258, "learning_rate": 7.91236881822067e-06, "loss": 0.2615, "step": 4505 }, { "epoch": 0.32, "grad_norm": 4.101490212265181, "learning_rate": 7.911424764356588e-06, "loss": 0.2522, "step": 4506 }, { "epoch": 0.32, "grad_norm": 3.087966566741037, "learning_rate": 7.91048055343216e-06, "loss": 0.2108, "step": 4507 }, { "epoch": 0.32, "grad_norm": 3.1935555164178853, "learning_rate": 7.909536185498323e-06, "loss": 0.2353, "step": 4508 }, { "epoch": 0.32, "grad_norm": 15.128756270593804, "learning_rate": 7.908591660606021e-06, "loss": 1.1872, "step": 4509 }, { "epoch": 0.32, "grad_norm": 3.8568795900136172, "learning_rate": 7.90764697880621e-06, "loss": 0.2727, "step": 4510 }, { "epoch": 0.32, "grad_norm": 4.973893409950621, "learning_rate": 7.906702140149848e-06, "loss": 1.1832, "step": 4511 }, { "epoch": 0.32, "grad_norm": 4.883704861413234, "learning_rate": 7.90575714468791e-06, "loss": 1.1088, "step": 4512 }, { "epoch": 0.32, "grad_norm": 18.957136385657254, "learning_rate": 7.904811992471373e-06, "loss": 0.6513, "step": 4513 }, { "epoch": 0.32, "grad_norm": 17.937643681660056, "learning_rate": 7.903866683551224e-06, "loss": 0.8064, "step": 4514 }, { "epoch": 0.32, "grad_norm": 20.120896427920368, "learning_rate": 7.902921217978458e-06, "loss": 1.1794, "step": 4515 }, { "epoch": 0.32, "grad_norm": 4.010286735364235, "learning_rate": 7.901975595804083e-06, "loss": 0.2624, "step": 4516 }, { "epoch": 0.32, "grad_norm": 4.652828358598974, "learning_rate": 7.901029817079106e-06, "loss": 0.3641, "step": 4517 }, { "epoch": 0.32, "grad_norm": 3.6770986436065827, "learning_rate": 7.900083881854554e-06, "loss": 0.1873, "step": 4518 }, { "epoch": 0.32, "grad_norm": 13.117412598358474, "learning_rate": 7.899137790181453e-06, "loss": 0.7747, "step": 4519 }, { "epoch": 0.32, "grad_norm": 12.724602323051709, "learning_rate": 7.898191542110841e-06, "loss": 1.1162, "step": 4520 }, { "epoch": 0.32, "grad_norm": 3.43537868745202, "learning_rate": 7.897245137693768e-06, "loss": 0.2358, "step": 4521 }, { "epoch": 0.32, "grad_norm": 4.846566840614456, "learning_rate": 7.896298576981285e-06, "loss": 1.0146, "step": 4522 }, { "epoch": 0.32, "grad_norm": 4.792330698282658, "learning_rate": 7.895351860024456e-06, "loss": 0.8284, "step": 4523 }, { "epoch": 0.32, "grad_norm": 3.2344974317020285, "learning_rate": 7.894404986874356e-06, "loss": 0.2042, "step": 4524 }, { "epoch": 0.32, "grad_norm": 3.038096517057507, "learning_rate": 7.89345795758206e-06, "loss": 0.2214, "step": 4525 }, { "epoch": 0.32, "grad_norm": 22.46892573003507, "learning_rate": 7.89251077219866e-06, "loss": 1.5463, "step": 4526 }, { "epoch": 0.32, "grad_norm": 3.742065263447551, "learning_rate": 7.891563430775256e-06, "loss": 0.2649, "step": 4527 }, { "epoch": 0.32, "grad_norm": 3.622593620678707, "learning_rate": 7.890615933362946e-06, "loss": 0.2404, "step": 4528 }, { "epoch": 0.32, "grad_norm": 5.775532210619643, "learning_rate": 7.88966828001285e-06, "loss": 1.1149, "step": 4529 }, { "epoch": 0.32, "grad_norm": 7.0924473561938814, "learning_rate": 7.888720470776088e-06, "loss": 0.8771, "step": 4530 }, { "epoch": 0.32, "grad_norm": 3.749487606066688, "learning_rate": 7.88777250570379e-06, "loss": 0.2149, "step": 4531 }, { "epoch": 0.32, "grad_norm": 3.210592253604511, "learning_rate": 7.886824384847098e-06, "loss": 0.1343, "step": 4532 }, { "epoch": 0.33, "grad_norm": 4.438720257008098, "learning_rate": 7.885876108257158e-06, "loss": 0.926, "step": 4533 }, { "epoch": 0.33, "grad_norm": 5.68635692219637, "learning_rate": 7.884927675985123e-06, "loss": 1.0416, "step": 4534 }, { "epoch": 0.33, "grad_norm": 20.841994846614515, "learning_rate": 7.88397908808216e-06, "loss": 1.2226, "step": 4535 }, { "epoch": 0.33, "grad_norm": 4.759289315134515, "learning_rate": 7.883030344599443e-06, "loss": 0.9951, "step": 4536 }, { "epoch": 0.33, "grad_norm": 15.584315551993996, "learning_rate": 7.882081445588152e-06, "loss": 0.9361, "step": 4537 }, { "epoch": 0.33, "grad_norm": 5.106072681371243, "learning_rate": 7.881132391099476e-06, "loss": 1.0234, "step": 4538 }, { "epoch": 0.33, "grad_norm": 3.7649042664664423, "learning_rate": 7.880183181184614e-06, "loss": 0.2196, "step": 4539 }, { "epoch": 0.33, "grad_norm": 31.66798685438568, "learning_rate": 7.879233815894772e-06, "loss": 0.7346, "step": 4540 }, { "epoch": 0.33, "grad_norm": 3.9490969095625648, "learning_rate": 7.87828429528116e-06, "loss": 0.2234, "step": 4541 }, { "epoch": 0.33, "grad_norm": 3.4668642509688086, "learning_rate": 7.877334619395009e-06, "loss": 0.1971, "step": 4542 }, { "epoch": 0.33, "grad_norm": 4.695967912590851, "learning_rate": 7.876384788287546e-06, "loss": 0.9008, "step": 4543 }, { "epoch": 0.33, "grad_norm": 4.287008108833107, "learning_rate": 7.87543480201001e-06, "loss": 0.242, "step": 4544 }, { "epoch": 0.33, "grad_norm": 35.30052122191297, "learning_rate": 7.874484660613652e-06, "loss": 1.1591, "step": 4545 }, { "epoch": 0.33, "grad_norm": 3.921502920662389, "learning_rate": 7.873534364149726e-06, "loss": 0.2444, "step": 4546 }, { "epoch": 0.33, "grad_norm": 3.7602182046339214, "learning_rate": 7.872583912669498e-06, "loss": 0.1474, "step": 4547 }, { "epoch": 0.33, "grad_norm": 3.5987649123646768, "learning_rate": 7.871633306224241e-06, "loss": 0.2435, "step": 4548 }, { "epoch": 0.33, "grad_norm": 4.855824311218067, "learning_rate": 7.870682544865237e-06, "loss": 0.298, "step": 4549 }, { "epoch": 0.33, "grad_norm": 21.135732171471332, "learning_rate": 7.869731628643777e-06, "loss": 0.9656, "step": 4550 }, { "epoch": 0.33, "grad_norm": 18.773446549811595, "learning_rate": 7.86878055761116e-06, "loss": 1.3782, "step": 4551 }, { "epoch": 0.33, "grad_norm": 5.538456540758223, "learning_rate": 7.867829331818686e-06, "loss": 1.2624, "step": 4552 }, { "epoch": 0.33, "grad_norm": 3.9680721893645674, "learning_rate": 7.866877951317678e-06, "loss": 0.1672, "step": 4553 }, { "epoch": 0.33, "grad_norm": 46.733107935769794, "learning_rate": 7.865926416159455e-06, "loss": 0.7008, "step": 4554 }, { "epoch": 0.33, "grad_norm": 4.876417563212222, "learning_rate": 7.864974726395351e-06, "loss": 0.9526, "step": 4555 }, { "epoch": 0.33, "grad_norm": 4.6648504592947, "learning_rate": 7.864022882076704e-06, "loss": 1.0264, "step": 4556 }, { "epoch": 0.33, "grad_norm": 3.496004385333202, "learning_rate": 7.863070883254863e-06, "loss": 0.1205, "step": 4557 }, { "epoch": 0.33, "grad_norm": 4.849712176227304, "learning_rate": 7.862118729981188e-06, "loss": 1.063, "step": 4558 }, { "epoch": 0.33, "grad_norm": 22.561876512579502, "learning_rate": 7.861166422307038e-06, "loss": 0.7875, "step": 4559 }, { "epoch": 0.33, "grad_norm": 16.769791826398663, "learning_rate": 7.860213960283792e-06, "loss": 0.6555, "step": 4560 }, { "epoch": 0.33, "grad_norm": 3.605869618545882, "learning_rate": 7.859261343962827e-06, "loss": 0.1919, "step": 4561 }, { "epoch": 0.33, "grad_norm": 5.28528609773345, "learning_rate": 7.858308573395536e-06, "loss": 0.6763, "step": 4562 }, { "epoch": 0.33, "grad_norm": 15.696415818773906, "learning_rate": 7.857355648633318e-06, "loss": 0.7336, "step": 4563 }, { "epoch": 0.33, "grad_norm": 5.134122169954804, "learning_rate": 7.856402569727578e-06, "loss": 1.0391, "step": 4564 }, { "epoch": 0.33, "grad_norm": 17.359111162877067, "learning_rate": 7.855449336729732e-06, "loss": 0.7493, "step": 4565 }, { "epoch": 0.33, "grad_norm": 4.621351998508689, "learning_rate": 7.854495949691201e-06, "loss": 0.9774, "step": 4566 }, { "epoch": 0.33, "grad_norm": 3.271826698343172, "learning_rate": 7.853542408663419e-06, "loss": 0.2086, "step": 4567 }, { "epoch": 0.33, "grad_norm": 14.261350803571258, "learning_rate": 7.852588713697826e-06, "loss": 0.6811, "step": 4568 }, { "epoch": 0.33, "grad_norm": 9.501884605311691, "learning_rate": 7.851634864845868e-06, "loss": 0.9495, "step": 4569 }, { "epoch": 0.33, "grad_norm": 5.642321179637739, "learning_rate": 7.850680862159005e-06, "loss": 0.9055, "step": 4570 }, { "epoch": 0.33, "grad_norm": 20.258359023880782, "learning_rate": 7.8497267056887e-06, "loss": 0.9421, "step": 4571 }, { "epoch": 0.33, "grad_norm": 5.053480585283848, "learning_rate": 7.848772395486424e-06, "loss": 0.7587, "step": 4572 }, { "epoch": 0.33, "grad_norm": 4.0851459547927105, "learning_rate": 7.847817931603662e-06, "loss": 0.266, "step": 4573 }, { "epoch": 0.33, "grad_norm": 4.626564652045534, "learning_rate": 7.846863314091902e-06, "loss": 0.8005, "step": 4574 }, { "epoch": 0.33, "grad_norm": 3.726635197407819, "learning_rate": 7.845908543002642e-06, "loss": 0.2119, "step": 4575 }, { "epoch": 0.33, "grad_norm": 3.7353495829258416, "learning_rate": 7.844953618387388e-06, "loss": 0.2293, "step": 4576 }, { "epoch": 0.33, "grad_norm": 4.0470087906911, "learning_rate": 7.843998540297654e-06, "loss": 0.2515, "step": 4577 }, { "epoch": 0.33, "grad_norm": 3.310324630091876, "learning_rate": 7.843043308784967e-06, "loss": 0.1774, "step": 4578 }, { "epoch": 0.33, "grad_norm": 2.7994889871449997, "learning_rate": 7.842087923900854e-06, "loss": 0.1229, "step": 4579 }, { "epoch": 0.33, "grad_norm": 4.269882909188061, "learning_rate": 7.841132385696853e-06, "loss": 0.1896, "step": 4580 }, { "epoch": 0.33, "grad_norm": 14.671644631913004, "learning_rate": 7.840176694224517e-06, "loss": 0.8653, "step": 4581 }, { "epoch": 0.33, "grad_norm": 34.37931869676478, "learning_rate": 7.839220849535398e-06, "loss": 1.827, "step": 4582 }, { "epoch": 0.33, "grad_norm": 4.327169665976989, "learning_rate": 7.83826485168106e-06, "loss": 0.1469, "step": 4583 }, { "epoch": 0.33, "grad_norm": 4.954822972865217, "learning_rate": 7.837308700713078e-06, "loss": 0.3655, "step": 4584 }, { "epoch": 0.33, "grad_norm": 6.0202964390100915, "learning_rate": 7.836352396683031e-06, "loss": 1.2483, "step": 4585 }, { "epoch": 0.33, "grad_norm": 5.3451946131531995, "learning_rate": 7.835395939642508e-06, "loss": 0.2957, "step": 4586 }, { "epoch": 0.33, "grad_norm": 40.68678958238465, "learning_rate": 7.834439329643106e-06, "loss": 1.0744, "step": 4587 }, { "epoch": 0.33, "grad_norm": 18.44001156165515, "learning_rate": 7.833482566736433e-06, "loss": 1.061, "step": 4588 }, { "epoch": 0.33, "grad_norm": 17.250968783901634, "learning_rate": 7.8325256509741e-06, "loss": 1.4338, "step": 4589 }, { "epoch": 0.33, "grad_norm": 5.201942365539642, "learning_rate": 7.831568582407729e-06, "loss": 1.0621, "step": 4590 }, { "epoch": 0.33, "grad_norm": 4.850325226458425, "learning_rate": 7.830611361088952e-06, "loss": 1.0048, "step": 4591 }, { "epoch": 0.33, "grad_norm": 4.65180505484248, "learning_rate": 7.829653987069407e-06, "loss": 0.9538, "step": 4592 }, { "epoch": 0.33, "grad_norm": 17.358996548760153, "learning_rate": 7.82869646040074e-06, "loss": 0.974, "step": 4593 }, { "epoch": 0.33, "grad_norm": 6.927450640402809, "learning_rate": 7.827738781134606e-06, "loss": 0.358, "step": 4594 }, { "epoch": 0.33, "grad_norm": 2.8592444146034413, "learning_rate": 7.826780949322668e-06, "loss": 0.1235, "step": 4595 }, { "epoch": 0.33, "grad_norm": 8.941558989420558, "learning_rate": 7.825822965016596e-06, "loss": 0.2354, "step": 4596 }, { "epoch": 0.33, "grad_norm": 4.481893075933115, "learning_rate": 7.824864828268073e-06, "loss": 0.2907, "step": 4597 }, { "epoch": 0.33, "grad_norm": 3.112891116221564, "learning_rate": 7.823906539128786e-06, "loss": 0.158, "step": 4598 }, { "epoch": 0.33, "grad_norm": 6.786326754706472, "learning_rate": 7.82294809765043e-06, "loss": 0.4846, "step": 4599 }, { "epoch": 0.33, "grad_norm": 12.443541842594191, "learning_rate": 7.821989503884709e-06, "loss": 0.6084, "step": 4600 }, { "epoch": 0.33, "grad_norm": 3.8981613456792146, "learning_rate": 7.821030757883337e-06, "loss": 0.3151, "step": 4601 }, { "epoch": 0.33, "grad_norm": 12.97406376859695, "learning_rate": 7.820071859698033e-06, "loss": 1.509, "step": 4602 }, { "epoch": 0.33, "grad_norm": 15.218740137697512, "learning_rate": 7.819112809380526e-06, "loss": 0.9059, "step": 4603 }, { "epoch": 0.33, "grad_norm": 3.3639306875469193, "learning_rate": 7.818153606982555e-06, "loss": 0.268, "step": 4604 }, { "epoch": 0.33, "grad_norm": 3.3358583457783406, "learning_rate": 7.817194252555864e-06, "loss": 0.2285, "step": 4605 }, { "epoch": 0.33, "grad_norm": 8.609432584413542, "learning_rate": 7.816234746152207e-06, "loss": 0.6287, "step": 4606 }, { "epoch": 0.33, "grad_norm": 3.4773364624758663, "learning_rate": 7.815275087823343e-06, "loss": 0.2645, "step": 4607 }, { "epoch": 0.33, "grad_norm": 4.665137413663499, "learning_rate": 7.814315277621047e-06, "loss": 0.2829, "step": 4608 }, { "epoch": 0.33, "grad_norm": 10.291076311549588, "learning_rate": 7.813355315597094e-06, "loss": 0.5847, "step": 4609 }, { "epoch": 0.33, "grad_norm": 2.8413705925162698, "learning_rate": 7.812395201803272e-06, "loss": 0.2126, "step": 4610 }, { "epoch": 0.33, "grad_norm": 6.439544897215517, "learning_rate": 7.81143493629137e-06, "loss": 1.1109, "step": 4611 }, { "epoch": 0.33, "grad_norm": 3.307603467985461, "learning_rate": 7.8104745191132e-06, "loss": 0.2408, "step": 4612 }, { "epoch": 0.33, "grad_norm": 3.436131225161918, "learning_rate": 7.809513950320566e-06, "loss": 0.3018, "step": 4613 }, { "epoch": 0.33, "grad_norm": 3.6829960025505146, "learning_rate": 7.80855322996529e-06, "loss": 0.2935, "step": 4614 }, { "epoch": 0.33, "grad_norm": 12.61515019969047, "learning_rate": 7.807592358099197e-06, "loss": 1.1468, "step": 4615 }, { "epoch": 0.33, "grad_norm": 10.98551048666334, "learning_rate": 7.806631334774125e-06, "loss": 0.845, "step": 4616 }, { "epoch": 0.33, "grad_norm": 8.91998444181872, "learning_rate": 7.805670160041914e-06, "loss": 0.8677, "step": 4617 }, { "epoch": 0.33, "grad_norm": 3.747800804888002, "learning_rate": 7.80470883395442e-06, "loss": 0.2823, "step": 4618 }, { "epoch": 0.33, "grad_norm": 11.720050022071964, "learning_rate": 7.803747356563498e-06, "loss": 1.0382, "step": 4619 }, { "epoch": 0.33, "grad_norm": 3.550787389545361, "learning_rate": 7.802785727921021e-06, "loss": 0.241, "step": 4620 }, { "epoch": 0.33, "grad_norm": 17.819182082107098, "learning_rate": 7.801823948078863e-06, "loss": 1.2316, "step": 4621 }, { "epoch": 0.33, "grad_norm": 4.560483002625191, "learning_rate": 7.800862017088908e-06, "loss": 0.3154, "step": 4622 }, { "epoch": 0.33, "grad_norm": 3.110372755542209, "learning_rate": 7.799899935003048e-06, "loss": 0.1664, "step": 4623 }, { "epoch": 0.33, "grad_norm": 19.659419032374128, "learning_rate": 7.798937701873188e-06, "loss": 1.5837, "step": 4624 }, { "epoch": 0.33, "grad_norm": 18.114356325053695, "learning_rate": 7.79797531775123e-06, "loss": 1.1236, "step": 4625 }, { "epoch": 0.33, "grad_norm": 3.62366999913574, "learning_rate": 7.797012782689097e-06, "loss": 0.1699, "step": 4626 }, { "epoch": 0.33, "grad_norm": 10.093782782150525, "learning_rate": 7.79605009673871e-06, "loss": 0.8309, "step": 4627 }, { "epoch": 0.33, "grad_norm": 16.419956710219974, "learning_rate": 7.795087259952005e-06, "loss": 0.7483, "step": 4628 }, { "epoch": 0.33, "grad_norm": 3.257021496909501, "learning_rate": 7.794124272380921e-06, "loss": 0.1687, "step": 4629 }, { "epoch": 0.33, "grad_norm": 4.359292447749283, "learning_rate": 7.79316113407741e-06, "loss": 0.277, "step": 4630 }, { "epoch": 0.33, "grad_norm": 3.3985075210237237, "learning_rate": 7.792197845093428e-06, "loss": 0.2582, "step": 4631 }, { "epoch": 0.33, "grad_norm": 3.675147885975896, "learning_rate": 7.791234405480942e-06, "loss": 0.214, "step": 4632 }, { "epoch": 0.33, "grad_norm": 3.273393221938819, "learning_rate": 7.790270815291924e-06, "loss": 0.1659, "step": 4633 }, { "epoch": 0.33, "grad_norm": 13.599373327499755, "learning_rate": 7.789307074578358e-06, "loss": 0.8041, "step": 4634 }, { "epoch": 0.33, "grad_norm": 18.52413589764355, "learning_rate": 7.788343183392232e-06, "loss": 1.4398, "step": 4635 }, { "epoch": 0.33, "grad_norm": 5.563379235478252, "learning_rate": 7.78737914178555e-06, "loss": 0.7581, "step": 4636 }, { "epoch": 0.33, "grad_norm": 4.00605499417119, "learning_rate": 7.786414949810311e-06, "loss": 0.2822, "step": 4637 }, { "epoch": 0.33, "grad_norm": 20.998662289036943, "learning_rate": 7.785450607518532e-06, "loss": 1.4926, "step": 4638 }, { "epoch": 0.33, "grad_norm": 4.568577966465275, "learning_rate": 7.784486114962237e-06, "loss": 0.3692, "step": 4639 }, { "epoch": 0.33, "grad_norm": 5.1102739321173525, "learning_rate": 7.783521472193456e-06, "loss": 1.0458, "step": 4640 }, { "epoch": 0.33, "grad_norm": 4.481375056716971, "learning_rate": 7.782556679264228e-06, "loss": 0.2137, "step": 4641 }, { "epoch": 0.33, "grad_norm": 4.949683002699103, "learning_rate": 7.7815917362266e-06, "loss": 1.0132, "step": 4642 }, { "epoch": 0.33, "grad_norm": 10.869719456790204, "learning_rate": 7.780626643132626e-06, "loss": 0.7907, "step": 4643 }, { "epoch": 0.33, "grad_norm": 2.9050756685570196, "learning_rate": 7.779661400034368e-06, "loss": 0.2005, "step": 4644 }, { "epoch": 0.33, "grad_norm": 5.5915965572094715, "learning_rate": 7.778696006983903e-06, "loss": 0.8766, "step": 4645 }, { "epoch": 0.33, "grad_norm": 3.548134322915602, "learning_rate": 7.777730464033304e-06, "loss": 0.2311, "step": 4646 }, { "epoch": 0.33, "grad_norm": 4.226013877359649, "learning_rate": 7.776764771234662e-06, "loss": 0.8766, "step": 4647 }, { "epoch": 0.33, "grad_norm": 2.5008005293918183, "learning_rate": 7.775798928640071e-06, "loss": 0.2148, "step": 4648 }, { "epoch": 0.33, "grad_norm": 3.77507348381353, "learning_rate": 7.774832936301632e-06, "loss": 0.3184, "step": 4649 }, { "epoch": 0.33, "grad_norm": 3.132965209175813, "learning_rate": 7.773866794271464e-06, "loss": 0.1597, "step": 4650 }, { "epoch": 0.33, "grad_norm": 4.700579619176154, "learning_rate": 7.772900502601678e-06, "loss": 0.846, "step": 4651 }, { "epoch": 0.33, "grad_norm": 12.45731377088042, "learning_rate": 7.771934061344407e-06, "loss": 0.6921, "step": 4652 }, { "epoch": 0.33, "grad_norm": 11.669487093011854, "learning_rate": 7.770967470551787e-06, "loss": 1.5287, "step": 4653 }, { "epoch": 0.33, "grad_norm": 5.0177820598566445, "learning_rate": 7.770000730275959e-06, "loss": 1.0022, "step": 4654 }, { "epoch": 0.33, "grad_norm": 4.688258400757773, "learning_rate": 7.769033840569078e-06, "loss": 0.7042, "step": 4655 }, { "epoch": 0.33, "grad_norm": 3.6403901763130926, "learning_rate": 7.768066801483303e-06, "loss": 0.2392, "step": 4656 }, { "epoch": 0.33, "grad_norm": 3.674857717182137, "learning_rate": 7.767099613070802e-06, "loss": 0.2613, "step": 4657 }, { "epoch": 0.33, "grad_norm": 12.558930747918138, "learning_rate": 7.76613227538375e-06, "loss": 1.0927, "step": 4658 }, { "epoch": 0.33, "grad_norm": 3.867758561813441, "learning_rate": 7.76516478847433e-06, "loss": 0.1843, "step": 4659 }, { "epoch": 0.33, "grad_norm": 11.590381041196254, "learning_rate": 7.764197152394738e-06, "loss": 1.0499, "step": 4660 }, { "epoch": 0.33, "grad_norm": 3.3366656405237025, "learning_rate": 7.763229367197173e-06, "loss": 0.2034, "step": 4661 }, { "epoch": 0.33, "grad_norm": 25.2193998489773, "learning_rate": 7.762261432933844e-06, "loss": 1.0457, "step": 4662 }, { "epoch": 0.33, "grad_norm": 4.509173362158864, "learning_rate": 7.761293349656966e-06, "loss": 0.176, "step": 4663 }, { "epoch": 0.33, "grad_norm": 16.682976085569326, "learning_rate": 7.760325117418763e-06, "loss": 1.1707, "step": 4664 }, { "epoch": 0.33, "grad_norm": 3.6200142920514793, "learning_rate": 7.759356736271469e-06, "loss": 0.1823, "step": 4665 }, { "epoch": 0.33, "grad_norm": 10.097254392805022, "learning_rate": 7.758388206267324e-06, "loss": 0.789, "step": 4666 }, { "epoch": 0.33, "grad_norm": 14.18620787701654, "learning_rate": 7.757419527458577e-06, "loss": 0.8773, "step": 4667 }, { "epoch": 0.33, "grad_norm": 47.29515749541948, "learning_rate": 7.75645069989748e-06, "loss": 0.7421, "step": 4668 }, { "epoch": 0.33, "grad_norm": 4.138264736282903, "learning_rate": 7.755481723636305e-06, "loss": 0.2152, "step": 4669 }, { "epoch": 0.33, "grad_norm": 3.399350566254681, "learning_rate": 7.75451259872732e-06, "loss": 0.2202, "step": 4670 }, { "epoch": 0.33, "grad_norm": 3.3853037747749313, "learning_rate": 7.753543325222807e-06, "loss": 0.1952, "step": 4671 }, { "epoch": 0.34, "grad_norm": 4.4389501966689835, "learning_rate": 7.752573903175055e-06, "loss": 0.2452, "step": 4672 }, { "epoch": 0.34, "grad_norm": 4.727801957455028, "learning_rate": 7.75160433263636e-06, "loss": 0.95, "step": 4673 }, { "epoch": 0.34, "grad_norm": 139.44101677475606, "learning_rate": 7.750634613659026e-06, "loss": 0.9652, "step": 4674 }, { "epoch": 0.34, "grad_norm": 35.05166221575381, "learning_rate": 7.749664746295364e-06, "loss": 1.0526, "step": 4675 }, { "epoch": 0.34, "grad_norm": 5.189741771212438, "learning_rate": 7.7486947305977e-06, "loss": 1.029, "step": 4676 }, { "epoch": 0.34, "grad_norm": 4.510096929874011, "learning_rate": 7.74772456661836e-06, "loss": 0.3127, "step": 4677 }, { "epoch": 0.34, "grad_norm": 11.497270920759519, "learning_rate": 7.746754254409679e-06, "loss": 0.9564, "step": 4678 }, { "epoch": 0.34, "grad_norm": 2.904135210061655, "learning_rate": 7.745783794024004e-06, "loss": 0.1869, "step": 4679 }, { "epoch": 0.34, "grad_norm": 3.4106403563812524, "learning_rate": 7.744813185513685e-06, "loss": 0.2256, "step": 4680 }, { "epoch": 0.34, "grad_norm": 10.089256764379646, "learning_rate": 7.743842428931086e-06, "loss": 0.8692, "step": 4681 }, { "epoch": 0.34, "grad_norm": 16.23123195093231, "learning_rate": 7.742871524328572e-06, "loss": 0.9399, "step": 4682 }, { "epoch": 0.34, "grad_norm": 3.6449947770110493, "learning_rate": 7.741900471758523e-06, "loss": 0.2843, "step": 4683 }, { "epoch": 0.34, "grad_norm": 3.6051419738989137, "learning_rate": 7.740929271273323e-06, "loss": 0.2696, "step": 4684 }, { "epoch": 0.34, "grad_norm": 3.655224037855001, "learning_rate": 7.739957922925362e-06, "loss": 0.1919, "step": 4685 }, { "epoch": 0.34, "grad_norm": 24.24944612986849, "learning_rate": 7.738986426767044e-06, "loss": 1.3135, "step": 4686 }, { "epoch": 0.34, "grad_norm": 3.511644632449795, "learning_rate": 7.738014782850773e-06, "loss": 0.2048, "step": 4687 }, { "epoch": 0.34, "grad_norm": 3.304208967122086, "learning_rate": 7.737042991228972e-06, "loss": 0.1612, "step": 4688 }, { "epoch": 0.34, "grad_norm": 2.9163058978713963, "learning_rate": 7.73607105195406e-06, "loss": 0.1504, "step": 4689 }, { "epoch": 0.34, "grad_norm": 3.6598364249683613, "learning_rate": 7.735098965078472e-06, "loss": 0.2537, "step": 4690 }, { "epoch": 0.34, "grad_norm": 33.46260557959653, "learning_rate": 7.734126730654648e-06, "loss": 0.8781, "step": 4691 }, { "epoch": 0.34, "grad_norm": 5.47952486489008, "learning_rate": 7.733154348735035e-06, "loss": 0.9863, "step": 4692 }, { "epoch": 0.34, "grad_norm": 29.560026299681766, "learning_rate": 7.73218181937209e-06, "loss": 1.1141, "step": 4693 }, { "epoch": 0.34, "grad_norm": 4.722780033632612, "learning_rate": 7.731209142618279e-06, "loss": 1.1184, "step": 4694 }, { "epoch": 0.34, "grad_norm": 4.2561105584914065, "learning_rate": 7.730236318526072e-06, "loss": 0.2931, "step": 4695 }, { "epoch": 0.34, "grad_norm": 15.661565202194536, "learning_rate": 7.729263347147952e-06, "loss": 0.862, "step": 4696 }, { "epoch": 0.34, "grad_norm": 3.112936181619445, "learning_rate": 7.728290228536404e-06, "loss": 0.1557, "step": 4697 }, { "epoch": 0.34, "grad_norm": 5.03337751992365, "learning_rate": 7.727316962743924e-06, "loss": 0.9012, "step": 4698 }, { "epoch": 0.34, "grad_norm": 19.284573528136402, "learning_rate": 7.72634354982302e-06, "loss": 1.258, "step": 4699 }, { "epoch": 0.34, "grad_norm": 4.228267161567216, "learning_rate": 7.7253699898262e-06, "loss": 0.2446, "step": 4700 }, { "epoch": 0.34, "grad_norm": 4.376984329984978, "learning_rate": 7.724396282805986e-06, "loss": 0.2558, "step": 4701 }, { "epoch": 0.34, "grad_norm": 4.627460045733042, "learning_rate": 7.723422428814903e-06, "loss": 0.859, "step": 4702 }, { "epoch": 0.34, "grad_norm": 4.636107319939676, "learning_rate": 7.722448427905489e-06, "loss": 0.3135, "step": 4703 }, { "epoch": 0.34, "grad_norm": 3.5380025177697414, "learning_rate": 7.721474280130288e-06, "loss": 0.2146, "step": 4704 }, { "epoch": 0.34, "grad_norm": 12.162860942782155, "learning_rate": 7.720499985541852e-06, "loss": 1.269, "step": 4705 }, { "epoch": 0.34, "grad_norm": 21.001799150808658, "learning_rate": 7.719525544192737e-06, "loss": 1.7964, "step": 4706 }, { "epoch": 0.34, "grad_norm": 4.081682231988608, "learning_rate": 7.718550956135515e-06, "loss": 0.2126, "step": 4707 }, { "epoch": 0.34, "grad_norm": 17.163530697913146, "learning_rate": 7.717576221422757e-06, "loss": 1.2016, "step": 4708 }, { "epoch": 0.34, "grad_norm": 4.254919530076667, "learning_rate": 7.716601340107052e-06, "loss": 0.2503, "step": 4709 }, { "epoch": 0.34, "grad_norm": 3.5420772595169736, "learning_rate": 7.715626312240985e-06, "loss": 0.1694, "step": 4710 }, { "epoch": 0.34, "grad_norm": 5.010494487299685, "learning_rate": 7.714651137877159e-06, "loss": 1.0813, "step": 4711 }, { "epoch": 0.34, "grad_norm": 7.459489183082076, "learning_rate": 7.713675817068176e-06, "loss": 0.8088, "step": 4712 }, { "epoch": 0.34, "grad_norm": 21.098399700615655, "learning_rate": 7.712700349866657e-06, "loss": 1.4622, "step": 4713 }, { "epoch": 0.34, "grad_norm": 15.243353398995408, "learning_rate": 7.711724736325223e-06, "loss": 0.7241, "step": 4714 }, { "epoch": 0.34, "grad_norm": 4.549434980842887, "learning_rate": 7.710748976496506e-06, "loss": 1.0223, "step": 4715 }, { "epoch": 0.34, "grad_norm": 7.082034676838645, "learning_rate": 7.709773070433139e-06, "loss": 0.3694, "step": 4716 }, { "epoch": 0.34, "grad_norm": 9.355891914860736, "learning_rate": 7.708797018187773e-06, "loss": 0.4889, "step": 4717 }, { "epoch": 0.34, "grad_norm": 9.512236680857267, "learning_rate": 7.70782081981306e-06, "loss": 0.5405, "step": 4718 }, { "epoch": 0.34, "grad_norm": 3.236464023628861, "learning_rate": 7.706844475361664e-06, "loss": 0.2747, "step": 4719 }, { "epoch": 0.34, "grad_norm": 3.347777901785547, "learning_rate": 7.705867984886255e-06, "loss": 0.2343, "step": 4720 }, { "epoch": 0.34, "grad_norm": 9.341351629353687, "learning_rate": 7.70489134843951e-06, "loss": 0.7292, "step": 4721 }, { "epoch": 0.34, "grad_norm": 2.858346271878001, "learning_rate": 7.703914566074117e-06, "loss": 0.2485, "step": 4722 }, { "epoch": 0.34, "grad_norm": 4.464948425564228, "learning_rate": 7.702937637842767e-06, "loss": 0.9829, "step": 4723 }, { "epoch": 0.34, "grad_norm": 3.6054152008606835, "learning_rate": 7.701960563798161e-06, "loss": 0.3307, "step": 4724 }, { "epoch": 0.34, "grad_norm": 3.70201823426002, "learning_rate": 7.700983343993012e-06, "loss": 0.2653, "step": 4725 }, { "epoch": 0.34, "grad_norm": 5.597936010418659, "learning_rate": 7.700005978480036e-06, "loss": 0.869, "step": 4726 }, { "epoch": 0.34, "grad_norm": 4.514754359284461, "learning_rate": 7.699028467311955e-06, "loss": 0.9151, "step": 4727 }, { "epoch": 0.34, "grad_norm": 4.021302449921561, "learning_rate": 7.698050810541506e-06, "loss": 0.304, "step": 4728 }, { "epoch": 0.34, "grad_norm": 4.518393085673552, "learning_rate": 7.697073008221428e-06, "loss": 0.1872, "step": 4729 }, { "epoch": 0.34, "grad_norm": 20.969136645822765, "learning_rate": 7.696095060404471e-06, "loss": 0.9281, "step": 4730 }, { "epoch": 0.34, "grad_norm": 3.587016983133738, "learning_rate": 7.69511696714339e-06, "loss": 0.2132, "step": 4731 }, { "epoch": 0.34, "grad_norm": 3.388363104165442, "learning_rate": 7.694138728490951e-06, "loss": 0.1558, "step": 4732 }, { "epoch": 0.34, "grad_norm": 5.341994469844949, "learning_rate": 7.693160344499924e-06, "loss": 0.9724, "step": 4733 }, { "epoch": 0.34, "grad_norm": 7.035903398189277, "learning_rate": 7.692181815223093e-06, "loss": 0.6119, "step": 4734 }, { "epoch": 0.34, "grad_norm": 14.690945761103512, "learning_rate": 7.691203140713242e-06, "loss": 0.8457, "step": 4735 }, { "epoch": 0.34, "grad_norm": 4.058582212056295, "learning_rate": 7.690224321023168e-06, "loss": 0.2735, "step": 4736 }, { "epoch": 0.34, "grad_norm": 5.75145958143347, "learning_rate": 7.689245356205675e-06, "loss": 1.1176, "step": 4737 }, { "epoch": 0.34, "grad_norm": 7.557045758428914, "learning_rate": 7.688266246313574e-06, "loss": 0.3647, "step": 4738 }, { "epoch": 0.34, "grad_norm": 16.133614054219713, "learning_rate": 7.687286991399686e-06, "loss": 1.5445, "step": 4739 }, { "epoch": 0.34, "grad_norm": 25.316674352600494, "learning_rate": 7.686307591516836e-06, "loss": 1.608, "step": 4740 }, { "epoch": 0.34, "grad_norm": 16.787398644944332, "learning_rate": 7.685328046717859e-06, "loss": 1.0628, "step": 4741 }, { "epoch": 0.34, "grad_norm": 3.160480346238986, "learning_rate": 7.6843483570556e-06, "loss": 0.1349, "step": 4742 }, { "epoch": 0.34, "grad_norm": 4.653220260545206, "learning_rate": 7.683368522582905e-06, "loss": 0.9742, "step": 4743 }, { "epoch": 0.34, "grad_norm": 11.620523558537336, "learning_rate": 7.682388543352638e-06, "loss": 1.3166, "step": 4744 }, { "epoch": 0.34, "grad_norm": 40.44496975107865, "learning_rate": 7.681408419417663e-06, "loss": 1.183, "step": 4745 }, { "epoch": 0.34, "grad_norm": 3.157361809560997, "learning_rate": 7.680428150830852e-06, "loss": 0.1732, "step": 4746 }, { "epoch": 0.34, "grad_norm": 4.105051550301777, "learning_rate": 7.679447737645088e-06, "loss": 0.295, "step": 4747 }, { "epoch": 0.34, "grad_norm": 4.239607024589858, "learning_rate": 7.678467179913263e-06, "loss": 0.1507, "step": 4748 }, { "epoch": 0.34, "grad_norm": 28.327791419431332, "learning_rate": 7.67748647768827e-06, "loss": 1.4088, "step": 4749 }, { "epoch": 0.34, "grad_norm": 15.440980050767921, "learning_rate": 7.676505631023017e-06, "loss": 1.1366, "step": 4750 }, { "epoch": 0.34, "grad_norm": 3.919660939377439, "learning_rate": 7.675524639970416e-06, "loss": 0.2574, "step": 4751 }, { "epoch": 0.34, "grad_norm": 3.6633215695620924, "learning_rate": 7.674543504583388e-06, "loss": 0.1611, "step": 4752 }, { "epoch": 0.34, "grad_norm": 4.240268484686846, "learning_rate": 7.673562224914864e-06, "loss": 0.2224, "step": 4753 }, { "epoch": 0.34, "grad_norm": 7.827845489560604, "learning_rate": 7.672580801017774e-06, "loss": 1.0834, "step": 4754 }, { "epoch": 0.34, "grad_norm": 3.65179597659082, "learning_rate": 7.671599232945069e-06, "loss": 0.149, "step": 4755 }, { "epoch": 0.34, "grad_norm": 19.43918997896183, "learning_rate": 7.670617520749697e-06, "loss": 1.595, "step": 4756 }, { "epoch": 0.34, "grad_norm": 3.6710143976141194, "learning_rate": 7.669635664484619e-06, "loss": 0.2212, "step": 4757 }, { "epoch": 0.34, "grad_norm": 6.135244179519968, "learning_rate": 7.6686536642028e-06, "loss": 0.3271, "step": 4758 }, { "epoch": 0.34, "grad_norm": 3.6535348148447895, "learning_rate": 7.66767151995722e-06, "loss": 0.1698, "step": 4759 }, { "epoch": 0.34, "grad_norm": 5.243811916882649, "learning_rate": 7.666689231800856e-06, "loss": 0.9959, "step": 4760 }, { "epoch": 0.34, "grad_norm": 4.214794009397422, "learning_rate": 7.665706799786705e-06, "loss": 0.2543, "step": 4761 }, { "epoch": 0.34, "grad_norm": 4.135226901249719, "learning_rate": 7.664724223967761e-06, "loss": 0.3257, "step": 4762 }, { "epoch": 0.34, "grad_norm": 4.439086118568784, "learning_rate": 7.663741504397032e-06, "loss": 0.2862, "step": 4763 }, { "epoch": 0.34, "grad_norm": 16.671015963197558, "learning_rate": 7.662758641127531e-06, "loss": 1.0425, "step": 4764 }, { "epoch": 0.34, "grad_norm": 5.135040207355265, "learning_rate": 7.66177563421228e-06, "loss": 0.8755, "step": 4765 }, { "epoch": 0.34, "grad_norm": 4.024629541184117, "learning_rate": 7.660792483704312e-06, "loss": 0.241, "step": 4766 }, { "epoch": 0.34, "grad_norm": 22.55426233030329, "learning_rate": 7.659809189656659e-06, "loss": 1.0771, "step": 4767 }, { "epoch": 0.34, "grad_norm": 16.36090900121713, "learning_rate": 7.658825752122367e-06, "loss": 1.2164, "step": 4768 }, { "epoch": 0.34, "grad_norm": 4.5207801759935515, "learning_rate": 7.65784217115449e-06, "loss": 0.2526, "step": 4769 }, { "epoch": 0.34, "grad_norm": 20.447696968800663, "learning_rate": 7.65685844680609e-06, "loss": 2.2294, "step": 4770 }, { "epoch": 0.34, "grad_norm": 16.05061425516581, "learning_rate": 7.655874579130234e-06, "loss": 0.9837, "step": 4771 }, { "epoch": 0.34, "grad_norm": 18.253061694577436, "learning_rate": 7.654890568179995e-06, "loss": 1.4132, "step": 4772 }, { "epoch": 0.34, "grad_norm": 3.3284789911023394, "learning_rate": 7.65390641400846e-06, "loss": 0.2062, "step": 4773 }, { "epoch": 0.34, "grad_norm": 9.655192078823735, "learning_rate": 7.652922116668722e-06, "loss": 0.6099, "step": 4774 }, { "epoch": 0.34, "grad_norm": 17.221760416905123, "learning_rate": 7.651937676213876e-06, "loss": 1.0575, "step": 4775 }, { "epoch": 0.34, "grad_norm": 5.858358579434193, "learning_rate": 7.650953092697029e-06, "loss": 0.9404, "step": 4776 }, { "epoch": 0.34, "grad_norm": 3.455187081387278, "learning_rate": 7.649968366171298e-06, "loss": 0.2016, "step": 4777 }, { "epoch": 0.34, "grad_norm": 4.031692301657086, "learning_rate": 7.648983496689804e-06, "loss": 0.2349, "step": 4778 }, { "epoch": 0.34, "grad_norm": 5.99745016457677, "learning_rate": 7.647998484305678e-06, "loss": 1.0739, "step": 4779 }, { "epoch": 0.34, "grad_norm": 8.686627733715873, "learning_rate": 7.647013329072058e-06, "loss": 1.011, "step": 4780 }, { "epoch": 0.34, "grad_norm": 4.623928749860881, "learning_rate": 7.646028031042088e-06, "loss": 1.0553, "step": 4781 }, { "epoch": 0.34, "grad_norm": 3.4054117377068382, "learning_rate": 7.645042590268921e-06, "loss": 0.1437, "step": 4782 }, { "epoch": 0.34, "grad_norm": 19.361078925635688, "learning_rate": 7.644057006805716e-06, "loss": 1.2549, "step": 4783 }, { "epoch": 0.34, "grad_norm": 4.709526189366129, "learning_rate": 7.643071280705647e-06, "loss": 0.292, "step": 4784 }, { "epoch": 0.34, "grad_norm": 4.325686713844528, "learning_rate": 7.642085412021886e-06, "loss": 0.1709, "step": 4785 }, { "epoch": 0.34, "grad_norm": 5.04273457034825, "learning_rate": 7.641099400807616e-06, "loss": 0.4025, "step": 4786 }, { "epoch": 0.34, "grad_norm": 14.702023632384106, "learning_rate": 7.640113247116034e-06, "loss": 1.3474, "step": 4787 }, { "epoch": 0.34, "grad_norm": 4.127187013741253, "learning_rate": 7.639126951000333e-06, "loss": 0.306, "step": 4788 }, { "epoch": 0.34, "grad_norm": 20.19913682074684, "learning_rate": 7.638140512513722e-06, "loss": 1.5235, "step": 4789 }, { "epoch": 0.34, "grad_norm": 12.269507739143075, "learning_rate": 7.63715393170942e-06, "loss": 1.1908, "step": 4790 }, { "epoch": 0.34, "grad_norm": 4.028706333389682, "learning_rate": 7.636167208640642e-06, "loss": 0.348, "step": 4791 }, { "epoch": 0.34, "grad_norm": 21.389879189129026, "learning_rate": 7.635180343360622e-06, "loss": 1.0891, "step": 4792 }, { "epoch": 0.34, "grad_norm": 4.690478330719508, "learning_rate": 7.634193335922599e-06, "loss": 0.1922, "step": 4793 }, { "epoch": 0.34, "grad_norm": 4.080549647223784, "learning_rate": 7.633206186379815e-06, "loss": 0.1467, "step": 4794 }, { "epoch": 0.34, "grad_norm": 19.310178328394514, "learning_rate": 7.632218894785526e-06, "loss": 1.1469, "step": 4795 }, { "epoch": 0.34, "grad_norm": 10.913059861969657, "learning_rate": 7.63123146119299e-06, "loss": 0.9595, "step": 4796 }, { "epoch": 0.34, "grad_norm": 12.014803446393193, "learning_rate": 7.630243885655476e-06, "loss": 0.7839, "step": 4797 }, { "epoch": 0.34, "grad_norm": 3.1421719658638194, "learning_rate": 7.62925616822626e-06, "loss": 0.1942, "step": 4798 }, { "epoch": 0.34, "grad_norm": 11.951174335969704, "learning_rate": 7.628268308958629e-06, "loss": 0.8027, "step": 4799 }, { "epoch": 0.34, "grad_norm": 4.3381135828403306, "learning_rate": 7.6272803079058675e-06, "loss": 0.3104, "step": 4800 }, { "epoch": 0.34, "grad_norm": 3.1872330982364363, "learning_rate": 7.6262921651212794e-06, "loss": 0.212, "step": 4801 }, { "epoch": 0.34, "grad_norm": 7.5984436113326455, "learning_rate": 7.62530388065817e-06, "loss": 0.7192, "step": 4802 }, { "epoch": 0.34, "grad_norm": 4.105756856307153, "learning_rate": 7.624315454569855e-06, "loss": 0.2825, "step": 4803 }, { "epoch": 0.34, "grad_norm": 3.8408148216510094, "learning_rate": 7.623326886909655e-06, "loss": 0.3315, "step": 4804 }, { "epoch": 0.34, "grad_norm": 3.6816973312384547, "learning_rate": 7.622338177730897e-06, "loss": 0.2887, "step": 4805 }, { "epoch": 0.34, "grad_norm": 3.02101582596002, "learning_rate": 7.621349327086921e-06, "loss": 0.2036, "step": 4806 }, { "epoch": 0.34, "grad_norm": 3.2241573619421806, "learning_rate": 7.620360335031071e-06, "loss": 0.3116, "step": 4807 }, { "epoch": 0.34, "grad_norm": 5.010837006794287, "learning_rate": 7.619371201616701e-06, "loss": 1.2095, "step": 4808 }, { "epoch": 0.34, "grad_norm": 4.422469445791845, "learning_rate": 7.6183819268971685e-06, "loss": 0.3061, "step": 4809 }, { "epoch": 0.34, "grad_norm": 25.018856942987085, "learning_rate": 7.617392510925843e-06, "loss": 1.0395, "step": 4810 }, { "epoch": 0.34, "grad_norm": 11.405305866781777, "learning_rate": 7.616402953756098e-06, "loss": 0.9017, "step": 4811 }, { "epoch": 0.35, "grad_norm": 3.394272237662542, "learning_rate": 7.615413255441317e-06, "loss": 0.2321, "step": 4812 }, { "epoch": 0.35, "grad_norm": 3.8941751186468374, "learning_rate": 7.6144234160348915e-06, "loss": 0.3145, "step": 4813 }, { "epoch": 0.35, "grad_norm": 9.329737165424802, "learning_rate": 7.6134334355902186e-06, "loss": 0.6176, "step": 4814 }, { "epoch": 0.35, "grad_norm": 10.329218984112222, "learning_rate": 7.612443314160704e-06, "loss": 1.1115, "step": 4815 }, { "epoch": 0.35, "grad_norm": 14.26180636580934, "learning_rate": 7.611453051799762e-06, "loss": 1.8278, "step": 4816 }, { "epoch": 0.35, "grad_norm": 4.869959571610339, "learning_rate": 7.610462648560812e-06, "loss": 0.3732, "step": 4817 }, { "epoch": 0.35, "grad_norm": 5.282912290892503, "learning_rate": 7.609472104497283e-06, "loss": 1.1145, "step": 4818 }, { "epoch": 0.35, "grad_norm": 4.141205515506342, "learning_rate": 7.608481419662612e-06, "loss": 0.1607, "step": 4819 }, { "epoch": 0.35, "grad_norm": 5.418782297483414, "learning_rate": 7.6074905941102405e-06, "loss": 0.8624, "step": 4820 }, { "epoch": 0.35, "grad_norm": 5.081443019759315, "learning_rate": 7.606499627893624e-06, "loss": 0.9111, "step": 4821 }, { "epoch": 0.35, "grad_norm": 10.751524960310608, "learning_rate": 7.605508521066216e-06, "loss": 0.8463, "step": 4822 }, { "epoch": 0.35, "grad_norm": 3.3942174467485113, "learning_rate": 7.604517273681489e-06, "loss": 0.2383, "step": 4823 }, { "epoch": 0.35, "grad_norm": 13.383344775256584, "learning_rate": 7.603525885792911e-06, "loss": 1.6405, "step": 4824 }, { "epoch": 0.35, "grad_norm": 3.6864267026865085, "learning_rate": 7.602534357453968e-06, "loss": 0.2454, "step": 4825 }, { "epoch": 0.35, "grad_norm": 6.220780978626373, "learning_rate": 7.601542688718146e-06, "loss": 0.7667, "step": 4826 }, { "epoch": 0.35, "grad_norm": 3.8207957999466697, "learning_rate": 7.600550879638944e-06, "loss": 0.2445, "step": 4827 }, { "epoch": 0.35, "grad_norm": 5.761075504210231, "learning_rate": 7.599558930269865e-06, "loss": 1.0437, "step": 4828 }, { "epoch": 0.35, "grad_norm": 3.961874309541914, "learning_rate": 7.598566840664423e-06, "loss": 0.193, "step": 4829 }, { "epoch": 0.35, "grad_norm": 4.384896435884626, "learning_rate": 7.597574610876135e-06, "loss": 0.8323, "step": 4830 }, { "epoch": 0.35, "grad_norm": 4.372791896735108, "learning_rate": 7.596582240958529e-06, "loss": 0.2589, "step": 4831 }, { "epoch": 0.35, "grad_norm": 3.38248532913966, "learning_rate": 7.595589730965139e-06, "loss": 0.1536, "step": 4832 }, { "epoch": 0.35, "grad_norm": 9.503365880376718, "learning_rate": 7.594597080949511e-06, "loss": 0.4275, "step": 4833 }, { "epoch": 0.35, "grad_norm": 3.3586559892635495, "learning_rate": 7.593604290965189e-06, "loss": 0.1451, "step": 4834 }, { "epoch": 0.35, "grad_norm": 10.358593505809555, "learning_rate": 7.592611361065732e-06, "loss": 0.7084, "step": 4835 }, { "epoch": 0.35, "grad_norm": 5.140826099086462, "learning_rate": 7.591618291304707e-06, "loss": 0.2329, "step": 4836 }, { "epoch": 0.35, "grad_norm": 3.852159238567869, "learning_rate": 7.590625081735685e-06, "loss": 0.234, "step": 4837 }, { "epoch": 0.35, "grad_norm": 3.063298973684845, "learning_rate": 7.589631732412246e-06, "loss": 0.1762, "step": 4838 }, { "epoch": 0.35, "grad_norm": 18.593485701538565, "learning_rate": 7.588638243387978e-06, "loss": 0.8853, "step": 4839 }, { "epoch": 0.35, "grad_norm": 2.9735121068962824, "learning_rate": 7.587644614716474e-06, "loss": 0.1791, "step": 4840 }, { "epoch": 0.35, "grad_norm": 5.007116943615835, "learning_rate": 7.586650846451338e-06, "loss": 1.1091, "step": 4841 }, { "epoch": 0.35, "grad_norm": 4.80560679812219, "learning_rate": 7.58565693864618e-06, "loss": 1.0305, "step": 4842 }, { "epoch": 0.35, "grad_norm": 5.049227949086875, "learning_rate": 7.584662891354617e-06, "loss": 0.8276, "step": 4843 }, { "epoch": 0.35, "grad_norm": 17.08366776999029, "learning_rate": 7.583668704630276e-06, "loss": 0.8598, "step": 4844 }, { "epoch": 0.35, "grad_norm": 3.360274070364165, "learning_rate": 7.582674378526786e-06, "loss": 0.2773, "step": 4845 }, { "epoch": 0.35, "grad_norm": 4.200046003477309, "learning_rate": 7.58167991309779e-06, "loss": 0.3338, "step": 4846 }, { "epoch": 0.35, "grad_norm": 3.4099204608394302, "learning_rate": 7.580685308396938e-06, "loss": 0.1906, "step": 4847 }, { "epoch": 0.35, "grad_norm": 14.934154910884164, "learning_rate": 7.579690564477877e-06, "loss": 0.9773, "step": 4848 }, { "epoch": 0.35, "grad_norm": 3.526295987615653, "learning_rate": 7.578695681394277e-06, "loss": 0.2288, "step": 4849 }, { "epoch": 0.35, "grad_norm": 11.064709802612812, "learning_rate": 7.577700659199807e-06, "loss": 0.8972, "step": 4850 }, { "epoch": 0.35, "grad_norm": 4.101830186025171, "learning_rate": 7.576705497948141e-06, "loss": 0.1939, "step": 4851 }, { "epoch": 0.35, "grad_norm": 5.153472830580551, "learning_rate": 7.5757101976929684e-06, "loss": 0.9801, "step": 4852 }, { "epoch": 0.35, "grad_norm": 13.500757860914607, "learning_rate": 7.574714758487981e-06, "loss": 0.8363, "step": 4853 }, { "epoch": 0.35, "grad_norm": 5.981396146721692, "learning_rate": 7.573719180386876e-06, "loss": 0.5539, "step": 4854 }, { "epoch": 0.35, "grad_norm": 11.158121271398873, "learning_rate": 7.572723463443365e-06, "loss": 0.8671, "step": 4855 }, { "epoch": 0.35, "grad_norm": 3.7586109270191357, "learning_rate": 7.571727607711161e-06, "loss": 0.243, "step": 4856 }, { "epoch": 0.35, "grad_norm": 3.212147444130551, "learning_rate": 7.5707316132439866e-06, "loss": 0.1604, "step": 4857 }, { "epoch": 0.35, "grad_norm": 14.96761798713061, "learning_rate": 7.569735480095572e-06, "loss": 0.6342, "step": 4858 }, { "epoch": 0.35, "grad_norm": 23.19984023624141, "learning_rate": 7.5687392083196575e-06, "loss": 1.3199, "step": 4859 }, { "epoch": 0.35, "grad_norm": 14.93352245751075, "learning_rate": 7.567742797969984e-06, "loss": 0.7107, "step": 4860 }, { "epoch": 0.35, "grad_norm": 3.6802112861504335, "learning_rate": 7.566746249100306e-06, "loss": 0.2151, "step": 4861 }, { "epoch": 0.35, "grad_norm": 4.412280858518089, "learning_rate": 7.565749561764384e-06, "loss": 0.3109, "step": 4862 }, { "epoch": 0.35, "grad_norm": 16.343847326846078, "learning_rate": 7.564752736015985e-06, "loss": 1.5952, "step": 4863 }, { "epoch": 0.35, "grad_norm": 4.189957428508128, "learning_rate": 7.563755771908884e-06, "loss": 0.3487, "step": 4864 }, { "epoch": 0.35, "grad_norm": 3.1622348367829467, "learning_rate": 7.562758669496863e-06, "loss": 0.1239, "step": 4865 }, { "epoch": 0.35, "grad_norm": 4.9888972308955415, "learning_rate": 7.561761428833713e-06, "loss": 1.1042, "step": 4866 }, { "epoch": 0.35, "grad_norm": 3.368803083442595, "learning_rate": 7.560764049973231e-06, "loss": 0.2321, "step": 4867 }, { "epoch": 0.35, "grad_norm": 14.573492641827054, "learning_rate": 7.55976653296922e-06, "loss": 1.3239, "step": 4868 }, { "epoch": 0.35, "grad_norm": 4.066443239827276, "learning_rate": 7.558768877875495e-06, "loss": 0.222, "step": 4869 }, { "epoch": 0.35, "grad_norm": 3.995732161884606, "learning_rate": 7.557771084745874e-06, "loss": 0.1475, "step": 4870 }, { "epoch": 0.35, "grad_norm": 4.799544802839965, "learning_rate": 7.556773153634185e-06, "loss": 0.8876, "step": 4871 }, { "epoch": 0.35, "grad_norm": 3.9455969384826886, "learning_rate": 7.555775084594263e-06, "loss": 0.2608, "step": 4872 }, { "epoch": 0.35, "grad_norm": 13.24886177339235, "learning_rate": 7.554776877679948e-06, "loss": 0.833, "step": 4873 }, { "epoch": 0.35, "grad_norm": 18.2137293496314, "learning_rate": 7.553778532945091e-06, "loss": 1.329, "step": 4874 }, { "epoch": 0.35, "grad_norm": 5.039383011272807, "learning_rate": 7.552780050443549e-06, "loss": 0.1948, "step": 4875 }, { "epoch": 0.35, "grad_norm": 3.884975536688908, "learning_rate": 7.551781430229186e-06, "loss": 0.2092, "step": 4876 }, { "epoch": 0.35, "grad_norm": 4.168858981676058, "learning_rate": 7.5507826723558746e-06, "loss": 0.221, "step": 4877 }, { "epoch": 0.35, "grad_norm": 5.076052950445692, "learning_rate": 7.549783776877493e-06, "loss": 1.0498, "step": 4878 }, { "epoch": 0.35, "grad_norm": 4.213237982114933, "learning_rate": 7.548784743847928e-06, "loss": 0.2918, "step": 4879 }, { "epoch": 0.35, "grad_norm": 13.666564368131239, "learning_rate": 7.547785573321072e-06, "loss": 0.7551, "step": 4880 }, { "epoch": 0.35, "grad_norm": 3.441502315236069, "learning_rate": 7.546786265350831e-06, "loss": 0.1905, "step": 4881 }, { "epoch": 0.35, "grad_norm": 3.867520564630271, "learning_rate": 7.5457868199911086e-06, "loss": 0.2915, "step": 4882 }, { "epoch": 0.35, "grad_norm": 4.876824487874206, "learning_rate": 7.544787237295825e-06, "loss": 0.3266, "step": 4883 }, { "epoch": 0.35, "grad_norm": 10.974872360112832, "learning_rate": 7.543787517318904e-06, "loss": 0.7866, "step": 4884 }, { "epoch": 0.35, "grad_norm": 3.5650290637464637, "learning_rate": 7.542787660114272e-06, "loss": 0.3337, "step": 4885 }, { "epoch": 0.35, "grad_norm": 3.170243354860824, "learning_rate": 7.541787665735871e-06, "loss": 0.2238, "step": 4886 }, { "epoch": 0.35, "grad_norm": 4.697615487964618, "learning_rate": 7.540787534237649e-06, "loss": 1.0188, "step": 4887 }, { "epoch": 0.35, "grad_norm": 5.196815509374834, "learning_rate": 7.539787265673555e-06, "loss": 0.2099, "step": 4888 }, { "epoch": 0.35, "grad_norm": 4.379288737302622, "learning_rate": 7.5387868600975535e-06, "loss": 0.8956, "step": 4889 }, { "epoch": 0.35, "grad_norm": 15.060022405687183, "learning_rate": 7.53778631756361e-06, "loss": 1.1917, "step": 4890 }, { "epoch": 0.35, "grad_norm": 4.647347665871973, "learning_rate": 7.5367856381257e-06, "loss": 0.8705, "step": 4891 }, { "epoch": 0.35, "grad_norm": 19.50645554935476, "learning_rate": 7.535784821837806e-06, "loss": 0.8416, "step": 4892 }, { "epoch": 0.35, "grad_norm": 8.30403642732486, "learning_rate": 7.534783868753921e-06, "loss": 1.1407, "step": 4893 }, { "epoch": 0.35, "grad_norm": 4.505873476411195, "learning_rate": 7.53378277892804e-06, "loss": 0.9946, "step": 4894 }, { "epoch": 0.35, "grad_norm": 9.539848619165006, "learning_rate": 7.532781552414171e-06, "loss": 0.771, "step": 4895 }, { "epoch": 0.35, "grad_norm": 3.2938869355579716, "learning_rate": 7.531780189266324e-06, "loss": 0.2369, "step": 4896 }, { "epoch": 0.35, "grad_norm": 4.563464296335966, "learning_rate": 7.530778689538517e-06, "loss": 0.9053, "step": 4897 }, { "epoch": 0.35, "grad_norm": 2.961541914474977, "learning_rate": 7.529777053284782e-06, "loss": 0.2158, "step": 4898 }, { "epoch": 0.35, "grad_norm": 3.8992274861529816, "learning_rate": 7.5287752805591476e-06, "loss": 0.2718, "step": 4899 }, { "epoch": 0.35, "grad_norm": 4.840469814957661, "learning_rate": 7.527773371415659e-06, "loss": 1.018, "step": 4900 }, { "epoch": 0.35, "grad_norm": 14.08121699740477, "learning_rate": 7.526771325908369e-06, "loss": 0.6707, "step": 4901 }, { "epoch": 0.35, "grad_norm": 14.375365383321684, "learning_rate": 7.525769144091326e-06, "loss": 1.2691, "step": 4902 }, { "epoch": 0.35, "grad_norm": 18.309931058564292, "learning_rate": 7.524766826018599e-06, "loss": 0.7706, "step": 4903 }, { "epoch": 0.35, "grad_norm": 21.166190723504887, "learning_rate": 7.52376437174426e-06, "loss": 0.7086, "step": 4904 }, { "epoch": 0.35, "grad_norm": 4.646552234183795, "learning_rate": 7.522761781322385e-06, "loss": 0.9232, "step": 4905 }, { "epoch": 0.35, "grad_norm": 3.734291451102557, "learning_rate": 7.521759054807061e-06, "loss": 0.2875, "step": 4906 }, { "epoch": 0.35, "grad_norm": 4.606186264395694, "learning_rate": 7.520756192252383e-06, "loss": 0.9624, "step": 4907 }, { "epoch": 0.35, "grad_norm": 3.9933079583668194, "learning_rate": 7.519753193712449e-06, "loss": 0.2143, "step": 4908 }, { "epoch": 0.35, "grad_norm": 3.5286120274373274, "learning_rate": 7.518750059241367e-06, "loss": 0.2158, "step": 4909 }, { "epoch": 0.35, "grad_norm": 3.4766819842353307, "learning_rate": 7.517746788893255e-06, "loss": 0.257, "step": 4910 }, { "epoch": 0.35, "grad_norm": 2.9556411813358903, "learning_rate": 7.516743382722232e-06, "loss": 0.2151, "step": 4911 }, { "epoch": 0.35, "grad_norm": 20.419859420747315, "learning_rate": 7.515739840782432e-06, "loss": 0.8476, "step": 4912 }, { "epoch": 0.35, "grad_norm": 16.1064877289014, "learning_rate": 7.5147361631279895e-06, "loss": 0.7469, "step": 4913 }, { "epoch": 0.35, "grad_norm": 14.392293145637359, "learning_rate": 7.51373234981305e-06, "loss": 0.9545, "step": 4914 }, { "epoch": 0.35, "grad_norm": 7.320859405534382, "learning_rate": 7.5127284008917645e-06, "loss": 1.0207, "step": 4915 }, { "epoch": 0.35, "grad_norm": 2.8378984670876575, "learning_rate": 7.511724316418292e-06, "loss": 0.1237, "step": 4916 }, { "epoch": 0.35, "grad_norm": 3.5243910809559944, "learning_rate": 7.510720096446802e-06, "loss": 0.2251, "step": 4917 }, { "epoch": 0.35, "grad_norm": 4.359826505768412, "learning_rate": 7.509715741031464e-06, "loss": 0.1856, "step": 4918 }, { "epoch": 0.35, "grad_norm": 6.538071693617603, "learning_rate": 7.508711250226465e-06, "loss": 0.2566, "step": 4919 }, { "epoch": 0.35, "grad_norm": 18.056796999874223, "learning_rate": 7.507706624085988e-06, "loss": 0.9452, "step": 4920 }, { "epoch": 0.35, "grad_norm": 4.113846310286607, "learning_rate": 7.50670186266423e-06, "loss": 0.2302, "step": 4921 }, { "epoch": 0.35, "grad_norm": 3.900081614131944, "learning_rate": 7.505696966015397e-06, "loss": 0.128, "step": 4922 }, { "epoch": 0.35, "grad_norm": 19.7818169783611, "learning_rate": 7.504691934193695e-06, "loss": 0.8112, "step": 4923 }, { "epoch": 0.35, "grad_norm": 3.278489808704319, "learning_rate": 7.503686767253346e-06, "loss": 0.1758, "step": 4924 }, { "epoch": 0.35, "grad_norm": 6.551055187763821, "learning_rate": 7.502681465248571e-06, "loss": 0.6013, "step": 4925 }, { "epoch": 0.35, "grad_norm": 3.894626067802878, "learning_rate": 7.501676028233605e-06, "loss": 1.0025, "step": 4926 }, { "epoch": 0.35, "grad_norm": 4.367450683580414, "learning_rate": 7.500670456262687e-06, "loss": 0.9694, "step": 4927 }, { "epoch": 0.35, "grad_norm": 2.3684919667711033, "learning_rate": 7.499664749390063e-06, "loss": 0.1696, "step": 4928 }, { "epoch": 0.35, "grad_norm": 2.5874692509802926, "learning_rate": 7.498658907669988e-06, "loss": 0.1677, "step": 4929 }, { "epoch": 0.35, "grad_norm": 4.0683799726751015, "learning_rate": 7.497652931156723e-06, "loss": 0.2641, "step": 4930 }, { "epoch": 0.35, "grad_norm": 17.759311052808854, "learning_rate": 7.496646819904536e-06, "loss": 1.0308, "step": 4931 }, { "epoch": 0.35, "grad_norm": 3.0804255433428187, "learning_rate": 7.495640573967703e-06, "loss": 0.2472, "step": 4932 }, { "epoch": 0.35, "grad_norm": 14.391665633808548, "learning_rate": 7.494634193400508e-06, "loss": 1.2478, "step": 4933 }, { "epoch": 0.35, "grad_norm": 27.39471294205535, "learning_rate": 7.49362767825724e-06, "loss": 1.1761, "step": 4934 }, { "epoch": 0.35, "grad_norm": 4.803559131028254, "learning_rate": 7.492621028592199e-06, "loss": 0.7569, "step": 4935 }, { "epoch": 0.35, "grad_norm": 14.132117380501938, "learning_rate": 7.491614244459688e-06, "loss": 0.7408, "step": 4936 }, { "epoch": 0.35, "grad_norm": 3.4055934006452717, "learning_rate": 7.490607325914021e-06, "loss": 0.2491, "step": 4937 }, { "epoch": 0.35, "grad_norm": 21.54628983225991, "learning_rate": 7.4896002730095154e-06, "loss": 0.9851, "step": 4938 }, { "epoch": 0.35, "grad_norm": 3.264096569685648, "learning_rate": 7.488593085800498e-06, "loss": 0.1883, "step": 4939 }, { "epoch": 0.35, "grad_norm": 11.008940589529868, "learning_rate": 7.487585764341303e-06, "loss": 0.5364, "step": 4940 }, { "epoch": 0.35, "grad_norm": 3.1777260249540533, "learning_rate": 7.486578308686273e-06, "loss": 0.238, "step": 4941 }, { "epoch": 0.35, "grad_norm": 4.279290380210669, "learning_rate": 7.485570718889755e-06, "loss": 0.8613, "step": 4942 }, { "epoch": 0.35, "grad_norm": 3.671059179328861, "learning_rate": 7.484562995006106e-06, "loss": 0.3567, "step": 4943 }, { "epoch": 0.35, "grad_norm": 3.3588879705431838, "learning_rate": 7.483555137089689e-06, "loss": 0.4224, "step": 4944 }, { "epoch": 0.35, "grad_norm": 3.337389744125833, "learning_rate": 7.48254714519487e-06, "loss": 0.254, "step": 4945 }, { "epoch": 0.35, "grad_norm": 13.62175730761331, "learning_rate": 7.4815390193760314e-06, "loss": 1.306, "step": 4946 }, { "epoch": 0.35, "grad_norm": 4.14424809100031, "learning_rate": 7.480530759687555e-06, "loss": 0.3151, "step": 4947 }, { "epoch": 0.35, "grad_norm": 4.281262281805822, "learning_rate": 7.479522366183834e-06, "loss": 0.2181, "step": 4948 }, { "epoch": 0.35, "grad_norm": 4.9449632540004975, "learning_rate": 7.478513838919268e-06, "loss": 0.8563, "step": 4949 }, { "epoch": 0.35, "grad_norm": 22.56729227644947, "learning_rate": 7.4775051779482615e-06, "loss": 1.0825, "step": 4950 }, { "epoch": 0.36, "grad_norm": 3.417012891457133, "learning_rate": 7.476496383325227e-06, "loss": 0.2152, "step": 4951 }, { "epoch": 0.36, "grad_norm": 4.535384694455368, "learning_rate": 7.475487455104589e-06, "loss": 0.8252, "step": 4952 }, { "epoch": 0.36, "grad_norm": 4.266014011946287, "learning_rate": 7.474478393340769e-06, "loss": 0.2764, "step": 4953 }, { "epoch": 0.36, "grad_norm": 4.3769236661615105, "learning_rate": 7.473469198088209e-06, "loss": 0.1918, "step": 4954 }, { "epoch": 0.36, "grad_norm": 3.914246593952578, "learning_rate": 7.47245986940135e-06, "loss": 0.2903, "step": 4955 }, { "epoch": 0.36, "grad_norm": 3.8024488357229247, "learning_rate": 7.471450407334637e-06, "loss": 0.2113, "step": 4956 }, { "epoch": 0.36, "grad_norm": 13.166274867897483, "learning_rate": 7.4704408119425296e-06, "loss": 0.6705, "step": 4957 }, { "epoch": 0.36, "grad_norm": 4.900024039115563, "learning_rate": 7.469431083279492e-06, "loss": 0.9515, "step": 4958 }, { "epoch": 0.36, "grad_norm": 14.081189058064334, "learning_rate": 7.468421221399995e-06, "loss": 0.6414, "step": 4959 }, { "epoch": 0.36, "grad_norm": 5.292526253260883, "learning_rate": 7.4674112263585166e-06, "loss": 1.2164, "step": 4960 }, { "epoch": 0.36, "grad_norm": 3.7288485786578915, "learning_rate": 7.466401098209544e-06, "loss": 0.2947, "step": 4961 }, { "epoch": 0.36, "grad_norm": 2.9881617516138794, "learning_rate": 7.465390837007566e-06, "loss": 0.1722, "step": 4962 }, { "epoch": 0.36, "grad_norm": 3.188582722317801, "learning_rate": 7.464380442807084e-06, "loss": 0.1734, "step": 4963 }, { "epoch": 0.36, "grad_norm": 4.2917046718990575, "learning_rate": 7.463369915662606e-06, "loss": 0.9006, "step": 4964 }, { "epoch": 0.36, "grad_norm": 2.943875378682345, "learning_rate": 7.462359255628645e-06, "loss": 0.1424, "step": 4965 }, { "epoch": 0.36, "grad_norm": 2.8561504181573465, "learning_rate": 7.461348462759723e-06, "loss": 0.2209, "step": 4966 }, { "epoch": 0.36, "grad_norm": 2.4701772828978528, "learning_rate": 7.460337537110369e-06, "loss": 0.1386, "step": 4967 }, { "epoch": 0.36, "grad_norm": 3.3057031970062374, "learning_rate": 7.459326478735116e-06, "loss": 0.2594, "step": 4968 }, { "epoch": 0.36, "grad_norm": 2.2878353770839697, "learning_rate": 7.45831528768851e-06, "loss": 0.1471, "step": 4969 }, { "epoch": 0.36, "grad_norm": 18.186319291049372, "learning_rate": 7.457303964025098e-06, "loss": 1.4852, "step": 4970 }, { "epoch": 0.36, "grad_norm": 24.4410140294192, "learning_rate": 7.456292507799438e-06, "loss": 1.1536, "step": 4971 }, { "epoch": 0.36, "grad_norm": 4.504946999117617, "learning_rate": 7.455280919066098e-06, "loss": 0.8293, "step": 4972 }, { "epoch": 0.36, "grad_norm": 17.560600617281363, "learning_rate": 7.4542691978796445e-06, "loss": 1.0683, "step": 4973 }, { "epoch": 0.36, "grad_norm": 3.7957588462527188, "learning_rate": 7.453257344294658e-06, "loss": 0.1948, "step": 4974 }, { "epoch": 0.36, "grad_norm": 4.604592503512674, "learning_rate": 7.4522453583657226e-06, "loss": 0.2572, "step": 4975 }, { "epoch": 0.36, "grad_norm": 3.3000540781198646, "learning_rate": 7.451233240147433e-06, "loss": 0.1445, "step": 4976 }, { "epoch": 0.36, "grad_norm": 18.771906248350138, "learning_rate": 7.4502209896943875e-06, "loss": 0.8719, "step": 4977 }, { "epoch": 0.36, "grad_norm": 4.986045790313431, "learning_rate": 7.449208607061197e-06, "loss": 1.097, "step": 4978 }, { "epoch": 0.36, "grad_norm": 16.306542933191803, "learning_rate": 7.448196092302471e-06, "loss": 1.0858, "step": 4979 }, { "epoch": 0.36, "grad_norm": 4.178037037612365, "learning_rate": 7.447183445472833e-06, "loss": 0.2324, "step": 4980 }, { "epoch": 0.36, "grad_norm": 5.529839205932373, "learning_rate": 7.446170666626912e-06, "loss": 0.9246, "step": 4981 }, { "epoch": 0.36, "grad_norm": 22.782702810574353, "learning_rate": 7.445157755819341e-06, "loss": 0.9617, "step": 4982 }, { "epoch": 0.36, "grad_norm": 4.511606736121004, "learning_rate": 7.4441447131047664e-06, "loss": 1.0153, "step": 4983 }, { "epoch": 0.36, "grad_norm": 3.769238454193082, "learning_rate": 7.443131538537834e-06, "loss": 0.189, "step": 4984 }, { "epoch": 0.36, "grad_norm": 13.924147930196574, "learning_rate": 7.442118232173206e-06, "loss": 1.1213, "step": 4985 }, { "epoch": 0.36, "grad_norm": 5.7585063878312, "learning_rate": 7.44110479406554e-06, "loss": 1.0712, "step": 4986 }, { "epoch": 0.36, "grad_norm": 4.83188399307158, "learning_rate": 7.440091224269512e-06, "loss": 0.2312, "step": 4987 }, { "epoch": 0.36, "grad_norm": 4.337648247921594, "learning_rate": 7.4390775228397985e-06, "loss": 0.4072, "step": 4988 }, { "epoch": 0.36, "grad_norm": 4.684539416165409, "learning_rate": 7.438063689831086e-06, "loss": 0.9396, "step": 4989 }, { "epoch": 0.36, "grad_norm": 4.513067293221739, "learning_rate": 7.437049725298063e-06, "loss": 1.0562, "step": 4990 }, { "epoch": 0.36, "grad_norm": 12.200617291437819, "learning_rate": 7.436035629295435e-06, "loss": 0.6936, "step": 4991 }, { "epoch": 0.36, "grad_norm": 3.239175950512088, "learning_rate": 7.435021401877906e-06, "loss": 0.139, "step": 4992 }, { "epoch": 0.36, "grad_norm": 23.499862392340876, "learning_rate": 7.434007043100188e-06, "loss": 1.2594, "step": 4993 }, { "epoch": 0.36, "grad_norm": 4.585000752636965, "learning_rate": 7.432992553017002e-06, "loss": 0.2158, "step": 4994 }, { "epoch": 0.36, "grad_norm": 18.78532757223045, "learning_rate": 7.4319779316830785e-06, "loss": 0.8374, "step": 4995 }, { "epoch": 0.36, "grad_norm": 4.394090566345815, "learning_rate": 7.430963179153151e-06, "loss": 0.2583, "step": 4996 }, { "epoch": 0.36, "grad_norm": 2.9634912518046304, "learning_rate": 7.429948295481964e-06, "loss": 0.1792, "step": 4997 }, { "epoch": 0.36, "grad_norm": 5.3066540647378915, "learning_rate": 7.428933280724262e-06, "loss": 0.8898, "step": 4998 }, { "epoch": 0.36, "grad_norm": 4.805527679359545, "learning_rate": 7.427918134934803e-06, "loss": 1.1672, "step": 4999 }, { "epoch": 0.36, "grad_norm": 4.886744228059118, "learning_rate": 7.426902858168351e-06, "loss": 1.0436, "step": 5000 }, { "epoch": 0.36, "grad_norm": 5.226067631272361, "learning_rate": 7.425887450479678e-06, "loss": 0.883, "step": 5001 }, { "epoch": 0.36, "grad_norm": 3.1316825368103314, "learning_rate": 7.4248719119235566e-06, "loss": 0.1905, "step": 5002 }, { "epoch": 0.36, "grad_norm": 8.93306825599153, "learning_rate": 7.4238562425547766e-06, "loss": 0.7728, "step": 5003 }, { "epoch": 0.36, "grad_norm": 8.030479693806688, "learning_rate": 7.422840442428126e-06, "loss": 0.4616, "step": 5004 }, { "epoch": 0.36, "grad_norm": 5.149547882091597, "learning_rate": 7.421824511598405e-06, "loss": 0.9848, "step": 5005 }, { "epoch": 0.36, "grad_norm": 3.110666803551705, "learning_rate": 7.420808450120419e-06, "loss": 0.1685, "step": 5006 }, { "epoch": 0.36, "grad_norm": 3.2370979485999234, "learning_rate": 7.419792258048979e-06, "loss": 0.2023, "step": 5007 }, { "epoch": 0.36, "grad_norm": 6.479931193883311, "learning_rate": 7.418775935438907e-06, "loss": 0.6049, "step": 5008 }, { "epoch": 0.36, "grad_norm": 22.14734606400215, "learning_rate": 7.417759482345028e-06, "loss": 1.5573, "step": 5009 }, { "epoch": 0.36, "grad_norm": 4.09336503610485, "learning_rate": 7.416742898822179e-06, "loss": 0.1938, "step": 5010 }, { "epoch": 0.36, "grad_norm": 15.364720913669712, "learning_rate": 7.415726184925196e-06, "loss": 0.6982, "step": 5011 }, { "epoch": 0.36, "grad_norm": 3.964027758035548, "learning_rate": 7.414709340708931e-06, "loss": 0.3361, "step": 5012 }, { "epoch": 0.36, "grad_norm": 10.09900950717284, "learning_rate": 7.413692366228236e-06, "loss": 0.7249, "step": 5013 }, { "epoch": 0.36, "grad_norm": 13.727958536591762, "learning_rate": 7.412675261537974e-06, "loss": 0.8282, "step": 5014 }, { "epoch": 0.36, "grad_norm": 27.70795952653584, "learning_rate": 7.411658026693016e-06, "loss": 1.3251, "step": 5015 }, { "epoch": 0.36, "grad_norm": 4.817513976276392, "learning_rate": 7.410640661748234e-06, "loss": 0.3257, "step": 5016 }, { "epoch": 0.36, "grad_norm": 5.831223296534801, "learning_rate": 7.4096231667585135e-06, "loss": 1.0021, "step": 5017 }, { "epoch": 0.36, "grad_norm": 4.348372181405149, "learning_rate": 7.408605541778746e-06, "loss": 0.3004, "step": 5018 }, { "epoch": 0.36, "grad_norm": 3.9133994444242246, "learning_rate": 7.407587786863823e-06, "loss": 0.2313, "step": 5019 }, { "epoch": 0.36, "grad_norm": 3.8917785189802214, "learning_rate": 7.406569902068657e-06, "loss": 0.281, "step": 5020 }, { "epoch": 0.36, "grad_norm": 5.485696242618189, "learning_rate": 7.40555188744815e-06, "loss": 1.0298, "step": 5021 }, { "epoch": 0.36, "grad_norm": 4.195551523871391, "learning_rate": 7.404533743057225e-06, "loss": 0.2648, "step": 5022 }, { "epoch": 0.36, "grad_norm": 3.9143204182750004, "learning_rate": 7.403515468950807e-06, "loss": 0.2014, "step": 5023 }, { "epoch": 0.36, "grad_norm": 5.304301810444222, "learning_rate": 7.402497065183827e-06, "loss": 1.0852, "step": 5024 }, { "epoch": 0.36, "grad_norm": 3.8168969303600466, "learning_rate": 7.401478531811223e-06, "loss": 0.1881, "step": 5025 }, { "epoch": 0.36, "grad_norm": 3.612127487730575, "learning_rate": 7.400459868887946e-06, "loss": 0.1869, "step": 5026 }, { "epoch": 0.36, "grad_norm": 13.600929145512131, "learning_rate": 7.399441076468941e-06, "loss": 0.6645, "step": 5027 }, { "epoch": 0.36, "grad_norm": 3.548109164135903, "learning_rate": 7.398422154609175e-06, "loss": 0.2736, "step": 5028 }, { "epoch": 0.36, "grad_norm": 30.470297030167806, "learning_rate": 7.397403103363611e-06, "loss": 1.077, "step": 5029 }, { "epoch": 0.36, "grad_norm": 11.609119832267202, "learning_rate": 7.396383922787225e-06, "loss": 0.9284, "step": 5030 }, { "epoch": 0.36, "grad_norm": 3.2978856871206, "learning_rate": 7.395364612934997e-06, "loss": 0.179, "step": 5031 }, { "epoch": 0.36, "grad_norm": 4.1790898204288665, "learning_rate": 7.394345173861915e-06, "loss": 0.8591, "step": 5032 }, { "epoch": 0.36, "grad_norm": 9.229692096140775, "learning_rate": 7.393325605622974e-06, "loss": 0.6159, "step": 5033 }, { "epoch": 0.36, "grad_norm": 2.8603244586854006, "learning_rate": 7.392305908273176e-06, "loss": 0.1457, "step": 5034 }, { "epoch": 0.36, "grad_norm": 2.5500141815419304, "learning_rate": 7.391286081867529e-06, "loss": 0.2271, "step": 5035 }, { "epoch": 0.36, "grad_norm": 3.159194175187927, "learning_rate": 7.39026612646105e-06, "loss": 0.1959, "step": 5036 }, { "epoch": 0.36, "grad_norm": 2.9862329944945167, "learning_rate": 7.38924604210876e-06, "loss": 0.2759, "step": 5037 }, { "epoch": 0.36, "grad_norm": 4.285305875252543, "learning_rate": 7.388225828865691e-06, "loss": 0.828, "step": 5038 }, { "epoch": 0.36, "grad_norm": 3.1540260420765986, "learning_rate": 7.387205486786878e-06, "loss": 0.2112, "step": 5039 }, { "epoch": 0.36, "grad_norm": 3.0384602195316908, "learning_rate": 7.386185015927365e-06, "loss": 0.1705, "step": 5040 }, { "epoch": 0.36, "grad_norm": 4.719981750443035, "learning_rate": 7.3851644163422036e-06, "loss": 1.1061, "step": 5041 }, { "epoch": 0.36, "grad_norm": 20.25140788866304, "learning_rate": 7.384143688086449e-06, "loss": 0.7578, "step": 5042 }, { "epoch": 0.36, "grad_norm": 4.360272453833547, "learning_rate": 7.383122831215168e-06, "loss": 0.9686, "step": 5043 }, { "epoch": 0.36, "grad_norm": 4.073577396373944, "learning_rate": 7.382101845783429e-06, "loss": 0.8847, "step": 5044 }, { "epoch": 0.36, "grad_norm": 3.9010791857534506, "learning_rate": 7.381080731846313e-06, "loss": 0.204, "step": 5045 }, { "epoch": 0.36, "grad_norm": 19.77382059999322, "learning_rate": 7.380059489458906e-06, "loss": 0.9232, "step": 5046 }, { "epoch": 0.36, "grad_norm": 2.9256072431456968, "learning_rate": 7.379038118676297e-06, "loss": 0.2369, "step": 5047 }, { "epoch": 0.36, "grad_norm": 3.4205500475333013, "learning_rate": 7.3780166195535874e-06, "loss": 0.2117, "step": 5048 }, { "epoch": 0.36, "grad_norm": 3.4626817358419286, "learning_rate": 7.376994992145881e-06, "loss": 0.225, "step": 5049 }, { "epoch": 0.36, "grad_norm": 3.496756023133445, "learning_rate": 7.3759732365082936e-06, "loss": 0.2112, "step": 5050 }, { "epoch": 0.36, "grad_norm": 4.76432479735536, "learning_rate": 7.374951352695943e-06, "loss": 1.1336, "step": 5051 }, { "epoch": 0.36, "grad_norm": 3.712489733818781, "learning_rate": 7.373929340763958e-06, "loss": 0.2286, "step": 5052 }, { "epoch": 0.36, "grad_norm": 4.939694178909087, "learning_rate": 7.3729072007674695e-06, "loss": 1.0211, "step": 5053 }, { "epoch": 0.36, "grad_norm": 4.0497454142805385, "learning_rate": 7.3718849327616195e-06, "loss": 0.2542, "step": 5054 }, { "epoch": 0.36, "grad_norm": 30.684224137448055, "learning_rate": 7.370862536801555e-06, "loss": 1.3202, "step": 5055 }, { "epoch": 0.36, "grad_norm": 5.373559126170542, "learning_rate": 7.369840012942432e-06, "loss": 1.0762, "step": 5056 }, { "epoch": 0.36, "grad_norm": 3.867341814090762, "learning_rate": 7.368817361239411e-06, "loss": 0.1501, "step": 5057 }, { "epoch": 0.36, "grad_norm": 15.778800758467538, "learning_rate": 7.367794581747658e-06, "loss": 1.0858, "step": 5058 }, { "epoch": 0.36, "grad_norm": 3.9572132921253047, "learning_rate": 7.366771674522351e-06, "loss": 0.2637, "step": 5059 }, { "epoch": 0.36, "grad_norm": 10.23245257177099, "learning_rate": 7.365748639618671e-06, "loss": 0.9743, "step": 5060 }, { "epoch": 0.36, "grad_norm": 2.9666005391000168, "learning_rate": 7.3647254770918065e-06, "loss": 0.1127, "step": 5061 }, { "epoch": 0.36, "grad_norm": 3.6897395640282586, "learning_rate": 7.363702186996953e-06, "loss": 0.3104, "step": 5062 }, { "epoch": 0.36, "grad_norm": 4.535745517770832, "learning_rate": 7.362678769389314e-06, "loss": 0.3905, "step": 5063 }, { "epoch": 0.36, "grad_norm": 44.80560683627233, "learning_rate": 7.361655224324099e-06, "loss": 0.3166, "step": 5064 }, { "epoch": 0.36, "grad_norm": 4.822964153556403, "learning_rate": 7.360631551856523e-06, "loss": 0.823, "step": 5065 }, { "epoch": 0.36, "grad_norm": 3.5490250336462825, "learning_rate": 7.359607752041812e-06, "loss": 0.3189, "step": 5066 }, { "epoch": 0.36, "grad_norm": 4.0830674516503915, "learning_rate": 7.358583824935193e-06, "loss": 0.2494, "step": 5067 }, { "epoch": 0.36, "grad_norm": 17.909034283404505, "learning_rate": 7.357559770591904e-06, "loss": 0.6413, "step": 5068 }, { "epoch": 0.36, "grad_norm": 5.798717077117766, "learning_rate": 7.356535589067191e-06, "loss": 0.9989, "step": 5069 }, { "epoch": 0.36, "grad_norm": 3.2011485153487427, "learning_rate": 7.355511280416301e-06, "loss": 0.1397, "step": 5070 }, { "epoch": 0.36, "grad_norm": 16.458309539987823, "learning_rate": 7.354486844694494e-06, "loss": 0.9086, "step": 5071 }, { "epoch": 0.36, "grad_norm": 12.245592561406752, "learning_rate": 7.353462281957035e-06, "loss": 0.9164, "step": 5072 }, { "epoch": 0.36, "grad_norm": 19.54846536099153, "learning_rate": 7.352437592259193e-06, "loss": 0.9892, "step": 5073 }, { "epoch": 0.36, "grad_norm": 11.411483587101372, "learning_rate": 7.351412775656248e-06, "loss": 0.9053, "step": 5074 }, { "epoch": 0.36, "grad_norm": 4.330774055817563, "learning_rate": 7.3503878322034834e-06, "loss": 0.7964, "step": 5075 }, { "epoch": 0.36, "grad_norm": 10.17072419596849, "learning_rate": 7.349362761956194e-06, "loss": 0.6639, "step": 5076 }, { "epoch": 0.36, "grad_norm": 23.099464796406014, "learning_rate": 7.348337564969674e-06, "loss": 1.052, "step": 5077 }, { "epoch": 0.36, "grad_norm": 15.445958963799214, "learning_rate": 7.347312241299234e-06, "loss": 0.5033, "step": 5078 }, { "epoch": 0.36, "grad_norm": 3.2929806105181245, "learning_rate": 7.346286791000181e-06, "loss": 0.286, "step": 5079 }, { "epoch": 0.36, "grad_norm": 3.4850619576133046, "learning_rate": 7.345261214127839e-06, "loss": 0.2152, "step": 5080 }, { "epoch": 0.36, "grad_norm": 19.820057730702857, "learning_rate": 7.3442355107375296e-06, "loss": 0.8969, "step": 5081 }, { "epoch": 0.36, "grad_norm": 12.058459001504588, "learning_rate": 7.34320968088459e-06, "loss": 0.6077, "step": 5082 }, { "epoch": 0.36, "grad_norm": 11.252516300817204, "learning_rate": 7.342183724624357e-06, "loss": 0.9086, "step": 5083 }, { "epoch": 0.36, "grad_norm": 4.53148052614586, "learning_rate": 7.3411576420121765e-06, "loss": 0.9217, "step": 5084 }, { "epoch": 0.36, "grad_norm": 3.203010429127624, "learning_rate": 7.3401314331034035e-06, "loss": 0.2093, "step": 5085 }, { "epoch": 0.36, "grad_norm": 2.8962551909800633, "learning_rate": 7.339105097953398e-06, "loss": 0.2405, "step": 5086 }, { "epoch": 0.36, "grad_norm": 12.183330004076499, "learning_rate": 7.338078636617526e-06, "loss": 0.9937, "step": 5087 }, { "epoch": 0.36, "grad_norm": 4.781363800526328, "learning_rate": 7.337052049151162e-06, "loss": 1.0481, "step": 5088 }, { "epoch": 0.36, "grad_norm": 16.575951337848792, "learning_rate": 7.3360253356096865e-06, "loss": 0.9476, "step": 5089 }, { "epoch": 0.36, "grad_norm": 3.2983245474580816, "learning_rate": 7.3349984960484856e-06, "loss": 0.1617, "step": 5090 }, { "epoch": 0.37, "grad_norm": 5.06934528591169, "learning_rate": 7.333971530522954e-06, "loss": 1.011, "step": 5091 }, { "epoch": 0.37, "grad_norm": 4.204692447556121, "learning_rate": 7.332944439088494e-06, "loss": 0.8269, "step": 5092 }, { "epoch": 0.37, "grad_norm": 3.1937571578466057, "learning_rate": 7.331917221800512e-06, "loss": 0.2646, "step": 5093 }, { "epoch": 0.37, "grad_norm": 41.95238839095512, "learning_rate": 7.3308898787144235e-06, "loss": 1.4706, "step": 5094 }, { "epoch": 0.37, "grad_norm": 12.350474739705989, "learning_rate": 7.3298624098856486e-06, "loss": 1.2305, "step": 5095 }, { "epoch": 0.37, "grad_norm": 22.7311629154556, "learning_rate": 7.3288348153696145e-06, "loss": 1.2186, "step": 5096 }, { "epoch": 0.37, "grad_norm": 21.975797928373904, "learning_rate": 7.327807095221757e-06, "loss": 0.5669, "step": 5097 }, { "epoch": 0.37, "grad_norm": 3.2860199425479455, "learning_rate": 7.326779249497519e-06, "loss": 0.184, "step": 5098 }, { "epoch": 0.37, "grad_norm": 14.575993973603527, "learning_rate": 7.325751278252346e-06, "loss": 0.8859, "step": 5099 }, { "epoch": 0.37, "grad_norm": 3.0748137329837464, "learning_rate": 7.324723181541698e-06, "loss": 0.1369, "step": 5100 }, { "epoch": 0.37, "grad_norm": 5.4942288360623435, "learning_rate": 7.323694959421033e-06, "loss": 1.0715, "step": 5101 }, { "epoch": 0.37, "grad_norm": 5.4369074840733305, "learning_rate": 7.322666611945819e-06, "loss": 0.8369, "step": 5102 }, { "epoch": 0.37, "grad_norm": 4.455761837092375, "learning_rate": 7.321638139171535e-06, "loss": 0.3, "step": 5103 }, { "epoch": 0.37, "grad_norm": 5.0793607409031285, "learning_rate": 7.320609541153661e-06, "loss": 1.06, "step": 5104 }, { "epoch": 0.37, "grad_norm": 3.741437702316067, "learning_rate": 7.3195808179476846e-06, "loss": 0.2198, "step": 5105 }, { "epoch": 0.37, "grad_norm": 4.374564551392739, "learning_rate": 7.318551969609104e-06, "loss": 0.7778, "step": 5106 }, { "epoch": 0.37, "grad_norm": 15.917662208224701, "learning_rate": 7.3175229961934215e-06, "loss": 0.8755, "step": 5107 }, { "epoch": 0.37, "grad_norm": 3.3949092630078956, "learning_rate": 7.316493897756145e-06, "loss": 0.1437, "step": 5108 }, { "epoch": 0.37, "grad_norm": 14.711275331957385, "learning_rate": 7.315464674352791e-06, "loss": 1.0099, "step": 5109 }, { "epoch": 0.37, "grad_norm": 4.280337383664087, "learning_rate": 7.3144353260388825e-06, "loss": 0.7414, "step": 5110 }, { "epoch": 0.37, "grad_norm": 3.5193819262114068, "learning_rate": 7.313405852869949e-06, "loss": 0.1935, "step": 5111 }, { "epoch": 0.37, "grad_norm": 4.646834712920007, "learning_rate": 7.312376254901528e-06, "loss": 0.2025, "step": 5112 }, { "epoch": 0.37, "grad_norm": 3.535896169463832, "learning_rate": 7.311346532189159e-06, "loss": 0.1492, "step": 5113 }, { "epoch": 0.37, "grad_norm": 23.667902969371283, "learning_rate": 7.310316684788393e-06, "loss": 1.2606, "step": 5114 }, { "epoch": 0.37, "grad_norm": 5.344868345113662, "learning_rate": 7.309286712754787e-06, "loss": 0.8796, "step": 5115 }, { "epoch": 0.37, "grad_norm": 3.969048256022415, "learning_rate": 7.308256616143905e-06, "loss": 0.2288, "step": 5116 }, { "epoch": 0.37, "grad_norm": 5.251773721581272, "learning_rate": 7.307226395011316e-06, "loss": 0.2179, "step": 5117 }, { "epoch": 0.37, "grad_norm": 3.465192253183377, "learning_rate": 7.306196049412595e-06, "loss": 0.2277, "step": 5118 }, { "epoch": 0.37, "grad_norm": 9.853690014936912, "learning_rate": 7.305165579403327e-06, "loss": 0.5981, "step": 5119 }, { "epoch": 0.37, "grad_norm": 2.0504779519341416, "learning_rate": 7.304134985039102e-06, "loss": 0.1241, "step": 5120 }, { "epoch": 0.37, "grad_norm": 2.9128428663060917, "learning_rate": 7.3031042663755156e-06, "loss": 0.2122, "step": 5121 }, { "epoch": 0.37, "grad_norm": 2.871396109858282, "learning_rate": 7.302073423468173e-06, "loss": 0.2072, "step": 5122 }, { "epoch": 0.37, "grad_norm": 3.2947380328759883, "learning_rate": 7.301042456372681e-06, "loss": 0.1976, "step": 5123 }, { "epoch": 0.37, "grad_norm": 8.174283652903691, "learning_rate": 7.3000113651446605e-06, "loss": 0.54, "step": 5124 }, { "epoch": 0.37, "grad_norm": 20.82368036806977, "learning_rate": 7.298980149839731e-06, "loss": 1.6937, "step": 5125 }, { "epoch": 0.37, "grad_norm": 21.622085173572877, "learning_rate": 7.297948810513526e-06, "loss": 0.9281, "step": 5126 }, { "epoch": 0.37, "grad_norm": 36.56502023832845, "learning_rate": 7.29691734722168e-06, "loss": 1.5361, "step": 5127 }, { "epoch": 0.37, "grad_norm": 17.922777380459593, "learning_rate": 7.295885760019838e-06, "loss": 0.538, "step": 5128 }, { "epoch": 0.37, "grad_norm": 6.307002448179862, "learning_rate": 7.294854048963649e-06, "loss": 1.0638, "step": 5129 }, { "epoch": 0.37, "grad_norm": 3.208674037614361, "learning_rate": 7.293822214108771e-06, "loss": 0.189, "step": 5130 }, { "epoch": 0.37, "grad_norm": 2.9053555261937936, "learning_rate": 7.2927902555108665e-06, "loss": 0.2206, "step": 5131 }, { "epoch": 0.37, "grad_norm": 26.049495249688007, "learning_rate": 7.291758173225608e-06, "loss": 0.982, "step": 5132 }, { "epoch": 0.37, "grad_norm": 3.43951628055524, "learning_rate": 7.290725967308669e-06, "loss": 0.2169, "step": 5133 }, { "epoch": 0.37, "grad_norm": 9.912767680598147, "learning_rate": 7.2896936378157355e-06, "loss": 0.8287, "step": 5134 }, { "epoch": 0.37, "grad_norm": 4.431801867606621, "learning_rate": 7.288661184802498e-06, "loss": 0.9606, "step": 5135 }, { "epoch": 0.37, "grad_norm": 20.161538012571928, "learning_rate": 7.287628608324652e-06, "loss": 1.3019, "step": 5136 }, { "epoch": 0.37, "grad_norm": 17.684352740451903, "learning_rate": 7.286595908437902e-06, "loss": 1.65, "step": 5137 }, { "epoch": 0.37, "grad_norm": 14.508390109041647, "learning_rate": 7.2855630851979575e-06, "loss": 0.815, "step": 5138 }, { "epoch": 0.37, "grad_norm": 13.134553559925775, "learning_rate": 7.284530138660536e-06, "loss": 0.8522, "step": 5139 }, { "epoch": 0.37, "grad_norm": 19.17983869127026, "learning_rate": 7.283497068881361e-06, "loss": 1.0116, "step": 5140 }, { "epoch": 0.37, "grad_norm": 10.874340239794837, "learning_rate": 7.282463875916162e-06, "loss": 0.5788, "step": 5141 }, { "epoch": 0.37, "grad_norm": 3.8581498043313145, "learning_rate": 7.281430559820678e-06, "loss": 0.219, "step": 5142 }, { "epoch": 0.37, "grad_norm": 14.54474265170544, "learning_rate": 7.28039712065065e-06, "loss": 1.1927, "step": 5143 }, { "epoch": 0.37, "grad_norm": 4.504510743288238, "learning_rate": 7.27936355846183e-06, "loss": 0.2949, "step": 5144 }, { "epoch": 0.37, "grad_norm": 4.968779391291306, "learning_rate": 7.278329873309973e-06, "loss": 0.903, "step": 5145 }, { "epoch": 0.37, "grad_norm": 4.549878506831841, "learning_rate": 7.277296065250843e-06, "loss": 0.2668, "step": 5146 }, { "epoch": 0.37, "grad_norm": 3.4086729391605517, "learning_rate": 7.276262134340211e-06, "loss": 0.1424, "step": 5147 }, { "epoch": 0.37, "grad_norm": 14.915151013700315, "learning_rate": 7.275228080633853e-06, "loss": 0.8783, "step": 5148 }, { "epoch": 0.37, "grad_norm": 4.8932200509455726, "learning_rate": 7.274193904187552e-06, "loss": 0.7164, "step": 5149 }, { "epoch": 0.37, "grad_norm": 15.158552392852803, "learning_rate": 7.273159605057098e-06, "loss": 1.0892, "step": 5150 }, { "epoch": 0.37, "grad_norm": 11.788672422336676, "learning_rate": 7.2721251832982875e-06, "loss": 1.0173, "step": 5151 }, { "epoch": 0.37, "grad_norm": 5.767924822395769, "learning_rate": 7.271090638966924e-06, "loss": 1.0603, "step": 5152 }, { "epoch": 0.37, "grad_norm": 4.707987161648192, "learning_rate": 7.270055972118817e-06, "loss": 0.693, "step": 5153 }, { "epoch": 0.37, "grad_norm": 4.9626421097397495, "learning_rate": 7.269021182809785e-06, "loss": 0.899, "step": 5154 }, { "epoch": 0.37, "grad_norm": 3.9021048110638197, "learning_rate": 7.2679862710956465e-06, "loss": 0.2027, "step": 5155 }, { "epoch": 0.37, "grad_norm": 10.739072141003804, "learning_rate": 7.266951237032232e-06, "loss": 0.4716, "step": 5156 }, { "epoch": 0.37, "grad_norm": 3.4716933711350153, "learning_rate": 7.265916080675383e-06, "loss": 0.2453, "step": 5157 }, { "epoch": 0.37, "grad_norm": 29.843788855883837, "learning_rate": 7.264880802080936e-06, "loss": 1.0703, "step": 5158 }, { "epoch": 0.37, "grad_norm": 37.10534714837151, "learning_rate": 7.263845401304742e-06, "loss": 0.363, "step": 5159 }, { "epoch": 0.37, "grad_norm": 5.198340871054921, "learning_rate": 7.2628098784026615e-06, "loss": 1.1383, "step": 5160 }, { "epoch": 0.37, "grad_norm": 4.916385730836263, "learning_rate": 7.26177423343055e-06, "loss": 0.8037, "step": 5161 }, { "epoch": 0.37, "grad_norm": 3.718772383008109, "learning_rate": 7.26073846644428e-06, "loss": 0.1839, "step": 5162 }, { "epoch": 0.37, "grad_norm": 3.7671931444241973, "learning_rate": 7.259702577499728e-06, "loss": 0.2956, "step": 5163 }, { "epoch": 0.37, "grad_norm": 3.651758184802983, "learning_rate": 7.258666566652774e-06, "loss": 0.174, "step": 5164 }, { "epoch": 0.37, "grad_norm": 13.76805951102535, "learning_rate": 7.257630433959311e-06, "loss": 1.4137, "step": 5165 }, { "epoch": 0.37, "grad_norm": 10.948466615857248, "learning_rate": 7.256594179475231e-06, "loss": 0.8988, "step": 5166 }, { "epoch": 0.37, "grad_norm": 16.318250292939066, "learning_rate": 7.255557803256434e-06, "loss": 0.8255, "step": 5167 }, { "epoch": 0.37, "grad_norm": 4.970367990487645, "learning_rate": 7.254521305358835e-06, "loss": 0.1665, "step": 5168 }, { "epoch": 0.37, "grad_norm": 19.16284864236269, "learning_rate": 7.253484685838343e-06, "loss": 1.456, "step": 5169 }, { "epoch": 0.37, "grad_norm": 19.501552101594847, "learning_rate": 7.252447944750881e-06, "loss": 1.0891, "step": 5170 }, { "epoch": 0.37, "grad_norm": 4.461089023036669, "learning_rate": 7.251411082152381e-06, "loss": 0.3561, "step": 5171 }, { "epoch": 0.37, "grad_norm": 14.121432298221702, "learning_rate": 7.250374098098773e-06, "loss": 1.375, "step": 5172 }, { "epoch": 0.37, "grad_norm": 4.003346496690391, "learning_rate": 7.249336992646002e-06, "loss": 0.2582, "step": 5173 }, { "epoch": 0.37, "grad_norm": 4.888921174187767, "learning_rate": 7.248299765850016e-06, "loss": 0.3173, "step": 5174 }, { "epoch": 0.37, "grad_norm": 3.7288720272118607, "learning_rate": 7.247262417766765e-06, "loss": 0.1954, "step": 5175 }, { "epoch": 0.37, "grad_norm": 3.544113963210604, "learning_rate": 7.246224948452213e-06, "loss": 0.2031, "step": 5176 }, { "epoch": 0.37, "grad_norm": 4.095649592736614, "learning_rate": 7.245187357962329e-06, "loss": 0.2595, "step": 5177 }, { "epoch": 0.37, "grad_norm": 6.628657723256945, "learning_rate": 7.2441496463530845e-06, "loss": 0.9305, "step": 5178 }, { "epoch": 0.37, "grad_norm": 19.34645139588672, "learning_rate": 7.243111813680461e-06, "loss": 1.2114, "step": 5179 }, { "epoch": 0.37, "grad_norm": 3.301417835337582, "learning_rate": 7.242073860000447e-06, "loss": 0.1617, "step": 5180 }, { "epoch": 0.37, "grad_norm": 3.704505788441558, "learning_rate": 7.241035785369034e-06, "loss": 0.2242, "step": 5181 }, { "epoch": 0.37, "grad_norm": 40.66728485181557, "learning_rate": 7.239997589842223e-06, "loss": 1.381, "step": 5182 }, { "epoch": 0.37, "grad_norm": 4.952349030419297, "learning_rate": 7.238959273476022e-06, "loss": 0.8661, "step": 5183 }, { "epoch": 0.37, "grad_norm": 4.691812706055327, "learning_rate": 7.237920836326443e-06, "loss": 0.4145, "step": 5184 }, { "epoch": 0.37, "grad_norm": 5.800417412252591, "learning_rate": 7.236882278449504e-06, "loss": 0.2483, "step": 5185 }, { "epoch": 0.37, "grad_norm": 13.284649358973857, "learning_rate": 7.235843599901235e-06, "loss": 1.0441, "step": 5186 }, { "epoch": 0.37, "grad_norm": 13.865041250934642, "learning_rate": 7.234804800737667e-06, "loss": 0.8615, "step": 5187 }, { "epoch": 0.37, "grad_norm": 11.229340207945246, "learning_rate": 7.233765881014838e-06, "loss": 0.7673, "step": 5188 }, { "epoch": 0.37, "grad_norm": 4.083958655565921, "learning_rate": 7.232726840788795e-06, "loss": 0.341, "step": 5189 }, { "epoch": 0.37, "grad_norm": 3.347894016751063, "learning_rate": 7.231687680115591e-06, "loss": 0.1894, "step": 5190 }, { "epoch": 0.37, "grad_norm": 4.069109263183965, "learning_rate": 7.230648399051285e-06, "loss": 0.1907, "step": 5191 }, { "epoch": 0.37, "grad_norm": 4.940075497163057, "learning_rate": 7.22960899765194e-06, "loss": 0.984, "step": 5192 }, { "epoch": 0.37, "grad_norm": 4.744394383785658, "learning_rate": 7.228569475973628e-06, "loss": 0.9938, "step": 5193 }, { "epoch": 0.37, "grad_norm": 3.8170851745848435, "learning_rate": 7.227529834072431e-06, "loss": 0.7017, "step": 5194 }, { "epoch": 0.37, "grad_norm": 7.393386404067674, "learning_rate": 7.226490072004428e-06, "loss": 0.7859, "step": 5195 }, { "epoch": 0.37, "grad_norm": 17.539261772405553, "learning_rate": 7.225450189825717e-06, "loss": 0.4801, "step": 5196 }, { "epoch": 0.37, "grad_norm": 17.795310102527715, "learning_rate": 7.22441018759239e-06, "loss": 1.2261, "step": 5197 }, { "epoch": 0.37, "grad_norm": 3.970080237977594, "learning_rate": 7.2233700653605534e-06, "loss": 0.1566, "step": 5198 }, { "epoch": 0.37, "grad_norm": 3.562452229571169, "learning_rate": 7.222329823186317e-06, "loss": 0.1463, "step": 5199 }, { "epoch": 0.37, "grad_norm": 3.4944685781130325, "learning_rate": 7.2212894611258e-06, "loss": 0.162, "step": 5200 }, { "epoch": 0.37, "grad_norm": 3.7034275480248917, "learning_rate": 7.220248979235123e-06, "loss": 0.1767, "step": 5201 }, { "epoch": 0.37, "grad_norm": 3.447013672021007, "learning_rate": 7.219208377570419e-06, "loss": 0.1983, "step": 5202 }, { "epoch": 0.37, "grad_norm": 3.7487646950484774, "learning_rate": 7.218167656187823e-06, "loss": 0.1648, "step": 5203 }, { "epoch": 0.37, "grad_norm": 4.021083052907453, "learning_rate": 7.217126815143478e-06, "loss": 0.2412, "step": 5204 }, { "epoch": 0.37, "grad_norm": 4.845960194503775, "learning_rate": 7.216085854493534e-06, "loss": 0.8573, "step": 5205 }, { "epoch": 0.37, "grad_norm": 4.753627591202705, "learning_rate": 7.215044774294145e-06, "loss": 0.3036, "step": 5206 }, { "epoch": 0.37, "grad_norm": 5.027800231410216, "learning_rate": 7.214003574601476e-06, "loss": 0.3805, "step": 5207 }, { "epoch": 0.37, "grad_norm": 4.631765047729997, "learning_rate": 7.212962255471696e-06, "loss": 0.6777, "step": 5208 }, { "epoch": 0.37, "grad_norm": 9.696754131105521, "learning_rate": 7.211920816960976e-06, "loss": 0.7473, "step": 5209 }, { "epoch": 0.37, "grad_norm": 23.763591303566663, "learning_rate": 7.210879259125502e-06, "loss": 1.3621, "step": 5210 }, { "epoch": 0.37, "grad_norm": 4.269839302523393, "learning_rate": 7.209837582021461e-06, "loss": 0.2459, "step": 5211 }, { "epoch": 0.37, "grad_norm": 3.8335693627740177, "learning_rate": 7.2087957857050454e-06, "loss": 0.1572, "step": 5212 }, { "epoch": 0.37, "grad_norm": 4.152879034851642, "learning_rate": 7.207753870232459e-06, "loss": 0.2436, "step": 5213 }, { "epoch": 0.37, "grad_norm": 4.804437669324109, "learning_rate": 7.20671183565991e-06, "loss": 0.3078, "step": 5214 }, { "epoch": 0.37, "grad_norm": 3.5476230300347664, "learning_rate": 7.205669682043608e-06, "loss": 0.1463, "step": 5215 }, { "epoch": 0.37, "grad_norm": 10.487479078564325, "learning_rate": 7.204627409439775e-06, "loss": 0.6709, "step": 5216 }, { "epoch": 0.37, "grad_norm": 2.877480013358229, "learning_rate": 7.20358501790464e-06, "loss": 0.1571, "step": 5217 }, { "epoch": 0.37, "grad_norm": 4.6853428314187235, "learning_rate": 7.202542507494432e-06, "loss": 0.8714, "step": 5218 }, { "epoch": 0.37, "grad_norm": 3.2584875192986944, "learning_rate": 7.201499878265396e-06, "loss": 0.2233, "step": 5219 }, { "epoch": 0.37, "grad_norm": 4.067847494552348, "learning_rate": 7.200457130273772e-06, "loss": 0.2701, "step": 5220 }, { "epoch": 0.37, "grad_norm": 16.272151360575954, "learning_rate": 7.199414263575817e-06, "loss": 0.9124, "step": 5221 }, { "epoch": 0.37, "grad_norm": 8.465107048118067, "learning_rate": 7.198371278227787e-06, "loss": 0.2865, "step": 5222 }, { "epoch": 0.37, "grad_norm": 5.03691721402367, "learning_rate": 7.197328174285947e-06, "loss": 1.0698, "step": 5223 }, { "epoch": 0.37, "grad_norm": 5.372686677673046, "learning_rate": 7.1962849518065695e-06, "loss": 0.9891, "step": 5224 }, { "epoch": 0.37, "grad_norm": 26.14621866344018, "learning_rate": 7.195241610845934e-06, "loss": 0.6952, "step": 5225 }, { "epoch": 0.37, "grad_norm": 2.7908025232340425, "learning_rate": 7.1941981514603195e-06, "loss": 0.1274, "step": 5226 }, { "epoch": 0.37, "grad_norm": 4.015064480430473, "learning_rate": 7.1931545737060234e-06, "loss": 0.1648, "step": 5227 }, { "epoch": 0.37, "grad_norm": 3.290868111372761, "learning_rate": 7.19211087763934e-06, "loss": 0.1943, "step": 5228 }, { "epoch": 0.37, "grad_norm": 3.8686133793774893, "learning_rate": 7.19106706331657e-06, "loss": 0.3584, "step": 5229 }, { "epoch": 0.38, "grad_norm": 3.1627913260344336, "learning_rate": 7.190023130794026e-06, "loss": 0.2157, "step": 5230 }, { "epoch": 0.38, "grad_norm": 5.751613564859613, "learning_rate": 7.188979080128024e-06, "loss": 0.9967, "step": 5231 }, { "epoch": 0.38, "grad_norm": 3.432239615097085, "learning_rate": 7.187934911374886e-06, "loss": 0.2002, "step": 5232 }, { "epoch": 0.38, "grad_norm": 3.3159488014717673, "learning_rate": 7.186890624590943e-06, "loss": 0.2649, "step": 5233 }, { "epoch": 0.38, "grad_norm": 3.0345670048145834, "learning_rate": 7.185846219832528e-06, "loss": 0.2446, "step": 5234 }, { "epoch": 0.38, "grad_norm": 3.5621431351877653, "learning_rate": 7.184801697155983e-06, "loss": 0.2178, "step": 5235 }, { "epoch": 0.38, "grad_norm": 7.1159670880593024, "learning_rate": 7.183757056617655e-06, "loss": 0.4245, "step": 5236 }, { "epoch": 0.38, "grad_norm": 2.6416290823608204, "learning_rate": 7.182712298273902e-06, "loss": 0.1831, "step": 5237 }, { "epoch": 0.38, "grad_norm": 3.3351809989458734, "learning_rate": 7.181667422181081e-06, "loss": 0.2513, "step": 5238 }, { "epoch": 0.38, "grad_norm": 11.38952324323738, "learning_rate": 7.180622428395562e-06, "loss": 0.8172, "step": 5239 }, { "epoch": 0.38, "grad_norm": 13.43335581554199, "learning_rate": 7.179577316973717e-06, "loss": 0.6954, "step": 5240 }, { "epoch": 0.38, "grad_norm": 3.700240452864187, "learning_rate": 7.1785320879719246e-06, "loss": 0.2841, "step": 5241 }, { "epoch": 0.38, "grad_norm": 14.218176395856368, "learning_rate": 7.177486741446573e-06, "loss": 1.0792, "step": 5242 }, { "epoch": 0.38, "grad_norm": 3.3383944029495254, "learning_rate": 7.176441277454054e-06, "loss": 0.2601, "step": 5243 }, { "epoch": 0.38, "grad_norm": 2.971703028670374, "learning_rate": 7.175395696050766e-06, "loss": 0.1783, "step": 5244 }, { "epoch": 0.38, "grad_norm": 24.134619998030512, "learning_rate": 7.174349997293115e-06, "loss": 1.1059, "step": 5245 }, { "epoch": 0.38, "grad_norm": 10.193851142426633, "learning_rate": 7.173304181237512e-06, "loss": 0.7666, "step": 5246 }, { "epoch": 0.38, "grad_norm": 2.873374693246783, "learning_rate": 7.172258247940373e-06, "loss": 0.2161, "step": 5247 }, { "epoch": 0.38, "grad_norm": 10.571570540225746, "learning_rate": 7.171212197458126e-06, "loss": 0.6747, "step": 5248 }, { "epoch": 0.38, "grad_norm": 10.111404026488936, "learning_rate": 7.170166029847199e-06, "loss": 0.9714, "step": 5249 }, { "epoch": 0.38, "grad_norm": 4.5659460948637465, "learning_rate": 7.169119745164028e-06, "loss": 0.9507, "step": 5250 }, { "epoch": 0.38, "grad_norm": 4.421525830983269, "learning_rate": 7.1680733434650605e-06, "loss": 0.1972, "step": 5251 }, { "epoch": 0.38, "grad_norm": 3.2707942868114963, "learning_rate": 7.16702682480674e-06, "loss": 0.2164, "step": 5252 }, { "epoch": 0.38, "grad_norm": 16.230631004667035, "learning_rate": 7.165980189245525e-06, "loss": 1.014, "step": 5253 }, { "epoch": 0.38, "grad_norm": 3.1252243158183304, "learning_rate": 7.164933436837879e-06, "loss": 0.1344, "step": 5254 }, { "epoch": 0.38, "grad_norm": 12.66703053716142, "learning_rate": 7.1638865676402665e-06, "loss": 0.9852, "step": 5255 }, { "epoch": 0.38, "grad_norm": 3.8469583313018925, "learning_rate": 7.162839581709167e-06, "loss": 0.3401, "step": 5256 }, { "epoch": 0.38, "grad_norm": 4.102554829235965, "learning_rate": 7.161792479101059e-06, "loss": 0.2692, "step": 5257 }, { "epoch": 0.38, "grad_norm": 13.665273771932286, "learning_rate": 7.160745259872427e-06, "loss": 0.8615, "step": 5258 }, { "epoch": 0.38, "grad_norm": 3.1260763779162013, "learning_rate": 7.159697924079769e-06, "loss": 0.2355, "step": 5259 }, { "epoch": 0.38, "grad_norm": 7.413353472606439, "learning_rate": 7.158650471779581e-06, "loss": 0.5987, "step": 5260 }, { "epoch": 0.38, "grad_norm": 3.8821926464107346, "learning_rate": 7.157602903028372e-06, "loss": 0.2554, "step": 5261 }, { "epoch": 0.38, "grad_norm": 3.8527831311805665, "learning_rate": 7.156555217882655e-06, "loss": 0.2536, "step": 5262 }, { "epoch": 0.38, "grad_norm": 2.561537453010208, "learning_rate": 7.155507416398946e-06, "loss": 0.1341, "step": 5263 }, { "epoch": 0.38, "grad_norm": 4.669244076375328, "learning_rate": 7.154459498633769e-06, "loss": 0.9538, "step": 5264 }, { "epoch": 0.38, "grad_norm": 6.359060824401418, "learning_rate": 7.1534114646436596e-06, "loss": 0.8567, "step": 5265 }, { "epoch": 0.38, "grad_norm": 4.212045951061379, "learning_rate": 7.152363314485151e-06, "loss": 0.7176, "step": 5266 }, { "epoch": 0.38, "grad_norm": 52.73182233061756, "learning_rate": 7.151315048214789e-06, "loss": 0.9336, "step": 5267 }, { "epoch": 0.38, "grad_norm": 15.537160868071705, "learning_rate": 7.150266665889125e-06, "loss": 0.9614, "step": 5268 }, { "epoch": 0.38, "grad_norm": 3.764403095492064, "learning_rate": 7.149218167564712e-06, "loss": 0.2489, "step": 5269 }, { "epoch": 0.38, "grad_norm": 2.570584086150343, "learning_rate": 7.1481695532981145e-06, "loss": 0.1898, "step": 5270 }, { "epoch": 0.38, "grad_norm": 6.580685987699798, "learning_rate": 7.147120823145901e-06, "loss": 0.9601, "step": 5271 }, { "epoch": 0.38, "grad_norm": 14.99546492927305, "learning_rate": 7.146071977164646e-06, "loss": 0.9059, "step": 5272 }, { "epoch": 0.38, "grad_norm": 8.380289437421112, "learning_rate": 7.1450230154109325e-06, "loss": 0.5585, "step": 5273 }, { "epoch": 0.38, "grad_norm": 2.3630438665474744, "learning_rate": 7.1439739379413445e-06, "loss": 0.19, "step": 5274 }, { "epoch": 0.38, "grad_norm": 2.989738964340704, "learning_rate": 7.14292474481248e-06, "loss": 0.2251, "step": 5275 }, { "epoch": 0.38, "grad_norm": 4.492982176466272, "learning_rate": 7.141875436080935e-06, "loss": 0.3731, "step": 5276 }, { "epoch": 0.38, "grad_norm": 3.81374394358383, "learning_rate": 7.140826011803319e-06, "loss": 0.2708, "step": 5277 }, { "epoch": 0.38, "grad_norm": 3.14923274722293, "learning_rate": 7.139776472036242e-06, "loss": 0.1818, "step": 5278 }, { "epoch": 0.38, "grad_norm": 7.333620260971328, "learning_rate": 7.138726816836325e-06, "loss": 0.6299, "step": 5279 }, { "epoch": 0.38, "grad_norm": 2.3640252756433235, "learning_rate": 7.137677046260191e-06, "loss": 0.1338, "step": 5280 }, { "epoch": 0.38, "grad_norm": 2.7941300630389407, "learning_rate": 7.136627160364473e-06, "loss": 0.1819, "step": 5281 }, { "epoch": 0.38, "grad_norm": 3.7004468204170893, "learning_rate": 7.135577159205807e-06, "loss": 0.2831, "step": 5282 }, { "epoch": 0.38, "grad_norm": 3.3086102865097775, "learning_rate": 7.134527042840835e-06, "loss": 0.1941, "step": 5283 }, { "epoch": 0.38, "grad_norm": 5.78269886721169, "learning_rate": 7.1334768113262096e-06, "loss": 0.7272, "step": 5284 }, { "epoch": 0.38, "grad_norm": 5.202119669798837, "learning_rate": 7.132426464718587e-06, "loss": 0.8177, "step": 5285 }, { "epoch": 0.38, "grad_norm": 17.75996669202589, "learning_rate": 7.131376003074627e-06, "loss": 1.0392, "step": 5286 }, { "epoch": 0.38, "grad_norm": 4.555338313752583, "learning_rate": 7.130325426451002e-06, "loss": 0.9707, "step": 5287 }, { "epoch": 0.38, "grad_norm": 5.6034184543979455, "learning_rate": 7.129274734904381e-06, "loss": 0.7582, "step": 5288 }, { "epoch": 0.38, "grad_norm": 3.481912593286658, "learning_rate": 7.128223928491449e-06, "loss": 0.2019, "step": 5289 }, { "epoch": 0.38, "grad_norm": 14.806555393380554, "learning_rate": 7.127173007268891e-06, "loss": 1.0294, "step": 5290 }, { "epoch": 0.38, "grad_norm": 4.743406533312644, "learning_rate": 7.126121971293402e-06, "loss": 0.7529, "step": 5291 }, { "epoch": 0.38, "grad_norm": 4.232729057974868, "learning_rate": 7.125070820621679e-06, "loss": 0.2106, "step": 5292 }, { "epoch": 0.38, "grad_norm": 16.169871947580972, "learning_rate": 7.124019555310431e-06, "loss": 0.7119, "step": 5293 }, { "epoch": 0.38, "grad_norm": 3.636502000917031, "learning_rate": 7.122968175416367e-06, "loss": 0.2148, "step": 5294 }, { "epoch": 0.38, "grad_norm": 19.532650032890825, "learning_rate": 7.121916680996205e-06, "loss": 0.8634, "step": 5295 }, { "epoch": 0.38, "grad_norm": 3.063209394418668, "learning_rate": 7.1208650721066705e-06, "loss": 0.1533, "step": 5296 }, { "epoch": 0.38, "grad_norm": 3.4029943503632483, "learning_rate": 7.119813348804494e-06, "loss": 0.173, "step": 5297 }, { "epoch": 0.38, "grad_norm": 14.203260103899021, "learning_rate": 7.118761511146411e-06, "loss": 1.5383, "step": 5298 }, { "epoch": 0.38, "grad_norm": 21.667298138512873, "learning_rate": 7.117709559189165e-06, "loss": 1.169, "step": 5299 }, { "epoch": 0.38, "grad_norm": 23.44558101154734, "learning_rate": 7.1166574929895026e-06, "loss": 1.4029, "step": 5300 }, { "epoch": 0.38, "grad_norm": 3.569918075234504, "learning_rate": 7.11560531260418e-06, "loss": 0.159, "step": 5301 }, { "epoch": 0.38, "grad_norm": 4.436183379797033, "learning_rate": 7.11455301808996e-06, "loss": 0.2181, "step": 5302 }, { "epoch": 0.38, "grad_norm": 3.7458556091768176, "learning_rate": 7.11350060950361e-06, "loss": 0.2454, "step": 5303 }, { "epoch": 0.38, "grad_norm": 5.22262385830625, "learning_rate": 7.1124480869018995e-06, "loss": 1.0341, "step": 5304 }, { "epoch": 0.38, "grad_norm": 3.328076062312059, "learning_rate": 7.111395450341614e-06, "loss": 0.163, "step": 5305 }, { "epoch": 0.38, "grad_norm": 3.511294529362287, "learning_rate": 7.110342699879533e-06, "loss": 0.1969, "step": 5306 }, { "epoch": 0.38, "grad_norm": 4.8712715617478395, "learning_rate": 7.109289835572452e-06, "loss": 0.2519, "step": 5307 }, { "epoch": 0.38, "grad_norm": 14.601186357796278, "learning_rate": 7.10823685747717e-06, "loss": 1.0424, "step": 5308 }, { "epoch": 0.38, "grad_norm": 3.3679215906219833, "learning_rate": 7.107183765650487e-06, "loss": 0.1987, "step": 5309 }, { "epoch": 0.38, "grad_norm": 30.230260944086744, "learning_rate": 7.106130560149218e-06, "loss": 1.0052, "step": 5310 }, { "epoch": 0.38, "grad_norm": 14.107025181257281, "learning_rate": 7.105077241030178e-06, "loss": 1.3772, "step": 5311 }, { "epoch": 0.38, "grad_norm": 5.260129738573525, "learning_rate": 7.104023808350188e-06, "loss": 0.9615, "step": 5312 }, { "epoch": 0.38, "grad_norm": 9.880728357486817, "learning_rate": 7.102970262166078e-06, "loss": 0.7701, "step": 5313 }, { "epoch": 0.38, "grad_norm": 4.576280800023818, "learning_rate": 7.101916602534682e-06, "loss": 0.7519, "step": 5314 }, { "epoch": 0.38, "grad_norm": 5.834409886772487, "learning_rate": 7.100862829512841e-06, "loss": 0.3509, "step": 5315 }, { "epoch": 0.38, "grad_norm": 4.45964102693828, "learning_rate": 7.099808943157403e-06, "loss": 0.7922, "step": 5316 }, { "epoch": 0.38, "grad_norm": 3.3832733364487284, "learning_rate": 7.0987549435252225e-06, "loss": 0.2798, "step": 5317 }, { "epoch": 0.38, "grad_norm": 3.2490994656955063, "learning_rate": 7.097700830673155e-06, "loss": 0.1843, "step": 5318 }, { "epoch": 0.38, "grad_norm": 2.049378818704419, "learning_rate": 7.096646604658068e-06, "loss": 0.1731, "step": 5319 }, { "epoch": 0.38, "grad_norm": 3.1696237208296347, "learning_rate": 7.095592265536834e-06, "loss": 0.3176, "step": 5320 }, { "epoch": 0.38, "grad_norm": 16.225235219985823, "learning_rate": 7.094537813366328e-06, "loss": 0.867, "step": 5321 }, { "epoch": 0.38, "grad_norm": 13.117792259759081, "learning_rate": 7.093483248203436e-06, "loss": 0.9839, "step": 5322 }, { "epoch": 0.38, "grad_norm": 3.8735293371045976, "learning_rate": 7.0924285701050475e-06, "loss": 0.2898, "step": 5323 }, { "epoch": 0.38, "grad_norm": 20.984714574761487, "learning_rate": 7.091373779128058e-06, "loss": 0.8058, "step": 5324 }, { "epoch": 0.38, "grad_norm": 3.824906556525467, "learning_rate": 7.090318875329369e-06, "loss": 0.2222, "step": 5325 }, { "epoch": 0.38, "grad_norm": 4.06491281655825, "learning_rate": 7.089263858765889e-06, "loss": 0.2468, "step": 5326 }, { "epoch": 0.38, "grad_norm": 12.781482596071932, "learning_rate": 7.0882087294945325e-06, "loss": 0.5385, "step": 5327 }, { "epoch": 0.38, "grad_norm": 12.719544721966962, "learning_rate": 7.087153487572218e-06, "loss": 1.5102, "step": 5328 }, { "epoch": 0.38, "grad_norm": 4.457154217274957, "learning_rate": 7.086098133055874e-06, "loss": 0.2466, "step": 5329 }, { "epoch": 0.38, "grad_norm": 4.83337402976926, "learning_rate": 7.085042666002432e-06, "loss": 0.2029, "step": 5330 }, { "epoch": 0.38, "grad_norm": 10.290622075703102, "learning_rate": 7.083987086468831e-06, "loss": 0.7845, "step": 5331 }, { "epoch": 0.38, "grad_norm": 3.4528463096784177, "learning_rate": 7.082931394512013e-06, "loss": 0.266, "step": 5332 }, { "epoch": 0.38, "grad_norm": 15.20522727534086, "learning_rate": 7.081875590188934e-06, "loss": 1.382, "step": 5333 }, { "epoch": 0.38, "grad_norm": 3.7514838239333703, "learning_rate": 7.080819673556545e-06, "loss": 0.1554, "step": 5334 }, { "epoch": 0.38, "grad_norm": 4.089088403022635, "learning_rate": 7.079763644671811e-06, "loss": 0.2866, "step": 5335 }, { "epoch": 0.38, "grad_norm": 24.318134886153974, "learning_rate": 7.078707503591702e-06, "loss": 0.8022, "step": 5336 }, { "epoch": 0.38, "grad_norm": 3.5684032417353633, "learning_rate": 7.07765125037319e-06, "loss": 0.2467, "step": 5337 }, { "epoch": 0.38, "grad_norm": 10.689595184153589, "learning_rate": 7.076594885073257e-06, "loss": 0.8879, "step": 5338 }, { "epoch": 0.38, "grad_norm": 2.9105994752880258, "learning_rate": 7.075538407748892e-06, "loss": 0.2435, "step": 5339 }, { "epoch": 0.38, "grad_norm": 3.1634947695820825, "learning_rate": 7.074481818457084e-06, "loss": 0.2586, "step": 5340 }, { "epoch": 0.38, "grad_norm": 12.187873905101258, "learning_rate": 7.073425117254838e-06, "loss": 0.8942, "step": 5341 }, { "epoch": 0.38, "grad_norm": 16.491283609690495, "learning_rate": 7.072368304199152e-06, "loss": 1.6802, "step": 5342 }, { "epoch": 0.38, "grad_norm": 16.213443927733394, "learning_rate": 7.0713113793470415e-06, "loss": 1.052, "step": 5343 }, { "epoch": 0.38, "grad_norm": 15.145360811237678, "learning_rate": 7.07025434275552e-06, "loss": 0.7869, "step": 5344 }, { "epoch": 0.38, "grad_norm": 3.0990526154784668, "learning_rate": 7.069197194481616e-06, "loss": 0.1495, "step": 5345 }, { "epoch": 0.38, "grad_norm": 4.662758271428803, "learning_rate": 7.068139934582355e-06, "loss": 1.1887, "step": 5346 }, { "epoch": 0.38, "grad_norm": 4.931421670843952, "learning_rate": 7.067082563114773e-06, "loss": 0.9807, "step": 5347 }, { "epoch": 0.38, "grad_norm": 28.75686334611396, "learning_rate": 7.066025080135912e-06, "loss": 0.9552, "step": 5348 }, { "epoch": 0.38, "grad_norm": 9.784131410954043, "learning_rate": 7.064967485702816e-06, "loss": 0.8059, "step": 5349 }, { "epoch": 0.38, "grad_norm": 3.3940416202917207, "learning_rate": 7.063909779872542e-06, "loss": 0.265, "step": 5350 }, { "epoch": 0.38, "grad_norm": 3.600561235326537, "learning_rate": 7.062851962702148e-06, "loss": 0.2731, "step": 5351 }, { "epoch": 0.38, "grad_norm": 2.7258641867692996, "learning_rate": 7.0617940342486965e-06, "loss": 0.1793, "step": 5352 }, { "epoch": 0.38, "grad_norm": 3.741168459244083, "learning_rate": 7.0607359945692645e-06, "loss": 0.8314, "step": 5353 }, { "epoch": 0.38, "grad_norm": 3.098991676449514, "learning_rate": 7.059677843720924e-06, "loss": 0.249, "step": 5354 }, { "epoch": 0.38, "grad_norm": 3.0884170313095347, "learning_rate": 7.0586195817607595e-06, "loss": 0.1346, "step": 5355 }, { "epoch": 0.38, "grad_norm": 2.5868777253381565, "learning_rate": 7.057561208745862e-06, "loss": 0.148, "step": 5356 }, { "epoch": 0.38, "grad_norm": 4.484606185079594, "learning_rate": 7.056502724733325e-06, "loss": 0.761, "step": 5357 }, { "epoch": 0.38, "grad_norm": 3.613615516662333, "learning_rate": 7.05544412978025e-06, "loss": 0.2004, "step": 5358 }, { "epoch": 0.38, "grad_norm": 3.6242970746390943, "learning_rate": 7.054385423943747e-06, "loss": 0.203, "step": 5359 }, { "epoch": 0.38, "grad_norm": 4.202443736304035, "learning_rate": 7.053326607280924e-06, "loss": 0.9917, "step": 5360 }, { "epoch": 0.38, "grad_norm": 3.302683411436027, "learning_rate": 7.052267679848903e-06, "loss": 0.2438, "step": 5361 }, { "epoch": 0.38, "grad_norm": 2.6625584372968265, "learning_rate": 7.05120864170481e-06, "loss": 0.1728, "step": 5362 }, { "epoch": 0.38, "grad_norm": 18.361515345263513, "learning_rate": 7.050149492905775e-06, "loss": 1.2332, "step": 5363 }, { "epoch": 0.38, "grad_norm": 4.344873398080498, "learning_rate": 7.049090233508934e-06, "loss": 0.2366, "step": 5364 }, { "epoch": 0.38, "grad_norm": 69.22526682491434, "learning_rate": 7.048030863571432e-06, "loss": 1.5612, "step": 5365 }, { "epoch": 0.38, "grad_norm": 4.128669843687885, "learning_rate": 7.0469713831504175e-06, "loss": 0.2203, "step": 5366 }, { "epoch": 0.38, "grad_norm": 4.308651674155898, "learning_rate": 7.045911792303044e-06, "loss": 0.357, "step": 5367 }, { "epoch": 0.38, "grad_norm": 3.653724157257202, "learning_rate": 7.044852091086474e-06, "loss": 0.1645, "step": 5368 }, { "epoch": 0.38, "grad_norm": 16.283554944551927, "learning_rate": 7.043792279557873e-06, "loss": 0.9747, "step": 5369 }, { "epoch": 0.39, "grad_norm": 20.426899692194795, "learning_rate": 7.042732357774416e-06, "loss": 1.4742, "step": 5370 }, { "epoch": 0.39, "grad_norm": 6.760083304184898, "learning_rate": 7.04167232579328e-06, "loss": 0.6335, "step": 5371 }, { "epoch": 0.39, "grad_norm": 3.339208348753242, "learning_rate": 7.04061218367165e-06, "loss": 0.2572, "step": 5372 }, { "epoch": 0.39, "grad_norm": 6.233723403735524, "learning_rate": 7.039551931466717e-06, "loss": 0.3484, "step": 5373 }, { "epoch": 0.39, "grad_norm": 3.1344914067486944, "learning_rate": 7.038491569235677e-06, "loss": 0.1911, "step": 5374 }, { "epoch": 0.39, "grad_norm": 16.27410621384887, "learning_rate": 7.037431097035733e-06, "loss": 0.6501, "step": 5375 }, { "epoch": 0.39, "grad_norm": 5.876677174263795, "learning_rate": 7.036370514924095e-06, "loss": 0.6769, "step": 5376 }, { "epoch": 0.39, "grad_norm": 4.25979899912492, "learning_rate": 7.035309822957974e-06, "loss": 0.737, "step": 5377 }, { "epoch": 0.39, "grad_norm": 27.080910199311152, "learning_rate": 7.0342490211945925e-06, "loss": 0.8558, "step": 5378 }, { "epoch": 0.39, "grad_norm": 15.806497536305349, "learning_rate": 7.033188109691177e-06, "loss": 1.1061, "step": 5379 }, { "epoch": 0.39, "grad_norm": 19.242117871760673, "learning_rate": 7.032127088504959e-06, "loss": 1.811, "step": 5380 }, { "epoch": 0.39, "grad_norm": 4.914049883866628, "learning_rate": 7.031065957693176e-06, "loss": 0.7635, "step": 5381 }, { "epoch": 0.39, "grad_norm": 4.627668512284662, "learning_rate": 7.030004717313072e-06, "loss": 0.1678, "step": 5382 }, { "epoch": 0.39, "grad_norm": 18.886884438524618, "learning_rate": 7.028943367421899e-06, "loss": 0.7441, "step": 5383 }, { "epoch": 0.39, "grad_norm": 18.944149919828728, "learning_rate": 7.02788190807691e-06, "loss": 1.0433, "step": 5384 }, { "epoch": 0.39, "grad_norm": 45.93445233112687, "learning_rate": 7.026820339335369e-06, "loss": 1.1054, "step": 5385 }, { "epoch": 0.39, "grad_norm": 4.888438702981629, "learning_rate": 7.0257586612545414e-06, "loss": 0.9123, "step": 5386 }, { "epoch": 0.39, "grad_norm": 4.14000907435613, "learning_rate": 7.024696873891704e-06, "loss": 0.2632, "step": 5387 }, { "epoch": 0.39, "grad_norm": 15.63163510949734, "learning_rate": 7.0236349773041315e-06, "loss": 0.8097, "step": 5388 }, { "epoch": 0.39, "grad_norm": 24.567101991262, "learning_rate": 7.0225729715491135e-06, "loss": 0.9362, "step": 5389 }, { "epoch": 0.39, "grad_norm": 5.087070863106937, "learning_rate": 7.021510856683939e-06, "loss": 0.9717, "step": 5390 }, { "epoch": 0.39, "grad_norm": 4.565902355601726, "learning_rate": 7.020448632765906e-06, "loss": 0.3001, "step": 5391 }, { "epoch": 0.39, "grad_norm": 4.562361336858531, "learning_rate": 7.019386299852316e-06, "loss": 0.2998, "step": 5392 }, { "epoch": 0.39, "grad_norm": 21.13853674987131, "learning_rate": 7.018323858000481e-06, "loss": 1.1248, "step": 5393 }, { "epoch": 0.39, "grad_norm": 5.403107425170702, "learning_rate": 7.0172613072677125e-06, "loss": 0.8219, "step": 5394 }, { "epoch": 0.39, "grad_norm": 15.72629780639952, "learning_rate": 7.016198647711332e-06, "loss": 0.8995, "step": 5395 }, { "epoch": 0.39, "grad_norm": 20.34064361599145, "learning_rate": 7.015135879388667e-06, "loss": 1.3856, "step": 5396 }, { "epoch": 0.39, "grad_norm": 12.131486771400775, "learning_rate": 7.014073002357047e-06, "loss": 0.7176, "step": 5397 }, { "epoch": 0.39, "grad_norm": 9.206270009239987, "learning_rate": 7.013010016673813e-06, "loss": 0.673, "step": 5398 }, { "epoch": 0.39, "grad_norm": 3.272247458572107, "learning_rate": 7.01194692239631e-06, "loss": 0.1894, "step": 5399 }, { "epoch": 0.39, "grad_norm": 3.419350788674439, "learning_rate": 7.0108837195818845e-06, "loss": 0.2906, "step": 5400 }, { "epoch": 0.39, "grad_norm": 33.10659209813253, "learning_rate": 7.009820408287895e-06, "loss": 0.5192, "step": 5401 }, { "epoch": 0.39, "grad_norm": 3.349803059983837, "learning_rate": 7.0087569885717015e-06, "loss": 0.1204, "step": 5402 }, { "epoch": 0.39, "grad_norm": 4.412599696530582, "learning_rate": 7.007693460490672e-06, "loss": 0.9333, "step": 5403 }, { "epoch": 0.39, "grad_norm": 30.368079661773116, "learning_rate": 7.00662982410218e-06, "loss": 1.1065, "step": 5404 }, { "epoch": 0.39, "grad_norm": 4.857216468745961, "learning_rate": 7.005566079463604e-06, "loss": 1.0938, "step": 5405 }, { "epoch": 0.39, "grad_norm": 3.166579376993311, "learning_rate": 7.004502226632329e-06, "loss": 0.1787, "step": 5406 }, { "epoch": 0.39, "grad_norm": 5.499910744764988, "learning_rate": 7.003438265665748e-06, "loss": 1.0438, "step": 5407 }, { "epoch": 0.39, "grad_norm": 20.655932953274384, "learning_rate": 7.0023741966212555e-06, "loss": 1.0114, "step": 5408 }, { "epoch": 0.39, "grad_norm": 4.158666889942012, "learning_rate": 7.0013100195562535e-06, "loss": 0.2423, "step": 5409 }, { "epoch": 0.39, "grad_norm": 3.898815185930569, "learning_rate": 7.0002457345281526e-06, "loss": 0.224, "step": 5410 }, { "epoch": 0.39, "grad_norm": 3.4258124311596934, "learning_rate": 6.9991813415943645e-06, "loss": 0.1973, "step": 5411 }, { "epoch": 0.39, "grad_norm": 18.62466579215663, "learning_rate": 6.99811684081231e-06, "loss": 1.1483, "step": 5412 }, { "epoch": 0.39, "grad_norm": 3.517906881144137, "learning_rate": 6.997052232239419e-06, "loss": 0.2294, "step": 5413 }, { "epoch": 0.39, "grad_norm": 11.47374817448957, "learning_rate": 6.9959875159331145e-06, "loss": 0.8768, "step": 5414 }, { "epoch": 0.39, "grad_norm": 2.5687863526087087, "learning_rate": 6.99492269195084e-06, "loss": 0.1305, "step": 5415 }, { "epoch": 0.39, "grad_norm": 3.811211486990551, "learning_rate": 6.993857760350039e-06, "loss": 0.2799, "step": 5416 }, { "epoch": 0.39, "grad_norm": 21.826863089975365, "learning_rate": 6.992792721188158e-06, "loss": 0.9303, "step": 5417 }, { "epoch": 0.39, "grad_norm": 3.5383651603496613, "learning_rate": 6.991727574522653e-06, "loss": 0.185, "step": 5418 }, { "epoch": 0.39, "grad_norm": 4.8356358928256356, "learning_rate": 6.990662320410985e-06, "loss": 0.8323, "step": 5419 }, { "epoch": 0.39, "grad_norm": 9.154181471983051, "learning_rate": 6.989596958910618e-06, "loss": 0.6332, "step": 5420 }, { "epoch": 0.39, "grad_norm": 3.0524313740271367, "learning_rate": 6.988531490079028e-06, "loss": 0.1724, "step": 5421 }, { "epoch": 0.39, "grad_norm": 10.49984010764323, "learning_rate": 6.98746591397369e-06, "loss": 0.8807, "step": 5422 }, { "epoch": 0.39, "grad_norm": 15.974761584880406, "learning_rate": 6.986400230652089e-06, "loss": 0.6552, "step": 5423 }, { "epoch": 0.39, "grad_norm": 3.386739542623246, "learning_rate": 6.985334440171716e-06, "loss": 0.2232, "step": 5424 }, { "epoch": 0.39, "grad_norm": 15.633932667165906, "learning_rate": 6.984268542590063e-06, "loss": 0.5974, "step": 5425 }, { "epoch": 0.39, "grad_norm": 3.279049122851898, "learning_rate": 6.983202537964634e-06, "loss": 0.1787, "step": 5426 }, { "epoch": 0.39, "grad_norm": 4.249590357978619, "learning_rate": 6.982136426352934e-06, "loss": 0.1331, "step": 5427 }, { "epoch": 0.39, "grad_norm": 17.186549506415584, "learning_rate": 6.981070207812477e-06, "loss": 1.0195, "step": 5428 }, { "epoch": 0.39, "grad_norm": 22.64016327267734, "learning_rate": 6.980003882400781e-06, "loss": 1.0644, "step": 5429 }, { "epoch": 0.39, "grad_norm": 12.960470048871379, "learning_rate": 6.978937450175372e-06, "loss": 0.9568, "step": 5430 }, { "epoch": 0.39, "grad_norm": 18.9117584652847, "learning_rate": 6.977870911193776e-06, "loss": 1.1428, "step": 5431 }, { "epoch": 0.39, "grad_norm": 4.532086801712328, "learning_rate": 6.976804265513532e-06, "loss": 0.8561, "step": 5432 }, { "epoch": 0.39, "grad_norm": 4.841350943509332, "learning_rate": 6.97573751319218e-06, "loss": 0.865, "step": 5433 }, { "epoch": 0.39, "grad_norm": 4.586161595872859, "learning_rate": 6.974670654287268e-06, "loss": 1.0464, "step": 5434 }, { "epoch": 0.39, "grad_norm": 3.693075501354293, "learning_rate": 6.973603688856349e-06, "loss": 0.1556, "step": 5435 }, { "epoch": 0.39, "grad_norm": 12.127182590130927, "learning_rate": 6.972536616956981e-06, "loss": 0.851, "step": 5436 }, { "epoch": 0.39, "grad_norm": 4.6332148467171255, "learning_rate": 6.971469438646729e-06, "loss": 0.9638, "step": 5437 }, { "epoch": 0.39, "grad_norm": 20.742291857715763, "learning_rate": 6.970402153983164e-06, "loss": 0.6875, "step": 5438 }, { "epoch": 0.39, "grad_norm": 17.531720099293537, "learning_rate": 6.96933476302386e-06, "loss": 1.0301, "step": 5439 }, { "epoch": 0.39, "grad_norm": 4.736295102892624, "learning_rate": 6.968267265826401e-06, "loss": 0.2754, "step": 5440 }, { "epoch": 0.39, "grad_norm": 3.906170966415475, "learning_rate": 6.967199662448372e-06, "loss": 0.3543, "step": 5441 }, { "epoch": 0.39, "grad_norm": 3.0959476914949517, "learning_rate": 6.966131952947368e-06, "loss": 0.1952, "step": 5442 }, { "epoch": 0.39, "grad_norm": 13.574774954122734, "learning_rate": 6.965064137380987e-06, "loss": 0.4742, "step": 5443 }, { "epoch": 0.39, "grad_norm": 3.8839364984565323, "learning_rate": 6.963996215806836e-06, "loss": 0.2396, "step": 5444 }, { "epoch": 0.39, "grad_norm": 4.400001597920436, "learning_rate": 6.962928188282522e-06, "loss": 0.9763, "step": 5445 }, { "epoch": 0.39, "grad_norm": 3.7218695660345897, "learning_rate": 6.961860054865661e-06, "loss": 0.3366, "step": 5446 }, { "epoch": 0.39, "grad_norm": 3.8929353778263347, "learning_rate": 6.960791815613878e-06, "loss": 0.2652, "step": 5447 }, { "epoch": 0.39, "grad_norm": 16.227333342996314, "learning_rate": 6.959723470584797e-06, "loss": 1.1723, "step": 5448 }, { "epoch": 0.39, "grad_norm": 5.04599024249817, "learning_rate": 6.958655019836052e-06, "loss": 0.975, "step": 5449 }, { "epoch": 0.39, "grad_norm": 4.568891425839777, "learning_rate": 6.9575864634252866e-06, "loss": 0.8265, "step": 5450 }, { "epoch": 0.39, "grad_norm": 12.64283469404187, "learning_rate": 6.9565178014101375e-06, "loss": 0.9709, "step": 5451 }, { "epoch": 0.39, "grad_norm": 4.451402591826437, "learning_rate": 6.95544903384826e-06, "loss": 0.9474, "step": 5452 }, { "epoch": 0.39, "grad_norm": 11.797122465139195, "learning_rate": 6.9543801607973095e-06, "loss": 0.8763, "step": 5453 }, { "epoch": 0.39, "grad_norm": 3.4631071249306706, "learning_rate": 6.953311182314946e-06, "loss": 0.2314, "step": 5454 }, { "epoch": 0.39, "grad_norm": 11.984443701882178, "learning_rate": 6.952242098458839e-06, "loss": 0.7679, "step": 5455 }, { "epoch": 0.39, "grad_norm": 4.5796074135697316, "learning_rate": 6.951172909286661e-06, "loss": 0.8877, "step": 5456 }, { "epoch": 0.39, "grad_norm": 31.410286031548733, "learning_rate": 6.950103614856087e-06, "loss": 0.8867, "step": 5457 }, { "epoch": 0.39, "grad_norm": 14.997892854937337, "learning_rate": 6.949034215224808e-06, "loss": 0.5429, "step": 5458 }, { "epoch": 0.39, "grad_norm": 5.129749548096679, "learning_rate": 6.947964710450509e-06, "loss": 0.9651, "step": 5459 }, { "epoch": 0.39, "grad_norm": 11.822399539845923, "learning_rate": 6.946895100590887e-06, "loss": 0.7477, "step": 5460 }, { "epoch": 0.39, "grad_norm": 7.20157244170983, "learning_rate": 6.9458253857036435e-06, "loss": 0.5556, "step": 5461 }, { "epoch": 0.39, "grad_norm": 2.374949372293154, "learning_rate": 6.944755565846488e-06, "loss": 0.1402, "step": 5462 }, { "epoch": 0.39, "grad_norm": 4.803492673976367, "learning_rate": 6.943685641077129e-06, "loss": 0.8667, "step": 5463 }, { "epoch": 0.39, "grad_norm": 17.204031444418995, "learning_rate": 6.942615611453287e-06, "loss": 0.7423, "step": 5464 }, { "epoch": 0.39, "grad_norm": 14.704681462163082, "learning_rate": 6.941545477032687e-06, "loss": 1.044, "step": 5465 }, { "epoch": 0.39, "grad_norm": 3.3943012796423067, "learning_rate": 6.940475237873057e-06, "loss": 0.1871, "step": 5466 }, { "epoch": 0.39, "grad_norm": 28.284706203217155, "learning_rate": 6.939404894032132e-06, "loss": 1.0612, "step": 5467 }, { "epoch": 0.39, "grad_norm": 3.8290955709399963, "learning_rate": 6.938334445567656e-06, "loss": 0.2608, "step": 5468 }, { "epoch": 0.39, "grad_norm": 4.626806868975624, "learning_rate": 6.9372638925373725e-06, "loss": 0.312, "step": 5469 }, { "epoch": 0.39, "grad_norm": 13.367134290233505, "learning_rate": 6.936193234999036e-06, "loss": 0.8032, "step": 5470 }, { "epoch": 0.39, "grad_norm": 5.527730268043688, "learning_rate": 6.935122473010402e-06, "loss": 0.8832, "step": 5471 }, { "epoch": 0.39, "grad_norm": 3.726570031572804, "learning_rate": 6.9340516066292364e-06, "loss": 0.2447, "step": 5472 }, { "epoch": 0.39, "grad_norm": 2.968555177546801, "learning_rate": 6.932980635913307e-06, "loss": 0.2201, "step": 5473 }, { "epoch": 0.39, "grad_norm": 25.65008837053793, "learning_rate": 6.931909560920389e-06, "loss": 1.2108, "step": 5474 }, { "epoch": 0.39, "grad_norm": 9.479444255190709, "learning_rate": 6.930838381708263e-06, "loss": 0.8094, "step": 5475 }, { "epoch": 0.39, "grad_norm": 3.5154677727843344, "learning_rate": 6.929767098334714e-06, "loss": 0.1787, "step": 5476 }, { "epoch": 0.39, "grad_norm": 4.877038655358977, "learning_rate": 6.928695710857535e-06, "loss": 0.9874, "step": 5477 }, { "epoch": 0.39, "grad_norm": 61.14122781754285, "learning_rate": 6.927624219334523e-06, "loss": 0.7954, "step": 5478 }, { "epoch": 0.39, "grad_norm": 5.482301056160709, "learning_rate": 6.926552623823482e-06, "loss": 0.7788, "step": 5479 }, { "epoch": 0.39, "grad_norm": 5.535884693449631, "learning_rate": 6.925480924382218e-06, "loss": 1.183, "step": 5480 }, { "epoch": 0.39, "grad_norm": 13.22975006382213, "learning_rate": 6.9244091210685475e-06, "loss": 0.9323, "step": 5481 }, { "epoch": 0.39, "grad_norm": 3.5693537298227196, "learning_rate": 6.923337213940289e-06, "loss": 0.2217, "step": 5482 }, { "epoch": 0.39, "grad_norm": 2.933561795569463, "learning_rate": 6.922265203055266e-06, "loss": 0.154, "step": 5483 }, { "epoch": 0.39, "grad_norm": 4.131297787474317, "learning_rate": 6.921193088471314e-06, "loss": 0.3004, "step": 5484 }, { "epoch": 0.39, "grad_norm": 10.281715113769494, "learning_rate": 6.920120870246265e-06, "loss": 0.5592, "step": 5485 }, { "epoch": 0.39, "grad_norm": 3.240012186020556, "learning_rate": 6.919048548437966e-06, "loss": 0.2046, "step": 5486 }, { "epoch": 0.39, "grad_norm": 12.903471817160685, "learning_rate": 6.91797612310426e-06, "loss": 1.0258, "step": 5487 }, { "epoch": 0.39, "grad_norm": 6.006588597559763, "learning_rate": 6.916903594303001e-06, "loss": 0.8024, "step": 5488 }, { "epoch": 0.39, "grad_norm": 3.288669005943346, "learning_rate": 6.91583096209205e-06, "loss": 0.147, "step": 5489 }, { "epoch": 0.39, "grad_norm": 4.925480322403427, "learning_rate": 6.914758226529271e-06, "loss": 1.1132, "step": 5490 }, { "epoch": 0.39, "grad_norm": 3.725450181314638, "learning_rate": 6.913685387672533e-06, "loss": 0.1545, "step": 5491 }, { "epoch": 0.39, "grad_norm": 18.088250238653405, "learning_rate": 6.912612445579712e-06, "loss": 1.149, "step": 5492 }, { "epoch": 0.39, "grad_norm": 4.192027892263247, "learning_rate": 6.911539400308689e-06, "loss": 0.2101, "step": 5493 }, { "epoch": 0.39, "grad_norm": 16.16308056457901, "learning_rate": 6.910466251917352e-06, "loss": 1.3994, "step": 5494 }, { "epoch": 0.39, "grad_norm": 14.044564058154748, "learning_rate": 6.909393000463592e-06, "loss": 0.7276, "step": 5495 }, { "epoch": 0.39, "grad_norm": 3.5695671294010483, "learning_rate": 6.908319646005305e-06, "loss": 0.2235, "step": 5496 }, { "epoch": 0.39, "grad_norm": 4.204086612321792, "learning_rate": 6.907246188600397e-06, "loss": 0.1903, "step": 5497 }, { "epoch": 0.39, "grad_norm": 4.992914768495658, "learning_rate": 6.906172628306779e-06, "loss": 0.3062, "step": 5498 }, { "epoch": 0.39, "grad_norm": 16.512931812362417, "learning_rate": 6.90509896518236e-06, "loss": 1.0732, "step": 5499 }, { "epoch": 0.39, "grad_norm": 3.3787512514699314, "learning_rate": 6.904025199285064e-06, "loss": 0.1833, "step": 5500 }, { "epoch": 0.39, "grad_norm": 3.8053774444626898, "learning_rate": 6.902951330672816e-06, "loss": 0.2086, "step": 5501 }, { "epoch": 0.39, "grad_norm": 4.522976773582381, "learning_rate": 6.901877359403546e-06, "loss": 0.8957, "step": 5502 }, { "epoch": 0.39, "grad_norm": 3.476585179109775, "learning_rate": 6.900803285535191e-06, "loss": 0.1917, "step": 5503 }, { "epoch": 0.39, "grad_norm": 4.1107430260543625, "learning_rate": 6.899729109125697e-06, "loss": 0.1654, "step": 5504 }, { "epoch": 0.39, "grad_norm": 3.2828291485087253, "learning_rate": 6.898654830233006e-06, "loss": 0.1575, "step": 5505 }, { "epoch": 0.39, "grad_norm": 3.3168108568731207, "learning_rate": 6.8975804489150735e-06, "loss": 0.2208, "step": 5506 }, { "epoch": 0.39, "grad_norm": 9.664547227273891, "learning_rate": 6.89650596522986e-06, "loss": 0.4127, "step": 5507 }, { "epoch": 0.39, "grad_norm": 3.170386371086322, "learning_rate": 6.895431379235327e-06, "loss": 0.1848, "step": 5508 }, { "epoch": 0.4, "grad_norm": 3.128088643151737, "learning_rate": 6.894356690989445e-06, "loss": 0.2224, "step": 5509 }, { "epoch": 0.4, "grad_norm": 4.07443872319752, "learning_rate": 6.893281900550192e-06, "loss": 0.2422, "step": 5510 }, { "epoch": 0.4, "grad_norm": 27.757704949147303, "learning_rate": 6.892207007975545e-06, "loss": 0.8295, "step": 5511 }, { "epoch": 0.4, "grad_norm": 3.977556648162725, "learning_rate": 6.891132013323494e-06, "loss": 0.2455, "step": 5512 }, { "epoch": 0.4, "grad_norm": 22.1366258068146, "learning_rate": 6.8900569166520285e-06, "loss": 1.6323, "step": 5513 }, { "epoch": 0.4, "grad_norm": 4.13304244268509, "learning_rate": 6.888981718019145e-06, "loss": 0.3173, "step": 5514 }, { "epoch": 0.4, "grad_norm": 10.676890556864956, "learning_rate": 6.88790641748285e-06, "loss": 0.8188, "step": 5515 }, { "epoch": 0.4, "grad_norm": 3.4006908497469603, "learning_rate": 6.886831015101151e-06, "loss": 0.2102, "step": 5516 }, { "epoch": 0.4, "grad_norm": 4.379874413599749, "learning_rate": 6.885755510932058e-06, "loss": 0.3631, "step": 5517 }, { "epoch": 0.4, "grad_norm": 3.0526024916387926, "learning_rate": 6.884679905033596e-06, "loss": 0.1513, "step": 5518 }, { "epoch": 0.4, "grad_norm": 3.238216359181118, "learning_rate": 6.883604197463784e-06, "loss": 0.204, "step": 5519 }, { "epoch": 0.4, "grad_norm": 3.8699259159247057, "learning_rate": 6.882528388280656e-06, "loss": 0.2972, "step": 5520 }, { "epoch": 0.4, "grad_norm": 3.2118602719674767, "learning_rate": 6.881452477542249e-06, "loss": 0.2152, "step": 5521 }, { "epoch": 0.4, "grad_norm": 3.0615270849270004, "learning_rate": 6.880376465306602e-06, "loss": 0.2484, "step": 5522 }, { "epoch": 0.4, "grad_norm": 3.0790617716339908, "learning_rate": 6.879300351631762e-06, "loss": 0.2247, "step": 5523 }, { "epoch": 0.4, "grad_norm": 3.577087903844319, "learning_rate": 6.878224136575781e-06, "loss": 0.2096, "step": 5524 }, { "epoch": 0.4, "grad_norm": 3.3603139528681187, "learning_rate": 6.877147820196717e-06, "loss": 0.1815, "step": 5525 }, { "epoch": 0.4, "grad_norm": 11.915818816146139, "learning_rate": 6.876071402552634e-06, "loss": 0.7121, "step": 5526 }, { "epoch": 0.4, "grad_norm": 4.427197608463616, "learning_rate": 6.8749948837016e-06, "loss": 0.8686, "step": 5527 }, { "epoch": 0.4, "grad_norm": 3.5975073153170305, "learning_rate": 6.873918263701691e-06, "loss": 0.1518, "step": 5528 }, { "epoch": 0.4, "grad_norm": 14.705303335873111, "learning_rate": 6.872841542610982e-06, "loss": 0.7902, "step": 5529 }, { "epoch": 0.4, "grad_norm": 3.610752738198982, "learning_rate": 6.871764720487564e-06, "loss": 0.1814, "step": 5530 }, { "epoch": 0.4, "grad_norm": 15.05451678699793, "learning_rate": 6.8706877973895215e-06, "loss": 1.1157, "step": 5531 }, { "epoch": 0.4, "grad_norm": 3.9552470776018227, "learning_rate": 6.8696107733749555e-06, "loss": 0.1383, "step": 5532 }, { "epoch": 0.4, "grad_norm": 8.665394576117697, "learning_rate": 6.868533648501962e-06, "loss": 0.7233, "step": 5533 }, { "epoch": 0.4, "grad_norm": 3.220101869112397, "learning_rate": 6.867456422828654e-06, "loss": 0.2528, "step": 5534 }, { "epoch": 0.4, "grad_norm": 3.3454773168321332, "learning_rate": 6.866379096413141e-06, "loss": 0.2027, "step": 5535 }, { "epoch": 0.4, "grad_norm": 28.220386366002412, "learning_rate": 6.865301669313538e-06, "loss": 0.7143, "step": 5536 }, { "epoch": 0.4, "grad_norm": 11.373913638879298, "learning_rate": 6.864224141587971e-06, "loss": 0.8618, "step": 5537 }, { "epoch": 0.4, "grad_norm": 4.1663607855703555, "learning_rate": 6.8631465132945685e-06, "loss": 0.7793, "step": 5538 }, { "epoch": 0.4, "grad_norm": 5.104548697635573, "learning_rate": 6.862068784491464e-06, "loss": 0.966, "step": 5539 }, { "epoch": 0.4, "grad_norm": 3.3589501406884947, "learning_rate": 6.860990955236797e-06, "loss": 0.2094, "step": 5540 }, { "epoch": 0.4, "grad_norm": 10.534523458744944, "learning_rate": 6.859913025588713e-06, "loss": 0.7531, "step": 5541 }, { "epoch": 0.4, "grad_norm": 3.4485541637881787, "learning_rate": 6.85883499560536e-06, "loss": 0.2339, "step": 5542 }, { "epoch": 0.4, "grad_norm": 3.3129264631460167, "learning_rate": 6.857756865344895e-06, "loss": 0.1816, "step": 5543 }, { "epoch": 0.4, "grad_norm": 8.652617363755544, "learning_rate": 6.85667863486548e-06, "loss": 1.1338, "step": 5544 }, { "epoch": 0.4, "grad_norm": 24.701108553689192, "learning_rate": 6.855600304225281e-06, "loss": 0.8164, "step": 5545 }, { "epoch": 0.4, "grad_norm": 4.065471233179358, "learning_rate": 6.8545218734824685e-06, "loss": 0.1758, "step": 5546 }, { "epoch": 0.4, "grad_norm": 3.3079108645822024, "learning_rate": 6.853443342695221e-06, "loss": 0.1407, "step": 5547 }, { "epoch": 0.4, "grad_norm": 5.126309228362135, "learning_rate": 6.85236471192172e-06, "loss": 0.9402, "step": 5548 }, { "epoch": 0.4, "grad_norm": 4.943516247736897, "learning_rate": 6.851285981220156e-06, "loss": 0.8645, "step": 5549 }, { "epoch": 0.4, "grad_norm": 3.4610434245505846, "learning_rate": 6.85020715064872e-06, "loss": 0.1678, "step": 5550 }, { "epoch": 0.4, "grad_norm": 3.5994599161940584, "learning_rate": 6.8491282202656106e-06, "loss": 0.1873, "step": 5551 }, { "epoch": 0.4, "grad_norm": 3.128650578608486, "learning_rate": 6.8480491901290356e-06, "loss": 0.1388, "step": 5552 }, { "epoch": 0.4, "grad_norm": 10.273163816578656, "learning_rate": 6.8469700602972e-06, "loss": 0.3223, "step": 5553 }, { "epoch": 0.4, "grad_norm": 14.931203074759932, "learning_rate": 6.84589083082832e-06, "loss": 0.701, "step": 5554 }, { "epoch": 0.4, "grad_norm": 4.284398816459083, "learning_rate": 6.844811501780619e-06, "loss": 0.3223, "step": 5555 }, { "epoch": 0.4, "grad_norm": 18.42786357199177, "learning_rate": 6.843732073212319e-06, "loss": 0.8199, "step": 5556 }, { "epoch": 0.4, "grad_norm": 5.221360922936301, "learning_rate": 6.842652545181652e-06, "loss": 0.8713, "step": 5557 }, { "epoch": 0.4, "grad_norm": 5.439637737042991, "learning_rate": 6.841572917746857e-06, "loss": 0.8721, "step": 5558 }, { "epoch": 0.4, "grad_norm": 13.996212954546033, "learning_rate": 6.840493190966173e-06, "loss": 0.5866, "step": 5559 }, { "epoch": 0.4, "grad_norm": 22.3850159548028, "learning_rate": 6.839413364897846e-06, "loss": 0.7205, "step": 5560 }, { "epoch": 0.4, "grad_norm": 3.4875931670864952, "learning_rate": 6.838333439600135e-06, "loss": 0.2102, "step": 5561 }, { "epoch": 0.4, "grad_norm": 12.378212729784943, "learning_rate": 6.83725341513129e-06, "loss": 1.009, "step": 5562 }, { "epoch": 0.4, "grad_norm": 18.058693736405974, "learning_rate": 6.836173291549578e-06, "loss": 0.9938, "step": 5563 }, { "epoch": 0.4, "grad_norm": 3.6495285331826577, "learning_rate": 6.835093068913267e-06, "loss": 0.2552, "step": 5564 }, { "epoch": 0.4, "grad_norm": 31.107650806520155, "learning_rate": 6.834012747280633e-06, "loss": 0.8461, "step": 5565 }, { "epoch": 0.4, "grad_norm": 9.186005553667732, "learning_rate": 6.8329323267099525e-06, "loss": 0.9112, "step": 5566 }, { "epoch": 0.4, "grad_norm": 3.9221481693971607, "learning_rate": 6.83185180725951e-06, "loss": 0.213, "step": 5567 }, { "epoch": 0.4, "grad_norm": 4.401694395450417, "learning_rate": 6.830771188987596e-06, "loss": 0.2045, "step": 5568 }, { "epoch": 0.4, "grad_norm": 4.263293956753445, "learning_rate": 6.829690471952508e-06, "loss": 0.2444, "step": 5569 }, { "epoch": 0.4, "grad_norm": 4.793911910419997, "learning_rate": 6.828609656212545e-06, "loss": 0.814, "step": 5570 }, { "epoch": 0.4, "grad_norm": 4.985194101420741, "learning_rate": 6.827528741826011e-06, "loss": 1.0394, "step": 5571 }, { "epoch": 0.4, "grad_norm": 13.97089898052416, "learning_rate": 6.82644772885122e-06, "loss": 0.7383, "step": 5572 }, { "epoch": 0.4, "grad_norm": 10.536131718075273, "learning_rate": 6.8253666173464876e-06, "loss": 0.7012, "step": 5573 }, { "epoch": 0.4, "grad_norm": 12.453189012972006, "learning_rate": 6.824285407370134e-06, "loss": 0.7227, "step": 5574 }, { "epoch": 0.4, "grad_norm": 3.4752696943291315, "learning_rate": 6.823204098980489e-06, "loss": 0.2105, "step": 5575 }, { "epoch": 0.4, "grad_norm": 3.3044341360494496, "learning_rate": 6.822122692235885e-06, "loss": 0.2361, "step": 5576 }, { "epoch": 0.4, "grad_norm": 22.913081031674828, "learning_rate": 6.821041187194658e-06, "loss": 0.822, "step": 5577 }, { "epoch": 0.4, "grad_norm": 3.818226894040759, "learning_rate": 6.8199595839151526e-06, "loss": 0.2132, "step": 5578 }, { "epoch": 0.4, "grad_norm": 11.769286824762547, "learning_rate": 6.818877882455716e-06, "loss": 1.2828, "step": 5579 }, { "epoch": 0.4, "grad_norm": 4.049688539620251, "learning_rate": 6.817796082874702e-06, "loss": 0.2173, "step": 5580 }, { "epoch": 0.4, "grad_norm": 15.654235871466577, "learning_rate": 6.8167141852304705e-06, "loss": 0.8468, "step": 5581 }, { "epoch": 0.4, "grad_norm": 3.9359755705754207, "learning_rate": 6.815632189581386e-06, "loss": 0.2196, "step": 5582 }, { "epoch": 0.4, "grad_norm": 3.038210149538424, "learning_rate": 6.814550095985817e-06, "loss": 0.1264, "step": 5583 }, { "epoch": 0.4, "grad_norm": 23.422974038089983, "learning_rate": 6.813467904502138e-06, "loss": 0.9333, "step": 5584 }, { "epoch": 0.4, "grad_norm": 4.847429159151815, "learning_rate": 6.81238561518873e-06, "loss": 1.0465, "step": 5585 }, { "epoch": 0.4, "grad_norm": 6.3347841822126485, "learning_rate": 6.811303228103979e-06, "loss": 0.8491, "step": 5586 }, { "epoch": 0.4, "grad_norm": 3.785422256149944, "learning_rate": 6.810220743306275e-06, "loss": 0.1819, "step": 5587 }, { "epoch": 0.4, "grad_norm": 5.040805081620316, "learning_rate": 6.809138160854014e-06, "loss": 0.8774, "step": 5588 }, { "epoch": 0.4, "grad_norm": 3.3156545507883073, "learning_rate": 6.808055480805597e-06, "loss": 0.1721, "step": 5589 }, { "epoch": 0.4, "grad_norm": 57.85265850811353, "learning_rate": 6.80697270321943e-06, "loss": 1.1587, "step": 5590 }, { "epoch": 0.4, "grad_norm": 15.338574049263993, "learning_rate": 6.8058898281539265e-06, "loss": 1.4438, "step": 5591 }, { "epoch": 0.4, "grad_norm": 4.028673328324577, "learning_rate": 6.804806855667501e-06, "loss": 0.2449, "step": 5592 }, { "epoch": 0.4, "grad_norm": 4.8860918092204155, "learning_rate": 6.803723785818579e-06, "loss": 0.8526, "step": 5593 }, { "epoch": 0.4, "grad_norm": 5.566937545150912, "learning_rate": 6.802640618665585e-06, "loss": 0.8493, "step": 5594 }, { "epoch": 0.4, "grad_norm": 5.277079851188706, "learning_rate": 6.801557354266955e-06, "loss": 0.8255, "step": 5595 }, { "epoch": 0.4, "grad_norm": 3.8870195947977937, "learning_rate": 6.800473992681122e-06, "loss": 0.2313, "step": 5596 }, { "epoch": 0.4, "grad_norm": 3.8572321157590235, "learning_rate": 6.799390533966533e-06, "loss": 0.2182, "step": 5597 }, { "epoch": 0.4, "grad_norm": 5.213533985287811, "learning_rate": 6.798306978181637e-06, "loss": 1.068, "step": 5598 }, { "epoch": 0.4, "grad_norm": 3.859626569601387, "learning_rate": 6.7972233253848855e-06, "loss": 0.232, "step": 5599 }, { "epoch": 0.4, "grad_norm": 18.345605836521266, "learning_rate": 6.7961395756347395e-06, "loss": 0.9206, "step": 5600 }, { "epoch": 0.4, "grad_norm": 3.461124796233085, "learning_rate": 6.795055728989662e-06, "loss": 0.1758, "step": 5601 }, { "epoch": 0.4, "grad_norm": 4.334999023858863, "learning_rate": 6.7939717855081225e-06, "loss": 0.9451, "step": 5602 }, { "epoch": 0.4, "grad_norm": 3.5108416902770023, "learning_rate": 6.792887745248595e-06, "loss": 0.2547, "step": 5603 }, { "epoch": 0.4, "grad_norm": 3.632078206889971, "learning_rate": 6.79180360826956e-06, "loss": 0.2317, "step": 5604 }, { "epoch": 0.4, "grad_norm": 3.178234167988559, "learning_rate": 6.7907193746295016e-06, "loss": 0.1643, "step": 5605 }, { "epoch": 0.4, "grad_norm": 9.893101346062325, "learning_rate": 6.789635044386915e-06, "loss": 0.6556, "step": 5606 }, { "epoch": 0.4, "grad_norm": 4.546250737996706, "learning_rate": 6.788550617600289e-06, "loss": 0.3197, "step": 5607 }, { "epoch": 0.4, "grad_norm": 3.103306574782077, "learning_rate": 6.787466094328128e-06, "loss": 0.2577, "step": 5608 }, { "epoch": 0.4, "grad_norm": 3.0787065558913707, "learning_rate": 6.786381474628937e-06, "loss": 0.2178, "step": 5609 }, { "epoch": 0.4, "grad_norm": 3.8450969266636528, "learning_rate": 6.785296758561226e-06, "loss": 0.2571, "step": 5610 }, { "epoch": 0.4, "grad_norm": 3.6507133660669684, "learning_rate": 6.784211946183513e-06, "loss": 0.2618, "step": 5611 }, { "epoch": 0.4, "grad_norm": 20.35765518605765, "learning_rate": 6.783127037554322e-06, "loss": 1.4904, "step": 5612 }, { "epoch": 0.4, "grad_norm": 10.961507705065987, "learning_rate": 6.782042032732175e-06, "loss": 1.0311, "step": 5613 }, { "epoch": 0.4, "grad_norm": 4.689136068327593, "learning_rate": 6.780956931775604e-06, "loss": 1.1375, "step": 5614 }, { "epoch": 0.4, "grad_norm": 12.302401039684522, "learning_rate": 6.77987173474315e-06, "loss": 0.9248, "step": 5615 }, { "epoch": 0.4, "grad_norm": 3.4614236532061944, "learning_rate": 6.778786441693351e-06, "loss": 0.1829, "step": 5616 }, { "epoch": 0.4, "grad_norm": 20.602743250275097, "learning_rate": 6.777701052684759e-06, "loss": 1.2484, "step": 5617 }, { "epoch": 0.4, "grad_norm": 9.872292677269767, "learning_rate": 6.7766155677759236e-06, "loss": 1.1089, "step": 5618 }, { "epoch": 0.4, "grad_norm": 3.352072809779184, "learning_rate": 6.775529987025402e-06, "loss": 0.1714, "step": 5619 }, { "epoch": 0.4, "grad_norm": 4.117759449202629, "learning_rate": 6.774444310491759e-06, "loss": 0.2394, "step": 5620 }, { "epoch": 0.4, "grad_norm": 4.0516731671272215, "learning_rate": 6.773358538233562e-06, "loss": 0.2632, "step": 5621 }, { "epoch": 0.4, "grad_norm": 5.929938121751038, "learning_rate": 6.7722726703093836e-06, "loss": 0.9202, "step": 5622 }, { "epoch": 0.4, "grad_norm": 3.254390237250316, "learning_rate": 6.7711867067778044e-06, "loss": 0.2199, "step": 5623 }, { "epoch": 0.4, "grad_norm": 3.3344700123973863, "learning_rate": 6.770100647697406e-06, "loss": 0.2055, "step": 5624 }, { "epoch": 0.4, "grad_norm": 2.9417721135337453, "learning_rate": 6.769014493126778e-06, "loss": 0.1413, "step": 5625 }, { "epoch": 0.4, "grad_norm": 3.8820073859625674, "learning_rate": 6.767928243124516e-06, "loss": 0.2951, "step": 5626 }, { "epoch": 0.4, "grad_norm": 18.917271340067323, "learning_rate": 6.766841897749213e-06, "loss": 1.0063, "step": 5627 }, { "epoch": 0.4, "grad_norm": 3.441268670471427, "learning_rate": 6.76575545705948e-06, "loss": 0.1806, "step": 5628 }, { "epoch": 0.4, "grad_norm": 23.124920489131444, "learning_rate": 6.764668921113923e-06, "loss": 1.1907, "step": 5629 }, { "epoch": 0.4, "grad_norm": 12.877620513238746, "learning_rate": 6.76358228997116e-06, "loss": 0.7727, "step": 5630 }, { "epoch": 0.4, "grad_norm": 3.031106432717313, "learning_rate": 6.762495563689806e-06, "loss": 0.2277, "step": 5631 }, { "epoch": 0.4, "grad_norm": 15.680242125261922, "learning_rate": 6.761408742328488e-06, "loss": 0.8996, "step": 5632 }, { "epoch": 0.4, "grad_norm": 3.6251179167418326, "learning_rate": 6.7603218259458344e-06, "loss": 0.2787, "step": 5633 }, { "epoch": 0.4, "grad_norm": 14.92150204767658, "learning_rate": 6.7592348146004836e-06, "loss": 0.9496, "step": 5634 }, { "epoch": 0.4, "grad_norm": 12.12975455540496, "learning_rate": 6.758147708351072e-06, "loss": 0.7159, "step": 5635 }, { "epoch": 0.4, "grad_norm": 4.144518712347315, "learning_rate": 6.757060507256248e-06, "loss": 0.2911, "step": 5636 }, { "epoch": 0.4, "grad_norm": 10.445938789405718, "learning_rate": 6.755973211374658e-06, "loss": 0.5801, "step": 5637 }, { "epoch": 0.4, "grad_norm": 4.011155301083908, "learning_rate": 6.754885820764962e-06, "loss": 0.2643, "step": 5638 }, { "epoch": 0.4, "grad_norm": 3.402622872850241, "learning_rate": 6.753798335485818e-06, "loss": 0.2998, "step": 5639 }, { "epoch": 0.4, "grad_norm": 11.131309291257953, "learning_rate": 6.752710755595893e-06, "loss": 0.6651, "step": 5640 }, { "epoch": 0.4, "grad_norm": 8.278604890107513, "learning_rate": 6.751623081153855e-06, "loss": 0.7721, "step": 5641 }, { "epoch": 0.4, "grad_norm": 21.69884885260278, "learning_rate": 6.750535312218385e-06, "loss": 1.3225, "step": 5642 }, { "epoch": 0.4, "grad_norm": 5.488921882434741, "learning_rate": 6.74944744884816e-06, "loss": 1.0045, "step": 5643 }, { "epoch": 0.4, "grad_norm": 15.609293242050116, "learning_rate": 6.748359491101866e-06, "loss": 1.2431, "step": 5644 }, { "epoch": 0.4, "grad_norm": 3.6357020555025747, "learning_rate": 6.747271439038195e-06, "loss": 0.1563, "step": 5645 }, { "epoch": 0.4, "grad_norm": 5.2653567006783915, "learning_rate": 6.746183292715845e-06, "loss": 0.8624, "step": 5646 }, { "epoch": 0.4, "grad_norm": 3.8236577644869025, "learning_rate": 6.745095052193515e-06, "loss": 0.2433, "step": 5647 }, { "epoch": 0.41, "grad_norm": 5.045224906005836, "learning_rate": 6.744006717529914e-06, "loss": 1.0503, "step": 5648 }, { "epoch": 0.41, "grad_norm": 15.239148909313725, "learning_rate": 6.742918288783752e-06, "loss": 1.3234, "step": 5649 }, { "epoch": 0.41, "grad_norm": 3.3077299669323197, "learning_rate": 6.741829766013744e-06, "loss": 0.2296, "step": 5650 }, { "epoch": 0.41, "grad_norm": 4.969779446759263, "learning_rate": 6.7407411492786135e-06, "loss": 1.0198, "step": 5651 }, { "epoch": 0.41, "grad_norm": 10.880215195463922, "learning_rate": 6.739652438637089e-06, "loss": 0.5722, "step": 5652 }, { "epoch": 0.41, "grad_norm": 3.285617358070621, "learning_rate": 6.738563634147898e-06, "loss": 0.246, "step": 5653 }, { "epoch": 0.41, "grad_norm": 25.457798572590825, "learning_rate": 6.7374747358697815e-06, "loss": 1.527, "step": 5654 }, { "epoch": 0.41, "grad_norm": 21.31283894710744, "learning_rate": 6.73638574386148e-06, "loss": 0.8962, "step": 5655 }, { "epoch": 0.41, "grad_norm": 17.261521311443953, "learning_rate": 6.735296658181738e-06, "loss": 0.8986, "step": 5656 }, { "epoch": 0.41, "grad_norm": 15.53404624953697, "learning_rate": 6.734207478889312e-06, "loss": 0.6217, "step": 5657 }, { "epoch": 0.41, "grad_norm": 3.5472108740261517, "learning_rate": 6.733118206042955e-06, "loss": 0.2783, "step": 5658 }, { "epoch": 0.41, "grad_norm": 4.4160085449850826, "learning_rate": 6.732028839701431e-06, "loss": 0.9432, "step": 5659 }, { "epoch": 0.41, "grad_norm": 7.431587753061681, "learning_rate": 6.730939379923509e-06, "loss": 0.4915, "step": 5660 }, { "epoch": 0.41, "grad_norm": 4.172207651735552, "learning_rate": 6.729849826767958e-06, "loss": 0.2616, "step": 5661 }, { "epoch": 0.41, "grad_norm": 8.16478527571973, "learning_rate": 6.728760180293557e-06, "loss": 0.4889, "step": 5662 }, { "epoch": 0.41, "grad_norm": 3.5408273751149175, "learning_rate": 6.727670440559089e-06, "loss": 0.2114, "step": 5663 }, { "epoch": 0.41, "grad_norm": 11.526209699243363, "learning_rate": 6.726580607623339e-06, "loss": 0.7894, "step": 5664 }, { "epoch": 0.41, "grad_norm": 18.941699833230302, "learning_rate": 6.725490681545099e-06, "loss": 0.7864, "step": 5665 }, { "epoch": 0.41, "grad_norm": 3.0682530110020787, "learning_rate": 6.724400662383172e-06, "loss": 0.2146, "step": 5666 }, { "epoch": 0.41, "grad_norm": 14.252935867270725, "learning_rate": 6.723310550196355e-06, "loss": 1.0154, "step": 5667 }, { "epoch": 0.41, "grad_norm": 17.981046921395453, "learning_rate": 6.7222203450434555e-06, "loss": 1.1077, "step": 5668 }, { "epoch": 0.41, "grad_norm": 18.43822294498831, "learning_rate": 6.721130046983288e-06, "loss": 0.6236, "step": 5669 }, { "epoch": 0.41, "grad_norm": 17.441101203745507, "learning_rate": 6.720039656074669e-06, "loss": 0.8706, "step": 5670 }, { "epoch": 0.41, "grad_norm": 3.439598144386139, "learning_rate": 6.718949172376423e-06, "loss": 0.2109, "step": 5671 }, { "epoch": 0.41, "grad_norm": 4.899353870007479, "learning_rate": 6.7178585959473745e-06, "loss": 0.2443, "step": 5672 }, { "epoch": 0.41, "grad_norm": 5.087234419807302, "learning_rate": 6.716767926846357e-06, "loss": 1.0026, "step": 5673 }, { "epoch": 0.41, "grad_norm": 10.625860791058507, "learning_rate": 6.715677165132208e-06, "loss": 0.6736, "step": 5674 }, { "epoch": 0.41, "grad_norm": 4.397460927711027, "learning_rate": 6.714586310863771e-06, "loss": 1.0192, "step": 5675 }, { "epoch": 0.41, "grad_norm": 4.274616249144246, "learning_rate": 6.713495364099892e-06, "loss": 0.8961, "step": 5676 }, { "epoch": 0.41, "grad_norm": 5.430431452671338, "learning_rate": 6.712404324899426e-06, "loss": 0.9124, "step": 5677 }, { "epoch": 0.41, "grad_norm": 17.44307792155619, "learning_rate": 6.711313193321225e-06, "loss": 0.8817, "step": 5678 }, { "epoch": 0.41, "grad_norm": 12.844590811691246, "learning_rate": 6.710221969424158e-06, "loss": 0.5291, "step": 5679 }, { "epoch": 0.41, "grad_norm": 2.9734930705130638, "learning_rate": 6.709130653267087e-06, "loss": 0.1525, "step": 5680 }, { "epoch": 0.41, "grad_norm": 3.103554337597358, "learning_rate": 6.708039244908888e-06, "loss": 0.2079, "step": 5681 }, { "epoch": 0.41, "grad_norm": 3.724431554776486, "learning_rate": 6.706947744408436e-06, "loss": 0.3248, "step": 5682 }, { "epoch": 0.41, "grad_norm": 100.84016586612503, "learning_rate": 6.705856151824615e-06, "loss": 1.2272, "step": 5683 }, { "epoch": 0.41, "grad_norm": 5.147219275307487, "learning_rate": 6.704764467216312e-06, "loss": 0.9377, "step": 5684 }, { "epoch": 0.41, "grad_norm": 5.476855038489945, "learning_rate": 6.7036726906424174e-06, "loss": 0.7797, "step": 5685 }, { "epoch": 0.41, "grad_norm": 20.162762772859082, "learning_rate": 6.7025808221618325e-06, "loss": 0.6857, "step": 5686 }, { "epoch": 0.41, "grad_norm": 11.792337942756348, "learning_rate": 6.701488861833454e-06, "loss": 0.8802, "step": 5687 }, { "epoch": 0.41, "grad_norm": 3.0692696636934182, "learning_rate": 6.7003968097161935e-06, "loss": 0.2083, "step": 5688 }, { "epoch": 0.41, "grad_norm": 2.5905262152868502, "learning_rate": 6.699304665868961e-06, "loss": 0.1487, "step": 5689 }, { "epoch": 0.41, "grad_norm": 4.728505271481918, "learning_rate": 6.698212430350673e-06, "loss": 0.9477, "step": 5690 }, { "epoch": 0.41, "grad_norm": 19.706104150377566, "learning_rate": 6.697120103220255e-06, "loss": 0.9123, "step": 5691 }, { "epoch": 0.41, "grad_norm": 4.586765689515778, "learning_rate": 6.69602768453663e-06, "loss": 0.2959, "step": 5692 }, { "epoch": 0.41, "grad_norm": 16.8707593704825, "learning_rate": 6.694935174358731e-06, "loss": 0.5414, "step": 5693 }, { "epoch": 0.41, "grad_norm": 4.623805911003456, "learning_rate": 6.693842572745496e-06, "loss": 0.3085, "step": 5694 }, { "epoch": 0.41, "grad_norm": 4.326637277877918, "learning_rate": 6.692749879755864e-06, "loss": 0.284, "step": 5695 }, { "epoch": 0.41, "grad_norm": 3.17189610389636, "learning_rate": 6.691657095448784e-06, "loss": 0.2354, "step": 5696 }, { "epoch": 0.41, "grad_norm": 3.8504446938186123, "learning_rate": 6.690564219883209e-06, "loss": 0.2718, "step": 5697 }, { "epoch": 0.41, "grad_norm": 14.147559198796406, "learning_rate": 6.6894712531180915e-06, "loss": 0.8627, "step": 5698 }, { "epoch": 0.41, "grad_norm": 3.1458247767143486, "learning_rate": 6.688378195212395e-06, "loss": 0.1798, "step": 5699 }, { "epoch": 0.41, "grad_norm": 3.6106331865357726, "learning_rate": 6.687285046225086e-06, "loss": 0.2776, "step": 5700 }, { "epoch": 0.41, "grad_norm": 3.6829910763038924, "learning_rate": 6.686191806215134e-06, "loss": 0.1858, "step": 5701 }, { "epoch": 0.41, "grad_norm": 4.989531491948055, "learning_rate": 6.685098475241518e-06, "loss": 0.208, "step": 5702 }, { "epoch": 0.41, "grad_norm": 3.8435683261028926, "learning_rate": 6.68400505336322e-06, "loss": 0.2246, "step": 5703 }, { "epoch": 0.41, "grad_norm": 3.54995785866879, "learning_rate": 6.682911540639221e-06, "loss": 0.1689, "step": 5704 }, { "epoch": 0.41, "grad_norm": 3.4127401659025747, "learning_rate": 6.681817937128514e-06, "loss": 0.2048, "step": 5705 }, { "epoch": 0.41, "grad_norm": 3.2958758437672286, "learning_rate": 6.680724242890097e-06, "loss": 0.2027, "step": 5706 }, { "epoch": 0.41, "grad_norm": 3.9008123079629438, "learning_rate": 6.6796304579829676e-06, "loss": 0.192, "step": 5707 }, { "epoch": 0.41, "grad_norm": 9.743826602971298, "learning_rate": 6.678536582466133e-06, "loss": 0.7234, "step": 5708 }, { "epoch": 0.41, "grad_norm": 14.421186691194391, "learning_rate": 6.677442616398605e-06, "loss": 0.3438, "step": 5709 }, { "epoch": 0.41, "grad_norm": 4.777558803983142, "learning_rate": 6.676348559839395e-06, "loss": 0.8923, "step": 5710 }, { "epoch": 0.41, "grad_norm": 20.488197313198494, "learning_rate": 6.675254412847527e-06, "loss": 0.9743, "step": 5711 }, { "epoch": 0.41, "grad_norm": 4.751073308921171, "learning_rate": 6.674160175482024e-06, "loss": 0.8331, "step": 5712 }, { "epoch": 0.41, "grad_norm": 4.0468160703707, "learning_rate": 6.673065847801916e-06, "loss": 0.2122, "step": 5713 }, { "epoch": 0.41, "grad_norm": 18.74137680149557, "learning_rate": 6.67197142986624e-06, "loss": 0.9412, "step": 5714 }, { "epoch": 0.41, "grad_norm": 5.040497744256587, "learning_rate": 6.670876921734034e-06, "loss": 0.8852, "step": 5715 }, { "epoch": 0.41, "grad_norm": 3.391992694748921, "learning_rate": 6.669782323464341e-06, "loss": 0.1984, "step": 5716 }, { "epoch": 0.41, "grad_norm": 31.09269515888735, "learning_rate": 6.668687635116214e-06, "loss": 0.8278, "step": 5717 }, { "epoch": 0.41, "grad_norm": 9.653219176896993, "learning_rate": 6.667592856748703e-06, "loss": 0.4121, "step": 5718 }, { "epoch": 0.41, "grad_norm": 4.829098251251197, "learning_rate": 6.6664979884208716e-06, "loss": 1.1309, "step": 5719 }, { "epoch": 0.41, "grad_norm": 17.536550828966966, "learning_rate": 6.665403030191782e-06, "loss": 1.3616, "step": 5720 }, { "epoch": 0.41, "grad_norm": 5.189423734738215, "learning_rate": 6.6643079821205025e-06, "loss": 0.2025, "step": 5721 }, { "epoch": 0.41, "grad_norm": 3.810508704093921, "learning_rate": 6.663212844266106e-06, "loss": 0.1887, "step": 5722 }, { "epoch": 0.41, "grad_norm": 3.787461824296271, "learning_rate": 6.6621176166876735e-06, "loss": 0.1985, "step": 5723 }, { "epoch": 0.41, "grad_norm": 12.325529328563489, "learning_rate": 6.661022299444286e-06, "loss": 0.6306, "step": 5724 }, { "epoch": 0.41, "grad_norm": 3.5035430313056466, "learning_rate": 6.659926892595033e-06, "loss": 0.2233, "step": 5725 }, { "epoch": 0.41, "grad_norm": 23.787579567325572, "learning_rate": 6.658831396199007e-06, "loss": 1.1823, "step": 5726 }, { "epoch": 0.41, "grad_norm": 3.166727572815394, "learning_rate": 6.657735810315307e-06, "loss": 0.1561, "step": 5727 }, { "epoch": 0.41, "grad_norm": 3.044707140317232, "learning_rate": 6.6566401350030344e-06, "loss": 0.1798, "step": 5728 }, { "epoch": 0.41, "grad_norm": 4.119986436328611, "learning_rate": 6.655544370321297e-06, "loss": 0.3292, "step": 5729 }, { "epoch": 0.41, "grad_norm": 4.558810571355245, "learning_rate": 6.654448516329208e-06, "loss": 0.8771, "step": 5730 }, { "epoch": 0.41, "grad_norm": 12.004429568677583, "learning_rate": 6.6533525730858835e-06, "loss": 0.7655, "step": 5731 }, { "epoch": 0.41, "grad_norm": 3.427613146891623, "learning_rate": 6.652256540650446e-06, "loss": 0.238, "step": 5732 }, { "epoch": 0.41, "grad_norm": 4.847463015330878, "learning_rate": 6.651160419082024e-06, "loss": 0.1291, "step": 5733 }, { "epoch": 0.41, "grad_norm": 4.692518849000452, "learning_rate": 6.650064208439746e-06, "loss": 0.1062, "step": 5734 }, { "epoch": 0.41, "grad_norm": 3.539938504962813, "learning_rate": 6.6489679087827485e-06, "loss": 0.2588, "step": 5735 }, { "epoch": 0.41, "grad_norm": 23.358935425827273, "learning_rate": 6.647871520170175e-06, "loss": 1.1171, "step": 5736 }, { "epoch": 0.41, "grad_norm": 9.871373064673616, "learning_rate": 6.646775042661171e-06, "loss": 0.9621, "step": 5737 }, { "epoch": 0.41, "grad_norm": 5.1031706485075805, "learning_rate": 6.645678476314886e-06, "loss": 1.0699, "step": 5738 }, { "epoch": 0.41, "grad_norm": 23.419544193160686, "learning_rate": 6.644581821190478e-06, "loss": 1.2349, "step": 5739 }, { "epoch": 0.41, "grad_norm": 3.572190026228455, "learning_rate": 6.643485077347105e-06, "loss": 0.1562, "step": 5740 }, { "epoch": 0.41, "grad_norm": 9.09196301561434, "learning_rate": 6.642388244843931e-06, "loss": 0.6452, "step": 5741 }, { "epoch": 0.41, "grad_norm": 2.7164857795001915, "learning_rate": 6.64129132374013e-06, "loss": 0.207, "step": 5742 }, { "epoch": 0.41, "grad_norm": 2.957861169306302, "learning_rate": 6.640194314094874e-06, "loss": 0.1972, "step": 5743 }, { "epoch": 0.41, "grad_norm": 5.19265585485569, "learning_rate": 6.639097215967344e-06, "loss": 0.9985, "step": 5744 }, { "epoch": 0.41, "grad_norm": 3.1793319169612193, "learning_rate": 6.638000029416724e-06, "loss": 0.2332, "step": 5745 }, { "epoch": 0.41, "grad_norm": 3.8154884416736725, "learning_rate": 6.636902754502203e-06, "loss": 0.8154, "step": 5746 }, { "epoch": 0.41, "grad_norm": 19.87959995602403, "learning_rate": 6.635805391282972e-06, "loss": 1.4333, "step": 5747 }, { "epoch": 0.41, "grad_norm": 5.039557757561722, "learning_rate": 6.634707939818233e-06, "loss": 1.0697, "step": 5748 }, { "epoch": 0.41, "grad_norm": 17.842924223005653, "learning_rate": 6.633610400167189e-06, "loss": 0.5942, "step": 5749 }, { "epoch": 0.41, "grad_norm": 3.338868577543168, "learning_rate": 6.632512772389046e-06, "loss": 0.2024, "step": 5750 }, { "epoch": 0.41, "grad_norm": 9.844956016864273, "learning_rate": 6.63141505654302e-06, "loss": 1.1855, "step": 5751 }, { "epoch": 0.41, "grad_norm": 4.918228576892649, "learning_rate": 6.630317252688326e-06, "loss": 0.8675, "step": 5752 }, { "epoch": 0.41, "grad_norm": 3.9312994951955313, "learning_rate": 6.629219360884187e-06, "loss": 0.2433, "step": 5753 }, { "epoch": 0.41, "grad_norm": 4.68035002068065, "learning_rate": 6.628121381189832e-06, "loss": 0.7837, "step": 5754 }, { "epoch": 0.41, "grad_norm": 11.487138146258754, "learning_rate": 6.627023313664489e-06, "loss": 0.7553, "step": 5755 }, { "epoch": 0.41, "grad_norm": 3.98932400791987, "learning_rate": 6.6259251583673975e-06, "loss": 0.8548, "step": 5756 }, { "epoch": 0.41, "grad_norm": 3.7758881046913015, "learning_rate": 6.6248269153578e-06, "loss": 0.2492, "step": 5757 }, { "epoch": 0.41, "grad_norm": 7.636076042995474, "learning_rate": 6.62372858469494e-06, "loss": 0.5987, "step": 5758 }, { "epoch": 0.41, "grad_norm": 5.435469526999406, "learning_rate": 6.622630166438068e-06, "loss": 0.3332, "step": 5759 }, { "epoch": 0.41, "grad_norm": 29.72607495402982, "learning_rate": 6.62153166064644e-06, "loss": 0.9275, "step": 5760 }, { "epoch": 0.41, "grad_norm": 3.2690741432021158, "learning_rate": 6.620433067379318e-06, "loss": 0.2547, "step": 5761 }, { "epoch": 0.41, "grad_norm": 8.171479763898029, "learning_rate": 6.619334386695966e-06, "loss": 0.7465, "step": 5762 }, { "epoch": 0.41, "grad_norm": 2.193759405608843, "learning_rate": 6.618235618655653e-06, "loss": 0.1146, "step": 5763 }, { "epoch": 0.41, "grad_norm": 8.11836894653957, "learning_rate": 6.617136763317653e-06, "loss": 0.8237, "step": 5764 }, { "epoch": 0.41, "grad_norm": 10.504001905228378, "learning_rate": 6.616037820741246e-06, "loss": 0.8498, "step": 5765 }, { "epoch": 0.41, "grad_norm": 4.8271200784758035, "learning_rate": 6.614938790985716e-06, "loss": 0.9726, "step": 5766 }, { "epoch": 0.41, "grad_norm": 7.538118049515018, "learning_rate": 6.61383967411035e-06, "loss": 0.2018, "step": 5767 }, { "epoch": 0.41, "grad_norm": 19.165838996416795, "learning_rate": 6.612740470174445e-06, "loss": 0.7804, "step": 5768 }, { "epoch": 0.41, "grad_norm": 3.959949281936674, "learning_rate": 6.6116411792372945e-06, "loss": 0.2444, "step": 5769 }, { "epoch": 0.41, "grad_norm": 3.3307774185759143, "learning_rate": 6.610541801358203e-06, "loss": 0.1893, "step": 5770 }, { "epoch": 0.41, "grad_norm": 8.718214561164956, "learning_rate": 6.609442336596476e-06, "loss": 0.6315, "step": 5771 }, { "epoch": 0.41, "grad_norm": 3.4121800783144107, "learning_rate": 6.6083427850114294e-06, "loss": 0.1789, "step": 5772 }, { "epoch": 0.41, "grad_norm": 3.2349518777153743, "learning_rate": 6.607243146662375e-06, "loss": 0.2514, "step": 5773 }, { "epoch": 0.41, "grad_norm": 4.5061417835776725, "learning_rate": 6.606143421608638e-06, "loss": 0.8791, "step": 5774 }, { "epoch": 0.41, "grad_norm": 2.805088841041371, "learning_rate": 6.6050436099095435e-06, "loss": 0.165, "step": 5775 }, { "epoch": 0.41, "grad_norm": 6.366175145052629, "learning_rate": 6.603943711624421e-06, "loss": 0.2867, "step": 5776 }, { "epoch": 0.41, "grad_norm": 22.200190089700445, "learning_rate": 6.602843726812606e-06, "loss": 1.377, "step": 5777 }, { "epoch": 0.41, "grad_norm": 18.278153702894137, "learning_rate": 6.6017436555334405e-06, "loss": 0.8745, "step": 5778 }, { "epoch": 0.41, "grad_norm": 2.752179711009133, "learning_rate": 6.600643497846268e-06, "loss": 0.1645, "step": 5779 }, { "epoch": 0.41, "grad_norm": 4.5043369970486244, "learning_rate": 6.599543253810435e-06, "loss": 0.8659, "step": 5780 }, { "epoch": 0.41, "grad_norm": 30.867583958556402, "learning_rate": 6.5984429234853e-06, "loss": 1.1763, "step": 5781 }, { "epoch": 0.41, "grad_norm": 13.118442300010113, "learning_rate": 6.5973425069302185e-06, "loss": 0.6547, "step": 5782 }, { "epoch": 0.41, "grad_norm": 13.752389161453522, "learning_rate": 6.596242004204556e-06, "loss": 0.6972, "step": 5783 }, { "epoch": 0.41, "grad_norm": 5.130708901993309, "learning_rate": 6.595141415367678e-06, "loss": 0.9522, "step": 5784 }, { "epoch": 0.41, "grad_norm": 2.895769801699168, "learning_rate": 6.594040740478962e-06, "loss": 0.165, "step": 5785 }, { "epoch": 0.41, "grad_norm": 4.520889187378987, "learning_rate": 6.5929399795977786e-06, "loss": 0.2912, "step": 5786 }, { "epoch": 0.41, "grad_norm": 18.089308498448833, "learning_rate": 6.591839132783514e-06, "loss": 1.1781, "step": 5787 }, { "epoch": 0.42, "grad_norm": 3.592995355184598, "learning_rate": 6.5907382000955546e-06, "loss": 0.2044, "step": 5788 }, { "epoch": 0.42, "grad_norm": 4.658677498507123, "learning_rate": 6.5896371815932895e-06, "loss": 0.8926, "step": 5789 }, { "epoch": 0.42, "grad_norm": 4.1467931666869475, "learning_rate": 6.588536077336114e-06, "loss": 0.2068, "step": 5790 }, { "epoch": 0.42, "grad_norm": 8.012558655594892, "learning_rate": 6.587434887383433e-06, "loss": 0.7241, "step": 5791 }, { "epoch": 0.42, "grad_norm": 3.6190144851148536, "learning_rate": 6.5863336117946465e-06, "loss": 0.2257, "step": 5792 }, { "epoch": 0.42, "grad_norm": 2.8109799311786814, "learning_rate": 6.585232250629167e-06, "loss": 0.176, "step": 5793 }, { "epoch": 0.42, "grad_norm": 16.53077698443912, "learning_rate": 6.584130803946408e-06, "loss": 0.7675, "step": 5794 }, { "epoch": 0.42, "grad_norm": 3.021709664873307, "learning_rate": 6.583029271805788e-06, "loss": 0.2211, "step": 5795 }, { "epoch": 0.42, "grad_norm": 2.6576174028708115, "learning_rate": 6.581927654266731e-06, "loss": 0.219, "step": 5796 }, { "epoch": 0.42, "grad_norm": 6.742090712400669, "learning_rate": 6.580825951388665e-06, "loss": 0.9266, "step": 5797 }, { "epoch": 0.42, "grad_norm": 3.425519483400429, "learning_rate": 6.579724163231023e-06, "loss": 0.2452, "step": 5798 }, { "epoch": 0.42, "grad_norm": 13.604233451015835, "learning_rate": 6.578622289853242e-06, "loss": 1.3017, "step": 5799 }, { "epoch": 0.42, "grad_norm": 12.608778413650017, "learning_rate": 6.577520331314764e-06, "loss": 0.5903, "step": 5800 }, { "epoch": 0.42, "grad_norm": 5.808423206167266, "learning_rate": 6.576418287675035e-06, "loss": 0.9998, "step": 5801 }, { "epoch": 0.42, "grad_norm": 2.5886747415430005, "learning_rate": 6.575316158993507e-06, "loss": 0.129, "step": 5802 }, { "epoch": 0.42, "grad_norm": 5.928025080569568, "learning_rate": 6.574213945329634e-06, "loss": 1.0352, "step": 5803 }, { "epoch": 0.42, "grad_norm": 11.796017104800248, "learning_rate": 6.573111646742878e-06, "loss": 0.7735, "step": 5804 }, { "epoch": 0.42, "grad_norm": 19.464129374035696, "learning_rate": 6.572009263292704e-06, "loss": 0.8414, "step": 5805 }, { "epoch": 0.42, "grad_norm": 3.890495105729724, "learning_rate": 6.57090679503858e-06, "loss": 0.3285, "step": 5806 }, { "epoch": 0.42, "grad_norm": 5.358738691703138, "learning_rate": 6.569804242039978e-06, "loss": 0.7493, "step": 5807 }, { "epoch": 0.42, "grad_norm": 4.851637133343964, "learning_rate": 6.568701604356383e-06, "loss": 1.0398, "step": 5808 }, { "epoch": 0.42, "grad_norm": 5.69468023732162, "learning_rate": 6.5675988820472725e-06, "loss": 0.886, "step": 5809 }, { "epoch": 0.42, "grad_norm": 11.037634741334232, "learning_rate": 6.5664960751721366e-06, "loss": 0.7723, "step": 5810 }, { "epoch": 0.42, "grad_norm": 2.2706205923786666, "learning_rate": 6.565393183790468e-06, "loss": 0.164, "step": 5811 }, { "epoch": 0.42, "grad_norm": 4.040960042247045, "learning_rate": 6.564290207961761e-06, "loss": 0.3635, "step": 5812 }, { "epoch": 0.42, "grad_norm": 2.476268515290046, "learning_rate": 6.563187147745518e-06, "loss": 0.1224, "step": 5813 }, { "epoch": 0.42, "grad_norm": 4.209148959029523, "learning_rate": 6.5620840032012465e-06, "loss": 0.2421, "step": 5814 }, { "epoch": 0.42, "grad_norm": 3.3027568723907983, "learning_rate": 6.560980774388455e-06, "loss": 0.2399, "step": 5815 }, { "epoch": 0.42, "grad_norm": 3.602879608731169, "learning_rate": 6.559877461366659e-06, "loss": 0.2461, "step": 5816 }, { "epoch": 0.42, "grad_norm": 12.582485614335397, "learning_rate": 6.558774064195381e-06, "loss": 1.163, "step": 5817 }, { "epoch": 0.42, "grad_norm": 2.875899193847006, "learning_rate": 6.557670582934139e-06, "loss": 0.2273, "step": 5818 }, { "epoch": 0.42, "grad_norm": 9.92732533047453, "learning_rate": 6.556567017642468e-06, "loss": 0.6536, "step": 5819 }, { "epoch": 0.42, "grad_norm": 4.197607803898969, "learning_rate": 6.555463368379897e-06, "loss": 1.076, "step": 5820 }, { "epoch": 0.42, "grad_norm": 14.370940512817924, "learning_rate": 6.554359635205965e-06, "loss": 1.1307, "step": 5821 }, { "epoch": 0.42, "grad_norm": 28.70664816111552, "learning_rate": 6.553255818180214e-06, "loss": 1.6212, "step": 5822 }, { "epoch": 0.42, "grad_norm": 15.689115032795396, "learning_rate": 6.552151917362193e-06, "loss": 0.6821, "step": 5823 }, { "epoch": 0.42, "grad_norm": 10.681885204381409, "learning_rate": 6.5510479328114485e-06, "loss": 0.7791, "step": 5824 }, { "epoch": 0.42, "grad_norm": 2.9895075668638107, "learning_rate": 6.54994386458754e-06, "loss": 0.1587, "step": 5825 }, { "epoch": 0.42, "grad_norm": 15.954683458439058, "learning_rate": 6.548839712750027e-06, "loss": 0.8757, "step": 5826 }, { "epoch": 0.42, "grad_norm": 3.7361680695133983, "learning_rate": 6.547735477358474e-06, "loss": 0.226, "step": 5827 }, { "epoch": 0.42, "grad_norm": 11.637134971501817, "learning_rate": 6.546631158472452e-06, "loss": 0.8654, "step": 5828 }, { "epoch": 0.42, "grad_norm": 3.6979799844107766, "learning_rate": 6.545526756151532e-06, "loss": 0.2584, "step": 5829 }, { "epoch": 0.42, "grad_norm": 3.476405615492388, "learning_rate": 6.544422270455293e-06, "loss": 0.2909, "step": 5830 }, { "epoch": 0.42, "grad_norm": 18.997578823077866, "learning_rate": 6.54331770144332e-06, "loss": 0.7913, "step": 5831 }, { "epoch": 0.42, "grad_norm": 12.922578780255124, "learning_rate": 6.542213049175197e-06, "loss": 0.8455, "step": 5832 }, { "epoch": 0.42, "grad_norm": 3.5705337957450825, "learning_rate": 6.541108313710519e-06, "loss": 0.205, "step": 5833 }, { "epoch": 0.42, "grad_norm": 7.645950419773876, "learning_rate": 6.54000349510888e-06, "loss": 0.728, "step": 5834 }, { "epoch": 0.42, "grad_norm": 3.546138496198238, "learning_rate": 6.538898593429883e-06, "loss": 0.2996, "step": 5835 }, { "epoch": 0.42, "grad_norm": 3.764728726644717, "learning_rate": 6.537793608733131e-06, "loss": 0.3733, "step": 5836 }, { "epoch": 0.42, "grad_norm": 4.575791274051024, "learning_rate": 6.536688541078236e-06, "loss": 1.0201, "step": 5837 }, { "epoch": 0.42, "grad_norm": 23.730127334029845, "learning_rate": 6.535583390524808e-06, "loss": 1.5104, "step": 5838 }, { "epoch": 0.42, "grad_norm": 4.448641670449938, "learning_rate": 6.534478157132471e-06, "loss": 0.2024, "step": 5839 }, { "epoch": 0.42, "grad_norm": 3.664619962669236, "learning_rate": 6.533372840960844e-06, "loss": 0.1554, "step": 5840 }, { "epoch": 0.42, "grad_norm": 4.191994819700733, "learning_rate": 6.532267442069558e-06, "loss": 0.2989, "step": 5841 }, { "epoch": 0.42, "grad_norm": 3.6648268420908185, "learning_rate": 6.531161960518243e-06, "loss": 0.212, "step": 5842 }, { "epoch": 0.42, "grad_norm": 5.467546985076111, "learning_rate": 6.530056396366535e-06, "loss": 0.9381, "step": 5843 }, { "epoch": 0.42, "grad_norm": 3.734462034633231, "learning_rate": 6.528950749674077e-06, "loss": 0.1974, "step": 5844 }, { "epoch": 0.42, "grad_norm": 4.2525905176213925, "learning_rate": 6.527845020500514e-06, "loss": 0.2538, "step": 5845 }, { "epoch": 0.42, "grad_norm": 3.281635502430485, "learning_rate": 6.526739208905494e-06, "loss": 0.1107, "step": 5846 }, { "epoch": 0.42, "grad_norm": 20.847069691420224, "learning_rate": 6.525633314948673e-06, "loss": 1.0557, "step": 5847 }, { "epoch": 0.42, "grad_norm": 19.807528174131434, "learning_rate": 6.52452733868971e-06, "loss": 1.1511, "step": 5848 }, { "epoch": 0.42, "grad_norm": 3.263556859388019, "learning_rate": 6.523421280188267e-06, "loss": 0.1806, "step": 5849 }, { "epoch": 0.42, "grad_norm": 3.336057533136698, "learning_rate": 6.522315139504013e-06, "loss": 0.2021, "step": 5850 }, { "epoch": 0.42, "grad_norm": 3.9258298608408215, "learning_rate": 6.5212089166966195e-06, "loss": 0.1926, "step": 5851 }, { "epoch": 0.42, "grad_norm": 4.708815886359908, "learning_rate": 6.520102611825763e-06, "loss": 0.2789, "step": 5852 }, { "epoch": 0.42, "grad_norm": 5.073632842952179, "learning_rate": 6.5189962249511244e-06, "loss": 0.8536, "step": 5853 }, { "epoch": 0.42, "grad_norm": 12.31538418670865, "learning_rate": 6.517889756132389e-06, "loss": 0.6985, "step": 5854 }, { "epoch": 0.42, "grad_norm": 2.5773604318733754, "learning_rate": 6.516783205429247e-06, "loss": 0.1186, "step": 5855 }, { "epoch": 0.42, "grad_norm": 3.1527341529460764, "learning_rate": 6.515676572901392e-06, "loss": 0.1937, "step": 5856 }, { "epoch": 0.42, "grad_norm": 6.606908586249752, "learning_rate": 6.514569858608524e-06, "loss": 0.8548, "step": 5857 }, { "epoch": 0.42, "grad_norm": 3.8059588249782186, "learning_rate": 6.513463062610344e-06, "loss": 0.2684, "step": 5858 }, { "epoch": 0.42, "grad_norm": 4.384257577426368, "learning_rate": 6.512356184966562e-06, "loss": 0.2297, "step": 5859 }, { "epoch": 0.42, "grad_norm": 3.472061855770975, "learning_rate": 6.511249225736888e-06, "loss": 0.2133, "step": 5860 }, { "epoch": 0.42, "grad_norm": 3.0908774268363124, "learning_rate": 6.510142184981036e-06, "loss": 0.1436, "step": 5861 }, { "epoch": 0.42, "grad_norm": 4.864689908478867, "learning_rate": 6.5090350627587325e-06, "loss": 0.8797, "step": 5862 }, { "epoch": 0.42, "grad_norm": 3.44318501891494, "learning_rate": 6.507927859129698e-06, "loss": 0.1995, "step": 5863 }, { "epoch": 0.42, "grad_norm": 4.799730490072334, "learning_rate": 6.506820574153662e-06, "loss": 0.8255, "step": 5864 }, { "epoch": 0.42, "grad_norm": 11.145375222561828, "learning_rate": 6.505713207890364e-06, "loss": 0.8924, "step": 5865 }, { "epoch": 0.42, "grad_norm": 3.8062334427827955, "learning_rate": 6.504605760399534e-06, "loss": 0.2526, "step": 5866 }, { "epoch": 0.42, "grad_norm": 5.518251122585073, "learning_rate": 6.503498231740919e-06, "loss": 0.2223, "step": 5867 }, { "epoch": 0.42, "grad_norm": 4.666593452659534, "learning_rate": 6.502390621974267e-06, "loss": 0.7799, "step": 5868 }, { "epoch": 0.42, "grad_norm": 3.2533286372635337, "learning_rate": 6.501282931159326e-06, "loss": 0.1879, "step": 5869 }, { "epoch": 0.42, "grad_norm": 19.04896250262296, "learning_rate": 6.500175159355855e-06, "loss": 0.9124, "step": 5870 }, { "epoch": 0.42, "grad_norm": 3.6767880710859027, "learning_rate": 6.499067306623612e-06, "loss": 0.2212, "step": 5871 }, { "epoch": 0.42, "grad_norm": 5.851505931275082, "learning_rate": 6.497959373022362e-06, "loss": 0.2555, "step": 5872 }, { "epoch": 0.42, "grad_norm": 3.3141029632792747, "learning_rate": 6.496851358611874e-06, "loss": 0.1826, "step": 5873 }, { "epoch": 0.42, "grad_norm": 4.443888179654615, "learning_rate": 6.495743263451921e-06, "loss": 0.7684, "step": 5874 }, { "epoch": 0.42, "grad_norm": 3.164962097197821, "learning_rate": 6.49463508760228e-06, "loss": 0.1719, "step": 5875 }, { "epoch": 0.42, "grad_norm": 5.424563584609726, "learning_rate": 6.493526831122735e-06, "loss": 1.0568, "step": 5876 }, { "epoch": 0.42, "grad_norm": 17.817959197995634, "learning_rate": 6.4924184940730696e-06, "loss": 0.8648, "step": 5877 }, { "epoch": 0.42, "grad_norm": 20.557814438239372, "learning_rate": 6.491310076513076e-06, "loss": 1.1462, "step": 5878 }, { "epoch": 0.42, "grad_norm": 3.8846878853394, "learning_rate": 6.4902015785025494e-06, "loss": 0.1603, "step": 5879 }, { "epoch": 0.42, "grad_norm": 3.2284391423355663, "learning_rate": 6.489093000101288e-06, "loss": 0.1686, "step": 5880 }, { "epoch": 0.42, "grad_norm": 21.307149847936863, "learning_rate": 6.487984341369093e-06, "loss": 0.848, "step": 5881 }, { "epoch": 0.42, "grad_norm": 8.954053603425708, "learning_rate": 6.486875602365778e-06, "loss": 0.6581, "step": 5882 }, { "epoch": 0.42, "grad_norm": 11.465686625641416, "learning_rate": 6.485766783151153e-06, "loss": 0.8127, "step": 5883 }, { "epoch": 0.42, "grad_norm": 5.515890641918209, "learning_rate": 6.484657883785032e-06, "loss": 0.7809, "step": 5884 }, { "epoch": 0.42, "grad_norm": 3.4142670653562557, "learning_rate": 6.4835489043272395e-06, "loss": 0.2063, "step": 5885 }, { "epoch": 0.42, "grad_norm": 4.8554769334451295, "learning_rate": 6.482439844837598e-06, "loss": 0.8053, "step": 5886 }, { "epoch": 0.42, "grad_norm": 3.943004367324808, "learning_rate": 6.481330705375939e-06, "loss": 0.288, "step": 5887 }, { "epoch": 0.42, "grad_norm": 2.9256915030428194, "learning_rate": 6.480221486002095e-06, "loss": 0.1308, "step": 5888 }, { "epoch": 0.42, "grad_norm": 4.421792899555339, "learning_rate": 6.4791121867759045e-06, "loss": 0.8141, "step": 5889 }, { "epoch": 0.42, "grad_norm": 21.847456450866378, "learning_rate": 6.478002807757211e-06, "loss": 1.351, "step": 5890 }, { "epoch": 0.42, "grad_norm": 21.060056859063824, "learning_rate": 6.476893349005861e-06, "loss": 0.8915, "step": 5891 }, { "epoch": 0.42, "grad_norm": 17.496018248589813, "learning_rate": 6.4757838105817035e-06, "loss": 0.9207, "step": 5892 }, { "epoch": 0.42, "grad_norm": 13.648773527076571, "learning_rate": 6.474674192544598e-06, "loss": 0.6108, "step": 5893 }, { "epoch": 0.42, "grad_norm": 3.536255188404024, "learning_rate": 6.473564494954399e-06, "loss": 0.2535, "step": 5894 }, { "epoch": 0.42, "grad_norm": 3.2915347756345907, "learning_rate": 6.472454717870974e-06, "loss": 0.2128, "step": 5895 }, { "epoch": 0.42, "grad_norm": 4.682000817479189, "learning_rate": 6.471344861354193e-06, "loss": 0.8391, "step": 5896 }, { "epoch": 0.42, "grad_norm": 3.524590107965362, "learning_rate": 6.470234925463923e-06, "loss": 0.2521, "step": 5897 }, { "epoch": 0.42, "grad_norm": 11.339948545275407, "learning_rate": 6.4691249102600454e-06, "loss": 0.3574, "step": 5898 }, { "epoch": 0.42, "grad_norm": 3.3492523750883536, "learning_rate": 6.468014815802439e-06, "loss": 0.1975, "step": 5899 }, { "epoch": 0.42, "grad_norm": 3.163920118678488, "learning_rate": 6.466904642150991e-06, "loss": 0.188, "step": 5900 }, { "epoch": 0.42, "grad_norm": 3.4490888119671443, "learning_rate": 6.465794389365589e-06, "loss": 0.2705, "step": 5901 }, { "epoch": 0.42, "grad_norm": 15.124962392621894, "learning_rate": 6.4646840575061295e-06, "loss": 0.9397, "step": 5902 }, { "epoch": 0.42, "grad_norm": 17.314577252670347, "learning_rate": 6.463573646632508e-06, "loss": 0.7002, "step": 5903 }, { "epoch": 0.42, "grad_norm": 26.351388701075706, "learning_rate": 6.462463156804628e-06, "loss": 1.4927, "step": 5904 }, { "epoch": 0.42, "grad_norm": 19.936468018544513, "learning_rate": 6.461352588082397e-06, "loss": 1.1974, "step": 5905 }, { "epoch": 0.42, "grad_norm": 16.912166739618417, "learning_rate": 6.460241940525724e-06, "loss": 0.8397, "step": 5906 }, { "epoch": 0.42, "grad_norm": 20.21208550677685, "learning_rate": 6.459131214194528e-06, "loss": 1.1113, "step": 5907 }, { "epoch": 0.42, "grad_norm": 4.932587688863323, "learning_rate": 6.458020409148727e-06, "loss": 0.8309, "step": 5908 }, { "epoch": 0.42, "grad_norm": 4.0808843442094, "learning_rate": 6.45690952544824e-06, "loss": 0.2506, "step": 5909 }, { "epoch": 0.42, "grad_norm": 4.809715852483498, "learning_rate": 6.455798563153002e-06, "loss": 1.0156, "step": 5910 }, { "epoch": 0.42, "grad_norm": 4.120252693306023, "learning_rate": 6.45468752232294e-06, "loss": 0.2102, "step": 5911 }, { "epoch": 0.42, "grad_norm": 5.139730329142542, "learning_rate": 6.453576403017992e-06, "loss": 1.0881, "step": 5912 }, { "epoch": 0.42, "grad_norm": 11.704411107950989, "learning_rate": 6.452465205298101e-06, "loss": 0.929, "step": 5913 }, { "epoch": 0.42, "grad_norm": 3.0740773673475528, "learning_rate": 6.45135392922321e-06, "loss": 0.1796, "step": 5914 }, { "epoch": 0.42, "grad_norm": 2.867899595596659, "learning_rate": 6.450242574853267e-06, "loss": 0.1741, "step": 5915 }, { "epoch": 0.42, "grad_norm": 2.870504573595666, "learning_rate": 6.4491311422482276e-06, "loss": 0.2204, "step": 5916 }, { "epoch": 0.42, "grad_norm": 2.988459614687739, "learning_rate": 6.448019631468048e-06, "loss": 0.1538, "step": 5917 }, { "epoch": 0.42, "grad_norm": 11.08163640154043, "learning_rate": 6.4469080425726885e-06, "loss": 0.7673, "step": 5918 }, { "epoch": 0.42, "grad_norm": 3.9142369017034304, "learning_rate": 6.44579637562212e-06, "loss": 0.2795, "step": 5919 }, { "epoch": 0.42, "grad_norm": 13.531104504174106, "learning_rate": 6.444684630676308e-06, "loss": 0.6131, "step": 5920 }, { "epoch": 0.42, "grad_norm": 5.152083131803346, "learning_rate": 6.443572807795229e-06, "loss": 0.8477, "step": 5921 }, { "epoch": 0.42, "grad_norm": 3.398681619326718, "learning_rate": 6.442460907038861e-06, "loss": 0.1688, "step": 5922 }, { "epoch": 0.42, "grad_norm": 17.967850163972848, "learning_rate": 6.441348928467187e-06, "loss": 1.2103, "step": 5923 }, { "epoch": 0.42, "grad_norm": 15.827255994232933, "learning_rate": 6.4402368721401955e-06, "loss": 0.7697, "step": 5924 }, { "epoch": 0.42, "grad_norm": 16.19611384016933, "learning_rate": 6.4391247381178744e-06, "loss": 0.7462, "step": 5925 }, { "epoch": 0.42, "grad_norm": 12.949500135675013, "learning_rate": 6.438012526460223e-06, "loss": 0.9234, "step": 5926 }, { "epoch": 0.43, "grad_norm": 3.1872651197621873, "learning_rate": 6.436900237227236e-06, "loss": 0.2038, "step": 5927 }, { "epoch": 0.43, "grad_norm": 4.747155213776615, "learning_rate": 6.435787870478922e-06, "loss": 0.3294, "step": 5928 }, { "epoch": 0.43, "grad_norm": 11.170980103022103, "learning_rate": 6.434675426275287e-06, "loss": 1.1543, "step": 5929 }, { "epoch": 0.43, "grad_norm": 4.06071910419386, "learning_rate": 6.433562904676345e-06, "loss": 0.3019, "step": 5930 }, { "epoch": 0.43, "grad_norm": 4.559243872784371, "learning_rate": 6.432450305742107e-06, "loss": 0.2357, "step": 5931 }, { "epoch": 0.43, "grad_norm": 10.646568499747596, "learning_rate": 6.4313376295326005e-06, "loss": 0.6608, "step": 5932 }, { "epoch": 0.43, "grad_norm": 4.110095141355345, "learning_rate": 6.430224876107846e-06, "loss": 0.2898, "step": 5933 }, { "epoch": 0.43, "grad_norm": 3.0254854433090106, "learning_rate": 6.429112045527872e-06, "loss": 0.1612, "step": 5934 }, { "epoch": 0.43, "grad_norm": 3.6270414003857705, "learning_rate": 6.427999137852712e-06, "loss": 0.307, "step": 5935 }, { "epoch": 0.43, "grad_norm": 4.568644372443428, "learning_rate": 6.426886153142407e-06, "loss": 0.8985, "step": 5936 }, { "epoch": 0.43, "grad_norm": 2.7096884697391124, "learning_rate": 6.4257730914569915e-06, "loss": 0.2141, "step": 5937 }, { "epoch": 0.43, "grad_norm": 2.7936349288864344, "learning_rate": 6.424659952856516e-06, "loss": 0.1732, "step": 5938 }, { "epoch": 0.43, "grad_norm": 16.545544831347275, "learning_rate": 6.4235467374010295e-06, "loss": 0.707, "step": 5939 }, { "epoch": 0.43, "grad_norm": 9.438219289236363, "learning_rate": 6.4224334451505845e-06, "loss": 0.435, "step": 5940 }, { "epoch": 0.43, "grad_norm": 11.236153012134166, "learning_rate": 6.4213200761652386e-06, "loss": 1.177, "step": 5941 }, { "epoch": 0.43, "grad_norm": 15.962751978336211, "learning_rate": 6.420206630505055e-06, "loss": 1.0214, "step": 5942 }, { "epoch": 0.43, "grad_norm": 17.34856417566295, "learning_rate": 6.419093108230099e-06, "loss": 1.8378, "step": 5943 }, { "epoch": 0.43, "grad_norm": 12.728724351386354, "learning_rate": 6.417979509400443e-06, "loss": 1.162, "step": 5944 }, { "epoch": 0.43, "grad_norm": 3.0375980247813756, "learning_rate": 6.416865834076158e-06, "loss": 0.1685, "step": 5945 }, { "epoch": 0.43, "grad_norm": 3.3515279866057344, "learning_rate": 6.415752082317326e-06, "loss": 0.221, "step": 5946 }, { "epoch": 0.43, "grad_norm": 5.510511454406904, "learning_rate": 6.414638254184027e-06, "loss": 0.9573, "step": 5947 }, { "epoch": 0.43, "grad_norm": 12.46467776225219, "learning_rate": 6.413524349736348e-06, "loss": 0.8389, "step": 5948 }, { "epoch": 0.43, "grad_norm": 11.017075999588075, "learning_rate": 6.412410369034382e-06, "loss": 0.6679, "step": 5949 }, { "epoch": 0.43, "grad_norm": 7.480771034376809, "learning_rate": 6.411296312138224e-06, "loss": 0.8474, "step": 5950 }, { "epoch": 0.43, "grad_norm": 14.079380078415632, "learning_rate": 6.41018217910797e-06, "loss": 0.9703, "step": 5951 }, { "epoch": 0.43, "grad_norm": 8.893693701492337, "learning_rate": 6.409067970003725e-06, "loss": 0.6836, "step": 5952 }, { "epoch": 0.43, "grad_norm": 4.932110829446761, "learning_rate": 6.407953684885598e-06, "loss": 0.9733, "step": 5953 }, { "epoch": 0.43, "grad_norm": 17.63093528451389, "learning_rate": 6.406839323813699e-06, "loss": 1.1659, "step": 5954 }, { "epoch": 0.43, "grad_norm": 18.660473132887187, "learning_rate": 6.405724886848143e-06, "loss": 1.1236, "step": 5955 }, { "epoch": 0.43, "grad_norm": 11.477564942115457, "learning_rate": 6.404610374049052e-06, "loss": 0.5079, "step": 5956 }, { "epoch": 0.43, "grad_norm": 14.835369856667718, "learning_rate": 6.403495785476546e-06, "loss": 0.6273, "step": 5957 }, { "epoch": 0.43, "grad_norm": 17.992755169977745, "learning_rate": 6.402381121190755e-06, "loss": 1.0139, "step": 5958 }, { "epoch": 0.43, "grad_norm": 3.6805991864299576, "learning_rate": 6.401266381251812e-06, "loss": 0.1985, "step": 5959 }, { "epoch": 0.43, "grad_norm": 3.9823169365740885, "learning_rate": 6.400151565719851e-06, "loss": 0.4302, "step": 5960 }, { "epoch": 0.43, "grad_norm": 12.705758112047025, "learning_rate": 6.399036674655014e-06, "loss": 0.5558, "step": 5961 }, { "epoch": 0.43, "grad_norm": 3.633705750061384, "learning_rate": 6.3979217081174425e-06, "loss": 0.202, "step": 5962 }, { "epoch": 0.43, "grad_norm": 3.2100345954964458, "learning_rate": 6.396806666167287e-06, "loss": 0.2462, "step": 5963 }, { "epoch": 0.43, "grad_norm": 13.200410936888275, "learning_rate": 6.395691548864699e-06, "loss": 0.6849, "step": 5964 }, { "epoch": 0.43, "grad_norm": 22.77614462380359, "learning_rate": 6.394576356269835e-06, "loss": 1.5564, "step": 5965 }, { "epoch": 0.43, "grad_norm": 4.127750154810139, "learning_rate": 6.3934610884428535e-06, "loss": 0.2416, "step": 5966 }, { "epoch": 0.43, "grad_norm": 3.781370538028891, "learning_rate": 6.392345745443924e-06, "loss": 0.279, "step": 5967 }, { "epoch": 0.43, "grad_norm": 3.7616771112558123, "learning_rate": 6.391230327333211e-06, "loss": 0.1929, "step": 5968 }, { "epoch": 0.43, "grad_norm": 3.4997169521435785, "learning_rate": 6.390114834170887e-06, "loss": 0.1986, "step": 5969 }, { "epoch": 0.43, "grad_norm": 11.66649567536985, "learning_rate": 6.388999266017131e-06, "loss": 0.8113, "step": 5970 }, { "epoch": 0.43, "grad_norm": 3.8055581259776825, "learning_rate": 6.3878836229321195e-06, "loss": 0.2187, "step": 5971 }, { "epoch": 0.43, "grad_norm": 4.068800858406206, "learning_rate": 6.386767904976041e-06, "loss": 0.3103, "step": 5972 }, { "epoch": 0.43, "grad_norm": 11.880020404983798, "learning_rate": 6.385652112209085e-06, "loss": 1.0616, "step": 5973 }, { "epoch": 0.43, "grad_norm": 2.6678745774533317, "learning_rate": 6.384536244691442e-06, "loss": 0.2897, "step": 5974 }, { "epoch": 0.43, "grad_norm": 4.709661108099903, "learning_rate": 6.383420302483308e-06, "loss": 1.1158, "step": 5975 }, { "epoch": 0.43, "grad_norm": 13.307511314511308, "learning_rate": 6.382304285644887e-06, "loss": 0.642, "step": 5976 }, { "epoch": 0.43, "grad_norm": 12.535055492346059, "learning_rate": 6.38118819423638e-06, "loss": 0.497, "step": 5977 }, { "epoch": 0.43, "grad_norm": 2.9909829465328457, "learning_rate": 6.380072028318e-06, "loss": 0.1602, "step": 5978 }, { "epoch": 0.43, "grad_norm": 25.710806079866483, "learning_rate": 6.378955787949956e-06, "loss": 1.0751, "step": 5979 }, { "epoch": 0.43, "grad_norm": 12.326584892704505, "learning_rate": 6.3778394731924684e-06, "loss": 0.9861, "step": 5980 }, { "epoch": 0.43, "grad_norm": 4.466631742966496, "learning_rate": 6.376723084105756e-06, "loss": 0.8536, "step": 5981 }, { "epoch": 0.43, "grad_norm": 3.493318531357907, "learning_rate": 6.375606620750045e-06, "loss": 0.2121, "step": 5982 }, { "epoch": 0.43, "grad_norm": 10.803662793298138, "learning_rate": 6.374490083185563e-06, "loss": 1.2334, "step": 5983 }, { "epoch": 0.43, "grad_norm": 11.27771514426524, "learning_rate": 6.3733734714725434e-06, "loss": 0.9988, "step": 5984 }, { "epoch": 0.43, "grad_norm": 5.594545731757319, "learning_rate": 6.372256785671224e-06, "loss": 0.8833, "step": 5985 }, { "epoch": 0.43, "grad_norm": 2.9848732974478267, "learning_rate": 6.371140025841845e-06, "loss": 0.1639, "step": 5986 }, { "epoch": 0.43, "grad_norm": 3.915012734934581, "learning_rate": 6.370023192044653e-06, "loss": 0.2446, "step": 5987 }, { "epoch": 0.43, "grad_norm": 4.2060588518531405, "learning_rate": 6.3689062843398945e-06, "loss": 0.7734, "step": 5988 }, { "epoch": 0.43, "grad_norm": 18.85358105143909, "learning_rate": 6.3677893027878234e-06, "loss": 1.074, "step": 5989 }, { "epoch": 0.43, "grad_norm": 18.0882998890718, "learning_rate": 6.366672247448697e-06, "loss": 1.1104, "step": 5990 }, { "epoch": 0.43, "grad_norm": 3.70862025927688, "learning_rate": 6.365555118382776e-06, "loss": 0.213, "step": 5991 }, { "epoch": 0.43, "grad_norm": 3.7644523320483434, "learning_rate": 6.364437915650326e-06, "loss": 0.225, "step": 5992 }, { "epoch": 0.43, "grad_norm": 3.506800667942387, "learning_rate": 6.363320639311615e-06, "loss": 0.2806, "step": 5993 }, { "epoch": 0.43, "grad_norm": 5.621470805256807, "learning_rate": 6.362203289426913e-06, "loss": 0.9527, "step": 5994 }, { "epoch": 0.43, "grad_norm": 5.613709875910908, "learning_rate": 6.361085866056502e-06, "loss": 0.2478, "step": 5995 }, { "epoch": 0.43, "grad_norm": 4.5822434008860204, "learning_rate": 6.359968369260662e-06, "loss": 0.8681, "step": 5996 }, { "epoch": 0.43, "grad_norm": 3.861337296451031, "learning_rate": 6.358850799099674e-06, "loss": 0.1349, "step": 5997 }, { "epoch": 0.43, "grad_norm": 12.921717081819256, "learning_rate": 6.357733155633831e-06, "loss": 0.8009, "step": 5998 }, { "epoch": 0.43, "grad_norm": 11.827530552150424, "learning_rate": 6.356615438923422e-06, "loss": 1.0396, "step": 5999 }, { "epoch": 0.43, "grad_norm": 3.8992673904947366, "learning_rate": 6.3554976490287456e-06, "loss": 0.2492, "step": 6000 }, { "epoch": 0.43, "grad_norm": 12.281347308693396, "learning_rate": 6.354379786010102e-06, "loss": 0.9134, "step": 6001 }, { "epoch": 0.43, "grad_norm": 13.698140001108845, "learning_rate": 6.353261849927796e-06, "loss": 0.7899, "step": 6002 }, { "epoch": 0.43, "grad_norm": 3.4015427360394366, "learning_rate": 6.352143840842133e-06, "loss": 0.2636, "step": 6003 }, { "epoch": 0.43, "grad_norm": 4.21074707151424, "learning_rate": 6.3510257588134315e-06, "loss": 0.8037, "step": 6004 }, { "epoch": 0.43, "grad_norm": 9.677217687045998, "learning_rate": 6.3499076039020025e-06, "loss": 0.4086, "step": 6005 }, { "epoch": 0.43, "grad_norm": 13.216311713306657, "learning_rate": 6.348789376168167e-06, "loss": 1.0875, "step": 6006 }, { "epoch": 0.43, "grad_norm": 17.68468688961123, "learning_rate": 6.347671075672252e-06, "loss": 0.9784, "step": 6007 }, { "epoch": 0.43, "grad_norm": 4.291310525417693, "learning_rate": 6.3465527024745825e-06, "loss": 0.7847, "step": 6008 }, { "epoch": 0.43, "grad_norm": 5.283879092308835, "learning_rate": 6.345434256635491e-06, "loss": 0.3418, "step": 6009 }, { "epoch": 0.43, "grad_norm": 10.698375563760752, "learning_rate": 6.344315738215316e-06, "loss": 0.661, "step": 6010 }, { "epoch": 0.43, "grad_norm": 4.775482043875996, "learning_rate": 6.343197147274395e-06, "loss": 0.814, "step": 6011 }, { "epoch": 0.43, "grad_norm": 4.520804071079497, "learning_rate": 6.3420784838730705e-06, "loss": 0.8068, "step": 6012 }, { "epoch": 0.43, "grad_norm": 3.852927338117474, "learning_rate": 6.340959748071695e-06, "loss": 0.31, "step": 6013 }, { "epoch": 0.43, "grad_norm": 20.49041506531838, "learning_rate": 6.339840939930614e-06, "loss": 1.2937, "step": 6014 }, { "epoch": 0.43, "grad_norm": 19.752327824693428, "learning_rate": 6.338722059510187e-06, "loss": 0.8666, "step": 6015 }, { "epoch": 0.43, "grad_norm": 9.678813539612726, "learning_rate": 6.337603106870772e-06, "loss": 0.172, "step": 6016 }, { "epoch": 0.43, "grad_norm": 13.118955846926282, "learning_rate": 6.336484082072732e-06, "loss": 1.3725, "step": 6017 }, { "epoch": 0.43, "grad_norm": 4.929121431205815, "learning_rate": 6.335364985176435e-06, "loss": 0.1751, "step": 6018 }, { "epoch": 0.43, "grad_norm": 4.523370710173607, "learning_rate": 6.334245816242253e-06, "loss": 0.2383, "step": 6019 }, { "epoch": 0.43, "grad_norm": 4.962200496492447, "learning_rate": 6.333126575330557e-06, "loss": 0.8249, "step": 6020 }, { "epoch": 0.43, "grad_norm": 31.54533193565364, "learning_rate": 6.332007262501731e-06, "loss": 0.8782, "step": 6021 }, { "epoch": 0.43, "grad_norm": 4.060765405297221, "learning_rate": 6.330887877816154e-06, "loss": 0.3281, "step": 6022 }, { "epoch": 0.43, "grad_norm": 12.941425250976812, "learning_rate": 6.3297684213342126e-06, "loss": 0.7343, "step": 6023 }, { "epoch": 0.43, "grad_norm": 11.06315781207901, "learning_rate": 6.3286488931163e-06, "loss": 0.6089, "step": 6024 }, { "epoch": 0.43, "grad_norm": 11.56251770746398, "learning_rate": 6.327529293222808e-06, "loss": 0.8228, "step": 6025 }, { "epoch": 0.43, "grad_norm": 3.6667363164673406, "learning_rate": 6.326409621714135e-06, "loss": 0.2426, "step": 6026 }, { "epoch": 0.43, "grad_norm": 17.84355049363492, "learning_rate": 6.325289878650686e-06, "loss": 1.1527, "step": 6027 }, { "epoch": 0.43, "grad_norm": 24.680710688255186, "learning_rate": 6.324170064092863e-06, "loss": 1.5633, "step": 6028 }, { "epoch": 0.43, "grad_norm": 4.019365702109752, "learning_rate": 6.323050178101076e-06, "loss": 0.1442, "step": 6029 }, { "epoch": 0.43, "grad_norm": 5.148214518387017, "learning_rate": 6.3219302207357425e-06, "loss": 0.9053, "step": 6030 }, { "epoch": 0.43, "grad_norm": 5.5287892152476115, "learning_rate": 6.320810192057275e-06, "loss": 0.428, "step": 6031 }, { "epoch": 0.43, "grad_norm": 5.018642931776574, "learning_rate": 6.3196900921260985e-06, "loss": 0.897, "step": 6032 }, { "epoch": 0.43, "grad_norm": 4.963325265923096, "learning_rate": 6.318569921002636e-06, "loss": 0.3504, "step": 6033 }, { "epoch": 0.43, "grad_norm": 5.130064723126205, "learning_rate": 6.317449678747318e-06, "loss": 0.9106, "step": 6034 }, { "epoch": 0.43, "grad_norm": 15.427883978563667, "learning_rate": 6.316329365420577e-06, "loss": 0.9875, "step": 6035 }, { "epoch": 0.43, "grad_norm": 19.357597887975878, "learning_rate": 6.315208981082849e-06, "loss": 0.8565, "step": 6036 }, { "epoch": 0.43, "grad_norm": 4.013582690022181, "learning_rate": 6.314088525794574e-06, "loss": 0.2189, "step": 6037 }, { "epoch": 0.43, "grad_norm": 3.4908298761618144, "learning_rate": 6.312967999616199e-06, "loss": 0.1686, "step": 6038 }, { "epoch": 0.43, "grad_norm": 3.5317452344619533, "learning_rate": 6.311847402608168e-06, "loss": 0.1537, "step": 6039 }, { "epoch": 0.43, "grad_norm": 3.6067659675477546, "learning_rate": 6.310726734830936e-06, "loss": 0.29, "step": 6040 }, { "epoch": 0.43, "grad_norm": 3.439198902995756, "learning_rate": 6.30960599634496e-06, "loss": 0.1407, "step": 6041 }, { "epoch": 0.43, "grad_norm": 19.690565489266273, "learning_rate": 6.308485187210696e-06, "loss": 1.2527, "step": 6042 }, { "epoch": 0.43, "grad_norm": 3.547815825740564, "learning_rate": 6.307364307488607e-06, "loss": 0.1819, "step": 6043 }, { "epoch": 0.43, "grad_norm": 24.753697549371246, "learning_rate": 6.306243357239166e-06, "loss": 1.1841, "step": 6044 }, { "epoch": 0.43, "grad_norm": 16.602064339463144, "learning_rate": 6.3051223365228376e-06, "loss": 1.2721, "step": 6045 }, { "epoch": 0.43, "grad_norm": 16.559373940338737, "learning_rate": 6.3040012454001e-06, "loss": 1.0708, "step": 6046 }, { "epoch": 0.43, "grad_norm": 3.2772027293752473, "learning_rate": 6.302880083931432e-06, "loss": 0.2332, "step": 6047 }, { "epoch": 0.43, "grad_norm": 3.030480932212935, "learning_rate": 6.301758852177313e-06, "loss": 0.1683, "step": 6048 }, { "epoch": 0.43, "grad_norm": 20.114036152465737, "learning_rate": 6.300637550198233e-06, "loss": 0.8669, "step": 6049 }, { "epoch": 0.43, "grad_norm": 2.9801334261009376, "learning_rate": 6.299516178054681e-06, "loss": 0.1043, "step": 6050 }, { "epoch": 0.43, "grad_norm": 4.9896766492795805, "learning_rate": 6.2983947358071485e-06, "loss": 0.8216, "step": 6051 }, { "epoch": 0.43, "grad_norm": 3.9665343079631112, "learning_rate": 6.297273223516136e-06, "loss": 0.1924, "step": 6052 }, { "epoch": 0.43, "grad_norm": 4.353187644737137, "learning_rate": 6.296151641242143e-06, "loss": 0.3693, "step": 6053 }, { "epoch": 0.43, "grad_norm": 3.3722849251665044, "learning_rate": 6.2950299890456755e-06, "loss": 0.1847, "step": 6054 }, { "epoch": 0.43, "grad_norm": 26.922230058458965, "learning_rate": 6.293908266987242e-06, "loss": 1.3, "step": 6055 }, { "epoch": 0.43, "grad_norm": 3.406385131772566, "learning_rate": 6.292786475127355e-06, "loss": 0.2215, "step": 6056 }, { "epoch": 0.43, "grad_norm": 6.925590062032104, "learning_rate": 6.291664613526531e-06, "loss": 0.5004, "step": 6057 }, { "epoch": 0.43, "grad_norm": 1.869648680906193, "learning_rate": 6.290542682245291e-06, "loss": 0.1133, "step": 6058 }, { "epoch": 0.43, "grad_norm": 4.69332391254574, "learning_rate": 6.289420681344157e-06, "loss": 0.7405, "step": 6059 }, { "epoch": 0.43, "grad_norm": 5.606191465680686, "learning_rate": 6.288298610883659e-06, "loss": 1.0795, "step": 6060 }, { "epoch": 0.43, "grad_norm": 4.092202449419159, "learning_rate": 6.2871764709243265e-06, "loss": 0.8664, "step": 6061 }, { "epoch": 0.43, "grad_norm": 2.6161088514195834, "learning_rate": 6.286054261526695e-06, "loss": 0.2036, "step": 6062 }, { "epoch": 0.43, "grad_norm": 3.110699523681992, "learning_rate": 6.2849319827513054e-06, "loss": 0.1372, "step": 6063 }, { "epoch": 0.43, "grad_norm": 4.008769062072912, "learning_rate": 6.283809634658699e-06, "loss": 0.8897, "step": 6064 }, { "epoch": 0.43, "grad_norm": 3.5027883887776285, "learning_rate": 6.282687217309423e-06, "loss": 0.248, "step": 6065 }, { "epoch": 0.43, "grad_norm": 2.1452907319873797, "learning_rate": 6.2815647307640244e-06, "loss": 0.1123, "step": 6066 }, { "epoch": 0.44, "grad_norm": 3.5640411148043145, "learning_rate": 6.280442175083063e-06, "loss": 0.2493, "step": 6067 }, { "epoch": 0.44, "grad_norm": 17.181353666113864, "learning_rate": 6.2793195503270895e-06, "loss": 0.7925, "step": 6068 }, { "epoch": 0.44, "grad_norm": 4.453514282778038, "learning_rate": 6.278196856556672e-06, "loss": 0.2383, "step": 6069 }, { "epoch": 0.44, "grad_norm": 3.481255225590659, "learning_rate": 6.277074093832371e-06, "loss": 0.2056, "step": 6070 }, { "epoch": 0.44, "grad_norm": 4.643621461959352, "learning_rate": 6.275951262214755e-06, "loss": 0.9013, "step": 6071 }, { "epoch": 0.44, "grad_norm": 3.8347463081796946, "learning_rate": 6.274828361764401e-06, "loss": 0.3189, "step": 6072 }, { "epoch": 0.44, "grad_norm": 21.380043523991013, "learning_rate": 6.27370539254188e-06, "loss": 1.1372, "step": 6073 }, { "epoch": 0.44, "grad_norm": 3.8795487249437373, "learning_rate": 6.272582354607775e-06, "loss": 0.1624, "step": 6074 }, { "epoch": 0.44, "grad_norm": 3.542415325107907, "learning_rate": 6.2714592480226696e-06, "loss": 0.203, "step": 6075 }, { "epoch": 0.44, "grad_norm": 4.305916868910422, "learning_rate": 6.270336072847151e-06, "loss": 0.3872, "step": 6076 }, { "epoch": 0.44, "grad_norm": 4.013334849780211, "learning_rate": 6.269212829141808e-06, "loss": 0.2589, "step": 6077 }, { "epoch": 0.44, "grad_norm": 4.975853847028649, "learning_rate": 6.268089516967239e-06, "loss": 1.103, "step": 6078 }, { "epoch": 0.44, "grad_norm": 7.949208110242334, "learning_rate": 6.266966136384038e-06, "loss": 0.4386, "step": 6079 }, { "epoch": 0.44, "grad_norm": 3.0201937972501636, "learning_rate": 6.26584268745281e-06, "loss": 0.2507, "step": 6080 }, { "epoch": 0.44, "grad_norm": 3.805261282375572, "learning_rate": 6.2647191702341625e-06, "loss": 0.8205, "step": 6081 }, { "epoch": 0.44, "grad_norm": 4.7519898372689635, "learning_rate": 6.263595584788703e-06, "loss": 0.8827, "step": 6082 }, { "epoch": 0.44, "grad_norm": 2.4834331167243633, "learning_rate": 6.262471931177043e-06, "loss": 0.1402, "step": 6083 }, { "epoch": 0.44, "grad_norm": 19.212089003790044, "learning_rate": 6.261348209459802e-06, "loss": 0.9942, "step": 6084 }, { "epoch": 0.44, "grad_norm": 4.766461479005219, "learning_rate": 6.260224419697599e-06, "loss": 0.776, "step": 6085 }, { "epoch": 0.44, "grad_norm": 21.289163601025315, "learning_rate": 6.259100561951059e-06, "loss": 1.1522, "step": 6086 }, { "epoch": 0.44, "grad_norm": 4.566714808711187, "learning_rate": 6.2579766362808105e-06, "loss": 0.2522, "step": 6087 }, { "epoch": 0.44, "grad_norm": 3.68768629689287, "learning_rate": 6.256852642747484e-06, "loss": 0.2501, "step": 6088 }, { "epoch": 0.44, "grad_norm": 3.465022347647706, "learning_rate": 6.255728581411713e-06, "loss": 0.1888, "step": 6089 }, { "epoch": 0.44, "grad_norm": 3.682150630214715, "learning_rate": 6.254604452334141e-06, "loss": 0.1422, "step": 6090 }, { "epoch": 0.44, "grad_norm": 5.182392491728919, "learning_rate": 6.253480255575405e-06, "loss": 0.1912, "step": 6091 }, { "epoch": 0.44, "grad_norm": 6.750418371362797, "learning_rate": 6.252355991196155e-06, "loss": 0.8453, "step": 6092 }, { "epoch": 0.44, "grad_norm": 3.9417402253619107, "learning_rate": 6.25123165925704e-06, "loss": 0.3319, "step": 6093 }, { "epoch": 0.44, "grad_norm": 4.214763204211231, "learning_rate": 6.250107259818711e-06, "loss": 0.244, "step": 6094 }, { "epoch": 0.44, "grad_norm": 4.427489919872372, "learning_rate": 6.248982792941832e-06, "loss": 0.7378, "step": 6095 }, { "epoch": 0.44, "grad_norm": 5.355711953739215, "learning_rate": 6.247858258687056e-06, "loss": 1.0014, "step": 6096 }, { "epoch": 0.44, "grad_norm": 3.914781072654251, "learning_rate": 6.246733657115049e-06, "loss": 0.2282, "step": 6097 }, { "epoch": 0.44, "grad_norm": 5.079916048062664, "learning_rate": 6.2456089882864834e-06, "loss": 0.9317, "step": 6098 }, { "epoch": 0.44, "grad_norm": 20.1929835360626, "learning_rate": 6.2444842522620256e-06, "loss": 0.8487, "step": 6099 }, { "epoch": 0.44, "grad_norm": 4.233147927203106, "learning_rate": 6.243359449102353e-06, "loss": 0.2536, "step": 6100 }, { "epoch": 0.44, "grad_norm": 4.4593397339464875, "learning_rate": 6.242234578868147e-06, "loss": 0.7831, "step": 6101 }, { "epoch": 0.44, "grad_norm": 25.429630343749547, "learning_rate": 6.241109641620086e-06, "loss": 1.1986, "step": 6102 }, { "epoch": 0.44, "grad_norm": 5.347195679452816, "learning_rate": 6.2399846374188566e-06, "loss": 0.9999, "step": 6103 }, { "epoch": 0.44, "grad_norm": 23.43594667363551, "learning_rate": 6.23885956632515e-06, "loss": 1.0088, "step": 6104 }, { "epoch": 0.44, "grad_norm": 8.87755958656619, "learning_rate": 6.23773442839966e-06, "loss": 0.4492, "step": 6105 }, { "epoch": 0.44, "grad_norm": 2.8478738555141008, "learning_rate": 6.236609223703083e-06, "loss": 0.2011, "step": 6106 }, { "epoch": 0.44, "grad_norm": 8.875282573290342, "learning_rate": 6.2354839522961186e-06, "loss": 0.5877, "step": 6107 }, { "epoch": 0.44, "grad_norm": 15.623769005986302, "learning_rate": 6.234358614239471e-06, "loss": 0.8483, "step": 6108 }, { "epoch": 0.44, "grad_norm": 3.6390430663688424, "learning_rate": 6.233233209593848e-06, "loss": 0.2376, "step": 6109 }, { "epoch": 0.44, "grad_norm": 22.795236144431446, "learning_rate": 6.232107738419962e-06, "loss": 1.0284, "step": 6110 }, { "epoch": 0.44, "grad_norm": 3.863040628524653, "learning_rate": 6.230982200778528e-06, "loss": 0.2013, "step": 6111 }, { "epoch": 0.44, "grad_norm": 4.636123766429715, "learning_rate": 6.2298565967302635e-06, "loss": 0.2012, "step": 6112 }, { "epoch": 0.44, "grad_norm": 4.008009856625868, "learning_rate": 6.228730926335891e-06, "loss": 0.1764, "step": 6113 }, { "epoch": 0.44, "grad_norm": 4.622280824315121, "learning_rate": 6.227605189656135e-06, "loss": 0.2001, "step": 6114 }, { "epoch": 0.44, "grad_norm": 19.802954394516167, "learning_rate": 6.226479386751727e-06, "loss": 1.0715, "step": 6115 }, { "epoch": 0.44, "grad_norm": 5.379879513239476, "learning_rate": 6.225353517683398e-06, "loss": 1.012, "step": 6116 }, { "epoch": 0.44, "grad_norm": 3.6585707362777478, "learning_rate": 6.224227582511884e-06, "loss": 0.209, "step": 6117 }, { "epoch": 0.44, "grad_norm": 4.309443208456638, "learning_rate": 6.223101581297928e-06, "loss": 0.2056, "step": 6118 }, { "epoch": 0.44, "grad_norm": 4.882743358477704, "learning_rate": 6.22197551410227e-06, "loss": 0.8021, "step": 6119 }, { "epoch": 0.44, "grad_norm": 19.246827975786232, "learning_rate": 6.220849380985658e-06, "loss": 0.6989, "step": 6120 }, { "epoch": 0.44, "grad_norm": 14.656089277177482, "learning_rate": 6.219723182008844e-06, "loss": 1.0432, "step": 6121 }, { "epoch": 0.44, "grad_norm": 24.928827074622404, "learning_rate": 6.2185969172325815e-06, "loss": 0.8358, "step": 6122 }, { "epoch": 0.44, "grad_norm": 3.8546220701512466, "learning_rate": 6.2174705867176275e-06, "loss": 0.1705, "step": 6123 }, { "epoch": 0.44, "grad_norm": 4.2053104696906916, "learning_rate": 6.216344190524743e-06, "loss": 0.2431, "step": 6124 }, { "epoch": 0.44, "grad_norm": 4.129865284870898, "learning_rate": 6.215217728714695e-06, "loss": 0.1827, "step": 6125 }, { "epoch": 0.44, "grad_norm": 3.505715354714149, "learning_rate": 6.214091201348249e-06, "loss": 0.1356, "step": 6126 }, { "epoch": 0.44, "grad_norm": 4.666825007590577, "learning_rate": 6.212964608486179e-06, "loss": 0.8141, "step": 6127 }, { "epoch": 0.44, "grad_norm": 10.339283297930201, "learning_rate": 6.211837950189259e-06, "loss": 0.51, "step": 6128 }, { "epoch": 0.44, "grad_norm": 49.937258958243824, "learning_rate": 6.21071122651827e-06, "loss": 1.1155, "step": 6129 }, { "epoch": 0.44, "grad_norm": 3.8538724183880664, "learning_rate": 6.209584437533992e-06, "loss": 0.1907, "step": 6130 }, { "epoch": 0.44, "grad_norm": 31.222387472402975, "learning_rate": 6.208457583297213e-06, "loss": 0.8735, "step": 6131 }, { "epoch": 0.44, "grad_norm": 4.737149233480443, "learning_rate": 6.207330663868721e-06, "loss": 0.7611, "step": 6132 }, { "epoch": 0.44, "grad_norm": 17.24012163494007, "learning_rate": 6.20620367930931e-06, "loss": 0.9073, "step": 6133 }, { "epoch": 0.44, "grad_norm": 11.230465163659108, "learning_rate": 6.205076629679775e-06, "loss": 0.4901, "step": 6134 }, { "epoch": 0.44, "grad_norm": 16.40147205642431, "learning_rate": 6.203949515040919e-06, "loss": 0.9364, "step": 6135 }, { "epoch": 0.44, "grad_norm": 11.294126340366804, "learning_rate": 6.2028223354535425e-06, "loss": 0.9124, "step": 6136 }, { "epoch": 0.44, "grad_norm": 12.01891738326737, "learning_rate": 6.201695090978454e-06, "loss": 0.6958, "step": 6137 }, { "epoch": 0.44, "grad_norm": 3.231965905347024, "learning_rate": 6.200567781676465e-06, "loss": 0.2726, "step": 6138 }, { "epoch": 0.44, "grad_norm": 22.648248198241397, "learning_rate": 6.199440407608388e-06, "loss": 1.4365, "step": 6139 }, { "epoch": 0.44, "grad_norm": 3.831726446531, "learning_rate": 6.198312968835041e-06, "loss": 0.2066, "step": 6140 }, { "epoch": 0.44, "grad_norm": 13.560163882617857, "learning_rate": 6.197185465417246e-06, "loss": 1.0358, "step": 6141 }, { "epoch": 0.44, "grad_norm": 4.0396514493840066, "learning_rate": 6.196057897415826e-06, "loss": 0.8833, "step": 6142 }, { "epoch": 0.44, "grad_norm": 3.2817863421387896, "learning_rate": 6.19493026489161e-06, "loss": 0.2027, "step": 6143 }, { "epoch": 0.44, "grad_norm": 3.7226581932081526, "learning_rate": 6.1938025679054295e-06, "loss": 0.1982, "step": 6144 }, { "epoch": 0.44, "grad_norm": 3.5381069783233365, "learning_rate": 6.1926748065181185e-06, "loss": 0.2683, "step": 6145 }, { "epoch": 0.44, "grad_norm": 13.28578228092013, "learning_rate": 6.191546980790517e-06, "loss": 0.8864, "step": 6146 }, { "epoch": 0.44, "grad_norm": 2.6018040425775273, "learning_rate": 6.190419090783466e-06, "loss": 0.1228, "step": 6147 }, { "epoch": 0.44, "grad_norm": 4.667601604360035, "learning_rate": 6.18929113655781e-06, "loss": 0.8123, "step": 6148 }, { "epoch": 0.44, "grad_norm": 13.948824810918621, "learning_rate": 6.188163118174402e-06, "loss": 1.1893, "step": 6149 }, { "epoch": 0.44, "grad_norm": 11.496003210790336, "learning_rate": 6.187035035694088e-06, "loss": 1.0721, "step": 6150 }, { "epoch": 0.44, "grad_norm": 17.19995389150501, "learning_rate": 6.185906889177727e-06, "loss": 1.8551, "step": 6151 }, { "epoch": 0.44, "grad_norm": 4.1606924463867925, "learning_rate": 6.184778678686179e-06, "loss": 0.2242, "step": 6152 }, { "epoch": 0.44, "grad_norm": 4.668936644220223, "learning_rate": 6.183650404280306e-06, "loss": 0.2749, "step": 6153 }, { "epoch": 0.44, "grad_norm": 16.245331296922423, "learning_rate": 6.182522066020974e-06, "loss": 0.7167, "step": 6154 }, { "epoch": 0.44, "grad_norm": 8.596464735717793, "learning_rate": 6.181393663969054e-06, "loss": 0.5097, "step": 6155 }, { "epoch": 0.44, "grad_norm": 11.441510509733735, "learning_rate": 6.180265198185416e-06, "loss": 0.8021, "step": 6156 }, { "epoch": 0.44, "grad_norm": 16.095120896534148, "learning_rate": 6.179136668730938e-06, "loss": 1.2294, "step": 6157 }, { "epoch": 0.44, "grad_norm": 8.614701385884336, "learning_rate": 6.178008075666502e-06, "loss": 0.5818, "step": 6158 }, { "epoch": 0.44, "grad_norm": 13.09043774533819, "learning_rate": 6.176879419052987e-06, "loss": 1.1225, "step": 6159 }, { "epoch": 0.44, "grad_norm": 3.9824000174166643, "learning_rate": 6.175750698951283e-06, "loss": 0.2061, "step": 6160 }, { "epoch": 0.44, "grad_norm": 3.8297118356079216, "learning_rate": 6.17462191542228e-06, "loss": 0.2796, "step": 6161 }, { "epoch": 0.44, "grad_norm": 2.821376282626193, "learning_rate": 6.173493068526869e-06, "loss": 0.1748, "step": 6162 }, { "epoch": 0.44, "grad_norm": 3.275202575089244, "learning_rate": 6.172364158325951e-06, "loss": 0.152, "step": 6163 }, { "epoch": 0.44, "grad_norm": 2.6349044016082717, "learning_rate": 6.171235184880422e-06, "loss": 0.1515, "step": 6164 }, { "epoch": 0.44, "grad_norm": 2.987038438882448, "learning_rate": 6.17010614825119e-06, "loss": 0.2012, "step": 6165 }, { "epoch": 0.44, "grad_norm": 3.976236559346294, "learning_rate": 6.1689770484991585e-06, "loss": 0.7845, "step": 6166 }, { "epoch": 0.44, "grad_norm": 17.24688727547328, "learning_rate": 6.167847885685242e-06, "loss": 1.1223, "step": 6167 }, { "epoch": 0.44, "grad_norm": 4.4051373142234285, "learning_rate": 6.166718659870352e-06, "loss": 0.8908, "step": 6168 }, { "epoch": 0.44, "grad_norm": 4.257556320729866, "learning_rate": 6.165589371115407e-06, "loss": 0.1622, "step": 6169 }, { "epoch": 0.44, "grad_norm": 3.2233745186169505, "learning_rate": 6.164460019481326e-06, "loss": 0.1325, "step": 6170 }, { "epoch": 0.44, "grad_norm": 11.175621757124084, "learning_rate": 6.1633306050290345e-06, "loss": 1.1419, "step": 6171 }, { "epoch": 0.44, "grad_norm": 4.118284284783964, "learning_rate": 6.1622011278194625e-06, "loss": 0.9312, "step": 6172 }, { "epoch": 0.44, "grad_norm": 4.140816873790416, "learning_rate": 6.1610715879135375e-06, "loss": 0.2345, "step": 6173 }, { "epoch": 0.44, "grad_norm": 16.18774753552178, "learning_rate": 6.159941985372195e-06, "loss": 1.0, "step": 6174 }, { "epoch": 0.44, "grad_norm": 3.353430713630056, "learning_rate": 6.158812320256374e-06, "loss": 0.1881, "step": 6175 }, { "epoch": 0.44, "grad_norm": 13.301082970398909, "learning_rate": 6.1576825926270135e-06, "loss": 0.908, "step": 6176 }, { "epoch": 0.44, "grad_norm": 5.669939768343096, "learning_rate": 6.156552802545061e-06, "loss": 0.9941, "step": 6177 }, { "epoch": 0.44, "grad_norm": 11.749509421281212, "learning_rate": 6.15542295007146e-06, "loss": 0.7087, "step": 6178 }, { "epoch": 0.44, "grad_norm": 4.09983364798862, "learning_rate": 6.154293035267167e-06, "loss": 0.7324, "step": 6179 }, { "epoch": 0.44, "grad_norm": 2.525396922785556, "learning_rate": 6.153163058193131e-06, "loss": 0.1417, "step": 6180 }, { "epoch": 0.44, "grad_norm": 3.305400886917632, "learning_rate": 6.152033018910316e-06, "loss": 0.2454, "step": 6181 }, { "epoch": 0.44, "grad_norm": 4.300285984462234, "learning_rate": 6.150902917479678e-06, "loss": 1.058, "step": 6182 }, { "epoch": 0.44, "grad_norm": 3.8184877908274286, "learning_rate": 6.149772753962185e-06, "loss": 0.2722, "step": 6183 }, { "epoch": 0.44, "grad_norm": 3.000345539833419, "learning_rate": 6.148642528418804e-06, "loss": 0.1481, "step": 6184 }, { "epoch": 0.44, "grad_norm": 11.049312441700186, "learning_rate": 6.147512240910507e-06, "loss": 0.8615, "step": 6185 }, { "epoch": 0.44, "grad_norm": 4.791889194349631, "learning_rate": 6.146381891498269e-06, "loss": 0.9288, "step": 6186 }, { "epoch": 0.44, "grad_norm": 3.5909355241696304, "learning_rate": 6.145251480243064e-06, "loss": 0.2569, "step": 6187 }, { "epoch": 0.44, "grad_norm": 7.512905999809122, "learning_rate": 6.144121007205879e-06, "loss": 0.5451, "step": 6188 }, { "epoch": 0.44, "grad_norm": 4.665696933498306, "learning_rate": 6.142990472447696e-06, "loss": 0.7431, "step": 6189 }, { "epoch": 0.44, "grad_norm": 3.1411959466316306, "learning_rate": 6.141859876029503e-06, "loss": 0.1473, "step": 6190 }, { "epoch": 0.44, "grad_norm": 3.105149550572731, "learning_rate": 6.140729218012292e-06, "loss": 0.1375, "step": 6191 }, { "epoch": 0.44, "grad_norm": 3.2338209664352933, "learning_rate": 6.139598498457059e-06, "loss": 0.1419, "step": 6192 }, { "epoch": 0.44, "grad_norm": 2.7541711819151904, "learning_rate": 6.138467717424798e-06, "loss": 0.2131, "step": 6193 }, { "epoch": 0.44, "grad_norm": 4.1472969400330255, "learning_rate": 6.137336874976515e-06, "loss": 0.227, "step": 6194 }, { "epoch": 0.44, "grad_norm": 3.5094413729144556, "learning_rate": 6.136205971173213e-06, "loss": 0.2832, "step": 6195 }, { "epoch": 0.44, "grad_norm": 11.982650271751737, "learning_rate": 6.135075006075898e-06, "loss": 0.7395, "step": 6196 }, { "epoch": 0.44, "grad_norm": 17.59764084159393, "learning_rate": 6.133943979745583e-06, "loss": 0.9179, "step": 6197 }, { "epoch": 0.44, "grad_norm": 3.634564569993271, "learning_rate": 6.132812892243285e-06, "loss": 0.1827, "step": 6198 }, { "epoch": 0.44, "grad_norm": 23.726820967617254, "learning_rate": 6.131681743630017e-06, "loss": 0.9764, "step": 6199 }, { "epoch": 0.44, "grad_norm": 3.7058057713622183, "learning_rate": 6.130550533966804e-06, "loss": 0.325, "step": 6200 }, { "epoch": 0.44, "grad_norm": 4.443363619436298, "learning_rate": 6.129419263314666e-06, "loss": 0.7785, "step": 6201 }, { "epoch": 0.44, "grad_norm": 17.036440634720723, "learning_rate": 6.128287931734636e-06, "loss": 0.6506, "step": 6202 }, { "epoch": 0.44, "grad_norm": 5.709591162890385, "learning_rate": 6.127156539287744e-06, "loss": 0.8823, "step": 6203 }, { "epoch": 0.44, "grad_norm": 4.417261255361114, "learning_rate": 6.126025086035022e-06, "loss": 0.8516, "step": 6204 }, { "epoch": 0.44, "grad_norm": 4.6424773775413755, "learning_rate": 6.124893572037509e-06, "loss": 0.9333, "step": 6205 }, { "epoch": 0.45, "grad_norm": 3.437192372455296, "learning_rate": 6.123761997356246e-06, "loss": 0.1696, "step": 6206 }, { "epoch": 0.45, "grad_norm": 11.745968187317736, "learning_rate": 6.122630362052276e-06, "loss": 0.6255, "step": 6207 }, { "epoch": 0.45, "grad_norm": 15.456792658656369, "learning_rate": 6.121498666186647e-06, "loss": 1.0279, "step": 6208 }, { "epoch": 0.45, "grad_norm": 4.325623542597812, "learning_rate": 6.120366909820413e-06, "loss": 0.9955, "step": 6209 }, { "epoch": 0.45, "grad_norm": 11.076378384838053, "learning_rate": 6.119235093014623e-06, "loss": 0.8283, "step": 6210 }, { "epoch": 0.45, "grad_norm": 18.859905519051733, "learning_rate": 6.118103215830336e-06, "loss": 0.7872, "step": 6211 }, { "epoch": 0.45, "grad_norm": 4.471262495705808, "learning_rate": 6.116971278328613e-06, "loss": 0.2426, "step": 6212 }, { "epoch": 0.45, "grad_norm": 3.6010889421634973, "learning_rate": 6.115839280570517e-06, "loss": 0.2166, "step": 6213 }, { "epoch": 0.45, "grad_norm": 4.418239215717993, "learning_rate": 6.114707222617117e-06, "loss": 0.9715, "step": 6214 }, { "epoch": 0.45, "grad_norm": 3.3842148799843743, "learning_rate": 6.113575104529481e-06, "loss": 0.21, "step": 6215 }, { "epoch": 0.45, "grad_norm": 6.3637464035720095, "learning_rate": 6.112442926368682e-06, "loss": 0.5634, "step": 6216 }, { "epoch": 0.45, "grad_norm": 5.47969010210108, "learning_rate": 6.111310688195799e-06, "loss": 1.0853, "step": 6217 }, { "epoch": 0.45, "grad_norm": 4.5878461448920165, "learning_rate": 6.1101783900719115e-06, "loss": 0.9625, "step": 6218 }, { "epoch": 0.45, "grad_norm": 13.183696990375779, "learning_rate": 6.1090460320581e-06, "loss": 0.671, "step": 6219 }, { "epoch": 0.45, "grad_norm": 16.219058342912007, "learning_rate": 6.107913614215455e-06, "loss": 0.7412, "step": 6220 }, { "epoch": 0.45, "grad_norm": 18.54533728981783, "learning_rate": 6.106781136605064e-06, "loss": 1.3718, "step": 6221 }, { "epoch": 0.45, "grad_norm": 14.409012931673546, "learning_rate": 6.10564859928802e-06, "loss": 0.7416, "step": 6222 }, { "epoch": 0.45, "grad_norm": 6.105714845951304, "learning_rate": 6.104516002325419e-06, "loss": 0.5635, "step": 6223 }, { "epoch": 0.45, "grad_norm": 4.5021587508662, "learning_rate": 6.1033833457783605e-06, "loss": 1.2084, "step": 6224 }, { "epoch": 0.45, "grad_norm": 6.747824635794003, "learning_rate": 6.102250629707946e-06, "loss": 0.4454, "step": 6225 }, { "epoch": 0.45, "grad_norm": 1.7780022836225073, "learning_rate": 6.101117854175285e-06, "loss": 0.0983, "step": 6226 }, { "epoch": 0.45, "grad_norm": 20.016097355019564, "learning_rate": 6.0999850192414815e-06, "loss": 1.0819, "step": 6227 }, { "epoch": 0.45, "grad_norm": 3.2125034648579485, "learning_rate": 6.09885212496765e-06, "loss": 0.1314, "step": 6228 }, { "epoch": 0.45, "grad_norm": 21.228275884627674, "learning_rate": 6.097719171414907e-06, "loss": 0.9569, "step": 6229 }, { "epoch": 0.45, "grad_norm": 5.808502647215503, "learning_rate": 6.096586158644368e-06, "loss": 0.8496, "step": 6230 }, { "epoch": 0.45, "grad_norm": 4.436831374106441, "learning_rate": 6.095453086717158e-06, "loss": 0.4049, "step": 6231 }, { "epoch": 0.45, "grad_norm": 4.717737840205864, "learning_rate": 6.094319955694399e-06, "loss": 0.2963, "step": 6232 }, { "epoch": 0.45, "grad_norm": 4.615146327192027, "learning_rate": 6.093186765637223e-06, "loss": 0.2427, "step": 6233 }, { "epoch": 0.45, "grad_norm": 3.8775647646423996, "learning_rate": 6.092053516606756e-06, "loss": 0.23, "step": 6234 }, { "epoch": 0.45, "grad_norm": 5.009600554915983, "learning_rate": 6.090920208664137e-06, "loss": 0.893, "step": 6235 }, { "epoch": 0.45, "grad_norm": 4.7697399900163235, "learning_rate": 6.089786841870502e-06, "loss": 0.3833, "step": 6236 }, { "epoch": 0.45, "grad_norm": 4.288409872267595, "learning_rate": 6.088653416286993e-06, "loss": 0.2918, "step": 6237 }, { "epoch": 0.45, "grad_norm": 35.469135532030634, "learning_rate": 6.087519931974751e-06, "loss": 0.7081, "step": 6238 }, { "epoch": 0.45, "grad_norm": 4.321293957966322, "learning_rate": 6.0863863889949275e-06, "loss": 0.2257, "step": 6239 }, { "epoch": 0.45, "grad_norm": 2.943279983915919, "learning_rate": 6.085252787408669e-06, "loss": 0.143, "step": 6240 }, { "epoch": 0.45, "grad_norm": 41.196440927932244, "learning_rate": 6.084119127277131e-06, "loss": 0.8278, "step": 6241 }, { "epoch": 0.45, "grad_norm": 3.181488782736408, "learning_rate": 6.082985408661468e-06, "loss": 0.1701, "step": 6242 }, { "epoch": 0.45, "grad_norm": 2.9199984173019455, "learning_rate": 6.0818516316228425e-06, "loss": 0.1847, "step": 6243 }, { "epoch": 0.45, "grad_norm": 17.836125070411402, "learning_rate": 6.080717796222417e-06, "loss": 0.9987, "step": 6244 }, { "epoch": 0.45, "grad_norm": 18.393396566263306, "learning_rate": 6.079583902521358e-06, "loss": 0.6943, "step": 6245 }, { "epoch": 0.45, "grad_norm": 11.57480932025939, "learning_rate": 6.078449950580833e-06, "loss": 0.6774, "step": 6246 }, { "epoch": 0.45, "grad_norm": 5.176515689858018, "learning_rate": 6.077315940462014e-06, "loss": 1.0004, "step": 6247 }, { "epoch": 0.45, "grad_norm": 4.2630447421701385, "learning_rate": 6.076181872226079e-06, "loss": 0.7134, "step": 6248 }, { "epoch": 0.45, "grad_norm": 3.6716359188429823, "learning_rate": 6.075047745934206e-06, "loss": 0.2659, "step": 6249 }, { "epoch": 0.45, "grad_norm": 13.252558986749362, "learning_rate": 6.073913561647575e-06, "loss": 0.8161, "step": 6250 }, { "epoch": 0.45, "grad_norm": 3.150887369832222, "learning_rate": 6.0727793194273744e-06, "loss": 0.2084, "step": 6251 }, { "epoch": 0.45, "grad_norm": 11.843877009521902, "learning_rate": 6.07164501933479e-06, "loss": 0.8448, "step": 6252 }, { "epoch": 0.45, "grad_norm": 4.132818400173673, "learning_rate": 6.070510661431012e-06, "loss": 0.8224, "step": 6253 }, { "epoch": 0.45, "grad_norm": 3.293553392948413, "learning_rate": 6.0693762457772366e-06, "loss": 0.1634, "step": 6254 }, { "epoch": 0.45, "grad_norm": 3.402921310210102, "learning_rate": 6.06824177243466e-06, "loss": 0.1843, "step": 6255 }, { "epoch": 0.45, "grad_norm": 4.405322296281176, "learning_rate": 6.067107241464483e-06, "loss": 0.9679, "step": 6256 }, { "epoch": 0.45, "grad_norm": 3.5440978232778138, "learning_rate": 6.065972652927911e-06, "loss": 0.2242, "step": 6257 }, { "epoch": 0.45, "grad_norm": 2.824346499796226, "learning_rate": 6.064838006886148e-06, "loss": 0.1145, "step": 6258 }, { "epoch": 0.45, "grad_norm": 3.0835322304247663, "learning_rate": 6.063703303400406e-06, "loss": 0.1507, "step": 6259 }, { "epoch": 0.45, "grad_norm": 5.790608366548334, "learning_rate": 6.062568542531896e-06, "loss": 0.8134, "step": 6260 }, { "epoch": 0.45, "grad_norm": 3.4150417122010954, "learning_rate": 6.0614337243418354e-06, "loss": 0.1932, "step": 6261 }, { "epoch": 0.45, "grad_norm": 5.828587907924215, "learning_rate": 6.060298848891442e-06, "loss": 0.6839, "step": 6262 }, { "epoch": 0.45, "grad_norm": 23.0580561526969, "learning_rate": 6.0591639162419434e-06, "loss": 1.3426, "step": 6263 }, { "epoch": 0.45, "grad_norm": 3.8329098839822398, "learning_rate": 6.058028926454558e-06, "loss": 0.1505, "step": 6264 }, { "epoch": 0.45, "grad_norm": 13.956747513562211, "learning_rate": 6.056893879590515e-06, "loss": 1.2027, "step": 6265 }, { "epoch": 0.45, "grad_norm": 13.624693434253551, "learning_rate": 6.055758775711052e-06, "loss": 1.2313, "step": 6266 }, { "epoch": 0.45, "grad_norm": 3.2542036529491707, "learning_rate": 6.054623614877395e-06, "loss": 0.2147, "step": 6267 }, { "epoch": 0.45, "grad_norm": 50.59041069535959, "learning_rate": 6.053488397150789e-06, "loss": 0.955, "step": 6268 }, { "epoch": 0.45, "grad_norm": 2.9815182781151846, "learning_rate": 6.052353122592471e-06, "loss": 0.1758, "step": 6269 }, { "epoch": 0.45, "grad_norm": 4.19615160240557, "learning_rate": 6.0512177912636835e-06, "loss": 0.427, "step": 6270 }, { "epoch": 0.45, "grad_norm": 3.3028228671820097, "learning_rate": 6.050082403225678e-06, "loss": 0.167, "step": 6271 }, { "epoch": 0.45, "grad_norm": 3.544126729697723, "learning_rate": 6.048946958539699e-06, "loss": 0.1601, "step": 6272 }, { "epoch": 0.45, "grad_norm": 3.777174037768067, "learning_rate": 6.047811457267003e-06, "loss": 0.1387, "step": 6273 }, { "epoch": 0.45, "grad_norm": 18.138714610660507, "learning_rate": 6.0466758994688454e-06, "loss": 0.8467, "step": 6274 }, { "epoch": 0.45, "grad_norm": 5.645256458258533, "learning_rate": 6.045540285206486e-06, "loss": 0.3268, "step": 6275 }, { "epoch": 0.45, "grad_norm": 3.3478652444708374, "learning_rate": 6.044404614541184e-06, "loss": 0.1739, "step": 6276 }, { "epoch": 0.45, "grad_norm": 19.586431926962685, "learning_rate": 6.043268887534208e-06, "loss": 0.6081, "step": 6277 }, { "epoch": 0.45, "grad_norm": 14.14812946586158, "learning_rate": 6.042133104246822e-06, "loss": 0.9061, "step": 6278 }, { "epoch": 0.45, "grad_norm": 17.36421412858388, "learning_rate": 6.040997264740301e-06, "loss": 1.2616, "step": 6279 }, { "epoch": 0.45, "grad_norm": 3.7162349597173114, "learning_rate": 6.039861369075919e-06, "loss": 0.2061, "step": 6280 }, { "epoch": 0.45, "grad_norm": 10.169701385691594, "learning_rate": 6.038725417314952e-06, "loss": 0.9199, "step": 6281 }, { "epoch": 0.45, "grad_norm": 3.3945382011898286, "learning_rate": 6.037589409518679e-06, "loss": 0.177, "step": 6282 }, { "epoch": 0.45, "grad_norm": 19.88695901145186, "learning_rate": 6.036453345748387e-06, "loss": 0.8918, "step": 6283 }, { "epoch": 0.45, "grad_norm": 3.26521348578076, "learning_rate": 6.035317226065358e-06, "loss": 0.1469, "step": 6284 }, { "epoch": 0.45, "grad_norm": 3.7689796586279734, "learning_rate": 6.034181050530885e-06, "loss": 0.1635, "step": 6285 }, { "epoch": 0.45, "grad_norm": 16.363721246718768, "learning_rate": 6.033044819206258e-06, "loss": 0.8181, "step": 6286 }, { "epoch": 0.45, "grad_norm": 3.879487293996061, "learning_rate": 6.031908532152775e-06, "loss": 0.2223, "step": 6287 }, { "epoch": 0.45, "grad_norm": 3.53689527565998, "learning_rate": 6.030772189431731e-06, "loss": 0.1487, "step": 6288 }, { "epoch": 0.45, "grad_norm": 2.8753276815327915, "learning_rate": 6.029635791104429e-06, "loss": 0.1397, "step": 6289 }, { "epoch": 0.45, "grad_norm": 4.0532372883613075, "learning_rate": 6.028499337232175e-06, "loss": 0.3509, "step": 6290 }, { "epoch": 0.45, "grad_norm": 3.2010637656223992, "learning_rate": 6.027362827876273e-06, "loss": 0.166, "step": 6291 }, { "epoch": 0.45, "grad_norm": 2.851253524347207, "learning_rate": 6.026226263098035e-06, "loss": 0.1385, "step": 6292 }, { "epoch": 0.45, "grad_norm": 16.748322109371635, "learning_rate": 6.0250896429587765e-06, "loss": 1.2298, "step": 6293 }, { "epoch": 0.45, "grad_norm": 8.062837929958256, "learning_rate": 6.023952967519812e-06, "loss": 0.7983, "step": 6294 }, { "epoch": 0.45, "grad_norm": 2.630460214343439, "learning_rate": 6.022816236842459e-06, "loss": 0.1748, "step": 6295 }, { "epoch": 0.45, "grad_norm": 2.852295378475987, "learning_rate": 6.021679450988042e-06, "loss": 0.1894, "step": 6296 }, { "epoch": 0.45, "grad_norm": 17.124626590913692, "learning_rate": 6.020542610017886e-06, "loss": 1.1559, "step": 6297 }, { "epoch": 0.45, "grad_norm": 3.170358177238514, "learning_rate": 6.019405713993319e-06, "loss": 0.1325, "step": 6298 }, { "epoch": 0.45, "grad_norm": 3.591614519753585, "learning_rate": 6.018268762975674e-06, "loss": 0.1894, "step": 6299 }, { "epoch": 0.45, "grad_norm": 4.92014898310044, "learning_rate": 6.017131757026282e-06, "loss": 0.807, "step": 6300 }, { "epoch": 0.45, "grad_norm": 9.927741994837712, "learning_rate": 6.015994696206482e-06, "loss": 0.6074, "step": 6301 }, { "epoch": 0.45, "grad_norm": 3.459514975539732, "learning_rate": 6.0148575805776134e-06, "loss": 0.2869, "step": 6302 }, { "epoch": 0.45, "grad_norm": 7.151009812054277, "learning_rate": 6.013720410201022e-06, "loss": 0.5619, "step": 6303 }, { "epoch": 0.45, "grad_norm": 3.313294573636142, "learning_rate": 6.012583185138049e-06, "loss": 0.226, "step": 6304 }, { "epoch": 0.45, "grad_norm": 29.190595960175266, "learning_rate": 6.011445905450047e-06, "loss": 0.9826, "step": 6305 }, { "epoch": 0.45, "grad_norm": 3.435309447248186, "learning_rate": 6.010308571198368e-06, "loss": 0.1553, "step": 6306 }, { "epoch": 0.45, "grad_norm": 12.276202563449173, "learning_rate": 6.009171182444364e-06, "loss": 1.1294, "step": 6307 }, { "epoch": 0.45, "grad_norm": 3.808570165452074, "learning_rate": 6.008033739249396e-06, "loss": 0.1389, "step": 6308 }, { "epoch": 0.45, "grad_norm": 3.9774704320820327, "learning_rate": 6.006896241674822e-06, "loss": 0.2533, "step": 6309 }, { "epoch": 0.45, "grad_norm": 3.924561118305422, "learning_rate": 6.005758689782008e-06, "loss": 0.24, "step": 6310 }, { "epoch": 0.45, "grad_norm": 3.4230844132128584, "learning_rate": 6.004621083632319e-06, "loss": 0.2071, "step": 6311 }, { "epoch": 0.45, "grad_norm": 7.588843687741181, "learning_rate": 6.003483423287126e-06, "loss": 0.5389, "step": 6312 }, { "epoch": 0.45, "grad_norm": 3.1561832108416383, "learning_rate": 6.0023457088078e-06, "loss": 0.2137, "step": 6313 }, { "epoch": 0.45, "grad_norm": 3.935236319802527, "learning_rate": 6.001207940255719e-06, "loss": 0.9371, "step": 6314 }, { "epoch": 0.45, "grad_norm": 10.596257016194482, "learning_rate": 6.0000701176922574e-06, "loss": 0.679, "step": 6315 }, { "epoch": 0.45, "grad_norm": 3.345256062731936, "learning_rate": 5.998932241178798e-06, "loss": 0.2752, "step": 6316 }, { "epoch": 0.45, "grad_norm": 9.073949785652639, "learning_rate": 5.997794310776727e-06, "loss": 0.5264, "step": 6317 }, { "epoch": 0.45, "grad_norm": 9.388309899932512, "learning_rate": 5.99665632654743e-06, "loss": 0.5897, "step": 6318 }, { "epoch": 0.45, "grad_norm": 3.292720799289365, "learning_rate": 5.995518288552294e-06, "loss": 0.153, "step": 6319 }, { "epoch": 0.45, "grad_norm": 3.095363642594842, "learning_rate": 5.9943801968527185e-06, "loss": 0.15, "step": 6320 }, { "epoch": 0.45, "grad_norm": 13.371869169433882, "learning_rate": 5.993242051510092e-06, "loss": 1.0525, "step": 6321 }, { "epoch": 0.45, "grad_norm": 10.229954916242505, "learning_rate": 5.992103852585818e-06, "loss": 0.7475, "step": 6322 }, { "epoch": 0.45, "grad_norm": 5.066211639963516, "learning_rate": 5.990965600141297e-06, "loss": 0.8221, "step": 6323 }, { "epoch": 0.45, "grad_norm": 3.761539568580388, "learning_rate": 5.989827294237931e-06, "loss": 0.2564, "step": 6324 }, { "epoch": 0.45, "grad_norm": 2.9600645219771837, "learning_rate": 5.98868893493713e-06, "loss": 0.1746, "step": 6325 }, { "epoch": 0.45, "grad_norm": 3.9395128855724066, "learning_rate": 5.987550522300304e-06, "loss": 0.2364, "step": 6326 }, { "epoch": 0.45, "grad_norm": 3.5524062188840997, "learning_rate": 5.986412056388865e-06, "loss": 0.1901, "step": 6327 }, { "epoch": 0.45, "grad_norm": 5.175571509369455, "learning_rate": 5.985273537264229e-06, "loss": 0.8101, "step": 6328 }, { "epoch": 0.45, "grad_norm": 3.3540771200199804, "learning_rate": 5.984134964987815e-06, "loss": 0.2154, "step": 6329 }, { "epoch": 0.45, "grad_norm": 3.406534691471873, "learning_rate": 5.982996339621047e-06, "loss": 0.2385, "step": 6330 }, { "epoch": 0.45, "grad_norm": 14.131251251906908, "learning_rate": 5.981857661225345e-06, "loss": 1.2713, "step": 6331 }, { "epoch": 0.45, "grad_norm": 3.2147555960996947, "learning_rate": 5.9807189298621395e-06, "loss": 0.1459, "step": 6332 }, { "epoch": 0.45, "grad_norm": 3.30162151527761, "learning_rate": 5.9795801455928584e-06, "loss": 0.1462, "step": 6333 }, { "epoch": 0.45, "grad_norm": 20.77123079984478, "learning_rate": 5.978441308478938e-06, "loss": 1.3045, "step": 6334 }, { "epoch": 0.45, "grad_norm": 8.545905251259923, "learning_rate": 5.977302418581812e-06, "loss": 0.5355, "step": 6335 }, { "epoch": 0.45, "grad_norm": 30.25762715784413, "learning_rate": 5.976163475962921e-06, "loss": 1.3682, "step": 6336 }, { "epoch": 0.45, "grad_norm": 4.238038639361727, "learning_rate": 5.975024480683704e-06, "loss": 0.2649, "step": 6337 }, { "epoch": 0.45, "grad_norm": 6.0625314211369, "learning_rate": 5.973885432805607e-06, "loss": 0.9254, "step": 6338 }, { "epoch": 0.45, "grad_norm": 3.9058799502357506, "learning_rate": 5.972746332390078e-06, "loss": 0.1906, "step": 6339 }, { "epoch": 0.45, "grad_norm": 3.6978842794890556, "learning_rate": 5.971607179498567e-06, "loss": 0.2223, "step": 6340 }, { "epoch": 0.45, "grad_norm": 12.489887719490733, "learning_rate": 5.970467974192525e-06, "loss": 0.4948, "step": 6341 }, { "epoch": 0.45, "grad_norm": 3.8832612964407596, "learning_rate": 5.96932871653341e-06, "loss": 0.3699, "step": 6342 }, { "epoch": 0.45, "grad_norm": 3.967533072333186, "learning_rate": 5.96818940658268e-06, "loss": 0.8847, "step": 6343 }, { "epoch": 0.45, "grad_norm": 21.65030748395819, "learning_rate": 5.967050044401796e-06, "loss": 1.5182, "step": 6344 }, { "epoch": 0.45, "grad_norm": 4.105482901237051, "learning_rate": 5.965910630052223e-06, "loss": 0.2918, "step": 6345 }, { "epoch": 0.46, "grad_norm": 5.098729912021252, "learning_rate": 5.964771163595427e-06, "loss": 0.9593, "step": 6346 }, { "epoch": 0.46, "grad_norm": 3.824517267620782, "learning_rate": 5.9636316450928775e-06, "loss": 0.2253, "step": 6347 }, { "epoch": 0.46, "grad_norm": 3.655111423508501, "learning_rate": 5.962492074606051e-06, "loss": 0.1969, "step": 6348 }, { "epoch": 0.46, "grad_norm": 3.7657923133439817, "learning_rate": 5.961352452196419e-06, "loss": 0.1739, "step": 6349 }, { "epoch": 0.46, "grad_norm": 4.196669532195956, "learning_rate": 5.960212777925459e-06, "loss": 0.2411, "step": 6350 }, { "epoch": 0.46, "grad_norm": 3.5953288036048723, "learning_rate": 5.959073051854658e-06, "loss": 0.2396, "step": 6351 }, { "epoch": 0.46, "grad_norm": 5.09578071474888, "learning_rate": 5.957933274045493e-06, "loss": 0.9405, "step": 6352 }, { "epoch": 0.46, "grad_norm": 17.794035555725976, "learning_rate": 5.956793444559454e-06, "loss": 1.0224, "step": 6353 }, { "epoch": 0.46, "grad_norm": 4.468720728612251, "learning_rate": 5.955653563458034e-06, "loss": 0.2442, "step": 6354 }, { "epoch": 0.46, "grad_norm": 5.108673089757748, "learning_rate": 5.954513630802719e-06, "loss": 0.9002, "step": 6355 }, { "epoch": 0.46, "grad_norm": 23.719664962753804, "learning_rate": 5.9533736466550054e-06, "loss": 0.829, "step": 6356 }, { "epoch": 0.46, "grad_norm": 2.726537400582276, "learning_rate": 5.952233611076394e-06, "loss": 0.1108, "step": 6357 }, { "epoch": 0.46, "grad_norm": 12.520370771847903, "learning_rate": 5.9510935241283825e-06, "loss": 0.8412, "step": 6358 }, { "epoch": 0.46, "grad_norm": 4.104456857541471, "learning_rate": 5.949953385872478e-06, "loss": 0.2312, "step": 6359 }, { "epoch": 0.46, "grad_norm": 4.115714834109404, "learning_rate": 5.948813196370182e-06, "loss": 0.8843, "step": 6360 }, { "epoch": 0.46, "grad_norm": 17.305907076060443, "learning_rate": 5.947672955683006e-06, "loss": 0.4784, "step": 6361 }, { "epoch": 0.46, "grad_norm": 4.477383204951544, "learning_rate": 5.946532663872462e-06, "loss": 0.8335, "step": 6362 }, { "epoch": 0.46, "grad_norm": 3.0481519241681596, "learning_rate": 5.945392321000062e-06, "loss": 0.1587, "step": 6363 }, { "epoch": 0.46, "grad_norm": 3.50020040558168, "learning_rate": 5.944251927127325e-06, "loss": 0.2468, "step": 6364 }, { "epoch": 0.46, "grad_norm": 3.9415867461172476, "learning_rate": 5.943111482315771e-06, "loss": 0.9806, "step": 6365 }, { "epoch": 0.46, "grad_norm": 4.330592952984665, "learning_rate": 5.941970986626923e-06, "loss": 0.7957, "step": 6366 }, { "epoch": 0.46, "grad_norm": 3.7088073189809005, "learning_rate": 5.940830440122305e-06, "loss": 0.2719, "step": 6367 }, { "epoch": 0.46, "grad_norm": 2.692389802822734, "learning_rate": 5.939689842863448e-06, "loss": 0.1969, "step": 6368 }, { "epoch": 0.46, "grad_norm": 3.476013714734162, "learning_rate": 5.938549194911878e-06, "loss": 0.259, "step": 6369 }, { "epoch": 0.46, "grad_norm": 14.321703044630866, "learning_rate": 5.937408496329132e-06, "loss": 1.0935, "step": 6370 }, { "epoch": 0.46, "grad_norm": 9.620565707668893, "learning_rate": 5.936267747176747e-06, "loss": 0.8505, "step": 6371 }, { "epoch": 0.46, "grad_norm": 3.743329316123732, "learning_rate": 5.935126947516262e-06, "loss": 0.2164, "step": 6372 }, { "epoch": 0.46, "grad_norm": 3.8675389587339155, "learning_rate": 5.933986097409215e-06, "loss": 0.6864, "step": 6373 }, { "epoch": 0.46, "grad_norm": 3.3444725486177123, "learning_rate": 5.9328451969171565e-06, "loss": 0.2025, "step": 6374 }, { "epoch": 0.46, "grad_norm": 21.15886510328795, "learning_rate": 5.9317042461016285e-06, "loss": 1.0851, "step": 6375 }, { "epoch": 0.46, "grad_norm": 4.071388580401543, "learning_rate": 5.930563245024184e-06, "loss": 0.1995, "step": 6376 }, { "epoch": 0.46, "grad_norm": 3.619961403220515, "learning_rate": 5.929422193746374e-06, "loss": 0.2248, "step": 6377 }, { "epoch": 0.46, "grad_norm": 2.991000434753614, "learning_rate": 5.928281092329755e-06, "loss": 0.1194, "step": 6378 }, { "epoch": 0.46, "grad_norm": 27.270118173195094, "learning_rate": 5.927139940835885e-06, "loss": 1.0234, "step": 6379 }, { "epoch": 0.46, "grad_norm": 13.033131242068343, "learning_rate": 5.925998739326325e-06, "loss": 0.7732, "step": 6380 }, { "epoch": 0.46, "grad_norm": 4.723667099353282, "learning_rate": 5.924857487862637e-06, "loss": 1.1082, "step": 6381 }, { "epoch": 0.46, "grad_norm": 4.766510821844196, "learning_rate": 5.92371618650639e-06, "loss": 0.964, "step": 6382 }, { "epoch": 0.46, "grad_norm": 3.4091656474010748, "learning_rate": 5.9225748353191495e-06, "loss": 0.1947, "step": 6383 }, { "epoch": 0.46, "grad_norm": 3.492655487601078, "learning_rate": 5.92143343436249e-06, "loss": 0.2151, "step": 6384 }, { "epoch": 0.46, "grad_norm": 3.548175624362465, "learning_rate": 5.920291983697984e-06, "loss": 0.2775, "step": 6385 }, { "epoch": 0.46, "grad_norm": 3.1863555613571983, "learning_rate": 5.919150483387209e-06, "loss": 0.1629, "step": 6386 }, { "epoch": 0.46, "grad_norm": 4.783915606006269, "learning_rate": 5.918008933491744e-06, "loss": 0.7011, "step": 6387 }, { "epoch": 0.46, "grad_norm": 4.504809444355974, "learning_rate": 5.916867334073173e-06, "loss": 0.2682, "step": 6388 }, { "epoch": 0.46, "grad_norm": 4.734126132015632, "learning_rate": 5.9157256851930785e-06, "loss": 0.9334, "step": 6389 }, { "epoch": 0.46, "grad_norm": 12.142909690876946, "learning_rate": 5.914583986913052e-06, "loss": 0.9063, "step": 6390 }, { "epoch": 0.46, "grad_norm": 2.9182074851917523, "learning_rate": 5.91344223929468e-06, "loss": 0.1644, "step": 6391 }, { "epoch": 0.46, "grad_norm": 4.6871193948568015, "learning_rate": 5.9123004423995554e-06, "loss": 0.8213, "step": 6392 }, { "epoch": 0.46, "grad_norm": 12.301464114357263, "learning_rate": 5.911158596289276e-06, "loss": 0.5194, "step": 6393 }, { "epoch": 0.46, "grad_norm": 3.0534485303612726, "learning_rate": 5.9100167010254394e-06, "loss": 0.1477, "step": 6394 }, { "epoch": 0.46, "grad_norm": 2.651439252438056, "learning_rate": 5.908874756669646e-06, "loss": 0.1819, "step": 6395 }, { "epoch": 0.46, "grad_norm": 3.795956706819546, "learning_rate": 5.9077327632835e-06, "loss": 0.1426, "step": 6396 }, { "epoch": 0.46, "grad_norm": 3.8601760670933047, "learning_rate": 5.906590720928609e-06, "loss": 0.2207, "step": 6397 }, { "epoch": 0.46, "grad_norm": 4.487253249942602, "learning_rate": 5.905448629666578e-06, "loss": 0.7888, "step": 6398 }, { "epoch": 0.46, "grad_norm": 23.501700250460633, "learning_rate": 5.904306489559022e-06, "loss": 1.0833, "step": 6399 }, { "epoch": 0.46, "grad_norm": 5.323802818147889, "learning_rate": 5.903164300667554e-06, "loss": 0.9823, "step": 6400 }, { "epoch": 0.46, "grad_norm": 17.33476840599258, "learning_rate": 5.902022063053788e-06, "loss": 1.1462, "step": 6401 }, { "epoch": 0.46, "grad_norm": 3.708118057287577, "learning_rate": 5.90087977677935e-06, "loss": 0.189, "step": 6402 }, { "epoch": 0.46, "grad_norm": 5.080742093284102, "learning_rate": 5.899737441905855e-06, "loss": 0.2344, "step": 6403 }, { "epoch": 0.46, "grad_norm": 3.8178138260481247, "learning_rate": 5.898595058494932e-06, "loss": 0.2423, "step": 6404 }, { "epoch": 0.46, "grad_norm": 3.447811917124558, "learning_rate": 5.897452626608207e-06, "loss": 0.1671, "step": 6405 }, { "epoch": 0.46, "grad_norm": 7.338769051808412, "learning_rate": 5.8963101463073094e-06, "loss": 0.4324, "step": 6406 }, { "epoch": 0.46, "grad_norm": 2.555349661328617, "learning_rate": 5.895167617653872e-06, "loss": 0.14, "step": 6407 }, { "epoch": 0.46, "grad_norm": 7.9307535397444076, "learning_rate": 5.8940250407095316e-06, "loss": 0.5143, "step": 6408 }, { "epoch": 0.46, "grad_norm": 3.7159127679203827, "learning_rate": 5.892882415535922e-06, "loss": 0.3079, "step": 6409 }, { "epoch": 0.46, "grad_norm": 14.28578148872454, "learning_rate": 5.8917397421946865e-06, "loss": 1.0943, "step": 6410 }, { "epoch": 0.46, "grad_norm": 3.665139924681717, "learning_rate": 5.890597020747468e-06, "loss": 0.2041, "step": 6411 }, { "epoch": 0.46, "grad_norm": 3.1815861122464577, "learning_rate": 5.88945425125591e-06, "loss": 0.1718, "step": 6412 }, { "epoch": 0.46, "grad_norm": 2.9320987727659378, "learning_rate": 5.888311433781665e-06, "loss": 0.1757, "step": 6413 }, { "epoch": 0.46, "grad_norm": 8.738424369083328, "learning_rate": 5.887168568386379e-06, "loss": 0.6488, "step": 6414 }, { "epoch": 0.46, "grad_norm": 4.79968770390852, "learning_rate": 5.886025655131706e-06, "loss": 1.1976, "step": 6415 }, { "epoch": 0.46, "grad_norm": 3.7232102132093416, "learning_rate": 5.884882694079304e-06, "loss": 0.2556, "step": 6416 }, { "epoch": 0.46, "grad_norm": 13.5837669002079, "learning_rate": 5.883739685290828e-06, "loss": 0.9039, "step": 6417 }, { "epoch": 0.46, "grad_norm": 15.973187659463381, "learning_rate": 5.882596628827941e-06, "loss": 1.0245, "step": 6418 }, { "epoch": 0.46, "grad_norm": 4.389844069182834, "learning_rate": 5.88145352475231e-06, "loss": 0.2671, "step": 6419 }, { "epoch": 0.46, "grad_norm": 10.184878063272018, "learning_rate": 5.880310373125595e-06, "loss": 0.7437, "step": 6420 }, { "epoch": 0.46, "grad_norm": 3.031589623788584, "learning_rate": 5.879167174009468e-06, "loss": 0.1738, "step": 6421 }, { "epoch": 0.46, "grad_norm": 3.7415161807600277, "learning_rate": 5.878023927465602e-06, "loss": 0.1769, "step": 6422 }, { "epoch": 0.46, "grad_norm": 14.12025849110487, "learning_rate": 5.876880633555665e-06, "loss": 0.9465, "step": 6423 }, { "epoch": 0.46, "grad_norm": 20.7902551586239, "learning_rate": 5.875737292341338e-06, "loss": 0.9492, "step": 6424 }, { "epoch": 0.46, "grad_norm": 9.78732339909211, "learning_rate": 5.8745939038843e-06, "loss": 0.9232, "step": 6425 }, { "epoch": 0.46, "grad_norm": 11.063103391100267, "learning_rate": 5.8734504682462325e-06, "loss": 0.5033, "step": 6426 }, { "epoch": 0.46, "grad_norm": 3.5092828237678515, "learning_rate": 5.872306985488816e-06, "loss": 0.2652, "step": 6427 }, { "epoch": 0.46, "grad_norm": 14.845110012062028, "learning_rate": 5.871163455673742e-06, "loss": 1.0217, "step": 6428 }, { "epoch": 0.46, "grad_norm": 10.513747049776763, "learning_rate": 5.870019878862694e-06, "loss": 0.5582, "step": 6429 }, { "epoch": 0.46, "grad_norm": 4.238481583278562, "learning_rate": 5.86887625511737e-06, "loss": 0.3387, "step": 6430 }, { "epoch": 0.46, "grad_norm": 3.0302303747998547, "learning_rate": 5.867732584499458e-06, "loss": 0.1821, "step": 6431 }, { "epoch": 0.46, "grad_norm": 2.371796048896874, "learning_rate": 5.86658886707066e-06, "loss": 0.1541, "step": 6432 }, { "epoch": 0.46, "grad_norm": 13.488677980829594, "learning_rate": 5.86544510289267e-06, "loss": 1.0274, "step": 6433 }, { "epoch": 0.46, "grad_norm": 4.035961017685475, "learning_rate": 5.864301292027196e-06, "loss": 0.1635, "step": 6434 }, { "epoch": 0.46, "grad_norm": 13.56964286197946, "learning_rate": 5.863157434535935e-06, "loss": 0.8788, "step": 6435 }, { "epoch": 0.46, "grad_norm": 3.4940676395324823, "learning_rate": 5.862013530480601e-06, "loss": 0.1688, "step": 6436 }, { "epoch": 0.46, "grad_norm": 5.00336512836228, "learning_rate": 5.860869579922899e-06, "loss": 1.0205, "step": 6437 }, { "epoch": 0.46, "grad_norm": 2.651429639284467, "learning_rate": 5.859725582924541e-06, "loss": 0.1882, "step": 6438 }, { "epoch": 0.46, "grad_norm": 10.744272337145654, "learning_rate": 5.858581539547244e-06, "loss": 0.7434, "step": 6439 }, { "epoch": 0.46, "grad_norm": 12.848588101011519, "learning_rate": 5.857437449852721e-06, "loss": 1.1116, "step": 6440 }, { "epoch": 0.46, "grad_norm": 4.717622225628916, "learning_rate": 5.856293313902692e-06, "loss": 0.6625, "step": 6441 }, { "epoch": 0.46, "grad_norm": 4.399763544501294, "learning_rate": 5.855149131758883e-06, "loss": 0.7286, "step": 6442 }, { "epoch": 0.46, "grad_norm": 4.498535452647451, "learning_rate": 5.854004903483013e-06, "loss": 0.7304, "step": 6443 }, { "epoch": 0.46, "grad_norm": 38.24527003927224, "learning_rate": 5.852860629136812e-06, "loss": 1.3414, "step": 6444 }, { "epoch": 0.46, "grad_norm": 3.8106276074065724, "learning_rate": 5.8517163087820084e-06, "loss": 0.1749, "step": 6445 }, { "epoch": 0.46, "grad_norm": 3.4125450553797014, "learning_rate": 5.850571942480333e-06, "loss": 0.1896, "step": 6446 }, { "epoch": 0.46, "grad_norm": 3.629705971564086, "learning_rate": 5.8494275302935215e-06, "loss": 0.1376, "step": 6447 }, { "epoch": 0.46, "grad_norm": 3.993631133928878, "learning_rate": 5.84828307228331e-06, "loss": 0.2113, "step": 6448 }, { "epoch": 0.46, "grad_norm": 3.5191596032091175, "learning_rate": 5.847138568511436e-06, "loss": 0.1289, "step": 6449 }, { "epoch": 0.46, "grad_norm": 4.436837040031869, "learning_rate": 5.845994019039645e-06, "loss": 0.7917, "step": 6450 }, { "epoch": 0.46, "grad_norm": 4.536523566594473, "learning_rate": 5.844849423929679e-06, "loss": 0.7833, "step": 6451 }, { "epoch": 0.46, "grad_norm": 15.232606704326338, "learning_rate": 5.843704783243284e-06, "loss": 1.0987, "step": 6452 }, { "epoch": 0.46, "grad_norm": 4.724085508375307, "learning_rate": 5.84256009704221e-06, "loss": 0.9539, "step": 6453 }, { "epoch": 0.46, "grad_norm": 14.985806471079554, "learning_rate": 5.8414153653882065e-06, "loss": 1.0011, "step": 6454 }, { "epoch": 0.46, "grad_norm": 3.1098143244575556, "learning_rate": 5.84027058834303e-06, "loss": 0.2215, "step": 6455 }, { "epoch": 0.46, "grad_norm": 15.068640124859034, "learning_rate": 5.839125765968436e-06, "loss": 0.8403, "step": 6456 }, { "epoch": 0.46, "grad_norm": 3.107823108647332, "learning_rate": 5.837980898326183e-06, "loss": 0.1584, "step": 6457 }, { "epoch": 0.46, "grad_norm": 3.9072382934660146, "learning_rate": 5.836835985478032e-06, "loss": 0.189, "step": 6458 }, { "epoch": 0.46, "grad_norm": 2.8386336153366263, "learning_rate": 5.8356910274857485e-06, "loss": 0.1878, "step": 6459 }, { "epoch": 0.46, "grad_norm": 2.7032956791519434, "learning_rate": 5.8345460244110955e-06, "loss": 0.1344, "step": 6460 }, { "epoch": 0.46, "grad_norm": 22.086457181203368, "learning_rate": 5.833400976315844e-06, "loss": 1.4927, "step": 6461 }, { "epoch": 0.46, "grad_norm": 4.230944385456459, "learning_rate": 5.8322558832617656e-06, "loss": 0.203, "step": 6462 }, { "epoch": 0.46, "grad_norm": 3.498352959011962, "learning_rate": 5.831110745310632e-06, "loss": 0.1189, "step": 6463 }, { "epoch": 0.46, "grad_norm": 10.841398861604478, "learning_rate": 5.829965562524218e-06, "loss": 0.8176, "step": 6464 }, { "epoch": 0.46, "grad_norm": 9.675019689589456, "learning_rate": 5.828820334964306e-06, "loss": 0.6814, "step": 6465 }, { "epoch": 0.46, "grad_norm": 2.6460628984146632, "learning_rate": 5.827675062692673e-06, "loss": 0.1135, "step": 6466 }, { "epoch": 0.46, "grad_norm": 3.0094271631333758, "learning_rate": 5.826529745771103e-06, "loss": 0.144, "step": 6467 }, { "epoch": 0.46, "grad_norm": 2.9947709331537005, "learning_rate": 5.825384384261384e-06, "loss": 0.2072, "step": 6468 }, { "epoch": 0.46, "grad_norm": 2.7390961508282774, "learning_rate": 5.824238978225299e-06, "loss": 0.2113, "step": 6469 }, { "epoch": 0.46, "grad_norm": 2.959908268350049, "learning_rate": 5.823093527724641e-06, "loss": 0.1978, "step": 6470 }, { "epoch": 0.46, "grad_norm": 4.417364404624074, "learning_rate": 5.821948032821203e-06, "loss": 0.971, "step": 6471 }, { "epoch": 0.46, "grad_norm": 2.958451331241044, "learning_rate": 5.82080249357678e-06, "loss": 0.1831, "step": 6472 }, { "epoch": 0.46, "grad_norm": 2.6246948446883605, "learning_rate": 5.819656910053171e-06, "loss": 0.1771, "step": 6473 }, { "epoch": 0.46, "grad_norm": 9.154293591671998, "learning_rate": 5.8185112823121735e-06, "loss": 0.6238, "step": 6474 }, { "epoch": 0.46, "grad_norm": 14.372186768972075, "learning_rate": 5.81736561041559e-06, "loss": 0.7107, "step": 6475 }, { "epoch": 0.46, "grad_norm": 4.6628620707173365, "learning_rate": 5.816219894425226e-06, "loss": 0.931, "step": 6476 }, { "epoch": 0.46, "grad_norm": 3.11423446726329, "learning_rate": 5.815074134402887e-06, "loss": 0.2302, "step": 6477 }, { "epoch": 0.46, "grad_norm": 18.69926856445956, "learning_rate": 5.8139283304103845e-06, "loss": 1.1427, "step": 6478 }, { "epoch": 0.46, "grad_norm": 5.55664231271022, "learning_rate": 5.81278248250953e-06, "loss": 0.4383, "step": 6479 }, { "epoch": 0.46, "grad_norm": 3.242636976158601, "learning_rate": 5.811636590762138e-06, "loss": 0.2216, "step": 6480 }, { "epoch": 0.46, "grad_norm": 3.4075008061477865, "learning_rate": 5.810490655230022e-06, "loss": 0.2217, "step": 6481 }, { "epoch": 0.46, "grad_norm": 4.167210915677501, "learning_rate": 5.809344675975005e-06, "loss": 0.7844, "step": 6482 }, { "epoch": 0.46, "grad_norm": 10.406682017189151, "learning_rate": 5.808198653058905e-06, "loss": 0.9638, "step": 6483 }, { "epoch": 0.46, "grad_norm": 2.694868383400973, "learning_rate": 5.807052586543548e-06, "loss": 0.2135, "step": 6484 }, { "epoch": 0.47, "grad_norm": 20.65201642349969, "learning_rate": 5.805906476490757e-06, "loss": 0.7645, "step": 6485 }, { "epoch": 0.47, "grad_norm": 30.81207651207101, "learning_rate": 5.804760322962364e-06, "loss": 2.1045, "step": 6486 }, { "epoch": 0.47, "grad_norm": 4.578108029496168, "learning_rate": 5.803614126020196e-06, "loss": 0.1418, "step": 6487 }, { "epoch": 0.47, "grad_norm": 4.856533429062606, "learning_rate": 5.8024678857260896e-06, "loss": 0.2081, "step": 6488 }, { "epoch": 0.47, "grad_norm": 14.810041195222801, "learning_rate": 5.801321602141876e-06, "loss": 0.8626, "step": 6489 }, { "epoch": 0.47, "grad_norm": 20.180776015710084, "learning_rate": 5.800175275329397e-06, "loss": 0.8965, "step": 6490 }, { "epoch": 0.47, "grad_norm": 3.4307839170462278, "learning_rate": 5.79902890535049e-06, "loss": 0.2146, "step": 6491 }, { "epoch": 0.47, "grad_norm": 14.114658042017549, "learning_rate": 5.797882492266999e-06, "loss": 0.5296, "step": 6492 }, { "epoch": 0.47, "grad_norm": 2.8058237237138846, "learning_rate": 5.7967360361407665e-06, "loss": 0.1963, "step": 6493 }, { "epoch": 0.47, "grad_norm": 4.443479370497395, "learning_rate": 5.795589537033641e-06, "loss": 0.342, "step": 6494 }, { "epoch": 0.47, "grad_norm": 25.705557966765472, "learning_rate": 5.794442995007471e-06, "loss": 0.7466, "step": 6495 }, { "epoch": 0.47, "grad_norm": 3.850584964671127, "learning_rate": 5.793296410124109e-06, "loss": 0.2673, "step": 6496 }, { "epoch": 0.47, "grad_norm": 29.51566563139527, "learning_rate": 5.792149782445408e-06, "loss": 0.9187, "step": 6497 }, { "epoch": 0.47, "grad_norm": 20.984823580501523, "learning_rate": 5.791003112033227e-06, "loss": 1.1624, "step": 6498 }, { "epoch": 0.47, "grad_norm": 4.932504314289925, "learning_rate": 5.7898563989494204e-06, "loss": 0.3579, "step": 6499 }, { "epoch": 0.47, "grad_norm": 16.55350576534982, "learning_rate": 5.78870964325585e-06, "loss": 1.0301, "step": 6500 }, { "epoch": 0.47, "grad_norm": 3.106423577109931, "learning_rate": 5.787562845014381e-06, "loss": 0.26, "step": 6501 }, { "epoch": 0.47, "grad_norm": 5.6722470631256625, "learning_rate": 5.786416004286877e-06, "loss": 0.2873, "step": 6502 }, { "epoch": 0.47, "grad_norm": 4.0380774862249105, "learning_rate": 5.785269121135206e-06, "loss": 0.9751, "step": 6503 }, { "epoch": 0.47, "grad_norm": 3.5221554310270924, "learning_rate": 5.784122195621239e-06, "loss": 0.1484, "step": 6504 }, { "epoch": 0.47, "grad_norm": 8.665934850404904, "learning_rate": 5.782975227806848e-06, "loss": 0.711, "step": 6505 }, { "epoch": 0.47, "grad_norm": 5.3354307153312766, "learning_rate": 5.781828217753906e-06, "loss": 0.3434, "step": 6506 }, { "epoch": 0.47, "grad_norm": 4.4981566561146105, "learning_rate": 5.780681165524294e-06, "loss": 0.8229, "step": 6507 }, { "epoch": 0.47, "grad_norm": 13.497964143921847, "learning_rate": 5.779534071179884e-06, "loss": 0.5544, "step": 6508 }, { "epoch": 0.47, "grad_norm": 26.500907014582868, "learning_rate": 5.7783869347825626e-06, "loss": 0.8623, "step": 6509 }, { "epoch": 0.47, "grad_norm": 3.8518164941533684, "learning_rate": 5.777239756394215e-06, "loss": 0.1838, "step": 6510 }, { "epoch": 0.47, "grad_norm": 12.98054828012862, "learning_rate": 5.776092536076723e-06, "loss": 0.6895, "step": 6511 }, { "epoch": 0.47, "grad_norm": 2.882962076468804, "learning_rate": 5.7749452738919754e-06, "loss": 0.2283, "step": 6512 }, { "epoch": 0.47, "grad_norm": 19.968106248247704, "learning_rate": 5.773797969901864e-06, "loss": 0.7304, "step": 6513 }, { "epoch": 0.47, "grad_norm": 4.602254330008244, "learning_rate": 5.772650624168281e-06, "loss": 0.9301, "step": 6514 }, { "epoch": 0.47, "grad_norm": 13.032957079419472, "learning_rate": 5.771503236753122e-06, "loss": 0.7738, "step": 6515 }, { "epoch": 0.47, "grad_norm": 2.8562525715097284, "learning_rate": 5.770355807718285e-06, "loss": 0.1177, "step": 6516 }, { "epoch": 0.47, "grad_norm": 4.754611380442773, "learning_rate": 5.769208337125666e-06, "loss": 0.9636, "step": 6517 }, { "epoch": 0.47, "grad_norm": 3.809493879037514, "learning_rate": 5.768060825037169e-06, "loss": 0.2122, "step": 6518 }, { "epoch": 0.47, "grad_norm": 9.648351382686217, "learning_rate": 5.766913271514699e-06, "loss": 0.785, "step": 6519 }, { "epoch": 0.47, "grad_norm": 3.6246312795587188, "learning_rate": 5.76576567662016e-06, "loss": 0.3911, "step": 6520 }, { "epoch": 0.47, "grad_norm": 2.9578799094032857, "learning_rate": 5.764618040415462e-06, "loss": 0.2421, "step": 6521 }, { "epoch": 0.47, "grad_norm": 9.915716717246223, "learning_rate": 5.763470362962515e-06, "loss": 0.8152, "step": 6522 }, { "epoch": 0.47, "grad_norm": 6.420370759731764, "learning_rate": 5.762322644323231e-06, "loss": 0.9391, "step": 6523 }, { "epoch": 0.47, "grad_norm": 14.483158818633115, "learning_rate": 5.761174884559526e-06, "loss": 0.891, "step": 6524 }, { "epoch": 0.47, "grad_norm": 7.073552961028206, "learning_rate": 5.760027083733317e-06, "loss": 0.5626, "step": 6525 }, { "epoch": 0.47, "grad_norm": 3.065492053241823, "learning_rate": 5.758879241906523e-06, "loss": 0.3014, "step": 6526 }, { "epoch": 0.47, "grad_norm": 5.247487121130845, "learning_rate": 5.757731359141069e-06, "loss": 0.4515, "step": 6527 }, { "epoch": 0.47, "grad_norm": 2.804094323391574, "learning_rate": 5.756583435498874e-06, "loss": 0.1249, "step": 6528 }, { "epoch": 0.47, "grad_norm": 3.623534068811072, "learning_rate": 5.7554354710418684e-06, "loss": 0.8362, "step": 6529 }, { "epoch": 0.47, "grad_norm": 3.102023777865649, "learning_rate": 5.754287465831977e-06, "loss": 0.2417, "step": 6530 }, { "epoch": 0.47, "grad_norm": 15.042428826887818, "learning_rate": 5.753139419931132e-06, "loss": 1.0746, "step": 6531 }, { "epoch": 0.47, "grad_norm": 12.689685659681599, "learning_rate": 5.751991333401266e-06, "loss": 0.5585, "step": 6532 }, { "epoch": 0.47, "grad_norm": 18.91514618625825, "learning_rate": 5.7508432063043155e-06, "loss": 1.2709, "step": 6533 }, { "epoch": 0.47, "grad_norm": 170.51495975563415, "learning_rate": 5.749695038702214e-06, "loss": 0.6125, "step": 6534 }, { "epoch": 0.47, "grad_norm": 19.859684887502052, "learning_rate": 5.7485468306569045e-06, "loss": 1.1054, "step": 6535 }, { "epoch": 0.47, "grad_norm": 4.724970464415759, "learning_rate": 5.747398582230328e-06, "loss": 0.8195, "step": 6536 }, { "epoch": 0.47, "grad_norm": 3.6647030637058706, "learning_rate": 5.7462502934844244e-06, "loss": 0.1681, "step": 6537 }, { "epoch": 0.47, "grad_norm": 4.418415749724829, "learning_rate": 5.745101964481144e-06, "loss": 0.3106, "step": 6538 }, { "epoch": 0.47, "grad_norm": 25.836134028384667, "learning_rate": 5.743953595282432e-06, "loss": 0.8775, "step": 6539 }, { "epoch": 0.47, "grad_norm": 4.633031305211974, "learning_rate": 5.742805185950239e-06, "loss": 0.7841, "step": 6540 }, { "epoch": 0.47, "grad_norm": 19.706231545033805, "learning_rate": 5.741656736546519e-06, "loss": 0.7564, "step": 6541 }, { "epoch": 0.47, "grad_norm": 4.032516440526068, "learning_rate": 5.740508247133226e-06, "loss": 0.177, "step": 6542 }, { "epoch": 0.47, "grad_norm": 3.10904262948874, "learning_rate": 5.739359717772314e-06, "loss": 0.1236, "step": 6543 }, { "epoch": 0.47, "grad_norm": 4.280011766806482, "learning_rate": 5.738211148525746e-06, "loss": 0.174, "step": 6544 }, { "epoch": 0.47, "grad_norm": 3.7494643033035873, "learning_rate": 5.737062539455479e-06, "loss": 0.1564, "step": 6545 }, { "epoch": 0.47, "grad_norm": 4.768542068997851, "learning_rate": 5.735913890623479e-06, "loss": 0.2558, "step": 6546 }, { "epoch": 0.47, "grad_norm": 5.139332779450016, "learning_rate": 5.734765202091711e-06, "loss": 0.1764, "step": 6547 }, { "epoch": 0.47, "grad_norm": 3.58076458012905, "learning_rate": 5.73361647392214e-06, "loss": 0.1628, "step": 6548 }, { "epoch": 0.47, "grad_norm": 4.838431792564614, "learning_rate": 5.732467706176736e-06, "loss": 0.2583, "step": 6549 }, { "epoch": 0.47, "grad_norm": 4.355403388431983, "learning_rate": 5.731318898917475e-06, "loss": 0.2588, "step": 6550 }, { "epoch": 0.47, "grad_norm": 5.099391583842165, "learning_rate": 5.730170052206325e-06, "loss": 0.379, "step": 6551 }, { "epoch": 0.47, "grad_norm": 11.607364195977931, "learning_rate": 5.7290211661052655e-06, "loss": 0.7485, "step": 6552 }, { "epoch": 0.47, "grad_norm": 2.661189283285072, "learning_rate": 5.727872240676275e-06, "loss": 0.1036, "step": 6553 }, { "epoch": 0.47, "grad_norm": 3.4349168431289967, "learning_rate": 5.7267232759813304e-06, "loss": 0.1432, "step": 6554 }, { "epoch": 0.47, "grad_norm": 3.511064186382821, "learning_rate": 5.725574272082416e-06, "loss": 0.2094, "step": 6555 }, { "epoch": 0.47, "grad_norm": 5.316543374738909, "learning_rate": 5.724425229041517e-06, "loss": 0.941, "step": 6556 }, { "epoch": 0.47, "grad_norm": 3.585074173246299, "learning_rate": 5.723276146920618e-06, "loss": 0.2143, "step": 6557 }, { "epoch": 0.47, "grad_norm": 40.670314410108595, "learning_rate": 5.72212702578171e-06, "loss": 1.8133, "step": 6558 }, { "epoch": 0.47, "grad_norm": 3.7297146412341617, "learning_rate": 5.720977865686781e-06, "loss": 0.1818, "step": 6559 }, { "epoch": 0.47, "grad_norm": 10.385435399181768, "learning_rate": 5.719828666697825e-06, "loss": 0.5638, "step": 6560 }, { "epoch": 0.47, "grad_norm": 7.792770697696976, "learning_rate": 5.718679428876839e-06, "loss": 0.9782, "step": 6561 }, { "epoch": 0.47, "grad_norm": 18.001546452832976, "learning_rate": 5.717530152285816e-06, "loss": 0.9087, "step": 6562 }, { "epoch": 0.47, "grad_norm": 3.519027757159143, "learning_rate": 5.716380836986758e-06, "loss": 0.1703, "step": 6563 }, { "epoch": 0.47, "grad_norm": 11.09692605856627, "learning_rate": 5.7152314830416654e-06, "loss": 0.9416, "step": 6564 }, { "epoch": 0.47, "grad_norm": 11.782619529309109, "learning_rate": 5.714082090512542e-06, "loss": 0.7812, "step": 6565 }, { "epoch": 0.47, "grad_norm": 4.61636685093967, "learning_rate": 5.712932659461393e-06, "loss": 0.8406, "step": 6566 }, { "epoch": 0.47, "grad_norm": 4.777554108168079, "learning_rate": 5.711783189950226e-06, "loss": 1.0919, "step": 6567 }, { "epoch": 0.47, "grad_norm": 16.36763216072745, "learning_rate": 5.710633682041049e-06, "loss": 0.9615, "step": 6568 }, { "epoch": 0.47, "grad_norm": 3.8551910443917086, "learning_rate": 5.709484135795874e-06, "loss": 0.2372, "step": 6569 }, { "epoch": 0.47, "grad_norm": 3.810995820010702, "learning_rate": 5.708334551276718e-06, "loss": 0.1583, "step": 6570 }, { "epoch": 0.47, "grad_norm": 4.653787022322391, "learning_rate": 5.707184928545593e-06, "loss": 0.7554, "step": 6571 }, { "epoch": 0.47, "grad_norm": 15.17932409100783, "learning_rate": 5.706035267664517e-06, "loss": 1.0654, "step": 6572 }, { "epoch": 0.47, "grad_norm": 3.6538576374216807, "learning_rate": 5.704885568695512e-06, "loss": 0.1928, "step": 6573 }, { "epoch": 0.47, "grad_norm": 3.8056159317424822, "learning_rate": 5.703735831700597e-06, "loss": 0.2135, "step": 6574 }, { "epoch": 0.47, "grad_norm": 15.970417114784894, "learning_rate": 5.702586056741799e-06, "loss": 1.0465, "step": 6575 }, { "epoch": 0.47, "grad_norm": 5.319861614227121, "learning_rate": 5.701436243881142e-06, "loss": 0.2391, "step": 6576 }, { "epoch": 0.47, "grad_norm": 16.480114508478216, "learning_rate": 5.700286393180656e-06, "loss": 0.8342, "step": 6577 }, { "epoch": 0.47, "grad_norm": 4.676301863081428, "learning_rate": 5.699136504702369e-06, "loss": 0.8007, "step": 6578 }, { "epoch": 0.47, "grad_norm": 19.826342464749448, "learning_rate": 5.697986578508314e-06, "loss": 0.6401, "step": 6579 }, { "epoch": 0.47, "grad_norm": 3.0837344303599563, "learning_rate": 5.696836614660525e-06, "loss": 0.1451, "step": 6580 }, { "epoch": 0.47, "grad_norm": 2.709766683679327, "learning_rate": 5.695686613221038e-06, "loss": 0.1566, "step": 6581 }, { "epoch": 0.47, "grad_norm": 13.306154718914714, "learning_rate": 5.694536574251892e-06, "loss": 1.0543, "step": 6582 }, { "epoch": 0.47, "grad_norm": 15.411430323488695, "learning_rate": 5.693386497815128e-06, "loss": 0.8054, "step": 6583 }, { "epoch": 0.47, "grad_norm": 4.923037486969645, "learning_rate": 5.692236383972786e-06, "loss": 0.9375, "step": 6584 }, { "epoch": 0.47, "grad_norm": 3.686386041281925, "learning_rate": 5.69108623278691e-06, "loss": 0.1853, "step": 6585 }, { "epoch": 0.47, "grad_norm": 6.971839099827852, "learning_rate": 5.6899360443195485e-06, "loss": 0.4982, "step": 6586 }, { "epoch": 0.47, "grad_norm": 3.6371494097892554, "learning_rate": 5.688785818632749e-06, "loss": 0.2412, "step": 6587 }, { "epoch": 0.47, "grad_norm": 3.2674985668692895, "learning_rate": 5.687635555788562e-06, "loss": 0.2153, "step": 6588 }, { "epoch": 0.47, "grad_norm": 4.461562781441946, "learning_rate": 5.6864852558490405e-06, "loss": 1.0005, "step": 6589 }, { "epoch": 0.47, "grad_norm": 4.7739048323544155, "learning_rate": 5.685334918876237e-06, "loss": 0.9535, "step": 6590 }, { "epoch": 0.47, "grad_norm": 2.411961889464227, "learning_rate": 5.6841845449322084e-06, "loss": 0.1454, "step": 6591 }, { "epoch": 0.47, "grad_norm": 4.098145948045164, "learning_rate": 5.683034134079013e-06, "loss": 1.0744, "step": 6592 }, { "epoch": 0.47, "grad_norm": 12.466799705762076, "learning_rate": 5.681883686378713e-06, "loss": 0.9238, "step": 6593 }, { "epoch": 0.47, "grad_norm": 16.40218208256095, "learning_rate": 5.680733201893368e-06, "loss": 0.8667, "step": 6594 }, { "epoch": 0.47, "grad_norm": 3.7166027399500914, "learning_rate": 5.679582680685044e-06, "loss": 0.1839, "step": 6595 }, { "epoch": 0.47, "grad_norm": 3.182862689910719, "learning_rate": 5.678432122815808e-06, "loss": 0.1347, "step": 6596 }, { "epoch": 0.47, "grad_norm": 3.341122566650163, "learning_rate": 5.6772815283477246e-06, "loss": 0.1645, "step": 6597 }, { "epoch": 0.47, "grad_norm": 3.245392056138796, "learning_rate": 5.676130897342868e-06, "loss": 0.1715, "step": 6598 }, { "epoch": 0.47, "grad_norm": 5.043102816644094, "learning_rate": 5.674980229863307e-06, "loss": 0.8806, "step": 6599 }, { "epoch": 0.47, "grad_norm": 4.095356565444046, "learning_rate": 5.673829525971118e-06, "loss": 0.343, "step": 6600 }, { "epoch": 0.47, "grad_norm": 3.4294258148771872, "learning_rate": 5.67267878572838e-06, "loss": 0.1843, "step": 6601 }, { "epoch": 0.47, "grad_norm": 3.3002314444391527, "learning_rate": 5.671528009197165e-06, "loss": 0.1559, "step": 6602 }, { "epoch": 0.47, "grad_norm": 3.3584691193314495, "learning_rate": 5.670377196439555e-06, "loss": 0.1956, "step": 6603 }, { "epoch": 0.47, "grad_norm": 3.737710165739889, "learning_rate": 5.669226347517634e-06, "loss": 0.1858, "step": 6604 }, { "epoch": 0.47, "grad_norm": 3.3712852816209447, "learning_rate": 5.668075462493484e-06, "loss": 0.1674, "step": 6605 }, { "epoch": 0.47, "grad_norm": 4.111265898720623, "learning_rate": 5.6669245414291915e-06, "loss": 0.2537, "step": 6606 }, { "epoch": 0.47, "grad_norm": 3.019057736036464, "learning_rate": 5.665773584386847e-06, "loss": 0.1767, "step": 6607 }, { "epoch": 0.47, "grad_norm": 15.705046077481912, "learning_rate": 5.6646225914285346e-06, "loss": 1.007, "step": 6608 }, { "epoch": 0.47, "grad_norm": 3.5959598089369265, "learning_rate": 5.6634715626163495e-06, "loss": 0.1719, "step": 6609 }, { "epoch": 0.47, "grad_norm": 4.589384828799845, "learning_rate": 5.662320498012386e-06, "loss": 0.8086, "step": 6610 }, { "epoch": 0.47, "grad_norm": 3.574859079338136, "learning_rate": 5.6611693976787385e-06, "loss": 0.261, "step": 6611 }, { "epoch": 0.47, "grad_norm": 3.165749273379793, "learning_rate": 5.6600182616775045e-06, "loss": 0.1427, "step": 6612 }, { "epoch": 0.47, "grad_norm": 4.51153551118243, "learning_rate": 5.658867090070784e-06, "loss": 0.7846, "step": 6613 }, { "epoch": 0.47, "grad_norm": 2.968577985599216, "learning_rate": 5.657715882920677e-06, "loss": 0.1323, "step": 6614 }, { "epoch": 0.47, "grad_norm": 4.084862833019276, "learning_rate": 5.6565646402892895e-06, "loss": 0.33, "step": 6615 }, { "epoch": 0.47, "grad_norm": 28.07005969294338, "learning_rate": 5.655413362238722e-06, "loss": 1.3375, "step": 6616 }, { "epoch": 0.47, "grad_norm": 3.630271963860434, "learning_rate": 5.6542620488310865e-06, "loss": 0.1592, "step": 6617 }, { "epoch": 0.47, "grad_norm": 15.59103068024695, "learning_rate": 5.653110700128489e-06, "loss": 0.8062, "step": 6618 }, { "epoch": 0.47, "grad_norm": 4.152837339271447, "learning_rate": 5.651959316193043e-06, "loss": 0.2756, "step": 6619 }, { "epoch": 0.47, "grad_norm": 13.055977629183467, "learning_rate": 5.6508078970868585e-06, "loss": 0.9191, "step": 6620 }, { "epoch": 0.47, "grad_norm": 2.753916428956951, "learning_rate": 5.649656442872051e-06, "loss": 0.1409, "step": 6621 }, { "epoch": 0.47, "grad_norm": 3.9256917634548043, "learning_rate": 5.6485049536107386e-06, "loss": 0.2092, "step": 6622 }, { "epoch": 0.47, "grad_norm": 14.606638909830435, "learning_rate": 5.647353429365036e-06, "loss": 1.1257, "step": 6623 }, { "epoch": 0.47, "grad_norm": 4.536688065458288, "learning_rate": 5.64620187019707e-06, "loss": 0.6856, "step": 6624 }, { "epoch": 0.48, "grad_norm": 4.576363216241882, "learning_rate": 5.645050276168958e-06, "loss": 0.8257, "step": 6625 }, { "epoch": 0.48, "grad_norm": 4.822034644995707, "learning_rate": 5.643898647342824e-06, "loss": 0.8935, "step": 6626 }, { "epoch": 0.48, "grad_norm": 11.499986298856465, "learning_rate": 5.642746983780797e-06, "loss": 0.7245, "step": 6627 }, { "epoch": 0.48, "grad_norm": 18.37738895246483, "learning_rate": 5.6415952855450005e-06, "loss": 0.5947, "step": 6628 }, { "epoch": 0.48, "grad_norm": 3.9202726457328256, "learning_rate": 5.64044355269757e-06, "loss": 0.1413, "step": 6629 }, { "epoch": 0.48, "grad_norm": 16.4693340700882, "learning_rate": 5.639291785300631e-06, "loss": 1.2875, "step": 6630 }, { "epoch": 0.48, "grad_norm": 3.2565214939813902, "learning_rate": 5.638139983416323e-06, "loss": 0.0972, "step": 6631 }, { "epoch": 0.48, "grad_norm": 10.86714430812804, "learning_rate": 5.636988147106777e-06, "loss": 0.7185, "step": 6632 }, { "epoch": 0.48, "grad_norm": 3.648208181623289, "learning_rate": 5.635836276434131e-06, "loss": 0.2186, "step": 6633 }, { "epoch": 0.48, "grad_norm": 9.865313540726154, "learning_rate": 5.634684371460525e-06, "loss": 0.6497, "step": 6634 }, { "epoch": 0.48, "grad_norm": 4.739789967664149, "learning_rate": 5.6335324322481e-06, "loss": 1.0225, "step": 6635 }, { "epoch": 0.48, "grad_norm": 3.829300467227345, "learning_rate": 5.632380458858998e-06, "loss": 0.2795, "step": 6636 }, { "epoch": 0.48, "grad_norm": 3.678205179055894, "learning_rate": 5.631228451355364e-06, "loss": 0.2062, "step": 6637 }, { "epoch": 0.48, "grad_norm": 5.5857470541059175, "learning_rate": 5.630076409799346e-06, "loss": 0.8157, "step": 6638 }, { "epoch": 0.48, "grad_norm": 3.5328752920039843, "learning_rate": 5.6289243342530875e-06, "loss": 0.2148, "step": 6639 }, { "epoch": 0.48, "grad_norm": 20.606982700521336, "learning_rate": 5.627772224778743e-06, "loss": 0.4997, "step": 6640 }, { "epoch": 0.48, "grad_norm": 11.993488425845369, "learning_rate": 5.626620081438464e-06, "loss": 0.9596, "step": 6641 }, { "epoch": 0.48, "grad_norm": 4.8325746292139495, "learning_rate": 5.625467904294403e-06, "loss": 0.8359, "step": 6642 }, { "epoch": 0.48, "grad_norm": 4.036831575836535, "learning_rate": 5.624315693408718e-06, "loss": 0.7736, "step": 6643 }, { "epoch": 0.48, "grad_norm": 3.6201025942728196, "learning_rate": 5.623163448843562e-06, "loss": 0.2525, "step": 6644 }, { "epoch": 0.48, "grad_norm": 3.63472649962016, "learning_rate": 5.622011170661097e-06, "loss": 0.2185, "step": 6645 }, { "epoch": 0.48, "grad_norm": 13.74347849580758, "learning_rate": 5.620858858923484e-06, "loss": 1.0379, "step": 6646 }, { "epoch": 0.48, "grad_norm": 22.79486190510949, "learning_rate": 5.6197065136928865e-06, "loss": 0.9653, "step": 6647 }, { "epoch": 0.48, "grad_norm": 3.383816790940976, "learning_rate": 5.6185541350314665e-06, "loss": 0.2447, "step": 6648 }, { "epoch": 0.48, "grad_norm": 4.892319213457521, "learning_rate": 5.6174017230013944e-06, "loss": 0.838, "step": 6649 }, { "epoch": 0.48, "grad_norm": 11.063997128957595, "learning_rate": 5.616249277664836e-06, "loss": 0.9333, "step": 6650 }, { "epoch": 0.48, "grad_norm": 11.192152817350525, "learning_rate": 5.61509679908396e-06, "loss": 1.1686, "step": 6651 }, { "epoch": 0.48, "grad_norm": 18.75718014520043, "learning_rate": 5.613944287320942e-06, "loss": 0.9867, "step": 6652 }, { "epoch": 0.48, "grad_norm": 5.151767463179571, "learning_rate": 5.612791742437952e-06, "loss": 0.755, "step": 6653 }, { "epoch": 0.48, "grad_norm": 4.27892655971617, "learning_rate": 5.611639164497167e-06, "loss": 0.1588, "step": 6654 }, { "epoch": 0.48, "grad_norm": 3.994842030268188, "learning_rate": 5.610486553560767e-06, "loss": 0.2528, "step": 6655 }, { "epoch": 0.48, "grad_norm": 18.813932481168838, "learning_rate": 5.609333909690925e-06, "loss": 0.953, "step": 6656 }, { "epoch": 0.48, "grad_norm": 2.955395807428071, "learning_rate": 5.608181232949825e-06, "loss": 0.1311, "step": 6657 }, { "epoch": 0.48, "grad_norm": 3.932242323073669, "learning_rate": 5.607028523399652e-06, "loss": 0.184, "step": 6658 }, { "epoch": 0.48, "grad_norm": 15.198973142269804, "learning_rate": 5.605875781102585e-06, "loss": 0.7677, "step": 6659 }, { "epoch": 0.48, "grad_norm": 7.8643035876619445, "learning_rate": 5.6047230061208145e-06, "loss": 0.4863, "step": 6660 }, { "epoch": 0.48, "grad_norm": 4.431299000393727, "learning_rate": 5.603570198516529e-06, "loss": 0.7399, "step": 6661 }, { "epoch": 0.48, "grad_norm": 2.0081879881673284, "learning_rate": 5.602417358351914e-06, "loss": 0.128, "step": 6662 }, { "epoch": 0.48, "grad_norm": 4.56693124861312, "learning_rate": 5.601264485689162e-06, "loss": 0.3105, "step": 6663 }, { "epoch": 0.48, "grad_norm": 17.14769135300656, "learning_rate": 5.600111580590468e-06, "loss": 1.3738, "step": 6664 }, { "epoch": 0.48, "grad_norm": 20.218242417959218, "learning_rate": 5.598958643118024e-06, "loss": 0.9474, "step": 6665 }, { "epoch": 0.48, "grad_norm": 11.173896036808049, "learning_rate": 5.597805673334031e-06, "loss": 0.696, "step": 6666 }, { "epoch": 0.48, "grad_norm": 9.629198322145646, "learning_rate": 5.596652671300684e-06, "loss": 0.6174, "step": 6667 }, { "epoch": 0.48, "grad_norm": 3.042223224580712, "learning_rate": 5.595499637080183e-06, "loss": 0.1777, "step": 6668 }, { "epoch": 0.48, "grad_norm": 3.282320253618913, "learning_rate": 5.594346570734731e-06, "loss": 0.1884, "step": 6669 }, { "epoch": 0.48, "grad_norm": 20.29374807709095, "learning_rate": 5.593193472326531e-06, "loss": 0.8658, "step": 6670 }, { "epoch": 0.48, "grad_norm": 3.3341477939432003, "learning_rate": 5.592040341917788e-06, "loss": 0.242, "step": 6671 }, { "epoch": 0.48, "grad_norm": 3.127623152847583, "learning_rate": 5.590887179570711e-06, "loss": 0.1766, "step": 6672 }, { "epoch": 0.48, "grad_norm": 5.357160192517218, "learning_rate": 5.589733985347507e-06, "loss": 0.1393, "step": 6673 }, { "epoch": 0.48, "grad_norm": 9.923052717995258, "learning_rate": 5.588580759310385e-06, "loss": 0.6125, "step": 6674 }, { "epoch": 0.48, "grad_norm": 15.193917623234265, "learning_rate": 5.587427501521562e-06, "loss": 1.0557, "step": 6675 }, { "epoch": 0.48, "grad_norm": 23.319525670364758, "learning_rate": 5.5862742120432455e-06, "loss": 0.8593, "step": 6676 }, { "epoch": 0.48, "grad_norm": 3.9263160256363947, "learning_rate": 5.5851208909376564e-06, "loss": 0.3517, "step": 6677 }, { "epoch": 0.48, "grad_norm": 5.264768224756138, "learning_rate": 5.58396753826701e-06, "loss": 0.7292, "step": 6678 }, { "epoch": 0.48, "grad_norm": 32.02767240064429, "learning_rate": 5.582814154093526e-06, "loss": 0.6839, "step": 6679 }, { "epoch": 0.48, "grad_norm": 18.877932412952237, "learning_rate": 5.581660738479422e-06, "loss": 0.7408, "step": 6680 }, { "epoch": 0.48, "grad_norm": 3.5310839510085876, "learning_rate": 5.580507291486925e-06, "loss": 0.2691, "step": 6681 }, { "epoch": 0.48, "grad_norm": 3.511105583589038, "learning_rate": 5.579353813178255e-06, "loss": 0.1921, "step": 6682 }, { "epoch": 0.48, "grad_norm": 37.203920287102896, "learning_rate": 5.578200303615643e-06, "loss": 1.2711, "step": 6683 }, { "epoch": 0.48, "grad_norm": 18.66198723239489, "learning_rate": 5.57704676286131e-06, "loss": 0.8662, "step": 6684 }, { "epoch": 0.48, "grad_norm": 26.41149777475218, "learning_rate": 5.575893190977491e-06, "loss": 1.9268, "step": 6685 }, { "epoch": 0.48, "grad_norm": 13.82833435779416, "learning_rate": 5.574739588026412e-06, "loss": 1.1474, "step": 6686 }, { "epoch": 0.48, "grad_norm": 4.502652767602151, "learning_rate": 5.573585954070309e-06, "loss": 0.9136, "step": 6687 }, { "epoch": 0.48, "grad_norm": 12.112186995058103, "learning_rate": 5.572432289171415e-06, "loss": 0.9152, "step": 6688 }, { "epoch": 0.48, "grad_norm": 4.85713517059971, "learning_rate": 5.571278593391965e-06, "loss": 0.8979, "step": 6689 }, { "epoch": 0.48, "grad_norm": 2.7588098296233348, "learning_rate": 5.570124866794196e-06, "loss": 0.1887, "step": 6690 }, { "epoch": 0.48, "grad_norm": 2.9169805916742635, "learning_rate": 5.568971109440351e-06, "loss": 0.1771, "step": 6691 }, { "epoch": 0.48, "grad_norm": 2.8216491641315486, "learning_rate": 5.5678173213926674e-06, "loss": 0.1889, "step": 6692 }, { "epoch": 0.48, "grad_norm": 2.993189420841952, "learning_rate": 5.5666635027133885e-06, "loss": 0.2258, "step": 6693 }, { "epoch": 0.48, "grad_norm": 3.7918279084703945, "learning_rate": 5.565509653464759e-06, "loss": 0.7893, "step": 6694 }, { "epoch": 0.48, "grad_norm": 21.891860777641817, "learning_rate": 5.564355773709023e-06, "loss": 0.9411, "step": 6695 }, { "epoch": 0.48, "grad_norm": 14.822465370986068, "learning_rate": 5.563201863508431e-06, "loss": 0.7901, "step": 6696 }, { "epoch": 0.48, "grad_norm": 4.309862650135718, "learning_rate": 5.56204792292523e-06, "loss": 0.2314, "step": 6697 }, { "epoch": 0.48, "grad_norm": 14.493187018483686, "learning_rate": 5.560893952021671e-06, "loss": 0.7384, "step": 6698 }, { "epoch": 0.48, "grad_norm": 4.868916207166774, "learning_rate": 5.559739950860005e-06, "loss": 1.087, "step": 6699 }, { "epoch": 0.48, "grad_norm": 2.9622196855436522, "learning_rate": 5.558585919502489e-06, "loss": 0.219, "step": 6700 }, { "epoch": 0.48, "grad_norm": 3.5621016421782064, "learning_rate": 5.557431858011375e-06, "loss": 0.2736, "step": 6701 }, { "epoch": 0.48, "grad_norm": 3.2168862557515556, "learning_rate": 5.556277766448923e-06, "loss": 0.1649, "step": 6702 }, { "epoch": 0.48, "grad_norm": 40.66268163500197, "learning_rate": 5.5551236448773925e-06, "loss": 0.8027, "step": 6703 }, { "epoch": 0.48, "grad_norm": 3.3048191454219356, "learning_rate": 5.553969493359043e-06, "loss": 0.1468, "step": 6704 }, { "epoch": 0.48, "grad_norm": 3.688718336059233, "learning_rate": 5.552815311956133e-06, "loss": 0.1923, "step": 6705 }, { "epoch": 0.48, "grad_norm": 17.605945712511417, "learning_rate": 5.551661100730933e-06, "loss": 1.1118, "step": 6706 }, { "epoch": 0.48, "grad_norm": 4.43591174980654, "learning_rate": 5.5505068597457024e-06, "loss": 0.7532, "step": 6707 }, { "epoch": 0.48, "grad_norm": 19.980871402341375, "learning_rate": 5.5493525890627105e-06, "loss": 0.9433, "step": 6708 }, { "epoch": 0.48, "grad_norm": 10.944123134818494, "learning_rate": 5.548198288744229e-06, "loss": 0.3526, "step": 6709 }, { "epoch": 0.48, "grad_norm": 17.274676721505337, "learning_rate": 5.547043958852522e-06, "loss": 0.9421, "step": 6710 }, { "epoch": 0.48, "grad_norm": 3.290303384305284, "learning_rate": 5.545889599449863e-06, "loss": 0.191, "step": 6711 }, { "epoch": 0.48, "grad_norm": 5.282802889451811, "learning_rate": 5.544735210598529e-06, "loss": 1.1388, "step": 6712 }, { "epoch": 0.48, "grad_norm": 3.1817054340024273, "learning_rate": 5.543580792360791e-06, "loss": 0.1626, "step": 6713 }, { "epoch": 0.48, "grad_norm": 3.8336464713337652, "learning_rate": 5.542426344798926e-06, "loss": 0.192, "step": 6714 }, { "epoch": 0.48, "grad_norm": 15.95136408839894, "learning_rate": 5.541271867975216e-06, "loss": 1.0005, "step": 6715 }, { "epoch": 0.48, "grad_norm": 4.5508565656978455, "learning_rate": 5.540117361951934e-06, "loss": 0.9761, "step": 6716 }, { "epoch": 0.48, "grad_norm": 3.602353187794849, "learning_rate": 5.538962826791366e-06, "loss": 0.2269, "step": 6717 }, { "epoch": 0.48, "grad_norm": 30.344632530826416, "learning_rate": 5.537808262555795e-06, "loss": 0.7935, "step": 6718 }, { "epoch": 0.48, "grad_norm": 2.936160501616326, "learning_rate": 5.536653669307502e-06, "loss": 0.1671, "step": 6719 }, { "epoch": 0.48, "grad_norm": 4.298704078235068, "learning_rate": 5.535499047108776e-06, "loss": 0.9275, "step": 6720 }, { "epoch": 0.48, "grad_norm": 10.883424742579697, "learning_rate": 5.5343443960219045e-06, "loss": 0.5128, "step": 6721 }, { "epoch": 0.48, "grad_norm": 3.2528743573760273, "learning_rate": 5.533189716109173e-06, "loss": 0.1452, "step": 6722 }, { "epoch": 0.48, "grad_norm": 10.189901953058238, "learning_rate": 5.532035007432876e-06, "loss": 0.8877, "step": 6723 }, { "epoch": 0.48, "grad_norm": 9.042086909520926, "learning_rate": 5.530880270055304e-06, "loss": 0.5067, "step": 6724 }, { "epoch": 0.48, "grad_norm": 8.641653686377593, "learning_rate": 5.52972550403875e-06, "loss": 0.7323, "step": 6725 }, { "epoch": 0.48, "grad_norm": 3.121429508333902, "learning_rate": 5.528570709445511e-06, "loss": 0.2877, "step": 6726 }, { "epoch": 0.48, "grad_norm": 13.359766574208633, "learning_rate": 5.527415886337883e-06, "loss": 0.8541, "step": 6727 }, { "epoch": 0.48, "grad_norm": 4.703587753317415, "learning_rate": 5.526261034778164e-06, "loss": 0.2352, "step": 6728 }, { "epoch": 0.48, "grad_norm": 8.269425432378942, "learning_rate": 5.525106154828656e-06, "loss": 0.7837, "step": 6729 }, { "epoch": 0.48, "grad_norm": 4.609586460949753, "learning_rate": 5.523951246551657e-06, "loss": 0.3151, "step": 6730 }, { "epoch": 0.48, "grad_norm": 21.35296027343275, "learning_rate": 5.522796310009469e-06, "loss": 0.8727, "step": 6731 }, { "epoch": 0.48, "grad_norm": 3.8540571902727483, "learning_rate": 5.521641345264402e-06, "loss": 0.2246, "step": 6732 }, { "epoch": 0.48, "grad_norm": 4.449312766346891, "learning_rate": 5.52048635237876e-06, "loss": 0.823, "step": 6733 }, { "epoch": 0.48, "grad_norm": 6.948497095165328, "learning_rate": 5.519331331414846e-06, "loss": 0.8026, "step": 6734 }, { "epoch": 0.48, "grad_norm": 3.6932080698826146, "learning_rate": 5.518176282434976e-06, "loss": 0.2321, "step": 6735 }, { "epoch": 0.48, "grad_norm": 3.7639078139336783, "learning_rate": 5.517021205501455e-06, "loss": 0.1831, "step": 6736 }, { "epoch": 0.48, "grad_norm": 10.2871473851953, "learning_rate": 5.5158661006765975e-06, "loss": 0.6135, "step": 6737 }, { "epoch": 0.48, "grad_norm": 3.706468437653883, "learning_rate": 5.5147109680227165e-06, "loss": 0.1815, "step": 6738 }, { "epoch": 0.48, "grad_norm": 3.064050723822974, "learning_rate": 5.513555807602126e-06, "loss": 0.1415, "step": 6739 }, { "epoch": 0.48, "grad_norm": 4.178909317758739, "learning_rate": 5.512400619477144e-06, "loss": 0.9098, "step": 6740 }, { "epoch": 0.48, "grad_norm": 3.6727427891837348, "learning_rate": 5.511245403710089e-06, "loss": 0.3329, "step": 6741 }, { "epoch": 0.48, "grad_norm": 3.020442975842254, "learning_rate": 5.51009016036328e-06, "loss": 0.2155, "step": 6742 }, { "epoch": 0.48, "grad_norm": 3.285328306285278, "learning_rate": 5.508934889499037e-06, "loss": 0.2541, "step": 6743 }, { "epoch": 0.48, "grad_norm": 5.512772236474786, "learning_rate": 5.507779591179681e-06, "loss": 0.255, "step": 6744 }, { "epoch": 0.48, "grad_norm": 4.746022392133572, "learning_rate": 5.506624265467542e-06, "loss": 0.9572, "step": 6745 }, { "epoch": 0.48, "grad_norm": 37.903502875721045, "learning_rate": 5.5054689124249394e-06, "loss": 1.3307, "step": 6746 }, { "epoch": 0.48, "grad_norm": 4.369165101016096, "learning_rate": 5.504313532114202e-06, "loss": 0.1332, "step": 6747 }, { "epoch": 0.48, "grad_norm": 13.163575707721296, "learning_rate": 5.503158124597658e-06, "loss": 0.6721, "step": 6748 }, { "epoch": 0.48, "grad_norm": 4.1721128095398345, "learning_rate": 5.50200268993764e-06, "loss": 0.2112, "step": 6749 }, { "epoch": 0.48, "grad_norm": 4.289457430579459, "learning_rate": 5.500847228196475e-06, "loss": 0.8496, "step": 6750 }, { "epoch": 0.48, "grad_norm": 3.5342641634395644, "learning_rate": 5.499691739436498e-06, "loss": 0.2497, "step": 6751 }, { "epoch": 0.48, "grad_norm": 6.015818874291548, "learning_rate": 5.4985362237200455e-06, "loss": 0.4098, "step": 6752 }, { "epoch": 0.48, "grad_norm": 3.0831813097344707, "learning_rate": 5.497380681109449e-06, "loss": 0.1777, "step": 6753 }, { "epoch": 0.48, "grad_norm": 11.475225124038836, "learning_rate": 5.496225111667046e-06, "loss": 0.8243, "step": 6754 }, { "epoch": 0.48, "grad_norm": 2.597835320725184, "learning_rate": 5.495069515455179e-06, "loss": 0.1286, "step": 6755 }, { "epoch": 0.48, "grad_norm": 15.701912460001198, "learning_rate": 5.493913892536184e-06, "loss": 0.6741, "step": 6756 }, { "epoch": 0.48, "grad_norm": 18.90367927271999, "learning_rate": 5.492758242972405e-06, "loss": 0.8415, "step": 6757 }, { "epoch": 0.48, "grad_norm": 4.967892813514565, "learning_rate": 5.491602566826184e-06, "loss": 0.9613, "step": 6758 }, { "epoch": 0.48, "grad_norm": 4.091296179376066, "learning_rate": 5.490446864159865e-06, "loss": 0.2612, "step": 6759 }, { "epoch": 0.48, "grad_norm": 4.575563539695101, "learning_rate": 5.489291135035795e-06, "loss": 0.7846, "step": 6760 }, { "epoch": 0.48, "grad_norm": 4.07594611651278, "learning_rate": 5.488135379516318e-06, "loss": 0.6119, "step": 6761 }, { "epoch": 0.48, "grad_norm": 3.8734046908364865, "learning_rate": 5.486979597663786e-06, "loss": 0.2095, "step": 6762 }, { "epoch": 0.48, "grad_norm": 4.911917326646009, "learning_rate": 5.485823789540548e-06, "loss": 0.8246, "step": 6763 }, { "epoch": 0.49, "grad_norm": 11.390344382747067, "learning_rate": 5.484667955208958e-06, "loss": 0.5555, "step": 6764 }, { "epoch": 0.49, "grad_norm": 3.461714273651461, "learning_rate": 5.483512094731362e-06, "loss": 0.201, "step": 6765 }, { "epoch": 0.49, "grad_norm": 16.898697326228604, "learning_rate": 5.482356208170122e-06, "loss": 0.6398, "step": 6766 }, { "epoch": 0.49, "grad_norm": 3.2877661957839526, "learning_rate": 5.48120029558759e-06, "loss": 0.144, "step": 6767 }, { "epoch": 0.49, "grad_norm": 3.6795663122690483, "learning_rate": 5.480044357046122e-06, "loss": 0.2738, "step": 6768 }, { "epoch": 0.49, "grad_norm": 13.336977198244442, "learning_rate": 5.4788883926080794e-06, "loss": 0.5858, "step": 6769 }, { "epoch": 0.49, "grad_norm": 28.759152563021186, "learning_rate": 5.47773240233582e-06, "loss": 0.7848, "step": 6770 }, { "epoch": 0.49, "grad_norm": 7.794934725027362, "learning_rate": 5.476576386291705e-06, "loss": 0.422, "step": 6771 }, { "epoch": 0.49, "grad_norm": 7.397505630012401, "learning_rate": 5.475420344538101e-06, "loss": 0.3425, "step": 6772 }, { "epoch": 0.49, "grad_norm": 2.6101680416405104, "learning_rate": 5.474264277137365e-06, "loss": 0.1346, "step": 6773 }, { "epoch": 0.49, "grad_norm": 3.553590885067085, "learning_rate": 5.47310818415187e-06, "loss": 0.2645, "step": 6774 }, { "epoch": 0.49, "grad_norm": 2.827026351893106, "learning_rate": 5.471952065643976e-06, "loss": 0.1451, "step": 6775 }, { "epoch": 0.49, "grad_norm": 4.201018558176043, "learning_rate": 5.470795921676056e-06, "loss": 1.1112, "step": 6776 }, { "epoch": 0.49, "grad_norm": 9.902053759077173, "learning_rate": 5.4696397523104786e-06, "loss": 0.7419, "step": 6777 }, { "epoch": 0.49, "grad_norm": 3.8470368883905413, "learning_rate": 5.468483557609614e-06, "loss": 0.8472, "step": 6778 }, { "epoch": 0.49, "grad_norm": 3.5535648226175645, "learning_rate": 5.4673273376358334e-06, "loss": 0.2041, "step": 6779 }, { "epoch": 0.49, "grad_norm": 4.214121463143138, "learning_rate": 5.466171092451514e-06, "loss": 0.8596, "step": 6780 }, { "epoch": 0.49, "grad_norm": 2.9809952396907247, "learning_rate": 5.465014822119028e-06, "loss": 0.1614, "step": 6781 }, { "epoch": 0.49, "grad_norm": 23.033318427773658, "learning_rate": 5.463858526700753e-06, "loss": 1.0866, "step": 6782 }, { "epoch": 0.49, "grad_norm": 4.790386015108892, "learning_rate": 5.462702206259065e-06, "loss": 0.2371, "step": 6783 }, { "epoch": 0.49, "grad_norm": 19.78213948107512, "learning_rate": 5.4615458608563444e-06, "loss": 0.6236, "step": 6784 }, { "epoch": 0.49, "grad_norm": 10.284108611484756, "learning_rate": 5.460389490554972e-06, "loss": 0.4582, "step": 6785 }, { "epoch": 0.49, "grad_norm": 3.3840811157228425, "learning_rate": 5.45923309541733e-06, "loss": 0.2508, "step": 6786 }, { "epoch": 0.49, "grad_norm": 4.7861426423973725, "learning_rate": 5.4580766755058e-06, "loss": 0.9342, "step": 6787 }, { "epoch": 0.49, "grad_norm": 2.299019852988881, "learning_rate": 5.456920230882768e-06, "loss": 0.1235, "step": 6788 }, { "epoch": 0.49, "grad_norm": 2.3416910551240773, "learning_rate": 5.455763761610618e-06, "loss": 0.118, "step": 6789 }, { "epoch": 0.49, "grad_norm": 40.10688109344811, "learning_rate": 5.454607267751739e-06, "loss": 1.452, "step": 6790 }, { "epoch": 0.49, "grad_norm": 8.396259133439266, "learning_rate": 5.453450749368518e-06, "loss": 0.4486, "step": 6791 }, { "epoch": 0.49, "grad_norm": 14.9787426839127, "learning_rate": 5.452294206523347e-06, "loss": 1.0018, "step": 6792 }, { "epoch": 0.49, "grad_norm": 3.4910964751184377, "learning_rate": 5.451137639278613e-06, "loss": 0.1621, "step": 6793 }, { "epoch": 0.49, "grad_norm": 3.128257508296302, "learning_rate": 5.449981047696713e-06, "loss": 0.1853, "step": 6794 }, { "epoch": 0.49, "grad_norm": 5.250848687304666, "learning_rate": 5.4488244318400395e-06, "loss": 0.2213, "step": 6795 }, { "epoch": 0.49, "grad_norm": 12.029465086229457, "learning_rate": 5.447667791770986e-06, "loss": 0.8854, "step": 6796 }, { "epoch": 0.49, "grad_norm": 4.104927162688115, "learning_rate": 5.446511127551949e-06, "loss": 0.1973, "step": 6797 }, { "epoch": 0.49, "grad_norm": 4.398894126496259, "learning_rate": 5.445354439245326e-06, "loss": 0.9766, "step": 6798 }, { "epoch": 0.49, "grad_norm": 4.408613627020539, "learning_rate": 5.444197726913517e-06, "loss": 0.7643, "step": 6799 }, { "epoch": 0.49, "grad_norm": 61.200237802093284, "learning_rate": 5.443040990618925e-06, "loss": 0.5966, "step": 6800 }, { "epoch": 0.49, "grad_norm": 19.636802456132646, "learning_rate": 5.441884230423946e-06, "loss": 0.6971, "step": 6801 }, { "epoch": 0.49, "grad_norm": 27.6211941063491, "learning_rate": 5.440727446390985e-06, "loss": 0.6689, "step": 6802 }, { "epoch": 0.49, "grad_norm": 10.711372748454433, "learning_rate": 5.439570638582448e-06, "loss": 0.6122, "step": 6803 }, { "epoch": 0.49, "grad_norm": 3.678402855158047, "learning_rate": 5.4384138070607374e-06, "loss": 0.1717, "step": 6804 }, { "epoch": 0.49, "grad_norm": 4.34615812151745, "learning_rate": 5.437256951888262e-06, "loss": 0.8996, "step": 6805 }, { "epoch": 0.49, "grad_norm": 3.1154376792986826, "learning_rate": 5.43610007312743e-06, "loss": 0.2237, "step": 6806 }, { "epoch": 0.49, "grad_norm": 3.048501185941304, "learning_rate": 5.43494317084065e-06, "loss": 0.1291, "step": 6807 }, { "epoch": 0.49, "grad_norm": 3.1569988049660056, "learning_rate": 5.43378624509033e-06, "loss": 0.2384, "step": 6808 }, { "epoch": 0.49, "grad_norm": 2.4909841771568475, "learning_rate": 5.432629295938886e-06, "loss": 0.173, "step": 6809 }, { "epoch": 0.49, "grad_norm": 16.046355694840006, "learning_rate": 5.4314723234487285e-06, "loss": 1.2131, "step": 6810 }, { "epoch": 0.49, "grad_norm": 3.485816134253986, "learning_rate": 5.430315327682273e-06, "loss": 0.1692, "step": 6811 }, { "epoch": 0.49, "grad_norm": 3.8844312186512164, "learning_rate": 5.429158308701934e-06, "loss": 0.1906, "step": 6812 }, { "epoch": 0.49, "grad_norm": 13.111997792601477, "learning_rate": 5.428001266570127e-06, "loss": 0.6859, "step": 6813 }, { "epoch": 0.49, "grad_norm": 3.5889015176766654, "learning_rate": 5.426844201349275e-06, "loss": 0.1705, "step": 6814 }, { "epoch": 0.49, "grad_norm": 2.9504558405926025, "learning_rate": 5.425687113101792e-06, "loss": 0.1081, "step": 6815 }, { "epoch": 0.49, "grad_norm": 4.474363579854776, "learning_rate": 5.4245300018901005e-06, "loss": 0.8785, "step": 6816 }, { "epoch": 0.49, "grad_norm": 11.478647039460114, "learning_rate": 5.423372867776623e-06, "loss": 1.013, "step": 6817 }, { "epoch": 0.49, "grad_norm": 5.191809564128324, "learning_rate": 5.422215710823782e-06, "loss": 0.3509, "step": 6818 }, { "epoch": 0.49, "grad_norm": 10.328714477860657, "learning_rate": 5.4210585310940004e-06, "loss": 0.7908, "step": 6819 }, { "epoch": 0.49, "grad_norm": 4.560366155083396, "learning_rate": 5.419901328649706e-06, "loss": 0.2793, "step": 6820 }, { "epoch": 0.49, "grad_norm": 15.08933026619589, "learning_rate": 5.4187441035533225e-06, "loss": 0.9532, "step": 6821 }, { "epoch": 0.49, "grad_norm": 30.527108819913845, "learning_rate": 5.41758685586728e-06, "loss": 0.7366, "step": 6822 }, { "epoch": 0.49, "grad_norm": 2.8994562403126745, "learning_rate": 5.41642958565401e-06, "loss": 0.2033, "step": 6823 }, { "epoch": 0.49, "grad_norm": 3.382349752912877, "learning_rate": 5.415272292975938e-06, "loss": 0.1932, "step": 6824 }, { "epoch": 0.49, "grad_norm": 19.837031348183928, "learning_rate": 5.4141149778954985e-06, "loss": 0.7524, "step": 6825 }, { "epoch": 0.49, "grad_norm": 3.369881424586303, "learning_rate": 5.412957640475124e-06, "loss": 0.2501, "step": 6826 }, { "epoch": 0.49, "grad_norm": 15.270429436416945, "learning_rate": 5.4118002807772465e-06, "loss": 0.9509, "step": 6827 }, { "epoch": 0.49, "grad_norm": 3.4643281684825378, "learning_rate": 5.410642898864303e-06, "loss": 0.1183, "step": 6828 }, { "epoch": 0.49, "grad_norm": 3.796955584884288, "learning_rate": 5.409485494798729e-06, "loss": 0.3518, "step": 6829 }, { "epoch": 0.49, "grad_norm": 3.5091697222668237, "learning_rate": 5.408328068642964e-06, "loss": 0.1599, "step": 6830 }, { "epoch": 0.49, "grad_norm": 4.910201030216575, "learning_rate": 5.407170620459443e-06, "loss": 0.9995, "step": 6831 }, { "epoch": 0.49, "grad_norm": 4.537137367761275, "learning_rate": 5.40601315031061e-06, "loss": 0.8928, "step": 6832 }, { "epoch": 0.49, "grad_norm": 4.125399507907067, "learning_rate": 5.404855658258904e-06, "loss": 0.8376, "step": 6833 }, { "epoch": 0.49, "grad_norm": 2.6669096135345742, "learning_rate": 5.403698144366768e-06, "loss": 0.1075, "step": 6834 }, { "epoch": 0.49, "grad_norm": 9.124765468820158, "learning_rate": 5.402540608696645e-06, "loss": 0.4957, "step": 6835 }, { "epoch": 0.49, "grad_norm": 2.905468331110728, "learning_rate": 5.40138305131098e-06, "loss": 0.1797, "step": 6836 }, { "epoch": 0.49, "grad_norm": 3.316014855213255, "learning_rate": 5.400225472272219e-06, "loss": 0.1827, "step": 6837 }, { "epoch": 0.49, "grad_norm": 3.047637336653636, "learning_rate": 5.399067871642808e-06, "loss": 0.1539, "step": 6838 }, { "epoch": 0.49, "grad_norm": 2.890666525042696, "learning_rate": 5.3979102494851955e-06, "loss": 0.127, "step": 6839 }, { "epoch": 0.49, "grad_norm": 10.69964680654664, "learning_rate": 5.396752605861831e-06, "loss": 0.5257, "step": 6840 }, { "epoch": 0.49, "grad_norm": 14.728117594016355, "learning_rate": 5.395594940835166e-06, "loss": 0.8244, "step": 6841 }, { "epoch": 0.49, "grad_norm": 16.524158512004316, "learning_rate": 5.394437254467653e-06, "loss": 0.7743, "step": 6842 }, { "epoch": 0.49, "grad_norm": 25.374666692460035, "learning_rate": 5.393279546821741e-06, "loss": 1.4259, "step": 6843 }, { "epoch": 0.49, "grad_norm": 3.8868097436429068, "learning_rate": 5.392121817959887e-06, "loss": 0.2409, "step": 6844 }, { "epoch": 0.49, "grad_norm": 3.47197170629648, "learning_rate": 5.390964067944543e-06, "loss": 0.1897, "step": 6845 }, { "epoch": 0.49, "grad_norm": 5.002364340498406, "learning_rate": 5.38980629683817e-06, "loss": 0.8614, "step": 6846 }, { "epoch": 0.49, "grad_norm": 3.906740838350072, "learning_rate": 5.388648504703221e-06, "loss": 0.1856, "step": 6847 }, { "epoch": 0.49, "grad_norm": 4.379536465374832, "learning_rate": 5.387490691602158e-06, "loss": 0.2727, "step": 6848 }, { "epoch": 0.49, "grad_norm": 5.181104478168938, "learning_rate": 5.386332857597439e-06, "loss": 0.9012, "step": 6849 }, { "epoch": 0.49, "grad_norm": 17.39212071586119, "learning_rate": 5.3851750027515235e-06, "loss": 0.9027, "step": 6850 }, { "epoch": 0.49, "grad_norm": 5.47234714125589, "learning_rate": 5.384017127126875e-06, "loss": 1.1284, "step": 6851 }, { "epoch": 0.49, "grad_norm": 3.8827648020097296, "learning_rate": 5.382859230785955e-06, "loss": 0.2853, "step": 6852 }, { "epoch": 0.49, "grad_norm": 7.539535200037921, "learning_rate": 5.38170131379123e-06, "loss": 0.5138, "step": 6853 }, { "epoch": 0.49, "grad_norm": 5.5769183931425585, "learning_rate": 5.380543376205165e-06, "loss": 0.3502, "step": 6854 }, { "epoch": 0.49, "grad_norm": 2.567538629304978, "learning_rate": 5.3793854180902225e-06, "loss": 0.1864, "step": 6855 }, { "epoch": 0.49, "grad_norm": 3.364906192295331, "learning_rate": 5.378227439508874e-06, "loss": 0.2128, "step": 6856 }, { "epoch": 0.49, "grad_norm": 2.6780510593913003, "learning_rate": 5.377069440523588e-06, "loss": 0.1547, "step": 6857 }, { "epoch": 0.49, "grad_norm": 2.922145625059473, "learning_rate": 5.375911421196831e-06, "loss": 0.1959, "step": 6858 }, { "epoch": 0.49, "grad_norm": 9.422688778809206, "learning_rate": 5.374753381591076e-06, "loss": 0.5524, "step": 6859 }, { "epoch": 0.49, "grad_norm": 5.031515252271272, "learning_rate": 5.3735953217687976e-06, "loss": 0.8565, "step": 6860 }, { "epoch": 0.49, "grad_norm": 2.095362271910932, "learning_rate": 5.372437241792462e-06, "loss": 0.121, "step": 6861 }, { "epoch": 0.49, "grad_norm": 2.886564385353262, "learning_rate": 5.371279141724548e-06, "loss": 0.1697, "step": 6862 }, { "epoch": 0.49, "grad_norm": 18.751419045896906, "learning_rate": 5.37012102162753e-06, "loss": 0.7572, "step": 6863 }, { "epoch": 0.49, "grad_norm": 3.6765504044234745, "learning_rate": 5.3689628815638835e-06, "loss": 0.2395, "step": 6864 }, { "epoch": 0.49, "grad_norm": 16.374332440714095, "learning_rate": 5.367804721596087e-06, "loss": 0.8288, "step": 6865 }, { "epoch": 0.49, "grad_norm": 27.925767453387923, "learning_rate": 5.366646541786619e-06, "loss": 1.7277, "step": 6866 }, { "epoch": 0.49, "grad_norm": 4.699813884360485, "learning_rate": 5.365488342197955e-06, "loss": 0.2794, "step": 6867 }, { "epoch": 0.49, "grad_norm": 3.5857220596704784, "learning_rate": 5.364330122892581e-06, "loss": 0.1326, "step": 6868 }, { "epoch": 0.49, "grad_norm": 4.034855629922351, "learning_rate": 5.363171883932974e-06, "loss": 0.095, "step": 6869 }, { "epoch": 0.49, "grad_norm": 3.9614182634689454, "learning_rate": 5.362013625381618e-06, "loss": 0.2193, "step": 6870 }, { "epoch": 0.49, "grad_norm": 3.8481965031382073, "learning_rate": 5.360855347301e-06, "loss": 0.2012, "step": 6871 }, { "epoch": 0.49, "grad_norm": 14.65099647240415, "learning_rate": 5.3596970497536004e-06, "loss": 0.5727, "step": 6872 }, { "epoch": 0.49, "grad_norm": 12.332546441332106, "learning_rate": 5.3585387328019055e-06, "loss": 0.6858, "step": 6873 }, { "epoch": 0.49, "grad_norm": 4.521580205292323, "learning_rate": 5.357380396508405e-06, "loss": 0.9501, "step": 6874 }, { "epoch": 0.49, "grad_norm": 4.103964232004236, "learning_rate": 5.3562220409355834e-06, "loss": 0.2808, "step": 6875 }, { "epoch": 0.49, "grad_norm": 2.667777195439538, "learning_rate": 5.355063666145931e-06, "loss": 0.1088, "step": 6876 }, { "epoch": 0.49, "grad_norm": 3.408262940905798, "learning_rate": 5.353905272201939e-06, "loss": 0.2136, "step": 6877 }, { "epoch": 0.49, "grad_norm": 4.429924944533404, "learning_rate": 5.352746859166097e-06, "loss": 0.8853, "step": 6878 }, { "epoch": 0.49, "grad_norm": 3.1173787500979975, "learning_rate": 5.351588427100897e-06, "loss": 0.1965, "step": 6879 }, { "epoch": 0.49, "grad_norm": 13.715793225895274, "learning_rate": 5.3504299760688315e-06, "loss": 0.9941, "step": 6880 }, { "epoch": 0.49, "grad_norm": 3.867915473868172, "learning_rate": 5.349271506132395e-06, "loss": 0.2762, "step": 6881 }, { "epoch": 0.49, "grad_norm": 3.159644851058857, "learning_rate": 5.348113017354084e-06, "loss": 0.1196, "step": 6882 }, { "epoch": 0.49, "grad_norm": 17.186247839783356, "learning_rate": 5.346954509796393e-06, "loss": 1.049, "step": 6883 }, { "epoch": 0.49, "grad_norm": 5.261866287263958, "learning_rate": 5.34579598352182e-06, "loss": 0.1942, "step": 6884 }, { "epoch": 0.49, "grad_norm": 3.78878194034738, "learning_rate": 5.344637438592861e-06, "loss": 0.1706, "step": 6885 }, { "epoch": 0.49, "grad_norm": 3.636706676635323, "learning_rate": 5.343478875072018e-06, "loss": 0.2044, "step": 6886 }, { "epoch": 0.49, "grad_norm": 4.021539499684146, "learning_rate": 5.34232029302179e-06, "loss": 0.256, "step": 6887 }, { "epoch": 0.49, "grad_norm": 10.693913825041141, "learning_rate": 5.341161692504677e-06, "loss": 0.5166, "step": 6888 }, { "epoch": 0.49, "grad_norm": 2.8391846256529076, "learning_rate": 5.340003073583182e-06, "loss": 0.2562, "step": 6889 }, { "epoch": 0.49, "grad_norm": 4.589467653732257, "learning_rate": 5.33884443631981e-06, "loss": 0.9715, "step": 6890 }, { "epoch": 0.49, "grad_norm": 4.755392367021096, "learning_rate": 5.337685780777062e-06, "loss": 0.9779, "step": 6891 }, { "epoch": 0.49, "grad_norm": 4.01175034230643, "learning_rate": 5.336527107017444e-06, "loss": 0.2532, "step": 6892 }, { "epoch": 0.49, "grad_norm": 3.420121452996098, "learning_rate": 5.335368415103461e-06, "loss": 0.3048, "step": 6893 }, { "epoch": 0.49, "grad_norm": 3.108942122360679, "learning_rate": 5.334209705097625e-06, "loss": 0.1944, "step": 6894 }, { "epoch": 0.49, "grad_norm": 3.2842299731288724, "learning_rate": 5.3330509770624375e-06, "loss": 0.1802, "step": 6895 }, { "epoch": 0.49, "grad_norm": 4.049496432300884, "learning_rate": 5.331892231060411e-06, "loss": 0.9366, "step": 6896 }, { "epoch": 0.49, "grad_norm": 2.8948252670180077, "learning_rate": 5.330733467154057e-06, "loss": 0.1503, "step": 6897 }, { "epoch": 0.49, "grad_norm": 18.537940657123606, "learning_rate": 5.329574685405883e-06, "loss": 0.7607, "step": 6898 }, { "epoch": 0.49, "grad_norm": 3.6393297217663094, "learning_rate": 5.328415885878401e-06, "loss": 0.2033, "step": 6899 }, { "epoch": 0.49, "grad_norm": 20.958615156948575, "learning_rate": 5.3272570686341265e-06, "loss": 0.6859, "step": 6900 }, { "epoch": 0.49, "grad_norm": 3.3133845476009736, "learning_rate": 5.326098233735571e-06, "loss": 0.2112, "step": 6901 }, { "epoch": 0.49, "grad_norm": 11.847567784233881, "learning_rate": 5.324939381245251e-06, "loss": 0.7588, "step": 6902 }, { "epoch": 0.49, "grad_norm": 4.887394632713505, "learning_rate": 5.323780511225682e-06, "loss": 0.8922, "step": 6903 }, { "epoch": 0.5, "grad_norm": 2.6986726146777396, "learning_rate": 5.322621623739378e-06, "loss": 0.1247, "step": 6904 }, { "epoch": 0.5, "grad_norm": 30.082479539634964, "learning_rate": 5.32146271884886e-06, "loss": 1.0664, "step": 6905 }, { "epoch": 0.5, "grad_norm": 5.862689957816708, "learning_rate": 5.320303796616644e-06, "loss": 0.1783, "step": 6906 }, { "epoch": 0.5, "grad_norm": 4.035727268521343, "learning_rate": 5.319144857105251e-06, "loss": 0.1533, "step": 6907 }, { "epoch": 0.5, "grad_norm": 4.034354375165078, "learning_rate": 5.317985900377203e-06, "loss": 0.2218, "step": 6908 }, { "epoch": 0.5, "grad_norm": 5.584175132565356, "learning_rate": 5.316826926495016e-06, "loss": 0.8233, "step": 6909 }, { "epoch": 0.5, "grad_norm": 15.9647851216879, "learning_rate": 5.315667935521217e-06, "loss": 0.8447, "step": 6910 }, { "epoch": 0.5, "grad_norm": 3.0307794644762267, "learning_rate": 5.314508927518328e-06, "loss": 0.1512, "step": 6911 }, { "epoch": 0.5, "grad_norm": 4.732343695901795, "learning_rate": 5.313349902548872e-06, "loss": 0.9184, "step": 6912 }, { "epoch": 0.5, "grad_norm": 13.196736641825883, "learning_rate": 5.312190860675376e-06, "loss": 0.8766, "step": 6913 }, { "epoch": 0.5, "grad_norm": 2.4200956572161574, "learning_rate": 5.311031801960366e-06, "loss": 0.1153, "step": 6914 }, { "epoch": 0.5, "grad_norm": 4.8755062442270996, "learning_rate": 5.3098727264663655e-06, "loss": 1.1406, "step": 6915 }, { "epoch": 0.5, "grad_norm": 3.567024789865397, "learning_rate": 5.3087136342559045e-06, "loss": 0.2125, "step": 6916 }, { "epoch": 0.5, "grad_norm": 4.315242879227545, "learning_rate": 5.307554525391512e-06, "loss": 0.9326, "step": 6917 }, { "epoch": 0.5, "grad_norm": 4.625474176269083, "learning_rate": 5.306395399935715e-06, "loss": 0.8544, "step": 6918 }, { "epoch": 0.5, "grad_norm": 16.359558201009968, "learning_rate": 5.305236257951049e-06, "loss": 0.9542, "step": 6919 }, { "epoch": 0.5, "grad_norm": 14.2294459784113, "learning_rate": 5.3040770995000415e-06, "loss": 0.7087, "step": 6920 }, { "epoch": 0.5, "grad_norm": 5.000673304885199, "learning_rate": 5.302917924645225e-06, "loss": 0.9538, "step": 6921 }, { "epoch": 0.5, "grad_norm": 3.3774320461764082, "learning_rate": 5.301758733449132e-06, "loss": 0.1227, "step": 6922 }, { "epoch": 0.5, "grad_norm": 4.472670595171264, "learning_rate": 5.300599525974299e-06, "loss": 0.3835, "step": 6923 }, { "epoch": 0.5, "grad_norm": 2.7203012373470337, "learning_rate": 5.299440302283258e-06, "loss": 0.1608, "step": 6924 }, { "epoch": 0.5, "grad_norm": 12.336501932950203, "learning_rate": 5.298281062438549e-06, "loss": 0.7467, "step": 6925 }, { "epoch": 0.5, "grad_norm": 3.6454713010364537, "learning_rate": 5.297121806502703e-06, "loss": 0.2183, "step": 6926 }, { "epoch": 0.5, "grad_norm": 4.651181664887551, "learning_rate": 5.295962534538261e-06, "loss": 0.8448, "step": 6927 }, { "epoch": 0.5, "grad_norm": 4.990141735148222, "learning_rate": 5.294803246607761e-06, "loss": 0.8274, "step": 6928 }, { "epoch": 0.5, "grad_norm": 3.647437954380082, "learning_rate": 5.293643942773741e-06, "loss": 0.2089, "step": 6929 }, { "epoch": 0.5, "grad_norm": 4.691885224610084, "learning_rate": 5.2924846230987416e-06, "loss": 0.8317, "step": 6930 }, { "epoch": 0.5, "grad_norm": 3.381622115021278, "learning_rate": 5.291325287645305e-06, "loss": 0.1786, "step": 6931 }, { "epoch": 0.5, "grad_norm": 4.620097241154215, "learning_rate": 5.290165936475971e-06, "loss": 0.9904, "step": 6932 }, { "epoch": 0.5, "grad_norm": 3.930819283813645, "learning_rate": 5.289006569653284e-06, "loss": 0.2202, "step": 6933 }, { "epoch": 0.5, "grad_norm": 4.046410075739999, "learning_rate": 5.287847187239786e-06, "loss": 0.2499, "step": 6934 }, { "epoch": 0.5, "grad_norm": 4.017177632871193, "learning_rate": 5.28668778929802e-06, "loss": 0.741, "step": 6935 }, { "epoch": 0.5, "grad_norm": 3.2257920971155856, "learning_rate": 5.285528375890536e-06, "loss": 0.13, "step": 6936 }, { "epoch": 0.5, "grad_norm": 3.3990874938881075, "learning_rate": 5.284368947079875e-06, "loss": 0.1512, "step": 6937 }, { "epoch": 0.5, "grad_norm": 3.7763978558015703, "learning_rate": 5.283209502928587e-06, "loss": 0.1703, "step": 6938 }, { "epoch": 0.5, "grad_norm": 23.15010193263203, "learning_rate": 5.2820500434992174e-06, "loss": 1.1329, "step": 6939 }, { "epoch": 0.5, "grad_norm": 3.5482321626039193, "learning_rate": 5.280890568854315e-06, "loss": 0.1824, "step": 6940 }, { "epoch": 0.5, "grad_norm": 5.718276340722618, "learning_rate": 5.27973107905643e-06, "loss": 0.3007, "step": 6941 }, { "epoch": 0.5, "grad_norm": 3.6699873761096575, "learning_rate": 5.278571574168113e-06, "loss": 0.1862, "step": 6942 }, { "epoch": 0.5, "grad_norm": 17.93783665761314, "learning_rate": 5.277412054251914e-06, "loss": 1.0118, "step": 6943 }, { "epoch": 0.5, "grad_norm": 23.148348550379954, "learning_rate": 5.276252519370384e-06, "loss": 1.2529, "step": 6944 }, { "epoch": 0.5, "grad_norm": 4.02802096597489, "learning_rate": 5.275092969586077e-06, "loss": 0.2424, "step": 6945 }, { "epoch": 0.5, "grad_norm": 3.7167047561250444, "learning_rate": 5.273933404961544e-06, "loss": 0.1753, "step": 6946 }, { "epoch": 0.5, "grad_norm": 13.424186059237634, "learning_rate": 5.2727738255593416e-06, "loss": 1.0704, "step": 6947 }, { "epoch": 0.5, "grad_norm": 3.7843172740048523, "learning_rate": 5.271614231442025e-06, "loss": 0.1697, "step": 6948 }, { "epoch": 0.5, "grad_norm": 4.243448424067839, "learning_rate": 5.270454622672147e-06, "loss": 0.1957, "step": 6949 }, { "epoch": 0.5, "grad_norm": 4.830847745529671, "learning_rate": 5.269294999312267e-06, "loss": 0.1981, "step": 6950 }, { "epoch": 0.5, "grad_norm": 3.587996945659655, "learning_rate": 5.268135361424941e-06, "loss": 0.2692, "step": 6951 }, { "epoch": 0.5, "grad_norm": 18.740195375751565, "learning_rate": 5.266975709072727e-06, "loss": 1.1654, "step": 6952 }, { "epoch": 0.5, "grad_norm": 4.927904459175427, "learning_rate": 5.265816042318183e-06, "loss": 0.9266, "step": 6953 }, { "epoch": 0.5, "grad_norm": 15.78445482182261, "learning_rate": 5.264656361223871e-06, "loss": 1.141, "step": 6954 }, { "epoch": 0.5, "grad_norm": 3.387330438416869, "learning_rate": 5.263496665852349e-06, "loss": 0.1024, "step": 6955 }, { "epoch": 0.5, "grad_norm": 4.274534185315741, "learning_rate": 5.262336956266181e-06, "loss": 0.1972, "step": 6956 }, { "epoch": 0.5, "grad_norm": 4.044851620031372, "learning_rate": 5.261177232527926e-06, "loss": 0.2426, "step": 6957 }, { "epoch": 0.5, "grad_norm": 4.000346476452341, "learning_rate": 5.260017494700148e-06, "loss": 0.2659, "step": 6958 }, { "epoch": 0.5, "grad_norm": 16.532128189246116, "learning_rate": 5.258857742845412e-06, "loss": 0.6402, "step": 6959 }, { "epoch": 0.5, "grad_norm": 2.530478149249893, "learning_rate": 5.257697977026277e-06, "loss": 0.0959, "step": 6960 }, { "epoch": 0.5, "grad_norm": 3.9654139270280795, "learning_rate": 5.256538197305313e-06, "loss": 0.1889, "step": 6961 }, { "epoch": 0.5, "grad_norm": 3.3682927645386425, "learning_rate": 5.255378403745087e-06, "loss": 0.169, "step": 6962 }, { "epoch": 0.5, "grad_norm": 11.355465454889966, "learning_rate": 5.25421859640816e-06, "loss": 0.7348, "step": 6963 }, { "epoch": 0.5, "grad_norm": 22.028481505397025, "learning_rate": 5.253058775357103e-06, "loss": 1.0598, "step": 6964 }, { "epoch": 0.5, "grad_norm": 23.83510177820254, "learning_rate": 5.251898940654483e-06, "loss": 1.2149, "step": 6965 }, { "epoch": 0.5, "grad_norm": 11.25175780070956, "learning_rate": 5.2507390923628675e-06, "loss": 0.682, "step": 6966 }, { "epoch": 0.5, "grad_norm": 3.317604569438033, "learning_rate": 5.249579230544827e-06, "loss": 0.1528, "step": 6967 }, { "epoch": 0.5, "grad_norm": 2.9409655576028313, "learning_rate": 5.248419355262934e-06, "loss": 0.2323, "step": 6968 }, { "epoch": 0.5, "grad_norm": 18.71167846589017, "learning_rate": 5.247259466579755e-06, "loss": 0.8929, "step": 6969 }, { "epoch": 0.5, "grad_norm": 13.852514737463627, "learning_rate": 5.246099564557864e-06, "loss": 0.7541, "step": 6970 }, { "epoch": 0.5, "grad_norm": 12.680450217857471, "learning_rate": 5.244939649259835e-06, "loss": 0.5681, "step": 6971 }, { "epoch": 0.5, "grad_norm": 2.693070963970384, "learning_rate": 5.2437797207482365e-06, "loss": 0.1305, "step": 6972 }, { "epoch": 0.5, "grad_norm": 18.345698916800558, "learning_rate": 5.242619779085646e-06, "loss": 0.9871, "step": 6973 }, { "epoch": 0.5, "grad_norm": 3.1560099855610884, "learning_rate": 5.241459824334636e-06, "loss": 0.1295, "step": 6974 }, { "epoch": 0.5, "grad_norm": 7.848028339575585, "learning_rate": 5.240299856557783e-06, "loss": 0.6131, "step": 6975 }, { "epoch": 0.5, "grad_norm": 2.938317704141151, "learning_rate": 5.239139875817662e-06, "loss": 0.2003, "step": 6976 }, { "epoch": 0.5, "grad_norm": 3.861972019168753, "learning_rate": 5.237979882176851e-06, "loss": 0.3325, "step": 6977 }, { "epoch": 0.5, "grad_norm": 12.69366679152077, "learning_rate": 5.236819875697925e-06, "loss": 0.3812, "step": 6978 }, { "epoch": 0.5, "grad_norm": 4.538834835900775, "learning_rate": 5.235659856443465e-06, "loss": 0.2859, "step": 6979 }, { "epoch": 0.5, "grad_norm": 3.4271984547078165, "learning_rate": 5.234499824476045e-06, "loss": 0.1657, "step": 6980 }, { "epoch": 0.5, "grad_norm": 3.203004025005146, "learning_rate": 5.2333397798582484e-06, "loss": 0.1538, "step": 6981 }, { "epoch": 0.5, "grad_norm": 3.222902111475795, "learning_rate": 5.2321797226526545e-06, "loss": 0.2189, "step": 6982 }, { "epoch": 0.5, "grad_norm": 4.6029743814914585, "learning_rate": 5.231019652921843e-06, "loss": 0.2035, "step": 6983 }, { "epoch": 0.5, "grad_norm": 4.068332672355117, "learning_rate": 5.229859570728395e-06, "loss": 0.7422, "step": 6984 }, { "epoch": 0.5, "grad_norm": 3.4223383017645617, "learning_rate": 5.228699476134894e-06, "loss": 0.3081, "step": 6985 }, { "epoch": 0.5, "grad_norm": 12.523479449110424, "learning_rate": 5.227539369203921e-06, "loss": 0.774, "step": 6986 }, { "epoch": 0.5, "grad_norm": 3.3035682687724557, "learning_rate": 5.2263792499980615e-06, "loss": 0.1883, "step": 6987 }, { "epoch": 0.5, "grad_norm": 3.613724482070412, "learning_rate": 5.225219118579899e-06, "loss": 0.1568, "step": 6988 }, { "epoch": 0.5, "grad_norm": 4.23191407867427, "learning_rate": 5.224058975012016e-06, "loss": 0.9622, "step": 6989 }, { "epoch": 0.5, "grad_norm": 3.064208673821898, "learning_rate": 5.222898819356999e-06, "loss": 0.1961, "step": 6990 }, { "epoch": 0.5, "grad_norm": 4.30660416957115, "learning_rate": 5.221738651677436e-06, "loss": 0.9602, "step": 6991 }, { "epoch": 0.5, "grad_norm": 15.811187474551028, "learning_rate": 5.22057847203591e-06, "loss": 0.6873, "step": 6992 }, { "epoch": 0.5, "grad_norm": 17.407433835190627, "learning_rate": 5.219418280495013e-06, "loss": 1.4048, "step": 6993 }, { "epoch": 0.5, "grad_norm": 14.095952744372276, "learning_rate": 5.218258077117328e-06, "loss": 0.7957, "step": 6994 }, { "epoch": 0.5, "grad_norm": 6.792285677631567, "learning_rate": 5.2170978619654475e-06, "loss": 0.7898, "step": 6995 }, { "epoch": 0.5, "grad_norm": 3.4948265507739817, "learning_rate": 5.215937635101959e-06, "loss": 0.1501, "step": 6996 }, { "epoch": 0.5, "grad_norm": 3.696539538677145, "learning_rate": 5.214777396589451e-06, "loss": 0.264, "step": 6997 }, { "epoch": 0.5, "grad_norm": 25.102607961220414, "learning_rate": 5.2136171464905164e-06, "loss": 0.8467, "step": 6998 }, { "epoch": 0.5, "grad_norm": 3.430991517223482, "learning_rate": 5.212456884867747e-06, "loss": 0.2078, "step": 6999 }, { "epoch": 0.5, "grad_norm": 10.524158186560216, "learning_rate": 5.211296611783731e-06, "loss": 0.6983, "step": 7000 }, { "epoch": 0.5, "grad_norm": 4.405224008624078, "learning_rate": 5.2101363273010606e-06, "loss": 0.9021, "step": 7001 }, { "epoch": 0.5, "grad_norm": 3.408710322335176, "learning_rate": 5.208976031482335e-06, "loss": 0.1863, "step": 7002 }, { "epoch": 0.5, "grad_norm": 2.5416322659590938, "learning_rate": 5.207815724390139e-06, "loss": 0.1484, "step": 7003 }, { "epoch": 0.5, "grad_norm": 3.494231202168802, "learning_rate": 5.206655406087074e-06, "loss": 0.2929, "step": 7004 }, { "epoch": 0.5, "grad_norm": 19.55082791490232, "learning_rate": 5.205495076635734e-06, "loss": 0.713, "step": 7005 }, { "epoch": 0.5, "grad_norm": 23.498258142595954, "learning_rate": 5.20433473609871e-06, "loss": 1.1734, "step": 7006 }, { "epoch": 0.5, "grad_norm": 3.8056977882899905, "learning_rate": 5.203174384538599e-06, "loss": 0.2419, "step": 7007 }, { "epoch": 0.5, "grad_norm": 4.66147470004159, "learning_rate": 5.202014022018002e-06, "loss": 0.2895, "step": 7008 }, { "epoch": 0.5, "grad_norm": 3.3942530530832893, "learning_rate": 5.200853648599511e-06, "loss": 0.1736, "step": 7009 }, { "epoch": 0.5, "grad_norm": 3.619542195239556, "learning_rate": 5.199693264345728e-06, "loss": 0.2101, "step": 7010 }, { "epoch": 0.5, "grad_norm": 6.195786872920454, "learning_rate": 5.198532869319248e-06, "loss": 0.3098, "step": 7011 }, { "epoch": 0.5, "grad_norm": 4.59354072873171, "learning_rate": 5.197372463582671e-06, "loss": 0.9173, "step": 7012 }, { "epoch": 0.5, "grad_norm": 4.573297548290401, "learning_rate": 5.196212047198597e-06, "loss": 0.8181, "step": 7013 }, { "epoch": 0.5, "grad_norm": 14.413439534725168, "learning_rate": 5.195051620229625e-06, "loss": 0.9847, "step": 7014 }, { "epoch": 0.5, "grad_norm": 5.003699820414836, "learning_rate": 5.193891182738357e-06, "loss": 1.0255, "step": 7015 }, { "epoch": 0.5, "grad_norm": 4.023182053857012, "learning_rate": 5.1927307347873935e-06, "loss": 0.173, "step": 7016 }, { "epoch": 0.5, "grad_norm": 3.327394407148881, "learning_rate": 5.191570276439337e-06, "loss": 0.2534, "step": 7017 }, { "epoch": 0.5, "grad_norm": 3.778125774515026, "learning_rate": 5.190409807756789e-06, "loss": 0.2559, "step": 7018 }, { "epoch": 0.5, "grad_norm": 3.8279908755186187, "learning_rate": 5.189249328802352e-06, "loss": 0.2686, "step": 7019 }, { "epoch": 0.5, "grad_norm": 19.829405088352367, "learning_rate": 5.188088839638631e-06, "loss": 1.2428, "step": 7020 }, { "epoch": 0.5, "grad_norm": 11.248985514662085, "learning_rate": 5.186928340328228e-06, "loss": 0.9674, "step": 7021 }, { "epoch": 0.5, "grad_norm": 11.212497775901287, "learning_rate": 5.185767830933749e-06, "loss": 0.5489, "step": 7022 }, { "epoch": 0.5, "grad_norm": 3.539501056373005, "learning_rate": 5.1846073115178e-06, "loss": 0.2597, "step": 7023 }, { "epoch": 0.5, "grad_norm": 3.7148141631782448, "learning_rate": 5.183446782142983e-06, "loss": 0.1974, "step": 7024 }, { "epoch": 0.5, "grad_norm": 4.019598313027799, "learning_rate": 5.182286242871908e-06, "loss": 0.8259, "step": 7025 }, { "epoch": 0.5, "grad_norm": 61.40361052116273, "learning_rate": 5.181125693767179e-06, "loss": 1.06, "step": 7026 }, { "epoch": 0.5, "grad_norm": 12.591573190970783, "learning_rate": 5.179965134891406e-06, "loss": 0.7669, "step": 7027 }, { "epoch": 0.5, "grad_norm": 4.873677878345933, "learning_rate": 5.1788045663071936e-06, "loss": 0.8225, "step": 7028 }, { "epoch": 0.5, "grad_norm": 4.604610894591143, "learning_rate": 5.177643988077154e-06, "loss": 0.9464, "step": 7029 }, { "epoch": 0.5, "grad_norm": 2.954137753014184, "learning_rate": 5.176483400263891e-06, "loss": 0.1047, "step": 7030 }, { "epoch": 0.5, "grad_norm": 3.7328993847354006, "learning_rate": 5.175322802930018e-06, "loss": 0.2358, "step": 7031 }, { "epoch": 0.5, "grad_norm": 3.4765941534132554, "learning_rate": 5.174162196138143e-06, "loss": 0.2293, "step": 7032 }, { "epoch": 0.5, "grad_norm": 3.4179981937556603, "learning_rate": 5.173001579950877e-06, "loss": 0.1918, "step": 7033 }, { "epoch": 0.5, "grad_norm": 12.287083107727913, "learning_rate": 5.17184095443083e-06, "loss": 0.899, "step": 7034 }, { "epoch": 0.5, "grad_norm": 3.3084422363468162, "learning_rate": 5.170680319640615e-06, "loss": 0.1775, "step": 7035 }, { "epoch": 0.5, "grad_norm": 3.034707306831654, "learning_rate": 5.1695196756428435e-06, "loss": 0.1823, "step": 7036 }, { "epoch": 0.5, "grad_norm": 25.22999013517966, "learning_rate": 5.168359022500126e-06, "loss": 1.1418, "step": 7037 }, { "epoch": 0.5, "grad_norm": 4.293972243846583, "learning_rate": 5.167198360275075e-06, "loss": 0.1784, "step": 7038 }, { "epoch": 0.5, "grad_norm": 15.066557327434772, "learning_rate": 5.166037689030308e-06, "loss": 0.7651, "step": 7039 }, { "epoch": 0.5, "grad_norm": 4.83103407144188, "learning_rate": 5.164877008828434e-06, "loss": 0.7308, "step": 7040 }, { "epoch": 0.5, "grad_norm": 3.350979381531514, "learning_rate": 5.163716319732071e-06, "loss": 0.1567, "step": 7041 }, { "epoch": 0.5, "grad_norm": 3.2402691126344405, "learning_rate": 5.162555621803832e-06, "loss": 0.2149, "step": 7042 }, { "epoch": 0.51, "grad_norm": 20.26957670078242, "learning_rate": 5.161394915106331e-06, "loss": 1.3246, "step": 7043 }, { "epoch": 0.51, "grad_norm": 10.361964147477565, "learning_rate": 5.160234199702186e-06, "loss": 0.6285, "step": 7044 }, { "epoch": 0.51, "grad_norm": 3.207508157904792, "learning_rate": 5.159073475654014e-06, "loss": 0.2151, "step": 7045 }, { "epoch": 0.51, "grad_norm": 4.65673031468508, "learning_rate": 5.157912743024428e-06, "loss": 0.9628, "step": 7046 }, { "epoch": 0.51, "grad_norm": 3.410192085922375, "learning_rate": 5.156752001876048e-06, "loss": 0.1436, "step": 7047 }, { "epoch": 0.51, "grad_norm": 5.3223460014762525, "learning_rate": 5.155591252271492e-06, "loss": 1.0947, "step": 7048 }, { "epoch": 0.51, "grad_norm": 27.016005423204927, "learning_rate": 5.154430494273376e-06, "loss": 0.7141, "step": 7049 }, { "epoch": 0.51, "grad_norm": 3.372909890310097, "learning_rate": 5.15326972794432e-06, "loss": 0.15, "step": 7050 }, { "epoch": 0.51, "grad_norm": 3.451778919792542, "learning_rate": 5.152108953346941e-06, "loss": 0.287, "step": 7051 }, { "epoch": 0.51, "grad_norm": 4.300164228550873, "learning_rate": 5.1509481705438605e-06, "loss": 0.8765, "step": 7052 }, { "epoch": 0.51, "grad_norm": 12.70147690447921, "learning_rate": 5.149787379597698e-06, "loss": 0.8896, "step": 7053 }, { "epoch": 0.51, "grad_norm": 3.7561137893085847, "learning_rate": 5.148626580571073e-06, "loss": 0.2341, "step": 7054 }, { "epoch": 0.51, "grad_norm": 2.494436604642709, "learning_rate": 5.147465773526605e-06, "loss": 0.1415, "step": 7055 }, { "epoch": 0.51, "grad_norm": 25.873441532247366, "learning_rate": 5.1463049585269195e-06, "loss": 0.9832, "step": 7056 }, { "epoch": 0.51, "grad_norm": 4.200949436096776, "learning_rate": 5.145144135634633e-06, "loss": 0.2906, "step": 7057 }, { "epoch": 0.51, "grad_norm": 4.316191547260211, "learning_rate": 5.14398330491237e-06, "loss": 0.2579, "step": 7058 }, { "epoch": 0.51, "grad_norm": 15.715050952052533, "learning_rate": 5.142822466422756e-06, "loss": 0.7675, "step": 7059 }, { "epoch": 0.51, "grad_norm": 3.05953350330586, "learning_rate": 5.141661620228406e-06, "loss": 0.2124, "step": 7060 }, { "epoch": 0.51, "grad_norm": 3.7368044055852128, "learning_rate": 5.140500766391949e-06, "loss": 0.2594, "step": 7061 }, { "epoch": 0.51, "grad_norm": 3.2353677442936957, "learning_rate": 5.13933990497601e-06, "loss": 0.1791, "step": 7062 }, { "epoch": 0.51, "grad_norm": 7.373557639648677, "learning_rate": 5.138179036043207e-06, "loss": 0.6792, "step": 7063 }, { "epoch": 0.51, "grad_norm": 9.244613116592573, "learning_rate": 5.137018159656171e-06, "loss": 1.1111, "step": 7064 }, { "epoch": 0.51, "grad_norm": 7.762012845505916, "learning_rate": 5.135857275877523e-06, "loss": 0.9678, "step": 7065 }, { "epoch": 0.51, "grad_norm": 3.0793306738838013, "learning_rate": 5.134696384769888e-06, "loss": 0.1035, "step": 7066 }, { "epoch": 0.51, "grad_norm": 25.00071952645177, "learning_rate": 5.133535486395895e-06, "loss": 1.2106, "step": 7067 }, { "epoch": 0.51, "grad_norm": 17.575710657284635, "learning_rate": 5.132374580818166e-06, "loss": 1.0468, "step": 7068 }, { "epoch": 0.51, "grad_norm": 4.960217789501668, "learning_rate": 5.13121366809933e-06, "loss": 1.0961, "step": 7069 }, { "epoch": 0.51, "grad_norm": 3.9111179958779734, "learning_rate": 5.130052748302013e-06, "loss": 0.7542, "step": 7070 }, { "epoch": 0.51, "grad_norm": 13.573052431189307, "learning_rate": 5.128891821488844e-06, "loss": 0.5347, "step": 7071 }, { "epoch": 0.51, "grad_norm": 19.249689985753186, "learning_rate": 5.127730887722449e-06, "loss": 0.5327, "step": 7072 }, { "epoch": 0.51, "grad_norm": 3.8853505316982164, "learning_rate": 5.126569947065455e-06, "loss": 0.2728, "step": 7073 }, { "epoch": 0.51, "grad_norm": 9.574042177767222, "learning_rate": 5.125408999580491e-06, "loss": 0.819, "step": 7074 }, { "epoch": 0.51, "grad_norm": 14.031903408184382, "learning_rate": 5.124248045330186e-06, "loss": 0.6304, "step": 7075 }, { "epoch": 0.51, "grad_norm": 3.208775899055357, "learning_rate": 5.1230870843771705e-06, "loss": 0.1285, "step": 7076 }, { "epoch": 0.51, "grad_norm": 3.9500102829530808, "learning_rate": 5.1219261167840725e-06, "loss": 0.303, "step": 7077 }, { "epoch": 0.51, "grad_norm": 16.011808176833924, "learning_rate": 5.120765142613522e-06, "loss": 0.8867, "step": 7078 }, { "epoch": 0.51, "grad_norm": 4.47444891572381, "learning_rate": 5.119604161928149e-06, "loss": 0.8308, "step": 7079 }, { "epoch": 0.51, "grad_norm": 3.2177362580546776, "learning_rate": 5.118443174790584e-06, "loss": 0.2367, "step": 7080 }, { "epoch": 0.51, "grad_norm": 3.9597838693120546, "learning_rate": 5.117282181263459e-06, "loss": 0.9457, "step": 7081 }, { "epoch": 0.51, "grad_norm": 17.479379124590544, "learning_rate": 5.116121181409403e-06, "loss": 0.6534, "step": 7082 }, { "epoch": 0.51, "grad_norm": 3.440751991062728, "learning_rate": 5.114960175291049e-06, "loss": 0.2579, "step": 7083 }, { "epoch": 0.51, "grad_norm": 5.167966918279585, "learning_rate": 5.1137991629710295e-06, "loss": 0.2205, "step": 7084 }, { "epoch": 0.51, "grad_norm": 5.3674722698719455, "learning_rate": 5.112638144511976e-06, "loss": 0.3034, "step": 7085 }, { "epoch": 0.51, "grad_norm": 3.4795224248504533, "learning_rate": 5.111477119976519e-06, "loss": 0.2471, "step": 7086 }, { "epoch": 0.51, "grad_norm": 2.8119363280021226, "learning_rate": 5.110316089427295e-06, "loss": 0.1593, "step": 7087 }, { "epoch": 0.51, "grad_norm": 3.365945899107389, "learning_rate": 5.109155052926935e-06, "loss": 0.2331, "step": 7088 }, { "epoch": 0.51, "grad_norm": 2.5774518951750403, "learning_rate": 5.1079940105380744e-06, "loss": 0.2147, "step": 7089 }, { "epoch": 0.51, "grad_norm": 3.44780497488636, "learning_rate": 5.1068329623233446e-06, "loss": 0.2689, "step": 7090 }, { "epoch": 0.51, "grad_norm": 2.409566422559508, "learning_rate": 5.105671908345381e-06, "loss": 0.149, "step": 7091 }, { "epoch": 0.51, "grad_norm": 13.415376596512921, "learning_rate": 5.104510848666817e-06, "loss": 0.6305, "step": 7092 }, { "epoch": 0.51, "grad_norm": 5.583160858989992, "learning_rate": 5.10334978335029e-06, "loss": 0.7519, "step": 7093 }, { "epoch": 0.51, "grad_norm": 22.08378695783893, "learning_rate": 5.102188712458431e-06, "loss": 0.8638, "step": 7094 }, { "epoch": 0.51, "grad_norm": 55.59762266662514, "learning_rate": 5.1010276360538804e-06, "loss": 0.8178, "step": 7095 }, { "epoch": 0.51, "grad_norm": 3.397018870374205, "learning_rate": 5.099866554199271e-06, "loss": 0.1542, "step": 7096 }, { "epoch": 0.51, "grad_norm": 3.634517156393646, "learning_rate": 5.098705466957236e-06, "loss": 0.1924, "step": 7097 }, { "epoch": 0.51, "grad_norm": 5.147153910734839, "learning_rate": 5.097544374390417e-06, "loss": 0.9722, "step": 7098 }, { "epoch": 0.51, "grad_norm": 7.706942277475465, "learning_rate": 5.09638327656145e-06, "loss": 0.4262, "step": 7099 }, { "epoch": 0.51, "grad_norm": 2.81579999713121, "learning_rate": 5.0952221735329665e-06, "loss": 0.1798, "step": 7100 }, { "epoch": 0.51, "grad_norm": 4.8320716983823, "learning_rate": 5.0940610653676095e-06, "loss": 0.8235, "step": 7101 }, { "epoch": 0.51, "grad_norm": 4.14685171737719, "learning_rate": 5.0928999521280144e-06, "loss": 0.2861, "step": 7102 }, { "epoch": 0.51, "grad_norm": 3.0821449354892247, "learning_rate": 5.091738833876816e-06, "loss": 0.1836, "step": 7103 }, { "epoch": 0.51, "grad_norm": 7.717431013116048, "learning_rate": 5.090577710676657e-06, "loss": 0.4994, "step": 7104 }, { "epoch": 0.51, "grad_norm": 4.097320988775142, "learning_rate": 5.089416582590172e-06, "loss": 0.2451, "step": 7105 }, { "epoch": 0.51, "grad_norm": 3.0136267023663965, "learning_rate": 5.088255449680001e-06, "loss": 0.2364, "step": 7106 }, { "epoch": 0.51, "grad_norm": 27.44584851891449, "learning_rate": 5.087094312008784e-06, "loss": 1.2369, "step": 7107 }, { "epoch": 0.51, "grad_norm": 21.920328352901237, "learning_rate": 5.085933169639158e-06, "loss": 0.8886, "step": 7108 }, { "epoch": 0.51, "grad_norm": 4.328077733618511, "learning_rate": 5.0847720226337614e-06, "loss": 0.8225, "step": 7109 }, { "epoch": 0.51, "grad_norm": 15.48687301817782, "learning_rate": 5.083610871055237e-06, "loss": 0.4203, "step": 7110 }, { "epoch": 0.51, "grad_norm": 2.479067803715843, "learning_rate": 5.082449714966221e-06, "loss": 0.1326, "step": 7111 }, { "epoch": 0.51, "grad_norm": 25.63883716289282, "learning_rate": 5.081288554429355e-06, "loss": 1.2177, "step": 7112 }, { "epoch": 0.51, "grad_norm": 4.446492474003896, "learning_rate": 5.080127389507281e-06, "loss": 0.2338, "step": 7113 }, { "epoch": 0.51, "grad_norm": 29.145488106033742, "learning_rate": 5.078966220262635e-06, "loss": 0.8493, "step": 7114 }, { "epoch": 0.51, "grad_norm": 28.817723068389657, "learning_rate": 5.077805046758062e-06, "loss": 1.0868, "step": 7115 }, { "epoch": 0.51, "grad_norm": 3.384288732053604, "learning_rate": 5.076643869056201e-06, "loss": 0.1749, "step": 7116 }, { "epoch": 0.51, "grad_norm": 5.347751930863891, "learning_rate": 5.075482687219693e-06, "loss": 0.9368, "step": 7117 }, { "epoch": 0.51, "grad_norm": 29.98248524846706, "learning_rate": 5.07432150131118e-06, "loss": 1.0276, "step": 7118 }, { "epoch": 0.51, "grad_norm": 58.89504510285276, "learning_rate": 5.073160311393305e-06, "loss": 1.4771, "step": 7119 }, { "epoch": 0.51, "grad_norm": 18.836924581237728, "learning_rate": 5.071999117528705e-06, "loss": 0.8897, "step": 7120 }, { "epoch": 0.51, "grad_norm": 4.855009214091739, "learning_rate": 5.070837919780027e-06, "loss": 0.9099, "step": 7121 }, { "epoch": 0.51, "grad_norm": 51.05000057540748, "learning_rate": 5.06967671820991e-06, "loss": 0.9134, "step": 7122 }, { "epoch": 0.51, "grad_norm": 4.3946947843327955, "learning_rate": 5.068515512880999e-06, "loss": 0.9047, "step": 7123 }, { "epoch": 0.51, "grad_norm": 24.667631633386613, "learning_rate": 5.067354303855936e-06, "loss": 1.3778, "step": 7124 }, { "epoch": 0.51, "grad_norm": 4.24149657427727, "learning_rate": 5.066193091197363e-06, "loss": 0.1943, "step": 7125 }, { "epoch": 0.51, "grad_norm": 4.717815302259036, "learning_rate": 5.0650318749679215e-06, "loss": 0.7724, "step": 7126 }, { "epoch": 0.51, "grad_norm": 3.6879711957613117, "learning_rate": 5.063870655230258e-06, "loss": 0.2558, "step": 7127 }, { "epoch": 0.51, "grad_norm": 4.0346860914947795, "learning_rate": 5.062709432047015e-06, "loss": 0.1878, "step": 7128 }, { "epoch": 0.51, "grad_norm": 17.43111611182097, "learning_rate": 5.061548205480834e-06, "loss": 1.0027, "step": 7129 }, { "epoch": 0.51, "grad_norm": 4.724385816805725, "learning_rate": 5.060386975594361e-06, "loss": 0.1617, "step": 7130 }, { "epoch": 0.51, "grad_norm": 14.77923980944202, "learning_rate": 5.05922574245024e-06, "loss": 1.0582, "step": 7131 }, { "epoch": 0.51, "grad_norm": 3.562581829264788, "learning_rate": 5.0580645061111135e-06, "loss": 0.2124, "step": 7132 }, { "epoch": 0.51, "grad_norm": 16.50671385556264, "learning_rate": 5.056903266639626e-06, "loss": 1.1147, "step": 7133 }, { "epoch": 0.51, "grad_norm": 3.325031706255285, "learning_rate": 5.055742024098423e-06, "loss": 0.1463, "step": 7134 }, { "epoch": 0.51, "grad_norm": 2.829318713486183, "learning_rate": 5.05458077855015e-06, "loss": 0.1153, "step": 7135 }, { "epoch": 0.51, "grad_norm": 8.757811830585645, "learning_rate": 5.053419530057449e-06, "loss": 0.6026, "step": 7136 }, { "epoch": 0.51, "grad_norm": 3.8270495733220673, "learning_rate": 5.052258278682969e-06, "loss": 0.3067, "step": 7137 }, { "epoch": 0.51, "grad_norm": 18.228920596430406, "learning_rate": 5.05109702448935e-06, "loss": 1.2503, "step": 7138 }, { "epoch": 0.51, "grad_norm": 17.83456285670302, "learning_rate": 5.049935767539241e-06, "loss": 0.6314, "step": 7139 }, { "epoch": 0.51, "grad_norm": 2.5939777664613914, "learning_rate": 5.0487745078952865e-06, "loss": 0.1664, "step": 7140 }, { "epoch": 0.51, "grad_norm": 15.236224065859023, "learning_rate": 5.047613245620132e-06, "loss": 0.9364, "step": 7141 }, { "epoch": 0.51, "grad_norm": 4.416530439113788, "learning_rate": 5.046451980776422e-06, "loss": 0.7205, "step": 7142 }, { "epoch": 0.51, "grad_norm": 4.391739391508066, "learning_rate": 5.045290713426805e-06, "loss": 0.2125, "step": 7143 }, { "epoch": 0.51, "grad_norm": 15.256314082066234, "learning_rate": 5.044129443633925e-06, "loss": 0.8942, "step": 7144 }, { "epoch": 0.51, "grad_norm": 4.464767326797669, "learning_rate": 5.042968171460428e-06, "loss": 0.9191, "step": 7145 }, { "epoch": 0.51, "grad_norm": 15.596289243658857, "learning_rate": 5.04180689696896e-06, "loss": 0.6076, "step": 7146 }, { "epoch": 0.51, "grad_norm": 4.649816322467306, "learning_rate": 5.040645620222171e-06, "loss": 0.8459, "step": 7147 }, { "epoch": 0.51, "grad_norm": 21.73978648148838, "learning_rate": 5.039484341282702e-06, "loss": 1.0049, "step": 7148 }, { "epoch": 0.51, "grad_norm": 27.46122762508077, "learning_rate": 5.038323060213204e-06, "loss": 1.2033, "step": 7149 }, { "epoch": 0.51, "grad_norm": 4.218479213701255, "learning_rate": 5.0371617770763214e-06, "loss": 0.2103, "step": 7150 }, { "epoch": 0.51, "grad_norm": 5.89243313571624, "learning_rate": 5.036000491934701e-06, "loss": 0.7028, "step": 7151 }, { "epoch": 0.51, "grad_norm": 10.536927321357805, "learning_rate": 5.034839204850989e-06, "loss": 0.4909, "step": 7152 }, { "epoch": 0.51, "grad_norm": 4.329068961747744, "learning_rate": 5.0336779158878364e-06, "loss": 0.7787, "step": 7153 }, { "epoch": 0.51, "grad_norm": 4.526632933599109, "learning_rate": 5.032516625107888e-06, "loss": 0.962, "step": 7154 }, { "epoch": 0.51, "grad_norm": 33.36747172358653, "learning_rate": 5.0313553325737886e-06, "loss": 1.1171, "step": 7155 }, { "epoch": 0.51, "grad_norm": 4.70613561788194, "learning_rate": 5.030194038348189e-06, "loss": 0.7794, "step": 7156 }, { "epoch": 0.51, "grad_norm": 4.0766678987399185, "learning_rate": 5.029032742493735e-06, "loss": 0.2637, "step": 7157 }, { "epoch": 0.51, "grad_norm": 4.951907316832405, "learning_rate": 5.027871445073075e-06, "loss": 0.8599, "step": 7158 }, { "epoch": 0.51, "grad_norm": 10.506667796906607, "learning_rate": 5.026710146148854e-06, "loss": 0.7722, "step": 7159 }, { "epoch": 0.51, "grad_norm": 3.784718414269091, "learning_rate": 5.025548845783723e-06, "loss": 0.2568, "step": 7160 }, { "epoch": 0.51, "grad_norm": 3.2962517712730017, "learning_rate": 5.02438754404033e-06, "loss": 0.2299, "step": 7161 }, { "epoch": 0.51, "grad_norm": 14.427419531644194, "learning_rate": 5.023226240981319e-06, "loss": 0.6768, "step": 7162 }, { "epoch": 0.51, "grad_norm": 4.295240261065666, "learning_rate": 5.022064936669341e-06, "loss": 0.7988, "step": 7163 }, { "epoch": 0.51, "grad_norm": 17.33545874432047, "learning_rate": 5.020903631167043e-06, "loss": 0.5051, "step": 7164 }, { "epoch": 0.51, "grad_norm": 3.3235360575366992, "learning_rate": 5.019742324537073e-06, "loss": 0.1633, "step": 7165 }, { "epoch": 0.51, "grad_norm": 3.750102753144056, "learning_rate": 5.01858101684208e-06, "loss": 0.311, "step": 7166 }, { "epoch": 0.51, "grad_norm": 8.408690018599339, "learning_rate": 5.0174197081447126e-06, "loss": 0.6178, "step": 7167 }, { "epoch": 0.51, "grad_norm": 22.24667961926797, "learning_rate": 5.0162583985076165e-06, "loss": 1.2563, "step": 7168 }, { "epoch": 0.51, "grad_norm": 4.877064661702152, "learning_rate": 5.0150970879934415e-06, "loss": 0.7263, "step": 7169 }, { "epoch": 0.51, "grad_norm": 3.4362444956389853, "learning_rate": 5.013935776664836e-06, "loss": 0.1527, "step": 7170 }, { "epoch": 0.51, "grad_norm": 3.8104076415460546, "learning_rate": 5.012774464584447e-06, "loss": 0.155, "step": 7171 }, { "epoch": 0.51, "grad_norm": 33.84371139608335, "learning_rate": 5.011613151814926e-06, "loss": 0.9853, "step": 7172 }, { "epoch": 0.51, "grad_norm": 9.003479226220172, "learning_rate": 5.0104518384189185e-06, "loss": 0.52, "step": 7173 }, { "epoch": 0.51, "grad_norm": 3.613808702712701, "learning_rate": 5.009290524459073e-06, "loss": 0.1754, "step": 7174 }, { "epoch": 0.51, "grad_norm": 2.2887458853341935, "learning_rate": 5.008129209998039e-06, "loss": 0.1032, "step": 7175 }, { "epoch": 0.51, "grad_norm": 3.701827635338937, "learning_rate": 5.006967895098466e-06, "loss": 0.2989, "step": 7176 }, { "epoch": 0.51, "grad_norm": 2.975556462714166, "learning_rate": 5.005806579822999e-06, "loss": 0.1712, "step": 7177 }, { "epoch": 0.51, "grad_norm": 4.8399237591297295, "learning_rate": 5.004645264234292e-06, "loss": 0.2667, "step": 7178 }, { "epoch": 0.51, "grad_norm": 4.049966482791155, "learning_rate": 5.0034839483949895e-06, "loss": 0.1219, "step": 7179 }, { "epoch": 0.51, "grad_norm": 4.132115929249343, "learning_rate": 5.002322632367741e-06, "loss": 0.2361, "step": 7180 }, { "epoch": 0.51, "grad_norm": 3.4502648598802734, "learning_rate": 5.001161316215195e-06, "loss": 0.1929, "step": 7181 }, { "epoch": 0.52, "grad_norm": 2.9883296812785574, "learning_rate": 5e-06, "loss": 0.1752, "step": 7182 }, { "epoch": 0.52, "grad_norm": 4.784485031789539, "learning_rate": 4.998838683784806e-06, "loss": 0.2755, "step": 7183 }, { "epoch": 0.52, "grad_norm": 2.6730192076916115, "learning_rate": 4.99767736763226e-06, "loss": 0.1641, "step": 7184 }, { "epoch": 0.52, "grad_norm": 4.629344464512999, "learning_rate": 4.996516051605011e-06, "loss": 0.9106, "step": 7185 }, { "epoch": 0.52, "grad_norm": 3.203914526457378, "learning_rate": 4.99535473576571e-06, "loss": 0.1555, "step": 7186 }, { "epoch": 0.52, "grad_norm": 9.859351812899622, "learning_rate": 4.994193420177001e-06, "loss": 0.38, "step": 7187 }, { "epoch": 0.52, "grad_norm": 4.200184341702598, "learning_rate": 4.993032104901535e-06, "loss": 0.2469, "step": 7188 }, { "epoch": 0.52, "grad_norm": 11.950087547606728, "learning_rate": 4.991870790001963e-06, "loss": 0.6864, "step": 7189 }, { "epoch": 0.52, "grad_norm": 3.4577903286358653, "learning_rate": 4.990709475540931e-06, "loss": 0.2838, "step": 7190 }, { "epoch": 0.52, "grad_norm": 4.2571637428182445, "learning_rate": 4.989548161581084e-06, "loss": 0.9375, "step": 7191 }, { "epoch": 0.52, "grad_norm": 4.46535142696509, "learning_rate": 4.988386848185076e-06, "loss": 0.8855, "step": 7192 }, { "epoch": 0.52, "grad_norm": 11.18887242494504, "learning_rate": 4.987225535415555e-06, "loss": 0.5274, "step": 7193 }, { "epoch": 0.52, "grad_norm": 3.267660588477854, "learning_rate": 4.986064223335166e-06, "loss": 0.1362, "step": 7194 }, { "epoch": 0.52, "grad_norm": 3.636281203644261, "learning_rate": 4.984902912006561e-06, "loss": 0.1819, "step": 7195 }, { "epoch": 0.52, "grad_norm": 4.1701761956995895, "learning_rate": 4.983741601492385e-06, "loss": 0.7501, "step": 7196 }, { "epoch": 0.52, "grad_norm": 3.0373673383852555, "learning_rate": 4.982580291855288e-06, "loss": 0.157, "step": 7197 }, { "epoch": 0.52, "grad_norm": 14.448129389055238, "learning_rate": 4.98141898315792e-06, "loss": 0.4913, "step": 7198 }, { "epoch": 0.52, "grad_norm": 2.5381147559287562, "learning_rate": 4.980257675462927e-06, "loss": 0.1407, "step": 7199 }, { "epoch": 0.52, "grad_norm": 29.62804576373166, "learning_rate": 4.979096368832958e-06, "loss": 1.1392, "step": 7200 }, { "epoch": 0.52, "grad_norm": 5.3184758582134375, "learning_rate": 4.977935063330661e-06, "loss": 0.9104, "step": 7201 }, { "epoch": 0.52, "grad_norm": 4.4402795776495925, "learning_rate": 4.9767737590186836e-06, "loss": 0.8254, "step": 7202 }, { "epoch": 0.52, "grad_norm": 11.644139673766201, "learning_rate": 4.9756124559596725e-06, "loss": 0.5222, "step": 7203 }, { "epoch": 0.52, "grad_norm": 14.291570366848076, "learning_rate": 4.974451154216279e-06, "loss": 0.9995, "step": 7204 }, { "epoch": 0.52, "grad_norm": 3.0168177260199656, "learning_rate": 4.973289853851147e-06, "loss": 0.1872, "step": 7205 }, { "epoch": 0.52, "grad_norm": 14.09804033729488, "learning_rate": 4.972128554926927e-06, "loss": 1.0366, "step": 7206 }, { "epoch": 0.52, "grad_norm": 3.9635052172459835, "learning_rate": 4.970967257506267e-06, "loss": 0.1847, "step": 7207 }, { "epoch": 0.52, "grad_norm": 4.59729863334452, "learning_rate": 4.969805961651812e-06, "loss": 0.9252, "step": 7208 }, { "epoch": 0.52, "grad_norm": 3.5151474342682114, "learning_rate": 4.968644667426213e-06, "loss": 0.1822, "step": 7209 }, { "epoch": 0.52, "grad_norm": 4.3749356424498975, "learning_rate": 4.967483374892113e-06, "loss": 0.7925, "step": 7210 }, { "epoch": 0.52, "grad_norm": 7.918246691068096, "learning_rate": 4.9663220841121635e-06, "loss": 0.1645, "step": 7211 }, { "epoch": 0.52, "grad_norm": 21.885864972228063, "learning_rate": 4.9651607951490106e-06, "loss": 1.2028, "step": 7212 }, { "epoch": 0.52, "grad_norm": 4.666134600058208, "learning_rate": 4.963999508065302e-06, "loss": 0.7952, "step": 7213 }, { "epoch": 0.52, "grad_norm": 4.498221670122784, "learning_rate": 4.962838222923681e-06, "loss": 0.2174, "step": 7214 }, { "epoch": 0.52, "grad_norm": 3.6516789593874153, "learning_rate": 4.961676939786798e-06, "loss": 0.1433, "step": 7215 }, { "epoch": 0.52, "grad_norm": 3.347138041749425, "learning_rate": 4.960515658717299e-06, "loss": 0.1484, "step": 7216 }, { "epoch": 0.52, "grad_norm": 3.919823755443359, "learning_rate": 4.959354379777831e-06, "loss": 0.1763, "step": 7217 }, { "epoch": 0.52, "grad_norm": 3.797876078521176, "learning_rate": 4.958193103031041e-06, "loss": 0.1876, "step": 7218 }, { "epoch": 0.52, "grad_norm": 13.550277109752107, "learning_rate": 4.957031828539573e-06, "loss": 0.5356, "step": 7219 }, { "epoch": 0.52, "grad_norm": 4.206328858125325, "learning_rate": 4.9558705563660755e-06, "loss": 1.0311, "step": 7220 }, { "epoch": 0.52, "grad_norm": 24.911109524454496, "learning_rate": 4.954709286573196e-06, "loss": 0.7647, "step": 7221 }, { "epoch": 0.52, "grad_norm": 3.450212962096687, "learning_rate": 4.953548019223578e-06, "loss": 0.1653, "step": 7222 }, { "epoch": 0.52, "grad_norm": 16.871129550728774, "learning_rate": 4.952386754379869e-06, "loss": 0.7315, "step": 7223 }, { "epoch": 0.52, "grad_norm": 3.7150741776781158, "learning_rate": 4.951225492104714e-06, "loss": 0.1712, "step": 7224 }, { "epoch": 0.52, "grad_norm": 19.367340482972878, "learning_rate": 4.9500642324607615e-06, "loss": 1.016, "step": 7225 }, { "epoch": 0.52, "grad_norm": 3.119237036548225, "learning_rate": 4.948902975510651e-06, "loss": 0.1645, "step": 7226 }, { "epoch": 0.52, "grad_norm": 14.638676355530116, "learning_rate": 4.947741721317034e-06, "loss": 1.0324, "step": 7227 }, { "epoch": 0.52, "grad_norm": 3.4788553598625347, "learning_rate": 4.9465804699425516e-06, "loss": 0.2456, "step": 7228 }, { "epoch": 0.52, "grad_norm": 9.71806906504467, "learning_rate": 4.945419221449851e-06, "loss": 0.6467, "step": 7229 }, { "epoch": 0.52, "grad_norm": 2.5899279855088526, "learning_rate": 4.944257975901578e-06, "loss": 0.1725, "step": 7230 }, { "epoch": 0.52, "grad_norm": 2.814644767143165, "learning_rate": 4.9430967333603745e-06, "loss": 0.1607, "step": 7231 }, { "epoch": 0.52, "grad_norm": 3.0738722988175735, "learning_rate": 4.941935493888888e-06, "loss": 0.1368, "step": 7232 }, { "epoch": 0.52, "grad_norm": 27.54301121707586, "learning_rate": 4.940774257549761e-06, "loss": 0.6194, "step": 7233 }, { "epoch": 0.52, "grad_norm": 14.276079728600644, "learning_rate": 4.93961302440564e-06, "loss": 0.8738, "step": 7234 }, { "epoch": 0.52, "grad_norm": 13.664852383241296, "learning_rate": 4.9384517945191665e-06, "loss": 0.6569, "step": 7235 }, { "epoch": 0.52, "grad_norm": 2.446790671360228, "learning_rate": 4.937290567952986e-06, "loss": 0.1007, "step": 7236 }, { "epoch": 0.52, "grad_norm": 3.9849285211011107, "learning_rate": 4.9361293447697435e-06, "loss": 0.2498, "step": 7237 }, { "epoch": 0.52, "grad_norm": 15.972301260740013, "learning_rate": 4.93496812503208e-06, "loss": 0.7837, "step": 7238 }, { "epoch": 0.52, "grad_norm": 3.61543539025328, "learning_rate": 4.933806908802639e-06, "loss": 0.2414, "step": 7239 }, { "epoch": 0.52, "grad_norm": 27.34902092511614, "learning_rate": 4.932645696144066e-06, "loss": 0.9898, "step": 7240 }, { "epoch": 0.52, "grad_norm": 4.456685242134273, "learning_rate": 4.9314844871190025e-06, "loss": 0.3031, "step": 7241 }, { "epoch": 0.52, "grad_norm": 10.88953370574277, "learning_rate": 4.93032328179009e-06, "loss": 0.5265, "step": 7242 }, { "epoch": 0.52, "grad_norm": 20.455272049928148, "learning_rate": 4.929162080219974e-06, "loss": 0.7339, "step": 7243 }, { "epoch": 0.52, "grad_norm": 15.792588008933501, "learning_rate": 4.9280008824712964e-06, "loss": 0.6289, "step": 7244 }, { "epoch": 0.52, "grad_norm": 4.95512815083364, "learning_rate": 4.9268396886066975e-06, "loss": 0.8405, "step": 7245 }, { "epoch": 0.52, "grad_norm": 9.95100923509515, "learning_rate": 4.925678498688821e-06, "loss": 0.2692, "step": 7246 }, { "epoch": 0.52, "grad_norm": 12.506331143706994, "learning_rate": 4.924517312780307e-06, "loss": 0.5507, "step": 7247 }, { "epoch": 0.52, "grad_norm": 19.984508561408635, "learning_rate": 4.923356130943799e-06, "loss": 1.0272, "step": 7248 }, { "epoch": 0.52, "grad_norm": 14.626787450268251, "learning_rate": 4.92219495324194e-06, "loss": 0.9842, "step": 7249 }, { "epoch": 0.52, "grad_norm": 4.994178355758792, "learning_rate": 4.9210337797373665e-06, "loss": 0.7576, "step": 7250 }, { "epoch": 0.52, "grad_norm": 14.213810416336058, "learning_rate": 4.919872610492721e-06, "loss": 1.2242, "step": 7251 }, { "epoch": 0.52, "grad_norm": 3.647179226054615, "learning_rate": 4.918711445570647e-06, "loss": 0.164, "step": 7252 }, { "epoch": 0.52, "grad_norm": 3.3930034464710594, "learning_rate": 4.917550285033781e-06, "loss": 0.2196, "step": 7253 }, { "epoch": 0.52, "grad_norm": 3.016259809761435, "learning_rate": 4.916389128944765e-06, "loss": 0.1415, "step": 7254 }, { "epoch": 0.52, "grad_norm": 3.639965978907688, "learning_rate": 4.915227977366239e-06, "loss": 0.1362, "step": 7255 }, { "epoch": 0.52, "grad_norm": 3.4534100974205444, "learning_rate": 4.914066830360844e-06, "loss": 0.1982, "step": 7256 }, { "epoch": 0.52, "grad_norm": 3.61603895804799, "learning_rate": 4.912905687991217e-06, "loss": 0.2115, "step": 7257 }, { "epoch": 0.52, "grad_norm": 11.703046725074119, "learning_rate": 4.911744550319999e-06, "loss": 1.003, "step": 7258 }, { "epoch": 0.52, "grad_norm": 4.272433422873963, "learning_rate": 4.910583417409828e-06, "loss": 0.2373, "step": 7259 }, { "epoch": 0.52, "grad_norm": 2.781954676468065, "learning_rate": 4.909422289323344e-06, "loss": 0.1499, "step": 7260 }, { "epoch": 0.52, "grad_norm": 28.874675251252114, "learning_rate": 4.908261166123186e-06, "loss": 0.535, "step": 7261 }, { "epoch": 0.52, "grad_norm": 10.033938199538873, "learning_rate": 4.907100047871989e-06, "loss": 0.638, "step": 7262 }, { "epoch": 0.52, "grad_norm": 2.387764526368929, "learning_rate": 4.905938934632392e-06, "loss": 0.1318, "step": 7263 }, { "epoch": 0.52, "grad_norm": 2.9819701540184957, "learning_rate": 4.904777826467035e-06, "loss": 0.1365, "step": 7264 }, { "epoch": 0.52, "grad_norm": 10.52851305027289, "learning_rate": 4.903616723438553e-06, "loss": 0.8329, "step": 7265 }, { "epoch": 0.52, "grad_norm": 2.8502617075484973, "learning_rate": 4.902455625609584e-06, "loss": 0.202, "step": 7266 }, { "epoch": 0.52, "grad_norm": 3.322645430678806, "learning_rate": 4.9012945330427645e-06, "loss": 0.2032, "step": 7267 }, { "epoch": 0.52, "grad_norm": 8.216650966698792, "learning_rate": 4.900133445800731e-06, "loss": 0.8933, "step": 7268 }, { "epoch": 0.52, "grad_norm": 3.7652097003769027, "learning_rate": 4.898972363946121e-06, "loss": 0.1564, "step": 7269 }, { "epoch": 0.52, "grad_norm": 11.472209098497546, "learning_rate": 4.897811287541569e-06, "loss": 0.8055, "step": 7270 }, { "epoch": 0.52, "grad_norm": 4.503255701368401, "learning_rate": 4.896650216649711e-06, "loss": 0.925, "step": 7271 }, { "epoch": 0.52, "grad_norm": 6.014530083852191, "learning_rate": 4.895489151333184e-06, "loss": 0.9089, "step": 7272 }, { "epoch": 0.52, "grad_norm": 15.667869405552707, "learning_rate": 4.894328091654622e-06, "loss": 1.0389, "step": 7273 }, { "epoch": 0.52, "grad_norm": 7.42744320659767, "learning_rate": 4.893167037676658e-06, "loss": 0.5512, "step": 7274 }, { "epoch": 0.52, "grad_norm": 20.64837539339915, "learning_rate": 4.892005989461929e-06, "loss": 1.0722, "step": 7275 }, { "epoch": 0.52, "grad_norm": 3.647915172340834, "learning_rate": 4.890844947073066e-06, "loss": 0.1701, "step": 7276 }, { "epoch": 0.52, "grad_norm": 3.6828622996813976, "learning_rate": 4.8896839105727055e-06, "loss": 0.2435, "step": 7277 }, { "epoch": 0.52, "grad_norm": 3.760424788547371, "learning_rate": 4.888522880023482e-06, "loss": 0.2081, "step": 7278 }, { "epoch": 0.52, "grad_norm": 23.532180161768192, "learning_rate": 4.887361855488026e-06, "loss": 0.5484, "step": 7279 }, { "epoch": 0.52, "grad_norm": 9.830887235556787, "learning_rate": 4.886200837028972e-06, "loss": 0.5663, "step": 7280 }, { "epoch": 0.52, "grad_norm": 27.635284264870215, "learning_rate": 4.885039824708952e-06, "loss": 0.7844, "step": 7281 }, { "epoch": 0.52, "grad_norm": 15.596463164089513, "learning_rate": 4.883878818590598e-06, "loss": 0.5299, "step": 7282 }, { "epoch": 0.52, "grad_norm": 3.2384470986516107, "learning_rate": 4.8827178187365425e-06, "loss": 0.2277, "step": 7283 }, { "epoch": 0.52, "grad_norm": 5.245018523728074, "learning_rate": 4.881556825209417e-06, "loss": 0.1444, "step": 7284 }, { "epoch": 0.52, "grad_norm": 4.019092366817288, "learning_rate": 4.880395838071853e-06, "loss": 0.2903, "step": 7285 }, { "epoch": 0.52, "grad_norm": 3.2900118808545273, "learning_rate": 4.8792348573864795e-06, "loss": 0.1973, "step": 7286 }, { "epoch": 0.52, "grad_norm": 4.1806586906745755, "learning_rate": 4.87807388321593e-06, "loss": 0.8761, "step": 7287 }, { "epoch": 0.52, "grad_norm": 4.75071079160524, "learning_rate": 4.87691291562283e-06, "loss": 1.0066, "step": 7288 }, { "epoch": 0.52, "grad_norm": 3.4135187865493433, "learning_rate": 4.875751954669815e-06, "loss": 0.1869, "step": 7289 }, { "epoch": 0.52, "grad_norm": 3.702078114785156, "learning_rate": 4.874591000419511e-06, "loss": 0.1417, "step": 7290 }, { "epoch": 0.52, "grad_norm": 3.5060939579193593, "learning_rate": 4.873430052934547e-06, "loss": 0.2437, "step": 7291 }, { "epoch": 0.52, "grad_norm": 2.651122315939013, "learning_rate": 4.872269112277554e-06, "loss": 0.2057, "step": 7292 }, { "epoch": 0.52, "grad_norm": 3.236033004063919, "learning_rate": 4.871108178511157e-06, "loss": 0.1452, "step": 7293 }, { "epoch": 0.52, "grad_norm": 3.6649446303315285, "learning_rate": 4.8699472516979875e-06, "loss": 0.1826, "step": 7294 }, { "epoch": 0.52, "grad_norm": 3.866063853532521, "learning_rate": 4.868786331900671e-06, "loss": 0.2407, "step": 7295 }, { "epoch": 0.52, "grad_norm": 2.990119995718422, "learning_rate": 4.867625419181834e-06, "loss": 0.2065, "step": 7296 }, { "epoch": 0.52, "grad_norm": 2.7821678475853138, "learning_rate": 4.8664645136041075e-06, "loss": 0.1727, "step": 7297 }, { "epoch": 0.52, "grad_norm": 9.670938786182298, "learning_rate": 4.865303615230114e-06, "loss": 0.4331, "step": 7298 }, { "epoch": 0.52, "grad_norm": 2.196523629769743, "learning_rate": 4.864142724122479e-06, "loss": 0.1429, "step": 7299 }, { "epoch": 0.52, "grad_norm": 6.439387717252638, "learning_rate": 4.86298184034383e-06, "loss": 0.8465, "step": 7300 }, { "epoch": 0.52, "grad_norm": 15.594568454125696, "learning_rate": 4.861820963956794e-06, "loss": 0.9393, "step": 7301 }, { "epoch": 0.52, "grad_norm": 8.806795437324073, "learning_rate": 4.860660095023992e-06, "loss": 0.4073, "step": 7302 }, { "epoch": 0.52, "grad_norm": 11.491861638161025, "learning_rate": 4.859499233608052e-06, "loss": 0.6276, "step": 7303 }, { "epoch": 0.52, "grad_norm": 28.176495473168842, "learning_rate": 4.858338379771595e-06, "loss": 0.9191, "step": 7304 }, { "epoch": 0.52, "grad_norm": 37.55144751718158, "learning_rate": 4.857177533577247e-06, "loss": 1.7783, "step": 7305 }, { "epoch": 0.52, "grad_norm": 3.7584447798995217, "learning_rate": 4.8560166950876305e-06, "loss": 0.1689, "step": 7306 }, { "epoch": 0.52, "grad_norm": 4.832852243669278, "learning_rate": 4.854855864365367e-06, "loss": 0.8792, "step": 7307 }, { "epoch": 0.52, "grad_norm": 24.683935283463608, "learning_rate": 4.853695041473082e-06, "loss": 0.7237, "step": 7308 }, { "epoch": 0.52, "grad_norm": 3.519449635227011, "learning_rate": 4.852534226473396e-06, "loss": 0.1854, "step": 7309 }, { "epoch": 0.52, "grad_norm": 17.72440193985303, "learning_rate": 4.85137341942893e-06, "loss": 1.0002, "step": 7310 }, { "epoch": 0.52, "grad_norm": 4.049756395610826, "learning_rate": 4.8502126204023044e-06, "loss": 0.8422, "step": 7311 }, { "epoch": 0.52, "grad_norm": 4.764335601620697, "learning_rate": 4.849051829456142e-06, "loss": 0.8215, "step": 7312 }, { "epoch": 0.52, "grad_norm": 6.796554469915449, "learning_rate": 4.84789104665306e-06, "loss": 0.4974, "step": 7313 }, { "epoch": 0.52, "grad_norm": 10.176699468576754, "learning_rate": 4.846730272055682e-06, "loss": 0.691, "step": 7314 }, { "epoch": 0.52, "grad_norm": 3.9759776998657124, "learning_rate": 4.845569505726625e-06, "loss": 0.9174, "step": 7315 }, { "epoch": 0.52, "grad_norm": 13.57931465032416, "learning_rate": 4.844408747728509e-06, "loss": 0.5726, "step": 7316 }, { "epoch": 0.52, "grad_norm": 16.214921301978343, "learning_rate": 4.843247998123953e-06, "loss": 0.8591, "step": 7317 }, { "epoch": 0.52, "grad_norm": 4.200095997926698, "learning_rate": 4.842087256975572e-06, "loss": 0.826, "step": 7318 }, { "epoch": 0.52, "grad_norm": 3.1794481443405456, "learning_rate": 4.8409265243459865e-06, "loss": 0.2237, "step": 7319 }, { "epoch": 0.52, "grad_norm": 14.343723726928333, "learning_rate": 4.839765800297814e-06, "loss": 0.6737, "step": 7320 }, { "epoch": 0.52, "grad_norm": 11.382734369395722, "learning_rate": 4.838605084893671e-06, "loss": 0.8062, "step": 7321 }, { "epoch": 0.53, "grad_norm": 13.940155167057052, "learning_rate": 4.837444378196171e-06, "loss": 1.1079, "step": 7322 }, { "epoch": 0.53, "grad_norm": 19.14294487852985, "learning_rate": 4.836283680267931e-06, "loss": 0.6756, "step": 7323 }, { "epoch": 0.53, "grad_norm": 43.75006576610923, "learning_rate": 4.835122991171567e-06, "loss": 0.7193, "step": 7324 }, { "epoch": 0.53, "grad_norm": 10.472764748143721, "learning_rate": 4.833962310969694e-06, "loss": 0.6404, "step": 7325 }, { "epoch": 0.53, "grad_norm": 2.6807658496167766, "learning_rate": 4.832801639724927e-06, "loss": 0.205, "step": 7326 }, { "epoch": 0.53, "grad_norm": 2.2277812371238808, "learning_rate": 4.831640977499876e-06, "loss": 0.1012, "step": 7327 }, { "epoch": 0.53, "grad_norm": 2.963926268985049, "learning_rate": 4.830480324357158e-06, "loss": 0.1873, "step": 7328 }, { "epoch": 0.53, "grad_norm": 8.306024266713584, "learning_rate": 4.829319680359386e-06, "loss": 0.7593, "step": 7329 }, { "epoch": 0.53, "grad_norm": 7.53155367676761, "learning_rate": 4.82815904556917e-06, "loss": 0.3964, "step": 7330 }, { "epoch": 0.53, "grad_norm": 2.9433008731716104, "learning_rate": 4.826998420049124e-06, "loss": 0.2232, "step": 7331 }, { "epoch": 0.53, "grad_norm": 17.479410928279755, "learning_rate": 4.825837803861857e-06, "loss": 1.149, "step": 7332 }, { "epoch": 0.53, "grad_norm": 3.4487831465470475, "learning_rate": 4.824677197069984e-06, "loss": 0.2015, "step": 7333 }, { "epoch": 0.53, "grad_norm": 17.440042806900074, "learning_rate": 4.82351659973611e-06, "loss": 1.5046, "step": 7334 }, { "epoch": 0.53, "grad_norm": 8.305801596335481, "learning_rate": 4.822356011922849e-06, "loss": 0.6239, "step": 7335 }, { "epoch": 0.53, "grad_norm": 8.507724329720157, "learning_rate": 4.821195433692807e-06, "loss": 0.9176, "step": 7336 }, { "epoch": 0.53, "grad_norm": 4.84191197392477, "learning_rate": 4.820034865108595e-06, "loss": 0.8773, "step": 7337 }, { "epoch": 0.53, "grad_norm": 4.548761275006913, "learning_rate": 4.818874306232822e-06, "loss": 0.7438, "step": 7338 }, { "epoch": 0.53, "grad_norm": 21.381993088549198, "learning_rate": 4.817713757128093e-06, "loss": 1.1367, "step": 7339 }, { "epoch": 0.53, "grad_norm": 4.882496095772485, "learning_rate": 4.816553217857018e-06, "loss": 0.6297, "step": 7340 }, { "epoch": 0.53, "grad_norm": 5.149286569429045, "learning_rate": 4.815392688482202e-06, "loss": 0.9502, "step": 7341 }, { "epoch": 0.53, "grad_norm": 5.51640292309537, "learning_rate": 4.81423216906625e-06, "loss": 0.6876, "step": 7342 }, { "epoch": 0.53, "grad_norm": 13.965332488148634, "learning_rate": 4.813071659671773e-06, "loss": 0.7503, "step": 7343 }, { "epoch": 0.53, "grad_norm": 19.491206761419445, "learning_rate": 4.81191116036137e-06, "loss": 1.2426, "step": 7344 }, { "epoch": 0.53, "grad_norm": 26.140560168850183, "learning_rate": 4.81075067119765e-06, "loss": 1.2042, "step": 7345 }, { "epoch": 0.53, "grad_norm": 10.951830850997519, "learning_rate": 4.809590192243213e-06, "loss": 0.4849, "step": 7346 }, { "epoch": 0.53, "grad_norm": 3.088789233857515, "learning_rate": 4.808429723560665e-06, "loss": 0.1671, "step": 7347 }, { "epoch": 0.53, "grad_norm": 3.4827791056847386, "learning_rate": 4.807269265212607e-06, "loss": 0.1834, "step": 7348 }, { "epoch": 0.53, "grad_norm": 4.480213086927034, "learning_rate": 4.806108817261645e-06, "loss": 0.8416, "step": 7349 }, { "epoch": 0.53, "grad_norm": 14.144438427766069, "learning_rate": 4.804948379770376e-06, "loss": 0.5827, "step": 7350 }, { "epoch": 0.53, "grad_norm": 5.74874075158457, "learning_rate": 4.803787952801404e-06, "loss": 0.3681, "step": 7351 }, { "epoch": 0.53, "grad_norm": 3.40476006069444, "learning_rate": 4.802627536417331e-06, "loss": 0.2029, "step": 7352 }, { "epoch": 0.53, "grad_norm": 4.277518959281359, "learning_rate": 4.801467130680753e-06, "loss": 0.8182, "step": 7353 }, { "epoch": 0.53, "grad_norm": 4.646684035324953, "learning_rate": 4.8003067356542745e-06, "loss": 0.1981, "step": 7354 }, { "epoch": 0.53, "grad_norm": 18.57119957748441, "learning_rate": 4.7991463514004895e-06, "loss": 0.9633, "step": 7355 }, { "epoch": 0.53, "grad_norm": 32.88147594172567, "learning_rate": 4.797985977981999e-06, "loss": 1.1384, "step": 7356 }, { "epoch": 0.53, "grad_norm": 4.504869859528523, "learning_rate": 4.7968256154614015e-06, "loss": 0.2807, "step": 7357 }, { "epoch": 0.53, "grad_norm": 23.410382378697953, "learning_rate": 4.7956652639012934e-06, "loss": 1.2504, "step": 7358 }, { "epoch": 0.53, "grad_norm": 8.962242869699095, "learning_rate": 4.794504923364269e-06, "loss": 0.5786, "step": 7359 }, { "epoch": 0.53, "grad_norm": 9.041436741473058, "learning_rate": 4.793344593912928e-06, "loss": 0.9731, "step": 7360 }, { "epoch": 0.53, "grad_norm": 3.7071061979196944, "learning_rate": 4.7921842756098615e-06, "loss": 0.1704, "step": 7361 }, { "epoch": 0.53, "grad_norm": 3.2155903248261275, "learning_rate": 4.791023968517668e-06, "loss": 0.2475, "step": 7362 }, { "epoch": 0.53, "grad_norm": 23.11886649448644, "learning_rate": 4.78986367269894e-06, "loss": 0.6318, "step": 7363 }, { "epoch": 0.53, "grad_norm": 3.8984430725734582, "learning_rate": 4.788703388216272e-06, "loss": 0.2927, "step": 7364 }, { "epoch": 0.53, "grad_norm": 7.132949459410912, "learning_rate": 4.787543115132255e-06, "loss": 0.6617, "step": 7365 }, { "epoch": 0.53, "grad_norm": 17.392823553547117, "learning_rate": 4.786382853509484e-06, "loss": 1.0035, "step": 7366 }, { "epoch": 0.53, "grad_norm": 21.334904071287085, "learning_rate": 4.7852226034105495e-06, "loss": 0.8581, "step": 7367 }, { "epoch": 0.53, "grad_norm": 10.480871348931549, "learning_rate": 4.784062364898043e-06, "loss": 0.6408, "step": 7368 }, { "epoch": 0.53, "grad_norm": 38.835458377843324, "learning_rate": 4.782902138034555e-06, "loss": 0.8756, "step": 7369 }, { "epoch": 0.53, "grad_norm": 3.555591288837996, "learning_rate": 4.781741922882673e-06, "loss": 0.2129, "step": 7370 }, { "epoch": 0.53, "grad_norm": 3.332081155914216, "learning_rate": 4.780581719504988e-06, "loss": 0.1817, "step": 7371 }, { "epoch": 0.53, "grad_norm": 3.6117860145490224, "learning_rate": 4.7794215279640906e-06, "loss": 0.1381, "step": 7372 }, { "epoch": 0.53, "grad_norm": 4.110786199279039, "learning_rate": 4.778261348322566e-06, "loss": 0.196, "step": 7373 }, { "epoch": 0.53, "grad_norm": 3.626226652849538, "learning_rate": 4.777101180643003e-06, "loss": 0.1771, "step": 7374 }, { "epoch": 0.53, "grad_norm": 17.94218043038131, "learning_rate": 4.775941024987985e-06, "loss": 0.7917, "step": 7375 }, { "epoch": 0.53, "grad_norm": 19.665536749238793, "learning_rate": 4.774780881420102e-06, "loss": 1.554, "step": 7376 }, { "epoch": 0.53, "grad_norm": 9.00599069463481, "learning_rate": 4.773620750001939e-06, "loss": 0.4873, "step": 7377 }, { "epoch": 0.53, "grad_norm": 10.856310940641572, "learning_rate": 4.772460630796079e-06, "loss": 0.8531, "step": 7378 }, { "epoch": 0.53, "grad_norm": 4.7509321625498435, "learning_rate": 4.771300523865106e-06, "loss": 0.8867, "step": 7379 }, { "epoch": 0.53, "grad_norm": 8.623975208257644, "learning_rate": 4.770140429271606e-06, "loss": 0.5762, "step": 7380 }, { "epoch": 0.53, "grad_norm": 9.644440717767928, "learning_rate": 4.7689803470781595e-06, "loss": 0.7997, "step": 7381 }, { "epoch": 0.53, "grad_norm": 4.732286375567001, "learning_rate": 4.767820277347347e-06, "loss": 0.7661, "step": 7382 }, { "epoch": 0.53, "grad_norm": 7.462040115897494, "learning_rate": 4.766660220141753e-06, "loss": 0.3775, "step": 7383 }, { "epoch": 0.53, "grad_norm": 2.6450757688676494, "learning_rate": 4.765500175523957e-06, "loss": 0.1351, "step": 7384 }, { "epoch": 0.53, "grad_norm": 3.093798905885654, "learning_rate": 4.764340143556537e-06, "loss": 0.2253, "step": 7385 }, { "epoch": 0.53, "grad_norm": 27.40054893446773, "learning_rate": 4.763180124302076e-06, "loss": 1.4757, "step": 7386 }, { "epoch": 0.53, "grad_norm": 9.461011912766589, "learning_rate": 4.76202011782315e-06, "loss": 0.5534, "step": 7387 }, { "epoch": 0.53, "grad_norm": 4.045134003234878, "learning_rate": 4.7608601241823385e-06, "loss": 0.878, "step": 7388 }, { "epoch": 0.53, "grad_norm": 3.1084750996615798, "learning_rate": 4.759700143442218e-06, "loss": 0.1468, "step": 7389 }, { "epoch": 0.53, "grad_norm": 6.579308630761731, "learning_rate": 4.758540175665365e-06, "loss": 0.428, "step": 7390 }, { "epoch": 0.53, "grad_norm": 4.639718804513899, "learning_rate": 4.757380220914355e-06, "loss": 0.2604, "step": 7391 }, { "epoch": 0.53, "grad_norm": 2.7942439079025156, "learning_rate": 4.756220279251764e-06, "loss": 0.1686, "step": 7392 }, { "epoch": 0.53, "grad_norm": 21.460299292433774, "learning_rate": 4.755060350740168e-06, "loss": 0.5502, "step": 7393 }, { "epoch": 0.53, "grad_norm": 3.273430734847947, "learning_rate": 4.753900435442137e-06, "loss": 0.1825, "step": 7394 }, { "epoch": 0.53, "grad_norm": 6.3489116351991415, "learning_rate": 4.752740533420247e-06, "loss": 0.8755, "step": 7395 }, { "epoch": 0.53, "grad_norm": 15.003698446787487, "learning_rate": 4.751580644737069e-06, "loss": 0.6953, "step": 7396 }, { "epoch": 0.53, "grad_norm": 3.944507354556899, "learning_rate": 4.750420769455174e-06, "loss": 0.2197, "step": 7397 }, { "epoch": 0.53, "grad_norm": 17.07772944917084, "learning_rate": 4.749260907637134e-06, "loss": 0.5203, "step": 7398 }, { "epoch": 0.53, "grad_norm": 2.822043857457953, "learning_rate": 4.748101059345519e-06, "loss": 0.1594, "step": 7399 }, { "epoch": 0.53, "grad_norm": 2.987265074444884, "learning_rate": 4.7469412246428996e-06, "loss": 0.1446, "step": 7400 }, { "epoch": 0.53, "grad_norm": 3.0939044313975645, "learning_rate": 4.745781403591841e-06, "loss": 0.159, "step": 7401 }, { "epoch": 0.53, "grad_norm": 15.421227883928937, "learning_rate": 4.744621596254914e-06, "loss": 0.9172, "step": 7402 }, { "epoch": 0.53, "grad_norm": 19.22571975792514, "learning_rate": 4.743461802694687e-06, "loss": 0.7671, "step": 7403 }, { "epoch": 0.53, "grad_norm": 4.71685609167171, "learning_rate": 4.742302022973723e-06, "loss": 0.8261, "step": 7404 }, { "epoch": 0.53, "grad_norm": 4.9258388137696585, "learning_rate": 4.741142257154591e-06, "loss": 0.8419, "step": 7405 }, { "epoch": 0.53, "grad_norm": 3.509653047958297, "learning_rate": 4.739982505299855e-06, "loss": 0.2561, "step": 7406 }, { "epoch": 0.53, "grad_norm": 29.77643878338538, "learning_rate": 4.738822767472076e-06, "loss": 0.772, "step": 7407 }, { "epoch": 0.53, "grad_norm": 16.188064815104585, "learning_rate": 4.7376630437338205e-06, "loss": 0.8383, "step": 7408 }, { "epoch": 0.53, "grad_norm": 3.30984636475674, "learning_rate": 4.736503334147652e-06, "loss": 0.1591, "step": 7409 }, { "epoch": 0.53, "grad_norm": 4.229771275264, "learning_rate": 4.735343638776131e-06, "loss": 0.8365, "step": 7410 }, { "epoch": 0.53, "grad_norm": 3.152871709157695, "learning_rate": 4.734183957681818e-06, "loss": 0.1888, "step": 7411 }, { "epoch": 0.53, "grad_norm": 26.09319594576119, "learning_rate": 4.733024290927275e-06, "loss": 0.5626, "step": 7412 }, { "epoch": 0.53, "grad_norm": 2.9795795317075218, "learning_rate": 4.73186463857506e-06, "loss": 0.2892, "step": 7413 }, { "epoch": 0.53, "grad_norm": 2.3197851096959745, "learning_rate": 4.730705000687734e-06, "loss": 0.1156, "step": 7414 }, { "epoch": 0.53, "grad_norm": 2.8069410356311475, "learning_rate": 4.729545377327853e-06, "loss": 0.2065, "step": 7415 }, { "epoch": 0.53, "grad_norm": 4.7164267690637, "learning_rate": 4.728385768557976e-06, "loss": 0.7531, "step": 7416 }, { "epoch": 0.53, "grad_norm": 3.128812923296037, "learning_rate": 4.727226174440658e-06, "loss": 0.2293, "step": 7417 }, { "epoch": 0.53, "grad_norm": 3.331227973521176, "learning_rate": 4.726066595038457e-06, "loss": 0.2247, "step": 7418 }, { "epoch": 0.53, "grad_norm": 8.823517841419365, "learning_rate": 4.724907030413925e-06, "loss": 0.7426, "step": 7419 }, { "epoch": 0.53, "grad_norm": 11.287457604707098, "learning_rate": 4.723747480629619e-06, "loss": 0.4732, "step": 7420 }, { "epoch": 0.53, "grad_norm": 3.0053244618281263, "learning_rate": 4.722587945748088e-06, "loss": 0.1965, "step": 7421 }, { "epoch": 0.53, "grad_norm": 19.78072054631874, "learning_rate": 4.721428425831888e-06, "loss": 0.6697, "step": 7422 }, { "epoch": 0.53, "grad_norm": 20.00137989790151, "learning_rate": 4.720268920943571e-06, "loss": 1.1159, "step": 7423 }, { "epoch": 0.53, "grad_norm": 12.822776870140846, "learning_rate": 4.719109431145686e-06, "loss": 0.6397, "step": 7424 }, { "epoch": 0.53, "grad_norm": 3.1209366621405, "learning_rate": 4.717949956500785e-06, "loss": 0.2096, "step": 7425 }, { "epoch": 0.53, "grad_norm": 5.3045528408448455, "learning_rate": 4.716790497071414e-06, "loss": 0.9552, "step": 7426 }, { "epoch": 0.53, "grad_norm": 3.6830384376844867, "learning_rate": 4.715631052920126e-06, "loss": 0.2019, "step": 7427 }, { "epoch": 0.53, "grad_norm": 4.328223732385981, "learning_rate": 4.714471624109465e-06, "loss": 0.7309, "step": 7428 }, { "epoch": 0.53, "grad_norm": 3.1286440355515235, "learning_rate": 4.713312210701979e-06, "loss": 0.1934, "step": 7429 }, { "epoch": 0.53, "grad_norm": 4.2498060638845, "learning_rate": 4.712152812760217e-06, "loss": 0.2771, "step": 7430 }, { "epoch": 0.53, "grad_norm": 4.082019071500093, "learning_rate": 4.710993430346718e-06, "loss": 0.2145, "step": 7431 }, { "epoch": 0.53, "grad_norm": 3.275297391509242, "learning_rate": 4.709834063524031e-06, "loss": 0.2297, "step": 7432 }, { "epoch": 0.53, "grad_norm": 2.8672899372015626, "learning_rate": 4.708674712354697e-06, "loss": 0.1477, "step": 7433 }, { "epoch": 0.53, "grad_norm": 3.9214553781919914, "learning_rate": 4.70751537690126e-06, "loss": 0.2774, "step": 7434 }, { "epoch": 0.53, "grad_norm": 3.9335828945122935, "learning_rate": 4.706356057226261e-06, "loss": 0.916, "step": 7435 }, { "epoch": 0.53, "grad_norm": 3.4974598652889934, "learning_rate": 4.7051967533922405e-06, "loss": 0.1872, "step": 7436 }, { "epoch": 0.53, "grad_norm": 15.698222079140967, "learning_rate": 4.7040374654617406e-06, "loss": 1.3554, "step": 7437 }, { "epoch": 0.53, "grad_norm": 6.14984216463681, "learning_rate": 4.702878193497298e-06, "loss": 0.799, "step": 7438 }, { "epoch": 0.53, "grad_norm": 3.8519564162513564, "learning_rate": 4.701718937561454e-06, "loss": 0.2066, "step": 7439 }, { "epoch": 0.53, "grad_norm": 4.181442619959871, "learning_rate": 4.700559697716742e-06, "loss": 0.2304, "step": 7440 }, { "epoch": 0.53, "grad_norm": 11.970605663224404, "learning_rate": 4.699400474025702e-06, "loss": 0.6707, "step": 7441 }, { "epoch": 0.53, "grad_norm": 8.46967409725048, "learning_rate": 4.69824126655087e-06, "loss": 0.7847, "step": 7442 }, { "epoch": 0.53, "grad_norm": 2.637537446574159, "learning_rate": 4.697082075354778e-06, "loss": 0.1681, "step": 7443 }, { "epoch": 0.53, "grad_norm": 11.835393956679349, "learning_rate": 4.695922900499961e-06, "loss": 0.692, "step": 7444 }, { "epoch": 0.53, "grad_norm": 3.3357814770346086, "learning_rate": 4.694763742048952e-06, "loss": 0.1875, "step": 7445 }, { "epoch": 0.53, "grad_norm": 2.728864137714398, "learning_rate": 4.6936046000642856e-06, "loss": 0.1465, "step": 7446 }, { "epoch": 0.53, "grad_norm": 2.780335856880941, "learning_rate": 4.69244547460849e-06, "loss": 0.1809, "step": 7447 }, { "epoch": 0.53, "grad_norm": 8.041677616556168, "learning_rate": 4.691286365744098e-06, "loss": 0.7044, "step": 7448 }, { "epoch": 0.53, "grad_norm": 19.93407805097311, "learning_rate": 4.690127273533636e-06, "loss": 0.9362, "step": 7449 }, { "epoch": 0.53, "grad_norm": 4.015413873808265, "learning_rate": 4.688968198039636e-06, "loss": 0.2827, "step": 7450 }, { "epoch": 0.53, "grad_norm": 4.575448165582346, "learning_rate": 4.687809139324624e-06, "loss": 0.6903, "step": 7451 }, { "epoch": 0.53, "grad_norm": 4.650405539543639, "learning_rate": 4.686650097451127e-06, "loss": 0.9899, "step": 7452 }, { "epoch": 0.53, "grad_norm": 11.036002934807112, "learning_rate": 4.685491072481673e-06, "loss": 0.7851, "step": 7453 }, { "epoch": 0.53, "grad_norm": 3.5661756053631866, "learning_rate": 4.684332064478784e-06, "loss": 0.1858, "step": 7454 }, { "epoch": 0.53, "grad_norm": 4.563720843223218, "learning_rate": 4.683173073504986e-06, "loss": 1.1045, "step": 7455 }, { "epoch": 0.53, "grad_norm": 3.0320905027216516, "learning_rate": 4.6820140996228e-06, "loss": 0.1831, "step": 7456 }, { "epoch": 0.53, "grad_norm": 8.427521126703587, "learning_rate": 4.680855142894751e-06, "loss": 0.1323, "step": 7457 }, { "epoch": 0.53, "grad_norm": 3.9855235920430183, "learning_rate": 4.679696203383358e-06, "loss": 0.2397, "step": 7458 }, { "epoch": 0.53, "grad_norm": 2.839749393821881, "learning_rate": 4.678537281151141e-06, "loss": 0.1327, "step": 7459 }, { "epoch": 0.53, "grad_norm": 11.72089324795949, "learning_rate": 4.677378376260623e-06, "loss": 0.8534, "step": 7460 }, { "epoch": 0.54, "grad_norm": 14.126796489998457, "learning_rate": 4.67621948877432e-06, "loss": 0.8408, "step": 7461 }, { "epoch": 0.54, "grad_norm": 2.866475252996303, "learning_rate": 4.67506061875475e-06, "loss": 0.1669, "step": 7462 }, { "epoch": 0.54, "grad_norm": 3.1801634931652623, "learning_rate": 4.67390176626443e-06, "loss": 0.2595, "step": 7463 }, { "epoch": 0.54, "grad_norm": 5.421037405035942, "learning_rate": 4.672742931365874e-06, "loss": 0.8406, "step": 7464 }, { "epoch": 0.54, "grad_norm": 5.137533875196127, "learning_rate": 4.671584114121599e-06, "loss": 0.7514, "step": 7465 }, { "epoch": 0.54, "grad_norm": 4.148806893646581, "learning_rate": 4.67042531459412e-06, "loss": 0.8927, "step": 7466 }, { "epoch": 0.54, "grad_norm": 3.2300916234669756, "learning_rate": 4.669266532845945e-06, "loss": 0.1409, "step": 7467 }, { "epoch": 0.54, "grad_norm": 9.562316937045482, "learning_rate": 4.6681077689395896e-06, "loss": 0.5583, "step": 7468 }, { "epoch": 0.54, "grad_norm": 3.3826055423149652, "learning_rate": 4.666949022937564e-06, "loss": 0.2857, "step": 7469 }, { "epoch": 0.54, "grad_norm": 2.777060035396032, "learning_rate": 4.665790294902377e-06, "loss": 0.2396, "step": 7470 }, { "epoch": 0.54, "grad_norm": 3.0245418489527895, "learning_rate": 4.66463158489654e-06, "loss": 0.2042, "step": 7471 }, { "epoch": 0.54, "grad_norm": 69.66993437655522, "learning_rate": 4.663472892982558e-06, "loss": 1.1279, "step": 7472 }, { "epoch": 0.54, "grad_norm": 8.43941476332614, "learning_rate": 4.662314219222939e-06, "loss": 0.4689, "step": 7473 }, { "epoch": 0.54, "grad_norm": 14.982836736457585, "learning_rate": 4.661155563680192e-06, "loss": 0.8817, "step": 7474 }, { "epoch": 0.54, "grad_norm": 13.093134329669807, "learning_rate": 4.6599969264168185e-06, "loss": 0.5595, "step": 7475 }, { "epoch": 0.54, "grad_norm": 4.207007973426964, "learning_rate": 4.658838307495324e-06, "loss": 0.247, "step": 7476 }, { "epoch": 0.54, "grad_norm": 3.5767260473131115, "learning_rate": 4.657679706978211e-06, "loss": 0.2124, "step": 7477 }, { "epoch": 0.54, "grad_norm": 18.03041313004919, "learning_rate": 4.656521124927984e-06, "loss": 1.2815, "step": 7478 }, { "epoch": 0.54, "grad_norm": 15.62945141926713, "learning_rate": 4.65536256140714e-06, "loss": 0.8509, "step": 7479 }, { "epoch": 0.54, "grad_norm": 32.636697535918, "learning_rate": 4.654204016478183e-06, "loss": 1.4987, "step": 7480 }, { "epoch": 0.54, "grad_norm": 4.804061144669372, "learning_rate": 4.653045490203609e-06, "loss": 1.1056, "step": 7481 }, { "epoch": 0.54, "grad_norm": 4.2556973508207925, "learning_rate": 4.651886982645917e-06, "loss": 0.1663, "step": 7482 }, { "epoch": 0.54, "grad_norm": 4.454954237536645, "learning_rate": 4.650728493867606e-06, "loss": 0.1877, "step": 7483 }, { "epoch": 0.54, "grad_norm": 27.693009132556217, "learning_rate": 4.64957002393117e-06, "loss": 0.9438, "step": 7484 }, { "epoch": 0.54, "grad_norm": 3.6141168487737843, "learning_rate": 4.6484115728991054e-06, "loss": 0.1573, "step": 7485 }, { "epoch": 0.54, "grad_norm": 4.029261487200071, "learning_rate": 4.647253140833904e-06, "loss": 0.2157, "step": 7486 }, { "epoch": 0.54, "grad_norm": 4.011275549535569, "learning_rate": 4.646094727798061e-06, "loss": 0.2357, "step": 7487 }, { "epoch": 0.54, "grad_norm": 5.225010870749044, "learning_rate": 4.644936333854069e-06, "loss": 0.9716, "step": 7488 }, { "epoch": 0.54, "grad_norm": 21.93667786994771, "learning_rate": 4.643777959064417e-06, "loss": 0.8942, "step": 7489 }, { "epoch": 0.54, "grad_norm": 3.895806090877178, "learning_rate": 4.642619603491598e-06, "loss": 0.279, "step": 7490 }, { "epoch": 0.54, "grad_norm": 3.3136095161860855, "learning_rate": 4.641461267198097e-06, "loss": 0.2046, "step": 7491 }, { "epoch": 0.54, "grad_norm": 6.214075622919922, "learning_rate": 4.640302950246402e-06, "loss": 0.593, "step": 7492 }, { "epoch": 0.54, "grad_norm": 2.2491069290360866, "learning_rate": 4.639144652699002e-06, "loss": 0.1926, "step": 7493 }, { "epoch": 0.54, "grad_norm": 3.1405424909208604, "learning_rate": 4.6379863746183826e-06, "loss": 0.1614, "step": 7494 }, { "epoch": 0.54, "grad_norm": 6.174878372773164, "learning_rate": 4.6368281160670276e-06, "loss": 0.1399, "step": 7495 }, { "epoch": 0.54, "grad_norm": 4.1338168880614905, "learning_rate": 4.635669877107421e-06, "loss": 0.7797, "step": 7496 }, { "epoch": 0.54, "grad_norm": 16.085967089679446, "learning_rate": 4.634511657802046e-06, "loss": 1.1117, "step": 7497 }, { "epoch": 0.54, "grad_norm": 2.9385177048647297, "learning_rate": 4.633353458213384e-06, "loss": 0.1571, "step": 7498 }, { "epoch": 0.54, "grad_norm": 33.400847524246906, "learning_rate": 4.632195278403914e-06, "loss": 0.3951, "step": 7499 }, { "epoch": 0.54, "grad_norm": 2.6295055519825827, "learning_rate": 4.6310371184361165e-06, "loss": 0.236, "step": 7500 }, { "epoch": 0.54, "grad_norm": 2.8754342847358383, "learning_rate": 4.629878978372469e-06, "loss": 0.1857, "step": 7501 }, { "epoch": 0.54, "grad_norm": 3.56302332289166, "learning_rate": 4.628720858275454e-06, "loss": 0.2925, "step": 7502 }, { "epoch": 0.54, "grad_norm": 3.128350351505575, "learning_rate": 4.62756275820754e-06, "loss": 0.3294, "step": 7503 }, { "epoch": 0.54, "grad_norm": 3.3289926370357583, "learning_rate": 4.626404678231206e-06, "loss": 0.2368, "step": 7504 }, { "epoch": 0.54, "grad_norm": 14.668938861217406, "learning_rate": 4.625246618408925e-06, "loss": 0.8169, "step": 7505 }, { "epoch": 0.54, "grad_norm": 4.72977458912884, "learning_rate": 4.62408857880317e-06, "loss": 0.3602, "step": 7506 }, { "epoch": 0.54, "grad_norm": 24.223920204648437, "learning_rate": 4.6229305594764135e-06, "loss": 1.0208, "step": 7507 }, { "epoch": 0.54, "grad_norm": 3.581175338102243, "learning_rate": 4.6217725604911274e-06, "loss": 0.2047, "step": 7508 }, { "epoch": 0.54, "grad_norm": 3.200722768033868, "learning_rate": 4.620614581909778e-06, "loss": 0.1257, "step": 7509 }, { "epoch": 0.54, "grad_norm": 3.4997511637426633, "learning_rate": 4.6194566237948365e-06, "loss": 0.2094, "step": 7510 }, { "epoch": 0.54, "grad_norm": 3.0376035763787455, "learning_rate": 4.618298686208771e-06, "loss": 0.1405, "step": 7511 }, { "epoch": 0.54, "grad_norm": 3.449772082167016, "learning_rate": 4.617140769214045e-06, "loss": 0.2441, "step": 7512 }, { "epoch": 0.54, "grad_norm": 8.538530842636638, "learning_rate": 4.615982872873127e-06, "loss": 0.6614, "step": 7513 }, { "epoch": 0.54, "grad_norm": 7.629105271475897, "learning_rate": 4.614824997248479e-06, "loss": 0.2101, "step": 7514 }, { "epoch": 0.54, "grad_norm": 2.6604031659815153, "learning_rate": 4.613667142402563e-06, "loss": 0.1591, "step": 7515 }, { "epoch": 0.54, "grad_norm": 2.672821993921933, "learning_rate": 4.6125093083978435e-06, "loss": 0.2069, "step": 7516 }, { "epoch": 0.54, "grad_norm": 4.379510092137377, "learning_rate": 4.6113514952967795e-06, "loss": 0.854, "step": 7517 }, { "epoch": 0.54, "grad_norm": 3.022581323723151, "learning_rate": 4.610193703161832e-06, "loss": 0.2165, "step": 7518 }, { "epoch": 0.54, "grad_norm": 7.737672800808506, "learning_rate": 4.6090359320554575e-06, "loss": 0.2065, "step": 7519 }, { "epoch": 0.54, "grad_norm": 21.96496223698355, "learning_rate": 4.607878182040115e-06, "loss": 0.6164, "step": 7520 }, { "epoch": 0.54, "grad_norm": 4.8947843370453255, "learning_rate": 4.6067204531782605e-06, "loss": 0.2198, "step": 7521 }, { "epoch": 0.54, "grad_norm": 3.654469706858648, "learning_rate": 4.6055627455323495e-06, "loss": 0.2637, "step": 7522 }, { "epoch": 0.54, "grad_norm": 2.921508948630713, "learning_rate": 4.6044050591648345e-06, "loss": 0.2492, "step": 7523 }, { "epoch": 0.54, "grad_norm": 3.016680749950015, "learning_rate": 4.603247394138168e-06, "loss": 0.2332, "step": 7524 }, { "epoch": 0.54, "grad_norm": 12.547841316264835, "learning_rate": 4.602089750514805e-06, "loss": 0.5885, "step": 7525 }, { "epoch": 0.54, "grad_norm": 3.1345703139366163, "learning_rate": 4.6009321283571954e-06, "loss": 0.1211, "step": 7526 }, { "epoch": 0.54, "grad_norm": 19.368605156826018, "learning_rate": 4.599774527727784e-06, "loss": 0.784, "step": 7527 }, { "epoch": 0.54, "grad_norm": 4.050498511920078, "learning_rate": 4.598616948689023e-06, "loss": 0.1887, "step": 7528 }, { "epoch": 0.54, "grad_norm": 2.8777326829675856, "learning_rate": 4.597459391303357e-06, "loss": 0.1574, "step": 7529 }, { "epoch": 0.54, "grad_norm": 3.567164302617554, "learning_rate": 4.596301855633233e-06, "loss": 0.2629, "step": 7530 }, { "epoch": 0.54, "grad_norm": 4.53338409307442, "learning_rate": 4.595144341741097e-06, "loss": 0.1936, "step": 7531 }, { "epoch": 0.54, "grad_norm": 5.857766404315423, "learning_rate": 4.59398684968939e-06, "loss": 0.3694, "step": 7532 }, { "epoch": 0.54, "grad_norm": 2.659149428665128, "learning_rate": 4.592829379540558e-06, "loss": 0.1306, "step": 7533 }, { "epoch": 0.54, "grad_norm": 2.2591915134248968, "learning_rate": 4.591671931357038e-06, "loss": 0.1326, "step": 7534 }, { "epoch": 0.54, "grad_norm": 4.066097252981131, "learning_rate": 4.590514505201271e-06, "loss": 1.062, "step": 7535 }, { "epoch": 0.54, "grad_norm": 5.331905209009681, "learning_rate": 4.589357101135699e-06, "loss": 0.2457, "step": 7536 }, { "epoch": 0.54, "grad_norm": 2.9795111596165813, "learning_rate": 4.588199719222754e-06, "loss": 0.1533, "step": 7537 }, { "epoch": 0.54, "grad_norm": 14.023256266606165, "learning_rate": 4.5870423595248795e-06, "loss": 1.0733, "step": 7538 }, { "epoch": 0.54, "grad_norm": 12.704364354762447, "learning_rate": 4.585885022104503e-06, "loss": 0.6441, "step": 7539 }, { "epoch": 0.54, "grad_norm": 3.8125430620609806, "learning_rate": 4.5847277070240635e-06, "loss": 0.2752, "step": 7540 }, { "epoch": 0.54, "grad_norm": 2.473374495556788, "learning_rate": 4.583570414345991e-06, "loss": 0.0948, "step": 7541 }, { "epoch": 0.54, "grad_norm": 3.182618300399041, "learning_rate": 4.5824131441327205e-06, "loss": 0.1755, "step": 7542 }, { "epoch": 0.54, "grad_norm": 11.62165778019586, "learning_rate": 4.581255896446678e-06, "loss": 0.746, "step": 7543 }, { "epoch": 0.54, "grad_norm": 9.016065155141087, "learning_rate": 4.580098671350296e-06, "loss": 0.4211, "step": 7544 }, { "epoch": 0.54, "grad_norm": 22.423302639328746, "learning_rate": 4.578941468906001e-06, "loss": 0.9591, "step": 7545 }, { "epoch": 0.54, "grad_norm": 16.613779284126192, "learning_rate": 4.57778428917622e-06, "loss": 0.327, "step": 7546 }, { "epoch": 0.54, "grad_norm": 3.189206043025753, "learning_rate": 4.576627132223379e-06, "loss": 0.1925, "step": 7547 }, { "epoch": 0.54, "grad_norm": 4.186749644953259, "learning_rate": 4.5754699981099e-06, "loss": 0.2804, "step": 7548 }, { "epoch": 0.54, "grad_norm": 11.865774975211234, "learning_rate": 4.574312886898209e-06, "loss": 0.9135, "step": 7549 }, { "epoch": 0.54, "grad_norm": 3.319070721750035, "learning_rate": 4.573155798650728e-06, "loss": 0.1428, "step": 7550 }, { "epoch": 0.54, "grad_norm": 2.998281585123664, "learning_rate": 4.571998733429874e-06, "loss": 0.157, "step": 7551 }, { "epoch": 0.54, "grad_norm": 7.75992040389267, "learning_rate": 4.5708416912980684e-06, "loss": 0.425, "step": 7552 }, { "epoch": 0.54, "grad_norm": 4.101221518424007, "learning_rate": 4.569684672317729e-06, "loss": 0.9027, "step": 7553 }, { "epoch": 0.54, "grad_norm": 7.585703548390409, "learning_rate": 4.568527676551273e-06, "loss": 0.9291, "step": 7554 }, { "epoch": 0.54, "grad_norm": 4.330208792029302, "learning_rate": 4.567370704061115e-06, "loss": 0.2701, "step": 7555 }, { "epoch": 0.54, "grad_norm": 4.243353644492875, "learning_rate": 4.566213754909671e-06, "loss": 0.7628, "step": 7556 }, { "epoch": 0.54, "grad_norm": 5.643224146522838, "learning_rate": 4.565056829159352e-06, "loss": 0.2485, "step": 7557 }, { "epoch": 0.54, "grad_norm": 4.30498242664271, "learning_rate": 4.5638999268725706e-06, "loss": 0.7509, "step": 7558 }, { "epoch": 0.54, "grad_norm": 3.3444442443267017, "learning_rate": 4.562743048111739e-06, "loss": 0.1626, "step": 7559 }, { "epoch": 0.54, "grad_norm": 3.2118618330691717, "learning_rate": 4.561586192939263e-06, "loss": 0.185, "step": 7560 }, { "epoch": 0.54, "grad_norm": 2.462774936727825, "learning_rate": 4.560429361417553e-06, "loss": 0.1804, "step": 7561 }, { "epoch": 0.54, "grad_norm": 7.176471861579383, "learning_rate": 4.559272553609016e-06, "loss": 0.4346, "step": 7562 }, { "epoch": 0.54, "grad_norm": 19.736363804433385, "learning_rate": 4.558115769576057e-06, "loss": 0.6452, "step": 7563 }, { "epoch": 0.54, "grad_norm": 16.18289799899416, "learning_rate": 4.556959009381077e-06, "loss": 0.5862, "step": 7564 }, { "epoch": 0.54, "grad_norm": 3.0129860150562457, "learning_rate": 4.5558022730864835e-06, "loss": 0.1841, "step": 7565 }, { "epoch": 0.54, "grad_norm": 5.121733232529207, "learning_rate": 4.5546455607546755e-06, "loss": 0.886, "step": 7566 }, { "epoch": 0.54, "grad_norm": 2.6620721897209987, "learning_rate": 4.5534888724480525e-06, "loss": 0.1216, "step": 7567 }, { "epoch": 0.54, "grad_norm": 2.866465885569533, "learning_rate": 4.552332208229017e-06, "loss": 0.1786, "step": 7568 }, { "epoch": 0.54, "grad_norm": 9.918525103659949, "learning_rate": 4.551175568159961e-06, "loss": 0.656, "step": 7569 }, { "epoch": 0.54, "grad_norm": 3.280085065690093, "learning_rate": 4.5500189523032875e-06, "loss": 0.2214, "step": 7570 }, { "epoch": 0.54, "grad_norm": 7.133853285213274, "learning_rate": 4.548862360721387e-06, "loss": 0.5041, "step": 7571 }, { "epoch": 0.54, "grad_norm": 2.8449725321269073, "learning_rate": 4.547705793476654e-06, "loss": 0.1498, "step": 7572 }, { "epoch": 0.54, "grad_norm": 3.492954987932931, "learning_rate": 4.546549250631483e-06, "loss": 0.1841, "step": 7573 }, { "epoch": 0.54, "grad_norm": 86.19906977195708, "learning_rate": 4.545392732248264e-06, "loss": 0.9323, "step": 7574 }, { "epoch": 0.54, "grad_norm": 3.5923014331572394, "learning_rate": 4.544236238389384e-06, "loss": 0.1737, "step": 7575 }, { "epoch": 0.54, "grad_norm": 3.571080289017431, "learning_rate": 4.543079769117233e-06, "loss": 0.1732, "step": 7576 }, { "epoch": 0.54, "grad_norm": 53.5143833133403, "learning_rate": 4.541923324494202e-06, "loss": 0.8555, "step": 7577 }, { "epoch": 0.54, "grad_norm": 2.8626254744503328, "learning_rate": 4.5407669045826714e-06, "loss": 0.14, "step": 7578 }, { "epoch": 0.54, "grad_norm": 3.8857093529359585, "learning_rate": 4.539610509445029e-06, "loss": 0.2634, "step": 7579 }, { "epoch": 0.54, "grad_norm": 4.377431815855463, "learning_rate": 4.538454139143656e-06, "loss": 0.1406, "step": 7580 }, { "epoch": 0.54, "grad_norm": 4.218688361350336, "learning_rate": 4.537297793740937e-06, "loss": 0.2763, "step": 7581 }, { "epoch": 0.54, "grad_norm": 3.384655502530925, "learning_rate": 4.53614147329925e-06, "loss": 0.1394, "step": 7582 }, { "epoch": 0.54, "grad_norm": 3.665269246099845, "learning_rate": 4.534985177880972e-06, "loss": 0.1611, "step": 7583 }, { "epoch": 0.54, "grad_norm": 5.776045318077126, "learning_rate": 4.533828907548487e-06, "loss": 0.9735, "step": 7584 }, { "epoch": 0.54, "grad_norm": 3.6242905075777094, "learning_rate": 4.532672662364166e-06, "loss": 0.2399, "step": 7585 }, { "epoch": 0.54, "grad_norm": 4.529879957876555, "learning_rate": 4.531516442390389e-06, "loss": 0.6847, "step": 7586 }, { "epoch": 0.54, "grad_norm": 3.9329686894910676, "learning_rate": 4.530360247689523e-06, "loss": 0.2195, "step": 7587 }, { "epoch": 0.54, "grad_norm": 3.705267979936853, "learning_rate": 4.529204078323946e-06, "loss": 0.1983, "step": 7588 }, { "epoch": 0.54, "grad_norm": 19.90984941410241, "learning_rate": 4.528047934356025e-06, "loss": 1.0248, "step": 7589 }, { "epoch": 0.54, "grad_norm": 15.034460512585378, "learning_rate": 4.526891815848133e-06, "loss": 0.8546, "step": 7590 }, { "epoch": 0.54, "grad_norm": 28.044714157617076, "learning_rate": 4.525735722862636e-06, "loss": 1.1731, "step": 7591 }, { "epoch": 0.54, "grad_norm": 3.4864027204178027, "learning_rate": 4.524579655461902e-06, "loss": 0.2135, "step": 7592 }, { "epoch": 0.54, "grad_norm": 4.6031943782685625, "learning_rate": 4.5234236137082955e-06, "loss": 0.2858, "step": 7593 }, { "epoch": 0.54, "grad_norm": 4.027201534956994, "learning_rate": 4.522267597664181e-06, "loss": 0.2728, "step": 7594 }, { "epoch": 0.54, "grad_norm": 21.693971705263298, "learning_rate": 4.521111607391921e-06, "loss": 1.2024, "step": 7595 }, { "epoch": 0.54, "grad_norm": 11.199673800859264, "learning_rate": 4.519955642953879e-06, "loss": 0.5053, "step": 7596 }, { "epoch": 0.54, "grad_norm": 12.600748735054031, "learning_rate": 4.518799704412411e-06, "loss": 0.5988, "step": 7597 }, { "epoch": 0.54, "grad_norm": 3.579162640464088, "learning_rate": 4.5176437918298796e-06, "loss": 0.1649, "step": 7598 }, { "epoch": 0.54, "grad_norm": 19.601028340413517, "learning_rate": 4.516487905268639e-06, "loss": 0.6806, "step": 7599 }, { "epoch": 0.54, "grad_norm": 2.910627109276003, "learning_rate": 4.515332044791045e-06, "loss": 0.2061, "step": 7600 }, { "epoch": 0.55, "grad_norm": 10.29702421410098, "learning_rate": 4.5141762104594525e-06, "loss": 0.8085, "step": 7601 }, { "epoch": 0.55, "grad_norm": 3.5790579696973848, "learning_rate": 4.513020402336215e-06, "loss": 0.1392, "step": 7602 }, { "epoch": 0.55, "grad_norm": 3.067971627558925, "learning_rate": 4.511864620483684e-06, "loss": 0.1493, "step": 7603 }, { "epoch": 0.55, "grad_norm": 4.003398744405604, "learning_rate": 4.510708864964207e-06, "loss": 0.809, "step": 7604 }, { "epoch": 0.55, "grad_norm": 4.900264353568015, "learning_rate": 4.509553135840137e-06, "loss": 0.9501, "step": 7605 }, { "epoch": 0.55, "grad_norm": 3.5872706071198626, "learning_rate": 4.508397433173817e-06, "loss": 0.1835, "step": 7606 }, { "epoch": 0.55, "grad_norm": 4.7332032938271285, "learning_rate": 4.507241757027596e-06, "loss": 1.1801, "step": 7607 }, { "epoch": 0.55, "grad_norm": 15.374378577496799, "learning_rate": 4.506086107463816e-06, "loss": 0.7441, "step": 7608 }, { "epoch": 0.55, "grad_norm": 3.9096371269188372, "learning_rate": 4.504930484544822e-06, "loss": 0.2023, "step": 7609 }, { "epoch": 0.55, "grad_norm": 3.1203440444454693, "learning_rate": 4.5037748883329555e-06, "loss": 0.1484, "step": 7610 }, { "epoch": 0.55, "grad_norm": 2.7137316749998104, "learning_rate": 4.502619318890554e-06, "loss": 0.0884, "step": 7611 }, { "epoch": 0.55, "grad_norm": 2.8601447769430934, "learning_rate": 4.501463776279957e-06, "loss": 0.0964, "step": 7612 }, { "epoch": 0.55, "grad_norm": 5.688818709348695, "learning_rate": 4.500308260563504e-06, "loss": 0.7951, "step": 7613 }, { "epoch": 0.55, "grad_norm": 23.556902506817956, "learning_rate": 4.499152771803527e-06, "loss": 0.9579, "step": 7614 }, { "epoch": 0.55, "grad_norm": 3.7257043374853596, "learning_rate": 4.497997310062362e-06, "loss": 0.1975, "step": 7615 }, { "epoch": 0.55, "grad_norm": 3.603152176759351, "learning_rate": 4.496841875402343e-06, "loss": 0.195, "step": 7616 }, { "epoch": 0.55, "grad_norm": 22.14972111875987, "learning_rate": 4.495686467885799e-06, "loss": 0.7153, "step": 7617 }, { "epoch": 0.55, "grad_norm": 2.476664401993516, "learning_rate": 4.494531087575062e-06, "loss": 0.1048, "step": 7618 }, { "epoch": 0.55, "grad_norm": 4.472031296415619, "learning_rate": 4.49337573453246e-06, "loss": 0.945, "step": 7619 }, { "epoch": 0.55, "grad_norm": 16.718451676434974, "learning_rate": 4.492220408820319e-06, "loss": 0.9747, "step": 7620 }, { "epoch": 0.55, "grad_norm": 17.491234781867025, "learning_rate": 4.491065110500965e-06, "loss": 0.7541, "step": 7621 }, { "epoch": 0.55, "grad_norm": 13.762417704646987, "learning_rate": 4.489909839636721e-06, "loss": 0.7841, "step": 7622 }, { "epoch": 0.55, "grad_norm": 15.826088320179766, "learning_rate": 4.488754596289913e-06, "loss": 1.112, "step": 7623 }, { "epoch": 0.55, "grad_norm": 12.445606793098468, "learning_rate": 4.487599380522856e-06, "loss": 0.4993, "step": 7624 }, { "epoch": 0.55, "grad_norm": 15.722918622040192, "learning_rate": 4.486444192397876e-06, "loss": 0.7276, "step": 7625 }, { "epoch": 0.55, "grad_norm": 22.884004433325238, "learning_rate": 4.485289031977286e-06, "loss": 0.4174, "step": 7626 }, { "epoch": 0.55, "grad_norm": 3.3170481078033975, "learning_rate": 4.484133899323404e-06, "loss": 0.2539, "step": 7627 }, { "epoch": 0.55, "grad_norm": 3.147274544790573, "learning_rate": 4.482978794498547e-06, "loss": 0.2241, "step": 7628 }, { "epoch": 0.55, "grad_norm": 3.5821945043057206, "learning_rate": 4.481823717565026e-06, "loss": 0.1624, "step": 7629 }, { "epoch": 0.55, "grad_norm": 2.936378719469817, "learning_rate": 4.4806686685851545e-06, "loss": 0.1229, "step": 7630 }, { "epoch": 0.55, "grad_norm": 14.006409421416532, "learning_rate": 4.479513647621241e-06, "loss": 0.6443, "step": 7631 }, { "epoch": 0.55, "grad_norm": 4.520433693931225, "learning_rate": 4.478358654735598e-06, "loss": 0.8365, "step": 7632 }, { "epoch": 0.55, "grad_norm": 4.117612923251746, "learning_rate": 4.47720368999053e-06, "loss": 0.1809, "step": 7633 }, { "epoch": 0.55, "grad_norm": 60.99966256161667, "learning_rate": 4.476048753448344e-06, "loss": 1.643, "step": 7634 }, { "epoch": 0.55, "grad_norm": 10.59618236328884, "learning_rate": 4.474893845171347e-06, "loss": 0.3861, "step": 7635 }, { "epoch": 0.55, "grad_norm": 2.533433680840275, "learning_rate": 4.473738965221838e-06, "loss": 0.2206, "step": 7636 }, { "epoch": 0.55, "grad_norm": 2.566296620976424, "learning_rate": 4.4725841136621185e-06, "loss": 0.156, "step": 7637 }, { "epoch": 0.55, "grad_norm": 20.612602076927278, "learning_rate": 4.47142929055449e-06, "loss": 0.8047, "step": 7638 }, { "epoch": 0.55, "grad_norm": 12.974220337397291, "learning_rate": 4.470274495961252e-06, "loss": 0.789, "step": 7639 }, { "epoch": 0.55, "grad_norm": 32.73564829092355, "learning_rate": 4.4691197299446975e-06, "loss": 0.7641, "step": 7640 }, { "epoch": 0.55, "grad_norm": 4.736974430834199, "learning_rate": 4.4679649925671244e-06, "loss": 0.6895, "step": 7641 }, { "epoch": 0.55, "grad_norm": 4.587026128076044, "learning_rate": 4.466810283890828e-06, "loss": 0.8804, "step": 7642 }, { "epoch": 0.55, "grad_norm": 10.358700291728969, "learning_rate": 4.465655603978097e-06, "loss": 0.5345, "step": 7643 }, { "epoch": 0.55, "grad_norm": 2.555773856493608, "learning_rate": 4.464500952891225e-06, "loss": 0.1373, "step": 7644 }, { "epoch": 0.55, "grad_norm": 18.733129930079283, "learning_rate": 4.463346330692498e-06, "loss": 0.9616, "step": 7645 }, { "epoch": 0.55, "grad_norm": 16.80867996611579, "learning_rate": 4.4621917374442055e-06, "loss": 0.8169, "step": 7646 }, { "epoch": 0.55, "grad_norm": 4.744840211009416, "learning_rate": 4.461037173208635e-06, "loss": 0.7722, "step": 7647 }, { "epoch": 0.55, "grad_norm": 15.653438825769582, "learning_rate": 4.459882638048068e-06, "loss": 0.9502, "step": 7648 }, { "epoch": 0.55, "grad_norm": 20.209348288500852, "learning_rate": 4.458728132024787e-06, "loss": 0.8434, "step": 7649 }, { "epoch": 0.55, "grad_norm": 2.7671689069157734, "learning_rate": 4.457573655201075e-06, "loss": 0.0914, "step": 7650 }, { "epoch": 0.55, "grad_norm": 4.801507465808787, "learning_rate": 4.456419207639211e-06, "loss": 0.8041, "step": 7651 }, { "epoch": 0.55, "grad_norm": 24.537977270657766, "learning_rate": 4.455264789401472e-06, "loss": 1.0839, "step": 7652 }, { "epoch": 0.55, "grad_norm": 3.644690935763437, "learning_rate": 4.454110400550138e-06, "loss": 0.1933, "step": 7653 }, { "epoch": 0.55, "grad_norm": 29.302792047623942, "learning_rate": 4.45295604114748e-06, "loss": 1.2286, "step": 7654 }, { "epoch": 0.55, "grad_norm": 26.165220597989894, "learning_rate": 4.451801711255773e-06, "loss": 0.5415, "step": 7655 }, { "epoch": 0.55, "grad_norm": 22.961186463354277, "learning_rate": 4.4506474109372894e-06, "loss": 0.774, "step": 7656 }, { "epoch": 0.55, "grad_norm": 3.2564598794969135, "learning_rate": 4.4494931402542975e-06, "loss": 0.1837, "step": 7657 }, { "epoch": 0.55, "grad_norm": 4.195336825034396, "learning_rate": 4.4483388992690685e-06, "loss": 0.2463, "step": 7658 }, { "epoch": 0.55, "grad_norm": 23.183768555612737, "learning_rate": 4.447184688043869e-06, "loss": 0.5659, "step": 7659 }, { "epoch": 0.55, "grad_norm": 4.824024152243695, "learning_rate": 4.446030506640961e-06, "loss": 0.9075, "step": 7660 }, { "epoch": 0.55, "grad_norm": 3.089602032525215, "learning_rate": 4.444876355122608e-06, "loss": 0.1463, "step": 7661 }, { "epoch": 0.55, "grad_norm": 4.1409387716968435, "learning_rate": 4.443722233551078e-06, "loss": 0.2932, "step": 7662 }, { "epoch": 0.55, "grad_norm": 3.4401623461437962, "learning_rate": 4.4425681419886255e-06, "loss": 0.3126, "step": 7663 }, { "epoch": 0.55, "grad_norm": 3.5020813118249894, "learning_rate": 4.441414080497514e-06, "loss": 0.2807, "step": 7664 }, { "epoch": 0.55, "grad_norm": 10.16970192649489, "learning_rate": 4.440260049139996e-06, "loss": 0.5862, "step": 7665 }, { "epoch": 0.55, "grad_norm": 2.813212384211566, "learning_rate": 4.439106047978331e-06, "loss": 0.1196, "step": 7666 }, { "epoch": 0.55, "grad_norm": 4.460936794057384, "learning_rate": 4.437952077074772e-06, "loss": 0.9259, "step": 7667 }, { "epoch": 0.55, "grad_norm": 3.465350801117944, "learning_rate": 4.436798136491569e-06, "loss": 0.2461, "step": 7668 }, { "epoch": 0.55, "grad_norm": 13.29369722370582, "learning_rate": 4.435644226290976e-06, "loss": 0.6126, "step": 7669 }, { "epoch": 0.55, "grad_norm": 29.727003209762888, "learning_rate": 4.434490346535242e-06, "loss": 0.7703, "step": 7670 }, { "epoch": 0.55, "grad_norm": 29.080542837850704, "learning_rate": 4.433336497286613e-06, "loss": 0.7113, "step": 7671 }, { "epoch": 0.55, "grad_norm": 3.719501273075459, "learning_rate": 4.432182678607334e-06, "loss": 0.1593, "step": 7672 }, { "epoch": 0.55, "grad_norm": 7.087286835470592, "learning_rate": 4.431028890559651e-06, "loss": 0.4937, "step": 7673 }, { "epoch": 0.55, "grad_norm": 10.466252108813617, "learning_rate": 4.4298751332058045e-06, "loss": 0.4619, "step": 7674 }, { "epoch": 0.55, "grad_norm": 5.365974382846155, "learning_rate": 4.428721406608036e-06, "loss": 0.7684, "step": 7675 }, { "epoch": 0.55, "grad_norm": 14.741145158052012, "learning_rate": 4.427567710828588e-06, "loss": 1.1646, "step": 7676 }, { "epoch": 0.55, "grad_norm": 3.288564281215861, "learning_rate": 4.426414045929692e-06, "loss": 0.2273, "step": 7677 }, { "epoch": 0.55, "grad_norm": 3.206777881165688, "learning_rate": 4.42526041197359e-06, "loss": 0.1739, "step": 7678 }, { "epoch": 0.55, "grad_norm": 18.010381826253667, "learning_rate": 4.424106809022511e-06, "loss": 0.4941, "step": 7679 }, { "epoch": 0.55, "grad_norm": 3.161477863397309, "learning_rate": 4.42295323713869e-06, "loss": 0.2404, "step": 7680 }, { "epoch": 0.55, "grad_norm": 2.8283952353395825, "learning_rate": 4.421799696384359e-06, "loss": 0.2062, "step": 7681 }, { "epoch": 0.55, "grad_norm": 3.401657638464138, "learning_rate": 4.420646186821744e-06, "loss": 0.2101, "step": 7682 }, { "epoch": 0.55, "grad_norm": 3.2512157385985088, "learning_rate": 4.4194927085130775e-06, "loss": 0.238, "step": 7683 }, { "epoch": 0.55, "grad_norm": 3.4067660838129488, "learning_rate": 4.4183392615205785e-06, "loss": 0.2515, "step": 7684 }, { "epoch": 0.55, "grad_norm": 3.0554993406264046, "learning_rate": 4.417185845906477e-06, "loss": 0.1692, "step": 7685 }, { "epoch": 0.55, "grad_norm": 21.720134017747323, "learning_rate": 4.4160324617329915e-06, "loss": 0.5927, "step": 7686 }, { "epoch": 0.55, "grad_norm": 4.134228686678323, "learning_rate": 4.414879109062345e-06, "loss": 0.2287, "step": 7687 }, { "epoch": 0.55, "grad_norm": 3.171874124830135, "learning_rate": 4.413725787956755e-06, "loss": 0.2269, "step": 7688 }, { "epoch": 0.55, "grad_norm": 2.788892833712293, "learning_rate": 4.4125724984784405e-06, "loss": 0.1006, "step": 7689 }, { "epoch": 0.55, "grad_norm": 3.6772475423211772, "learning_rate": 4.411419240689615e-06, "loss": 0.1668, "step": 7690 }, { "epoch": 0.55, "grad_norm": 3.140235779334527, "learning_rate": 4.410266014652495e-06, "loss": 0.1771, "step": 7691 }, { "epoch": 0.55, "grad_norm": 3.112497319358312, "learning_rate": 4.40911282042929e-06, "loss": 0.1348, "step": 7692 }, { "epoch": 0.55, "grad_norm": 3.651228587969006, "learning_rate": 4.407959658082213e-06, "loss": 0.1509, "step": 7693 }, { "epoch": 0.55, "grad_norm": 4.818739376423802, "learning_rate": 4.4068065276734695e-06, "loss": 0.9108, "step": 7694 }, { "epoch": 0.55, "grad_norm": 18.752557293403658, "learning_rate": 4.405653429265272e-06, "loss": 0.7383, "step": 7695 }, { "epoch": 0.55, "grad_norm": 4.12992073793284, "learning_rate": 4.40450036291982e-06, "loss": 0.7068, "step": 7696 }, { "epoch": 0.55, "grad_norm": 18.421124413357603, "learning_rate": 4.403347328699319e-06, "loss": 0.8103, "step": 7697 }, { "epoch": 0.55, "grad_norm": 4.903587336473342, "learning_rate": 4.40219432666597e-06, "loss": 0.9776, "step": 7698 }, { "epoch": 0.55, "grad_norm": 3.611678113415652, "learning_rate": 4.401041356881977e-06, "loss": 0.1476, "step": 7699 }, { "epoch": 0.55, "grad_norm": 23.106693760136302, "learning_rate": 4.399888419409534e-06, "loss": 1.6598, "step": 7700 }, { "epoch": 0.55, "grad_norm": 3.93459992690095, "learning_rate": 4.39873551431084e-06, "loss": 0.2517, "step": 7701 }, { "epoch": 0.55, "grad_norm": 3.289836590184524, "learning_rate": 4.397582641648088e-06, "loss": 0.1876, "step": 7702 }, { "epoch": 0.55, "grad_norm": 3.6100591284974946, "learning_rate": 4.396429801483472e-06, "loss": 0.1605, "step": 7703 }, { "epoch": 0.55, "grad_norm": 4.068374190084175, "learning_rate": 4.395276993879185e-06, "loss": 0.21, "step": 7704 }, { "epoch": 0.55, "grad_norm": 4.1459241239539475, "learning_rate": 4.394124218897414e-06, "loss": 0.3362, "step": 7705 }, { "epoch": 0.55, "grad_norm": 3.840723001696674, "learning_rate": 4.392971476600349e-06, "loss": 0.2357, "step": 7706 }, { "epoch": 0.55, "grad_norm": 3.141799261869704, "learning_rate": 4.391818767050176e-06, "loss": 0.1394, "step": 7707 }, { "epoch": 0.55, "grad_norm": 10.880383719406614, "learning_rate": 4.390666090309078e-06, "loss": 0.7331, "step": 7708 }, { "epoch": 0.55, "grad_norm": 33.651476234938826, "learning_rate": 4.3895134464392365e-06, "loss": 0.8192, "step": 7709 }, { "epoch": 0.55, "grad_norm": 3.3235564871184415, "learning_rate": 4.388360835502835e-06, "loss": 0.1478, "step": 7710 }, { "epoch": 0.55, "grad_norm": 4.081904302877537, "learning_rate": 4.3872082575620494e-06, "loss": 0.2764, "step": 7711 }, { "epoch": 0.55, "grad_norm": 4.549061461485676, "learning_rate": 4.38605571267906e-06, "loss": 0.7116, "step": 7712 }, { "epoch": 0.55, "grad_norm": 5.162209885340765, "learning_rate": 4.3849032009160415e-06, "loss": 0.792, "step": 7713 }, { "epoch": 0.55, "grad_norm": 4.812421426986124, "learning_rate": 4.383750722335166e-06, "loss": 1.0159, "step": 7714 }, { "epoch": 0.55, "grad_norm": 4.360297382532928, "learning_rate": 4.382598276998607e-06, "loss": 0.1248, "step": 7715 }, { "epoch": 0.55, "grad_norm": 3.5137776910699983, "learning_rate": 4.3814458649685335e-06, "loss": 0.1284, "step": 7716 }, { "epoch": 0.55, "grad_norm": 11.55468711399097, "learning_rate": 4.380293486307114e-06, "loss": 0.4624, "step": 7717 }, { "epoch": 0.55, "grad_norm": 3.1120765680284084, "learning_rate": 4.379141141076517e-06, "loss": 0.2241, "step": 7718 }, { "epoch": 0.55, "grad_norm": 8.985515305758556, "learning_rate": 4.377988829338906e-06, "loss": 0.6432, "step": 7719 }, { "epoch": 0.55, "grad_norm": 3.6486855771045534, "learning_rate": 4.3768365511564405e-06, "loss": 0.2043, "step": 7720 }, { "epoch": 0.55, "grad_norm": 13.922476411338742, "learning_rate": 4.375684306591285e-06, "loss": 0.5517, "step": 7721 }, { "epoch": 0.55, "grad_norm": 5.192500207324022, "learning_rate": 4.374532095705598e-06, "loss": 0.2221, "step": 7722 }, { "epoch": 0.55, "grad_norm": 4.888315966751965, "learning_rate": 4.373379918561537e-06, "loss": 0.8888, "step": 7723 }, { "epoch": 0.55, "grad_norm": 20.77772216792418, "learning_rate": 4.3722277752212585e-06, "loss": 1.2303, "step": 7724 }, { "epoch": 0.55, "grad_norm": 3.542800382894834, "learning_rate": 4.371075665746913e-06, "loss": 0.1904, "step": 7725 }, { "epoch": 0.55, "grad_norm": 6.797188081357745, "learning_rate": 4.369923590200656e-06, "loss": 1.0648, "step": 7726 }, { "epoch": 0.55, "grad_norm": 4.782129943696552, "learning_rate": 4.368771548644637e-06, "loss": 0.9355, "step": 7727 }, { "epoch": 0.55, "grad_norm": 11.268038988910686, "learning_rate": 4.367619541141003e-06, "loss": 0.6893, "step": 7728 }, { "epoch": 0.55, "grad_norm": 3.502399747042577, "learning_rate": 4.366467567751901e-06, "loss": 0.2544, "step": 7729 }, { "epoch": 0.55, "grad_norm": 4.342636945888162, "learning_rate": 4.365315628539475e-06, "loss": 0.664, "step": 7730 }, { "epoch": 0.55, "grad_norm": 17.196195275852602, "learning_rate": 4.364163723565872e-06, "loss": 0.57, "step": 7731 }, { "epoch": 0.55, "grad_norm": 3.5905612554738795, "learning_rate": 4.363011852893226e-06, "loss": 0.2219, "step": 7732 }, { "epoch": 0.55, "grad_norm": 2.4456599940734387, "learning_rate": 4.36186001658368e-06, "loss": 0.165, "step": 7733 }, { "epoch": 0.55, "grad_norm": 21.26057861478246, "learning_rate": 4.36070821469937e-06, "loss": 0.797, "step": 7734 }, { "epoch": 0.55, "grad_norm": 11.781293280923483, "learning_rate": 4.359556447302432e-06, "loss": 0.6984, "step": 7735 }, { "epoch": 0.55, "grad_norm": 3.555338989695431, "learning_rate": 4.358404714455e-06, "loss": 0.223, "step": 7736 }, { "epoch": 0.55, "grad_norm": 7.797733222086011, "learning_rate": 4.357253016219205e-06, "loss": 0.1498, "step": 7737 }, { "epoch": 0.55, "grad_norm": 4.110336230152493, "learning_rate": 4.356101352657178e-06, "loss": 0.2538, "step": 7738 }, { "epoch": 0.55, "grad_norm": 9.773753126731593, "learning_rate": 4.354949723831043e-06, "loss": 0.799, "step": 7739 }, { "epoch": 0.56, "grad_norm": 3.1283617351490043, "learning_rate": 4.3537981298029305e-06, "loss": 0.1221, "step": 7740 }, { "epoch": 0.56, "grad_norm": 3.0888013670972345, "learning_rate": 4.352646570634964e-06, "loss": 0.1614, "step": 7741 }, { "epoch": 0.56, "grad_norm": 4.207316816323747, "learning_rate": 4.351495046389262e-06, "loss": 0.9266, "step": 7742 }, { "epoch": 0.56, "grad_norm": 3.96108381967765, "learning_rate": 4.35034355712795e-06, "loss": 0.9556, "step": 7743 }, { "epoch": 0.56, "grad_norm": 12.525028306722305, "learning_rate": 4.349192102913145e-06, "loss": 0.708, "step": 7744 }, { "epoch": 0.56, "grad_norm": 14.741168748545366, "learning_rate": 4.34804068380696e-06, "loss": 0.6996, "step": 7745 }, { "epoch": 0.56, "grad_norm": 45.74263479641248, "learning_rate": 4.3468892998715114e-06, "loss": 0.8949, "step": 7746 }, { "epoch": 0.56, "grad_norm": 3.1608131479269908, "learning_rate": 4.345737951168916e-06, "loss": 0.1689, "step": 7747 }, { "epoch": 0.56, "grad_norm": 2.745664061264096, "learning_rate": 4.344586637761279e-06, "loss": 0.1115, "step": 7748 }, { "epoch": 0.56, "grad_norm": 23.4306615101827, "learning_rate": 4.343435359710713e-06, "loss": 1.0799, "step": 7749 }, { "epoch": 0.56, "grad_norm": 3.8226884323684134, "learning_rate": 4.3422841170793244e-06, "loss": 0.1559, "step": 7750 }, { "epoch": 0.56, "grad_norm": 8.978416207572172, "learning_rate": 4.341132909929217e-06, "loss": 0.5854, "step": 7751 }, { "epoch": 0.56, "grad_norm": 2.526263225076896, "learning_rate": 4.339981738322497e-06, "loss": 0.1147, "step": 7752 }, { "epoch": 0.56, "grad_norm": 3.20962424888474, "learning_rate": 4.338830602321262e-06, "loss": 0.2095, "step": 7753 }, { "epoch": 0.56, "grad_norm": 3.196628487799195, "learning_rate": 4.337679501987614e-06, "loss": 0.1809, "step": 7754 }, { "epoch": 0.56, "grad_norm": 9.022016527045425, "learning_rate": 4.336528437383652e-06, "loss": 0.1297, "step": 7755 }, { "epoch": 0.56, "grad_norm": 14.373230165965, "learning_rate": 4.335377408571468e-06, "loss": 0.6834, "step": 7756 }, { "epoch": 0.56, "grad_norm": 28.209600704024847, "learning_rate": 4.334226415613156e-06, "loss": 1.3226, "step": 7757 }, { "epoch": 0.56, "grad_norm": 21.988120881704134, "learning_rate": 4.33307545857081e-06, "loss": 1.3755, "step": 7758 }, { "epoch": 0.56, "grad_norm": 21.590945322540705, "learning_rate": 4.331924537506518e-06, "loss": 0.9887, "step": 7759 }, { "epoch": 0.56, "grad_norm": 25.15903183494245, "learning_rate": 4.330773652482367e-06, "loss": 0.5107, "step": 7760 }, { "epoch": 0.56, "grad_norm": 5.015993353256276, "learning_rate": 4.329622803560447e-06, "loss": 0.8438, "step": 7761 }, { "epoch": 0.56, "grad_norm": 4.511734411620741, "learning_rate": 4.328471990802837e-06, "loss": 0.8414, "step": 7762 }, { "epoch": 0.56, "grad_norm": 12.996604148736774, "learning_rate": 4.327321214271622e-06, "loss": 0.524, "step": 7763 }, { "epoch": 0.56, "grad_norm": 3.7096204893333447, "learning_rate": 4.326170474028881e-06, "loss": 0.1847, "step": 7764 }, { "epoch": 0.56, "grad_norm": 3.3949268143855065, "learning_rate": 4.325019770136692e-06, "loss": 0.1557, "step": 7765 }, { "epoch": 0.56, "grad_norm": 13.149212072654597, "learning_rate": 4.323869102657133e-06, "loss": 0.8089, "step": 7766 }, { "epoch": 0.56, "grad_norm": 4.188914753944762, "learning_rate": 4.322718471652278e-06, "loss": 0.9336, "step": 7767 }, { "epoch": 0.56, "grad_norm": 2.8726779992203277, "learning_rate": 4.3215678771841955e-06, "loss": 0.1897, "step": 7768 }, { "epoch": 0.56, "grad_norm": 21.534865476544354, "learning_rate": 4.3204173193149565e-06, "loss": 0.9045, "step": 7769 }, { "epoch": 0.56, "grad_norm": 4.210255589352442, "learning_rate": 4.319266798106634e-06, "loss": 0.7819, "step": 7770 }, { "epoch": 0.56, "grad_norm": 3.391485701736269, "learning_rate": 4.318116313621288e-06, "loss": 0.185, "step": 7771 }, { "epoch": 0.56, "grad_norm": 13.20387666973887, "learning_rate": 4.316965865920989e-06, "loss": 1.0158, "step": 7772 }, { "epoch": 0.56, "grad_norm": 3.6133379665649907, "learning_rate": 4.315815455067793e-06, "loss": 0.2128, "step": 7773 }, { "epoch": 0.56, "grad_norm": 3.3138126514559128, "learning_rate": 4.3146650811237645e-06, "loss": 0.1906, "step": 7774 }, { "epoch": 0.56, "grad_norm": 11.986815677759031, "learning_rate": 4.313514744150961e-06, "loss": 0.4265, "step": 7775 }, { "epoch": 0.56, "grad_norm": 4.436544012073227, "learning_rate": 4.312364444211438e-06, "loss": 0.9985, "step": 7776 }, { "epoch": 0.56, "grad_norm": 3.086019722703584, "learning_rate": 4.311214181367251e-06, "loss": 0.1584, "step": 7777 }, { "epoch": 0.56, "grad_norm": 7.429697783922166, "learning_rate": 4.3100639556804515e-06, "loss": 0.5396, "step": 7778 }, { "epoch": 0.56, "grad_norm": 11.124537282438416, "learning_rate": 4.308913767213092e-06, "loss": 0.5291, "step": 7779 }, { "epoch": 0.56, "grad_norm": 2.871964087253803, "learning_rate": 4.307763616027217e-06, "loss": 0.1395, "step": 7780 }, { "epoch": 0.56, "grad_norm": 4.413197220212672, "learning_rate": 4.3066135021848756e-06, "loss": 0.6162, "step": 7781 }, { "epoch": 0.56, "grad_norm": 4.33263182010812, "learning_rate": 4.305463425748109e-06, "loss": 0.8207, "step": 7782 }, { "epoch": 0.56, "grad_norm": 18.29585717794325, "learning_rate": 4.3043133867789625e-06, "loss": 1.1011, "step": 7783 }, { "epoch": 0.56, "grad_norm": 3.362343556192269, "learning_rate": 4.303163385339477e-06, "loss": 0.1577, "step": 7784 }, { "epoch": 0.56, "grad_norm": 4.179585612550113, "learning_rate": 4.302013421491687e-06, "loss": 0.1354, "step": 7785 }, { "epoch": 0.56, "grad_norm": 4.2455108500935825, "learning_rate": 4.300863495297632e-06, "loss": 0.2148, "step": 7786 }, { "epoch": 0.56, "grad_norm": 3.6117555991323265, "learning_rate": 4.299713606819345e-06, "loss": 0.2082, "step": 7787 }, { "epoch": 0.56, "grad_norm": 3.1538418896030596, "learning_rate": 4.298563756118858e-06, "loss": 0.2069, "step": 7788 }, { "epoch": 0.56, "grad_norm": 4.605233344415508, "learning_rate": 4.297413943258202e-06, "loss": 0.6747, "step": 7789 }, { "epoch": 0.56, "grad_norm": 3.366020872600739, "learning_rate": 4.296264168299404e-06, "loss": 0.1316, "step": 7790 }, { "epoch": 0.56, "grad_norm": 4.60781910458323, "learning_rate": 4.2951144313044915e-06, "loss": 0.8662, "step": 7791 }, { "epoch": 0.56, "grad_norm": 23.045247104602836, "learning_rate": 4.293964732335485e-06, "loss": 0.9978, "step": 7792 }, { "epoch": 0.56, "grad_norm": 3.753418449053727, "learning_rate": 4.29281507145441e-06, "loss": 0.2471, "step": 7793 }, { "epoch": 0.56, "grad_norm": 21.99861408812565, "learning_rate": 4.291665448723285e-06, "loss": 1.0369, "step": 7794 }, { "epoch": 0.56, "grad_norm": 3.548590652509928, "learning_rate": 4.290515864204127e-06, "loss": 0.1611, "step": 7795 }, { "epoch": 0.56, "grad_norm": 3.2264885758251487, "learning_rate": 4.289366317958953e-06, "loss": 0.1661, "step": 7796 }, { "epoch": 0.56, "grad_norm": 30.857066245914602, "learning_rate": 4.2882168100497755e-06, "loss": 0.8381, "step": 7797 }, { "epoch": 0.56, "grad_norm": 4.395461748451165, "learning_rate": 4.2870673405386085e-06, "loss": 0.1639, "step": 7798 }, { "epoch": 0.56, "grad_norm": 20.236096237299734, "learning_rate": 4.2859179094874585e-06, "loss": 0.4571, "step": 7799 }, { "epoch": 0.56, "grad_norm": 2.6580400096570886, "learning_rate": 4.284768516958335e-06, "loss": 0.1438, "step": 7800 }, { "epoch": 0.56, "grad_norm": 13.472645433026736, "learning_rate": 4.283619163013243e-06, "loss": 0.6503, "step": 7801 }, { "epoch": 0.56, "grad_norm": 3.485183098900156, "learning_rate": 4.282469847714185e-06, "loss": 0.1491, "step": 7802 }, { "epoch": 0.56, "grad_norm": 2.7675071301085743, "learning_rate": 4.281320571123164e-06, "loss": 0.1147, "step": 7803 }, { "epoch": 0.56, "grad_norm": 4.356286256887831, "learning_rate": 4.2801713333021775e-06, "loss": 0.1702, "step": 7804 }, { "epoch": 0.56, "grad_norm": 3.751304155712385, "learning_rate": 4.279022134313221e-06, "loss": 0.1784, "step": 7805 }, { "epoch": 0.56, "grad_norm": 3.8818307488757005, "learning_rate": 4.277872974218292e-06, "loss": 0.6737, "step": 7806 }, { "epoch": 0.56, "grad_norm": 8.758253177602969, "learning_rate": 4.276723853079384e-06, "loss": 0.569, "step": 7807 }, { "epoch": 0.56, "grad_norm": 2.7657281097737685, "learning_rate": 4.275574770958485e-06, "loss": 0.1505, "step": 7808 }, { "epoch": 0.56, "grad_norm": 2.8370501406513253, "learning_rate": 4.274425727917586e-06, "loss": 0.2008, "step": 7809 }, { "epoch": 0.56, "grad_norm": 5.058387910376397, "learning_rate": 4.27327672401867e-06, "loss": 1.1888, "step": 7810 }, { "epoch": 0.56, "grad_norm": 3.661945072542115, "learning_rate": 4.272127759323726e-06, "loss": 0.278, "step": 7811 }, { "epoch": 0.56, "grad_norm": 14.028820763515785, "learning_rate": 4.270978833894735e-06, "loss": 1.19, "step": 7812 }, { "epoch": 0.56, "grad_norm": 20.71934715940188, "learning_rate": 4.269829947793675e-06, "loss": 0.7847, "step": 7813 }, { "epoch": 0.56, "grad_norm": 4.651267629855309, "learning_rate": 4.268681101082525e-06, "loss": 0.8764, "step": 7814 }, { "epoch": 0.56, "grad_norm": 2.687619331004777, "learning_rate": 4.267532293823265e-06, "loss": 0.1554, "step": 7815 }, { "epoch": 0.56, "grad_norm": 2.904098608274552, "learning_rate": 4.266383526077863e-06, "loss": 0.1575, "step": 7816 }, { "epoch": 0.56, "grad_norm": 16.188884689566514, "learning_rate": 4.265234797908291e-06, "loss": 0.7604, "step": 7817 }, { "epoch": 0.56, "grad_norm": 3.5318966643081495, "learning_rate": 4.264086109376523e-06, "loss": 0.1934, "step": 7818 }, { "epoch": 0.56, "grad_norm": 2.8271839058689894, "learning_rate": 4.262937460544522e-06, "loss": 0.1733, "step": 7819 }, { "epoch": 0.56, "grad_norm": 24.378880290054738, "learning_rate": 4.261788851474255e-06, "loss": 0.7353, "step": 7820 }, { "epoch": 0.56, "grad_norm": 14.299557214754593, "learning_rate": 4.260640282227687e-06, "loss": 0.6691, "step": 7821 }, { "epoch": 0.56, "grad_norm": 3.849555233615288, "learning_rate": 4.259491752866776e-06, "loss": 0.183, "step": 7822 }, { "epoch": 0.56, "grad_norm": 4.897372816890437, "learning_rate": 4.258343263453482e-06, "loss": 0.7698, "step": 7823 }, { "epoch": 0.56, "grad_norm": 3.0809821521690255, "learning_rate": 4.257194814049761e-06, "loss": 0.1768, "step": 7824 }, { "epoch": 0.56, "grad_norm": 12.563305684872633, "learning_rate": 4.256046404717569e-06, "loss": 0.598, "step": 7825 }, { "epoch": 0.56, "grad_norm": 4.2171517089444315, "learning_rate": 4.254898035518857e-06, "loss": 0.7217, "step": 7826 }, { "epoch": 0.56, "grad_norm": 6.713091097704949, "learning_rate": 4.253749706515578e-06, "loss": 0.8802, "step": 7827 }, { "epoch": 0.56, "grad_norm": 3.636811879321625, "learning_rate": 4.252601417769675e-06, "loss": 0.203, "step": 7828 }, { "epoch": 0.56, "grad_norm": 2.6719917790420395, "learning_rate": 4.251453169343096e-06, "loss": 0.103, "step": 7829 }, { "epoch": 0.56, "grad_norm": 13.455315885145257, "learning_rate": 4.2503049612977874e-06, "loss": 0.8762, "step": 7830 }, { "epoch": 0.56, "grad_norm": 3.220376350277266, "learning_rate": 4.249156793695686e-06, "loss": 0.207, "step": 7831 }, { "epoch": 0.56, "grad_norm": 5.162139792537025, "learning_rate": 4.248008666598735e-06, "loss": 0.7398, "step": 7832 }, { "epoch": 0.56, "grad_norm": 3.8002241790310998, "learning_rate": 4.2468605800688685e-06, "loss": 0.1787, "step": 7833 }, { "epoch": 0.56, "grad_norm": 27.537272092255733, "learning_rate": 4.245712534168024e-06, "loss": 0.9357, "step": 7834 }, { "epoch": 0.56, "grad_norm": 8.669140054215934, "learning_rate": 4.244564528958134e-06, "loss": 0.5312, "step": 7835 }, { "epoch": 0.56, "grad_norm": 26.436207528447696, "learning_rate": 4.243416564501127e-06, "loss": 0.9501, "step": 7836 }, { "epoch": 0.56, "grad_norm": 4.294723168216259, "learning_rate": 4.242268640858933e-06, "loss": 0.1736, "step": 7837 }, { "epoch": 0.56, "grad_norm": 5.006900758003667, "learning_rate": 4.241120758093477e-06, "loss": 1.0086, "step": 7838 }, { "epoch": 0.56, "grad_norm": 3.172472161331908, "learning_rate": 4.239972916266684e-06, "loss": 0.1798, "step": 7839 }, { "epoch": 0.56, "grad_norm": 34.45663165691037, "learning_rate": 4.238825115440476e-06, "loss": 1.37, "step": 7840 }, { "epoch": 0.56, "grad_norm": 3.6693562144559233, "learning_rate": 4.237677355676772e-06, "loss": 0.2113, "step": 7841 }, { "epoch": 0.56, "grad_norm": 3.6908410166706007, "learning_rate": 4.236529637037488e-06, "loss": 0.1929, "step": 7842 }, { "epoch": 0.56, "grad_norm": 3.7540736236844516, "learning_rate": 4.23538195958454e-06, "loss": 0.1971, "step": 7843 }, { "epoch": 0.56, "grad_norm": 16.57648640056134, "learning_rate": 4.234234323379842e-06, "loss": 0.8173, "step": 7844 }, { "epoch": 0.56, "grad_norm": 3.6312336543112727, "learning_rate": 4.233086728485303e-06, "loss": 0.2167, "step": 7845 }, { "epoch": 0.56, "grad_norm": 3.419722241363778, "learning_rate": 4.231939174962833e-06, "loss": 0.1537, "step": 7846 }, { "epoch": 0.56, "grad_norm": 3.4206361185313137, "learning_rate": 4.230791662874335e-06, "loss": 0.2298, "step": 7847 }, { "epoch": 0.56, "grad_norm": 20.57393273424891, "learning_rate": 4.229644192281717e-06, "loss": 1.1766, "step": 7848 }, { "epoch": 0.56, "grad_norm": 2.85051279330546, "learning_rate": 4.228496763246878e-06, "loss": 0.1251, "step": 7849 }, { "epoch": 0.56, "grad_norm": 26.61434614903441, "learning_rate": 4.227349375831719e-06, "loss": 1.5536, "step": 7850 }, { "epoch": 0.56, "grad_norm": 4.3406025958916645, "learning_rate": 4.2262020300981375e-06, "loss": 0.2172, "step": 7851 }, { "epoch": 0.56, "grad_norm": 4.300494566474166, "learning_rate": 4.225054726108028e-06, "loss": 0.2044, "step": 7852 }, { "epoch": 0.56, "grad_norm": 4.56494863664597, "learning_rate": 4.22390746392328e-06, "loss": 0.2557, "step": 7853 }, { "epoch": 0.56, "grad_norm": 13.958444548693326, "learning_rate": 4.222760243605787e-06, "loss": 0.8253, "step": 7854 }, { "epoch": 0.56, "grad_norm": 8.565223929601203, "learning_rate": 4.221613065217438e-06, "loss": 0.4632, "step": 7855 }, { "epoch": 0.56, "grad_norm": 15.097022209401409, "learning_rate": 4.220465928820117e-06, "loss": 0.7643, "step": 7856 }, { "epoch": 0.56, "grad_norm": 3.1824388711000293, "learning_rate": 4.219318834475709e-06, "loss": 0.1536, "step": 7857 }, { "epoch": 0.56, "grad_norm": 3.367003208352913, "learning_rate": 4.218171782246095e-06, "loss": 0.1303, "step": 7858 }, { "epoch": 0.56, "grad_norm": 3.4009678199499147, "learning_rate": 4.217024772193153e-06, "loss": 0.1607, "step": 7859 }, { "epoch": 0.56, "grad_norm": 4.840716200462646, "learning_rate": 4.215877804378762e-06, "loss": 0.9493, "step": 7860 }, { "epoch": 0.56, "grad_norm": 4.5410292413537094, "learning_rate": 4.2147308788647944e-06, "loss": 0.9064, "step": 7861 }, { "epoch": 0.56, "grad_norm": 4.77976882761405, "learning_rate": 4.213583995713123e-06, "loss": 1.148, "step": 7862 }, { "epoch": 0.56, "grad_norm": 10.967811130957932, "learning_rate": 4.21243715498562e-06, "loss": 0.666, "step": 7863 }, { "epoch": 0.56, "grad_norm": 4.169614679525814, "learning_rate": 4.211290356744152e-06, "loss": 0.9008, "step": 7864 }, { "epoch": 0.56, "grad_norm": 3.0391327505149395, "learning_rate": 4.210143601050583e-06, "loss": 0.1883, "step": 7865 }, { "epoch": 0.56, "grad_norm": 22.173427532893, "learning_rate": 4.2089968879667756e-06, "loss": 1.1597, "step": 7866 }, { "epoch": 0.56, "grad_norm": 4.35807663952612, "learning_rate": 4.207850217554592e-06, "loss": 0.9221, "step": 7867 }, { "epoch": 0.56, "grad_norm": 4.124769211927917, "learning_rate": 4.2067035898758915e-06, "loss": 0.2457, "step": 7868 }, { "epoch": 0.56, "grad_norm": 4.145946616928089, "learning_rate": 4.205557004992531e-06, "loss": 0.2293, "step": 7869 }, { "epoch": 0.56, "grad_norm": 4.134143922706683, "learning_rate": 4.20441046296636e-06, "loss": 0.2927, "step": 7870 }, { "epoch": 0.56, "grad_norm": 3.795034307347986, "learning_rate": 4.203263963859234e-06, "loss": 0.259, "step": 7871 }, { "epoch": 0.56, "grad_norm": 3.933247640629749, "learning_rate": 4.202117507733003e-06, "loss": 0.21, "step": 7872 }, { "epoch": 0.56, "grad_norm": 4.609139893549734, "learning_rate": 4.2009710946495104e-06, "loss": 0.2284, "step": 7873 }, { "epoch": 0.56, "grad_norm": 4.492968770266474, "learning_rate": 4.199824724670604e-06, "loss": 0.8416, "step": 7874 }, { "epoch": 0.56, "grad_norm": 3.5652615948032427, "learning_rate": 4.198678397858124e-06, "loss": 0.1713, "step": 7875 }, { "epoch": 0.56, "grad_norm": 13.38865540875418, "learning_rate": 4.197532114273914e-06, "loss": 0.5534, "step": 7876 }, { "epoch": 0.56, "grad_norm": 25.56165797300205, "learning_rate": 4.1963858739798054e-06, "loss": 1.0154, "step": 7877 }, { "epoch": 0.56, "grad_norm": 5.2740516889401405, "learning_rate": 4.195239677037639e-06, "loss": 0.9618, "step": 7878 }, { "epoch": 0.56, "grad_norm": 9.681602981474684, "learning_rate": 4.194093523509244e-06, "loss": 0.1588, "step": 7879 }, { "epoch": 0.57, "grad_norm": 2.9006916910776215, "learning_rate": 4.192947413456454e-06, "loss": 0.1272, "step": 7880 }, { "epoch": 0.57, "grad_norm": 4.73833934302268, "learning_rate": 4.191801346941097e-06, "loss": 0.9287, "step": 7881 }, { "epoch": 0.57, "grad_norm": 4.760743577271051, "learning_rate": 4.1906553240249965e-06, "loss": 0.8112, "step": 7882 }, { "epoch": 0.57, "grad_norm": 3.416227898200276, "learning_rate": 4.189509344769979e-06, "loss": 0.1246, "step": 7883 }, { "epoch": 0.57, "grad_norm": 2.994470221384417, "learning_rate": 4.188363409237863e-06, "loss": 0.1457, "step": 7884 }, { "epoch": 0.57, "grad_norm": 3.9150588324798217, "learning_rate": 4.18721751749047e-06, "loss": 0.1986, "step": 7885 }, { "epoch": 0.57, "grad_norm": 20.406041239797375, "learning_rate": 4.1860716695896155e-06, "loss": 0.9085, "step": 7886 }, { "epoch": 0.57, "grad_norm": 3.331890638503594, "learning_rate": 4.184925865597113e-06, "loss": 0.1914, "step": 7887 }, { "epoch": 0.57, "grad_norm": 4.274392791973141, "learning_rate": 4.183780105574776e-06, "loss": 0.2642, "step": 7888 }, { "epoch": 0.57, "grad_norm": 3.6621792432116718, "learning_rate": 4.182634389584413e-06, "loss": 0.1427, "step": 7889 }, { "epoch": 0.57, "grad_norm": 4.512821606462495, "learning_rate": 4.181488717687829e-06, "loss": 0.8052, "step": 7890 }, { "epoch": 0.57, "grad_norm": 9.626434480496872, "learning_rate": 4.180343089946831e-06, "loss": 0.3516, "step": 7891 }, { "epoch": 0.57, "grad_norm": 5.110929739470367, "learning_rate": 4.179197506423221e-06, "loss": 0.915, "step": 7892 }, { "epoch": 0.57, "grad_norm": 3.1227480353449852, "learning_rate": 4.1780519671787974e-06, "loss": 0.1844, "step": 7893 }, { "epoch": 0.57, "grad_norm": 2.6548725897312755, "learning_rate": 4.17690647227536e-06, "loss": 0.1306, "step": 7894 }, { "epoch": 0.57, "grad_norm": 4.137892853968675, "learning_rate": 4.175761021774704e-06, "loss": 0.9806, "step": 7895 }, { "epoch": 0.57, "grad_norm": 4.267345590378054, "learning_rate": 4.174615615738619e-06, "loss": 0.9694, "step": 7896 }, { "epoch": 0.57, "grad_norm": 15.168567102718299, "learning_rate": 4.173470254228899e-06, "loss": 0.5759, "step": 7897 }, { "epoch": 0.57, "grad_norm": 3.3556648184189632, "learning_rate": 4.172324937307328e-06, "loss": 0.1526, "step": 7898 }, { "epoch": 0.57, "grad_norm": 11.46649650001053, "learning_rate": 4.171179665035694e-06, "loss": 0.667, "step": 7899 }, { "epoch": 0.57, "grad_norm": 17.97294903459322, "learning_rate": 4.170034437475783e-06, "loss": 1.2315, "step": 7900 }, { "epoch": 0.57, "grad_norm": 4.395125125869611, "learning_rate": 4.168889254689371e-06, "loss": 0.7428, "step": 7901 }, { "epoch": 0.57, "grad_norm": 16.5615641842204, "learning_rate": 4.167744116738236e-06, "loss": 0.7278, "step": 7902 }, { "epoch": 0.57, "grad_norm": 4.071487020784662, "learning_rate": 4.166599023684157e-06, "loss": 0.2347, "step": 7903 }, { "epoch": 0.57, "grad_norm": 3.9509700683504754, "learning_rate": 4.165453975588905e-06, "loss": 0.3406, "step": 7904 }, { "epoch": 0.57, "grad_norm": 3.954527038779464, "learning_rate": 4.164308972514253e-06, "loss": 0.3422, "step": 7905 }, { "epoch": 0.57, "grad_norm": 3.4173787723224818, "learning_rate": 4.163164014521969e-06, "loss": 0.0858, "step": 7906 }, { "epoch": 0.57, "grad_norm": 3.0604340311746343, "learning_rate": 4.162019101673818e-06, "loss": 0.2078, "step": 7907 }, { "epoch": 0.57, "grad_norm": 19.296350967444816, "learning_rate": 4.1608742340315644e-06, "loss": 0.6288, "step": 7908 }, { "epoch": 0.57, "grad_norm": 12.886976637489923, "learning_rate": 4.159729411656971e-06, "loss": 0.4074, "step": 7909 }, { "epoch": 0.57, "grad_norm": 3.299546603382184, "learning_rate": 4.1585846346117935e-06, "loss": 0.2056, "step": 7910 }, { "epoch": 0.57, "grad_norm": 39.07390151388489, "learning_rate": 4.157439902957791e-06, "loss": 1.4775, "step": 7911 }, { "epoch": 0.57, "grad_norm": 4.997666456000443, "learning_rate": 4.156295216756718e-06, "loss": 1.0283, "step": 7912 }, { "epoch": 0.57, "grad_norm": 4.242347184023535, "learning_rate": 4.155150576070323e-06, "loss": 0.2983, "step": 7913 }, { "epoch": 0.57, "grad_norm": 3.6307069537083754, "learning_rate": 4.154005980960355e-06, "loss": 0.1559, "step": 7914 }, { "epoch": 0.57, "grad_norm": 4.111265384980245, "learning_rate": 4.152861431488564e-06, "loss": 0.2061, "step": 7915 }, { "epoch": 0.57, "grad_norm": 15.019575711170795, "learning_rate": 4.151716927716692e-06, "loss": 0.6072, "step": 7916 }, { "epoch": 0.57, "grad_norm": 2.929335131227282, "learning_rate": 4.150572469706481e-06, "loss": 0.1668, "step": 7917 }, { "epoch": 0.57, "grad_norm": 12.057347444764968, "learning_rate": 4.149428057519669e-06, "loss": 0.5909, "step": 7918 }, { "epoch": 0.57, "grad_norm": 4.126838675432013, "learning_rate": 4.148283691217993e-06, "loss": 0.1983, "step": 7919 }, { "epoch": 0.57, "grad_norm": 3.296300658696932, "learning_rate": 4.14713937086319e-06, "loss": 0.1891, "step": 7920 }, { "epoch": 0.57, "grad_norm": 3.08426176409165, "learning_rate": 4.145995096516988e-06, "loss": 0.1619, "step": 7921 }, { "epoch": 0.57, "grad_norm": 3.4661787112302562, "learning_rate": 4.144850868241118e-06, "loss": 0.2291, "step": 7922 }, { "epoch": 0.57, "grad_norm": 8.714361403378343, "learning_rate": 4.143706686097308e-06, "loss": 0.4967, "step": 7923 }, { "epoch": 0.57, "grad_norm": 3.2885817941518267, "learning_rate": 4.142562550147282e-06, "loss": 0.148, "step": 7924 }, { "epoch": 0.57, "grad_norm": 4.381944920240485, "learning_rate": 4.141418460452759e-06, "loss": 0.1571, "step": 7925 }, { "epoch": 0.57, "grad_norm": 13.959648129678806, "learning_rate": 4.140274417075461e-06, "loss": 0.5372, "step": 7926 }, { "epoch": 0.57, "grad_norm": 12.923269499502062, "learning_rate": 4.139130420077103e-06, "loss": 0.6183, "step": 7927 }, { "epoch": 0.57, "grad_norm": 9.25820078077636, "learning_rate": 4.1379864695194e-06, "loss": 0.5929, "step": 7928 }, { "epoch": 0.57, "grad_norm": 4.779979362856991, "learning_rate": 4.136842565464066e-06, "loss": 0.73, "step": 7929 }, { "epoch": 0.57, "grad_norm": 1.8641997888029929, "learning_rate": 4.135698707972806e-06, "loss": 0.0918, "step": 7930 }, { "epoch": 0.57, "grad_norm": 24.694202829775325, "learning_rate": 4.1345548971073305e-06, "loss": 1.384, "step": 7931 }, { "epoch": 0.57, "grad_norm": 3.9503841957012797, "learning_rate": 4.133411132929342e-06, "loss": 0.2054, "step": 7932 }, { "epoch": 0.57, "grad_norm": 15.938286336063454, "learning_rate": 4.1322674155005426e-06, "loss": 0.9332, "step": 7933 }, { "epoch": 0.57, "grad_norm": 13.599368540271204, "learning_rate": 4.131123744882633e-06, "loss": 0.688, "step": 7934 }, { "epoch": 0.57, "grad_norm": 12.28733410653313, "learning_rate": 4.129980121137306e-06, "loss": 0.3924, "step": 7935 }, { "epoch": 0.57, "grad_norm": 2.947858044580015, "learning_rate": 4.128836544326262e-06, "loss": 0.1763, "step": 7936 }, { "epoch": 0.57, "grad_norm": 8.887728587607477, "learning_rate": 4.127693014511185e-06, "loss": 0.5363, "step": 7937 }, { "epoch": 0.57, "grad_norm": 9.164495458924135, "learning_rate": 4.12654953175377e-06, "loss": 0.5601, "step": 7938 }, { "epoch": 0.57, "grad_norm": 3.4424658433194235, "learning_rate": 4.125406096115701e-06, "loss": 0.284, "step": 7939 }, { "epoch": 0.57, "grad_norm": 3.574027641661105, "learning_rate": 4.124262707658663e-06, "loss": 0.2074, "step": 7940 }, { "epoch": 0.57, "grad_norm": 3.7950042478869985, "learning_rate": 4.123119366444336e-06, "loss": 0.2673, "step": 7941 }, { "epoch": 0.57, "grad_norm": 3.0014476457313988, "learning_rate": 4.121976072534401e-06, "loss": 0.174, "step": 7942 }, { "epoch": 0.57, "grad_norm": 4.219384422054025, "learning_rate": 4.1208328259905326e-06, "loss": 0.7978, "step": 7943 }, { "epoch": 0.57, "grad_norm": 22.697681179405873, "learning_rate": 4.119689626874406e-06, "loss": 0.963, "step": 7944 }, { "epoch": 0.57, "grad_norm": 4.381285611549313, "learning_rate": 4.1185464752476925e-06, "loss": 0.1568, "step": 7945 }, { "epoch": 0.57, "grad_norm": 4.264420488993958, "learning_rate": 4.117403371172059e-06, "loss": 0.2307, "step": 7946 }, { "epoch": 0.57, "grad_norm": 18.862504052902423, "learning_rate": 4.116260314709172e-06, "loss": 0.8105, "step": 7947 }, { "epoch": 0.57, "grad_norm": 3.3127993570703165, "learning_rate": 4.1151173059207e-06, "loss": 0.2602, "step": 7948 }, { "epoch": 0.57, "grad_norm": 24.186348622493192, "learning_rate": 4.113974344868297e-06, "loss": 0.676, "step": 7949 }, { "epoch": 0.57, "grad_norm": 4.662581047114629, "learning_rate": 4.112831431613624e-06, "loss": 0.8822, "step": 7950 }, { "epoch": 0.57, "grad_norm": 4.7772673378665305, "learning_rate": 4.111688566218337e-06, "loss": 1.0137, "step": 7951 }, { "epoch": 0.57, "grad_norm": 3.5206721201632862, "learning_rate": 4.110545748744091e-06, "loss": 0.2052, "step": 7952 }, { "epoch": 0.57, "grad_norm": 24.931791948152114, "learning_rate": 4.109402979252533e-06, "loss": 1.057, "step": 7953 }, { "epoch": 0.57, "grad_norm": 4.875472102236982, "learning_rate": 4.108260257805314e-06, "loss": 0.8743, "step": 7954 }, { "epoch": 0.57, "grad_norm": 3.610368777385985, "learning_rate": 4.107117584464079e-06, "loss": 0.1632, "step": 7955 }, { "epoch": 0.57, "grad_norm": 10.538147794375723, "learning_rate": 4.105974959290469e-06, "loss": 0.5272, "step": 7956 }, { "epoch": 0.57, "grad_norm": 21.148848358784385, "learning_rate": 4.104832382346129e-06, "loss": 1.0999, "step": 7957 }, { "epoch": 0.57, "grad_norm": 3.077358125131569, "learning_rate": 4.103689853692691e-06, "loss": 0.1425, "step": 7958 }, { "epoch": 0.57, "grad_norm": 3.4954367513738496, "learning_rate": 4.102547373391794e-06, "loss": 0.1363, "step": 7959 }, { "epoch": 0.57, "grad_norm": 4.25644042527023, "learning_rate": 4.101404941505069e-06, "loss": 0.3039, "step": 7960 }, { "epoch": 0.57, "grad_norm": 3.320515408388499, "learning_rate": 4.100262558094147e-06, "loss": 0.1044, "step": 7961 }, { "epoch": 0.57, "grad_norm": 16.636023756989438, "learning_rate": 4.0991202232206526e-06, "loss": 0.5842, "step": 7962 }, { "epoch": 0.57, "grad_norm": 4.838043752603084, "learning_rate": 4.097977936946213e-06, "loss": 0.7811, "step": 7963 }, { "epoch": 0.57, "grad_norm": 20.38751324613349, "learning_rate": 4.096835699332449e-06, "loss": 0.7074, "step": 7964 }, { "epoch": 0.57, "grad_norm": 3.1922359684554005, "learning_rate": 4.09569351044098e-06, "loss": 0.2552, "step": 7965 }, { "epoch": 0.57, "grad_norm": 2.8690039714462263, "learning_rate": 4.094551370333424e-06, "loss": 0.1559, "step": 7966 }, { "epoch": 0.57, "grad_norm": 11.027923725674858, "learning_rate": 4.093409279071392e-06, "loss": 0.668, "step": 7967 }, { "epoch": 0.57, "grad_norm": 9.426945750772191, "learning_rate": 4.092267236716501e-06, "loss": 0.5346, "step": 7968 }, { "epoch": 0.57, "grad_norm": 13.918444645117141, "learning_rate": 4.091125243330355e-06, "loss": 0.7076, "step": 7969 }, { "epoch": 0.57, "grad_norm": 20.303486677794496, "learning_rate": 4.0899832989745605e-06, "loss": 0.9758, "step": 7970 }, { "epoch": 0.57, "grad_norm": 4.52705822363258, "learning_rate": 4.088841403710725e-06, "loss": 0.1871, "step": 7971 }, { "epoch": 0.57, "grad_norm": 3.8522296023630624, "learning_rate": 4.087699557600447e-06, "loss": 0.17, "step": 7972 }, { "epoch": 0.57, "grad_norm": 2.774121450667093, "learning_rate": 4.086557760705323e-06, "loss": 0.095, "step": 7973 }, { "epoch": 0.57, "grad_norm": 3.955929382935799, "learning_rate": 4.08541601308695e-06, "loss": 0.1934, "step": 7974 }, { "epoch": 0.57, "grad_norm": 10.563785837006428, "learning_rate": 4.084274314806922e-06, "loss": 0.4076, "step": 7975 }, { "epoch": 0.57, "grad_norm": 4.425652106658501, "learning_rate": 4.083132665926829e-06, "loss": 0.2087, "step": 7976 }, { "epoch": 0.57, "grad_norm": 4.805970227961957, "learning_rate": 4.0819910665082575e-06, "loss": 0.8569, "step": 7977 }, { "epoch": 0.57, "grad_norm": 3.022859711212388, "learning_rate": 4.080849516612792e-06, "loss": 0.1945, "step": 7978 }, { "epoch": 0.57, "grad_norm": 4.608071857102902, "learning_rate": 4.079708016302017e-06, "loss": 0.9567, "step": 7979 }, { "epoch": 0.57, "grad_norm": 12.349494098142214, "learning_rate": 4.078566565637511e-06, "loss": 0.6169, "step": 7980 }, { "epoch": 0.57, "grad_norm": 22.98821455729083, "learning_rate": 4.0774251646808505e-06, "loss": 1.0803, "step": 7981 }, { "epoch": 0.57, "grad_norm": 14.441158810078951, "learning_rate": 4.076283813493611e-06, "loss": 0.7678, "step": 7982 }, { "epoch": 0.57, "grad_norm": 4.643091912703462, "learning_rate": 4.075142512137363e-06, "loss": 0.8884, "step": 7983 }, { "epoch": 0.57, "grad_norm": 3.8596910360109202, "learning_rate": 4.074001260673678e-06, "loss": 0.8044, "step": 7984 }, { "epoch": 0.57, "grad_norm": 2.633382269454563, "learning_rate": 4.0728600591641165e-06, "loss": 0.1351, "step": 7985 }, { "epoch": 0.57, "grad_norm": 3.168680832261562, "learning_rate": 4.0717189076702466e-06, "loss": 0.1718, "step": 7986 }, { "epoch": 0.57, "grad_norm": 3.261945635611612, "learning_rate": 4.070577806253627e-06, "loss": 0.1595, "step": 7987 }, { "epoch": 0.57, "grad_norm": 3.520550433767513, "learning_rate": 4.069436754975817e-06, "loss": 0.2115, "step": 7988 }, { "epoch": 0.57, "grad_norm": 3.06560264713283, "learning_rate": 4.068295753898373e-06, "loss": 0.1047, "step": 7989 }, { "epoch": 0.57, "grad_norm": 3.5465661017399452, "learning_rate": 4.067154803082845e-06, "loss": 0.2547, "step": 7990 }, { "epoch": 0.57, "grad_norm": 2.9203417142384747, "learning_rate": 4.0660139025907855e-06, "loss": 0.1695, "step": 7991 }, { "epoch": 0.57, "grad_norm": 18.54609675778797, "learning_rate": 4.0648730524837396e-06, "loss": 0.5868, "step": 7992 }, { "epoch": 0.57, "grad_norm": 3.179210458638859, "learning_rate": 4.063732252823253e-06, "loss": 0.2015, "step": 7993 }, { "epoch": 0.57, "grad_norm": 4.008406642747796, "learning_rate": 4.062591503670868e-06, "loss": 0.887, "step": 7994 }, { "epoch": 0.57, "grad_norm": 12.226473711823118, "learning_rate": 4.061450805088122e-06, "loss": 0.6116, "step": 7995 }, { "epoch": 0.57, "grad_norm": 6.086618142066293, "learning_rate": 4.0603101571365555e-06, "loss": 0.8841, "step": 7996 }, { "epoch": 0.57, "grad_norm": 4.812883882377866, "learning_rate": 4.059169559877697e-06, "loss": 1.0487, "step": 7997 }, { "epoch": 0.57, "grad_norm": 14.216399395494543, "learning_rate": 4.058029013373079e-06, "loss": 0.7052, "step": 7998 }, { "epoch": 0.57, "grad_norm": 9.348456609493768, "learning_rate": 4.05688851768423e-06, "loss": 0.5167, "step": 7999 }, { "epoch": 0.57, "grad_norm": 8.037351807301572, "learning_rate": 4.055748072872677e-06, "loss": 0.555, "step": 8000 }, { "epoch": 0.57, "grad_norm": 12.11355327008387, "learning_rate": 4.05460767899994e-06, "loss": 0.9155, "step": 8001 }, { "epoch": 0.57, "grad_norm": 4.09126459476802, "learning_rate": 4.05346733612754e-06, "loss": 0.2264, "step": 8002 }, { "epoch": 0.57, "grad_norm": 10.195141378189948, "learning_rate": 4.052327044316996e-06, "loss": 0.6835, "step": 8003 }, { "epoch": 0.57, "grad_norm": 19.55617304914119, "learning_rate": 4.051186803629819e-06, "loss": 0.8691, "step": 8004 }, { "epoch": 0.57, "grad_norm": 4.067978159068093, "learning_rate": 4.050046614127524e-06, "loss": 0.2574, "step": 8005 }, { "epoch": 0.57, "grad_norm": 4.985627985529814, "learning_rate": 4.048906475871617e-06, "loss": 0.8045, "step": 8006 }, { "epoch": 0.57, "grad_norm": 16.1102114880716, "learning_rate": 4.047766388923606e-06, "loss": 0.9653, "step": 8007 }, { "epoch": 0.57, "grad_norm": 3.50350153290623, "learning_rate": 4.046626353344995e-06, "loss": 0.1792, "step": 8008 }, { "epoch": 0.57, "grad_norm": 3.2440441545041137, "learning_rate": 4.045486369197284e-06, "loss": 0.2404, "step": 8009 }, { "epoch": 0.57, "grad_norm": 2.4820532678056137, "learning_rate": 4.044346436541968e-06, "loss": 0.1086, "step": 8010 }, { "epoch": 0.57, "grad_norm": 13.350705467976159, "learning_rate": 4.043206555440547e-06, "loss": 0.7534, "step": 8011 }, { "epoch": 0.57, "grad_norm": 3.346086259521417, "learning_rate": 4.042066725954508e-06, "loss": 0.1846, "step": 8012 }, { "epoch": 0.57, "grad_norm": 9.959893921686426, "learning_rate": 4.040926948145344e-06, "loss": 0.4606, "step": 8013 }, { "epoch": 0.57, "grad_norm": 19.02498562449279, "learning_rate": 4.039787222074542e-06, "loss": 0.8782, "step": 8014 }, { "epoch": 0.57, "grad_norm": 50.31517189773321, "learning_rate": 4.038647547803583e-06, "loss": 0.6162, "step": 8015 }, { "epoch": 0.57, "grad_norm": 3.471061910304369, "learning_rate": 4.0375079253939505e-06, "loss": 0.1527, "step": 8016 }, { "epoch": 0.57, "grad_norm": 3.21407244601555, "learning_rate": 4.036368354907123e-06, "loss": 0.2306, "step": 8017 }, { "epoch": 0.57, "grad_norm": 2.7680777449066998, "learning_rate": 4.035228836404574e-06, "loss": 0.2329, "step": 8018 }, { "epoch": 0.58, "grad_norm": 2.7433888382540084, "learning_rate": 4.034089369947779e-06, "loss": 0.1456, "step": 8019 }, { "epoch": 0.58, "grad_norm": 18.02291347303725, "learning_rate": 4.032949955598207e-06, "loss": 0.7681, "step": 8020 }, { "epoch": 0.58, "grad_norm": 3.442514012298415, "learning_rate": 4.031810593417322e-06, "loss": 0.2258, "step": 8021 }, { "epoch": 0.58, "grad_norm": 4.020099765378402, "learning_rate": 4.030671283466592e-06, "loss": 0.8508, "step": 8022 }, { "epoch": 0.58, "grad_norm": 29.92088279598027, "learning_rate": 4.029532025807476e-06, "loss": 0.7854, "step": 8023 }, { "epoch": 0.58, "grad_norm": 7.364977838596616, "learning_rate": 4.028392820501435e-06, "loss": 0.4553, "step": 8024 }, { "epoch": 0.58, "grad_norm": 8.113585847541572, "learning_rate": 4.027253667609923e-06, "loss": 0.695, "step": 8025 }, { "epoch": 0.58, "grad_norm": 4.183078688818598, "learning_rate": 4.0261145671943935e-06, "loss": 0.9949, "step": 8026 }, { "epoch": 0.58, "grad_norm": 4.095085065889974, "learning_rate": 4.024975519316297e-06, "loss": 0.2049, "step": 8027 }, { "epoch": 0.58, "grad_norm": 10.994246333116195, "learning_rate": 4.023836524037081e-06, "loss": 0.7041, "step": 8028 }, { "epoch": 0.58, "grad_norm": 4.153585232195974, "learning_rate": 4.022697581418188e-06, "loss": 0.3224, "step": 8029 }, { "epoch": 0.58, "grad_norm": 9.833031101829254, "learning_rate": 4.0215586915210616e-06, "loss": 0.3133, "step": 8030 }, { "epoch": 0.58, "grad_norm": 3.8338653285940443, "learning_rate": 4.0204198544071415e-06, "loss": 0.7505, "step": 8031 }, { "epoch": 0.58, "grad_norm": 7.53855742036802, "learning_rate": 4.019281070137864e-06, "loss": 0.5427, "step": 8032 }, { "epoch": 0.58, "grad_norm": 13.730226115673373, "learning_rate": 4.018142338774657e-06, "loss": 1.0873, "step": 8033 }, { "epoch": 0.58, "grad_norm": 3.2057997702724284, "learning_rate": 4.017003660378957e-06, "loss": 0.1345, "step": 8034 }, { "epoch": 0.58, "grad_norm": 14.775360224746809, "learning_rate": 4.015865035012186e-06, "loss": 0.7912, "step": 8035 }, { "epoch": 0.58, "grad_norm": 3.6302345925967323, "learning_rate": 4.014726462735772e-06, "loss": 0.1308, "step": 8036 }, { "epoch": 0.58, "grad_norm": 3.8906044629351313, "learning_rate": 4.013587943611137e-06, "loss": 0.2205, "step": 8037 }, { "epoch": 0.58, "grad_norm": 3.427830308801061, "learning_rate": 4.012449477699697e-06, "loss": 0.2093, "step": 8038 }, { "epoch": 0.58, "grad_norm": 4.093735200983856, "learning_rate": 4.011311065062871e-06, "loss": 0.7648, "step": 8039 }, { "epoch": 0.58, "grad_norm": 8.150958727673988, "learning_rate": 4.01017270576207e-06, "loss": 0.3862, "step": 8040 }, { "epoch": 0.58, "grad_norm": 4.322517765802978, "learning_rate": 4.009034399858704e-06, "loss": 0.9674, "step": 8041 }, { "epoch": 0.58, "grad_norm": 24.69364015079323, "learning_rate": 4.007896147414183e-06, "loss": 0.7981, "step": 8042 }, { "epoch": 0.58, "grad_norm": 3.6033115046059003, "learning_rate": 4.0067579484899084e-06, "loss": 0.1458, "step": 8043 }, { "epoch": 0.58, "grad_norm": 3.4681506343073725, "learning_rate": 4.005619803147285e-06, "loss": 0.2015, "step": 8044 }, { "epoch": 0.58, "grad_norm": 3.5018584400149324, "learning_rate": 4.004481711447707e-06, "loss": 0.1907, "step": 8045 }, { "epoch": 0.58, "grad_norm": 3.7787023635948005, "learning_rate": 4.003343673452573e-06, "loss": 0.1846, "step": 8046 }, { "epoch": 0.58, "grad_norm": 3.16385061416171, "learning_rate": 4.002205689223274e-06, "loss": 0.1087, "step": 8047 }, { "epoch": 0.58, "grad_norm": 3.688832458768538, "learning_rate": 4.001067758821203e-06, "loss": 0.1311, "step": 8048 }, { "epoch": 0.58, "grad_norm": 10.905641763660695, "learning_rate": 3.999929882307744e-06, "loss": 0.6483, "step": 8049 }, { "epoch": 0.58, "grad_norm": 2.907094704026213, "learning_rate": 3.9987920597442834e-06, "loss": 0.2155, "step": 8050 }, { "epoch": 0.58, "grad_norm": 17.158529743690593, "learning_rate": 3.997654291192201e-06, "loss": 0.7655, "step": 8051 }, { "epoch": 0.58, "grad_norm": 4.429294626207626, "learning_rate": 3.9965165767128745e-06, "loss": 0.8586, "step": 8052 }, { "epoch": 0.58, "grad_norm": 3.1343996329133073, "learning_rate": 3.995378916367681e-06, "loss": 0.2075, "step": 8053 }, { "epoch": 0.58, "grad_norm": 4.719023164354062, "learning_rate": 3.994241310217993e-06, "loss": 0.9875, "step": 8054 }, { "epoch": 0.58, "grad_norm": 4.106991958469744, "learning_rate": 3.993103758325179e-06, "loss": 0.6851, "step": 8055 }, { "epoch": 0.58, "grad_norm": 25.121589313713567, "learning_rate": 3.991966260750605e-06, "loss": 0.873, "step": 8056 }, { "epoch": 0.58, "grad_norm": 3.2216069108531156, "learning_rate": 3.990828817555639e-06, "loss": 0.137, "step": 8057 }, { "epoch": 0.58, "grad_norm": 10.192282650952688, "learning_rate": 3.9896914288016344e-06, "loss": 0.6678, "step": 8058 }, { "epoch": 0.58, "grad_norm": 15.589521294194146, "learning_rate": 3.988554094549955e-06, "loss": 0.8128, "step": 8059 }, { "epoch": 0.58, "grad_norm": 3.8587620350042293, "learning_rate": 3.987416814861953e-06, "loss": 0.1999, "step": 8060 }, { "epoch": 0.58, "grad_norm": 5.1071926994091585, "learning_rate": 3.986279589798981e-06, "loss": 1.0111, "step": 8061 }, { "epoch": 0.58, "grad_norm": 21.70845565961581, "learning_rate": 3.985142419422387e-06, "loss": 0.9769, "step": 8062 }, { "epoch": 0.58, "grad_norm": 4.207903627339113, "learning_rate": 3.98400530379352e-06, "loss": 0.3032, "step": 8063 }, { "epoch": 0.58, "grad_norm": 5.037273935605159, "learning_rate": 3.982868242973719e-06, "loss": 0.2616, "step": 8064 }, { "epoch": 0.58, "grad_norm": 14.157548430089662, "learning_rate": 3.981731237024328e-06, "loss": 0.5909, "step": 8065 }, { "epoch": 0.58, "grad_norm": 3.2707922182482907, "learning_rate": 3.980594286006681e-06, "loss": 0.1708, "step": 8066 }, { "epoch": 0.58, "grad_norm": 2.645335155205544, "learning_rate": 3.979457389982114e-06, "loss": 0.1355, "step": 8067 }, { "epoch": 0.58, "grad_norm": 21.183190805184246, "learning_rate": 3.978320549011959e-06, "loss": 0.9243, "step": 8068 }, { "epoch": 0.58, "grad_norm": 14.827806316481652, "learning_rate": 3.977183763157544e-06, "loss": 0.8353, "step": 8069 }, { "epoch": 0.58, "grad_norm": 4.6281604976385005, "learning_rate": 3.976047032480191e-06, "loss": 0.8113, "step": 8070 }, { "epoch": 0.58, "grad_norm": 4.123958965200312, "learning_rate": 3.974910357041226e-06, "loss": 0.704, "step": 8071 }, { "epoch": 0.58, "grad_norm": 3.347440403358507, "learning_rate": 3.973773736901966e-06, "loss": 0.2319, "step": 8072 }, { "epoch": 0.58, "grad_norm": 14.17808477576156, "learning_rate": 3.972637172123728e-06, "loss": 0.5453, "step": 8073 }, { "epoch": 0.58, "grad_norm": 15.550218645159529, "learning_rate": 3.971500662767828e-06, "loss": 0.6283, "step": 8074 }, { "epoch": 0.58, "grad_norm": 16.68146200461594, "learning_rate": 3.970364208895572e-06, "loss": 0.7171, "step": 8075 }, { "epoch": 0.58, "grad_norm": 3.5543603640967962, "learning_rate": 3.969227810568271e-06, "loss": 0.2093, "step": 8076 }, { "epoch": 0.58, "grad_norm": 3.2820507986824436, "learning_rate": 3.968091467847227e-06, "loss": 0.2233, "step": 8077 }, { "epoch": 0.58, "grad_norm": 3.7712502887327513, "learning_rate": 3.966955180793743e-06, "loss": 0.227, "step": 8078 }, { "epoch": 0.58, "grad_norm": 18.74403778933608, "learning_rate": 3.965818949469116e-06, "loss": 1.5003, "step": 8079 }, { "epoch": 0.58, "grad_norm": 5.044005150274433, "learning_rate": 3.964682773934642e-06, "loss": 0.9786, "step": 8080 }, { "epoch": 0.58, "grad_norm": 14.502388455694597, "learning_rate": 3.963546654251616e-06, "loss": 0.7629, "step": 8081 }, { "epoch": 0.58, "grad_norm": 3.4873740801199955, "learning_rate": 3.962410590481322e-06, "loss": 0.2158, "step": 8082 }, { "epoch": 0.58, "grad_norm": 3.7495416535088903, "learning_rate": 3.961274582685051e-06, "loss": 0.2782, "step": 8083 }, { "epoch": 0.58, "grad_norm": 3.3304219938629616, "learning_rate": 3.960138630924083e-06, "loss": 0.1959, "step": 8084 }, { "epoch": 0.58, "grad_norm": 4.261072935848948, "learning_rate": 3.9590027352597e-06, "loss": 0.8229, "step": 8085 }, { "epoch": 0.58, "grad_norm": 6.274729096879156, "learning_rate": 3.957866895753179e-06, "loss": 0.2567, "step": 8086 }, { "epoch": 0.58, "grad_norm": 7.508504411042576, "learning_rate": 3.9567311124657935e-06, "loss": 0.7693, "step": 8087 }, { "epoch": 0.58, "grad_norm": 12.300630807447645, "learning_rate": 3.955595385458817e-06, "loss": 0.7564, "step": 8088 }, { "epoch": 0.58, "grad_norm": 3.767510329095467, "learning_rate": 3.954459714793515e-06, "loss": 0.256, "step": 8089 }, { "epoch": 0.58, "grad_norm": 17.04230857018448, "learning_rate": 3.953324100531155e-06, "loss": 0.5988, "step": 8090 }, { "epoch": 0.58, "grad_norm": 4.21438641561187, "learning_rate": 3.952188542732998e-06, "loss": 0.9863, "step": 8091 }, { "epoch": 0.58, "grad_norm": 9.979142017064563, "learning_rate": 3.951053041460301e-06, "loss": 0.5825, "step": 8092 }, { "epoch": 0.58, "grad_norm": 3.121327053167468, "learning_rate": 3.9499175967743244e-06, "loss": 0.1416, "step": 8093 }, { "epoch": 0.58, "grad_norm": 2.2008079049017284, "learning_rate": 3.948782208736319e-06, "loss": 0.137, "step": 8094 }, { "epoch": 0.58, "grad_norm": 2.922656296421236, "learning_rate": 3.947646877407532e-06, "loss": 0.2142, "step": 8095 }, { "epoch": 0.58, "grad_norm": 2.4046752822323394, "learning_rate": 3.946511602849213e-06, "loss": 0.1692, "step": 8096 }, { "epoch": 0.58, "grad_norm": 10.628886452580627, "learning_rate": 3.945376385122606e-06, "loss": 0.6977, "step": 8097 }, { "epoch": 0.58, "grad_norm": 4.134579251760826, "learning_rate": 3.944241224288951e-06, "loss": 0.2654, "step": 8098 }, { "epoch": 0.58, "grad_norm": 4.113184356764322, "learning_rate": 3.943106120409486e-06, "loss": 0.2425, "step": 8099 }, { "epoch": 0.58, "grad_norm": 3.676196523933489, "learning_rate": 3.9419710735454444e-06, "loss": 0.157, "step": 8100 }, { "epoch": 0.58, "grad_norm": 3.937431806739016, "learning_rate": 3.940836083758058e-06, "loss": 0.2428, "step": 8101 }, { "epoch": 0.58, "grad_norm": 4.7516287237271335, "learning_rate": 3.939701151108557e-06, "loss": 0.8128, "step": 8102 }, { "epoch": 0.58, "grad_norm": 15.365756424122322, "learning_rate": 3.938566275658165e-06, "loss": 0.4348, "step": 8103 }, { "epoch": 0.58, "grad_norm": 2.576154622518415, "learning_rate": 3.937431457468105e-06, "loss": 0.1828, "step": 8104 }, { "epoch": 0.58, "grad_norm": 4.5211995572128405, "learning_rate": 3.936296696599597e-06, "loss": 0.948, "step": 8105 }, { "epoch": 0.58, "grad_norm": 11.944712832464145, "learning_rate": 3.9351619931138545e-06, "loss": 0.9208, "step": 8106 }, { "epoch": 0.58, "grad_norm": 3.627005593850142, "learning_rate": 3.934027347072091e-06, "loss": 0.309, "step": 8107 }, { "epoch": 0.58, "grad_norm": 3.6257327210889425, "learning_rate": 3.932892758535519e-06, "loss": 0.1875, "step": 8108 }, { "epoch": 0.58, "grad_norm": 3.483830854990033, "learning_rate": 3.931758227565342e-06, "loss": 0.1882, "step": 8109 }, { "epoch": 0.58, "grad_norm": 4.211347765681148, "learning_rate": 3.930623754222765e-06, "loss": 0.2678, "step": 8110 }, { "epoch": 0.58, "grad_norm": 18.7217661842868, "learning_rate": 3.92948933856899e-06, "loss": 0.7299, "step": 8111 }, { "epoch": 0.58, "grad_norm": 2.9994741128808453, "learning_rate": 3.9283549806652115e-06, "loss": 0.1408, "step": 8112 }, { "epoch": 0.58, "grad_norm": 4.710515165809931, "learning_rate": 3.927220680572626e-06, "loss": 0.6604, "step": 8113 }, { "epoch": 0.58, "grad_norm": 3.123605548334863, "learning_rate": 3.926086438352425e-06, "loss": 0.1153, "step": 8114 }, { "epoch": 0.58, "grad_norm": 3.918090823944724, "learning_rate": 3.9249522540657945e-06, "loss": 0.2038, "step": 8115 }, { "epoch": 0.58, "grad_norm": 15.552186341563871, "learning_rate": 3.923818127773921e-06, "loss": 1.3588, "step": 8116 }, { "epoch": 0.58, "grad_norm": 5.201159308919701, "learning_rate": 3.9226840595379875e-06, "loss": 0.2123, "step": 8117 }, { "epoch": 0.58, "grad_norm": 4.766764223998268, "learning_rate": 3.92155004941917e-06, "loss": 0.9984, "step": 8118 }, { "epoch": 0.58, "grad_norm": 18.609708474138223, "learning_rate": 3.920416097478644e-06, "loss": 1.1255, "step": 8119 }, { "epoch": 0.58, "grad_norm": 4.135004313812845, "learning_rate": 3.9192822037775845e-06, "loss": 0.7866, "step": 8120 }, { "epoch": 0.58, "grad_norm": 13.030277065405949, "learning_rate": 3.918148368377158e-06, "loss": 1.2229, "step": 8121 }, { "epoch": 0.58, "grad_norm": 3.414065862609482, "learning_rate": 3.917014591338534e-06, "loss": 0.1573, "step": 8122 }, { "epoch": 0.58, "grad_norm": 3.3797247714097836, "learning_rate": 3.915880872722872e-06, "loss": 0.1987, "step": 8123 }, { "epoch": 0.58, "grad_norm": 3.5593830128665407, "learning_rate": 3.914747212591332e-06, "loss": 0.1811, "step": 8124 }, { "epoch": 0.58, "grad_norm": 13.877334685393436, "learning_rate": 3.913613611005075e-06, "loss": 0.6709, "step": 8125 }, { "epoch": 0.58, "grad_norm": 2.992425690113595, "learning_rate": 3.9124800680252496e-06, "loss": 0.1136, "step": 8126 }, { "epoch": 0.58, "grad_norm": 3.8117112374221795, "learning_rate": 3.911346583713009e-06, "loss": 0.2067, "step": 8127 }, { "epoch": 0.58, "grad_norm": 4.1511499743138325, "learning_rate": 3.910213158129498e-06, "loss": 0.1824, "step": 8128 }, { "epoch": 0.58, "grad_norm": 16.936126814986974, "learning_rate": 3.909079791335864e-06, "loss": 0.4626, "step": 8129 }, { "epoch": 0.58, "grad_norm": 4.043720245996776, "learning_rate": 3.907946483393245e-06, "loss": 0.7972, "step": 8130 }, { "epoch": 0.58, "grad_norm": 4.545091092668529, "learning_rate": 3.90681323436278e-06, "loss": 0.8592, "step": 8131 }, { "epoch": 0.58, "grad_norm": 3.653304062827689, "learning_rate": 3.905680044305602e-06, "loss": 0.1556, "step": 8132 }, { "epoch": 0.58, "grad_norm": 4.399101742765978, "learning_rate": 3.904546913282844e-06, "loss": 0.8209, "step": 8133 }, { "epoch": 0.58, "grad_norm": 4.1942603185440355, "learning_rate": 3.903413841355633e-06, "loss": 0.1906, "step": 8134 }, { "epoch": 0.58, "grad_norm": 2.8269088065629195, "learning_rate": 3.902280828585095e-06, "loss": 0.2233, "step": 8135 }, { "epoch": 0.58, "grad_norm": 18.798697276644813, "learning_rate": 3.9011478750323515e-06, "loss": 1.0312, "step": 8136 }, { "epoch": 0.58, "grad_norm": 20.91950073112408, "learning_rate": 3.900014980758519e-06, "loss": 1.0182, "step": 8137 }, { "epoch": 0.58, "grad_norm": 3.025284747743773, "learning_rate": 3.898882145824716e-06, "loss": 0.1496, "step": 8138 }, { "epoch": 0.58, "grad_norm": 10.04147440441566, "learning_rate": 3.897749370292054e-06, "loss": 0.4726, "step": 8139 }, { "epoch": 0.58, "grad_norm": 5.650291558532678, "learning_rate": 3.89661665422164e-06, "loss": 0.3824, "step": 8140 }, { "epoch": 0.58, "grad_norm": 2.478054680160912, "learning_rate": 3.895483997674583e-06, "loss": 0.1268, "step": 8141 }, { "epoch": 0.58, "grad_norm": 2.7989432214931678, "learning_rate": 3.894351400711982e-06, "loss": 0.1553, "step": 8142 }, { "epoch": 0.58, "grad_norm": 2.8794802563679522, "learning_rate": 3.893218863394937e-06, "loss": 0.1797, "step": 8143 }, { "epoch": 0.58, "grad_norm": 16.478752447598062, "learning_rate": 3.892086385784546e-06, "loss": 0.495, "step": 8144 }, { "epoch": 0.58, "grad_norm": 2.8813305562395293, "learning_rate": 3.8909539679419016e-06, "loss": 0.135, "step": 8145 }, { "epoch": 0.58, "grad_norm": 2.4587960694994377, "learning_rate": 3.88982160992809e-06, "loss": 0.1064, "step": 8146 }, { "epoch": 0.58, "grad_norm": 15.08737695463999, "learning_rate": 3.8886893118042015e-06, "loss": 0.8576, "step": 8147 }, { "epoch": 0.58, "grad_norm": 22.916464996312104, "learning_rate": 3.887557073631319e-06, "loss": 0.6055, "step": 8148 }, { "epoch": 0.58, "grad_norm": 6.344065463832404, "learning_rate": 3.886424895470521e-06, "loss": 0.573, "step": 8149 }, { "epoch": 0.58, "grad_norm": 11.195358937146104, "learning_rate": 3.885292777382885e-06, "loss": 0.6453, "step": 8150 }, { "epoch": 0.58, "grad_norm": 3.590016163321932, "learning_rate": 3.884160719429484e-06, "loss": 0.7495, "step": 8151 }, { "epoch": 0.58, "grad_norm": 3.4380421423797394, "learning_rate": 3.883028721671387e-06, "loss": 0.1975, "step": 8152 }, { "epoch": 0.58, "grad_norm": 2.919869315733224, "learning_rate": 3.881896784169666e-06, "loss": 0.1947, "step": 8153 }, { "epoch": 0.58, "grad_norm": 3.1012987240813312, "learning_rate": 3.88076490698538e-06, "loss": 0.215, "step": 8154 }, { "epoch": 0.58, "grad_norm": 3.501892088678202, "learning_rate": 3.87963309017959e-06, "loss": 0.1492, "step": 8155 }, { "epoch": 0.58, "grad_norm": 3.17285004158812, "learning_rate": 3.878501333813354e-06, "loss": 0.2151, "step": 8156 }, { "epoch": 0.58, "grad_norm": 19.500129813178965, "learning_rate": 3.877369637947725e-06, "loss": 0.4619, "step": 8157 }, { "epoch": 0.58, "grad_norm": 3.261887373898156, "learning_rate": 3.876238002643756e-06, "loss": 0.1545, "step": 8158 }, { "epoch": 0.59, "grad_norm": 4.151788893911445, "learning_rate": 3.875106427962493e-06, "loss": 0.839, "step": 8159 }, { "epoch": 0.59, "grad_norm": 4.32327003601512, "learning_rate": 3.873974913964979e-06, "loss": 0.9108, "step": 8160 }, { "epoch": 0.59, "grad_norm": 2.7327490635631295, "learning_rate": 3.8728434607122566e-06, "loss": 0.1339, "step": 8161 }, { "epoch": 0.59, "grad_norm": 15.071254398896565, "learning_rate": 3.871712068265364e-06, "loss": 0.7705, "step": 8162 }, { "epoch": 0.59, "grad_norm": 2.705139815275869, "learning_rate": 3.870580736685334e-06, "loss": 0.1535, "step": 8163 }, { "epoch": 0.59, "grad_norm": 3.195051342871945, "learning_rate": 3.869449466033198e-06, "loss": 0.1637, "step": 8164 }, { "epoch": 0.59, "grad_norm": 3.203112134454353, "learning_rate": 3.8683182563699866e-06, "loss": 0.1565, "step": 8165 }, { "epoch": 0.59, "grad_norm": 16.25449880514245, "learning_rate": 3.867187107756718e-06, "loss": 0.6058, "step": 8166 }, { "epoch": 0.59, "grad_norm": 3.6948679446848045, "learning_rate": 3.866056020254418e-06, "loss": 0.1647, "step": 8167 }, { "epoch": 0.59, "grad_norm": 2.9091724147318665, "learning_rate": 3.864924993924104e-06, "loss": 0.1371, "step": 8168 }, { "epoch": 0.59, "grad_norm": 28.21104831983351, "learning_rate": 3.863794028826789e-06, "loss": 1.5906, "step": 8169 }, { "epoch": 0.59, "grad_norm": 3.6668049792355175, "learning_rate": 3.862663125023487e-06, "loss": 0.1659, "step": 8170 }, { "epoch": 0.59, "grad_norm": 4.619281628669042, "learning_rate": 3.8615322825752025e-06, "loss": 0.2619, "step": 8171 }, { "epoch": 0.59, "grad_norm": 20.365077464463752, "learning_rate": 3.860401501542943e-06, "loss": 0.5942, "step": 8172 }, { "epoch": 0.59, "grad_norm": 13.373051902727527, "learning_rate": 3.8592707819877084e-06, "loss": 0.6776, "step": 8173 }, { "epoch": 0.59, "grad_norm": 15.40488100584979, "learning_rate": 3.858140123970498e-06, "loss": 1.0854, "step": 8174 }, { "epoch": 0.59, "grad_norm": 4.1023211819328305, "learning_rate": 3.8570095275523045e-06, "loss": 0.3122, "step": 8175 }, { "epoch": 0.59, "grad_norm": 3.933836931057112, "learning_rate": 3.855878992794122e-06, "loss": 0.1548, "step": 8176 }, { "epoch": 0.59, "grad_norm": 3.809706977675242, "learning_rate": 3.8547485197569375e-06, "loss": 0.2201, "step": 8177 }, { "epoch": 0.59, "grad_norm": 3.590227710234069, "learning_rate": 3.853618108501735e-06, "loss": 0.1758, "step": 8178 }, { "epoch": 0.59, "grad_norm": 3.567714593988948, "learning_rate": 3.852487759089495e-06, "loss": 0.1922, "step": 8179 }, { "epoch": 0.59, "grad_norm": 3.9787775042475793, "learning_rate": 3.851357471581197e-06, "loss": 0.237, "step": 8180 }, { "epoch": 0.59, "grad_norm": 3.724737037295292, "learning_rate": 3.850227246037816e-06, "loss": 0.1873, "step": 8181 }, { "epoch": 0.59, "grad_norm": 3.2273271462518247, "learning_rate": 3.849097082520323e-06, "loss": 0.1346, "step": 8182 }, { "epoch": 0.59, "grad_norm": 3.1803522280482257, "learning_rate": 3.847966981089686e-06, "loss": 0.0988, "step": 8183 }, { "epoch": 0.59, "grad_norm": 18.17351843049882, "learning_rate": 3.846836941806869e-06, "loss": 0.8331, "step": 8184 }, { "epoch": 0.59, "grad_norm": 3.1751205235949516, "learning_rate": 3.845706964732835e-06, "loss": 0.1842, "step": 8185 }, { "epoch": 0.59, "grad_norm": 10.996300295397367, "learning_rate": 3.84457704992854e-06, "loss": 0.6069, "step": 8186 }, { "epoch": 0.59, "grad_norm": 4.5679137133083, "learning_rate": 3.843447197454941e-06, "loss": 0.9914, "step": 8187 }, { "epoch": 0.59, "grad_norm": 3.282614397853298, "learning_rate": 3.8423174073729865e-06, "loss": 0.1382, "step": 8188 }, { "epoch": 0.59, "grad_norm": 33.63627255559627, "learning_rate": 3.8411876797436285e-06, "loss": 1.0215, "step": 8189 }, { "epoch": 0.59, "grad_norm": 3.129320108983332, "learning_rate": 3.840058014627806e-06, "loss": 0.1033, "step": 8190 }, { "epoch": 0.59, "grad_norm": 3.505994582541549, "learning_rate": 3.838928412086464e-06, "loss": 0.1577, "step": 8191 }, { "epoch": 0.59, "grad_norm": 4.862469710372817, "learning_rate": 3.837798872180539e-06, "loss": 0.9222, "step": 8192 }, { "epoch": 0.59, "grad_norm": 21.553560057304157, "learning_rate": 3.836669394970966e-06, "loss": 1.0659, "step": 8193 }, { "epoch": 0.59, "grad_norm": 12.236020136415801, "learning_rate": 3.835539980518675e-06, "loss": 0.8583, "step": 8194 }, { "epoch": 0.59, "grad_norm": 4.827314706174382, "learning_rate": 3.834410628884595e-06, "loss": 0.9131, "step": 8195 }, { "epoch": 0.59, "grad_norm": 3.7300245341790994, "learning_rate": 3.83328134012965e-06, "loss": 0.2077, "step": 8196 }, { "epoch": 0.59, "grad_norm": 18.571823392983372, "learning_rate": 3.832152114314759e-06, "loss": 0.799, "step": 8197 }, { "epoch": 0.59, "grad_norm": 23.38994038706052, "learning_rate": 3.831022951500842e-06, "loss": 0.5256, "step": 8198 }, { "epoch": 0.59, "grad_norm": 11.84541911838744, "learning_rate": 3.829893851748811e-06, "loss": 0.5693, "step": 8199 }, { "epoch": 0.59, "grad_norm": 3.41620294308459, "learning_rate": 3.828764815119579e-06, "loss": 0.1529, "step": 8200 }, { "epoch": 0.59, "grad_norm": 2.481157325171247, "learning_rate": 3.827635841674052e-06, "loss": 0.129, "step": 8201 }, { "epoch": 0.59, "grad_norm": 4.529959323646252, "learning_rate": 3.8265069314731335e-06, "loss": 0.8721, "step": 8202 }, { "epoch": 0.59, "grad_norm": 3.565674161996713, "learning_rate": 3.825378084577723e-06, "loss": 0.1729, "step": 8203 }, { "epoch": 0.59, "grad_norm": 3.1770402638337476, "learning_rate": 3.8242493010487185e-06, "loss": 0.1623, "step": 8204 }, { "epoch": 0.59, "grad_norm": 11.02424967181584, "learning_rate": 3.8231205809470154e-06, "loss": 0.6957, "step": 8205 }, { "epoch": 0.59, "grad_norm": 3.687844419018487, "learning_rate": 3.8219919243335005e-06, "loss": 0.279, "step": 8206 }, { "epoch": 0.59, "grad_norm": 4.218365886432751, "learning_rate": 3.820863331269064e-06, "loss": 0.7829, "step": 8207 }, { "epoch": 0.59, "grad_norm": 3.829824998407282, "learning_rate": 3.819734801814585e-06, "loss": 0.1719, "step": 8208 }, { "epoch": 0.59, "grad_norm": 2.4981302536291503, "learning_rate": 3.8186063360309475e-06, "loss": 0.0963, "step": 8209 }, { "epoch": 0.59, "grad_norm": 3.339400627043039, "learning_rate": 3.817477933979027e-06, "loss": 0.2238, "step": 8210 }, { "epoch": 0.59, "grad_norm": 39.874807313802016, "learning_rate": 3.816349595719694e-06, "loss": 0.7281, "step": 8211 }, { "epoch": 0.59, "grad_norm": 2.969062854162501, "learning_rate": 3.8152213213138215e-06, "loss": 0.1253, "step": 8212 }, { "epoch": 0.59, "grad_norm": 3.3030656628910995, "learning_rate": 3.8140931108222744e-06, "loss": 0.2103, "step": 8213 }, { "epoch": 0.59, "grad_norm": 3.148745625853992, "learning_rate": 3.8129649643059146e-06, "loss": 0.1821, "step": 8214 }, { "epoch": 0.59, "grad_norm": 2.822815209896524, "learning_rate": 3.8118368818256012e-06, "loss": 0.1449, "step": 8215 }, { "epoch": 0.59, "grad_norm": 2.5192144197589994, "learning_rate": 3.8107088634421913e-06, "loss": 0.1266, "step": 8216 }, { "epoch": 0.59, "grad_norm": 5.092079154371582, "learning_rate": 3.809580909216536e-06, "loss": 0.7748, "step": 8217 }, { "epoch": 0.59, "grad_norm": 4.247885019701444, "learning_rate": 3.808453019209484e-06, "loss": 0.8642, "step": 8218 }, { "epoch": 0.59, "grad_norm": 4.015287679896396, "learning_rate": 3.8073251934818827e-06, "loss": 0.2781, "step": 8219 }, { "epoch": 0.59, "grad_norm": 8.872920242044811, "learning_rate": 3.8061974320945717e-06, "loss": 0.6712, "step": 8220 }, { "epoch": 0.59, "grad_norm": 11.012310080319574, "learning_rate": 3.8050697351083916e-06, "loss": 0.6813, "step": 8221 }, { "epoch": 0.59, "grad_norm": 15.23199293668817, "learning_rate": 3.803942102584175e-06, "loss": 0.7482, "step": 8222 }, { "epoch": 0.59, "grad_norm": 18.709646031805587, "learning_rate": 3.8028145345827543e-06, "loss": 0.8341, "step": 8223 }, { "epoch": 0.59, "grad_norm": 4.71387493812111, "learning_rate": 3.8016870311649597e-06, "loss": 0.8075, "step": 8224 }, { "epoch": 0.59, "grad_norm": 8.900702507102839, "learning_rate": 3.800559592391614e-06, "loss": 0.6841, "step": 8225 }, { "epoch": 0.59, "grad_norm": 4.501561497704077, "learning_rate": 3.7994322183235367e-06, "loss": 0.8354, "step": 8226 }, { "epoch": 0.59, "grad_norm": 4.3338034718879666, "learning_rate": 3.798304909021547e-06, "loss": 0.6981, "step": 8227 }, { "epoch": 0.59, "grad_norm": 10.047678814926645, "learning_rate": 3.797177664546459e-06, "loss": 0.671, "step": 8228 }, { "epoch": 0.59, "grad_norm": 4.275707389235973, "learning_rate": 3.796050484959083e-06, "loss": 0.7829, "step": 8229 }, { "epoch": 0.59, "grad_norm": 2.82813713453407, "learning_rate": 3.7949233703202265e-06, "loss": 0.1393, "step": 8230 }, { "epoch": 0.59, "grad_norm": 22.31733099787106, "learning_rate": 3.793796320690692e-06, "loss": 0.9387, "step": 8231 }, { "epoch": 0.59, "grad_norm": 22.587096503575047, "learning_rate": 3.7926693361312806e-06, "loss": 1.0032, "step": 8232 }, { "epoch": 0.59, "grad_norm": 3.2054860742420193, "learning_rate": 3.7915424167027885e-06, "loss": 0.1896, "step": 8233 }, { "epoch": 0.59, "grad_norm": 13.168323225010697, "learning_rate": 3.790415562466009e-06, "loss": 0.6419, "step": 8234 }, { "epoch": 0.59, "grad_norm": 4.524055129426256, "learning_rate": 3.7892887734817315e-06, "loss": 0.8399, "step": 8235 }, { "epoch": 0.59, "grad_norm": 4.486626166854766, "learning_rate": 3.788162049810741e-06, "loss": 0.9021, "step": 8236 }, { "epoch": 0.59, "grad_norm": 2.409580850201946, "learning_rate": 3.787035391513823e-06, "loss": 0.1316, "step": 8237 }, { "epoch": 0.59, "grad_norm": 2.913353469728561, "learning_rate": 3.7859087986517524e-06, "loss": 0.1427, "step": 8238 }, { "epoch": 0.59, "grad_norm": 4.892078834165305, "learning_rate": 3.7847822712853077e-06, "loss": 0.2581, "step": 8239 }, { "epoch": 0.59, "grad_norm": 50.13796233624127, "learning_rate": 3.7836558094752583e-06, "loss": 1.1764, "step": 8240 }, { "epoch": 0.59, "grad_norm": 4.4790001241805575, "learning_rate": 3.7825294132823738e-06, "loss": 0.7806, "step": 8241 }, { "epoch": 0.59, "grad_norm": 4.39388879349024, "learning_rate": 3.78140308276742e-06, "loss": 0.8399, "step": 8242 }, { "epoch": 0.59, "grad_norm": 5.575712045418694, "learning_rate": 3.7802768179911566e-06, "loss": 0.7568, "step": 8243 }, { "epoch": 0.59, "grad_norm": 13.369876462882397, "learning_rate": 3.7791506190143427e-06, "loss": 0.5267, "step": 8244 }, { "epoch": 0.59, "grad_norm": 3.627676479482943, "learning_rate": 3.778024485897731e-06, "loss": 0.2006, "step": 8245 }, { "epoch": 0.59, "grad_norm": 9.80996371226092, "learning_rate": 3.7768984187020728e-06, "loss": 0.5129, "step": 8246 }, { "epoch": 0.59, "grad_norm": 14.183746919400932, "learning_rate": 3.775772417488116e-06, "loss": 0.8069, "step": 8247 }, { "epoch": 0.59, "grad_norm": 4.0408361268788795, "learning_rate": 3.7746464823166024e-06, "loss": 0.3101, "step": 8248 }, { "epoch": 0.59, "grad_norm": 4.663802412631463, "learning_rate": 3.7735206132482748e-06, "loss": 0.8213, "step": 8249 }, { "epoch": 0.59, "grad_norm": 3.782870137117762, "learning_rate": 3.7723948103438675e-06, "loss": 0.1532, "step": 8250 }, { "epoch": 0.59, "grad_norm": 3.573295196553494, "learning_rate": 3.7712690736641113e-06, "loss": 0.1934, "step": 8251 }, { "epoch": 0.59, "grad_norm": 3.363355215803308, "learning_rate": 3.770143403269738e-06, "loss": 0.1478, "step": 8252 }, { "epoch": 0.59, "grad_norm": 4.4043017427706745, "learning_rate": 3.769017799221474e-06, "loss": 0.7694, "step": 8253 }, { "epoch": 0.59, "grad_norm": 3.379535540618407, "learning_rate": 3.767892261580039e-06, "loss": 0.1627, "step": 8254 }, { "epoch": 0.59, "grad_norm": 3.333269111388738, "learning_rate": 3.7667667904061524e-06, "loss": 0.1252, "step": 8255 }, { "epoch": 0.59, "grad_norm": 21.06958386264654, "learning_rate": 3.765641385760531e-06, "loss": 1.1472, "step": 8256 }, { "epoch": 0.59, "grad_norm": 8.05338461671239, "learning_rate": 3.7645160477038835e-06, "loss": 0.484, "step": 8257 }, { "epoch": 0.59, "grad_norm": 4.04907482154923, "learning_rate": 3.763390776296919e-06, "loss": 0.7053, "step": 8258 }, { "epoch": 0.59, "grad_norm": 2.8936404030290643, "learning_rate": 3.762265571600341e-06, "loss": 0.1796, "step": 8259 }, { "epoch": 0.59, "grad_norm": 34.01962309670117, "learning_rate": 3.7611404336748497e-06, "loss": 0.5386, "step": 8260 }, { "epoch": 0.59, "grad_norm": 2.902887287375405, "learning_rate": 3.760015362581145e-06, "loss": 0.1975, "step": 8261 }, { "epoch": 0.59, "grad_norm": 23.7401854778171, "learning_rate": 3.7588903583799172e-06, "loss": 1.2222, "step": 8262 }, { "epoch": 0.59, "grad_norm": 3.4307488636978145, "learning_rate": 3.7577654211318556e-06, "loss": 0.0909, "step": 8263 }, { "epoch": 0.59, "grad_norm": 4.2280953792947145, "learning_rate": 3.7566405508976483e-06, "loss": 0.1987, "step": 8264 }, { "epoch": 0.59, "grad_norm": 3.7906943223241085, "learning_rate": 3.7555157477379757e-06, "loss": 0.1827, "step": 8265 }, { "epoch": 0.59, "grad_norm": 3.807534705524911, "learning_rate": 3.754391011713518e-06, "loss": 0.1241, "step": 8266 }, { "epoch": 0.59, "grad_norm": 11.418665741104169, "learning_rate": 3.7532663428849514e-06, "loss": 0.5827, "step": 8267 }, { "epoch": 0.59, "grad_norm": 3.008583406107897, "learning_rate": 3.752141741312946e-06, "loss": 0.1574, "step": 8268 }, { "epoch": 0.59, "grad_norm": 15.575323154842584, "learning_rate": 3.751017207058169e-06, "loss": 1.3394, "step": 8269 }, { "epoch": 0.59, "grad_norm": 18.404560932628872, "learning_rate": 3.7498927401812877e-06, "loss": 0.976, "step": 8270 }, { "epoch": 0.59, "grad_norm": 2.8290320861563627, "learning_rate": 3.7487683407429605e-06, "loss": 0.1334, "step": 8271 }, { "epoch": 0.59, "grad_norm": 2.721517879356468, "learning_rate": 3.7476440088038456e-06, "loss": 0.1165, "step": 8272 }, { "epoch": 0.59, "grad_norm": 15.895558193960948, "learning_rate": 3.7465197444245953e-06, "loss": 0.6779, "step": 8273 }, { "epoch": 0.59, "grad_norm": 4.371680134870567, "learning_rate": 3.745395547665862e-06, "loss": 0.9598, "step": 8274 }, { "epoch": 0.59, "grad_norm": 11.88522564953361, "learning_rate": 3.744271418588288e-06, "loss": 0.9377, "step": 8275 }, { "epoch": 0.59, "grad_norm": 2.3215484717720853, "learning_rate": 3.743147357252519e-06, "loss": 0.0902, "step": 8276 }, { "epoch": 0.59, "grad_norm": 6.092620836093496, "learning_rate": 3.7420233637191916e-06, "loss": 0.1311, "step": 8277 }, { "epoch": 0.59, "grad_norm": 2.842654960759407, "learning_rate": 3.7408994380489415e-06, "loss": 0.1628, "step": 8278 }, { "epoch": 0.59, "grad_norm": 5.174312419599919, "learning_rate": 3.7397755803024023e-06, "loss": 0.7584, "step": 8279 }, { "epoch": 0.59, "grad_norm": 10.613377562278018, "learning_rate": 3.7386517905401988e-06, "loss": 0.4654, "step": 8280 }, { "epoch": 0.59, "grad_norm": 4.092237992111348, "learning_rate": 3.7375280688229585e-06, "loss": 0.7097, "step": 8281 }, { "epoch": 0.59, "grad_norm": 4.391228108133745, "learning_rate": 3.7364044152112987e-06, "loss": 0.879, "step": 8282 }, { "epoch": 0.59, "grad_norm": 2.63066760515764, "learning_rate": 3.735280829765837e-06, "loss": 0.1668, "step": 8283 }, { "epoch": 0.59, "grad_norm": 4.697575627870537, "learning_rate": 3.7341573125471895e-06, "loss": 0.2302, "step": 8284 }, { "epoch": 0.59, "grad_norm": 16.513316493603803, "learning_rate": 3.7330338636159626e-06, "loss": 1.0078, "step": 8285 }, { "epoch": 0.59, "grad_norm": 3.2074846889069106, "learning_rate": 3.7319104830327647e-06, "loss": 0.1365, "step": 8286 }, { "epoch": 0.59, "grad_norm": 4.543273518094403, "learning_rate": 3.7307871708581945e-06, "loss": 0.8576, "step": 8287 }, { "epoch": 0.59, "grad_norm": 15.944794458376188, "learning_rate": 3.7296639271528523e-06, "loss": 1.0329, "step": 8288 }, { "epoch": 0.59, "grad_norm": 12.164532658440777, "learning_rate": 3.728540751977332e-06, "loss": 0.6875, "step": 8289 }, { "epoch": 0.59, "grad_norm": 2.3642592872567567, "learning_rate": 3.7274176453922263e-06, "loss": 0.0995, "step": 8290 }, { "epoch": 0.59, "grad_norm": 5.773124912872595, "learning_rate": 3.726294607458121e-06, "loss": 0.8219, "step": 8291 }, { "epoch": 0.59, "grad_norm": 14.101181925761066, "learning_rate": 3.7251716382356007e-06, "loss": 0.6586, "step": 8292 }, { "epoch": 0.59, "grad_norm": 18.059186979093255, "learning_rate": 3.724048737785245e-06, "loss": 0.7246, "step": 8293 }, { "epoch": 0.59, "grad_norm": 13.925306764094806, "learning_rate": 3.7229259061676305e-06, "loss": 1.1395, "step": 8294 }, { "epoch": 0.59, "grad_norm": 3.787054258158034, "learning_rate": 3.72180314344333e-06, "loss": 0.1672, "step": 8295 }, { "epoch": 0.59, "grad_norm": 3.482827529321633, "learning_rate": 3.72068044967291e-06, "loss": 0.1584, "step": 8296 }, { "epoch": 0.59, "grad_norm": 12.000741162309472, "learning_rate": 3.719557824916937e-06, "loss": 0.6913, "step": 8297 }, { "epoch": 0.6, "grad_norm": 4.343980128174196, "learning_rate": 3.718435269235976e-06, "loss": 0.1736, "step": 8298 }, { "epoch": 0.6, "grad_norm": 10.431177268157684, "learning_rate": 3.7173127826905796e-06, "loss": 0.375, "step": 8299 }, { "epoch": 0.6, "grad_norm": 11.710126925339981, "learning_rate": 3.7161903653413024e-06, "loss": 0.8399, "step": 8300 }, { "epoch": 0.6, "grad_norm": 13.98992629901049, "learning_rate": 3.7150680172486962e-06, "loss": 0.997, "step": 8301 }, { "epoch": 0.6, "grad_norm": 2.781405663581147, "learning_rate": 3.7139457384733057e-06, "loss": 0.0933, "step": 8302 }, { "epoch": 0.6, "grad_norm": 12.473638815107401, "learning_rate": 3.7128235290756747e-06, "loss": 0.8334, "step": 8303 }, { "epoch": 0.6, "grad_norm": 4.5213338364420235, "learning_rate": 3.711701389116343e-06, "loss": 0.8583, "step": 8304 }, { "epoch": 0.6, "grad_norm": 3.4852458445574572, "learning_rate": 3.7105793186558447e-06, "loss": 0.2099, "step": 8305 }, { "epoch": 0.6, "grad_norm": 20.42921555199547, "learning_rate": 3.7094573177547105e-06, "loss": 1.1099, "step": 8306 }, { "epoch": 0.6, "grad_norm": 3.6223680389247637, "learning_rate": 3.7083353864734706e-06, "loss": 0.1244, "step": 8307 }, { "epoch": 0.6, "grad_norm": 3.7261988848614656, "learning_rate": 3.707213524872646e-06, "loss": 0.1395, "step": 8308 }, { "epoch": 0.6, "grad_norm": 3.718319115108073, "learning_rate": 3.7060917330127592e-06, "loss": 0.1393, "step": 8309 }, { "epoch": 0.6, "grad_norm": 25.295575695040593, "learning_rate": 3.7049700109543274e-06, "loss": 1.2303, "step": 8310 }, { "epoch": 0.6, "grad_norm": 12.211161422070147, "learning_rate": 3.703848358757859e-06, "loss": 0.62, "step": 8311 }, { "epoch": 0.6, "grad_norm": 2.9896406394640374, "learning_rate": 3.7027267764838653e-06, "loss": 0.1562, "step": 8312 }, { "epoch": 0.6, "grad_norm": 18.78178781486696, "learning_rate": 3.7016052641928528e-06, "loss": 0.8058, "step": 8313 }, { "epoch": 0.6, "grad_norm": 3.3005749803751994, "learning_rate": 3.700483821945321e-06, "loss": 0.1373, "step": 8314 }, { "epoch": 0.6, "grad_norm": 3.1952498205014708, "learning_rate": 3.699362449801768e-06, "loss": 0.1433, "step": 8315 }, { "epoch": 0.6, "grad_norm": 8.716238286099696, "learning_rate": 3.6982411478226874e-06, "loss": 0.2697, "step": 8316 }, { "epoch": 0.6, "grad_norm": 2.8389745235542376, "learning_rate": 3.6971199160685694e-06, "loss": 0.1744, "step": 8317 }, { "epoch": 0.6, "grad_norm": 19.052516705822864, "learning_rate": 3.695998754599901e-06, "loss": 0.6914, "step": 8318 }, { "epoch": 0.6, "grad_norm": 3.1269337507243073, "learning_rate": 3.6948776634771633e-06, "loss": 0.1559, "step": 8319 }, { "epoch": 0.6, "grad_norm": 3.1194779523771206, "learning_rate": 3.6937566427608352e-06, "loss": 0.1574, "step": 8320 }, { "epoch": 0.6, "grad_norm": 15.048416687404895, "learning_rate": 3.6926356925113926e-06, "loss": 0.6321, "step": 8321 }, { "epoch": 0.6, "grad_norm": 2.3592167055821682, "learning_rate": 3.6915148127893073e-06, "loss": 0.0957, "step": 8322 }, { "epoch": 0.6, "grad_norm": 3.324587618531621, "learning_rate": 3.690394003655043e-06, "loss": 0.2422, "step": 8323 }, { "epoch": 0.6, "grad_norm": 2.579761714790495, "learning_rate": 3.6892732651690655e-06, "loss": 0.1515, "step": 8324 }, { "epoch": 0.6, "grad_norm": 3.9544287257482207, "learning_rate": 3.6881525973918336e-06, "loss": 0.6999, "step": 8325 }, { "epoch": 0.6, "grad_norm": 3.1846496392165973, "learning_rate": 3.6870320003838034e-06, "loss": 0.2149, "step": 8326 }, { "epoch": 0.6, "grad_norm": 3.6791727110937695, "learning_rate": 3.685911474205427e-06, "loss": 0.3172, "step": 8327 }, { "epoch": 0.6, "grad_norm": 25.407254039681867, "learning_rate": 3.684791018917152e-06, "loss": 1.1849, "step": 8328 }, { "epoch": 0.6, "grad_norm": 10.411718094523346, "learning_rate": 3.6836706345794247e-06, "loss": 0.6605, "step": 8329 }, { "epoch": 0.6, "grad_norm": 14.307856751821202, "learning_rate": 3.6825503212526824e-06, "loss": 0.5704, "step": 8330 }, { "epoch": 0.6, "grad_norm": 4.3305210520313, "learning_rate": 3.681430078997364e-06, "loss": 0.7312, "step": 8331 }, { "epoch": 0.6, "grad_norm": 23.0203604294776, "learning_rate": 3.6803099078739023e-06, "loss": 0.6557, "step": 8332 }, { "epoch": 0.6, "grad_norm": 4.641865402619698, "learning_rate": 3.679189807942725e-06, "loss": 0.7558, "step": 8333 }, { "epoch": 0.6, "grad_norm": 9.145556656658396, "learning_rate": 3.6780697792642605e-06, "loss": 0.8041, "step": 8334 }, { "epoch": 0.6, "grad_norm": 2.2369811300460323, "learning_rate": 3.6769498218989252e-06, "loss": 0.1072, "step": 8335 }, { "epoch": 0.6, "grad_norm": 4.5376151983947, "learning_rate": 3.67582993590714e-06, "loss": 0.8583, "step": 8336 }, { "epoch": 0.6, "grad_norm": 16.60136490709481, "learning_rate": 3.6747101213493163e-06, "loss": 1.3004, "step": 8337 }, { "epoch": 0.6, "grad_norm": 6.982272951933264, "learning_rate": 3.6735903782858663e-06, "loss": 0.5208, "step": 8338 }, { "epoch": 0.6, "grad_norm": 3.541938398276591, "learning_rate": 3.6724707067771936e-06, "loss": 0.3214, "step": 8339 }, { "epoch": 0.6, "grad_norm": 20.244984547564453, "learning_rate": 3.671351106883701e-06, "loss": 0.5582, "step": 8340 }, { "epoch": 0.6, "grad_norm": 22.834994979052095, "learning_rate": 3.6702315786657882e-06, "loss": 0.5968, "step": 8341 }, { "epoch": 0.6, "grad_norm": 3.3475875016704895, "learning_rate": 3.6691121221838475e-06, "loss": 0.1724, "step": 8342 }, { "epoch": 0.6, "grad_norm": 10.539660995951872, "learning_rate": 3.667992737498271e-06, "loss": 0.4496, "step": 8343 }, { "epoch": 0.6, "grad_norm": 3.2463942056955353, "learning_rate": 3.6668734246694433e-06, "loss": 0.2338, "step": 8344 }, { "epoch": 0.6, "grad_norm": 4.223984447331439, "learning_rate": 3.665754183757748e-06, "loss": 1.0146, "step": 8345 }, { "epoch": 0.6, "grad_norm": 19.379273763481056, "learning_rate": 3.6646350148235665e-06, "loss": 0.6422, "step": 8346 }, { "epoch": 0.6, "grad_norm": 3.4766707758429773, "learning_rate": 3.6635159179272705e-06, "loss": 0.1974, "step": 8347 }, { "epoch": 0.6, "grad_norm": 4.5286561495573165, "learning_rate": 3.6623968931292296e-06, "loss": 0.8777, "step": 8348 }, { "epoch": 0.6, "grad_norm": 2.439730404227265, "learning_rate": 3.6612779404898146e-06, "loss": 0.0948, "step": 8349 }, { "epoch": 0.6, "grad_norm": 4.350453416069234, "learning_rate": 3.660159060069388e-06, "loss": 0.6267, "step": 8350 }, { "epoch": 0.6, "grad_norm": 13.293122757876061, "learning_rate": 3.6590402519283075e-06, "loss": 0.577, "step": 8351 }, { "epoch": 0.6, "grad_norm": 17.900581557209215, "learning_rate": 3.6579215161269303e-06, "loss": 1.0466, "step": 8352 }, { "epoch": 0.6, "grad_norm": 53.85231823268146, "learning_rate": 3.656802852725606e-06, "loss": 0.599, "step": 8353 }, { "epoch": 0.6, "grad_norm": 4.577857689167011, "learning_rate": 3.6556842617846845e-06, "loss": 0.2864, "step": 8354 }, { "epoch": 0.6, "grad_norm": 3.014898123892212, "learning_rate": 3.654565743364509e-06, "loss": 0.1772, "step": 8355 }, { "epoch": 0.6, "grad_norm": 24.336996901762642, "learning_rate": 3.6534472975254175e-06, "loss": 0.7536, "step": 8356 }, { "epoch": 0.6, "grad_norm": 4.898274165607196, "learning_rate": 3.6523289243277493e-06, "loss": 0.7757, "step": 8357 }, { "epoch": 0.6, "grad_norm": 14.524138811170697, "learning_rate": 3.651210623831834e-06, "loss": 0.7769, "step": 8358 }, { "epoch": 0.6, "grad_norm": 5.4899770028487085, "learning_rate": 3.650092396098e-06, "loss": 0.8388, "step": 8359 }, { "epoch": 0.6, "grad_norm": 3.2817129013670074, "learning_rate": 3.64897424118657e-06, "loss": 0.1916, "step": 8360 }, { "epoch": 0.6, "grad_norm": 10.224251904543173, "learning_rate": 3.6478561591578676e-06, "loss": 0.6623, "step": 8361 }, { "epoch": 0.6, "grad_norm": 10.851618349596139, "learning_rate": 3.6467381500722065e-06, "loss": 0.9438, "step": 8362 }, { "epoch": 0.6, "grad_norm": 20.73516728633794, "learning_rate": 3.645620213989899e-06, "loss": 0.8842, "step": 8363 }, { "epoch": 0.6, "grad_norm": 3.5429503348578235, "learning_rate": 3.6445023509712557e-06, "loss": 0.2417, "step": 8364 }, { "epoch": 0.6, "grad_norm": 3.172723489804151, "learning_rate": 3.6433845610765785e-06, "loss": 0.1549, "step": 8365 }, { "epoch": 0.6, "grad_norm": 4.302820235739094, "learning_rate": 3.642266844366171e-06, "loss": 0.7765, "step": 8366 }, { "epoch": 0.6, "grad_norm": 6.648977409327113, "learning_rate": 3.6411492009003256e-06, "loss": 0.3061, "step": 8367 }, { "epoch": 0.6, "grad_norm": 11.078632949943005, "learning_rate": 3.640031630739338e-06, "loss": 0.4352, "step": 8368 }, { "epoch": 0.6, "grad_norm": 3.6263907758500036, "learning_rate": 3.6389141339434976e-06, "loss": 0.274, "step": 8369 }, { "epoch": 0.6, "grad_norm": 4.259136600828489, "learning_rate": 3.6377967105730884e-06, "loss": 0.7876, "step": 8370 }, { "epoch": 0.6, "grad_norm": 2.7652387070980224, "learning_rate": 3.6366793606883883e-06, "loss": 0.1401, "step": 8371 }, { "epoch": 0.6, "grad_norm": 3.899299968390435, "learning_rate": 3.6355620843496764e-06, "loss": 0.8072, "step": 8372 }, { "epoch": 0.6, "grad_norm": 3.617359688623361, "learning_rate": 3.634444881617226e-06, "loss": 0.3331, "step": 8373 }, { "epoch": 0.6, "grad_norm": 15.198673666282756, "learning_rate": 3.6333277525513044e-06, "loss": 0.2967, "step": 8374 }, { "epoch": 0.6, "grad_norm": 3.2448236993255795, "learning_rate": 3.6322106972121782e-06, "loss": 0.2147, "step": 8375 }, { "epoch": 0.6, "grad_norm": 4.074753411073381, "learning_rate": 3.631093715660107e-06, "loss": 0.808, "step": 8376 }, { "epoch": 0.6, "grad_norm": 30.386541476719334, "learning_rate": 3.6299768079553484e-06, "loss": 1.6137, "step": 8377 }, { "epoch": 0.6, "grad_norm": 16.401568964331297, "learning_rate": 3.6288599741581554e-06, "loss": 0.7418, "step": 8378 }, { "epoch": 0.6, "grad_norm": 3.4934523546261076, "learning_rate": 3.6277432143287765e-06, "loss": 0.2009, "step": 8379 }, { "epoch": 0.6, "grad_norm": 22.147961486091553, "learning_rate": 3.6266265285274582e-06, "loss": 1.0273, "step": 8380 }, { "epoch": 0.6, "grad_norm": 3.620919736273041, "learning_rate": 3.625509916814438e-06, "loss": 0.2204, "step": 8381 }, { "epoch": 0.6, "grad_norm": 3.693185873886618, "learning_rate": 3.624393379249958e-06, "loss": 0.274, "step": 8382 }, { "epoch": 0.6, "grad_norm": 3.2714944904720817, "learning_rate": 3.623276915894246e-06, "loss": 0.1859, "step": 8383 }, { "epoch": 0.6, "grad_norm": 4.91239876027257, "learning_rate": 3.6221605268075345e-06, "loss": 0.9011, "step": 8384 }, { "epoch": 0.6, "grad_norm": 3.5425682983009135, "learning_rate": 3.6210442120500457e-06, "loss": 0.1548, "step": 8385 }, { "epoch": 0.6, "grad_norm": 5.277018754369779, "learning_rate": 3.619927971682002e-06, "loss": 0.8304, "step": 8386 }, { "epoch": 0.6, "grad_norm": 4.565997963008798, "learning_rate": 3.6188118057636213e-06, "loss": 0.214, "step": 8387 }, { "epoch": 0.6, "grad_norm": 15.87191507795931, "learning_rate": 3.617695714355115e-06, "loss": 1.1468, "step": 8388 }, { "epoch": 0.6, "grad_norm": 4.050262331339389, "learning_rate": 3.6165796975166932e-06, "loss": 0.1242, "step": 8389 }, { "epoch": 0.6, "grad_norm": 4.962344256064241, "learning_rate": 3.6154637553085596e-06, "loss": 0.3998, "step": 8390 }, { "epoch": 0.6, "grad_norm": 4.057423547855635, "learning_rate": 3.6143478877909153e-06, "loss": 0.1793, "step": 8391 }, { "epoch": 0.6, "grad_norm": 3.7476757513722343, "learning_rate": 3.6132320950239584e-06, "loss": 0.3011, "step": 8392 }, { "epoch": 0.6, "grad_norm": 17.5749821857652, "learning_rate": 3.61211637706788e-06, "loss": 0.7188, "step": 8393 }, { "epoch": 0.6, "grad_norm": 13.583399976306898, "learning_rate": 3.6110007339828723e-06, "loss": 0.7859, "step": 8394 }, { "epoch": 0.6, "grad_norm": 41.315286933571436, "learning_rate": 3.6098851658291155e-06, "loss": 0.9757, "step": 8395 }, { "epoch": 0.6, "grad_norm": 3.113577908561095, "learning_rate": 3.6087696726667913e-06, "loss": 0.1435, "step": 8396 }, { "epoch": 0.6, "grad_norm": 3.1581369701546014, "learning_rate": 3.607654254556077e-06, "loss": 0.1042, "step": 8397 }, { "epoch": 0.6, "grad_norm": 5.320186536857683, "learning_rate": 3.606538911557147e-06, "loss": 0.9147, "step": 8398 }, { "epoch": 0.6, "grad_norm": 3.4911437523557898, "learning_rate": 3.6054236437301667e-06, "loss": 0.1495, "step": 8399 }, { "epoch": 0.6, "grad_norm": 35.05311277066853, "learning_rate": 3.6043084511353023e-06, "loss": 0.9519, "step": 8400 }, { "epoch": 0.6, "grad_norm": 19.837332349653497, "learning_rate": 3.6031933338327146e-06, "loss": 0.5661, "step": 8401 }, { "epoch": 0.6, "grad_norm": 3.533019566701329, "learning_rate": 3.6020782918825583e-06, "loss": 0.1846, "step": 8402 }, { "epoch": 0.6, "grad_norm": 3.0073505985775326, "learning_rate": 3.6009633253449882e-06, "loss": 0.2005, "step": 8403 }, { "epoch": 0.6, "grad_norm": 3.2430036719727964, "learning_rate": 3.5998484342801498e-06, "loss": 0.2281, "step": 8404 }, { "epoch": 0.6, "grad_norm": 4.612437125950197, "learning_rate": 3.598733618748188e-06, "loss": 0.245, "step": 8405 }, { "epoch": 0.6, "grad_norm": 3.3939577773446596, "learning_rate": 3.5976188788092465e-06, "loss": 0.1299, "step": 8406 }, { "epoch": 0.6, "grad_norm": 3.0341767450370574, "learning_rate": 3.5965042145234562e-06, "loss": 0.2084, "step": 8407 }, { "epoch": 0.6, "grad_norm": 12.033137900064435, "learning_rate": 3.5953896259509506e-06, "loss": 0.743, "step": 8408 }, { "epoch": 0.6, "grad_norm": 15.909309535989394, "learning_rate": 3.594275113151859e-06, "loss": 0.4659, "step": 8409 }, { "epoch": 0.6, "grad_norm": 3.1385645226291907, "learning_rate": 3.5931606761863026e-06, "loss": 0.1064, "step": 8410 }, { "epoch": 0.6, "grad_norm": 15.028597810236311, "learning_rate": 3.5920463151144024e-06, "loss": 0.739, "step": 8411 }, { "epoch": 0.6, "grad_norm": 4.201533150947177, "learning_rate": 3.5909320299962757e-06, "loss": 0.8633, "step": 8412 }, { "epoch": 0.6, "grad_norm": 3.3311421676229553, "learning_rate": 3.5898178208920313e-06, "loss": 0.1643, "step": 8413 }, { "epoch": 0.6, "grad_norm": 4.2703613124311115, "learning_rate": 3.5887036878617776e-06, "loss": 0.7427, "step": 8414 }, { "epoch": 0.6, "grad_norm": 9.482192021082826, "learning_rate": 3.587589630965619e-06, "loss": 0.2937, "step": 8415 }, { "epoch": 0.6, "grad_norm": 2.771172589668282, "learning_rate": 3.586475650263652e-06, "loss": 0.1736, "step": 8416 }, { "epoch": 0.6, "grad_norm": 13.02716439825513, "learning_rate": 3.5853617458159744e-06, "loss": 0.6763, "step": 8417 }, { "epoch": 0.6, "grad_norm": 3.213953994966628, "learning_rate": 3.584247917682677e-06, "loss": 0.1806, "step": 8418 }, { "epoch": 0.6, "grad_norm": 3.010152400023144, "learning_rate": 3.583134165923844e-06, "loss": 0.0944, "step": 8419 }, { "epoch": 0.6, "grad_norm": 4.746855871132046, "learning_rate": 3.5820204905995593e-06, "loss": 0.8374, "step": 8420 }, { "epoch": 0.6, "grad_norm": 3.744620553132419, "learning_rate": 3.5809068917699026e-06, "loss": 0.2647, "step": 8421 }, { "epoch": 0.6, "grad_norm": 15.106583456799719, "learning_rate": 3.579793369494946e-06, "loss": 0.6489, "step": 8422 }, { "epoch": 0.6, "grad_norm": 3.8103320129334217, "learning_rate": 3.5786799238347635e-06, "loss": 0.2566, "step": 8423 }, { "epoch": 0.6, "grad_norm": 16.02985724182002, "learning_rate": 3.5775665548494176e-06, "loss": 0.9221, "step": 8424 }, { "epoch": 0.6, "grad_norm": 18.228443830507292, "learning_rate": 3.5764532625989714e-06, "loss": 0.8589, "step": 8425 }, { "epoch": 0.6, "grad_norm": 16.286181588666267, "learning_rate": 3.5753400471434853e-06, "loss": 0.7687, "step": 8426 }, { "epoch": 0.6, "grad_norm": 3.430814732049235, "learning_rate": 3.5742269085430093e-06, "loss": 0.2035, "step": 8427 }, { "epoch": 0.6, "grad_norm": 3.515740283641421, "learning_rate": 3.5731138468575944e-06, "loss": 0.2198, "step": 8428 }, { "epoch": 0.6, "grad_norm": 3.1895450213525844, "learning_rate": 3.572000862147288e-06, "loss": 0.2194, "step": 8429 }, { "epoch": 0.6, "grad_norm": 24.33411804490577, "learning_rate": 3.5708879544721308e-06, "loss": 0.6376, "step": 8430 }, { "epoch": 0.6, "grad_norm": 21.592602833025108, "learning_rate": 3.5697751238921567e-06, "loss": 1.1396, "step": 8431 }, { "epoch": 0.6, "grad_norm": 4.070187301789489, "learning_rate": 3.5686623704674024e-06, "loss": 0.1629, "step": 8432 }, { "epoch": 0.6, "grad_norm": 11.053611294163723, "learning_rate": 3.5675496942578934e-06, "loss": 0.4953, "step": 8433 }, { "epoch": 0.6, "grad_norm": 16.622894734075796, "learning_rate": 3.566437095323657e-06, "loss": 0.6848, "step": 8434 }, { "epoch": 0.6, "grad_norm": 2.8853473315422917, "learning_rate": 3.5653245737247143e-06, "loss": 0.1936, "step": 8435 }, { "epoch": 0.6, "grad_norm": 3.5021676340476984, "learning_rate": 3.5642121295210785e-06, "loss": 0.2046, "step": 8436 }, { "epoch": 0.6, "grad_norm": 20.120712886646825, "learning_rate": 3.5630997627727647e-06, "loss": 0.763, "step": 8437 }, { "epoch": 0.61, "grad_norm": 3.969386037364166, "learning_rate": 3.5619874735397795e-06, "loss": 0.2495, "step": 8438 }, { "epoch": 0.61, "grad_norm": 5.4734511735983995, "learning_rate": 3.5608752618821264e-06, "loss": 0.1363, "step": 8439 }, { "epoch": 0.61, "grad_norm": 3.57255767117786, "learning_rate": 3.5597631278598066e-06, "loss": 0.32, "step": 8440 }, { "epoch": 0.61, "grad_norm": 3.827313806919275, "learning_rate": 3.5586510715328134e-06, "loss": 0.2803, "step": 8441 }, { "epoch": 0.61, "grad_norm": 2.7203724750902287, "learning_rate": 3.5575390929611407e-06, "loss": 0.1333, "step": 8442 }, { "epoch": 0.61, "grad_norm": 44.308335683158674, "learning_rate": 3.5564271922047727e-06, "loss": 1.3109, "step": 8443 }, { "epoch": 0.61, "grad_norm": 11.43759852288486, "learning_rate": 3.5553153693236943e-06, "loss": 0.6135, "step": 8444 }, { "epoch": 0.61, "grad_norm": 5.879117122103382, "learning_rate": 3.554203624377882e-06, "loss": 0.9198, "step": 8445 }, { "epoch": 0.61, "grad_norm": 4.526979840119844, "learning_rate": 3.5530919574273127e-06, "loss": 0.9375, "step": 8446 }, { "epoch": 0.61, "grad_norm": 4.659649326721015, "learning_rate": 3.5519803685319545e-06, "loss": 0.7808, "step": 8447 }, { "epoch": 0.61, "grad_norm": 4.0684964992313475, "learning_rate": 3.5508688577517737e-06, "loss": 0.76, "step": 8448 }, { "epoch": 0.61, "grad_norm": 3.7055636239542515, "learning_rate": 3.549757425146735e-06, "loss": 0.2901, "step": 8449 }, { "epoch": 0.61, "grad_norm": 3.0121459881121035, "learning_rate": 3.5486460707767913e-06, "loss": 0.1264, "step": 8450 }, { "epoch": 0.61, "grad_norm": 4.370267213485449, "learning_rate": 3.5475347947019e-06, "loss": 0.741, "step": 8451 }, { "epoch": 0.61, "grad_norm": 3.5356998429781235, "learning_rate": 3.546423596982008e-06, "loss": 0.2266, "step": 8452 }, { "epoch": 0.61, "grad_norm": 3.4619310026833956, "learning_rate": 3.545312477677061e-06, "loss": 0.1502, "step": 8453 }, { "epoch": 0.61, "grad_norm": 3.6527133897511352, "learning_rate": 3.544201436847001e-06, "loss": 0.1711, "step": 8454 }, { "epoch": 0.61, "grad_norm": 4.892693445440644, "learning_rate": 3.5430904745517623e-06, "loss": 1.0488, "step": 8455 }, { "epoch": 0.61, "grad_norm": 3.3317174725188843, "learning_rate": 3.5419795908512766e-06, "loss": 0.2219, "step": 8456 }, { "epoch": 0.61, "grad_norm": 3.1555117942029782, "learning_rate": 3.540868785805473e-06, "loss": 0.161, "step": 8457 }, { "epoch": 0.61, "grad_norm": 8.875371044341053, "learning_rate": 3.5397580594742766e-06, "loss": 0.7279, "step": 8458 }, { "epoch": 0.61, "grad_norm": 2.9556388012791164, "learning_rate": 3.538647411917604e-06, "loss": 0.1803, "step": 8459 }, { "epoch": 0.61, "grad_norm": 10.944983197622614, "learning_rate": 3.537536843195374e-06, "loss": 0.6074, "step": 8460 }, { "epoch": 0.61, "grad_norm": 4.036341054944832, "learning_rate": 3.5364263533674936e-06, "loss": 0.8128, "step": 8461 }, { "epoch": 0.61, "grad_norm": 2.594478936642903, "learning_rate": 3.535315942493872e-06, "loss": 0.0844, "step": 8462 }, { "epoch": 0.61, "grad_norm": 3.629089601526326, "learning_rate": 3.5342056106344126e-06, "loss": 0.1826, "step": 8463 }, { "epoch": 0.61, "grad_norm": 18.77398793945695, "learning_rate": 3.53309535784901e-06, "loss": 0.9251, "step": 8464 }, { "epoch": 0.61, "grad_norm": 4.926887694330639, "learning_rate": 3.5319851841975622e-06, "loss": 0.8547, "step": 8465 }, { "epoch": 0.61, "grad_norm": 4.556604894337666, "learning_rate": 3.5308750897399566e-06, "loss": 0.8249, "step": 8466 }, { "epoch": 0.61, "grad_norm": 3.499716337377744, "learning_rate": 3.529765074536079e-06, "loss": 0.1715, "step": 8467 }, { "epoch": 0.61, "grad_norm": 19.229619813000365, "learning_rate": 3.52865513864581e-06, "loss": 1.0226, "step": 8468 }, { "epoch": 0.61, "grad_norm": 2.748251697038906, "learning_rate": 3.5275452821290275e-06, "loss": 0.1178, "step": 8469 }, { "epoch": 0.61, "grad_norm": 4.131330631183319, "learning_rate": 3.5264355050456023e-06, "loss": 0.1821, "step": 8470 }, { "epoch": 0.61, "grad_norm": 3.232011963296692, "learning_rate": 3.5253258074554044e-06, "loss": 0.1886, "step": 8471 }, { "epoch": 0.61, "grad_norm": 4.3103569082792115, "learning_rate": 3.524216189418298e-06, "loss": 0.2937, "step": 8472 }, { "epoch": 0.61, "grad_norm": 4.80238026643842, "learning_rate": 3.52310665099414e-06, "loss": 0.9421, "step": 8473 }, { "epoch": 0.61, "grad_norm": 14.861527589598381, "learning_rate": 3.5219971922427904e-06, "loss": 0.7236, "step": 8474 }, { "epoch": 0.61, "grad_norm": 5.037809910417515, "learning_rate": 3.520887813224096e-06, "loss": 0.9337, "step": 8475 }, { "epoch": 0.61, "grad_norm": 12.437518692662433, "learning_rate": 3.5197785139979057e-06, "loss": 0.5777, "step": 8476 }, { "epoch": 0.61, "grad_norm": 4.531126352805681, "learning_rate": 3.5186692946240618e-06, "loss": 0.8984, "step": 8477 }, { "epoch": 0.61, "grad_norm": 4.485773836535443, "learning_rate": 3.5175601551624037e-06, "loss": 0.1916, "step": 8478 }, { "epoch": 0.61, "grad_norm": 4.377907696885328, "learning_rate": 3.5164510956727626e-06, "loss": 0.5109, "step": 8479 }, { "epoch": 0.61, "grad_norm": 3.136592466791275, "learning_rate": 3.5153421162149693e-06, "loss": 0.1553, "step": 8480 }, { "epoch": 0.61, "grad_norm": 3.4638776568514853, "learning_rate": 3.514233216848849e-06, "loss": 0.2174, "step": 8481 }, { "epoch": 0.61, "grad_norm": 3.4737025066930505, "learning_rate": 3.513124397634222e-06, "loss": 0.1874, "step": 8482 }, { "epoch": 0.61, "grad_norm": 4.82696090460721, "learning_rate": 3.5120156586309072e-06, "loss": 0.7551, "step": 8483 }, { "epoch": 0.61, "grad_norm": 4.854430435496094, "learning_rate": 3.5109069998987145e-06, "loss": 0.9028, "step": 8484 }, { "epoch": 0.61, "grad_norm": 3.006428018758374, "learning_rate": 3.5097984214974522e-06, "loss": 0.1548, "step": 8485 }, { "epoch": 0.61, "grad_norm": 4.057063228814731, "learning_rate": 3.5086899234869253e-06, "loss": 0.2225, "step": 8486 }, { "epoch": 0.61, "grad_norm": 16.461687345334617, "learning_rate": 3.507581505926931e-06, "loss": 0.7398, "step": 8487 }, { "epoch": 0.61, "grad_norm": 4.335325650030094, "learning_rate": 3.506473168877267e-06, "loss": 0.7658, "step": 8488 }, { "epoch": 0.61, "grad_norm": 3.768084694201329, "learning_rate": 3.5053649123977202e-06, "loss": 0.1804, "step": 8489 }, { "epoch": 0.61, "grad_norm": 3.4042050612304235, "learning_rate": 3.504256736548082e-06, "loss": 0.1385, "step": 8490 }, { "epoch": 0.61, "grad_norm": 5.334716490887455, "learning_rate": 3.5031486413881276e-06, "loss": 0.356, "step": 8491 }, { "epoch": 0.61, "grad_norm": 8.274942947398749, "learning_rate": 3.5020406269776406e-06, "loss": 0.4726, "step": 8492 }, { "epoch": 0.61, "grad_norm": 2.8189659426471745, "learning_rate": 3.50093269337639e-06, "loss": 0.1605, "step": 8493 }, { "epoch": 0.61, "grad_norm": 8.674549068238044, "learning_rate": 3.499824840644146e-06, "loss": 0.515, "step": 8494 }, { "epoch": 0.61, "grad_norm": 4.287441525297207, "learning_rate": 3.4987170688406753e-06, "loss": 0.8186, "step": 8495 }, { "epoch": 0.61, "grad_norm": 4.075461539021687, "learning_rate": 3.4976093780257347e-06, "loss": 0.7072, "step": 8496 }, { "epoch": 0.61, "grad_norm": 3.082970132470427, "learning_rate": 3.496501768259082e-06, "loss": 0.2121, "step": 8497 }, { "epoch": 0.61, "grad_norm": 3.117916619248666, "learning_rate": 3.495394239600467e-06, "loss": 0.2138, "step": 8498 }, { "epoch": 0.61, "grad_norm": 6.692637287169395, "learning_rate": 3.494286792109638e-06, "loss": 0.4167, "step": 8499 }, { "epoch": 0.61, "grad_norm": 3.3311555195797755, "learning_rate": 3.4931794258463376e-06, "loss": 0.1242, "step": 8500 }, { "epoch": 0.61, "grad_norm": 15.5011770853593, "learning_rate": 3.4920721408703028e-06, "loss": 0.9067, "step": 8501 }, { "epoch": 0.61, "grad_norm": 43.15791794478365, "learning_rate": 3.490964937241269e-06, "loss": 1.1974, "step": 8502 }, { "epoch": 0.61, "grad_norm": 18.121694991808194, "learning_rate": 3.4898578150189653e-06, "loss": 1.2639, "step": 8503 }, { "epoch": 0.61, "grad_norm": 3.937474303745838, "learning_rate": 3.4887507742631154e-06, "loss": 0.1776, "step": 8504 }, { "epoch": 0.61, "grad_norm": 29.797166483798172, "learning_rate": 3.48764381503344e-06, "loss": 0.9048, "step": 8505 }, { "epoch": 0.61, "grad_norm": 13.24322922712316, "learning_rate": 3.4865369373896572e-06, "loss": 0.5286, "step": 8506 }, { "epoch": 0.61, "grad_norm": 4.581610223422983, "learning_rate": 3.4854301413914775e-06, "loss": 0.9627, "step": 8507 }, { "epoch": 0.61, "grad_norm": 3.501606667685288, "learning_rate": 3.484323427098608e-06, "loss": 0.2922, "step": 8508 }, { "epoch": 0.61, "grad_norm": 3.366853802433284, "learning_rate": 3.483216794570754e-06, "loss": 0.2451, "step": 8509 }, { "epoch": 0.61, "grad_norm": 3.5489562738249387, "learning_rate": 3.4821102438676117e-06, "loss": 0.2081, "step": 8510 }, { "epoch": 0.61, "grad_norm": 3.6185267461209185, "learning_rate": 3.481003775048877e-06, "loss": 0.2408, "step": 8511 }, { "epoch": 0.61, "grad_norm": 3.3903212192927947, "learning_rate": 3.4798973881742383e-06, "loss": 0.1791, "step": 8512 }, { "epoch": 0.61, "grad_norm": 3.5470217512045377, "learning_rate": 3.478791083303381e-06, "loss": 0.1896, "step": 8513 }, { "epoch": 0.61, "grad_norm": 3.458216347001551, "learning_rate": 3.4776848604959877e-06, "loss": 0.228, "step": 8514 }, { "epoch": 0.61, "grad_norm": 2.6264518497127844, "learning_rate": 3.4765787198117353e-06, "loss": 0.1136, "step": 8515 }, { "epoch": 0.61, "grad_norm": 3.489831999591181, "learning_rate": 3.4754726613102917e-06, "loss": 0.2652, "step": 8516 }, { "epoch": 0.61, "grad_norm": 13.90072749699889, "learning_rate": 3.474366685051328e-06, "loss": 0.6227, "step": 8517 }, { "epoch": 0.61, "grad_norm": 18.769669056031024, "learning_rate": 3.473260791094508e-06, "loss": 0.7914, "step": 8518 }, { "epoch": 0.61, "grad_norm": 3.5930921656906727, "learning_rate": 3.472154979499488e-06, "loss": 0.181, "step": 8519 }, { "epoch": 0.61, "grad_norm": 3.3119400509434875, "learning_rate": 3.4710492503259248e-06, "loss": 0.1593, "step": 8520 }, { "epoch": 0.61, "grad_norm": 8.979935678083077, "learning_rate": 3.469943603633466e-06, "loss": 0.5586, "step": 8521 }, { "epoch": 0.61, "grad_norm": 9.692733205112763, "learning_rate": 3.4688380394817578e-06, "loss": 0.3643, "step": 8522 }, { "epoch": 0.61, "grad_norm": 3.103660754975861, "learning_rate": 3.4677325579304434e-06, "loss": 0.2374, "step": 8523 }, { "epoch": 0.61, "grad_norm": 2.8042631793569144, "learning_rate": 3.4666271590391566e-06, "loss": 0.1905, "step": 8524 }, { "epoch": 0.61, "grad_norm": 3.8610171236711266, "learning_rate": 3.4655218428675307e-06, "loss": 0.2652, "step": 8525 }, { "epoch": 0.61, "grad_norm": 4.311663135041268, "learning_rate": 3.4644166094751925e-06, "loss": 0.8909, "step": 8526 }, { "epoch": 0.61, "grad_norm": 3.8573938214603367, "learning_rate": 3.4633114589217677e-06, "loss": 0.2499, "step": 8527 }, { "epoch": 0.61, "grad_norm": 4.343091184320195, "learning_rate": 3.462206391266871e-06, "loss": 0.9182, "step": 8528 }, { "epoch": 0.61, "grad_norm": 3.840527939109492, "learning_rate": 3.4611014065701192e-06, "loss": 0.2088, "step": 8529 }, { "epoch": 0.61, "grad_norm": 3.0976327276547906, "learning_rate": 3.459996504891121e-06, "loss": 0.2261, "step": 8530 }, { "epoch": 0.61, "grad_norm": 4.358346461551397, "learning_rate": 3.4588916862894817e-06, "loss": 0.8963, "step": 8531 }, { "epoch": 0.61, "grad_norm": 2.8266825859172107, "learning_rate": 3.457786950824804e-06, "loss": 0.176, "step": 8532 }, { "epoch": 0.61, "grad_norm": 3.8957311596678315, "learning_rate": 3.4566822985566817e-06, "loss": 0.844, "step": 8533 }, { "epoch": 0.61, "grad_norm": 4.5970820481846895, "learning_rate": 3.455577729544708e-06, "loss": 0.8445, "step": 8534 }, { "epoch": 0.61, "grad_norm": 16.259721047916525, "learning_rate": 3.4544732438484697e-06, "loss": 1.0205, "step": 8535 }, { "epoch": 0.61, "grad_norm": 4.765217388004382, "learning_rate": 3.453368841527549e-06, "loss": 0.7526, "step": 8536 }, { "epoch": 0.61, "grad_norm": 3.8041849119598403, "learning_rate": 3.4522645226415265e-06, "loss": 0.1626, "step": 8537 }, { "epoch": 0.61, "grad_norm": 11.085050608700815, "learning_rate": 3.4511602872499727e-06, "loss": 0.4711, "step": 8538 }, { "epoch": 0.61, "grad_norm": 4.533677419558109, "learning_rate": 3.450056135412462e-06, "loss": 0.9534, "step": 8539 }, { "epoch": 0.61, "grad_norm": 2.8426031682880333, "learning_rate": 3.448952067188554e-06, "loss": 0.1116, "step": 8540 }, { "epoch": 0.61, "grad_norm": 3.190546651985684, "learning_rate": 3.4478480826378103e-06, "loss": 0.1467, "step": 8541 }, { "epoch": 0.61, "grad_norm": 3.5909646926783245, "learning_rate": 3.4467441818197873e-06, "loss": 0.236, "step": 8542 }, { "epoch": 0.61, "grad_norm": 3.794705465074233, "learning_rate": 3.4456403647940375e-06, "loss": 0.6874, "step": 8543 }, { "epoch": 0.61, "grad_norm": 3.7656950809176823, "learning_rate": 3.4445366316201046e-06, "loss": 0.7753, "step": 8544 }, { "epoch": 0.61, "grad_norm": 3.601268817980964, "learning_rate": 3.443432982357533e-06, "loss": 0.2105, "step": 8545 }, { "epoch": 0.61, "grad_norm": 2.751711836178535, "learning_rate": 3.442329417065862e-06, "loss": 0.1672, "step": 8546 }, { "epoch": 0.61, "grad_norm": 4.612160153401434, "learning_rate": 3.4412259358046207e-06, "loss": 0.8755, "step": 8547 }, { "epoch": 0.61, "grad_norm": 2.6976134442520783, "learning_rate": 3.4401225386333415e-06, "loss": 0.1519, "step": 8548 }, { "epoch": 0.61, "grad_norm": 8.348781397821398, "learning_rate": 3.4390192256115457e-06, "loss": 0.4316, "step": 8549 }, { "epoch": 0.61, "grad_norm": 3.0353026872683055, "learning_rate": 3.437915996798754e-06, "loss": 0.1901, "step": 8550 }, { "epoch": 0.61, "grad_norm": 3.9364748599285777, "learning_rate": 3.4368128522544837e-06, "loss": 0.8446, "step": 8551 }, { "epoch": 0.61, "grad_norm": 18.46955670094345, "learning_rate": 3.435709792038242e-06, "loss": 0.7913, "step": 8552 }, { "epoch": 0.61, "grad_norm": 3.675782891837804, "learning_rate": 3.4346068162095346e-06, "loss": 0.1917, "step": 8553 }, { "epoch": 0.61, "grad_norm": 3.9558884111127286, "learning_rate": 3.433503924827865e-06, "loss": 0.1362, "step": 8554 }, { "epoch": 0.61, "grad_norm": 3.4239768394064893, "learning_rate": 3.4324011179527283e-06, "loss": 0.1671, "step": 8555 }, { "epoch": 0.61, "grad_norm": 3.5706827238228267, "learning_rate": 3.431298395643618e-06, "loss": 0.1409, "step": 8556 }, { "epoch": 0.61, "grad_norm": 13.354768310598212, "learning_rate": 3.430195757960022e-06, "loss": 0.3679, "step": 8557 }, { "epoch": 0.61, "grad_norm": 3.5717847927434536, "learning_rate": 3.4290932049614225e-06, "loss": 0.1972, "step": 8558 }, { "epoch": 0.61, "grad_norm": 4.310075271664534, "learning_rate": 3.4279907367072974e-06, "loss": 0.2215, "step": 8559 }, { "epoch": 0.61, "grad_norm": 2.6948828164918504, "learning_rate": 3.426888353257123e-06, "loss": 0.1492, "step": 8560 }, { "epoch": 0.61, "grad_norm": 2.4890583250361416, "learning_rate": 3.4257860546703663e-06, "loss": 0.1113, "step": 8561 }, { "epoch": 0.61, "grad_norm": 11.762463923662812, "learning_rate": 3.4246838410064948e-06, "loss": 0.9327, "step": 8562 }, { "epoch": 0.61, "grad_norm": 4.298929927820857, "learning_rate": 3.4235817123249674e-06, "loss": 0.2499, "step": 8563 }, { "epoch": 0.61, "grad_norm": 5.148552980514055, "learning_rate": 3.422479668685238e-06, "loss": 0.0969, "step": 8564 }, { "epoch": 0.61, "grad_norm": 3.883383867061265, "learning_rate": 3.4213777101467593e-06, "loss": 0.2151, "step": 8565 }, { "epoch": 0.61, "grad_norm": 3.741772950275049, "learning_rate": 3.4202758367689786e-06, "loss": 0.248, "step": 8566 }, { "epoch": 0.61, "grad_norm": 3.275643475034828, "learning_rate": 3.419174048611336e-06, "loss": 0.1551, "step": 8567 }, { "epoch": 0.61, "grad_norm": 14.600548634454308, "learning_rate": 3.418072345733271e-06, "loss": 0.484, "step": 8568 }, { "epoch": 0.61, "grad_norm": 2.7819226389902667, "learning_rate": 3.4169707281942134e-06, "loss": 0.1871, "step": 8569 }, { "epoch": 0.61, "grad_norm": 3.344077392188117, "learning_rate": 3.4158691960535933e-06, "loss": 0.1876, "step": 8570 }, { "epoch": 0.61, "grad_norm": 16.480663501692423, "learning_rate": 3.414767749370835e-06, "loss": 0.6887, "step": 8571 }, { "epoch": 0.61, "grad_norm": 3.694298414122483, "learning_rate": 3.4136663882053543e-06, "loss": 0.1946, "step": 8572 }, { "epoch": 0.61, "grad_norm": 2.6518377608504844, "learning_rate": 3.4125651126165682e-06, "loss": 0.1389, "step": 8573 }, { "epoch": 0.61, "grad_norm": 5.277134986258166, "learning_rate": 3.4114639226638863e-06, "loss": 1.0511, "step": 8574 }, { "epoch": 0.61, "grad_norm": 14.282546845596622, "learning_rate": 3.410362818406714e-06, "loss": 0.6052, "step": 8575 }, { "epoch": 0.61, "grad_norm": 14.58298892104882, "learning_rate": 3.409261799904449e-06, "loss": 1.0681, "step": 8576 }, { "epoch": 0.62, "grad_norm": 16.478533172923562, "learning_rate": 3.4081608672164883e-06, "loss": 0.5026, "step": 8577 }, { "epoch": 0.62, "grad_norm": 3.763586965489225, "learning_rate": 3.407060020402223e-06, "loss": 0.2149, "step": 8578 }, { "epoch": 0.62, "grad_norm": 24.623660568836105, "learning_rate": 3.40595925952104e-06, "loss": 1.0865, "step": 8579 }, { "epoch": 0.62, "grad_norm": 13.061287229767798, "learning_rate": 3.4048585846323223e-06, "loss": 0.4416, "step": 8580 }, { "epoch": 0.62, "grad_norm": 3.0296782567293, "learning_rate": 3.403757995795445e-06, "loss": 0.197, "step": 8581 }, { "epoch": 0.62, "grad_norm": 26.83761611456052, "learning_rate": 3.4026574930697827e-06, "loss": 1.0048, "step": 8582 }, { "epoch": 0.62, "grad_norm": 3.326620478484256, "learning_rate": 3.401557076514701e-06, "loss": 0.118, "step": 8583 }, { "epoch": 0.62, "grad_norm": 3.0248595710313735, "learning_rate": 3.4004567461895654e-06, "loss": 0.1049, "step": 8584 }, { "epoch": 0.62, "grad_norm": 3.4652991185187565, "learning_rate": 3.3993565021537343e-06, "loss": 0.1195, "step": 8585 }, { "epoch": 0.62, "grad_norm": 4.187281085169272, "learning_rate": 3.39825634446656e-06, "loss": 0.2169, "step": 8586 }, { "epoch": 0.62, "grad_norm": 3.2422845798331137, "learning_rate": 3.3971562731873954e-06, "loss": 0.1482, "step": 8587 }, { "epoch": 0.62, "grad_norm": 4.885530411186063, "learning_rate": 3.3960562883755797e-06, "loss": 0.9313, "step": 8588 }, { "epoch": 0.62, "grad_norm": 3.944778828480892, "learning_rate": 3.394956390090458e-06, "loss": 0.2486, "step": 8589 }, { "epoch": 0.62, "grad_norm": 3.47475984332537, "learning_rate": 3.3938565783913623e-06, "loss": 0.1042, "step": 8590 }, { "epoch": 0.62, "grad_norm": 4.530729449610913, "learning_rate": 3.3927568533376265e-06, "loss": 0.8526, "step": 8591 }, { "epoch": 0.62, "grad_norm": 3.848897669417766, "learning_rate": 3.391657214988573e-06, "loss": 0.174, "step": 8592 }, { "epoch": 0.62, "grad_norm": 3.830944668077142, "learning_rate": 3.3905576634035244e-06, "loss": 0.1311, "step": 8593 }, { "epoch": 0.62, "grad_norm": 24.255081183857286, "learning_rate": 3.389458198641799e-06, "loss": 0.7919, "step": 8594 }, { "epoch": 0.62, "grad_norm": 9.893499427643254, "learning_rate": 3.388358820762707e-06, "loss": 0.7057, "step": 8595 }, { "epoch": 0.62, "grad_norm": 3.414309953356682, "learning_rate": 3.3872595298255566e-06, "loss": 0.2845, "step": 8596 }, { "epoch": 0.62, "grad_norm": 20.33292417188185, "learning_rate": 3.386160325889649e-06, "loss": 0.3871, "step": 8597 }, { "epoch": 0.62, "grad_norm": 9.694351630544059, "learning_rate": 3.385061209014284e-06, "loss": 0.4673, "step": 8598 }, { "epoch": 0.62, "grad_norm": 5.034503052095931, "learning_rate": 3.3839621792587553e-06, "loss": 0.8914, "step": 8599 }, { "epoch": 0.62, "grad_norm": 4.388470461138641, "learning_rate": 3.382863236682349e-06, "loss": 0.928, "step": 8600 }, { "epoch": 0.62, "grad_norm": 4.769313518683253, "learning_rate": 3.381764381344349e-06, "loss": 0.7662, "step": 8601 }, { "epoch": 0.62, "grad_norm": 3.4985371424411897, "learning_rate": 3.380665613304035e-06, "loss": 0.2123, "step": 8602 }, { "epoch": 0.62, "grad_norm": 4.132488326213628, "learning_rate": 3.379566932620683e-06, "loss": 0.8257, "step": 8603 }, { "epoch": 0.62, "grad_norm": 4.366235318964845, "learning_rate": 3.37846833935356e-06, "loss": 0.8687, "step": 8604 }, { "epoch": 0.62, "grad_norm": 4.765060830102377, "learning_rate": 3.3773698335619337e-06, "loss": 1.0308, "step": 8605 }, { "epoch": 0.62, "grad_norm": 19.38771246129921, "learning_rate": 3.376271415305062e-06, "loss": 0.8609, "step": 8606 }, { "epoch": 0.62, "grad_norm": 11.336625698238468, "learning_rate": 3.3751730846422006e-06, "loss": 0.6339, "step": 8607 }, { "epoch": 0.62, "grad_norm": 3.000338907810098, "learning_rate": 3.3740748416326024e-06, "loss": 0.1693, "step": 8608 }, { "epoch": 0.62, "grad_norm": 3.6500864560995674, "learning_rate": 3.3729766863355107e-06, "loss": 0.1753, "step": 8609 }, { "epoch": 0.62, "grad_norm": 15.315147663340888, "learning_rate": 3.3718786188101692e-06, "loss": 0.7971, "step": 8610 }, { "epoch": 0.62, "grad_norm": 13.382240451247233, "learning_rate": 3.3707806391158137e-06, "loss": 0.3789, "step": 8611 }, { "epoch": 0.62, "grad_norm": 20.774841036806578, "learning_rate": 3.3696827473116762e-06, "loss": 1.32, "step": 8612 }, { "epoch": 0.62, "grad_norm": 3.315220454123826, "learning_rate": 3.368584943456982e-06, "loss": 0.1663, "step": 8613 }, { "epoch": 0.62, "grad_norm": 10.137006806917011, "learning_rate": 3.3674872276109554e-06, "loss": 0.4871, "step": 8614 }, { "epoch": 0.62, "grad_norm": 11.98448191126571, "learning_rate": 3.3663895998328135e-06, "loss": 0.5954, "step": 8615 }, { "epoch": 0.62, "grad_norm": 14.855634788409134, "learning_rate": 3.365292060181768e-06, "loss": 0.6544, "step": 8616 }, { "epoch": 0.62, "grad_norm": 16.390324864329838, "learning_rate": 3.3641946087170297e-06, "loss": 0.9133, "step": 8617 }, { "epoch": 0.62, "grad_norm": 11.13411980370992, "learning_rate": 3.3630972454977997e-06, "loss": 0.7227, "step": 8618 }, { "epoch": 0.62, "grad_norm": 4.048352772888098, "learning_rate": 3.3619999705832773e-06, "loss": 0.2762, "step": 8619 }, { "epoch": 0.62, "grad_norm": 14.117335486604814, "learning_rate": 3.360902784032656e-06, "loss": 0.7478, "step": 8620 }, { "epoch": 0.62, "grad_norm": 3.2474870889865084, "learning_rate": 3.359805685905125e-06, "loss": 0.2817, "step": 8621 }, { "epoch": 0.62, "grad_norm": 3.165755950670987, "learning_rate": 3.3587086762598704e-06, "loss": 0.243, "step": 8622 }, { "epoch": 0.62, "grad_norm": 2.5906489361590257, "learning_rate": 3.3576117551560707e-06, "loss": 0.1298, "step": 8623 }, { "epoch": 0.62, "grad_norm": 3.4543410233685856, "learning_rate": 3.3565149226528978e-06, "loss": 0.2137, "step": 8624 }, { "epoch": 0.62, "grad_norm": 3.2553393089448224, "learning_rate": 3.355418178809524e-06, "loss": 0.152, "step": 8625 }, { "epoch": 0.62, "grad_norm": 2.8519733242059284, "learning_rate": 3.3543215236851155e-06, "loss": 0.1126, "step": 8626 }, { "epoch": 0.62, "grad_norm": 5.021449305217326, "learning_rate": 3.3532249573388303e-06, "loss": 0.9283, "step": 8627 }, { "epoch": 0.62, "grad_norm": 24.083470544641184, "learning_rate": 3.3521284798298265e-06, "loss": 1.3104, "step": 8628 }, { "epoch": 0.62, "grad_norm": 3.4700630159933192, "learning_rate": 3.3510320912172523e-06, "loss": 0.1076, "step": 8629 }, { "epoch": 0.62, "grad_norm": 19.63622693536505, "learning_rate": 3.3499357915602558e-06, "loss": 1.1626, "step": 8630 }, { "epoch": 0.62, "grad_norm": 13.507263690983509, "learning_rate": 3.348839580917978e-06, "loss": 0.6433, "step": 8631 }, { "epoch": 0.62, "grad_norm": 3.3666952400739105, "learning_rate": 3.347743459349554e-06, "loss": 0.184, "step": 8632 }, { "epoch": 0.62, "grad_norm": 4.075271392937431, "learning_rate": 3.346647426914117e-06, "loss": 0.781, "step": 8633 }, { "epoch": 0.62, "grad_norm": 6.131790397797989, "learning_rate": 3.3455514836707926e-06, "loss": 0.7189, "step": 8634 }, { "epoch": 0.62, "grad_norm": 11.148099021598291, "learning_rate": 3.3444556296787044e-06, "loss": 0.5742, "step": 8635 }, { "epoch": 0.62, "grad_norm": 3.458505816800703, "learning_rate": 3.343359864996967e-06, "loss": 0.2014, "step": 8636 }, { "epoch": 0.62, "grad_norm": 5.53920017641138, "learning_rate": 3.342264189684695e-06, "loss": 0.1632, "step": 8637 }, { "epoch": 0.62, "grad_norm": 12.576154447593781, "learning_rate": 3.3411686038009938e-06, "loss": 0.8074, "step": 8638 }, { "epoch": 0.62, "grad_norm": 14.784943705286762, "learning_rate": 3.340073107404968e-06, "loss": 0.6875, "step": 8639 }, { "epoch": 0.62, "grad_norm": 3.188508025245733, "learning_rate": 3.3389777005557156e-06, "loss": 0.1411, "step": 8640 }, { "epoch": 0.62, "grad_norm": 20.03067309867494, "learning_rate": 3.337882383312328e-06, "loss": 0.6931, "step": 8641 }, { "epoch": 0.62, "grad_norm": 3.433961232823078, "learning_rate": 3.3367871557338952e-06, "loss": 0.1792, "step": 8642 }, { "epoch": 0.62, "grad_norm": 22.37662012100605, "learning_rate": 3.3356920178794996e-06, "loss": 1.0083, "step": 8643 }, { "epoch": 0.62, "grad_norm": 9.034356669466506, "learning_rate": 3.3345969698082186e-06, "loss": 0.5651, "step": 8644 }, { "epoch": 0.62, "grad_norm": 3.844819651220712, "learning_rate": 3.333502011579129e-06, "loss": 0.814, "step": 8645 }, { "epoch": 0.62, "grad_norm": 10.189667948818881, "learning_rate": 3.3324071432512963e-06, "loss": 0.4711, "step": 8646 }, { "epoch": 0.62, "grad_norm": 27.331713333197012, "learning_rate": 3.331312364883789e-06, "loss": 0.69, "step": 8647 }, { "epoch": 0.62, "grad_norm": 2.9527892070640887, "learning_rate": 3.3302176765356607e-06, "loss": 0.1344, "step": 8648 }, { "epoch": 0.62, "grad_norm": 3.776737715114601, "learning_rate": 3.3291230782659684e-06, "loss": 0.2264, "step": 8649 }, { "epoch": 0.62, "grad_norm": 4.277295674251441, "learning_rate": 3.328028570133761e-06, "loss": 0.6183, "step": 8650 }, { "epoch": 0.62, "grad_norm": 3.9335041317961235, "learning_rate": 3.3269341521980847e-06, "loss": 0.2327, "step": 8651 }, { "epoch": 0.62, "grad_norm": 4.022496852042886, "learning_rate": 3.325839824517977e-06, "loss": 0.2802, "step": 8652 }, { "epoch": 0.62, "grad_norm": 4.398161282590985, "learning_rate": 3.3247455871524735e-06, "loss": 0.2612, "step": 8653 }, { "epoch": 0.62, "grad_norm": 3.118843971454462, "learning_rate": 3.3236514401606056e-06, "loss": 0.1312, "step": 8654 }, { "epoch": 0.62, "grad_norm": 3.1046649588673447, "learning_rate": 3.322557383601397e-06, "loss": 0.18, "step": 8655 }, { "epoch": 0.62, "grad_norm": 14.31689178277294, "learning_rate": 3.321463417533868e-06, "loss": 1.0788, "step": 8656 }, { "epoch": 0.62, "grad_norm": 9.589082771137173, "learning_rate": 3.3203695420170333e-06, "loss": 0.6913, "step": 8657 }, { "epoch": 0.62, "grad_norm": 3.2905437299345643, "learning_rate": 3.319275757109904e-06, "loss": 0.1952, "step": 8658 }, { "epoch": 0.62, "grad_norm": 3.316008046989301, "learning_rate": 3.318182062871488e-06, "loss": 0.1801, "step": 8659 }, { "epoch": 0.62, "grad_norm": 4.392413746460844, "learning_rate": 3.317088459360782e-06, "loss": 0.8951, "step": 8660 }, { "epoch": 0.62, "grad_norm": 4.672939435340451, "learning_rate": 3.315994946636783e-06, "loss": 0.6314, "step": 8661 }, { "epoch": 0.62, "grad_norm": 15.25848529414826, "learning_rate": 3.3149015247584835e-06, "loss": 0.9704, "step": 8662 }, { "epoch": 0.62, "grad_norm": 6.9027552559137275, "learning_rate": 3.3138081937848666e-06, "loss": 0.3474, "step": 8663 }, { "epoch": 0.62, "grad_norm": 2.9100675277114156, "learning_rate": 3.312714953774916e-06, "loss": 0.1673, "step": 8664 }, { "epoch": 0.62, "grad_norm": 3.5382619017229118, "learning_rate": 3.311621804787607e-06, "loss": 0.2316, "step": 8665 }, { "epoch": 0.62, "grad_norm": 3.2324494518823035, "learning_rate": 3.31052874688191e-06, "loss": 0.3168, "step": 8666 }, { "epoch": 0.62, "grad_norm": 2.8986047050781325, "learning_rate": 3.3094357801167925e-06, "loss": 0.1704, "step": 8667 }, { "epoch": 0.62, "grad_norm": 10.73965374428321, "learning_rate": 3.308342904551216e-06, "loss": 0.6052, "step": 8668 }, { "epoch": 0.62, "grad_norm": 2.1363734038555, "learning_rate": 3.3072501202441365e-06, "loss": 0.124, "step": 8669 }, { "epoch": 0.62, "grad_norm": 4.727886061688437, "learning_rate": 3.306157427254506e-06, "loss": 0.7667, "step": 8670 }, { "epoch": 0.62, "grad_norm": 6.253005485833949, "learning_rate": 3.3050648256412714e-06, "loss": 0.3642, "step": 8671 }, { "epoch": 0.62, "grad_norm": 9.100854336746082, "learning_rate": 3.303972315463372e-06, "loss": 0.4379, "step": 8672 }, { "epoch": 0.62, "grad_norm": 4.325927102889292, "learning_rate": 3.3028798967797464e-06, "loss": 0.9046, "step": 8673 }, { "epoch": 0.62, "grad_norm": 3.4408302709891467, "learning_rate": 3.3017875696493274e-06, "loss": 0.1643, "step": 8674 }, { "epoch": 0.62, "grad_norm": 3.4102728155590794, "learning_rate": 3.30069533413104e-06, "loss": 0.1701, "step": 8675 }, { "epoch": 0.62, "grad_norm": 5.066594813454459, "learning_rate": 3.299603190283808e-06, "loss": 0.6158, "step": 8676 }, { "epoch": 0.62, "grad_norm": 15.394682690865471, "learning_rate": 3.298511138166547e-06, "loss": 0.7806, "step": 8677 }, { "epoch": 0.62, "grad_norm": 3.642035711972994, "learning_rate": 3.297419177838169e-06, "loss": 0.1924, "step": 8678 }, { "epoch": 0.62, "grad_norm": 4.481293943980722, "learning_rate": 3.296327309357583e-06, "loss": 0.2088, "step": 8679 }, { "epoch": 0.62, "grad_norm": 3.080161624881787, "learning_rate": 3.2952355327836892e-06, "loss": 0.2043, "step": 8680 }, { "epoch": 0.62, "grad_norm": 3.9255874928633774, "learning_rate": 3.294143848175385e-06, "loss": 0.3052, "step": 8681 }, { "epoch": 0.62, "grad_norm": 3.7048940859581365, "learning_rate": 3.2930522555915644e-06, "loss": 0.1763, "step": 8682 }, { "epoch": 0.62, "grad_norm": 4.576845913494811, "learning_rate": 3.291960755091115e-06, "loss": 1.0234, "step": 8683 }, { "epoch": 0.62, "grad_norm": 6.9392571628848465, "learning_rate": 3.2908693467329144e-06, "loss": 0.611, "step": 8684 }, { "epoch": 0.62, "grad_norm": 7.697644314246079, "learning_rate": 3.289778030575845e-06, "loss": 0.2906, "step": 8685 }, { "epoch": 0.62, "grad_norm": 3.976201224821137, "learning_rate": 3.288686806678776e-06, "loss": 0.2259, "step": 8686 }, { "epoch": 0.62, "grad_norm": 3.446136984573256, "learning_rate": 3.2875956751005764e-06, "loss": 0.2611, "step": 8687 }, { "epoch": 0.62, "grad_norm": 11.560264536709349, "learning_rate": 3.286504635900109e-06, "loss": 0.6981, "step": 8688 }, { "epoch": 0.62, "grad_norm": 3.23487437658642, "learning_rate": 3.28541368913623e-06, "loss": 0.1808, "step": 8689 }, { "epoch": 0.62, "grad_norm": 20.443795604767338, "learning_rate": 3.284322834867793e-06, "loss": 0.548, "step": 8690 }, { "epoch": 0.62, "grad_norm": 2.856491662689582, "learning_rate": 3.2832320731536437e-06, "loss": 0.1624, "step": 8691 }, { "epoch": 0.62, "grad_norm": 2.9445414408133574, "learning_rate": 3.2821414040526263e-06, "loss": 0.1456, "step": 8692 }, { "epoch": 0.62, "grad_norm": 11.484065764408653, "learning_rate": 3.2810508276235787e-06, "loss": 1.0261, "step": 8693 }, { "epoch": 0.62, "grad_norm": 3.501734416371643, "learning_rate": 3.2799603439253303e-06, "loss": 0.1627, "step": 8694 }, { "epoch": 0.62, "grad_norm": 4.433204242124856, "learning_rate": 3.2788699530167135e-06, "loss": 0.2172, "step": 8695 }, { "epoch": 0.62, "grad_norm": 40.17170741219283, "learning_rate": 3.277779654956546e-06, "loss": 0.7568, "step": 8696 }, { "epoch": 0.62, "grad_norm": 17.475213607550838, "learning_rate": 3.2766894498036484e-06, "loss": 0.5536, "step": 8697 }, { "epoch": 0.62, "grad_norm": 16.276991951215972, "learning_rate": 3.2755993376168306e-06, "loss": 0.9021, "step": 8698 }, { "epoch": 0.62, "grad_norm": 3.6961310906777345, "learning_rate": 3.2745093184549015e-06, "loss": 0.2422, "step": 8699 }, { "epoch": 0.62, "grad_norm": 4.55659787292978, "learning_rate": 3.273419392376663e-06, "loss": 0.7634, "step": 8700 }, { "epoch": 0.62, "grad_norm": 3.4350788321047694, "learning_rate": 3.2723295594409134e-06, "loss": 0.1897, "step": 8701 }, { "epoch": 0.62, "grad_norm": 4.295676384112759, "learning_rate": 3.2712398197064444e-06, "loss": 0.1671, "step": 8702 }, { "epoch": 0.62, "grad_norm": 9.961578067358063, "learning_rate": 3.270150173232043e-06, "loss": 0.5025, "step": 8703 }, { "epoch": 0.62, "grad_norm": 4.065420665251559, "learning_rate": 3.2690606200764918e-06, "loss": 0.9104, "step": 8704 }, { "epoch": 0.62, "grad_norm": 2.9927710992580825, "learning_rate": 3.2679711602985697e-06, "loss": 0.1822, "step": 8705 }, { "epoch": 0.62, "grad_norm": 20.35135398834106, "learning_rate": 3.266881793957046e-06, "loss": 0.8137, "step": 8706 }, { "epoch": 0.62, "grad_norm": 11.87647328128738, "learning_rate": 3.265792521110691e-06, "loss": 0.5692, "step": 8707 }, { "epoch": 0.62, "grad_norm": 2.7071299080851023, "learning_rate": 3.2647033418182643e-06, "loss": 0.1393, "step": 8708 }, { "epoch": 0.62, "grad_norm": 4.347076899105067, "learning_rate": 3.263614256138523e-06, "loss": 0.8948, "step": 8709 }, { "epoch": 0.62, "grad_norm": 18.361762799766975, "learning_rate": 3.26252526413022e-06, "loss": 0.9275, "step": 8710 }, { "epoch": 0.62, "grad_norm": 4.7515686418206595, "learning_rate": 3.261436365852103e-06, "loss": 0.8288, "step": 8711 }, { "epoch": 0.62, "grad_norm": 5.393134511412783, "learning_rate": 3.2603475613629134e-06, "loss": 0.7727, "step": 8712 }, { "epoch": 0.62, "grad_norm": 4.443875136556763, "learning_rate": 3.2592588507213873e-06, "loss": 0.2803, "step": 8713 }, { "epoch": 0.62, "grad_norm": 4.651940305569615, "learning_rate": 3.2581702339862575e-06, "loss": 0.7243, "step": 8714 }, { "epoch": 0.62, "grad_norm": 119.87044360127342, "learning_rate": 3.2570817112162496e-06, "loss": 1.683, "step": 8715 }, { "epoch": 0.63, "grad_norm": 3.873135053123107, "learning_rate": 3.255993282470087e-06, "loss": 0.1666, "step": 8716 }, { "epoch": 0.63, "grad_norm": 9.532350641068668, "learning_rate": 3.254904947806485e-06, "loss": 0.6646, "step": 8717 }, { "epoch": 0.63, "grad_norm": 2.3441644948099447, "learning_rate": 3.2538167072841552e-06, "loss": 0.1421, "step": 8718 }, { "epoch": 0.63, "grad_norm": 19.558289204775036, "learning_rate": 3.2527285609618055e-06, "loss": 0.7077, "step": 8719 }, { "epoch": 0.63, "grad_norm": 4.771627570509164, "learning_rate": 3.251640508898137e-06, "loss": 0.7529, "step": 8720 }, { "epoch": 0.63, "grad_norm": 3.710850025745178, "learning_rate": 3.2505525511518427e-06, "loss": 0.1688, "step": 8721 }, { "epoch": 0.63, "grad_norm": 3.355621677472135, "learning_rate": 3.2494646877816177e-06, "loss": 0.1418, "step": 8722 }, { "epoch": 0.63, "grad_norm": 14.226959463737366, "learning_rate": 3.2483769188461457e-06, "loss": 0.9857, "step": 8723 }, { "epoch": 0.63, "grad_norm": 5.260617193093416, "learning_rate": 3.2472892444041086e-06, "loss": 0.3297, "step": 8724 }, { "epoch": 0.63, "grad_norm": 3.315342223523702, "learning_rate": 3.2462016645141838e-06, "loss": 0.2865, "step": 8725 }, { "epoch": 0.63, "grad_norm": 11.546928087461058, "learning_rate": 3.2451141792350393e-06, "loss": 0.5522, "step": 8726 }, { "epoch": 0.63, "grad_norm": 16.774083293477045, "learning_rate": 3.2440267886253428e-06, "loss": 0.5067, "step": 8727 }, { "epoch": 0.63, "grad_norm": 6.282292286184014, "learning_rate": 3.2429394927437542e-06, "loss": 0.446, "step": 8728 }, { "epoch": 0.63, "grad_norm": 2.040105326370044, "learning_rate": 3.241852291648928e-06, "loss": 0.0781, "step": 8729 }, { "epoch": 0.63, "grad_norm": 3.2020014698132386, "learning_rate": 3.240765185399518e-06, "loss": 0.2861, "step": 8730 }, { "epoch": 0.63, "grad_norm": 3.627862077597827, "learning_rate": 3.2396781740541655e-06, "loss": 0.2334, "step": 8731 }, { "epoch": 0.63, "grad_norm": 3.2262436711757183, "learning_rate": 3.2385912576715143e-06, "loss": 0.1418, "step": 8732 }, { "epoch": 0.63, "grad_norm": 49.83383360136963, "learning_rate": 3.237504436310196e-06, "loss": 0.6143, "step": 8733 }, { "epoch": 0.63, "grad_norm": 4.69605110601361, "learning_rate": 3.2364177100288424e-06, "loss": 0.8129, "step": 8734 }, { "epoch": 0.63, "grad_norm": 15.322012318631074, "learning_rate": 3.235331078886077e-06, "loss": 0.6832, "step": 8735 }, { "epoch": 0.63, "grad_norm": 24.08159564689988, "learning_rate": 3.234244542940521e-06, "loss": 0.5886, "step": 8736 }, { "epoch": 0.63, "grad_norm": 4.457251539784411, "learning_rate": 3.2331581022507875e-06, "loss": 0.2299, "step": 8737 }, { "epoch": 0.63, "grad_norm": 3.465201943997248, "learning_rate": 3.232071756875487e-06, "loss": 0.1547, "step": 8738 }, { "epoch": 0.63, "grad_norm": 3.021080163066953, "learning_rate": 3.230985506873223e-06, "loss": 0.1306, "step": 8739 }, { "epoch": 0.63, "grad_norm": 3.4382235362376656, "learning_rate": 3.2298993523025944e-06, "loss": 0.1814, "step": 8740 }, { "epoch": 0.63, "grad_norm": 2.6955314311904592, "learning_rate": 3.2288132932221972e-06, "loss": 0.1354, "step": 8741 }, { "epoch": 0.63, "grad_norm": 2.973745360953695, "learning_rate": 3.2277273296906164e-06, "loss": 0.1129, "step": 8742 }, { "epoch": 0.63, "grad_norm": 4.903578957400309, "learning_rate": 3.2266414617664387e-06, "loss": 0.8771, "step": 8743 }, { "epoch": 0.63, "grad_norm": 4.100327625108729, "learning_rate": 3.225555689508243e-06, "loss": 0.8038, "step": 8744 }, { "epoch": 0.63, "grad_norm": 4.344588868796161, "learning_rate": 3.224470012974601e-06, "loss": 0.9283, "step": 8745 }, { "epoch": 0.63, "grad_norm": 3.1061499702961193, "learning_rate": 3.2233844322240794e-06, "loss": 0.1068, "step": 8746 }, { "epoch": 0.63, "grad_norm": 3.206239726309305, "learning_rate": 3.222298947315243e-06, "loss": 0.1921, "step": 8747 }, { "epoch": 0.63, "grad_norm": 3.0005611975871145, "learning_rate": 3.22121355830665e-06, "loss": 0.1577, "step": 8748 }, { "epoch": 0.63, "grad_norm": 4.91726631018366, "learning_rate": 3.220128265256852e-06, "loss": 0.2163, "step": 8749 }, { "epoch": 0.63, "grad_norm": 2.299720565392333, "learning_rate": 3.2190430682243976e-06, "loss": 0.0866, "step": 8750 }, { "epoch": 0.63, "grad_norm": 3.1239837816281835, "learning_rate": 3.217957967267827e-06, "loss": 0.1895, "step": 8751 }, { "epoch": 0.63, "grad_norm": 3.1355635461704585, "learning_rate": 3.2168729624456795e-06, "loss": 0.2274, "step": 8752 }, { "epoch": 0.63, "grad_norm": 3.3062666596239447, "learning_rate": 3.215788053816487e-06, "loss": 0.1992, "step": 8753 }, { "epoch": 0.63, "grad_norm": 16.67668942954114, "learning_rate": 3.214703241438774e-06, "loss": 0.6045, "step": 8754 }, { "epoch": 0.63, "grad_norm": 2.8108685416283463, "learning_rate": 3.213618525371065e-06, "loss": 0.1519, "step": 8755 }, { "epoch": 0.63, "grad_norm": 3.832555464321777, "learning_rate": 3.212533905671874e-06, "loss": 0.2049, "step": 8756 }, { "epoch": 0.63, "grad_norm": 3.3288616490809657, "learning_rate": 3.2114493823997132e-06, "loss": 0.1633, "step": 8757 }, { "epoch": 0.63, "grad_norm": 3.204057666848846, "learning_rate": 3.2103649556130878e-06, "loss": 0.1904, "step": 8758 }, { "epoch": 0.63, "grad_norm": 3.1499719249960467, "learning_rate": 3.209280625370499e-06, "loss": 0.1291, "step": 8759 }, { "epoch": 0.63, "grad_norm": 1.9243416531754995, "learning_rate": 3.2081963917304416e-06, "loss": 0.0927, "step": 8760 }, { "epoch": 0.63, "grad_norm": 4.4775255978699775, "learning_rate": 3.2071122547514067e-06, "loss": 0.7948, "step": 8761 }, { "epoch": 0.63, "grad_norm": 13.32115012322924, "learning_rate": 3.2060282144918796e-06, "loss": 0.4602, "step": 8762 }, { "epoch": 0.63, "grad_norm": 2.6882494356179705, "learning_rate": 3.2049442710103395e-06, "loss": 0.1533, "step": 8763 }, { "epoch": 0.63, "grad_norm": 17.170454795235848, "learning_rate": 3.2038604243652617e-06, "loss": 0.7172, "step": 8764 }, { "epoch": 0.63, "grad_norm": 4.574360930998386, "learning_rate": 3.2027766746151144e-06, "loss": 0.7732, "step": 8765 }, { "epoch": 0.63, "grad_norm": 13.63338694579094, "learning_rate": 3.2016930218183628e-06, "loss": 0.975, "step": 8766 }, { "epoch": 0.63, "grad_norm": 3.5764412642834422, "learning_rate": 3.200609466033467e-06, "loss": 0.1581, "step": 8767 }, { "epoch": 0.63, "grad_norm": 22.505489813862365, "learning_rate": 3.19952600731888e-06, "loss": 1.1109, "step": 8768 }, { "epoch": 0.63, "grad_norm": 3.2153133463603347, "learning_rate": 3.1984426457330475e-06, "loss": 0.1798, "step": 8769 }, { "epoch": 0.63, "grad_norm": 47.16967090851763, "learning_rate": 3.1973593813344162e-06, "loss": 0.4171, "step": 8770 }, { "epoch": 0.63, "grad_norm": 4.533038413415603, "learning_rate": 3.196276214181423e-06, "loss": 0.8483, "step": 8771 }, { "epoch": 0.63, "grad_norm": 11.667624070994576, "learning_rate": 3.195193144332499e-06, "loss": 0.5641, "step": 8772 }, { "epoch": 0.63, "grad_norm": 8.850698762133502, "learning_rate": 3.1941101718460756e-06, "loss": 0.3464, "step": 8773 }, { "epoch": 0.63, "grad_norm": 2.9435543986034034, "learning_rate": 3.193027296780571e-06, "loss": 0.1794, "step": 8774 }, { "epoch": 0.63, "grad_norm": 3.279773050670282, "learning_rate": 3.191944519194404e-06, "loss": 0.2707, "step": 8775 }, { "epoch": 0.63, "grad_norm": 21.441240546431942, "learning_rate": 3.1908618391459874e-06, "loss": 0.5452, "step": 8776 }, { "epoch": 0.63, "grad_norm": 3.500096019561475, "learning_rate": 3.189779256693726e-06, "loss": 0.2543, "step": 8777 }, { "epoch": 0.63, "grad_norm": 20.285102681871045, "learning_rate": 3.1886967718960214e-06, "loss": 0.6785, "step": 8778 }, { "epoch": 0.63, "grad_norm": 4.109803922232466, "learning_rate": 3.1876143848112702e-06, "loss": 0.6627, "step": 8779 }, { "epoch": 0.63, "grad_norm": 3.5764149357692436, "learning_rate": 3.186532095497864e-06, "loss": 0.165, "step": 8780 }, { "epoch": 0.63, "grad_norm": 11.985629248939098, "learning_rate": 3.185449904014185e-06, "loss": 0.4896, "step": 8781 }, { "epoch": 0.63, "grad_norm": 25.908403466978175, "learning_rate": 3.1843678104186167e-06, "loss": 1.0334, "step": 8782 }, { "epoch": 0.63, "grad_norm": 12.240133148490935, "learning_rate": 3.183285814769531e-06, "loss": 0.7108, "step": 8783 }, { "epoch": 0.63, "grad_norm": 4.38323443652533, "learning_rate": 3.182203917125299e-06, "loss": 0.1938, "step": 8784 }, { "epoch": 0.63, "grad_norm": 18.10554481443324, "learning_rate": 3.1811221175442864e-06, "loss": 0.649, "step": 8785 }, { "epoch": 0.63, "grad_norm": 2.517726902509289, "learning_rate": 3.1800404160848487e-06, "loss": 0.0888, "step": 8786 }, { "epoch": 0.63, "grad_norm": 4.376358404691623, "learning_rate": 3.1789588128053433e-06, "loss": 0.6651, "step": 8787 }, { "epoch": 0.63, "grad_norm": 18.55573413867512, "learning_rate": 3.1778773077641157e-06, "loss": 0.7228, "step": 8788 }, { "epoch": 0.63, "grad_norm": 13.26540243386919, "learning_rate": 3.17679590101951e-06, "loss": 0.5712, "step": 8789 }, { "epoch": 0.63, "grad_norm": 3.593273525296932, "learning_rate": 3.175714592629866e-06, "loss": 0.2436, "step": 8790 }, { "epoch": 0.63, "grad_norm": 2.613100684523751, "learning_rate": 3.1746333826535137e-06, "loss": 0.0974, "step": 8791 }, { "epoch": 0.63, "grad_norm": 17.047987694171287, "learning_rate": 3.173552271148782e-06, "loss": 0.8537, "step": 8792 }, { "epoch": 0.63, "grad_norm": 10.11086462689075, "learning_rate": 3.1724712581739914e-06, "loss": 0.5242, "step": 8793 }, { "epoch": 0.63, "grad_norm": 3.6136434426521413, "learning_rate": 3.1713903437874578e-06, "loss": 0.3038, "step": 8794 }, { "epoch": 0.63, "grad_norm": 8.281495739166099, "learning_rate": 3.1703095280474933e-06, "loss": 0.9372, "step": 8795 }, { "epoch": 0.63, "grad_norm": 12.330590759023751, "learning_rate": 3.169228811012405e-06, "loss": 0.5623, "step": 8796 }, { "epoch": 0.63, "grad_norm": 3.0292183390263223, "learning_rate": 3.1681481927404916e-06, "loss": 0.1625, "step": 8797 }, { "epoch": 0.63, "grad_norm": 17.247964444254844, "learning_rate": 3.1670676732900496e-06, "loss": 1.013, "step": 8798 }, { "epoch": 0.63, "grad_norm": 3.0890542193841024, "learning_rate": 3.165987252719369e-06, "loss": 0.1252, "step": 8799 }, { "epoch": 0.63, "grad_norm": 25.466769488003102, "learning_rate": 3.164906931086733e-06, "loss": 0.5492, "step": 8800 }, { "epoch": 0.63, "grad_norm": 3.547567137718335, "learning_rate": 3.1638267084504233e-06, "loss": 0.2254, "step": 8801 }, { "epoch": 0.63, "grad_norm": 21.700556079603942, "learning_rate": 3.1627465848687113e-06, "loss": 0.5059, "step": 8802 }, { "epoch": 0.63, "grad_norm": 9.028134549539715, "learning_rate": 3.1616665603998664e-06, "loss": 0.4939, "step": 8803 }, { "epoch": 0.63, "grad_norm": 3.5060639579587556, "learning_rate": 3.160586635102154e-06, "loss": 0.2025, "step": 8804 }, { "epoch": 0.63, "grad_norm": 4.009161135996136, "learning_rate": 3.1595068090338292e-06, "loss": 0.343, "step": 8805 }, { "epoch": 0.63, "grad_norm": 4.669208780062693, "learning_rate": 3.158427082253145e-06, "loss": 0.9733, "step": 8806 }, { "epoch": 0.63, "grad_norm": 14.742349550075213, "learning_rate": 3.1573474548183493e-06, "loss": 0.6202, "step": 8807 }, { "epoch": 0.63, "grad_norm": 22.741217348013834, "learning_rate": 3.1562679267876827e-06, "loss": 0.8361, "step": 8808 }, { "epoch": 0.63, "grad_norm": 4.225628333219183, "learning_rate": 3.1551884982193827e-06, "loss": 0.7924, "step": 8809 }, { "epoch": 0.63, "grad_norm": 3.7481446882312826, "learning_rate": 3.1541091691716806e-06, "loss": 0.246, "step": 8810 }, { "epoch": 0.63, "grad_norm": 8.940325188241319, "learning_rate": 3.153029939702802e-06, "loss": 0.5668, "step": 8811 }, { "epoch": 0.63, "grad_norm": 2.3162694071051813, "learning_rate": 3.151950809870966e-06, "loss": 0.1238, "step": 8812 }, { "epoch": 0.63, "grad_norm": 3.2287591116088974, "learning_rate": 3.1508717797343903e-06, "loss": 0.2363, "step": 8813 }, { "epoch": 0.63, "grad_norm": 30.874263616085983, "learning_rate": 3.1497928493512808e-06, "loss": 0.877, "step": 8814 }, { "epoch": 0.63, "grad_norm": 3.89652794231246, "learning_rate": 3.1487140187798453e-06, "loss": 0.2113, "step": 8815 }, { "epoch": 0.63, "grad_norm": 4.248318228493418, "learning_rate": 3.147635288078282e-06, "loss": 0.2089, "step": 8816 }, { "epoch": 0.63, "grad_norm": 4.610378356045033, "learning_rate": 3.1465566573047814e-06, "loss": 1.0414, "step": 8817 }, { "epoch": 0.63, "grad_norm": 16.151831182176608, "learning_rate": 3.1454781265175327e-06, "loss": 0.7169, "step": 8818 }, { "epoch": 0.63, "grad_norm": 14.986717591104632, "learning_rate": 3.1443996957747215e-06, "loss": 0.8772, "step": 8819 }, { "epoch": 0.63, "grad_norm": 13.482184139011334, "learning_rate": 3.1433213651345206e-06, "loss": 0.8318, "step": 8820 }, { "epoch": 0.63, "grad_norm": 3.298738610112816, "learning_rate": 3.1422431346551064e-06, "loss": 0.2712, "step": 8821 }, { "epoch": 0.63, "grad_norm": 4.279730146335051, "learning_rate": 3.1411650043946417e-06, "loss": 0.1961, "step": 8822 }, { "epoch": 0.63, "grad_norm": 3.4310240824456018, "learning_rate": 3.1400869744112885e-06, "loss": 0.1933, "step": 8823 }, { "epoch": 0.63, "grad_norm": 9.873327775436296, "learning_rate": 3.139009044763204e-06, "loss": 0.8854, "step": 8824 }, { "epoch": 0.63, "grad_norm": 3.396744612417202, "learning_rate": 3.137931215508536e-06, "loss": 0.1683, "step": 8825 }, { "epoch": 0.63, "grad_norm": 2.9889350760488615, "learning_rate": 3.136853486705431e-06, "loss": 0.1679, "step": 8826 }, { "epoch": 0.63, "grad_norm": 15.732435418045375, "learning_rate": 3.1357758584120294e-06, "loss": 0.7585, "step": 8827 }, { "epoch": 0.63, "grad_norm": 4.839915077845547, "learning_rate": 3.1346983306864646e-06, "loss": 0.8259, "step": 8828 }, { "epoch": 0.63, "grad_norm": 12.415841687910861, "learning_rate": 3.1336209035868626e-06, "loss": 0.4426, "step": 8829 }, { "epoch": 0.63, "grad_norm": 2.948033309134594, "learning_rate": 3.1325435771713485e-06, "loss": 0.1605, "step": 8830 }, { "epoch": 0.63, "grad_norm": 3.900467332622024, "learning_rate": 3.1314663514980393e-06, "loss": 0.8546, "step": 8831 }, { "epoch": 0.63, "grad_norm": 13.43487083169094, "learning_rate": 3.130389226625047e-06, "loss": 0.9738, "step": 8832 }, { "epoch": 0.63, "grad_norm": 3.2059631680269356, "learning_rate": 3.1293122026104806e-06, "loss": 0.1718, "step": 8833 }, { "epoch": 0.63, "grad_norm": 4.500123358700665, "learning_rate": 3.128235279512438e-06, "loss": 0.1478, "step": 8834 }, { "epoch": 0.63, "grad_norm": 5.197030323865688, "learning_rate": 3.1271584573890192e-06, "loss": 0.2236, "step": 8835 }, { "epoch": 0.63, "grad_norm": 4.7645972929100555, "learning_rate": 3.126081736298311e-06, "loss": 0.8166, "step": 8836 }, { "epoch": 0.63, "grad_norm": 3.7488631462554802, "learning_rate": 3.1250051162983996e-06, "loss": 0.2766, "step": 8837 }, { "epoch": 0.63, "grad_norm": 7.35532802120582, "learning_rate": 3.1239285974473665e-06, "loss": 0.5042, "step": 8838 }, { "epoch": 0.63, "grad_norm": 2.877939406193913, "learning_rate": 3.1228521798032833e-06, "loss": 0.1557, "step": 8839 }, { "epoch": 0.63, "grad_norm": 3.6172216575368012, "learning_rate": 3.1217758634242218e-06, "loss": 0.2528, "step": 8840 }, { "epoch": 0.63, "grad_norm": 15.746544794676838, "learning_rate": 3.1206996483682404e-06, "loss": 0.3358, "step": 8841 }, { "epoch": 0.63, "grad_norm": 4.2505292307427816, "learning_rate": 3.1196235346934013e-06, "loss": 0.7936, "step": 8842 }, { "epoch": 0.63, "grad_norm": 3.499498580725548, "learning_rate": 3.118547522457753e-06, "loss": 0.1986, "step": 8843 }, { "epoch": 0.63, "grad_norm": 2.957285696453485, "learning_rate": 3.1174716117193447e-06, "loss": 0.1528, "step": 8844 }, { "epoch": 0.63, "grad_norm": 3.2979453800542022, "learning_rate": 3.1163958025362173e-06, "loss": 0.131, "step": 8845 }, { "epoch": 0.63, "grad_norm": 4.424555119929277, "learning_rate": 3.1153200949664065e-06, "loss": 0.9901, "step": 8846 }, { "epoch": 0.63, "grad_norm": 12.852735598751098, "learning_rate": 3.1142444890679436e-06, "loss": 0.6677, "step": 8847 }, { "epoch": 0.63, "grad_norm": 15.909742354408197, "learning_rate": 3.1131689848988513e-06, "loss": 0.8024, "step": 8848 }, { "epoch": 0.63, "grad_norm": 2.820741687939107, "learning_rate": 3.1120935825171505e-06, "loss": 0.1294, "step": 8849 }, { "epoch": 0.63, "grad_norm": 15.977432527702458, "learning_rate": 3.111018281980855e-06, "loss": 0.6431, "step": 8850 }, { "epoch": 0.63, "grad_norm": 3.300528562911868, "learning_rate": 3.1099430833479728e-06, "loss": 0.187, "step": 8851 }, { "epoch": 0.63, "grad_norm": 9.904508853028071, "learning_rate": 3.108867986676508e-06, "loss": 0.5424, "step": 8852 }, { "epoch": 0.63, "grad_norm": 2.980362542473442, "learning_rate": 3.107792992024457e-06, "loss": 0.1768, "step": 8853 }, { "epoch": 0.63, "grad_norm": 3.286736423792628, "learning_rate": 3.1067180994498104e-06, "loss": 0.2303, "step": 8854 }, { "epoch": 0.63, "grad_norm": 3.3409397733747936, "learning_rate": 3.105643309010556e-06, "loss": 0.1458, "step": 8855 }, { "epoch": 0.64, "grad_norm": 20.76694726448807, "learning_rate": 3.104568620764675e-06, "loss": 0.8557, "step": 8856 }, { "epoch": 0.64, "grad_norm": 23.084203353788403, "learning_rate": 3.103494034770142e-06, "loss": 0.8704, "step": 8857 }, { "epoch": 0.64, "grad_norm": 13.018920340204192, "learning_rate": 3.102419551084928e-06, "loss": 0.722, "step": 8858 }, { "epoch": 0.64, "grad_norm": 17.952842604502568, "learning_rate": 3.101345169766995e-06, "loss": 0.6726, "step": 8859 }, { "epoch": 0.64, "grad_norm": 4.322844312532762, "learning_rate": 3.100270890874304e-06, "loss": 0.2231, "step": 8860 }, { "epoch": 0.64, "grad_norm": 8.315935332650948, "learning_rate": 3.099196714464808e-06, "loss": 0.5493, "step": 8861 }, { "epoch": 0.64, "grad_norm": 2.793247248840243, "learning_rate": 3.098122640596454e-06, "loss": 0.1633, "step": 8862 }, { "epoch": 0.64, "grad_norm": 7.9753988010201855, "learning_rate": 3.097048669327185e-06, "loss": 0.4483, "step": 8863 }, { "epoch": 0.64, "grad_norm": 3.7378804878466636, "learning_rate": 3.095974800714937e-06, "loss": 0.1504, "step": 8864 }, { "epoch": 0.64, "grad_norm": 2.5563123053387797, "learning_rate": 3.094901034817642e-06, "loss": 0.127, "step": 8865 }, { "epoch": 0.64, "grad_norm": 2.819691367606965, "learning_rate": 3.093827371693224e-06, "loss": 0.1134, "step": 8866 }, { "epoch": 0.64, "grad_norm": 4.117783096696119, "learning_rate": 3.092753811399605e-06, "loss": 0.1846, "step": 8867 }, { "epoch": 0.64, "grad_norm": 2.940280683095799, "learning_rate": 3.0916803539946972e-06, "loss": 0.1607, "step": 8868 }, { "epoch": 0.64, "grad_norm": 4.308661083072382, "learning_rate": 3.090606999536411e-06, "loss": 0.9362, "step": 8869 }, { "epoch": 0.64, "grad_norm": 15.494385227537633, "learning_rate": 3.0895337480826503e-06, "loss": 0.6813, "step": 8870 }, { "epoch": 0.64, "grad_norm": 3.4840618675659787, "learning_rate": 3.088460599691312e-06, "loss": 0.1291, "step": 8871 }, { "epoch": 0.64, "grad_norm": 3.1462853686250383, "learning_rate": 3.0873875544202895e-06, "loss": 0.1642, "step": 8872 }, { "epoch": 0.64, "grad_norm": 4.325289127963288, "learning_rate": 3.086314612327468e-06, "loss": 0.815, "step": 8873 }, { "epoch": 0.64, "grad_norm": 5.612984779983343, "learning_rate": 3.0852417734707296e-06, "loss": 0.3118, "step": 8874 }, { "epoch": 0.64, "grad_norm": 4.099810720327698, "learning_rate": 3.08416903790795e-06, "loss": 0.1487, "step": 8875 }, { "epoch": 0.64, "grad_norm": 17.47240220604797, "learning_rate": 3.0830964056970007e-06, "loss": 1.0985, "step": 8876 }, { "epoch": 0.64, "grad_norm": 12.065798940752542, "learning_rate": 3.082023876895742e-06, "loss": 0.5916, "step": 8877 }, { "epoch": 0.64, "grad_norm": 3.760574369044922, "learning_rate": 3.080951451562036e-06, "loss": 0.678, "step": 8878 }, { "epoch": 0.64, "grad_norm": 15.220195373935107, "learning_rate": 3.0798791297537356e-06, "loss": 1.0026, "step": 8879 }, { "epoch": 0.64, "grad_norm": 22.59376150496832, "learning_rate": 3.0788069115286873e-06, "loss": 0.7284, "step": 8880 }, { "epoch": 0.64, "grad_norm": 14.49743320310795, "learning_rate": 3.0777347969447346e-06, "loss": 0.7827, "step": 8881 }, { "epoch": 0.64, "grad_norm": 3.6718189464415647, "learning_rate": 3.0766627860597133e-06, "loss": 0.2682, "step": 8882 }, { "epoch": 0.64, "grad_norm": 4.79372615009069, "learning_rate": 3.0755908789314537e-06, "loss": 0.2163, "step": 8883 }, { "epoch": 0.64, "grad_norm": 5.908900427911576, "learning_rate": 3.0745190756177833e-06, "loss": 0.2247, "step": 8884 }, { "epoch": 0.64, "grad_norm": 21.16753339680398, "learning_rate": 3.0734473761765195e-06, "loss": 0.6229, "step": 8885 }, { "epoch": 0.64, "grad_norm": 2.4890266077239147, "learning_rate": 3.0723757806654776e-06, "loss": 0.1024, "step": 8886 }, { "epoch": 0.64, "grad_norm": 4.52112476617366, "learning_rate": 3.071304289142465e-06, "loss": 0.9299, "step": 8887 }, { "epoch": 0.64, "grad_norm": 9.960412949238258, "learning_rate": 3.070232901665288e-06, "loss": 0.5002, "step": 8888 }, { "epoch": 0.64, "grad_norm": 20.969963043915648, "learning_rate": 3.0691616182917394e-06, "loss": 0.8876, "step": 8889 }, { "epoch": 0.64, "grad_norm": 3.5935159104771017, "learning_rate": 3.068090439079614e-06, "loss": 0.152, "step": 8890 }, { "epoch": 0.64, "grad_norm": 12.481527092916807, "learning_rate": 3.067019364086695e-06, "loss": 0.8949, "step": 8891 }, { "epoch": 0.64, "grad_norm": 4.499131067683668, "learning_rate": 3.0659483933707656e-06, "loss": 0.835, "step": 8892 }, { "epoch": 0.64, "grad_norm": 15.556061157708816, "learning_rate": 3.0648775269895997e-06, "loss": 0.7249, "step": 8893 }, { "epoch": 0.64, "grad_norm": 13.702472977346524, "learning_rate": 3.0638067650009657e-06, "loss": 0.3462, "step": 8894 }, { "epoch": 0.64, "grad_norm": 12.865390564014108, "learning_rate": 3.0627361074626288e-06, "loss": 0.5682, "step": 8895 }, { "epoch": 0.64, "grad_norm": 10.500762236102519, "learning_rate": 3.0616655544323447e-06, "loss": 0.2626, "step": 8896 }, { "epoch": 0.64, "grad_norm": 13.751074658429301, "learning_rate": 3.060595105967867e-06, "loss": 1.0382, "step": 8897 }, { "epoch": 0.64, "grad_norm": 15.218350535304852, "learning_rate": 3.0595247621269443e-06, "loss": 0.965, "step": 8898 }, { "epoch": 0.64, "grad_norm": 16.63811719153045, "learning_rate": 3.0584545229673133e-06, "loss": 1.1196, "step": 8899 }, { "epoch": 0.64, "grad_norm": 3.535921622536025, "learning_rate": 3.0573843885467147e-06, "loss": 0.2237, "step": 8900 }, { "epoch": 0.64, "grad_norm": 17.693924808409932, "learning_rate": 3.056314358922874e-06, "loss": 0.7714, "step": 8901 }, { "epoch": 0.64, "grad_norm": 3.799135665441515, "learning_rate": 3.055244434153515e-06, "loss": 0.1851, "step": 8902 }, { "epoch": 0.64, "grad_norm": 3.163142685436623, "learning_rate": 3.054174614296357e-06, "loss": 0.1296, "step": 8903 }, { "epoch": 0.64, "grad_norm": 15.386166965317924, "learning_rate": 3.053104899409115e-06, "loss": 0.4912, "step": 8904 }, { "epoch": 0.64, "grad_norm": 20.67387823310968, "learning_rate": 3.0520352895494925e-06, "loss": 1.2925, "step": 8905 }, { "epoch": 0.64, "grad_norm": 4.090353305916967, "learning_rate": 3.050965784775193e-06, "loss": 0.2205, "step": 8906 }, { "epoch": 0.64, "grad_norm": 4.495523976636465, "learning_rate": 3.0498963851439136e-06, "loss": 0.121, "step": 8907 }, { "epoch": 0.64, "grad_norm": 4.584016231111106, "learning_rate": 3.0488270907133412e-06, "loss": 0.9693, "step": 8908 }, { "epoch": 0.64, "grad_norm": 6.315282002782171, "learning_rate": 3.0477579015411625e-06, "loss": 0.2667, "step": 8909 }, { "epoch": 0.64, "grad_norm": 8.841518160631031, "learning_rate": 3.046688817685054e-06, "loss": 0.6677, "step": 8910 }, { "epoch": 0.64, "grad_norm": 15.820299811903284, "learning_rate": 3.045619839202691e-06, "loss": 0.8212, "step": 8911 }, { "epoch": 0.64, "grad_norm": 14.216060006183346, "learning_rate": 3.0445509661517416e-06, "loss": 0.5168, "step": 8912 }, { "epoch": 0.64, "grad_norm": 21.163815121794997, "learning_rate": 3.043482198589864e-06, "loss": 0.8412, "step": 8913 }, { "epoch": 0.64, "grad_norm": 4.702085295333963, "learning_rate": 3.0424135365747164e-06, "loss": 0.8212, "step": 8914 }, { "epoch": 0.64, "grad_norm": 10.530548863303968, "learning_rate": 3.0413449801639484e-06, "loss": 0.9183, "step": 8915 }, { "epoch": 0.64, "grad_norm": 13.859999669153858, "learning_rate": 3.0402765294152048e-06, "loss": 1.1069, "step": 8916 }, { "epoch": 0.64, "grad_norm": 3.5790480472839237, "learning_rate": 3.039208184386124e-06, "loss": 0.1825, "step": 8917 }, { "epoch": 0.64, "grad_norm": 7.417447352339009, "learning_rate": 3.038139945134341e-06, "loss": 0.4313, "step": 8918 }, { "epoch": 0.64, "grad_norm": 19.292614150684486, "learning_rate": 3.03707181171748e-06, "loss": 0.6683, "step": 8919 }, { "epoch": 0.64, "grad_norm": 5.220047061798587, "learning_rate": 3.0360037841931654e-06, "loss": 0.9482, "step": 8920 }, { "epoch": 0.64, "grad_norm": 12.752681146924937, "learning_rate": 3.0349358626190136e-06, "loss": 0.5337, "step": 8921 }, { "epoch": 0.64, "grad_norm": 2.7943168585607046, "learning_rate": 3.033868047052632e-06, "loss": 0.1806, "step": 8922 }, { "epoch": 0.64, "grad_norm": 3.317221622842274, "learning_rate": 3.032800337551629e-06, "loss": 0.1732, "step": 8923 }, { "epoch": 0.64, "grad_norm": 3.9773414433591214, "learning_rate": 3.0317327341736024e-06, "loss": 0.1955, "step": 8924 }, { "epoch": 0.64, "grad_norm": 28.236723415702095, "learning_rate": 3.0306652369761425e-06, "loss": 1.1255, "step": 8925 }, { "epoch": 0.64, "grad_norm": 4.525143682470397, "learning_rate": 3.029597846016838e-06, "loss": 0.6393, "step": 8926 }, { "epoch": 0.64, "grad_norm": 9.732680086938842, "learning_rate": 3.028530561353273e-06, "loss": 0.7171, "step": 8927 }, { "epoch": 0.64, "grad_norm": 5.159535027389822, "learning_rate": 3.0274633830430205e-06, "loss": 0.9629, "step": 8928 }, { "epoch": 0.64, "grad_norm": 3.1441196987567332, "learning_rate": 3.026396311143653e-06, "loss": 0.173, "step": 8929 }, { "epoch": 0.64, "grad_norm": 4.370796288686838, "learning_rate": 3.025329345712733e-06, "loss": 0.8087, "step": 8930 }, { "epoch": 0.64, "grad_norm": 3.077538947466888, "learning_rate": 3.024262486807821e-06, "loss": 0.2122, "step": 8931 }, { "epoch": 0.64, "grad_norm": 2.5471652003098915, "learning_rate": 3.0231957344864693e-06, "loss": 0.1345, "step": 8932 }, { "epoch": 0.64, "grad_norm": 2.808378564116779, "learning_rate": 3.022129088806225e-06, "loss": 0.1519, "step": 8933 }, { "epoch": 0.64, "grad_norm": 2.7558699003157736, "learning_rate": 3.0210625498246295e-06, "loss": 0.1568, "step": 8934 }, { "epoch": 0.64, "grad_norm": 2.5515544456742942, "learning_rate": 3.0199961175992187e-06, "loss": 0.1722, "step": 8935 }, { "epoch": 0.64, "grad_norm": 12.025289676491123, "learning_rate": 3.0189297921875226e-06, "loss": 0.724, "step": 8936 }, { "epoch": 0.64, "grad_norm": 13.94624652993344, "learning_rate": 3.0178635736470674e-06, "loss": 1.054, "step": 8937 }, { "epoch": 0.64, "grad_norm": 2.827474540542823, "learning_rate": 3.0167974620353687e-06, "loss": 0.1168, "step": 8938 }, { "epoch": 0.64, "grad_norm": 3.767301218320063, "learning_rate": 3.0157314574099384e-06, "loss": 0.2013, "step": 8939 }, { "epoch": 0.64, "grad_norm": 3.6423464785528235, "learning_rate": 3.014665559828286e-06, "loss": 0.1052, "step": 8940 }, { "epoch": 0.64, "grad_norm": 4.300360351666357, "learning_rate": 3.0135997693479126e-06, "loss": 0.8084, "step": 8941 }, { "epoch": 0.64, "grad_norm": 3.5797644964735507, "learning_rate": 3.0125340860263107e-06, "loss": 0.2079, "step": 8942 }, { "epoch": 0.64, "grad_norm": 4.1623249742158, "learning_rate": 3.011468509920973e-06, "loss": 0.7758, "step": 8943 }, { "epoch": 0.64, "grad_norm": 3.2460899596134705, "learning_rate": 3.0104030410893827e-06, "loss": 0.1872, "step": 8944 }, { "epoch": 0.64, "grad_norm": 3.696486233573506, "learning_rate": 3.0093376795890167e-06, "loss": 0.2023, "step": 8945 }, { "epoch": 0.64, "grad_norm": 24.513287485418125, "learning_rate": 3.0082724254773487e-06, "loss": 0.7496, "step": 8946 }, { "epoch": 0.64, "grad_norm": 4.304501588819714, "learning_rate": 3.0072072788118427e-06, "loss": 0.2578, "step": 8947 }, { "epoch": 0.64, "grad_norm": 11.6206676092139, "learning_rate": 3.0061422396499618e-06, "loss": 0.7734, "step": 8948 }, { "epoch": 0.64, "grad_norm": 2.92453432424145, "learning_rate": 3.0050773080491615e-06, "loss": 0.1266, "step": 8949 }, { "epoch": 0.64, "grad_norm": 2.1003937207732033, "learning_rate": 3.0040124840668876e-06, "loss": 0.0959, "step": 8950 }, { "epoch": 0.64, "grad_norm": 4.04186064881225, "learning_rate": 3.0029477677605846e-06, "loss": 0.8203, "step": 8951 }, { "epoch": 0.64, "grad_norm": 6.947650204890768, "learning_rate": 3.0018831591876907e-06, "loss": 0.2609, "step": 8952 }, { "epoch": 0.64, "grad_norm": 3.0746175481436793, "learning_rate": 3.000818658405637e-06, "loss": 0.1996, "step": 8953 }, { "epoch": 0.64, "grad_norm": 1.8095334504808969, "learning_rate": 2.999754265471849e-06, "loss": 0.1049, "step": 8954 }, { "epoch": 0.64, "grad_norm": 6.4219585718476235, "learning_rate": 2.9986899804437473e-06, "loss": 0.3605, "step": 8955 }, { "epoch": 0.64, "grad_norm": 2.4043049957687708, "learning_rate": 2.9976258033787457e-06, "loss": 0.1124, "step": 8956 }, { "epoch": 0.64, "grad_norm": 19.27115064783564, "learning_rate": 2.9965617343342525e-06, "loss": 0.838, "step": 8957 }, { "epoch": 0.64, "grad_norm": 18.673384368621775, "learning_rate": 2.9954977733676714e-06, "loss": 0.5553, "step": 8958 }, { "epoch": 0.64, "grad_norm": 3.225193552387687, "learning_rate": 2.9944339205363966e-06, "loss": 0.1659, "step": 8959 }, { "epoch": 0.64, "grad_norm": 2.986009861356497, "learning_rate": 2.9933701758978214e-06, "loss": 0.1717, "step": 8960 }, { "epoch": 0.64, "grad_norm": 2.951148108981512, "learning_rate": 2.992306539509331e-06, "loss": 0.1196, "step": 8961 }, { "epoch": 0.64, "grad_norm": 2.6312555442581185, "learning_rate": 2.991243011428301e-06, "loss": 0.1054, "step": 8962 }, { "epoch": 0.64, "grad_norm": 2.7004213852342036, "learning_rate": 2.990179591712107e-06, "loss": 0.1061, "step": 8963 }, { "epoch": 0.64, "grad_norm": 2.1443540859245136, "learning_rate": 2.989116280418117e-06, "loss": 0.0904, "step": 8964 }, { "epoch": 0.64, "grad_norm": 3.0183980524621954, "learning_rate": 2.9880530776036916e-06, "loss": 0.2023, "step": 8965 }, { "epoch": 0.64, "grad_norm": 3.711731674645402, "learning_rate": 2.986989983326188e-06, "loss": 0.1321, "step": 8966 }, { "epoch": 0.64, "grad_norm": 15.566300954819255, "learning_rate": 2.985926997642954e-06, "loss": 0.581, "step": 8967 }, { "epoch": 0.64, "grad_norm": 4.64443069904531, "learning_rate": 2.984864120611335e-06, "loss": 0.2562, "step": 8968 }, { "epoch": 0.64, "grad_norm": 3.7354963355541075, "learning_rate": 2.98380135228867e-06, "loss": 0.1614, "step": 8969 }, { "epoch": 0.64, "grad_norm": 3.2340283469902755, "learning_rate": 2.9827386927322888e-06, "loss": 0.1822, "step": 8970 }, { "epoch": 0.64, "grad_norm": 8.611754190893192, "learning_rate": 2.98167614199952e-06, "loss": 0.3907, "step": 8971 }, { "epoch": 0.64, "grad_norm": 3.01332180217453, "learning_rate": 2.9806137001476845e-06, "loss": 0.224, "step": 8972 }, { "epoch": 0.64, "grad_norm": 4.59101178340665, "learning_rate": 2.9795513672340963e-06, "loss": 0.2052, "step": 8973 }, { "epoch": 0.64, "grad_norm": 4.580881390673392, "learning_rate": 2.9784891433160623e-06, "loss": 0.7291, "step": 8974 }, { "epoch": 0.64, "grad_norm": 16.876863507617593, "learning_rate": 2.9774270284508886e-06, "loss": 1.1449, "step": 8975 }, { "epoch": 0.64, "grad_norm": 18.484525798755133, "learning_rate": 2.9763650226958697e-06, "loss": 1.1188, "step": 8976 }, { "epoch": 0.64, "grad_norm": 3.876846628993028, "learning_rate": 2.975303126108298e-06, "loss": 0.1556, "step": 8977 }, { "epoch": 0.64, "grad_norm": 16.77978257601603, "learning_rate": 2.97424133874546e-06, "loss": 1.1979, "step": 8978 }, { "epoch": 0.64, "grad_norm": 3.610015529168651, "learning_rate": 2.9731796606646325e-06, "loss": 0.1605, "step": 8979 }, { "epoch": 0.64, "grad_norm": 3.889654414870889, "learning_rate": 2.972118091923092e-06, "loss": 0.1669, "step": 8980 }, { "epoch": 0.64, "grad_norm": 3.0960170175823407, "learning_rate": 2.9710566325781026e-06, "loss": 0.1074, "step": 8981 }, { "epoch": 0.64, "grad_norm": 4.8799240074758785, "learning_rate": 2.9699952826869285e-06, "loss": 0.2292, "step": 8982 }, { "epoch": 0.64, "grad_norm": 12.608011442405278, "learning_rate": 2.968934042306826e-06, "loss": 0.8752, "step": 8983 }, { "epoch": 0.64, "grad_norm": 2.9987182825287197, "learning_rate": 2.967872911495042e-06, "loss": 0.2044, "step": 8984 }, { "epoch": 0.64, "grad_norm": 4.211053498891578, "learning_rate": 2.966811890308825e-06, "loss": 0.7804, "step": 8985 }, { "epoch": 0.64, "grad_norm": 10.649057706533638, "learning_rate": 2.9657509788054084e-06, "loss": 0.5351, "step": 8986 }, { "epoch": 0.64, "grad_norm": 4.485947813041525, "learning_rate": 2.9646901770420277e-06, "loss": 0.847, "step": 8987 }, { "epoch": 0.64, "grad_norm": 3.759488043469371, "learning_rate": 2.963629485075907e-06, "loss": 0.2846, "step": 8988 }, { "epoch": 0.64, "grad_norm": 2.930733203412976, "learning_rate": 2.962568902964268e-06, "loss": 0.1984, "step": 8989 }, { "epoch": 0.64, "grad_norm": 3.173427630097427, "learning_rate": 2.961508430764324e-06, "loss": 0.1674, "step": 8990 }, { "epoch": 0.64, "grad_norm": 27.121300939433066, "learning_rate": 2.960448068533284e-06, "loss": 1.09, "step": 8991 }, { "epoch": 0.64, "grad_norm": 11.510237475681505, "learning_rate": 2.9593878163283513e-06, "loss": 0.4772, "step": 8992 }, { "epoch": 0.64, "grad_norm": 2.7179762520224577, "learning_rate": 2.9583276742067213e-06, "loss": 0.2685, "step": 8993 }, { "epoch": 0.64, "grad_norm": 2.795872136456684, "learning_rate": 2.9572676422255854e-06, "loss": 0.1185, "step": 8994 }, { "epoch": 0.65, "grad_norm": 10.175213066261293, "learning_rate": 2.9562077204421273e-06, "loss": 0.6857, "step": 8995 }, { "epoch": 0.65, "grad_norm": 3.647421142844456, "learning_rate": 2.9551479089135268e-06, "loss": 0.1988, "step": 8996 }, { "epoch": 0.65, "grad_norm": 19.71145826617354, "learning_rate": 2.954088207696958e-06, "loss": 0.8148, "step": 8997 }, { "epoch": 0.65, "grad_norm": 21.325758777074554, "learning_rate": 2.953028616849586e-06, "loss": 1.0284, "step": 8998 }, { "epoch": 0.65, "grad_norm": 3.409595609391943, "learning_rate": 2.9519691364285696e-06, "loss": 0.1509, "step": 8999 }, { "epoch": 0.65, "grad_norm": 15.433604570134722, "learning_rate": 2.950909766491067e-06, "loss": 0.5849, "step": 9000 }, { "epoch": 0.65, "grad_norm": 3.262948556777192, "learning_rate": 2.9498505070942274e-06, "loss": 0.2272, "step": 9001 }, { "epoch": 0.65, "grad_norm": 3.298580826556606, "learning_rate": 2.9487913582951912e-06, "loss": 0.1947, "step": 9002 }, { "epoch": 0.65, "grad_norm": 22.322805276127482, "learning_rate": 2.947732320151098e-06, "loss": 1.0405, "step": 9003 }, { "epoch": 0.65, "grad_norm": 3.438339643227646, "learning_rate": 2.9466733927190773e-06, "loss": 0.1561, "step": 9004 }, { "epoch": 0.65, "grad_norm": 4.513471624263991, "learning_rate": 2.945614576056254e-06, "loss": 0.8208, "step": 9005 }, { "epoch": 0.65, "grad_norm": 4.521399603267733, "learning_rate": 2.9445558702197495e-06, "loss": 0.7698, "step": 9006 }, { "epoch": 0.65, "grad_norm": 76.9403219041721, "learning_rate": 2.943497275266675e-06, "loss": 0.5569, "step": 9007 }, { "epoch": 0.65, "grad_norm": 8.380431272203618, "learning_rate": 2.9424387912541385e-06, "loss": 0.4643, "step": 9008 }, { "epoch": 0.65, "grad_norm": 12.89934524106118, "learning_rate": 2.9413804182392413e-06, "loss": 0.9865, "step": 9009 }, { "epoch": 0.65, "grad_norm": 14.193360481034844, "learning_rate": 2.9403221562790787e-06, "loss": 0.8266, "step": 9010 }, { "epoch": 0.65, "grad_norm": 2.5996437950501994, "learning_rate": 2.9392640054307376e-06, "loss": 0.0871, "step": 9011 }, { "epoch": 0.65, "grad_norm": 7.093885622518923, "learning_rate": 2.9382059657513047e-06, "loss": 0.4276, "step": 9012 }, { "epoch": 0.65, "grad_norm": 13.504801263644023, "learning_rate": 2.9371480372978544e-06, "loss": 0.444, "step": 9013 }, { "epoch": 0.65, "grad_norm": 2.6997401216716836, "learning_rate": 2.936090220127459e-06, "loss": 0.1257, "step": 9014 }, { "epoch": 0.65, "grad_norm": 10.718856703288953, "learning_rate": 2.9350325142971847e-06, "loss": 0.5589, "step": 9015 }, { "epoch": 0.65, "grad_norm": 4.812560587725336, "learning_rate": 2.93397491986409e-06, "loss": 0.2495, "step": 9016 }, { "epoch": 0.65, "grad_norm": 2.5311172151097807, "learning_rate": 2.9329174368852283e-06, "loss": 0.1133, "step": 9017 }, { "epoch": 0.65, "grad_norm": 3.119878097370484, "learning_rate": 2.9318600654176454e-06, "loss": 0.204, "step": 9018 }, { "epoch": 0.65, "grad_norm": 2.127985861685959, "learning_rate": 2.9308028055183836e-06, "loss": 0.0752, "step": 9019 }, { "epoch": 0.65, "grad_norm": 21.43191554123789, "learning_rate": 2.9297456572444796e-06, "loss": 1.1742, "step": 9020 }, { "epoch": 0.65, "grad_norm": 4.121244095992815, "learning_rate": 2.928688620652962e-06, "loss": 0.1801, "step": 9021 }, { "epoch": 0.65, "grad_norm": 12.811599534094746, "learning_rate": 2.9276316958008498e-06, "loss": 0.6807, "step": 9022 }, { "epoch": 0.65, "grad_norm": 2.8824326480312608, "learning_rate": 2.9265748827451647e-06, "loss": 0.1223, "step": 9023 }, { "epoch": 0.65, "grad_norm": 3.8943725183418354, "learning_rate": 2.9255181815429168e-06, "loss": 0.2293, "step": 9024 }, { "epoch": 0.65, "grad_norm": 23.818206554517992, "learning_rate": 2.92446159225111e-06, "loss": 0.7044, "step": 9025 }, { "epoch": 0.65, "grad_norm": 3.576991679188169, "learning_rate": 2.9234051149267446e-06, "loss": 0.1017, "step": 9026 }, { "epoch": 0.65, "grad_norm": 3.4361155426237993, "learning_rate": 2.922348749626812e-06, "loss": 0.2227, "step": 9027 }, { "epoch": 0.65, "grad_norm": 11.65867066226637, "learning_rate": 2.9212924964083e-06, "loss": 1.101, "step": 9028 }, { "epoch": 0.65, "grad_norm": 3.6145071127993025, "learning_rate": 2.9202363553281904e-06, "loss": 0.2039, "step": 9029 }, { "epoch": 0.65, "grad_norm": 4.445983164403661, "learning_rate": 2.919180326443456e-06, "loss": 0.8128, "step": 9030 }, { "epoch": 0.65, "grad_norm": 3.473654593399228, "learning_rate": 2.918124409811068e-06, "loss": 0.213, "step": 9031 }, { "epoch": 0.65, "grad_norm": 14.18804318532798, "learning_rate": 2.9170686054879853e-06, "loss": 0.5407, "step": 9032 }, { "epoch": 0.65, "grad_norm": 3.72145229964589, "learning_rate": 2.9160129135311717e-06, "loss": 0.2377, "step": 9033 }, { "epoch": 0.65, "grad_norm": 13.306293379695578, "learning_rate": 2.9149573339975708e-06, "loss": 0.9912, "step": 9034 }, { "epoch": 0.65, "grad_norm": 3.580917089975282, "learning_rate": 2.913901866944127e-06, "loss": 0.15, "step": 9035 }, { "epoch": 0.65, "grad_norm": 18.414810079900096, "learning_rate": 2.912846512427784e-06, "loss": 0.4855, "step": 9036 }, { "epoch": 0.65, "grad_norm": 4.105728685396394, "learning_rate": 2.911791270505471e-06, "loss": 0.7383, "step": 9037 }, { "epoch": 0.65, "grad_norm": 5.266778672971065, "learning_rate": 2.910736141234112e-06, "loss": 0.8173, "step": 9038 }, { "epoch": 0.65, "grad_norm": 3.05938672679265, "learning_rate": 2.9096811246706324e-06, "loss": 0.1484, "step": 9039 }, { "epoch": 0.65, "grad_norm": 3.2654178793055713, "learning_rate": 2.908626220871944e-06, "loss": 0.1837, "step": 9040 }, { "epoch": 0.65, "grad_norm": 3.42110325044552, "learning_rate": 2.907571429894954e-06, "loss": 0.1716, "step": 9041 }, { "epoch": 0.65, "grad_norm": 20.568337642139024, "learning_rate": 2.906516751796564e-06, "loss": 0.949, "step": 9042 }, { "epoch": 0.65, "grad_norm": 3.216506734759576, "learning_rate": 2.9054621866336723e-06, "loss": 0.1389, "step": 9043 }, { "epoch": 0.65, "grad_norm": 27.966855403754696, "learning_rate": 2.9044077344631678e-06, "loss": 0.772, "step": 9044 }, { "epoch": 0.65, "grad_norm": 4.47486974976348, "learning_rate": 2.9033533953419337e-06, "loss": 0.8971, "step": 9045 }, { "epoch": 0.65, "grad_norm": 3.373800535191071, "learning_rate": 2.9022991693268476e-06, "loss": 0.1134, "step": 9046 }, { "epoch": 0.65, "grad_norm": 12.966789214085715, "learning_rate": 2.901245056474779e-06, "loss": 0.5446, "step": 9047 }, { "epoch": 0.65, "grad_norm": 2.95186116392203, "learning_rate": 2.9001910568425974e-06, "loss": 0.1247, "step": 9048 }, { "epoch": 0.65, "grad_norm": 3.377668188138449, "learning_rate": 2.8991371704871607e-06, "loss": 0.1821, "step": 9049 }, { "epoch": 0.65, "grad_norm": 3.0542801634763928, "learning_rate": 2.8980833974653206e-06, "loss": 0.1965, "step": 9050 }, { "epoch": 0.65, "grad_norm": 2.514795147379304, "learning_rate": 2.8970297378339236e-06, "loss": 0.1391, "step": 9051 }, { "epoch": 0.65, "grad_norm": 5.819815270506184, "learning_rate": 2.8959761916498136e-06, "loss": 0.8118, "step": 9052 }, { "epoch": 0.65, "grad_norm": 18.204485885905004, "learning_rate": 2.894922758969824e-06, "loss": 0.5551, "step": 9053 }, { "epoch": 0.65, "grad_norm": 6.186890040940323, "learning_rate": 2.8938694398507816e-06, "loss": 0.1945, "step": 9054 }, { "epoch": 0.65, "grad_norm": 20.66426015504128, "learning_rate": 2.8928162343495124e-06, "loss": 1.0601, "step": 9055 }, { "epoch": 0.65, "grad_norm": 3.8945687151078334, "learning_rate": 2.8917631425228323e-06, "loss": 0.1423, "step": 9056 }, { "epoch": 0.65, "grad_norm": 8.048511711186803, "learning_rate": 2.8907101644275493e-06, "loss": 0.4615, "step": 9057 }, { "epoch": 0.65, "grad_norm": 3.1270761612488953, "learning_rate": 2.88965730012047e-06, "loss": 0.1236, "step": 9058 }, { "epoch": 0.65, "grad_norm": 4.049004310349068, "learning_rate": 2.8886045496583903e-06, "loss": 0.9109, "step": 9059 }, { "epoch": 0.65, "grad_norm": 3.4988763205736206, "learning_rate": 2.8875519130981013e-06, "loss": 0.1722, "step": 9060 }, { "epoch": 0.65, "grad_norm": 16.636339344892292, "learning_rate": 2.886499390496392e-06, "loss": 0.5891, "step": 9061 }, { "epoch": 0.65, "grad_norm": 11.871178612183627, "learning_rate": 2.8854469819100416e-06, "loss": 0.3719, "step": 9062 }, { "epoch": 0.65, "grad_norm": 3.986453190008836, "learning_rate": 2.88439468739582e-06, "loss": 0.2365, "step": 9063 }, { "epoch": 0.65, "grad_norm": 3.9058059058000105, "learning_rate": 2.883342507010499e-06, "loss": 0.2613, "step": 9064 }, { "epoch": 0.65, "grad_norm": 5.170094584582562, "learning_rate": 2.8822904408108375e-06, "loss": 0.3333, "step": 9065 }, { "epoch": 0.65, "grad_norm": 2.3524308138586654, "learning_rate": 2.8812384888535893e-06, "loss": 0.0891, "step": 9066 }, { "epoch": 0.65, "grad_norm": 3.0845044790732987, "learning_rate": 2.8801866511955066e-06, "loss": 0.1383, "step": 9067 }, { "epoch": 0.65, "grad_norm": 4.107362702010508, "learning_rate": 2.87913492789333e-06, "loss": 0.2709, "step": 9068 }, { "epoch": 0.65, "grad_norm": 19.391344441200534, "learning_rate": 2.878083319003796e-06, "loss": 0.6848, "step": 9069 }, { "epoch": 0.65, "grad_norm": 3.332395130249187, "learning_rate": 2.8770318245836353e-06, "loss": 0.165, "step": 9070 }, { "epoch": 0.65, "grad_norm": 3.999862034183467, "learning_rate": 2.8759804446895724e-06, "loss": 0.2033, "step": 9071 }, { "epoch": 0.65, "grad_norm": 9.671308802129893, "learning_rate": 2.874929179378322e-06, "loss": 0.5084, "step": 9072 }, { "epoch": 0.65, "grad_norm": 2.736619441733797, "learning_rate": 2.8738780287066e-06, "loss": 0.1561, "step": 9073 }, { "epoch": 0.65, "grad_norm": 3.3837270288056858, "learning_rate": 2.872826992731111e-06, "loss": 0.2945, "step": 9074 }, { "epoch": 0.65, "grad_norm": 3.3894471174394667, "learning_rate": 2.8717760715085517e-06, "loss": 0.311, "step": 9075 }, { "epoch": 0.65, "grad_norm": 2.125891086974366, "learning_rate": 2.8707252650956202e-06, "loss": 0.0904, "step": 9076 }, { "epoch": 0.65, "grad_norm": 4.97398768237791, "learning_rate": 2.8696745735490004e-06, "loss": 0.963, "step": 9077 }, { "epoch": 0.65, "grad_norm": 4.0278623715039705, "learning_rate": 2.868623996925374e-06, "loss": 0.1961, "step": 9078 }, { "epoch": 0.65, "grad_norm": 2.5093470020929787, "learning_rate": 2.867573535281413e-06, "loss": 0.1836, "step": 9079 }, { "epoch": 0.65, "grad_norm": 38.182853502427484, "learning_rate": 2.86652318867379e-06, "loss": 0.9448, "step": 9080 }, { "epoch": 0.65, "grad_norm": 5.071169374873738, "learning_rate": 2.8654729571591655e-06, "loss": 0.9039, "step": 9081 }, { "epoch": 0.65, "grad_norm": 5.242575295355469, "learning_rate": 2.8644228407941954e-06, "loss": 0.776, "step": 9082 }, { "epoch": 0.65, "grad_norm": 3.2983708381189376, "learning_rate": 2.86337283963553e-06, "loss": 0.1248, "step": 9083 }, { "epoch": 0.65, "grad_norm": 3.723670155912121, "learning_rate": 2.862322953739809e-06, "loss": 0.1555, "step": 9084 }, { "epoch": 0.65, "grad_norm": 4.162366883639351, "learning_rate": 2.8612731831636763e-06, "loss": 0.1506, "step": 9085 }, { "epoch": 0.65, "grad_norm": 4.443682809614598, "learning_rate": 2.8602235279637592e-06, "loss": 0.1518, "step": 9086 }, { "epoch": 0.65, "grad_norm": 9.989999520636076, "learning_rate": 2.859173988196683e-06, "loss": 0.5533, "step": 9087 }, { "epoch": 0.65, "grad_norm": 21.966277442840767, "learning_rate": 2.858124563919066e-06, "loss": 0.272, "step": 9088 }, { "epoch": 0.65, "grad_norm": 6.637030064966366, "learning_rate": 2.857075255187522e-06, "loss": 0.2956, "step": 9089 }, { "epoch": 0.65, "grad_norm": 2.955239396929394, "learning_rate": 2.8560260620586568e-06, "loss": 0.2369, "step": 9090 }, { "epoch": 0.65, "grad_norm": 40.78021434072126, "learning_rate": 2.8549769845890684e-06, "loss": 0.8337, "step": 9091 }, { "epoch": 0.65, "grad_norm": 3.304319255106418, "learning_rate": 2.8539280228353546e-06, "loss": 0.1397, "step": 9092 }, { "epoch": 0.65, "grad_norm": 24.481124353980814, "learning_rate": 2.852879176854102e-06, "loss": 0.702, "step": 9093 }, { "epoch": 0.65, "grad_norm": 13.917261733007289, "learning_rate": 2.851830446701887e-06, "loss": 0.4395, "step": 9094 }, { "epoch": 0.65, "grad_norm": 16.738893813859384, "learning_rate": 2.85078183243529e-06, "loss": 0.9814, "step": 9095 }, { "epoch": 0.65, "grad_norm": 2.8375593040722618, "learning_rate": 2.849733334110878e-06, "loss": 0.1149, "step": 9096 }, { "epoch": 0.65, "grad_norm": 4.6932495069698765, "learning_rate": 2.848684951785211e-06, "loss": 0.6956, "step": 9097 }, { "epoch": 0.65, "grad_norm": 4.104348439476477, "learning_rate": 2.84763668551485e-06, "loss": 0.7895, "step": 9098 }, { "epoch": 0.65, "grad_norm": 3.585498944146723, "learning_rate": 2.8465885353563425e-06, "loss": 0.2119, "step": 9099 }, { "epoch": 0.65, "grad_norm": 4.216288232131194, "learning_rate": 2.8455405013662308e-06, "loss": 0.8425, "step": 9100 }, { "epoch": 0.65, "grad_norm": 7.74577403895497, "learning_rate": 2.844492583601056e-06, "loss": 0.6489, "step": 9101 }, { "epoch": 0.65, "grad_norm": 3.13621784358889, "learning_rate": 2.8434447821173473e-06, "loss": 0.3291, "step": 9102 }, { "epoch": 0.65, "grad_norm": 6.849408489887868, "learning_rate": 2.8423970969716274e-06, "loss": 0.4485, "step": 9103 }, { "epoch": 0.65, "grad_norm": 3.379229831884512, "learning_rate": 2.841349528220419e-06, "loss": 0.1443, "step": 9104 }, { "epoch": 0.65, "grad_norm": 1.9880419158098215, "learning_rate": 2.840302075920235e-06, "loss": 0.0894, "step": 9105 }, { "epoch": 0.65, "grad_norm": 10.583185700198724, "learning_rate": 2.8392547401275744e-06, "loss": 0.5491, "step": 9106 }, { "epoch": 0.65, "grad_norm": 18.856612743249276, "learning_rate": 2.838207520898944e-06, "loss": 0.9268, "step": 9107 }, { "epoch": 0.65, "grad_norm": 23.218761816115812, "learning_rate": 2.837160418290835e-06, "loss": 1.0127, "step": 9108 }, { "epoch": 0.65, "grad_norm": 3.297215402935514, "learning_rate": 2.8361134323597326e-06, "loss": 0.1264, "step": 9109 }, { "epoch": 0.65, "grad_norm": 13.863884213604535, "learning_rate": 2.8350665631621222e-06, "loss": 0.4886, "step": 9110 }, { "epoch": 0.65, "grad_norm": 3.1932222504231667, "learning_rate": 2.834019810754476e-06, "loss": 0.2185, "step": 9111 }, { "epoch": 0.65, "grad_norm": 2.5095314413039005, "learning_rate": 2.83297317519326e-06, "loss": 0.0907, "step": 9112 }, { "epoch": 0.65, "grad_norm": 4.566770798753645, "learning_rate": 2.8319266565349408e-06, "loss": 0.9276, "step": 9113 }, { "epoch": 0.65, "grad_norm": 2.6672321300314445, "learning_rate": 2.8308802548359716e-06, "loss": 0.1232, "step": 9114 }, { "epoch": 0.65, "grad_norm": 28.387651712282636, "learning_rate": 2.829833970152802e-06, "loss": 1.2156, "step": 9115 }, { "epoch": 0.65, "grad_norm": 4.65934720909287, "learning_rate": 2.8287878025418736e-06, "loss": 0.7728, "step": 9116 }, { "epoch": 0.65, "grad_norm": 26.295077751656645, "learning_rate": 2.8277417520596285e-06, "loss": 0.5883, "step": 9117 }, { "epoch": 0.65, "grad_norm": 5.085017468383851, "learning_rate": 2.826695818762489e-06, "loss": 0.8009, "step": 9118 }, { "epoch": 0.65, "grad_norm": 17.94751670300157, "learning_rate": 2.8256500027068867e-06, "loss": 0.9005, "step": 9119 }, { "epoch": 0.65, "grad_norm": 3.407725539881127, "learning_rate": 2.824604303949236e-06, "loss": 0.2132, "step": 9120 }, { "epoch": 0.65, "grad_norm": 3.9310158030776865, "learning_rate": 2.8235587225459487e-06, "loss": 0.2622, "step": 9121 }, { "epoch": 0.65, "grad_norm": 23.699170967736794, "learning_rate": 2.8225132585534277e-06, "loss": 0.9016, "step": 9122 }, { "epoch": 0.65, "grad_norm": 4.723499071340649, "learning_rate": 2.8214679120280763e-06, "loss": 0.8565, "step": 9123 }, { "epoch": 0.65, "grad_norm": 4.515500097938249, "learning_rate": 2.8204226830262855e-06, "loss": 0.8447, "step": 9124 }, { "epoch": 0.65, "grad_norm": 3.195552649569968, "learning_rate": 2.819377571604438e-06, "loss": 0.1532, "step": 9125 }, { "epoch": 0.65, "grad_norm": 3.5063144549078316, "learning_rate": 2.818332577818919e-06, "loss": 0.2835, "step": 9126 }, { "epoch": 0.65, "grad_norm": 4.394531491227969, "learning_rate": 2.817287701726099e-06, "loss": 0.1581, "step": 9127 }, { "epoch": 0.65, "grad_norm": 3.340600845603857, "learning_rate": 2.8162429433823433e-06, "loss": 0.1588, "step": 9128 }, { "epoch": 0.65, "grad_norm": 3.344875748717547, "learning_rate": 2.8151983028440194e-06, "loss": 0.1818, "step": 9129 }, { "epoch": 0.65, "grad_norm": 3.9690504476768744, "learning_rate": 2.8141537801674747e-06, "loss": 0.2039, "step": 9130 }, { "epoch": 0.65, "grad_norm": 5.687937311591359, "learning_rate": 2.8131093754090577e-06, "loss": 0.2337, "step": 9131 }, { "epoch": 0.65, "grad_norm": 8.591619776581123, "learning_rate": 2.8120650886251143e-06, "loss": 0.4102, "step": 9132 }, { "epoch": 0.65, "grad_norm": 3.1378086089615134, "learning_rate": 2.8110209198719773e-06, "loss": 0.2017, "step": 9133 }, { "epoch": 0.65, "grad_norm": 12.607378042573234, "learning_rate": 2.8099768692059747e-06, "loss": 0.8294, "step": 9134 }, { "epoch": 0.66, "grad_norm": 4.136019995154856, "learning_rate": 2.808932936683432e-06, "loss": 0.8085, "step": 9135 }, { "epoch": 0.66, "grad_norm": 9.203781364848819, "learning_rate": 2.807889122360663e-06, "loss": 0.3667, "step": 9136 }, { "epoch": 0.66, "grad_norm": 28.760792829202135, "learning_rate": 2.806845426293977e-06, "loss": 0.5378, "step": 9137 }, { "epoch": 0.66, "grad_norm": 3.633721487334935, "learning_rate": 2.8058018485396797e-06, "loss": 0.1644, "step": 9138 }, { "epoch": 0.66, "grad_norm": 16.772482496169964, "learning_rate": 2.8047583891540685e-06, "loss": 1.2032, "step": 9139 }, { "epoch": 0.66, "grad_norm": 3.3672734539737412, "learning_rate": 2.80371504819343e-06, "loss": 0.1335, "step": 9140 }, { "epoch": 0.66, "grad_norm": 31.087217981688138, "learning_rate": 2.8026718257140556e-06, "loss": 0.7049, "step": 9141 }, { "epoch": 0.66, "grad_norm": 17.120745121557793, "learning_rate": 2.8016287217722165e-06, "loss": 0.7574, "step": 9142 }, { "epoch": 0.66, "grad_norm": 3.5894231626049833, "learning_rate": 2.800585736424185e-06, "loss": 0.2178, "step": 9143 }, { "epoch": 0.66, "grad_norm": 2.873227844084406, "learning_rate": 2.7995428697262295e-06, "loss": 0.2245, "step": 9144 }, { "epoch": 0.66, "grad_norm": 4.029889731843494, "learning_rate": 2.798500121734607e-06, "loss": 0.2191, "step": 9145 }, { "epoch": 0.66, "grad_norm": 21.10373527106122, "learning_rate": 2.7974574925055676e-06, "loss": 0.467, "step": 9146 }, { "epoch": 0.66, "grad_norm": 3.1863137264977084, "learning_rate": 2.796414982095362e-06, "loss": 0.2025, "step": 9147 }, { "epoch": 0.66, "grad_norm": 5.337329600085596, "learning_rate": 2.795372590560227e-06, "loss": 0.7801, "step": 9148 }, { "epoch": 0.66, "grad_norm": 2.4661345488833715, "learning_rate": 2.7943303179563928e-06, "loss": 0.1069, "step": 9149 }, { "epoch": 0.66, "grad_norm": 5.1723073829010815, "learning_rate": 2.793288164340092e-06, "loss": 0.8894, "step": 9150 }, { "epoch": 0.66, "grad_norm": 4.371527064166149, "learning_rate": 2.792246129767542e-06, "loss": 0.6922, "step": 9151 }, { "epoch": 0.66, "grad_norm": 26.73988176521455, "learning_rate": 2.791204214294956e-06, "loss": 0.4868, "step": 9152 }, { "epoch": 0.66, "grad_norm": 2.657957631933697, "learning_rate": 2.7901624179785402e-06, "loss": 0.1308, "step": 9153 }, { "epoch": 0.66, "grad_norm": 2.4490648632591787, "learning_rate": 2.789120740874501e-06, "loss": 0.1201, "step": 9154 }, { "epoch": 0.66, "grad_norm": 4.245755612449312, "learning_rate": 2.7880791830390254e-06, "loss": 0.8506, "step": 9155 }, { "epoch": 0.66, "grad_norm": 6.481436953989683, "learning_rate": 2.787037744528307e-06, "loss": 0.1255, "step": 9156 }, { "epoch": 0.66, "grad_norm": 9.528108411252642, "learning_rate": 2.785996425398526e-06, "loss": 0.6109, "step": 9157 }, { "epoch": 0.66, "grad_norm": 3.2429831092767345, "learning_rate": 2.7849552257058572e-06, "loss": 0.1269, "step": 9158 }, { "epoch": 0.66, "grad_norm": 2.6383416511887767, "learning_rate": 2.7839141455064678e-06, "loss": 0.1253, "step": 9159 }, { "epoch": 0.66, "grad_norm": 3.77236503901608, "learning_rate": 2.7828731848565237e-06, "loss": 0.1743, "step": 9160 }, { "epoch": 0.66, "grad_norm": 4.0805550313958285, "learning_rate": 2.781832343812179e-06, "loss": 0.3541, "step": 9161 }, { "epoch": 0.66, "grad_norm": 13.45697371297717, "learning_rate": 2.7807916224295817e-06, "loss": 0.6459, "step": 9162 }, { "epoch": 0.66, "grad_norm": 2.862878826261619, "learning_rate": 2.779751020764877e-06, "loss": 0.2265, "step": 9163 }, { "epoch": 0.66, "grad_norm": 22.21694741926033, "learning_rate": 2.778710538874202e-06, "loss": 0.8446, "step": 9164 }, { "epoch": 0.66, "grad_norm": 3.4728613282373297, "learning_rate": 2.7776701768136826e-06, "loss": 0.1282, "step": 9165 }, { "epoch": 0.66, "grad_norm": 3.3188018313214473, "learning_rate": 2.7766299346394495e-06, "loss": 0.1131, "step": 9166 }, { "epoch": 0.66, "grad_norm": 3.3931624266714806, "learning_rate": 2.775589812407613e-06, "loss": 0.1279, "step": 9167 }, { "epoch": 0.66, "grad_norm": 4.881757229960341, "learning_rate": 2.774549810174285e-06, "loss": 0.8168, "step": 9168 }, { "epoch": 0.66, "grad_norm": 3.628088030965066, "learning_rate": 2.7735099279955724e-06, "loss": 0.1897, "step": 9169 }, { "epoch": 0.66, "grad_norm": 4.009878868858438, "learning_rate": 2.7724701659275716e-06, "loss": 0.2024, "step": 9170 }, { "epoch": 0.66, "grad_norm": 4.676148982832304, "learning_rate": 2.7714305240263716e-06, "loss": 0.9256, "step": 9171 }, { "epoch": 0.66, "grad_norm": 10.25141884705433, "learning_rate": 2.770391002348062e-06, "loss": 0.5409, "step": 9172 }, { "epoch": 0.66, "grad_norm": 3.317258638260505, "learning_rate": 2.7693516009487174e-06, "loss": 0.2782, "step": 9173 }, { "epoch": 0.66, "grad_norm": 2.5949253965776142, "learning_rate": 2.7683123198844084e-06, "loss": 0.1398, "step": 9174 }, { "epoch": 0.66, "grad_norm": 11.798226668821393, "learning_rate": 2.767273159211205e-06, "loss": 0.2979, "step": 9175 }, { "epoch": 0.66, "grad_norm": 2.798156573600318, "learning_rate": 2.7662341189851626e-06, "loss": 0.1322, "step": 9176 }, { "epoch": 0.66, "grad_norm": 2.4789256896635004, "learning_rate": 2.7651951992623327e-06, "loss": 0.1095, "step": 9177 }, { "epoch": 0.66, "grad_norm": 5.184743872249308, "learning_rate": 2.764156400098767e-06, "loss": 0.906, "step": 9178 }, { "epoch": 0.66, "grad_norm": 3.3877784454854933, "learning_rate": 2.763117721550498e-06, "loss": 0.1463, "step": 9179 }, { "epoch": 0.66, "grad_norm": 8.388899976650846, "learning_rate": 2.762079163673559e-06, "loss": 0.4776, "step": 9180 }, { "epoch": 0.66, "grad_norm": 4.4525367646170615, "learning_rate": 2.7610407265239793e-06, "loss": 1.0244, "step": 9181 }, { "epoch": 0.66, "grad_norm": 13.962151375787032, "learning_rate": 2.7600024101577784e-06, "loss": 1.314, "step": 9182 }, { "epoch": 0.66, "grad_norm": 4.718118725436064, "learning_rate": 2.7589642146309665e-06, "loss": 0.9022, "step": 9183 }, { "epoch": 0.66, "grad_norm": 3.16050959993965, "learning_rate": 2.757926139999554e-06, "loss": 0.1652, "step": 9184 }, { "epoch": 0.66, "grad_norm": 10.822445785581372, "learning_rate": 2.7568881863195397e-06, "loss": 0.484, "step": 9185 }, { "epoch": 0.66, "grad_norm": 4.1990380630290405, "learning_rate": 2.755850353646917e-06, "loss": 0.2892, "step": 9186 }, { "epoch": 0.66, "grad_norm": 2.635126862775609, "learning_rate": 2.754812642037672e-06, "loss": 0.1649, "step": 9187 }, { "epoch": 0.66, "grad_norm": 2.784715154529542, "learning_rate": 2.753775051547788e-06, "loss": 0.1237, "step": 9188 }, { "epoch": 0.66, "grad_norm": 2.8581097160965045, "learning_rate": 2.752737582233237e-06, "loss": 0.1244, "step": 9189 }, { "epoch": 0.66, "grad_norm": 3.055185739843952, "learning_rate": 2.7517002341499876e-06, "loss": 0.1789, "step": 9190 }, { "epoch": 0.66, "grad_norm": 11.26732672437621, "learning_rate": 2.7506630073539996e-06, "loss": 0.4887, "step": 9191 }, { "epoch": 0.66, "grad_norm": 3.3348147609406147, "learning_rate": 2.749625901901227e-06, "loss": 0.2127, "step": 9192 }, { "epoch": 0.66, "grad_norm": 4.453362399675088, "learning_rate": 2.748588917847621e-06, "loss": 0.1919, "step": 9193 }, { "epoch": 0.66, "grad_norm": 3.9566835774333504, "learning_rate": 2.7475520552491196e-06, "loss": 0.1581, "step": 9194 }, { "epoch": 0.66, "grad_norm": 4.7562830298715255, "learning_rate": 2.74651531416166e-06, "loss": 0.9591, "step": 9195 }, { "epoch": 0.66, "grad_norm": 2.6449584089537037, "learning_rate": 2.7454786946411673e-06, "loss": 0.1388, "step": 9196 }, { "epoch": 0.66, "grad_norm": 2.945439055543972, "learning_rate": 2.7444421967435663e-06, "loss": 0.1782, "step": 9197 }, { "epoch": 0.66, "grad_norm": 3.2399120228499227, "learning_rate": 2.7434058205247714e-06, "loss": 0.2068, "step": 9198 }, { "epoch": 0.66, "grad_norm": 10.677057435730632, "learning_rate": 2.7423695660406895e-06, "loss": 0.5524, "step": 9199 }, { "epoch": 0.66, "grad_norm": 5.454308619024404, "learning_rate": 2.7413334333472254e-06, "loss": 0.6992, "step": 9200 }, { "epoch": 0.66, "grad_norm": 4.4955648351754975, "learning_rate": 2.7402974225002733e-06, "loss": 0.9623, "step": 9201 }, { "epoch": 0.66, "grad_norm": 12.27956378368481, "learning_rate": 2.7392615335557217e-06, "loss": 0.4639, "step": 9202 }, { "epoch": 0.66, "grad_norm": 3.6176399996100628, "learning_rate": 2.738225766569452e-06, "loss": 0.2425, "step": 9203 }, { "epoch": 0.66, "grad_norm": 3.03381619093737, "learning_rate": 2.7371901215973423e-06, "loss": 0.2031, "step": 9204 }, { "epoch": 0.66, "grad_norm": 11.076326881417236, "learning_rate": 2.7361545986952578e-06, "loss": 0.5464, "step": 9205 }, { "epoch": 0.66, "grad_norm": 2.904852310890011, "learning_rate": 2.7351191979190663e-06, "loss": 0.1295, "step": 9206 }, { "epoch": 0.66, "grad_norm": 25.79812694599571, "learning_rate": 2.7340839193246195e-06, "loss": 0.822, "step": 9207 }, { "epoch": 0.66, "grad_norm": 3.2499077651911548, "learning_rate": 2.7330487629677673e-06, "loss": 0.1728, "step": 9208 }, { "epoch": 0.66, "grad_norm": 4.078019865363935, "learning_rate": 2.732013728904356e-06, "loss": 0.21, "step": 9209 }, { "epoch": 0.66, "grad_norm": 3.2954324788666405, "learning_rate": 2.730978817190218e-06, "loss": 0.1916, "step": 9210 }, { "epoch": 0.66, "grad_norm": 18.686426961177638, "learning_rate": 2.729944027881183e-06, "loss": 0.8937, "step": 9211 }, { "epoch": 0.66, "grad_norm": 18.909327951245345, "learning_rate": 2.7289093610330764e-06, "loss": 0.6627, "step": 9212 }, { "epoch": 0.66, "grad_norm": 3.3546765578840465, "learning_rate": 2.7278748167017133e-06, "loss": 0.2195, "step": 9213 }, { "epoch": 0.66, "grad_norm": 18.663428405400676, "learning_rate": 2.7268403949429036e-06, "loss": 0.5349, "step": 9214 }, { "epoch": 0.66, "grad_norm": 15.538951821227217, "learning_rate": 2.72580609581245e-06, "loss": 1.0923, "step": 9215 }, { "epoch": 0.66, "grad_norm": 4.2188139519031465, "learning_rate": 2.72477191936615e-06, "loss": 0.6585, "step": 9216 }, { "epoch": 0.66, "grad_norm": 15.377078204400382, "learning_rate": 2.7237378656597903e-06, "loss": 0.3586, "step": 9217 }, { "epoch": 0.66, "grad_norm": 2.4879562952534875, "learning_rate": 2.722703934749158e-06, "loss": 0.1077, "step": 9218 }, { "epoch": 0.66, "grad_norm": 3.1851635654870645, "learning_rate": 2.721670126690029e-06, "loss": 0.1861, "step": 9219 }, { "epoch": 0.66, "grad_norm": 12.523002056714157, "learning_rate": 2.720636441538171e-06, "loss": 0.8611, "step": 9220 }, { "epoch": 0.66, "grad_norm": 3.7796569632882364, "learning_rate": 2.719602879349351e-06, "loss": 0.1036, "step": 9221 }, { "epoch": 0.66, "grad_norm": 23.321211349066896, "learning_rate": 2.7185694401793234e-06, "loss": 0.7528, "step": 9222 }, { "epoch": 0.66, "grad_norm": 3.5572805892700865, "learning_rate": 2.717536124083839e-06, "loss": 0.1251, "step": 9223 }, { "epoch": 0.66, "grad_norm": 17.050627117087082, "learning_rate": 2.716502931118639e-06, "loss": 0.8059, "step": 9224 }, { "epoch": 0.66, "grad_norm": 20.509848036808087, "learning_rate": 2.715469861339465e-06, "loss": 0.7066, "step": 9225 }, { "epoch": 0.66, "grad_norm": 22.191665730926676, "learning_rate": 2.7144369148020437e-06, "loss": 0.7941, "step": 9226 }, { "epoch": 0.66, "grad_norm": 3.9119165563283618, "learning_rate": 2.7134040915621e-06, "loss": 0.2603, "step": 9227 }, { "epoch": 0.66, "grad_norm": 3.0033153201833134, "learning_rate": 2.71237139167535e-06, "loss": 0.1087, "step": 9228 }, { "epoch": 0.66, "grad_norm": 3.5862329660630756, "learning_rate": 2.7113388151975027e-06, "loss": 0.2113, "step": 9229 }, { "epoch": 0.66, "grad_norm": 4.564071550775924, "learning_rate": 2.710306362184265e-06, "loss": 0.8495, "step": 9230 }, { "epoch": 0.66, "grad_norm": 3.045882022963058, "learning_rate": 2.7092740326913325e-06, "loss": 0.1571, "step": 9231 }, { "epoch": 0.66, "grad_norm": 16.622651395028054, "learning_rate": 2.708241826774395e-06, "loss": 0.4944, "step": 9232 }, { "epoch": 0.66, "grad_norm": 3.119892297637865, "learning_rate": 2.707209744489133e-06, "loss": 0.1831, "step": 9233 }, { "epoch": 0.66, "grad_norm": 3.336683109439704, "learning_rate": 2.7061777858912297e-06, "loss": 0.1883, "step": 9234 }, { "epoch": 0.66, "grad_norm": 3.8202883268010273, "learning_rate": 2.705145951036352e-06, "loss": 0.7744, "step": 9235 }, { "epoch": 0.66, "grad_norm": 2.6330756435497213, "learning_rate": 2.704114239980162e-06, "loss": 0.1599, "step": 9236 }, { "epoch": 0.66, "grad_norm": 2.4937501450110164, "learning_rate": 2.7030826527783205e-06, "loss": 0.1289, "step": 9237 }, { "epoch": 0.66, "grad_norm": 4.105466372476865, "learning_rate": 2.7020511894864775e-06, "loss": 0.8303, "step": 9238 }, { "epoch": 0.66, "grad_norm": 8.593601531259678, "learning_rate": 2.7010198501602703e-06, "loss": 0.6002, "step": 9239 }, { "epoch": 0.66, "grad_norm": 2.0534804795475523, "learning_rate": 2.6999886348553415e-06, "loss": 0.0736, "step": 9240 }, { "epoch": 0.66, "grad_norm": 1.5613130695177133, "learning_rate": 2.698957543627321e-06, "loss": 0.0779, "step": 9241 }, { "epoch": 0.66, "grad_norm": 20.65014406586283, "learning_rate": 2.6979265765318286e-06, "loss": 0.9772, "step": 9242 }, { "epoch": 0.66, "grad_norm": 15.260999031051837, "learning_rate": 2.6968957336244857e-06, "loss": 0.6331, "step": 9243 }, { "epoch": 0.66, "grad_norm": 4.855789632743954, "learning_rate": 2.6958650149609e-06, "loss": 0.9447, "step": 9244 }, { "epoch": 0.66, "grad_norm": 5.838121539611944, "learning_rate": 2.6948344205966732e-06, "loss": 0.2709, "step": 9245 }, { "epoch": 0.66, "grad_norm": 26.764747027895968, "learning_rate": 2.693803950587406e-06, "loss": 1.4585, "step": 9246 }, { "epoch": 0.66, "grad_norm": 3.8846105593667826, "learning_rate": 2.692773604988686e-06, "loss": 0.1072, "step": 9247 }, { "epoch": 0.66, "grad_norm": 4.607372701397667, "learning_rate": 2.691743383856095e-06, "loss": 0.3299, "step": 9248 }, { "epoch": 0.66, "grad_norm": 4.311985795623159, "learning_rate": 2.690713287245213e-06, "loss": 0.2305, "step": 9249 }, { "epoch": 0.66, "grad_norm": 11.739790347869317, "learning_rate": 2.6896833152116096e-06, "loss": 0.8507, "step": 9250 }, { "epoch": 0.66, "grad_norm": 2.778646655118126, "learning_rate": 2.6886534678108435e-06, "loss": 0.151, "step": 9251 }, { "epoch": 0.66, "grad_norm": 19.544631076346402, "learning_rate": 2.6876237450984747e-06, "loss": 0.5626, "step": 9252 }, { "epoch": 0.66, "grad_norm": 7.84104944271005, "learning_rate": 2.686594147130053e-06, "loss": 0.6406, "step": 9253 }, { "epoch": 0.66, "grad_norm": 5.64770527621605, "learning_rate": 2.6855646739611175e-06, "loss": 0.356, "step": 9254 }, { "epoch": 0.66, "grad_norm": 2.788109898000323, "learning_rate": 2.6845353256472094e-06, "loss": 0.09, "step": 9255 }, { "epoch": 0.66, "grad_norm": 2.807601774743306, "learning_rate": 2.6835061022438567e-06, "loss": 0.1699, "step": 9256 }, { "epoch": 0.66, "grad_norm": 2.935907236422837, "learning_rate": 2.6824770038065785e-06, "loss": 0.1943, "step": 9257 }, { "epoch": 0.66, "grad_norm": 12.946592432341598, "learning_rate": 2.681448030390896e-06, "loss": 0.5017, "step": 9258 }, { "epoch": 0.66, "grad_norm": 2.5333254430744936, "learning_rate": 2.6804191820523163e-06, "loss": 0.1145, "step": 9259 }, { "epoch": 0.66, "grad_norm": 5.7168013121587125, "learning_rate": 2.6793904588463414e-06, "loss": 0.2858, "step": 9260 }, { "epoch": 0.66, "grad_norm": 12.335852082173346, "learning_rate": 2.6783618608284658e-06, "loss": 0.6444, "step": 9261 }, { "epoch": 0.66, "grad_norm": 10.731790497660844, "learning_rate": 2.677333388054183e-06, "loss": 1.0628, "step": 9262 }, { "epoch": 0.66, "grad_norm": 12.752865708094731, "learning_rate": 2.676305040578968e-06, "loss": 0.6802, "step": 9263 }, { "epoch": 0.66, "grad_norm": 15.08395394240405, "learning_rate": 2.6752768184583033e-06, "loss": 0.6522, "step": 9264 }, { "epoch": 0.66, "grad_norm": 2.836531175711805, "learning_rate": 2.674248721747654e-06, "loss": 0.1727, "step": 9265 }, { "epoch": 0.66, "grad_norm": 3.1762587621119733, "learning_rate": 2.673220750502483e-06, "loss": 0.1362, "step": 9266 }, { "epoch": 0.66, "grad_norm": 2.530270354259472, "learning_rate": 2.6721929047782434e-06, "loss": 0.1523, "step": 9267 }, { "epoch": 0.66, "grad_norm": 4.096838895091768, "learning_rate": 2.6711651846303876e-06, "loss": 0.933, "step": 9268 }, { "epoch": 0.66, "grad_norm": 4.0101909567208684, "learning_rate": 2.6701375901143544e-06, "loss": 0.1469, "step": 9269 }, { "epoch": 0.66, "grad_norm": 4.347577288700491, "learning_rate": 2.6691101212855773e-06, "loss": 0.795, "step": 9270 }, { "epoch": 0.66, "grad_norm": 3.9280459403651977, "learning_rate": 2.6680827781994885e-06, "loss": 0.1783, "step": 9271 }, { "epoch": 0.66, "grad_norm": 21.129643302446954, "learning_rate": 2.6670555609115065e-06, "loss": 0.8886, "step": 9272 }, { "epoch": 0.66, "grad_norm": 16.592866636140513, "learning_rate": 2.666028469477045e-06, "loss": 0.5436, "step": 9273 }, { "epoch": 0.67, "grad_norm": 18.713328173846993, "learning_rate": 2.6650015039515165e-06, "loss": 0.872, "step": 9274 }, { "epoch": 0.67, "grad_norm": 9.568705090875033, "learning_rate": 2.6639746643903164e-06, "loss": 0.4986, "step": 9275 }, { "epoch": 0.67, "grad_norm": 2.4904780700679328, "learning_rate": 2.6629479508488393e-06, "loss": 0.1562, "step": 9276 }, { "epoch": 0.67, "grad_norm": 4.159907103055147, "learning_rate": 2.661921363382476e-06, "loss": 0.6586, "step": 9277 }, { "epoch": 0.67, "grad_norm": 2.5339004939610006, "learning_rate": 2.6608949020466048e-06, "loss": 0.1461, "step": 9278 }, { "epoch": 0.67, "grad_norm": 4.641589918992852, "learning_rate": 2.6598685668965973e-06, "loss": 0.1801, "step": 9279 }, { "epoch": 0.67, "grad_norm": 3.2301168047856867, "learning_rate": 2.658842357987825e-06, "loss": 0.2091, "step": 9280 }, { "epoch": 0.67, "grad_norm": 2.2958506152173546, "learning_rate": 2.657816275375645e-06, "loss": 0.1084, "step": 9281 }, { "epoch": 0.67, "grad_norm": 15.969671357890723, "learning_rate": 2.656790319115411e-06, "loss": 0.9532, "step": 9282 }, { "epoch": 0.67, "grad_norm": 23.731692172332487, "learning_rate": 2.655764489262471e-06, "loss": 1.0947, "step": 9283 }, { "epoch": 0.67, "grad_norm": 3.798245246352172, "learning_rate": 2.654738785872163e-06, "loss": 0.1199, "step": 9284 }, { "epoch": 0.67, "grad_norm": 18.510362809834657, "learning_rate": 2.6537132089998184e-06, "loss": 0.6843, "step": 9285 }, { "epoch": 0.67, "grad_norm": 3.560285817296301, "learning_rate": 2.652687758700769e-06, "loss": 0.1641, "step": 9286 }, { "epoch": 0.67, "grad_norm": 3.2524235496276743, "learning_rate": 2.6516624350303278e-06, "loss": 0.1735, "step": 9287 }, { "epoch": 0.67, "grad_norm": 2.817002979460029, "learning_rate": 2.650637238043807e-06, "loss": 0.1843, "step": 9288 }, { "epoch": 0.67, "grad_norm": 42.58123907172789, "learning_rate": 2.649612167796517e-06, "loss": 0.3799, "step": 9289 }, { "epoch": 0.67, "grad_norm": 4.089696915598682, "learning_rate": 2.6485872243437537e-06, "loss": 0.7887, "step": 9290 }, { "epoch": 0.67, "grad_norm": 11.465656393295067, "learning_rate": 2.6475624077408074e-06, "loss": 0.8094, "step": 9291 }, { "epoch": 0.67, "grad_norm": 4.1437735323560085, "learning_rate": 2.6465377180429664e-06, "loss": 0.7909, "step": 9292 }, { "epoch": 0.67, "grad_norm": 3.157604511939448, "learning_rate": 2.645513155305507e-06, "loss": 0.1649, "step": 9293 }, { "epoch": 0.67, "grad_norm": 3.3726576015919685, "learning_rate": 2.644488719583701e-06, "loss": 0.1272, "step": 9294 }, { "epoch": 0.67, "grad_norm": 3.468783435947657, "learning_rate": 2.64346441093281e-06, "loss": 0.145, "step": 9295 }, { "epoch": 0.67, "grad_norm": 11.036915945678164, "learning_rate": 2.6424402294080963e-06, "loss": 0.6294, "step": 9296 }, { "epoch": 0.67, "grad_norm": 10.727048133728774, "learning_rate": 2.6414161750648088e-06, "loss": 0.6101, "step": 9297 }, { "epoch": 0.67, "grad_norm": 4.309566301057196, "learning_rate": 2.6403922479581905e-06, "loss": 0.7357, "step": 9298 }, { "epoch": 0.67, "grad_norm": 8.259996684638402, "learning_rate": 2.639368448143479e-06, "loss": 0.5282, "step": 9299 }, { "epoch": 0.67, "grad_norm": 15.658887521906685, "learning_rate": 2.638344775675902e-06, "loss": 0.9422, "step": 9300 }, { "epoch": 0.67, "grad_norm": 6.584910741217851, "learning_rate": 2.6373212306106873e-06, "loss": 0.7961, "step": 9301 }, { "epoch": 0.67, "grad_norm": 25.55643594616221, "learning_rate": 2.6362978130030485e-06, "loss": 1.0379, "step": 9302 }, { "epoch": 0.67, "grad_norm": 21.896963146967586, "learning_rate": 2.635274522908196e-06, "loss": 0.7905, "step": 9303 }, { "epoch": 0.67, "grad_norm": 13.00868546057357, "learning_rate": 2.63425136038133e-06, "loss": 0.717, "step": 9304 }, { "epoch": 0.67, "grad_norm": 2.819208282167623, "learning_rate": 2.63322832547765e-06, "loss": 0.171, "step": 9305 }, { "epoch": 0.67, "grad_norm": 2.807354309381124, "learning_rate": 2.6322054182523438e-06, "loss": 0.1322, "step": 9306 }, { "epoch": 0.67, "grad_norm": 2.3481315486045573, "learning_rate": 2.63118263876059e-06, "loss": 0.125, "step": 9307 }, { "epoch": 0.67, "grad_norm": 17.254098891183112, "learning_rate": 2.630159987057569e-06, "loss": 0.5055, "step": 9308 }, { "epoch": 0.67, "grad_norm": 10.391132733114269, "learning_rate": 2.629137463198446e-06, "loss": 0.539, "step": 9309 }, { "epoch": 0.67, "grad_norm": 3.1645937820352668, "learning_rate": 2.6281150672383826e-06, "loss": 0.2087, "step": 9310 }, { "epoch": 0.67, "grad_norm": 22.790127320701405, "learning_rate": 2.627092799232533e-06, "loss": 0.9948, "step": 9311 }, { "epoch": 0.67, "grad_norm": 3.4972472807104023, "learning_rate": 2.6260706592360453e-06, "loss": 0.2493, "step": 9312 }, { "epoch": 0.67, "grad_norm": 4.552982887047514, "learning_rate": 2.625048647304057e-06, "loss": 0.8257, "step": 9313 }, { "epoch": 0.67, "grad_norm": 4.039085447169016, "learning_rate": 2.6240267634917077e-06, "loss": 0.3638, "step": 9314 }, { "epoch": 0.67, "grad_norm": 5.461755717092112, "learning_rate": 2.6230050078541203e-06, "loss": 0.266, "step": 9315 }, { "epoch": 0.67, "grad_norm": 3.1781022473649907, "learning_rate": 2.6219833804464134e-06, "loss": 0.1449, "step": 9316 }, { "epoch": 0.67, "grad_norm": 17.490116267538674, "learning_rate": 2.6209618813237035e-06, "loss": 0.8203, "step": 9317 }, { "epoch": 0.67, "grad_norm": 4.13382919389049, "learning_rate": 2.619940510541096e-06, "loss": 0.2211, "step": 9318 }, { "epoch": 0.67, "grad_norm": 4.263118074915521, "learning_rate": 2.6189192681536857e-06, "loss": 0.1873, "step": 9319 }, { "epoch": 0.67, "grad_norm": 3.7123029108521686, "learning_rate": 2.6178981542165705e-06, "loss": 0.1848, "step": 9320 }, { "epoch": 0.67, "grad_norm": 7.61025966468879, "learning_rate": 2.6168771687848337e-06, "loss": 0.4921, "step": 9321 }, { "epoch": 0.67, "grad_norm": 18.02909148352623, "learning_rate": 2.615856311913553e-06, "loss": 0.3914, "step": 9322 }, { "epoch": 0.67, "grad_norm": 4.154661764979098, "learning_rate": 2.614835583657799e-06, "loss": 0.9392, "step": 9323 }, { "epoch": 0.67, "grad_norm": 3.4997235189033504, "learning_rate": 2.613814984072637e-06, "loss": 0.2549, "step": 9324 }, { "epoch": 0.67, "grad_norm": 3.6990569250872793, "learning_rate": 2.612794513213123e-06, "loss": 0.2311, "step": 9325 }, { "epoch": 0.67, "grad_norm": 4.209128393441639, "learning_rate": 2.6117741711343103e-06, "loss": 0.8182, "step": 9326 }, { "epoch": 0.67, "grad_norm": 2.5243583782893153, "learning_rate": 2.610753957891241e-06, "loss": 0.1791, "step": 9327 }, { "epoch": 0.67, "grad_norm": 21.102026208523387, "learning_rate": 2.609733873538951e-06, "loss": 0.4369, "step": 9328 }, { "epoch": 0.67, "grad_norm": 2.587970744505364, "learning_rate": 2.6087139181324717e-06, "loss": 0.1237, "step": 9329 }, { "epoch": 0.67, "grad_norm": 3.135314594626842, "learning_rate": 2.6076940917268257e-06, "loss": 0.1455, "step": 9330 }, { "epoch": 0.67, "grad_norm": 3.585205844240912, "learning_rate": 2.606674394377028e-06, "loss": 0.2172, "step": 9331 }, { "epoch": 0.67, "grad_norm": 39.12085916750367, "learning_rate": 2.6056548261380856e-06, "loss": 0.4984, "step": 9332 }, { "epoch": 0.67, "grad_norm": 7.02373172994518, "learning_rate": 2.6046353870650036e-06, "loss": 0.4381, "step": 9333 }, { "epoch": 0.67, "grad_norm": 4.289733654140636, "learning_rate": 2.603616077212776e-06, "loss": 0.8748, "step": 9334 }, { "epoch": 0.67, "grad_norm": 23.827061653855026, "learning_rate": 2.6025968966363903e-06, "loss": 0.5218, "step": 9335 }, { "epoch": 0.67, "grad_norm": 8.620120124448492, "learning_rate": 2.6015778453908267e-06, "loss": 0.5167, "step": 9336 }, { "epoch": 0.67, "grad_norm": 3.30029114432116, "learning_rate": 2.600558923531059e-06, "loss": 0.1952, "step": 9337 }, { "epoch": 0.67, "grad_norm": 8.472112312933152, "learning_rate": 2.599540131112056e-06, "loss": 0.6192, "step": 9338 }, { "epoch": 0.67, "grad_norm": 3.1866065481981076, "learning_rate": 2.598521468188777e-06, "loss": 0.1809, "step": 9339 }, { "epoch": 0.67, "grad_norm": 2.0909559062705045, "learning_rate": 2.597502934816175e-06, "loss": 0.1366, "step": 9340 }, { "epoch": 0.67, "grad_norm": 3.67526469979227, "learning_rate": 2.596484531049194e-06, "loss": 0.1406, "step": 9341 }, { "epoch": 0.67, "grad_norm": 2.5055770567519193, "learning_rate": 2.5954662569427756e-06, "loss": 0.1754, "step": 9342 }, { "epoch": 0.67, "grad_norm": 11.201202252124311, "learning_rate": 2.594448112551852e-06, "loss": 0.6793, "step": 9343 }, { "epoch": 0.67, "grad_norm": 3.9200646941878476, "learning_rate": 2.5934300979313443e-06, "loss": 0.8353, "step": 9344 }, { "epoch": 0.67, "grad_norm": 2.6643028167622553, "learning_rate": 2.5924122131361763e-06, "loss": 0.1152, "step": 9345 }, { "epoch": 0.67, "grad_norm": 2.7914421741601294, "learning_rate": 2.591394458221258e-06, "loss": 0.1352, "step": 9346 }, { "epoch": 0.67, "grad_norm": 4.631442068424137, "learning_rate": 2.590376833241487e-06, "loss": 0.673, "step": 9347 }, { "epoch": 0.67, "grad_norm": 13.5593508430383, "learning_rate": 2.589359338251768e-06, "loss": 0.8968, "step": 9348 }, { "epoch": 0.67, "grad_norm": 4.131164848647766, "learning_rate": 2.588341973306987e-06, "loss": 0.9288, "step": 9349 }, { "epoch": 0.67, "grad_norm": 3.0111928443811897, "learning_rate": 2.587324738462026e-06, "loss": 0.184, "step": 9350 }, { "epoch": 0.67, "grad_norm": 4.910883434019324, "learning_rate": 2.5863076337717654e-06, "loss": 0.8221, "step": 9351 }, { "epoch": 0.67, "grad_norm": 2.8922644179405044, "learning_rate": 2.5852906592910707e-06, "loss": 0.1062, "step": 9352 }, { "epoch": 0.67, "grad_norm": 15.499367388500938, "learning_rate": 2.5842738150748044e-06, "loss": 0.4075, "step": 9353 }, { "epoch": 0.67, "grad_norm": 2.419042765479894, "learning_rate": 2.5832571011778227e-06, "loss": 0.1391, "step": 9354 }, { "epoch": 0.67, "grad_norm": 12.849207961800992, "learning_rate": 2.5822405176549724e-06, "loss": 0.8324, "step": 9355 }, { "epoch": 0.67, "grad_norm": 3.1301372870960282, "learning_rate": 2.5812240645610933e-06, "loss": 0.2076, "step": 9356 }, { "epoch": 0.67, "grad_norm": 6.825960574119559, "learning_rate": 2.5802077419510205e-06, "loss": 0.2292, "step": 9357 }, { "epoch": 0.67, "grad_norm": 3.286017565708699, "learning_rate": 2.5791915498795844e-06, "loss": 0.2005, "step": 9358 }, { "epoch": 0.67, "grad_norm": 4.123481426665956, "learning_rate": 2.5781754884015963e-06, "loss": 0.6916, "step": 9359 }, { "epoch": 0.67, "grad_norm": 2.789358630257882, "learning_rate": 2.5771595575718754e-06, "loss": 0.1753, "step": 9360 }, { "epoch": 0.67, "grad_norm": 21.178908590019535, "learning_rate": 2.576143757445226e-06, "loss": 1.0657, "step": 9361 }, { "epoch": 0.67, "grad_norm": 17.876979433997523, "learning_rate": 2.575128088076444e-06, "loss": 0.7672, "step": 9362 }, { "epoch": 0.67, "grad_norm": 14.324270606333142, "learning_rate": 2.5741125495203244e-06, "loss": 0.9593, "step": 9363 }, { "epoch": 0.67, "grad_norm": 3.095311801397357, "learning_rate": 2.57309714183165e-06, "loss": 0.165, "step": 9364 }, { "epoch": 0.67, "grad_norm": 3.1075625527423933, "learning_rate": 2.5720818650651975e-06, "loss": 0.0884, "step": 9365 }, { "epoch": 0.67, "grad_norm": 3.5853603116979302, "learning_rate": 2.57106671927574e-06, "loss": 0.1442, "step": 9366 }, { "epoch": 0.67, "grad_norm": 3.207018575231187, "learning_rate": 2.5700517045180385e-06, "loss": 0.1657, "step": 9367 }, { "epoch": 0.67, "grad_norm": 4.01430655852783, "learning_rate": 2.5690368208468496e-06, "loss": 0.18, "step": 9368 }, { "epoch": 0.67, "grad_norm": 3.972314583797799, "learning_rate": 2.568022068316921e-06, "loss": 0.7736, "step": 9369 }, { "epoch": 0.67, "grad_norm": 3.361343138684298, "learning_rate": 2.567007446982998e-06, "loss": 0.1922, "step": 9370 }, { "epoch": 0.67, "grad_norm": 18.24195280056374, "learning_rate": 2.5659929568998145e-06, "loss": 0.6074, "step": 9371 }, { "epoch": 0.67, "grad_norm": 4.292566954589873, "learning_rate": 2.5649785981220965e-06, "loss": 0.9061, "step": 9372 }, { "epoch": 0.67, "grad_norm": 3.016971615868042, "learning_rate": 2.5639643707045667e-06, "loss": 0.1794, "step": 9373 }, { "epoch": 0.67, "grad_norm": 2.6006933096186358, "learning_rate": 2.5629502747019367e-06, "loss": 0.1168, "step": 9374 }, { "epoch": 0.67, "grad_norm": 8.972690771483066, "learning_rate": 2.5619363101689164e-06, "loss": 0.4016, "step": 9375 }, { "epoch": 0.67, "grad_norm": 2.2657700551385433, "learning_rate": 2.5609224771602036e-06, "loss": 0.0994, "step": 9376 }, { "epoch": 0.67, "grad_norm": 7.407655883125067, "learning_rate": 2.55990877573049e-06, "loss": 0.6238, "step": 9377 }, { "epoch": 0.67, "grad_norm": 4.233666688446255, "learning_rate": 2.5588952059344608e-06, "loss": 0.8703, "step": 9378 }, { "epoch": 0.67, "grad_norm": 9.414123412427662, "learning_rate": 2.5578817678267966e-06, "loss": 0.6912, "step": 9379 }, { "epoch": 0.67, "grad_norm": 2.7761216948504077, "learning_rate": 2.5568684614621673e-06, "loss": 0.14, "step": 9380 }, { "epoch": 0.67, "grad_norm": 3.9399835452861556, "learning_rate": 2.555855286895235e-06, "loss": 0.7548, "step": 9381 }, { "epoch": 0.67, "grad_norm": 20.64023886433469, "learning_rate": 2.55484224418066e-06, "loss": 0.8072, "step": 9382 }, { "epoch": 0.67, "grad_norm": 26.830935452481633, "learning_rate": 2.5538293333730918e-06, "loss": 1.4656, "step": 9383 }, { "epoch": 0.67, "grad_norm": 3.984465667579972, "learning_rate": 2.552816554527169e-06, "loss": 0.1007, "step": 9384 }, { "epoch": 0.67, "grad_norm": 4.1063957971437866, "learning_rate": 2.551803907697531e-06, "loss": 0.137, "step": 9385 }, { "epoch": 0.67, "grad_norm": 8.294123892997888, "learning_rate": 2.5507913929388063e-06, "loss": 0.4381, "step": 9386 }, { "epoch": 0.67, "grad_norm": 3.21416445746745, "learning_rate": 2.549779010305612e-06, "loss": 0.31, "step": 9387 }, { "epoch": 0.67, "grad_norm": 22.462561853195762, "learning_rate": 2.5487667598525686e-06, "loss": 0.5601, "step": 9388 }, { "epoch": 0.67, "grad_norm": 5.084341316187035, "learning_rate": 2.547754641634279e-06, "loss": 0.2803, "step": 9389 }, { "epoch": 0.67, "grad_norm": 2.925634019330194, "learning_rate": 2.5467426557053433e-06, "loss": 0.1391, "step": 9390 }, { "epoch": 0.67, "grad_norm": 10.206710778777706, "learning_rate": 2.5457308021203564e-06, "loss": 0.346, "step": 9391 }, { "epoch": 0.67, "grad_norm": 26.082576228347733, "learning_rate": 2.5447190809339035e-06, "loss": 0.8817, "step": 9392 }, { "epoch": 0.67, "grad_norm": 5.6288741705007626, "learning_rate": 2.54370749220056e-06, "loss": 0.2119, "step": 9393 }, { "epoch": 0.67, "grad_norm": 16.191790868630058, "learning_rate": 2.542696035974902e-06, "loss": 0.852, "step": 9394 }, { "epoch": 0.67, "grad_norm": 3.875920777782579, "learning_rate": 2.5416847123114927e-06, "loss": 0.2978, "step": 9395 }, { "epoch": 0.67, "grad_norm": 4.383696608360766, "learning_rate": 2.5406735212648855e-06, "loss": 0.3016, "step": 9396 }, { "epoch": 0.67, "grad_norm": 7.193063234675008, "learning_rate": 2.5396624628896336e-06, "loss": 0.4991, "step": 9397 }, { "epoch": 0.67, "grad_norm": 2.8891298670670844, "learning_rate": 2.5386515372402793e-06, "loss": 0.1389, "step": 9398 }, { "epoch": 0.67, "grad_norm": 4.258574197630992, "learning_rate": 2.537640744371356e-06, "loss": 0.9695, "step": 9399 }, { "epoch": 0.67, "grad_norm": 2.8188611858476227, "learning_rate": 2.5366300843373955e-06, "loss": 0.1575, "step": 9400 }, { "epoch": 0.67, "grad_norm": 3.0196117972828853, "learning_rate": 2.5356195571929177e-06, "loss": 0.2329, "step": 9401 }, { "epoch": 0.67, "grad_norm": 12.375285994110188, "learning_rate": 2.5346091629924345e-06, "loss": 0.4622, "step": 9402 }, { "epoch": 0.67, "grad_norm": 16.16579721117358, "learning_rate": 2.5335989017904574e-06, "loss": 0.6228, "step": 9403 }, { "epoch": 0.67, "grad_norm": 3.3166484952894044, "learning_rate": 2.5325887736414834e-06, "loss": 0.1323, "step": 9404 }, { "epoch": 0.67, "grad_norm": 3.4294143057675046, "learning_rate": 2.531578778600006e-06, "loss": 0.1423, "step": 9405 }, { "epoch": 0.67, "grad_norm": 4.006563570740644, "learning_rate": 2.5305689167205073e-06, "loss": 0.1844, "step": 9406 }, { "epoch": 0.67, "grad_norm": 41.03970818939971, "learning_rate": 2.529559188057472e-06, "loss": 0.7925, "step": 9407 }, { "epoch": 0.67, "grad_norm": 4.759948761858017, "learning_rate": 2.528549592665364e-06, "loss": 0.8663, "step": 9408 }, { "epoch": 0.67, "grad_norm": 3.6335080489242944, "learning_rate": 2.5275401305986524e-06, "loss": 0.1753, "step": 9409 }, { "epoch": 0.67, "grad_norm": 3.704895924151666, "learning_rate": 2.5265308019117917e-06, "loss": 0.1701, "step": 9410 }, { "epoch": 0.67, "grad_norm": 4.861094174377079, "learning_rate": 2.5255216066592326e-06, "loss": 0.8572, "step": 9411 }, { "epoch": 0.67, "grad_norm": 18.748076686562047, "learning_rate": 2.5245125448954133e-06, "loss": 0.449, "step": 9412 }, { "epoch": 0.67, "grad_norm": 3.805466948582517, "learning_rate": 2.5235036166747745e-06, "loss": 0.2275, "step": 9413 }, { "epoch": 0.68, "grad_norm": 3.37816071304794, "learning_rate": 2.5224948220517415e-06, "loss": 0.18, "step": 9414 }, { "epoch": 0.68, "grad_norm": 4.822103397865944, "learning_rate": 2.521486161080733e-06, "loss": 0.7875, "step": 9415 }, { "epoch": 0.68, "grad_norm": 4.8317342779989785, "learning_rate": 2.520477633816166e-06, "loss": 0.8498, "step": 9416 }, { "epoch": 0.68, "grad_norm": 3.0195261703482617, "learning_rate": 2.5194692403124455e-06, "loss": 0.1632, "step": 9417 }, { "epoch": 0.68, "grad_norm": 2.7715794342981765, "learning_rate": 2.518460980623968e-06, "loss": 0.1101, "step": 9418 }, { "epoch": 0.68, "grad_norm": 2.2035365644256157, "learning_rate": 2.517452854805132e-06, "loss": 0.1043, "step": 9419 }, { "epoch": 0.68, "grad_norm": 4.127640177816875, "learning_rate": 2.516444862910315e-06, "loss": 0.6973, "step": 9420 }, { "epoch": 0.68, "grad_norm": 4.634699291381333, "learning_rate": 2.5154370049938946e-06, "loss": 0.955, "step": 9421 }, { "epoch": 0.68, "grad_norm": 2.6661239234128082, "learning_rate": 2.5144292811102462e-06, "loss": 0.1632, "step": 9422 }, { "epoch": 0.68, "grad_norm": 2.948043148998042, "learning_rate": 2.5134216913137287e-06, "loss": 0.1583, "step": 9423 }, { "epoch": 0.68, "grad_norm": 17.70742153899896, "learning_rate": 2.5124142356586972e-06, "loss": 0.9145, "step": 9424 }, { "epoch": 0.68, "grad_norm": 4.034232641009924, "learning_rate": 2.5114069141995035e-06, "loss": 0.2444, "step": 9425 }, { "epoch": 0.68, "grad_norm": 5.243666335891412, "learning_rate": 2.510399726990487e-06, "loss": 1.1426, "step": 9426 }, { "epoch": 0.68, "grad_norm": 13.362999079184386, "learning_rate": 2.5093926740859797e-06, "loss": 0.7164, "step": 9427 }, { "epoch": 0.68, "grad_norm": 3.3501497398737574, "learning_rate": 2.5083857555403122e-06, "loss": 0.1858, "step": 9428 }, { "epoch": 0.68, "grad_norm": 12.002442892418966, "learning_rate": 2.5073789714078023e-06, "loss": 0.6243, "step": 9429 }, { "epoch": 0.68, "grad_norm": 9.374749933952225, "learning_rate": 2.5063723217427595e-06, "loss": 0.5358, "step": 9430 }, { "epoch": 0.68, "grad_norm": 4.0722210207679925, "learning_rate": 2.5053658065994944e-06, "loss": 1.0044, "step": 9431 }, { "epoch": 0.68, "grad_norm": 4.114357198754406, "learning_rate": 2.5043594260323e-06, "loss": 0.7689, "step": 9432 }, { "epoch": 0.68, "grad_norm": 2.6680874014373583, "learning_rate": 2.5033531800954657e-06, "loss": 0.1266, "step": 9433 }, { "epoch": 0.68, "grad_norm": 3.0100524872031715, "learning_rate": 2.5023470688432795e-06, "loss": 0.1577, "step": 9434 }, { "epoch": 0.68, "grad_norm": 3.321428150852471, "learning_rate": 2.5013410923300143e-06, "loss": 0.234, "step": 9435 }, { "epoch": 0.68, "grad_norm": 4.690176293313395, "learning_rate": 2.5003352506099376e-06, "loss": 0.8646, "step": 9436 }, { "epoch": 0.68, "grad_norm": 3.1452894276884678, "learning_rate": 2.499329543737314e-06, "loss": 0.1833, "step": 9437 }, { "epoch": 0.68, "grad_norm": 3.1981935456484574, "learning_rate": 2.498323971766396e-06, "loss": 0.1819, "step": 9438 }, { "epoch": 0.68, "grad_norm": 4.198235351446625, "learning_rate": 2.4973185347514308e-06, "loss": 0.7719, "step": 9439 }, { "epoch": 0.68, "grad_norm": 23.762120217217092, "learning_rate": 2.4963132327466547e-06, "loss": 0.9791, "step": 9440 }, { "epoch": 0.68, "grad_norm": 3.6347506541924637, "learning_rate": 2.4953080658063055e-06, "loss": 0.2122, "step": 9441 }, { "epoch": 0.68, "grad_norm": 28.732243754246955, "learning_rate": 2.4943030339846054e-06, "loss": 1.2304, "step": 9442 }, { "epoch": 0.68, "grad_norm": 3.563244664286911, "learning_rate": 2.4932981373357713e-06, "loss": 0.1194, "step": 9443 }, { "epoch": 0.68, "grad_norm": 5.223818933460581, "learning_rate": 2.492293375914015e-06, "loss": 0.7089, "step": 9444 }, { "epoch": 0.68, "grad_norm": 12.932123004110819, "learning_rate": 2.4912887497735367e-06, "loss": 0.6456, "step": 9445 }, { "epoch": 0.68, "grad_norm": 3.31228688047071, "learning_rate": 2.490284258968536e-06, "loss": 0.2057, "step": 9446 }, { "epoch": 0.68, "grad_norm": 2.8701658871785085, "learning_rate": 2.4892799035532e-06, "loss": 0.0975, "step": 9447 }, { "epoch": 0.68, "grad_norm": 10.03220257423549, "learning_rate": 2.4882756835817096e-06, "loss": 0.7267, "step": 9448 }, { "epoch": 0.68, "grad_norm": 3.1080054204585217, "learning_rate": 2.4872715991082367e-06, "loss": 0.1113, "step": 9449 }, { "epoch": 0.68, "grad_norm": 4.121590556121575, "learning_rate": 2.486267650186952e-06, "loss": 0.6922, "step": 9450 }, { "epoch": 0.68, "grad_norm": 7.08789763139817, "learning_rate": 2.4852638368720126e-06, "loss": 0.2533, "step": 9451 }, { "epoch": 0.68, "grad_norm": 19.700254659688277, "learning_rate": 2.4842601592175685e-06, "loss": 0.8636, "step": 9452 }, { "epoch": 0.68, "grad_norm": 3.5630472575605454, "learning_rate": 2.4832566172777676e-06, "loss": 0.2286, "step": 9453 }, { "epoch": 0.68, "grad_norm": 17.995106218001112, "learning_rate": 2.4822532111067466e-06, "loss": 0.7557, "step": 9454 }, { "epoch": 0.68, "grad_norm": 3.0695166296435574, "learning_rate": 2.4812499407586336e-06, "loss": 0.1174, "step": 9455 }, { "epoch": 0.68, "grad_norm": 12.699485286161309, "learning_rate": 2.4802468062875534e-06, "loss": 0.8154, "step": 9456 }, { "epoch": 0.68, "grad_norm": 10.813646350941607, "learning_rate": 2.4792438077476193e-06, "loss": 0.544, "step": 9457 }, { "epoch": 0.68, "grad_norm": 2.159902024865778, "learning_rate": 2.478240945192939e-06, "loss": 0.0845, "step": 9458 }, { "epoch": 0.68, "grad_norm": 3.4210056306461984, "learning_rate": 2.4772382186776163e-06, "loss": 0.2478, "step": 9459 }, { "epoch": 0.68, "grad_norm": 4.2675297629484445, "learning_rate": 2.476235628255742e-06, "loss": 0.878, "step": 9460 }, { "epoch": 0.68, "grad_norm": 4.115761251768394, "learning_rate": 2.475233173981401e-06, "loss": 0.7516, "step": 9461 }, { "epoch": 0.68, "grad_norm": 4.44546777299574, "learning_rate": 2.474230855908675e-06, "loss": 0.8586, "step": 9462 }, { "epoch": 0.68, "grad_norm": 4.387791778914951, "learning_rate": 2.4732286740916345e-06, "loss": 1.0399, "step": 9463 }, { "epoch": 0.68, "grad_norm": 2.951470056715702, "learning_rate": 2.47222662858434e-06, "loss": 0.1726, "step": 9464 }, { "epoch": 0.68, "grad_norm": 8.94274349365192, "learning_rate": 2.4712247194408533e-06, "loss": 0.2386, "step": 9465 }, { "epoch": 0.68, "grad_norm": 4.161959061613525, "learning_rate": 2.4702229467152207e-06, "loss": 0.8929, "step": 9466 }, { "epoch": 0.68, "grad_norm": 5.563569104328432, "learning_rate": 2.4692213104614843e-06, "loss": 0.2261, "step": 9467 }, { "epoch": 0.68, "grad_norm": 8.779942245759608, "learning_rate": 2.468219810733679e-06, "loss": 0.3703, "step": 9468 }, { "epoch": 0.68, "grad_norm": 2.1462148960145324, "learning_rate": 2.4672184475858312e-06, "loss": 0.1532, "step": 9469 }, { "epoch": 0.68, "grad_norm": 26.367870410653065, "learning_rate": 2.466217221071959e-06, "loss": 0.9866, "step": 9470 }, { "epoch": 0.68, "grad_norm": 4.39012988250183, "learning_rate": 2.46521613124608e-06, "loss": 0.9754, "step": 9471 }, { "epoch": 0.68, "grad_norm": 3.5861228855861387, "learning_rate": 2.464215178162195e-06, "loss": 0.167, "step": 9472 }, { "epoch": 0.68, "grad_norm": 4.272258696066529, "learning_rate": 2.463214361874301e-06, "loss": 0.3664, "step": 9473 }, { "epoch": 0.68, "grad_norm": 21.322113408034024, "learning_rate": 2.462213682436391e-06, "loss": 0.8241, "step": 9474 }, { "epoch": 0.68, "grad_norm": 4.327367914444516, "learning_rate": 2.461213139902448e-06, "loss": 0.9417, "step": 9475 }, { "epoch": 0.68, "grad_norm": 16.964143794676435, "learning_rate": 2.460212734326446e-06, "loss": 0.7285, "step": 9476 }, { "epoch": 0.68, "grad_norm": 6.338357036836748, "learning_rate": 2.459212465762351e-06, "loss": 0.1454, "step": 9477 }, { "epoch": 0.68, "grad_norm": 12.237009883585529, "learning_rate": 2.4582123342641283e-06, "loss": 0.6257, "step": 9478 }, { "epoch": 0.68, "grad_norm": 3.048877387112114, "learning_rate": 2.4572123398857287e-06, "loss": 0.1852, "step": 9479 }, { "epoch": 0.68, "grad_norm": 3.3347840514133527, "learning_rate": 2.456212482681099e-06, "loss": 0.1705, "step": 9480 }, { "epoch": 0.68, "grad_norm": 22.778737290115387, "learning_rate": 2.4552127627041768e-06, "loss": 0.7817, "step": 9481 }, { "epoch": 0.68, "grad_norm": 3.225391221192769, "learning_rate": 2.4542131800088914e-06, "loss": 0.1365, "step": 9482 }, { "epoch": 0.68, "grad_norm": 5.094690701097532, "learning_rate": 2.453213734649171e-06, "loss": 0.9291, "step": 9483 }, { "epoch": 0.68, "grad_norm": 4.355430695057973, "learning_rate": 2.4522144266789287e-06, "loss": 0.8108, "step": 9484 }, { "epoch": 0.68, "grad_norm": 2.9412267689446057, "learning_rate": 2.451215256152075e-06, "loss": 0.1424, "step": 9485 }, { "epoch": 0.68, "grad_norm": 7.748607759154363, "learning_rate": 2.450216223122508e-06, "loss": 0.4918, "step": 9486 }, { "epoch": 0.68, "grad_norm": 10.396645914736332, "learning_rate": 2.4492173276441267e-06, "loss": 0.5762, "step": 9487 }, { "epoch": 0.68, "grad_norm": 3.1144146267891375, "learning_rate": 2.448218569770815e-06, "loss": 0.2078, "step": 9488 }, { "epoch": 0.68, "grad_norm": 3.5073672709428667, "learning_rate": 2.447219949556451e-06, "loss": 0.1968, "step": 9489 }, { "epoch": 0.68, "grad_norm": 2.6039472841285347, "learning_rate": 2.446221467054909e-06, "loss": 0.1187, "step": 9490 }, { "epoch": 0.68, "grad_norm": 4.065616107845036, "learning_rate": 2.445223122320055e-06, "loss": 0.9049, "step": 9491 }, { "epoch": 0.68, "grad_norm": 3.6985964588584532, "learning_rate": 2.444224915405739e-06, "loss": 0.2431, "step": 9492 }, { "epoch": 0.68, "grad_norm": 32.626117294711754, "learning_rate": 2.4432268463658165e-06, "loss": 0.7694, "step": 9493 }, { "epoch": 0.68, "grad_norm": 4.673439513104212, "learning_rate": 2.4422289152541277e-06, "loss": 0.8226, "step": 9494 }, { "epoch": 0.68, "grad_norm": 21.918432227440395, "learning_rate": 2.4412311221245063e-06, "loss": 1.2142, "step": 9495 }, { "epoch": 0.68, "grad_norm": 15.805371055441132, "learning_rate": 2.440233467030781e-06, "loss": 0.7209, "step": 9496 }, { "epoch": 0.68, "grad_norm": 16.109275694689636, "learning_rate": 2.4392359500267714e-06, "loss": 0.9324, "step": 9497 }, { "epoch": 0.68, "grad_norm": 2.7217041477947763, "learning_rate": 2.4382385711662874e-06, "loss": 0.099, "step": 9498 }, { "epoch": 0.68, "grad_norm": 21.149259564247448, "learning_rate": 2.437241330503138e-06, "loss": 0.804, "step": 9499 }, { "epoch": 0.68, "grad_norm": 14.874364373110302, "learning_rate": 2.4362442280911174e-06, "loss": 0.5223, "step": 9500 }, { "epoch": 0.68, "grad_norm": 4.381185054807476, "learning_rate": 2.435247263984015e-06, "loss": 0.8239, "step": 9501 }, { "epoch": 0.68, "grad_norm": 3.408126501075501, "learning_rate": 2.4342504382356163e-06, "loss": 0.1468, "step": 9502 }, { "epoch": 0.68, "grad_norm": 24.327292212958657, "learning_rate": 2.4332537508996966e-06, "loss": 0.7711, "step": 9503 }, { "epoch": 0.68, "grad_norm": 4.1509174484603175, "learning_rate": 2.432257202030018e-06, "loss": 0.1826, "step": 9504 }, { "epoch": 0.68, "grad_norm": 3.7060968926183837, "learning_rate": 2.431260791680345e-06, "loss": 0.211, "step": 9505 }, { "epoch": 0.68, "grad_norm": 2.7321806999007734, "learning_rate": 2.4302645199044296e-06, "loss": 0.121, "step": 9506 }, { "epoch": 0.68, "grad_norm": 5.077239582375984, "learning_rate": 2.4292683867560142e-06, "loss": 0.8701, "step": 9507 }, { "epoch": 0.68, "grad_norm": 3.5368413223882844, "learning_rate": 2.4282723922888407e-06, "loss": 0.2516, "step": 9508 }, { "epoch": 0.68, "grad_norm": 3.054146876648377, "learning_rate": 2.427276536556637e-06, "loss": 0.1527, "step": 9509 }, { "epoch": 0.68, "grad_norm": 3.280287553598671, "learning_rate": 2.426280819613124e-06, "loss": 0.1372, "step": 9510 }, { "epoch": 0.68, "grad_norm": 3.1312821701619855, "learning_rate": 2.4252852415120204e-06, "loss": 0.1413, "step": 9511 }, { "epoch": 0.68, "grad_norm": 12.039544330214643, "learning_rate": 2.424289802307032e-06, "loss": 0.8546, "step": 9512 }, { "epoch": 0.68, "grad_norm": 21.40074555765129, "learning_rate": 2.42329450205186e-06, "loss": 0.3709, "step": 9513 }, { "epoch": 0.68, "grad_norm": 17.042869404743712, "learning_rate": 2.422299340800194e-06, "loss": 0.5277, "step": 9514 }, { "epoch": 0.68, "grad_norm": 4.030337475262566, "learning_rate": 2.421304318605725e-06, "loss": 0.2026, "step": 9515 }, { "epoch": 0.68, "grad_norm": 2.795856763811079, "learning_rate": 2.4203094355221234e-06, "loss": 0.1488, "step": 9516 }, { "epoch": 0.68, "grad_norm": 2.49214505152415, "learning_rate": 2.4193146916030653e-06, "loss": 0.1321, "step": 9517 }, { "epoch": 0.68, "grad_norm": 21.005000290611772, "learning_rate": 2.4183200869022112e-06, "loss": 0.4487, "step": 9518 }, { "epoch": 0.68, "grad_norm": 3.8929749010970296, "learning_rate": 2.4173256214732155e-06, "loss": 0.7965, "step": 9519 }, { "epoch": 0.68, "grad_norm": 4.134394417205276, "learning_rate": 2.4163312953697254e-06, "loss": 0.8518, "step": 9520 }, { "epoch": 0.68, "grad_norm": 3.132303491869447, "learning_rate": 2.4153371086453835e-06, "loss": 0.2241, "step": 9521 }, { "epoch": 0.68, "grad_norm": 3.214918815736351, "learning_rate": 2.4143430613538215e-06, "loss": 0.1319, "step": 9522 }, { "epoch": 0.68, "grad_norm": 13.076142033585702, "learning_rate": 2.4133491535486624e-06, "loss": 0.6428, "step": 9523 }, { "epoch": 0.68, "grad_norm": 4.502426291983294, "learning_rate": 2.4123553852835267e-06, "loss": 0.7927, "step": 9524 }, { "epoch": 0.68, "grad_norm": 4.170219363342825, "learning_rate": 2.411361756612024e-06, "loss": 0.7514, "step": 9525 }, { "epoch": 0.68, "grad_norm": 2.7267656611644844, "learning_rate": 2.4103682675877536e-06, "loss": 0.1248, "step": 9526 }, { "epoch": 0.68, "grad_norm": 4.330926487485075, "learning_rate": 2.4093749182643167e-06, "loss": 0.8608, "step": 9527 }, { "epoch": 0.68, "grad_norm": 4.081346807718165, "learning_rate": 2.4083817086952953e-06, "loss": 0.1834, "step": 9528 }, { "epoch": 0.68, "grad_norm": 4.927781769664992, "learning_rate": 2.407388638934269e-06, "loss": 1.0832, "step": 9529 }, { "epoch": 0.68, "grad_norm": 12.099694878572041, "learning_rate": 2.4063957090348134e-06, "loss": 0.6655, "step": 9530 }, { "epoch": 0.68, "grad_norm": 3.7844898336634216, "learning_rate": 2.4054029190504925e-06, "loss": 0.3062, "step": 9531 }, { "epoch": 0.68, "grad_norm": 10.247160868241492, "learning_rate": 2.404410269034861e-06, "loss": 0.4648, "step": 9532 }, { "epoch": 0.68, "grad_norm": 5.0465792852282565, "learning_rate": 2.4034177590414727e-06, "loss": 0.9028, "step": 9533 }, { "epoch": 0.68, "grad_norm": 2.9350042326135544, "learning_rate": 2.402425389123867e-06, "loss": 0.1491, "step": 9534 }, { "epoch": 0.68, "grad_norm": 18.77241271761607, "learning_rate": 2.4014331593355776e-06, "loss": 1.1504, "step": 9535 }, { "epoch": 0.68, "grad_norm": 24.43886053999722, "learning_rate": 2.4004410697301358e-06, "loss": 0.7501, "step": 9536 }, { "epoch": 0.68, "grad_norm": 2.6976696381716287, "learning_rate": 2.3994491203610577e-06, "loss": 0.0785, "step": 9537 }, { "epoch": 0.68, "grad_norm": 4.740105502237538, "learning_rate": 2.3984573112818543e-06, "loss": 0.8183, "step": 9538 }, { "epoch": 0.68, "grad_norm": 4.616879269460515, "learning_rate": 2.397465642546035e-06, "loss": 0.8057, "step": 9539 }, { "epoch": 0.68, "grad_norm": 3.617172287782943, "learning_rate": 2.3964741142070917e-06, "loss": 0.1392, "step": 9540 }, { "epoch": 0.68, "grad_norm": 31.395070025471107, "learning_rate": 2.395482726318513e-06, "loss": 0.9829, "step": 9541 }, { "epoch": 0.68, "grad_norm": 4.1991020674538655, "learning_rate": 2.3944914789337847e-06, "loss": 0.2589, "step": 9542 }, { "epoch": 0.68, "grad_norm": 18.201048571723128, "learning_rate": 2.393500372106378e-06, "loss": 1.1072, "step": 9543 }, { "epoch": 0.68, "grad_norm": 19.712427719476043, "learning_rate": 2.392509405889759e-06, "loss": 0.9184, "step": 9544 }, { "epoch": 0.68, "grad_norm": 3.67709406134781, "learning_rate": 2.3915185803373893e-06, "loss": 0.1782, "step": 9545 }, { "epoch": 0.68, "grad_norm": 2.571763891300494, "learning_rate": 2.390527895502718e-06, "loss": 0.0709, "step": 9546 }, { "epoch": 0.68, "grad_norm": 18.252280480169155, "learning_rate": 2.3895373514391906e-06, "loss": 0.9701, "step": 9547 }, { "epoch": 0.68, "grad_norm": 4.81517264545261, "learning_rate": 2.388546948200239e-06, "loss": 0.8424, "step": 9548 }, { "epoch": 0.68, "grad_norm": 4.8085100998301264, "learning_rate": 2.3875566858392975e-06, "loss": 0.7641, "step": 9549 }, { "epoch": 0.68, "grad_norm": 3.7783313902956617, "learning_rate": 2.386566564409783e-06, "loss": 0.2191, "step": 9550 }, { "epoch": 0.68, "grad_norm": 27.492639436179388, "learning_rate": 2.3855765839651106e-06, "loss": 0.5519, "step": 9551 }, { "epoch": 0.68, "grad_norm": 2.963236578524655, "learning_rate": 2.384586744558685e-06, "loss": 0.1006, "step": 9552 }, { "epoch": 0.69, "grad_norm": 2.2132424579557153, "learning_rate": 2.3835970462439036e-06, "loss": 0.0769, "step": 9553 }, { "epoch": 0.69, "grad_norm": 12.773764558299517, "learning_rate": 2.3826074890741592e-06, "loss": 0.3924, "step": 9554 }, { "epoch": 0.69, "grad_norm": 17.62183339459306, "learning_rate": 2.381618073102833e-06, "loss": 0.7312, "step": 9555 }, { "epoch": 0.69, "grad_norm": 5.1592955182642495, "learning_rate": 2.3806287983833017e-06, "loss": 0.9249, "step": 9556 }, { "epoch": 0.69, "grad_norm": 9.049464195103576, "learning_rate": 2.379639664968929e-06, "loss": 0.5464, "step": 9557 }, { "epoch": 0.69, "grad_norm": 3.563107401668519, "learning_rate": 2.3786506729130803e-06, "loss": 0.1254, "step": 9558 }, { "epoch": 0.69, "grad_norm": 3.2688687770519373, "learning_rate": 2.377661822269105e-06, "loss": 0.1498, "step": 9559 }, { "epoch": 0.69, "grad_norm": 17.897371470832336, "learning_rate": 2.3766731130903465e-06, "loss": 0.8742, "step": 9560 }, { "epoch": 0.69, "grad_norm": 2.9225501482134812, "learning_rate": 2.375684545430146e-06, "loss": 0.1132, "step": 9561 }, { "epoch": 0.69, "grad_norm": 3.296631477970186, "learning_rate": 2.3746961193418306e-06, "loss": 0.1222, "step": 9562 }, { "epoch": 0.69, "grad_norm": 18.87990269869585, "learning_rate": 2.373707834878722e-06, "loss": 0.7098, "step": 9563 }, { "epoch": 0.69, "grad_norm": 3.0763457327881762, "learning_rate": 2.372719692094134e-06, "loss": 0.1056, "step": 9564 }, { "epoch": 0.69, "grad_norm": 3.9375609305531447, "learning_rate": 2.371731691041375e-06, "loss": 0.2087, "step": 9565 }, { "epoch": 0.69, "grad_norm": 4.2993470907513185, "learning_rate": 2.3707438317737403e-06, "loss": 0.7812, "step": 9566 }, { "epoch": 0.69, "grad_norm": 2.7552599499655117, "learning_rate": 2.369756114344526e-06, "loss": 0.1152, "step": 9567 }, { "epoch": 0.69, "grad_norm": 14.080047727513715, "learning_rate": 2.3687685388070124e-06, "loss": 0.8508, "step": 9568 }, { "epoch": 0.69, "grad_norm": 11.527235135942233, "learning_rate": 2.367781105214475e-06, "loss": 0.4619, "step": 9569 }, { "epoch": 0.69, "grad_norm": 3.375637747485289, "learning_rate": 2.3667938136201855e-06, "loss": 0.1667, "step": 9570 }, { "epoch": 0.69, "grad_norm": 12.810470611936026, "learning_rate": 2.3658066640774026e-06, "loss": 0.4289, "step": 9571 }, { "epoch": 0.69, "grad_norm": 3.4324989889619615, "learning_rate": 2.364819656639377e-06, "loss": 0.2844, "step": 9572 }, { "epoch": 0.69, "grad_norm": 15.623034137320548, "learning_rate": 2.3638327913593577e-06, "loss": 0.9652, "step": 9573 }, { "epoch": 0.69, "grad_norm": 4.24612900534094, "learning_rate": 2.3628460682905817e-06, "loss": 0.8216, "step": 9574 }, { "epoch": 0.69, "grad_norm": 2.794743823404047, "learning_rate": 2.3618594874862777e-06, "loss": 0.0966, "step": 9575 }, { "epoch": 0.69, "grad_norm": 2.820143389993457, "learning_rate": 2.360873048999669e-06, "loss": 0.1484, "step": 9576 }, { "epoch": 0.69, "grad_norm": 2.7178476319056073, "learning_rate": 2.3598867528839694e-06, "loss": 0.1004, "step": 9577 }, { "epoch": 0.69, "grad_norm": 4.628465557754426, "learning_rate": 2.3589005991923838e-06, "loss": 0.2571, "step": 9578 }, { "epoch": 0.69, "grad_norm": 6.490487384855726, "learning_rate": 2.3579145879781155e-06, "loss": 0.6865, "step": 9579 }, { "epoch": 0.69, "grad_norm": 3.2845811783969467, "learning_rate": 2.3569287192943553e-06, "loss": 0.1601, "step": 9580 }, { "epoch": 0.69, "grad_norm": 9.931943488813422, "learning_rate": 2.3559429931942833e-06, "loss": 0.5361, "step": 9581 }, { "epoch": 0.69, "grad_norm": 9.243151179712221, "learning_rate": 2.354957409731081e-06, "loss": 0.6263, "step": 9582 }, { "epoch": 0.69, "grad_norm": 3.9305642263444556, "learning_rate": 2.3539719689579144e-06, "loss": 0.232, "step": 9583 }, { "epoch": 0.69, "grad_norm": 3.308545386976956, "learning_rate": 2.3529866709279435e-06, "loss": 0.182, "step": 9584 }, { "epoch": 0.69, "grad_norm": 4.739811631532437, "learning_rate": 2.3520015156943216e-06, "loss": 0.7903, "step": 9585 }, { "epoch": 0.69, "grad_norm": 2.569768123105129, "learning_rate": 2.3510165033101956e-06, "loss": 0.158, "step": 9586 }, { "epoch": 0.69, "grad_norm": 4.597617377835982, "learning_rate": 2.3500316338287025e-06, "loss": 0.9458, "step": 9587 }, { "epoch": 0.69, "grad_norm": 2.507178275099004, "learning_rate": 2.3490469073029726e-06, "loss": 0.1368, "step": 9588 }, { "epoch": 0.69, "grad_norm": 17.555829345311242, "learning_rate": 2.3480623237861273e-06, "loss": 0.8636, "step": 9589 }, { "epoch": 0.69, "grad_norm": 21.97978730650181, "learning_rate": 2.3470778833312795e-06, "loss": 0.6166, "step": 9590 }, { "epoch": 0.69, "grad_norm": 4.943487664739519, "learning_rate": 2.34609358599154e-06, "loss": 0.1549, "step": 9591 }, { "epoch": 0.69, "grad_norm": 24.65660020472393, "learning_rate": 2.3451094318200056e-06, "loss": 0.8945, "step": 9592 }, { "epoch": 0.69, "grad_norm": 3.474441761068235, "learning_rate": 2.344125420869769e-06, "loss": 0.1723, "step": 9593 }, { "epoch": 0.69, "grad_norm": 4.151332592370387, "learning_rate": 2.34314155319391e-06, "loss": 0.1512, "step": 9594 }, { "epoch": 0.69, "grad_norm": 4.997731389455298, "learning_rate": 2.34215782884551e-06, "loss": 0.201, "step": 9595 }, { "epoch": 0.69, "grad_norm": 4.122508072154237, "learning_rate": 2.3411742478776345e-06, "loss": 0.2316, "step": 9596 }, { "epoch": 0.69, "grad_norm": 6.052124291038023, "learning_rate": 2.3401908103433422e-06, "loss": 0.7629, "step": 9597 }, { "epoch": 0.69, "grad_norm": 3.035710471802146, "learning_rate": 2.3392075162956896e-06, "loss": 0.1049, "step": 9598 }, { "epoch": 0.69, "grad_norm": 11.360071629768301, "learning_rate": 2.33822436578772e-06, "loss": 0.559, "step": 9599 }, { "epoch": 0.69, "grad_norm": 22.453966644514594, "learning_rate": 2.3372413588724703e-06, "loss": 0.9738, "step": 9600 }, { "epoch": 0.69, "grad_norm": 11.039876552299843, "learning_rate": 2.3362584956029704e-06, "loss": 0.3142, "step": 9601 }, { "epoch": 0.69, "grad_norm": 5.574145044853343, "learning_rate": 2.3352757760322416e-06, "loss": 0.3839, "step": 9602 }, { "epoch": 0.69, "grad_norm": 2.301575664803555, "learning_rate": 2.334293200213296e-06, "loss": 0.163, "step": 9603 }, { "epoch": 0.69, "grad_norm": 2.9724266156670254, "learning_rate": 2.3333107681991445e-06, "loss": 0.1554, "step": 9604 }, { "epoch": 0.69, "grad_norm": 5.42602321998773, "learning_rate": 2.3323284800427833e-06, "loss": 0.3561, "step": 9605 }, { "epoch": 0.69, "grad_norm": 10.55936582521585, "learning_rate": 2.3313463357972004e-06, "loss": 0.3025, "step": 9606 }, { "epoch": 0.69, "grad_norm": 4.844795406769601, "learning_rate": 2.330364335515383e-06, "loss": 1.0203, "step": 9607 }, { "epoch": 0.69, "grad_norm": 7.664446884216308, "learning_rate": 2.3293824792503056e-06, "loss": 0.5029, "step": 9608 }, { "epoch": 0.69, "grad_norm": 3.874992561573831, "learning_rate": 2.3284007670549317e-06, "loss": 0.7631, "step": 9609 }, { "epoch": 0.69, "grad_norm": 2.739400648736866, "learning_rate": 2.3274191989822265e-06, "loss": 0.1332, "step": 9610 }, { "epoch": 0.69, "grad_norm": 17.824600600701025, "learning_rate": 2.3264377750851385e-06, "loss": 0.7684, "step": 9611 }, { "epoch": 0.69, "grad_norm": 4.685789307521357, "learning_rate": 2.325456495416613e-06, "loss": 0.9366, "step": 9612 }, { "epoch": 0.69, "grad_norm": 4.594523631972976, "learning_rate": 2.324475360029586e-06, "loss": 0.8907, "step": 9613 }, { "epoch": 0.69, "grad_norm": 3.350505912294641, "learning_rate": 2.3234943689769858e-06, "loss": 0.1972, "step": 9614 }, { "epoch": 0.69, "grad_norm": 3.709790643128217, "learning_rate": 2.3225135223117313e-06, "loss": 0.1898, "step": 9615 }, { "epoch": 0.69, "grad_norm": 3.2957813311006343, "learning_rate": 2.3215328200867393e-06, "loss": 0.1837, "step": 9616 }, { "epoch": 0.69, "grad_norm": 4.229005426119518, "learning_rate": 2.3205522623549133e-06, "loss": 0.8544, "step": 9617 }, { "epoch": 0.69, "grad_norm": 3.3738281617152275, "learning_rate": 2.3195718491691483e-06, "loss": 0.1471, "step": 9618 }, { "epoch": 0.69, "grad_norm": 25.674871558919587, "learning_rate": 2.3185915805823378e-06, "loss": 0.8698, "step": 9619 }, { "epoch": 0.69, "grad_norm": 3.6172391580544274, "learning_rate": 2.3176114566473624e-06, "loss": 0.2046, "step": 9620 }, { "epoch": 0.69, "grad_norm": 3.9654904901819212, "learning_rate": 2.3166314774170952e-06, "loss": 0.1982, "step": 9621 }, { "epoch": 0.69, "grad_norm": 9.108962426186944, "learning_rate": 2.315651642944401e-06, "loss": 0.7391, "step": 9622 }, { "epoch": 0.69, "grad_norm": 2.5282913841863803, "learning_rate": 2.314671953282141e-06, "loss": 0.1028, "step": 9623 }, { "epoch": 0.69, "grad_norm": 25.91594791966956, "learning_rate": 2.3136924084831653e-06, "loss": 1.0679, "step": 9624 }, { "epoch": 0.69, "grad_norm": 4.462269992902622, "learning_rate": 2.3127130086003156e-06, "loss": 0.1622, "step": 9625 }, { "epoch": 0.69, "grad_norm": 3.790946633350719, "learning_rate": 2.3117337536864276e-06, "loss": 0.1573, "step": 9626 }, { "epoch": 0.69, "grad_norm": 5.43832281371441, "learning_rate": 2.310754643794326e-06, "loss": 0.8176, "step": 9627 }, { "epoch": 0.69, "grad_norm": 3.903943994931677, "learning_rate": 2.309775678976834e-06, "loss": 0.2039, "step": 9628 }, { "epoch": 0.69, "grad_norm": 4.367713291218591, "learning_rate": 2.3087968592867604e-06, "loss": 0.1922, "step": 9629 }, { "epoch": 0.69, "grad_norm": 4.73988827342186, "learning_rate": 2.3078181847769097e-06, "loss": 0.1737, "step": 9630 }, { "epoch": 0.69, "grad_norm": 9.361185136956383, "learning_rate": 2.306839655500076e-06, "loss": 0.5071, "step": 9631 }, { "epoch": 0.69, "grad_norm": 4.6798624596871266, "learning_rate": 2.3058612715090507e-06, "loss": 1.0252, "step": 9632 }, { "epoch": 0.69, "grad_norm": 3.288157179691327, "learning_rate": 2.3048830328566112e-06, "loss": 0.1628, "step": 9633 }, { "epoch": 0.69, "grad_norm": 3.0146757595687763, "learning_rate": 2.3039049395955293e-06, "loss": 0.1289, "step": 9634 }, { "epoch": 0.69, "grad_norm": 2.5809572819554543, "learning_rate": 2.3029269917785725e-06, "loss": 0.1603, "step": 9635 }, { "epoch": 0.69, "grad_norm": 13.494468639950984, "learning_rate": 2.301949189458497e-06, "loss": 0.733, "step": 9636 }, { "epoch": 0.69, "grad_norm": 10.440977341348749, "learning_rate": 2.3009715326880467e-06, "loss": 0.4134, "step": 9637 }, { "epoch": 0.69, "grad_norm": 4.289627294801096, "learning_rate": 2.2999940215199675e-06, "loss": 0.7288, "step": 9638 }, { "epoch": 0.69, "grad_norm": 3.5582263207526963, "learning_rate": 2.2990166560069903e-06, "loss": 0.1725, "step": 9639 }, { "epoch": 0.69, "grad_norm": 2.7792903740457646, "learning_rate": 2.2980394362018392e-06, "loss": 0.1541, "step": 9640 }, { "epoch": 0.69, "grad_norm": 3.4044173884744398, "learning_rate": 2.297062362157235e-06, "loss": 0.2464, "step": 9641 }, { "epoch": 0.69, "grad_norm": 9.206345501489155, "learning_rate": 2.2960854339258853e-06, "loss": 0.3977, "step": 9642 }, { "epoch": 0.69, "grad_norm": 8.746403019728549, "learning_rate": 2.2951086515604897e-06, "loss": 0.5791, "step": 9643 }, { "epoch": 0.69, "grad_norm": 5.047718397468638, "learning_rate": 2.2941320151137455e-06, "loss": 1.1681, "step": 9644 }, { "epoch": 0.69, "grad_norm": 4.326021616854802, "learning_rate": 2.293155524638337e-06, "loss": 0.2621, "step": 9645 }, { "epoch": 0.69, "grad_norm": 24.025097209228413, "learning_rate": 2.29217918018694e-06, "loss": 0.4403, "step": 9646 }, { "epoch": 0.69, "grad_norm": 9.891860547147921, "learning_rate": 2.2912029818122282e-06, "loss": 0.4278, "step": 9647 }, { "epoch": 0.69, "grad_norm": 2.678786689279197, "learning_rate": 2.2902269295668644e-06, "loss": 0.0956, "step": 9648 }, { "epoch": 0.69, "grad_norm": 3.172508923396076, "learning_rate": 2.2892510235034964e-06, "loss": 0.1663, "step": 9649 }, { "epoch": 0.69, "grad_norm": 28.442924629609262, "learning_rate": 2.2882752636747773e-06, "loss": 0.8862, "step": 9650 }, { "epoch": 0.69, "grad_norm": 3.5640880177634626, "learning_rate": 2.287299650133343e-06, "loss": 0.1858, "step": 9651 }, { "epoch": 0.69, "grad_norm": 3.9435404910654355, "learning_rate": 2.2863241829318234e-06, "loss": 0.2017, "step": 9652 }, { "epoch": 0.69, "grad_norm": 4.123820079883887, "learning_rate": 2.285348862122844e-06, "loss": 0.1878, "step": 9653 }, { "epoch": 0.69, "grad_norm": 34.17890351728108, "learning_rate": 2.2843736877590174e-06, "loss": 0.7677, "step": 9654 }, { "epoch": 0.69, "grad_norm": 6.932826134635222, "learning_rate": 2.2833986598929493e-06, "loss": 0.2071, "step": 9655 }, { "epoch": 0.69, "grad_norm": 2.8509424041289533, "learning_rate": 2.2824237785772424e-06, "loss": 0.2165, "step": 9656 }, { "epoch": 0.69, "grad_norm": 4.149382834907604, "learning_rate": 2.281449043864486e-06, "loss": 0.8073, "step": 9657 }, { "epoch": 0.69, "grad_norm": 17.528606838786406, "learning_rate": 2.2804744558072637e-06, "loss": 0.6491, "step": 9658 }, { "epoch": 0.69, "grad_norm": 30.50611225152073, "learning_rate": 2.2795000144581485e-06, "loss": 0.5266, "step": 9659 }, { "epoch": 0.69, "grad_norm": 3.5897212396795566, "learning_rate": 2.278525719869714e-06, "loss": 0.1789, "step": 9660 }, { "epoch": 0.69, "grad_norm": 13.103721161003913, "learning_rate": 2.277551572094512e-06, "loss": 0.6765, "step": 9661 }, { "epoch": 0.69, "grad_norm": 27.326506833579227, "learning_rate": 2.2765775711850983e-06, "loss": 0.717, "step": 9662 }, { "epoch": 0.69, "grad_norm": 9.731231274349703, "learning_rate": 2.2756037171940172e-06, "loss": 0.4071, "step": 9663 }, { "epoch": 0.69, "grad_norm": 7.088686073426599, "learning_rate": 2.2746300101738027e-06, "loss": 0.533, "step": 9664 }, { "epoch": 0.69, "grad_norm": 3.196813712512601, "learning_rate": 2.273656450176981e-06, "loss": 0.2036, "step": 9665 }, { "epoch": 0.69, "grad_norm": 28.356606415009633, "learning_rate": 2.272683037256076e-06, "loss": 1.2279, "step": 9666 }, { "epoch": 0.69, "grad_norm": 16.722429868366962, "learning_rate": 2.2717097714635983e-06, "loss": 0.9095, "step": 9667 }, { "epoch": 0.69, "grad_norm": 10.295003754621849, "learning_rate": 2.2707366528520488e-06, "loss": 0.4911, "step": 9668 }, { "epoch": 0.69, "grad_norm": 4.6562418024973, "learning_rate": 2.2697636814739278e-06, "loss": 1.0126, "step": 9669 }, { "epoch": 0.69, "grad_norm": 2.6222270421575407, "learning_rate": 2.2687908573817217e-06, "loss": 0.2008, "step": 9670 }, { "epoch": 0.69, "grad_norm": 3.3901682240217643, "learning_rate": 2.2678181806279093e-06, "loss": 0.2138, "step": 9671 }, { "epoch": 0.69, "grad_norm": 13.9306938515437, "learning_rate": 2.2668456512649677e-06, "loss": 0.8822, "step": 9672 }, { "epoch": 0.69, "grad_norm": 3.5464550684343616, "learning_rate": 2.2658732693453558e-06, "loss": 0.2052, "step": 9673 }, { "epoch": 0.69, "grad_norm": 20.371273310111626, "learning_rate": 2.2649010349215293e-06, "loss": 0.9974, "step": 9674 }, { "epoch": 0.69, "grad_norm": 9.80965285344417, "learning_rate": 2.263928948045942e-06, "loss": 0.3856, "step": 9675 }, { "epoch": 0.69, "grad_norm": 15.072863419009778, "learning_rate": 2.2629570087710305e-06, "loss": 1.1395, "step": 9676 }, { "epoch": 0.69, "grad_norm": 4.10591280085538, "learning_rate": 2.2619852171492266e-06, "loss": 0.2343, "step": 9677 }, { "epoch": 0.69, "grad_norm": 7.157742328523476, "learning_rate": 2.261013573232958e-06, "loss": 0.4364, "step": 9678 }, { "epoch": 0.69, "grad_norm": 43.61869085609793, "learning_rate": 2.2600420770746397e-06, "loss": 0.6437, "step": 9679 }, { "epoch": 0.69, "grad_norm": 4.4444032264644004, "learning_rate": 2.2590707287266783e-06, "loss": 0.7891, "step": 9680 }, { "epoch": 0.69, "grad_norm": 4.773067293029106, "learning_rate": 2.2580995282414777e-06, "loss": 0.8972, "step": 9681 }, { "epoch": 0.69, "grad_norm": 3.5825666930340643, "learning_rate": 2.257128475671429e-06, "loss": 0.153, "step": 9682 }, { "epoch": 0.69, "grad_norm": 15.426004625839687, "learning_rate": 2.256157571068915e-06, "loss": 1.0544, "step": 9683 }, { "epoch": 0.69, "grad_norm": 3.667682726283871, "learning_rate": 2.255186814486317e-06, "loss": 0.1348, "step": 9684 }, { "epoch": 0.69, "grad_norm": 15.283819668493866, "learning_rate": 2.254216205976e-06, "loss": 0.4083, "step": 9685 }, { "epoch": 0.69, "grad_norm": 21.641984040525017, "learning_rate": 2.2532457455903224e-06, "loss": 0.7461, "step": 9686 }, { "epoch": 0.69, "grad_norm": 17.525012757645328, "learning_rate": 2.2522754333816423e-06, "loss": 1.0757, "step": 9687 }, { "epoch": 0.69, "grad_norm": 7.458194584627364, "learning_rate": 2.251305269402301e-06, "loss": 0.3192, "step": 9688 }, { "epoch": 0.69, "grad_norm": 3.58482463821585, "learning_rate": 2.2503352537046354e-06, "loss": 0.2656, "step": 9689 }, { "epoch": 0.69, "grad_norm": 2.781676543478426, "learning_rate": 2.249365386340976e-06, "loss": 0.1359, "step": 9690 }, { "epoch": 0.69, "grad_norm": 19.704783304397278, "learning_rate": 2.2483956673636427e-06, "loss": 0.872, "step": 9691 }, { "epoch": 0.69, "grad_norm": 7.783015422864076, "learning_rate": 2.247426096824947e-06, "loss": 0.3383, "step": 9692 }, { "epoch": 0.7, "grad_norm": 2.760258892784224, "learning_rate": 2.246456674777193e-06, "loss": 0.1816, "step": 9693 }, { "epoch": 0.7, "grad_norm": 14.810121099724045, "learning_rate": 2.2454874012726805e-06, "loss": 0.9569, "step": 9694 }, { "epoch": 0.7, "grad_norm": 4.206761917433484, "learning_rate": 2.2445182763636965e-06, "loss": 0.6967, "step": 9695 }, { "epoch": 0.7, "grad_norm": 4.065379478861302, "learning_rate": 2.243549300102521e-06, "loss": 0.6622, "step": 9696 }, { "epoch": 0.7, "grad_norm": 16.528004798037642, "learning_rate": 2.2425804725414267e-06, "loss": 0.6191, "step": 9697 }, { "epoch": 0.7, "grad_norm": 3.695232567312929, "learning_rate": 2.2416117937326777e-06, "loss": 0.1565, "step": 9698 }, { "epoch": 0.7, "grad_norm": 3.35216639370303, "learning_rate": 2.2406432637285323e-06, "loss": 0.1488, "step": 9699 }, { "epoch": 0.7, "grad_norm": 4.739421078491588, "learning_rate": 2.2396748825812387e-06, "loss": 0.1899, "step": 9700 }, { "epoch": 0.7, "grad_norm": 14.962271188497196, "learning_rate": 2.2387066503430365e-06, "loss": 0.6659, "step": 9701 }, { "epoch": 0.7, "grad_norm": 3.102613559820934, "learning_rate": 2.2377385670661567e-06, "loss": 0.1447, "step": 9702 }, { "epoch": 0.7, "grad_norm": 4.296945878784298, "learning_rate": 2.2367706328028273e-06, "loss": 0.9286, "step": 9703 }, { "epoch": 0.7, "grad_norm": 9.498998031027716, "learning_rate": 2.2358028476052627e-06, "loss": 0.6915, "step": 9704 }, { "epoch": 0.7, "grad_norm": 6.580882381663226, "learning_rate": 2.23483521152567e-06, "loss": 0.9571, "step": 9705 }, { "epoch": 0.7, "grad_norm": 3.0498248553521243, "learning_rate": 2.233867724616252e-06, "loss": 0.1416, "step": 9706 }, { "epoch": 0.7, "grad_norm": 22.178466255290523, "learning_rate": 2.2329003869292004e-06, "loss": 1.0444, "step": 9707 }, { "epoch": 0.7, "grad_norm": 3.51906308979127, "learning_rate": 2.231933198516699e-06, "loss": 0.1664, "step": 9708 }, { "epoch": 0.7, "grad_norm": 4.27244949883074, "learning_rate": 2.230966159430924e-06, "loss": 0.178, "step": 9709 }, { "epoch": 0.7, "grad_norm": 14.293916723440963, "learning_rate": 2.229999269724043e-06, "loss": 0.4756, "step": 9710 }, { "epoch": 0.7, "grad_norm": 15.855863068772269, "learning_rate": 2.229032529448214e-06, "loss": 0.6488, "step": 9711 }, { "epoch": 0.7, "grad_norm": 4.2415622289625015, "learning_rate": 2.228065938655594e-06, "loss": 0.7524, "step": 9712 }, { "epoch": 0.7, "grad_norm": 6.85866813389085, "learning_rate": 2.227099497398324e-06, "loss": 0.3864, "step": 9713 }, { "epoch": 0.7, "grad_norm": 2.9386410436357377, "learning_rate": 2.2261332057285384e-06, "loss": 0.1662, "step": 9714 }, { "epoch": 0.7, "grad_norm": 2.6682321535498015, "learning_rate": 2.2251670636983684e-06, "loss": 0.1388, "step": 9715 }, { "epoch": 0.7, "grad_norm": 14.52065154591396, "learning_rate": 2.2242010713599322e-06, "loss": 0.7402, "step": 9716 }, { "epoch": 0.7, "grad_norm": 3.659683364182931, "learning_rate": 2.223235228765339e-06, "loss": 0.2399, "step": 9717 }, { "epoch": 0.7, "grad_norm": 2.6050980381993507, "learning_rate": 2.2222695359666967e-06, "loss": 0.0762, "step": 9718 }, { "epoch": 0.7, "grad_norm": 9.1865591666379, "learning_rate": 2.2213039930160985e-06, "loss": 0.2348, "step": 9719 }, { "epoch": 0.7, "grad_norm": 4.132202282835932, "learning_rate": 2.2203385999656325e-06, "loss": 0.8993, "step": 9720 }, { "epoch": 0.7, "grad_norm": 3.5752275877218365, "learning_rate": 2.2193733568673763e-06, "loss": 0.2507, "step": 9721 }, { "epoch": 0.7, "grad_norm": 5.263873454032688, "learning_rate": 2.218408263773403e-06, "loss": 0.8709, "step": 9722 }, { "epoch": 0.7, "grad_norm": 3.878508037109764, "learning_rate": 2.2174433207357732e-06, "loss": 0.305, "step": 9723 }, { "epoch": 0.7, "grad_norm": 14.02892047400649, "learning_rate": 2.216478527806545e-06, "loss": 0.8866, "step": 9724 }, { "epoch": 0.7, "grad_norm": 4.970872361269043, "learning_rate": 2.215513885037765e-06, "loss": 0.9008, "step": 9725 }, { "epoch": 0.7, "grad_norm": 3.0468819826034204, "learning_rate": 2.2145493924814685e-06, "loss": 0.1896, "step": 9726 }, { "epoch": 0.7, "grad_norm": 9.487906691503083, "learning_rate": 2.213585050189691e-06, "loss": 0.4726, "step": 9727 }, { "epoch": 0.7, "grad_norm": 5.028320588941807, "learning_rate": 2.2126208582144525e-06, "loss": 0.9903, "step": 9728 }, { "epoch": 0.7, "grad_norm": 14.134817212212333, "learning_rate": 2.2116568166077677e-06, "loss": 0.5508, "step": 9729 }, { "epoch": 0.7, "grad_norm": 3.8777717400682956, "learning_rate": 2.210692925421642e-06, "loss": 0.2283, "step": 9730 }, { "epoch": 0.7, "grad_norm": 2.787909334496422, "learning_rate": 2.2097291847080765e-06, "loss": 0.1421, "step": 9731 }, { "epoch": 0.7, "grad_norm": 4.113578211038479, "learning_rate": 2.2087655945190595e-06, "loss": 0.1659, "step": 9732 }, { "epoch": 0.7, "grad_norm": 5.437115058963347, "learning_rate": 2.2078021549065737e-06, "loss": 0.6303, "step": 9733 }, { "epoch": 0.7, "grad_norm": 2.9147386895597602, "learning_rate": 2.206838865922592e-06, "loss": 0.1109, "step": 9734 }, { "epoch": 0.7, "grad_norm": 4.348711187389663, "learning_rate": 2.2058757276190795e-06, "loss": 0.8155, "step": 9735 }, { "epoch": 0.7, "grad_norm": 4.4090083749843165, "learning_rate": 2.204912740047996e-06, "loss": 0.7639, "step": 9736 }, { "epoch": 0.7, "grad_norm": 4.465216048472551, "learning_rate": 2.2039499032612915e-06, "loss": 0.8503, "step": 9737 }, { "epoch": 0.7, "grad_norm": 21.927076352498762, "learning_rate": 2.2029872173109056e-06, "loss": 0.5369, "step": 9738 }, { "epoch": 0.7, "grad_norm": 2.8726477070841905, "learning_rate": 2.20202468224877e-06, "loss": 0.1309, "step": 9739 }, { "epoch": 0.7, "grad_norm": 9.107349047438257, "learning_rate": 2.201062298126814e-06, "loss": 0.7186, "step": 9740 }, { "epoch": 0.7, "grad_norm": 3.170664213455837, "learning_rate": 2.2001000649969524e-06, "loss": 0.1809, "step": 9741 }, { "epoch": 0.7, "grad_norm": 19.715596333166996, "learning_rate": 2.199137982911092e-06, "loss": 0.5135, "step": 9742 }, { "epoch": 0.7, "grad_norm": 3.132559683807636, "learning_rate": 2.198176051921138e-06, "loss": 0.2134, "step": 9743 }, { "epoch": 0.7, "grad_norm": 3.6419595701150764, "learning_rate": 2.197214272078982e-06, "loss": 0.2063, "step": 9744 }, { "epoch": 0.7, "grad_norm": 9.343550083631664, "learning_rate": 2.1962526434365035e-06, "loss": 0.6525, "step": 9745 }, { "epoch": 0.7, "grad_norm": 2.8089911340235516, "learning_rate": 2.1952911660455834e-06, "loss": 0.1442, "step": 9746 }, { "epoch": 0.7, "grad_norm": 14.314979469786492, "learning_rate": 2.1943298399580885e-06, "loss": 0.4656, "step": 9747 }, { "epoch": 0.7, "grad_norm": 4.024263536664181, "learning_rate": 2.193368665225877e-06, "loss": 0.1146, "step": 9748 }, { "epoch": 0.7, "grad_norm": 12.802601711889883, "learning_rate": 2.1924076419008043e-06, "loss": 0.6665, "step": 9749 }, { "epoch": 0.7, "grad_norm": 4.642396102208649, "learning_rate": 2.1914467700347126e-06, "loss": 0.6374, "step": 9750 }, { "epoch": 0.7, "grad_norm": 20.336422429602084, "learning_rate": 2.190486049679434e-06, "loss": 0.5143, "step": 9751 }, { "epoch": 0.7, "grad_norm": 4.623285689998815, "learning_rate": 2.1895254808868004e-06, "loss": 0.8563, "step": 9752 }, { "epoch": 0.7, "grad_norm": 16.738077684116092, "learning_rate": 2.188565063708629e-06, "loss": 0.4516, "step": 9753 }, { "epoch": 0.7, "grad_norm": 2.652836702084732, "learning_rate": 2.187604798196729e-06, "loss": 0.1637, "step": 9754 }, { "epoch": 0.7, "grad_norm": 3.7342147942352675, "learning_rate": 2.1866446844029067e-06, "loss": 0.205, "step": 9755 }, { "epoch": 0.7, "grad_norm": 4.584102401026861, "learning_rate": 2.1856847223789557e-06, "loss": 0.856, "step": 9756 }, { "epoch": 0.7, "grad_norm": 13.758837587699375, "learning_rate": 2.1847249121766574e-06, "loss": 0.5404, "step": 9757 }, { "epoch": 0.7, "grad_norm": 2.0705415930262507, "learning_rate": 2.183765253847796e-06, "loss": 0.0883, "step": 9758 }, { "epoch": 0.7, "grad_norm": 14.51660967402974, "learning_rate": 2.182805747444139e-06, "loss": 0.6922, "step": 9759 }, { "epoch": 0.7, "grad_norm": 4.9075738315669915, "learning_rate": 2.1818463930174464e-06, "loss": 0.8641, "step": 9760 }, { "epoch": 0.7, "grad_norm": 13.175963445834087, "learning_rate": 2.1808871906194755e-06, "loss": 0.5686, "step": 9761 }, { "epoch": 0.7, "grad_norm": 15.169389528452028, "learning_rate": 2.1799281403019695e-06, "loss": 0.7351, "step": 9762 }, { "epoch": 0.7, "grad_norm": 3.188215723020622, "learning_rate": 2.1789692421166642e-06, "loss": 0.158, "step": 9763 }, { "epoch": 0.7, "grad_norm": 2.7598568280017446, "learning_rate": 2.178010496115292e-06, "loss": 0.1445, "step": 9764 }, { "epoch": 0.7, "grad_norm": 3.095132914794193, "learning_rate": 2.177051902349571e-06, "loss": 0.1807, "step": 9765 }, { "epoch": 0.7, "grad_norm": 3.952411416768845, "learning_rate": 2.176093460871215e-06, "loss": 0.3161, "step": 9766 }, { "epoch": 0.7, "grad_norm": 3.1508209868827004, "learning_rate": 2.175135171731926e-06, "loss": 0.1102, "step": 9767 }, { "epoch": 0.7, "grad_norm": 2.506456808141142, "learning_rate": 2.174177034983406e-06, "loss": 0.1306, "step": 9768 }, { "epoch": 0.7, "grad_norm": 4.442257470980588, "learning_rate": 2.173219050677334e-06, "loss": 0.8469, "step": 9769 }, { "epoch": 0.7, "grad_norm": 4.4589376185278615, "learning_rate": 2.1722612188653963e-06, "loss": 0.7932, "step": 9770 }, { "epoch": 0.7, "grad_norm": 4.2331126661031595, "learning_rate": 2.1713035395992628e-06, "loss": 0.218, "step": 9771 }, { "epoch": 0.7, "grad_norm": 5.329410651993841, "learning_rate": 2.170346012930596e-06, "loss": 0.232, "step": 9772 }, { "epoch": 0.7, "grad_norm": 8.37046907798947, "learning_rate": 2.1693886389110487e-06, "loss": 0.325, "step": 9773 }, { "epoch": 0.7, "grad_norm": 15.650039723248161, "learning_rate": 2.1684314175922715e-06, "loss": 0.6961, "step": 9774 }, { "epoch": 0.7, "grad_norm": 5.428464631350765, "learning_rate": 2.167474349025902e-06, "loss": 0.2582, "step": 9775 }, { "epoch": 0.7, "grad_norm": 12.349824745371343, "learning_rate": 2.1665174332635674e-06, "loss": 0.6665, "step": 9776 }, { "epoch": 0.7, "grad_norm": 2.7520824996627993, "learning_rate": 2.165560670356894e-06, "loss": 0.1414, "step": 9777 }, { "epoch": 0.7, "grad_norm": 5.515539881539053, "learning_rate": 2.1646040603574935e-06, "loss": 0.9885, "step": 9778 }, { "epoch": 0.7, "grad_norm": 2.8840314359238004, "learning_rate": 2.1636476033169694e-06, "loss": 0.1577, "step": 9779 }, { "epoch": 0.7, "grad_norm": 16.597168648954316, "learning_rate": 2.1626912992869247e-06, "loss": 0.6262, "step": 9780 }, { "epoch": 0.7, "grad_norm": 2.905847138693259, "learning_rate": 2.1617351483189426e-06, "loss": 0.1966, "step": 9781 }, { "epoch": 0.7, "grad_norm": 3.2447485076532776, "learning_rate": 2.160779150464604e-06, "loss": 0.1679, "step": 9782 }, { "epoch": 0.7, "grad_norm": 12.647407764812932, "learning_rate": 2.159823305775485e-06, "loss": 0.542, "step": 9783 }, { "epoch": 0.7, "grad_norm": 4.491959576778361, "learning_rate": 2.1588676143031483e-06, "loss": 0.7887, "step": 9784 }, { "epoch": 0.7, "grad_norm": 18.064581769963514, "learning_rate": 2.157912076099148e-06, "loss": 0.4636, "step": 9785 }, { "epoch": 0.7, "grad_norm": 20.730386217945604, "learning_rate": 2.1569566912150345e-06, "loss": 0.6033, "step": 9786 }, { "epoch": 0.7, "grad_norm": 3.4104970996202324, "learning_rate": 2.156001459702346e-06, "loss": 0.1964, "step": 9787 }, { "epoch": 0.7, "grad_norm": 2.8555781623351297, "learning_rate": 2.155046381612612e-06, "loss": 0.0963, "step": 9788 }, { "epoch": 0.7, "grad_norm": 6.636293489681209, "learning_rate": 2.154091456997359e-06, "loss": 0.9542, "step": 9789 }, { "epoch": 0.7, "grad_norm": 2.7610888354248737, "learning_rate": 2.153136685908099e-06, "loss": 0.1386, "step": 9790 }, { "epoch": 0.7, "grad_norm": 3.7601720955649984, "learning_rate": 2.152182068396338e-06, "loss": 0.2154, "step": 9791 }, { "epoch": 0.7, "grad_norm": 3.6320480892341123, "learning_rate": 2.1512276045135777e-06, "loss": 0.205, "step": 9792 }, { "epoch": 0.7, "grad_norm": 19.433533319368458, "learning_rate": 2.150273294311303e-06, "loss": 0.7232, "step": 9793 }, { "epoch": 0.7, "grad_norm": 2.3373036463989214, "learning_rate": 2.149319137840996e-06, "loss": 0.1152, "step": 9794 }, { "epoch": 0.7, "grad_norm": 2.7440006154648193, "learning_rate": 2.1483651351541328e-06, "loss": 0.1141, "step": 9795 }, { "epoch": 0.7, "grad_norm": 42.86147913244138, "learning_rate": 2.147411286302176e-06, "loss": 0.55, "step": 9796 }, { "epoch": 0.7, "grad_norm": 4.098439795092645, "learning_rate": 2.1464575913365814e-06, "loss": 0.1953, "step": 9797 }, { "epoch": 0.7, "grad_norm": 2.9819611301671345, "learning_rate": 2.1455040503088004e-06, "loss": 0.1605, "step": 9798 }, { "epoch": 0.7, "grad_norm": 14.00498353668516, "learning_rate": 2.1445506632702705e-06, "loss": 0.6428, "step": 9799 }, { "epoch": 0.7, "grad_norm": 4.162527117247614, "learning_rate": 2.143597430272422e-06, "loss": 0.2464, "step": 9800 }, { "epoch": 0.7, "grad_norm": 8.47256291090701, "learning_rate": 2.1426443513666828e-06, "loss": 1.0212, "step": 9801 }, { "epoch": 0.7, "grad_norm": 12.568888667058637, "learning_rate": 2.141691426604464e-06, "loss": 0.8996, "step": 9802 }, { "epoch": 0.7, "grad_norm": 8.74186363614195, "learning_rate": 2.1407386560371747e-06, "loss": 0.4828, "step": 9803 }, { "epoch": 0.7, "grad_norm": 50.3035626239062, "learning_rate": 2.139786039716209e-06, "loss": 0.6304, "step": 9804 }, { "epoch": 0.7, "grad_norm": 3.9167017598826845, "learning_rate": 2.1388335776929643e-06, "loss": 0.1472, "step": 9805 }, { "epoch": 0.7, "grad_norm": 23.173370368229925, "learning_rate": 2.1378812700188146e-06, "loss": 0.6793, "step": 9806 }, { "epoch": 0.7, "grad_norm": 7.722183131193633, "learning_rate": 2.136929116745138e-06, "loss": 0.7479, "step": 9807 }, { "epoch": 0.7, "grad_norm": 41.275230687495, "learning_rate": 2.135977117923298e-06, "loss": 0.9359, "step": 9808 }, { "epoch": 0.7, "grad_norm": 3.5316902045878167, "learning_rate": 2.135025273604652e-06, "loss": 0.1592, "step": 9809 }, { "epoch": 0.7, "grad_norm": 3.995858156459854, "learning_rate": 2.1340735838405455e-06, "loss": 0.2925, "step": 9810 }, { "epoch": 0.7, "grad_norm": 3.478866404808177, "learning_rate": 2.133122048682324e-06, "loss": 0.16, "step": 9811 }, { "epoch": 0.7, "grad_norm": 4.336105320635075, "learning_rate": 2.1321706681813155e-06, "loss": 0.7966, "step": 9812 }, { "epoch": 0.7, "grad_norm": 13.222845006215131, "learning_rate": 2.131219442388842e-06, "loss": 0.5751, "step": 9813 }, { "epoch": 0.7, "grad_norm": 12.097814900588777, "learning_rate": 2.1302683713562234e-06, "loss": 0.795, "step": 9814 }, { "epoch": 0.7, "grad_norm": 10.575831735622327, "learning_rate": 2.129317455134763e-06, "loss": 0.4399, "step": 9815 }, { "epoch": 0.7, "grad_norm": 13.621936606074897, "learning_rate": 2.128366693775758e-06, "loss": 0.7157, "step": 9816 }, { "epoch": 0.7, "grad_norm": 17.53087017974597, "learning_rate": 2.127416087330504e-06, "loss": 0.7316, "step": 9817 }, { "epoch": 0.7, "grad_norm": 17.129456366362042, "learning_rate": 2.1264656358502765e-06, "loss": 0.9761, "step": 9818 }, { "epoch": 0.7, "grad_norm": 3.123489020132875, "learning_rate": 2.1255153393863496e-06, "loss": 0.159, "step": 9819 }, { "epoch": 0.7, "grad_norm": 2.788036520006472, "learning_rate": 2.1245651979899916e-06, "loss": 0.1211, "step": 9820 }, { "epoch": 0.7, "grad_norm": 3.103930677409132, "learning_rate": 2.123615211712457e-06, "loss": 0.1615, "step": 9821 }, { "epoch": 0.7, "grad_norm": 8.842766279764588, "learning_rate": 2.122665380604992e-06, "loss": 0.2798, "step": 9822 }, { "epoch": 0.7, "grad_norm": 2.830174279506613, "learning_rate": 2.12171570471884e-06, "loss": 0.1734, "step": 9823 }, { "epoch": 0.7, "grad_norm": 29.35874469580877, "learning_rate": 2.1207661841052314e-06, "loss": 0.9669, "step": 9824 }, { "epoch": 0.7, "grad_norm": 4.61622658030266, "learning_rate": 2.1198168188153868e-06, "loss": 0.8107, "step": 9825 }, { "epoch": 0.7, "grad_norm": 11.147336055824848, "learning_rate": 2.118867608900524e-06, "loss": 0.3598, "step": 9826 }, { "epoch": 0.7, "grad_norm": 3.1672455998726843, "learning_rate": 2.1179185544118484e-06, "loss": 0.1762, "step": 9827 }, { "epoch": 0.7, "grad_norm": 8.997663686834803, "learning_rate": 2.116969655400556e-06, "loss": 0.2815, "step": 9828 }, { "epoch": 0.7, "grad_norm": 22.934476031549043, "learning_rate": 2.116020911917841e-06, "loss": 0.5702, "step": 9829 }, { "epoch": 0.7, "grad_norm": 10.973969040696828, "learning_rate": 2.1150723240148797e-06, "loss": 0.5354, "step": 9830 }, { "epoch": 0.7, "grad_norm": 25.043330003793336, "learning_rate": 2.1141238917428443e-06, "loss": 0.5387, "step": 9831 }, { "epoch": 0.71, "grad_norm": 18.668425135970832, "learning_rate": 2.1131756151529035e-06, "loss": 0.9535, "step": 9832 }, { "epoch": 0.71, "grad_norm": 4.064057343219835, "learning_rate": 2.1122274942962107e-06, "loss": 0.2643, "step": 9833 }, { "epoch": 0.71, "grad_norm": 25.045641724289936, "learning_rate": 2.1112795292239125e-06, "loss": 1.2164, "step": 9834 }, { "epoch": 0.71, "grad_norm": 28.473702063020312, "learning_rate": 2.1103317199871507e-06, "loss": 0.5027, "step": 9835 }, { "epoch": 0.71, "grad_norm": 3.254653632088014, "learning_rate": 2.109384066637055e-06, "loss": 0.1485, "step": 9836 }, { "epoch": 0.71, "grad_norm": 14.157205127860955, "learning_rate": 2.108436569224747e-06, "loss": 0.4264, "step": 9837 }, { "epoch": 0.71, "grad_norm": 11.654899820759475, "learning_rate": 2.107489227801339e-06, "loss": 0.4102, "step": 9838 }, { "epoch": 0.71, "grad_norm": 5.1733173140599, "learning_rate": 2.1065420424179408e-06, "loss": 1.0257, "step": 9839 }, { "epoch": 0.71, "grad_norm": 3.8230568296512475, "learning_rate": 2.1055950131256466e-06, "loss": 0.1267, "step": 9840 }, { "epoch": 0.71, "grad_norm": 2.7747316990613697, "learning_rate": 2.104648139975545e-06, "loss": 0.1901, "step": 9841 }, { "epoch": 0.71, "grad_norm": 3.7232459683446613, "learning_rate": 2.103701423018718e-06, "loss": 0.1287, "step": 9842 }, { "epoch": 0.71, "grad_norm": 3.561314745242627, "learning_rate": 2.1027548623062337e-06, "loss": 0.1702, "step": 9843 }, { "epoch": 0.71, "grad_norm": 2.9144861981096213, "learning_rate": 2.1018084578891597e-06, "loss": 0.1238, "step": 9844 }, { "epoch": 0.71, "grad_norm": 6.482853124519792, "learning_rate": 2.100862209818549e-06, "loss": 0.7274, "step": 9845 }, { "epoch": 0.71, "grad_norm": 36.3427154974813, "learning_rate": 2.0999161181454485e-06, "loss": 1.2614, "step": 9846 }, { "epoch": 0.71, "grad_norm": 3.9778912828779696, "learning_rate": 2.098970182920894e-06, "loss": 0.2358, "step": 9847 }, { "epoch": 0.71, "grad_norm": 4.70862943962036, "learning_rate": 2.0980244041959192e-06, "loss": 0.7968, "step": 9848 }, { "epoch": 0.71, "grad_norm": 4.334071200091753, "learning_rate": 2.097078782021543e-06, "loss": 0.1473, "step": 9849 }, { "epoch": 0.71, "grad_norm": 4.410189816158131, "learning_rate": 2.096133316448777e-06, "loss": 0.8145, "step": 9850 }, { "epoch": 0.71, "grad_norm": 3.8998928934856893, "learning_rate": 2.095188007528628e-06, "loss": 0.2246, "step": 9851 }, { "epoch": 0.71, "grad_norm": 11.999816664792926, "learning_rate": 2.094242855312091e-06, "loss": 0.8018, "step": 9852 }, { "epoch": 0.71, "grad_norm": 4.085804091436882, "learning_rate": 2.093297859850153e-06, "loss": 0.7517, "step": 9853 }, { "epoch": 0.71, "grad_norm": 25.4823112261196, "learning_rate": 2.0923530211937933e-06, "loss": 0.7942, "step": 9854 }, { "epoch": 0.71, "grad_norm": 13.191276871182854, "learning_rate": 2.0914083393939816e-06, "loss": 0.7862, "step": 9855 }, { "epoch": 0.71, "grad_norm": 3.1352749014604755, "learning_rate": 2.0904638145016785e-06, "loss": 0.1086, "step": 9856 }, { "epoch": 0.71, "grad_norm": 3.9752044079550473, "learning_rate": 2.0895194465678412e-06, "loss": 0.1569, "step": 9857 }, { "epoch": 0.71, "grad_norm": 4.61144849447002, "learning_rate": 2.0885752356434137e-06, "loss": 0.803, "step": 9858 }, { "epoch": 0.71, "grad_norm": 2.6221424062072898, "learning_rate": 2.08763118177933e-06, "loss": 0.0946, "step": 9859 }, { "epoch": 0.71, "grad_norm": 18.75949857849664, "learning_rate": 2.0866872850265212e-06, "loss": 0.7278, "step": 9860 }, { "epoch": 0.71, "grad_norm": 16.915474727015607, "learning_rate": 2.0857435454359065e-06, "loss": 0.546, "step": 9861 }, { "epoch": 0.71, "grad_norm": 3.44842649559725, "learning_rate": 2.0847999630583944e-06, "loss": 0.2467, "step": 9862 }, { "epoch": 0.71, "grad_norm": 9.918236448702686, "learning_rate": 2.0838565379448918e-06, "loss": 0.688, "step": 9863 }, { "epoch": 0.71, "grad_norm": 5.540140439748312, "learning_rate": 2.0829132701462903e-06, "loss": 0.4314, "step": 9864 }, { "epoch": 0.71, "grad_norm": 2.775216729674137, "learning_rate": 2.081970159713476e-06, "loss": 0.2028, "step": 9865 }, { "epoch": 0.71, "grad_norm": 7.394802081125276, "learning_rate": 2.081027206697327e-06, "loss": 0.4232, "step": 9866 }, { "epoch": 0.71, "grad_norm": 3.013638943015118, "learning_rate": 2.0800844111487107e-06, "loss": 0.1014, "step": 9867 }, { "epoch": 0.71, "grad_norm": 10.899080530704866, "learning_rate": 2.079141773118486e-06, "loss": 0.4117, "step": 9868 }, { "epoch": 0.71, "grad_norm": 3.638543917152954, "learning_rate": 2.078199292657509e-06, "loss": 0.1792, "step": 9869 }, { "epoch": 0.71, "grad_norm": 9.665086783668821, "learning_rate": 2.07725696981662e-06, "loss": 0.4699, "step": 9870 }, { "epoch": 0.71, "grad_norm": 5.227127507835806, "learning_rate": 2.0763148046466523e-06, "loss": 0.8177, "step": 9871 }, { "epoch": 0.71, "grad_norm": 3.1915029083468758, "learning_rate": 2.0753727971984365e-06, "loss": 0.2199, "step": 9872 }, { "epoch": 0.71, "grad_norm": 4.379337549201956, "learning_rate": 2.074430947522788e-06, "loss": 0.8496, "step": 9873 }, { "epoch": 0.71, "grad_norm": 2.8522183829864733, "learning_rate": 2.073489255670516e-06, "loss": 0.1243, "step": 9874 }, { "epoch": 0.71, "grad_norm": 3.2945361498184633, "learning_rate": 2.0725477216924188e-06, "loss": 0.3308, "step": 9875 }, { "epoch": 0.71, "grad_norm": 2.6527816284614403, "learning_rate": 2.0716063456392933e-06, "loss": 0.2092, "step": 9876 }, { "epoch": 0.71, "grad_norm": 3.2130692114904127, "learning_rate": 2.070665127561921e-06, "loss": 0.1552, "step": 9877 }, { "epoch": 0.71, "grad_norm": 13.889611728031788, "learning_rate": 2.069724067511077e-06, "loss": 0.5291, "step": 9878 }, { "epoch": 0.71, "grad_norm": 24.267084602380784, "learning_rate": 2.0687831655375273e-06, "loss": 0.4166, "step": 9879 }, { "epoch": 0.71, "grad_norm": 5.3197819896511485, "learning_rate": 2.067842421692029e-06, "loss": 0.9832, "step": 9880 }, { "epoch": 0.71, "grad_norm": 24.691863748040255, "learning_rate": 2.066901836025335e-06, "loss": 1.2215, "step": 9881 }, { "epoch": 0.71, "grad_norm": 3.5316240392428835, "learning_rate": 2.0659614085881847e-06, "loss": 0.0836, "step": 9882 }, { "epoch": 0.71, "grad_norm": 5.917672614393281, "learning_rate": 2.0650211394313107e-06, "loss": 0.7623, "step": 9883 }, { "epoch": 0.71, "grad_norm": 4.522213569001434, "learning_rate": 2.0640810286054347e-06, "loss": 0.2438, "step": 9884 }, { "epoch": 0.71, "grad_norm": 19.13319061749379, "learning_rate": 2.0631410761612756e-06, "loss": 0.9311, "step": 9885 }, { "epoch": 0.71, "grad_norm": 5.3325305889977255, "learning_rate": 2.062201282149539e-06, "loss": 0.7724, "step": 9886 }, { "epoch": 0.71, "grad_norm": 4.6794695365720935, "learning_rate": 2.061261646620922e-06, "loss": 0.425, "step": 9887 }, { "epoch": 0.71, "grad_norm": 4.411772347748666, "learning_rate": 2.060322169626116e-06, "loss": 0.7129, "step": 9888 }, { "epoch": 0.71, "grad_norm": 3.5767331156504443, "learning_rate": 2.0593828512158044e-06, "loss": 0.2034, "step": 9889 }, { "epoch": 0.71, "grad_norm": 3.587899374968582, "learning_rate": 2.058443691440653e-06, "loss": 0.1765, "step": 9890 }, { "epoch": 0.71, "grad_norm": 8.75289168506577, "learning_rate": 2.0575046903513324e-06, "loss": 0.6756, "step": 9891 }, { "epoch": 0.71, "grad_norm": 4.4112207936651835, "learning_rate": 2.056565847998495e-06, "loss": 0.9707, "step": 9892 }, { "epoch": 0.71, "grad_norm": 12.536736671979769, "learning_rate": 2.0556271644327874e-06, "loss": 0.4387, "step": 9893 }, { "epoch": 0.71, "grad_norm": 3.265928304922543, "learning_rate": 2.054688639704851e-06, "loss": 0.1871, "step": 9894 }, { "epoch": 0.71, "grad_norm": 3.5903672514527294, "learning_rate": 2.0537502738653137e-06, "loss": 0.2359, "step": 9895 }, { "epoch": 0.71, "grad_norm": 4.088359691328724, "learning_rate": 2.0528120669647955e-06, "loss": 0.7314, "step": 9896 }, { "epoch": 0.71, "grad_norm": 3.849388193777664, "learning_rate": 2.051874019053912e-06, "loss": 0.1799, "step": 9897 }, { "epoch": 0.71, "grad_norm": 15.798213795280422, "learning_rate": 2.050936130183266e-06, "loss": 0.6033, "step": 9898 }, { "epoch": 0.71, "grad_norm": 4.143824385958544, "learning_rate": 2.049998400403451e-06, "loss": 0.7864, "step": 9899 }, { "epoch": 0.71, "grad_norm": 2.7219078263565266, "learning_rate": 2.0490608297650585e-06, "loss": 0.0959, "step": 9900 }, { "epoch": 0.71, "grad_norm": 4.678485714386562, "learning_rate": 2.0481234183186656e-06, "loss": 0.852, "step": 9901 }, { "epoch": 0.71, "grad_norm": 4.286535582316111, "learning_rate": 2.047186166114837e-06, "loss": 1.015, "step": 9902 }, { "epoch": 0.71, "grad_norm": 2.5101030632804537, "learning_rate": 2.0462490732041396e-06, "loss": 0.1268, "step": 9903 }, { "epoch": 0.71, "grad_norm": 3.0269483024498545, "learning_rate": 2.045312139637124e-06, "loss": 0.1724, "step": 9904 }, { "epoch": 0.71, "grad_norm": 3.0725143200984943, "learning_rate": 2.0443753654643326e-06, "loss": 0.2282, "step": 9905 }, { "epoch": 0.71, "grad_norm": 2.3752486467128993, "learning_rate": 2.0434387507363054e-06, "loss": 0.1159, "step": 9906 }, { "epoch": 0.71, "grad_norm": 11.123120836637732, "learning_rate": 2.042502295503565e-06, "loss": 0.4265, "step": 9907 }, { "epoch": 0.71, "grad_norm": 6.610716272193876, "learning_rate": 2.04156599981663e-06, "loss": 0.3863, "step": 9908 }, { "epoch": 0.71, "grad_norm": 5.102694641452542, "learning_rate": 2.040629863726013e-06, "loss": 0.2768, "step": 9909 }, { "epoch": 0.71, "grad_norm": 20.270762720912156, "learning_rate": 2.0396938872822124e-06, "loss": 1.2029, "step": 9910 }, { "epoch": 0.71, "grad_norm": 15.452548467865052, "learning_rate": 2.0387580705357212e-06, "loss": 0.4058, "step": 9911 }, { "epoch": 0.71, "grad_norm": 2.4303806091419085, "learning_rate": 2.0378224135370224e-06, "loss": 0.1306, "step": 9912 }, { "epoch": 0.71, "grad_norm": 16.493131240839258, "learning_rate": 2.0368869163365947e-06, "loss": 0.4447, "step": 9913 }, { "epoch": 0.71, "grad_norm": 18.315176464271502, "learning_rate": 2.035951578984898e-06, "loss": 0.7469, "step": 9914 }, { "epoch": 0.71, "grad_norm": 3.693281443741191, "learning_rate": 2.0350164015323963e-06, "loss": 0.2321, "step": 9915 }, { "epoch": 0.71, "grad_norm": 3.7470447809098717, "learning_rate": 2.034081384029537e-06, "loss": 0.1533, "step": 9916 }, { "epoch": 0.71, "grad_norm": 36.47671707152087, "learning_rate": 2.0331465265267597e-06, "loss": 1.0333, "step": 9917 }, { "epoch": 0.71, "grad_norm": 12.271026528620592, "learning_rate": 2.032211829074496e-06, "loss": 0.7699, "step": 9918 }, { "epoch": 0.71, "grad_norm": 4.003694481099229, "learning_rate": 2.0312772917231716e-06, "loss": 0.2023, "step": 9919 }, { "epoch": 0.71, "grad_norm": 4.451587123439188, "learning_rate": 2.0303429145232007e-06, "loss": 0.7678, "step": 9920 }, { "epoch": 0.71, "grad_norm": 4.729307577990793, "learning_rate": 2.0294086975249865e-06, "loss": 0.8168, "step": 9921 }, { "epoch": 0.71, "grad_norm": 5.222846227128009, "learning_rate": 2.0284746407789307e-06, "loss": 0.8613, "step": 9922 }, { "epoch": 0.71, "grad_norm": 4.18320352605196, "learning_rate": 2.0275407443354204e-06, "loss": 0.2706, "step": 9923 }, { "epoch": 0.71, "grad_norm": 4.6516965671557955, "learning_rate": 2.0266070082448334e-06, "loss": 0.7846, "step": 9924 }, { "epoch": 0.71, "grad_norm": 2.8126716000844567, "learning_rate": 2.0256734325575477e-06, "loss": 0.1046, "step": 9925 }, { "epoch": 0.71, "grad_norm": 3.909859392518363, "learning_rate": 2.0247400173239194e-06, "loss": 0.1735, "step": 9926 }, { "epoch": 0.71, "grad_norm": 3.8197990495754275, "learning_rate": 2.0238067625943035e-06, "loss": 0.7794, "step": 9927 }, { "epoch": 0.71, "grad_norm": 11.799005005653735, "learning_rate": 2.0228736684190496e-06, "loss": 0.4825, "step": 9928 }, { "epoch": 0.71, "grad_norm": 3.1867246292463705, "learning_rate": 2.021940734848492e-06, "loss": 0.2438, "step": 9929 }, { "epoch": 0.71, "grad_norm": 8.59155237425045, "learning_rate": 2.021007961932957e-06, "loss": 0.422, "step": 9930 }, { "epoch": 0.71, "grad_norm": 5.353810269605527, "learning_rate": 2.0200753497227683e-06, "loss": 0.9998, "step": 9931 }, { "epoch": 0.71, "grad_norm": 2.8317041433647057, "learning_rate": 2.019142898268235e-06, "loss": 0.1539, "step": 9932 }, { "epoch": 0.71, "grad_norm": 14.679823776055219, "learning_rate": 2.018210607619657e-06, "loss": 0.3294, "step": 9933 }, { "epoch": 0.71, "grad_norm": 3.128030016101704, "learning_rate": 2.0172784778273326e-06, "loss": 0.222, "step": 9934 }, { "epoch": 0.71, "grad_norm": 9.278632053252181, "learning_rate": 2.0163465089415434e-06, "loss": 0.4965, "step": 9935 }, { "epoch": 0.71, "grad_norm": 3.7873040913128073, "learning_rate": 2.0154147010125654e-06, "loss": 0.7503, "step": 9936 }, { "epoch": 0.71, "grad_norm": 16.83548611337223, "learning_rate": 2.0144830540906695e-06, "loss": 0.9323, "step": 9937 }, { "epoch": 0.71, "grad_norm": 32.28787658687442, "learning_rate": 2.013551568226111e-06, "loss": 0.8815, "step": 9938 }, { "epoch": 0.71, "grad_norm": 8.760808095522473, "learning_rate": 2.012620243469139e-06, "loss": 0.3252, "step": 9939 }, { "epoch": 0.71, "grad_norm": 25.791579518079217, "learning_rate": 2.0116890798699986e-06, "loss": 1.0397, "step": 9940 }, { "epoch": 0.71, "grad_norm": 23.55404810833015, "learning_rate": 2.0107580774789215e-06, "loss": 0.7251, "step": 9941 }, { "epoch": 0.71, "grad_norm": 4.1080575651573445, "learning_rate": 2.009827236346129e-06, "loss": 0.834, "step": 9942 }, { "epoch": 0.71, "grad_norm": 3.9424857815756273, "learning_rate": 2.0088965565218406e-06, "loss": 0.7589, "step": 9943 }, { "epoch": 0.71, "grad_norm": 4.325578407131187, "learning_rate": 2.007966038056261e-06, "loss": 0.88, "step": 9944 }, { "epoch": 0.71, "grad_norm": 4.078163664219083, "learning_rate": 2.0070356809995877e-06, "loss": 0.7167, "step": 9945 }, { "epoch": 0.71, "grad_norm": 4.9631496153467785, "learning_rate": 2.0061054854020095e-06, "loss": 0.1735, "step": 9946 }, { "epoch": 0.71, "grad_norm": 21.71348814040366, "learning_rate": 2.0051754513137095e-06, "loss": 1.2558, "step": 9947 }, { "epoch": 0.71, "grad_norm": 5.020466712260427, "learning_rate": 2.0042455787848575e-06, "loss": 0.8757, "step": 9948 }, { "epoch": 0.71, "grad_norm": 3.8290747505949203, "learning_rate": 2.0033158678656173e-06, "loss": 0.2138, "step": 9949 }, { "epoch": 0.71, "grad_norm": 16.152474474764198, "learning_rate": 2.0023863186061425e-06, "loss": 1.0387, "step": 9950 }, { "epoch": 0.71, "grad_norm": 3.5219743631721023, "learning_rate": 2.001456931056578e-06, "loss": 0.1173, "step": 9951 }, { "epoch": 0.71, "grad_norm": 3.322024727525596, "learning_rate": 2.0005277052670633e-06, "loss": 0.1673, "step": 9952 }, { "epoch": 0.71, "grad_norm": 25.267891970074913, "learning_rate": 1.9995986412877256e-06, "loss": 1.6032, "step": 9953 }, { "epoch": 0.71, "grad_norm": 3.635312382932724, "learning_rate": 1.9986697391686837e-06, "loss": 0.1732, "step": 9954 }, { "epoch": 0.71, "grad_norm": 9.554407234890977, "learning_rate": 1.9977409989600473e-06, "loss": 0.3683, "step": 9955 }, { "epoch": 0.71, "grad_norm": 8.84071580398641, "learning_rate": 1.9968124207119217e-06, "loss": 0.3299, "step": 9956 }, { "epoch": 0.71, "grad_norm": 24.046277530741275, "learning_rate": 1.9958840044743976e-06, "loss": 0.6138, "step": 9957 }, { "epoch": 0.71, "grad_norm": 20.5860913812348, "learning_rate": 1.994955750297559e-06, "loss": 1.2016, "step": 9958 }, { "epoch": 0.71, "grad_norm": 3.103994206167641, "learning_rate": 1.994027658231485e-06, "loss": 0.0875, "step": 9959 }, { "epoch": 0.71, "grad_norm": 16.478171917640992, "learning_rate": 1.9930997283262406e-06, "loss": 0.525, "step": 9960 }, { "epoch": 0.71, "grad_norm": 3.131174314264321, "learning_rate": 1.992171960631884e-06, "loss": 0.1823, "step": 9961 }, { "epoch": 0.71, "grad_norm": 9.114322243908163, "learning_rate": 1.9912443551984646e-06, "loss": 0.36, "step": 9962 }, { "epoch": 0.71, "grad_norm": 3.8646093010326683, "learning_rate": 1.9903169120760235e-06, "loss": 0.2078, "step": 9963 }, { "epoch": 0.71, "grad_norm": 15.708573720240686, "learning_rate": 1.989389631314591e-06, "loss": 0.5011, "step": 9964 }, { "epoch": 0.71, "grad_norm": 19.837641968955378, "learning_rate": 1.988462512964194e-06, "loss": 0.8642, "step": 9965 }, { "epoch": 0.71, "grad_norm": 15.867757047881176, "learning_rate": 1.987535557074845e-06, "loss": 0.5788, "step": 9966 }, { "epoch": 0.71, "grad_norm": 14.434143544977614, "learning_rate": 1.9866087636965485e-06, "loss": 0.7835, "step": 9967 }, { "epoch": 0.71, "grad_norm": 3.6690344234987764, "learning_rate": 1.985682132879304e-06, "loss": 0.1872, "step": 9968 }, { "epoch": 0.71, "grad_norm": 4.465073786303766, "learning_rate": 1.984755664673099e-06, "loss": 0.9112, "step": 9969 }, { "epoch": 0.71, "grad_norm": 4.813452022981269, "learning_rate": 1.983829359127911e-06, "loss": 0.9298, "step": 9970 }, { "epoch": 0.71, "grad_norm": 20.37432310788009, "learning_rate": 1.982903216293714e-06, "loss": 0.7071, "step": 9971 }, { "epoch": 0.72, "grad_norm": 6.261695187566692, "learning_rate": 1.981977236220468e-06, "loss": 0.2642, "step": 9972 }, { "epoch": 0.72, "grad_norm": 20.066649162466685, "learning_rate": 1.9810514189581265e-06, "loss": 0.4582, "step": 9973 }, { "epoch": 0.72, "grad_norm": 4.745565016453964, "learning_rate": 1.980125764556633e-06, "loss": 1.1405, "step": 9974 }, { "epoch": 0.72, "grad_norm": 2.3819914543854788, "learning_rate": 1.9792002730659242e-06, "loss": 0.1289, "step": 9975 }, { "epoch": 0.72, "grad_norm": 3.1480670618768265, "learning_rate": 1.9782749445359244e-06, "loss": 0.1661, "step": 9976 }, { "epoch": 0.72, "grad_norm": 5.4803916199607485, "learning_rate": 1.977349779016555e-06, "loss": 0.9432, "step": 9977 }, { "epoch": 0.72, "grad_norm": 16.800536260284638, "learning_rate": 1.9764247765577237e-06, "loss": 0.6764, "step": 9978 }, { "epoch": 0.72, "grad_norm": 4.353747136509646, "learning_rate": 1.97549993720933e-06, "loss": 0.2274, "step": 9979 }, { "epoch": 0.72, "grad_norm": 18.964938305598906, "learning_rate": 1.974575261021267e-06, "loss": 0.5481, "step": 9980 }, { "epoch": 0.72, "grad_norm": 9.790479204618668, "learning_rate": 1.9736507480434175e-06, "loss": 0.6974, "step": 9981 }, { "epoch": 0.72, "grad_norm": 3.33527796504761, "learning_rate": 1.9727263983256545e-06, "loss": 0.2391, "step": 9982 }, { "epoch": 0.72, "grad_norm": 10.838709783490772, "learning_rate": 1.9718022119178415e-06, "loss": 0.636, "step": 9983 }, { "epoch": 0.72, "grad_norm": 21.072744382117776, "learning_rate": 1.9708781888698386e-06, "loss": 0.8887, "step": 9984 }, { "epoch": 0.72, "grad_norm": 3.657102938205426, "learning_rate": 1.9699543292314916e-06, "loss": 0.1292, "step": 9985 }, { "epoch": 0.72, "grad_norm": 2.9614361166459258, "learning_rate": 1.9690306330526395e-06, "loss": 0.0727, "step": 9986 }, { "epoch": 0.72, "grad_norm": 3.3249130939134677, "learning_rate": 1.9681071003831115e-06, "loss": 0.2071, "step": 9987 }, { "epoch": 0.72, "grad_norm": 17.925591395352942, "learning_rate": 1.9671837312727276e-06, "loss": 0.8173, "step": 9988 }, { "epoch": 0.72, "grad_norm": 13.07107831437408, "learning_rate": 1.9662605257713035e-06, "loss": 0.5325, "step": 9989 }, { "epoch": 0.72, "grad_norm": 3.1547362901390037, "learning_rate": 1.9653374839286405e-06, "loss": 0.1533, "step": 9990 }, { "epoch": 0.72, "grad_norm": 2.776737748268746, "learning_rate": 1.9644146057945336e-06, "loss": 0.1731, "step": 9991 }, { "epoch": 0.72, "grad_norm": 11.991002241874346, "learning_rate": 1.963491891418766e-06, "loss": 0.227, "step": 9992 }, { "epoch": 0.72, "grad_norm": 28.90508728776273, "learning_rate": 1.9625693408511193e-06, "loss": 0.7193, "step": 9993 }, { "epoch": 0.72, "grad_norm": 11.394515554639835, "learning_rate": 1.9616469541413595e-06, "loss": 0.7538, "step": 9994 }, { "epoch": 0.72, "grad_norm": 3.5108939321396937, "learning_rate": 1.9607247313392436e-06, "loss": 0.1301, "step": 9995 }, { "epoch": 0.72, "grad_norm": 16.89058693545121, "learning_rate": 1.9598026724945264e-06, "loss": 0.6816, "step": 9996 }, { "epoch": 0.72, "grad_norm": 15.436074245531614, "learning_rate": 1.9588807776569486e-06, "loss": 0.7628, "step": 9997 }, { "epoch": 0.72, "grad_norm": 2.904633277461371, "learning_rate": 1.9579590468762382e-06, "loss": 0.1184, "step": 9998 }, { "epoch": 0.72, "grad_norm": 3.3935163393523933, "learning_rate": 1.957037480202124e-06, "loss": 0.1818, "step": 9999 }, { "epoch": 0.72, "grad_norm": 3.778609979982102, "learning_rate": 1.95611607768432e-06, "loss": 0.1979, "step": 10000 }, { "epoch": 0.72, "grad_norm": 3.006835086332341, "learning_rate": 1.95519483937253e-06, "loss": 0.1906, "step": 10001 }, { "epoch": 0.72, "grad_norm": 40.01649168759682, "learning_rate": 1.9542737653164546e-06, "loss": 0.6971, "step": 10002 }, { "epoch": 0.72, "grad_norm": 9.377924497254805, "learning_rate": 1.953352855565781e-06, "loss": 0.5869, "step": 10003 }, { "epoch": 0.72, "grad_norm": 17.300733337708195, "learning_rate": 1.952432110170187e-06, "loss": 0.7359, "step": 10004 }, { "epoch": 0.72, "grad_norm": 12.716496999140155, "learning_rate": 1.951511529179347e-06, "loss": 0.5515, "step": 10005 }, { "epoch": 0.72, "grad_norm": 3.306796967160343, "learning_rate": 1.9505911126429204e-06, "loss": 0.2211, "step": 10006 }, { "epoch": 0.72, "grad_norm": 9.438664249710802, "learning_rate": 1.949670860610559e-06, "loss": 0.5327, "step": 10007 }, { "epoch": 0.72, "grad_norm": 2.949530020771437, "learning_rate": 1.9487507731319105e-06, "loss": 0.1548, "step": 10008 }, { "epoch": 0.72, "grad_norm": 4.204647147030251, "learning_rate": 1.9478308502566096e-06, "loss": 0.2414, "step": 10009 }, { "epoch": 0.72, "grad_norm": 2.4679285497319197, "learning_rate": 1.946911092034278e-06, "loss": 0.1123, "step": 10010 }, { "epoch": 0.72, "grad_norm": 3.89123866599998, "learning_rate": 1.9459914985145377e-06, "loss": 0.1827, "step": 10011 }, { "epoch": 0.72, "grad_norm": 3.9391004085624624, "learning_rate": 1.9450720697469967e-06, "loss": 0.923, "step": 10012 }, { "epoch": 0.72, "grad_norm": 2.812179767883588, "learning_rate": 1.944152805781252e-06, "loss": 0.1915, "step": 10013 }, { "epoch": 0.72, "grad_norm": 11.163791912449229, "learning_rate": 1.9432337066668982e-06, "loss": 0.645, "step": 10014 }, { "epoch": 0.72, "grad_norm": 4.471223502955501, "learning_rate": 1.9423147724535152e-06, "loss": 0.6689, "step": 10015 }, { "epoch": 0.72, "grad_norm": 15.420181629372548, "learning_rate": 1.941396003190675e-06, "loss": 0.3638, "step": 10016 }, { "epoch": 0.72, "grad_norm": 47.96305103917783, "learning_rate": 1.940477398927944e-06, "loss": 0.7871, "step": 10017 }, { "epoch": 0.72, "grad_norm": 17.86021322795581, "learning_rate": 1.9395589597148767e-06, "loss": 0.9993, "step": 10018 }, { "epoch": 0.72, "grad_norm": 11.114888861894263, "learning_rate": 1.938640685601019e-06, "loss": 0.3731, "step": 10019 }, { "epoch": 0.72, "grad_norm": 3.06070484300682, "learning_rate": 1.9377225766359077e-06, "loss": 0.1446, "step": 10020 }, { "epoch": 0.72, "grad_norm": 14.219485560445674, "learning_rate": 1.9368046328690724e-06, "loss": 0.5335, "step": 10021 }, { "epoch": 0.72, "grad_norm": 5.09506685065878, "learning_rate": 1.935886854350033e-06, "loss": 0.9675, "step": 10022 }, { "epoch": 0.72, "grad_norm": 3.867617171112302, "learning_rate": 1.9349692411282995e-06, "loss": 0.1461, "step": 10023 }, { "epoch": 0.72, "grad_norm": 2.851806100991467, "learning_rate": 1.9340517932533736e-06, "loss": 0.1318, "step": 10024 }, { "epoch": 0.72, "grad_norm": 4.732972224171169, "learning_rate": 1.9331345107747486e-06, "loss": 0.1338, "step": 10025 }, { "epoch": 0.72, "grad_norm": 2.755769683517298, "learning_rate": 1.9322173937419058e-06, "loss": 0.1695, "step": 10026 }, { "epoch": 0.72, "grad_norm": 3.9387085976312615, "learning_rate": 1.931300442204324e-06, "loss": 0.19, "step": 10027 }, { "epoch": 0.72, "grad_norm": 2.686633634611206, "learning_rate": 1.930383656211468e-06, "loss": 0.154, "step": 10028 }, { "epoch": 0.72, "grad_norm": 3.0971949895074204, "learning_rate": 1.929467035812793e-06, "loss": 0.1692, "step": 10029 }, { "epoch": 0.72, "grad_norm": 9.181582760417356, "learning_rate": 1.9285505810577505e-06, "loss": 0.6647, "step": 10030 }, { "epoch": 0.72, "grad_norm": 3.606817139841977, "learning_rate": 1.927634291995778e-06, "loss": 0.1548, "step": 10031 }, { "epoch": 0.72, "grad_norm": 42.939289881093764, "learning_rate": 1.9267181686763037e-06, "loss": 0.5452, "step": 10032 }, { "epoch": 0.72, "grad_norm": 25.051026615953482, "learning_rate": 1.9258022111487533e-06, "loss": 0.7968, "step": 10033 }, { "epoch": 0.72, "grad_norm": 7.656873543555542, "learning_rate": 1.9248864194625392e-06, "loss": 0.2569, "step": 10034 }, { "epoch": 0.72, "grad_norm": 2.576466552438447, "learning_rate": 1.923970793667059e-06, "loss": 0.1689, "step": 10035 }, { "epoch": 0.72, "grad_norm": 4.229346288088699, "learning_rate": 1.923055333811713e-06, "loss": 0.292, "step": 10036 }, { "epoch": 0.72, "grad_norm": 3.163261086328545, "learning_rate": 1.922140039945885e-06, "loss": 0.1892, "step": 10037 }, { "epoch": 0.72, "grad_norm": 3.769945332647322, "learning_rate": 1.9212249121189498e-06, "loss": 0.7356, "step": 10038 }, { "epoch": 0.72, "grad_norm": 4.233367876760837, "learning_rate": 1.920309950380278e-06, "loss": 0.1935, "step": 10039 }, { "epoch": 0.72, "grad_norm": 21.492976242339353, "learning_rate": 1.9193951547792273e-06, "loss": 0.7019, "step": 10040 }, { "epoch": 0.72, "grad_norm": 4.857762528850556, "learning_rate": 1.9184805253651456e-06, "loss": 0.874, "step": 10041 }, { "epoch": 0.72, "grad_norm": 4.179634937579628, "learning_rate": 1.9175660621873777e-06, "loss": 0.2013, "step": 10042 }, { "epoch": 0.72, "grad_norm": 3.8544757120195245, "learning_rate": 1.916651765295252e-06, "loss": 0.2001, "step": 10043 }, { "epoch": 0.72, "grad_norm": 4.080763134739516, "learning_rate": 1.915737634738092e-06, "loss": 0.2151, "step": 10044 }, { "epoch": 0.72, "grad_norm": 16.79681946604653, "learning_rate": 1.9148236705652128e-06, "loss": 0.6292, "step": 10045 }, { "epoch": 0.72, "grad_norm": 3.221593144888047, "learning_rate": 1.9139098728259205e-06, "loss": 0.1821, "step": 10046 }, { "epoch": 0.72, "grad_norm": 14.566092270125027, "learning_rate": 1.912996241569506e-06, "loss": 0.4617, "step": 10047 }, { "epoch": 0.72, "grad_norm": 3.6236625945489966, "learning_rate": 1.9120827768452606e-06, "loss": 0.1959, "step": 10048 }, { "epoch": 0.72, "grad_norm": 26.239868230047716, "learning_rate": 1.911169478702461e-06, "loss": 0.9144, "step": 10049 }, { "epoch": 0.72, "grad_norm": 18.37568774884481, "learning_rate": 1.9102563471903744e-06, "loss": 0.8896, "step": 10050 }, { "epoch": 0.72, "grad_norm": 3.5618173996590436, "learning_rate": 1.9093433823582646e-06, "loss": 0.1688, "step": 10051 }, { "epoch": 0.72, "grad_norm": 4.746724048505069, "learning_rate": 1.9084305842553802e-06, "loss": 0.8708, "step": 10052 }, { "epoch": 0.72, "grad_norm": 5.469170456947858, "learning_rate": 1.907517952930962e-06, "loss": 0.8767, "step": 10053 }, { "epoch": 0.72, "grad_norm": 22.807382866031514, "learning_rate": 1.9066054884342461e-06, "loss": 0.6341, "step": 10054 }, { "epoch": 0.72, "grad_norm": 3.728183509973753, "learning_rate": 1.9056931908144555e-06, "loss": 0.1126, "step": 10055 }, { "epoch": 0.72, "grad_norm": 4.4093150821525455, "learning_rate": 1.904781060120804e-06, "loss": 0.8345, "step": 10056 }, { "epoch": 0.72, "grad_norm": 3.5930560557529847, "learning_rate": 1.903869096402497e-06, "loss": 0.125, "step": 10057 }, { "epoch": 0.72, "grad_norm": 30.605789541458577, "learning_rate": 1.9029572997087358e-06, "loss": 1.5238, "step": 10058 }, { "epoch": 0.72, "grad_norm": 17.016060206318624, "learning_rate": 1.9020456700887024e-06, "loss": 1.0197, "step": 10059 }, { "epoch": 0.72, "grad_norm": 12.274080544894893, "learning_rate": 1.90113420759158e-06, "loss": 0.4458, "step": 10060 }, { "epoch": 0.72, "grad_norm": 18.681807820872184, "learning_rate": 1.9002229122665377e-06, "loss": 0.5668, "step": 10061 }, { "epoch": 0.72, "grad_norm": 2.9785263698319984, "learning_rate": 1.8993117841627357e-06, "loss": 0.1377, "step": 10062 }, { "epoch": 0.72, "grad_norm": 2.952108272019605, "learning_rate": 1.8984008233293243e-06, "loss": 0.1413, "step": 10063 }, { "epoch": 0.72, "grad_norm": 2.691766517008156, "learning_rate": 1.8974900298154502e-06, "loss": 0.1365, "step": 10064 }, { "epoch": 0.72, "grad_norm": 19.11668187178378, "learning_rate": 1.8965794036702456e-06, "loss": 1.0671, "step": 10065 }, { "epoch": 0.72, "grad_norm": 3.159727819487083, "learning_rate": 1.8956689449428333e-06, "loss": 0.1442, "step": 10066 }, { "epoch": 0.72, "grad_norm": 3.304035856829418, "learning_rate": 1.8947586536823326e-06, "loss": 0.1247, "step": 10067 }, { "epoch": 0.72, "grad_norm": 3.481903991951205, "learning_rate": 1.8938485299378484e-06, "loss": 0.1687, "step": 10068 }, { "epoch": 0.72, "grad_norm": 4.1625863407678745, "learning_rate": 1.8929385737584765e-06, "loss": 0.2393, "step": 10069 }, { "epoch": 0.72, "grad_norm": 3.6656517304781677, "learning_rate": 1.8920287851933117e-06, "loss": 0.1241, "step": 10070 }, { "epoch": 0.72, "grad_norm": 3.772435374198758, "learning_rate": 1.8911191642914279e-06, "loss": 0.2261, "step": 10071 }, { "epoch": 0.72, "grad_norm": 20.077207314310527, "learning_rate": 1.8902097111018952e-06, "loss": 0.7405, "step": 10072 }, { "epoch": 0.72, "grad_norm": 10.074112085675644, "learning_rate": 1.8893004256737796e-06, "loss": 0.5872, "step": 10073 }, { "epoch": 0.72, "grad_norm": 19.55461427156388, "learning_rate": 1.8883913080561312e-06, "loss": 0.4741, "step": 10074 }, { "epoch": 0.72, "grad_norm": 3.139384094372276, "learning_rate": 1.8874823582979918e-06, "loss": 0.2034, "step": 10075 }, { "epoch": 0.72, "grad_norm": 3.065079010559057, "learning_rate": 1.8865735764483995e-06, "loss": 0.2159, "step": 10076 }, { "epoch": 0.72, "grad_norm": 2.401132836539505, "learning_rate": 1.8856649625563778e-06, "loss": 0.0755, "step": 10077 }, { "epoch": 0.72, "grad_norm": 5.242803725494899, "learning_rate": 1.8847565166709413e-06, "loss": 0.1393, "step": 10078 }, { "epoch": 0.72, "grad_norm": 16.14899858655588, "learning_rate": 1.8838482388411005e-06, "loss": 1.2092, "step": 10079 }, { "epoch": 0.72, "grad_norm": 9.373619530908995, "learning_rate": 1.8829401291158523e-06, "loss": 0.314, "step": 10080 }, { "epoch": 0.72, "grad_norm": 4.651952235587754, "learning_rate": 1.8820321875441832e-06, "loss": 0.8241, "step": 10081 }, { "epoch": 0.72, "grad_norm": 9.861179373744234, "learning_rate": 1.8811244141750795e-06, "loss": 0.5891, "step": 10082 }, { "epoch": 0.72, "grad_norm": 8.892073037320323, "learning_rate": 1.8802168090575056e-06, "loss": 0.4763, "step": 10083 }, { "epoch": 0.72, "grad_norm": 22.24696595662173, "learning_rate": 1.8793093722404249e-06, "loss": 0.5456, "step": 10084 }, { "epoch": 0.72, "grad_norm": 2.8953450599499875, "learning_rate": 1.8784021037727924e-06, "loss": 0.1835, "step": 10085 }, { "epoch": 0.72, "grad_norm": 3.3752536509227546, "learning_rate": 1.87749500370355e-06, "loss": 0.1448, "step": 10086 }, { "epoch": 0.72, "grad_norm": 4.155389627243045, "learning_rate": 1.876588072081632e-06, "loss": 0.3182, "step": 10087 }, { "epoch": 0.72, "grad_norm": 8.0498215138883, "learning_rate": 1.875681308955966e-06, "loss": 0.1682, "step": 10088 }, { "epoch": 0.72, "grad_norm": 3.0640820355672425, "learning_rate": 1.8747747143754675e-06, "loss": 0.2644, "step": 10089 }, { "epoch": 0.72, "grad_norm": 30.997640541333634, "learning_rate": 1.873868288389043e-06, "loss": 1.2347, "step": 10090 }, { "epoch": 0.72, "grad_norm": 5.0769560078234965, "learning_rate": 1.8729620310455898e-06, "loss": 0.7187, "step": 10091 }, { "epoch": 0.72, "grad_norm": 4.011262768540066, "learning_rate": 1.8720559423940004e-06, "loss": 0.2621, "step": 10092 }, { "epoch": 0.72, "grad_norm": 3.1885814555144014, "learning_rate": 1.871150022483153e-06, "loss": 0.0976, "step": 10093 }, { "epoch": 0.72, "grad_norm": 11.153036884436407, "learning_rate": 1.8702442713619184e-06, "loss": 0.7114, "step": 10094 }, { "epoch": 0.72, "grad_norm": 13.014732795750097, "learning_rate": 1.8693386890791587e-06, "loss": 0.7722, "step": 10095 }, { "epoch": 0.72, "grad_norm": 13.919820203234224, "learning_rate": 1.8684332756837249e-06, "loss": 0.541, "step": 10096 }, { "epoch": 0.72, "grad_norm": 7.996747533267251, "learning_rate": 1.8675280312244638e-06, "loss": 0.3912, "step": 10097 }, { "epoch": 0.72, "grad_norm": 4.965339925425855, "learning_rate": 1.8666229557502086e-06, "loss": 0.8434, "step": 10098 }, { "epoch": 0.72, "grad_norm": 3.1221037051620075, "learning_rate": 1.8657180493097842e-06, "loss": 0.1341, "step": 10099 }, { "epoch": 0.72, "grad_norm": 20.635052788632116, "learning_rate": 1.8648133119520056e-06, "loss": 0.8426, "step": 10100 }, { "epoch": 0.72, "grad_norm": 28.01374897138542, "learning_rate": 1.863908743725683e-06, "loss": 0.5954, "step": 10101 }, { "epoch": 0.72, "grad_norm": 4.39366963103133, "learning_rate": 1.8630043446796132e-06, "loss": 0.7599, "step": 10102 }, { "epoch": 0.72, "grad_norm": 2.869324523030478, "learning_rate": 1.8621001148625827e-06, "loss": 0.1804, "step": 10103 }, { "epoch": 0.72, "grad_norm": 6.914695864037026, "learning_rate": 1.8611960543233753e-06, "loss": 0.3156, "step": 10104 }, { "epoch": 0.72, "grad_norm": 46.52591364858745, "learning_rate": 1.8602921631107602e-06, "loss": 0.6016, "step": 10105 }, { "epoch": 0.72, "grad_norm": 3.303013084649899, "learning_rate": 1.859388441273498e-06, "loss": 0.18, "step": 10106 }, { "epoch": 0.72, "grad_norm": 3.776160609146842, "learning_rate": 1.8584848888603418e-06, "loss": 0.1905, "step": 10107 }, { "epoch": 0.72, "grad_norm": 4.318773770332027, "learning_rate": 1.8575815059200342e-06, "loss": 0.7372, "step": 10108 }, { "epoch": 0.72, "grad_norm": 17.247207212910872, "learning_rate": 1.8566782925013082e-06, "loss": 0.6025, "step": 10109 }, { "epoch": 0.72, "grad_norm": 12.01859343759623, "learning_rate": 1.8557752486528918e-06, "loss": 0.4644, "step": 10110 }, { "epoch": 0.73, "grad_norm": 2.802283416347296, "learning_rate": 1.8548723744234993e-06, "loss": 0.1379, "step": 10111 }, { "epoch": 0.73, "grad_norm": 5.347110722983089, "learning_rate": 1.8539696698618354e-06, "loss": 0.9507, "step": 10112 }, { "epoch": 0.73, "grad_norm": 15.03451609690565, "learning_rate": 1.8530671350166012e-06, "loss": 0.9251, "step": 10113 }, { "epoch": 0.73, "grad_norm": 4.792089118932734, "learning_rate": 1.8521647699364837e-06, "loss": 0.8589, "step": 10114 }, { "epoch": 0.73, "grad_norm": 4.3277534056621105, "learning_rate": 1.8512625746701595e-06, "loss": 0.1435, "step": 10115 }, { "epoch": 0.73, "grad_norm": 3.0874659062706753, "learning_rate": 1.8503605492663023e-06, "loss": 0.1068, "step": 10116 }, { "epoch": 0.73, "grad_norm": 49.36648902796341, "learning_rate": 1.849458693773572e-06, "loss": 1.0403, "step": 10117 }, { "epoch": 0.73, "grad_norm": 3.752819376591786, "learning_rate": 1.848557008240619e-06, "loss": 0.1939, "step": 10118 }, { "epoch": 0.73, "grad_norm": 4.048361329025615, "learning_rate": 1.8476554927160872e-06, "loss": 0.1624, "step": 10119 }, { "epoch": 0.73, "grad_norm": 4.660552877796638, "learning_rate": 1.8467541472486089e-06, "loss": 0.5719, "step": 10120 }, { "epoch": 0.73, "grad_norm": 27.60856704110569, "learning_rate": 1.8458529718868072e-06, "loss": 0.8041, "step": 10121 }, { "epoch": 0.73, "grad_norm": 3.164996275949656, "learning_rate": 1.8449519666793003e-06, "loss": 0.1632, "step": 10122 }, { "epoch": 0.73, "grad_norm": 3.2542591384321304, "learning_rate": 1.8440511316746924e-06, "loss": 0.1416, "step": 10123 }, { "epoch": 0.73, "grad_norm": 3.719227009193997, "learning_rate": 1.8431504669215777e-06, "loss": 0.2475, "step": 10124 }, { "epoch": 0.73, "grad_norm": 3.236056952014402, "learning_rate": 1.8422499724685483e-06, "loss": 0.1061, "step": 10125 }, { "epoch": 0.73, "grad_norm": 3.110359192116609, "learning_rate": 1.84134964836418e-06, "loss": 0.1219, "step": 10126 }, { "epoch": 0.73, "grad_norm": 3.9246978518929025, "learning_rate": 1.8404494946570424e-06, "loss": 0.2125, "step": 10127 }, { "epoch": 0.73, "grad_norm": 8.206114385803279, "learning_rate": 1.8395495113956923e-06, "loss": 0.4064, "step": 10128 }, { "epoch": 0.73, "grad_norm": 3.22516570007026, "learning_rate": 1.8386496986286856e-06, "loss": 0.2036, "step": 10129 }, { "epoch": 0.73, "grad_norm": 2.578566759284842, "learning_rate": 1.8377500564045615e-06, "loss": 0.1551, "step": 10130 }, { "epoch": 0.73, "grad_norm": 2.875232486614445, "learning_rate": 1.8368505847718516e-06, "loss": 0.2169, "step": 10131 }, { "epoch": 0.73, "grad_norm": 9.990842003721786, "learning_rate": 1.8359512837790794e-06, "loss": 0.8504, "step": 10132 }, { "epoch": 0.73, "grad_norm": 4.634825950470816, "learning_rate": 1.8350521534747573e-06, "loss": 0.6323, "step": 10133 }, { "epoch": 0.73, "grad_norm": 13.95873056906726, "learning_rate": 1.834153193907393e-06, "loss": 0.7423, "step": 10134 }, { "epoch": 0.73, "grad_norm": 16.600515240592873, "learning_rate": 1.8332544051254813e-06, "loss": 0.7979, "step": 10135 }, { "epoch": 0.73, "grad_norm": 5.559672207811562, "learning_rate": 1.8323557871775072e-06, "loss": 0.158, "step": 10136 }, { "epoch": 0.73, "grad_norm": 15.047554222300128, "learning_rate": 1.8314573401119462e-06, "loss": 0.5489, "step": 10137 }, { "epoch": 0.73, "grad_norm": 3.4258708413126286, "learning_rate": 1.8305590639772701e-06, "loss": 0.1557, "step": 10138 }, { "epoch": 0.73, "grad_norm": 3.129199170804912, "learning_rate": 1.8296609588219354e-06, "loss": 0.157, "step": 10139 }, { "epoch": 0.73, "grad_norm": 3.598035111962246, "learning_rate": 1.8287630246943899e-06, "loss": 0.1723, "step": 10140 }, { "epoch": 0.73, "grad_norm": 4.3463051177705925, "learning_rate": 1.8278652616430776e-06, "loss": 0.7847, "step": 10141 }, { "epoch": 0.73, "grad_norm": 4.161637352595786, "learning_rate": 1.8269676697164285e-06, "loss": 0.7162, "step": 10142 }, { "epoch": 0.73, "grad_norm": 3.6979212164362467, "learning_rate": 1.8260702489628595e-06, "loss": 0.2064, "step": 10143 }, { "epoch": 0.73, "grad_norm": 4.3197237003113145, "learning_rate": 1.8251729994307888e-06, "loss": 0.8029, "step": 10144 }, { "epoch": 0.73, "grad_norm": 3.3600097255908117, "learning_rate": 1.8242759211686168e-06, "loss": 0.1938, "step": 10145 }, { "epoch": 0.73, "grad_norm": 3.37032055986898, "learning_rate": 1.8233790142247366e-06, "loss": 0.1752, "step": 10146 }, { "epoch": 0.73, "grad_norm": 2.901560601026379, "learning_rate": 1.822482278647536e-06, "loss": 0.1194, "step": 10147 }, { "epoch": 0.73, "grad_norm": 4.425780375886859, "learning_rate": 1.8215857144853894e-06, "loss": 0.9297, "step": 10148 }, { "epoch": 0.73, "grad_norm": 10.908967793085036, "learning_rate": 1.8206893217866606e-06, "loss": 0.3967, "step": 10149 }, { "epoch": 0.73, "grad_norm": 13.207532614033422, "learning_rate": 1.81979310059971e-06, "loss": 0.5391, "step": 10150 }, { "epoch": 0.73, "grad_norm": 3.2896967245429902, "learning_rate": 1.8188970509728842e-06, "loss": 0.1874, "step": 10151 }, { "epoch": 0.73, "grad_norm": 4.437359086497791, "learning_rate": 1.8180011729545193e-06, "loss": 0.9511, "step": 10152 }, { "epoch": 0.73, "grad_norm": 5.373998899323001, "learning_rate": 1.8171054665929488e-06, "loss": 1.0088, "step": 10153 }, { "epoch": 0.73, "grad_norm": 9.141616678022183, "learning_rate": 1.8162099319364923e-06, "loss": 0.3118, "step": 10154 }, { "epoch": 0.73, "grad_norm": 4.218281862496716, "learning_rate": 1.8153145690334544e-06, "loss": 0.7894, "step": 10155 }, { "epoch": 0.73, "grad_norm": 12.51963733164666, "learning_rate": 1.8144193779321435e-06, "loss": 0.6343, "step": 10156 }, { "epoch": 0.73, "grad_norm": 2.460624965256467, "learning_rate": 1.8135243586808482e-06, "loss": 0.0982, "step": 10157 }, { "epoch": 0.73, "grad_norm": 2.513131951798802, "learning_rate": 1.8126295113278513e-06, "loss": 0.1356, "step": 10158 }, { "epoch": 0.73, "grad_norm": 12.270601301790206, "learning_rate": 1.8117348359214287e-06, "loss": 0.4784, "step": 10159 }, { "epoch": 0.73, "grad_norm": 4.22613983887695, "learning_rate": 1.8108403325098434e-06, "loss": 0.7953, "step": 10160 }, { "epoch": 0.73, "grad_norm": 20.042121637096294, "learning_rate": 1.809946001141349e-06, "loss": 0.9403, "step": 10161 }, { "epoch": 0.73, "grad_norm": 40.37534378639496, "learning_rate": 1.8090518418641945e-06, "loss": 0.4023, "step": 10162 }, { "epoch": 0.73, "grad_norm": 4.152482474767009, "learning_rate": 1.8081578547266148e-06, "loss": 0.8756, "step": 10163 }, { "epoch": 0.73, "grad_norm": 4.378046314768873, "learning_rate": 1.8072640397768371e-06, "loss": 0.7586, "step": 10164 }, { "epoch": 0.73, "grad_norm": 2.694023312880524, "learning_rate": 1.8063703970630775e-06, "loss": 0.1132, "step": 10165 }, { "epoch": 0.73, "grad_norm": 19.626842827884815, "learning_rate": 1.8054769266335498e-06, "loss": 0.8749, "step": 10166 }, { "epoch": 0.73, "grad_norm": 22.05973091711054, "learning_rate": 1.8045836285364465e-06, "loss": 0.8202, "step": 10167 }, { "epoch": 0.73, "grad_norm": 12.145689795340662, "learning_rate": 1.803690502819963e-06, "loss": 0.7274, "step": 10168 }, { "epoch": 0.73, "grad_norm": 2.209128606374317, "learning_rate": 1.8027975495322785e-06, "loss": 0.091, "step": 10169 }, { "epoch": 0.73, "grad_norm": 3.9509286703411193, "learning_rate": 1.8019047687215634e-06, "loss": 0.6313, "step": 10170 }, { "epoch": 0.73, "grad_norm": 2.646787640369455, "learning_rate": 1.8010121604359794e-06, "loss": 0.1284, "step": 10171 }, { "epoch": 0.73, "grad_norm": 3.6377877569951225, "learning_rate": 1.8001197247236824e-06, "loss": 0.155, "step": 10172 }, { "epoch": 0.73, "grad_norm": 16.13701609297321, "learning_rate": 1.7992274616328137e-06, "loss": 1.0022, "step": 10173 }, { "epoch": 0.73, "grad_norm": 13.992536892036371, "learning_rate": 1.798335371211506e-06, "loss": 0.4694, "step": 10174 }, { "epoch": 0.73, "grad_norm": 22.921062552566326, "learning_rate": 1.7974434535078876e-06, "loss": 0.3658, "step": 10175 }, { "epoch": 0.73, "grad_norm": 16.63556619513915, "learning_rate": 1.7965517085700735e-06, "loss": 1.19, "step": 10176 }, { "epoch": 0.73, "grad_norm": 3.6382575587552113, "learning_rate": 1.7956601364461662e-06, "loss": 0.138, "step": 10177 }, { "epoch": 0.73, "grad_norm": 3.8081281670236033, "learning_rate": 1.7947687371842698e-06, "loss": 0.2244, "step": 10178 }, { "epoch": 0.73, "grad_norm": 4.065297054317171, "learning_rate": 1.7938775108324653e-06, "loss": 0.7965, "step": 10179 }, { "epoch": 0.73, "grad_norm": 30.48586680892871, "learning_rate": 1.7929864574388318e-06, "loss": 0.9865, "step": 10180 }, { "epoch": 0.73, "grad_norm": 4.183397007963294, "learning_rate": 1.7920955770514408e-06, "loss": 0.1562, "step": 10181 }, { "epoch": 0.73, "grad_norm": 4.898706152698156, "learning_rate": 1.7912048697183516e-06, "loss": 0.7794, "step": 10182 }, { "epoch": 0.73, "grad_norm": 3.752930511171949, "learning_rate": 1.7903143354876117e-06, "loss": 0.169, "step": 10183 }, { "epoch": 0.73, "grad_norm": 5.00071532262765, "learning_rate": 1.789423974407266e-06, "loss": 0.7945, "step": 10184 }, { "epoch": 0.73, "grad_norm": 3.712346308096151, "learning_rate": 1.7885337865253439e-06, "loss": 0.1624, "step": 10185 }, { "epoch": 0.73, "grad_norm": 3.52319971582224, "learning_rate": 1.7876437718898665e-06, "loss": 0.1582, "step": 10186 }, { "epoch": 0.73, "grad_norm": 20.890814487921475, "learning_rate": 1.7867539305488496e-06, "loss": 1.0227, "step": 10187 }, { "epoch": 0.73, "grad_norm": 5.126635628061291, "learning_rate": 1.7858642625502958e-06, "loss": 1.0197, "step": 10188 }, { "epoch": 0.73, "grad_norm": 3.322067459309955, "learning_rate": 1.784974767942197e-06, "loss": 0.2003, "step": 10189 }, { "epoch": 0.73, "grad_norm": 23.355015748726515, "learning_rate": 1.7840854467725433e-06, "loss": 0.7348, "step": 10190 }, { "epoch": 0.73, "grad_norm": 5.359227126061947, "learning_rate": 1.7831962990893047e-06, "loss": 0.3944, "step": 10191 }, { "epoch": 0.73, "grad_norm": 8.89526826226362, "learning_rate": 1.7823073249404481e-06, "loss": 0.4347, "step": 10192 }, { "epoch": 0.73, "grad_norm": 2.6997187783051584, "learning_rate": 1.781418524373934e-06, "loss": 0.1985, "step": 10193 }, { "epoch": 0.73, "grad_norm": 3.734866915977488, "learning_rate": 1.7805298974377067e-06, "loss": 0.7842, "step": 10194 }, { "epoch": 0.73, "grad_norm": 1.9049584601862075, "learning_rate": 1.7796414441797038e-06, "loss": 0.0865, "step": 10195 }, { "epoch": 0.73, "grad_norm": 2.848149296567831, "learning_rate": 1.7787531646478567e-06, "loss": 0.1563, "step": 10196 }, { "epoch": 0.73, "grad_norm": 1.6478760515590665, "learning_rate": 1.7778650588900837e-06, "loss": 0.0657, "step": 10197 }, { "epoch": 0.73, "grad_norm": 3.9631836892356804, "learning_rate": 1.7769771269542941e-06, "loss": 0.2377, "step": 10198 }, { "epoch": 0.73, "grad_norm": 14.736411725104634, "learning_rate": 1.7760893688883868e-06, "loss": 0.7103, "step": 10199 }, { "epoch": 0.73, "grad_norm": 16.32928937934259, "learning_rate": 1.775201784740257e-06, "loss": 0.2932, "step": 10200 }, { "epoch": 0.73, "grad_norm": 11.891166741292022, "learning_rate": 1.774314374557784e-06, "loss": 0.8152, "step": 10201 }, { "epoch": 0.73, "grad_norm": 27.773394318138457, "learning_rate": 1.7734271383888407e-06, "loss": 1.0536, "step": 10202 }, { "epoch": 0.73, "grad_norm": 3.9028526823562846, "learning_rate": 1.77254007628129e-06, "loss": 0.1641, "step": 10203 }, { "epoch": 0.73, "grad_norm": 3.3359047346018986, "learning_rate": 1.7716531882829841e-06, "loss": 0.1317, "step": 10204 }, { "epoch": 0.73, "grad_norm": 3.469988155104739, "learning_rate": 1.7707664744417702e-06, "loss": 0.1179, "step": 10205 }, { "epoch": 0.73, "grad_norm": 34.94695104355312, "learning_rate": 1.7698799348054818e-06, "loss": 0.733, "step": 10206 }, { "epoch": 0.73, "grad_norm": 4.609447530692274, "learning_rate": 1.7689935694219446e-06, "loss": 0.1291, "step": 10207 }, { "epoch": 0.73, "grad_norm": 3.028330005286833, "learning_rate": 1.768107378338972e-06, "loss": 0.1055, "step": 10208 }, { "epoch": 0.73, "grad_norm": 3.9027875818413453, "learning_rate": 1.7672213616043748e-06, "loss": 0.2175, "step": 10209 }, { "epoch": 0.73, "grad_norm": 17.006333917839633, "learning_rate": 1.7663355192659488e-06, "loss": 1.1654, "step": 10210 }, { "epoch": 0.73, "grad_norm": 3.2749592049114167, "learning_rate": 1.765449851371479e-06, "loss": 0.0892, "step": 10211 }, { "epoch": 0.73, "grad_norm": 10.561900327194822, "learning_rate": 1.7645643579687488e-06, "loss": 0.461, "step": 10212 }, { "epoch": 0.73, "grad_norm": 3.4279007087856557, "learning_rate": 1.7636790391055242e-06, "loss": 0.0824, "step": 10213 }, { "epoch": 0.73, "grad_norm": 2.1873064556255675, "learning_rate": 1.7627938948295654e-06, "loss": 0.1388, "step": 10214 }, { "epoch": 0.73, "grad_norm": 2.697605703255765, "learning_rate": 1.761908925188623e-06, "loss": 0.1337, "step": 10215 }, { "epoch": 0.73, "grad_norm": 14.056096601917991, "learning_rate": 1.7610241302304364e-06, "loss": 0.7723, "step": 10216 }, { "epoch": 0.73, "grad_norm": 3.002689761529251, "learning_rate": 1.7601395100027368e-06, "loss": 0.1476, "step": 10217 }, { "epoch": 0.73, "grad_norm": 3.663220571656648, "learning_rate": 1.759255064553248e-06, "loss": 0.267, "step": 10218 }, { "epoch": 0.73, "grad_norm": 15.148271346233669, "learning_rate": 1.7583707939296824e-06, "loss": 0.9732, "step": 10219 }, { "epoch": 0.73, "grad_norm": 5.111899012423692, "learning_rate": 1.7574866981797401e-06, "loss": 0.8182, "step": 10220 }, { "epoch": 0.73, "grad_norm": 3.2392653354146104, "learning_rate": 1.7566027773511185e-06, "loss": 0.1618, "step": 10221 }, { "epoch": 0.73, "grad_norm": 5.49738960124646, "learning_rate": 1.7557190314915007e-06, "loss": 0.8453, "step": 10222 }, { "epoch": 0.73, "grad_norm": 3.619667317074064, "learning_rate": 1.7548354606485585e-06, "loss": 0.1965, "step": 10223 }, { "epoch": 0.73, "grad_norm": 17.307597972100005, "learning_rate": 1.7539520648699615e-06, "loss": 0.5095, "step": 10224 }, { "epoch": 0.73, "grad_norm": 4.919906232836372, "learning_rate": 1.7530688442033638e-06, "loss": 0.8559, "step": 10225 }, { "epoch": 0.73, "grad_norm": 6.669987759480661, "learning_rate": 1.7521857986964113e-06, "loss": 0.3126, "step": 10226 }, { "epoch": 0.73, "grad_norm": 45.379693605774875, "learning_rate": 1.7513029283967414e-06, "loss": 0.594, "step": 10227 }, { "epoch": 0.73, "grad_norm": 3.879240771235177, "learning_rate": 1.750420233351982e-06, "loss": 0.1778, "step": 10228 }, { "epoch": 0.73, "grad_norm": 3.588596090834798, "learning_rate": 1.7495377136097486e-06, "loss": 0.1371, "step": 10229 }, { "epoch": 0.73, "grad_norm": 4.589097976930378, "learning_rate": 1.7486553692176539e-06, "loss": 0.8808, "step": 10230 }, { "epoch": 0.73, "grad_norm": 6.421450370397516, "learning_rate": 1.7477732002232944e-06, "loss": 0.6349, "step": 10231 }, { "epoch": 0.73, "grad_norm": 3.671021138195882, "learning_rate": 1.7468912066742594e-06, "loss": 0.2041, "step": 10232 }, { "epoch": 0.73, "grad_norm": 13.83694227990024, "learning_rate": 1.7460093886181312e-06, "loss": 0.4872, "step": 10233 }, { "epoch": 0.73, "grad_norm": 13.465629463525808, "learning_rate": 1.7451277461024796e-06, "loss": 0.68, "step": 10234 }, { "epoch": 0.73, "grad_norm": 7.994040661432973, "learning_rate": 1.744246279174866e-06, "loss": 0.4278, "step": 10235 }, { "epoch": 0.73, "grad_norm": 2.3243520385863192, "learning_rate": 1.74336498788284e-06, "loss": 0.098, "step": 10236 }, { "epoch": 0.73, "grad_norm": 16.270663701498297, "learning_rate": 1.7424838722739478e-06, "loss": 1.0207, "step": 10237 }, { "epoch": 0.73, "grad_norm": 25.203712457387493, "learning_rate": 1.7416029323957207e-06, "loss": 1.1326, "step": 10238 }, { "epoch": 0.73, "grad_norm": 4.478484706736238, "learning_rate": 1.740722168295681e-06, "loss": 0.2392, "step": 10239 }, { "epoch": 0.73, "grad_norm": 3.0648907938304957, "learning_rate": 1.7398415800213436e-06, "loss": 0.1034, "step": 10240 }, { "epoch": 0.73, "grad_norm": 3.676890361071535, "learning_rate": 1.7389611676202107e-06, "loss": 0.1439, "step": 10241 }, { "epoch": 0.73, "grad_norm": 4.3454533497684364, "learning_rate": 1.7380809311397807e-06, "loss": 0.2357, "step": 10242 }, { "epoch": 0.73, "grad_norm": 10.797377492102685, "learning_rate": 1.7372008706275378e-06, "loss": 0.4807, "step": 10243 }, { "epoch": 0.73, "grad_norm": 3.439828505804822, "learning_rate": 1.7363209861309571e-06, "loss": 0.1383, "step": 10244 }, { "epoch": 0.73, "grad_norm": 33.16193289749751, "learning_rate": 1.7354412776975044e-06, "loss": 1.0033, "step": 10245 }, { "epoch": 0.73, "grad_norm": 4.8103700812170365, "learning_rate": 1.7345617453746388e-06, "loss": 0.8773, "step": 10246 }, { "epoch": 0.73, "grad_norm": 5.146767783417098, "learning_rate": 1.733682389209807e-06, "loss": 0.2708, "step": 10247 }, { "epoch": 0.73, "grad_norm": 20.392128534804197, "learning_rate": 1.7328032092504455e-06, "loss": 0.8484, "step": 10248 }, { "epoch": 0.73, "grad_norm": 18.17673200176472, "learning_rate": 1.7319242055439844e-06, "loss": 0.7522, "step": 10249 }, { "epoch": 0.74, "grad_norm": 3.312808453880608, "learning_rate": 1.7310453781378429e-06, "loss": 0.153, "step": 10250 }, { "epoch": 0.74, "grad_norm": 3.624523662942896, "learning_rate": 1.7301667270794293e-06, "loss": 0.1206, "step": 10251 }, { "epoch": 0.74, "grad_norm": 4.628713624097228, "learning_rate": 1.7292882524161435e-06, "loss": 0.1687, "step": 10252 }, { "epoch": 0.74, "grad_norm": 3.9228097477316113, "learning_rate": 1.728409954195377e-06, "loss": 0.2026, "step": 10253 }, { "epoch": 0.74, "grad_norm": 3.1483965935831413, "learning_rate": 1.7275318324645074e-06, "loss": 0.1289, "step": 10254 }, { "epoch": 0.74, "grad_norm": 4.97255149343406, "learning_rate": 1.7266538872709104e-06, "loss": 0.9466, "step": 10255 }, { "epoch": 0.74, "grad_norm": 3.338853650988144, "learning_rate": 1.725776118661946e-06, "loss": 0.1372, "step": 10256 }, { "epoch": 0.74, "grad_norm": 4.893099127196768, "learning_rate": 1.7248985266849644e-06, "loss": 0.1759, "step": 10257 }, { "epoch": 0.74, "grad_norm": 12.539637452042058, "learning_rate": 1.7240211113873123e-06, "loss": 0.5818, "step": 10258 }, { "epoch": 0.74, "grad_norm": 3.4579749685225334, "learning_rate": 1.7231438728163208e-06, "loss": 0.2073, "step": 10259 }, { "epoch": 0.74, "grad_norm": 8.645937029724289, "learning_rate": 1.7222668110193126e-06, "loss": 0.4184, "step": 10260 }, { "epoch": 0.74, "grad_norm": 13.503516408025304, "learning_rate": 1.721389926043604e-06, "loss": 0.5415, "step": 10261 }, { "epoch": 0.74, "grad_norm": 23.237085036201577, "learning_rate": 1.7205132179364992e-06, "loss": 0.7905, "step": 10262 }, { "epoch": 0.74, "grad_norm": 13.743721980498668, "learning_rate": 1.7196366867452923e-06, "loss": 0.481, "step": 10263 }, { "epoch": 0.74, "grad_norm": 3.783270916866699, "learning_rate": 1.7187603325172697e-06, "loss": 0.1884, "step": 10264 }, { "epoch": 0.74, "grad_norm": 3.427537289031876, "learning_rate": 1.7178841552997073e-06, "loss": 0.2129, "step": 10265 }, { "epoch": 0.74, "grad_norm": 24.93655857606423, "learning_rate": 1.7170081551398687e-06, "loss": 0.8334, "step": 10266 }, { "epoch": 0.74, "grad_norm": 46.54378164514651, "learning_rate": 1.7161323320850154e-06, "loss": 0.9266, "step": 10267 }, { "epoch": 0.74, "grad_norm": 10.74773879821657, "learning_rate": 1.7152566861823928e-06, "loss": 0.4182, "step": 10268 }, { "epoch": 0.74, "grad_norm": 11.715775100845725, "learning_rate": 1.7143812174792368e-06, "loss": 0.7312, "step": 10269 }, { "epoch": 0.74, "grad_norm": 4.617759653318615, "learning_rate": 1.7135059260227794e-06, "loss": 0.7871, "step": 10270 }, { "epoch": 0.74, "grad_norm": 4.38389672538335, "learning_rate": 1.7126308118602365e-06, "loss": 0.7222, "step": 10271 }, { "epoch": 0.74, "grad_norm": 11.249280768820896, "learning_rate": 1.7117558750388187e-06, "loss": 0.6699, "step": 10272 }, { "epoch": 0.74, "grad_norm": 5.503221460017465, "learning_rate": 1.7108811156057226e-06, "loss": 0.9471, "step": 10273 }, { "epoch": 0.74, "grad_norm": 4.358240289105805, "learning_rate": 1.7100065336081417e-06, "loss": 0.8314, "step": 10274 }, { "epoch": 0.74, "grad_norm": 3.5414032614549025, "learning_rate": 1.7091321290932556e-06, "loss": 0.1524, "step": 10275 }, { "epoch": 0.74, "grad_norm": 3.2737944916844945, "learning_rate": 1.7082579021082347e-06, "loss": 0.1823, "step": 10276 }, { "epoch": 0.74, "grad_norm": 3.9571412871557023, "learning_rate": 1.7073838527002396e-06, "loss": 0.6909, "step": 10277 }, { "epoch": 0.74, "grad_norm": 3.519371239143408, "learning_rate": 1.7065099809164205e-06, "loss": 0.1297, "step": 10278 }, { "epoch": 0.74, "grad_norm": 2.877688401770212, "learning_rate": 1.7056362868039234e-06, "loss": 0.1578, "step": 10279 }, { "epoch": 0.74, "grad_norm": 2.685006318351355, "learning_rate": 1.704762770409879e-06, "loss": 0.1021, "step": 10280 }, { "epoch": 0.74, "grad_norm": 19.182967697121285, "learning_rate": 1.7038894317814097e-06, "loss": 1.061, "step": 10281 }, { "epoch": 0.74, "grad_norm": 3.199741708170355, "learning_rate": 1.7030162709656273e-06, "loss": 0.1053, "step": 10282 }, { "epoch": 0.74, "grad_norm": 5.17312848126498, "learning_rate": 1.7021432880096395e-06, "loss": 0.8917, "step": 10283 }, { "epoch": 0.74, "grad_norm": 13.49009414913881, "learning_rate": 1.7012704829605375e-06, "loss": 0.6687, "step": 10284 }, { "epoch": 0.74, "grad_norm": 17.10938170057873, "learning_rate": 1.7003978558654055e-06, "loss": 0.599, "step": 10285 }, { "epoch": 0.74, "grad_norm": 3.6253716657749844, "learning_rate": 1.6995254067713214e-06, "loss": 0.2236, "step": 10286 }, { "epoch": 0.74, "grad_norm": 3.984799899050516, "learning_rate": 1.6986531357253493e-06, "loss": 0.1807, "step": 10287 }, { "epoch": 0.74, "grad_norm": 2.764782171316059, "learning_rate": 1.6977810427745417e-06, "loss": 0.1372, "step": 10288 }, { "epoch": 0.74, "grad_norm": 13.72247083892969, "learning_rate": 1.696909127965949e-06, "loss": 0.3027, "step": 10289 }, { "epoch": 0.74, "grad_norm": 2.9047245286309806, "learning_rate": 1.6960373913466054e-06, "loss": 0.1647, "step": 10290 }, { "epoch": 0.74, "grad_norm": 4.446475269919242, "learning_rate": 1.6951658329635373e-06, "loss": 0.1773, "step": 10291 }, { "epoch": 0.74, "grad_norm": 12.545209546424921, "learning_rate": 1.6942944528637644e-06, "loss": 0.3189, "step": 10292 }, { "epoch": 0.74, "grad_norm": 8.074083162705586, "learning_rate": 1.6934232510942933e-06, "loss": 0.2989, "step": 10293 }, { "epoch": 0.74, "grad_norm": 2.2353693494797433, "learning_rate": 1.6925522277021195e-06, "loss": 0.1292, "step": 10294 }, { "epoch": 0.74, "grad_norm": 8.781265764667408, "learning_rate": 1.6916813827342355e-06, "loss": 0.4867, "step": 10295 }, { "epoch": 0.74, "grad_norm": 3.4056660190645576, "learning_rate": 1.6908107162376187e-06, "loss": 0.1147, "step": 10296 }, { "epoch": 0.74, "grad_norm": 3.3891381695471394, "learning_rate": 1.6899402282592352e-06, "loss": 0.2091, "step": 10297 }, { "epoch": 0.74, "grad_norm": 2.693065758693978, "learning_rate": 1.6890699188460497e-06, "loss": 0.1363, "step": 10298 }, { "epoch": 0.74, "grad_norm": 3.1130596026973114, "learning_rate": 1.6881997880450108e-06, "loss": 0.1721, "step": 10299 }, { "epoch": 0.74, "grad_norm": 4.2421577068006995, "learning_rate": 1.6873298359030543e-06, "loss": 0.7427, "step": 10300 }, { "epoch": 0.74, "grad_norm": 3.7441468876057002, "learning_rate": 1.6864600624671156e-06, "loss": 0.2129, "step": 10301 }, { "epoch": 0.74, "grad_norm": 11.96563930469888, "learning_rate": 1.685590467784114e-06, "loss": 0.3733, "step": 10302 }, { "epoch": 0.74, "grad_norm": 4.62856508725038, "learning_rate": 1.6847210519009593e-06, "loss": 0.8237, "step": 10303 }, { "epoch": 0.74, "grad_norm": 3.268429137588091, "learning_rate": 1.6838518148645566e-06, "loss": 0.1587, "step": 10304 }, { "epoch": 0.74, "grad_norm": 2.7047660238875664, "learning_rate": 1.682982756721796e-06, "loss": 0.1361, "step": 10305 }, { "epoch": 0.74, "grad_norm": 6.14782343863057, "learning_rate": 1.682113877519559e-06, "loss": 0.2848, "step": 10306 }, { "epoch": 0.74, "grad_norm": 3.5478191933193703, "learning_rate": 1.6812451773047205e-06, "loss": 0.2125, "step": 10307 }, { "epoch": 0.74, "grad_norm": 20.694981303184207, "learning_rate": 1.6803766561241425e-06, "loss": 0.5442, "step": 10308 }, { "epoch": 0.74, "grad_norm": 2.565966575583242, "learning_rate": 1.6795083140246788e-06, "loss": 0.0981, "step": 10309 }, { "epoch": 0.74, "grad_norm": 10.541159210910996, "learning_rate": 1.6786401510531708e-06, "loss": 0.4331, "step": 10310 }, { "epoch": 0.74, "grad_norm": 17.643925261131226, "learning_rate": 1.6777721672564585e-06, "loss": 0.5739, "step": 10311 }, { "epoch": 0.74, "grad_norm": 3.480375222720595, "learning_rate": 1.6769043626813585e-06, "loss": 0.1941, "step": 10312 }, { "epoch": 0.74, "grad_norm": 4.863345210583974, "learning_rate": 1.6760367373746917e-06, "loss": 0.7787, "step": 10313 }, { "epoch": 0.74, "grad_norm": 3.2811255133480453, "learning_rate": 1.6751692913832608e-06, "loss": 0.1075, "step": 10314 }, { "epoch": 0.74, "grad_norm": 3.0995024059634044, "learning_rate": 1.674302024753861e-06, "loss": 0.1724, "step": 10315 }, { "epoch": 0.74, "grad_norm": 15.030255443406825, "learning_rate": 1.6734349375332775e-06, "loss": 0.8028, "step": 10316 }, { "epoch": 0.74, "grad_norm": 4.972574297442262, "learning_rate": 1.672568029768289e-06, "loss": 0.908, "step": 10317 }, { "epoch": 0.74, "grad_norm": 3.265266886651391, "learning_rate": 1.6717013015056598e-06, "loss": 0.1362, "step": 10318 }, { "epoch": 0.74, "grad_norm": 3.224169342405201, "learning_rate": 1.6708347527921459e-06, "loss": 0.1426, "step": 10319 }, { "epoch": 0.74, "grad_norm": 4.606104230184437, "learning_rate": 1.6699683836744973e-06, "loss": 0.7984, "step": 10320 }, { "epoch": 0.74, "grad_norm": 2.819774831019619, "learning_rate": 1.6691021941994489e-06, "loss": 0.1117, "step": 10321 }, { "epoch": 0.74, "grad_norm": 4.649124910048432, "learning_rate": 1.6682361844137279e-06, "loss": 0.8455, "step": 10322 }, { "epoch": 0.74, "grad_norm": 21.510511651206482, "learning_rate": 1.6673703543640562e-06, "loss": 0.8818, "step": 10323 }, { "epoch": 0.74, "grad_norm": 3.0124711903117043, "learning_rate": 1.6665047040971383e-06, "loss": 0.1131, "step": 10324 }, { "epoch": 0.74, "grad_norm": 14.39082781712359, "learning_rate": 1.665639233659671e-06, "loss": 0.7277, "step": 10325 }, { "epoch": 0.74, "grad_norm": 4.446679354269836, "learning_rate": 1.6647739430983483e-06, "loss": 0.6422, "step": 10326 }, { "epoch": 0.74, "grad_norm": 3.1110077298701864, "learning_rate": 1.6639088324598463e-06, "loss": 0.1947, "step": 10327 }, { "epoch": 0.74, "grad_norm": 4.193579311691185, "learning_rate": 1.663043901790834e-06, "loss": 0.8223, "step": 10328 }, { "epoch": 0.74, "grad_norm": 6.567545152714915, "learning_rate": 1.6621791511379732e-06, "loss": 0.3036, "step": 10329 }, { "epoch": 0.74, "grad_norm": 15.052194883771802, "learning_rate": 1.661314580547913e-06, "loss": 0.7147, "step": 10330 }, { "epoch": 0.74, "grad_norm": 4.203188938841879, "learning_rate": 1.660450190067292e-06, "loss": 0.2251, "step": 10331 }, { "epoch": 0.74, "grad_norm": 4.512091110527563, "learning_rate": 1.6595859797427437e-06, "loss": 0.1115, "step": 10332 }, { "epoch": 0.74, "grad_norm": 2.8702616729713024, "learning_rate": 1.6587219496208879e-06, "loss": 0.124, "step": 10333 }, { "epoch": 0.74, "grad_norm": 18.73245616772204, "learning_rate": 1.657858099748334e-06, "loss": 0.5741, "step": 10334 }, { "epoch": 0.74, "grad_norm": 2.7626450478373807, "learning_rate": 1.6569944301716877e-06, "loss": 0.1382, "step": 10335 }, { "epoch": 0.74, "grad_norm": 8.973837292780813, "learning_rate": 1.656130940937537e-06, "loss": 0.4502, "step": 10336 }, { "epoch": 0.74, "grad_norm": 2.771436022697406, "learning_rate": 1.6552676320924627e-06, "loss": 0.1362, "step": 10337 }, { "epoch": 0.74, "grad_norm": 2.827397295917479, "learning_rate": 1.654404503683041e-06, "loss": 0.1574, "step": 10338 }, { "epoch": 0.74, "grad_norm": 11.171889749701384, "learning_rate": 1.653541555755832e-06, "loss": 0.3234, "step": 10339 }, { "epoch": 0.74, "grad_norm": 16.787444865522513, "learning_rate": 1.652678788357388e-06, "loss": 0.8132, "step": 10340 }, { "epoch": 0.74, "grad_norm": 4.249295546760046, "learning_rate": 1.6518162015342542e-06, "loss": 0.6331, "step": 10341 }, { "epoch": 0.74, "grad_norm": 4.724417133600386, "learning_rate": 1.6509537953329635e-06, "loss": 0.9387, "step": 10342 }, { "epoch": 0.74, "grad_norm": 11.41053177574208, "learning_rate": 1.650091569800038e-06, "loss": 0.6708, "step": 10343 }, { "epoch": 0.74, "grad_norm": 3.413808086989411, "learning_rate": 1.649229524981991e-06, "loss": 0.0886, "step": 10344 }, { "epoch": 0.74, "grad_norm": 16.53798598628024, "learning_rate": 1.6483676609253297e-06, "loss": 1.0616, "step": 10345 }, { "epoch": 0.74, "grad_norm": 12.266051194342744, "learning_rate": 1.6475059776765456e-06, "loss": 0.8908, "step": 10346 }, { "epoch": 0.74, "grad_norm": 3.541973617341928, "learning_rate": 1.6466444752821248e-06, "loss": 0.1378, "step": 10347 }, { "epoch": 0.74, "grad_norm": 3.5189166742713973, "learning_rate": 1.645783153788541e-06, "loss": 0.2588, "step": 10348 }, { "epoch": 0.74, "grad_norm": 3.5142460320564517, "learning_rate": 1.6449220132422582e-06, "loss": 0.2067, "step": 10349 }, { "epoch": 0.74, "grad_norm": 28.327340271423896, "learning_rate": 1.644061053689735e-06, "loss": 0.4277, "step": 10350 }, { "epoch": 0.74, "grad_norm": 4.219225500951752, "learning_rate": 1.6432002751774151e-06, "loss": 0.7387, "step": 10351 }, { "epoch": 0.74, "grad_norm": 4.6790814941065735, "learning_rate": 1.6423396777517337e-06, "loss": 0.1958, "step": 10352 }, { "epoch": 0.74, "grad_norm": 3.859321620439286, "learning_rate": 1.6414792614591157e-06, "loss": 0.1715, "step": 10353 }, { "epoch": 0.74, "grad_norm": 2.960221162389875, "learning_rate": 1.640619026345981e-06, "loss": 0.1211, "step": 10354 }, { "epoch": 0.74, "grad_norm": 3.112603861475817, "learning_rate": 1.6397589724587331e-06, "loss": 0.1502, "step": 10355 }, { "epoch": 0.74, "grad_norm": 15.747955755105547, "learning_rate": 1.6388990998437682e-06, "loss": 1.0771, "step": 10356 }, { "epoch": 0.74, "grad_norm": 3.2947227002904613, "learning_rate": 1.638039408547476e-06, "loss": 0.1664, "step": 10357 }, { "epoch": 0.74, "grad_norm": 4.99243673481708, "learning_rate": 1.6371798986162325e-06, "loss": 1.0475, "step": 10358 }, { "epoch": 0.74, "grad_norm": 3.608021047336959, "learning_rate": 1.6363205700964036e-06, "loss": 0.1818, "step": 10359 }, { "epoch": 0.74, "grad_norm": 13.372639025098119, "learning_rate": 1.6354614230343484e-06, "loss": 0.9108, "step": 10360 }, { "epoch": 0.74, "grad_norm": 2.812994602005497, "learning_rate": 1.6346024574764136e-06, "loss": 0.1472, "step": 10361 }, { "epoch": 0.74, "grad_norm": 42.66841312007523, "learning_rate": 1.6337436734689361e-06, "loss": 0.9408, "step": 10362 }, { "epoch": 0.74, "grad_norm": 3.592920869170344, "learning_rate": 1.6328850710582467e-06, "loss": 0.1318, "step": 10363 }, { "epoch": 0.74, "grad_norm": 4.75358739253702, "learning_rate": 1.6320266502906629e-06, "loss": 0.2206, "step": 10364 }, { "epoch": 0.74, "grad_norm": 3.5933130565887144, "learning_rate": 1.631168411212491e-06, "loss": 0.1572, "step": 10365 }, { "epoch": 0.74, "grad_norm": 3.334598947729408, "learning_rate": 1.6303103538700331e-06, "loss": 0.1191, "step": 10366 }, { "epoch": 0.74, "grad_norm": 3.330616608091637, "learning_rate": 1.6294524783095772e-06, "loss": 0.1593, "step": 10367 }, { "epoch": 0.74, "grad_norm": 3.8415041283609095, "learning_rate": 1.6285947845773998e-06, "loss": 0.2277, "step": 10368 }, { "epoch": 0.74, "grad_norm": 4.183179062350343, "learning_rate": 1.6277372727197739e-06, "loss": 0.1275, "step": 10369 }, { "epoch": 0.74, "grad_norm": 3.0707588219220043, "learning_rate": 1.626879942782958e-06, "loss": 0.0892, "step": 10370 }, { "epoch": 0.74, "grad_norm": 3.8791597501239696, "learning_rate": 1.6260227948132006e-06, "loss": 0.2125, "step": 10371 }, { "epoch": 0.74, "grad_norm": 4.7006728286991635, "learning_rate": 1.6251658288567423e-06, "loss": 0.738, "step": 10372 }, { "epoch": 0.74, "grad_norm": 3.7371003488980445, "learning_rate": 1.624309044959813e-06, "loss": 0.1289, "step": 10373 }, { "epoch": 0.74, "grad_norm": 3.1821649739079088, "learning_rate": 1.6234524431686315e-06, "loss": 0.084, "step": 10374 }, { "epoch": 0.74, "grad_norm": 4.844149448839117, "learning_rate": 1.622596023529412e-06, "loss": 0.1353, "step": 10375 }, { "epoch": 0.74, "grad_norm": 15.359361281783315, "learning_rate": 1.621739786088352e-06, "loss": 0.8869, "step": 10376 }, { "epoch": 0.74, "grad_norm": 3.7510315080621215, "learning_rate": 1.620883730891642e-06, "loss": 0.2077, "step": 10377 }, { "epoch": 0.74, "grad_norm": 17.46108224924886, "learning_rate": 1.6200278579854655e-06, "loss": 0.7537, "step": 10378 }, { "epoch": 0.74, "grad_norm": 37.70776042671991, "learning_rate": 1.6191721674159921e-06, "loss": 0.5356, "step": 10379 }, { "epoch": 0.74, "grad_norm": 3.576540645047738, "learning_rate": 1.6183166592293837e-06, "loss": 0.1258, "step": 10380 }, { "epoch": 0.74, "grad_norm": 2.997913408381967, "learning_rate": 1.6174613334717893e-06, "loss": 0.1959, "step": 10381 }, { "epoch": 0.74, "grad_norm": 3.0651986329555587, "learning_rate": 1.6166061901893537e-06, "loss": 0.1229, "step": 10382 }, { "epoch": 0.74, "grad_norm": 2.868475110531536, "learning_rate": 1.6157512294282074e-06, "loss": 0.11, "step": 10383 }, { "epoch": 0.74, "grad_norm": 11.487508574571988, "learning_rate": 1.6148964512344722e-06, "loss": 0.7346, "step": 10384 }, { "epoch": 0.74, "grad_norm": 3.936969126925137, "learning_rate": 1.6140418556542597e-06, "loss": 0.7437, "step": 10385 }, { "epoch": 0.74, "grad_norm": 2.7016130450391787, "learning_rate": 1.6131874427336714e-06, "loss": 0.1928, "step": 10386 }, { "epoch": 0.74, "grad_norm": 4.43488240189037, "learning_rate": 1.6123332125188024e-06, "loss": 0.8065, "step": 10387 }, { "epoch": 0.74, "grad_norm": 11.105078975332951, "learning_rate": 1.6114791650557338e-06, "loss": 0.3236, "step": 10388 }, { "epoch": 0.74, "grad_norm": 2.9743315623051974, "learning_rate": 1.6106253003905376e-06, "loss": 0.1908, "step": 10389 }, { "epoch": 0.75, "grad_norm": 15.392811458963111, "learning_rate": 1.6097716185692758e-06, "loss": 0.9254, "step": 10390 }, { "epoch": 0.75, "grad_norm": 4.77790837662395, "learning_rate": 1.6089181196380034e-06, "loss": 0.728, "step": 10391 }, { "epoch": 0.75, "grad_norm": 10.201193525086397, "learning_rate": 1.6080648036427631e-06, "loss": 0.4347, "step": 10392 }, { "epoch": 0.75, "grad_norm": 2.205662828124759, "learning_rate": 1.6072116706295854e-06, "loss": 0.1139, "step": 10393 }, { "epoch": 0.75, "grad_norm": 3.572690889070111, "learning_rate": 1.6063587206444976e-06, "loss": 0.2664, "step": 10394 }, { "epoch": 0.75, "grad_norm": 2.8583785778266715, "learning_rate": 1.6055059537335128e-06, "loss": 0.1623, "step": 10395 }, { "epoch": 0.75, "grad_norm": 3.364620915438358, "learning_rate": 1.6046533699426297e-06, "loss": 0.5706, "step": 10396 }, { "epoch": 0.75, "grad_norm": 3.924039238379069, "learning_rate": 1.6038009693178464e-06, "loss": 0.7809, "step": 10397 }, { "epoch": 0.75, "grad_norm": 2.0616214343770087, "learning_rate": 1.6029487519051462e-06, "loss": 0.0724, "step": 10398 }, { "epoch": 0.75, "grad_norm": 2.3923875025439982, "learning_rate": 1.6020967177505003e-06, "loss": 0.0982, "step": 10399 }, { "epoch": 0.75, "grad_norm": 3.288291096559981, "learning_rate": 1.6012448668998759e-06, "loss": 0.1429, "step": 10400 }, { "epoch": 0.75, "grad_norm": 4.0124805785008935, "learning_rate": 1.6003931993992267e-06, "loss": 0.1331, "step": 10401 }, { "epoch": 0.75, "grad_norm": 2.6903669875162395, "learning_rate": 1.5995417152944942e-06, "loss": 0.1363, "step": 10402 }, { "epoch": 0.75, "grad_norm": 13.890388407411173, "learning_rate": 1.5986904146316162e-06, "loss": 0.6158, "step": 10403 }, { "epoch": 0.75, "grad_norm": 12.858326067617238, "learning_rate": 1.597839297456516e-06, "loss": 0.5397, "step": 10404 }, { "epoch": 0.75, "grad_norm": 3.5088335551065564, "learning_rate": 1.5969883638151063e-06, "loss": 0.1344, "step": 10405 }, { "epoch": 0.75, "grad_norm": 11.344364302932261, "learning_rate": 1.5961376137532946e-06, "loss": 0.4646, "step": 10406 }, { "epoch": 0.75, "grad_norm": 15.460389289264661, "learning_rate": 1.5952870473169757e-06, "loss": 0.7345, "step": 10407 }, { "epoch": 0.75, "grad_norm": 3.588981379792408, "learning_rate": 1.5944366645520304e-06, "loss": 0.1407, "step": 10408 }, { "epoch": 0.75, "grad_norm": 4.928089261797033, "learning_rate": 1.5935864655043376e-06, "loss": 0.7897, "step": 10409 }, { "epoch": 0.75, "grad_norm": 10.494315763395344, "learning_rate": 1.592736450219761e-06, "loss": 0.5272, "step": 10410 }, { "epoch": 0.75, "grad_norm": 2.6754566079931497, "learning_rate": 1.5918866187441546e-06, "loss": 0.119, "step": 10411 }, { "epoch": 0.75, "grad_norm": 4.162180707819621, "learning_rate": 1.5910369711233653e-06, "loss": 0.1111, "step": 10412 }, { "epoch": 0.75, "grad_norm": 3.666808412029571, "learning_rate": 1.5901875074032285e-06, "loss": 0.2568, "step": 10413 }, { "epoch": 0.75, "grad_norm": 3.376631540966282, "learning_rate": 1.5893382276295672e-06, "loss": 0.1215, "step": 10414 }, { "epoch": 0.75, "grad_norm": 7.606239692232317, "learning_rate": 1.5884891318482e-06, "loss": 0.2952, "step": 10415 }, { "epoch": 0.75, "grad_norm": 2.9856737240089952, "learning_rate": 1.5876402201049307e-06, "loss": 0.1917, "step": 10416 }, { "epoch": 0.75, "grad_norm": 4.792743707028027, "learning_rate": 1.5867914924455553e-06, "loss": 0.9044, "step": 10417 }, { "epoch": 0.75, "grad_norm": 12.436628489284901, "learning_rate": 1.5859429489158574e-06, "loss": 0.7952, "step": 10418 }, { "epoch": 0.75, "grad_norm": 4.9917434585307445, "learning_rate": 1.5850945895616176e-06, "loss": 0.1474, "step": 10419 }, { "epoch": 0.75, "grad_norm": 3.689157854468442, "learning_rate": 1.5842464144285957e-06, "loss": 0.1217, "step": 10420 }, { "epoch": 0.75, "grad_norm": 5.477650870637154, "learning_rate": 1.5833984235625515e-06, "loss": 0.1943, "step": 10421 }, { "epoch": 0.75, "grad_norm": 11.368764663358434, "learning_rate": 1.5825506170092303e-06, "loss": 0.5942, "step": 10422 }, { "epoch": 0.75, "grad_norm": 2.414349765576141, "learning_rate": 1.5817029948143674e-06, "loss": 0.113, "step": 10423 }, { "epoch": 0.75, "grad_norm": 3.2124316124096155, "learning_rate": 1.5808555570236877e-06, "loss": 0.1759, "step": 10424 }, { "epoch": 0.75, "grad_norm": 4.019659606822016, "learning_rate": 1.5800083036829096e-06, "loss": 0.3289, "step": 10425 }, { "epoch": 0.75, "grad_norm": 29.532543990070167, "learning_rate": 1.5791612348377383e-06, "loss": 0.9283, "step": 10426 }, { "epoch": 0.75, "grad_norm": 3.5143340509930723, "learning_rate": 1.5783143505338683e-06, "loss": 0.152, "step": 10427 }, { "epoch": 0.75, "grad_norm": 3.4757861830932106, "learning_rate": 1.577467650816989e-06, "loss": 0.2507, "step": 10428 }, { "epoch": 0.75, "grad_norm": 4.509826751131623, "learning_rate": 1.576621135732775e-06, "loss": 0.1764, "step": 10429 }, { "epoch": 0.75, "grad_norm": 3.040825852615364, "learning_rate": 1.5757748053268905e-06, "loss": 0.1593, "step": 10430 }, { "epoch": 0.75, "grad_norm": 20.72525305487804, "learning_rate": 1.5749286596449976e-06, "loss": 0.4079, "step": 10431 }, { "epoch": 0.75, "grad_norm": 3.46599722225734, "learning_rate": 1.5740826987327369e-06, "loss": 0.1502, "step": 10432 }, { "epoch": 0.75, "grad_norm": 3.5469979155172813, "learning_rate": 1.5732369226357452e-06, "loss": 0.1259, "step": 10433 }, { "epoch": 0.75, "grad_norm": 11.337626939258245, "learning_rate": 1.5723913313996526e-06, "loss": 0.3272, "step": 10434 }, { "epoch": 0.75, "grad_norm": 3.0177393958892087, "learning_rate": 1.5715459250700727e-06, "loss": 0.1419, "step": 10435 }, { "epoch": 0.75, "grad_norm": 13.130834901767374, "learning_rate": 1.5707007036926114e-06, "loss": 0.4187, "step": 10436 }, { "epoch": 0.75, "grad_norm": 11.65532794861083, "learning_rate": 1.5698556673128684e-06, "loss": 0.4308, "step": 10437 }, { "epoch": 0.75, "grad_norm": 29.099515431973334, "learning_rate": 1.5690108159764283e-06, "loss": 1.2066, "step": 10438 }, { "epoch": 0.75, "grad_norm": 3.940850175798656, "learning_rate": 1.568166149728866e-06, "loss": 0.1572, "step": 10439 }, { "epoch": 0.75, "grad_norm": 3.8292448975749824, "learning_rate": 1.5673216686157517e-06, "loss": 0.2164, "step": 10440 }, { "epoch": 0.75, "grad_norm": 5.522788478427724, "learning_rate": 1.5664773726826393e-06, "loss": 0.9922, "step": 10441 }, { "epoch": 0.75, "grad_norm": 23.756018420361244, "learning_rate": 1.5656332619750746e-06, "loss": 0.969, "step": 10442 }, { "epoch": 0.75, "grad_norm": 4.7082283881877895, "learning_rate": 1.5647893365385991e-06, "loss": 0.7423, "step": 10443 }, { "epoch": 0.75, "grad_norm": 25.811508338312716, "learning_rate": 1.5639455964187345e-06, "loss": 0.8995, "step": 10444 }, { "epoch": 0.75, "grad_norm": 3.9656677409276315, "learning_rate": 1.5631020416609966e-06, "loss": 0.2425, "step": 10445 }, { "epoch": 0.75, "grad_norm": 12.470411076317557, "learning_rate": 1.5622586723108963e-06, "loss": 0.4682, "step": 10446 }, { "epoch": 0.75, "grad_norm": 3.1022941328936287, "learning_rate": 1.5614154884139281e-06, "loss": 0.157, "step": 10447 }, { "epoch": 0.75, "grad_norm": 4.195729682954328, "learning_rate": 1.5605724900155766e-06, "loss": 0.246, "step": 10448 }, { "epoch": 0.75, "grad_norm": 3.980088280311471, "learning_rate": 1.559729677161322e-06, "loss": 0.2221, "step": 10449 }, { "epoch": 0.75, "grad_norm": 3.512820744956742, "learning_rate": 1.558887049896629e-06, "loss": 0.1999, "step": 10450 }, { "epoch": 0.75, "grad_norm": 3.208442538308042, "learning_rate": 1.5580446082669532e-06, "loss": 0.1603, "step": 10451 }, { "epoch": 0.75, "grad_norm": 3.13947269725815, "learning_rate": 1.5572023523177438e-06, "loss": 0.192, "step": 10452 }, { "epoch": 0.75, "grad_norm": 3.75362657505968, "learning_rate": 1.5563602820944357e-06, "loss": 0.698, "step": 10453 }, { "epoch": 0.75, "grad_norm": 4.319460429740921, "learning_rate": 1.5555183976424553e-06, "loss": 0.8569, "step": 10454 }, { "epoch": 0.75, "grad_norm": 19.214876182703556, "learning_rate": 1.5546766990072176e-06, "loss": 1.0511, "step": 10455 }, { "epoch": 0.75, "grad_norm": 4.210898240806304, "learning_rate": 1.5538351862341345e-06, "loss": 0.1415, "step": 10456 }, { "epoch": 0.75, "grad_norm": 3.3632000502849433, "learning_rate": 1.5529938593685951e-06, "loss": 0.1236, "step": 10457 }, { "epoch": 0.75, "grad_norm": 3.117584378953446, "learning_rate": 1.5521527184559908e-06, "loss": 0.1127, "step": 10458 }, { "epoch": 0.75, "grad_norm": 4.664631738373832, "learning_rate": 1.551311763541697e-06, "loss": 0.263, "step": 10459 }, { "epoch": 0.75, "grad_norm": 3.473119569384821, "learning_rate": 1.5504709946710795e-06, "loss": 0.0992, "step": 10460 }, { "epoch": 0.75, "grad_norm": 5.169272401113344, "learning_rate": 1.5496304118894922e-06, "loss": 0.7378, "step": 10461 }, { "epoch": 0.75, "grad_norm": 6.914618512416904, "learning_rate": 1.5487900152422864e-06, "loss": 0.3842, "step": 10462 }, { "epoch": 0.75, "grad_norm": 3.9927181887502745, "learning_rate": 1.5479498047747953e-06, "loss": 0.9356, "step": 10463 }, { "epoch": 0.75, "grad_norm": 3.666243210341405, "learning_rate": 1.5471097805323442e-06, "loss": 0.7634, "step": 10464 }, { "epoch": 0.75, "grad_norm": 2.6127570640560305, "learning_rate": 1.5462699425602517e-06, "loss": 0.1249, "step": 10465 }, { "epoch": 0.75, "grad_norm": 3.5280852862658625, "learning_rate": 1.5454302909038232e-06, "loss": 0.2116, "step": 10466 }, { "epoch": 0.75, "grad_norm": 5.6521152744536165, "learning_rate": 1.5445908256083525e-06, "loss": 0.8213, "step": 10467 }, { "epoch": 0.75, "grad_norm": 1.922542437708379, "learning_rate": 1.5437515467191305e-06, "loss": 0.0805, "step": 10468 }, { "epoch": 0.75, "grad_norm": 2.1408721688148193, "learning_rate": 1.5429124542814283e-06, "loss": 0.0885, "step": 10469 }, { "epoch": 0.75, "grad_norm": 2.725624355591029, "learning_rate": 1.5420735483405115e-06, "loss": 0.1793, "step": 10470 }, { "epoch": 0.75, "grad_norm": 2.3438930255352277, "learning_rate": 1.5412348289416396e-06, "loss": 0.0942, "step": 10471 }, { "epoch": 0.75, "grad_norm": 8.884769219990341, "learning_rate": 1.5403962961300566e-06, "loss": 0.4413, "step": 10472 }, { "epoch": 0.75, "grad_norm": 2.2419459250680984, "learning_rate": 1.5395579499509966e-06, "loss": 0.1026, "step": 10473 }, { "epoch": 0.75, "grad_norm": 20.065699928973004, "learning_rate": 1.5387197904496887e-06, "loss": 0.9203, "step": 10474 }, { "epoch": 0.75, "grad_norm": 10.80655298621421, "learning_rate": 1.5378818176713455e-06, "loss": 0.4449, "step": 10475 }, { "epoch": 0.75, "grad_norm": 3.436099437850214, "learning_rate": 1.5370440316611723e-06, "loss": 0.3202, "step": 10476 }, { "epoch": 0.75, "grad_norm": 3.0808450582542393, "learning_rate": 1.5362064324643676e-06, "loss": 0.1091, "step": 10477 }, { "epoch": 0.75, "grad_norm": 3.752081747880478, "learning_rate": 1.535369020126114e-06, "loss": 0.7433, "step": 10478 }, { "epoch": 0.75, "grad_norm": 16.062691328880568, "learning_rate": 1.5345317946915867e-06, "loss": 0.7915, "step": 10479 }, { "epoch": 0.75, "grad_norm": 3.0865379756306637, "learning_rate": 1.533694756205954e-06, "loss": 0.1594, "step": 10480 }, { "epoch": 0.75, "grad_norm": 10.379854381935996, "learning_rate": 1.5328579047143672e-06, "loss": 1.0988, "step": 10481 }, { "epoch": 0.75, "grad_norm": 4.056189199436716, "learning_rate": 1.5320212402619704e-06, "loss": 0.8113, "step": 10482 }, { "epoch": 0.75, "grad_norm": 14.401152450158694, "learning_rate": 1.5311847628939031e-06, "loss": 0.7897, "step": 10483 }, { "epoch": 0.75, "grad_norm": 3.094451273114595, "learning_rate": 1.5303484726552876e-06, "loss": 0.1279, "step": 10484 }, { "epoch": 0.75, "grad_norm": 4.136077346319301, "learning_rate": 1.5295123695912366e-06, "loss": 0.2268, "step": 10485 }, { "epoch": 0.75, "grad_norm": 4.093420565720189, "learning_rate": 1.5286764537468585e-06, "loss": 0.1607, "step": 10486 }, { "epoch": 0.75, "grad_norm": 3.757737357039967, "learning_rate": 1.527840725167246e-06, "loss": 0.1069, "step": 10487 }, { "epoch": 0.75, "grad_norm": 3.5386975023369205, "learning_rate": 1.5270051838974837e-06, "loss": 0.1149, "step": 10488 }, { "epoch": 0.75, "grad_norm": 2.904586821926373, "learning_rate": 1.5261698299826444e-06, "loss": 0.0949, "step": 10489 }, { "epoch": 0.75, "grad_norm": 3.9233088231419644, "learning_rate": 1.5253346634677946e-06, "loss": 0.1838, "step": 10490 }, { "epoch": 0.75, "grad_norm": 8.314819269409343, "learning_rate": 1.524499684397988e-06, "loss": 0.339, "step": 10491 }, { "epoch": 0.75, "grad_norm": 4.356146698291222, "learning_rate": 1.5236648928182673e-06, "loss": 0.7302, "step": 10492 }, { "epoch": 0.75, "grad_norm": 15.004704235187805, "learning_rate": 1.5228302887736674e-06, "loss": 0.6172, "step": 10493 }, { "epoch": 0.75, "grad_norm": 4.911595602401472, "learning_rate": 1.5219958723092099e-06, "loss": 1.0191, "step": 10494 }, { "epoch": 0.75, "grad_norm": 3.1634085096290665, "learning_rate": 1.5211616434699117e-06, "loss": 0.1614, "step": 10495 }, { "epoch": 0.75, "grad_norm": 3.981192541687538, "learning_rate": 1.520327602300775e-06, "loss": 0.6316, "step": 10496 }, { "epoch": 0.75, "grad_norm": 14.332628044993964, "learning_rate": 1.5194937488467926e-06, "loss": 0.8025, "step": 10497 }, { "epoch": 0.75, "grad_norm": 22.680640488408656, "learning_rate": 1.5186600831529463e-06, "loss": 0.6339, "step": 10498 }, { "epoch": 0.75, "grad_norm": 9.1402524354708, "learning_rate": 1.5178266052642132e-06, "loss": 0.5731, "step": 10499 }, { "epoch": 0.75, "grad_norm": 29.965827802906766, "learning_rate": 1.5169933152255535e-06, "loss": 1.049, "step": 10500 }, { "epoch": 0.75, "grad_norm": 3.0451194226975007, "learning_rate": 1.5161602130819186e-06, "loss": 0.1038, "step": 10501 }, { "epoch": 0.75, "grad_norm": 21.03173616526007, "learning_rate": 1.5153272988782553e-06, "loss": 0.7079, "step": 10502 }, { "epoch": 0.75, "grad_norm": 4.421382178691701, "learning_rate": 1.514494572659494e-06, "loss": 0.7837, "step": 10503 }, { "epoch": 0.75, "grad_norm": 19.187444880714985, "learning_rate": 1.5136620344705566e-06, "loss": 0.8081, "step": 10504 }, { "epoch": 0.75, "grad_norm": 4.35881407321841, "learning_rate": 1.5128296843563568e-06, "loss": 0.7545, "step": 10505 }, { "epoch": 0.75, "grad_norm": 21.56722935585993, "learning_rate": 1.5119975223617949e-06, "loss": 1.0669, "step": 10506 }, { "epoch": 0.75, "grad_norm": 4.329873720684628, "learning_rate": 1.5111655485317628e-06, "loss": 0.7332, "step": 10507 }, { "epoch": 0.75, "grad_norm": 3.3884757108044044, "learning_rate": 1.5103337629111453e-06, "loss": 0.134, "step": 10508 }, { "epoch": 0.75, "grad_norm": 3.3583792244746773, "learning_rate": 1.5095021655448116e-06, "loss": 0.1604, "step": 10509 }, { "epoch": 0.75, "grad_norm": 3.669207123738959, "learning_rate": 1.5086707564776227e-06, "loss": 0.1111, "step": 10510 }, { "epoch": 0.75, "grad_norm": 11.949705104307744, "learning_rate": 1.5078395357544323e-06, "loss": 0.9785, "step": 10511 }, { "epoch": 0.75, "grad_norm": 3.216290736015354, "learning_rate": 1.5070085034200804e-06, "loss": 0.1827, "step": 10512 }, { "epoch": 0.75, "grad_norm": 25.576640877948236, "learning_rate": 1.5061776595193966e-06, "loss": 1.0976, "step": 10513 }, { "epoch": 0.75, "grad_norm": 13.641593671856814, "learning_rate": 1.5053470040972051e-06, "loss": 0.7555, "step": 10514 }, { "epoch": 0.75, "grad_norm": 3.4291217717584326, "learning_rate": 1.5045165371983145e-06, "loss": 0.1322, "step": 10515 }, { "epoch": 0.75, "grad_norm": 21.390146746919925, "learning_rate": 1.5036862588675254e-06, "loss": 0.8113, "step": 10516 }, { "epoch": 0.75, "grad_norm": 14.074294788682616, "learning_rate": 1.5028561691496286e-06, "loss": 0.6871, "step": 10517 }, { "epoch": 0.75, "grad_norm": 3.8615111155992796, "learning_rate": 1.5020262680894043e-06, "loss": 0.7024, "step": 10518 }, { "epoch": 0.75, "grad_norm": 31.111436913682073, "learning_rate": 1.5011965557316199e-06, "loss": 0.5306, "step": 10519 }, { "epoch": 0.75, "grad_norm": 9.897330327804726, "learning_rate": 1.50036703212104e-06, "loss": 0.3905, "step": 10520 }, { "epoch": 0.75, "grad_norm": 3.331898998940974, "learning_rate": 1.4995376973024118e-06, "loss": 0.2278, "step": 10521 }, { "epoch": 0.75, "grad_norm": 2.8616585478457592, "learning_rate": 1.498708551320473e-06, "loss": 0.1623, "step": 10522 }, { "epoch": 0.75, "grad_norm": 13.508768001587406, "learning_rate": 1.4978795942199564e-06, "loss": 0.4036, "step": 10523 }, { "epoch": 0.75, "grad_norm": 2.8059812142490297, "learning_rate": 1.4970508260455796e-06, "loss": 0.1816, "step": 10524 }, { "epoch": 0.75, "grad_norm": 19.91965459402314, "learning_rate": 1.496222246842051e-06, "loss": 0.7049, "step": 10525 }, { "epoch": 0.75, "grad_norm": 3.246327472776704, "learning_rate": 1.4953938566540683e-06, "loss": 0.128, "step": 10526 }, { "epoch": 0.75, "grad_norm": 15.27463810372922, "learning_rate": 1.4945656555263233e-06, "loss": 0.535, "step": 10527 }, { "epoch": 0.75, "grad_norm": 4.774583610172025, "learning_rate": 1.4937376435034917e-06, "loss": 0.8118, "step": 10528 }, { "epoch": 0.76, "grad_norm": 17.824903261329112, "learning_rate": 1.4929098206302429e-06, "loss": 0.4534, "step": 10529 }, { "epoch": 0.76, "grad_norm": 3.3088385279729957, "learning_rate": 1.4920821869512337e-06, "loss": 0.2266, "step": 10530 }, { "epoch": 0.76, "grad_norm": 2.609782652371177, "learning_rate": 1.4912547425111113e-06, "loss": 0.1339, "step": 10531 }, { "epoch": 0.76, "grad_norm": 8.751280361562163, "learning_rate": 1.4904274873545155e-06, "loss": 0.3906, "step": 10532 }, { "epoch": 0.76, "grad_norm": 2.814663878955241, "learning_rate": 1.4896004215260723e-06, "loss": 0.1193, "step": 10533 }, { "epoch": 0.76, "grad_norm": 4.280874950637123, "learning_rate": 1.488773545070399e-06, "loss": 0.7575, "step": 10534 }, { "epoch": 0.76, "grad_norm": 5.166150775351717, "learning_rate": 1.4879468580321006e-06, "loss": 0.1828, "step": 10535 }, { "epoch": 0.76, "grad_norm": 1.864834597149628, "learning_rate": 1.4871203604557767e-06, "loss": 0.1012, "step": 10536 }, { "epoch": 0.76, "grad_norm": 3.023363081710344, "learning_rate": 1.486294052386012e-06, "loss": 0.1558, "step": 10537 }, { "epoch": 0.76, "grad_norm": 11.404029357244635, "learning_rate": 1.4854679338673821e-06, "loss": 0.6682, "step": 10538 }, { "epoch": 0.76, "grad_norm": 8.954812207284743, "learning_rate": 1.4846420049444553e-06, "loss": 0.5074, "step": 10539 }, { "epoch": 0.76, "grad_norm": 2.686106567624104, "learning_rate": 1.4838162656617878e-06, "loss": 0.166, "step": 10540 }, { "epoch": 0.76, "grad_norm": 3.86699392507079, "learning_rate": 1.48299071606392e-06, "loss": 0.2025, "step": 10541 }, { "epoch": 0.76, "grad_norm": 29.224620438403086, "learning_rate": 1.482165356195392e-06, "loss": 0.4652, "step": 10542 }, { "epoch": 0.76, "grad_norm": 2.9695889385534984, "learning_rate": 1.4813401861007277e-06, "loss": 0.1343, "step": 10543 }, { "epoch": 0.76, "grad_norm": 2.58142983627663, "learning_rate": 1.4805152058244392e-06, "loss": 0.0994, "step": 10544 }, { "epoch": 0.76, "grad_norm": 27.89968893361345, "learning_rate": 1.4796904154110353e-06, "loss": 0.6606, "step": 10545 }, { "epoch": 0.76, "grad_norm": 6.484070908091142, "learning_rate": 1.4788658149050083e-06, "loss": 0.0913, "step": 10546 }, { "epoch": 0.76, "grad_norm": 4.649111652231298, "learning_rate": 1.4780414043508412e-06, "loss": 0.3305, "step": 10547 }, { "epoch": 0.76, "grad_norm": 13.99292267491419, "learning_rate": 1.4772171837930104e-06, "loss": 0.4716, "step": 10548 }, { "epoch": 0.76, "grad_norm": 2.889421317336539, "learning_rate": 1.476393153275978e-06, "loss": 0.165, "step": 10549 }, { "epoch": 0.76, "grad_norm": 4.459554204891358, "learning_rate": 1.4755693128441956e-06, "loss": 0.754, "step": 10550 }, { "epoch": 0.76, "grad_norm": 4.859421428130851, "learning_rate": 1.4747456625421096e-06, "loss": 0.917, "step": 10551 }, { "epoch": 0.76, "grad_norm": 13.577268293408096, "learning_rate": 1.4739222024141536e-06, "loss": 0.3778, "step": 10552 }, { "epoch": 0.76, "grad_norm": 27.512286178792436, "learning_rate": 1.473098932504744e-06, "loss": 1.0344, "step": 10553 }, { "epoch": 0.76, "grad_norm": 6.833189836898929, "learning_rate": 1.4722758528582986e-06, "loss": 0.4753, "step": 10554 }, { "epoch": 0.76, "grad_norm": 3.2809781869592674, "learning_rate": 1.4714529635192182e-06, "loss": 0.1236, "step": 10555 }, { "epoch": 0.76, "grad_norm": 3.2872226724072657, "learning_rate": 1.470630264531892e-06, "loss": 0.1512, "step": 10556 }, { "epoch": 0.76, "grad_norm": 4.986905207215766, "learning_rate": 1.4698077559407059e-06, "loss": 0.8471, "step": 10557 }, { "epoch": 0.76, "grad_norm": 6.446992031566262, "learning_rate": 1.468985437790028e-06, "loss": 0.2873, "step": 10558 }, { "epoch": 0.76, "grad_norm": 2.162147391520608, "learning_rate": 1.4681633101242194e-06, "loss": 0.1081, "step": 10559 }, { "epoch": 0.76, "grad_norm": 8.449690446276472, "learning_rate": 1.4673413729876324e-06, "loss": 0.2842, "step": 10560 }, { "epoch": 0.76, "grad_norm": 2.546357475225605, "learning_rate": 1.4665196264246067e-06, "loss": 0.0914, "step": 10561 }, { "epoch": 0.76, "grad_norm": 4.40111042933875, "learning_rate": 1.4656980704794727e-06, "loss": 0.308, "step": 10562 }, { "epoch": 0.76, "grad_norm": 15.334344268996473, "learning_rate": 1.4648767051965473e-06, "loss": 1.3777, "step": 10563 }, { "epoch": 0.76, "grad_norm": 3.2636102236269573, "learning_rate": 1.4640555306201464e-06, "loss": 0.1115, "step": 10564 }, { "epoch": 0.76, "grad_norm": 5.233329585581898, "learning_rate": 1.463234546794562e-06, "loss": 0.1693, "step": 10565 }, { "epoch": 0.76, "grad_norm": 14.633788535784914, "learning_rate": 1.4624137537640877e-06, "loss": 0.811, "step": 10566 }, { "epoch": 0.76, "grad_norm": 10.885629548098555, "learning_rate": 1.4615931515730008e-06, "loss": 0.4491, "step": 10567 }, { "epoch": 0.76, "grad_norm": 9.361248119739999, "learning_rate": 1.4607727402655703e-06, "loss": 0.4208, "step": 10568 }, { "epoch": 0.76, "grad_norm": 5.0350061141361975, "learning_rate": 1.4599525198860515e-06, "loss": 0.8372, "step": 10569 }, { "epoch": 0.76, "grad_norm": 26.31789302989978, "learning_rate": 1.459132490478697e-06, "loss": 0.7396, "step": 10570 }, { "epoch": 0.76, "grad_norm": 8.99925628148459, "learning_rate": 1.4583126520877406e-06, "loss": 0.5178, "step": 10571 }, { "epoch": 0.76, "grad_norm": 14.453139836085558, "learning_rate": 1.45749300475741e-06, "loss": 0.5822, "step": 10572 }, { "epoch": 0.76, "grad_norm": 11.588440564338848, "learning_rate": 1.4566735485319233e-06, "loss": 0.4421, "step": 10573 }, { "epoch": 0.76, "grad_norm": 3.103068774762991, "learning_rate": 1.4558542834554873e-06, "loss": 0.2218, "step": 10574 }, { "epoch": 0.76, "grad_norm": 8.087552578461203, "learning_rate": 1.4550352095722952e-06, "loss": 0.4162, "step": 10575 }, { "epoch": 0.76, "grad_norm": 15.023844828765505, "learning_rate": 1.4542163269265386e-06, "loss": 0.3649, "step": 10576 }, { "epoch": 0.76, "grad_norm": 2.5636133336561784, "learning_rate": 1.453397635562388e-06, "loss": 0.1049, "step": 10577 }, { "epoch": 0.76, "grad_norm": 3.0081894703424297, "learning_rate": 1.4525791355240088e-06, "loss": 0.1902, "step": 10578 }, { "epoch": 0.76, "grad_norm": 16.87464641859355, "learning_rate": 1.4517608268555594e-06, "loss": 0.5111, "step": 10579 }, { "epoch": 0.76, "grad_norm": 11.832241641703714, "learning_rate": 1.4509427096011818e-06, "loss": 0.6642, "step": 10580 }, { "epoch": 0.76, "grad_norm": 8.20132822622267, "learning_rate": 1.4501247838050103e-06, "loss": 0.558, "step": 10581 }, { "epoch": 0.76, "grad_norm": 14.311962248716675, "learning_rate": 1.4493070495111711e-06, "loss": 0.4939, "step": 10582 }, { "epoch": 0.76, "grad_norm": 64.71637308556478, "learning_rate": 1.448489506763776e-06, "loss": 0.4617, "step": 10583 }, { "epoch": 0.76, "grad_norm": 2.824961394944865, "learning_rate": 1.447672155606928e-06, "loss": 0.1139, "step": 10584 }, { "epoch": 0.76, "grad_norm": 3.3297230473223958, "learning_rate": 1.4468549960847216e-06, "loss": 0.1804, "step": 10585 }, { "epoch": 0.76, "grad_norm": 2.6743571975389866, "learning_rate": 1.4460380282412395e-06, "loss": 0.1267, "step": 10586 }, { "epoch": 0.76, "grad_norm": 20.011316236362074, "learning_rate": 1.4452212521205516e-06, "loss": 0.7928, "step": 10587 }, { "epoch": 0.76, "grad_norm": 10.602912778759567, "learning_rate": 1.4444046677667245e-06, "loss": 0.5601, "step": 10588 }, { "epoch": 0.76, "grad_norm": 3.8613667808309446, "learning_rate": 1.4435882752238055e-06, "loss": 0.7896, "step": 10589 }, { "epoch": 0.76, "grad_norm": 3.2488965870380753, "learning_rate": 1.442772074535836e-06, "loss": 0.2075, "step": 10590 }, { "epoch": 0.76, "grad_norm": 4.391581983100564, "learning_rate": 1.441956065746849e-06, "loss": 0.684, "step": 10591 }, { "epoch": 0.76, "grad_norm": 7.638700230326836, "learning_rate": 1.4411402489008652e-06, "loss": 0.3807, "step": 10592 }, { "epoch": 0.76, "grad_norm": 3.5643716950878295, "learning_rate": 1.4403246240418917e-06, "loss": 0.1512, "step": 10593 }, { "epoch": 0.76, "grad_norm": 2.1954463865907416, "learning_rate": 1.4395091912139325e-06, "loss": 0.1739, "step": 10594 }, { "epoch": 0.76, "grad_norm": 17.398709457172636, "learning_rate": 1.4386939504609754e-06, "loss": 0.7222, "step": 10595 }, { "epoch": 0.76, "grad_norm": 3.247343548029793, "learning_rate": 1.4378789018269996e-06, "loss": 0.1428, "step": 10596 }, { "epoch": 0.76, "grad_norm": 4.274819696243151, "learning_rate": 1.4370640453559714e-06, "loss": 0.7883, "step": 10597 }, { "epoch": 0.76, "grad_norm": 3.607313734492282, "learning_rate": 1.436249381091853e-06, "loss": 0.176, "step": 10598 }, { "epoch": 0.76, "grad_norm": 5.308395948162153, "learning_rate": 1.435434909078592e-06, "loss": 0.9024, "step": 10599 }, { "epoch": 0.76, "grad_norm": 5.919551308158774, "learning_rate": 1.4346206293601245e-06, "loss": 0.2561, "step": 10600 }, { "epoch": 0.76, "grad_norm": 12.942194547568148, "learning_rate": 1.4338065419803782e-06, "loss": 0.5528, "step": 10601 }, { "epoch": 0.76, "grad_norm": 3.201127969863803, "learning_rate": 1.4329926469832683e-06, "loss": 0.1337, "step": 10602 }, { "epoch": 0.76, "grad_norm": 3.9441124693291765, "learning_rate": 1.4321789444127055e-06, "loss": 0.7034, "step": 10603 }, { "epoch": 0.76, "grad_norm": 5.03752583557945, "learning_rate": 1.4313654343125833e-06, "loss": 0.898, "step": 10604 }, { "epoch": 0.76, "grad_norm": 3.4750836042606914, "learning_rate": 1.4305521167267882e-06, "loss": 0.1209, "step": 10605 }, { "epoch": 0.76, "grad_norm": 2.4665924095982303, "learning_rate": 1.4297389916991939e-06, "loss": 0.104, "step": 10606 }, { "epoch": 0.76, "grad_norm": 4.263653274600042, "learning_rate": 1.4289260592736676e-06, "loss": 1.0443, "step": 10607 }, { "epoch": 0.76, "grad_norm": 2.870736467255477, "learning_rate": 1.4281133194940638e-06, "loss": 0.1747, "step": 10608 }, { "epoch": 0.76, "grad_norm": 5.593447083254364, "learning_rate": 1.4273007724042244e-06, "loss": 0.361, "step": 10609 }, { "epoch": 0.76, "grad_norm": 2.614152107702655, "learning_rate": 1.4264884180479865e-06, "loss": 0.1686, "step": 10610 }, { "epoch": 0.76, "grad_norm": 2.4290844686275097, "learning_rate": 1.4256762564691718e-06, "loss": 0.0975, "step": 10611 }, { "epoch": 0.76, "grad_norm": 3.295893992639529, "learning_rate": 1.4248642877115937e-06, "loss": 0.2358, "step": 10612 }, { "epoch": 0.76, "grad_norm": 24.30779679428163, "learning_rate": 1.4240525118190547e-06, "loss": 1.0415, "step": 10613 }, { "epoch": 0.76, "grad_norm": 38.44377488771993, "learning_rate": 1.4232409288353467e-06, "loss": 0.3181, "step": 10614 }, { "epoch": 0.76, "grad_norm": 3.3109456248004205, "learning_rate": 1.4224295388042503e-06, "loss": 0.1609, "step": 10615 }, { "epoch": 0.76, "grad_norm": 3.8930032573361615, "learning_rate": 1.42161834176954e-06, "loss": 0.8178, "step": 10616 }, { "epoch": 0.76, "grad_norm": 3.093435597262616, "learning_rate": 1.4208073377749753e-06, "loss": 0.1983, "step": 10617 }, { "epoch": 0.76, "grad_norm": 2.7957359706044627, "learning_rate": 1.4199965268643057e-06, "loss": 0.1281, "step": 10618 }, { "epoch": 0.76, "grad_norm": 3.903825252625034, "learning_rate": 1.4191859090812732e-06, "loss": 0.1963, "step": 10619 }, { "epoch": 0.76, "grad_norm": 3.907339088543511, "learning_rate": 1.4183754844696074e-06, "loss": 0.7911, "step": 10620 }, { "epoch": 0.76, "grad_norm": 10.390014547017827, "learning_rate": 1.4175652530730249e-06, "loss": 0.4679, "step": 10621 }, { "epoch": 0.76, "grad_norm": 4.5232544256505065, "learning_rate": 1.4167552149352386e-06, "loss": 1.0097, "step": 10622 }, { "epoch": 0.76, "grad_norm": 16.6189485523489, "learning_rate": 1.4159453700999453e-06, "loss": 0.7179, "step": 10623 }, { "epoch": 0.76, "grad_norm": 3.462151908065913, "learning_rate": 1.4151357186108327e-06, "loss": 0.1801, "step": 10624 }, { "epoch": 0.76, "grad_norm": 4.510339904770555, "learning_rate": 1.4143262605115788e-06, "loss": 0.8752, "step": 10625 }, { "epoch": 0.76, "grad_norm": 18.23577592849459, "learning_rate": 1.4135169958458505e-06, "loss": 0.4661, "step": 10626 }, { "epoch": 0.76, "grad_norm": 5.2980144743450195, "learning_rate": 1.4127079246573033e-06, "loss": 0.7859, "step": 10627 }, { "epoch": 0.76, "grad_norm": 3.0375364240336054, "learning_rate": 1.4118990469895866e-06, "loss": 0.1688, "step": 10628 }, { "epoch": 0.76, "grad_norm": 3.0442798636979598, "learning_rate": 1.4110903628863343e-06, "loss": 0.1501, "step": 10629 }, { "epoch": 0.76, "grad_norm": 3.176689165627601, "learning_rate": 1.4102818723911704e-06, "loss": 0.0988, "step": 10630 }, { "epoch": 0.76, "grad_norm": 5.687466444617511, "learning_rate": 1.4094735755477135e-06, "loss": 0.8676, "step": 10631 }, { "epoch": 0.76, "grad_norm": 22.65013346625524, "learning_rate": 1.4086654723995663e-06, "loss": 0.9402, "step": 10632 }, { "epoch": 0.76, "grad_norm": 3.969674882330943, "learning_rate": 1.4078575629903229e-06, "loss": 0.233, "step": 10633 }, { "epoch": 0.76, "grad_norm": 17.961157132363113, "learning_rate": 1.4070498473635657e-06, "loss": 0.6514, "step": 10634 }, { "epoch": 0.76, "grad_norm": 3.256799456142964, "learning_rate": 1.4062423255628704e-06, "loss": 0.1505, "step": 10635 }, { "epoch": 0.76, "grad_norm": 22.4104026085497, "learning_rate": 1.4054349976317988e-06, "loss": 0.8126, "step": 10636 }, { "epoch": 0.76, "grad_norm": 3.257221894969381, "learning_rate": 1.4046278636139026e-06, "loss": 0.1319, "step": 10637 }, { "epoch": 0.76, "grad_norm": 3.0194210600735496, "learning_rate": 1.403820923552724e-06, "loss": 0.2149, "step": 10638 }, { "epoch": 0.76, "grad_norm": 3.2129792017366867, "learning_rate": 1.4030141774917926e-06, "loss": 0.1106, "step": 10639 }, { "epoch": 0.76, "grad_norm": 18.493599448659605, "learning_rate": 1.4022076254746331e-06, "loss": 0.9926, "step": 10640 }, { "epoch": 0.76, "grad_norm": 15.999146571915345, "learning_rate": 1.401401267544753e-06, "loss": 0.8752, "step": 10641 }, { "epoch": 0.76, "grad_norm": 3.0200515520365845, "learning_rate": 1.4005951037456538e-06, "loss": 0.0857, "step": 10642 }, { "epoch": 0.76, "grad_norm": 11.38032559513147, "learning_rate": 1.3997891341208226e-06, "loss": 0.4815, "step": 10643 }, { "epoch": 0.76, "grad_norm": 14.17713307748143, "learning_rate": 1.3989833587137414e-06, "loss": 0.9973, "step": 10644 }, { "epoch": 0.76, "grad_norm": 10.979210791726464, "learning_rate": 1.3981777775678773e-06, "loss": 0.4971, "step": 10645 }, { "epoch": 0.76, "grad_norm": 4.839131574254719, "learning_rate": 1.3973723907266873e-06, "loss": 0.2467, "step": 10646 }, { "epoch": 0.76, "grad_norm": 2.985897667274232, "learning_rate": 1.3965671982336216e-06, "loss": 0.163, "step": 10647 }, { "epoch": 0.76, "grad_norm": 10.74654152057079, "learning_rate": 1.395762200132117e-06, "loss": 0.6672, "step": 10648 }, { "epoch": 0.76, "grad_norm": 37.2218602615346, "learning_rate": 1.394957396465596e-06, "loss": 1.0056, "step": 10649 }, { "epoch": 0.76, "grad_norm": 3.382556230845674, "learning_rate": 1.3941527872774796e-06, "loss": 0.1825, "step": 10650 }, { "epoch": 0.76, "grad_norm": 13.885518260974477, "learning_rate": 1.3933483726111713e-06, "loss": 0.735, "step": 10651 }, { "epoch": 0.76, "grad_norm": 9.635180615210244, "learning_rate": 1.3925441525100647e-06, "loss": 0.5587, "step": 10652 }, { "epoch": 0.76, "grad_norm": 8.190070312358959, "learning_rate": 1.3917401270175473e-06, "loss": 0.4937, "step": 10653 }, { "epoch": 0.76, "grad_norm": 9.250957356964403, "learning_rate": 1.3909362961769922e-06, "loss": 0.3097, "step": 10654 }, { "epoch": 0.76, "grad_norm": 12.877673435624017, "learning_rate": 1.3901326600317615e-06, "loss": 0.496, "step": 10655 }, { "epoch": 0.76, "grad_norm": 4.455800261249573, "learning_rate": 1.3893292186252116e-06, "loss": 0.796, "step": 10656 }, { "epoch": 0.76, "grad_norm": 2.7829530658382735, "learning_rate": 1.3885259720006828e-06, "loss": 0.1473, "step": 10657 }, { "epoch": 0.76, "grad_norm": 4.229516851766469, "learning_rate": 1.3877229202015064e-06, "loss": 0.7249, "step": 10658 }, { "epoch": 0.76, "grad_norm": 12.39637523260589, "learning_rate": 1.3869200632710061e-06, "loss": 0.6493, "step": 10659 }, { "epoch": 0.76, "grad_norm": 4.600013456806901, "learning_rate": 1.3861174012524942e-06, "loss": 0.7862, "step": 10660 }, { "epoch": 0.76, "grad_norm": 3.5467949064822877, "learning_rate": 1.3853149341892664e-06, "loss": 0.279, "step": 10661 }, { "epoch": 0.76, "grad_norm": 2.802297512593086, "learning_rate": 1.3845126621246164e-06, "loss": 0.1406, "step": 10662 }, { "epoch": 0.76, "grad_norm": 2.8201273689167263, "learning_rate": 1.3837105851018235e-06, "loss": 0.1469, "step": 10663 }, { "epoch": 0.76, "grad_norm": 4.121527714923286, "learning_rate": 1.3829087031641542e-06, "loss": 0.6639, "step": 10664 }, { "epoch": 0.76, "grad_norm": 2.701345986962685, "learning_rate": 1.3821070163548705e-06, "loss": 0.124, "step": 10665 }, { "epoch": 0.76, "grad_norm": 13.758386421083037, "learning_rate": 1.3813055247172192e-06, "loss": 0.8327, "step": 10666 }, { "epoch": 0.76, "grad_norm": 3.124530143138668, "learning_rate": 1.3805042282944353e-06, "loss": 0.1981, "step": 10667 }, { "epoch": 0.76, "grad_norm": 7.712099240815089, "learning_rate": 1.3797031271297495e-06, "loss": 0.8704, "step": 10668 }, { "epoch": 0.77, "grad_norm": 11.201132859838387, "learning_rate": 1.3789022212663766e-06, "loss": 0.7348, "step": 10669 }, { "epoch": 0.77, "grad_norm": 3.4878244087842463, "learning_rate": 1.378101510747522e-06, "loss": 0.1249, "step": 10670 }, { "epoch": 0.77, "grad_norm": 3.737561144797206, "learning_rate": 1.3773009956163802e-06, "loss": 0.2507, "step": 10671 }, { "epoch": 0.77, "grad_norm": 35.00341779859813, "learning_rate": 1.3765006759161397e-06, "loss": 0.8298, "step": 10672 }, { "epoch": 0.77, "grad_norm": 13.456109022284792, "learning_rate": 1.3757005516899691e-06, "loss": 0.4946, "step": 10673 }, { "epoch": 0.77, "grad_norm": 3.0129458904617357, "learning_rate": 1.3749006229810364e-06, "loss": 0.1371, "step": 10674 }, { "epoch": 0.77, "grad_norm": 3.859262895092709, "learning_rate": 1.374100889832493e-06, "loss": 0.2566, "step": 10675 }, { "epoch": 0.77, "grad_norm": 2.8806010155802126, "learning_rate": 1.3733013522874826e-06, "loss": 0.1563, "step": 10676 }, { "epoch": 0.77, "grad_norm": 14.08179425178552, "learning_rate": 1.3725020103891346e-06, "loss": 0.9415, "step": 10677 }, { "epoch": 0.77, "grad_norm": 6.816715090963509, "learning_rate": 1.3717028641805736e-06, "loss": 0.2905, "step": 10678 }, { "epoch": 0.77, "grad_norm": 8.518166501089384, "learning_rate": 1.3709039137049097e-06, "loss": 0.3973, "step": 10679 }, { "epoch": 0.77, "grad_norm": 3.1023243659308246, "learning_rate": 1.3701051590052406e-06, "loss": 0.1693, "step": 10680 }, { "epoch": 0.77, "grad_norm": 4.67977004635033, "learning_rate": 1.3693066001246601e-06, "loss": 0.1572, "step": 10681 }, { "epoch": 0.77, "grad_norm": 2.0590747306900354, "learning_rate": 1.368508237106246e-06, "loss": 0.0872, "step": 10682 }, { "epoch": 0.77, "grad_norm": 2.5394819502973145, "learning_rate": 1.3677100699930647e-06, "loss": 0.1007, "step": 10683 }, { "epoch": 0.77, "grad_norm": 3.327674872309306, "learning_rate": 1.3669120988281776e-06, "loss": 0.1611, "step": 10684 }, { "epoch": 0.77, "grad_norm": 3.1892566822734008, "learning_rate": 1.3661143236546325e-06, "loss": 0.2363, "step": 10685 }, { "epoch": 0.77, "grad_norm": 4.307331429565649, "learning_rate": 1.3653167445154613e-06, "loss": 0.8876, "step": 10686 }, { "epoch": 0.77, "grad_norm": 3.397543646271601, "learning_rate": 1.3645193614536955e-06, "loss": 0.1941, "step": 10687 }, { "epoch": 0.77, "grad_norm": 3.295273926889126, "learning_rate": 1.3637221745123492e-06, "loss": 0.1482, "step": 10688 }, { "epoch": 0.77, "grad_norm": 4.22248337612284, "learning_rate": 1.362925183734426e-06, "loss": 0.847, "step": 10689 }, { "epoch": 0.77, "grad_norm": 14.154117452905423, "learning_rate": 1.3621283891629234e-06, "loss": 0.6075, "step": 10690 }, { "epoch": 0.77, "grad_norm": 11.432717044446454, "learning_rate": 1.361331790840824e-06, "loss": 0.3571, "step": 10691 }, { "epoch": 0.77, "grad_norm": 5.122886183242533, "learning_rate": 1.3605353888111e-06, "loss": 0.7236, "step": 10692 }, { "epoch": 0.77, "grad_norm": 3.3834365563310955, "learning_rate": 1.3597391831167174e-06, "loss": 0.2551, "step": 10693 }, { "epoch": 0.77, "grad_norm": 3.6218940489298035, "learning_rate": 1.3589431738006264e-06, "loss": 0.5757, "step": 10694 }, { "epoch": 0.77, "grad_norm": 2.4220648854490934, "learning_rate": 1.3581473609057677e-06, "loss": 0.1044, "step": 10695 }, { "epoch": 0.77, "grad_norm": 16.359311306663965, "learning_rate": 1.357351744475075e-06, "loss": 0.6328, "step": 10696 }, { "epoch": 0.77, "grad_norm": 3.5339394279636833, "learning_rate": 1.356556324551469e-06, "loss": 0.1763, "step": 10697 }, { "epoch": 0.77, "grad_norm": 27.654752187285336, "learning_rate": 1.3557611011778554e-06, "loss": 0.6234, "step": 10698 }, { "epoch": 0.77, "grad_norm": 7.900150442551394, "learning_rate": 1.3549660743971377e-06, "loss": 0.4418, "step": 10699 }, { "epoch": 0.77, "grad_norm": 11.342945682201714, "learning_rate": 1.3541712442522025e-06, "loss": 0.3903, "step": 10700 }, { "epoch": 0.77, "grad_norm": 6.670414873414035, "learning_rate": 1.3533766107859275e-06, "loss": 0.9131, "step": 10701 }, { "epoch": 0.77, "grad_norm": 3.0597647157619847, "learning_rate": 1.3525821740411826e-06, "loss": 0.1926, "step": 10702 }, { "epoch": 0.77, "grad_norm": 2.1360808652033567, "learning_rate": 1.351787934060823e-06, "loss": 0.1156, "step": 10703 }, { "epoch": 0.77, "grad_norm": 4.8512732650688, "learning_rate": 1.350993890887694e-06, "loss": 0.7457, "step": 10704 }, { "epoch": 0.77, "grad_norm": 40.247118321360034, "learning_rate": 1.3502000445646336e-06, "loss": 1.1595, "step": 10705 }, { "epoch": 0.77, "grad_norm": 21.97115527935559, "learning_rate": 1.3494063951344654e-06, "loss": 0.7752, "step": 10706 }, { "epoch": 0.77, "grad_norm": 14.891546836488786, "learning_rate": 1.3486129426400042e-06, "loss": 0.4977, "step": 10707 }, { "epoch": 0.77, "grad_norm": 17.977283726978193, "learning_rate": 1.3478196871240518e-06, "loss": 0.6559, "step": 10708 }, { "epoch": 0.77, "grad_norm": 13.374796936035361, "learning_rate": 1.3470266286294065e-06, "loss": 0.7296, "step": 10709 }, { "epoch": 0.77, "grad_norm": 2.471976343401538, "learning_rate": 1.3462337671988434e-06, "loss": 0.106, "step": 10710 }, { "epoch": 0.77, "grad_norm": 3.023285841528439, "learning_rate": 1.3454411028751408e-06, "loss": 0.1694, "step": 10711 }, { "epoch": 0.77, "grad_norm": 16.017318890132845, "learning_rate": 1.344648635701057e-06, "loss": 0.7123, "step": 10712 }, { "epoch": 0.77, "grad_norm": 4.28176935018823, "learning_rate": 1.3438563657193427e-06, "loss": 0.6388, "step": 10713 }, { "epoch": 0.77, "grad_norm": 3.0416137789735314, "learning_rate": 1.3430642929727373e-06, "loss": 0.1502, "step": 10714 }, { "epoch": 0.77, "grad_norm": 4.572502484738679, "learning_rate": 1.342272417503972e-06, "loss": 0.7939, "step": 10715 }, { "epoch": 0.77, "grad_norm": 28.496399056903407, "learning_rate": 1.341480739355765e-06, "loss": 0.86, "step": 10716 }, { "epoch": 0.77, "grad_norm": 3.5107440908199066, "learning_rate": 1.340689258570822e-06, "loss": 0.1794, "step": 10717 }, { "epoch": 0.77, "grad_norm": 4.673732893234264, "learning_rate": 1.339897975191844e-06, "loss": 0.8691, "step": 10718 }, { "epoch": 0.77, "grad_norm": 3.766794834615157, "learning_rate": 1.339106889261516e-06, "loss": 0.1614, "step": 10719 }, { "epoch": 0.77, "grad_norm": 14.247691438577302, "learning_rate": 1.3383160008225126e-06, "loss": 0.5714, "step": 10720 }, { "epoch": 0.77, "grad_norm": 4.422791982625853, "learning_rate": 1.3375253099175035e-06, "loss": 0.2424, "step": 10721 }, { "epoch": 0.77, "grad_norm": 3.446005779212047, "learning_rate": 1.3367348165891397e-06, "loss": 0.1373, "step": 10722 }, { "epoch": 0.77, "grad_norm": 5.117319573013353, "learning_rate": 1.3359445208800642e-06, "loss": 0.335, "step": 10723 }, { "epoch": 0.77, "grad_norm": 2.52058472210535, "learning_rate": 1.3351544228329144e-06, "loss": 0.1906, "step": 10724 }, { "epoch": 0.77, "grad_norm": 3.012440235937686, "learning_rate": 1.3343645224903113e-06, "loss": 0.1496, "step": 10725 }, { "epoch": 0.77, "grad_norm": 4.172809194262789, "learning_rate": 1.3335748198948656e-06, "loss": 0.1704, "step": 10726 }, { "epoch": 0.77, "grad_norm": 3.099799407529779, "learning_rate": 1.3327853150891817e-06, "loss": 0.2348, "step": 10727 }, { "epoch": 0.77, "grad_norm": 3.8334064949377606, "learning_rate": 1.331996008115849e-06, "loss": 0.1872, "step": 10728 }, { "epoch": 0.77, "grad_norm": 2.8612075608638667, "learning_rate": 1.3312068990174453e-06, "loss": 0.103, "step": 10729 }, { "epoch": 0.77, "grad_norm": 17.675458636998318, "learning_rate": 1.3304179878365447e-06, "loss": 0.6167, "step": 10730 }, { "epoch": 0.77, "grad_norm": 7.523101737542959, "learning_rate": 1.329629274615703e-06, "loss": 0.2702, "step": 10731 }, { "epoch": 0.77, "grad_norm": 23.700060749662416, "learning_rate": 1.3288407593974673e-06, "loss": 0.8262, "step": 10732 }, { "epoch": 0.77, "grad_norm": 4.4458867151735575, "learning_rate": 1.3280524422243796e-06, "loss": 0.7723, "step": 10733 }, { "epoch": 0.77, "grad_norm": 3.5032964815002163, "learning_rate": 1.327264323138962e-06, "loss": 0.1564, "step": 10734 }, { "epoch": 0.77, "grad_norm": 8.217001303435797, "learning_rate": 1.3264764021837313e-06, "loss": 0.3271, "step": 10735 }, { "epoch": 0.77, "grad_norm": 14.475338925210373, "learning_rate": 1.3256886794011942e-06, "loss": 0.4367, "step": 10736 }, { "epoch": 0.77, "grad_norm": 9.87405454122727, "learning_rate": 1.3249011548338452e-06, "loss": 0.6162, "step": 10737 }, { "epoch": 0.77, "grad_norm": 3.0165217970297946, "learning_rate": 1.3241138285241668e-06, "loss": 0.1568, "step": 10738 }, { "epoch": 0.77, "grad_norm": 11.547831283718331, "learning_rate": 1.3233267005146343e-06, "loss": 0.2537, "step": 10739 }, { "epoch": 0.77, "grad_norm": 12.998011793396769, "learning_rate": 1.3225397708477094e-06, "loss": 0.9247, "step": 10740 }, { "epoch": 0.77, "grad_norm": 3.0607150277405375, "learning_rate": 1.321753039565844e-06, "loss": 0.1349, "step": 10741 }, { "epoch": 0.77, "grad_norm": 16.300212394712517, "learning_rate": 1.320966506711478e-06, "loss": 0.9505, "step": 10742 }, { "epoch": 0.77, "grad_norm": 49.72931823603606, "learning_rate": 1.320180172327044e-06, "loss": 0.7607, "step": 10743 }, { "epoch": 0.77, "grad_norm": 3.374181889930973, "learning_rate": 1.3193940364549613e-06, "loss": 0.1486, "step": 10744 }, { "epoch": 0.77, "grad_norm": 7.374216313204301, "learning_rate": 1.3186080991376382e-06, "loss": 0.74, "step": 10745 }, { "epoch": 0.77, "grad_norm": 14.043914258938795, "learning_rate": 1.3178223604174734e-06, "loss": 0.496, "step": 10746 }, { "epoch": 0.77, "grad_norm": 5.053811678747211, "learning_rate": 1.3170368203368534e-06, "loss": 0.6761, "step": 10747 }, { "epoch": 0.77, "grad_norm": 4.167261868169029, "learning_rate": 1.316251478938157e-06, "loss": 0.7532, "step": 10748 }, { "epoch": 0.77, "grad_norm": 13.553125980694803, "learning_rate": 1.3154663362637498e-06, "loss": 1.0625, "step": 10749 }, { "epoch": 0.77, "grad_norm": 6.636470988964357, "learning_rate": 1.314681392355987e-06, "loss": 0.29, "step": 10750 }, { "epoch": 0.77, "grad_norm": 3.5514459804297567, "learning_rate": 1.3138966472572124e-06, "loss": 0.3251, "step": 10751 }, { "epoch": 0.77, "grad_norm": 10.336675300352885, "learning_rate": 1.313112101009762e-06, "loss": 0.551, "step": 10752 }, { "epoch": 0.77, "grad_norm": 3.2484987013838817, "learning_rate": 1.312327753655958e-06, "loss": 0.2009, "step": 10753 }, { "epoch": 0.77, "grad_norm": 2.9010144715190465, "learning_rate": 1.311543605238112e-06, "loss": 0.1244, "step": 10754 }, { "epoch": 0.77, "grad_norm": 17.980726809385217, "learning_rate": 1.3107596557985281e-06, "loss": 0.817, "step": 10755 }, { "epoch": 0.77, "grad_norm": 4.383290372674291, "learning_rate": 1.3099759053794963e-06, "loss": 0.902, "step": 10756 }, { "epoch": 0.77, "grad_norm": 19.189235927813883, "learning_rate": 1.3091923540232971e-06, "loss": 0.5769, "step": 10757 }, { "epoch": 0.77, "grad_norm": 2.9117848679501286, "learning_rate": 1.3084090017722e-06, "loss": 0.1228, "step": 10758 }, { "epoch": 0.77, "grad_norm": 3.5814535133178116, "learning_rate": 1.3076258486684635e-06, "loss": 0.1486, "step": 10759 }, { "epoch": 0.77, "grad_norm": 3.2299515225543667, "learning_rate": 1.3068428947543348e-06, "loss": 0.1435, "step": 10760 }, { "epoch": 0.77, "grad_norm": 3.0643446798166156, "learning_rate": 1.306060140072054e-06, "loss": 0.2056, "step": 10761 }, { "epoch": 0.77, "grad_norm": 3.620500460352851, "learning_rate": 1.305277584663846e-06, "loss": 0.2604, "step": 10762 }, { "epoch": 0.77, "grad_norm": 4.534893414594236, "learning_rate": 1.304495228571926e-06, "loss": 0.8854, "step": 10763 }, { "epoch": 0.77, "grad_norm": 3.243807607675401, "learning_rate": 1.3037130718385016e-06, "loss": 0.1459, "step": 10764 }, { "epoch": 0.77, "grad_norm": 3.285213244092131, "learning_rate": 1.3029311145057655e-06, "loss": 0.1538, "step": 10765 }, { "epoch": 0.77, "grad_norm": 8.640827838086842, "learning_rate": 1.3021493566159004e-06, "loss": 0.2966, "step": 10766 }, { "epoch": 0.77, "grad_norm": 2.9329110990385634, "learning_rate": 1.3013677982110817e-06, "loss": 0.1315, "step": 10767 }, { "epoch": 0.77, "grad_norm": 7.0871272669616445, "learning_rate": 1.3005864393334704e-06, "loss": 0.3163, "step": 10768 }, { "epoch": 0.77, "grad_norm": 29.077875657846704, "learning_rate": 1.2998052800252176e-06, "loss": 0.6663, "step": 10769 }, { "epoch": 0.77, "grad_norm": 6.88237062950187, "learning_rate": 1.2990243203284636e-06, "loss": 0.6772, "step": 10770 }, { "epoch": 0.77, "grad_norm": 5.421094000433625, "learning_rate": 1.2982435602853389e-06, "loss": 0.797, "step": 10771 }, { "epoch": 0.77, "grad_norm": 10.846999833668322, "learning_rate": 1.2974629999379607e-06, "loss": 0.4768, "step": 10772 }, { "epoch": 0.77, "grad_norm": 5.068295232805877, "learning_rate": 1.2966826393284404e-06, "loss": 0.8264, "step": 10773 }, { "epoch": 0.77, "grad_norm": 3.2840214212566945, "learning_rate": 1.2959024784988739e-06, "loss": 0.1415, "step": 10774 }, { "epoch": 0.77, "grad_norm": 3.0044245570306405, "learning_rate": 1.2951225174913462e-06, "loss": 0.1748, "step": 10775 }, { "epoch": 0.77, "grad_norm": 2.8415831078728675, "learning_rate": 1.294342756347936e-06, "loss": 0.125, "step": 10776 }, { "epoch": 0.77, "grad_norm": 3.1356329428213696, "learning_rate": 1.293563195110708e-06, "loss": 0.1654, "step": 10777 }, { "epoch": 0.77, "grad_norm": 3.362463877756069, "learning_rate": 1.2927838338217164e-06, "loss": 0.1479, "step": 10778 }, { "epoch": 0.77, "grad_norm": 4.209818229655647, "learning_rate": 1.2920046725230023e-06, "loss": 0.6496, "step": 10779 }, { "epoch": 0.77, "grad_norm": 41.451953028056295, "learning_rate": 1.2912257112566019e-06, "loss": 1.1848, "step": 10780 }, { "epoch": 0.77, "grad_norm": 38.054374461684766, "learning_rate": 1.290446950064536e-06, "loss": 0.7443, "step": 10781 }, { "epoch": 0.77, "grad_norm": 21.892082098717786, "learning_rate": 1.2896683889888161e-06, "loss": 1.049, "step": 10782 }, { "epoch": 0.77, "grad_norm": 3.462186908930284, "learning_rate": 1.2888900280714422e-06, "loss": 0.0894, "step": 10783 }, { "epoch": 0.77, "grad_norm": 21.57608848992775, "learning_rate": 1.2881118673544024e-06, "loss": 1.0996, "step": 10784 }, { "epoch": 0.77, "grad_norm": 3.7809125218182347, "learning_rate": 1.2873339068796786e-06, "loss": 0.1925, "step": 10785 }, { "epoch": 0.77, "grad_norm": 4.544272817439608, "learning_rate": 1.2865561466892373e-06, "loss": 0.3016, "step": 10786 }, { "epoch": 0.77, "grad_norm": 5.134664577616902, "learning_rate": 1.2857785868250356e-06, "loss": 0.2553, "step": 10787 }, { "epoch": 0.77, "grad_norm": 13.315719136382766, "learning_rate": 1.2850012273290186e-06, "loss": 0.7443, "step": 10788 }, { "epoch": 0.77, "grad_norm": 2.969444945536656, "learning_rate": 1.2842240682431251e-06, "loss": 0.1668, "step": 10789 }, { "epoch": 0.77, "grad_norm": 14.435022733117233, "learning_rate": 1.2834471096092778e-06, "loss": 0.7358, "step": 10790 }, { "epoch": 0.77, "grad_norm": 3.6175294597325904, "learning_rate": 1.2826703514693894e-06, "loss": 0.1876, "step": 10791 }, { "epoch": 0.77, "grad_norm": 3.374876105354169, "learning_rate": 1.281893793865367e-06, "loss": 0.1719, "step": 10792 }, { "epoch": 0.77, "grad_norm": 4.176810599013184, "learning_rate": 1.2811174368391012e-06, "loss": 0.7695, "step": 10793 }, { "epoch": 0.77, "grad_norm": 3.683692647757938, "learning_rate": 1.2803412804324705e-06, "loss": 0.1916, "step": 10794 }, { "epoch": 0.77, "grad_norm": 3.6947157678131277, "learning_rate": 1.279565324687349e-06, "loss": 0.2071, "step": 10795 }, { "epoch": 0.77, "grad_norm": 14.698279233465572, "learning_rate": 1.2787895696455966e-06, "loss": 0.6859, "step": 10796 }, { "epoch": 0.77, "grad_norm": 2.978655574384722, "learning_rate": 1.2780140153490589e-06, "loss": 0.1794, "step": 10797 }, { "epoch": 0.77, "grad_norm": 3.2782807474706415, "learning_rate": 1.2772386618395788e-06, "loss": 0.1679, "step": 10798 }, { "epoch": 0.77, "grad_norm": 33.06626248386065, "learning_rate": 1.2764635091589812e-06, "loss": 0.6995, "step": 10799 }, { "epoch": 0.77, "grad_norm": 4.064916053292379, "learning_rate": 1.2756885573490814e-06, "loss": 0.1703, "step": 10800 }, { "epoch": 0.77, "grad_norm": 19.853610417395213, "learning_rate": 1.2749138064516881e-06, "loss": 0.9123, "step": 10801 }, { "epoch": 0.77, "grad_norm": 3.0305367329892756, "learning_rate": 1.2741392565085943e-06, "loss": 0.1411, "step": 10802 }, { "epoch": 0.77, "grad_norm": 3.1356555248449745, "learning_rate": 1.2733649075615833e-06, "loss": 0.1802, "step": 10803 }, { "epoch": 0.77, "grad_norm": 10.657846389150547, "learning_rate": 1.272590759652431e-06, "loss": 0.4412, "step": 10804 }, { "epoch": 0.77, "grad_norm": 13.281651387892136, "learning_rate": 1.2718168128229002e-06, "loss": 0.8569, "step": 10805 }, { "epoch": 0.77, "grad_norm": 3.169880766498499, "learning_rate": 1.2710430671147366e-06, "loss": 0.116, "step": 10806 }, { "epoch": 0.77, "grad_norm": 3.353922780638119, "learning_rate": 1.2702695225696865e-06, "loss": 0.1425, "step": 10807 }, { "epoch": 0.78, "grad_norm": 25.46473446424915, "learning_rate": 1.269496179229478e-06, "loss": 1.0486, "step": 10808 }, { "epoch": 0.78, "grad_norm": 3.0474382044537283, "learning_rate": 1.2687230371358284e-06, "loss": 0.1673, "step": 10809 }, { "epoch": 0.78, "grad_norm": 3.3249327451951762, "learning_rate": 1.2679500963304487e-06, "loss": 0.1531, "step": 10810 }, { "epoch": 0.78, "grad_norm": 4.11390230879204, "learning_rate": 1.267177356855035e-06, "loss": 0.21, "step": 10811 }, { "epoch": 0.78, "grad_norm": 2.9197391638341554, "learning_rate": 1.2664048187512712e-06, "loss": 0.1033, "step": 10812 }, { "epoch": 0.78, "grad_norm": 5.312237138102292, "learning_rate": 1.2656324820608374e-06, "loss": 0.7154, "step": 10813 }, { "epoch": 0.78, "grad_norm": 3.6278760234616163, "learning_rate": 1.264860346825395e-06, "loss": 0.2171, "step": 10814 }, { "epoch": 0.78, "grad_norm": 3.8508959828076774, "learning_rate": 1.2640884130865988e-06, "loss": 0.2125, "step": 10815 }, { "epoch": 0.78, "grad_norm": 16.854641371870198, "learning_rate": 1.2633166808860903e-06, "loss": 0.6897, "step": 10816 }, { "epoch": 0.78, "grad_norm": 19.354210481139113, "learning_rate": 1.2625451502655055e-06, "loss": 1.1707, "step": 10817 }, { "epoch": 0.78, "grad_norm": 8.118741481377578, "learning_rate": 1.2617738212664604e-06, "loss": 0.2733, "step": 10818 }, { "epoch": 0.78, "grad_norm": 4.029320447318949, "learning_rate": 1.2610026939305686e-06, "loss": 0.7743, "step": 10819 }, { "epoch": 0.78, "grad_norm": 34.37226120122269, "learning_rate": 1.2602317682994285e-06, "loss": 0.2489, "step": 10820 }, { "epoch": 0.78, "grad_norm": 10.021116158334566, "learning_rate": 1.2594610444146293e-06, "loss": 0.5145, "step": 10821 }, { "epoch": 0.78, "grad_norm": 25.281129113982058, "learning_rate": 1.2586905223177458e-06, "loss": 1.0068, "step": 10822 }, { "epoch": 0.78, "grad_norm": 3.2324513186959396, "learning_rate": 1.2579202020503494e-06, "loss": 0.1307, "step": 10823 }, { "epoch": 0.78, "grad_norm": 2.9327437597198394, "learning_rate": 1.257150083653993e-06, "loss": 0.1169, "step": 10824 }, { "epoch": 0.78, "grad_norm": 10.122346988732103, "learning_rate": 1.256380167170221e-06, "loss": 0.3806, "step": 10825 }, { "epoch": 0.78, "grad_norm": 3.006446101810794, "learning_rate": 1.2556104526405694e-06, "loss": 0.142, "step": 10826 }, { "epoch": 0.78, "grad_norm": 9.286259667673654, "learning_rate": 1.2548409401065609e-06, "loss": 0.7197, "step": 10827 }, { "epoch": 0.78, "grad_norm": 3.058076098914738, "learning_rate": 1.254071629609706e-06, "loss": 0.1555, "step": 10828 }, { "epoch": 0.78, "grad_norm": 3.4839637837212267, "learning_rate": 1.2533025211915106e-06, "loss": 0.2567, "step": 10829 }, { "epoch": 0.78, "grad_norm": 2.760074474566807, "learning_rate": 1.2525336148934608e-06, "loss": 0.147, "step": 10830 }, { "epoch": 0.78, "grad_norm": 21.038540920350485, "learning_rate": 1.2517649107570361e-06, "loss": 1.0413, "step": 10831 }, { "epoch": 0.78, "grad_norm": 9.371021279722989, "learning_rate": 1.250996408823708e-06, "loss": 0.3565, "step": 10832 }, { "epoch": 0.78, "grad_norm": 4.966079930128353, "learning_rate": 1.2502281091349327e-06, "loss": 0.9756, "step": 10833 }, { "epoch": 0.78, "grad_norm": 4.234370738054585, "learning_rate": 1.249460011732156e-06, "loss": 0.7404, "step": 10834 }, { "epoch": 0.78, "grad_norm": 12.916729280206981, "learning_rate": 1.2486921166568167e-06, "loss": 0.6533, "step": 10835 }, { "epoch": 0.78, "grad_norm": 3.063283646565542, "learning_rate": 1.247924423950338e-06, "loss": 0.1723, "step": 10836 }, { "epoch": 0.78, "grad_norm": 20.190056071754572, "learning_rate": 1.2471569336541329e-06, "loss": 0.9652, "step": 10837 }, { "epoch": 0.78, "grad_norm": 4.747656520316795, "learning_rate": 1.246389645809607e-06, "loss": 0.2712, "step": 10838 }, { "epoch": 0.78, "grad_norm": 3.902640974069244, "learning_rate": 1.2456225604581524e-06, "loss": 0.1621, "step": 10839 }, { "epoch": 0.78, "grad_norm": 18.98463834754695, "learning_rate": 1.2448556776411474e-06, "loss": 0.9324, "step": 10840 }, { "epoch": 0.78, "grad_norm": 3.179703488410313, "learning_rate": 1.2440889973999676e-06, "loss": 0.2133, "step": 10841 }, { "epoch": 0.78, "grad_norm": 36.001127171133696, "learning_rate": 1.243322519775968e-06, "loss": 0.5859, "step": 10842 }, { "epoch": 0.78, "grad_norm": 32.59240108051671, "learning_rate": 1.2425562448104977e-06, "loss": 0.9998, "step": 10843 }, { "epoch": 0.78, "grad_norm": 4.091884869782185, "learning_rate": 1.2417901725448961e-06, "loss": 0.7297, "step": 10844 }, { "epoch": 0.78, "grad_norm": 4.69273661488417, "learning_rate": 1.2410243030204893e-06, "loss": 0.7561, "step": 10845 }, { "epoch": 0.78, "grad_norm": 13.995788797247013, "learning_rate": 1.2402586362785918e-06, "loss": 0.6286, "step": 10846 }, { "epoch": 0.78, "grad_norm": 3.0623009183868817, "learning_rate": 1.2394931723605103e-06, "loss": 0.1305, "step": 10847 }, { "epoch": 0.78, "grad_norm": 4.409392210240674, "learning_rate": 1.2387279113075384e-06, "loss": 0.1549, "step": 10848 }, { "epoch": 0.78, "grad_norm": 4.557967622755561, "learning_rate": 1.2379628531609584e-06, "loss": 0.8824, "step": 10849 }, { "epoch": 0.78, "grad_norm": 3.2314472658098996, "learning_rate": 1.2371979979620408e-06, "loss": 0.1737, "step": 10850 }, { "epoch": 0.78, "grad_norm": 2.8841592451063156, "learning_rate": 1.2364333457520495e-06, "loss": 0.1704, "step": 10851 }, { "epoch": 0.78, "grad_norm": 3.265215651208285, "learning_rate": 1.2356688965722336e-06, "loss": 0.1825, "step": 10852 }, { "epoch": 0.78, "grad_norm": 3.438314887235421, "learning_rate": 1.2349046504638318e-06, "loss": 0.1725, "step": 10853 }, { "epoch": 0.78, "grad_norm": 2.863554108008442, "learning_rate": 1.234140607468073e-06, "loss": 0.0988, "step": 10854 }, { "epoch": 0.78, "grad_norm": 2.8468362272753907, "learning_rate": 1.2333767676261715e-06, "loss": 0.1481, "step": 10855 }, { "epoch": 0.78, "grad_norm": 4.453523810093792, "learning_rate": 1.2326131309793383e-06, "loss": 0.8074, "step": 10856 }, { "epoch": 0.78, "grad_norm": 12.055994903765232, "learning_rate": 1.2318496975687654e-06, "loss": 0.308, "step": 10857 }, { "epoch": 0.78, "grad_norm": 4.290213536347343, "learning_rate": 1.2310864674356387e-06, "loss": 0.3651, "step": 10858 }, { "epoch": 0.78, "grad_norm": 5.8187395256634105, "learning_rate": 1.2303234406211295e-06, "loss": 0.1912, "step": 10859 }, { "epoch": 0.78, "grad_norm": 4.828362670901011, "learning_rate": 1.2295606171664032e-06, "loss": 0.1231, "step": 10860 }, { "epoch": 0.78, "grad_norm": 13.194077959179834, "learning_rate": 1.2287979971126097e-06, "loss": 0.6637, "step": 10861 }, { "epoch": 0.78, "grad_norm": 19.45230233137056, "learning_rate": 1.2280355805008875e-06, "loss": 0.7132, "step": 10862 }, { "epoch": 0.78, "grad_norm": 3.1191900869197213, "learning_rate": 1.2272733673723702e-06, "loss": 0.1197, "step": 10863 }, { "epoch": 0.78, "grad_norm": 2.397714895092171, "learning_rate": 1.2265113577681736e-06, "loss": 0.0944, "step": 10864 }, { "epoch": 0.78, "grad_norm": 63.99509628918436, "learning_rate": 1.225749551729406e-06, "loss": 0.9666, "step": 10865 }, { "epoch": 0.78, "grad_norm": 4.712564108052201, "learning_rate": 1.2249879492971639e-06, "loss": 0.7769, "step": 10866 }, { "epoch": 0.78, "grad_norm": 20.235640859847592, "learning_rate": 1.2242265505125329e-06, "loss": 0.8002, "step": 10867 }, { "epoch": 0.78, "grad_norm": 14.79040438718522, "learning_rate": 1.2234653554165859e-06, "loss": 0.5711, "step": 10868 }, { "epoch": 0.78, "grad_norm": 3.3700206256774594, "learning_rate": 1.2227043640503895e-06, "loss": 0.1618, "step": 10869 }, { "epoch": 0.78, "grad_norm": 3.193384491757744, "learning_rate": 1.2219435764549948e-06, "loss": 0.2227, "step": 10870 }, { "epoch": 0.78, "grad_norm": 13.536016319852422, "learning_rate": 1.2211829926714413e-06, "loss": 0.5306, "step": 10871 }, { "epoch": 0.78, "grad_norm": 3.1324506239450494, "learning_rate": 1.2204226127407643e-06, "loss": 0.1596, "step": 10872 }, { "epoch": 0.78, "grad_norm": 4.429361546228547, "learning_rate": 1.21966243670398e-06, "loss": 0.2085, "step": 10873 }, { "epoch": 0.78, "grad_norm": 2.9292468286509745, "learning_rate": 1.218902464602097e-06, "loss": 0.1031, "step": 10874 }, { "epoch": 0.78, "grad_norm": 17.867527082456274, "learning_rate": 1.218142696476115e-06, "loss": 0.2848, "step": 10875 }, { "epoch": 0.78, "grad_norm": 13.380617825149619, "learning_rate": 1.2173831323670194e-06, "loss": 0.6408, "step": 10876 }, { "epoch": 0.78, "grad_norm": 4.837616404724494, "learning_rate": 1.216623772315786e-06, "loss": 0.8182, "step": 10877 }, { "epoch": 0.78, "grad_norm": 4.438812040610197, "learning_rate": 1.2158646163633787e-06, "loss": 0.1746, "step": 10878 }, { "epoch": 0.78, "grad_norm": 4.998931695214306, "learning_rate": 1.2151056645507519e-06, "loss": 0.8944, "step": 10879 }, { "epoch": 0.78, "grad_norm": 3.0320159122655803, "learning_rate": 1.214346916918847e-06, "loss": 0.0829, "step": 10880 }, { "epoch": 0.78, "grad_norm": 12.273863644646209, "learning_rate": 1.2135883735085973e-06, "loss": 0.7586, "step": 10881 }, { "epoch": 0.78, "grad_norm": 11.561126374671757, "learning_rate": 1.2128300343609228e-06, "loss": 0.404, "step": 10882 }, { "epoch": 0.78, "grad_norm": 11.512305438387456, "learning_rate": 1.2120718995167308e-06, "loss": 0.3063, "step": 10883 }, { "epoch": 0.78, "grad_norm": 45.88688560752062, "learning_rate": 1.2113139690169234e-06, "loss": 0.8713, "step": 10884 }, { "epoch": 0.78, "grad_norm": 42.43850943130593, "learning_rate": 1.210556242902387e-06, "loss": 0.588, "step": 10885 }, { "epoch": 0.78, "grad_norm": 3.382427232886164, "learning_rate": 1.2097987212139967e-06, "loss": 0.1844, "step": 10886 }, { "epoch": 0.78, "grad_norm": 2.852317745656201, "learning_rate": 1.209041403992618e-06, "loss": 0.1261, "step": 10887 }, { "epoch": 0.78, "grad_norm": 15.999672579482159, "learning_rate": 1.2082842912791065e-06, "loss": 0.4008, "step": 10888 }, { "epoch": 0.78, "grad_norm": 18.961985127666726, "learning_rate": 1.2075273831143058e-06, "loss": 0.4411, "step": 10889 }, { "epoch": 0.78, "grad_norm": 3.358032387735153, "learning_rate": 1.2067706795390472e-06, "loss": 0.148, "step": 10890 }, { "epoch": 0.78, "grad_norm": 4.531362092058744, "learning_rate": 1.2060141805941523e-06, "loss": 0.8349, "step": 10891 }, { "epoch": 0.78, "grad_norm": 4.621854912287872, "learning_rate": 1.20525788632043e-06, "loss": 0.7125, "step": 10892 }, { "epoch": 0.78, "grad_norm": 2.957323861613493, "learning_rate": 1.2045017967586826e-06, "loss": 0.1636, "step": 10893 }, { "epoch": 0.78, "grad_norm": 20.66989542324108, "learning_rate": 1.2037459119496959e-06, "loss": 1.3237, "step": 10894 }, { "epoch": 0.78, "grad_norm": 4.861151281972011, "learning_rate": 1.202990231934248e-06, "loss": 0.7917, "step": 10895 }, { "epoch": 0.78, "grad_norm": 3.473801538525049, "learning_rate": 1.202234756753103e-06, "loss": 0.0892, "step": 10896 }, { "epoch": 0.78, "grad_norm": 3.3001726172313126, "learning_rate": 1.201479486447019e-06, "loss": 0.1094, "step": 10897 }, { "epoch": 0.78, "grad_norm": 19.700867363706152, "learning_rate": 1.2007244210567387e-06, "loss": 0.8813, "step": 10898 }, { "epoch": 0.78, "grad_norm": 4.574633823970542, "learning_rate": 1.1999695606229932e-06, "loss": 0.9466, "step": 10899 }, { "epoch": 0.78, "grad_norm": 5.923409353197446, "learning_rate": 1.1992149051865076e-06, "loss": 0.8461, "step": 10900 }, { "epoch": 0.78, "grad_norm": 62.1372465168516, "learning_rate": 1.1984604547879913e-06, "loss": 0.7468, "step": 10901 }, { "epoch": 0.78, "grad_norm": 3.645784297463166, "learning_rate": 1.1977062094681436e-06, "loss": 0.1545, "step": 10902 }, { "epoch": 0.78, "grad_norm": 11.848652288046882, "learning_rate": 1.1969521692676533e-06, "loss": 0.3425, "step": 10903 }, { "epoch": 0.78, "grad_norm": 10.257909674586172, "learning_rate": 1.1961983342271981e-06, "loss": 0.4287, "step": 10904 }, { "epoch": 0.78, "grad_norm": 10.075777988012325, "learning_rate": 1.1954447043874434e-06, "loss": 0.3186, "step": 10905 }, { "epoch": 0.78, "grad_norm": 2.4850056701921344, "learning_rate": 1.1946912797890476e-06, "loss": 0.1784, "step": 10906 }, { "epoch": 0.78, "grad_norm": 2.414966153477718, "learning_rate": 1.193938060472653e-06, "loss": 0.0908, "step": 10907 }, { "epoch": 0.78, "grad_norm": 6.720563154512063, "learning_rate": 1.193185046478892e-06, "loss": 0.3038, "step": 10908 }, { "epoch": 0.78, "grad_norm": 28.078084452308637, "learning_rate": 1.1924322378483894e-06, "loss": 0.5323, "step": 10909 }, { "epoch": 0.78, "grad_norm": 3.460249522458672, "learning_rate": 1.1916796346217552e-06, "loss": 0.2379, "step": 10910 }, { "epoch": 0.78, "grad_norm": 9.85089496996007, "learning_rate": 1.1909272368395886e-06, "loss": 0.4033, "step": 10911 }, { "epoch": 0.78, "grad_norm": 6.507360263414868, "learning_rate": 1.1901750445424803e-06, "loss": 0.3296, "step": 10912 }, { "epoch": 0.78, "grad_norm": 4.267497311014455, "learning_rate": 1.1894230577710076e-06, "loss": 0.7193, "step": 10913 }, { "epoch": 0.78, "grad_norm": 4.653603085328101, "learning_rate": 1.1886712765657366e-06, "loss": 0.1878, "step": 10914 }, { "epoch": 0.78, "grad_norm": 2.8794839407556263, "learning_rate": 1.187919700967224e-06, "loss": 0.1209, "step": 10915 }, { "epoch": 0.78, "grad_norm": 2.037924404606032, "learning_rate": 1.1871683310160136e-06, "loss": 0.0763, "step": 10916 }, { "epoch": 0.78, "grad_norm": 3.0639193340898285, "learning_rate": 1.1864171667526385e-06, "loss": 0.1818, "step": 10917 }, { "epoch": 0.78, "grad_norm": 10.685397357651897, "learning_rate": 1.1856662082176235e-06, "loss": 0.3435, "step": 10918 }, { "epoch": 0.78, "grad_norm": 3.135715659321347, "learning_rate": 1.1849154554514786e-06, "loss": 0.1781, "step": 10919 }, { "epoch": 0.78, "grad_norm": 3.2689020026991393, "learning_rate": 1.1841649084947022e-06, "loss": 0.2215, "step": 10920 }, { "epoch": 0.78, "grad_norm": 4.468910277553638, "learning_rate": 1.1834145673877867e-06, "loss": 0.7997, "step": 10921 }, { "epoch": 0.78, "grad_norm": 11.004915073497472, "learning_rate": 1.182664432171209e-06, "loss": 0.4931, "step": 10922 }, { "epoch": 0.78, "grad_norm": 4.225836327173883, "learning_rate": 1.1819145028854357e-06, "loss": 0.1449, "step": 10923 }, { "epoch": 0.78, "grad_norm": 3.050119202522936, "learning_rate": 1.1811647795709208e-06, "loss": 0.1494, "step": 10924 }, { "epoch": 0.78, "grad_norm": 14.945496151070868, "learning_rate": 1.180415262268113e-06, "loss": 0.7471, "step": 10925 }, { "epoch": 0.78, "grad_norm": 4.270939371510511, "learning_rate": 1.1796659510174435e-06, "loss": 0.6832, "step": 10926 }, { "epoch": 0.78, "grad_norm": 4.145987489115378, "learning_rate": 1.1789168458593353e-06, "loss": 0.1967, "step": 10927 }, { "epoch": 0.78, "grad_norm": 6.352496107915023, "learning_rate": 1.1781679468342e-06, "loss": 0.2672, "step": 10928 }, { "epoch": 0.78, "grad_norm": 2.8441315417926254, "learning_rate": 1.1774192539824358e-06, "loss": 0.1238, "step": 10929 }, { "epoch": 0.78, "grad_norm": 7.50940114734407, "learning_rate": 1.176670767344435e-06, "loss": 0.3118, "step": 10930 }, { "epoch": 0.78, "grad_norm": 4.112060365981185, "learning_rate": 1.1759224869605746e-06, "loss": 0.9037, "step": 10931 }, { "epoch": 0.78, "grad_norm": 23.817431540116228, "learning_rate": 1.1751744128712205e-06, "loss": 1.4176, "step": 10932 }, { "epoch": 0.78, "grad_norm": 4.64448525291698, "learning_rate": 1.174426545116728e-06, "loss": 0.1789, "step": 10933 }, { "epoch": 0.78, "grad_norm": 4.772595602714017, "learning_rate": 1.1736788837374447e-06, "loss": 0.1749, "step": 10934 }, { "epoch": 0.78, "grad_norm": 4.0408795241856925, "learning_rate": 1.1729314287737014e-06, "loss": 0.1748, "step": 10935 }, { "epoch": 0.78, "grad_norm": 28.19830724397369, "learning_rate": 1.17218418026582e-06, "loss": 0.7117, "step": 10936 }, { "epoch": 0.78, "grad_norm": 3.34437024468429, "learning_rate": 1.1714371382541151e-06, "loss": 0.1758, "step": 10937 }, { "epoch": 0.78, "grad_norm": 3.4810190958635667, "learning_rate": 1.1706903027788858e-06, "loss": 0.2016, "step": 10938 }, { "epoch": 0.78, "grad_norm": 5.057744617160387, "learning_rate": 1.1699436738804166e-06, "loss": 0.9433, "step": 10939 }, { "epoch": 0.78, "grad_norm": 4.424388324636553, "learning_rate": 1.1691972515989902e-06, "loss": 0.8694, "step": 10940 }, { "epoch": 0.78, "grad_norm": 2.993985966631148, "learning_rate": 1.1684510359748725e-06, "loss": 0.1147, "step": 10941 }, { "epoch": 0.78, "grad_norm": 17.783633871661948, "learning_rate": 1.1677050270483158e-06, "loss": 0.461, "step": 10942 }, { "epoch": 0.78, "grad_norm": 3.051012807180511, "learning_rate": 1.1669592248595685e-06, "loss": 0.1749, "step": 10943 }, { "epoch": 0.78, "grad_norm": 23.878582155003063, "learning_rate": 1.1662136294488624e-06, "loss": 0.9555, "step": 10944 }, { "epoch": 0.78, "grad_norm": 11.816361247618554, "learning_rate": 1.1654682408564177e-06, "loss": 0.4978, "step": 10945 }, { "epoch": 0.78, "grad_norm": 4.537984190505659, "learning_rate": 1.164723059122449e-06, "loss": 1.0185, "step": 10946 }, { "epoch": 0.78, "grad_norm": 2.927629236117007, "learning_rate": 1.1639780842871534e-06, "loss": 0.1528, "step": 10947 }, { "epoch": 0.79, "grad_norm": 2.840734634562666, "learning_rate": 1.1632333163907184e-06, "loss": 0.1727, "step": 10948 }, { "epoch": 0.79, "grad_norm": 3.1248927172601686, "learning_rate": 1.162488755473325e-06, "loss": 0.1554, "step": 10949 }, { "epoch": 0.79, "grad_norm": 2.78991082985156, "learning_rate": 1.1617444015751384e-06, "loss": 0.18, "step": 10950 }, { "epoch": 0.79, "grad_norm": 15.335220173154934, "learning_rate": 1.1610002547363103e-06, "loss": 0.5958, "step": 10951 }, { "epoch": 0.79, "grad_norm": 17.173340780059746, "learning_rate": 1.1602563149969881e-06, "loss": 0.6625, "step": 10952 }, { "epoch": 0.79, "grad_norm": 23.11493595872363, "learning_rate": 1.1595125823973035e-06, "loss": 1.3588, "step": 10953 }, { "epoch": 0.79, "grad_norm": 21.95812454596423, "learning_rate": 1.158769056977377e-06, "loss": 1.0473, "step": 10954 }, { "epoch": 0.79, "grad_norm": 11.954472571434714, "learning_rate": 1.1580257387773208e-06, "loss": 0.4763, "step": 10955 }, { "epoch": 0.79, "grad_norm": 3.0885838159617807, "learning_rate": 1.157282627837234e-06, "loss": 0.1979, "step": 10956 }, { "epoch": 0.79, "grad_norm": 9.702977737898735, "learning_rate": 1.1565397241972021e-06, "loss": 0.3478, "step": 10957 }, { "epoch": 0.79, "grad_norm": 2.9102772985543357, "learning_rate": 1.1557970278973058e-06, "loss": 0.2217, "step": 10958 }, { "epoch": 0.79, "grad_norm": 2.328669300502002, "learning_rate": 1.155054538977608e-06, "loss": 0.1067, "step": 10959 }, { "epoch": 0.79, "grad_norm": 3.8811208456586637, "learning_rate": 1.1543122574781645e-06, "loss": 0.1923, "step": 10960 }, { "epoch": 0.79, "grad_norm": 16.525896170586044, "learning_rate": 1.1535701834390161e-06, "loss": 0.913, "step": 10961 }, { "epoch": 0.79, "grad_norm": 4.220122241436668, "learning_rate": 1.1528283169001997e-06, "loss": 0.9975, "step": 10962 }, { "epoch": 0.79, "grad_norm": 18.037395076690274, "learning_rate": 1.1520866579017309e-06, "loss": 0.9248, "step": 10963 }, { "epoch": 0.79, "grad_norm": 4.025373183459858, "learning_rate": 1.1513452064836227e-06, "loss": 0.1432, "step": 10964 }, { "epoch": 0.79, "grad_norm": 3.7031278563764776, "learning_rate": 1.1506039626858729e-06, "loss": 0.1271, "step": 10965 }, { "epoch": 0.79, "grad_norm": 3.063828830897891, "learning_rate": 1.1498629265484684e-06, "loss": 0.1181, "step": 10966 }, { "epoch": 0.79, "grad_norm": 14.811869107498616, "learning_rate": 1.1491220981113842e-06, "loss": 0.6261, "step": 10967 }, { "epoch": 0.79, "grad_norm": 4.2161511847360265, "learning_rate": 1.148381477414588e-06, "loss": 0.1862, "step": 10968 }, { "epoch": 0.79, "grad_norm": 3.4508763649150813, "learning_rate": 1.1476410644980314e-06, "loss": 0.2764, "step": 10969 }, { "epoch": 0.79, "grad_norm": 7.479215412494044, "learning_rate": 1.1469008594016563e-06, "loss": 0.3884, "step": 10970 }, { "epoch": 0.79, "grad_norm": 12.981833827345401, "learning_rate": 1.1461608621653964e-06, "loss": 0.5922, "step": 10971 }, { "epoch": 0.79, "grad_norm": 66.70889237218078, "learning_rate": 1.1454210728291704e-06, "loss": 1.2634, "step": 10972 }, { "epoch": 0.79, "grad_norm": 3.541347309254879, "learning_rate": 1.1446814914328858e-06, "loss": 0.1035, "step": 10973 }, { "epoch": 0.79, "grad_norm": 11.555481410723425, "learning_rate": 1.1439421180164446e-06, "loss": 0.8764, "step": 10974 }, { "epoch": 0.79, "grad_norm": 4.837883535270541, "learning_rate": 1.1432029526197285e-06, "loss": 0.936, "step": 10975 }, { "epoch": 0.79, "grad_norm": 3.5627852012478254, "learning_rate": 1.1424639952826128e-06, "loss": 0.1535, "step": 10976 }, { "epoch": 0.79, "grad_norm": 5.333203529748305, "learning_rate": 1.1417252460449646e-06, "loss": 0.7642, "step": 10977 }, { "epoch": 0.79, "grad_norm": 8.94320538808635, "learning_rate": 1.140986704946635e-06, "loss": 0.5829, "step": 10978 }, { "epoch": 0.79, "grad_norm": 28.575803760248192, "learning_rate": 1.1402483720274642e-06, "loss": 0.9368, "step": 10979 }, { "epoch": 0.79, "grad_norm": 3.44047509824729, "learning_rate": 1.1395102473272851e-06, "loss": 0.1342, "step": 10980 }, { "epoch": 0.79, "grad_norm": 4.993842016374861, "learning_rate": 1.138772330885915e-06, "loss": 0.926, "step": 10981 }, { "epoch": 0.79, "grad_norm": 4.9143903026352165, "learning_rate": 1.1380346227431615e-06, "loss": 0.9747, "step": 10982 }, { "epoch": 0.79, "grad_norm": 4.559859475558607, "learning_rate": 1.1372971229388223e-06, "loss": 0.814, "step": 10983 }, { "epoch": 0.79, "grad_norm": 7.5874156332416005, "learning_rate": 1.136559831512683e-06, "loss": 0.3561, "step": 10984 }, { "epoch": 0.79, "grad_norm": 12.31572371598341, "learning_rate": 1.1358227485045148e-06, "loss": 0.7025, "step": 10985 }, { "epoch": 0.79, "grad_norm": 8.859439992795235, "learning_rate": 1.1350858739540859e-06, "loss": 0.5204, "step": 10986 }, { "epoch": 0.79, "grad_norm": 2.6359179076414168, "learning_rate": 1.134349207901143e-06, "loss": 0.1258, "step": 10987 }, { "epoch": 0.79, "grad_norm": 3.315547823911087, "learning_rate": 1.1336127503854266e-06, "loss": 0.2403, "step": 10988 }, { "epoch": 0.79, "grad_norm": 2.754265115550695, "learning_rate": 1.1328765014466692e-06, "loss": 0.1853, "step": 10989 }, { "epoch": 0.79, "grad_norm": 4.390050088967316, "learning_rate": 1.1321404611245857e-06, "loss": 0.7606, "step": 10990 }, { "epoch": 0.79, "grad_norm": 24.60395161248869, "learning_rate": 1.1314046294588831e-06, "loss": 0.3561, "step": 10991 }, { "epoch": 0.79, "grad_norm": 8.15904899918894, "learning_rate": 1.1306690064892583e-06, "loss": 0.3752, "step": 10992 }, { "epoch": 0.79, "grad_norm": 2.804026100055878, "learning_rate": 1.1299335922553945e-06, "loss": 0.1736, "step": 10993 }, { "epoch": 0.79, "grad_norm": 2.2387639098500736, "learning_rate": 1.1291983867969646e-06, "loss": 0.1009, "step": 10994 }, { "epoch": 0.79, "grad_norm": 3.6458814957501633, "learning_rate": 1.128463390153628e-06, "loss": 0.1809, "step": 10995 }, { "epoch": 0.79, "grad_norm": 2.3893934440489275, "learning_rate": 1.127728602365039e-06, "loss": 0.0957, "step": 10996 }, { "epoch": 0.79, "grad_norm": 3.4153310966772685, "learning_rate": 1.1269940234708343e-06, "loss": 0.196, "step": 10997 }, { "epoch": 0.79, "grad_norm": 3.002545568372544, "learning_rate": 1.1262596535106418e-06, "loss": 0.1628, "step": 10998 }, { "epoch": 0.79, "grad_norm": 2.26143501131002, "learning_rate": 1.1255254925240783e-06, "loss": 0.1003, "step": 10999 }, { "epoch": 0.79, "grad_norm": 4.256829083966532, "learning_rate": 1.124791540550747e-06, "loss": 0.7582, "step": 11000 }, { "epoch": 0.79, "grad_norm": 2.173204579577306, "learning_rate": 1.1240577976302453e-06, "loss": 0.1016, "step": 11001 }, { "epoch": 0.79, "grad_norm": 3.0538829034607446, "learning_rate": 1.1233242638021541e-06, "loss": 0.1964, "step": 11002 }, { "epoch": 0.79, "grad_norm": 2.747789214361793, "learning_rate": 1.1225909391060453e-06, "loss": 0.1134, "step": 11003 }, { "epoch": 0.79, "grad_norm": 4.4873193578609065, "learning_rate": 1.1218578235814765e-06, "loss": 0.8989, "step": 11004 }, { "epoch": 0.79, "grad_norm": 15.214909593679923, "learning_rate": 1.121124917268e-06, "loss": 0.2958, "step": 11005 }, { "epoch": 0.79, "grad_norm": 2.1305494603433623, "learning_rate": 1.120392220205152e-06, "loss": 0.0822, "step": 11006 }, { "epoch": 0.79, "grad_norm": 4.306614926265992, "learning_rate": 1.1196597324324576e-06, "loss": 0.7973, "step": 11007 }, { "epoch": 0.79, "grad_norm": 4.270627180045432, "learning_rate": 1.1189274539894335e-06, "loss": 0.7202, "step": 11008 }, { "epoch": 0.79, "grad_norm": 4.117671791994789, "learning_rate": 1.118195384915583e-06, "loss": 0.1333, "step": 11009 }, { "epoch": 0.79, "grad_norm": 54.07767674824041, "learning_rate": 1.117463525250398e-06, "loss": 0.617, "step": 11010 }, { "epoch": 0.79, "grad_norm": 19.963232375623022, "learning_rate": 1.11673187503336e-06, "loss": 0.6265, "step": 11011 }, { "epoch": 0.79, "grad_norm": 4.224485255589866, "learning_rate": 1.1160004343039377e-06, "loss": 0.6302, "step": 11012 }, { "epoch": 0.79, "grad_norm": 13.299076295837866, "learning_rate": 1.1152692031015895e-06, "loss": 0.8115, "step": 11013 }, { "epoch": 0.79, "grad_norm": 3.120356971910358, "learning_rate": 1.1145381814657641e-06, "loss": 0.1444, "step": 11014 }, { "epoch": 0.79, "grad_norm": 2.777644564194622, "learning_rate": 1.1138073694358965e-06, "loss": 0.1076, "step": 11015 }, { "epoch": 0.79, "grad_norm": 3.3658670316615087, "learning_rate": 1.1130767670514103e-06, "loss": 0.1811, "step": 11016 }, { "epoch": 0.79, "grad_norm": 13.28177063212878, "learning_rate": 1.112346374351721e-06, "loss": 0.8453, "step": 11017 }, { "epoch": 0.79, "grad_norm": 3.3733409264863785, "learning_rate": 1.1116161913762292e-06, "loss": 0.1212, "step": 11018 }, { "epoch": 0.79, "grad_norm": 7.423653144959766, "learning_rate": 1.1108862181643238e-06, "loss": 0.4611, "step": 11019 }, { "epoch": 0.79, "grad_norm": 11.831197071828214, "learning_rate": 1.1101564547553878e-06, "loss": 0.7028, "step": 11020 }, { "epoch": 0.79, "grad_norm": 6.592162275013737, "learning_rate": 1.1094269011887871e-06, "loss": 0.3493, "step": 11021 }, { "epoch": 0.79, "grad_norm": 3.735339330532026, "learning_rate": 1.1086975575038783e-06, "loss": 0.1943, "step": 11022 }, { "epoch": 0.79, "grad_norm": 14.44671343291347, "learning_rate": 1.1079684237400069e-06, "loss": 0.3698, "step": 11023 }, { "epoch": 0.79, "grad_norm": 3.2860552792233264, "learning_rate": 1.1072394999365065e-06, "loss": 0.2479, "step": 11024 }, { "epoch": 0.79, "grad_norm": 2.935754350047992, "learning_rate": 1.1065107861326996e-06, "loss": 0.1722, "step": 11025 }, { "epoch": 0.79, "grad_norm": 3.2307398706794883, "learning_rate": 1.1057822823678988e-06, "loss": 0.1915, "step": 11026 }, { "epoch": 0.79, "grad_norm": 2.6478706435821557, "learning_rate": 1.1050539886814039e-06, "loss": 0.1413, "step": 11027 }, { "epoch": 0.79, "grad_norm": 4.100914260242588, "learning_rate": 1.1043259051125017e-06, "loss": 0.281, "step": 11028 }, { "epoch": 0.79, "grad_norm": 4.59488162346132, "learning_rate": 1.1035980317004724e-06, "loss": 0.7735, "step": 11029 }, { "epoch": 0.79, "grad_norm": 3.6604563016465894, "learning_rate": 1.1028703684845805e-06, "loss": 0.191, "step": 11030 }, { "epoch": 0.79, "grad_norm": 17.967498976100092, "learning_rate": 1.1021429155040809e-06, "loss": 0.6011, "step": 11031 }, { "epoch": 0.79, "grad_norm": 3.6460337940374683, "learning_rate": 1.101415672798215e-06, "loss": 0.1288, "step": 11032 }, { "epoch": 0.79, "grad_norm": 5.266184020620794, "learning_rate": 1.1006886404062183e-06, "loss": 0.7733, "step": 11033 }, { "epoch": 0.79, "grad_norm": 10.257327943537282, "learning_rate": 1.09996181836731e-06, "loss": 0.3962, "step": 11034 }, { "epoch": 0.79, "grad_norm": 10.14798577113647, "learning_rate": 1.099235206720699e-06, "loss": 0.4612, "step": 11035 }, { "epoch": 0.79, "grad_norm": 3.248166837341038, "learning_rate": 1.0985088055055832e-06, "loss": 0.1888, "step": 11036 }, { "epoch": 0.79, "grad_norm": 4.1823987181751185, "learning_rate": 1.0977826147611482e-06, "loss": 0.1742, "step": 11037 }, { "epoch": 0.79, "grad_norm": 33.76047379421178, "learning_rate": 1.0970566345265714e-06, "loss": 0.6575, "step": 11038 }, { "epoch": 0.79, "grad_norm": 3.1130476351956515, "learning_rate": 1.0963308648410164e-06, "loss": 0.157, "step": 11039 }, { "epoch": 0.79, "grad_norm": 4.997978761766487, "learning_rate": 1.0956053057436344e-06, "loss": 0.7422, "step": 11040 }, { "epoch": 0.79, "grad_norm": 2.9360031331882257, "learning_rate": 1.0948799572735664e-06, "loss": 0.1328, "step": 11041 }, { "epoch": 0.79, "grad_norm": 13.139046872160064, "learning_rate": 1.0941548194699442e-06, "loss": 0.5339, "step": 11042 }, { "epoch": 0.79, "grad_norm": 13.565666209344762, "learning_rate": 1.093429892371885e-06, "loss": 0.57, "step": 11043 }, { "epoch": 0.79, "grad_norm": 10.902366364347657, "learning_rate": 1.0927051760184948e-06, "loss": 0.3312, "step": 11044 }, { "epoch": 0.79, "grad_norm": 13.505777959130318, "learning_rate": 1.0919806704488716e-06, "loss": 0.2725, "step": 11045 }, { "epoch": 0.79, "grad_norm": 11.529762839916867, "learning_rate": 1.0912563757021e-06, "loss": 0.3952, "step": 11046 }, { "epoch": 0.79, "grad_norm": 16.386417014216157, "learning_rate": 1.090532291817249e-06, "loss": 0.5358, "step": 11047 }, { "epoch": 0.79, "grad_norm": 3.3645372643803184, "learning_rate": 1.0898084188333834e-06, "loss": 0.2046, "step": 11048 }, { "epoch": 0.79, "grad_norm": 8.32523633266304, "learning_rate": 1.0890847567895536e-06, "loss": 0.608, "step": 11049 }, { "epoch": 0.79, "grad_norm": 9.244564832791667, "learning_rate": 1.0883613057247955e-06, "loss": 0.6382, "step": 11050 }, { "epoch": 0.79, "grad_norm": 2.4283343027090187, "learning_rate": 1.0876380656781404e-06, "loss": 0.1004, "step": 11051 }, { "epoch": 0.79, "grad_norm": 30.236202339546427, "learning_rate": 1.0869150366886023e-06, "loss": 0.9899, "step": 11052 }, { "epoch": 0.79, "grad_norm": 3.772890409085906, "learning_rate": 1.0861922187951846e-06, "loss": 0.0845, "step": 11053 }, { "epoch": 0.79, "grad_norm": 8.065530089029846, "learning_rate": 1.085469612036884e-06, "loss": 0.5766, "step": 11054 }, { "epoch": 0.79, "grad_norm": 10.27060080942993, "learning_rate": 1.0847472164526802e-06, "loss": 0.5112, "step": 11055 }, { "epoch": 0.79, "grad_norm": 2.8033668051573706, "learning_rate": 1.0840250320815427e-06, "loss": 0.1027, "step": 11056 }, { "epoch": 0.79, "grad_norm": 11.453203444891013, "learning_rate": 1.0833030589624333e-06, "loss": 0.6061, "step": 11057 }, { "epoch": 0.79, "grad_norm": 10.171034784045155, "learning_rate": 1.0825812971342997e-06, "loss": 0.6554, "step": 11058 }, { "epoch": 0.79, "grad_norm": 9.857734815615986, "learning_rate": 1.0818597466360742e-06, "loss": 0.5898, "step": 11059 }, { "epoch": 0.79, "grad_norm": 4.731411080398899, "learning_rate": 1.081138407506686e-06, "loss": 0.8169, "step": 11060 }, { "epoch": 0.79, "grad_norm": 9.104375374412118, "learning_rate": 1.080417279785047e-06, "loss": 0.4698, "step": 11061 }, { "epoch": 0.79, "grad_norm": 3.5070542918943115, "learning_rate": 1.0796963635100577e-06, "loss": 0.1866, "step": 11062 }, { "epoch": 0.79, "grad_norm": 12.133241229709817, "learning_rate": 1.0789756587206118e-06, "loss": 0.3235, "step": 11063 }, { "epoch": 0.79, "grad_norm": 2.931579139955609, "learning_rate": 1.0782551654555873e-06, "loss": 0.127, "step": 11064 }, { "epoch": 0.79, "grad_norm": 2.822570092454609, "learning_rate": 1.0775348837538507e-06, "loss": 0.1762, "step": 11065 }, { "epoch": 0.79, "grad_norm": 5.1089702855027825, "learning_rate": 1.076814813654261e-06, "loss": 1.0011, "step": 11066 }, { "epoch": 0.79, "grad_norm": 4.039941813803772, "learning_rate": 1.0760949551956623e-06, "loss": 0.8283, "step": 11067 }, { "epoch": 0.79, "grad_norm": 3.0449754475203177, "learning_rate": 1.0753753084168877e-06, "loss": 0.115, "step": 11068 }, { "epoch": 0.79, "grad_norm": 13.635419531874135, "learning_rate": 1.0746558733567586e-06, "loss": 0.9152, "step": 11069 }, { "epoch": 0.79, "grad_norm": 27.15073586561342, "learning_rate": 1.0739366500540898e-06, "loss": 0.7564, "step": 11070 }, { "epoch": 0.79, "grad_norm": 11.585696538862837, "learning_rate": 1.0732176385476745e-06, "loss": 0.5005, "step": 11071 }, { "epoch": 0.79, "grad_norm": 2.3757803191785802, "learning_rate": 1.0724988388763057e-06, "loss": 0.1441, "step": 11072 }, { "epoch": 0.79, "grad_norm": 2.660847515404371, "learning_rate": 1.0717802510787579e-06, "loss": 0.0768, "step": 11073 }, { "epoch": 0.79, "grad_norm": 4.087150429304298, "learning_rate": 1.0710618751937968e-06, "loss": 0.7739, "step": 11074 }, { "epoch": 0.79, "grad_norm": 21.0596470095541, "learning_rate": 1.0703437112601738e-06, "loss": 0.4832, "step": 11075 }, { "epoch": 0.79, "grad_norm": 2.7211237766599385, "learning_rate": 1.0696257593166347e-06, "loss": 0.1175, "step": 11076 }, { "epoch": 0.79, "grad_norm": 3.7851809247914745, "learning_rate": 1.0689080194019087e-06, "loss": 0.2107, "step": 11077 }, { "epoch": 0.79, "grad_norm": 3.471104628170101, "learning_rate": 1.0681904915547137e-06, "loss": 0.2076, "step": 11078 }, { "epoch": 0.79, "grad_norm": 5.996686933514308, "learning_rate": 1.0674731758137597e-06, "loss": 0.8805, "step": 11079 }, { "epoch": 0.79, "grad_norm": 17.90439323803486, "learning_rate": 1.0667560722177434e-06, "loss": 0.6151, "step": 11080 }, { "epoch": 0.79, "grad_norm": 4.522347683803841, "learning_rate": 1.0660391808053467e-06, "loss": 0.9511, "step": 11081 }, { "epoch": 0.79, "grad_norm": 3.2432729781903635, "learning_rate": 1.0653225016152485e-06, "loss": 0.0904, "step": 11082 }, { "epoch": 0.79, "grad_norm": 3.558624374197309, "learning_rate": 1.0646060346861064e-06, "loss": 0.1489, "step": 11083 }, { "epoch": 0.79, "grad_norm": 16.2434487589395, "learning_rate": 1.0638897800565707e-06, "loss": 0.4649, "step": 11084 }, { "epoch": 0.79, "grad_norm": 10.482274470326345, "learning_rate": 1.063173737765284e-06, "loss": 0.757, "step": 11085 }, { "epoch": 0.79, "grad_norm": 3.0049829460304176, "learning_rate": 1.0624579078508728e-06, "loss": 0.1697, "step": 11086 }, { "epoch": 0.8, "grad_norm": 22.850517844270907, "learning_rate": 1.0617422903519513e-06, "loss": 0.4976, "step": 11087 }, { "epoch": 0.8, "grad_norm": 4.003623025907629, "learning_rate": 1.0610268853071277e-06, "loss": 0.2595, "step": 11088 }, { "epoch": 0.8, "grad_norm": 4.958625617005108, "learning_rate": 1.0603116927549934e-06, "loss": 0.789, "step": 11089 }, { "epoch": 0.8, "grad_norm": 2.753761281250887, "learning_rate": 1.05959671273413e-06, "loss": 0.1058, "step": 11090 }, { "epoch": 0.8, "grad_norm": 6.739365915608794, "learning_rate": 1.0588819452831095e-06, "loss": 0.4508, "step": 11091 }, { "epoch": 0.8, "grad_norm": 2.4933934461532554, "learning_rate": 1.0581673904404903e-06, "loss": 0.1152, "step": 11092 }, { "epoch": 0.8, "grad_norm": 2.9581286758651775, "learning_rate": 1.0574530482448185e-06, "loss": 0.1635, "step": 11093 }, { "epoch": 0.8, "grad_norm": 2.8631900482895007, "learning_rate": 1.0567389187346343e-06, "loss": 0.0995, "step": 11094 }, { "epoch": 0.8, "grad_norm": 21.073186449879575, "learning_rate": 1.0560250019484574e-06, "loss": 0.4152, "step": 11095 }, { "epoch": 0.8, "grad_norm": 4.0317039998447095, "learning_rate": 1.055311297924802e-06, "loss": 0.1891, "step": 11096 }, { "epoch": 0.8, "grad_norm": 3.303234305687881, "learning_rate": 1.0545978067021722e-06, "loss": 0.1566, "step": 11097 }, { "epoch": 0.8, "grad_norm": 5.80606528044855, "learning_rate": 1.0538845283190563e-06, "loss": 0.1744, "step": 11098 }, { "epoch": 0.8, "grad_norm": 6.816181630634084, "learning_rate": 1.0531714628139317e-06, "loss": 0.2154, "step": 11099 }, { "epoch": 0.8, "grad_norm": 1.8838151546436501, "learning_rate": 1.0524586102252687e-06, "loss": 0.1266, "step": 11100 }, { "epoch": 0.8, "grad_norm": 9.302432969072848, "learning_rate": 1.0517459705915211e-06, "loss": 0.277, "step": 11101 }, { "epoch": 0.8, "grad_norm": 2.2988329659717293, "learning_rate": 1.0510335439511327e-06, "loss": 0.1066, "step": 11102 }, { "epoch": 0.8, "grad_norm": 3.335885106892875, "learning_rate": 1.050321330342538e-06, "loss": 0.2669, "step": 11103 }, { "epoch": 0.8, "grad_norm": 7.035919267073583, "learning_rate": 1.0496093298041571e-06, "loss": 0.3642, "step": 11104 }, { "epoch": 0.8, "grad_norm": 3.7000508288681373, "learning_rate": 1.0488975423743996e-06, "loss": 0.1769, "step": 11105 }, { "epoch": 0.8, "grad_norm": 72.59243633225833, "learning_rate": 1.048185968091664e-06, "loss": 0.6831, "step": 11106 }, { "epoch": 0.8, "grad_norm": 11.561072199747136, "learning_rate": 1.0474746069943376e-06, "loss": 0.6435, "step": 11107 }, { "epoch": 0.8, "grad_norm": 5.9312074791566785, "learning_rate": 1.0467634591207932e-06, "loss": 0.2832, "step": 11108 }, { "epoch": 0.8, "grad_norm": 4.222076360639854, "learning_rate": 1.0460525245093978e-06, "loss": 0.3256, "step": 11109 }, { "epoch": 0.8, "grad_norm": 3.3930861875837777, "learning_rate": 1.0453418031985018e-06, "loss": 0.2726, "step": 11110 }, { "epoch": 0.8, "grad_norm": 17.075135315682243, "learning_rate": 1.044631295226447e-06, "loss": 0.5938, "step": 11111 }, { "epoch": 0.8, "grad_norm": 3.6325612060096546, "learning_rate": 1.0439210006315603e-06, "loss": 0.2329, "step": 11112 }, { "epoch": 0.8, "grad_norm": 3.598827343047673, "learning_rate": 1.0432109194521622e-06, "loss": 0.1175, "step": 11113 }, { "epoch": 0.8, "grad_norm": 19.677344198250804, "learning_rate": 1.0425010517265583e-06, "loss": 1.404, "step": 11114 }, { "epoch": 0.8, "grad_norm": 3.5445547009268505, "learning_rate": 1.0417913974930405e-06, "loss": 0.1732, "step": 11115 }, { "epoch": 0.8, "grad_norm": 4.580522304455954, "learning_rate": 1.0410819567898962e-06, "loss": 0.637, "step": 11116 }, { "epoch": 0.8, "grad_norm": 15.565971227595075, "learning_rate": 1.040372729655395e-06, "loss": 0.4323, "step": 11117 }, { "epoch": 0.8, "grad_norm": 13.05894722207939, "learning_rate": 1.0396637161277955e-06, "loss": 0.5592, "step": 11118 }, { "epoch": 0.8, "grad_norm": 10.65904989162703, "learning_rate": 1.0389549162453506e-06, "loss": 0.6331, "step": 11119 }, { "epoch": 0.8, "grad_norm": 13.955351620465741, "learning_rate": 1.038246330046293e-06, "loss": 0.2331, "step": 11120 }, { "epoch": 0.8, "grad_norm": 9.19742177244975, "learning_rate": 1.0375379575688489e-06, "loss": 0.3677, "step": 11121 }, { "epoch": 0.8, "grad_norm": 2.2270096356824434, "learning_rate": 1.0368297988512343e-06, "loss": 0.1082, "step": 11122 }, { "epoch": 0.8, "grad_norm": 13.66959224036218, "learning_rate": 1.0361218539316504e-06, "loss": 0.6315, "step": 11123 }, { "epoch": 0.8, "grad_norm": 12.551005708804496, "learning_rate": 1.0354141228482872e-06, "loss": 0.9837, "step": 11124 }, { "epoch": 0.8, "grad_norm": 9.113341212156055, "learning_rate": 1.0347066056393269e-06, "loss": 0.4632, "step": 11125 }, { "epoch": 0.8, "grad_norm": 16.586930220892302, "learning_rate": 1.0339993023429357e-06, "loss": 0.6613, "step": 11126 }, { "epoch": 0.8, "grad_norm": 4.460328333722019, "learning_rate": 1.0332922129972689e-06, "loss": 0.7493, "step": 11127 }, { "epoch": 0.8, "grad_norm": 20.385367169383812, "learning_rate": 1.032585337640473e-06, "loss": 0.5463, "step": 11128 }, { "epoch": 0.8, "grad_norm": 4.6520568566542355, "learning_rate": 1.0318786763106815e-06, "loss": 0.8503, "step": 11129 }, { "epoch": 0.8, "grad_norm": 22.199373962407027, "learning_rate": 1.0311722290460135e-06, "loss": 0.8102, "step": 11130 }, { "epoch": 0.8, "grad_norm": 18.67887045772859, "learning_rate": 1.0304659958845835e-06, "loss": 0.9716, "step": 11131 }, { "epoch": 0.8, "grad_norm": 3.684347952437074, "learning_rate": 1.0297599768644866e-06, "loss": 0.2069, "step": 11132 }, { "epoch": 0.8, "grad_norm": 3.597962165329852, "learning_rate": 1.029054172023809e-06, "loss": 0.1477, "step": 11133 }, { "epoch": 0.8, "grad_norm": 3.586320831573965, "learning_rate": 1.0283485814006293e-06, "loss": 0.1704, "step": 11134 }, { "epoch": 0.8, "grad_norm": 3.6487006605123615, "learning_rate": 1.0276432050330104e-06, "loss": 0.1434, "step": 11135 }, { "epoch": 0.8, "grad_norm": 14.63405950948327, "learning_rate": 1.026938042959003e-06, "loss": 0.9592, "step": 11136 }, { "epoch": 0.8, "grad_norm": 8.970361304101559, "learning_rate": 1.026233095216651e-06, "loss": 0.5543, "step": 11137 }, { "epoch": 0.8, "grad_norm": 2.8113516155354037, "learning_rate": 1.0255283618439815e-06, "loss": 0.2202, "step": 11138 }, { "epoch": 0.8, "grad_norm": 3.956516006100425, "learning_rate": 1.0248238428790125e-06, "loss": 0.2396, "step": 11139 }, { "epoch": 0.8, "grad_norm": 3.6685605643141255, "learning_rate": 1.0241195383597497e-06, "loss": 0.2286, "step": 11140 }, { "epoch": 0.8, "grad_norm": 21.235679886981348, "learning_rate": 1.0234154483241893e-06, "loss": 0.8556, "step": 11141 }, { "epoch": 0.8, "grad_norm": 4.846163095381687, "learning_rate": 1.0227115728103126e-06, "loss": 0.8713, "step": 11142 }, { "epoch": 0.8, "grad_norm": 4.21254443012837, "learning_rate": 1.0220079118560927e-06, "loss": 0.1341, "step": 11143 }, { "epoch": 0.8, "grad_norm": 3.939032441856249, "learning_rate": 1.021304465499488e-06, "loss": 0.1959, "step": 11144 }, { "epoch": 0.8, "grad_norm": 20.824916269477185, "learning_rate": 1.020601233778446e-06, "loss": 0.7531, "step": 11145 }, { "epoch": 0.8, "grad_norm": 3.5470261031132035, "learning_rate": 1.0198982167309056e-06, "loss": 0.1134, "step": 11146 }, { "epoch": 0.8, "grad_norm": 3.5208494657124003, "learning_rate": 1.0191954143947913e-06, "loss": 0.1454, "step": 11147 }, { "epoch": 0.8, "grad_norm": 7.802479843242074, "learning_rate": 1.0184928268080157e-06, "loss": 0.5272, "step": 11148 }, { "epoch": 0.8, "grad_norm": 2.779127376015473, "learning_rate": 1.0177904540084804e-06, "loss": 0.1223, "step": 11149 }, { "epoch": 0.8, "grad_norm": 3.7016277739231014, "learning_rate": 1.0170882960340777e-06, "loss": 0.2445, "step": 11150 }, { "epoch": 0.8, "grad_norm": 3.4065458078669937, "learning_rate": 1.0163863529226852e-06, "loss": 0.2213, "step": 11151 }, { "epoch": 0.8, "grad_norm": 15.860588237316794, "learning_rate": 1.0156846247121694e-06, "loss": 0.9133, "step": 11152 }, { "epoch": 0.8, "grad_norm": 11.403047739780622, "learning_rate": 1.0149831114403873e-06, "loss": 0.8792, "step": 11153 }, { "epoch": 0.8, "grad_norm": 3.7192885756144736, "learning_rate": 1.0142818131451827e-06, "loss": 0.1944, "step": 11154 }, { "epoch": 0.8, "grad_norm": 3.833269409983314, "learning_rate": 1.0135807298643874e-06, "loss": 0.2077, "step": 11155 }, { "epoch": 0.8, "grad_norm": 4.579316207690428, "learning_rate": 1.0128798616358227e-06, "loss": 0.2621, "step": 11156 }, { "epoch": 0.8, "grad_norm": 4.543504712725051, "learning_rate": 1.0121792084972964e-06, "loss": 0.8319, "step": 11157 }, { "epoch": 0.8, "grad_norm": 19.118418229716184, "learning_rate": 1.0114787704866063e-06, "loss": 0.5794, "step": 11158 }, { "epoch": 0.8, "grad_norm": 4.526123823500318, "learning_rate": 1.010778547641541e-06, "loss": 0.7247, "step": 11159 }, { "epoch": 0.8, "grad_norm": 2.783625752286263, "learning_rate": 1.0100785399998725e-06, "loss": 0.0904, "step": 11160 }, { "epoch": 0.8, "grad_norm": 3.49558313935303, "learning_rate": 1.0093787475993627e-06, "loss": 0.1701, "step": 11161 }, { "epoch": 0.8, "grad_norm": 4.394670976627208, "learning_rate": 1.0086791704777653e-06, "loss": 0.9163, "step": 11162 }, { "epoch": 0.8, "grad_norm": 4.258803528070314, "learning_rate": 1.0079798086728194e-06, "loss": 0.8961, "step": 11163 }, { "epoch": 0.8, "grad_norm": 10.906308891814241, "learning_rate": 1.0072806622222502e-06, "loss": 0.6422, "step": 11164 }, { "epoch": 0.8, "grad_norm": 3.227599790477113, "learning_rate": 1.0065817311637776e-06, "loss": 0.1922, "step": 11165 }, { "epoch": 0.8, "grad_norm": 3.459761221981098, "learning_rate": 1.0058830155351047e-06, "loss": 0.1463, "step": 11166 }, { "epoch": 0.8, "grad_norm": 4.40311253123029, "learning_rate": 1.0051845153739248e-06, "loss": 0.1151, "step": 11167 }, { "epoch": 0.8, "grad_norm": 13.03745585232755, "learning_rate": 1.0044862307179188e-06, "loss": 0.4526, "step": 11168 }, { "epoch": 0.8, "grad_norm": 4.296857785226154, "learning_rate": 1.0037881616047567e-06, "loss": 0.6669, "step": 11169 }, { "epoch": 0.8, "grad_norm": 3.2469380417478946, "learning_rate": 1.0030903080720956e-06, "loss": 0.1916, "step": 11170 }, { "epoch": 0.8, "grad_norm": 13.774591472658393, "learning_rate": 1.0023926701575848e-06, "loss": 0.813, "step": 11171 }, { "epoch": 0.8, "grad_norm": 4.507710933303915, "learning_rate": 1.0016952478988573e-06, "loss": 0.7746, "step": 11172 }, { "epoch": 0.8, "grad_norm": 7.912010806215655, "learning_rate": 1.0009980413335358e-06, "loss": 0.3192, "step": 11173 }, { "epoch": 0.8, "grad_norm": 12.582418693927158, "learning_rate": 1.0003010504992338e-06, "loss": 0.6717, "step": 11174 }, { "epoch": 0.8, "grad_norm": 19.111391455522295, "learning_rate": 9.996042754335511e-07, "loss": 0.3884, "step": 11175 }, { "epoch": 0.8, "grad_norm": 8.070242096938799, "learning_rate": 9.989077161740751e-07, "loss": 0.5188, "step": 11176 }, { "epoch": 0.8, "grad_norm": 3.066539198629823, "learning_rate": 9.982113727583814e-07, "loss": 0.2548, "step": 11177 }, { "epoch": 0.8, "grad_norm": 4.90550804812969, "learning_rate": 9.975152452240378e-07, "loss": 0.1916, "step": 11178 }, { "epoch": 0.8, "grad_norm": 12.513039035870177, "learning_rate": 9.968193336085968e-07, "loss": 0.4127, "step": 11179 }, { "epoch": 0.8, "grad_norm": 3.906342115012532, "learning_rate": 9.961236379495997e-07, "loss": 0.3632, "step": 11180 }, { "epoch": 0.8, "grad_norm": 2.7443627628031693, "learning_rate": 9.954281582845765e-07, "loss": 0.1771, "step": 11181 }, { "epoch": 0.8, "grad_norm": 66.42802799136257, "learning_rate": 9.947328946510455e-07, "loss": 0.5337, "step": 11182 }, { "epoch": 0.8, "grad_norm": 4.975587075709441, "learning_rate": 9.940378470865148e-07, "loss": 0.89, "step": 11183 }, { "epoch": 0.8, "grad_norm": 3.568939782785417, "learning_rate": 9.93343015628479e-07, "loss": 0.1811, "step": 11184 }, { "epoch": 0.8, "grad_norm": 3.813882410993355, "learning_rate": 9.926484003144217e-07, "loss": 0.2265, "step": 11185 }, { "epoch": 0.8, "grad_norm": 16.27364060467717, "learning_rate": 9.91954001181813e-07, "loss": 0.8618, "step": 11186 }, { "epoch": 0.8, "grad_norm": 13.448875186389163, "learning_rate": 9.91259818268116e-07, "loss": 0.9735, "step": 11187 }, { "epoch": 0.8, "grad_norm": 3.468275416002643, "learning_rate": 9.90565851610778e-07, "loss": 0.1806, "step": 11188 }, { "epoch": 0.8, "grad_norm": 21.773304219196994, "learning_rate": 9.898721012472345e-07, "loss": 0.8113, "step": 11189 }, { "epoch": 0.8, "grad_norm": 4.145232530996839, "learning_rate": 9.891785672149129e-07, "loss": 0.7074, "step": 11190 }, { "epoch": 0.8, "grad_norm": 4.475625824205055, "learning_rate": 9.884852495512276e-07, "loss": 0.7915, "step": 11191 }, { "epoch": 0.8, "grad_norm": 10.761972484878248, "learning_rate": 9.877921482935759e-07, "loss": 0.4289, "step": 11192 }, { "epoch": 0.8, "grad_norm": 2.6759796802404763, "learning_rate": 9.870992634793518e-07, "loss": 0.0924, "step": 11193 }, { "epoch": 0.8, "grad_norm": 31.632522274604366, "learning_rate": 9.864065951459335e-07, "loss": 1.2025, "step": 11194 }, { "epoch": 0.8, "grad_norm": 3.33799525881131, "learning_rate": 9.85714143330685e-07, "loss": 0.104, "step": 11195 }, { "epoch": 0.8, "grad_norm": 10.955990063111875, "learning_rate": 9.85021908070965e-07, "loss": 0.4934, "step": 11196 }, { "epoch": 0.8, "grad_norm": 10.430960604248481, "learning_rate": 9.84329889404116e-07, "loss": 0.5251, "step": 11197 }, { "epoch": 0.8, "grad_norm": 4.2390709667637605, "learning_rate": 9.836380873674679e-07, "loss": 0.1153, "step": 11198 }, { "epoch": 0.8, "grad_norm": 4.236466927624268, "learning_rate": 9.82946501998343e-07, "loss": 0.198, "step": 11199 }, { "epoch": 0.8, "grad_norm": 3.1476670007845815, "learning_rate": 9.822551333340496e-07, "loss": 0.1762, "step": 11200 }, { "epoch": 0.8, "grad_norm": 4.059921761413497, "learning_rate": 9.815639814118822e-07, "loss": 0.2526, "step": 11201 }, { "epoch": 0.8, "grad_norm": 2.167441871778299, "learning_rate": 9.808730462691285e-07, "loss": 0.0763, "step": 11202 }, { "epoch": 0.8, "grad_norm": 3.0037190872072794, "learning_rate": 9.801823279430622e-07, "loss": 0.1108, "step": 11203 }, { "epoch": 0.8, "grad_norm": 3.1438456841099724, "learning_rate": 9.794918264709413e-07, "loss": 0.1921, "step": 11204 }, { "epoch": 0.8, "grad_norm": 3.555682958812967, "learning_rate": 9.788015418900188e-07, "loss": 0.24, "step": 11205 }, { "epoch": 0.8, "grad_norm": 16.664206938993722, "learning_rate": 9.781114742375319e-07, "loss": 0.9574, "step": 11206 }, { "epoch": 0.8, "grad_norm": 4.977242282342794, "learning_rate": 9.774216235507062e-07, "loss": 0.7443, "step": 11207 }, { "epoch": 0.8, "grad_norm": 14.061799604390206, "learning_rate": 9.767319898667589e-07, "loss": 0.6396, "step": 11208 }, { "epoch": 0.8, "grad_norm": 5.522087544240121, "learning_rate": 9.760425732228917e-07, "loss": 0.1648, "step": 11209 }, { "epoch": 0.8, "grad_norm": 7.691246879348281, "learning_rate": 9.753533736562953e-07, "loss": 0.4201, "step": 11210 }, { "epoch": 0.8, "grad_norm": 2.1685455948687857, "learning_rate": 9.746643912041515e-07, "loss": 0.0962, "step": 11211 }, { "epoch": 0.8, "grad_norm": 9.633574152289373, "learning_rate": 9.739756259036265e-07, "loss": 0.5476, "step": 11212 }, { "epoch": 0.8, "grad_norm": 4.439984782154816, "learning_rate": 9.732870777918779e-07, "loss": 0.8147, "step": 11213 }, { "epoch": 0.8, "grad_norm": 2.091923608423853, "learning_rate": 9.725987469060478e-07, "loss": 0.107, "step": 11214 }, { "epoch": 0.8, "grad_norm": 2.72456889868327, "learning_rate": 9.719106332832733e-07, "loss": 0.1531, "step": 11215 }, { "epoch": 0.8, "grad_norm": 9.400233504025405, "learning_rate": 9.712227369606703e-07, "loss": 0.5979, "step": 11216 }, { "epoch": 0.8, "grad_norm": 5.140857765397046, "learning_rate": 9.705350579753526e-07, "loss": 0.1673, "step": 11217 }, { "epoch": 0.8, "grad_norm": 9.271280688141688, "learning_rate": 9.698475963644161e-07, "loss": 0.3632, "step": 11218 }, { "epoch": 0.8, "grad_norm": 4.120186286011137, "learning_rate": 9.691603521649463e-07, "loss": 0.7847, "step": 11219 }, { "epoch": 0.8, "grad_norm": 2.8542311199613577, "learning_rate": 9.684733254140172e-07, "loss": 0.1252, "step": 11220 }, { "epoch": 0.8, "grad_norm": 13.585375984133492, "learning_rate": 9.677865161486933e-07, "loss": 0.4423, "step": 11221 }, { "epoch": 0.8, "grad_norm": 2.165305088389518, "learning_rate": 9.670999244060237e-07, "loss": 0.0873, "step": 11222 }, { "epoch": 0.8, "grad_norm": 3.6039256153908346, "learning_rate": 9.66413550223047e-07, "loss": 0.1211, "step": 11223 }, { "epoch": 0.8, "grad_norm": 2.5710291484039036, "learning_rate": 9.657273936367922e-07, "loss": 0.1286, "step": 11224 }, { "epoch": 0.8, "grad_norm": 2.9801905314065924, "learning_rate": 9.65041454684274e-07, "loss": 0.1319, "step": 11225 }, { "epoch": 0.8, "grad_norm": 10.48158352050901, "learning_rate": 9.64355733402495e-07, "loss": 0.6872, "step": 11226 }, { "epoch": 0.81, "grad_norm": 3.0034540926813076, "learning_rate": 9.636702298284513e-07, "loss": 0.1585, "step": 11227 }, { "epoch": 0.81, "grad_norm": 3.204978813386365, "learning_rate": 9.62984943999119e-07, "loss": 0.1342, "step": 11228 }, { "epoch": 0.81, "grad_norm": 4.288945248086135, "learning_rate": 9.622998759514661e-07, "loss": 0.2867, "step": 11229 }, { "epoch": 0.81, "grad_norm": 8.327408424310647, "learning_rate": 9.616150257224533e-07, "loss": 0.2767, "step": 11230 }, { "epoch": 0.81, "grad_norm": 4.243989622277644, "learning_rate": 9.609303933490237e-07, "loss": 1.0015, "step": 11231 }, { "epoch": 0.81, "grad_norm": 30.438689307397215, "learning_rate": 9.60245978868109e-07, "loss": 1.0712, "step": 11232 }, { "epoch": 0.81, "grad_norm": 3.0165026357681457, "learning_rate": 9.595617823166342e-07, "loss": 0.0997, "step": 11233 }, { "epoch": 0.81, "grad_norm": 20.18665755928999, "learning_rate": 9.588778037315073e-07, "loss": 0.6779, "step": 11234 }, { "epoch": 0.81, "grad_norm": 3.3423241336749188, "learning_rate": 9.58194043149625e-07, "loss": 0.1729, "step": 11235 }, { "epoch": 0.81, "grad_norm": 3.978894138810723, "learning_rate": 9.57510500607876e-07, "loss": 0.2434, "step": 11236 }, { "epoch": 0.81, "grad_norm": 23.99182417669606, "learning_rate": 9.568271761431342e-07, "loss": 0.7598, "step": 11237 }, { "epoch": 0.81, "grad_norm": 13.172106481954605, "learning_rate": 9.561440697922608e-07, "loss": 0.6335, "step": 11238 }, { "epoch": 0.81, "grad_norm": 2.896410984045872, "learning_rate": 9.5546118159211e-07, "loss": 0.1342, "step": 11239 }, { "epoch": 0.81, "grad_norm": 3.646035274014027, "learning_rate": 9.547785115795184e-07, "loss": 0.2058, "step": 11240 }, { "epoch": 0.81, "grad_norm": 3.3392439089393595, "learning_rate": 9.540960597913125e-07, "loss": 0.1296, "step": 11241 }, { "epoch": 0.81, "grad_norm": 4.966624817584027, "learning_rate": 9.534138262643111e-07, "loss": 0.65, "step": 11242 }, { "epoch": 0.81, "grad_norm": 3.931316261928019, "learning_rate": 9.527318110353162e-07, "loss": 0.2956, "step": 11243 }, { "epoch": 0.81, "grad_norm": 14.112468328280865, "learning_rate": 9.520500141411193e-07, "loss": 0.7094, "step": 11244 }, { "epoch": 0.81, "grad_norm": 3.30143790746558, "learning_rate": 9.513684356185032e-07, "loss": 0.1408, "step": 11245 }, { "epoch": 0.81, "grad_norm": 12.692715704194667, "learning_rate": 9.506870755042352e-07, "loss": 0.6853, "step": 11246 }, { "epoch": 0.81, "grad_norm": 2.9837233433633883, "learning_rate": 9.50005933835072e-07, "loss": 0.1674, "step": 11247 }, { "epoch": 0.81, "grad_norm": 4.96322603566102, "learning_rate": 9.49325010647757e-07, "loss": 0.3254, "step": 11248 }, { "epoch": 0.81, "grad_norm": 4.359179604493433, "learning_rate": 9.486443059790262e-07, "loss": 0.8889, "step": 11249 }, { "epoch": 0.81, "grad_norm": 4.493621192945069, "learning_rate": 9.479638198656004e-07, "loss": 0.7328, "step": 11250 }, { "epoch": 0.81, "grad_norm": 2.425340473543828, "learning_rate": 9.472835523441886e-07, "loss": 0.1367, "step": 11251 }, { "epoch": 0.81, "grad_norm": 2.1959261860306625, "learning_rate": 9.466035034514886e-07, "loss": 0.0887, "step": 11252 }, { "epoch": 0.81, "grad_norm": 25.333153193018607, "learning_rate": 9.459236732241855e-07, "loss": 0.9474, "step": 11253 }, { "epoch": 0.81, "grad_norm": 5.374250917228139, "learning_rate": 9.452440616989555e-07, "loss": 0.8883, "step": 11254 }, { "epoch": 0.81, "grad_norm": 15.202836809108526, "learning_rate": 9.44564668912461e-07, "loss": 0.6539, "step": 11255 }, { "epoch": 0.81, "grad_norm": 3.6656076641564046, "learning_rate": 9.438854949013516e-07, "loss": 0.1599, "step": 11256 }, { "epoch": 0.81, "grad_norm": 2.2023322569811246, "learning_rate": 9.432065397022655e-07, "loss": 0.1369, "step": 11257 }, { "epoch": 0.81, "grad_norm": 3.014149085973004, "learning_rate": 9.425278033518315e-07, "loss": 0.1862, "step": 11258 }, { "epoch": 0.81, "grad_norm": 2.0574257248912042, "learning_rate": 9.418492858866645e-07, "loss": 0.0837, "step": 11259 }, { "epoch": 0.81, "grad_norm": 16.706016580345068, "learning_rate": 9.411709873433666e-07, "loss": 0.3997, "step": 11260 }, { "epoch": 0.81, "grad_norm": 9.806780491780195, "learning_rate": 9.404929077585312e-07, "loss": 0.7261, "step": 11261 }, { "epoch": 0.81, "grad_norm": 4.037191112207576, "learning_rate": 9.398150471687378e-07, "loss": 0.2516, "step": 11262 }, { "epoch": 0.81, "grad_norm": 3.318194249655774, "learning_rate": 9.39137405610554e-07, "loss": 0.2083, "step": 11263 }, { "epoch": 0.81, "grad_norm": 5.007559815862972, "learning_rate": 9.384599831205354e-07, "loss": 0.2484, "step": 11264 }, { "epoch": 0.81, "grad_norm": 4.853617230186594, "learning_rate": 9.377827797352274e-07, "loss": 0.8293, "step": 11265 }, { "epoch": 0.81, "grad_norm": 14.099302292433087, "learning_rate": 9.371057954911606e-07, "loss": 0.7392, "step": 11266 }, { "epoch": 0.81, "grad_norm": 3.503038534135882, "learning_rate": 9.364290304248591e-07, "loss": 0.149, "step": 11267 }, { "epoch": 0.81, "grad_norm": 12.168983828075183, "learning_rate": 9.357524845728294e-07, "loss": 0.5358, "step": 11268 }, { "epoch": 0.81, "grad_norm": 15.773790477426283, "learning_rate": 9.350761579715678e-07, "loss": 0.8706, "step": 11269 }, { "epoch": 0.81, "grad_norm": 4.110550927338189, "learning_rate": 9.344000506575623e-07, "loss": 0.1948, "step": 11270 }, { "epoch": 0.81, "grad_norm": 11.152991468183895, "learning_rate": 9.33724162667285e-07, "loss": 0.4888, "step": 11271 }, { "epoch": 0.81, "grad_norm": 2.3539095379032817, "learning_rate": 9.330484940371964e-07, "loss": 0.1295, "step": 11272 }, { "epoch": 0.81, "grad_norm": 2.7217795559427636, "learning_rate": 9.323730448037482e-07, "loss": 0.1182, "step": 11273 }, { "epoch": 0.81, "grad_norm": 4.0375001587387125, "learning_rate": 9.316978150033773e-07, "loss": 0.2419, "step": 11274 }, { "epoch": 0.81, "grad_norm": 2.762019276362371, "learning_rate": 9.310228046725101e-07, "loss": 0.1015, "step": 11275 }, { "epoch": 0.81, "grad_norm": 28.82342644913045, "learning_rate": 9.303480138475612e-07, "loss": 1.1151, "step": 11276 }, { "epoch": 0.81, "grad_norm": 7.231768135352706, "learning_rate": 9.296734425649318e-07, "loss": 0.3368, "step": 11277 }, { "epoch": 0.81, "grad_norm": 10.21451262859199, "learning_rate": 9.28999090861012e-07, "loss": 0.7558, "step": 11278 }, { "epoch": 0.81, "grad_norm": 13.372677157688367, "learning_rate": 9.283249587721832e-07, "loss": 0.5801, "step": 11279 }, { "epoch": 0.81, "grad_norm": 3.172647303780197, "learning_rate": 9.276510463348104e-07, "loss": 0.1462, "step": 11280 }, { "epoch": 0.81, "grad_norm": 4.235633872121106, "learning_rate": 9.269773535852478e-07, "loss": 0.1803, "step": 11281 }, { "epoch": 0.81, "grad_norm": 2.764235602674284, "learning_rate": 9.263038805598413e-07, "loss": 0.1142, "step": 11282 }, { "epoch": 0.81, "grad_norm": 8.813623221676542, "learning_rate": 9.256306272949206e-07, "loss": 0.4738, "step": 11283 }, { "epoch": 0.81, "grad_norm": 2.1735983308862883, "learning_rate": 9.249575938268052e-07, "loss": 0.1135, "step": 11284 }, { "epoch": 0.81, "grad_norm": 8.962373634506864, "learning_rate": 9.242847801918014e-07, "loss": 0.5274, "step": 11285 }, { "epoch": 0.81, "grad_norm": 4.080346718264186, "learning_rate": 9.236121864262071e-07, "loss": 0.2055, "step": 11286 }, { "epoch": 0.81, "grad_norm": 4.288933801344575, "learning_rate": 9.229398125663058e-07, "loss": 0.7215, "step": 11287 }, { "epoch": 0.81, "grad_norm": 2.7942038112273546, "learning_rate": 9.222676586483692e-07, "loss": 0.1479, "step": 11288 }, { "epoch": 0.81, "grad_norm": 8.532979304576024, "learning_rate": 9.215957247086577e-07, "loss": 0.6425, "step": 11289 }, { "epoch": 0.81, "grad_norm": 10.047793835669392, "learning_rate": 9.209240107834172e-07, "loss": 0.4784, "step": 11290 }, { "epoch": 0.81, "grad_norm": 3.2297211552224665, "learning_rate": 9.202525169088878e-07, "loss": 0.2079, "step": 11291 }, { "epoch": 0.81, "grad_norm": 19.323656264376975, "learning_rate": 9.195812431212925e-07, "loss": 0.6493, "step": 11292 }, { "epoch": 0.81, "grad_norm": 4.756026778101369, "learning_rate": 9.189101894568442e-07, "loss": 0.1249, "step": 11293 }, { "epoch": 0.81, "grad_norm": 12.03707304411393, "learning_rate": 9.182393559517417e-07, "loss": 0.4427, "step": 11294 }, { "epoch": 0.81, "grad_norm": 2.409591415184252, "learning_rate": 9.17568742642177e-07, "loss": 0.1228, "step": 11295 }, { "epoch": 0.81, "grad_norm": 15.452580947180804, "learning_rate": 9.168983495643258e-07, "loss": 0.9399, "step": 11296 }, { "epoch": 0.81, "grad_norm": 4.757296267083468, "learning_rate": 9.162281767543518e-07, "loss": 0.9351, "step": 11297 }, { "epoch": 0.81, "grad_norm": 5.202560861216639, "learning_rate": 9.155582242484112e-07, "loss": 0.7339, "step": 11298 }, { "epoch": 0.81, "grad_norm": 6.035592443623348, "learning_rate": 9.148884920826451e-07, "loss": 0.3127, "step": 11299 }, { "epoch": 0.81, "grad_norm": 3.3418094180635594, "learning_rate": 9.142189802931795e-07, "loss": 0.1625, "step": 11300 }, { "epoch": 0.81, "grad_norm": 13.691070112210687, "learning_rate": 9.135496889161361e-07, "loss": 0.7214, "step": 11301 }, { "epoch": 0.81, "grad_norm": 10.087387391890557, "learning_rate": 9.128806179876181e-07, "loss": 0.2981, "step": 11302 }, { "epoch": 0.81, "grad_norm": 3.8521482951991852, "learning_rate": 9.122117675437192e-07, "loss": 0.2407, "step": 11303 }, { "epoch": 0.81, "grad_norm": 17.327633544576894, "learning_rate": 9.115431376205236e-07, "loss": 0.6666, "step": 11304 }, { "epoch": 0.81, "grad_norm": 11.429736602572087, "learning_rate": 9.108747282541003e-07, "loss": 0.6805, "step": 11305 }, { "epoch": 0.81, "grad_norm": 3.38244466295835, "learning_rate": 9.10206539480506e-07, "loss": 0.0694, "step": 11306 }, { "epoch": 0.81, "grad_norm": 3.6319933225083494, "learning_rate": 9.09538571335789e-07, "loss": 0.2516, "step": 11307 }, { "epoch": 0.81, "grad_norm": 31.62378271846809, "learning_rate": 9.088708238559829e-07, "loss": 1.0972, "step": 11308 }, { "epoch": 0.81, "grad_norm": 3.5734946796557714, "learning_rate": 9.082032970771093e-07, "loss": 0.1904, "step": 11309 }, { "epoch": 0.81, "grad_norm": 12.167124865694928, "learning_rate": 9.075359910351806e-07, "loss": 0.6324, "step": 11310 }, { "epoch": 0.81, "grad_norm": 14.963751955510077, "learning_rate": 9.068689057661955e-07, "loss": 0.7818, "step": 11311 }, { "epoch": 0.81, "grad_norm": 2.4315827018825793, "learning_rate": 9.062020413061368e-07, "loss": 0.0655, "step": 11312 }, { "epoch": 0.81, "grad_norm": 4.56730726074165, "learning_rate": 9.055353976909837e-07, "loss": 0.8922, "step": 11313 }, { "epoch": 0.81, "grad_norm": 8.37436770173705, "learning_rate": 9.048689749566975e-07, "loss": 0.3766, "step": 11314 }, { "epoch": 0.81, "grad_norm": 14.370624043841556, "learning_rate": 9.042027731392278e-07, "loss": 0.8235, "step": 11315 }, { "epoch": 0.81, "grad_norm": 3.4286707758540063, "learning_rate": 9.035367922745165e-07, "loss": 0.1891, "step": 11316 }, { "epoch": 0.81, "grad_norm": 3.0584292763050924, "learning_rate": 9.028710323984891e-07, "loss": 0.1269, "step": 11317 }, { "epoch": 0.81, "grad_norm": 2.9814922306564795, "learning_rate": 9.022054935470598e-07, "loss": 0.16, "step": 11318 }, { "epoch": 0.81, "grad_norm": 3.569675045171879, "learning_rate": 9.015401757561349e-07, "loss": 0.1751, "step": 11319 }, { "epoch": 0.81, "grad_norm": 3.6415670538844562, "learning_rate": 9.008750790616033e-07, "loss": 0.1816, "step": 11320 }, { "epoch": 0.81, "grad_norm": 3.422995161223985, "learning_rate": 9.002102034993454e-07, "loss": 0.1538, "step": 11321 }, { "epoch": 0.81, "grad_norm": 3.4517819759983026, "learning_rate": 8.995455491052274e-07, "loss": 0.1771, "step": 11322 }, { "epoch": 0.81, "grad_norm": 9.09490506431575, "learning_rate": 8.988811159151084e-07, "loss": 0.4002, "step": 11323 }, { "epoch": 0.81, "grad_norm": 2.9130581085310627, "learning_rate": 8.98216903964827e-07, "loss": 0.1426, "step": 11324 }, { "epoch": 0.81, "grad_norm": 9.235184280664564, "learning_rate": 8.975529132902189e-07, "loss": 0.4653, "step": 11325 }, { "epoch": 0.81, "grad_norm": 6.181814473593463, "learning_rate": 8.968891439271022e-07, "loss": 0.3425, "step": 11326 }, { "epoch": 0.81, "grad_norm": 2.7901066017407934, "learning_rate": 8.962255959112854e-07, "loss": 0.149, "step": 11327 }, { "epoch": 0.81, "grad_norm": 7.357990102945346, "learning_rate": 8.955622692785626e-07, "loss": 0.2774, "step": 11328 }, { "epoch": 0.81, "grad_norm": 13.459967598140478, "learning_rate": 8.948991640647209e-07, "loss": 0.8726, "step": 11329 }, { "epoch": 0.81, "grad_norm": 11.960470524807391, "learning_rate": 8.942362803055299e-07, "loss": 0.5491, "step": 11330 }, { "epoch": 0.81, "grad_norm": 2.7386652673917515, "learning_rate": 8.935736180367499e-07, "loss": 0.1175, "step": 11331 }, { "epoch": 0.81, "grad_norm": 3.7228117134580563, "learning_rate": 8.929111772941307e-07, "loss": 0.2127, "step": 11332 }, { "epoch": 0.81, "grad_norm": 3.200337948512874, "learning_rate": 8.92248958113407e-07, "loss": 0.1918, "step": 11333 }, { "epoch": 0.81, "grad_norm": 22.770294206282923, "learning_rate": 8.915869605303018e-07, "loss": 0.9193, "step": 11334 }, { "epoch": 0.81, "grad_norm": 23.62959591005656, "learning_rate": 8.909251845805305e-07, "loss": 0.6981, "step": 11335 }, { "epoch": 0.81, "grad_norm": 12.462572510273626, "learning_rate": 8.902636302997936e-07, "loss": 0.4369, "step": 11336 }, { "epoch": 0.81, "grad_norm": 3.9188517635884805, "learning_rate": 8.896022977237745e-07, "loss": 0.1651, "step": 11337 }, { "epoch": 0.81, "grad_norm": 10.974346678633523, "learning_rate": 8.889411868881543e-07, "loss": 0.6569, "step": 11338 }, { "epoch": 0.81, "grad_norm": 3.5124646239494774, "learning_rate": 8.882802978285954e-07, "loss": 0.1842, "step": 11339 }, { "epoch": 0.81, "grad_norm": 3.4430255809256263, "learning_rate": 8.8761963058075e-07, "loss": 0.2171, "step": 11340 }, { "epoch": 0.81, "grad_norm": 3.342173990700256, "learning_rate": 8.8695918518026e-07, "loss": 0.1863, "step": 11341 }, { "epoch": 0.81, "grad_norm": 6.160227677358714, "learning_rate": 8.862989616627538e-07, "loss": 0.2722, "step": 11342 }, { "epoch": 0.81, "grad_norm": 11.686562231619115, "learning_rate": 8.85638960063846e-07, "loss": 0.5838, "step": 11343 }, { "epoch": 0.81, "grad_norm": 2.3439175496400804, "learning_rate": 8.849791804191437e-07, "loss": 0.1005, "step": 11344 }, { "epoch": 0.81, "grad_norm": 134.98407848605393, "learning_rate": 8.843196227642386e-07, "loss": 0.6618, "step": 11345 }, { "epoch": 0.81, "grad_norm": 19.220324229996418, "learning_rate": 8.836602871347099e-07, "loss": 0.5623, "step": 11346 }, { "epoch": 0.81, "grad_norm": 25.69075593606668, "learning_rate": 8.830011735661287e-07, "loss": 0.6603, "step": 11347 }, { "epoch": 0.81, "grad_norm": 4.055822364175358, "learning_rate": 8.82342282094052e-07, "loss": 0.1852, "step": 11348 }, { "epoch": 0.81, "grad_norm": 4.137898667243659, "learning_rate": 8.816836127540202e-07, "loss": 0.2282, "step": 11349 }, { "epoch": 0.81, "grad_norm": 3.923852679565913, "learning_rate": 8.810251655815699e-07, "loss": 0.2094, "step": 11350 }, { "epoch": 0.81, "grad_norm": 4.340048968098989, "learning_rate": 8.803669406122211e-07, "loss": 0.5932, "step": 11351 }, { "epoch": 0.81, "grad_norm": 3.1858135277551365, "learning_rate": 8.797089378814804e-07, "loss": 0.0869, "step": 11352 }, { "epoch": 0.81, "grad_norm": 3.455567800512745, "learning_rate": 8.790511574248484e-07, "loss": 0.1785, "step": 11353 }, { "epoch": 0.81, "grad_norm": 3.8796388498668053, "learning_rate": 8.783935992778075e-07, "loss": 0.2395, "step": 11354 }, { "epoch": 0.81, "grad_norm": 4.328256055966652, "learning_rate": 8.77736263475829e-07, "loss": 0.1957, "step": 11355 }, { "epoch": 0.81, "grad_norm": 15.680583074211912, "learning_rate": 8.770791500543774e-07, "loss": 0.6989, "step": 11356 }, { "epoch": 0.81, "grad_norm": 3.014206602737682, "learning_rate": 8.764222590488997e-07, "loss": 0.1453, "step": 11357 }, { "epoch": 0.81, "grad_norm": 2.338332888422316, "learning_rate": 8.75765590494832e-07, "loss": 0.078, "step": 11358 }, { "epoch": 0.81, "grad_norm": 2.6548093300407287, "learning_rate": 8.751091444275983e-07, "loss": 0.1345, "step": 11359 }, { "epoch": 0.81, "grad_norm": 18.637345557927844, "learning_rate": 8.744529208826158e-07, "loss": 0.8195, "step": 11360 }, { "epoch": 0.81, "grad_norm": 7.019891553098966, "learning_rate": 8.737969198952795e-07, "loss": 0.8248, "step": 11361 }, { "epoch": 0.81, "grad_norm": 3.6637456100146646, "learning_rate": 8.731411415009817e-07, "loss": 0.1318, "step": 11362 }, { "epoch": 0.81, "grad_norm": 51.67322540181307, "learning_rate": 8.724855857350983e-07, "loss": 0.7276, "step": 11363 }, { "epoch": 0.81, "grad_norm": 8.952334806020787, "learning_rate": 8.718302526329947e-07, "loss": 0.2737, "step": 11364 }, { "epoch": 0.81, "grad_norm": 3.411773483563645, "learning_rate": 8.711751422300213e-07, "loss": 0.2385, "step": 11365 }, { "epoch": 0.82, "grad_norm": 14.670006246410352, "learning_rate": 8.705202545615216e-07, "loss": 0.6771, "step": 11366 }, { "epoch": 0.82, "grad_norm": 3.405281219725024, "learning_rate": 8.698655896628233e-07, "loss": 0.1519, "step": 11367 }, { "epoch": 0.82, "grad_norm": 15.890588934234819, "learning_rate": 8.692111475692422e-07, "loss": 0.8465, "step": 11368 }, { "epoch": 0.82, "grad_norm": 4.519402935549822, "learning_rate": 8.685569283160849e-07, "loss": 0.8254, "step": 11369 }, { "epoch": 0.82, "grad_norm": 3.261870499715079, "learning_rate": 8.679029319386429e-07, "loss": 0.1386, "step": 11370 }, { "epoch": 0.82, "grad_norm": 3.1343114601687825, "learning_rate": 8.672491584721954e-07, "loss": 0.178, "step": 11371 }, { "epoch": 0.82, "grad_norm": 6.466606797907774, "learning_rate": 8.665956079520149e-07, "loss": 0.1006, "step": 11372 }, { "epoch": 0.82, "grad_norm": 9.761878387268265, "learning_rate": 8.659422804133549e-07, "loss": 0.295, "step": 11373 }, { "epoch": 0.82, "grad_norm": 3.1151473759689754, "learning_rate": 8.652891758914589e-07, "loss": 0.1458, "step": 11374 }, { "epoch": 0.82, "grad_norm": 17.62159437377214, "learning_rate": 8.646362944215625e-07, "loss": 0.5476, "step": 11375 }, { "epoch": 0.82, "grad_norm": 26.956091705561157, "learning_rate": 8.639836360388848e-07, "loss": 1.3137, "step": 11376 }, { "epoch": 0.82, "grad_norm": 3.8740142253779686, "learning_rate": 8.633312007786332e-07, "loss": 0.212, "step": 11377 }, { "epoch": 0.82, "grad_norm": 3.770407011189253, "learning_rate": 8.626789886760062e-07, "loss": 0.1542, "step": 11378 }, { "epoch": 0.82, "grad_norm": 3.598852779424617, "learning_rate": 8.620269997661873e-07, "loss": 0.138, "step": 11379 }, { "epoch": 0.82, "grad_norm": 4.955201912501022, "learning_rate": 8.613752340843473e-07, "loss": 0.8545, "step": 11380 }, { "epoch": 0.82, "grad_norm": 9.591168981511204, "learning_rate": 8.607236916656491e-07, "loss": 0.2512, "step": 11381 }, { "epoch": 0.82, "grad_norm": 22.39565641347222, "learning_rate": 8.600723725452397e-07, "loss": 0.5726, "step": 11382 }, { "epoch": 0.82, "grad_norm": 3.4605886746133687, "learning_rate": 8.594212767582543e-07, "loss": 0.1191, "step": 11383 }, { "epoch": 0.82, "grad_norm": 3.578583920595402, "learning_rate": 8.58770404339821e-07, "loss": 0.157, "step": 11384 }, { "epoch": 0.82, "grad_norm": 3.393071741322175, "learning_rate": 8.58119755325047e-07, "loss": 0.1678, "step": 11385 }, { "epoch": 0.82, "grad_norm": 2.849531121926335, "learning_rate": 8.574693297490333e-07, "loss": 0.1491, "step": 11386 }, { "epoch": 0.82, "grad_norm": 13.615588452655807, "learning_rate": 8.568191276468701e-07, "loss": 0.4806, "step": 11387 }, { "epoch": 0.82, "grad_norm": 4.912843418949837, "learning_rate": 8.561691490536323e-07, "loss": 0.8863, "step": 11388 }, { "epoch": 0.82, "grad_norm": 27.51514947730469, "learning_rate": 8.555193940043821e-07, "loss": 0.7579, "step": 11389 }, { "epoch": 0.82, "grad_norm": 4.321253060869117, "learning_rate": 8.548698625341745e-07, "loss": 0.7803, "step": 11390 }, { "epoch": 0.82, "grad_norm": 2.7742908108404003, "learning_rate": 8.542205546780475e-07, "loss": 0.1284, "step": 11391 }, { "epoch": 0.82, "grad_norm": 10.056635818336877, "learning_rate": 8.53571470471029e-07, "loss": 0.4676, "step": 11392 }, { "epoch": 0.82, "grad_norm": 2.179745289973495, "learning_rate": 8.52922609948133e-07, "loss": 0.085, "step": 11393 }, { "epoch": 0.82, "grad_norm": 10.184557323451537, "learning_rate": 8.522739731443663e-07, "loss": 0.6106, "step": 11394 }, { "epoch": 0.82, "grad_norm": 3.251328090337382, "learning_rate": 8.516255600947187e-07, "loss": 0.1567, "step": 11395 }, { "epoch": 0.82, "grad_norm": 8.65233550002426, "learning_rate": 8.509773708341695e-07, "loss": 0.1593, "step": 11396 }, { "epoch": 0.82, "grad_norm": 4.1412231530848445, "learning_rate": 8.503294053976868e-07, "loss": 0.8915, "step": 11397 }, { "epoch": 0.82, "grad_norm": 1.8967022476008362, "learning_rate": 8.496816638202238e-07, "loss": 0.0733, "step": 11398 }, { "epoch": 0.82, "grad_norm": 4.6048755317161465, "learning_rate": 8.490341461367263e-07, "loss": 0.9156, "step": 11399 }, { "epoch": 0.82, "grad_norm": 2.120966111036423, "learning_rate": 8.483868523821243e-07, "loss": 0.0725, "step": 11400 }, { "epoch": 0.82, "grad_norm": 3.4269146396597, "learning_rate": 8.477397825913369e-07, "loss": 0.1739, "step": 11401 }, { "epoch": 0.82, "grad_norm": 4.333584538951088, "learning_rate": 8.470929367992697e-07, "loss": 0.7172, "step": 11402 }, { "epoch": 0.82, "grad_norm": 2.924483895199214, "learning_rate": 8.464463150408209e-07, "loss": 0.1653, "step": 11403 }, { "epoch": 0.82, "grad_norm": 15.468523848841022, "learning_rate": 8.457999173508707e-07, "loss": 0.6885, "step": 11404 }, { "epoch": 0.82, "grad_norm": 3.574407113835521, "learning_rate": 8.451537437642893e-07, "loss": 0.1252, "step": 11405 }, { "epoch": 0.82, "grad_norm": 25.619200311365383, "learning_rate": 8.44507794315938e-07, "loss": 0.7177, "step": 11406 }, { "epoch": 0.82, "grad_norm": 4.788257350380811, "learning_rate": 8.438620690406612e-07, "loss": 0.1644, "step": 11407 }, { "epoch": 0.82, "grad_norm": 3.8108136329498365, "learning_rate": 8.432165679732945e-07, "loss": 0.1193, "step": 11408 }, { "epoch": 0.82, "grad_norm": 3.8118242563255715, "learning_rate": 8.425712911486594e-07, "loss": 0.1766, "step": 11409 }, { "epoch": 0.82, "grad_norm": 3.8005070109849135, "learning_rate": 8.41926238601567e-07, "loss": 0.1706, "step": 11410 }, { "epoch": 0.82, "grad_norm": 2.988459470005291, "learning_rate": 8.412814103668127e-07, "loss": 0.1045, "step": 11411 }, { "epoch": 0.82, "grad_norm": 15.833630004337152, "learning_rate": 8.406368064791865e-07, "loss": 0.4819, "step": 11412 }, { "epoch": 0.82, "grad_norm": 5.238286566709133, "learning_rate": 8.399924269734605e-07, "loss": 0.9296, "step": 11413 }, { "epoch": 0.82, "grad_norm": 18.54653337472811, "learning_rate": 8.393482718843954e-07, "loss": 1.0127, "step": 11414 }, { "epoch": 0.82, "grad_norm": 3.602306766394731, "learning_rate": 8.387043412467432e-07, "loss": 0.1535, "step": 11415 }, { "epoch": 0.82, "grad_norm": 3.616407634482233, "learning_rate": 8.380606350952403e-07, "loss": 0.2029, "step": 11416 }, { "epoch": 0.82, "grad_norm": 15.471313515416083, "learning_rate": 8.374171534646114e-07, "loss": 0.7595, "step": 11417 }, { "epoch": 0.82, "grad_norm": 3.33141559416526, "learning_rate": 8.367738963895722e-07, "loss": 0.1836, "step": 11418 }, { "epoch": 0.82, "grad_norm": 4.401130618469618, "learning_rate": 8.361308639048222e-07, "loss": 0.5461, "step": 11419 }, { "epoch": 0.82, "grad_norm": 3.2997238492173757, "learning_rate": 8.354880560450512e-07, "loss": 0.1506, "step": 11420 }, { "epoch": 0.82, "grad_norm": 25.567550667824797, "learning_rate": 8.348454728449362e-07, "loss": 0.6612, "step": 11421 }, { "epoch": 0.82, "grad_norm": 25.36793236734448, "learning_rate": 8.342031143391416e-07, "loss": 1.2521, "step": 11422 }, { "epoch": 0.82, "grad_norm": 17.46429228956417, "learning_rate": 8.335609805623196e-07, "loss": 0.5587, "step": 11423 }, { "epoch": 0.82, "grad_norm": 3.451728079063249, "learning_rate": 8.329190715491131e-07, "loss": 0.1664, "step": 11424 }, { "epoch": 0.82, "grad_norm": 2.6770828089683256, "learning_rate": 8.322773873341494e-07, "loss": 0.0747, "step": 11425 }, { "epoch": 0.82, "grad_norm": 25.354170955068454, "learning_rate": 8.316359279520436e-07, "loss": 0.4422, "step": 11426 }, { "epoch": 0.82, "grad_norm": 13.275555878846868, "learning_rate": 8.30994693437403e-07, "loss": 0.398, "step": 11427 }, { "epoch": 0.82, "grad_norm": 10.759834728949796, "learning_rate": 8.303536838248177e-07, "loss": 0.5913, "step": 11428 }, { "epoch": 0.82, "grad_norm": 14.402182578791166, "learning_rate": 8.297128991488684e-07, "loss": 0.2595, "step": 11429 }, { "epoch": 0.82, "grad_norm": 2.8666062374087673, "learning_rate": 8.290723394441214e-07, "loss": 0.1807, "step": 11430 }, { "epoch": 0.82, "grad_norm": 3.7927599052063816, "learning_rate": 8.284320047451349e-07, "loss": 0.2055, "step": 11431 }, { "epoch": 0.82, "grad_norm": 4.80770733305481, "learning_rate": 8.277918950864522e-07, "loss": 0.1311, "step": 11432 }, { "epoch": 0.82, "grad_norm": 2.7134346717876676, "learning_rate": 8.271520105026037e-07, "loss": 0.1819, "step": 11433 }, { "epoch": 0.82, "grad_norm": 107.16368904709559, "learning_rate": 8.265123510281093e-07, "loss": 0.7044, "step": 11434 }, { "epoch": 0.82, "grad_norm": 2.978658456622087, "learning_rate": 8.258729166974744e-07, "loss": 0.1345, "step": 11435 }, { "epoch": 0.82, "grad_norm": 16.028880150218907, "learning_rate": 8.252337075451971e-07, "loss": 0.5917, "step": 11436 }, { "epoch": 0.82, "grad_norm": 9.331566489775122, "learning_rate": 8.245947236057595e-07, "loss": 0.2966, "step": 11437 }, { "epoch": 0.82, "grad_norm": 6.15718308966119, "learning_rate": 8.239559649136314e-07, "loss": 0.293, "step": 11438 }, { "epoch": 0.82, "grad_norm": 4.029704298694671, "learning_rate": 8.233174315032705e-07, "loss": 0.9169, "step": 11439 }, { "epoch": 0.82, "grad_norm": 22.165140973328743, "learning_rate": 8.226791234091259e-07, "loss": 0.8041, "step": 11440 }, { "epoch": 0.82, "grad_norm": 3.7689464113923803, "learning_rate": 8.220410406656304e-07, "loss": 0.0971, "step": 11441 }, { "epoch": 0.82, "grad_norm": 3.240974454548984, "learning_rate": 8.214031833072051e-07, "loss": 0.1321, "step": 11442 }, { "epoch": 0.82, "grad_norm": 4.070167344943989, "learning_rate": 8.207655513682627e-07, "loss": 0.2006, "step": 11443 }, { "epoch": 0.82, "grad_norm": 18.750052697719564, "learning_rate": 8.201281448832004e-07, "loss": 1.0829, "step": 11444 }, { "epoch": 0.82, "grad_norm": 2.4127431203291705, "learning_rate": 8.194909638864007e-07, "loss": 0.0631, "step": 11445 }, { "epoch": 0.82, "grad_norm": 25.654390786781388, "learning_rate": 8.188540084122409e-07, "loss": 0.7359, "step": 11446 }, { "epoch": 0.82, "grad_norm": 3.2519581696603757, "learning_rate": 8.182172784950803e-07, "loss": 0.1118, "step": 11447 }, { "epoch": 0.82, "grad_norm": 30.134183731700666, "learning_rate": 8.175807741692681e-07, "loss": 0.7401, "step": 11448 }, { "epoch": 0.82, "grad_norm": 4.075724647145497, "learning_rate": 8.169444954691424e-07, "loss": 0.7735, "step": 11449 }, { "epoch": 0.82, "grad_norm": 4.094048947394591, "learning_rate": 8.16308442429028e-07, "loss": 0.6443, "step": 11450 }, { "epoch": 0.82, "grad_norm": 4.809391167643615, "learning_rate": 8.156726150832356e-07, "loss": 0.8384, "step": 11451 }, { "epoch": 0.82, "grad_norm": 11.81728236974456, "learning_rate": 8.150370134660679e-07, "loss": 0.3405, "step": 11452 }, { "epoch": 0.82, "grad_norm": 9.58148611814071, "learning_rate": 8.144016376118131e-07, "loss": 0.4695, "step": 11453 }, { "epoch": 0.82, "grad_norm": 6.246422586363604, "learning_rate": 8.137664875547447e-07, "loss": 0.2084, "step": 11454 }, { "epoch": 0.82, "grad_norm": 12.455519768393195, "learning_rate": 8.131315633291298e-07, "loss": 0.7725, "step": 11455 }, { "epoch": 0.82, "grad_norm": 11.473769355762323, "learning_rate": 8.124968649692205e-07, "loss": 0.6269, "step": 11456 }, { "epoch": 0.82, "grad_norm": 2.7159160798191415, "learning_rate": 8.118623925092523e-07, "loss": 0.168, "step": 11457 }, { "epoch": 0.82, "grad_norm": 5.401138173879388, "learning_rate": 8.112281459834558e-07, "loss": 1.0535, "step": 11458 }, { "epoch": 0.82, "grad_norm": 2.514779074170204, "learning_rate": 8.105941254260463e-07, "loss": 0.0989, "step": 11459 }, { "epoch": 0.82, "grad_norm": 3.470581821445979, "learning_rate": 8.09960330871224e-07, "loss": 0.1414, "step": 11460 }, { "epoch": 0.82, "grad_norm": 3.478698234038923, "learning_rate": 8.093267623531831e-07, "loss": 0.1591, "step": 11461 }, { "epoch": 0.82, "grad_norm": 3.590588279795515, "learning_rate": 8.086934199061008e-07, "loss": 0.0896, "step": 11462 }, { "epoch": 0.82, "grad_norm": 4.03090722695491, "learning_rate": 8.080603035641426e-07, "loss": 0.8656, "step": 11463 }, { "epoch": 0.82, "grad_norm": 5.053825576686835, "learning_rate": 8.074274133614645e-07, "loss": 0.7915, "step": 11464 }, { "epoch": 0.82, "grad_norm": 4.263642327042843, "learning_rate": 8.067947493322082e-07, "loss": 0.6598, "step": 11465 }, { "epoch": 0.82, "grad_norm": 3.007372547746549, "learning_rate": 8.061623115105027e-07, "loss": 0.1551, "step": 11466 }, { "epoch": 0.82, "grad_norm": 4.559180202747493, "learning_rate": 8.055300999304649e-07, "loss": 0.9176, "step": 11467 }, { "epoch": 0.82, "grad_norm": 12.716786361309914, "learning_rate": 8.048981146262036e-07, "loss": 0.503, "step": 11468 }, { "epoch": 0.82, "grad_norm": 3.6820730140620226, "learning_rate": 8.04266355631807e-07, "loss": 0.1575, "step": 11469 }, { "epoch": 0.82, "grad_norm": 1.8232636180298123, "learning_rate": 8.036348229813606e-07, "loss": 0.0721, "step": 11470 }, { "epoch": 0.82, "grad_norm": 2.5111592154918587, "learning_rate": 8.030035167089311e-07, "loss": 0.1215, "step": 11471 }, { "epoch": 0.82, "grad_norm": 2.5980651552958784, "learning_rate": 8.023724368485753e-07, "loss": 0.1104, "step": 11472 }, { "epoch": 0.82, "grad_norm": 31.619117053939995, "learning_rate": 8.017415834343362e-07, "loss": 1.1299, "step": 11473 }, { "epoch": 0.82, "grad_norm": 4.17317813716771, "learning_rate": 8.01110956500249e-07, "loss": 0.1937, "step": 11474 }, { "epoch": 0.82, "grad_norm": 4.974538597228024, "learning_rate": 8.00480556080332e-07, "loss": 0.1745, "step": 11475 }, { "epoch": 0.82, "grad_norm": 18.318966626084542, "learning_rate": 7.998503822085918e-07, "loss": 0.5578, "step": 11476 }, { "epoch": 0.82, "grad_norm": 4.16501512988275, "learning_rate": 7.992204349190258e-07, "loss": 0.1661, "step": 11477 }, { "epoch": 0.82, "grad_norm": 3.941329794513546, "learning_rate": 7.985907142456173e-07, "loss": 0.2042, "step": 11478 }, { "epoch": 0.82, "grad_norm": 4.818559109875494, "learning_rate": 7.979612202223347e-07, "loss": 0.2024, "step": 11479 }, { "epoch": 0.82, "grad_norm": 23.31254630256185, "learning_rate": 7.973319528831414e-07, "loss": 0.7871, "step": 11480 }, { "epoch": 0.82, "grad_norm": 3.146098782673737, "learning_rate": 7.967029122619802e-07, "loss": 0.136, "step": 11481 }, { "epoch": 0.82, "grad_norm": 18.601188811150198, "learning_rate": 7.96074098392785e-07, "loss": 0.8104, "step": 11482 }, { "epoch": 0.82, "grad_norm": 3.1227508959366377, "learning_rate": 7.954455113094806e-07, "loss": 0.1169, "step": 11483 }, { "epoch": 0.82, "grad_norm": 3.5735734632787257, "learning_rate": 7.948171510459763e-07, "loss": 0.0901, "step": 11484 }, { "epoch": 0.82, "grad_norm": 7.1604125122740205, "learning_rate": 7.941890176361672e-07, "loss": 0.7505, "step": 11485 }, { "epoch": 0.82, "grad_norm": 2.0212358722294983, "learning_rate": 7.935611111139424e-07, "loss": 0.059, "step": 11486 }, { "epoch": 0.82, "grad_norm": 3.092321720415323, "learning_rate": 7.929334315131737e-07, "loss": 0.1554, "step": 11487 }, { "epoch": 0.82, "grad_norm": 3.941045185362747, "learning_rate": 7.923059788677201e-07, "loss": 0.2019, "step": 11488 }, { "epoch": 0.82, "grad_norm": 3.7477069012168323, "learning_rate": 7.916787532114334e-07, "loss": 0.1476, "step": 11489 }, { "epoch": 0.82, "grad_norm": 3.1300804337749075, "learning_rate": 7.910517545781487e-07, "loss": 0.1467, "step": 11490 }, { "epoch": 0.82, "grad_norm": 9.425217827422971, "learning_rate": 7.904249830016885e-07, "loss": 0.6285, "step": 11491 }, { "epoch": 0.82, "grad_norm": 3.395733766517612, "learning_rate": 7.897984385158691e-07, "loss": 0.1874, "step": 11492 }, { "epoch": 0.82, "grad_norm": 6.828974098342997, "learning_rate": 7.891721211544862e-07, "loss": 0.156, "step": 11493 }, { "epoch": 0.82, "grad_norm": 3.375749631433692, "learning_rate": 7.885460309513276e-07, "loss": 0.1502, "step": 11494 }, { "epoch": 0.82, "grad_norm": 2.715898458045288, "learning_rate": 7.879201679401699e-07, "loss": 0.1048, "step": 11495 }, { "epoch": 0.82, "grad_norm": 11.01800111547032, "learning_rate": 7.872945321547765e-07, "loss": 0.5007, "step": 11496 }, { "epoch": 0.82, "grad_norm": 12.884552495098209, "learning_rate": 7.866691236288953e-07, "loss": 0.8853, "step": 11497 }, { "epoch": 0.82, "grad_norm": 3.038170231004002, "learning_rate": 7.860439423962679e-07, "loss": 0.0847, "step": 11498 }, { "epoch": 0.82, "grad_norm": 4.446501100290045, "learning_rate": 7.854189884906188e-07, "loss": 0.6043, "step": 11499 }, { "epoch": 0.82, "grad_norm": 4.244235246178877, "learning_rate": 7.847942619456628e-07, "loss": 0.157, "step": 11500 }, { "epoch": 0.82, "grad_norm": 4.533501100416585, "learning_rate": 7.84169762795099e-07, "loss": 0.7047, "step": 11501 }, { "epoch": 0.82, "grad_norm": 16.191663982379534, "learning_rate": 7.835454910726203e-07, "loss": 0.6186, "step": 11502 }, { "epoch": 0.82, "grad_norm": 3.5748628608535338, "learning_rate": 7.829214468119023e-07, "loss": 0.2245, "step": 11503 }, { "epoch": 0.82, "grad_norm": 4.140779049838657, "learning_rate": 7.822976300466101e-07, "loss": 0.2198, "step": 11504 }, { "epoch": 0.82, "grad_norm": 5.289550754293374, "learning_rate": 7.816740408103951e-07, "loss": 0.7916, "step": 11505 }, { "epoch": 0.83, "grad_norm": 3.7680537859986036, "learning_rate": 7.810506791368977e-07, "loss": 0.2372, "step": 11506 }, { "epoch": 0.83, "grad_norm": 3.117132933651364, "learning_rate": 7.804275450597476e-07, "loss": 0.1727, "step": 11507 }, { "epoch": 0.83, "grad_norm": 3.6547967567180537, "learning_rate": 7.798046386125596e-07, "loss": 0.1377, "step": 11508 }, { "epoch": 0.83, "grad_norm": 22.20727652508343, "learning_rate": 7.791819598289368e-07, "loss": 0.7598, "step": 11509 }, { "epoch": 0.83, "grad_norm": 4.779001335683365, "learning_rate": 7.785595087424696e-07, "loss": 0.7832, "step": 11510 }, { "epoch": 0.83, "grad_norm": 11.673636365800302, "learning_rate": 7.77937285386739e-07, "loss": 0.7329, "step": 11511 }, { "epoch": 0.83, "grad_norm": 18.497931549571224, "learning_rate": 7.773152897953107e-07, "loss": 0.6275, "step": 11512 }, { "epoch": 0.83, "grad_norm": 2.7430494397176477, "learning_rate": 7.766935220017374e-07, "loss": 0.1291, "step": 11513 }, { "epoch": 0.83, "grad_norm": 5.310001568364786, "learning_rate": 7.760719820395635e-07, "loss": 0.905, "step": 11514 }, { "epoch": 0.83, "grad_norm": 3.7219368001325175, "learning_rate": 7.754506699423181e-07, "loss": 0.2193, "step": 11515 }, { "epoch": 0.83, "grad_norm": 3.7013779237832343, "learning_rate": 7.748295857435184e-07, "loss": 0.1688, "step": 11516 }, { "epoch": 0.83, "grad_norm": 4.603244228965546, "learning_rate": 7.742087294766687e-07, "loss": 0.1892, "step": 11517 }, { "epoch": 0.83, "grad_norm": 21.10009675730082, "learning_rate": 7.735881011752633e-07, "loss": 0.7711, "step": 11518 }, { "epoch": 0.83, "grad_norm": 18.472850581569823, "learning_rate": 7.729677008727804e-07, "loss": 0.7818, "step": 11519 }, { "epoch": 0.83, "grad_norm": 20.799294402404563, "learning_rate": 7.72347528602691e-07, "loss": 0.7177, "step": 11520 }, { "epoch": 0.83, "grad_norm": 3.0784472985658247, "learning_rate": 7.717275843984501e-07, "loss": 0.1278, "step": 11521 }, { "epoch": 0.83, "grad_norm": 3.525682361473285, "learning_rate": 7.711078682934997e-07, "loss": 0.1862, "step": 11522 }, { "epoch": 0.83, "grad_norm": 18.21205712009639, "learning_rate": 7.704883803212743e-07, "loss": 0.512, "step": 11523 }, { "epoch": 0.83, "grad_norm": 2.8281212676427696, "learning_rate": 7.698691205151909e-07, "loss": 0.0965, "step": 11524 }, { "epoch": 0.83, "grad_norm": 14.213854372740416, "learning_rate": 7.692500889086551e-07, "loss": 0.5259, "step": 11525 }, { "epoch": 0.83, "grad_norm": 12.058358552532516, "learning_rate": 7.686312855350642e-07, "loss": 0.8931, "step": 11526 }, { "epoch": 0.83, "grad_norm": 20.370730251126044, "learning_rate": 7.68012710427799e-07, "loss": 0.8033, "step": 11527 }, { "epoch": 0.83, "grad_norm": 4.929284053212142, "learning_rate": 7.673943636202286e-07, "loss": 0.8599, "step": 11528 }, { "epoch": 0.83, "grad_norm": 2.989326286064737, "learning_rate": 7.667762451457117e-07, "loss": 0.0651, "step": 11529 }, { "epoch": 0.83, "grad_norm": 4.511828521765688, "learning_rate": 7.661583550375923e-07, "loss": 0.7276, "step": 11530 }, { "epoch": 0.83, "grad_norm": 6.0723514627547015, "learning_rate": 7.655406933292026e-07, "loss": 0.7228, "step": 11531 }, { "epoch": 0.83, "grad_norm": 3.1959698430140566, "learning_rate": 7.649232600538653e-07, "loss": 0.1046, "step": 11532 }, { "epoch": 0.83, "grad_norm": 3.989789533253229, "learning_rate": 7.643060552448873e-07, "loss": 0.1521, "step": 11533 }, { "epoch": 0.83, "grad_norm": 30.117726890240558, "learning_rate": 7.636890789355638e-07, "loss": 0.7876, "step": 11534 }, { "epoch": 0.83, "grad_norm": 3.8976127283816786, "learning_rate": 7.630723311591803e-07, "loss": 0.1835, "step": 11535 }, { "epoch": 0.83, "grad_norm": 4.967346888893819, "learning_rate": 7.624558119490066e-07, "loss": 0.9116, "step": 11536 }, { "epoch": 0.83, "grad_norm": 3.415740468466332, "learning_rate": 7.618395213383023e-07, "loss": 0.1349, "step": 11537 }, { "epoch": 0.83, "grad_norm": 5.375796889627593, "learning_rate": 7.612234593603123e-07, "loss": 0.6911, "step": 11538 }, { "epoch": 0.83, "grad_norm": 13.706965052366563, "learning_rate": 7.606076260482731e-07, "loss": 0.4621, "step": 11539 }, { "epoch": 0.83, "grad_norm": 2.6203306121585523, "learning_rate": 7.599920214354056e-07, "loss": 0.0935, "step": 11540 }, { "epoch": 0.83, "grad_norm": 3.3030003594883794, "learning_rate": 7.59376645554919e-07, "loss": 0.2075, "step": 11541 }, { "epoch": 0.83, "grad_norm": 2.7037818578354784, "learning_rate": 7.58761498440011e-07, "loss": 0.1229, "step": 11542 }, { "epoch": 0.83, "grad_norm": 5.24376449614833, "learning_rate": 7.581465801238647e-07, "loss": 0.2552, "step": 11543 }, { "epoch": 0.83, "grad_norm": 20.420981048059296, "learning_rate": 7.575318906396556e-07, "loss": 0.9209, "step": 11544 }, { "epoch": 0.83, "grad_norm": 3.1277037527685705, "learning_rate": 7.569174300205423e-07, "loss": 0.1147, "step": 11545 }, { "epoch": 0.83, "grad_norm": 4.948388352564651, "learning_rate": 7.563031982996727e-07, "loss": 0.2622, "step": 11546 }, { "epoch": 0.83, "grad_norm": 8.072370504633165, "learning_rate": 7.556891955101809e-07, "loss": 0.3289, "step": 11547 }, { "epoch": 0.83, "grad_norm": 4.3417239186888175, "learning_rate": 7.55075421685193e-07, "loss": 0.8062, "step": 11548 }, { "epoch": 0.83, "grad_norm": 2.64919072584529, "learning_rate": 7.544618768578177e-07, "loss": 0.1186, "step": 11549 }, { "epoch": 0.83, "grad_norm": 6.441130644872199, "learning_rate": 7.538485610611529e-07, "loss": 0.9905, "step": 11550 }, { "epoch": 0.83, "grad_norm": 6.106102180146796, "learning_rate": 7.532354743282872e-07, "loss": 0.2414, "step": 11551 }, { "epoch": 0.83, "grad_norm": 12.195540964548412, "learning_rate": 7.526226166922923e-07, "loss": 0.4436, "step": 11552 }, { "epoch": 0.83, "grad_norm": 3.669704695642848, "learning_rate": 7.520099881862302e-07, "loss": 0.1967, "step": 11553 }, { "epoch": 0.83, "grad_norm": 33.10645810675977, "learning_rate": 7.513975888431502e-07, "loss": 1.2845, "step": 11554 }, { "epoch": 0.83, "grad_norm": 5.63069622177669, "learning_rate": 7.507854186960878e-07, "loss": 0.8734, "step": 11555 }, { "epoch": 0.83, "grad_norm": 13.928718832896925, "learning_rate": 7.501734777780667e-07, "loss": 0.6061, "step": 11556 }, { "epoch": 0.83, "grad_norm": 5.030990462915481, "learning_rate": 7.495617661221016e-07, "loss": 0.7946, "step": 11557 }, { "epoch": 0.83, "grad_norm": 8.365709383538652, "learning_rate": 7.489502837611906e-07, "loss": 0.4607, "step": 11558 }, { "epoch": 0.83, "grad_norm": 30.427249835694926, "learning_rate": 7.48339030728319e-07, "loss": 0.6911, "step": 11559 }, { "epoch": 0.83, "grad_norm": 3.77553184284032, "learning_rate": 7.477280070564646e-07, "loss": 0.1311, "step": 11560 }, { "epoch": 0.83, "grad_norm": 8.584199775275398, "learning_rate": 7.471172127785886e-07, "loss": 0.3497, "step": 11561 }, { "epoch": 0.83, "grad_norm": 2.7510469462042795, "learning_rate": 7.465066479276395e-07, "loss": 0.1714, "step": 11562 }, { "epoch": 0.83, "grad_norm": 2.251992814873611, "learning_rate": 7.458963125365581e-07, "loss": 0.1115, "step": 11563 }, { "epoch": 0.83, "grad_norm": 2.8163338784058602, "learning_rate": 7.452862066382676e-07, "loss": 0.1258, "step": 11564 }, { "epoch": 0.83, "grad_norm": 19.94130741705083, "learning_rate": 7.446763302656813e-07, "loss": 1.1451, "step": 11565 }, { "epoch": 0.83, "grad_norm": 4.741834101071004, "learning_rate": 7.440666834516996e-07, "loss": 0.8611, "step": 11566 }, { "epoch": 0.83, "grad_norm": 5.208556178336088, "learning_rate": 7.434572662292111e-07, "loss": 0.8573, "step": 11567 }, { "epoch": 0.83, "grad_norm": 3.886440640262248, "learning_rate": 7.428480786310899e-07, "loss": 0.1362, "step": 11568 }, { "epoch": 0.83, "grad_norm": 15.912671102339065, "learning_rate": 7.422391206902018e-07, "loss": 0.6395, "step": 11569 }, { "epoch": 0.83, "grad_norm": 3.0703897556525033, "learning_rate": 7.416303924393964e-07, "loss": 0.1387, "step": 11570 }, { "epoch": 0.83, "grad_norm": 4.307265371797523, "learning_rate": 7.410218939115116e-07, "loss": 0.8362, "step": 11571 }, { "epoch": 0.83, "grad_norm": 2.753376253874173, "learning_rate": 7.404136251393757e-07, "loss": 0.1205, "step": 11572 }, { "epoch": 0.83, "grad_norm": 5.19801368323218, "learning_rate": 7.398055861558013e-07, "loss": 0.9636, "step": 11573 }, { "epoch": 0.83, "grad_norm": 2.5883448289704565, "learning_rate": 7.391977769935899e-07, "loss": 0.1564, "step": 11574 }, { "epoch": 0.83, "grad_norm": 4.115501200315276, "learning_rate": 7.385901976855292e-07, "loss": 0.6975, "step": 11575 }, { "epoch": 0.83, "grad_norm": 12.181250902675243, "learning_rate": 7.379828482643981e-07, "loss": 0.7825, "step": 11576 }, { "epoch": 0.83, "grad_norm": 3.724706461692632, "learning_rate": 7.373757287629596e-07, "loss": 0.1888, "step": 11577 }, { "epoch": 0.83, "grad_norm": 3.9750416710189156, "learning_rate": 7.367688392139655e-07, "loss": 0.1928, "step": 11578 }, { "epoch": 0.83, "grad_norm": 2.872762399301401, "learning_rate": 7.361621796501556e-07, "loss": 0.0953, "step": 11579 }, { "epoch": 0.83, "grad_norm": 1.9814943658365927, "learning_rate": 7.355557501042553e-07, "loss": 0.0535, "step": 11580 }, { "epoch": 0.83, "grad_norm": 5.782462743590023, "learning_rate": 7.349495506089821e-07, "loss": 0.8898, "step": 11581 }, { "epoch": 0.83, "grad_norm": 3.132273313889291, "learning_rate": 7.343435811970362e-07, "loss": 0.0918, "step": 11582 }, { "epoch": 0.83, "grad_norm": 17.830374569415635, "learning_rate": 7.337378419011076e-07, "loss": 0.4554, "step": 11583 }, { "epoch": 0.83, "grad_norm": 3.9482727023143185, "learning_rate": 7.331323327538731e-07, "loss": 0.1623, "step": 11584 }, { "epoch": 0.83, "grad_norm": 9.85056953596787, "learning_rate": 7.325270537879986e-07, "loss": 0.2995, "step": 11585 }, { "epoch": 0.83, "grad_norm": 15.886697359587416, "learning_rate": 7.31922005036137e-07, "loss": 0.8162, "step": 11586 }, { "epoch": 0.83, "grad_norm": 4.862805551357267, "learning_rate": 7.313171865309266e-07, "loss": 0.8504, "step": 11587 }, { "epoch": 0.83, "grad_norm": 5.003414335046643, "learning_rate": 7.307125983049968e-07, "loss": 0.7307, "step": 11588 }, { "epoch": 0.83, "grad_norm": 5.101676772732031, "learning_rate": 7.301082403909631e-07, "loss": 0.8515, "step": 11589 }, { "epoch": 0.83, "grad_norm": 5.1231902208199385, "learning_rate": 7.295041128214258e-07, "loss": 0.7557, "step": 11590 }, { "epoch": 0.83, "grad_norm": 11.945340907880466, "learning_rate": 7.289002156289776e-07, "loss": 0.6469, "step": 11591 }, { "epoch": 0.83, "grad_norm": 22.354307229933077, "learning_rate": 7.282965488461962e-07, "loss": 0.6865, "step": 11592 }, { "epoch": 0.83, "grad_norm": 5.275507917189025, "learning_rate": 7.276931125056453e-07, "loss": 0.7292, "step": 11593 }, { "epoch": 0.83, "grad_norm": 10.056245148955677, "learning_rate": 7.270899066398807e-07, "loss": 0.3811, "step": 11594 }, { "epoch": 0.83, "grad_norm": 4.669444760334393, "learning_rate": 7.264869312814415e-07, "loss": 0.7647, "step": 11595 }, { "epoch": 0.83, "grad_norm": 4.462533073610011, "learning_rate": 7.258841864628551e-07, "loss": 0.9518, "step": 11596 }, { "epoch": 0.83, "grad_norm": 2.7040340251435486, "learning_rate": 7.252816722166401e-07, "loss": 0.1569, "step": 11597 }, { "epoch": 0.83, "grad_norm": 13.944422056033353, "learning_rate": 7.246793885752979e-07, "loss": 0.6821, "step": 11598 }, { "epoch": 0.83, "grad_norm": 3.290516282285179, "learning_rate": 7.240773355713182e-07, "loss": 0.1796, "step": 11599 }, { "epoch": 0.83, "grad_norm": 10.17493786612402, "learning_rate": 7.234755132371824e-07, "loss": 0.6769, "step": 11600 }, { "epoch": 0.83, "grad_norm": 3.119987849845651, "learning_rate": 7.228739216053565e-07, "loss": 0.11, "step": 11601 }, { "epoch": 0.83, "grad_norm": 19.644099425279823, "learning_rate": 7.222725607082904e-07, "loss": 0.6744, "step": 11602 }, { "epoch": 0.83, "grad_norm": 8.391983379628092, "learning_rate": 7.216714305784283e-07, "loss": 0.4315, "step": 11603 }, { "epoch": 0.83, "grad_norm": 1.500889504079284, "learning_rate": 7.210705312481986e-07, "loss": 0.0542, "step": 11604 }, { "epoch": 0.83, "grad_norm": 3.2006652498541324, "learning_rate": 7.204698627500162e-07, "loss": 0.1653, "step": 11605 }, { "epoch": 0.83, "grad_norm": 13.74869562430814, "learning_rate": 7.198694251162863e-07, "loss": 0.761, "step": 11606 }, { "epoch": 0.83, "grad_norm": 3.0648025027968444, "learning_rate": 7.192692183793998e-07, "loss": 0.1308, "step": 11607 }, { "epoch": 0.83, "grad_norm": 3.258707739507673, "learning_rate": 7.186692425717351e-07, "loss": 0.1139, "step": 11608 }, { "epoch": 0.83, "grad_norm": 4.2162977804741075, "learning_rate": 7.180694977256597e-07, "loss": 0.2722, "step": 11609 }, { "epoch": 0.83, "grad_norm": 3.028995046060997, "learning_rate": 7.174699838735266e-07, "loss": 0.1342, "step": 11610 }, { "epoch": 0.83, "grad_norm": 11.431280436555369, "learning_rate": 7.168707010476778e-07, "loss": 0.5402, "step": 11611 }, { "epoch": 0.83, "grad_norm": 4.628335546643255, "learning_rate": 7.162716492804412e-07, "loss": 0.7863, "step": 11612 }, { "epoch": 0.83, "grad_norm": 4.305198185689251, "learning_rate": 7.156728286041365e-07, "loss": 0.7845, "step": 11613 }, { "epoch": 0.83, "grad_norm": 5.3044963636785845, "learning_rate": 7.150742390510635e-07, "loss": 0.2445, "step": 11614 }, { "epoch": 0.83, "grad_norm": 22.730431526463512, "learning_rate": 7.144758806535168e-07, "loss": 0.4969, "step": 11615 }, { "epoch": 0.83, "grad_norm": 21.956040697245967, "learning_rate": 7.138777534437752e-07, "loss": 0.6817, "step": 11616 }, { "epoch": 0.83, "grad_norm": 4.84016461331668, "learning_rate": 7.13279857454105e-07, "loss": 0.733, "step": 11617 }, { "epoch": 0.83, "grad_norm": 2.7719447507297437, "learning_rate": 7.126821927167587e-07, "loss": 0.1113, "step": 11618 }, { "epoch": 0.83, "grad_norm": 8.404445698374285, "learning_rate": 7.120847592639812e-07, "loss": 0.4248, "step": 11619 }, { "epoch": 0.83, "grad_norm": 11.452726709852652, "learning_rate": 7.11487557128e-07, "loss": 0.5533, "step": 11620 }, { "epoch": 0.83, "grad_norm": 4.34958655138867, "learning_rate": 7.108905863410314e-07, "loss": 0.903, "step": 11621 }, { "epoch": 0.83, "grad_norm": 17.77096875430123, "learning_rate": 7.102938469352816e-07, "loss": 0.688, "step": 11622 }, { "epoch": 0.83, "grad_norm": 4.502895731324728, "learning_rate": 7.096973389429413e-07, "loss": 0.1901, "step": 11623 }, { "epoch": 0.83, "grad_norm": 4.371867372650565, "learning_rate": 7.091010623961886e-07, "loss": 0.917, "step": 11624 }, { "epoch": 0.83, "grad_norm": 22.18073281678125, "learning_rate": 7.08505017327194e-07, "loss": 0.9875, "step": 11625 }, { "epoch": 0.83, "grad_norm": 10.311168792770168, "learning_rate": 7.079092037681085e-07, "loss": 0.3163, "step": 11626 }, { "epoch": 0.83, "grad_norm": 3.8814955922817385, "learning_rate": 7.073136217510734e-07, "loss": 0.1994, "step": 11627 }, { "epoch": 0.83, "grad_norm": 3.226299798066124, "learning_rate": 7.067182713082216e-07, "loss": 0.1331, "step": 11628 }, { "epoch": 0.83, "grad_norm": 19.84037389591835, "learning_rate": 7.061231524716678e-07, "loss": 1.1237, "step": 11629 }, { "epoch": 0.83, "grad_norm": 14.047225126885726, "learning_rate": 7.055282652735152e-07, "loss": 0.6624, "step": 11630 }, { "epoch": 0.83, "grad_norm": 3.2326636598017675, "learning_rate": 7.049336097458586e-07, "loss": 0.2053, "step": 11631 }, { "epoch": 0.83, "grad_norm": 3.24772820298811, "learning_rate": 7.043391859207765e-07, "loss": 0.1676, "step": 11632 }, { "epoch": 0.83, "grad_norm": 8.971708880912415, "learning_rate": 7.037449938303337e-07, "loss": 0.7696, "step": 11633 }, { "epoch": 0.83, "grad_norm": 2.923376591712542, "learning_rate": 7.031510335065878e-07, "loss": 0.1267, "step": 11634 }, { "epoch": 0.83, "grad_norm": 5.131759099956439, "learning_rate": 7.025573049815787e-07, "loss": 0.1859, "step": 11635 }, { "epoch": 0.83, "grad_norm": 2.7080989689578936, "learning_rate": 7.019638082873353e-07, "loss": 0.1372, "step": 11636 }, { "epoch": 0.83, "grad_norm": 4.335054750058308, "learning_rate": 7.013705434558782e-07, "loss": 0.1609, "step": 11637 }, { "epoch": 0.83, "grad_norm": 4.27208237100881, "learning_rate": 7.007775105192071e-07, "loss": 0.7766, "step": 11638 }, { "epoch": 0.83, "grad_norm": 3.450523837054277, "learning_rate": 7.001847095093156e-07, "loss": 0.1398, "step": 11639 }, { "epoch": 0.83, "grad_norm": 17.775548823711596, "learning_rate": 6.995921404581841e-07, "loss": 0.6253, "step": 11640 }, { "epoch": 0.83, "grad_norm": 2.629104803635288, "learning_rate": 6.989998033977786e-07, "loss": 0.1144, "step": 11641 }, { "epoch": 0.83, "grad_norm": 5.072776611048794, "learning_rate": 6.984076983600524e-07, "loss": 0.901, "step": 11642 }, { "epoch": 0.83, "grad_norm": 11.992342062243319, "learning_rate": 6.978158253769501e-07, "loss": 0.3081, "step": 11643 }, { "epoch": 0.83, "grad_norm": 4.400148928025634, "learning_rate": 6.97224184480399e-07, "loss": 0.8324, "step": 11644 }, { "epoch": 0.84, "grad_norm": 15.323902640720025, "learning_rate": 6.966327757023161e-07, "loss": 0.5965, "step": 11645 }, { "epoch": 0.84, "grad_norm": 3.443807028336484, "learning_rate": 6.960415990746044e-07, "loss": 0.1995, "step": 11646 }, { "epoch": 0.84, "grad_norm": 10.764149050526017, "learning_rate": 6.954506546291589e-07, "loss": 0.4478, "step": 11647 }, { "epoch": 0.84, "grad_norm": 10.533428159092548, "learning_rate": 6.94859942397857e-07, "loss": 0.503, "step": 11648 }, { "epoch": 0.84, "grad_norm": 19.887688422956668, "learning_rate": 6.942694624125651e-07, "loss": 0.9992, "step": 11649 }, { "epoch": 0.84, "grad_norm": 42.084094590485655, "learning_rate": 6.936792147051375e-07, "loss": 0.7993, "step": 11650 }, { "epoch": 0.84, "grad_norm": 24.05125541317249, "learning_rate": 6.930891993074146e-07, "loss": 0.7801, "step": 11651 }, { "epoch": 0.84, "grad_norm": 4.506983720909783, "learning_rate": 6.924994162512283e-07, "loss": 0.1322, "step": 11652 }, { "epoch": 0.84, "grad_norm": 3.067143572294217, "learning_rate": 6.919098655683937e-07, "loss": 0.1063, "step": 11653 }, { "epoch": 0.84, "grad_norm": 2.6270441781337186, "learning_rate": 6.913205472907153e-07, "loss": 0.0647, "step": 11654 }, { "epoch": 0.84, "grad_norm": 3.4434530659809064, "learning_rate": 6.907314614499821e-07, "loss": 0.1056, "step": 11655 }, { "epoch": 0.84, "grad_norm": 13.30621486202283, "learning_rate": 6.901426080779772e-07, "loss": 0.2928, "step": 11656 }, { "epoch": 0.84, "grad_norm": 24.412576455490314, "learning_rate": 6.895539872064644e-07, "loss": 0.4447, "step": 11657 }, { "epoch": 0.84, "grad_norm": 3.845414180137233, "learning_rate": 6.889655988671967e-07, "loss": 0.2186, "step": 11658 }, { "epoch": 0.84, "grad_norm": 5.042345375312243, "learning_rate": 6.883774430919182e-07, "loss": 0.8218, "step": 11659 }, { "epoch": 0.84, "grad_norm": 3.7353411731293287, "learning_rate": 6.877895199123563e-07, "loss": 0.2244, "step": 11660 }, { "epoch": 0.84, "grad_norm": 18.544652043085474, "learning_rate": 6.872018293602267e-07, "loss": 0.634, "step": 11661 }, { "epoch": 0.84, "grad_norm": 3.430443085844685, "learning_rate": 6.866143714672341e-07, "loss": 0.1412, "step": 11662 }, { "epoch": 0.84, "grad_norm": 19.111310651442057, "learning_rate": 6.860271462650692e-07, "loss": 0.6627, "step": 11663 }, { "epoch": 0.84, "grad_norm": 4.7481726178489385, "learning_rate": 6.854401537854094e-07, "loss": 1.1279, "step": 11664 }, { "epoch": 0.84, "grad_norm": 22.06692563049285, "learning_rate": 6.848533940599222e-07, "loss": 0.5965, "step": 11665 }, { "epoch": 0.84, "grad_norm": 4.384065431952569, "learning_rate": 6.842668671202613e-07, "loss": 0.8025, "step": 11666 }, { "epoch": 0.84, "grad_norm": 3.252742770258685, "learning_rate": 6.836805729980655e-07, "loss": 0.1028, "step": 11667 }, { "epoch": 0.84, "grad_norm": 21.203768186803224, "learning_rate": 6.830945117249666e-07, "loss": 0.7209, "step": 11668 }, { "epoch": 0.84, "grad_norm": 4.0273156977867615, "learning_rate": 6.825086833325773e-07, "loss": 0.2521, "step": 11669 }, { "epoch": 0.84, "grad_norm": 17.976116265908747, "learning_rate": 6.819230878525013e-07, "loss": 0.8304, "step": 11670 }, { "epoch": 0.84, "grad_norm": 7.210164721148102, "learning_rate": 6.813377253163311e-07, "loss": 0.3052, "step": 11671 }, { "epoch": 0.84, "grad_norm": 17.12779324751947, "learning_rate": 6.807525957556438e-07, "loss": 0.6525, "step": 11672 }, { "epoch": 0.84, "grad_norm": 2.926864041101215, "learning_rate": 6.801676992020045e-07, "loss": 0.0798, "step": 11673 }, { "epoch": 0.84, "grad_norm": 4.255439482877788, "learning_rate": 6.795830356869665e-07, "loss": 0.2773, "step": 11674 }, { "epoch": 0.84, "grad_norm": 33.42847047706558, "learning_rate": 6.789986052420694e-07, "loss": 0.859, "step": 11675 }, { "epoch": 0.84, "grad_norm": 3.1104830894309905, "learning_rate": 6.784144078988414e-07, "loss": 0.1207, "step": 11676 }, { "epoch": 0.84, "grad_norm": 20.37233504042895, "learning_rate": 6.778304436887984e-07, "loss": 1.1394, "step": 11677 }, { "epoch": 0.84, "grad_norm": 3.527964319973474, "learning_rate": 6.772467126434435e-07, "loss": 0.2123, "step": 11678 }, { "epoch": 0.84, "grad_norm": 4.739123252032527, "learning_rate": 6.766632147942643e-07, "loss": 0.5303, "step": 11679 }, { "epoch": 0.84, "grad_norm": 3.6013907238747365, "learning_rate": 6.760799501727411e-07, "loss": 0.1927, "step": 11680 }, { "epoch": 0.84, "grad_norm": 34.83131534765182, "learning_rate": 6.754969188103378e-07, "loss": 0.8475, "step": 11681 }, { "epoch": 0.84, "grad_norm": 14.842328099598046, "learning_rate": 6.749141207385063e-07, "loss": 0.685, "step": 11682 }, { "epoch": 0.84, "grad_norm": 4.245527248732805, "learning_rate": 6.743315559886854e-07, "loss": 0.1657, "step": 11683 }, { "epoch": 0.84, "grad_norm": 2.871254615199029, "learning_rate": 6.73749224592305e-07, "loss": 0.1588, "step": 11684 }, { "epoch": 0.84, "grad_norm": 13.221632436331072, "learning_rate": 6.731671265807782e-07, "loss": 0.6271, "step": 11685 }, { "epoch": 0.84, "grad_norm": 3.204114629183009, "learning_rate": 6.725852619855067e-07, "loss": 0.1351, "step": 11686 }, { "epoch": 0.84, "grad_norm": 11.20701749223786, "learning_rate": 6.720036308378808e-07, "loss": 0.4557, "step": 11687 }, { "epoch": 0.84, "grad_norm": 3.217263448737003, "learning_rate": 6.714222331692749e-07, "loss": 0.1557, "step": 11688 }, { "epoch": 0.84, "grad_norm": 11.211891538496081, "learning_rate": 6.708410690110562e-07, "loss": 0.4206, "step": 11689 }, { "epoch": 0.84, "grad_norm": 4.117839586372035, "learning_rate": 6.70260138394575e-07, "loss": 0.7078, "step": 11690 }, { "epoch": 0.84, "grad_norm": 3.3793595260401736, "learning_rate": 6.69679441351171e-07, "loss": 0.172, "step": 11691 }, { "epoch": 0.84, "grad_norm": 20.892110639796485, "learning_rate": 6.690989779121682e-07, "loss": 0.6206, "step": 11692 }, { "epoch": 0.84, "grad_norm": 3.926673409894006, "learning_rate": 6.685187481088834e-07, "loss": 0.2059, "step": 11693 }, { "epoch": 0.84, "grad_norm": 13.411701944795853, "learning_rate": 6.679387519726165e-07, "loss": 0.7664, "step": 11694 }, { "epoch": 0.84, "grad_norm": 22.39980381417035, "learning_rate": 6.673589895346555e-07, "loss": 0.7186, "step": 11695 }, { "epoch": 0.84, "grad_norm": 3.535688098066517, "learning_rate": 6.667794608262778e-07, "loss": 0.2216, "step": 11696 }, { "epoch": 0.84, "grad_norm": 15.660215854778805, "learning_rate": 6.662001658787476e-07, "loss": 1.1742, "step": 11697 }, { "epoch": 0.84, "grad_norm": 2.696961951779752, "learning_rate": 6.656211047233113e-07, "loss": 0.0982, "step": 11698 }, { "epoch": 0.84, "grad_norm": 3.0215785363399728, "learning_rate": 6.650422773912113e-07, "loss": 0.1148, "step": 11699 }, { "epoch": 0.84, "grad_norm": 4.774797271499911, "learning_rate": 6.64463683913672e-07, "loss": 0.2241, "step": 11700 }, { "epoch": 0.84, "grad_norm": 20.04425545055363, "learning_rate": 6.638853243219045e-07, "loss": 0.6017, "step": 11701 }, { "epoch": 0.84, "grad_norm": 4.509766508089323, "learning_rate": 6.633071986471123e-07, "loss": 0.8442, "step": 11702 }, { "epoch": 0.84, "grad_norm": 3.396205665116898, "learning_rate": 6.62729306920481e-07, "loss": 0.1811, "step": 11703 }, { "epoch": 0.84, "grad_norm": 4.834622910951188, "learning_rate": 6.621516491731849e-07, "loss": 1.0225, "step": 11704 }, { "epoch": 0.84, "grad_norm": 17.289643392894412, "learning_rate": 6.615742254363894e-07, "loss": 0.969, "step": 11705 }, { "epoch": 0.84, "grad_norm": 4.998366273148457, "learning_rate": 6.609970357412421e-07, "loss": 0.197, "step": 11706 }, { "epoch": 0.84, "grad_norm": 26.46817601737441, "learning_rate": 6.604200801188799e-07, "loss": 0.4818, "step": 11707 }, { "epoch": 0.84, "grad_norm": 5.373007696588976, "learning_rate": 6.598433586004288e-07, "loss": 0.8309, "step": 11708 }, { "epoch": 0.84, "grad_norm": 3.641984858285857, "learning_rate": 6.59266871217002e-07, "loss": 0.1505, "step": 11709 }, { "epoch": 0.84, "grad_norm": 15.976492416846758, "learning_rate": 6.586906179996938e-07, "loss": 0.5338, "step": 11710 }, { "epoch": 0.84, "grad_norm": 6.631379927483508, "learning_rate": 6.581145989795956e-07, "loss": 0.2466, "step": 11711 }, { "epoch": 0.84, "grad_norm": 18.703579322463515, "learning_rate": 6.575388141877804e-07, "loss": 0.7417, "step": 11712 }, { "epoch": 0.84, "grad_norm": 3.892642040678578, "learning_rate": 6.569632636553069e-07, "loss": 0.1878, "step": 11713 }, { "epoch": 0.84, "grad_norm": 10.2734488092979, "learning_rate": 6.563879474132279e-07, "loss": 0.4436, "step": 11714 }, { "epoch": 0.84, "grad_norm": 27.620668446402103, "learning_rate": 6.558128654925772e-07, "loss": 0.5431, "step": 11715 }, { "epoch": 0.84, "grad_norm": 12.538189208142162, "learning_rate": 6.552380179243778e-07, "loss": 0.7225, "step": 11716 }, { "epoch": 0.84, "grad_norm": 3.55668826959231, "learning_rate": 6.546634047396427e-07, "loss": 0.1569, "step": 11717 }, { "epoch": 0.84, "grad_norm": 2.5061025963735206, "learning_rate": 6.540890259693694e-07, "loss": 0.0845, "step": 11718 }, { "epoch": 0.84, "grad_norm": 9.840204008566973, "learning_rate": 6.535148816445425e-07, "loss": 0.387, "step": 11719 }, { "epoch": 0.84, "grad_norm": 15.676917588219629, "learning_rate": 6.529409717961343e-07, "loss": 0.6947, "step": 11720 }, { "epoch": 0.84, "grad_norm": 10.057697469703887, "learning_rate": 6.523672964551081e-07, "loss": 0.5821, "step": 11721 }, { "epoch": 0.84, "grad_norm": 2.860082755167475, "learning_rate": 6.517938556524079e-07, "loss": 0.1868, "step": 11722 }, { "epoch": 0.84, "grad_norm": 4.288109136646405, "learning_rate": 6.51220649418971e-07, "loss": 0.6285, "step": 11723 }, { "epoch": 0.84, "grad_norm": 3.1558187061740375, "learning_rate": 6.50647677785719e-07, "loss": 0.1864, "step": 11724 }, { "epoch": 0.84, "grad_norm": 2.5304918578959175, "learning_rate": 6.500749407835622e-07, "loss": 0.0791, "step": 11725 }, { "epoch": 0.84, "grad_norm": 6.9684056554794225, "learning_rate": 6.495024384433951e-07, "loss": 0.3223, "step": 11726 }, { "epoch": 0.84, "grad_norm": 4.800187281121672, "learning_rate": 6.489301707961054e-07, "loss": 0.8454, "step": 11727 }, { "epoch": 0.84, "grad_norm": 11.560564919127863, "learning_rate": 6.483581378725635e-07, "loss": 1.1196, "step": 11728 }, { "epoch": 0.84, "grad_norm": 7.694571482835537, "learning_rate": 6.477863397036272e-07, "loss": 0.3409, "step": 11729 }, { "epoch": 0.84, "grad_norm": 28.72425494074221, "learning_rate": 6.472147763201442e-07, "loss": 1.4329, "step": 11730 }, { "epoch": 0.84, "grad_norm": 4.4869882693285685, "learning_rate": 6.466434477529482e-07, "loss": 0.7079, "step": 11731 }, { "epoch": 0.84, "grad_norm": 4.251588974175203, "learning_rate": 6.46072354032859e-07, "loss": 0.1848, "step": 11732 }, { "epoch": 0.84, "grad_norm": 3.288677570194808, "learning_rate": 6.455014951906879e-07, "loss": 0.1071, "step": 11733 }, { "epoch": 0.84, "grad_norm": 4.1630037637913135, "learning_rate": 6.449308712572272e-07, "loss": 0.1623, "step": 11734 }, { "epoch": 0.84, "grad_norm": 13.764571087763773, "learning_rate": 6.443604822632604e-07, "loss": 0.648, "step": 11735 }, { "epoch": 0.84, "grad_norm": 2.8854155953156426, "learning_rate": 6.437903282395591e-07, "loss": 0.1383, "step": 11736 }, { "epoch": 0.84, "grad_norm": 9.976250354296031, "learning_rate": 6.432204092168808e-07, "loss": 0.298, "step": 11737 }, { "epoch": 0.84, "grad_norm": 3.0139419204153373, "learning_rate": 6.426507252259689e-07, "loss": 0.1769, "step": 11738 }, { "epoch": 0.84, "grad_norm": 16.275185443955742, "learning_rate": 6.420812762975581e-07, "loss": 0.739, "step": 11739 }, { "epoch": 0.84, "grad_norm": 3.4779443394531437, "learning_rate": 6.415120624623672e-07, "loss": 0.1696, "step": 11740 }, { "epoch": 0.84, "grad_norm": 3.4059033302417125, "learning_rate": 6.40943083751101e-07, "loss": 0.1094, "step": 11741 }, { "epoch": 0.84, "grad_norm": 11.959726768980563, "learning_rate": 6.403743401944573e-07, "loss": 0.6598, "step": 11742 }, { "epoch": 0.84, "grad_norm": 4.3615158559276415, "learning_rate": 6.398058318231154e-07, "loss": 0.7049, "step": 11743 }, { "epoch": 0.84, "grad_norm": 3.665625940320625, "learning_rate": 6.392375586677435e-07, "loss": 0.2343, "step": 11744 }, { "epoch": 0.84, "grad_norm": 3.2600104019604568, "learning_rate": 6.38669520759001e-07, "loss": 0.234, "step": 11745 }, { "epoch": 0.84, "grad_norm": 13.221289102741784, "learning_rate": 6.381017181275284e-07, "loss": 0.8566, "step": 11746 }, { "epoch": 0.84, "grad_norm": 4.400044483629993, "learning_rate": 6.375341508039567e-07, "loss": 0.7032, "step": 11747 }, { "epoch": 0.84, "grad_norm": 10.855518216907946, "learning_rate": 6.369668188189054e-07, "loss": 0.5643, "step": 11748 }, { "epoch": 0.84, "grad_norm": 8.942817220934518, "learning_rate": 6.363997222029794e-07, "loss": 0.284, "step": 11749 }, { "epoch": 0.84, "grad_norm": 14.034725495264633, "learning_rate": 6.358328609867703e-07, "loss": 0.5626, "step": 11750 }, { "epoch": 0.84, "grad_norm": 19.51928447865235, "learning_rate": 6.352662352008599e-07, "loss": 0.7131, "step": 11751 }, { "epoch": 0.84, "grad_norm": 3.2776425133272036, "learning_rate": 6.346998448758146e-07, "loss": 0.1032, "step": 11752 }, { "epoch": 0.84, "grad_norm": 19.945728391157182, "learning_rate": 6.341336900421896e-07, "loss": 0.5228, "step": 11753 }, { "epoch": 0.84, "grad_norm": 19.723147786235206, "learning_rate": 6.335677707305249e-07, "loss": 0.5961, "step": 11754 }, { "epoch": 0.84, "grad_norm": 15.842717863317207, "learning_rate": 6.330020869713521e-07, "loss": 0.9105, "step": 11755 }, { "epoch": 0.84, "grad_norm": 8.471637107416361, "learning_rate": 6.324366387951869e-07, "loss": 0.6271, "step": 11756 }, { "epoch": 0.84, "grad_norm": 6.396426390663739, "learning_rate": 6.318714262325331e-07, "loss": 0.3196, "step": 11757 }, { "epoch": 0.84, "grad_norm": 4.801326352669704, "learning_rate": 6.31306449313881e-07, "loss": 1.024, "step": 11758 }, { "epoch": 0.84, "grad_norm": 2.823860093359432, "learning_rate": 6.307417080697093e-07, "loss": 0.1375, "step": 11759 }, { "epoch": 0.84, "grad_norm": 4.0268568188186, "learning_rate": 6.301772025304843e-07, "loss": 0.9447, "step": 11760 }, { "epoch": 0.84, "grad_norm": 5.954448676217252, "learning_rate": 6.296129327266586e-07, "loss": 0.3011, "step": 11761 }, { "epoch": 0.84, "grad_norm": 2.8735673036005482, "learning_rate": 6.290488986886728e-07, "loss": 0.191, "step": 11762 }, { "epoch": 0.84, "grad_norm": 3.5466598748919536, "learning_rate": 6.284851004469528e-07, "loss": 0.1882, "step": 11763 }, { "epoch": 0.84, "grad_norm": 15.936960480458966, "learning_rate": 6.279215380319154e-07, "loss": 0.457, "step": 11764 }, { "epoch": 0.84, "grad_norm": 3.3157328635555254, "learning_rate": 6.273582114739618e-07, "loss": 0.1373, "step": 11765 }, { "epoch": 0.84, "grad_norm": 4.157762849373139, "learning_rate": 6.267951208034801e-07, "loss": 0.0969, "step": 11766 }, { "epoch": 0.84, "grad_norm": 9.969412894760968, "learning_rate": 6.262322660508496e-07, "loss": 0.7005, "step": 11767 }, { "epoch": 0.84, "grad_norm": 2.6079163989446483, "learning_rate": 6.256696472464324e-07, "loss": 0.1234, "step": 11768 }, { "epoch": 0.84, "grad_norm": 65.08830858078083, "learning_rate": 6.251072644205796e-07, "loss": 0.6075, "step": 11769 }, { "epoch": 0.84, "grad_norm": 9.772772800675266, "learning_rate": 6.245451176036316e-07, "loss": 0.4517, "step": 11770 }, { "epoch": 0.84, "grad_norm": 7.856565498789182, "learning_rate": 6.239832068259116e-07, "loss": 0.6193, "step": 11771 }, { "epoch": 0.84, "grad_norm": 4.7883869803218175, "learning_rate": 6.234215321177323e-07, "loss": 0.1965, "step": 11772 }, { "epoch": 0.84, "grad_norm": 25.485853201365757, "learning_rate": 6.22860093509396e-07, "loss": 0.9757, "step": 11773 }, { "epoch": 0.84, "grad_norm": 16.94325867697581, "learning_rate": 6.222988910311895e-07, "loss": 0.636, "step": 11774 }, { "epoch": 0.84, "grad_norm": 2.647831945835592, "learning_rate": 6.217379247133864e-07, "loss": 0.1283, "step": 11775 }, { "epoch": 0.84, "grad_norm": 13.665910323480373, "learning_rate": 6.211771945862505e-07, "loss": 0.4874, "step": 11776 }, { "epoch": 0.84, "grad_norm": 3.238003236944586, "learning_rate": 6.206167006800301e-07, "loss": 0.1723, "step": 11777 }, { "epoch": 0.84, "grad_norm": 3.4991055732366823, "learning_rate": 6.200564430249612e-07, "loss": 0.1749, "step": 11778 }, { "epoch": 0.84, "grad_norm": 15.014062370006894, "learning_rate": 6.194964216512689e-07, "loss": 0.7095, "step": 11779 }, { "epoch": 0.84, "grad_norm": 10.22442913288886, "learning_rate": 6.189366365891635e-07, "loss": 0.8183, "step": 11780 }, { "epoch": 0.84, "grad_norm": 8.091877118405588, "learning_rate": 6.183770878688422e-07, "loss": 0.4682, "step": 11781 }, { "epoch": 0.84, "grad_norm": 2.9360389614605245, "learning_rate": 6.178177755204945e-07, "loss": 0.0961, "step": 11782 }, { "epoch": 0.84, "grad_norm": 16.919022645919284, "learning_rate": 6.172586995742886e-07, "loss": 0.6192, "step": 11783 }, { "epoch": 0.85, "grad_norm": 17.048797036015106, "learning_rate": 6.166998600603852e-07, "loss": 0.7394, "step": 11784 }, { "epoch": 0.85, "grad_norm": 4.219058891207059, "learning_rate": 6.161412570089343e-07, "loss": 0.2374, "step": 11785 }, { "epoch": 0.85, "grad_norm": 4.404039759564226, "learning_rate": 6.155828904500683e-07, "loss": 0.7371, "step": 11786 }, { "epoch": 0.85, "grad_norm": 4.600681049702704, "learning_rate": 6.150247604139081e-07, "loss": 0.6834, "step": 11787 }, { "epoch": 0.85, "grad_norm": 25.102195386552125, "learning_rate": 6.144668669305659e-07, "loss": 1.0051, "step": 11788 }, { "epoch": 0.85, "grad_norm": 4.467777430467385, "learning_rate": 6.139092100301352e-07, "loss": 0.7402, "step": 11789 }, { "epoch": 0.85, "grad_norm": 3.4555424055764163, "learning_rate": 6.133517897427011e-07, "loss": 0.1833, "step": 11790 }, { "epoch": 0.85, "grad_norm": 4.772277887321884, "learning_rate": 6.127946060983325e-07, "loss": 0.917, "step": 11791 }, { "epoch": 0.85, "grad_norm": 3.400314942553233, "learning_rate": 6.12237659127089e-07, "loss": 0.1762, "step": 11792 }, { "epoch": 0.85, "grad_norm": 6.591210641391886, "learning_rate": 6.116809488590153e-07, "loss": 0.4751, "step": 11793 }, { "epoch": 0.85, "grad_norm": 15.71821647955643, "learning_rate": 6.111244753241441e-07, "loss": 0.8421, "step": 11794 }, { "epoch": 0.85, "grad_norm": 4.000851655514531, "learning_rate": 6.105682385524942e-07, "loss": 0.2517, "step": 11795 }, { "epoch": 0.85, "grad_norm": 10.648680826780671, "learning_rate": 6.100122385740726e-07, "loss": 0.4057, "step": 11796 }, { "epoch": 0.85, "grad_norm": 3.5401634934724426, "learning_rate": 6.094564754188742e-07, "loss": 0.1495, "step": 11797 }, { "epoch": 0.85, "grad_norm": 18.73842231117508, "learning_rate": 6.089009491168796e-07, "loss": 0.829, "step": 11798 }, { "epoch": 0.85, "grad_norm": 4.513875221148527, "learning_rate": 6.083456596980581e-07, "loss": 0.2446, "step": 11799 }, { "epoch": 0.85, "grad_norm": 17.109467611919722, "learning_rate": 6.077906071923633e-07, "loss": 0.7603, "step": 11800 }, { "epoch": 0.85, "grad_norm": 3.2317162445008654, "learning_rate": 6.072357916297417e-07, "loss": 0.174, "step": 11801 }, { "epoch": 0.85, "grad_norm": 3.1945461549211176, "learning_rate": 6.066812130401212e-07, "loss": 0.1314, "step": 11802 }, { "epoch": 0.85, "grad_norm": 3.413362267744799, "learning_rate": 6.061268714534185e-07, "loss": 0.12, "step": 11803 }, { "epoch": 0.85, "grad_norm": 8.215898567746686, "learning_rate": 6.055727668995404e-07, "loss": 0.3615, "step": 11804 }, { "epoch": 0.85, "grad_norm": 14.71955610780586, "learning_rate": 6.050188994083777e-07, "loss": 0.4966, "step": 11805 }, { "epoch": 0.85, "grad_norm": 2.86719705493606, "learning_rate": 6.044652690098096e-07, "loss": 0.0902, "step": 11806 }, { "epoch": 0.85, "grad_norm": 2.764134800918565, "learning_rate": 6.039118757337026e-07, "loss": 0.11, "step": 11807 }, { "epoch": 0.85, "grad_norm": 3.3121203711677905, "learning_rate": 6.033587196099094e-07, "loss": 0.1089, "step": 11808 }, { "epoch": 0.85, "grad_norm": 6.621215986036071, "learning_rate": 6.028058006682702e-07, "loss": 0.3472, "step": 11809 }, { "epoch": 0.85, "grad_norm": 3.7013835206499817, "learning_rate": 6.022531189386144e-07, "loss": 0.1897, "step": 11810 }, { "epoch": 0.85, "grad_norm": 8.658684672194052, "learning_rate": 6.01700674450757e-07, "loss": 0.4381, "step": 11811 }, { "epoch": 0.85, "grad_norm": 3.038148566330133, "learning_rate": 6.01148467234498e-07, "loss": 0.1832, "step": 11812 }, { "epoch": 0.85, "grad_norm": 4.1770317542783255, "learning_rate": 6.005964973196304e-07, "loss": 0.2223, "step": 11813 }, { "epoch": 0.85, "grad_norm": 3.3802130521434437, "learning_rate": 6.000447647359286e-07, "loss": 0.1948, "step": 11814 }, { "epoch": 0.85, "grad_norm": 2.9929512298933223, "learning_rate": 5.994932695131567e-07, "loss": 0.1336, "step": 11815 }, { "epoch": 0.85, "grad_norm": 4.886099069032137, "learning_rate": 5.989420116810663e-07, "loss": 0.7588, "step": 11816 }, { "epoch": 0.85, "grad_norm": 4.109011403018523, "learning_rate": 5.983909912693959e-07, "loss": 0.7512, "step": 11817 }, { "epoch": 0.85, "grad_norm": 2.642849172018695, "learning_rate": 5.978402083078705e-07, "loss": 0.1724, "step": 11818 }, { "epoch": 0.85, "grad_norm": 1.9887484099036346, "learning_rate": 5.972896628262026e-07, "loss": 0.1062, "step": 11819 }, { "epoch": 0.85, "grad_norm": 5.991935372961224, "learning_rate": 5.967393548540923e-07, "loss": 0.3503, "step": 11820 }, { "epoch": 0.85, "grad_norm": 3.175321937073813, "learning_rate": 5.961892844212258e-07, "loss": 0.228, "step": 11821 }, { "epoch": 0.85, "grad_norm": 2.8678981945310644, "learning_rate": 5.956394515572789e-07, "loss": 0.1264, "step": 11822 }, { "epoch": 0.85, "grad_norm": 10.4346704085874, "learning_rate": 5.950898562919122e-07, "loss": 0.5894, "step": 11823 }, { "epoch": 0.85, "grad_norm": 14.223645381234011, "learning_rate": 5.945404986547731e-07, "loss": 0.5669, "step": 11824 }, { "epoch": 0.85, "grad_norm": 34.04664321795518, "learning_rate": 5.939913786754997e-07, "loss": 1.1133, "step": 11825 }, { "epoch": 0.85, "grad_norm": 3.9939149321111613, "learning_rate": 5.934424963837138e-07, "loss": 0.1725, "step": 11826 }, { "epoch": 0.85, "grad_norm": 3.7759722001555796, "learning_rate": 5.928938518090255e-07, "loss": 0.1526, "step": 11827 }, { "epoch": 0.85, "grad_norm": 3.309017377586742, "learning_rate": 5.923454449810306e-07, "loss": 0.0753, "step": 11828 }, { "epoch": 0.85, "grad_norm": 4.726859721908704, "learning_rate": 5.917972759293166e-07, "loss": 0.6648, "step": 11829 }, { "epoch": 0.85, "grad_norm": 29.712708740155353, "learning_rate": 5.912493446834533e-07, "loss": 1.3363, "step": 11830 }, { "epoch": 0.85, "grad_norm": 12.504596363698782, "learning_rate": 5.907016512730002e-07, "loss": 0.5786, "step": 11831 }, { "epoch": 0.85, "grad_norm": 14.115425664529743, "learning_rate": 5.901541957275026e-07, "loss": 0.4853, "step": 11832 }, { "epoch": 0.85, "grad_norm": 12.830637788993835, "learning_rate": 5.896069780764935e-07, "loss": 0.2844, "step": 11833 }, { "epoch": 0.85, "grad_norm": 2.2044812255584536, "learning_rate": 5.890599983494938e-07, "loss": 0.0934, "step": 11834 }, { "epoch": 0.85, "grad_norm": 23.13726777400695, "learning_rate": 5.885132565760116e-07, "loss": 0.7925, "step": 11835 }, { "epoch": 0.85, "grad_norm": 4.858943971664273, "learning_rate": 5.879667527855404e-07, "loss": 0.286, "step": 11836 }, { "epoch": 0.85, "grad_norm": 3.521507881371635, "learning_rate": 5.874204870075617e-07, "loss": 0.1463, "step": 11837 }, { "epoch": 0.85, "grad_norm": 4.233450075386712, "learning_rate": 5.868744592715464e-07, "loss": 0.1289, "step": 11838 }, { "epoch": 0.85, "grad_norm": 13.278343494053928, "learning_rate": 5.863286696069492e-07, "loss": 0.8312, "step": 11839 }, { "epoch": 0.85, "grad_norm": 8.321991887551588, "learning_rate": 5.857831180432122e-07, "loss": 0.2144, "step": 11840 }, { "epoch": 0.85, "grad_norm": 3.1340584490340646, "learning_rate": 5.85237804609769e-07, "loss": 0.1049, "step": 11841 }, { "epoch": 0.85, "grad_norm": 4.796771419821607, "learning_rate": 5.84692729336036e-07, "loss": 0.6287, "step": 11842 }, { "epoch": 0.85, "grad_norm": 3.4991241762457026, "learning_rate": 5.841478922514154e-07, "loss": 0.1939, "step": 11843 }, { "epoch": 0.85, "grad_norm": 2.7718526197125186, "learning_rate": 5.836032933853019e-07, "loss": 0.2013, "step": 11844 }, { "epoch": 0.85, "grad_norm": 5.4347297002206245, "learning_rate": 5.830589327670744e-07, "loss": 0.6666, "step": 11845 }, { "epoch": 0.85, "grad_norm": 16.770249537909734, "learning_rate": 5.825148104260969e-07, "loss": 0.5551, "step": 11846 }, { "epoch": 0.85, "grad_norm": 3.044181918591866, "learning_rate": 5.819709263917256e-07, "loss": 0.0906, "step": 11847 }, { "epoch": 0.85, "grad_norm": 29.62608060512738, "learning_rate": 5.814272806932997e-07, "loss": 0.7422, "step": 11848 }, { "epoch": 0.85, "grad_norm": 14.033095846297897, "learning_rate": 5.808838733601458e-07, "loss": 0.4895, "step": 11849 }, { "epoch": 0.85, "grad_norm": 13.211593469218645, "learning_rate": 5.803407044215803e-07, "loss": 0.7677, "step": 11850 }, { "epoch": 0.85, "grad_norm": 4.1364811362128195, "learning_rate": 5.79797773906905e-07, "loss": 0.6237, "step": 11851 }, { "epoch": 0.85, "grad_norm": 4.277195073960571, "learning_rate": 5.792550818454068e-07, "loss": 0.2846, "step": 11852 }, { "epoch": 0.85, "grad_norm": 2.535599181251336, "learning_rate": 5.787126282663647e-07, "loss": 0.0729, "step": 11853 }, { "epoch": 0.85, "grad_norm": 3.534424196489536, "learning_rate": 5.781704131990423e-07, "loss": 0.125, "step": 11854 }, { "epoch": 0.85, "grad_norm": 27.070616310337236, "learning_rate": 5.776284366726858e-07, "loss": 0.5231, "step": 11855 }, { "epoch": 0.85, "grad_norm": 4.408527914340486, "learning_rate": 5.770866987165374e-07, "loss": 0.8163, "step": 11856 }, { "epoch": 0.85, "grad_norm": 22.643031564020585, "learning_rate": 5.765451993598198e-07, "loss": 0.6717, "step": 11857 }, { "epoch": 0.85, "grad_norm": 4.104032107987074, "learning_rate": 5.760039386317434e-07, "loss": 0.1749, "step": 11858 }, { "epoch": 0.85, "grad_norm": 5.063478674935331, "learning_rate": 5.7546291656151e-07, "loss": 0.1334, "step": 11859 }, { "epoch": 0.85, "grad_norm": 6.452496804817486, "learning_rate": 5.74922133178305e-07, "loss": 0.2554, "step": 11860 }, { "epoch": 0.85, "grad_norm": 21.575473312800302, "learning_rate": 5.743815885112997e-07, "loss": 0.6656, "step": 11861 }, { "epoch": 0.85, "grad_norm": 18.260299750714786, "learning_rate": 5.738412825896566e-07, "loss": 0.568, "step": 11862 }, { "epoch": 0.85, "grad_norm": 4.9455939013957995, "learning_rate": 5.733012154425227e-07, "loss": 0.8664, "step": 11863 }, { "epoch": 0.85, "grad_norm": 3.0829196787942257, "learning_rate": 5.727613870990323e-07, "loss": 0.091, "step": 11864 }, { "epoch": 0.85, "grad_norm": 4.4582906542977305, "learning_rate": 5.72221797588306e-07, "loss": 0.9004, "step": 11865 }, { "epoch": 0.85, "grad_norm": 3.103970371963963, "learning_rate": 5.716824469394555e-07, "loss": 0.1661, "step": 11866 }, { "epoch": 0.85, "grad_norm": 3.0380780196843844, "learning_rate": 5.711433351815732e-07, "loss": 0.1083, "step": 11867 }, { "epoch": 0.85, "grad_norm": 3.8401637771047556, "learning_rate": 5.706044623437446e-07, "loss": 0.2792, "step": 11868 }, { "epoch": 0.85, "grad_norm": 4.369506159237035, "learning_rate": 5.700658284550387e-07, "loss": 0.1958, "step": 11869 }, { "epoch": 0.85, "grad_norm": 15.909939235908363, "learning_rate": 5.695274335445139e-07, "loss": 0.5374, "step": 11870 }, { "epoch": 0.85, "grad_norm": 2.9762042249734284, "learning_rate": 5.689892776412126e-07, "loss": 0.254, "step": 11871 }, { "epoch": 0.85, "grad_norm": 14.606711554881997, "learning_rate": 5.684513607741682e-07, "loss": 0.6182, "step": 11872 }, { "epoch": 0.85, "grad_norm": 3.2074227099730255, "learning_rate": 5.679136829723986e-07, "loss": 0.1722, "step": 11873 }, { "epoch": 0.85, "grad_norm": 18.78991270727046, "learning_rate": 5.673762442649084e-07, "loss": 0.3635, "step": 11874 }, { "epoch": 0.85, "grad_norm": 3.4520821101275367, "learning_rate": 5.668390446806926e-07, "loss": 0.1333, "step": 11875 }, { "epoch": 0.85, "grad_norm": 17.104483004925054, "learning_rate": 5.663020842487299e-07, "loss": 0.8431, "step": 11876 }, { "epoch": 0.85, "grad_norm": 4.134001154079176, "learning_rate": 5.657653629979865e-07, "loss": 0.2196, "step": 11877 }, { "epoch": 0.85, "grad_norm": 5.601659739016702, "learning_rate": 5.652288809574191e-07, "loss": 0.2505, "step": 11878 }, { "epoch": 0.85, "grad_norm": 4.364129173187812, "learning_rate": 5.646926381559658e-07, "loss": 0.7895, "step": 11879 }, { "epoch": 0.85, "grad_norm": 6.808178038099225, "learning_rate": 5.641566346225557e-07, "loss": 0.1969, "step": 11880 }, { "epoch": 0.85, "grad_norm": 11.956689556778565, "learning_rate": 5.636208703861051e-07, "loss": 0.4629, "step": 11881 }, { "epoch": 0.85, "grad_norm": 4.281204924736169, "learning_rate": 5.630853454755164e-07, "loss": 0.9176, "step": 11882 }, { "epoch": 0.85, "grad_norm": 3.572475168293842, "learning_rate": 5.625500599196771e-07, "loss": 0.1846, "step": 11883 }, { "epoch": 0.85, "grad_norm": 18.85532822405509, "learning_rate": 5.620150137474667e-07, "loss": 0.8414, "step": 11884 }, { "epoch": 0.85, "grad_norm": 3.5310358669347557, "learning_rate": 5.614802069877478e-07, "loss": 0.2094, "step": 11885 }, { "epoch": 0.85, "grad_norm": 5.739793283726222, "learning_rate": 5.609456396693697e-07, "loss": 0.2514, "step": 11886 }, { "epoch": 0.85, "grad_norm": 9.435410585729695, "learning_rate": 5.604113118211724e-07, "loss": 0.4798, "step": 11887 }, { "epoch": 0.85, "grad_norm": 19.287051967259064, "learning_rate": 5.598772234719807e-07, "loss": 1.0869, "step": 11888 }, { "epoch": 0.85, "grad_norm": 3.105637436686062, "learning_rate": 5.593433746506044e-07, "loss": 0.1311, "step": 11889 }, { "epoch": 0.85, "grad_norm": 3.1032521343964747, "learning_rate": 5.588097653858465e-07, "loss": 0.1189, "step": 11890 }, { "epoch": 0.85, "grad_norm": 4.895180443255786, "learning_rate": 5.582763957064902e-07, "loss": 0.7796, "step": 11891 }, { "epoch": 0.85, "grad_norm": 3.4220315372612284, "learning_rate": 5.577432656413084e-07, "loss": 0.1815, "step": 11892 }, { "epoch": 0.85, "grad_norm": 3.600614849663183, "learning_rate": 5.572103752190632e-07, "loss": 0.1463, "step": 11893 }, { "epoch": 0.85, "grad_norm": 3.9556871329795555, "learning_rate": 5.566777244685017e-07, "loss": 0.1858, "step": 11894 }, { "epoch": 0.85, "grad_norm": 3.33856553879353, "learning_rate": 5.56145313418357e-07, "loss": 0.0853, "step": 11895 }, { "epoch": 0.85, "grad_norm": 21.594909030677496, "learning_rate": 5.556131420973531e-07, "loss": 0.4833, "step": 11896 }, { "epoch": 0.85, "grad_norm": 4.856710245584923, "learning_rate": 5.550812105341968e-07, "loss": 0.6193, "step": 11897 }, { "epoch": 0.85, "grad_norm": 18.623991849304467, "learning_rate": 5.545495187575845e-07, "loss": 0.6933, "step": 11898 }, { "epoch": 0.85, "grad_norm": 4.558002664625495, "learning_rate": 5.540180667961975e-07, "loss": 0.2931, "step": 11899 }, { "epoch": 0.85, "grad_norm": 6.32739638458224, "learning_rate": 5.534868546787081e-07, "loss": 0.2789, "step": 11900 }, { "epoch": 0.85, "grad_norm": 4.879188649476771, "learning_rate": 5.529558824337716e-07, "loss": 0.7994, "step": 11901 }, { "epoch": 0.85, "grad_norm": 2.965456088995654, "learning_rate": 5.524251500900324e-07, "loss": 0.2091, "step": 11902 }, { "epoch": 0.85, "grad_norm": 3.9185027687161664, "learning_rate": 5.518946576761213e-07, "loss": 0.7217, "step": 11903 }, { "epoch": 0.85, "grad_norm": 12.802615030804033, "learning_rate": 5.513644052206557e-07, "loss": 0.6846, "step": 11904 }, { "epoch": 0.85, "grad_norm": 18.2806300696795, "learning_rate": 5.508343927522425e-07, "loss": 0.6978, "step": 11905 }, { "epoch": 0.85, "grad_norm": 13.856025302365863, "learning_rate": 5.503046202994722e-07, "loss": 0.314, "step": 11906 }, { "epoch": 0.85, "grad_norm": 3.109093314381208, "learning_rate": 5.497750878909253e-07, "loss": 0.123, "step": 11907 }, { "epoch": 0.85, "grad_norm": 10.405115116906384, "learning_rate": 5.492457955551661e-07, "loss": 0.4274, "step": 11908 }, { "epoch": 0.85, "grad_norm": 35.963414089580105, "learning_rate": 5.487167433207502e-07, "loss": 0.8314, "step": 11909 }, { "epoch": 0.85, "grad_norm": 18.690403621945535, "learning_rate": 5.48187931216217e-07, "loss": 0.3083, "step": 11910 }, { "epoch": 0.85, "grad_norm": 3.2575246316084585, "learning_rate": 5.47659359270093e-07, "loss": 0.1594, "step": 11911 }, { "epoch": 0.85, "grad_norm": 3.0019468388653037, "learning_rate": 5.471310275108943e-07, "loss": 0.1483, "step": 11912 }, { "epoch": 0.85, "grad_norm": 3.5973819006554453, "learning_rate": 5.466029359671215e-07, "loss": 0.121, "step": 11913 }, { "epoch": 0.85, "grad_norm": 16.989689594781904, "learning_rate": 5.460750846672635e-07, "loss": 0.3666, "step": 11914 }, { "epoch": 0.85, "grad_norm": 10.875517392074569, "learning_rate": 5.455474736397959e-07, "loss": 0.8824, "step": 11915 }, { "epoch": 0.85, "grad_norm": 5.287296520273689, "learning_rate": 5.450201029131808e-07, "loss": 0.724, "step": 11916 }, { "epoch": 0.85, "grad_norm": 14.541841661270581, "learning_rate": 5.444929725158671e-07, "loss": 1.116, "step": 11917 }, { "epoch": 0.85, "grad_norm": 4.963200968205714, "learning_rate": 5.439660824762938e-07, "loss": 0.731, "step": 11918 }, { "epoch": 0.85, "grad_norm": 12.972123157328369, "learning_rate": 5.434394328228831e-07, "loss": 0.4184, "step": 11919 }, { "epoch": 0.85, "grad_norm": 13.572152650300875, "learning_rate": 5.429130235840451e-07, "loss": 0.4952, "step": 11920 }, { "epoch": 0.85, "grad_norm": 8.871300981135814, "learning_rate": 5.423868547881795e-07, "loss": 0.4421, "step": 11921 }, { "epoch": 0.85, "grad_norm": 2.466723878537534, "learning_rate": 5.418609264636698e-07, "loss": 0.1211, "step": 11922 }, { "epoch": 0.85, "grad_norm": 4.187695098029085, "learning_rate": 5.413352386388876e-07, "loss": 0.6737, "step": 11923 }, { "epoch": 0.86, "grad_norm": 25.67891408156512, "learning_rate": 5.408097913421934e-07, "loss": 0.8739, "step": 11924 }, { "epoch": 0.86, "grad_norm": 3.340565927940079, "learning_rate": 5.402845846019317e-07, "loss": 0.1071, "step": 11925 }, { "epoch": 0.86, "grad_norm": 3.1345218351346875, "learning_rate": 5.397596184464354e-07, "loss": 0.1002, "step": 11926 }, { "epoch": 0.86, "grad_norm": 3.5784963363206255, "learning_rate": 5.392348929040254e-07, "loss": 0.1274, "step": 11927 }, { "epoch": 0.86, "grad_norm": 3.6722909063592932, "learning_rate": 5.387104080030076e-07, "loss": 0.1302, "step": 11928 }, { "epoch": 0.86, "grad_norm": 3.0508097188508336, "learning_rate": 5.381861637716757e-07, "loss": 0.121, "step": 11929 }, { "epoch": 0.86, "grad_norm": 4.826810630600062, "learning_rate": 5.37662160238312e-07, "loss": 0.191, "step": 11930 }, { "epoch": 0.86, "grad_norm": 4.239595503850786, "learning_rate": 5.371383974311839e-07, "loss": 0.1494, "step": 11931 }, { "epoch": 0.86, "grad_norm": 8.6626179869553, "learning_rate": 5.36614875378546e-07, "loss": 0.5446, "step": 11932 }, { "epoch": 0.86, "grad_norm": 2.3051104741423094, "learning_rate": 5.360915941086414e-07, "loss": 0.0827, "step": 11933 }, { "epoch": 0.86, "grad_norm": 3.9337970081951, "learning_rate": 5.35568553649698e-07, "loss": 0.7428, "step": 11934 }, { "epoch": 0.86, "grad_norm": 16.551589999103562, "learning_rate": 5.350457540299331e-07, "loss": 0.4238, "step": 11935 }, { "epoch": 0.86, "grad_norm": 16.46206676292899, "learning_rate": 5.345231952775476e-07, "loss": 0.5265, "step": 11936 }, { "epoch": 0.86, "grad_norm": 13.396747843041952, "learning_rate": 5.34000877420734e-07, "loss": 0.8137, "step": 11937 }, { "epoch": 0.86, "grad_norm": 17.588342997617424, "learning_rate": 5.334788004876684e-07, "loss": 0.7557, "step": 11938 }, { "epoch": 0.86, "grad_norm": 4.534650490627462, "learning_rate": 5.329569645065147e-07, "loss": 0.1678, "step": 11939 }, { "epoch": 0.86, "grad_norm": 19.526842300186544, "learning_rate": 5.324353695054241e-07, "loss": 0.5518, "step": 11940 }, { "epoch": 0.86, "grad_norm": 3.4389193396496474, "learning_rate": 5.319140155125341e-07, "loss": 0.1689, "step": 11941 }, { "epoch": 0.86, "grad_norm": 3.6124423185958325, "learning_rate": 5.313929025559705e-07, "loss": 0.1918, "step": 11942 }, { "epoch": 0.86, "grad_norm": 19.363413665362053, "learning_rate": 5.308720306638459e-07, "loss": 0.3864, "step": 11943 }, { "epoch": 0.86, "grad_norm": 2.734651851862434, "learning_rate": 5.303513998642584e-07, "loss": 0.1022, "step": 11944 }, { "epoch": 0.86, "grad_norm": 3.6452949067359035, "learning_rate": 5.298310101852938e-07, "loss": 0.1493, "step": 11945 }, { "epoch": 0.86, "grad_norm": 4.956673910830199, "learning_rate": 5.293108616550263e-07, "loss": 0.9039, "step": 11946 }, { "epoch": 0.86, "grad_norm": 2.9136382990183436, "learning_rate": 5.287909543015152e-07, "loss": 0.1501, "step": 11947 }, { "epoch": 0.86, "grad_norm": 10.20152653445203, "learning_rate": 5.282712881528068e-07, "loss": 0.5148, "step": 11948 }, { "epoch": 0.86, "grad_norm": 3.3054718673005663, "learning_rate": 5.277518632369372e-07, "loss": 0.1558, "step": 11949 }, { "epoch": 0.86, "grad_norm": 3.703328089590816, "learning_rate": 5.272326795819266e-07, "loss": 0.2569, "step": 11950 }, { "epoch": 0.86, "grad_norm": 16.271057661783836, "learning_rate": 5.26713737215781e-07, "loss": 0.2089, "step": 11951 }, { "epoch": 0.86, "grad_norm": 5.077888450457996, "learning_rate": 5.26195036166498e-07, "loss": 0.234, "step": 11952 }, { "epoch": 0.86, "grad_norm": 20.411676846468396, "learning_rate": 5.256765764620586e-07, "loss": 0.8167, "step": 11953 }, { "epoch": 0.86, "grad_norm": 3.479207897270621, "learning_rate": 5.251583581304304e-07, "loss": 0.1488, "step": 11954 }, { "epoch": 0.86, "grad_norm": 2.8148414910859065, "learning_rate": 5.24640381199572e-07, "loss": 0.0803, "step": 11955 }, { "epoch": 0.86, "grad_norm": 25.517908813843654, "learning_rate": 5.241226456974247e-07, "loss": 0.9119, "step": 11956 }, { "epoch": 0.86, "grad_norm": 6.936749120188616, "learning_rate": 5.236051516519175e-07, "loss": 0.4601, "step": 11957 }, { "epoch": 0.86, "grad_norm": 4.5813369510675255, "learning_rate": 5.230878990909688e-07, "loss": 0.7359, "step": 11958 }, { "epoch": 0.86, "grad_norm": 4.454760981507435, "learning_rate": 5.225708880424823e-07, "loss": 0.7649, "step": 11959 }, { "epoch": 0.86, "grad_norm": 5.7453959876630325, "learning_rate": 5.220541185343475e-07, "loss": 0.2482, "step": 11960 }, { "epoch": 0.86, "grad_norm": 17.88045307541127, "learning_rate": 5.215375905944442e-07, "loss": 0.4745, "step": 11961 }, { "epoch": 0.86, "grad_norm": 16.535870990653812, "learning_rate": 5.210213042506368e-07, "loss": 0.5783, "step": 11962 }, { "epoch": 0.86, "grad_norm": 51.26428607988013, "learning_rate": 5.205052595307741e-07, "loss": 1.1685, "step": 11963 }, { "epoch": 0.86, "grad_norm": 8.928056007829811, "learning_rate": 5.199894564626984e-07, "loss": 0.3168, "step": 11964 }, { "epoch": 0.86, "grad_norm": 13.560491075684379, "learning_rate": 5.194738950742328e-07, "loss": 0.8271, "step": 11965 }, { "epoch": 0.86, "grad_norm": 2.903005097602227, "learning_rate": 5.189585753931908e-07, "loss": 0.0908, "step": 11966 }, { "epoch": 0.86, "grad_norm": 3.706748288837731, "learning_rate": 5.184434974473723e-07, "loss": 0.2456, "step": 11967 }, { "epoch": 0.86, "grad_norm": 4.338220791537947, "learning_rate": 5.17928661264564e-07, "loss": 0.723, "step": 11968 }, { "epoch": 0.86, "grad_norm": 23.89366708348035, "learning_rate": 5.174140668725374e-07, "loss": 0.775, "step": 11969 }, { "epoch": 0.86, "grad_norm": 2.9190968487271487, "learning_rate": 5.168997142990556e-07, "loss": 0.0908, "step": 11970 }, { "epoch": 0.86, "grad_norm": 12.729931300964756, "learning_rate": 5.163856035718645e-07, "loss": 0.3317, "step": 11971 }, { "epoch": 0.86, "grad_norm": 4.4187170993660425, "learning_rate": 5.158717347186993e-07, "loss": 0.8322, "step": 11972 }, { "epoch": 0.86, "grad_norm": 4.046784225571753, "learning_rate": 5.153581077672792e-07, "loss": 0.817, "step": 11973 }, { "epoch": 0.86, "grad_norm": 22.953362947870804, "learning_rate": 5.148447227453157e-07, "loss": 0.4393, "step": 11974 }, { "epoch": 0.86, "grad_norm": 3.621087403046301, "learning_rate": 5.14331579680501e-07, "loss": 0.1255, "step": 11975 }, { "epoch": 0.86, "grad_norm": 14.842751262096229, "learning_rate": 5.138186786005184e-07, "loss": 0.6582, "step": 11976 }, { "epoch": 0.86, "grad_norm": 13.92576588527017, "learning_rate": 5.133060195330375e-07, "loss": 0.8122, "step": 11977 }, { "epoch": 0.86, "grad_norm": 8.14451199916439, "learning_rate": 5.12793602505714e-07, "loss": 0.2598, "step": 11978 }, { "epoch": 0.86, "grad_norm": 4.4742828897801985, "learning_rate": 5.12281427546189e-07, "loss": 0.1926, "step": 11979 }, { "epoch": 0.86, "grad_norm": 4.333854498106117, "learning_rate": 5.117694946820956e-07, "loss": 0.9444, "step": 11980 }, { "epoch": 0.86, "grad_norm": 2.8069383179895935, "learning_rate": 5.112578039410487e-07, "loss": 0.1676, "step": 11981 }, { "epoch": 0.86, "grad_norm": 3.7825935803836446, "learning_rate": 5.10746355350652e-07, "loss": 0.2707, "step": 11982 }, { "epoch": 0.86, "grad_norm": 3.1139832840883375, "learning_rate": 5.102351489384972e-07, "loss": 0.1355, "step": 11983 }, { "epoch": 0.86, "grad_norm": 17.220985659302627, "learning_rate": 5.097241847321615e-07, "loss": 0.413, "step": 11984 }, { "epoch": 0.86, "grad_norm": 9.991275490048174, "learning_rate": 5.092134627592088e-07, "loss": 1.3525, "step": 11985 }, { "epoch": 0.86, "grad_norm": 4.870903292544525, "learning_rate": 5.087029830471918e-07, "loss": 0.8622, "step": 11986 }, { "epoch": 0.86, "grad_norm": 3.78572619827005, "learning_rate": 5.081927456236496e-07, "loss": 0.1915, "step": 11987 }, { "epoch": 0.86, "grad_norm": 3.245272114798643, "learning_rate": 5.076827505161042e-07, "loss": 0.1835, "step": 11988 }, { "epoch": 0.86, "grad_norm": 8.944143852472784, "learning_rate": 5.071729977520712e-07, "loss": 0.1795, "step": 11989 }, { "epoch": 0.86, "grad_norm": 2.3495151425286918, "learning_rate": 5.066634873590486e-07, "loss": 0.1261, "step": 11990 }, { "epoch": 0.86, "grad_norm": 4.252308543544083, "learning_rate": 5.061542193645219e-07, "loss": 0.1742, "step": 11991 }, { "epoch": 0.86, "grad_norm": 22.68429824197646, "learning_rate": 5.056451937959656e-07, "loss": 0.6421, "step": 11992 }, { "epoch": 0.86, "grad_norm": 3.0939634517870194, "learning_rate": 5.051364106808392e-07, "loss": 0.1169, "step": 11993 }, { "epoch": 0.86, "grad_norm": 3.452673350099406, "learning_rate": 5.046278700465884e-07, "loss": 0.1611, "step": 11994 }, { "epoch": 0.86, "grad_norm": 2.848267796368961, "learning_rate": 5.041195719206493e-07, "loss": 0.1257, "step": 11995 }, { "epoch": 0.86, "grad_norm": 2.893429424622065, "learning_rate": 5.036115163304411e-07, "loss": 0.0701, "step": 11996 }, { "epoch": 0.86, "grad_norm": 10.31392654961928, "learning_rate": 5.031037033033709e-07, "loss": 0.3254, "step": 11997 }, { "epoch": 0.86, "grad_norm": 9.723586770179375, "learning_rate": 5.025961328668355e-07, "loss": 0.6946, "step": 11998 }, { "epoch": 0.86, "grad_norm": 4.1425013397808454, "learning_rate": 5.020888050482158e-07, "loss": 0.1897, "step": 11999 }, { "epoch": 0.86, "grad_norm": 4.528439241724901, "learning_rate": 5.01581719874878e-07, "loss": 0.8347, "step": 12000 }, { "epoch": 0.86, "grad_norm": 3.8678187722406254, "learning_rate": 5.010748773741798e-07, "loss": 0.1885, "step": 12001 }, { "epoch": 0.86, "grad_norm": 32.01141948724884, "learning_rate": 5.005682775734627e-07, "loss": 0.856, "step": 12002 }, { "epoch": 0.86, "grad_norm": 4.704062006672816, "learning_rate": 5.000619205000545e-07, "loss": 0.7928, "step": 12003 }, { "epoch": 0.86, "grad_norm": 4.611794029984421, "learning_rate": 4.995558061812734e-07, "loss": 0.729, "step": 12004 }, { "epoch": 0.86, "grad_norm": 4.6506700840690325, "learning_rate": 4.990499346444216e-07, "loss": 0.8368, "step": 12005 }, { "epoch": 0.86, "grad_norm": 4.642887955933612, "learning_rate": 4.985443059167877e-07, "loss": 0.8933, "step": 12006 }, { "epoch": 0.86, "grad_norm": 26.644033585269355, "learning_rate": 4.980389200256508e-07, "loss": 0.6303, "step": 12007 }, { "epoch": 0.86, "grad_norm": 4.797502272382234, "learning_rate": 4.975337769982736e-07, "loss": 0.774, "step": 12008 }, { "epoch": 0.86, "grad_norm": 4.9351686215277, "learning_rate": 4.970288768619058e-07, "loss": 0.8716, "step": 12009 }, { "epoch": 0.86, "grad_norm": 17.096252762556798, "learning_rate": 4.965242196437848e-07, "loss": 0.796, "step": 12010 }, { "epoch": 0.86, "grad_norm": 14.313287504173363, "learning_rate": 4.960198053711374e-07, "loss": 0.6035, "step": 12011 }, { "epoch": 0.86, "grad_norm": 12.57617041647316, "learning_rate": 4.955156340711709e-07, "loss": 0.6762, "step": 12012 }, { "epoch": 0.86, "grad_norm": 4.104579423873017, "learning_rate": 4.950117057710863e-07, "loss": 0.7456, "step": 12013 }, { "epoch": 0.86, "grad_norm": 3.0753957089064525, "learning_rate": 4.945080204980685e-07, "loss": 0.1462, "step": 12014 }, { "epoch": 0.86, "grad_norm": 3.1103037045661863, "learning_rate": 4.940045782792885e-07, "loss": 0.2084, "step": 12015 }, { "epoch": 0.86, "grad_norm": 6.368030235930793, "learning_rate": 4.935013791419041e-07, "loss": 0.2398, "step": 12016 }, { "epoch": 0.86, "grad_norm": 12.320545073487855, "learning_rate": 4.929984231130635e-07, "loss": 0.7825, "step": 12017 }, { "epoch": 0.86, "grad_norm": 7.397282164394771, "learning_rate": 4.92495710219898e-07, "loss": 0.5414, "step": 12018 }, { "epoch": 0.86, "grad_norm": 4.077458014380638, "learning_rate": 4.919932404895267e-07, "loss": 0.2308, "step": 12019 }, { "epoch": 0.86, "grad_norm": 2.137351827007736, "learning_rate": 4.914910139490564e-07, "loss": 0.0901, "step": 12020 }, { "epoch": 0.86, "grad_norm": 17.855718930482997, "learning_rate": 4.90989030625581e-07, "loss": 0.6193, "step": 12021 }, { "epoch": 0.86, "grad_norm": 11.66111685303585, "learning_rate": 4.904872905461782e-07, "loss": 0.7412, "step": 12022 }, { "epoch": 0.86, "grad_norm": 4.452473945278823, "learning_rate": 4.899857937379193e-07, "loss": 0.2206, "step": 12023 }, { "epoch": 0.86, "grad_norm": 3.098445389121844, "learning_rate": 4.894845402278542e-07, "loss": 0.1872, "step": 12024 }, { "epoch": 0.86, "grad_norm": 93.96727836158391, "learning_rate": 4.889835300430241e-07, "loss": 1.3044, "step": 12025 }, { "epoch": 0.86, "grad_norm": 23.76617641027894, "learning_rate": 4.884827632104577e-07, "loss": 0.527, "step": 12026 }, { "epoch": 0.86, "grad_norm": 3.898554870611414, "learning_rate": 4.879822397571698e-07, "loss": 0.2302, "step": 12027 }, { "epoch": 0.86, "grad_norm": 5.188417689653673, "learning_rate": 4.874819597101605e-07, "loss": 0.8475, "step": 12028 }, { "epoch": 0.86, "grad_norm": 3.0541813117614063, "learning_rate": 4.869819230964191e-07, "loss": 0.1088, "step": 12029 }, { "epoch": 0.86, "grad_norm": 3.9797918917405823, "learning_rate": 4.864821299429207e-07, "loss": 0.1558, "step": 12030 }, { "epoch": 0.86, "grad_norm": 18.91148865265908, "learning_rate": 4.859825802766255e-07, "loss": 0.5509, "step": 12031 }, { "epoch": 0.86, "grad_norm": 2.8163714734141463, "learning_rate": 4.854832741244841e-07, "loss": 0.1375, "step": 12032 }, { "epoch": 0.86, "grad_norm": 3.2645854039941877, "learning_rate": 4.849842115134323e-07, "loss": 0.1969, "step": 12033 }, { "epoch": 0.86, "grad_norm": 3.650981346958607, "learning_rate": 4.844853924703913e-07, "loss": 0.1657, "step": 12034 }, { "epoch": 0.86, "grad_norm": 4.701023810720173, "learning_rate": 4.839868170222728e-07, "loss": 0.7975, "step": 12035 }, { "epoch": 0.86, "grad_norm": 2.9034643280878702, "learning_rate": 4.834884851959703e-07, "loss": 0.1238, "step": 12036 }, { "epoch": 0.86, "grad_norm": 3.9324613759119402, "learning_rate": 4.829903970183675e-07, "loss": 0.1723, "step": 12037 }, { "epoch": 0.86, "grad_norm": 2.853745373461182, "learning_rate": 4.82492552516336e-07, "loss": 0.1165, "step": 12038 }, { "epoch": 0.86, "grad_norm": 3.2236708051080396, "learning_rate": 4.81994951716731e-07, "loss": 0.1389, "step": 12039 }, { "epoch": 0.86, "grad_norm": 4.4080322933827105, "learning_rate": 4.814975946463968e-07, "loss": 0.7376, "step": 12040 }, { "epoch": 0.86, "grad_norm": 2.330358220109559, "learning_rate": 4.81000481332164e-07, "loss": 0.1079, "step": 12041 }, { "epoch": 0.86, "grad_norm": 14.368470651135942, "learning_rate": 4.805036118008499e-07, "loss": 0.5581, "step": 12042 }, { "epoch": 0.86, "grad_norm": 3.2321956815311466, "learning_rate": 4.800069860792594e-07, "loss": 0.18, "step": 12043 }, { "epoch": 0.86, "grad_norm": 3.4143129725422643, "learning_rate": 4.795106041941816e-07, "loss": 0.1909, "step": 12044 }, { "epoch": 0.86, "grad_norm": 2.6570761282946838, "learning_rate": 4.790144661723967e-07, "loss": 0.1014, "step": 12045 }, { "epoch": 0.86, "grad_norm": 3.757142232551042, "learning_rate": 4.785185720406683e-07, "loss": 0.1666, "step": 12046 }, { "epoch": 0.86, "grad_norm": 9.404679120843415, "learning_rate": 4.780229218257481e-07, "loss": 0.4605, "step": 12047 }, { "epoch": 0.86, "grad_norm": 11.557742739370976, "learning_rate": 4.775275155543752e-07, "loss": 0.2865, "step": 12048 }, { "epoch": 0.86, "grad_norm": 19.415904221886443, "learning_rate": 4.770323532532728e-07, "loss": 1.3641, "step": 12049 }, { "epoch": 0.86, "grad_norm": 8.831574235717811, "learning_rate": 4.765374349491558e-07, "loss": 0.2433, "step": 12050 }, { "epoch": 0.86, "grad_norm": 10.810587055140143, "learning_rate": 4.760427606687218e-07, "loss": 0.8235, "step": 12051 }, { "epoch": 0.86, "grad_norm": 5.695057910348585, "learning_rate": 4.755483304386571e-07, "loss": 0.1549, "step": 12052 }, { "epoch": 0.86, "grad_norm": 2.387023428847431, "learning_rate": 4.750541442856327e-07, "loss": 0.0886, "step": 12053 }, { "epoch": 0.86, "grad_norm": 9.408339214094507, "learning_rate": 4.745602022363105e-07, "loss": 0.6253, "step": 12054 }, { "epoch": 0.86, "grad_norm": 3.9457860321478986, "learning_rate": 4.740665043173359e-07, "loss": 0.7159, "step": 12055 }, { "epoch": 0.86, "grad_norm": 4.321935644982515, "learning_rate": 4.735730505553404e-07, "loss": 0.7697, "step": 12056 }, { "epoch": 0.86, "grad_norm": 5.039028267997543, "learning_rate": 4.7307984097694657e-07, "loss": 1.0173, "step": 12057 }, { "epoch": 0.86, "grad_norm": 2.9974158297203077, "learning_rate": 4.7258687560875984e-07, "loss": 0.1563, "step": 12058 }, { "epoch": 0.86, "grad_norm": 3.057160699610223, "learning_rate": 4.720941544773744e-07, "loss": 0.1331, "step": 12059 }, { "epoch": 0.86, "grad_norm": 16.384174566791454, "learning_rate": 4.716016776093696e-07, "loss": 0.5321, "step": 12060 }, { "epoch": 0.86, "grad_norm": 17.562484241046327, "learning_rate": 4.7110944503131363e-07, "loss": 0.6515, "step": 12061 }, { "epoch": 0.86, "grad_norm": 3.08409240573469, "learning_rate": 4.7061745676975966e-07, "loss": 0.1335, "step": 12062 }, { "epoch": 0.87, "grad_norm": 3.5057269596546865, "learning_rate": 4.7012571285124985e-07, "loss": 0.182, "step": 12063 }, { "epoch": 0.87, "grad_norm": 4.255758750221587, "learning_rate": 4.6963421330231187e-07, "loss": 0.2232, "step": 12064 }, { "epoch": 0.87, "grad_norm": 4.016205374635897, "learning_rate": 4.6914295814945834e-07, "loss": 0.2198, "step": 12065 }, { "epoch": 0.87, "grad_norm": 2.9417354747617557, "learning_rate": 4.686519474191925e-07, "loss": 0.134, "step": 12066 }, { "epoch": 0.87, "grad_norm": 3.4917262413901042, "learning_rate": 4.681611811380027e-07, "loss": 0.1528, "step": 12067 }, { "epoch": 0.87, "grad_norm": 9.586979604209196, "learning_rate": 4.676706593323621e-07, "loss": 0.4888, "step": 12068 }, { "epoch": 0.87, "grad_norm": 2.6280595380392113, "learning_rate": 4.6718038202873463e-07, "loss": 0.1164, "step": 12069 }, { "epoch": 0.87, "grad_norm": 3.9818044499382115, "learning_rate": 4.6669034925356794e-07, "loss": 0.1569, "step": 12070 }, { "epoch": 0.87, "grad_norm": 4.119406667412748, "learning_rate": 4.66200561033297e-07, "loss": 0.929, "step": 12071 }, { "epoch": 0.87, "grad_norm": 3.3092333305226664, "learning_rate": 4.6571101739434465e-07, "loss": 0.1059, "step": 12072 }, { "epoch": 0.87, "grad_norm": 13.424241448599197, "learning_rate": 4.652217183631197e-07, "loss": 0.5389, "step": 12073 }, { "epoch": 0.87, "grad_norm": 15.382528924748684, "learning_rate": 4.647326639660171e-07, "loss": 0.3707, "step": 12074 }, { "epoch": 0.87, "grad_norm": 4.171634685136099, "learning_rate": 4.642438542294214e-07, "loss": 0.1535, "step": 12075 }, { "epoch": 0.87, "grad_norm": 26.05293980253564, "learning_rate": 4.6375528917970093e-07, "loss": 1.2243, "step": 12076 }, { "epoch": 0.87, "grad_norm": 7.605068285443158, "learning_rate": 4.632669688432112e-07, "loss": 0.414, "step": 12077 }, { "epoch": 0.87, "grad_norm": 5.350637244496776, "learning_rate": 4.6277889324629677e-07, "loss": 0.7859, "step": 12078 }, { "epoch": 0.87, "grad_norm": 2.1353326444127716, "learning_rate": 4.622910624152865e-07, "loss": 0.1394, "step": 12079 }, { "epoch": 0.87, "grad_norm": 10.956544999368148, "learning_rate": 4.618034763764978e-07, "loss": 0.8531, "step": 12080 }, { "epoch": 0.87, "grad_norm": 17.62878116180258, "learning_rate": 4.6131613515623233e-07, "loss": 1.4844, "step": 12081 }, { "epoch": 0.87, "grad_norm": 3.125359378070105, "learning_rate": 4.6082903878078235e-07, "loss": 0.0695, "step": 12082 }, { "epoch": 0.87, "grad_norm": 3.627510678899019, "learning_rate": 4.603421872764241e-07, "loss": 0.1615, "step": 12083 }, { "epoch": 0.87, "grad_norm": 4.074281778679872, "learning_rate": 4.5985558066942104e-07, "loss": 0.1205, "step": 12084 }, { "epoch": 0.87, "grad_norm": 10.781303239989075, "learning_rate": 4.5936921898602437e-07, "loss": 0.6461, "step": 12085 }, { "epoch": 0.87, "grad_norm": 3.1260585275988673, "learning_rate": 4.5888310225247037e-07, "loss": 0.1301, "step": 12086 }, { "epoch": 0.87, "grad_norm": 3.981637787307232, "learning_rate": 4.5839723049498465e-07, "loss": 0.1519, "step": 12087 }, { "epoch": 0.87, "grad_norm": 4.52275475159596, "learning_rate": 4.579116037397774e-07, "loss": 0.8806, "step": 12088 }, { "epoch": 0.87, "grad_norm": 4.272271876994329, "learning_rate": 4.57426222013046e-07, "loss": 0.2107, "step": 12089 }, { "epoch": 0.87, "grad_norm": 11.626575639196478, "learning_rate": 4.5694108534097505e-07, "loss": 0.3263, "step": 12090 }, { "epoch": 0.87, "grad_norm": 3.6348595823447316, "learning_rate": 4.564561937497369e-07, "loss": 0.2115, "step": 12091 }, { "epoch": 0.87, "grad_norm": 2.4833220399634577, "learning_rate": 4.5597154726548853e-07, "loss": 0.0936, "step": 12092 }, { "epoch": 0.87, "grad_norm": 2.814204172728082, "learning_rate": 4.554871459143745e-07, "loss": 0.1739, "step": 12093 }, { "epoch": 0.87, "grad_norm": 13.98327344958925, "learning_rate": 4.5500298972252776e-07, "loss": 0.4021, "step": 12094 }, { "epoch": 0.87, "grad_norm": 10.101229434411037, "learning_rate": 4.545190787160675e-07, "loss": 0.4201, "step": 12095 }, { "epoch": 0.87, "grad_norm": 3.872541075722584, "learning_rate": 4.54035412921095e-07, "loss": 0.8013, "step": 12096 }, { "epoch": 0.87, "grad_norm": 4.416860467665359, "learning_rate": 4.535519923637055e-07, "loss": 0.6789, "step": 12097 }, { "epoch": 0.87, "grad_norm": 4.111107668451683, "learning_rate": 4.5306881706997653e-07, "loss": 0.7745, "step": 12098 }, { "epoch": 0.87, "grad_norm": 17.930393479010156, "learning_rate": 4.525858870659733e-07, "loss": 0.4138, "step": 12099 }, { "epoch": 0.87, "grad_norm": 3.444459325901743, "learning_rate": 4.5210320237774943e-07, "loss": 0.2527, "step": 12100 }, { "epoch": 0.87, "grad_norm": 2.980149297784189, "learning_rate": 4.5162076303134295e-07, "loss": 0.1591, "step": 12101 }, { "epoch": 0.87, "grad_norm": 2.523896191318291, "learning_rate": 4.511385690527786e-07, "loss": 0.0875, "step": 12102 }, { "epoch": 0.87, "grad_norm": 14.814967504559437, "learning_rate": 4.5065662046807126e-07, "loss": 1.075, "step": 12103 }, { "epoch": 0.87, "grad_norm": 6.407917356900683, "learning_rate": 4.501749173032194e-07, "loss": 0.3515, "step": 12104 }, { "epoch": 0.87, "grad_norm": 10.629681853196663, "learning_rate": 4.496934595842073e-07, "loss": 0.7375, "step": 12105 }, { "epoch": 0.87, "grad_norm": 4.723454165905812, "learning_rate": 4.492122473370103e-07, "loss": 0.7779, "step": 12106 }, { "epoch": 0.87, "grad_norm": 2.4467670826941226, "learning_rate": 4.4873128058758766e-07, "loss": 0.0999, "step": 12107 }, { "epoch": 0.87, "grad_norm": 3.258132025587258, "learning_rate": 4.482505593618836e-07, "loss": 0.1337, "step": 12108 }, { "epoch": 0.87, "grad_norm": 4.29343922513512, "learning_rate": 4.4777008368583286e-07, "loss": 0.8476, "step": 12109 }, { "epoch": 0.87, "grad_norm": 2.8449988499889476, "learning_rate": 4.472898535853554e-07, "loss": 0.1062, "step": 12110 }, { "epoch": 0.87, "grad_norm": 4.729287546603903, "learning_rate": 4.468098690863565e-07, "loss": 0.9748, "step": 12111 }, { "epoch": 0.87, "grad_norm": 3.911744868795029, "learning_rate": 4.4633013021473105e-07, "loss": 0.1837, "step": 12112 }, { "epoch": 0.87, "grad_norm": 3.371914733404015, "learning_rate": 4.458506369963583e-07, "loss": 0.1592, "step": 12113 }, { "epoch": 0.87, "grad_norm": 4.982659443449195, "learning_rate": 4.453713894571049e-07, "loss": 1.0126, "step": 12114 }, { "epoch": 0.87, "grad_norm": 4.639444235955144, "learning_rate": 4.448923876228256e-07, "loss": 0.7365, "step": 12115 }, { "epoch": 0.87, "grad_norm": 2.5863493682719203, "learning_rate": 4.4441363151935977e-07, "loss": 0.13, "step": 12116 }, { "epoch": 0.87, "grad_norm": 3.5018409433415196, "learning_rate": 4.4393512117253455e-07, "loss": 0.2736, "step": 12117 }, { "epoch": 0.87, "grad_norm": 4.049711968712372, "learning_rate": 4.434568566081632e-07, "loss": 0.2118, "step": 12118 }, { "epoch": 0.87, "grad_norm": 17.9708761244436, "learning_rate": 4.429788378520489e-07, "loss": 0.3785, "step": 12119 }, { "epoch": 0.87, "grad_norm": 25.849930796196364, "learning_rate": 4.4250106492997547e-07, "loss": 0.68, "step": 12120 }, { "epoch": 0.87, "grad_norm": 3.6376939905279957, "learning_rate": 4.4202353786771903e-07, "loss": 0.1313, "step": 12121 }, { "epoch": 0.87, "grad_norm": 2.871591319200956, "learning_rate": 4.4154625669104003e-07, "loss": 0.1224, "step": 12122 }, { "epoch": 0.87, "grad_norm": 14.332046480583777, "learning_rate": 4.4106922142568567e-07, "loss": 0.7469, "step": 12123 }, { "epoch": 0.87, "grad_norm": 3.504117919306241, "learning_rate": 4.4059243209738924e-07, "loss": 0.1812, "step": 12124 }, { "epoch": 0.87, "grad_norm": 2.8239557092498395, "learning_rate": 4.4011588873187407e-07, "loss": 0.0939, "step": 12125 }, { "epoch": 0.87, "grad_norm": 5.1648847309168024, "learning_rate": 4.396395913548462e-07, "loss": 0.7746, "step": 12126 }, { "epoch": 0.87, "grad_norm": 3.5428573825791227, "learning_rate": 4.3916353999199967e-07, "loss": 0.17, "step": 12127 }, { "epoch": 0.87, "grad_norm": 11.350803672601767, "learning_rate": 4.386877346690177e-07, "loss": 0.3575, "step": 12128 }, { "epoch": 0.87, "grad_norm": 3.9991851616876977, "learning_rate": 4.38212175411567e-07, "loss": 0.2472, "step": 12129 }, { "epoch": 0.87, "grad_norm": 2.7551886051465964, "learning_rate": 4.37736862245301e-07, "loss": 0.1925, "step": 12130 }, { "epoch": 0.87, "grad_norm": 8.946283833285811, "learning_rate": 4.3726179519586354e-07, "loss": 0.4753, "step": 12131 }, { "epoch": 0.87, "grad_norm": 2.969213865487328, "learning_rate": 4.367869742888808e-07, "loss": 0.1297, "step": 12132 }, { "epoch": 0.87, "grad_norm": 16.986783948856747, "learning_rate": 4.363123995499674e-07, "loss": 0.813, "step": 12133 }, { "epoch": 0.87, "grad_norm": 3.108454675371158, "learning_rate": 4.358380710047261e-07, "loss": 0.1416, "step": 12134 }, { "epoch": 0.87, "grad_norm": 5.063611716769467, "learning_rate": 4.353639886787447e-07, "loss": 0.9882, "step": 12135 }, { "epoch": 0.87, "grad_norm": 2.910041933635842, "learning_rate": 4.348901525975974e-07, "loss": 0.095, "step": 12136 }, { "epoch": 0.87, "grad_norm": 3.648638438449448, "learning_rate": 4.344165627868474e-07, "loss": 0.1996, "step": 12137 }, { "epoch": 0.87, "grad_norm": 11.721077519187068, "learning_rate": 4.3394321927204163e-07, "loss": 0.7282, "step": 12138 }, { "epoch": 0.87, "grad_norm": 4.014822761937559, "learning_rate": 4.334701220787152e-07, "loss": 0.1702, "step": 12139 }, { "epoch": 0.87, "grad_norm": 33.80891117059956, "learning_rate": 4.3299727123239156e-07, "loss": 0.9942, "step": 12140 }, { "epoch": 0.87, "grad_norm": 2.7298369019788757, "learning_rate": 4.325246667585775e-07, "loss": 0.092, "step": 12141 }, { "epoch": 0.87, "grad_norm": 5.077500288887929, "learning_rate": 4.320523086827683e-07, "loss": 0.7831, "step": 12142 }, { "epoch": 0.87, "grad_norm": 4.803754480936804, "learning_rate": 4.315801970304484e-07, "loss": 0.7975, "step": 12143 }, { "epoch": 0.87, "grad_norm": 4.630336929565878, "learning_rate": 4.3110833182708313e-07, "loss": 0.8829, "step": 12144 }, { "epoch": 0.87, "grad_norm": 4.081680135753121, "learning_rate": 4.306367130981287e-07, "loss": 0.2088, "step": 12145 }, { "epoch": 0.87, "grad_norm": 16.126284271578868, "learning_rate": 4.3016534086902815e-07, "loss": 0.5277, "step": 12146 }, { "epoch": 0.87, "grad_norm": 3.307201580247972, "learning_rate": 4.2969421516520946e-07, "loss": 0.2284, "step": 12147 }, { "epoch": 0.87, "grad_norm": 4.377576555866037, "learning_rate": 4.2922333601208785e-07, "loss": 0.7984, "step": 12148 }, { "epoch": 0.87, "grad_norm": 5.6531933496383395, "learning_rate": 4.287527034350669e-07, "loss": 1.0077, "step": 12149 }, { "epoch": 0.87, "grad_norm": 20.008748359349802, "learning_rate": 4.2828231745953407e-07, "loss": 0.8089, "step": 12150 }, { "epoch": 0.87, "grad_norm": 12.792398567526394, "learning_rate": 4.2781217811086574e-07, "loss": 0.6424, "step": 12151 }, { "epoch": 0.87, "grad_norm": 4.521937958428288, "learning_rate": 4.2734228541442267e-07, "loss": 0.7843, "step": 12152 }, { "epoch": 0.87, "grad_norm": 3.586924463768963, "learning_rate": 4.268726393955558e-07, "loss": 0.1061, "step": 12153 }, { "epoch": 0.87, "grad_norm": 3.5549651345601334, "learning_rate": 4.2640324007959977e-07, "loss": 0.1286, "step": 12154 }, { "epoch": 0.87, "grad_norm": 4.833057599304132, "learning_rate": 4.259340874918771e-07, "loss": 0.7058, "step": 12155 }, { "epoch": 0.87, "grad_norm": 18.56274909082471, "learning_rate": 4.2546518165769657e-07, "loss": 0.6791, "step": 12156 }, { "epoch": 0.87, "grad_norm": 4.2417076446468664, "learning_rate": 4.249965226023528e-07, "loss": 0.1707, "step": 12157 }, { "epoch": 0.87, "grad_norm": 4.45407845454494, "learning_rate": 4.245281103511306e-07, "loss": 0.191, "step": 12158 }, { "epoch": 0.87, "grad_norm": 12.38236293913637, "learning_rate": 4.24059944929297e-07, "loss": 0.5583, "step": 12159 }, { "epoch": 0.87, "grad_norm": 2.5326379070526417, "learning_rate": 4.2359202636210906e-07, "loss": 0.0858, "step": 12160 }, { "epoch": 0.87, "grad_norm": 20.171696487083008, "learning_rate": 4.231243546748076e-07, "loss": 0.4474, "step": 12161 }, { "epoch": 0.87, "grad_norm": 2.656706455276512, "learning_rate": 4.226569298926242e-07, "loss": 0.1062, "step": 12162 }, { "epoch": 0.87, "grad_norm": 4.672923620586587, "learning_rate": 4.221897520407725e-07, "loss": 0.7202, "step": 12163 }, { "epoch": 0.87, "grad_norm": 9.956382014310602, "learning_rate": 4.217228211444552e-07, "loss": 0.9912, "step": 12164 }, { "epoch": 0.87, "grad_norm": 16.680092834454804, "learning_rate": 4.212561372288632e-07, "loss": 0.6931, "step": 12165 }, { "epoch": 0.87, "grad_norm": 2.0774152727783033, "learning_rate": 4.207897003191702e-07, "loss": 0.0561, "step": 12166 }, { "epoch": 0.87, "grad_norm": 4.631147201525199, "learning_rate": 4.2032351044054064e-07, "loss": 0.7856, "step": 12167 }, { "epoch": 0.87, "grad_norm": 4.932336431056388, "learning_rate": 4.1985756761812215e-07, "loss": 0.8696, "step": 12168 }, { "epoch": 0.87, "grad_norm": 4.2659707191860425, "learning_rate": 4.193918718770512e-07, "loss": 0.8838, "step": 12169 }, { "epoch": 0.87, "grad_norm": 14.128965007863165, "learning_rate": 4.1892642324244894e-07, "loss": 0.4248, "step": 12170 }, { "epoch": 0.87, "grad_norm": 4.537695764248221, "learning_rate": 4.184612217394274e-07, "loss": 0.9869, "step": 12171 }, { "epoch": 0.87, "grad_norm": 2.8975682748772313, "learning_rate": 4.1799626739308043e-07, "loss": 0.1108, "step": 12172 }, { "epoch": 0.87, "grad_norm": 3.3210431614810463, "learning_rate": 4.1753156022849016e-07, "loss": 0.1981, "step": 12173 }, { "epoch": 0.87, "grad_norm": 21.745938307606853, "learning_rate": 4.170671002707277e-07, "loss": 0.6233, "step": 12174 }, { "epoch": 0.87, "grad_norm": 4.900145434724561, "learning_rate": 4.1660288754484743e-07, "loss": 0.8406, "step": 12175 }, { "epoch": 0.87, "grad_norm": 4.27734385691894, "learning_rate": 4.161389220758916e-07, "loss": 0.2202, "step": 12176 }, { "epoch": 0.87, "grad_norm": 18.88525938085115, "learning_rate": 4.1567520388889114e-07, "loss": 0.9818, "step": 12177 }, { "epoch": 0.87, "grad_norm": 16.869276163773257, "learning_rate": 4.1521173300886066e-07, "loss": 0.995, "step": 12178 }, { "epoch": 0.87, "grad_norm": 14.101898106215437, "learning_rate": 4.147485094608028e-07, "loss": 0.664, "step": 12179 }, { "epoch": 0.87, "grad_norm": 3.0993628234344506, "learning_rate": 4.1428553326970713e-07, "loss": 0.1523, "step": 12180 }, { "epoch": 0.87, "grad_norm": 3.46636109354735, "learning_rate": 4.138228044605486e-07, "loss": 0.1945, "step": 12181 }, { "epoch": 0.87, "grad_norm": 8.93446957797605, "learning_rate": 4.133603230582894e-07, "loss": 0.3589, "step": 12182 }, { "epoch": 0.87, "grad_norm": 3.3727328991830956, "learning_rate": 4.128980890878803e-07, "loss": 0.1403, "step": 12183 }, { "epoch": 0.87, "grad_norm": 18.381418235455193, "learning_rate": 4.1243610257425626e-07, "loss": 0.6182, "step": 12184 }, { "epoch": 0.87, "grad_norm": 6.332400908575437, "learning_rate": 4.119743635423384e-07, "loss": 0.1829, "step": 12185 }, { "epoch": 0.87, "grad_norm": 3.797685556952819, "learning_rate": 4.115128720170386e-07, "loss": 0.2306, "step": 12186 }, { "epoch": 0.87, "grad_norm": 3.029643451892448, "learning_rate": 4.1105162802325015e-07, "loss": 0.1291, "step": 12187 }, { "epoch": 0.87, "grad_norm": 7.878880908285493, "learning_rate": 4.1059063158585657e-07, "loss": 0.4615, "step": 12188 }, { "epoch": 0.87, "grad_norm": 2.9968795829789103, "learning_rate": 4.1012988272972564e-07, "loss": 0.2263, "step": 12189 }, { "epoch": 0.87, "grad_norm": 4.733996676407047, "learning_rate": 4.096693814797154e-07, "loss": 1.0359, "step": 12190 }, { "epoch": 0.87, "grad_norm": 17.42683522660595, "learning_rate": 4.092091278606658e-07, "loss": 0.2559, "step": 12191 }, { "epoch": 0.87, "grad_norm": 3.4779205230401335, "learning_rate": 4.087491218974071e-07, "loss": 0.2564, "step": 12192 }, { "epoch": 0.87, "grad_norm": 2.3848386132799755, "learning_rate": 4.0828936361475447e-07, "loss": 0.1194, "step": 12193 }, { "epoch": 0.87, "grad_norm": 3.092905143513855, "learning_rate": 4.078298530375091e-07, "loss": 0.2108, "step": 12194 }, { "epoch": 0.87, "grad_norm": 14.708416184560338, "learning_rate": 4.0737059019046176e-07, "loss": 0.8491, "step": 12195 }, { "epoch": 0.87, "grad_norm": 51.73849570429828, "learning_rate": 4.069115750983871e-07, "loss": 0.9918, "step": 12196 }, { "epoch": 0.87, "grad_norm": 2.714709059706247, "learning_rate": 4.064528077860469e-07, "loss": 0.065, "step": 12197 }, { "epoch": 0.87, "grad_norm": 9.091397824811105, "learning_rate": 4.0599428827818975e-07, "loss": 0.5066, "step": 12198 }, { "epoch": 0.87, "grad_norm": 3.2157588973849602, "learning_rate": 4.05536016599552e-07, "loss": 0.1614, "step": 12199 }, { "epoch": 0.87, "grad_norm": 3.2484337084883292, "learning_rate": 4.0507799277485546e-07, "loss": 0.1567, "step": 12200 }, { "epoch": 0.87, "grad_norm": 5.254247430622249, "learning_rate": 4.046202168288077e-07, "loss": 0.1139, "step": 12201 }, { "epoch": 0.87, "grad_norm": 4.130713714519037, "learning_rate": 4.0416268878610555e-07, "loss": 0.2697, "step": 12202 }, { "epoch": 0.88, "grad_norm": 26.559837355946446, "learning_rate": 4.037054086714298e-07, "loss": 0.8163, "step": 12203 }, { "epoch": 0.88, "grad_norm": 17.405545759302566, "learning_rate": 4.032483765094497e-07, "loss": 0.8366, "step": 12204 }, { "epoch": 0.88, "grad_norm": 3.2625819877335704, "learning_rate": 4.027915923248199e-07, "loss": 0.11, "step": 12205 }, { "epoch": 0.88, "grad_norm": 15.784054871848193, "learning_rate": 4.0233505614218237e-07, "loss": 0.624, "step": 12206 }, { "epoch": 0.88, "grad_norm": 10.704500971656492, "learning_rate": 4.0187876798616464e-07, "loss": 0.3406, "step": 12207 }, { "epoch": 0.88, "grad_norm": 19.242693264426364, "learning_rate": 4.0142272788138315e-07, "loss": 0.3278, "step": 12208 }, { "epoch": 0.88, "grad_norm": 2.6219246546326875, "learning_rate": 4.0096693585243926e-07, "loss": 0.1516, "step": 12209 }, { "epoch": 0.88, "grad_norm": 18.062569801916258, "learning_rate": 4.005113919239195e-07, "loss": 0.5096, "step": 12210 }, { "epoch": 0.88, "grad_norm": 3.0910651853909124, "learning_rate": 4.000560961204014e-07, "loss": 0.1013, "step": 12211 }, { "epoch": 0.88, "grad_norm": 2.305555383931108, "learning_rate": 3.996010484664453e-07, "loss": 0.1264, "step": 12212 }, { "epoch": 0.88, "grad_norm": 13.704849610260288, "learning_rate": 3.9914624898659827e-07, "loss": 0.6359, "step": 12213 }, { "epoch": 0.88, "grad_norm": 3.371961230738556, "learning_rate": 3.9869169770539675e-07, "loss": 0.0882, "step": 12214 }, { "epoch": 0.88, "grad_norm": 4.230135859884381, "learning_rate": 3.982373946473611e-07, "loss": 0.1738, "step": 12215 }, { "epoch": 0.88, "grad_norm": 14.977648207551697, "learning_rate": 3.9778333983699957e-07, "loss": 0.6758, "step": 12216 }, { "epoch": 0.88, "grad_norm": 6.832789469952064, "learning_rate": 3.973295332988064e-07, "loss": 0.2699, "step": 12217 }, { "epoch": 0.88, "grad_norm": 6.546409899587706, "learning_rate": 3.968759750572626e-07, "loss": 0.3295, "step": 12218 }, { "epoch": 0.88, "grad_norm": 11.813166117829883, "learning_rate": 3.964226651368358e-07, "loss": 0.3916, "step": 12219 }, { "epoch": 0.88, "grad_norm": 2.0526916983872834, "learning_rate": 3.9596960356198143e-07, "loss": 0.0861, "step": 12220 }, { "epoch": 0.88, "grad_norm": 2.4031409089991635, "learning_rate": 3.955167903571394e-07, "loss": 0.0925, "step": 12221 }, { "epoch": 0.88, "grad_norm": 2.631236142507359, "learning_rate": 3.950642255467374e-07, "loss": 0.1062, "step": 12222 }, { "epoch": 0.88, "grad_norm": 2.919923444301165, "learning_rate": 3.946119091551903e-07, "loss": 0.1872, "step": 12223 }, { "epoch": 0.88, "grad_norm": 19.17836172288559, "learning_rate": 3.941598412068981e-07, "loss": 0.6243, "step": 12224 }, { "epoch": 0.88, "grad_norm": 3.435258599792539, "learning_rate": 3.93708021726249e-07, "loss": 0.1714, "step": 12225 }, { "epoch": 0.88, "grad_norm": 2.5473413212330507, "learning_rate": 3.932564507376152e-07, "loss": 0.0654, "step": 12226 }, { "epoch": 0.88, "grad_norm": 26.745075501470623, "learning_rate": 3.928051282653594e-07, "loss": 0.5302, "step": 12227 }, { "epoch": 0.88, "grad_norm": 3.5499026992753904, "learning_rate": 3.923540543338272e-07, "loss": 0.1727, "step": 12228 }, { "epoch": 0.88, "grad_norm": 8.54827754220235, "learning_rate": 3.9190322896735357e-07, "loss": 0.4207, "step": 12229 }, { "epoch": 0.88, "grad_norm": 2.0983485699586284, "learning_rate": 3.9145265219025785e-07, "loss": 0.095, "step": 12230 }, { "epoch": 0.88, "grad_norm": 41.72712271760743, "learning_rate": 3.9100232402684624e-07, "loss": 0.3738, "step": 12231 }, { "epoch": 0.88, "grad_norm": 23.783126556621184, "learning_rate": 3.905522445014143e-07, "loss": 0.6307, "step": 12232 }, { "epoch": 0.88, "grad_norm": 4.540473016507622, "learning_rate": 3.9010241363824144e-07, "loss": 0.2098, "step": 12233 }, { "epoch": 0.88, "grad_norm": 17.362004864127755, "learning_rate": 3.8965283146159325e-07, "loss": 1.4256, "step": 12234 }, { "epoch": 0.88, "grad_norm": 5.249498317535852, "learning_rate": 3.892034979957232e-07, "loss": 0.7895, "step": 12235 }, { "epoch": 0.88, "grad_norm": 3.682103138438281, "learning_rate": 3.887544132648724e-07, "loss": 0.147, "step": 12236 }, { "epoch": 0.88, "grad_norm": 3.7095426171536503, "learning_rate": 3.883055772932665e-07, "loss": 0.1323, "step": 12237 }, { "epoch": 0.88, "grad_norm": 31.851712209602955, "learning_rate": 3.8785699010511766e-07, "loss": 1.0138, "step": 12238 }, { "epoch": 0.88, "grad_norm": 10.133955057892928, "learning_rate": 3.874086517246273e-07, "loss": 0.544, "step": 12239 }, { "epoch": 0.88, "grad_norm": 7.032394923004306, "learning_rate": 3.869605621759809e-07, "loss": 0.1942, "step": 12240 }, { "epoch": 0.88, "grad_norm": 3.8235079247545913, "learning_rate": 3.865127214833497e-07, "loss": 0.8289, "step": 12241 }, { "epoch": 0.88, "grad_norm": 29.209234589487803, "learning_rate": 3.8606512967089507e-07, "loss": 0.5105, "step": 12242 }, { "epoch": 0.88, "grad_norm": 4.674467029154379, "learning_rate": 3.856177867627614e-07, "loss": 0.2089, "step": 12243 }, { "epoch": 0.88, "grad_norm": 4.601106534788779, "learning_rate": 3.851706927830817e-07, "loss": 0.843, "step": 12244 }, { "epoch": 0.88, "grad_norm": 3.36642237578521, "learning_rate": 3.8472384775597505e-07, "loss": 0.1523, "step": 12245 }, { "epoch": 0.88, "grad_norm": 3.7757553354558873, "learning_rate": 3.842772517055476e-07, "loss": 0.146, "step": 12246 }, { "epoch": 0.88, "grad_norm": 4.969699504508349, "learning_rate": 3.8383090465589024e-07, "loss": 0.1562, "step": 12247 }, { "epoch": 0.88, "grad_norm": 3.324970978237964, "learning_rate": 3.83384806631083e-07, "loss": 0.0933, "step": 12248 }, { "epoch": 0.88, "grad_norm": 17.9329025342752, "learning_rate": 3.829389576551906e-07, "loss": 0.6055, "step": 12249 }, { "epoch": 0.88, "grad_norm": 18.55441941968895, "learning_rate": 3.824933577522638e-07, "loss": 0.8142, "step": 12250 }, { "epoch": 0.88, "grad_norm": 3.3471786735412894, "learning_rate": 3.8204800694634335e-07, "loss": 0.1268, "step": 12251 }, { "epoch": 0.88, "grad_norm": 17.224551913386815, "learning_rate": 3.8160290526145396e-07, "loss": 0.7364, "step": 12252 }, { "epoch": 0.88, "grad_norm": 3.2444658973913554, "learning_rate": 3.8115805272160413e-07, "loss": 0.1339, "step": 12253 }, { "epoch": 0.88, "grad_norm": 12.616983914599077, "learning_rate": 3.8071344935079524e-07, "loss": 0.5485, "step": 12254 }, { "epoch": 0.88, "grad_norm": 14.796296697387504, "learning_rate": 3.8026909517301093e-07, "loss": 0.5971, "step": 12255 }, { "epoch": 0.88, "grad_norm": 13.761600265089395, "learning_rate": 3.798249902122214e-07, "loss": 0.5923, "step": 12256 }, { "epoch": 0.88, "grad_norm": 4.346462932466726, "learning_rate": 3.793811344923859e-07, "loss": 0.8807, "step": 12257 }, { "epoch": 0.88, "grad_norm": 4.324128310566319, "learning_rate": 3.7893752803744855e-07, "loss": 0.7367, "step": 12258 }, { "epoch": 0.88, "grad_norm": 4.416823810676487, "learning_rate": 3.784941708713391e-07, "loss": 0.2835, "step": 12259 }, { "epoch": 0.88, "grad_norm": 3.4313425779113524, "learning_rate": 3.7805106301797623e-07, "loss": 0.1154, "step": 12260 }, { "epoch": 0.88, "grad_norm": 4.566780103280946, "learning_rate": 3.7760820450126356e-07, "loss": 0.7474, "step": 12261 }, { "epoch": 0.88, "grad_norm": 15.26335408072407, "learning_rate": 3.7716559534509143e-07, "loss": 0.8082, "step": 12262 }, { "epoch": 0.88, "grad_norm": 16.39742414344863, "learning_rate": 3.767232355733369e-07, "loss": 0.6381, "step": 12263 }, { "epoch": 0.88, "grad_norm": 11.415947124170911, "learning_rate": 3.7628112520986473e-07, "loss": 0.4769, "step": 12264 }, { "epoch": 0.88, "grad_norm": 13.469180795989523, "learning_rate": 3.7583926427852247e-07, "loss": 0.7131, "step": 12265 }, { "epoch": 0.88, "grad_norm": 5.527364781135173, "learning_rate": 3.7539765280315e-07, "loss": 0.1489, "step": 12266 }, { "epoch": 0.88, "grad_norm": 6.988150104543242, "learning_rate": 3.7495629080756826e-07, "loss": 0.1451, "step": 12267 }, { "epoch": 0.88, "grad_norm": 3.4926187223849627, "learning_rate": 3.745151783155887e-07, "loss": 0.2082, "step": 12268 }, { "epoch": 0.88, "grad_norm": 5.860637095260375, "learning_rate": 3.740743153510057e-07, "loss": 0.3062, "step": 12269 }, { "epoch": 0.88, "grad_norm": 17.42757382936848, "learning_rate": 3.736337019376035e-07, "loss": 0.4601, "step": 12270 }, { "epoch": 0.88, "grad_norm": 19.38773588844337, "learning_rate": 3.7319333809915204e-07, "loss": 0.7246, "step": 12271 }, { "epoch": 0.88, "grad_norm": 3.2231616701013786, "learning_rate": 3.727532238594056e-07, "loss": 0.1506, "step": 12272 }, { "epoch": 0.88, "grad_norm": 44.373390908482456, "learning_rate": 3.723133592421085e-07, "loss": 1.2684, "step": 12273 }, { "epoch": 0.88, "grad_norm": 5.038796241818373, "learning_rate": 3.718737442709891e-07, "loss": 0.8032, "step": 12274 }, { "epoch": 0.88, "grad_norm": 5.126125598360395, "learning_rate": 3.7143437896976165e-07, "loss": 0.1475, "step": 12275 }, { "epoch": 0.88, "grad_norm": 11.733148507622003, "learning_rate": 3.709952633621311e-07, "loss": 0.6932, "step": 12276 }, { "epoch": 0.88, "grad_norm": 3.420819520693698, "learning_rate": 3.705563974717835e-07, "loss": 0.1234, "step": 12277 }, { "epoch": 0.88, "grad_norm": 15.019470656303334, "learning_rate": 3.701177813223944e-07, "loss": 0.9093, "step": 12278 }, { "epoch": 0.88, "grad_norm": 3.760160237477991, "learning_rate": 3.696794149376265e-07, "loss": 0.2516, "step": 12279 }, { "epoch": 0.88, "grad_norm": 3.6057416706331686, "learning_rate": 3.69241298341127e-07, "loss": 0.1456, "step": 12280 }, { "epoch": 0.88, "grad_norm": 3.961707272541501, "learning_rate": 3.6880343155653085e-07, "loss": 0.2096, "step": 12281 }, { "epoch": 0.88, "grad_norm": 3.0518047712277654, "learning_rate": 3.6836581460746034e-07, "loss": 0.1367, "step": 12282 }, { "epoch": 0.88, "grad_norm": 2.4635019100426785, "learning_rate": 3.679284475175221e-07, "loss": 0.0605, "step": 12283 }, { "epoch": 0.88, "grad_norm": 7.013495643265519, "learning_rate": 3.6749133031031e-07, "loss": 0.4225, "step": 12284 }, { "epoch": 0.88, "grad_norm": 2.627104201383844, "learning_rate": 3.67054463009407e-07, "loss": 0.1702, "step": 12285 }, { "epoch": 0.88, "grad_norm": 1.9277851356220945, "learning_rate": 3.6661784563837845e-07, "loss": 0.0797, "step": 12286 }, { "epoch": 0.88, "grad_norm": 10.687171696092234, "learning_rate": 3.661814782207784e-07, "loss": 0.2556, "step": 12287 }, { "epoch": 0.88, "grad_norm": 1.6994269048513997, "learning_rate": 3.657453607801492e-07, "loss": 0.0608, "step": 12288 }, { "epoch": 0.88, "grad_norm": 15.076016827109454, "learning_rate": 3.653094933400153e-07, "loss": 0.9156, "step": 12289 }, { "epoch": 0.88, "grad_norm": 2.636810845000528, "learning_rate": 3.648738759238901e-07, "loss": 0.0947, "step": 12290 }, { "epoch": 0.88, "grad_norm": 17.021012989460683, "learning_rate": 3.6443850855527487e-07, "loss": 0.6919, "step": 12291 }, { "epoch": 0.88, "grad_norm": 3.0335121217269467, "learning_rate": 3.6400339125765625e-07, "loss": 0.1092, "step": 12292 }, { "epoch": 0.88, "grad_norm": 12.16913444734814, "learning_rate": 3.6356852405450505e-07, "loss": 0.7344, "step": 12293 }, { "epoch": 0.88, "grad_norm": 3.123047382720992, "learning_rate": 3.6313390696928355e-07, "loss": 0.1495, "step": 12294 }, { "epoch": 0.88, "grad_norm": 2.0376834159558275, "learning_rate": 3.626995400254357e-07, "loss": 0.0921, "step": 12295 }, { "epoch": 0.88, "grad_norm": 4.091114703266801, "learning_rate": 3.6226542324639457e-07, "loss": 0.7399, "step": 12296 }, { "epoch": 0.88, "grad_norm": 3.9945045108006862, "learning_rate": 3.6183155665557855e-07, "loss": 0.6654, "step": 12297 }, { "epoch": 0.88, "grad_norm": 4.464367104928807, "learning_rate": 3.613979402763945e-07, "loss": 0.749, "step": 12298 }, { "epoch": 0.88, "grad_norm": 3.0490133994464004, "learning_rate": 3.6096457413223317e-07, "loss": 0.1731, "step": 12299 }, { "epoch": 0.88, "grad_norm": 2.9547062854259902, "learning_rate": 3.605314582464736e-07, "loss": 0.0924, "step": 12300 }, { "epoch": 0.88, "grad_norm": 16.296101722358365, "learning_rate": 3.600985926424805e-07, "loss": 0.2616, "step": 12301 }, { "epoch": 0.88, "grad_norm": 3.563765874871481, "learning_rate": 3.5966597734360454e-07, "loss": 0.2121, "step": 12302 }, { "epoch": 0.88, "grad_norm": 3.29482947850213, "learning_rate": 3.5923361237318545e-07, "loss": 0.1222, "step": 12303 }, { "epoch": 0.88, "grad_norm": 4.337456466250533, "learning_rate": 3.5880149775454673e-07, "loss": 0.8028, "step": 12304 }, { "epoch": 0.88, "grad_norm": 3.2386462428385014, "learning_rate": 3.583696335109987e-07, "loss": 0.0971, "step": 12305 }, { "epoch": 0.88, "grad_norm": 12.084275522405626, "learning_rate": 3.579380196658394e-07, "loss": 0.6255, "step": 12306 }, { "epoch": 0.88, "grad_norm": 3.7323217179239094, "learning_rate": 3.5750665624235293e-07, "loss": 0.195, "step": 12307 }, { "epoch": 0.88, "grad_norm": 21.264035318424863, "learning_rate": 3.570755432638101e-07, "loss": 1.0991, "step": 12308 }, { "epoch": 0.88, "grad_norm": 5.020596837846739, "learning_rate": 3.566446807534663e-07, "loss": 0.1833, "step": 12309 }, { "epoch": 0.88, "grad_norm": 3.572666023329617, "learning_rate": 3.5621406873456664e-07, "loss": 0.1327, "step": 12310 }, { "epoch": 0.88, "grad_norm": 4.791677784309104, "learning_rate": 3.5578370723033994e-07, "loss": 0.1021, "step": 12311 }, { "epoch": 0.88, "grad_norm": 4.222837014748731, "learning_rate": 3.553535962640037e-07, "loss": 0.2268, "step": 12312 }, { "epoch": 0.88, "grad_norm": 3.4482649494859303, "learning_rate": 3.5492373585875927e-07, "loss": 0.0693, "step": 12313 }, { "epoch": 0.88, "grad_norm": 3.951361733960289, "learning_rate": 3.544941260377971e-07, "loss": 0.2295, "step": 12314 }, { "epoch": 0.88, "grad_norm": 3.9968056141852197, "learning_rate": 3.540647668242919e-07, "loss": 0.6975, "step": 12315 }, { "epoch": 0.88, "grad_norm": 20.62112847913219, "learning_rate": 3.5363565824140747e-07, "loss": 1.0634, "step": 12316 }, { "epoch": 0.88, "grad_norm": 11.352665348543518, "learning_rate": 3.5320680031229185e-07, "loss": 0.6028, "step": 12317 }, { "epoch": 0.88, "grad_norm": 16.821205990969958, "learning_rate": 3.5277819306007933e-07, "loss": 0.7082, "step": 12318 }, { "epoch": 0.88, "grad_norm": 2.7546826854046658, "learning_rate": 3.523498365078931e-07, "loss": 0.0746, "step": 12319 }, { "epoch": 0.88, "grad_norm": 2.896169085320266, "learning_rate": 3.519217306788414e-07, "loss": 0.1039, "step": 12320 }, { "epoch": 0.88, "grad_norm": 24.412695094985857, "learning_rate": 3.5149387559601733e-07, "loss": 0.4116, "step": 12321 }, { "epoch": 0.88, "grad_norm": 3.1897639279338112, "learning_rate": 3.5106627128250413e-07, "loss": 0.1117, "step": 12322 }, { "epoch": 0.88, "grad_norm": 9.47224189372129, "learning_rate": 3.506389177613684e-07, "loss": 0.4243, "step": 12323 }, { "epoch": 0.88, "grad_norm": 23.684415373741544, "learning_rate": 3.502118150556644e-07, "loss": 0.7395, "step": 12324 }, { "epoch": 0.88, "grad_norm": 3.3906014035223238, "learning_rate": 3.49784963188432e-07, "loss": 0.1693, "step": 12325 }, { "epoch": 0.88, "grad_norm": 14.274776355231749, "learning_rate": 3.493583621826996e-07, "loss": 0.4117, "step": 12326 }, { "epoch": 0.88, "grad_norm": 4.012427236300459, "learning_rate": 3.4893201206147864e-07, "loss": 0.2218, "step": 12327 }, { "epoch": 0.88, "grad_norm": 2.948265693584574, "learning_rate": 3.4850591284777133e-07, "loss": 0.1518, "step": 12328 }, { "epoch": 0.88, "grad_norm": 3.1047286395894123, "learning_rate": 3.4808006456456265e-07, "loss": 0.1459, "step": 12329 }, { "epoch": 0.88, "grad_norm": 3.2561448736534278, "learning_rate": 3.476544672348259e-07, "loss": 0.1552, "step": 12330 }, { "epoch": 0.88, "grad_norm": 2.9562611748840073, "learning_rate": 3.4722912088152096e-07, "loss": 0.1193, "step": 12331 }, { "epoch": 0.88, "grad_norm": 21.500005134675465, "learning_rate": 3.468040255275934e-07, "loss": 0.8475, "step": 12332 }, { "epoch": 0.88, "grad_norm": 13.230430516602876, "learning_rate": 3.463791811959755e-07, "loss": 0.6921, "step": 12333 }, { "epoch": 0.88, "grad_norm": 3.0597776205029628, "learning_rate": 3.459545879095849e-07, "loss": 0.1416, "step": 12334 }, { "epoch": 0.88, "grad_norm": 10.891435202344608, "learning_rate": 3.455302456913284e-07, "loss": 0.5257, "step": 12335 }, { "epoch": 0.88, "grad_norm": 2.8851291728189983, "learning_rate": 3.451061545640971e-07, "loss": 0.1442, "step": 12336 }, { "epoch": 0.88, "grad_norm": 9.454209227371468, "learning_rate": 3.4468231455076937e-07, "loss": 0.4906, "step": 12337 }, { "epoch": 0.88, "grad_norm": 2.83105782266648, "learning_rate": 3.4425872567420913e-07, "loss": 0.1296, "step": 12338 }, { "epoch": 0.88, "grad_norm": 3.859663394587689, "learning_rate": 3.43835387957267e-07, "loss": 0.1497, "step": 12339 }, { "epoch": 0.88, "grad_norm": 4.2202814008915395, "learning_rate": 3.4341230142278195e-07, "loss": 0.8025, "step": 12340 }, { "epoch": 0.88, "grad_norm": 4.659524647126663, "learning_rate": 3.429894660935773e-07, "loss": 0.3391, "step": 12341 }, { "epoch": 0.89, "grad_norm": 3.481715147430967, "learning_rate": 3.4256688199246267e-07, "loss": 0.1483, "step": 12342 }, { "epoch": 0.89, "grad_norm": 4.435480538896312, "learning_rate": 3.4214454914223483e-07, "loss": 0.9167, "step": 12343 }, { "epoch": 0.89, "grad_norm": 10.272743948934746, "learning_rate": 3.4172246756567885e-07, "loss": 0.6109, "step": 12344 }, { "epoch": 0.89, "grad_norm": 3.268044300555383, "learning_rate": 3.413006372855626e-07, "loss": 0.1864, "step": 12345 }, { "epoch": 0.89, "grad_norm": 2.984029429976761, "learning_rate": 3.408790583246424e-07, "loss": 0.1568, "step": 12346 }, { "epoch": 0.89, "grad_norm": 17.00411538773698, "learning_rate": 3.4045773070566224e-07, "loss": 0.4198, "step": 12347 }, { "epoch": 0.89, "grad_norm": 21.67100792705525, "learning_rate": 3.4003665445135056e-07, "loss": 0.6937, "step": 12348 }, { "epoch": 0.89, "grad_norm": 3.0216475413000876, "learning_rate": 3.3961582958442095e-07, "loss": 0.166, "step": 12349 }, { "epoch": 0.89, "grad_norm": 13.288858466494961, "learning_rate": 3.39195256127578e-07, "loss": 0.5338, "step": 12350 }, { "epoch": 0.89, "grad_norm": 4.657816633957036, "learning_rate": 3.387749341035085e-07, "loss": 0.7115, "step": 12351 }, { "epoch": 0.89, "grad_norm": 15.711495156250756, "learning_rate": 3.3835486353488656e-07, "loss": 0.3219, "step": 12352 }, { "epoch": 0.89, "grad_norm": 3.967915788081252, "learning_rate": 3.3793504444437576e-07, "loss": 0.1575, "step": 12353 }, { "epoch": 0.89, "grad_norm": 20.489508519171896, "learning_rate": 3.375154768546224e-07, "loss": 0.8179, "step": 12354 }, { "epoch": 0.89, "grad_norm": 4.626265307665008, "learning_rate": 3.3709616078825946e-07, "loss": 0.7448, "step": 12355 }, { "epoch": 0.89, "grad_norm": 9.304361784068815, "learning_rate": 3.3667709626790945e-07, "loss": 0.4156, "step": 12356 }, { "epoch": 0.89, "grad_norm": 4.0166859149922995, "learning_rate": 3.362582833161787e-07, "loss": 0.7094, "step": 12357 }, { "epoch": 0.89, "grad_norm": 9.217883539479809, "learning_rate": 3.3583972195565975e-07, "loss": 0.4199, "step": 12358 }, { "epoch": 0.89, "grad_norm": 15.707643925331428, "learning_rate": 3.3542141220893334e-07, "loss": 0.2256, "step": 12359 }, { "epoch": 0.89, "grad_norm": 2.946112123861699, "learning_rate": 3.3500335409856643e-07, "loss": 0.1653, "step": 12360 }, { "epoch": 0.89, "grad_norm": 4.737309156004313, "learning_rate": 3.3458554764710885e-07, "loss": 0.8683, "step": 12361 }, { "epoch": 0.89, "grad_norm": 16.522631636606246, "learning_rate": 3.341679928771019e-07, "loss": 1.2132, "step": 12362 }, { "epoch": 0.89, "grad_norm": 3.7589734308001863, "learning_rate": 3.3375068981107094e-07, "loss": 0.1563, "step": 12363 }, { "epoch": 0.89, "grad_norm": 3.3232461894997125, "learning_rate": 3.333336384715269e-07, "loss": 0.1165, "step": 12364 }, { "epoch": 0.89, "grad_norm": 4.371424272804809, "learning_rate": 3.329168388809695e-07, "loss": 0.1385, "step": 12365 }, { "epoch": 0.89, "grad_norm": 5.527108527170871, "learning_rate": 3.325002910618824e-07, "loss": 0.7785, "step": 12366 }, { "epoch": 0.89, "grad_norm": 3.63809155089455, "learning_rate": 3.320839950367372e-07, "loss": 0.1438, "step": 12367 }, { "epoch": 0.89, "grad_norm": 4.172818895333893, "learning_rate": 3.316679508279913e-07, "loss": 0.2343, "step": 12368 }, { "epoch": 0.89, "grad_norm": 3.538549343001547, "learning_rate": 3.3125215845808913e-07, "loss": 0.1585, "step": 12369 }, { "epoch": 0.89, "grad_norm": 17.644714819168055, "learning_rate": 3.3083661794946096e-07, "loss": 0.3014, "step": 12370 }, { "epoch": 0.89, "grad_norm": 37.16761292935801, "learning_rate": 3.3042132932452285e-07, "loss": 0.8613, "step": 12371 }, { "epoch": 0.89, "grad_norm": 16.801151807690182, "learning_rate": 3.300062926056802e-07, "loss": 1.0479, "step": 12372 }, { "epoch": 0.89, "grad_norm": 3.7264022294919235, "learning_rate": 3.2959150781531944e-07, "loss": 0.122, "step": 12373 }, { "epoch": 0.89, "grad_norm": 11.976901653435968, "learning_rate": 3.2917697497581946e-07, "loss": 0.5766, "step": 12374 }, { "epoch": 0.89, "grad_norm": 4.700677719978456, "learning_rate": 3.2876269410954176e-07, "loss": 0.9435, "step": 12375 }, { "epoch": 0.89, "grad_norm": 4.641358644842256, "learning_rate": 3.2834866523883457e-07, "loss": 0.8969, "step": 12376 }, { "epoch": 0.89, "grad_norm": 4.099602198563807, "learning_rate": 3.2793488838603336e-07, "loss": 0.7042, "step": 12377 }, { "epoch": 0.89, "grad_norm": 10.696839623581354, "learning_rate": 3.2752136357346087e-07, "loss": 0.5073, "step": 12378 }, { "epoch": 0.89, "grad_norm": 2.2892400724453585, "learning_rate": 3.271080908234242e-07, "loss": 0.0998, "step": 12379 }, { "epoch": 0.89, "grad_norm": 1.861979258052713, "learning_rate": 3.2669507015821775e-07, "loss": 0.0599, "step": 12380 }, { "epoch": 0.89, "grad_norm": 3.013176573094458, "learning_rate": 3.2628230160012376e-07, "loss": 0.1548, "step": 12381 }, { "epoch": 0.89, "grad_norm": 14.431431210930556, "learning_rate": 3.2586978517140824e-07, "loss": 1.0459, "step": 12382 }, { "epoch": 0.89, "grad_norm": 2.7000985492726985, "learning_rate": 3.254575208943245e-07, "loss": 0.1048, "step": 12383 }, { "epoch": 0.89, "grad_norm": 3.8096236518461626, "learning_rate": 3.250455087911153e-07, "loss": 0.6681, "step": 12384 }, { "epoch": 0.89, "grad_norm": 10.736294083583314, "learning_rate": 3.24633748884004e-07, "loss": 0.6189, "step": 12385 }, { "epoch": 0.89, "grad_norm": 2.2169309779767876, "learning_rate": 3.242222411952045e-07, "loss": 0.0877, "step": 12386 }, { "epoch": 0.89, "grad_norm": 2.7409584127007527, "learning_rate": 3.2381098574691684e-07, "loss": 0.1166, "step": 12387 }, { "epoch": 0.89, "grad_norm": 4.990952878155614, "learning_rate": 3.2339998256132597e-07, "loss": 0.8538, "step": 12388 }, { "epoch": 0.89, "grad_norm": 23.00475163344102, "learning_rate": 3.229892316606037e-07, "loss": 1.1032, "step": 12389 }, { "epoch": 0.89, "grad_norm": 4.053859469094586, "learning_rate": 3.2257873306690944e-07, "loss": 0.1692, "step": 12390 }, { "epoch": 0.89, "grad_norm": 3.386623293604715, "learning_rate": 3.221684868023878e-07, "loss": 0.1266, "step": 12391 }, { "epoch": 0.89, "grad_norm": 3.6368055945604567, "learning_rate": 3.2175849288916873e-07, "loss": 0.2241, "step": 12392 }, { "epoch": 0.89, "grad_norm": 17.671531897818525, "learning_rate": 3.2134875134937184e-07, "loss": 0.7738, "step": 12393 }, { "epoch": 0.89, "grad_norm": 22.415772906247152, "learning_rate": 3.2093926220509996e-07, "loss": 0.4523, "step": 12394 }, { "epoch": 0.89, "grad_norm": 16.754705457787292, "learning_rate": 3.205300254784427e-07, "loss": 0.3783, "step": 12395 }, { "epoch": 0.89, "grad_norm": 29.37805969992594, "learning_rate": 3.201210411914796e-07, "loss": 0.5063, "step": 12396 }, { "epoch": 0.89, "grad_norm": 35.66842187416367, "learning_rate": 3.197123093662713e-07, "loss": 0.4566, "step": 12397 }, { "epoch": 0.89, "grad_norm": 8.097469326579484, "learning_rate": 3.1930383002486744e-07, "loss": 0.5596, "step": 12398 }, { "epoch": 0.89, "grad_norm": 13.037347802501905, "learning_rate": 3.1889560318930537e-07, "loss": 0.7417, "step": 12399 }, { "epoch": 0.89, "grad_norm": 2.569604173567066, "learning_rate": 3.1848762888160636e-07, "loss": 0.1164, "step": 12400 }, { "epoch": 0.89, "grad_norm": 9.668923668834264, "learning_rate": 3.1807990712377833e-07, "loss": 0.6273, "step": 12401 }, { "epoch": 0.89, "grad_norm": 3.277594724346284, "learning_rate": 3.176724379378182e-07, "loss": 0.0982, "step": 12402 }, { "epoch": 0.89, "grad_norm": 2.7644233637960083, "learning_rate": 3.172652213457067e-07, "loss": 0.1126, "step": 12403 }, { "epoch": 0.89, "grad_norm": 24.547982268113948, "learning_rate": 3.168582573694112e-07, "loss": 0.9505, "step": 12404 }, { "epoch": 0.89, "grad_norm": 23.630314888161905, "learning_rate": 3.164515460308848e-07, "loss": 0.8311, "step": 12405 }, { "epoch": 0.89, "grad_norm": 4.073893745200298, "learning_rate": 3.1604508735207093e-07, "loss": 0.6919, "step": 12406 }, { "epoch": 0.89, "grad_norm": 3.5099912219491154, "learning_rate": 3.156388813548944e-07, "loss": 0.1779, "step": 12407 }, { "epoch": 0.89, "grad_norm": 11.699052265304863, "learning_rate": 3.1523292806126926e-07, "loss": 0.4658, "step": 12408 }, { "epoch": 0.89, "grad_norm": 4.601475636855972, "learning_rate": 3.148272274930941e-07, "loss": 0.1624, "step": 12409 }, { "epoch": 0.89, "grad_norm": 4.792492708083033, "learning_rate": 3.144217796722554e-07, "loss": 0.6491, "step": 12410 }, { "epoch": 0.89, "grad_norm": 16.484751683089403, "learning_rate": 3.1401658462062665e-07, "loss": 0.5242, "step": 12411 }, { "epoch": 0.89, "grad_norm": 4.58152599289572, "learning_rate": 3.13611642360066e-07, "loss": 0.7761, "step": 12412 }, { "epoch": 0.89, "grad_norm": 14.315629773815425, "learning_rate": 3.132069529124182e-07, "loss": 0.5892, "step": 12413 }, { "epoch": 0.89, "grad_norm": 13.898316076365052, "learning_rate": 3.128025162995135e-07, "loss": 0.9324, "step": 12414 }, { "epoch": 0.89, "grad_norm": 23.36583244423653, "learning_rate": 3.123983325431723e-07, "loss": 0.9892, "step": 12415 }, { "epoch": 0.89, "grad_norm": 3.527275727563059, "learning_rate": 3.119944016651971e-07, "loss": 0.1935, "step": 12416 }, { "epoch": 0.89, "grad_norm": 4.830325784357155, "learning_rate": 3.1159072368737886e-07, "loss": 0.8789, "step": 12417 }, { "epoch": 0.89, "grad_norm": 3.279259695145326, "learning_rate": 3.111872986314946e-07, "loss": 0.1558, "step": 12418 }, { "epoch": 0.89, "grad_norm": 4.312598027482513, "learning_rate": 3.1078412651930734e-07, "loss": 0.7215, "step": 12419 }, { "epoch": 0.89, "grad_norm": 3.6000411304696214, "learning_rate": 3.103812073725665e-07, "loss": 0.108, "step": 12420 }, { "epoch": 0.89, "grad_norm": 28.602303826781814, "learning_rate": 3.0997854121300965e-07, "loss": 0.9752, "step": 12421 }, { "epoch": 0.89, "grad_norm": 16.807267748764875, "learning_rate": 3.0957612806235717e-07, "loss": 0.2903, "step": 12422 }, { "epoch": 0.89, "grad_norm": 2.8880857172632357, "learning_rate": 3.091739679423178e-07, "loss": 0.2324, "step": 12423 }, { "epoch": 0.89, "grad_norm": 23.13243395612931, "learning_rate": 3.0877206087458755e-07, "loss": 1.2956, "step": 12424 }, { "epoch": 0.89, "grad_norm": 12.249576701316531, "learning_rate": 3.083704068808474e-07, "loss": 0.6017, "step": 12425 }, { "epoch": 0.89, "grad_norm": 3.2798683765316237, "learning_rate": 3.079690059827639e-07, "loss": 0.1478, "step": 12426 }, { "epoch": 0.89, "grad_norm": 3.810799499675485, "learning_rate": 3.0756785820199365e-07, "loss": 0.1447, "step": 12427 }, { "epoch": 0.89, "grad_norm": 11.283616623564745, "learning_rate": 3.071669635601754e-07, "loss": 0.6111, "step": 12428 }, { "epoch": 0.89, "grad_norm": 3.1520308002984514, "learning_rate": 3.067663220789352e-07, "loss": 0.176, "step": 12429 }, { "epoch": 0.89, "grad_norm": 2.9029750965729697, "learning_rate": 3.06365933779888e-07, "loss": 0.125, "step": 12430 }, { "epoch": 0.89, "grad_norm": 9.194481674950756, "learning_rate": 3.0596579868463204e-07, "loss": 0.5724, "step": 12431 }, { "epoch": 0.89, "grad_norm": 21.473465546785597, "learning_rate": 3.055659168147529e-07, "loss": 0.5228, "step": 12432 }, { "epoch": 0.89, "grad_norm": 4.189693223988119, "learning_rate": 3.051662881918249e-07, "loss": 0.1585, "step": 12433 }, { "epoch": 0.89, "grad_norm": 3.5537014121143926, "learning_rate": 3.047669128374037e-07, "loss": 0.1801, "step": 12434 }, { "epoch": 0.89, "grad_norm": 5.349424182795345, "learning_rate": 3.0436779077303413e-07, "loss": 0.8531, "step": 12435 }, { "epoch": 0.89, "grad_norm": 9.056438583751747, "learning_rate": 3.0396892202024906e-07, "loss": 0.3284, "step": 12436 }, { "epoch": 0.89, "grad_norm": 3.187037921635775, "learning_rate": 3.0357030660056574e-07, "loss": 0.1569, "step": 12437 }, { "epoch": 0.89, "grad_norm": 10.042666884874725, "learning_rate": 3.031719445354864e-07, "loss": 0.3363, "step": 12438 }, { "epoch": 0.89, "grad_norm": 2.663417316779527, "learning_rate": 3.0277383584650264e-07, "loss": 0.1101, "step": 12439 }, { "epoch": 0.89, "grad_norm": 11.862511979862669, "learning_rate": 3.023759805550908e-07, "loss": 0.5647, "step": 12440 }, { "epoch": 0.89, "grad_norm": 12.252851943719289, "learning_rate": 3.0197837868271295e-07, "loss": 0.5354, "step": 12441 }, { "epoch": 0.89, "grad_norm": 2.740425710728999, "learning_rate": 3.015810302508182e-07, "loss": 0.1353, "step": 12442 }, { "epoch": 0.89, "grad_norm": 11.79221314989532, "learning_rate": 3.0118393528084323e-07, "loss": 0.6971, "step": 12443 }, { "epoch": 0.89, "grad_norm": 4.422651340327637, "learning_rate": 3.0078709379420866e-07, "loss": 0.8172, "step": 12444 }, { "epoch": 0.89, "grad_norm": 6.864436689528525, "learning_rate": 3.0039050581232353e-07, "loss": 0.2554, "step": 12445 }, { "epoch": 0.89, "grad_norm": 4.677012440467809, "learning_rate": 2.999941713565807e-07, "loss": 1.0369, "step": 12446 }, { "epoch": 0.89, "grad_norm": 3.7603949364412284, "learning_rate": 2.99598090448362e-07, "loss": 0.7243, "step": 12447 }, { "epoch": 0.89, "grad_norm": 5.356296331946362, "learning_rate": 2.992022631090341e-07, "loss": 0.1837, "step": 12448 }, { "epoch": 0.89, "grad_norm": 5.109451111949096, "learning_rate": 2.9880668935995117e-07, "loss": 0.9823, "step": 12449 }, { "epoch": 0.89, "grad_norm": 24.020786280036102, "learning_rate": 2.9841136922245274e-07, "loss": 0.4903, "step": 12450 }, { "epoch": 0.89, "grad_norm": 26.782112579696882, "learning_rate": 2.9801630271786344e-07, "loss": 0.9599, "step": 12451 }, { "epoch": 0.89, "grad_norm": 4.228057145419519, "learning_rate": 2.976214898674967e-07, "loss": 0.1348, "step": 12452 }, { "epoch": 0.89, "grad_norm": 5.639211834436086, "learning_rate": 2.972269306926517e-07, "loss": 0.9212, "step": 12453 }, { "epoch": 0.89, "grad_norm": 10.728387143033709, "learning_rate": 2.968326252146125e-07, "loss": 0.4186, "step": 12454 }, { "epoch": 0.89, "grad_norm": 19.275316258218176, "learning_rate": 2.96438573454651e-07, "loss": 0.6636, "step": 12455 }, { "epoch": 0.89, "grad_norm": 3.194609554292521, "learning_rate": 2.9604477543402456e-07, "loss": 0.1343, "step": 12456 }, { "epoch": 0.89, "grad_norm": 4.6879698390798215, "learning_rate": 2.9565123117397675e-07, "loss": 0.8256, "step": 12457 }, { "epoch": 0.89, "grad_norm": 3.023962963923674, "learning_rate": 2.952579406957384e-07, "loss": 0.1306, "step": 12458 }, { "epoch": 0.89, "grad_norm": 4.266123694453022, "learning_rate": 2.948649040205259e-07, "loss": 0.1574, "step": 12459 }, { "epoch": 0.89, "grad_norm": 11.979283753031897, "learning_rate": 2.944721211695406e-07, "loss": 0.6664, "step": 12460 }, { "epoch": 0.89, "grad_norm": 92.7333747529862, "learning_rate": 2.940795921639744e-07, "loss": 1.0546, "step": 12461 }, { "epoch": 0.89, "grad_norm": 22.996317164406964, "learning_rate": 2.9368731702500096e-07, "loss": 0.5701, "step": 12462 }, { "epoch": 0.89, "grad_norm": 2.92866993397727, "learning_rate": 2.932952957737817e-07, "loss": 0.0909, "step": 12463 }, { "epoch": 0.89, "grad_norm": 5.884905733818743, "learning_rate": 2.9290352843146574e-07, "loss": 0.2451, "step": 12464 }, { "epoch": 0.89, "grad_norm": 21.90328034223117, "learning_rate": 2.925120150191879e-07, "loss": 1.1422, "step": 12465 }, { "epoch": 0.89, "grad_norm": 14.012107024443235, "learning_rate": 2.9212075555806685e-07, "loss": 0.5888, "step": 12466 }, { "epoch": 0.89, "grad_norm": 4.2331530282791485, "learning_rate": 2.9172975006921125e-07, "loss": 0.2235, "step": 12467 }, { "epoch": 0.89, "grad_norm": 3.185122065567626, "learning_rate": 2.913389985737142e-07, "loss": 0.1072, "step": 12468 }, { "epoch": 0.89, "grad_norm": 4.5796147014136475, "learning_rate": 2.9094850109265493e-07, "loss": 0.7998, "step": 12469 }, { "epoch": 0.89, "grad_norm": 12.916963465188617, "learning_rate": 2.9055825764709946e-07, "loss": 0.6591, "step": 12470 }, { "epoch": 0.89, "grad_norm": 3.0865256228967826, "learning_rate": 2.9016826825809917e-07, "loss": 0.115, "step": 12471 }, { "epoch": 0.89, "grad_norm": 3.096697674851065, "learning_rate": 2.897785329466929e-07, "loss": 0.1441, "step": 12472 }, { "epoch": 0.89, "grad_norm": 4.029450814185677, "learning_rate": 2.8938905173390653e-07, "loss": 0.2761, "step": 12473 }, { "epoch": 0.89, "grad_norm": 5.237034793517681, "learning_rate": 2.8899982464074994e-07, "loss": 0.2701, "step": 12474 }, { "epoch": 0.89, "grad_norm": 2.8891107963425786, "learning_rate": 2.886108516882197e-07, "loss": 0.2033, "step": 12475 }, { "epoch": 0.89, "grad_norm": 15.084391238154444, "learning_rate": 2.8822213289730115e-07, "loss": 0.6013, "step": 12476 }, { "epoch": 0.89, "grad_norm": 13.913029154098442, "learning_rate": 2.878336682889638e-07, "loss": 0.3558, "step": 12477 }, { "epoch": 0.89, "grad_norm": 15.999998599177722, "learning_rate": 2.8744545788416346e-07, "loss": 0.6227, "step": 12478 }, { "epoch": 0.89, "grad_norm": 4.829093028101982, "learning_rate": 2.8705750170384236e-07, "loss": 0.9443, "step": 12479 }, { "epoch": 0.89, "grad_norm": 4.416784794820546, "learning_rate": 2.8666979976892997e-07, "loss": 0.7521, "step": 12480 }, { "epoch": 0.89, "grad_norm": 17.63307417437362, "learning_rate": 2.862823521003405e-07, "loss": 0.8522, "step": 12481 }, { "epoch": 0.9, "grad_norm": 3.328023281646916, "learning_rate": 2.8589515871897623e-07, "loss": 0.1941, "step": 12482 }, { "epoch": 0.9, "grad_norm": 14.006053271794395, "learning_rate": 2.8550821964572426e-07, "loss": 0.5289, "step": 12483 }, { "epoch": 0.9, "grad_norm": 3.389746068144768, "learning_rate": 2.85121534901458e-07, "loss": 0.1003, "step": 12484 }, { "epoch": 0.9, "grad_norm": 3.4837662960588416, "learning_rate": 2.8473510450703844e-07, "loss": 0.1919, "step": 12485 }, { "epoch": 0.9, "grad_norm": 2.612081079501245, "learning_rate": 2.8434892848331176e-07, "loss": 0.1236, "step": 12486 }, { "epoch": 0.9, "grad_norm": 2.797560883085866, "learning_rate": 2.839630068511107e-07, "loss": 0.1161, "step": 12487 }, { "epoch": 0.9, "grad_norm": 10.552753759667164, "learning_rate": 2.835773396312541e-07, "loss": 0.3734, "step": 12488 }, { "epoch": 0.9, "grad_norm": 3.4623766528551534, "learning_rate": 2.831919268445477e-07, "loss": 0.1762, "step": 12489 }, { "epoch": 0.9, "grad_norm": 11.76315661051682, "learning_rate": 2.8280676851178245e-07, "loss": 0.5909, "step": 12490 }, { "epoch": 0.9, "grad_norm": 4.677794130314371, "learning_rate": 2.824218646537363e-07, "loss": 0.702, "step": 12491 }, { "epoch": 0.9, "grad_norm": 15.805783262462135, "learning_rate": 2.820372152911738e-07, "loss": 0.6265, "step": 12492 }, { "epoch": 0.9, "grad_norm": 4.048933255443557, "learning_rate": 2.81652820444846e-07, "loss": 0.1706, "step": 12493 }, { "epoch": 0.9, "grad_norm": 7.963998688213515, "learning_rate": 2.812686801354875e-07, "loss": 0.4893, "step": 12494 }, { "epoch": 0.9, "grad_norm": 3.0931935124321215, "learning_rate": 2.8088479438382275e-07, "loss": 0.135, "step": 12495 }, { "epoch": 0.9, "grad_norm": 2.177896199049973, "learning_rate": 2.805011632105603e-07, "loss": 0.1073, "step": 12496 }, { "epoch": 0.9, "grad_norm": 2.8460000317972045, "learning_rate": 2.8011778663639524e-07, "loss": 0.1978, "step": 12497 }, { "epoch": 0.9, "grad_norm": 4.731964340257252, "learning_rate": 2.7973466468201093e-07, "loss": 0.1668, "step": 12498 }, { "epoch": 0.9, "grad_norm": 4.51224302501596, "learning_rate": 2.793517973680737e-07, "loss": 0.854, "step": 12499 }, { "epoch": 0.9, "grad_norm": 6.39376471387419, "learning_rate": 2.7896918471523814e-07, "loss": 0.8023, "step": 12500 }, { "epoch": 0.9, "grad_norm": 2.8243089129042107, "learning_rate": 2.785868267441455e-07, "loss": 0.1509, "step": 12501 }, { "epoch": 0.9, "grad_norm": 12.631850499250046, "learning_rate": 2.782047234754226e-07, "loss": 0.6494, "step": 12502 }, { "epoch": 0.9, "grad_norm": 18.220299053874047, "learning_rate": 2.778228749296802e-07, "loss": 0.3707, "step": 12503 }, { "epoch": 0.9, "grad_norm": 2.7167560075103223, "learning_rate": 2.7744128112752067e-07, "loss": 0.1166, "step": 12504 }, { "epoch": 0.9, "grad_norm": 2.9914188494400995, "learning_rate": 2.770599420895287e-07, "loss": 0.1169, "step": 12505 }, { "epoch": 0.9, "grad_norm": 4.205207844764214, "learning_rate": 2.7667885783627446e-07, "loss": 0.7101, "step": 12506 }, { "epoch": 0.9, "grad_norm": 3.5122315415262286, "learning_rate": 2.762980283883171e-07, "loss": 0.2388, "step": 12507 }, { "epoch": 0.9, "grad_norm": 13.466799712549312, "learning_rate": 2.7591745376620125e-07, "loss": 0.6198, "step": 12508 }, { "epoch": 0.9, "grad_norm": 8.891441625830888, "learning_rate": 2.7553713399045666e-07, "loss": 0.7471, "step": 12509 }, { "epoch": 0.9, "grad_norm": 4.087471169123137, "learning_rate": 2.7515706908160135e-07, "loss": 0.707, "step": 12510 }, { "epoch": 0.9, "grad_norm": 10.728385068605304, "learning_rate": 2.747772590601372e-07, "loss": 0.4882, "step": 12511 }, { "epoch": 0.9, "grad_norm": 4.117713339735028, "learning_rate": 2.7439770394655397e-07, "loss": 0.282, "step": 12512 }, { "epoch": 0.9, "grad_norm": 10.923319377044315, "learning_rate": 2.740184037613275e-07, "loss": 0.7299, "step": 12513 }, { "epoch": 0.9, "grad_norm": 16.12832488653355, "learning_rate": 2.736393585249192e-07, "loss": 1.0311, "step": 12514 }, { "epoch": 0.9, "grad_norm": 13.874048008595484, "learning_rate": 2.732605682577777e-07, "loss": 0.9445, "step": 12515 }, { "epoch": 0.9, "grad_norm": 3.850138116683119, "learning_rate": 2.7288203298033556e-07, "loss": 0.1818, "step": 12516 }, { "epoch": 0.9, "grad_norm": 59.365394024054886, "learning_rate": 2.7250375271301644e-07, "loss": 1.1097, "step": 12517 }, { "epoch": 0.9, "grad_norm": 3.3066224422641928, "learning_rate": 2.72125727476224e-07, "loss": 0.1414, "step": 12518 }, { "epoch": 0.9, "grad_norm": 3.987787978247231, "learning_rate": 2.717479572903531e-07, "loss": 0.2052, "step": 12519 }, { "epoch": 0.9, "grad_norm": 9.001791647092743, "learning_rate": 2.7137044217578235e-07, "loss": 0.4842, "step": 12520 }, { "epoch": 0.9, "grad_norm": 2.6272768947427076, "learning_rate": 2.709931821528777e-07, "loss": 0.0976, "step": 12521 }, { "epoch": 0.9, "grad_norm": 10.942157812420104, "learning_rate": 2.706161772419896e-07, "loss": 0.4883, "step": 12522 }, { "epoch": 0.9, "grad_norm": 12.94661456771474, "learning_rate": 2.702394274634584e-07, "loss": 0.6313, "step": 12523 }, { "epoch": 0.9, "grad_norm": 4.570730724755251, "learning_rate": 2.698629328376062e-07, "loss": 0.8364, "step": 12524 }, { "epoch": 0.9, "grad_norm": 3.946861450470238, "learning_rate": 2.69486693384744e-07, "loss": 0.1692, "step": 12525 }, { "epoch": 0.9, "grad_norm": 11.178958147809237, "learning_rate": 2.691107091251693e-07, "loss": 0.4373, "step": 12526 }, { "epoch": 0.9, "grad_norm": 2.91975046082824, "learning_rate": 2.6873498007916445e-07, "loss": 0.095, "step": 12527 }, { "epoch": 0.9, "grad_norm": 3.3058643724884385, "learning_rate": 2.6835950626699803e-07, "loss": 0.1362, "step": 12528 }, { "epoch": 0.9, "grad_norm": 22.13044078243067, "learning_rate": 2.679842877089273e-07, "loss": 1.1823, "step": 12529 }, { "epoch": 0.9, "grad_norm": 18.561772629668905, "learning_rate": 2.676093244251915e-07, "loss": 0.9885, "step": 12530 }, { "epoch": 0.9, "grad_norm": 4.31724501507047, "learning_rate": 2.6723461643601953e-07, "loss": 0.2058, "step": 12531 }, { "epoch": 0.9, "grad_norm": 3.386142765668106, "learning_rate": 2.668601637616258e-07, "loss": 0.1003, "step": 12532 }, { "epoch": 0.9, "grad_norm": 5.900691621114796, "learning_rate": 2.664859664222108e-07, "loss": 0.4244, "step": 12533 }, { "epoch": 0.9, "grad_norm": 4.3347547474268024, "learning_rate": 2.6611202443795947e-07, "loss": 0.7224, "step": 12534 }, { "epoch": 0.9, "grad_norm": 3.151097797060864, "learning_rate": 2.6573833782904625e-07, "loss": 0.2021, "step": 12535 }, { "epoch": 0.9, "grad_norm": 3.327411129767896, "learning_rate": 2.6536490661562995e-07, "loss": 0.1583, "step": 12536 }, { "epoch": 0.9, "grad_norm": 18.413549189333015, "learning_rate": 2.649917308178546e-07, "loss": 0.3454, "step": 12537 }, { "epoch": 0.9, "grad_norm": 8.385849507569356, "learning_rate": 2.646188104558528e-07, "loss": 0.5838, "step": 12538 }, { "epoch": 0.9, "grad_norm": 4.546730444886402, "learning_rate": 2.6424614554974195e-07, "loss": 0.7169, "step": 12539 }, { "epoch": 0.9, "grad_norm": 8.23116462618102, "learning_rate": 2.6387373611962476e-07, "loss": 0.353, "step": 12540 }, { "epoch": 0.9, "grad_norm": 26.041829525979423, "learning_rate": 2.635015821855941e-07, "loss": 0.5233, "step": 12541 }, { "epoch": 0.9, "grad_norm": 4.748371051874682, "learning_rate": 2.6312968376772384e-07, "loss": 0.7521, "step": 12542 }, { "epoch": 0.9, "grad_norm": 4.749901652123044, "learning_rate": 2.627580408860764e-07, "loss": 0.8157, "step": 12543 }, { "epoch": 0.9, "grad_norm": 4.755271746597521, "learning_rate": 2.623866535607017e-07, "loss": 0.1695, "step": 12544 }, { "epoch": 0.9, "grad_norm": 3.677716138273899, "learning_rate": 2.6201552181163435e-07, "loss": 0.1971, "step": 12545 }, { "epoch": 0.9, "grad_norm": 9.62936582791369, "learning_rate": 2.6164464565889446e-07, "loss": 0.6521, "step": 12546 }, { "epoch": 0.9, "grad_norm": 14.31770416931309, "learning_rate": 2.61274025122491e-07, "loss": 0.4852, "step": 12547 }, { "epoch": 0.9, "grad_norm": 8.054952669615345, "learning_rate": 2.609036602224169e-07, "loss": 0.4149, "step": 12548 }, { "epoch": 0.9, "grad_norm": 3.1119485798529816, "learning_rate": 2.6053355097865216e-07, "loss": 0.1573, "step": 12549 }, { "epoch": 0.9, "grad_norm": 7.8044427033982355, "learning_rate": 2.6016369741116153e-07, "loss": 0.3419, "step": 12550 }, { "epoch": 0.9, "grad_norm": 2.7502257116709226, "learning_rate": 2.5979409953989845e-07, "loss": 0.1314, "step": 12551 }, { "epoch": 0.9, "grad_norm": 18.380878563467014, "learning_rate": 2.594247573848019e-07, "loss": 0.7273, "step": 12552 }, { "epoch": 0.9, "grad_norm": 5.37947602962449, "learning_rate": 2.5905567096579486e-07, "loss": 0.225, "step": 12553 }, { "epoch": 0.9, "grad_norm": 3.706127890179712, "learning_rate": 2.5868684030278926e-07, "loss": 0.189, "step": 12554 }, { "epoch": 0.9, "grad_norm": 3.204108254394831, "learning_rate": 2.5831826541568127e-07, "loss": 0.2189, "step": 12555 }, { "epoch": 0.9, "grad_norm": 3.4627635333705835, "learning_rate": 2.579499463243551e-07, "loss": 0.1231, "step": 12556 }, { "epoch": 0.9, "grad_norm": 42.95066095060908, "learning_rate": 2.5758188304867984e-07, "loss": 0.4321, "step": 12557 }, { "epoch": 0.9, "grad_norm": 5.3188282105318265, "learning_rate": 2.5721407560851066e-07, "loss": 0.8619, "step": 12558 }, { "epoch": 0.9, "grad_norm": 24.316741766743128, "learning_rate": 2.5684652402368893e-07, "loss": 0.5941, "step": 12559 }, { "epoch": 0.9, "grad_norm": 4.016298681849045, "learning_rate": 2.564792283140444e-07, "loss": 0.1451, "step": 12560 }, { "epoch": 0.9, "grad_norm": 4.523028032857943, "learning_rate": 2.561121884993906e-07, "loss": 0.6336, "step": 12561 }, { "epoch": 0.9, "grad_norm": 4.172072727132234, "learning_rate": 2.557454045995267e-07, "loss": 0.1885, "step": 12562 }, { "epoch": 0.9, "grad_norm": 4.101277202946402, "learning_rate": 2.5537887663424075e-07, "loss": 0.2359, "step": 12563 }, { "epoch": 0.9, "grad_norm": 5.106356247173029, "learning_rate": 2.550126046233048e-07, "loss": 0.7978, "step": 12564 }, { "epoch": 0.9, "grad_norm": 3.7044283293576967, "learning_rate": 2.546465885864785e-07, "loss": 0.14, "step": 12565 }, { "epoch": 0.9, "grad_norm": 3.5844374636557563, "learning_rate": 2.5428082854350664e-07, "loss": 0.0972, "step": 12566 }, { "epoch": 0.9, "grad_norm": 14.262941476287846, "learning_rate": 2.539153245141207e-07, "loss": 0.6905, "step": 12567 }, { "epoch": 0.9, "grad_norm": 14.217463002450138, "learning_rate": 2.535500765180371e-07, "loss": 0.8534, "step": 12568 }, { "epoch": 0.9, "grad_norm": 3.0884342260354076, "learning_rate": 2.5318508457496116e-07, "loss": 0.1098, "step": 12569 }, { "epoch": 0.9, "grad_norm": 4.246239316718338, "learning_rate": 2.528203487045822e-07, "loss": 0.7537, "step": 12570 }, { "epoch": 0.9, "grad_norm": 4.609208041724086, "learning_rate": 2.5245586892657603e-07, "loss": 0.9226, "step": 12571 }, { "epoch": 0.9, "grad_norm": 4.331424443881578, "learning_rate": 2.520916452606059e-07, "loss": 0.7339, "step": 12572 }, { "epoch": 0.9, "grad_norm": 8.69803075120849, "learning_rate": 2.5172767772631946e-07, "loss": 0.4186, "step": 12573 }, { "epoch": 0.9, "grad_norm": 9.718180473080444, "learning_rate": 2.5136396634335093e-07, "loss": 0.3742, "step": 12574 }, { "epoch": 0.9, "grad_norm": 20.558533671216296, "learning_rate": 2.5100051113132294e-07, "loss": 1.1118, "step": 12575 }, { "epoch": 0.9, "grad_norm": 9.373548855273118, "learning_rate": 2.5063731210984097e-07, "loss": 0.5684, "step": 12576 }, { "epoch": 0.9, "grad_norm": 2.4881288903236465, "learning_rate": 2.5027436929849933e-07, "loss": 0.1445, "step": 12577 }, { "epoch": 0.9, "grad_norm": 5.364055508457846, "learning_rate": 2.499116827168763e-07, "loss": 0.863, "step": 12578 }, { "epoch": 0.9, "grad_norm": 7.189298282877122, "learning_rate": 2.4954925238453776e-07, "loss": 0.3878, "step": 12579 }, { "epoch": 0.9, "grad_norm": 2.9045200005900664, "learning_rate": 2.491870783210354e-07, "loss": 0.1504, "step": 12580 }, { "epoch": 0.9, "grad_norm": 2.946683092322426, "learning_rate": 2.48825160545908e-07, "loss": 0.1628, "step": 12581 }, { "epoch": 0.9, "grad_norm": 2.8291269668666756, "learning_rate": 2.4846349907867886e-07, "loss": 0.1099, "step": 12582 }, { "epoch": 0.9, "grad_norm": 11.14918368101397, "learning_rate": 2.4810209393885786e-07, "loss": 0.4043, "step": 12583 }, { "epoch": 0.9, "grad_norm": 5.289311747819481, "learning_rate": 2.4774094514594225e-07, "loss": 0.8358, "step": 12584 }, { "epoch": 0.9, "grad_norm": 23.08813569476439, "learning_rate": 2.473800527194148e-07, "loss": 0.6145, "step": 12585 }, { "epoch": 0.9, "grad_norm": 3.444130777942361, "learning_rate": 2.4701941667874375e-07, "loss": 0.2011, "step": 12586 }, { "epoch": 0.9, "grad_norm": 2.4802786473975167, "learning_rate": 2.4665903704338303e-07, "loss": 0.0794, "step": 12587 }, { "epoch": 0.9, "grad_norm": 22.659141741258438, "learning_rate": 2.462989138327765e-07, "loss": 0.6293, "step": 12588 }, { "epoch": 0.9, "grad_norm": 9.831914776094218, "learning_rate": 2.459390470663492e-07, "loss": 0.3463, "step": 12589 }, { "epoch": 0.9, "grad_norm": 4.641383900917197, "learning_rate": 2.4557943676351505e-07, "loss": 0.776, "step": 12590 }, { "epoch": 0.9, "grad_norm": 3.7266283792199197, "learning_rate": 2.4522008294367406e-07, "loss": 0.1725, "step": 12591 }, { "epoch": 0.9, "grad_norm": 3.264361420381991, "learning_rate": 2.448609856262113e-07, "loss": 0.1793, "step": 12592 }, { "epoch": 0.9, "grad_norm": 3.5686056561607877, "learning_rate": 2.445021448304996e-07, "loss": 0.1806, "step": 12593 }, { "epoch": 0.9, "grad_norm": 2.941391807871777, "learning_rate": 2.441435605758968e-07, "loss": 0.1509, "step": 12594 }, { "epoch": 0.9, "grad_norm": 3.429752012963929, "learning_rate": 2.4378523288174695e-07, "loss": 0.128, "step": 12595 }, { "epoch": 0.9, "grad_norm": 19.609134981515233, "learning_rate": 2.4342716176738e-07, "loss": 0.5816, "step": 12596 }, { "epoch": 0.9, "grad_norm": 7.7881718523001195, "learning_rate": 2.430693472521134e-07, "loss": 0.5334, "step": 12597 }, { "epoch": 0.9, "grad_norm": 2.616024996070755, "learning_rate": 2.4271178935524997e-07, "loss": 0.1039, "step": 12598 }, { "epoch": 0.9, "grad_norm": 3.5145012563171907, "learning_rate": 2.4235448809607707e-07, "loss": 0.2337, "step": 12599 }, { "epoch": 0.9, "grad_norm": 3.1578561708039183, "learning_rate": 2.419974434938721e-07, "loss": 0.0984, "step": 12600 }, { "epoch": 0.9, "grad_norm": 2.228782408704419, "learning_rate": 2.416406555678952e-07, "loss": 0.0764, "step": 12601 }, { "epoch": 0.9, "grad_norm": 9.71634470224675, "learning_rate": 2.412841243373926e-07, "loss": 0.6049, "step": 12602 }, { "epoch": 0.9, "grad_norm": 3.081410824443223, "learning_rate": 2.4092784982159947e-07, "loss": 0.1687, "step": 12603 }, { "epoch": 0.9, "grad_norm": 9.219551736052699, "learning_rate": 2.4057183203973433e-07, "loss": 0.3086, "step": 12604 }, { "epoch": 0.9, "grad_norm": 2.9729498295920096, "learning_rate": 2.4021607101100296e-07, "loss": 0.1507, "step": 12605 }, { "epoch": 0.9, "grad_norm": 2.9302331597392968, "learning_rate": 2.398605667545983e-07, "loss": 0.2067, "step": 12606 }, { "epoch": 0.9, "grad_norm": 2.511531618964688, "learning_rate": 2.3950531928969847e-07, "loss": 0.1337, "step": 12607 }, { "epoch": 0.9, "grad_norm": 4.069872322708259, "learning_rate": 2.391503286354663e-07, "loss": 0.8179, "step": 12608 }, { "epoch": 0.9, "grad_norm": 2.5885335528349134, "learning_rate": 2.3879559481105374e-07, "loss": 0.1152, "step": 12609 }, { "epoch": 0.9, "grad_norm": 2.9736641603706957, "learning_rate": 2.3844111783559666e-07, "loss": 0.176, "step": 12610 }, { "epoch": 0.9, "grad_norm": 4.225706800692584, "learning_rate": 2.380868977282169e-07, "loss": 0.3034, "step": 12611 }, { "epoch": 0.9, "grad_norm": 20.542224689967185, "learning_rate": 2.3773293450802538e-07, "loss": 0.5198, "step": 12612 }, { "epoch": 0.9, "grad_norm": 7.325806458042042, "learning_rate": 2.3737922819411619e-07, "loss": 0.3818, "step": 12613 }, { "epoch": 0.9, "grad_norm": 15.148313257097625, "learning_rate": 2.370257788055691e-07, "loss": 0.8146, "step": 12614 }, { "epoch": 0.9, "grad_norm": 3.992781804187287, "learning_rate": 2.3667258636145273e-07, "loss": 0.2388, "step": 12615 }, { "epoch": 0.9, "grad_norm": 2.963639726912379, "learning_rate": 2.363196508808202e-07, "loss": 0.1956, "step": 12616 }, { "epoch": 0.9, "grad_norm": 11.626536387725277, "learning_rate": 2.3596697238271015e-07, "loss": 0.5392, "step": 12617 }, { "epoch": 0.9, "grad_norm": 9.390723544166102, "learning_rate": 2.3561455088615014e-07, "loss": 0.3796, "step": 12618 }, { "epoch": 0.9, "grad_norm": 12.413034166820507, "learning_rate": 2.352623864101511e-07, "loss": 0.672, "step": 12619 }, { "epoch": 0.9, "grad_norm": 3.7312049103279032, "learning_rate": 2.3491047897371e-07, "loss": 0.2734, "step": 12620 }, { "epoch": 0.91, "grad_norm": 3.1060415815140443, "learning_rate": 2.3455882859581225e-07, "loss": 0.1058, "step": 12621 }, { "epoch": 0.91, "grad_norm": 9.850374626790849, "learning_rate": 2.342074352954271e-07, "loss": 0.4438, "step": 12622 }, { "epoch": 0.91, "grad_norm": 11.850873230435381, "learning_rate": 2.3385629909151164e-07, "loss": 0.7885, "step": 12623 }, { "epoch": 0.91, "grad_norm": 5.733587451855553, "learning_rate": 2.335054200030079e-07, "loss": 0.225, "step": 12624 }, { "epoch": 0.91, "grad_norm": 2.7261091335662733, "learning_rate": 2.3315479804884522e-07, "loss": 0.0939, "step": 12625 }, { "epoch": 0.91, "grad_norm": 14.747816784834313, "learning_rate": 2.3280443324793677e-07, "loss": 0.2821, "step": 12626 }, { "epoch": 0.91, "grad_norm": 11.047253137050863, "learning_rate": 2.324543256191847e-07, "loss": 0.3656, "step": 12627 }, { "epoch": 0.91, "grad_norm": 4.146468244084492, "learning_rate": 2.3210447518147549e-07, "loss": 0.6954, "step": 12628 }, { "epoch": 0.91, "grad_norm": 3.4187296797856037, "learning_rate": 2.3175488195368302e-07, "loss": 0.1874, "step": 12629 }, { "epoch": 0.91, "grad_norm": 6.263044541365186, "learning_rate": 2.314055459546649e-07, "loss": 0.3572, "step": 12630 }, { "epoch": 0.91, "grad_norm": 2.9346886271512775, "learning_rate": 2.310564672032678e-07, "loss": 0.1493, "step": 12631 }, { "epoch": 0.91, "grad_norm": 14.939712894980303, "learning_rate": 2.3070764571832326e-07, "loss": 0.4916, "step": 12632 }, { "epoch": 0.91, "grad_norm": 3.340395223516975, "learning_rate": 2.303590815186474e-07, "loss": 0.1937, "step": 12633 }, { "epoch": 0.91, "grad_norm": 3.0136545162116977, "learning_rate": 2.300107746230462e-07, "loss": 0.1544, "step": 12634 }, { "epoch": 0.91, "grad_norm": 2.7296544753934207, "learning_rate": 2.2966272505030806e-07, "loss": 0.1762, "step": 12635 }, { "epoch": 0.91, "grad_norm": 3.8984533467164377, "learning_rate": 2.2931493281920847e-07, "loss": 0.1334, "step": 12636 }, { "epoch": 0.91, "grad_norm": 11.51405943207085, "learning_rate": 2.289673979485113e-07, "loss": 0.4564, "step": 12637 }, { "epoch": 0.91, "grad_norm": 4.7506040741267865, "learning_rate": 2.286201204569638e-07, "loss": 0.1131, "step": 12638 }, { "epoch": 0.91, "grad_norm": 5.942972289956833, "learning_rate": 2.282731003632993e-07, "loss": 0.8017, "step": 12639 }, { "epoch": 0.91, "grad_norm": 4.849735950524334, "learning_rate": 2.279263376862395e-07, "loss": 0.6665, "step": 12640 }, { "epoch": 0.91, "grad_norm": 3.195156494141563, "learning_rate": 2.2757983244449e-07, "loss": 0.1708, "step": 12641 }, { "epoch": 0.91, "grad_norm": 3.6880167847756375, "learning_rate": 2.2723358465674417e-07, "loss": 0.1696, "step": 12642 }, { "epoch": 0.91, "grad_norm": 9.70946855462673, "learning_rate": 2.268875943416804e-07, "loss": 0.3794, "step": 12643 }, { "epoch": 0.91, "grad_norm": 4.847354820651541, "learning_rate": 2.2654186151796432e-07, "loss": 0.9062, "step": 12644 }, { "epoch": 0.91, "grad_norm": 3.5769154747003666, "learning_rate": 2.2619638620424544e-07, "loss": 0.1195, "step": 12645 }, { "epoch": 0.91, "grad_norm": 15.39455258559176, "learning_rate": 2.2585116841916222e-07, "loss": 0.5985, "step": 12646 }, { "epoch": 0.91, "grad_norm": 10.563512194274404, "learning_rate": 2.255062081813375e-07, "loss": 0.5471, "step": 12647 }, { "epoch": 0.91, "grad_norm": 2.4499782838956023, "learning_rate": 2.2516150550937977e-07, "loss": 0.0899, "step": 12648 }, { "epoch": 0.91, "grad_norm": 4.604046146759887, "learning_rate": 2.2481706042188578e-07, "loss": 0.77, "step": 12649 }, { "epoch": 0.91, "grad_norm": 25.230876377174287, "learning_rate": 2.2447287293743681e-07, "loss": 1.0094, "step": 12650 }, { "epoch": 0.91, "grad_norm": 3.3479294963574864, "learning_rate": 2.2412894307459855e-07, "loss": 0.143, "step": 12651 }, { "epoch": 0.91, "grad_norm": 3.595496630993987, "learning_rate": 2.2378527085192725e-07, "loss": 0.1308, "step": 12652 }, { "epoch": 0.91, "grad_norm": 14.773660853238896, "learning_rate": 2.2344185628796145e-07, "loss": 0.7031, "step": 12653 }, { "epoch": 0.91, "grad_norm": 12.889674106279688, "learning_rate": 2.230986994012263e-07, "loss": 0.5969, "step": 12654 }, { "epoch": 0.91, "grad_norm": 18.09268111639125, "learning_rate": 2.2275580021023535e-07, "loss": 0.7224, "step": 12655 }, { "epoch": 0.91, "grad_norm": 3.230738677264715, "learning_rate": 2.2241315873348658e-07, "loss": 0.1284, "step": 12656 }, { "epoch": 0.91, "grad_norm": 3.4189483020538116, "learning_rate": 2.2207077498946294e-07, "loss": 0.1131, "step": 12657 }, { "epoch": 0.91, "grad_norm": 3.880850599947329, "learning_rate": 2.2172864899663636e-07, "loss": 0.1657, "step": 12658 }, { "epoch": 0.91, "grad_norm": 15.575681004247167, "learning_rate": 2.213867807734621e-07, "loss": 0.4962, "step": 12659 }, { "epoch": 0.91, "grad_norm": 3.9699194146068306, "learning_rate": 2.2104517033838314e-07, "loss": 0.2209, "step": 12660 }, { "epoch": 0.91, "grad_norm": 2.730290935148788, "learning_rate": 2.2070381770982752e-07, "loss": 0.1016, "step": 12661 }, { "epoch": 0.91, "grad_norm": 3.378676133768837, "learning_rate": 2.2036272290621109e-07, "loss": 0.1638, "step": 12662 }, { "epoch": 0.91, "grad_norm": 17.068961744956347, "learning_rate": 2.2002188594593298e-07, "loss": 1.0089, "step": 12663 }, { "epoch": 0.91, "grad_norm": 5.317370527070516, "learning_rate": 2.196813068473813e-07, "loss": 0.1864, "step": 12664 }, { "epoch": 0.91, "grad_norm": 3.2609695653471777, "learning_rate": 2.1934098562892858e-07, "loss": 0.1038, "step": 12665 }, { "epoch": 0.91, "grad_norm": 7.270985285906631, "learning_rate": 2.1900092230893454e-07, "loss": 0.4293, "step": 12666 }, { "epoch": 0.91, "grad_norm": 17.729089973069048, "learning_rate": 2.186611169057423e-07, "loss": 0.7523, "step": 12667 }, { "epoch": 0.91, "grad_norm": 12.626050451017218, "learning_rate": 2.1832156943768557e-07, "loss": 0.7999, "step": 12668 }, { "epoch": 0.91, "grad_norm": 2.725447607591509, "learning_rate": 2.1798227992308018e-07, "loss": 0.0845, "step": 12669 }, { "epoch": 0.91, "grad_norm": 9.247007513202377, "learning_rate": 2.1764324838022932e-07, "loss": 0.2869, "step": 12670 }, { "epoch": 0.91, "grad_norm": 45.9228958885095, "learning_rate": 2.173044748274239e-07, "loss": 0.9912, "step": 12671 }, { "epoch": 0.91, "grad_norm": 4.756656671611684, "learning_rate": 2.1696595928293817e-07, "loss": 0.2848, "step": 12672 }, { "epoch": 0.91, "grad_norm": 4.738116434937684, "learning_rate": 2.1662770176503367e-07, "loss": 0.7137, "step": 12673 }, { "epoch": 0.91, "grad_norm": 4.896077456992213, "learning_rate": 2.1628970229196023e-07, "loss": 0.7374, "step": 12674 }, { "epoch": 0.91, "grad_norm": 9.929253511203449, "learning_rate": 2.159519608819488e-07, "loss": 0.331, "step": 12675 }, { "epoch": 0.91, "grad_norm": 10.781601422537225, "learning_rate": 2.156144775532204e-07, "loss": 0.7271, "step": 12676 }, { "epoch": 0.91, "grad_norm": 5.2573778625463525, "learning_rate": 2.1527725232398154e-07, "loss": 0.9897, "step": 12677 }, { "epoch": 0.91, "grad_norm": 2.9496626331843627, "learning_rate": 2.1494028521242328e-07, "loss": 0.1093, "step": 12678 }, { "epoch": 0.91, "grad_norm": 2.5303281529281603, "learning_rate": 2.146035762367238e-07, "loss": 0.1215, "step": 12679 }, { "epoch": 0.91, "grad_norm": 9.251917651218053, "learning_rate": 2.1426712541504858e-07, "loss": 0.3186, "step": 12680 }, { "epoch": 0.91, "grad_norm": 17.770569987603192, "learning_rate": 2.13930932765547e-07, "loss": 0.7735, "step": 12681 }, { "epoch": 0.91, "grad_norm": 14.177207170138297, "learning_rate": 2.1359499830635455e-07, "loss": 0.7344, "step": 12682 }, { "epoch": 0.91, "grad_norm": 11.009543394135981, "learning_rate": 2.1325932205559508e-07, "loss": 0.6989, "step": 12683 }, { "epoch": 0.91, "grad_norm": 4.937161312095255, "learning_rate": 2.129239040313763e-07, "loss": 1.1247, "step": 12684 }, { "epoch": 0.91, "grad_norm": 9.528344035911827, "learning_rate": 2.125887442517921e-07, "loss": 0.5866, "step": 12685 }, { "epoch": 0.91, "grad_norm": 2.977454204369908, "learning_rate": 2.1225384273492522e-07, "loss": 0.1285, "step": 12686 }, { "epoch": 0.91, "grad_norm": 13.901014304716377, "learning_rate": 2.1191919949884067e-07, "loss": 0.5772, "step": 12687 }, { "epoch": 0.91, "grad_norm": 3.6974507959415575, "learning_rate": 2.1158481456159008e-07, "loss": 0.1722, "step": 12688 }, { "epoch": 0.91, "grad_norm": 2.8456680615333103, "learning_rate": 2.112506879412146e-07, "loss": 0.1166, "step": 12689 }, { "epoch": 0.91, "grad_norm": 2.4881140574666127, "learning_rate": 2.109168196557382e-07, "loss": 0.0955, "step": 12690 }, { "epoch": 0.91, "grad_norm": 2.82922925549843, "learning_rate": 2.105832097231708e-07, "loss": 0.1421, "step": 12691 }, { "epoch": 0.91, "grad_norm": 3.2441938924660447, "learning_rate": 2.1024985816151146e-07, "loss": 0.1445, "step": 12692 }, { "epoch": 0.91, "grad_norm": 10.072373736704959, "learning_rate": 2.0991676498874125e-07, "loss": 0.393, "step": 12693 }, { "epoch": 0.91, "grad_norm": 4.335519330823689, "learning_rate": 2.0958393022283085e-07, "loss": 0.7293, "step": 12694 }, { "epoch": 0.91, "grad_norm": 9.618546218190122, "learning_rate": 2.0925135388173367e-07, "loss": 0.467, "step": 12695 }, { "epoch": 0.91, "grad_norm": 7.373879269311964, "learning_rate": 2.0891903598339314e-07, "loss": 0.3402, "step": 12696 }, { "epoch": 0.91, "grad_norm": 3.2648137736561633, "learning_rate": 2.085869765457349e-07, "loss": 0.2414, "step": 12697 }, { "epoch": 0.91, "grad_norm": 15.45083821403892, "learning_rate": 2.0825517558667297e-07, "loss": 0.474, "step": 12698 }, { "epoch": 0.91, "grad_norm": 17.181106545824044, "learning_rate": 2.0792363312410636e-07, "loss": 0.8364, "step": 12699 }, { "epoch": 0.91, "grad_norm": 3.5936417761976673, "learning_rate": 2.0759234917592019e-07, "loss": 0.2277, "step": 12700 }, { "epoch": 0.91, "grad_norm": 15.6971364233362, "learning_rate": 2.0726132375998742e-07, "loss": 0.5867, "step": 12701 }, { "epoch": 0.91, "grad_norm": 9.922832035328238, "learning_rate": 2.0693055689416431e-07, "loss": 0.4751, "step": 12702 }, { "epoch": 0.91, "grad_norm": 13.890892598692245, "learning_rate": 2.0660004859629544e-07, "loss": 0.8071, "step": 12703 }, { "epoch": 0.91, "grad_norm": 3.4500849641304527, "learning_rate": 2.0626979888420884e-07, "loss": 0.1025, "step": 12704 }, { "epoch": 0.91, "grad_norm": 2.814598375608968, "learning_rate": 2.0593980777572243e-07, "loss": 0.1143, "step": 12705 }, { "epoch": 0.91, "grad_norm": 17.271574132370755, "learning_rate": 2.056100752886364e-07, "loss": 0.7964, "step": 12706 }, { "epoch": 0.91, "grad_norm": 2.5879412152051513, "learning_rate": 2.052806014407388e-07, "loss": 0.0844, "step": 12707 }, { "epoch": 0.91, "grad_norm": 3.595409713818144, "learning_rate": 2.0495138624980427e-07, "loss": 0.1865, "step": 12708 }, { "epoch": 0.91, "grad_norm": 12.07904332267769, "learning_rate": 2.0462242973359193e-07, "loss": 0.7163, "step": 12709 }, { "epoch": 0.91, "grad_norm": 2.9310391808849605, "learning_rate": 2.0429373190984814e-07, "loss": 0.1228, "step": 12710 }, { "epoch": 0.91, "grad_norm": 11.950711701621088, "learning_rate": 2.039652927963054e-07, "loss": 0.6711, "step": 12711 }, { "epoch": 0.91, "grad_norm": 2.674944442592895, "learning_rate": 2.0363711241068064e-07, "loss": 0.1242, "step": 12712 }, { "epoch": 0.91, "grad_norm": 13.848322051523768, "learning_rate": 2.0330919077067747e-07, "loss": 0.4787, "step": 12713 }, { "epoch": 0.91, "grad_norm": 3.805927582086556, "learning_rate": 2.0298152789398784e-07, "loss": 0.2852, "step": 12714 }, { "epoch": 0.91, "grad_norm": 5.530466458449905, "learning_rate": 2.0265412379828764e-07, "loss": 0.1751, "step": 12715 }, { "epoch": 0.91, "grad_norm": 2.4926932672453614, "learning_rate": 2.0232697850123718e-07, "loss": 0.1281, "step": 12716 }, { "epoch": 0.91, "grad_norm": 4.688857036145698, "learning_rate": 2.0200009202048676e-07, "loss": 1.0541, "step": 12717 }, { "epoch": 0.91, "grad_norm": 4.260892769325777, "learning_rate": 2.0167346437367065e-07, "loss": 0.7838, "step": 12718 }, { "epoch": 0.91, "grad_norm": 2.352023718055542, "learning_rate": 2.0134709557840691e-07, "loss": 0.1192, "step": 12719 }, { "epoch": 0.91, "grad_norm": 4.27420260232721, "learning_rate": 2.0102098565230487e-07, "loss": 0.9879, "step": 12720 }, { "epoch": 0.91, "grad_norm": 2.964170594985151, "learning_rate": 2.0069513461295486e-07, "loss": 0.1462, "step": 12721 }, { "epoch": 0.91, "grad_norm": 3.0017202714341518, "learning_rate": 2.0036954247793672e-07, "loss": 0.1722, "step": 12722 }, { "epoch": 0.91, "grad_norm": 15.722477797555214, "learning_rate": 2.000442092648136e-07, "loss": 0.7479, "step": 12723 }, { "epoch": 0.91, "grad_norm": 4.276767690159596, "learning_rate": 1.9971913499113705e-07, "loss": 0.197, "step": 12724 }, { "epoch": 0.91, "grad_norm": 4.958137881976813, "learning_rate": 1.993943196744419e-07, "loss": 0.9097, "step": 12725 }, { "epoch": 0.91, "grad_norm": 3.3167853258228086, "learning_rate": 1.9906976333225302e-07, "loss": 0.1388, "step": 12726 }, { "epoch": 0.91, "grad_norm": 3.3223490746179247, "learning_rate": 1.9874546598207754e-07, "loss": 0.1711, "step": 12727 }, { "epoch": 0.91, "grad_norm": 2.738582455715516, "learning_rate": 1.9842142764141037e-07, "loss": 0.0783, "step": 12728 }, { "epoch": 0.91, "grad_norm": 2.920936659461116, "learning_rate": 1.9809764832773248e-07, "loss": 0.0861, "step": 12729 }, { "epoch": 0.91, "grad_norm": 10.672601314002437, "learning_rate": 1.9777412805850993e-07, "loss": 0.5468, "step": 12730 }, { "epoch": 0.91, "grad_norm": 3.436767941332311, "learning_rate": 1.9745086685119596e-07, "loss": 0.1757, "step": 12731 }, { "epoch": 0.91, "grad_norm": 19.764906532297033, "learning_rate": 1.9712786472322886e-07, "loss": 0.569, "step": 12732 }, { "epoch": 0.91, "grad_norm": 2.65421585366441, "learning_rate": 1.9680512169203413e-07, "loss": 0.0979, "step": 12733 }, { "epoch": 0.91, "grad_norm": 14.769679379923389, "learning_rate": 1.9648263777502173e-07, "loss": 0.6775, "step": 12734 }, { "epoch": 0.91, "grad_norm": 3.3532935670410464, "learning_rate": 1.9616041298958887e-07, "loss": 0.1627, "step": 12735 }, { "epoch": 0.91, "grad_norm": 3.0531965268834176, "learning_rate": 1.958384473531183e-07, "loss": 0.1719, "step": 12736 }, { "epoch": 0.91, "grad_norm": 20.700003460853946, "learning_rate": 1.955167408829778e-07, "loss": 0.7991, "step": 12737 }, { "epoch": 0.91, "grad_norm": 37.40858379997267, "learning_rate": 1.95195293596524e-07, "loss": 0.5546, "step": 12738 }, { "epoch": 0.91, "grad_norm": 2.9811907036281697, "learning_rate": 1.9487410551109697e-07, "loss": 0.0953, "step": 12739 }, { "epoch": 0.91, "grad_norm": 14.774646832692593, "learning_rate": 1.945531766440234e-07, "loss": 0.5594, "step": 12740 }, { "epoch": 0.91, "grad_norm": 25.223101425222453, "learning_rate": 1.9423250701261608e-07, "loss": 0.7516, "step": 12741 }, { "epoch": 0.91, "grad_norm": 2.6287609297139727, "learning_rate": 1.9391209663417455e-07, "loss": 0.0903, "step": 12742 }, { "epoch": 0.91, "grad_norm": 2.8062066672018626, "learning_rate": 1.9359194552598326e-07, "loss": 0.0923, "step": 12743 }, { "epoch": 0.91, "grad_norm": 3.2289995584234057, "learning_rate": 1.9327205370531287e-07, "loss": 0.1399, "step": 12744 }, { "epoch": 0.91, "grad_norm": 4.951967334058101, "learning_rate": 1.9295242118942125e-07, "loss": 0.7593, "step": 12745 }, { "epoch": 0.91, "grad_norm": 8.535080534099592, "learning_rate": 1.9263304799555127e-07, "loss": 0.4582, "step": 12746 }, { "epoch": 0.91, "grad_norm": 4.351806921681064, "learning_rate": 1.923139341409308e-07, "loss": 0.8425, "step": 12747 }, { "epoch": 0.91, "grad_norm": 2.8356299076405476, "learning_rate": 1.9199507964277552e-07, "loss": 0.1534, "step": 12748 }, { "epoch": 0.91, "grad_norm": 12.780138526675785, "learning_rate": 1.9167648451828669e-07, "loss": 0.779, "step": 12749 }, { "epoch": 0.91, "grad_norm": 4.838350781613299, "learning_rate": 1.9135814878465053e-07, "loss": 0.9353, "step": 12750 }, { "epoch": 0.91, "grad_norm": 15.478765189075892, "learning_rate": 1.910400724590411e-07, "loss": 0.714, "step": 12751 }, { "epoch": 0.91, "grad_norm": 5.131323184393199, "learning_rate": 1.907222555586169e-07, "loss": 0.9281, "step": 12752 }, { "epoch": 0.91, "grad_norm": 4.004484029766244, "learning_rate": 1.9040469810052199e-07, "loss": 0.6819, "step": 12753 }, { "epoch": 0.91, "grad_norm": 3.1267532877868724, "learning_rate": 1.900874001018893e-07, "loss": 0.1183, "step": 12754 }, { "epoch": 0.91, "grad_norm": 2.956741168029429, "learning_rate": 1.8977036157983463e-07, "loss": 0.0955, "step": 12755 }, { "epoch": 0.91, "grad_norm": 4.80237206747951, "learning_rate": 1.894535825514604e-07, "loss": 0.8755, "step": 12756 }, { "epoch": 0.91, "grad_norm": 3.3482377542502233, "learning_rate": 1.8913706303385736e-07, "loss": 0.2814, "step": 12757 }, { "epoch": 0.91, "grad_norm": 12.731551018303724, "learning_rate": 1.888208030441002e-07, "loss": 0.6232, "step": 12758 }, { "epoch": 0.91, "grad_norm": 10.079251964697287, "learning_rate": 1.8850480259924863e-07, "loss": 0.514, "step": 12759 }, { "epoch": 0.91, "grad_norm": 14.050059116746679, "learning_rate": 1.8818906171635064e-07, "loss": 0.4922, "step": 12760 }, { "epoch": 0.92, "grad_norm": 3.182985761833646, "learning_rate": 1.878735804124393e-07, "loss": 0.1473, "step": 12761 }, { "epoch": 0.92, "grad_norm": 3.914541071252888, "learning_rate": 1.8755835870453264e-07, "loss": 0.1902, "step": 12762 }, { "epoch": 0.92, "grad_norm": 4.883735476602394, "learning_rate": 1.8724339660963707e-07, "loss": 0.8246, "step": 12763 }, { "epoch": 0.92, "grad_norm": 20.86851395202052, "learning_rate": 1.8692869414474292e-07, "loss": 0.6664, "step": 12764 }, { "epoch": 0.92, "grad_norm": 18.717227710645307, "learning_rate": 1.8661425132682654e-07, "loss": 0.6896, "step": 12765 }, { "epoch": 0.92, "grad_norm": 14.361937428003033, "learning_rate": 1.8630006817285275e-07, "loss": 0.6384, "step": 12766 }, { "epoch": 0.92, "grad_norm": 4.502074740366812, "learning_rate": 1.8598614469976904e-07, "loss": 0.1423, "step": 12767 }, { "epoch": 0.92, "grad_norm": 11.422035202866098, "learning_rate": 1.8567248092451086e-07, "loss": 0.4021, "step": 12768 }, { "epoch": 0.92, "grad_norm": 4.54384442104577, "learning_rate": 1.8535907686399846e-07, "loss": 0.6584, "step": 12769 }, { "epoch": 0.92, "grad_norm": 13.390342038563878, "learning_rate": 1.8504593253514059e-07, "loss": 0.6301, "step": 12770 }, { "epoch": 0.92, "grad_norm": 4.050118317459008, "learning_rate": 1.8473304795482816e-07, "loss": 0.251, "step": 12771 }, { "epoch": 0.92, "grad_norm": 24.07166323011159, "learning_rate": 1.8442042313994157e-07, "loss": 1.5213, "step": 12772 }, { "epoch": 0.92, "grad_norm": 11.087207195376896, "learning_rate": 1.8410805810734512e-07, "loss": 0.8333, "step": 12773 }, { "epoch": 0.92, "grad_norm": 4.3484317003806305, "learning_rate": 1.8379595287388974e-07, "loss": 0.3531, "step": 12774 }, { "epoch": 0.92, "grad_norm": 4.207630479280456, "learning_rate": 1.8348410745641144e-07, "loss": 0.7456, "step": 12775 }, { "epoch": 0.92, "grad_norm": 31.811212696678563, "learning_rate": 1.8317252187173505e-07, "loss": 1.1537, "step": 12776 }, { "epoch": 0.92, "grad_norm": 5.531460795333346, "learning_rate": 1.828611961366683e-07, "loss": 0.6778, "step": 12777 }, { "epoch": 0.92, "grad_norm": 13.958612790179393, "learning_rate": 1.8255013026800607e-07, "loss": 0.6815, "step": 12778 }, { "epoch": 0.92, "grad_norm": 3.70734469320083, "learning_rate": 1.8223932428252933e-07, "loss": 0.1886, "step": 12779 }, { "epoch": 0.92, "grad_norm": 4.721151543515703, "learning_rate": 1.8192877819700473e-07, "loss": 0.9354, "step": 12780 }, { "epoch": 0.92, "grad_norm": 2.9200152268791255, "learning_rate": 1.8161849202818493e-07, "loss": 0.1553, "step": 12781 }, { "epoch": 0.92, "grad_norm": 16.78498120873803, "learning_rate": 1.813084657928099e-07, "loss": 0.8068, "step": 12782 }, { "epoch": 0.92, "grad_norm": 9.376288216132389, "learning_rate": 1.8099869950760295e-07, "loss": 0.16, "step": 12783 }, { "epoch": 0.92, "grad_norm": 2.7756190765136193, "learning_rate": 1.806891931892746e-07, "loss": 0.1309, "step": 12784 }, { "epoch": 0.92, "grad_norm": 3.234648513221352, "learning_rate": 1.8037994685452254e-07, "loss": 0.1376, "step": 12785 }, { "epoch": 0.92, "grad_norm": 4.956100443010996, "learning_rate": 1.800709605200296e-07, "loss": 0.8223, "step": 12786 }, { "epoch": 0.92, "grad_norm": 3.5991556971588987, "learning_rate": 1.7976223420246297e-07, "loss": 0.1495, "step": 12787 }, { "epoch": 0.92, "grad_norm": 2.8971647448408673, "learning_rate": 1.794537679184788e-07, "loss": 0.1282, "step": 12788 }, { "epoch": 0.92, "grad_norm": 17.893774046020916, "learning_rate": 1.7914556168471709e-07, "loss": 0.6074, "step": 12789 }, { "epoch": 0.92, "grad_norm": 3.463226335374661, "learning_rate": 1.7883761551780398e-07, "loss": 0.124, "step": 12790 }, { "epoch": 0.92, "grad_norm": 4.49177112866331, "learning_rate": 1.785299294343529e-07, "loss": 0.8094, "step": 12791 }, { "epoch": 0.92, "grad_norm": 16.88172820400989, "learning_rate": 1.782225034509616e-07, "loss": 1.0224, "step": 12792 }, { "epoch": 0.92, "grad_norm": 5.28000042838768, "learning_rate": 1.7791533758421463e-07, "loss": 0.8341, "step": 12793 }, { "epoch": 0.92, "grad_norm": 3.593046730556484, "learning_rate": 1.7760843185068378e-07, "loss": 0.1773, "step": 12794 }, { "epoch": 0.92, "grad_norm": 16.12587461912664, "learning_rate": 1.773017862669235e-07, "loss": 0.42, "step": 12795 }, { "epoch": 0.92, "grad_norm": 18.184448399756803, "learning_rate": 1.7699540084947674e-07, "loss": 0.571, "step": 12796 }, { "epoch": 0.92, "grad_norm": 2.2915726716019886, "learning_rate": 1.7668927561487193e-07, "loss": 0.0721, "step": 12797 }, { "epoch": 0.92, "grad_norm": 3.3906731023261143, "learning_rate": 1.7638341057962415e-07, "loss": 0.1511, "step": 12798 }, { "epoch": 0.92, "grad_norm": 17.449547733515715, "learning_rate": 1.7607780576023193e-07, "loss": 0.3141, "step": 12799 }, { "epoch": 0.92, "grad_norm": 7.588579871808047, "learning_rate": 1.757724611731837e-07, "loss": 0.4578, "step": 12800 }, { "epoch": 0.92, "grad_norm": 3.0998372558724814, "learning_rate": 1.754673768349502e-07, "loss": 0.1609, "step": 12801 }, { "epoch": 0.92, "grad_norm": 3.5326617952115043, "learning_rate": 1.7516255276198934e-07, "loss": 0.1663, "step": 12802 }, { "epoch": 0.92, "grad_norm": 3.848263962058969, "learning_rate": 1.748579889707458e-07, "loss": 0.7145, "step": 12803 }, { "epoch": 0.92, "grad_norm": 17.321846078682665, "learning_rate": 1.7455368547765028e-07, "loss": 1.2272, "step": 12804 }, { "epoch": 0.92, "grad_norm": 11.887964838063176, "learning_rate": 1.74249642299118e-07, "loss": 0.5831, "step": 12805 }, { "epoch": 0.92, "grad_norm": 4.654089261768295, "learning_rate": 1.7394585945155083e-07, "loss": 0.7724, "step": 12806 }, { "epoch": 0.92, "grad_norm": 8.475642690776226, "learning_rate": 1.736423369513368e-07, "loss": 0.3304, "step": 12807 }, { "epoch": 0.92, "grad_norm": 2.7608730643768933, "learning_rate": 1.7333907481484947e-07, "loss": 0.1454, "step": 12808 }, { "epoch": 0.92, "grad_norm": 4.563303814896455, "learning_rate": 1.7303607305845016e-07, "loss": 0.8724, "step": 12809 }, { "epoch": 0.92, "grad_norm": 3.4920500449221312, "learning_rate": 1.7273333169848306e-07, "loss": 0.1456, "step": 12810 }, { "epoch": 0.92, "grad_norm": 2.0394180047738804, "learning_rate": 1.724308507512812e-07, "loss": 0.0945, "step": 12811 }, { "epoch": 0.92, "grad_norm": 2.727408396279956, "learning_rate": 1.721286302331604e-07, "loss": 0.1887, "step": 12812 }, { "epoch": 0.92, "grad_norm": 13.969330785088484, "learning_rate": 1.7182667016042655e-07, "loss": 0.6872, "step": 12813 }, { "epoch": 0.92, "grad_norm": 12.066673547350474, "learning_rate": 1.7152497054936767e-07, "loss": 0.3756, "step": 12814 }, { "epoch": 0.92, "grad_norm": 5.8619082404973915, "learning_rate": 1.712235314162597e-07, "loss": 0.8114, "step": 12815 }, { "epoch": 0.92, "grad_norm": 3.218903474337861, "learning_rate": 1.7092235277736513e-07, "loss": 0.254, "step": 12816 }, { "epoch": 0.92, "grad_norm": 10.651718663281448, "learning_rate": 1.7062143464892988e-07, "loss": 0.6006, "step": 12817 }, { "epoch": 0.92, "grad_norm": 18.775640465983543, "learning_rate": 1.7032077704718875e-07, "loss": 0.8158, "step": 12818 }, { "epoch": 0.92, "grad_norm": 3.986567551748615, "learning_rate": 1.7002037998835985e-07, "loss": 0.1265, "step": 12819 }, { "epoch": 0.92, "grad_norm": 4.442689271842332, "learning_rate": 1.6972024348864912e-07, "loss": 0.8635, "step": 12820 }, { "epoch": 0.92, "grad_norm": 3.0965176258842066, "learning_rate": 1.6942036756424695e-07, "loss": 0.0954, "step": 12821 }, { "epoch": 0.92, "grad_norm": 5.4270675813885365, "learning_rate": 1.6912075223133205e-07, "loss": 0.7456, "step": 12822 }, { "epoch": 0.92, "grad_norm": 3.9486937744780146, "learning_rate": 1.6882139750606652e-07, "loss": 0.1275, "step": 12823 }, { "epoch": 0.92, "grad_norm": 26.887152400596506, "learning_rate": 1.6852230340459907e-07, "loss": 0.6906, "step": 12824 }, { "epoch": 0.92, "grad_norm": 3.5881598218731856, "learning_rate": 1.6822346994306572e-07, "loss": 0.181, "step": 12825 }, { "epoch": 0.92, "grad_norm": 4.6743857756370355, "learning_rate": 1.679248971375874e-07, "loss": 0.3436, "step": 12826 }, { "epoch": 0.92, "grad_norm": 4.783860562518035, "learning_rate": 1.676265850042691e-07, "loss": 0.8725, "step": 12827 }, { "epoch": 0.92, "grad_norm": 20.20038039068971, "learning_rate": 1.673285335592062e-07, "loss": 0.8597, "step": 12828 }, { "epoch": 0.92, "grad_norm": 5.695067466519982, "learning_rate": 1.6703074281847587e-07, "loss": 0.7799, "step": 12829 }, { "epoch": 0.92, "grad_norm": 5.117751298535736, "learning_rate": 1.6673321279814357e-07, "loss": 0.7124, "step": 12830 }, { "epoch": 0.92, "grad_norm": 6.343048216407667, "learning_rate": 1.664359435142593e-07, "loss": 0.3441, "step": 12831 }, { "epoch": 0.92, "grad_norm": 2.772336160329683, "learning_rate": 1.661389349828596e-07, "loss": 0.1711, "step": 12832 }, { "epoch": 0.92, "grad_norm": 1.6043737920056176, "learning_rate": 1.6584218721996726e-07, "loss": 0.0754, "step": 12833 }, { "epoch": 0.92, "grad_norm": 12.69352099264146, "learning_rate": 1.6554570024159057e-07, "loss": 0.5885, "step": 12834 }, { "epoch": 0.92, "grad_norm": 3.8181524908692923, "learning_rate": 1.6524947406372448e-07, "loss": 0.1892, "step": 12835 }, { "epoch": 0.92, "grad_norm": 21.364078739721588, "learning_rate": 1.6495350870234794e-07, "loss": 1.0672, "step": 12836 }, { "epoch": 0.92, "grad_norm": 11.428449784529834, "learning_rate": 1.6465780417342814e-07, "loss": 0.5597, "step": 12837 }, { "epoch": 0.92, "grad_norm": 30.048279491150844, "learning_rate": 1.643623604929173e-07, "loss": 0.4443, "step": 12838 }, { "epoch": 0.92, "grad_norm": 3.3714358992363467, "learning_rate": 1.6406717767675385e-07, "loss": 0.2087, "step": 12839 }, { "epoch": 0.92, "grad_norm": 3.7708971540088463, "learning_rate": 1.6377225574085998e-07, "loss": 0.2401, "step": 12840 }, { "epoch": 0.92, "grad_norm": 2.7896930320352786, "learning_rate": 1.6347759470114744e-07, "loss": 0.1291, "step": 12841 }, { "epoch": 0.92, "grad_norm": 18.443832408953117, "learning_rate": 1.631831945735113e-07, "loss": 0.4873, "step": 12842 }, { "epoch": 0.92, "grad_norm": 12.847590379399113, "learning_rate": 1.6288905537383382e-07, "loss": 0.4115, "step": 12843 }, { "epoch": 0.92, "grad_norm": 2.9115036677130988, "learning_rate": 1.6259517711798178e-07, "loss": 0.1711, "step": 12844 }, { "epoch": 0.92, "grad_norm": 2.8333689067482997, "learning_rate": 1.6230155982180916e-07, "loss": 0.131, "step": 12845 }, { "epoch": 0.92, "grad_norm": 1.384781580909973, "learning_rate": 1.6200820350115665e-07, "loss": 0.0486, "step": 12846 }, { "epoch": 0.92, "grad_norm": 3.0889683754742565, "learning_rate": 1.617151081718482e-07, "loss": 0.1471, "step": 12847 }, { "epoch": 0.92, "grad_norm": 2.819328000445926, "learning_rate": 1.6142227384969566e-07, "loss": 0.0965, "step": 12848 }, { "epoch": 0.92, "grad_norm": 3.612818471158618, "learning_rate": 1.6112970055049636e-07, "loss": 0.2408, "step": 12849 }, { "epoch": 0.92, "grad_norm": 2.887520028028182, "learning_rate": 1.6083738829003383e-07, "loss": 0.1328, "step": 12850 }, { "epoch": 0.92, "grad_norm": 6.919675215327488, "learning_rate": 1.6054533708407704e-07, "loss": 0.324, "step": 12851 }, { "epoch": 0.92, "grad_norm": 3.0805898785952874, "learning_rate": 1.602535469483807e-07, "loss": 0.1467, "step": 12852 }, { "epoch": 0.92, "grad_norm": 12.202636125298051, "learning_rate": 1.599620178986866e-07, "loss": 0.6193, "step": 12853 }, { "epoch": 0.92, "grad_norm": 16.49757696203467, "learning_rate": 1.5967074995072107e-07, "loss": 0.7299, "step": 12854 }, { "epoch": 0.92, "grad_norm": 3.934070390443579, "learning_rate": 1.5937974312019655e-07, "loss": 0.2627, "step": 12855 }, { "epoch": 0.92, "grad_norm": 16.08412395470947, "learning_rate": 1.5908899742281213e-07, "loss": 0.4152, "step": 12856 }, { "epoch": 0.92, "grad_norm": 11.84662787350699, "learning_rate": 1.587985128742531e-07, "loss": 0.6338, "step": 12857 }, { "epoch": 0.92, "grad_norm": 2.8688337147852163, "learning_rate": 1.5850828949018794e-07, "loss": 0.1201, "step": 12858 }, { "epoch": 0.92, "grad_norm": 19.220357292028943, "learning_rate": 1.5821832728627584e-07, "loss": 0.7237, "step": 12859 }, { "epoch": 0.92, "grad_norm": 3.275965953227884, "learning_rate": 1.5792862627815763e-07, "loss": 0.2122, "step": 12860 }, { "epoch": 0.92, "grad_norm": 25.800361030882453, "learning_rate": 1.5763918648146082e-07, "loss": 0.6991, "step": 12861 }, { "epoch": 0.92, "grad_norm": 17.959328963178457, "learning_rate": 1.5735000791180176e-07, "loss": 0.4361, "step": 12862 }, { "epoch": 0.92, "grad_norm": 20.24989103066656, "learning_rate": 1.5706109058477915e-07, "loss": 0.4013, "step": 12863 }, { "epoch": 0.92, "grad_norm": 2.1880935369926875, "learning_rate": 1.5677243451597823e-07, "loss": 0.0592, "step": 12864 }, { "epoch": 0.92, "grad_norm": 7.199457138705029, "learning_rate": 1.564840397209727e-07, "loss": 0.4106, "step": 12865 }, { "epoch": 0.92, "grad_norm": 4.154833266583721, "learning_rate": 1.5619590621531954e-07, "loss": 0.9941, "step": 12866 }, { "epoch": 0.92, "grad_norm": 3.103692429355203, "learning_rate": 1.5590803401456244e-07, "loss": 0.1566, "step": 12867 }, { "epoch": 0.92, "grad_norm": 4.0733570741956, "learning_rate": 1.556204231342312e-07, "loss": 0.8294, "step": 12868 }, { "epoch": 0.92, "grad_norm": 12.909029317413564, "learning_rate": 1.5533307358984117e-07, "loss": 0.7107, "step": 12869 }, { "epoch": 0.92, "grad_norm": 4.612250465548584, "learning_rate": 1.5504598539689274e-07, "loss": 0.8405, "step": 12870 }, { "epoch": 0.92, "grad_norm": 16.398796846037616, "learning_rate": 1.547591585708752e-07, "loss": 0.7005, "step": 12871 }, { "epoch": 0.92, "grad_norm": 2.747400969540252, "learning_rate": 1.544725931272606e-07, "loss": 0.094, "step": 12872 }, { "epoch": 0.92, "grad_norm": 2.8556883287788373, "learning_rate": 1.5418628908150767e-07, "loss": 0.102, "step": 12873 }, { "epoch": 0.92, "grad_norm": 6.897192167357113, "learning_rate": 1.5390024644906297e-07, "loss": 0.2722, "step": 12874 }, { "epoch": 0.92, "grad_norm": 4.091614897788435, "learning_rate": 1.536144652453564e-07, "loss": 0.729, "step": 12875 }, { "epoch": 0.92, "grad_norm": 1.5149285393596281, "learning_rate": 1.5332894548580447e-07, "loss": 0.0779, "step": 12876 }, { "epoch": 0.92, "grad_norm": 10.775834658000829, "learning_rate": 1.5304368718580987e-07, "loss": 0.4729, "step": 12877 }, { "epoch": 0.92, "grad_norm": 4.369097688692362, "learning_rate": 1.5275869036076251e-07, "loss": 0.7672, "step": 12878 }, { "epoch": 0.92, "grad_norm": 18.80352868750018, "learning_rate": 1.524739550260357e-07, "loss": 0.5968, "step": 12879 }, { "epoch": 0.92, "grad_norm": 11.309444136823645, "learning_rate": 1.521894811969904e-07, "loss": 0.4551, "step": 12880 }, { "epoch": 0.92, "grad_norm": 4.223439510292065, "learning_rate": 1.5190526888897217e-07, "loss": 0.8216, "step": 12881 }, { "epoch": 0.92, "grad_norm": 2.877679490885975, "learning_rate": 1.5162131811731374e-07, "loss": 0.1188, "step": 12882 }, { "epoch": 0.92, "grad_norm": 38.197563280132016, "learning_rate": 1.513376288973334e-07, "loss": 0.772, "step": 12883 }, { "epoch": 0.92, "grad_norm": 3.3386583065575826, "learning_rate": 1.5105420124433445e-07, "loss": 0.1546, "step": 12884 }, { "epoch": 0.92, "grad_norm": 19.789489850947326, "learning_rate": 1.5077103517360746e-07, "loss": 1.0333, "step": 12885 }, { "epoch": 0.92, "grad_norm": 4.029952463619769, "learning_rate": 1.5048813070042688e-07, "loss": 0.1442, "step": 12886 }, { "epoch": 0.92, "grad_norm": 4.011592595265739, "learning_rate": 1.5020548784005608e-07, "loss": 0.1712, "step": 12887 }, { "epoch": 0.92, "grad_norm": 3.33593608907633, "learning_rate": 1.4992310660774167e-07, "loss": 0.1254, "step": 12888 }, { "epoch": 0.92, "grad_norm": 18.66602943757234, "learning_rate": 1.4964098701871653e-07, "loss": 0.6001, "step": 12889 }, { "epoch": 0.92, "grad_norm": 3.212043817552855, "learning_rate": 1.4935912908820128e-07, "loss": 0.1697, "step": 12890 }, { "epoch": 0.92, "grad_norm": 4.605804352729818, "learning_rate": 1.490775328314009e-07, "loss": 0.2571, "step": 12891 }, { "epoch": 0.92, "grad_norm": 3.7225209922697355, "learning_rate": 1.4879619826350444e-07, "loss": 0.3114, "step": 12892 }, { "epoch": 0.92, "grad_norm": 3.008073944335045, "learning_rate": 1.4851512539969138e-07, "loss": 0.1404, "step": 12893 }, { "epoch": 0.92, "grad_norm": 3.040302120780481, "learning_rate": 1.482343142551229e-07, "loss": 0.1662, "step": 12894 }, { "epoch": 0.92, "grad_norm": 3.210654110198743, "learning_rate": 1.4795376484494805e-07, "loss": 0.0699, "step": 12895 }, { "epoch": 0.92, "grad_norm": 3.765634486859832, "learning_rate": 1.4767347718430191e-07, "loss": 0.1346, "step": 12896 }, { "epoch": 0.92, "grad_norm": 9.57328935736722, "learning_rate": 1.4739345128830463e-07, "loss": 0.4724, "step": 12897 }, { "epoch": 0.92, "grad_norm": 2.116467116366418, "learning_rate": 1.4711368717206186e-07, "loss": 0.0801, "step": 12898 }, { "epoch": 0.92, "grad_norm": 4.250737567839699, "learning_rate": 1.4683418485066715e-07, "loss": 0.1791, "step": 12899 }, { "epoch": 0.93, "grad_norm": 2.455570073291168, "learning_rate": 1.4655494433919838e-07, "loss": 0.1282, "step": 12900 }, { "epoch": 0.93, "grad_norm": 4.433744188654156, "learning_rate": 1.4627596565271796e-07, "loss": 0.2771, "step": 12901 }, { "epoch": 0.93, "grad_norm": 17.56613125035712, "learning_rate": 1.4599724880627775e-07, "loss": 0.3397, "step": 12902 }, { "epoch": 0.93, "grad_norm": 3.0236803263073244, "learning_rate": 1.4571879381491293e-07, "loss": 0.1689, "step": 12903 }, { "epoch": 0.93, "grad_norm": 6.6383222739835395, "learning_rate": 1.4544060069364374e-07, "loss": 0.1689, "step": 12904 }, { "epoch": 0.93, "grad_norm": 4.559867643643816, "learning_rate": 1.451626694574787e-07, "loss": 0.8231, "step": 12905 }, { "epoch": 0.93, "grad_norm": 4.403447459613871, "learning_rate": 1.4488500012141195e-07, "loss": 0.1508, "step": 12906 }, { "epoch": 0.93, "grad_norm": 7.9733426668492795, "learning_rate": 1.4460759270042035e-07, "loss": 0.281, "step": 12907 }, { "epoch": 0.93, "grad_norm": 12.232767465090848, "learning_rate": 1.44330447209472e-07, "loss": 0.5059, "step": 12908 }, { "epoch": 0.93, "grad_norm": 2.79567052213876, "learning_rate": 1.4405356366351543e-07, "loss": 0.1365, "step": 12909 }, { "epoch": 0.93, "grad_norm": 3.1752295440681477, "learning_rate": 1.4377694207748817e-07, "loss": 0.1165, "step": 12910 }, { "epoch": 0.93, "grad_norm": 8.200697561425315, "learning_rate": 1.4350058246631382e-07, "loss": 0.4022, "step": 12911 }, { "epoch": 0.93, "grad_norm": 4.4802287027690095, "learning_rate": 1.4322448484489993e-07, "loss": 0.6465, "step": 12912 }, { "epoch": 0.93, "grad_norm": 3.410205865407241, "learning_rate": 1.4294864922814122e-07, "loss": 0.0991, "step": 12913 }, { "epoch": 0.93, "grad_norm": 2.4558110258461263, "learning_rate": 1.4267307563091747e-07, "loss": 0.1123, "step": 12914 }, { "epoch": 0.93, "grad_norm": 14.212519985557625, "learning_rate": 1.423977640680968e-07, "loss": 0.3305, "step": 12915 }, { "epoch": 0.93, "grad_norm": 3.4902717698671712, "learning_rate": 1.4212271455452786e-07, "loss": 0.1939, "step": 12916 }, { "epoch": 0.93, "grad_norm": 4.364002812244834, "learning_rate": 1.4184792710505158e-07, "loss": 0.8324, "step": 12917 }, { "epoch": 0.93, "grad_norm": 3.0611549596477734, "learning_rate": 1.4157340173448997e-07, "loss": 0.1454, "step": 12918 }, { "epoch": 0.93, "grad_norm": 3.3834053575582477, "learning_rate": 1.4129913845765396e-07, "loss": 0.2506, "step": 12919 }, { "epoch": 0.93, "grad_norm": 4.445614639734738, "learning_rate": 1.4102513728933677e-07, "loss": 0.7783, "step": 12920 }, { "epoch": 0.93, "grad_norm": 2.266027112533171, "learning_rate": 1.407513982443226e-07, "loss": 0.1418, "step": 12921 }, { "epoch": 0.93, "grad_norm": 10.561647567337959, "learning_rate": 1.4047792133737693e-07, "loss": 0.4305, "step": 12922 }, { "epoch": 0.93, "grad_norm": 3.3143772816653208, "learning_rate": 1.4020470658325235e-07, "loss": 0.206, "step": 12923 }, { "epoch": 0.93, "grad_norm": 2.8770116973971898, "learning_rate": 1.3993175399668934e-07, "loss": 0.1171, "step": 12924 }, { "epoch": 0.93, "grad_norm": 2.766047431587891, "learning_rate": 1.3965906359241165e-07, "loss": 0.1565, "step": 12925 }, { "epoch": 0.93, "grad_norm": 19.58946244843749, "learning_rate": 1.3938663538512976e-07, "loss": 0.8418, "step": 12926 }, { "epoch": 0.93, "grad_norm": 3.6850235899608657, "learning_rate": 1.391144693895413e-07, "loss": 0.1799, "step": 12927 }, { "epoch": 0.93, "grad_norm": 5.122199437723162, "learning_rate": 1.388425656203274e-07, "loss": 0.7344, "step": 12928 }, { "epoch": 0.93, "grad_norm": 9.272156228320698, "learning_rate": 1.3857092409215567e-07, "loss": 0.4009, "step": 12929 }, { "epoch": 0.93, "grad_norm": 2.5124612017428736, "learning_rate": 1.3829954481968223e-07, "loss": 0.1254, "step": 12930 }, { "epoch": 0.93, "grad_norm": 42.45665699410231, "learning_rate": 1.3802842781754533e-07, "loss": 0.7166, "step": 12931 }, { "epoch": 0.93, "grad_norm": 3.539799761022384, "learning_rate": 1.3775757310037107e-07, "loss": 0.1236, "step": 12932 }, { "epoch": 0.93, "grad_norm": 17.856249983415495, "learning_rate": 1.3748698068277166e-07, "loss": 0.9826, "step": 12933 }, { "epoch": 0.93, "grad_norm": 11.35759246777353, "learning_rate": 1.3721665057934375e-07, "loss": 0.4266, "step": 12934 }, { "epoch": 0.93, "grad_norm": 10.923619412594272, "learning_rate": 1.369465828046701e-07, "loss": 0.5637, "step": 12935 }, { "epoch": 0.93, "grad_norm": 14.925959180241978, "learning_rate": 1.3667677737332187e-07, "loss": 0.7601, "step": 12936 }, { "epoch": 0.93, "grad_norm": 3.682862619417845, "learning_rate": 1.3640723429985238e-07, "loss": 0.2332, "step": 12937 }, { "epoch": 0.93, "grad_norm": 3.2110718255751407, "learning_rate": 1.3613795359880278e-07, "loss": 0.1208, "step": 12938 }, { "epoch": 0.93, "grad_norm": 8.50828997823272, "learning_rate": 1.3586893528470036e-07, "loss": 0.5062, "step": 12939 }, { "epoch": 0.93, "grad_norm": 3.378126248538408, "learning_rate": 1.3560017937205682e-07, "loss": 0.2415, "step": 12940 }, { "epoch": 0.93, "grad_norm": 2.9964177826175264, "learning_rate": 1.353316858753706e-07, "loss": 0.1291, "step": 12941 }, { "epoch": 0.93, "grad_norm": 15.539835837934008, "learning_rate": 1.350634548091262e-07, "loss": 0.6721, "step": 12942 }, { "epoch": 0.93, "grad_norm": 7.478716954755643, "learning_rate": 1.3479548618779425e-07, "loss": 0.281, "step": 12943 }, { "epoch": 0.93, "grad_norm": 11.12190083307188, "learning_rate": 1.3452778002582934e-07, "loss": 0.5509, "step": 12944 }, { "epoch": 0.93, "grad_norm": 2.825972011559221, "learning_rate": 1.3426033633767377e-07, "loss": 0.1143, "step": 12945 }, { "epoch": 0.93, "grad_norm": 21.32482702927076, "learning_rate": 1.3399315513775602e-07, "loss": 0.7321, "step": 12946 }, { "epoch": 0.93, "grad_norm": 3.7290046226306677, "learning_rate": 1.3372623644048844e-07, "loss": 0.1937, "step": 12947 }, { "epoch": 0.93, "grad_norm": 3.107450075667746, "learning_rate": 1.3345958026027005e-07, "loss": 0.1089, "step": 12948 }, { "epoch": 0.93, "grad_norm": 4.198418405998958, "learning_rate": 1.3319318661148662e-07, "loss": 0.8025, "step": 12949 }, { "epoch": 0.93, "grad_norm": 3.737208106539486, "learning_rate": 1.3292705550850937e-07, "loss": 0.1389, "step": 12950 }, { "epoch": 0.93, "grad_norm": 4.485416681777706, "learning_rate": 1.3266118696569407e-07, "loss": 0.2659, "step": 12951 }, { "epoch": 0.93, "grad_norm": 4.164322797322174, "learning_rate": 1.3239558099738426e-07, "loss": 0.1903, "step": 12952 }, { "epoch": 0.93, "grad_norm": 3.0358292878975384, "learning_rate": 1.3213023761790788e-07, "loss": 0.1464, "step": 12953 }, { "epoch": 0.93, "grad_norm": 4.494572082960997, "learning_rate": 1.3186515684157907e-07, "loss": 0.8373, "step": 12954 }, { "epoch": 0.93, "grad_norm": 2.1093286513742084, "learning_rate": 1.3160033868269863e-07, "loss": 0.1165, "step": 12955 }, { "epoch": 0.93, "grad_norm": 1.6810163122792527, "learning_rate": 1.3133578315555174e-07, "loss": 0.0657, "step": 12956 }, { "epoch": 0.93, "grad_norm": 12.174407328376171, "learning_rate": 1.310714902744098e-07, "loss": 0.6763, "step": 12957 }, { "epoch": 0.93, "grad_norm": 3.8619854539107945, "learning_rate": 1.3080746005353194e-07, "loss": 0.2458, "step": 12958 }, { "epoch": 0.93, "grad_norm": 3.0801921498425227, "learning_rate": 1.3054369250716071e-07, "loss": 0.1962, "step": 12959 }, { "epoch": 0.93, "grad_norm": 13.069718589387751, "learning_rate": 1.3028018764952465e-07, "loss": 0.7376, "step": 12960 }, { "epoch": 0.93, "grad_norm": 25.97974091800649, "learning_rate": 1.3001694549484023e-07, "loss": 0.7935, "step": 12961 }, { "epoch": 0.93, "grad_norm": 4.865337542292215, "learning_rate": 1.2975396605730773e-07, "loss": 0.9224, "step": 12962 }, { "epoch": 0.93, "grad_norm": 4.69790402045018, "learning_rate": 1.2949124935111356e-07, "loss": 0.8631, "step": 12963 }, { "epoch": 0.93, "grad_norm": 5.130772044281704, "learning_rate": 1.2922879539043086e-07, "loss": 0.2239, "step": 12964 }, { "epoch": 0.93, "grad_norm": 2.900411482752643, "learning_rate": 1.2896660418941775e-07, "loss": 0.0769, "step": 12965 }, { "epoch": 0.93, "grad_norm": 16.932890432658375, "learning_rate": 1.287046757622179e-07, "loss": 0.9028, "step": 12966 }, { "epoch": 0.93, "grad_norm": 3.8158829022087386, "learning_rate": 1.2844301012296278e-07, "loss": 0.1419, "step": 12967 }, { "epoch": 0.93, "grad_norm": 4.38474611817287, "learning_rate": 1.2818160728576722e-07, "loss": 0.8694, "step": 12968 }, { "epoch": 0.93, "grad_norm": 3.9680129990393938, "learning_rate": 1.2792046726473218e-07, "loss": 0.136, "step": 12969 }, { "epoch": 0.93, "grad_norm": 3.61881073247627, "learning_rate": 1.2765959007394692e-07, "loss": 0.1277, "step": 12970 }, { "epoch": 0.93, "grad_norm": 3.2463145746744093, "learning_rate": 1.2739897572748404e-07, "loss": 0.1527, "step": 12971 }, { "epoch": 0.93, "grad_norm": 3.5253095763636106, "learning_rate": 1.2713862423940237e-07, "loss": 0.1221, "step": 12972 }, { "epoch": 0.93, "grad_norm": 17.989648756328986, "learning_rate": 1.268785356237473e-07, "loss": 1.0377, "step": 12973 }, { "epoch": 0.93, "grad_norm": 14.676431800946697, "learning_rate": 1.2661870989454983e-07, "loss": 0.3353, "step": 12974 }, { "epoch": 0.93, "grad_norm": 3.5382396826502553, "learning_rate": 1.2635914706582598e-07, "loss": 0.2465, "step": 12975 }, { "epoch": 0.93, "grad_norm": 2.31606474002253, "learning_rate": 1.2609984715157842e-07, "loss": 0.1011, "step": 12976 }, { "epoch": 0.93, "grad_norm": 10.769674313692146, "learning_rate": 1.2584081016579486e-07, "loss": 0.3044, "step": 12977 }, { "epoch": 0.93, "grad_norm": 3.510886371389053, "learning_rate": 1.2558203612245022e-07, "loss": 0.281, "step": 12978 }, { "epoch": 0.93, "grad_norm": 28.651890153506226, "learning_rate": 1.2532352503550393e-07, "loss": 0.653, "step": 12979 }, { "epoch": 0.93, "grad_norm": 2.2880107815620416, "learning_rate": 1.250652769189026e-07, "loss": 0.0564, "step": 12980 }, { "epoch": 0.93, "grad_norm": 3.4859999194292333, "learning_rate": 1.2480729178657558e-07, "loss": 0.1781, "step": 12981 }, { "epoch": 0.93, "grad_norm": 3.931953746917364, "learning_rate": 1.2454956965244236e-07, "loss": 0.2024, "step": 12982 }, { "epoch": 0.93, "grad_norm": 3.4314107369786795, "learning_rate": 1.242921105304057e-07, "loss": 0.1998, "step": 12983 }, { "epoch": 0.93, "grad_norm": 4.633812219028004, "learning_rate": 1.2403491443435332e-07, "loss": 0.947, "step": 12984 }, { "epoch": 0.93, "grad_norm": 3.448649458699727, "learning_rate": 1.2377798137816087e-07, "loss": 0.1281, "step": 12985 }, { "epoch": 0.93, "grad_norm": 4.925912461999226, "learning_rate": 1.2352131137568945e-07, "loss": 0.8224, "step": 12986 }, { "epoch": 0.93, "grad_norm": 3.1865552232909287, "learning_rate": 1.2326490444078465e-07, "loss": 0.1322, "step": 12987 }, { "epoch": 0.93, "grad_norm": 3.254105767639242, "learning_rate": 1.2300876058727874e-07, "loss": 0.158, "step": 12988 }, { "epoch": 0.93, "grad_norm": 4.956451916027988, "learning_rate": 1.2275287982898953e-07, "loss": 0.8658, "step": 12989 }, { "epoch": 0.93, "grad_norm": 3.944165606803622, "learning_rate": 1.2249726217972047e-07, "loss": 0.1252, "step": 12990 }, { "epoch": 0.93, "grad_norm": 73.70744996782598, "learning_rate": 1.2224190765326272e-07, "loss": 0.7366, "step": 12991 }, { "epoch": 0.93, "grad_norm": 3.835891807599332, "learning_rate": 1.219868162633908e-07, "loss": 0.1372, "step": 12992 }, { "epoch": 0.93, "grad_norm": 4.31426393471822, "learning_rate": 1.2173198802386542e-07, "loss": 0.3028, "step": 12993 }, { "epoch": 0.93, "grad_norm": 3.8461285143243273, "learning_rate": 1.214774229484339e-07, "loss": 0.1397, "step": 12994 }, { "epoch": 0.93, "grad_norm": 4.628574025154759, "learning_rate": 1.2122312105082967e-07, "loss": 0.7453, "step": 12995 }, { "epoch": 0.93, "grad_norm": 4.165399378565865, "learning_rate": 1.2096908234477066e-07, "loss": 0.1528, "step": 12996 }, { "epoch": 0.93, "grad_norm": 19.671093188567244, "learning_rate": 1.2071530684396148e-07, "loss": 0.6056, "step": 12997 }, { "epoch": 0.93, "grad_norm": 19.56243560619769, "learning_rate": 1.2046179456209229e-07, "loss": 0.7937, "step": 12998 }, { "epoch": 0.93, "grad_norm": 3.5482843030188005, "learning_rate": 1.202085455128399e-07, "loss": 0.2669, "step": 12999 }, { "epoch": 0.93, "grad_norm": 2.817313811132414, "learning_rate": 1.199555597098645e-07, "loss": 0.1392, "step": 13000 }, { "epoch": 0.93, "grad_norm": 2.6519679799576075, "learning_rate": 1.1970283716681519e-07, "loss": 0.1141, "step": 13001 }, { "epoch": 0.93, "grad_norm": 11.315640683197437, "learning_rate": 1.194503778973255e-07, "loss": 0.5891, "step": 13002 }, { "epoch": 0.93, "grad_norm": 3.374989522149523, "learning_rate": 1.1919818191501286e-07, "loss": 0.1412, "step": 13003 }, { "epoch": 0.93, "grad_norm": 7.922939900571241, "learning_rate": 1.1894624923348418e-07, "loss": 0.3304, "step": 13004 }, { "epoch": 0.93, "grad_norm": 12.598010620674456, "learning_rate": 1.1869457986632915e-07, "loss": 0.6542, "step": 13005 }, { "epoch": 0.93, "grad_norm": 4.110533266675777, "learning_rate": 1.1844317382712467e-07, "loss": 0.1303, "step": 13006 }, { "epoch": 0.93, "grad_norm": 4.2211281957749325, "learning_rate": 1.1819203112943378e-07, "loss": 0.2756, "step": 13007 }, { "epoch": 0.93, "grad_norm": 3.482635158083589, "learning_rate": 1.1794115178680398e-07, "loss": 0.2214, "step": 13008 }, { "epoch": 0.93, "grad_norm": 2.8940059008985384, "learning_rate": 1.1769053581276946e-07, "loss": 0.1812, "step": 13009 }, { "epoch": 0.93, "grad_norm": 6.9103269338947895, "learning_rate": 1.1744018322084993e-07, "loss": 0.3681, "step": 13010 }, { "epoch": 0.93, "grad_norm": 20.025618336417466, "learning_rate": 1.1719009402455183e-07, "loss": 1.1696, "step": 13011 }, { "epoch": 0.93, "grad_norm": 19.648268612173904, "learning_rate": 1.1694026823736438e-07, "loss": 0.7597, "step": 13012 }, { "epoch": 0.93, "grad_norm": 3.595163377394137, "learning_rate": 1.1669070587276676e-07, "loss": 0.2448, "step": 13013 }, { "epoch": 0.93, "grad_norm": 20.86405766926683, "learning_rate": 1.1644140694422101e-07, "loss": 0.4059, "step": 13014 }, { "epoch": 0.93, "grad_norm": 37.798676545204344, "learning_rate": 1.1619237146517637e-07, "loss": 0.7067, "step": 13015 }, { "epoch": 0.93, "grad_norm": 22.219235534039576, "learning_rate": 1.1594359944906652e-07, "loss": 1.0662, "step": 13016 }, { "epoch": 0.93, "grad_norm": 4.204627425497549, "learning_rate": 1.1569509090931296e-07, "loss": 0.2657, "step": 13017 }, { "epoch": 0.93, "grad_norm": 3.825776757486409, "learning_rate": 1.1544684585932053e-07, "loss": 0.1657, "step": 13018 }, { "epoch": 0.93, "grad_norm": 3.2355966937294385, "learning_rate": 1.151988643124824e-07, "loss": 0.1065, "step": 13019 }, { "epoch": 0.93, "grad_norm": 4.763693053475156, "learning_rate": 1.1495114628217507e-07, "loss": 0.8319, "step": 13020 }, { "epoch": 0.93, "grad_norm": 4.700937838406348, "learning_rate": 1.1470369178176289e-07, "loss": 0.6605, "step": 13021 }, { "epoch": 0.93, "grad_norm": 3.613774558355277, "learning_rate": 1.1445650082459403e-07, "loss": 0.1233, "step": 13022 }, { "epoch": 0.93, "grad_norm": 3.6062773411583557, "learning_rate": 1.1420957342400507e-07, "loss": 0.1189, "step": 13023 }, { "epoch": 0.93, "grad_norm": 22.38418478548239, "learning_rate": 1.1396290959331535e-07, "loss": 0.8211, "step": 13024 }, { "epoch": 0.93, "grad_norm": 4.291775462535165, "learning_rate": 1.13716509345832e-07, "loss": 0.1153, "step": 13025 }, { "epoch": 0.93, "grad_norm": 5.100071422449239, "learning_rate": 1.1347037269484772e-07, "loss": 0.8046, "step": 13026 }, { "epoch": 0.93, "grad_norm": 25.363883459661924, "learning_rate": 1.1322449965364024e-07, "loss": 0.5365, "step": 13027 }, { "epoch": 0.93, "grad_norm": 8.693754878852005, "learning_rate": 1.1297889023547281e-07, "loss": 0.4219, "step": 13028 }, { "epoch": 0.93, "grad_norm": 3.3324301660853854, "learning_rate": 1.1273354445359652e-07, "loss": 0.1458, "step": 13029 }, { "epoch": 0.93, "grad_norm": 2.887181084989627, "learning_rate": 1.1248846232124633e-07, "loss": 0.1259, "step": 13030 }, { "epoch": 0.93, "grad_norm": 2.108123129147638, "learning_rate": 1.1224364385164276e-07, "loss": 0.0756, "step": 13031 }, { "epoch": 0.93, "grad_norm": 3.2895719425438625, "learning_rate": 1.1199908905799418e-07, "loss": 0.1554, "step": 13032 }, { "epoch": 0.93, "grad_norm": 20.903555872006848, "learning_rate": 1.117547979534922e-07, "loss": 1.0461, "step": 13033 }, { "epoch": 0.93, "grad_norm": 3.389698542837565, "learning_rate": 1.1151077055131576e-07, "loss": 0.1598, "step": 13034 }, { "epoch": 0.93, "grad_norm": 4.402608631119811, "learning_rate": 1.1126700686462988e-07, "loss": 0.2295, "step": 13035 }, { "epoch": 0.93, "grad_norm": 3.473873263487527, "learning_rate": 1.1102350690658403e-07, "loss": 0.181, "step": 13036 }, { "epoch": 0.93, "grad_norm": 3.0000475870310574, "learning_rate": 1.1078027069031383e-07, "loss": 0.095, "step": 13037 }, { "epoch": 0.93, "grad_norm": 19.249371607264674, "learning_rate": 1.1053729822894155e-07, "loss": 0.5469, "step": 13038 }, { "epoch": 0.93, "grad_norm": 5.160174669729856, "learning_rate": 1.1029458953557449e-07, "loss": 0.9213, "step": 13039 }, { "epoch": 0.94, "grad_norm": 3.366809656510096, "learning_rate": 1.1005214462330494e-07, "loss": 0.2211, "step": 13040 }, { "epoch": 0.94, "grad_norm": 4.747903788507213, "learning_rate": 1.0980996350521356e-07, "loss": 0.2775, "step": 13041 }, { "epoch": 0.94, "grad_norm": 26.8472308541634, "learning_rate": 1.0956804619436434e-07, "loss": 1.0406, "step": 13042 }, { "epoch": 0.94, "grad_norm": 4.05638159731334, "learning_rate": 1.0932639270380685e-07, "loss": 0.1861, "step": 13043 }, { "epoch": 0.94, "grad_norm": 20.395629494885593, "learning_rate": 1.0908500304657899e-07, "loss": 0.364, "step": 13044 }, { "epoch": 0.94, "grad_norm": 16.883335607012626, "learning_rate": 1.08843877235702e-07, "loss": 0.6953, "step": 13045 }, { "epoch": 0.94, "grad_norm": 3.7319773674865004, "learning_rate": 1.0860301528418383e-07, "loss": 0.1339, "step": 13046 }, { "epoch": 0.94, "grad_norm": 7.016686397107189, "learning_rate": 1.0836241720501849e-07, "loss": 0.3196, "step": 13047 }, { "epoch": 0.94, "grad_norm": 2.729987179683953, "learning_rate": 1.0812208301118454e-07, "loss": 0.0815, "step": 13048 }, { "epoch": 0.94, "grad_norm": 1.928122258277512, "learning_rate": 1.0788201271564713e-07, "loss": 0.0846, "step": 13049 }, { "epoch": 0.94, "grad_norm": 3.4249417891779195, "learning_rate": 1.0764220633135758e-07, "loss": 0.1183, "step": 13050 }, { "epoch": 0.94, "grad_norm": 2.4931848418529996, "learning_rate": 1.0740266387125275e-07, "loss": 0.0955, "step": 13051 }, { "epoch": 0.94, "grad_norm": 2.647906433006463, "learning_rate": 1.07163385348254e-07, "loss": 0.1404, "step": 13052 }, { "epoch": 0.94, "grad_norm": 3.339178831729228, "learning_rate": 1.0692437077527096e-07, "loss": 0.1939, "step": 13053 }, { "epoch": 0.94, "grad_norm": 16.838781138015815, "learning_rate": 1.0668562016519668e-07, "loss": 1.0713, "step": 13054 }, { "epoch": 0.94, "grad_norm": 3.528054040337739, "learning_rate": 1.0644713353091085e-07, "loss": 0.116, "step": 13055 }, { "epoch": 0.94, "grad_norm": 9.727108670370797, "learning_rate": 1.0620891088527873e-07, "loss": 0.3972, "step": 13056 }, { "epoch": 0.94, "grad_norm": 3.7934490831478223, "learning_rate": 1.0597095224115228e-07, "loss": 0.7024, "step": 13057 }, { "epoch": 0.94, "grad_norm": 4.407569720541911, "learning_rate": 1.0573325761136788e-07, "loss": 0.8223, "step": 13058 }, { "epoch": 0.94, "grad_norm": 3.3634687389459303, "learning_rate": 1.0549582700874861e-07, "loss": 0.1593, "step": 13059 }, { "epoch": 0.94, "grad_norm": 7.760523285024333, "learning_rate": 1.0525866044610311e-07, "loss": 0.5049, "step": 13060 }, { "epoch": 0.94, "grad_norm": 4.5685475780135665, "learning_rate": 1.0502175793622449e-07, "loss": 0.7276, "step": 13061 }, { "epoch": 0.94, "grad_norm": 16.89755883097661, "learning_rate": 1.0478511949189362e-07, "loss": 0.4677, "step": 13062 }, { "epoch": 0.94, "grad_norm": 3.1201762179855557, "learning_rate": 1.0454874512587698e-07, "loss": 0.1703, "step": 13063 }, { "epoch": 0.94, "grad_norm": 13.744195279689086, "learning_rate": 1.0431263485092436e-07, "loss": 0.7411, "step": 13064 }, { "epoch": 0.94, "grad_norm": 29.593964088461224, "learning_rate": 1.0407678867977389e-07, "loss": 0.3119, "step": 13065 }, { "epoch": 0.94, "grad_norm": 4.7438566903146215, "learning_rate": 1.038412066251493e-07, "loss": 0.163, "step": 13066 }, { "epoch": 0.94, "grad_norm": 11.088189456933232, "learning_rate": 1.0360588869975819e-07, "loss": 0.5575, "step": 13067 }, { "epoch": 0.94, "grad_norm": 8.805537400454662, "learning_rate": 1.033708349162954e-07, "loss": 0.3637, "step": 13068 }, { "epoch": 0.94, "grad_norm": 2.8284925477418694, "learning_rate": 1.0313604528744192e-07, "loss": 0.2082, "step": 13069 }, { "epoch": 0.94, "grad_norm": 4.48501164090303, "learning_rate": 1.0290151982586261e-07, "loss": 0.7114, "step": 13070 }, { "epoch": 0.94, "grad_norm": 2.276091111726464, "learning_rate": 1.0266725854421012e-07, "loss": 0.1204, "step": 13071 }, { "epoch": 0.94, "grad_norm": 10.815260341444583, "learning_rate": 1.0243326145512156e-07, "loss": 0.3909, "step": 13072 }, { "epoch": 0.94, "grad_norm": 2.6249730360466677, "learning_rate": 1.0219952857122017e-07, "loss": 0.1439, "step": 13073 }, { "epoch": 0.94, "grad_norm": 2.8616106552144083, "learning_rate": 1.0196605990511477e-07, "loss": 0.0797, "step": 13074 }, { "epoch": 0.94, "grad_norm": 3.2742417167681936, "learning_rate": 1.0173285546940081e-07, "loss": 0.1981, "step": 13075 }, { "epoch": 0.94, "grad_norm": 2.9062850720474147, "learning_rate": 1.0149991527665826e-07, "loss": 0.1735, "step": 13076 }, { "epoch": 0.94, "grad_norm": 9.01763155196935, "learning_rate": 1.0126723933945315e-07, "loss": 0.6802, "step": 13077 }, { "epoch": 0.94, "grad_norm": 7.087008927083226, "learning_rate": 1.0103482767033768e-07, "loss": 0.8088, "step": 13078 }, { "epoch": 0.94, "grad_norm": 3.554329593308779, "learning_rate": 1.0080268028185014e-07, "loss": 0.1734, "step": 13079 }, { "epoch": 0.94, "grad_norm": 3.88886390831599, "learning_rate": 1.005707971865133e-07, "loss": 0.7329, "step": 13080 }, { "epoch": 0.94, "grad_norm": 2.8556201035022304, "learning_rate": 1.003391783968366e-07, "loss": 0.2156, "step": 13081 }, { "epoch": 0.94, "grad_norm": 2.2628590366085315, "learning_rate": 1.0010782392531504e-07, "loss": 0.0908, "step": 13082 }, { "epoch": 0.94, "grad_norm": 21.43416609229644, "learning_rate": 9.987673378442919e-08, "loss": 0.7339, "step": 13083 }, { "epoch": 0.94, "grad_norm": 4.635080926404771, "learning_rate": 9.96459079866463e-08, "loss": 0.9343, "step": 13084 }, { "epoch": 0.94, "grad_norm": 3.5969567811166954, "learning_rate": 9.941534654441699e-08, "loss": 0.207, "step": 13085 }, { "epoch": 0.94, "grad_norm": 33.2301287072219, "learning_rate": 9.918504947018014e-08, "loss": 1.3292, "step": 13086 }, { "epoch": 0.94, "grad_norm": 3.8833469377832, "learning_rate": 9.895501677635921e-08, "loss": 0.0731, "step": 13087 }, { "epoch": 0.94, "grad_norm": 4.125761536705114, "learning_rate": 9.872524847536424e-08, "loss": 0.1764, "step": 13088 }, { "epoch": 0.94, "grad_norm": 21.874621875848344, "learning_rate": 9.849574457958866e-08, "loss": 0.8801, "step": 13089 }, { "epoch": 0.94, "grad_norm": 14.60717413120826, "learning_rate": 9.826650510141533e-08, "loss": 0.4844, "step": 13090 }, { "epoch": 0.94, "grad_norm": 2.6999578034487457, "learning_rate": 9.803753005320993e-08, "loss": 0.0794, "step": 13091 }, { "epoch": 0.94, "grad_norm": 4.4904799048808, "learning_rate": 9.780881944732479e-08, "loss": 0.7682, "step": 13092 }, { "epoch": 0.94, "grad_norm": 3.275109103714346, "learning_rate": 9.758037329609782e-08, "loss": 0.1024, "step": 13093 }, { "epoch": 0.94, "grad_norm": 4.900173650262777, "learning_rate": 9.73521916118536e-08, "loss": 0.8123, "step": 13094 }, { "epoch": 0.94, "grad_norm": 8.46660645410571, "learning_rate": 9.712427440690064e-08, "loss": 0.4118, "step": 13095 }, { "epoch": 0.94, "grad_norm": 12.980795386212238, "learning_rate": 9.689662169353464e-08, "loss": 0.6579, "step": 13096 }, { "epoch": 0.94, "grad_norm": 2.984916292896164, "learning_rate": 9.66692334840369e-08, "loss": 0.139, "step": 13097 }, { "epoch": 0.94, "grad_norm": 3.4574057079133818, "learning_rate": 9.64421097906737e-08, "loss": 0.1978, "step": 13098 }, { "epoch": 0.94, "grad_norm": 3.303093744222966, "learning_rate": 9.621525062569748e-08, "loss": 0.1449, "step": 13099 }, { "epoch": 0.94, "grad_norm": 2.6285794467543733, "learning_rate": 9.598865600134732e-08, "loss": 0.105, "step": 13100 }, { "epoch": 0.94, "grad_norm": 2.4011396367440305, "learning_rate": 9.576232592984624e-08, "loss": 0.0849, "step": 13101 }, { "epoch": 0.94, "grad_norm": 10.523196085499883, "learning_rate": 9.553626042340336e-08, "loss": 0.2805, "step": 13102 }, { "epoch": 0.94, "grad_norm": 23.640020701279195, "learning_rate": 9.531045949421558e-08, "loss": 0.6035, "step": 13103 }, { "epoch": 0.94, "grad_norm": 11.31745910956163, "learning_rate": 9.508492315446372e-08, "loss": 0.4804, "step": 13104 }, { "epoch": 0.94, "grad_norm": 15.716469160606374, "learning_rate": 9.485965141631304e-08, "loss": 0.9499, "step": 13105 }, { "epoch": 0.94, "grad_norm": 3.241980071095749, "learning_rate": 9.463464429191771e-08, "loss": 0.1246, "step": 13106 }, { "epoch": 0.94, "grad_norm": 3.9492228914264538, "learning_rate": 9.440990179341581e-08, "loss": 0.7342, "step": 13107 }, { "epoch": 0.94, "grad_norm": 3.2733852372399697, "learning_rate": 9.418542393293095e-08, "loss": 0.2348, "step": 13108 }, { "epoch": 0.94, "grad_norm": 3.2774804855860897, "learning_rate": 9.396121072257236e-08, "loss": 0.1438, "step": 13109 }, { "epoch": 0.94, "grad_norm": 2.5923959075865164, "learning_rate": 9.373726217443701e-08, "loss": 0.0788, "step": 13110 }, { "epoch": 0.94, "grad_norm": 5.081997552973055, "learning_rate": 9.351357830060415e-08, "loss": 0.2335, "step": 13111 }, { "epoch": 0.94, "grad_norm": 3.0843002436550897, "learning_rate": 9.329015911314243e-08, "loss": 0.1198, "step": 13112 }, { "epoch": 0.94, "grad_norm": 36.73617418529035, "learning_rate": 9.306700462410334e-08, "loss": 0.8767, "step": 13113 }, { "epoch": 0.94, "grad_norm": 23.45401642863268, "learning_rate": 9.284411484552558e-08, "loss": 1.0338, "step": 13114 }, { "epoch": 0.94, "grad_norm": 4.644664823065026, "learning_rate": 9.262148978943342e-08, "loss": 0.835, "step": 13115 }, { "epoch": 0.94, "grad_norm": 4.534070324092167, "learning_rate": 9.239912946783613e-08, "loss": 0.6116, "step": 13116 }, { "epoch": 0.94, "grad_norm": 4.168017088639331, "learning_rate": 9.217703389272969e-08, "loss": 0.2021, "step": 13117 }, { "epoch": 0.94, "grad_norm": 5.710302552081164, "learning_rate": 9.195520307609507e-08, "loss": 0.2375, "step": 13118 }, { "epoch": 0.94, "grad_norm": 25.221047750409543, "learning_rate": 9.17336370298999e-08, "loss": 0.6669, "step": 13119 }, { "epoch": 0.94, "grad_norm": 24.870076849274778, "learning_rate": 9.151233576609575e-08, "loss": 0.7943, "step": 13120 }, { "epoch": 0.94, "grad_norm": 5.225117421313255, "learning_rate": 9.129129929662139e-08, "loss": 0.8381, "step": 13121 }, { "epoch": 0.94, "grad_norm": 4.559014924006107, "learning_rate": 9.107052763340118e-08, "loss": 0.2272, "step": 13122 }, { "epoch": 0.94, "grad_norm": 15.64676233816105, "learning_rate": 9.085002078834449e-08, "loss": 0.7158, "step": 13123 }, { "epoch": 0.94, "grad_norm": 2.4243162599055426, "learning_rate": 9.062977877334677e-08, "loss": 0.0674, "step": 13124 }, { "epoch": 0.94, "grad_norm": 16.43827741268249, "learning_rate": 9.040980160029023e-08, "loss": 0.4655, "step": 13125 }, { "epoch": 0.94, "grad_norm": 3.2068544097120557, "learning_rate": 9.019008928104034e-08, "loss": 0.1347, "step": 13126 }, { "epoch": 0.94, "grad_norm": 11.332546127007488, "learning_rate": 8.997064182745096e-08, "loss": 0.581, "step": 13127 }, { "epoch": 0.94, "grad_norm": 1.9201965881626102, "learning_rate": 8.975145925135986e-08, "loss": 0.0735, "step": 13128 }, { "epoch": 0.94, "grad_norm": 2.7202053667200117, "learning_rate": 8.953254156459145e-08, "loss": 0.1508, "step": 13129 }, { "epoch": 0.94, "grad_norm": 3.3935774738608644, "learning_rate": 8.931388877895464e-08, "loss": 0.1412, "step": 13130 }, { "epoch": 0.94, "grad_norm": 3.723497043225758, "learning_rate": 8.909550090624607e-08, "loss": 0.1749, "step": 13131 }, { "epoch": 0.94, "grad_norm": 4.066190845751961, "learning_rate": 8.887737795824692e-08, "loss": 0.8607, "step": 13132 }, { "epoch": 0.94, "grad_norm": 3.1347152494572827, "learning_rate": 8.865951994672273e-08, "loss": 0.1291, "step": 13133 }, { "epoch": 0.94, "grad_norm": 8.982676667652548, "learning_rate": 8.844192688342745e-08, "loss": 0.3797, "step": 13134 }, { "epoch": 0.94, "grad_norm": 11.693972329595368, "learning_rate": 8.822459878009893e-08, "loss": 0.5681, "step": 13135 }, { "epoch": 0.94, "grad_norm": 2.015454066093674, "learning_rate": 8.800753564846109e-08, "loss": 0.0909, "step": 13136 }, { "epoch": 0.94, "grad_norm": 2.5835041957691596, "learning_rate": 8.779073750022405e-08, "loss": 0.1268, "step": 13137 }, { "epoch": 0.94, "grad_norm": 3.9775280471834638, "learning_rate": 8.757420434708285e-08, "loss": 0.8773, "step": 13138 }, { "epoch": 0.94, "grad_norm": 3.518358606724006, "learning_rate": 8.735793620071875e-08, "loss": 0.1175, "step": 13139 }, { "epoch": 0.94, "grad_norm": 3.6410968538328095, "learning_rate": 8.714193307279851e-08, "loss": 0.1243, "step": 13140 }, { "epoch": 0.94, "grad_norm": 2.8504521653964865, "learning_rate": 8.692619497497501e-08, "loss": 0.1264, "step": 13141 }, { "epoch": 0.94, "grad_norm": 7.60390260153524, "learning_rate": 8.671072191888619e-08, "loss": 0.7017, "step": 13142 }, { "epoch": 0.94, "grad_norm": 18.96855682850444, "learning_rate": 8.649551391615607e-08, "loss": 0.599, "step": 13143 }, { "epoch": 0.94, "grad_norm": 4.250012927547489, "learning_rate": 8.628057097839482e-08, "loss": 0.8673, "step": 13144 }, { "epoch": 0.94, "grad_norm": 3.47587946759895, "learning_rate": 8.606589311719705e-08, "loss": 0.2397, "step": 13145 }, { "epoch": 0.94, "grad_norm": 3.4298850569240713, "learning_rate": 8.585148034414403e-08, "loss": 0.1716, "step": 13146 }, { "epoch": 0.94, "grad_norm": 9.388055739239707, "learning_rate": 8.563733267080265e-08, "loss": 0.5709, "step": 13147 }, { "epoch": 0.94, "grad_norm": 4.128344068736736, "learning_rate": 8.542345010872477e-08, "loss": 0.1058, "step": 13148 }, { "epoch": 0.94, "grad_norm": 4.8932527974994695, "learning_rate": 8.520983266944949e-08, "loss": 0.1421, "step": 13149 }, { "epoch": 0.94, "grad_norm": 3.7719114796501936, "learning_rate": 8.499648036450036e-08, "loss": 0.7051, "step": 13150 }, { "epoch": 0.94, "grad_norm": 24.921944681978676, "learning_rate": 8.478339320538653e-08, "loss": 0.6531, "step": 13151 }, { "epoch": 0.94, "grad_norm": 3.705199336166404, "learning_rate": 8.457057120360379e-08, "loss": 0.2202, "step": 13152 }, { "epoch": 0.94, "grad_norm": 13.552454532975839, "learning_rate": 8.435801437063296e-08, "loss": 0.677, "step": 13153 }, { "epoch": 0.94, "grad_norm": 3.082870567919918, "learning_rate": 8.414572271793986e-08, "loss": 0.15, "step": 13154 }, { "epoch": 0.94, "grad_norm": 2.406793117731951, "learning_rate": 8.393369625697811e-08, "loss": 0.1071, "step": 13155 }, { "epoch": 0.94, "grad_norm": 19.037927094052996, "learning_rate": 8.372193499918524e-08, "loss": 1.3365, "step": 13156 }, { "epoch": 0.94, "grad_norm": 5.240281638896841, "learning_rate": 8.351043895598432e-08, "loss": 0.8874, "step": 13157 }, { "epoch": 0.94, "grad_norm": 18.243641867292112, "learning_rate": 8.329920813878567e-08, "loss": 0.6951, "step": 13158 }, { "epoch": 0.94, "grad_norm": 15.565541509658006, "learning_rate": 8.308824255898407e-08, "loss": 0.5495, "step": 13159 }, { "epoch": 0.94, "grad_norm": 3.5968229787099393, "learning_rate": 8.287754222795985e-08, "loss": 0.1863, "step": 13160 }, { "epoch": 0.94, "grad_norm": 3.0855222278755625, "learning_rate": 8.266710715708059e-08, "loss": 0.1307, "step": 13161 }, { "epoch": 0.94, "grad_norm": 13.378283350172127, "learning_rate": 8.245693735769721e-08, "loss": 0.5121, "step": 13162 }, { "epoch": 0.94, "grad_norm": 14.119708769071295, "learning_rate": 8.224703284114787e-08, "loss": 0.8273, "step": 13163 }, { "epoch": 0.94, "grad_norm": 3.928817648528844, "learning_rate": 8.203739361875684e-08, "loss": 0.3116, "step": 13164 }, { "epoch": 0.94, "grad_norm": 4.4049114123339885, "learning_rate": 8.182801970183341e-08, "loss": 0.6819, "step": 13165 }, { "epoch": 0.94, "grad_norm": 4.37224434018418, "learning_rate": 8.16189111016713e-08, "loss": 0.8593, "step": 13166 }, { "epoch": 0.94, "grad_norm": 2.9614952353453514, "learning_rate": 8.141006782955207e-08, "loss": 0.1534, "step": 13167 }, { "epoch": 0.94, "grad_norm": 12.691532882704255, "learning_rate": 8.120148989674225e-08, "loss": 0.482, "step": 13168 }, { "epoch": 0.94, "grad_norm": 22.050223518328266, "learning_rate": 8.099317731449286e-08, "loss": 0.6441, "step": 13169 }, { "epoch": 0.94, "grad_norm": 4.37711240562248, "learning_rate": 8.07851300940421e-08, "loss": 0.6793, "step": 13170 }, { "epoch": 0.94, "grad_norm": 15.162514837133589, "learning_rate": 8.057734824661379e-08, "loss": 0.8396, "step": 13171 }, { "epoch": 0.94, "grad_norm": 11.408672710148508, "learning_rate": 8.036983178341617e-08, "loss": 0.4189, "step": 13172 }, { "epoch": 0.94, "grad_norm": 3.4846533228812366, "learning_rate": 8.016258071564364e-08, "loss": 0.2276, "step": 13173 }, { "epoch": 0.94, "grad_norm": 4.587374728366351, "learning_rate": 7.995559505447836e-08, "loss": 0.1778, "step": 13174 }, { "epoch": 0.94, "grad_norm": 4.903065198224361, "learning_rate": 7.974887481108472e-08, "loss": 0.7839, "step": 13175 }, { "epoch": 0.94, "grad_norm": 4.486697611011033, "learning_rate": 7.95424199966155e-08, "loss": 0.7744, "step": 13176 }, { "epoch": 0.94, "grad_norm": 4.029527943735407, "learning_rate": 7.93362306222073e-08, "loss": 0.7724, "step": 13177 }, { "epoch": 0.94, "grad_norm": 31.665822457419452, "learning_rate": 7.913030669898402e-08, "loss": 0.9153, "step": 13178 }, { "epoch": 0.95, "grad_norm": 3.6556370140845993, "learning_rate": 7.892464823805401e-08, "loss": 0.2078, "step": 13179 }, { "epoch": 0.95, "grad_norm": 3.6550018834938514, "learning_rate": 7.871925525051283e-08, "loss": 0.1728, "step": 13180 }, { "epoch": 0.95, "grad_norm": 4.939383141648227, "learning_rate": 7.851412774743938e-08, "loss": 1.0643, "step": 13181 }, { "epoch": 0.95, "grad_norm": 3.236627883940877, "learning_rate": 7.830926573989928e-08, "loss": 0.0999, "step": 13182 }, { "epoch": 0.95, "grad_norm": 4.839034772090345, "learning_rate": 7.810466923894478e-08, "loss": 0.7968, "step": 13183 }, { "epoch": 0.95, "grad_norm": 7.34140552636922, "learning_rate": 7.790033825561372e-08, "loss": 0.2777, "step": 13184 }, { "epoch": 0.95, "grad_norm": 3.3174274039823395, "learning_rate": 7.769627280092728e-08, "loss": 0.2037, "step": 13185 }, { "epoch": 0.95, "grad_norm": 20.870370013076702, "learning_rate": 7.749247288589557e-08, "loss": 0.953, "step": 13186 }, { "epoch": 0.95, "grad_norm": 4.930855135647345, "learning_rate": 7.728893852151253e-08, "loss": 0.9167, "step": 13187 }, { "epoch": 0.95, "grad_norm": 3.6446413524445407, "learning_rate": 7.708566971875663e-08, "loss": 0.0971, "step": 13188 }, { "epoch": 0.95, "grad_norm": 3.57739921425694, "learning_rate": 7.688266648859577e-08, "loss": 0.1561, "step": 13189 }, { "epoch": 0.95, "grad_norm": 6.132663112735527, "learning_rate": 7.66799288419795e-08, "loss": 0.3062, "step": 13190 }, { "epoch": 0.95, "grad_norm": 4.562631954507303, "learning_rate": 7.647745678984519e-08, "loss": 0.9079, "step": 13191 }, { "epoch": 0.95, "grad_norm": 8.11030240761941, "learning_rate": 7.627525034311634e-08, "loss": 0.3837, "step": 13192 }, { "epoch": 0.95, "grad_norm": 2.52187995659405, "learning_rate": 7.607330951270031e-08, "loss": 0.1177, "step": 13193 }, { "epoch": 0.95, "grad_norm": 3.883807600799346, "learning_rate": 7.587163430949008e-08, "loss": 0.7797, "step": 13194 }, { "epoch": 0.95, "grad_norm": 3.9153349034103413, "learning_rate": 7.567022474436691e-08, "loss": 0.7018, "step": 13195 }, { "epoch": 0.95, "grad_norm": 4.7019716902691275, "learning_rate": 7.546908082819604e-08, "loss": 0.7271, "step": 13196 }, { "epoch": 0.95, "grad_norm": 5.188949911783052, "learning_rate": 7.52682025718271e-08, "loss": 0.2999, "step": 13197 }, { "epoch": 0.95, "grad_norm": 3.0366437034502947, "learning_rate": 7.506758998609809e-08, "loss": 0.1859, "step": 13198 }, { "epoch": 0.95, "grad_norm": 19.82076779600217, "learning_rate": 7.486724308183036e-08, "loss": 1.0386, "step": 13199 }, { "epoch": 0.95, "grad_norm": 12.026108631516074, "learning_rate": 7.466716186983247e-08, "loss": 0.4849, "step": 13200 }, { "epoch": 0.95, "grad_norm": 14.510830535624548, "learning_rate": 7.446734636089747e-08, "loss": 0.7149, "step": 13201 }, { "epoch": 0.95, "grad_norm": 27.637654471441223, "learning_rate": 7.426779656580507e-08, "loss": 0.7025, "step": 13202 }, { "epoch": 0.95, "grad_norm": 3.9157159141575493, "learning_rate": 7.406851249532054e-08, "loss": 0.2246, "step": 13203 }, { "epoch": 0.95, "grad_norm": 3.7461806405494587, "learning_rate": 7.386949416019418e-08, "loss": 0.1607, "step": 13204 }, { "epoch": 0.95, "grad_norm": 4.727669333340698, "learning_rate": 7.367074157116183e-08, "loss": 0.9108, "step": 13205 }, { "epoch": 0.95, "grad_norm": 10.913025842233067, "learning_rate": 7.347225473894549e-08, "loss": 0.4396, "step": 13206 }, { "epoch": 0.95, "grad_norm": 3.556046826102608, "learning_rate": 7.327403367425323e-08, "loss": 0.1863, "step": 13207 }, { "epoch": 0.95, "grad_norm": 11.181889628287765, "learning_rate": 7.307607838777875e-08, "loss": 0.8668, "step": 13208 }, { "epoch": 0.95, "grad_norm": 4.4446504050647, "learning_rate": 7.287838889020016e-08, "loss": 0.2424, "step": 13209 }, { "epoch": 0.95, "grad_norm": 5.043969181167479, "learning_rate": 7.268096519218171e-08, "loss": 1.1028, "step": 13210 }, { "epoch": 0.95, "grad_norm": 10.08015844598161, "learning_rate": 7.248380730437487e-08, "loss": 0.3364, "step": 13211 }, { "epoch": 0.95, "grad_norm": 2.976094520360665, "learning_rate": 7.228691523741505e-08, "loss": 0.1004, "step": 13212 }, { "epoch": 0.95, "grad_norm": 9.639312256787917, "learning_rate": 7.209028900192317e-08, "loss": 0.4398, "step": 13213 }, { "epoch": 0.95, "grad_norm": 3.181303244697896, "learning_rate": 7.189392860850741e-08, "loss": 0.1293, "step": 13214 }, { "epoch": 0.95, "grad_norm": 35.12989466189376, "learning_rate": 7.169783406776043e-08, "loss": 0.9225, "step": 13215 }, { "epoch": 0.95, "grad_norm": 2.893743997596025, "learning_rate": 7.150200539026042e-08, "loss": 0.1341, "step": 13216 }, { "epoch": 0.95, "grad_norm": 3.0732133646365254, "learning_rate": 7.130644258657227e-08, "loss": 0.1724, "step": 13217 }, { "epoch": 0.95, "grad_norm": 10.255833707878626, "learning_rate": 7.111114566724531e-08, "loss": 0.4822, "step": 13218 }, { "epoch": 0.95, "grad_norm": 13.28159961694587, "learning_rate": 7.091611464281444e-08, "loss": 0.6911, "step": 13219 }, { "epoch": 0.95, "grad_norm": 21.915783936134087, "learning_rate": 7.072134952380238e-08, "loss": 0.8655, "step": 13220 }, { "epoch": 0.95, "grad_norm": 8.0402296148345, "learning_rate": 7.052685032071516e-08, "loss": 0.3726, "step": 13221 }, { "epoch": 0.95, "grad_norm": 3.317039561680805, "learning_rate": 7.033261704404493e-08, "loss": 0.2123, "step": 13222 }, { "epoch": 0.95, "grad_norm": 4.391869777384619, "learning_rate": 7.013864970427053e-08, "loss": 0.7022, "step": 13223 }, { "epoch": 0.95, "grad_norm": 4.143460879849803, "learning_rate": 6.994494831185528e-08, "loss": 0.1984, "step": 13224 }, { "epoch": 0.95, "grad_norm": 4.458028956099246, "learning_rate": 6.975151287724858e-08, "loss": 0.748, "step": 13225 }, { "epoch": 0.95, "grad_norm": 2.190278774625394, "learning_rate": 6.955834341088596e-08, "loss": 0.1155, "step": 13226 }, { "epoch": 0.95, "grad_norm": 4.520459133542004, "learning_rate": 6.9365439923188e-08, "loss": 0.8737, "step": 13227 }, { "epoch": 0.95, "grad_norm": 2.485855778973186, "learning_rate": 6.917280242456137e-08, "loss": 0.1125, "step": 13228 }, { "epoch": 0.95, "grad_norm": 2.460694614781245, "learning_rate": 6.898043092539719e-08, "loss": 0.1163, "step": 13229 }, { "epoch": 0.95, "grad_norm": 8.924376197841614, "learning_rate": 6.87883254360744e-08, "loss": 0.3788, "step": 13230 }, { "epoch": 0.95, "grad_norm": 3.2833792605070578, "learning_rate": 6.859648596695522e-08, "loss": 0.0922, "step": 13231 }, { "epoch": 0.95, "grad_norm": 2.418589530926973, "learning_rate": 6.840491252838976e-08, "loss": 0.1164, "step": 13232 }, { "epoch": 0.95, "grad_norm": 3.1559295282958137, "learning_rate": 6.821360513071251e-08, "loss": 0.211, "step": 13233 }, { "epoch": 0.95, "grad_norm": 3.1089813172744676, "learning_rate": 6.802256378424244e-08, "loss": 0.1028, "step": 13234 }, { "epoch": 0.95, "grad_norm": 4.33443041688174, "learning_rate": 6.78317884992874e-08, "loss": 0.777, "step": 13235 }, { "epoch": 0.95, "grad_norm": 4.533458472468362, "learning_rate": 6.76412792861375e-08, "loss": 0.877, "step": 13236 }, { "epoch": 0.95, "grad_norm": 17.86479752521288, "learning_rate": 6.74510361550712e-08, "loss": 0.7387, "step": 13237 }, { "epoch": 0.95, "grad_norm": 12.285300707110482, "learning_rate": 6.726105911635028e-08, "loss": 0.6521, "step": 13238 }, { "epoch": 0.95, "grad_norm": 4.929106772483287, "learning_rate": 6.707134818022377e-08, "loss": 0.6932, "step": 13239 }, { "epoch": 0.95, "grad_norm": 3.497022009101306, "learning_rate": 6.688190335692623e-08, "loss": 0.1741, "step": 13240 }, { "epoch": 0.95, "grad_norm": 19.87657866239657, "learning_rate": 6.669272465667675e-08, "loss": 0.6713, "step": 13241 }, { "epoch": 0.95, "grad_norm": 4.3795489127262694, "learning_rate": 6.65038120896816e-08, "loss": 0.7478, "step": 13242 }, { "epoch": 0.95, "grad_norm": 19.82492975627288, "learning_rate": 6.631516566613094e-08, "loss": 1.012, "step": 13243 }, { "epoch": 0.95, "grad_norm": 4.219974842327192, "learning_rate": 6.612678539620277e-08, "loss": 0.18, "step": 13244 }, { "epoch": 0.95, "grad_norm": 12.098803737891105, "learning_rate": 6.593867129005838e-08, "loss": 0.9341, "step": 13245 }, { "epoch": 0.95, "grad_norm": 1.848976394091587, "learning_rate": 6.575082335784632e-08, "loss": 0.0649, "step": 13246 }, { "epoch": 0.95, "grad_norm": 7.488040451596364, "learning_rate": 6.556324160970018e-08, "loss": 0.2626, "step": 13247 }, { "epoch": 0.95, "grad_norm": 3.158230952777871, "learning_rate": 6.537592605573905e-08, "loss": 0.1367, "step": 13248 }, { "epoch": 0.95, "grad_norm": 4.487894446124088, "learning_rate": 6.518887670606877e-08, "loss": 0.1136, "step": 13249 }, { "epoch": 0.95, "grad_norm": 2.8568133606801207, "learning_rate": 6.500209357077847e-08, "loss": 0.0651, "step": 13250 }, { "epoch": 0.95, "grad_norm": 2.149051728512213, "learning_rate": 6.481557665994564e-08, "loss": 0.0921, "step": 13251 }, { "epoch": 0.95, "grad_norm": 14.800445617161332, "learning_rate": 6.46293259836328e-08, "loss": 0.8196, "step": 13252 }, { "epoch": 0.95, "grad_norm": 3.8592564913685634, "learning_rate": 6.44433415518847e-08, "loss": 0.1242, "step": 13253 }, { "epoch": 0.95, "grad_norm": 4.576586569977401, "learning_rate": 6.425762337473717e-08, "loss": 0.8142, "step": 13254 }, { "epoch": 0.95, "grad_norm": 5.023224210258699, "learning_rate": 6.407217146220835e-08, "loss": 0.6947, "step": 13255 }, { "epoch": 0.95, "grad_norm": 21.32432657323335, "learning_rate": 6.38869858243013e-08, "loss": 0.4024, "step": 13256 }, { "epoch": 0.95, "grad_norm": 13.76907230391099, "learning_rate": 6.370206647100807e-08, "loss": 0.3429, "step": 13257 }, { "epoch": 0.95, "grad_norm": 25.267915418801397, "learning_rate": 6.351741341230288e-08, "loss": 0.5071, "step": 13258 }, { "epoch": 0.95, "grad_norm": 14.068026789417425, "learning_rate": 6.333302665814778e-08, "loss": 0.675, "step": 13259 }, { "epoch": 0.95, "grad_norm": 12.815549513085115, "learning_rate": 6.314890621848979e-08, "loss": 0.7056, "step": 13260 }, { "epoch": 0.95, "grad_norm": 3.058818438957053, "learning_rate": 6.296505210326153e-08, "loss": 0.1328, "step": 13261 }, { "epoch": 0.95, "grad_norm": 23.234671940452866, "learning_rate": 6.278146432238063e-08, "loss": 0.9795, "step": 13262 }, { "epoch": 0.95, "grad_norm": 3.2631596237022427, "learning_rate": 6.25981428857514e-08, "loss": 0.1205, "step": 13263 }, { "epoch": 0.95, "grad_norm": 10.226288078757051, "learning_rate": 6.241508780326366e-08, "loss": 0.5771, "step": 13264 }, { "epoch": 0.95, "grad_norm": 2.8967819862861903, "learning_rate": 6.223229908479178e-08, "loss": 0.1008, "step": 13265 }, { "epoch": 0.95, "grad_norm": 33.311812940970654, "learning_rate": 6.204977674019675e-08, "loss": 1.0285, "step": 13266 }, { "epoch": 0.95, "grad_norm": 15.763345702114743, "learning_rate": 6.186752077932567e-08, "loss": 0.9761, "step": 13267 }, { "epoch": 0.95, "grad_norm": 4.070774695625814, "learning_rate": 6.168553121200904e-08, "loss": 0.1653, "step": 13268 }, { "epoch": 0.95, "grad_norm": 5.858019081515323, "learning_rate": 6.150380804806622e-08, "loss": 0.2332, "step": 13269 }, { "epoch": 0.95, "grad_norm": 6.014836367628685, "learning_rate": 6.132235129729991e-08, "loss": 0.8307, "step": 13270 }, { "epoch": 0.95, "grad_norm": 4.802766063471751, "learning_rate": 6.114116096949785e-08, "loss": 0.8639, "step": 13271 }, { "epoch": 0.95, "grad_norm": 6.401826112723365, "learning_rate": 6.096023707443666e-08, "loss": 0.1948, "step": 13272 }, { "epoch": 0.95, "grad_norm": 4.228633036163275, "learning_rate": 6.077957962187464e-08, "loss": 0.9146, "step": 13273 }, { "epoch": 0.95, "grad_norm": 2.763370714652961, "learning_rate": 6.0599188621559e-08, "loss": 0.189, "step": 13274 }, { "epoch": 0.95, "grad_norm": 14.211316619477824, "learning_rate": 6.041906408321973e-08, "loss": 0.7025, "step": 13275 }, { "epoch": 0.95, "grad_norm": 2.5307211508700953, "learning_rate": 6.023920601657574e-08, "loss": 0.0795, "step": 13276 }, { "epoch": 0.95, "grad_norm": 4.139116546269103, "learning_rate": 6.005961443132757e-08, "loss": 0.1793, "step": 13277 }, { "epoch": 0.95, "grad_norm": 8.10213515366297, "learning_rate": 5.988028933716473e-08, "loss": 0.674, "step": 13278 }, { "epoch": 0.95, "grad_norm": 3.907030591542554, "learning_rate": 5.97012307437611e-08, "loss": 0.2219, "step": 13279 }, { "epoch": 0.95, "grad_norm": 2.8579844304049002, "learning_rate": 5.952243866077623e-08, "loss": 0.0958, "step": 13280 }, { "epoch": 0.95, "grad_norm": 3.5055687452192013, "learning_rate": 5.934391309785403e-08, "loss": 0.1324, "step": 13281 }, { "epoch": 0.95, "grad_norm": 2.652690412156685, "learning_rate": 5.9165654064627374e-08, "loss": 0.1021, "step": 13282 }, { "epoch": 0.95, "grad_norm": 3.534589677878949, "learning_rate": 5.898766157071134e-08, "loss": 0.144, "step": 13283 }, { "epoch": 0.95, "grad_norm": 3.7404554447749367, "learning_rate": 5.880993562570769e-08, "loss": 0.2668, "step": 13284 }, { "epoch": 0.95, "grad_norm": 4.02456417220456, "learning_rate": 5.8632476239205424e-08, "loss": 0.6628, "step": 13285 }, { "epoch": 0.95, "grad_norm": 14.03284157522616, "learning_rate": 5.845528342077633e-08, "loss": 0.752, "step": 13286 }, { "epoch": 0.95, "grad_norm": 9.586609328692191, "learning_rate": 5.827835717997998e-08, "loss": 0.4376, "step": 13287 }, { "epoch": 0.95, "grad_norm": 12.829197216592664, "learning_rate": 5.810169752636152e-08, "loss": 0.5153, "step": 13288 }, { "epoch": 0.95, "grad_norm": 3.153522756938072, "learning_rate": 5.7925304469449994e-08, "loss": 0.1235, "step": 13289 }, { "epoch": 0.95, "grad_norm": 4.018879231997138, "learning_rate": 5.774917801876112e-08, "loss": 0.1631, "step": 13290 }, { "epoch": 0.95, "grad_norm": 5.028341487438817, "learning_rate": 5.75733181837973e-08, "loss": 0.8307, "step": 13291 }, { "epoch": 0.95, "grad_norm": 2.6983869319656866, "learning_rate": 5.739772497404483e-08, "loss": 0.1186, "step": 13292 }, { "epoch": 0.95, "grad_norm": 24.158318117566157, "learning_rate": 5.722239839897559e-08, "loss": 0.4917, "step": 13293 }, { "epoch": 0.95, "grad_norm": 12.167156044716902, "learning_rate": 5.704733846804922e-08, "loss": 0.6075, "step": 13294 }, { "epoch": 0.95, "grad_norm": 6.993890380809275, "learning_rate": 5.6872545190708727e-08, "loss": 0.2792, "step": 13295 }, { "epoch": 0.95, "grad_norm": 17.682356721565828, "learning_rate": 5.669801857638324e-08, "loss": 0.705, "step": 13296 }, { "epoch": 0.95, "grad_norm": 4.634287962627782, "learning_rate": 5.652375863448856e-08, "loss": 0.2874, "step": 13297 }, { "epoch": 0.95, "grad_norm": 4.294894167288765, "learning_rate": 5.634976537442494e-08, "loss": 0.5895, "step": 13298 }, { "epoch": 0.95, "grad_norm": 4.348911555017516, "learning_rate": 5.6176038805578756e-08, "loss": 0.7978, "step": 13299 }, { "epoch": 0.95, "grad_norm": 4.360541368955451, "learning_rate": 5.6002578937321974e-08, "loss": 0.8925, "step": 13300 }, { "epoch": 0.95, "grad_norm": 4.610001842395092, "learning_rate": 5.582938577901209e-08, "loss": 1.0166, "step": 13301 }, { "epoch": 0.95, "grad_norm": 5.0785739767756155, "learning_rate": 5.56564593399922e-08, "loss": 1.0498, "step": 13302 }, { "epoch": 0.95, "grad_norm": 3.700535651669732, "learning_rate": 5.548379962959094e-08, "loss": 0.3041, "step": 13303 }, { "epoch": 0.95, "grad_norm": 2.733202579468023, "learning_rate": 5.531140665712254e-08, "loss": 0.1225, "step": 13304 }, { "epoch": 0.95, "grad_norm": 13.024172442675836, "learning_rate": 5.5139280431886765e-08, "loss": 0.4105, "step": 13305 }, { "epoch": 0.95, "grad_norm": 2.6957576811041344, "learning_rate": 5.496742096317009e-08, "loss": 0.1157, "step": 13306 }, { "epoch": 0.95, "grad_norm": 3.3984232868078363, "learning_rate": 5.479582826024288e-08, "loss": 0.2016, "step": 13307 }, { "epoch": 0.95, "grad_norm": 21.398371062487016, "learning_rate": 5.462450233236161e-08, "loss": 0.5849, "step": 13308 }, { "epoch": 0.95, "grad_norm": 9.764626906006237, "learning_rate": 5.445344318876999e-08, "loss": 0.6911, "step": 13309 }, { "epoch": 0.95, "grad_norm": 2.932048506874375, "learning_rate": 5.428265083869455e-08, "loss": 0.0945, "step": 13310 }, { "epoch": 0.95, "grad_norm": 16.10016292060807, "learning_rate": 5.411212529135012e-08, "loss": 0.5097, "step": 13311 }, { "epoch": 0.95, "grad_norm": 19.989968984278168, "learning_rate": 5.394186655593492e-08, "loss": 0.5003, "step": 13312 }, { "epoch": 0.95, "grad_norm": 4.323554435134182, "learning_rate": 5.377187464163436e-08, "loss": 0.6706, "step": 13313 }, { "epoch": 0.95, "grad_norm": 6.077620191285605, "learning_rate": 5.360214955761833e-08, "loss": 0.2634, "step": 13314 }, { "epoch": 0.95, "grad_norm": 4.136663544076232, "learning_rate": 5.343269131304396e-08, "loss": 0.2312, "step": 13315 }, { "epoch": 0.95, "grad_norm": 3.0314382515055334, "learning_rate": 5.326349991705171e-08, "loss": 0.1321, "step": 13316 }, { "epoch": 0.95, "grad_norm": 7.8527605180432785, "learning_rate": 5.309457537876928e-08, "loss": 0.4041, "step": 13317 }, { "epoch": 0.96, "grad_norm": 3.1427855716443043, "learning_rate": 5.292591770730882e-08, "loss": 0.1556, "step": 13318 }, { "epoch": 0.96, "grad_norm": 4.178130420440268, "learning_rate": 5.2757526911770275e-08, "loss": 0.1907, "step": 13319 }, { "epoch": 0.96, "grad_norm": 4.0313314979312045, "learning_rate": 5.2589403001236385e-08, "loss": 0.2454, "step": 13320 }, { "epoch": 0.96, "grad_norm": 14.289116409415593, "learning_rate": 5.242154598477711e-08, "loss": 0.2398, "step": 13321 }, { "epoch": 0.96, "grad_norm": 3.354108986783177, "learning_rate": 5.225395587144799e-08, "loss": 0.1961, "step": 13322 }, { "epoch": 0.96, "grad_norm": 1.9442377044722585, "learning_rate": 5.208663267029013e-08, "loss": 0.0605, "step": 13323 }, { "epoch": 0.96, "grad_norm": 7.00339256664746, "learning_rate": 5.191957639032852e-08, "loss": 0.3776, "step": 13324 }, { "epoch": 0.96, "grad_norm": 1.9218258601920972, "learning_rate": 5.1752787040577625e-08, "loss": 0.0727, "step": 13325 }, { "epoch": 0.96, "grad_norm": 2.4301815522781935, "learning_rate": 5.158626463003302e-08, "loss": 0.1243, "step": 13326 }, { "epoch": 0.96, "grad_norm": 3.418234135461022, "learning_rate": 5.142000916767864e-08, "loss": 0.1459, "step": 13327 }, { "epoch": 0.96, "grad_norm": 3.9211770923998457, "learning_rate": 5.125402066248342e-08, "loss": 0.6735, "step": 13328 }, { "epoch": 0.96, "grad_norm": 19.295701414047524, "learning_rate": 5.1088299123401854e-08, "loss": 0.9883, "step": 13329 }, { "epoch": 0.96, "grad_norm": 3.6009761831830547, "learning_rate": 5.0922844559373466e-08, "loss": 0.1411, "step": 13330 }, { "epoch": 0.96, "grad_norm": 5.202258148265915, "learning_rate": 5.0757656979325e-08, "loss": 0.1722, "step": 13331 }, { "epoch": 0.96, "grad_norm": 4.847644217645336, "learning_rate": 5.059273639216655e-08, "loss": 0.7332, "step": 13332 }, { "epoch": 0.96, "grad_norm": 3.8775541049853697, "learning_rate": 5.0428082806795454e-08, "loss": 0.242, "step": 13333 }, { "epoch": 0.96, "grad_norm": 5.034726358696125, "learning_rate": 5.026369623209459e-08, "loss": 0.7823, "step": 13334 }, { "epoch": 0.96, "grad_norm": 3.1415651441069183, "learning_rate": 5.009957667693133e-08, "loss": 0.1286, "step": 13335 }, { "epoch": 0.96, "grad_norm": 3.6655621824486526, "learning_rate": 4.9935724150159125e-08, "loss": 0.181, "step": 13336 }, { "epoch": 0.96, "grad_norm": 5.753264709861386, "learning_rate": 4.97721386606187e-08, "loss": 0.2705, "step": 13337 }, { "epoch": 0.96, "grad_norm": 4.8020951121051105, "learning_rate": 4.960882021713298e-08, "loss": 0.1333, "step": 13338 }, { "epoch": 0.96, "grad_norm": 2.478304292693647, "learning_rate": 4.944576882851271e-08, "loss": 0.144, "step": 13339 }, { "epoch": 0.96, "grad_norm": 2.350415108045164, "learning_rate": 4.9282984503554735e-08, "loss": 0.1114, "step": 13340 }, { "epoch": 0.96, "grad_norm": 2.164571428551478, "learning_rate": 4.912046725104092e-08, "loss": 0.0762, "step": 13341 }, { "epoch": 0.96, "grad_norm": 17.797392647880734, "learning_rate": 4.895821707973702e-08, "loss": 0.5004, "step": 13342 }, { "epoch": 0.96, "grad_norm": 19.01420761405127, "learning_rate": 4.879623399839662e-08, "loss": 0.7518, "step": 13343 }, { "epoch": 0.96, "grad_norm": 9.378700941095397, "learning_rate": 4.8634518015758244e-08, "loss": 0.4945, "step": 13344 }, { "epoch": 0.96, "grad_norm": 2.7866199716577156, "learning_rate": 4.847306914054606e-08, "loss": 0.1812, "step": 13345 }, { "epoch": 0.96, "grad_norm": 2.5280481457580763, "learning_rate": 4.8311887381468636e-08, "loss": 0.1213, "step": 13346 }, { "epoch": 0.96, "grad_norm": 14.65199788875699, "learning_rate": 4.8150972747221804e-08, "loss": 0.7239, "step": 13347 }, { "epoch": 0.96, "grad_norm": 2.5817068619354306, "learning_rate": 4.799032524648639e-08, "loss": 0.1264, "step": 13348 }, { "epoch": 0.96, "grad_norm": 12.561899781640532, "learning_rate": 4.7829944887928804e-08, "loss": 0.7538, "step": 13349 }, { "epoch": 0.96, "grad_norm": 3.9745342123128204, "learning_rate": 4.766983168020045e-08, "loss": 0.3043, "step": 13350 }, { "epoch": 0.96, "grad_norm": 2.968607159666815, "learning_rate": 4.750998563193887e-08, "loss": 0.124, "step": 13351 }, { "epoch": 0.96, "grad_norm": 17.3675319740442, "learning_rate": 4.735040675176772e-08, "loss": 0.9059, "step": 13352 }, { "epoch": 0.96, "grad_norm": 3.488903906226554, "learning_rate": 4.7191095048295664e-08, "loss": 0.1811, "step": 13353 }, { "epoch": 0.96, "grad_norm": 3.114932267397488, "learning_rate": 4.7032050530116394e-08, "loss": 0.1544, "step": 13354 }, { "epoch": 0.96, "grad_norm": 3.408209481022463, "learning_rate": 4.6873273205809144e-08, "loss": 0.1152, "step": 13355 }, { "epoch": 0.96, "grad_norm": 15.510629543570097, "learning_rate": 4.6714763083940964e-08, "loss": 0.8825, "step": 13356 }, { "epoch": 0.96, "grad_norm": 5.004168416341726, "learning_rate": 4.655652017306222e-08, "loss": 0.8647, "step": 13357 }, { "epoch": 0.96, "grad_norm": 9.681749299235895, "learning_rate": 4.6398544481708885e-08, "loss": 0.3683, "step": 13358 }, { "epoch": 0.96, "grad_norm": 3.20460020083693, "learning_rate": 4.6240836018404124e-08, "loss": 0.1332, "step": 13359 }, { "epoch": 0.96, "grad_norm": 15.261638251136686, "learning_rate": 4.608339479165558e-08, "loss": 0.5869, "step": 13360 }, { "epoch": 0.96, "grad_norm": 9.068929191753003, "learning_rate": 4.59262208099559e-08, "loss": 0.2321, "step": 13361 }, { "epoch": 0.96, "grad_norm": 2.408341095664471, "learning_rate": 4.5769314081784424e-08, "loss": 0.1393, "step": 13362 }, { "epoch": 0.96, "grad_norm": 2.6922172862149765, "learning_rate": 4.561267461560548e-08, "loss": 0.1014, "step": 13363 }, { "epoch": 0.96, "grad_norm": 5.14541679862455, "learning_rate": 4.545630241986954e-08, "loss": 0.2527, "step": 13364 }, { "epoch": 0.96, "grad_norm": 2.6109854058490103, "learning_rate": 4.5300197503012066e-08, "loss": 0.1106, "step": 13365 }, { "epoch": 0.96, "grad_norm": 4.5615357237107315, "learning_rate": 4.514435987345411e-08, "loss": 0.7389, "step": 13366 }, { "epoch": 0.96, "grad_norm": 2.80861962190204, "learning_rate": 4.498878953960284e-08, "loss": 0.1562, "step": 13367 }, { "epoch": 0.96, "grad_norm": 2.341434470582217, "learning_rate": 4.483348650985098e-08, "loss": 0.0981, "step": 13368 }, { "epoch": 0.96, "grad_norm": 2.892990346683554, "learning_rate": 4.467845079257627e-08, "loss": 0.1723, "step": 13369 }, { "epoch": 0.96, "grad_norm": 10.992621168348364, "learning_rate": 4.452368239614202e-08, "loss": 0.3421, "step": 13370 }, { "epoch": 0.96, "grad_norm": 2.6755695738008054, "learning_rate": 4.436918132889767e-08, "loss": 0.1433, "step": 13371 }, { "epoch": 0.96, "grad_norm": 2.0619291989338158, "learning_rate": 4.421494759917766e-08, "loss": 0.0757, "step": 13372 }, { "epoch": 0.96, "grad_norm": 15.327603237411012, "learning_rate": 4.406098121530311e-08, "loss": 0.5385, "step": 13373 }, { "epoch": 0.96, "grad_norm": 4.592553684071719, "learning_rate": 4.390728218557905e-08, "loss": 0.1779, "step": 13374 }, { "epoch": 0.96, "grad_norm": 4.4876609261274885, "learning_rate": 4.375385051829717e-08, "loss": 0.6588, "step": 13375 }, { "epoch": 0.96, "grad_norm": 13.273031530128138, "learning_rate": 4.3600686221734743e-08, "loss": 0.6858, "step": 13376 }, { "epoch": 0.96, "grad_norm": 5.190748879048548, "learning_rate": 4.344778930415461e-08, "loss": 0.1777, "step": 13377 }, { "epoch": 0.96, "grad_norm": 8.647565805103207, "learning_rate": 4.329515977380405e-08, "loss": 0.4316, "step": 13378 }, { "epoch": 0.96, "grad_norm": 4.04273135157998, "learning_rate": 4.31427976389176e-08, "loss": 0.2891, "step": 13379 }, { "epoch": 0.96, "grad_norm": 4.687561842252615, "learning_rate": 4.2990702907714787e-08, "loss": 0.1351, "step": 13380 }, { "epoch": 0.96, "grad_norm": 4.206314064510006, "learning_rate": 4.2838875588400166e-08, "loss": 0.7826, "step": 13381 }, { "epoch": 0.96, "grad_norm": 25.15578626436013, "learning_rate": 4.26873156891644e-08, "loss": 0.5354, "step": 13382 }, { "epoch": 0.96, "grad_norm": 3.2231942247142924, "learning_rate": 4.253602321818373e-08, "loss": 0.1775, "step": 13383 }, { "epoch": 0.96, "grad_norm": 11.648704190477083, "learning_rate": 4.2384998183618854e-08, "loss": 0.4531, "step": 13384 }, { "epoch": 0.96, "grad_norm": 3.87504492124276, "learning_rate": 4.223424059361825e-08, "loss": 0.2566, "step": 13385 }, { "epoch": 0.96, "grad_norm": 12.098355417196734, "learning_rate": 4.20837504563143e-08, "loss": 0.8858, "step": 13386 }, { "epoch": 0.96, "grad_norm": 3.1938433628002816, "learning_rate": 4.193352777982551e-08, "loss": 0.1906, "step": 13387 }, { "epoch": 0.96, "grad_norm": 23.329278960704457, "learning_rate": 4.178357257225485e-08, "loss": 0.7837, "step": 13388 }, { "epoch": 0.96, "grad_norm": 5.068665534561664, "learning_rate": 4.163388484169306e-08, "loss": 0.9142, "step": 13389 }, { "epoch": 0.96, "grad_norm": 20.521585817305144, "learning_rate": 4.1484464596214804e-08, "loss": 1.2173, "step": 13390 }, { "epoch": 0.96, "grad_norm": 5.399075290351531, "learning_rate": 4.1335311843880286e-08, "loss": 0.1713, "step": 13391 }, { "epoch": 0.96, "grad_norm": 2.837124966220087, "learning_rate": 4.118642659273642e-08, "loss": 0.2229, "step": 13392 }, { "epoch": 0.96, "grad_norm": 11.151353617892823, "learning_rate": 4.1037808850814544e-08, "loss": 0.3454, "step": 13393 }, { "epoch": 0.96, "grad_norm": 5.093833581872986, "learning_rate": 4.088945862613214e-08, "loss": 0.76, "step": 13394 }, { "epoch": 0.96, "grad_norm": 3.0061852029497955, "learning_rate": 4.074137592669225e-08, "loss": 0.1372, "step": 13395 }, { "epoch": 0.96, "grad_norm": 2.6224791764978486, "learning_rate": 4.059356076048348e-08, "loss": 0.1642, "step": 13396 }, { "epoch": 0.96, "grad_norm": 3.159565778326934, "learning_rate": 4.044601313548002e-08, "loss": 0.128, "step": 13397 }, { "epoch": 0.96, "grad_norm": 13.993199790445201, "learning_rate": 4.0298733059640496e-08, "loss": 0.6409, "step": 13398 }, { "epoch": 0.96, "grad_norm": 3.1286367543767475, "learning_rate": 4.015172054091132e-08, "loss": 0.1453, "step": 13399 }, { "epoch": 0.96, "grad_norm": 2.98656943552379, "learning_rate": 4.0004975587222824e-08, "loss": 0.1021, "step": 13400 }, { "epoch": 0.96, "grad_norm": 3.0397773139892785, "learning_rate": 3.98584982064909e-08, "loss": 0.1712, "step": 13401 }, { "epoch": 0.96, "grad_norm": 2.299100907894208, "learning_rate": 3.971228840661812e-08, "loss": 0.0586, "step": 13402 }, { "epoch": 0.96, "grad_norm": 3.1449793507155883, "learning_rate": 3.956634619549149e-08, "loss": 0.1151, "step": 13403 }, { "epoch": 0.96, "grad_norm": 4.405852693814918, "learning_rate": 3.942067158098417e-08, "loss": 0.2455, "step": 13404 }, { "epoch": 0.96, "grad_norm": 2.825797266505263, "learning_rate": 3.927526457095488e-08, "loss": 0.1036, "step": 13405 }, { "epoch": 0.96, "grad_norm": 20.302755574217425, "learning_rate": 3.913012517324788e-08, "loss": 0.382, "step": 13406 }, { "epoch": 0.96, "grad_norm": 3.8007171599063865, "learning_rate": 3.898525339569248e-08, "loss": 0.3168, "step": 13407 }, { "epoch": 0.96, "grad_norm": 4.303431017892282, "learning_rate": 3.884064924610464e-08, "loss": 0.8337, "step": 13408 }, { "epoch": 0.96, "grad_norm": 35.72348135692865, "learning_rate": 3.86963127322848e-08, "loss": 0.5197, "step": 13409 }, { "epoch": 0.96, "grad_norm": 3.715622937794781, "learning_rate": 3.855224386201895e-08, "loss": 0.1811, "step": 13410 }, { "epoch": 0.96, "grad_norm": 3.238849997319549, "learning_rate": 3.840844264307919e-08, "loss": 0.137, "step": 13411 }, { "epoch": 0.96, "grad_norm": 3.5463698645247703, "learning_rate": 3.826490908322378e-08, "loss": 0.1598, "step": 13412 }, { "epoch": 0.96, "grad_norm": 3.191997573620288, "learning_rate": 3.812164319019485e-08, "loss": 0.1594, "step": 13413 }, { "epoch": 0.96, "grad_norm": 18.924516467850474, "learning_rate": 3.797864497172232e-08, "loss": 0.8276, "step": 13414 }, { "epoch": 0.96, "grad_norm": 4.804494373507003, "learning_rate": 3.7835914435518925e-08, "loss": 0.8123, "step": 13415 }, { "epoch": 0.96, "grad_norm": 3.807542305384726, "learning_rate": 3.769345158928517e-08, "loss": 0.1361, "step": 13416 }, { "epoch": 0.96, "grad_norm": 3.5820627126944333, "learning_rate": 3.755125644070656e-08, "loss": 0.1428, "step": 13417 }, { "epoch": 0.96, "grad_norm": 19.6946412411294, "learning_rate": 3.74093289974542e-08, "loss": 1.0388, "step": 13418 }, { "epoch": 0.96, "grad_norm": 17.429779634133176, "learning_rate": 3.7267669267183616e-08, "loss": 0.5514, "step": 13419 }, { "epoch": 0.96, "grad_norm": 3.951066083259172, "learning_rate": 3.7126277257537036e-08, "loss": 0.2055, "step": 13420 }, { "epoch": 0.96, "grad_norm": 8.746397273343463, "learning_rate": 3.698515297614336e-08, "loss": 0.3145, "step": 13421 }, { "epoch": 0.96, "grad_norm": 18.032933334393455, "learning_rate": 3.6844296430613714e-08, "loss": 0.8097, "step": 13422 }, { "epoch": 0.96, "grad_norm": 4.800242121270714, "learning_rate": 3.670370762854758e-08, "loss": 0.7367, "step": 13423 }, { "epoch": 0.96, "grad_norm": 2.8957929647476655, "learning_rate": 3.6563386577530005e-08, "loss": 0.0912, "step": 13424 }, { "epoch": 0.96, "grad_norm": 3.083727097193489, "learning_rate": 3.642333328512992e-08, "loss": 0.2178, "step": 13425 }, { "epoch": 0.96, "grad_norm": 3.5129046719354333, "learning_rate": 3.628354775890242e-08, "loss": 0.1364, "step": 13426 }, { "epoch": 0.96, "grad_norm": 2.9861118482360487, "learning_rate": 3.614403000638922e-08, "loss": 0.1426, "step": 13427 }, { "epoch": 0.96, "grad_norm": 11.223519738540187, "learning_rate": 3.6004780035116537e-08, "loss": 0.6994, "step": 13428 }, { "epoch": 0.96, "grad_norm": 3.2121425198194093, "learning_rate": 3.586579785259559e-08, "loss": 0.2199, "step": 13429 }, { "epoch": 0.96, "grad_norm": 8.951254172492357, "learning_rate": 3.572708346632481e-08, "loss": 0.4578, "step": 13430 }, { "epoch": 0.96, "grad_norm": 10.766533552098748, "learning_rate": 3.5588636883787664e-08, "loss": 0.6112, "step": 13431 }, { "epoch": 0.96, "grad_norm": 23.261940828098716, "learning_rate": 3.5450458112451515e-08, "loss": 0.6757, "step": 13432 }, { "epoch": 0.96, "grad_norm": 3.159076276693505, "learning_rate": 3.531254715977206e-08, "loss": 0.1491, "step": 13433 }, { "epoch": 0.96, "grad_norm": 20.680978195974305, "learning_rate": 3.5174904033187795e-08, "loss": 0.8092, "step": 13434 }, { "epoch": 0.96, "grad_norm": 3.7168360874928963, "learning_rate": 3.5037528740124446e-08, "loss": 0.1508, "step": 13435 }, { "epoch": 0.96, "grad_norm": 3.114798351250379, "learning_rate": 3.490042128799331e-08, "loss": 0.1757, "step": 13436 }, { "epoch": 0.96, "grad_norm": 3.445706871443687, "learning_rate": 3.476358168419069e-08, "loss": 0.1294, "step": 13437 }, { "epoch": 0.96, "grad_norm": 4.5447693971364655, "learning_rate": 3.462700993609791e-08, "loss": 0.7848, "step": 13438 }, { "epoch": 0.96, "grad_norm": 5.59596929867451, "learning_rate": 3.4490706051082955e-08, "loss": 0.1374, "step": 13439 }, { "epoch": 0.96, "grad_norm": 3.42291287853005, "learning_rate": 3.435467003649939e-08, "loss": 0.1546, "step": 13440 }, { "epoch": 0.96, "grad_norm": 8.440607234969844, "learning_rate": 3.4218901899685245e-08, "loss": 0.4919, "step": 13441 }, { "epoch": 0.96, "grad_norm": 171.5645823296567, "learning_rate": 3.408340164796464e-08, "loss": 0.5391, "step": 13442 }, { "epoch": 0.96, "grad_norm": 2.4655020396888254, "learning_rate": 3.3948169288647856e-08, "loss": 0.0769, "step": 13443 }, { "epoch": 0.96, "grad_norm": 4.6258705767419, "learning_rate": 3.3813204829029055e-08, "loss": 0.7896, "step": 13444 }, { "epoch": 0.96, "grad_norm": 2.9359358223730028, "learning_rate": 3.3678508276390744e-08, "loss": 0.1282, "step": 13445 }, { "epoch": 0.96, "grad_norm": 4.551347339779237, "learning_rate": 3.3544079637998214e-08, "loss": 0.8221, "step": 13446 }, { "epoch": 0.96, "grad_norm": 3.6548793495855567, "learning_rate": 3.34099189211029e-08, "loss": 0.1486, "step": 13447 }, { "epoch": 0.96, "grad_norm": 13.692165643038443, "learning_rate": 3.327602613294345e-08, "loss": 0.6833, "step": 13448 }, { "epoch": 0.96, "grad_norm": 2.3864003224173733, "learning_rate": 3.314240128074242e-08, "loss": 0.1057, "step": 13449 }, { "epoch": 0.96, "grad_norm": 3.728225323332184, "learning_rate": 3.300904437170793e-08, "loss": 0.1798, "step": 13450 }, { "epoch": 0.96, "grad_norm": 3.124645507137426, "learning_rate": 3.2875955413034234e-08, "loss": 0.1373, "step": 13451 }, { "epoch": 0.96, "grad_norm": 7.49256386739201, "learning_rate": 3.274313441190169e-08, "loss": 0.2082, "step": 13452 }, { "epoch": 0.96, "grad_norm": 2.6434919702634976, "learning_rate": 3.2610581375474574e-08, "loss": 0.1329, "step": 13453 }, { "epoch": 0.96, "grad_norm": 20.089853699274386, "learning_rate": 3.247829631090438e-08, "loss": 0.9629, "step": 13454 }, { "epoch": 0.96, "grad_norm": 3.148295482177022, "learning_rate": 3.234627922532652e-08, "loss": 0.1403, "step": 13455 }, { "epoch": 0.96, "grad_norm": 2.7800465754124497, "learning_rate": 3.221453012586362e-08, "loss": 0.0857, "step": 13456 }, { "epoch": 0.96, "grad_norm": 4.6347443799892885, "learning_rate": 3.208304901962278e-08, "loss": 0.787, "step": 13457 }, { "epoch": 0.97, "grad_norm": 4.0890817853266785, "learning_rate": 3.195183591369666e-08, "loss": 0.1591, "step": 13458 }, { "epoch": 0.97, "grad_norm": 3.275721407739138, "learning_rate": 3.182089081516404e-08, "loss": 0.216, "step": 13459 }, { "epoch": 0.97, "grad_norm": 3.4826034905940397, "learning_rate": 3.1690213731088715e-08, "loss": 0.0689, "step": 13460 }, { "epoch": 0.97, "grad_norm": 4.6646334858900635, "learning_rate": 3.15598046685206e-08, "loss": 0.8299, "step": 13461 }, { "epoch": 0.97, "grad_norm": 4.018987826612586, "learning_rate": 3.142966363449407e-08, "loss": 0.1503, "step": 13462 }, { "epoch": 0.97, "grad_norm": 7.678171442575762, "learning_rate": 3.129979063602962e-08, "loss": 0.3097, "step": 13463 }, { "epoch": 0.97, "grad_norm": 7.967124279546804, "learning_rate": 3.117018568013497e-08, "loss": 0.4106, "step": 13464 }, { "epoch": 0.97, "grad_norm": 1.671485039549939, "learning_rate": 3.104084877380009e-08, "loss": 0.0753, "step": 13465 }, { "epoch": 0.97, "grad_norm": 3.4407057134974313, "learning_rate": 3.091177992400329e-08, "loss": 0.2415, "step": 13466 }, { "epoch": 0.97, "grad_norm": 3.2057588742417518, "learning_rate": 3.078297913770678e-08, "loss": 0.1412, "step": 13467 }, { "epoch": 0.97, "grad_norm": 4.3769981402073395, "learning_rate": 3.0654446421858884e-08, "loss": 0.8251, "step": 13468 }, { "epoch": 0.97, "grad_norm": 28.963049416439663, "learning_rate": 3.05261817833935e-08, "loss": 0.8311, "step": 13469 }, { "epoch": 0.97, "grad_norm": 3.918036976652232, "learning_rate": 3.039818522923066e-08, "loss": 0.146, "step": 13470 }, { "epoch": 0.97, "grad_norm": 4.174460967307593, "learning_rate": 3.027045676627427e-08, "loss": 0.1841, "step": 13471 }, { "epoch": 0.97, "grad_norm": 3.1357454324950176, "learning_rate": 3.014299640141549e-08, "loss": 0.1873, "step": 13472 }, { "epoch": 0.97, "grad_norm": 5.784796601192804, "learning_rate": 3.0015804141529934e-08, "loss": 0.1437, "step": 13473 }, { "epoch": 0.97, "grad_norm": 9.352703172998083, "learning_rate": 2.988887999347989e-08, "loss": 0.4652, "step": 13474 }, { "epoch": 0.97, "grad_norm": 3.454305926776258, "learning_rate": 2.9762223964110993e-08, "loss": 0.3093, "step": 13475 }, { "epoch": 0.97, "grad_norm": 17.689239505456346, "learning_rate": 2.9635836060257774e-08, "loss": 0.5376, "step": 13476 }, { "epoch": 0.97, "grad_norm": 3.744431756352085, "learning_rate": 2.9509716288736445e-08, "loss": 0.2019, "step": 13477 }, { "epoch": 0.97, "grad_norm": 3.012661060428034, "learning_rate": 2.9383864656352123e-08, "loss": 0.177, "step": 13478 }, { "epoch": 0.97, "grad_norm": 2.950260694728473, "learning_rate": 2.9258281169893266e-08, "loss": 0.2006, "step": 13479 }, { "epoch": 0.97, "grad_norm": 10.835224876051269, "learning_rate": 2.9132965836135007e-08, "loss": 0.4777, "step": 13480 }, { "epoch": 0.97, "grad_norm": 16.089764385077142, "learning_rate": 2.9007918661837496e-08, "loss": 0.4948, "step": 13481 }, { "epoch": 0.97, "grad_norm": 29.143797881847277, "learning_rate": 2.888313965374645e-08, "loss": 0.7493, "step": 13482 }, { "epoch": 0.97, "grad_norm": 2.800820104800784, "learning_rate": 2.8758628818593704e-08, "loss": 0.0961, "step": 13483 }, { "epoch": 0.97, "grad_norm": 2.6339936189392166, "learning_rate": 2.8634386163094997e-08, "loss": 0.1004, "step": 13484 }, { "epoch": 0.97, "grad_norm": 7.237300518187775, "learning_rate": 2.8510411693953855e-08, "loss": 0.2064, "step": 13485 }, { "epoch": 0.97, "grad_norm": 5.48618860667754, "learning_rate": 2.8386705417858264e-08, "loss": 0.9072, "step": 13486 }, { "epoch": 0.97, "grad_norm": 8.010370906634957, "learning_rate": 2.826326734148066e-08, "loss": 0.5763, "step": 13487 }, { "epoch": 0.97, "grad_norm": 3.875087909572343, "learning_rate": 2.8140097471481276e-08, "loss": 0.739, "step": 13488 }, { "epoch": 0.97, "grad_norm": 7.484683181048119, "learning_rate": 2.8017195814503684e-08, "loss": 0.2957, "step": 13489 }, { "epoch": 0.97, "grad_norm": 4.0185717885784555, "learning_rate": 2.7894562377179247e-08, "loss": 0.798, "step": 13490 }, { "epoch": 0.97, "grad_norm": 3.860926706676001, "learning_rate": 2.7772197166121563e-08, "loss": 0.2804, "step": 13491 }, { "epoch": 0.97, "grad_norm": 3.099328944015459, "learning_rate": 2.7650100187933683e-08, "loss": 0.1721, "step": 13492 }, { "epoch": 0.97, "grad_norm": 4.580625774440502, "learning_rate": 2.7528271449201448e-08, "loss": 0.8441, "step": 13493 }, { "epoch": 0.97, "grad_norm": 9.090255477046492, "learning_rate": 2.7406710956497383e-08, "loss": 0.5078, "step": 13494 }, { "epoch": 0.97, "grad_norm": 2.755302852654525, "learning_rate": 2.7285418716378466e-08, "loss": 0.143, "step": 13495 }, { "epoch": 0.97, "grad_norm": 2.5210364636190215, "learning_rate": 2.7164394735388344e-08, "loss": 0.08, "step": 13496 }, { "epoch": 0.97, "grad_norm": 18.019147390521795, "learning_rate": 2.7043639020056246e-08, "loss": 0.7485, "step": 13497 }, { "epoch": 0.97, "grad_norm": 14.396849169261303, "learning_rate": 2.69231515768964e-08, "loss": 0.4538, "step": 13498 }, { "epoch": 0.97, "grad_norm": 3.4513309500496403, "learning_rate": 2.6802932412408057e-08, "loss": 0.1301, "step": 13499 }, { "epoch": 0.97, "grad_norm": 3.068196769679602, "learning_rate": 2.6682981533077133e-08, "loss": 0.185, "step": 13500 }, { "epoch": 0.97, "grad_norm": 25.808708819815326, "learning_rate": 2.6563298945374015e-08, "loss": 0.6623, "step": 13501 }, { "epoch": 0.97, "grad_norm": 5.0623299175832965, "learning_rate": 2.644388465575576e-08, "loss": 0.9335, "step": 13502 }, { "epoch": 0.97, "grad_norm": 3.8670809623584717, "learning_rate": 2.6324738670663874e-08, "loss": 0.2282, "step": 13503 }, { "epoch": 0.97, "grad_norm": 3.5378311762318866, "learning_rate": 2.6205860996526e-08, "loss": 0.1601, "step": 13504 }, { "epoch": 0.97, "grad_norm": 11.480899930568476, "learning_rate": 2.608725163975534e-08, "loss": 0.4752, "step": 13505 }, { "epoch": 0.97, "grad_norm": 5.083631668701765, "learning_rate": 2.5968910606749553e-08, "loss": 0.9727, "step": 13506 }, { "epoch": 0.97, "grad_norm": 2.4061650713482345, "learning_rate": 2.585083790389409e-08, "loss": 0.1203, "step": 13507 }, { "epoch": 0.97, "grad_norm": 9.69183587258964, "learning_rate": 2.5733033537557185e-08, "loss": 0.8185, "step": 13508 }, { "epoch": 0.97, "grad_norm": 4.777586155739093, "learning_rate": 2.5615497514094313e-08, "loss": 0.9077, "step": 13509 }, { "epoch": 0.97, "grad_norm": 9.860878881373552, "learning_rate": 2.5498229839847065e-08, "loss": 0.3776, "step": 13510 }, { "epoch": 0.97, "grad_norm": 26.079847281646817, "learning_rate": 2.5381230521140387e-08, "loss": 0.763, "step": 13511 }, { "epoch": 0.97, "grad_norm": 4.655182386250775, "learning_rate": 2.5264499564285892e-08, "loss": 0.7808, "step": 13512 }, { "epoch": 0.97, "grad_norm": 8.793258140559479, "learning_rate": 2.514803697558188e-08, "loss": 0.2659, "step": 13513 }, { "epoch": 0.97, "grad_norm": 9.878472758373867, "learning_rate": 2.5031842761309988e-08, "loss": 0.5414, "step": 13514 }, { "epoch": 0.97, "grad_norm": 3.5746695223244345, "learning_rate": 2.4915916927739093e-08, "loss": 0.1572, "step": 13515 }, { "epoch": 0.97, "grad_norm": 4.406839905195818, "learning_rate": 2.480025948112308e-08, "loss": 0.6731, "step": 13516 }, { "epoch": 0.97, "grad_norm": 4.6811484856010095, "learning_rate": 2.468487042770029e-08, "loss": 0.9328, "step": 13517 }, { "epoch": 0.97, "grad_norm": 10.305410581794062, "learning_rate": 2.456974977369686e-08, "loss": 0.481, "step": 13518 }, { "epoch": 0.97, "grad_norm": 3.4814043374015027, "learning_rate": 2.4454897525322263e-08, "loss": 0.2175, "step": 13519 }, { "epoch": 0.97, "grad_norm": 20.824863021254707, "learning_rate": 2.43403136887721e-08, "loss": 0.5077, "step": 13520 }, { "epoch": 0.97, "grad_norm": 10.632936093463568, "learning_rate": 2.422599827022809e-08, "loss": 0.4634, "step": 13521 }, { "epoch": 0.97, "grad_norm": 5.3058729754235365, "learning_rate": 2.4111951275857527e-08, "loss": 0.8888, "step": 13522 }, { "epoch": 0.97, "grad_norm": 3.398406868670049, "learning_rate": 2.399817271181215e-08, "loss": 0.1906, "step": 13523 }, { "epoch": 0.97, "grad_norm": 3.0111908864693793, "learning_rate": 2.388466258423039e-08, "loss": 0.1771, "step": 13524 }, { "epoch": 0.97, "grad_norm": 4.331592475819171, "learning_rate": 2.377142089923512e-08, "loss": 0.7259, "step": 13525 }, { "epoch": 0.97, "grad_norm": 4.707961016610933, "learning_rate": 2.36584476629359e-08, "loss": 0.9774, "step": 13526 }, { "epoch": 0.97, "grad_norm": 3.3494288624971196, "learning_rate": 2.3545742881426748e-08, "loss": 0.1756, "step": 13527 }, { "epoch": 0.97, "grad_norm": 3.0373311718498424, "learning_rate": 2.343330656078835e-08, "loss": 0.1746, "step": 13528 }, { "epoch": 0.97, "grad_norm": 15.403757131840987, "learning_rate": 2.33211387070853e-08, "loss": 0.7369, "step": 13529 }, { "epoch": 0.97, "grad_norm": 3.628093551871112, "learning_rate": 2.3209239326369427e-08, "loss": 0.1816, "step": 13530 }, { "epoch": 0.97, "grad_norm": 4.473787077761858, "learning_rate": 2.3097608424676456e-08, "loss": 0.1887, "step": 13531 }, { "epoch": 0.97, "grad_norm": 3.017441149538655, "learning_rate": 2.2986246008029343e-08, "loss": 0.0954, "step": 13532 }, { "epoch": 0.97, "grad_norm": 3.534587461055645, "learning_rate": 2.287515208243496e-08, "loss": 0.2235, "step": 13533 }, { "epoch": 0.97, "grad_norm": 13.491463393874573, "learning_rate": 2.2764326653886836e-08, "loss": 0.4884, "step": 13534 }, { "epoch": 0.97, "grad_norm": 2.7866451974064277, "learning_rate": 2.2653769728362974e-08, "loss": 0.1214, "step": 13535 }, { "epoch": 0.97, "grad_norm": 4.661883928687054, "learning_rate": 2.2543481311828595e-08, "loss": 0.6701, "step": 13536 }, { "epoch": 0.97, "grad_norm": 4.461473291392034, "learning_rate": 2.2433461410232282e-08, "loss": 0.3096, "step": 13537 }, { "epoch": 0.97, "grad_norm": 3.4740249182534906, "learning_rate": 2.2323710029509836e-08, "loss": 0.1516, "step": 13538 }, { "epoch": 0.97, "grad_norm": 4.49844547191831, "learning_rate": 2.2214227175581525e-08, "loss": 0.7034, "step": 13539 }, { "epoch": 0.97, "grad_norm": 4.932528237416027, "learning_rate": 2.2105012854353736e-08, "loss": 0.8717, "step": 13540 }, { "epoch": 0.97, "grad_norm": 3.8447377761821437, "learning_rate": 2.1996067071718417e-08, "loss": 0.6069, "step": 13541 }, { "epoch": 0.97, "grad_norm": 2.375883455220095, "learning_rate": 2.188738983355254e-08, "loss": 0.0927, "step": 13542 }, { "epoch": 0.97, "grad_norm": 4.2564065608772, "learning_rate": 2.1778981145718635e-08, "loss": 0.7551, "step": 13543 }, { "epoch": 0.97, "grad_norm": 19.586738930443044, "learning_rate": 2.1670841014064802e-08, "loss": 0.6073, "step": 13544 }, { "epoch": 0.97, "grad_norm": 19.00263677561075, "learning_rate": 2.1562969444425263e-08, "loss": 0.9083, "step": 13545 }, { "epoch": 0.97, "grad_norm": 3.258167775321072, "learning_rate": 2.145536644261925e-08, "loss": 0.1681, "step": 13546 }, { "epoch": 0.97, "grad_norm": 3.9669316767950855, "learning_rate": 2.134803201445157e-08, "loss": 0.2128, "step": 13547 }, { "epoch": 0.97, "grad_norm": 3.013710327779311, "learning_rate": 2.124096616571203e-08, "loss": 0.1091, "step": 13548 }, { "epoch": 0.97, "grad_norm": 3.9221412048113335, "learning_rate": 2.1134168902176565e-08, "loss": 0.2119, "step": 13549 }, { "epoch": 0.97, "grad_norm": 4.076936003853364, "learning_rate": 2.102764022960724e-08, "loss": 0.1531, "step": 13550 }, { "epoch": 0.97, "grad_norm": 31.732304830398085, "learning_rate": 2.0921380153750004e-08, "loss": 1.2841, "step": 13551 }, { "epoch": 0.97, "grad_norm": 5.346009193122248, "learning_rate": 2.08153886803375e-08, "loss": 0.705, "step": 13552 }, { "epoch": 0.97, "grad_norm": 4.195564693580674, "learning_rate": 2.0709665815087375e-08, "loss": 0.1845, "step": 13553 }, { "epoch": 0.97, "grad_norm": 5.221436390178417, "learning_rate": 2.0604211563703402e-08, "loss": 0.7309, "step": 13554 }, { "epoch": 0.97, "grad_norm": 22.539798488472776, "learning_rate": 2.0499025931873807e-08, "loss": 0.5458, "step": 13555 }, { "epoch": 0.97, "grad_norm": 2.799091544366348, "learning_rate": 2.039410892527349e-08, "loss": 0.1057, "step": 13556 }, { "epoch": 0.97, "grad_norm": 4.656571283719118, "learning_rate": 2.028946054956238e-08, "loss": 0.785, "step": 13557 }, { "epoch": 0.97, "grad_norm": 2.796897101537819, "learning_rate": 2.0185080810385393e-08, "loss": 0.1038, "step": 13558 }, { "epoch": 0.97, "grad_norm": 3.772798232805194, "learning_rate": 2.0080969713374142e-08, "loss": 0.2026, "step": 13559 }, { "epoch": 0.97, "grad_norm": 12.509156961172522, "learning_rate": 1.9977127264144136e-08, "loss": 0.3512, "step": 13560 }, { "epoch": 0.97, "grad_norm": 4.346548980693828, "learning_rate": 1.9873553468297557e-08, "loss": 0.7649, "step": 13561 }, { "epoch": 0.97, "grad_norm": 2.9096531189458616, "learning_rate": 1.977024833142216e-08, "loss": 0.0795, "step": 13562 }, { "epoch": 0.97, "grad_norm": 2.8436773858725704, "learning_rate": 1.9667211859090707e-08, "loss": 0.1806, "step": 13563 }, { "epoch": 0.97, "grad_norm": 3.46118278831248, "learning_rate": 1.9564444056861532e-08, "loss": 0.2295, "step": 13564 }, { "epoch": 0.97, "grad_norm": 4.379050269463572, "learning_rate": 1.9461944930278533e-08, "loss": 0.7872, "step": 13565 }, { "epoch": 0.97, "grad_norm": 2.2678322773678667, "learning_rate": 1.9359714484871726e-08, "loss": 0.0902, "step": 13566 }, { "epoch": 0.97, "grad_norm": 2.6575762312526723, "learning_rate": 1.9257752726155042e-08, "loss": 0.1322, "step": 13567 }, { "epoch": 0.97, "grad_norm": 8.656092036022498, "learning_rate": 1.9156059659629633e-08, "loss": 0.291, "step": 13568 }, { "epoch": 0.97, "grad_norm": 3.1823422006538653, "learning_rate": 1.9054635290781663e-08, "loss": 0.2407, "step": 13569 }, { "epoch": 0.97, "grad_norm": 3.4481426583696075, "learning_rate": 1.8953479625081762e-08, "loss": 0.2592, "step": 13570 }, { "epoch": 0.97, "grad_norm": 4.82269515140315, "learning_rate": 1.8852592667987225e-08, "loss": 0.7438, "step": 13571 }, { "epoch": 0.97, "grad_norm": 10.53354758649268, "learning_rate": 1.875197442494092e-08, "loss": 0.4556, "step": 13572 }, { "epoch": 0.97, "grad_norm": 19.59938788711966, "learning_rate": 1.8651624901370736e-08, "loss": 0.5885, "step": 13573 }, { "epoch": 0.97, "grad_norm": 19.043541928567993, "learning_rate": 1.8551544102689556e-08, "loss": 0.5875, "step": 13574 }, { "epoch": 0.97, "grad_norm": 4.715266861591209, "learning_rate": 1.8451732034296956e-08, "loss": 0.824, "step": 13575 }, { "epoch": 0.97, "grad_norm": 4.447882980565077, "learning_rate": 1.8352188701576957e-08, "loss": 0.5821, "step": 13576 }, { "epoch": 0.97, "grad_norm": 14.385450934232676, "learning_rate": 1.8252914109900265e-08, "loss": 0.6612, "step": 13577 }, { "epoch": 0.97, "grad_norm": 4.196907048916603, "learning_rate": 1.8153908264622045e-08, "loss": 0.2369, "step": 13578 }, { "epoch": 0.97, "grad_norm": 3.788456862241957, "learning_rate": 1.8055171171082464e-08, "loss": 0.2292, "step": 13579 }, { "epoch": 0.97, "grad_norm": 5.539887428656388, "learning_rate": 1.795670283460893e-08, "loss": 0.161, "step": 13580 }, { "epoch": 0.97, "grad_norm": 15.156461613480825, "learning_rate": 1.785850326051386e-08, "loss": 0.4404, "step": 13581 }, { "epoch": 0.97, "grad_norm": 12.80611009664442, "learning_rate": 1.776057245409357e-08, "loss": 0.2868, "step": 13582 }, { "epoch": 0.97, "grad_norm": 3.3975828706164237, "learning_rate": 1.7662910420631617e-08, "loss": 0.1158, "step": 13583 }, { "epoch": 0.97, "grad_norm": 23.04406726222915, "learning_rate": 1.756551716539656e-08, "loss": 1.0725, "step": 13584 }, { "epoch": 0.97, "grad_norm": 5.380986894504453, "learning_rate": 1.7468392693641978e-08, "loss": 0.7948, "step": 13585 }, { "epoch": 0.97, "grad_norm": 4.725471145058794, "learning_rate": 1.7371537010608118e-08, "loss": 0.2142, "step": 13586 }, { "epoch": 0.97, "grad_norm": 3.616923125902312, "learning_rate": 1.727495012151914e-08, "loss": 0.1149, "step": 13587 }, { "epoch": 0.97, "grad_norm": 4.125860583305109, "learning_rate": 1.7178632031586427e-08, "loss": 0.1276, "step": 13588 }, { "epoch": 0.97, "grad_norm": 3.603282600143518, "learning_rate": 1.708258274600527e-08, "loss": 0.145, "step": 13589 }, { "epoch": 0.97, "grad_norm": 12.095048668774616, "learning_rate": 1.6986802269957637e-08, "loss": 0.697, "step": 13590 }, { "epoch": 0.97, "grad_norm": 2.777591045689233, "learning_rate": 1.68912906086105e-08, "loss": 0.1162, "step": 13591 }, { "epoch": 0.97, "grad_norm": 3.715575351574549, "learning_rate": 1.6796047767115298e-08, "loss": 0.1651, "step": 13592 }, { "epoch": 0.97, "grad_norm": 4.012249819755517, "learning_rate": 1.6701073750611807e-08, "loss": 0.1871, "step": 13593 }, { "epoch": 0.97, "grad_norm": 10.508254574473295, "learning_rate": 1.660636856422204e-08, "loss": 0.7331, "step": 13594 }, { "epoch": 0.97, "grad_norm": 4.275742691045085, "learning_rate": 1.6511932213055247e-08, "loss": 0.812, "step": 13595 }, { "epoch": 0.97, "grad_norm": 4.689497644181111, "learning_rate": 1.6417764702206797e-08, "loss": 0.9248, "step": 13596 }, { "epoch": 0.98, "grad_norm": 3.6429500671297466, "learning_rate": 1.6323866036755954e-08, "loss": 0.163, "step": 13597 }, { "epoch": 0.98, "grad_norm": 4.488331882101061, "learning_rate": 1.6230236221768114e-08, "loss": 0.8274, "step": 13598 }, { "epoch": 0.98, "grad_norm": 2.938648798222052, "learning_rate": 1.6136875262294238e-08, "loss": 0.1429, "step": 13599 }, { "epoch": 0.98, "grad_norm": 3.1141236491233055, "learning_rate": 1.604378316337141e-08, "loss": 0.1808, "step": 13600 }, { "epoch": 0.98, "grad_norm": 12.598526118901091, "learning_rate": 1.595095993002116e-08, "loss": 0.4838, "step": 13601 }, { "epoch": 0.98, "grad_norm": 4.321757915815953, "learning_rate": 1.5858405567250602e-08, "loss": 0.8281, "step": 13602 }, { "epoch": 0.98, "grad_norm": 3.359855275881563, "learning_rate": 1.5766120080053516e-08, "loss": 0.1714, "step": 13603 }, { "epoch": 0.98, "grad_norm": 4.185426900245417, "learning_rate": 1.5674103473407586e-08, "loss": 0.731, "step": 13604 }, { "epoch": 0.98, "grad_norm": 3.253156517866239, "learning_rate": 1.558235575227718e-08, "loss": 0.1388, "step": 13605 }, { "epoch": 0.98, "grad_norm": 4.178723590398584, "learning_rate": 1.5490876921611664e-08, "loss": 0.2076, "step": 13606 }, { "epoch": 0.98, "grad_norm": 2.310059199812936, "learning_rate": 1.5399666986345983e-08, "loss": 0.0765, "step": 13607 }, { "epoch": 0.98, "grad_norm": 21.95180289276702, "learning_rate": 1.5308725951400095e-08, "loss": 0.9685, "step": 13608 }, { "epoch": 0.98, "grad_norm": 3.984293017993808, "learning_rate": 1.5218053821680624e-08, "loss": 0.2085, "step": 13609 }, { "epoch": 0.98, "grad_norm": 3.0810992811539903, "learning_rate": 1.5127650602078658e-08, "loss": 0.11, "step": 13610 }, { "epoch": 0.98, "grad_norm": 5.391317018879871, "learning_rate": 1.503751629747141e-08, "loss": 0.3218, "step": 13611 }, { "epoch": 0.98, "grad_norm": 31.33144963769091, "learning_rate": 1.49476509127211e-08, "loss": 0.9217, "step": 13612 }, { "epoch": 0.98, "grad_norm": 10.808719573833379, "learning_rate": 1.4858054452675519e-08, "loss": 0.5342, "step": 13613 }, { "epoch": 0.98, "grad_norm": 3.1341549445954904, "learning_rate": 1.4768726922168019e-08, "loss": 0.1433, "step": 13614 }, { "epoch": 0.98, "grad_norm": 3.7705632865221244, "learning_rate": 1.4679668326017527e-08, "loss": 0.1962, "step": 13615 }, { "epoch": 0.98, "grad_norm": 2.830033209920084, "learning_rate": 1.459087866902853e-08, "loss": 0.138, "step": 13616 }, { "epoch": 0.98, "grad_norm": 19.312021910514876, "learning_rate": 1.4502357955991086e-08, "loss": 0.8139, "step": 13617 }, { "epoch": 0.98, "grad_norm": 3.040392676270233, "learning_rate": 1.4414106191679711e-08, "loss": 0.154, "step": 13618 }, { "epoch": 0.98, "grad_norm": 3.822033728523964, "learning_rate": 1.4326123380856149e-08, "loss": 0.1823, "step": 13619 }, { "epoch": 0.98, "grad_norm": 3.185488751298225, "learning_rate": 1.4238409528266606e-08, "loss": 0.0864, "step": 13620 }, { "epoch": 0.98, "grad_norm": 3.4155711599940597, "learning_rate": 1.4150964638642296e-08, "loss": 0.1803, "step": 13621 }, { "epoch": 0.98, "grad_norm": 3.1231920560291493, "learning_rate": 1.4063788716701665e-08, "loss": 0.1372, "step": 13622 }, { "epoch": 0.98, "grad_norm": 3.2615745114580132, "learning_rate": 1.3976881767145955e-08, "loss": 0.0981, "step": 13623 }, { "epoch": 0.98, "grad_norm": 3.192709867628937, "learning_rate": 1.3890243794665303e-08, "loss": 0.1198, "step": 13624 }, { "epoch": 0.98, "grad_norm": 4.729111122255743, "learning_rate": 1.3803874803932083e-08, "loss": 0.8716, "step": 13625 }, { "epoch": 0.98, "grad_norm": 4.040634040152941, "learning_rate": 1.3717774799605898e-08, "loss": 0.2236, "step": 13626 }, { "epoch": 0.98, "grad_norm": 3.0951101416387012, "learning_rate": 1.3631943786331925e-08, "loss": 0.0942, "step": 13627 }, { "epoch": 0.98, "grad_norm": 2.932868146185084, "learning_rate": 1.3546381768739791e-08, "loss": 0.097, "step": 13628 }, { "epoch": 0.98, "grad_norm": 3.246329140156411, "learning_rate": 1.3461088751445806e-08, "loss": 0.1595, "step": 13629 }, { "epoch": 0.98, "grad_norm": 15.277813245855974, "learning_rate": 1.3376064739051287e-08, "loss": 0.4907, "step": 13630 }, { "epoch": 0.98, "grad_norm": 11.842598488905036, "learning_rate": 1.3291309736142566e-08, "loss": 0.4981, "step": 13631 }, { "epoch": 0.98, "grad_norm": 13.29368534982156, "learning_rate": 1.3206823747291542e-08, "loss": 0.7156, "step": 13632 }, { "epoch": 0.98, "grad_norm": 4.824822812567741, "learning_rate": 1.312260677705679e-08, "loss": 0.7829, "step": 13633 }, { "epoch": 0.98, "grad_norm": 16.36706895228007, "learning_rate": 1.3038658829980788e-08, "loss": 0.5937, "step": 13634 }, { "epoch": 0.98, "grad_norm": 3.3453549307931816, "learning_rate": 1.295497991059269e-08, "loss": 0.1724, "step": 13635 }, { "epoch": 0.98, "grad_norm": 4.616419797840763, "learning_rate": 1.287157002340611e-08, "loss": 0.1623, "step": 13636 }, { "epoch": 0.98, "grad_norm": 3.4741211871204696, "learning_rate": 1.2788429172921336e-08, "loss": 0.1396, "step": 13637 }, { "epoch": 0.98, "grad_norm": 11.361319311897622, "learning_rate": 1.270555736362311e-08, "loss": 0.5415, "step": 13638 }, { "epoch": 0.98, "grad_norm": 7.255838104484499, "learning_rate": 1.2622954599982307e-08, "loss": 0.222, "step": 13639 }, { "epoch": 0.98, "grad_norm": 3.175821507746599, "learning_rate": 1.2540620886454801e-08, "loss": 0.1724, "step": 13640 }, { "epoch": 0.98, "grad_norm": 31.466224547825735, "learning_rate": 1.2458556227482044e-08, "loss": 0.6326, "step": 13641 }, { "epoch": 0.98, "grad_norm": 5.245363263607044, "learning_rate": 1.2376760627491047e-08, "loss": 0.8164, "step": 13642 }, { "epoch": 0.98, "grad_norm": 3.4308164217773376, "learning_rate": 1.2295234090894947e-08, "loss": 0.1521, "step": 13643 }, { "epoch": 0.98, "grad_norm": 8.569527371107837, "learning_rate": 1.2213976622091895e-08, "loss": 0.4291, "step": 13644 }, { "epoch": 0.98, "grad_norm": 4.669975635397872, "learning_rate": 1.2132988225464492e-08, "loss": 0.9173, "step": 13645 }, { "epoch": 0.98, "grad_norm": 17.238858239027593, "learning_rate": 1.2052268905382025e-08, "loss": 0.5885, "step": 13646 }, { "epoch": 0.98, "grad_norm": 4.205843405512096, "learning_rate": 1.1971818666199898e-08, "loss": 0.1247, "step": 13647 }, { "epoch": 0.98, "grad_norm": 3.2897270154195577, "learning_rate": 1.189163751225686e-08, "loss": 0.0872, "step": 13648 }, { "epoch": 0.98, "grad_norm": 15.052347300346705, "learning_rate": 1.1811725447879453e-08, "loss": 0.5343, "step": 13649 }, { "epoch": 0.98, "grad_norm": 2.347217739122812, "learning_rate": 1.1732082477378114e-08, "loss": 0.0853, "step": 13650 }, { "epoch": 0.98, "grad_norm": 10.679550081034908, "learning_rate": 1.1652708605048857e-08, "loss": 0.4454, "step": 13651 }, { "epoch": 0.98, "grad_norm": 11.370047307691442, "learning_rate": 1.1573603835174363e-08, "loss": 0.4267, "step": 13652 }, { "epoch": 0.98, "grad_norm": 3.5333308240490484, "learning_rate": 1.1494768172021776e-08, "loss": 0.1632, "step": 13653 }, { "epoch": 0.98, "grad_norm": 2.882310893980596, "learning_rate": 1.1416201619843802e-08, "loss": 0.1937, "step": 13654 }, { "epoch": 0.98, "grad_norm": 13.080780697703023, "learning_rate": 1.1337904182879278e-08, "loss": 0.387, "step": 13655 }, { "epoch": 0.98, "grad_norm": 3.0787202167230987, "learning_rate": 1.1259875865350932e-08, "loss": 0.1962, "step": 13656 }, { "epoch": 0.98, "grad_norm": 4.261626774617109, "learning_rate": 1.118211667146929e-08, "loss": 0.7482, "step": 13657 }, { "epoch": 0.98, "grad_norm": 25.66673035972616, "learning_rate": 1.1104626605428769e-08, "loss": 1.3739, "step": 13658 }, { "epoch": 0.98, "grad_norm": 5.256989844489481, "learning_rate": 1.1027405671409919e-08, "loss": 0.8209, "step": 13659 }, { "epoch": 0.98, "grad_norm": 4.166129206298516, "learning_rate": 1.0950453873577738e-08, "loss": 0.2328, "step": 13660 }, { "epoch": 0.98, "grad_norm": 3.845028328372938, "learning_rate": 1.0873771216084461e-08, "loss": 0.1243, "step": 13661 }, { "epoch": 0.98, "grad_norm": 7.309750786273567, "learning_rate": 1.0797357703066225e-08, "loss": 0.4554, "step": 13662 }, { "epoch": 0.98, "grad_norm": 2.926070104120978, "learning_rate": 1.0721213338645287e-08, "loss": 0.0878, "step": 13663 }, { "epoch": 0.98, "grad_norm": 15.201373451029966, "learning_rate": 1.0645338126929473e-08, "loss": 0.9692, "step": 13664 }, { "epoch": 0.98, "grad_norm": 3.5500283851389423, "learning_rate": 1.0569732072012174e-08, "loss": 0.1933, "step": 13665 }, { "epoch": 0.98, "grad_norm": 2.927453459765775, "learning_rate": 1.0494395177971239e-08, "loss": 0.1224, "step": 13666 }, { "epoch": 0.98, "grad_norm": 10.140834544344274, "learning_rate": 1.0419327448871752e-08, "loss": 0.6122, "step": 13667 }, { "epoch": 0.98, "grad_norm": 29.181771810715112, "learning_rate": 1.034452888876325e-08, "loss": 0.8779, "step": 13668 }, { "epoch": 0.98, "grad_norm": 4.731727215783691, "learning_rate": 1.0269999501680284e-08, "loss": 0.2492, "step": 13669 }, { "epoch": 0.98, "grad_norm": 5.46015519432637, "learning_rate": 1.0195739291643526e-08, "loss": 0.8048, "step": 13670 }, { "epoch": 0.98, "grad_norm": 15.819259018564988, "learning_rate": 1.0121748262659215e-08, "loss": 1.0388, "step": 13671 }, { "epoch": 0.98, "grad_norm": 14.640300942387999, "learning_rate": 1.0048026418719159e-08, "loss": 0.7434, "step": 13672 }, { "epoch": 0.98, "grad_norm": 21.175238890950272, "learning_rate": 9.974573763799621e-09, "loss": 0.4136, "step": 13673 }, { "epoch": 0.98, "grad_norm": 19.181930829954542, "learning_rate": 9.9013903018641e-09, "loss": 0.7893, "step": 13674 }, { "epoch": 0.98, "grad_norm": 16.851498165890025, "learning_rate": 9.828476036859436e-09, "loss": 0.9915, "step": 13675 }, { "epoch": 0.98, "grad_norm": 13.718563945783577, "learning_rate": 9.755830972720259e-09, "loss": 0.3987, "step": 13676 }, { "epoch": 0.98, "grad_norm": 3.799995521331063, "learning_rate": 9.683455113364548e-09, "loss": 0.83, "step": 13677 }, { "epoch": 0.98, "grad_norm": 3.4514978137292363, "learning_rate": 9.611348462696956e-09, "loss": 0.1435, "step": 13678 }, { "epoch": 0.98, "grad_norm": 4.416198631067048, "learning_rate": 9.539511024607706e-09, "loss": 0.167, "step": 13679 }, { "epoch": 0.98, "grad_norm": 20.87257018684411, "learning_rate": 9.46794280297203e-09, "loss": 0.7141, "step": 13680 }, { "epoch": 0.98, "grad_norm": 15.104005611174609, "learning_rate": 9.396643801650174e-09, "loss": 0.4233, "step": 13681 }, { "epoch": 0.98, "grad_norm": 4.315815559822798, "learning_rate": 9.32561402448906e-09, "loss": 0.146, "step": 13682 }, { "epoch": 0.98, "grad_norm": 2.851785897474868, "learning_rate": 9.254853475320623e-09, "loss": 0.1733, "step": 13683 }, { "epoch": 0.98, "grad_norm": 4.765474225044386, "learning_rate": 9.184362157961812e-09, "loss": 0.6031, "step": 13684 }, { "epoch": 0.98, "grad_norm": 16.613225033641974, "learning_rate": 9.11414007621514e-09, "loss": 0.6969, "step": 13685 }, { "epoch": 0.98, "grad_norm": 8.810078680066692, "learning_rate": 9.04418723386924e-09, "loss": 0.3886, "step": 13686 }, { "epoch": 0.98, "grad_norm": 16.795029038593135, "learning_rate": 8.974503634697762e-09, "loss": 0.7729, "step": 13687 }, { "epoch": 0.98, "grad_norm": 3.6269487892270984, "learning_rate": 8.905089282459923e-09, "loss": 0.1172, "step": 13688 }, { "epoch": 0.98, "grad_norm": 3.3417688291219174, "learning_rate": 8.835944180899947e-09, "loss": 0.1445, "step": 13689 }, { "epoch": 0.98, "grad_norm": 2.606626780866028, "learning_rate": 8.767068333748185e-09, "loss": 0.0921, "step": 13690 }, { "epoch": 0.98, "grad_norm": 2.441090184048966, "learning_rate": 8.698461744719999e-09, "loss": 0.0848, "step": 13691 }, { "epoch": 0.98, "grad_norm": 19.602208379002033, "learning_rate": 8.630124417516871e-09, "loss": 0.4904, "step": 13692 }, { "epoch": 0.98, "grad_norm": 23.232367187388338, "learning_rate": 8.562056355825298e-09, "loss": 1.1337, "step": 13693 }, { "epoch": 0.98, "grad_norm": 3.931534138667245, "learning_rate": 8.494257563317343e-09, "loss": 0.2413, "step": 13694 }, { "epoch": 0.98, "grad_norm": 3.393768408961373, "learning_rate": 8.426728043650079e-09, "loss": 0.0898, "step": 13695 }, { "epoch": 0.98, "grad_norm": 14.771894570648806, "learning_rate": 8.359467800466703e-09, "loss": 0.4654, "step": 13696 }, { "epoch": 0.98, "grad_norm": 29.982669759007216, "learning_rate": 8.292476837395425e-09, "loss": 1.417, "step": 13697 }, { "epoch": 0.98, "grad_norm": 4.373872379841674, "learning_rate": 8.225755158050574e-09, "loss": 0.2009, "step": 13698 }, { "epoch": 0.98, "grad_norm": 7.757021152341624, "learning_rate": 8.159302766031497e-09, "loss": 0.4164, "step": 13699 }, { "epoch": 0.98, "grad_norm": 4.747536265245255, "learning_rate": 8.093119664922544e-09, "loss": 0.88, "step": 13700 }, { "epoch": 0.98, "grad_norm": 20.47739827769594, "learning_rate": 8.02720585829475e-09, "loss": 0.6532, "step": 13701 }, { "epoch": 0.98, "grad_norm": 9.835931794380745, "learning_rate": 7.961561349703607e-09, "loss": 0.3241, "step": 13702 }, { "epoch": 0.98, "grad_norm": 2.2009594464133855, "learning_rate": 7.896186142690166e-09, "loss": 0.0752, "step": 13703 }, { "epoch": 0.98, "grad_norm": 3.728411608670489, "learning_rate": 7.831080240781053e-09, "loss": 0.5941, "step": 13704 }, { "epoch": 0.98, "grad_norm": 2.96113457689016, "learning_rate": 7.766243647489013e-09, "loss": 0.1655, "step": 13705 }, { "epoch": 0.98, "grad_norm": 4.337660264701831, "learning_rate": 7.701676366311805e-09, "loss": 0.8892, "step": 13706 }, { "epoch": 0.98, "grad_norm": 45.161601934660716, "learning_rate": 7.637378400732198e-09, "loss": 0.9494, "step": 13707 }, { "epoch": 0.98, "grad_norm": 4.978771969954928, "learning_rate": 7.57334975421853e-09, "loss": 0.9095, "step": 13708 }, { "epoch": 0.98, "grad_norm": 5.019532067598731, "learning_rate": 7.509590430225255e-09, "loss": 1.0918, "step": 13709 }, { "epoch": 0.98, "grad_norm": 4.398445183896871, "learning_rate": 7.446100432192404e-09, "loss": 0.7774, "step": 13710 }, { "epoch": 0.98, "grad_norm": 3.6805062202758787, "learning_rate": 7.382879763544459e-09, "loss": 0.247, "step": 13711 }, { "epoch": 0.98, "grad_norm": 3.6282472144605293, "learning_rate": 7.319928427692025e-09, "loss": 0.1732, "step": 13712 }, { "epoch": 0.98, "grad_norm": 9.831188898839844, "learning_rate": 7.2572464280312725e-09, "loss": 0.4767, "step": 13713 }, { "epoch": 0.98, "grad_norm": 19.034336819420393, "learning_rate": 7.194833767943388e-09, "loss": 0.8542, "step": 13714 }, { "epoch": 0.98, "grad_norm": 4.448266766339373, "learning_rate": 7.132690450795121e-09, "loss": 0.1036, "step": 13715 }, { "epoch": 0.98, "grad_norm": 5.224396379188854, "learning_rate": 7.070816479939347e-09, "loss": 0.8095, "step": 13716 }, { "epoch": 0.98, "grad_norm": 4.091825507787856, "learning_rate": 7.009211858713949e-09, "loss": 0.1417, "step": 13717 }, { "epoch": 0.98, "grad_norm": 3.2191338151468942, "learning_rate": 6.947876590442382e-09, "loss": 0.153, "step": 13718 }, { "epoch": 0.98, "grad_norm": 3.4548245722419963, "learning_rate": 6.886810678432554e-09, "loss": 0.0941, "step": 13719 }, { "epoch": 0.98, "grad_norm": 3.342437907705243, "learning_rate": 6.826014125979608e-09, "loss": 0.1359, "step": 13720 }, { "epoch": 0.98, "grad_norm": 36.11732689233431, "learning_rate": 6.765486936362586e-09, "loss": 1.4581, "step": 13721 }, { "epoch": 0.98, "grad_norm": 4.764952831985201, "learning_rate": 6.7052291128477665e-09, "loss": 0.7231, "step": 13722 }, { "epoch": 0.98, "grad_norm": 4.85829848937561, "learning_rate": 6.645240658684771e-09, "loss": 0.7, "step": 13723 }, { "epoch": 0.98, "grad_norm": 4.1379173875388044, "learning_rate": 6.585521577110454e-09, "loss": 0.1869, "step": 13724 }, { "epoch": 0.98, "grad_norm": 18.16133941528796, "learning_rate": 6.526071871345574e-09, "loss": 0.6631, "step": 13725 }, { "epoch": 0.98, "grad_norm": 3.4164579937917305, "learning_rate": 6.466891544598675e-09, "loss": 0.153, "step": 13726 }, { "epoch": 0.98, "grad_norm": 3.3449706579051384, "learning_rate": 6.407980600060537e-09, "loss": 0.1435, "step": 13727 }, { "epoch": 0.98, "grad_norm": 3.3388563526940427, "learning_rate": 6.34933904091084e-09, "loss": 0.158, "step": 13728 }, { "epoch": 0.98, "grad_norm": 15.525586930129158, "learning_rate": 6.2909668703120545e-09, "loss": 0.6221, "step": 13729 }, { "epoch": 0.98, "grad_norm": 6.796068013840843, "learning_rate": 6.232864091412772e-09, "loss": 0.4707, "step": 13730 }, { "epoch": 0.98, "grad_norm": 8.60559211356765, "learning_rate": 6.175030707348817e-09, "loss": 0.3317, "step": 13731 }, { "epoch": 0.98, "grad_norm": 8.48486801143914, "learning_rate": 6.117466721238807e-09, "loss": 0.4035, "step": 13732 }, { "epoch": 0.98, "grad_norm": 4.069379222305758, "learning_rate": 6.0601721361891464e-09, "loss": 0.7295, "step": 13733 }, { "epoch": 0.98, "grad_norm": 18.91023055980306, "learning_rate": 6.003146955289585e-09, "loss": 0.9363, "step": 13734 }, { "epoch": 0.98, "grad_norm": 4.856400467938809, "learning_rate": 5.946391181617106e-09, "loss": 0.1533, "step": 13735 }, { "epoch": 0.98, "grad_norm": 3.1371196050796137, "learning_rate": 5.889904818233705e-09, "loss": 0.0949, "step": 13736 }, { "epoch": 0.99, "grad_norm": 9.954495615111389, "learning_rate": 5.833687868185833e-09, "loss": 0.6905, "step": 13737 }, { "epoch": 0.99, "grad_norm": 3.2289477327826117, "learning_rate": 5.7777403345066206e-09, "loss": 0.1656, "step": 13738 }, { "epoch": 0.99, "grad_norm": 6.034906528635373, "learning_rate": 5.722062220214209e-09, "loss": 0.8431, "step": 13739 }, { "epoch": 0.99, "grad_norm": 12.197492319047054, "learning_rate": 5.6666535283117495e-09, "loss": 0.4111, "step": 13740 }, { "epoch": 0.99, "grad_norm": 4.626077204430881, "learning_rate": 5.611514261789075e-09, "loss": 0.3265, "step": 13741 }, { "epoch": 0.99, "grad_norm": 3.6444811608786796, "learning_rate": 5.556644423620472e-09, "loss": 0.2152, "step": 13742 }, { "epoch": 0.99, "grad_norm": 4.139217409955151, "learning_rate": 5.502044016765795e-09, "loss": 1.0409, "step": 13743 }, { "epoch": 0.99, "grad_norm": 2.683937517364717, "learning_rate": 5.447713044170466e-09, "loss": 0.1464, "step": 13744 }, { "epoch": 0.99, "grad_norm": 14.203857262380575, "learning_rate": 5.393651508766029e-09, "loss": 0.4887, "step": 13745 }, { "epoch": 0.99, "grad_norm": 3.935231993651036, "learning_rate": 5.339859413468485e-09, "loss": 0.164, "step": 13746 }, { "epoch": 0.99, "grad_norm": 5.112629669669172, "learning_rate": 5.2863367611794005e-09, "loss": 0.7943, "step": 13747 }, { "epoch": 0.99, "grad_norm": 18.90033261250447, "learning_rate": 5.233083554786466e-09, "loss": 0.8496, "step": 13748 }, { "epoch": 0.99, "grad_norm": 5.874207357515349, "learning_rate": 5.18009979716294e-09, "loss": 0.1764, "step": 13749 }, { "epoch": 0.99, "grad_norm": 3.592367410599165, "learning_rate": 5.127385491165981e-09, "loss": 0.5683, "step": 13750 }, { "epoch": 0.99, "grad_norm": 2.0412663873774974, "learning_rate": 5.074940639639981e-09, "loss": 0.0753, "step": 13751 }, { "epoch": 0.99, "grad_norm": 3.8255252774612827, "learning_rate": 5.022765245414341e-09, "loss": 0.7284, "step": 13752 }, { "epoch": 0.99, "grad_norm": 8.31354567915484, "learning_rate": 4.970859311303478e-09, "loss": 0.4618, "step": 13753 }, { "epoch": 0.99, "grad_norm": 5.659639504414805, "learning_rate": 4.919222840107929e-09, "loss": 1.0758, "step": 13754 }, { "epoch": 0.99, "grad_norm": 5.215742574332375, "learning_rate": 4.867855834612134e-09, "loss": 0.6898, "step": 13755 }, { "epoch": 0.99, "grad_norm": 4.2960042612021025, "learning_rate": 4.8167582975883195e-09, "loss": 0.7513, "step": 13756 }, { "epoch": 0.99, "grad_norm": 11.128225382780213, "learning_rate": 4.7659302317926144e-09, "loss": 0.5001, "step": 13757 }, { "epoch": 0.99, "grad_norm": 23.7271081026618, "learning_rate": 4.71537163996727e-09, "loss": 0.6627, "step": 13758 }, { "epoch": 0.99, "grad_norm": 2.9290587442808076, "learning_rate": 4.665082524838993e-09, "loss": 0.1107, "step": 13759 }, { "epoch": 0.99, "grad_norm": 17.397000129927193, "learning_rate": 4.615062889121169e-09, "loss": 0.9931, "step": 13760 }, { "epoch": 0.99, "grad_norm": 24.546619543086717, "learning_rate": 4.565312735512195e-09, "loss": 0.8371, "step": 13761 }, { "epoch": 0.99, "grad_norm": 4.813641432844756, "learning_rate": 4.515832066696035e-09, "loss": 0.7789, "step": 13762 }, { "epoch": 0.99, "grad_norm": 2.736077399113728, "learning_rate": 4.466620885341666e-09, "loss": 0.0666, "step": 13763 }, { "epoch": 0.99, "grad_norm": 3.318912856941653, "learning_rate": 4.4176791941041855e-09, "loss": 0.1247, "step": 13764 }, { "epoch": 0.99, "grad_norm": 2.9656299816844762, "learning_rate": 4.369006995623148e-09, "loss": 0.0778, "step": 13765 }, { "epoch": 0.99, "grad_norm": 3.9175348676418547, "learning_rate": 4.3206042925253436e-09, "loss": 0.1349, "step": 13766 }, { "epoch": 0.99, "grad_norm": 9.09944529155802, "learning_rate": 4.272471087420904e-09, "loss": 0.512, "step": 13767 }, { "epoch": 0.99, "grad_norm": 2.66495151468212, "learning_rate": 4.224607382906642e-09, "loss": 0.1372, "step": 13768 }, { "epoch": 0.99, "grad_norm": 13.354462562406711, "learning_rate": 4.1770131815649376e-09, "loss": 0.638, "step": 13769 }, { "epoch": 0.99, "grad_norm": 5.9407934833973535, "learning_rate": 4.129688485963179e-09, "loss": 0.1451, "step": 13770 }, { "epoch": 0.99, "grad_norm": 23.133564297760667, "learning_rate": 4.082633298654326e-09, "loss": 1.0119, "step": 13771 }, { "epoch": 0.99, "grad_norm": 22.702581138926483, "learning_rate": 4.035847622176347e-09, "loss": 0.9618, "step": 13772 }, { "epoch": 0.99, "grad_norm": 42.88657076894663, "learning_rate": 3.989331459054446e-09, "loss": 0.556, "step": 13773 }, { "epoch": 0.99, "grad_norm": 3.3968115732430286, "learning_rate": 3.943084811796616e-09, "loss": 0.0756, "step": 13774 }, { "epoch": 0.99, "grad_norm": 3.5934244624325737, "learning_rate": 3.897107682898638e-09, "loss": 0.1609, "step": 13775 }, { "epoch": 0.99, "grad_norm": 5.000976920532406, "learning_rate": 3.851400074840195e-09, "loss": 0.1654, "step": 13776 }, { "epoch": 0.99, "grad_norm": 4.854365851168264, "learning_rate": 3.805961990087093e-09, "loss": 0.936, "step": 13777 }, { "epoch": 0.99, "grad_norm": 2.3427772397321407, "learning_rate": 3.760793431091258e-09, "loss": 0.0535, "step": 13778 }, { "epoch": 0.99, "grad_norm": 16.173332170969594, "learning_rate": 3.715894400288522e-09, "loss": 0.7947, "step": 13779 }, { "epoch": 0.99, "grad_norm": 3.2364978494616445, "learning_rate": 3.671264900100835e-09, "loss": 0.1202, "step": 13780 }, { "epoch": 0.99, "grad_norm": 4.503248157202642, "learning_rate": 3.6269049329368256e-09, "loss": 0.89, "step": 13781 }, { "epoch": 0.99, "grad_norm": 17.318346320061487, "learning_rate": 3.5828145011890247e-09, "loss": 0.6321, "step": 13782 }, { "epoch": 0.99, "grad_norm": 4.292458861246766, "learning_rate": 3.5389936072360853e-09, "loss": 0.8517, "step": 13783 }, { "epoch": 0.99, "grad_norm": 8.869572967068434, "learning_rate": 3.495442253441117e-09, "loss": 0.3176, "step": 13784 }, { "epoch": 0.99, "grad_norm": 4.746653817760634, "learning_rate": 3.452160442155017e-09, "loss": 0.736, "step": 13785 }, { "epoch": 0.99, "grad_norm": 21.577378572428298, "learning_rate": 3.409148175711474e-09, "loss": 0.6319, "step": 13786 }, { "epoch": 0.99, "grad_norm": 14.072395606887904, "learning_rate": 3.36640545643141e-09, "loss": 0.5609, "step": 13787 }, { "epoch": 0.99, "grad_norm": 9.858252212113161, "learning_rate": 3.323932286620202e-09, "loss": 0.506, "step": 13788 }, { "epoch": 0.99, "grad_norm": 3.2290110929709748, "learning_rate": 3.2817286685699057e-09, "loss": 0.1629, "step": 13789 }, { "epoch": 0.99, "grad_norm": 19.846492300639095, "learning_rate": 3.2397946045559237e-09, "loss": 0.9444, "step": 13790 }, { "epoch": 0.99, "grad_norm": 4.960252055042294, "learning_rate": 3.1981300968420002e-09, "loss": 0.8368, "step": 13791 }, { "epoch": 0.99, "grad_norm": 5.654140839511596, "learning_rate": 3.156735147674672e-09, "loss": 0.1581, "step": 13792 }, { "epoch": 0.99, "grad_norm": 4.093914051307108, "learning_rate": 3.1156097592877077e-09, "loss": 0.1173, "step": 13793 }, { "epoch": 0.99, "grad_norm": 17.05322618388386, "learning_rate": 3.0747539338993326e-09, "loss": 0.8165, "step": 13794 }, { "epoch": 0.99, "grad_norm": 3.4083557486957643, "learning_rate": 3.0341676737133397e-09, "loss": 0.1523, "step": 13795 }, { "epoch": 0.99, "grad_norm": 5.285525260234732, "learning_rate": 2.9938509809196435e-09, "loss": 0.6784, "step": 13796 }, { "epoch": 0.99, "grad_norm": 8.184860825913194, "learning_rate": 2.9538038576926166e-09, "loss": 0.3056, "step": 13797 }, { "epoch": 0.99, "grad_norm": 3.6099959114199476, "learning_rate": 2.9140263061938624e-09, "loss": 0.1346, "step": 13798 }, { "epoch": 0.99, "grad_norm": 4.04839890188053, "learning_rate": 2.874518328567777e-09, "loss": 0.1903, "step": 13799 }, { "epoch": 0.99, "grad_norm": 10.83768445082131, "learning_rate": 2.8352799269459884e-09, "loss": 0.4665, "step": 13800 }, { "epoch": 0.99, "grad_norm": 11.971170479517383, "learning_rate": 2.7963111034462476e-09, "loss": 0.804, "step": 13801 }, { "epoch": 0.99, "grad_norm": 16.705360798300198, "learning_rate": 2.757611860169651e-09, "loss": 0.6408, "step": 13802 }, { "epoch": 0.99, "grad_norm": 3.945487325033875, "learning_rate": 2.7191821992045285e-09, "loss": 0.1541, "step": 13803 }, { "epoch": 0.99, "grad_norm": 4.142999968304159, "learning_rate": 2.6810221226236666e-09, "loss": 0.1699, "step": 13804 }, { "epoch": 0.99, "grad_norm": 4.245028052657386, "learning_rate": 2.643131632485418e-09, "loss": 0.8396, "step": 13805 }, { "epoch": 0.99, "grad_norm": 2.8561260111837514, "learning_rate": 2.605510730834815e-09, "loss": 0.1535, "step": 13806 }, { "epoch": 0.99, "grad_norm": 3.9078339537328737, "learning_rate": 2.5681594197007888e-09, "loss": 0.2148, "step": 13807 }, { "epoch": 0.99, "grad_norm": 3.0406593660825427, "learning_rate": 2.5310777010978395e-09, "loss": 0.102, "step": 13808 }, { "epoch": 0.99, "grad_norm": 3.3839439131475104, "learning_rate": 2.494265577026589e-09, "loss": 0.1047, "step": 13809 }, { "epoch": 0.99, "grad_norm": 29.09001697352834, "learning_rate": 2.4577230494737814e-09, "loss": 0.5112, "step": 13810 }, { "epoch": 0.99, "grad_norm": 3.3222466960476176, "learning_rate": 2.4214501204095076e-09, "loss": 0.1719, "step": 13811 }, { "epoch": 0.99, "grad_norm": 3.102315468359057, "learning_rate": 2.3854467917916457e-09, "loss": 0.1065, "step": 13812 }, { "epoch": 0.99, "grad_norm": 3.006819397746877, "learning_rate": 2.3497130655614208e-09, "loss": 0.1523, "step": 13813 }, { "epoch": 0.99, "grad_norm": 16.861141100959873, "learning_rate": 2.3142489436472906e-09, "loss": 0.8608, "step": 13814 }, { "epoch": 0.99, "grad_norm": 3.4851987498388666, "learning_rate": 2.2790544279621686e-09, "loss": 0.1332, "step": 13815 }, { "epoch": 0.99, "grad_norm": 5.383959001465604, "learning_rate": 2.2441295204050916e-09, "loss": 0.9055, "step": 13816 }, { "epoch": 0.99, "grad_norm": 9.181985181685878, "learning_rate": 2.2094742228589984e-09, "loss": 0.2408, "step": 13817 }, { "epoch": 0.99, "grad_norm": 13.125469788961183, "learning_rate": 2.175088537194614e-09, "loss": 0.6096, "step": 13818 }, { "epoch": 0.99, "grad_norm": 6.403883351212438, "learning_rate": 2.1409724652660114e-09, "loss": 0.2528, "step": 13819 }, { "epoch": 0.99, "grad_norm": 12.636489569526745, "learning_rate": 2.10712600891394e-09, "loss": 0.6198, "step": 13820 }, { "epoch": 0.99, "grad_norm": 4.385069963822319, "learning_rate": 2.073549169964717e-09, "loss": 0.7628, "step": 13821 }, { "epoch": 0.99, "grad_norm": 12.331723444285544, "learning_rate": 2.040241950229671e-09, "loss": 0.5851, "step": 13822 }, { "epoch": 0.99, "grad_norm": 4.155383622172587, "learning_rate": 2.007204351504588e-09, "loss": 0.696, "step": 13823 }, { "epoch": 0.99, "grad_norm": 4.030712265528127, "learning_rate": 1.974436375572486e-09, "loss": 0.6343, "step": 13824 }, { "epoch": 0.99, "grad_norm": 4.233612842053331, "learning_rate": 1.9419380242013954e-09, "loss": 0.8479, "step": 13825 }, { "epoch": 0.99, "grad_norm": 2.676350320608069, "learning_rate": 1.9097092991438025e-09, "loss": 0.1343, "step": 13826 }, { "epoch": 0.99, "grad_norm": 3.1438844083424016, "learning_rate": 1.8777502021383174e-09, "loss": 0.2067, "step": 13827 }, { "epoch": 0.99, "grad_norm": 2.5318848726784218, "learning_rate": 1.8460607349096715e-09, "loss": 0.1206, "step": 13828 }, { "epoch": 0.99, "grad_norm": 2.785286349808898, "learning_rate": 1.8146408991670528e-09, "loss": 0.1236, "step": 13829 }, { "epoch": 0.99, "grad_norm": 2.9939013756428876, "learning_rate": 1.7834906966052169e-09, "loss": 0.1209, "step": 13830 }, { "epoch": 0.99, "grad_norm": 4.51931220099882, "learning_rate": 1.7526101289050411e-09, "loss": 0.1365, "step": 13831 }, { "epoch": 0.99, "grad_norm": 4.009264628346131, "learning_rate": 1.7219991977318606e-09, "loss": 0.9089, "step": 13832 }, { "epoch": 0.99, "grad_norm": 13.991743512682659, "learning_rate": 1.6916579047371317e-09, "loss": 0.3449, "step": 13833 }, { "epoch": 0.99, "grad_norm": 4.047036194458751, "learning_rate": 1.6615862515584336e-09, "loss": 0.7536, "step": 13834 }, { "epoch": 0.99, "grad_norm": 5.210195453175073, "learning_rate": 1.6317842398172468e-09, "loss": 0.351, "step": 13835 }, { "epoch": 0.99, "grad_norm": 32.656085292684566, "learning_rate": 1.6022518711211742e-09, "loss": 0.6526, "step": 13836 }, { "epoch": 0.99, "grad_norm": 3.1014973513965898, "learning_rate": 1.5729891470633863e-09, "loss": 0.1232, "step": 13837 }, { "epoch": 0.99, "grad_norm": 14.645149532899831, "learning_rate": 1.5439960692231747e-09, "loss": 0.4363, "step": 13838 }, { "epoch": 0.99, "grad_norm": 12.362337138038729, "learning_rate": 1.5152726391642891e-09, "loss": 0.366, "step": 13839 }, { "epoch": 0.99, "grad_norm": 2.259038963241693, "learning_rate": 1.4868188584360454e-09, "loss": 0.1003, "step": 13840 }, { "epoch": 0.99, "grad_norm": 3.990597345641571, "learning_rate": 1.4586347285733272e-09, "loss": 0.222, "step": 13841 }, { "epoch": 0.99, "grad_norm": 18.9855682780806, "learning_rate": 1.4307202510971396e-09, "loss": 0.7995, "step": 13842 }, { "epoch": 0.99, "grad_norm": 3.992231418417598, "learning_rate": 1.4030754275129454e-09, "loss": 0.1655, "step": 13843 }, { "epoch": 0.99, "grad_norm": 17.52856246771985, "learning_rate": 1.3757002593117742e-09, "loss": 0.6152, "step": 13844 }, { "epoch": 0.99, "grad_norm": 3.5038942574536227, "learning_rate": 1.3485947479707773e-09, "loss": 0.1411, "step": 13845 }, { "epoch": 0.99, "grad_norm": 3.4268274387065607, "learning_rate": 1.3217588949526738e-09, "loss": 0.1795, "step": 13846 }, { "epoch": 0.99, "grad_norm": 6.856061772599295, "learning_rate": 1.2951927017040845e-09, "loss": 0.4422, "step": 13847 }, { "epoch": 0.99, "grad_norm": 12.930168766924476, "learning_rate": 1.268896169658862e-09, "loss": 0.7757, "step": 13848 }, { "epoch": 0.99, "grad_norm": 2.5479105700785096, "learning_rate": 1.2428693002358716e-09, "loss": 0.0998, "step": 13849 }, { "epoch": 0.99, "grad_norm": 24.00513843341013, "learning_rate": 1.2171120948384352e-09, "loss": 0.5737, "step": 13850 }, { "epoch": 0.99, "grad_norm": 4.163546374030315, "learning_rate": 1.1916245548559967e-09, "loss": 0.7675, "step": 13851 }, { "epoch": 0.99, "grad_norm": 4.936821184864747, "learning_rate": 1.166406681664678e-09, "loss": 0.9077, "step": 13852 }, { "epoch": 0.99, "grad_norm": 4.574901688829418, "learning_rate": 1.1414584766233916e-09, "loss": 0.2845, "step": 13853 }, { "epoch": 0.99, "grad_norm": 4.519455424134272, "learning_rate": 1.1167799410793934e-09, "loss": 0.3416, "step": 13854 }, { "epoch": 0.99, "grad_norm": 17.346610303585337, "learning_rate": 1.0923710763627304e-09, "loss": 0.9161, "step": 13855 }, { "epoch": 0.99, "grad_norm": 10.073792436454854, "learning_rate": 1.0682318837906824e-09, "loss": 0.6925, "step": 13856 }, { "epoch": 0.99, "grad_norm": 9.395131117065148, "learning_rate": 1.0443623646660961e-09, "loss": 0.5333, "step": 13857 }, { "epoch": 0.99, "grad_norm": 3.883258432723736, "learning_rate": 1.02076252027572e-09, "loss": 0.856, "step": 13858 }, { "epoch": 0.99, "grad_norm": 9.279929136431138, "learning_rate": 9.974323518929795e-10, "loss": 0.6873, "step": 13859 }, { "epoch": 0.99, "grad_norm": 6.701746113504047, "learning_rate": 9.743718607768682e-10, "loss": 0.345, "step": 13860 }, { "epoch": 0.99, "grad_norm": 7.229325340667933, "learning_rate": 9.515810481708353e-10, "loss": 0.7754, "step": 13861 }, { "epoch": 0.99, "grad_norm": 3.2948089936053675, "learning_rate": 9.290599153044533e-10, "loss": 0.1353, "step": 13862 }, { "epoch": 0.99, "grad_norm": 15.668948313439536, "learning_rate": 9.06808463393416e-10, "loss": 0.7695, "step": 13863 }, { "epoch": 0.99, "grad_norm": 10.94196963912871, "learning_rate": 8.848266936367645e-10, "loss": 0.765, "step": 13864 }, { "epoch": 0.99, "grad_norm": 10.993814773319375, "learning_rate": 8.63114607221327e-10, "loss": 0.5029, "step": 13865 }, { "epoch": 0.99, "grad_norm": 3.2993499087745253, "learning_rate": 8.416722053178339e-10, "loss": 0.1585, "step": 13866 }, { "epoch": 0.99, "grad_norm": 3.600307348451471, "learning_rate": 8.204994890836926e-10, "loss": 0.135, "step": 13867 }, { "epoch": 0.99, "grad_norm": 5.142161326308924, "learning_rate": 7.995964596602124e-10, "loss": 0.8458, "step": 13868 }, { "epoch": 0.99, "grad_norm": 16.228915613461403, "learning_rate": 7.789631181759349e-10, "loss": 0.4503, "step": 13869 }, { "epoch": 0.99, "grad_norm": 2.8257883100880874, "learning_rate": 7.585994657433038e-10, "loss": 0.0995, "step": 13870 }, { "epoch": 0.99, "grad_norm": 2.0172381693241785, "learning_rate": 7.385055034608846e-10, "loss": 0.0681, "step": 13871 }, { "epoch": 0.99, "grad_norm": 4.464323846818163, "learning_rate": 7.186812324133651e-10, "loss": 0.1113, "step": 13872 }, { "epoch": 0.99, "grad_norm": 11.799659943999735, "learning_rate": 6.991266536693353e-10, "loss": 0.4311, "step": 13873 }, { "epoch": 0.99, "grad_norm": 27.20586495154389, "learning_rate": 6.798417682840619e-10, "loss": 1.0149, "step": 13874 }, { "epoch": 0.99, "grad_norm": 4.851017973139592, "learning_rate": 6.608265772978239e-10, "loss": 0.9255, "step": 13875 }, { "epoch": 1.0, "grad_norm": 3.902349402459062, "learning_rate": 6.420810817364676e-10, "loss": 0.1928, "step": 13876 }, { "epoch": 1.0, "grad_norm": 5.06492777911885, "learning_rate": 6.236052826114058e-10, "loss": 0.8236, "step": 13877 }, { "epoch": 1.0, "grad_norm": 9.78651714066697, "learning_rate": 6.05399180919064e-10, "loss": 0.4201, "step": 13878 }, { "epoch": 1.0, "grad_norm": 9.546285323495352, "learning_rate": 5.874627776419895e-10, "loss": 0.4432, "step": 13879 }, { "epoch": 1.0, "grad_norm": 4.355633001201917, "learning_rate": 5.697960737471864e-10, "loss": 0.8303, "step": 13880 }, { "epoch": 1.0, "grad_norm": 3.912054609493351, "learning_rate": 5.523990701883364e-10, "loss": 0.2102, "step": 13881 }, { "epoch": 1.0, "grad_norm": 3.3808709754134725, "learning_rate": 5.352717679030228e-10, "loss": 0.167, "step": 13882 }, { "epoch": 1.0, "grad_norm": 3.1853236271497196, "learning_rate": 5.184141678160614e-10, "loss": 0.192, "step": 13883 }, { "epoch": 1.0, "grad_norm": 2.6745032822273256, "learning_rate": 5.01826270836725e-10, "loss": 0.0817, "step": 13884 }, { "epoch": 1.0, "grad_norm": 2.304792816558091, "learning_rate": 4.855080778598531e-10, "loss": 0.1035, "step": 13885 }, { "epoch": 1.0, "grad_norm": 6.945442502361783, "learning_rate": 4.694595897652976e-10, "loss": 0.2798, "step": 13886 }, { "epoch": 1.0, "grad_norm": 3.4690386277629592, "learning_rate": 4.536808074190324e-10, "loss": 0.1568, "step": 13887 }, { "epoch": 1.0, "grad_norm": 2.3015397499257797, "learning_rate": 4.3817173167315374e-10, "loss": 0.1118, "step": 13888 }, { "epoch": 1.0, "grad_norm": 7.415860537008874, "learning_rate": 4.229323633625493e-10, "loss": 0.4112, "step": 13889 }, { "epoch": 1.0, "grad_norm": 3.1185224432697036, "learning_rate": 4.079627033110045e-10, "loss": 0.1254, "step": 13890 }, { "epoch": 1.0, "grad_norm": 3.0771367562810434, "learning_rate": 3.932627523250965e-10, "loss": 0.2289, "step": 13891 }, { "epoch": 1.0, "grad_norm": 4.6701465403488625, "learning_rate": 3.788325111980795e-10, "loss": 0.95, "step": 13892 }, { "epoch": 1.0, "grad_norm": 4.562091846170025, "learning_rate": 3.6467198070877506e-10, "loss": 0.3284, "step": 13893 }, { "epoch": 1.0, "grad_norm": 10.342956701244322, "learning_rate": 3.507811616204615e-10, "loss": 0.8426, "step": 13894 }, { "epoch": 1.0, "grad_norm": 3.385369130566439, "learning_rate": 3.3716005468309443e-10, "loss": 0.197, "step": 13895 }, { "epoch": 1.0, "grad_norm": 50.48501327902202, "learning_rate": 3.238086606310864e-10, "loss": 0.7046, "step": 13896 }, { "epoch": 1.0, "grad_norm": 22.67155185170702, "learning_rate": 3.1072698018441703e-10, "loss": 1.1515, "step": 13897 }, { "epoch": 1.0, "grad_norm": 4.433168891521563, "learning_rate": 2.979150140497433e-10, "loss": 0.2196, "step": 13898 }, { "epoch": 1.0, "grad_norm": 10.493801002847531, "learning_rate": 2.8537276291706883e-10, "loss": 0.4144, "step": 13899 }, { "epoch": 1.0, "grad_norm": 3.4758929594598227, "learning_rate": 2.7310022746418473e-10, "loss": 0.1996, "step": 13900 }, { "epoch": 1.0, "grad_norm": 3.651045555282577, "learning_rate": 2.6109740835222887e-10, "loss": 0.1458, "step": 13901 }, { "epoch": 1.0, "grad_norm": 3.1047534405180897, "learning_rate": 2.493643062290163e-10, "loss": 0.1375, "step": 13902 }, { "epoch": 1.0, "grad_norm": 4.693909419271494, "learning_rate": 2.379009217273742e-10, "loss": 0.2909, "step": 13903 }, { "epoch": 1.0, "grad_norm": 2.427585036500806, "learning_rate": 2.2670725546625194e-10, "loss": 0.1255, "step": 13904 }, { "epoch": 1.0, "grad_norm": 4.134280178657389, "learning_rate": 2.1578330804905566e-10, "loss": 0.8926, "step": 13905 }, { "epoch": 1.0, "grad_norm": 4.142116570844552, "learning_rate": 2.0512908006475873e-10, "loss": 0.8601, "step": 13906 }, { "epoch": 1.0, "grad_norm": 2.804818642689913, "learning_rate": 1.9474457208901176e-10, "loss": 0.1425, "step": 13907 }, { "epoch": 1.0, "grad_norm": 15.075226749651488, "learning_rate": 1.8462978468081205e-10, "loss": 1.1196, "step": 13908 }, { "epoch": 1.0, "grad_norm": 4.886131654691152, "learning_rate": 1.7478471838694445e-10, "loss": 0.1091, "step": 13909 }, { "epoch": 1.0, "grad_norm": 4.999531947310394, "learning_rate": 1.6520937373809555e-10, "loss": 0.7902, "step": 13910 }, { "epoch": 1.0, "grad_norm": 4.293450480414804, "learning_rate": 1.5590375125107416e-10, "loss": 0.1502, "step": 13911 }, { "epoch": 1.0, "grad_norm": 5.0520925273009745, "learning_rate": 1.46867851427146e-10, "loss": 0.9757, "step": 13912 }, { "epoch": 1.0, "grad_norm": 14.492113524546633, "learning_rate": 1.3810167475425406e-10, "loss": 1.0433, "step": 13913 }, { "epoch": 1.0, "grad_norm": 4.740859463972718, "learning_rate": 1.2960522170535338e-10, "loss": 0.191, "step": 13914 }, { "epoch": 1.0, "grad_norm": 13.19295582331631, "learning_rate": 1.2137849273896607e-10, "loss": 0.4248, "step": 13915 }, { "epoch": 1.0, "grad_norm": 3.799709610444513, "learning_rate": 1.134214882980711e-10, "loss": 0.1528, "step": 13916 }, { "epoch": 1.0, "grad_norm": 3.9307081805537063, "learning_rate": 1.0573420881287988e-10, "loss": 0.2201, "step": 13917 }, { "epoch": 1.0, "grad_norm": 4.319258582213797, "learning_rate": 9.831665469750562e-11, "loss": 0.7894, "step": 13918 }, { "epoch": 1.0, "grad_norm": 4.402036770408262, "learning_rate": 9.116882635273883e-11, "loss": 0.1876, "step": 13919 }, { "epoch": 1.0, "grad_norm": 3.3961016417486487, "learning_rate": 8.429072416327177e-11, "loss": 0.1335, "step": 13920 }, { "epoch": 1.0, "grad_norm": 3.6650526339906824, "learning_rate": 7.768234850102919e-11, "loss": 0.1926, "step": 13921 }, { "epoch": 1.0, "grad_norm": 3.8268034343240505, "learning_rate": 7.134369972183753e-11, "loss": 0.3015, "step": 13922 }, { "epoch": 1.0, "grad_norm": 2.799700054321997, "learning_rate": 6.527477816764549e-11, "loss": 0.1198, "step": 13923 }, { "epoch": 1.0, "grad_norm": 20.819301591884052, "learning_rate": 5.947558416652399e-11, "loss": 1.0687, "step": 13924 }, { "epoch": 1.0, "grad_norm": 10.56942206155229, "learning_rate": 5.3946118030445694e-11, "loss": 0.5434, "step": 13925 }, { "epoch": 1.0, "grad_norm": 3.0702473821204954, "learning_rate": 4.86863800586157e-11, "loss": 0.179, "step": 13926 }, { "epoch": 1.0, "grad_norm": 4.227815004073505, "learning_rate": 4.369637053414089e-11, "loss": 0.1744, "step": 13927 }, { "epoch": 1.0, "grad_norm": 20.988630134415434, "learning_rate": 3.897608972625033e-11, "loss": 0.8389, "step": 13928 }, { "epoch": 1.0, "grad_norm": 14.189044916691822, "learning_rate": 3.452553788974022e-11, "loss": 0.7497, "step": 13929 }, { "epoch": 1.0, "grad_norm": 17.590562738984527, "learning_rate": 3.0344715264973845e-11, "loss": 0.6555, "step": 13930 }, { "epoch": 1.0, "grad_norm": 3.4170324867735733, "learning_rate": 2.643362207677136e-11, "loss": 0.1442, "step": 13931 }, { "epoch": 1.0, "grad_norm": 10.956076823769655, "learning_rate": 2.279225853663025e-11, "loss": 0.4293, "step": 13932 }, { "epoch": 1.0, "grad_norm": 3.484899433359579, "learning_rate": 1.942062484106e-11, "loss": 0.224, "step": 13933 }, { "epoch": 1.0, "grad_norm": 21.757292585374636, "learning_rate": 1.6318721172137174e-11, "loss": 0.7822, "step": 13934 }, { "epoch": 1.0, "grad_norm": 4.279028729094106, "learning_rate": 1.3486547696395235e-11, "loss": 0.651, "step": 13935 }, { "epoch": 1.0, "grad_norm": 16.5313718457164, "learning_rate": 1.0924104567044957e-11, "loss": 0.6035, "step": 13936 }, { "epoch": 1.0, "grad_norm": 22.007397765997577, "learning_rate": 8.631391922309107e-12, "loss": 0.7159, "step": 13937 }, { "epoch": 1.0, "grad_norm": 19.07959125563077, "learning_rate": 6.608409886532663e-12, "loss": 0.6281, "step": 13938 }, { "epoch": 1.0, "grad_norm": 20.73833720424769, "learning_rate": 4.855158567407259e-12, "loss": 0.914, "step": 13939 }, { "epoch": 1.0, "grad_norm": 3.036839349109236, "learning_rate": 3.3716380609671863e-12, "loss": 0.0698, "step": 13940 }, { "epoch": 1.0, "grad_norm": 3.0679417395275927, "learning_rate": 2.15784844603828e-12, "loss": 0.071, "step": 13941 }, { "epoch": 1.0, "grad_norm": 3.782458876863421, "learning_rate": 1.2137897892339211e-12, "loss": 0.1734, "step": 13942 }, { "epoch": 1.0, "grad_norm": 4.864184598976832, "learning_rate": 5.394621405141465e-13, "loss": 0.8037, "step": 13943 }, { "epoch": 1.0, "grad_norm": 3.3308421542445914, "learning_rate": 1.348655370714269e-13, "loss": 0.1177, "step": 13944 }, { "epoch": 1.0, "step": 13944, "total_flos": 103289085389312.0, "train_loss": 0.6157975893083487, "train_runtime": 78786.2953, "train_samples_per_second": 2.832, "train_steps_per_second": 0.177 } ], "logging_steps": 1.0, "max_steps": 13945, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 2000, "total_flos": 103289085389312.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }